Volume 188

THE

Number 1

BIOLOGICAL
BULLETIN

FEBRUARY/MARCH, 1995

Published by the Marine Biological Laboratory

75V?-

THE

BIOLOGICAL BULLETIN

PUBLISHED BY
THE MARINE BIOLOGICAL LABORATORY

Associate Editors

FEB 1

PETER A. V. ANDERSON. The Whitney Laboratory. University of Florida

WILLIAM D. COHEN. Hunter College. City University of New York

Woods Hda, (*
DAVID EPEL. Hopkins Marine Station. Stanford University

.i. MALCOLM SHICK, University of Maine, Orono

Editorial Board

THOMAS H. DIETZ, Louisiana State Llniversity BARUCH RINKEVICH, Israel Oceanographic &

Limnological Research Ltd.

DAPHNE GAIL FAUTIN, University of Kansas

RICHARD STRATHMANN. Friday Harbor Laboratories,

WILLIAM F. GILLV, Hopkins Marine Station, Stanford University of Washington

University STEVEN VOGEL. Duke University

ROGER T. HANLON, Marine Biomedical Institute. J. HERBERT WAITE, Llniversity of Delaware

University of Texas Medical Branch

SARAH ANN WOODIN, University of South Carolina

CHARLES B. METZ, University of Miami

RICHARD K. ZIMMER-FALIST. Llniversity of South

K. RANGA RAO, University of West Florida Carolina

Editor MICHAEL J. GREENBERG. The Whitney Laboratory. University of Rorida
Managing Editor: PAMELA L. CLAPP, Marine Biological Laboratory

FEBRUARY/MARCH, 1995

Printed and Issued by
LANCASTER PRESS, Inc.

3575 HEMPLAND ROAD
LANCASTER, PA

THE BIOLOGICAL BULLETIN

THE BIOLOGICAL BULLETIN is published six times a year by the Marine Biological Laboratory, MBL
Street, Woods Hole, Massachusetts 02543.

Subscriptions and similar matter should be addressed to Subscription Manager. THE BIOLOGICAL BUL-
LETIN, Marine Biological Laboratory. Woods Hole, Massachusetts 02543. Single numbers, $37.50. Sub-
scription per volume (three issues). $92.50 ($185.00 per year for six issues).

Communications relative to manuscripts should be sent to Michael J. Greenberg, Editor-in-Chief, or
Pamela L. Clapp. Managing Editor, at the Marine Biological Laboratory, Woods Hole, Massachusetts 02543.
Telephone: (508) 548-3705, ext. 428. FAX: 508-540-6902. E-mail: pclappw-hoh. mbl.edu.

THE BIOLOGICAL BLILLETIN is indexed in bibliographic services including Imte.\ Mciticus and MEDLINE,
Chemical Abstracts, and Current Contents.

Printed on acid free paper,
effective with Volume 1 80, Issue 1 . 1 99 1

POSTMASTER: Send address changes to THE BIOLOGICAL BLILLETIN, Manne Biological Laboratory,

Woods Hole, MA 02543.

Copyright 1995, by the Manne Biological Laboratory

Second-class postage paid at Woods Hole. MA. and additional mailing offices.

ISSN 0006-3185

INSTRUCTIONS TO AUTHORS

The Biological Bulletin accepts outstanding original research
reports of general interest to biologists throughout the world.
Papers are usually of intermediate length (10-40 manuscript
pages). A limited number of solicited review papers may be ac-
cepted after formal review. A paper will usually appear within
four months after its acceptance.

Very short, especially topical papers (less than 9 manuscript
pages including tables, figures, and bibliography) will be pub-
lished in a separate section entitled "Research Notes." A Re-
search Note in The Biological Bul/ciin follows the format of
similar notes in Nature. It should open with a summary para-
graph of 150 to 200 words comprising the introduction and the
conclusions. The rest of the text should continue on without
subheadings, and there should be no more than 30 references.
References should be referred to in the text by number, and
listed in the Literature Cited section in the order that they appear
in the text. Unlike references in Nature, references in the Re-
search Notes section should conform in punctuation and ar-
rangement to the style of recent issues of The Biological Bulletin
Materials and Methods should be incorporated into appropriate
figure legends. See the article by Lohmann et al. (October 1990,
Vol. 179: 214-218) for sample style. A Research Note will usually
appear wr.hin two months after its acceptance.

The Editorial Board requests that regular manuscripts con-
form to the requirements set below; those manuscripts that do
not conform will be returned to authors for correction before
review.

1. Manuscripts. Manuscripts, including figures, should
be submitted in triplicate. (Xerox copies of photographs are not
acceptable for review purposes.) The submission letter accom-
panying the manuscript should include a telephone number, a
FAX number, and (if possible) an E-mail address for the cor-
responding author. The original manuscript must be typed in
no smaller than 12 pitch or 10 point, using double spacing (in-
cluding figure legends, footnotes, bibliography, etc.) on one side

of 16- or 20-lb. bond paper, 8'/2 by 11 inches. Please, no right
justification. Manuscripts should be proofread carefully and er-
rors corrected legibly in black ink. Pages should be numbered
consecutively. Margins on all sides should be at least 1 inch (2.5
cm). Manuscripts should conform to the Council of Biology Ed-
itors Style Manual. 5th Edition (Council of Biology Editors.
1983) and to American spelling. Unusual abbreviations should
be kept to a minimum and should be spelled out on first reference
as well as defined in a footnote on the title page. Manuscripts
should be divided into the following components: Title page.
Abstract (of no more than 200 words). Introduction, Materials
and Methods, Results, Discussion. Acknowledgments, Literature
Cited, Tables, and Figure Legends. In addition, authors should
supply a list of words and phrases under which the article should
be indexed.

2. Title page. The title page consists of a condensed title
or running head of no more than 35 letters and spaces, the
manuscript title, authors' names and appropriate addresses, and
footnotes listing present addresses, acknowledgments or contri-
bution numbers, and explanation of unusual abbreviations.

3. Figures. The dimensions of the printed page, 7 by 9
inches, should be kept in mind in preparing figures for publi-
cation. We recommend that figures be about 1 '/: times the linear
dimensions of the final printing desired, and that the ratio of
the largest to the smallest letter or number and of the thickest
to the thinnest line not exceed 1:1.5. Explanatory matter gen-
erally should be included in legends, although axes should always
be identified on the illustration itself. Figures should be prepared
for reproduction as either line cuts or halftones. Figures to be
reproduced as line cuts should be unmounted glossy photo-
graphic reproductions or drawn in black ink on white paper,
good-quality tracing cloth or plastic, or blue-lined coordinate
paper. Those to be reproduced as halftones should be mounted
on board, with both designating numbers or letters and scale
bars affixed directly to the figures. All figures should be numbered

in consecutive order, with no distinction between text and plate
figures. The author's name and an arrow indicating orientation
should appear on the reverse side of all figures.

4. Tables, footnotes, figure legends, etc. Authors should
follow the style in a recent issue of The Biological Bulletin in
preparing table headings, figure legends, and the like. Because
of the high cost of setting tabular material in type, authors are
asked to limit such material as much as possible. Tables, with
their headings and footnotes, should be typed on separate sheets,
numbered with consecutive Roman numerals, and placed after
the Literature Cited. Figure legends should contain enough in-
formation to make the figure intelligible separate from the text.
Legends should be typed double spaced, with consecutive Arabic
numbers, on a separate sheet at the end of the paper. Footnotes
should be limited to authors' current addresses, acknowledg-
ments or contribution numbers, and explanation of unusual
abbreviations. All such footnotes should appear on the title page.
Footnotes are not normally permitted in the body of the text.

5. Literature cited. In the text, literature should be cited
by the Harvard system, with papers by more than two authors
cited as Jones el al., 1980. Personal communications and ma-
terial in preparation or in press should be cited in the text only,
with author's initials and institutions, unless the material has
been formally accepted and a volume number can be supplied.
The list of references following the text should be headed Lit-
erature Cited, and must be typed double spaced on separate
pages, conforming in punctuation and arrangement to the style
of recent issues of The Biological Bulletin. Citations should in-
clude complete titles and inclusive pagination. Journal abbre-
viations should normally follow those of the U. S. A. Standards
Institute (USASI), as adopted by BIOLOGICAL ABSTRACTS and
CHEMICAL ABSTRACTS, with the minor differences set out below.
The most generally useful list of biological journal titles is that
published each year by BIOLOGICAL ABSTRACTS (BIOSIS List of
Serials; the most recent issue). Foreign authors, and others who
are accustomed to using THE WORLD LIST OF SCIENTIFIC PE-
RIODICALS, may find a booklet published by the Biological
Council of the U.K. (obtainable from the Institute of Biology.
41 Queen's Gate, London, S.W.7. England, U.K.) useful, since
it sets out the WORLD LIST abbreviations for most biological
journals with notes of the USASI abbreviations where these differ.

CHEMICAL ABSTRACTS publishes quarterly supplements of ad-
ditional abbreviations. The following points of reference style
for THE BIOLOGICAL BULLETIN differ from USASI (or modified
WORLD LIST) usage:

A. Journal abbreviations, and book titles, all underlined
(for italics)

B. All components of abbreviations with initial capitals
(not as European usage in WORLD LIST e.g.. J. Cell. Camp.
Physio! . NOT / cell. comp. Physiol.)

C. All abbreviated components must be followed by a pe-
riod, whole word components must not (i.e.. J. Cancer Res.)

D. Space between all components (e.g.. J Cell. Comp
Physiol., not JCell.Comp.Physiol )

E. Unusual words in journal titles should be spelled out
in full, rather than employing new abbreviations invented by
the author. For example, use Rit I fsindqffelags Islendinga with-
out abbreviation.

F. All single word journal titles in full (e.g.. Veliger. Ecol-
ogy. Brain).

G. The order of abbreviated components should be the
same as the word order of the complete title (i.e.. Proc. and
Trans, placed where they appear, not transposed as in some
BIOLOGICAL ABSTRACTS listings).

H. A few well-known international journals in their pre-
ferred forms rather than WORLD LIST or USASI usage (e.g..
Nature. Science. Evolution NOT Nature. Loud.. Science. N.Y.:
Evolution. Lancaster. Pa.)

6. Reprints, page proofs, and charges. Authors receive their
first 100 reprints (without covers) free of charge. Additional re-
prints may be ordered at time of publication and normally will
be delivered about two to three months after the issue date.
Authors (or delegates for foreign authors) will receive page proofs
of articles shortly before publication. They will be charged the
current cost of printers' time for corrections to these (other than
corrections of printers' or editors' errors). Other than these
charges for authors' alterations. The Biological Bulletin does not
have page charges.

iii

Reference: Biol. Bull 188: 1-4. (February/March.

Responses of the Medaka Fish Egg (Oryzias latipes) to

the Photolysis of Microinjected Nitrophenyl-EGTA, a

Photolabile Calcium Chelator

RICHARD A. FLUCK

Biology Department, Franklin and Marshall College, P.O. Box 3003. Lancaster, Pennsylvania 17604

Photolabile calcium dictators (calcium cages) can be
used to elevate cytosolic [Ca 2+ ] at specific sites and times
(I. 2, 3). They have been especially valuable in flash pho-
tolysis studies of muscle contraction (2) and secretion (4,
5). In the present report. I describe several responses of
medaka eggs to the photolysis ofmicroinjectednitrophenyl-
EGTA (NP-EGTA), a new calcium cage (6). IVhen unfer-
tilized eggs injected with NP-EGTA were irradiated with
ultraviolet irradiation in a small region of the egg, the eggs
were activated and ooplasm within the irradiated region
contracted and accumulated there. Eggs into which NP-
EGTA was injected could a/so be fertilized. Subsequent
irradiation of such eggs, in addition to causing the con-
traction and accumulation of ooplasm, also caused a global
contraction of dividing blastomeres and the contraction
and blebbing of embryonic cells for up to 4 days after fer-
tilization. Injection of NP-EGTA had no apparent effect
on the maturation of fertilized eggs, which developed nor-
mally and hatched.

The methods for dissection of gonads from breeding
medaka, preparation of gametes, and in vitro fertilization
of eggs have been described previously (7). Gonads, ga-
metes, and zygotes were prepared in a balanced saline
solution (BSS: 1 1 1 mM NaCl; 5.37 mM KG; 1.0 mM
CaCV, 0.6 mM MgSO 4 ; 5 mM HEPES, pH 7.3). A nor-
malized time (T n ) scale in which the time between fertil-
ization and the beginning of cytokinesis is 1 unit was used
to indicate the relative temporal positions of events. A
total of 80 eggs from 6 females were used in these exper-
iments; 44 of the eggs were monitored in detail and an
additional 36 were monitored intermittently. The exper-
iments were conducted at room temperature (23-25C).

Received 30 August 1994; accepted 14 October 1994.

Methods described previously (8, 9, 10) were used to
microinject 1.4-5.6 nl of an aqueous solution of NP-
EGTA/Ca 2+ (50 mM NP-EGTA, tetrapotassium salt;
39.6 mM CaCl 2 ; 10 mM HEPES. pH 7.3) into the thin
peripheral layer of ooplasm. Assuming an accessible vol-
ume of 27.6 nl (9). injection of these volumes would give
a final ooplasm ic concentration of 2. 5-10.0 mM NP-
EGTA. After microinjection. the eggs were either placed
in a darkened cabinet for subsequent use or fertilized
within 5 min. During the experiments, the laboratory was
only dimly illuminated with incandescent lamps.

For microscopic observation and irradiation, the eggs
were transferred to a microscope slide on which a cover
glass was supported by four pillars of petroleum jelly (7).
An Osram 100 W mercury arc lamp was used to irradiate
the eggs with ultraviolet light. The light from the lamp
was passed through a filter cube (Omega Optical) con-
taining a 360 DF 40 exciter filter, a DC 405 dichroic mir-
ror, and a 486 DF32 barrier filter. An octagonal diaphragm
was used to control the size of the light beam (in most
experiments, it was either 200 //m or 475 /urn), and neutral
density filters (Omega Optical) were used to reduce the
light intensity by either 34-fold or 286-fold. Ultraviolet
light was projected onto the egg via one of three objective
lenses (Nikon): Plan 4X.N.A. = 0. 1 ; Fluor/Ph 2 DL 10X,
N.A. = 0.5; Fluor/Ph 3 DL 20X. N.A. = 0.75. In most
experiments, the equatorial region of the egg that is, a
region along a meridian and midway between the animal
and vegetal poles was illuminated en profit (an edge of
the egg was irradiated) via the 10x objective lens. Light
intensity was measured with a UVX radiometer with a
long wave sensor (UVP, Inc.). Given a light intensity of
523 //W cm~- (referred to as "high intensity" hereafter)
and assuming that all the ultraviolet light was of wave-
length 360 nm, I calculated an incident light intensity of

R. A. FLUCK

4.9 X 10 4 quanta second" ' ^m 2 ( 10X objective lens, no
neutral density filter). To monitor the eggs during irra-
diation, they were transilluminated with light from a
quartz-halogen lamp, using a heat filter (KG5) and a 486
DF32 filter, and the images were recorded via a SIT cam-
era (Dage/MTI) and a time-lapse videocassette recorder.

To monitor the development of fertilized eggs after the
first cell cycle, they were transferred to embryo rearing
medium (17 mM NaCl; 0.4 mM KCI; 0.3 mM CaCl : ;
0.67 mM MgSCV, 0.001 g/1 methylene blue).

When unfertilized eggs into which 1.4-5. 6 nl NP-
EGTA had been injected were irradiated with UV light,
they activated within 16 3 s (X SD, n = 5), as evi-
denced by the exocytosis of cortical vesicles. Exocytosis
began within the irradiated region and spread as a wave
over the rest of the egg. Eggs were photoactivated even
after the light intensity was reduced 34-fold with a neutral
density filter; but when a second neutral density filter was
added, reducing the light intensity an additional 8.5-fold,
the eggs were not activated even after 3 min of continuous
irradiation. Irradiation of unfertilized eggs that had not
received NP-EGTA did not cause them to activate.

Continued irradiation of photoactivated eggs caused
both ooplasm and oil droplets to accumulate in and next
to the irradiated zone (Fig. 1 A). Staining with rhodamine
phalloidin showed that these accumulations of ooplasm

contained filamentous actin (F-actin, Fig. IB). Such ac-
cumulations of ooplasm and F-actin were identified in 1 8
eggs, 13 of which had been photoactivated and 5 of which
had been fertilized. Moreover, the caps of ooplasm formed
in eggs into which either 1.4 nl or 4.2 nl of NP-EGTA
had been injected and in eggs that were irradiated either
intermittently (5 s on/1 15 s off) with a high intensity of
light or continuously with a 34-fold lower light intensity.
However, when the light intensity was lowered 289-fold,
ooplasm neither contracted nor accumulated in the ir-
radiated zone. When UV irradiation was intermittent, the
ooplasm within the irradiated zone usually contracted
each time it was irradiated. For example, the region in
the egg shown in Figure 1 was irradiated 29 times for 5 s
and contracted 16 times, and a sibling egg contracted each
of the 3 1 times it was irradiated for 5 s. Each contraction
appeared to pull ooplasm and nearby oil droplets toward
the irradiated region. Eggs that were parthenogenetically
activated by the injection itself and grown in the dark
segregated normally (as do eggs that have been parthe-
nogenetically activated by pricking; Fluck, unpub. obs.),
with a cap of ooplasm forming at the animal pole and the
oil droplets segregating toward the vegetal pole.

Eggs into which NP-EGTA had been injected could
also be fertilized. In most fertilized eggs, cortical vesicles
in one small region of the egg, presumably near the in-

Kifjure 1. Accumulation of ooplasm within a UV-irradiated region of an egg. NP-EGTA/Ca ;+ (4.2 nl)
was microimected into this egg, which was then irradiated en pro/i/ in a region (approximately defined by
the filled circles) along a meridian and midway between its animal and vegetal poles. The UV light was
projected through the 10x objective lens with no neutral density filter in the light path. Ultraviolet irradiation
was intermittent, with light pulses delivered for 5 s at 2-min intervals fora total of 79 min (until 7 n = 1.0).
The egg was then fixed overnight at room temperature with formaldehyde dissolved in an actin-stabilizing
buffer(24): 3.7% formaldehyde (Electron Microscopy Sciences. Fort Washington. PA). 100 mM KCI, 5 mM
MgCl,. 2 mM EGTA. 10 mM PIPES. pH 6.8. It was then dechorionated with fine forceps, permeabilized
for 15 min with 0.3% Triton X-KH) in BSS, and stained for 30 min in 0.25 ^M rhodamine phalloidin
(Molecular Probes, Inc., Eugene. OR) dissolved in BSS. The irradiated region of the egg was photographed
with bnghtfield optics just before fixation I A) and with epi-illumination after staining the egg with rhodamine
phalloidin (B). Note the accumulation of ooplasm, oil droplets, and F-actin in the irradiated region. Scale
bars. 100 /jm

CALCIUM CAGE IN MEDAKA EGGS

jection site, did not undergo exocytosis. Irradiation of fer-
tilized eggs caused ooplasm to accumulate within the ir-
radiated region, but no such accumulations were seen in
fertilized eggs that were grown in the dark; in such dark-
grown eggs, the ooplasm and its contents appeared to seg-
regate normally. Moreover, irradiation of unfertilized eggs
that had not received NP-EGTA caused neither contrac-
tion of the ooplasm nor accumulation of ooplasm within
the irradiated region.

All eggs that received 1.4 nl of NP-EGTA underwent
cytokinesis. Of eight such fertilized eggs whose subsequent
development was monitored, four hatched and the other
four underwent extensive morphogenesis but did not
hatch. Cleavage was abnormal in eggs that received either
4.2 or 5.6 nl of NP-EGTA, and the embryos did not de-
velop further. Irradiation of eggs (that received 1.4 nl of
NP-EGTA and were subsequently fertilized) during early
cleavage caused cells in the light beam to contract globally.
Moreover, irradiation of early gastrulae caused blebbing
and global contraction of deep blastomeres, but irradiation
of the yolk sac in stage 19, 22. and 25 embryos (that is,
up to 4 days after fertilization) caused contractions that
appeared similar to those seen during the rhythmic con-
traction waves that occur in the stellate layer of the me-
daka embryo (11). Cells of embryos that did not receive
NP-EGTA failed to contract when they were irradiated
with UV light.

Taken together, these findings show NP-EGTA to be
a useful new reagent for cell and developmental biologists.
Several properties of this compound its high affinity for
Ca :+ (6), the approximately 10,000-fold decrease in its
affinity for Ca 24 upon photolysis (6), its weak fluorescence
(6), and its persistence and low toxicity in the teleost em-
bryo (this study) appear to make it particularly suitable
for studying events that have been linked to elevations in
cytosolic [Ca 2+ ]: egg activation (12, 13), ooplasmic seg-
regation (9, 10), nuclear envelope breakdown (14), mitosis
(15, 16, 17), cytokinesis (8, 18), and neuronal growth cone
motility (19, 20).

The accumulation of ooplasm and F-actin within the
UV-irradiated region of the egg is consistent with the hy-
pothesis that cytosolic calcium gradients organize devel-
opmental localization in eggs ( 10, 2 1 , 22, 23). At present,
however, the evidence consistent with this hypothesis, in-
cluding that presented in the present report, is indirect
and must be extended by using aequorin or a fluorescent
calcium indicator to measure cytosolic [Ca : ' ] in eggs dur-
ing and after photolysis of NP-EGTA. Full exploitation
of the apparent promise of NP-EGTA will require the
development of a dextran-conjugated form of the mole-
cule and the exploration of wider ranges of intracellular
concentrations of the cage, shapes of the irradiated region
(for example, a narrow rectangle that could elevate cy-

tosolic [Ca 2+ ] in a narrow band in a cell), and light inten-
sities.

Acknowledgments

Lionel Jaffe suggested the use of calcium cages to gen-
erate zones of elevated calcium in medaka eggs. I am
grateful to Jack Kaplan and Graham Ellis-Davies for pro-
viding the NP-EGTA used in these studies; to Alan Bruns
for helpful discussions about units of light intensity; and
to Andrew Miller for helping to improve the text of the
manuscript. This work was supported by NSF DCB-
9017210 and NSF MCB-9316125.

Literature Cited

1. McCray, J. A., and D. R. Trenlham. 1989. Properties and uses of
photoreactive caged compounds. Aiuui Rev Biophys Biophys.
Clwm. 18: 239-270.

2 Ashley, C. C ., P. J. Griffiths, I. J. Lea, I. P. Mulligan, R. E. Palmer,
and S. J. Simmons. 1991. Use of fluorescent TnC derivatives and
'caged' compounds to study cellular phenomena. Pp. 177-203 in
Cellular Calcium: A Pruetiail Approach. J. G. McCormack and P.
H. Cobbold, eds. 1RL Press, Oxford.

3. Zucker, R. 199-4. Photorelease techniques for raising or lowering
intracellular Ca :+ . Mcth Cell Biol. 40: 31-63.

4. Zucker, R. S., and P. G. llaydon. 1988. Membrane potential has
no direct role in evoking neurotransmitter release. Nature 335: 360-
362.

5. Thomas, P., A. K. Lee, J. G. Wong, and \V. Aimers. 1994. A
triggered mechanism retrieves membrane in seconds after Ca 2+ -
stimulated exocytosis in single pituitary cells. J Cell Biol. 124: 667-
675.

6. Ellis-Davies, C. R., and J. 11. Kaplan. 1994. Nitrophenyl-EGTA.
a photoiabile chelator that selectively binds Ca 2+ with high affinity
and releases it rapidly upon photolysis. Proc. Nat! Acud Sci. US.-t
91: 187-191.

7. Abraham, V.C., S.Gupta, and R. A. Muck. 1993. Ooplasmic seg-
regation in the medaka (Ory:ias latipex) Biol. Bull 184: 1 15-124.

8. Eluck, R. A., A. L. Miller, and L. E. Jaffe. 1991. Slow calcium
waves accompany cytokinesis in medaka fish eggs. J Cell Biol. 115:
1259-1265.

9. Eluck, R. A., A. L. Miller, and L. E. Jaffe. 1992. High calcium
zones at the poles of developing medaka eggs. Biol Bull 183: 70-
77.

10. Fluck, R. A., A. L. Miller, V. C. Abraham, and L. F. Jaffe.

1994. Calcium buffer injections inhibit ooplasmic segregation in

medaka eggs. Biol Bull 186: 254-262.
1 1 Cope, J., R. Eluck, L. Nicklas, L. A. Plumhoff, and S. Sincock.

1990. The stellate layer and rhythmic contractions of the Oryiias

lanpes embryo. J E\f>. tool 254: 270-275.

12. Gilkey, J. C, L. F. Jaffe, E. B. Ridgway, and G. T. Reynolds.
1978. A free calcium wave traverses the activating egg of the me-
daka, Oryiw* lalipe\ ./ Cell Biol 76:448-466.

13. Shen, S. S., and \V. R. Buck. 1993. Sources of calcium in sea
urchin eggs during the fertilization response. Dev. Biol. 157: 157-
169.

14 Browne, C. L., A. L. Miller, R. E. Palazzo, and L. F. Jaffe. 1992. On
the calcium pulse during nuclear envelope breakdown (NEB) in sea
urchin eggs. Biol Bull 183:370-371.

15. Hepler, P. K. 1989. Calcium transients during mitosis: observations
in fluv ./ Cell Biol 109: 2567-2573.

4

R. A. FLUCK

16. VVhitaker, M., and R. Patel. 1990. Calcium and cell cycle control.
Development 108: 525-542.

17. Harris, P. J. 199-4. Caffeine-induced calcium release in sea urchin
eggs and the effect of continuous versus pulsed application on the
mitotic apparatus. Dc\ fiial 161: 370-378.

18. Miller, A. L., R. A. Fluck, J. A. McLaughlin. and L. F. Jaffe.
1993. Calcium buffer injections inhibit cytokinesis in \enornts eggs.
J. Cell Buil 106: 523-534.

19. Eisen, J. S., P. Z. Myers, and M. VVesterfield. 1986. Pathway
selection by growth cones of identified motoneurones in live zebra
fish eggs. Nature 320: 269-27 1 .

20. Lankford, K. L.,andP.C. Lclourneau. 1989. Evidence that calcium

may control neurite outgrowth by regulating the stability of actin
filaments./ Cell Biol. 109: 1229-1243.

2 1 . Jeffery. \V. R. 1982. Calcium lonophore polanzes ooplasmic seg-
regation in ascidian eggs. Science 216: 545-547.

22. Jaffe, L. F. 1986. Calcium and morphogenetic fields. Pp. 271-
288 in Calcium und the Cell. D. Evered and J. Whelan, eds. John
Wiley and Sons, Chichester, England.

23. Bates, \V. R., and \V. R. Jeffery. 1988. Polarization of ooplasmic
segregation and dorsal-ventral axis determination in ascidian em-
bryos. Dev Bin! 130:98-107.

24. Yonemura, S., and S. Kinoshita. 1986. Actin filament organization
in the sand dollar egg cortex. Dev. Bint. 115: 1 7 1 - 1 83.

Reference: Biol. Bull 188: 5-7. (February/March, 1995)

Hemoglobin in the Symbiont-Harboring Gill of the
Marine Gastropod Alviniconcha hessleri

JONATHAN B. WITTENBERG 1 AND JEFFREY L. STEIN 2

1 Department of Physiology and Biophysics, Albert Einstein College of Medicine, Bronx,
New York 10461, and 2 The Agouron Institute, La Jolla, California 92037

Hydrogen sulfide of geochemical origin, mixing at
oceanic hydrothermal vents with oxygen from oceanic sea-
water, supports dense populations ofchemoautrophic, sul-
fur-oxidizing bacteria. Those animals, the vest i men! iferan
worm Riftia pachyptila, certain bivalve molluscs, and the
recently discovered Pacific gastropod Alviniconcha hess-
leri, that interiorize the bacteria as intracellular symbionts
dominate the vent fauna (I, 2). The immense size of these
animals, the large standing crop represented in their dense
communities, and the rapid growth of individuals all attest
to the effective use of an abundant food base. Dense con-
centrations of the mesogastropod Alviniconcha hessleri
(2, 3) were recently discovered at deep-sea hydrothermal
vents at the spreading center in the Mariana Back- Arc
Basin of the Western Pacific. These animals house che-
moantrophic. sulfide-oxidizing bacteria within specialized
cells (bacteriocytes) of their modified gills (2). They are
the only reported example of a symbiotic association be-
tween a gastropod mollusc and intracellular chemoauto-
trophic bacteria. H 'e now show that the modified gill of
Alviniconcha contains hemoglobin at a concentration of
about 65 n/nol/kg wet weight gill. This value falls within
the range. 20-250 nmol hemoglobin per kilogram, en-
countered in the modified symbiont-harboring gills of
many of the sulfide- dependent clams examined but is short
of the very high concentrations, 550 ami 1200 ^mol/kg,
found in Myrtea spinifera and Lucina pectinata respec-
tively (4). Accordingly, bacteriocyte hemoglobin is a feature
common to both gastropod and bivalve symbioses.

Symbioses with intracellular carbon-fixing bacteria,
believed to be dependent on bacteriocyte hemoglobin,
have heretofore been described only in clams of the fam-
ilies Solemyidae, Lucinidae, and Vesicomyidae and in a

Received 2 September 1994; accepted 23 November 1994.

few mussels, Mytilidae, restricted to the genus Bathy-
modiolus (4). The molluscan symbionts fix carbon from
oceanic carbon dioxide into hexoses and supply almost
all of the carbon nutrition of the host (5). Ribulose bis-
phosphate carboxylase/oxygenase (RuBisCO), the enzyme
responsible for carbon dioxide fixation, has been cloned
from the Alviniconcha symbiont and expressed in Esch-
erichia coli (6). The relatively low specificity of the purified
enzyme for carbon dioxide indicates that the intracellular
environment of the endosymbionts may be microaero-
philic for RuBisCO to maintain net carboxylation (7).

Hemoglobins, probably located in the host cytoplasm
and excluded from the peribacterial space and probably
coded by host genes, are a near-constant feature of sym-
bioses between bivalve molluscs and intracellular che-
moautotrophic bacteria (4, 8); such hemoglobins are also
a constant feature of symbioses between plants and in-
tracellular prokaryotic nitrogen-fixing symbionts (9).
Clam gill hemoglobins have been investigated intensively
(10-13), and the three-dimensional structure of one is
known from x-ray diffraction analysis ( 14). The probable
role of the hemoglobin is to bring oxygen and hydrogen
sulfide to the symbiont (8, 15). In the giant tube worm
Riftia. this function is served by blood or coelomic he-
moglobin, which bathe the symbiont-harboring cells and
transport oxygen at the heme and sulfide at a site remote
from the heme (16). In the bacteria-housing clam gill,
these functions are probably served by two separate he-
moglobins of the bacteriocyte cytoplasm ( 1 7). Cytoplasmic
hemoglobins are believed to supplement the diffusion of
free oxygen by adding to it a contribution from oxygen
combined with the protein. In those few hemoglobin-
containing tissues that have been studied intensively, the
hemoglobin is maintained partially desaturated with oxy-
gen, the larger part of the oxygen flow to the intracellular
organelle is carried in combination with the protein, and

J. B. WITTENBERG AND J. L. STEIN

the oxygen pressure is held low (18). This is in accord
with the suggestion that low oxygen pressure is probably
required to allow RuBisCO to maintain net carboxylation
in the Alviniconcha gill (7). The concentration of hydrogen
sultide likewise is probably low, perhaps in the nanomolar
range (8), and the concentration of free hydrogen sulfide
may not be sufficient to support the flux of hydrogen sul-
fide to the symbiont.

A specimen of the Western Pacific hydrothermal vent
gastropod Alviniconcha hessleri was collected by the sub-
mersible Alvin at the "Snail Pit" site (18 10.95' N,
144 43.20' E. about 3650m depth) on Dive number
1 837. 23 April 1987. The gill, 0.58 g wet weight, was stored
frozen in liquid nitrogen and a clear soluble extract pre-
pared in 1.5 ml of buffer (19). Optical direct spectra of
this extract display a prominent narrow feature at 55 1 nm.
This unchanging feature ascribed to reduced soluble sym-

o

CD

tr

8

CD

500

550 600

WAVELENGTH, nm

650

Figure 1. Optical difference spectra of the extract of Alviniconcha
gill. Features ascribable to oxyhemoglobin were inconspicuous in the
direct spectrum of the extract, suggesting that some component of the
solution had consumed the dissolved oxygen. Accordingly the solution
was equilibrated with oxygen gas. The difference spectrum of the oxy-
genated solution minus that of the initial (deoxygenuted) solution (spec-
trum not shown) exhibits conspicuous features at 541 and 579 nm. di-
agnostic of oxygenated hemoglobin. A small feature near 622 nm sug-
gested the presence of feme hemoglobin. This feature increased in
magnitude with time. The difference spectrum in the Soret region (spec-
trum not shown) of a portion of the sample that had been stored for 60
mm at ice temperature minus that of a portion of the solution to which
sodium dithiomte (a reagent that removes oxygen and reduces most
hemeproteins) had been added resembled the difference spectrum of
ferric minus deoxy Lucina Hb II. and exhibited a maximum at 409 nm
and a conspicuous minimum at 434 nm. diagnostic of hemoglobin. This
confirms the presence of a hemoglobin in the solution. The oxygenated
solution was then equilibrated with carbon monoxide. (A) Difference
spectrum: Carbon-monoxide-equilibraled gill extract minus that of the
oxygenated extract. Features at 545, 567, and 581 nm are diagnostic of
hemoglobin. Conversion of oxy- to carbon monoxy hemoglobin shows
that oxygen binding by the hemoglobin is reversible. (B) Difference spec-
trum: Carbon monoxide minus oxy Liicinu Hh II.

419

0.05 ABS

0.01 ABS

400

450

500 550

WAVELENGTH, nm

600

650

Figure 2. B> adding sodium dithionite. all of the hemoglobin present
in the extract ofAlvinianuiiu gill is converted into single chemical species,
permitting quantitative estimation of concentration. Absorbance in the
visible region is amplified sixfold relative to the Soret region. (A) Dif-
ference spectrum: Carbon-monoxide-equilibrated gill extract containing
sodium dilhionile minus the same extract containing sodium dithionite
alone. Well-defined features of 419, 435, 535. 554, 570 and 590 nm are
diagnostic of hemoglobin. (B) Difference spectrum: Carbon monoxide
Lucina lib II minus deoxv (ferrous) Lucina Hb II.

biont bacterial cytochrome 1-552 (20), vanishes in the dif-
ference spectra. Optical difference spectra (Figs. 1 and
2)unequivocally establish the presence of hemoglobin in
the extract. To confirm the identity of the spectral entities,
optical difference spectra of the extract of Alviniconcha
gill are compared to those of purified Lucina Hb II, iso-
lated from the modified symbiont-harboring gill of the
clam Lucina pcciinata ( 10). The concentration of Alvini-
concha hemoglobin in the solution, about 19 ^.U(heme),
was estimated from the spectrum presented in Fig. 2A by
using molar extinction coefficients appropriate for the dif-
ference: carbon monoxide Lucina Hb II minus ferrous
Lucina Hb II. Optical spectra in the visible and soret re-
gions are expected to differ only slightly (about 10%)
among similar hemoglobins. The concentration in the tis-
sue was estimated ( 19) to be about 65 ^mol Alviniconcha
hemoglobin per kilogram wet weight gill.

Acknowledgments

We thank Dr. Robert R. Hessler for helpful discussion.
This work was supported in part by Research grants DCB
90-17722 (to JBW) and OCE-93- 17734 (to JLS) from The
National Science Foundation. JBW was a Research Career
Awardee 1-K6-733 of The U.S. Public Health Service,
National Heart Lung and Blood Institute.

Literature Cited

1 Messier, R. R., and \V. M. Smithey, Jr. 1983. The distribution
and community structure of megafauna at the Galapagos nft hy-

ALI'1MCO\CHA GILL HEMOGLOBIN

drothermal vents. Pp. 73>-770 in Hydrothermal Processes at Sea
Floor Spreading Centers. P. A. Rona el a/., eds. Plenum Press, New
York.

2. Stein, J. L., S. C. Cary, R. R. Hesslcr, S. Ohta, R. D. Vetter, J. J.
Childress. and H. Felbeck. 1988. Chemoautotrophic symbiosis in
a hydrothermal \ent gastropod. Bi< >/ Bull 174:373-378.

3. Okutani, T., and S. Ohta. 1988. A new gastropod mollusk asso-
ciated with hydrothermal vents in the Mariana back-arc basin.
Western Pacific. JPN .1 Malacol 47: 1 -9.

4. Wittenberg, J. B. 1985. Oxygen supply to intracellular bacteria
symbionts. In: Hydrolhermal rents of the Eastern Pacific: An Over-
view. M. L. Jones (ed.). Bull. Biol Sm: Washington 6: 301-310.

5. Childress. J. J., and C. R. Fisher. 1992. The biology of hydro-
thermal vent animals: physiology, biochemistry and autotrophic
symbioses. Oceanogr Mm Bin/ Annii Rev 30:337-441

6. Stein, J. L., M. Haygood. and H. Felbeck. 1990. Nucleotide se-
quence and expression of a deep-sea ribulose-1.5-bisphosphate car-
boxylase gene cloned from a Chemoautotrophic bacterial endosym-
biont. Proc. Null. Acad. Sci. USA 87: 8850-8854.

7. Stein, J.L.. and H. Felbeck. 199.3. Kinetic and physical properties
of a recombmanl RuBisCO from a chemoautrophic endosymbiont.
A/i)/ Mar Biol Biolcchnol 2: 280-290.

8. Wittenberg, J. B., and I). W. Kraus. 1991. Hemoglobins of eu-
karyote/prokaryote symbioses. Chapter 43. pp. 323-330. in Structure
and Function of Invertebrate Oxygen Carriers, S. N. Vinogradov
and O. H. Kapp. eds. Springer-Verlag. New York.

9. Appleby, C. 1984. Leghemoglobin and Rhtiohium respiration. Ann
Rev Plant Plmiol. 35: 443-478.

10. Kraus, D. W'., and J. B. Wittenberg. 1990. Hemoglobins of the
Lucina /'iv/wirtn/bacteria symbiosis. I. Molecular properties, kinetics
and equilibria of reactions with ligands. J Bin! C/iem 265: 16043-
16053.

11. Kraus, D. W'., J. B. Wittenberg, J. F. Lu. and J. Peisach.
1990. Hemoglobins of the Lucina peclinala/baclena symbiosis. II.
An electron paramagnetic resonance and optical spectral study of
the ferric proteins. / Biol. Client 265: 16054-16059.

12. Hockenhull-Johnson, J., M. S. Stern, P. Martin, C. Dass, D. M.
Desiderio, J. B. Wittenberg, S. N. Vinogradov, and D. A. W'alz.
1991. The amino acid sequence of hemoglobin II from the sym-
biont-harboring clam Lucina peclinala. J Protein Chcm 10: 609-
622.

13. Hockenhull-Johnson. J. D., M. S. Stern, J. B. Wittenberg. S. N.
Vinogradov, O. II. Kapp, and D. A. Walz. 1993. The ammo acid
sequence of hemoglobin III from the symbiont-harbonng clam. Lu-
cina pcclinala ./ Protein Chcm. 12: 261-277.

14 Rizzi, M., J. B. Wittenberg, A. Coda. M. Fasano, P. Ascenzi, and
M. Bolognesi. 1994. Structure of the sultide-reactive hemoglobin
from the clam Lucina peclina/a. Crystallographic analysis at 1.5 A
resolution. J Mol Biol 244: 86-89.

15. Wittenberg, J. B. 1991. Functions of cytoplasmic hemoglobins
and myohemerythrin. Chapter 3, pp. 59-85 in Advances in Com-
parative and Environmental Physiology. Vol. 13: Oxygen Carriers
in Blood and Tissues. C. P. Mangum, ed. Springer Verlag, Berlin.
Heidelberg, New York.

16. Arp, A. J., J. J. Childress, and R. D. Vetter. 1987. The sulfide-
binding protein in the blood of the vestimentiferan tube-worm. Rillia
pachypn/a. is the extracellular haemoglobin. J E.\p Biol 128: 139-
158.

17. Doeller, J. E., D. \V. Kraus, J. M. Colacino, and J. B. Wittenberg.
1988. Gill hemoglobin may deliver sulfide to bacterial symbionts
oiSolcmyu iv/Hw(Bivalvia, Mollusca). Biol Bull 175: 388-396.

18. Wittenberg, J. B., and B. A. Wittenberg. 1990. Mechanisms of
cytoplasmic hemoglobin and myoglobin function. Ann. Rev Biophys.
Biophys Chcm 19: 217-241.

19. Schuder, S., J. B. Wittenberg, B. llaseltine, and B. A. Wittenberg.
1979. Spectrophotometnc determination of myoglobin in cardiac
and skeletal muscle: separation from hemoglobin by subunit-ex-
change chromatography. Anal Bioc/iein. 92: 473-481.

20. Kraus, D. W ., J. E. Doeller, and J. B. Wittenberg. 1992. Hydrogen
sulfide reduction of symbiont cytochrome c\$ : ' n gills of Solemya
reuh (Mollusca). Biol. Bull. 182: 435-443.

Reference: Binl. Hull 188: 8-15. (February/March. 1995)

Inorganic Overgrowth of Aragonite on Molluscan
Nacre Examined by Atomic Force Microscopy

R. GILES 1 *, S. MANNE 1 , S. MANN 2 , D. E. MORSE 3 , G. D. STUCKY 4 ,

AND P. K. HANSMA 1 f

1 Department o/ Physics. University of California, Santa Barbara, California 93106,

2 School oj 'Chemistry. University of Bath. Claverton Down. Bath BA2 JAY. United Kingdom,

^Marine Biotechnology Center. Marine Science Institute. University of California,

San/a Barbara. California 93106. ami 4 Department of Chemistry.

University of California, Santa Barbara. California 93106

Abstract. The nacre (mother-of-pearl) that forms the ir-
ridescent inner layers of mollusc shells is a highly ordered
microlaminate composite of aragonite crystals and bio-
polymers with a strength and fracture resistance that far
exceed those of the mineral crystals themselves. The pro-
cesses governing the biofabrication of this material by the
secretory cells of the mantle are complex and only partially
understood. We have used the atomic force microscope
( AFM) to investigate the aqueous solution conditions un-
der which mineral growth can occur on the nacreous layer
of the shell of the bivalve mollusc Atrina sp. In situ im-
aging of the mature nacre surface exposed to a pH-con-
trolled environment of natural seawater with added car-
bonate ions reveals that inorganic overgrowth of aragonite
can occur within the ranges of pH and inorganic ion con-
centrations found in the molluscan extrapallial fluid from
which the mineral is produced during biological shell
growth. Thus, we posit that once nucleation has occurred,
nacreous tablets could grow inorganically in the extra-
pallial space; the role of proteins and other macromole-
cules may be limited to initiating growth or controlling
morphology through selective adsorption and spatial
constraint on the growing crystal.

Introduction

The mineral shells of a variety of molluscs are com-
posite biomaterials consisting of crystals of calcium car-
Received 7 March 1994; accepted 4 November 1994.
* Present address: Department of Physics. Simon Fraser University,
Burnaby. British Columbia, Canada V5A 1S6.

f Author to whom correspondence should be addressed.

bonate (CaCO,) intercalated with organic materials, pri-
marily proteins and glycoproteins (reviewed in Wilbur,
1972; Towe, 1972;Watabe, 1981;Weiner, 1986;Simkiss
and Wilbur, 1989; Lowenstam and Weiner, 1989). The
CaCOi occurs in two predominate crystal phases within
shells: calcite and aragonite. A shell may contain one phase
or the other, or both, depending on the animal species,
but commonly the stronger, denser aragonite forms an
inner structural layer, while calcite forms the outer layer.
The inner structural layer, called nacre or mother-of-pearl,
is a complex microlaminate composed of polygonal "tab-
lets" of aragonite that measure 5 to 15 ^m across, but
only 0.5 to 1 /urn in thickness, packed together with a thin
(40 nm) "mortar" of organic macromolecules. The or-
ganic component thus amounts to a small portion of the
total shell: less than 10% (Addadi and Weiner, 1992). It
nevertheless is responsible for the excellent strength and
resistance to crack propagation of the molluscan shell.
Crystallographically, the a and b axes lie in the plane of
the aragonitic tablets, with the e axis uniformly perpen-
dicular to the surface.

The nacreous layer of molluscan shell has been studied
extensively for several decades, principally with x-ray and
electron microscopic techniques. This work has been
largely successful in describing the microstructure of nacre
(see Lowenstam and Weiner, 1989; Weiner. 1986; Wa-
tabe, 1981; Towe. 1972; and Wise, 1970, for reviews).
Various calcium-binding, highly acidic, water-soluble
proteins have been isolated from the shell in various de-
velopmental stages (Cariolou and Morse, 1988). Water-
insoluble proteins from the shell have been characterized

GROWTH OF ARAGONITE ON NACRE

with x-ray diffraction, leading to the conclusion that they
resemble silk fibroin (Weiner and Traub, 1980). Both the
water-soluble and the water-insoluble proteins have been
proposed as multilaminar templates for the mineral tablets
(Nakahara t>/fl/., 1982).

The mechanism of growth of the nacreous layer is
complex and not well understood. It is known that both
organic and inorganic components are secreted by epi-
thelial cells in the mantle tissue into the extrapallial
space (the extracellular cavity between the mantle and
the shell, which is sealed from the surrounding envi-
ronment), bathing the growing shell in a mixture called
the extrapallial fluid. Although the inorganic compo-
nents of the extrapallial fluid are obviously necessary
for mineral growth, it is not known whether they are
sufficient: i.e., the role of the organic components is not
well known. Kitano and Hood (1962) showed that ara-
gonite is the most favorable phase of CaCO 3 to nucleate
in seawater supersaturated with respect to that mineral:
the presence of Mg 2+ in solution apparently acts to se-
lect aragonite over calcite. Others have measured nu-
cleation rates of aragonite crystals in seawater and ar-
tificial extrapallial fluid (Pytkowicz, 1965; Wilbur and
Bernhardt, 1984). We extended these studies by making
two experimental modifications relevant to nacreous
growth. First, we studied crystallization directly on a
nacreous surface (rather than unseeded nucleation in
solution). (Recently, Sabbides and Koutsoukos [1993]
also investigated seeded growth of aragonite on a variety
of substrates in seawater.) Second, we controlled both
pH and total carbonate concentration simultaneously,
and we compare growth conditions to those values re-
ported for extrapallial fluid.

We have used the atomic force microscope (AFM)
(Binnig el al. 1986) to examine the conditions for inor-
ganic growth of the nacre tablets. (For reviews of the AFM,
see Rugar and Hansma, 1990; Sarid, 1991; Hoh and
Hansma, 1992). The AFM (also known as the scanning
force microscope) is a member of the family of scanning
probe microscopes; these instruments form images by
raster scanning a tiny probe over the surface of the sample
while mapping some local interaction, such as electron
tunneling or near-field optical effects, as a function of
position. The AFM probe consists of a flexible cantilever,
~ 100 jum long, with a sharp tip attached at the end; the
probe measures (through the elastic response of the can-
tilever) the interaction forces between the tip and the
sample. The probe can thus map surface topography by
scanning in gentle contact with the sample; the displace-
ment of the cantilever (as the tip slides over surface fea-
tures) is detected from the motion of a laser beam reflected
from the back of the cantilever onto a position-sensing
photodiode. The AFM can operate in solution and hence

allows /// situ imaging of samples from the micrometer
to the nanometer scale. It recently has been applied to
biomineralized composites such as diatom shells (Linder
et al., 1992), bone (Tao and Lindsay, 1992). teeth (Kasas
el til., 1993), pressed powders of clam shells and sea urchin
shells (Friedbacher et al.. 1991), and molluscan nacre
(Manne et ul.. 1994). It has imaged in situ dynamic pro-
cesses on relevant systems such as calcite (Gratz et al..
1993; Hillner el al.. 1992), fluorite (Hillner et al.. 1993),
and hydroxyapatite (Kasas et al.. 1993), as well as calcite
growth modification in the presence of polyamino acids
(Wierzbicki et al.. 1993) and inorganic poisons (Gratz
and Hillner, 1993; Dove and Hochella, 1993). By ex-
amining the exposed aragonite surface of mature nacreous
tablets for signs of growth under various solutions, we
bracketed and thus defined the conditions under which
aragonite growth can occur. These conditions are biolog-
ically relevant: the solutions used approximate the inor-
ganic components of the extrapallial fluid in which new
nacre is formed.

Materials and Methods

Samples of nacre from the bivalve Atrina sp. were
kindly provided by Prof. S. Weiner at the Weizmann In-
stitute of Science in Israel. For imaging, small pieces (ap-
prox. 1 X 0.5 X 0.1 mm) of mature nacre were prepared
by mechanically cleaving a shell fragment with a razor
blade and then fracturing the resulting chip down to the
desired dimensions.

The solutions tested were based on natural seawater
collected locally from the Pacific Ocean along the Santa
Barbara coast. The water was coarsely filtered, irradiated
with ultraviolet light, passed through a 0.2-nm filter, and
stored at 2-4C in a sterilized, lightproof container until
just before use. To the seawater various amounts of
NaHCO 3 were added, and the pH was adjusted to the
desired value by addition of HC1 or NaOH.

Cation concentrations for the filtered seawater were
measured by atomic absorption spectroscopy; total car-
bonate ion concentration was determined by titration
with HC1. Table I lists concentrations for the measured
ions (at about 20C). These agree well with previously
published concentrations for seawater (Crenshaw, 1972;
Smith, 1974; Wada and Fujinuki, 1976). In addition,
the cation concentrations are all within about 10% of
the published values for the extrapallial fluid of bivalves.
In particular, the concentration of Ca : + ion we deter-
mined (Table I) is about the same as found in extra-
pallial fluid (Crenshaw, 1972; Wada and Fujinuki,
1976). The major difference between seawater and the
inorganic composition of extrapallial fluid is the higher
concentration of carbonate ion, which is approximately

10

R. GILES ET AL

Table I

Concentrations o\ f inorganic ions in the filtered natural seawater

Ion

Concentration

(mA/) standard dev.

Na +

463

-i- 2

K +

10.9

0.9

Mg 2+
Ca 2+

63.1
10.3

1.4
0.1

Sr +

0.085

0.007

HCO 3 ~ + CO 3 2 ~ or total carbonate

2.3

0.1

2-fold higher in extrapallial fluid. Therefore, only car-
bonate ion was added to natural seawater to create the
growth solutions. A comparison of the values of ion
concentrations determined for seawater and extrapallial
fluid is included in Figure 5.

The samples of nacre were glued to a stainless steel disk
with epikot resin and placed in the fluid cell of a com-
mercial AFM (Nanoscope III from Digital Instruments,
Santa Barbara, CA 93 103). Samples were always oriented
with the proximal side (the side facing the animal during
life) exposed for imaging. After an appropriate area had
been selected by imaging in air, growth solution was added
to the fluid cell. The sample was examined for signs of
growth for 15-20 min under a steady gravity flow of this
solution. Typical flow rates were 5 n\/s, which corresponds
to a replacement of the fluid cell volume every few sec-
onds. The tip was then withdrawn and flow stopped for
1.5 h to allow more time for growth to occur. Only the
exit line was blocked so that the cell remained in chemical
contact with at least 40 ml of the solution in a reservoir
above the cell. The sample was then examined again under
flow for signs of growth. The procedure was repeated with
an alternate growth solution; either the pH was raised
while maintaining the same carbonate concentration, or
vice versa. After growth had occurred with a given solu-
tion, only one or two more solutions could be tried with
a given sample before the surface became so rough that
further growth could not be analyzed. All growth exper-
iments were conducted at about 20C.

Results

Figure 1 shows an AFM image of the nacreous surface.
Most of one polygonal tablet and portions of two others
can be seen. The characteristic features of bivalve nacre
(Manne el ai, 1994), such as concavity of the proximal
tablet surfaces, a depression in the center of each of tablet,
and elongate rings surrounding the depressions, are visible.
Growth assessment was somewhat difficult at this scale;
usually the characterization was made on the basis of im-

ages only 3-j/m square, such as the area outlined in the
figure.

Figure 2 illustrates the changes in surface roughness
considered indicative of growth. It shows the same area
(the inset in Fig. 1 ) before and after incubation for 1 .5 h
in seawater with a total carbonate concentration of 4.3
mM. In the sample imaged in Figure 2a, the pH was 7.9,
at which point the surface had already changed somewhat
from its initial appearance under plain seawater. The
sample then was incubated for 1.5 h in a solution with
the same total carbonate concentration, but at pH 8.1,
and then shifted once more to pH 8.3. In Figure 2b the
surface is shown 6 min after raising the pH to 8.3; further
growth had occurred. As the images are shaded propor-
tionally to height, the new growth can be seen by com-
paring the relative brightness of corresponding features
in the two images; a few prominent pairs are indicated
by arrows.

It is important to recognize that tip convolution
(Grutter et ai , 1992; Allen et al. 1992) dominates the
growth image. This is indicated by the similarity of
shape and orientation among the bumps on the surface.

Figure 1. Atomic force microscope (AFM) image of the nacre of

Atnnu sp. The image is 1 1 /am 2 . Shading is proportional to elevation,
500 nm from dark to light, with the brightest regions being the highest.
The high ridge coincides with the boundary between nacreous tablets.
Most of a tablet is visible in the center and left of the image, along with
portions of two others. Note the general concavity, central depression
and the elongate rings characteristic of bivalve nacre. The black square
marks the area shown in Figure 2. centered on the intersection of the
three tablet boundaries.

GROWTH OF ARAGONITE ON NACRE

Figure 2. AFM images of the area indicated in Figure 1 before and
after mineral overgrowth. Both images are 2.5 ^m square; the height is
200 nm trom dark to white. Comparison of corresponding features be-
tween (a) (before overgrowth) and (b) (after overgrowth) demonstrates
the characteristic increase in the height of the surface asperities indicative
of mineral growth on the nacreous surface. Details in text. Arrows indicate
corresponding prominent asperities in the two images. These images
demonstrate the topographic changes used to characterize specific solution
conditions as productive of mineral overgrowth (Fig. 5).

Figure 3. Illustration of the convolution between the AFM tip and
a sharp asperity on the sample surface as it is scanned. The tip is shown
passing from right to left over the asperity. The measured topography
(dashed line) is a "convolution" of the tip shape and the asperity; it is
largely an image of the tip itself, except at the very top of the asperity.
Note that while the true width of the asperity is completely obscured,
the height measurement is accurate.

Tip convolution is a common AFM imaging artifact;
Figure 3 illustrates the mechanism responsible for this
effect. The AFM measures topography by scanning a
tip over the surface and measuring the vertical deflection
of that tip as it slides over surface features. However,
as the tip slides over an asperity with a higher aspect
ratio than the tip itself, the deflection of the tip traces
the tip's profile, rather than that of the asperity. In Fig-
ure 3, the dashed line indicates the path that will be
traced by the tip as it passes over an asperity. Although
as a consequence of tip convolution the lateral dimen-
sions of a sharp asperity are not resolved, the overall
height of that asperity is correctly measured, as is to-
pography on the flat top (Griitter ct ai. 1992; Allen et
a/.. 1992); thus, reliance on changes in local height de-
tected by the AFM is justified as a measure of mineral
growth. Figure 2b is consistent with the observation of
surface asperities lengthening normal to the imaging
plane, as expected for aragonite needles growing in their
normal crystallographic habit along the c 1 axis.

Although our classification of specific solution condi-
tions into growth/no-growth categories was based on
qualitative comparison of surface topography between
"before" and "after" images (of which the images in Fig.
2 are an example), this method is further supported by
quantitative measurements of surface roughness. Mea-
surements of the root-mean-square deviation of height
values from their collective mean was made for corre-
sponding areas of the images in Figure 2, as well as for

12

R. GILES ET AL

Figure -4. Unfiltered AFM image of the aragonite lattice, viewed along
the c axis to show the (001) crystallographic plane, 30 nm square. The
image was obtained on an area where overgrowth had been observed.
The inset shows a Fourier transform of the image. All peaks in the trans-
form are consistent with the expected reciprocal lattice of aragonite, and
the fundamental translation sectors (circled) that define the unit cell
agree with the aragonite a and /' axis spacing (0.495 nm and 0.796 nm,
respectively) to within 5%.

one intermediate between the two (at pH 8. 1 ). Combining
such measurements from two different areas, both away
from the high tablet boundary in the center of the image,
yielded a monotonically increasing surface roughness (of
2.9 nm for Fig. 2a, 3.9 nm for Fig. 2b, and 3.4 nm for the
intermediate image) in agreement with the qualitative as-
sessment of growth.

The observed changes in topography of the nacreous
tablets we imaged were caused by crystal growth on the
tablet surfaces, rather than by ilc nnvo nucleation and
precipitation from the growth solution. Parallel growth
experiments performed on nacreous particles embedded
in epoxy showed changes in surface roughness (growth of
crystal asperities) only on the nacre, and not on the sur-
rounding epoxy. These results are described in detail else-
where (Giles ct al.. 1 993).

Atomic lattice resolution could sometimes be obtained
atop the asperities, indicating that they are terminated by
small (<50 nm) flat areas. However, a wide variety of lat-
tices were observed, perhaps indicating the presence of
high-index planes on the sidewalls of the asperities. Oc-
casionally lattices (Fig. 4) did show periodicities corre-
sponding to the expected unit cell of the (00 1) plane of
aragonite (i.e.. viewed along the c axis).

Figure 5 presents a summary of the growth results. The
dashed lines separate the values of total carbonate con-
centration and pH that define the growth and no-growth
conditions. The rectangular bands designate the ranges of
these values previously reported for molluscan extrapallial
fluid and natural seawater. Note that growth never occurs
at the carbonate concentrations of seawater, but that the
boundary between growth and no-growth cuts across the
extrapallial fluid range, suggesting the potential for dy-
namic control of shell formation by changes in extrapallial
fluid composition. This is consistent with previous data
on seasonal variation in the acidity of the extrapallial fluid,
in which high pH was correlated with a high rate of shell
growth and low pH with slow growth or shell dissolution
(Wada, 1 96 1).

Supersaturation values of specific ions with respect to
aragonite were estimated with the ION PRODUCT com-
puter program (Shellis, 1988) for each of the solutions
tested. Saturation fractions ranged from 0.3 to 3.6, but in
general (with two exceptions) overgrowth occurred at sat-
uration values greater than 1, and no overgrowth occurred
at saturation values less than 1 .

Discussion

The nacre of molluscan shell is a highly organized mi-
crolaminate composite of proteins, glycoproteins, and
calcium carbonate crystals in the aragonite phase; the

12 -

extrapalli

al range

\ I-A-"

8,2

pH of growth solution

Figure 5. Aragonite growth/no-growth results as a function of pH
and total carbonate concentration ([HCO 3 ~] + [CO 3 : ~]). Open circles
indicate conditions al which no growth was observed; filled triangles
indicate conditions at which growth occurred; no symbol (two horizontal
lines) indicates indeterminate results. Error bars account for pH drift
over the course of the 1. 5 h incubation. Dashed curves approximately
separate the regions of growth and no growth. The labeled bands indicate
the ranges of published values for molluscan extrapallial fluid and natural
seawater. Note that the range for extrapallial fluid, in which shell growth
occurs biologically, spans the boundary between the growth and no-
growth conditions.

GROWTH OF ARAGONITE ON NACRE

13

biological mechanisms that control its formation are
complex and only partially understood. Previous research
(Wilbur, 1972; Weiner, 1986; Lowenstam and Weiner.
1989; Simkiss and Wilbur. 1989; Weiner and Addadi,
1991; Addadi and Weiner. 1992) suggests that shell min-
eralization commences with isolation of the site of min-
eralization from the external seawater environment by an
insoluble matrix of macromolecules and the mantle tissue
from which these molecules are secreted as an extension
of the growing shell edge. The shell and the mantle epi-
thelium enclose the extrapallial fluid, which is ionically
enriched and pH-controlled by enzymatic pumping across
the cell membranes of the mantle epithelium (Weiner and
Traub. 1 984; Weiner and Addadi, 1 99 1 ). It has been sug-
gested that insoluble matrix molecules play essential roles
both in the control of crystal nucleation and by establish-
ing compartments that limit the spaces in which the crys-
tals grow (e.g., Wilbur, 1972; Weiner, 1984; Nakahara,
1989). New aragonite crystals on the growth surface seem
to be nucleated in pores of the organic matrix (Nakahara
el ai. 1982), forming as small crystallites that grow to
become the next nacreous layer (Wada, 1972). Crystal
growth apparently also is controlled by several polyanionic
proteins found associated with (and in some cases, oc-
cluded within) the mineral crystals; these proteins also
have been suggested to act both in nucleation and in se-
lective inhibition of crystal growth (Addadi and Weiner,
1985; Sikes and Wheeler, 1988; Addadi et ai, 1990; Wei-
ner and Addadi, 1991; Morse et ai. 1993; Albeck et ai.
1993; Herman el ai. 1993).

The experiments reported here demonstrate that an ex-
posed surface of mature nacre can continue to grow by
purely inorganic means when the ion concentrations
present in the extrapallial fluid favor the aragonite phase
(primarily due to the presence of magnesium) and are
supersaturated with respect to that form. This indicates
that once the nucleation of the mineral phase has begun,
the crystal can continue to grow without direct biological
control. Therefore the requirements for nucleation, pos-
sibly including a template of acidic proteins (Aizenberg
et ai., 1994; Morse et ai. 1993: Weiner et ai. 1983) or
mineral bridges between the aragonite tablets (Manne et
ai, 1994), can be independent of growth. Because the
supersaturation levels required for overgrowth can be quite
low, it is plausible that cells of the mantle could regulate
growth simply by adjusting carbonate concentration and
pH in the extrapallial fluid.

Although evidence for growth was decisive in most of
the observed cases, quantification of the rates of mineral
growth proved difficult since growth was not observed to
occur uniformly over the experimental period. In some
instances, growth occurred in the first few minutes of im-
aging after introducing a solution; in others, growth was

not apparent until after the 1.5-h incubation. Unlike the
case of the cleavage plane of geological calcite (Hillner et
ai, 1993), on which growth occurs by quantifiable accre-
tion of widely spaced steps, the aragonite tablet is much
rougher and has a far greater step density. This high con-
centration of reactive sites may set up complex concen-
tration gradients that affect the reaction rate in unpre-
dictable ways.

It is interesting that the aragonitic overgrowth of na-
cre occurred in the form of needlelike extensions of the
(001 ) surface, rather than as the layered growth char-
acteristic of nacreous tablets in molluscan shells. The
former is the common growth morphology of abiogenic
aragonite and often results in extensive lateral inter-
growth, producing fanlike aggregates of misaligned
needles. Development of the highly coherent tablet
morphology found in biogenic nacre, characterized by
a high degree of orientation of the crystallographic c
axes of the aragonite tablets, thus would require
suppression of the tendency for disorder observed in
the overgrowth process seen in our experiments. One
possibility is that the highly anionic, soluble proteins
(and possibly other macromolecules) found intimately
associated with the aragonite crystals in molluscan nacre
may prevent long-range incoherent intergrowth of the
needles by specific interactions at the growing crystal
surfaces (Addadi and Weiner, 1985). This would pro-
duce the coherent (001) surface and single crystal com-
position observed in mature nacre. In addition, the in-
soluble polymers of the matrix may help determine the
final crystal form by creating a preformed microstruc-
ture that delimits the space in which the tablets grow
(Wada, 1972: Nakahara, 1989). As neither the soluble
acidic proteins nor empty sheaths of the insoluble ma-
trix proteins were present in our experiments, growth
along the e axis was persistently the fastest, as in abio-
genic aragonite.

Acknowledgments

We thank Robert Petty (of the Marine Science In-
stitute Analytical Laboratory. University of California,
Santa Barbara) for performing the atomic absorption
measurements; Monika Fritz, Angela Belcher, Charlotte
Zaremba, and Deron Walters for useful discussions; and
J. M. Didymus (School of Chemistry, University of
Bath, UK) for kindly performing the calculations of
supersaturation fractions. This work was supported by
grants from the Materials Research Laboratory program
of the National Science Foundation (DMR-9 123048);
the Molecular and Cellular Biosciences and Materials
Research Divisions of the National Science Foundation
(MCB-9202775 to G.D.S., D.E.M., and P.K.H.); the

14

R. GILES ET AL

Office of Naval Research (NOOO 14-93- 1-0584 to
D.E.M., G.D.S., and P.K.H.); and a fellowship from
AT&T (S. Mamie).

Literature Cited

Addadi, L., and S. Weiner. 1985. Interactions between acidic proteins

and crystals: stereochemical requirements in hiomincralization. Proe

Nail. Acad. Sci. USA 82: 41 10-41 14.
Addadi, I,., and S. \\ einer. 1992. Control and design principles in

biological mineralization. Angew Chcm Inl /:'</ lingl 31: 153-

169.
Addadi, L., A. Bcrman. J. Moradian-Oldak, and S. Weiner.

1990. Tuning of crystal nucleation and growth by proteins: molec-
ular interactions at solid-liquid interfaces in biomineralization. Croat.
Chcm Ada 63: 539-544.

Aizenberg, J., S. Albeck, S. \\einer, and I.. Addadi. 1994. Crystal-
protein interactions studied by overgrowth of calcite on biogenic
skeletal elements. ./ ( 'ry.s/al Growth 1-12: 156-164.

Albeck, S., J. Aizenberg, I.. Addadi, and S. Weiner. 1993. Interactions
of various skeletal mtracrystalline components with calcite crystals.
J Am. Chcm. Sue 115: I 1691-1 1697.

Allen, M. J., N. V. Hud, M. Balooch, R. .J. Tench, W. J. Siekhaus,
and R. Balhorn. 1992. Tip-radius artifacts in AFM images of
protamine-complexed DNA fibers. Ultramicroscopy 42-44: 1095.

Herman, A., J. Hanson, I,. Leiserowitz, I. F. koet/le, S. Weiner, and
L. Addadi. 1993. Biological control of crystal texture: a widespread
strategy for adapting crystal properties to function. Science 259: 776-
777.

Binnig. G., C. F. Quate, and Ch. Gerber. 1986. Atomic Force Micro-
scope. Phys Rev Lett. 56: 930-933.

Cariolou. M. A., and D. E. Morse. 1988. Purification and character-
ization of calcium-binding conchiolin shell peptides from the mollusc.
Haliolis nilescens. as a function of development. ./ Comp Phvsiol
B 157: 717-729.

Crenshaw, M. A. 1972. The inorganic composition of molluscan ex-
trapallial fluid. Biol Bull 143:506-512.

Dove, P. M., and M. F. Hochella Jr. 1993. Calcite precipitation mech-
anisms and inhibition by orthophosphate: in situ observations by
scanning force microscopy. Gencliini Cosmoclnm Ada 57: 705-
714.

Friedbacher, G.. P. K. Hansma, F. Ramli, and G. D. Stuck).

1991. Imaging powders with the atomic force microscope: from
biommerals to commercial materials. Science 253: 1261-1263.

Giles, R., S. Manne, C. M. /.aremba, A. Butcher, S. Mann, D. E. Morse,

G. D. Stueky, and P. K. Hansma. 1993. Imaging single nacreous

tablets with the atomic force microscope. Materials Re-- Sue Proc

332: (Boston, Fall 1993. in press).
Gratz. A. J., and P. F. I lillner. 1993. Poisoning of calcile growth viewed

in the AFM. J. Crvstat Growth 129: 789-793.
Gratz, A. J., P. F. I lillner, and P. K. Hansma. 1993. Step dynamics

and spiral growth on calcite. Geoclinn Cosmoclum Ida 57: 491-

495.
Grutter, P., \V. Zinunermann-Edling, and D. Brodbeek. 1992. Tip ar-

titacts ol microfabricated force sensors for atomic force microscopy.

.!/'/>/ Phys l.cll 60: 2741-2743.
I lillner, P. F., A. .1. Gratz, S. Manne, and P. K. I laiisma. 1992. Atomic

scale imaging of calcite growth and dissolution in real time. Geology

20: 359-362.
I lillner, P. E., S. Manne, A. J. Gralz, and P. k. I lansma. 1993. Atomic

force microscope: a new tool for imaging crystal growth processes.

l-'araday l>is,n^ ( hem. Sue. 95: 191-197.

Hob, .). II., and P. K. Hansma. 1992. Atomic Force Microscopy for
high-resolution imaging in cell biology. Trends Cell Biol 2: 208-
213.

Kasas, S., A. Berdal. and M. R. Celio. 1993. Tooth structure studied
using the atomic force microscope. Scanning Probe Microscopies 11.
C. C. Williams, ed.. Proc. SHE 1855: 17-25.

kitano, Y., and D. \V. Hood. 1962. Calcium carbonate crystal forms
formed from seawater by inorganic processes. J. Oceanogr. Soc. JPN
18: 35-39.

Under, A., .1. Colehero, II.-.I. Apell, O. Marti, and J. Mlynek.
1992. Scanning force microscopy of diatom shells. Ultramicroscopy
42-44: 329-332.

Lowenstam, H. A., and S. Weiner. 1989. On Bioi)iiiii-rali:atioii. Oxford
University Press, New York.

Manne, S.. C. M. Zaremha. R. Giles, L. Huggins, D. A. Walters, A.
Belcher, D. E. Morse, G. D. Slucky, J. M. Didymus, S. Mann,
and P. k. Hansma. 1994. Atomic force microscopy of the na-
creous layer in mollusc shells. Proc. R. Soc. l.onil. B 256: I 7-23.

Morse, D. E., M. A. Cariolou, G. D. Stueky, C. M. Zaremba, and
P. k. Hansma. 1993. Genetic coding in biomineralization of
microlaminate composites. In Biomolecular Materials, C. Viney,
S. T. Case, and J. H. Waite. eds. Materials Res Soc Proc 292:
59-67.

Nakahara, II. 1989. Nacre formation in bivalve and gastropod mol-
luscs. Pp. 343-350 in Mechanisms and Phytogeny of Mineralization
in Biological Systems. S. Suga and H. Nakahara, eds. Springer- Verlag.
New York.

Nakahara, H., G. Bevelander, and M. kakei. 1982. Electron micro-
scopic and amino acid studies on the outer and inner shell layers of
Iliiliolis nilescens I cnns 41: 33-46.

Pytkoicz, R. M. 1965. Rates of inorganic calcium carbonate nucle-
ation. ./ Geol 73: 196-199.

Rugar, 1).. and P. k. I lansma. 199(1. Atomic Force Microscopy. Physics
Today 43: 23-30 (October).

Sabbides, T. G., and P. G. koutsoukos. 1993. The crystallization of
calcium carbonate in artificial seawater; role of the substrate. J Crystal
Growth 133: 13-22.

Sarid, I). 1991. Scanning f-'orce Microscopy: \\'nh Applications to
lileiinc. Magnetic, and Atomic l-'orccs Oxford University Press, New
York.

Sikev C. S., and A. P. Wheeler. 1988. Regulators of biomineralization.
Chemteeh 1988: 620-626.

Simkiss, k., and k. M. Wilbur. 1989. Biomineralization. Academic
Press, New York.

Sbellis, R. P. 1988. A microcomputer program to evaluate the satu-
ration of complex solutions with respect to biominerals. Cambios 4:
373-379.

Smith, F. G. W., ed. 1974. CRC Ham/hook ol Marine Science. Vol.
1, p. 4. CRC Press, Cleveland. Ohio.

Tao, N. J., and S. M. Lindsay. 1992. Measuring the microelastic prop-
erties of biological material. Biophys J 63: 1 165-1 169.

I oe, k. M. 1972. Invertebrate shell structure and the organic matrix
concept. Biomineralisation 4: 1-14.

Wada, k. 1961. Crystal growth of molluscan shells. Bull Nail Pearl
Res Lah .fPN 7: 703-828.

Wada, k. 1972. Nucleation and growth of aragonite crystals in the
nacre of some bivalve molluscs. Biomineralisation 6: 141-159.

Wada, k., and I. Fujinuki. 1976. Biomineralization in bivalve
molluscs with emphasis on the chemical composition of the e.x-
trapallial fluid. Pp. 175-190 in The Mechanisms of Mineralization
in llic Invertehrales anil Plants, N. Walabe and K.. M. Wilbur,
eds. University of South Carolina Press, Columbia, South Caro-
lina.

GROWTH OF ARAGONITE ON NACRE

15

Watabe, N. 1981. Crystal growth of calcium carbonate in the inver-
tebrates. Prog. Crystal Growth Characl 4: 99-147.

Weiner, S. 1984. Organization of organic matrix components in min-
eralized tissues. Am. Zoo/. 24: 945-952.

Weiner, S. 1986. Organization of extracellularly mineralized tissues:
a comparative study of biological crystal growth. CRC Cril. Rev
Biochem 20: 365-408.

Weiner, S., and L. Addadi. 1991. Acidic macromolecules of mineralized
tissues: the controllers of crystal formation. Trends Biochem. Sci 16:
252-256.

Weiner, S., and W. Traub. 1980. X-ray diffraction study of the in-
soluble organic matrix of mollusk shells. FEBS Lett. Ill: 311-
316.

Weiner, S., and W. Traub. 1984. Macromolecules in mollusc shells
and their functions in biomineralization. Phil. Trans. R. Sot: Lond.
B 304: 425-434.

Weiner, S., Y. Talmon, and W. Traub. 1983. Electron diffraction of
mollusc shell organic matrices and their relationship to the mineral
phase. Int. J Bio/. Macmmol 5: 325-328.

Wierzbicki, A., C. S. Sikes, J. D. Madura, and B. Drake. 1993. Atomic
force microscopy and molecular modeling of protein and peptide
binding to calcite. Calcif. Tiss. Inl 54: 133-141.

Wilbur, K. M. 1972. Shell formation in mollusks. Pp. 103-145 in
Chemical Zoology, \'ol. 6: Mollusca, M. Florkin and B. T. Scheer,
eds. Academic Press, New York.

Wilbur, K. M., and A. M. Bernhardt. 1984. Effects of amino acids,
magnesium, and molluscan extrapallial fluid on crystallization of
calcium carbonate: in vitro experiments. Biol Bull. 166: 251-259.

Wise, S. W. 1970. Microarchitecture and mode of formation of nacre
(mother-of-pearl) in pelecypods, gastropods, and cephalopods. Eclo-
gue Geol. Helv. 63: 775-797.

Reference: S/W Bull. 188: 16-22. (February/March, 1995)

Direct Development in the Ascidian Molgula
retortiformis (Verrill, 1871)

WILLIAM R. BATES*

Department of Biology. Carlelon University. Ottawa. Ontario, Canada K1S ?B6.
ami IlnniMiiiin Marine Science Centre. St. Andrews. New Brunswick. Canada EOG 2X0

Abstract. The cellular features of the ascidian Molgula
retortiformis (Verrill, 1871), a direct developing species,
were investigated with the aid of transmission electron
microscopy, histochemistry, and immunocytochemistry.
Developmental comparisons between direct and indirect
developing ascidians will further our understanding of how
developmental processes evolve. M. relortiformis eggs are
surrounded by a follicular envelope comprising a layer of
outer follicle cells attached to an acellular chorion. The
cytoplasm of M. retortiformis eggs contains large quan-
tities of yolk and glycogen. Immediately after hatching,
at day 2.5 of development, the cells constituting a juvenile
exhibited similar ultrastructural features, except that the
larger, deeper cells contained more yolk and glycogen than
the epidermal cells. Differentiated muscle cells were absent
in newly hatched M. retortiformis juveniles, and acetyl-
cholinesterase (AChE) activity was not detected. Immu-
nocytochemistry experiments using a vertebrate inter-
mediate filament antibody (NN18) support the idea that
the failure of newly hatched M. retortiformis juveniles to
develop muscle cells may be due to the absence of a factor
localized in the egg myoplasm. This paper concludes with
a discussion of the "substrate hypothesis" and the evo-
lution of ascidian direct development.

Introduction

Most ascidians produce eggs that develop into chordate
larvae that swim for a brief time and subsequently me-
tamorphose into adulls. During metamorphosis the
chordate features of a larva are selectively destroyed, and
the adult morphology develops (Grave, 1935; Cloney,
1978, 1982). Ascidians that produce swimming larvae

Received 18 January 1994; accepted 4 November 1994.
* Present address: Bamfield Marine Station, Bamfield, British Colum-
bia. Canada VOR 1BO

are termed indirect-developing species. In striking con-
trast to indirect-developing species, about a dozen species
produce fertilized eggs that develop directly into juve-
niles, bypassing the development of a swimming larva
(de Lacaze-Duthiers, 1874; Berrill, 1931; Jeffery and
Swalla, 1990; Bates and Mallett, 199 la). Here I report
on the cellular features of a direct-developing species,
Molgula retortiformis.

N. J. Berrill (1931) wrote that M. retortiformis has a
direct mode of development; however, he provided only
one line drawing of a juvenile. His drawing shows a M.
retortiformis juvenile without a tail, lacking a sensory ves-
icle, having partially extended epidermal ampullae, and
containing a cluster of large, opaque cells, which he terms
"tail phagocytes," in the posterior region. Aside from these
general features, no information was given on the cellular
features of eggs, embryos, and juveniles in this species.
Although Berrill was not concerned primarily about the
cellular features of direct-developing ascidians. he was
among the first to recognize that comparisons between
indirect and direct modes of ascidian development can
provide valuable insights about chordate evolution. In his
1931 paper, Berrill suggested that direct development in
ascidians evolved by the elimination of the larval sensory
vesicle and larval tail structures. He argued that the de-
velopment of a swimming tadpole larva capable of se-
lecting a habitat would be unnecessary if the adult lived
in a uniform habitat. This idea, which is termed the "sub-
strate hypothesis," is based primarily on studies of Mol-
gula occitlla. a direct-developing species that inhabits the
sand flats of Brittany. I reexamine Ben-ill's substrate hy-
pothesis in the present study of M. retortiformis.

Interest in ascidian direct development was renewed
when Whittaker (1979) reported that Molgula arrenata
embryos, embryos exhibiting direct development, can ex-
press acetylcholinesterase despite the lack of tail devel-
opment. AChE activity in a species with direct develop-

16

DIRECT DEVELOPMENT IN MOLGULA

17

ment suggested to Whittaker that AChE activity is a ves-
tigial trait that has not been eliminated from an ancestral
program responsible for larval muscle cell development.
The present study tested the possibility that newly hatched
M. retortiformis juveniles can express AChE activity. Just-
hatched tadpoles from three indirect-developing species,
Halocynthia pyriformis, Boltenia echinata, and Ciona in-
testinalis, were also tested for AChE activity.

Since the publication of Whittaker's exciting results in
1979, a number of studies on ascidian direct development
have been reported, including those by Young el al.
(1988), Jeffery and Swalla ( 1990, 1991, 1992), Bates and
Mallett (1991a,b), Bates (1991), and others. In 1988,
Young e! al. were the first to report that Molgula pacifica
is a direct developer. Many of the cellular features of M.
pacifica development have been described (Bates and
Mallett, 1 99 la,b; Bates, 1991. 1993). The postfertilization
movements of the egg cytoplasm, termed ooplasmic seg-
regation, and early cleavage patterns in M. pacifica were
similar to those in eggs and embryos having indirect de-
velopment. Although most features of early development
were similar to those in indirect developers, ampulla de-
velopment in M. pacifica juveniles was triggered before
hatching (Bates and Mallett, 199 la; Bates, 1993, 1994)
instead of after larval settlement (Cloney, 1978; Grosberg,
1981; Grosberg and Quinn, 1986).

The elimination of larval muscle cell development in
direct-developing ascidians was recently studied in Mol-
gula oculata (an indirect-developer) and Molgula occulta
(a direct-developer), the same species studied by Berrill
(1931). Results of these studies suggested that the lack of
larval muscle cell development in M. occulta may be due
to the absence of a protein that is recognized by a verte-
brate intermediate filament antibody (NN18) localized in
the myoplasm of M. oculata eggs (Swalla ct al., 1991). In
the present study, I used M. retortiformis and an indirect-
developing species, Boltenia villosa. to test the correlation
between the antigen recognized by NN18 and AChE
activity.

In summary, the threefold aim of the present study was
( 1 ) to examine the general cellular features of A/, retor-
tiformis eggs, embryos, and juveniles; (2) to determine if
there is a correlation between AChE activity and a factor
localized in the egg myoplasm that reacts with NN18 in
M. retortiformis juveniles and B. villosa tadpoles; and (3)
to test Benin's substrate hypothesis by examining the
habitats of A/, retortiformis adults.

Materials and Methods

Collection of adults, eggs and sperm, and embryo
cultures

Molgula retortiformis, Halocynthia pyriformis, Boltenia
echinata, and Ciona intestinalis adults were collected in
the Bay of Fundy near Huntsman Marine Station, St.

Andrews, New Brunswick, Canada. Collections were
made with a dredge at depths ranging from 50 to 100 feet.
Boltenia villosa adults were purchased from Westwind
Sealab Supplies, Victoria, British Columbia. Adults were
maintained in aquaria containing flowing seawater under
conditions of constant light to prevent spawning. Testes
and ovaries were removed from adults and placed in a
Syracuse dish containing seawater; eggs and sperm were
collected by using forceps to macerate the gonads. Van
Name (1945) described M. retortiformis (Verrill, 1871).
The testis on the left side of an adult was situated alongside
the inner side of the lower branch of the intestinal loop
and the left ovary was situated outside the intestinal loop
along the upper branch of the intestinal loop. On the right
side, the testis was situated ventral to the kidney and the
ovary was situated along the dorsal border of the kidney.
Fertilized eggs were obtained by mixing together eggs and
sperm from two or more individuals in a Syracuse dish
containing Millipore-filtered seawater. Eggs were insem-
inated for 10 min, washed with large volumes of seawater,
and cultured at 1 1 C. Embryos were viewed at frequent
intervals with an Olympus SZ stereomicroscope.

Transmission electron microscopy

Embryos and juveniles were prepared for light micros-
copy and transmission electron microscopy as previously
described by Bates and Mallett (1991a). Specimens were
fixed in 2% glutaraldehyde in 0.1 A/ sodium phosphate
buffer, pH 7.4, for 30 min. After a wash in the same buffer,
the specimens were immersed in 1% osmium tetroxide in
the same buffer for 1 h. Specimens were dehydrated
through a graded series of ethanol dilutions ( 10%- 100%),
then immersed in propylene oxide and gradually infil-
trated with Spurr low-viscocity resin. Thick and thin sec-
tions were cut; the thick sections were stained with
methylene blue and azure B, and the thin sections were
immersed in uranyl acetate. The thin sections were viewed
with a Phillips electron microscope at 80 kV. As a positive
control, hatched B. villosa larvae were prepared for trans-
mission electron microscopy along with hatched A/, re-
tortiformis juveniles. In every B. villosa preparation ex-
amined, sarcomeres were clearly evident within the tail
muscle cells.

Acetylcholinesterase histochemistry

Day 2 M. retortiformis juveniles, Boltenia echinata.
Halocynthia pyriformis, and Ciona intestinalis larvae were
tested for acetylcholinesterase activity as previously de-
scribed by Karnowski and Roots ( 1964), Whittaker ( 1973),
and Bates and Jeffery (1987). Wholemount preparations
were viewed with an Olympus microscope and photo-
graphed with Plus X film.

18

W. R BATtS

Figures I and 2. Transmission electron micrographs of a sectioned Mnlt;iilti ri'inriilnnni.\ follicle cell ( 1 ;
and a sectioned M rclornfrmiiix gastrula (2). The swirl patterns of follicle cell droplets (d) are evident in ( 1 !
and a test cell (tc) is seen within the perivitelline space (ps) in (2). X 3300 in ( 1 ) and in (2).

Immunocytochemistry

M. retortiformis and B. villosu eggs were prepared for
immunocytochemistry, as previously described by
Mita-Miyazawa et al. (1987). Eggs were immersed for
20 min in absolute methanol, and then for 20 min in
cold absolute ethanol. Fixed eggs were infiltrated with
50% polyester wax (BDH Limited, Poole, England): ab-
solute ethanol for 1 h at 40C and then infiltrated with
100% polyester wax for 1 h at 40C. Specimens were
embedded in BEEM capsules, and 8-/jm sections were
cut from the blocks. Sections were mounted on gelatin-
coated coverslips, de-waxed through a graded series of
ethanol dilutions (100%; 90%; 80%; 70%; 50%; 30%),
and rinsed in phosphate buffered saline (PBS). The
specimens were incubated with a monoclonal antibody
( 1:25 dilution of NN18 from Sigma Chemicals) for 1 h
at room temperature, washed with PBS, and incubated
for 50 min in a 1:60 dilution of FITC-conjugated IgG
(Sigma Chemical Company), as previously described
by Swalla el til. ( 1 99 1 ). The specimens were washed in
PBS for 30 min, mounted in 80% glycerol dissolved in
PBS, and viewed with an Olympus fluorescence micro-
scope. Sections were photographed with Tri X film,
ASA 400.

Results

A large population of M retoriiformis adults was dis-
covered living on an underwater hill near Huntsman Ma-
rine Station at a depth of 50-100 feet. The animals were

attached directly to rocks and lived close to several other
ascidian species, Bolicnia ovifera. Molgitla citrina, Ascidia
callosti. and Halocynthia pyriformis. The A/, retortiformis
adults collected from the underwater hill ranged from
about 20 to 75 mm in diameter. Only a few specimens
were collected from sand and gravel sites dredged near
the underwater hill, suggesting that M. retortiformis adults
prefer a hard substrate.

Maximum egg diameters (not including the surround-
ing follicular envelope) were 230-240 ^m. The ultra-
structural features of M. retortiformis follicle cells are
shown in Figure 1. The cytoplasm of follicle cells con-
tained droplets of various sizes, the contents of which dis-
play swirl patterns. Follicle cells are attached to an acel-
lular chorion separated from the plasmalemma of the egg
by a narrow perivitelline space. Cells within the perivi-
telline space, termed test cells, were observed in a few
sections (Fig. 2).

The cytoplasm of M. retortiformis eggs contains large
quantities of yolk and glycogen. After an egg was cross-
fertilized, a thick coat of sticky adhesive material anchored
it to the bottom of the glass culture dish. Fertilization
triggered a rapid rearrangement of the egg cytoplasm,
known as ooplasmic segregation. Opaque cytoplasm
moved into one region of the egg and subsequently, just
before first cleavage, formed a narrow belt of opaque cy-
toplasm in the equatorial region. Unlike the eggs of several
other species, including B. villosa. the egg of M. retorti-
formis does not have colored pigment granules in its cor-
tex. In some of the fertilized eggs, ooplasmic movements

DIRECT DEVELOPMENT IN MOUiL'L.l

19

Kifjures 3 and 4. Transmission electron micrographs of sectioned Molgula rclorlitorniix gastrulae showing
the outer epidermal cells (ep) containing less yolk and glycogen than the large, centrally located cells (cc).
X 3300 in Fig. 3; X 4900 in Fig. 4.

were accompanied by changes in the overall shape of
the egg.

The early cleavage patterns exhibited by M. relortifor-
mis embryos appeared similar to those exhibited by other
ascidian embryos. The first cleavage plane bissected the
narrow belt of ectoplasm into two equal regions. The two
equal-sized blastomeres of a two-celled embryo continued
cell division and formed a gastrula. Cells in the vegetal
pole region invaginated in a manner similar to that seen
in Boltenia villosa gastrulae. As a result of these vegetal
cell movements, an archenteron resembling that of B. vil-
losa formed. The ultrastructural features of the various
cells that constitute a M. retortiformis gastrula are shown
in Figures 3 and 4. The cytoplasm of the large, centrally
located cells was packed with yolk and glycogen. Ecto-
dermal cells contained less yolk and glycogen than these
central cells. Other cell types, based on distinct ultrastruc-
tural features, were not evident.

Tail development was completely absent in M. retor-
tiformis. No indication of a shape change of the posterior
region or of notochord elongation was observed. Ampulla
outgrowth was always triggered at a fixed time in devel-
opment, immediately before hatching. Each juvenile de-
veloped a maximum of eight ampullae. Rhythmic con-
traction waves were evident in each ampulla by day 4 of
development. Blood cells were evident within each am-
pullar lumen.

Figures 5 and 6 show the ultrastructural features of
various cell types constituting day 2.5 juveniles. Yolk and
glycogen stored in the egg cytoplasm persisted through
day 2.5 of development and were not partitioned into any
particular cell type, but were present in varying amounts
in all cells. Epidermal cells contained fewer yolk granules
and glycogen than the larger, central cells of a juvenile.
Given that M. retortiformis juveniles do not start feeding
until after one week of development, the energy required
for all of the morphogenetic processes is likely derived
from the large, yolky cells. These cells probably make up
part of the adult rudiment. In striking contrast to species
that produce planktonic larvae, in M. retortiformis ju-
veniles have no differentiated muscle cells (compare Figs.
5 and 6 and Fig. 7). I tested the possibility that despite
the absence of differentiated muscle cells, these juveniles
might be able to express AChE activity. AChE histochem-
istry was performed on newly hatched M. retortiformis
juveniles at day 2 of development and on day-2 larvae
produced by Hulocynthia pyriformis, Boltenia echinata,
or dona intent inalis. The results of these experiments are
shown in Figures 8 through 1 1 and Table I. Larvae from
all three species that have indirect development showed
AChE activity in tail muscle cells (Fig. 9), whereas M.
retortiformis juveniles did not express AChE activity (Fig.
1 1 ). One hundred and sixty-three M. retortiformis juve-
niles from eight egg clutches collected during four sum-

20

W. R. BATES

Figures 5 and 6. Transmission electron micrographs of sectioned day 2.5 Molgula rclorlilorniis juveniles.
Yolk and glycogen were the predominant cytoplasmic feature of juvenile cells. Centrally located cells (cc)
contain large quantities of yolk and glycogen. Differentiated muscle cells were not observed in M. ri-lortil<>rnn\
sections. - 3300 in Fig. 5: 4900 in Fig. 6.

mers were tested. .17. rctortiformis juveniles lack not only
larval muscle cells, but also the sensory structures present
in the head region of tadpole larvae. NN 1 8, a monoclonal
antibody raised to vertebrate neurofilament protein,
stained the cortical region of B. villosa eggs (Fig. 8). In
contrast, NN 1 8 did not stain the cortical cytoplasm of M.
retortiformis eggs (Fig. 10). More than 100 sectioned eggs

,.,... .

* - : - ' -

V-.*V*V

;>..

e$ .

' ct - .-J'.'jjjj.. .1

"{ V, *

Figure 7. Transmission electron micrograph of a sectioned Bullcniu
f(//iiMi larva. Differentiated muscle cells were evident in It. villnsu larvae,
in contrast to A/, retortiformis preparations that lacked differentiated
muscle cells, my striated myofihril.

from different clutches were examined together with sec-
tioned B. ri/losu eggs.

Discussion

In summary, this report ( 1 ) provides new information
on the ultrastructural features of A7. ret<>nifon)ii.<i eggs,
gastrulae, and day-2.5 juveniles; (2) presents ultrastruc-
tural and histochemical evidence suggesting that M. rc-
i/>n//i>nni\ embryos do not produce differentiated larval
muscle cells; (3) burnishes immunocytochemical evi-
dence that M. retortiformis eggs lack a cortical protein
that is recognized by NN18 antibody; and (4) suggests
that Berrill's substrate hypothesis is in need of revision,
because M. rclorti/hmiis adults live on a hard, nonuni-
form substrate.

Large quantities of yolk and glycogen were present in
the cytoplasm of eggs and most cells constituting gastrulae
and day-2.5 juveniles. Two other direct-developing ascid-
ians. Mdlgulti puci/icu (Bates and Mallett, 1991a,b) and
Molgiiln oirii/ui (Jeftery and Swalla, 1990). produce eggs
containing large quantities of yolk and glycogen. In all
three of these direct-developing molgulids, as in ascidians
having indirect development (Berrill, 1975;Cloney, 1982),
feeding does not begin until after the development of adult
organs. Large quantities of yolk present in the cytoplasm

DIRECT DEVELOPMENT IN MOUiVLA

Figures 8-11. NNI8 antibody staining of Bulletin: villow and Mol-
gula retortiformis eggs and AChE expressions of B villosa larvae and
M. retortiformis juveniles. The cortical region of/?, villoxa eggs was stained
with NN18 antibody (Fig. 8). whereas the cortical region of M. retorti-
formis eggs did not stain with NNI8 antibody (Fig. 10). M. rcturti/imni.i
follicle cells are autofluorescent. g germinal vesicle. Fig. 9: Dark-stained
AChE positive muscle cells in the tail of a B villwa larva. Fig. 1 1: M
retortiformis juvenile exhibiting no AChE activity. Scale bars equal 50 ^/m
in (X); 100 jjm in (9): 50 /jm in (10). 100 ^m in ( I I )

of meroblastic types of eggs, such as those produced by
birds and reptiles, directly affect patterns of cell division
and modify cell movements associated with gastrulation.
The presence of a few test cells within the perivitelline
space of M. retortiformis eggs was surprising because such
cells are thought to be involved in the development of a
larval tail fin (Cloney. 1 982). Despite the yolky cytoplasm
of M. retortiformis eggs, early cell divisions were holo-
blastic. and gastrulation was similar to that in indirect-
developing embryos containing less yolk. Vegetal pole cells
invaginated to form an archenteron. In contrast, gastru-
lation in M. pacified embryos is highly modified (Bates
and Mallett, 199 la) and a typical archenteron never de-
velops. Instead, the large, yolky endoderm cells within the
central region of the embryo appear to physically impede
the inward movements of vegetal pole cells.

Ooplasmic segregation movements and early cleavage
patterns in M. retortiformis are similar to those in eggs
and embryos that have indirect development (Conklin,
1905; Bates and Jeffery, 1988). Unlike the eggs produced
by several species of Stye/a and by Boltenia villosa, the
eggs of Af. retortiformis do not contain colored pigment
granules associated with the cortical region. However, the
postfertilization movements of the egg cytoplasm of M.
retortiformis could be studied in live eggs due to the pres-
ence of an opaque cytoplasm presumably derived from
the contents of the germinal vesicle, as in other ascidians
(Conklin, 1905). Opaque cytoplasm first accumulated in
one region of the egg and was subsequently moved into
the equatorial region where it spread out and formed a
narrow cytoplasmic region. These cytoplasmic move-
ments that have been described in the fertilized eggs of
indirect-developing ascidians are thought to be important

in the specification of cell fates and axial development
(Conklin, 1905; Bates and Jeffery. 1988). It appears that
in M. pacified (Bates and Mallett, 1991a) and M retor-
tiformis. these precise movements of egg cytoplasm have
been evolutionarily conserved.

The absence of myofilaments and AChE activity in M.
retortiformis juveniles suggests that the developmental
program responsible for the specification of larval muscle
cells was eliminated. Myofilaments and AChE activity
were also absent in M. pacifica juveniles (Bates and Mal-
lett, 1991b). But at least two other molgulids that have
direct development can express low levels of AChE activity
(Whittaker, 1979; Jeffery and Swalla, 1990; Bates and
Mallett, 1 99 1 b). The interpretation that AChE activity in
a direct-developing ascidian is a vestige of larval muscle
cell expression is based on Berrill's assumption that direct
development evolved from species that have indirect de-
velopment (1931 ). This assumption is being tested in sev-
eral laboratories by comparing ascidian gene sequences.
DNA sequence comparisons may suggest that Af. retor-
tiformis is most closely related to another molgulid that
has direct development or to a molgulid with indirect
development. Maybe M. retortiformis is closely related to
Molgula eitrina, an indirect-developing species that lives
on the same underwater hill as M. retortiformis.

The elimination of differentiated muscle cells in M.
retortiformis may be due to an evolutionary modification
of the egg cytoskeleton, an idea first suggested by Swalla
ct a/ ( 1 99 1 ) in their study of direct-developing Af. occult a
embryos. NN18, an antibody raised to vertebrate neu-
rofilament protein, stains the cortical myoplasmic region
of B. villosa eggs, but did not stain M. retortiformis eggs.
This result suggests that a cytoplasmic factor recognized
by NN18 antibody, absent in M. retortiformis eggs, may
be involved in larval muscle cell specification. The ques-
tion of whether the antigens recognized by NN 1 8 antibody
are attached to the myoplasmic cytoskeletal domain, an
egg cytoplasmic region thought to be involved in muscle
cell specification (Jeffery and Meier, 1983), must await
future studies.

Table I

The failure / m/r hutched, day 2 Molgula retortiformis juveniles to
express acclylcholinesl erase activity

Number

Number

Species

tested

positive

Molgula retortiformis (D)

163

Ilalocynlhia pynlormis (I)

65

56

Biillenia c'chinaui (1)

78

77

Ciona intestinalii (I)

8

8

D species with direct development; I species with indirect devel-
opment. Tested at day 2 of development.

22

W. R. BATES

Data collected from field sites in the Atlantic and Pacific
oceans, on adults of M. retortiformis (present study) and
M. padficu (Bates and Mallett, 1 99 1 a) respectively, appear
to conflict with Berrill's substrate hypothesis (1931). Berrill
based his hypothesis on field and developmental studies
of Molgula ncciiltii (a direct developer) and Molgula ocu-
lata (an indirect developer), species that live on the sand-
flats along the coast of Brittany. The occurrence of M.
occulta was attributed to habitat uniformity. Berrill sug-
gested that tadpole development was eliminated from the
life cycle because tadpoles capable of selecting a habitat
are unnecessary in a uniform environment. But the largest
populations of M. retortiformis adults live in a rocky,
nonuniform habitat. Seven summers of field collections
along the west coast of Vancouver Island near Bamfield
Marine Station indicate that M pucificu adults thrive on
rocky, nonuniform habitats (Young ci <;/., 1988; Bates
and Mallett, 199 la; Bates, 1993). The evolution of mor-
phogenetic processes in ascidians has been discussed at
length elsewhere (Bates, 1993, 1994), with the suggestion
that the evolution of a fixed timing mechanism for trig-
gering a rapid deployment of ampullae may be important
to the reproductive success of direct -developing ascidians.
The finding described in the present report that ampulla
morphogenesis occurs at a fixed time in M retortiformis
development supports this idea.

Acknowledgments

Mike Swallow is thanked for providing Figure 7. Tech-
nical help with transmission electron microscopy and
darkroom assistance were provided by J. Mallett. I am
grateful to Fred Purton at Huntsman Marine Science
Centre, St. Andrews. New Brunswick, for making my visits
productive. I am grateful for the critical reviews of a pre-
vious form of this manuscript. W.R.B. is supported by
an operating grant provided by the Natural Sciences and
Engineering Council of Canada.

Literature Cited

Bates, \V. R. 1991. Ampulla morphogenesis in anural and urodele
ascidians. /)nv/ (inmih /)///</ 33:401-410.

Bates, VV. R. 1993. Evolutionary modifications of morphogenelic
mechanisms and alternate life history strategies in ascidians. Micrmc
Rc\ lech 26: 285-300.

Bales, \V. R. 199-4. Ecological consequences of altering the timing
mechanism for metamorphosis in anural ascidians. Am /.mil 34:
333_342.

Bates, VV. R., and \V. R. .letter). 19S7. I ocali/ation of axial deter-
minants in the vegetal pole of ascidian eggs. />IT liinl 124: 65-7<v

Bates, VV. R., and W. R. Jeffery. 1988. Polarisation of ooplasmic seg-
regation and dorsal-ventral axis determination in ascidian embryos.
Dcv liml 130: 98-107.

Bates, VV. R.. and ,1. K. Mallett. I991a. Anural development of the
ascidian Mulgulu pacilica (Huntsman), dm J //ml 69: 618-627.

Bates, VV. R., and J. E. Mailed. I991b. Ultrastructural and histo-
chemical study of anural development in the ascidian Mnlxitlu pucilicu
(Huntsman). Rou\\ Arch Dcv Biul 200: 143-201.

Berrill, N. J. 1931. Studies in tunicate development. Part II. Abbre-
viation of development in the Molgulidae. Pint Tram Sue B 219:
281-346.

Berrill, N. J. 1975. Chordata: Tunicata. Pp. 24 1 -282 in Reproduction
l Marine Invertebrate*. Vol. 2. A. C. Giese and J. S. Pearse. eds.
Academic Press, New York.

Cloney. R. A. 1978. Ascidian metamorphosis: review and analysis. Pp.
255-282 in Settlement nut/ Metamorphosis in Marine Invertebrate
Larvae. F. S. Chia and M. Rice. eds. Elsevier, New York.

Cloney, R. A. 1982. Ascidian larvae and the events of metamorphosis.
Am /ool 22: 817-826.

C'onklin, E. G. 1905. The organisation and cell lineage of the ascidian
egg../, .lead Null Sci 13: 1-119.

Grave, C. 1935. Metamorphosis of ascidian larvae. Papers from Tor-
tugas Laboratory, Vol. XXIX. Cumexic Inililitlion of Washington
riihlicalion 452: 209-292.

Grosberg, R. K. 1981. Competitive ability influences habitat choice in
marine invertebrates. Nan/re 290: 700-702.

Grosberg, R. K., and J. K. Quinn. 1986. The genetic control and con-
sequences of kin recognition by the larvae of a colonial marine in-
vertebrate. Nature 322: 456-459.

Jeffery, VV. R., and S. Meier. 1983. A yellow crescent cytoskeletal
domain in ascidian eggs and its role in early development. Dev. Biol.
96: 125-143.

.lettery, VV. R., and B. J. Swalla. 1990. Anural development in ascidians:
evolutionary modification and elimination of the tadpole larva. Sem.
Dev Biul. I: 253-261.

Jeffery, VV . R., and B. J. Swalla. 1991. An evolutionary change in the
muscle lineage of an anural ascidian embryo is restored by interspecific
hybridization with a urodele ascidian. Dev Biol 145: 328-337.

Jeffery, VV. R., and B. J. Swalla. 1992. Evolution of alternate modes
of development in ascidians. BioE<^t/ys 14: 219-226.

Karnovsky, M. J., and I.. Roots. 1964. A direct-coloring thiocholine
method for cholinesterase. 7 Ilivochcin Cnochcm, 12:219-221.

de I.aca/e-Duthiers, F. J. H. 1874. Histoire des ascidies simples des
cotes de France. 1. Arch //ml /-.'v/> (I en 3: 119-656.

Mitam-Miyazawa, I., 1 . Nishikata, and N. Satoh. 1987. Cell- and tissue-
specific monoclonal antibodies in eggs and embryos of the ascidian
Jliil/icynlhiti roicl:i l>c\'C/n/>ienl 99: 155-162.

Swalla, B. J., M. R. Badgett, and VV. R. Jeffery. 1991. Identification
of a cytoskeletal protein localized in the myoplasm of ascidian eggs:
Localization is modified during anural development. Development
111:425-436.

Van Name, VV. G. 1945. The North and South American Ascidians.
Hull, tin ,1/in Nm Union- 84: 422-424.

VV hittaker. J. R. 1973. Segregation during ascidian embryogenesis of
egg cytoplasmic information for tissue specific enzyme development.
Proc \iiil Ic/ul Si i. USA 70: 2096-2100.

VV hittaker, J. R. 1979. Development of vestigial tail muscle acetyl-
cholinesterase in embryos of an anural ascidian species. Biol Bull.
156: 393-407.

Young, C. M., R. K. Gowan, J. Dalby, C. A. Pennachetti, and I). Gagliardi.
1988. Distributional consequences of adhesive eggs and anural de-
velopment m the ascidian Mol^u/u /hici/icii (Huntsman 1912). Biol
Bull 174: 39-46.

Reference: Bn>l Hull 188: 23-31. (February/March, 1995)

Isolation of Biologically Functional RNA During
Programmed Death of a Colonial Ascidian

WEN-TEH CHANG 1 AND ROBERT J. LAUZON' : *

1 Department of Microbiology, Immunology and Molecular Genetics, and 2 Department of Pediatrics,
Albany Medical College, Albany. New York 12208

Abstract. The blastogenic (asexual) cycle of the colonial
ascidian Botryllus schlosseri (Tunicata. Ascidiaceae) con-
cludes in a cyclical phase of programmed cell and zooid
death called takeover, in which all asexually derived adults
die synchronously by apoptosis. The characterization of
developmental^ regulated genes whose expression pat-
terns are selectively modulated during this process could
pave the way to understand how this model organism
dies. However, isolation of biologically functional RNA
in this and other colonial ascidians with conventional
phenol/chloroform-based procedures is hampered by ex-
tensive contamination of RNA preparations by pigments.
Upon cell lysis, pigments that normally reside within spe-
cialized cells in the mantle wall of the adult are released
and tightly associate with nucleic acids. Here, we report
on the usefulness of a single-step RNA isolation method
in which acid guanidinium isothiocyanate is used as an
extraction medium, followed by preparative cesium chlo-
ride ultracentrifugation. This procedure successfully iso-
lated biologically active, high-purity total RNA (OD 26 o/
OD :s(l = 1 .9-2. 1 ) from Botryllus colonies during takeover,
as well as other species of colonial ascidians (Diplosoma
macdonaldii, Botrylloides diegense) irrespective of pig-
mentation. Northern blot analysis performed with a 32 P-
labeled tunicate actin probe detected two polyadenylated
transcripts of 1.5 and 1.7 kilobases in length from both
growth phase and takeover colonies. Two-dimensional
protein gel assays from //; vitro translated mRNA prep-
arations further revealed that specific transcripts were up-
regulated during takeover, while others were repressed or
down-regulated. Growth phase and takeover-specific
cDNA libraries were constructed from pooled poly(A) +
RNA with a complexity of 1.0 X 10 7 and 1.2 X 10 7 re-
Received 8 July 1994; accepted 22 November 1994.
* Author to whom correspondence should be addressed.

combinants respectively per 100 ng of cDNA before am-
plification. The procedure described herein renders fea-
sible the cloning of developmentally regulated genes in
this organism. In addition, our findings raise the possibility
that zooid death in Botryllus involves modulated gene
expression.

Introduction

Programmed cell death is a fundamental morphoge-
netic process within developing multicellular animals ( El-
lis et al.. 1991; Schwartz and Osborne, 1993). In adult
tissues, cell death also functions as a homeostatic mech-
anism complementary to mitosis: changes to this balance
bring about pathologic abnormalities (Ellis et al.. 1991).
Recent studies in vertebrates (Owens et al.. 1991; Miura
et al.. 1993; Woronicz et al.. 1994; Liu et al.. 1994) and
invertebrates (Ellis and Horvitz, 1986; Schwartz et al..
1990; White et al.. 1994) strongly suggest that cell death
is an active process dependent on modulated gene expres-
sion. One of the most characteristic forms of cell death is
a dynamic morphological process known as apoptosis,
characterized by nuclear chromatin condensation and
margination, cellular fragmentation into membrane-
bounded bodies followed by engulfment and digestion
within phagocytic cells (Kerr, 1972).

The colonial ascidian Botryllus schlosseri contributes
a unique perspective to the study of cell death: adult col-
onies, derived from a chordate tadpole through palleal
budding and which at peak size consist of approximately
1000 asexually derived clones (zooids), undergo weekly
phases of regression (Milkman, 1967). Every 5 days at
2 1 C, the blastogenic (asexual) cycle concludes in a phase
of programmed cell and zooid death called takeover, dur-
ing which all zooids, each containing a functional heart,
nervous and digestive systems simultaneously die by an

23

24

W.-T. CHANG AND R. L. LAUZON

apoptotic process over a 24-h period and are replaced by
a new asexual generation of zooids (Lauzon et a!., 1993).

Because the temporal and morphological events of
takeover can be predicted in detail in Botryllus, we have
undertaken a multidisciplinary investigation of the mo-
lecular mechanisms underlying programmed death in this
model organism. Thus far, molecular studies have been
impeded by the lack of appropriate methods for isolation
of biologically active mRNA. Unfortunately, Botryllus
and other colonial ascidians harbor polyphenolic and
DOPA-containing pigments that bind tightly with nucleic
acids following cellular lysis with detergents and chao-
tropic agents, thus interfering with the isolation process
as well as its subsequent analysis and cloning (Kumar et
ai, 1988). Moreover, isolation of intact RNA molecules
from regressing tissues may also prove to be difficult be-
cause, during cell death, a substantial fraction of the RNA
pool is rapidly degraded through the enhanced activity of
ribonucleases (Cidlowski, 1982; Owens et a/.. 1991).
Therefore, to ensure isolation of biologically active RNA
from these organisms, a strategy had to be developed that
would eliminate both ribonuclease activity and pigments.

Here, we report on the success of a procedure by which
biologically functional RNA suitable for cDNA cloning
and other molecular applications can be rapidly isolated
from various colonial ascidian species, including Botryllus.
In addition, we present evidence which indicates that
changes in gene expression occur during takeover.

Materials and Methods

Animals

Ascidians (Botryllus schlosseri. Botrylloides diegense,
Diplosoma macdonaldii, Molgula manhattensis) were
collected on glass microscope slides contained within
wooden enclosures submerged in the Eel Pond (Woods
Hole, MA) and Monterey Bay (CA). They were subse-
quently maintained in a refrigerated aquarium ( 1 50-gallon
capacity) containing artificial sea salts, trace and bio-ele-
ments (hw-marine mix: Hawaiian Marine Imports Inc.,
Houston, TX), and were continuously fed with an algal
scrubber irradiated for 12 h each day with two 15-watt
Aurora 50:50 bulbs (Fritz Pet Products, Dallas, TX). In-
dividual Botryllus colonies were developmentally staged
with the use of a stereomicroscope (Stemi SV 6, Carl Zeiss,
Germany). Following removal of debris and encrusting
organisms, all animals were subsequently snap-frozen in
liquid nitrogen and stored at 70C until needed.

RNA extraction

RNA was extracted using a modification of the method
by Chirgwin et ai ( 1978). Individual colonies (0.5-1.0 g)
were initially ground to a fine powder with liquid nitrogen

in a precooled mortar and pestle, and subsequently ho-
mogenized in 2.0 ml of a lysis solution containing the
following components: 4 M guanidium isothiocyanate
(Gibco/BRL, Gaithersburg, MD) predissolved in a 0.75 M
sodium citrate solution (pH = 7.0; 25 mA/ final concen-
tration), 0.2 M sodium acetate (pH = 5.0). and 0.1 M /3-
mercaptoethanol. Following transfer of the lysate to a 50-
ml polypropylene tube, the DNA was sheared with a 23-
gauge needle and syringe, and sodium lauryl sarcosinate
(10% stock) was added to a final concentration of 0.5%.
The homogenate was incubated on ice for 1 5 min and
centrifuged at 3500 X g for 5 min at 4C. The supernatant
was layered on a 1 .3-ml cesium chloride (CsCl; Boehringer
Mannheim, Indianapolis, IN) solution (5.7 AI CsCl, 0.5 M
EDTA, pH = 8.0) in a 1 3- X 5 1-mm ultracentrifuge poly-
allomer tube ( Beckman Instruments Inc., Palo Alto, CA),
and centrifuged at 40,000 rpm at 20C for 1 2 h in an
SW50. 1 rotor (Beckman Instruments Inc.). Following
centrifugation, the RNA pellet was resuspended in dieth-
ylpyrocarbonate (DEPC (-treated water (Sigma Chemical
Co., St. Louis, MO), precipitated overnight at -20C in
100% ethanol, and washed in 70% ethanol. The pellet was
subsequently dried under vacuum at room temperature,
resuspended in DEPC-treated water, and stored at 70C.
For phenol/chloroform-based extractions, the method
described by Chomcynski and Sacchi ( 1987) was used.

Spectrophotometric analysis

An aliquot from each RNA preparation (1-5 /ul) was
diluted into 250 jul of DEPC-treated water, transferred to
a quartz cuvette, and scanned between 240 and 320 nm
with a Perkin-Elmer X-2 microprocessor-controlled spec-
trophotometer (Perkin-Elmer Inc., Foster City. CA). A
GeneQuant spectrophotometer (Pharmacia, Piscataway,
NJ) was used to determine total RNA concentrations as
outlined in Sambrook et al. (1989). During the course of
our studies, we observed that OD2 6 o/OD 280 ratios were
greatly affected by pH. For instance, DEPC-treated water
samples that still contained residual levels of DEPC fol-
lowing autoclaving (pH = 5.0) gave aberrant OD 260 ab-
sorbance readings (between 1.3 and 1.5). Consequently,
we routinely autoclaved all our DEPC-treated solutions
twice for 30 min each. Poly-A + RNA was isolated with
the poly-A tract mRNA isolation system from Promega
(Promega Corp., Madison, WI), and concentrations were
determined with the Dipstick kit by Invitrogen (Invitrogen
Corp., San Diego, CA). Both were used according to the
manufacturer's specifications.

Northern blot hybridization

Ten micrograms of total RNA was denatured in for-
maldehyde, size-fractionated in 1% agarose/formaldehyde
gels (4 V/cm), and transferred onto nitrocellulose mem-

ISOLATION OF RNA FROM BOTRVLLVS

25

hranes (Schleicher and Schuell, Keene, NH) with 20 x
SSC. Blots were hybridized overnight with a cytoactin
cDNA probe from Styelu c/ava (SpCAS; Beach and Jef-
fery, 1990) under high-stringency conditions ( 1 M NaCl,
10% dextran sulfate, 1% SDS. 100A<g/ml denatured
salmon sperm DNA) at 65C. The SpCAS probe was la-
beled by random hexadeoxynudeotide priming to a spe-
cific activity of 10 9 cpm/^g DNA (Feinberg and Vogel-
stein, 1983). Blots were washed three times in IX SSC
(prepared from a 20X SSC stock solution: 3.0 M sodium
chloride, 0.3 M sodium citrate. pH = 7.0) and 0.1% SDS
at 65C 5 min each, followed by two additional washes
inO.lx SSC and 0.1% SDS (65 C) 15 min each, and au-
toradiographed with Kodak XAR film.

Two-dimensional protein gel electrophoresis

Poly-A + RNA (0.2 ^g/sample) was translated /// vitro
with a rabbit reticulocyte lysate from Promega using 35 S-
methionine. Protein products were analyzed by two-di-
mensional polyacrylamide gel electrophoresis (O'Farrell,
1975; O'Farrell ft al.. 1977) on a protein II xi slab cell
(Bio-Rad, San Diego, CA). //; v/m>-labeled samples were
focused with wide range ampholytes (pH = 3-10; Bio-
Lyte 3/10, Bio-Rad) in 4 M urea and 10% NP-40, and
size-fractionated on 10% sodium-dodecyl-sulfate poly-
acrylamide gels (SDS-PAGE) along with Brome Mosaic
Virus (BMV) molecular weight markers ( 1 10, 97. 35 and
20 Kd) provided as a control with //; vitro translation kits
(Promega). Gels were fixed in methanol/acetic acid, en-
hanced with RESOLUTION (EM Corp, Chestnut Hill,
MA), dried under vacuum, and autoradiographed with
Kodak XAR film at -70C. All samples were run at least
twice to ensure reproducibility of translational profiles
observed by autoradiography.

cDNA library construction

cDNA was synthesized from poly-A + RNA of pooled
colonies isolated at onset and early stages of takeover
(stages D-l and D-2) or from representative growth stages
(A, B-l, B-2 and C-l) with the unizap cDNA synthesis
kit from Stratagene (La Jolla, CA) using 3: P-dATP. The
cDNA products were ligated to Eco RI linkers, restricted
with Xho I, and cloned unidirectionally into lambda zap
vector, according to the manufacturer's specifications. Size
range of first and second strand cDNA products was de-
termined by alkaline agarose gel electrophoresis by the
slide technique. Briefly, 10 ml of 1% molten alkaline aga-
rose (containing 1 ml of 10X alkaline agarose buffer: 3 ml
of 5 N NaOH, 2 ml of 0.5 M EDTA and 45 ml of sterile
milli-Q water) was added near the upper center of a 5- X
7.5-cm glass slide, to which a mini-gel comb had been
attached over it with high-tension clips. The gel was run
in 1 X alkaline buffer at 75 V for 2 h at room temperature.

Following electrophoresis. the gel was blotted dry with
several changes of Kimwipes EX-L (Kimberly-Clark,
Roswell, GA), sealed in an air-tight hybridization bag,
and autoradiographed with Kodak XAR film at room
temperature. The library was packaged and titered ac-
cording to Stratagene's specifications. The level of non-
recombinants was determined by plating various phage
dilutions with XL 1 -Blue MRF cells along with IPTG
(200 mg/ml in water) and X-gal (20 mg/ml in dimethyl-
formamide). For either takeover or growth phase cDNA
library, blue background plaques were not observed on
plates containing up to 10 ? PFUs (plaque forming units),
indicating that the percentage of non-recombinants was
very low (less than 1 x 10 5 PFUs/Vg of phage arms).
Lastly, the primary library was amplified in XL 1 -Blue,
phage suspensions were stored at -70C, and an aliquot
was prepared to assess the quality of the cDNA library.
The quality of each cDNA library was assayed by probing
nitrocellulose plaque lifts for representation of actin-
complementary sequences using the SpCAS cytoactin
cDNA clone. Phage transfer was performed for 1 min at
room temperature, and filters were sequentially placed
for 3 min each onto sheets of 3MM paper saturated with
the following solutions: ( 1 ) 0.5 N sodium hydroxide and
1.5 M sodium chloride, (2) 10% SDS, (3)0.5 A/Tris-HCl
pH = 8.0 and 1.5 M NaCl, and (4) 2x SSC. Membranes
were subsequently baked at 80C under vacuum for
30 min, hybridized with the 32 P-labeled SpCAS cDNA
clone, and autoradiographed with XAR film at -70C.
Hybridization conditions and post-hybridization washes
were identical to those used in the northern blot analysis.

Results

The blastogenic cycle ofB. schlosseri

Developmental staging of B. schlosseri colonies was
adapted from the nomenclature used by Mukai and Wa-
tanabe (1976), as well as Izzard (1973), and is described
in Table I and depicted in Figure 1. Following metamor-
phosis of the free-swimming tadpole, a colony arises by
weekly cycles of palleal budding, in which the bud evag-
inates from the wall of its parent zooid. Under optimal
growth conditions, two to three primary buds are pro-
duced per zooid and can be easily observed dorsally by
stage B-2 (Fig. 1, panel B). By stage C-l. organogenesis
begins in the secondary bud with the formation of primary
atrial folds, and at this time it exhibits an elongated ap-
pearance as primary organs (gut rudiment) begin to form
(not shown). At 21C. the cycle concludes on the fifth
day with the synchronous death of all parent zooids, a
process called takeover (Lauzon et al.. 1992). The onset
of takeover is characterized by the shutdown of both oral
and excurrent siphons (Fig. 1, panel C). At this stage,
while buds begin to move dorsally, zooids are still re-

26

W.-T. CHANG AND R. L. LAUZON

Table I
Developmental stages of the blaslogenic cycle

Stage

Characteristic

A Onset of new cycle; opening of oral and excurrent

siphons.
B-l Secondary hud skewing to parent /ooid's anterior

hemisphere. Heartbeat begins in primary bud
B-2 Secondary bud is a closed double-layered vesicle.

C-l Organogenesis (atnal folds) begins in secondary bud.

Secondary bud elongates along its anteroposterior axis.
C-2 Primary subdivisions completed in secondary bud.

D-l Onset of takeover: shutdown of zooid's oral and

excurrent siphons. Primary buds move dorsally.
D-2 Early takeover; contraction of zooid along its

anteropostenor axis.
D-3 Mid-takeover (zooid involution): visceral organs are

being resorbed. Apoptotic cell death and macrophage

phagocytosis are prevalent.
D-4 End of takeover: cessation of heartbeat in zooid. Siphons

of new asexual generation not yet open.

sponsive to mechanical stimulus. In the early stages of
takeover (3-5 h post-onset), the zooids contract along their
anteroposterior axis and begin to shrink in size (Fig. 1,
panel D). Pigment cells, which normally reside in the
zooid's mantle wall, begin to accumulate in the vascular
ampullae. In the middle stages of takeover (12- 15 h post-
onset), visceral organs die principally through an apoptotic
process (Lauzon el a/., 1993), although necrotic changes
can also be observed alone or in combination with an
apoptotic morphology. Takeover concludes with the ces-
sation of heartbeat in zooids, and a new cycle begins with
the opening of siphons in the next asexual generation of
zooids (panel A).

RNA isolated by preparative ultracentrifugation is
biologically functional

As shown in Figure 2A, when RNA was extracted with
the guanidine isothiocyanate/phenol/chloroform proce-
dure (Chomcynski and Sacchi, 1987), ;he spectrophoto-
metric absorbance pattern was severely disrupted, exhib-
iting a peak absorbance at 268 nm instead of 260 nm. All
preparations extracted in this manner were significantly
contaminated with blue and red pigments that could not
be removed upon further phenol/chloroform extraction.
In addition, the yields from these preparations were very
poor (10-20 jug of total RNA/g tissue), and often displayed
OD 2 6o/OD 2 8o ratios greater than 2.5, suggesting that pig-
ments were contributing to the altered ratios. Further-
more, when the preparations were size-fractionated on
formaldehyde/agarose gels, most of the sample remained
in the loading well. We surmise that since many pigments
have been reported to be polyphenolic in nature (Kumar

et /., 1988), RNA was most likely sequestered to the or-
ganic phase along with them. In contrast, RNA isolated
by means of cesium chloride ultracentrifugation was
spectrophotometrically pure (Fig. 2B), exhibited optimal
ODibo/ODigu ratios between 1 .9 and 2.1, and consistently
produced yields ranging between 0.5 and 0.8 Aig/mg of
colony. When cesium-chloride-purified samples were
electrophoresed, prominent 28S and 18S ribosomal RNA
bands were visualized irrespective of the developmental
stage of the colony (Fig. 3, lanes 2, 3, 4, 7, 8) or species
(Fig. 3, lanes 5, 6). To test the integrity of the RNA, sam-
ples from various ascidians were size-fractionated by aga-
rose/formaldehyde gel electrophoresis, transferred to a
nitrocellulose membrane, and hybridized to a 3: P-labeled
cytoactin cDNA probe from Styela clava. The results,
which are shown in Figure 4, indicate that all colonial
ascidians (Botryllus, Botrylloides, and Diplosoma) ex-
pressed two polyadenylated transcripts of 1.5 and 1.7 kb
in length (panel A, lanes 1 -4; panel B, lane 5). In contrast,
the solitary ascidian Molgula manhattensis expressed only
a single 1.5-kb transcript. Furthermore, both transcripts
were expressed during all stages of the blastogenic cycle
in Botryllus (Fig. 4, panel B).

We next sought to determine whether Botryllus RNA
isolated by this method could also be in w/ro-translated
into protein. Samples (0.2 jug) of poly-A 4 RNA from var-
ious developmental stages were translated with a rabbit
reticulocyte lysate with 35 S-methionine and analyzed by
two-dimensional polyacrylamide gel electrophoresis to
examine patterns of gene expression between different
stages of the blastogenic cycle. The results (Fig. 5) dem-
onstrate that RNA preparations could be successfully
translated and focused into a wide spectrum of acidic-
and basic-range polypeptides during both the growth phase
of the cycle and takeover. Furthermore, at the onset of a
new blastogenic cycle, several different spots were iden-
tified from the acidic and basic range that were absent in
the early stages of takeover (panel A). Additional tran-
scripts (for instance, arrow in panels A and B, Fig. 5)
appeared to be significantly down-regulated during take-
over. Conversely, other transcripts were expressed in the
early stages of takeover, but absent at the beginning of a
new asexual cycle (panel B) or other stages (not shown).

Lastly, in order to determine whether mRNA from
takeover colonies was suitable for cDNA synthesis, poly-
A + RNA was reverse-transcribed, and cDNA products
were subsequently size-fractionated by alkaline gel elec-
trophoresis. The results (Fig. 6) indicate that both first-
and second-strand cDNA products exhibited an appro-
priate size range, with the bulk distributed between 300
bases and 2.5 kb. Pooled cDNA products were then
used to construct a unidirectional library into lambda
zap. The size of the primary nonamplified library was
1.2 X 10 7 PFUs/100 ng of cDNA, as determined from

ISOLATION OF RNA FROM BOTRYLLUS

27

Figure I. The blastogenic cycle ofBotryllus M'/i/m.stv/ Individual colonies were developmental!)" staged
by stereomicroscopy and are depicted dorsally in panels A through D. Panel A shows a colony at the onset
of a new cycle. Note that the primary buds are not visible from the dorsal plane. Panel B shows a colony
during stage B-2 (see Table I for specific details of individual stages), in which primary buds are now visible.
The onset of takeover (panel C) is characterized by the shutdown of oral and excurrent siphons in all zooids
(arrow) and star-shaped systems. Buds (arrowhead) have begun their dorsal migration. In the early stages of
takeover (panel D), each zooid undergoes a synchronous polarized contraction along its anteroposterior axis.
Zooid regression is completed in approximately 24 h at 2 1 C. and a new cycle begins with the opening of
siphons from the new asexual generation of zooids. Bar represents 1 mm.

plaque counts using serial dilutions of phage suspensions.
A cDNA probe encoding cytoplasmic actin from Styela
clava (as a prototype abundant sequence) was then used
to screen nitrocellulose plaque lifts to ensure adequate
representation of this sequence. A comparable screen was
performed with a growth phase (pooled stages A, B-l and
C-l) cDNA library (1.0 X 10 7 PFUs nonamplified). The
percentages of actin-positive clones (Fig. 7) were found
to be comparable in both libraries, namely 3.0% ( 1 50 pos-
itive/5 X 10 3 plaques) in the takeover and 3.2% in the
growth phase cDNA library (58 positive/ 1.8 : 10'
plaques).

Discussion

The findings presented in this paper indicate that RNA
isolated by cesium chloride ultracentrifugation is opti-

mally suited for a wide range of molecular applications,
including northern blot analysis, /// vitro translation, and
cDNA synthesis from dying tissues ofBotryllus schlosseri
and other colonial ascidians. Because colonial ascidians
(Kumar el at., 1988) and other marine invertebrates
(Groppe and Morse, 1993) exhibit a spectacular range of
pigmentation patterns, they have been reported to pose a
distinct problem in the isolation of nucleic acids. Poly-
phenolic compounds and DOPA-containing proteins,
which interfere with nucleic acid isolation, have been
found in the adult tunic and mantle wall of both solitary
and colonial ascidians (Kumar el ul., 1988). In Botryllus
colonies found on the eastern coast of the United States,
the problem is intensified because most colonies contain
an alcohol-insoluble red pigment that cannot be removed
from nucleic acids with conventional lysis buffers followed
by phenol/chloroform-based extractions. The addition of

28

W.-T. CHANG AND R. L. LAUZON

153-
123
093-
063-
033-
003-

Figure 2. Spectrophotometnc scanning analysis of isolated total RNA
from Bolrvllus schlmseri. (A) RNA isolated from B schlosseri with a
conventional extraction method that utilizes phenol/chloroform/gua-
nidme isothiocyanate (Chomcynski and Sacchi, 1987) demonstrates an
altered ultraviolet absorption spectrum. In contrast (B), RNA isolated
with the single-step cesium-chlonde method is free of contaminants and
exhibits an optimal OD 26 o/OD 280 ratio ( 1 .9-2. 1 ).

a cesium chloride ultracentrifugation step permitted the
recovery of high yields of spectrophotometrically and
electrophoretically pure RNA preparations, irrespective
of pigmentation or species. Groppe and Morse (1993) re-
cently described a two-step cold method of isolating RNA
from Haliotis ntfescens (red abalone); the method pro-
vided high yields of pigment-free, undegraded material
suitable for cDNA cloning. The first step, a phenol/chlo-
roform extraction performed at 0C, was crucial for the
removal of ribonuclease activity, and the second step,
employing ultracentrifugation through a cesium chloride
gradient, removed an inhibitor of reverse transcriptase.
The observations reported herein indicate that in Botryllus
and other colonial ascidians, only a single preparative ul-
tracentrifugation step through cesium chloride is required
for isolation of biologically functional RNA.

However, our findings are in contrast with those of Ku-
mar el al. (1988), who reported that they successfully iso-
lated RNA from various ascidians by using only a phenol/
chloroform-based procedure. In our hands, all prepara-
tions isolated using phenol/chloroform were significantly
contaminated with pigments and gave very poor yields.
Furthermore, much of the original sample was left in the
loading well during formaldhyde/agarose gel electropho-
resis, and the efficiencies for in vitro translation reactions
and reverse-transcription for cDNA library construction
were significantly impaired (W-T.C and R.J.L., unpub.
obs.). At present, we have no explanation for the discrep-
ancy between our results and those of Kumar et al. ( 1988).

3911
2800

1898

872
562

Figure 3. Ethidium bromide staining of total RNA isolated by pre-
parative ultracentrifugation. Lane I. RNA markers. Lanes 2, 3, 7, and
8. Bolryllus vhhnscn from Eel Pond (Woods Hole. MA): lane 2 (stage
A), lane 3 (stage B-2), lane 7 (early takeover; 3 h post-onset), and lane
8 (mid-takeover, 12 h post-onset). Lane 4, B schlosseri from Monterey
Bay, CA (stage C-2). Lane 5, Bolrylloidcx dicgense (growth phase). Lane
6, Diplosoma nuicdniinklii

One possibility is that the composition of pigments found
in Botryllus and other botryllid ascidians may differ from
those found in other solitary or colonial species reported

1.7 kb
1.5kb

1.7 kb
1.5kb

.-UK-

B

H

Figure 4. Northern blot analysis of RNA isolated from various species
of colonial and solitary ascidians. Samples were electrophoresed on a
1% agarose/formaldehyde gel, transferred to nitrocellulose, and hybridized
with a 32 P-labeled cytoactin probe SpCAS from Slre/a cluru (Beach and
Jeffery, 199(1). (A) Lane 1. Bniryllns scltlm.seri (stage A) from Eel Pond
(Woods Hole, MA); lane 2. B .ichliaxcri (stage C-2) from Monterey Bay
(CA); lane 3. Boirylloides diegense (growth phase); lane 4. /)//'/< worna
macdonaldn. lane 5, Molgula iinin/nillcnsis (B) Total RNA samples of
B schlosseri from Eel Pond isolated at various stages of the blastogenic
cycle (lanes 1-4), and pooled poly-A* RNA from stages A-C colonies
(lane 5). Lane 1. early takeover, lane 2. mid-takeover; lane 3, stage A;
lane 4, stage B- 1 .

ISOLATION OF RNA FROM BOTRYLLUS

29

IEF

SDS

110 kd -

97 kd -r

35 kd -

20 kd -

B

'r^I *<

* 4*. * * \m

Figure 5. Two-dimensional protein gel analysis of in two-translated RNA with 35 S-methionine reveal
changes in gene expression during takeover. Panel A depicts a colony at the onset of a new blastogenic cycle
(stage A), whereas panel B is from a colony in early takeover (3 h post-onset). The circled spots in panel A
represent transcripts that are repressed in the early stages of takeover. The arrows in panels A and B depict
a representative polypeptide whose mRNA is down-regulated during takeover. Conversely, the circled spots
in panel B are transcripts that appear to be induced de nova during takeover. Abbreviations: SDS. sodium
dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) dimension; IEF, isoelectric focusing di-
mension.

by Kumar et al. (1988). The presence of contaminating
pigments markedly altered OD: W)/ 2 8 o ratios, presumably
by absorbing in the range that is optimal for nucleic acids
(i.e., 260 nm). In support of this hypothesis, we have re-
cently observed that pigment cells from live colonies are
fluorescent under ultraviolet light (R.J.L., unpub. obs.).
Northern blot analysis with cesium-chloride-purified
material further revealed that the RNA was not degraded
by ribonuclease activity present during zooid regression.
Previous studies have cautioned that isolation of intact
RNA molecules from dying tissues can be significantly
impeded by ribonucleases (Cidlowski, 1982; Owens et ai,
1991). In addition, all colonial ascidian species reported
in this paper (Botryllus schlosseri from the East and West
coasts, Botryloides diegense, and Diplosoma macdonaldii)
expressed two poly-A* transcripts of 1.5 and 1.7 kb in
length that hybridized to a cytoplasmic cDNA clone from
Styela clava. This was in contrast to the single 1.5 kb-
message found in the solitary ascidian Mo/gula manhal-
tensis. The significance of two mRNAs in colonial ascid-
ians is unclear, although another solitary tunicate. Stye/a
clava, was previously reported to express a single 1.8-kb

message during both embryonic and post-metamorphic
development (Beach and Jeffery, 1990). The additional
1.7-kb band in colonial species may represent a cross-
hybridizing muscle actin transcript. Tomlinson et al.
(1987) showed that a probe made exclusively from the 3'
untranslated region of a Styela muscle actin clone detected
transcripts exclusively in muscle cells, whereas one made
from the coding region, such as the cytoplasmic cDNA
clone used in this study (e.g.. SpCAS), detected both mus-
cle and nonmuscle transcripts. However, several lines of
evidence argue against this scenario. First, although both
transcripts were expressed at all phases of the blastogenic
cycle in Botryllus including takeover, the relative intensity
of the bands varied at different stages of the cycle. Second,
if the 1.7-kb transcript represented a cross-reactive muscle
mRNA. one would expect the intensity of the hybridizing
band to be less than the 1.5-kb transcript at any given
time under high-stringency conditions. This condition was
clearly not observed. Alternatively, both transcripts could
result from alternative splicing. The expression of an ad-
ditional 1.7-kb transcript could be functionally related to
the colonial life style, but seems unlikely to be associated

30

W.-T. CHANG AND R. L. LAUZON

with zooid death since Diplosoma species do not undergo
takeover. An intriguing possibility is that it may be ex-
pressed during bud development. Therefore, determina-
tion of the complete nucleotide sequences of both cDNA
clones followed by in situ hybridization with non-cross-
hybridizing probes will be required to resolve this issue.
Studies with invertebrate (Wadewitz and Lockshin,
1988) and vertebrate (Wang and Brown, 1991) develop-
mental systems indicate that individual death programs
may involve fewer than 40 up-regulated genes. For in-
stance, thyroid-hormone-mediated changes leading to tail
resorption in Xenopus laevis involve two periods of gene
expression during which all genes belonging to a specific
group are induced with identical kinetics. Conversely,
about 10 additional genes are down-regulated with iden-
tical decay kinetics (Wang and Brown, 1993). These ob-
servations indicate that in amphibians the death program
reflects a relatively simple pattern of gene expression. The
initial findings reported here with two-dimensional protein
gels from Botryllus suggest that modulated gene expression
occurs during the takeover phase of blastogenesis. We have
previously demonstrated that takeover involves the po-
larized breakdown of the perivisceral extracellular matrix
along the zooid's anteroposterior axis, followed by apop-
totic and necrotic morphological changes within dying
visceral tissues (Lauzon et a/., 1992, 1993). Changes in
gene expression may thus be associated with these mor-
phological events. Unfortunately, the shutdown of oral
siphons during takeover precluded us from analyzing 35 S-
methionine incorporation patterns in vivo. Therefore, the
possibility cannot be ruled out that differences in 2-D

B

23,130
9,146
6,557

4,361

2,320
2,037

567

Figure 6. Alkaline agarose gel electrophoresis assay of first- and sec-
ond-strand cDNA synthesis in Botryllus schlosseri. First- (lane A) and
second-strand (lane B) cDNA products were converted using pooled poly-
A* RNA isolated from colonies during takeover (onset and early take-
over). Note that both lanes exhibit a broad size distribution of cDNA
products, with the majority of material ranging between I and 2 Kb.

B

Figure 7. Nitrocellulose plaque lifts from growth stages (panel A)
and takeover (panel B) cDNA libraries hybridized with an ascidian 32 P-
labeled cytoactin probe from Si vela clava (SpCAS). Percent positive
plaque forming units in (B) was 3.0% (1 50 positive out of 5 X 10 3 PFUs)
compared to 3.2"! for the growth phase cDNA library (58 positive out of
1. 8 x I0 3 PFUs), indicating that actin was adequately represented.

protein profiles between the onset of blastogenesis and
takeover are due to inherent limitations of the in vitro
translation kits. In addition, since clonal replicates were
not used in any of these studies, the differences observed
may represent intra-species polymorphisms. Lastly, since
takeover involves the simultaneous regression of adult
zooids along with asexual growth of the future parental
generation, the possibility cannot be excluded that tran-
scriptional changes also occur in buds or in the colonial
vasculature. Therefore, assessing the specificity of tran-
scriptional changes will require isolation of takeover-spe-
cific mRNAs and analysis of their spatial distribution pat-
tern by in situ hybridization. We are currently using dif-
ferential mRNA display (Liang and Pardee, 1 992) as a
means for ultimately characterizing full-length transcripts
from the cDNA libraries. Interestingly, the percentages of
actin-positive PFUs were similar in the growth phase and
takeover libraries (3.2 versus 3.0). Libraries with reported
actin cDNA-positive frequencies above 0. l% have yielded
clones of interest for sequences of moderate to low abun-
dance, whereas percentages below 0.05% have not (Hagen
et a/.. 1988). Collectively, our findings strongly suggest
that both libraries are likely to contain cDNAs corre-
sponding to single-copy gene transcripts. The character-
ization of genes involved in zooid regression could provide
a fundamental understanding of molecular mechanisms
of programmed cell death in Botryllus and other meta-
zoans.

Acknowledgments

The authors gratefully acknowledge Dr. Craig Tomlin-
son for his generous gift of the SpCAS cytoactin cDNA
clone, and two anonymous reviewers for their efforts in
improving the focus of this manuscript. This work was
supported by a Basil O'Connor Starter Scholar Award
from the March of Dimes Birth Defects Foundation #5-

ISOLATION OF RNA FROM BOTRYU.L'S

31

FY94-0813, and from a Frederick Bang Fellowship at the
Marine Biological Laboratory, Woods Hole, Massachu-
setts.

Literature Cited

Beach, R. L., and W. R. Jeffery. 1990. Temporal and spatial expression
of a cytoskeletal actin gene in the ascidian Styela clava. Devel. Genet.
11: 2-14.

Chirgwin, J. M., A. E. Przybyla, R. J. MacDonald, and W. J. Rutter.
1978. Isolation of biologically active ribonucleic acid from sources
enriched by ribonuclease. Biochemistry 18: 5294-5299.

Chomczynski, P., and N. Sacchi. 1987. Single-step method of RNA
isolation by acid guanidinium thiocyanate-phenol-chloroform ex-
traction. . -tnal. Bmchem. 162: 156-159.

Cidlowski, J. A. 1982. Glucocorticoids stimulate ribonucleic acid deg-
radation in isolated rat thymic lymphocytes in vitro. Endocrinology
111: 184-190.

Ellis, H. M., and H. R. Horvitz. 1986. Genetic control of programmed
cell death in the nematode C. elegans. Cell 44: 817-829.

Ellis, R. E., J. Yuan, and H. R. Horvitz. 1991 . Mechanisms and func-
tions of cell death. Ann Rev Cell Bioi 7: 663-698.

Feinberg, A. P., and B. Vogelstein. 1983. A technique for radiolabelling
DNA restriction endonuclease fragments to high specific activity.
Anal. Biochem. 132: 6-13.

Groppe, J. C., and D. E. Morse. 1993. Isolation of full-length RNA
templates for reverse transcriptase from tissues rich in RNase and
proteoglycans. Anal. Biochem 210: 337-343.

Hagen, F. S., C. L. Gray, and J. L. Kuijper. 1988. Assaying the quality
ofcDNA libraries. Biotechniques 6: 340-345.

Izzard, C. S. 1973. Development of polarity and bilateral asymmetry
in the palleal bud of Botryllus schlosseri (Pallas). J. Morph. 139: 1-
26.

Kerr, J. F. R., A. H. Wyllie, and A. R. Currie. 1972. Apoptosis: a basic
biological phenomenon with wide-ranging implications in tissue ki-
netics. Br J Cancer 26: 239-257.

Kumar S., B. M. Degnan, I. L. Ross, C. J. Hawkins, and M. F. Lavin.
1988. Isolation of DNA and RNA from ascidians. Mar. Bio/ 98:
95-100.

Lauzon, R. J., K. J. Ishizuka, and I. I.. Weissman. 1992. A cyclical,
developmentally-regulated death phenomenon in a colonial uro-
chordate. Dev. Dyn 194: 71-83.

Lauzon, R. J., C. W. Patton, and I. L. Weissman. 1993. A morpho-
logical and immunohistochemical study of programmed cell death
in Botryllus schlosseri. Cell Tis. Res. 272: 1 15-127.

Liang, P., and A. B. Pardee. 1992. Differential display of eukaryotic
messenger RNA by means of the polymerase chain reaction. Science
257: 967-970.

l.iu, Z-G., S. \V. Smith, K. A. McLaughlin, L. M. Schwartz, and B. A.

Osborne. 1994. Apoptotic signals delivered through the T-cell re-
ceptor of a T-cell hybrid require the immediate-early gene nur77.

\alitn-367: 281-284.
Milkman, R. 1967. Genetic and developmental studies on Botrvllus

seliln.^en. Bint Bull 132:229-243.
Miura, M., H. Zhu, R. Rotello, E. A. Hartwieg, and J. Yuan.

1993. Induction ofapoptosis in fibroblasts by I L-l -beta-con verting

enzyme, a mammalian homolog of the C elegans cell death gene

ced-3. Cell 75: 653-660.
Mukai, II., and II. \\atanabe. 1976. Studies on the formation of germ

cells in a compound ascidian Botryllus primigenus Oka. J. Morphol.

148: 337-362.
O'Farrell. P. H. 1975. High resolution two-dimensional electrophoresis

of proteins. J Bioi Chem 250:4007-4021.
O'Farrell, P. Z., H. M. Goddman, and P. H. O'Farrell. 1977. High

resolution two-dimensional electrophoresis of basic as well as acidic

proteins. Cell 12: 1 133-1 142.
Owens, G. P., W. E. Hahn, and J. J. Cohen. 1991. Identification of

mRNAs associated with programmed cell death in immature thy-

mocytes. Mnlec Cell Bioi 11: 4177-4188.

Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular Clon-
ing A Laboratory Manual, 2nd ed. Cold Spring Harbor Laboratory.

Cold Spring Harbor Lab Press. Cold Spring Harbor, NY.
Schwartz, L. M., L. Kosz, and B. K. Kay. 1990. Gene activation is

required for developmentally programmed cell death. Proc Nail.

Acad Sri. USA 87: 6594-6598.
Schwartz, L. M., and B. A. Osborne. 1993. Programmed cell death,

apoptosis and killer genes. Immunol. Today 14: 582-590.
Tomlinson, C. R., R. L. Beach, and W. R. Jeffery. 1987. Differential

expression of a muscle actin gene in muscle cell lineages of ascidian

embryos. Development 101: 751-765.
Wadewitz, A. G., and R. A. Lockshin. 1988. Programmed cell death:

dying cells synthesize a co-ordinated, unique set of proteins in two

different episodes of cell death. f'EBS Letters 241: 19-23.
Wang, Z., and D. D. Brown. 1991. A gene expression screen. Proc.

Nail. Acad. Sa L'SA 88: 1 1505-1 1509.
Wang, Z., and D. D. Brown. 1993. Thyroid hormone-induced gene

expression program for amphibian tail resorption. J Bioi Chem.

268: 16270-16278.
White, K., M. E. Grether, J. M. Abrams, L. Young, K. Farrell, and H.

Steller. 1994. Genetic control of programmed cell death in Dro-

sophilu. Science 264: 677-683.
Woronicz, J. D., B. Cain, V. Ngo. and A. Winoto. 1994. Requirement

for the orphan steroid receptor Nur77 in apoptosis of T-cell hybri-

domas. Nature 367: 277-28 1 .

Reference: Binl. Bull 188: 32-45. (February/March. 1995)

Prespawning Behavior, Spawning, and Development
of the Brooding Starfish Leptasterias polaris

JEAN-FRANCOIS HAMEL AND ANNIE MERCIER

Depart ement d'Oceanographie, Universite dii Quebec a Rimouski, Centre Oceanographique
de Rimouski, 310 allee des Ursulines. Rimouski (Quebec), Canada G5L 3A1

Abstract. Our study focused on the precise reproductive
behavior of the starfish Leptasterias polaris (Miiller and
Troschel) before and during spawning a subject of much
speculation and evident ecological importance. Between
the third week of December 1992 and mid-January 1993,
we observed spawning in the laboratory that roughly cor-
responded to field observations in the Lower St. Lawrence
Estuary. In experimental tanks provided with natural en-
vironmental conditions, the spawning was preceded by 7
to 8 weeks of complex aggregative interactions among the
starfish. The individuals, which usually avoid each other,
began to make discreet arm contact, which intensified
with time and eventually led to the superposition of two
or more starfish, independently of sex. The interactions
seem to be associated with decreasing temperature, be-
cause aggregative and spawning behaviors were not ob-
served under stable temperature conditions. Male spawn-
ing is first initiated when the temperature falls to about
2C during minimum daylength (<9 h-d '). In seawater,
the spermatozoa are negatively buoyant and tend to de-
posit as a sticky film on the substrate, where they enter a
state of low activity. Stimulated by male spawning, females
spawn on the layer of sperm, which is reactivated by con-
tact with the oocytes, ensuring fertilization. In the labo-
ratory, the fertilized eggs undergo first cleavage in 45 h,
become brachiolaria in 40 days, and form fully developed
young starfish within 5.5 to 6 months, synchronously with
populations in the field. The embryos develop at the same
rate even when not brooded, suggesting that the brooding
behavior in L. polaris serves mainly to keep the eggs clean,
healthy, and protected against predation.

Introduction

Successful fertilization constitutes a critical stage in
marine invertebrate reproduction, and many organisms

Received 7 December 1993; accepted 4 November 1994.

develop strategies to maximize this important step (Him-
melman. 1981; Giese and Kanatani, 1987). Starfish show
diversified reproductive behaviors. In many species, ga-
metes are broadcasted by both sexes, with fertilization in
the water being enhanced by synchronization of spawning
(Hyman, 1955; Strathmann, 1987; Chia and Walker.
199 1 ). In other starfish, males broadcast spawn in the usual
fashion, and females emit fewer gametes but brood their
embryos to fully developed young starfish (McClary and
Mladenov, 1990; Chia and Walker, 1991). Leptasterias
polaris, which protects its embryos for 5 to 6 months, is
among the few species that brood by overlaying the eggs
deposited on the substrate (Emerson, 1977; Himmelman
et ai, 1982; Boivin el a/.. 1986). Although brooding star-
fish are generally small-sized, with lecithotrophic devel-
opment (Chia and Walker, 1991). L. polaris can reach
diameters up to 50 cm (Boivin et al., 1986) and are prob-
ably among the largest brooders.

Prespawning and spawning behaviors are very impor-
tant to reproductive success in marine invertebrates.
Breeding aggregations have been observed in a number
of asteroids (Chia, 1968; Komatsu, 1983; Minchin, 1987;
Young et a/., 1992; Slattery and Bosch, 1993). Many au-
thors suggest that such aggregations could minimize sperm
dilution and increase fertilization success (Ormond et al.,
1973; Levitan, 1991; Levitan et al., 1992), as exemplified
by the pairing strategies in Archaster typicu.i (Run et al..
1988) and Neosmilaster georgianus (Slattery and Bosch,
1993). In those species, the male, after finding a female,
mounts her before spawning (Ohshima and Ikeda, 1934;
Komatsu, 1983; Run et al.. 1988; Slattery and Bosch,
1993). There is also evidence that the spatial distribution
of broadcast spawners has a major influence on the prob-
ability of fertilization due to gamete viability (Pennington,
1985; Yund, 1990; Levitan et al.. 1992; Young et al..
1992). Young et al. (1992) suggested that aggregations

32

REPRODUCTIVE BEHAVIOR OF LEPTASTER1AS POLARIS

33

could be useful in overcoming the absence of the usual
spawning cues (e.g., light, temperature) in bathyal echi-
noid populations. The possible role of pheromones and
other possible attractants on clustering and related
spawning inducement in starfish has been examined
(Lewis, 1958; Miller, 1989). Komatsu (1983) suggested
that initial heterosexual recognition and pairing in A. typ-
icus allow male spawning to be induced by release of ma-
ture oocytes from females. However, most studies on ag-
gregation have associated it with cooperative feeding or
predation avoidance (Ormond el a/., 1973; Blankley and
Branch, 1984; Sloan, 1984; Pearse and Cameron, 1991).

Although different kinds of aggregations during spawn-
ing have been observed, little is known about the involve-
ment of grouping prior to spawning other than recent
work on the bathyal sea urchin Stylocidaris lineata (Young
el a/., 1992). Moreover, prespawning interactions have
never been discussed in respect to environmental factors
such as photoperiod and temperature, which are known
to influence gametogenesis and spawning, respectively
(Giese and Pearse, 1974; Himmelman, 1981; Pearse and
Walker, 1986; Pearse el al. 1986; Pearse and Cameron,
1991). As for fertilization strategies, observations on ga-
mete interactions, other than mutual recognition and at-
traction, remain scarce. Sperm motility and respiration
activation by egg extracts have been studied in sea urchins
(Suzuki el ai, 1982) and in horseshoe crabs (Clapper and
Epel, 1981). but only sperm chemotaxis has been de-
scribed in detail for starfish (Miller, 1985).

The starfish Leptasterias polaris can be kept in labo-
ratory facilities that reproduce natural conditions, and
this has provided a chance to record and describe its ag-
gregative behavior both before and during spawning. Fur-
ther evidence from experiments on gamete behavior and
embryonic development allowed us to better understand
the evolutionary strategy that seems to link spawning, ga-
mete fertilization, and brooding activities in this species.

Materials and Methods

Using scuba, we collected 60 specimens of Leptasterias
polaris from a depth of about 1 m on the south shore of
the Lower St. Lawrence Estuary (48 21' N: 68 47' W),
eastern Canada. The animals, ranging from 150 to
200 mm in diameter, were collected in May 1992, to en-
sure that they were acclimatized well before the December
spawning that we expected on the basis of previous ob-
servations by Boivin el at. (1986). The starfish were kept
in tanks to which seawater from the collect site was sup-
plied by a flow-through system and light on natural pho-
toperiod was provided through large windows. Physical
and chemical conditions were therefore similar to the
natural environment. Preliminary determination of sex
in L. polaris demonstrated a natural sex ratio close to 1:

1 , and this was carefully reproduced in the tanks. To es-
tablish the importance of environmental factors on pre-
spawning aggregative behavior, spawning, and develop-
ment, the temperature and the salinity of the circulating
water were continuously recorded. The data for the day-
length were provided by the Canadian government (En-
vironment Canada; Atmospheric Environmental Service,
Quebec airport). The starfish were given an unlimited
quantity of mussels (Mytilus edulis, =20 mm in shell
length), their favorite prey (Himmelman and Dutil, 1991 ).
Frequently very abundant in subtidal environments
(Himmelman and Dutil, 1991), L. polaris adapts ex-
tremely well to experimental conditions.

Prespawning behavior

Starfish behavior was recorded on a regular basis be-
tween 4 and 14 times a week depending on activities ob-
served, from November 1992 to February 1993, with
complementary observations before and after this period.
The number of individuals preying on the mussels, resting
on the bottom, and climbing on the sides of the tank were
noted. The number of starfish in contact was recorded
and categorized as light, when arms touched from the
middle part to the tip; intimate, when the arms intertwined
for more then half their length; or superposition, when
the individuals overlaid one another (Fig. 1). Particular
attention was given to reactions of the starfish after re-
newal of food supply, about twice a month. Data were
subsequently combined for weekly comparison of contact
intensities. Two control groups were also observed during
all prespawning and spawning experiments. Group 1 was
maintained under constant conditions in the Quebec
Aquarium at a temperature of 6C, a salinity of 28% and
a daylength of 10 h; group 2 was kept in continuous dark-
ness with natural conditions of salinity and temperature.

Spawning behavior

As the spawning period approached (Boivin el al.,
1986), observations focused on detecting gamete release
in relation to the postures adopted by the starfish and to
the prevailing environmental conditions. When observed,
spawning events were carefully described.

To test the hypothesis that sperm induces spawning in
females, a solution (1.2 X 10 1 spermatozoa ml ' deter-
mined with a hemacytometer under a light microscope)
was prepared with freshly collected sperm from a single
male. This sperm solution was poured into a tank (4 m 3 )
among mature females, and their subsequent behavior
was recorded. This procedure was carried out four times
at different periods under the conditions previously de-
scribed for the starfish in control group 1. Male starfish
were exposed to the same sperm concentrations to see if
they would be induced to spawn.

34

J.-F. HAMEL AND A. MERCIER

Figure 1. Photographs illustrating the prespavvning aggregations of
Leptasterias polaris in the laboratory. (A) Superposition of two individ-
uals. (B) Massive aggregation of starfish.

Sperm behavior

The term dry sperm refers to undiluted, freshly removed
sperm from the mature gonads of at least two males. Ac-
tivated oocytes were obtained by spreading freshly re-
moved female gonads (including gonoducts) in a petri
dish filled with 5 ml of 1-methyladenine 1(T 5 M, yielding
naturally spawned oocytes after about 45 min of exposure.
The buoyancy, capacity for adherence to the substrate,
and motility of the sperm were used to describe its be-
havior in seawater. All experiments on sperm behavior
were conducted at a temperature close to the one recorded
during the spawning period (2-4C). All live observations
of sperm samples were made under a light microscope
( 100-400X) and the sperm was kept cool by surrounding
the slide with ice cubes. The speed of the spermatozoa
was evaluated with a graduated lens, by calculating the
number of bars traveled per second. Verifications were
undertaken to discard any behavior induced by the light
of the microscope or false sperm velocity induced by the
sperm-glass interaction (thigmotaxis).

For more than 7 h we continuously recorded the speed,
orientation, and flagellar activity of sperm before and after
the samples were diluted in seawater ( 1 50 /jl of dry sperm
in 1 I). To examine the behavior of sperm in the water

column and on the bottom, the protocol was carried out
in both still and agitated (current of 7cm-s~') water.
Samples were taken from the bottom of the beaker and
at a middle depth every minute for 15 min, then every
1 5 min for 7 h, and finally at 24-h intervals until the sper-
matozoa were dead. This protocol, which was repeated
twice using repetitive independent measurements, also
allowed the estimation of the sperm concentration, with
the proportion of resting and agglomerated sperm in the
water and on the bottom over time. To estimate the
maintenance of sperm potency, samples of sperm depos-
ited on the bottom of the beaker were collected just after
the sperm was released in the seawater and regularly over
72 h. The samples were tested on two replicates of 5-10
freshly activated oocytes. Fertilization success was deter-
mined by staining the eggs with the DNA-specific flu-
orescent dye Hoechst 33258. Using a Leitz Diaplan flu-
orescence microscope, we determined the proportion of
eggs showing a male pronucleus.

After spermatozoa reached a state of low activity (>80%
barely moved) in the beaker of the above experiment, we
tried to stimulate the sperm by exposing it to oocytes. In
separate trials, a sample of sperm collected on the bottom
of the beaker was exposed to oocytes of different levels of
maturity and of different origins. The conspecific previ-
tellogenic and mature oocytes, classified according to the
studies of Boivin ft til. (1986) and Mercier ct ul. (1994),
were first assayed. In the case of mature oocytes we tried
both activated oocytes (showing germinal vesicle break-
down) and surgically removed unactivated ones. The ac-
tivated oocytes of Asterias vulgaris. another species of
starfish, were also tested for stimulation of Leptasterias
polaris sperm. The oocytes of both species were routinely
washed in filtered seawater and immediately used for the
reactivation experiments. The speed and the flagellar ac-
tivity of sperm were noted every 10 min for the first hour,
then periodically until no movement was detectable, using
replicates and going through the whole protocol twice. A
control sample with no oocytes was tested in the same
manner.

To investigate sperm sinking, dry sperm was diluted in
seawater ( 1:60) to a concentration of 120 X 10 3 sperma-
tozoa-mi '. Five homogeneous replicates (10/^1) of this
solution were prepared from eight males. The samples
were deposited at a middle depth in a 1000-ml beaker
filled with seawater. and the time needed for about 50%
of the sperm (in visible filaments) to reach the bottom
was recorded. These times were compared to those of
sperm collected from three species in which fertilization
occurs in the water column, the starfish Asterias vulgaris.
the sea urchin Strongylocentrotus droebachiensis, and the
sea cucumber Cucumaria frondosa.

To examine sperm dispersion over time, we deposited
dry sperm on the bottom of a large dish filled with 1000 ml

REPRODUCTIVE BEHAVIOR OF LEPTASTER1AS POLARIS

35

18"

16-

14

12

ID-
S'
6
4
2"

Prespawning

Spawning

Developmental biology

Temperature (C) Metamorphosis

O

N
1992

D

M A

1993

M

Figure 2. Variation of environmental factors during the reproductive season of Leptasterias polaris in
the experimental tanks. The arrow points to the beginning of embryo metamorphosis.

of seawater and periodically collected water samples at
the surface and on the sides of the dish. The spermatozoan
concentrations of the samples were evaluated with a he-
macytometer under a light microscope. This experiment
was simultaneously performed on live and dead dry sperm
of Leptasterias polaris and on dry sperm from Asterias
vulgaris. Strongylocentrotus droebachiensis, and Cucu-
nuiria fmndosa. Sperm longevity could thus be compared
for those species under the same conditions.

Development

Whenever we discovered a naturally spawned egg mass,
whether brooded or not, it was left undisturbed in the
tank so that we could examine development under natural
variations of environmental factors. Nonbrooded egg
masses were kept clean by periodically agitating the water.
Samples were regularly collected with pipettes, from fer-
tilization to young starfish stage, and transferred to 4%
formaldehyde/seawater for later examination with light
microscopy. These embryos also served to determine de-
velopmental kinetics and growth. During the first hour,
samples were collected every 2-5 min, then about every
day until the brachiolaria stage, and finally once a week.
A new stage was considered attained when 50-60% of the
embryos reached it. Maximum embryo diameters were
measured under a light microscope equipped with a grad-
uated lens.

Results

Prespawning behavior

During summer and early fall, the well-fed starfish
clearly avoided each other. Contacts among starfish began
in mid-October, coinciding with the first significant de-
crease of temperature (Fig. 2). The proportion of contact-
free starfish decreased, reaching a minimum plateau be-
tween 1 November and 15 December (Fig. 3) when tem-
peratures fell to about 4C. In that same interval, the
proportion of starfish involved in intimate contacts in-
creased progressively, with the maximum recorded in the
last week of November. Superposition became more fre-
quent in early December and was observed most often
just before the main spawning period of late December
(Figs. 1,3) when the water temperature fluctuated between
3 and 4C. When the temperature fell below 3C (end
of December), the number of superpositions and intimate
contacts decreased while spawning was recorded (Fig. 3).
Superposition behavior ceased after January and most in-
dividuals resumed avoiding one another (Fig. 3), with
contact-free starfish accounting for more than 85% of the
observed individuals. Only some light contacts and a few
intimate ones were observed. No particular sex-specific
patterns were noticed for the aggregations, which involved
both males and females. Addition of prey to the experi-
mental tanks always provoked a migration of the starfish
toward the food source, except 1 week before spawning

36

J.-F. HAMEL AND A. MERCIER

100

u

o

40"

20

Spawning

Types of contact

B Superposition

Q Intimate

B Light

D No contact

N

D

1992

1993

Figure 3. Leptasterias polaris. Temporal evolution of the prespawning aggregative behavior recorded
several times a week for a 3-month period. For each date, the percentage of individuals involved in a
particular type of contact was recorded.

when feeding did not disturb the contact behavior between
the individuals.

Environmental factors seem to be involved in the ini-
tiation and development of the prespawning aggregative
behaviors. No aggregation occurred among individuals
maintained at constant temperature and photoperiod, but
grouping did take place among individuals kept in total
darkness with natural temperature.

Spawning

Our experimental information includes actual obser-
vations of spawning events in 4 females and 3 males and
additional indications from sperm agglutinates on many
males and on the substrate. We also observed more than
20 brooding females, which were always discovered within
24 h of spawning. We examined the correlations of all
our observations with environmental factors (Fig. 2),
which were similar in the laboratory and in the field.
Spawning occurred in our tanks from 1 9 December to 1 2
January, which roughly corresponded to the period when
we observed spawning in the field. During spawning
events, the starfish stayed close together, although there
was a net decrease in frequency of contact (Fig. 3). The
spawning individuals were not paired or superposed.

Spawning events. Figure 4 schematically illustrates the
spawning behavior we observed in the experimental tanks.
When a male spawned, it elevated the central disk, stand-

ing on the curved tip of its arms, and emitted sperm as a
whitish stream from the six interradial aboral gonopores.
Emission continued for more than an hour. Qualitative
observations showed the sperm to be negatively buoyant,
with a tendency to deposit on the substrate. The first fe-
male spawning was observed after male spawning. The
female remained flat on the substrate to release eggs while
its arms were extended, then progressively adopted the
characteristic brooding posture in "pinwheel" shape (Figs.
4, 5a, b). The average 300-500 spawned oocytes emerged
individually at a rate of about 1 oocyte every 2-5 s from
each of the six aboral gonopores located between the arms.
Almost all the spawning starfish observed were on a ver-
tical substrate (aquarium wall or rock face). The eggs had
a tendency to fall and were retained by the ambulacral
podia and the curved arms of the female (Fig. 5c). A num-
ber of eggs (possibly 50%) were, however, lost before the
end of a given spawning event. The final posture adopted
by the female was not always a perfect pinwheel shape
but seemed designed to cover the egg mass (around 20-
25 eggs -cm : ) in the best way possible. Consequently,
some brooding individuals were seen with two or three
arms extended somewhat to cover isolated groups of oo-
cytes.

Induction <>/ 'spawning. Although the first individuals
to spawn were males, spawning events were subsequently
recorded in both sexes alternately throughout the spawn-
ing period (Fig. 2). Nevertheless, some correlations con-

REPRODUCTIVE BEHAVIOR OF LEPTASTERIAS POLARIS

37

2 C

0Male spawning induced by
decreasing temperature

Active sperm in the
-^ water column (moving at

Female spawning stimulated
by sperm in the water column

Aggregation and settling
of sperm (resting state)

Spterm reactivated by
fresrrty-soawned oocytes
" rtilizationT?

Brooding and
embryonic
development

Figure 4. Schematic illustration of spawning behavior for male and female Lcplastvrias polans, showing
the relation between them.

cerning the induction of spawning could be made in the
course of our experiments. A few isolated male spawnings
occurred during minimum daylength (<9 h) when water
temperature was around 2-3C (first sight of sperm fil-
aments) in late December (Fig. 2), but most spawnings
were observed in January as the temperature fell further.
The temperature fluctuated between 2 and 4C (Fig. 2)
throughout the following weeks of spawning, and gamete
release seemed to be closely related to these variations.
The same spawning pattern was observed in control group
2, maintained in natural temperature and total darkness,
whereas no spawning occurred in control group 1, kept
at steady temperature and photoperiod. As a result of the
seasonally low primary production, the tanks provided
with seawater from the estuary, where spawnings were
recorded, contained virtually no phytoplankton. Salinity
continuously fluctuated between 26 and 32%o without any
consistent increasing or decreasing trends (data not pre-
sented). During qualitative observations, the presence of
sperm filaments seemed to be correlated with subsequent
female spawning within a few hours. Complementary ex-
periments conducted in replicates showed that the intro-
duction of sperm in the water induced spawning of several
females in the controlled environment (control group 1 ),
therefore minimizing the importance of temperature in
female spawning. No male spawning was induced by the
presence of sperm.

Sperm behavior

A microscopic examination of the sperm showed that
the head (more or less spherical) measured 3.25 0.25 ^m

and the flagellum 62 3 ^m. The negative buoyancy of
the male spawn caused it to sink (mainly as white fila-
ments), at a rate of about 2.1 mm-s" 1 . Only a small por-
tion was resuspended after reaching the bottom. We ex-
amined the motility of sperm artificially maintained in
the water column compared to the motility of sperm set-
tled on the bottom (Fig. 6). Freshly extracted dry sperm
contained nonmotile spermatozoa. Upon introduction to
the seawater, the spermatozoa immediately displayed a
major increase of activity, both in the water and on the
bottom (Fig. 6). Within the first 30 min of water contact,
100% of the spermatozoa had reached a velocity of 250-
350 p.m s~' and showed intense flagellar activity, resem-
bling a helical movement with 6-7 revolutions -s~'. The
spermatozoa maintained in suspension continued to show
the same high velocity and activity with no marked net
decrease for the whole 425 min of observation. In contrast,
after reaching a maximum velocity (after 40-50 min) in
synchrony with the sperm in suspension, the sperm on
the bottom showed an abrupt decrease of activity (reduced
velocity and flagellar movement). The settled spermatozoa
attained a low velocity (=s50Mm-s~') after 120 min of
contact with seawater (Fig. 6a). This corresponded to an
increase in the inactive population of spermatozoa, which
rose from 7% to 56% in the same period (Fig. 6a). The
spermatozoa seemed to gradually reach a state of almost
null velocity (about 0-15^m-s ') in which they only
quivered and moved by a wave along the flagellum (prox-
imal to distal) with a 10 angle. The percentage of settled
spermatozoa that attained a state of low activity was 47%

38

J.-F. HAMEL AND A. MERCIER

Figure 5. Underwater photographs showing (A) Li'ptasterias polari* brooding in its natural habitat
surrounded by sea urchins; (B) close-up of a brooding female on the rocky bottom at 30m depth: (C)
fertilized eggs under a female: (D) young starfish after about 5.5 months of growth in their natural habitat
(the female was previously removed). The scale bar represents 15mm and applies to photographs (C)
and (D).

after 120 min of contact with seawater, 62% after 250 min,
and a maximum of 80% after 380 min (Fig. 6a). On the
bottom and most probably on the glass walls, the increased
number of nearly inactive spermatozoa was responsible
for the decrease in overall velocity of the population. This
correlation could also be made for the sperm in the water
column, although the small apparent decrease in velocity
could only be visually associated with the slight increase
in inactive spermatozoa (Fig. 6b). Another progressive
phenomenon was the formation of sperm conglomerates
(dense aggregations), which began about 180 min after
the sperm came in contact with seawater and reached a
maximum (100 and more spermatozoa together) after
380 min. About 8 h after entry into still seawater, the
sperm covered the bottom and glass walls of the dish.
This also occurred in agitated conditions, although after
a longer period, showing that the sperm of Leptasterias
polaris is very adhesive.

The sperm behavior of Leptasterias polaris differed
from that observed in the other species of echinoderms

tested: the sperm of Asterias vulgaris was diluted within
2 min, before reaching bottom, and those of Cucumaria
Jrondosa and Strongylocentrotus droebachiensis sank at
1 mm-s~' and 1.5 mm-s~' respectively. The sperm of
these species did not adhere to the bottom, but was im-
mediately and almost totally redispersed, becoming well
dispersed in seawater within 80 min. In contrast, the ma-
jority of L. polaris sperm stayed on the bottom until death.
After reaching a state of almost null activity (conglom-
erated or not), the sperm of Leptasterias polaris could be
reactivated by contact with conspecific activated mature
oocytes (Fig. 7). Within 10 min of exposure to these oo-
cytes, sperm motility was reinitiated. Spermatozoa veloc-
ity increased significantly, by 485% (P < 0.01, Student's
/ test), after 20 min of contact, and attained a peak of
230 nm s~' after 50 min. This was almost 12 times the
original speed (Fig. 7). Subsequently, the sperm velocity
decreased progressively and reached a minimum after
1020 min (Fig. 7). Reactivation was unsuccessful with
unactivated conspecific mature and previtellogenic oo-

REPRODUCTIVE BEHAVIOR OF LEPTASTERIAS POLARIS

39

100 150 200 250 300 350 400 450

Time (min)

Figure 6. Leptasterias polaris Temporal changes in velocity and
activity of sperm settled on the bottom and kept in suspension. The
sperm velocity ( O ) upon contact with seawater is given with the
corresponding percentage of inactive spermatozoa ($) The error
bars represent the confidence intervals (95%).

cytes as well as with mature activated oocytes otAsterias
vulgaris. The spermatozoa exposed to those oocytes
showed a constant low velocity, comparable to that ob-
served in the unexposed sperm serving as control (Fig. 7).
The sperm of Leptasterias polaris was more resistant
than that of other species. Most spermatozoa collected
from other echinoderms were dead after 8-20 h, whereas
those of L. polaris remained capable of high fertilization
success (57%) for as long as 34 h in seawater (Fig. 8) and
showed a high percentage of mortality only after 6-7 days.

Development

Early development. The complete chronology of em-
bryonic development of Leptasterias polaris is presented
in Table I and Figure 9. The large unfertilized mature
oocytes (^0.85 mm in diameter) were mainly spherical
and yellowish, or occasionally light orange. They were
covered with a rather thick outer membrane (average of
7.04 ^m). After their fertilization, the eggs were attached
to one another by the fertilization membrane, showing
that the membrane was sticky, especially after reaching
the 2-cell stage (Fig. 9a). All the cleavages were of the
radial holoblastic type. Later in its development, from the
blastula to young gastrula, the embryo decreased in size

300 n

^ 250-

u

<u

s 20

Jl

>.

n iso

<u
Q.
CD

100

50

Control

Previiellogemc oocyks
UnacDvaled oocytes
Activated oocytes
Astenas oocytes

10

100

1000

Time (min)

Figure 7. Leptasterias polaris Influence ofoocyte maturity on sperm
reactivation. The horizontal axis has a log lo scale and the error bars
represent the confidence intervals (95%).

(zx6%) due to blastomere compaction (Table I). The fur-
rows became shallower on the pole, but tended to deepen
on the other pole, eventually forming the blastopore. After
7 1 1 h, the embryo's surface became uniform and ciliated.
The embryo then began to spin inside the fertilization
membrane and hatched 96 h later (Fig. 9e).

Many females moved away from their brood within a
week, leaving their embryos uncared for in the experi-
mental tank. No significant difference was observed be-

20

40 60

Time (h)

80

100

Figure 8. Leptaslerias polaris. Sperm effectiveness to fertilize acti-
vated mature oocytes over time. The error bars represent the confidence
intervals (95%).

40

J.-F. HAMEL AND A. MERC1ER

Table I

Leptasterias polans: A'/mY;o / cnihryaiuc i/c\-c/<>piiiciii in
experimental lank', supplied hy running Mwatcr

Developmental Stages

Time

Size (jim)

Spawning

Early stage of the fertilization

membrane elevation
Fertilization membrane completely

elevated

Emission of the first polar body
Second polar body
2-cell
4-cell
8-cell
16-cell
32-cell
64-cell
128-cell
256-cell

Blastula (compaction)
Wrinkled-blastula
Young gastrula
Late gastrula
Spinning
Hatching
Brachiolaria
Metamorphosis
Young starfish (2 pairs of ambulacra!

podia/arm)
Young starfish (preoral lobe

disappears and ocelli are present)
Free-moving starfish (visible pyloric

caeca and opening of buccal

cavity)
Small starfish (uprighting

movements)

852 36
840 38

27 min

940 48

45 min

439 26

45 h

1 079 [ 5

86 h

1032 15

92 h

1046 12

106 h (4d)

1031 18

121 h(5d)

1062 9

133 h (5-6d)

1103 62

146 h (6.1 d)

1080 23

156 h (6.5 d)

1 101 40

209 h(8-9d)

1 1 20 5 1

260 h (10-13 d)

1 144 45

493 h (20-21 d)

1056 31

666 h (27-28 d)

1288 52

711 h ( 29-30 d)

1375 27

807 h (33-34 d)

1121 42

38-84 d

1 1 90 3 1

75-40 d

1207 70

120-132 d
150-170 d

180-195 d
more than 200 d

1348 53
1534 77

2102 102
over 2500

A new stage was considered attained when 50'" -60^ of the embryos
reached it. The standard deviations about the mean size are given.

tween the developmental rate of brooded and nonbrooded
embryos during early development up to hatching (P
= 0.392, Student's / test), which is the latest brooded stage
we observed in the laboratory (Table II).

Lute development. After the loss of the fertilization
membrane upon hatching, the unciliated portion of the
late gastrula enabled it to attach to the substrate, well
before the appearance of the brachiolar arms (Fig. 9f).
The hatched larvae immediately settled on the bottom;
however, many embryos were lost by the female at this
time of development (especially for those brooding on
vertical surfaces). With the growth of the embryo, the arms
elongated, becoming very distinct from the dorsal ciliated
bulb (larval body), and served the purpose of adhesion to
the substrate. Cilia were still present, so the fixation was
mainly with the sticky ramified tips that had developed
at the end of each arm (Fig. 9g). A depression began to

grow in the central portion delimited by the arms (fixing
disk), and was used for later fixation on the substrate with
the brachiolar arms. The brachiolar stage was prolonged
as long as the water temperature remained around 1C
(February and March; Fig. 2), until a sudden warming
coincided with metamorphosis (day 75-90). Although
there was a concurrent elevated photoperiod, this factor
had been increasing for months and cannot be the decid-
ing inducer (Fig. 2). About 50% of the embryos died during
the gradual metamorphosis, which was completed around
mid-May following the complete disappearance of the
brachiolar arms (Fig. 9h-j). At this stage the characteristic
yellow color had been lost and the embryo was whitish
or translucent, indicating that a large amount of vitelline
reserves had been consumed. One month later (mid-June),
the young starfish possessed a well-developed buccal cav-
ity, stomach, and pyloric caeca, which were easily observed
across the transparent body wall on the oral surface (Fig.
91) with the madreporite and anus on the aboral surface.
The ambulacra! podia, baring suckers, became effective
in helping the uprighting movement and displacement of
the growing starfish, which were capable of coordinated
locomotion. Having the capacity to feed and move on
their own, the young were self-sufficient about 6 months
after fertilization (Figs. 5d, 9k, 1).

Discussion

Temporary aggregative behavior is common among
marine invertebrates. It has been observed in echinoderms
such as echinoids ( Pearse and Cameron, 1 99 1 ; Levitan et
til.. 1992: Young </<//.. 1992), ophiuroids( Warner. 1979)
and asteroids (Ormond et al.. 1973; Sloan, 1980, 1984;
Blankley and Branch, 1984; Run et al.. 1988) as well as
in crustaceans (Gherardi and Vannini, 1993). Those
studies have proposed many hypotheses about the mean-
ing and usefulness of aggregation, but only a few have
discussed a relationship with reproduction and spawning.

Most of the few reports of aggregation related to repro-
duction in echinoderms refer to aggregative spawning
events in the field (Hendler and Meyer. 1982; McEuen,
1988; Pearse et al.. 1988). Pseudocopulation or pairing
has been observed in Arc/Ulster typieiis (Ohshima and
Ikeda. 1934; Komatsu. 1983; Run et al.. 1988) and in
Neosmilaster georgianus (Slattery and Bosch, 1993), but
no such behavior could be detected during our study. Like
Chia (1968) in observations of Leptasteriii* lie.\actis. we
noticed that the groupings occurred only as the breeding
season approached. Grouping behavior initiated well be-
fore spawning, such as observed in L. polaris, has not
been often reported. Young et al. (1992) observed this
behavior in the bathyal sea urchin Stylocidaris lineata. in
which the individuals aggregate during autumn before
spawning. Orton (1914) and Lewis (1958) also mentioned

REPRODUCTIVE BEHAVIOR OF LEPTASTERIAS POLARIS

Figure 9. Lcplastcrias polaris. Photographic sequence showing the principal steps of development from
fertilization to young starfish (4X); see Table I for corresponding age and size of the embryos. (A) Fertilized
eggs with fertilization membrane completely elevated (arrows). (B) 4-cell stage. (C) 8-cell stage. (D) Wrinkled-
blastula stage on which it is possible to observe the furrows (arrows). (E) Early brachiolaria stage (newly
hatched) showing early shaping of brachiolar arms (ba) and the persistent blastopore (arrow). (F) Growth
of the three brachiolar arms (ba). (G) Fully developed brachiolarian embryo with the larval body (Ib) and
the slightly ramified brachiolar arms (ba). (H) Metamorphosing embryo with brachiolar arms (ba), showing
the development of the first five hydrocoelic lobes (hi). (I) Further metamorphosed embryo with two pairs
of ambulacral podia (p) and a terminal tentacle (t) on each of the five arms. The six hydrocoelic lobes are
transforming into the radial canals (re). A distinct oral disk (od) and the residual brachiolar arms (ba) with
central fixing disk (f) are visible. (J) Aboral view of a free-moving, six-rayed young starfish showing clearly
visible dorsal spines (ds), madreporite (m), and regressing preoral lobe (pi). (K) Aboral view of a small starfish
showing the well-developed terminal spines (ts). (L) Oral view of a small starfish ready to leave the brood.
The buccal cavity (be) has opened and the stomach (s) is visible surrounded by the ring canal (ri). We can
also see the amhulacral podia (p). the well-developed radial canals (re), and the ocelli (o).

42

J.-F. HAMEL AND A. MERCIER

the occurrence of this prespawning aggregative behavior
in echinoids.

What cues make the starfish come together and why
do they display this behavior? Pheromones have often
been proposed as the proximate cause (Kanatani and Shi-
rai, 1968), principally acting to synchronize the liberation
of gametes (Ormonde/ al., 1973; Young el al, 1992; Slat-
tery and Bosch, 1993). For broadcast spawners, aggrega-
tion and synchronous gamete release appear to be im-
portant in minimizing gamete dispersion, ensuring good
fertilization success (Levitan, 1988; Levitan el al, 1992).
For a protective brooder like Leptasterias polaris, syn-
chronous spawning would not be very advantageous as
the eggs laid by the female are maintained under the body
at all times. Therefore, aggregation is more likely to be
related to preparatory recognition. Contact chemorecep-
tion is suggested to be a strong sensory stimulus in aster-
oids (Sloan and Campbell, 1982), maybe to ensure ade-
quate recognition of males and females before spawning.
In Arcliaster typicus and Neosmilaster georgiamis, this
recognition is of prime importance because fertilization
is ensured by the close superposition of a male and a fe-
male (Run et al.. 1988; Slattery and Bosch, 1993). In L.
polaris, our results demonstrate a totally different pattern,
in which the prolonged intimate contacts could be related
to the chemical induction of synchronized gamete devel-
opment as demonstrated in sea cucumbers Citcumaria
frondosa (Hamel et al., unpub. manuscript).

The initiation of aggregative behaviors in Leptasterias
polaris appears to be correlated with decreasing temper-
ature. Aggregations were observed among all starfish that
were supplied with natural seawater and exposed to sea-
sonal changes of water temperature, both when main-
tained in darkness and when exposed to natural photo-
period. Lower temperatures possibly trigger or enable the
liberation of hormones through a pathway otherwise sup-
pressed and may favor the formation of clumps of starfish.
This would assure that a fairly good proportion of male
and female individuals would be close together and ready
to release their gametes during the winter spawning events.
We saw no evidence of recognition between sexes but can
assume that the male/female ratio close to equality ensures
a 50% chance of random heterosexual encounters. Young
ct al. ( 1992) found that individuals of Stylocidaris lineata
also aggregated without regard to sex. The spawning events
were not synchronized as the male began liberating
sperm before any female spawning could be detected
which differs from spawning sequences mentioned for
Hymenaster membranaceus (Pain et al., 1982) and for
Archasler typicus (Run et al., 1988). Male spawning
seemed to be triggered when the falling temperature
reached about 2C, an inducer previously suggested by
O'Bnen (1976) for L. linoralis.

The sperm behavior of Leptasterias polaris possibly
further explains the need for aggregation. Upon its release,
a portion of sperm can be dispersed by currents and re-
main active as long as it is maintained in the water column
(Fig. 6), potentially limiting the genetic isolation in a pop-
ulation. From the spawnings successfully induced by
sperm in our experimental tanks, we infer that this active
fraction is the stimulus for females to spawn. Sperm as a
stimulus of spawning has also been discussed by Starr et
al. (1990) for the sea urchin Strongylocentrotus droeba-
cliiensis. In fact, sperm suspension in seawater is suggested
to be the spawning inducer in many species of ophiuroids
and echinoids (Thorson, 1950; Lewis, 1958). The lack of
strong epidemic spawnings during natural breeding activ-
ities in our tanks could be explained by fluctuations of
water temperature around the 2C threshold at that time
(Fig. 2). A few male spawnings could have been triggered
in late December, then delayed by the rising temperature
(to almost 4C) before being induced again in the second
week of January. Females followed this scattered pattern
because they probably need to be close to a sperm source
for spawning to be induced. The negative buoyancy and
stickiness of sperm causes most spermatozoa to settle on
the substrate where they gradually enter an inactive state.
The settling ensures a minimal dispersion of sperm but
makes fertilization dependant on the proximity of indi-
viduals, which is achieved by aggregation. Further, sperm
inactivation seems an effective energy-saving behavior,
extending the viability of settled sperm up to 6 or 7 days,
which is much longer than the 2- to 3-day longevity of
sperm maintained in the water column. The extreme en-
durance of L. polaris sperm is further emphasised upon
comparison with that of the other asteroids, holothuroids,
and echinoids tested, for which the spermatozoa did not
survive longer than a day at the temperature normally
recorded during their spawning periods. These short-lived
spermatozoa display a dispersion behavior that we could
probably associate with organisms having synchronously
spawning males and females. The longevity of sperm from
L. polaris is probably an advantage given the asynchro-
nous spawning of the sexes in this species.

We could not determine whether females were attracted
by deposited sperm, but the delay between male and fe-
male spawnings seems advantageous. Because the sper-
matozoa are present on the medium before eggs are emit-
ted, they do not have to overcome the protective barrier
maintained by the brooder. Thanks to its adhesiveness,
sperm also covered the vertical substrates favored by many
spawning females in our experiments. It is probable that
as soon as a female spawns on the sperm-covered sub-
strate, the oocytes can reactivate the inactive sperm (Fig.
7), and fertilization takes place. Experimentally, the best
success (>75%) was achieved when the delay between the
male and female spawnings was no more than 1 1 h; how-

REPRODUCTIVE BEHAVIOR OF LEPTASTERIAS POLARIS

43

ever, success was still good (>50%) after as long as 30 h
(Fig. 8). A contact of 20-50 min with the oocyte seemed
to be necessary for the spermatozoan to attain an optimum
speed that probably maximizes its ability to fertilize.
Sperm inactivity and reactivation appears to be very rare
in marine habitats. Although sperm chemotaxis has been
shown in echinoderms (Miller, 1985), no significant ve-
locity increase or activation of the attracted sperm has
ever been mentioned, except in cnidarians (Miller, 1979a,
b) and larvaceans (Miller and King, 1983). The closest
example with a similarity to Leptasterias polaris is the
sperm of the horseshoe crab (Limulus polyphemus), which
undergoes a brief flurry of motility and remains nonmotile
until it encounters a sperm motility initiating molecule
(SMI) emanating from eggs (Clapper and Epel, 198 1 ). In
contrast, both our observations and previous studies (Chia,
1968) show that the sperm of most echinoderms becomes
active upon release in the water and disperses quickly.
This is probably the best strategy when both male and
female gametes are released in great numbers in the sea-
water at close intervals.

In Leptasterias polaris, the gamete behavior seems well
adapted to the brooding mode, which in turn has a pro-
tective function. Brooded and nonbrooded embryos
showed almost perfect synchrony in development through
the gastrula stage (Table II). This suggests the absence of
the obligatory exchange of nutrients between parent and
young that is seen in Pteraster militaris (McClary and
Mladenov, 1990), where a brood chamber is present. In
L. hexactis, another brooding asteroid overlaying its eggs,
the maternal presence is proposed to be essential to help
the hatching embryo tear the fertilization membrane
(Chia, 1966). Although this was thought to be the case
for L. polaris (Himmelman el ai, 1982), we observed no
evidence of that phenomenon. The hatching was not de-
layed and no loss of embryos was evident in the unbrooded
group. Brooding in L. polaris probably serves mainly to
aerate the embryos and prevent them from being covered
with sediment as they lie on the substrate. Observations
in the field (Himmelman el ai, 1982) support this hy-
pothesis, as the substrate under a brooding starfish was
always found clear of debris. Brood protection also seems
to be in direct relation to adverse environmental condi-
tions and predatory pressures. Extremely active grazers
such as sea urchins, which are abundant wherever L. po-
laris is found, would rapidly decimate any unprotected
starfish embryos exposed on a rock (Fig. 5a).

The embryonic development observed in Leptasterias
polaris is similar to that described for L. hexactis (Chia,
1968) and L. acceptances similispinis (Kubo, 1951). The
developmental kinetics of L. polaris is characteristically
slower (Table I) than in all other reports for this genus,
perhaps because of the lower temperatures (0- 1 C) found
during the breeding and the subsequent development of

Table II

Leptasterias polaris: Development of brooded and nonbrooded embryos

Nonbrooded

Brooded Embryos

Embryos

Developmental Stages

Time

Size i

:Mm)

Time

Sizei

,pm)

Fertilization

870

43

852

36

2-cell

43 h

1111

+ 22

46 h

1079

15

4-cell

81 h

1092

18

86 h

1032

15

8-cell

93 h

1032

9

92 h

1046

12

16-cell

104 h

1044

24

106 h

1032

18

32-cell

125 h

1050

33

121 h

1062

9

64-cell

140 h

1082

39

133 h

1103

62

128-cell

152 h

1086

41

146 h

1080

3

256-cell

164 h

1092

52

156 h

1101

40

Blastula

8-9 d

1090

62

8-9 d

1120

51

Gastrula

20-21 d

1288

52

20-21 d

mi

33

Hatching

33-34 d

1121

42

32-35 d

1199

63

A new stage was considered attained when 50%-60% of the embryos
reached it. The standard deviations about the mean size are given.

this species. The major differences between our results
and those of Kubo (1951) and Chia (1968) are the oc-
currence of a spinning stage and the much earlier hatching
of L. polaris. Because L. polaris embryos hatch in late
gastrula, before they develop brachiolar arms, such arms
cannot contribute to the tearing of the fertilization mem-
brane, as they are said to do in the two other species.

The freshly spawned eggs are negatively buoyant and
do not adhere to one another until a few moments later,
after fertilization. This stickiness was also observed by
Chia ( 1968) for Leptasterias hexactis, but was correlated
with a reaction to seawater rather than with fertilization.
Through its growth, the embryo undergoes many changes
in attachment capacity, which is first provided by the
sticky fertilization membrane, then by small unciliated
body areas after hatching, and later by the fixing disk and
ramified tips of the brachiolar arms. The parental protec-
tion is probably useful in preventing dispersion of embryos
during these changes in fixation ability, for instance during
hatching, when attachment to the substrate may be inef-
fective for a short time. After the metamorphosis of the
embryos (4-5 months), brooding individuals are still ob-
served in the field for at least one month. The free-moving
young starfish seem to remain under protection through
the development of the pyloric caeca and the opening of
the mouth. Environmental factors apparently play a role
in the development of the embryos, especially in initiating
metamorphosis by a considerable increase in temperature
(Fig. 2). As previously mentioned by Boivin el at. (1986),
this correlation seems to ensure that the young starfish
are ready for release at the proper time, namely spring,
when conditions are most favorable for their survival as

44

J.-F. HAMEL AND A. MERCIER

self-sufficient individuals. This timing control, together
with the possibly lower energetic cost required from the
parent under cold temperatures, is probably an advantage
of winter brooding.

Acknowledgments

We thank Dr. C. Bouland for her help during the ex-
periments, A. Caron for helpful criticism and assistance
with statistical analysis, and J.-L. Theberges for photo-
processing. We greatly appreciated the helpful comments
of Drs. S. Demers, J. H. Himmelman, C. Bouland, M. I.
El-Sabh, and two anonymous reviewers on the manu-
script. We are also grateful to J. Noel for her collaboration
on the sketches and to N. Piche for photographs of Lep-
tasterias and field observations. This work was carried
out thanks to the space and material lent to us by Drs. E.
Pelletier and F. Dube at the Station Aquicole de Pointe-
au-Pere and was supported by personal funds of J.-F. Ha-
mel and A. Mercier.

Literature Cited

Blankley, W. O., and G. M. Branch. 1984. Co-operative prey capture
and unusual brooding habits of Anaxlt'nax ntpicola (Verrill) (Aster-
oidea) at sub-Antarctic Marion Island. Mar Ecoi Prog. Set: 20: 171-
176.

Boivin, V., D. Larrivee, and J. H. Ilimmclman. 1986. Reproductive
cycle of the subarctic brooding asteroid Leplasterias polaris. Mar.
Biol. 92: 329-337.

Chia, F.-S. 1966. Brooding ofa six-rayed starfish, Leptasterias liexuclis.
Bml Bull 130: 304-305.

C'hia, F.-S. 1968. The embryology ofa brooding starfish, Leplasterias
hexactis (Stimpson). Ada Zool. Bd XLI.X; 1-44.

Chia, F.-S., and C. \V. Walker. 1991. Echinodermata: Asteroidae. Pp.
301-353 in Reproduction <>/ Marine Invertebrates A. C. Giese. J. S.
Pearse, and V. B. Pearse, eds. Boxwood Press, Pacific Grove. Cali-
fornia.

Clapper, D. L., and D. Epel. 1981 . Isolation and utilization ofa sperm
motility initiating peptide in the horseshoe crab Limitlux polyp/tennis:
evidence for involvement of Ca 2+ but not intracellular pH or mem-
brane potential in motility initiation. / Cell Biol 91: 179.

Emerson, C. J. 1977. Larval development of the sea star. Leplaslerias
polaris. with particular reference to the optic cushion and ocelli. Sean.
Elec Micro 2: 631-638.

Giese, A. C"., and H. kanatani. 1987. Maturation and spawning. Pp.
251-313 in Reproduction of Marine Invertebrates, Vol. 9, A. C. Giese,
J. S. Pearse, and V. B. Pearse, eds. Blackwell Scientific Publications,
Palo Alto, California.

Giese, A. C., and J. S. Pearse. 1974. Introduction: general principles.
Pp. 1 -49 in Reproduction o/ Marine Invertebrates, Vol. I , Acoeloinate
and Pseudocoelomate \lcta:oans A. C. Giese and J. S. Pearse. eds.
Academic Press, New York.

Gherardi, F., and M. Yannini. 1993. Hermit crabs in a mangrove
swamp: proximate and ultimate factors in the clustering ofClihuiiariit?,
liievinumus J Exp Mar Biol Ecol 168: 167-187.

)l. mil, i. G., and D. I.. Meyer. 1982. Ophiuroids //<a;n;/;/<' de/iclo and
notes on the spawning behavior ol'other cchmoderms in their natural
habitat. Bull Mar Set 32: 600-607.

Himmelman, .1. II. 1981. Synchronization ol spawning in marine in-
vertebrates by phytoplankton. Pp. 3-19 in Advances in Invertebrate

Reproduction W. H. Clark, Jr. and T. S. Adams, eds. Elsevier North
Holland. New York.

Mimmelman, J. H., and C. Dutil. 1991 . Distribution, population struc-
ture and feeding of subtidal seastars in the northern Gulf of St. Law-
rence. Mar Ecu!. Prog. Ser 76: 61-72.

Himmelman, J. H., Y. Lavergne, A. Cardinal, G. Martel, and P. Jalbert.
1982. Brooding behaviour of the northern sea star Leptasterias po-
lans Mm Biol 68: 235-240.

Hyman, 1,. H. 1955. The Invertebrates: Echinodermata. McGraw-Hill.
New York, 763 pp.

Kanatani, H., and II. Shirai. 1968. Does GSS or MIS act as a phero-
mone? Problems concerning the participation of a pheromone in
starfish spawning. 7.ool. Mag. 77: 207-212.

Komatsu, M. 1983. Development of the sea-star, Areha.sler tvpicus,
with a note on male-on-female superposition, Annot. Zool. JPN 56:
187-195.

Kubo, K. 1951. Some observations on the development of the sea-star,
Leptasterias acceptances similispinis (Clarck). J. Fac. Set. Hokkaido
L'ntv. Ser 17. Zoo/. 10:97-105.

Levitan, D. R. 1988. Asynchronous spawning and aggregative behavior
in the sea urchin Diadema annllanim (Philippi). Pp. 181-186 in
Echinoderm Biology. R. D. Burke. P. V. Mladenov, P. Lambert, and
R. L. Parsley, eds. Balkema, Rotterdam.

Li-iitan, D. R. 1991. Influence of body size and population density on
fertilization success and reproductive output in a free-spawning in-
vertebrate. Biol Bull 181: 261-268.

Levitan, D. R., M. A. Sewell, and F.-S. Chia. 1992. How distribution
and abundance influence fertilization success in the sea urchin Stron-
gylocentrotus franciscanus. Ecology 73: 248-254.

Lewis, J. B. 1958. The biology of the tropical sea urchin Tripneustes
escnlentits Leske in Barbabos, West Indies. Can. J. Zool. 36: 607-
621.

Mi-Clary, I). J., and P. V. Mladenov. 1990. Brooding biology of the
sea star Plerasler miltlaris (O. F. Miiller): energetic and histological
evidence for nutrient translocation to brooded juveniles. J. E\p. Mar.
Biol Ecol 142: 183-199.

McEuen, F. S. 1988. Spawning behaviors of northeast Pacific sea cu-
cumbers (Hulothuroidea: Echinodermata). Mar Biol 98: 565-585.

Mercier, A., F. Pelletier, and J.-F. I hum I 1994. Metabolism and subtle
toxic effects of butyltin compounds in starfish. Aquat. Toxiail. 28:
259-273.

Miller, R. L. 1979a. Sperm chemotaxis in the Hydromedusae. I. Spe-
cies-specificity and sperm behavior. Alar. Biol. 53: 99- 1 1 4.

Miller, R. L. I979b. Sperm chemotaxis in the Hydromedusae. II. Some
chemical properties of sperm attractants. Mar. Biol. 53: 115-1 24.

Miller, R. L. 1985. Demonstration of sperm chemotaxis in Echino-
dermata: Asteroidea, Holothuroidea, Ophiuroidea. J. Exp. Zool 234:
383-414.

Miller, R. I.. 1989. Evidence for the presence of sexual pheromones
in free-spawning starfish. / Exp. Mar Biol Ecol 130:205-221.

Miller, R. I... and K. King. 1983. Sperm chemotaxis in Oikupteura
t/ioica (Urochordata: Larvaceae). Biol. Bull 165:419-428.

Minchin, I). 1987. Sea-water temperature and spawning behaviour in
the seastar Marlhaslenas glacialis Mar. Biol 95: 139-143.

O'Brien, F. \. 1976. Some adaptations of the seastar. Leplaslerias
(Stimpson) to life in the intertidal zone. Thalassia Jugosl 12: 237-
243.

Ohshima, H., and H. Ikeda. 1934. Male-female superposition of the
sea-star Archaster typicus Miiller etTroschel. Zool Lab Kyushu Im-
perial r/iMvrw/r54: 125-128.

Ormond, R. I . G., A. C. Campbell, S. H. Head, R. J. Moore, P. R.
Rninbim, and A. P. Saunders. 1973. Formation and breakdown of
aggregations of the crown-of-thorns starfish, Acanthasler planci (L.).
Nature 246: 167-169.

REPRODUCTIVE BEHAVIOR OF LEPTASTERIAS POLARIS

45

Orton, J. H. 1914. On the breeding habit of Echinus miliaris, with
note on the feeding habits of Patella vulgata. J Mar Bin/. Ass.
T A 10:254-257.

Pain, S. L., P. A. Tyler, and J. D. Gage. 1982. The reproductive biology
of Hymenaster membranaceus from the Rockall Trough, North East
Atlantic Ocean, with notes on // gennaeus. Mar. Biol 70: 41-50.

Pearse, J. S., and R. A. Cameron. 1991. Echinodermata: Echinoidea.
Pp. 5 1 3-662 in Reproduction of Marine Invertebrates: Lophophomtes,
Echinoderms Vol. VI, A. C. Giese. J. S. Pearse, and V. B. Pearse,
eds. Boxwood Press. Pacific Grove. California.

Pearse, J. S., and C. \V. Walker. 1986. Photoperiodic regulation of
gametogenesis in a North Atlantic sea star Asterias vulgaris Im J
Invert Reprod Develop. 9: 71-77.

Pearse, J. S., D. J. Eernisse, V. B. Pearse. and K. A. Brauchamp.
1986. Photoperiodic regulation of gametogenesis in sea stars, with
evidence for an annual calendar independent of fixed daylength. Am.
Zool. 26:417-431.

Pearse, J. S., D. J. McClary, M. A. Sewell, W. C. Austin, A. Perez-
Ruzafa, and M. Byrne. 1988. Simultaneous spawning of six species
of echinoderms in Barkley Sound, British Columbia. Inl J Invert.
Repmd Develop. 14: 279-288.

Pennington, J. T. 1985. The ecology of fertilization of echinoid eggs:
the consequences of sperm dilution, adult aggregation and synchro-
nous spawning. Biol. Butt. 169:417-430.

Run, J.-Q., C.-P. Chen, K.-H. Chang, and F.-S. Chia. 1988. Mating
behaviour and reproductive cycle ofArchaster typicus (Echinoderm-
ata: Asteroidea). Afar Biol. 99: 247-253.

Slattery, M., and I. Bosch. 1993. Mating behavior of a brooding Ant-
arctic asteroid, Neosmilaster georgianitx. Int. J Invert Reprod. De-
velop. 24: 97-102.

Sloan, N. A. 1980. Aspects of the feeding biology of asteroids. Oceanogr.

Mar. Biol Ann Rev 18: 57-124.
Sloan, N. A. 1984. Interference and aggregation: close encounters of

the starfish kind. Ophelia 23: 23-31.
Sloan, N. A., and A. C. Campbell. 1982. Perception of food. Pp 3-23

in Kchinoderm Nutrition M. Jangoux and J. M. Lawrence, eds.

A. A. Balkema, Rotterdam.

Starr, M., J. H. Himmelman, and J.-C. Therriault. 1990. Direct cou-
pling of marine invertebrate spawning with phytoplankton blooms.
Science 247: 1071-1074.

Strathmann, R. R. 1987. Evolutionary aspects of marine invertebrate
reproduction. Pp. 609-665 in Reproduction of Marine Invertebrates
Vol. 9. A. C. Giese, J. S. Pearse, and V. B. Pearse, eds. Blackwell
Scientific Publ., Palo Alto, and Boxwood Press, Pacific Grove, Cal-
ifornia.

Suzuki, N., M. Hoshi, K. Nomura, and S. Isaka. 1982. Respiratory
stimulation of sea urchin spermatozoa by egg extracts, egg jelly extracts
and egg jelly peptides from various species of sea urchins: taxonomical
significance. Comp. Binchem. Phy.iiol. 72: 489-495.

Thorson, G. 1950. Reproductive and larval ecology of marine bottom
invertebrates. Biol Rev 25: 387-397.

Warner, G. F. 1979. Aggregation in echinoderms. Pp. 375-396 in Bi-
ology and Systematic! of Colonial Organisms. G. Larwood and

B. R. Rosen, eds. Academic Press, London.

Young, C. M., P. A. Tyler, J. L. Cameron, and S. G. Rumrill.

1992. Seasonal breeding aggregations in low-density populations

of the bathyal echinoid Slylocidanf lineata. Mar. Biol 113: 603-

612.
Yund, P. O. 1990. An in \ilu measurement of sperm dispersal in a

colonial marine hydroid. J Exp. Zool. 253: 102-106.

Reference: Biol. Bull 188: 46-56. (February/March, 1995)

The Effects of Hydrodynamic Shear Stress on

Fertilization and Early Development of the Purple

Sea Urchin Strongylocentrotus purpumtus

KRISTINA S. MEAD AND MARK W. DENNY

Department of Biological Sciences, Stanford University, Hopkins Marine Station,
Pacific Grove, California 93950

Abstract. Life in the highly turbulent surf zone poses a
severe challenge to reproduction in free-spawning animals.
Not only can breaking waves quickly dilute the gametes
shed by spawning organisms, but turbulence-induced
shear stresses may limit fertilization and interfere with
normal development. A Couette cell was used to re-create
some of the effects of turbulent water motion to study
effects of environmentally relevant shear stresses on fer-
tilization in the purple sea urchin (Strongylocentrotus
purpuratiis). Although low shear stresses improved fertil-
ization success (presumably by increasing mixing), ex-
posure to high shear stresses (of the magnitude found in
the surf zone) substantially decreased fertilization success,
probably by interfering with contact between egg and
sperm. Furthermore, eggs fertilized at high shear stresses
often showed abnormal development and low survival of
eggs through the blastula stage.

Introduction

Like most intertidal invertebrates, Strongylocentrotus
purpuratiis. the purple sea urchin, reproduces by means
of external fertilization. S. purpuratiis is commonly found
on exposed intertida! rocky shores and in nearshore
benthic environments. In both of these habitats, its ga-
metes are subject to the energetic turbulent mixing char-
acteristic of these environments. Although a small amount
of turbulence can provide beneficial mixing, bringing eggs
and sperm into contact, the intense turbulence of the surf
zone or the benthic boundary layer may interfere with
fertilization through a variety of mechanisms. These in-
clude gamete dilution, gamete damage, decreased en-
Received 28 July 1994; accepted 28 November 1994.

counter rate between the egg and the sperm, and removal
of the sperm from the egg's surface before fertilization is
complete.

To date, the best studied of these mechanisms concerns
the effects of gamete dilution. The percentage of eggs fer-
tilized during spawning depends on the local concentra-
tion of viable sperm (Vogel et al., 1982, Levitan el a/.,
1 99 1 ). When subjected to the type of turbulent flows found
commonly in nearshore waters, sperm concentrations can
decrease by several orders of magnitude within a meter
of a spawning male (Denny, 1988; Denny and Shibata,
1 989; Grosberg, 1 99 1 ), resulting in low rates of fertilization
(Pennington, 1985; Levitan et al., 1992). The dilution of
sperm increases with an increase in distance downstream
from a spawning male and with an increase in current
velocity, resulting in a decrease in fertilization (Penning-
ton, 1985; Grosberg, 1987. 1991; Yund, 1990; Petersen,
1991; Petersen et al.. 1992; Levitan et at.. 1992; Oliver
and Babcock, 1992). The detrimental effects of sperm di-
lution can be partially offset through an increase in the
overall number, size, spatial density, synchronicity, and
fecundity of spawning animals (Denny and Shibata, 1989;
Levitan, 1991; Levitan et al.. 1992; Babcock and Mundy,
1992; Sewell and Levitan. 1992; Babcock et al., 1994).
Nonetheless, field studies have shown that the fraction of
eggs fertilized in the water column is typically low, often
less than 15% (Pennington, 1985; Levitan, 1991; Levitan
el al., 1992).

Under certain circumstances, however, gamete dilution
may be greatly reduced. Denny and coworkers (1992)
showed that the presence of surge channels can drastically
reduce the rate of dilution in trre surf zone of rocky shores,
to the extent that gamete concentrations could locally be
high enough to result in efficient fertilization. They tem-

46

TURBULENCE AND FERTILIZATION

47

pered this prediction, however, by noting that gamete di-
lution is only one of several factors affecting external fer-
tilization and suggested that the effects of small-scale hy-
drodynamics could impose additional limits to the efficacy
of this common form of reproduction.

Here we explore this possibility by recreating in the
laboratory one important aspect of turbulent flow (hy-
drodynamic shear stress) and measuring its effect on fer-
tilization and early development in the common purple
sea urchin, 5. purpwatus. In the next section, we describe
the eddy sizes, shear stresses, and energy dissipation rates
found in the surf zone. We then discuss how the average
values of these parameters make it possible to mimic some
aspects of turbulent flow in a rotating device called a
Couette cell. Finally, by examining fertilization under a
variety of conditions in the Couette cell, we show that
surf zone turbulence may have important consequences
for both the fertilization and the early development of sea
urchin embryos.

Theoretical Background

Surf-zone turbulence

When a wave breaks on a rocky shore, its orderly mo-
tion is disrupted as the waveform degenerates into a col-
lection of turbulent eddies (Denny, 1988). The interaction
of eddies causes the water to be sheared, and the friction
between layers of fluid (a result of viscosity) causes a shear
stress, T, to be exerted on the fluid and any gametes in it
(Okubo, 1978; Fischer, 1979). As an additional result of
friction, the wave's energy is lost to heat at a rate , called
the energy dissipation rate. Shear stress has the units
N/nr, the energy dissipation rate has the units W/m 3 ,
and for a given fluid, the two are related by viscosity:

* = T 2 /M, (1)

where n is the dynamic viscosity of seawater at 1 2C, here
taken to be 1.24 X 10~ 3 Pa -s (Denny et al., 1992). Shear
stress and energy dissipation rate are different descriptors
of the same phenomenon: the small-scale deformation of
fluid by turbulence.

Rates of energy dissipation, averaged over both time
and space, have been predicted for the surf zone (Thornton
and Guza, 1983; Svendsen, 1987; Denny et al., 1992;
George et al., 1994); increases with both wave height
and beach slope, with expected values ranging from 10-
3000 W/m 3 on wave-swept rocky shores (Fig. 1 ). The only
direct data for energy dissipation rates in the surf zone
come from hotfilm anemometer measurements of the
turbulence generated by small waves breaking on the
gently sloping beach at Scripps Pier (George et al., 1994).
For 1-m-high waves, a time-averaged value of t of about
10 W/m 3 was measured, with instantaneous values rang-
ing up to 100 W/m 3 .

3000

0.50 1.00 1.50 2.00

Wave Height (m)

Figure 1. Energy dissipation rates in the surf zone. Predicted values
given typical beach slopes and wave heights (Denny el al.. 1992).

The bent hie boundary layer

While the intense turbulence characteristic of the surf
zone is apparent to an observer, habitats occupied by 5.
purpuralits in the benthic boundary layer near shore also
experience high energy dissipation rates. In this case en-
ergy is dissipated as water is sheared against the solid sub-
stratum. The shear stress exerted on the substratum by
turbulent flow is pu\, where p is the density of the water
(about 1025 kg/m 3 for seawater) and /<*, an index of tur-
bulence intensity, is the friction velocity (Schlichting,
1979). From Equation 1, we see that this shear stress re-
sults in an energy dissipation rate of p 2 */M- The friction
velocity for flow over the rough substrata typical of rocky
shores is typically 5% of the mainstream velocity (Mid-
dleton and Southard, 1984), which in turn depends on
the height and period of the waves and the depth of the
water column. Using linear wave theory to estimate
mainstream water velocities near the substratum (Denny,
1988), it is thus possible to estimate energy dissipation
rates in near-shore benthic boundary layers for a typical
wave period of 10 s (Fig. 2). For the depths and wave
heights shown, peak wave-induced velocities vary from
0.02 to 1.35m/s, friction velocities from 0.001 to
0.067 m/s, and peak energy dissipation rates from 0.001
to 1 7,600 W/m 3 (resulting in average energy dissipation
rates of 0.0005 to 8,800 W/m 3 ).

Although energy dissipation rates predicted for the
benthic boundary layer can be quite high, their effect is
very localized. The thickness of the benthic boundary layer
under waves is approximately 0.41 u* 7/2 TT, where T is
the wave period (Grant and Madsen. 1986). Thus, for T

48

K. S. MEAD AND M W. DENNY

2000

10 m

0.50

1.00

1.50

2.00

Wave Height (m)

Figure 2. Energy dissipation rates in the benthic boundary layer.
Predicted values given typical wave heights and water depths of habitats
favored by Strongylocentrotus purpuratits. The wave period is assumed
to be 10 s.

= 10 s and the values of u* assumed here, the boundary
layer is expected to be only 0.7 mm to 4.2 cm thick. This
narrow localization of high energy dissipation rates in the
benthic boundary layer is in contrast to the situation in
the surf zone, where the entire volume encompassed by
the breaking wave is subject to high energy dissipation
rates.

In either case, once gametes or embryos leave the surf
zone or the benthic boundary layer, they are no longer
subject to high energy dissipation rates. Most other marine
habitats have turbulent energy dissipation rates that are
several orders of magnitude smaller than those found in
the surf zone and in the benthic boundary layer. For ex-
ample, energy dissipation rates in oceanic waters typically
vary from 10^ 7 to 10"' W/m 3 (Belyaev rta/., 1975; Dillon
etal.. 1981;Kitaigorodskii etal.. 1983;Osborn andLueck,
1985; Baker and Gibson, 1987), and energy dissipation
rates in tidally mixed areas vary from 10~ 2 to 10"' W/m 3
(Rothschild and Osborn, 1988).

Eddy size

To visualize how eggs and sperm respond to turbulent
flow and understand how it is possible to re-create some
of the effects of oceanic turbulence in the laboratory, it is
useful to consider the size of the smallest vortices, or ed-
dies, in a flow. This length, called the Kolmogorov length,
is

(2)

(Tennekes and Lumley, 1972) where i> is the fluid's kine-
matic viscosity (v = n/p, approximately 1.2 X 10~ 6 m 2 /s
for seawater at 12C). For example, using values for e
from 10 to 3000 W/m 3 , the Kolmogorov length in the
surf zone is predicted to be 22 to 5 /urn. Note that eddy
size decreases as the energy dissipation rate increases.

Few eddies have a diameter of less than 5-10 times the
Kolmogorov length, and the maximum shear energy den-
sity occurs in eddies about 40 times the size of the Kol-
mogorov length (Lazier and Mann, 1989; Osborn el a/.,
1 990). Therefore, the most energetically significant eddies
in the surf zone and benthic boundary layer are 200 to
880 ^m in diameter, and are larger still in less turbulent
flows. Most eddies are thus considerably bigger than a sea
urchin egg (80-1 10 nm) and sperm (head. 3 jum; flagel-
lum. 40-45 ,um). Within an eddy, the rotational nature
of the flow results in a velocity gradient (a shear) extending
from the center of the eddy to its periphery (Vogel, 1981).
It is reasonable to suppose, therefore, that the gametes
experience turbulent eddies as a temporally variable ve-
locity gradient with an associated shear stress and energy
dissipation rate (Denny ft ai. 1992).

Effects of shear

As yet, it has not been possible to directly observe eggs
and sperm in a velocity gradient of the sort found in the
surf zone. Nevertheless, general predictions about their
behavior can be made. Like all spherical objects in velocity
gradients, eggs are likely to tumble at many cycles per
second (Happel and Brenner. 1983; Kessler. 1986; Denny
et u/.. 1992), and the axis of rotation is likely to change
rapidly as different eddies shear. The rotation of the eggs
induces a secondary velocity gradient (a boundary layer)
around the egg. Although sperm are motile, their swim-
ming velocity (about 150-200 /um/s; Levitan etal., 1991)
is substantially lower than the small-scale velocities in-
duced by turbulence (approximately u*\ Denny, 1988).
Therefore, sperm are also expected to move according to
the local water motion. Due to the elongated shape and
flexibility of the sperm, however, their motion is expected
to differ from that of eggs, and it is possible that the effect
of a velocity gradient will be to align sperm with the di-
rection of flow. In the boundary layer surrounding the
egg, this alignment would cause sperm to move tangen-
tially to the egg's surface. It is therefore easy to imagine
how the induced motion of the egg and sperm might con-
spire to hinder contact between the gametes.

Shear stress in the laboratory: Taylor-Couette flow

A simple way to expose eggs and sperm to shear stress
(and thereby to mimic one aspect of turbulent flow) is to
place them in the well-defined velocity gradient of a
Couette cell (Coles, 1965; Donnelly, 1991). Couette cells

TURBULENCE AND FERTILIZATION

49

and similar instruments were recently used in several bio-
logical investigations. For example, Thomas and Gibson,
(1990a.b. 1992) used Couette cells to examine how tur-
bulent motion inhibits dinoflagellate growth, cell count,
and chlorophyll content, and Edwards el a/. (1989) used
a combination of Couette and cone and plate flows to
observe changes in cell length, septal length, hyphal di-
ameter, and branching frequency in two species of bac-
teria.

Materials and Methods

The Couette cell

The Couette cell consists of an inner stationary cylinder
and a rotating coaxial outer cylinder (Fig. 3). The inner
cylinder (50.5-mm outer radius) is constructed of stainless
steel with an acrylic plastic base. Cold water circulates
through the lumen of the inner cylinder to control the
temperature of the test solution, and an air line runs to a
small hole at the bottom of the inner cylinder, allowing
air to be bubbled into the test solution at a controlled
rate. The air bubbles keep the eggs from settling and get-
ting caught between the bases of the two cylinders. The
outer cylinder (54-mm inner radius) is made of clear
acrylic plastic and has a base that fits onto a motor shaft.
The distance between the two cylinders (3.5 mm) is more
than 30 times larger than the diameter of the 5. piirjniratiis
egg. The cylinders are 20 cm long, allowing an experi-
mental volume of more than 200 ml.

When the motor turns, the outer cylinder rotates,
shearing the liquid between the two cylinders. By treating
the cylinders as two wide plates a small distance apart,
the shear stress T in the test solution is calculated to be

(3)

where u> is the angular velocity (radians/s) of the outer
cylinder and r is its inner radius (here 54 mm), n is the
dynamic viscosity of the test solution, and /; is the radial
distance between cylinders (here 3.5 mm). Equivalently,
the flow inside the Couette cell can be described by the
energy dissipation rate t:

Air Inlet

Coolant Outlet

Coolant Inlet

ur\

~h

(4)

Note that Equations 3 and 4 hold whether flow in the cell
is laminar or turbulent (Schlichting, 1979). Both T and t
will be used to describe the flow inside the Couette cell.
By varying w (or ^) it is possible to recreate the shear
stresses (and energy dissipation rates) found in the surf
zone. In the experiments described here, filtered seawater
was used at 12C, and the dynamic viscosity n was taken
to be 1 .24 X 10~ 3 Pa s. The velocity of the outer cylinder

UL

22cm

=3

t Inner Cylinder

Outer Cylinder
( 108 mm m djameter)

! Working Space
(3 5 mm between cylinders,
100 ml working volume)

Mptor

.-L.l...^ Beanng

1

V ..III

1 , ^ ,

, b- =3 U

Figure 3. The Couette cell.

was measured by means of a magnetic pickup system.
Each time a tooth of a gear attached to the motor shaft
passed by the pickup, a current pulse was induced. The
pulses were amplified, counted, and converted into an-
gular velocities. The Couette cell was run at shear stresses
of 0.06 to 1 .45 Pa, corresponding to energy dissipation
rates of 2.8 to 1591 W/m 3 . These rates cover a large por-
tion of the range of turbulent energy dissipation rates ex-
pected on exposed rocky shores.

The Reynolds number describing the flow inside the
Couette cell is Re = pu>r/;//u- At the maximum angular
velocity used in this study (75.4 rad/s). Re = 11.600.
Dye studies indicated that flow became turbulent at
Re =a 4400.

Experimental design

S. piirpuratus males and females were induced to spawn
by injection with 0.5 A/KC1. Sperm were collected and
stored undiluted on ice. Eggs (jelly intact) were collected
in a beaker filled with filtered seawater, washed three times,
and diluted to a 0.5%- 1% suspension (by volume), which
was stirred gently and kept at 12C. Sperm concentrations
were determined by hemacytometer counts. All experi-
ments were performed within 8 h of gamete collection.
To determine the relative concentrations of eggs and
sperm to be used in each experiment, a standard fertil-
ization curve was created for each pair of urchins (Fig.
4). Small volumes of egg suspension were exposed to so-

50

K. S, MEAD AND M W DENNY

T3

S

U)
O)
O)
LU

O

c

Q)
0.

100

80

60

40

20

10 5

10 s

Sperm/mL

10'

Figure 4. Representative standard fertilization curve. Concentrations
and volumes of egg and sperm solutions giving 80%-90% fertilization
in a test tube were used in all experiments.

lutions of sperm for 2 min in gently stirred test tubes.
Fertilization was stopped by the addition of an equal vol-
ume of 0.5 Af KCI, which renders the sperm immobile
without harming the eggs (Schuel, 1 984). Concentrations
and relative volumes of eggs and sperm giving rise to 80%-
90% fertilization in a test tube were used in the Couette
cell experiments, to ensure that the decrease in fertilization
success expected as a result of shear stress would not be
concealed by an overabundance of sperm.

The rates of energy dissipation used in the experiments
ranged from 2.8 to 1 591 W/m\ In each experiment, 45 ml
of the egg suspension was put in the Couette cell. The
outer cylinder was brought up to speed over about 10 s.
after which 5 ml of newly diluted sperm was added to the
egg suspension. Fertilization was allowed to take place at
the specified energy dissipation rate for 2 min before the
reaction was stopped with 50 ml 0.5 M KCI. Eggs were
subsequently washed with filtered seawater and examined
under a microscope 4 h after fertilization. Because shear
stress can induce artificial activation and concomitant
formation of both the fertilization envelope and the hya-
line membrane (normally indicators of fertilization) in
the absence of sperm, only eggs that had divided were
counted as fertilized, possibly resulting in a slight under-
estimate of fertilization. Two hundred eggs were counted
per sample.

Each experiment was repeated between 3 and 16 times
at the same energy dissipation rates, using gametes from
different pairs of urchins. Because the relative concentra-

tion and the "fertilizability" of the gametes varied slightly
between pairs, all data were normalized to the percent
fertilization obtained when the experimental concentra-
tions of eggs and sperm were combined in a test tube in
the absence of appreciable shear.

Some urchins, such as those living on exposed rocky
shores, experience turbulence almost constantly. Other
animals, for instance those in tide pools, may experience
significant turbulence only as the waves break, or just
when the largest waves break on the shore. To approxi-
mate the time-dependent nature of this kind of environ-
mental turbulence more accurately, eggs were fertilized
under intermittent shear stress. As above, 45 ml of the
egg suspension was put in the Couette cell. Once the outer
cylinder had been brought up to speed, 5 ml of newly
diluted sperm was added to the egg suspension. Fertiliza-
tion was allowed to occur for 2 min, during which time
the Couette cell was alternately spun for 10 s, then allowed
to stop for 10s throughout the 2-min trial. Experiments
were repeated and data were normalized as above.

To determine if exposure to shear stress decreases fer-
tilization success by damaging gametes (as opposed to
some other mechanism, such as interfering with egg-sperm
binding), S pitrpnnitits eggs were sheared prior to fertil-
ization and then combined with unsheared, newly diluted
sperm in a test tube. Similarly, S. purpitralus sperm were
sheared prior to fertilization and immediately combined
with unsheared eggs in a test tube. Fertilization was
stopped after 2 min by the addition of an equal volume
of 0.5 M KCI. For comparison, eggs and sperm from the
same animals were fertilized under shear in the Couette
cell. All eggs were washed with filtered seawater and in-
cubated at 12C for 4 h before being examined. Experi-
ments were repeated three times and data were normalized
as above.

Eggs were reexamined after 24 h to determine whether
exposure to shear stress (either before or during fertiliza-
tion) had any effects extending past fertilization. Samples
were counted, and the percentage of fertilized eggs that
had developed into normal blastulae was recorded. Nor-
mal blastulae were characterized as clear, hollow balls of
cells spinning rapidly about their animal-vegetal pole axes.
Two hundred embryos were counted per sample.

Results

Fertilisation under constant shear stress

Water motion associated with low energy dissipation
rates (<70 W/m 3 ) enhanced fertilization success, pre-
sumably as a result of mixing. As the energy dissipation
rate increased from 2.8 to 69.2 W/m 3 , the mean fertiliza-
tion success increased from 78%- to 96%. Fertilization
success decreased when fertilization occurred during ex-
posure to moderate and high energy dissipation rates. As

TURBULENCE AND FERTILIZATION

51

100

T3

H
N

V)

O)
O)
UJ

C.

O

O

D.

50 75 100

500 1000 1500

Energy Dissipation Rate (W/m 3 )

Figure 5. The effect of shear stress on fertilization. Low energy dis-
sipation rates enhance fertilization success, whereas moderate and high
energy dissipation rates decrease fertilization success. Data from 16 pairs
of sea urchins. Error bars indicate the standard error.

the energy dissipation rate increased from 69.2 to
1 59 1 W/m\ the percentage of eggs that were fertilized
decreased from 96% to 19% (Fig. 5).

Intermittent shear stress

When eggs were fertilized under conditions of inter-
mittent shear stress, fertilization success decreased with
increasing energy dissipation rate, although not as dra-
matically as when the shear stress was constant. As the
maximum energy dissipation rate experienced during
the 10-s pulses of turbulence increased from 44.1 to
1 59 1 W/m\ the mean fertilization success decreased from
90% to 39% (Fig. 6). In comparison, when eggs and sperm
from the same pair of urchins were fertilized under con-
stant shear, mean fertilization success decreased from 92%
to 19%. At moderate energy dissipation rates (up to
397.7 W/m 3 ). there was no significant difference between
the effects of constant and intermittent shear stress. At
high energy dissipation rates (707 W/m 3 and above), eggs
fertilized under intermittent shear stress had greater fer-
tilization success than eggs fertilized under constant shear.

In this set of experiments, few measurements were made
at low energy dissipation rates. Therefore, the pattern seen
in constant shear stress of an increase in fertilization suc-
cess at low energy dissipation rates was not observed.

Shearing gametes before fertilization

When S. purpuratus eggs and sperm were sheared sep-
arately before fertilization and then combined with un-

"D

N

O)
O>
111

C

O

Q_

100

80

60

40

20

- Intermittent
A Constant

400 800

1200

1600

Energy Dissipation Rate (W/m 3 )

Figure 6. The effect of intermittent shear stress on fertilization. Data
are from three pairs of sea urchins. Error bars indicate standard error.

sheared gametes in a test tube in the absence of appre-
ciable shear stress, fertilization success was very high.
As the energy dissipation rate increased from 44. 1 to
1 59 1 W/m\ the fertilizability of the sheared eggs decreased
from 98% to 86%., and the fertilizability of the sheared
sperm decreased from 100%. to 92% (Fig. 7). In compar-

-o

in

O)

C

O

Q-

100

80

60

40

20

- - Sperm Sheared
-- Eggs Sheared
A Fertilized in Shear

400

800

1200

1600

Energy Dissipation Rate (W/m 3 )

Figure 7. The fertilizability of presheared gametes. Sperm and eggs
lose little fertilizability when sheared before fertilization. Data are from
three pairs of sea urchins. Error bars indicate standard error.

52

K. S. MEAD AND M. W. DENNY

ison, the fertilization success of eggs from the same female
fertilized under shear decreased from 94% to 8%.

Effect oj shear stress on early development

Although almost 100% of the 5". purpuratits eggs fertil-
ized in the absence of shear developed into normal blas-
tulae, many of the eggs fertilized while exposed to shear
stress showed the effects of shear-stress-induced damage;
typically their development was arrested at about the 64-
cell stage. As the energy dissipation rate during fertilization
increased from 44. 1 to 1591 W/m 3 , the percentage of fer-
tilized eggs that developed into normal blastulae decreased
from 92% to 22% (Fig. 8). Given that many eggs were not
fertilized, the overall fraction of eggs that developed nor-
mally was lower still. For example, as the energy dissi-
pation rate during fertilization increased from 44. 1 to
1591 W/m 1 , the percentage of eggs that developed into
normal blastulae decreased from 88% to 2%.

In comparison, almost all fertilized eggs developed
normally when the gametes were sheared separately and
then combined in a test tube (Fig. 9A). As the energy
dissipation rate increased from 44.1 to 1591 W/m 3 , the
percentage of sheared eggs that (once fertilized) developed
into normal blastulae decreased only from 99% to 89%>,
and the percentage of eggs fertilized by sheared sperm
that developed into normal blastulae decreased only from
96% to 94%;. These data can be graphed to reflect the total

100

CD

o.
o

<D
Q

15

100

80

60

40

20

- % Of Fertilized Eggs
A- % Of All Eggs

400

800

1200

1600

Energy Dissipation Rate (W/m 3 )

Figure 8. The effect of shear stress on development. Many eggs fer-
tilized in shear do not develop normally. Combining fertilization and
developmental data shows that the percentage of all eggs developing
normally at 24 h is very low. Data are from 12 pairs of urchins. Error
bars indicate standard error.

o

X
CO

<

O)

c

3

W

E

LLJ

O

X

o
<

W

<n

o>

D)
111

80 -

60

40

20

- - Sperm Sheared

-- Eggs Sheared

* Fertilization In Shear

400

800

1200
Energy Dissipation Rate (W/m 3 )

1600

100

80

60

40

20

- - Sperm Sheared
-- Eggs Sheared
A Fertilized In Shear

400

800

1200

1600

Energy Dissipation Rate (W/m 3 )

Figure 9. The effect of shearing gametes on early development. Al-
most all eggs fertilized in the absence of shear develop into normal blas-
tulae, but eggs fertilized in shear do not. (A) Development of fertilized
eggs. (B) Survival of all eggs to blastula, including eggs that were not
successfully fertilized. Data are from three pairs of urchins. Error bars
indicate standard error.

number of eggs that developed into normal blastulae, in-
cluding the effect of reduced fertilization (Fig. 9B). As
the energy dissipation rate increased from 44.1 to
1591 W/m 3 , the percentage of sheared eggs that developed
into normal blastulae decreased from 97% to 77%, and
the percentage of eggs fertilized by sheared sperm that
developed into normal blastulae decreased only from 96%
to 87%.

TURBULENCE AND FERTILIZATION

53

Discussion

How can low levels of turbulence increase fertilization
success?

Although intense turbulence limits fertilization success,
low energy dissipation rates enhance fertilization success.
This is presumably because turbulent mixing increases
contact rates between the egg and sperm. It is reasonable
to suppose that at low turbulent intensities the tumbling
of the egg in response to shifting velocity gradients may
not be rapid or abrupt enough to inhibit binding of the
sperm to the egg. Similarly, investigators have shown that
low energy dissipation rates increase rates of contact be-
tween predator and prey in the plankton (Rothschild and
Osborn, 1988; Marrase et al., 1990; Costello et at., 1990;
Saiz et al., 1992). At high energy dissipation rates, the
beneficial aspects of turbulence are outweighed by other
factors.

Why does excessive turbulence decrease fertilization
success?

Despite observations that turbulence decreases fertil-
ization success by diluting gamete concentration, the data
presented above indicate that environmentally relevant
turbulence can dramatically decrease fertilization success
even when eggs and sperm are in high concentrations.
There are several possible mechanisms for this shear-in-
duced decrease in fertilization success, including gamete
damage, a decrease in the encounter rate between egg and
sperm, and a removal of sperm from the egg's surface
during the latent period, the period after the sperm has
made initial contact with the egg, but before fertilization
is complete. The fact that shearing gametes before fertil-
ization does not dramatically reduce their fertilizability
indicates that exposure to shear does not substantially
damage gametes (Fig. 7). This suggests that high levels of
turbulence instead may limit fertilization by reducing the
effective encounter rate between gametes. If, as expected,
turbulence causes the eggs to tumble rapidly and the sperm
to align themselves in the direction of flow tangential to
the egg surface, small-scale turbulence could hinder con-
tact between the gametes. Alternatively (or additionally),
exposure to turbulence might limit the ability of the sperm
to attach to an egg once contact has been made, or it
might cause attached sperm to be removed before fertil-
ization is complete.

Our experiments provide no direct evidence that allows
us to choose among these possibilities, but indirect evi-
dence allows for reasonable speculation. We begin by ask-
ing whether it is likely that hydrodynamic shear stresses
are sufficient to tear an attached sperm from an egg. Al-
though no data are yet available for sea urchins, bond
strengths between a sperm and the zona pellucida of an

egg range in mammalian systems from 6 to 250 dyn, with
40 dyn being the most likely average value (Baltz el ui,
1988), and for the sake of argument we assume a similar
value for urchin sperm. This force can be compared to
the hydrodynamic force exerted on an urchin sperm as
follows: The shear force pulling at a sperm attached to an
egg is approximately equal to the hydrodynamic shear
stress to which the sperm is exposed multiplied by the
sperm's surface area. If a sea urchin sperm is modeled as
a 41-^m-long cylinder with a radius of 0.1 j/m (the fla-
gellum) attached to a conical head with a height of 3 /urn
and a radius of 0.5 ^m (Gray, 1955; Brokaw, 1965), its
surface area is 3 1 .4 /urn 2 . At a shear stress of 1 .45 Pa (the
maximum shear stress to which the gametes were exposed
in the Couette cell, equivalent to an energy dissipation
rate of 1591 W/m 3 ) the resulting shear force is 4.6 dyn,
much less than the expected bond strength. A still smaller
force would be exerted at lower shear stresses. It thus seems
unlikely that sperm, once well attached, are sheared off
before fertilization is complete. This suggests that the pri-
mary effect of hydrodynamic shear on fertilization is to
reduce the encounter rate between gametes, to disrupt
contact between sperm and egg before final sperm at-
tachment, or both.

This proposition is supported by our experiment re-
garding the "dose" of turbulence to which gametes are
subjected. A comparison of the effects of constant and
intermittent shear stress reveals a time-dependent re-
sponse. At high shear stresses, a smaller percentage of eggs
is fertilized if shear is constant than if shear is applied in
10-s periods alternating with 10-s periods of quiescence
(Fig. 6). In both cases gametes are exposed to the same
peak shear stress. If shear stress acted primarily by tearing
well-attached sperm from the egg before they could fer-
tilize (a process that takes about 20-30 s [Epel, 1989, 1990;
Ruiz-Bravo and Lennarz, 1989]), one would expect that
the effects of intermittent shear would be similar to those
of constant shear. If, on the other hand, the primary effect
of shear stress is to prevent sperm from encountering or
becoming strongly attached to eggs (processes that can
occur effectively in periods less than 10s [Epel, 1989,
1990; Ruiz-Bravo and Lennarz, 1989]), periods of qui-
escence would allow some sperm to encounter eggs and
attach well enough to be immune from later disruption
by shear stress. If this is indeed the case, fertilization rates
will be higher under intermittent shear, and this is the
effect that we observed. The time-dependence of the effect
of shear stress is consistent with the hypothesis that shear
stress limits fertilization by interfering with contact be-
tween the egg and sperm.

Further experiments will be necessary to test this spec-
ulation and to differentiate between the effects of turbu-
lence on encounter rates and the effects on subsequent
adhesion.

54

R. S. MEAD AND M. W. DENNY

Why does turbulence affect development'.'

Only a fraction of the eggs that are fertilized under tur-
bulent conditions develop into normal blastulae (Fig. 8),
indicating that shear stress, whether constant or inter-
mittent, affects reproductive success beyond fertilization.
Somehow, exposure to shear stress curtails or irrevocably
alters some process essential to normal development. Al-
though the mechanism or mechanisms behind this effect
are uncertain, reasonable speculation is again possible.
Hiramoto (1974) noted a brief fivefold stiffening of the
egg membrane 90 s after fertilization in the urchin Tem-
nopleurus toreumaticus. If S. purpiiratus eggs experience
a similar increase in stiffness, and if the hardened mem-
branes result in increased damage, eggs exposed to shear
stress during fertilization could sustain injuries not seen
when eggs were exposed to shear stress prior to fertiliza-
tion. These injuries could result in leakage of some essen-
tial cytoplasmic compound or in infection, either of which
could prevent the fertilized egg from completing more
than a few rounds of cell division.

Not only might different types or degrees of injury arise,
depending on whether the egg is exposed to turbulence
before or during fertilization, but the ability of the egg to
heal itself after injury might vary. Mechanically injured
cells heal themselves by releasing exocytotic vesicles,
which fuse to the plasma membrane (Steinhardt el a/.,
1994). Shortly after fertilization, these vesicles are depleted
(Vacquier, 198 1 ), presumably decreasing the cell's ability
to heal itself. This could help explain why a large fraction
of the eggs fertilized under turbulent conditions developed
abnormally.

Two further possibilities can be discounted. First, in-
juries resulting in parthenogenic stimulation or poly-
spermy cannot explain the effect of shear stress on de-
velopment, because the fertilized eggs usually divide to
about the 64-cell stage before degrading. In contrast, ar-
tificially activated sea urchin eggs tend not to divide, and
polyspermic eggs have very characteristic disruptions in
their cleavage patterns (e.g.. irregular or incomplete di-
vision, multiple asters, lobe formation), none of which
were observed here. Second, if exposure to shear stress
altered egg membranes in some way, eggs sheared prior
to fertilization would show the same defects in their de-
velopment, and this was not seen.

Ecological implications of the effect of turbulence
on fertilization

These experiments indicate that although low levels of
shear stress can aid mixing and fertilization, the intense
shear stress found in the surf zone and in the benthic
boundary layer can have severe mechanical effects on fer-
tilization and early development. These effects interact
with those of gamete dilution to narrow any window of

opportunity open to urchins attempting to reproduce by
external fertilization. For instance, although surge chan-
nels may restrict gamete dilution (Denny el al. 1992),
thereby promoting efficient fertilization, the deleterious
effects of shear stress may still limit the fraction of eggs
fertilized. Furthermore, any eggs that are fertilized are
very likely to develop abnormally. Gametes shed outside
of surge channels may be quickly swept out of the surf
zone, thereby avoiding prolonged exposure to high shear
stresses, but dilution will be rapid; consequently, the frac-
tion of eggs fertilized will again be low. Similarly, while
gametes released above the benthic boundary layer (by
large or aggregated individuals) experience lower energy
dissipation rates, they too are quickly diluted, and fertil-
ization rates will still be low.

The picture is not quite as bleak as it appears. Turbu-
lence is patchy, and even areas with high temporally av-
eraged rates of energy dissipation may experience (short)
periods of relative calm. Because of their great fecundity,
urchins able to time it right may end up with abundant
offspring.

There are plenty of sea urchins along our rocky shores,
but the chanciness of external fertilization in a turbulent
environment suggests that only a small number of them
contribute to the next generation. If this is true, then the
effective population is much smaller than the actual pop-
ulation a situation that could have populational, con-
servational, and evolutionary consequences (Quinn el al.
1993; Hedgecock, 1994).

Acknowledgments

We thank D. Epel for his constant moral and technical
support and for his excellent advice. Both he and Friday
Harbor Laboratories (University of Washington) provided
generous access to laboratory facilities. H. Crenshaw made
helpful suggestions, and B. Rees, P. Sund, J. Geller, and
B. Podolsky made insightful comments on earlier versions
ol this manuscript. Two anonymous reviewers had
thoughtful observations and helped clarify parts of the
paper. This study was funded by grants to K. Mead from
the Myers Oceanographic and Marine Biological Trust,
and by NSF grants to M. Denny (OCE-91 15688) and D.
Epel (IBN-9205393).

Literature Cited

Babcock, R. C., and C. N. Mundy. 1992. Reproductive biology,
spawning, and field fertilization rates of Acanthaster planci. Auxi J
Mar h'n'slnvalci- o 43: 525-534.

Babcock, R. C., C. N. Mundy. and D. \Vhitehead. 1994. Sperm diffusion
models and in \itii confirmation of long-distance fertilization in the
free-spawning asteroid Acanllmxlcr planci Biol Bull. 186: 17-28.

Baltz, J. M., D. F. Katz, and R. A. Cone. 1988. Mechanics of sperm-
egg interaction at the :nna pelluada. Buiphr.i. J 54: 643-654.

TURBULENCE AND FERTILIZATION

55

Baker, M. A., and C. H. Gibson. 1987. Sampling turbulence in the
stratified ocean: statistical consequences of strong intermittency. /
Phys Ocean 17: 1817-1856.

Belyaev, V. S., M. M. Lubimtzev, and R. V. Ozmidov. 1975. The rate
of dissipation of turbulent energy in the upper layer of the ocean.
J I'hy, Ocean 5: 499-505.

Brokaw, C. J. 1965. Non-sinusoidal bending waves of sperm flagella.
/ Exp. Biol. 43: 155-169.

Coles, D. 1965. Transition in circular Couette flow. J Fluid Mec/i 21:
385-425.

Costello, J. H., J. R. Strickler, C. Marrase, G. Trager, R. Zeller, and
A. J. Friese. 1990. Grazing in a turbulent environment: Behavioral
response of a calanoid copepod, Centmpages hamalus Proc Nail.
Acad. Sci. 87: 1648-1652.

Denny, M. W. 1988. Biology and the Mechanic* ol the W'are-Swept
Environment. Princeton LIniversity Press, Princeton, NJ. 329 pp.

Denny, M., and M. Shibata. 1989. Consequences of surf-zone turbu-
lence for settlement and external fertilization. Am. Nut. 134: 859-
889.

Denny, M., .1. Dairiki, and S. Distefano. 1992. Biological consequences
of topography on wave-swept rocky shores: I. Enhancement of ex-
ternal fertilization. Biol. Bull 183: 220-232.

Dillon, T. M., J. G. Richman, C. G. Hansen, and M. D. Pearson.
1981. Near-surface turbulence measurements in a lake. Nature 290:
390-392.

Donnelly, R. J. 1991. Taylor-Couette flow: The early days. Physics
Today 44: 32-39.

Edwards, N., S. Beeton, A. T. Bull, and ,J. C. Merchuk. 1989. A novel
device for the assessment of shear effects on suspended microbial
cultures. Appl Microhiol. Biotechnol. 30: 190-195.

Epel, D. 1989. Arousal of activity in sea urchin eggs at fertilization.
Chapter 1 5 in The Cell Biology oj Fertilization, H. Schatten and G.
Schatten, eds. Academic Press, San Diego. 404 pp.

Epel, D. 1990. The initiation of development at fertilization. Cell Differ.
Dev 29: 1-12.

Fischer, H. B. 1979. mixing in Inland and Coastal M'atenvays. Aca-
demic Press, New York. 483 pp.

George, R., R. E. Flick, and T. Guza. 1994. Observations of turbulence
in the surf zone. / Geophys. Re* 99: 801-810.

Grant, \V. D., and O. S. Madsen. 1986. The continental-shelf bottom
boundary layer. Ann Rev Fluid Mec/i 18: 265-305.

Gray, J. 1955. The movement of sea-urchin spermatozoa. / Exp. Biol.
32: 775-801.

Grosberg, R. K. 1987. Limited dispersal and proximity-dependent
mating success in the colonial ascidian Boiryllus schlosseri. Evolution
41: 372-384.

Grosberg, R. K. 1991. Sperm-mediated gene flow and the genetic
structure of a population of the colonial ascidian Botryllus schlosseri.
Evolution 45: 130-142.

Happel, J., and H. Brenner. 1983. Low Reynolds Number Hydrody-
namics. Martinus Nijhoff Publishers, Dordrecht. 553 pp.

Hedgecock, D. 1994. Does variance in reproductive success limit ef-
fective population sizes of marine organisms? Pp. 122-143 in Ge-
netics and Evolution of Aquatic Organisms, A. R. Beaumont, ed.
Chapman and Hall. London.

Hiramoto, Y. 1974. Mechanical properties of the surface of the sea
urchin egg at fertilization and during cleavage. Exp Cell Res 89:
320-326.

Kessler, J. O. 1986. The external dynamics of swimming micro-or-
ganisms. Pp. 257-307 in Progress in Phycological Research, Vol. 4,
F. E. Round and D. J. Chapman, eds. Biopress Ltd, Bristol. 48 1 pp.

Kitaigorodskii, S. A., M. A. Donelan, J. L. Lumley, and E. A. Terray.
1983. Wave-turbulence interactions in the upper ocean. Part II:
Statistical characteristics of wave and turbulent components of the

random velocity field in the marine surface layer. J 1'hys. Ocean
13: 1988-1999.

Lazier, J. R. N., and K. II. Mann. 1989. Turbulence and the diffusive
layers around small organisms. Deep-Sea Rex. 36: 1721-1733.

Levitan, D. R. 1991. Influence of body size and population density on
fertilization success and reproductive output in a free-spawning in-
vertebrate. Biol Bull 181:261-268.

Levitan, D. R., M. A. Sewell, and F.-S. Chia. 1991. Kinetics of fertil-
ization in the sea urchin Strongylocentrotus franciscanus: interaction
of gamete dilution, age, and contact time. Biol. Bull. 181: 371-378.

Levitan, D. R., M. A. Sewell, and F.-S. Chia. 1992. How distribution
and abundance influence fertilization success in the sea urchin
Strongylocentrotus Irunciscunus Ecology 73: 248-254.

Marrase, C., J. H. Costello, T. Granata, and J. R. Strickler.
1990. Grazing in a turbulent environment: energy dissipation, en-
counter rates, and efficacy of feeding currents in Ceniropage.i ha-
malus Proc Nail. Acad Set 87: 1653-1657.

Middlcton, G. V., and J. B. Southard. 1984. Mechanics ot Sediment
Movement. Society of Economic Paleontologists and Mineralogists,
Tulsa, Oklahoma. 401 pp.

Okubo, A. 1978. Horizontal dispersion and critical scales for phyto-
plankton patches. Pp. 1-42 in Spatial Pattern in Plankton Com-
munities, from the NATO Conference on Marine Biology. Erice,
Italy, 1977. J. H. Steele. ed. Plenum Press, New York. 470 pp.

Oliver, J., and R. Babcock. 1992. Aspects of the fertilization ecology
of broadcast spawning corals: sperm dilution effects and in situ mea-
surements of fertilization. Biol. Bull 183:409-417.

Osborn, T. R., and R. G. Lueck. 1 985. Turbulence measurements with
a submarine. / Phys. Ocean. 15: 1502-1520.

Osborn, T. R., H. Yamazaki, and K. Squires. 1990. Direct simulation
of the effect of turbulence on planktonic contact rates. Pp. 99-103
in I. urge Marine Ecosystems: Patterns, Processes, and Yields. K.
Sherman. L. M. Alexander, and B. D. Gold, eds. Publication 90-
30S of the American Association for the Advancement of Science.
Washington, DC. 242 pp.

Pennington, J. T. 1985. The ecology of fertilization of echinoid eggs:
the consequences of sperm dilution, adult aggregation, and syn-
chronous spawning. Biol Bull 169: 417-430.

Petersen, C. VV. 1991 . Variation in fertilization rate in the tropical reef
fish, Halichoeres bivatlaliis: correlates and implications. Biol. Bull.
181: 232-237.

Petersen, C. W., R. R. Warner, S. Cohen, H. C. Hess, and A. T. Sewell.
1992. Variable pelagic fertilization success: implications for mate
choice and spatial patterns of mating. Ecology 73: 391-401.

Quinn, J. F., S. R. Wing, and L. VV. Botsford. 1993. Harvest refugia
in marine invertebrate fisheries: models and applications to the red
sea urchin, Strongylocentrotus franciscanus. Am. Zool. 33: 537-550.

Rothschild, B. J., and T. R. Osborn. 1988. Small-scale turbulence and
plankton contact rates. ./ Plank Res 10: 465-474.

Ruiz-Bravo, N., and VV. J. Lennarz. 1989. Receptors and membrane
interactions during fertilization. Chapter 2 in The Molecular Biology
oj Fertilization H. Schatten and G. Schatten, eds. Academic Press.
San Diego. 384 pp.

Saiz, E., M. Alcaraz, and G.-A. Paffenhofer. 1992. Effects of small-
scale turbulence on feeding rate and gross-growth efficiency of three
Acartia species (Copepoda: Calanoida). J Plunk Res 14: 1085-
1097.

Schlichting, H. 1987. Boundary-Layer Theory 7th ed. McGraw-Hill,
New York. 817 pp.

Schuel, H. 1984. The prevention of polyspermic fertilization in sea
urchins. Biol. Bull. 167: 271-309.

Sewell, M. A., and D. R. Levitan. 1992. Fertilization success during a
natural spawning of the dendrochirote sea cucumber Cucumaria
nunicita. Bull Mar Sci 51: 161-166.

56

K. S. MEAD AND M. W. DENNY

Steinhardt, R. A., B. Guoqiang, and J. M. Alderlun. 1994. Cell mem-
brane resealing by a vesicular mechanism similar to neurotransmitter
release. Science 263: 390-393.

Svendsen, I. A. 1987. Analysis of surf zone turbulence. J Geophvx.
Rex 92: 5115-5124.

Tennekes, H., and J. 1,. Lumley. 1972. A First Course in Turbulence.
MIT Press, Cambridge. 300 pp.

Thomas, VV. H., and C. H. Gibson. 1990a. Effects of small-scale tur-
bulence on microalgae. / Appl Phycol 2: 71-77.

Thomas, \V. H., and C. H. Gibson. I990b. Quantified small-scale tur-
bulence inhibits a red tide dinoflagellate. Gonyaulax polyedra Stein.
Deep-Sea Res .17: 1583-1593.

Thomas, VV. H., and C. H. Gibson. 1992. Effects of quantified small-

scale turbulence on the dinoflagellate, Gymnodinium xangiiineinu

(splendens): contrasts with Gonyauliix (Lingulodiinnm) pulveiira.

and the fishery implication. Deer-Sea Res 39: 1429-1437.
Thornton, E. B., and R. I. Gu/a. 1983. Transformation of wave height

distribution. / Gtvi/'/i.r.v. Rex. 88: 5925-5938.
Vacquier, \'. D. 1981. Dynamic changes of the egg cortex. Dcv. Biol.

84: 1-26.
V'ogel, S. 1981. Lite in Moving Hinds. Princeton University Press,

Princeton. NJ. 352 pp.
Vogel, H., G. Chizak, P. Chang, and P. Wolf. 1982. Fertilization kinetics

of sea urchin eggs. Math. BIOXCI 58: 189-216.
Yund, P. O. 1990. An in .xint measurement of sperm dispersal in a

colonial marine hydroid. J E.\p. Zoo. 253: 102-106.

Reference: Bio/. Bull. 188: 57-67. (February/March, 1995)

Elemental Distributions in Marine Bivalve Shells as
Measured by Synchrotron X-Ray Fluorescence

KURT THORN 1 *. ROBERT M. CERRATO 2 , AND MARK L. RIVERS'

1 Department of Geophysical Sciences. University of Chicago, Chicago, Illinois 60637, and 2 Marine
Sciences Research Center, State University of New York, Stony Brook, New York 11794-5000

Abstract. The concentrations of elements from Mn to
Pb in the shells of Mercenaria mercenaria. Mya arenaria,
and Argopecten irradiam were measured using synchro-
tron x-ray fluorescence. This technique provides sensitiv-
ity as low as 1 ppm and resolution of 8 ^m. Elements
were heterogeneously distributed, both on a large scale
(several millimeters) and on a small scale (tens of mi-
crometers). Large-scale variations were observed in the
compositions of shell layers and in seasonal variations in
strontium concentration. Small-scale changes in com-
position included elevated iron levels at the boundary be-
tween the prismatic and inner homogeneous shell of the
hard clam, Mercenaria mercenaria. Variations in stron-
tium concentrations were seen over time spans of several
months, suggesting that this technique can be used to de-
termine historical water temperatures. Elemental maps
with a resolution of less than 10 ^m were produced.

Introduction

The use of marine bivalves as environmental indicators
has received considerable attention. Bivalves are believed
to incorporate trace elements into their shells in propor-
tion to the concentration of those elements in the water
(Rhoads and Lutz, 1980). This incorporation is also in-
fluenced by other circumstances, including water tem-
perature and salinity (Rosenberg. 1980). The incorpora-
tion of trace elements into marine bivalves could be used
to monitor temporal changes in aspects of the marine
environment, including the elemental composition and

Received 29 October 1993; accepted 22 November 1994.

* Current Address: Whitespruce Dr., Wading River, NY 1 1792.

Abbreviations: SXRF, synchrotron x-ray fluorescence; PIXE, proton
induced x-ray emission; EPMA, electron probe microanalysis; MDL;
minimum detectable level.

temperature of the water. Moreover, recent pollution
could be studied, as well as paleoceanographic conditions.
Strontium/calcium ratios have already been used as a
method in paleothermometry that overcomes the limi-
tations of the more common oxygen isotope paleother-
mometry (Beck el al, 1992). Additionally, many bivalves
produce visible growth increments in their shells, often
as frequently as once per day. These growth increments
provide a method of dating any observed changes in el-
emental composition of the shell.

When techniques like instrumental neutron activation
analysis (INAA) and atomic absorption analysis are ap-
plied to whole shells of marine bivalves, heavy metal con-
centrations of less than 5 ppm by weight are reported (Al-
Dabbas el al, 1984; Koide et al, 1982; Milliman, 1974).
Carriker el al (1982), using proton induced x-ray emission
analysis (PIXE), reported concentrations of Cu and Zn
in oysters (Crassostrea virginica) of 20 to 30 ppm weight.
When such concentrations are measured by bulk analysis,
fairly large amounts of shell are required for analysis, so
resolution over short time scales is impossible.

All previous attempts to measure the distribution of
elements within bivalve shells have been made with either
PIXE microprobes or electron probe microanalysis
(EPMA). Because the minimum elemental concentration
measurable with EPMA is high, only those elements oc-
curring in fairly high concentrations in the shell (Ca, Sr,
Mg, S) can be observed (Wada and Suga. 1976; Rosenberg
and Hughes, 1991). Three PIXE microprobe studies
(Carriker et a!., 1982; Carell et al., 1987; Swann et al..
1991) report variations in heavy metal concentrations,
but over fairly long (several month) time scales, although
the excellent spatial resolution of the proton microprobe
(3 ^m) used by Carell et al. (1987) would have enabled

57

58

K. THORN ET AL

them to study periods corresponding to 6-8 hours in
young mussels.

In his analysis of bivalve shell chemistry, Rosenberg
( 1980) concludes that whole-shell analysis represents the
elemental concentration incorrectly because elements are
not necessarily distributed uniformly throughout the shell.
Carriker el al (1982) list several factors that influence the
distribution of elements in shells, including structural and
chemical changes that result from environmental fluc-
tuations, fouling, adsorption, and weathering of the shell
surface, elements adsorbed at shell surfaces and integrated
into the shell matrix, and the heterogeneous distribution
of elements within the shell.

The metal concentrations typically found in bivalve
shells are considerably less than those measurable by
EPMA and are near the 20 ppm detection limit of PIXE
microprobes. Synchrotron x-ray fluorescence has mini-
mum detection limits near 1 ppm, which should permit
detection of these elements and resolution of variations
in their concentration over short time scales.

This study attempts to evaluate the potential of syn-
chrotron x-ray fluorescence as a technique for micro-
analysis of trace element distributions in bivalve shells of
three species. Some samples were also analyzed with a
PIXE microprobe in a pilot attempt to measure elemental
distributions on scales down to 1 //m.

Materials and Methods

Preparation of shell sections

Three species were examined in this study: hard clams
(Mercenaria mercenaria), soft-shell clams (Mya arenaria),
and bay scallops (Argopecten irradians). Hard clams were
obtained from a population in Moriches Bay, soft-shell
clams from a population in Stony Brook Harbor, and
scallops from a population in Peconic Bay. All collection
locations were on Long Island, New York. Two types of
sections were used in this study: thin sections (less than
250 nm thick) and thick sections (about 1 mm thick). For
studies of Mya arenaria, sections were made of the chon-
drophore, an internal structure projecting from the hinge
region. Sections of the chondrophore were used for anal-
ysis because the chondrophore contains better preserved
growth lines than the main shell (Cerrato et al.. 1991 ). In
addition, it is isolated from the environment, preventing
damage and elemental contamination.

Shells were cleaned to remove adherent debris by
scrubbing with a nylon brush in tap water. Thick sections
were prepared by cutting a section from the shells along
the axis of maximum growth. The cut surfaces were
ground with successively finer silicon carbide grits and
polished with aluminum oxide. The sections were then
attached to cardboard mounts for synchrotron x-ray flu-

orescence analysis. Thin sections were prepared in the
same manner as thick sections except that the shell sec-
tions, after grinding, were glued to pure SiO 2 glass slides
with 5-minute epoxy. X-ray fluorescence analyses of the
epoxy and slide indicated that they contained only trace
amounts of zirconium and titanium. The mounted sec-
tions were resectioned to several hundred micrometers
and ground to the desired thickness with successively finer
silicon carbide grits. The sections were then hand polished
with aluminum oxide. Previous x-ray fluorescence studies
have shown that this process neither introduces significant
amounts of elemental contamination nor smears the el-
emental distribution of the sample.

Shell microstmctiire

The shells of Mercenaria mercenaria are composed of
three distinct shell layers: an outer prismatic layer and a
middle and an inner homogeneous layer. The middle and
inner homogeneous layers are separated by the pallial
myostracum (Panella and MacClintock, 1968). All layers
of the shell are aragonitic (Taylor et al.. 1973). During
periods of stress, the organism ceases its growth and re-
tracts its mantle, producing a translucent band, known
as a growth break, in the shell. Daily growth lines are also
visible in the shells of Mercenaria mercenaria. These con-
sist of an aragonite-rich increment followed by a protein-
rich line(Kennish, 1980).

Both the shell and the chondrophore of Mya arenaria
contain daily growth patterns; however, the growth pat-
terns in the chondrophore are much better preserved than
those in the main shell. These growth increments are sim-
ilar in form to those in Mercenaria mercenaria: they are
composed of a thin line followed by a broad increment.
It is not known whether they also exhibit the same pattern
of protein-rich lines and calcium-carbonate-rich incre-
ments. The chondrophores also show strong seasonal pat-
terns: they are opaque in spring, translucent in summer,
and opaque in fall-winter. Winter and spring are separated
by a prominent translucent spawning band (Cerrato et
al. 1991).

The shell of Argopecten irradians consists of an inner
and an outer calcitic foliated layer separated by a middle
aragonitic crossed-lamellar layer (Kennedy et al.. 1969).
To our knowledge, these shells do not contain growth
lines.

Synchrotron x-ray fluorescence analyses

Synchrotron x-ray fluorescence analyses were per-
formed at beamline X26-A of the National Synchrotron
Light Source at Brookhaven National Laboratory. Sam-
ples were mounted in a sample holder in air. X rays pro-
duced by the synchrotron travel 9 m through an evacuated

ELEMENTAL DISTRIBUTIONS IN BIVALVES

59

pipe to the sample. The x rays produced by the synchro-
tron have a critical energy of 5 keV and travel through
several beryllium windows (total Be thickness = 1 mm)
and 35 mm of air before striking the sample. The air path
of the fluorescent x rays sets the effective lower detection
limit at argon (Z = 18). The x-ray beam is collimated by
an 8 nm tantalum pinhole immediately before striking
the sample. This produces a microbeam, allowing reso-
lution of fine structures. Immediately upstream of the
collimator, the intensity of the x-ray beam is measured
by an ion chamber.

The samples were mounted on an X, Y, Z, 6 stage.
They could thus be rastered under the x-ray beam, al-
lowing the measurement of elemental concentrations at
different points as well as along one- and two-dimen-
sional scans over the specimen. An optical microscope
imaging the specimen permitted precise determination
of the location being analyzed. The secondary x rays
emitted by the sample are collected in a Si(Li) detector.
The detector was filtered with 500 ^m of Kapton film,
to reduce the count rate from the Ca x rays. This prevents
saturation of the detector. The spectra are then stored
in a Micro Vax computer for further analysis. Minimum
detectable levels (MDLs) were 5 ppm for Mn, 2.5 ppm
for Fe and Pb, 2 ppm for Sr, and 1 .5 ppm for Ni through
Br, for a spectrum collected for 4 min. MDLs were cal-
culated by measuring the background under these peaks
in a typical x-ray spectrum and computing the elemental
concentration of a peak with an area equal to three stan-
dard deviations of the background. The calculated con-
centrations from the spectra were used to standardize
these measurements.

The sample is mounted with the surface oriented at 45
degrees to the incoming x-ray beam and the detector. This
exploits the angular distribution of the scattered radiation
to minimize background. However, it also results in the
beam traveling relative to the face of the sample as it
penetrates the sample. This prevents the detection of
changes in elemental concentration over small areas per-
pendicular to the beam. This effect is more pronounced
at higher x-ray energies. At Ca (3.69 keV) the horizontal
travel is only 1 1 ^m; at Sr (14.14 keV) it is 55 j/m. To
reduce this effect, one can make thinner samples; however,
it becomes difficult to optically resolve growth lines in
samples thinner than 100 ^m. Thus, the samples were
oriented so that the interfaces of interest were parallel to
the incoming beam whenever possible.

Spectra can be recorded in one of two ways: either the
entire spectrum collected by the detector is recorded, or
the net areas of the peaks of interest are recorded. Gen-
erally, when analyzing individual points, the entire spec-
trum was saved, as this allowed a more sophisticated
background and peak-fitting routine to be used. However,

when scanning a region, generally only the net areas of
the peaks were saved, because this greatly reduces data
analysis time. The Micro Vax implements automated
scanning routines that allow the collection of data over
long scans with no user intervention. The setup of the x-
ray microprobe is discussed more comprehensively by
Gordon et al. (1990).

Absolute concentrations were determined from the
peak areas fit to the spectra by use of a "standardless"
analysis program (NRLXRF). Using fundamental pa-
rameters, this program predicts the relative intensities
for the different x-ray lines in the spectrum. These pre-
dictions are based on the thickness of the sample, the
filtering conditions, and the assumed concentrations.
The concentration of Ca was assumed to be 40% by
weight, and was used as an internal standard to calculate
the composition by comparing the predicted intensities
and the actual count intensities. It was checked by pre-
dicting known standards that were measured at the be-
ginning of each measurement session. The error was
typically less than 5%.

PIXE microprobe analyses

Several samples were also analyzed by proton induced
x-ray emission (PIXE) at the University of Melbourne.
Analyses were performed with a 3 MeV proton beam,
magnetically focused to 1 and 4 jum. Samples were thick
shell sections, mounted in vacuum. The secondary x rays
were collected with a Si(Li) detector, which was generally
filtered with 50 nm Al to reduce the Ca count rate. How-
ever, some analyses were performed without filters in an
attempt to measure sulfur concentrations. Rutherford
backscattering spectroscopy (RBS) analyses were also
made to allow carbon and oxygen concentrations to be
measured.

PIXE analyses were made by sweeping the beam over
a predetermined region of interest in the shape of an un-
closed lissajous figure. Every time an x-ray event is de-
tected by the Si(Li) detector, the energy of the event and
the A and y coordinates of the beam are recorded. The x-
ray spectrum is generated by disregarding the .v and y
values and simply taking the energy spectrum. A map of
the concentration of a given element can then be generated
by placing a window over a given peak and plotting the
intensity of events in that energy range as a function of
A and y. Analyses were performed on scales from 30 by
30 ^m to 100 by 100 /urn and were generally collected for
30 min. The MDLs in calcium carbonate matrix were
about 1 8 ppm for Zn and Fe. This microprobe is described
in more detail in Legge et al. ( 1986).

Although the PIXE microprobe does not have elemen-
tal sensitivity as good as the synchrotron x-ray fluorescence

60

K. THORN ET AL

microprobe, it does have substantially better position res-
olution (a beam spot of 1 /urn rather than 8 /urn). The mean
x-ray production depth for proton excitation in calcium
carbonate is also less than that of x-ray excitation. Both
of these factors enabled higher resolution of trace element
distributions in the shell, but increased the MDLs for
many elements.

Results

Surface contamination

Both shell surfaces of all species studied were found to
be extremely contaminated with many trace elements.
Table I lists the concentrations of these elements on both
the exterior and interior surfaces of several organisms.
Transverse scans across sections of scallop and hard clam
shells showed that metal concentrations were much higher
on both the interior and exterior surfaces of the shell than
they were within the main shell. A representative transect
is shown in Figure 1. Most elements were more concen-
trated on the exterior surface than the interior surface.
However, the ratio of surface to main shell concentrations
varied widely from element to element. The Ni concen-
tration was only 1.5 times greater on the exterior surface,
but the Cu concentration was 4.5 times greater. This is
not unexpected, as the processes by which the elements
are deposited on these surfaces are very complex. Scans
in hard clams showed similar results. Trace element con-
tamination on the interior shell surface is present in only

a very thin layer. Within 100 ^m the trace element con-
tamination drops by a factor of 20-50, typically to within
a factor of 2 of its average concentration. Most elements
on the exterior surface of the hard clam behave in the
same way as those on the interior surface, except iron
(Fig. 2). Iron concentrations peaked both near the surface
(30 ^m) and farther in ( 180 ^m). This is probably due to
the periostracum and adherent debris on the shell surface;
but why only Fe is affected is not known. The other trace
elements are present in a surface layer that contains little
calcium. This effect was most pronounced on the external
surface; the maximum trace metal concentrations were
at 30 /urn from the edge of the shell and about 240 ^m
from where calcium concentrations became constant.

Shell layers

Most trace and minor elements (including Sr) are pref-
erentially incorporated into the prismatic shell of the hard
clam. The ratio of concentration in the prismatic shell to
concentration in the homogeneous shell varied consid-
erably from element to element, from about 20 for Fe to
1.5 for Sr. The reason for this increased concentration is
unknown. Similar differences between shell layers were
observed in scallops. Sr concentration in scallops was in-
versely related to both Mn and Fe concentrations; and Sr
concentrations were highest in the outer 500 ^m of shell,
while Mn concentrations were highest in the inner 400 ^m
of shell. Some feature of the biological precipitation pro-
cess is influencing the trace element concentration, as the

Table I

Elemental concentrations (ppm) in selected marine bivalves

Location

Mn

Fe

Ni

Cu

Zn

Br

Sr

Ph

Scallops (Argopecten imuiians)

Exterior surface

262

1080

152

902

308

17

814

21

Interior surface

208

636

102

201

106

21

1300

21

Outer shell layer

34 1 .4

8.5 3.5

2 +

2.5 2.1

3 1.4

1

1495 106

2

Inner shell layer

48.5 5

8.5 3.5

20

20

5 1.4

2 1.4

911.5 18

5 2.1

Hard-shell clams (Mercenana mercenana)

Prismatic shell

12.4 3

219 59

<1.5

3.6 2.2

<1.5

5.4 0.8

1247 448

<2.5

Prismatic (DEC)

8.3 3.9

38.8 35

2.8 1.5

1.5 0.6

1.5 0.6

30

1139 + 211

1

Prismatic (PIXE)

9.8 15

349 518

<18

<18

13 6.8

12.8 + 8.3

3832 961

Prism. /homog. boundary

3. 5 2.1

1085 302

<l.5

3.3 0.4

<1.5

5.5 1.1

1 1 1 5 345

<2.5

Middle homog. shell

7.2 + 2.2

11.6 + 7.3

<1.5

3 0.9

<l.5

5.4 0.5

984 353

<2.5

Middle homog. (DEC)

6.8 8.2

4 4.1

1 0.8

1.8 0.5

1.3 0.5

20

1007 336

2.3 2.5

Middle homog. (Thick)

1.25 0.5

<2.5

1

1

1.5 0.6

2.5 1.3

725 + 132

1

Inner homog. (DEC)

1.8 1.3

2. 8 2. 5

1

3 2.6

1.4 0.6

1.4 0.6

1673 + 180

1.7 + 0.6

Soft-shell clams (Atyn arenana)

Chondrophore

10 4.4

<2.5

<1.5

<1.5

2 0.6

<1.5

1383 133

<2.5

All concentrations are in parts per million. Errors are given as the standard deviation of multiple measurements within the same shell layer; they
do not reflect measurement error. Numbers without listed errors are based on single measurements. DEC: Shells came from collections in waters
closed by the Department of Environmental Conservation. Thick: Data were taken on shell thick sections.

ELEMENTAL DISTRIBUTIONS IN BIV-MVES

61

Figure 1 . Representative transects along which scans were taken in Mcrccnaria nwirciuiria (A) Transect
across daily growth increments in prismatic shell. (B) Transect across growth break. (C) Transect across
prismatic-homogeneous boundary. (D) Transect across daily growth increments in homogeneous shell. (E)
Transect across exterior shell surface.

chemistry of the scallop shell does not behave like geo-
logical carbonates. However, differences in chemical
composition between biologically precipitated carbonates
and geologically precipitated carbonates are not unknown:
Buchardt and Fritz ( 1978) described aragonitic shell pre-
cipitated by a freshwater gastropod in which the strontium
concentration of the shell is 5 times less than that of geo-
logically precipitated aragonite under the same conditions.
In addition. White el al. ( 1977) showed that Mn can re-
place Ca in aragonite formed by Mya arenaria, whereas
this is not possible in geologically precipitated aragonite.
At the boundary between the prismatic shell and the
middle homogeneous shell in the hard clam, very pro-
nounced changes in elemental composition were observed
(Figs. 3, 1 ). At this boundary, iron concentrations are typ-
ically 4.5 times greater than in the prismatic shell and
100 times greater than in the homogeneous shell. This
iron peak is very narrow, about 40 /jm wide, and was
present in every measurement made across this boundary.
The peak was always within 50 yum of where the boundary

was located optically. Strontium concentrations declined
between the prismatic and the homogeneous boundary,
but this occurred more gradually, over about 100 /urn, and
appears to be associated only with the difference in Sr
concentration in the two shell layers. Sr concentration
reaches its nominal concentration in homogeneous shell
at the same location as the peak in iron concentration.
Manganese concentrations are lower at this boundary than
in either the prismatic shell or the homogeneous shell,
perhaps because the much greater concentration of iron
is occupying the sites in the crystal lattice that are normally
occupied by manganese. No feature similar to this has
been observed in other organisms to our knowledge.

Sr concent nil ion maps

Both the x-ray microprobe and the proton microprobe
can be used to make high-resolution elemental maps, so
changes in elemental concentration can be explored over
very small areas. And with bivalve shells, fine variations

62

K. THORN ET AL.

<n

o
O

30 60 90 120 150 180 210 240 270 300

Depth (urn)

Figure 2. Scan across the outer surface of a Mercenaria mercenana
thick section. Data were recorded as individual spectra at each point.
The spectra were then fitted to a sum of gaussian peaks and a background.
The counts reported are the net areas of the peaks, less the background.
Data was recorded for 40 s per point (detector live time).

with ontogeny can also be studied. We have used this
scanning ability to elucidate changes within the shell
structure, as with the prismatic-homogeneous boundary
described above; but we have also measured the variations
of trace element concentrations with time. The measure-
ments of strontium were most successful. Because of its
high concentration in the shell, strontium can be measured
very accurately with counting times of only 5 s at each
point in the shell. One-dimensional scans can therefore
be made very quickly: a millimeter can be scanned at 10-
/xm resolution in less than 10 min. We could easily map
a 2-year area of shell deposition at a resolution of a few
days (Fig. 4) and produce very high-resolution two-di-
mensional elemental maps. A map of the hinge region of
M. mercenaria is shown in Figure 5. Note the very good
correlation between the strontium concentration and the
banding in the thin section. Such scans are also possible
for other elements, though reduced concentrations require
proportionally longer counting times. Scans to measure
Fe concentration were typically recorded at 30 s per point.
For elements at levels of a few parts per million, counting
times of several minutes are necessary.

High-resolution maps of Sr concentration made with
the PIXE microprobe have shown variations in Sr con-
centration over scales as small as 1 nm. These variations
appear to be correlated with the presence of growth breaks.
However, because only a few PIXE measurements were
made, this is not conclusive.

Growth lines

A comprehensive attempt was made to find trace ele-
ment correlations with the daily microgrowth lines in the
prismatic shell of M. mercenaria. The results strongly
suggest that, although there is variation in trace element
concentration in the prismatic shell of M. mercenaria, it
is not necessarily correlated with the presence of micro-
growth lines. All scans showed that iron concentration in
the prismatic shell was nonuniform, but in most cases,
no correlation with the microgrowth lines could be made.
However, one scan (nsls!06.10) showed very high cor-
relation between the microgrowth lines and fluctuations
in the iron content. In that scan, every microgrowth line
was accompanied by a drop in iron concentration. Other
scans showed some correlation between the presence of
microgrowth lines and iron concentration, but none was
as pronounced as this one. In an effort to improve the
detectability of any iron fluctuations, we oriented samples
so that the microgrowth lines were parallel to the incident
beam. Thus, as the beam penetrated, it would stay within
the microgrowth lines rather than transversing them. In
addition, images were captured digitally after the beam
position had been determined, enabling precise compar-
ison between the image and the elemental concentration.
Neither of these improvements showed further correla-
tions between iron concentration and the presence of mi-
crogrowth lines. PIXE microprobe analyses were not sen-
sitive enough to detect variations in the iron concentration
of less than 40 ppm. and no structures were seen in the
PIXE measurements of iron concentration. A correlation
between microgrowth lines and iron concentration in the
shell seems unlikely, but this is an area deserving further
studv.

Discussion

Synchrotron x-ray fluorescence (SXRF) microprobes
provide lower minimum detectable levels than PIXE and
EPMA microprobes. The experimentally determined
MDL for Fe in thick clamshell sections was 3 ppm for
SXRF and 1 8 ppm for PIXE. Our experience is that PIXE
microprobes can make much larger and higher resolution
maps than SXRF microprobes, although at lower sensi-
tivity. PIXE measurements are generally made for long
times (30 min) over large areas ( 1000 j/nr ), whereas SXRF
measurements are made for short times ( 1 min) over small
areas (64 /jm 2 ). Electron microprobes have MDLs of sev-
eral hundred parts per million, worse than either PIXE
or SXRF. But electron microprobes are typically more
sensitive to low-Z elements than SXRF and PIXE micro-
probes (Janssens ct ai. 1992). The minimum spot size of
the NSLS microprobe is 8 ^m; for EPMA and PIXE mi-

ELEMENTAL DISTRIBUTIONS IN BIVALVES

63

.15000

30000

25000

w

D

o
o

20000

15000 -

10000

5000

100 200 300 400

Position (urn)

500

600

Figure 3. Scan across the prismatic/homogeneous boundary in a Mercenaria mcrcenaria thin section.
Data were recorded as the area under user-defined regions of interest, with the background subtracted. Data
was recorded for 1 5 s per point (detector live time). Initial Fe concentration is 2 1 1 ppm, initial Sr concentration
is 1550 ppm. The Fe peak concentration is 1420 ppm.

croprobes it is 1 ^m. New PIXE microprobes claim spot
sizes of 300 nm.

For low-Z elements (Z < 18) EPMA has the greatest
sensitivity as well as small spot size. Because of the ready
availability of electron microprobes, EPMA is also suited
to measurements of higher-Z elements present in quan-
tities of several hundred parts per million or more. For
measurements of elements present in concentrations of a
few parts per million, SXRF is the method of choice. This
is the only technique that combines MDLs of a few parts
per million with spot sizes of a few micrometers. However,
for some measurements, the spot size of the NSLS mi-
croprobe (8 ^m) is too large. In these cases, PIXE micro-
probes can be used, if the elements are present in large
enough concentrations. The greater accessibility of PIXE
microprobes also makes them attractive for studies of ele-
ments present in concentrations of several tens of parts
per million.

Bulk analysis techniques cannot provide the same spa-
tial resolution as microprobes. The neutron activation
studies of Carell ct ai (1987) were performed on shell
samples of 10-50 mg. This corresponds to a shell section

0.5 mm thick by 6 mm long by 3 mm deep. If this section
is cut with the thin direction normal to the ventral margin,
it represents approximately 1 month of growth. Atomic
absorption analysis can detect much smaller amounts of
trace elements, hence the limiting factor is the size of the
shell section that can be obtained. This probably limits
bulk analysis techniques to time resolutions of a few
weeks. In addition, many clamshell sections must be pre-
pared to make a study at this resolution over a significant
period. In contrast, microprobe techniques require only
one section per shell. Consequently, bulk analysis tech-
niques are suitable for situations where resolution of small
time scales is not necessary and integration of elemental
concentrations over all dimensions of the shell is not a
problem. In practice, microprobe techniques are probably
more suitable for time-resolved studies of trace elements
in bivalve shells. Bulk analysis methods can, however,
detect elemental concentrations a factor of a thousand
smaller than microprobes can.

Wavelength-dispersive detectors should improve the
usefulness of SXRF microprobes for analyzing trace ele-
ments in bivalve shells. Because wavelength-dispersive

64

K. THORN ET AL.

Distance (mm)
0.0 0.5 1.0 1.5 2.6 2.5
15

10

50 100 150

Days Before Collection

15

E 10

W

< 5

15

10

I 5
w

Distance (mm)
2 3

100

200
Days Before Collection

300

400
Distance (mm)

200

400
Days Before Collection

600

800

Figure 4. Variation in Sr concentration across growth increments in the chondrophore ofAIya arenaria.
Each of the three scans was taken on a different shell; all shells were harvested at the same time. Distance
is measured in millimeters from the ventral margin of the chondrophore. Time before harvesting was calculated
hy dividing the distance from the margin by the average growth rate per day, which was determined by
measuring the distance from the margin to the spawning break which occurred approximately 90 days before
harvest. Each plot consists of three separate scans separated by up to 1 50 /jm parallel to the growth increments.
Notice the strong correlation of variations in each of the separate scans. Note also the correlations between
the shells. We believe that these variations represent changes in the water temperature over time.

spectrometers detect only a single x-ray wavelength, they
eliminate the filtering necessary to prevent saturation of
an energy-dispersive detector by Ca x rays. This filtering
also attenuates other low-energy x rays, preventing the
detection of low-Z elements and resulting in high
(200 ppm) MDLs for Ti. Use of a wavelength-dispersive
detector would allow the detection of much smaller quan-
tities of low-Z elements, and possibly improve MDLs for
higher-Z elements as well. Such a device is now in use at
the National Synchrotron Light Source at Brookhaven
National Laboratory (Rivers et at., 1992). Future SXRF
microprobes will become even more useful as they are

developed at third-generation synchrotrons, such as the
Advanced Photon Source (APS) at Argonne National
Laboratory. A planned SXRF microprobe at the APS will
have 10 4 times more photons in a given energy range.
This will allow the use of extremely small spot sizes
(0.1 nm) at current detection limits, or lower detection
limits at small spot sizes ( 1 nm).

SXRF microprobes are not limited merely to the ap-
plications discussed here. It is also possible to use SXRF
to study other calcined samples, such as corals or fish
otoliths. Studies offish otoliths with EPMA (Gunn et a/.,
1992) and PIXE (Gauldie et at.. 1992) suggest that the

ELEMENTAL DISTRIBUTIONS IN BIVALVES

65

Figure 5. Two-dimensional scan (map) of Sr and Ca concentration in the hinge region of Merccnaria
incrci'iiaria. Each pixel measures 20 ^m on a side, and was collected for 20 s. The data were recorded as
background-subtracted regions of interest and have been normalized to beam intensity and detector live
time. The box on the photomicrograph approximately corresponds to the region scanned. Note the correlation
of the variations in strontium concentration with the seasonal banding in the chondrophore.

SXRF microprobe would he a valuable tool for such stud-
ies. Spatial resolution and elemental sensitivity similar to
that obtained here should be achievable in studies of other
calcined samples. Furthermore, it is not necessary to sac-
rifice the molluscs and section their shells to use the SXRF
microprobe. Because the beam deposits relatively little
energy in the shell and stops within the first few hundred
micrometers of the shell, live molluscs can be studied,
though this raises the problems of removing surface con-
tamination and dating the sampled positions accurately.
SXRF can also be used to detect organometallic com-
pounds, provided the metals are of sufficiently high Z.
Techniques such as extended x-ray absorption fine struc-
ture (EXAFS) also promise the ability to determine the
oxidation state of metals detected bv SXRF.

The measurements made on bivalve shells have re-
vealed several interesting features. The contamination of
the exterior surface is most likely due to the adsorption
of elements onto the calcium carbonate from seawater.
However, the contamination could also be due to adherent
debris on the surface of the shell. The contamination of
the interior surface is probably not due to elements ad-
sorbed onto the calcium carbonate during active growth,
because they would then have to be removed before ad-
ditional shell was deposited. Although there is some evi-
dence that the shell undergoes structural changes after
deposition (Gordon and Carriker. 1978). it seems more
likely that the elemental contamination arose after the
shells were collected, either from adsorption of metals
from the pallial fluid or from contamination after collection.

66

K. THORN ET AL

Seasonal variations in strontium concentration were
observed in the shells of both hard- and soft-shell clams
(Fig. 4). Differences in strontium concentration of 30%
to 40% were typical between winter and summer. Each
of the high and low Sr regions was about 180 days long.
We believe these variations are a result of the seasonal
temperature change of the seawater. In addition, shorter-
term variations (100 nm long) were observed; these prob-
ably correspond to shorter-term temperature variations.
As with paleothermometry done with Sr/Ca ratios in cor-
als (Beck el al., 1992), our results show that increases in
Sr concentration correspond to decreases in temperature.
Assuming that the change in Sr concentration is linearly
related to the change in water temperature, the change in
the observed Sr count rate is 18 counts/s/C. In a 10-s
per point scan, changes of 6 counts/s are observable. This
corresponds to a temperature change of 0.33C. The use
of x-ray fluorescence to measure small (0.3C) tempera-
ture variations over short (3-day) time scales seems quite
simple. In practice, the resolution is probably limited by
the rate at which shell strontium concentrations equili-
brate to the ambient water temperature. Our scans show
good correlation between different transects on an indi-
vidual shell and between different shells harvested from
the same location at the same time. This strongly suggests
that we are observing environmental effects upon the shell.
Figure 5 also shows very good correlation between Sr con-
centration and seasonal banding in the shell.

One problem complicating the use of x-ray fluorescence
as a means of paleothermometry using Sr/Ca ratios is that
for samples thinner than 200 /urn. sample thickness influ-
ences the Sr count rate. This is especially a problem near
the edges of shells, which were often inadvertently ground
thinner during preparation than other parts of the shell.
Such "thickness effects" show up as a gradual increase in
the Sr (and in extreme cases Ca) count rate as the sample
is scanned. In practice, the best way to avoid such effects
is to use samples thicker than the mean fluorescence depth
for Sr K x rays, because preparing samples uniform in
thickness to within a few percent is difficult to do. In that
case, the Sr count rate is unaffected by small variations
in sample thickness.

The synchrotron x-ray fluorescence microprobe has
significant advantages for trace element analyses in cal-
cined structures. It combines very low detection limits
with very small spot sizes. This allows the measurement
of elemental concentrations as low as 1 ppm over scales
as small as 10 ^m. Synchrotron x-ray fluorescence mi-
croprobes have significantly better detection limits than
EPMA and PIXE microprobes, and are probably the best
instrument for most studies of trace element distributions
in calcified structures. When structures smaller than
10 nm are to be resolved, the extremely small spot size

of PIXE microprobes makes them the analytical technique
of choice.

Acknowledgments

We are grateful to David Jamieson for performing the
PIXE and RBS analyses at the University of Melbourne.

Literature Cited

AI-Dabbas, M. A. M., F. H. Hubbard, and J. McManus. 1984. The

shell ofAfytilits as an indicator of zonal variations of water quality
within an estuary. Esiuarine Coastal Shelf Sci. 18: 263-270.

Beck, J. W., I,. R. Edwards, E. Ito, F. W. Taylor, J. Recy, F. Rougerie,
P. Joannot, and C. Hcnin. 1992. Sea-surface temperature from coral
skeletal strontium/calcium ratios. Science 257: 644-647.

Buchardt, B., and P. Fritz. 1978. Strontium uptake in shell aragomte
of the freshwater gastropod Limnaea stagnalis. Science' 199: 291-
292.

C'arell, B., S. Forberg, E. Grundelius, L. Henrikson, A. Johnels, II. Lindh,
H. Mutvei, M. Olsson, K. Svaerdslroem, and T. Westermark.
1987. Can mussel shells reveal environmental history? Amhio 16:
2-10.

Carrikcr, M. R., C. P. Swann, and J. W. Ewart. 1982. An exploratory
study with the proton microprobe of the ontogenetic distribution of
16 elements in the shell of living oysters. Mar Biol 69: 235-246.

Cerrato, R. M., H. V. E. Wallace, and K. G. Lighlfool. 1991. Tidal
and seasonal patterns in the chondrophore of the soft-shell clam Mya
urcihii-ui liiol Hull 181: 307-31 1.

Gauldie, R. W., G. C. Coote, K. P. Mulligan, and I. F. West. 1992. A
chemical probe of the microstructural organization offish otoliths.
Comp Hiochem f'hysiol. 102A: 533-545.

Gunn, J. S., I. R. Harrowfield, C. H. Proctor, and R. E. Thresher.
1992. Electron probe microanalysis offish otoliths evaluation of
techniques for studying age and stock discrimination. / E.\p Mar.
Biol Ecol. 158: 1-36.

Gordon, J., and M. R. Carriker. 1978. Growth lines in a bivalve mol-
lusk: subdaily patterns and dissolution of the shell. Science 202: 519-
521.

Gordon, B., A. L. Hanson, K. \V. Jones, J. G. Pounds, M. L. Rivers, G.
Schidlovsky, P. Spanne, and S. R. Sutton. 1990. The application
of synchrotron radiation to microprobe trace element analysis of
biological samples. Niicl. lust. Melh. B45: 527-531.

Janssens, K. H., F. van Langevelde, F. C. Adams, R. D. Vis, S. R. Sutton,
M. L. Rivers, K. W. Jones, and D. K. Bowen. 1992. Comparison
of synchrotron X-ray microanalysis with electron and proton mi-
croscopy for individual particle analysis. Adv. \-Ray Anal 35B: 1265-
1275.

Kennedy, \V. J., J. D. Taylor, and A. Hall. 1969. Environmental and
biological controls on bivalve shell mineralogy. Biol. Rev 44: 499-
530.

kennish, M. J. 1980. Shell nucrogrowth analysis: Mercenaria mercen-
iinii as a type example for research in population dynamics. Pp. 255-
294 in Skeletal (irowth of Aquatic Organisms. D. C. Rhoads and R.
A. Lutz. eds. Plenum, New York.

Koide, M., D. S. Lee, and E. D. Goldberg. 1982. Melal and Iransuramc
records in mussel shells, byssal threads and tissues. Esiuarine Coastal
Shell Sci 15:679-695.

Legge, G. J. F., C. D. Mackenzie, A. P. Mazzolini, R. M. Sealock,
D. N. Jamieson, P. M. O'Brien, J. C. McCallum, G. 1.. Allan,
R. A. Brown, R. A. Colman, B. J. Kirby, M. A. Lucas, J. 7>hu, and
J. Cerini. 1986. Microheam imaging at micron and submicron res-
olution. Kitcl I nut. Melh. BIS: 669-674.

ELEMENTAL DISTRIBUTIONS IN BIVALVES

67

Milliman, J. D. 1974. Marine Carbonates, Part I of Recent Sedimentary
Carbonates Springer- Verlag. New York. 375 pp.

Panella, G., and C. MacClintock. 1968. Biological and environmental
rhythms reflected in molluscan shell growth. / Paleonlol. 42: 64-
80.

Rivers, M. L., K. Thorn, S. R. Sutton, and K. VV. Jones.
1992. Wavelength dispersive detectors in synchrotron x-ray fluo-
rescence microprobe analysis. EOS, Trans Am. Geophys. Union.
73(43): 620.

Rhoads, D. C., and R. A. Lutz. 1980. Skeletal records of environmental
change. Pp. 1-19 in Skeletal Growth of Aquatic Organisms. D. C.
Rhoads and R. A. Lutz, eds. Plenum, New York.

Rosenberg, G. D. 1980. An ontogenetic approach to the environmental
significance of bivalve shell chemistry. Pp. 1 33- 1 68 in Skeletal Growth

of Aquatic Organisms. D. C. Rhoads and R. A. Lutz, eds. Plenum,

New York.
Rosenberg, G. D., and W. W. Hughes. 1991. A metabolic model for

the determination of shell composition in the bivalve mollusc, MytHus

edulis. Lethaia 24: 83-96.
Swann, C. P., K. M. Hansen, K. Price, and R. Lutz. 1991. Application

of PIXE in the study of shellfish. Nucl. lust AM/; B56/57: 683-

686.
\Vada, K., and S. Suga. 1976. The distribution of some elements in

the shell of freshwater and marine bivalves by electron microprobe

analysis. Bull. Nail Pearl Res Lab 20: 2219-2240.
White, L. K., A. Szabo, P. Carkner, and N. D. Chasteen. 1977. An

electron paramagnetic resonance study of manganese(ll) in the ara-

gonite lattice of a clam shell, Mya arenaria. J. Phys. Chem. 81(14):

1420-1424.

Reference: Biol. Bull 188: 68-77. (February/March, 1995)

Life History Patterns of Discorsopagurus schmitti,
a Hermit Crab Inhabiting Polychaete Tubes

FRANCESCA GHERARDI 1 AND PAUL M. CASSIDY 2

[ Dipartimento ill Biologia Animate e Genetica "Leo Pardi, " Universita di Firenie, Via Romana 17,

50125 Firenic, Italy: and 2 Shannon Point Marine Center, Western Washington University.

1900 Shannon Point Rd.. Anacortcs, Washington 98221

Abstract. Discorsopagurus .schmitti is a hermit crab that
inhabits empty polychaete tubes in the North Pacific. Here
we describe some aspects of its lite history (relative growth,
population structure, reproductive biology, and incidence
of parasitism) and discuss the relationships among them.
Unlike most hermits, the two sexes of this species have
similar size distributions. In both sexes, larger body size
is accompanied by a higher reproductive output (larger
clutch size in females and more intrasex competitive po-
tential in males). The energy the females expend in egg
production might be equaled in this species by the energy
the males expend in supporting parasites. In fact, the ex-
tent of infestation by two rhizocephalans [Peltogaster
boschmae and Thilacoplethus ( = Thompsonia) reinhardi]
is more pronounced in males, especially those in the larger
size classes. However, rhizocephalans have little effect on
their hosts; growth and secondary sexual characters are
not influenced. The only morphological modification is
the more frequent loss of the second pleopod. Infected
hermits also showed a mock parental behavior, fanning
the externae with the pleopods as ovigerous females fan
their eggs. Larvae are released in sequential bursts, and
hatching occurs exclusively at night, possibly to minimize
predation by diurnal fishes. Hatching is also synchronized
with neap tides, which might keep the larvae from being
flushed out into open waters. In a species whose habitat
(sabellarian bioherms) is rare and quite unpredictable, it
is beneficial to retain larvae near the parental population.

Introduction

Discorsopagurus schmitti (Stevens, 1925) is an ano-
muran crab that occurs widely along the North Pacific

Received 10 December 1993; accepted IX November 1994.

coasts from Japan to Puget Sound (McLaughlin, 1974).
In both its geographical distribution and its ecological role,
this species is strictly dependent on the polychaete Sa-
hellaria ceinentariitin Moore, 1906. The hermit uses at-
tached worm tubes as housing and occupies a niche within
the community associated with sabellarian bioherms
(Gherardi and Cassidy, 1994a). A bioherm is a rock
formed by accretions from sedentary organisms and sur-
rounded by other kinds of rocks. Within the habitat, D
schmitti has a contagious distribution, the crabs occurring
with a density averaging 6 specimens per dm 2 (Gherardi
and Cassidy. 1994b).

Despite its peculiar habits and widespread distribution,
the main life history traits of the species are still unknown.
Previous papers were concerned only with its adaptations
to the sessile worm tubes (Caine, 1980) and its ecology
(Gherardi and Cassidy, 1994b). Our study investigates the
relative growth, population structure, reproductive biol-
ogy, and incidence of parasitism by rhizocephalans in D.
schmitti.

Materials and Methods

D. schmitti was collected from a wide sabellarian bio-
herm in Burrows Channel. Fidalgo Island (northern Puget
Sound, Washington). A total of 440 specimens were col-
lected: 329 from June to August 1992, and 1 1 1 from Jan-
uary to April 1993. See Gherardi and Cassidy ( 1994a. b)
for details on habitat and sampling procedure.

Sixty-four animals were individually weighed to the
nearest 0.01 g. Chelipeds were excluded from the weight
because they are variable and sometimes absent. For each
specimen, we recorded sex, size (shield length, SL, to the
nearest 0.1 mm), missing chelipeds (i.e.. the number of
"injured" specimens), and the number and maximum

68

LIFE HISTORY OF A TUBE-DWELLING HERMIT CRAB

69

diameter of any egg present. When possible, the maximum
axis of the occupied polychaete tube was measured with
a caliper. The number and position of externae of the
parasitic rhizocephalans Peltogaster boschmae Reinhard
and Thilacopk'thus ( = Thompsonia) reinhardi Lutzen were
also noted. Because we did not assess the presence of root-
lets penetrating major organs (stage of interna), which
precedes the parasite's sexual development, we may have
underestimated the extent of infestation within the
sample.

To describe the format of relative growth (i.e.. the
change in shape with growth; Hartnoll, 1982), we mea-
sured the length of the dactyl (DL) and palm (PA) of both
chelae, and their depth (DE) in 1 30 hermits. To represent
the patterns of the relative growth of these measures (y)
with respect to the SL as an independent variable (.\). the
natural logarithmic transformation (In y = In a + h In x)
of the exponential function y = a x h was used. This re-
lationship fits nearly all the instances of allometric growth
in crustaceans (Hartnoll, 1982). The values of h define
the type of allometry (b = 1: isometry; b < 1: negative
allometry; b > 1: positive allometry). This and the other
parameters of In y on In A, calculated using the Least-
Squares Method, allowed us to use standard tests for sig-
nificance and to compare slopes and intercepts between
groups.

Within winter samples, pleopods were examined and
their configuration related to the occurrence of parasites.
The configuration of pleopods in D. schmitti was first de-
scribed by McLaughlin (1974); the species shows pleopods
2-5, with the exception of some males, in which the sec-
ond pleopod is absent.

The behavior of animals occupying pieces of transpar-
ent glass tubing was recorded with a Panasonic color
camera and played back on a Mitsubishi recorder.

1.5

-i r

25 3 3.5

SHIELD LENGTH (mm)

Figure 1. Relationship between size (shield length) and body (without
chelipeds) weight, compared between sexes. A positive correlation was
found in both males (r = 0.692. df = 41, /> < 0.01 ), and females
(r = 0.809, df = 40, /><0.01).

1.2

2 24 28 32

SIZE CLASSES (SL, mm)
n = 104

i = 100

Figure 2. Size class distributions compared between sexes from
summer collections.

Data on egg incubation and hatching were obtained
from 19 ovigerous females, collected on January 18 (5),
February 8 (5), and April 27 (9) 1993. In the laboratory,
the females were placed in individual glass bowls 20 cm
in diameter, in filtered seawater with a salinity of 28-
3 1 %o. The bowls were kept in a constant temperature unit
at 10C under a light:dark regime of 14:10. Until they
released larvae, ovigerous females were checked twice a
day for hatching, placed in bowls of clean seawater, and
fed Anemia. The number of larvae released daily was re-
corded for ten of the females.

For statistical analysis, we followed the methods and
recommendations of Siegel (1956) and Zar (1984). The
level of significance under which the null hypothesis was
rejected is a = 0.05.

Results

Population struct we

Figure 1 shows the relationship between SL and body
weight (excluding chelipeds) compared between sexes. No
between-sex difference was found in either the slope (32.12
vs. 43.96, / = 1.549. df = 81. ns) or the intercept (-49.90
vs. -78.98. t = 1.781, df = 82, ns) of the regression line.

The sizes of crabs (SL) during summer were analyzed
(Fig. 2). No significant difference in size distributions was
found between the two sexes (G = 4.442, df = 7, ns). The
smallest specimens measured 1.1 (prepubertal, showing
no gonopores), 1.4 (male), and 1.9 mm SL (female), and
the maximum size attained 3.9 mm SL in the two sexes.

The sex ratio was 50.98% (104 males to 100 females),
which did not differ from 1:1 (X 2 = 0.044, df = 1, ns).
Similarly, the sexes remained balanced when three size
classes were distinguished (<2. 6 mm SL: X : = 0.5, df= 1,
ns; 2.6-3.4 mm SL: X : = 0. df = 1, ns; >3.4 mm SL:
X 2 = 0, df = 1, ns).

70

F. GHERARDI AND P. M. CASSIDY
Table I

Isometric or allomeiric growth with sue of Discorsopagurus schmitti (shield length) of three measures of both the major (right) and the minor (left)
chela, compared between sexes

In shield length vs.:

Si

99

b* 1

In DC (major chela)

0.721

0.79

2.173

0.13

0.337

0.40

4.367*

0.65

In PA (major chela)

0.878

0.66

7.450*

0.12

0.624

0.47

7.435*

0.36

In DE (major chela)

0.900

0.84

3.010*

-0.04

0.740

0.57

6.727*

0.33

In DC (minor chela)

0.831

0.70

4.928*

-0.07

0.484

0.41

6.590*

0.31

In PA (minor chela)

0.811

0.60

7.238*

-0.04

0.682

0.51

7.371*

0.07

In DE (minor chela)

0.634

0.48

6.831*

0.19

0.752

0.56

7.445*

0.11

DC = dactyl length; PA = palm length; DE = chela depth
a = intercept of the regression line.
* /><0.01.

All the correlation coefficients (r) are significant (P < 0.01). Isometry is satisfied when the regression coefficient (b) equals 1 after Student's (-test
((), otherwise an allometric growth (here only negative) occurs. The numbers of males and females are 63 and 69. respectively.

The size and sex of specimens in three individually
collected clumps were separately analyzed; the compari-
son did not show any difference in either size distribution
(G = 6.08, df = 4, ns) or sex ratio (X 2 = 1.458, df = 2, ns).

Chelipeds

The growth of both chelae relative to hermit size (SL)
was always negatively allometric for DC, PA, and DE,
with the exception of the DC of the major chela in the
males, where it was isometric (Table I). Table II gives the
between-sex differences in the parameters of the regression
lines; the males had a more voluminous major chela than
similarly sized females, as well as a longer dactyl in the
minor chela. However, the male and female regression

lines crossed at a large crab size (major chela: 3.8, 3.4,
4. 1 mm SL for DC, PA, and DE, respectively; minor chela
DC: 3.7 mm SL).

In all the examined specimens, the right chela was the
major one. This had constantly higher values than the
left one with SL increment (equal slope, but a higher in-
tercept), with the exception of the major chela DE, where
the growth increased with size (Table III).

Injured specimens represented 24.11% of the sample,
without any significant difference between sexes (25% in
males, 24.26% in females: X 2 = 1.762, df = 1, ns). Right

Table III
Between-side comparison of chelar growth in Discorsopagurus schmitti

Table II
Between-sex comparison of chelar growth in Discorsopagurus schmitti

r vs. I

rvs.l

6

a

( t vs. 9

I i vs. 9

DC (major chela)

2.298*

PA (major chela)

2.265* >

DE (major chela)

3.190"

DC (minor chela)

2.734"

PA (minor chela)

1.003

0.301

DE (minor chela)

0.713

1.428

test not applicable.

DC = dactyl length; PA = palm length; DE = chela depth.

* /><0.05, " /><0.01.

Comparisons after Student's (-test (t) between sexes in both the slope
(b) and the intercept (a) of the regression lines (after a In-ln transfor-
mation) describing the relationships between hermit size (shield length)
and three chelar measures. The degrees of freedom are 127 and 128.

DC

0.755

16.342"

>

PA

0.847

20.505**

>

DE

3.935**

99:

b

a

t r vs. 1

l

r vs. I

DC

0.037

17.964"

>

PA

0.418

21.289"

>

DE

1.636

24.428**

>

test not applicable.

DC = dactyl length; PA = palm length; DE = chela depth.

" P<0.0\.

Comparison after Student's (-test (() between the right (r) and the left
(/) chela in both the slope (b) and the intercept (a) of the regression lines
(alter a In-ln transformation) describing the relationships between hermit
size (shield length) and three chelar measures. The degrees of freedom
are 121, 122 in males, and 134, 135 in females.

LIFE HISTORY OF A TUBE-DWELLING HERMIT CRAB

71

N

co

n = 30

i 1 1 1

OS 1 12

^SHIELD LENGTH (mm)

B

1000-1 n = 30

1
1 5

25

35

SHIELD LENGTH (mm)

Figure 3. Relationships between the size of ovigerous females (shield
length) and both the number (after a In-ln transformation. A) and the
diameter (maximum axis, B) of the spawned eggs.

chelipeds were missing more often than left ones (65.96%
vs. 34.04%, X 2 = 4.17, df = 1, P<0.05).

Eggs

Egg-bearing females occurred in winter samples only.
They were first found in January and were still present at
the end of April, though their percentage was low (20%,
9 out of 45). Their numbers did not differ from those of
nonovigerous females (January 18: 19 vs. 10, X 2 = 2.207,
df = 1 , ns; February 1:15 vs. 2 1 , X 2 = 0.694, df = 1 . ns),
and specimens in the two reproductive states shared the
same size distribution (G = 2.906, df = 5, ns) and fre-
quency per size class (<2. 6 mm SL 45.10%, vs. a uniform
distribution: G = 0.486, df = 1, ns; >2.6 mm SL 70%,
G = 1.567, df = 1, ns). The smallest and largest females
found bearing eggs measured, respectively, 1.1 and
3.2mm SL.

Egg number per clutch ranged from 1 4 to 496, averaging
287. Female size (SL) was positively correlated with the
number of eggs (after a In-ln transformation: r = 0.478,
df = 28, P < 0.01, b = 2.56, a = 2.48) (Fig. 3A). The

value of the correlation coefficient did not significantly
differ from 3 (/ = 0.495. df = 28, ns): that is, clutch size
is proportional to the cube of the SL (roughly equaling
the body mass).

The mean egg diameter was 722 ^m (SE = 19, n = 30),
ranging from 455 to 990 ^m. A positive correlation was
also found between the SL of the female and the average
diameter of her eggs (r = 0.586, df == 28, P <0.01.
b = 0. 14, a = 0.40) (Fig. 3B), showing that bigger females
produce larger (and more numerous) eggs.

Eggs are attached to the second through the fourth
pleopods, about 100 per pleopod, in bunches of 7 to 15.
They are slightly ovate and attached by a funiculus, mea-
suring around 1.2 mm. Ovigerous females kept inside
transparent tubing were seen fanning the eggs with a re-
versing current created by the second and third pleopods.

Hatching

Hatching occurred between 1 and 75 days (n = 1 7) after
collection. Because all the analyzed females bore eggs
when collected, this is only a minimum estimate of the
actual length of egg incubation.

The number of larvae per individual ranged from 80
to 541 (average = 226, SE = 46) in the 10 females ana-
lyzed, and did not differ significantly from the number of
eggs per batch (/ = 1.31. df = 38. ns). being on average
98.74%- of the eggs spawned. Larvae were released in 3-
6 days (average = 5.1 day. SE = 0.3). with a maximum
of 209 larvae in the fourth day. No correlation was found
between the length of the hatching period and the female
size (r = 0.074, df = 8, ns), but the former was positively
related to the overall number of larvae (Spearman rank
correlation test: i\ = 0.742, / = 3.134, df = 8, P < 0.02).
Larvae were not released at a constant rate; the percentage
released (Fig. 4) differed significantly throughout the
hatching period (Kruskal-Wallis one-way analysis of vari-

40 -i

TIME (days)

Figure 4. Percentage (average SE) of the larvae released by K)
females plotted against the length of hatching (in days).

72

F. GHERARDI AND P. M. CASSIDY

ance: H = 29.975, df = 5. P < 0.00 1 ). peaking in the third
day and then falling off abruptly after the fifth day.

Hatching occurred exclusively at night, and mostly
during the neap phase of the tide (Mann-Whitney test: U
= 0, H = 7 and 7. P < 0.01 ). when the mean tidal current
is consistently slower (Fig. 5). For our comparison, we
defined neap (or spring) phase as the day of the minimum
(or maximum) tidal excursion for a lunar tidal cycle plus
the 3 days preceding and following that date.

Parasite distribution

Peltogaster boschmae was the most common rhizoce-
phalan parasite in our samples, affecting 18.5% of the
specimens; Tliilacopletlnis ( = Thoipsonici) reinhardi in-
fected 6.8%; and the two rhizocephalans co-occurred in
2.5%. These figures are similar to the percentages reported
by Lutzen (1992) for a previous study in the same area.

A sexual difference in the degree of infestation was seen
in both the number of externae per individual (males:
average = 2.9, SE = 0.6; females: average = 2.3, SE = 0.9;
X 2 = 9.312, df = 2, P < 0.01) (Fig. 6) and the prevalence
of parasitism (i.e.. the percentage of the parasitized her-
mits; Margolis el al.. 1982) (males v.v. females: 28.17% vs.
14.39%, X 2 = 7.144, df = 1, P < 0.01). Similar results
were obtained from the winter samples, where parasitized
males and females scored 28.26% and 9.37%. respectively
(X 2 = 5.424, df = l,P< 0.02). The number of parasitized
specimens did not differ between sampling periods (sum-
mer: 60 out of 281. winter: 19 out of 1 10; X 2 = 0.01,
df = 1, ns).

The maximum number of externae from the summer
samples was 15 in males and 18 in females. We counted
23 externae in one female collected in winter. None of
the ovigerous females in our sample had parasites (0 vs.
20% in nonovigerous ones: X 2 = 5.334, df = \.P< 0.05).
Only one female has been collected bearing both eggs and

hatching days = 93
tide cycles = 4

LU
u_

O

SPRING

1

NEAP

^
Q

<r

D
O

z
<

01

SPRING

TIDE CYCLE (days)

Figure?. Occurrence of hatching within a lunar tidal cycle, compared
ith the mean tidal current speed per day (average SE).

180 -

160 -

140 -

S. 120 -

>

O

100 -

LU

8(H

o: 60 -

LL

40 -

20 -

NUMBER OF EXTERNAE PER HERMIT

Figure 6. Number of the externae of rhizocephalan parasites per
hermit host, compared between sexes.

three externae (P. M. Cassidy. pers. obs.), but it is plausible
that infestation occurred after spawning. The number of
externae was significantly correlated with the host size if
the host was male (Spearman rank correlation test: r s
= 0.413, / = 2.759, df = 37, P < 0.01). but not if it was
female (r, = 0.299, / = 1.331, df = 18. ns).

The frequency distribution per size class of the infested
specimens compared with respect to the healthy ones did
not show any difference in the males (G = 4.816, df = 3,
ns) (Fig. 7 A); a difference (though slight) was found in the
females, where parasites occurred more often within
smaller size classes (G = 6.913. df = 3. P cu. 0.05) (Fig.
7B). When three size classes were distinguished, no be-
tween-sex difference was found in small specimens (G
= 0.004, df = 1, ns). but the difference was significant in
larger classes) intermediate: G = 10.643, df= 1,.P<0.01;
biggest: G = 2.744. df = 1. P ca. 0.05). The minimum
size of parasitized specimens was 1.8 in females and
1.4 mm SL in males.

Peltogaster hoshnuie externae never exceeded two per
individual. They were more frequently found on the left
side of the hermit abdomen (left, center, and right vs. a
uniform distribution: X 2 = 76.513, df = 2, P < 0.001),
and at the proximal end, close to the carapace (proximal,
middle, distal v.v. a uniform distribution: X 2 = 34.241, df
= 1 . P < 0.00 1 ), without any difference between sexes
(side: G = 1.1 18, df = 2, ns; extremity: X 2 = 0.029, df
= 1, ns). Their first point of eruption corresponded to the
position of the second pleopod. In contrast, the externae
of Thihicopletluis ( = Tlioinpsoniu) rcinluinli. ranging from
1 to 23, were more clumped on the host, equally distrib-
uted on the right and left halves of the hermit body, and
more diffused, involving the dorsal side of the abdomen,
the cephalothorax, and even the pereiopods and chelipcds.

LIFE HISTORY OF A TUBE-DWELLING HERMIT CRAB

73

70 -

24 3.2

SIZE CLASSES (SL, mm)

UNINFECTEDn = 102

INFECTED n = 40

B

24 32

SIZE CLASSES (SL, mm)

I I UNINFECTEDn=119

INFECTED n = 20

Figure 7. Size frequency distributions compared between hermits
that were either uninfected or infected by rhizocephalan parasites in
males (A) and females (B).

Parasites ' effects on the host external morphology and
behavior

To evaluate the effect of parasites on their hosts, we
examined various aspects of the hermit external mor-
phology. The pleopod number did not show any signifi-
cant difference between infested and noninfested speci-
mens in either sex (distinguishing animals with 4, 3, and
less than 3 pleopods, males: G = 1 .645, df = 2, ns; females:
G = 0.31, df = 2, ns). However, the second pleopod was
absent more often in parasitized specimens (7 out of 16
vs. 5 out of 54: G= 8.333, df = l,P< 0.01). The difference
was more pronounced in the males (6 out of 12 vs. 5 out
of 3 1 : G = 4.576, df = 1 , P < 0.05) than in females ( 1
out of 5 v.v. out of 23: G = 1 .839, df = 1 , ns).

The relative growth of the DE of the major chela, one
"maleness" character, was analyzed. No difference was
found between parasitized and unparasitized specimens.

either in males (after a In-ln transformation: b, 0.63 vs.
0.93,? = 1.772, df = 56, ns; a, 0.21 vs. -0.15,? = 1.847,
df = 57. ns) or in females (/>, 0.52 vs. 0.57, / = 0.237, df
= 63, ns: a, 0.40 vs. 0.33. / = 0.01 1, df = 64, ns). Anal-
ogously, parasites seemed not to affect hermit relative body
weight in either sex (after In-ln transformation. SL vs.
weight without chelae, males: b, 3.45 vs. 5.08, / = 1.539,
df = 39, ns; a, -1.13 vs. -3.31, / = 0.043, df = 40, ns;
females: b, 1.70 vs. 3.59, / = 0.778, df = 38, ns; a, 1.32
vs. - 1 .04, / = 1 .2 1 6, df = 39, ns). The same was also true
when cheliped weight was examined (males: b, 8.80 vs.
7.26, / = 0.537. df = 24, ns; a, -15.41 vs. -11.97,
t = 0.858, df = 25, ns: females: b, -2.19 vs. 6.41, / = 1.664,
df = 22, ns, a, 13.89 v.v. -9.34, / = 1.043, df = 23, ns).

Several parasitized hermits were seen molting and pre-
served their externae after the ecdysis, even when externae
belonged to the latest stages, already containing oocytes
and embryos (Liitzen. 1992).

One possible behavioral effect of parasites is lethargy,
which might reduce the ability of naked hermits to find
empty tubes. However, in pilot experiments in the labo-
ratory, parasitized and unparasitized D. schmitti individ-
uals were about equally matched when competing for a
single empty tube. In the field, no difference was seen in
the relative opening diameter of the occupied tubes be-
tween hermits belonging to the two conditions (males: b,
t = 0.025, df = 64, ns, a, t = 0. 169, df = 65, ns; females:
b. t = 0.86, df = 68, ns, a,t= 1 .9 1 3, df = 69, ns).

Parasitized specimens of both sexes placed inside
transparent tubing were seen fanning the externae of Pel-
togaster boschiuae. The behavior was the same as that
already described for ovigerous females fanning their eggs.

Discussion

The structure oj the D. schmitti population

In the species of hermit crabs studied to date (Table
IV), sex ratio in relation to size mostly follows the "anom-
alous" pattern described by Wenner (1972). This implies
that sexes in small size classes are approximately balanced,
a large excess of females is found in intermediate size
classes, and an excess of males is found in the largest ones.
Exceptions are reported by Wenner ( 1972) in Clibanarius
zebra and Calcinus latens, by Abrams (1988) in Pagurus
ochoiensis and P. aleitticus. and by Gherardi and Mc-
Laughlin (1994) in Calcinus laevimanus from the Mas-
carenes. A further exception is D. schmitti. in which an
equal number of males and females are represented in
each size class.

One obvious bias in the size distribution analysis is the
large- or small-scale habitat segregation between sexes.
Species have been reported to show between-sex differ-
ences in habitat utilization (e.g., males of Pagurus hir-
siitiuscnlits occupy high tidepools, but females are dom-

74

F. GHERARDI AND P. M. CASSIDY

Table IV

Species ot hermit crabs reported from the literature following the
"anomalous" pattern (Wetiner. 1972) in the se.\-ratio-lo-si:e relation

Genus

Species

Reference

Coenobila coinpressus Wenner. 1972

Calcinux laevimanus Wenner, 1972

lalens Gherardi and McLaughlin, 1994

Clibananus digueti Harvey, 1988

erythropus Gherardi, 1991

laevimanus Gherardi el at., 1994

Intmilt.s Gherardi and McLaughlin, 1994

Diogenes breviroslris Walters and Griffiths, 1987

Elassochims tenuimanita Abrams, 1988

Pagurisles turgidus Abrams, 1988

Pagurux grano.si mii/uis Abrams, 1988

hirsutiusculus Abrams, 1988

samuelis Abrams, 1988

kenncrly Abrams. 1988

beringanus Abrams, 1988

dalli Abrams, 1988

inant in microhabitats without standing water at low tide;
Abrams, 1988). The size of clustering species is also seg-
regated within clumps (in Clibanarius laevimanus, Ghe-
rardi et ai, 1994). However, both sexes of/), schmitti are
restricted within sabellarian bioherms (Gherardi and Cas-
sidy, 1994b), starting from the late megalopa stage, and
although the species has a contagious distribution (Ghe-
rardi and Cassidy, 1994b), clumps do not significantly
differ in either sex ratio or size.

The sexual selection hypothesis

Under the rationale of the sexual selection hypothesis
( Bertness, 1 98 1 a), the between-sex balance in the size dis-
tribution of D. schmitti implies that the two sexes get the
same benefits (or handicaps) from larger dimension.

Reproductive potential might be enhanced with size.
From the perspective of D. schmitti females, clutch size
significantly increases with the body mass, and larger fe-
males also bear more voluminous eggs.

Larger size might also provide a higher reproductive
potential to males. A sexual dimorphism was evident in
the major chela dimensions (dactyl length, and palm
length and width): the chela (especially the biggest) was
more massive in the male than in the female. The func-
tional significance of this sexual difference has been widely
discussed for Brachyura (Hartnoll, 1974), where it was
related to the use of chelipeds in territorial defense, com-
bat, display, and courtship. In several hermit crabs, males
showed complex precopulatory behaviors, involving the
chelae, for example, either rotating and shaking the female
(Diogenidae) or jerking her toward himself (Paguridae)
(Hazlett, 1966, 1968). Sexual behavior has not yet been

observed in D. schmitti. but the importance for males of
having larger chelipeds might be associated with the in-
trasexual competition to mate. Chelipeds are widely used
in aggressive interactions, both in displays (cheliped ex-
tension, waving, and wig-wag display; F. Gherardi, in
prep.), and in fights (hits and grasps), where the bigger
and stronger the chelipeds are, the more likely the hermit
is to win.

In hermit crabs, factors that could reduce the tendency
to grow are the interspecific competition for shells and
the scarcity of large housings within the habitat. By its
ability to occupy empty polychaete tubes as a new housing,
D. schmitti has freed itself from the harsh war for shells
that occurs within the subtidal hermit crab assemblage in
northern Puget Sound (Abrams et ai, 1986). Its small
relative size must have preadapted this species to this nar-
row microhabitat, but its body mass is certainly con-
strained by the size distribution of the available empty
tubes. In his ecological notes on the endemic Bermuda
hermit Ca/cinus verrilli. Markham (1977) observed that
the mean size of the crabs occupying attached vermetid
shells was far smaller than that of crabs in mobile Ceri-
thium shells. Members of the D. schmitti population an-
alyzed here occupy the largest tubes at their disposal in
the bioherm, and size in both sexes was positively cor-
related with tube opening, suggesting that crabs must
change their housing with growth (Gherardi and Cassidy,
1994b).

The growth hypothesis

The growth hypothesis (Abrams, 1988) refers to the
between-sex difference in the available energy for growth;
the male-biased sex ratio in larger size classes in most
hermit species is attributed to the additional energy that
males can allocate to growth because they do not have to
produce eggs (Bertness, 1981b). Data are still missing for
the extent of growth through molts in D. schmitti and its
energy-time budget is unknown, but a number of clues
suggest that the distribution of the rhizocephalan parasites
might affect growth in this species.

In the population we examined, the extent of infestation
and parasite prevalence varied significantly between sexes,
reaching in the males an average of 2.9 externae per in-
dividual and a percentage of 28 infested specimens. Prev-
alence is unaffected by the male host size, but the fre-
quency of infested females decreases in the intermediate
and larger size classes, where the infestation is significantly
less diffused than in similar sized males.

Within the framework of the growth hypothesis, one
likely conclusion drawn from these data is that if (1) the
males are more frequently infected than the females, and
if (2) parasites cause a reduction in the growth rate of the
host, then the two sexes grow to the same extent because

LIFE HISTORY OF A TUBE-DWELLING HERMIT CRAB

75

the energy the females expend in producing eggs equals
that which the males consume to support parasites. How-
ever, the two assumptions require further clarification and
open new questions.

First, we do not know why the parasites are unequally
distributed between the sexes. The observed pattern could
not be explained by either an increased mortality rate of
infested females or the occurrence of sex reversal, because
the sex ratio was 50% in all the size classes. The attachment
of the parasite larvae may be impeded by the efficiency
of cleaning and grooming (Bauer, 1981), but the two sexes
did not differ in either the extent or the modes of cleaning
behavior (Gherardi, 1994). As a third explanation, im-
munological responses by the hosts might vary between
sexes. Parasitized, but not normal, Carcimis mediterra-
neus have a substance in their blood that fixes complement
in the presence of extracts ofSacculina (reviewed in Bang.
1983). However, in that parasitic relationship, electro-
phoregrams did not show any marked difference between
the parasitized males and females (Herberts. 1978). D.
schmitti females differ in their susceptibility to infection
according to their reproductive states. No parasitized fe-
males have been found in ovigerous condition (other ex-
amples in Hoggarth, 1990, and Liitzen and Jespersen,
1992; exceptions in Hoeg and Liitzen, 1985); one expla-
nation is that parasitized females lose their eggs after a
few days (Liitzen and Jespersen, 1992), but the reasons
remain unknown.

The second assumption, that growth rate of the host is
affected by the parasite, is supported by the previous lit-
erature on rhizocephalan infestation (O'Brien and Van
Wyk, 1984; Hawkes et ai. 1986; Hoggarth, 1990; Abello
and Macpherson, 1992; Bang, 1983; Overstreet, 1983).
Nevertheless, a direct investigation of molt frequency is
lacking and figures on the relative increase at ecdysis
compared between infected and uninfected individuals
are provided only by Liitzen and Jespersen ( 1992). Our
findings that parasites do not inhibit molting in D. schmitti
or influence either body or cheliped weight in either sex
make the growth hypothesis questionable, at least in this
species.

Other effects of parasites

A variety of morphological and behavioral alterations
exhibited by rhizocephalan-infected decapods and the
hormonal involvement in those phenomena are exten-
sively described by Hartnoll (1967), Nielsen (1970), and
Phillips and Cannon (1978) among others (see, e.g.. bib-
liography by Overstreet, 1983). D. schmitti males do not
undergo the process of feminization observed in other
species (Hartnoll, 1982; O'Brien and Van Wyk. 1984). as
evidenced by the preservation of some "maleness" char-
acters (e.g., the high relative depth of the major chela).

The only alteration is the frequent absence of the second
pleopod, which cannot result from an attempt by the par-
asite to provide a safe accommodation for the externae,
but seems instead to be a consequence of the eruption of
Peltogaster boschmae externae within the soft tissue lining
the host abdomen, which corresponds to the attachment
point of the second pleopod.

Neither do infected hermits exhibit behavioral altera-
tions, such as lethargy, that could decrease their ability
in direct or exploitative competition: in the laboratory,
parasitized and healthy D. schmitti had the same proba-
bility of getting an empty polychaete tube, and in the
field, they occupied equally sized housings. Besides, rel-
ative weight, and thus possibly feeding efficiency, was un-
affected by the presence of rhizocephalans. The only be-
havioral result of parasite manipulation is the initiation
of mock parental care, in which infected hermits of both
sexes ventilate Peltogaster boschmae externae in the same
way that gravid females ventilate their eggs.

Reproductive patterns

D. schmitti females attain maturity at a relatively small
size: the smallest egg-bearing specimen measured 1 . 1 mm
SL. On the other hand, the allometry of chela growth
should indicate that maturity (at least, functional matu-
rity; Hartnoll, 1969) occurs in males at larger size (over
3.4 mm SL).

A precocious onset of sexual behavior in females has
been reported in the decapod literature and associated
with a reduced possibility of encountering males; in the
parasitic females of the Pinnotheridae (Christensen and
McDermott, 1958) and in the freshwater crab Potamon
Jhtviatile(Miche\i et al.. 1990) copulation can occur even
in prepubertal females, and sperm are kept in the seminal
receptacles until ovulation.

A second remarkable feature of reproduction is the low
frequency (50%) of gravid females in all the size classes.
This is particularly evident when we consider that D.
schmitti breeds only once per year (Nyblade, 1974), and
that the breeding period (January-April) is short, but the
time necessary for eggs to mature is relatively long (ex-
ceeding, on average, 1 month).

One explanation is that due to the shortage of food,
females may have limited energy for producing clutches,
causing them to skip the reproductive season. If this were
the case here, we should expect a gradient in the clutch
size depending on the available energy. Nonetheless, egg
number is a function of female size, and the latter is not
related to feeding efficiency (Gherardi, 1 994). In addition,
the annual egg production (52.8 mg of eggs per 100 mg
female weight per year; Nyblade 1974) is high compared
with that of the other hermit crabs in northern Puget
Sound.

76

F. GHERARDI AND P. M. CASSIDV

Another hypothesis refers again to the difficulty that
this sedentary species encounters in finding a mate. D
schmitti is gonochoristic and mating in hermit crabs
requires copulation (Hazlett, 1966). hut it is still unclear
how this is effected in this species. Males, females, or
both are assumed to leave the attached tube and roam
about with their abdomens naked (or at best in broken
pieces of tubes; Nyblade as reported by Caine, 1980) to
seek receptive mates. Despite the clumped distribution
of the population, this is a risky behavior; in the labo-
ratory, wandering hermits inhabiting loose tubes are
easy prey for the crabs and fishes (F. Gherardi, in prep.),
that frequent sabellarian bioherms (Gherardi and
Cassidy. 1994a).

Hatching lasts from 1 to 6 nights, the length of time
being related to the overall number of larvae. This suggests
either that development of embryos belonging to the same
batch is out of phase or that hatching is controlled by the
embryos themselves, by the females, or by both (Saigusa,
1992). Such an extension of hatching in sequential bursts
might be a mechanism to allow survival of at least a num-
ber of larvae in a difficult, predator-filled, and unpredict-
able environment, such as the current-swept channels of
Puget Sound.

In D. schmitti. hatching occurs exclusively at night,
possibly to minimize predation on the newly released lar-
vae by diurnal fishes. In contrast to the other decapods
inhabiting enclosed habitats (estuaries and mangrove
swamps; Forward, 1987; Hartnoll. 1988). in this species
larval release is synchronized with neap tides, when the
tidal current is consistently lower than in the spring phase.
This timing seems to be controlled by an endogenous
clock (De Vries and Forward, 1989). persisting under lab-
oratory conditions in which the tidal cycle corresponding
to the rhythm is absent. This pattern of larval release seems
unrelated to salinity tolerance (Forward el ai. 1982), be-
cause salinity is nearly constant in the examined area
(SPMC, 1992). Its adaptive meaning is suggested by D.
schmiiii '.v behavioral ecology. For this species so depen-
dent upon a habitat (sabellarian bioherms) that is rare
and quite unpredictable (Gherardi and Cassidy. 1994a)
it is more beneficial if larvae are retained within the basin
near the parental population than if they are flushed out
to open waters for planktonic development (see Mc-
Conaugha. 1992. for a discussion of larval retention vs.
dispersal in decapods).

Acknowledgments

We thank Dr. Jorgen Liitzen (University of Copen-
hagen) who kindly identified parasites of D. schmitli. The
study was encouraged by Dr. Patsy A. McLaughlin
(Shannon Point Marine Center, WWLJ), to whom we are
greatly indebted. Part of the work was conducted at the

Shannon Point Marine Center, Anacortes, Washington.
Partial funding was provided by M.U.R.S.T. to the first
author.

Literature Cited

Abello, P., and F. Macpherson. 1992. Epihiosis and rhizocephalan in-
festation patterns in relation to the rep-oductive biology of Lithodes
/<TA (F : ilhol. 1885) (Anomura: Lithodidae). / Crustacean Biol 12:
561-570.

Ahrams, P. A. 1988. Sexual difference in resource use in hermit crabs;
consequences and causes. Pp. 283-296 in Behavioural adaptation to
l/ic liueriidal Lite. G. Chelazzi and M. Vannmi, eds. Plenum Press,
New York.

Abrams, P., C. Nyblade. and S. Sheldon. 1986. Resource partitioning
and competition tor shells in a subtidal hermit crab species assemblage.
Oecolofita 69: 424-445.

Bans, F. B. 1983. Crustacean disease responses. Pp. 1 13-153 in The
Biology ol Crustacea, Vol. 6. Pathobiology, A. J. Provenzano. Jr.,
ed. Academic Press, New York.

Bauer, R. T. 1981 . Grooming behavior and morphology in the decapod
Crustacea. J Crustacean Biol. 1: 153-173.

Bertness. M. D. I981a. Interference, exploitation, and sexual com-
ponents of competition in a tropical hermit crab assemblage. J. Exp.
Mar Biol Lcol 49: 189-202.

Bertness, M. I). 1981b. Competitive dynamics of a tropical hermit
crab assemblage. Ecology 62: 751-761.

Caine, K. A. 1980. Adaptations of a species of hermit crab (Decapoda.
Paguridea) inhabiting sessile worm tubes. Cruslaceana 38: 306-310.

Christensen, A. M., and J. J. McDermott. 1958. Life-history and bi-
ology of the oyster crab. Pinnotheres uslreiun Say. Biol Bull. 114:
146-174.

De Vries, M. C., and R. B. Forward, Jr. 1989. Rhythms in larval release
of the sublittoral crab Neopanope sayi and the supralittoral crab Se-
\arnia ciiiereuni (Decapoda: Brachyura). Mar. Biol 100: 241-248.

Forward. R. B., Jr. 1987. Larval release rhythms of decapod crusta-
ceans: an overview. Bull. Mar Sci. 41: 165-176.

Forward, R. B., Jr., K. I.ohmann, and T. VV. Cronin. 1982. Rhythms
in larval release by an estuarine crab (Rhithropanopew liarrisii). Biol.
Hull 163: 287-300.

Gherardi. F. 1991. Relative growth, population structure, and shell-
utili/ation of the hermit crab Clihanarius eryihropns in the Medi-
terranean. Oehalia 17: 181-196.

Gherardi, F. 1994. Multiple feeding techniques in the sessile hermit
crab Discorsopagurus schmitti. inhabiting polychate tubes. Oecologia
98: 134-146.

Gherardi, F., and P. M. Cassidy. I994a. Macrobenthic associates to
the polychaete Sahellaria eenientanuin bioherrn from northern Puget
Sound, Washington. Can J /.ool. 72: 516-525.

Gherardi. F., and P. M. Cassidy. 1994b. Sabellarian tubes as the housing
o( Discorsopagurus scliniilti. Canad. J Zoo/ 72: 526-532.

Gherardi. F., and P. A. McLaughlin. 1994. Shallow-water hermit crabs
from Mauritius and Rodrigues Islands, with the description of a new
species ofCalcinus. Rallies Bull 7.ooi 42: 613-636.

Gherardi, F.. F. Zatteri, and M. Vannini. 1994. Hermit crabs in a man-
grove swamp: the structure of Clihanarius laevimaniix clusters. Mar.
Biol 121:41-52.

I hum, >ll, R. G. 1967. The effects of sacculinid parasites on two Ja-
maican crabs. ./ Linn Soc London /.ool. 46: 275-295.

1 1. ii in, ,11 R. G. 1969. Mating in the Brachyura. Crustaceana 16: 1 1-
181.

llarlnoll, R. G. 1974. Variation in growth pattern between some sec-
ondary sexual characters in crabs ( Decapoda Brachyura). C 'rustaceana
27: 131-136.

LIFE HISTORY OF A TUBE-DWELLING HERMIT CRAB

77

Hartnoll, R. G. 1982. Growth. Pp. 1 1 1 - 1 96 in The Biology of Crustacea,
Vol. 2, Embrvologv, Morphology and Genetics, L. G. Ahele. ed. Ac-
ademic Press. New York.

Hartnoll, R. G. 1988. Eco-ethology of mangroves. Pp. 477-489 in Be-
havioural Adaptation to the Intertidal Life, G. Chelazzi and M. Van-
nini, eds. Plenum Press. New York.

Harvey, A. \V. 1988. Size- and sex-related aspects of ecology of the
hermit crab Clibananus digueti Bouvier (Decapoda: Anomura: Di-
ogenidae). Ph.D. dissertation. University of Arizona, Tucson, AZ.

Hawkes, C. R., T. R. Meyers, T. C. Shirley, and T. M. Koeneman.
1986. Prevalence of the parasitic barnacle Briarosaccus cal/osus on
king crabs of southeastern Alaska. Trans. Am. Fish. Soc. 115: 252-
257.

Hazlett, B. A. 1966. Social behaviour of the Pagundaeand Diogenidae
of Curacao. Stud Fauna Curasao Other Caribb. 1st. 23: 1-143.

Hazlett, B. A. 1968. The sexual behavior of some European hermit
crabs (Anomura: Paguridae). Pubbi Sin. Zool. Napoli 36: 238-252.

Herberts, C. 1978. Relation hote-parasite entre Carcinus mediterraneus
and Sacculina carcini: analyse immunochimique el mise en evidence
d'une precipitine antisacculine. C. R Acad. Sci. 286: 725-728.

Heeg, J. T., and J. Liitzen. 1985. Crustacea Rhizocephala. Pp. 1-92
in Marine Invertebrates of Scandinavia. Vol. 6. Norwegian University
Press, Oslo.

Hoggarth, D. D. 1990. The effects of parasitism by the rhizocephalan,
Briarosaccus ca/losus Boschma on the lithodid crab. Paralomis gran-
ulosa (Jacquinot) in the Falkland Islands. Cnistaceana 59: 1 56-170.

Liitzen, J. 1992. Morphology of Thompsonia reinhardi. new species
(Cirripedia: Rhizocephala). parasitic on the northeast Pacific hermit
crab Discorsopagurus schmitti (Stevens). J Crustacean Bin/. 12: 83-
93.

Liitzen, J., and A. Jespersen. 1992. A study of the morphology and
biology of Thompsonia littoralis (Crustacea: Cirripedia: Rhizo-
cephala). Ada Zool 73: 1-23.

Margolis, L., G. W. Esch, J. C. Homes, A. M. Kuris, and G. A. Schad.
1982. The use of ecological terms in parasitology. J Parasitol 68:
131-133.

Markham. J. C. 1977. Preliminary note on the ecology of Calcinut
verrilli, an endemic Bermuda hermit crab occupying attached ver-
metid shells. J. Zool. (Land.) 181: 131-136.

McConaugha, J. R. 1992. Decapod larvae: dispersal, mortality, and

ecology. A working hypothesis. Am. Zool. 32: 512-523.
McLaughlin, P. A. 1974. The hermit crabs (Crustacea Decapoda. Pa-

guridea) of Northwestern North America. Zool I'erhand 130: I-

396.

Micheli, F., F. Gherardi, and M. Vannini. 1990. Growth and repro-
duction in the freshwater crab, Polamon fluvialile (Decapoda.

Brachyura). Freshwater Biol 23:491-503.
Nielsen, S. O. 1970. The effects of the rhizocephalan parasites Pelto-

gaster paguri Rathke and Gennnosticciis siilcalus (Lilljeborg) on five

species of paguridan hosts (Crustacea Decapoda). Sarsia 42: 17-32.
Nyblade, C. F. 1974. Coexistence in Sympalric Hermit Crabs. Ph.D.

dissertation. University of Washington, Seattle. WA.
O'Brien, J., and P. Van Wyk. 1984. Effect of crustacean parasitic cas-

trators (Epicandean isopods and Rhyzocephalan barnacles) on growth

of crustacean hosts. Crustacean Issues 3: 191-218.
Overstreet, R. M. 1983. Metazoan symbionts of crustaceans. Pp. 155-

250 in The Biology of Crustacea. Vol. 6, Pathobiology. A. J. Proven-

zano. Jr., ed. Academic Press, New York.
Phillips, W. J., and L. R. G. Cannon. 1978. Ecological observations

on the commercial sand crab. Portumis pelagicus (L.), and its parasite.

Sacculina gram/era Boschma, 1973 (Cirripedia: Rhizocephala). /

Fish. Dis. 1: 137-149.
Saigusa, M. 1992. Control of hatching in an estuarine terrestrial crab.

I. Hatching of embryos detached from the female and emergence of

mature larvae. Biol. Bull. 183: 401-408.
Siegel, S. 1956. Nonparametric Statistics for the Behavioral Sciences.

McGraw-Hill, New York.
SPMC. 1992. U'ater Quality Record. Annual Report from Shannon

Point Marine Center, Anacortes. WA.
Walters, W. L., and C. L. Griffiths. 1987. Patterns of distribution,

abundance and shell utilization amongst hermit crabs, Diogenes bre-

virostris S Afr. J Zool. 22: 269-277.
\Venner, A. M. 1972. Sex ratio as a function of size in marine Crustacea.

Am. Nalitr. 106: 321-350.
Zar, J. H. 1984. Biostatistical Analysis. Prentice-Hall, Englewood Cliffs,

NJ.

Reference: Biol Bull 188: 78-82. (February/March, 1995)

Taurine-like Immunoreactivity in the Motor Nerve Net
of the Jellyfish Cyanea capillata

MATS CARLBERG 1 , KARIN ALFREDSSON 1 , SVEN-OLLE NIELSEN 1 ,
AND PETER A. V. ANDERSON 2

^Department of Zoology, University of Lund, Helgonavdgen 3, S-223 62 Lund, Sweden, and

2 Whitney Laboratory and Departments of Physiology and Neuroscience.

University of Florida. Si. Augustine, Florida 32086

Abstract. Two antisera against the sulfonated amino
acid taurine were applied to subumbrella tissue of the
jellyfish Cyanea capillata. Taurine-immunoreactive nerve
nets were found in both the ectoderm and endoderm. The
ectoderm had two morphologically and immunocyto-
chemically distinct populations of neurons, the motor
nerve net (MNN), which was immunoreactive to the tau-
rine-like molecule, and the diffuse nerve net (DNN), which
was immunoreactive to the neuropeptide Phe-Met-Arg-
Phe-NH : (FMRFamide). In the endoderm, immunoreac-
tivity was found in the endodermal DNN. This localiza-
tion was confirmed by double-labeling experiments, which
also revealed that the endodermal DNN neurons may
contain both taurine and FMRFamide-related peptide.
The presence of a taurine immunoreactivity in the MNN
supports the hypothesis that taurine or some chemically
related compound is the neurotransmitter at synapses
within the MNN of Cyanea.

Introduction

Cnidarians are the earliest extant animals to have a
nervous system and, as such, they may provide useful
information about the evolution of the nervous system
and its components. Furthermore, their structural sim-
plicity affords opportunities for studying functional as-
pects of these nervous systems, including the cellular
mechanisms underlying chemical synaptic transmission
(Anderson, 1985; Spencer et ai, 1989; Anderson and
Spencer, 1989). A focus of considerable interest in recent
years has been the identity of neurotransmitters in the
Cnidaria. Neuropeptides are known to be common within

Received 24 May 1994; accepted 27 October 1994.

the phylum (Grimmelikhuijzen et ai. 1989a, b: 1992),
and evidence for a role of small molecules and amino
acids as neurotransmitters is growing (Anctil, 1989;
Scemes, 1989; Chung el ai. 1989; Chung and Spencer,
1990; Umbriaco et ai, 1990), but remains limited.

The sulfonated amino acid taurine, which is ubiquitous
in animals and prokaryotes and has been implicated as
an inhibitory neurotransmitter in both vertebrates (Huxt-
able, 1989) and invertebrates (Nistri and Constant!, 1976;
Hue?/ a/.. 1979; Giles and Usherwood, 1985), has recently
been shown to depolarize neurons in the motor nerve net
(MNN) of the scyphozoan jellyfish Cyanea capillata (An-
derson and Trapido-Rosenthal, 1990). The mode of action
of taurine on these neurons is very similar to that of the
endogenous neurotransmitter, raising the possibility that
taurine may serve as an excitatory neurotransmitter at
these synapses. To determine whether taurine is present
in the tissues of Cyanea and if so, to delineate its distri-
bution, we used antisera raised against a taurine-bovine
serum albumin complex (Campistron et ai. 1986: Madsen
cl ai. 1985). The results indicate that taurine, or a taurine-
like molecule, is indeed present in the MNN and that its
distribution is consistent with a role as a neurotransmitter
in the Cyanea MNN.

Materials and Methods

Specimens of Cyanea were collected at the Tja'rno Ma-
rine Biological Laboratory on the west coast of Sweden.
Pieces of perirhopalial tissue (Anderson and Schwab,
198 1 ) were removed from the animal and pinned out to
prevent curling. In some preparations the myoepithelium
that envelops the MNN neurons was removed to expose
the nerve net (Anderson and Schwab, 1 984). Tissues were

78

TAURINE-LIKE MATERIAL IN NERVES OF A JELLYFISH

79

fixed for 3 h in freshly prepared 5% glutaraldehyde in
0.05 M Na-cacodylate buffer containing 1% sodium
metabisulphite and 2.4% sodium chloride (pH 7.5). After
fixation, the tissues were given three 15-min washes in
Tris-buffered saline (TBS; 0.05 A/TRIS-HCI buffer, pH
7.5, 1% sodium metabisulphite and 2.4% sodium chlo-
ride), followed by 30 min in 0.1 M sodium borohydride
in TBS, then a further three 1 5-min rinses in TBS. Twelve
specimens from 5 to 20 cm in diameter were used for
immunocytochemical investigations.

Samples were incubated for 4-6 days in rabbit anti-
taurine antisera diluted 1:200 (Chemicon) or l:1000(Im-
munotech S.A.) in TBS with 0.2% Triton X-100 (TBS/
TX) and 1% bovine serum albumin (BSA). After three
15-min rinses in TBS/TX, the samples were incubated
for 3 h with fluorescine isothiocyanate- (FlTC)-conjugated
swine anti-rabbit IgG (Dakopatts, Denmark) diluted 1:10
in TBS. The samples were then given three 15-min rinses
in TBS, stained in a 1% solution of Evans blue (Merck)
in phosphate-buffered saline (PBS), pH 7.4, rinsed for 2 h
in PBS, and mounted in phosphate-buffered glycerol.
Specificity was tested by preabsorbtion of antiserum with
1 [iM taurine-glutaraldehyde-BSA conjugate.

Double labeling with rabbit antisera raised against tau-
rine and the neuropeptide FMRFamide was carried out
in the manner developed by Wiirden and Homberg
(1993). Specifically, tissues were first stained with anti-
bodies to taurine. and the location of the primary antibody
was visualized by a 3-h incubation with Texas-red-con-
jugated donkey anti-rabbit IgG (Jackson Immuno Re-
search) diluted 1:40. After a 3-h incubation in rabbit IgG
(Dakopatts) diluted at 1:25, the tissues were then incu-
bated for 24 h with biotinylated (Bayer and Wilcheck.
1980) anti-FMRFamide antibodies (Incstar), diluted 1:
800. The FMRFamide immunostaining was then visu-
alized by treatment with streptavidin-FITC (Dakopatts)
at 1 :20 for 3 h. All light microscopical observations were
made with a Leitz Aristoplan microscope.

The specificity of taurine antiserum from Chemicon
has been characterized by the company. Cross-reactivity
with glutaraldehyde-conjugated hypotaurine was 0.067
(1:15) and was less than 0.002 for other glutaraldehyde-
conjugated amino acids including GABA, beta-alanine,
aspartate, glycine, cysteine. and glutamate.

Results

Light microscopy

Taurine-like immunoreactivity (Tau-IR) was found in
both the ectoderm and endoderm of the perirhopalial tis-
sue of Cyanea.

Ectodermal-specific Tau-IR was found in neurons and,
to a lesser extent, in myoepithelial cells. Ectodermal myo-
epithelial cells in this species contain a large central vac-

uole (Anderson and Schwab, 1981). Immunoreactivity
was restricted to the narrow layer of cytoplasm that sur-
rounds each vacuole; vacuolar contents were not im-
munoreactive. The Tau-IR neurons were large, bipolar
cells with lengths up to 2 mm. cell-body diameters of 15
to 20 nm, and axonal diameters from 1 to 5 /urn (Fig. 1 A).
These were clearly motor nerve net (MNN) neurons (An-
derson and Schwab, 1981) and were easily distinguished
from the FMRFamide-immunoreactive (FMRF-IR) cells
that form the diffuse nerve net (DNN) (Fig. 1C), the other
nerve net present in the perirhopalial tissue ectoderm. In
the MNN, synapses occur wherever two neurons are in
physical contact with one another (Anderson, 1985), and,
as can be seen in these micrographs (Fig. IB), such con-
tacts are abundant. The Tau-IR within the MNN was
restricted to the perirhopalial tissue; although MNN neu-
rons are known to extend into the radial and circular
muscle bands that surround the perirhopalial tissue (An-
derson and Schwab, 1981), no Tau-IR neurons were found
in the radial or circular muscle bands.

In the endoderm. at least two cells types were immu-
noreactive. One was a population of bipolar neurons.
These cells, which had cell-body diameters of 1 to 15 ^m
and axon diameters of 0.5 to 2 ^m, were at least 0.6 mm
long and formed a loose nerve net (Fig. ID). Their overall
appearance is consistent with that of the diffuse nerve net
(DNN) known to be present in this tissue. The other ob-
viously immunoreactive endodermal cell type was more
difficult to characterize. The cells in question occurred
relatively densely, and their immunoreactivity appeared
as a rather amorphous, frequently circular mass. Whether
this mass represents an intracellular compartment of the
cell or the true dimensions of the cell was not clear. Both
of these cell types were surrounded by a low level of back-
ground immunoreactivity interspersed with occasional
nonfluorescent areas that are presumably spaces in the
endodermal epithelium.

Preabsorbtion of antiserum with taurine-GA-BSA con-
jugate completely abolished all immunostaining.

Double labeling

Double labeling revealed two distinct nerve nets in the
ectoderm. Again. Tau-IR was restricted to MNN neurons,
whereas FMRF-IR was localized to a separate population
of smaller, multipolar cells (Fig. 2A). FMRF-IR was also
evident in the marginal rhopalia and in regions covered
by the circular and radial muscle bands. At no time were
the two signals co-localized in the ectoderm.

In the endoderm, both antibodies stained what ap-
peared to be the DNN. In smaller animals, the neurons
were FMRF-IR, but in larger animals they were apparently
Tau-IR. In one specimen, both FMRF-IR and Tau-IR
were evident in the same cells, indicating co-localization
(Fig. 2B).

80

M. CARLBERG ET AL

Figure 1. Whole-mount immunostaming of penrhopalial tissue of Cyanea capillata. (A) Low power
micrograph of Tau-IR in the ectoderm. Neurons in the MNN stain readily. Scale bar = 0. 1 mm. (B) Micrograph
of the Tau-IR MNN in the ectoderm. Several apparent contact sites between the axons (arrows) and thinner
elements twined together (arrowheads) can be seen. Scale bar = 50 ^m. (C) Low-power micrograph of
FMRFamide-immunoreactive diffuse nerve net (DNN) in the ectoderm. Scale bar = 0.1 mm. (D) Tau-IR
in the endoderm of the penrhopalial tissue. Immunoreactivity was present in bipolar neurons and in un-
differentiated endodermal cells. Scale bar = 50 nm.

Discussion

In cnidarians, nerve nets are located under an overlying
epithelium that forms a permeability barrier for phar-
macological agents and microelectrodes. The ectodermal
MNN in the penrhopalial tissue of Cyanea is one of the
very few instances in which a coelenterate nerve net can
be exposed, permitting access for electrophysiological and
pharmacological studies (Anderson and Schwab. 1984:
Anderson, 1985). The size of the neurons makes them
suitable for electrophysiological recordings. The accessi-

bility of this nerve net and, in particular, its synapses pro-
vides a useful preparation for studying the pharmacology
of chemical neurotransmission in a cnidarian.

The MNN is a plexus of large bipolar neurons that
innervates the swimming muscle bands and serves as the
pathway to coordinate swimming motor activity. Synapses
between the MNN neurons are fast, chemical synapses
that are bidirectional (Anderson, 1985). Previous work
(Anderson and Trapido-Rosenthal, 1990) has implicated
taurine or a closely related molecule as a potential neuro-

TAURINE-LIK.E MATERIAL IN NERVES OF A JELLYFISH

81

Figure 2. Double labeling. (A) Whole mount of perirhopalial tissue
ectoderm, stained with antibodies against taurine (red) and FMRFamide
(green ). The clear anatomical separation between the Tau-IR MNN neu-
rons and FMRFamide DNN neurons is evident. (B) Endodermal tissues
stained in the same manner. The majority of cells in this preparation
are Tau-IR. Two neurons (arrows) were more (yellow) or less (green)
intensely immunoreactive to FMRFamide, but at least one (arrowhead)
was reactive to both antibodies. Scale bars = 50 ^m.

transmitter at these synapses. The current investigation
provides immunocytochemical evidence that a taurine-
like molecule is indeed present in MNN neurons and in
an endodermal nerve net, but is not present, at least in
detectable quantities, in neurons of another ectodermal
nerve net, the diffuse nerve net (DNN). This was partic-
ularly obvious in the double-labeling experiments, in
which the ectodermal FMRFamide-IR was clearly re-
stricted to the DNN (Anderson et ai. 1992). Although
Tau-IR was also present in the ectodermal myoepithelial
cells, its absence in the DNN indicates that taurine is not
a constituent component of all nerve nets in this animal.
This distinction is important because taurine acts as an
osmoregulator in many marine organisms (Thurston et

ii/., 1980). If it were serving the same function in Cyanea,
one might expect it to be widespread in different cell types
and present in all nerve nets.

The presence of Tau-IR in the endodermal DNN in
Cyiineu was unexpected considering that immunoreac-
tivity to antibodies raised against the sea anemone neuro-
peptide AnthoRFamide was found in these neurons
(Anderson ct at.. 1992). In the present study, endodermal
Tau-IR neurons also were found to be immunoreactive
to antibodies to FMRFamide. In addition, however, mor-
phologically similar neurons in larger animals were found
to have Tau-IR, and one specimen showed apparent co-
localization of the two transmitter candidates. It may be
worth further investigations to find out if there is a pro-
gression from an FMRFamide or AnthoRFamide-like
peptide to taurine (or a related compound) as the animal
grows. To meet the requirement for faster transmission
in a large medusa, a switch from peptidergic metabo-
trophic receptors to fast excitatory ionotrophic receptors
would be functional. In either case, it is clear from this
work that cnidarian synapses may have a hitherto un-
appreciated complexity in the number of neurotransmit-
ters present in single neurons.

Tau-IR was also found in a very abundant, non-neu-
ronal cell type in the endoderm (Fig. ID). The identity
of this cell is unclear. The presence of Tau-IR in these
cells, and perhaps all endodermal cells if the light back-
ground fluorescence is indeed indicative of low levels of
taurine, may imply that it has an alternative function such
as osmoregulation. It is also possible, however, that some
of the small circular profiles represent interstitial cells dif-
ferentiating into DNN neurons.

The antibody used in the light microscopical compo-
nent of this study is known to have low cross reactivity
to the most abundant metabolites of taurine including
hypotaurine and cysteine. However, one cannot as yet
exclude the possibility that the antigen is a closely related
compound or a small taurine-containing oligopeptide
(Marnela et a/., 1985). Free taurine is, however, an abun-
dant constituent of MNN neurons (Anderson and Trap-
ido-Rosenthal, unpub.).

The MNN extends over the entire subumbrella surface,
forming a network that connects all eight marginal ganglia,
or rhopalia, with the circular and radial swimming muscle
bands. To do this, the nerve net must transit the muscle
bands, and individual, Lucifer-yellow-filled neurons have
been seen to extend from the perirhopalial tissue into ra-
dial muscle bands (Anderson and Schwab, 1981). How-
ever, Tau-IR neurons were never observed in either the
radial or circular muscle bands. Although the MNN neu-
rons located within the confines of these muscle bands
may employ a different neurotransmitter in this region,
it is also possible that failure to observe Tau-IR in these
areas is due to a technical problem. To get adequate stain-

82

M. CARLBERG F.T AL

ing of neurons in these preparations, they had to be in-
cubated with the antibodies for as long as 6 days. In con-
trast, anti-FMRFamide antibodies usually penetrate the
tissues easily, typically requiring 24 h (Anderson et a/.,
1992). The staining difficulty may reflect either a low an-
tibody liter or poor penetration by anti-Tau antibody. In
either case, the thick layer of muscle that overlies the
MNN in the radial and circular muscle bands may have
compromised the ability of the anti-Tau antibodies to
reach their targets in this region.

The major conclusion of this study is that MNN neu-
rons are Tau-IR. This, together with electrophysiological
evidence that taurine depolarizes the MNN neurons in a
manner consistent with that of the endogenous neuro-
transmitter, provides compelling evidence that taurine,
or a taurine-like molecule, is the neurotransmitter in
Cyanea. This possibility has evolutionary implications.
Taurine is one of the most abundant amino acids in the
animal cell, and it is conceivable that carnivores and
scavengers developed olfactory receptors for taurine very
early in evolution. Receptors for taurine are, indeed,
known to be present on the olfactory antennae of lobsters
(Derby and Atema, 1982), and one could envisage how
neurotransmitter receptors might have developed from
external chemoreceptors (Carr, 1989).

Acknowledgments

We are very grateful to the Director, Dr. Larz Afzelius,
and to Drs. Lars Hagstrom and Benno Magnusson for
providing us with facilities at the Tjarno Marine Station.
This work was supported by grant B-BU 1781-303 from
the Swedish Natural Research Council to Mats Carlberg
and by NSF grant BNS 91091 55 to Peter Anderson.

References

Anctil, M. 1989. The antiquity of monoaminergic neurotransmitters:
evidence from Cnidaria. Pp 141-155 in Evolution of the First Nervous
Systems. P. A. V. Anderson, ed. Plenum Press, New York.

Anderson, P. A. V. 1985. The physiology of a bidirectional, excitatory,
chemical synapse. / Neurophvsiol. 53: 821-835.

Anderson, P. A. V., and W. E. Schwab. 1981. The organization and
structure of nerve and muscle in the jellyfish Cyanea capillata (Coe-
lenterata; Scyphozoa). J. Morphol. 170: 383-399.

Anderson, P. A. V., and W. E. Schwab. 1984. An epithelial cell-free
preparation of the motor nerve net of Cyanea (Coelenterata; Scy-
phozoa). Bid. Bull 166: 396-408.

Anderson, P. A. V., and A. N. Spencer. 1989. The importance of cni-
danan synapses for neurobiology. / Neurobiol. 20: 435-457.

Anderson, P. A. V., and H. G. Trapido-Rosenthal. 1990. Electro-
physiological evidence that a taurine-like amino acid is the neuro-
transmitter at a fast excitatory synapse. Stic. Neitrosci. Ahxir. 16:
1185.

Anderson, P. A. V., A. Moosler, and C. J. P. Grimmelikhuijzen. 1992.
The presence and distribution of Antho-RFamide-like material in
scyphomedusae. Cell Tissue Res. 267: 67-74.

Bayer, E. A., and M. \Vilchek. 1980. The use of avidin-biotin complex
as a tool in molecular biology. Alelh. Biochem. Anal. 26: 1-45.

Campistron, G., M. Geffard, and R. M. Buijs. 1986. Immunological
approach to the detection of taurine and immunocytochemical results.
J Neuroehem. 46: 862-868.

Carr, \V. E. S. 1989. Chemical signalling systems in lower organisms:
a prelude to the evolution of chemical communication in the nervous
system. Pp. 81-94 in Evolution of the First Nervous Systems, P. A. V.
Anderson, ed. Plenum Press, New York.

Chung, J. M., and A. N. Spencer. 1990. On the potassium conductance
increase activated by dopamine in the jellyfish Polyorchis penicillatus.
Soc. Neurosci. Ahstr. 16: 1020.

Chung, J. M., A. N. Spencer, and K. H. Gahm. 1989. Dopamine in
tissues of the hydrozoan jellyfish Polyorchis penicillatus as revealed
by HPLC and GC/MS. / Comp. Physiol. B 159: 173-181.

Derby, C. D., and J. Atema. 1982. Narrow-spectrum chemoreceptor
cells in walking legs of the lobster Homarus americanus: Taste spe-
cialists. J Ctimp Physiol A 146: 181-189.

Giles, D. P., and P. N. R. Usherwood. 1985. The effects of putative
neurotransmitters on somata isolated from neurons of the locust
central nervous system. Comp Biochem. Physiol. SOC: 231-236.

Grimmelikhuijzen, C. J. P., D. Graff, O. kiozumi, J. A. Weslfall, and
1. D. McFarlane. 1989a. Neurones and their peptide neurotrans-
mitters in coelenterates. Pp. 95-109 in Evolution of the First Nervous
Systems. P. A. V. Anderson, ed. Plenum Press, New York.

Grimmelikhuijzen, C. J. P., D. Graff, and I. D. McFarlane. 1989b.
Neurones and neuropeptides in coelenterates. Arch. Histol. Cviol.
52: 265-276.

Grimmelikhuijzen, C. J. P., K. Cartensen, D. Darmer, A. Moosler,
M-P. Nothacker, R. K. Reinscheid, C. Schmutzler, and H. Vollert.
1992. Coelenterate neuropeptides: structure, action and biosyn-
thesis. Am. Zoo/. 32: 1-12.

Hue, B., M. Pelhate, and J. Chanelet. 1979. Pre- and postsynaptic
effects of taurine and GABA in the cockroach central nervous system.
Can. J. A two/. Sci 6: 243-250.

Huxtable, R. J. 1989. Taunne in the central nervous system and the
mammalian actions of taurine. Prog. Neurobiol. 32: 471-533.

Madsen, S., O. P. Ottersen, and J. Storm-Mathisen. 1985. Immu-
nocytochemical visualization of taurine: neuronal localization in the
rat cerebellum. Neurosci Lett. 60: 255-260.

Marncla, K-M., H. R. Morris, M. Panico, M. Timonen, and P. Lah-
desmarki. 1985. Glutamyl-taurine is the predominant synaptic
taurine peptide. ./ \citrochcm. 44: 752-754.

Nistri, A., and A. Conslanti. 1976. The action of taurine on the lobster
muscle fibre and the frog spinal cord. Newopharmacology 15: 635-
641.

Scemes, E. 1989. Rethinking the role of cholinergic neurotransmission
in theCnidana. Pp. 157-166 m Evolution of the First Nervous Systems
P. A. V. Anderson, ed. Plenum Press, New York.

Spencer. A. N., J. Przysiezniak, J. Acosta-Urquidi, and T. A. Basarsky.
1989. Presynaptic spike broadening reduces junctional potential
amplitude. Nature 340: 636-638.

Thurston, J. H., R. E. Hauhart. and J. A. Dirgo. 1980. Taurine: a role
in osmotic regulation of mammalian brain and possible clinical sig-
nificance. Life Sci. 26: 1561-1568.

I mbriaco, D., M. Anctil, and L. Descarres. 1990. Serotonin-immu-
noreactive neurons in the cnidarian Renilla koelliken J Comp. N en-
rol. 291: 167-178.

\\urdcn, S., and l : . llomhcrg. 1993. A simple method for immuno-
lluorescent double staining with primary antisera from the same spe-
cies. J. Histochem Cytochcm 41:627-630.

Reference: Biol Bull 188: 83-88. (February/March, 1995)

The Influence of Opponent-Related and Outcome-
Related Memory on Repeated Aggressive Encounters
in the Paradise Fish (Macropodus opercularis)

ADAM MIKLOSI*, JOZSEF HALLER**, AND VILMOS CSANYI*

*Depanment of Ethology, Eotvos Lorand University and **Imtitute of Experimental Medicine,

Hungarian Academy of Sciences

Abstract. The aggressive behavior of male paradise fish
(Macropodus opercularis) was studied. Fish were subjected
to three aggressive encounters on consecutive days. If
submissive males encountered the same opponent three
times, the last aggressive encounter was very different than
the first one. When the animals faced a new opponent
each day, the changes were much less pronounced. We
conclude that ( 1 ) fish are able to recognize their opponents
at least one day after the encounter ("social recognition"),
and (2) social recognition modifies the effect of prior defeat
("status- related memory") in subsequent encounters.

Introduction

An overwhelming amount of evidence indicates that
prior agonistic experience influences the outcome of future
aggressive encounters (Beacham and Newman. 1987;
Frank and Ribowsi, 1987). One can hypothesize that prior
aggressive experience may influence subsequent aggressive
encounters by two kinds of processes: one related to the
outcome of the encounter ("winner" or "loser" effect)
and the other specifically related to the opponent. The
significance of the former process was recently examined
in detail (Bevan et a/., 1960; Poll el ai, 1982; Francis,
1983; Beaugrand and Zayan, 1985; Beacham and New-
man, 1987; Bakker et ai. 1989). Most studies demonstrate
an asymmetrical effect of prior winning or losing on sub-
sequent winning probability. For example, in paradise fish
(Macropodus opercularis; Francis, 1983) and in stickle-
backs (Gasterosteus acideatits; Bakker and Sevenster,
1983) losing greatly enhances the probability of also losing

Received 15 April 1993; accepted 18 November 1994.
Address for correspondence: Adam Miklosi, Eotvos Lorand University,
Department of Ethology, God, Javorka S. u 14, H-2131. Hungary.

the subsequent contest. Winning usually has no strong
effect, but under some experimental conditions it might
increase the probability of winning again (Bakker and
Sevenster, 1983; Bakker et ai, 1989).

The possibility of the involvement of the second pro-
cess individual recognition in agonistic encounters has
been also demonstrated. Fricke (1973) showed that Am-
phiprion bicinctus males more frequently attacked un-
known individuals than known ones in a two-choice ex-
periment. The importance of individual recognition in
the stickleback was demonstrated by Peeke and Veno
( 1973), who observed that a resident male that had been
habituated to an intruder presented in a glass cylinder
would resume aggressive behavior if the individual in the
cylinder was changed. Thresher (1979) used a similar
method to study rival recognition in the threespot dam-
selfish (Eupomaceus planifrons), and those field obser-
vations further confirm that some fish species might be
able to recognize individuals. Myrberg and Riggio (1985)
showed that coral reef fish (Pomacentrus partitus) rec-
ognize territorial neighbors acoustically. Recently Waas
and Colgan (1994) provided experimental evidence that
male sticklebacks can distinguish between familiar rivals
on the basis of visual cues alone.

Assuming that the effects of previous encounters are
mediated by memory and that the behavioral differences
are not due to energetic consequences of aggression (Haller
and Wittenberger, 1988; Haller, 1991) the problem of
interference between social recognition and status-related
memory arises.

As a continuation of a recent study (Miklosi et ai, 1992)
on aggressive behavior in the paradise fish, we experi-
mentally examined these processes in the aggressive be-
havior of the paradise fish. To clarify the relationship be-

83

84

A. MIKLOSI ET AL

tween social recognition and status-related memory, two
questions have been posed: ( 1 ) Does the behavior of the
fish change if it encounters the same or different opponents
for two subsequent encounters? (2) Are there differences
between these experimental manipulations?

Materials and Methods

Experiments were conducted with 6-month-old, 8-cm
to 10-cm-long Macropodus operadaris males, which were
raised and kept in our laboratory. Three days before the
start of the experiment, pairs of size-matched fish were
placed in 40 X 20 X 20 cm glass tanks provided with
filtration and aeration. Each tank was separated into two
equal parts by a green opaque screen, and a single fish
was kept in each part of the tank. The animals were vi-
sually isolated from each other: all sides but the front side
of the tanks were covered with green plastic sheets. This
isolation lasted 3 days prior to the experiment. Water
temperature was kept at 28C, and a 14:10 h light:dark
cycle was maintained. The fish were fed daily on Tubifex
worms.

Animals were exposed to three aggressive encounters
on successive days. Encounters were begun by removing
the plastic partition. All encounters were videotaped until
dominance relationships were established. We defined this
as the point at which one of the fighting males no longer
participated in simultaneous or reciprocal threatening and
fighting but instead became the "subordinate," showed
fleeing and escaping behavior when approached by the
"dominant," which in turn chased and bit its opponent.
A submissive fish remains motionless for a long time in
"oblique position" near to the water surface and does not
retaliate against the winner (Miklosi el a/., unpub. obs.;
Forselius, 1957). Both fish were observed for an additional
hour to monitor the stability of the dominant-subordinate
relationship. One hour after the end of fighting, the contact
between the fish was interrupted by lowering the plastic
door, and the animals were kept in isolation for the next
24 h.

Two groups were tested: the fish in group A (n = 10)
faced the same opponent throughout the experiment; in
group B (n = 10) the dominant animals were randomly
changed between the tanks after each fight. Thus, in group
B, the submissive fish remained in the tank and faced a
new. previously dominant opponent each day. The dom-
inant fish was changed immediately after the end of the
encounter and remained isolated in its new tank until the
following day when the partition was removed again. The
time between two consecutive encounters was long enough
for the dominant to acclimate to the new place (Csanyi
ct ai, 1985), thus the advantage of prior residency of the
submissive fish was minimized.

Videotapes were later analyzed by recording behavioral
units with an event recorder (Nagy et til., 1985). On the

basis of earlier findings, each aggressive encounter was
divided into three main phases: latency, threatening, and
fighting. After the latency for initiation of the first display,
a second phase was defined, which lasted until the first
appearance of contact behavior (biting or mouthlocking).
This was called threatening, which in turn was followed
by the escalation of the fighting fighting phase until
one of the males showed submissive behavior. The be-
havior units we identified are as follows:

Display at distance (DIS): The fish stay in head-tail
position with erected tailfin, but the distance between
them is larger then one body length.

Head-head display (HHD): The fish are oriented par-
allel to one other and face in the same direction, with one
fish slightly behind the other.

Parallel swimming (PAS): The fish swim very close to
each other in the same direction.

Head-tail display (HTD): The fish in parallel orienta-
tion are facing opposite directions. Sometimes this be-
havior is associated with circling.

Shaking (SHA): This behavior is similar to the head-
tail display, but it is associated with fast circling, vigorous
body-shaking, and a downward movement of the pair;
the pattern stops when the animals reach the bottom.

Bite (BIT): One fish makes a swift dart and slashes at
the other fish.

Mouthlock (MOU): The fish reciprocally bite and hold
one another's mouths for up to 2 min.

Air gulping (AG): A fish takes an air bubble in its mouth
by breaking the surface of the water.

Each of these behavioral units was recorded in all of
the pairs investigated. Two samples of behavior were reg-
istered. The first sample, which characterized behavior
during the threatening phase, lasted for 10 min from the
raising of the door or until the first instance of contact
behavior (biting or mouthlocking). The second sample,
which characterized the fighting phase, was a 20-min ob-
servation following the first observed bite.

To permit comparison between pairs, the values of ob-
served behavior units were divided by the sampling time.
This adjustment was necessary because in many contests
fish finished the threatening or fighting phase before our
predetermined interval (10 or 20 min) of observations
ended, resulting in shorter time samples. Thus the relative
duration (minutes per hour) or frequency (number per
minute) of behavior units was used for statistical analysis.

Because the measured variables were not normally dis-
tributed (according to the Kolgomorov-Smirnov test),
nonparametric statistical methods were used. Kruskal-
Wallis's one-way ANOVA was used separately for groups
A and B to examine the change in the general pattern of
aggressive behavior.

OPPONENT-RELATED AND OUTCOME-RELATED MEMORY

85

Results

The difference between the two groups that is, the
different effects of the "treatments" can be seen in Ta-
ble I.

Repeated encounters with the same opponent, group
A, caused marked change in aggressive behavior. Although
the duration of the threatening phase did not change sig-
nificantly in the course of the three encounters, shaking
and air-gulping were significantly reduced. All measured
variables (with the exception of head-head display) of
fighting, including its duration, decreased significantly
when submissive fish faced the same opponent three times.

Interestingly, the changes in the other group (B) were
much less pronounced. When the submissive fish repeat-
edly faced new opponents, only minor changes could be
observed in their aggressive behavior. The threatening
phase did not change significantly; only the relative du-
ration of shaking and the frequency of biting showed a
marked decrease.

A comparison with current literature showed that some
behavioral elements and parameters are of special im-
portance. Thus head-tail display (e.g.. Baerends and Baer-
ends-Van Roon, 1950; Barlow, 1962; Enquist and Ja-
kobsson, 1986), biting (e.g.. Peeke and Veno, 1973; Frank
et a/.. 1985; Enquist and Jakobsson, 1986; Halperin and
Dunham, 1994), mouthlocking (e.g.. Baerends and Baer-
ends-Van Roon, 1950; Enquist and Jakobsson, 1986),
duration of threatening (e.g., Frank et al.. 1985), and du-
ration of fighting (e.g.. Enquist et al.. 1990; Haller, 1992)

were examined further when we used the nonparametric
Mann-Whitney test to compare the behavior of the two
groups in the first, second, and third encounters (Fig. 1).

The two groups did not differ in the first and second
encounter; however, with the exception of head-tail dis-
play, they differed markedly in the third encounter. The
time spent with mouthlocking (z = -2.4, P < 0.02), the
number of bites (z = -2.3, P < 0.02), and the duration
of threatening (z = -2.6, P < 0.01) and fighting (z =
-l.9,P< 0.05) were lower in the group (A) with the same
opponent than in the group (B) with different opponents.

The same variables were compared by the nonpara-
metric Wilcoxon test to show within-group differences
during the three encounters. In group A same opponent
in each encounter we found a significant change from
the first to the second encounter only in fighting duration
(z = -2.8, P < 0.01). However, significant changes oc-
curred between the second and the third encounters in
all of the selected variables (head-tail display: z = -2.8,
P < 0.01; mouthlocking: z = -2.6, P < 0.01; biting: z =
-2.8, P < 0.0 1 ; threatening: z = -2.5, P < 0.02; fighting:
z = -2. 1, P < 0.04). In contrast, no significant differences
could be found in the group (B) with unknown opponents.

Discussion

The results clearly show that the type of opponent (fa-
miliar versus nonfamiliar) has a major effect on the ag-
gressive behavior of male paradise fish. In the case of fa-
miliar opponents (group A), three consecutive encounters

Table I

Analysis of elements of aggressive behavior shown by fighting paradise fish pairs during three consecutive contests in both groups

Group A: familiar opponent

Group B: unfamiliar opponent

encounter 1
Mean (SE)

encounter 2
Mean (SE)

encounter 3
Mean (SE)

Chi

Signif.

encounter 1
Mean (SE)

encounter 2
Mean (SE)

encounter 3
Mean (SE)

Chi

Signif

Dur. of threatening

12.6(2.1)

13.4(2.9)

5.5(1.8)

4.7

ns

12.3(1.9)

12.1 (2.2)

11 (2.2)

1.7

ns

Head-head display

2.3 (0.9)

2.5 (0.7)

3.7(1.4)

0.02

ns

4.4(0.8)

6.2(1.6)

3.6(0.8)

0.7

ns

Shaking

2.3(0.6)

1.7(0.4)

0.4(0.2)

7.3

P < 0.03

2.1 (0.7)

1.4(0.3)

1.1 (0.3)

0.8

ns

Air-gulping

1.7(0.4)

1.9(0.7)

0.3 (0.1)

10.5

P<0.01

3.2(1.4)

1.9(0.8)

2.5(0.6)

2.5

ns

Parallel swimming

1.4(0.7)

3.5 (1.3)

0.7 (0.3)

2.5

ns

2.1 (0.6)

4.4 (1.6)

1.6(0.6)

2.1

ns

Head-tail display

32.2(4.5)

36.1 (6.6)

19 (4.7)

4.7

ns

28.7(4.4)

25.5(5.1)

26.5 (4.3)

0.4

ns

Display at distance

4.1 (1.3)

3.1 (0.7)

5.1 (1.9)

0.2

ns

5.5 (1.8)

6.9(3.5)

6.3(2.1)

0.2

ns

Dur. of fighting

142.6(40.1)

59.8(22.8)

7.3(3.5)

18.1

/><0.01

115 (42.4)

50.1 (16.5)

54.8(32.7)

4.9

ns

Head-head display

1.9(0.8)

1.8(0.8)

2.1 (1.4)

4.7

ns

4.4 (0.8)

6.2(1.7)

3.6(0.8)

0.1

ns

Shaking

1.3(0.5)

1.2(0.5)

0.5 (0.4)

5.9

P<Q.Q\

1.3(0.2)

0.5 (0.2)

0.5 (0.2)

9.2

P < 0.01

Air-gulping

3.4 (0.9)

1.7(0.5)

0.4(0.3)

15.1

F<0.01

3.1 (0.6)

2.6(0.9)

1.6 (0.6)

4.0

ns

Parallel swimming

22.4(6.2)

16.2 (5.7)

12.3(2.8)

8.8

P < 0.01

20.4 (7.9)

3.1 (1.4)

2.5 (0.4)

0.9

ns

Head-tail display

39.6 (2.6)

31.4(3.2)

10 (5.2)

8.8

P < 0.02

32.8 (2.2)

22 (5.5)

17 (5.4)

0.2

ns

Biting

1 (0.2)

1.1 (0.2)

0.1 (O.I)

15.3

P < 0.01

1.6(0.3)

0.9 (0.3)

0.9(0.3)

16.9

P < 0.01

Mouthlocking

13.8(2.5)

9.2(3.4)

0.8 (0.8)

17.3

P<0.01

9.1 (1.8)

6.1 (2.4)

5.9(2.2)

3.3

ns

Note: As a summary of the results, the mean (standard error) and the result of the one-way Kruskal-Wallis ANOVAs are given. The level of
significance was greater than 0.05.

86

A. MIK.LOSI ET AL

20

18

16

14

12

10

8

6

4

2

Duration of threatening (a)

1.

encounter

2.
encounter

3.
encounter

200

180

160

140

120

100

80

60

40

20

Duration of fighting (b)

1.
encounter

2.
encounter

3.

encounter

Head-tail display (c)

Mouthlocking (d)

60

, N

to .-A
50

| 40
re

I 30

$ 20
+:

m
15 1 U

1.

encounter

3.

encounter

o

a

n>

20

18

16

14

12

10

8

6

4

2

1.

encounter

3.
encounter

1.6
'P 1.2

10

I 0.8

| 0.6

0.4

0.2

Biting (e)

i.

encounter

2.

encounter

3.
encounter

Group A: familiar opponents
Group B: unfamiliar opponents

Figure 1. The mean duration of threatening (a) and fighting (h), and the mean relative duration of head-
tail display (c) and mouthlocking (d) and biting (e) in the three consecutive encounters of the experimental
groups. In group A the opponents were the same for each contest: in group B the former dominant was
replaced hy a new dominant male for each fight.

were needed to induce significant changes in aggression.
On the other hand, repeated encounters with unfamiliar
individuals (group B) caused significant changes in only
some parameters of fighting. However, the changes that

occurred in group B are much less dramatic than those
in group A.

The behavior of the contestants was markedly similar
during the first two encounters in both groups. This means

OPPONENT-RELATED AND OUTCOME-RELATED MEMORY

87

that (1) there was no significant change within a group
from the first to the second encounter, and (2) behavior
did not seem to depend on the familiarity of the opponent
during the second encounter. The significant decrease in
the duration of fighting can be explained by noting that
the first encounter occurred after 3 days of isolation, dur-
ing which fish could build up energy reserves depleted
during the rather long fight (about 2 h on average) at the
first encounter (Haller and Wittenberger, 1988; Haller,
1 99 1 ). Without these energy reserves the second encounter
became shorter. However, it is also possible that isolation
increased aggression levels.

The defeated fish in group A (same opponents) fought
very similarly during the first and second encounters. Al-
though all defeated fish lost the fight again, they gave up
fighting only after a considerable time and engaged in
both signaling behavior (e.g.. head-tail display) and
strength-testing behavior (e.g.. mouthlocking). The same
happened in the group with unfamiliar opponents (B):
although defeated fish lost against the formerly dominant
opponents, the previous defeat did not seem to change
their behavior significantly.

Thus comparing the first two encounters in both groups
we find the effect of previous experience on behavior
"status- related memory" but no direct evidence of social
recognition. Since the initial work of Ginsburg and Allee
(1942), many studies have documented the effects of prior
experience (e.g., Bakker and Sevenster, 1983, Beacham
and Newman, 1987). In the case of the paradise fish, defeat
decreases the probability of subsequent winning in an ag-
gressive encounter, but prior winning has no influence
(Francis, 1983). However, three other factors might de-
crease the difference between a first and second encounter.
(1) Following longer isolation before the first contest
(3 days) and between contests (about 22 h), fish fight
longer in both the first and second encounters (Miklosi
et a/., unpub. obs.). (2) The encounter was terminated
1 h after fighting had finished, and fish were fed only fol-
lowing separation, thus opportunities for expressing
dominance or submission were limited. (3) The weight
symmetry between contestants rendered mutual assess-
ment more difficult, according to the resource holding
power (RHP) hypothesis (Parker, 1974). Usually larger
animals initiate aggressive encounters and are more likely
to win in a shorter fight. Thus similarity in size will in-
crease both the latency of initiation and the duration of
a contest.

The third encounter separates the two groups clearly.
For fish facing familiar opponents (group A ), the duration
of the threatening phase decreased by half, and previously
submissive fish gave up fighting soon after they began. In
contrast, no significant change was observed in the be-
havior of fish facing unfamiliar opponents (group B).
There was about a sixfold difference in biting, mouth-

locking, and duration of fighting between the two groups,
which rules out the role of exhaustion. In both groups,
submissive fish lost two fights before engaging in the third
contest: thus experience in submission or dominance
cannot explain the observed difference. As a result, the
involvement of some form of social recognition should
also be taken into account for the third encounter.

Nevertheless, this experiment does not directly prove
that individual recognition takes place. As Waas and Col-
gan (1994) recently noted, "Individual recognition implies
that subjects can distinguish between animals that belong
to the same social and physical class." But it is very difficult
to tell the exact basis of this form of recognition because
individuals can be categorized into several subcategories,
and the same animal can use different arrays of variables
to categorize its opponents. Because opponents were al-
ways of the same social class in both groups (submissive
or dominant), the recognition might have occurred on a
different level, which suggests that paradise fish are capable
of categorization within dominants or submissives.
Whether this can be described as individual recognition
remains to be seen, and Waas and Colgan (1994) show a
good way to examine this subject.

On the other hand, we already have some evidence that
individual recognition exists in fish (Gandolfi et a/.. 1973;
Zayan, 1975; Myrberg and Riggio. 1985). For example,
as shown by Zayan (1975), individual recognition of
formerly dominant fish can reverse the effect of prior res-
idence. Thus, the process of individual recognition inter-
acts with the effects of both prior experience and prior
residence in a way similar to that in our present experi-
ment.

It is usually assumed that the end of the fight depends
on the decision of the future submissive fish. This idea
stems from the classical conditioning view of aggression,
in which contact behaviors (biting, mouthlocking) are seen
as punishment for the opponent, which learns during the
aggressive encounter to avoid these aversive effects
(McDonald eta/.. 1968; Bakker el <//.. 1989). In this con-
text an individual opponent becomes a conditioned signal
for future punishment, but a new opponent acts as a dis-
criminative signal, which does not predict aversive stim-
ulation.

Another theory interprets aggressive encounters in
terms of an associative habituation process (Peeke, 1969;
Lorenz, 1981). New opponents (or territorial neighbors)
disrupt habituation and release an aggressive response.
For example, Peeke and Veno ( 1973) found that resident
male sticklebacks attacked unknown territorial neighbor-
ing sticklebacks more often than they attacked familiar
ones.

The problem of small changes in a complex stimulus,
like an opponent, raises a major problem to the incor-
poration of individual recognition into the learning mod-

A. MIKLOSI ET AL.

els presented above. Either dishabituation or discrimi-
nation suppose some form of recognition of the opponent,
but we do not know whether this categorization process
is similar in the two models or not. As is the case with
other processes described mainly on a behavioral level
(e.g.. imprinting), it is very difficult to explain them in
the framework of classical learning models.

At least in paradise fish, it seems that submissive fish
try to use every occasion that offers the possibility of win-
ning. Winning a contest presumably has many advantages.

Paradise fish males defend territories and build bubble-
nests in shallow waters of rice-fields, where several males
breed at the same time near each other (Forselius. 1957).
Recognizing neighbors could be advantageous because
males spend less time in aggression, leaving time for
courtship and later caring for the fry.

Our results support the hypothesis that aggressive ex-
perience in the paradise fish influences subsequent ag-
gressive encounters by means of two kinds of memory:
one related to the outcome of the encounter ("status-re-
lated memory") and the other related to the opponent
("social recognition").

Acknowledgments

This study was supported by the Hungarian Academy
of Sciences by an OTKA grant (No. 368-08 13\ 1991).
Three anonymous reviewers gave helpful comments on
the earlier version of this paper.

Literature Cited

Bacrends, G. P., and .1. M. Baerends-Van Roon. 1950. An introduction
to the study of the ethology of cichlid fishes. Behaviour 1 (suppl,):
1-243.

Bakker, Th. C. M., and P. Sevenster. 1983. Determinants of dominance
in male sticklebacks (Gaslerosteus aeulealus L.) Behaviour 86: 55-
71.

Bakker, Th. C. M., E. Feuth-De Bruijn, and P. Sevenster.
1989. Asymmetrical effects of prior winning and losing on domi-
nance in sticklebacks (Gasteroxtetix aeuleutitx). Ethology 82: 224-
229.

Barlow, G. W. 1962. Ethology of the Asian teleost Badix hmlix III.
Aggressive behaviour. Z Tierpxychol 19: 29-55.

Beacham, J. L., and J. A. Newman. 1987. Social experience and the
formation of dominance relationships in the pumkinseed suntish.
Lepomix xil'htixux Anim. Behav 35: 1560-1563.

Beaugrand, .). P., and R. Zayan. 1985. An experimental model of ag-
gressive dominance in \iphophorux hellen ( Pisces. Poeciliidae). Behav
Proc. 10: [-52

Bevan, VV., W. F. Daves, and G. W. Levy. 1960. The relation ol cas-
tration, androgen therapy and pre-test fighting experience to com-
petitive aggression in male C57 Bl/IO mice. Ainm. Behav 8: 6-12.

Csanyi, V., P. Toth, V. Altbacker, A. Doka, and J. Gervai.
1985. Behavioural elements of the paradise fish (Macropodnx op-
ercularis) I. Regularities of defensive behaviour. Ada Bin/ Hung
36: 115-130.

Enquist, M., and S. Jakobsson. 1986. Decision making and assessment
in the fighting behaviour ofNannacara anmnala (Cichlidae. Pisces).
Ethology 12: 143-153.

Enquist, M., O. I I'imar, T. I.jungberg, Y. Mallner, and N. Segerdahl.

1990. A test of the sequential assessment game: fighting in the cichlid

fish Nannacara anomala .tiuni Behav 40: 1-14.
Forselius, S. 1957. Studies of anabantoid fishes. I. A quantitative de-
scription of the reproductive behaviour. Zoologixka Bidrag fran

i'ppxa/a 32: 93-302.
Frank, D. R.-P. Hannes, II. LanfTermann, and A. Ribowski.

1985. Effects of social isolation on aggressiveness in fish with special

reference to the swordtail (Xiphophorus helleri). Behav Proc. 10: 4 1 5-

427.

Frank, D., and A. Ribonski. 1987. Influences of prior agonistic expe-
riences on aggression measures in the male swordtail (Xiphophorus

hellen! Behaviour 9\: 218-239.
Francis, R. C. 1983. Experiential effects on agonistic behavior in the

paradise fish. Maeropodits upm'iilaris. Behaviour 85: 292-313.
Fricke, H. \V. 1973. Individual partner recognition in fish: field studies

on Amphiprion bicinelus Naturwissenschaften 60: 204-205.
Gandolfi, G., R.Gandolfi, and K. I.e Moli. 1973. Short-term individual

recognition in Padogobius murteiixi. Lincei Rend Sc. Fis. Mat. e

A'< 54: 281-285.
Ginsburg, B., and \V. C. Alive. 1942. Some effects of conditioning on

social dominance and subordination in inbred strains of mice. Physiol.

/ool 15:485-506.

Mailer, J., and C. \\ittenberger. 1988. Biochemical energetics of hi-
erarchy formation in Bella xplendens. Physiol. Behav 43: 447-450.
Mailer, J. 1991. Muscle metabolic changes during the first six hours

of cohabitation in pairs of male Bella splendent. Physiol. Behav 49:

1301-1305.
Mailer, J. 1992. Group size modifies the patterns and carbohydrate

effects of aggression in Bella sp/endens. Physiol. Behav. 52: 278-290.
Malperin, J. R. P., and D. W. Dunham. 1994. Social overstimulation

reduces subsequent aggression in Bella splendens. Aggressive Behav.

20: 135-142.

Lorenz, K. 1981. Foundations ol Ethology Springer- Verlag, Wien.
McDonald, A. L., N. \V. Meimslra, and D. K. Damkot. 1968. Social

modification of agonistic behaviour in fish. Anim Behav 16: 437-

441.
Miklosi, A., J. Mailer, and V. Csanyi. 1992. Different duration of

memory for conspecific and heterospecific fish in the paradise fish

(Macropodus opeirii/anx). Ethology 90: 29-36.

Myrberg, A. A., and R. J. Riggio. 1985. Acoustically mediated indi-
vidual recognition by a coral reef fish (Pomacentrus partitas). Anim.

Behav 33: 41 1-416.
Nagy, A., A. Doka, and V. Csanyi. 1985. A microcomputer method

for recording and analysing behavioural elements. Ada Biol. Hung

36: 239-244.
Parker, G. A. 1974. Assessment strategy and the evolution of fighting

behaviour. ./ Theor. Iliol 47: 223-43.
Peeke, H. V. S. 1969. Habituation of aggression in the three-spined

stickleback (Gasterosteus aculeatus L.). Behaviour 35: 137-156.
Peeke, H. V. S., and A. Venn. 1973. Stimulus specificity of habituated

aggression in the stickleback (Gaxlerosleus aeuleatus) Behav. Biol.

8: 427-432.
Poll, van de, L., F. Jonge. H. G. van Oyen, and J. van Pelt.

1982. Aggressive behaviour in rats. Effects of winning or losing on

subsequent aggressive interactions. Behav Proe. 7: 143-155.
Thresher, R. E. 1979. The role of individual recognition in the territorial

behaviour of the threespot damselfish (Eupomacentrus plaiufrons).

Mar Behav Phyxiol 6: 83-93.
Waas, J. R., and P. \V. Colgan. 1994. Male sticklebacks can distinguish

between familiar rivals on the basis of visual cues alone. Anim Behav.

47: 7-13.
/ayan. R. C. 1975. Defense du territorielle et reconnaissance indivi-

duellechez.V(/)/!/7/iomv( Pisces, Pecillidae). Behaviour 52: 266-313.

Reference: Bio/ Bull 188: 89-97. (February/March. 1995)

Rapid Arm Movements in Stalked Crinoids

CRAIG M. YOUNG 1 AND ROLAND H. EMSON :

'Department of Larval Ecology. Harbor Branch Oceanographic Institution, 5600 U.S. //ivy / N..

Ft. Pierce, Florida 34946, and 2 Division of Life Sciences, King's College. London,

Campden Hill Road, London. W8 7 AH, United Kingdom

Abstract. Stalked crinoids in the family Isocrinidae have
been observed to wave individual arms actively. Using
video cameras mounted on a manned submersible, we
studied these movements and investigated the factors that
elicit them. Crinoids wave their arms in response to sand
or detritus dropped on their crowns, to entanglement in
tentacles of adjacent sea anemones, and to contact by
small crustaceans that might steal from the food grooves.
There was no evidence that arm waving functions in food
collection. In most cases, the movements could be attrib-
uted directly to mechanical stimulation by some natural
stimulus. The rapid effective stroke of an arm flexure is
caused by contraction of dorsal longitudinal arm muscles.
The slower return stroke results from the elastic recoil of
large ligaments near the aboral sides of the arms.

Introduction

Stalked crinoids are passive suspension feeders with
limited mobility but are nevertheless capable of several
kinds of movements. The most characteristic behaviors
are slow movements used to orient with respect to currents
and to hold the arms and pinnules in a parabolic feeding-
fan posture (Macurda and Meyer, 1974, 1976; Conan et
al.. 1981). The mechanisms by which these postures are
maintained and controlled are poorly understood. Ori-
entation of the stalk, which contains no muscles, is de-
pendent on mutable collagenous tissues (Wilkie et al.,
1993). The tonic posture of the parabolic feeding fan is
probably maintained by a similar mechanism, but there
is as yet no morphological or physiological evidence for
mutable arm ligaments (I. Wilkie, pers. comm.).

Stalked crinoids occasionally demonstrate fast muscular
movements. Several species are thought to be capable of
moving between attachment sites (Carpenter, 1884:

Received 16 March 1993; accepted 2 December 1994.

Conan et al., 1981: Roux. 1976), and stalked crinoids
have recently been observed crawling across the bottom
(Messing, 1985; Messing et al.. 1988). When stimulated
by the manipulator arm of a submersible or by very bright
lights, this same species, Endoxocrinits parrae, rapidly
flexes some or all of its arms in an adoral direction (Mess-
ing et al.. 1988; Young and Emson, unpub.). Except for
an unpublished anecdotal observation suggesting that cri-
noids may respond to suspended sediment (W. I. Ausich.
pers. comm.), all reports of rapid active arm movements
have involved strong artificial stimuli. The natural roles
of rapid arm movements remain undocumented. Here,
we describe in detail rapid arm flexures of some bathyal
isocrinids and present evidence that this behavior defends
crinoids against various biotic and abiotic threats.

Materials and Methods

Several species of stalked crinoid were observed from
Johnson-Sea-Link (JSL) submersibles at depths ranging
from 400 to 900 m in the northern Bahamas (see map in
Young, 1992). Still photographs were taken with a Benthos
35-mm camera equipped with an 80-mm lens, mounted
on the front of the submersible and focused with twin
laser beams that converged on a fixed focal plane. Video
footage was obtained with a Photosea Camera on a pan-
and-tilt mechanism and was recorded on W or hi-8 vid-
eotape. Video still sequences were taken from the tape
with a Seikosha VP-1500 video printer.

We obtained numerical data on arm-waving frequency
and crustacean abundance directly from the videotape.
We stopped the tape every 30 s and counted the number
of arm movements, the number of crinoids involved in
arm-waving behavior, and the total number of crinoids
visible in the frame. We ran the tape forwards and back-
wards a few frames at each census point to be certain that
arms counted as waving were really in motion and not

89

90

C. M. YOUNG AND R. H. EMSON

Figure 1. (A I Ijii/ii\ncniuit pumic with arms drooping in slack current. Note the single arm waving in
the water column (arrow). (B) Cenocriniis aslerius in current, showing parabolic feeding tan characteristic
of all Bahamian isocrinids. (C) E ptimic engaged in arm-waving behavior (arrow indicates moving arm).
(D) A dense population of E purrac with numerous individuals waving arms (indicated by arrow).

being held in a static posture. The number of small crus-
taceans in a frame was estimated by repeatedly passing
the video forward and back, frame by frame, while scan-
ning each part of the frame in succession for moving or-
ganisms.

The velocity of arm movement during effective and
recovery strokes was documented by laying down a time
code on the videotape with a hi-8 video editing machine
(Sony EVO-9700), then, during frame-by-frame analysis,
recording the time that movements were initiated and
completed (resolution: 0.067 s).

To investigate the possibility that sediment particles
might elicit arm waving, we used a suction tube on the
manipulator arm of the submersible to pick up a small
amount of sediment and release it about 1 m above an
aggregation of crinoids. This experiment was repeated on

six different occasions, while recording the responses of
crinoids on videotape. On some occasions, the sediment
consisted of fine silt; at other times, it was dominated
either by coarse sand or coarse, flocculent organic parti-
cles.

Crinoid arm pieces were fixed in 4% neutral buffered
formalin, decalcified in 70% acid alcohol, then embedded
in paraffin by standard histological procedures. Sections
were cut at a thickness of 8 ^m and stained with Milligan's
trichrome (Humason, 1972).

Results

Description and mechanics oj arm waving

At times of slack current, three Bahamian isocrinids,
f'~nc/o.\ocnnu,\ parrac, Cenocriniis asterius, and Diplocri-

ARM WAVING IN SEA LILIES

91

Figure 2. Video sequence of characteristic arm waving behavior in Endo.\ocnnii.\ parrac (A-C) Sequential
steps of the effective stroke. (D) Maximum arm extension. (E-F) Recovery stroke.

mis maclearanus, stand erect with arms drooping down
near the stalk (Fig. 1 A). In a current, these same species
form their arms into a parabolic fan for feeding (Fig. IB;
see also Macurda and Meyer, 1974, 1976), though the
uppermost few arms of the fan may sometimes be ex-
tended straight up into the water column. All three species
have been observed with individual arms waving up and
down rapidly (Fig. 1A, 1C). In dense populations, large
numbers of individuals have been observed to engage in
arm-waving behavior simultaneously (Fig. ID), particu-
larly after several minutes of illumination by the sub-
mersible.

Although we have occasionally observed arm flicking
or waving in animals with their arms extended in the
feeding posture, arm-waving behavior has been observed
more commonly in animals with drooping arms. The arm
is moved rapidly away from the stalk, sweeping outward
and upward until it is fully extended above or to the side
of the calyx (Fig. 2). The arm pauses only briefly at the
end of the stroke before reflexing downward more slowly
to its initial position. This entire movement may take as
little as 2 s or as much as 21 s. Frequency histograms of

the durations of effective and recovery strokes (Fig. 3)
show that the recovery strokes were more variable and
often longer than the effective strokes, but the two distri-
butions overlapped substantially. For individual strokes,
the ratio of the effective component to the recovery was
nearly always greater than 1 (Fig. 4). and the difference
between the durations of paired effective and recovery
strokes was highly significant (paired Student's / test,
54d.f., i = 5.75, P < 0.0000). The arms were flexed
through arcs ranging from a few degrees to more than
1 80 degrees. Most arms were flexed only once before an-
other arm was brought into play. Often, one arm was
flexed while another on the same animal was in its re-
covery stroke.

Examination of histological sections of the arm of E.
parrae revealed the presence of large dorsal (oral) longi-
tudinal muscles linking the arm segments (Fig. 5). These
muscles, which are described elsewhere (Hyman, 1955)
as flexor muscles, are clearly responsible for the flexure
of the arms. There are no opposing longitudinal muscles,
but large ligaments are found ventral (aboral) to the flexor
muscles (Fig. 5). The recovery phase of arm waving must

92

C. M. YOUNG

25
20 -

p

0)

D Effect

ve Stroke

2 15-

Recoi

ery Stroke

"o

1 10-

:

5 -

,

r

III In

1 P

5 10

15

Duration of Stroke (s)

Figure 3. Frequency histogram comparing the durations of the ef-
fective strokes (hatchedl and recovery strokes (solid black) during arm
flexure of EndoMicrinus parrae.

therefore be achieved by elastic recoil. Small dark-staining
cell bodies at the insertions of the ligaments (Fig. 5B)
appear to be juxtaligamental cells (Wilkie, 1984), which
are known to regulate collagen viscosity in other echi-
noderms. The largest axons in arm cross sections mea-
sured 3.75 ^m in diameter, and most were between 2.5
and 3.5 //m in diameter.

Functions of arm waving

With the use of a close-up video camera, we determined
that flexures often occurred in response to mechanical
stimuli caused by various organisms and particles. For
example, when the arms of stalked crinoids become en-
trapped in the tentacles of adjacent sea anemones, arm
flexures allow them to escape. Arms are also flexed in
response to contacts by small crustaceans. Such crusta-
ceans are always attracted to the lights of the submersible
in large numbers, affording us increased opportunities for
observing encounters between crinoids and crustaceans.
On 17 February 1980, we came upon a rocky ndge sup-
porting more than 200 E. parrae and C. asterius at a
depth between 409 and 500 m off Booby Rocks, New
Providence Channel, Bahamas. As we passed up the ridge
without stopping, we filmed 49 crinoids in two aggrega-
tions, observing all the while only two instances of arm-
waving behavior. We then rested the submersible near a
third large aggregation and filmed it from a distance of
3 m for 7 min. The percentage of animals participating
in arm waving and the number of arm waves per indi-
vidual increased linearly with the number of crustaceans
visible (Fig. 6). Although these regressions are consistent

with the idea that crustaceans stimulate arm waving, we
could not dismiss the possibility that density of crustaceans
covaried with some other factor (e.g., illumination time)
until video cameras with higher resolution were installed
in 1991.

On 24 October 1991 at a depth of 642 m off Egg Island,
we located a large aggregation of E. parrae. By focusing
on inactive individuals, we recorded 10 instances of arm
waving that were clearly stimulated by a single crustacean.
A representative encounter is shown in Figure 7. The time
required for initiation of a visible response to the impact
of this crustacean was 0.47 s. In every case, the crustacean
contacted the crinoid on the oral side of the arm between
the pinnules and in the region of the food groove. In one
observed encounter, the crustacean remained attached
during three sequential flexures before becoming dis-
lodged; in all other instances, the crustacean was dislodged
by the initial arm movement and swam away. On sub-
sequent dives, crustacean-induced arm movements were
also recorded for C. asterius, one of which is shown in
Figure 8. Here, the crustacean was swimming upstream
in the turbulent downstream wake of a crinoid feeding
passively in the current. When the crustacean contacted
the oral side of the arm, a small flexure was elicited im-
mediately (Fig. 8), and the crustacean moved downstream.

We dropped sediment from the manipulator on six
separate occasions with two to four attempts on each ex-
periment. Sediment containing a mixture of particles
ranging in size from 1 to several millimeters elicited dis-
crete flexures of individual arms when individual particles
struck (Fig. 9). Small amounts of very fine silt did not

20

-!-

0)

n.

15 "
CO

10 -

LLJ

C

o
ro
D

5 -

10

I
15

20

Duration of Recovery Stroke (s)

Figure 4. Durations of individual effective strokes plotted against
durations of corresponding individual recovery strokes for individual
arm flexures of Endoxocrinus parrae. If flexure and recovery were of the
same duration, all points would fall on the dashed line. Most points lie
below the line, indicating that recovery strokes are generally, but not
always, longer than effective strokes.

ARM WAVING IN SEA LILIES

93

200pm

Figure 5. Cross section (A) and longitudinal section (B) of an arm
of Endoxocrimtx parrae showing longitudinal flexor muscles (M), and
extensor ligaments (L) connecting portions of ossicles (O). Individual
bundles of collagen (CB) are visible in the ligaments. Cell bodies of jux-
taligamental cells ( JC) are visible at the points where collagen fiber bundles
insert into the brachial ossicles. PM: longitudinal muscle of a pinnule
cut in cross section.

stimulate waving, but fine sediment in large quantities
sometimes elicited a dramatic arm-waving response in-
volving numerous arms. Figure 10 shows the response of
one E. parrae individual to a large piece of flocculent
organic matter that lodged firmly on an arm. The crinoid
moved the affected arm as well as adjacent arms several
times until the material was dislodged.

Various kinds of crabs and ophiuroids (e.g.. euryalids)
commonly perch on sessile organisms, including large
sponges, gorgonians, and antipatharians, on the Bahamian
slope. These same organisms live on the stalks of crinoids.
but we have never seen a single individual occupying the
crown region. We suppose that arm waving might deter
occupation of the crown by ophiuroids and crabs, but
cannot prove this with observations.

Discussion

Virtually all sessile animals have neuromuscular
mechanisms for ridding themselves of impinging organ-
isms or objects that threaten them or that interfere with
the feeding process. It is not surprising, therefore, that
stalked crinoids would have an active mechanism of pro-
tection appropriate to their form and life style. In echi-
noderms, some protective mechanisms involve the use of
giant nerve fibers and very rapid (0.25 s) reaction times
(Cobb, 1985; Cobb and Ghyoot, 1993). The nerve fibers
of E. parrae measured between 2.5 and 3.75 /urn in di-
ameter, only about 30% as large as the giant fibers in
Ophiwa ophiura (Cobb, 1985). However, these are larger
than the 1 jum diameter neurons found in most echino-
derms (Cobb, 1985). Reaction times of stalked crinoids
(about 0.5 s) were about twice as long as those reported
for ophiuroids (Moore and Cobb. 1985).

On the basis of behavioral and histological observations,
it appears that arm flexure results from the contraction
of large flexor muscles, and that recovery results from the
elastic recoil of ligaments. This interpretation is consistent

2.5

'

. 2,0 -

g. 15H
en

I 1<H

0.5 -

00

r 2 =0725

10

15

1 I '
20

25

30

30

5 10 15 20 25

Number of Crustaceans Visible

Figure 6. The relationships between crustacean density and the
number of arms waving per crinoid (top panel) and proportion of crinoids
waving arms (bottom panel) during a single 7-min taping session in a
dense bed of Endoxwrinm pumic at 409 m depth at Booby Rocks. Ba-
hamas.

94

C. M. YOUNG AND R. H. EMSON

Figure 7. Response of Endoxocrinus parrae to an individual crustacean contacting the arm. (A-C)
Crustacean (position and direction of swimming indicated by arrows) enters rapidly from the right side of
the field and moves into crown region of crinoid. (D) Arm begins to flex immediately after crustacean
contacts it, and crustacean responds by swimming rapidly away (arrow). (E-F) Arm contacted by crustacean
continues to Ilex until it is maximally extended.

with the views of most previous workers (e.g., Muller.
1843; Breimer, 1978; Grimmer and Holland, 1987). An
alternative hypothesis, invoking hydraulic pressure in the
coelomic canals of the arms as a mechanism of arm ex-

tension, has been put forward by Candia Carnevali and
Saita (1985). Grimmer and Holland (1987), however,
showed experimentally that destruction of these coelomic
canals did not affect recovery from flexure in the coma-

Figure 8. (A) Small crustacean swimming upstream (in direction of arrow) within the turbulent wake
of a feeding Ccnucrinnx aslerius. (B) Small arm flexure following contact by crustacean (arrow), which is
thrown downstream.

ARM WAVING IN SEA LILIES

95

Figure 9. Arm flexures of Cenocrimis axicrinx in response to sand particles dropped on the crown. Sand
particles (one is indicated by arrow in A) were falling throughout this sequence, which involved movements
by arms on all sides of the crown.

tulid Florometra serratissima. In fact, their experimental
results demonstrate that elastic forces in the aboral soft
tissues of the arm of a comatulid were alone sufficient to
drive a power stroke in active swimming. We suggest that
a similar mechanism is involved in the recovery stroke of
stalked crinoids such as E. parrae.

Existing evidence does not support the idea that arm
waving is involved in food collection. Unlike some other
sedentary filter-feeding echinoderms at bathyal depths
(e.g., brisingid asteroids: Emson and Young. 1994),
stalked crinoids are not known to capture large particles
(Meyer, 1982; Lawrence, 1987). Indeed, our observa-
tions indicate that large particles contacting the crown
are thrown away from the crown, not toward the mouth.
Arm waving could enhance encounters with small par-
ticles, particularly in still water. However, such a feeding
strategy would seem inefficient in the oligotrophic hab-
itats where these animals live, since arm flexure requires
considerable muscular involvement and might result in
a net energy loss. By contrast, the maintenance of pos-
ture for passive feeding may involve catch connective

tissues (Wilkie and Emson, 1988) which require little
energy expenditure. Our observation of apparent jux-
taligamental cells (Fig. 5B) is equivocal evidence for
mutable collagenous tissues in stalked crinoid arms. The
use of arm flexing for filter feeding seems unlikely, but
cannot be discounted completely.

Arm-waving behavior appears on present evidence
to be principally a mechanism to eliminate inorganic
particles from the crown and to deter small organisms
from settling on the arms of the crinoid. Specifically,
we have demonstrated that the behavior has a deterrent
effect on small crustaceans, preventing them from acting
as predators, food thiefs, or opportunistic commensals.
Arm movements may prevent colonization of the crown
by crabs, ophiuroids, and other epifauna, and they per-
mit escape from the feeding structures of adjacent sessile
organisms such as sea anemones. Inorganic and organic
particles that fall on the arms also elicit arm waving,
so the behavior may have evolved as a general mech-
anism for ridding the crown of unwanted particles. As
all stalked crinoids that have been examined histolog-

96

C. M. YOUNG AND R. H EMSON

Figure 10. Response of Cenocrinus aslerius to fine sediment and a large flocculent mass of detritus
dropped on the crown. (A) Sediment contacts the crown, and a large particle of detritus (arrow) falls toward
one arm. (B) Detrital mass lands on the arm. Note that the fine silt elicits no dramatic responses from the
arms. (C) Arm (asterisk) with large detntal particle flexes, dislodging a portion of the mass. (D) Remainder
of detrital mass (arrow) remains on the arm. (E. F) Various arms in the region of the detrital mass flex
repeatedly, apparently attempting to dislodge the attached detritus.

ically have similar flexor muscles, we suspect that arm
waving may be a behavior of great antiquity and com-
mon to many living and extinct crinoids.

Acknowledgments

We thank the skilled pilots of the JSL submersibles for
assistance with close-up video. This work was funded by
the National Science Foundation (OCE-89 16264 and
OCE-91 16560) and by a NATO Collaborative Research
Grant (CRG-900628). Harbor Branch Contribution No.
1064.

Literature Cited

Brcimer, A. 1978. Recent cnnoids. Pp. 9-58 in Treatise on Invertebrate
Paleontology. Part T. Eclunodermalu. I . R. C. Moore and C. Teichert,
eds. University of Kansas Press, Lawrence, and the Geological Society
of America, Inc., Boulder, CO.

Candia Carnevali, M. D., and A. Saita. 1985. Muscle system organi-
zation in the echinoderms: II. Microscopic anatomy and functional

significance of the muscle-hgament-skeleton system in the arm of

the comatulids (Anledon medilerranea). J Morph 185: 59-74.
Carpenter, P. II. 188-4. Report on the Crinoidea collected during the

voyage of HMS 'Challenger' during the years 1874-1876. The stalked

crinoids. Challenger Rep Zoo/. 11, 442 pp.
Cobb, J. I.. S. 1985. The neurobiology of the ectoneural/hyponeural

synaptic connection in an echinoderm. Bio/. Bull 168: 432-446.
Cobb, J. {.. S., and M. Ghyoot. 1993. The giant through conducting

neuron of the hnttlestar Opliinru ophiura a key neuron? Coitifi.

Biiielieni Phyii'l 105A: 697-703.
Conan, G., M. Rouv and M. Sibuel. 1981. A photographic survey of

a population of the stalked crinoid Diphcrimis (Annacrinu.i) wyvil-

lcllinin\iini (Echinodermata) from the bathyal slope of the Bay of

Biscay. Deep-Sea Res 28A(5): 441-453.
I ins. .11. R. II., and C. M. Young. 1994. The feeding mechanism of a

deep-water hrisingid asteroid, Novodinea antillensis. Mar Bio/ 1 18:

433-442.
Grimmer, J. C., and N. I). Holland. 1987. The role of ligaments in

arm extension in feather stars (Echinodermata: Cnnoidea). Ada Zool.

68: 79-82.
Ilumason, G. I.. 1972. Animal Tissue Techniques. W. H. Freeman,

San Francisco. 641 pp.

Hyman, I,. II. 1955. The Invertebrates. II'. Echinodermata. McGraw-
Hill, New York. 763 pp.

ARM WAVING IN SEA LILIES

97

Lawrence,.). 1987. A Functional Biology of Echinodemu Johns Hop-
kins University Press. Baltimore. 340 pp.

Macurda, D. B., and I). 1.. Meyer. 197-1. Feeding posture of modern
stalked conoids, .\\nurc 247: 394-396.

Macurda. I). B.. and I). I.. Meyer. 1976. The identification and inter-
pretation of stalked conoids from deep-water photographs. Bull Mar,
Sci 26:205-215.

Meyer, D. L. 1982. Food and feeding mechanisms: Crinozoa. Pp. 25-
42 in Echinoderm Nutrition. M. Jangoux and J. M. Lawrence, eds.
A. A. Balkema, Rotterdam.

Messing, C. 1985. Submersible observation of deep-water crinoid as-
semblages in the tropical western Atlantic Ocean. Pp. 185-193 in
Echinodcrmata, B. F. Keegan and B. D. S. O'Connor, eds. A. A.
Balkema. Rotterdam.

Messing, C. G., M. C. RoseSmyth, S. R. Mailer, and J. E. Miller.
1988. Relocation movement in a stalked conoid (Echinodermata).
Bull Mm Sd. 42: 480-487.

Moore, A., and J. L. S. Cobb. 1985. Neurophysiological studies on

photic responses in Ophnira ophiura Comp. Biochem. Physiot 80A:
11-16.

Miiller, J. 1843. Uber den Bau des Penlacrinus carnil medusae- Abd.
A'.?/ Akud IIV.v.v. Berlin 1841 (Part 1): 177-248.

Rou\, M. 1976. Aspects de la varibilite et de la croissance au sein d'une
population de la Pentacrine actuelle: Annacrinus wyvillethomsoni Jef-
freys (Conoidea). Tluilassia Jugosl. 12:307-320.

\\ ilkie, 1. C. 1984. Vaoahle tensility in echinoderm collagenous tissues:
a review. Mar Behav Phvuol II: 1-34.

\\ilkie, I.C., and R. H. Kmson. 1988. Mutable collagenous tissues and
their significance for echinoderm paleobiology and phylogeny. Pp.
311-330 in Echinoderm Phylogeny and Evolutionary Biology,
C. R. C Paul and A. B. Smith, eds. Clarendon Press, Oxford.

\\ilkie, I. C., R. H. Emson, and C. M. Young. 1993. Smart collagen
in sea lilies. Nature 366: 519-520.

Young, C. M. 1992. Episodic recruitment and cohort dominance in
echinoid populations at bathyal depths. Pp. 239-246 in Marine Eu-
trophication ami Population Dynamics. G. Colombo. I. Ferran,
V. U. Ceccherelli, and R. Rossi, eds. Olsen and Olsen. Fredensborg.

Reference: Bin/ Bull 188: 98-1 10. (February/March, 1995)

The Structure and Mode of Function of the

Water Vascular System of a Brittlestar,

Ophioderma appressum

JOHN C. FERGUSON

Galbraith Marine Science Laboratories. Eckerd College, P.O. Box 12560'
St. Petersburg. Florida 33733

Abstract. Unlike the asteroids, which have large mad-
reporite structures, the ophiuroid Ophioderma appressum
possesses only two small hidden madreporite pores. Ex-
periments with labeled amino acids, fluorescent micro-
beads, and surgical obstruction show that small amounts
of seawater do routinely enter these pores and become
distributed throughout the water vascular system; but this
uptake does not seem essential. The flagellated stone canal
draws its fluid from the axial sinus, to which the pores
connect through a tortuous ampulla. Thus, the stone canal
mainly recirculates fluid from hyponeural (perihemal)
passages. That perihemal fluid is augmented by seawater
from the pores. As perihemal fluid moves towards the
stone canal, it passes by or through the axial organ, where
nutritive materials may be removed and passed into the
hemal channels. Pressure generated by the stone canal
forces flow out to the oral tube feet, polian vesicles, and,
through valves, eventually to the arm tube feet. Inflation
of the tube feet also might occur through osmotic mech-
anisms, but their activity was not impeded by raising the
external osmotic level with dextran. Observations indicate
that negative coelomic pressures must be generated during
respiratory movements, and these could lead to sufficient
body fluid production (by nitration) that the need for sub-
stantive madreporitic inflows would be alleviated.

Introduction

One of the most distinctive characteristics of echino-
derms is their water vascular system a unique arrange-
ment of fluid-filled coelomic passages and associated parts.
The general form of these structures in the different echi-

Received 25 July 1994; accepted 4 November 1994.

noderm classes has been summarized by Hyman ( 1955)
and Nichols (1966) from the works of pioneer micros-
copists. In most types, the water vascular passages open
to the exterior through a hydropore or compound mad-
reporite. and fluid within the system is used hydraulically
to extend agile appendages, the tube feet. With their va-
riety of functional adaptations, the tube feet are probably
responsible for much of the evolutionary success of this
major animal group. The water vascular system, however,
is much more complex in both structure and function
than is generally realized, and it presents many attributes
that are puzzling and. as yet, little studied.

It is generally assumed, for example, that the madre-
porite admits seawater into the system, and that ciliary
pumping by the stone canal connected to it forces the
seawater taken in to flow out to the tube feet where it
replenishes fluid losses due to their activity. That view,
however, was challenged by Binyon ( 1964, 1966, 1976a,
b, 1980, 1984), who observed that starfish tube feet remain
active for long periods without access to the madreporite.
He noted that ionic elevation of the water vascular fluid,
particularly of potassium ion, could produce an osmotic
entry of water into the structures, making madreporitic
inflow unnecessary. Binyon opined that the madreporite
might be a relief valve, a supplemental mechanism, or
have some other unknown functions.

More recently, my own work with fluorescent molec-
ular tracers and microbeads, and other methods, has
shown that seawater does routinely enter the madreporite
of all starfishes tested; but this uptake is not specifically
directed at tube foot inflation. Rather, the uptake plays a
more important role in keeping the spacious asteroid body
filled with fluid (Ferguson, 1988, 1989, 1990a, 1992,
1 994). Further, the stone canal, through connections with

98

BRITTLESTAR WATER VASCULAR SYSTEM

99

Figure 1. Oral disk of Op/iioderma appressnm. The arrows mark a pair of central (C) and peripheral
(P) slits that open into one of the 10 genital bursae. Seawater is drawn from the side of the disk through the
peripheral genital bursal slits and is vented by the central ones.

Figure 2. Magnified view of the oral shield overlaying the axial complex, visible as a slightly darkish
area (arrow). The madreporite pores are out of sight in the crevice near the asterisk. Extended tube feet (TF)
can be seen in the mouth and on the arms.

HC

UAO

CRC

P V

V^-^v J

I. V

OHR

FIG.
-15

-14

-13
-12

-11

Figure 3. Organization of the water vascular system. The oral tube
feet have been omitted and the arm tube feet and axial complex are only
partially shown. A, ampulla: CRC. circumoral nng canal; OTC, oral
tube feet canal; PP. primary madreporite pore; PV, polian vesicle; RAS.
right axial sinus; RC, radial canal; SC, stone canal; SP, secondary mad-
reponte pore; TC, transverse canal.

-10

-9

-8
-7
-6
-5

Figure -4. Detailed diagram of the madreporite-axial complex, with
a scale showing the approximate level of serial sections (Figs. 5-15). A.
ampulla; CRC, circumoral ring canal; GHS, genital hemal strand; GS.
genital bursal slit; HC, hyponeural coelom; LAO, lower axial organ;
LAS, left axial sinus; OHR, oral hemal ring; PP. primary madreponte
pore; UAO, upper axial organ; RAS, right axial sinus; S, slit in axial
organ connecting left and right axial sinus; SC, stone canal; SP, secondary
madreporite pore.

100

J. C. FERGUSON

.*'/>* i 1 '- ' '

: i**-'<fe-' rii
4 1^3^ ;i\ '

Figure 5. First of a set of serial frontal sections from the madreporite oral shield upwards. It shows part
of the large, angular ampulla (A) that lies within the oral shield. The edge of the secondary madreporite
pore (SP) can be seen on left, facing the cleft formed by the edge of the genital bursal slit. (All scale bars
= 50 ^m).

Figure 6. Section located 2()(im higher than Figure 5, showing the secondary pore opening (SP) and
the lowest portion of the left axial sinus (LAS). The pore canal lends towards the site of the primary pore,
away from the direction of inflow through the tortuous ampulla (A).

BRITTLESTAR WATER VASCULAR SYSTEM

101

other parts, must maintain an internal system of circu-
lation and filtration of the body fluids. Some starfish, such
as the intertidal Pi\ii\ter ncliruccnx, cannot adequately
maintain their body fluid volume without access to the
madreporite, while others, such as Pycnopodia helian-
thoides, rely more heavily on osmotic uptake (Ferguson,
PN100,,,L,I01992, 1994). Physical differences between
species largely reflect their individual adaptations to fluid
volume maintenance. Pinaster has a heavy, stiff, and im-
permeable integument, while that of Pycnopodia is light,
flexible, and much more permeable.

Because the function of madreporite mechanism in dif-
ferent asteroids varies considerably, similar diversity
should be expected in the other classes; indeed, differences
in body form between the classes might relate to their
individual approaches to solving the problems of body
fluid maintenance. Although ophiuroids share many
characteristics with the asteroids, they are also quite dif-
ferent. Here I begin a detailed examination of the structure
and function of an ophiuroid water vascular system-
thai of Ophioderma appressnm. Comparison of its prop-
erties with those of other echinoderms should broaden
our understanding of the adaptations of these diverse or-
ganisms.

Materials and Methods

Specimens of Ophioderma appressnm were collected
from tidal grass flats in Tampa Bay and maintained in
running, fresh seawater until used for experiments. This
species is very similar to Ophioderma hrevispinum (re-
ferred to in earlier work), but grows larger and possesses
more equal-dimensioned oral shields (five large plates
surrounding the mouth). As in other ophiodermids. the
openings into its genital bursae (pairs of pouches invagi-
nated into the disk on either side of the arm bases) are
each separated into two slits a "central" one near the
mouth and a "peripheral" one on the edge of the disk. Its
madreporite cannot be seen externally, and the structure
of its water vascular system must be inferred from old
descriptions of other ophiuroid species (cf. Hyman, 1955;
Nichols, 1966).

Because the internal parts of the water vascular system
are all small and embedded in ossicle and fibrous con-
nective tissue, they are best studied with histological
methods. To locate the site of the madreporite and ex-
amine the organization of the entire system, two small
specimens (6-7 mm disk diameter) were fixed in Helly's
fixative and decalcified for one week in changes of cold
(4C) 5% disodium EDTA. After paraffin embedding, se-
rial frontal sections (from the mouth through the top of
the disk) were cut at 10 /^m thickness and stained in
Delafield's hetnatoxylin, with eosin and orange G as
counter stains.

The extent of seawater entry through the madreporite
was evaluated in two sets of experiments. The first of
these was actually an unpublished part of a former study
dealing with ingestion and hemal transport by Ophio-
derma( Ferguson, 1985). A pair of brittlestars was placed
for 8 h in seawater containing 100 ^Ci' 4 C synthetic algal
protein hydrolysate, 375 mg glycine, 990 mg glucose,
and 10 mg chloramphenicol per liter. They were then
fixed and decalcified in Bouin's solution, embedded in
paraffin, sectioned (8 ^m). and exposed for up to
6 months to Kodak AR-10 autoradiographic film. In
the second set of experiments, fluorescent microbeads
were employed. These had been used to study madre-
porite function of the starfish Leptasterias hexactis (see
Ferguson. 1990a). Two brittlestars were placed in a dish
containing about 100 ml of seawater along with about
0.5 ml of 0.2-/jm YG "Fluoresbrite" carboxylate beads
(Polysciences, Inc.). After 48 h they were removed,
rinsed in seawater, fixed for 24 h in 10% formalin, de-
calcified for 7 days in cold 5% disodium EDTA,
embedded in gelatin, and cut as 20-Mm frozen sections.
The distribution of the beads was then observed by epi-
fluorescent UV microscopy.

Further tests were conducted to see if madreporite
oblation would lead to diminished functions. Specimens
(groups of 6) were anesthetized by a brief immersion
in 8% MgCl in tap water. The region of the madreporite
and stone canal (identified by the pigmentation pattern)
was then surgically destroyed and sealed with a small

Figure 7. Section just above Figure 6, showing the primary madreporite pore (PP) in the crevice of the
genital hursal slit. A, ampulla.

Figure 8. Section located 40 jim higher than Figure 7. The ampulla (A) is opening into the bottom of
the right axial sinus (RAS). opposite to the opening of the stone canal (SO, which contains long flagella.
The lower axial organ (AO) can now be seen in the left axial sinus (LAS).

Figure 9. Section located 60 ^m above Figure 8, showing typical form of the axial complex in this region.
nThe axial organ (AO) here is thin and contains only a few cell nuclei. It is applied closely to the flagellated
columnar cells of the stone canal (SC). LAS, left axial sinus; RAS. right axial sinus.

Figure 10. Section about 100 |im higher than Figure 9. At this level the axial complex lies between the
perivisceral coelom (PC) and the interradial muscle (IM). The axial organ (AO) gives off a pair of genital
hemal strands (GHS), surrounded by extensions of the left axial sinus (LAS). The right axial sinus (RAS) is
beginning to expand.

''>P?

-

/ ::./>

f s ,*; 1J V'-r? |iS ,' * ' /

'fc \ J j >t%^C- ^ * ''ft 1 * f v*'+ ** "^t* J ^

tw^-^.s*?

/' *: * iV -. ^X ,H - 1 :'*'

\ >t ^ : -.<

/.' i| I ^v*

1 v. :: ^^

sf ^ m^

'.-' .&' ^z~~ ,*PjS' -^

P'igure 11. Section about 170 ^m higher than Figure 10. In this region the nght axial sinus (RAS) has
expanded laterally and nearly surrounds the rest of the axial complex. It is separated by only a very thin
double peritoneal membrane from the extension of the perivisceral coelom (PC) that accompanies the large
interradial muscle (IM). GB, genital bursa; LAS, left axial sinus.

BRITTLESTAR WATER VASCULAR SYSTEM

103

cement plug. Control animals were treated similarly,
but a comparable wound was made through an oral
shield on the opposite side of the body from the mad-
reporite. All of the animals recovered from the imme-
diate effects of the anesthetic in about an hour. They
were then observed for 7 days to see if their tube feet
would extend and make normal stepping motions, and
if the specimens would right themselves when over-
turned. The animals also were periodically blotted and
weighed to see if weight variations could demonstrate
any changes in body fluid content. The cement plugs
were usually lost after about the second day as growing
tissue sealed the wound. Otherwise, all animals except
one, which fractured in two, appeared healthy through-
out the study period.

In a final set of observations, groups of three specimens
(tests and controls) previously operated on in the above
manner were placed in seawater to which sufficient dex-
tran (M.W. 5300) had been added to raise the osmotic
concentration 20 mosmoles/kg. The purpose of this ap-
proach was to try to counter any osmotic inflow by using
an osmolyte that should be neither permeable nor chem-
ically harmful to the integument. The animals were ob-
served and weighed as before until, on the third day, the
experiment was terminated.

Results

As previously noted madreporitic pores cannot be seen
in intact specimens of Ophioderma appressum. Exami-
nation of the serial sections revealed that they are small
and hidden by the outer edge of one of the oral shields.
It was found, however, that their position on the body
could usually be determined by the slightly greater amount
of pigmentation on that oral shield compared to the others,
or if the pigmentation were absent, by the dark shadow
of the ampullary complex showing through the translucent
oral shield (Figs. 1,2). From the oral perspective, the pores
lie just within the crevice of the central genital bursal slit
on the distal left side of the oral shield. Normally, water

currents could be seen passing into the peripheral slits of
the genital bursae and exiting through the central ones.
Thus, while the animals lie on the silty bay bottom, the
madreporitic pores are mainly bathed by seawater that
has passed through the genital bursae from the presumably
cleaner source at the edge of the disk. The water currents
within the genital bursae are maintained by cilia as well
as by rhythmical expansions and contractions of the whole
upper part of the disk.

Structure

The organization of the entire water vascular system is
shown in Figure 3, and a more detailed representation of
the complex madreporite-axial structure in Figure 4. Fig-
ures 5-18 show views of some of the serial sections from
which these two diagrams were developed (differences be-
tween the two specimens studied were negligible). To
avoid confusion in the following descriptions, the pho-
tomicrographs have been reversed in printing to compen-
sate for the normal optical inversion of the compound
microscope.

The primary opening into the system from the exterior
is a 10-15 nm diameter pore located in the folded edge
of the central genital bursal slit, just under the lip of the
oral shield (Fig. 7). Several undulations occur in the layer
of cuboidal cells that line the pore passage; raising the
possibility that more pores might develop in specimens
more fully grown then the two studied. A second pore of
nearly the same diameter is located about 150 ^m lateral
and slightly lower than the first (Fig. 6). Its duct points
towards the first and away from the stone canal. This
orientation suggests that the secondary pore may be a
rejection pathway, but there is no way of verifying that.

Both pores lead to a discrete lobe of the madreporite
"ampulla." This spacious chamber lies just under the sur-
face of the oral shield. It is lined with a distinctive cuboidal
epithelium, and it has several regions separated by sharp
angles (Figs. 5-8). After making tortuous turns, it opens
broadly into the lower end of the "right" axial sinus (Fig.

Figure 12. Section 70 Mm above Figure I I. The upper axial organ (UAO) has formed large cellular lobes
that hang down into the right axial sinus (RAS). The left axial sinus (LAS) penetrates the axial organ with
slit-like spaces (S) that continue through to the surrounding right axial sinus. The lower axial organ (LAO),
containing only scattered cell nuclei, still lies close to the stone canal (SO.

Figure 13. Next section above Figure 12. The stone canal (SC) here connects to the circumoral ring
canal (CRC). The cellular lobes of the axial organ and the slits between them connecting the two parts of
the axial sinus are quite evident.

Figure 14. Section 40 ^m above Figure 1 3. showing the top of the axial organ forming its cellular lobes.
Parts of the undulating circumoral ring canal (CRC) are visible on either side of it.

Figure 15. Section 60 nm above Figure 14, showing extensions from the upper axial organ (UAO) and
the right axial sinus (RAS) toward the circumoral hyponeural coelom (HC) and the oral hemal ring (not
visible). N. nerve ring.

Figure 16. A polian vesicle (PV) lying in the perivisceral coelom (PC) next to an interradial muscle
(IM). Note the heavy wall of the vesicle and its elastic membrane (arrow). GB, genital bursa.

104

OTC

J. C. FERGUSON

*.

r ^

If

v

7- \W i*X ' V f' r ' ''

-'.. ' '. ^ V'Vu " ' . ' * 'TC

y^ ; vvl.:%? -^ :'--.-feH*

_ ' ,? S, ,'-'* ' ' V- ' 'H " V*^

; //. ^-.--. ;4f. ' ^/>'. -M:* :^X> |

i^/ :#- T, ^>. ^v - ; .

-f . /'.' '.' v>'.-. >:> K'.-',:.; : ,:' x-.':v : .'-

' y ** ','^ fc '.vx ^' v *''. v' '''' '

/_ ~* . { ^ '/:.., 'i;',.,.": r !?*. ' . - -

17,

RC
\

N

UAO

UAO

/.,*^"H4
. 'j

Figure 17. The connection of an oral tube foot (TF) with its canal (OTC) from the circumoral ring
canal. There is a muscular restriction at the neck (arrow), hut no valve. N, nerve ring.

Figure 18. A transverse canal (TC) in an arm connecting to a tube foot (TF). It has a well-developed
valve. A strand of hemal tissue (H) connecting the tube foot to the radial hemal vessel lies nearby.

BRII IILSTAR WATER VASCULAR SYSTEM

105

8). Opposite this opening is the entrance to the lower end
of the stone canal (Fig. 8), which thus does not directly
connect with the ampulla, but rather with the right axial
sinus. On the far side of the stone canal (and attached
separately to it) is the "left" axial sinus, a somewhat
smaller chamber. It has no opening at the lower end, but
contains the axial organ (part of the hemal system),
broadly attached to the wall of the stone canal. Both parts
of the axial sinus are lined with a thin, simple squamous
peritoneum. The cells forming the stone canal stain
densely. They are cuboidal adjacent to the right axial sinus,
and columnar next to the axial organ. Long flagella extend
from the columnar cells and reach nearly across the 15-
20 nm width of the somewhat flattened lumen (Figs. 9-
1 2). The close proximity of the hemal tissue to the position
of the flagellated cells suggests that hemal fluid might
supply the high levels of energy nutrients that must be
needed by these cells to produce a current within the stone
canal.

The four structures (the two separate parts of the axial
sinus, the stone canal, and the axial organ) all rise in a
long arch towards the ring complex located high up in
the mouth frame. As they rise, they sweep over the large
interradial muscles (Figs. 10, 1 1). About one third of the
way up, the axial organ gives off the genital hemal strands
surrounded by extensions of the left axial sinus (Fig. 10).
These are thought to supply nutritive materials to the go-
nads (Walker, 1982; Byrne, 1988, 1989). Above that point,
the right axial sinus begins to expand and stretch out lat-
erally (Fig. 1 1 ). It comes to surround, almost completely,
the other three structures, and medially it forms a long
thin boundary with the perivisceral coelomic extension
that partially encases the interradial muscle. On the distal
side, it is separated from the perivisceral coelom by a thin
layer of connective tissue (Figs. 11-15).

After the right axial sinus begins its expansion, the left
axial sinus splits into a series of slit-like passages that enter
the axial organ, which now becomes highly cellular and
bulges out like a cauliflower (Figs. 1 2, 1 3). These passages
then open extensively into the right axial sinus, completing
the connection between the two portions of that cavity.
At its upper end, the axial organ becomes more solid again

(Fig. 14) and gives off an extension to the oral hemal ring
(Fig. 15). This extension is surrounded by a portion of
the right axial sinus, which appears to connect with the
hyponeural (perihemal) compartment that lies adjacent
to the nerve ring. This connection is not a spacious open-
ing, but a complex grouping of peritoneal cells and hemal
tissues that were difficult to resolve in the preparations
studied.

At the same level in which the two parts of the axial
sinus join, the top of the stone canal bends over and joins
the ring canal that encircles the mouth (Fig. 1 3). The ring
canal, like the other canals of the water vascular system,
is lined with squamous cells surrounded by an elastic
membrane, fibrous connective tissue, and a few muscle
cells (Figs. 13. 14). In the other four interradii, the ring
canal gives off canals that open into bulbous polian ves-
icles. These lie in the perivisceral coelom next to the in-
terradial muscles. Their walls possess a conspicuous elastic
membrane like that of the water canals (Fig. 16). In each
radius, the ring canal gives off three branches a radial
canal that descends and runs out the arm, and, well to
either side of it, canals that connect to upper and lower
oral tube feet that lie horizontally between the jaws, within
the mouth frame. There are no valves between the oral
tube feet and their connecting passages, although sphincter
muscles may be able to restrict the openings (Fig. 17).

Along the arms, transverse canals extend in pairs from
the radial canal. At the opening of each canal into a tube
foot, there is a well-developed valve that holds fluid within
the appendage (Fig. 18). Accessory elastic vesicles, such
as Woodley ( 1967) described as extending from the radial
canals of Amp/iiiim lililornns. were not seen. When the
tube foot retracts, it pulls up into a surrounding sheath
that is then closed over by two or three flattened spines.
When it extends, it slides out of this sheath as a unit and
then stretches out as a dexterous tentacle. Reiger and
Lombardi (1987) have reported on the ultrastructure of
the wall of the tube feet of Ophioderma brevispinum and
other species.

The radial canals lie in loose connective tissue in the
lower portion of the arm. Below them (above the nerve
cord) are extensions of the hemal tissues and hyponeural

Figure 19. An autoradiograph of an unstained radial section of the disk of an animal exposed to I4 C-
amino acids in seawater for 8 h. Note the considerable darkening (radioactivity) in the ampulla (A). A small
amount of label, perhaps from ingestion, is in the upper axial organ (UAO). The exposed epidermis (E) is
intensely labeled.

Figure 20. Autoradiograph of a section near that of Figure 18. Label is seen in the ampulla (A), lightly
in the stone canal (SC), and a bit more intensely in the upper axial organ (UAO).

Figure 21. Autoradiograph of a transverse section of an arm of the same specimen as Figures I 1 '. 20.
No radioactivity is seen in the radial canal (RC), but it is found clearly in the radial hemal vessel (H). Some
label may be in the lining of the tube foot (TF). but there is strong background from the heavily labeled
epidermis. N, radial nerve cord.

Figure 22. Autoradiograph of a radial section of arm base of the same specimen as others. Note the
high level of uptake in the radial hemal vessel (H). N. radial nerve cord.

106

J. C. FERGUSON

Figure 23. This and the next five figures show epifluorescent views of specimens exposed for 48 h to
fluorescent microbeads in seawater. Here a pore canal (C) extends towards the ampulla (arrow ). Numerous
beads are in the pore canal and epidermis (E); smaller numbers are in the cellular lining of the ampulla.

Figure 24. A view of the ampulla (A) just above the epidermis (E) of the oral shield. A few beads are
found in the ampullary chamber and some in its cellular lining.

BRITTLESTAR WATER VASCULAR SYSTEM

107

Table I

Weight variations (g) </Ophioderma oppression following destruction
of the madreporite area, and control animals with a comparable injury
to another site

Specimen Start 1 day 2 days 3 days 4 days 7 days

Matlreporite destroyed

1

1.14

1.12

1.11

1.07

1.07

1.09

2

1.01

1.01

1.02

1.02

1.07

1.04

3

1.29

1.28

1.29

1.26

1.32

1.28

4

1.49

1.50

1.57

1.50

1.57

1.55

5

0.61

0.61

0.60

0.62

0.66

0.63

6

1.10

1.09

(fragmented)

Controls

1

0.98

0.98

0.97

0.97

0.98

0.94

2

1.97

1.96

1.98

1.97

1.99

1.96

3

0.83

0.83

0.87

0.85

0.83

0.83

4

0.58

0.50

0.48

0.50

0.50

0.50

5

0.59

0.56

0.57

0.58

0.58

0.59

6

0.64

0.59

0.58

0.59

0.58

0.59

coelomic spaces. Hemal tissue extends laterally to each
tube foot (Fig. 18) where it appears to join, near the valve,
with the middle connective tissue layer of the appendage.

Tracer studies

Autoradiographs of animals placed for 8 h in ' 4 C-amino
acid mixture showed intense incorporation of the tracer
within all exposed epidermal cells including those of the
tube feet and their sheaths. This result was expected from
earlier studies (Ferguson, 1967; Fontaine and Chia, 1968).
The concern here was whether the tracer could demon-
strate seawater entry into the water vascular system. That
was the case, as notable radioactivity was found in the
lining cells of the madreporite ampulla (Figs. 19, 20). This
labeling diminished rapidly up the stone canal (Fig. 20).
[In animals fed labeled surf clams, radioactivity was found
abundantly in the hemal tissues, but not extensively in
the water vascular passages (Ferguson, 1985)]. No label
was visible within the radial canals, and amounts in the
lumen of tube feet were low and not readily distinguished

from background (Fig. 2 1 ). The radial hemal vessels ac-
cumulated considerable amounts of label (Figs. 21, 22),
and a small amount was seen in the upper, cellular portion
of the axial organ (Fig. 19). These accumulations could
have come from an oral route (cf. Ferguson, 1985).

Fluorescent microbeads were employed as a means
of more directly following the flow of seawater into the
water vascular passages. Unlike the labeled amino acids,
these beads should not easily penetrate membranes or
become rapidly depleted by cells as they take up nu-
trients from the fluid flowing by them. However, the
entry of beads into the body must overcome the ani-
mal's natural defenses against foreign particles. A very
large number of beads was used in the medium with
the expectation that a proportion of them would get
through if bulk fluids were indeed entering the body
through the pores. After 48 h of exposure, fluorescent
beads were found in many water vascular passages, but
not in great abundance (Figs. 23-28). Beads were seen
moderately distributed within the lining of the ampulla
(Figs. 23, 24). Some were found free in the radial canals
(Fig. 25) and in the tube feet (Fig. 27). More commonly,
beads were found engorged in clumps of coelomocytes.
These often accumulated in the lateral canals just before
the valves (Fig. 26), or within the lumen of tube feet
(Fig. 28). Many tube feet, however, could be seen with-
out any beads within them. Coelomocytes bearing many
beads collected in the hemal tissues of the arms, es-
pecially between the radial hemal vessels and the tube
feet, and beneath the radial canals (Fig. 25). A few
clumps of coelomocytes containing beads were found
in the perivisceral coelomic spaces of the arms. All ex-
posed epidermal areas were filled with beads, including
the areas just outside the madreporitic pores (Fig. 23)
and the sheaths of the tube feet (Figs. 27, 28).

Experiments

Because the tracer studies showed that seawater must
enter the madreporitic pores, at least in small quantities,
a study was made to see whether that uptake was es-
sential to the animals. It was found that surgical oblit-
eration of the madreporite, ampulla, and lower stone

Figure 25. Transverse section through the ambulacrum of an arm. Several discrete microbeads lie in
the radial canal (arrow). Below the arrow is part of a hemal strand extending towards a tube foot. It contains
many coelomocytes engorged with the beads. Most of the other fluorescence is natural.

Figure 26. A clump of coelomocytes engorged with beads lying in a transverse canal just before the
valve. Two or three coelomocytes containing beads are nearby, within the tube foot.

Figure 27. Transverse section through a tube foot lying in its sheath. A number of the fluorescent beads
are seen in the lumen (arrow). The epidermal tissue (E) of the tube foot and its sheath contains many beads
that were taken up directly from the seawater.

Figure 28. Transverse section through another tube foot. This one contains, in its lumen, many free
beads as well as coelomocytes engorged with beads.

108

J. C. FERGUSON

Table II

II t>/i,'/;; varuiliim* (g) l Ophioderma appressum in scauatcr raised 20
inouiMle.s/kH with de\nwi (5300 Jl/.H'J; specimen* mill inailrci'onlc.s
destroyed uiul control animal* with a comparable injury in another site

Specimen

Start

4 hours

1 day

2 days

3 days

Madri'poriie clem roved

1

1.09

1.04

1.05

1.06

1.12*

2

1.28

1.24

1.23

1.23

1.28*

3

0.65

0.61

0.60

0.60

0.69*

Controls

1

0.69

0.66

0.64

0.64

0.65*

2

0.45

0.43

0.41

0.40

0.42*

3

1.09

1 .05

1.01

1.02

1.04*

* Animals abnormal arched up and rigid.

canal had very little effect on tube foot function for at
least a week. For the first two or three days, tube foot
activity declined somewhat, especially in the oral tube
feet, but there was no consistent difference in observed
behavior when compared to controls. All tube feet could
extend and bend, and when animals were overturned,
righting movements involving the arms were unaffected.
The number of tube feet active at any one moment may
have diminished, but with the diversity of individual
behavioral responses demonstrated by the specimens,
that could not be quantified. Nor were there any con-
sistent variations in body weights that would reflect fluid
volume changes (Table I). Both test animals and con-
trols varied a few percent from day to day, but not sig-
nificantly.

If tube foot inflation and body fluid content are main-
tained by osmotic elevation, as by a potassium ion pump
in the tube feet (if. Prusch, 1977), the mechanism could
be sensitive to elevated colloidal osmotic pressure in
the medium. To test this possibility, animals with obliter-
ated madreporites, and controls, were placed in dishes
of seawater in which the osmotic levels had been ele-
vated a modest amount (20 mosmoles/kg) with dextran
(5300 M.W.). In both groups the immediate effect was a
small loss in weight (2 to 4%) over the first few hours and
then stability within a normal range (Table II). For the
next two days there was no diminishment of tube foot
function or other observable effects. On the third day.
animals in both groups showed some arching rigidity ot
their arms, rather similar to that seen previously in animals
placed in ionically altered seawaters. At that point the
experiment was terminated.

Discussion

This study has shown that Ophindcnna has a complex
water vascular system, but one in which the passages from

the exterior (the madreporite pores) appear to be of minor
importance compared to provisions for internal recircu-
lation of fluid via the axial sinus and its extensions. Al-
though it was shown that seawater routinely enters the
pores, it must do so only in small quantities. Under the
laboratory conditions employed, this uptake seemed to
provide little advantage. Perhaps under the stresses of the
natural environment uptake may be important in some
circumstances, but such conditions have yet to be dis-
covered. Fluid might also exit the pores when pressures
in the system become too great.

In contrast to the minimal madreporite of Ophioder-
ma, the madreporites of asteroids have many pores and
complex arrangements of ciliated gutters to keep them
free of suspended foreign particles (Ferguson and
Walker, 199 1 ). In one study on Echinaster graminicola,
seawater uptake through the madreporite was equal to
about 5.5% of the animal's body weight per day; and
more than half of it went into replacing the general
body fluid (Ferguson. 1989). The two small pores of
Ophioclcrnui. located very inconspicuously at the edge
of a genital bursal slit, clearly cannot allow much sea-
water into the system. Reduced inflow would alleviate
contamination from the silty environment in which the
animals normally live. Further, the genital bursa might
protect the pores from silt, as the asteroid gutters do.
Likewise, the complex form of the ampulla probably
plays a sanitary role, because many beads were seen
taken up by its cells.

The stone canal itself is fairly well developed. It is
smaller in diameter than those of asteroids, and it does
not have the internal ridges that make a larger diameter
tube more efficient as a ciliary pump. It also does not have
as much bony ossicle material, needed by a larger structure
for strength. However, for the much smaller body size of
Ophioflcmui. the stone canal seems proportionately
scaled. Its lumen is somewhat flattened, which allows the
flagellated cells to be very efficient in sweeping fluid
through the canal.

As noted, the stone canal does not connect directly
with the ampulla, but rather with the right axial sinus.
If fluid flows down this sinus into the stone canal, as it
appears to, it would be drawn from three sources: ( 1 )
the left axial sinus (and the genital perihemal vessels),
through the slits of the axial organ; (2) the circumoral
hyponeural (perihemal) coelom (and its connections
with the radial hyponeural coelomic spaces of the arms),
over the surface of the axial organ; and (3) through the
delicate peritoneal membranes separating much of the
axial sinus from the perivisceral coelom. I conclude,
then, that most of the fluid pumped by the stone canal
is coelomic not seawater flowing in through the two
madreporitic pores. As the fluid flows over the axial
organ it is probably purified, and it probably gives up

BRITTl FSTAR WATER VASCULAR SYSTEM

109

nutritive materials to form hemal fluid. [The transport
of nutritive material by hemal systems was previously
described (Ferguson, 1984, 1985)].

The fluid pumped by the stone canal then passes
through the water vascular canals, and most of it even-
tually travels out the arms. Some is diverted to the oral
tube feet and the polian vesicles, which among other
functions, probably collectively serve as a general reservoir
and pressure stabilizer for the system. In the arms, the
fluid passes through the valves into the tube feet only
when the hydrostatic pressure of the system surpasses that
maintained in the appendages. The tube feet also might
be kept inflated by osmotic inflow. Based on the obser-
vations on other animals by Robertson (1949), Binyon
(1976b), Prusch (1977), and Ferguson (1990b), the tube
feet likely contain a higher osmotic pressure than the sur-
rounding seawater. In the present experiments, however,
the tube feet failed to collapse when the external osmotic
pressure was increased with dextran, though by the third
day the treatment appeared to produce pervasive detri-
mental effects on the bodies of the animals. Although the
water vascular vessels can deliver fluid to the tube feet,
they may be equally valuable in providing a more general
circulatory flow by permeation to all the tissues of the
lower arms, and return via the hyponeural spaces. Ad-
ditional circulation is achieved by the perivisceral coe-
lomic passages.

When compared with the body cavities of asteroids,
the perivisceral coelom of Ophioderma is not large.
Within the arms it consists mostly of canals that ex-
pand into larger spaces between the vertebral-like os-
sicles. Asteroid perivisceral fluid is kept under low,
but positive, hydrostatic pressure (Ferguson, 1988).
That cannot be the case in Ophioderma, Distinct
"breathing" motions of the aboral disk alternately
stretch and compress the coelomic space, and that
movement pumps seawater in and out of the genital
bursae. The ventilation must produce negative coe-
lomic pressures that should lead to the accumulation
of fluid in the coelomic space by nitration. Pressure
from pumping by cilia in the genital bursae would have
the same effect. [Net negative coelomic pressures have
recently been described in sea urchins by Ellers and
Telford ( 1992), but these are produced by a different
mechanism.] It appears, then, that Ophioderma has
little need to take up seawater through its madreporite
either to support its tube feet or to maintain its peri-
visceral coelomic fluid, and it has a limited ability to
do so. Asteroids, on the other hand, often do tend to
lose large amounts of fluid from their bodies and must
replace it. For them, the madreporite system (together
with the Tiedemann's bodies) is a much more impor-
tant and well-developed mechanism.

Literature Cited

Binyon, J. 196-4. On the mode of functioning of the water vascular
system of Asterias nihcns L. ./ Mar. Bio/. Assoc. U. A" 44: 577-
588.

Binyon, J. 1966. Salinity tolerance and ionic regulation. Pp. 359-378
in Physiology of Echinodermata, R. A. Boolootian, ed. Wiley-Inter-
science. New York.

Binyon, J. 1976a. The permeability of the podial wall to water and
potassium ions. J. Mar Biol Assoc V. K 56: 639-647.

Binyon, J. 1976b. The effects of reduced salinity upon the starfish
Aslenas rubens L. together with a special consideration of the
integument and its permeability to water. Thalassia Jugoslav
12: 15-20.

Binyon. J. 1980. Osmotic and hydrostatic permeability of the integu-
ment of the starfish A stenas ruhens ./. Mar Biol Assoc. U. A 60:
627-630.

Binyon, J. 1984. A re-appraisal of the fluid loss resulting from the
operation of the water vascular system of the starfish. Asterias rubens
J. Mar Biol Assoc V K 64: 726.

Byrne, M. 1988. Evidence for endocytotic incorporation of nutrients
from the hemal sinus by the oocytes of the brittlestar Ophiolepix
paucispimi. Pp. 557-563 in Echinoderm Biology: Proceedings oj the
St.\th International Echinoderm Conference, l'ictoria/23-28 August
1987. R. D. Burke, P. Mladenov, P. Lambert and R. L. Parsley, eds.
Balkema. Rotterdam.

Byrne, M. 1989. Infrastructure of the ovary and oogenesis in the ovo-
viparous ophiuroid Ophiolepis nuticispina (Echinodermata). Biol.
Bull 176: 79-95.

Ellers, O., and M. Telford 1992. Causes and consequences of fluctuating
coelomic pressure in sea urchins. Biol. Bull 182: 424-434.

Ferguson, J. C. 1967. An autoradiographic study of the utilization
of free exogenous amino acids by starfishes. Biol. Bull 133: 317-
329.

Ferguson, J. C. 1984. Translocative functions of the enigmatic or-
gans of starfish the axial organ, hemal vessels. Tiedemann's
bodies, and rectal caeca: an autoradiographic study. Biol. Bull
166: 140-155.

Ferguson, J. C. 1985. Hemal transport of ingested nutrients by the
ophiuroid, Ophioderma brevispinum Pp. 663-626 in Proceedings of
the l-'itih International Echinoderm Conference Galway/24-29 Sep-
tember 1984. D. F. Keegan and B. D. S. O'Connor, eds. Balkema.
Rotterdam.

F'erguson, J. C. 1988. Madreporite and anus function in fluid vol-
ume regulation of a starfish (Echinaster graminicola). Pp. 603-
609 in Echmotlenn Biology: Proceedings of the Sixth International
Ecliiiwdcrm Conference. \'ictoria/23-28 August 19S7. R D.
Burke, P. Mladenov, P. Lambert and R. L. Parsley, eds. Balkema,
Rotterdam

Ferguson, J. C. 1989. Rate of water admission through the madreporite
of a starfish. / E.\p. Biol. 145: 147-156.

F'erguson, J. C. 1990a. Hyperosmotic properties of the fluids of the
perivisceral coelom and watervascular system of starfish kept under
stable conditions. Comp. Physio/. Biochem 95A: 245-248.

F'erguson, J. C. I990b. Sea water inflow through the madreponte and
internal body regions of a starfish (Leptasteriax hexactis) as dem-
onstrated with fluorescent microbeads. / E.\p. Zool 255: 262-271.

F'erguson, J. C. 1992. The function of the madreporite system in body
fluid volume maintenance by an intertidal starfish. Pisaster ochraceus.
Biol. Bull. 183: 482-489.

F'erguson, J. C. 1994. Madreponte inflow of seawater to maintain body
fluids in five species of starfish. Proceedings of the 8th International
Echinoderms Conference. Dijon. France. D. Bruno and A. Guille,
eds. Balkema, Rotterdam (in press).

110

J. C. FERGUSON

Ferguson, J. C., and C. VV. Walker. 1991. Cytology and function of
the madreponte systems of the starfish Henricia sanguinolenta and
Aslerias vulgaris. J Morphol 210: 1-11.

Fontaine, A. R., and F. S. Chia. 1968. Echinoderms: an autoradio-
graphic study of the assimilation of dissolved organic molecules. Sci-
ence 161: 1153-1155.

Hyman, L. 1955. The Invertebrates. II' Echinodermata. McGraw-Hill,
New York.

Nichols, D. 1966. Functional morphology of the water-vascular system.
Pp. 219-244 in Physiology of Echinodermala, R. A. Boolootian, ed.
Wiley-Interscience, New York.

Prusch, R. D. 1977. Solute secretion by the tube foot epithelium in
the starfish Asteriax forbesi. J Exp. Biol 68: 35-43.

Reiger, R. M., and J. Lombard!. 1987. Ultrastructure of coelomic lining
in echinoderm podia: significance for concepts in evolution of muscle
and peritoneal cells. Zoomorphology 107: 191-208.

Robertson, J. D. 1949. Ionic regulation in some marine invertebrates.
/ Exp. Biol. 26: 182-200.

Walker, C. W. 1982. Nutrition of gametes. Pp. 449-468 in Echi-
noderm Nutrition. M. Jangoux and J. M. Lawrence, eds. Balkema,
Rotterdam.

Woodley, J. D. 1967. Problems in the ophiuroid water-vascular system.
Svmp. Zoo/. Soc. Lund. 20: 75-104.

CONTENTS

No. 1, FEBRUARY/MARCH 1995

RESEARCH NOTES

Fluck, Richard A.

Responses of the medaka fish egg (Oryzias latipes)
to the photolysis of microinjected nitrophenyl-

EGTA, a photolabile calcium chelator

Wittenberg, Jonathan B., and Jeffrey L. Stein
Hemoglobin in the symbiont-harboring gill of the
marine gastropod Alvinichoncha hessleri

BIOMINERALIZATION

Giles, R., S. Manne, S. Mann, D. E. Morse, G. D.
Stucky, and P. K. Hansma

Inorganic overgrowth of aragonite on molluscan
nacre examined by atomic force microscopy . . .

Thorn, Kurt, Robert M. Gerrato, and Mark L. Rivers

Elemental distributions in marine bivalve shells as
measured by synchrotron x-ray fluorescence ... 57
Gherardi, Francesca, and Paul M. Cassidy

Life history patterns of Discorsopagurus schmitti. a
hermit crab inhabiting polychaete tubes .... 68

NEUROBIOLOGY AND BEHAVIOR

Carlberg, Mats, Karin Alfredsson, Sven-Olle Nielsen,
and Peter A. V. Anderson

Taurine-like immunoreactivity in the motor nerve

net of the jellyfish Cyanea capillata 78

DEVELOPMENT AND REPRODUCTION

Bates, William R.

Direct development in the ascidian Molgula retor-
tiformis (Verriil, 1871) 16

Chang, Wen-Teh, and Robert J. Lauzon

Isolation of biologically functional RNA during
programmed death of a colonial ascidian 23

Hamel, Jean-Francois, and Annie Mercier

Prespawning behavior, spawning, and development

of the brooding starfish Leptasterias polaris 32

ECOLOGY AND EVOLUTION

Mead, Kristina S., and Mark W. Denny

The effects of hydrodynamic shear stress on fer-
tilization and early development of the purple sea
urchin Strongylocentrotus purpuratus 46

Miklosi, Adam, Jozsef Haller, and Vilmos Csanyi

The influence of opponent-related and outcome-
related memory on repeated aggressive encounters
in the paradise fish (Macropodus opercularis) .... 83

Young, Craig M., and Roland H. Emson

Rapid arm movements in stalked crinoids 89

PHYSIOLOGY

Ferguson, John C.

The structure and mode of function of the water
vascular system of a brittlestar, Qphioderma apressum 98

Volume 188

THE

Number 2

BIOLOGICAL
BULLETIN

995

APRIL, 1995

Published by the Marine Biological Laboratory

THE

BIOLOGICAL BULLETIN

PUBLISHED BY
THE MARINE BIOLOGICAL LABORATORY

Associate Editors

PETER A. V. ANDERSON. The Whitney Laboratory. University of Florida

WILLIAM D. COHEN, Hunter College, City University of New York

DAVID EPEL, Hopkins Marine Station, Stanford University

J. MALCOLM SHICK, University of Maine, Orono

Editorial Board

PETER B. ARMSTRONG. University of California, Davis
THOMAS H. DIETZ, Louisiana State University
DAPHNE GAIL FAUTIN, University of Kansas

WILLIAM F. GILLY, Hopkins Marine Station, Stanford

University

ROGER T. HANLON, Marine Biomcdical Institute,
University of Texas Medical Branch

MICHAEL LABARBERA, University of Chicago
CHARLES B. METZ. University of Miami

K. RANGA RAO. University of West Florida

BARUCH RINKEVICH, Israel Oceanographic &
l.imnological Research Ltd.

RICHARD STRAIHM ANN, Friday Harbor Laboratories,
University of Washington

STEVEN VOGEL, Duke University

J. HERBERT WAITE, LJniversity of Delaware

SARAH ANN WOODIN, University of South Carolina

RICHARD K.. ZIMMER-FAUST. University of South

Carolina

Editor: MICHAEL J. GREENBERG, The Whitney Laboratory. University of Florida
Managing Editor PAMELA L. CLAPP. Marine Biological Laboratory

APRIL, 1995

Printed and Issued by
LANCASTER PRESS, Inc.

3575 HEMPLAND ROAD
LANCASTER, PA

THE BIOLOGICAL BULLETIN

THE BIOLOGICAL BULLETIN is published six times a year by the Marine Biological Laboratory. MBL
Street, Woods Hole, Massachusetts 02543.

Subscriptions and similar matter should be addressed to Subscription Manager. THE BIOLOGICAL BUL-
LETIN, Marine Biological Laboratory, Woods Hole, Massachusetts 0254.1. Single numbers, $37.50. Sub-
scription per volume (three issues), $92.50 ($185.00 per year for six issues).

Communications relative to manuscripts should be sent to Michael J. Greenberg, Editor-in-Chief, or
Pamela L. Clapp, Managing Editor, at the Marine Biological Laboratory, Woods Hole, Massachusetts 02543.
Telephone: (508) 548-3705, ext. 428. FAX: 508-540-6902. E-mail: pclappwihoh.mbl.edu.

THE BIOLOGICAL BULLETIN is indexed in bibliographic services including Im/c.\ Medicus and MEDLINE,
Chemical Ahstriicts. and Current Contents.

Printed on acid tree paper,
effective with Volume 180, Issue I, 1991.

POSTMASTER: Send address changes to THE BIOLOGICAL BULLETIN, Marine Biological Laboratory.

Woods Hole, MA 02543.

Copyright 1995, by the Marine Biological Laboratory

Second-class postage paid at Woods Hole, MA, and additional mailing offices.

ISSN 0006-31 85

INSTRUCTIONS TO AUTHORS

TVie Biological Bulletin accepts outstanding original research
reports of general interest to biologists throughout the world.
Papers are usually of intermediate length (10-40 manuscript
pages). A limited number of solicited review papers may be ac-
cepted after formal review. A paper will usually appear within
four months after its acceptance.

Very short, especially topical papers (less than 9 manuscript
pages including tables, figures, and bibliography) will be pub-
lished in a separate section entitled "Research Notes." A Re-
search Note in The Biological Bulletin follows the format of
similar notes in Nature. It should open with a summary para-
graph of 150 to 200 words comprising the introduction and the
conclusions. The rest of the text should continue on without
subheadings, and there should be no more than 30 references.
References should be referred to in the text by number, and
listed in the Literature Cited section in the order that they appear
in the text. Unlike references in Nature, references in the Re-
search Notes section should conform in punctuation and ar-
rangement to the style of recent issues of The Biological Bulletin
Materials and Methods should be incorporated into appropriate
figure legends. See the article by Lohmann el al. (October 1990,
Vol. 179: 214-218) for sample style. A Research Note will usually
appear within two months after its acceptance.

The Editorial Board requests that regular manuscripts con-
form to the requirements set below, those manuscripts that do
not conform will be returned to authors for correction before
review.

1. Manuscripts. Manuscripts, including figures, should
be submitted in triplicate. (Xerox copies of photographs are not
acceptable for review purposes.) The submission letter accom-
panying the manuscript should include a telephone number, a
FAX number, and (if possible) an E-mail address for the cor-
responding author. The original manuscript must be typed in
no smaller than 12 pitch or 10 point, using double spacing (in-
cluding figure legends, footnotes, bibliography, etc.) on one side

of 16- or 20-lb. bond paper. 8 1 /? by 1 1 inches. Please, no right
justification. Manuscripts should be proofread carefully and er-
rors corrected legibly in black ink. Pages should be numbered
consecutively. Margins on all sides should be at least 1 inch (2.5
cm). Manuscripts should conform to the Council of Biology Ed-
itors Style Manual. 5th Edition (Council of Biology Editors,
1983) and to American spelling. Unusual abbreviations should
be kept to a minimum and should be spelled out on first reference
as well as denned in a footnote on the title page. Manuscripts
should be divided into the following components: Title page.
Abstract (of no more than 200 words). Introduction. Materials
and Methods, Results, Discussion. Acknowledgments, Literature
Cited, Tables, and Figure Legends. In addition, authors should
supply a list of words and phrases under which the article should
be indexed.

2. Title page. The title page consists of a condensed title
or running head of no more than 35 letters and spaces, the
manuscript title, authors' names and appropriate addresses, and
footnotes listing present addresses, acknowledgments or contri-
bution numbers, and explanation of unusual abbreviations.

3. Figures. The dimensions of the printed page, 7 by 9
inches, should be kept in mind in preparing figures for publi-
cation. We recommend that figures be about I '/: times the linear
dimensions of the final printing desired, and that the ratio of
the largest to the smallest letter or number and of the thickest
to the thinnest line not exceed 1:1.5. Explanatory matter gen-
erally should be included in legends, although axes should always
be identified on the illustration itself. Figures should be prepared
for reproduction as either line cuts or halftones. Figures to be
reproduced as line cuts should be unmounted glossy photo-
graphic reproductions or drawn in black ink on white paper,
good-quality tracing cloth or plastic, or blue-lined coordinate
paper. Those to be reproduced as halftones should be mounted
on hoard, with both designating numbers or letters and scale
bars affixed directly to the figures. All figures should be numbered

in consecutive order, with no distinction between text and plate
figures. The author's name and an arrow indicating orientation
should appear on the reverse side of all figures.

4. Tables, footnotes, figure legends, etc. Authors should
follow the style in a recent issue of The Biological Bulletin in
preparing table headings, figure legends, and the like. Because
of the high cost of setting tabular material in type, authors are
asked to limit such material as much as possible. Tables, with
their headings and footnotes, should be typed on separate sheets,
numbered with consecutive Roman numerals, and placed after
the Literature Cited. Figure legends should contain enough in-
formation to make the figure intelligible separate from the text.
Legends should be typed double spaced, with consecutive Arabic
numbers, on a separate sheet at the end of the paper. Footnotes
should be limited to authors' current addresses, acknowledg-
ments or contribution numbers, and explanation of unusual
abbreviations. All such footnotes should appear on the title page.
Footnotes are not normally permitted in the body of the text.

5. Literature cited. In the text, literature should be cited
by the Harvard system, with papers by more than two authors
cited as Jones et ai, 1980. Personal communications and ma-
terial in preparation or in press should be cited in the text only,
with author's initials and institutions, unless the material has
been formally accepted and a volume number can be supplied.
The list of references following the text should be headed Lit-
erature Cited, and must be typed double spaced on separate
pages, conforming in punctuation and arrangement to the style
of recent issues of The Biological Bitllelin. Citations should in-
clude complete titles and inclusive pagination. Journal abbre-
viations should normally follow those of the U. S. A. Standards
Institute (USASI), as adopted by BIOLOGICAL ABSTRACTS and
CHEMICAL ABSTRACTS, with the minor differences set out below.
The most generally useful list of biological journal titles is that
published each year by BIOLOGICAL ABSTRACTS (BIOSIS List of
Serials; the most recent issue). Foreign authors, and others who
are accustomed to using THE WORLD LIST OF SCIENTIFIC PE-
RIODICALS, may find a booklet published by the Biological
Council of the U.K.. (obtainable from the Institute of Biology,
41 Queen's Gate, London, S.W.7, England, U.K..) useful, since
it sets out the WORLD LIST abbreviations for most biological
journals with notes of the USASI abbreviations where these differ.

CHEMICAL ABSTRACTS publishes quarterly supplements of ad-
ditional abbreviations. The following points of reference style
for THE BIOLOGICAL BULLETIN differ from USASI (or modified
WORLD LIST) usage:

A. Journal abbreviations, and book titles, all underlined
(for italics)

B. All components of abbreviations with initial capitals
(not as European usage in WORLD LIST e.g.. J Cell Camp.
Physiol. NOT/ cell, comp Physiol.)

C. All abbreviated components must be followed by a pe-
riod, whole word components musl not (i.e.. J Cancer Res.)

D. Space between all components (e.g.. J. Cell. Comp.
Physiol.. not J Cell Comp .Physiol )

E. Unusual words in journal titles should be spelled out
in full, rather than employing new abbreviations invented by
the author. For example, use Ril I 'Isiniiqfjelags fslendinga with-
out abbreviation.

F. All single word journal titles in full (e.g.. I'eliger, Ecol-
ogy. Brain).

G. The order of abbreviated components should be the
same as the word order of the complete title (i.e., Proc. and
Trans, placed where they appear, not transposed as in some
BIOLOGICAL ABSTRACTS listings).

H. A few well-known international journals in their pre-
ferred forms rather than WORLD LIST or USASI usage (e.g..
Nature. Science, Evolution NOT Nature. Loud.. Science. N.Y.;
Evolution. Lancaster. Pa)

6. Reprints, page proofs, and charges. Authors receive their
first 100 reprints (without covers) free of charge. Additional re-
prints may be ordered at time of publication and normally will
be delivered about two to three months after the issue date.
Authors (or delegates for foreign authors) will receive page proofs
of articles shortly before publication. They will be charged the
current cost of printers' time for corrections to these (other than
corrections of printers' or editors' errors). Other than these
charges for authors' alterations. The Biological Bulletin does not
have page charges.

Reference: Biitl. Bull 188: 111-116. (April,

Odor Plumes and Animal Navigation in Turbulent
Water Flow: A Field Study

RICHARD K. ZIMMER-FAUST 1 : \ CHRISTOPHER M. FINELLI",
N. DEAN PENTCHEFF 1 \ AND DAVID S. WETHEY 1 21

1 Department of Biological Sciences, 2 Marine Science Program, and* Belle (('. Barnch Institute

for Marine Biology and Coastal Research, University of South Carolina,

Columbia, South Carolina 2V208

Turbulence causes chemical stimuli to he highly variable
in time and space: hence the study of animal orientation
in odor plumes presents a formidable challenge. Through
combined chemical and physical measurements, we char-
acterized the transport of at tract ant released by clam prey
in a turbulent aquatic environment. Concurrently, we
quantified the locomotory responses of predatory crabs
successfully searching for sources of clam attractant. Our
results demonstrate that both rheotaxis and chemotaxis
are necessary for successful orientation. Perception of
chemical cues causes crabs to move in the upstream di-
rection, but feedback from attractant distributions directly
regulates movement across-stream in the plume. Orien-
tation mechanisms used by crabs differ from those em-
ployed by flying insects, the only other system in which
navigation relative to odor plumes has been coupled with
fluid dynamics. Insects respond to odors by moving up-
stream, but they do not use chemical distributions to de-
termine across-stream direction, whereas crabs do. Tur-
bulent eddy diflusivities in crab habitats are 100 to 1000
times lower than those of terrestrial grasslands and forests
occupied by insects. Insects must respond to plumes con-
sisting of highly dispersed, tiny filaments or parcels of odor.
Crabs rely more heavily on spatial aspects of chemical
stimulus distributions because their fluid dynamic envi-
ronment creates a more stable plume structure, thus per-
mitting chemotaxis.

The ability to detect chemical stimuli is nearly universal
among animals. Chemical signals are generated by lib-
erating stimulus molecules into a fluid, where they are
transported by advection and diffusion (eddy and molec-

Received 8 December 1994; accepted 25 January 1995.

ular) until detected and acted upon by a biological sensor.
Thus the physical process governing chemical transport
has a profound impact on the nature and success of che-
mosensory-mediated behavior ( 1, 2, 3, 4, 5, 6). Except at
microscopic scales, turbulence causes a stimulus pattern
whose concentration is highly variable in time and space
(3. 4. 5, 6. 7). As a result, the study of chemoreceptive
behavior presents a formidable challenge. It demands si-
multaneous measurements of stimulus release, fluid dy-
namics, and animal responses to perceived chemical cues.
Until now, however, these determinations have not been
combined in a single field study.

We experimentally establish a direct link between the
transport of chemical stimuli and animal navigation in a
natural, turbulent flow environment. Attraction of pred-
atory blue crabs (Callinectes sapidus) to odor released by
clam prey (Mercenaria mercenaria) was investigated. The
mechanisms directing predator to prey were identified for
crabs foraging naturally in estuarine tidal creeks, where
water flows unidirectionally for hours at a time. These
creeks were shallow enough to permit direct, noninvasive
observations of crab locomotory behavior.

Experiments were conducted in the North Inlet Estuary,
near Georgetown, South Carolina, USA (3220'N,
79 15' W). Flow records were obtained with an electro-
magnetic flow meter (Marsh-McBirney 523) equipped
with a two-dimensional sensor ( 1 cm diam), mounted on
a flat base, and submerged in the tidal creek. The base
was positioned flush with the sandy bottom, and the sensor
was placed 5 cm above the substrate. Sensor height was
chosen to match the elevation of a typical, adult crab
body. A data logger (Campbell CR10) was used to record
both horizontal and vertical flow velocities continuously
measured by the sensor at 1 Hz, between 26 June and 8

1 1 1

112

R K.. Z1MMER-FAUST ET AL

August 1993. Horizontal flow (downstream advection)
typically ranged from cm/s at slack tides, to 30 cm/s
during peak ebb and flood tides. We applied the eddy
correlation method to calculate shear velocities, at 1-min
intervals, using the correlation between the horizontal and
vertical flow velocities (8, 9). Shear velocity (;/*) is a mea-
sure of the strength and correlation of turbulent fluctua-
tions in flow speed near the substratum. Finally, we de-
termined the coefficients of turbulent mixing as the prod-
ucts of shear velocity, sensor height above the substratum,
and Von Karman's constant (9). These mixing coefficients
were remarkably low, ranging from 0.5 to 1.5 cirr/s. They
indicated that across-stream mixing occurred very slowly
in the tidal creek even though water flow was turbulent.

The dynamics of odor plumes were characterized by
measuring the transport of fluorescent dye (sodium flu-
orescein) and an electrochemical (dopamine) following
their combined release from a point source. Input con-
centrations of fluorescein and dopamine were l.Og/liter
and 2 mmol/1, respectively, with ascorbic acid added to
the mixture at 20 mmol/1 as an antioxidant. The mixture
was introduced through polyethylene tubing (0.5 mm ID)
at 6 ml/min.

Fluorescein provided a visual marker, and fluorometric
determinations were used to establish the time-averaged
distributions of dye at downstream and at across-stream
sites, relative to the release point. Water samples were
collected (at 1 ml/s) simultaneously over 1-min intervals
by syringes placed at each of 1 8 to 30 sites per trial. These
sites were distributed in a grid, with six sites placed across-
stream (0, 2, 4, 6, 8, 10 cm distant from the plume mid-
line) at three to five locations downstream of the release
point (5, 25. 50, and in some trials, 100 and 275 cm distant
from the source). The bell-shaped distribution of fluores-
cein concentration with gradual decay downstream (Fig.
1 ) is what a Gaussian plume model would predict (Pear-
son's product-moment correlation: /- : > 0.95; P < 0.001 ;
all replicate plume measurements). Concentration
dropped sharply at the plume's visible lateral edges
(Fig. 1).

Fluorometric measurements also provided an alter-
native method of calculating the mixing coefficient for
comparison with determinations made using the electro-
magnetic flow meter. Temporal changes in the across-
stream variance in fluorescein concentration were used
to estimate the mixing coefficient (9). Results from the
two methods matched well. For example, estimates
based on fluorometric determinations ranged from 0.5
to 1.2cnr/s during a time when estimates of 0.5 to
0.8 cm 2 /s were made from electromagnetic flow meter
records.

When measured at last temporal scales, chemical stim-
uli in odor plumes are patchily distributed due to tur-
bulence. Mean concentrations and time-averaged distri-

butions of fluorescein dye, therefore, may not be indicative
of the information available for crabs attempting to orient
towards an odor source (3, 4, 5, 6. 7). Because arthropod
chemoreceptors detect intermittent (or pulsed) chemical
stimuli applied at a maximum frequency of 4- 10 Hz (10.
1 1 ), we employed carbon fiber microelectrodes ( 150 /urn
diam) and a computer recording system (MedSystems
Corp. I VEC-1 Otto sample dopamine at 10 Hz (12). Elec-
trode recordings were made at the fluorescein sampling
sites (see above). Turbulent mixing caused the concen-
tration of dopamine sampled downstream of the source
to fluctuate strongly in time and space. Bursts of highly
concentrated chemical passed over the sensor, alternating
with periods of low or zero concentration (Fig. 1). The
plume's lateral edge, as defined by our high-speed dopa-
mine measurements, was positioned identically relative
to the edge detected both by our time-averaged fluorescein
measurements and by our visual observations. This lateral
edge, separating clean from chemical-laden water, was
very narrow (2-4 cm wide) compared to the body size of
an adult crab (10-15 cm carapace width). Thus a steep
concentration gradient was found across-stream, but not
downstream of chemical release.

We previously demonstrated that some crabs search
for and find intact live clams, and that these crabs are
responding to odor plumes created by the excurrent release
of attractant metabolites at low concentration (13). Once
a clam is found, however, it is chipped open by a crab
and attractants are released to form a plume of high con-
centration. High-concentration plumes then immediately
attract other crabs to the predation site. Hence, depending
on the situation, crabs may be exposed either to low or
high attractant concentrations, and crabs respond effec-
tively in each case.

Concurrently with hydrodynamic and chemical mea-
surements, we assessed crab orientation in odor plumes.
Our field studies focused on plumes characteristic of
chipped clams. We chose to work with chipped clams
because high attractant concentrations would better en-
sure an effective stimulus through the broad range of hy-
drodynamic conditions encountered by crabs in the field.
Stimulus plumes (dyed with fluorescein for visibility) were
created, either presenting a chipped clam or introducing
clam mantle fluid (membrane filtered to 0.22 ^m) at a
rate mimicking its release from a chipped clam. Each
stimulus plume was always paired with a control plume
that delivered fluorescein in filtered seawater. Both stim-
ulus and control solutions were introduced at 6 ml/min,
with inputs separated by 60 cm across-stream.

Free amino acid compositions of effluent leaking from
chipped clams (/; = 8 clams assayed), mantle fluid of intact
live clams (/; = 8 clams), and homogenized clam flesh
(n = : 7 clams) were all determined using a Beckman
6300 System Gold high-performance liquid chromato-

TURBULENT ODOR PLUMES AND ANIMAL NAVIGATION

50

25-

o

a
o
u

o - 1

Distance From Midline (cm)
-10 -5 5 10

50cm Y = 0cm X = SOcm Y=2cm X = 50cm Y = 4 cm X=50cm Y = 8 cm X = 50 cm Y=10cm

^ ou

_a

ou

c

^ 0.2 -
^

o

B 40 -

- 40

E

1|

|

c

u

5 0.1 -
X

820.

1

I ll

i u i

- 20

Q)

1 ll

O

D

"- nn -

III., I n

dLLllkU

uL.llkll

1

4M

i

Iliiijiiiv

. n

20 40 20 40 20 40 20 40 20 40 60

Time (s)

( = 25 cm Y = 0cm X = 25cm Y = 2cm X = 25 cm Y = 4 cm X = 25 cm Y = 6cm

^ 0.2 -

o>
E

c

'o> 0.1 -

Fluoresc

3
3

1,,

5 w

c
o

"5 40

o 20
O

60

40

20

20 40 20 40 20 40

Time (s)

X5cm Y-Ocm X = 5 cm Y = 1 cm X = 5 cm Y = 2 cm

20 40 60

D>

E

5 2 '

o

3

"- n -

fr

^ ou

3.

1

- ou

C

O

|40-

-40

1
o

c

020

ll

i i

i HI

-20

0)

UN

ii ,

i

8

kj 'M

y

m

i

ii i ,i

20 40 20 40 20 40 60

Time (s)

Plume Source

Figure I. Representative concentration distributions downstream from a point source in a tidal creek.
Histograms represent fluorescein concentrations (mg/1) in samples collected at 5 cm (bottom row), 25 cm
(middle row), and 50 cm (top row) downstream from the source and at 0, 2, 4, 6, 8, and 10 cm from the
midlmc of the plume. Note the scale difference between fluorescein concentrations at downstream locations.
The visible region of the fluorescein plume at each position is denoted by the shading. Panels to the right
of the histograms represent ftO-s records of instantaneous fluctuations in dopamine (tracer) concentration,
measured at 10 Hz with a carbon fiber electrode, at locations where the fluorescein was sampled. The left-
most panel in each row is the sample from the midline of the plume, and successive panels are samples
from 2, 4, 8, and 10cm from the midline (see tick marks on histogram axes for sampling sites). Highly
concentrated bursts of dopamine were common in all samples taken within the visible portion of the plume.

graph with a sodium ion-exchange column (4-mm ID
X 120mm; Beckman) for separation. In this system,
amino acids were monitored spectrophotometrically after
post-column reaction with ninhydrin. Compositions of
clam effluent and mantle fluid were almost identical
(Pearson's product-moment correlation: r 2 = 0.998; P
< 0.001; n = 18 amino acids chromatographed), indicat-
ing that mantle fluid was the source of effluent material.
Because taurine was by far the most abundant amino acid
in both clam effluent and mantle fluid (accounting for
>50% of the total amino acid composition), we used it
as a marker to measure the rates of fluid release from
chipped clams. In the laboratory, clams (n = 12) of various
sizes were chipped by using a metal rod to deliver a single

firm blow to the lateral shell margin. The resulting chip
was similar in size and shape to one produced by a blue
crab as it begins to feed. Each chipped clam was then
placed individually into a separate beaker of artificial sea-
water. The beakers were stirred, and they were maintained
at the same temperature and salinity as seawater in the
tidal creek from which clams were collected. Artificial
seawater was sampled from each beaker before placement
of the clam, and again at 30- to 60-s intervals for 15 min
after placement; HPLC analysis of this seawater indicated
that taurine (and mantle fluid) release was constant
throughout the trial period. The relation between taurine
release rate and clam size was then used to scale our de-
livery of mantle fluid in field experiments, simulating the

14

R k. ZIMMER-FAUST ET AL

40

X-N

H o
5

ce

o

"to

b

o>

O 40
<D

2! o

U = 4.2 cm/s

~w^^

U = 1 cm/s

**.

^ o
</>

-40

U = 25 cm/s

50 100 150 200 250

Distance Downstream from Source (cm)

Figure 2. Representative tracks of crabs following odor plumes. The crab symbols represent the positions
and orientations of individuals at 1-s intervals as they moved upstream toward the odor source. Naturally
foraging crabs normally walk sideways as well as forward. The visible region of the fluorescein/odor plume
is noted by the shading. The water velocity (U) at ? cm above the bottom was 4.2 cm/s (top panel). 10 cm/s
(center panel), and 25 cm/s (bottom panel). Distances downstream and across stream are in centimeters.

300

fluid input from an intermediate-sized chipped clam (6 cm
total shell length).

Quantitative observations of foraging crabs were made
noninvasively with a video camera (Sony TR8 1 ) mounted
4 m above the tidal creek. Video records of crabs re-
sponding to plumes were made during ebbing tides. A
3-m field of view was dictated by the resolution of the
video camera and the size of the crabs (10-15 cm carapace
width), whose positions could not be reliably quantified
in wider field images. A scale bar in the field of view was
employed to measure distance and to correct for distortion
due to perspective. In the laboratory, plume edges and
crab positions at 1-s intervals were traced onto acetate
sheets from video playbacks to a monitor. Both crab lo-

cation, in relation to the plume edge, and crab locomotory
kinematics were measured (Fig. 2).

Crab responses to the plumes were dramatic and un-
ambiguous: 29 crabs contacted the control plumes, but
only 4 of these crabs walked upstream towards the input
source. The near absence of positive responses to control
plumes demonstrated a lack of attractant effect by fluo-
rescein dye. In comparison, after contacting stimulus
plumes, 68 of 80 crabs walked upstream to the input
source. Crabs turned upstream within 1.5 s (0.3 SD) of
contacting an odor plume. Percentages of crabs respond-
ing positively to plumes from either chipped clams or
mantle fluid were nearly identical, being 86% (n = 52
crabs) or 82% (n = 28 crabs), respectively (G-test for ho-

TURBULENT ODOR PLUMES AND ANIMAL NAVIGATION

115

mogeneity: G 2 = 0.270; df = 1; P > 0.50). Our results
demonstrate an odor-conditioned rheotaxis that orients
crabs upstream. Previously, we reached an identical con-
clusion for blue crabs foraging in a laboratory flume. Crabs
walked upstream to find intact live clams in flowing water,
but they oriented indiscriminately and searched unsuc-
cessfully for clams in still water (13).

Oriented movements by crabs lateral to water flow are
controlled by chemotaxis. As crabs walked upstream to-
wards an attractant source, they frequently approached
the lateral edges of the plume. When crabs did reach the
edge, they nearly always turned directly back to the plume
(50 of 6 1 turns; G-test for goodness-of-fit. 1 : 1 hypothesized
ratio, G 2 == 14.97; df = 1; P < 0.001 ), without exhibiting
either casting or zigzagging (Fig. 2). Lateral movements
were initiated as crabs began to exit a plume and partially
contacted clean water. Fluorescein did not act as a visual
cue, because crabs displayed identical oriented responses
when tests were conducted in the dark (under infrared
illumination) and without fluorescein (13, and in prep.).
It took, on average, less than 1 s (0.8 0.2 s SD) for crabs
to renew upstream walking after they had begun moving
laterally towards the plume midline. Remarkably, we did
not observe walking speeds to change significantly as crabs
moved closer to attractant sources (analysis of covariance:
F = 0.60; df = 4,237; P = 0.66; walking speed: 12.8
0.4 cm/s SD), and we found no significant correlation
between walking speed and water flow velocity (Pearson's
product-moment correlation: /- : = 0.037; df = 1,64; P
= 0.12). We hypothesize that crabs perceive clam attrac-
tant as a binary cue (present/absent), both in their up-
stream movement and in their across-stream walking.
Because the plume edges were very sharp, when crabs
partially exited the plume, some pereiopods (legs or claws)
were outside the plume while others remained inside. A
comparison of simultaneous chemosensory inputs from
the appendages inside and outside the plume would pre-
sumably allow the crabs to determine the correct direction
and return to the plume. This binary response would
lead crabs to locations of higher concentration of clam
attractant.

Orientation mechanisms used by crabs in upstream
movement are similar to those of flying insects. However,
crabs differ from insects in their across-stream response.
Insects provide the only other system in which navigation
relative to odor plumes has been coupled with fluid dy-
namics. Flying insects locate a source of chemical attrac-
tant by moving upwind upon contacting a filamentous
trace of attractant odor (14, 15). After several seconds of
flying in clean air, insects shift to casting (regular reversals
of flight directed across-stream) until contact with another
odor trace causes a return to upwind flight (16). Flying
insects, therefore, do not use chemical concentration gra-
dients to determine either their upwind or across-wind

directions (16, 17, 18). The use of chemotaxis may be
impractical in their environment, where complex fluid
dynamics do not permit stable zones of high attractant
concentrations to exist. Crabs in contrast, consistently turn
back into the attractant plume rather than zigzagging after
losing the plume signal.

The difference between estuarine tidal creek flow and
atmospheric winds may explain why blue crabs and insects
use contrasting mechanisms for successful navigation to-
wards an odor source. The crop fields and forests used as
experimental models for insect flight are hydraulically
rough, with high advection. Eddy diffusivities in insect
habitats are 100 to 1000 times greater than those we re-
corded in estuarine tidal creeks (19, 20). Higher diffusiv-
ities yield plumes consisting of tiny, highly dispersed fila-
ments or parcels of odor. Wind direction changes fre-
quently, causing plumes to meander (3, 6). The dispersal
pattern of odor, coupled with the relatively fast flight speed
of insects, means that a flying insect has little chance to
detect more than the occasional pulse of passing odor.
Casting, zigzagging, and rapid behavioral modulation in
response to fine-scale changes in odor concentration may
be strategies appropriate for situations in which the entire
plume meanders away from the animal.

In contrast, the flow environment of estuarine tidal
creeks is markedly less turbulent, yielding relatively stable,
straight, and sharply delineated odor plumes. Plumes
cannot meander substantially, because flow is constrained
by water depth and by the sides of the creeks. A stable
plume structure permits direct binary comparisons of
chemical concentration inside and outside the plume, to
guide movement lateral to flow. The more direct plume-
following behavior and across-stream chemotactic re-
sponses shown by crabs reflect a strategy appropriate to
the plume structure characteristic of their environment.
Mechanisms of plume-following behavior, therefore, arise
in response to chemical stimulus distributions, as deter-
mined by the specific fluid dynamic environments in
which animals must naturally navigate.

Acknowledgments

We thank Dr. D. M. Allen, Director of the USC Baruch
Field Laboratory, for providing laboratory space and lo-
gistical support for field work. Dr. Y. Ishikawa, USC In-
stitute for Biological Research and Technology, performed
amino acid composition analyses. We also thank Dr.
S. A. Woodin, whose comments on earlier drafts greatly
improved this manuscript. This research was sponsored
by the National Science Foundation (IBN 92-22225) and
the University of South Carolina Research and Productive
Scholarship Fund.

Literature Cited

I Wright, R. H. 1958. The olfactory guidance of flying insects. Can
Entomtil. 90: 81-89.

116

R K. ZIMMER-FAUST KT AL

2. Bossert, \V. H., and E. O. Wilson. 1963. The analysis of olfactory'
communication among animals. / Theor. Biol 5: 443-469.

3. Murlis, J., and C. D. .(ones. 1981. Fine-scale structure of odour
plumes in relation to insect orientation to distant pheromone and
other attractant sources. Physiol. Entomol 6: 71-86.

4. Elkinton, J. S., R. T. Carde, and C. J. Mason. 1984. Evaluation
of time-average dispersion models for estimating pheromone
concentration in a deciduous forest. J. C/ie/n Ecu/. 10: 1081-
1108.

5. Zimmer-Faust, R. K., J. M. Stanfill, and S. B. Collard, III. 1988. A
fast, multi-channel fluorometer for investigating aquatic chemore-
ception and odor trails. I. annul Oceant;r. 33: 1586-1595.

6. Murlis, J., J. S. Elkinton, and R. I. Carde. 1992. Odor plumes
and how insects use them. Anmi Rev Entomol. 37: 505-532.

7. Moore, P. A., and J. Atema. 1991. Spatial information in the three-
dimensional fine structure of an aquatic odor plume. Biol. Bull 181:
408-418.

8. Schlichting, H. 1979. Boundary Layer Theory. McGraw-Hill. New
York. 486 pp.

9. Denny, M. W. 1988. Biolot<y and Mechanics <>/' ilie IIV/iv-
Sivepl Environment Princeton University Press. Princeton. NJ.
329 pp.

10. Kaissling, K. E., C. Z. Straussfeld, and E. Rumbo. 1987.
Adaptation processes in insect olfactory receptors: mechanisms and
behavioral significance. Annals N Y Acail Set 510: 104-1 12.

I I. Gomez, G., R. Voigl, and .J. Alema. 1994. Frequency filter prop-
erties of lobster chemoreceptor cells determined with high-resolution
stimulus measurement. ./ Comp PliYsml A 17-1: 803-81 1.

1 2. Moore, P. A., G. A. Gerhardt, and J. Alema. 1989. High resolution
spatio-temporal analysis of aquatic chemical signals using micro-
electrochemical electrodes. Chem Senses 14: 829-840.

13. VVeissburg, M. J., and R. K. Zimmer-Faust. 1993. Life and death
in moving fluids: hydrodynamic effects on chemosensory-mediated
predation. Ecolivi 1-4: 1428-1443.

14. Mafra-Nelo, A., and R. T. Carde. 199-4. Fine-scale structure of
pheromone plumes modulates upwind orientation of flying moths.
Nut lire 3W: 142-144

15. Vickers, N. J., and T. C. Baker. 1992. Male lleloilus vircsccns
maintain upwind flight in response to experimentally pulsed filaments
of their sex pheromone (Lepidoptera: Nocturidae). ,/ Insect Behav.
5: 669-687.

16. Baker, T. C. 1986. Pheromone-modulated movements of flying
moths. Pp. 39-48, in Mechanisms of Insect Olfaeliim. T. L. Payne.
M. C. Birch, and C. E. Kennedy, eds. Clarendon Press. Oxford.

17. David, C. T.. J. S. Kennedy, and A. R. Ludlow. 1983. Finding of
a se\ pheromone source by gypsy moths released in the held, \nliire
303: 804-806.

18. Arbas, E. A., M. A. \\ illis, and R. Kanazaki. 1993. Organization
of goal-oriented locomotion: pheromone-modulated flight behavior
of moths. Pp. 159-198. in Biological Neural Networks in Invertebrate
Neiirnellidlnny ami Roholics. R. D. Beer, R. E. Rit/mann. and T.
McKenna, eds. Academic Press, New York.

19. Shaw, R. H., J. Tavangar, and D. P. Ward. 1983. Structure of the
Reynolds stress in a canopy layer. J Climate Anpl Meteornl. 22:
1922-1931.

20. Arya, S. P. 1988. Introduction to Micrometeorology Academic
Press, San Diego. 303 pp.

Reference: Biol Bull 188: 117-119. (April, 1995)

Evidence for Selection Against Heterozygotes:

Post-Settlement Excess of Allozyme Homozygosity

in a Cohort of the Chilean Oyster,

Ostrea chilensis Philippi, 1845

J. E. TORO AND A. M. VERGARA

Instituto de Biologia Manna, L'niversidad Austral de Chile. Casilla 567, I 'aldivia. Chile

Reports of heterozygote deficiencies in electrophoretic
survevs carried mil in marine bivalves abound in the lit-
erature (1-6), but the mechanism or mechanisms produc-
ing this phenomenon have not been well defined. H 'e report
that, in the Chilean oyster (Ostrea chilensis), heterozygote
deficiencies in a cohort obtained hy mass spawning in the
laboratory are not randomly distributed in lime among
genotypes. The eggs of the Chilean oyster are internally
fertilised, and the larvae, which are brooded within the
mantle cavitv. have limited dispersal capabilities because
of their extremely short pelagic stage (7). These features
could allow mechanisms such as inbreeding or Wahlwnl
effect to produce heterozygote deficiencies. However, we
observed no significant heteroiygote deficiencies in juve-
niles at 6 months of age; instead allozyme heterozygosiiy
decreased over time. Inbreeding. H 'ahlund effect, aneu-
ploidy. and null alleles are unlikely to be main causes of
the heterozygosity deficiency in this cohort; if they were,
the deficiency should be evident from the juvenile stage
and would not necessarily increase over time (2, 5. 8, 9.
10). We suggest that selection against heterozygotes is the
most probable cause of the increasing degree of hetero-
zvgote deficiency with age in this cohort ofO. chilensis, a
proposition that accords with data for other marine bivalve
species (2, 4. 1 1).

Populations of marine bivalves exhibit deficiency of
allozyme heterozygotes. This deficiency has been dem-
onstrated in laboratory studies of mussels and clams (2.
3), in studies using wild populations (8, 12) of Mytilus
edu/is. and in several studies of oysters (Crassoslrea vir-
ginica) (13-15). In the Chilean oyster (Ostrea chilensis) a

Received 10 May 1994; accepted 26 January 1995.

heterozygote deficiency was found in the carbonic an-
hydrase (C A) locus from a southern population (Melinka,
4353'S) ( 1 ). The time at which heterozygote deficiency
first appears in the population can help distinguish caus-
ative mechanisms ( 16). In laboratory studies with mussels,
an overall significant deficiency of heterozygotes was
found at the juvenile stage but not at the spat stage (4).

In the present study, we used a cohort of O. chilensis
settled on artificial collectors in the Quempillen hatchery,
Ancud, Chiloe (4552'S, 7346'W). The parental stock
was a cohort of O. chilensis collected during December
1987 from a natural spatfall in the wild population at
Quempillen estuary. The Chilean oyster becomes sexually
mature at the beginning of the second year of life with a
shell length of about 27 mm (7). After three years of
growth under uniform conditions, 800 randomly chosen
oysters were mass spawned in the laboratory. The tem-
perature and salinity used in the experiment were within
the range of those in the natural environment (10-18C
and 27-32 ppt). Although the use of mass spawning pre-
vents one from knowing how many individuals contribute
genes to the offspring obtained, the female contribution
can be estimated by keeping track of the number in each
brood of eyed larvae. Fecundity in O. chilensis ranges
between 10.000 and 1 15,000, with an average of 60,000
(7). The number of larvae released, more than 8.2 X 10 6 ,
indicates that at least 130 females contributed larvae. This
number of females may be an underestimation because
some of the eyed larvae released set within 5 min (7); thus
this cohort cannot be treated as a product of restricted
matings.

From an initial population size of 4050 randomly
tagged juveniles grown at the Hueihue location (4158'S,

117

18

J. E. TORO AND A. M. VERGARA

Table I

Helero:ygi>te Im/iu-iiey and D values tor lour loci at three stages of the
lite cvcle ot Ostrea chilensis tf>. IX. and 30 months of age)

Age

(months)

Locus

OH.

E.H.

D

OP)

6

LAP

0.410

0.398

0.053

NS

GPI

0.581

0.458

0.283

*

CA

0.645

0.617

0.046

NS

PGM

0.155

0.167

-0.072

NS

18

LAP

0.338

0.457

-0.260

*

GPI

0.373

0.389

-0.041

NS

CA

0.514

0.607

-0.153

*

PGM

0.247

0.383

-0.355

*

30

LAP

0.447

0.591

-0.243

*

GPI

0.268

0.292

-0.082

NS

CA

0.408

0.631

-0.353

#

PGM

0.231

0.374

-0.382

*

Genotype frequencies were investigated using random samples of 1 50
oysters taken from each class interval. Each locus was tested individually,
using the X 2 goodness of fit test with D as an index of heterozygote
deviation. Starch gel electrophoresis was used (18, 19) to score the loci
leucine aminopeptidase (LAP. EC 3.4. I.I), glucose phosphate isomerase
(GPI, EC 5.3.1.9), carbonic anhydrase (AC, EC 4.2.1.1), and phospho-
glucomutase (PGM, EC 2.5.7. 1 ).

O.H. = proportion of observed heterozygotes; E.H. = proportion of
expected heterozygotes; D = heterozygote deviation index denned as
(O.H. - E.H.l/E.H.; (P) = probability of the X 2 goodness of fit to the
Hardy Weinberg model (NS = nonsignificant; * = significant at alpha
= 0.05).

7330'W), the percentages of mortality at ages from 6 to
18 and 18 to 30 months were 25% and 17% respectively.
At each class interval, 1 50 oysters were sampled without
replacement. Neither significant deficiencies nor an excess
of heterozygotes was found in three of four loci in the 6-
month-old oysters; the exception was glucose phosphate
isomerase (GPI), which showed an excess of heterozygotes
(Table I). At 18 months, significant deficiencies of het-
erozygotes were found at LAP (D = -0.260), CA (D

-0.150), and PGM (D = -0.355) (Table I). In adult
oysters (30 months), negative values of D were present at
three of four analyzed loci, presenting significant values
at LAP (D = -0.243), AC (D = -0.353), and PGM (D

-0.382) (Table I). The data showed that between the
age of 6 and 18 months, three out of four loci studied
showed a genotype-dependent mortality. This differential
mortality produces a significant overall deficiency of het-
erozygosity in the cohort. One of the loci studied (GPI)
showed an excess of heterozygotes at 6 months and neither
an excess nor a deficiency of heterozygotes at 1 8 and 30
months; this result agrees with other studies carried out
on this locus in natural populations of bivalve molluscs
(I, 17, 18). High fecundity, external fertilization, and ex-
tensive larval dispersal characteristics common to most
of the bivalves molluscs make it unlikely that inbreeding

or the Wahlund effect could be the main cause of hetero-
zygote deficiencies. The reproductive features of O clii-
lensis favor mechanisms such as inbreeding or Wahlund
effect to act and produce heterozygote deficiencies. How-
ever, in accord with data for other marine bivalve species,
we suggest that selection against heterozygotes is the most
probable cause of the heterozygote deficiencies (2, 4, 1 1 ),
because the deficiency is not evident at the juvenile stage
but increases over time. We discarded inbreeding (which
would affect the whole genome), Wahlund effect, aneu-
ploidy, and null alleles as possible causes for the hetero-
zygosity deficiency in this cohort because a deficiency
produced by these factors should be evident from the ju-
venile stage and not necessarily increase over time (2, 5,
8,9, 10).

Acknowledgments

We thank Dr. J. B. Mitton and two anonymous re-
viewers for their valuable suggestions and advice for im-
proving our manuscript. This work was supported by the
Fondo Nacional de Desarrollo Cientifico y Tecnologico
(Fondecyt 91/0897) and by the Direccion de Investigation
y Desarrollo, U.A.Ch. (S-94-18).

Literature Cited

1. Guinea, R., and R. Gallegillos. 1985. Clinal variation in morpho-
logical distance between genotypes at the carbonic anhydrase locus
in the Chilean oyster, Tioslrea chilensis (Philippi. 1845) Chanley
and Dinamam. 1980. Bra: J Genet. 8: 609-616.

2 Mallet, A. L., E. /ouros, K. E. Gartner-Kepkay, K. R. Freeman,
and L. M. Dickie. 1985. Larval viability and heterozygote defi-
ciency in populations of marine bivalves: evidence from pair matings
of mussels. Mar. Biol 87: 165-172.

3. Gaffney, P. M., T. M. Scott, R. K. Koehn, and W. J. Diehl.
1990. Interrelationships of heterozygosity. growth rate and het-
erozygote deficiencies in the Coot Clam, Mu/inia lateralts. Genetics
124: 687-699.

4. Beaumont, A. R. 1991. Genetics studies of laboratory reared mus-
sels, Mylilus edulis heterozygote deficiencies, heterozygosity and
growth. Biol .1 Linn Soc. 44: 273-285.

5 Borsa, P., M. Zainuri, and B. Delay. 1991. Heterozygote deficiency
and population structure in the bivalve Ruditapcs ileciissatus
lleredtl\(tf>: 1-8.

6. Pogson, G. H., and E. Zouros. 1994. Allozyme and RFLP hetero-
zygosities as correlates of growth rate in the scallop Placopecten ma-
Kelliiiuenx. a test of the associative overdommance hypothesis. Ge-
netics 137: 221-231.

7. Toro, J. E., and O. R. Chaparro. 1990. Conocimiento hiologico
de Ostrea chilensis (Philippi, 1845): Impacto y Perspectives en el
Desarrollo de la ostricultura en Chile. Pp. 231-264 in Cultivo de
Moluscos en America Lalina. A. Hernadez, ed. Memonas segunda
reunion grupo trabajo tecnico, Ancud, Chile. Nov. 1989.

8. Gosling, E. M., and N. P. Wilkins. 1985. Genetics of settling co-
horts of Mvtilus edulis (L.): preliminary observations. Ai/iiacu//ure
44: I 15-123.

9. Gaffney, P. M. 1993. Heterosis and heterozygote deficiencies in
marine bivalves: more light? In Genetics and Evolution oj Aquatic
Organisms. A. R. Beaumont, ed. Chapman & Hall, London.

HETEROZYGOTE DEFICIENCY IN OYSTERS

19

10. Thriot-Quievreux, C, T. Noel, S. Bougrier, and S. Dalloi.

1988. Relationship between aneuploidy and growth rate in pair

matings of the oyster Crassostrea gigas. Ayuaculture 75: 89-91.
1 1 Alvarez, G., C. Zapala, R. Amaro, and A. Guerra. 1989. Multilocus

heterozygosity at protein loci and fitness in the European oyster.

Oslmi cilulis L. Heredity 63: 359-372.

12. Gosling, E. M., and D. McGrath. 1990. Genetic variability in ex-
posed-shore mussels, Mytilus spp. along an environmental gradient.
Mar. liii'l 104: 413-418.

13. Singh, S. M., and E. Zouros. 1978. Genetic variation associated
with growth rate in the American oyster (Crassostrea virginica},
Evolution 32: 342-353.

14. Zouros, E., S. M. Singh, D. W. Foltz, and A. L. Mallei. 1983. Post-
settlement viability in the American oyster (Crassostrea virginica):
an overdommant phenotype. Genet Res 41: 259-270.

15. Zouros, E., S. M. Singh, and II. E. Miles. 1980. An overdominant
phenotype and its possible explanations. Evolution 34: 856-867.

16. Zouros, E., and D. \V. Foltz. 1984. Minimal selection requirements
tor the correlation between heterozygosity and growth, and for the
deficiency of heterozygotes, in oyster populations. Dev. Genet 4:
393-405.

17. Koehn, R. W., R. Milkman, and J. B. Milton. 1976. Population
genetics of marine pelecypods. IV. Selection, migration and genetic
differentiation in the blue mussel Mylilus cdulis. Evolution 30: 2-
32.

18. Shaw, C. H., and R. Prasad. 1970. Starch gel electrophoresis of
enzymes: a compilation of recipies. Biochem. Genet- 4: 297-320.

19. Selander, R. K., M. II. Smith, S. Y. Yang, W. E. Johnson, and
J. B. Gentry. 1971. Biochemical polymorphism and systematics in
the genus Peromysenx 1. Variation in the oldfield mouse (Peromyseti.'i
polionotus). Stud. Genet. 6: 49-90.

Reference: Bi, >l. Bull 188: 120-127. (April. 1995)

Catch in the Primary Spines of the Sea Urchin

Eucidaris tribuloides: A Brief Review

and a New Interpretation

JOSE DEL CASTILLO*. DAVID S. SMITH**, ADA M. VIDAL*. AND CESAR SIERRA*

*lnstiliile of Newohiology, University of Puerto Rico M.S.C., Blvd. del 1'iille 201. Sun ./nan,

Puerto Rieo 00901; tint/ **Departnient of Zoology, University of Oxford,

South Parks Road. Oxford. OXl 3PS. United Kingdom

Abstract. Previous models of reversible catch in echi-
noid spines, as a property of muscle or of collagen, are
briefly reviewed and discussed. This brief review offers a
new interpretation of catch in primary spines of Eucidaris
tribuloides. viewing the collagen and small muscles of the
catch ligament working together as a variable-length ten-
don. In the model presented, changes in ligament length
when out of catch are accommodated by sliding of dis-
continuous, interdigitating and cross-link-stabilized col-
umns of collagen fibrils, the muscle layer external to the
ligament effecting spine movement. Catch is viewed as a
consequence of contraction of small muscles inserted on
the collagen columns within the ligament. Ligament
shortening tightens the profuse (en. 30,000/mrrr) and
highly ordered collagen insertion loops within the ster-
eoms of the spine base and test, and catch results from
the multiplicative effect of these friction sites in series.
New data are presented on novel structural cross-links
between collagen fibrils. The cross-links stabilize the liga-
ment columns. The central ligament in Eucidaris plays a
purely passive mechanical role in maintaining the align-
ment of the spine-test articulation. It contains no muscle
and neither contracts nor undergoes catch: its insertions
are simple, unlike the complex stereom insertions of the
main ligament.

Introduction

From the time that it was first recognized, the phenom-
enon of catch in sea urchin spines has attracted the interest
of investigators, but its basis has remained unclear. Two
seemingly contradictory theories have been proposed to

Received 5 January 1995; accepted 7 February 1995.

explain catch; but recent experimental observations allow
a new interpretation that reconciles the two discrepant
hypotheses.

Catch is an operational concept that can be defined in
this instance as a reversible, neurally controlled enhance-
ment of the passive mechanical resistance offered by the
spine test articulation (Fig. 1 ) to forces tending to change
the position of the spine. The sudden inducement of catch
freezes the primary spines in their respective positions,
whether normal to the test surface or angled from this
axis, thereby allowing the animal to maintain a fixed pos-
ture for long periods.

von Ui'.\kiil/'s catch muscle

At the turn of the last century. Count Jakob von
Uexkiill, a self-supporting German biologist noted for his
strong vitalist convictions, published a paper ( 1 900) titled
"The Physiology of the Sea-urchin Spine" in which he
reported that the voluntary and reflex movements of the
spine are powered by a thin layer of muscle fibers that
surrounds the thick articular capsule. In addition, he
found that the integrity of this capsule, which is also
known as the spine ligament or catch apparatus, is essen-
tial for the development of catch.

Von Uexkull described the breakage of the capsule by
forcible displacement of the spine while in catch: spines
treated in this manner failed to show catch, but they re-
tained the ability to perform voluntary and reflex move-
ments because the thin muscle layer was not disrupted.
Accordingly, von Uexkull called this muscle layer Be-
wegungsmuskulatur (motion-supporting muscle) as op-
posed to the articular capsule, which he believed also to
be a muscle, the Sperrmuskulatur (catch or holding mus-
cle). As we shall see below, this was an inspired guess that

CATCH IN SEA URCHIN SPINES

121

Figure 1. The spine-test articulation of Eucidaris. In this Chlorox-
digested preparation a small area ofligament remains, maintaining the
ball-and-socket arrangement. 18

defied contemporary evidence, because the latter tissue
had been studied by 19th century microscopists (Prouho.
1887; Hamann, 1887) and was recognized by them as
being primarily a connective tissue.

Takahaslu's mutable connective tissue

The problem of catch in echinoderm spines was studied
again in the 1960s by Takahashi (1966, 1967a, b, c), who
confirmed von Uexkiill's results while disagreeing with
him on the nature of the ligament. In the discussion of
his landmark paper (1967b) on "Responses to stimuli,"
Takahashi gave an account of the experimental results
that led him to propose a new hypothesis to explain catch.

Because at that time the ligament was still regarded as
a muscle, Takahashi first attempted to record its contrac-
tion following the application of chemical or electrical
stimuli. He was not successful. Yet Takahashi was greatly
impressed by the effects of the same chemical stimuli on
the rate of elongation of ligaments subjected to a constant
load (isotonic recording; creep test). In his words "the
effects were