Volume 200 THE Number 1 BIOLOGICAL BULLETIN Aerobic and Anaerobic Circular Muscles Anaerobic Circular Muscle „./? 1.00 1.25 1.50 1.75 Time (s) Time (s) FEBRUARY 2001 Published by the Marine Biological Laboratory Thanks Galileo, Hooke and Leeuwenhoek for the incentive. We walk in the footsteps of visionaries. And see with the eyes of disciples. ; Galileo up close: astronomer, physicist, j mathematician, dethroner of Aristotle, defender of Copernicus and, yes, credited with making „ ,*~. double convex lenses and the first biological observations with a microscope. A compound microscope. In 1624. Hooke: 43 years later, the compound microscope reveals living things are composed of cells. Leeuwenhoek: invents the first practical microscope and discovers bacteria. Armed with the knowledge and inspiration of these and other Olympians, secure in the history of our own accomplishments and aware that the mind's eye should always focus on challenging the enigmatic as well as the undeniable, Olympus has created the BX2 series of clinical and research microscopes, noteworthy for their enhancements in optics, imaging and ergonomics. And each so individual, it deserves the nomenclature My Microscope. From possibility to actuality, it is wonderful to behold. A posse ad esse est mirabile visu. The new standard in laboratory . microscopy vis-a-vis " performance, ver- satility, operation. Uncanny digitized imaging: enhanced with the DP11 digital camera, delivering high-resolution image recording beyond the norm. Unusually insightful optics: observe magnifications from 1.25x to lOOx without changing the condenser. Choice of condensers includes Abbe, swing- out, phase-contrast and darkfieH Unerringly precise- motion stage: selectable for right/ left hand. Discover speed, ease, touch, comfort, and clinical accuracy. Uncommonly easy-to-operate controls: extended fine-focus knob; l-positioned illumination intensity dial; tactile X-Y control. Welcqrne to the next golden age icroscope Series The new standard in ergonomic microscopy. A dramatic metamor- phosis: innova- •&£$ tive, intelligent, insightful design improvements specifically addressing the needs of those spending hours looking through a microscope. The lowest low-position stage: three inches lower than standard microscopes, offering the promise of minimum effort and hand movement when changing specimens. The tilting, telescoping observation tube: a marvel of adjustability, it can be changed to accommodate any observation position. 80mm higher •"id 70mm oser to the operator than a convention- al tilting tube. And the eye- i be moved forward by 45mm .-•lined from 0-25 degrees. The new start- s* "•"•a-i it Ht ™ dard in research microscopy. The words "new" and "standard" really don't do it justice. How about "ingenious" or "original," "paragon" or "prototypical." Prepare yourself for a higher power. The 6-position filter cube turret: Not 4, not 5, but 6, so single and multi- band imaging of new fluorochromes and fluorescent proteins is faster, simpler and exclusively tailored. The rectangular field s ours alone. By cleverly matching the camera's field of view, only the area requiring fluorescence excita- tion is exposed. Which means that the surrounding areas are protected from photo bleaching. The fluorescence excitation bal- ancers: another Olympus exclusive. A continuously variable excitation bandwidth. Visualization is enhanced, differentiation of multi-labels is immeasurably improved, capturing of images is at an unheard-of level. Triple Labeled Specimen (1 to r): with prevalent FITC and Cy3 with prevalent DAPI and Cy3 with balanced fluorescence emission •ror aspherical lens herical collector lens: sui generic, in a class by itself. Specimens appear brighter and more evenly illuminated; and achromatic performance now incorporates wavelengths from UV to IR. . The universal -8%^- condenser: -with 8 positions, it can accept optical inserts for DIG, phase, brightfield and dark-, field illumination: And the advanced Nomarski DIG system has been expanded and optimized to encompass prisms for enhanced contrast and image resolution. OLYMPUS FOCUS ON LIFE Visit us at www.olympusamcrica.com or call 1-800-455-8236 © Olympus America Inc. THE BIOLOGICAL BULLETIN Editor Associate Editors Section Editor Online Editors Editorial Board Editorial Office FEBRUARY 2001 MICHAEL J. GREENBERG Louis E. BURNETT R. ANDREW CAMERON CHARLES D. DERBY MICHAEL LABARBERA SHINYA INOUE, Imaging and Microscopy JAMES A. BLAKE, Keys to Marine Invertebrates of the Woods Hole Region WILLIAM D. COHEN, Marine Models Electronic Record and Compendia PETER B. ARMSTRONG ERNEST S. CHANG THOMAS H. DIETZ RICHARD B. EMLET DAVID EPEL GREGORY HINKLE MAKOTO KOBAYASHI ESTHER M. LEISE DONAL T. MANAHAN MARGARET MCFALL-NGAI MARK W. MILLER TATSUO MOTOKAWA YOSHITAKA NAGAHAMA SHERRY D. PAINTER J. HERBERT WAITE RICHARD K. ZIMMER PAMELA CLAPP HINKLE VICTORIA R. GIBSON CAROL SCHACHINGER CAROL MARRAMA FEB ^ 6 2001 The Whitney Laboratory. University of Florida Grice Marine Biological Laboratory. College of Charleston California Institute of Technology Georgia State University University of Chicago Marine Biological Laboratory ENSR Marine & Coastal Center. Woods Hole Hunter College, City University of New York University of California, Davis Bodega Marine Lab., University of California, Davis Louisiana State University Oregon Institute of Marine Biology, Univ. of Oregon Hopkins Marine Station, Stanford University Cereon Genomics, Cambridge, Massachusetts Hiroshima University of Economics, Japan University of North Carolina Greensboro University of Southern California Kewalo Marine Laboratory, University of Hawaii Institute of Neurobiology, University of Puerto Rico Tokyo Institute of Technology, Japan National Institute for Basic Biology, Japan Marine Biomed. Inst., Univ. of Texas Medical Branch University of California, Santa Barbara University of California, Los Angeles Managing Editor Staff Editor Editorial Associate Subscription & Advertising Secretary Published by MARINE BIOLOGICAL LABORATORY WOODS HOLE, MASSACHUSETTS Cover The brief squid, Lolliguncula brevis, is a small cephalopod mollusc found in shallow coastal waters from Delaware Bay south to Texas and Brazil. It is the only species of cephalopod frequently caught in estuaries, and is noteworthy for its ability to tolerate euryhaline waters. The brief squid on the cover is swimming by jet propulsion in a water tunnel at intermediate speeds (~15 cm s"1); the image is from a single video frame. The circular muscles of the mantle of squids and cuttlefishes are responsible for body contractions during jetting and consist of three anatomically and metabolically distinct layers. The inner and outer (peripheral) layers have structural and biochemical features characteristic of aerobic muscle, whereas the middle (central) layer has characteristics of an- aerobic muscle. This has led to the hypothesis that the peripheral muscle layers - analogous to the red muscle in fish - are employed in slow, steady swim- ming and in rhythmic, respiratory contractions, whereas the central layer is analogous to the white muscle in fish and is used during burst swimming, escape, and in the capture of prey. I.K. Bartol (see p. 59 in this issue) has now tested these predictions about the functional roles of the circular muscle layers of the mantle in swimming squid. Bartol captured brief squid in the Chesapeake Bay and fitted them with paired electrodes embed- ded either in both the peripheral aerobic and central anaerobic circular muscles or only in the latter. Electromyograms (EMGs) were recorded from the electrodes, while video recordings were made of the squid swimming against a range of steady currents in a water tunnel. The waveforms on the left side of the cover are EMGs recorded over five jet cycles from both muscle layers, whereas the EMGs on the right are recorded over four jet cycles from the central anaerobic circular muscle. In both cases, the steady current was 15 cm s~'. Although central anaerobic circular muscle activity was first detected at these intermediate swimming speeds, it does not occur with each contraction of the mantle at such speeds. Bartol presents evidence that the peripheral aerobic and the central anaerobic circular muscle fibers in squid have distinct functional roles during swim- ming that are indeed similar to those of red and white myotomal fibers in fishes. Unlike fishes, how- ever, squids appear to use anaerobic muscle at sub- critical speeds. This discovery of functional "gears" in squids - yet another instance of convergent evo- lution in fishes and cephalopods - suggests that the locomotive muscle in squids is specialized and ef- ficiently used. CONTENTS VOLUME 200, No. 1: FEBRUARY 2001 BIOMECHANICS Etnier, Shelley Flexural and torsional stiffness in multi-jointed bio- logical beams Nauen. Jennifer C., and George V. Lauder Three-dimensional analysis of finlet kinematics in the chub mackerel (Scomber japmicus) DEVELOPMENT AND REPRODUCTION Amano, Shigetoyo, and Isao Hori Metamorphosis of coeloblastula performed by multi- potential larval flagellated cells in the calcareous sponge Leucosolenia laxa 20 ECOLOGY AND EVOLUTION McEdward, L. R., and K. H. Morgan Interspecific relationships between egg size and the level of parental investment per offspring in echino- derms. . 33 Kinzie, Robert A., Ill, Michelle Takayama, Scott R. Santos, and Mary Alice Coffroth The adaptive bleaching hypothesis: experimental tests of critical assumptions 51 PHYSIOLOGY Bartol, Ian K. Role of aerobic and anaerobic circular mantle mus- cle fibers in swimming squid: electromyography ... 59 Zielinski, S., F. J. Sartoris, and H. O. Portner Temperature effects on hemocyanin oxygen binding in an Antarctic cephalopod 67 Medler, Scott, and Harold Silverman Muscular alteration of gill geometry in vitro: implica- tions for bivalve pumping processes 77 RESEARCH NOTES Browne, Kenneth A., and Richard K. Zimmer Controlled field release of a waterborne chemical signal stimulates planktonic larvae to settle 87 Dawson, Michael N., and David K. Jacobs Molecular evidence for cryptic species of Amelia au- nta (C.nidaria, Scyphozoa) 92 THE BIOLOGICAL BULLETIN THE BIOLOGICAL BULLETIN is published six times a year by the Marine Biological Laboratory. 7 MBL Street. Woods Hole, Massachusetts 02543. Subscriptions and similar matter should be addressed to Subscription Manager. THE BIOLOGICAL BULLETIN. Marine Biological Laboratory, 7 MBL Street, Woods Hole, Massachusetts 02543. Subscription per year (six issues, two volumes): $235 for libraries; $95 for individuals. Subscription per volume (three issues): $1 17.50 for libraries; $47.50 for individuals. Back and single issues (subject to availability): $40 for libraries; $20 for individuals. Communications relative to manuscripts should be sent to Michael J. Greenberg, Editor-in-Chief, or Pamela Clapp Hinkle. Managing Editor, at the Marine Biological Laboratory. 7 MBL Street. Woods Hole. Massachusetts 02543. Telephone: (508) 289-7428. FAX: 508-289-7922. E-mail: pclapp@mbl.edu. http://www.mbl.edu/BiologicalBulletin/ The home page for the electronic companion to THE BIOLOGICAL BULLETIN — the Marine Models Electronic Record — and other BIOLOGICAL BULLETIN publications is available on the World Wide Web at the address shown above. THE BIOLOGICAL BULLETIN is indexed in bibliographic services including Index Medicus and MEDLINE, Chemical Abstracts, Current Contents, Elsevier BlOBASE/Current Awareness in Biological Sciences, and Ge o Abstracts. Printed on acid free paper, effective with Volume 180. Issue 1. 1991. POSTMASTER: Send address changes to THE BIOLOGICAL BULLETIN. Marine Biological Laboratory, 7 MBL Street, Woods Hole. MA 02543. Copyright © 2001. by the Marine Biological Laboratory Periodicals 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 accepted 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 published in a separate section entitled "Research Notes." A Research Note in The Biological Bulletin follows the format of similar notes in Nature. It should open with a summary paragraph 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 Research Notes section should conform in punctuation and arrangement to the style of recent issues of The Biological Bulletin. Materials and Methods should be incorpo- rated into appropriate figure legends. See the article by Loh- mann et 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 quadruplicate, with the originals clearly marked. (Xerox copies of photographs are not acceptable for review pur- poses.) The submission letter accompanying the manuscript should include a telephone number, a FAX number, and (if possible) an E-mail address for the corresponding author. The original manu- script must be typed in no smaller than 12 pitch or 10 point, using double spacing (including figure legends, footnotes, bibliography, etc.) on one side of 16- or 20-lb. bond paper, 8 by 1 1 inches. Please, no right justification. Manuscripts should be proofread carefully and errors 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 Editors 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 contribution num- bers, 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 publication. 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 generally should be included in legends, although axes should always be identified on the illustra- tion itself. Figures should be prepared for reproduction as either line cuts or halftones. Figures to be reproduced as line cuts should be unmounted glossy photographic 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 and cited, in order, in the text. The author's name and an arrow indicating orientation should appear on the reverse side of all figures. Digital art: The Biological Bulletin will accept figures sub- mitted in electronic form; however, digital art must conform to the following guidelines. Authors who create digital images are wholly responsible for the quality of their material, including color and halftone accuracy. Format. Acceptable graphic formats are TIFF and EPS. Color submissions must be in EPS format, saved in CMKY mode. Software. Preferred software is Adobe Illustrator or Adobe Photoshop for the Mac and Adobe Photoshop for Windows. Spe- cific instructions for artwork created with various software pro- grams are available on the Web at the Digital Art Information Site maintained by Cadmus Professional Communications at hup:// cjs. cadmus.com/da/home. html Resolution. The minimum requirements for resolution are 1200 DPI for line art and 300 for halftones. Size. All digital artwork must be submitted at its actual printed size so that no scaling is necessary. Multipanel figures. Figures consisting of individual parts (e.g., panels A, B, C) must be assembled into final format and submitted as one file. Hard copy. Files must be accompanied by hard copy for use in case the electronic version is unusable. Disk identification. Disks must be clearly labeled with the following information: author name and manuscript number; for- mat (PC or Macintosh); name and version of software used. Color: The Biological Bulletin will publish color figures and plates, but must bill authors for the actual additional cost of printing in color. The process is expensive, so authors with more than one color image should — consistent with editorial concerns, especially citation of figures in order — combine them into a single plate to reduce the expense. On request, when supplied with a copy of a color illustration, the editorial staff will provide a pre-publi- cation estimate of the printing cost. 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, num- bered with consecutive Roman numerals, and placed after the Literature Cited. Figure legends should contain enough informa- tion 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, acknowledgments or contri- bution 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 al. 1980. Personal communications and material in prep- aration 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 Literature 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 include complete titles and inclusive pagination. Journal abbreviations 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 bio- logical 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 PERIODICALS, 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 addi- tional 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 J. cell. comp. Physiol. ) C. All abbreviated components must be followed by a period, whole word components must not (i.e.. J. Cancer Res.) D. Space between all components (e.g., J. Cell. Comp. Plivsiol.. not J.CelLComp.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 Visindafjelags Is/endinga without 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 6. Reprints, page proofs, and charges. Authors may pur- as the word order of the complete title (i.e., Proc. and Trims. chase reprints in lots of 100. Forms for placing reprint orders are placed where they appear, not transposed as in some BIOLOGICAL sent with page proofs. Reprints normally will be delivered about 2 ABSTRACTS listings). to 3 months after the issue date. Authors (or delegates for foreign authors) will receive page proofs of articles shortly before publi- H. A tew well-known international ournals in their preferred . ^, cation. They will be charged the current cost ot printers time tor forms rather than WORLD LIST or USASI usage (e.e.. Nature. , , .- ,. corrections to these (other than corrections ot printers or editors Science, Evolution NOT Nature, Lond.. Science. N.Y.: Evolution. mQK) Q^ than these charg£s fof authors- alterations_ The Lancaster. Pa. ) Biological Bulletin does not have page charges. Reference: Bini Hull. 200: 1-8. (February 2001) Flexural and Torsional Stiffness in Multi-Jointed Biological Beams SHELLEY A. ETNIER* Department of Biology, Duke University. Durham, North Carolina 27708-0338 Abstract. Flexibility, the ability to deform in response to loads, is a common property of biological beams. This paper investigates the mechanical behavior of multi-jointed beams, which are characterized by a linear series of mor- phologically similar joints. Flexural stiffness and torsional stiffness were measured in two structurally distinct beams, crinoid arms (Echinodermata, Comatulida) and crustacean antennae ( Arthropoda, Decapoda). Morphological data from these beams were used to determine the relative contribu- tions of beam diameter and joint density (number of joints per millimeter of beam length) to the flexural and torsional stiffness of these two structures. As predicted by beam theory, beam diameter influenced stiffness in both crinoid arms and crustacean antennae. In crinoid arms, increases in joint density were associated with decreases in stiffness, but joint density had no significant influence on stiffness in crustacean antennae. In both crinoid arms and crustacean antennae, the magnitudes of flexural and torsional stiffness, as well as the ratio of these two variables, were similar to previously reported values for non-jointed biological beams. These results suggest that the structural design of a biological beam is not a limiting factor determining its mechanical properties. Introduction Biological beams, which are structures that are long rel- ative to their width, can be divided into three broad groups based on their structural design. Continuous beams have no distinct discontinuities in either material or geometry along the length of the beam. These continuous beams, such as leaf petioles and tree trunks, deform relatively evenly along Received 1 3 July 2000; accepted 1 3 December 200(1. * Present address: Department of Biological Sciences. University of North Carolina, 601 S. College Rd.. Wilmington. NC 28403. E-mail: etniers@uncuil.edu their entire length when loaded. Continuous beams contrast with jointed beams, which are characterized by one or a few material or structural discontinuities along their length which allow the beam to deform at particular points. Ex- amples include vertebrate limbs and arthropod appendages. The third structural design is a multi-jointed beam, which is characterized by a linear series of morphologically similar joints connected by a series of stiff elements. The vertebrate backbone is the most extensively studied example of a multi-jointed beam (see Gal, 1993; Long et al.. 1997); other examples are found in echinoderms (Baumiller and LaBar- bera. 1993). cnidarians (Muzik and Wainwright. 1977). and plants (Niklas. 1997). When loaded, a multi-jointed beam will deform at each of the joints. Yet, because the joints are arrayed in a linear series, the deformation is evenly distributed along the length of the beam. Thus, the overall deformation of the beam resembles that of a continuous beam, although the underly- ing mechanism (deformation at individual joints) differs dramatically. Ideally, studies of the mechanical properties of a multi-jointed beam should include information on the overall flexibility of the beam as well as on the relative contribution of the individual joints. Regardless of their structural design, most biological beams are flexible; that is. they deform in response to loads (Vogel, 1984; Denny, 1988). One dictionary (New Lexicon Webster's Dictionary, 1987) defines flexibility as the qual- ity of being easily bent, and rigidity as the quality of resisting deformation. These iwo contrasting states actually reflect a continuum that can be described with a single, more precise term borrowed from engineering beam theory. Stiff- ness is defined as a mechanical property indicating the resistance of a material or structure to deformation under a given load (Roark, 1943). Flexural stiffness is a measure of the resistance of a structure to bending, and torsional stiff- ness is a measure of the resistance of a structure to twisting (Roark. 1943). S. A. ETNIER Flexural stiffness and torsional stiffness are composite variables whose magnitudes are determined both by mate- rial and structural properties (Wainwright et ai, 1976). Flexural stiffness is expressed as El. where E is Young's modulus of elasticity and / is the second moment of area relative to the neutral plane of bending (i.e., the plane that neither shortens nor lengthens during deformation). Tor- sional stiffness is expressed as GJ. where G is the shear modulus and J is the polar moment of area about the central axis of rotation. / and J reflect the cross-sectional geometry of a beam and are influenced by both size and shape. Size is a very strong determinant of beam stiffness (Roark, 1943: Wainwright et til, 1976), as both / and J are proportional to radius to the fourth power. Flexural stiffness and torsional stiffness are engineering terms used to describe the stiffness of ideal beams that meet a certain set of criteria (Roark, 1943), none of which are met by either continuous or multi-jointed biological beams (Wainwright et ai, 1976). For biological beams, flexural stiffness and torsional stiffness are descriptors of the overall mechanical behavior of the beam, reflecting the amount of deformation observed under a given load. As such, the presence of joints does not negatively impact the overall usefulness of these terms as a means to quantify the stiffness of biological beams. The stiffness of biological beams can also be character- ized by the dimensionless. and thus size invariant, ratio EIIGJ, called the twist-to-bend ratio (Niklas, 1992; Vogel, 1995; Etnier and Vogel. 2000). The twist-to-bend ratio indicates the relative resistance of a beam to bending versus twisting, without reference to the absolute magnitude of either. Intuitively, a higher twist-to-bend ratio indicates a structure that twists more readily than it bends. Previous studies have found that the twist-to-bend ratio is a biolog- ically relevant parameter that provides insight into the func- tional demands on a structure. For example, daffodil stems (Etnier and Vogel. 2000). leaf petioles (Vogel. 1992). and sedges (Ennos, 1993) are characterized by relatively high twist-to-bend ratios. These structures easily twist into low- drag configurations when the wind blows, thus reducing the potential for damage to the organism. In this paper, values for flexural and torsional stiffness, as well as the twist-to-bend ratio, are reported for crinoid arms (Echinodermata, Comatulida) and crustacean antennae (Ar- thropoda, Decapoda). Mechanical and morphological data from these beams are used to determine the relative contri- butions of beam diameter and the number of joints to the flexural and torsional stiffness of these two structures. This broadly comparative study of morphologically distinct structures provides insight into the contribution of a serial arrangement of joints to the mechanical properties of multi- jointed beams. Comatulid crinoid have long, slender arms composed of a linear series of ossicles connected by muscles and liga- ments (Breimer, 1978). The arms move actively in response to muscular contractions (Breimer, 1978) or passively in response to external forces. Crinoids extend their arms into the currents to filter feed passively (Liddell. 1982). thus the arms must be flexible enough to be positioned for feeding, yet stiff enough to maintain that position once obtained (Meyer. 1971). The flexibility of the arms is also relevant during locomotion (Motokawa, 1988), because the arms are bent during swimming and crawling (Breimer, 1978; Shaw and Fontaine. 1990). The dependence of feeding and loco- motion on the mechanical properties of the arms suggests that these properties are important to the survival and re- productive success of comatulid crinoids. The only experi- mental studies of arm mechanics have explored the behavior of isolated joints in the arm (Birenheide and Motokawa, 1994, 1996) without considering how the serial arrangement of the joints affects the mechanics of this multi-jointed beam. The second antenna of a lobster or crayfish consists of a series of calcified rings, or annuli, connected by flexible arthrodial membranes (Tautz et ai, 1981). There are no muscles associated with the annuli of the antennae, although the entire structure can be moved by muscles located at its base (Sandeman, 1985). Hair receptors on each annulus are sensitive to gross water movements, water vibrations, and mechanical deformation of the antenna (Vedel, 1985). In palinurid lobsters, the antennae are not only used as sense organs, but are also actively used as a defensive mechanism to ward off predators (Atema and Cobb, 1980). These two functions may have conflicting mechanical demands in lob- sters, because the antenna must be flexible enough to de- form in response to water vibrations, but also stiff enough to serve as a defensive weapon. Again, very little is known about how the multi-jointed nature of these beams affects their mechanical properties. Materials and Methods Specimens Crinoids. Specimens of Comactinia echinoptera were obtained from the invertebrate collection at Duke University Marine Laboratory, Beaufort, North Carolina, and freshly preserved specimens of Florometra serratissima were ob- tained from Sea Life Supply in Sand City. California. Both species were initially preserved in formalin, and then stored in 107c alcohol. Prior to testing, arms were disarticulated proximally and then transferred through a series of increas- ingly dilute alcohol solutions, culminating in seawater. CnisUicciins. Specimens of Procambarus sp. (freshwater crayfish) were obtained commercially from a local grocer (Wellspring Grocery), and specimens of Pamdirus argus (spiny lobster) were obtained from investigators at the Uni- versity of North Carolina, Chapel Hill. Antennae were dis- articulated from the carapace and frozen until tested. The FLEXIBILITY IN MULTI-JOINTED BEAMS A. B. String Beam length L String \ Pin LVDT LVDT Applied Load Applied Load String Figure 1. Mechanical apparatus for measuring flexural and torsional stiffness. (Al Flexural stiffness is measured by applying a load to the specimen's free end, causing it to bend. The two pulleys on the left are coaxial. Shown below is a cross section through the free end of the embedding cap, illustrating how the load is applied during flexural tests. (B) Torsional stiffness is measured by applying a load to the specimen's free end, causing it to twist. Shown below is a cross section through the free end of the embedding cap, illustrating how the load is applied during torsional tests. The load is applied simultaneously to the two ends of pin extending through the embedding cap, causing the specimen to twist, but not bend. L = Beam length, d = moment arm, LVDT = linear variable differential transformer. (Based on Vogel, 1992.) antennae were thawed in seawater or fresh water, as appro- priate, before testing. Mechanical measures For both flexural and torsional stiffness, the free ends of the specimen were embedded in hard, plastic caps (diameter approximately 11 mm. length 15 mm) with 5-minute epoxy resin. As the epoxy cured, a pin was inserted in the distal cap. The proximal end of the pre- pared specimen was fixed in place. A load was applied to the pin at the free distal end, causing the specimen to bend 01 twist (Fig. 1). Through a series of pulleys and strings, the deformation of the specimen was linked to movement of a linear variable differential transformer (LVDT. Pickering 7308-W2-AO). winch provided a volt- age output proportional to distance traveled (see Vogel, 1992, for full details). The LVD 1 was calibrated with a micrometer, and each readii r \Vas corrected for load- dependent stretching of tru- suing (Berkley Gorilla Braid 10-lb-test fishing line made of braided gelspun polyeth- ylene fibers). All readings were taken on moist specimens 60 s after loading, to allow for initial creep. Specimens were tested with two loads of different magnitude, and data for deformation relative to the load were averaged. This experimental protocol assumes that flexural and torsional stiffness are linear functions of load. S. A. ETNIER Flexural stiffness (in N • nr) was calculated with the usual formula for end-loaded cantilever beams: Flexural stiffness = (i) where F is the force applied, L is the length of the beam, and v is the deflection at the free end of the beam (Fig. 1A). The formula is satisfactory for deflections up to about 10% of total specimen length (Gere and Thnoshenko, 1984). Crinoid arms and crustacean antennae are radially asym- metric, so flexural stiffness was measured both dorsoven- trally (oraL/aboral) and laterally. These values were then averaged to provide an overall measure of the flexural stiffness of the system. Torsional stiffness (in N • nr) was determined by apply- ing a load to the distal end of the beam, causing the specimen to twist (Fig. IB). Calculations of torsional stiff- ness were based on the following formula: Torsional stiffness = Fd (2) where F is the force applied at a moment arm d, L is the length of the beam, and 9 is the resulting rotation in radians. A test of the apparatus using a piece of spring steel (0.65 mm in diameter) gave an average EI/GJ value of 1.4 (standard deviation 0.3), close to the expected value of 1.3 (Gere and Timoshenko. 1984). This error indicates a sys- tematic overestimate of about 8% in the ratios given here. In some cases, the applied loads were so small that they did not overcome the frictional resistance of the pulleys in the test apparatus. For measures of flexural stiffness, data were omitted if ( 1 ) the total deflection was less than 1% of the total specimen length, and (2) the load applied over the moment arm of the pulley was less than or equal to 4.9 X 10"" N (0.5 g). For torsional measures, data were omitted if (1) the total rotation was less than 0.06 radians (3.6°), and (2) the load applied over the moment arm of the pulley was less than or equal to 9.8 X 10~3 N ( 1 g). Morphological measures Immediately after the mechanical tests, beam length (L) and moment arm (d) were measured using digital calipers (resolution 0.1 mm). The beam diameter was recorded as the average diameter of the specimen at the midpoint of the beam. The number of joints in each specimen was counted to calculate joint density (number of joints per millimeter of beam length). Data analysis The . i'iage twist-to-bend ratio for each species was calculate :Vom the twist-to-bend ratio for each individual. Distributions and variances were normalized using log- transformed values for flexural stiffness, torsional stiffness, the twist-to-bend ratio, and beam diameter (Sokal and Rohlf, 1981 ). Joint densities were unaltered because of their normal distribution. Student's t tests were used to determine whether there were significant differences in these variables within the crinoids and within the Crustacea. Differences between phyla were not tested because distinct preservation methods were used for crinoids and crustaceans. Morphological and mechanical data from the two species of crinoids and the two species of crustaceans were com- bined to provide information on the general structural de- sign of each system. Although there are assuredly mechan- ical and morphological differences within each phylum, these are assumed to be relatively minor compared to the differences between the two. Multiple regressions were used to determine the relative contribution of beam diameter and joint density to flexural and torsional stiffness (Sokal and Rohlf. 1981). Multiple regressions determine whether a linear combination of independent variables explain a sig- nificant portion of the variability observed in the original data (Sokal and Rohlf, 1981 ). Additionally, multiple regres- sions indicate which of the independent variables contribute significantly to the overall model. Because of the unequal variances of these data, differ- ences in joint density between crinoid arms and crustacean antennae were examined using a Wilcoxon rank sum test for nonparametric data (Glantz, 1992). Microsoft Excel 5.0 ( 1995) was used for all data manip- ulation, and JMP In (SAS Institute. Inc., Gary, NC) was used for all statistical analyses. Results Mechanical measures Flexural stiffness and torsional stiffness varied greatly from specimen to specimen, as expected based on differ- ences in beam diameter between individuals (Table 1). The coefficients of variation for flexural stiffness ranged from 84% to 196%, whereas those for torsional stiffness ranged from 61% to 1 10%. The twist-to-bend ratio was also highly variable (Table 1). with coefficients of variation ranging from 41% to 88%. The variability of the twist-to-bend ratio far exceeded that of the test apparatus, suggesting that the variability was real. The magnitudes of flexural stiffness and torsional stiffness varied among all of the species. Procam- barus was the most flexible beam in both bending and twisting, and Panulirus was by far the stiffest in both measures. The two crinoids fall in the middle, with Flo- romelra being stiffer than Comactinia. This pattern was not observed in the twist-to-bend ratio, where Panulirus had the lowest ratio and Florometra had the highest. The values for flexural stiffness, torsional stiffness, and twist-to-bend ratio are comparable to those of other continuous biological beams (Table 2), such as leaf petioles and plant stems. FLEXIBILITY IN MULTI-JOINTED BEAMS Table I Morphological and mechanical values for multi-jointed heamx Flexural stiffness Torsional stiffness Twist-to-bend Species n Diameter (mm) (N • nr - 10"") (N • nr x 10"6) ratio Comactinia echinoptera 10 1.33(0.13) 50(47) 15(11) 4.3 (3.8) Florometra serratissima 6 1.75(0.10) 392(383) 51 (31) 6.6(2.7) Procambarus sp. 8 0.99 (0.07) 25(21) 6(4) 4.5(3.7) Panulirus argus 13 2.83(0.36) 5827(11466) 2036(22421 1.8(1.3) n = 12 H = 12 Mean values for beam diameter, flexural stiffness, and torsionul stiffness for n specimens. In one case, as noted, the number of specimens varied. The mean tvvist-to-bend ratio is the average calculated from the individual twist-to-bend ratios. The standard deviation for each parameter is given in parentheses. Mechanical differences benveen crinoid species The arms of Florometra had significantly higher flexural (/ = -3.88. P < 0.002) and torsional (t = -3.40. P < 0.004) stiffness than those of Comactinia (Table 1 ). These differences may be associated with differences in the size of the two species, because both flexural and torsional stiffness are highly size dependent. Florometra specimens had sig- nificantly thicker arms (Table 1 ) than Comactinia speci- mens (t = -2.21. P < 0.04). Despite the differences in the magnitudes of flexural and torsional stiffness, the twist- to-bend ratios of these two species were not significantly different U = - 1.70. P < 0.1 12). Mechanical differences benreen crustacean species Panulirus antennae were significantly stiffer in flexural (t = -5.63. P < 0.0001 ) and torsional (t = -7.99, P < 0.0001) stiffness than Procambarus antennae (Table 1). Again, these differences may be associated with size, be- cause Panulirus antennae were significantly larger in diam- eter (Table 1) than Procambarus antennae (t = —5.05, P < 0.0001). The twist-to-bend ratio of Panulirus was significantly higher than that of Procambarus (t = 2.43, P < 0.026). As in the crinoid arms, the magnitudes of Table 2 Mechanical properties of biological beams Flexural stiffness Torsional stiffness Twist-to-bend Biological beam (N • nr x 10~6) (N • nr x 10~6) ratio Daffodils' 11900 890 13.3 Sweet gum petiole2 984 199 5.1 Sunflower shoot2 73 53 1.4 Mean flexural stiffness, torsional stiffness, and twist-to-bend ratios for a variety of continuous biological beams. 1 Data from Etnier and Vogel (2000). 2 Data from Vogel (1992). flexural and torsional stiffness varied between species, but in this case, the twist-to-bend ratios also differed. Differences between structural designs Morphology. Joint density differed significantly (Z - 3. 28, P < 0.001) between the crinoid arms and crustacean antennae. Crustacean antennae had an average joint density of 2.04 mm"' (SD = 1.18), whereas crinoid arms had an average joint density of 0.82 mm"' (SD = 0.17). These results correspond to about 200 joints in a 10-cm-long crustacean antenna, but only about 80 joints in a crinoid arm of the same length. Mechanics, In crinoids. the multiple regression model accounted for a significant portion of the observed variation in both flexural (R2 = 0.87) and torsional (R2 = 0.74) stiffness. Both joint density and diameter contributed sig- nificantly to the overall model. Increases in diameter were associated with increases in stiffness, and increases in joint density were associated with decreases in stiffness (Table 3). In crustacean antennae, the model also explained a significant portion of the observed variation in flexural (R2 = 0.92) and torsional (R2 = 0.91 ) stiffness, although in this case, only beam diameter contributed significantly to the overall model (Table 3). Joint density did not contribute significantly to either flexural or torsional stiffness in the crustacean antennae (Table 3). In summary, beam diameter was an important determinant of stiffness in both multi- jointed beams, as expected for a continuous beam. Joint density, which was expected to r.UX- ihe mechanical prop- erties of multi-jointed beams, w, s a determining factor only in the crinoid arms. Discussion The results of this study suggest that flexibility in multi- jointed beams is influenced by variables associated with both continuous and jointed beams. As predicted by engi- neering beam theory, beam diameter contributes to flexural and torsional stiffness in both crinoid arms and crustacean Factors affecting stiffness S. A. ETNIER Table . Structural design R- Model for log (stiffness) P values Regression coefficients log (diameter) Joint density Flexural stiffness Crinoids 16 0.87 0.0001* 2.4* -2.5* Crustaceans 15 0.92 0.0001* 4.0* 0.0 Torsional stiffness Crinoids 16 0.74 0.0002* 2.1* -1.3* Crustaceans 15 0.4 1 0.0001* 4.4* 0.1 Statistical values for multiple regressions of log (diameter) and joint density against log (stiffness) for both structural designs (n ~ sample size). The model tests the significance of the overall regression equation, which is in the form Y = A + BX, + CX2. Values given for log (diameter) and joint density are standard partial regression coefficients for flexural and torsional stiffness. Regression coefficients that contribute significantly to the overall model (P < 0.05) are marked with asterisks. For example, in crustaceans, the linear equation, log (£/) = a + 4.0*log (diameter) + 0.0*joint density, explained 92% of the observed variation in flexural stiffness. Joint density, with a regression coefficient of 0.0, did not contribute significantly to the model. antennae. In contrast, joint density contributes to beam stiffness only in the crinoid arms. Future studies of multi- jointed beams must consider both types of variables to fully understand the mechanical properties of these structures. The flexibility of crinoid arms is influenced both by beam diameter and by joint density, so morphological variation at the level of the entire arm or at the level of the individual ossicles will affect the mechanical properties of the arms. This result is particularly interesting because the evolution- ary success of the comatulid crinoids has been attributed to morphological changes that increased arm flexibility rela- tive to the stalked crinoids (Meyer and Macurda. 1977). Increased arm flexibility, coupled with the loss of a stalk. enabled comatulid crinoids to become secondarily mobile (Motokawa. 1988). Today, this mobility allows comatulids to feed in variable water currents and to move in response to undesirable environmental conditions or predators (Meyer and Macurda. 1977). The hypothesis that comatulid crinoid arms are more flexible than stalked crinoid arms could be directly tested in extant crinoids by using the techniques developed in this study. Similarly, arm diameter and joint density may serve as useful proxies for investigating flexi- bility in fossil specimens, providing valuable insight into changes in arm flexibility over evolutionary time. The mechanical properties of crinoid arms may deter- mine, in part, the distribution of individuals in the environ- ment. Comatulid crinoids differentially distribute them- selves on the basis of flow regime, with some species preferring exposed sites with relatively predictable flows and others preferring more sheltered areas, with unpredict- able flow (Meyer. 1973a). Comactinia is typically found in the in1 ''structure of the reef in sheltered areas (Meyer, I ' ' arms bend and twist relatively easily, potentially allowi ;inimals to adjust to flows that change in both velocity direction. In contrast, Florometra tends to be found at I apex of rocky outcroppings (Shaw and Fon- taine, 1990) where it is exposed to more regular flow (LaBarbera, 1982). The arms of Florometra are stiffer in both bending and twisting than those of Comactinia. The flexibility needed to respond to constantly changing flows may not be necessary in a steady current. Instead, the arms may simply need to be stiff enough to maintain their posi- tion. Because stiffness is related to arm diameter, these results also suggest that individuals within a species may be differentially distributed on the basis of their size. Larger individuals, with stiffer arms, may be able to live and feed in different areas than smaller, more flexible, individuals. These results suggest that the mechanical properties of crinoid arms may be tuned to the specific loading regime of the animal's preferred habitat. Although the magnitudes of flexural and torsional stiff- ness differed between these crinoid species, there were no detectable differences in their twist-to-bend ratios. This constancy suggests that there is a functional advantage to maintaining the relationship between flexural and torsional stiffness across comatulid crinoid species, although the bi- ological relevance of this constancy has yet to be deter- mined. One possibility is that the constant twist-to-bend ratio reflects selective pressure on the ability to passively orient the arms with respect to water currents during feed- ing. The ability to passively orient the arms may be partic- ularly important for crinoids that live where flow oscillates continuously. Passive orientation of entire organisms has been described in gorgonian corals (Theodor and Denizot, 1965; Wainwright and Dillon, 1969) and in stalked crinoids (Baumiller and Plotnick, 1989), although the mechanisms differ in these two systems. In comatulid crinoids, passive orientation has been observed in the individual pinnules radiating off of each crinoid arm (Baumiller and Meyer, 2000), suggesting that the mechanical properties of the arms may be less important than the mechanical properties of the pinnules themselves. Passive orientation of comatulid arms FLEXIBILITY IN MULTI-JOINTED BEAMS has apparently not been investigated. Alternatively, the con- stant twist-to-bend ratio of comatulid crinoid arms may reflect functional demands of locomotion. Comatulid cri- noids use their arms to crawl along the surface of the reef and to swim freely through the water. Each of these activ- ities place different functional demands on the arms, and these demands may be reflected in the constant twist-to- bend ratios. The mechanical properties discussed here represent the passive properties of the arms. Crinoids may be able to actively alter the mechanical properties of their arms by contracting the muscles that cross from one ossicle to the next or by making changes in the properties of their con- nective tissue. As in all echinoderms, crinoid ligaments are made up of catch connective tissue that can exhibit dramatic changes in mechanical properties (Diab and Gilly, 1984; Motokawa, 1984). Crinoids might be able to use muscular contractions to position the arms, and then use the mutable properties of catch connective tissue to maintain that posi- tion with little muscular effort (Willkie. 1983). Thus, the ability of a crinoid to feed and locomote may depend on both the passive and active properties of the arms. The crinoids used in this study were initially preserved in formalin, which increases the number of cross-links be- tween the proteins in connective tissue (Presnell and Schreibman, 1997). Such treatment will greatly influence the mechanical properties of these tissues (Wainwright et til., 1976). The reported values for flexural stiffness are within an order of magnitude of values reported for the stalks of living stalked crinoids (Baumiller and LaBarbera. 1993). Yet the stalks were about 3 times larger in diameter than the comatulid arms, suggesting that the values for the flexural stiffness of arms given here are overestimates of the true values. No values were available for comparisons with torsional stiffness. In the multi-jointed vertebrate spine, the magnitudes of both flexural and torsional stiffness increased after preservation in formalin, yet the twist-to-bend ratio remained the same (Wilkie et ai, 1996). Thus, although the magnitudes of flexural and torsional stiffness in preserved crinoids may be overestimates, the twist-to-bend ratios may accurately reflect values for living crinoids. The factors affecting beam stiffness in crustacean anten- nae differed in part from those affecting crinoid arms. The flexibility of crustacean antennae is dependent on their diameter but not their joint density, suggesting that these beams effectively function as continuous beams. Increases or decreases in the number of joints in the antennae do not significantly affect their mechanical properties. I suggest that the joints of the antenna are not designed to allow appreciable bending or twisting. Previous studies have shown that the mechanoreceptors on the antenna are sensi- tive to bending deformations of 0.06 degrees at each joint (Tautz et til,, 1981 ). Thus, the joints may allow only very small deformations. Deformations of this magnitude, if they occurred, were below the resolution of the present equip- ment. Both Procambants and Panulims use their antennae to collect sensory information, but Panulirus also uses its antennae in aggressive interactions with predators (Atema and Cobb, 1980). The antennae of Panulirus do not bend or twist easily, which may make them more effective tools for warding off predators. Yet, because stiffness is related to diameter, the ability to ward off predators may be size dependent. In contrast, the antennae of Procambarus de- form easily in response to loads, ensuring their sensitivity to environmental stimuli. The twist-to-bend ratios differed sig- nificantly, with Panulims having a lower ratio than Pro- cambarus. A variable twist-to-bend ratio suggests that flex- ural stiffness and torsional stiffness are not tightly coupled in the crustacean antenna. This decoupling may be indica- tive of the different functional demands placed on the two antennae, or alternatively, may simply reflect the functional irrelevancy of torsional stiffness in these structures. Mechanistically, the decoupling of flexural and torsional stiffness in crustacean antennae may be a function of subtle morphological or material changes that affect one of the stiffness variables to a greater degree than the other. For example, the stiffness of a hollow beam, such as a crusta- cean antenna, is influenced not only by its external cross- sectional shape, but also by the thickness of its walls (Roark, 1943; Niklas, 1998). This study did not investigate wall thickness or other morphological features within each hollow antenna. Thus, the assumption that the two different antennae have the same structural design may be inappro- priate despite their external similarity. The antennae used in this study were frozen and thawed prior to mechanical tests, which could have affected the results. The flexural stiffness reported for Prociimbants was very similar to values reported for other similarly sized freshwater crayfish that were tested immediately after death (Sandeman, 1989). For torsional stiffness, no data from fresh specimens were available for comparisons. Again, the values reported here reflect the passive properties of the antennae. In crustaceans, no muscles crovs over the joints in the antenna, so there is no possibility of active control of the mechanical properties of the stru The mechanical properties n1 nulti jointed beams do not differ dramatically from tin; .if continuous biological beams, despite large different in underlying morphology. Both multi-jointed and continuous beams have similar me- chanical properties despite vastly different structural de- signs and functional demands. This convergence suggests that the structural design of a biological beam does not necessarily determine the resultant mechanical properties. This commonality also emphasizes the importance of flex- ibility, both in bending and in twisting, biological beams. S. A. ETNIER Acknowledgments I thank Shaun Cain of the University of North Carolina. Chapel Hill, for assistance in obtaining Panulinis argus antennae, and Duke University Marine Lab for loaning me specimens of Conuictiniti echinoptera from their inverte- brate collection. I am grateful for the comments of Dr. Steve Vogel and Dr. Bill Hoese on early drafts of this manuscript, as well as for the critical comments of two anonymous reviewers. Literature Cited Atema, J., and J. S. Cobb. 1980. Social behavior. Pp. 409-450 in The Biology and Management of Lobsters. Vol I : Physiology and Behavior. ). S. Cobb and B. F. Phillips, eds. Academic Press. New York. Baumiller, T. K., and M. LaBarbera. 1993. Mechanical properties of the stalk and cirri of the sea lily Cenocrinus asterius. Comp. Biochem. Physiol. 106A: 91-95. Baumiller. T. K.. and D. L. Meyer. 2000. Crinoid postures in oscillat- ing current: insights into some biomechanical aspects of passive filter feeding. Ceo/. Soc. Am. Abs. Progs. 32(7): A372. Baumiller, T. K., and R. E. Plotnick. 1989. Rotational stability in stalked crinoids and the function of wing plates in Pterotocrinus depressus. Let/lain 22: 317-326. Birenheide, R., and T. Motokawa. 1994. Morphological basis and mechanics of arm movement in the stalked crinoid Meracrintis rotun- tli/s (Echinodermata, Crinoida). Mar. Biol. 121: 273-283. Birenheide, R., and T. Motokawa. 1996. Contractile connective tissue in crinoids. Bint. Bull. 191: 1-4. Breimer, A. 1978. General morphology, recent crinoids. Pp. 9-58 in Treatise on Invertebrate Paleontology. Parr T. Echinodermata 2. Geo- logical Society of America. Boulder, CO. Denny, M. VV. 1988. Biology and the Mechanics of the Wave-swept Environment. Princeton University Press. Princeton. NJ. Diab, M., and VV. F. Gilly. 1984. Mechanical properties and control of non-muscular catch in spine ligaments of the sea urchin. Strongvlocen- trotus francisc amis. J. Exp. Biol. Ill: 155-170. Ennos, A. R. 1993. The mechanics of the flower stem of the sedge Carex acutiformis. Ann. Bot. 72: 123-127. Etnier, S. A., and S. Vogel. 2000. Reorientation of daffodil (Narcissus: Amaryllidaceae) flowers in wind: drag reduction and torsional flexi- bility. Am. J. Bot. 87: 29-32. Gal, J. 1993. Mammalian spinal biomechanics. I. Static and dynamic mechanical properties of intact intervertebral joints. J. Exp. Biol. 174: 247-280. Gere, J. M., and S. P. Timoshenko. 1984. Mechanics of Materials. Brooks/Cole Engineering Division, Monterey, CA. Glantz. S. A. 1992. Primer of Biostatistics. 3rd ed. McGraw-Hill. New York. LaBarbera, M. 1982. Metabolic rates of suspension feeding crinoids and ophiuroids (Echinodermata) in a unidirectional laminar flow. Comp. Biochem. Physio/. 71A: 303-307. Liddell, \V. D. 1982. Suspension feeding by Caribbean comatulid crinoids. Pp. 33-40 in Ecliinodenns: Proceedings of the International Conference, Tampa Bay. J. M. Lawrence, ed. A. A. Balkema. Rotterdam. Long, J. H., Jr., D. A. Pabst, VV. R. Shepherd, and W. A. McLellan. 1997. Locomotor design of dolphin vertebral columns: bending mechanics and morphology of Delphinus delphis. J. Exp. Biol. 200: 65-81. Meyer, D. L. 1971. The collagenous nature of problematical ligaments in crinoids (Echinodermata). Mar. Biol. 9: 235-241. Meyer, D. L. 1973a. Feeding behavior and ecology of shallow-water unstalked crinoids (Echinodermata) in the Caribbean Sea. Mar. Biol. 22: 105-129. Meyer, D. L. 1973b. Distribution and living habits of comatulid crinoids near Discovery Bay. Jamaica. Bull. Mar. Sci. 23: 244-259. Meyer, D. L., and D. B. Macurda, Jr. 1977. Adaptive radiation of the comatulid crinoids. Paleohiolog\ 3: 74-82. Motokawa, T. 1984. Connective tissue catch in echinoderms. Biol. Rev. 59: 255-270. Motokawa, T. 1988. Catch connective tissue: a key character for echi- noderm success. Pp. 39-54 in Echinoderm Biologv. R. D. Burke. P. V. Mladenov. P. Lambert, and R. B. Parsley, eds. Proceedings of the 6th International Echinoderm Conference, Victoria. A. A. Balkema. Rot- terdam. Muzik, K., and S. A. Wainwright. 1977. Morphology and habitat of the Fijian sea fans. Bull. Mar. Sci. 27: 308-337. New Lexicon Webster's Dictionary of the English Language. 1987. Lex- icon Publications, New York. Niklas, K. J. 1992. Plant Biomechanics: An Engineering Approach to Plant Form and Function. University of Chicago Press, Chicago, IL. Niklas, K. J. 1997. Relative resistance of hollow, septate internodes to twisting and bending. Ann. Bot. 80: 275-287. Niklas. K. J. 1998. Modes of mechanical failure of hollow, septate stems. Ann. Bot. 81: 1 1-21. Pressnell, J. K., and M. P. Schreibman. 1997. Humason's Animal Tissue Techniques. John Hopkins University Press. Baltimore, MD. Roark, R. J. 1943. Formulas for Stress and Strain, 2nd ed. McGraw- Hill. New York. Sandeman, D. C. 1985. Crayfish antennae as tactile organs: their mo- bility and the responses of their proprioceptors. J. Comp. Phvsiol. 157A: 363-373. Sandeman, D. C. 1989. Physical properties, sensory receptors and tactile reflexes of the antenna of the Australian freshwater crayfish Che rax destructor. J. Exp. Biol. 141: 197-218. Shaw, G. D., and A. R. Fontaine. 1990. The locomotion of the comatu- lid Florometra serratissima (Echinodermata: Crinoidea) and its adap- tive significance. Can. J. Zool. 68: 942-950. Sokal, R. R., and F. J. Rohlf. 1981. Biometry: The Principles and Practice of Statistics in Biological Research. 2nd ed. W. H. Freeman. New York. Tautz, J., W. M. Masters, B. Aicher, and H. Markl. 1981. A new type of water vibration receptor on the crayfish antenna. I. Sensory physi- ology. J. Comp. Physiol. 144: 533-541. Theodor, J., and M. Denizot. 1965. Contribution a 1'etude des Gor- gones. I. A propos de 1'orientation d'organismesmarins fixes vegetaux et animaux en fonction du courant. Vie Milieu 16: 237-24 1 . Vedel, J.-P. 1985. Cuticular mechanoreception in the antennal flagellum of the rock lobster Palinurus vulgaris. Comp. Biochem. Phvsiol. 80A: 151-158. Vogel, S. 1984. Drag and flexibility in sessile organisms. Am. Zool. 24: 37-44. Vogel, S. 1992. Twist-to-bend ratios and cross-sectional shapes of peti- oles and stems. J. Exp. Bot. 43: 1527-1532. Vogel, S. 1995. Twist-to-bend ratios of woody structures. J. Exp. Bot. 46: 981-985. Wainwright, S. A., and J. R. Dillon. 1969. On the orientation of sea fans (Genus Gorgonia}. Biol. Bull. 136: 130-139. Wainwright. S. A., W. D. Biggs, J. D. Currey, and J. M. Gosline. 1976. Mechanical Design in Organisms. Princeton University Press. Prince- ton. NJ. VV'ilkie, H. J., S. Krischak, and L. E. Claes. 1996. Formalin fixation strongly influences biomechanical properties of the spine. J. Biomech. 29: 1629-1631. Wilkie, I. C. 1983. Nervously mediated change in the mechanical properties of the cirral ligaments of a crinoid. Mar. Be/iav. Physiol. 9: 229-248. Reference: Biul. Bull. 200: 9-19. (Fehruarv 2001) Three-Dimensional Analysis of Finlet Kinematics in the Chub Mackerel (Scomber japonicus) JENNIFER C. NAUEN* AND GEORGE V. LAUDER Department of Organismic ciiul Evolutionaiy Biologv, Hun-unl University; 26 Oxford Street, Cambridge. Massachusetts 02138 Abstract. Finlets, which are small non-retractable fins located on the body margins between the second dorsal and anal fins and the caudal fin of scombrid fishes, have been hypothesized to improve swimming performance. The ki- nematics of three posterior finlets of the chub mackerel, Scomber japonicus, were examined using three-dimensional measurement techniques to test hypotheses on finlet rigidity and function during steady swimming. Finlet bending and finlet planar orientation to the A:, vc, and AY planes were measured during steady swimming at 1.2 lengths s~' in a flow tank. Despite very similar morphology among the individual finlets. there was considerable variability in finlet flexure during a stroke. Several of the finlets were relatively rigid and flat (with intrafinlet angles close to 180° during the stroke), although intrafinlet angle of the proximal portion of the most posterior finlet varied considerably over the stroke and was as low as 140° midstroke. Finlets showed complex orientations in three-dimensional space over a stroke, and these orientations differed among the finlets. For example, during tail deceleration the proximal portion of the fifth finlet achieves a mean angle of approximately 75° with the A; plane, while the distal portion of this finlet is oriented at 1 10". Our data suggest that the trajectory of local water flow varies among finlets and that the most posterior finlet is oriented to redirect flow into the developing tail vortex, which may increase thrust produced by the tail of swimming mackerel. Received 30 May 2000; accepted 1 December 2000. * To whom correspondence should he addressed. E-mail: jnauen® oeb.harvard.edu. Introduction Finlets are small non-retractable fins characteristic of scombrid fishes including mackerel, bonitos, and tuna (Col- lette and Nauen. 1983; Joseph el ill., 1988). The finlets are situated on the dorsal and ventral body margins adjacent to the tail (spanning the region between the second dorsal and anal fins and the caudal fin. Fig. I ). In the case of the five dorsal and five ventral finlets of the chub mackerel. Scomber japonicus (Fig. 1 ). the summed surface area of the finlets is about 15% of the surface area of the caudal fin (Nauen and Lauder, 2000). Muscles that may actively control finlet motion insert at the base of each finlet (Nauen and Lauder. 2000). Scombrid fishes are capable of high locomotory perfor- mance, including burst speeds from 18 body lengths per second (bl s~') for mackerel (Wardle and He. 1988) to up to 27 bl s~' for tuna (Fierstine and Walters, 1968; also see Magnuson, 1978), and cruising speeds from 3.5 bl s~' for mackerel (Wardle and He. 1988) to 6-10 bl s~' for tuna (Yuen. 1970; summarized in Beamish. 1978). Given the close proximity of finlets to the caudal fin. previous inves- tigators have suggested that finlets play a role in locomo- tion. Walters ( 1962) proposed that finlets direct flow longi- tudinally along the body, and Magnuson (1970) and Lindsey (1978) suggested that finlets direct flow across the caudal peduncle and caudal keels. A recent study by Nauen anu Lauder (2000) using two- dimensional (2-D) kinematic ti?\i!ysis methods to quantify the kinematics of finlets of S. jiiptmicits showed that during steady forward locomotion at speeds from 1.2 to 3.0 fork lengths (/) s~', finlet kinematics in the vertical (AT) and horizontal (.\z) plane were independent of speed. Angle of attack calculations using the kinematic measurements and the assumption that the direction of flow incident to the finlet was equal and opposite to the path of motion of the 10 J. C. NAUEN AND G. V. LAUDER Figure 1. A schematic of the flow tank and camera system showing camera A (dashed line) viewing the mirror (C), which was situated at a 45° angle to the A; plane and showed a dorsal view of the fish. The origin of the dorsal view image (white asterisk in the left image) was in the upper left comer of the image. The viewing area of camera B (dotted line) was a lateral view of the left side of the fish: the origin of the lateral image (black asterisk in the right image) was in the lower left corner of the image. The three posterior (inlets are identified in both of the images. Scale bars indicate 1 cm. fish indicated that finlet oscillation in the horizontal plane is largely passive, and thrust is not created by lift-based mech- anisms. However, the position of the finlets as the tail decelerates (at the end of each stroke) suggests that the finlets might direct flow into the developing caudal fin vortex, thus enhancing vortex circulation and thrust. The 2-D method was useful for determining basic patterns of finlet movement and the independence of these patterns from swimming speed (Nauen and Lander. 2000) A limitation of this method, however, was that each lini'.-i was considered to be a flat plate that acted as a single functional unit during the tail beat cycle. Thus, any flexion of the finlet in response to hydrodynamic load was neglected. Furthermore, the orientations of the finlets in three-dimensional (3-D) space were not determined. This information is important for understanding finlet hydrodynamic function because it is the motion and orientation of the surface of a fin that creates fluid motion and generates force (Dickinson, 1996). For examples of how 2-D kinematic measurements can be misleading for evaluating fin hydrodynamics, see Ferry and Lander (1996), Lauder and Jayne (1996), Walker and Westneat. 3-D FINLET KINEMATICS 11 (1997), Gibb et cil. ( 1999), Wilga and Lauder (1999). and Lauder (2000). Thus, the primary goal of this paper is to quantify the movement of finlets in three dimensions and to describe the orientation of the finlet surfaces with respect to three exter- nal earth reference planes (.vv, xz, and vc). To test the hypothesis that each finlet acts as a single rigid flat plate, we divided each finlet into two separate elements and calcu- lated the internal angle of these elements to each other as an approximation of finlet curvature. A priori we expected that finlet deformation would be low, because a dense assembly of fin rays support each finlet (Nauen and Lauder, 2000), and that the magnitude of flexion and 3-D orientation of the finlets would be similar, because the individual (inlets are very similar in morphology (see Fig. 1, Fig. 2, and Nauen and Lauder, 2000). Using the 3-D data, we determined the position of the finlets during a critical portion of the stroke cycle when, as predicted by the vorticity enhancement hy- pothesis described above, the finlets may redirect water flow towards the caudal fin vortex. We then use these data to predict the direction of water motion in the region of the finlets. Materials and Methods Animals Chub mackerel. Scomber japonicus (Houttuyn) were col- lected, using rod and reel, from various locations in coastal southern California. The animals were fed chopped smelt and housed in 1200-1 tanks at a water temperature of 18 ± 2 °C in a photoperiod of 12:12 h light:dark. Three individ- uals (numbered 7, 9, and 10) ranging in fork length ( / ) from 20 to 26 cm were studied here. 3-D kinematic measurements Experiments were conducted using a 600-1 flow tank with a working area 82 cm long X 28 cm wide X 28 cm high (Fig. 1) and a water temperature of 19 ± 1 °C. The speed profile of the flow across and along the working section of the tank has been determined by tracking dye streams on images collected using high-speed video (for details, see Jayne et ai, 1996). To accurately image the motion of the finlets for a series of tailbeats, it was necessary that the fish maintain a consistent position relative to the field of view of the cameras. Thus, we used a flow tank rather than have the fish swim in still water through the field of view. Two cameras that were part of a N AC HS V 500 C1 video system were mounted on a vertical frame and aimed per- pendicular to the flow tank (Fig. 1 ). The upper camera (Fig. 1A) was focused on a front-surface mirror (Fig. 1C) that was immersed in the flow at a 45° angle to the bottom of the tank (the .v; plane) and showed a dorsal view of the fish. The lower camera (Fig. IB) provided a lateral view (the vv plane) of the finlets. Using Nikon Micro-Nikkor 55-mm lenses with these cameras, we were able to image the finlets clearly in a field of view that was about 5 cm X 4 cm (Fig. 1 ). When a mackerel was in the field of view of the cameras and the image was in focus, the animal was necessarily swimming in the center of the working section of the tank. Thus, no data were obtained near the walls or floor of the flow tank, or the upper surface of the water. The fields of view of both cameras were scaled at the start of the exper- iment using two perpendicularly oriented rulers. The video system electronically synchronized the two cameras and recorded images at 250 Hz. About 12-15 images were collected per stroke of the tail. Video images were recorded continuously until sufficient sequences of steady swimming with the finlets in the fields of view of both cameras were obtained. We swam the mackerel at speeds of 1.2 and 2.2 fork lengths per second (/ s~'). These speeds are within the range of swimming speeds (0.4-3.5 bl s~') that mackerel can sustain for longer than 200 min (Wardle and He, 1988), and match the speeds used in previous kinematic studies of mackerel finlets (Nauen and Lauder, 2000) and tail (Gibb et «/., 1999). The video images were imported into a computer using DT-Acquire software with a Data Translation video card (Data Translation. Inc.). The procedures for calculating 3-D kinematics were adopted from those used in previous stud- ies (Lauder and Jayne, 1996; Wilga and Lauder, 1999; Lauder, 2000). With the Cartesian coordinate system, any point on the video images can be identified by .v, v, and ; values. The origin was assigned to the lower left corner of the lateral view and the upper left corner of the dorsal view because the dorsal view was recorded using a mirror (the origin is denoted by asterisks in the images in Fig. 1 ). Because the finlets move over the body midline with each stroke (Nauen and Lauder, 2000), and we viewed the left side of the fish, the finlets were in full view of camera B as the tail was beating from left to right. There is a phase lag in the movement of the finlets relative to the body (Nauen and Lauder, 2000), thus the finlets are in view from about the start of the second quarter of one stroke to the end of the first quarter of the next (as determined by digitizing the dorsal insertion of finlet 5 A, see Figs. 3 and 5). The movements of finlets 3, 4, 5A, and 5B were quantified in this study. Previous kinematic measurements (Nauen and Lauder, 2000) indicated that finlet size and amplitude of finlet movement decrease anteriorly, with finlets 1 and 2 showing small excursions compared to those of finlet 5. In addition, the body of S. japonicus tapers posteriorly (Fig. 1). For example, for the fish 23 cm in fork length examined here, the depth of the body at the insertion of finlet 5 was 0.72 cm, which is 30% of the depth of the body at the position of finlet 1 . The posterior decrease in the depth of the body and increase in the size and excursion of the finlets result in the posterior finlets moving 12 J. C. NAUEN AND G. V. LAUDER over a much greater area of the body. The tips of dorsal and ventral finlet 5B actually meet at the lateral midline of the body on the caudal peduncle during their maximum vertical excur- sion (see fig. 12 of Nauen and Lauder, 2000). Thus, because the posterior finlets have a much larger potential hydrodynamic effect than the anterior finlets. we quantified the 3-D kinemat- ics of finlets 3, 4, and 5. The fifth finlet in 5. juponicus is composed of two distinct groups of fin rays joined by a thin, clear membrane (Nauen and Lauder, 2000). Finlet 5 was treated here as two separate elements, 5A and 5B. The geometric relationship between those two elements (the internal angle of finlet 5) was also quantified. Single finlets 3 and 4 were an interesting com- parison to the double-finlet structure of finlet 5. Each finlet was divided into two triangles that were defined by a series of points (Fig. 2). This method gives a very good representation of finlet shape (Fig. 2) and allowed us to estimate finlet curvature by calculating the angle between the two triangular surfaces (angle a in the animal's frame of reference. Fig. 3), given the assumption of span- wise rigidity of the two triangles. Angle a for finlet 5 as a whole was the angle between finlets 5 A and 5B (Fig. 2). The angles made by each of the eight triangular surfaces to the three orthogonal planes in the earth frame of reference (.vv, .vc. and yz) were also determined (Fig. 3). Downloaded video images were digitized using a cus- tomized program. The coordinates were imported into Excel (Microsoft) to calculate the internal angles of the finlets and the angles of the finlet triangles to the three external refer- ence planes. Each calculated angle was verified in a cus- Caudal keels Figure 2. An outline of Scomber japonicus (gray lines) traced from a video image showing finlets 3, 4, 5A, 5B, and the caudal keels tor reference. Note that the fifth finlet is morphologically composed of two distinct units (5A and 5B) that are bound by a clear membrane (Nauen and Lander. 2000). Each set of four points (black circles) defined two triangular suifai.es shown by the solid and dotted lines. Three of the points on nnlets 5A and ?B (illustrated on the ventral finlets for clarity) were used to determine the rigidity of the fifth dorsal finlet as a whole. The thin, clear membrane that covers each of finlets 1-5 and attaches to the body is depicted in gray on ventral finlets 3 and 4. Figure 3. Diagram of a finlet (dark gray triangle) from a dorsal (A) and lateral (B) view. From the dorsal view (A) the v axis (indicated by an asterisk) is coming straight out of the page; in both views the .re plane is light gray. The internal angle (a), and the .vv. .v,-. and y- angles (dotted lines), shown here in the dorsal view of triangle A. were calculated from digitized points (Fig. 2). See the text for further explanation. tomized 3-D visualization program. All angle measure- ments were made from the upstream left side of the triangular surface to the plane of interest (Fig. 3). Under this measurement convention, if the left finlet surface was po- sitioned to the right of the body midline (as shown in Fig. 3) the yc angle was less than 90°; a yc angle greater than 90° indicated that the left finlet surface was on the left side of the body midline. An A; measurement greater than 90° indicated that the surface was tilted away from the floor of the tank (or the frontal plane of the fish, Fig. 3). An .vv angle measurement less than 90° indicated that the surface was oriented to the left of the body midline. Values of .vv angles decreased to 0° as the finlet moved parallel to the body midline and then abruptly increased to about 120° as the finlet crossed to the right of the body midline. This abrupt increase (see Fig. 5 A) is solely due to the measurement convention and does not reflect a large change in orientation of the finlet. 3-D FINLET KINEMATICS 13 Initial comparisons of a time series of the 3-D angles over a complete stroke for a single individual swimming at 1.2 and 2.2 / s ' indicated that the general patterns of finlet kinematics were not affected by speed. This observation is supported by our previous finding (based on a statistical analysis of 2-D measurements) that finlet kinematics were independent of speed over a speed range of 1.2 to 3.0 / s (Nauen and Lauder. 2000). On the basis of this information we focused on the 3-D kinematics of the finlets at the speed of 1.2 / s"1. Three to six tail strokes were digitized for each fish. The strokes were from sequential tailbeats for two of the fish; for the third fish we analyzed strokes from two sequential tailbeats and a third, single tailbeat. The digitized position values were not filtered. To determine the digitizing error, we digitized a single finlet 5 times. The calculated angle to the AV. AC. and y^planes were 23.0° ± 0.7°. 91.1° ± 1.0°. and 113.0° ± 0.8° (mean + SD, /; = 5). Therefore, the digitizing error is approximately 1°. Finlet movement over an entire stroke was determined for one individual; finlet position as the tail was decelerating was determined for multiple strokes from all three individuals. Statistics Statistical analyses were performed using Statgraphics (v. 3.0 for Windows, STSC, USA). To determine if the mean values of intrafinlet angles averaged over a tailbeat cycle for a single individual were significantly different from 180°. t tests were performed on the time series data. The purpose of this analysis was to determine the general trend of finlet flexibility over the course of a tail stroke. The probability values of the / tests were established using the sequential Bonferroni method of Rice ( 1989) to control for conducting multiple comparisons. A multivariate ANOVA could not be performed on all of the angle data for the multiple individ- uals because of insufficient degrees of freedom; therefore, the data for each plane (AV, A:, and yc) were separately analyzed using two-way ANOVAS. The position of the finlet on the body (finlet number) was considered a fixed effect, and the individuals were considered random effects. The data at the specific time at 157c of the tail beat cycle — the time when the posterior finlets are in position to influ- ence flow according to the vorticity enhancement hypothe- sis (Nauen and Lauder, 2000) — were also analyzed using this method. Tukey-Kramer post hoc tests were performed on each variable that showed significant effects of speed, finlet number, or structure. Results Kinematics over a stroke Given that the magnitude of digitizing error was about 1° (detailed in the methods section), intrafinlet angle (a) values over a complete stroke for one individual (fish 10) swim- ming at a slow cruising speed of 1.2 / s"1 indicate changes in finlet flexure over the stroke and variability in flexure among finlets. The variability in finlet flexibility is not directly attributable to the finlets' position on the body (Fig. 4) given that the greatest difference in a values from 180° (representing a flat plate) were about 40° for finlet 5A and about 15° for finlet 5B, and these two finlets are directly adjacent to each other (Fig. 2). Mean a values over a stroke, which represent a general index of finlet flexion, were not significantly different from 180° for finlets 3 (a = 179 ± 6°. mean ± SD, n = 11) and 5B (a = 182 ± 10°, mean ± SD. ;; = 10. t test. P = 0.61 and 0.52. respectively). The mean a values of finlet 4 ( 174 ± 6°. mean ± SD. n = 1 3 ) and finlet 5 A (160 ± 18°. mean ± SD, n == 10) were significantly different from 180° (t test. P = 0.003, and 0.006, respectively). The flexion of finlet 5A was largest three-quarters of the way through the stroke (at about 0.75 s in Fig. 4) and decreased to close to zero (a = 180°) at the end of the stroke (at about 0.1 s in Fig. 4). The mean a value of finlet 5 over one stroke for fish 10, measured as the angle between finlet 5 A and finlet 5B (see Fig. 2). was significantly different from 180° (t test. P < 0.001 ), indicating that the coupling between the two finlets is not rigid. The a for finlet 5 was relatively low throughout the stroke (a = 157 ± 8°, mean ± SD. n = 10). The time series of the orientation of the finlets to the planes AV (Fig. 5 A, B). yz (Fig. 5C. D). and xz (Fig. 5E. F) also indicate variability in orientation among the different finlets during different stages in the stroke. The more pos- terior finlets tended to make greater angles to the AV (lon- gitudinal) plane (Fig. 5A. B). The phased increase in AV angle to values greater than 90° reflects the finlets crossing the longitudinal body midline to the right side of the fish at the end of the stroke (see Fig. 2). Relative to the yc (trans- verse) plane (Fig. 5C, D). the angles of the finlets decreased throughout the stroke, and were less than 90° for the first quarter of the next stroke. As the AV angle increases to greater than 90°. the yc angle decreases to less than 90° (Fig. 5 A. B). indicating that the finlet has crossed to the right side of the body midline. The time series of the angle of finlets 3, 4. and 5B to the .YC plane during a stroke (Fig. 5E. F) suggests that the angles of finlets 3, 4. and 5B were at a slightly obtuse angle to the xz plane, while finlet 5A made an acute angle to the AC plane. However, when the mean values of the AC angle over a stroke were tested using unlisted probability values to control for conducting a series of simultaneous t tests (Rice. 1989), this observation was not statistically significant (Ta- ble 1 ). When averaged over an entire tail-beat, the angle of triangles 3A, 3B. 4A, and 4B to the AC plane were signifi- cantly different from 90° (Table 1 ). whereas the angles of finlets 5A and 5B were not (Table 1). This surprising statistical result is due to ( 1 ) averaging values over the 14 J. C. NAUEN AND G. V. LAUDER 210- r34 140 14 0.02 0.04 0.06 Time (s) O.OS 0.12 Figure 4. The internal angle (a) of (inlets 3 (triangle). 4 (diamond). 5A (gray square), and 5B (black square) of a single individual over a tail stroke at 1.2 / s"'. The angle between 5A and 5B (crosses) is also plotted. The dashed line indicates 180°. Axial body bending is indicated by the position of the body (at the insertion ot'rinlet 5A) on the ; axis (circles, 5A Z). entire tail stroke and (2) the fact that the P value at which significance is achieved decreases with an increasing num- ber of tests (Rice, 1989). We investigated this further by determining whether the intratinlet angle differed from 90° at a specific time in the stroke. Kinematics during tail deceleration Three-quarters of the way through a stroke, the tail de- celerates (Nauen and Lauder, 2000). Intrafinlet angle (a) values averaged from three fish indicate that finlet flexure was low at this point in the stroke, because a values were generally close to 180° (Fig. 6). Individual variation in the a value for finlet 5A was high, however: two of the indi- viduals showed relatively low mean a values (mean ± SD of 151 ± 6° and 153 ± 7°, for fish 10 (n = 6) and fish 9 (n = 3). respectively), similar to the value of 144° seen at that point in the time series of a single stroke for Fish 10 (Fig. 4): in contrast, individual 7 showed a mean a value of 182 ± 1° (H = 3). This variation was reflected in the significant individual effect on a for finlets 4, 5A, 5B, and 5 (F 15.9, P < 0.0001, Table 2). The significant interaction effect (F = 5.4, P = 0.0024, Table 2) indi- cates that there was no consistent change among individuals in a with position on the body (i.e., finlet number. Fig. 6). There was no significant effect of finlet position on the body on a (F == 3.0, P = 0.16). At this point in the stroke, the AT angle of the posterior finlets tended to be larger than those of the anterior finlets (Fig. 7 A). This difference is reflected in the significant finlet effects (F = 9.2, P = 0.0016, Table 3). There were also significant individual (F = 30.6, P < 0.0001, Table 3) and individual X finlet interaction effects (F = 2.98, P = 0.0046, Table 3). The y- angles tended to be greater than 90° (Fig. 7B); the ANOVA indicated significant individual (F = 20.2, P < 0.0001 ) and individual X finlet (F = 9.5, P < 0.0001) effects but not significant finlet effects (F = 2.53, P = 0.0996) on the yc angles (Table 3). The A; angles of finlet 5B tended to be greater than 90°, whereas finlet 5A tended to be less than 90° (Fig. 7C). The ANOVA indicated significant finlet (F = 8.12, P = 0.0027, Table 3) and individual X finlet effects (F ••-- 3.64, P ---- 0.0009, Table 3) but not significant individual effects (F = 1.56, P = 0.2190, Table 3) on the .re angles. To illustrate the position of the finlets three-quarters of the way through the tail stroke, the coordinates of finlets 4, 5A, and 5B from the time series in Figure 5 are plotted in 3-D space in Figure 8. The data shown here are the high- lighted points in the time series (Fig. 5). Although from a lateral view the finlets appear roughly flat and oriented normal to the AT plane (Fig. 8 A), the flexure of finlets 4 and 5, the flexure between 5 A and 5B, and the acute angle of 5A 3-D FINLET KINEMATICS 15 -pi r-pi i— [- 0.02 0.04 0.06 0.08 14 0.1 0.12 14 0.12 150-1 S 120-1 00 1> "u" 90- N 60- 30- -30 -26 150- -= « 120- 00 (L> T3 „ I « 90-^ 22 c oo 3 5 .2 < _lg Q N 60- 14 0 0.02 0.04 0.06 O.OS 0.1 0.12 30- 1 i ' r\l r\l r\^ ' i ' 0.02 0.04 0.06 0.08 0.1 0. :22 | to tl8 5 14 12 120-1 S 60 0.02 0.04 0.06 0.08 Time (s) 0.1 0.12 120-1 90-- 18° g 14 60- -30 -18 14 0 0.02 0.04 0.06 0.08 Time (s) Figure 5. The angle of finlets 3 (triangle). 4 (diamond). 5A (gray square) and 5B (black square) to the .vv (A, anterior triangles. B, posterior triangles), v; (C. D), and .v; (E, F) planes over a stroke by a single individual at 1.2 / s~ '. Axial body bending is indicated by the position of the body (at the insertion of rinlet 5A) on the : axis (circles). The tail stroke is defined from minimum to maximum values of 5A Z; the duration of this stroke was approximately 0.09 s. The angles at the time highlighted by the gray bar are plotted in 3-D space in Figure 8. The abrupt change in angles to the .vv plane in panels A and B (indicated by the dotted lines) is due to our measurement convention (see Fig. 3 and the methods section) and reflects the transition of the planar Imlct orientation across the body midline relative to the .vv reference plane, not a large movement by the (inlets. to the x~ plane are visible when the lateral view is rotated about 30° clockwise (Fig. 8B). Discussion Finlet morphology A detailed morphological description of the finlets of Scomberjaponicus is available in Nauen and Lauder (2000). and is useful for interpreting the three-dimensional pattern of movement. In brief, the finlets are on the order of I cm in length. A thin, clear membrane covers each finlet and attaches to the body. Jointed bony fin rays that extend to the distal tip of the fin stiffen each finlet. These rays articulate on a cartilaginous pad and are associated with muscles that appear to be homologous to the inclinator, depressor, and 16 J. C. NAUEN AND G. V. LAUDER Table 1 Results of t rests of the angle of each finlet to the XZ plane over a stroke Finlet P 4A 0.000035 4B 0.000086 3A 0.0087 3B 0.0021 5Aa NS 5Ab NS 5Ba NS 5Bb NS NS indicates not significant according to P = 0.05/fc-/ where k = the number of t tests performed and / = the order number of the variable based on its calculated P value (Rice. 1989). erector muscles of the dorsal fin (Jayne et ill., 1996). This structure is identical for rinlets 1 through 4. The fifth (most posterior) finlet of 5. japonicus has an interesting structural difference from the anterior four fin- lets. Finlet 5 is actually two finlets (5 A and 5B) that each have a separate set of the structural components described above (Fig. 2). Finlets 5 A and 5B are bound together by the clear membrane that covers the other finlets. Considered as a single functional unit, the fifth finlet is significantly larger than the anterior four finlets. Finlet kinematics On the basis of the presence of the fin rays throughout each of the finlets, the membrane binding finlets 5A and 5B. and the movement patterns of the finlets inferred from separate measurements in the .vv and v; planes, we previ- ously considered the fifth rinlet to be a single functional unit and each finlet to move as a flat plate normal to the ,vc and 200 T) 180 . 160- &MO- f-i -120- 100- 2 60- _c 40- 20- 0 3 4 5A 5B Fish 10 Fish 9 Fish ' Figure 6. Means and standard deviations of the intrafinlet angle la) for the finlets of three individuals at a single time point three-quarters of the way through the stroke as the tail is decelerating (n = 6. 3. and 3 strokes for fish 10, 9. and 7. respectively). The dashed line indicates an intrafinlet angle of ISO when the finlet elements are coplanar. Table 2 Results (F values) of the three-way ANOVA on intmfinlet angle Ji/ruig [nil deceleration Individual Variable Individual Finlet # x finlet df 2. 27 2, 4 4, 27 Intrafinlet angle (a) 15.9* 3.0 5.4* df is the degrees of freedom. * Statistically significant effect {P < 0.05). .vv planes (Nauen and Lauder, 2000). The present data show that this characterization was oversimplified. All of the (inlets display considerable flexion and tilting to the xz and v; planes during locomotion, with finlets 5A and 5B show- ing distinct patterns of flexion and movement. 60 n Fish 10 140- B 130- - 120- §iio- u 100- cjj 1 J f | § 90- s: > 80- 70- ^n- Fish 10 4b 5Aa 5 Ah Fish 10 Fish? Figure 7. Means and standard deviations of the angles of the anterior (a. solid bars) and the posterior (b. doited bars) elements of finlets 3 (white). 4 (light gray). 5 A (dark gray), and 5B (black) to the three reference planes as the tail is decelerating three-quarters of the way through the stroke (n = 6. 3, and 3 strokes for fish 10, 9. and 7. respectively). The dashed lines in B and C indicate 90°. 3-D FINLET KINEMATICS 17 Table 3 Results (F values) of the three-way ANOVAs on xy. yz and xz angles during tail deceleration Individual Variable Individual Finlet # x (inlet df 2. 54 5. 10 10. 54 xy angle v; angle AC angle 30.6* 20.2* 1.6 9.2* 2.5 8.12* 2.98* 9.49* 3.64* df is the degrees of freedom. * Statistically significant effect (P < 0.05). A priori we expected that the stiffness of finlets 3, 4, 5A. and 5B would he similar because they are structurally identical, hut that the stiffness of finlet 5 as a whole would be less than that of the individual finlets because the double- finlet structure is supported only by a clear membrane. In some cases this expectation was corroborated by the lack of bending seen in finlets 3 and 5B. and the slight bending (a = 174 ± 6°, mean ± SD. n = 13) of finlet 4. However, in two of the three fish examined, finlet 5A displayed significant bending during the tail stroke, with a values as low as 145° (compared with 180° for a flat plate). This variability in the stiffness of finlet 5 among individuals was unexpected. As anticipated, there was considerable flexion 15 12 14 16 18 20 X axis (mm) 22 24 26 2S axis (mm) Figure 8. Three-dimensional orientation of finlets 4. 5A, and 5B as the tail is decelerating three-quarters of the way through the stroke. The thick solid line indicates the position of the body midline. The tail is beating into the page. The anterior element of each finlet is shaded light gray, while the posterior element is shaded dark gray. The lateral view (A) is rotated 30° clockwise relative to the x- plane in panel B. The arrows in B depict the hypothesized fluid motion based on finlet orientation. 18 J. C. NAUEN AND G. V. LAUDER between finlet 5A and 5B, given that mean a values were about 160°. This resulted in a complex, concave structure for the entire fifth finlet for over half of the stroke (see Fig. 8). Fin deformation during movement has been demonstrated for the pectoral fins of elasmobranchs (Rosenberger and Westneat, 2000; Wilga and Lauder, 2000), chondrosteans (Wilga and Lauder, 1999), and teleosts (Webb, 1973; Geer- link, 1983: Gibb et . Biol. 200: 1549-1569. Walker, J. A., and M. W. Westneat. 2000. Mechanical performance of aquatic rowing and flying. Proc. R. Soc. Loud. B. 267: 1875-1881. Walters, V. 1962. Body form and swimming performance in the scom- broid fishes. Am. Zool. 2: 143-149. Wardle, C. S., and P. He. 1988. Burst swimming speeds of mackerel. Scomber scombrus L. J. Fish Biol. 32: 471-478. Webb, P. W. 1973. Kinematics of pectoral fin propulsion in Cymato- gaster aggregata. J. Exp. Biol. 59: 697-710. Wilga, C. D., and G. V. Lauder. 1999. Locomotion in sturgeon: func- tion of the pectoral fins. J. Exp. Biol. 202: 2413-2432. Wilga, C. D., and G. V. Lauder. 2000. Three-dimensional kinematics and wake structure of the pectoral fins during locomotion in leopard sharks Triakis senufasciutu. J. Exp. Biol 203: 2261-2278. Yuen, H. S. H. 1970. Behavior of a skipjack tuna. Katsuivuiws pclumis. as determined by tracking with ultrasonic devices. J. Fish. Res. Board Can. 27: 2071-2079. Reference: Biol. Bull. 2(10: 20-32. (February 2001) Metamorphosis of Coeloblastula Performed by Multipotential Larval Flagellated Cells in the Calcareous Sponge Leucosolenia laxa SHIGETOYO AMANO1 * AND ISAO HORI~ 1 Cancer Research Institute, Kanazawa Universit\, 13-1 Takaramachi, Kanuzcnva, Isliikuwu 920-0934; Department of Biologv, Kanazawa Medical University, Uchinada, Ishikiwa 920-0265, Japan Abstract. The calcareous sponge Leucosolenia laxa re- leases free-swimming hollow larvae called coeloblastulae that are the characteristic larvae of the subclass Calcinea. Although the coeloblastula is a major type of sponge larva, our knowledge about its development is scanty. Detailed electron microscopic studies on the metamorphosis of the coeloblastula revealed that the larva consists of four types of cells: flagellated cells, bottle cells, vesicular cells, and free cells in a central cavity. The flagellated cells, the principal cell type of the larva, are arranged in a pseudostratified layer around a large central cavity. The larval flagellated cells characteristically have glutinous granules that are used as internal markers during metamorphosis. After a free-swim- ming period the larva settles on the substratum, and settle- ment apparently triggers the initiation of metamorphosis. The larval flagellated cells soon lose their flagellum and begin the process of dedifferentiation. Then the larva be- comes a mass of dedifferentiated cells in which many au- tophagosomes are found. Within 18 h after settlement, the cells at the surface of the cell mass differentiate to pinaco- cytes. The cells beneath the pinacoderm differentiate to scleroblasts that form triradiate spicules. Finally, the cells of the inner cell mass differentiate to choanocytes and are arranged in a choanoderm that surrounds a newly formed large gastral cavity. We found glutinous granules in these three principal cell types of juvenile sponges, thus indicat- ing the multipotency of the flagellated cells of the coelo- blastula. * To whom correspondence should samano@kenroku.kanazawa-u.ac.jp he addressed. E-mail: Introduction All major sponge groups emerged on the earth in the late Precambrian, about 580 million years ago (Li et ai, 1998). Thereafter they have changed in different ways and adapted the structure of their larvae and process of metamorphosis to survive through the subsequent geologic eras. As a result of adaptation, sponge development is greatly diversified among higher taxonomic groups (Brusca and Brusca, 1990). Most sponges are viviparous, but some are oviparous. Some viviparous sponges release crawling larvae, but most release swimming larvae, of which there are a number of types. An amphiblastula or coeloblastula is produced by calcareous sponges: a parenchymella is the larva of the majority of demosponges, but several other larval types are known; and a trichimella is exclusive to hexactinellid sponges. The modes of metamorphosis of these larvae are also quite variable (Simpson, 1984). In spite of such variations, recent molecular evidence shows that all sponges are of monophy- letic origin (Mu'ller. 1997). Because all adult sponges are sessile animals, their free- swimming larval period is a critical one for habitat selec- tion. After the free-swimming period, larvae settle on the substratum and initiate metamorphosis. The layer of flagel- lated cells is a larval organ for swimming, therefore those cells disappear soon after settlement. Because the sponge larvae are generally minute and their nutritive materials are limited, it seems very disadvantageous to discard the flag- ellated cells during metamorphosis. The developmental fate of these cells in the parenchymellae of demosponges has long been debated. One view holds that the flagellated cells are transformed into the choanocytes of juvenile sponges (Borojevic and Levi, 1965; Boury-Esnault, 1976; Amano and Hori, 1994, 1996). but the opposing view contends that 20 METAMORPHOSIS OF CALCAREOUS SPONGE 21 they are lost by exfoliation or phagocytosis during meta- morphosis (Bergquist and Glasgow 1986: Misevic et al., 1990; Kaye and Reiswig, 1991 ). The flagellated cells in the coeloblastulae of calcareous sponges cannot, however, be discarded, because these cells make up the greater part of the larva. The coeloblastulae of calcareous sponges have been stud- ied very rarely, and we know little about the process of their metamorphosis (Minchin. 1896; Tuzet, 1947; Borojevic, 1969). In this study we will show the fine structure of the coeloblastula, metamorphosing larva, and juvenile sponge of Leucosolenia ld.\n in clear electron micrographs. After settlement, the larval flagellated cells dedifferentiate into a simple cell mass on the substratum; thereafter they differ- entiate again into the three principal cell types of a juvenile sponge. We discuss the multipotency of the larval flagel- lated cells of L. laxa by comparing them with the develop- mental potency of the larval flagellated cells of other sponges. Materials and Methods Sponges and larvae In early September, specimens of Leucosolenia laxa were collected from rafts in Mutsu bay in northern Japan. L. laxa is one of the most common calcareous sponges in this region. Immediately after collection, the sponges were placed in containers with seawater, brought to the laboratory of Asamushi Marine Biological station within 1 h. and kept in running seawater. To collect larvae, in the early morning sponges were placed in glassware with seawater. Under the natural illu- mination, about one-tenth of them released larvae, begin- ning soon after dawn and ceasing by noon. So the larval release of this calcareous sponge is probably controlled by light cycles, as it is in certain demosponges (Amano, 1986, 1988; Maldonado and Young, 1996). The larvae, which began swimming immediately upon release, were placed in petri dishes with filtered seawater; the seawater was changed daily. The larvae and juveniles can be reared for a week under these conditions. Electron microscopv Released larvae swam around mainly under the water surface, and after a swimming period, settled on the glass surface or beneath the air-water interface. Free larvae were picked up on a platinum loop and placed into fixative. Settled larvae were collected from beneath the air-water interface in the same way. This procedure minimized the mechanical stress of the larvae during collection. It is quite difficult to preserve the infrastructure of the larvae of this species satisfactorily. After trying a number of fixatives, we found the following formula to be the best: 1% paraformaldehyde. 7.5% glutaraldehyde. and 14% sucrose in 0.1 M cacodylate buffer. pH 7.4. Larvae or juvenile sponges were fixed in the ice-cold fixative for 1 h. Then the samples were rinsed twice in the same buffer and postfixed in 1% osmium tetroxide in 0.1 M cacodylate buffer, pH 7.4, for 1 h. They were dehydrated through a graded ethanol series, cleared in propylene oxide, and embedded in Spurr epoxy resin (Spurr, 1969). Semithin sections were stained with toluidine blue for light microscopy. Ultrathin sections were stained with uranyl acetate and lead citrate, then ex- amined and photographed under a Hitachi H-500 electron microscope. Results Free-swimming larva The calcareous sponge Leucosolenia laxa releases free- swimming coeloblastulae through the osculum with the excurrent water. The free larvae were fixed soon after re- lease and on the next day. We examined larvae collected after different periods of swimming but found no morpho- logical difference; thus we show only the images of the free larvae fixed soon after release. The coeloblastula is a hollow spheroid consisting of a layer of flagellated cells around a large central cavity (Fig. 1). This larva, which is about 90 /u,m in length and about 50 /urn in width, is entirely cili- ated— including its posterior pole. The coeloblastula con- tains three additional types of cells: bottle cells, vacuolar cells, and free cells in the central cavity. Figure 1 also shows the polarity of the larva. Its anterior pole consists exclu- sively of flagellated cells, and its posterior pole includes several vacuolar cells. Flagellated cells. Figure 2 shows the flagellated cells of the coeloblastula of L. la.\a. These cells are organized in a pseudostratified cell layer that is about 15-fj,m thick and V F B ,- Figure 1. Coeloblastula of Leucosolenia IUMI. The pseudostratified layer of flagellak-J cell1, sum .unds a large central cavity. B: bottle cell. F: free cell in the central cavity. V: vacuolar cell. Scale bar = 10 /urn. S. AMANO AND I. HORI Figure 2. The pseudostratitied layer of flagellated cells in the coeloblastula of Leucosolenia laxu. F: fibrous granule in the outermost cytoplasm, G: glutinous granule in the apical cytoplasm, P: phagosome-like granule in the basal cytoplasm. Scale bar = 2 /nm. lacks a basal lamina. One flagellum emerges from the outer surface of the cell. There are glutinous granules in the apical cytoplasm and fibrous granules in the outermost cytoplasm. An elongated nucleus, about 5x2 /urn, has a nucleolus and heterochromatin masses. The basal cytoplasm contains pha- gosome-like granules, and the innermost region of the cell is filled with many small (about 0.25 /Ltm) vesicles and a few lipid droplets (Fig. 3). There are mitochondria in both the apical cytoplasm and the basal cytoplasm. Figures 4 and 5 show magnified images of the apical region of flagellated cells. It is apparent that one flagellum emerges from the flat outer surface of the cell. The most proximal portion of the flagellum is somewhat electron- dense. From a basal body, two rootlets extend downward to the sides of a nucleus. The rootlets are striated, with a distance of about 40 nm between the midpoints of two neighboring striae. There is a basal foot running parallel to the outer surface. The Golgi apparatus is always located vertically along the rootlet. Only in the apical region are the neighboring flagellated cells united, by simple apposition (Figs. 4 and 5); in the deeper region, there is a wide space between the cells (Fig. 2). We found a consistent arrangement and orientation in the apical region of larval flagellated cells. One side of the outermost portion of the cell protrudes toward a neighbor cell, and the protrusion is regularly directed to the anterior end of the larva (Figs. 4 and 5). A basal foot consistently extends in the opposite direction — that is, to the posterior METAMORPHOSIS OF CALCAREOUS SPONGE '*- 23 " - ''-'•' - 4 , . • . • w* m R • .s R Figure 3. The innermost region of flagellated cells in the coeloblastulu ot If»u'Mi Leucosolenia laxa about 18 h after settlement. Scale bar = 2 jj.ni. Figure 16. Autophagosome (A) with a degenerative rootlet (R) in the inner cell mass of the settled lar\;i nl I.t'iii uMileiiUi la\a about 18 h after settlement. Scale bar = 1 /urn. loblastulae; therefore, we thoroughly studied the fate of these cells during metamorphosis. Examination of a large number of electron micrographs revealed that the flagellated cells become the three principal cell types of juvenile sponges — that is, pinacocytes. scleroblasts. and choano- cytes. This conclusion is supported by the fact that these three cell types contain glutinous granules, which only the flagellated cells have in the free-swimming larva. This fact shows clearly that the flagellated cells of the coeloblastulae hold multipotency, althoi i:ive been specialized for swimming. Experimental Approaches in future research will indisputably demonstrate the developmental potency of the larval cells. The flagellated cells of coeloblastulae pass through the intermediate stage of dedifferentiation in a settled larva before they differentiate to the cells of a juvenile sponge. 28 S. AMANO AND I. HORI >' "•• p Figure 17. Immature choanocytes (C) of the juvenile sponge of Lcucuxulcniii lu.\ti about 36 h after settlement. G: glutinous granule. P: pinucoderm. Scale bar = 2 /MTU. Figure 18. Choanoderm of the juvenile sponge of Leucosolenia la.\a about 72 h after settlement. A: autophagosome. C: collar. F: flagellum of a choanocyte. N: nucleus, P: pmacoderm. Scale bar = 2 /urn. The flagellated cells are arranged in a pseudostratified layer in a swimming larva. This cell layer is not a true epithelium because — like similar layers in other sponges — it lacks a basal lamina (Gaino et ai, 1985: Woollacott and Pinto, 1995). One flagellum emerges from the flat outer surface of the cell. In most sponges, however, larval flagella emerge from a pit at an outer surface (Amano and Hori. 1992, 1994: Woollacott, 1993: Woollacott and Pinto, 1995). After set- tlement, the flagellated cells soon loose their flagellum and dedifferentiate. The dedifferentiation is apparently triggered by the settlement of the larva. Autophagosomes play a central role in the intracellular processes of dedifferentia- tion. These autophagosomes are different in origin and structure from the phagosome-like granules in the basal cytoplasm of the larval flagellated cells: the latter were probably formed to take in nutritive materials by phagocy- tosis during ovogenesis (Fell. 1969): the former are formed de novo after settlement and are usually larger than the latter. Glutinous granules, fibrous yolk granules, and basal rootlets in various stages of digestion are observed in the METAMORPHOSIS OF CALCAREOUS SPONGE 29 5 /•"' 5^1 ^** OK %:J* .4:.:* ' • •.--, • M I »•• .X Figure 19. Basal apparatus ot a choanocyte in the juvenile sponge of Leucosolenia lti.\a about 72 h after settlement. C: collar. G: Golgi apparatus. Scale bar = 1 /urn. Figure 20. Fuzzy coat (F) of a choanocyte in the juvenile .sponge of Leucosolenia la\u about 72 h after settlement. Scale bar = 0.5 /am. Figure 21. Scleroblast in the juvenile sponge of Leucosolenia la.\a about 72 h after settlement. G glutinous granule. S: space occupied by a triradiate spicule. Scale bar = 2 jim. Figure 22. Two scleroblasts forming a triradiate spicule in the juvenile sponge of Leu, ; • « about 72 h after settlement. A multivesicular body (M) opens into the space occupied by a spicul. •: cell junction between scleroblasts. Scale bar = 2 /urn. autophagosomes of the inner cell mass of settled larvae. The choanocytes of juvenile sponges contain autophagosomes whose contents have been almost entirely digested. Bottle cells are described for the first time in this report. These large cells, like the cruciform cells of amphiblastulae (Duboscq and Tuzet. !9.^/; Amano and Hori, 1992), are located among the fl.igeliuted cells. The function of the bottle cells is not known; they apparently do not contribute to the development of the juvenile sponge and disappear during metamorphosis, as do the cruciform cells. Both cell 30 S. AMANO AND I. HORI Figure 23. Spicules surrounded by spongin fibers and by discontinuous cytoplasm in the juvenile sponge of Li'iicoxolenia luxti about 72 h after settlement. S: space occupied by a triradiate spicule. Scale bar = I /iin. Figure 24. A vacuolar cell in the mesohyl of the juvenile sponge of Leucosolenia laxa about 72 h after settlement. Scale bar = 2 jum. types may have similar functions, but the bottle cells cannot be named cruciform cells because the former cells are not arranged in each quadrant in a horizontal plane. Vesicular cells in the coeloblastula are also described for the first time. Their function is not known, but we know that they partic- ipate in metamorphosis, because they are found in the mesohyl of juvenile sponges. Free cells in a central cavity are probably somatic cells derived from the mother sponge. Similar maternally derived nutritive cells are found in the larvae of other sponges (Warburton, 1961; Gallissian and Vacelet, 1992). These cells disappear during metamorpho- sis. Juvenile sponge formation After settlement, a coeloblastula becomes a mass of ded- ifferentiated cells on the substratum. The first cells to dif- ferentiate are pinacocytes, which are apparently formed from cells at the surface of the cell mass. The pinacocytes are flattened, except at their nuclear region. As a juvenile sponge becomes larger, the cytoplasm of the pinacocytes becomes thinner. We found no ostia in the juvenile sponge; however, after osculum formation, seawater can flow into the gastral cavity through the spaces between the cells and the small holes of the thin cytoplasm. Scleroblasts are the next cells to differentiate in the cell mass settled on the substratum. Cells beneath the pinaco- derm become scleroblasts. These cells form triradiates, a characteristic spicule of calcareous sponges, in the mesohyl of juvenile sponges. The scleroblasts always enclose a large space delimited by a membrane. This space had been occu- pied by a calcareous spicule that was dissolved in the fixative. Similarly, siliceous spicules are formed in a space surrounded by a membrane called the silicalemma (Simpson and Vaccaro, 1974). An axial filament is essential for sili- ceous spicule formation (Garrone. 1969; Cha et «/., 1999), but the calcareous triradiate is formed without the aid of an axial filament (Ilan et ul., 1966; Jones, 1967; Amano and Hori, 1993). Here — as in other calcareous spicules — no axial filament was found in the space formerly occupied by a spicule. The scleroblasts usually have a small number of multivesicular bodies. We showed that the multivesicular bodies fuse with the membrane around the space and open into it. The multivesicular bodies probably supply calcium and membrane components for spicule formation. Similarly. in the metamorphosis of amphiblastulae, a number of small vesicles open into the space for spicule formation (Amano and Hori, 1993). Scleroblasts that form siliceous spicules also have a number of vesicles in the cytoplasm (Simpson and Vaccaro, 1974). Although the spicules are formed within the scleroblasts, the finished spicules free of cells are usually covered by spongin filaments (Jones, 1967; Gar- rone. 1985). Lastly, the cells of an inner cell mass differentiate to choanocytes. In the inner cell mass, a space develops de novo among the cells and becomes larger as metamorphosis progresses. A layer of immature choanocytes, not yet pos- sessing a flagellum or collar, encircles the large space, which is a gastral cavity. The gastral cavity is different in METAMORPHOSIS OF CALCAREOUS SPONGE 31 origin and structure from the central cavity of coeloblastu- lae, which is a retained blastocoel (Borojevic, 1969). As the gastral cavity develops, the choanocytes are flattened except at their thick central region. Then one flagellum emerges from the center of the central region and a collar of mi- crovilli develops around the thick central region. The basal body of the flagellum lacks rootlets and a foot: other sponges also lack these structures (Amano and Hori. 1993. 1996). On the cell membrane we found a fuzzy coat encir- cled by a collar; a similar fuzzy coat is found in demo- sponges (Bergquist and Green, 1977; Lethias et al, 1983). About 72 h after settlement, an osculum opens at the up- permost part of the metamorph and water flow begins. Thus an asconoid juvenile sponge, an olynthus, is formed. Coeloblastiilae of other calcinean sponges While the coeloblastulae of L. laxti are swimming, the flagellated cells do not migrate into a central cavity and the cells do not lose their flagellum. When the larvae adhere to the substratum for settlement, their flagella still continue active beating. In the coeloblastulae of other calcinean sponges, however, the larval flagellated cells have been reported to migrate into a central cavity while the larvae are swimming (Minchin, 1896; Borojevic, 1969). The reason for this discrepancy is not known. Time to initiate the dedifferentiation of larval flagellated cells may differ among calcinean sponges. A more likely explanation is that bottle cells and vesicular cells were erroneously judged to be the flagellated cells under migration into the central cavity. We showed that the bottle cells and vesicular cells of L. la.\a protrude extensively into the central cavity. Indeed, Borojevic (1969) noted that larval cells migrate into the blastocoel firstly from the posterior region of the coeloblas- tula, where vesicular cells are concentrated. If the resolution of larval images is not high enough, it seems to be difficult to know which is the case. In the calcinean sponge Clathrina reticitlnm. Borojevic (1969) described pigment granules in larval flagellated cells; however, these are apparently glutinous granules. Although the granules in question are indeed very electron- dense, they cannot be pigment granules, because coeloblas- tulae are colorless. We showed that the granules can be released from the cells and that their content is a sticky substance, so they may participate in larval settlement. We know that some of them are, however, digested intracellu- larly, because they were found in autophagosomes in the inner cell mass of settled larvae. Fates of lan'al flagellated cells After settlement, flagella and basal apparatus meet dif- ferent fates of degeneration depending on the types of larvae. In the parenchymellae of demosponges, flagella are withdrawn into the cell and decomposed in the cytoplasm after they are severed at the transitional region between a basal body and the proximal end of an axoneme (Boury- Esnault, 1976; Amano and Hori, 1996). In the coeloblastu- lae and amphiblastulae of calcareous sponges, however, the axonemes are probably shed soon after settlement (Amano and Hori, 1993). The basal bodies and rootlets, severed from the axoneme, are decomposed in the cytoplasm in the am- phiblastulae and parenchymellae (Amano and Hori, 1993, 1996). In the coeloblastula of L. la.\a, however, they are digested in autophagosomes. The developmental potency of larval flagellated cells differs greatly among the types of sponge larvae. In the coeloblastula of L. la.\a, the flagellated cells apparently become all three principal cell types of juvenile sponges. In amphiblastulae, another larval type of calcareous sponges, the flagellated cells become scleroblasts and choanocytes but not pinacocyles (Amano and Hori, 1993). The pinaco- cytes are derived from larval granular cells. In the paren- chymellae of demosponges, the flagellated cells are not multipotential and become only choanocytes (Borojevic, 1966; Amano and Hori, 1996). The parenchymellae usually consist of many more cell types than do coeloblastulae or amphiblastulae. and in such complex larvae, juvenile sponge cells are derived from distinctive larval cells. During the geologic eras after the Vendian age, each type of sponge larva evolved to metamorphose in different ways (Li et al., 1998). Acknowledgments We thank Mr. S. Tamura and Mr. M. Washio of the Asamushi Marine Biological Station for help in collecting sponges. Literature Cited Amano, S. 1986. Larval release in response to a light signal by the intertidul sponge Halichondria panicea. Biol. Bull. 171: 371-378. Amano, S. 1988. Morning release of larvae controlled by the light in an intertidal sponge, Callyspongia rarnosa. Biol. Bull 175: 181-184. Amano. S., and I. Hori. 1992. Metamorphosis of calcareous sponges I. Ultrastructure of free-swimming larvae, lin-ertehr. Reprod. Dev. 21: 81-90. Amano, S., and I. Hori. 1993. Metamorphosis of calcareous sponges II. Cell rearrangement and differentiation in metamorphosis. Invertebr. Reprod. Dev. 24: 13-26. Amano, S.. and I. Hori. 1994. Metamorphosis ol a demosponge I. Cells and structure of swimnun- larva. Invertebr. Ri'/imd. Dev. 25: 193-204. Amano, S., and I. Hori. !"->6. T,,.i: -.iii'erentiation of larval flagellated cells to choanocytes in the n , tiosis ut the demosponge Huli- clona pcrimillis. Kiol. Bull. 19(1: 161-172. Bergquist, P. R., and K. i.Uisumv. 1986. Developmental potential of ciliated celK <". eractinomorph -pongc larvae. E\p. Biol. 45: 111-122. BergquisI, P. R.. and ('. R. Green. 1977. An ultrastructural study of settlement and nu-Kiincipliosis in sponge larvae. Call. Biol. Mar. 18: 289-302. Borojevic, R. 1966. Etude experimentale de la differenciation des eel- 32 S. AMANO AND I. HORI lules de 1'eponge au cours de son developpement. Dc\: Biol. 14: 130-154. Borojevic, R. 1969. Etude du developpement el de la differenciation cellulaire d'eponges calcaires calcineennes (generes Clmhrina et As- candra). Ann. Embiyol. Morphog. 2: 15-36. Borojevic, R., and C. Levi. 1965. Morphogenese experimental d'une eponge a partir de cellules de la larve nageante dissociee. Z. Zellfurscli. Mikroxk. Amu. 68: 57-69. Boury-Esnault, N. 1976. Ultrastrueture de la larve parenchymella d'Haniigera hamigera (Schmidt) (Demosponge, Poecilosclerida). Origine des cellules grises. Cull. Biol. Mar. 17: 9-20. Brusca. R. C., and G. J. Brusca. 1990. Invertebrates. Sinauer Associ- ates. Sunderland, MA. Cha, J. N., K. Shimizu, Y. Zhou, S. C. Christiansen, B. F. Chmelka, G. D. Stucky, and D. E. Morse. 1999. Silicatein filaments and subunits from a marine sponge direct the polymerization of silica and silicones in vitro. Proc. Natl. Acad. Sci. 96: 361-365. Duboscq. O., and O. Tuzet. 1937. L'ovogenese, la fecondation et les premiers stades du developpement des eponges calcaires. Arch. Zoo/. Exp. Gen. 79: 157-316. Fell, P. E. 1969. The involvement of nurse cells in oogenesis and embryonic development in the marine sponge. Haliclona ecbasis. J. Morphol. 127: 133-150. Gaino, E., B. Burlando, M. A. Sabatini, and P. Buffa. 1985. Cytoskel- eton and morphology of dissociated sponge cells. A whole-mount and scanning electron microscopic study. Eur. J. Cell Biol. 39: 328-332. Gallissian, M.-F., and J. Vacelet. 1992. Ultrastrueture of the oocyte and embryo of the calcified sponge. Petrobionu massiluinn (Porifera, Cal- carea). Zoomorphology 112: 133-141. Garrone, R. 1969. Collagene. spongine et squelette Mineral chez 1'ep- onge Haliclona msea (O.S.) (Demosponge. Haploscleride). J. Microsc. 8: 581-598. Garrone, R. 1985. The collagen of the Porifera. Pp. 157-175 in Biolog\ of Invertebrale anil Lower Vertebrate Collagens. A. Bairati and R. Garrone. eds. Plenum Press. New York. Ilan, M., J. Aizenberg, and O. Gilor. 1966. Dynamics and growth patterns of calcareous sponge spicules. Proc. R. Soc. Land. B 263: 133-139. Jones. W. C. 1967. Sheath and axial filament of calcareous sponge spicules. Nature 214: 365-368. Kaye, H. R., and H. M. Reiswig. 1991. Sexual reproduction in four Caribbean commercial sponges. III. Larval behaviour, settlement and metamorphosis. Invertebr. Reprod. Dev. 19: 25-35. Lethias, C., R. Garrone, and M. Mazzorana. 1983. Fine structure of sponge cell membranes: comparative study with freeze-fracture and conventional thin section methods. Tissue Cell 15: 523-535. Li, C.-W., J.-Y. Chen, and T.-E. Hua. 1998. Precambrian sponges with cellular structures. Science 279: 879-882. Maldonado, M., and C. M. Young. 1996. Effects of physical factors on larval behavior, settlement and recruitment of four tropical demo- sponges. Mar. Ecol. Prog. Ser. 138: 169-180. Minchin. M. A. 1896. Note on the larva and the postlarval development of Leucosolenia variabilis, H. sp., with remarks on the development of other Asconidae. Proc. R. Sue. Land. 60: 42-52. Misevic, G. N., V. Schlup, and M. M. Burger. 1990. Larval metamor- phosis of Microciona prolifera: evidence against the reversal of layers. Pp. 182-187 in New Perspectives in Sponge Biology. K. Riitzler. ed. Smilhonian Institution Press. Washington, DC. Miiller, W. E. G. 1997. Origin of metazoan adhesion molecules and adhesion receptors as deduced from cDNA analysis in the marine sponge Geodia c\d<>nium: a review. Cell Tissue Res. 289: 383-395. Simpson, T. L. 1984. The Cell Biology of Sponges. Springer- Verlag, New York. Simpson, T. L., and C. A. Vaccaro. 1974. An ultrastructural study of silica deposition in the freshwater sponge Spongillu lacustris. J. Ultra- smut R^. 47: 296-309. Spurr, A. R. 1969. A low-viscosity epoxy resin embedding medium for electron microscopy. J. Ultrastruct. Res. 26: 31-43. Tuzet, O. 1947. L'ovogenese et la fecondation de 1'eponge calcaire Leucosolenia (Clathrina) coriacea Mont, et de I'eponge siliceuse Re- niera e/egans Bow. Arch. Zoo/. Exp. Gen. 85: 127-148. Warburton. F. E. 1961. Inclusion of parental somatic cells in sponge larvae. Nature 191: 1317. \Voollacott, R. M. 1993. Structure and swimming behavior of the larva of Huliclonu tuhifera (Porifera: Demospongiae). J. Morphol. 218: 301-321. Woollacott, R. M., and R. L. Pinto. 1995. Flagellar basal apparatus and its utility in phylogenetic analysis of the Porifera. J. Morphol. 226: 247-265. Reference: Biol. Bull. 200: 33-50. (Februan 20(1 1 I Interspecific Relationships Between Egg Size and the Level of Parental Investment per Offspring in Echinoderms L. R. McEDWARD* AND K. H. MORGAN Department of Zoologv, Universitv of Florida, Gainesville, Florida 3261 1 Abstract. The relationship between the size of an egg and its energy content was analyzed using published data for 47 species of echinoderms. Scaling relationships were evalu- ated for all species, as well as for subsets of the species, based on mode of development. Regressions were calcu- lated using linear, power function, full allometric, and sec- ond-order polynomial models. The full allometric model is preferred because it is relatively simple and the most gen- eral. Among these species of echinoderms, larger eggs contain more energy. Egg energy content scales isometri- cally across a wide range of egg sizes both among and within different modes of development. The only exception is among species with feeding larval development, where there does not seem to be a clear scaling relationship. In most cases, the regressions were statistically significant and explained a very large proportion of the variance in energy content. However, there were wide confidence intervals around the estimated regression parameters. In all cases, the predictive power of the regression was poor, requiring large differences in egg size to yield significantly different pre- dictions of energy content. Consequently, egg size is of limited value for the quantitative prediction of egg energy content and should be used with caution in life-history studies. Introduction A major goal of ecological research is to explain the evolution of life histories, i.e., how natural selection mod- ifies reproduction and development to yield the patterns that are observed in nature. Quantitative, theoretical models Received 27 December 1999: accepted 27 October 2(). *To whom correspondence should be addressed. mcedward@zoo.ufl.edu E-mail: have been the primary tool used to explore hypotheses on the effects of selection on life-history traits. As a result, theoretical developments became fairly sophisticated and are now considerably ahead of the empirical database. In addition to the emphasis on modeling, the difficulty of obtaining the appropriate data is also responsible for the discrepancy between theory and data. Pianka ( 1976, p. 782) argued that "A great deal of ingenuity and effort will be required to design and execute research to test this elegant body of theory." Beyond that, Stearns ( 1977) contended that there might well be fundamental limitations on the kinds of information that ecologists can obtain about life-history evolution. The relationship between parental investment per off- spring and offspring fitness is one of the central tenets of life-history theory that has been particularly difficult to evaluate empirically. Given that the total resources that the parent devotes to reproduction are limited, there should be an inverse relationship between the investment made in each offspring and the number of offspring that can be produced. Further, it has been assumed that as parents allocate more material and energy to each individual off- spring, the fitness of the offspring increases because of an increase in the quality of the young U'.t;.. Smith and Fretwell, 1974). This is a seemingly straightforward and intuitively reasonable assumption. Wh\ has it been so dif- ficult to test empirically? An important obstacle to the critical evaluation of the relationship between parental investment and offspring fit- ness is the difficulty in drlming parental investment in precise and measurable terms. What exactly are the re- sources that a parent de\otes to its offspring? How can we effectively measure the variation in the level of investment that exists among offspring of a given parent or among the offspring of different parents? Parental care is common 33 34 L. R. McEDWARD AND K, H MORGAN among animals, and it is extremely difficult to identify components of care that represent quantifiable resources. Time and energy budgets of parents are particularly difficult to measure precisely and even more difficult to quantify in terms of the actual resources that were spent on the off- spring. The problem of parental care can be avoided by studying free-spawning benthic marine organisms. Although some marine organisms do provide protection and care for their young, a very large number are free-spawners. and do not. The latter simply release gametes into the surrounding sea- water; fertilization and development occur independently of the parents. In this case, parental investment can reasonably be defined as the material and energy contained in the egg, because that is the only contribution that the mother makes to the young. The paternal contribution consists of the sperm pronucleus and functional centrioles, but this is quan- titatively insignificant relative to the contents of the egg. Given the suitability of free-spawning organisms for measurement and analysis of parental investment, have ma- rine ecologists succeeded in evaluating the relationship be- tween investment and offspring fitness? Unfortunately, no. We have been remarkably unsuccessful in our attempts to acquire data necessary to test and further develop this aspect of life-history theory. This is in spite of a tremendous increase in the interest and research effort in "larval ecol- ogy" in recent years. The failure is due in large part to the fact that marine ecologists have not measured parental investment (as egg energy content) directly; rather they have relied on mea- surements of egg size as an index of investment. In fact, this is a central assumption of the theory that underlies most of the quantitative models of life-history evolution in marine benthic invertebrates (e.g.. Vance, 1973; Christensen and Fenchel, 1979; Pechenik, 1979; Perron and Carrier, 1981; Grant. 1983; Emlet et (3) Where Yh represents the I - « confidence limits for the prediction generated from the new measurement of the explanatory variable (X,,). MSE is the residual mean square from the regression ANOVA. and t is the Student's t sta- tistic with n — 2 degrees of freedom. X B xf Figure 1. Evaluation of predictive power of a regression using overlap of 95% confidence intervals. Bold, solid diagonal line = least squares regression; bold, dashed diagonal lines = 95% confidence belts for regres- sion. .YA. XB. Xc = additional measurements of egg size for which predicted energy contents ( f'A, YB. Yc) are calculated using the regression equalion (solid vertical and horizontal lines). The 95% confidence intervals for predicted energy contents are obtained from the 95% confidence belts around the regression (dashed horizontal lines). Overlap of the 95% con- fidence intervals indicates that two predicted egg energy contents are not significantly different. In this example. >', and Yc are significantly differ- ent ( )\ + CI < Yc — C7). but neither is significantly different from YB. To evaluate the predictive power of a regression, we asked: what is the minimum difference in egg size that yields significantly different predictions of egg content? Two predictions of egg energy content were considered to be significantly different (i.e., statistically distinguishable) only if the confidence intervals around them did not overlap (Fig. 1 ). Starting at the midpoint of the observed range of egg volumes, predictions were calculated for pairs of egg sizes, each pair progressively farther apart, until the confi- dence intervals around the two predictions did not overlap. or until the entire range of egg sizes had been evaluated. The resolution of this analysis was 0.1% of the observed egg volume range. The 95% confidence belts around the regression represent the confidence intervals around the predicted egg energy content for all values of the explanatory variable and are needed for our evaluation of predictive power. Both the linear and nonlinear regression analyses generated a table of the single-prediction confidence limits for the actual egg volumes in the data set. From these, we generated equations for the upper and lower confidence belts by fitting the same regression model to the upper and lower (respectively) values of the prediction confidence limits. Given confidence belt equations, it was possible to calculate the confidence limits around predicted egg energy content for any egg EGG SIZE-ENERGY RELATIONSHIP 41 volume between the minimum and maximum observed val- ues for linear and allometric regression models. The predictive power of a regression depends on the width of the confidence belts around the regression and the slope of the regression. Narrow confidence belts result in greater predictive power because nonoverlapping predic- tions occur with smaller differences in egg volume (Fig. 1 ). Likewise, steeper slopes yield greater predictive power be- cause the absolute difference between predicted values is greater for a given difference in egg size than with a regression that has a shallow slope. Hence, predictive power does not have a simple relationship with other regression statistics, such as the tit (proportion of variance in the egg energy that is explained by egg size). This approach was not feasible with the power function regression model. The reason for this is that the power function regression must pass through the origin, which influences the way that predictive power is estimated. Like- wise, the estimation of predictive power with the polyno- mial model can be misleading because of strong curvature in the regressions (e.g.. Fig. 2D). The problem with using overlap of confidence belts with strongly nonlinear relation- ships is that as the slope of the relationship changes, the overlap of confidence belts changes, and therefore the pre- dictive power changes across the range of egg sizes. Eval- uations of predictive power would be valid only over small regions of the egg size range. Rather than report what would seem to be anomalous values for predictions by the power and polynomial functions, we report predictive values only for the linear and full allometric models. Comparison of means ANOVAs and Student-Newman-Kuels multiple range tests were used to evaluate differences among taxonornic classes or among developmental modes in mean egg volume or mean egg energy content. All tests were conducted at the 5% significance level. Results Brooders make eggs (6.348 ju.1) that are significantly larger than the eggs of species with planktotrophic (0.0016 /LI!) or pelagic lecithotrophic development (0.417 jul). Brooders also provision eggs with significantly more energy (66.910 J) than do planktotrophs (0.0087 J) or pelagic lecithotrophs (4.363 J). However, there are no significant differences among the taxonornic classes (excluding Cri- noidea with only a single species) in egg size (P = 0. 165) or energy content (P = 0.124). Scaling relationships for all species Among these 47 species of echinoderms. both egg size and energy content vary across five orders of magnitude. 150 100 50 ~ 200 1 150 O 100 te I 10 15 Volume ml) 5 2°o| C I 150 | I 50 Volumelull Volumel^D Figure 2. Plots of egg volume and energy content for all 47 species of echinoderms. Panels show different regression models tit to the data and the 95% confidence belts around the regression line. A. linear regression; B, power function: C. full allometric model: D. polynomial model. There is a strong trend for larger eggs to contain more energy (Fig. 2A). The linear model indicates that the inter- cept is not significantly different from zero and that egg energy (J) is approximately 8.6-fold greater than the egg volume (jul) (see Table 3 for parameter estimates and re- gression statistics for all models). The linear regression explains most of the variation in egg content (fit = 84%), but eggs must differ by more than 8 /nl in volume («=35% of the egg size range) to allow the confident prediction that they contain different amounts of energy. This does not provide a useful means of predicting content from measure- ments of volume. The power function (Fig. 2B) yields a better fit to the data than the linear model, explaining 91% of the variance in egg energy content. The exponent of the power function (0.70) indicates that egg energy content scales with negative allometry relative to egg volume. This means that energy content does not change in proportion to egg volume and that a linear model is inappropriate for Figure 3. Standardized residual plots lor all 47 species of echino- derms. Panels show residuals tor ditk'icnt iVLiivssion models fit to the data. A. linear regression; B, power function: C, full allometric model; D. polynomial model. 42 L. R. McEDWARD AND K. H. MORGAN Table 3 Regression parameters for all 47 species of echinoderms Model a b c k Fit A.Y predict Y Linear 2.607 8.610 _ 0.838 8.11 /ul [-2.179-7.393] [7.474-9.746] [35%] Power — 20.381 0.702 0.907 — [14.331-26.432] [0.592-0.812] Allometric -3.141 23.128 0.666 0.911 8.36 jul [-8.194-1.912] [14.925-31.331] [0.543-0.789] [36%] Polynomial -2.201 19.478 -0.561 0.984 — [-3.922—0.481] [18.229-20.727] [-0.622—0.500] Regression parameters are illustrated for each model in Table 2. Fit is the proportion of the total variance in egg energy content that is explained by the regression against egg volume. A.Y predict Y gives the minimum difference in egg volume needed to yield significantly different predictions of egg energy content. these data. The full allometric model (Fig. 2C) yields a negligibly better fit than the power function, with nearly the same parameter values. The allometric model has weak predictive power; the minimum difference in egg size nec- essary to yield significantly different predictions is more than 8 /nl. The polynomial regression provides the best fit to the data among the four models tested, with a fit of 98%. It shows that the data are strongly curvilinear (Fig. 2D). Do any of these models fit well enough to justify con- cluding that we have described the pattern of scaling be- tween egg size and content? In one sense, yes. using all of the data with fairly simple, general models that were ex- plicitly chosen beforehand yields an objective analysis. Be- sides, the fit in terms of the explained variance is quite good for all models tested (84%-98%). However, the confidence intervals for the regression parameters are broad in all cases, and this cautions against putting too much confidence in the described scaling relationship. Can the fit of any of the ISO 125 100 150 ~ 125 | 100 8 75 2 125 S 100 Figure 4. Plots of egg volume and energy content for 46 species of echinoderms. excluding Notasterius annala. Panels show different regres- sion models fit to the data and the 95% confidence belts around the regression line. A, linear regression; B. power function; C. full allometric model; D. polynomial model. models be improved substantially? Examination of the data (Fig. 2), the standardized residuals (Fig. 3). and especially the Hat Diagonals shows that there is one point which strongly influences the estimated regression parameters. On the basis of this finding, the asteroid Notasterias armata was removed from the data set, and the regressions were recal- culated. The most striking result of removing this species is that the relationship between egg size and energy content be- comes linear (Fig. 4). With the modified data set, the fits (% variance explained) are exceptionally high (99%), the con- fidence intervals for all of the regression parameters are quite narrow, and the four models generate very similar parameter values (e.g.. the intercept is zero; the slope is in the low teens; and the allometric exponent is only slightly greater than one. indicating nearly proportional or isometric scaling of energy content with volume) (Table 4). In addi- tion, the standardized residuals are more evenly distributed (Fig. 5), and the 95% confidence belts are much narrower. This suggests that Notasterias armata is an anomalous Figure 5. Standardized residual plots for 46 species of echinoderms. excluding Notasterias armata. Panels show residuals for different regres- sion models fit to the data. A. linear regression; B. power function; C. full allometric model; D, polynomial model. EGG SIZE-ENERGY RELATIONSHIP Table 4 parameters for 46 species <>/ ' ccliinodenns, excluding Notastenas urniuta 43 Model a /' C k Fit \X predict Y Linear -0.640 13.755 0.995 0.75 ju.1 [-1.360-O.OSOI [13.470-14.039] [7%] Power — 1 1 .582 1.073 0.997 [10.594-12.570] [1.036-1.109] Allometnc 0.094 1 1 .444 1 .076 0.997 0.69 Ml [-0.646-0.833] [10.283-12.705] [1.032-1.119] [6%] Polynomial -0.038 11.721 0.186 — 0.997 [-0.728-0.652] [10.673-12.769] [0.093-0.280] Regression parameters are illustrated for each model in Table 2. Fit is the proportion of the total variance in egg energy content that is explained by the regression against egg volume. A.V predict >' gives the minimum difference in egg volume needed to yield significantly different predictions of egg energy content. species, in terms of egg size, energy content, or both. All subsequent analyses were conducted excluding this species. The biological basis of the anomalous egg characteristics is not known (see below). How good is the predictive power of these regressions'? Both the linear and the allometric models indicate that significantly different predictions of content require differ- ences in egg volume of 0.69-0.75 ;u,l, which comprise only about 6%-7% of the egg size range. This is substantially less than in the previous analyses, but just how useful is it'.' Given that the 35 species with the smallest eggs all lie within a range of volumes that spans only 0.52 pil. then it is clear that we are only able to conclude with confidence that very large eggs of species with lecithotrophic larval devel- opment (benthic brooding or pelagic) contain more energy than very small eggs of species with planktotrophic (feed- ing) larval development. These regressions, even given ex- tremely good fit with the data, have poor predictive power. Scaling relationships within developmental patterns Is the predictive power of the relationship improved by restricting the analysis to just those species with particular modes of development? In general, egg size and content of planktotrophic and lecithotrophic species are very different (Fig. 2). All species with planktotrophic larval development produce eggs that contain less energy (<4.64 X 10~2 J • egg"') than eggs from all species with pelagic or benthic (brooded) lecithotrophic development (>82.5 X 10~2 J • egg ), except for the crinoid Florometra serratissiina (4.55 X 10~2 J • egg~') (Table 1 ). There is overlap in size or content among a few species that represent the extremes within planktotrophy and lecithotrophy. The echinoid C/v- peaster rosaceus has larvae that are lecithotrophic. in that they do not require exogenous food for larval development and metamorphosis (Emlet. 1986). However, unlike all other lecithotrophic larvae, they have functional feeding structures and can acquire and utilize exogenous, paniculate food. This species, being a facultative feeder, has a mixture of traits characteristic of planktotrophic and nonfeeding lecithotrophic larvae. It has a larger egg (1.15 X 10~2 p.]) than any obligately planktotrophic species and, with only a single exception, it has greater egg energy (2.06 x 10~2 J • egg"1) than any obligate planktotroph. Only the echinoid Encope mic/ielini (4.64 X 10 2 } • egg"') has greater egg energy than C. rosaceus. Encope michelini is an obligate planktotroph, but it can complete most of larval develop- ment using endogenous reserves and has only a minimal need for exogenous food (Eckert. 1995). Clypeaster rosaceus produces eggs that are larger than one of the pelagic lecithotrophs, the crinoid Florometra serratissima. The eggs of Florometra are unusually small (7.42 X 10~3 ju.1. 4.55 X 10~2 J • egg~') for nonfeeding lecithotrophs, containing 8.6-fold less volume and 18.1-fold less energy than the next largest egg (Cucumaria miniata, 6.40 X 10"2 ju,l. 8.25 X I0~' J • egg"1). If Clypeaster and Florometra are discounted, then the distinction between the eggs of planktotrophs and lecithotrophs is very striking: there is an 12.5-fold difference in egg volume and an 17.8-fold difference in energy content between the largest egg from planktotrophic species (Encope michelini) com- pared to the smallest egg found among lecithotrophs (Cu- cumaria miniata) (Table 1). This difference in egg charac- teristics is consistent with the nutritional strategies of planktotrophy and lecithotrophy. Obligately planktotrophic larvae must supplement the material provided in the egg by feeding on planktonic particles. Nonfeeding lecithotrophs cannot capture or ingest particles and must rely on nutri- tional reserves provided in the egg by the parent, possibly supplemented by uptake of dissolved organic matter from the surrounding seawater. The eggs of nearly all brooders are larger and contain more energy than the eggs of pelagic lecithotrophs (Table 1 ). This is surprising because, whereas pelagic lecithotrophic larvae are independent of the parent throughout development and are incapable of utilizing 44 L. R. McEDWARD AND K. H. MORGAN -OT2 0.4 0.6 C Volume (Ml) -0.2 0.4 0.6 ( Volume (ull r 12-5 I 10 5 7.5 g 5 I 2.5 -2.5 Figure 6. Plots of egg volume and energy content for the 38 species of echinoderms with planktotrophic or lecithotrophic development. Panels show different regression models fit to the data and the 95% confidence belts around the regression line. A. linear regression; B, power function; C, full allometric model: D. polynomial model. planktonic food particles, brooders retain the offspring on or in the body of the parent and therefore have the potential to provide nutrition to the developing young. Scaling relationships among free-spawners Among free-spawning species, linear regression explains 84% of the variance in egg energy content in this sample of 38 species (Fig. 6). The best-fit linear regression passes through the origin (the intercept confidence interval in- cludes zero) and has a slope of 9.87 (Table 5). The power function and the allometric models each explain 88% of the variance. Their slopes (9.64) are not different from the slope of the linear regression, and their scaling exponents (0.87) are not significantly different from 1. This indicates that energy content increases roughly in proportion to egg vol- ume. The predictive power of these scaling relationships does not improve substantially by restricting the analysis to de- velopment modes without offspring retention. Eggs that differ by 0.56-0.62 /xl in volume have different predicted energy contents. Note that this is still more than half of the entire range of egg sizes in these two modes of develop- ment. How good is this? Volume is the most useful measure of egg size for comparison with content, but it is not easy to visualize. Biologists routinely report egg size as diameter because it is easy to measure and it is what one sees in an optical section under the microscope. Obviously, the diam- eter difference required to generate a difference of 0.56 /xl in volume will depend on the size range of eggs considered. To illustrate the point, consider an egg with a diameter of 1000 /urn; it has a volume of 0.52 /xl. Eggs must be larger than 1250 ju,m in diameter in order to yield different pre- dictions of content. However the egg of 1000 /xm has a predicted content that is indistinguishable from the contents of all eggs that are smaller. In effect, all that can be con- cluded from the regression of size and content is that most lecithotrophs produce eggs with more energy in them than do planktotrophs. Seal ing relationships among species with nonfeeding development The pelagic lecithotrophs and brooders together comprise the species with nonfeeding modes of development. Among these 24 species, all four regression models yield excellent fit, narrow parameter confidence intervals, and reasonably good predictive power (Fig. 7, Table 6). All of the models account for more than 99% of the variance in energy con- tent. Both the power function (1.07) and allometric (1.08) model exponents are only slightly different from 1, indicat- ing nearly proportional scaling of content with volume. The predictive power is good in that the minimum difference in egg volume needed for significantly different predictions of content is only 1 .0 ju.1. 9% of the egg volume range in this sample. But as can be seen from the plot (Fig. 7), this difference allows predictions only between the lecithotrophs Table 5 Regression parameters for the 38 species of echinoderms with planktotrnphic or lecithotrophic development Model a /> c k Fit A.V predict }' Linear 0.099 9.870 0.839 0.56 n\ [-0.400-0.599] [8.407-11.333] [55%] Power — 9.636 0.871 0.882 — [8.167-11.105] [0.610-1. 132] Allometric -0.012 9.644 0.868 0.882 0.62 Ml [-0.592-0.567] [8.097-11.190] [0.569-1.167] [61%] Polynomial 0.066 10.688 -0.957 0.880 — [-0.476-0.609] [5.620-15.756] [-6.627-1.712] Regression parameters arc illustrated for each model in Table 2. Fit is the proportion of the total variance in egg energy content that is explained by the regression against egg volume. AX predict >' gives the minimum difference in egg volume needed to yield significantly different predictions of egg energy content. EGG SIZE-ENERGY RELATIONSHIP 45 150 125 100 0.0 0.0 0.0 0.0 0. 00: 0.006 0.01 Volume (^I) 0.05 0.04 0.03 0.02 o.oi 0.002 ~ 0.006 Volume! nl ~ 150 r 125 J100 75 Volume ml) Volume lull Figure 7. Plots of egg volume and energy content for the 24 species of echinoderms with lecithotropic or brooded development, excluding Notas- lerias armata. Panels show different regression models fit to the data and the 95% confidence belts around the regression line. A. linear regression; B. power function; C. full allometric model; D, polynomial model. *: 0.05 c 0-04 5 0.03 p 0.02 rS 0.01 0.006 0.01 Volume (nit Figure 8. Plots of egg volume and energy content for the 22 species of echinoderms with feeding larvae, including Clypeaster roseaceus. Panels show different regression models fit to the data and the 95% confidence belts around the regression line. A. linear regression; B, power function; C. full allometric model; D. polynomial model. as a group and between the four brooders with the largest eggs. Scaling relationships for species with planktotrophic lan'al development Planktotrophic species produce small eggs that develop into feeding larvae. The echinoid Clypeaster rosaecus has, for all previous analyses, been considered to be lecithotro- phic because the energy content of the egg is sufficient to support complete larval development to metamorphosis (Emlet. 1986). However, since it has feeding larvae, it could be considered a facultative planktotroph. We analyzed the scaling relationships among planktotrophs with C. rosaceits included and excluded from the data. When Clypeaster is included there are 22 species with feeding larvae (Table 1). The linear regression model yields a relatively poor fit to the data, explaining only 37% of the variance in content (Fig. 8, Table 7). The power function and the allometric models explain 70%-72% of the variance, and the polynomial re- gression generates the best fit (77%). For all of these models the parameter confidence intervals are very broad, and none of the models have any predictive power (Table 7). Even the extremes in the range of egg sizes do not yield significantly different predictions of content. Excluding Clypeaster im- proves the fit of the regressions, but the parameter confi- dence intervals remain very large. Based on Hat Diagonals and residuals, we removed three additional species (Encope aberrans, Encope michelini, Parastichopus califomicus) and then two more species from the planktotrophic dataset (Luidia clathrata, Odontaster valulus). These changes did not result in better estimates of the regression parameters (CIs). Furthermore, the nonlinear models were drastically different for each subset of the data. The allometric model exponent varied from 0.335 (negative allometry) to 1.97 (positive allometry) as species were included or excluded from the analysis. However, the allometric exponent was not significantly different from isometric ( 1 .0) for any of the Table 6 Regression parameters for the 24 species of echinoderms with lecilholrophic or brooded development, excluding Notasterias armata Model a b c A Fit \X predict Y Linear -1.318 13.848 _ 0.995 1.00 nl [-2.735-0.098] [13,435-14.261] [9%] Power — 11.582 1.073 0.997 [10.180-12.985] (1.021-1.124! Allometric 0.326 11.279 1 .083 0.997 1.03 fi\ [-l.hl4-2.265] [8.998-13.5601 [1.002-1.164] [9%] Polynomial -0.088 11.759 0.183 0.997 (-1.745-1.5681 (9.951-13.568] [0.028-0.339] Regression parameters are illustrated for each model in Table 2. Fit is the proportion of the total variance in egg energy content that is explained by the regression against egg volume. AA" predict Y gives the minimum difference in egg volume needed to yield significantly different predictions of egg energy content. 46 L. R. McEDWARD AND K. H. MORGAN Table 7 tur the 22 species of echinodenns with feeding larvae, including Clypeaster roseaceus Model b c k Fit AX predict Y Linear 0.004 2.482 — — 0.367 [0.000-0.009[ 10.960-4.004] [>100%] Power 0.403 0.583 0.703 [-0.340-1.145] [0.256-0.910] Allometric -0.013 0.151 0.289 0.717 [-0.079-0.052] [-0.200-0.502] [-0.510-1.089] [>100%] Polynomial -0.001 7.600 -481.456 0.765 [-0.007-0.004] [3.719-11.480] [-825.424—137.488] Regression parameters are illustrated for each model in Table 2. Fit is the proportion of the total variance in egg energy content that is explained by the regression against egg volume. AX predict Y gives the minimum difference in egg volume needed to yield significantly different predictions of egg energy content. subsets of planktotrophic species that we analyzed. Like- wise, the coefficient of the second-order term of the poly- nomial regression varied from positive (upward curvature) to negative (downward curvature), but the 95% confidence interval for that parameter ranged from positive to negative for most analyses. The behavior of these models indicates that there is not a clear scaling trend among species with feeding larvae. This is consistent with the growing appre- ciation that planktotrophic species utilize a wide range of nutritional strategies and are characterized by different life history trade-offs across those strategies (Herrera et ill.. 1996; McEdward and Janies, 1997; McEdward, 1997). At this point, with existing data, it is not possible to character- ize the scaling relationships between egg size and energy content, and it is certainly impossible to use regression analysis to predict energy content from measurements of egg volume. Data on additional planktotrophic species might provide a clear picture of how egg energy scales with volume. However, it is also possible that energetic diversity among planktotrophs is considerable and clear trends tor the entire suite of species with feeding larvae do not exist. Sen ling relationships among species with pelagic lecithotrophic development There are data for 17 species (Clypeaster included) of echinoderms that have pelagic lecithotrophic development (Table 1). Egg volumes range from 0.007 to 1.01 ju.1, and energy contents range between 0.021 and 13.7 J • egg"1 (Fig. 9). The linear model explains 73% of the variance in egg energy content and the nonlinear models explain 88% (Table 8). The scaling exponents are 0.87 and 0.90 respec- tively for the power and allometric models, suggesting negative allometry, but neither is statistically different from isometry. The predictive power of all of the models is low, requiring differences in egg size on the order of the entire observed range of sizes (Table 8). Scaling relationships among brooders The eggs of nearly all brooders are larger and contain more enerey than the eggs of pelagic lecithotrophs (Table 1 ). The only exception to this generalization is the brooding holothu- roid Cucnmaria cnnitu. which has small eggs (0.52 fil, 4.58 J, Table 1 ). This is surprising because brooders have the potential to provide nutrition to the developing young (e.g., Sewell and Qua, 1994; Byrne, 1996), whereas pelagic nonfeeding larvae are independent of the parent throughout development and are incapable of utilizing planktonic food particles. Is this a general result? Emlet et til. (1987) reported egg sizes, but not energy content, for nearly 200 species of asteroids and echinoids. They found that there was very little, if any, overlap in the egg size ranges of planktotrophs and lecithotrophs, but there was considerable overlap in the egg sizes produced by pelagic lecithotrophs and brooders (Emlet et al., 1987, p. 108). In spite of the overlap, the largest eggs produced by both asteroids and echinoids occurred in brooding species. These observations 0 -Z- -"0". 4 0.6 0.8 Volume (jil ) 0.2 ff.4 0.6 0.8 Volume (til) § 10 O% 4 0.6 Volume (nD ZtFTf oTe 0,8 Volume (pi) Figure 9. Plots of egg volume and energy content for the 17 species of echinoderms with lecithotrophic development, including Clypeaster rosea- ceus. Panels show different regression models fit to the data and the 95% confidence belts around the regression line. A. linear regression; B, power function; C. full allometric model; D, polynomial model. EGG SIZE-ENERGY RELATIONSHIP Table 8 Regression parameters for the 17 species of echinoderms with lecithotrophic development, including Clypeaster roseaceus 47 Model (/ b c k Fit AX predict Y Linear 0.421 9.385 0.727 0.97 n\ [-1.196-2.038] [6.217-12.554] [97%] Power — 9.631 0.869 0.882 — [7.237-12.025] [0.443-1.296] Allometric 0.137 9.524 0.900 0.882 — (-2.937-3.212] [9.580-13.091] [0.087-1.714] [>100%] Polynomial 0.510 8.722 0.659 0.882 — [-1.857-2.877) [-4.122-21.567] [-11.680-12.998] Regression parameters are illustrated for each model in Table 2. Fit is the proportion of the total variance in egg energy content that is explained by the regression against egg volume. AX predict Y gives the minimum difference in egg volume needed to yield significantly different predictions of egg energy content. suggest that brooding, with its potential for post-spawning parental investment, does not necessarily result in lower levels of parental investment by means of the egg. Rather, it seems to !" 175 150 I provide the opportunity to increase levels of parental invest- ment per offspring over those found in pelagic lecithotrophs. Possibly the functional requirements of pelagic life (e.g., buoy- ancy control, swimming, settlement) preclude very large eggs and offspring. Alternatively, parental protection of brooded young could reduce the risk of offspring mortality sufficiently to allow allocation of more resources to individual young. The eight species of brooders (Notasterias excluded) are fit well by both the linear and nonlinear models (Fig. 10, Table 9). More than 99% of the variance in content is explained by each regression model. However, parameter confidence intervals are still quite wide: for example, the scaling exponent of the allometric model (1.15) is not sig- nificantly different from 1. The predictive power is reason- ably good for both models, requiring egg size differences of 16%-17% of the observed range. Figure 10. Plots of egg volume and energy content for the 8 species of echinoderms with brooded development, excluding Notasterias armata. Panels show different regression models fit to the data and the 95% confidence belts around the regression line. A. linear regression; B, power function; C, full allometric model; D. polynomial model. Discussion Among echinoderms, larger eggs do contain more en- ergy. The general pattern that is suggested by this study is Table 9 Regression parameters for the 8 species of echinoderms with brooded development, without Notasterias armata Model a b c k Fit AX predict Y Linear - 1 .495 13.897 0.996 1.74 jul [-6.788-3.798] [13.018-14.775] [16%] Power — 12.152 1.053 0.998 — [9.111-15.192] [0.946-1.159] Allometric 4.414 9.277 1.153 0.998 1.86 /M! [-6.389-15.218] 12.102-16.453] [0.858-1.447] [17%] Polynomial 3.056 10.566 0.264 0.999 — [-5.483-11.595] [5.376-15.757] [-0.142-0.669] Regression parameters are illustrated for each model in Table 2. Fit is the proportion of the total variance in egg energy content that is explained by the regression against egg volume. AX predict Y gives the minimum difference in egg volume needed to yield significantly different predictions of egg energy content. 48 L. R. McEDWARD AND K. H. MORGAN that egg energy scales very nearly in direct proportion to egg volume across a remarkable range of egg sizes both within and among different modes of development. This seems to be a robust result that emerges from different analyses using different regression models and different methods and assumptions for calculating egg energy (see Jaeckle, 1995). Statistically significant linear and nonlinear regressions were obtained in nearly all of these interspecific analyses. This demonstrates that some, and often a large proportion, of the variance in content is related to egg size. However, in many cases there were wide confidence intervals around the estimated regression parameters, suggesting caution in in- terpretation of the parameters. Whether the precision of regression parameter estimates will improve with new data for additional species or whether new data will confirm an underlying diversity in the egg size-content relationship is not clear. We believe that the latter possibility is likely, especially for moderate increases in the number of species analyzed. Once a substantial number of new species are added to the dataset and the underlying diversity is ade- quately sampled, then additional data should lead to more precise estimates of scaling patterns. A statistically significant regression provides a means to predict content from a measurement of size. But a significant regression, even with a very good fit (% variance explained), does not ensure that predictions based on it will be useful or reliable. Although we have tried to bias the outcome by ex- cluding some species or restricting the analysis to subsets of the data, in all of the cases examined here the predictive power of the regressions was poor. We strongly advocate explicit tests of regression results, whether the regression is used to describe scaling patterns or to make predictions from future measure- ments. At a minimum, we advocate evaluation of parameter confidence intervals and some test of the predictive power of the regression. The analyses presented here clearly show how poor the predictive power of regressions can be. even when there is a good fit to the data. In some cases, the predictive power is poor because many of the data points are clus- tered at the low end of the egg size range. So that even if the minimum difference in egg size needed to yield significant predictions of content is a small percentage of the egg size range, the difference is quite large relative to the egg size differences among many of the species {e.g., nonfeeding development. Fig. 7. Table 6). In other cases, the predictive power is low because the regression slope is shallow relative to the width of the confidence bands and the range of egg sizes (e.g.. lecithotrophs. Fig. 9, Table 8). In still other cases, the poor predictive power is the result of a poor fit between the regression and the data (e.g., linear model for planktotrophs. Fig. 8A. Table 7) or a strongly curvilinear regression (e.g., allometric model for planktotrophs. Fig. 8C, Table 7). On the basis of our results in this study, we advocate use of the full allometric model for studies of scaling relation- ships. This model has the advantage of being simple and general (see also Emlet. 1989: Albrecht et study, are hard to interpret biologically. One asteroid species, Notusterias annata (a brooder), is strikingly different in egg size or energy content relative to all other species. It potentially exerts extraordinary influ- ence on the regression analyses. Given that these data are not the result of serious measurement error, it become an interesting problem to explain the basis for such different levels of egg provisioning in this Antarctic species. It has exceptionally large eggs (23.2 /ul " ' ). but the energy content is the same as that of Diplasterius meridionalis and Diplas- terias brucei (Table 1 ). These two species are also Antarctic asteroids that brood their young. Their eggs are 3 times greater in volume than those of any other species in our data set. but their eggs are only half the size of Notasterias eggs. It is not obvious what selection pressures would favor substantially increased egg volume without increased egg EGG SIZE-ENERGY RELATIONSHIP 49 energy content, especially in an organism that broods its offspring. Our conclusion from this study is that egg size is of limited value for tine-scale quantitative predictions of egg energy content and should be used with caution. This conclusion challenges a fundamental assumption in re- productive ecology and. as a result, increases the diffi- culty of understanding life-history patterns in marine invertebrates. Acknowledgments We thank G. Albrecht for kindly providing advice and source code for the iterative method for fitting the full allometric model. The assistance of B. Bolker, D. Brazeau. A. Kohn, C. Lanciani, C. Osenberg. D. Padilla. R. Strathmann. and especially C. Solari with data analy- sis is greatly appreciated. L. McEdward and M. McEd- ward assisted with proofreading the manuscript. This work was supported by grants from the University of Florida and the National Science Foundation (OCE- 9818593) to L. McEdward. Literature Cited Albrecht. G. H., B. R. Gelvin, and S. E. Hartman. 1993. Ratios as a size adjustment in moiphometrics. J. Phvsiol. Anthropol. 39: 441- 468. Belsley, D. A., E. Kuh, and R. E. Welsch. 1980. Regression Diagnos- tics: Identifying Influential Data and Sources of Collinearitv. John Wiley. New York. Byrne, M. 1996. Viviparity and intragonadal cannibalism in the dimin- utive sea stars Patiriella vivipara and P. par\-ivipara (family Asterini- dae). Mar. Biol. 125: 551-567. Christiansen, F. B., and T. M. Fenchel. 1979. Evolution of marine invertebrate reproductive patterns. Theor. Popul. Biol. 16: 267-2X2. Ebert, T. A., and M. P. Russell. 1994. Allometry and model-II nonlin- ear regression. J. Tlieor. Biol. 168: 367-372. Eckert, G. L. 1995. A novel larval feeding strategy of the tropical sand dollar, Encope nuehelini (Agassiz): adaptation to food limitation and an evolutionary link between planktotrophy and lecithotrophy. ./. E.\p. Mar. Biol. Ecol. 187: 103-128. Emlet. R. B. 1986. Facultative planktotrophy in the tropical echinoid Clypeaster rosacen.s (Linnaeus) and a comparison with obligate plank- totrophy in Clypeaster subdepressus (Gray) (Clypeasteroidea: Echi- noidea). J. E.\p. Mar. Biol. Ecol. 95: 183-202. Emlet, R. B. 1989. Apical skeletons of sea urchins (Echinodermata: Echinoidea): two methods tor inferring mode of larval development. Paleohiology 15: 223-254. Emlet, R. B., L. R. McEdward. and R. R. Strathmann. 1987. Echi- noderm larval ecology viewed from the egg. Ecliinodcrin Stud. 2: 55-136. George, S. B., C. M. Young, and L. Fenaux. 1997. Proximate compo- sition of eggs and larvae of the sand dollar Encope michelini (Agassiz): the advantage of higher investment in planktotrophic eggs. Invertehr. Reprod. Dev. 32: I 1-19. Gnaiger, E. 1983. Calculation of energetic and biochemical equivalents of respiratory oxygen consumption. Pp. 337-345 in Polarograpliii Oxygen Sensors, E. Gnaiger and H. Forstner. eds. Springer-Verlag. Berlin. Grant, A. 1983. On the evolution of brood protection in marine benthic invertebrates, ,4m. Nat. 122: 549-555. Havenhand, J. N. 1995. Evolutionary ecology of larval types. Pp. 79- 122 in Ecology, nl Marine Invertebrate Larvae. L. R. McEdward, ed. CRC Press. Boca Raton. EL. Herrera, J., S. McVVeeney, and L. R. McEdward. 1996. Diversity of energetic strategies among echinoid larvae and the transition from feeding to nonteedmg development. Oceanol. Ada 19: 313-321. Huxley, J. 1932. Problem^ ,>/ Relative Growth. Dover Publications. New York. Jaeckle. W. B. 1995. Variation m the size, energy content, and biochem- ical composition of invertebrate eggs: correlates to the mode of larval development. Pp. 49-78 in Ecology of Marine Invertebrate Larvae. L. R. McEdward. ed. CRC Press. Boca Raton. FL. Lawrence, J. M.. J. B. McCIintock, and A. Guille. 1984. Organic level and caloric content of eggs of brooding asteroids and an echinoid (Echinodermata) from Kerguelen (South Indian Ocean). Int. J. Inver- tebr. Reprod. Dev. 7: 249-257. Levin, L. A., and T. S. Bridges. 1995. Pattern and diversity in repro- duction and development. Pp. 1-48 in Ecolog\ of Marine Invertebrate Lan-ae. L. R. McEdward. ed. CRC Press. Boca Raton. FL. Levitan, D. R. 1996. Predicting optimal and unique egg sizes in free- spawning marine invertebrates. Am. Nat. 1-18: 174-188. Manaster, B. J., and S. Manaster. 1975. Techniques for estimating allometric equations. ,/. Morphol. 147: 299-308. McClary, D. J., and P. V. Mladenov. 1990. Brooding biology of the sea star Pteraster mi/ilaris (O. F. Muller): energetic and histological evi- dence for nutrient translocation to brooded juveniles. J. Exp. Mar. Biol. Ecol. 142: 183-199. McCIintock, J. B., and J. S. Pearse. 1986. Organic and energetic- content of eggs and juveniles of Antarctic echinoids and asteroids with lecithotrophic development. Comp. Biochem. Physiol. 85 A: 341-345. McEdward, L. R. 1997. Reproductive strategies of marine benthic in- vertebrates revisited: facultative feeding by planktotrophic larvae. Am. Nat. 150: 48-72. McEdward, L. R.. and S. F. Carson. 1987. Variation in egg organic content and its relationship with egg size in the starfish Solaster stimpsoni. Mar. Ecol. Prog. Ser. 37: 159-168. McEdward, L. R.. and F. S. Chia. 1991. Size and energy content of eggs from echinoderms with pelagic lecithotrophic development. J. E.xp. Mar. Biol. Ecol. 147: 95-102. McEdward, L. R., and L. K. Coulter. 1987. Egg volume and energetic content are not correlated among sibling offspring of starfish: implica- tions for life-history theory. Evolution 41: 914-917. McEdward, L. R.. and D. A. Janies. 1997. Relationships among de- velopment, ecology, and morphology in the evolution of Echinoderm larvae and life cycles. Biol. J. Lmn. Soc. 6(1: 381-400. McEdward, L. R., S. F. Carson, and F. S. Chia. 1988. Energetic content of eggs, larvae, and juveniles of Florometra serrati.\\iina and the implications for the evolution of crmoid life histories. Int. J. Inver- tebr. Reprod. Dev. 13: 9-22. Neter, J., W. Wasserman, and M. H. Kutner. 1990. Applied Linear Statistical Models. Irwin. Bun Ridge. IL. Parsons, T. R., M. Takahashi. and B. Margrave. 1984. Biological Oceanographic /Voir.v.wv Pciiiaiii '•'!.. New York. Pechenik. J. A. 1979. K"l - ni ,-IK ,ip-.uhition in invertebrate life histories. An,. Nat- 114: 859-870. Perron, F. E., and R. H. Carrier. 1981. Egg size distributions among closely related marine invertebrate species: Are they bimodal or uni- modal'Mm. Nut. 118: 749-755. Pianka. E. R. 197f>. Natural selection of optimal reproductive tactics. Am. Zoo/. 16: 775-7S4. Sewell, M. A., and F. S. Chia. 1994. Reproduction of the intraovarian 50 L. R. McEDWARD AND K. H. MORGAN brooding apodid Leptosynapra clarki (Echinodermata; Holothuroidea) in British Columbia. Mar. Biol. 121: 285-300. Shilling, F. M., and D. T. Manahan. 1994. Energy metabolism and amino acid transport during early development of Antarctic and tem- perate echinoderms. Biol. Bull. 187: 398-407. Smith, C. C., and S. D. Fretwell. 1974. The optimal balance between size and number of offspring. Am. Nat. 108: 499-506. Stearns, S. C. 1977. The evolution of life history traits: a critique of the theory and a review of the data. Annu. Rev. Ecol. Syst. 8: 145-171. Strathmann, R. R. 1985. Feeding and nonfeedmg larval development and life-history evolution in marine invertebrates. Annu. Rev. Ecol. S\st. 16: 339-361. Strathmann, R. R., and K. Vedder. 1977. Size and organic content of eggs of echinoderms and other invertebrates as related to developmen- tal strategies and egg eating. Mar. Biol. 39: 305-309. Turner. R. L., and J. M. Lawrence. 1979. Volume and composition of echinoderm eggs: implications for the use of egg size in life-history models. Pp. 25-40 in Reproductive Ecology of Marine Invertebrates, S. E. Stancyk. ed. University of South Carolina Press. Columbia. SC. Turner, R. L., and J. C. Rutherford. 1976. Organic, inorganic, and caloric composition of eggs, pentaculae, and adults of the brooding sea cucumber Cucumaria curata Cowles (Echinodermata: Holothuroidea). J. Exp. Mar. Biol. Ecol. 24: 49-60. Vance, R. R. 1973. On reproductive strategies in marine benthic inver- tebrates. Am. Nat. 107: 339-352. Reference: Biol. Bull. 200: 51-58. (February 20011 The Adaptive Bleaching Hypothesis: Experimental Tests of Critical Assumptions ROBERT A. KINZIE III1-*, MICHELLE TAKAYAMA2, SCOTT R. SANTOS3, AND MARY ALICE COFFROTH3 ' Zoology Department and Hawaii Institute of Marine Biology. University of Hawaii, Honolulu. Hawaii 96822; 2Claremont McKenna College. Clarewont, California 9171 1-6400: and 'Department of Biological Sciences. The Universin- at Buffalo, Buffalo, New York 14260-1300 Abstract. Coral bleaching, the loss of color due to loss of symbiotic zooxanthellae or their pigment, appears to be increasing in intensity and geographic extent, perhaps re- lated to increasing sea surface temperatures. The adaptive bleaching hypothesis (ABH) posits that when environmen- tal circumstances change, the loss of one or more kinds of zooxanthellae is rapidly, sometimes unnoticeably, followed by formation of a new symbiotic consortium with different zooxanthellae that are more suited to the new conditions in the host's habitat. Fundamental assumptions of the ABH include ( 1 ) different types of zooxanthellae respond differ- ently to environmental conditions, specifically temperature, and (2) bleached adults can secondarily acquire zooxanthel- lae from the environment. We present simple tests of these assumptions and show that ( 1 ) genetically different strains of zooxanthellae exhibit different responses to elevated tem- perature. (2) bleached adult hosts can acquire algal symbi- onts with an apparently dose-dependent relationship be- tween the concentration of zooxanthellae and the rate of establishment of the symbiosis, (3) and finally, bleached adult hosts can acquire symbionts from the water column. Introduction Coral bleaching has increased in frequency, intensity, and geographical extent over the last two decades (Huppert and Stone, 1998). Concern is being expressed both by reef scientists (Brown. 1990. 1997; Goreau and Hayes. 1994) and the general public (Brown and Ogden, 1993; Wilson. 1999) about the future of the planet's reefs. Bleaching is a Received 12 April 2000; accepted 14 November 2000. * To whom correspondence should be addressed. kinzie@hawaii.edu E-mail: general term describing the loss of color in corals and other symbiotic reef invertebrates, usually due to the loss of their intracellular dinoflagellate symbionts, colloquially termed zooxanthellae. A reduction in the density or pig- ment content of symbiotic algae can be elicited by a variety of environmental factors including low (Goreau, 1964) and high (Nakano et a/.. 1997) salinity, low and high levels of illumination — especially ultraviolet radiation (Ban- aszak and Trench. 1995; Lesser and Shick, 1989), disease (Kushmaro et al.. 1996; Rosenberg and Loya, 1999), and high (Iglesias-Prieto et al.. 1992; Warner et al.. 1996) and low (Steen and Muscatine, 1987; Kobluk and Ly- senko, 1994) temperatures, as well as combinations of these factors. Although any of these may act on a local scale, widespread bleaching has generally been linked to high water temperatures at larger geographic scales, particu- larly those associated with El Nino-Southern Oscillation (ENSO) events (Glynn, 1984. 1991: Coffroth et al.. 1988). Most recently, the high sea surface temperatures (SSTs) of the summer of 1998 resulted in massive mortality of reefs in Okinawa. Tahiti, and Belau (ISRS, 1998; Hoegh- Guldberg, 1999; Tsuchiya, 1999). Such events may be be- coming more frequent, and it has been suggested that they will become even more common if global warming persists and summer SSTs increase (Glynn, 1991). The loss, in association with such events, of zooxanthellae from a host may be partial or essentially complete. In the latter case recovery frequently does run occur, and the subsequent death of large numbers of symbiotic reef invertebrates can drastically alter the composition of the reef commu- nity (Goreau. 1992). Thus, the potential for increased inci- dence of bleaching may have negative effects on these ecosystems. Here we present results from experiments 51 52 R. A. KINZIE ET AL relating zooxanthellar diversity to potential recovery from loss of symbionts. Our original view of the symbiotic system of reef inver- tebrates and their algal symbionts was based on the idea that, at least in cnidarians, there was a single algal partner, Symbiodinium microadriaticwn (Freudenthal) (Taylor. 1974). The elucidation of the complexities of coral and zooxanthellar systematics and taxonomy has made the pic- ture more complex (Rowan and Powers 1991a; Knowlton et al., 1992; Rowan, 1998). About a dozen named dinoflagel- late taxa are currently known to associate with temperate and tropical invertebrates (Blank and Trench. 1985; Ban- aszaketal., 1993; McNally etui., 1994), but the diversity of the algae is probably much higher than this (Rowan, 1998; Darius et al.. 1998; Carlos et al.. 1999; Baillie et al.. 2000). A broad classification of zooxanthellae. based on sequence differences in the small subunit of the ribosomal nuclear genes (ssrDNA) has been proposed (Rowan and Powers, 199 la). This region of the genome contains a range of domains that have evolved at different rates; these domains are useful in characterizing phylogenetic relationships and have been used to characterize populations of zooxanthellae (Rowan, 1991; Rowan and Powers, 199 la. b; Sadler et a I.. 1992; McNally et al.. 1994; Zardoya et al.. 1995; Carlos et al.. 1999). Analysis of restriction fragment length polymor- phism (RFLP) of the ssrDNA has shown that zooxanthellae can be divided into at least four clades (Rowan and Powers, 1991a, b; Rowan, 1998), each probably including many species (Rowan. 1998). In addition, we now know that although some invertebrates may host a single algal clade, others can harbor at least three distinctly different types of zooxanthellae (Rowan and Knowlton, 1995). When more than one clade occurs, the balance among the population densities and the location of the clades within a host coral colony can be altered by changes in microscale environ- mental conditions (Rowan and Powers, 199 la; Rowan and Knowlton, 1995; Rowan et al. 1997; Buddemeier, 1997). Furthermore, evidence suggests that the consortium may readily alter membership (Davy et al.. 1997: Hill and Wil- cox, 1998; Belda-Baillie et al., 1999; Bates. 2000). Thus, this symbiosis, which — at least in scleractinians — has per- sisted since the late Triassic (Stanley and Swart. 1995). appears to be both complex and mutable. This combination of persistence over geological time scales (Kinzie and Buddemeier, 1996; Buddemeier et al.. 1997) and the apparently multifaceted array of symbiotic partnerships that may occur in some hosts (Rowan and Powers, 199 la: Baker and Rowan, 1997; Rowan. 1998) has given rise to the "adaptive bleaching hypothesis" (ABH) (Buddemeier and Fautin, 1993). This hypothesis posits that different algal partners within hosts shift in space — over a range of scales — and time, resulting in a symbiotic complex with many potential combinations, some being more func- tional under certain environmental conditions than others. According to the ABH, as environmental conditions change, the makeup of the symbiotic unit responds, tracking envi- ronmental changes by shuffling the algal membership of the symbiotic consortium. This hypothesis, presented 7 years ago, was explicitly stated so as to be experimentally test- able, but despite substantial work both in the field and in laboratories around the world on causes and mechanisms of coral bleaching, surprisingly little effort has been devoted to such tests. Here we present the results of experimental tests of some fundamental assumptions underlying the ABH. One assumption of the ABH is that the various algal types have physiological differences that are related to the etiol- ogy of bleaching. High temperatures are known to nega- tively affect zooxanthellae through degradation of photo- system II (Iglesias-Prieto et al.. 1992). but there has been little study of temperature effects on different zooxanthellar types. It has been known for some time that different zooxanthellae differ in their ability to grow in different hosts (Schoenberg and Trench. 1980) and to support growth and reproduction of the animal hosts (Kinzie and Chee, 1979; Fitt, 1985). Additionally, algae differ in their photo- synthetic abilities (Chang et al.. 1983) and in their produc- tion of mycosporine-like amino acids in response to ultra- violet irradiance (Banaszak et al., 2000). A second assumption of the ABH is that bleached adult hosts can obtain new symbionts from the reef habitat and reestablish a symbiosis; that is, secondary acquisition of algae is possible. Many studies have demonstrated initial acquisition of newly settled aposymbiotic juveniles (Kinzie. 1974: Fitt and Trench. 1981; Muller-Parker and D'Elia, 1996; Benayahu et al., 1989) or freely swimming planulae (Schwarz et al., 1999). Numerous symbiotic reef inverte- brates utilize horizontal transmission (i.e., each generation obtains its algal symbionts from the environment rather than maternally), showing that there must be substantial popula- tions of potential zooxanthellae partners in reef waters (Goulet and Coffroth, 1997). These potential symbiotic partners exist outside of hosts and may occur in fish feces I Muller-Parker, 1984), be released from corals (Stimson and Kinzie, 1991; Hoegh-Guldberg et al., 1987). or occur free living (Carlos et al., 1999). Direct study of free-living zooxanthellae on reefs is difficult because we currently have very little data on their density, spatial distribution, or temporal behavior. We wished to determine whether bleached hosts could acquire symbiotic algae from popula- tions of zooxanthellae in more natural situations. Although bleached adults have been shown to take up zooxanthellae under laboratory conditions, these experiments either used extremely high concentrations of zooxanthellae or the algae were injected into the coelenteron. The ability of adult cnidarians. once they have become bleached, to re-acquire symbionts from populations at concentrations likely to oc- cur in situ is less studied (Franzisket. 1970). Because understanding of the symbiotic unit is based on ADAPTIVE BLEACHING HYPOTHESIS knowledge about the individual partners (Iglesias-Prieto el nl., 1992). we examined the algal partner under culture conditions to test the first assumption of the ABH. Although certain physiological traits of zooxanthellae can change in culture, (Stochaj and Grossman. 1997). use of well-charac- terized and controlled partners in isolation provides strong experimental control. In testing the other assumptions we used bleached hosts and naturally occurring populations of zooxanthellae. Materials and Methods Algal cultures We refer to the zooxanthellae used in our experiments as "isolates." referencing the host from which they were iso- lated rather than a dinoflagellate species because the sys- tematics and nomenclature of the entire group are in forma- tive stages (Lee et al., 1995). The isolates used are from a range of hosts (Table 1 ) and have been in culture for 4 to 20 years. Each culture was initiated with zooxanthellae from a single host. Because some hosts may harbor more than a single zooxanthellar genotype (Rowan and Powers. 199 la; Rowan and Knowlton, 1995; Rowan et al.. 1997; Budde- meier. 1997; Goulet and Coffroth. 1997; Carlos et al., 2000). the isolates could have been initiated from more than a single genetic strain of zooxanthellae. In addition, the culturing process itself may exert strong selection on which genotypes thrive. To determine the number and types of clades present in the isolates as well as those that colonized bleached hosts, zooxanthellar genotypes were characterized using RFLP analysis of zooxanthellar ssrDNA (Rowan and Powers. 1991a). Cladal membership of the stock cultures was determined before the start of the experiment and one month after the experiments were terminated. The experi- mental cultures were analyzed at the end of the experiment. Table 1 Isolates used in temperature tolerance experiments Isolate designation Clade Host Source Cassiopea xamachana A Scyphozoan Caribbean (Symbiodinium mn roadriaticum) Montipora verruciisa C Scleractinian Hawaii (Svmbiodiniiim luzwagutii) Pocillopora B Scleractinian Hawaii damicomis Cassiopea KB8 A Scyphozoan Hawaii Zoanthus sociatus A Zoanthid Caribbean (S\mbir>diiiiiitn pilosuni) Tndacna gigas A Giant clam Central Pacific Aipknui pulchellu B Anemone Hawaii (S\inbii>dinium pulchrorurri) Specific names in parentheses are species names given to zooxanthellae from these species (Blank and Trench. 1985. and Banaszak et al.. 1993). The resultant RFLP groups were classified by comparison with known zooxanthellae (i.e., Symbiodinium clades A. B. and C, Rowan and Powers, 1991b). Each stock culture contained a single algal clade (sensit Rowan and Powers, 199 la; Table 1). In the analysis, algal DNA extraction followed the protocols of Coffroth et al. (1992) and Goulet and Coffroth ( 1997). and RFLP analysis followed the pro- tocols of Rowan and Powers (1991b). Stock cultures were grown in f/2 medium (Guillard and Ryther, 1962) held at 25°C under 160 juE irT2 s~' (160 /u,mol m~2 s"1) on a 12:12 light-dark cycle. Cultures were transferred to new medium monthly. Periodically, cultures were treated with antibiotics (Polne-Fuller, 1991) but were not axenic. As discussed below, even though some isolates have been in culture for more than two decades, they retain their ability to infect their normal host. Temperature sensitivity of different algal isolates Aliquots from each cultured isolate were diluted to 1 X 104 cells ml"' with sterile f/2 medium and put into twelve 150-ml culture tubes. Four tubes for each isolate were randomly assigned to one of the three temperature treat- ments. Zooxanthellae were exposed to temperatures of 27. 29, and 31°C over a 2-week period during which growth rates of the cultures were determined fluorometrically. Tem- perature treatments consisted of 28 tubes (4 per isolate and 7 isolates) in a test tube rack immersed in a 10-1 water bath with an aquarium heater; an aquarium pump kept the water mixed. The three water baths were placed by an east-facing window that provided natural light. Solar radiation within the culture tubes (measured with a BioSpherical quantum meter, model QS100 with a 4ir sensor inserted into a tube) ranged from 800 juE m 2 s" ' in the morning to 100 /xE m~" s (100/imolm ~s ) in the afternoon. Variability within a tube, from the top of the culture medium to the bottom of the tube, was about 30%. Each day. when tubes were read, the positions of the three water baths were rotated, and the positions of the tubes in the racks within each bath were shuffled. Aside from the temperature effects, growth was probably light-limited (unpubl. data) because we wished to reduce the potential for interaction between radiation stress and the experimental treatment. Growth was followed using in vivo fluorescence with a Turner Designs model 10 fluor- ometer. Fluorescence was measured daily starting at 0830. Racks were held in the dark at the experimental temperature for 30 min before fluorescence was read. Just before the reading, cultures in the tubes were dispersed with a vortex mixer to ensure even suspensions. Readings were taken from day 0. when inoculation took place, until day 13. when several of the cultures became difficult to disperse. At the termination of the experiment. 1-ml samples were preserved in Lugol's solution for cell enumeration with a Spiers-Levi eosinophil hemacytometer. Growth was measured as daily 54 R. A. KINZIE ET AL increase in fluorescence. We used growth rate as a measure of each strain's ability to tolerate high temperatures because taster growing zooxanthellae would directly con-elate with a coral's recovery from bleaching. Population growth rate is generally considered to be a primary measure of fitness (Cole. 1954). At the end of the experiment, the cell counts were used to convert fluorescence to cell density. Specific growth rate (cell cell'1 d ') was determined from regres- sions of cell density over time. Growth rate was determined as the slope of the natural log of fluorescence over time. Subsequent analysis indicated that growth was logarithmic over this time period. Ability of bleached hosts to acquire new synibionts To demonstrate that secondan acquisition of algal s\m- bionts b\ bleached adults can occur at low concentrations ot zooxanthellae. we used the anemone Aiptasia pnlcliella as a model organism. This symbiotic antho/oan can be bleached completeh and maintained indefinite!) in tins condition. making it particularly useful for reinfection studies. Aipta- sia pohps \\ere held individual!) in 50-ml beakers \\ith 0.45-M'n-filtered seawater (FSW) in the dark and fed Ar- temia nauplii once a \veek. After at least one year in the dark, bleached Aiptasia were moved to the light (natural sunlight through north-facing windows*, and held in 0.22-ju.m FSW with weekly feeding of Aneniiii nauplii. No anemones were used in experiments unless the) remained aposymbiotic in the light for at least 5 months. We have kept bleached Aiptasia for more than a v ear w ith no signs of zooxanthellae. Bleached Aiptasia polyps were exposed to algal svmbi- onts isolated fiom A. pulchella at concentrations that ranged from 1C5 to 10" cells nil" '. Stock culture of the A. pulchella isolate (in culture for >20 yearsi was diluted to 1 X 10" cells ml"1 (\enfied by direct hemacx tometer counts), then diluted to the different experimental concentrations with 0.22-/j.m FSW. The seawater in the beakers with the Aipta- sia was replaced with the algal suspensions 01 = 3 per treatment). Anemones remained in the suspension for 48 h. after which the suspension w as removed and replaced with 0.22-^tm FSW. Anemones were maintained in 0.22-/im FSW with the same light and feeding conditions, and were checked dailv . We measured the time required for the zoox- anthellae to proliferate to the point that the first brow n color was visible. We termed this "visible infection." None of the controls e\er showed signs of establishment of symbiosis with algae. EviV. ^:iitable strain* .<:iht'llae in natural watt To test the possibilitv that bleached adult hosts can ac- quire akal sxmbionts from natural waters, we exposed bleached Aipuitia to unhTtered seawater in three situations: Table 2 Exposure and infection of bleached Aiptasia pulchella polyps Time to infection (days) Number of Number Treatment Aiptasia reint'ected Mean Range Water table 9 6 41.2 5-76 Lagoon 3 3 54 41-61 Coral reel 10 3 28 18-46 on a sea\\ ater table in the laboratory, in the lagoon, and on the coral reef (Table 2). The anemones used in these treat- ments were randomly selected from the set of bleached Aiptasia described above. In the laboratory. 9 anemones were placed in individual uncovered 50-ml beakers that were then immersed in running unnTtered seawater (from the laboratory's system) on a water table that had a popu- lation of naturally occurring symbiotic Aiptasia polyps. For the field exposures. Aiptasiu were placed in plastic tubes (ca. 50 ml) with coarse (5-mm-mesh) netting over the ends. A tube with 3 anemones was suspended at a depth of 0.5 m in the lagoon at a site that had moderate densities of sym- biotic Aiptasia. A tube with 10 anemones was suspended 1 m above the reef at a site where no Aiptasia were visible. The anemones w ere held in these treatment conditions for 1 to 7 days, then returned to 0.22-^im FSW and kept in the conditions described in the previous section. Results Temperature sensitivity of different algal isolates All seven isolates showed positive growth at all three temperatures. e\en at 31:C. which is often associated with bleaching events (Hoegh-Guldberg and Salvat. 1995: Da- \ies et al.. 1997: Drollet et ai.. 1995) (Fig. It. Although all the zooxanthellae were able to grow at these three temper- atures, there were significant differences among the isolates (Table 3). Clade B zooxanthellae showed decreasing growth at higher temperatures, whereas the single clade C isolate showed increasing growth rates at higher temperatures. Among the clade A isolates tested, the response to temper- ature was variable. Clade A isolates included the fastest growing isolate, which showed increasing growth rates with increasing temperatures, and the slowest growing isolate, which showed decreased growth rates at the higher temper- atures. RFLP anal) sis of ssrDNA of the stock cultures both before and after the experiments verified that each stock culture contained a single algal clade. RFLP analysis of the experimental cultures at the termination of the experiment revealed that sev eral of the treatments showed low levels of contamination w ith a second clade. Because the cladal type of the stock cultures did not change during this time and the growth rates of replicate isolates were similar, we conclude ADAPTIVE BLEACHING HYPOTHESIS 70 55 Temperature (C°) Figure 1. Growth rates of zooxanthellae isolated from 7 hosts at three temperatures. Points represent means: error bars represent ±1 SD (n = 4). Data were obtained as increase in fluorescence over the 13-day growth period. Growth was logarithmic during this time. Fluorescence units were converted to cell density at the end of the growth period. Host: zs — /tuinihiis sociatus. mv — Montipora verrucosa, ex — Cassiopea \am- achana, KB8 — Cassiopea KB8, pd — Pocillopora damicomis, ap — Aipta- sia pulchella, tg — Tridacna gigas. The dotted lines represent algae in clade A, the solid lines algae in clade B. and the dashed line the alga in strain C. that the contamination was introduced at the termination of the experiments when samples were being processed for chlorophyll a measurements, cell counts, and photosyn- thetic rate studies, which was also when samples were taken for cladal identification of the experimental cultures. Ahilit\ of bleached hosts to acquire new symbionts at low zooxanthellae concentrations Symbiosis was successfully established at concentrations as low as 10 cells ml"1 (Fig. 2). Anemones exposed to higher concentrations became visibly infected sooner than those exposed to lower concentrations, suggesting that re- population was due to multiple infection events during the 48-h exposure period and that the time to visible infection is dose-dependent. Although it is not known whether recovery following natural bleaching events comes from a residual population within the host or from free-living populations in Table 3 Analysis of variance of growth rates by temperature for the seven Zooxanthella isolates Source DF SS MS Isolate 6 0.089 0.015 101.03 <0.0001 Temperature 2 0.001 0.000 3.32 0.043 Isolate x Temperature 12 0.006 0.000 3.31 0.0009 (A £ Q 30 Zooxanthella concentration vs Time until visibly infected Days = 56.55 - 2.19 " In(Concentration) r2 = 0.47 In (Zooxanthella Concentration) Figure 2. Time to infection of bleached Aiptasia pulchella polyps vs. concentration of zooxanthellae isolated from Aiptasia pulchella. reef waters, the minimum effective dose we demonstrated of 10 cells ml" ' implies that the concentrations of zooxanthel- lae repopulating bleached hosts could be almost undetect- ably low. yet still be sufficient for the bleached host to recover. Existence of infective fonns of zooxanthellae in natural waters Two-thirds of the bleached anemones exposed to unfil- tered seawater in the water table became infected, as did one-third of the bleached anemones suspended above the reef and all of the bleached anemones suspended in the lagoon (Table 2). All Aiptasia that became infected when exposed to natural seawater were found to harbor clade B zooxanthellae, which is the zooxanthellar clade normally found in this anemone. We emphasize that in both experi- ments, in which Aiptasia did regain zooxanthellae. none of the control anemones showed any sign of infection at the termination of the experiments, and the same polyps remain aposymbiotic 10 months later. There was no correlation between the time in the treatment and the time to infection. Discussion Temperature sensitivity nf different algal isolates The algal isolates we studied responded differently to the different temperatures. Some showed lowered growth rate with increasing temperature, some showed the same growth rate at all three temperatures, and two showed higher growth rates with increasing temperatures. Given that temperatures of 30°C. even when applied for short periods (2 days), can 56 R. A. K1NZIE ET AL cause breakdown in the photosynthetic machinery of some zooxanthellae (Iglesias-Prieto el ai, 1992; Iglesias-Prieto, 1997; Warner et ai, 1996, 1999), the ability of these isolates to grow at 31°C, and for some to apparently show their highest growth rate at this temperature, is surprising. Also unexpected is the apparent lack of relationship between the habitat of the host and the temperature tolerance of the zooxanthellae. For example, algae from Montipom verni- cosa, isolated from a coral from Hawaii, from relatively cool waters, showed increased growth rates at higher tem- peratures, whereas an isolate from the giant clam Tridacna gigas, typically found in shallow, more equatorial waters, was one of the most sensitive to high temperatures. When the data are examined in terms of zooxanthellar clade, several trends are apparent. First, clade A zooxan- thellae, for which we tested the largest number of isolates, had the widest range of growth rates. This is consistent with previous characterizations of clade A zooxanthellae as "weedy" (Rowan, 1998). Some clade A zooxanthellae showed increased growth with increasing temperatures (e.g.. Zoanthus sociatus), others showed decreasing growth (Triilacnci gigas). and isolates from the two species of Cassiopea showed little change in growth rate over the experimental temperature range. Clade B zooxanthellae, for which we tested only two isolates, showed less variability in response to temperature, with both isolates growing slower at higher temperatures. Growth rates in the single clade C isolate tested showed a slightly higher growth rate at higher temperatures and was the second fastest growing of all the isolates we tested. The results from clade A zooxanthellae demonstrate that, at least in some algal groups, the variabil- ity in physiological response within a clade may be as great as or greater than between eludes. Further, these data sug- gest that the potential for selection for the most ecologically suitable taxon or genotype could occur within a clade. If this proves to be generally true, it might be impossible to detect ecologically significant shifts in the algal complement of an animal host by using the identification techniques we em- ployed in this study. Ahilitv of bleached hosts to reestablish svmhiosis at low zooxanthellae concentrations Even though the isolate used in these experiments has been in culture for more than two decades, it retained the ability to successfully reestablish a symbiosis with its nor- mal host, even when at exceedingly low densities. Thus, although some physiological changes may occur in the course of the culturing process (Stochaj and Grossman. 1997), the basic mechanisms of recognition and entry to host cells appear to remain intact. The dose-response rela- tionship demonstrated here is ecologically important be- cause it suggests that the chances of secondary acquisition, such as may occur during recovery from bleaching, will be greater at sites where zooxanthellae are abundant in the waters overlying the reef (and presumably their hosts). This implies an effect of spatial scale and community composi- tion in the probability of recovery from different sorts of bleaching events. The experiments described here were done in closed containers, so the concentration at which a 48-h exposure results in infection may not easily translate to concentrations in the water overlying the reef. Existence of infective forms of zooxanthellae in natural waters On the reef, water motion is an important factor in ameliorating the effects of low concentration of nutrients on algal growth (Larned and Atkinson. 1997). The same phe- nomenon is likely to be important in exposing bleached animals to free-living zooxanthellae, potentially resulting in reestablishing the symbiosis even when zooxanthellae are at very low concentrations. It is noteworthy that on the reef at the Hawaii Institute of Marine Biology pier where water motion is greatest, the density of Aiptasia is much lower than in the calm lagoon, and yet the mean time to estab- lishment of the symbiosis was the shortest in this location. However, the percentage of polyps that secondarily ac- quired zooxanthellae was lowest at the reef site. These apparently conflicting results could mean that in open wa- ters, free-living zooxanthellae are sometimes at very low densities and perhaps patchily distributed. Because all 10 anemones were within the same tube, it is also possible that only one took up algae from the water column and the rest somehow obtained their zooxanthellae from this individual. Conclusion In summary, we have provided experimental tests of fundamental assumptions of the ABH. The differential growth rates of the isolates in response to increasing tem- peratures suggest that different responses to a factor sus- pected as a cause of recent large-scale bleaching events do exist among zooxanthellae. Furthermore, the variability in this response can be as great within a clade as between zooxanthellae from different clades. The uptake of zooxan- thellae in situ by bleached hosts demonstrates that second- ary acquisition can occur, and the results of the laboratory studies show that new symbionts can be acquired when the alga concentrations are very low. These findings address important aspects of the ABH. The assumptions tested here are necessary but not sufficient conditions of the ABH. In addition, the lack of clear cladal patterns in the temperature response demonstrates that the ability to withstand high temperatures is not correlated with membership in a specific clade and implies that finer genetic differentiation will be required to understand the dynamics of these symbionts. A clearer understanding of the significance of the different ADAPTIVE BLEACHING HYPOTHESIS 57 temperature responses and the bearing these differences have on the ABH will require studies of intact syrnbioses. Acknowledgments We thank H. Lasker and E. F. Cox for comments on the manuscript. This study was partly supported by Grants-in- Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan, and the A. Y. Yoshi- naga foundation (to RAKI1I); REU support from NSF OCE- 96-22782 (to MT): an NSF minority fellowship (to SRS); and NSF OCE-95-30057 (to MAC). Heather Reed assisted with both laboratory and field aspects of the study. Literature Cited Baillic, B. K., C. A. Belda-Baillie, V. Silvestre, M. Sisin, A. V. Gomez, E. D. Gomez, V. Monje. 20(10. Genetic variation in Synihiodinimit isolates from giant clams based on random-amplified-polymoiphic DNA (RAPID) patterns. Mar. Bio/. 136: 829-836. Baker, A. C., and R. Rowan. 1997. Diversity of symbiotic dinoflagel- lates (zooxanthellae) in scleractinian corals of the Caribbean and east- ern Pacific. Proceedings of the Eighth International Coral Reef Symp. I Panama} 2: 1301-1306. Banaszak. A. T.. and R. K. Trench. 1995. Effects of ultraviolet (UV) radiation on marine microalgal-invertebrate symbioses. I. Response of the algal symbionts in culture and in hospite. J. Exp. Mar. Bio/. Ecol. 194: 213-232. Banaszak, A. T.. T. R. Iglesias-Prieto, and R. K. Trench. 1993. Scrippsiella ve/ellae sp. nov. (Peridiniales) and Gloeodiniuin viscinn sp. nov. (Phytodinmlesi. dinorlagellate symbionts of two hydrozoans (Cni- daria). J. Phycol. 29: 517-528. Banaszak. A. T., T. C. Lajeunesse, and R. K. Trench. 2000. The synthesis of mycosporine-like amino acids (MAAs) by cultured, sym- biotic dinorlagellates. J. Ev/>. Mar. Bio/. Ecol. 249: 219-233. Bates, A. 20(10. The intertidal distribution of two algal symbionts hosted by Anthopleura xanthogrammica (Brandt 1935). J. Exp. Mar. Biol. Ecol. 249: 249-262. Belda-Baillie. C. A., M. Sison, V. Silvestre, K. Villamor, V. Monje, E. D. Gomez, and B. K. Baillie. 1999. Evidence for changing symbiotic algae in juvenile tridacnids. J. Exp. Mai. Biol. Ecol. 241: 207-221. Benayahu, Y., Y. Achituv, and T. Berner. 1989. Metamorphosis of an octocoral primary polyp and its infection by algal symbiosis. S\mbiosi.\ 1: 159-169. Blank. R. J., and R. K. Trench. 1985. Speciation and symbiotic dinorlagellates. Science 229: 656-658. Brown, B. E. (Ed). 1990. Coral bleaching: special issue. Coral Reefs 8: 153-232. Brown, B. E. 1997. Coral bleaching: causes and consequences. Coral Reefs 16Suppl.: S129-S138. Brown, B. E.. and J. C. Ogden. 1993. Coral bleaching. Sci. Am. 268: 64-70. Buddemeier. R. \V. 1997. Making light work of adaptation. Nature 388: 65-73. Buddemeier, R. W.. and D. G. Fautin. 1993. Coral bleaching as an adaptive mechanism. BioScience 43: 320-326. Buddemeier, R. W., D. G. Fautin, and J. R. Ware. 1997. Acclimation, adaptation and algal symbiosis in reef-building scleractinian corals. Proceedings of the Si\th International Conference on Coelenterale Biology. Noordwijkerhout. The Netherlands. 1995 1: 3-10. Carlos, A. A.. B. K. Baillie. M. Kawachi, and T. Maruvama. 1999. Phylogenetic position of Symbiodinium (Dinophyceae) isolates from Tridacnids (Bivalvia). Cardids (Bivalvia). a sponge (Porifera). a soft coral (Anthozoa). and a free-living strain. J. Ph\col. 35: 1054-1062. Carlos, A. A., B. K. Baillie, and T. Maruyama. 2000. Diversity of dinoflagellate symbionts (/ooxamhellae) in a host individual. Mar. Ecol. Prog. Set: 195: 93-100. Chang, S. S., B. B. Prezelin. and R. K. Trench. 1983. Mechanisms ,>| photoadaptation in three strains of the symbiotic dinoflagellate Sym- hiodiniiim microudrialicum. Mar. Biol. 76: 219-224. Coffroth, M. A., H. R. Lasker, and J. Oliver. 1988. Coral mortality outside of the Eastern Pacific during 1982- S3: Relationship to El Nino. Pp. 141-182 in Some Ecological Consct/iicnccs of the 1982/1983 El Niiio Event to Marine Life. P. W. Glynn, ed. Elsevier. Amsterdam. Coffroth, M. A., H. R. Lasker, M. E. Diamond. J. A. Bruenn. and E. Bermingham. 1992. DNA fingerprints of a gorgonian coral: a method for detecting clonal structure in a vegetative species. Mar. Biol. 114: 317-325. Cole, L. C. 1954. The population consequences of life history phenom- ena. Q. Rev. Biol. 29: 103-135. Darius. H. T., C. Dauga, P. A. D. Grimont, E. Chungue, and P. V. Martin. 1998. Diversity in symbiotic dinorlagellates (Pyrrophyta) from seven scleractinian coral species: Restriction enzyme analysis of small subunit rihosoma! RNA genes. J. Eiikaryat. Micrahiol. 45: 619- 627. Davies, J. M., R. P. Dunne, and B. E. Brown. 1997. Coral bleaching and elevated sea-water temperature in Milne Ba> Province. Papua New Guinea, 1996. Mar. Fres/m. Res. 48: 513-516. Davy, S. K., I. A. N. Lucas, and J. R. Turner. 1997. Uptake and persistance of homologous and heterologous zooxanthellae in the tem- perate sea anemone Cereus pedimciilatiis (Pennant). Biol. Bull. 192: 208-216. Drollet, J., H. M. Faucou, and P. M. V. Martin. 1995. Elevated sea-water temperature and solar UV-B flux associated with two suc- cessive coral mass bleaching events in Tahiti. Mar. Freslm: Kes. 46: 1153-1157. Fitt, W. K. 1985. Effect of different strains of the zooxanthella Symbio- diniiun microadriaticum on growth and survival of their coelenterate and molluscan hosts. Proceedings of the Fifth International Coral Reef Congress (Tahiti) 6: 131-136. Fitt, W. K., and R. K. Trench. 1981. Spawning, development, and acquisition of zooxanthellae by Tridacna squamosa (Mollusca. Bi- valvia). Biol. Bull. 161: 213-235. Franzisket, L. 1970. The atrophy of hermatypic reef corals maintained in darkness and their subsequent regeneration in light. hit Rev. Gesam- ten Hyilrohiol. 55: 1-12. Glynn, P. W. 1984. Widespread coral mortality and the 1982-1983 El Nino warming event. Environ. Consen: 11: 133-146 Glynn, P. W. 1991. Coral reef bleaching in the 1980's and possible connections with global warming. Trends l-'n'l. I-' vol. 6: 175-179. Goreau, T. F. 1964. Mass expulsion of zooxanthellae from Jamaican reef communities after Hurricane Finn 5 ience 145: 383-386. Goreau, T. J. 1992. Bleaching and reel community change in Jamaica: 1951-1991. ,4m. Zoo/. 32: 683-695. Goreau, T. J., and R. L. Hayes. 1994. Coral bleaching and ocean "hot spots." Ainhio 23: 176-180 Gnulel. T. L.. and M. A. Coffroih. 1997. A within colony compansnn of zooxanthella genotvrx's M the Caribbean gorgonian Plexaiira kiinn. Proceedings of ihc f-.i^htu Intt rn.ilional Coral Reef Symposium I Pan- ama) 2: 1331-1334, Guillard, R. R. L., and J. H. Ryther. 1962. Studies of marine plank- tonic diatoms. 1: Cycloiella nana Hustedt and Detoniila confervacea (Cleve) Gran. Can. J. \lu rohiul. 8: 229-239. Mill. M.. and I . V\ ik c>\. 19«)8. I Inusual mode of svmbiont repopulation 'ie Oceanographiclnsfiiuion | Library FEB*6 2001 58 R. A. KINZIE ET AL. after bleaching in Anthosigmella various: acquisition of different zoox- amhellae strains. Symbiosis 25: 279-289. Hoegh-Guldberg, O. 1999. Climate Change. Coral Bleaching and the Future of the World's Coral Reefs. Greenpeace, Washington DC. Hoegh-Guldberg. O., and B. Salvat. 1995. Periodic mass-bleaching and elevated sea temperatures: bleaching of outer reef slope communities in Moorea. French Polynesia. Mar. Ecol. Prog. Ser. 121: 181-190. Hoegh-Guldberg, O., L. R. McCloskey, and L. Muscatine. 1987. Ex- pulsion of zooxanthellae by symbiotic cnidarians from the Red Sea. Coral Reefs 5: 201-204. Huppert, A., and L. Stone. 1998. Chaos in the Pacific's coral reef bleaching cycle. Am. Nat. 152: 447-459. Iglesias-Prieto, R. 1997. Temperature-dependent inactivation of photo- system II in symbiotic dinoflagellates. Proceedings of the Eighth In- ternational Coral Reef Symposium (Panama} 2: 1313-1318. Iglesias-Prieto, R., J. L. Malta, W. A. Robis, and R. K. Trench. 1992. Photosynthetic response to elevated temperature in the symbiotic dinoflagellate Symbiodinium microadriaticum in culture. Proc. Nat. Acad. Sci. USA 89: 10302-10305. ISRS (International Society for Reef Studies). 1998. Statement on bleaching. Reef Encounter 24: 19-20. Kinzie, R. A. 1974. Experimental infection of aposymbiotic gorgonian polyps with zooxanthellae. J. E.\p. Mar. Biol. Ecol. 15: 335-345. Kinzie, R. A. Ill, and R. \V. Buddemeier. 1996. Reefs Happen. Global Change Biol. 2: 479-494. Kinzie, R. A. Ill, and G. S. Chee. 1979. The effect of different zoo- xanthellae on the growth of experimentally reinfected hosts. Biol. Bull. 156: 315-327. Knowlton, N., E. Weil, L. A. Weight, and H. M. Guzman. 1992. Sibling species in Montastrea annnlaris. coral bleaching, and the coral climate record. Science 255: 330-333. Kobluk, I). R., and M. A. Lysenko. 1994. "Ring" bleaching in southern Caribbean Agaricia agaricites during rapid water cooling. Bull. Mar. Sci. 54: 142-150. Kushmaro, A., V. Loya, M. Fine, and E. Rosenberg. 1996. Bacterial infection and coral bleaching. Nature 380: 396. Lamed, S. T., and M. J. Atkinson. 1997. Effects of water velocity on NH4 and PO4 uptake and nutrient-limited growth in the macroalga Dictyosphaeria cavemosa. Mar. Ecol. Prog. Ser. 157: 295-302. Lee, J. J., C. G. Wray, and C. Lawrence. 1995. Could foraminiferal zooxanthellae be derived from environmental pools contributed by different coelenterate hosts? Ada Protocol. 34: 75-85. Lesser. M. P.. and J. M. Murk. 1989. Effects of irradiance and ultra- violet radiation on photoadaptation in the zooxanthellae of Aiptasia pallida: primary production, photo-inhibition, and enzymic defenses against oxygen toxicity. Mar. Biol. 102: 243-255. McNally, K. L., N. S. Govind, P. E. Thome, and R. K. Trench. 1994. Smali-subunit ribosomal DNA sequence analysis and a reconstruction of the inferred phylogeny among symbiotic dinoflagellates (Pyrro- phyta). ,/. Phycol. 30: 316-329. Muller-Parker, G. 1984. Dispersal of zooxanthellae on coral reefs by predators on cnidarians. Biol. Bull. 167: 159-167. Muller-Parker. G., and C. F. D'Elia. 1996. Interactions between corals and their symbiotic algae. Pp. 96-1 1 3 in Life and Death of Coral Reefs. C. Birkland, ed. Chapman and Hall, New York. Nakano, Y., K. Yamazato, H. Masuhara, and S. Ito. 1997. Responses of Okinawan reef-building corals to artificial high salinity. Galaxia 13: 181-195. Polne-Fuller, M. 1991. A novel technique for preparation of axenic cultures of Symbiodinium (Pyrrophyta) through selective digestion by amoebae. J. Pliycul. 27: 552-554. Rosenberg, E., and Y. Loya. 1999. \'ihrio sln/oi is the etiological (causative) agent of Oculina patagonica bleaching: General implica- tions. Reef Encounter 25: 8-10. Rowan, R. 1991. Molecular systematics of symbiotic algae. J. Phycol. 27: 661-666. Rowan, R. 1998. Diversity and ecology of zooxanthellae on coral reefs. 7. Plm-ol. 34: 407-417. Rowan, R., and N. Knowlton. 1995. Intraspecific diversity and ecolog- ical zonation in coral-algal symbiosis. Proc. Nat. Acad. Sci. 92: 2850- 2853. Rowan. R., and J. D. Powers. 1991a. A molecular genetic classification of zooxanthellae and the evolution of animal-algal symbioses. Science 251: 1348-1351. Rowan, R., and J. D. Powers. 1991b. Molecular genetic identification of symbiotic dinoflagellates (zooxanthellae). Mar. Ecol. Prog. Ser. 71: 65-73. Rowan, R., N. Knowlton, A. Baker, and J. Jara. 1997. Landscape ecology of algal symbionts creates variation in episodes of coral bleaching. Nature 388: 265-269. Sadler, L. A., K. L. McNally, N. S. Govind, C. F. Brunk, and R. K. Trench. 1992. The nucleotide sequence of the small subunit ribo- somal RNA gene from Symbiodinium pi/osum, a symbiotic dinoflagel- late. Curr. Genet. 21: 409-416. Schoenberg, D. A., and R. K. Trench. 1980. Genetic variation in Symbiodinium (=Gymnodinium) microadriaticum Freudenthal. and specificity in its symbiosis with marine invertebrates. III. Specificity and infectivity of Symbiodinium microadriaticum. Proc. R. Soc. Land. B 207: 445-460. Sehwarz, J. A., D. A. Krupp, and V. M. Weis. 1999. Late larval development and onset of symbiosis in the scleractinian coral Fungia scuraria. Biol. Bull. 196: 70-79. Stanlej, G. D. Jr., and P. K. Swart. 1995. Evolution of the coral- zooxanthellae symbiosis during the Triassic: A geochemical approach. Paleobiology 21: 179-199. Steen, R. G., and L. Muscatine. 1987. Low temperature evokes rapid exocytosis of symbiotic algae by a sea anemone. Biol. Bull. 172: 246-263. Stimson, J., and R. A. Kinzie III. 1991. The temporal pattern and rate of release of zooxanthellae from the reef coral Pocillopora damicornis (Linnaeus) under nitrogen-enrichment and control conditions. 7. Exp. Mar. Biol. Ecol. 153: 63-74. Stochaj. W. R., and A. R. Grossman. 1997. Differences in the protein profiles of cultured and endosymbiotic Symbiotliniiim sp. (Pyrrophyta) from the anemone Aiptasia pallida (Anthozoa). 7. Phycol. 33: 44-58. Taylor. D. L. 1974. Symbiotic marine algae: taxonomy and biological fitness. Pp. 245-262 in Symbiosis in the Sea. W. B. Vernberg. ed. University of South Carolina Press. Columbia. SC. Tsuchiya, M. 1999. Warning from the coral reefs. Galaxia 1: 27-29 (in Japanese and English). Warner, M. E., W. K. Fitt, and G. W. Schmidt. 1996. The effects of elevated temperature on the photosynthetic efficiency of zooxanthellae in hospite from four different species of reef coral: a novel approach. Plant Cell Em iron. 19: 291-299. Warner. M. E.. W. K. Fitt. and G. W. Schmidt. 1999. Damage to photosystem II in symbiotic dinoflagellates: A determinant of coral bleaching. Proc. Nat. Acad. Sci. 96: 8007-8012. Wilson, A. D. 1999. Coral reefs: A threatened resource. Sea Techno/. 40: 10-15. Zardoya, R., E. Costas. V. Lopez-Rodas. A. Garrido-Pertierra, and J. M. Bautisa. 1995. Revised dinoflagellate phylogeny inferred from molecular analysis of large-subunit ribosomal RNA gene sequences. 7. Mol. Evol. 41: 637-645. Reference: Bio/. Bull. 200: 59-66. (February 2001) Role of Aerobic and Anaerobic Circular Mantle Muscle Fibers in Swimming Squid: Electromyography IAN K. BARTOL Department of Orgaiiismic Biology. Ecology, and Evolution, University of California, Los Angeles, 621 Charles E. Young Drive South, Los Angeles, California 90095-1606 Abstract. Circular mantle muscle of squids and cuttle- fishes consists of distinct zones of aerobic and anaerobic muscle fibers that are thought to have functional roles analogous to red and white muscle in fishes. To test predic- tions of the functional role of the circular muscle zones during swimming, electromyograms (EMGs) in conjunction with video footage were recorded from brief squid Lolli- gitncnla hrevis (5.0-6.8 cm dorsal mantle length, 10.9- 18.3 g) swimming in a flume at speeds of 3-27 cm s '. In one set of experiments, in which EMGs were recorded from electrodes intersecting both the central anaerobic and pe- ripheral aerobic circular mantle muscles, electrical activity was detected during each mantle contraction at all swim- ming speeds, and the amplitude and frequency of responses increased with speed. In another set of experiments, in which EMGs were recorded from electrodes placed in the central anaerobic circular muscle fibers alone, electrical activity was not detected during mantle contraction until speeds of about 15 cm s~', when EMG activity was spo- radic. At speeds greater than 15 cm s ', the frequency of central circular muscle activity subsequently increased with swimming speed until maximum speeds of 21-27 cm s~', when muscular activity coincided with the majority of man- tle contractions. These results indicate that peripheral aero- bic circular muscle is used for low, intermediate, and prob- ably high speeds, whereas central anaerobic circular muscle is recruited at intermediate speeds and used progressively more with speed for powerful, unsteady jetting. This is significant because it suggests that there is specialization and efficient use of locomotive muscle in squids. Received 16 June 2000; accepted ikbartol@lifesci.ucla.edu 10 October 2000. E-mail: Introduction Fishes have three myotomal muscle types, that is, red. pink, and white fibers, that are structurally, metabolically. and functionally distinct. The red muscle is characterized by a high myoglobin content, extensive capillary beds, a high density of mitochondria, and high levels of oxidative en- zymes (Bone. 1966; 1978). It operates aerobically, fatigues slowly, has low rates of activation and relaxation, and has low shortening velocities (Johnston et ai, 1977). White muscle is characterized by a high myofibrillar density, a low content of mitochondria, poor blood supply, and high levels of glycolytic enzymes (Flitney and Johnston, 1979). It op- erates anaerobically, fatigues rapidly, has high rates of activation and relaxation, and has high shortening velocities (Curtin and Woledge. 1988). Pink muscle is intermediate in structure, metabolism, and contractile properties between red and white muscle (Johnston et a!., 1977: Coughlin et al.. 1996). Generally, red muscle is recruited for the lowest, steady, undulatory swimming speeds (Rome et al,, 1984): pink muscle is recruited at intermediate swimming speeds (Johnston et al., 1977): both red and pink muscle are re- cruited at maximum steady swimming speeds (Coughlin et al.. 1996): and white muscle is recruited at higher r.wim- ming speeds when swimming becomes unsteady (Rome et al., 1984: Juyne and Lauder, 1996). Squids and cuttlefishes have muscle types in the mantle musculature similar to red and vhite muscle. The mantle musculature of cephalopods functions as a constant volume system with a three-dimensional array of tightly bundled muscles commonly called a inuscular-hydrostat (Kier and Smith, 1985; Smith and Kier, 1989). During jet propulsion, circular muscles \\iihin the mantle contract, which conse- quently decreases mantle diameter and forces intramantle water out of the funnel. Because changes in mantle length during contraction are negligible and the mantle muscula- 59 60 I. K. BARTOL ture is constant in volume, contraction of circular muscle results in concomitant thickening of the mantle wall. Obliquely oriented collagen fibers traversing the thickness of the mantle wall are strained and store energy. This energy together with contraction of radially oriented muscle fibers, which extend from the inner to outer surface of the mantle, power mantle refilling (Gosline and Shadwick, 1983: Gos- line et al., 1983; Kier, 1988). The circular muscle respon- sible for mantle contraction within this system consists of three anatomically and metabolically distinct muscle layers when viewed in cross section: an inner, outer, and middle layer (Bone et al., 1981: Mommsen et al.. 1981 ). The inner and outer (peripheral) layers are thin and rich in mitochon- dria; they have high succinic dehydrogenase (SDH) activity, a high ratio of oxidative to glycolytic enzyme activity, and heavy vascularization. The middle (central) layer is thick and poor in mitochondria; it has low SDH activity, a high ratio of glycolytic to oxidative enzyme activity, and sparse vascularization. Because of these structural and biochemical differences. Bone et al. ( 1981 ) and Mommsen el al. ( 1981 ) suggested that the inner and outer layers are analogous to aerobic red muscle in fish and are used for slow, steady swimming and rhythmic, respiratory contractions, while the central layer is analogous to anaerobic white muscle and is used during burst swimming, escape, and capture events. This is consistent with the classic view (Wilson, 1960) that squid switch between quiet "respiratory" jetting and giant fiber escape responses. However, O'Dor (1982, 1988a) and Bartol ( 1999) have reported fully graded swimming in squid swim-tunnel studies, and Otis and Gilly (1990) and Preuss and Gilly (2000) have shown that interplay between giant and non-giant axon systems, which innervate the circular mantle musculature of squid, allow fpr considerable flexi- bility in jetting behavior and may facilitate smooth speed transitions. Although peripheral and central circular muscle have several structural and biochemical similarities to red and white muscle in fishes, the functional roles of circular mus- cle types during swimming in squids and cuttlefishes remain unresolved. Bone et al. (1994) found large-amplitude elec- tromyographic activity in the mantle of cuttlefish Sepia officinulis during escape jetting and concluded that both the central anaerobic and peripheral aerobic circular muscles were active. However, since electrodes were not inserted exclusively in either muscle layer, the functional role of the circular muscle types is still uncertain. Accumulations of metabolic end products within mantle musculature of squids during routine behavior, exercise, and after exercise have been the focus of many studies (Hochachka et al.. 1975: Grieshaber and Gade, 1976; Storey and Storey, 1978: Fields and Quinn, 1981; Portner et al.. 1991, 1993; Finke et al.. 1996), but no distinctions were made in those studies be- tween end products found within the various layers of circular muscle. On the basis of mantle pressure measure- ments. O'Dor ( 1988b) concluded that both types of circular muscle fiber must be active in Illex illecebrosits at subcriti- cal swimming speeds to generate the necessary swimming power. Furthermore. Finke et al. (1996) suggested that Lolliguncula brevis, which tends to oscillate between high and low muscular activity at increased swimming velocities, relies on both types of circular muscle fiber at high speeds. However, once again, direct documentation of the func- tional roles of the two circular mantle muscle fibers is lacking. In this study electromyograms (EMGs) were collected in conjunction with video footage to determine the role of the peripheral aerobic and central anaerobic circular muscle layers in brief squid Lolliguncula brevis swimming over a range of speeds. Two experiments are reported: in one, EMG activity was recorded from electrodes traversing both peripheral and central circular muscle layers; in the other, EMG activity was recorded from electrodes embedded ex- clusively in the central circular muscle layer. It was not possible to embed electrodes solely in peripheral circular muscle because the muscle layer was too thin (—0.1 mm) for reliable placement. However, by comparing results from the two experiments, it was possible to postulate the func- tional roles of the circular muscle zones. Materials and Methods Experimental animals In August-October 1998. brief squid Lolliguncula brevis (Blainville) were captured by trawl within embay ments along the seaside of Virginia's Eastern Shore and within the York River. Virginia. Squid captured along the Eastern Shore of Virginia were transported to the Virginia Institute of Marine Science (VIMS) Eastern Shore Laboratory in Wachapreague. Virginia; squid captured in the York River were transported to the VIMS main campus in Gloucester Point. Virginia. Squid were kept alive in the field using 1 20-quart coolers equipped with filtration and aeration sys- tems, which were powered by 12-V, sealed, gel-cell batter- ies. At Wachapreague and Gloucester Point, squid were kept in flow-through raceway tanks for at least one week prior to experimentation and fed a diet of grass shrimp Palaemonetes pugio. Experiments on 12 squid (3.5-7.8 cm dorsal mantle length [DML], 4.3-27.6 g) were performed, but results from only 5 squid (5.0-6.8 cm DML. 10.9-18.3 g) are reported here. Some squid were eliminated from consideration because of one or more of the following: ( 1 ) the location of the electrodes in the muscle was somewhat uncertain. (2) ambient electrical noise was excessive. (3) the squid was not cooperative in swim tunnels. Electromyogram {EMG) recordings Disposable, paired hook- wire electrodes ( 1 50-jum-gauge insulated nickel alloy, Nicolet Biomedical. Madison. WI) ROLE OF SQUID CIRCULAR MUSCLE FIBERS 61 were used for EMG recordings. The hook portions of the manufactured electrode wires were too large for recording EMGs from the peripheral and central circular muscle lay- ers, which were about 0.1 mm and 1.0 mm thick, respec- tively. Therefore, the hook portions of the electrodes were modified so that they were 1.0 mm in total length with 0.5 mm of insulation removed at the tip. The squid were anes- thetized (~2 mini in an isotonic solution of MgCl2 (7.5% MgCK • 6H:O) and seawater (Messenger el til., 1985). and each pair of electrodes was inserted obliquely, using a hypodermic needle, into the lateral mantle wall at a point 60% of the mantle length from posterior. Electrode spacing within the muscle was about 2-6 mm. Insertion of two pairs of electrodes for simultaneous recording from different muscle groups was attempted on several occasions, but these attempts were unsuccessful because dissections re- vealed that at least one pair of electrodes was not in target muscle fibers or the two sets of electrodes interfered with swimming. Thus, recordings from only one pair of elec- trodes during each experiment are reported. The electrode pair from which EMGs were recorded was embedded in either the central circular muscle layer or both the central circular muscle layer and the peripheral circular muscle layer adjacent to the skin. It was not possible to embed elec- trodes exclusively in the peripheral layer of circular muscle because it was too thin (—0.1 mm) for hook-wire electrodes. Fine wire needle electrodes like those described in Kier et al. ( 1989). which allow for very precise electrode placement, were not used because of difficulties associated with anchoring the electrodes to the mantle of L brevis and the high probability that anchoring tools (e.g., clamping blocks) would disrupt swimming. During electrode placement, no attempt was made to avoid radial muscles, which extend from the inner to outer mantle surface and partition the circular muscle into 0. 1 -mm sections. Electrical activity of radial muscles, however, did not obscure electrical activity from the circular muscle layers. because radial muscles are active during mantle expansion, whereas circular muscles are active during mantle contraction (Gosline and Shadwick, 1983; Gosline et at., 1983). To help ensure reliable placement of electrodes, many practice inser- tions were performed on preserved squid prior to the experi- ments, and depth references were marked on hypodermic nee- dles used for implantation. Sufficient slack in the electrode wires was provided so that anchoring electrodes at other loca- tions along the body was not necessary to prevent dislodgment. After successful implantation, squid were placed in a 20 X 10 X 10 cm holding section within a 16-1 recirculating water tunnel (Vogel and LaBarbera, 1978) filled with aerated seawa- ter (24%o, 22 °C) and allowed to recover from surgery at low flow velocity (1 cm s :). Flow velocity in the tunnel was controlled using two propellers arranged in a rotor-stator con- figuration and a 14 hp variable-speed motor, which was shielded using aluminum sheeting to reduce radiated electro- magnetic noise. A ground electrode was attached to a down- stream collimator within the flume. The ground electrode to- gether with the embedded bipolar hook-wire electrodes, which were attached to insulated Nicolet micrograbbers, were plugged into a two-channel differential amplifier using Nicolet DIN 42 802 connectors. The differential amplifier was part of a Nicolet Compass II PortaBook system, which is designed to record electromyograms and somatosensory, auditory, and vi- sual evoked potentials from human patients. The PortaBook was used to record, amplify, and filter electromyograms col- lected at various frequencies. For this study, EMGs were collected in the free-run mode using a bandpass filter of 20- 10,000 Hz, which was the most effective Nicolet filter avail- able for electromyographic recordings, and all EMG traces were recorded on the hard drive. After the squid became active in the holding section, flow velocity within the flume was increased to 6 cm s~ ' and the animal was allowed to acclimate to flow, which generally occurred within 15 min. After the acclimation period, squid were exposed to a flow velocity of 3 cm s~' for 10 min. Speed was subsequently increased by 3 cm s~' every 10 min until the squid could no longer keep pace with free- stream flow. During each of the 10-min speed increments, EMGs were collected. Because of limitations of the Porta- Book, it was not possible to record EMG signals on a channel of a video recorder or precisely timelock video and EMG recordings. Therefore, to provide a synchronized record of EMG responses and swimming behavior, the computer screen of the PortaBook was positioned below the holding section of the flume, and both Nicolet display output and swimming behavior of the squid were video- taped simultaneously from a lateral perspective using a Sony Hi-8 video camera. After squid were exhausted and could no longer maintain free-stream velocity, they were removed from the flume and over-anesthetized in an iso- tonic solution of MgCI2 (7.51 MgCl2 • 6H;O) and seawater. The squid then were transferred to 10% buffered formalin for later dissection to determine precise electrode location. Some cooperative squid were not euthanized after the initial recording session. Instead, the electrode wires were cut. leaving the embedded section undisturbed within the mantle so that it could be examined later to determine placement, and the squid were returned to the raceway tanks. After several hours these squid were retrieved, electrodes were embedded in a different circular muscle layer, and EMGs were recorded during another swimming session. Following the completion of the second swimming session, these an- imals were over-anesthetized, fixed, and dissected as de- scribed above to determine c ectrode placement. Kinematic (inul\\i\ Video footage of trials in which squid were cooperative at all swimming speeds, ambient electrical noise was low. and electrode placement was unambiguous was analyzed using a 62 I. K. BARTOL Sony EVO-9700 editing deck and a Peak Motus video and computer motion measurement system (Peak Performance Technologies, Englewood, CO). At each speed several video sequences with clear mantle profiles and representa- tive EMG recordings were selected. Mantle diameter at the location of electrode placement, swimming velocity, and acceleration of the squid were measured for each frame of video (30 frames s~') within the sequences, and frames where EMG responses began and ended were marked using the event feature in the Peak Motus system. Mantle diam- eter, velocity, and acceleration data were smoothed using a fourth-order Butterworth filter to account for video jitter and digitization error. Peak video data and Nicolet EMG data were imported into Microsoft Excel and aligned using event markers from the Peak file and timebase frequencies from both files. Synchronization of the data was reliable only to within 33.3 ms because the video camera records at 30 frames s~'. Since the objective of this study was simply to determine the presence or absence of EMG activity within the circular muscle layers during mantle contractions, more precise analysis of the timing of EMG activity was unnec- essary. Results The five specimens of Lolliguncula brevis (5.0-6.8 cm DML) considered in this study matched free-stream flow well at speeds less than 2 1 cm s~ ', but at higher speeds they had some difficulty matching flow velocity and eventually collapsed against the downstream weir. Prior to collapse, the squid accelerated and decelerated erratically, and the contraction frequencies were more irregular than those ob- served at lower velocities. At low (3-9 cm s~') and inter- mediate (12-18 cm s"1) swimming velocities, squid relied on both fin and jet propulsion, whereas at high subcritical speeds (21-27 cm s~'). they relied exclusively on jet pro- pulsion and, like Loligo opalescens and Ille.x illecebrosus (O'Dor. 1988a), wrapped their fins against the mantle. The electrodes did not appear to impede swimming in the five squid considered; however, some squid, which were ex- cluded from this analysis, were visibly agitated after sur- gery, swam inconsistently at most swimming speeds, and frequently swam forcibly into the flume sides. The amplitude and frequency of electrical activity varied from animal to animal, but the basic patterns of activity presented in this paper are representative of all the squid considered. Results from a squid with electrodes inserted into the central and peripheral circular muscle layers (Fig. 1 ) show significant EMG activity at low (6 cm s~'), interme- diate (15 cm s~'), and high speeds (24 cm"1); this activity was correlated with mantle contraction. The amplitude of EMG activity increased with speed and, as was typical of other squid examined, frequency of mantle contraction was greater at the highest speed tested (24 cm s"1) than at low speed (6 cm s '). Moreover, maximum accelerations throughout the jet cycle were closely coupled with high- magnitude EMG bursts recorded during mantle contraction. At high speeds (24 cm s"1). lower amplitude electrical activity sometimes occurred during mantle expansion. Electromyographic recordings from a squid with elec- trodes embedded in only central circular muscle (Fig. 2), but with settings for bandpass filter, sensitivity, timebase, etc., identical to those used for recordings from both central and peripheral layers, show no EMG activity during mantle expansion and contraction at 6 cm s~'. At an intermediate speed of 15 cm s~', occasional EMG activity was detected. (The mean central circular muscle recruitment speed for all squid considered was 15.3 ± 3.7 [SD] cm s~'.) Most of the large-amplitude electrical activity at intermediate speeds occurred during the contraction phase of the jet cycle, but sometimes smaller amplitude activity occurred during man- tle expansion (Fig. 2). Furthermore, the large-amplitude EMGs at intermediate speeds were coupled with large changes in mantle diameter and high peaks of acceleration. Generally, the frequency of EMG bursts increased with speed up to maximum speeds of 21-27 cm s~', when muscular activity coincided with the majority of mantle contractions. Again large-amplitude signals correlated with mantle contraction, whereas less frequent low-amplitude signals correlated with mantle expansion. Contraction am- plitude and frequency were more irregular at 24 cm s~ ' than at lower velocities, and high-amplitude EMG peaks were strongly coupled with high accelerations. Discussion The results of this study indicate that the peripheral aerobic and the central anaerobic circular muscle fibers in squid probably have distinct functional roles during swim- ming. Electrodes implanted in both peripheral and central circular muscle fibers recorded EMG activity with each mantle contraction at all swimming speeds (3-27 cm s"1). However, electrodes implanted exclusively in central (an- aerobic) muscle did not record EMG activity during mantle contraction until speeds of about 15 cm s~', when activity was sporadic. At speeds above 15 cm s \ the frequency of electrical activity increased with swimming speed up to maximum speeds of 21-27 cm s 1, when anaerobic mus- cular activity coincided with the majority of mantle con- tractions and with high accelerations. These results indicate that peripheral aerobic circular muscle is used for slow, intermediate, and probably high speeds, whereas central anaerobic muscle is recruited at intermediate speeds and used progressively more with increasing speed for power- ful, unsteady jetting typical of higher speed swimming. The observed periodic detection of central anaerobic cir- cular muscle activity at intermediate speeds is interesting given that anaerobic muscle is often associated only with ROLE OF SQUID CIRCULAR MUSCLE FIBERS 15cm s'1 63 0 00 0 25 0 50 0 75 1 00 1 25 1 50 1 75 2 00 E245 o • S 235 OJ £ 230 S2: m 220 0 00 0 25 0 50 0 75 1 00 1 25 1 50 1 75 2 00 "» 100 §80 _ 60 C 40 O 20 • c5 o o3 -2I 0> "° O -60 - 0 00 0 25 0.50 0 75 1 00 1 25 1 50 1 75 2 00 24 cm s" 000 025 050 075 100 125 150 175 200 Time (s) 120 80 60 40 20 0 -20 • -40 -60 -80 100 0 00 025 0 50 0 75 1 00 1 25 1 50 1 75 2 00 000 025 050 075 100 125 150 175 200 0 00 0 25 0 50 0 75 1 00 125 1 50 175 2 00 Time (s) 120 100 80 60 40 20 0 -20 -40 -60 -80 100 000 025 050 075 1 00 1 25 1 50 1 75 2 00 0 00 0 25 0 50 0 75 1 00 1 .25 1 50 1 75 2 00 Time (s) Figure 1. Electromyograms recorded from electrodes embedded in both peripheral aerobic circular muscle and central anaerobic circular muscle of a brief squid (5.5 cm dorsal mantle length) while swimming at 6, 15, and 24 cm s" '. Mantle diameter and acceleration are plotted underneath each recording. Electrical activity occurs during each mantle contraction, and at high speeds electrical activity is sometimes present during mantle expansion. burst swimming in fishes (Rome el al., 1984; Jayne and Lauder, 1996) and escape jets in cephalopods (Bone et al., 1981. 1994; Mommsen et al.. 1981). Finke et al. (1996) determined that anaerobic end products, such as a-glycero- phosphate. succinate. and octopine. begin to accumulate in Lolliguncula brevis at speeds of 1.5-2.0 mantle lengths s~' (8.3-1 1 .0 cm s" ' for a specimen of 5.5-cm DML), suggest- ing that anaerobic metabolism may occur in the mantle at low intermediate speeds. Moreover, intramantle pressure records revealed that at speeds of 3.2 mantle lengths s~' (17.6 cm s~' for a specimen of 5.5-cm DML) high mantle pressures comparable to those recorded at speeds of 4.3 mantle lengths s~' (23.7 cm s~' for a specimen of 5.5-cm DML) are periodically generated (Finke et al., 1996). Ac- cumulation of anaerobic end products and the occasional detection of high intramantle pressures at intermediate speeds are consistent with the activity of anaerobic circular muscle observed at intermediate speeds in this study. Peri- odic anaerobic circular mantle activity at intermediate speeds boosted power production, which helped squid keep pace with free-stream flow in the tunnels. Periodic anaero- bic activity was probably not energetically deleterious since anaerobic metabolic changes can be rapidly reversed in squid (Portner et al., 1993). At higher speeds, central an- aerobic circular muscle activity was more frequent, but there was still some oscillation between aerobic and anaer- obic muscular activity. Finke et al. (1996) suggest that oscillating between aerobic and anaerobic circular muscle recruitment rather than simply relying exclusively on anaer- obic muscle at a critical speed allows for an extended net use of anaerobic resources before fatigue sets in. One limitation of this study was the level of background noise (±40 /J.V at between f>() and 120 Hz) in EMG record- ings despite attempts to reduce such interference by shield- ing the variable-speed flume motor and performing trials in laboratories with minimal electrical equipment. The noise probably came from interference produced by the flume motor and other electrical laboratory equipment (e.g., fluo- rescent lights, proportional controllers, and rheostats) and was allowed through the 20-10.000-Hz bandpass filter. A 64 I. K. BARTOL 250 200 150 I 100 50 •50 -100 -150 •200 -250 I -300 I •350 I -400 J— 000 6 cm s 0 25 0 50 0 75 1 00 1 25 1 50 1 75 2 00 025 050 075 100 125 150 175 200 120 100 80 60 40 20 0 -20 -40 -60 -80 -100 0 00 0 25 0 50 0 75 1 00 1 25 1 50 1 75 2 00 24cm s 255 250 245 240 235 230 225 0 00 0 25 0 50 0 75 1 00 1 25 1 50 1 75 2 00 Time (s) 220 - 0 00 0 25 0 50 0 75 1 00 1 25 1 50 1 75 2 00 Time (s) [20 100 40 20 0 •20 •40 •60 •80 |0 0 00 0 25 0 50 0 75 1 00 1 25 1 50 1 75 2 00 Time (s) 000 025 050 075 100 125 150 175 200 0 00 0 25 0 50 0 75 1 00 1 25 1 50 1 75 2 00 120 100 80 60 40 20 0 -20 •40 60 0 00 0 25 0 50 0 75 1 00 1 25 1 50 1 75 2 00 Figure 2. Electromyograms recorded from electrodes embedded in central anaerobic circular muscle of a brief squid (5.7 cm dorsal mantle length) while swimming at 6, 15, and 24 cm s"'. Mantle diameter and acceleration are plotted underneath each recording. There are no obvious waveforms at 6 cm s~'. occasional waveforms at 15 cm s~', and frequent waveforms at high speeds (24 cm s ' ). At high speeds, high-amplitude electrical activity occurs during each mantle contraction, and lower amplitude electrical activity sometimes occurs during mantle expansion. Furthermore, mantle contractions are more erratic at 24 cm s"1 than at 6 cm s~'. significant proportion of the noise might have been removed with a 60-Hz notch filter (Loeb and Guns. 1986), but un- fortunately the electromyographic system used could not be configured with a notch filter. Considering the high level of background noise, it is possible that some low-level EMG activity was undetected. However, it is unlikely that anaer- obic EMG activity was absent at low speeds because it was concealed in the noise. EMG activity from the thin periph- eral (aerobic) muscle zones was easily distinguishable and consistently recorded at low speeds when muscular activity was lowest in spite of the electrical noise, and thus EMG activity from the thick, powerful central (anaerobic) muscle zone should have been visible if those fibers were active. Although electrical noise probably did not mask circular muscle activity, which was the focus of this study, it inter- fered with detection of EMG activity from the smaller radial muscles. Dissections revealed that electrodes crossed radial muscle bands, which are active during mantle expansion and hyperinflation (i.e., a sharp increase in mantle diameter just prior to contraction) (Gosline et al, 1983; Bone et ci/.. 1994). However, no radial EMG activity was detected at low speeds. This is not surprising given that Wilson ( 1960) and Ward (1972). who were the first to record electrical activity in the squid mantle musculature, were unable to detect radial muscle activity, and Gosline et al. (1983) and Bone et al. (1994) were able to detect consistent radial muscle activity in squid and cuttlefish only when using a 60-Hz notch filter and a Faraday cage, respectively. Radial muscle activity at low speeds detected by Gosline et al. (1983) and Bone et al. (1994), who examined squid and cuttlefish in dissecting pans and small aquaria and conse- quently did not have to contend with a motor-driven swim tunnel, was often less than 40 ^.V. Therefore, low-speed radial muscle activity presumably was hidden by the elec- trical noise. From squid engaged in escape jets, Gosline et ul. (1983). without using a 60-Hz notch filter, recorded ROLE OF SQUID CIRCULAR MUSCLE FIBERS radial muscle activity during refilling and hyperinflation. At high speeds, EMG activity frequently was observed in the present study during mantle expansion. Since circular mus- cle is not active during mantle expansion (Gosline et nl.. 1983), radial muscles — the only other muscle into which electrodes were imbedded — were probably responsible for the EMG activity observed during mantle refilling. The detection of radial muscle activity at high speeds when anaerobic circular muscle was active suggests that signifi- cant radial muscle activity is used to expand the mantle during vigorous jetting. Elastic energy stored within the connective tissue of the mantle also probably plays an important role during mantle expansion (Gosline and Shad- wick, 1983: Gosline et ai, 1983; Pabst. 1996). Using electromyography, Kier et al. (1989) demonstrated that anaerobic and aerobic zones of muscle in cuttlefish Sepia officinal is fins have distinct functional roles, with aerobic fibers responsible for gentle fin movements and anaerobic muscles responsible for vigorous fin movements and support for that movement. The results of the present study indicate that peripheral aerobic circular muscle and central anaerobic circular muscle in brief squid Lolliguncula brevis also have distinct functional roles that are analogous to red and white myotomal fibers in fishes. Contrary to the situation in many fishes, however, in squids the anaerobic muscle appears to be used at subcritical speeds. The dis- covery of functional "gears" in squids, which is yet another instance of convergent evolution in fishes and cephalopods, is significant because it suggests that there is specialization and efficient use of locomotive muscle in squids. Future study on the innervation and neural control of these differ- ent muscle fibers during unrestrained swimming should provide further insight into the locomotive system of cepha- lopods. Acknowledgments I gratefully acknowledge intellectual input from R. Mann and M. R. Patterson and electromyographic advice from S. L. Sanderson and W. M. Kier. I thank M. Vecchione and M. Luckenbach for constructive criticisms of earlier drafts, and I am especially appreciative of the generosity of M. Lenhardt, who allowed me to use his Nicolet Compass PortaBook II. This study was supported in part by the VIMS/SMS Minor Research Grants, the Eastern Shore Sea- side Corporate Scholarship, and Newport News Shipbuild- ing. Financial support during the writing phase of this project was provided by the Office of Naval Research under contract NOOO 14-96-0607. Literature Cited Bartol. I. K. 1999. Distribution, Swimming Ph\siolog\. anil Si\iiiu>iinx Mechanics uf the Brief Si/ttid Lolliguncula brevis. Ph.D. dissertation. College of William and Mary, Virginia Institute of Marine Science. Gloucester Point. VA. 215 pp. Bone, Q. 1966. On the function of the two types of myotomal muscle fibers in elasmobranch fish. J. Mar. Biol. Assoc. UK 46: 321-349. Bone, Q. 1978. Locomotor muscle. Pp. 361-424 in Fish Physiology, vol. 7. W. S. Hoar and D. J. Randall, eds. Academic Press. New York. Bone, Q., A. Pulsford, and A. D. Chub. 1981. Squid mantle muscle. J. Mar. Biol. 61: 327-342. Bone, Q.. K. R. Brown, and G. Travers. 1994. On the respiratory flow in the cuttlefish Sepia officinalis. J. £.\y>. Biol. 194: 153-165. Coughlin, D. J., L. Valdes, and L. C. Rome. 1996. Contraction dynam- ics and power production ot pink muscle of scup (Stenotomus chry- sops}. ./. E.\p. Biol. 199: 2703-2712. Curtin, N. A., and R. C. Woledge. 1988. Power output and force- velocity relationships of live fibers from white myotomal muscle of the dogfish. Scyliorhimis caniculu. J. E\p. Biol. 140: 187-197. Fields. J. H. A., and J. F. Quinn. 1981. Some theoretical considerations on cytosolic redox balance during anaerobiosis in marine invertebrates. J. Theor Biol. 88: 35-45. Finke, E., H. O. Portner, P. G. Lee, and D. M. Webber. 1996. Squid (Lollignnciila hrevis) life in shallow waters: oxygen limitation of metabolism and swimming performance. J. E.\p. Biol. 199: 911-921. Flitney, F. W., and I. A. Johnston. 1979. Mechanical properties of isolated fish red and white muscle fibers. J. Physiol. 295: 49-50. Gosline, J. M., and R. E. Shadwick. 1983. The role of elastic energy storage mechanisms in swimming: an analysis of mantle elasticity in escape jetting in the squid. Loligo opalescens. Can. J. Zool. 61: 1421-1431. Gosline. J. M., J. D. Sleeves, A. D. Harman, and E. DeMont. 1983. Patterns of circular and radial mantle muscle activity in respiration and jetting of the squid Loligo opalescens. J. Exp. Biol. 104: 97-109. Grieshaber. M. E., and G. Gade. 1976. The biological role of octopine in the squid, Loligo vulgaris (Lamarck). J. Comp. Physiol. 108: 225- 232. Hochachka, P. W.. T. W. Moon, T. Mustafa, and K. B. Storey. 1975. Metabolic sources of power for mantle muscle of a fast swimming squid. Comp. Biocliem. Physiol. 52B: 151-158. Jayne, B. C., and G. V. Lauder. 1996. New data on axial locomotion in fishes: how speed affects diversity of kinematics and motor patterns. Am. Zool. 36: 642-655. Johnston, I. A.. W. Davison, and G. Goldspink. 1977. Energy metab- olism of carp swimming muscles. J. Comp. Physio/. 114: 203-216. Kier, VV. M. 1988. The arrangement and function of molluscan muscle. Pp. 211-252 in The Mollnsca. Form and Function, vol. 11, E. R. Trueman and M. R. Clarke, eds. Academic Press. New York. Kier, W. M., and K. K. Smith. 1985. Tongues, tentacles and trunks: the biomechanics of movement in muscular hydrostats. Zool. J. Linn. Soc. 83: 307-324. Kier, W. M., K. K. Smith, and J. A. Miyan. 1989. Electromyography of the tin musculature of the cuttlefish Sepia officinulis. J. E.\p. Biol. 143: 17-31. Loeb, G. E., and C. (Jans. 1986. F.i* ••. raphy for Experimental- ists. University of Chicago Press. ( Messenger, J. B., M. Nixon, and K. }'. i<;.;i!i. i985. Magnesium chlo- ride as an anaesthetic for cephalop'"'- G mp Hiochein. Physiol. 82C: 203-205. Mommsen, T. P., J. Ballantyne, !>. Mat-Donald, J. Gosline, and P. VV. Hochachka. 1981. Analo t J ."id white muscle in squid mantle. Proc. Null. Acad Si , .' SA 78: 3274-3278. O'Dor. R. K. 1982. The respiratory metabolism and swimming perfor- mance of the squid. L.ilixo opalescens. Can. ./. Fish. Ai/nai. Sci. 39: 580-5X7 O'Dor, R. K. 1988a. The forces acting on swimming squid. J. Exp. Biol. 137: 421-442. 66 I. K. BARTOL O'Dor, R. K. 1988b. Limitations on locomotor performance in squid. J. Appl. Physiol. 64: 128-134. Otis, T. S., and W. F. Gilly. 1990. Jel-propelled escape in the squid Loligo opa/escens: concerted control by giant and non-giant motor axon pathways. Proc. Nat/. Acad. Sci. USA 87: 291 1-2915. Pabst, D. A. 1996. Springs in swimming animals. Am. Zoo/. 36: 723- 735. Portner, H. O., D. M. Webber, R. G. Boutilier, and R. K. O'Dor. 1991. Acid-base regulation in exercising squid (Hlex il/ecehrosux, Loligo pea/eh. Am. J. Physio/. 261: R239-R246. Portner, H. O., D. M. Webber, R. K. O'Dor, and R. G. Boutilier. 1993. Metabolism and energetics in squid (lltex illecebrosus, Loligo pealei) during muscular fatigue and recovery. Am J. Physiol. 265: R157-R165. Preuss, T., and W. F. Gilly. 2000. Role of prey-capture experience in the development of the escape response in the squid Loligo opalescens: a physiological correlate in an identified neuron. J. E.\p. Biol. 203: 559-565. Rome, L. C., P. T. Loughna, and G. Goldspink. 1984. Muscle fiber activity of carp as a function of swimming speed and muscle temper- ature. Am. J. Physiol. 247: R272-R279. Smith, K. K., and W. M. Kier. 1989. Trunks, tongues and tentacles: moving with skeletons of muscle. Am. Sci. 77: 28-35. Storey, K. B., and J. M. Storey. 1978. Energy metabolism in the mantle muscle of the squid. Loligo pealeii. J. Comp. Phvsiol. 123: 169-175. Vogel, S., and M. LaBarbera. 1978. Simple flow tanks for research and teaching. BioScience 10: 638-643. Ward, D. V. 1972. Locomotory function of the squid mantle. J. Zool. (Land). 167: 487-499. Wilson, D. M. 1960. Nervous control of movement in cephalopods. J. Exp. Biol. 37: 57-72. Reference: Biol. Bull 200: 67-76. (February 2001) Temperature Effects on Hemocyanin Oxygen Binding in an Antarctic Cephalopod S. ZIELINSKI, F. J. SARTORIS, AND H. O. PORTNER* Alfred-Wegener-Institute for Marine and Polar Research, Ecophysiology and Ecotoxicology, Colwnbusstrasse, 27568 Bremerhaven, Gennanv Abstract. The functional relevance of oxygen transport by hemocyanin of the Antarctic octopod Megaleledone se- noi and of the eurythermal cuttlefish Sepia officinalis was analyzed by continuous and simultaneous recordings of changes in pH and hemocyanin oxygen saturation in whole blood at various temperatures. These data were compared to literature data on other temperate and cold-water cephalo- pods (octopods and giant squid). In S. officinalis, the oxygen affinity of hemocyanin changed at AP50/°C = 0.12 kPa (pH 7.4) with increasing temperatures; this is similar to observations in temperate octopods. In M. senoi. thermal sensitivity was much smaller (<0.01 kPa, pH 7.2). Furthermore, M. senoi hemocyanin displayed one of the highest levels of oxygen affinity (P30 < 1 kPa, pH 7.6, 0 °C) found so far in cephalopods and a rather low cooperativity (/!50 = 1 .4 at 0 °C). The pH sensitivity of oxygen binding (A log PSO/A pH) increased with increasing temperature in both the cuttlefish and the Antarctic octopod. At low PQ-, ( 1.0 kPa) and pH (7.2), the presence of a large venous oxygen reserve (43% saturation) insensitive to pH reflects reduced pH sensitivity and high oxygen affinity in M. senoi hemocyanin at 0 °C. In S. officinalis, this reserve was 19% at pH 7.4, 20 °C, and 1.7 kPa O2, a level still higher than in squid. These findings suggest that the lower metabolic rate of octopods and cuttlefish compared to squid is reflected in less pH-dependent oxygen transport. Results of the hemocyanin analysis for the Antarctic octopod were similar to those reported for Vampyroteuthis — an extremely high oxygen affinity supporting a very low metabolic rate. In contrast to findings in cold-adapted giant squid, the minimized thermal sensitivity of oxygen transport in Antarctic octopods will Received 3 July 1999: accepted 10 October 2000. * To whom correspondence should be addressed. hpoertner@awi-bremerhaven.de E-mail: reduce metabolic scope and thereby contribute to their stenothermality. Introduction Cephalopods are found throughout the seas of the world. from warm tropical waters to polar oceans (Roper et al., 1984). Representatives of this group, especially squids, usu- ally display the highest metabolic rates among marine in- vertebrates, even higher than those of fishes of similar size and mode of life (Webber and O'Dor, 1985; O'Dor and Webber, 1986). Oxygen delivery via the blood is maxi- mized to cover metabolic requirements (Portner, 1994). However, the capacity of their blood pigment, hemocyanin, for carrying oxygen is constrained by the low concentration of an extracellular pigment. This limitation is due to the unfavorable increase in colloidal osmotic pressure and blood viscosity at high pigment concentrations (Mangum, 1983, 1990). Although cephalopods, in accordance with their high rate of oxygen consumption, display the highest hemocyanin concentrations in the animal kingdom, the level of bound oxygen in squid (up to 3 mmol 1 ; Brix et al., 1989) remains below the 4-5 mmol I"1 of active fishes (Urich, 1990). Therefore, the hearts of squids pump large volumes of blood (Wells et al., 1988; Shadwick et al., 1990) and the tissues extract most of the oxygen (Portner. 1994). Compared to that of squids, the oxygen-binding capacity of octopod blood is somewhat reduced, ranging between 0.6 and 1.6 mmol P1. depending on hemocyanin levels (Senozan et al.. 1988; Brix et al., 1989). In most cephalopods cooperativity and temperature- and pH-dependent changes in affinity are the only means of modulating hemocyanin function and adjusting oxygen transport (e.g., Brix et al., 1989; Mangum, 1990; Portner, 1990). Low-molecular-weight organic substances that con- tribute to blood pigment function in vertebrates or crusta- 67 68 S. ZIELINSKI ET AL ceans are not found in this group. In consequence, ex- tremely large Bohr shifts (A log P50/ApH < -1: Bridges. 1994) and very high levels of pH-dependent cooperativity are common (Miller. 1985: Portner. 1990). Binding of CO2 together with Oo in arterial blood has been suggested to support pigment function on the venous side in sepioid species (Brix et al.. 1981). where both O: and CO; are released, and this CO-, helps to exploit the large Bohr effect. In squid, supplementary oxygen uptake via the skin sup- ports the excessive oxygen demand and provides the excess CO, required for the Bohr effect to function (Portner. 1994). In some cephalopods. an increase in ambient temperature has a large effect on oxygen transport by hemocyanin; this effect is reflected by an increase in cooperativity and a fall in oxygen affinity (Brix et al., 1989. 1994: Mangum. 1990). If a rise in metabolic rate with temperature is supported by an adequate rise in P?n. the species should be able to live at a broader range of temperatures than a species in which P50 remains constant or in which the change in P50 is too large. For example, the high thermal sensitivity of the oxygen affinity of hemocyanin in the giant squid Architenthis niona- chits suggests that arterial saturation becomes impossible at high temperatures (Brix. 1983). This question has gained general importance since comparative studies in Antarctic and temperate fish and invertebrates (including cephalo- pods; Portner and Zielinski. 1998) revealed that the limits of thermal tolerance are characterized by oxygen limitation, owing to the inability of circulation or ventilation to provide sufficient oxygen at extreme temperatures (for review, see Portner et al.. 2000). Comparison of hemocyanin oxygen binding in cephalopods of different metabolic rates and from various latitudes should show how hemocyanin oxy- gen transport has adapted to different temperature regimes at various levels of metabolic activity and how blood pig- ment function contributes to the oxygen limitation of ther- mal tolerance. These questions are especially interesting for an under- standing of physiological adaptations to life in Antarctica. Here the marine environment is characterized by very stable water temperatures that are close to freezing (Clarke, 1988). Under these conditions more oxygen is physically dis- solved, thereby facilitating oxygen uptake and supply to tissues. At the same time metabolic rate is reduced at lower temperatures, with the consequence that in some species blood pigments may be less important (for hemocyanin. see Mauro and Mangum. 1982b: Burnett et al.. 1988). Some Antarctic fishes, the icefishes (Channichtyidae) have even lost their respiratory pigments (Ruud. 1954). The question arises as to whether the importance of blood pigment func- tion is also reduced in Antarctic cephalopods. Live specimens of the octopod Megaleledone senoi be- came available during a recent expedition to Antarctica with the RV Polarstern. This species is found in the indo-atlantic sector of the Antarctic Ocean (Taki. 1961: Kubodera and Okutani, 1986. 1994). In our study we investigated oxygen binding to the hemocyanin of this stenothermal Antarctic octopod by using a technique that allows continuous and simultaneous recordings of blood pH and oxygenation and the construction of diagrams depicting changes in oxygen saturation with pH (pH/saturation diagrams: Portner. 1990). Such an approach is most suitable for cephalopod blood owing to the extremely large pH dependence of oxygen binding (see above). It avoids the use of artificial buffers that may lead to a change in oxygen-binding properties (Portner, 1990: Brix et al.. 1994). At the same time, the amount of blood required is reduced such that more sophis- ticated data can be collected from the very few animals accessible in remote environments like the Antarctic. The oxygen-binding properties of M. senoi hemocyanin were compared with those from other eurythermal and stenother- mal cephalopods. For eurythermal octopods some literature data were available. Temperature effects on oxygen binding were studied experimentally in the cuttlefish Sepia officina- lis to complement the data set available in the literature (Lykkeboe et al., 1980; Johansen et al., 1982a). To some extent, cuttlefish display a mode of life similar to that of octopods. Like octopods, they live close to the bottom of the sea (von Boletzky, 1983), but they have a larger scope for activity and metabolism, which might influence the thermal adaptation of hemocyanin function. Materials and Methods Animals Antarctic octopods (Megaleledone senoi, up to 9 kg body weight) were caught in November 1996 north of Elephant Island, Antarctica, during expedition ANT XI V/2 of the RV Polarstern. The animals were collected from bottom trawls. Samples were taken immediately after capture. Cuttlefish (Sepia officinalis. 470 to 960 g body weight) were obtained from the Marine Biomedical Institute of the University of Texas, Galveston. Texas, where this species has been bred and grown for several consecutive genera- tions. They were kept at a salinity of 359£r at temperatures of 20 to 22 °C. Sampling procedure Animals were anesthetized by transferring them into sea- water containing 2% ethanol (v/v). The animals were then removed from the seawater and the mantle was opened by a ventral incision. Blood was collected from the vena cava, the systemic heart, and the gill hearts. Blood samples from all animals were pooled, frozen, and stored for up to one year at close to —20 °C until utilized for in vitro studies of oxygen binding. ANTARCTIC CEPHALOPOD HEMOCYANIN FUNCTION 69 X Gas mixture layer of 0.45 mm Stirrer I Blood Figure 1. Cuvette used for the measurement of oxygen-binding prop- erties. Dashed areas show the compartments of the cuvette filled with blood. Absorbance was measured through a thin layer, 0.45-mm thick, in the center part of the cuvette. Analysis of oxygen binding Oxygen-binding characteristics of cephalopod hemocya- nin were studied using a specially constructed cuvette, built by Hellma GmbH & Co. (Mulheim, Germany; Fig. 1). The cuvette consisted of an upper and a lower compartment connected by two shafts ( 1 .5 and 2 mm in diameter) in the left and right periphery of the cuvette, as well as a central compartment between the shafts, where blood formed a thin layer of only 0.45 mm. Stirring bars operating in the upper and lower compartments ensured continuous exchange of blood between all compartments and thus uniform mixture of the blood. Oxygen saturation was monitored continu- ously by using a diode array spectrophotometer with fiber optics (X-dap. IKS Optoelektronik MeGgeriite GmbH, Waldbronn, Germany) to measure absorbance at 345 nm through the thin layer. Blood samples were equilibrated by introducing humidified gas mixtures through a hole in the lid of the cuvette. Gas mixtures of variable Po, (between 1 .0 and 20.0 kPa) were prepared from pure O,, CO,, and N, by gas-mixing pumps (type 2M303/a-F, Wosthoff, Bochum. Germany); complete deoxygenation occurred under pure N,. Blood pH was varied by changing Pco, (between 0.09 and 1.01 kPa) or by replacing small volumes (<10 ^\ per 2 ml of blood) of supernatant plasma after ultracentrifugation (1 h at 120.000 X g; Beckman Airfuge, Beckman Instru- ments, Inc., Fullerton, CA) with fixed acid (1 mol 1~' HC1) or base (2 mol 1~' NaOH: Morris el ai, 1985; Portner, 1990). Changes in blood pH during oxygenation and deox- ygenation of hemocyanin were measured continuously by using a needle pH electrode (long micro needle electrode #811, Diamond General Corp., Ann Arbor. MI) that was introduced into one of the shafts via a second hole in the lid. Total CO, was analyzed at 0 °C in 50-jid blood samples of M. senoi; the gas chromatography method of Lenfant and Aucutt (1966) modified after Boutilier el al. (1985) was used. Measurements of oxygen-binding properties were car- ried out at 0, 5. and 10 °C for samples of M. senoi and at 0, 10. and 20 °C for samples of S. officinalis; 10 °C was chosen since this temperature can be reached in the northern part of Sepia's distribution range (Isemer and Hasse, 1985). Graphical analysis and calculations Hemocyanin concentrations were measured photometri- cally and calculated using the extinction coefficients of Nickerson and van Holde (1971). Oxygen capacity was estimated using the molecular weights for octopods and Sepia as compiled by Miller (1994) and the assumption that there are 70 O,-binding sites per hemocyanin molecule in octopods and 80 in Sepia. For the evaluation of hemocyanin oxygen saturation, constant absorbance levels in the range of the highest values of Po, and pH were set to 100% saturation. Changes in hemocyanin oxygenation and pH were plotted in a pH/saturation diagram according to Port- ner (1990). The resulting oxygen-binding curves represent isobars delineating the change in oxygenation with pH at constant Po,. The points of intersection of the oxygen isobars with the line of half saturation quantify P50, because it depends on pH. These P50 and pH values were used to evaluate the Bohr coefficient, A log Pso/ApH by linear regression analysis. For comparison, and owing to the pres- ence of a large pH-insensitive oxygen reserve at low tem- peratures, the coefficient A log P80/ApH was evaluated following the same procedure. The Haldane coefficient (AHCO3~/AHcyO,) was evaluated from the vertical dis- tance between buffer lines in a pH/bicarbonate diagram (as used by Brix el al.. 1981 ). To assess whether oxygen-linked CO, binding to the hemocyanin occurs (Lykkeboe el al., 1980). "measured" and calculated apparent bicarbonate lev- els were compared. Apparent "bicarbonate" (the sum of HCO3 and CO,2" levels) was calculated from measured CO, concentrations (Ceo,) using the applied Pco, and the measured pH according to the formula [HCOr] = Ceo, - aPco; ( 1 ) where a is the solubility of CO2. For comparison, bicarbon- ate levels were also calculated according to equation ( 1 ) with Ceo, values derived from the Henderson-Hasselbalch equation: Cco: = Pco, • I a • 1 OpH~pK"' + a) ( 2 ) a and pK'" were calculated according to Heisler (1986). Along each isobar in the pH/saturation diagram, values of saturation S depend on pH values and the Po, of the isobar. The pH/saturation diagram allows comparison of S and log Po, with P5(l at the same pH (=pH?0). This leads to an 70 S. ZIELINSKI ET AL analysis of cooperativity at a specific pH. If this is done in the range of saturation S between 0.4 and 0.6, the analysis leads to an estimate of Hill coefficients (H5()) according to log(S/l - S) = H5(,(log Po: - log P50) (3) where Po-, is the Po2 of the isobar, S results from Pen at a specific pH (pH50), and P50 is the Po, for S = 0.5 at the same pH (Former, 1990). Results The concentration of hemocyanin in native blood (he- molymph) was 93 g 1~' for Megaleledone senoi and 142 g 1 ' for Sepia officinalis. This is equivalent to a maximum level of hemocyanin-bound oxygen of 1.86 mmol O: 1~' in the octopod and of 2.84 mmol 1~' in Sepia. For M. senoi hemocyanin, the highest pH sensi- tivity of oxygen binding was found at 10 °C, as indicated by maximum slopes AS/ApH (Fig. 2). Lower tempera- tures resulted in a somewhat decreased pH sensitivity of oxygen affinity, with a maximum of AS/ApH = 13% per pH unit at 10 °C, compared to a maximum of 10% per pH unit at 0 °C. Saturation at 0 °C did not fall below 43% even at low pH (6.4 and 6.6) and low Po2 ( 1 kPa). At 10 °C, saturation dropped to a minimum of 32% at the same 100 r — 50% S- 6.2 6.4 6.6 6.8 7.0 7.2 7.4 7.6 7.8 8.0 8.2 PH Figure 2. Continuous analysis of the relationships between oxygen binding, pH. and Po, in whole blood of the Antarctic octopod Megaleledone senoi at temperatures between 0 and 10 °C (S = saturation; points were chosen at regular intervals from the continuous recordings). ANTARCTIC CEPHALOPOD HEMOCYANIN FUNCTION Table 1 Oxygen affinity (Pw-), Bohr, Haldane, and Hi/I coefficients evaluated for the hemocyanins of the Antarctic octopoil Megaleledone senoi and of the cuttlefish Sepia officinalis at different temperatures 71 Temperature (°C) />,„ (kPa) Bohr coefficient (A log/VApH) (A log /V,/ApH) Haldane coefficient (AHCOr/AHcyO,) Hill coefficient ("50) Megaleledone senoi 0 0.98 (pH 7.2) -0.9 -1.51 0.66 (pH6.2) 1.4 (pH 7.43) 0.50 (pH6.8) 0.39 (pH7.2) 5 ND -1.46 ND 1.5 (PH7.24) 1.0 (PH7.31) 10 1.10(pH7.2) -2.33 -2.13 ND 1.2 ± 0.1 (pH 6.83-7.31; Sepia officinalis 10 5.3 (pH 7.4) -0.99 -1.44 4.6 (pH 7.29) 20 6.5 (pH 7.4) -1.33 -1.94 5.9 (pH 7.48) Pfa, Haldane. and Hill coefficients are valid for the pH values given in brackets. ND, not determined; HcyCK. concentration of oxygenated hemocyanin. Po2. Intermediate values of pH sensitivity and maximum unloading were found at 5 °C. A large Bohr coefficient of —2.33 was found at 10 °C, similar to the coefficient A log P80/A pH (Table 1 ). The experimental evaluation of the Bohr coefficient was not possible at lower temperatures due to the fact that pH- dependent saturation did not drop below 50% at most partial pressures of oxygen. An extrapolation of binding data to very low partial pressures of oxygen revealed a Bohr coef- ficient of approximately —0.9 at 0 °C, below the level of A log /^,,/ApH. Furthermore, oxygen affinity (P50) at pH 7.2 changed only at AP50/°C < 0.01 kPa, from 0.98 kPa at 0 °C to l.lOkPaat 10 °C (Table 1). For S. officinalis hemocyanin. pH sensitivity was high at 20 °C, reaching a maximum AS/ApH of 41% per pH unit (Fig. 3). Especially in the pH range between 7.4 and 7.8, very small pH changes were sufficient to cause maximal unloading of oxygen, down to 19% saturation. The pH sensitivity at 20 °C was higher than found for M. senoi at all temperatures. As in M. senoi. lower temperatures decreased the pH sensitivity of oxygen binding with a decreased Bohr factor and level of A log P80/ApH (Table 1 ) and an increase in the pH-insensitive reserve at the same Po2. At 0 °C, AS/ApH reached a maximal value of only 7% per pH unit. Oxygen saturation remained above 50% at all investigated partial pressures of oxygen and values of pH. At pH 7.4. oxygen affinity fell from />5() = 5.3 kPa at 10 °C to PMI = 6.5 kPa at 20 °C (AP50/°C = 0.12 kPa. Table 1 ). The change in cooperativity with pH and temperature for S. officinalis is shown in Figure 4. At 20 °C, the largest Hill coefficient (»50) of 5.9 was found at a pH (7.48) where pH sensitivity (AS/ApH) was also high. A decrease in temper- ature to 10 °C resulted in a decrease of the maximal Hill coefficient to ;;_„, == 4.6 (pH 7.29). The maximum was shifted to lower pH. In contrast to cuttlefish, M. sensoi had much lower Hill coefficients (Table 1 ). At 0 °C. «50 was 1.4 (pH 7.43). and it varied between 1.0 (pH 6.83) and 1.4 (pH 7.31 ) at 10 °C. No clear maximum could be found. Analysis of total CO2 in M. senoi blood during variations of Po2 and Fco-, yields the buffer lines depicted in the pH/bicarbonate diagram (Fig. 5). The position of the buffer line shifts between oxygenated and deoxygenated blood according to the quantity of H+ bound by the pigment. The vertical distance between the buffer lines yields the Haldane coefficient (AHCO,~/AHcyO2). For M. senoi hemocyanin at 0 °C, the Haldane coefficient rose with falling pH (Table 1 ). The calculated apparent bicarbonate levels for oxygen- ated and deoxygenated blood diverge only slightly from the measured values, suggesting that O^-linked CCK binding does not exist (Fig. 5). The non-bicarbonate buffer value (|3NB) of 4.25 mmol 1~' pH units ~' of M. senoi blood at 0 °C is in the same range as in the squids Illex illecebrosus and Loligo pealei (5.0 and 5.8 mmol spectively; Portner, 1990). 1 pH units , re- Discussion At the low temperatures of Antarctic.! icefishes rely ex- clusively on the transport of oxygen that is physically dis- solved in the blood (see Introduction). The presence of hemocyanin-bound oxygen in Megaleledone senoi blood at levels similar to those seen in squids and temperate octo- pods (cf. Brix et a!.. 1989) suggests that oxygen transport via hemocyanin is as important in this Antarctic species as in temperate and warm-water cephalopods. In contrast to Antarctic fishes, the unchanged requirement for blood oxy- gen transport in Antarctic cephalopods may be related to the 72 s .111 IVSKI i I AL -• "•• 00 A^-^ o :o.o v 40 - V l.V3kPa 5/5 D S->kPa 20 v 'JiJ officinalis O 4.3 kl\i n i i i i i i i i i > i Cv: 0.4 cvt. cvS 7.0 7.2 7.4 7.6 7.S S.O S.I 8.4 pH Kigurv 3. IVpiciion of oxygen-binding properties of ,Siii <>{fh'iiwlis hemocyanin at temperatures between 0 and 20 ~C in a pH saturation digram. Each lino indicates an oxygen isobar and shows the changes in oxygen binding depending on ,-il s vaturation: p*>ims were chosen ai regular intervals faun the continuous rcvotu low level of capitalization ofcephalopod niuseulanuv eoui- \ ed to tish nuiseU-> iHone <•; .;/.. 1^811. In eoph.ilo(vxis. pH and tctn|vrature arc most inipottani taotoi-s in the regulation of henuvvunin o\\v:en transport ,Bn\ <•: .;.'.. l^S^; Brukes, 1^4; Portlier. l^>4>. The teni- [X-ratitrc dependence of /\ , \ancs greatly between s|\\ es -•vaniple. the o\\cen at'tinitx of the heniocyaiiin of iiiant squid (Archittuthis monac'-:. - den eases at A/%,, A/' 1 ,S° kPa per derive Celsius »pH ~.4». while a \alue of onh v\:0 kPa C i.pH ~.4* was found for the octopixl Octopus vulgaris (calculated after Brix . . l°Sk)l. A lower value of A,'\ > AT == 0.10 kPa >.' »pll 7 4) was found for the octopod l-'lction? cirrliosu Calculated after Bridges. 1OU4>; tin- \\a- similar to the value of 0.12 kPa/°C (pH ".41 calculated for >Y;>;a iijfficinalis heiuocyanin in the prest study. In eur\ thermal cephalopods like Sepia ofticintilis. •:is vulgaris. or Eledon? cirrhosa and in some squids. a nuxlerate rise in P50 with temperature occurs (cf. Bri\ ct al.. \W4\ In this wa\ capillary Po2 is maintained ("buf- fered" i at progressively higher levels, which are required for elevated diffusive oxygen rlu\ to mitochondria during in- . e-.-ed rates of oxygen consumption. Such changes in P50 with temperature allow S. officinalis to be distributed over a \\ idc range, from the Mediterranean to the North Sea (von I" I If < I I-HM',I'<;I, III -.\i,< VANIN M '.' TON 73 f. 5 4 = '' 2 1 0 6.4 ' 6 ' 7.0 7.2 7.4 7 /, 7.8 pH Figure 4. Mill coefficient'. '•, hemocyanin de- pending li%o pealei bkxxl in vitro (Howell and Gilbert. 1976;, the in vivo pH range for M. senoi is between pH 7.7 and 7.9 at 0 ''C. With the highest pH sensitivity in this pH range, oxygen unloading would occur at very low oxygen tensions (<] kPa;, supporting only very low metabolic rates (Fig. 2;. A P5I> below 1 kPa (pH 7.6; 0 r'C) reflects one of the highest oxygen affinities reported so far for cephalopods. This value is close to the /' of 0.47 to 0.55 kPa evaluated for the cold-water vam- pire squid Vampyroteuthis infernalis (5 °C; Seibel el al., 1 999 1. These findings suggest that M. senoi displays a low metabolic rate similar to that of the Antarctic octopod Pareledone charcoli (0.3 /nmol g ' h ' at 0 °C and about nody weight; H. O. Portner. T. Hirse, V. Wegewit/, unpubl data. I S times lower than similar si/ed .V. offu imilis at 17 f. 1 .4 //mol •.' ' h '. Johan-.en rial.. 19H2b;oreven lower and close to the O.I //mol g ' h ' measured at ' in the deep-sea squid Vn/n/> . mii-mlu- //// 'Sei1 al 1997). The Bohr coefficient evaluated for both investigated cephalopod species dropped when temperature decreased. This result is similar to findings in the crustaceans Cancer n IIT and Cancer anthonyi (Burnett el al., 1988; and in the octopod i.li-il<,n<- • mhusa 'Bridges. 1994;. In .V. o/fit t nulls, A lo;.> /'.,/ApH decreased moderately, from —1.33 at 20 Cto 099 at 10 ( ' 'I able \>.\nM. Knot th-.- Bohr coefficient fell drastically, from an extremely high value of — 2.33 at 10 '•(„' to a much smaller value evaluated by extrapolation to be 0.9 at 0 ( r'fable );. The Bohr (actor in the vampire squid was found to be even lov.c-r < 0.22: Seibel et al., 1999;. These results indicate that the Bohr effect becomes less important at low temperature and low metabolic rate. A mechanism of oxygen-linked CO2 binding has been proposed for Sepia hemocyanin. which transports both O, and CO to the tissues. The CO:, produced in metabolism and th CO leased during deoxygenation would elicit a drop in pH. as required for the large Bohr effect (< - 1.0; to function normally (Lykkeboe el al.. 1980: Brix. <> »/ 1981;. No oxygen-linked CO, transport was found in M. senoi (Fig. 5;. At 0 ''C. the estimated Bohr cm-fficient of A Joy I' ,,/ApH ' 09 v.ould reflect normal function of the Bohr effect, whereas the extremely high Bohr coefficient at 10 ( v.ould be counterproductive for oxygen transport. 7 6 "£" 5 | 4 *• i T 3 Z 2 1 P.NB = 4-25 mmol • H pH-units-' Deoxygenated 6.0 6.2 6.4 6.6 ' - 7.0 7.2 7.4 7.6 7.8 8.0 PH Figure 5. pH/bitarbonate diagram for whole blood of Me^aleltdtmc •>em>i. Experimental buffer lines of oxygenated and deoxygenated blood match those derived from calculated apparent bicarbonate levels ^broken lines: see text;. 74 S. ZIELINSKI ET AL a finding consistent with the stenothermality of Antarctic animals. A reduced pH sensitivity of hemocyanin oxygen binding in M. senoi and other octopods compared to S. officinalis and squids is reflected in the magnitude of the pH-indepen- dent venous reserve, which rises as temperature falls (Figs. 2. 3). This reserve represents the amount of oxygen that remains bound to the respiratory pigment at constant Po2, even when pH falls to very low values. Comparison of this venous reserve for several cephalopod species at normal environmental temperatures and low Po2 shows that it is below 5% (at 1.7 kPa) for the squid II lex illecebrosus (15 °C; Portner, 1990). 19% (at 1.7 kPa) for S. officinalis (20 °C; this study), and 43% (at 1 kPa) for the Antarctic octopod M. senoi (0 °C; this study). A value of below 10% results for the hemocyanin of the octopod Octopus dofleini (at 1.7 kPa and 20 °C; Portner. 1990: calculated after Miller and Man- gum, 1988); however, the /;; vivo value may be higher for this species because it lives at lower temperatures. The high pH sensitivity of squid hemocyanins maximizes the release of oxygen in the tissues and supports their high metabolic rate (Portner, 1990, 1994). Sepioids and, even more so. octopods display a less active life style with lower meta- bolic rates (for example: Houlihan el til.. 1982; Webber and O'Dor. 1985, 1986; Finke et al.. 1996; Seibel el al. 1997). A low-activity mode of life may eliminate the necessity to maximize pH-dependent oxygen transport to the extent seen in squids. With falling pH. the pH-independent venous reserve in- creased and was reached at higher Po2 (88% at pH 6.8, 20 kPa O2, and 0 °C in M. senoi. or 40% at pH 6.8, 20 kPa O2. and 20 °C in S. officinalis). At normoxic Po2 (20 kPa CM and low pH, this resembles a Root effect (Bridges, 1994) but at the same time, further pH sensitivity (the Bohr effect) is eliminated and deoxygenation depends exclusively on Po2 (Figs. 2 and 3). The cooperativity of respiratory pigments is characterized by the Hill coefficient («5()). In 5. officimilis at 20 °C (Fig. 4) and in the squids 1 1 lex illecehrosus and Lo/igo vnlgaris. the highest cooperativity correlates with the highest pH sensitivity of oxygen binding ( AS/ApH) in the range of in vivo pH (Portner, 1990). Here maximal deoxygenation oc- curs at minimal pH change (Portner. 1990. 1994). A de- crease in temperature caused the maximal Hill coefficient of S. officinalis hemocyanin to drop from /;50 = 5.9 at 20 °C to ;i50 == 4.6 at 10 °C. At the same time, maximum cooperativity was shifted to lower pH values, when in vivo pH should rise according to a-stat predictions (Reeves, 1972). A similar temperature dependence of the Hill coef- ficient was found for several crustaceans (Mauro and Man- gum, 1982a.b). As with the Bohr effect, the progressive mismatch between the pH range of maximum cooperativity and the actual blood pH suggests that cooperativity becomes less important in oxygen transport at lower temperatures. Accordingly, a low cooperativity of n5(, = 1 .4 was found for M. senoi. at 0 °C (Table 1) and of /i50 = 2.2 for the vampire squid (Seibel et til.. 1999). Surprisingly, cooperat- ivity did not increase with temperature in M. senoi hemo- cyanin. The question arises as to why thermal sensitivity is so low in M. senoi hemocyanin but so high in the blood pigment of cold-adapted giant squid (cf. Brix, 1983). A high value of AP?0 °C~' reflects a high heat of oxygenation (cf. Brix et al.. 1994) or Arrhenius activation energy. Giant squid prob- ably display higher metabolic rates and thus must maintain P50 levels higher than those of M. senoi. A high heat of oxygenation may be required for setting P50 values high at low temperature as in giant squid. In that respect the low thermal sensitivity of hemocyanin in the Antarctic octopod is again in accordance with the low metabolic rate of this group. In summary, the pH sensitivity of oxygen binding in cephalopod hemocyanins is adjusted to metabolic rate. The pH-insensitive oxygen reserve in hemocyanin was largest in M. senoi and intermediate in S. officinalis, if compared to squids (this study and Portner, 1990). Furthermore, the Bohr effect is reduced and the pH-insensitive oxygen reserve rises during cooling, suggesting that pH sensitivity falls in the cold. The temperature dependence of the Bohr factor is less pronounced in the eurythermal S. officinalis. which would, together with an appropriate change in P50, ensure a supply of oxygen at changing temperatures. In M. senoi. a high oxygen affinity of hemocyanin. a moderately high Bohr coefficient, and a low cooperativity at 0 °C cause blood Po2 to be maintained ("buffered") at low values matching a low rate of oxygen consumption. In this species the low thermal sensitivity of oxygen affinity prevents an upward shift of the buffered Po2 at higher temperatures, suggesting that oxygen transfer to tissues may become limiting when oxygen demand rises. This observation is in contrast to the findings in giant squid, where arterial oxygen uptake is hampered by an excessive drop in oxygen affinity, thereby limiting heat tolerance (Brix, 1983). Accordingly, hemocyanin function probably contributes to an oxygen limitation of heat tolerance that sets in early and character- izes thermal tolerance in Antarctic octopods (Portner and Zielinski. 1998) and probably also in giant squid. Acknowledgments The authors thank Iris Hardewig and Boris Klein for sampling hemolymph from Megaleledone senoi during the expedition with RV PoUirstern. The technical and logistical help by the staff of the Marine Biomedical Institute is gratefully acknowledged. Supported by grants of the Deut- sche Forschungsgemeinschaft to H.O. Portner (Po 278). ANTARCTIC CEPHALOPOD HEMOCYANIN FUNCTION 75 Literature Cited Bone, Q., A. Pulsford, and A. D. Chubb. 1981. Squid mantle muscle. J. Mar. Bio/. Assoc. UK 61: 327-342. Boutilier, R. G., G. K. Iwama, T. A. Heming, and D. J. Randall. 1985. The apparent pK of carbonic acid in rainbow trout blood plasma between 5 and 15 "C. Respir. Phvsiol. 61: 237-254. Bridges, C. R. 1994. Bohr and Root effects in cephalopod haemocya- nins — paradox or pressure in Sepia officinalis? Pp. 121-130 in Physi- ology of Cephalopod Molluscs — Lifestyle and Performance Adapta- tions. H. O. Portner. R. K. O'Dor. and D. MacMillan. eds. Gordon and Breach, Basel. Brix, O. 1983. Giant squids may die when exposed to warm water currents. Nature 303: 422-423. Brix, O., G. Lykkeboe, and K. Johansen. 1981. The significance of the linkage between the Bohr and Haldane effects in cephalopod bloods. Respir. Physiol. 44: 177-186. Brix, O., A. Bardgard, A. Cau, A. Colosimo, S. G. Condo, and B. Giardina. 1989. Oxygen-binding properties of cephalopod blood with special reference to environmental temperatures and ecological distribution. 7. Exp. Zoo/. 252: 34-42. Brix, O., A. Colosimo, and B. Giardina. 1994. Temperature depen- dence of oxygen binding to cephalopod haemocyanins: ecological implications. Pp. 149-162 in Physiology of Cephalopod Molluscs- Lifestyle and Performance Adaptations. H. O. Portner, R. K. O'Dor. and D. MacMillan. eds. Gordon and Breach. Basel. Burnett, L. E., D. A. Scholnick, and C. P. Mangum. 1988. Tempera- ture sensitivity of molluscan and arthropod hemocyanins. Dial. Bull. 174: 153-162. Clarke, A. 1988. Seasonality in the Antarctic marine environment. Camp. Biochem. Physiol. 90B: 461-473. Finke, E., H. O. Portner, P. G. Lee, and D. M. Webber. 1996. Squid (Lolliguncula brevis) life in shallow waters: oxygen limitation of metabolism and swimming performance. J. Exp. Bio/. 199: 91 1-921. Heisler, N. 1986. Buffering and transmembrane ion transfer processes. Pp. 3 — VI in Acid-Base Regulation in Animals, N. Heisler, ed. Elsevier, Amsterdam. Houlihan, D. G., A. J. Innes, M. J. Wells, and J. Wells. 1982. Oxygen consumption and blood gases of Octopus vulgaris in hypoxic condi- tions. J. Comp. Physiol. 148: 35-40. Howell, B. J., and D. L. Gilbert. 1976. pH-temperature dependence of the hemolymph of the squid. Loligo pealei. Comp. Biochem. Phvsiol. 55A: 287-289. Isemer, H. J., and L. Hasse. 1985. The Bunker Climate Atlas of the North Atlantic Ocean. Vol 1: Obsen'ations. Springer Verlag, Berlin. 218pp. Johansen, K., O. Brix, and G. Lykkeboe. 1982a. Blood gas transport in the cephalopod. Sepia officinalis. J. Exp. Biol. 99: 331-338. Johansen, K., O. Brix, S. Kornerup, and G. Lykkeboe. 1982b. Factors affecting O2 uptake in the cuttlefish. Sepia officinalis. J. Mar. Biol. Assoc. UK 62: 1X7-191. Kubodera, T., and T. Okutani. 1986. New and rare cephalopods from the Antarctic waters. Mem. Natl. Insl. Polar Res.. Spec. Issue 44: 129-143. Kubodera, T., and T. Okutani. 1994. Eledonine octopods from the Southern Ocean: systematic^ and distribution. In Southern Ocean Cephalopods: Life Cycles and Population*, P. G. Rodhouse. U. Piat- kowski and C. C. Lu. eds. Antarct. Sci. 6: 205-214. Lenfant, C., and C. Aucutt. 1966. Measurement of blood gases by gas chromatography. Respir. Physiol. 1: 398-407. Lykkeboe, G., O. Brix, and K. Johansen. 1980. Oxygen-linked CO, binding independent of pH in cephalopod blood. Nature 287: 330-331. Mangum, C. P. 1983. Oxygen transport in the blood. Pp. 373-429 in The Biology of Crustacea. Vol. 3, L. H. Mantel, ed. Academic Press, New York. Mangum, C. P. 1990. Gas transport in the blood. Pp. 443-468 in Squid as Experimental Animals, D. L. Gilbert. E. J. Adelman. Jr.. and J. M. Arnold, eds. Plenum. New York. Mauro, N. A., and C. P. Mangum. 1982a. The role of the blood in the temperature dependence of oxidative metabolism in decapod crusta- ceans. I. Intraspecific responses to seasonal differences in temperature. J. Exp. Zool. 219: 179-188. Mauro, N. A., and C. P. Mangum. 1982b. The role of the blood in the temperature dependence of oxidative metabolism in decapod crusta- ceans. II. Interspecific adaptations to latitudinal changes. J. Exp. Zool. 219: 189-195. Miller. K. I. 1985. Oxygen equilibria of Octopus dofleini hemocyanin. Biochemistry 24: 4582-4586. Miller, K. I. 1994. Cephalopod hemocyanins. A review of structure and function. Pp. 101-120 in Physiology of Cephalopod Molluscs— Life- style and Performance Adaptations. H. O. Portner. R. K. O'Dor, and D. MacMillan, eds. Gordon and Breach, Basel. Miller, K., and C. P. Mangum. 1988. An investigation of the nature of Bohr, Root, and Haldane effects in Octopus dofleini hemocyanin. J. Comp. Physiol. 158B: 547-552. Morris, S., A. C. Taylor, C. R. Bridges, and M. K. Grieshaber. 1985. Respiratory properties of the haemolymph of the intertidal prawn Palaemon elegans (Rathke). /. Exp. Zool. 233: 175-186. Nickerson, K. W., and K. E. van Holde. 1971. A comparison of molluscan and arthropod hemocyanin. I. Circular dichroism and ab- sorption spectra. Comp. Biochem. Physiol. 39B: 855-872. O'Dor, R. K., and D. M. Webber. 1986. The constraints on cephalo- pods: why squid aren't fish. Can. J. Zool. 64: 1591-1605. Portner, H. O. 1990. An analysis of the effects of pH on oxygen binding by squid (I/lex illecebrosus. Loligo pealei) haemocyanin. J. Exp. Biol. 150: 407-424. Portner, H. O. 1994. Coordination of metabolism, acid-base regulation and haemocyanin function in cephalopods. Pp. 131-148 in Physiology of Cephalopod Molluscs — Lifestyle and Performance Adaptations, H. O. Portner, R. K. O'Dor, and D. MacMillan. eds. Gordon and Breach. Basel. Portner, H. O., and S. Zielinski. 1998. Environmental constraints and the physiology of performance in squids. In Cephalopod Biodiversity. Ecology and Evolution. A. I. L. Payne, M. R. Lipinski, M. R. Clarke, and M. A. C. Roeleveld, eds. S. Afr. J. Mar. Sci. 20: 207-221. Portner, H. O., P. L. M. van Dijk, I. Hardewig, and A. Sommer. 2000. Levels of metabolic cold adaptation: tradeoffs in eurythermal and stenothermal ectotherms. Pp. 109-122 in Antarctic Ecosystems: Mod- els for Wider Ecological Understanding, W. Davison and C. Howard Williams, eds. Caxton Press, Christchurch, New Zealand. Reeves, R. B. 1972. An imidazole alphastat hypothesis for vertebrate acid-base-regulation: tissue carbon dioxide content and body tempera- ture in bullfrogs. Respir. Physio/. 81: 255-274. Roper, C. F. E., M. J. Sweeney, and C. E. Nauen. 1984. FAO Species Catalogue. Vol. 3: Cephalopods of the World. An Annotated and Illustrated Catalogue of Species of Interest to Fisht-ries. FAO Fisheries Synopses 125. FAO/UNDP. Rome. 277 pp. Ruud, J. T. 1954. Vertebrates without erythnx > tes and blood pigment. Nature 173: 848-850. Seibel, B. A., E. V. Thuesen, J. J. Children, and L. A. Gorodezky. 1997. Decline in pelagic cephalop<"J " tiabolism with habitat depth reflects differences in locomotory elliciency. Biol. Bull. 192: 262-278. Seibel, B. A., F. Chausson, F. H. L;i!lier, F. Zal, and J. J. Childress. 1999. Vampire blood: respiratory physiology of the vampire squid (Cephalopoda: Vampyromorpha) in relation to the oxygen minimum layer. Exp. Biol. Online 4: 1 . Senozan, N. M., A. Avinc, and Z. Unver. 1988. Hemocyanin levels in Octopus vulgaris and the cuttlefish Sepia officinalis from the Aegean sea. Comp. Biochem. Physiol. 91A: 581-585. 76 S. ZIELINSKI ET AL Mi. iil« H I,, R. E., R. K. O'Dor, and J. M. Gosline. 1990. Respiratory and performance of short-finned squid (lllex illecebrosus). Northwest All. cardiac function during exercise in squid. Can. J. Zoo/. 68: 192-198. Fish. Organ. Sci. Count: Stud. 9: 133-138. Urich. K. 1990. Vergleichende Biochemie der Tiere. Gustav Fischer. Webber, D. M., and R. K. O'Dor. 1986. Monitoring the metabolic rate Stuttgart. 708 pp. and activity of free-swimming squid with telemetered jet pressure. J. Taki, I. 1961. On two new eledonid octopods from the Antarctic sea. ./. E.\p. Biol. 126: 203-224. Fac. Fish. Anim. Hush. Hiroshima Univ. 3: 297-316. Wells, M. J.. R. T. Hanlon, P. G. Lee, and F. P. Dimarco. 1988. von Boletzky, S. 1983. Sepia officinalis. Pp. 31-52 in Cephalopod Life Respiratory and cardiac performance in Lolliguncula bre\'is (Cepha- C\clt'x. Vol. I. P. R. Boyle, ed. Academic Press, London. lopoda. Myopsida): the effects of activity, temperature and hypoxia. /. Webber, D. M., and R. K. O'Dor. 1985. Respiration and swimming Exp. Biol. 138: 17-36. Reference: Biol. Bull. 200: 77-86. (February 2001) Muscular Alteration of Gill Geometry in vitro: Implications for Bivalve Pumping Processes SCOTT MEDLER* AND HAROLD SILVERMAN Louisiana State University, Baton Rouge, Louisiana 70803 Abstract. In bivalves, water-pumping potential is deter- mined both by ciliary activity and by the geometry of the system of passageways that acts as a conduit for water flow. Smooth muscles intrinsic to the gills of eulamellibranch bivalves possess the anatomical organization needed to reg- ulate the dimensions of these water passageways. The tone of these muscles can be controlled experimentally using excitatory neurotransmitters to elicit muscle contraction and by removing Ca++ from the Ringer's solution to induce muscular relaxation. These experimental methods were used to investigate the effects of smooth muscle tone on the gill dimensions of two freshwater bivalves, Dreissena poly- morpha and Corbicula fliuninea, and one marine bivalve, Mercenaria mercenaries In addition, endoscopic observa- tions were made from the suprabranchial chamber of a freshwater unionid, Lampsilis anodontoides. Contraction of gill muscles led to a significant reduction in interfilament width, internal ostial area, and the cross-sectional area of the water tubes. Endoscopic observation from minimally dis- turbed L. anodontoides revealed rapid constriction of the water tubes upon contraction of the muscles of the gill and gill axis. Taken together, these data support the idea that alteration of smooth muscle tone in the gill provides a mechanism for controlling water-pumping activities. Introduction Filter-feeding is a complex process, employed by a di- verse assemblage of aquatic animals, in which small parti- cles are separated from the water suspending these food items (reviewed by J0rgensen, 1990; Riisgard and Larsen, 1995). Filter-feeding animals exhibit a variety of conver- Received 23 March 2000; accepted 19 October 2000. * Author to whom correspondence should be addressed. Current ad- dress: Department of Biology. Colorado State University. Ft. Collins. CO 80523. E-mail: Skmedler@aol.com gent designs for pumping water, shaped both by the intrinsic limitations of biological systems and by extrinsic con- straints such as those arising from the physical nature of the environment (LaBarbera, 1990: Vogel, 1994). Bivalves, brachiopods, ascidians, and other distinct phyla all use the same type of low-velocity ciliary pump in their filter-feed- ing processes (LaBarbera, 1990: Vogel, 1994; Riisgard and Larsen, 1995). Most biological pumps consist of a pump and a system, where the pump represents the pressure- generating component and the series of vessels or "pipes" acting as conduits for water flow constitute the system (J0rgensen, 1989; LaBarbera, 1990; Riisgard and Larsen, 1995: Grunbaum el at., 1998). In animals that use low- velocity ciliary pumps, the beating cilia collectively repre- sent the pump, and the system consists of an incurrent region, a transfer region, and an excurrent region (LaBar- bera, 1990; Riisgard and Larsen, 1995). The term "bivalve pump" is used to describe the water-pumping processes of filter-feeding bivalves, independent of other events such as particle capture and feeding (J0rgensen et al.. 1986). Extensive research has been earned out to better under- stand bivalve pumping processes, and data on the rate of water pumping by various bivalves provide an important component needed to develop general models of the pump- ing process (Foster-Smith. 1976: M0hlenber» and Riisgard, 1979; Silvester and Sleigh, 1984: Meyhofer, 1985; J0r- gensen et al., 1986, 1990; J0rgensen and Riisgard, 1988; Kryger and Riisgard, 1988: Sihostei. 1988; J0rgensen, 1989; Jones et al., 1992: Nielsen et al.. 1993). There is general agreement that the lateral ciliated cells provide the driving force for water flow and that changes in valve gape and siphon dimension contribute to adjustments in pumping rate. However, little work has addressed the potential role of other system components in regulating pumping processes, and a detailed understanding of the overall control of pump- ing remains elusive. Debate continues as to whether pump- 77 78 S. MEDLER AND H. SILVERMAN ing and feeding are regulated only in an on/off mode (J0r- gensen et al., 1988; J0rgensen, 1996) or whether a more sophisticated physiological regulation of these processes is possible (Ward and Targett, 1989; Wildish and Saulnier, 1993; Navarro et al., 1994; Ward et al, 1997). Further refinement of bivalve pumping models requires more de- tailed knowledge of the gill itself. Models of the bivalve pump have relied heavily on the principles of fluid mechanics to develop mathematical de- scriptions of pumping processes (Foster-Smith, 1976; Sil- vester and Sleigh, 1984; J0rgensen et al., 1986, 1988; Sil- vester, 1988; Griinbaum et al.. 1998). This type of analysis requires a thorough knowledge of the animal's morphology to accurately estimate pumping properties. For example, the length and width of water passageways are used to estimate pressure losses stemming from the frictional resistance to flow. These system components are generally approached as rigid structures, with models applying fixed estimates of gill geometry to calculate system characteristics (Foster-Smith, 1976; Silvester and Sleigh. 1984; J0rgensen et al., 1986, 1988; Silvester, 1988). Although this is a logical simplifi- cation for modeling purposes, a body of evidence indicates that smooth muscles are important in affecting the geometry of the passages that constitute the system (Setna, 1930; Elsey, 1935; Atkins, 1943: Gardiner et al., 1991; Medler and Silverman, 1997, 1998). Poiseuille's law describes fluid flow in a circular pipe as: Q = 7rAfV4/8L/n,, where Q is flow rate, AP is pressure difference, /• is pipe radius, L is pipe length, and JJL, is dynamic viscosity of the fluid (LaBar- bera, 1990). One of the implications of Poiseuille's law is that relatively small changes in vessel radius result in sig- nificant alteration of fluid flow. Many organisms take ad- vantage of this principle to regulate flow by contracting or relaxing smooth muscles or muscle-like cells that line the vessels. For example, vertebrates modify arteriole diameter to regulate blood flow into capillary beds (Eckert et al., 1988), sponges change water flow by contracting or relaxing porocytes and myocytes as water enters the animal (Bagby, 1964; Pearse et al., 1987), and bivalves adjust flow through the alteration of siphon dimensions (Foster-Smith, 1976; J0rgensen et al., 1986, 1988). In this study, we focus on the ability of the eulamellibranch gill to change the dimensions of water passageways by altering the tone of integral smooth muscles. Bivalves possess several muscles that have the potential to affect water flow. The most widely recognized of these are the adductor muscles and muscles of the mantle edges and siphons that are important for controlling valve gape and siphon dimensions. The integral gill muscles are not as widely appreciated, but are also likely to play a role in basic pumping processes. Each of the individual conduits for water flow is closely associated with smooth muscle fibers that can alter the dimensions of these passageways (Setna, 1930; Elsey. 1935; Atkins. 1943; Gardiner et al., 1991; Medler and Silverman. 1997, 1998). In addition, extensive muscles in the gill axis lying dorsal to the suprabranchial chamber are important in shortening the gill (Setna, 1930; Atkins, 1943). Previous endoscopic studies have docu- mented changes in interfilament width, ostial dimension, and water-tube dimension in living animals (Tankersley and Dimock, 1993; Ward et al., 1994; Tankersley, 1996). Al- though the exact mechanism of such movements has not been clearly defined, these are consistent with the types of movements observed after the contraction of smooth mus- cles in the gill (Setna. 1930; Elsey, 1935: Atkins. 1943; Gardiner et al., 1991: Medler and Silverman, 1997). Exam- ination of gills in their fully relaxed and contracted states is useful for defining the extremes in a continuum of gill dimensions. We report here 2- to 6-fold differences between the dimensions of the water passages in fully relaxed and fully contracted gills. Materials and Methods Animals Two freshwater bivalves, Dreissena polymorpha and Corbicula flitminea, and one marine species, Mercenaria mercenaria, were used for gill measurements. Large spec- imens (shell length about 10 cm) of the freshwater unionid Lampsilis anodontoides were used for endoscopic observa- tions. C. fluminea and L. anodontoides were collected lo- cally from ponds in Baton Rouge, Louisiana. D. polymor- pha was collected from the Huron River in Michigan, and M. mercenaria was purchased from the Marine Biological Laboratory in Woods Hole. Massachusetts. Solutions Freshwater animals were maintained in aerated aquaria with artificial pondwater (APW) at 22-24°C as described in Dietz et al. (1994). Specimens of M. mercenaria were held in aerated aquaria with artificial seawater (ASW) at 4°C as specified in Chambers and De Armendi (1979). Ringer's solutions for the freshwater bivalves were prepared as de- scribed for D. polvmorpha in Dietz et al. (1994), with differences in hemolymph osmolality between species being corrected by adding NaCl to the solutions as needed. For M. mercenaria, ASW was used in place of Ringer's solution. Calcium-free solutions were prepared by omitting Ca+H from the solutions and adding 4 mM EDTA in its place (Medler et al.. 1999). Neurotransmitters were added to Ringer's solution or ASW at a concentration of 1 mM, a pharmacological dose that has been shown to elicit maximal contraction of these muscles (Medler and Silverman, 1997; Gainey et al., 1998). Acetylcholine is effective as an exci- tatory neurotransmitter in the gills of D. polymorpha (Med- ler and Silverman, 1997) and C. fluminea (Medler, unpub. obs.) and was used to stimulate contraction in these species. BIVALVE GILL MUSCLES 79 Serotonin acts as an excitatory neurotransmitter in the gill muscles of M. mercenaria (Gainey et at., 1998) and was used to stimulate contraction in this species. Gill preparation Gills were removed from animals by cutting along the dorsal connection to the body with surgical scissors and were placed in the appropriate Ringer's solution or ASW. The gills from one side of an animal were placed in a solution containing Ca++ before exposing them to an exci- tatory neurotransmitter; those of the opposite side were placed in a Ca++-free solution. Interfilament width and internal ostial area were measured from live gills as de- scribed below. Gills normally exhibit severe muscular contraction upon exposure to fixatives such as glutaraldehyde. This was not the case in the present study, since the gills with Ca+" available to trigger contraction were already contracted through exposure to excitatory neurotransmitter. In fact, any further contraction that occurred helped to ensure that the gill was in a fully contracted state. We have recently dem- onstrated that the removal of extracellular Ca++ blocks muscle contraction in the gills of D. pol\morpha (Medler et ai. 1999), and this effect is also evident for C.flitminea and M. mercenaria. Exposure of these relaxed gills to glutaral- dehyde failed to initiate any muscular contraction during the fixation process. Water-tube measurements were made from gill sections. Excised gills were fixed with a 2% glutaraldehyde solution in isosmotic phosphate buffer for the freshwater species and a 2% glutaraldehyde solution in ASW for M. mercenaria. Tissues were fixed for 1 h in glutaraldehyde, rinsed twice in either buffer or ASW, and postfixed for 1 h in 1 % OsO4. After fixation, the gills were rinsed twice in deionized water and dehydrated in a graded ethanol series. Whole gills were embedded in LR White (London Resin Co.) medium-grade resin by first placing them in a 1:1 mixture of ethanol and resin for 24 h. They were then transferred to 100% resin for 12 h and embedded flat in fresh resin at 60°C for 24-48 h. Water-tube dimension A small portion of the central region of relaxed and contracted gills (approximate midpoint along both the dor- soventral and anterioposterior axes) was cut from embed- ded, fixed gills and cross-sectioned. Sections were cut using a Reichert-Jung ultracut E ultramicrotome at 1-2 /urn thick- ness and stained with toluidine blue. The sections were viewed with a Nikon Microphot FXA using bright field optics, and the cross-sectional area of the water tube was measured from digitized video images using Image- 1 com- puter software (Universal Imaging Corp.). Because muscle contraction causes a shortening of the gill in an anteriopos- terior direction, it was necessary to account for this change in our water-tube measurements. This adjustment was made by standardizing water-tube area per unit length of gill (number of filaments spanning the anterioposterior direc- tion). Thus, water-tube area is given as (/u,m2/filaments). This correction would be unnecessary if the total cross- sectional area of the gill were measured, but measurements were made from only a portion of the gill. Failing to correct for the shortening would overestimate the cross-sectional area in contracted gills. Water-tube dimensions from re- laxed and contracted gills from individuals of each species were compared with paired t tests (n = 5). Studies using fixed and dehydrated tissues are sometimes criticized for introducing artifacts due to shrinkage. Indeed, these procedures do lead to changes in gill dimensions that should be noted when an accurate measure of absolute dimensions is critical (Silverman et ai, 1995). In the present study, we were interested only in comparing the relative differences between important dimensions in the fully re- laxed and fully contracted states. When the gills were fixed for measurement of water-tube area, each gill pair came from an individual animal and was processed with the same fixatives, dehydration steps, and embedding. Thus, any shrinkage is expected to be proportional in the relaxed and contracted gills, leaving the relative change unaffected. Interfilament width Live gills were placed on microscopic slides in the ap- propriate solution and covered by coverslips elevated on posts of petroleum jelly to prevent the gills from being compressed. Gills were examined with differential interfer- ence contrast (DIC) optics on a Nikon Microphot FXA. Interfilament distances from digitized video images were measured as described above, and distances were calibrated with a stage micrometer. Interfilament widths from relaxed and contracted gills from individuals of each species were compared by paired / tests (n = 5). Internal ostial area Live gills were split into single lamellae and placed in an irrigation chamber as described in a previous study (Medler and Silverman, 1997). Gill lamellae were placed in Ca+" free solutions to completely relax their musculature before any measurements were made and remained in this solution when placed into the chamber. The internal water-channel epithelium was placed toward the bottom of the chamber so that the internal ostia could be observed using an inverted Nikon microscope with Hoffman modulation optics. After the relaxed gill ostia were measured, the chamber was irrigated with Ringer's or ASW containing an excitatory neurotransmitter. The gill was observed and videotaped as it contracted; the ostia were remeasured once contraction was complete (about 1 min later). Ostial areas (jinn2) were measured from digitized video images as described above 80 S. MEDLER AND H. SILVERMAN and were calibrated with a stage micrometer in the experi- mental set-up. Ostial areas from relaxed and contracted gills from individuals of each species were compared with paired t tests (n = 5). Scanning electron microscopy Gills in contracted and relaxed states were fixed as de- scribed above. After dehydration, the gills were wrapped in lens paper, critical-point dried, and mounted on stubs. De- hydrated gills were either sectioned or split apart to reveal relevant regions of the gills. Specimens were sputter coated with a mixture of gold and palladium (20 nm) and viewed with a Cambridge S-260 scanning electron microscope. Digitized video images were enhanced for optimal bright- ness and contrast using Adobe Photoshop 5.0 software (Adobe Systems, Inc.). Video endoscopy of live gills An optical insertion tube (OIT) was inserted into the suprabranchial chamber of large specimens of L. ancdon- toides using the general approach described by Tankersley ( 1996). This species was selected because individuals tend to gape widely, allowing observations to be made without wedging the valves open. Animals were placed in an aerated container of APW (about 4 1) and fixed in position by means of a nylon bolt cemented to one valve. The OIT (1.7 mm diam. X 101 mm long; AEI North America) was attached to a 150 W halogen fiberoptic light source and inserted through the exhalent aperture into the suprabranchial cham- ber. A mirror sleeve was attached to the OIT to provide the 90° view needed for direct observation of the water tubes. The OIT was attached to a zoom adapter that provided a maximal magnification of about 150X. Maximum resolu- tion was estimated to be approximately 5 /urn at maximum magnification. The OIT and zoom adapter were coupled to a Costar color video camera (0.85 cm CCD model CV-730) mounted on a microscope stage. The microscope stage served as a micromanipulator, allowing movements in the X. Y, and Z planes. Observations were recorded on VMS videotape, and digitized video images of portions of these recordings were captured using Image- 1 computer software. Images were adjusted for brightness and contrast using Adobe Photoshop 5.0 software. No pharmacological agents were used during endoscopy. Animals were held in a darkened room, and once the endo- scope was positioned, there was minimal disturbance to the animals. Changes in the geometry of the gills and su- prabranchial chamber were spontaneous, not resulting from any discernible stimulus. Results Water flow through the eulamellibranch gill begins as water moves between parallel filaments and into the exter- nal ostia that lead into the water canals of the gill (Fig. 1 ). These canals empty through internal ostia into the central water channel that separates the ascending and descending gill lamellae. For clarification, some studies use the term ••interfilament canal" to mean the space between filaments leading all the way into the central water channel of the gill (Jorgensen el ai. 1986), whereas others use the term "ostia" for the same canal system (Foster-Smith, 1976). We use the term '•interfilament space" more strictly, as defining the region between filaments on the frontal face of the gill, and the term "ostia" to mean the openings of the water canals that lead from the interfilament space to the central water channel. The central water channel is partitioned into water tubes by septae that connect the two opposing gill lamellae. Water moves dorsally through the water tubes before emp- tying into the suprabranchial chamber and then out of the excurrent siphon. Each of the dimensions from gills with relaxed muscles was significantly larger than those with muscles contracted. Interfilament width was approximately 20 /nm in relaxed gills from each species and decreased to less than 10 /im after muscle contraction (Fig. 2). Contracted gills had sig- nificantly narrower interfilament widths than relaxed gills for each of the three species examined (mean ± SE): Corbicnlti flwmneu (23.3 ± 0.7 vs. 8.7 ± 1.1 /urn; P < 0.002), Dreissemi polyinorplui (19.2 ± 2.1 v.v. 3.3 ± 0.3 JU.ITT. P < 0.001 ), Mercenaria mercenaria (22.7 ± 1.7 vs. 6.5 ± 0.6 /urn: P < 0.001) (Fig. 2). In fully contracted sills, filaments viewed from the frontal surface appear to abut one another, often producing a zig-zag pattern along the length of the filaments (Fig. 2c). At the level of the lateral ciliated cells, even these gills have about 5 /xm between the apical cell surfaces. This apparent discrepancy results from the fact that extended cilia project from the apical cell surfaces and obscure the small gap remaining between filaments. Internal ostial dimension (per ostium) was quite variable between species, with C. ftuminea and M. mercenaria hav- ing larger ostia than D. polymorpha (Fig. 3a). After mus- cular contraction, gills had significantly smaller individual ostial areas for each of the three species examined (mean ± SE): C.flnminea (9052 ± 701 v.v. 3674 ± 404 junr; P < 0.002), D. pohmorpha (3175 ± 436 vs. 1123 ± 393 ju,m2; P < 0.022), M. mercenaria (16082 ± 1283 vs. 9949 ± 1186 /xnr: P < °-002> (Fi£- 3>- Water-tube dimension was also variable between species, with larger species having larger gills and water tubes (Fig. 4a). As the muscles of the gill contracted, the cross-sec- tional area of the water tubes significantly decreased in each species (mean ± SE): C.flwnineu (8191 ±727 vs. 4164 ± 364 janr/filament; P < 0.002), D. polymorpha (3486 ± 545 v.v. 502 ± 161 M,m2/filament: P < 0.005). M. merce- naria (142982 ± 31212 v.v. 22156 ± 5084 /inr/filament: P < 0.001) (Fig. 4). BIVALVE GILL MUSCLES 81 ,( «'" ""• •C A t S SL « I • Hi ^V Ar dt • , o .. «' M. V Llr-rOSS-StrutS— ; r • Figure 1. Eulamellibranch gill organization and water flow. (a. b) General organization of a single gill lamella with epithelial tissue removed to reveal the organization of the supporting structures (modified from Medler and Silverman. 1998, with permission from Invertebrate Biology), (a) Dreissena polymorpha gill (SEM) showing parallel filaments ( /') held apart by connective tissue cross-struts. Bands of smooth muscles are found below the cross-struts (b); during muscle contraction, the cross-struts bend inwardly and the filaments are drawn closer to one another. In addition to these muscle bands, more diffuse smooth muscle fibers (depicted as thin wavy lines) are found in the connective tissue sheets that enclose the hemocoel (/i) at the base of the filaments (b). Water is drawn into the central water channel of the gill through water canals (indicated by arrows in b). Ostia (dark openings perforating the gill lamellae) represent the ends of the water canals and provide a route for water to flow into the central region of the gill. (c, d) Frontal (c) and cross-sectional (d) views (SEM) of Merci'inirin nn-rcenaria gill, demonstrating the movement of water (indicated by arrows in c) through the water tubes of the gill after entering the gill through the water canals. The route of water flow through the gill is between the gill filaments and then through the gill ostia and associated water canals. Water moves to the suprabranchial chamber (not shown) though the water tubes of the central water channel. The suprabranchial chamber at the top of each gill is connected directly to the excurrent aperture where the water leaves the animal. Endoscopic observations from the suprabranchial chamber of Lampsilis anodontoidt-s revealed distinct changes in the geometry of the chamber and water tubes after spontaneous muscle contractions that caused obvious shortening of the gill and suprabranchial chamber. One such change was a rapid reduction in the cross-sectional area of the water tubes, which occurred during a time period of less than 5 s (Fig. 5). Part of the reduction in water-tube area resulted from a shortening of the gill axis, caused by contraction of a large bundle of muscle fibers located dorsal to the suprabranchial chamber and running in the anterioposterior direction. These muscle bundles were described for a number of species by Atkins (1943) and were identified in transverse sections of the dorsal gill region of L mwdontoidcs (data not shown). Many of the observed gill responses were not accompanied by any other obvious changes, such as valve closure. In fact, the valves usually continued to gape widely throughout the observational period. 82 2a 0) S, MEDLER AND H. SILVERMAN relaxed contracted Figure 2. Intertilament distance, (a) Relaxed gills from each of the species (Corbicitln fluminea. Dreissena polymorpha. and Mercenaria mercenaria) had an intertilament distance of about 20 /xm, but this distance was significantly reduced to less than 10 /j.m upon muscle contraction (mean ± SE: ;; = 5 ). (b, c) Frontal face (SEM I of a C. fluminea gill: (b) relaxed gill; (c) gill from the same animal following muscle contraction. Discussion Two of the most important factors affecting bivalve pumping potential are the dimensions of the intertilament passages and the exit loss (derived from the kinetic energy carried by the water jet leaving the excurrent siphon) (Fos- ter-Smith. 1976: J0rgensen et ai.. 1986, 1988; J0rgensen and Riisgard. 1988; J0rgensen, 1989; Riisgard and Larsen, 1995). Both of these factors are ultimately controlled by muscle tone, since integral gill muscles control gill dimen- sions and exit loss is controlled by the muscles affecting siphon dimensions. It is reasonable to expect higher pump- ing rates when the gill muscles are fully relaxed, because in this condition the passageways for water flow are signifi- cantly more open than when the muscles contract. J0rgensen and colleagues have long held that pumping activities are correlated with the degree of valve gape and the associated changes of the gill (J0rgensen et til.. 1986, 1988: J0rgensen and Riisgard. 1988; J0rgensen, 1989. 1990). Their interpre- tation is that gill dimensions are controlled secondarily to the contraction of muscles within the gill axis and that the muscles of the gill axis contract when the adductor muscles reduce valve gape (Jorgensen et ai. 1988; J0rgensen, 1989, 1990). We agree with this general description, but would refine it by stipulating that intertilament distance and other gill dimensions are controlled directly by integral smooth muscles. Part of this distinction relates to the fact that J0rgensen's model is based on animals that possess fili- branch gills, rather than the eulamellibranch organization described here. In filibranch gills, there is no direct connec- tion between adjacent filaments aside from that made through the ciliary discs. Nevertheless, these discs clearly cause changes in interfilament distance even in excised gill fragments independent of the gill axis (J0rgensen, 1976; Jones et ai. 1992; Medler, unpub. obs.). In addition, such movements are probably aided by the muscles that attach at the base of the ciliary discs in many filibranch species (Atkins. 1943). Jorgensen and colleagues have highlighted alteration of interfilament distance as a central control mechanism for regulating pumping activities (J0rgensen et ai, 1986. 1988; J0rgensen and Riisgard. 1988; J0rgensen, 1989, 1990). There are several possible consequences of changes in 3a 10000 7500- BIVALVE GILL MUSCLES 83 relaxed contracted Figure 3. Ostial area, (a) Area of the internal ostia of relaxed gills from each of the species (Corbicula fluminea, Dreissena polymorpha, and Mercenaria mercenaria) was significantly reduced upon muscular con- traction (mean ± SE: n — 5 ). (b. c) Internal face (SEM) of a D. polymorpha gill: (b) relaxed gill; (c) gill from the same animal following muscle contraction. The relative position of septa (s ) that connect opposing lamellae and form the water tubes give an indication of shortening in the anterior to posterior axis. interfilament distance, with the most obvious being that a change in distance will affect the resistance to water flow (Foster-Smith, 1976; J0rgensen el at., 1986). In addition, it has been suggested that as the filaments move toward one another, the lateral ciliated cells responsible for establishing water flow begin to interfere with one another (J0rgensen et ai, 1988; J0rgensen, 1989, 1990). Recent mathematical modeling of gill dimensions has provided further insight into the specific consequences of a particular interfilament distance (Griinbaum et ai, 1998). One of the predictions of this model is that the optimal interfilament width for max- imizing water flow depends upon the pressure gradient producing the flow. At low pressure differences, a ciliary gap of near 20 jum is optimal; as pressure differences increase, the optimal gap decreases toward 5-10 /am (Griin- baum et al., 1998). This range corresponds well with the distances observed for each of the species in this study (Fig. 2a). When the gill muscles are relaxed, the interfilament width is close to 20 p,m, but can quickly change to 10 ;u.m or less as the muscles contract (Medler and Silverman, 1997; Fig. 2a). An environmental variable that may dictate adjustments of pump properties is water temperature, since water viscosity changes inversely with temperature and has direct effects on pumping activities (J0rgensen et al., 1990). It has also been suggested that changes in interfilament width may provide a mechanism to adjust feeding rate and efficiency (Ward et al., 1998). The data in this and previous studies provide evidence that interfilament distance is con- trolled directly by the activity of smooth muscles, allowing these animals to adjust interfilament distance as needed (Gardiner et ai, 1991; Medler and Silverman, 1997). Connecting the interfilament spaces with the su- prabranchial chamber are the passages that constitute the "pipes" of the gill. As indicated by Griinbaum et al. ( 1998). 84 S. MEDLER AND H. SILVERMAN 4a c is the "association" parameter of H,O ( =2.6); and V is the molal volume of the solute at its normal boiling point (370 cm3 g mol"1 for disodium fluorescein). Through computer simulations, the rates of solute release from acrylamide were determined by applying Eq. 1 to cylindrical gels with lengths ranging from 0.1 to 10 cm for up to 5 days of field exposure. Fluorescein was chosen as a tracer because its molecular mass (and. hence, diffusion coefficient) is close to that of the small peptides used in the settlement assays (330 g moP ' versus 269 and 288 g mol~' for GGH and GGR. respectively). The diffusion coefficients for fluorescein in 8.0% acrylamide (Dr) and H2O (D0) were calculated from Eq. 2 and Eq. 3, respectively, to be 2.9 X 10~6 and 4.4 X 10~6 cm2 s~' at 27 °C. From these simu- lations, it was clear that at lengths greater than 4-5 cm, gels release very little (<50%) of their included organic com- pounds in 3-5 d (Figs. 1 and 2). If the gel is too thin (<0.8 cm), however, more than 90% of the solute is lost within 0.5 d. Chemical-releasing collectors (CRCs) were engineered to test the effects of waterborne cues on larval settlement in natural habitats. Each CRC was built of a 4.0-cm-long polyvinyl chloride (PVC) cylinder, 9.5-cm i.d., sealed at one end with a flat sheet of PVC. Three 12-cm-long eye- bolts were fastened symmetrically around the cylinder. 2 cm from either end. A 1.0-cm layer of acrylamide (71 ml) in ASW, with or without added stimulant, was polymerized nearest the PVC sheet. One centimeter of crushed, cleaned C,/C0 Figure 1. Fluorescein remaining in an 8.0% polyacrylamide gel as a function of time and gel length. These data were generated from Eq. 1 using Dp = 2.9 x 10 " cm2 s~' and four terms (0-3) in the Bessel series. The fraction of dye in the gel has been non-dimensionalized by division with the ; = 0 values (C,/C,,K oyster shell hash (pieces 0.5-1.0 mm in diam) was layered over the acrylamide to form a diffusional dampener. The remaining 2 cm of the CRC were filled with six large, flat oyster shells (surface area cti. 240 cm2) set flush with the collector opening to provide settlement substrata. Oyster shells (sun bleached >6 months) from the local estuary 4 6 Length (cm) Kjgure 2. Plot of TI/:, versus gel length. The nonlinear curve has been fitted to a third-order expression relating the variation in the residency time (TI of a small solute within a gel to the gel length (TI/: = -0.00911 x Length3 + 0.202 X Length2 + 0.196 X Length - 0.127; R2 > 0.999, n = 24). The slope of the polynomial regression of log C,/C,, versus time is inversely proportional to the half-life, or chemical residency time, in the gel matrix (rl/2 = 0.3/slope). FIELD EXPERIMENTS ON WATERBORNE SETTLEMENT CUES 89 were prepared for use as settlement substrata by scrubbing free all sediment, flora and fauna, bathing for 5 min in 2%-3*7r NaOCl, rinsing >12 h in continuously running H,O: and autoclaving to sterilize. These shells were found to be superior to preliminary systems containing scoured glass or PVC substrata (data not shown). Monofi lament mesh (0.6-cm pore size) was secured over the shells with cable ties tightened around the cylinder portion of each CRC. The field experiments were conducted in the North Inlet Estuary (Town Creek; 33° 19.5' N, 79° 11.5' W) near Georgetown, South Carolina. This tidal creek is shallow and rapidly flushed, with flow speeds up to 35 cm s~ ' (and shear velocities to 3.8 cm s"1) about 2200 m from the mouth of the inlet. The creek bank falls steeply. 2-3 m, to a mud bottom. Deployments occurred during August, when the mean (±SEM) water temperature and salinity were 28.3 ± 0.2 °C and 30.3 ± O.&Zc. respectively (P. Kenny and D. Allen, unpubl. data). The CRCs were held above the creek bed, at least 20 cm below mean low tide level, by harnesses constructed of 0.6-cm-diam. nylon rope and brass fittings attached to wooden posts embedded into the sediment. Concentrations of diffused solute from acrylamide gel were measured using disodium fluorescein and GGR as chemical tracers in the field. The experiment employing fluorescein was performed to elucidate the effects of tracer release rate. GGR determinations were subse- quently made for seawater sampled during tests of larval settlement. Three replicate 1-ml aliquots of seawater were removed via sterile syringes from 0.5, 1.1, and 3.5 cm above the gels. The samples were filtered on site with ().45-jum syringe filter cartridges, stored in sterile vials, and placed on ice in the dark until analysis was per- formed (GGR samples were maintained at —80 °C upon return to the laboratory). The sampling was repeated for 12 or 4 replicate collectors at 12-h and 24-h intervals in the fluorescein and GGR experiments, respectively, for a total of 12 h. Fluorescein concentrations were determined from fluorescent output (Turner Designs field fluorome- ter, model 10-AU-005. excitation wavelengths 455-500 nm. emission wavelengths 510-700 nm) by comparison to a standard concentration curve. GGR was quantified by high-performance liquid chromatography (HPLC) with a Beckman System Gold 126 binary solvent module/507 autosampler and a Jasco FP-920 fluorescence detector. Briefly, the guanidinium group of the arginine on GGR reacted with benzoin under basic conditions to quantita- tively form a highly fluorescent derivative (21. 22). Sep- aration through a silica-based bonded phenyl column (Waters Corp.) was induced by increasing the concentra- tion of ethyl alcohol in the 75 mM Tris-HCl (pH 8.1) mobile phase from 25% to 85% in a series of gradients over 23 min at 0.8 ml min"'. HPLC peaks (excitation A = 325 nm, emission A = 435 nm) were identified by their retention times relative to previously run standards (guanidinium. arginine, glycyl-L-arginine, and GGR). and concentrations were calculated from the areas under the peaks (23). Previous laboratory investigations with barnacle and oys- ter larvae indicated maximum settlement in response to peptides at 10~7 to 10~8 M ( 15, 16). Concentrations greater or less than two orders of magnitude from these target doses substantially diminished the effectiveness of GGR to stim- ulate settlement. A gel concentration of 1 mg mP1 fluores- cein produced mean levels that were nearly constant through 72-h field deployments. These values were as fol- lows: for shell-hash pore waters: 1.1 (±0.1 SEM) X 10~: mg ml"1; for settlement (shell) substrata: 2.1 (±0.1 SEM) X 10~4mg ml"1; and, for the water column at 0.5 cm above the collector: 1.0 (±0.1 SEM) X 10~6 mg ml"1. On the basis of this information, a concentration of 2 X 10~4 M peptide was established for the gels. A 3700-fold dilution led to a mean GGR concentration of 5.4 (±0.5 SEM) X 10~8 M in the settlement substrata during the field trials, in good agreement with our measurements of fluorescein re- lease. The inclusion of the shell-hash layer above the gel resulted in near steady-state concentrations of fluorescein and GGR. Solute concentrations in shell-hash, settlement substrata, and water at 0.5 cm above the collector did not significantly change as a function of time since placing the CRCs in the field (Table 1, and one-way ANOVAs: F < 1.23. P > 0.37, all comparisons). This field study was intended to determine the settlement responses of barnacle and oyster larvae to a waterborne chemical cue. The densities of larvae in the water column were estimated at the beginning and end of the deployment from the number of larvae collected at the study site in paired 3-6 min oblique tows (>4 m3 filtered seawater) taken with 30-cm-diam., 153-ju.m-mesh Nitex plankton nets. The results indicated that oyster larvae in the water column were rare (<2.1 individuals m"3) during the course of the investigation, so only data for barnacle larvae (mean con- centration: 40.4 larvae m ) are presented here. The effects of ASW, glycyl-glycyl-L-arginine in ASW (GGR: settlement stimulant), and glycyl-glycyl-L-histidine in ASW (GGH; organic enrichment control) were deter- mined by assaying eight CRCs of each chemical treatment. The exact position of each CRC was established according to a randomized block design in which eight sets of three collectors (one collector of each treatment) were placed along a 1 20-m transect parallel to and 4 m from shore. The CRCs were incubated in the fk K! \vith the open ends ori- ented upward for 72 h. At the end of the incubation time, the chambers were removed from the estuary and the number of newly metamorphosed barnacle (Balanim amphhrite} juve- niles on the settlement (shell) substrates were counted under a dissecting microscope. GGR was found to stimulate substrate colonization by sig- 90 K A. BROWNE AND R. K. ZIMMER Table 1 Mean (±SEM) concentrations of disodium fluorescein (fjLg/ml) and gl\c\/-g/vcv/-L-arginine (GGR; nM) in waters sampled from among the shell substrata of lan-al collectors Time (h) since collector was placed in the field Compound 1 2 24 36 48 60 72 Disodium fiuorescein 0. 18 ± 0.05 GGR — 0.22 ± 0.06 58 ± 13 0.21 ± 0.04 0.25 ± 0.08 59 ± 16 0.22 ± 0.07 0.20 ± 0.04 45 ± 18 Time = 0 is defined as the point at which collectors were first submerged in field habitats and for which neither fluorescein nor GGR had yet been released from polyacrylamide gels. Separate experiments were performed for fluorescein and GGR. Initial concentrations of the compounds in acrylamide gels were 1 mg/ml (fluorescein) and 2 x 10~4 M (GGR). Seawater samples of fluorescein and GGR were collected at 12-h and 24-h intervals, respectively. nificantly more barnacle larvae than did either ASW or GGH (Fig. 3). Because GGR and GGH are basic peptides with nearly identical molecular weights and chemical functional- ities, settlement induction cannot be attributed to stimulation by organic compounds in general but is due. specifically, to the arginine moiety at the carboxy terminus of the peptide. These results show that even weakly swimming iarvae like barnacle cyprids (mean speed: 2.9 mm s '; 24) can 25- 20- I 15- 10- ASW GGH GGR Figure 3. The numbers (mean ± SEM; n = 8) of newly metamor- phosed barnacle juveniles (Balanus amphitrite) on clean oyster shells, provided as settlement substrata, were counted for chemical-releasing collectors (CRCs) with either no organic enrichment (ASW, control) or with 2 x 10~4 M GGR or GGH. Significantly more barnacle juveniles were found in response to GGR than to the control or GGH (P < 0.001. one-way ANOVA followed by multiple, paired comparisons using Bon- ferroni's correction; 27). All CRCs were oriented with their openings facing upward with an 8.0% acrylamide gel (2.4% crosslinking) prepared using sterile ASW. ASW, artificial seawater: GGH. glycyl-glycyl-L-histi- dine; GGR. glycyl-glycyl-L-arginine. select their settlement sites upon receipt of an appropriate chemical stimulus in turbulent flow. After being released from a benthic source, such as patches of the conspecinc target organism, a peptide signal molecule can be trans- ported a considerable distance above the bottom by turbu- lent eddies (25). Larval delivery to the seabed may thus result from hydrodynamic transport of larvae that also may swim down in response to dissolved signal molecules. The latter process tends to concentrate larvae near the bed, enhancing settlement rate (26). In this manner, water-solu- ble cues might remotely entrain larvae to settle in suitable adult habitat, and therefore warrant careful attention as putative agents mediating supply and delivery of planktonie larvae to benthic environments. Acknowledgments This study was supported by an award from the NOAA Sea Grant College program (R/CZ-152) through the Na- tional Marine Biotechnology Initiative. D. Allen and P. Kenny generously provided unpublished data on tempera- ture, salinity, and plankton counts for North Inlet waters. Earlier drafts of this manuscript were greatly improved by comments from M. S. Gordon, and especially from C. A. Zimmer. Literature Cited 1. Crisp, D. J. 1974. Factors influencing settlement of marine inverte- brate larvae. Pp. 1 77—265 in Chemoreception in Marine Organisms. P. T. Grant and A. M. Mackie. eds. Academic Press, London. 2. Morse, A. N. C., and D. E. Morse. 1984. Recruitment and metamor- phosis of Halintis larvae induced by molecules uniquely available at surfaces of crustose red algae. J. E.\p. Mar. Bio/. Ecol. 75: 191-215. 3. Butman. C. A. 1987. Larval settlement of soft-bottom invertebrates: the spatial scales of pattern explained by active habitat selection and the emerging role of hydrodynamical processes. Oceanogr. Mar. Biol. Awui. Rev. 25: 113-165. 4. Gaines, S. D., and M. D. Bertness. 1992. Dispersal of juveniles and variable recruitment in sessile marine species. Nature 360: 579-580. 5 Pawlik, J. R.. and C. A. Butman. 1993. Settlement of a marine tube worm as a function of current velocity: interacting effects of hydrody- namics and behavior. Limnol. Oceanogr. 38: 1730-1741) FIELD EXPERIMENTS ON WATERBORNE SETTLEMENT CUES 91 6. Gross, T. F., F. E. Werner, and J. E. Eckman. 1992. Numerical modeling of larval settlement in turbulent bottom boundary layers. ./. Mar. Res. 50: 61 1-642. 7. Eckman, J. E., F. E. Werner, and T. F. Gross. 1994. Modelling some effects of behavior on larval settlement in a turbulent boundary layer. Deep-Sea Res. II— Top. Stud. Oceanogr. 41: 1 85-208. X. Hadfield, M. G., and D. Scheuer. 1985. Evidence for a soluble metamorphic inducer in Phestilla: ecological, chemical and biological data. Bull. Mar. Sci. 37: 556-566. 9. Forward. R. B.. Jr., R. A. Tankersley, D. Blonde), and D. Rittschof. 1997. Metamorphosis of the blue crab Callinectes sapidus: effects of humic acid and ammonium. Mar. Ecol. Prog. Ser. 157: 277-286. 10. Fleck, J., and W. K. Fitt. 1999. Degrading mangrove leaves of Rhizophora mangle Linne provide a natural cue for settlement and metamorphosis of the upside down jellyfish Cassiopea xamaclnma Bigelovv. J. Exp. Mar. Biol. Ecol. 234: 83-94. I I . Pawlik, J. R. 1992. Chemical ecology of the settlement of benthic marine invertebrates. Oceanogr. Mar. Biol. Anna. Rev. 30: 273-335. 12. Steinberg, P. D., R. de Nys, and S. Kjelleberg. 2001. Chemical mediation of surface colonisation. In Marine Chemical Ecology. I. B. McClintock and B. J. Baker, eds. CRC Press. St. Petersberg, FL. (In press). 13. Rittschof, D. 1985. Oyster drills and the frontiers of chemical ecol- ogy: unsettling ideas. Am. Malacol. Bull. Special Edition No. 1: 111-116. 14 Zimmer-Faust, R. K., and M. N. Tamburri. 1994. Chemical iden- tity and ecological implications of a waterborne. larval settlement cue. Limnol. Oceanogr. 39: 1075-1087. 15. Tegtmeyer, K., and D. Rittschof. 1989. Synthetic peptide analogs to barnacle settlement pheromone. Peptides 9: 1403-1406. 16. Browne, K. A., M. N. Tamburri, and R. K. Zimmer-Faust. 1998. Modelling quantitative structure-activity relationships between animal behaviour and environmental signal molecules. J. Exp. Biol. 2(11: 245-258. 17. Ogston, A. G. 1958. The spaces in a uniform random suspension of fibres. Train. Faruday Soc. 54: 1754-1757. 18. White, M. L., and G. H. Dorion. 1961. Diffusion in a crosslinked acrylamide polymer gel. J. Polymer Sci. 55: 731-740. 19. Davis, B. K. 1974. Diffusion in polymer gel implants. Proc. Null. Acad. Sci. USA 71: 3120-3123. 20. Fawcett. J. S., and C. J. O. R. Morris. 1966. Molecular-sieve chromatography ot proteins on granulated polv acrylamide gels. Sep. Sci. 1: 9-26. 21. Ohkura, V., and M. Kai. 1979. Fluorometric determination of monosubstituted guanidino compounds with benzoin-dimethylform- amide reagent. Anal. C/um. Acta 196: 89-94. 22. Kai, M., T. Miyazaki, M. Yamaguchi. and Y. Ohkura. 1983. High-performance liquid chromatography of guanidino compounds using benzoin as a pre-column fluorescent derivitization reagent. J. Chromatogr. 268: 417-424. 23 Decho, A. W., K. A. Browne, and R. K. Zimmer-Faust. 1998. Chemical cues: why basic peptides are signal molecules in marine environments. Limnol. Oceanogr. 43: 1410-1417. 24. Tamburri, M. N., and R. K. Zimmer-Faust. 1996. Suspension feeding: basic mechanisms controlling recognition and ingestion of larvae. Limnol. Oceanogr. 41: 1188-1197. 25. Zimmer-Faust, R. K., M. N. Tamburri. and A. W. Decho. 1997. Chemosensory ecology of oyster larvae: benthic-pelagic coupling. Pp. 37-50 in Sensory Ecology and Physiology of Zooplankton, D. L. Hartline. P. Lenz, and J. E. Purcell. eds. Gordon and Breach. Toronto. 26 Tamburri, M. N., C. M. Finelli, D. S. Wethey, and R. K. Zimmer- Faust. 1996. Chemical induction of larval settlement behavior in flow. Biol. Bull. 191: 367-373. 27. Sokal, R. R., and F. J. Rohlf. 1981. Biometry. 2nd ed. W. H Freeman. San Francisco. CA. Reference: Bio/. Bull 200: 92-96. (February 2001) Molecular Evidence for Cryptic Species of Aurelia aurita (Cnidaria, Scyphozoa) MICHAEL N. DAWSON' AND DAVID K. JACOBS' lOrganismic Biology, Ecology, and Evolution, University of California, 621 Charles E. Young Drive South. Los Angeles, California 90095-1606; and 2Coral Reef Research Foundation, Koror, Pa/mi Morphological taxonomy suggests that marine faunas are species poor compared to terrestrial and freshwater faunas ill This dichotomy has been attributed to the unique po- tential of marine plankters for distant dispersal across homogenous oceans with few barriers to gene flow (2). The relative scarcity of opportunities for allopatric divergence has resulted in depauperate marine faunas characterized by a high proportion of widespread or cosmopolitan species. Aurelia aurita (Linnaeus) /HIS been considered a good ex- ample of such a cosmopolite (3. 4, 5. 6). However, recent molecular studies have revealed cryptic species in many marine ta.\a (7), suggesting that marine biodiversity is higher and opportunities for speciation have been more frequent than generally recognized. Here, we present nu- clear and mitochondrial DNA sequence evidence of seven sibling species of Aurelia aurita and two additional species, A. limbata Brandt and A. labiata Chamisso & Evsenhanlt. These sequence data indicate speciation events as early as the late Cretaceous or earlv Tertian', consistent with the formation of well-recognized biogeographic barriers to gene flow in the seas. Traditionally, the genus Aurelia comprises two species: A. limbata, a polar species, and A. aurita. a common inhab- itant of nearshore waters circumglobally between about 50 °N and 55 °S (3. 4. 5; Fig. 1 ). Perhaps due to its ubiquity, /-I. aurita has become a popular research organism for stud- ies as diverse as protein chemistry, development, ecology, ethology, and hydrodynamics (6). A. aurita also is econom- ically important because worldwide it preys on or competes with larvae of commercial fisheries and because swarms of medusae may impede trawling or block power-plant intakes (8). Furthermore, this "pest" has been introduced at least Received 7 August 2000; accepted 19 October 2000. * To whom correspondence should be addressed at Coral Reef Research Foundation, Box 1765. Koror. PW 96940. Palau. into San Francisco Bay (9) and possibly many other places ( 10). A. aurita is also familiar to nonspecialists because it is the most commonly displayed medusa in public aquaria. The systematics of A. aurita therefore is of considerable scientific, economic, and general interest. Aurelia has a typical bipartite scyphozoan life history in which benthic scyphopolyps asexually strobilate ephyrae that grow into sexual medusae, the females of which brood larvae that settle into the shallow coastal benthos within a few days of being released. Of these life stages, the medusa probably is the principal dispersal phase because only the medusa is both long-lived (several months to more than one year; 1 1 ) and planktonic (6). The potential of medusae for distant dispersal is consistent with the current classification of A. aurita as a circumglobal. almost cosmopolitan, species (4. 5 1. However, A. aurita medusae in Saanich Inlet, British Columbia, and perhaps elsewhere, migrate directionally (12). maintaining breeding aggregations within isolated in- lets and probably limiting gene flow among populations. Consistent with limited gene flow, allozyme differences have been found between populations of A. aurita in the eastern and western Atlantic Ocean, the Gulf of Mexico, and the eastern and western Pacific Ocean (9. 13). In addi- tion, one species. A. labiata. was recently recognized as native to Pacific North America and distinct from A. aurita (10). Novel DNA sequence data from nuclear internal tran- scribed spacer one (ITS-1) and mitochondrial cytochrome oxidase c subunit I (COI) reveal highly structured gene genealogies and at least nine distinct clades of Aurelia (Figs. 1,2). Several lines of argument suggest that these clades warrant recognition as distinct species. First, the length of ITS-1 varies from 240 nucleotides (Charlestown. RI) to 360 nucleotides (Cananeia, Brazil). Such length variation is comparable to that found among congeneric species of 92 AUREL1A AURITA SPECIES COMPLEX 93 180° 120°W 60°W 60°E 60°N 0 --V 30°S okyo Bay ©Palau , ©Helen Reef Newport BayJ© Tomales Bay Monterey Bay ^yBoston Harbor Narragansett Bay Marina del Rey Long Beach Newport BeaclA ^Qn F^iAor* » V San Cananeia Fjord|-600N Bjornsund Fjord •30°N -0° 180° 120°W 60°W 30°S 0° 60°E Figure 1. Sample locations of Aitrelia in this study include sites within the ranges of the three currently recognized morphospecies: A. labiata (Pacific North America), ,4. limbaia (polar seas below 70 °N from western Greenland to Siberia), and A. aitrita (Europe [the type locality], northern Atlantic, and all remaining regions). Samples from Narragansett Bay comprise two sites: the semi-enclosed Charlestown Salt Pond and the open Greenwich Cove. Samples from Palau comprise several sites from three different habitats: land-locked, generally meromictic. marine lakes (Big Jellyfish Lake. Ongael Lake, Ongeim'l Tketau. Hotwater Lake): semi-enclosed holomictic 'coves' or similar waters (Tab Kukau. Malakal. Tketau); and the lagoon, which is open to the ocean (Ngell Channel) [see refs. 29. 31]. The population in Mljet inhabits a semi-enclosed meromictic "cove." Gut. bell margin, or oral arm tissues were cleaned, biopsied. and preserved in DMSO + NaCI (29). Total DNA was extracted using a CTAB extraction protocol. PCRs were set up according to the guidelines issued with Taq polymerase (Perkin Elmer) in MJ Research MiniCyclers and amplification products cloned using Invitrogen's Topo TA Cloning Kit. Cloned DNA was purified with Pharmacia's Flexiprep kit and sequenced on Applied Biosystems 373 autosequencers according to protocols in the ABI PRISM manual. Sequencing employed the PCR primers (HCO [32]. AaCOIi-L = 5'-gcccgtyytaataggrgggtttgg. jfITSl-5f = 5'-ggtttccgtaggtgaacctgcggaag- gatc. and jfITSl-3r = 5'-cgcacgagccgagtgatccaccttagaag [29]) and Invitrogen's MI3Reverse and M13 Forward primers. Corallimorpharia (286 to 350 nucleotides; 14) and between genera within the Scleractinia (214 to 297 nucleotides; 15) and Octocorallia (216 to 336 nucleotides; 15). Second, inter-clade percent sequence differences in ITS-1 (exclud- ing regions containing microsatellites), which range from 10% (A. labiata [Kachemak, AK] vs. Aurelia sp. 1 [New- port Beach. CA]) to 40% (Aitrelia sp. 2 [Cananeia] vs. Aurelia sp. 4 [Ongeim'l Tketau]), also are more similar to inter-specific and inter-generic than intra-specific differ- ences in other taxa. For example, percent sequence differ- ence in ITS-1 among congeneric species ranges from 4.5% to 14.9% in the Corallimorpharia (14) and from 5.8% to 19.5% in mosquitoes (16). and is less than 29% in the scleractinian Acropom (17). Inter-generic sequence differ- ences in ITS-1 range from 27% to 79% in the Anthozoa ( 14, 1 5 ). Thus, if a difference of from 5% to 1 5% between ITS- 1 sequences indicates species-level divergence, the sequences reported here are evidence of at least nine species of Aure- lia: A. limbata, A. labiata, and A. aitrita, plus six cryptic species previously identified as A. aitrita (Fig. 2 A). Sequence variation in COI in Aurelia, which ranges from 13% (A. limbata v.v. Aurelia sp. 1 [Marina del Rey. CA]) to 24% (Aurelia sp. 3 [Tab Kukau Cove, Palau] vs. Aurelia sp. 2 [Cananeia]) also is comparable to inter-specific differ- ences in other metazoans. For example. 18% sequence difference in COI distinguishes congeneric species of cope- pods (18), 4% to 19% distinguishes species of the snapping shrimp A Ipheus (19), about 7.5% indicates species of deep- sea clam (Vesicomyidae; 20), and 1 1.5% to 30.8% charac- terizes congeneric species of the amphipod Gammarns (21). Thus, if 10% to 20% sequence difference is taken as a benchmark of distinct species, these COI data also support recognition of A. limbata, A. labiata, and seven sibling species of A. aitrita (Fig. 2B). Several other observations suggest that the nine clades of Aurelia distinguished by sequence data represent distinct species. For example, A. labiata and A. limbata are sympa- tric in Kachemak Bay, providing strong evidence that their molecular differences reflect reproductive isolation. Simi- larly, Palau cove Aurelia occasionally are advected into the lagoon, but have not. according to the molecular data, interbred with lagoon Aurelia for millions of years, again implying reproductive isolation. Reproduction by Aurelia in Palau coves, lagoon, and lakes is also ecologically, usually geographically, and often temporally isolated (pers. obs.), and brackish, marine lake Aurelia may be physiologically inhibited from encountering congeners in seawater (5). For similar reasons. Aurelia in the fjord-like Mljet "lake" are probably reproductively isolated from Aurelia that occur elsewhere in Europe, including the Black Sea and probably 94 M. N. DAWSON AND D. K. JACOBS Aureliasp 2 Cananeia t N = 4, 8 B Aureliasp.'} Manna del Rev. Long Beach. Newport Beach. San Diego. Japan N= 14 Aureliasp 5 Mljet N = 6. 10 8:99 Aurella labiata Kachemak Bay. Sooke Basin. Todd Inlel. Newport Ba\ . Tomales Bay . Montere> Bay. N= 13. 20 Boston Harbor. Narragansetl Ba\ N = 8, 12 Aurella aurlta Aurelia limbata Kachemak Ba\ N= 1.1 White Sea, Gullmar Fjord. Bjomsund Fjord. Black Sea Bosporus N= 15. 22 Aureliasp 6 PaJau Lagoon & Helen Reef N- 3,3 2.74" Aureliasp.'} Manna del Rev, Long Beach. Newport Beach. San Diego. Japan N = 8. 15 Aurelia limbata Kachemak Ba; N = I Monterey Bay . Tomales Bay . Newport Bay N = Aurelia labiata Kachemak Bay. Sooke Basin. Todd Inlet N= II Aureliasp 6 Palau Lagoon & Helen Reef N = 7 — 1 change Aureliasp 3 Palau coves N = 4, 7 Aureliasp.2 Cananeia N = 4 aurita White Sea Gullmar Fjord, Bjomsund Fjord. Black Sea, Bosporus N= 10 .9 ,00 /"ffasp.3 f Palau coves — 5 changes Aureliasp 4 45 100 Palau marine lakes N- II Aureliasp 5 Mljet N = 6 Aureliasp 4 Palau marine lakes N=8. IS Figure 2. DNA sequence analyses of ITS-1 and COI are consistent with previous allozyme studies and morphological descriptions of Aurelia (3, 9, 10, 13) that indicated European and northeastern Atlantic clades within A. aurita. northeastern Pacific (A. labiatu). polar (A. limbata), and Japanese or southern California lineages. DNA sequence analyses of ITS-1 and COI also indicate an additional six cryptic species of A. aurita. (A) Unrooted ITS-1 gene tree. One of 571 shortest trees (length = 168 steps: CI = 0.7738) recovered by unweighted maximum parsimony analysis (PAUP ver. 4.0b4a; 33) of sequences aligned in CLUSTALW (34; gap opening:extension penalty = 5: 1 1 and amended by eye; gapped positions were excluded from the parsimony analysis, therefore emphasizing the better aligned portions of sequences (i.e.. 164 characters of which 83 were parsimony informative). Major branches present in the strict consensus tree are annotated with branch-length and bootstrap support (if >50 0.05) between Ea X EC and EC X Ea. Measurements in both of the reciprocal hybrids were closer to those of Ea X Ea than to EC X EC. These measurements mirror the different growth rates found for these animals by Rahman et al. (2000). The aboral body coloration differed between conspecifics and hybrids (Fig. 2. upper). In Ea X Ea specimens, test 100 M. A. RAHMAN ET AL. Table 2 Weight, test si:t, and spine length comparison <>j Echmometra s/>. ,4 (Ea X Ea), Echinometra sp. C I EC X EC), and their reciprocal hybrids 1 year after metamorphosis Character Ea x Ea Ea x EC EC X Ea EC X EC Wet weight (gm) Test length (mm) Test width (mm) Test height (mm) Spine length (mm) 14.7 29.4 27.3 14.2 26.8 ± 0.7" (14.6-14.8) ±0.8;l* (28.2-3 1.0) ±0.8-' (26.1-29.0) ±0.4J (13.6-15.0) ± 1.0" (24.4-28.2) 13.6 28.4 26.3 13.5 23.4 ± 0.2h(13.6-13.7) ± 0.9h( 27.0-30.0) ± 0.9b( 25.0-28.1) ± 0.4h( 12.8-13.9) ±0.7h (22.3-24.9) 13.9 28.7 26.5 13.6 24.3 ± 0.5h( 13.8-14.0) ± 0.8" (27.5-30.5) ± 0.8" (25.3-28.5) ± 0.3" (13. 1-14. 2) ± 0.6" (23.4-25.5) 9.2 24.2 22.2 12.0 20.7 ± 0.5C (9.1-9.3) ± l.lc (23. 0-26.0) ± 1.0° (2 1.0-23.9) ±0.5' (11. 2- 13.0) ±0.7° (19.3-2 1.9) * Thirty adult specimens were measured for each treatment; mean ± SD. ranges in parentheses. Wet weight data from Rahman et at. (2000). Numbers in the same row having the same superscripts are not significantly different (P > 0.05). color was dark brownish, and each spine had a white tip and a translucent white ring at its base. EC X EC specimens were uniformly greenish, and each spine was uniformly colored except for a faded basal white ring. Ea X EC hybrids were more similar to Ea X Ea conspecifics. having dark brownish tests and white-tipped spines with a translucent white, basal ring. On the other hand. EC X Ea hybrids were more similar to EC X EC conspecirics; the tests were uniformly deep greenish, and the spines had barely detectable white tips and basal white rings. Orally (Fig. 2. lower), Ea X Ea conspecifics had white- tipped spines around the mouth and brownish dark tests, whereas EC X EC conspecifics had yellowish green spines around the mouth and a greenish test color. Ea X EC hybrids were more similar to Ea X Ea, whereas EC X Ea hybrids were more similar to EC X EC; that is, hybrid coloration was most similar to maternal coloration. Tubefoot spicules in Ea X Ea were always bihamate (C-shaped), whereas those in EC X EC were always triradi- ate (Fig. 3). Tubefoot spicules of Ea X EC hybrids were bihamate (58%), bihamate-like (17%). triradiate-bihamate (20%), and triradiate (5%), whereas those of EC X Ea were bihamate (32%), bihamate-like (12%). triradiate-bihamate (44%), and triradiate (13%). Therefore, the tubefoot spicule morphologies of the hybrids were intermediate and tended toward maternal affinities. There was little intraspecific variation in gonadal spicule morphology (Fig. 4). The spicules in Ea gonads were almost all spindle-shaped (98%; other spicules seen: bihamate, <1%; irregular, 1%; no triradiates were found). In contrast, those in EC gonads were nearly all triradiate (94%; other spicules seen: spindle, 5%; bihamate, 1%). Gonads in both Ea X EC and EC X Ea had high proportions of spindle- shaped spicules (62% and 46%. respectively), with spindle- like (16% and 11%), triradiate-spindle (13% and 25%), triradiate (7% and 15%), bihamate (1% in both), and irreg- ular (1%- in both) spicules in smaller proportions. Gamete sizes of conspecific and hybrid individuals were different. Egg diameters of Ea X Ea were smallest among the four treatments, whereas the EC X EC eggs were largest (Table 3). Hybrids contained intermediate-sized eggs that were significantly different from the eggs of either of the conspecifics. The sizes of sperm heads were also smallest in Ea X Ea, and were significantly different among all the crosses, including between the hybrids. Existence of natural hvbrids The characteristics of reproductively mature hybrids are summarized in Table 4. Two hundred individuals with col- oration more-or-less intermediate between the two species were collected from the field. However, detailed compari- sons of spicule shapes and gamete sizes revealed that none actually had character combinations common to the exper- imentally obtained hybrids; that is, all could be assigned to either Ea or EC on the basis of both spicule shapes and gamete sizes. Similarly, Asian and Uehara (1997) did not find any natural hybrids between Ea and Ed in the field, even though they were able to rear hybrids of these species in the laboratory. Backcrosses There was no indication that the hybrids were any less fertile than either of their conspecific half-siblings (Table 5). Eggs from both Ea X EC and EC X Ea yielded higher percentages of fertilization with Ea X Ea sperm (93% and 98%), than with EC X EC sperm (82.2% and 89.2%), similar to the findings with the parental crosses in which Ea sperm more readily fertilized EC eggs than EC sperm fertilized Ea eggs (Fig. 1 ). Moreover, backcrosses by sperm from males of Ea X EC and EC X Ea yielded higher percentages of fertilization with EC X EC ova (99.3% and 99.7%) than with Ea X Ea ova (84% and 86.5%); Ea ova appear to be more discriminating than EC ova. The higher fertilization rates between the same types of hybrids versus the different types of hybrids (99.5% and 99% versus 95% and 86%) indicate the presence of a complex sorting of gamete compatibility genes. HYBRIDIZATION BETWEEN SEA URCHINS 101 Figure 2. Ahoral (upper) and oral (lower) color patterns of Echinninciia sp. A and E. sp. C, and their reciprocal hybrids, I year after metamorphosis; maternal species named first. (A) Ea X Ea; (B) EC X EC; (C) Ea X EC; (D) EC X Ea. 102 M. A. RAHMAN ET AL bl V bl bl Figure 3. Morphology of tubefoot spicules of Echinomelra sp. A and Echinometra sp. C and their reciprocal hybrids, 1 year after metamorpho- sis; maternal species named first. (A) Ea X Ea; (B) EC X EC; (C) Ea X EC; (D) EC x Ea. b. bihamute; bl, bihamate-like; I, triradiate; tb, triradiate- hihamate. Scale bar, 50 /LUII. Discussion The crosses between Ea-Ec showed a distinct asymmetry in fertilization success, as reported previously by Uehara et al. (1990) and Rahman et al. (2000). Heterogamic fertiliza- tion rate was high when eggs of EC were mixed with sperm of Ea, but much lower when eggs of Ea were mixed with sperm of EC. The reduction in fertilization with the Ea egg X EC sperm crosses indicates the presence of a protein- binding system for gamete recognition, as reported by Metz ct al. (1994). and Metz and Palumbi (1996). Such a system might eventually lead to gamete incompatibility and repro- ductive isolation, and thus provide a mechanism for main- taining species integrity, as proposed by Metz et al. ( 1994), Vacquier et al. (1995), Metz and Palumbi (1996). and Palumbi ( 1998). However, with the high fertilization of EC eggs by Ea sperm, prezygotic isolation by gamete incom- patibility hardly appears to be present between these two species, or among most of the other species of Echinometra on Okinawa, most of which, like the two sympatric species of Caribbean Echinometra, also show high asymmetry in fertilization among hybrid crosses (Uehara et al., 1990; Lessios and Cunningham, 1990). Consequently, it seems unlikely that gamete incompatibility, by itself, provides a mechanism for reproductive isolation in Echinometra. Re- productive isolation and speciation apparently occurred be- fore gametic incompatibility closed off the possibility of gene flow, and as in two sympatic species of asteroids in the genus Patiriella that do not have gamete incompatibility (Byrne and Anderson, 1994), other mechanisms need to be found to explain how species integrity is maintained in these closely related, interfertile species. Furthermore, the Ea-Ec crosses had higher fertilization rates in the F, backcrosses than Asian and Uehara (1997) found for Ea-Ed F, backcrosses. The higher fertilization rate may be due to the higher genetic similarity between Ea and EC than between Ea and Ed, as proposed by Matsuoka and Hatanaka ( 1991 ). If this is true, it provides evidence for the eventual evolution of complete gamete incompatibility in these species. Supporting the idea of a transition of gamete compatibility to incompatibility after speciation, Uehara et al. ( 1990) reported that fertilization between EC and Ed was high in both directions, while crosses between Ea and Eb were completely infertile; other combinations among the four species were asymmetrical. Whether these differences reflect evolutionary distances among the species, or indicate that other factors — such as some sort of reinforcement se- lection coupled to microhabitat proximity — are involved, or D Figure 4. Morphology of gonadal spicules of Echinometra sp. A and Echinometra sp. C and their reciprocal hybrids, 1 year after metamorpho- sis; maternal species named lirst. (A) Ea x Ea; (B) EC x EC; (C) Ea X EC; (D) EC x Ea. b, bihamate, ir, irregular, s, spindle; si, spindle-like; st, spmdle-tnradiate; t, triradiate. Scale bar, KM) /^m. HYBRIDIZATION BETWEEN SEA URCHINS Table 3 Gamete sizes of Echinometra sp. A {Ea X Ea). Echinometra sp. C (Ec X EC), and their reciprocal hybrids 1 year after metamorphosis Measurement Ea X Ea Ea x EC EC X Ea EC x EC Egg diameter Sperm head length 66.9 4.1 ± 1.0 ± 0.5 a* (65.0-67.5) "(3.1-5.0) 71.0 5.7 ± 1.2" ± 0.7h (70.0-73.8) (4.3-6.8) 70.2 5.2 ± I.Ih (67.5-72.5) ± 0.8C (4.3-6.8) 72.6 ± 6.5 ± 1.3C 0.8d (71.3-75.0) (5.0-7.4) * Twenty individuals were examined from each cross with 25 eggs and 25 sperm from each individual; mean ± SD in /iun. ranges in parentheses. Figures in the same row having the same superscripts are not significantly different (P < 0.05). suggest that gametic incompatibility in one or the other species "simply happened by accident" (Lessios and Cun- ningham. 1990, p. 938) remains to be determined. The successful culturing of hybrids to fertile adults in laboratory conditions suggests that postzygotic isolating mechanisms, such as hybrid inviability or sterility, are not involved in maintaining species integrity. However. Rah- man et al. (2000) found that Ea X EC hybrids had lower larval survival, metamorphosis success, and juvenile sur- vival than either EC X Ea hybrids or conspecifk half- siblings. Consequently, not only gamete incompatibility but postzygotic fitness parameters act against Ea X EC hybrids. Nevertheless, the surviving Ea X EC hybrids grow at the same rate as conspecifics (Rahman et ai. 2000), and as shown in the present study, are as fertile as the conspecifics in backcrosses. Moreover, EC X Ea hybrids are as viable and fertile as the conspecifics. indicating that there are neither gametic nor postzygotic blocks to introgression for at least half of the possible hybrid events. Table 4 Characterization of hybrids produced experimentally through cross fertilisation benveen Echinometra sp. A (Ea) and Echinometra sp. C (Ec) Distinctive features Ea -• EC EC X Ea Body color a. Oral Ea-like Ec-like b. Aboral Ea-hke Ec-like Test sizes a. Length Intermediate Intermediate or Ea-like b. Width Intermediate Intermediate or Ea-like c. Height Intermediate Intermediate or Ea-like Spines a. Length Intermediate Intermediate or Ea-like b. Color Ea-like with white Ec-like or intermediate tip or intermediate with faint white spine tip c. Milled ring color Ea-like Ea-like Spicules a. Tubefoot Intermediate and Intermediate and Ec-like Ea-hke b. Gonad Intermediate and Intermediate and Ea-like Ea-like Gametes a. Sperm sizes/form Intermediate Intermediate h. Egg sizes Intermediate Intermediate Besides gamete incompatibility, potential prezygotic mechanisms that may occur between these two species include ecological separation and asynchronous reproduc- tive cycles (Mayr, 1970; Palumbi, 1994). The two species of sea urchins in this study, Ea and EC. live relatively close to each other but occupy different microhabitats; Ea inhabits the calmer moat and tidepools of the reef-flat, generally below the mean low water level, whereas EC inhabits bur- rows or crevices in the wave-swept intertidal of the reef margin (Table 1 ). These differences in microhabitats may be enough to prevent most chances of cross fertilization. In broadcast spawning invertebrates such as echinoids, for example, fertilization success drops dramatically with dis- tance between spawning individuals (Pennington. 1985; Levitan, 1998a, b). However, individuals of Ea are occa- sionally found near and within the reef margin where EC predominates, and it remains unclear, whether microhabitat separation, by itself, is sufficient to prevent introgression between the two species, or at least through EC ova and Ea sperm where there is little or no gamete incompatibility, hybrid viability, or hybrid fertility. In addition to the problem of being separated enough in adjacent microhabitat to prevent gamete mixing after spawning, there is the problem of maintaining microhabitat differentiation that could assure reproductive isolation. This problem is especially acute for species with long-lived, widely dispersing larvae, such as species of Echinometra. The different larvae would have to have exquisite settling cues that assured that they would be established in their exact microhabitats or suffer highly selective postsettlement mortality outside their particular microhabitats. There is little evidence of highly selective habitat selection 'n sea urchin larvae (Pearse and Cameron. 1991). Indeed. Rahman and Uehara (2001) found no discrimination in settling pref- erences by competent larvae of the four species of Echi- nometra on Okinawa; all showed settlement rates of 86%- 89% on coralline red algae, and much lower rates on other algae. The mechanism resulting in and maintaining micro- habitat differentiation among broadcast spawning species such as those of Echinometra in the tropical west Pacific remain unknown and unexplored. Echinoid sperm are active for only a short time after spawning, much less than an hour in temperate species 104 M. A. RAHMAN ET AL Table 5 Percentage of eggs fertilized in backcrosses among laboratory-reared Ft generation of conspecifics and hybrids of Echinometra sp. A (Ea) and Echinometra sp. C (Ec) Sperm from Ea x Ea Eggs from Ea X EC EC x Ea EC X EC Ea x Ea Ea x EC EC x Ea EC X EC 99.7 ± 0.5 (99-100) 84.0 ± 1.6(82-86) 86.5 ± 1.9(84-88) 30.7 ± 2.0(28-33) 93.0 ± 1.4(91-95) 99.0 ± 0.9(98-100) 86.0 ± 1.8(83-88) 82.2 ± 1.8(80-84) 98.0 ± 1.4(96-100) 95.0 ± 1.4(93-97) 99.5 ± 0.8 (98-100) 89.2 ± 2.1 (87-93) 86.8 ± 1.7(84-88) 99.3 ± 0.8(98-100) 99.7 ± 0.5(99-100) 99.8 ± 0.4 (99-100) Each value represents six replicate crosses with gametes from different individuals in each replicate; mean ± SD, ranges in parentheses. (Hinegardner, 1975; Levitan et ai, 1991), and probably even less in tropical species. Asynchrony in spawning, therefore, would ensure that these two species could exist in sympatry as separate species whether or not their gametes are capable of fertilization (Lessios, 1984). However, Ea and EC mature at the same time and their spawning periods overlap extensively (Arakaki and Uehara, 1991: Table 1). Consequently, seasonal separation of spawning is not a mechanism of reproductive isolation. On the other hand, gametes may be released at different times of the day or specific pheromonal spawning cues may differ between the species, and these factors may prevent near simultaneous spawning that could lead to hybridization. Natural spawning in the field has not yet been observed. The sperm concentration used in this study was probably much higher than would be encountered under usual natural conditions, especially considering the dilution that would occur if individuals were separated by 1 m or more (Pen- nington, 1985; Levitan, 1998b). However, Uehara et al. ( 1990) provide data showing that fertilization rates of eggs of both Ea and EC mixed with sperm from Ea were similar over a wide range of sperm concentrations, whereas the sperm of EC had very low fertilization success with Ea eggs, even at high concentrations, up to 10~~. Consequently, it is equivocal whether inappropriate sperm concentrations were used in this study, and Ea sperm, at least, appear as likely to fertilize EC eggs as Ea eggs over concentrations found in the field. In addition, the eggs in this study were exposed to sperm at relatively high concentrations for 1 .5 h. much longer than they would likely be so exposed in the field. Levitan et al. ( 1991 ) found that fertilization increases in sea urchins with time of exposure of eggs to sperm over a range of 10-60 s. It is possible, therefore, that fertilization rates would be lower in the EC x Ea crosses if the gametes had been kept together for shorter times. Preliminary experiments indicate that fertilization reaches the asymptote in less than 2 min in conspecific crosses of Ea and EC, within 4 min in EC X Ea crosses, and by about 6 min in Ea X EC crosses (Rahman, unpubl. data). Therefore, more extended exposure times. like those used in the study, would not increase the fertili- zation rates. However, if gametes are spawned in the field at the same time by equal numbers of adjacent individuals of both sexes of both species, most conspecific gametes would probably be fertilized before much hybridization could oc- cur. On the other hand, where the populations intermingle it seems as likely that a spawning EC female could be closer to a spawning Ea male than to spawning EC males, so that hybrids could be expected to occur. To date there have been no experiments with mixed populations of gametes. Palumbi ( 1998) showed that at low concentrations where less than near 100% fertilization is achieved, sperm from different males of the same species of sea urchins fertilized different proportions of eggs from the same female, and eggs from different females were fertil- ized in different proportions by sperm from the same male. These differences were further related to differences in the sequence of the bindin alleles. This opens the possibility of some sort of interlocus antagonistic coevolution between gametes, as was proposed by Rice (1998) and which was demonstrated to be occurring in multiple-mating, internally fertilizing species such as crickets (Howard et ai, 1998). How such sperm competition could be achieved in broad- cast spawning, externally fertilizing species such as sea urchins, if it occurs at all, remains to be determined. How- ever, if it does occur, and if conspecific sperm are always at an advantage when mixed at low concentrations with sperm from another species because of more compatible gametes, a mechanism for maintaining species integrity in sympatric species may be present. Some characteristics of the hybrid progeny found in the present study were mainly maternal. Color patterns of both hybrids, for instance, tended to be maternally inherited. Maternal inheritance of color pattern also was observed in hybrids between Strongylocentrotus nitdus and S. interme- diiis (Osanai 1974). On the other hand, other characteristics, such as spine length and spicule shapes, especially those from the gonads, showed intermediate features, and can be used to distinguish the hybrids from either parent. Such distinct hybrid phenotypes are important for finding hybrids HYBRIDIZATION BETWEEN SEA URCHINS ' in the field (e.g., Hagstrom and L0nning, 1961; Menge, 1986; Palumbi and Metz. 1991; Lessios and Pearse. 1996). We searched for distinctive intermediate phenotypes in the field, and although suggestive color morphs were found, examination of the spicules revealed no hybrids. Similarly, although hybrids with distinguishing characteristics can be formed in the laboratory between Ea and Ed, Asian and Uehara ( 1997) were unable to find such hybrids in the field. Moreover, although two species of the tropical long-spined sea urchins, Diadama savignyi and D. setosum, readily hybridize in the laboratory (Uehara et al., 1990) and often occur in mixed populations in the field (Pearse, 1998), genetic analyses, using allozyme analyses, showed that hy- brids only occur rarely in the field and there is limited introgression (Lessios and Pearse, 1996). Genetic analyses using allozyme or DNA markers should now be done with sympatric species of Echinometra to determine whether some introgression is occurring that has not been detectable by morphological characters. The adult hybrids of these two species were completely fertile, indicating that genetic differences between them are not large enough to cause developmental or gamete incom- patibility, in spite of their morphological differences and microhabitat segregation. The high fertilization rates in backcrossing further suggests that they are genetically very close to each other. Similar results were obtained from crosses that produced fertile hybrids between Ea and the putatively more distantly related Ed (Asian and Uehara, 1997). There is no reason to believe that fertile hybrids could not be produced in crosses between most other spe- cies of Echinometra on Okinawa, including between EC and Ed, which show high fertilization with each other in both directions (Uehara et ai, 1990). Only Ea and Eb (=Echi- nometra mathaei) showed very low fertilization with each other in either direction, and may be isolated by gamete incompatibility (Uehara et ai. 1990; Metz et al.. 1994). These latter two species co-occur in pools and channels below mean low water on the reef flats of Okinawa, and they may have the greatest opportunity for gamete mixing in the field. Consequently, for these two species, gamete in- compatibility may be necessary to maintain species integ- rity. Reproductive isolation of the other species, however, probably depends on additional prezygotic isolating mech- anisms, such as microhabitat separation. New biological species are formed when populations become reproductively isolated from other, previously con- specific populations. How and when reproductive isolation develops has yet to be established for broadcast spawning marine invertebrates such as sea urchins (Lessios, 1984; Knowlton. 1993; Palumbi, 1994). Although not isolated by full gamete incompatibility, or by hybrid inviability or in- fertility, Ea and EC are both morphologically and geneti- cally distinct, and hybridization between them appears to be either very low or nonexistent — that is, they appear to be effectively reproductively isolated. The morphological dis- tinctness and apparent genetic integrity maintained between them warrants their recognition as distinct biological spe- cies despite their ability to produce viable and fertile hy- brids in the laboratory. They should be given taxonomic descriptions and appropriate species names that are distinct from Echinometra mathaei, sensu stricto (=Eb; Arakaki et al. 1998) and Ed ( = the Okinawan component of the Echi- nometra oblonga species complex; Arakaki and Uehara, 1999). Acknowledgments We are greatly indebted to L. M. Asian. H. A. Lessios, H. Ota, and two anonymous reviewers for their critical and valuable suggestions on this manuscript. We also extend our appreciation and grateful thanks to the Director and staff of the Sesoko Station of the Tropical Biosphere Research Center, University of the Ryukyus, who provided space and facilities for culturing sea urchins; and to Motobu Fisheries Research Center for providing samples of Chaetoceros gra- cilis and instructions for culturing it. The first author is also grateful to the Rotary Yoneyama Memorial Foundation, Japan, for scholarship grants, and especially to Dr. Tomio Iha. Naha Rotary Club, Okinawa, for his continuous coop- eration and encouragement during the period of the grants. Literature Cited Amy, R. L. 1983. Gamete sizes and developmental time tables of five tropical sea urchins. Bull. Mar. Sci. 33: 173-176. Arakaki, Y. 1989. A comparative ecological and reproductive study on the four types of sea urchin Echinometra mathaei (Blainville) on Okinawan reef flats. Master's thesis. University of the Ryukyus. Japan. Arakakj, Y.. and T. Uehara. 1991. Physiological adaptation and repro- duction of the four types of Echinometra mathaei (Blainville). Pp. 105-1 1 1 in Biology ofEchinode nnata, T. Yanagisawa, I. Yasumasu. C. Oguro. N. Suzuki, and T. Motokawa, eds. A. A. Balkema, Rotterdam. Arakaki, Y., and T. Uehara. 1999. Morphological comparison of black Echinometra individuals among those in the Indo-west Pacific. Zoo/. Sci. Japan 16: 551-558. Arakaki, Y., T. Uehara, and I. Fagoone. 1998. Comparative studies of the genus Echinometra from Okinawa and Mauritius. Zoo/. Sci. (Tokyo) 15: 159-168. Asian, L. M., and T. Uehara. 1997. Hybridization and F, backcrosses between two closely related tropical species of sea urchins (genus Echinometra) in Okinawa. Invertebr. Keprod. Div. 31: 319-324. Bartlett, M. S. 1937. Some examples of statistical methods of research in agriculture and applied biology. J. Roy. >,/<ana implexa (NMNH #4462) Filograna sp. (NMNH #79634) Filograna sp. (NMNH #79633) Filograna sp. (NMNH #81507) Filograna sp. (NMNH #98500) Siilimtcimi timphidenlata (author's collection) Salmacina dysteri (NMNH #67631) Salmacina dysteri (author's collection) Siilmiicina incrustans (NMNH #50436) Salmacina tribranchiuta (author's collection) Atlantic Ocean: Martha's Vineyard. Massachusetts, USA, 55 m Atlantic Ocean: Tangier Harbor, Morocco. 5 m Indian Ocean: Mauritius, 18 m Pacific Ocean: South Pacific (43°48 S, 147°34 E). 146 m Pacific Ocean: Galapagos Islands, 0-15 m Atlantic Ocean: Fort Pierce, Florida. USA, 0-1 m Atlantic Ocean: Charleston, South Carolina, USA, 32 m Pacific Ocean: Pearl Harbor, Hawai'i, USA, 0-1 m Atlantic Ocean: La Parguera. Puerto Rico. 1 m Pacific Ocean: Santa Cruz. California, USA, 0-1 m Tubes branch distal to peristomes 1893 Tubes branch distal to peristomes 1982 Tubes branch distal to peristomes 1974 Escape hatches present; tubes branch 1968 distal to peristomes Tubes branch distal to peristomes 1966 Escape hatches present: tubes brunch 2000 distal to peristomes Tubes branch distal to peristomes 1978 Escape hatches present; tubes branch 2000 distal to peristomes Tubes branch distal to peristomes 1963 Escape hatches present; tubes branch 2000 distal to peristomes * Names listed are those used in the collections of the U.S. National Museum of Natural History (NMNH), or, for specimens in the author's collection, those used in regional guides (California: Blake, 1975; Hawai'i: Bailey-Brock and Hartman, 1987). Both generic and specific identifications should be viewed with skepticism. See Materials and Methods for additional comments on systematics. of branching tubes. It should be noted that aggregations of Filograna and Salmacina spp. are not necessarily composed only of the genetically identical offspring of a single founder individual. Recruitment of planktonic larvae may contribute substantially to their growth (Nishi et ol, 1996; Nishi and Nishihira, 1997). The branching tubes character- istic of aggregations of Filograna and Salmacina spp., however, are undoubtedly the consequence of repeated bouts of asexual reproduction. Many species of serpulids that do not reproduce asexually also form aggregations (ten Hove and van den Hurk, 1993), but these do not contain branching tubes. Mechanisms and timing of escape hatch formation in Salmacina amphidentata Though the escape hatches formed by Salmacina amphi- dentata are among the most complex tube structures known from the serpulids, the actual mechanism of their formation does not seem to have required great modifications of the routine processes of tube growth. The ability to carefully control where calcareous tube material is deposited around the margin of the tube, which permits S. amphidentata to build escape hatches, is widespread in serpulids. Many species, for example, deposit more material at some loca- tions around the tube margin than at others, resulting in the formation of sculptural elements like keels or spines (e.g.. Pomatoceros triqiteter: Hedley, 1958; Filogranula gracilis (as Omphalopoma): Zibrowius, 1968; Spiraserpula spp.: Pillai and ten Hove, 1994). A few species form even more complex structures such as brood chambers for developing embryos (Ben-Eliahu and ten Hove, 1989; Nishi, 1993). Adults of 5. amphidentata avoid depositing any calcareous tube material at all in the narrow gaps between the calcar- eous disc and the rest of the tube, while depositing it elsewhere along the tube margin. Such precise control over the location of deposition may be mediated by two pro- cesses. First, the slurry of calcareous crystals in a fluid organic matrix may be deposited directly at the correct sites on the tube margin by the pair of calcium-secreting glands on the ventral lobe of the collar, with the adult rotating in the tube to position the gland openings correctly. However, in each of the two episodes of escape hatch formation I observed in detail by time-lapse video, the adult worm rotated in its tube infrequently. Further, for most of the time spent building hatches, the openings of calcium-secreting glands were approximately adjacent to the gaps between the calcareous disc and the rest of the tube, exactly where no calcareous tube material was deposited. A more plausible possibility is that muscular movements of the collar mold the slurry of tube material into the correct positions on the tube margin. I did not observe such movements, but would expect them to be subtle and difficult to see at the low TUBE GROWTH IN CLONAL SERPULIDS magnifications used in this study. The slurry of tube mate- rial is not molded onto the tube margin by the action of cilia on the collar, as its relevant surfaces lack cilia (not shown). In Salmacina amphidentata, the calcareous discs of the escape hatches were supported in place by a thin organic membrane (Fig. 3E). This organic layer lined the inner surface of the entire tube. Though it is rarely mentioned in accounts of tube structure or growth, a similar "coating membrane" has been found lining the tubes of at least five other species of serpulids (Muzii, 1968; Fernet, pers. obs.), and is probably an integral component of the tube in all species of serpulids. This is presumably the material that the chaetae of serpulids interact with during routine movement up and down the tube, and also during anchoring when a predator attempts to pull the worm from the tube (e.g., Merz and Woodin, 2000). Two observations suggest that the coating membrane is deposited prior to or at the same time as the calcareous tube material. First, during the process of escape hatch formation, Filograna and Salmacina spp. con- struct a calcareous disc that is discontinuous with the rest of the calcareous tube. As noted earlier, however, it appears that the growth of calcareous tubes in all serpulids requires a pre-existing tube margin onto which new calcareous ma- terial can be added. During the deposition of calcareous discs, it seems likely that such a pre-existing tube margin is present in the form of a coating membrane. Second, the calcareous disc is supported in place by the coating mem- brane; it seems likely that this is true throughout its con- struction. That the coating membrane has been found in all serpulids that have been carefully examined (Muzii, 1968; Fernet, pers. obs.) and that it appears to be deposited prior to or simultaneously with the calcareous portion of the tube suggest that it plays an important and rarely considered role in tube formation. It may also be important in considerations of the evolution of the calcareous tube, a defining feature of the serpulids (Rouse, 2000). The close temporal coordination of events in asexual reproduction and tube growth in Salmacina amphidentata suggests the possibility of signaling pathways between buds and parents. Only after a developing bud had reached a certain stage (with unbranched radioles, 0-3 thoracic seg- ments, and rudimentary thoracic membranes) did the parent worm begin to form an escape hatch. Adults never built escape hatches unless they were in the midst of a budding cycle. What stimulates the parent worm to alter its pattern of tube deposition at this particular stage in bud development9 It is possible that bud and escape hatch formation are part of a single developmental pathway in adult worms, such that once asexual reproduction is initiated, escape hatch forma- tion several days later is inevitable. Alternatively, once a bud reaches a certain stage, it may somehow signal to the adult that it is an appropriate time to form an escape hatch. For example, buds at the stage shown in Figure 2 may already be producing their own hormonal products, perhaps involved in controlling such processes as the transformation of former adult abdominal segments into thoracic segments for the bud. Adult worms, which are still attached to the buds at this stage, might use such signals as a cue to alter their patterns of tube deposition. Endocrine substances pro- duced in the prostomium (or in other parts of the anterior end) are well known as signals controlling the timing of asexual reproduction and changes in chaetal morphology in other polychaetes (e.g., Schroeder, 1967; Franke and Pfan- nenstiel. 1984). A test of this hypothesis might involve implanting the prostomium of a developing bud into the body of a non-budding adult, and examining the effects on tube growth. Once buds have detached from the parent, they are free to move forward and backward in the parental tube. Within a few days, however, they must gain access to the outside via the escape hatch. I did not determine which worm actually dislodged the calcareous disc from the escape hatch, though it seems likely that the bud does this rather than the parent, which has usually grown well beyond the escape hatch by this time. Buds may identify the site of the escape hatch by simply moving around and testing the tube walls until they find a weak spot. Dislodging the calcareous disc presumably does not require much force, as it is held in place only by a very thin coating membrane. I did not measure rates of asexual reproduction, or how these might be related to rates of tube growth. In laboratory cultures, both the frequency of asexual reproduction and the rate of tube growth appeared to vary within and among clones, resulting in great variation in the distance between side branches from parental tubes. Many factors, both en- vironmental (food levels, temperature, water flow) and ge- netic, might affect these parameters, as has been described in other clonal organisms (e.g., Sebens, 1980; Keen and Gong, 1989). Understanding the determinants of rates of asexual reproduction and tube growth in Filograna and Salmacina spp. is of particular interest since these parame- ters should strongly affect the eventual form of clonal aggregations. In nature, aggregations are quite variable in form. Distribution and evolution of escape hatches in asexually reproducing serpulids Some Filograna and Salmacina spp. from the Atlantic, Indian, and Pacific oceans form escape hatches for their clonal offspring (Table 1 ). In three lots of relatively recently preserved material, escape hatches identical in form to those described above in Salmacina amphidentata were present. In the remaining six lots of material I examined, which had been in preservative for 18-107 years, branching pat- terns of tubes were identical to those described above, but no intact escape hatches were found. The absence of intact escape hatches in this older material is not too surprising, 116 B. FERNET even if escape hatches were present at the time of collection. The tubes of these species are delicate and are often dam- aged during collection and handling. It is also likely that they are gradually dissolved during storage, unless preser- vatives are well buffered. Escape hatches in particular are prone to damage or dislodgement, as the organic membrane that supports calcareous discs in the tube is extremely thin (Fig. 3E). These comparative data suggest that the forma- tion of escape hatches for clonal offspring is widespread in members of Filograna and Salmacina. Asexual reproduction is also known from several other genera of serpulids (Filogranella, Filogranula, Josepliella. Rhoilopsis, and Spiraserpula). Members of these genera that reproduce asexually do so by posterior budding, as in Filograna and Salmacina spp. (ten Hove, 1979; Ben-Efiahu and ten Hove, 1989; Nishi, 1992; Pillai and ten Hove, 1994). Their clonal offspring should also face the dual problems of escaping from the parental tubes and starting to build their own tubes. How they solve these problems is not known. Aggregations of branching tubes similar to those of Filograna and Salmacina spp., suggesting some mechanism of reaching the exterior of the parental tube but not dispers- ing from it, are reported in members of three of these genera (Filogranella, Josephella, and Spiraserpula: ten Hove, pers. comm.; George. 1974; Pillai and ten Hove, 1994). A few members of a closely related family of tube- dwelling polychaetes, the Sabellidae, also reproduce asex- ually by posterior budding (Knight-Jones and Bowden. 1984). Sabellid tubes are typically constructed of secreted organic compounds and sediment particles, and are not calcified. How the clonal offspring of sabellids escape the parental tubes is not known. Adults of several species are apparently capable of boring holes in their own tubes (Fitzsimmons, 1965), and it is possible that buds are able to do this as well. Once buds have gained access to the exterior of parental tubes, they may be free to disperse far from their parents before building new tubes, as many sabellids are able to build tubes de novo throughout their lives (Fitzsim- mons, 1965; Lewis, 1968). In serpulids, asexual reproduction by posterior budding is only profitable if buds have some way of gaining access to the exterior of the parent's tube (such as escape hatches), but it is difficult to imagine how escape hatches might have evolved in the absence of asexual reproduction. One sce- nario for the evolution of both of these traits (which seem to be tightly correlated in extant species of Filograna and Salmacina, at least) involves the initial evolution of poste- rior budding. Tubes of serpulids — especially the very thin tubes characteristic of species that have small maximum body sizes (Nishi, 1993) — are easily damaged by chance biological or physical insults, and buds may have originally used such fortuitous openings to gain access to the exterior of parental tubes. Adult worms that secreted generally weaker skeletons, or skeletons with weak points, would have had more consistent success in fledging clonal off- spring, though this might have been balanced by greater vulnerability to predators. Selection might then eventually lead to adults that built extremely weak points — escape hatches — into their tubes to ensure that buds would find exits. This is obviously speculative. However, this scenario might be tested with comparative data on living clonal serpulids examined in the context of a well-supported phy- logeny. Neither sort of data is yet available. Such an anal- ysis might reveal persistent and informative variation in the mechanisms buds use to gain access to the exterior of parental tubes. Acknowledgments I thank the director and staff of the Smithsonian Marine Station at Fort Pierce for support during this study, espe- cially S. Reed for guidance in the field and J. Piraino for operating the electron microscope. K. Fauchald, E. Kupriy- anova, and especially H. ten Hove provided advice on taxonomic problems. K. Fauchald and L. Ward made mu- seum collections available, and K. del Carmen and B. Ned- ved (Hawaii) and K. Wasson (California) kindly collected worms for me. I thank R. Brodie, K. Fauchald, H. ten Hove, A. J. Kohn. E. Kupriyanova. E. Nishi, S. Woodin, and a reviewer for discussion or comments on the manuscript. Smithsonian Marine Station at Fort Pierce Contribution #510. Literature Cited Bailey-Brock. J. H., and O. Hartman. 1987. Polychaeta. Pp. 216-454 in Reef and Shore Fauna of Hawai'i. Section 2: Platyhelminthes through Phuronida and Section 3: Sipuncula through Annelida. D. M. Devaney and L. G. Eldredge, eds. Bishop Museum Press. Honolulu. Hawai'i. Ben-Eliahu, M. N., and H. A. ten Hove. 1989. Redescription of Rho- dopsis pusilla Bush, a little known but widely-distributed species of Serpulidae. Zoo/. Sir. 18: 381-395. Benham, \V. B. 1927. Polychaeta. British Antarctic ("Terra Nova") Expedition, 1910. Nat. Hist. Rep. Zoo/.. Br. Mils. Nat. Hist. 7: 47-182. Blake, J. A. 1975. Phylum Annelida: Class Polychaeta. Pp. 151-243 in Light's Manual: Intenidal Invertebrates of the Central California Coast. 3rd ed. R. I. Smith and J. T. Carlton. eds. University of California Press, Berkeley. Carpizo-Ituarte, E., and M. G. Hadfleld. 1998. Stimulation of meta- morphosis in the polychaete Hydroides elegans Haswell (Serpulidae). Biol. Bull. 194: 14-24. Clark. G. R.. Jr. 1976. Shell growth in the marine environment: ap- proaches to the problem of marginal calcification. Am. Zoo/. 16: 617- 626. Cresp, J. 1964. Etudes experimentales et histologiques sur la regenera- tion et le hourgeonnement chez les Serpulides Hydroides norvegica et Salmacina imriiMans. Hull. Biol. Fr. Belt;. 98: 3-152. Faouzi, H. 1931. Tube formation in Pomatoceros triaueter. J. Mar. Biol. Assoc. UK 17: 379-384. Faulkner, G. H. 1930. The anatomy and histology of bud-formation in the serpulid Filtigrana implexa. together with some cytological obser- TUBE GROWTH IN CLONAL SERPULIDS 1 vations on (he nuclei of the neoblasts. J. Linn. Soc. Lund. Zool. 37: 109-190. Fitzhugh, K. 1989. A systematic revision of the Sabellidae-Caobangi- dae-Sabellongidae complex (Annelida: Polychaeta). Bull. Am. Mus. Nat. Hist. 192: 1-104. Fit/.simmons, G. 1965. Feeding and tube-building in Sabellastarte mag- ni/ica. Bull. Mar. Sci. 15: 642-671. Franke, H.-D., and H.-D. Pfannenstiel. 198-4. Some aspects of endo- crine control of polychaete reproduction. Pp. 53-72 in Polychaete Reproduction: Progress in Comparative Reproductive Biology, A. Fischer and H.-D. Pfannenstiel, eds. Gustav Fischer Verlag, Stuttgart. Gee, J. M. 1963. On the taxonomy and distribution in South Wales of Filograna, Hydmides and Mercierelia. Ann. Mug. Nat. Hist. 13: 705- 715. George, J. D. 1974. The marine fauna of Lundy. Polychaeta (marine bristleworms). Reports of the Lundy Field Society 25: 33-48. Hanson, J. 1948. Formation and breakdown of serpulid tubes. Nature 161: 610-611. Hedley, R. H. 1956. Studies of serpulid tube formation. II. The calcium secreting glands in the peristomium of Spirorbis, Hydmides, and Ser- pula. Q. J. Microsc. Sci. 97: 421-427. Hedley, R. H. 1958. Tube formation by Pomatoceros triqueter. J. Mar. Bio/. Assoc. UK 37: 315-322. Hove, H. A. ten. 1979. Different causes of mass occurrence in serpulids. Pp. 281-298 in Biology and Systematics of Colonial Organisms, G. Larwood and B. R. Rosen, eds. Academic Press. London. Hove, H. A. ten, and P. van den Hurk. 1993. A review of Recent and fossil serpulid 'reefs': actuopalaentology and the 'Upper Malm' serpu- lid limestones in NW Germany. Geol. Mijnbomv 72: 23-67. Hove, H. A. ten, and P. S. Wolf. 1984. Serpulidae. Chapter 55, pp. 1-34 in Taxonomic Guide to the Polychaetes of the Northern Gulf of Mexico, J. M. Uebelacker and P. G. Johnson, eds. Barry A. Vittor and Associ- ates, Mobile. AL. Huxley, T. H. 1855. On a hermaphrodite and fissiparous species of tubicolar annelid Prontla dysteri. Edinburgh Phil. J. New Series 1: 113-129. Jones, M. L. 1962. On some polychaetous annelids from Jamaica, the West Indies. Bull. Am. Mus. Nat. Hist. 124: 169-212. Keen, S. L., and A. J. Gong. 1989. Genotype and feeding frequency affect clone formation in a marine cnidarian (Atireliu aurita Lamarck 1816). Fund. Ecol 3: 735-745. Knight-Jones, P., and N. Bowden. 1984. Incubation and scissiparity in Sabellidae. J. Mar. Biol. Assoc. UK 64: 809-818. Lewis, D. B. 1968. Feeding and tube-building in the Fabriciinae. Pmc. Linn. Soc. Loud. 179: 37-49. Malaquin, A. 1895. La formation du schizozoite dans la scissiparite chez les Filigranes et les Salmacynes. C. R. Hebd. Seances Acad. Sci. 121: 953-955. Merz, R. A., and S. A. Woodin. 2000. Hooked setae: tests of the anchor hypothesis. Invertebr. Biol. 119: 67-82. Mn/ii, E. O. 1968. A coating membrane in the calcareous tubes ol serpulids. Pnhbi Sla:. Zool. Napo/i 36: 135-137. Neff, J. M. 1971. Ultrastructural studies of the secretion of calcium carbonate by the serpulid polychaete worm, Pomatoceros caerulus. Z. Zellforsch. Mikrosk. Anal. 120: 160-186. Nishi, E. 1992. Notes on the reproductive biology of some serpulid polychaetes at Sesoko Island. Okinawa, with brief accounts of setal morphology ot three species ol Salmacina and Filograna implexa. Mar. Foul. 10: 11-16. Nishi, E. 1993. On the origin of brooding characteristics in spirorbids with the phylogeny of sabellids and serpulids. Pmc. Jpn. Soc. Svsl. Zool. 49: 6-12. Nishi, E., and M. Nishihira. 1994. Colony formation via sexual and asexual reproduction in Salmacina dysteri. Zool. Sci. 11: 589-595. Nishi, E., and M. Nishihira. 1997. Multi-clonal pseudo-colony forma- tion in the calcareous tube worm Salmacina dysteri. Nat. Hist. Res. 4: 93-100. Nishi, E., and T. Vamasu. 1992. Brooding and development of a ser- pulid tube worm Salmucinu dysteri. Bull. Coll. Sci.. Univ. R\uk\n\ 54: 107-121. Nishi, E., T. Vamasu, and T. Kikuchi. 1996. Intra-colomal allozyme variation in the serpulid polychaete Salmacina dysteri. Publ. Amakusa Mar. Biol. Lab., Kyushu University 12: 55-62. Nott, J. A., and K. R. Parkes. 1975. Calcium accumulation and secre- tion in the serpulid polychaete Spirorbis spirorbis at settlement. J. Mar. Biol. Assoc. UK 55: 911-923. Pillai, T. G., and H. A. ten Hove. 1994. On Recent species of Spiraser- pula Rcgenhardt, 1961, a serpulid polychaete genus hitherto known only from Cretaceous and Tertiary fossils. Bull. Nat. Hist. Mus. Land. 60: 39-104. Richardson, K. C., L. Jarett, and E. H. Finke. 1960. Embedding in epoxy resins for ultrathin sectioning in electron microscopy. Stain Techno!. 35: 313-323. Rouse, G. W. 2000. Family Serpulidae. Pp. 184-189 in Polychaetes and Allies: The Southern Synthesis. Volume 4A Polychaeta, My:.ostomida. Pogonophora, Echiura. Sipiincula. P. L. Beesley, G. J. B. Ross, and C. J. Glasby, eds. CSIRO Publishing. Melbourne. Schroeder, P. C. 1967. Morphogenesis of epitokous setae during normal and induced metamorphosis in the polychaete annelid Nereis grubei. Biol. Bull. 133: 426-437. Sebens, K. P. 1980. The regulation of asexual reproduction and indeter- minate body size in the sea anemone. Anthopleura elegantissima (Brandt). Biol. Bull. 158: 370-382. Vannini, E., and F. Ranzoli. 1954. Rigenerazione. schizogonia e matu- razione sessulae nel polichete Salmacina incrustans. Boll. Zool. 21: 467-479. Zibrowius, H. 1968. Etude morphologique, systematique et ecologique, des Serpulidae de la region de Marseille. Reel. Trav. Sta. Mar. En- donme Bull. 43: 81-252. Zibrowius, H. 1973. Serpulidae des cotes ouest de 1'Afnque et des archipels voisins. Ann. Mus. R. Afr. Cent. Ser. Quaiin Zool. 207: 1-93. Reference: Bio/ Bull. 200: 118-126. (April 2001) Factors Controlling the Expansion Behavior of Favia favus (Cnidaria: Scleractinia): Effects of Light, Flow, and Planktonic Prey O. LEVY*. L. MIZRAHI, N. E. CHADWICK-FURMAN, AND Y. ACHITUV Faculty of Life Sciences, Bar-Han University, Ramat Can 52900, Israel Abstract. Colonies of the massive stony coral Faviafavus were exposed to different flow speeds and levels of light, and to the addition of zooplankton prey. The relative im- portance of each factor in controlling polyp expansion be- havior was tested. The coral polyps fully expanded when they were exposed to low light intensity (0-40 ju,mol m s~') and high flow speed (15 cm s~'), regardless of prey presence. They also partially expanded under low and me- dium light (40-80 jumol m~2 s~') at medium flow speed (10 cm s~'). The corals expanded their polyps only when they were exposed to light levels below compensation irra- diance (IL.om: light level at which photosynthesis = respira- tion), which was determined to be about 107 ± 24 /nmol m~2 s~'. The results presented here indicate that high flow, low light, and the presence of planktonic prey induce coral expansion. There is a hierarchy of response to these stimuli, in which light level and flow speed are dominant over prey presence. Coral response to these three factors is probably due to the relative importance of gas exchange and zoo- plankton prey. Introduction Diel patterns of polyp expansion and contraction vary between species of anthozoans, including reef-building cor- als (Kawaguti, 1954; Porter, 1974; Lasker, 1979), gorgon- ians (Wainwright, 1967), and sea anemones (Pearse, 1974; Sebens and DeRiemer, 1977). Most reef corals expand their tentacles nocturnally, but some expand continuously or only during the day (Abe, 1939; Porter, 1974; Lasker, 1979). Received 25 April 2000; accepted 14 December 2000. * To whom correspondence should he addressed. levysher@netvision.net.il. E-mail: Corals that expand tentacles at night remain open until dawn, but a beam of light or mechanical stimulation may cause them to contract immediately (Abe, 1939). Sea anem- ones show a similar range of behavior. Structures containing high densities of zooxanthellae (such as pseudotentacles and column vesicles) may expand under conditions favorable for photosynthesis, whereas structures that contain few or no zooxanthellae (such as feeding tentacles) contract under these conditions (Sebens and DeRiemer, 1977). In the stony coral Plerog\ra sinuosa, vesicles with high zooxanthellae density expand only during daytime, and feeding tentacles expand at night (Vareschi and Fricke, 1986). Likewise, polyps of the encrusting gorgonian Erythropodium carib- baeorum have pinnate tentacles with dense populations of zooxanthellae that do not function in nocturnal feeding and are expanded during the day (Sebens and Miles, 1988). It is assumed that, since such structures contract when their main function is impeded, their expansion involves some form of energy cost. When feeding tentacles themselves contain dense populations of zooxanthellae, they serve both func- tions of prey capture and photosynthesis, and generally are expanded continuously. Zooplankton prey are most abundant at night on coral reefs (Sorokin, 1990). so it has been assumed that most corals expand their tentacles nocturnally in order to capture prey (Porter, 1974; Lewis and Price, 1975). Tentacular expansion also is affected by water flow, which reduces the thickness of the boundary layer over tissues and enhances gas exchange (Shashar et ai, 1993). Increases in water flow result in higher (approximately doubled) rates of respiration by sea anemones, stony corals, and soft corals (Patterson and Sebens, 1989; Patterson et ai, 1991 ). The effect of flow on respiration is likely to be magnified when the exposed tissue surface area is increased by exten- sion of either tentacles or specialized column structures. 118 EXPANSION BEHAVIOR OF FAVIA FAVUS Figure 1. Levels of tentacle expansion in colonies of the stony coral Faviafavus. Five levels were arbitrarily designated as follows: 0 = full contraction. 1 = 25% expansion. 2 = 50% expansion, 3 = 75% expansion, 4 = 100% expansion. Each score refers to a whole coral colony. Low flow rate limits the diffusion of materials in the water column to and from coral tissues, thereby regulating and reducing metabolic rates (Patterson and Sebens, 1989; Patterson et ai, 1991). zooxanthellae productivity (Denni- son and Barnes, 1988; Shick, 1990; Patterson et ai, 1991), and nutrient uptake (Atkinson and Bilger, 1992). Oxygen builds up to high levels within and above coral tissue surfaces during the day as a result of photosynthesis, and is extremely depleted near those surfaces at night due to respiration; both gradients are reduced, improving delivery of dissolved substances to and from tissues, with even small increases of flow speed (Shashar and Stambler, 1992; Kuhl et ai, 1995; Shashar et ai, 1996). Polyp size, tentacle morphology, and tissue surface area vary widely among coral species, and all three factors contribute to prey encounter and capture rate. Porter (1976) suggested that large polyps with elongate tentacles special- ize in zooplankton capture, at the expense of high surface area conducive to photosynthesis. However, more recent evidence shows that certain small-polyped corals are able to capture almost the same size range of prey as do those with larger polyps, and actually have much higher rates of prey capture per unit biomass because of the high surface area of the tentacles that they expose to moving water (Sebens. 1987a). The small-polyped coral Madracis mirabilis cap- tures 30 times more zooplankton per coral biomass than does the much larger-polyped Montastrea cavernosa (Se- bens et ai, 1996). The biomass of polyps of M. mirabilis is only l/90th of the biomass of M. cavernosa polyps, but an equal biomass of M. mirabilis colony has 12 times the tentacular surface area of a M. cavernosa colony. A primary cost associated with expansion in corals is increased respiration, facilitated by a larger surface area of tissue in contact with the overlying water, including water inside the coelenteron, than in contracted corals. Polyp expansion is energy consuming, in that it is achieved by pressurization of the coelenteron. Pressurization is brought about by ciliary pumping of water into the coelenteron, which acts on the viscoelastic material of the polyp wall (Barnes. 1987). Rates of respiration are known to be significantly reduced in contracted corals (Brafield and Chapman. 1965; Pearse, 1974; Robbins, and Shick, 1980; Lasker, 1981). The mag- nitude of this difference depends on ambient oxygen con- centration (Beattie, 1971; Sassaman and Mangum, 1972, 1973, 1974; Shumway. 1978) and on the ratio of expanded to contracted tissue area, rather than being a simple response to low external oxygen concentration alone. Contraction may be a means of producing a low oxygen concentration in the coelenteron, resulting in lower metabolic rates and in savings of energy or carbon regardless of the external oxygen level (Sebens, 1987b). In contrast, expansion may aid in the diffusion of excess oxygen away from photosyn- thesizing coral tissues, thus reducing local hyperoxia and its negative effect on metabolic rates or damage due to super- oxide radicals (Lesser and Shick, 1989; Shick. 1991). On 120 O. LEVY ET AL. the other hand, contraction could result in anaerobic metab- olism and oxygen debt. The effects of polyp expansion and contraction on gas exchange and on the key metabolic processes of photosyn- thesis and respiration are unknown in stony corals. In some cases, photosynthesis contributes all of the required carbon for corals, as in the shallow-water, high-light-adapted S/v/o- phora pistillate! (Falkowski et al., 1984), whereas in low- light-adapted S. pistillata (Falkowski et al.. 1984) and in other coral species that depend mostly on predation, photo- synthesis supplies only a small fraction of energy demand. Since, in the course of photosynthesis, oxygen is evolved while carbon dioxide is consumed, it is possible that in some corals and other zooxanthellate organisms, the extension and contraction of tentacles is tightly coupled to photosyn- thetic rates. Tentacle expansion status in corals affects the amount of light absorbed by the zooxanthellae, and in turn photosynthetic rates determine O2 and CO2 concentrations in the coral tissue and surrounding waters. Predation on coral tentacles may cause their retraction at times when predators are active. However, a considerable number of coral species with zooxanthellate tentacles ex- pand during the daytime when many coral predators (e.g., chaetodontid fish) are active, suggesting that predation on tentacles is not a currently active selective force on these behavioral patterns. On coral reefs at Eilat (Gulf of Aqaba, northern Red Sea), colonies of the massive stony coral Favia favus ex- pand their tentacles only after sunset and remain open until sunrise (O. Levy, unpub. obs.). In the present study, we determined the relative importance of flow speed, photosyn- thetically active radiation (PAR), and planktonic prey avail- ability in controlling tentacle expansion behavior in Favia favus. Materials and Methods Coral collection and flow tank setup Colonies of the common coral Faviu favus were collected from the reef adjacent to the Interuniversity Institute for Marine Science at Eilat, Israel, northern Red Sea, from depths of 5-7 m. Individuals of this species were chosen for study because they are abundant on coral reefs at Eilat, their polyps are large (about 1 cm diameter) and easy to observe, and they exhibit regular diel behavior under laboratory conditions. The collected coral colonies, each up to 10 cm diameter, were transferred to the laboratory and placed in shallow tanks supplied with running seawater for 10 h of acclimation before experiments were begun. New corals were collected for each combination of experimental treat- ments. Experiments on tentacle expansion behavior were per- formed in two 25-1 recirculating flow tanks; procedures were based on those described by Vogel and LaBarbera Low flow speed A. low light Time (min) Coral number B medium light 3 -s Time (min) 3A 4A5A Coral number C high light lime (min) Figure 2. Tentacle expansion behavior of the stony coral Favia favus at low flow (5 cm s"'). with variation in light intensity and prey presence (Anemia nauplii). Corals marked with A were fed Anemia after 90 mm. (A) Light intensity 40 /j.mol m " s s"'. (C) Light intensity 120 ^mol m (B) Light intensity 80 fimol m EXPANSION BEHAVIOR OF FAVIA FAVUS , (1978). Each tank was 100 cm long X 10 cm wide X 25 cm high, and seawater was circulated with a 12-V DC motor. Electronic function generators, consisting of computer pro- grams that were interfaced through a digital-to-analog cir- cuit, controlled the motor speed. Flow speed was measured by following the movement of brightly illuminated particles that were added to the water. Particle movements were recorded using a video camera (Sony CCD 2000E, Hi8 PAL system) with a close-up lens. Flow velocity was measured as described by Trager et al. (1990), using a videocassette recorder (Sony EVO-9800P) for frame-to-frame tracking of the movement of the back-lit particles suspended in the flowing water. Coral expansion experiments Colonies of F. favus were tested at three flow speeds: 5, 10. and 15 cm s~ '. These speeds were selected as similar to natural flow speeds on coral reefs at Eilat, which range between about 5 and 20 cm s~' (Genin et al.. 1994). For each set of experiments, 20 colonies were used; five corals were placed in each of four tanks. Two of the tanks served as controls with no flow, while in the other two tanks flow was generated as described above. Halogen lamps coupled to fiber optics were set above all four tanks within each experiment, and three light irradiance levels were tested — 40. 80, and 120 jumol rrT2 s~' (equivalent to 40. 80. and 120 jitE irT2 s~'). Irradiance was measured as photon flux, with a QSI-140 quantum scalar irradiance meter. Temper- ature was maintained at 24 ± 0.1 °C. Each experiment lasted 180 min. and coral expansion behavior was recorded every 10 min. The tanks were rinsed between experiments. In feeding experiments, freshly hatched Anemia nauplii were added ad libidwn. Tentacle expansion of F. favus was scored on a scale based on Lasker (1979). ranging from 0 to 4. where 0 represents 0% expansion (full contraction), 1 is 25% expansion, 2 is 50% expansion, 3 is 75% expansion, and 4 is 100% expansion (Fig. 1). Each score refers to a whole coral colony (Fig. 1 ). The first set of three experiments was conducted at a flow speed of 5 cm s~', with three levels of irradiance (40, 80, 120 /xmol m~2 s~"). In one of the two tanks with water flow, and in one of the control tanks without flow, Artemia nauplii serving as zooplankton prey were added after 90 min. In the second set of three experiments, the flow speed was set to 10 cm s~'. and the last set of three experiments was run with new colonies of F. favus at a flow speed of 15 cm s~'. Thus, in the three sets of experiments, corals were exposed to all possible combi- nations of water flow, light intensity, and prey presence. The corals were replaced with new colonies in each experiment to avoid changes in behavior with experience. At the beginning of all experiments, the tentacles of the F. favus colonies were contracted (Fig. 1). Medium flow speed A. low light Time (min) 5A Coral number B medium light Time (min) 4A5A Coral number C. high light 2A3A 4A5A Coral number Figure 3. Tentacle expansion behavior of the stony coral Favia favus at medium flow (10 cm s~'), with variation in light intensity and prey presence (Artemia nauplii). Corals marked with A were fed Anemia after 90 min. (A) Light intensity 40 fimol m ~ s . (B) Light intensity 80 /nmol s . (C) Light intensity 120 ;u.mol m - s 122 O. LEVY ET AL High flow speed A. low light 3 I Time (min) Cora] number B medium light Time (mm) 1A 2A 3A 4A 5A Coral number C high light Time (min) Coral number Figure 4. Tentacle expansion behavior of the stony coral Favia favus at high flow ( 15 cm s~ ' ), with variation in light intensity and prey presence (Anemia nauplii). Corals marked with A were fed Anemia after 90 mm. (A) Light intensity 40 jiunol trT2 s~'. (B) Light intensity 80 /Mtnol m~2 We assessed possible interaction effects between light intensity, prey presence, and flow speed on tentacle expan- sion behavior in F. favus by performing three-way ANOVA tests on the results. Determination of light compensation point (/„„„) The light compensation point for F. favus was determined by measuring coral respiration in a closed system (Hara- maty et ai, 1997). Photosynthesis (under different light intensities) and dark respiration were measured as changes in oxygen concentration, using a Clark-type O^ electrode (YSI 5331) (Dubinsky et ai, 1987). Measurements were made in a double-walled, cylindrical 230-ml chamber filled with filtered seawater (GF/C 0.45 ^im). Temperature was maintained at 24 ± 0.1 °C. The water in each respirometry chamber was stirred by a magnetic bar beneath a perforated plate supporting the coral. An oxygen electrode detected the rate of oxygen decline in the chamber with an amplifier, which was connected to a data logger (Furier, Israel), cou- pled to an oxygen meter. Oxygen uptake was measured in the dark for 30 min, and then light intensity was increased, in steps of 30 min. to 10. 40, 80, 120. and 200 /Ainol m~2 r1. Colony surface area was measured by covering each coral colony with plasticene, then removing the cover and spread- ing it flat. The outline of the plasticene sheet was then copied onto aluminum foil and cut; surface area was deter- mined by the weight of the foil in relation to the weight of a piece of foil of known surface area (Marsh, 1970). For calculation of light compensation point (Icom), we used computer software attached to the data logger, and the inverse quadratic equation for unit surface area, as described by Ben-Zion and Dubinsky (1988). Results Coral expansion behavior The expansion behavior of F. favus colonies varied with each combination of stimuli. In the first set of experiments at low flow speed (5 cm s~ ' ) under low irradiance levels (40 /nmol m 2 S '), the coral tentacles expanded only partially with or without the addition of Anemia nauplii (Fig. 2A), up to level 1 (25%) (Fig. 1). Most tentacles remained fully contracted under medium irradiance level (Fig. 2B); when prey were added, only one coral responded and expanded up to level 1 (25%). In the control tanks with no water flow, the corals remained contracted during the entire 180 min of the experiment. When corals were exposed to higher irradiance ( 120 /nmol m ), tentacle expansion did not occur at all. and the corals remained contracted during the entire exper- iment (180 min) (Fig. 2C). The same response occurred in the two control tanks in which corals were exposed to no flow at 120 /u,mol m~2 s~'. EXPANSION BEHAVIOR OF FAVIA FAVUS , Table 1 Three-wav ANOVA. with interactions among the variables of light intensity, prey presence, and flow speed on the behavior of tentacle i:\pansiiin in the stony coral Favia favus after ISO nun Source of variation DF Mean square F P Light 2 3298.61 6.55 <0.01 Prev 1 2777.77 5.52 <0.05 Light X Prey 2 1173.61 2.33 Not significant Flow 2 5361.11 10.65 0.001 Light x Flow 4 1027.77 2.04 Not significant Prey X Flow 2 694.44 1.38 Not significant Light x Prey x Flow 4 590.27 1.17 Not significant P < 0.0001. F,I789, = 3.54. In the second set of experiments at intermediate water flow rate (10 cm s"1). at a low light intensity of 40 jumol m " s , and in the absence of Anemia nauplii, two corals expanded up to level 1 (25%). When Anemia nauplii were added, four corals expanded their tentacles — two to level 1 (25%) and two up to level 2 (50%) (Fig. 3A). In the control tanks with no flow, all corals remained contracted during the entire experiment, even when prey was added. When light intensity was increased to 80 ju,mol m " s , only three corals responded. Two colonies out of the group fed with Anemia expanded to level 3 (75%); the third colony from the unfed group expanded only to level 1 (25%) (Fig. 3B). No expansion responses occurred in the two control groups at 80 jLtmol m"2 s~ '. At high light intensity ( 120 /j,mol m^2 s~ ' ) and intermediate flow speed, only four corals expanded their tentacles slightly, two in the Anemia-fed group and two in the non-fed group (Fig. 3C). Again, there was no expansion response by the control corals. In the third set of experiments run at high flow speed (15 cm S ), with low irradiance (40 /zmol m"2 s~ ' ), most of the corals showed positive response and expanded their tentacles. In the group fed with Anemia, four corals re- sponded: two fully expanded to level 4 (100%), and two opened to level 3 (75%). In the non-fed group, four corals responded: two ex- panded to level 1 (25%), and the other two expanded up to level 2 (50%) (Fig. 4A). In the two control groups, with no flow, the corals remained contracted during the entire ex- periment. Similar responses occurred at medium light in- tensity 80 jamol m~2 s~" (Fig. 4B), but no expansion response occurred in the two (no-flow) control groups. Only one coral expanded its tentacles to level 3 (75%) when light irradiance was increased to 120 /xmol m~2 s~', and all the corals, including those in the other control groups, remained contracted during the entire 180 min (Fig. 4C). A three-way ANOVA of expansion behavior with the three variables tested (light, flow, and prey) revealed no significant interactions between the variables (P < 0.0001 Table 2 Three-way ANOVA. without interaction among the variables of light intensity, prey presence, and flow speed on the behavior of tentacle expansion in the stony coral Favia favus after 180 nun Source of variation DF Mean square F P Light 2 3298.61 5.96 <0.01 Prey 1 2777.77 5.02 <0.005 Flow T 5361.11 9.69 <0.001 P < 0.0001, FI5.84, = 7.27. ^(17.89) = 3-54' (Table 1). A three-way ANOVA without interactions showed significant variation in expansion with each variable (P < 0.0001, F(5 84) == 7.27) (Table 2). Duncan's test (Table 3) showed significantly higher fre- quencies of coral expansion at high flow ( 1 5 cm s ~ ' ) than at low (5 cm s"1) and medium flow (10 cm s"1). Expansion rates also were significantly higher at low-to-medium irra- diance levels (40 and 80 /xmol m^2 s"') than at high irradiance (120 /imol m"2 s"1). When prey (Anemia) was present, expansion rates were significantly higher than for corals not fed with prey (Table 3). Light compensation point (7C.,,,J Calculation of the P/I curve revealed that the irradiance level needed to achieve compensation in F. favus was 107 ± 24 p,mol m " s (n = 5 corals tested). We calculated only the Icom, therefore the maximum light intensity that the corals were exposed to was only 200 ju.mol m"2 s"1. Table .1 Duncan's test on the effect of three types of stimuli (light intensify, prey presence, and flow speed) on the tentacle expansion behavior of the stonv coral Favia favus Stimulus Tentacle expansion (mean %) Duncan grouping Light intensity (/xmol m 2 s ') 40 27.5 A 80 19.16 A 120 6.66 B Prey (Anemia nauplii) Present 23.33 A Absent 122: B Flow speed (cm s"1) 5 5 B 10 16.66 B 15 31.66 A Maximum coral expansion occurred when light intensity was low (40 /nmol m~2 s~'), prey were present, and water flow speed was high (15 cm s"1). See Fig. 1 for description of % tentacle expansion. Means with the same letter are not significantly different at P = 0.05. 124 O. LEVY ET AL Figure 5. A conceptual diagram of tentacle expansion behavior in the stony coral Faria favus in response to different levels of light intensity, flow speed, and the addition of prey (Anemia nauplii). Maximum tentacle expansion occurred at dark. At low irradiance. under the compensation point (Icom), corals will expand with correlation to the flow rate. Furthermore, the addition of prey can enlarge tentacle expansion response in those conditions. At high irradiance level, above Icom the tentacles remain contracted regardless of the presence of neither flow nor prey. In still water the corals remain contracted. Discussion We demonstrate here the relative importance of light, flow, and zooplankton prey in controlling the expansion behavior of a stony coral. Although previous studies have examined these factors separately, the present study is the first to test all three factors simultaneously and to elucidate that prey presence is of secondary importance to light and flow in inducing the expansion of coral tissue. Several reports have demonstrated that stony corals and sea anemones expand their tentacles in the presence of food (Mariscal and Lenhoff. 1968; Reimer, 1970. 1971; Lewis and Price, 1975). Electrophysiological studies using suction electrodes in the coral Meandrina meandrites have shown that repeated stimulation of the nerve net evokes oral disk expansion and tentacle extension, similar to the expansion behavior that normally occurs at night when the corals feed (McFarlane, 1978). It is assumed that the nervous responses occur due to prey stimulation of chemoreceptors. as de- scribed for the mushroom coral Heliofungia actiniformis by McFarlane and Lawn ( 1991 ). Flow also is known to cause expansion by corals; high flow speed increases both the potential for encountering plantonic prey and the mass transfer rate of dissolved gases and nutrients (Sebens et al., 1997). Our analysis of three variables simultaneously shows that each factor has an impact on the expansion behavior of F. favus, but that flow and light are the main triggers. Maxi- mum tentacle expansion in F. favus corals occurred with high flow, low irradiance. and the addition of prey (Fig. 4 A). When there was no addition of prey, but light intensity remained low and flow velocity high, the corals responded and expanded their tentacles up to 757c. Positive responses also occurred in medium flow, with medium light (Fig. 3A, B), but there was almost no response at all when light intensity was maximal (120 jumol irT2 s~"), regardless of flow speed or prey presence. At high flow, the benefits of flow to photosynthesis are greater than the costs associated with increased respiration, thus energy available to the coral for growth and reproduc- tion is greater, up to some asymptote, with increasing flow speed (Patterson et al.. 1991; Patterson, 1992). Coral ener- getics depend upon tissue expansion state, which alters both the surface area exposed to flow and the thickness of the tissue layers through which materials diffuse. In high-speed flow environments, flow modulation of coral energetics is minimal. When water flow is low, dense branches or other morphologies that increase the total volume and thickness of a colony's diffusional boundary layer may negatively EXPANSION BEHAVIOR OF FAVIA r'AVI'S I affect the coral's energy balance. In low-flow environments, the diffusional boundary layer may become severely limit- ing to metabolic processes; corals in these environments show morphologies that maximize the exposure of their tissue surfaces to flow (Lesser et a!.. 1994; Sebens et al, 1997). We suggest that expansion behavior in zooxanthellate corals relates to their energy equilibrium between hetero- trophy and autotrophy. The expanded tentacles of F. favus corals have low densities of zooxanthellae and may produce a self-shadow on the rest of the coral tissue that contains dense algae (O. Levy, unpub. obs.). Thus, under conditions of high light, expansion may lead to a decrease in overall photosynthetic rate and an increase in coral metabolic rate, resulting in a net loss of energy. The expansion responses that we observed here lead us to propose a hierarchy of response to the three variables ex- amined. According to our results, in still water, corals will not expand their tentacles even if light and prey levels vary (Fig. 5). When the light level is too high (over Icom), they also will not expand, even when flow and prey vary. When light levels are low (. Zool. 188: 297-306. Sebens, K. P. 1987a. Coelenterata. Pp. 55-120 in Animal Energetics. F. J. Vernberg and T. J. Pandian, eds. Academic Press, New York. Sebens, K. P. 1987b. The ecology of indeterminate growth in animals. Annu. Rev. Ecol. Syst. 371-407. Sebens, K. P., and K. DeRiemer. 1977. Diel cycles of expansion and contraction in coral reef anthozoans. Mar. Biol. 43: 247-256. Sebens, K. P., and J. S. Miles. 1988. Sweeper tentacles in a gorgonian octocoral: morphological modifications for interference competition. Biol. Bull. 175: 378-387. Sebens, K. P., K. S. Vandersall, L. A. Savina, and K. R. Graham. 1996. Zooplankton capture by two scleractinian corals, Madracis mirabilis and Montastrea cavernosa, in a field enclosure. Mar. Biol. 127: 303- 317. Sebens, K. P., J. Witting, and B. Helmuth. 1997. Effects of water flow and branch spacing on particle capture by the reef coral Madracis mirahilis. J. Exp. Mar. Biol. Ecol. 211: 1-28. Shashar, N., and N. Stambler. 1992. Endolithic algae within corals — life in an extreme environment. J. Exp. Mar. Biol. Ecol. 163: 277-286. Shashar, N., Y. Cohen, and Y. Loya. 1993. Extreme diel fluctuations of oxygen in diffusive boundary layers surrounding stony corals. Biol. Bull 185: 455-461. Shashar, N., S. Kinane, P. L. Jokiel, and M. R. Patterson. 1996. Hydromechamcal boundary layers over a coral reef. J. Exp. Mar. Biol. Ecol. 199: 17-28. Shick, J. M. 1990. Diffusion limitation and hyperoxic enhancement of oxygen consumption in zooxanthellate sea anemones, zoanthids, and corals. Biol. Bull. 179: 148-158. Shick, J. M. 1991. A Functional Biology of Sea Anemones. Chapman & Hall, New York. Shumway, S. E. 1978. Activity and respiration in the anemone Metridium senile (L.) exposed to salinity fluctuations. J. Exp. Mar. Biol. Ecol. 33: 85-92. Sorokin, Y. U. 1. 1990. Plankton in the reef ecosystems. Pp. 291-327 in Ecosystems of the World: Coral Reefs. Z. Dubinsky. ed. Elsevier, Amsterdam. Trager, G., J. S. Hawang, and J. R. Strickler. 1990. Barnacle sus- pension feeding in variable flow. Mar. Biol. 105: 1 17-127. Vareschi, E., and H. Fricke. 1986. Light responses of a scleractinian coral (Plemgyra sinuosal Mar. Biol. 90: 395-402. Vogel, S., and M. LaBarbera. 1978. Simple flow tanks for research and teaching. Bioscience 28: 638-643. Wainwright, S. A. 1967. Diurnal activity of hermatypic gorgomans. Nature 216: 1941. Reference: Biol. Bull. 200: 127-143. (April 20011 Development of Host- and Symbiont-Specific Monoclonal Antibodies and Confirmation of the Origin of the Symbiosome Membrane in a Cnidarian-Dinoflagellate Symbiosis TIMOTHY S. WAKEFIELD1 * AND STEPHEN C. KEMPF2 1 Division of Natural Sciences, John Brown University, Siloam Springs, Arkansas 72761; and 'Department of Biological Sciences, Auburn University, Alabama 36849-5414 Abstract. The "symbiosome membrane" as defined by Roth et al. (1988) is a single, host-derived membrane that surrounds an endosymbiotic organism, separating it from the cytoplasm of the host cell. However, in the case of cnidarian-dinoflagellate endosymbioses, clear identifica- tion of the symbiosome membrane is complicated by the fact that each algal symbiont is surrounded by multiple layers of apparent membrane. The origin and molecular nature of these membranes has been the subject of consid- erable debate in the literature. Here we report the develop- ment of host-specific (G12) and symbiont-specific (PC3) monoclonal antibodies that allow separation of the host and symbiont components of these multiple membranes. Using immunocytochemistry at both the light and the electron microscopic level, we present data supporting the conclu- sion that the definitive symbiosome membrane is a single, host-derived membrane, whereas the remainder of the un- derlying apparent membranes surrounding the algal cell are symbiont-derived. The potential for macromolecules asso- ciated with these membranes to act as cellular signals crit- ical to recruiting symbionts and maintaining established symbioses is discussed. Introduction Endocytobiotic associations (symbioses in which one or- ganism lives inside the cell cytoplasm of another organism) are fairly common in nature. They include many pro- karyote-eukaryote associations (Jeon and Jeon, 1976; Received 23 June 2000; accepted 21 December 2000. * To whom correspondence should be addressed. E-mail: twakefie@ jbu.edu. Lerouge et al., 1990; Regensburg-Tuink and Hooykaas, 1993; Udvardi and Day, 1997; Ferrari et al., 1999) as well as a wide variety of eukaryote-eukaryote associations in- cluding parasite-vertebrate (Sibley et al., 1986; Hall et al., 1991; Russell et al.. 1992; Sam-Yellowe, 1992; Collins et al., 1997; Forero et al., 1999) and microalgal-invertebrate (for reviews see Lee and Anderson, 1991 ; Trench, 1993). A common feature of many of these associations is the mem- branes that surround the microsymbiont, effectively sepa- rating it from the cytoplasm of the host cell. These mem- branes form the boundary across which all cell-to-cell communication between the symbiotic partners must occur; thus the membranes become the critical interface for com- munication and control between symbiont and host (Day and Udvardi. 1992). An understanding of how these mem- branes mediate interactions between the partners of a sym- biotic association requires a knowledge of their origin and molecular nature. In an attempt to consolidate endosymbiotic research so that homologies in all the systems could be investigated, Roth et al. (1988) introduced the terms "symbiosome" and "symbiosome membrane" to describe the intracellular sym- biotic compartment in which symbionts reside. In zooxan- thellae-cnidarian symbioses. the symbiosome is commonly recognized to include a continuous outer, or vacuole, mem- brane underlain by multiple layers of additional membrane that surround the "pellicle," or cell wall, of the internal symbiont (Battey, 1992). The origin and nature of these membranes have been debated since they were first de- scribed. Taylor (1968) stated that some of these membranes were formed from host tissue, but others were contributed by the symbiont itself; however, he also stated that deter- 127 128 T. S. WAKEFIELD AND S. C. KEMPF mining which membranes came from which source was difficult. Dodge and Crawford (1970), in their investigation of the "periplast" of zooxanthellae, agreed with Taylor, stating that "it is impossible to tell whether these [mem- branes] belong to the host or alga." Undaunted by this proclamation, many papers have been published suggesting possible origins for these membranes. These suggestions have naturally centered on two possible origins: (1) the host cell (Colley and Trench, 1983; Palincsar et ai, 1988; Rands et ai, 1993; Tripodi and Santisi, 1982), and (2) the algal cell (Kev'metal., 1969; Taylor, 1 97 1 ; Trench and Winsor, 1987; Wakefield et ai. 1998, 2000). Unfortunately, in all cases, suggested origins are presented without any supporting ex- perimental evidence. In this study, immunocytochemical methods were em- ployed to obtain evidence needed to support or dispute the claims of symbiont versus host origins of the multiple layers of apparent membrane surrounding the endosymbiont. The marine anthozoan Aiptasia pallida and its endosymbiont Symbiodinium bennudense (Banaszak et ai, 1993) were used in confocal and electron microscopic investigations that employed two monoclonal antibodies produced against antigens associated with the membranes surrounding the algal symbiont. The data obtained provide the experimental evidence needed to reject one of the proposed hypotheses for the origin of the multiple layers of membrane. The significance of these findings is discussed. Materials and Methods Animal collection and care Specimens of Aiptasia pallida were collected in the Flor- ida Keys and transported to Auburn University, Alabama. Collected animals ranged in length from about 4.0 to 8.0 cm, with an oral disc radius of about 2.5 to 5.0 cm. The animals were maintained in two sets of established 300- gallon aquarium systems with both external trickle and undergravel filtration. Artificial seawater for the aquaria was prepared with deionized water and Reef Crystals (Aquarium Systems) at a salinity of about 30 ppt. One aquarium system was maintained on a 12-h light/1 2-h dark cycle; the other was kept in total darkness. The constant dark condition causes a gradual loss of all photosynthetic symbionts within the host cells; these bleached animals are said to be apo- symbiotic. Animals were fed with newly hatched brine shrimp nauplii every other day. Tank temperature ranged between 25 and 30 °C. Individual animals reproduce asex- ually by producing pedal lacerations: small pieces of tissue that pull away from the pedal disc and grow into a new individual. Thus, within the aquaria, animal sizes ranged from less than 0.5 to about 8 cm in length. Monoclonal antibody production Antigen preparation. Eight specimens of Aiptasia pal- lida, each about 5.0-8.0 cm long, were removed from the aquaria. Two specimens were placed in 6.0 ml of a homog- enization solution consisting of 1.0 mM phenylmethylsul- fonyl fluoride (PMSF) in 0.45-/j,m Millipore-filtered sea- soned aquarium water (MFSAW) pre-cooled to 4 °C. The tissues of the specimens were fully dispersed using a Hand- ishear (Virtis) tissue homogenizer; an ice bath was used to maintain the pre-cooled temperature. This process was re- peated for all eight specimens, which were then combined and split into four samples. Each sample was poured into 15-ml glass centrifuge tubes and centrifuged at about 250 X g for 3 min to pellet the extracted symbiosomes. The su- pernatant was discarded, and the pelleted symbiosomes were resuspended in homogenization solution. This process was repeated three times until a firm pellet of extracted symbiosomes was obtained in each tube. Each symbiosome pellet was resuspended in 1.0 ml of nanopure filtered water (np-H-,0) and either remained untreated or was processed in one of the following ways: (1) fixed in 4% paraformalde- hyde containing 0.2 M Millonigs phosphate buffer (MPB) for 10 min and then rinsed three times in 0.2 M MPB and finally resuspended in np-H2O, (2) boiled for 1 min, or (3) repeatedly frozen and thawed by placing the sample in an Eppendorf tube, dropping it into liquid nitrogen, removing and wanning until it thawed, and then repeating. All four types of samples were then combined to produce about 4.0 ml of "symbiosome slurry." This slurry was used as a source of antigen for the production of antibodies. Antibody production. The production of monoclonal an- tibodies followed standard procedures (Kohler and Milstein, 1975). Briefly, BALB\C mice were injected subcutaneously with 0.1-0.2 ml of the symbiosome slurry in Freund's complete adjuvent. At 3 weeks and 6 weeks, symbiosome slurry in Freund's incomplete adjuvent was injected into the peritoneal cavity. Seven days after the last injection, blood was taken from the tail vein of the inoculated mice. Blood sera were titered 1:50 through 1:10,000 and placed in a 96-well plate coated with fixed symbiosomes to undergo an enzyme-linked immunosorbant assay (ELISA) for the pres- ence of symbiosome antibodies. Once levels of serum antibodies were sufficient, a mouse was selected and sacrificed. Lymphocytes were removed from the spleen, washed to remove red blood cells, and then suspended in RPMI 1640 tissue culture medium (RPMI). Throughout the fusion procedure, this suspension was main- tained at 37 °C by using a dry heating block. A 4: 1 mixture of lymphocytes and AG-8.653 myeloma cells was added to a sterile centrifuge tube and pelleted. The supernatant was removed, and a solution of 50% polyethylene glycol in 20 mM phosphate buffered saline (PEGM) was slowly added while the cell suspension was constantly stirred. Additional ORIGIN OF THE SYMBIOSOME MEMBRANE RPMI was then added while stirring continued. After sev- eral minutes the suspension was again centrit'uged to pellet the cells, and the PEGM + RPMI medium was removed and replaced with RPMI supplemented with BSA (15% bovine serum albumin), OPI (0.15% oxaloacetate, 0.05% pyruvate, 0.0082% bovine insulin), and HAT (5 X 10~5 M hypoxan- thme. 2 X 10~7 M aminopterin, 8 X 10"" M thymidine). The resulting fusion mixture of cells was aliquoted in 100-jul amounts into the wells of 96- well plates that were then incubated at 37 °C. Developing cell cultures were periodically fed and checked as they increased in size. Once the 96-well plate cultures had reached one-third confluency they were screened by an immunofluorescent dot-spot test to identify culture wells that were producing antibodies against sym- biosome antigens (see below). Screening. Symbiosomes that had been extracted using the above-outlined procedure were fixed in 4% paraformal- dehyde and 0.2 M MPB for 10 min. Fixed symbiosomes were rinsed three times in 0.2 M MPB and then resuspended and diluted in np-H,O. Spots of fixed symbiosomes were dried onto spot slides. Each spot was incubated first in 30 /LI! of blocking media (BM) consisting of 5% heat-inactivated goat serum in 20 mM phosphate buffered saline solution containing 0.1% Triton X-100 (PBS+). Subsequently, spots were incubated in 30 /u,l each of RPMI supernatant from cell-culture wells for about 1 h. Spots were then washed three times in PBS+ and incubated in goat anti-mouse fluorescence isothiocyanate (FITC (-conjugated secondary antibody (1:100 in BM) for 1 h, followed by three more rinses in PBS + . Spots were then examined with a Zeiss epifluorescence microscope. Cells in culture wells produc- ing symbiosome antibodies were suspended in RPMI and cloned by limiting dilution. Upon reaching one-third con- fluency, clone wells were screened and positive wells cloned again. This process of recloning was repeated until all of the clones in a freshly inoculated 96-well plate showed the same degree of antigen recognition. The final outcome was the isolation of two hybridoma cell lines (G12 and PC3) producing antibodies that recognized antigens associated with the symbiosome membranes. The two cell lines were then expanded, allowing collection of antibody containing spent culture medium for immunohistochemistry and providing cells for frozen storage in liquid nitrogen. The two cell lines were isotyped using an ISO-1 isotyping kit purchased from SIGMA chemical company. The G12 antibody was found to be an IgG, and the PC3 antibody was an IgM. Immunocytochemistry and confocal microscopy Cultured symbionts. Axenic cultures of Symbiodininm microadriaticum (Freudenthal, 1962) extracted from Cas- siopea xamachana, S. pulchrorum (Banaszak et ai, 1993) extracted from Aiptasia pulchella, and S. bennudense ex- tracted from A. tagetes (= A. pallida Muller-Parker et ai, 1996) have been continuously maintained in incubation chambers within our laboratory for about 3 years (cultures were initially provided by Dr. William Fitt, University of Georgia). Incubation temperatures were maintained at 28.5 °C and illumination of the cultures was on a 12-h light/12-h dark cycle. Cultures were grown in 10 ml of ASP-8A culture medium (Blank, 1987) in sterile 15-ml culture tubes. New cultures were started from existing cul- tures roughly every 2 months, using sterile techniques. Samples from each culture were removed and fixed in 4% paraformaldehyde in 0.2 M MPB. Fixed algal cell samples were rinsed three times in 0.2 M MPB and then incubated in the following solutions: BM for 1 h, PC3 primary antibody (1:100 in BM) or G12 primary antibody (1:100 in BM) for 3 h. rinsed 3 times in PBS+ for 10 min each, FITC- conjugated-IgM-specific secondary antibody ( 1 : 100 in BM) for PC3 antibody or FITC-conjugated-IgG-specific second- ary antibody (1:1000 in BM) for G12 antibody for 1 h. followed by three rinses in PBS+ for 10 min each. Controls were also run using either BM only or a nonspecific anti- body-producing hybridoma cell supernatant in place of the primary antibody solution. Extracted symbiosomes. Symbiotic Aiptasia pallida spec- imens were homogenized, and the symbiosomes were ex- tracted and fixed using the previously described procedures. Labeling of the fixed symbiosomes with the same controls was done using the procedures described above for cultured symbionts. Cn-ostat samples. Small (0.5-1.0 cm) symbiotic and apo- symbiotic specimens of Aiptasia pallida were fixed and rinsed following the same procedure as used on the ex- tracted symbiosomes. After rinsing, these specimens were placed in a 1:1 mixture of 0.2 M MPB and Fisherbrand tissue freezing medium (TFM) and left overnight. Speci- mens were then removed from the 1:1 mixture and placed in 100% TFM. Specimens were frozen, and thin sections were cut using a Reichert-Jung 2800 Frigocut N cryostat. The sections were collected and placed on warm, 4% gelatin- coated slides. The tissues were labeled using the same procedures as those used to label extracted symbiosomes. Sections of symbiotic tissue and aposymbiotic tissue were tested for positive labeling using both PC3 and G12 primary antibodies. The same controls as those used on the extracted symbiosomes were used on the cryosectioned tissue. Whole-tissue samples. Tentacles (2.0-3.0 cm long) from A. pallida were removed from symbiotic specimens. Small (0.5-1.0 cm) whole, aposymbiotic A. pallida specimens were also collected. Both types of tissue were fixed and rinsed following the same procedure as that used for ex- tracted symbiosomes and sectioned tissue. After the final rinse, the aposymbiotic specimens were cut into small pieces (2.0-3.0 mm). Symbiotic tentacles were also cut into 130 T. S. WAKEFIELD AND S. C. KEMPF small pieces (2.0-3.0 mm), and additional cuts was made longitudinally through the tentacle so that the symbiotic endodermal layer was fully exposed to the incubating solu- tions. Single- and double-labeling experiments using G12 and PC3 primary antibody were performed using the same procedures and controls as described above for extracted symbiosomes and sections. Imaging. Labeled symbiosomes, sections, and whole- tissue samples were placed on glass slides and infiltrated with either glycerol mounting medium containing 4% n- propyl gallate (pH 9.5. Giloh and Sadat, 1982) or Per- mafluor aqueous mounting medium (Beckman Coulter). Fluorescent images were captured using a BioRad MRC- 1000 laser scanning confocal microscope equipped with a krypton-argon laser. For those tissues that were labeled only with FITC-conjugated antibodies, the FITC-labeled tissue was visualized first with 488-nm laser excitation and a 500-nm fluorescent emission filter. A second image was then acquired of the autofluorescence of the algal chloro- plast by using 514-nm laser excitation and a 660-nm fluo- rescent emission filter. When the filter sets provided with the confocal microscope are used, autofluorescence due to chlorophyll is very dim; thus it was necessary to maximize the gain to capture images of chloroplast fluorescence. Each image was saved, pseudocolored (green for FITC; red for autofluorescence), and merged using BioRad's confocal as- sistant. For double-labeled tissues the same procedures were employed as for single-labeled tissues, except that maxi- mizing the gain for the RITC-labeled tissue was unneces- sary because its fluorescence was much brighter than the apparent autofluorescence of the chloroplast. To ensure that the labeling could be compared accurately, the positively labeled tissue was imaged first. The laser intensity, iris diameter, gain, black level, and enhancement were all optimized for the FITC label. After switching the filter set, each parameter (i.e., laser intensity, iris diameter, ere. ) was again optimized to capture either the autofluores- cent algal chloroplast or the RITC label. Once both sets of parameters were established (one for FITC and the other for either autofluorescence or RITC label), they were not changed during the image acquisition of other experimental tissues. Electron microscopy Standard fixation and embedding. Small pieces (2-4 mm) of symbiotic tentacle were removed from a live anem- one and fixed in 0.5% glutaraldehyde + 4% paraformalde- hyde + 0.2 M sucrose in 20 mM phosphate buffered saline (PBS) solution at 37 °C for 4 h. Tissues were rinsed in 20 mM PBS + 0.2 M sucrose three times for 15 min each time and dehydrated in an ethanol series (30%, 50%, 70%, 85%, 90%, 95%, 100%) for 30 min in each, at room temperature, with constant rotation. Tissues were then incubated in a 1:1 mixture of 1007f ethanol and LR White and left overnight with constant rotation. The next day. tissues were placed in 100% LR White, incubated for 6 h with constant rotation, and then placed in fresh LR White in an aluminum pan; the pan was placed inside an acrylic plastic chamber and sealed. Intake and outflow valves were used to flood the chamber with N2 and exclude oxygen. Finally, the sealed chamber was incubated at 70 °C overnight to polymerize the plastic. Cryofixation and embedding. As above, small pieces (2-4 mm) of symbiotic tentacle were removed from a live anemone. Two or three pieces were placed in a brass planchet. inserted into a Balzers 010 HPM high pressure freezing machine, and quick-frozen. Frozen tissues were transferred to liquid NT and held there briefly before being transferred into a Leica EM AFS automatic freeze-substitu- tion system. The tissues were placed in a freeze-substitution medium of methanol (MeOH) + 0.5% uranyl acetate. The freeze-substitution medium and tissues were maintained at -90 °C for 2 days before the temperature of the system was slowly ramped up 5 °C/h until it was stabilized at -45 °C. The freeze-substitution medium was removed and the tis- sues were rinsed twice, for 30 min each time, in pure MeOH. The tissues were then incubated at -45 °C in the following: a 1:1 mixture of MeOH:Lowicryl HM20 resin for 2 h. a 1:2 mixture of MeOH:HM20 for 2 h, pure HM20 for 2 h, and finally pure HM20 overnight. Tissues were placed into molds filled with fresh HM20 and polymerized at -45 °C by ultraviolet light over a 3-day period. Sectioning. labeling, and staining. Blocks of the stan- dard-fixed and cryofixed tissues were trimmed, and ultrathin sections were cut using a Reichert-Jung Ultracut E ultrami- crotome. Sections were collected on carbon-coated, Forni- var-covered slot and 100-mesh grids. Grid-mounted sec- tions were incubated on drops of the following solutions: 0.05 M glycine in incubation buffer (10 mM phosphate buffer + 150 mM NaCl + 0.1-0.2% Aurion BSA-c [acety- lated bovine serum albumin] + 20 mM NaN3) for 15 min. blocking medium (incubation buffer + 5% BSA + 0.1 % cold water fish skin gelatin + 5% heat inactivated goat serum) for 30 min. incubation buffer 3X5 min, diluted primary antibody (see below) in incubation buffer over- night, incubation buffer 6 X 5 min, immunogold-conjugated secondary antibodies (see below) diluted in incubation buffer for 2 h. incubation buffer 6x5 min, 10 mM PBS 3 X 5 min, postfixed in 2.5% glutaraldehyde in 10 mM PBS for 5 min and then rinsed in 10 mM PBS for 5 min, nanopure H,O 5x2 min, Aurion silver enhancement for 20 min, nanopure H2O (np-H2O) 3X5 min, saturated uranyl acetate in 50% MeOH for 20 min, and nanopure H:O 3X5 min. Grids were then carefully blotted by touching the edge of the grid to filter paper and allowing most of the water to be wicked away. Grids were then placed on filter paper and allowed to dry. Labeled tissues were viewed and photo- ORIGIN OF THE SYMBIOSOME MEMBRANE graphed using a Zeiss EM- 10 transmission electron micro- scope. Primary antibody dilutions used in the procedure above were either PC3 1:100, G12 1:100. or both 1:100 in incu- bation buffer. Secondary antibodies used were goat anti- mouse (GAM) IgM-specific antibody conjugated to an ul- trasmall ( — 1.0 nm) gold particle (Aurion). GAM IgG- specific antibody conjugated to an ultrasmall gold particle (Aurion). or a combination of the GAM IgM-specific anti- body-ultrasmall gold with a GAM IgG-specitic antibody conjugated to a 20-nm gold particle (ICN). Some sections were treated with protein A conjugated to a 15-nm gold particle (Aurion). Sections were incubated in both 1:10 or 1:100 dilutions of secondary antibody or protein A in incu- bation buffer. Single-label controls included incubations in BM or non- specific primary antibodies in place of PC3 or G12. Controls for nonspecific cross reaction of the secondary antibodies were also done. This included incubation in PC3 followed by the IgG-specific secondary antibody and incubation in G12 fol- lowed by the IgM-specific secondary antibody. Double-label controls included incubations in BM. in nonspecific primary antibodies, in PC3 only, or in G12 only; each was followed by incubations in both secondary antibodies. Ultra-cryomicrotomy. Small pieces (2-4 mm) of symbi- otic A. pallida tentacle were removed from a live specimen. Tentacle pieces were fixed using the same protocol as that used in the standard fixation and embedding procedure described above. However, after the tissue had been rinsed it was placed in an 80% sucrose solution at 4 °C and left overnight. The following day the tentacle pieces were re- moved and placed on small metal stubs. Care was taken to remove excess sucrose solution, and then the stub and tissue were plunged into liquid nitrogen (LN2) until completely frozen. Following this initial freezing the tissues remained frozen at all times, either in LN^ or in the cold chamber of the ultra-cryomicrotomy apparatus. The frozen tissue was trimmed and sectioned on the Reichert-Jung Ultracut E ultramicrotome fitted with an FC4E ultra-cryomicrotomy unit. Trimming was done at -70 °C with glass knives; sections were cut at -100°C with a Diatome 45° angle ultra-cryomicrotomy diamond knife. A Diatome Static Line I was also used to reduce the build-up of static charge on the knife surface, allowing a fine ribbon of 75- to 85-nm sections to be produced. Each ribbon was pulled from the knife's edge with an eyelash brush and picked up by using a tiny drop of 80% sucrose suspended in a wire loop. (The sucrose drop begins to freeze as it is brought near the knife surface, and the tissue was picked up just before the drop was completely frozen. Thus the tissue was frozen onto the sucrose drop and removed from the knife.) The drop was removed from the cryo-chamber and al- lowed to thaw. The drop was then applied, tissue side down. to a 100-mesh, Formvar + carbon-coated nickel grid. The grid adhered to the drop so that it could be flipped over and floated off the wire loop onto a plate of 4 °C solidified 2% gelatin in np-H2O. Grids could be stored on the gelatin for several days before being used in the immunogold labeling procedure. In preparation for labeling the grids, a small amount of the solidified gelatin was removed from around and under each grid. The grids and gelatin were placed on a small piece of Parafilm attached to a glass petri dish and placed in a 37 °C oven for about 30 min. When the plate was re- moved, the grids could be easily picked out of the now- liquid gelatin with fine forceps. The grids were immediately placed into the 0.05% glycine solution and were incubated in all of the same solutions used on the plastic-embedded tissues. After the final rinse in np-H:O, the grids were floated onto a solution of 2% methyl cellulose in np-H2O + 2% uranyl acetate (MU). The grids were then picked up with a wire loop that was just slightly larger than the grid itself, and the excess MU was wicked away by touching the edge of the loop to a piece of filter paper. Care was taken to ensure that only a thin layer of MU was left over the tissue. The MU was then allowed to dry completely, and the grids were removed from the loops and stored in grid boxes until they were viewed and photographed using the Zeiss EM- 10 transmission electron microscope. Results Immunofluorescence single label Table 1 presents the overall results for the immunofluo- rescent labeling of the various tissues used in these exper- iments. The cultured symbionts of the two species of Aipta- sia showed positive labeling with the PC3 antibody. In those cultures that did show positive labeling, only a few cells were actually labeled among a larger population of unlabeled algal cells (Fig. la). However, none of the cul- tured symbionts showed any positive labeling with the G12 antibody (Fig. Ib). The symbionts extracted from Cassio- peia also did not show positive labeling to the PC3 antibody and appeared the same as negative controls (Fig. Ic, d). Extracted symbiosomes from Aiptasia showed very strong positive labeling to the PC3 antibody (Fig. le), whereas most of the labeling with the G12 antibody appeared to be in the contaminating host tissues (Fig. If). Cryosectioned aposymbiotic tissues and pieces of tenta- cle from aposymbiotic animals all showed strong positive labeling for the G12 antibody (Fig. 2a). Neither sections nor whole mounts of aposymbiotic A. pallida showed any pos- itive labeling with the PC3 antibody, and they appeared the same as negative controls (Fig. 2b). Sections and whole mounts of symbiotic A. pallida showed positive labeling with both the G12 antibody (Fig. 2c) and the PC3 antibody (Fig. 2d). As seen in Figure 2c, the G12 antigen was found 132 T. S. WAKEFIELD AND S. C. KEMPF Table 1 Results of imimtnofluorescence imaging of cultured symbionts. extracted symbiosome. anil whole tissue using PC3 and G12 antibodies Tissue or cell tested PC 3 antibody G12 antihody Secondary antibody only Nonspecific primary antihody Cultured symbionts Symbiodinium microadriaticum S. piilchronim S. bermudense Extracted symbiosomes from Aiptasiu pallula Cryosections Aposymbiotic tissue Symbiotic tissue Whole fixed tissue Aposymbiotic tissue Symbiotic tissue around the symbionts but was apparent in other portions of the host tissue as well. In contrast, as seen in Figure 2d, the PC3 antigen was found only directly around the symbionts and was not found throughout the host tissue. Controls for cryosectioned symbiotic and aposymbiotic tissues and whole pieces of tissue that were incubated in either BM or a nonspecific primary antibody showed either no fluorescence in aposymbiotic tissues (Fig. 2b) or only the autofluorescence of symbiont chloroplasls (Fig. 2e). Immunofluorescence double label Double-labeled tissues (Fig. 2f) revealed the PC3 antibody labeling (green) to be confined to a very narrow band imme- diately surrounding the symbionts. Alternatively, G12 labeling (red) was distributed evenly throughout the entire endodermal layer of the host and also appeared as a thin yellow line (the overlap color of red and green) that was immediately adjacent and external to the PC3 label (Fig. 20. [Note that since the autofluorescence of the algal cell chloroplast is emitted at the same wavelength as the much brighter RITC label used for G12 identification, it was not possible to boost the gain and capture it. This resulted in the symbionts appearing as "black holes" (e.g.. Fig. 20. in contrast to what is seen in the single- labeling experiments (e.g., Fig. 2c).| Immunogold single label Table 2 summarizes the results of the immunogold label- ing in these experiments. Standard fixed tissue and rapid- Figure 1. Cultured symbionts and extracted symbiosomes from Aiplasia pallula. labeled with monoclonal primary antibodies PC3 or G12 and fluorescent secondary antibodies. Every picture, including the negative control, is a false color merger of green FITC labeling and red autofluorescence of the algal chloroplast. (a) Specific peripheral labeling of a minority of cultured Syinbiodinium bermudense treated with PC3 antibody, (b) Total lack of labeling in cultured S. bermudense treated withG12 antibody, (c) Total lack of labeling in cultured S. microadriaticum treated with PC3 antibody, (d) Total lack of labeling in the negative control of cultured 5. bermudense treated with secondary antibody only (all negative controls appeared the same), (e) Specific peripheral labeling of extracted symbiosomes from tissues of A. pallida, treated with PC3 antibody, (f) Extracted symbiosomes and contaminating host tissue from A. pallida. labeled with G12 and FITC. Scale bars = 20 /im. Figure 2. Cryosections and whole-tissue pieces of Aiplasia pallida labeled with PC3 antibody, G12 antibody, or both. Each picture, including the negative controls, is a merged image of green FITC labeling and red algal autofluorescence or RITC labeling, (a) General labeling of host tissue in a cryosection of aposymbiotic A. pallula treated with G12 antibody, (b) Total lack of labeling in a cryosection of aposymbiotic A. pallida treated with PC3 antibody, (c) General labeling of host tissue, as well as symbiosome membrane, in whole tissue from symbiotic A. pallida treated with G12 antibody, (d) Specific labeling of symbiosome-associated membranes without host tissue labeling in whole tissue from symbiotic A. pallida treated with PC3 antibody, (e) Total lack of labeling in a negative control of whole tissue from symbiotic A. pal/ida treated only with secondary FITC-conjugated antibody, (f) Double labeling of whole tissue from symbiotic A. pallida with PC3 (FITC— green) and G12 (RITC— red) antibodies; PC3 labeling is clearly limited to the symbiosome, whereas G12 labeling is mainly a characteristic of host tissue, but slightly overlaps the symbiosome, as represented by a thin yellow line surrounding the periphery of some symbiosomes. The black holes are the algal symbionts, the autofluorescence of which cannot be seen at the lower gain levels used to image the RITC signal. The bright orange spot in the center of many algal cells (c, d, e) is either the pyrenoid or the accumulation body ot the cell. Data gathered using tissues labeled with acridine orange, a fluorescent nuclear stain, indicate that this structure is not the nucleus (Wakefield. unpubl. data). Scale bars: (a, b) = 50 /am; (c-f) = 10 /am. ORIGIN OF THE SYMBIOSOME MEMBRANE f >-V W T O— -4 iSgpi F*" .,&&fe * 134 T. S. WAKEFIELD AND S. C. KEMPF Table 2 Results of immunogold labeling of in situ symbionts using PC3 and G12 antibodies Tissue components PC 3 label G12 label 2° ab. only Nonspecific primary PC3 + IgG 2° G12 + IgM 2° Host tissues Ectoderm Mesoglea Endoderm Symbiosome/Symbiont components Outermost symbiosome membrane Multiple layers of membrane Cell wall Thecal vesicles Cytoplasm freeze-freeze-substituted-low-temperature-embedded tissue labeled with G12 and examined with the electron micro- scope showed positive labeling of the host cell plasma- lemma and internal membranes. In particular, the outermost symbiosome membrane was clearly labeled (Fig. 3a). There was also a strong labeling within the mesoglea of the host tissue (Fig. 3b). However, there was no labeling of the symbiont. A close examination revealed that the multiple layers of membrane that are immediately adjacent to the symbiont were not labeled by the G12 antibody (Fig. 3c); the ectodermal layer of the host tissue also lacked label. In fact, long extensions of the mesoglea could be traced into the ectodermal layer due to the strong label found therein and the absence of label within the ectodermal cells (Fig. 3b). Controls that were incubated in either BM or a nonspe- cific primary antibody did not show positive labeling any- where in the tissues (Fig. 3d). The tissue sectioned by ultra-cryomicrotomy confirmed our findings that the G12 antigen indicated specific binding in the host membranes and mesoglea (Fig. 4a. b). It also revealed the specificity of the PC3 antibody to the algal symbiont (Fig. 4c. d). The labeling with the PC3 antibody was significant because our early attempts to label standard- fixed, LR-White-embedded tissue with PC3 had totally failed. Thus, the ultra-cryomicrotomy technique provided our first glimpse of the specificity of the PC3 antibody to the algal symbiont. Unfortunately, the preservation of ultra- structure in the ultra-cryomicrotomy technique was less than favorable, necessitating the use of the freeze-substitut- ed-Lowicryl-embedding technique. Labeling of the freeze-substituted-Lowicryl-embedded tissues with the PC3 antibody revealed strong labeling of the symbiont (Fig. 5a). Although there was a small amount of labeling present in the chloroplast and other cell or- ganelles, the vast majority of label was found in the cyto- plasm of the symbiont. Although the internal thecal vesicles and plates of the symbiont did not appear to be labeled, the PC3 antigen was concentrated around the periphery of the cell, between the chloroplast profiles (Fig. 5b). It was some- times also seen in the cell wall directly adjacent to these peripheral cytoplasmic concentrations (Fig. 5b). Dense la- beling was also present in the space between the symbiont and the outermost symbiosome membrane, including areas where multiple layers of membrane were present (Fig. 5b). Host cell cytoplasm and associated membranes surrounding the symbiosomes were essentially devoid of label (Fig. 5a). The minute amount of labeling in the host cell cytoplasm was similar to that of the negative controls (Fig. 5c) and was thus considered to be background labeling. Immunogold double label Tissues labeled with both primary antibodies and differ- ent-sized gold-conjugated secondary antibodies confirmed our single-labeled experiments. The outermost symbiosome membrane, host endodermal cell membranes, and mesoglea showed dense labeling with large gold particles, indicating the presence of the G12 antigen. Conversely, the multiple layers of membrane directly adjacent to the cell wall, the cytoplasm between the chloroplast profiles, and the cell wall adjacent to these cytoplasmic areas were heavily labeled with small gold particles, indicating the presence of the PC3 antigen (Fig. 6a. b). Negative controls in which the primary and secondary antibodies were switched did show a small amount of cross reactivity. This was most evident with PC3 and the IgG-specific antibody, perhaps due to its larger particle size. However, the cross reaction of the IgG-specific antibody to PC3 confirmed that the PC3 antibody was specific for the algal cell (Fig. 6c). There was some cross reactivity of the IgM antibody to G12. but this also con- firmed that the G 1 2 antibody was specific for host tissue and not for the algal cell. Discussion The results of both the immunofluorescent and immuno- gold labeling experiments clearly support the hypothesis ORIGIN OF THE SYMB1OSOME MEMBRANE 13: Figure 3. Labeling of host membranes in plastic-embedded tissues of symbiotic Aiptasiti pallida treated with G12 antibody and visualized with immunogold particles, (al General labeling of host cell membranes in standard aldehyde-fixed and dehydrated A. pullula, embedded in LR White (G12 visualized with protein A-15-nm gold particles), (b) Extension of labeled mesoglea (arrowheads) into unlabeled ectoderm in freeze- substituted A. pallida, embedded in Lowicryl (G12 localization visualized with silver-enhanced ultrasmall gold particles), (cl Labeling of the outermost symbiosome membrane in standard aldehyde-fixed and dehydrated A. pallida, embedded in LR White (G12 visualized with protein A-15-nm gold particles); insert shows labeling of the outermost membranes of two adjacent symbiosomes. Note that host membrane (white arrowhead) is labeled, but inner, symbiont-associated membranes (black arrowheads) are not. (d) Lack of specific labeling in a negative control section of standard aldehyde-fixed and dehydrated A. pallida tentacle embedded in LR White, incubated without G12 antibody followed with incubation with protein A-15-nm gold particles; all negative controls appeared the same with only minor background scattering of gold particles, ch = chloroplast. ec = ectodermal cells, en = endodermal cell cytoplasm, ms = mesoglea, nu = nucleus. that the two monoclonal antibodies, G12 and PC3. are specific for the different partners in this symbiotic associa- tion. The fact that G12 shows positive labeling of host tissue in whole mounts and sections of Aiptasiu pallida and no labeling of cultured symbionts demonstrates that it is bind- ing to some antigen that is found exclusively in host tissue. Also, since it shows positive labeling of both symbiotic and aposymbiotic tissues, it is clear that this antigen is not symbiont-induced and is present in the host tissue regardless of its symbiotic condition. By contrast, PC3 labeling is apparent only in symbiotic tissues and is completely absent in the aposymbiotic condition. One possible explanation is that the PC3 antigen is induced in the husi by the presence of the symbionts and then somehow transferred to them. However, since in vitro cultured Symbiodinium bermudense and 5. piilchrorum also show positive (albeit reduced) la- beling with the PC3 antibod) . it is clear that this antigen is symbiont-derived. The finding that cultured 5. microadriaticum is not la- beled by the PC3 antibody but both 5. piilchrorum and S. bermudense are labeled may reflect phylogenetic differ- ences between the three tested algal species. In past studies. 136 T. S. WAKEFIELD AND S. C. KEMPF Figure 4. Ultra-cryomicrotomy-sectioned symbiotic Aiplasia pallida labeled with G12 or PC3 and immu- nogold particles indicating the specificity of the antibodies to their respective tissues. Fixation quality varied greatly, but cell morphology was adequate to distinguish host versus symbiont. (a) Host endodermal cell cytoplasm labeled in tentacle tissue of A. pallida, incubated with G12 and an IgG-specific secondary antibody conjugated to ultrasmall gold particles. Note that algal symbionts are not labeled, (b) Higher magnification of the ultra-cryomicrotomy-sectioned A. pallida. labeled with G12 and an IgG-specific secondary antibody conjugated to an ultrasmall gold particle (1:10). (c) Labeling of algal symbionts and associated material in tentacle tissue of A. pallida incubated with PC3 and an IgM-specific secondary antibody conjugated to an ultrasmall gold particle. Note the labeling of material between the symbiont and the host cytoplasm as well as internally within the algal symbiont. Label within the symbiont is localized in the cytoplasm surrounding the chloroplasts. Host cytoplasm is devoid of label, (d) Higher magnification of (he ultra-cryomicrotomy-sectioned A. pallida. labeled with PC3 and an IgM-specific secondary antibody conjugated to an ultrasmall gold particle (1:10). ch = chloroplast, en = endoderm. evidence based on karyotyping (Blank and Trench. 1985). morphology (Schoenberg and Trench. 1980b; Trench and Blank, 1987), isoenzyme patterns (Schoenberg and Trench, 1980a), physiology (Fitt and Trench. 1981: Chang et al.. 1983; Fitt and Trench, 1983), and biochemistry (Chang and Trench, 1982; Govind et al.. 1990) has indicated that the genus Symbiodinium consists of many different species. Also, an examination of similarities and differences in the genes that encode for the small ribosomal subunit of algal symbionts has revealed that S. bermudense and S. piilchro- rum are closely related but different species, and both have been placed into clade B of the "zooxanthellae" phyloge- netic tree (Rowan, 1998). S. microadriaticum, on the other hand, is more distantly related and has been identified as belonging to clade A (McNally et al.. 1994; Rowan, 1991, 1998; Rowan and Powers, 1991). The results of our immu- nofluorescent studies support the concept of phylogenetic diversity among the symbionts and may be added to the list of biochemical differences that exist among some Symbio- dinium species. The double-labeled immunorluorescent and immunogold results clearly demonstrate that the algal symbionts are surrounded by layers of material that originate from the different symbiotic partners. The narrow region of PC3 ORIGIN OF THE SYMBIOSOME MEMBRANE 0. cw py Figure 5. Freeze-substituted symbiotic Aipuiua pa/lula labeled with PC3 and an IgM-specitic secondary antibody conjugated to an ultrasmall gold particle. The fixation of this tissue was extremely good for EM cytochemical studies. Both host and symbiont tissues were well preserved, and distinct morphological features were easily discernible, (a) Freeze- substituted symbiotic ,4. pallidu tissue labeled with PC3 and ultrasmall gold antibody ( 1 : 100); arrowheads indicate label of muluple layers of membrane; note the absence of label in either the host endodermal layer or the symbiont's cell v\ oil i b) Specific labeling of symbiont membranes and no host-tissue labeling in freeze-subsututed ,4. /%////Mr -v .>;-•?. .""ik-.^. :.- J >t-iai&^N * • -:;/ ;; : • ,;;:|v: _ 0.5urh . - • Figure 6. Freeze-suhslitutcd symbiotic Aiplasiu pallida double labeled with both G12 and PC3 and with IgG-specific-antibody-20-nm gold particle and IgM-specific-antibody-ultrasmall gold particle, (al Note the difference in particle size between the host membranes and the symbiont membranes and cytoplasm that clearly defines the distribution of the G12 (large particles) and PC? (small particles) antigens; arrowheads indicate abundance of PC3 antigen at the periphery of the algal cell, particularly between the chloroplast profiles, (b) Higher magnification of the freeze-substituted symbiotic A. pallida double label, clearly showing the PC3- labeled symbiont membranes and the 012-labeled host membranes, (cl Freeze-substituted symbiotic A. pallida cross reactivity control in which PC3 was incubated with the IgG-specific secondary antibody: the slight cross reactivity of this nonspecific secondary antibody supports the assertion that PC3 is an algal-cell-specific antibody (note the lack of label in the host tissues), ch = chloroplast. cw = cell wall, en = host endodermal tissue, ml = multiple layer of symbiont membranes. algal cell wall, but only in areas directly adjacent to high cytoplasmic levels of this antigen. This suggests that these may be areas of secretion for a molecule or molecules containing the PC3 antigen. The fact that the PC3 antigen is found both within the algal cell cytoplasm and also closely associated with the multiple layers of membrane, beneath ORIGIN OF THE SYMBIOSOME MEMBRANE i: the G12-labeled host symbiosome membrane, supports the hypothesis that these membranous layers are of symbiont rather than host origin. The results of this research should help clarify some confusing terminology currently used in the field of cni- darian-dinoflagellate endosymbiosis. The term "symbio- some," as introduced by Roth et al. ( 1988). was defined as "a membrane-bound compartment containing one or more symbionts and certain metabolic components located in the cytoplasm of eukaryotic cells." They further describe the "symbiosome membrane" as being a host-derived mem- brane that surrounds the microsymbiont, creating a "sym- biosome space" (Roth et al., 1988). This terminology has been adopted by researchers in various fields of endosym- biotic research, including plant- bacteria interactions (Day and Udvardi, 1992; Udvardi and Day, 1997). protozoan- bacteria interactions (Choi and Jeon. 1992; Jeon, 1992), and invertebrate-algal interactions (Rands et a I., 1993; Wake- field et al., 1998; Wakefield et al., 2000). For plant-bacte- rial and protozoan- bacterial interactions, these terms are adequate and clearly define the endosymbiotic situation of these partners. However, for invertebrate-algal symbioses. particularly cnidarian-dinoflagellate symbioses, use of the term "symbiosome membrane" has been problematic. In situ dinoflagellate symbionts have been identified as existing predominantly in a coccoid, or vegetative, state within the host cells (Trench and Blank, 1987). The coccoid state is defined by the presence of the spherical "pellicle," or cell wall. However, exterior to the cell wall exist multiple layers of apparent membranes. Historically, these layers have been referred to by a variety of terms, including the periplast (Dodge and Crawford. 1970; Kevin et al., 1969: Palincsar et al.. 1988: Taylor. 1968: Trench and Blank. 1987), the amphiesma (Schoenberg and Trench, 1980a; Thinh et al.. 1986), the theca (Battey, 1992; Taylor. 1971). the vacuolar membranes (Colley and Trench, 1983), or simply the cell covering (Tripodi and Santisi, 1982). The origin of these membranes has also been debated for quite some time. Tripodi and Santisi (1982). working with the symbiotic octocoral Eunicella stricta, assigned the multiple mem- branes to the host cell. They stated that the location of recognizable host organelles on the "inner side" of the membranes indicated that, in this host species, the mem- branes were composed of host plasmalemma. They also stated that the "many times folded" plasmalemma would have an increased surface area and thus could have a func- tional significance; however, they did not relate what this might be. nor did they indicate how the plasmalemma would accomplish this folding process (Tripodi and Santisi, 1982). Palinscar et al. ( 1988), working with Aiptasia pallida, also assigned the multiple membranes to the host but suggested that they were formed from internal host cell membranes rather than from the plasmalemma. They hypothesized that membrane-bound cytoplasmic extensions of host tissue could envelope the symbiont. followed by the extrusion of the cytoplasm. This would produce the multiple layers of membrane with no cytoplasmic separations (Palincsar et al., 1988). Other authors, including Colley and Trench (1983) and Rands et al. ( 1 993 ), have also assigned the origin of the multiple membranes to the host cell. On the other hand, Kevin et al. (1969), Schoenberg and Trench (1980b). Trench and Blank (1987). and Muller- Parker et al. (1996) have identified a single host-derived membrane and assigned the origin of the multiple layers of membrane to the algal symbiont. Most do so without pos- tulating a mechanism for how these membranes might be produced; however. Trench and Blank (1987) have sug- gested one possibility. They stated that the cell wall of symbionts is covered by an acidic polysaccharide or glyco- protein layer that resembles a membrane at the ultrastruc- tural level, but in fact contains no lipids. They further suggested that this layer is occasionally sloughed off the outer surface of the cell wall. As these layers accumulate, they appear as multiple membranes exterior to the cell wall and produce the peripheral scroll-like layers of apparent membrane often seen in ultrastructural sections (Trench and Blank. 1987). In earlier papers (Wakefield et al.. 1998, 2000). we also suggested that the membranes were derived from algal cells. but we proposed a very different mechanism for the pro- duction of these membranes than that put forth by Trench and Blank (1987). Briefly, we hypothesized that the algal symbionts were continuing through their normal ecdysis cycle within the host cell vacuole. By cycling through the coccoid-to-thecate stage and back again, a layer of apparent multiple membranes would accumulate between the host cell symbiosome membrane and the symbiont. The mem- branous layers would be composed of the sloughed plasma membrane, thecal vesicles, and plates from each ecdysis event. Therefore, we feel that any assignment of the multi- ple layers of membrane to the host (with the exception of the single outermost symbiosome membrane) is unsubstan- tiated and probably in error. If the host-derived origin for the membranes were correct, then each and every membrane in the layer could be defined as the "symbiosome membrane." as defined by Roth et al. (1988). However, the results of labeling with our PC3 antibody indicate that the multiple layers of membrane lying between the host vacuole membrane and the symbiont cell wall originate from the algal symbiont and are held in place by the presence of a single host-derived membrane. There- fore, the host-derived membrane is the definitive symbio- some membrane, while the other apparent membranes could be defined as symbiont-derived "metabolic components" within the symbiosome space (see Roth's [1988] original definition of the symbiosome). However, designating these membranes as mere "metabolic components" may belie 140 T. S. WAKEFIELD AND S. C. KEMPF their potential importance to the symbiotic system. In sub- sequent text, the term "symbiont membranes" is used to describe these structures. On the basis of the presence of the PC3 antigen within the symbiont membranes and the cytoplasm, it may be hypoth- esized that, during the ecdysis process, cytoplasmic material containing the PC3 antigen is also shed and becomes trapped between the layers of plasmalemma and thecal vesicles described by Waketield et al. (2000). Another possibility is that the PC3 antigen is simply a substance exuded through the cell wall during the coccoid stage of the life cycle. This material could accumulate and become trapped between the layers of shed plasmalemma and thecal vesicles during ecdysis. Alternatively, if Trench and Blank (1987) are correct, exuded PC3 antigen would become trapped between the layers of polysaccharide or glycopro- tein material that is occasionally sloughed from the outer cell wall. Although the identification of the PC3 and G12 antigens is still under investigation in our laboratory, recent data indicate that the PC3 antigen is. in fact, an exuded material (Waketield and Kempf, unpubl. data). Markell and Trench ( 1993) have demonstrated that cultured symbionts do exude large water-soluble glycoconjugates. If this exudate is pro- duced in situ (in the symbiotic state), then these molecules could provide essential nutrients to the host (provided the host could digest them), and might also serve as molecular signals between the host and the symbiont (Markell and Trench, 1993). The potential of the symbiont membranes (or trapped exudates) to act as molecular signals is suggested by the findings of Colley and Trench (1983). They performed a series of experiments in which freshly isolated symbio- somes from Cassiopeia xamachana, Aiptasia pallida, A. piilchella, Anthopleura elegantissima, and Zoanthus sp. were introduced into the gastrovascular cavity of aposym- biotic scyphistomae of C. xamachana. They also introduced cultured algae isolated from the cnidarian hosts C. .\am- achana, A. pallida, A. tagetes. and Zoanthus sociatus. They found that all cultured algae, even those that were specific for C. xamachana, were either completely rejected or phagocytosed at almost immeasurably low rates. By con- trast, the freshly isolated symbiosomes from each host spe- cies were all phagocytosed at relatively rapid rates. When viewed with the transmission electron microscope, all of the freshly isolated symbionts were surrounded by multiple layers of apparent membrane. These membranes were com- pletely absent from the cultured algal cells. After several experiments, the authors concluded that a recognition pro- cess must be occurring at the surface between the host endodermal cell and the specific membrane layers that sur- round the algal symbiont (Colley and Trench. 1983). Believing that there were no algal cell membranes exte- rior to the cell wall, Collev and Trench ( 1983) assigned the origin of these membranes to the host and called them "vacuolar membranes." They hypothesized that the en- hanced uptake of freshly isolated symbiosomes was a result of the presence of these host membranes, and that there must be a specific molecular factor involved. Since the host endodermal cell was making its initial contact with various "host membranes" of injected symbiosomes, they con- cluded that there was no actual intercellular (host-symbi- ont) recognition occurring, and that the uptake of symbio- somes might be indistinguishable from the phagocytosis of food particles. As we have clearly demonstrated in this study, \nAiptasia pallida there are a multitude of symbiont-produced mem- branes or membrane-like layers that surround the algal cell wall and are associated with high concentrations of the symbiont-derived PC3 antigen. We postulate that the rapid uptake of isolated symbiosomes compared to the low uptake of cultured algae observed by Colley and Trench (1983) may have resulted from the absence of the correct molecular signal at the external surface of most cultured algal cells. Since our results reveal the PC3 antigen to be absent from most (but not all) suspended, cultured algal symbionts from A. pallida and A. pulchella and present in all symbiosomes in situ, it is possible that such an antigen may be the actual signal that stimulated rapid phagocytosis in the experiments of Colley and Trench (1983). Similarly, since cultured sym- bionts lack the multiple layers of symbiont membranes found in situ, it is also possible that these membranes may contain the molecular signal that stimulates phagocytosis. We feel that it would be useful to reexamine the question of host recognition of algal symbionts with attention to these possibilities. It is unknown how aposymbiotic cnidarian offspring that must recruit their symbionts from the environment achieve an initial infection (Berner et al., 1993; Davy et ai, 1997; Schwarz et al.. 1999); however, there is some support for the hypothesis that the motile thecate stage of the symbiont rather than the sessile coccoid stage is capable of infecting (Kinzie. 1974). If the plasmalemma of the thecate stage contains the appropriate molecular signal, these algal cells would be readily phagocytosed, thus resulting in the initial infection. If such signal molecules can stimulate phagocy- tosis, then other host- or symbiont-specific molecules must be involved in maintaining the symbiosis once it is estab- lished (Trench. 1993). Although Colley and Trench (1983) found that all of the freshly isolated symbionts were phago- cytosed by Cassiopeia xamachana scyphistomae. only Sym- hiodinium microadriaticum and 5. bermudense were ulti- mately maintained in the endodermal cells and could stimulate strobilation. Therefore, the signals that stimulate phagocytosis could not be the same as those that maintain the symbiosis. Host-symbiont-specific signaling has been shown to oc- cur in many other symbiotic associations such as the Rlii- ORIGIN OF THE SYMBIOSOME MEMBRANE zobium— legume association (Lerouge et ai, 1990; Sanchez et a!., 1991), pathogenic-bacteria-plant interactions (Dixon and Lamb, 1990: Regensburg-Tuink and Hooykaas, 1993), Mycobacterium-macmphage invasions (Ferrari et ai, 1999), X-bacteria-amoeba symbiosis (Jeon, 1992, 1997), parasite-host infections (Sibley et ai, 1986; Sibley and Krahenbuhl, 1988; Perkins, 1989; Sam- Yellowe, 1992; Col- lins et al.. 1997), and diatom-foraminifera interactions (Lee et al., 1988, 1997; Lee. 1998; Chai and Lee, 1999, 2000). In many of these associations, the specific signals leading to successful symbiotic interactions have been elucidated. Agrobacterium tumefaciens, a bacterium that induces tu- mors in many types of plants, is dependent upon the VirF protein for tumor formation (Regensburg-Tuink and Hooy- kaas, 1993). Plasmodium falciparum uses several proteins, including the 155-kDa ring-infected surface antigen and the 140/130/1 10/-kDa rhoptry complex, to bind to appropriate erythrocyte membranes (Sam- Yellowe, 1992). Rhi~obiiim meliloti uses a (3-1,4-tetrasaccharide of o-glucosamine to induce root nodules in alfalfa (Lerouge et al., 1990). My- cobacterium species use the TACO protein to inhibit lyso- somal fusion in macrophages (Ferrari et al., 1999). The frustules of all endosymbiotic diatoms of large foraminifera display a 104-kDa polypeptide that prevents their digestion by the host (Chai and Lee, 1999). Attempts to locate and characterize similar types of mo- lecular signals required for establishing and maintaining zooxanthella-cnidarian symbioses have met with little suc- cess; however, the two reported attempts to identify these signals concentrated on the algal cell wall (Markell et ai, 1992) and on water-soluble glycoconjugates released by algal cells in culture (Markell and Trench, 1993). Because we have identified the symbiont membranes to be of algal- cell origin and intimately associated with a symbiont-de- rived antigen (PC3), and because an earlier investigation (Kinzie, 1974) supports the hypothesis that molecular sig- nals associated with the thecate motile stage must stimulate phagocytosis, it may be hypothesized that these symbiont membranes are involved in this signal transduction event. It is likely that knowledge of the identity and function of such molecular signals will be critical to our understanding of the cellular mechanisms orchestrating events such as coral bleaching (Bunkley-Williams and Williams, 1990; Ghiold, 1990; Gleason and Wellington, 1993; Goreau and Hayes, 1994; Brown et ai, 1995), where such signaling capabilities have apparently been lost. Acknowledgments The authors thank David Werneke for many hours of hard work during the production of the two monoclonal antibod- ies. We also thank Dr. Hong Yi and Jan L. M. Leunissen for their aid, expertise, and generosity in supplying materials and advice concerning immunogold labeling. Dr. William Fitt and the Key Largo Marine Research Laboratory, and Dr. Mike Miller and the Auburn University Biological Elec- tron Microscope Imaging Facility provided assistance with some aspects of this research. This work was supported by NSF #9018698 (SCK), NSF-BIR #9220230 (SCK), ONR #3231114 (SCK), and a grant from the Alabama Agricul- tural Experiment Station (SCK). Literature Cited Banaszak, A. T., R. Iglesias-Prieto, and R. K. Trench. 1993. Scripp- siella velellae sp. nov. (Peridmiales) and Gloedodinium viscum sp. nov. (Phytodiniales), dinoflagellate symbionts of two hydrozoans (Cni- daria). / Phycol. 29: 517-528. Baltey, J. F. 1992. Carbon metabolism in zooxanthellae-coelenterate symbiosis. Pp. 153-187 in Algae and Symbioses, W. Reiser, ed. Bio- press, Bristol. England. Berner, T., G. Baghdasarian, and L. Muscatine. 1993. Repopulation of a sea anemone with symbiotic dinoflagellates: analysis by in vivo fluorescence. / Exp. Mar. Biol. Ecol. 170: 145-158. Blank, R. J. 1987. Cell architecture of the dinoflagellate Symbiodinium sp. inhabiting the Hawaiian stony coral Moniipora verrucosa. Mar. Biol. 94: 143-155. Blank, R. J., and R. K. Trench. 1985. Speciation and symbiotic dinoflagellates. Science 229: 656-658. Brown, B. E., M. D. A. Letissier, and J. C. Bythell. 1995. Mechanisms of bleaching deduced from histological studies of reef corals sampled during a natural bleaching event. Mar. Biol. 122: 655-663. Bunkley-Williams, L., and E. H. Williams. 1990. Global assault on coral reefs. Nat. Hist. 99: 46-54. Chai, J., and J. J. Lee. 1999. Initial recognition of endosymbiotic diatom surface antigens by the larger foraminifer Amphistegina lobif- era. Symbiosis 26: 39-53. Chai, J., and J. J. Lee. 2000. Recognition, establishment and mainte- nance of diatom endosymbiosis in foraminifera. Micropaleonlology 46: 182-195. Chang, S. S., and R. K. Trench. 1982. Peridinin-chlorophyll a-proteins from the symbiotic dinoflagellate Symbiodinium (= Gymnodinium) microadriaticum Freudenthal. Pn>c. R. Soc. Lond. B. 215: 191-210. Chang, S. S., B. B. Prezelin, and R. K. Trench. 1983. Mechanisms of photoadaptation in three strains of the symbiotic dinoflagellate Sym- biodinium microadriaticum. Mar. Biol. 76: 219-231. Choi, E. Y., and K. W. Jeon. 1992. Bacterial-endosymbiote-derived lipopolysaccharides on amoeba symbiosome membranes. J. Protozool. 39: 205-210. Colley, N. J., and R. K. Trench. 1983. Selectivity in phagocytosis and persistence of symbiotic algae by the scyphistoma stage of the jellyfish Cassiopeia xamachana. Proc. R. Soc. Land. B 219: 61-82. Collins, H. L., U. E. Schaible, J. D. Ernst, and D. G. Russell. 1997. Transfer of phagocytosed particles to the parasitophorous vacuole of Leishmania mexicana is a transient phenomenon preceding the acqui- sition of annexin I by the phagosome. J. Cell .SV/'. 110: 191-200. Davy, S. K., I. A. N. Lucas, and J. R. Turner. 1997. Uptake and persistence of homologous and heterologous zooxanthellae in the tem- perate sea anemone Cereus pedunculatus (Pennant). Biol. Bull. 192: 208-216. Day, D. A., and M. K. Udvardi. 1992. Metabolite exchange across symbiosome membranes. Svm/ww.v 14: 175-189. Dixon, R. A., and C. J. Lamb. 1990. Molecular communication in interactions between plants and microbial pathogens. Anna. Rev. Plant Physiol. 41: 339-367. Dodge, J. D., and R. M. Crawford. 1970. A survey of thecal fine structure in the Dinophyceae. Bol. J. Linn. Soc. 63: 53-67. 142 T. S. WAKEFIELD AND S. C. KEMPF Ferrari, G., H. Langen, M. Naito, and J. Pielers. 1999. A coat protein on phagosomes involved in the intracellular survival of mycobacteria. Cell 97: 435-447. Fitt, VV. K., and R. K. Trench. 1981. Spawning, development, and acquisition of zooxanthellae by Triducmi squamosu (Mollusca, Bi- valvia). Bio/. Bull. 161: 213-235. Fitt, W. K., and R. K. Trench. 1983. Endocytosis of the symbiotic dinoflagellate Symbiodinium microadriaticum Freudenthal by endoder- mal cells of the scyphistomae of Cassiopeia xamachana and resistance of the algae to host digestion. J. Cell Sci. 64: 195-212. Forero, M. E., M. Marin, A. Corrales, I. Llano, H. Moreno, and M. Camacho. 1999. Lcislnnania amazonensis infection induces changes in the electrophysiological properties of macrophage-like cells. / Membr. Biol. 170: 173-180. FYeudenthal, H. D. 1962. Symbiodinium gen. nov. and Symbiodinium microadriaticum sp. nov., a zooxanthella: taxonomy, life cycle and morphology. J. Protozoa!. 9: 45-52. Ghiold, J. 1990. White death: the fate of a deserted coral. New Sci. 126: 46. Giloh, H., and J. \V. Sedat. 1982. Fluorescence microscopy: reduced photo-bleaching of rhodamine and fluorescein protein conjugates by ;i-propyl gallate. Science 217: 1252-1255. Gleason, D. F., and G. M. Wellington. 1993. Ultraviolet radiation and coral bleaching. Nature 365: 836-838. Goreau. T. J., and R. L. Haves. 1994. Coral bleaching and ocean "hot spots." Amhio 23: 176-180. Govind, N. S., S. J. Roman, R. Iglesias-Prieto, R. K. Trench, E. L. Triplet!, and B. B. Prezelin. 1990. An analysis of the light-harvest- ing peridinin-chlorophyll a-proteins from dinoflagellates by immuno- blotting techniques. Prnc. R. Soc. Loud. B 240: 187-195. Hall, B. F., G. C. Furtado, and K. A. Joiner. 1991. Characterization of host cell-derived membrane proteins of the vacuole surrounding dif- ferent intracellular forms of Trypanosoma cruzi in J774 cells. J. Im- munol. 147: 4313-4321. Jeon, K. \V. 1992. Macromolecules involved in the amoeba-bacteria symbiosis. J. Protozool. 39: 199-204. Jeon, K. W. 1997. Symbiosis and macromolecules. Pp. 359-366 in Eukaryolism and Symbiosis, H. E. A. Schenk, R. G. Herrman, K. W. Jeon, N. E. Muller, and W. Schwemmler, eds. Springer- Verlag, Hei- delberg. Jenn, K. W., and M. S. Jeon. 1976. Endosymbiosis in amoebae: re- cently established endosymbionts have become required cytoplasmic components. J. Cell. Physiol. 89: 337-347. Kevin, M. J., W. T. Hall, J. J. A. McLaughlin, and P. A. Zahl. 1969. Symbiodinium microadriaticum Freudenthal, a revised taxonomic de- scription, ullrastructure. J. Phycol. 5: 341-350. kinzie, R. A., III. 1974. Experimental infection of aposymbiotic gorgo- nian polyps with zooxanthellae. J. Exp. Mar. Biol. Ecol. 15: 333-345. Kohler. G., and C. Milstein. 1975. Continuous cultures of fused cells secreting antibody of predefined specificity. Nature 256: 495-497. Lee, J. J. 1998. "Living sands" — larger forammifera and their endosym- biotic algae. Symbiosis 25: 71-100. Lee, J. J., and O. R. Anderson. 1991. Symbiosis in foraminifera. Pp. 157-220 in Biology ofFoiaminifcra. J. J. Lee and O. R. Anderson, eds. Academic Press, London. Lee, J. J., Y. Chan, and A. Lagziel. 1988. An immunofluorescence approach toward the identification of endosymbiolic diatoms in several species of larger formaminifera. Pp. 230-241 in Immunocliemical Approaches to Coastal, Estuarine, and Oceanographic Questions, C. M. Yentsch, F. C. Mague. and P. K. Horan, eds. Springer- Verlag, New York. Lee, J. J., J. Morales, J. Chai, C. Wray, and R. Rottger. 1997. Progress in the studies of endosymbiotic algae from larger foraminif- era. Pp. 329-344 in Eulairyntism ami Symbiosis, H. E. A. Schenk. R. G. Herrman, K. W. Jeon, N. E. Muller, and W. Schwemmler. eds. Springer-Verlag, Heidelberg. Lerouge, P., P. Roche, C. Faucher, F. Maillet, G. Truchet, J. C. Prome, and J. Denaire. 1990. Symbiotic host-specificity of Rhizobium me- liloti is determined by a sulfated and acylated glucosamine oligosac- charide signal. Nature 344: 781-784. Markell, D. A., and R. K. Trench. 1993. Macromolecules exuded by symbiotic dinoflagellates in culture: amino acid and sugar composition. J. Phycol. 29: 64-68. Markell, D. A., R. K. Trench, and R. Iglesias-Prieto. 1992. Macro- molecules associated with the cell walls of symbiotic dinoflagellates. Symbiosis 12: 19-31. McNally, K. L., N. S. Govind, P. E. Thome, and R. K. Trench. 1994. Small-subunit ribosomal DNA sequence analyses and a reconstruction of the inferred phylogeny among symbiotic dinoflagellates (Pyrro- phyta). J. P/iycol. 30: 316-329. Muller-Parker, G., K. \V. Lee, and C. B. Cook. 1996. Changes in the ultrastructure of symbiotic zooxanthellae (Symbiodinium sp., Dino- phyceae) in fed and starved sea anemones maintained under high and low light. J. Phycol. 32: 987-994. Palincsar, J. S., W. R. Jones, and E. E. Palincsar. 1988. Effects of isolation of the endosymbiont Symbiodinium microadriaticum {Dino- phyceae) from its host Aiptasia pallida (Anthozoa) on cell wall ultra- structure and mitotic rates. Trans. Am. Microsc. Soc. 107: 53-66. Perkins, M. 1989. Erythrocyte invasion by the malarial merozoite: re- cent advances. Exp. Parasitol. 69: 94-99. Rands, M. L., B. C. Loughman, and A. E. Douglas. 1993. The sym- biotic interface in an alga-invertebrate symbiosis. Proc. R. Soc. Land. B 253: 161-165. Regensburg-Tuink, A. J. G., and P. J. J. Hooykaas. 1993. Transgenic N. glauca plants expressing bacterial virulence gene virF are con- verted into hosts for nopaline strains of A. tumefaciens. Nature 363: 69-71. Roth, K. E.. K. Jeon, and G. Stacey. 1988. Homology in endosymbiotic systems: the term "Symbiosome." Pp. 220-225 in Molecular Genetics of Plant-Microbe Interactions, R. Palacios and D. P. S. Verma. eds. APS Press, St. Paul. MN. Rowan, R. 1991. Molecular systematics of symbiotic algae. J. Phycol. 27: 661-666. Rowan, R. 1998. Diversity and ecology of zooxanthellae on coral reefs. J. Phycol. 34: 407-417. Rowan, R., and D. A. Powers. 1991. A molecular genetic classification of zooxanthellae and the evolution of animal-algal symbioses. Science 251: 1348-1351. Russell, D. G., S. Xu, and P. Chakraborty. 1992. Intracellular traffick- ing and the parasitophorous vacuole of Leishmania mexicana-'mfected macrophages. J. Cell Sci. 103: 1193-1210. Sam-Yellowe, T. Y. 1992. Molecular factors responsible for host cell recognition and invasion in PUismodium falciparum. J. Protozool. 39: 181-189. Sanchez. F.. J. E. Padilla, H. Perez, and M. Lara. 1991. Control of nodulin genes in root-nodule development and metabolism. Annu. Rev. Plant Physiol. 42: 507-528. Schoenberg. D. A., and R. K. Trench. 1980a. Genetic variation in Symbiodmuim (=Gymnodinium) microadriaticum Freudenthal. and specificity in its symbiosis with marine invertebrates. I. Isoenzyme and soluble protein patterns of axenic cultures of S\mbiodinium microad- riaticum. Proc. R. Soc. Lond. B 207: 405-427. Schoenberg, D. A., and R. K. Trench. 1980b. Genetic variation in Symbiodinium ( = Gymnodinium) microadriaticum Freudenthal, and specificity in its symbiosis with marine invertebrates. II Morphological variation in Symbiodinium microadriaticum. Proc. R. Soc. Lond. B 207: 429-444. ORIGIN OF THE SYMB1OSOME MEMBRANE 14' Schwarz, J. A., D. A. Krupp, and V. M. Weis. 1999. Late larval development and onset of symbiosis in the scleractinian coral Funt>iu scutaria. Biol. Bull. 196: 70-79. Sibley, L. D., and J. L. Krahenbuhl. 1988. Modification of host cell phagosomes by Toxoplasma gondii involves redistribution of sur- face proteins and secretion of a 32 kDa prolein. Eur. J. Cell Biol. 47: 81-87. Sibley, L. D., J. L. Krahenbuhl, G. M. W. Adams, and E. Weidner. 1986. Toxoplasma modifies macrophage phagosomes by secretion of a vesicular network rich in surface proteins. J. Cell Biol. 103: 867-874. Taylor, D. L. 1968. In situ studies on the cytochemistry and ultrastruc- ture of a symbiotic marine dinofiagellate. J. Mar. Biol. Assoc. UK 48: 34Q-366. Taylor, D. L. 1971. infrastructure of the 'zooxanthella' Endodinium chattonii in situ. J. Mar. Biol. Assoc. UK 51: 227-234. Thinh. L. V., D. J. Griffiths, and H. Winsor. 1986. Ultrastructure of Symbiodinium microadriaticum (Dmophyceae) symbiotic with Zoan- rluts sp. (Zoanthidea). Phycologia 25: 178-184. Trench, R. K. 1993. Microalgal-invertebrate symbioses: a review. En- Jocytobiosis Cell Res. 9: 135-175. Trench, R. K., and R. J. Blank. 1987. Symbiodinium microadriuticum Freudenthal, S. goreauii, sp. nov., S. kawagulii. sp. nov.. and 5. />//«- sum. sp. nov.: gymnodinioid dinofiagellate symbionts of marine inver- tebrates. J. Pliycol. 23: 469-481. Trench, R. K., and H. Winsor. 1987. Symbiosis with dinoflagellates in two pelagic flat worms Amphiscolops sp. and Haplodiscus sp. Symbi- osis 3: 1-22. Tripodi, G., and S. Santisi. 1982. A study on the cell covering of Symbiodinium, a symbiote of the octocoral Eimicella. J. Siibmicroic. Cytol. 14: 613-620. Udvardi, M. K., and D. A. Day. 1997. Metabolite transport across symbiotic membranes of legume nodules. Anna. Rev. Plant Pli\.\ial. Plant Mol. Biol. 48: 493-523. Wakefield, T. S., S. C. Kempf, and M. A. Farmer. 1998. The contri- bution of Symbiodinium microadriaticum to the symbiosome mem- brane in the anthozoan host Aiptasia pullidti. Microsc. Microanal. 4: 1182-1183. Wakefield, T. S.. M. A. Farmer, and S. C. Kempf. 2000. Revised description of the fine structure of in situ "zooxanthellae" genus Svm- biodinium. Biol. Bull. 199: 76-84. INVERTEBRATE SENSORY INFORMATION PROCESSING: Implications for Biologically Inspired Autonomous Systems Proceedings of a workshop sponsored by THE CENTER FOR ADVANCED STUDIES IN THE SPACE LIFE SCIENCES AT THE MBL 15 to 17 June 2000 J. Erik Jonsson Center for the National Academy of Sciences, Woods Hole, Massachusetts Funded by THE NATIONAL AERONAUTICS AND SPACE ADMINISTRATION under Cooperative Agreement NCC 2-896. Reprints of this symposium can be obtained from the Center for Advanced Studies in the Space Life Sciences, 7 MBL Street, Woods Hole, MA 02543. CONTENTS Invertebrate Sensoiy Information Processing: Implications for Biologically Inspired Autonomous Systems Blazis, Diana E. J., and Frank W. Grasso Introduction 147 Breithaupt, Thomas Fan organs of crayfish enhance chemical information flow 150 Fraser, Peter J. Statocysts in crabs: short-term control of locomotion and long-term monitoring of hydrostatic pressure . . 155 Grasso, Frank W. Invertebrate-inspired sensory-motor systems and au- tonomous, olfactory-guided exploration 160 Barlow, Robert B., James M. Hitt, and Frederick A. Dodge Limulus vision in the marine environment 169 Cronin, Thomas W., and Justin Marshall Parallel processing and image analysis in the eyes of mantis shrimps 177 Webb, Barbara View from the boundary 184 Robert, Daniel Innovative biomechanics for directional hearing in small flies HIO Schmitz, Josef, Jeffrey Dean, Thomas Kindermann, Michael Schumm, and Hoik Cruse A biologically inspired controller for hexapod walking: simple solutions by exploiting physical properties. . . . 195 Macmillan, David L., and Blair W. Patullo Insights for robotic design from studies of the con- trol of abdominal position in crayfish 201 Birmingham, J. T. Increasing sensor flexibility through neuromodula- tion 206 Derby, Charles D., and Pascal Steullet Why do animals have so many receptors? The role of multiple chemosensors in animal perception 211 Srinivasan, Mandyam V., Shaowu Zhang, and Javaan S. Chahl Landing strategies in honeybees and possible appli- cations to autonomous airborne vehicles 216 Ishida, H., T. Nakamoto, T. Moriizumi, T. Kikas, and J. Janata Plume-tracking robots: a new application of chemical sensors 222 Mountain, David C., and Allyn E. Hubbard Sensing scenes with silicon 227 Higgins, Charles M. Sensory architectures for biologically inspired auton- omous robotics 235 LIST OF PARTK:IPA.\TS . 243 146 Reference: Biol. Bull. 200: 147-149. (April 2nii| i Introduction DIANA E. J. BLAZIS1 AND FRANK W. GRASSO; < lThe Center for Advanced Studies in the Space Life Sciences, Marine Biological Laboratory, Woods Hole, Massachusetts: and Boston University Marine Program, Marine Biological Laboratory. Woods Hole. Massachusetts 02453 This workshop — entitled "Invertebrate Sensory Informa- tion Processing: Implications for Biologically Inspired Au- tonomous Systems" — was sponsored by the Center for Ad- vanced Studies in the Space Life Sciences at the Marine Biological Laboratory. At this meeting, scientific leaders in the fields of invertebrate sensory biology and sensory- guided behavior were brought together with biologists and engineers who develop robotic systems that are based on biologically inspired algorithms and architectures. The par- ticipants discussed how moderately complex metazoan or- ganisms (e.g., insects, crustaceans, and cephalopods) pro- cess the sensory information that is required to perform natural tasks in a changeable, dynamic environment. They also critiqued the extant, man-made autonomous systems that were built to emulate various invertebrate abilities (e.g., fly in-flight control of air-speed for landing or maintaining course: crustacean ability to track a turbulent odor plume to its source; sensory compensation for lost appendages in walking insects: cricket and parasitic fly recognition; and tracking of a specific acoustic signature in a noisy back- ground). The participants considered the gaps in our knowl- edge about biological sensory information processing and the limitations in our present technology, both of which limit the transfer of biological competence to man-made autonomous systems. They also identified and debated es- tablished and speculative biological concepts and technol- ogies that will facilitate the development of autonomous systems with invertebrate levels of capabilities. The proceedings of this workshop comprise a substantial subset of the presentations delivered at the meeting, and thus reflect the major themes that emerged during those discussions. These themes are highlighted in the following summary of the papers making up the proceedings. * Current address: Department of Psychology. Brooklyn College. CUNY, 2900 Bedford Ave.. Brooklyn. New York 1 12 Ml. The variety of modalities of information processing that are used by invertebrates are well represented in the pro- ceedings. In many cases, these modalities may be useful in the exploration of novel environments. For example, inver- tebrates use polarization and multi-spectral imaging (Cronin and Marshall, 2001). Invertebrate chemoreception includes examples of the identification and classification of complex odors via a range of bimodal sensors (Derby and Steullet. 2001). Tunable sensors are used in acoustic processing to expand the available range, while filtering out background noise and locking onto important or new signals (Mountain and Hubbard. 2001; Robert. 2001; Webb. 2001). Under- standing these forms of information processing and their roles in certain environmental niches can illuminate our knowledge of sensory biology and expand our options for sensory processing in autonomous systems. Many invertebrate systems are robust across a wide range of conditions and make efficient use of limited material and energy. This robustness results, in part, from close coupling between sensory and motor systems: vision (Srinivasan et nl.. 2001; Barlow et ai, 2001); olfaction (Derby and Steul- let. 2001; Grasso. 2001): audition (Mountain and Hubbard. 2001 : Robert, 2001 ); vestibular senses (Fraser, 2001 ); rheo- sense (Breithaupt. 2001). Natural organisms seem not to calculate exact solutions to equations: rather they perform tasks only approximately — just exactly enough to sur- vive— as stick insects do when they target the posterior leg onto the position of the anterior leg during walking (Schmitz et til.. 2001 ). Elegant studies of visual guidance for fly landing provide a powerful example of this coupling (Srinivasan et «/.. 2001): simple mechanisms suffice be- cause the motor systems provide a predictable sampling context. Examples like this clearly show that, as the func- tional origins of robust invertebrate behavior in dynamic environments are better understood, valuable insights about 147 148 INTRODUCTION the engineering of autonomous systems that can explore unfamiliar or variable environments will be provided. Invertebrate sensory systems are flexible, and dynamism is built into every processing stage. Contrary to traditional neurobiological dogma, we find plasticity present in abun- dance in invertebrate systems. Sensory receptors are plastic and show adaptation (Barlow et al., 2001; Eraser. 2001: Mountain and Hubbard. 2001: Robert, 2001): central pro- cesses, including learning and memory, augment sensory processing, and the animals themselves show behavioral plasticity (Macmillan and Patullo, 2001; Breithaupt, 2001). Changes in sensory networks have been interpreted as in- stances of a system adapting to ("discovering") causal re- lationships in the environment (Birmingham, 2001; Schmitz ft nl., 2001). Sensory systems can adapt to the statistics of natural scenes, as noted by Breithaupt in crayfish. Indeed, the ability to express these levels of dynamic processing means that an animal is capable of engineering new circuits "on the fly." Redundancy is a feature of biological sensory systems that assures the attainability of a particular goal. During olfactory learning in spiny lobsters, for example, discrimi- nation, local searching, and distance orientation are carried out by two separate pathways, as described by Derby and Steullet (2001). But apparent redundancy can, in fact, dis- guise an organization that produces deeper computational and energetic savings (Higgins. 2001). In general, invertebrates make greater use than verte- brates of multi-modal, multifunctional sensor arrays. Thus, statocysts are used for angular and linear acceleration, vi- bration, and hydrostatic pressure sensing (Fraser, 2001). Birmingham described the stretch receptor in the crustacean foregut that functions at different times, either as a strain gauge or a volume sensor. Crustacean antennules combine information from flow and chemical sensors located on the same anatomical structure, and can therefore make spatial and temporal correlations across these two modalities (Breithaupt. 2001; Grasso. 2001; Derby and Steullet, 2001). This ability to multitask sensory inputs has implications for the design and implementation of light, efficient systems. Our understanding of sensory systems and our develop- ment of useful autonomous systems depend upon having the best information possible about the to-be-explored environ- ment, as well as stimulus dynamics in those environments. Understanding the stimulus environment speaks to the "de- sign" of sensory receptors and processors and clarifies the evolutionary pressures that shape both the sensory biology and the behavioral choices of an organism. Several inves- tigators, particularly those examining chemosensory behav- iors (e.g., Breithaupt, 2001; Grasso. 2001: and Ishida et iiiiiiis Systems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences. Woods Hole, Massachusetts. I'nini 15-17 April 2000. was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded hy the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. nutrient uptake by and waste removal from living organ- isms, the acquisition of information that serves to modify the behavior of organisms with respect to environmental resources (e.g., water, food, mates) and threats (e.g., pred- ators, obstacles) is important for any living organism as well as for autonomous robots. The oldest method of information exchange is by means of chemical stimuli. Microorganisms (e.g., motile bacteria) need to search for micropatches of higher nutrient concen- tration and adjust their motility with respect to chemical stimuli (Blackburn et al, 1998). Microorganisms also ex- change chemical signals ("pheromones") with conspecifics to synchronize their mating activity (Agosta, 1992). The movement of chemicals from a source to a receiver at such small size scales (below 0.1 mm. Reynolds number below 10 2) is driven by molecular diffusion, or random molecu- lar motion. Cohesion among water molecules at that scale makes it almost impossible to influence the direction ot chemical information flow except by locomotion ot the whole organism. With increasing size and velocity of organisms (increas- ing Reynolds number) molecular diffusion loses in impor- tance for the dispersal of chemicals (Weissburg, 2000). In water, molecular diffusion is a very slow process that on average would displace a molecule by no more than 0.7 mm (70 mm in air) in 1 min and by only 5 mm (500 mm in air) in 1 h (Dusenbery. 1992). Water flow is much more effec- tive in dispersing odor molecules, and animals can influence the flow in their immediate neighborhood by using undu- lating, beating, or fanning appendages. By modulating the nearby flow pattern, they can facilitate the exchange ot chemical information with their environment. Insects as well as some vertebrates use wing-fanning for delivering pheromones to their mates (bees: Free, 1987; butterflies: Boppre, 1984; bats: Voigt and von Helversen, 1999). Wing- fanning may facilitate odor perception by drawing nearby 150 CRAYFISH FANNING BEHAVIOR I antennule nephropores fan organs Figure 1. (A) Location of fan organs, urinary pores (nephropores), and major chemoreceptor organs (antennules) of crayfish. (B) The fan organs are multi-segmental flagella of the mouthparts (maxillipeds) with feathered setae on the distal part. During the power stroke (scanning electron microscopy "SEM" picture of Procambams clarkii) the feathered hairs are extended. (C) During the recovery stroke (SEM picture of Procambarus clarkii) the feathered hairs are tilted downstream. White scale bars in SEM pictures represent 1 mm. molecules towards the receptors. Wing-fanning of silkworm moths in response to pheromone (Agosta, 1992) may aid in stimulus acquisition (Ishida et ai, 1996). Crustaceans are well known for their ability to create directed water currents by pumping or fanning appendages (Brock, 1926; Burrows and Willows, 1969; Budd et al., 1979: Koehl and Strickler, 1981; Atema, 1985; Lavalli and Factor, 1995). This behav- ior can be used for gill ventilation, locomotion, suspension feeding, and chemoreception or chemical signaling. Plank- tonic copepods, small crustaceans of 1 to 10 mm, generate water currents that are used for capturing food particles and that carry odor information from the food particles to the chemoreceptors (Yen, 2000). Here I describe the fanning behavior of crayfish, typical of larger crustaceans, that is used for both sending and receiving chemical signals in aquatic environments with stagnant flow conditions. Maxilliped Flagella, the Fan Organs of Crayfish One of the first reactions of many crustaceans to chemical stimulation is the onset of the rhythmic beating of three pairs of flagella of the maxillipeds (mouthparts). These so-called fan organs (Atema, 1985) in crayfish are distrib- uted around the mouth opening below the major chemore- ceptor organs ("antennules") and the urinary pore ("nephro- pore") (Fig. 1A). Each fan organ consists of a multi-seg- mental, flattened stem. It is feathered in one plane, with setae emerging laterally on both sides of the flattened stem (Fig. IB). The setae are feathered themselves, and their rami overlap with those of adjacent setae to form a dense layer (Fig. IB). During the power stroke, the fan — with stem erect and setae extended — acts like a paddle propelling fluid past the organism (Fig. IB; Cheer and Koehl, 1987). During the recovery stroke, the setae are tilted downstream and the stem is flexed (Fig. 1C). This behavior reduces the area of the leading, water-displacing surface and therefore effec- tively reduces the drag during the return stroke. This fan- ning activity is similar to that of a rowing paddle and can generate unidirectional flow. All three flagella on one side usually beat simultaneously and with the same frequency (between 5 and 8 Hz in Pmciimbarus clarkii). The rhythmic movements of the three flagella are metachronally phase- shifted by about 0.2 to 0.5 wavelengths (data from high- speed video-recordings; Breithaupt, unpubl. data). By changing the angle of the mouthparts. the flagella can mod- ify the axis of fanning activity. By fanning only one side or 152 T. BREITHAUPT Figure 2. Examples of horizontal flow fields generated by the fan organs of an adult male crayfish (Pi-ociimbunix tUirkii) without claws. Flow pictures were generated by superimposing 25 video-frames (Breit- haupt and Ayers. 1998) of neutrally buoyant particles moving in a 5-mm-thick horizontal light sheet, and tracing the flow lines. (A) Fans of both sides beat backward and generate bilateral backward-directed water jets (lighl sheet at 1 cm above ground lasting for 2s). Large arrows indicate fluid velocities of 1 em/s; small arrows represent velocities of 0.5 cm/s. (B) Fans of left side only are beating forward and produce a forward-directed jet (at 3.5 cm for 2 s; arrows as in A). (C) Fans of both sides beat upward and in the horizontal plane draw water from all directions towards the frontal mouthparts (at 1 cm for 10 s). Arrows denote fluid velocities of 0.1 cm/s. (D) Fans of right side only beat upward and draw water from right to left (at 2.5 cm for 12 s; arrows as in CK both sides and by changing the direction of fanning activity, crayfish can generate a variety of flow fields. Flow Fields Generated by the Fan Organs Visualization of the flow fields using small (50 /xm). neutrally buoyant particles illuminated within a horizontal and a vertical light sheet (Breithaupt and Ayers, 1998) revealed that crayfish can modulate the flow environment in their immediate vicinity in a variety of ways (Fig. 2). In the horizontal plane that is probably most significant for most of the crayfish's activities, the flow can be directed either away from (Fig. 2A, B) or toward (Fig. 2C. D) the head region. Flow velocity of the outgoing water jets can be up to 4 cm/s. Incoming water in the horizontal plane is usually much slower, with velocities between 0.1 and 0.5 cm/s depending on the distance to the fan organs. By beating the flagella on only one side, crayfish can draw in water exclusively from this side (Fig. 2D). Biological Significance of Fanning Behavior Many crayfish species live in environments with stagnant flow conditions. The flow velocity in a lake inhabited by the European narrow-clawed crayfish (Astacns leptodactylus) rarely exceeds 1 cm/s even on stormy days (T. Breithaupt CRAYFISH FANNING BEHAVIOR I. and E. Ebert. unpubl. data). In such environments, flow fields created by the animals themselves help in getting access to distant odor stimuli and provide directional infor- mation. Fanning is a major activity of crayfish during food search in stagnant water. Blindfolded crayfish (Astacus lep- todactylus) with restrained fan organs that could not gener- ate water currents were not able to find the odor source (T. Breithaupt and E. Ebert, unpubl. data). This demonstrates the importance of the fan organs in actively scanning the environment and transporting the odor molecules towards the chemoreceptors on the antennules (Fig. 1 ). The crayfish may identify the direction of the odor source by comparing the arrival time of the moving odor patches at different receptor locations on the bilateral antennules. Recent stud- ies suggested bilateral directional information as a critical element in chemical orientation of decapod crustaceans at turbulent flow conditions (Atema, 1996; Grasso et al., 1998). A second important function of the fan organs is to carry urine-borne pheromone from the nephropores to the chemo- receptors of nearby conspecifics. Visualization of urine re- lease during fights of American lobsters and crayfish re- vealed that signaling with urine is an integral component of aggressive communication in these animals (Breithaupt and Atema, 2000; T. Breithaupt and P. Rohleder, unpubl. data). The crayfish Procambarus clarkii uses frontally projecting fan organ currents to carry urine signals toward the oppo- nent (T. Breithaupt and P. Rohleder, unpubl. data). The currents are also used to direct urine signals in directions other than forward. Lessons to Learn From the Fan Organs of Crayfish Only by using such elaborate appendages can crustaceans acquire and send chemical information in environments with stagnant flow conditions. Fan organs are widely used in the animal kingdom to propel and direct chemical stimuli away from and toward organisms. Flow-generating devices can also greatly enhance the search success of autonomous robots orienting to chemical sources. The use of a rotating propeller fan has been shown to help terrestrial autonomous robots to locate chemical sources by actively drawing air to their chemical sensor and scanning different directions for the presence of chemicals (Nakamoto et al.. 1999; Ishida et al., 2001). Recent progress in the development of aquatic robots has been inspired by research on crustaceans (Grasso, 2001). The versatility of the paddlelike fan design of the crayfish appendage goes beyond that of propeller fans by enabling the exchange of chemical information in three dimensions. A thorough analysis of the behavior and the mechanics of flow production as well as the underlying muscular and neural activity should further illuminate the functional organization of this beautiful system and may trigger biomimetic designs for artificial fans. Acknowledgments Thanks to Frank Grasso for organizing the inter-disci- plinary symposium. I would also like to thank Myriam Schmid and Dr. Joachim Hentschel for assistance in the SEM study of fan organs, the graphics workshop of the University of Konstanz for the crayfish drawing in Fig. 1, Kirsten Pohlmann and two anonymous reviewers for helpful comments on the manuscript, and Prof. Dr. Axel Meyer for supporting the study. The research was funded by the Deut- sche Forschungsgemeinschaft (Br 1321/3-1). Literature Cited Agosta, W. C. 1992. Chemical Communication: the Language of Pher- omones. Scientific American Library, New York. Atema, J. 1985. Chemoreception in the sea: adaptations of chemorecep- tors and behaviour to aquatic stimulus conditions. Soc. E.\p. Biol. Symp. 39: 386-423. Atema, J. 1996. Eddy chemotaxis and odor landscapes: exploration of nature with animal sensors. Biol. Bull. 191: 129-138. Blackburn. N., T. Fenchel, and J. Mitchell. 1998. Microscale nutrient patches in planktonic habitats shown by chemotactic bacteria. Science 282: 2254-2256. Boppre, M. 1984. Chemically mediated interactions between butterflies. Pp. 259-275 in The Biology of Butterflies. R. I. Vane-Wright and P. R. Ackery, eds. Academic Press, London. Breithaupt, T., and J. Atema. 2000. The timing of chemical signaling with urine in dominance fights of male lobsters (Homarus americanus). Behav. Ecol. Sociobiol. 49: 67-78. Breithaupt, T., and J. Ayers. 1998. Visualization and quantification of biological flow fields through video-based digital motion-analysis tech- niques. Mar. Freshw. Behav. Physiol. 31: 55-61. Brock, F. 1926. Das Verhalten des Einsiedlerkrebses Paguriis arrosor Herbst wahrend der Suche und Aufnahme der Nahrung. 2. Morph. Oekol. Tiered: 415-552. Budd, T. W., J. C. Lewis, and M. L. Tracey. 1979. Filtration feeding in Orconectes propinquity and Cambarus robustus (Decapoda, Cam- baridae). Crustaceans Suppl. 5: 131-134. Burrows, M., and A. O. D. Willows. 1969. Neuronal co-ordination of rhythmic maxilliped beating in brachyuran and anomuran Crustacea. Comp. Biochem. Physiol. 31: 121-135. Cheer, A. Y. L., and M. A. R. Koehl. 1987. Paddles and rakes: fluid flow through bristled appendages of small organisms. J. Theor. Biol. 129: 17-39. Dusenbery, D. B. 1992. Sensory Ecology: How Organisms Acquire and Respond to Information. W. H. Freeman. New York. Free, J. B. 1987. Pheromones of Social Bees. Chapman and Hall, Lon- don. Grasso, Frank W. 2001. Invertebrate inspired sensory-motor systems and autonomous, olfactory-guided exploration. Biol. Bull- 200: 160-168. Grasso, F. W., J. A. Basil, and J. Atema. 1998. Toward the conver- gence: robot and lobster perspectives of tracking odors to their source in the turbulent marine environment. Pp. 259-263 in Proceedings of the 199S IEEE. Gaithersburg. MD. Ishida, H., Y. Kagawa, T. Nakamoto, and T. Moriizumi. 1996. Odor source localization in the clean room by an autonomous mobile sensing system. Sens. Actuutors B 33: I 15-121. 154 T BREITHAUPT Ishida, H., T. Nakamoto, T. Moriizumi, T. Kikas, and J. Janata. 2001. Nakamoto, T., H. Ishida, and T. Moriizumi. 1999. A sensing system Chemical sensors for plume tracing robots. Biol. Bull. 200: for odor plumes. Anal. Chem. 71: 531A-537A. 222-226. Voigt, C. C., and O. von Helversen. 1999. Storage and display of odour Koehl, M. A. R., and J. R. Strickler. 1981. Copepod feeding currents: by male Saccopteryx bilineata (Chiroptera, Emballonundae). Behav. food capture at low Reynolds number. Limnol. Oceanogr. 26: 1062- Ecol. Sociobiol 47: 29-40. 1073. Weissburg, M. J. 2000. The fluid dynamical context of chemosensory Lavalli, K. L., and J. R. Factor. 1995. The feeding appendages. Pp behavior. Biol Bull. 198: 188-202. 349-393 in Biology of the Lobster Homarus americanus. J. R. Factor. Yen, J. 2000. Life in transition: balancing inertial and viscous forces by ed. Academic Press, Orlando. FL. planktonic copepods. Biol. Bull. 198: 213-224. Reference: Biol. Bull. 200: 155-159. (April 2001) Statocysts in Crabs: Short-Term C<» ol of Locomotion and Long-Term Monitoring of Hydrostatic Pressure PETER J. ERASER Zoology Department, Aberdeen University, Tilly-drone Avenue. Aberdeen AB24 2TZ Scotland Abstract. Crabs show well-coordinated locomotion. They have proprioceptors similar to those of lobsters, but they differ in terms of their balancing systems and their con- densed nervous system, which allows rapid interganglionic conduction. Typically they exhibit dynamically stable loco- motion with a highly developed semicircular canal system that codes angular acceleration in each of three orthogonal planes (horizontal and vertical at 45° and 135° to the pitch- ing plane). Left and right interneurons each code one direc- tion of angular acceleration, carrying information between the brain and the thoracic ganglia. Cell A codes head-up vertical plane angular accelerations. Cell B codes rotations in the horizontal plane. Interneurons C and D code head- down vertical plane information, carrying it ipsilaterally and contralaterally respectively. These interneurons have a cen- tral role in locomotion. They are activated and have their responsiveness to angular acceleration enhanced before and during locomotion. Such simple activation pathways point to how an angular-acceleration-controlled robot (CRABOT) could be constructed. Hydrostatic pressure information car- ried by the thread hairs, which also sense angular acceler- ation, is filtered out from direct pathways onto the interneu- rons, but spectral analysis shows that it still has an influence via central pathways. Long-term recordings from equilib- rium interneurons in free-walking crabs taken from the wild into constant conditions show tidally changing frequencies E-mail: p.fraser@abdn.ac.uk This paper was originally presented at a workshop titled Invertebrate Sensor,- Information Processing: Implications for Biologically Inspired Autonomous Systems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences, Woods Hole. Massachusetts, from 15-17 April 2000. was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. of bouts of activity, with peaks that correspond to times of high tide. Introduction Brachyuran crabs are conspicuous animals that occupy a variety of habitats and show well-coordinated locomotor ability while walking, crawling, climbing, swimming, or burrowing under a variety of hydrostatic pressure condi- tions. Compared with the ladder-like nerve cord of macru- rans (crayfish and lobsters), their nervous system shows condensation and fusion of thoracic ganglia, allowing rapid interganglionic conduction of impulses. The abdomen has been much reduced and, in contrast to that of macrurans, is little used in normal or escape swimming. These crabs can survive leg autotomy: 13%-43% of individual Carcimis sampled from the wild have missing legs (McVean, 1982). Typically they exhibit dynamically stable locomotion, with statocyst (or balancing organ) interneurons having a central role (Fig. 1; Eraser, 1982; Eraser et al., 1987). Thread hair receptors, which respond to fluid displacements in the sta- tocyst, have been shown to have their spontaneous activity and sensitivity to angular acceleration modulated by small changes in hydrostatic pressure (Eraser and Macdonald. 1994; Macdonald and Eraser, 1999). Clearly, elucidating the simplifying principles underlying such versatility and diver- sification of function is an important aim. One approach to dissecting such complexity is to concentrate on a fundamen- tal difference between crabs and lobsters — that is, the elab- oration in crabs of the statocyst and its central pathways. Statocysts and Leg Proprioceptors Involved in Locomotion The statocyst contains a statolith and angular acceleration receptors. In the crab, the crescent of statolith hairs (Sand- 155 156 P. J. ERASER A Thread Hairs *Esophageal Connective Fused Thoracic Ganglia Figure 1. Equilibrium cell A in brain (A) and thoracic ganglia (B) of the crab Carcinus maenas. The large axon runs in the esophageal connec- tive. Thread hair afferents (C), which are activated by angular acceleration in the plane of the vertical canal of the statocyst, contact the equilibrium interneurons directly. (Redrawn from Fraser, 1989). eman and Okajima. 1972) is much smaller than the macru- ran crescent (see Lemmnitz and Wolff, 1990). The fluid- filled sac has been shaped from a simple cavity in the macruran statocyst into two orthogonal semicircular canals. Angular accelerations in the planes of the horizontal or vertical canals are monitored by a row of long, slender thread hairs (Fig. 1C; Sandeman and Okajima. 1972; Fraser and Takahata, 2001 ). Typically the response of thread hairs to an angular acceleration is unidirectional, with two oppo- sitely responding directional classes. Analogous to those of vertebrates, the receptors integrate angular acceleration and code angular velocity in the horizontal plane or in vertical planes either at 45° or at 135° to the median line. Thread hairs are the main sensory component involved in a variety of statocyst-driven behaviors ranging from walking to eye- stalk movement, with free hook hairs and statolith hairs playing a lesser role (Cohen and Dijkgraaf. 1961; Fraser, 1989; Fraser and Takahata, 2001). An inertial angular ac- celeration detector based on an orthogonal semicircular canal system could be easily added to a robot by using connectivity and coding principles derived from crab equi- librium interneurons, which are described below. Crabs normally walk sideways. The leading legs pull and the trailing legs push to move the body (Clarac and Coul- mance, 1971; Evoy and Ayers. 1982). Crabs do not exhibit regular gait patterns, but have a great variety of gaits and leg activation sequences. Hence Evoy and Ayers (1982) state that "changes in gait occur and conditions strictly satisfying the alternating tetrapod gait are rarely seen." During high- speed running, ghost crabs may even become bipedal, with only two legs alternating on the trailing side (Burrows and Hoyle, 1973). Statocyst ablation affects sideways walking and the rearing reflex (Cohen and Dijkgraaf, 1961; Fraser, 1974). Swimming activity is more stereotyped and again highly dependent on statocyst input (Fraser el al., 1987). During swimming, all muscles in the fifth legs are active in a sequence that causes cyclical sculling. Only muscles that operate joints controlling dorsolateral movements are in- volved in sideways walking, and they are used in a different temporal pattern than in swimming (Fraser et al., 1987). Leg proprioceptors are used in the fine motor control to produce coordinated behavior, but during rhythmic behav- ior, sensory information does not vary much on each cycle unless a leg hits an obstacle. A variety of receptors include muscle receptor organs, which measure muscle stretch (Bush, 1977); chordotonal organs, which are articular (joint rotation) receptors measuring stretch around joints; and apodeme receptors and cuticular stress receptors, which record muscular tension via tendons and cuticle respectively (Wales et al., 1971 ). The funnel canal organs at the ends of the dactylopodites must be silent for the animal to swim, whereas their rhythmic firing in walking is used for step regulation (Libersat et al., 1987). Although it is not possible here to do justice to the complexity of information relayed by these proprioceptors, in general their role in locomotion in crabs is broadly similar to that in lobsters and crayfish. Equilibrium Interneurons Involved in Locomotion The angular acceleration sensors, the thread hairs, project onto a small set of eight large interneurons that link the brain to the fused thoracic and abdominal ganglia (Fig. 1; Fraser, 1974, 1990). These fire in the short term during angular accelerations caused by locomotion (Fig. 2A, B; Fraser. 1982). but are also activated centrally before and during bouts of walking or swimming (Fig. 2B; Fraser et al., 1987; Fraser, 1975. 1989). Their activity long term consists of quiescent periods with short bouts of extremely elevated activity. Cells are not coupled and are more likely to act in anti-phase during short-term oscillations that form part of the locomotion. Over a longer time scale, on the order of minutes, their overall activity is well synchronized. Vector information is hence preserved (Fraser et al., 1997. 2001). During these bouts of firing, the gain in the statocyst path- ways alters greatly so that perturbations of angular acceler- ation have a large effect (Fig. 3). In the context of discussing the role of the interneurons in STATOCYSTS AND CRAB LOCOMOTION 157 Left Walking 1 2 3 Time in Seconds A Cell A, Left Connective Figure 2. (A) Extension phases of left and right 2nd to 5th legs during a short sequence of walking to the left by a crab. The extracellular recording shows cell A in the left esophageal connective. The series of bursts of activity are more related to slight angular excursions of the body () than to the gait pattern. is the clockwise angular displacement of the body of the crab around the horizontal longitudinal axis. (B) The effect of cutting the right esophageal connective on locomotion in the crab Carcinus maenas. Gait alters and the crab prefers walking left. Regardless of the direction of walking, the crab always turns counterclockwise (Bethe, 1897). This is the direction that optimally excites Cell B in the right connective. Electrical stimulation of the cut end of the connective will straighten or reverse the counterclockwise movement. locomotion. Fraser (1982) points out that "where a behavior involves vectorial output in more than one dimension, or where feedback loops are involved, then the output mea- sured by displacement of appendages or by muscle activity is no longer an adequate measure of the behavior ... in terms of understanding the underlying neural activity, loco- motion in crabs may be better described in terms of com- ponents of force and torque in the planes used by the equilibrium cells rather than in terms of the gaits em- ployed." The angular displacements of crabs have been monitored in orthogonal planes with a miniature triaxial accelerometer. Although proprioceptor input affects the activity of the interneurons, their outputs in free-walking animals are not particularly complicated (Fraser, unpubl. data; Fraser. 1995). Interestingly, Cruse et al. (1998) and Schmitz (2001 ) state that in the stick insect there is little evidence for a central pattern generator. Instead sensory information de- rived from leg position and ground contact, together with six different coupling mechanisms, allows selection be- tween swing networks and stance networks. Lobster walk- ing has been successfully modeled with five main classes of components: central pattern generators, command systems, coordinating systems, proprioceptive and exteroceptive sen- sors, and phase- and amplitude-modulating sensory feed- back (Ayers et al., 1998). Stepping rates for such models are low. It may be that the crab has more similarities to the stick insect model than to the lobster model in terms of its reliance on sensory control, and an orthogonal command set based on the equilibrium interneurons may be sufficient to control a large part of its locomotion. The finding that the statocyst thread hair receptors alter their activity under the influence of hydrostatic pressure makes understanding of their role more complicated (Fraser and Macdonald, 1994; Fraser et al., 1996; Macdonald and Fraser, 1999). Positive-going and negative-going modula- tion of thread hair spike frequency is thought to arise from activation of the two directional classes of mechanorecep- tors via the linking chorda because of the slight volume changes associated with differential compressibility of the cuticle and other tissue components. Although thread hairs respond directly to alterations in hydrostatic pressure, thread hair interneurons do not seem to do so. Tidal period rhythms of firing pattern in equilibrium interneurons occur and match those found for general locomotor activity (Nay- lor and Atkinson, 1972; Fraser and Takahata. 2001). These are affected by hydrostatic pressure pulses or cycles, and spectral analyses of long sequences show that some of the 158 P. J. ERASER 0 0 10 15 20 25 Time (hours) Figure 3. Long-term recording from equilibrium interneurons (spike frequency measured every 100 s) in a free-walking crab. Can inns maenas, taken into constant conditions in an aquarium after capture at low tide. The recording starts 6 h after capture. Arrows mark the occurrence of high tides. High bouts of activity in the cells correspond to loccmotor behavior in the crab, and there is clear tidal rhythmicity. (Redrawn from Eraser and Takahata, 2001). pressure information is getting through, presumably via higher order inputs (Fraser et at., 2001 ). CRABOTS in the Future Crabs show well-coordinated but highly variable walking patterns and easy transitions between walking and swim- ming. Activation patterns of equilibrium intemeurons dur- ing such movement are simple and should be easily copied. This points the way to how an angular acceleration con- trolled walking robot (CRABOT) could be constructed, which would advance maneuverability of such robots and should save energy. Optimism for the success of this ap- proach comes from early experiments showing circling movements in crabs when one esophageal connective was cut and reversal of this circling behavior with stimulation of the cut connective. This circling may be explained in terms of the effects on equilibrium cells (Fig. 2B; Bethe, 1897; Fraser, 1982). At present, a major limitation regarding re- alistic robotic modeling is the stepping frequency of artifi- cial limbs and activators. Literature Cited Ayers, J., P. Zavracky, N. McGruer, D. P. Massa. W. S. Vorus, R. Mukherjee. and S. Currie. 1998. A modular behavioral-based ar- chitecture for biomimetic autonomous underwater robots. Pp. 15-21 in Proceedings of the Autonomous Vehicles in Mine Countermeasures Symposium, 5-9 April 1998. Naval Postgraduate School. Monterey, CA. Bethe, A. 1897. Das Nervensystem von Carcinus maenas. Ein anato- misch-physiologischer Versuch. I. Theil, II. Mittheil. Arch. Mikrosk. Amu. Eiihvickliingsmech. 50: 589-639. Burrows, M., and G. Hoyle. 1973. The mechanism of rapid running in the ghost crab, Ocypode ceratothalma. J. Exp. Biol. 58: 327-349. Bush, B. M. H. 1977. Non-impulsive afferent coding and stretch reflexes in crabs. Pp. 439-460 in Identified Neurons anil Behavior of Arthro- pods. G. Hoyle. ed. Plenum Press. New York. Clarac, F., and M. Coulmance. 1971. La marche lateral du crabe (Carcinus}: Coordination des movements articulaires et regulation pro- prioceptive. Z. Vgl. Physial. 73: 408-438. Cohen. M. J., and S. Dijkgraaf. 1961. Pp. 65-108 in The Physiology of Crustacea. Vol. II, T. H. Waterman, ed. Academic Press, New York. Cruse, H., T. Kindermunn. M. Schumm, J. Dean, and J. Schmitz. 1998. Walknet — a biologically inspired network to control six-legged walk- ing. Neural Networks 11: 1435-1447. Evoy, \V. H., and J. Ayers. 1982. Locomotion and control of limb STATOCYSTS AND CRAB LOCOMOTION 159 movements. Pp. 61-105 in The Biology of Crustacea. Vol. 4: Neural Integration and Behavior. D. C. Sandeman and H. L. Atwood, eds. Academic Press, New York. Eraser, P. J. 1974. Interneurons in crab connectives (Carcinus maenas (L.)): directional statocyst fibres. J. Exp. Biol. 61: 615-628. Fraser, P. J. 1975. Three classes of input to a semicircular canal inter- neuron in the crab, Scylla serrala and a possible output. J. Comp. Physiol. 104: 261-271. Fraser. P. J. 1978. Vector coding and command fibres. In The Command Neuron Concept. I. Kupfermann and K. R. Weiss, eds. Beliav. Brain Set. 1: 3-39. Fraser, P. J. 1982. Views on the nervous control of complex behavior. Pp. 293-319 in The Biology of Crustacea. Vol. 4: Neural Integration and Behavior. D. C. Sandeman and H. L. Atwood, eds. Academic Press, New York. Fraser, P. J. 1989. Vector coding and multiplicative gain control in the nervous system of the crab. Pp. 95-102 in New Developments in Neural Computing. J. G. Taylor and C. L. T. Mannion. eds. Adam Hiliger, New York. Fraser, P. J. 1990. Equilibrium control by statocyst activated mterneu- rons. Pp. 187-192 in Frontiers in Crustacean Neurobiology, K. Wiese, W. D. Krenz. J. Tautz, H. Reichert, and B. Mulloney, eds. Birkhauser Verlag, Basel. Fraser, P. J., and A. G. Macdonald. 1994. Crab hydrostatic pressure sensors. Nature 371: 383-384. Fraser, P. J., and M. Takahata. 2001. Statocysts and statocyst control of motor pathways in crayfish and crabs. In The Crustacean Nen-ous System. K. Wiese el al.. eds. Springer Verlag, Berlin. (In press). Fraser, P. J., M. Bevengut, and F. Clarac. 1987. Swimming patterns and the activity of identified equilibrium interneurons in the shore crab, Carcinus maenas. J. Exp. Biol. 130: 305-330. Fraser. P. J., A. G. Macdonald, and R. N. Gibson. 1996. Low pressure hydrostatic pressure receptors in the crab Carcinus maenas (L.). In Basic and Applied High Pressure Biology IV. J. C. Rostain. A. G. Macdonald, and R. E. Marquis, eds. Medsubhyp. Int. 5: 59-68. Fraser, P. J., A. G. Macdonald, S. F. Cruickshank, and M. P. Schraner. 1997. Integration of hydrostatic pressure information by identified interneurons in the crab Carcinus maenas (L.): long term recordings. Proc. R. Inst. Navigation RIN 97: 25.1-25.10. Fraser. P. J., A. G. MacdonaSd. S. F. Cruickshank, and M. P. Schraner. 2001. Integration ot h_»d.rstatic pressure information by identified interneurons in the crab Carcinus maenas (L.); long term recordings. J. Navigation (In press). Lemmnitz, G., and H. G. Wolff. 1990. Recording from sensory cells in the statocyst of Astacus. Pp. 97-105 in Frontiers in Crustacean Neu- robiology. K. Wiese, W. D. Krenz, J. Tauz, H. Reichert, and B. Mulloney, eds. Birkhauser Verlag, Basel. Libersat, F., F. Clarac, and S. Zill. 1987. Force sensitive mechanore- ceptors of the dactyl of the crab: single unit responses during walking and evaluation of function. J. Neurophysiol. 57: 1618-1637. Macdonald, A. C., and P. J. Fraser. 1999. The transduction of very small hydrostatic pressures. Comp. Biochem. Physiol. A 122: 13-36. McVean, A. 1982. Autotomy. Pp. 107-132 in The Biology of Crustacea, Vol. 4: Neural Integration and Behavior. D. C. Sandeman and H. L. Atwood, eds. Academic Press. New York. Naylor, E., and R. J. VV. Atkinson. 1972. Pressure and rhythmic be- havior of inshore marine animals. Symp. Soc. Exp. Biol. 26: 395-416. Sandman, D. C., and A. Okajima. 1972. Statocyst-induced eye move- ments in the crab Sc\/la serrala I. The sensory input from the statocyst. / Exp. Biol. 57: 187-204. Schmitz. J., J. Dean, T. Kinderman, M. Schumm, and H. Cruse. 2001. A biologically inspired controller for hexapod walking: simple solu- tions by exploiting physical properties. Biol. Bull. 200: 195-200. Wales, W., F. Clarac, and M. S. Laverack. 1971. Stress detection at the autotomy plane in the decapod Crustacea. 1 . Comparative anatomy of the receptors of the basi-ischiopodite region. Z. Vgl. Physiol. 73: 357-382. Reference: Biol Bull. 200: 160-168. (April 2001) Invertebrate-Inspired Sensory-Motor Systems and Autonomous, Olfactory-Guided Exploration FRANK W. GRASSO* Boston University Marine Program, Marine Biological Laboratory, Woods Hole, Massachusetts 02453 Abstract. The localization of resources in a natural envi- ronment is a multifaceted problem faced by both inverte- brate animals and autonomous robots. At a first approxima- tion, locomotion through natural environments must be guided by reliable sensory information. But natural envi- ronments can be unpredictable, so from time to time, infor- mation from any one sensory modality is likely to become temporarily unreliable. Fortunately, compensating mecha- nisms ensure that such signals are replaced or disambigu- ated by information from more reliable modalities. For invertebrates and robots to rely primarily on chemical senses has advantages and pitfalls, and these are discussed. The role of turbulence, which makes tracking a single odor to its source a complex problem, is contrasted with the high-fidelity identification of stimulus quality by the inver- tebrate chemoreceptor and by artificial sensors. Chemical Senses in the Resource Localization Problem The ability of an animal or an autonomous robot to secure resources (e.g., a supply of energy, a source of water, a shelter from dangerous weather conditions or other hazards) is critical to its behavior and survival. In natural environ- ments critical resources are rarely stable for long. As they are depleted, become inaccessible, or simply disappear, the problem of securing resources is one that autonomous * Current address: Dept. of Psychology, Brooklyn College. CUNY, 2900 Bedford Ave.. Brooklyn. NY 11210. E-mail: fgrassoS1 biomimetics.mbl.edu This paper was originally presented at a workshop titled Invertebrate Sensory Information Processing: Implications for Biologically Inspired Autonomous Systems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences, Woods Hole. Massachusetts, from 15-17 April 2000, was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. agents must solve repeatedly. Whether all of the resources in a given region are mapped at once for future use or new resources are found sequentially by foraging as identified ones expire, the activity is exploration. The first step in exploring for spatially distributed re- sources is localizing them with appropriate sensory equip- ment. That animals and robots rely on their senses to locate resources is a tautology: an entity without sensors cannot determine the location of anything. Simple, single-modality search mechanisms have been well studied in robots and invertebrates. If the resource is obvious and detectable from a unique cue, a simple scan with a ranged (distance) sensor from a single site will suffice to locate it. If the resource, and therefore the diagnostic cue, is hidden, a systematic probe of each location in the search region may be required. In the former case no locomotion is required of the animal or robot to localize the resource; in the latter the animal or robot must locomote everywhere. Robots and animals typically employ mechanisms that combine or lie between these two extremes, depending on the complexity of the sensory task. Optical illusions are the result of a deceived visual sense (Gibson, 1982). Echoes from the substrate can confound auditory localization senses (Mountain and Hubbard, 2001; Robert. 2001 ). Though the mixing of odors from different sources may muddle olfactory recognition under some con- ditions, olfaction offers an advantage over vision and audi- tion in sensing the nature of distant objects. By definition, a chemical sensor is detecting a sample that was a physical constituent of the resource. Once detected, therefore, the resource must certainly be in the vicinity, and the agent may then make inferences about the resource's quality. Thus, the chemical senses are advantageous when used in resource localization, because the rate of false positives is low. Although invertebrates are much more effective than existing robots in chemical tracking, scientists and engi- neers face three formidable and interrelated challenges that 160 OLFACTORY-GUIDED EXPLORATION 161 must be surmounted before invertebrate solutions can be put to work for us in autonomous systems designed for explo- ration. First, we must thoroughly understand the sensory and behavioral repertoires of organisms that are proficient chemical trackers. Second, we must be able to transfer the principles abstracted from animals to implementation in autonomous systems. The principal pitfall, however, is tur- bulence, which governs spatial distributions of chemicals in most macroscopic environments. In this essay I discuss the advantages and the pitfalls of identifying the chemical sens- ing strategies employed in or inspired by invertebrate sys- tems and using them to solve the resource localization problem. Natural fluid-dispersal processes are a two-edged sword: turbulence complicates the chemical tracking prob- lem, but turbulence simultaneously provides a "dynamic similarity" that is invariant in its characteristics across en- vironments. The Physics of Tracking Chemicals to Their Source Tracking an odor (a chemical or chemical cocktail per- ceived by an olfactory system) to its source might, naively, be imagined to involve the maintenance of the chemical sensors in continuous contact with the odor. The analogy to keeping the image of a visually tracked object fixated on the retina or the vibrations arising from an acoustically tracked object on the cochlea is tempting. However, given that the odor track can be stationary on a solid substrate or moving in the fluid medium above, the olfactory counterpart of the analogy is trail following, not plume tracking. Among invertebrates, chemical trails are used by conspe- cifics for social purposes but also by predators or parasites to exploit the trail-leaving individual; and these are distinct behaviors demanding distinct strategies. Ants, for example, follow the pheromone trails left on the substrate by their nest mates (Wilson. 1962). Male snakes locate potential mates by following the chemical signatures left on the ground by females (Ford and Low. 1984; Ford, 1986). Snails (Wells and Buckley. 1972) track similarly, though they sometimes vary their search for mates by indulging in cannibalistic predation (Snyder and Snyder. 1971). Not- withstanding that the strategies underlying these activities are widely recognized for their effectiveness in autonomous systems, they may be of limited utility for exploration of an unknown chemical landscape. In all of these examples — and in the case of cooperative foraging by eusocial in- sects— salient trails are deliberately generated by one indi- vidual for another to follow. Therefore, localization strategies that draw on examples from foraging or hunting creatures hold more promise for autonomous exploration. Shrimp track the odor trail left in the wake of sinking pieces of food (Hamner and Hamner. 1977). For small organisms like bacteria (Alder. 1969) and copepods (Yen, 2000). which operate at spatial scales dominated by diffusion (rather than turbulent dispersal), odor forms a single, sta- tionary patch contiguous with the source that these creatures exploit. Similarly, nematodes (Caenorhabditis elegans) may track the gradients of particles between sand grains to locate mates and sources of nourishment (Morse et al, 1998; Pierce-Shimomura et al., 1999). In any event, how- ever, following a stationary odor trail is only a small part of the behavioral invertebrate repertoire. Although tracking a stationary odor source may appear to be an easier task than tracking a moving one, the principles of fluid mechanics, upon which olfaction depends, make the problem of the moving trail more tractable. When a stationary object or resource donates portions of itself to the ambient medium through the release of volatile compounds or through surface erosion, that region of space containing the detached chemical from the source is termed a plume (Fig. 1 ). Long-lasting odor plumes from a stationary source can leave traces that accumulate on the substrate. This precipi- tated odor creates onto the substrate a two-dimensional imprint resembling a time-average image of the plume (Fig. 1 ). Depending on the conditions, the image might resemble a trail or an annulus, with its highest concentrations near the source. Production of such an image would require stable release and environmental conditions for long periods rela- tive to the dispersal process. Above the micro-scale of nematodes mentioned earlier, no studies of animals tracking such plume images appear in the literature. This is probably because conditions to permit such precipitated plume im- ages are rarely realized physically in natural environments (terrestrial aquatic or marine). This method of inferring a plume from long-term time averages could find application in exploration for persistent resources under conditions fa- vorable to such long-term deposition such as in the vicinity of deep-sea vents or in subterranean caverns. Most animals that used chemo-orientation to locate re- sources do so by making use of plumes borne on the fluid medium rather than deposited on the substrate. Among invertebrates, plume tracking is a common olfactory-guided behavior. Because it depends on transport by a fluid, it is fundamentally different from acoustic tracking, visual tracking, or stationary trail-following. For an agent to locate sources of odor, the odor must be earned by the ambient fluid from the surface of the object to the surface of the chemical sensors of the robot or the animal. The extent to which the continuity of the odor plume is maintained depends on the viscosity of the fluid (v), the ve- locity of the fluid (U), and the length scale of interest (/). The relationship is summarized in the Reynolds number. Re: Re = lUlv When the value of Re < 1 viscous forces dominate, and the continuity of the odor in space will be preserved. This is the regime to which bacteria, copepods. and nematodes are 162 F. W. GRASSO Figure 1. "Leaky" plume characterization in the laboratory. This plume models the odor from an immobile object on the bottom. It was generated in a flume that produced approximately laminar flow. The source, a narrow tube on the floor of the flume, released dye at an "isokinetic" rate matched to that of the ambient flow. We refer to such plumes as leaky to contrast them with jet plumes and wakes that add momentum coincident with the injection of tracer. This is thus a good model for a passive resource on the substrate of a fluid environment. This figure illustrates the spatial distribution of this plume as visualized from trace dye concentration averaged over 5 min at 30 Hz. It shows the plume as a very slow moving animal would experience it with its olfactory organs. This is contrasted with an inset instantaneous ( < 10 /is) section of the same plume as visualized with the laser-induced fluorescence (LIF) method (see text). The inset gives an indication of the intermittent nature of this plume as faster moving animals might experience it. adapted. At Re > 1 inertial forces dominate, and plume coherence begins to break down. The higher the value of Re, the greater the degree of turbulence. Most resource localization problems solved by macroscopic invertebrates and those that are of interest for human-scale exploration fall within fluid mechanical regimes that exceed Re > 10. Intermittent signaling results directly from the action on inertial forces of chemical distribution. When turbulence acts to fragment a continuous stream of chemicals released from an odor source, patches of ambient fluid are interposed between patches of odor (Murlis and Jones, 1981; Moore and Atema, 1991: Murlis et ai, 1992). From the perspective of a sensor in the plume, the patches of chemical in space are detected as pulses in time. The problem of tracking odors to their source is thus a temporal one, requiring the use of a time series of encounters with patches of odor for guidance to their source. The temporal or spatial dispersal patterns produced by different Reynolds regimes are diverse. However, the Reynolds number points to a principle of similarity as well as diversity. Regardless of the specific length scale, viscos- ity, or velocity, conditions with Re within an order of magnitude are "dynamically similar." Put another way, any strategy that can lead an agent to the source of a chemical or odor at one combination of viscosity, velocity, and length scale is predicted to work for all combinations of these three variables that have similar Reynolds regimes. A dramatic demonstration of this principle was the huge smoke plume generated by the recent fire at Los Alamos. Imaged from an orbiting remote sensing platform, the plume from the fire spread eastward over the Rocky Moun- tains and covered six states. The overall shape was remi- niscent of many other plumes (Fig. 1 ). My colleagues and I could pick out features and characteristic structures that could as easily have been found in a plume from a camprire. in a bathtub, or in one of our laboratory flumes. It is not clear at this time whether animals can recognize the Reyn- olds regime in which they find themselves and adjust their behavior appropriately: indeed, the problem may be ill- posed. Yet, if this is possible it would be a powerful means OLFACTORY-GUIDED EXPLORATION 163 for selection of appropriate strategies by biological or hu- man fabricated autonomous agents. Invertebrates and Chemical Plume-Tracking Plume-tracking behavior G. S. Fraenkel and D. L. Gunn defined a set of theoretical mechanisms and proposed that these could explain the ori- entation behavior of animals (Fraenkel and Gunn, 1961). The kineses are defined as undirected reactions to a stimu- lus: The animal may change direction or speed, systemati- cally or randomly, to the cue or source, but this does not lead to systematic orientation of the animal's body. Kineses are characteristic of bacteria (Berg and Purcell, 1977; Alder, 1987). Taxes involve directed reactions to a stimulus. In WiHotaxis the animal takes sequential samples of a stimulus and estimates whether movement is with or against the spatial gradient of stimulus intensity. Insect maggots and Euglena are examples of creatures that use this mechanism (Mast. 1938). TVopotaxis is a mechanism by which the animal makes simultaneous comparisons with two or more sensors to yield an instantaneous estimate for guidance to move with or against a cue gradient. The use of two or more spatially separated sensors by odor-tracking decapod crus- taceans (Reeder and Ache, 1980; Devine and Atema, 1982; Beglane el ai, 1997) and by adult flying insects (Srinivasan etal., 1996, 2001) and walking insects (McCoy, 1984; Bell. 1986) has been interpreted as indicative of a tropotactic mechanism. These theories, already advanced in the 1940s, have had a long and productive influence on ideas of chemo-orientation in animals since. There is a growing understanding, however, that their intuitive appeal often exceeds their practical utility as explanations of specific tracking behavior in turbulent plumes. Since the pioneering work of J. Murlis (Murlis and Jones, 198 1 ), biologists have been aware of the problems posed by the intermittency caused by the natural odor dispersal and have proposed alternative mechanisms to single modality taxes and kineses. Odor gated rheotaxis (OGR) (Kennedy, 1986) combines the guidance provided by two sensory modalities. Thus, while the animal is sensing odor, it moves "upstream" against the ambient fluid flow. When contact with the odor is lost, the motion is directed cross-stream, perpendicular to the mean flow. This "casting," if in the correct direction, can recover contact with the plume and cause a new upstream surge. This mechanism is common among invertebrates; it is found in many species of moths (Kennedy, 1986; Baker el ai., 1988; Vickers and Baker, 1992; Mafra-Neto and Carde, 1994; Mafra-Neto and Carde, 1996) and in blue crabs (Weissburg and Zimmer-Faust, 1994; Zimmer-Faust etal.. 1995). In recognition of the costs associated with the degree of persistence involved in casting and surging upstream, a variant of OGR called counter- turner, which incorporates an internal oscillator, has also been advanced (Belanger and Willis, 1996). Atema has proposed eddy-chemo-rheotaxis (ECR) (Atema, 1996), a mechanism in which the microflow-associated eddies pro- duced by turbulence (as contrasted with the mean flow used in OGR) are coi'ilmed with the "flavor" of the eddy to provide guidance toi ••ment. Eddies dissipate into the ambient flow (have bi! d deaths) as they move from the source. Thus, dependmi' e distance from the source at which the animal encountu . they will possess differ- ent characteristics. This is a potentially rich source of in- formation about the nature of the source and its distance from the agent. Speculations about other sources of infor- mation for guidance have also been advanced (Grasso et al., 1996a). As studies of chemo-orientation break away from the classical notions of taxes and kineses, work on OGR and ECR points to two major themes for the future. First is multi-modal research; that information from multiple sen- sory modalities must inform investigations of olfactory searching behavior. The second theme is the exploitation by searching agents of the structures produced by turbulence as it carries the odor from the source. Further research in these new areas is likely to find many other novel — and to judge from the data provided by animal studies — much more efficient strategies. Switching between behaviors. Animals alter their behav- ior as they progress through a plume. In the course of a successful tracking episode an animal may show variations in speed or direction that are characteristic of a specific phase of tracking behavior. Lobsters show an initial increase in speed, followed by a steady speed phase, and a final decrease, as they progress toward the source (Moore et al., 1991; Basil, 1994). Their headings also show systematic variation as they track certain types of plumes. Such shifts may be attributed to differences in the spatial organization of the plume (Grasso et al.. 1996b). As blue crabs and lobsters near the source, they shift from using their anten- nules, which are elevated above the substrate, to using the chemo-receptors on their dactyls (feet), which are in contact with the substrate (Moore et al., 1991; Basil. 1994; Weiss- burg and Zimmer-Faust, 1994). These behavioral observations, and the changes in the structure of the plume as it evolves on its way downstream, suggest that a critical problem in the design of artificial autonomous systems will be the need to match the tracking algorithm to the ambient conditions. To do this, the robot, like an animal, must be able to recognize a given condition, and particularly those conditions that require a change from one strategy to another. Solutions to the problem may entail (1) shifting modalities; (2) giving a different weight to the information coming from one sensory modality relative to another; or (3) identifying more subtle shifts in the infor- mation extracted by a given modality. 164 F. W. GRASSO The spatial biomechanics of olfactory sampling The use and integration of multiple sensors is a faculty at which invertebrates seem to excel. Invertebrates bristle with a diversity of chemosensors and fluid-mechanosensors. In- deed, this and similar observations have led some to spec- ulate that invertebrates have pushed their neural processing peripherally to lighten the central processing load and thereby to increase response speed (Wehner, 1987). The preceding discussion of turbulent fluid dispersal under- scores the complexity of the problems that animals face in placing and maintaining these sensor arrays in contact with a plume (Fig. 1, inset). To maintain that contact, many invertebrates have evolved modes of body locomotion that are expressed during searching. In particular, the sensors are moved through space (a principle they share with verte- brates) and moveable antennae that place some of the sen- sors away from the body occur (structures that are atypical of the vertebrate bauplan; though catfish, sea robins, etc., possess chemical sensors at the ends of moveable append- ages). Antennae make the experience of odors very different from our own. Humans, and vertebrates in general, have internal nasal passages that homogenize and obliterate the natural spatial and temporal structure of odor signals before they contact the chemoreceptor. Invertebrates with chemical sensors on the ends of long thin stalks (arthropods, cepha- lopods, and certain gastropods being common examples) have more direct access to the temporal structure of odor signals produced by turbulence. Distal placement frees the sensors from the boundary layers and hydrodynamic shadow of the body, and movement of the appendages enables a search for contact with the plume in ways that are informed by recent experience. The motions of the whole body are also critical to determining the patterns of spatial sampling experienced by the receptor surface. To the extent that motions of an animal's body and sensory antennae may be important components of any active searching plume- tracking strategy, similar mechanisms would be of great importance in developing autonomous plume-tracking sys- tems. Neural processing of olfactory signals That chemoreceptors reside on the ends of antennae, free from body boundary layers, is not evidence that inverte- brates actually utilize data about the structure of the odor plume. The chemoreceptors must be fast enough, and sen- sitive enough, to capture the information and transfer it centrally. Recordings from peripheral chemoreceptors in lobsters (Gomez and Atema, 1996a; Gomez and Atema. 1996b; Gomez et ai, 1999) and cockroaches (Lemon and Getz. 1996, 1999) suggest that, to a first approximation, this is the case in both terrestrial and marine invertebrates. Modeling studies of both species, constrained by physio- logical data, indicate that dynamic information on the time scale of ambient turbulence can be passed centrally and is not filtered out by the chemoreceptors (Getz. 1999; Grasso. unpubl. data). We know very little about how the natural dynamics of odor signals are processed centrally. Arthropod brains, like those of vertebrates, contain olfactory information-process- ing centers that are subdivided into glomeruli (Hildebrand and Shepherd. 1997; Derby and Steullet. 2001). These dense regions of neuropile are thought to be the sites of convergence for information about the quality or identity of an odor. Recent research (particularly that employing mo- lecular techniques) has mapped the pathways that carry olfactory information to the glomeruli, and the results sup- port the long-held notion that the coding of odor quality is somehow mirrored in the spatial arrangement of glomeruli (Hildebrand and Shepherd, 1997). A recent study in the three-toed box turtle suggests that concentration coding is independent of this level of anatomical organization in the olfactory bulb (Wachowiak et ai, 2000). Variation in odor concentration rather than information about odor quality is the stimulus dimension along which natural odor dynamics information is expected to be encoded in the central nervous system. The central processing and representation of the dynam- ics of natural odor has received virtually no attention. Once an animal has solved the quality problem (i.e., an odor has been recognized as worth tracking), information reflecting the temporal series of odor patch encounters must be re- ceived, interpreted, and translated into motor commands that lead to the source. An explanation of any piece of this process will aid our understanding of olfactory systems and not just plume tracking. The speed and efficiency with which certain invertebrates (notably arthropods) track a turbulent plume to its source demonstrates that the requisite temporal to spatial transformation can be carried out by the nervous system. Autonomous Systems and Chemical Tracking Many simulation studies of chemo-orientation behavior by man-made systems have confirmed that the classic taxes and kineses proposed by Fraenkel and Gunn ( 1961) can lead to localization of a source in a gradient and have quantita- tively delineated the conditions under which taxes and kine- ses will succeed or fail (Rohlf and Davenport, 1969; Brait- enberg. 1984; Beer et ai. 1991; Holland and Melhuish, 1996: Ferree et al.. 1997). A small number of laboratory robot implementations (Deveza et al., 1994; Ishida et al. 1996; Grasso et al., 2000: Ishida et ai, 2001) designed to explain animal chemo-orientation behavior or to produce practical chemical tracking devices have explicitly or implicitly employed various single-modality taxes and kineses. The robot performance, however, has lagged far OLFACTORY-GUIDED EXPLORATION behind the expectations raised by successful simulations. The paths taken by robots that have successfully tracked a turbulent plume are very inefficient compared to the paths taken by animals under comparable conditions; and the algorithms used have been effective only in limited contexts (i.e.. they do not generalize). These performance discrepan- cies are not due to sloppy hardware or software implemen- tations of the algorithms; they result from the failure of these mechanisms to account for the realistic fluid dynamics the robots encounter. The gradients present in naively sim- ulated plumes (and tacitly assumed by the single-modality kineses and taxes by Fraenkel and Gunn) are absent from physical settings where turbulence dominates the patterns of odor dispersal at reasonable time-scales (Warhaft, 2000). The problem rests with our understanding and representa- tion of the environment in simulation studies. The solution requires algorithms that are designed to operate in real plumes rather than smooth gradients. Improved simulations of real environmental conditions is not a straightforward matter. Simulation of fluid dynamics. with sufficient accuracy, over the required range of temporal and spatial scales is not computationally feasible and per- haps impossible (Warhatt, 2000). An alternative is to cap- ture the dispersal of a given plume in digital form and to use the resulting "digital movie" as the simulation environment. This method is achieved in practice by putting a fluorescent dye into the plume source and capturing the laser-induced fluorescence (LIF) of the dye with digital imaging tech- niques. From the digital movies, flow fields can simulta- neously be obtained for the same plume (Cowen and Mo- nismith, 1997). This "one plume at a time" approach does not lead to generality, but it does allow for the evaluation of many candidate plume-tracking algorithms and for explor- ing the parameter sensitivity of those algorithms under realistic fluid conditions. This simulation prescreening can be done on a scale that would be impracticable with robots. These LIF methods, however, cannot include the agent's own influence on turbulent dispersal as it moves through the plume. For this, and to avoid the false sense of progress described in the first paragraph in this section, robots must provide the final test. Robotic studies of chemical plume tracking progress more slowly from simulation studies (but faster than animal studies!). Therefore, a combination of prescreening of algorithms in simulation to weed out pa- tently unsuccessful strategies, followed by definitive testing with robots seems a sound approach. Compared to the toy problems with which robots and experimental animals are challenged in the laboratory (where experimental control requires variables to be held to a minimum), resource localization problems in natural set- tings are vastly more complex at many levels. Physical hazards, obstacles, competitors, and predators require be- havioral responses that interrupt the smooth application of a foraging strategy. The recognition of false cues, early aban- donment of false trails, and maintenance of a stream of appropriate guidance information in unpredictable environ- ments may also require real-world animals and robots to employ a hierarchy of behavioral responses that may devi- ate from the application of a single simple mechanism. A mechanism that switches between strategies when condi- tions change, or that varies the weighting of sensory mo- dalities with their reliability, offers solutions to many of these difficulties. Such strategies — so essential for the suc- cess of autonomous systems — have scarcely been addressed in animal and biomimetic robotic studies. Biomimetic Robots and Hypothesis Testing Methods The preceding sections show that fluid mechanics, animal behavior, neuroscience, and cybernetics can each contribute to the understanding that will finally permit us to construct autonomous systems that can locate resources based on their chemical signature and under chaotic environmental condi- tions. Each field has reached a point where its methodology has provided tantalizing hints about the robust plume-track- ing strategies of invertebrates; but each faces difficult tech- nical and conceptual hurdles before it can advance further. Biomimetic robots offer a methodology that can answer some of the questions raised by each of these approaches. In my own studies, conducted with colleagues in the laboratory of L Atema at the Marine Biological Laboratory in Woods Hole, the behavior of a robot told us something unsuspected about the plume structure (Grasso el ai. 1996b). The behavior of this robot changed in different regions of the plume as it was executing a simple tropotactic tracking algorithm. This result indicated a qualitative dif- ference in plume structure from the robot's perspective and serves notice that changes in animal behavior must be interpreted with caution. We have also explored spatial sampling issues by varying the separation between paired sensors (Grasso et al., 1997). This work had an engineering aspect — finding an optimal sampling configuration — and a biological aspect — determining the information that ani- mals obtain from the inter-sensor separations they possess. We have also studied tropotactic algorithms in one (jet) plume (Grasso er al., 2000) and OGR in two (leaky) plumes (Grasso. unpubl. data) that American lobsters are known to track. As these robot experiments accumulate, they raise new biological questions and narrow the possible biological and physical explanations of plume tracking by lobsters. The performance of the robots used in these studies was quantitative/v evaluated against the performance of the American lobster. This unique comparison was possible because we tested the robot under physical conditions vir- tually identical to those under which animal experiments had previously been conducted (i.e.. the plume was com- posed of a natural food extract for the lobster and a synthetic tracer for the robot, but all other experimental conditions 166 F. W. GRASSO eft and Right Plume Sensors Algorithms Discrete StContinuous forms of: 1. Single sensor Odor-Gated "Rheotaxis' 2. Dual sensor Odor-Gated "Rheotaxis" 3. Plume-Edge-T racking OGR Guidance from: A; Onboard gyro provides a frame of reference to simulate estimation of / flow direction. B. Three "c provide indications of pl extent in the boundar Figure 2. One of our biomimetic robots, in the large flume facility at the Marine Biological Laboratory, Woods Hole. The biologically inspired chemical sensing aquatic autonomous robots (BISAAR) are equipped with two antenna-like fluorescence sensors that detect fluorescein dye with the spatial (1 mm) and temporal (>5 Hz) resolution of the lobster lateral antennule. From this information, and an approximation of the mean flow in the tank provided by a gyroscope, the BICSAAR robots are a platform for exploring plume-tracking algorithms. The third chemical sensor, the "bottom sensor," pointed toward the flume floor is intended to fulfill part of the role of the chemical sensors on the legs of real lobsters. It is used in studies of behavioral switching between antennule and leg-guided chemo-orientation. The wheeled chassis is capable of propelling and turning faster than its biological counterparts, so to a first approximation it achieves locomotory biomimetic scaling. were identical). The comparison was also possible because the critical factors of robot hardware and software required to implement the algorithms were matched to the corre- sponding biological characteristics of the lobster (Fig. 2). I refer to this as biomimetic scaling. My use of the term biomimetic is not intended to denote a slavish mimicry of all the biological details. The mimicry is primarily conceptual and is structural only as needed to support the conception. The idea of biomimetic scaling is to reproduce only those points that are essential to test a given hypothesis. Thus, to test the power of OGR or tropotaxis as explanations of plume tracking by lobsters, we constrained the following critical features to known lobster values: body size and shape, sensor arrangement in space, speed and pattern of locomotion, and temporal and spatial resolution of the sensors in the robots' hardware and software to known lobster values (Grasso a ai, 2000). These experiments were conducted to evaluate behavioral theories. The results, while germane to the plume-tracking problem, had broader implications for the study of fluid mechanics, cybernetics, and neuroscience, as noted above. Continued studies with more strenuous constraints — from implementation of a realistic neural architecture, to software or sensors configured to take advantage of fluid flow around the robot in ways that mimic antennal morphology — will lead to the evaluation of theories with ever-greater explan- atory (or exclusionary) power. In future studies, biomimetic scaling will allow us to use robots to explore issues of neural processing and biomechanical strategies of spatial sampling. Though space prohibits a detailed review, note that af- fordable technologies are available from commercial sup- pliers to make small processors and actuators as well as biologically scaled sensors that are suitable to meet the demands of biomimetic scaling (Ishida et ai, 2001; Robert, 2001). Conclusions With the methodology described in this paper, we cannot "prove" that the algorithms biomimetic robots implement are the exact analogs of biological mechanisms. Those mechanisms can only be determined from biological exper- iments. But in the application of the scientific method, we OLFACTORY-GUIDED EXPLORATION 16" can exclude untenable biological explanations, simply stat- ing that they fail to account for the biological behavior. As this process of excluding the biologically implausible proceeds, something unrelated will occur in parallel. The ability of the robots to track chemical plumes to their source will improve as their performance approaches that ot real animals. Thus, although there is no scientific guarantee that the robots will ever match their biological models, a real- izable benchmark will be systematically approached. The algorithms gleaned from these biologically inspired studies can find practical application as components in autonomous systems intended for exploration. Though the "one plume at a time" approach is a slow and inductive one, the principle of "dynamic similarity" indi- cates that strategies developed in one context may find general utility in autonomous systems that are designed to operate at a variety of scales with coincidental Reynolds numbers. This extends to regimes in fluids of different viscosity like water and air (or any fluid for that matter). Studies of behavioral switching by animals may help to demarcate these boundaries and may also suggest control hierarchies that will recognize strategies appropriate to par- ticular fluid-dynamic contexts. Finally, invertebrates offer excellent models for the study of plume tracking. Their external sensors, borne on movable antennae, provide access to dynamic information associated with plume structures that is simply not available to verte- brates with internal noses. This may mean a richer diversity of information, and by analogy to the significance of eye movements for visual information processing (Yarbus, 1967). the motions of the antennae provide an easily as- sayed indication of information-processing strategies. Acknowledgments This work was supported in part by a grant from the joint DARPA/ONR plume-tracing program administered by Keith Ward (ONR) and Regina Dugan (DARPA). I thank Jelle Atema, Jennifer Basil, David Mountain, and Rainer Voigt for stimulating and inspiring discussions; Lisa Bor- den, Todd Cowen, Jonathan Dale, Kevin Dittmer, and John Hanna, for technical assistance; Diana Blazis and Pam Old- ham for making the ISIP-BIAS symposium possible and successful; and Todd Cowen, Jeff Kosseff, and Don Web- ster for discussions of fluid dynamics and for drawing my attention to the Los Alamos plume. Literature Cited Alder, J. 1969. Chemoreceptors in bacteria. Science 166: 1588-1597. Alder, J. 1987. How motile bacteria are attracted and repelled by chem- icals: an approach to neurobiology. Biol. Chem. 386: 163-173. Atema, J. 1996. Eddy chemotaxis and odor landscapes: exploration of nature with animal sensors. Biol. Bull. 191: 129-138. Baker, T. C., B. S. Hansson, C. Lofstedt, and J. Lofqvist. 19SS. Adaptation of antennal neurons in moth is associated with cessation ot pheromone-mediated upwind flight. Neurobiology 85: 9826-9830. Basil, J., and J. Atema. 1994. Lobster orientation in turbulent odor plumes: simultaneous measurement of tracking behavior and temporal odor patterns. Biol. Bull. 187: 272-273. Beer, R. D.. H. J. Chiel. and L. S. Sterling. 1991. An artificial insect: computer-simulated insects that adapt to their environment may be the next stage in evolution of artificial intelligence. Am. Sci. 79: 444-452. Beglane, P. F., F. W. Grasso, J. A. Basil, and J. Atema. 1997. Far field chemo-orientation in the American lobster, Homarus americanus: ef- fects of unilateral ablation and lesioning of the lateral antennule. Biol. Bull. 193: 214-215. Belanger, J. H., and M. A. Willis. 1996. Adaptive control of odor- guided locomotion: behavioral flexibility as an antidote to environmen- tal unpredictability. Adapt. Behav. 4: 217-253. Bell, W. J. 1986. Responses of arthropods to temporal chemical stimulus changes: simulation of a humidity differential and a pheromone plume. Pp. 139-148 in Mechanisms in Insect Olfaction, T. L. Payne, M. C. Birch, and C. E. J. Kennedy, eds. Clarendon Press, Oxford Berg, R. C., and E. M. Purcell. 1977. The physics of chemoreception. Biophysics 20: 193-219. Braitenberg, V. 1984. Vehicle's: Experiments in Synthetic Psychology. Bradford Books. MIT Press, Cambridge. MA. Cowen, E. A., and S. G. Monismith. 1997. A hybrid digital particle tracking velocimetry technique. Exp. Fluids 22: 199-211. Derby, C. D., and P. Steullet. 2001. Why do animals have so many receptors? The role of multiple chemosensors in animal perception. Biol. Bull. 200: 211-215. Deveza, R., D. Thiel, A. Russell, and A. Mackay-Sim. 1994. Odor sensing for robot guidance. Int. J. Rohm. Res. 13: 232-239. Devine, D. V., and J. Atema. 1982. Function of chemoreceptor organs in spatial orientation of the lobster. Homarus americanus: differences and overlap. Biol. Bull. 163: 144-153. Ferree, T. C., T. M. Morse, and S. R. Lockery. 1997. Neural networks for chemotaxis in C. elegans: rule extraction and robotics. Society for Neuroscience: 27th Annual Meeting. New Orleans. 25-30 October 1997. Abstr. #527.10. Ford, N. B. 1986. The role of pheromone trails in the sociobiology of snakes. Pp. 261-278 in Chemical Signals in Vertebrates, D. Duvall, D. Muller-Schwartze. and R. M. Silverstien, eds. Plenum, New York. Ford, N. B., and J. R. Low. 1984. Sex pheromone source location by garter snakes: a mechanism of detection of direction in nonvolatile trails. J. Chem. Ecol. 10: 1193-1199. Fraenkel, G. S., and D. L. Gunn. 1961. The Orientation <>/ Animals: Kineses, Taxes and Compass Reactions. Dover, New York. Getz, W. 1999. A kinetic model of the transient phase in the response of olfactory receptor neurons. Chem. Senses 24: 497-508. Gibson, J. J. 1982. Ecological physics, magic and reality. Pp. 217-223 in Reasons for Realism: Selected Essays of James J. Gibson. R. Edward and R. Jones, eds. Lawrence Erlbaum Associates. London. Gomez, G., and J. Atema. 1996a. Temporal resolution in olfaction: stimulus integration time of lobster chemoreceptor cells. J. Exp. Biol. 199: 1771-1779. Gomez, G., and J. Atema. 1996b. Temporal resolution in olfaction II: time course of recovery from adaptation in lobster chemoreceptor cells. J. Neurophysiol 76: 1340-1343. Gomez, G., R. Voigt. and J. Atema. 1999. Temporal resolution in olfaction III: flicker fusion and concentration-dependent synchroniza- tion with stimulus pulse trains of antennular chemoreceptor cells in the American lobster. J. Camp. Physml. 185: 427-436. Grasso, F. W., T. Consi, D. Mountain, and J. Atema. 1996a. Locating odor sources in turbulence with a lobster inspired robot. Pp. 104-1 12 in Front Animals to Animals 4: Proceedings of the Fourth International Conference on Simulation of Adaptive Behavior. P. Maes, M. J. Ma- 168 F. W. GRASSO taric, J.-A. Meyer, J. Pollack, and S. W. Wilson, eds. MIT Press, Cambridge, MA. Grasso, F. W., J. H. Dale, T. R. Consi, D. C. Mountain, and J. Atema. 1996b. Behavior of purely chemotactic robot lobster reveals different odor dispersal patterns in the jet region and the patch field of a turbulent plume. Biol. Bull. 191: 312-313. Grasso, F. W., J. H. Dale, T. R. Consi, D. C. Mountain, and J. Atema. 1997. Effectiveness of continuous bilateral sampling for robot che- motaxis in a turbulent odor plume: implications for lobster chemo- onentation. Biol. Bull. 193: 215-216. Grasso, F. W., T. R. Consi, D. C. Mountain, and J. Atema. 2000. Biomimetic robot lobster performs chemo-orientation in turbulence using a pair of spatially separated sensors: progress and challenges. Robot. Aut. Sys. 30: 115-131. 1 1 jimic i. P., and W. M. Hamner. 1977. Chemosensory tracking of scent trails by the planktonic shrimp Acetes sibogae australis. Science 195: 886-888. Hildehrand, J. G., and G. M. Shepherd. 1997. Mechanisms of olfac- tory discrimination: converging evidence for common principles across phyla. Annu, Re\: Neurnsci. 20: 595- 631. Holland. O., and C. Melhuish. 1996. Some adaptive movements of animats with single symmetrical sensors. Pp. 53-64 in From Animals to Animats: Proceedings of the Fourth International Conference on Simulation of Adaptive Behavior, P. Maes, M. J. Matanc, J.-A. Meyer, J. Pollack, and S. W. Wilson, eds. MIT Press, Cambridge, MA. Ishida. H., Y. Kagawa. T. Nakamoto, and T. Moriizumi. 1996. Odor- source localization in the clean room by an autonomous mobile sensing system. Sensors ami Activators B 33: 1 15-121. Ishida, H., T. Nakamoto, T. Moriizumi, T. kikas. and J. Janata. 2001. Plume-tracing robots: a new application of chemical sensors. Biol. Bull. 200: 222-226. Kennedy, J. S. 1986. Some current issues in orientation to odour sources. Pp. 11-25 in Mechanisms in Insect Olfaction. T. L. Payne. M. C. Birch, and C. E. J. Kennedy, eds. Clarendon Press, Oxford. Lemon, W. C., and W. M. Getz. 1996. Temporal resolution of general odor pulses by olfactory sensory neurons in American cockroaches. J. £.v/». Biol. 200: 1809-1819. Lemon, W. C., and W. M. Getz. 1999. Responses of cockroach anten- nal lobe projection neurons to pulsatile olfactory stimuli. Ann. N. Y. Acad. Sci. 855: 517-520. Mafra-Neto, A., and R. T. Carde. 1994. Fine-scale structure of pher- omone plumes modulates upwind orientation of flying moths. Nature 369: 142-144. Mafra-Neto, A., and R. T. Carde. 1996. Dissection of the pheromone- modulated flight of moths using single-pulse response as a template. Experientia 52: 373-379. Mast, S. O. 1938. Factors involved in the orientation of lower organisms to light. Biol. Rev. 13: 186-224. McCoy, M. M. 1984. Antennal movements of the American cockroach Periplaneta americana. Ph.D. dissertation. University of Kansas. Law- rence. 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. Moore. P. A.. N. Scholz, and J. Atema. 1991. Chemical orientation of lobsters. Homarus americanus, in turbulent odor plumes. J. Chem. Ecol. 17: 1293-1307. Morse, T. M., T. C. Ferree, and S. R. Lockery. 1998. Robust spatial navigation in a robot inspired by chemotaxis in Caenorhabditis el- egans. Adapt. Behav. 6: 393-410. Mountain, D. C., and A. E. Hubbard. 2001. Sensing scenes with silicon. Biol. Bull. 200: 227-234. Murlis, J., and C. D. Jones. 1981. Fine-scale structure of odour plumes in relation to insect orientation to distant pheromones and other attract- ant sources. Physiol. Entomol. 6: 71-86. Murlis, J., J. S. Elkinton, and R. T. Carde. 1992. Odor plumes and how insects use them. Annu. Rev. Entomol. 37: 505-532. Pierce-Shimomura, J. T., T. M. Morse, and S. R. Lockery. 1999. The fundamental role of pirouettes in Caenorhabditis elegans chemotaxis. J. Neumsci. 19: 9557-9569. Reeder. P. B., and B. VV. Ache. 1980. Chemotaxis in the Florida spiny lobster, Panulims argus. Anim. Behav. 28: 831-839. Robert, D. 2001. Innovative biomechanics for directional hearing in small flies. Biol. Bull. 200: 190-194. Rohlf. J. F., and D. Davenport. 1969. Simulation of simple models of animal behavior. J. Theor. Biol 23: 400-424 Snyder, N. F. R.. and H. A. Snyder. 1971. Pheromone-mediated be- haviour of Fasciolaria tulipa. Anim. Behav. 19: 257-268. Srinivasan, M. V., S. Zhang, and J. S. Chahl. 2001. Landing strategies in honeybees and possible applications to autonomous airborne vehi- cles. Biol. Bull. 200: 216-221. Srinivasan. M. V.. S. \V. Zhang. M. Lehrer. and T. S. Collett. 1996. Honeybee navigation en route to the goal: visual flight control and odometry. J. £v/>. Biol. 199: 237-244. Vickers, N. J., and T. C. Baker. 1992. Male Heliot/tis virescens main- tain upwind flight in response to experimentally pulsed filaments of their sex pheromone (Lepidoptera: Noctuidae). J. Insect Behav. 5: 669-687. Wachowiak, M., M. Zochowski, L. B. Cohen, and C. X. Falk. 2000. The spatial representation of odors by olfactory receptor neuron input to the olfactory bulb is concentration invariant. Biol. Bull. 199: 162- 163. \Varhaft, Z. 2000. Passi\e scalars in turbulent flows. Annu. Rev. Fluid Mech. 32: 203-240. Wehner. R. 1987. Matched filters — neural models of the external world. J. Comp. Physiol. A 161: 511-531. Weissburg, M. J., and R. K. Zimmer-Faust. 1994. Odor plumes and how blue crabs use them in finding prey. J. Exp. Biol. 197: 349-375. Wells. M. J., and S. K. L. Buckley. 1972. Snails and trails. Anim. Behav. 20: 345-355. Wilson, E. O. 1962. Chemical communication among workers of the fire ant Solenopsis saevissima (Fr. Smith). 1. The organization of mass foraging. Anim. Behav. 10: 134-147. Yarbus, A. L. 1967. Eve Movements and Vision. Plenum Press. New York. Yen, J. 2000. Life in transition: balancing inertial and viscous forces by planktonic copepods. Biol. Bull. 198: 213-224. Zimmer-Faust, R. K., C. M. Finelli, N. D. Pentcheff. and D. S. Wethey. 1995. Odor plumes and animal navigation in turbulent water flow: a field study. Biol. Bull. 188: 111-116. Reference: Biol. Bull. 200: 169-176. (April 2001) Limulus Vision in the Marine Environment ROBERT B. BARLOW*. JAMES M. HITT. AND FREDERICK A. DODGE Center for Vision Research, Department of Ophthalmology, Upstate Medical University, 750 Adams Street, Svracuse, New York 132 JO Abstract. Horseshoe crabs use vision to find mates. They can reliably detect objects resembling potential mates under a variety of lighting conditions. To understand how they achieve this remarkable performance, we constructed a cell- based realistic model of the lateral eye to compute the ensembles of optic nerve activity ("neural images") it trans- mits to the brain. The neural images reveal a robust encod- ing of mate-like objects that move underwater during the day. The neural images are much less clear at night, even though the eyes undergo large circadian increases of sensi- tivity that nearly compensate for the millionfold decrease in underwater lighting after sundown. At night the neural images are noisy, dominated by bursts of nerve impulses from random photon events that occur at low nighttime levels of illumination. Deciphering the eye's input to the brain begins at the first synaptic level with lowpass temporal and spatial filtering. Both neural filtering mechanisms im- prove the signal-to-noise properties of the eye's input, yield- ing clearer neural images of potential mates, especially at night. Insights about visual processing by the relatively simple visual system of Limulus may aid in the design of robotic sensors for the marine environment. Introduction The world is rich with sensory information, and animals are highly efficient at extracting what is essential for their survival. The retina begins the processing of visual infor- * To whom correspondence should be addressed. E-mail: barlowr@upstate.edu. This paper was originally presented at a workshop titled Invenebrate Sensory Information Processing: Implications for Biologically Inspired Autonomous Systems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences. Woods Hole. Massachusetts, from 15-17 April 2000. was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. mation by transforming patterns of incident light intensity into trains of impulses in optic nerve fibers (Dowling, 1987). The retina encodes information it receives in a reli- able and efficient manner (Rieke et ai, 1997) but does not encode all of it. Rather, the retina extracts certain features in the visual scene at the expense of others (Lettvin et ai. 1959). An important first step for exploring the neural code the eye transmits to the brain when an animal sees is to understand what an animal can see in its natural habitat. The next step is to examine the retinal coding of natural scenes in activity of optic nerve fibers. Recordings from single nerve fibers have indeed yielded useful insights about reti- nal function: however, it is difficult to infer from them the information transmitted by arrays of optic nerve fibers to the brain about the complex patterns of illumination animals encounter in their natural habitat. Techniques such as multi- electrode arrays (Meister et ai, 1994) and voltage-sensitive dyes (Wong et ai. 1995) can access patterns of activity generated by ensembles of retinal neurons, but they are not practical for recording from large numbers of optic nerve fibers in behaving animals. A Computational Model of the Limulus Eye An alternative approach is to construct a realistic com- putational model of the eye. The relative simplicity of the eyes of lower vertebrates and invertebrates offers the best opportunities (Werblin, 1991: Teeters et ai. 1997). The lateral eye of the horseshoe crab Limulus polyphenuis is a particularly attractive model system for the following rea- sons: first, its visually guided behavior is well known (Bar- low et ai, 1982: Powers et ai, 1991; Herzog et ai. 1996): second, it processes visual information with integrative mechanisms shared by more complex systems (Barlow. 1969; Ratliff. 1974); and third, its lateral eye contains the largest neural network (-1000 neurons) for which a quan- titative cell-based model exists (Hartline and Ratliff. 1957, 1958; Barlow and Quarles, 1975; Barlow et ai, 1993b). 169 170 R. B. BARLOW ET AL Figure 1. A horseshoe crab. Limuliis polyphemus, mounted with a video camera. "CrabCam." for recording underwater movies and a microsuction electrode for recording responses from a single optic nerve fiber. A white Teflon cap (2.5 cm diameter) seals the recording chamber, which is attached to the carapace anterior to the right lateral eye. The barrel of the microsuction electrode protrudes from the recording chamber to the right. Tethers lead the video and optic nerve signals to recording electronics located on shore or in an overhead skiff as the animal moves about underwater at depths of 0.5 to 1 m. Experiments were carried out in an estuary near the Marine Biological Laboratory, Woods Hole. Massachusetts. We have constructed a computational model of the lateral eye that predicts optic nerve responses with good accuracy (Passaglia et al., 1998). In brief, the model treats the retina as an array of neurons that samples visual space as the compound eye does, incorporates the known excitatory and inhibitory integrative mechanisms of the retina, and adapts to changes in ambient illumination. Our strategy for examining the retinal code underlying be- havior is to first videotape the lateral eye's view of its under- water world with an animal-mounted camera ("CrabCam") while simultaneously recording from a single optic nerve fiber of an ommatidium viewing the central region of the videotaped scene (Passaglia et al., 1997a). Figure 1 shows the CrabCam and the watertight recording chamber mounted on an animal before it enters the water and passes near submerged mate-like objects near the water's edge in Woods Hole, Massachusetts. Back in the laboratory we digitize the CrabCam recordings and feed them to the cell-based model, which then computes the arrays of optic nerve activities in response to the underwater scenes. The computed arrays of activities are converted to a grey scale and mapped to their appropriate retinal location, generating "neural images" of the eye's input to the brain (Fig. 2). Finally, we assess the accuracy of the model's predictions by comparing the response recorded from a single optic nerve fiber to that computed by the model for the corresponding receptor. Correlation coefficients between recorded and corn- Figure 2. Computed responses of the Limulus eye and brain to moving mate-like objects day and night. Top panel shows the results for a low-contrast object, and the bottom panel shows those for a high-contrast object. The two objects approximate the size (0.3 m diameter, 0.15 m high) and range of contrasts of adult female crabs. They move across the visual field at a distance of 0.6 m, where most visual detection occurs (Herzog el al, 1996). The left column ("Visual stimulus") shows CrabCam images of the high- and low-contrast objects after sampling by the eye's optical apparatus. The arrays of pixels indicate the light intensities incident on the 16 X 16 array of ommatidia viewing the videotaped scene. The adjacent column ("Neural image") shows the ensembles of optic nerve activities computed by the retinal model in response to the visual stimuli on the left. The arrays of pixels in the neural images give the computed firing rates of optic nerve fibers mapped onto a gray scale with black set to 0 impulses/s and white set to twice the mean firing rate. Photon fluxes were reduced by — 10" in the model calculations to simulate the "Night" state of the eye. The neural images represent snapshots of the responses of the 16X16 array of ommatidia to the visual stimulus. The third column shows the computed neural images of synaptic activity in the brain after "temporal integration" of the retinal neural image with an integration time of 400 ms. The synaptic activities are mapped onto a gray scale with black set to 0 mv and white set to twice the mean amplitude of the synaptic potential. The fourth column displays the computed neural images of synaptic activity in the brain after "spatial summation" within the excitatory centers of the presumptive receptive fields of laminar cells. Note that at night, phototransduction noise obscures the neural images of the visual stimuli, but temporal and spatial integration partially recovers them. VISION IN HORSESHOE CRABS 171 puted responses are typically greater than 95% (n == 5) under controlled laboratory conditions but lower for field experiments because of the difficulty of precisely determin- ing the stimulus to the recorded ommatidium. Once satisfied with the accuracy of the model's predictions, we analyze the neural images for information the eye sends to the brain when the animal sees. The Limulus Eye Functions as a Global Feature Detector The eye transmits to the brain robust "neural images" of objects having the size, contrast, and motion of potential mates (Passaglia et ai. 1997a). Inspection of the neural images computed for the daytime state of the eye in Figure 2 shows that the eye is highly sensitive to images of crab- EYE Visual stimulus Neural image BRAIN Temporal integration Spatial integration 172 Circadian rlnthms in the Limulus lateral eye R. B. BARLOW ET AL. Table 1 Retinal property Day Night Reference Efferent input Gain Absent Low Present High Barlow el ai. 1977; Barlow, 1983 Renninger el a/.. 1984; Barlow el ai. 1987 Noise High Low Barlow el ul.. 1977; Kaplan and Barlow. 1980; Barlow el ai. 1993a Quantum bumps Short Long Kaplan et ai. 1990 Frequency response Fast Slow Batra and Barlow, 1990 Dark adaptation Fast Slow Kass and Berent, 1988 Lateral inhibition Strong Weak Renninger and Barlow. 1979; Ruta et ai. 1999 Cell position Proximal Distal Barlow and Chamberlain, 1980; Barlow et ai. 1980 Pigment granules Clustered Dispersed Barlow and Chamberlain, 1980 Aperture Constricted Dilated Chamberlain and Barlow, 1977, 1987 Acceptance angle 6° 13° Barlow et ai. 1980 Photomechanical movements Trigger Prime Chamberlain and Barlow. 1987 Photon catch Low High Barlow et ai. 1980 Membrane shedding Trigger Prime Chamberlain and Barlow. 1979. 1984 Arrestin mRNA level High Low Battelle et ai. 2000 Intense light effects Protected Labile Barlow et ai. 1989 Visual sensitivity Low High Powers and Barlow. 1985; Herzog et ai. 1996 size objects moving within the animal's visual range at about the speed of a horseshoe crab (15 cm/s). Indeed, measurements of the spatial and temporal transfer functions of the eye using linear systems analysis show that it func- tions as a tuned spatiotemporal filter. These filtering prop- erties can readily account for the animal's ability to see high-contrast objects but not low-contrast ones. Natural fluctuations of underwater lighting enhance the visibility of low-contrast objects. Beams of light created by overhead waves strobe the underwater scene in a range of frequencies ( — 2-6 Hz) for which the temporal transfer function shows the eye is maximally sensitive. Such wave-induced flicker increases the visibility of low-contrast, crab-sized objects during the day, as observed in field studies (Passaglia et al., 1997b; Krutky et ai, 2000). The strobic light evokes co- herent bursts of nerve impulses from clusters of neighboring ommatidia as the object moves across the visual field. These coherent bursts of activity are equal in amplitude to those evoked by moving, high-contrast objects, which is consis- tent with the animal's ability to detect mate-like objects regardless of their contrast. Stationary objects, either high or low contrast, are hardly recognizable in the computed neural images (not shown). Limulus vision requires both stimulus motion and activ- ities generated over ensembles of retinal receptors. The neural code for moving crab-like objects is not found in ambiguous messages of individual optic nerve fibers but in the coherent activity of small ensembles of nerve fibers. Such "distributed coding" has been detected in amphibian (Warland et al., 1997; Brivanlou et al., 1998) and mamma- lian visual systems (Field, 1994; Alonso et al.. 1996; Berry et al.. 1997; De Vries, 1999; Meister and Berry, 1999). In Limulns. stimulus motion binds these coherent activities together, sending a robust signal to the brain about potential mates. Endowed with spatiotemporal filtering properties, the eye is a sensitive detector of moving, crab-sized objects. A Circadian Clock Modulates Lateral Eye Sensitivity The Limulus eye operates in two distinct states: daytime and nighttime. It not only responds to changes in illumina- tion, it anticipates them. At dusk a circadian oscillator in the brain transmits efferent optic-nerve signals to the lateral eye, influencing almost every physiological and anatomical property of the retina (Table 1 ). The endogenous rhythms of the retina combine with mechanisms of light and dark adaptation to increase visual sensitivity by about 106 at night, nearly compensating for the decrease in the intensity of illumination in the animal's marine environment. In our initial theoretical analysis of retinal coding discussed above, we constructed a computa- tional model to simulate the daytime state of the eye. To examine the retinal coding that underlies vision at night, we must modify our computational model so that it can account for the circadian changes in lateral eye function. We have developed a "nighttime" model that includes most of the circadian changes listed in Table 1. Photon Noise Dominates Retinal Responses at Night Neural images of mate-like objects are less clear at night. They are dominated by bursts of spikes triggered by random photon events that characterize low nighttime levels of illumination (Hitt et al., 2000). The high-contrast object is detectable in the computed neural image in Figure 2, but the low-contrast object is almost completely obscured by ran- dom photon events. Computations require the setting of VISION IN HORSESHOE CRABS 173 ou "55 n 8 20 Q. E CD c/3 8. 10 ' I T Jl 1 Cfl L CD r1 o: J u - - - - - A / r r y °c — I 246 Seconds B Figure 3. (A) Response of a single optic nerve fiber to a black mate-like object moving across the visual field. Plotted are the instantaneous frequencies, reciprocals of the intervals between nerve impulses, of the train of spikes over a 7-s period. The highly variable discharge reflects phototransduction noise caused by random photon events at low nighttime light levels. (B) Average of the responses of the single optic nerve fiber to 15 repetitions of the visual stimulus. Modulation by the moving black object is evident in the average response but not in the single one. initial conditions, and an important initial condition is the mean light level of the visual scene. To simulate the night- time state we reduce the mean light level by 106 relative to daytime levels. At such low light levels the Poisson nature of photon absorption events produces substantial photo- transduction noise that can obscure signals generated by visual stimuli. The response of a single optic nerve fiber in Figure 3A recorded under nighttime light levels is indeed variable. It shows no clear sign that a high-contrast crab-sized object had moved across the visual field. This experiment was performed in the laboratory because the Crabcam is insen- sitive to the low light levels in the animal's habitat at night. To overcome this problem we simulated nighttime condi- tions by aligning an animal in front of a monitor that played back CrabCam recordings made in the animal's habitat. The output of the monitor was attenuated by 106 with neutral density filters to approximate nighttime levels of illumina- tion. The highly variable optic nerve response shown in Figure 3A is consistent with the noisy neural images in Figure 2 that were computed with the nighttime model of the eye. Brain Processing Enhances Retinal Signals How does the brain extract a reliable signal from such a noisy retinal input? Optic nerve fibers carrying the retinal signals synapse on neurons of the lamina of the brain. Single-cell recordings show that the laminar synapses inte- grate retinal signals with a time constant of about 400 ms (Passaglia et a!., 1997a). To assess the effect of these "slow" synapses we added a stage of temporal integration to the computational models. Figure 2 shows that synaptic inte- gration with such a long time constant suppresses the high- frequency fluctuations in optic nerve activity both day and night. The effect is especially striking in the neural images computed for the daytime state where temporal integration nearly recovers the visual stimulus. It is interesting that in another invertebrate visual system, that of the fly Lucilia cuprina, temporal integration reliably recovered the stimu- lus-induced response component from noisy neuronal sig- nals in the motion pathway (Warzecha and Egelhaaf, 1997). Returning to Limulus, temporal integration of the neural images generated by the retina in its nighttime state does not recover the visual stimulus. The burstiness evoked by ran- dom photon events (Fig. 2) remains prominent and obscures information about the visual stimulus. Because horseshoe crabs can see potential mates nearly as well at night as during the day. the brain must possess additional neural mechanisms for processing the retinal input, although these mechanisms need not be located in the lamina. One such mechanism could be spatial integration within the receptive fields of laminar neurons (Hitt et al., 174 R. B. BARLOW ET AL. 2000). Although the dimensions of receptive fields in the lamina have not been mapped with precision, we assumed for preliminary calculations that each laminar neuron sums optic nerve inputs from a 3 X 3 matrix of retinal receptors. Figure 2 shows that adding such a stage of spatial integra- tion significantly improves the signal-to-noise properties of the neural images in the brain computed for the nighttime state of the eye. Integrating the optic nerve responses from an array of retinal receptors viewing partially overlapping regions of visual space is nearly equivalent to averaging the responses from a single receptor. Indeed, Figure 3B shows that averaging responses of a single receptor to repeated visual stimuli yielded a clear modulation of optic nerve activity, whereas a single response exhibited no detectable modulation. We conclude that circadian increases in the sensitivity of the lateral eye in combination with lowpass spatial and temporal filtering in the brain can yield detect- able visual signals in the presence of high phototransduction noise caused by low nighttime light levels. The circadian and neural integrative mechanisms may help explain how Limitlus can see so well at night. What Is the Neural Basis of Behavior? How does the intricate circuitry of a nervous system receive sensory information, process it, and generate a be- havioral response? Analyzing a relatively simple nervous system may yield important insights about the functioning of more complex ones. Indeed the visual system of Limulus has proven complex enough to be interesting, yet simple enough to be understood. Using a computational approach, we unraveled its coding properties and determined the neu- ral image it sends to the brain about behaviorally relevant stimuli during the day. The Limulus eye, however, turned out not to be so "simple" after all. A circadian clock increases its sensitivity at night, enabling the animal to detect potential mates, a critical task that it performs equally well day and night. The clock does so by modulating almost every property of the retina, from stabilizing rhodopsin to weakening lateral inhibition and increasing photoreceptor gain (see Table 1 ). The challenge addressed in this paper is to understand how the eye efficiently encodes information about potential mates under the photon-limited conditions of the animal's marine environment at night. The answer in part appears to be that coding mechanisms in the eye to- gether with integrative mechanisms in the brain overcome environmental noise to enhance the neural images of behav- iorally important visual stimuli. The neural basis of visually guided behavior has been studied extensively in another invertebrate, the fly. Partic- ular attention has been paid to understanding how the fly visual system adapts to. encodes, and processes natural stimuli (Review: Rieke et ui, 1997). Adaptative mecha- nisms in the fly retina appear to enhance the efficiency of coding information about natural scenes (Review: Laughlin, 1994). In more central pathways, adaptive motion-sensitive mechanisms rescale the dynamic range of neural responses to match that of stimuli and thereby maximize information transmission (Brenner et ai, 2000). Adaptation of the mo- tion-sensitive mechanisms decreases contrast sensitivity while preventing saturation of neural responses and preserv- ing receptive field response properties (Harris et e shape, optical axes and resolution. J. Comp. Physiol. A 173: 565-5^2 Marshall. J.. and J. Oberwinkler. 1999. The colourful world of the mantis shrimp, \arure 401: 873-874. Marshall. N. J.. M. F. Land. C. A. King, and T. W. Cronin. 1991a. The compound eyes of mantis shrimps i Crustacea. Hoplocarida. Sto- matopoda). I. Compound eye structure: the detection of polarised light- Phil. Trans. R. < . Ser. B 334: Marshall. N. J.. M. F. Land. C. A. King, and T. W. Cronin. 1991b. The compound eyes of mantis shrimps (Crustacea. Hoplocarida. Sto- matopoda i. n. Colour pigments in the eyes of stomatopod crustaceans: polychromatic vision by serial and lateral filtering. PhiL Trans. R. Soc. Ser. B 334: 57-84. Marshall. N. J.. J. P. Jones, and T. \V. Cronin. 1996. Behavioural evidence for color vision in stomatopod crustaceans. J. Comp. Physiol. A 179: 4-3-481. Milne. L. J.. and M. Milne. 1961. Scanning movements of the stalked compound eyes in crustaceans of the order Stomatopoda. FT 422— 42< in Progress in Photobiology. B. C. Christensen and B- Bnchmann. eds. Proceedings of the Third International Congress on Photobiology. Elsevier. New York. Osorio. D.. N. J. Marshall, and T. \V. Cronin. 1997. Stomatopod photoreceptor spectral tuning as an adaptation for colour constancy in water. Vision Res. 37: 3299-3309. On. M.. and F. Schaeffel. 1995. A negatively powered lens in the chameleon. Sature 373: 692-694. Pettigrew. J. D.. S. P. Collin. and M. On. 1999. Convergence of specialised behaviour, eye movements and visual optics in sandlance iTeleosteii and the chameleon. Curr. Biol. 9: 421--1- Srinivasan. M. V.. S. Zhang, and J. S. Chahl. 2001. Landing strategies in honeybees and possible applications to autonomous airborne vehi- cles. Biol. Bull. 200: 216-221. Waterman. T. H. 1981. Polarization sensitivity. Pp. 281-469 in Han, book of Sensory Physiology. Vol. V1I/6B. H. Autrum. ed. Springer. New York. Wolff. L. B. 1997. Polarization vision: a new sensory approach to image understanding. Image Vision Comput. 15: 81-93. Yamaguchi. T.. Y. Katagiri. and K. Ochi. 1976. Polarized light re- sponses from retinular cells and sustaining fibers of the mantis shrimp. Biol J. Oka\ama L'nh. 17: 61-66. Reference: Biol. Bull. 200: 184-189. (April 2001) View From the Boundary BARBARA WEBB Center for Cognitive and Computational Neuroscience, Department of Psychology, University of Stirling, Stirling FK9 4LA, Scotland, UK Abstract. Re-implementing biological mechanisms on ro- bots not only has technological application but can provide a unique perspective on the nature of sensory processing in animals. To make a robot work, we need to understand the function as part of an embodied, behaving system. I argue that this perspective suggests that the terms "representation" and "information processing" can be misleading when we seek to understand how neurobiological mechanisms carry out perceptual processes. This argument is presented here with reference to a robot model of cricket behavior, which has demonstrated competence comparable to that of the insect, but utilizes surprisingly simple central processing. Instead it depends on sensory interfaces that are well matched to the task, and on the link between environment, action, and perception. Introduction The intersection of biology and robotics — the position of my own research — is often characterized as taking informa- tion from neuroethological investigations of natural systems to implement as new technology for man-made systems. However, another aspect of work in this area is to use the robotic implementations as a means of exploring biological hypotheses (Webb. 2000). This approach can provide a perspective on fundamental issues that is complementary to the view of the biologist engaged in primary research on the animal. This includes ideas on the most promising routes by E-mail: b-h.webh@stir.ac.uk This paper was originally presented at a workshop titled Invertebrate Sensory Information Processing: Implications for Biologically Inspired Autonomous Svstems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences. Woods Hole. Massachusetts, from 15-17 April 2000. was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. which biological understanding might inform technological developments. My main thesis will be that examining invertebrate sen- sory systems from this perspective teaches us that they do not actually do much "information processing" or "repre- sentation"— depending, of course, on how you define these terms (see below). When we look at invertebrates, it appears that the function of the sensory systems is not to inform the animal generally but to control specific behaviors; that the means by which they do so is often determined as much by peripheral sensory physics as by central computation; and that appreciating the problem in terms of an embodied animal interacting with an environment is more appropriate than approaching it in terms of building an internal repre- sentation of the external stimuli. Wehner (1987) used the term "matched filters" to describe how animals may be faced with problems that apparently need sophisticated in- formation processing solutions, but actually solve them by exploiting sensor mechanisms and behaviors that are uniquely matched to the required tasks. Further examples presented in this collection of symposium papers included the simple visual variables exploited by the bee to control flight (Srinivasan, 2001), and the use of "fanning" by cray- fish (Breithaupt, 2001) or moth (Ishida, 2001) to improve chemical plume tracking. Given that the terms "representation" and "information processing" are nevertheless commonly used by inverte- brate neuroethologists (e.g., during the symposium Thomas Cronin discussed the scanning movements of the mantis shrimp eye as implying a relatively sophisticated system for registering the information properly onto a subjective rep- resentation of space [Cronin and Marshall, 2001]), a dis- tinction may need to be drawn between this usage and the kind of full-blown symbolic encoding and manipulation that characterizes the "information processing" view of percep- tion and cognition in traditional Artificial Intelligence. The claim that a pattern of neural firing represents a stimulus is 184 VIEW FROM THE BOUNDARY 18 often only a claim that the firing and the presence of the stimuli causally co-vary: in the same way the electric cur- rent in a wire might be said to represent the position of an on-off switch. However, some deplore this as a misuse of the term "representation." For example. Maze (1981; p. 87) says that "the connection between the brain state and the external fact the knowledge of which it subserves ... is just that of cause and effect, not representation." and Clancey (1991: p. 110) argues that "structures in the brain that cannot be perceived [by the agent] have no representational status to the agent." The background for this disagreement over appropriate usage reflects two distinguishable senses of the relationship of representation. The first I will call "intentional represen- tation." defined in the theory of signs by Peirce (cited in Fetzer, 1988) as "something that stands in for something (else) in some respect or other for somebody" (p. 1 34) — for example, use of the term "LGN" by a scientist to represent a part of the brain. The second I will call "causal represen- tation," which describes indirect or mediated presentation, for example, the activity of ganglion cells presenting retinal stimulation patterns to LGN. The critical distinction be- tween these two is that the "intentional" case requires that the thing represented can be directly experienced by the representor: the scientist can hear the sound "LGN." or look at the brain part, and this is why he can use one to represent the other. In the "causal" case the LGN cannot access the retinal activity independently — for example, to confirm that the "representation" by the ganglion cells is correct. To illustrate the distinction another way: an ant may use a pattern of landmarks as a representation of a nest position, in which case it can know about the presence of the land- marks and the presence of the nest in the same way (i.e.. through its senses). If the ant is also said to use the response pattern of neurons in its brain to "represent" the presence of the landmarks, the ant's relationship to the neural firing and to the landmarks are not comparable. We are using different levels of description when we say it "recognizes" the land- marks or "recognizes" the pattern of neural firing. (A pos- sible source of confusion here is that looking at the ant's behavior and its neural processes from our point of view, we may well find that one (the firing) seems to stand in for the other (the landmark): but this is "intentional representation" only to the experimenter; to the ant it is merely "causal representation.") Similarly, there are distinctions to be drawn between usages of "information processing." There is the formal communication theory sense as defined by Shannon (1948); there is the everyday sense in which information is taken to be something containing meaning; and then there is the more recent identification of information processing with computation — that is, involving syntactic manipulation. None of these maps directly onto the usage whereby, for example, lateral inhibition in the retina is called information processing (there is no well-defined sender, receiver, or probability function; the meaning is opaque in the same way that the "representation" is non-intentional: and the process- ing is governed by physical rather than syntactic rules). The more apposite term here would seem to be signal process- ing, but "information processing" has become ubiquitous. Do these distinctions matter, or are they mere semantics? I would argue they are important because the explanatory power of applying the terms is very different. It is an empirical, and somewhat controversial, hypothesis to say that invertebrate behavior is controlled by intentional inter- nal representations, manipulated in meaningful information processing. Whereas to say that behavior is controlled by "causal" representations and involves "signal" processing is merely to say that the activity of the nervous system has a role in controlling the behavior, which was not in doubt. The same point has been expressed by Beer (2000. p. 97) with regard to cognitive science: "If any internal state is a representation and any systematic process is a computation then a computational theory of mind loses its force." Moreover, it is not always clear that insect neuroetholo- gists, in their usage, are not drawing conclusions that rest on conflating the meanings. An example is the tendency to start from the observation that an animal behaves differently in the presence of some stimulus, go on to describe the process involved as the animal internally "identifying" that stimulus before responding to it, and from this end up looking inside the brain for the neural mechanism that carries out the "identification." If the use of "identify" is only metaphori- cal, then it should not constrain the interpretation of find- ings, but it does. As an illustration, we can consider a classic piece of neuroethology in cricket phonotaxis research, the discovery of "recognition" neurons in the cricket brain whose firing rate response corresponds remarkably well to the likelihood of tracking by the cricket when it is presented with songs of different syllable rates (Schildberger. 1984a). Although this discovery is certainly of significance in trying to disentangle the neural wiring underlying the behavior, this "representation" of the "attractiveness" of syllable rates by the firing rate of an identified neuron is by no means an explanation of the behavior. First, it is not surprising, given that the animal behaves in different ways to different songs, that we find some neurons active under conditions when it does respond and not active when it does not — this is simply to say that its motor behavior is under some kind of neural control. Furthermore, the result does not in itself tell us how the neuron comes to have this property: understanding the mechanism of "recognition" requires understanding the neural connectivity leading to this property, which to date is still not fully resolved. Finally, in the neural model de- scribed below, we found highly comparable property of correlation of firing rate in certain neurons with syllable rate preference — yet the firing rate here had no functional role in the behavior but was simply a side-effect (Fig. 1). In fact. 186 B WEBB 42 50 58 66 74 Syllable Repetition Interval (ms) Figure 1. The "firing rate" of a neuron in the robot model matches the "phonotactic preference" displayed in behavior. This looks like the "rec- ognition" neuron discovered in the cricket (Schildberger, 1984) but in fact plays no functional role in the behavior. (Adapted from Webb and Scutt, 2000.) there need not be any explicit "identifier" in the brain for the animal to single out and approach a specific signal, as I will now describe in more detail. Modeling Cricket Behavior Cricket phonotaxis — the ability of females to track down male calling songs — includes a significant range of the problems of responding appropriately to specific sensory signals: identifying the signal against a noisy background; recognizing that it is the correct one; localizing its source; possibly choosing between rival signals. An information processing approach to this problem identifies the problems to be solved by the cricket's neural system as filtering for the right carrier frequency and filtering for the right repeti- tion rate to recognize the signal (Popov and Shuvalov, 1977; Thorson el ai, 1982; Stout and McGhee, 1988); comparing the amplitude of the auditory signal between two sensors to determine the direction of the source or at least which way to turn (Schmitz el ai, 1982; Schildberger and Horner, 1988; Huber. 1992); and separating simultaneous sound sources sufficiently to assess and approach the more attrac- tive one (Doherty, 1985; Simmons, 1988; Pollack, 1998). However, closer examination of the peripheral sensing system in the animal suggests that it may solve at least some of these problems directly, without any explicit representa- tion of the song. The pressure difference receiver mecha- nism that enables the animal to detect the sound direction (Michelsen et ai, 1994) is inherently dependent on that sound being within a particular range of wavelengths. The neural encoding of the subsequent intensity difference be- tween the ears is potentially in the form of a temporal code (Schildberger, 1984b; Stumpner et ai, 1995) that could explain the pattern dependency of the response. Finally, the animal's behavior in response to sound will position it in the sound field in such a way that it is likely to end up at the most attractive source rather than confused between them (see below). In other words, the behavior does not require any internal representation of the nature or position of the sound source. That this is indeed possible has been demonstrated in a robot implementation of this suggested mechanism for pho- notaxis (Webb, 1995; Lund et ai. 1997; Webb and Scutt, 2000). The robot has an auditory system that, like the cricket's ears, uses cross-delay and summation of the two signals to produce a strongly directional response despite small receptor separation. Because the delay is fixed, the wavelength of the signal is a crucial determinant of the effectiveness of the device. Thus the robot will, for exam- ple, locate a 4.7-kHz signal better than one at higher or lower frequencies, and will preferentially approach a 4.7- kHz signal when a song of differing frequency is simulta- neously presented, with no other form of frequency filtering. The behavior of the robot is controlled by a spiking neural network consisting of only four units. Two input units integrate the auditory signal and initiate firing above a threshold (their behavior is closely modeled on the response properties of identified neurons [AN1] in the cricket). They respectively excite two output units, but cross inhibit each other's axons. Thus the unit that fires first effectively sup- presses the effect of the other side. The input-output con- nection is further modulated by synaptic suppression — that is to say, successive spikes have progressively less effect on the postsynaptic membrane potential, unless there is a gap in which the synapse can recover. The result is that unless the input has an appropriate on-off pattern, it is not effective in generating an appropriate motor response as controlled by the output units. For example, the robot will show consistent tracking behavior only to songs that fall within a particular band of syllable repetition rates, the same as that preferred by the cricket (Fig. 2). Although this behavioral preference has a corresponding neural "representation" in the firing rates of the output units (Fig. 1), the actual explanation of the behavior lies in the interactions of the neural time courses of summation and decay, and indeed these generate the appropriate response much faster than the time that would be needed to get a reasonable estimate of the firing rate. Having the model implemented in a physical device allowed us to test the behavior in realistic sound fields that would be difficult to simulate convincingly. Further char- acteristics of cricket behavior could thus be shown to emerge from the interaction of the controller, the physical interface, and the environment, without requiring further elaboration of the model. With sound from directly above (i.e., lacking any horizontal directional difference), the VIEW FROM THE BOUNDARY SRI=18ms SRI=26ms 18 i SRI=34ms 45 0 -45 ( §45 n ? 0 1-45 ( 45 0 -45 ( 45 0 -45 ^^ 45 0 ^»5 AjvwAA7VY^NV\/vV' 5 10 0 5 10 0 5 10 SRI-42ms SRI-50ms SRI=58ms AAAAAAAyVMAAA/vw 45 0 MAAAAAAMAAAAM 45 0 -45 ywvwvww 5 10 0 5 10 0 5 1C SRU66ms SRI=74ms SRI«82ms r—H 45 0 45 0 -45 J ) 5 10 0 5 10 0 5 1C seconds Figure 2. Tracking behavior of the robot in response to cricket songs at different syllable repetition intervals (SRI). The sound is at 45 degrees to the starting position of the robot. A SRI between 26 and 58 ms (comparable to the cricket) is needed for the robot to consistently turn and meander in the sound direction. (From Webb and Scutt, 2000.) robot, like the cricket (Weber et al., 1 981), showed a ten- dency to perform tracking-like behavior without actually following one consistent direction. When the sound from above was paired with a continuous (i.e., unattractive) stim- ulus from one side, the robot, like the cricket (Stabel et al.. 1989), tracked away from the lateral stimulus. When two similar sounds were played simultaneously, the robot could choose and track one of them because once it had turned slightly more to one side; the sound from that side captured the response. If the sounds differed slightly in temporal 90 45 g o _SRI SRI=60 15 20 25 90 45 _SRI=60 -45 -90 SRI=40 10 15 seconds 25 Figure 3. Tracking behavior of the robot to simultaneous cricket songs at different syllable rates. The robot (like the cricket) turns and tracks the faster repetition rate (SRI = 40) whether it is on the left (upper plot) or the right (lower plot). (Adapted from Webb and Scutt, 2000.) 188 B WEBB pattern, the robot, like the cricket (Doherty, 1985), could consistently choose one as the more attractive signal (Fig. 3). Integrating Sensory Systems One argument advanced in favor of (real) information processing solutions is that they are more amenable to scaling up to explain more complex, flexible behaviors such as the integration of different sensory sources to control behavior. From an engineering or designer point of view, this might indeed be the case. Whether it is true of biology is another question: perhaps biological systems can offer us alternative schemes — perhaps more specialized to the ani- mal's task niche, but on the other hand flexible and robust — for solving these kinds of problems. As a preliminary start- ing point for investigating these issues, I will describe some recent work done in collaboration with Reid Harrison (Webb and Harrison. 2000a,b) to look at the integration of the phonotaxis behavior on the robot with another funda- mental sensorimotor reflex, the optomotor response. Like many other insects, crickets will rotate in response to rotation of their visual surroundings. Normally this serves as a basic stabilization mechanism. The underlying sensor and neural circuitry for this response has been closely studied, particularly in the fly (Gotz, 1975; Reichardt and Poggio, 1976: Heisenberg and Wolf, 1988; Egelhaaf and Borst, 1993). It has been suggested that, in lit conditions, crickets will additively integrate their phonotaxis response and their optomotor response (Bohm et ai, 1991). which could improve the accuracy of their approach to sound (Weber el ai. 1981) by controlling for unintended course deviations. A sensor that embodies the hypothesized mechanism of the optomotor response has been built in analog VLSI (very large scale integration) hardware (Harrison and Koch, 1998). This is a single chip that contains photoreceptors, temporal filters, comparison units, and widefield summa- tion. The output can be used as a "torque" signal for the direction and approximate velocity of motion that would compensate for the visual rotation. We interfaced this chip to a robot that also had the sound-sensing circuit and neural model for phonotaxis described above. The two behaviors were initially combined in a directly additive way; that is. the motor output was a weighted sum of the signal given by the phonotactic turning decision and the signal given by the optomotor torque. However, this caused some problems, because turns in response to sound would generate strong visual rotation signals that the robot would attempt to cor- rect, thus negating the initial turn. As a second approach, we used an inhibition scheme in which the robot would ignore the optomotor signal while turning in response to sound (other possible solutions are discussed in Webb and Harri- son, 2000b). 'EBSOUND SOURCE - -150 -100 x[cml 0SOUND SOURCE - Figure 4. Tracks of a robot with a 20% motor bias. Top, using phonotaxis only. Bottom, with an optomotor response added. The optomo- tor behavior significantly improves the ability to go directly to the sound source. (From Webb and Harrison. 2000b.) With this simple interaction scheme it was possible to show that the added optomotor capability could signifi- cantly improve phonotaxis, more obviously so under con- ditions where the motor capability was made less reliable. Thus Figure 4 shows the behavior of the robot when ap- proaching a sound source with an induced bias in its motor output that makes the left wheel turn 20% faster than the right. Without the optomotor response the robot had some difficulty reaching the speaker: with the response added it successfully and directly reached the speaker on all but one trial. Because the two hardware sensor systems are well tuned to executing their specific tasks, it was relatively simple to combine the behaviors to produce a robust per- formance without any explicit representation of the "fused" auditory and visual information. VIEW FROM THE BOUNDARY Conclusion Robotics engineers already know a lot about information processing on representations. It is the standard computa- tional paradigm, but it has proved difficult to employ to get robots to display behavioral competence comparable to even "mere" invertebrates. What they can learn from biol- ogy is how to build smart sensors that are matched to tasks; how to devise control systems that include patterns of behavior as part of the sensing process; and how to design internal nervous systems that exploit these factors. Calling these latter kinds of processes "representation" and "infor- mation processing" obscures the distinctive character of the mechanisms on offer. There is much yet to learn about the interplay of environments, behaviors, physics, and physiol- ogy. Biologists may have as much to learn from attempts to implement these mechanisms as do engineers. Literature Cited Beer, R. D. 2000. Dynamical approaches to cognitive science. Trends Cognit. Sci. 4: 91-98. Bohm, H., K. Schildberger, and F. Huber. 1991. Visual and acoustic course control in the cricket Gryllus bimaculatus. J. Exp. Biol. 159: 235-248. Breithaupt, T. 2001. Fan organs of crayfish: devices supporting chem- ical information exchange with the environment. Biol. Bull. 200: 150- 154. Clancey, W. J. 1991. Situated cognition: stepping out of representational flatland. Ai Commun. 4: 109-112. Cronin, T. W., and J. Marshall. 2001. Parallel processing and image analysis in the eyes of mantis shrimps. Biol. Bull. 200: 177-183. Doherty, J. A. 1985. Phonotaxis in the cricket, Gryllus bimaculatus De Geer: comparisons of choice and no choice paradigms. J. Comp. Physiol. A 157: 279-289. Egelhaaf, M., and A. Borsl. 1993. A look into the cockpit of the fly — visual orientation, algorithms, and identified neurons. J. Neurosci. 13: 4563-4574. Fetzer, J. H. 1988. Signs and minds: an introduction to the theory of semiotic systems. Pp. 133-161 in Aspects of Artificial Intelligence, J H. Fetzer. ed.. Kluwer, Berlin. Gotz, K. G. 1975. The optomotor equilibrium of the Drosophila navi- gation system. J. Comp. Physiol. A 99: 187-210. Harrison, R. R., and C. Koch. 1998. An analog VLSI model of the fly elementary motion detector. Pp. 880-886 in Advances in Neural In- formation Processing Systems 10. M. I. Jordan, M. J. Kearns. and S. A. Solla, eds., MIT Press, Cambridge, MA. Heisenberg, M., and R. Wolf. 1988. Reafferent control of optomotor yaw torque in Drosophila melanogaster. J. Comp. Physiol. A 163: 373-388. Huber, F. 1992. Behavior and neurobiology of acoustically oriented insects. Naturwissenschaften 79: 393-406. Ishida. H., T. Nakamoto, T. Moriizumi, T. Kikas, and J. Janata. 2001. Plume-tracking robots: a new application of chemical sensors. Biol. Bull. 200: 222-226. Lund, H. H.. B. Webb, and J. 1 1. ill, mi 1997. A robot attracted to the cricket species Gryllus bimaculatus. Pp. 246-255 in Fourth European Conference on Artificial Life. P. Husbands and I. Harvey, eds. MIT Press, Cambridge, MA. Maze, J. 1981. The Meaning of Behaviour. Allen and Unwin, London. Michelsen, A., A. V. Popov, and B. Lewis. 1994. Physics of directional hearing in the cricket Gryllus bimaculatus. J. Comp. Phvsiol. A 175: 153-164. Pollack. G. S. 1998. Neural processing of acoustic signals. Pp. 139-196 in Comparative Hearing: Insects, R. R. Hoy, A. N. Popper, and R. R. Fay, eds. Springer. Berlin. Popov, A. V., and V. F. Shuvalov. 1977. Phonotactic behaviour of crickets. J. Comp. Physiol. A 119: 111-126. Reichardt, W., and T. Poggio. 1976. Visual control in the fly. A quantitative analysis. Q. Rev. Bioi'hvx. 9: 311-375. Schildberger, K. 1984a. Multimodal interneurons in the cricket brain: properties of identified extrinsic mushroom body cells. / Comp. Physiol. A 154: 71-79. Schildberger, K. 1984b. Temporal selectivity of identified auditory in- terneurons in the cricket brain. J. Comp. Physio/. A 155: 171-185. Schildberger, K., and M. Horner. 1988. The function of auditory neurons in cricket phonotaxis. I. Influence of hyperpolarization of identified neurons on sound localization. / Comp. Physiol. A 163: 621-631. Schmitz, B., H. Scharstein, and G. Wendler. 1982. Phonotaxis in Gryllus campestris L. (Orthoptera. Gryllidae). I. Mechanism of acous- tic orientation in intact female crickets. J. Comp. Physiol. A 148: 431-444. Shannon, C. E. 1948. A mathematical theory of communication. Bell System Tech. J. 27: 379-423. Simmons, L. W. 1988. The calling song of the field cricket Gryllus bimaculatus (de Geer): constraints on transmission and its role in intermale competition and female choice. Anim. Behav. 36: 380-394. Srinivasan, M. V., S. Zhang, and J. S. Chahl. 2001. Landing strategies in honeybees and possible applications to autonomous airborne vehi- cles. Biol. Bull. 200: 216-221. Stabel, J., G. Wendler, and H. Scharstein. 1989. Cricket phonotaxis: localization depends on recognition of the calling song pattern. J. Comp. Physiol. A 165: 165-177. Stout, J. F., and R. McGhee. 1988. Attractiveness of the male Acheta domestica calling song to females. II. The relative importance of syllable period, intensity, and chirp rate. / Comp. Physiol. A 164: 277-287. Stumpner, A., G. Atkins, and J. F. Stout. 1995. Processing of unilateral and bilateral auditory inputs by the ONI and LI interneurons of the cricket Acheta domesticus and comparison to other cricket species. J. Comp. Phvsiol. A 177: 379-388. Thorson, J., T. Weber, and F. Huber. 1982. Auditory behaviour of the cricket. II. Simplicity of calling-song recognition in Gryllus and anom- alous phonotaxis at abnormal carrier frequencies. J. Comp. Physiol. A 146: 361-378. Webb, B. 1995. Using robots to model animals: a cricket test. Robot. Anton. Syst. 16: 117-134. Webb, B. 2000. What does robotics offer animal behaviour? Anim. Behav. 60: 545-558. Webb, B., and R. R. Harrison. 2000a. Eyes and ears: combining sensory motor systems modelled on insect physiology. Pp. 3913-3918 in IEEE International Conference on Robotics and Automation. San Francisco. 24-28 April 2000. Webb, B., and R. R. Harrison. 2000b. Integrating sensorimotor sys- tems in a robot model of cricket behavior. Sensor Fusion and Decen- tralised Control in Robotic Systems 111. SPIE, Boston, 6-8 Nov. 2000. Webb, B., and T. Scutt. 2000. A simple latency dependent spiking neuron model of cricket phonotaxis. Biol. Cybern. 82: 247-269. Weber, T., J. Thorson, and F. Huber. 1981. Auditory behaviour of the cricket. I. Dynamics of compensated walking and discrimination par- adigms on the Kramer treadmill. J. Comp. Physiol. A 141: 215-232. Wehner, R. 1987. Matched filters — neural models of the external world. J. Comp. Physiol. A 161: 511-531. Reference: Biol. Bull. 200: 190-194. (April 20(11) Innovative Biomechanics for Directional Hearing in Small Flies DANIEL ROBERT Laboratory of Bioacoustics, Institute of Zoology, University of Zurich, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland Abstract. In humans and animals alike, the localization of sound constitutes a fundamental processing task of the auditory system. Directional hearing relies on acoustic cues such as the interaural amplitude and time differences and also, sometimes, the signal spectral composition. In small animals, such as insects, the auditory receptors are forcibly set close together, a design constraint imposing very short interaural distances. Due to the physics of sound propaga- tion, the close proximity of the sound receivers results in vanishingly small amplitude and time cues. Yet. because of their directionality, small auditory systems embed original and innovative solutions that can be of inspirational value to some acute problems of technological miniaturization. Such ears are found in a parasitoid fly that acoustically locates its singing cricket host. Anatomically rather unconventional, the fly's auditory system is endowed with a directional sensitivity that is based on the mechanical coupling between its two hemilateral tympanal membranes. The functional principle permitting this directionality may be of particular relevance for technological applications necessitating sen- sors that are low cost, low weight, and low energy. Based on silicon-etching technology, early prototypes of sub-millime- ter acoustic sensors provide evidence for directional me- chanical responses. Further developments hold the promise of applications in hearing aid technology, vibration sensors, and miniature video-acoustic surveillance systems. E-mail: drobert@zool.umzh. eh This paper was originally presented at a workshop titled Invertebrate Sensory Information Processing: Implications for Biologically Inspired Autonomous Svstems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences, Woods Hole, Massachusetts, from 15-17 April 2000, was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. Introduction In the sense of evolution, sensory systems constitute "adaptive packages" of biological engineering that have been proven sufficiently refined and efficient by the com- bined processes of natural and sexual selections. From this evolutionary vantage point, the considerable diversity of insect auditory systems (Hoy and Robert, 1996) can be regarded as an array of original and adapted solutions to the particular problems of acoustic detection using small sen- sors. Because they evolved under the particular constraints imposed by small body size, insects are likely to be of significant inspirational value for general problems of min- iaturization. Our investigations are focused on questioning whether insects at the small end of the biometric range (millimeter and less) can be endowed with auditory organs, whether they can be directionally sensitive, and if so, what mechanisms are responsible for such directionality? Design Constraints for a Miniature Auditory System One such small insect is a parasitoid fly that acoustically locates and attacks singing field crickets (Cade, 1975). As part of their reproductive cycle, female parasitoid flies Ormia ochracea must find their cricket hosts as a source of food for their larval offspring. Remarkably, successful song localization can take place in complete darkness, that is, in the presence of acoustic information alone, but in the ab- sence of visual and olfactory cues (Ramsauer and Robert, 1999. 2000). Hence, because of its crucial role in reproduc- tion, audition is of great importance in the evolutionary history of these flies: promoting an increased reproductive success, efficient audition is likely to be under acute selec- tive pressure. Both detection and localization of the singing host are mediated by a pair of auditory sensory organs situated on 190 INNOVATIVE AUDITORY MICROMECHANICS the anterior thorax just above the first pair of legs, which are directly below the fly's neck (Robert ft al.. 1992) (Fig. 1 A). The ears of the fly are endowed with two thin tympanal membranes, the eardrums, that are set very close together across the midline of the animal. In fact, while the tympanal membranes together span about 1.2 mm in width, the mechanoreceptive sensory organs are separated from each other by only 520 /im (Robert et al.. 1994) (Fig. 1A). As expected from the fly's small size and the resulting minute interaural distance, the largest interaural time difference (ITD) generated by a passing sound wave does not exceed 2 /is (Robert et al., 1996). In the best possible case, for an azimuthal angle of inci- dence of 90° off the longitudinal axis of the animal, the ITD was measured with probe microphones to be 1 .45 /is (SD: 0.49 /is. n = 10) (Fig. IB). Such time difference is clearly much too small to be encoded by the nervous Contralateral -90 -60 -40 -20 0 20 40 60 Angle of Incidence [degrees] Figure 1. Auditory anatomy and temporal acoustic cues. (A) The auditory organs of the parasitoid fly Ormia ochracea are located on the anterior thorax, between the first pair of legs (L) and the neck (N). The tympanal membranes (TM) are adjacent to each other and set close to- gether by the midline of the animal (vertical dashed line). Providing a connection between the two TMs across the midline, the intertympanal bridge is made of thicker cuticle than TMs and has the shape of a coat hanger. Two depressions at both ends of the intertympanal bridge indicate the insertion point of the sensory organs (arrows). Arrows also point to the interaural distance. (B) Interaural time difference (ITD) as a function of the angle of incidence of the sound stimulus. Right ordinate: ITDs calculated for humans (ear separation of 1 70 mm). Left ordinate: ITDs at the fly's ears calculated for an interaural distance of 0.6 mm. Data point (with standard deviation) shows ITD measurement made at —90° azimuth and 5 kHz tone, with two probe microphones located at the TMs. system of the fly. For humans, in comparison, ITDs at such an angle of incidence are much larger (Fig. IB), varying from 500 /is to 700 /is, depending on head size, the sound path considered, and the frequency considered. It thus becomes apparent that microsecond-range ITDs, and possibly smaller ones, ought to constitute a major difficulty for the auditor) system of the fly, both at the mechanical and neural coding levels. Another difficulty arises when the second directional cue, the interaural intensity difference (IID), is considered. The behavior- ally relevant frequency range is that of the cricket's calling song, which spans from 4.6 to 5.2 kHz. Since at such frequencies the ratio of wavelength (66 mm at 5.2 kHz) to interaural distance (520 /im) is larger than 100, diffraction is very unlikely to occur around the fly's body, or at her ears (Morse and Ingard, 1968). Measure- ments made with custom probe microphones (100-/xm acceptance diameter) and with Vx-in microphones (Briiel & Kjaer type 4138, Naerum, Denmark) failed to indicate diffraction-elicited sound pressure differences around the fly that could constitute IID cues. Admittedly, the minuscule ITD (1.7 jus) and IID (<1 dB) generated from an incident sound wave constitute less than reliable acoustic cues for directional hearing. This finding, however, stands in sharp contrast to the behavioral capacity of this fly to localize the song of her host, as demonstrated in the field and laboratory (Robert et al., 1992; Walker. 1993; Ramsauer and Robert, 2000). Innovative Tympanal Mechanics Taking place at the peripheral level of the tympanal membranes, the first step in the process of hearing per- tains to the conversion of acoustic energy into mechani- cal energy. Thus, tympanal mechanics was examined in response to incident sound waves, either with pure tones mimicking cricket songs, or with band-limited white noise. Using microscanning laser vibrometry (Robert and Lewin, 1998), displacement velocities were assessed over several hundred measurement locations on both tympanal membranes and on the associated cuticular elements. In response to random noise as well as pure tone stimuli, the amplitude and phase responses of both tympanal mem- branes differ strongly from those expected for two inde- pendent pressure receivers set 0.5 mm apart (Fig. 2). Firstly, independent receivers would indicate immeasur- ably small differences in amplitude, whereas the fly's tympana undergo large differences in their vibrations above about 4.5 kHz (Fig. 2A). Secondly, independent receivers such as microphones indicate interaural time delays on the order of 2 /is (Fig. IB), an ITD significantly smaller than the delays (also measured as the phase lag at maximal deflection) observed in the mechanical response of the tympanal membranes (on average; 55 /is; SD: 12 192 D. ROBERT 1 3 5 7 9 11 13 15 17 19 21 23 25 Frequency [Hz] B. 100 1 1 1 1 1 1 1 1 1 1 3 5 7 9 11 13 15 17 19 21 23 25 Frequency [Hz] Figure 2. Mechanical response of the tympanal membranes in ampli- tude and time. (A) Difference amplitude spectrum of the mechanical response at the anatomical locations indicated in the inset (arrows). The contralateral mechanical response is subtracted from the ipsilateral one. (B) Time delay between the ipsilateral and contralateral mechanical responses, calculated from the difference phase spectrum. Sound stimulus at 45° incidence in azimuth, band-limited white noise (1-30 kHz). ;u.s, n = 769 frequency points) (Fig. 2B). In effect, these measurements indicate that the mechanical response of the tympanal membranes has a pronounced directionality. Interaural time and intensity differences in the mechan- ical response of the tympana are significantly larger than those available in the acoustic field. The analysis of tympanal deflection shapes (see Fig. 3A) provides evi- dence that such differences between the ipsilateral and contralateral responses are due to the particular mechan- ical properties of this small tympanal system. Reflecting the asymmetrical responses reported earlier (Fig. 2). de- flection shapes at frequencies around 5 kHz indicate that while the ipsilateral tympanum undergoes an outward deflection, the contralateral tympanum experiences an inward deflection, but of much lesser amplitude (Fig. 3B). Such asymmetry in the response is due to the fact that the two hemilateral tympanal membranes are not vibrating independently. The tympana are in fact con- nected, across the midline, by a specialized cuticular structure (Robert et al., 1994, 1996). The Process of Intertympanal Coupling In an effort to unite structure and function in a functional explanation, the reconstitution of actual tympanal deflec- tions provides key information on the anatomical basis for the observed mechanical behavior. The particularity — and perhaps uniqueness — of these ears is that they are physi- cally connected by an unpaired cuticular structure, the in- tertympanal bridge (Figs. 1A, 4A). This bridge is thicker (2-10 jrni), and therefore stiffer, than the surrounding tym- panal membranes (0.2-1 /urn) to which it attaches. Thus, the tympana are mechanically linked by a relatively rigid cu- ticular bridge, a fact that can also be easily assessed by gently deflecting one tympanic membrane downwards with midline contra ipsi midline contra 300 Figure 3. Tympanal deflection shapes. (Al Tympanal vibration veloc- ity was monitored at 364 locations over tympanal and non-tympanal areas. Coherence values exceeded 0.95, indicating little contamination by uncor- related noise. Stimulus: band-limited random noise (1 to 30 kHz; 94 dB SPL (sound pressure level]) at 90° azimuth to the longitudinal axis of the fly. Ipsilateral and contralateral conventions as indicated. (B) Upper panel shows maximum outward deflection of the ipsilateral tympanal membrane at 5 kHz excitation frequency. Lower panel depicts the maximum inward deflection reached half a period later, e.g., 100 /J.s later. X axis: lateral; Y: ventral; Z: axis of measured tympanal deflections. For animation of de- flection, please visit: www.unizh.ch/~ormia/research_flies.html INNOVATIVE AUDITORY MICROMECHANICS The intertympanal bridge B. p^t) Flexible hinge F2(t) IRdl ] — Fulcrum mode 1 Figure 4. The anatomy of mechanical coupling, and vibrational modes of the flexible intertympanal bridge. (A) Close-up of the intertympanal bridge connecting the tympanal membranes. (B) Simple mechanical model of the bridge as a seesaw endowed with two rigid bars connected by a flexible central hinge ( = ). (C) On the basis of the laser vibrometric micromechanical analysis, it is suggested that two basic modes can char- acterize the observed mechanical response. Bending occurred at low fre- quencies (mode 1; <4 kHz), whereas rocking was measured at intermedi- ate frequencies (mode 2: 5-1 kHz). At higher frequencies (\5 kHz and above), bending and rocking modes combine to elicit motion in one tympanum only (mode 1 + 2). a human hair, and observing the other deflect upwards as a result. Deflection shapes obtained by microscanning laser Dopp- ler vibrometry for different stimulus frequencies (e.g., 2, 5 or 15 kHz) reveal that this micromechanical system can produce several different patterns of deflection that are reminiscent of the movements of a flexible seesaw (Fig. 4B). The simple mechanical model shown in Figure 4B has been proposed as a reasonable functional, and intuitively accessible, approximation of this unconventional peripheral auditory system (Miles et al, 1995: Robert et al. 1996). The physical action of the intertympanal bridge is to convert small acoustic ITDs into larger time and amplitude differ- ences at the mechanical level. The functional principle for this effect resides in the somewhat complex linear interac- tion between two coupled oscillators — the tympanal mem- branes. Very briefly, for low frequencies of stimulation (2 kHz), the deflections of the ipsilateral and contralateral membranes show little difference in amplitude and phase. In this case, the forces applied to the bridge (Fig. 4B) have a phase difference of only 1° and result in sympathetic de- flections, owing to the bending of the bridge (Fig. 4C, mode 1 ). At this frequency of excitation, the first mode of deflec- tion dominates the response. Excitation of this first mode results in little directionality — or asymmetry — in the me- chanical response, as shown for frequencies lower than about 3 kHz (Fig. 2A). As stimulus frequency increases, a transition to the second mode of deflection is expressed by the rocking motion of the bridge (Fig. 4, mode 2). Note- worthy is the fact that for intermediate frequencies (i.e., 5 kHz) and a given effective interaural distance, the phase difference in the driving forces increases. At 5 kHz, this phase is very small — 2.6° — yet, it is sufficient to drive this system to its rocking mode (Fig. 3B). A combination of these two modes (rocking and bending) will take place at even higher frequencies (e.g., \5 kHz) and result in the deflection of the ipsilateral tympanum, but in the silencing of the contralateral one (Fig. 4C, mode 1+2). Hence, the relative flexibility of the bridge, be it localized at the ful- crum point or distributed along the lateral arms, constitutes the key to the directionality of the observed mechanical response. In addition, experiments of mechanical actuation in the absence of sound have demonstrated that mechanical cou- pling is mediated and can be explained by the action of the intertympanal bridge alone (Robert et at., 1998). The me- chanical actuation of one tympanum at amplitudes of about 10 nanometers — mimicking the displacement amplitudes observed during acoustic stimulation — elicits a correlated mechanical displacement of the other tympanum. The finite flexibility of the intertympanal bridge accounts for the IIDs and ITDs in the mechanical deflections, for which two modes dominate at different frequencies (Robert et al.. 1996). Hence, in summary, the mechanical structure of the hearing organs increases the minimal acoustic interaural time and amplitude difference cues into more substantial mechanical cues that can be processed by the nervous system (Robert et al., 1996). Intertympanal Coupling as an Evolutionary Novelty As such, the process of mechanical coupling between tympanal membranes, and its amplification effect on inter- aural auditory cues, seems to be unique to flies. In essence, intertympanal coupling constitutes a novel mechanism for directional hearing, representing a third kind of directional receiver for terrestrial animals. One other kind of directional system is found in most vertebrates, and is based on two acoustically isolated pressure receivers set apart on the head of the animal. In other small animals, such as frogs (Narins et ui. 1988) and some birds (Calford and Piddington. 1988), 194 D. ROBERT the auditory receivers are acoustically coupled (pressure- difference receivers), a system that has been shown to result in the amplification of the perceivable ITDs and IIDs avail- able in the sound field. Conclusions One of the outcomes of the research presented here has been to explore the possibility of transferring the mecha- nism of intertympanal coupling to microsensor technology. Abstracting the functional principle unraveled in the para- sitoid fly may indeed contribute to the development of subminiature microphones. Besides the inherent direction- ality of the receiver's mechanics despite its small size, two other, definite and foreseeable, advantages of fly-inspired acoustic receivers would be their low cost and low energy consumption. Indeed, one of the possible attractive aspects of such a system is the fact that the first stage of acoustic processing is made on a mechanical basis, without the intervention of energy-consuming electronics. Using micro- electromechanical systems technology (MEMS), our early prototypes etched on silicon wafers provide evidence that coupled pressure receivers no larger than 500 ^un are di- rectionally sensitive at about 5 kHz. Further optimization promises the development of subminiature microphones endowed with directionality in the range of human speech. Finally, possible applications can be envisaged that could contribute to the development of hearing aids or miniature personal communication devices endowed with improved receiver acoustics, or small surveillance systems capable of pinpointing a source of noise. Acknowledgments This work was supported by the Swiss National Science Foundation. Literature Cited Cade, VV. H. 1975. Acoustically orienting parasitoid: fly phonotaxis to cricket song. Science 190: 1312-1313. Calford, M. B., and R. W. Piddington. 1988. Avian interaural canal enhances interaural delay. J. Comp. Physiol. A 162: 503-510. Hoy, R. R., and D. Robert. 1996. Tympanal hearing in insects. Anna. Rev. Emomol. 41: 433-450. Miles, R. N., D. Robert, and R. R. Hoy. 1995. Mechanically coupled ears for directional hearing in the parasitoid fly O. ochracea. J. Acousl. Soc. Am. 98(6): 3059-3070. Morse, P. M., and K. U. Ingard. 1968. Theoretical Acoustics. McGraw Hill, New York. Pp. 418-422. Narins, P. M., G. Ehret, and J. Tautz. 1988. Accessory pathway for sound transfer in a neotropical frog. Proc. Natl. Acad. Sci. USA 85: 1508-1512. Ramsauer, N., and D. Robert. 1999. Acoustic attraction of the parasi- toid fly Ormia ochracea to the song of its host. J. Acoust. Soc. Am. 105: 1319. Ramsauer, N., and D. Robert. 2000. Phonotactic orientation in a para- sitoid fly. Naturwissenschqften 87: 315-319. Robert, D., and A. Lewin. 1998. Microscannmg laser vibrometry ap- plied to the biomechanical study of small auditory systems. Pp. 564- 571 in Third International Conference of Vibration Measurements by Laser Techniques. Ancona. Italy. E. P. Tomasini. ed. Proceedings of the International Society of Optical Engineers. Vol. 341 1 . Bellingham, WA. Robert, D., J. Amoroso, and R. R. Hoy. 1992. The evolutionary con- vergence of hearing in a parasitoid fly and its cricket host. Science 258: 1135-1137. Robert, D., M. P. Read, and R. R. Hoy. 1994. The tympanal hearing organ of the parasitoid fly Ormia ochracea (Diptera, Tachinidae, Ornii- ini). Cell Tissue Res. 275: 63-78. Robert, D., R. N. Miles, and R. R. Hoy. 1996. Directional hearing by mechanical coupling in the parasitoid fly Ormia ochracea. J. Comp. Physio/. A 179: 29-44. Robert, D., R. N. Miles, and R. R. Hoy. 1998. Tympanal mechanics in the parasitoid fly Ormia ochracea: intertympanal coupling during mechanical vibration. J. Comp. Physiol. A 183: 443-452. Walker. T. J. 1993. Phonotaxis in female Ormia ochracea (Diptera, Tachinidae). a parasitoid of field crickets. J. Insect Behav. 6: 389-410. Reference: Biol. Bull. 200: 195-200. (April 2001) A Biologically Inspired Controller for Hexapod Walking: Simple Solutions by Exploiting Physical Properties JOSEF SCHMITZ*. JEFFREY DEAN, THOMAS KINDERMANN, MICHAEL SCHUMM. AND HOLK CRUSE Department of Biological Cybernetics, University of Bielefeld, P.O. Box 100131, D-33501 Bielefeld, FRG Abstract. The locomotor system of slowly walking in- sects is well suited for coping with highly irregular terrain and therefore might represent a paragon for an artificial six-legged walking machine. Our investigations of the stick insect Carausius morosus indicate that these animals gain their adaptivity and flexibility mainly from the extremely decentralized organization of the control system that gen- erates the leg movements. Neither the movement of a single leg nor the coordination of all six legs (i.e., the gait) appears to be centrally pre-programmed. Thus, instead of using a single, central controller with global knowledge, each leg appears to possess its own controller with only procedural knowledge for the generation of the leg's movement. This is possible because exploiting the physical properties avoids the need for complete information on the geometry of the system that would be a prerequisite for explicitly solving the problems. Hence, production of the gait is an emergent property of the whole system, in which each of the six single-leg controllers obeys a few simple and local rules in processing state-dependent information about its neighbors. * To whom correspondence should he addressed. E-mail: Josef.Schmitz@Biologie.Uni-Bielefeld.De This paper was originally presented at a workshop titled Invertebrate Sensory Information Processing: Implications for Biologically Inspired Autonomous Systems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences, Woods Hole, Massachusetts, from 15-17 April 2000. was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. Introduction The advantages of legged locomotion in contrast to wheel-driven locomotion are particularly obvious in rough terrain. A disadvantage, however, is that legs can provide only discontinuous support of the body. At a certain point within its posterior working range, each leg has to be lifted off the ground and swung forward to regain ground contact for the next step. Hence during each step cycle each leg has to go through two mutually exclusive phases: the stance, in which the leg supports the body and can exert propulsion forces; and the swing, in which the leg cannot contribute to the movement of the body. The control system that generates the leg movements of an animal faces several problems. During the stance, the trajectories of all leg endpoints have to be coordinated in such a way that each leg with ground contact contributes efficiently to propel the central body in the desired direction and with the desired velocity. Thus, in a six-legged animal with 3 joints per leg (Fig. la), as many as 18 joints have to be controlled. (For simplicity, the tarsus is considered as a passive foothold only, and the tip of the tibia is considered as the leg endpoint throughout this paper.) Four factors further complicate this control task: (1) the number and combinations of legs in stance vary continuously, (2) during curve walking, the legs move at different speeds, (3) on compliant substrates the movement of a leg may vary un- predictably. and (4) the geometry of the system may vary due to nonrigid suspension of the joints or due to injuries of leg segments. For the swing movement, it appears to be easier to solve the problems because no mechanical cou- plings via the substrate have to be considered. However, for the stability of an animal, the timing of the switch from 195 196 J. SCHMITZ ET AL target net swing net femur Figure 1. Summary of leg geometry and the single leg control network, (a) Schematic drawing of a stick insect leg showing the arrangement of the joints and their axes of rotation, (b) The leg controller consists of three parts: the swing net. which controls the leg movement during swing; the stance net, which controls the stance movement; and the selector net, which determines whether the swing or the stance net can control the motor output, i.e., the velocity of the three joints a. j3, and y. The selector net contains four units: the PEP unit, which signals posterior extreme positions; the GC unit, which signals ground contact; the RS unit, which controls the return stroke (swing movement); and the PS unit, which controls the power stroke (stance movement). The target net transforms information on the configuration of the anterior target leg, given by the angles a,, (3,, 7,. into angular values for the next caudal leg («,, £),. y,). which place the two tarsi close together. These desired final values and the current values of the leg angles, a. /3. and -y, are input to the swing net together with bias inputs and sensory inputs (rl-r4) that are activated by an obstruction blocking the swing and thereby initiate appropriate avoidance movements. ML is a nonlinear compensation term. The stance net uses the /3 joint for height control via a negative-feedback system. Walking velocity is controlled by a negative-feedback system, which compares vrel and vscns, and influences the a and y gains. The a channels are further subject to a yaw control (angle 6), which stabilizes straight and curve walking. The sign of this influence is opposite for the left and right legs. Walking is switched on by another central influence, which passes through a maximum detector and influences the a channels of all legs. CONTROLLER FOR WALKING HEXAPODS 19 stance to swing is crucial. A control system therefore also has to ensure proper spatiotemporal coordination among all legs. Each single problem mentioned above can be solved by means of classical control algorithms; this is, however, very costly in terms of computational time. In contrast, insects, like all other animals equipped with a slow-processing neuronal system, solve all these problems in real time and, beyond this, are able to react immediately even to such extreme changes as, for example, the loss of a whole leg. The solutions invented by nature are therefore not only interesting for the biologist, but could also be useful for solving problems in robotics. Control of the Gait Results from biological investigations show that each leg is controlled by its own controller (e.g., Cruse, 1990) and may indicate that even each leg controller consists of sep- arate control networks (e.g.. Bassler. 1988). The relaxation oscillator making up the step pattern generator is assumed to consist of a swing net that controls the swing movement and a stance net that controls the movement of the leg during stance (Fig. Ib). The transition between swing and stance is controlled by the selector net. The swing net and the stance net are always active, but the selector net determines which of the two has access to the motor output at any given time. The decision of when to switch from one state into the other is completely determined by sensory input. The selector net consists of a two-layer feedforward net with positive feed- back connections in the second layer. These positive-feed- back connections serve to stabilize the ongoing activity, namely stance or swing. The three most important coordi- nating mechanisms used in our present model influence the selector net so as to modulate the beginning of a swing movement, and therefore the end-point of a stance move- ment (the posterior extreme position. PEP). These influ- ences (numbers 1-3, summarized in Fig. 2) are described in detail elsewhere (e.g.. Cruse, 1990; Cruse et ai, 1998). The end of the swing movement in the animal is modulated by a single, caudally directed influence (number 4 in Fig. 2) that depends on the position of the next rostral leg. This mechanism is responsible for the targeting behavior — the placement of the tarsus at the end of a swing close to the tarsus of the adjacent rostral leg. A computer simulation of this system showed proper temporal coordination of the legs when these are walking at different speeds on a horizontal plane (Cruse et ai, 1998). The steps of ipsilateral legs are organized in triplets forming "metachronal waves," which proceed from back to front, whereas the steps of the contralateral legs on each segment step approximately in alternation. With increasing walking speed, the typical change in coordination from the tetrapod to a tripod-like gait is found. The coordination pattern is 1 Return stroke inhibits start of return stroke 2 Start of power stroke excites start of return stroke 3. Caudal positions excite start of return stroke 4 Position influences position at end of return stroke ("targeting") 5a Increased resistance increases force ("coactivation") 5b Increased load prolongs power stroke 6 Treading-on-tarsus reflex Figure 2. Summary of the mam coordinating influences between ad- jacent legs during walking. very stable. For example, when the movement of one leg is interrupted briefly during the stance, the normal coordina- tion is regained immediately after the perturbation. Note that in our approach, the temporal sequence of the activities of the legs is not explicitly produced by a master timer but is implicitly determined by the connections between the leg controllers. These coordinating mechanisms had been suc- cessfully tested on a variety of six-legged robots (e.g., Espenschied et ai. 1993; Ferrell, 1995; Pfeiffer et ai, 1995; Flannigan et ai, 1998; Frik et ai. 1999). Control of the Swing Movement The geometry of the stick insect's leg is shown in Figure la. The coxa-trochanter (/3) and femur-tibia (7) joints, the two distal joints, are simple hinge joints with one degree of freedom corresponding to elevation and to extension of the tarsus, respectively. The subcoxal (a) joint is more com- plex, but during forward walking most of its movement is in a rostrocaudal direction around the nearly vertical axis. The leg can thus be considered as a manipulator with three degrees of freedom for movement in three dimensions, and the control network must have at least three output channels, one for each leg joint. As has been shown by Cruse et ai (1998), a simple, two-layer feedforward net with three out- put units and six input units can produce movements that closely resemble the swing movements observed in walking stick insects. In the simulation, the three outputs of this net, interpreted as the angular velocities da/dt. d/3/dr. and d7/dr, are fed into an integrator (not shown in Fig. Ib). which in the animal corresponds to the leg itself, to obtain the joint angles. The actual angles are measured and fed back into the net. Via this loop through the environment, the network becomes a recurrent system. In addition to the actual angles of the three leg joints, three other input units (a,, j3,, 7,) represent the target of the swing movement, i.e., the leg position that should be achieved at the end of the return stroke (targeting influence, number 4 in Fig. 2). 198 J. SCHMITZ ET AL. This system with only eight or nine nonzero weights is able to generalize over a considerable range of untrained situations. Furthermore, the swing net is remarkably tolerant with respect to external disturbances (Fig. 3). The learned trajectories represent a kind of attractor to which the dis- turbed trajectory returns. This compensation for distur- bances is possible because the system does not compute explicit trajectories but — using actual sensor data — calcu- lates only the next movement increment. This ability to compensate for external disturbances permits simple exten- sions of the swing net in order to simulate avoidance be- haviors observed in insects (additional inputs "rl-r4" in Fig. Ib). Control of the Stance Movement To control the stance movement, it is not enough simply to specify a movement for each leg on its own: the mechan- ical coupling through the substrate means that efficient locomotion requires coordinated movement of all the joints of all the legs in contact with the substrate — that is, a total of 18 joints when all legs of an insect are on the ground. However, the number and combination of mechanically coupled joints varies from one moment to the next, depend- ing on which legs are lifted. A further complication occurs when the animal negotiates a curve, because then the dif- ferent legs move at different speeds. 10 , 5 E E TO n c « 1 8 N -5 -10 -15 -10 -5 10 15 20 25 x-coordinate [mm] Figure 3. Behavior of the swing net. The tine stippled line represents a trajectory (in the A-C plane, i.e., in a side view) of the tibia endpoint of a middle leg of an undisturbed swing movement. The leg lifts off at the posterior extreme position (PEP I and swings forward to the anterior ex- treme position (AEP). The dashed line represents a swing trajectory of the middle leg in which the movement of the tibia was obstructed (arrow). The swing net instantaneously generates a new trajectory, which avoids the obstruction by retracting and lifting the tibia and which then reaches the AEP on this new trajectory. The filled circles (with .v-c error bars) con- nected by the continuous line show the reaction of a stick insect whose swing movement was obstructed at the same moment in swing as it was applied to the simulation. In robots, these problems can be solved using traditional, though computationally costly, methods, because these ap- proaches consider the ground-reaction forces of all legs in stance and seek to optimize some additional criteria, such as minimizing the tension or compression exerted by the legs on the substrate. Due to the nature of the mechanical inter- actions, and inherent in the search for a globally optimal control strategy, such algorithms require a single, central controller; they do not lend themselves to distributed pro- cessing. This makes real-time control difficult, even in the simple case of walking on a rigid substrate. Despite the evident complexity of the task, it is mastered even by insects. Therefore, there must be a solution fast enough that on-line computation is possible even for slow- processing neuronal systems. How can this be done? To solve the particular problem at hand, we propose to replace a central controller by distributed control in the form of local positive feedback (Cruse et ai, 1995a). Positive feed- back (or in other words, reflex reversals) had already been reported to occur during walking in the stick insect (e.g., Bassler. 1988; Schmitz et ai, 1995). Compared to earlier versions of the leg controller (Cruse et al. 1995b). this change permits the stance net to be radically simplified. The positive feedback occurs at the level of single joints: the position signal of each is fed back to control the motor output of the same joint (Fig. Ib, stance net). How does this system work? Let us assume that any one of the joints is moved actively. Then, because of the mechanical connec- tions, all other joints begin to move passively, but in exactly the proper way. Thus, the movement direction and speed of each joint does not have to be computed, because this information is already provided by the physics. The positive feedback then transforms this passive movement into an active movement. There are, however, several problems to be solved that are inherent in positive feedback systems. The first is that positive feedback using the raw position signal would lead to unpredictable changes in movement speed. This problem can be solved by introducing a kind of bandpass filter into the feedback loop. The effect is to make the feedback proportional to the angular velocity of joint movement, not the angular position. In the stick insect, the bandpass filter property can be attributed to the phasic response character- istic of the sense organs and the lowpass filter properties of the neuromuscular system. The second problem is that using positive feedback for all three leg joints leads to unpredict- able changes in body height, even in a computer simulation neglecting gravity. A physical system would, of course, be pulled downward by gravity, and the positive feedback would accelerate this movement. In the stick insect, body height is controlled by a distributed system in which each leg acts like an independent, proportional controller (Cruse et nl., 1993). Thus, no master height controller is necessary; the only central information is the invariant reference value CONTROLLER FOR WALKING HEXAPODS 19! a) straight and curve walking b) walking over obstacle c) stumbling Figure 4. Strobe-shot video of a simulation of the six-legged system with negative feedback applied to all six /3 joints and positive feedback to all a and y joints. Leg positions are shown only during stance and only for each fifth time interval. In (a), straight (angle Oref = 0) and curve walking (angle 6rt., ^ 0) is shown from both a top and a side view. Walking direction is from left to right. In (b), a straight walk over an obstacle (obstacle height is 90^ of the normal body-substrate clearance) is shown from a side view. In (c), three single frames (both top and side view each) of a situation in which the system was forced to fall to the ground are shown. The fall caused an extremely disordered arrangement of the legs, but within a few seconds the system stood up by itself and resumed proper coordinated walking. for each leg. In the WalkNet we implemented this such that only the a and the y joints are under positive feedback control, whereas the /3 joint remains under classical negative feedback as in the standing animal. In this way, it is possible to solve the problems mentioned above in an easy and computationally simple manner. This hypothesis is also compatible with biological results obtained from animals subject to disturbances during stance movements (Battling and Schmitz. 2000). Finally, we have to address the question of how walking speed is determined in such a positive feedback controller. We assume a central value that represents the desired walk- ing speed vref. This is compared with the actual speed, which could be derived from the optical flow or by moni- 200 J. SCHMITZ ET AL. toring leg movement. An error signal, resulting from a deviation, modifies the gain of the positive feedback for all a and 7 joints of all six legs accordingly (Fig. Ib, stance net). With our controller, curve walking is achieved in an easy way. Again we assume a central value that represents the intended angle of body yaw. A traditional negative feedback system controls the rate of yaw (0yaw, Fig. Ib). The error signal modifies the gain of the positive feedback of the legs. Keeping fjri.t at zero results in straight walks: a bias to one or the other side results in curve walking with the curvature proportional to the amplitude of the bias (Fig. 4a). Walking over irregularities in the terrain is performed sufficiently well due to the adaptive properties inherent in our control approach (Fig. 4b). Moreover, we found further emergent properties of the decentralized control structure. If the system was made to stumble and fall to the ground, it stood up by itself and resumed proper walking (Fig. 4c). This happened even when the fall placed the six legs in an extremely disordered arrangement. This means that the sim- ple solution proposed here also eliminates the need for a special supervisory system to rearrange leg positions after such an emergency. Summary Considering the problems with which an adaptive control system for walking must cope, one could be inclined to propose a highly centralized system with global knowledge. Our biologically inspired model shows, however, that the most complex task (stance movement) can be handled by a rather simple system (local positive feedback). The simpli- fication is possible because instead of explicit calculations, the physical properties of the system and its environment are exploited. This is shown in our model at several stages: • The rhythmic movements of single legs are not pro- duced by a top-down approach but emerge from the interaction of the neuromuscular system with the en- vironment. • For the swing movement, no explicit trajectory is pre- calculated and then spooled out: rather the trajectory is calculated on the basis of the actual sensor data of the ongoing swing. • The discrepancy between the complexity of the task and the simplicity of the solution is most obvious for the control of the stance movement. Local rules for each single joint are sufficient to solve the given task in an adaptive way. Furthermore, our simulation shows that at the systems level, complex behavior can emerge from the cooperation of local rules: • The coordination mechanisms produce proper tripod or tetrapod gait. These gaits are robust against distur- bances. • The system, by combining local positive and negative feedback loops, can cope with a variety of disturbances such as restricting movement of a single leg or removal of leg segments. • If the system was forced to stumble and fall to the ground, it stood up by itself and resumed proper walk- ing. These findings encourage us to propose a decentralized control scheme as a basis for legged robot locomotion. This basis can be extended to integrate further reflexes and behaviors, in order to achieve in the robot a level of agility similar to that found in insects. Acknowledgments Supported by Deutsche Forschungsgemeinschaft (Cr 58/ 9-2) and the Korber-Foundation. Literature Cited Bartling, C., and J. Schmitz. 2000. Reaction to disturbances of a walking leg during stance. J. E.xp. Biol. 203: 121 1-1233. Bassler, U. 1988. Functional principles of pattern generation for walking movements of stick insect forelegs: the role of the femoral chordotonal organ afferences. J. E\p. Biol. 136: 125-147. Cruse, H. 1990. What mechanisms coordinate leg movement in walking arthropods? Trends Neurosci. 13: 15-21. Cruse, H., J. Schmitz. U. Braun, and A. Schweins. 1993. Control of body height in a stick insect walking on a treadwheel. J. Exp. Biol. 181: 141-155. Cruse, H., C. Bartling, and T. Kindermann. 1995a. High-pass filtered positive feedback for decentralized control of cooperation. Pp. 668- 678 in Advances in Artificial Life. F. Moran. A. Moreno. J. J. Merelo. and P. Chacon, eds. Springer. New York. Cruse, H., C. Bartling. D. E. Brunn, J. Dean, M. Dreifert, T. Kinder- mann, and J. Schmitz. 1995b. Walking: a complex behavior con- trolled by simple systems. Adapt. Behav. 3: 385 — U8. Cruse, H., T. Kindermann. M. Schumm, J. Dean, and J. Schmitz. 1998. Walknet — a biologically inspired network to control six-legged walk- ing. Neural Networks 11: 1435-1447. Espenschied, K. S., R. D. Quinn, H. J. Chiel, and R. D. Beer. 1993. Leg coordination mechanisms in the stick insect applied to hexapod robot locomotion. Adapt. Behav. 1: 455-468. Ferrell, C. 1995. A comparison of three insect-inspired locomotion controllers. Robot. Anton. Syst. 16: 135-159. Flannigan, W. C., G. M. Nelson, and R. D. Quinn. 1998. Locomotion controller for a crab-like robot. Pp. 152-162 in IEEE Proc. Robotics and Automation 1998. Leuven. Belgium. Frik, M., M. Guddat, M. Karatas, and C. D. Losch. 1999. A novel approach to autonomous control of walking machines. Pp. 333-342 in Proc. 2nd Conference on Climbing and Walking Robots. CLAWAR 1999. Professional Engineering Publishing, Bury St. Edmunds. UK. Pfeiffer, F., J. Eltze, and H. J. Weidemann. 1995. Six-legged technical walking considering biological principles. Robot. Auton. Syst. 14: 223-232. Schmitz, J., C. Bartling, D. E. Brunn, H. Cruse, J. Dean, T. Kinder- mann, M. Schumm, and H. Wagner. 1995. Adaptive properties of "hard-wired" neuronal systems. Verh. Dtsch. Zoo/. Ces. 88.2: 165-179. Reference: Biol. Bull. 200: 201-205. (April 2001) Insights for Robotic Design From Studies of the Control of Abdominal Position in Crayfish DAVID L. MACMILLAN* AND BLAIR W. PATULLO Department of Zoology, University of Melbourne, Park\'ille, Victoria 3052, Australia Abstract. Studies of the control of position and move- ment of the abdomen of crayfish illustrate a number of features of invertebrate sensory-motor systems that have implications for their use to inform robotic design. We use the abdominal slow extensor motor system to illustrate three of them here: first, the way in which a behaviorally flexible length-servo device can be achieved with very few ele- ments; second, the importance of knowledge of the biolog- ical and behavioral context in which the elements operate; third, that design solutions resulting from natural selection have been constrained by the previous evolutionary history of the animal, which can affect the outcomes in ways that may not be immediately apparent in a design context. Introduction Crayfish, and other crustaceans with tubular bodies, have proprioceptive structures called muscle receptor organs (MROs) spanning the articulations between the segments (Alexandrowicz. 1951 ). Each MRO signals the relative po- sition and movement of the two segments to which it attaches. Since the discovery of MROs. many different aspects of their biology have attracted the interest of neu- robiologists (Fields, 1976; Macmillan, 2001). In this paper we combine data from the literature with insights from our current work on crayfish to highlight three aspects of MRO *Author to whom correspondence should be addressed. E-mail: dlmacm@unimelb.edu.au. This paper was originally presented at a workshop titled Invertebrate Season Information Processing: Implications for Biologically Inspired Autonomous Systems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences. Woods Hole. Massachusetts, from 15-17 April 2000. was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. Abbreviations: ACC. accessory motor neuron; MRO. muscle receptor organ; SEMN, slow extensor motor neuron; SR, stretch receptor. biology with interesting implications for biologically in- spired robots: that arthropods achieve complex behavioral outcomes with few neurons, that assumptions derived from reduced preparations or parts of animals need to be tested in biologically meaningful situations in intact animals, and that it is important to have an appreciation of the impact of natural selection when examining biological solutions to engineering design problems. The abdomen, or tail, of the crayfish is composed of five similar, articulated tubes connected by simple lateral hinges that permit movement only in the pitch plane (Fig. 1 A). The connection between the thorax and the first abdominal seg- ment is different. It is surrounded by strong, flexible mem- branes enclosing stout muscles anchored deep within the thorax. In addition to movement above and below the hor- izontal line of the body, the joint allows limited rotation in the roll and yaw planes so that it acts as a limited-range universal joint. The functional outcome is flexibility of position and movement at the other end of the tail where the biramous appendages (uropods) of the sixth abdominal seg- ment, together with the telson. form a tailfan (Fig. 1 A). The tailfan elements elevate and depress and also slide over each other laterally so that the whole tailfan can vary in size and can change shape from paddle to scoop. This flexibility at the end of the hinged lever that is the tail permits the animal to modify the direction and magnitude of the forces gener- ated when it flexes and extends its tail for swimming and balancing. The arrangement presents an interesting case study for biological solutions to problems associated with the dynamics and control of multi-jointed levers. The muscles controlling the movement of the tail are divided into four matching sets: the slow and fast flexors and extensors. The fast flexors ventrally and fast extensors dorsally occupy most of the abdominal cavity (Fig. IB) and are responsible for escape tail flip behavior and swimming. Although related developmentally to their segment of ori- 201 202 D. L. MACMILLAN AND B. W. PATULLO SEGMENTAL ARTICULATION Figure 1. (A) Lateral view of the crayfish abdomen showing the changing position and shape of the segments and tailt'an during a movement from a mid-flexed position to full extension. The joints between the abdominal segments permit movement only in the pitch plane, whereas that between the thorax and first abdominal segment permits movement in pitch, roll, and yaw planes. ( B ) Diagram of a section of the abdomen showing placement of the main muscle blocks in cross-section and the ventral nerve cord and its connection with the MRO in one segment. (C) Summary of the nerve-muscle connections operating when (i) the SEMN4 program is operating and the SR load-compensation loop can be recruited, and (iil the SEMN3 program is operating and the SR load-compensation loop is not recruited. Triangles and circles represent excitatory and inhibitory synapses respectively. (D) Pattern of activity of the SEMNs and SRs recorded from the dorsal nerves during a platform drop extension (i) in abdominal segment 4 when the movement is unhindered, and (ii) in abdominal segment 2 when a rod and mechanical stop prevent straightening of the joint spanned by this MRO throughout the extension taking place at the other abdominal joints. Note that these recordings are displayed on different time and amplitude scales. gin, they are organized into the overlapping muscle spirals familiar to lovers of lobster as a gourmet delicacy. Common connective tissue strands that run the length of the tail ensure that they act as a coordinated unit when delivering their power. The slow flexor and extensor muscles, which are responsible for slow movements and postural adjust- ments, lie ventrally and dorsally respectively, in thin sheets external to the bulk of the fast muscles and immediately beneath the surface of the exoskeleton (Fig. IB). The Remarkable Parsimony of Arthropod Neuromuscular Control Vertebrates achieve fine control in muscles primarily by having large numbers of motor units that can be recruited sequentially and, to a lesser extent, by heterogeneity of their nerve and muscle. The pools of motor neurons employed by vertebrates to achieve fine control of their muscles need space that the small-bodied arthropods do not have. Arthro- pods achieve comparable levels of control with far fewer neurons (insects typically have only two or three motor neurons per muscle and crustaceans have up to five or six). They achieved this reduction in part by employing nerve- muscle combinations ranging in their contraction character- istics from slow and gradual to fast and twitching, and partly by evolving motor neurons that inhibit muscle contraction in parallel with the more familiar ones that cause contrac- tion. These developments massively increase the possibili- ties for a continuously graded range of contraction outputs CRAYFISH POSITIONAL CONTROL 20 and for the integration of excitatory and inhibitory inputs throughout the control mechanism without a concomitant increase in the number of motor neurons. These ingenious and effective ways of reducing the number of neurons could well be of interest in robotic design. The fact that variations of these arrangements are found throughout the arthropods suggests that they evolved early. Comparative studies show that they have not necessarily been advantageous in all subsequent behavioral situations and provide interesting insights into ways in which design flexibility is achieved from a base model constrained by its evolutionary history to very few motor neurons (Paul, 1991). Each of the segmental muscle groups in the crayfish is innervated by just six motor neurons per segment; these are conventionally numbered from 6 to 1 in descending order of axonal diameter (Fields and Kennedy. 1965; Fields. 1966; Drummond and Macmillan. 1998a,b). Each displays partic- ular functional characteristics exemplified by the slow ex- tensor sensory-motor system about which most is known and on which we will concentrate here. Slow extensor motor neuron 6 (SEMN6) is the largest motor neuron in the group. It innervates predominantly the fastest fibers in the slow extensor muscle. At low firing frequencies it produces large membrane responses that facilitate rapidly and sum when the frequency of stimulation is high to produce strong, twitch-type contractions. SEMN5 is slightly smaller in di- ameter than SEMN6. Its effect on the muscle potential is inhibitory — that is. it produces ionic permeability changes that hyperpolarize the muscle membrane and move its ex- citability away from the threshold for contraction, thereby modifying the nature of the contraction produced by exciter activity. SEMN4 and SEMN3 are almost indistinguishable in size and innervate almost the same population of slow extensor muscle fibers throughout all the slow extensor muscle bundles. They both evoke small to medium-sized responses that show modest facilitation and sum readily to produce smooth contractions at higher frequencies. SEMN2 is a small neuron that innervates almost all the slow exten- sor muscle fibers. Its activation produces small, non-facili- tating muscle responses that sum to produce strong, slow muscle contractions at high frequencies. SEMN1 is a neuron of about the same size, but its muscle potentials and con- tractions are small, so its physiology is not well docu- mented. From a design perspective, these elements provide for a full range of control options in the slow-fast and fine-coarse spectra while preserving remarkable component economy. This brief outline of the innervation of this major muscle series shows that although the slow extensors in each seg- ment are controlled by bilateral pairs of only six motor neurons, the flexibility inherent in their properties and pat- terns of innervation together with the trick of peripheral inhibition provide the same sort of flexibility available to multi-neuronal vertebrate equivalents. SEMN6 can be used for twitch or brief, strong contractions; SEMN4 and SEMN3 for rapidly developing and decaying smooth con- tractions; and SEMN2 for slowly developing, sustained smooth contractions. Activation of SEMN5 reduces the contractions evoked by any contemporaneous exciter activ- ity and also returns the membrane potential of slower fibers to their resting level more rapidly following slow exciter activity, thereby increasing the rate of muscle relaxation. What do we know about how the potential flexibility of this parsimonious arrangement is employed by the crayfish? To answer this question we have to make a brief side trip into the sensory biology of the MROs. The Sensory Biology of the Muscle Receptor Organs A pair of MROs span the articulations between adjacent abdominal segments on each side of the body (Fig. IB). Each pair of MROs has two stretch receptor (SR) neurons, a large, high-threshold phasic neuron and a smaller, low- threshold, tonic one. These respond to changes in the ten- sion of a small, innervated receptor muscle strand in which they are embedded. Because the muscle spans the articula- tion, the SRs monitor the position of the abdominal seg- ments relative to one another, and because the strand is innervated, their level of activity, or set point, can be adjusted from the central nervous system. SR activity levels can also be adjusted by the axons of inhibitory cells called accessory motor neurons (ACC) that synapse directly onto the SR surface and inhibit or delay sensory spike activation (Fig. ICi). The receptor muscles lie in parallel with the slow extensor muscle but are not powerful enough to assist with joint movement. These elements and their arrangement sug- gest those an engineer might employ in designing a servo- controlled load-compensating device (Fig. ICi). Analysis reveals that this sophisticated and economical arrangement of elements has the potential to deliver flexible abdominal control. Experiments on semi-intact and isolated abdominal preparations (mainly in Procambarus clarkii and Che rax destructor) reveal a number of connections and reflexes involving the tonic SRs and the motor neurons in the local and adjacent segments (Fields. 1976; Drummond and Macmillan, 1998a; summarized in Fig. 1C). Activation of the tonic SR excites ACC, which inhibits the sensory neuron itself (Eckert. 1961) — a negative feedback loop that radiates weakly to adjacent segments. Functionally, this loop provides a classic myotatic reflex: it damps small displacements of the joint by externally imposed force, and it may also damp SR reflex loops. The tonic SR also excites the ipsilateral SEMN2 in its own segment, the motor neuron that produces slow sustained contractions in most of the slow extensor muscle fibers. This is the basis of a "resis- tance reflex." If the joint is flexed so that the tonic SR is activated, it will excite SEMN2, which will fire and extend 204 D. L. MACMILLAN AND B. W. PATULLO the joint until the receptor is unloaded and stops firing (Fields, 1966; Fields et at.. 1967). Another important feature of the neuromuscular wiring is that SEMN4 innervates both the slow extensor muscle and the receptor muscle so that both could contract synchro- nously when it is activated. This relationship spawned a hypothesis about the way in which feedback could operate on the tonic sensor to produce different behavioral outputs. If central drive for extension included SEMN4, the tonic SR would not be activated during the extension because the MRO receptor muscle would contract with the main exten- sor muscle. If the movement met with resistance and the rate of extension slowed, the receptor muscle would continue to contract at the rate determined by the central drive onto SEMN4. Rising tension in the receptor muscle would acti- vate the SR to recruit SEMN2 or increase its firing rate until the receptor was again unloaded. The circuitry is also con- figured to permit the same movement without activation of SR-mediated load compensation (Fig. ICii). If the central drive for extension were directed through SEMN3 rather than SEMN4. the same muscle fibers would be activated but the receptor muscle would not contract during the move- ment, and thus the length-servo loop would not be activated. This is how this sensory-motor cluster appears to operate in isolated C. destructor preparations. The SEMN3 and SEMN4 are not active together during spontaneous activity (Dnimmond and Macmillan, 1998a). and SEMN6 only fires at the peak of bursts of spontaneous activity in SEMN4 or SEMN3. SEMN2, on the other hand, can be recruited at any stage during ongoing activity in either unit. The pattern of firing in SEMN5, the inhibitor, suggests that it is used both to inhibit extensor contractions — for example, by firing during flexor muscle activity — and also to move the con- tractions due to repetitive activation of the intermediate fast-slow units SEMN3 and SEMN4 towards the more phasic end of their spectrum by suppressing their tendency to sum responses. In summary, there is ample potential for remarkable functional flexibility, notwithstanding the small number of neurons involved. The Operation of the Receptors in Freely Behaving Animals The finding that circuitry is capable of producing outputs predicted by a hypothesis and that it does so in non-behav- ioral, dissected preparations is a valuable step towards un- derstanding how it might function, but from a robotics viewpoint, observation of its function and role in the bio- logical and behavioral world of the animal is essential. We therefore set out to examine how the MROs and SRs operate in intact, freely behaving animals. Using surface markings to locate the position of the MROs, we implanted fine wire electrodes to make long-term recordings from the SRs and the SEMNs in C. destructor (McCarthy and Macmillan, 1999a,b). When an immersed crayfish loses contact with the substrate, it extends its abdomen so that it falls in a bal- anced, feet-down attitude, ready to walk as soon as it lands. This type of extension movement can be evoked by drop- ping a hinged platform on which the animal is standing (Larimer and Eggleston. 1971). The behavior was therefore dubbed "a platform-drop extension." The outcome of our platform-drop experiments provided a sharp reminder of the importance of evaluating hypotheses derived from reduc- tionist analysis by testing them in the real world of the animal. This caution is particularly germane if the interpre- tations are to inform or lead robotic design. In C. destructor, the tonic SR was invariably active when the animal was in the resting position with its abdomen curled beneath the body. The neuron fired very regularly, with a frequency around 16 Hz (Fig. ID), the actual rate being slightly higher or lower depending on the whether the abdomen was more or less flexed. Activity was seen in the SEMNs as soon as the platform was dropped to initiate the extension, but by the time extension movement was appar- ent to an observer, the SR had ceased firing (Fig. IDi). This result surprised us because previous results from other spe- cies suggested that the SR would fire throughout the exten- sion (Sokolove, 1973). It could, however, be explained in system terms if the SR falls silent because the receptor muscle and the working extensor muscle are contracting at about the same rate but with the straightening of the joint just slightly leading the contraction of the receptor muscle. To test this possibility, we recorded from the SR in a segment that was prevented from straightening at all be- cause of a small rod inserted between two acrylic plastic blocks glued either side of the joint. In this situation, the expected load compensation occurred, as evidenced by the increase in output by the motor neurons compared with their firing rates in the unblocked situation (Fig. IDii). It was not, however, due to recruitment of the local MRO. because the rate of firing in the SR remained constant throughout the extension movement (Fig. IDii). Some other receptors must have been responsible. This outcome was also unexpected because of earlier indications from other species that the MROs might be involved in load compensation (Sokolove, 1973). We hypothesized, however, that the reason the SR load-compensation pathway is not selected for platform- drop extensions is that, in this situation, the animal uses the SEMN3 mode rather than the SEMN4 one. Extension in this context is part of a stability or balancing behavior that gives the body a parachute-like profile. In the unlikely event that the animal encounters resistance during its fall through the water, it can then change posture to prepare for defense or other behaviors, some of which may involve the SR- SEMN4 load-compensation loop. To test this hypothesis, we used our ability to record from freely moving animals to study the activity of the SRs and SEMNs during a range of other behaviors, including some — such as the extension that CRAYFISH POSITIONAL CONTROL 20:. accompanies the assumption of the defense posture — in which load compensation, possibly mediated by the SR loop, could be expected to occur. As yet. we have found no behavior in C. destructor in which the SR-SEMN4 loop is activated when SEMN output increases in response to in- creased load (Patullo et al, 2001 ). Behavioral Context — the Key to Design Insights The message is clear. This elegant circuitry works in a dish and we can see ways in which it could be employed to control movement, but we do not yet know enough about the biology to understand completely how the animal uses the components. Such understanding is necessary if we are to exploit the full potential of this sensory-motor system to inform our design of fine movement control devices with component economies suitable for robotic applications. It may be that the capabilities of the circuitry are not primarily targeted at the control of slow, postural movements. We have evidence, for example, that the SRs are cyclically active during non-giant swimming and that interference with their function alters such movements (Daws. Mc- Carthy and Macmillan. unpubl. data). It is also possible that the normal function of the local feedback loops is evident only when they are operating as part of a concatenated chain involving the whole of the jointed abdominal lever. In support of this possibility is evidence that the cord stretch receptors have different effects on motor output when they are activated in groups rather than singly (Drummond and Macmillan, 1997). The study of abdominal movement in these animals has already provided insights into the way in which fine control can be achieved with very few elements and the necessity of fully understanding the context in which control mechanisms are to function. It promises many more as we achieve a better understanding of the biology. Acknowledgments We gratefully acknowledge the support of NASA for underwriting the stimulating cross-disciplinary ISIP-BIAS meeting that generated this paper and Dr. Diana Blazis and Dr. Frank Grasso for their assistance and support in its conception and publication. Supported by a grant from the Australian Research Council to DLM and from NASA under Cooperative Agreement NCC 2-896. Literature Cited Alexandrowicz, J. S. 1951. Muscle receptor organs in the abdomen of Homarus vulgaris and Pantilirus vulgaris. Q. J. Microsc. Sci. 92: 163-199. Drummond, J. M., and D. L. Macmillan. 1997. Cord stretch receptors of the Australian freshwater crayfish. Cln-ra.\ destructor. Soc. Neurosci. Abstr. 23: 1569. Drummond, J. M., and D. L. Macmillan. 1998a. The abdominal motor system of the crayfish, Cherax destructor. I. Morphology and physiol- ogy of the superficial extensor motor neurons. J. Comp. Physiol. A 183: 583-601. Drummond, J. M., and D. L. Macmillan. 1998b. The abdominal motor system of the crayfish. Cherax destructor. II. Morphology and physi- ology of the deep extensor motor neurons. J. Comp. Physiol. A 183: 603-619. Eckert, 1961. Reflex relationships of the abdominal stretch receptors of the crayfish. I. Feedback inhibition of the receptors. / Cell. Comp. Physiol. 57: 149-162. Fields, H. L. 1966. Propnoceptive control of posture in the crayfish abdomen. J. Exp. Biol. 44: 455-468. Fields, H. L. 1976. Crustacean abdominal and thoracic muscle receptor organs. Pp. 65-1 14 in Structure and Function of Proprioceptors in the Invertebrates. P. J. Mill. ed. Chapman and Hall. London. Fields. H. I... and D. Kennedy. 1965. Functional role of the muscle receptor organs in crayfish. Nature 206: 1235-1237. Fields, H. L., W. H. Evoy, and D. Kennedy. 1967. Reflex role played by efferent control of an invertebrate receptor. J. Neurophysiol. 30: 859-874. Larimer, J. L., and A. C. Eggleston. 1971. Motor programs for abdom- inal positioning in crayfish. Z. Vgl. Physio/. 74: 388-402. Macmillan. D. L. 2001. The abdominal muscle receptor organ of cray- fish and lobsters: old models new challenges. In Frontiers in Crusta- cean Neurobiology. Springer, Berlin. (In press). McCarthy, B. J., and D. L. Macmillan. 1999a. Control of abdominal extension in the freely moving intact crayfish Cherax destructor. I. Activity of the tonic stretch receptor. J. Exp. Biol. 202: 171-181. McCarthy, B. J., and D. L. Macmillan. 1999b. Control of abdominal extension in the freely moving intact crayfish Cherax destructor. II. Activity of the superficial extensor motor neurones. / Exp. Biol. 202: 183-191. Patullo, B., Z. Faulkes, and D. L. Macmillan. 2001. The muscle receptor organs do not mediate load compensation during body roll and defence. J. Exp. Zoo/, (in press). Paul, D. H. 1991. Pedigrees of neurobehavioral circuits: tracing the evolution of novel behaviors by comparing motor patterns, muscles, and neurons in members of related taxa. Brain Behav. Evol. 38: 226-239. Sokolove, P. G. 1973. Crayfish stretch receptor and motor unit behaviour during abdominal extensions. J. Comp. Physiol. 84: 251-266. Reference: Bio!. Bull. 200: 206-210. (April 2001) Increasing Sensor Flexibility Through Neuromodulation J. T. BIRMINGHAM* Volen Center and Biology Department, Brandeis University. Wcilllmm. Massachusetts 02454-91 10 Abstract. Both biological and man-made motor control networks require input from sensors to allow for modifica- tion of the motor program. Real sensory neurons are more flexible than typical robotic sensors because they are dy- namic rather than static. The membrane properties of neu- rons and hence their excitability can be modified by the presence of neuromodulatory substances. In the case of a sensory neuron, this can change, in a functionally significant way. the code used to describe a stimulus. For instance, extension of the neuron's dynamic range or modification of its filtering characteristics can result. This flexibility has an apparent cost. The code used may be situation-dependent and hence difficult to interpret. To address this issue and to understand how neuromodulation is used effectively in a motor control network, I am studying the GPR2 stretch receptor in the crustacean stomatogastric nervous system. Several different neuromodulatory substances can modify its encoding properties. Comparisons of physiological and anatomical evidence suggest that neuromodulation can be effected both by GPR2 itself and by other neurons in the network. These results suggest that the analog of neuro- modulation might be useful for improving sensor perfor- mance in an artificial motor control system. Introduction Many forms of locomotion in both invertebrates and vertebrates result from the activity of central pattern gener- * Present address: Department of Physics. Santa Clara University, 500 El Camino Real, Santa Clara, CA 95053. This paper was originally presented at a workshop titled Iiivertehrute Sensory Information Pmci.-s-.inK: Implications for Biologically Inspired Autonomous S\stems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences, Woods Hole, Massachusetts, from 15-17 April 2000. was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. ator (CPG) networks of neurons (see review by Marder and Calabrese [1996]). CPGs produce rhythmic motor patterns by virtue of the intrinsic properties of the neurons and the connections between them, even when disconnected from sensory afferents (Wilson, 1961). Sensory feedback, how- ever, is usually essential for modifying the details of the basic rhythm to shape a physiologically relevant output in response to particular environmental or body conditions (Grillner and Wallen, 1977; Foth and Graham, 1983a,b; Pearson et «/.. 1983). Man-made CPG networks are used to generate movement in biologically inspired robots, as reflected in the papers in this collection (CASSLS, 2001 ). An artificial CPG has both the power and limitations of a biological one: it has a built-in robustness, but if there is to be flexibility in its output, it too will require input from sensors. The sensors employed in many robots differ from biological sensory neurons in a fundamental way. A typical man-made sensor is static. It always gives the same unambiguous response to a particular stimulus. Real sensory neurons are dynamic. The spike trains they generate depend not solely on the stimulus but also can be influenced by other factors. One of these factors is the presence of neuromodulators, substances that modify a neuron's membrane properties and hence its excitability (Kaczmarek and Levitan, 1987). Neuromodula- tion affects neurons in the sensory, central, and motor systems. In the case of a sensory neuron, neuromodulation can modify the spike train generated in response to a par- ticular sensory stimulus. This can have functional signifi- cance. For example, when primary olfactory receptors in the newt are exposed to micromolar concentrations of adrena- line, the receptor's threshold for and differential sensitivity to odor perception are affected (Kawai et al.. 1999). This modulation may very well be crucial for an animal trying to find a mate or avoid becoming a meal. The fact that sensory neurons can be modulated prompts an interesting line of questioning. Two distinct stimuli pre- 206 MODULATION OF A CRAB STRETCH RECEPTOR 20 sented in two different neuromodulatory environments could quite conceivably result in very similar spike trains. Does that imply that the meaning of the spike train is ill-defined? How might what appears to be ambiguity in the code be removed? In particular, how can sensor modulation be used to advantage in a CPG network? The crab stomatogastric nervous system (STNS) is an ideal preparation in which to address some of these questions. The STNS controls the movements of stomach muscles used to grind and filter food and is today probably the most carefully studied and best understood small CPG network. Gastropyloric receptor 2 (GPR2) is a stretch-sensitive neuron that monitors the movement of two stomach muscles (Katz et al, 1989) and provides this information to the CPG. There are several rea- sons why GPR2 is an attractive candidate for studying the computational significance of neuromodulation. ( 1 ) The stim- ulus (muscle tension) is well defined and one-dimensional (Katz et al. 1989). (2) The neuron's response to muscle stretch can be modulated in several different ways, as will be de- scribed below. (3) The synaptic targets of GPR2 are known and accessible (Katz and Harris-Warrick. 1989, 1990). and signal processing of the sensory information can be investi- gated directly. The results discussed below imply that neuro- modulation of GPR2 may allow the neuron to interact both with the stimulus environment and the central nervous system so that it can remain sensitive to important stimuli. Our inves- tigations suggest that incorporation of the analog of neuro- modulation into a robotic nervous system might similarly be useful for introducing dynamics to sensors to increase their functionality. All experiments were done using male Cancer borealis crabs purchased from local seafood suppliers in Boston, Mas- sachusetts. From the gut of the crab, a preparation consisting of two muscles and the nerve containing the GPR2 cell body was removed, placed flat in 5-ml silicone elastomer-coated petri dishes and continuously superfused with cooled physiological saline. One side of the cpv3a muscle (nomenclature from Maynard and Dando [ 1974]) was pinned to the dish. The other side was attached to a force-displacement transducer. This transducer in turn was attached to the arm of a computer- controlled pen motor. The muscle was stretched using various waveforms (steps, sine waves, white noise). The resulting spikes generated by the sensory neuron were measured using an extracellular electrode on the nerve. Neuromodulatory sub- stances were introduced to the preparation via switching ports in the supervision system. A detailed description of the exper- imental setup and technique has previously been published (Birmingham et al.. 1999). Modulation of a Stretch Receptor I have discovered that the response of GPR2 to stretch can be modulated in two very different ways. In one case, the neuron's sensitivity can be continuously tuned through the application of several neuroactive substances. In the Muscle stretch Control 2.5 sec 0.15 mm Modulated response (by AST-3) Figure 1. Stretch receptor response can be modulated by peptide application. Extracellular recordings of the activity of crab GPR2 neuron in control saline (middle) and in saline containing 10~x M AST-3 (bottom) in response to stretch of the cpv3a muscle (top). other case, an unknown internal mechanism drives the neu- ron between two qualitatively distinct modes of firing (Katz et al.. 1989; Birmingham et al.. 1999). The sensitivity of the GPR2 response to muscle stretch can be modified by at least four distinct substances: serotonin, y-aminobutyric acid (GAB A), and the peptides TNRNFLRF- amide and allatostatin-3 (AST-3). I will focus on the effects of AST-3 in this paper. Figure 1 shows that the number of GPR2 spikes generated in response to muscle stretch is reduced in the presence of 1CT8 M AST-3. The decrease in response is a continuous function of modulator concentration. The threshold for the AST-3 effect on firing rate is l(Tq - 1CTS M. and saturation occurs above 10"7 M. In most cases, the GPR2 stretch response is completely eliminated in 10~6 M AST-3. Real sensory neurons have a maximum firing rate and thus a limited-amplitude dynamic range. Beyond the stretch amplitude corresponding to this rate, the neuron no longer is sensitive to additional stretch. One result of AST-3 appli- cation is to extend the dynamic range, as shown in Figure 2. Reduction in sensitivity causes the curve of rate versus stretch to be linearized. Large-amplitude stretches that were indistinguishable under control conditions can be differen- tiated in the modulator. AST-3 application may also qualitatively change the fea- ture detection capability of the GPR2 neuron. Figure 3 shows control and modulated responses to a cpv3a muscle stretch generated from filtered white noise. The correspond- ing muscle stretch velocity is shown beneath the spike trains. Under control conditions, the amplitude of the stretch is well described in the spike times (Birmingham el al.. 1999). In a high concentration ( 1(T6 M) of AST-3, this is no longer the case. Many fewer spikes are generated, and it appears that those spikes might be correlated with large positive stretch velocities. What is the effect of AST-3 application on the GPR2 code? More experiments are required to answer this ques- tion. One possibility is that AST-3 simply reduces the 208 J. T BIRMINGHAM N D) C 5 4 - 3 - 0 - Control AST-3 0.08 0.10 0.12 0.14 0.16 Stretch (mm) Figure 2. Neuromodulation can increase dynamic range. GPR2 firing rate as a function of cpv3a stretch amplitude in control saline and in saline containing 1(T8 M AST-3. sensitivity of GPR2 in a linear fashion, so that essentially the same features in the stimulus are being encoded as under control conditions. Far more interesting is the possibility that AST-3 changes the filtering properties of the neuron. The presence of the peptide might change GPR2 from what is basically a position detector into a sensor with a response that has a stronger dependence on muscle velocity. GPR2 operates in two modes: the conventional "spiking" mode and a novel "bursting" mode (Katz et ai, 1989). In the spiking mode, which was the behavior shown in Figures 1-3, GPR2 generates spikes only during the stretch and is silent when not under tension. In contrast, in the bursting mode, GPR2 generates bursts of spikes (several seconds in duration) even when no stretch is imposed. In the initial description of the bursting mode, the average burst period was found to be -17 s (Katz et ai, 1989). More recently, burst periods ranging between 12 and 101 s have been reported (Birmingham et al.. 1999). GPR2 can be driven from the spiking mode into the bursting mode though pro- longed muscle activity (Birmingham et a!., 1999). The encoding properties of the neuron in the two modes are complementary. In the spiking mode, the neuron is best able to encode rapidly varying stretches, but not slower ones. In the bursting mode, fast stretches are ignored, while slower stretches (varying over minutes or hours) are encoded in the burst times (Birmingham et 600 /j,m and guard setae can be > 1200 /xm) (Gate and Derby, 2000). These different sensillar types often have restricted distributions. For exam- ple, aesthetascs are only on the distal half of antennular lateral flagella, and hedgehog and fringed sensilla are only on the distal two segments of some legs. The specific distributions, structure, and innervation of different sensilla undoubtedly have functional conse- quences. For example, the location of aesthetascs at the anterior end of the animal places them where odor-laden currents often first reach an animal and also where fan organs can direct odors toward or away from their own sensors (Breithaupt, 2001). In the same way, chemorecep- tors on the tips of legs are more likely to encounter chem- icals associated with the objects in the substrate, and che- moreceptors around the mouth will receive maximal stimulation when food is held in the mouthparts. The fact that some sensilla are innervated by both che- moreceptor neurons and mechanoreceptor neurons makes these bimodal sensilla ideal for identifying the spatial loca- tion of chemotactile stimuli, especially if the axons of these neurons project topotopically into the central nervous sys- tem. There is evidence that mechanoreceptor neurons and chemoreceptor neurons from bimodal sensilla on the anten- nules project to the same central neuropils (Schmidt et al., 1992; Schmidt and Ache, 1996a), but it is not known if their maps are overlapping as in some insects (Newland et al., 2000), or what is the spatial relationship between projec- tions from mechanoreceptor neurons and chemoreceptor neurons from the same sensillum. A sensory appendage can have a variety of chemorecep- tor neurons that differ in their sensitivities. Antennular chemoreceptor neurons of lobsters have diverse sensitivities to different types or qualities of odorants, thus broadening the range of chemical signals to which the entire organ is sensitive (Derby and Atema, 1988; Voigt and Atema, 1992; Derby. 2000). Antennular chemoreceptor neurons also dif- fer from each other in their sensitivities to odor concentra- tions (Derby and Atema. 1988; Merrill et al., 1994; Daniel ft ill., 1996) and in their ability to follow high-frequency odor pulses (e.g.. flicker-fusion properties) (Gomez et al., 1999). Thus, different neurons extract different stimulus features, together building an image of the chemical and mechanical features of the external world. Multiple Sensors Increase Response Sensitivity Through Response Summation Having multiple sensors allows animals to detect lower amplitude signals. This is because central neurons can sum responses from many sensory neurons, increasing signal capturing and the signal-to-noise ratio and thus allowing finer resolution between related signals (Van Drongelen et ill.. 1978; Meisami, 1989). For example, olfactory systems typically have at least tens to hundreds of thousands of receptor cells (Chase, 1986; Ache, 1991; Farbman, 1992; Hildebrand and Shepherd, 1997). If each neuron responded to one chemical with one spike above spontaneous activity and to another chemical with two spikes, a summation involving all neurons would dramatically amplify the dif- ference in the magnitude of the responses to the two stimuli. ROLE OF MULTIPLE CHEMOSENSORS thus providing the multineuronal system with a resolving capacity far beyond that of one or a few neurons. Multiple Sensors Ensure Functional Integrity of a Sense Organ Following Localized Damage Multiple sensors may serve to compensate for local dam- age to sensory organs. Given that chemosensors are used to probe an animal's external environment for chemical sig- nals, these sensors are often highly exposed and vulnerable to physical, chemical, and biological damage. Damage can affect sensory function (Hamilton and Case, 1983; Bauer, 1989), but multiple sensors minimize the impact of damage by increasing the probability that some sensors are undam- aged and functional (Daniel et ai, 2000). As an example, aesthetascs are repeating functional units, each containing a broad complement of receptor neurons with different sen- sitivities (Spencer, 1986; Mellon and Alones. 1993; Steullet er al., 2000b). This type of organization ensures that local- ized damage does not have significant impact on an ani- mal's ability to respond to and discriminate different types of chemicals (Steullet er al., 1999, 2000c; Horner et ai, 2000). In addition, many chemoreceptor systems, including the aesthetasc system, have mechanisms for self-renewal. Continuous turnover, including addition and loss, of che- moreceptors occurs in the vertebrates (Weiler and Farbman, 1997: Calof et al., 1998) and invertebrates (Chase, 1986; Chase and Rieling, 1986; Sandeman and Sandeman, 1996; Steullet et al., 2000a; Harrison et al.. 2001a. b). This turn- over enables the renewal of neurons that have a finite life span, which helps to maintain function of a system over time in spite of damage. Similar turnover also occurs in crustacean mechanoreceptor sensilla (Macmillan et al., 1998: Steullet et al., 2000a). Multiple Sensors Compensate for Nonfunctioning Developmental Stages of Sensors As animals develop and grow, they must add new sensors to maintain sensitivity over their enlarging body surfaces. To ensure that the animals can detect chemicals that im- pinge anywhere on the body surface, the number of sensors must increase both during development (Schafer and Sanchez. 1973; Laverack, 1988b) and throughout the lives of those species with indeterminate growth (Mellon and Alones, 1993). The antennular receptors of crustaceans are added as units in two ways (Sandeman and Sandeman. 1996; Steullet et al.. 2000a). First, the antennule is com- posed of many segments called annuli. and new annuli are added at molting. Second, new sensilla are added to existing annuli. In some cases, such as aesthetascs and their associ- ated chemo-mechanoreceptive sensilla, new sensilla are added in highly organized spatial arrays. In this way. the addition of annuli and sensilla creates a multiplicity of repeating, modular packages of sensors. In addition, turnover of neurons in olfactory systems causes some neurons to be nonfunctional because they are either immature or senescing (Chase, 1986; Farbman, 1992; Steullet et al.. 2000a). Since the percentage of nonfunc- tional cells is high in the vertebrates and invertebrates, there could be compensation by maintaining large numbers of sensors (Chase, 1986). Multiple Sensors Enable Formation of Specialized Central Processing Centers With Different Behavioral Functions If different sensor types are differentially coupled to central processors and motor programs, then a greater rich- ness in behavioral outputs should be possible. For example, it has been demonstrated that the function of a chemosen- sory neuron in Caenorhabditis elegans is determined by that neuron's central connections rather than by the receptor molecules that it expresses (Troemel et al., 1997). Crusta- ceans have not only a diversity of antennular chemo- and mechanoreceptor types but also a diversity of sensory neu- ropils that receive their projections. These sensory neuropils include olfactory lobes (OLs), which receive input from aesthetasc chemoreceptors (Mellon and Munger, 1990; San- deman et al.. 1992; Schmidt and Ache, 1996b); lateral antennular neuropils (LANs), which are thought to receive input from non-aesthetasc chemoreceptors and mechanore- ceptors on both lateral and medial antennular flagella (Schmidt et al., 1992; Schmidt and Ache, 1996a; Roye et al., 2000): and median antennular neuropils (MANs), which receive projections from statocysts, equilibrium receptors, and receptors from the antennular proximal segments (San- deman etal. 1992; Schmidt et al., 1992: Schmidt and Ache. 1993, 1996a; Cate and Roye. 1997). The organization of these neuropils gives hints about their function. The OLs have a glomerular neuropil, similar to the olfactory neuro- pils of insects (antennal neuropils) and vertebrates (olfac- tory bulbs), in which the glomeruli represent different but overlapping chemical sensitivities (Hildebrand and Shep- herd, 1997). This suggests that OLs encode chemical quality and are involved in olfactory discrimination. The LANs have a bilobed organization, with each lobe receiving input from one of the two antennular flagella (Schmidt tt al., 1992). Additionally, the lobes have a stratified organization, reflecting regionalized sensory input and stratified motor output from antennular motor neurons (Schmidt et al.. 1992; Schmidt and Ache, 1996a). This organization suggests that LANs may function in the sensori-motor control of anten- nular behavior, including flicking (Maynard. 1966) or dis- criminating the location of chemo-mechanostimulation on the antennule. The MANs are involved in maintaining equi- librium (Sandeman et ai, 1992; Schmidt and Ache, 1993; Cate and Roye. 1997), but their function may be broader than this (Fraser, 2001). 214 C. D. DERBY AND P. STEULLET Our analyses of the function of the different types of antennular sensors and antennular sensory neuropils in Ca- ribbean spiny lobsters suggest that there is some overlap in chemosensory function of the aesthetasc/OL pathway and the non-aesthetasc/LAN pathway (Steullet et at.. 1999. 2000c: Horner et al., 2000). Our working hypothesis is that antennular chemosensory neuropils have both unique and overlapping functions. Overlapping functions may include resolution of temporal characteristics of odors and media- tion of discrimination of odors such as food-related chem- icals. Functions unique to a neuropil may include, for the OLs. processing of pheromones (Gleeson, 1991 ). and for the LANs, determining location of chemotactile antennular stimulation and thus controlling local antennular reflexes (Maynard, 1966; Schmidt et al., 1992; Schmidt and Ache, 1993, 1996a; Roye et al., 2000). Multiple Sensors and the Design of Autonomous Robots Based on the assumption that the design of animals provides a guide for the principles to be used in constructing autonomous robots, we argue that such robots should have multiple sensors with the following characteristics. The sensors are sufficiently spatially distributed to sample stim- ulus space. They are sufficiently redundant to allow for signal summation and the resultant enhancement in sensi- tivity and resolving power. They are sufficiently diverse to sample the different stimulus qualities, intensities, and tem- poral profiles. Self-repair of sensors in autonomous robots may be too complicated to be feasible, but the functional equivalent could be achieved by having redundant sensors, only some of which are physically or functionally opera- tional at any one time. Dysfunction of some sensors could be detected and automatically compensated by bringing back-up sensors of similar types on-line. Such a system could function as a self-repair mechanism while limiting the energy demands and processing requirements of having all sensors simultaneously and constantly functional. Sensors with different sensitivities should be connected to the inte- grative and motor systems related to their function, such as attraction or avoidance. Acknowledgments We thank Paul Harrison. Holly Gate, and Amy Horner for discussions and comments on the manuscript, and acknowl- edge the grant support of NIH (DC00312). NSF (IBN 0077474), and the Georgia Research Alliance. Literature Cited Ache, B. W. 1991. Phylogeny of smell and taste. Pp. 3-18 in Smell and Time in Health and Disease. T. V. Getchell. R. L. Doty. L. M. Bartoshuk. and J. B. Snow. Jr.. eds. Raven Press. New York. Altner, H., C. Routil, and R. l.oftus. 1981. The structure of bimodal chemo-, thermo-. and hygroreceptive sensilla on the antenna of Locusta mignitoriii. Ceil Tissue Res. 215: 289-308. Altner, I., H. Halt, and H. Altner. 1983. Structural properties of bi- modal chemo- and mechanoreceptive setae on the pereiopods chelae of the crayfish, Austropotamobius torrentium. Cell Tissue Res. 228: 357- 374. Bauer, R. T. 1989. Decapod crustacean grooming: functional morphol- ogy, adaptive value, and phylogenetic significance. Pp. 48-73 in Func- tionci! Morphology of Feeding and Grooming in Crustacea, B. E. Felgenhauer. L. Watling. and A. B. Thistle, eds. A. A. Balkema, Rotterdam. Breithaupt, T. 2001. The fan organs of crayfish enhance chemical in- formation flow. Binl. Bull. 200: 150-154. Calof, A. L., J. S. M i. M i MI. P. C. Rim, and J. Shou. 1998. The neuronal stem cell of the olfactory epithelium. / Neurabiol. 36: 190-205. Caprio, J. 1988. Peripheral filters and chemoreceptor cells in fishes. Pp. 313-338 in Sensory Biology of Aquatic Animals. J. Atema. R. R. Fay, A. N. Popper, and W. N. Tavolga. eds. Springer, New York. Gate, H. S., and C. D. Derby. 2000. A novel chemo-/mechanosensillum that is widely distributed on the Caribbean spiny lobster and other lobsters. C/iem. Senses 25: 633-634 (abstract). Cate, H. S., and D. B. Roye. 1997. Ultrastructure and physiology of the outer row statolith sensilla of the blue crab Callinectes sapidua. J. Cruslac. Biol. 17: 398-411. Chase, R. 1986. Lessons from snail tentacles. Chem. Senses 11: 41 I - 426. Chase, R., and J. Rieling. 1986. Autoradiographic evidence for receptor cell renewal in the olfactory epithelium of a snail. Brain Res. 384: 232-239. Daniel. P. C.. M. F. Burgess, and C. D. Derby. 1996. Responses of olfactory receptor neurons in the spiny lobster to binary mixtures are predictable using a noncompetitive model that incorporates excitatory and inhibitory transduction pathways. J. Comp. Physiol. A 178: 523- 536. Daniel, P. C., C. D. Derby, D. Naram, and S. Saul. 2000. Function of antennular grooming behaviour in Caribbean spiny lobsters. Interna- tional Symposium on Olfaction and Taste XIII, 20-24 July, 2000. Brighton, UK. Int. Society for Neuroethology (Abstr]. Derby, C. D. 1989. Physiology of sensory neurons in morphologically identified cuticular sensilla of crustaceans. Pp. 27-47 in Functional Morphology of Feeding and Grooming in Crustacea, B. E. Felgen- hauer. L. Watling, and A. B. Thistle, eds. A. A. Balkema. Rotterdam. Derby, C. D. 2000. Learning from spiny lobsters about chemosensory coding of mixtures. Physiol. Behav. 69: 203-209. Derby, C. D., and J. Atema. 1988. Chemoreceptor cells in aquatic invertebrates: peripheral filtering mechanisms in decapod crustaceans. Pp. 365-388 in Sensory Biology of Aquatic Animals, 1. Atema, R. R. Fay, A. N. Popper, and W. N. Tavolga, eds. Springer. New York. Dethier, V. G. 1976. The Hungry Fly. A Physiological Study of the Behavior Associated with Feeding. Harvard University Press, Cam- bridge, MA. 1 .11 Inn. in. A. 1992. Cell Biology of Olfaction. Cambridge University Press, Cambridge. Fraser, P. J. 2001. Statocysts in crabs: short-term control of locomotion and long-term monitoring of hydrostatic pressure. Biol Bull. 200: 155-159. Gleeson, R. A. 1991. Intrinsic factors mediating pheromone communi- cation in the blue crab, Callinectes sapidus. Pp. 17-32 in Crustacean Sexual Biology. R. T. Bauer and J. W. Martin, eds. Columbia Univer- sity Press, New York. Gomez, G., R. Voigt, and J. Atema. 1999. Temporal resolution in olfaction. Ill: Flicker fusion and concentration-dependent synchroniza- tion with stimulus pulse trains of antennular chemoreceptor cells in the American lobster. J. Comp. Physiol. A 185: 427-436. ROLE OF MULTIPLE CHEMOSENSORS 2 Griinert. U., and B. W. Ache. 1988. Ultrastructure of the aesthetasc (olfactory) sensilla of the spiny lobster, Pain/lints argus. Cell Tissue Res. 251: 95-103. Hallberg, K., K. U. I. Johansson, and P. VVallen. 1997. Olfactory sensilla in crustaceans: morphology, sexual dimorphism and distribu- tion patterns. Int. .1. Insect Morphol. Embiyol. 26: 173-180. Hamilton, K. A., and J. F. Case. 1983. Effects of abrasion and Na+ on dactyl-mediated chemoreception in mature kelp crabs, Pugettin pro- Jut u: (Randall). J. Exp. Zoo/. 226: 363-372. Harrison, P. J. H., H. S. Cate, E. S. Swanson, and C. D. Derby. 2001a. Post-embryonic proliferation m the spiny lobster antennular epithe- lium: rate of genesis of olfactory receptor neurons is dependent on molt-stage. J. Neurobiol. (In press). Harrison, P. J. H., H. S. Cate, P. Steullet, and C. D. Derby. 200 Ib. Continuous change in the olfactory system of the Caribbean spiny lobster Pamilim\ tirgns: multiple mechanisms exist for growth, turn- over, and response to injury. Mar. Freshw. Res. 51. (In press). Hildebrand, J. G., and G. M. Shepherd. 1997. Mechanisms of olfac- tory discrimination: converging evidence for common principles across phyla. Anna. Rev. Neurosci. 20: 595-631. Horner, A. J., V. Ngo, P. Steullet, T. Keller, M. J. Weissburg, and C. D. Derby. 2000. The role of different types of antennular sensilla in orientation by Caribbean spiny lobsters to a natural odor stimulus under controlled flow conditions. Chem. Senses 25: 670-671 (abstract). Laverack, M. S. 1988a. The diversity of chemoreceptors. Pp. 287-312 in Sensory Biology of Aquatic Animals, 3. Atema, R. R. Fay, A. N. Popper, and W. N. Tavolga, eds. Springer. New York. Laverack, M. S. 1988b. The numbers of neurones in decapod Crustacea. J. Crustac. Biol. 8: 1-11. Macmillan, D. L., T. Stuart, and M. Thomas. 1998. Development of a proprioceptive organ on the walking legs of the rock lobster Jusus edwardsii (Decapoda: Palinuridae) by ordered addition and loss of receptor elements. J. Crustac. Biol. 18: 1-9. Maynard, D. M. 1966. Integration in crustacean ganglia. Symp. Soc. Exp. Biol. 20: 111-149. Meisami, E. 1989. A proposed relationship between increases in the number of olfactory receptor neurons, convergence ratio and sensitivity in the developing rat. Dev. Brain Res. 46: 9-19. Mellon, D., Jr., and V. Alones. 1993. Cellular organization and growth- related plasticity of the crayfish olfactory midbrain. Microsc. Res. Tech. 24: 231-259. Mellon, D., Jr., and S. D. Munger. 1990. Nontopographic projection of olfactory sensory neurons in the crayfish brain. J. Comp. Neurol. 296: 253-262. Merrill, C. L., R. Voigt, and J. Atema. 1994. Reliability of chemore- ceptor cell response. I. Intensity coding by pattern and response mag- nitude with a comparison of analytical methods. J. Comp. Physio/. A 175: 95-105. Newland, P. L., S. M. Rogers, I. Goaboub, and T. Matheson. 2000. Parallel somatotopic maps of gustatory and mechanosensory neurons in the central nervous system of an insect. J. Comp. Neural. 425: 82-96. Rove, D. B., C. M. Kilroy, J. M. Doyle, and L. E. Reuss. 2000. Decussating interneurons mediate antennular withdrawal in the blue crab. Callinectes sapidus. J. Crustac. Biol. 20: 603-613. Sandeman, D., R. Sandeman, C. Derby, and M. Schmidt. 1992. Mor- phology of the brain of crayfish, crabs, and spiny lobsters: a common nomenclature for homologous structures. Biol. Bull. 183: 304-326. Sandeman. R., and D. C. Sandeman. 1996. Pre- and postembryonic development, growth and turnover of olfactory receptor neurones in crayfish antennules. J. Exp. Biol. 199: 2409-2418. Schafer, R., and T. Sanchez. 1973. Antennal sensory system of the cockroach. Periplaneta ainericana: postembryonic development and morphology of the sense organs. J. Comp. Neural. 149: 335-354. Schmidt, M. 1989. The hair-peg organs of the shore crab, Carcinus maenas (Crustacea, Decapoda): ultrastructure and functional properties of sensilla sensitive to changes in seawater concentration. Cell Tissue Res. 257: 609-621. Schmidt, M., and B. W. Ache. 1993. Antennular projections to the midbrain of the spiny lobster. III. Central arborizations of motoneurons. J. Comp. Neural. 336: 583-594. Schmidt, M., and B. VV. Ache. 1996a. Processing of antennular input in the brain of the spiny lobster, Puiiulinis urgus. I. Non-olfactory che- mosensory and mechanosensory pathway of the lateral and median antennular neuropils. J. Comp. Physiol. A 178: 579-604. Schmidt, M., and B. VV. Ache. 1996b. Processing of antennular input in the brain of the spiny lobster. Panulirus argus. II. The olfactory pathway. J. Comp. Physiol. A 178: 605-628. Schmidt, M., and VV. Gnatzy. 1984. Are the funnel-canal organs the "campaniform sensilla" of the shore crab, Carcinus maenas (Decapoda, Crustacea)? II. Ultrastructure. Cell Tissue Res. 237: 81-93. Schmidt, M., L. Van Ekeris. and B. VV. Ache. 1992. Antennular projections to the midbrain of the spiny lobster. I. Sensory innervation of the lateral and medial antennular neuropils. J. Comp. Neural. 318: 277-290. Shanbhag, S. R., B. Mueller, and R. A. Steinbrecht. 1999. Atlas of olfactory organs of Drosophila nielanogaster. 1 . Types, external orga- nization, innervation and distribution of olfactory sensilla. Int. J. Insect Morphol. Embiyol. 28: 377-397. Spencer, M. 1986. The innervation and chemical sensitivity of single aesthetasc hairs. J. Comp. Physiol. A 158: 59-68. Steullet, P., T. Flavus, D. Radman, G. Hamidani, M. Zhou, O. Dudar, R. Hill, and C. D. Derby. 1999. The aesthetasc-olfactory lobe path- way of spiny lobsters is not necessary for odor-activated searching behavior, odor-associative learning, and discrimination of complex odors. Chem. Senses 24: 613 (abstract). Steullet. P., H. S. Cate, and C. D. Derby. 2000a. A spatio-temporal wave of turnover and functional maturation of olfactory receptor neu- rons in the spiny lobster Panulirus argus. J. Neurosci. 20: 3282-3294. Steullet, P., H. S. Cate, W. C. Michel, and C. D. Derby. 2000b. Functional units of a compound nose: aesthetasc sensilla house similar populations of olfactory receptor neurons on the crustacean antennule. J. Comp. Neurol. 418: 270-280. Steullet, P., D. R. Kruetzfeldt, G. Hamidani, T. Flavus, and C. I). Derby. 2000c. Functional overlap of two antennular chemosensory pathways in food odor discrimination behavior of spiny lobsters. Che m. Senses 25: 671 (abstract). Troemel, E. R., B. E. Kimmd, and C. I. Bargmann. 1997. Reprogram- ming chemotaxis responses: sensory neurons define olfactory prefer- ences in C. elegans. Cell 91: 161-169. Van Drongelen, W., A. Holley, and K. B. Deving. 1978. Convergence in the olfactory system: quantitative aspects of odour sensitivity. J. Theor. Biol. 71: 39-48. Voigt, R., and J. Atema. 1992. Tuning «il Jiemoreceptor cells of the second antenna of the American lobster (Humunts amencunus) with a comparison of four of its other chemoreceptor organs. J. Comp. Physiol. A 171: 673-683. Weiler. E., and A. I. Farbman. 1997. Proliferation in the rat olfactory epithelium: age-dependent changes. J. Neurosci, 17: 3610-3622. Reference: Biol. Bull. 200: 216-221. (April 2001) Landing Strategies in Honeybees, and Possible Applications to Autonomous Airborne Vehicles MANDYAM V. SRINIVASAN*. SHAOWU ZHANG, AND JAVAAN S. CHAHL Centre for Visual Science, Research School of Biological Sciences, Australian National University, P.O. Box 475, Canberra, A.C.T. 2601, Australia Abstract. Insects, being perhaps more reliant on image motion cues than mammals or higher vertebrates, are prov- ing to be an excellent organism in which to investigate how information on optic flow is exploited to guide locomotion and navigation. This paper describes one example, illustrat- ing how bees perform grazing landings on a flat surface. A smooth landing is achieved by a surprisingly simple and elegant strategy: image velocity is held constant as the surface is approached, thus automatically ensuring that flight speed is close to zero at touchdown. No explicit knowledge of flight speed or height above the ground is necessary. The feasibility of this landing strategy is tested by implementation in a robotic gantry, and its applicability to autonomous airborne vehicles is discussed. Introduction Unlike vertebrates, insects have immobile eyes with fixed-focus optics. Therefore, they cannot infer the dis- tances to objects or surfaces from the extent to which the directions of gaze must converge to view the object, or by monitoring the refractive power that is required to bring the image of the object into focus on the retina. Furthermore, compared with human eyes, the eyes of insects are posi- tioned much closer together and have inferior spatial acuity (Horridge. 1977). Therefore, the precision with which in- * To whom correspondence should be addressed. E-mail: M.Srinivasan@anu.edu.au. This paper was originally presented al a workshop titled Invertebrate Sensory Information Processing: Implications for Biologically Inspired Autonomous Systems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences, Woods Hole, Massachusetts, from 15-17 April 2000, was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. sects could estimate range through binocular stereopsis would be much poorer and restricted to relatively small distances, even if they possessed the requisite neural appa- ratus (Srinivasan, 1993). Not surprisingly, then, insects have evolved alternative strategies for dealing with the problems of visually guided flight. Many of these strategies rely on using image motion, generated by the insect's own motion, to infer the distances to obstacles and to control various maneuvers (Horridge. 1987; Srinivasan. 1993. 1998). Here we describe how honeybees use image motion cues to perform smooth landings on a flat surface. How Bees Perform Smooth Landings The seminal work of Gibson ( 1950) highlighted the optic- flow cues that can be brought to bear in controlling the landing of an aircraft. Studies of landing behavior in flies have revealed that, as a surface is approached, the expansion of the image of the surface provides strong cues that are used to control deceleration and trigger extension of the legs in preparation for contact (Goodman, 1960; Eckert and Hamdorf, 1980; Wagner, 1982; Borst and Bahde, 1988). There is also evidence that the rate of expansion of the image is used to infer the time to contact the surface, even if the insect does not possess explicit information about the speed of its flight or the distance to the surface (Wagner, 1982). However, when an insect makes a grazing landing on a flat surface, cues derived from image expansion are rela- tively weak. This is because the dominant pattern of image motion is then a translatory flow in the front-to-back direc- tion. Given that flying insects often make grazing landings on flat surfaces, what are the processes by which such landings are orchestrated? Recently, this question was investigated by Srinivasan et 216 HONEYBEE LANDING STRATEGIES 21, 10 y, cm 10 y, cm V (Forward speed) -y. (Descent angle) Figure 1. (a. b) Three-dimensional reconstruction of two typical landing trajectories, from video films. Vertical lines depict height above surface, (c) Illustration of some of the variables analyzed to investigate the control of landing, h (cm): height above surface; Vf(cm/s): horizontal (forward) flight speed; Vj (cm/s): vertical (descent) speed; Tan~'(Vd/Vf) (deg or radl: descent angle. Adapted from Srinivasan el al. (2000). al. (2000), who video-filmed trajectories, in three dimen- sions, of bees landing on a flat, horizontal surface. Two examples of landing trajectories, reconstructed from the data, are shown in Figure la, b. A number of such landing trajectories were analyzed to examine the variation of the instantaneous height above the surface (h). instanta- neous horizontal (forward) flight speed (Vf), instantaneous descent speed (Vt/) and descent angle (a). These variables are illustrated in Figure Ic. Analysis of the landing trajectories revealed that the descent angles were indeed quite shallow. The average value measured in 26 trajectories was about 28° (Srinivasan et al., 2000). Figure 2a, b shows the variation of flight speed with height above the surface, analyzed for two landing trajec- tories. These data reveal one of the most striking and consistent observations with regard to landing bees: Hori- zontal speed is roughly proportional to height, as indicated by the linear regression on the data. When a bee flies at a horizontal speed of Vf cm/s at a height of /; cm. the angular velocity w of the image of the surface directly beneath the eye is given by u> = Vflh rad/s. From this relationship it is clear that, if the bee's horizontal flight speed is proportional to her height above the surface (as shown by the data), then the angular velocity of the image of the surface, as seen by the eye, must be constant as the bee approaches it. This angular velocity is given by the slope of the regression line. The angular velocity of the image varies from one trajectory to another, but is maintained al an approximately constant value in any given landing. An analysis of 26 landing trajectories revealed a mean image angular velocity of about 500°/s (Srinivasan el al, 2000). These results reveal two important characteristics. First, bees landing on a horizontal surface tend to approach the 218 M. V. SRINIVASAN ET AL Iand09 Iand11 Mean image angular vel: 241deg/s 10 20 Height, cm 140 •Sj o 12° 1 100 ™ 80 01 £ 60 c 40 o N 20 'Vf = 11.43.h -9.81 r=0.96 Mean image angular vel: 655deg/s 5 10 Height, cm 15 Iand09 Iand17 120 100 in I 80 1 60 a. | 40 sel, as it descends. In the first time step, the gantry moves the camera head along the direction of descent at an arbitrarily chosen initial speed. The image velocity is measured during this step, using the image interpolation algorithm. Let us denote the measured image velocity by w,m.(M. In the next step, the speed of motion of the head is increased or decreased, depending The speed of descent is also corrected by the same factor, since the forward and descent speeds are proportional to each other and linked by the desired angle of descent. This speed correction ensures that the image velocity during the next step will have the desired value w^r provided the camera maintains its present altitude. However, since the camera continues to descend during the new step, the for- ward speed in the following step would have to be reduced further. Thus, both the forward and descent speeds decrease continuously as the camera descends, reaching very low values when the camera is close to the ground. Landing trajectories generated, using this procedure, by the gantry are shown in Figure 4a for three descent angles. The image velocities maintained during these three landings are shown in Figure 4b. It is clear that the image velocity, though somewhat noisy, is held approximately constant. The height of the camera decreases exponentially with time (Fig. 4c), as do the forward speed and the speed of descent. These behaviors are as expected: a quantitative model of the landing strategy predicts and verifies that these variables do indeed vary exponentially with time (Srinivasan et al., 2000). The results with the robotic gantry suggest that the strat- egy proposed is a feasible one for landing on flat surfaces, provided the surface carries visual texture that will enable the measurement of image motion. In undulating terrain, the system reduces the forward and descent speeds when the ground rises toward the camera, and increases them when the ground falls away. This is obviously a desirable feature, but has limitations in that the system cannot cope with a situation in which the ground in front rises abruptly to a level above the camera's current height. A little reflection will reveal that the landing strategy described here can be used by an aerial vehicle to dock with any flat surface, regardless of its orientation: horizontal, vertical, or oblique. All that is required is that the vehicle approach the surface in a straight line and hold the image velocity constant during the approach. This will automati- cally ensure that the vehicle's speed decreases as the surface is approached, ensuring smooth docking. In the special case in which the surface is approached perpendicularly, the image velocity will be zero in the "straight ahead" direction: the flow field has a pole there. However, the strategy can still be implemented by holding constant the image speed in an annular region surrounding the pole, or in a large region 220 M. V. SRINIVASAN ET AL. 1000 800 1 600 en '5 400 200 0 500 1000 1500 2000 Horizontal distance travelled, mm 1000 50 100 Time, arbitrary steps 150 50 100 Time, arbitrary steps 50 100 Time, arbitrary steps 150 Figure 4. Landing trajectories generated by the robotic gantry, (a) Height versus distance traveled for three descent angles: —26.5° (circles), —45° (squares), and —63.5° (triangles), (b) Variation of image angular velocity as a function of time. The symbols in this and other panels refer to the three different descent angles, as in (a). (c) Variation of height with time, (d) Variation of forward speed with time. The line curves in (c) and (d) depict least-squares fits of exponential functions to the data. Descent speed also declines exponentially (data not shown). Adapted from Srinivasan el ul. (2000). centered on the pole. Although the present study does not reveal whether bees are actually "aware" of the orientation of the surface in relation to the direction of their approach, it is clear from the above discussion that this information is not necessary for executing the landing process. Conclusions Analysis of vision in simple natural systems, such as those found in insects, can often point to novel ways of tackling tenacious problems in autonomous navigation. This is probably because insects, with their "stripped down" nervous systems, have been forced to evolve ingenious strategies to cope with visual challenges within their envi- ronment. This article has outlined a surprisingly simple way in which insects use motion cues to perform smooth land- ings on flat surfaces. The next step is to investigate whether this principle can be used to advantage in the design of visually based control systems for autonomously flying vehicles. Acknowledgments We thank the anonymous referees for their assistance in improving the manuscript. This research was supported partly by a grant from the Australian Defence Science and Technology Organisation, Salisbury, Grant RG 84/97 from the Human Frontiers in Science Program, and Grant NOOO 14-99- 1-0506 from the U.S. Defense Advanced Re- search Projects Agency and the Office of Naval Research. Literature Cited Borst, A., and S. Bahde. 1988. Visual information processing in the fly's landing system. / OWH/J. Physiol. A 163: 167-173. Eckert, H., and K. Hamdorf. 1980. Excitatory and inhibitory response components in the landing response of the blowfly. Calliphora eryth- rocephalu. J. Com/7. Physiol. 138: 253-264. Gibson, J. J. 1950. The Perception of the Visual World. Houghton Mifrlin. Boston. Goodman, L. J. 1960. The landing responses of insects. I. The landing response of the fly, Lucilia sericata, and other Calliphoridae. J. E.\p. Binl. 37: 854-878. HONEYBEE LANDING STRATEGIES Gotz, K. G., and U. Wandel. 1984. Optomotor control of the force of Srinivasan, M. V. 1994. An image-interpolation technique for the corn- flight in Drosophila and Musca. II. Covanance of lift and thrust in still putation of optic flow and egomotion. Bio/. Cybern. 71: 401-416. air. Bial. Cvbern. 51: 135-139. Srinivasan, M. V. 1998. Insects as Gibsonian animals. Ecol. Psycho/. Horridge, G. A. 1977. Insects which turn and look. Endeavour 1: 7-17. 10: 251-270. Horridge. G. A. 1987. The evolution of visual processing and the Srinivasan, M. V., S. W. Zhang, J. S. Chahl, E. Barth, and S. Yen- construction of seeing systems. Proc. R. Soc. Land. B 230: 279-292. katesh. 2000. How honeybees make grazing landings on flat sur- Srinivasan, M. V. 1993. How insects infer range from visual motion. Pp. faces. Biol. Cybern. 83: 171-183. 139-156 in Visual Motion and Its Role in the Stabilization of Gaze, Wagner, H. 1982. Flow-field variables trigger landing flies. Nature 297: F. A. Miles and J. Wallman. eds. Elsevier. Amsterdam. 147-148. Reference: Bio/. Bull. 200: 222-226. (April 2001) Plume-Tracking Robots: A New Application of Chemical Sensors HIROSHI ISHIDA*'1, TAKAMICHI NAKAMOTO2, TOYOSAKA MORIIZUMI2, TIMO KIKAS1, AND JIRI JANATA1 1 School of Chemistry and Biochemistry, Georgia Institute of Technology, Atlanta, GA 30332-0400; and 'Department of Physical Electronics, Tokyo Institute of Technology, 2-12-1 Ookayanui, Meguro-ku, Tokyo 152-8552, Japan Abstract. Many animals have the ability to search for odor sources by tracking their plumes. Some of the key features of this search behavior have been successfully transferred to robot platforms, although the capabilities of animals are still beyond the current level of sensor technol- ogies. The examples described in this paper are ( 1 ) incor- porating into a wheeled robot the upwind surges and casting used by moths in tracking pheromone plumes, (2) extracting useful information from the response patterns of a chemical sensor array patterned after the spatially distributed chemo- receptors of some animals, and (3) mimicking the fanning behavior of silkworm moths to enhance the reception of chemical signals by drawing molecules from one direction. The achievements so far and current efforts are reviewed to illustrate the steps to be taken toward future development of this technology. Introduction Many species of animals rely for their survival on their ability to track odor plumes. Male moths follow sexual pheromones to find their mates (Willis and Arbas, 1991) and Received 23 August 2000; accepted 1 1 January 2001. * To whom correspondence should be addressed. Current address: De- partment of Physical Electronics. Tokyo Institute of Technology. 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8552, Japan. E-mail: ishida@ ee.titech.ac.jp This paper was originally presented at a workshop titled Invertebrate Sensoty Information Processing: Implications for Biologically Inspired Autonomous Systems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences, Woods Hole, Massachusetts, from 15-17 April 2000, was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. marine crustaceans track smells to search for food (Weiss- burg and Zimmer-Faust, 1993; Atema. 1996; Grasso, 2001 ). Although their mechanisms are not yet fully revealed, these examples suggest that, if the underlying mechanisms are successfully transferred to robotic platforms, the chemical senses can be as effective as the visual (Viollet and France- schini. 1999) or auditory senses (Webb and Scutt, 2000) in directing navigation of autonomous robots. The purpose of this paper is to give a general idea of the current state of development of robots that track chemical plumes. The potential applications for such robots include searching for hazardous chemicals and pollutant sources. Signals Available in Chemical Plumes The most fundamental characteristics of a chemical plume are concentration of the target chemical and flow direction. Since molecular diffusion is almost always slower than convection, a chemical plume trails downstream from its source, as shown in Figure 1A. There is no smooth concentration gradient in this instantaneous image since the eddies in the turbulent flow stretch and twist the plume. If an array of eight chemical sensors is immersed into the plume, it is exposed to the intermittent stimuli shown in Figure 2. The importance of flow direction is also seen in Figure 1 A. When a searcher is in the plume, the source is always in the upstream direction. This is why many species of animals perform anemotaxis in air and rheotaxis in water (Willis and Arbas, 1991; Dusenbery. 1992; Arbas et «/.. 1993; Weiss- burg and Zimmer-Faust, 1994). Whereas animals have keen senses for detecting chemical and flow stimuli (Dusenbery, 1992), sensors for a robot to detect those signals with capabilities similar to those of animals are not yet available. In the case of chemical 222 PLUME-TRACKING ROBOTS Source location -Flow direction Sensor array in Figs. 2B-D ,\ r ^ 1 0 cm " Source location / — ^- Flow direction 10 cm Figure 1. Images of a chemical plume in a turbulent flow. A fluores- cent dye, rhodamine 6G. was released from a nozzle inio a fully developed open-channel water flow. The effluent velocity of the dye matched the channel flow velocity, thus creating a passive source and avoiding the production of additional turbulence by the effluent itself. The images of the dye plume were captured using the planar laser-induced fluorescence (PLIF) measurement technique (Webster et al.. 1999). (A) is an instanta- neous image, and (B) is the average of 6000 images measured at 10 frames/s. sensors, a compromise has been made on the times of response onset and recovery, which can affect the perfor- mance of robots. For example, for our wheeled robot we chose semiconductor gas sensors that are commonly used for gas alarms to detect flammable gases and organic sol- vent vapors. Although these sensors have high sensitivity and fast response, it takes more than 30 s for their responses to return to the initial levels after gas is removed. Therefore, the crawling speed of the robot had to be slowed down to a few centimeters per second (Ishida et al., 1996). Such slow sensors can be beneficial when time is not a critical issue. When averaged over long enough times (typically several minutes), chemical plumes have continuous concentration gradients, as shown in Figure IB. By smoothing the fluc- tuating signals, slow sensors make it easier to acquire this gradient information. Iterative measurements of time-averaged gradients until a robot reaches an odor source may need hours, however, and most applications do not permit such long search. Attempts have been made to extract rapidly changing signals from dull sensor responses by filtering or differentiation of sensor outputs (Nakamoto et al.. 1996b; Webb, 1998). Another option is to use faster sensors, such as quartz crystal mi- crobalance sensors, although their sensitivities tend to be lower than those of semiconductor gas sensors. For under- water plume tracking, amperometric microelectrode sensors are promising. In the amperometric mode of sensing, a constant voltage is applied between a microelectrode and a counter electrode so that electrochemical reaction of a target chemical substance takes place on the electrodes. The signal measured is a current proportional to the concentration. A short response time comparable to that of animals' chemo- receptors (100 ms) can be easily achieved (Moore and Atema, 1991; Kikas et al., 2000). The direction of airflow can be detected by an array of hot wire sensors (Ishida et al., 1994). However, wind slower than 5 cm/s is hard to detect (Lomas, 1986), and detection limits lower than this are required in some indoor environ- ments. If a robot is used to detect gas leaks in industrial or domestic buildings with moderate air-conditioning, it often PLIF image Pixel- 1 cm 4 5 678 _ c, a a Pixel values at eight locations jj \2\ |3| ~ei ITI IT B D D D D D D D D E D D LJ D cnc n i : n i n n n D n n n n G D D D n n Figure 2. Concentration change in a plume represented by stimulus patterns exposed to chemical sensors. Eight pixels are chosen from PLIF images representing an array of eight sensors. The gray-scale value of each pixel is then drawn in a blown-up picture as shown in (A). Darker pixels represent higher concentrations. (B)-(D) are a series of stimulus patterns when the array position is 50 cm downstream from the source (Fig. I A), and (E)-(G) are the patterns when the eight pixels surrounding the source are observed in successive 0.2-s intervals. The size and the location of the array shown in (B)-(D) are depicted in Figure 1A. 224 H. ISHIDA ET AL encounters wind fields with velocities less than 5 cm/s (Pluijm et al., 1986). Search strategies that do not require flow sensors should be considered in this case. Chemical sensor arrays and the odor compass described later have been developed to address this issue (Nakamoto et al., 1996a; Ishida et al.. 2000). The detection limits of various types of flow sensors are lowered in water because of the higher viscosity, heat capacity, and density of water com- pared to air (Lomas, 1986; Roberson and Crowe, 1997). Realization of Various Aspects of Plume Tracking Multiphase search algorithm Animals show different types of behaviors in different situations (Dusenbery. 1992). Our multiphase algorithm for a wheeled robot was divided into anemotactic and chemo- tactic strategies (Ishida et al., 1996). Although this combi- nation was devised after many experiments using robots, each of those two is a fundamental search strategy seen in many animal species (Dusenbery. 1992; Arbas et al., 1993). The anemotactic strategy is based on moths tracking pher- omone plumes. The key feature implemented into the robot is to track a chemical plume by using wind direction, as a moth shows upwind surges when it perceives a pheromone (Willis and Arbas, 1991 ). When the robot by chance leaves the plume, it tries to relocate the lost plume by moving back and forth across the wind. This fail-safe mechanism, known in moths as casting flight (Willis and Arbas, 1991), is extremely important for successful plume tracking because of the random nature of a turbulent plume. Another important feature of the multiphase algorithm is chemotactic search. It was added to the multiphase algo- rithm to cope with winds coming from multiple directions simultaneously (Ishida et al., 1996). In a domestic or indus- trial building, the main source of wind is an air conditioner, and a robot often encounters winds from multiple air-supply openings. However, the anemotactic strategy described above failed in a clean room with two air-supply openings (Ishida et al., 1996). In the experiments, an ethanol vapor source was placed near one of the openings. A plume extended from it approximately 1 m until the wind from another opening merged into a side of the plume. In such a case, care should be taken to employ the anemotactic strat- egy. When the detected concentration is low, the robot might be in the merging area where unstable winds often direct an anemotactic robot to wrong directions. Anemo- taxis should be activated only when a high concentration is detected, thus only the wind from the source direction is considered to exist. In the multiphase algorithm, the robot is programmed to employ the chemotactic strategy first, and anemotaxis is activated only when a concentration above a predefined threshold is detected. This change in strategies helps to prevent the robot from being trapped in the area where the wind direction is not stable (Ishida et al., 1996). The robot is also programmed to change its strategies from anemotaxis to chemotaxis, or vice versa, when one strategy makes no significant progress for 60 s (Ishida et al.. 1996). This ensures timely changes in strategies even when the pre- defined threshold is inappropriate. To accomplish fully autonomous search, there still re- main many questions, including how to locate a plume for the first time in the absence of any chemical signals, and how to decide when the odor source has been located so as to terminate the search. These issues have been more fully discussed by Dusenbery (1989) and Ishida et al. (1998). Array mode of information acquisition Use of an array of sensors should provide more informa- tion than a pair or a limited number of chemical sensors, thus enhancing simple gradient detection. Extensive and spatially distributed arrays of chemoreceptors are used by many animals, including arthropods (Keller and Weissburg, 2000). Various array configurations and signal processing algorithms are being investigated in simulations (Ishida et al., 2000), and they are being evaluated by their ability to acquire flow vector and source location. For example, con- sider an array of amperometric sensors with fast and linear response. Since their outputs faithfully trace the input stim- uli, the stimulus pattern shown in Figure 2 can be regarded as the output pattern of the sensors. Keller and Weissburg (2000) showed that blue crabs use not only their antennules but also chemoreceptors on other appendages for tracking odor plumes. A crab of 1 cm in size would experience similar response patterns. As mentioned in the previous section, amperometric microelectrode sensors have re- sponse fast enough to yield such output patterns, and an array of this size can be easily fabricated and mounted on an underwater vehicle. As seen in Figure 2, the observed responses are highly intermittent since they show instantaneous concentrations with no integration over time. This intermittency enables tracking of patches of the plume. When a patch passes over the sensor array from sensor 6 through 7 to 8, the flow direction and speed can be determined from this change in the response pattern (Fig. 2B-D). Figure 2E-G shows the response pattern when the array is placed over the source location. From the responses of sensors 2 and 3. the flow can be determined to move from left to right. However, sensors 1, 4, and 6 at the upstream edge of the array do not show any response. Considering the fact that the plume always trails in the downstream direc- tion, this response pattern indicates that the source is located within the array. PLUME-TRACKING ROBOTS 225 Modulation of chemical signals Marine crustaceans flick their antennules, and terrestrial vertebrates show sniffing behavior. These actions modulate the dynamics of chemical signals at the animals' sensors. An interesting example of this signal modulation is the wing fanning of a male silkworm moth tracking a pheromone plume. Attempts are also being made to understand the fanning behavior of other animals (Breithaupt and Ayers. 1998: Breithaupt, 2001). Mimicking this mechanism, a sensing probe, termed an "odor compass," consisting of two gas sensors and a small fan to draw air to the sensors was fabricated (Nakamoto et al.. 1996a). Experiments showed that the effect of the fan is significant in obtaining direc- tional cues. When an odor source is in front of the compass, odor molecules are carried towards both the left and right sensors by airflow produced by the fan. On the other hand, they are repelled by the airflow when a source is behind the compass. The obtained sensor outputs thus change accord- ing to the direction of the compass with respect to an odor source. Fanning was also found to enhance differences between the outputs of the left and right sensors. When a source is to the right of the compass, turning on the fan increases the output of the right sensor and decreases that of the left sensor. Therefore, the direction of an odor source is determined by rotating the compass and finding the direc- tion where the sensors show the largest and matched re- sponse (Nakamoto et al.. 1996a). Simulations Simulation is a useful way to accelerate the development of plume-tracking systems, since it takes a long time to develop a new system by trial and error. There are several levels of simulations. The most common way is to simulate all the dynamics of fluid and the kinetics of sensors and robots in a computer. However, this is impractical because the simulation of turbulent flow with fine resolution requires massive computer power. Visualization techniques are more realistic methods of obtaining concentration data on a fine scale (Femeretal.. 1993: Webster et al, 1999). If computer models for the kinetics of sensors or robots (or animals) are used, the simulation can be based on visualized plume data (Ishida et al, 2000). This level of simulation is effective in testing algorithms and determining the specifications for robots. The next level of simulation can be performed with a "virtual plume ( VP)." which is a test bench for an array of real chemical sensors (Kikas et al, 2000). This system is based on the principles of multichannel flow injection anal- ysis (FIA)/sequential injection analysis (SIA) (Ruzicka and Hansen, 1988) and the array of amperometric microelec- trode sensors. A plug of chemical solution is introduced into a flow of carrier solution in tubing, divided into eight channels, and delivered to the sensors after passing through delay and dispersion elements. The delay element is a coil of tube that adjusts the timing with which a divided plug reaches the sensor at the end of the channel. The dispersion element is a small mixing chamber that adjusts the onset slope of the plug. By using those elements, VP can create various stimulus patterns that the sensor array would expe- rience in real plumes. Discussion Research on plume trucking is still at its initial stages of development. Animals show a variety of behaviors, each of which is optimized for the habitat of that species. There seems to be no single search algorithm that is effective in every environment. Considering the variety of situations in which plume-tracking robots will be used, future work should use diverse algorithms and situations. Various as- pects of the mechanisms underlying animal behavior have been transferred to robots through the interaction between biology and engineering. Although we should continue such efforts, we are now at a point where we can begin a synthesis of the data to develop a unified strategy for the design of plume-tracking systems. The first step toward this goal would be to evaluate the applicability of each algo- rithm in various situations. The limit of that algorithm will, then, be clarified, and quantitative comparison between different algorithms will be possible. The simulation tech- niques are expected to facilitate this progress by providing a way to compare a variety of situations quickly. Acknowledgments We thank Drs. Frank Grasso and Diana Blazis for inviting us for the insightful interdisciplinary workshop. We are grateful to Drs. Philip Roberts and Donald Webster for providing access to their LIF plume data sets. Enlightening discussions with Drs. Marc Weissburg, David Dusenbery, and Troy Keller are also gratefully acknowledged. We also thank two anonymous reviewers for valuable suggestions and fruitful criticisms. This work was partially supported by a grant from DARPA/ONR, Project Number NOOO 14-98-1- 0776. Literature Cited Arbas, E. A., M. A. Willis, and R. Kanzaki. 1993. Organization of goal-oriented locomotion: pheromone-modulaled flight behavior of moths. Pp. 159-198 in Biological Neural Ncnwrks in Invertebrate Neiiroetliology and Robotic.',. R. D. Beer. R. E. Ritzmann. and T. McKenna, eds. Academic Press. San Diego. Atema, J. 1996. Eddy chemotaxis and odor landscapes: exploration of nature with animal sensors. [iiol. Bull. 191: 129-138. Breithaupt, Thomas. 2001. Fan organs of crayfish enhance chemical information flow. Bi»l. Bull. 200: 150-154. Breithaupt, T., and J. Ayers. 1998. Visualization and quantification of biological flow fields through video-based digital motion-analysis tech- niques. Mar. Freshw. Behav. Physiol. 31: 55-61. 226 H. ISHIDA ET AL Dusenbery, D. B. 1989. Optimal search direction for an animal flying or swimming in a wind or current. J. Chum. Ecol. 15: 2511-2519. Dusenbery, D. B. 1992. Sensory Ecology: How Organisms Acquire and Respond to Information. W. H. Freeman. New York. Ferrier. A. J.. D. R. Funk, and P. J. VV. Roberts. 1993. Application of optical techniques to the study of plumes in stratified fluids. Dyn. Atmos. Oceans 20: 155-183. Grasso, Frank W. 2001. Invertebrate-inspired sensory-motor systems and autonomous, olfactory-guided exploration. Bio/. Bull. 200: 160— 168. Khnhi, H., K. Suetsugu, T. Nakamoto. and T. Moriizumi. 1994. Study of autonomous mobile sensing system for localization of odor source using gas sensors and anemometric sensors. Sens. Actuators A 45: 153-157. Ishida, H., Y. Kagawa, T. Nakamoto, and T. Moriizumi. 1996. Odor- source localization in the clean room by an autonomous mobile sensing system. Sens. Actuators B 33: 115-121. Ishida, H., T. Nakamoto, and T. Moriizumi. 1998. Remote sensing of gas/odor source location and concentration distribution using mobile system. Sens. Actuators B 49: 52-57. Ishida, H., T. Yamanaka, N. Kushida, T. Nakamoto, and T. Moriizumi. 2000. Study of real-time visualization of gas/odor flow image using gas sensor array. Sens. Actuators B 65: 14-16. Keller, T. A., and M. J. Weissburg. 2000. Chemically-mediated search strategies: elucidating the nature of stimulus dynamics for searching blue crabs (Cal/inectes sapidus). (Abstr.] Aquatic Science Meeting, Copenhagen, 5-9 June 2000. Am. Soc. Limnol. Oceanogr. Kikas, T., H. Ishida, P. J. W. Roberts. D. R. Webster, and J. Janata. 2000. Virtual plume. Electrowiul\\is 12: 974-979. Lomas, C. G. 1986. Fundamentals of Hot Wire Anemometry. Cambridge University Press, Cambridge. Moore, P. A., and J. Atema. 1991. Spatial information in the three- dimensional tine structure of an aquatic odor plume. Biol. Bull. 181: 408-418. Nakamoto, T., H. Ishida, and T. Moriizumi. 1996a. An odor compass for localizing an odor source. Sens. Actuators B 35: 32-36. Nakamoto, T., T. Yamanaka, H. Ishida, and T. Moriizumi. 1996b. Speed up of odor-source localization system using Kalman-filter sensor signal processing method. [Abstr.] Annual Meeting. San Antonio, TX, Fall 1996. Electrochem. Soc. 96-2: 1163. Pluijm, M. J. F. P., G. J. A. Sars, and C. H. Massen. 1986. Calibration unit for micro-anemometers at very low air velocities. Appl. Sci. Res. 43: 227-234. Roberson, J. A., and C. T. Crowe. 1997. Engineering Fluid Mechanics, 6th ed. John Wiley. New York. Ruzicka. J., and E. H. Hansen. 1988. Flow Injection Anal\sis, 2nd ed. John Wiley. New York. Viollet, S., and N. Franceschini. 1999. Visual servo system based on a biologically-inspired scanning sensor. Pp. 144-155 in Sensor Fusion aiul Deccntrali:cJ Control in Robotic Systems II. G. T. McKee and P. S. Schenker, eds. Prtu: SPIE Int. Soc. Opt. Eng. 3839. Webb, B. 1998. Robots, crickets and ants: models of neural control of chemotaxis and phonotaxis. Neural Nenwrks 11: 1479-1496. Webb, B.. and T. Scutt. 2000. A simple latency-dependent spikmg- neuron model of cricket phonotaxis. Biol. Cybern. 82: 247-269. Webster. D. R.. P. J. W. Roberts, S. Rahman, and L. P. Dasi. 1999. Simultaneous P1V/PLIF measurements of a turbulent chemical plume. [Abstr.] ASCE Int. Water Resources Eng. Conf.. Seattle, WA, August 1999. Am. Soc. Civil Engineers. Weissburg, M. J., and R. K. Zimmer-Faust. 1993. Life and death in moving fluids: hydrodynamic effects on chemosensory-mediated pre- dation. Ecol,>g\ 74: 1428-1443. Weissburg, M. J., and R. K. Zimmer-Faust. 1994. Odor plumes and how blue crabs use them in finding prey. J. Exp. Biol. 197: 349-375. Willis, M. A., and E. A. Arbas. 1991. Odor-modulated upwind flight of the sphinx moth, Mandncu sexta L. J. Comp. Physiol. A 169: 427-440. Reference: Biol. Bull. 200: 227-234. (April 2001) Sensing Scenes With Silicon DAVID C. MOUNTAIN1 * AND ALLYN E. HUBBARD2 1 Boston University Hearing Research Center and Department of Biomedical Engineering. 44 Cuminington St., Boston, Massachusetts 02215; and 2Boston University Hearing Research Center and Department of Electrical and Systems Engineering and Department of Biomedical Engineering, 8 St. Marv's Street, Boston Uni\'ersit\, Boston, Massachusetts 02215 Abstract. Scene analysis, the process of converting sen- sory information from peripheral receptors into a represen- tation of objects in the external world, is central to our human experience of perception. Through our efforts to design systems for object recognition and for robot naviga- tion, we have come to appreciate that a number of common themes apply across the sensory modalities of vision, audi- tion, and olfaction; and many apply across species ranging from invertebrates to mammals. These themes include the need for adaptation in the periphery and trade-offs between selectivity for frequency or molecular structure with reso- lution in time or space. In addition, neural mechanisms involving coincidence detection are found in many different subsystems that appear to implement cross-correlation or autocorrelation computations. Introduction As we walk in a busy city or even a pristine forest, our senses are bombarded by signals from many sources. The acoustic signals entering our ears are a mixture of sounds produced by many sources as well as innumerable echoes. The photons reaching our retina have been reflected off a complicated montage of clothing, faces, automobiles, and buildings or perhaps off a mixture of leaves, stems, insects, birds, soil, and flowers. Likewise, the molecules reaching our olfactory epithelium may be a mixture of burnt hydro- *To whom correspondence should be addressed. E-mail: dcm@bu.edu. This paper was originally presented at a workshop titled Invertebrate Sensory Information Processing: Implications for Biologically Inspired Autonomous Systems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences, Woods Hole. Massachusetts, from 15-17 April 2000, was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. carbons, perfume, and the smell of decaying trash or a combination of fragrances from flowers, musk from ani- mals, and byproducts of the breakdown of leaves. We refer to the problem of interpreting this jumble of sensory input and relating it to the physical world as scene analysis. Many of the current ideas about scene analysis in general started with experimental and theoretical work on vertebrate vision. David Mart (1982) introduced a conceptual frame- work that spanned the entire range of issues from perception down through the physiological mechanisms to the actual underlying computations. The core idea is that sensory systems carry out specific computations that can be de- scribed mathematically, and that if these computations are understood, then they can be implemented as computer programs or in electronic hardware. Our own approach to designing artificial systems for scene analysis follows Marr's lead. We start with physio- logically based models that replicate the responses of the sensory receptors and neural structures that appear to be involved with the early stages of sensory processing. These models are then further abstracted to a form in which they can be used as the starting point for the design of very large-scale integrated circuits (VLSI). The VLSI circuits, after fabrication, are then integrated with appropriate sen- sors, and the outputs are fed to a microprocessor for tasks such as grouping, object localization, and object classifica- tion. Visual Scene Analysis Visual scene analysis in mammals is believed to take place through a series of parallel pathways (Fig. 1 ). The image projected by the lens onto the retina is transduced by photoreceptors, and then contrast is enhanced by neural processing before the visual information is split into spe- cialized pathways that appear to extract important features 227 228 D. C. MOUNTAIN AND A. E. HUBBARD Feature Grouping Transaction and Contrast Enhancement t Light Figure 1. Visual feature analysis consists of first transducing the light projected onto the photoreceptor array and enhancing the contrast of the projected image. This is followed by parallel pathways of feature extrac- tion, the outputs of which are then processed to group related elements to form visual objects. such as distance, orientation, velocity, color, and size (Marr, 1982). Individual regions of the visual image are analyzed for these different features, and then selected portions are grouped together through selective attention to form visual objects that can be identified. Similar processes may also be taking place in inverte- brates. For example, cells from the third optic ganglion of dragonflies respond selectively to different target classes with properties that are remarkably similar to those of cells from the mammalian visual cortex (O'Carroll, 1993). Also, bees — like mammals — can recognize a familiar shape un- der a variety of viewing conditions regardless of whether it is initially sensed by color contrast, luminance contrast, or motion contrast (Zhang ct ai, 1995). The visual system must be able to cope with the large changes in ambient light level that take place due to time of day, presence or absence of clouds, and moving in and out of the shade. Even with fixed lighting conditions, some parts of the visual scene may be brightly lit while others may be in the shade. The image projected onto a receptor array is the product of the illumination falling upon the objects within the visual scene, multiplied by the reflectivities of these objects. Since it is the reflectivity (both overall mag- nitude and spectrum) that provides the useful information about object identity, the visual system needs a method to minimize the effects of varying illumination. These illumination problems must be dealt with in the first stages of processing, before object formation can take place. The large changes in ambient light level appear to be handled at the receptor level through adaptation. Adaptation is a process whereby the sensitivity of the photoreceptor depends on the time-averaged light level. In biological photoreceptors, biochemical processes provide the needed automatic gain control. The outputs of small groups of photoreceptors are then combined so as to enhance the differences in reflectivity of objects within the scene by using a "center-surround" organization (Fig. 2, column I). This is done by combining an excitatory input from a receptor or small cluster of receptors with inhibitory inputs from the surrounding neighbors (on-center receptive field) or by combining an inhibitory input from a receptor or cluster with excitatory inputs from the surrounding neigh- bors (off-center receptive field). Mathematically, the com- bination of adaptation and center-surround organization is equivalent to performing the combination of local normal- ization and a two-dimensional second spatial derivative on the output of the receptor array. This process has the effect of emphasizing contrast boundaries in the image. The spa- tial extent of the receptors contributing to the receptive field can be varied at the design stage to achieve different degrees of resolution (image smoothing). Alternatively, the scene can be processed by parallel pathways each with a different resolution. If appropriate weights are used for the excitation and inhibition, then the center-surround spatial filters can be approximated mathematically as Gabor functions (Weldon and Higgins. 1999). The multi-resolution approach can be thought of as taking a two-dimensional wavelet transform of the image (Porat and Zeevi, 1989). Distance information is not available to the visual system directly, because the external three-dimensional world is mapped onto a two-dimensional array of receptors. If a three-dimensional internal representation is needed, say for navigational purposes, then the third dimension must be synthesized from the information available from the recep- tors. If the system has two eyes with overlapping visual fields, then differences due to parallax between the images from the two eyes can be exploited (binocular disparity) to estimate distance; otherwise, vergence or more subtle cues must be used. To estimate binocular disparity, the visual system appears to perform a spatial cross-correlation be- tween corresponding regions of the two retinas (Marr. 1982). Spatial cross-correlation is also used to detect motion. SENSING SCENES WITH SILICON 2_ II III Off-Cell N Figure 2. Orientation processing consists of combining the outputs of cells with center-surround organiza- tion (column I) to create oriented receptive fields (column II). These oriented receptive fields are then combined to form oriented edge detectors (column III). Coincidence detection between the output of a cell and the delayed outputs of other cells with nearby receptive fields is mathematically equivalent to computing the spatial cross- correlation between the current visual frame and a previous visual frame on a region-by-region basis. Orientation processing involves detecting lines and edges and estimating their angular orientation. Hubel and Wiesel (1962), working with cat visual cortex, showed that detec- tion of oriented edges can be accomplished by a sequence of processing stages that combine the outputs of groups of cells with similar center-surround characteristics. By using groups of cells arranged as short linear arrays, short linear segments of light or dark can be detected (Fig. 2, column II). Different arrays have different orientations (orientation tun- ing), so that all possible edge segments within a region can be detected. If we then combine the output of pairs of these arrays that are slightly offset from each other and have the same orientation but with one array being of the "on" type and the other being of the "off type, we have a system that detects edge segments between areas of different reflectivi- ties (Fig. 2, column III). This process can be performed a second time to detect line segments. Higher-level process- ing can then be used to group the edge or line segments into longer lines and arcs (Pasupathy and Connor, 1999). We have implemented this type of processing in silicon by designing a set of integrated circuits that implement the processing illustrated in Figure 2 (Hinck and Hubbard, 1999). We do not have space here to go into the details of the silicon implementation, but one significant difference between the biological and silicon system must be men- tioned. In biological systems, the information between pro- cessing units (cells) is carried by axons that are self routing; in other words, they can work their way through the nervous tissue and find their targets. With silicon processing sys- tems, the wiring problem becomes serious. The processing described within a single column of Figure 2 only requires communication between nearby elements on the chip. How- ever, when we need to move information from one process- ing level or chip to another (from one column to another in Fig. 2), then we run into problems due to the sheer number of wires involved. To reduce this bottleneck, a technique known as address event representation (AER) is used (Boa- hen, 2000). When a silicon cell is "excited," it broadcasts its address (identity) to all listeners, which may be a one-to-one or a one-to-many mapping. Each broadcast event is equiv- alent to the production of a single action potential (spike) in the biological system, and given the bandwidth (speed) of the circuitry we have the ability to transmit the identity of all the spikes from all the cells on a chip. Because the processing is taking place in real time, there is no need to record a time stamp for the events. For simulations that do not run in real time, each event may need both a time stamp and an address. With AER, signaling takes place only if a spike is gen- erated; this minimizes power consumption because, for a single cell, spikes are relatively rare events. This minimi- zation of power consumption is important, especially for small robots (as well as for biological systems), since low power consumption allows operation for longer periods of time without replenishment of energy stores. 230 D. C. MOUNTAIN AND A. E. HUBBARD Auditory Scene Analysis A major challenge in auditory scene analysis is that acoustic signals from different sources can overlap in direc- tion, frequency, and time. We believe that biological sys- tems meet this challenge by dividing up the received signals in frequency and time and — through the use of appropriate grouping principles — enhance the signal-to-noise ratio for individual sources to the point where the bearing and iden- tification of the source can be determined. In many appli- cations, both transient and long-duration signals are of in- terest. In auditory scene analysis, each frequency band can be analyzed for the presence of specific features, and then the grouping rules can be used to combine information from selected frequency bands to produce the features vector that represents an auditory object. Audition, unlike vision, has no method by which even two of the three physical dimensions of the external acoustic world can be projected directly onto the receptor array. To determine the direction of a sound source, one either needs to compare signals acquired by directional ears (micro- phones) with different orientations or compare measure- ments of pressure taken at different locations in space. In the latter case, the ears or microphones must be spaced suffi- ciently that the time delay due to the speed of sound is large enough to be sensed or measured. If only two ears or microphones are used, then directional ambiguities are present, but these can generally be resolved through rotation of the head or microphone array. The third dimension (source distance) is much more difficult to estimate in audition. Experiments with human listeners suggest that the ratio of direct to reverberant sound energy may be an important distance cue. How this ratio might be estimated is not clear. Each frequency channel is analyzed in parallel through the computation of multiple features (Fig. 3). These features are likely to be similar for frequency channels that contain signals from the same sound source and are likely to differ for signals from different sound sources. For example, the differences in time delay between the arrival of the signals (interaural time differences, ITD) as well as differences in intensity (interaural intensity differences, IID) at two sen- sors will be similar across frequency channels for a single source because these features depend on source direction. Frequency components with similar onsets, offsets, dura- tion, and envelope period are also most likely to be from a single sound source. For many vertebrates, the head size is sufficient to create significant time delays (ITD) between the ears that can be used for localization; at higher frequencies the head shadow effect is large, producing a significant IID. For very small animals, especially insects, the ears are very close together, making ITD estimation via neural circuits impractical, and the animal's size precludes creating a sound shadow. These Feature Grouping Frequency Filtering" and Transduction t Sound Figure 3. Auditory feature analysis consists of first filtering and trans- ducing the sound received by the peripheral organs. This is followed by parallel pathways of feature extraction, the outputs of which are then processed to group related elements to form auditory objects. animals appear to use mechanical or acoustic means, or both, to detect the subtle pressure differences between the two sides of their body (Michelsen, 1998). As was the case for visual processing, the final step before auditory source identification is the grouping process (Bregman, 1990). In each of the features maps described above, timing information is preserved. This enables the grouping process to use common bearing, as determined by the ITD and IID maps, and synchrony across maps as the major cues for grouping specific components together. This grouping process results in a simplified set of features that includes target direction, the major peaks in the target signal spectrum, and temporal features such as the period of the signal envelope. This set of features can then be compared to stored signatures to complete the identification process. Signatures in this context can be hardwired (acquired through evolution at the species level), learned through experience at the individual level, or derived from a com- bination of the two methods. If the system is hardwired, then it is possible to imple- SENSING SCENES WITH SILICON 2 ment the entire analysis/tracking system with simple cir- cuits. For example, the Webb and Scutt (2000) model of cricket phonotaxis implements pattern recognition and source localization with a system comprising two receptors followed by four neurons. The pattern of interest in this case is the mating call of the male, which is characterized by a limited range of carrier frequencies and a limited range of syllable repetition intervals (SRI) (modulation periods). Fil- tering for the appropriate carrier frequencies takes place in the hearing organ, and subsequent filtering for SRI takes place using a pair (one for each ear) of output neurons that act as lowpass filters, followed by another pair of neurons that act as a highpass filters. Source localization is accom- plished by using directional ears and a combination of excitation and inhibition in the same neurons that perform the highpass filtering. For auditory scene analysis, it is essential that the filters that perform the frequency separation be designed to have impulse responses that are compact both in frequency and time. The performance measure commonly used to describe this feature is the time-bandwidth product. Simple, single mode resonances, although narrow in frequency, do not have good temporal performance and hence do not have good time-bandwidth products. The impulse response that achieves the theoretical time-bandwidth product limit is a sinusoid with a Gaussian envelope (Gabor function). Such an impulse response is physically unrealizable, but it is possible to combine multiple resonances to create a re- sponse that comes close to the ideal. Also, for a general purpose signal processing system, it is generally better to use filters with a constant ratio of bandwidth to center frequency (constant Q) rather than a constant bandwidth like that obtained with a Fourier transform. The widespread use of approximately constant-Q filtering across the ears of many species ranging from bush crickets (Hoy, 1992) to mammals (Javel, 1986) suggests that this approach offers significant survival value. The use of a constant-Q filter bank is very similar mathematically to taking a wavelet transform of the acoustic time signal. It should be noted that most of the acoustic frequencies of biological significance are higher than what most cells can follow, so the filtering is generally done mechanically before detection by the receptor cells. The number of frequency channels may vary from very few in insects (Michelsen, 1992) to hundreds in many vertebrates (Echteler el al.. 1994). Typically this filtering process is implemented in silicon using a cascade of second-order filters with progressively lower resonant frequencies. This cascade is intended to simulate the traveling wave of the mammalian cochlea, which starts in the basal (high-frequency) end of the cochlea and propagates towards the apical (low-frequency) end. For this purpose, subthreshold circuits have been most com- monly used (Mead, 1989: Fragniere el uL. 1997; Sarpeshkar et al.. 1998). Like the visual system, the auditory system must also deal with a wide range of signal levels. Here again, adaptation (automatic gain control) plays an important role. In mam- malian auditory systems the adaptation is specific to each frequency channel (Javel, 1986). In insects, responses of neurons in the central nervous system can also exhibit adaptation (e.g.. see Lewis, 1992). Unlike the visual system, however, the auditory system is processing a very rapidly changing signal, one that often changes much faster than the biological hardware can fol- low. To circumvent the problem of following high-fre- quency signals, the receptor cells (hair cells) act as soft half-wave rectifiers (Mountain and Hubbard, 1996) so that at high frequencies they respond to the envelope of the acoustic signal rather than to the fine structure of the signal. In the auditory system, temporal cross-correlation and autocorrelation-like processing is believed to play an im- portant role (Colburn, 1996; Lyon and Shamma, 1996). In vertebrates, the time delay between the two ears (IID) is an important cue for localization. The combination of neural delay lines and coincidence detection is used to cross- correlate the signals from the two ears for each frequency channel. Periodicity analysis is believed to take place also using delays and coincidence detection. Periodicity analysis no doubt plays an important role for many species from insects to man, because so many communication sounds involve periodic amplitude modulation (AM). Figure 4 il- lustrates time waveforms in which AM is a prominent feature for a cricket call (panel A) and for a human vowel (panel C). Panels B and D show the results of spectral analysis using a constant-Q filter bank, and except for center frequency and modulation rate, the AM signals are remark- ably similar. Olfactory Scene Analysis By analogy to the visual and auditory systems, we refer to the problem of identifying and localizing odor sources in complex environments as olfactoiy scene analysis. Unlike vision and hearing, in which the signal propagates via wave phenomena, olfaction is characterized by mass transport by currents in water or air and the associated turbulence found in these media (Grasso, 2001). No direct information about source location is present in the received signal, but approx- imate direction can be estimated by sensing wind or water- flow direction. The only way a source can be located with any certainty is to trace the odor plume back to its source. In general, individual odor sources release mixtures of compounds into the environment, and the signal at the sensory organ is the result of the mixing of turbulent plumes from multiple sources. Due to the nature of turbulent trans- port, the plume produced by a single odor source is made up of a series of patches or filaments distributed within the plume; these move past the olfactory organ, creating a series 232 D. C. MOUNTAIN AND A. E. HUBBARD Cricket Q) +1 1— 3 (fl „ o o If Y f "lf';*|f N o c 0> 3 CT 0 4.0 0.1 B 50 100 150 200 250 Speech 60 80 100 Time (ms) Figure 4. Spectral analysis of animal communication sounds. Time waveforms for a cricket call (panel A) and for a human vowel (panel C) are plotted along with the results of spectral analysis using a constant-Q filter hank (panels B and D). of odor pulses at the receptors with random arrival times, durations, and amplitudes (Moore and Atema, 1991). The patchy nature of odor concentration signals can be seen in the two concentration signals shown in Figure 5. In a multi-source environment, the odor pulses from one source will be intermixed with pulses from other sources. In such an environment, the average concentration of a compound is not a useful feature for olfactory scene analysis. Even if only one odor source is present, the statistical nature of the plume is such that several minutes of signal averaging are necessary to get an accurate estimate of average concentra- tion. However, behavioral experiments in plumes of this sort indicate that animals make olfactory decisions on the order of a few seconds (Basil and Atema. 1994). Like the visual and auditory systems, the olfactory sys- tem must be able to cope with wide ranges in signal (con- centration) level. Olfactory receptors, like their counterparts in the other sensory systems, also exhibit adaptation that adjusts the sensitivity of individual receptors on the basis of background concentration levels. Olfactory systems have many different receptor types, ranging from a few dozen in insects to approximately 1000 receptor types in mammals. Some receptors, mainly those that have evolved to detect pheromones, are extremely selective, but most will respond to a number of different compounds. The higher the selec- tivity of a receptor, the higher the affinity for the odor molecule and the slower the release of the odor molecule after it has bound to the receptor (Lauffenburger and Lin- derman, 1993). The relationship between high affinity and slow release comes about because affinity depends on the ratio of the binding to unbinding rates. The rate of binding is limited by the rate at which the odorant can access the binding site, a rate that is similar for all receptors. Affinity, therefore, varies from receptor to receptor, largely due to differences in the unbinding rate. This relationship means C o "oo -I—* c / Auditory Research. H. L. Hawkins. T. A. McMullen. A. N. Popper, and R. R. Fay, eds. Springer Verlag, New York. 234 D. C. MOUNTAIN AND A. E. HUBBARD Marr, D. 1982. Vision: A Computational Investigation into the Human Representation and Processing of Visual Information. W. H. Freeman, San Francisco. Mead, C. 1989. Analog VLSI and Neural Systems. Addison Wesley, Reading, MA. Michelsen, A. 1992. Hearing and sound communication in small ani- mals: evolutionary adaptations to the laws of physics. Pp. 61-77 in The Evolutionary Biology of Hearing. D. B. Webster, R. R. Fay. and A. N. Popper, eds. Springer Verlag. New York. Michelsen, A. 1998. Biophysics of sound localization in insects. Pp. 18-62 in Comparative Hearing: Insects, Springer Handbook of Audi- tory Research, R. R. Hoy, A. N. Popper, and R. R. Fay. eds. Springer Verlag, New York. 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. Mountain, D. C., and A. E. Hubbard. 1996. Computational analysis of hair cell and auditory nerve processes. Pp. 121-156 in Auditory Com- putation, Springer Handbook of Auditory Research, H. L. Hawkins, T. A. McMullen, A. N. Popper, and R. R. Fay, eds. Springer Verlag, New York. O'Carroll, D. 1993. Feature-detecting neurons in dragonflies. Nature 362: 541-543. Pasupathy, A., and C. E. Connor. 1999. Responses to contour features in macaque area V<1. J. Neiirophysiol. 82: 2490-2502. Porat, M., and Y. Y. Zeevi. 1989. Localized texture processing in vision: analysis and synthesis in the Gaborian space. IEEE Trans Biomed. Eng. 36: 115-129. Sarpeshkar, R., R. F. Lyon, and C. Mead. 1998. A low-power wide- dynamic-range analog VLSI cochlea. Analog Integrated Circuits Sig- nal Processing 16: 245-274. Webb, B., and T. Scutt. 2000. A simple latency-dependent spiking- neuron model of cricket phonotaxis. Biol. C\bern. 82: 247-269. Weldon, T., and W. E. Higgins. 1999. Designing multiple Gabor filters for multitexture image segmentation. Optical Eng. 38: 1478-1489. Zhang, S. W., M. V. Srinivasan, and T. Collett. 1995. Convergent processing in honeybee vision: multiple channels tor the recognition of shape. Proc. Natl. Acad. Sci. USA 92: 3029-3031. Reference: Binl. Bull. 200: 235-242. (April 2001) Sensory Architectures for Biologically Inspired Autonomous Robotics CHARLES M. HIGGINS Neuromorphic Vision and Robotic Systems Laboratory, Department of Electrical and Computer Engineering. The University of Arizona, 1230 E. Speedwav Blvd.. Tucson. Arizona 85721 Abstract. Engineers have a lot to gain from studying biology. The study of biological neural systems alone pro- vides numerous examples of computational systems that are far more complex than any man-made system and perform real-time sensory and motor tasks in a manner that humbles the most advanced artificial systems. Despite the evolution- ary genesis of these systems and the vast apparent differ- ences between species, there are common design strategies employed by biological systems that span taxa, and engi- neers would do well to emulate these strategies. However, biologically-inspired computational architectures, which are continuous-time and parallel in nature, do not map well onto conventional processors, which are discrete-time and serial in operation. Rather, an implementation technology that is capable of directly realizing the layered parallel structure and nonlinear elements employed by neurobiology is re- quired for power- and space-efficient implementation. Cus- tom neuromorphic hardware meets these criteria and yields low-power dedicated sensory systems that are small, light, and ideal for autonomous robot applications. As examples of how this technology is applied, this article describes both a low-level neuromorphic hardware emulation of an ele- mentary visual motion detector, and a large-scale, system- level spatial motion integration system. E-mail: higgins@ece.arizona.edu This paper was originally presented at a workshop titled Invertebrate Sensory Information Processing: Implications for Biologically Inspired Autonomous Systems. The workshop, which was held at the J. Erik Jonsson Center for the National Academy of Sciences. Woods Hole, Massachusetts, from 15-17 April 2000, was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-896. Engineers Have a Lot to Gain From Studying Biology The architecture of biological sensory systems has some- thing to teach conventional engineering about system-level design. Neurobiological representations of visual modalities including depth, motion, color, and form are quite unlike those employed by conventional computer vision systems. Neurobiological computational architectures are not modu- lar, linear, or feedforward. And yet biological organisms routinely accomplish complex visual tasks such as object recognition, obstacle avoidance, and target tracking, which continue to challenge artificial systems. Dealing with the complexity posed by the availability of a multitude of parallel sensors and a large number of re- dundant actuators is a major unsolved problem in modern computational systems. As we become more ambitious in our inclusion of computational systems into every possible device, more sensory inputs are available: more actuators are controlled by the computing system; more possible conditions must be recognized and appropriately dealt with. Engineers in fields from space-bound robotics and guided missile design to automotive navigation and prosthetics are beginning to realize the value of paying attention to biolog- ical solutions. The biological literature is replete wiih de- tailed analyses of systems that deal with complex sensory and motor control tasks and still manage to outperform, by any metric, anything created by mankind. In the papers in this collection (CASSLS. 2001). for example, Robert Bar- low and colleagues describe a finely elaborated model of horseshoe crab visual processing, revealing how the crab manages to navigate and find mates in a highly variable and noisy underwater environment (Barlow et al, 2001); Thomas Cronin and Justin Marshall give us insight into the marvelously complex mantis shrimp visual system, a 235 236 C. M. HIGGINS unique and highly specialized tool for predation (Cronin and Marshall, 2001); Mandyam Srinivasan and co-work- ers introduce us to the secrets of insect vision (Srinivasan et til.. 2001). The daunting task for engineers is to produce — by ab- straction from the specific organism — a set of general de- sign principles to replace the top-down, linear, functional- block oriented strategies that characterize conventional design. These principles may include layered massively parallel processing, population coding, probabilistic infor- mation transmission, a mix of analog and digital coding strategies, connection-oriented architectures, removal of re- dundant information at every processing stage, positive as well as negative feedback, and the use of nonlinear compu- tational elements (such as voltage-dependent conductances and neurons with adaptive properties), even though these elements make the system more difficult to describe analyt- ically in closed form. Although the mathematical principles underlying conventional design are well known, it is not clear that we have the mathematical tools to analyze and design systems such as those suggested by neurobiology. However, these tools are being developed in the theory of nonlinear dynamical systems, as well as in the study of neurobiological systems themselves. By building artificial sensory systems true to the repre- sentations and computational architectures used by neuro- biology, we will be able to produce novel, highly capable autonomous robots, while at the same time suggesting test- able hypotheses as to how biological systems accomplish sensory and motor tasks. vices, highly efficient computations can be performed on signals represented by currents and voltages. Neuromorphic technology focuses on analog computation for efficiency, but in general mixes both analog and digital transistor circuitry. The implementation can be at a level of abstrac- tion from the biophysical (Hahnloser et al, 2000) to the neural (Boahen. 1999) to the mathematical (Higgins and Korrapati, 2000). The individual physical components in such a design are on the scale of micrometers and are fabricated together on a single piece of silicon measuring millimeters on a side, resulting in a physically robust com- pact package. While general-purpose digital processors grow ever more powerful, special-purpose hardware will always be more efficient in power consumption and size for specific prob- lems. This is because the former, by its nature, is not optimized for any specific computation, whereas the latter includes only the circuitry that is necessary for the problem at hand. The efficiency of a task-specific analog implemen- tation comes at the price of precision and flexibility. The analog implementation of a computation is subject to inev- itable noise that limits the possible precision, whereas a digital implementation can (potentially but usually not prac- tically) be performed with virtually unlimited precision. The software implementation of a computation clearly has flex- ibility to change at the programmer's whim, whereas a dedicated hardware implementation is fixed at fabrication time. With awareness of these trade-offs, a designer will choose a special-purpose hardware solution where the power and size constraints are most important. This is often an appropriate choice in real-time vision applications on autonomous robots. Efficient Implementation of Biological Computational Architectures Because the primary computational strategy used by neu- robiology is layered, massively parallel processing, an effi- cient parallel implementation is essential to make effective use of such computational architectures. Biological compu- tational architectures, which are continuous-time parallel algorithms, map very badly onto conventional discrete-time serial processors. The conventional real-time implementa- tion of a biological vision computation involves an imager providing discrete-time frames to a high-speed signal pro- cessor that serially performs a large number of operations that could be performed in parallel. In contrast, true parallel architectures may be straightforwardly implemented in neu- romorphic VLSI (very large scale integration) hardware (Mead, 1989). This type of special-purpose hardware im- plements biological algorithms directly and efficiently in patterns of transistors, resistors, and capacitors. By making use of the physical primitives provided by electronic de- An Example of Neuromorphic Implementation: Low-Level Motion Detection The neurobiological representation of visual motion in organisms from insects to primates is in terms of oriented spatiotemporal-frequency-tuned cells. The conventional op- tical flow representation is mathematically a vector field, allowing only one velocity of motion in every local area of the visual scene. The biological representation of visual motion is more powerful in that it allows multiple simulta- neous directions of motion as long as they are distinguished in spatial or temporal frequency, making possible our per- ception of transparent motion and occlusion boundaries. However, the biological representation of motion requires a range of spatial- and temporal-frequency tunings to cover all stimuli of interest. Primates have a tremendous number of neurons (on the order of 10") and thus can afford to populate the whole spatiotemporal frequency space with a BIOLOGICALLY INSPIRED ROBOTICS 237 large bank of tuned neurons for each spatial location. In- sects, with a much smaller set of neurons (on the order of 105), are forced to use a much smaller set of tunings, yet they operate very effectively in the same environment. Thus insects may provide engineers with a more neuron-efficient example of a visual motion system. In a seminal 1985 paper, Adelson and Bergen proposed a model of the motion response of primate complex cells and showed it to be equivalent to the Reichardt (1961) model of elementary motion detection in the fly. We have imple- mented this motion energy model in neuromorphic VLSI hardware (Higgins and Korrapati, 2000), both to allow system-level hardware modeling of biological motion sys- tems, addressed in the next section, and to provide a sensi- tive, real-time visual motion detector for robotic applica- tions. Because of the lack of any digital thresholding step in the algorithm, motion energy sensors (like the motion- sensitive cells they model) are limited in contrast response only by signal-to-noise ratio, allowing them to perceive the direction of motion even for very low contrasts by integrat- ing over time. The original algorithm, shown in Figure 1A, combines the response of four simple cells with a rectifying nonlinearity (a square) to create a spatiotemporal-fre- quency-tuned estimate of motion direction. Figure IB shows how this model was implemented in hardware: sev- eral simplifications were made in the pursuit of a compact implementation, but the algorithm still performs a highly sensitive directional motion computation, as shown in Fig- ure 2. This sensor was implemented with only 41 transis- tors, allowing hundreds or thousands of parallel sensors on a reasonably sized silicon substrate. Because of the analog implementation, each sensor consumes less than 40 micro- watts of power, allowing large-scale parallel implementa- tion without problematic power consumption or special cooling needs. System-Level Neuromorphic Design: Spatial Motion Integration As mentioned earlier, the spatiotemporal filter represen- tation of motion common to a wide variety of organisms requires a range of spatial- and temporal-frequency tunings to cover all stimuli of interest and thus, unlike the conven- tional representation, must be implemented with a bank of filters for each spatial location. Once this bank of filters is computed, the information must be spatially integrated to be of use to the organism; in other words, information about spatial patterns of visual motion is more useful to the animal than the low-level motion information itself. This integration occurs in insect tangential neurons (Krapp and Hengstenberg, 1996) and in the medial superior temporal area (MST) of primate visual cortex (Perrone and Stone, 1998). This strategy suggests a computational architecture like that shown in Figure 3A, which includes a first-stage photosensitive array, a second stage of multiple parallel motion-processor arrays computing different spatial fre- quencies, orientations, and temporal frequencies, and a third integration stage that can synthesize units sensitive to wide- field spatial patterns of motion by combining spatial regions of low-level motion detectors. We have implemented a preliminary version of such an architecture in neuromorphic VLSI hardware (Higgins and Shams, unpubl. data), as shown in Figure 3B. Each two- dimensional processing unit is implemented as a separate VLSI chip; chips communicate using trains of spikes mod- eling action potentials. This multi-chip neuromorphic VLSI system is capable of synthesizing units sensitive to complex arbitrary patterns of visual motion, including expansion, contraction, rotation, and translation, as shown in Figure 4. These patterns, along with more elaborate ones, allow de- termination of imager self-motion through the world for postural stability and navigation (Zemel and Sejnowski, 1998), tracking of objects (Gronenberg and Strausfeld, 1991), and obstacle avoidance (Gabbiani et al. 1999). De- spite the multi-chip nature of the system, it is still a com- pact, robust, low-power special-purpose computing system ideal for robotic applications. Summary This article contends that biological representations and computational architectures are useful in a wide range of engineering problems. However, in practice, it is difficult to separate biological algorithms from their implementation, and so we also present an implementation technology, neu- romorphic VLSI, which is well suited for physical realiza- tion of these algorithms. A low-level motion processor based on the Adelson-Bergen model was described to illus- trate how a biological model may be directly mapped onto an analog VLSI transistor circuit for highly efficient imple- mentation. From an engineer's perspective, this sensor ben- efits not only from the biological representation, but also from the analog nature of the biological model computation. A multi-stage motion-processing system was also de- scribed, exemplifying a biologically inspired computing architecture that also uses representations inspired by biol- ogy. This architecture is very well matched to the vision problem and allows a specific useful computation to be done with very low requirements for power consumption and size. It is the author's contention that systems of this type are the best hope for future highly capable small autono- mous robotic systems. 238 C. M. HIGGINS /\ Quadrature pair of spatial filters Quadrature pairs of temporal filters Combine filters nonlinearly Motion energy output Approximate quadrature spatial pair Temporal lowpass filtering Combine filters with nonlinearity Motion energy output Figure 1. Low-level motion detection. (A) The canonical Adelson-Bergen motion energy model combines each of an even and odd spatial Gahor tiller with two temporal bandpass filters to create me output of four linear simple cells. The output of these simple cells is combined nonlinearly in quadrature pairs to create a phase-independent motion direction estimate, which models a primate complex cell. (Bl For hardware imple- mentation, the spatial filters were implemented with a center-surround diffusive network, and the temporal filters by using a single lowpass filter. Rather than using a squaring operation, a simpler rectifying nonlinearity (the absolute value) was employed. The output of the implementation is quite comparable to the original model. BIOLOGICALLY INSPIRED ROBOTICS 239 -200 - 150 - 100 -50 -100 -150 -0.4 -0.3 -0.2 -0.1 0 0.1 Spatial Frequency (cycles/pixel) 0.3 Figure 2. Output of the low-level motion detector. (A) The output of the hardware motion sensm i .'< >v.n in response to no stimulus (under fluorescent lighting), an orthogonal, preferred and null direction sinusoid grating. The lighter traces represent raw output: darker lines are averaged output. (B) As the spatial and temporal frequency of a moving sinusoidal stimulus is varied, the sensor shows a strong preference in a certain hand of frequencies. 240 C. M. HIGGINS Visuid scene FIRST STAGE B FIRST STAGE SECOND STAGE THIRD STAGE SECOND STAGE THIRD STAGE Figure 3. Architecture for spatial motion integration. (A) In the idealized system, a first-stage silicon retina is followed by multiple second-stage motion processors tuned for different orientations and spatial and temporal frequencies. Interaction is allowed between second-stage motion processors. The final stage integrates over space to synthesize sensitivity to patterns of visual motion. (B) In the current system, the silicon retina detects moving edges and transmits this information to the next stage. The second-stage motion processors use the moving edge information to compute the direction of edge motion, with each of four processors tuned for a different direction of motion. The final stage synthesizes spatial motion sensitivity. BIOLOGICALLY INSPIRED ROBOTICS Contraction-sensitive unit Expansion-sensitive unit 24 i 0 degree translation-sensitive unit 180 degree translation-sensitive unit 90 degree translation-sensitive unit 270 degree translation-sensitive unit 50 -50 50 Figure 4. Performance of the spatial motion integration system. This figure shows the response of eight simultaneously synthesized units to a wide-field stimulus — a periodically contracting circle — as the focus of contraction is moved around the visual field. Lighter shading indicate stronger responses. The upper left unit is tuned for contraction at the center of the visual field and thus shows a strong response there. The upper right unit is tuned for expansion and thus shows no response. In the second row from the top. units tuned for clockwise and counterclockwise rotation likewise show little or no response. However, units shown at the bottom tuned tor translation in four different directions show a strong response when the focus of contraction nears the appropriate edge of the visual field, because the farther the contracting pattern gets from the center, the more similar it becomes to a simple translating pattern. 242 C. M. HIGGINS Acknowledgments The author gratefully acknowledges support for this re- search by the National Science Foundation through the California Institute of Technology Center for Neuromorphic Systems Engineering as part of the Engineering Research Centers program, by the Office of Naval Research under grant number NOOO 14-97- 1-0970, and by the National Sci- ence Foundation's State/Industry/University Cooperative Research Center for Low Power Electronics (CLPE). CLPE is supported by NSF (Grant #EEC-9523338), the State of Arizona, and the following companies and foundations: Burr-Brown, Inc., Conexant, Gain Technology, Intel Cor- poration. Medtronic Microelectronics Center, Microchip Technology, Motorola, Inc., The Motorola Foundation, Raytheon. Texas Instruments, and Western Design Center. The author also thanks the anonymous reviewers for their assistance in clarifying this manuscript. Literature Cited Adelson, E. H.. and J. R. Bergen. 1985. Sputiotemporal energy models for the perception of motion. J. Opt. Soc. Am. A 2(2): 284-299. Barlow, R. B.. J. M. Hitt, and F. Dodge. 2(101. Linniliis vision in the marine environment. Biol. Bull. 200: 169-176. Boahen. K. 1999. Retinomorphic chips that see quadruple images. In Proceedings of the 7th International Conference on Microelectronics for Neural, Fuzzy and Bio-inspired Systems. University of Granada. Spam. 7-9 April 1999. CASSLS (The Center for Advanced Studies in the Space Life Sci- ences). 2001. Invertebrate sensory information processing: implica- tions for biologically inspired autonomous systems. Biol. Bull. 200: 145-245. Cronin, T., and J. Marshall. 2001. Parallel processing and image anal- ysis in the eyes of mantis shrimps. Biol. Bull. 200: 177-183. (iahbiani. F., H. Krapp, and G. Laurent. 1999. Computation of object approach by a wide-field, visual neuron. J. Neurosci. 19: 1122-1141. Gronenberg, W., and N. J. Strausfeld. 1991. Descending pathways connecting the male-specific visual system of flies to the neck and flight motor. J. Comp. Physiol. A 169: 413-426. Hahnloser, R. H. R., R. Sarpeshkar, M. A. Mahowald, R. J. Douglas, and H. S. Seung. 2000. Digital selection and analogue amplification coexist in a cortex-inspired silicon circuit. Nature 405: 947-951. Higgins, C. M., and S. K. Korrapati. 2000. An analog VLSI motion energy sensor based on the Adelson-Bergen algorithm. In Proceedings of the Internationa/ ICSC Symposium on Biologically-Inspired Sys- tems. Wollongong. Australia. 12-15 December 2000. Krapp, H. G., and R. Hengstenberg. 1996. Estimation of self-motion by optic flow processing in single visual intemeurons. Nature 384: 463-466. Mead. C. A. 1989. Analog VLSI and Neural Systems. Addison-Wesley. Reading, MA. Perrone, J. A., and L. S. Stone. 1998. Emulating the visual receptive- field properties of MST neurons with a template model of heading estimation. / Neurosci. 18(15): 5958-5975. Reichardt, W. 1961. Autocorrelation, a principle for the evaluation of sensory information by the central nervous system. Pp. 303-317 in Sensory Communication, W. A. Rosenbhth. ed. MIT Press. Cambridge. MA. Srinivasan, M.. S. Zhang, and J. S. Chahl. 2001. Landing strategies in honeybees, and possible applications to autonomous airborne vehicles Biol. Bull. 200: 216-221. Zeniel, R. S., and T. J. Sejnowski. 1998. A model for encoding multiple object motions and self-motion in area MST of primate visual cortex. J. Neurosci. 18(1): 531-547. Reference: Biol. Bull. 200: 243-24?. (April 2001) Participants BARRY W. ACHE University of Florida Whitney Laboratory 9505 Ocean Shore Blvd. St. Augustine. FL 32086 RAYMOND ARVIDSON Washington University in St. Louis McDonnell Hall One Brookings Drive, Box 1169 St. Louis, MO 63130 JELLE ATEMA Boston University Marine Program 7 MBL Street Woods Hole, MA 02543 JOSEPH AYERS Northeastern University Marine Science Center Nahant. MA 01908 ROBERT BARLOW SUNY Upstate Medical University at Syracuse Center for Vision Research 750 E. Adams Street, 3258 Weiskotten Hall Syracuse, NY 13210 JENNIFER BASIL Brooklyn College Department of Biology 2900 Bedford Avenue Brooklyn, NY 11210 JIM BELANGER Tufts University Department of Biology Dana Lab Medford, MA 02155 JOHN BIRMINGHAM Brandeis University Volen Center for Complex Systems MS-013 Waltham, MA 02454 DIANA BLAZIS Center for Advanced Studies in the Space Life Sciences 7 MBL Street Woods Hole, MA 02543 THOMAS BREITHAUPT University of Konstanz Faculty of Biology D-78457 Konstanz, Germany CHUAN-CHIN CHIAO University of Maryland, Baltimore County Department of Biological Sciences Baltimore, MD 21250 THOMAS CONSI Massachusetts Institute of Technology Department of Ocean Engineering 77 Massachusetts Avenue, Room 5-221 Cambridge, MA 02139 THOMAS CRONIN University of Maryland. Baltimore County Department of Biological Sciences 100 Hilltop Circle Baltimore, MD 21250 JONATHAN DALE Boston University Marine Program Marine Biological Laboratory 7 MBL Street Woods Hole, MA 02543 CHARLES DERBY Georgia State University Biology Department P.O. Box 4010 Atlanta, GA 30302-4010 FREDERICK DODGE SUNY Upstate Medical University at Syracuse Center for Vision Research Department of Physiology, 3260 Weiskotten Hall Syracuse, NY 13210 243 244 PARTICIPANTS NICOLAS FRANCESCHINI CNRS / LNB3 Neurocybernetics Research Group 31, Chemin J. Aiguier J. Aiguier Avenue 31 Marseille, France 13009 PETER FRASER University of Aberdeen, UK Department of Zoology Tillydrone Avenue Aberdeen, AB24 2TZ United Kingdom CATHERINE GOLDEN Fundamental Biology Research Program NASA Ames Research Center MS 19-20 Moffett Field, CA 94035 FRANK GRASSO Visiting Assistant Professor Department of Psychology Brooklyn College Brooklyn, NY 11210 ROGER HANLON Marine Biological Laboratory Marine Resources Center (MRC) 7 MBL Street Woods Hole, MA 02543 ADAM HAYES California Institute of Technology (136-93) Microsystems Lab Pasadena, CA 91125 CHARLES HIGGINS University of Arizona Department of Electrical & Computer Engineering 1230 E. Speedway Blvd. Tucson, AZ 85721-0104 JOHN HINES Fundamental Biology Research Program NASA Ames Research Center MS 19-20 Moffett Field, CA 94035 OWEN HOLLAND California Institute of Technology (136-93) Microsystems Lab Pasadena, CA 91125 ALLYN E. HUBBARD Boston University Electrical & Computer Engineering Boston. MA 02215 HlROSHI ISHIDA Georgia Institute of Technology School of Chemistry and Biochemistry Atlanta, GA 30332-0400 ELLIS LOEW Cornell University Department of Biomedical Sciences SI -086 Schurman Hall Ithaca, NY 14853 DAVID MACMILLAN The University of Melbourne Dept. of Zoology Parkville Victoria 3052 Australia MEGAN MAHAFFY Montana State University Center for Computational Biology 30 AJM Johnson Hall Bozeman. MT 597 1 7 JOHN P. MILLER Montana State University Center for Computational Biology 30 AJM Johnson Hall Bozeman, MT 59717 DAVID MOUNTAIN Boston University Department of Biomedical Engineering 44 Cummington Street Boston. MA 022 1 5 LESLIE OSBORNE University of California - San Francisco Dept. of Physiology - Sloan Center for Theoretical Neurobiology 513 Parnassus Ave., Room 0444 San Francisco, CA 94143-0444 DANIEL ROBERT Laboratory of Bioacoustics Institute for Zoology, Universitat Zuerich CH-8057 Zurich, Switzerland JOSEF SCHMITZ University of Bielefeld Department of Biological Cybernetics D-33501 Bielefeld, Germany NADAV SHASAR The Interuniversity Inst. of Eilat Eilat, 88103. Israel INVERTEBRATE SENSORY INFORMATION PROCESSING 245 BRIAN SMITH Ohio State University Department of Entomology 1735 Neil Avenue Columbus, OH 432 10 MANDYAM SRINIVASAN Australian National University Centre for Visual Science - Research School P.O. Box 475 Canberra, ACT 2601 Australia PASCAL STEULLET Georgia State University Biology Department P.O. Box 4010 Atlanta. GA 30302-4010 MARK A. STOPPER California Institute of Technology Pasadena. CA 91125 BARBARA WEBB University of Stirling Department of Psychology Scotland. UK WILLIAM "RED" WHITTAKER Carnegie Mellon University Robotics Institute 5000 Forbes Avenue Pittsburg. PA 15213 JAN WITTING Marine Science Center East Point Nahant, MA 01908 MARK WILLIS University of Arizona Hildebrand Lab - ARL Division of Neurobiology P.O. Box 210077 Tucson, AZ 8572 1-0077 GERALDINE WRIGHT Ohio State University Department of Entomology 1735 Neil Avenue Columbus, OH 43210 MBL EDUCATIONAL OPPORTUNITIES Educational Programs at the Marine Biological Laboratory (MBL) provide opportunities for intensive academic inquiry under the guidance of expert and distinguished faculty The teaching staff of each MBL course is drawn from a variety of reseach institutions and frequently outnumbers the student enrollment. Summer Courses have been a central part of the Laboratory's mission since its founding more than a century ago. The opportunity for exchange with members of the summer research community, as well as for intensive individual research, makes these programs unique in science education Summer courses last from six to nine weeks, and consist of formal lectures integrated with laboratory instruction and research. Laboratories emphasize application of the most current techniques. Summer courses are designed for graduate students, postdocs and independent investigators actively engaged in research Summer course topics include: Biology of Parasitism ~ Embryology -~ Microbial Diversity ~ \eural Systems Jf Behavior ~ Neurobiology ~ Physiology Special Topics Courses are intensive educational programs, one to five weeks long, providing experience in specialized research techniques, or are lecture and laboratory courses in topics of current interest. Generally, these courses are designed for graduate students, postdoctoral fellows, and other professionals actively engaged in research MBL courses are supported by a wide range of donors The overall MBL educational program has been underwritten by gifts from the Howard Hughes Medical Institute and the Burroughs Wellcome Fund. Visit our website to learn more about courses offered at the Marine Biological Laboratory. http://courses.mbl.edu Marine Biological Laboratory ~ 7 MBL Street ~ Woods Hole. AL-l 02543 ~ (508) 548-3705 D i g i t a I Imaging FHatform Meet the challenges of the digital age with Axioplan 2 Imaging. Not just a new microscope but a flexible system designed for the future - the digital imaging platform. A system that is set up and controlled by computer, with images captured and stored by computer, that communicates with other computers - truly digital from start to finish. Combine with AxioVision software, with its built-in microscope control, and the new AxioCam ultra-high resolution digital camera and you have pure synergy - microscopy and more. Make yours the future-proof choice for high resolution image capture, image processing, archiving - even tele-microscopy. The digital age is with us. Are you? 800.233.2343, ext. 7858 micro@zeiss.com www.zeiss.com/micro THE BIOLOGICAL BULLETIN '"' JUNK 2001 'ublislu'd l>\ ilu- M.ninc Hiolc.-u al I.aliotalorv $£ We walk in the footsteps of visionaries. And see with the eyes of disciples. Galileo up close: astronomer, physicist, mathematician, dethroner of Aristotle, defender ' of Copernicus and, yes, credited with making 'is double convex lenses and the first biological observations with a microscope. A compound microscope. In 1624. Hooke: 43 years later, the compound microscope reveals living things are composed of cells. Leeuwenhoek: invents the first practical microscope and discovers bacteria. Armed with the knowledge and inspiration of these and other Olympians, secure in the history of our own accomplishments and aware that the mind's eye should always focus on challenging the enigmatic as well as the undeniable, Olympus has created the BX2 series of clinical and research microscopes, noteworthy for their enhancements in optics, imaging and ergonomics. And each so individual, it deserves the nomenclature My Microscope. From possibility to actuality, it is wonderful to behold. iA posse ad esse est mirabile visu. |TheBX4l| The new standard """ in laboratory " microscopy vis-a-vis " ; •' .-"" 'I performance, ver- satility, operation. Uncanny digitized imaging: enhanced with the DP11 digital camera, delivering high-resolution image recording beyond the norm. Unusually insightful optics: observe magnifications from 1.25x to lOOx without changing the condenser. Choice of condensers includes Abbe, swing- out, phase-contrast and darkfield. - Unerringly motion sta selectable f< left hand. D speed, ease, comfort, an accuracy. Uncommonly easy-to-operate controls: extended fine-focu0 b^ .arded-pu intensity dial; tac Series The new standard in ergonomic microscopy. A dramatic metamor- phosis: innova- <£ft^ live, intelligent, insightful design improvements specifically addressing the needs of those spending hours looking through a microscope. The lowest low-position stage: three inches lower than standard microscopes, offering the promise of minimum effort and hand movement when changing specimens. The tilting, telescoping observation tube: a marvel of adjustability, it can be changed to accommodate any observation position. 80mm higher °nd 70mm loser to the operator than a convention- al tilting tube. And the eye- point can be moved forward by 45mm and inclined from 0-25 degrees. The new stan- ^ f dard in research ^ microscopy. The words "new" and "standard" really don't do it justice. How about "ingenious" or "original, ""paragon" or "prototypical." Prepare yourself for a higher power. The 6-position filter cube turret: Not 4, not 5, but 6, so single and multi- band imaging of new fluorochromes and fluorescent proteins is faster, simpler and exclusively tailored. The rectangular field stop: ours alone. By cleverly matching the camera's field of view, only the area requiring fluorescence excita- tion is exposed. Which means that the surrounding areas are protected from photo bleaching. The fluorescence excitation bal- ancers: another Olympus exclusive. A continuously variable excitation bandwidth. Visualization is enhanced, differentiation of multi-labels is immeasurably improved, capturing of images is at an unheard-of level. Triple Labeled Specimen (1 to r): • with prevalent FITC and Cy3 • with prevalent DAPI and Cy3 • with balanced fluorescence emission r mirror aspherical lens The aspherical collector lens: sui generis, in a class by itself. Specimens appear brighter and more evenly illuminated; and achromatic- performance now incorporates wavelengths from UV to IR. The universal -^,sC; condenser: with 8 positions, it can accept optical inserts for DIG, phase, brightfield and dark- field illumination. And the advanced Nomarski DIC system has been expanded arid optimized to encompass prisms for enhanced contrast and image resolution. OLYMPUS FOCUS ON LIFE Visit us at www.olympusamerica.com or call 1-800-455-8236 Olympus America Inc. ARINE IXESOURCES CENTER MARINE BIOLOGICAL LABORATORY • WOODS HOLE, MA 02543 • (508)289-7700 WWW.MBL.EDU/SERVICES/MRC/INDEX.HTML Animal and Tissue Supply for Education & Research • 150 aouatic species available for shipment via online catalog: //m nnmbilis. Proc. Natl. Acad. Sci. USA 98: 1993-1998. 16. Kuo, A. Y., and B. J. Neilson. 1987. Hypoxia and salinity in Virginia estuaries. Estuaries 10: 277-283. PUTATIVE HRE IN FUNDULUS LDH-B 251 17. Powers, D. A., M. Smith, I. Gonzalez-Villasenor, L. DiMichele, D. Crawford, G. Bernard!, and T. Lauerman. 1993. A multidisci- plinary approach to the selectionist/neutralist controversy using the model teleost, Fundulus heteroclinis. Pp. 43-107 in Oxford Sunrv.v in Evolutionary Biology, vol. 8., D. Futuyama and J. Antonovics, eds. Oxford University Press, Oxford. 18. Segal, J. A., P. M. Schulte, D. A. Powers, and D. L. Crawford. 1996. Descriptive and functional characterization of variation in the Fundu- lus heteroclitim Lilh-B proximal promoter. J. E.\p. Zo/>/. 275: 355-364. 19. Schulte, P. M., M. Gomez-Chiarri, and D. A. Powers. 1997. Structural and functional differences in the promoter and 5' flanking region of Ldh-B within and between populations of the teleost Fun- dulus heteroclinis. Genetics 145: 759-769. 20. Crawford, D. L., J. A. Segal, and J. L. Barnett. 1999. Evolution- ary analysis of TATA-less proximal promoter function. Mol. Biol. Eml. 16: 194-207. 21. Semenza, G. L., P. H. Roth, H. M. Fang, and G. L. Wang. 1994. Transcriptional regulation of genes encoding glycoytic enzymes by hypoxia-inducible factor 1. J. Biol. Chem. 269: 23757-23763. 22 Semenza, G. L., B. H. Jiang, S. W. Leung, R. Passantino, J. P. Concordet, P. Marie, and A. Giallongo. 1996. Hypoxia response elements in the aldolase A, enolase I, and lactate dehydrogenase A gene promoters contain essential binding sites for hypoxia-inducible factor 1. J. Biol. Chem. 271: 32529-32537. 23. Latchman, D. S. 1995. F.ukun-otic Transcription Factors. Aca- demic Press. London. 24. Magor, B. G., D. A. Ross, D. L. Middleton, and G. W. Warr. 1997. Functional motifs in the IgH enhancer of the channel catfish. Inwiu- nogenetics 46: 192-198. 25. Muller, F., B. Chang, S. Albert, N. Fischer, L. Tora, and U. Strahle. 1999. Intronic enhancers control expression of zebrarish sonic hedgehog in floor plate and notochord. Development 126: 2103- 2106. 26. Medrano, J. F., E. Aasen, and L. Sharrow. 1990. DNA extraction from nucleated red blood cells. Biotechniques 8: 43. 27 Bernard!, G., P. Sordino, and D. A. Powers. 1993. Concordant mitochondria! and nuclear DNA phylogenies for populations of the teleost fish Fundulus heteroclinis. Proc. Nntl. Acad. Sci. USA 90: 9271-9274. Reference: Biol. Bull. 200: 252-256. (June 2001) Neuronal Form in the Central Nervous System of the Tadpole Larva of the Ascidian Ciona intestinalis T. OKADA'-*, S. STANLEY MAcISAACt. Y. KATSUYAMA", Y. OKAMURAt AND I. A. MEINERTZHAGEN'-§ lNeuroscience Institute, Life Sciences Centre, Dalhousie University, Halifax, Nova Scotia, Canada B3H 4J1; and ' Department of Developmental and Cell Biology, Developmental Biology Center, BioScience II, University of California Irvine, Irvine, California 92697 The dorsal tubular central nen'ous system (CNS) of the ascidian tadpole larva is a diagnostic feature by which the chordate affinities of this group, as a whole, are recognized. We have used two methods to identify larval neurons of Ciona intestinalis. The first is serial electron microscopy (EM), as part of a dedicated study of the visceral ganglion (1), and the second is the transient transfection of neural plate progeny with green fluorescent protein (GFP) (2), to visualize the sorna and its neurites of individual neurons in whole-mounted lan-ae ofC. intestinalis. Our observations reveal that ascidian lan-al neurons are simple inform, with a single axonal neurite arising from a soma that is either monopolar or has only very few, relatively simple neurites arising from it, as part of a presumed dendritic arbor. Somata in the visceral ganglion giving rise to axons de- scending in the caudal nerve cord are presumed to be those of motor neurons. Regardless of whether ancestral tunicates more closely resembled the adult than its larva (3, 4), or vice versa (5), urochordates form a sister group to chordates (e.g., ref. Received 3 November 2000; accepted 15 March 2001. * Present address: Howard Hughes Medical Institute. Department of Neurobiology and Behavior, State University of New York. Stony Brook. NY 11794-5230. t Present address: 405 Penn Circle. Apt. H. Allemown. PA 18102. t Present address: Biomolecular Engineering Dept. (Ion Channel Group). National Institute of Bioscience and Human Technology, Agency iif Industrial Science and Technology, Higashi l-l, Tsukuba-shi, Ibaraki 305 X566, Japan. § To whom correspondence should be addressed. E-mail: IAN: (S.DAL.CA. 6). Yet. possibly because of their small size and transient life, the neurobiology of ascidian larvae has not received sufficient attention, even though they may be better known than other urochordates (7). This is not only a taxonomic omission but is also an opportunity missed, given that these simple nervous systems contain rela- tively few cells, only about 340 in the larva of C. intes- tinalis, an estimated 100 of which are neurons in that species (8). The criteria by which cells are recognized are, however, weak. Although cell maps exist for the entire larval CNS (8, 9), few studies identify these cells in any way. Markers identify a number of cell types (10), and the number of these has increased in recent years, but reports on the morphology of individual neurons are either lacking or are limited to partial studies on the receptor cells of the sensory vesicle (11, 12). Of partic- ular interest are the motor neuron outputs of the CNS, because these generate undulations of the tail, rhythmical and otherwise (13), that underlie the larva's swimming behavior. The motor neurons have been widely presumed to lie in the visceral ganglion, the middle division of the CNS, between the rostral swelling of the sensory vesicle and the caudal nerve cord (8, 14). To visualize neurons in their entirety, in whole- mounted larvae, we used a modification of the GFP transfection method of Corbo el ul. (2). Alternatively, neurons in the visceral ganglion were reconstructed in three dimensions from serial-section EM. For GFP ob- servations, using fluorescence microscopy we first se- lected those with intense GFP fluorescence from among several hundred transfected, fixed larvae. About 70% of 252 ASCIDIAN LARVAL NEURONS 253 the larvae had such a positive signal, and we then ob- served about 50 of these by confocal microscopy. Those that were finally selected had either a single neuronal transfection or transfected neurons that were well sepa- rated, so that we could observe the entire morphology of the individual neurons, without the possibility of confu- sion with other GFP profiles. Some larvae had multiple transfections, but the number of cells simultaneously transfected was always restricted and never included all the neurons of a region in the CNS, possibly because of the low concentration of DNA we used. Various forms of cell outline were visualized after GFP transfection (Fig. 1A-D). Confirming the neuron-specific promoter for the GFP, the cytological profiles were identi- fied as those of neurons on the basis of the defining feature of a long slender neurite. the presumed axon. The terminals of these were rarely conspicuous. In addition, we saw few presumed dendritic neurites arising from the soma. Most somata therefore appeared to be monopolar. with a shape approximating a simple prolate spheroid, although two neu- rons were bipolar (data not shown). Soma size in GFP transfected neurons varied between about 5.5 and 8 jam for the long axis and 2.5 and 5 jum for the minor axis (Fig. 1A-D); motor neurons recon- structed in three dimensions by computer were somewhat larger than this, with minor axis diameters about 5 ^irn (Fig. IE). Neuron size in the ascidian larva is thus quite small, similar to the sizes of motor neurons in, for ex- ample, the salp ganglion (15) or the arthropod brain. For the latter, an average density of 4.6 X 106 per cubic millimeter in the housefly Musca domestica (16) corre- sponds to a mean cell diameter of 7.5 p,m; and neuronal diameters in the Drosophila optic lobe fall between 3.6 and 5.0 /xm for lamina monopolar cells and 2.8 and 4.2 jLtm for medulla cells (measurements from ref. 17). In the vertebrate brain, the density of the cerebellar granule cell layer is, for reference, 3-7 X 106 (18). The small size of ascidian larval neurons is compatible with the restricted extent of their arborizations, which are among the small- est examples of deuterostome neuron. Different forms of neurons were identified with respect to the location of their somata and the direction of their axons (Fig. 1). Most neurons were descending with axons passing either from the sensory vesicle to the visceral ganglion region or from the visceral ganglion to the caudal nerve cord. For some reason, transfection of cells in the anterior region of the sensory vesicle occurred much less frequently than in either the posterior region of the sensory vesicle or the smaller visceral ganglion, even though there are many more cells in the anterior sensory vesicle. To confirm that the morphological picture of ascidian neurons was correct, we compared images of GFP trans- fected neurons with three-dimensional reconstructions of motor neurons obtained from serial EM (Fig. IE). The two neurons illustrated are one of five bilaterally sym- metrical pairs of motor neurons in the visceral ganglion that have been reconstructed from one individual ( 1 ). Observations from two other serially sectioned larvae confirm the number of cells and their general morphology but were not analyzed in detail (Stanley Maclsaac, un- pub.). The EM series contained images from only every third section, insufficient for fully tracing the finest neu- rites but sufficient to have seen coarse basal dendrites; the absence of the latter supports the picture, of simple somata lacking an extensive dendritic arbor, seen from GFP transfection. Neuromuscular terminals traced out to sites of presynaptic contact with the tail muscle do, however, indicate that these cells have bulbous terminals (Fig. IE), in at least two sites each, that were not clearly seen after GFP transfection (Fig. 1 A). Thus, GFP expres- sion in a neuron may not always reveal the axon termi- nals, perhaps because there is insufficient GFP protein transported down the axon. On the other hand, the tra- jectory of the axon is clearly revealed. Compared with the typical multipolar neuron of the ver- tebrate CNS, the form of neurons in the ascidian larval nervous system seems to be remarkably simple. All exam- ples of neurons have somata with few slender neurites, or sometimes none, and a single axon with simple terminals. Although it is possible that GFP protein does not transport completely into the finest dendrites, serial EM of the motor neurons bears out the general impression that most neurons lack well-developed dendrites. In addition to monopolar neurons, at least two transfected cells each had two neurites, an axon and perhaps a stout dendrite, and thus appeared bipolar. In the general level of their complexity, the transfected neurons resemble motor neurons in the salp ganglion (15) or in the amphioxus larva (19). By comparison, neurons in another deuterostome group, echinoderms (20, 21), seem morphologically more like vertebrate neurons. Thus neuron form does not relate closely to phylogenetic position. It does, however, correlate in general with neuron numbei, the simplest morphological types of neuron correlating with the least populous nervous systems. Indeed, one view of the dendritic tree is that it provides a structure to segregate different inputs to different dendritic limbs (22), the size and complexity of neuronal branching in a parasympathetic neuron, for example, increasing with the body size of a particular species (23, 24) and with the number of inputs on the neuron (25). One distinction between vertebrate and invertebrate nervous systems lies in the placement of the soma with respect to its neurites. The monopolar cells of inverte- brates can be seen as the product of segregating somata 254 T. OKADA ET AL Figure 1. Neurons in newly hatched larvae of dona. The rostral tip of the trunk lies to the left in each figure. (A-D) Left lateral views of wholemount larvae of C. iittesriiialis expressing green fluorescent protein (GFP) in selected neurons of the larval CNS and stained with BOBO-3. viewed as collapsed stacks of confocal images. Scale bars: 25 /xm. GFP transfection was achieved with a plasmid containing a 3.7-kb sequence upstream of ascidian synaptotagmin cDNA (Katsuyama et ai. unpublished), fused to a mutated bright form of GFP. pEGFP-Nl (Clontech) in pBluescnpt. Synaptotagmin is a protein essential for synaptic function, and in the larva is exclusively expressed in neurons, so that we used the synaptotagmin promoter to drive GFP expression in neurons. To obtain transfections, fertilized eggs were cultured in dechorionation solution, comprising 1% Na thioglycorate and 0.05% actinase E in filtered seawater to 10 ml of which 20 drops of 1 N NaOH were initially added, for 3-10 min at 18-20°C. Then eggs were dechorionated by repeated pipetting. After being washed several times, eggs were concentrated in a low-speed centrifuge and finally resuspended in 300 fj,\ of filtered seawater and subjected to electroporation in the cuvette of a BioRad gene pulser 11 in a mixture of 250 /u.1 each of DNA (100-200 /ig/ml| and 1.54 M mannitol. Successful electroporation was attained with a time constant of ca 15 ms. Surviving eggs were then incubated, at 15-18°C in an agar-coated petri dish, spread out in a layer so that they did not touch each other. Emerging larvae were fixed in 2% paraformuldehyde in PBS, mounted in 70% ASCIDIAN LARVAL NEURONS 255 into a cortex of cell bodies surrounding the neuropile formed by their neurites. Because the neuropile region is very thin in the ascidian larval CNS (1, 26), in practice this distinction is hard to make in our case. A corollary of the arrangement in invertebrates is that the axonal and dendritic regions of the neuron are not strictly segregated at different regions of the neuronal arbor (27). In con- trast, inputs in vertebrate neurons generally are exclu- sively received over the soma and its dendritic tree, quite separate from the output region at the axon terminal (28, 29). Our evidence does not yet allow us to examine whether this distinction is upheld by all neurons in the ascidian larval CNS. Available evidence does, however, indicate the presence of synaptic inputs at the soma ( 1 , 11, 12) and outputs at neuromuscular terminals (1, 30). Our GFP evidence also does not allow us to determine whether axons arise from the apical or basal region of the soma (15), with respect to the neural canal, because this feature is not resolved in wholemount preparations. The cells of the visceral ganglion are the subject of a recent study ( 1 ). Further documentation of the cells in the nervous system of the ascidian tadpole larva will, however, require a dedicated study of the sensory vesicle, where at least 60% of the cells in the CNS reside (8). Application of the GFP method we report here is a first step towards a systematic cataloging of cell types, and their correlation with cell maps derived either from histological methods (8) or from nuclear stains (31) will be a first step toward the analysis of the neural circuits these form and the behavior for which these are a substrate. Likewise, if expression turns on sufficiently early in the embryo and persists sufficiently long in the larva, the GFP method may be a way to examine the growth of neurites and their possible later regression during larval metamorphosis. We currently lack evidence on these points. Acknowledgments We thank Ms. Alison Cole for sharing unpublished work on BOBO-3 staining, Ms. Grazyna Tokarczyk for help with confocal imaging of larvae, and Ms. Jane Anne Home for support with computer 3-D reconstructions. Supported by NSERC grant 0000065 (to I. A.M.). Literature Cited 6. 1 . Stanley Maclsaac, S. 1999. Ultrastructure of the visceral ganglion in the ascidian larva dona intestinalis: cell circuitry and synaptic distri- bution. Master's thesis, Dalhousie University. Halifax. NS. Canada. 2. Corbo, J. C., M. Levine, and R. W. Zeller. 1997. Characterization of a notochord-specific enhancer from the Brachyury promoter region of the ascidian, dona intestinalis. Development 124: 589-602. 3. Garstang. W. 1928. The morphology of the Tunicata. and its bearings on the phylogeny of the Chordata. Q. J. Micmsc. Sci. 72: 51-187. 4. Bt i rill. N. J. 1950. The Tunicata with an Account of the British Species. The Ray Society, London. 5. Wada, H., and N. Satoh. 1994. Details of the evolutionary history from invertebrates to vertebrates, as deduced from the sequences of 18S rDNA. Proc. Nail Acail. Sci. USA 91: 1801-1804. Cameron, C. B., J. R. Garey, and B. K. Swalla. 2000. Evolution of the chordate body plan: new insights from phylogenetic analyses of deuterostome phyla. Proc. Nail. Acail. Sci. USA 97: 4469-4474. 7. Bone, Q., and G. O. Mackie. 1982. Urochordata. Pp. 473-535 in Electrical Conduction and Behaviour in 'Simple' Invertebrates. G. A. B. Shelton, ed. Clarendon Press, Oxford. 8. Nicol, D., and I. A. Meinertzhagen. 1991. Cell counts and maps in the larval central nervous system of the ascidian dona intestinalis (L.). J. Comp. Neurol. 309: 415-429. 9. Cole, A. G. 2000. Cell-lineage of the larval nervous system in the ascidian dona intestinalis: neurula stage through to hatched larva. Master's thesis. Dalhousie University, Halifax, NS. Canada. 10. Chiba, S., and T. Nishikata. 1998. Genes of the ascidian: an anno- tated list as of 1997. Zoo/. Sci. 15: 625-643. 1 1 . Barnes, S. N. 1971. Fine structure of the photoreceptor and cerebral ganglion of the tadpole larva of Amaroucium constellation (Verrill) (Subphylum: Urochordata; Class: Ascidiacea). Z. Zellforsch. 117: 1-16. glycerin in PBS beneath a #00 coverglass raised up from the slide by spacers of #1.5 coverglass. and visualized using laser scanning confocal microscopy (Zeiss LSM512) with a 63 X/ 1.4 Plan Neofluar objective. To confirm that GFP labeling was cellular, and to localize such labeling with respect to the cellular features of the CNS. larvae were counterstained with a nuclear stain (9). For this, larvae were permeabilized for 30 min in 4% Triton X-100 in PBS, washed in PBS, and incubated for ca 60 min in a 500-/xg/ml solution of RNAse in PBS. They were then incubated overnight in the dark in 0.01% BOBO-3 (Cat # B-3586, Molecular Probes) in PBS at 4°C. (A) Motor neuron in the region of the visceral ganglion with an axon extending into the caudal nerve cord. (B) Several cells in the posterior sensory vesicle extend their axons towards the visceral ganglion. Other cells anterior to these in the sensory vesicle have axons that may remain locally. (C) Neuron in the posterior region of the sensory vesicle with an axon that appears to branch (arrowheadl. (D) GFP-labeled cells in the visceral ganglion (arrowhead) and, anterior to this, in the sensory vesicle, with axons extending towards the visceral ganglion. In the rostral region of the tail, an irregular GFP outline is too large for a neuromuscular terminal, and so appears to be a cell in the caudal nerve cord. In the rostral tip of the trunk, a papilla exhibits GFP labeling, which could be neuronal except that no axon is visible. (E) Dorsal view of a pair of motor neurons in the visceral ganglion of C. intestinalis reconstructed in three dimensions from their profiles in the visceral ganglion, from a series of more than 2000 electron micrograph montages of 70-nm sections. The sections were cut from a fully hatched, 2-h-old larva of C. intestinalis fixed in osmium tetroxide and glutaraldehyde (8). Each motor neuron pair forms neuromuscular terminals (arrowhead) on both the dorsal and median muscle bands. Scale bar: 10 /xm. 256 T. OKADA ET AL 12. Eakin, R. M., and A. Kuda. 1971. Ultrastructure of sensory recep- tors in ascidian tadpoles. Z. Zellforsch. 112: 287-312. 13. Bone, Q. 1992. On the locomotion of ascidian tadpole larvae. J. Mar. Biol Assoc. UK 72: 161-186. 14. Katz, M. J. 1983. Comparative anatomy of the tunicate tadpole. dona intestinalis. Biol. Bull. 164: 1-27. 15. Lacalli. T. C., and L. Z. Holland. 1998. The developing dorsal ganglion of the salp Thalia democratica, and the nature of the ancestral chordate brain. Philos. Trans. R. Soc. Land. B 353: 1943-1967. 16. Strausfeld, N. J. 1976. Atlas of an Insect Brain. Springer-Verlag, Berlin. 17. Fischbach, K.-F., and A. P. M. Dittrich. 1989. The optic lobe of Drosophila melanogaster. I. A golgi analysis of wild-type structure. Cell Tissue Res. 258: 441-475. 18. Braitenberg, V., and R. P. Atwood. 1958. Morphological observa- tions on the cerebellar cortex. J. Comp. Neural. 109: 2-27. 19. Lacalli, T. C., and S. J. Kelly. 1999. Somatic motoneurones in ainphioxus larvae: cell types, cell position and innervation patterns. Acta Zool. (Stockh.) 80: 113-124. 20. Lacalli, T. C., T. H. J. Gilmour, and J. E. West. 1990. Ciliary band innervation in the bipinnaria larva of Pisaster ochraceus. Philos. Trans. R. Soc. Loud. B 330: 371-390. 21 Ghyoot, M., J. L. S. Cobb, and M. C. Thorndyke. 1994. Local- ization of neuropeptides in the nervous system of the brittle star Ophiura ophiura. Philos. Trans. R. Soc. Land. B 346: 433-444. 22. Purves, D. 1988. Body and Brain. A Trophic Theory of Neural Connections. Harvard University Press, Cambridge. 23 Purves, D., and J. W. Lichtman. 1985. Geometrical differences among homologous neurons in mammals. Science 228: 298-302. 24. Snider, W. D. 1987. The dendritic complexity and innervation of submandibular neurons in five species of mammals. J. Neurosci. 7: 1760-1768. 25. Purves, D., E. Rubin, W. D. Snider, and J. Lichtman. 1986. Re- lation of animal size to convergence, divergence, and neuronal number in peripheral sympathetic pathways. /. Neurosci. 6: 158-163. 26. Torrence, S. A. 1983. Ascidian larval nervous system: anatomy, ultrastructure and metamorphosis. Ph.D. dissertation. University of Washington, Seattle. 178 pp. 27. Bullock, T. H., and G. A. Horridge. 1965. Structure and Function in the Nen'ous Systems of Invertebrates, Vol. 2. W. H. Freeman, San Francisco. 28. Bodian, D. 1962. The generalized vertebrate neuron. Science 137: 323-326. 29. Bullock, T. H. 1974. Comparisons between vertebrates and inverte- brates in nervous organization. Pp. 343-431 in The Neurosciences: Third Study Program. F. O. Schmitt and F. G. Worden, eds. MIT Press, Cambridge. MA. 30. Tannenbaum, A. S., and J. Rosenbluth. 1972. Myoneural junc- tions in larval ascidian tail. Experientia 28: 1210-1212. 31. Meinertzhagen, I. A., A. G. Cole, and S. Stanley. 2000. The central nervous system, its cellular organisation and development, in the tadpole larva of the ascidian Ciona intestinalis. Acta Biol. Hung. 51: 417-431. Reference: Bio/. Bull. 200: 257-260. (June 2001) Larval Muscle Contraction Fails to Produce Torsion in a Trochoidean Gastropod CAROLE S. HICKMAN1-* AND MICHAEL G. HADFIELD2 Department of Integrative Biology and Museum of Paleontology, University of California, Berkeley, California 94720; and 2Kewalo Marine Laboratory, University of Hawaii, 41 Ahui Street, Honolulu, Hawaii 96813, and Department of Zoology, University of Hawaii at Manoa, Honolulu, Hawaii 96822 The causes and effects of ontogenetic torsion in gastro- pods have been debated intensely for more than a century (1-19). Occurring rapidly and very early in development, torsion figures prominently in shaping both the lan'al and adult bod\ plans. We show that mechanical explanations of the ontogenetic event that invoke contraction of larval re- tractor muscles are inadequate to explain the observed consequences in some gastropods. The classic mechanical explanation of Crofts (4, 5) and subsequent refinements of her explanation have been based on species with rigid lan'al shell properties (18, 19) that cannot be extrapolated to all gastropods. We present visual evidence of the lack of rigidity of the uncalcified lan'al shell in a basal trochid gastropod, Margarites pupillus (Gould), and provide pho- tographic confirmation of our prediction that lan'al retrac- tor muscle contraction is insufficient to produce more than local deformation or dimpling at the site of muscle inser- tion. These findings do not refute muscular contraction as a primary cause of ontogenetic torsion in gastropods that calcify their larval shells prior to the onset of torsion, nor do they refute the monophyly of torsion. They do, however, suggest that torsion may be a loosely constrained develop- mental process with multiple pathways to the more con- strained end result (20, 21). Torsion is defined as an ontogenetic process that twists the gastropod larval head and foot counterclockwise by 180° relative to the underlying (in the swimming larva) shell with its mantle epithelium and visceral mass. The process is recognized morphologically by its developmental conse- quence: a new anatomical configuration in which the mantle cavity, anus, and gill rudiment are positioned over the head. Received 21 July 2000; accepted 12 February 2001. * To whom correspondence should be addressed. E-mail: carolehfe' socrates.berkeley.edu. Torsion has been regarded as the hallmark of the Gas- tropoda (8), the evolutionary novelty or shared derived character defining this class of molluscs. When treated as a homologous feature throughout Gastropoda, torsion tends to take on a structural definition synonymous with the torted anatomical condition. As an ontogenetic event, torsion occurs early in larval development and may be completed in as little as 2 min (22). Crofts' explanation (4, 5) of the mechanism initiating developmental rotation in the trochid gastropod Calliostoma zizyphinum, the abalone Haliotis tuberculata, and the patel- lid limpet Patella vulgata involved contraction of a larval retractor muscle and assumed that the larval shell on which the muscle is inserted is calcified or at least rigid. Crofts (5) constructed wax models of gastropod larvae that were de- signed to illustrate the rotation by pulling on a string that was attached to the foot and threaded through a hole in the shell at the point of insertion of the larval retractor muscle. The use of working models of torsion appears to have been widespread in Britain. Bales (7) referred to a working model constructed much earlier by Ray Lancaster, and she bor- rowed Crofts' models and made a set that she used in her presidential address to the Malacological Society of London (7). Physical models used "to illustrate the torsion process" (5) did not accurately simulate muscle contraction; but because they replicated the observed twist, they supported the theory of muscle contraction over differential growth as an explanation of torsion. Although the onset of torsion and its completion are readily observed in populations of developing larvae of many species (pers. obs.), the underlying mechanism and details of the morphogenetic movements cannot be resolved (18). Debates over both the morphological changes and the underlying mechanisms have been based more on conjec- 257 258 C. S. HICKMAN AND M. G. HADFIELD a Figure 1. Individual frames from a videotape of the developing pre-torsional veliger larvae of the troch- oidean gastropod Margariies pupillus. (a, b) Typical appearance of the larvae in most orientations, with the shell appearing evenly rounded, (c, d) Frames in which the larval retractor muscle is rotated into view while contracting to produce a temporary local dimpling (arrow) of the shell at the site of muscle attachment. ture than on actual observation. Bandel (23) hypothesized that torsion in trochoidean gastropods was solely a conse- quence of differential growth, in contrast to the mechanism Wanninger et al. (19) described in patelloidean gastropods, which involved an alternation of muscle contractions and hydraulic swelling of the larval foot. Page ( 18). in a careful study of sectioned larvae of the haliotid gastropod Haliotis kamtschatkana Jonas, presented evidence that is consistent with potential roles for both muscle contraction and differ- ential growth. In both Haliotis and Patella the larval shell is calcified at the onset of torsion (18, 19). Can larval muscle contraction provide a global explana- tion of ontogenetic torsion? Our observations of early larval development in trochoidean gastropods led us to hypothe- size that the thin organic larval shell was insufficiently calcified at the onset of torsion and lacked the mechanical rigidity to antagonize contraction of the larval retractor muscles. Contraction of a muscle inserted on a thin organic shell should produce, instead, only a local deformation. During the late spring and early summer of 1994. repre- sentatives of the major clades of primitive marine gastro- pods (informally known as archaeogastropods) were spawned at the Friday Harbor Laboratories for comparative observations of larval development and changes in shell secretion at metamorphosis. Larvae of the margaritine tro- chid Margarites pupillus Dall were videotaped as tro- chophores and pre- and post-torsional veligers; further ob- servations were made using a combination of standard brightfield light microscopy, polarizing microscopy, and differential interference contrast microscopy. Larvae developed rapidly (at 12°C) to the pre-torsional veliger stage within the egg envelope prior to hatching, and had secreted a cap-shaped, transparent organic shell by 48 h post-fertilization (Fig. la). Although the initial shell was weakly birefringent. it collapsed and crumpled at this stage (between 48 and 72 h) when attempts were made to make preparations for scanning electron microscopy. We inferred that the primordial shell was not rigid at this stage, even if weakly calcified. Immediately prior to hatching and torsion, the active GASTROPOD TORSION 259 veligers were rotating and constantly changing position within the egg envelope so that it was impossible to make continuous observations in a standard orientation. Two lar- val retractor muscles appeared to be well-developed, but only one was visible at a time in the rotating veliger, and for only a fraction of a second. Five different observers of larvae during this period were unable to agree on the num- ber of muscles and the position of their insertion on the primordial shell. Only by videotaping larvae and examining individual frames was it possible to identify sites of apparent muscle attachment (Fig. Ib). The central observation of this paper is that the normally rounded larval shells (Fig. la, b) show a periodic transitory dimpling at the site of attachment of a larval retractor muscle (Fig. Ib, c). We argue that the dimpling refutes a mechanical explanation of torsion in this species. Because the dimpling is not permanent, because the active larva is constantly and rapidly changing its orienta- tion, and because the entire shell cannot be kept in focus, the dimple is visible only periodically and for a fraction of a second at a time. We originally intended to address the further question of whether larval muscle contraction could explain the putative mechanical deformation of the larval shell first proposed by Bandel (24). Subsequently formalized by Morita (25) as "BandeT s Rule," the idea of mechanical deformation has been linked both with torsion and with the initiation of shell coiling. It is based on the observation that the larval shells of many "archaeogastropods" have protuberant lobes bor- dered by lateral grooves, giving the shell a coiled appear- ance. Our observations of transitory dimpling of the larval shell in response to muscle contraction suggest that much more powerful forces are required to permanently deform the symmetrical larval shell as a means of giving rise to the initial coil. At least in Margarites pupillus, shell morpho- genesis is not explained by Bandel's Rule. Two other studies now provide a more direct critique of Bandel's Rule (18, 26). Collin and Voltzow (26) have shown that mechanical deformation cannot explain the ini- tiation of shell coiling in the abalone Haliotis kamischat- kana, because the larval shell is fully calcified and rigid when the alleged deformation occurs. Page (18) provides an elegant alternative explanation of the underlying observa- tions through detailed histological investigation of develop- ment in H. kamtschatkana. The creases in the lateral flank of the protoconch correspond precisely with deep visceral clefts (indentations of the shell field epithelium) in the developing larva. Her observations show that the topogra- phy of the larval shell is superimposed on the indented topography of the visceral lobe of the larva. Our observations of transitory dimpling at the site of muscle insertion on the flexible larval shell in a phylo- genetically basal trochid gastropod clade (27) immedi- ately prior to torsion show that muscle contraction is inadequate to explain the 180° rotation of the larval head and foot (cephalopodium) relative to the shell, primordial mantle complex, and visceral mass (visceropallium). However, extrapolation of these results to all gastropods is unwarranted. In patellid and haliotid gastropods the shell is calcified and rigid at the onset of torsion and may facilitate the process in an immediate mechanical sense (18, 19). In Calliostoma ligatuw, a member of a more derived trochid clade (27), the larval shell is not com- pletely calcified at the time of torsion (26), and we suspect that it also may lack the rigidity required to produce torsion by muscle contraction. Further investigations of torsion, both as a developmental event and as a key evolutionary innovation, require a phy- logenetic context in which comparative morphogenesis is pursued at a much higher level of resolution. The phenom- enon of torsion becomes more interesting if loose con- straints on the evolution of development have indeed per- mitted the emergence of multiple morphogenetic pathways to a single, more tightly constrained body plan of the torted larva and adult gastropod. Acknowledgments We thank R. and M. Strathmann for space and facilities at the Friday Harbor Laboratories and for their ideas and suggestions during the course of this study. Two anony- mous reviewers provided helpful suggestions for improve- ment of the manuscript. Literature Cited 1. Spengel, J. W. 1881. Die Geruchsorgane und das Nervensystem der Mollusken. Zeitschrift fiir senschaflliche Zoologie 35: 333-383. 2. Amaudrut, A. 1898. La partie anterieure du tube digestive et la torsion chez les mollusques gasteropodes. Ann. Sci. Nat. Zool. 7: 1-391. 3. Garstang, W. 1928. Origin and evolution of larval forms. Nature 122: 366. 4. Crofts, D. 1937. The development of Haliotis tuberculata with special reference to organogenesis during torsion. Philos. Trans. R. Soc. Land. B 228: 219-268. 5. Crofts, D. 1955. Muscle morphogenesis and torsion. Proc. Zool. Soc. Land 125: 711-750. 6. Yonge, C. M. 1947. The pallial organs in the aspidobranch Gas- tropoda, and their evolution throughout the Mollusca. Philos. Trans. R. Soc. Land. B 232: 443-518. 7. Eales, N. B. 1949. Torsion in the Gastropoda. Proc. Malac. Soc. Land. 28: 53-61. 8. Morton, J. E. 1958. Torsion and the adult snail; a re-evaluation. Proc. Malac. Soc. Land. 33: 2-10. 9. Ghiselin, M. T. 1966. The adaptive significance of gastropod tor- sion. Evolution 20: 337-348. 10. Batten, R., H. B. Rollins, and S. J. Gould. 1967. Comments on "the adaptive significance of gastropod torsion." Evolution 21: 405-406. 1 1. Thompson, T. E. 1972. Adaptive significance of gastropod torsion. Malacologia 5: 423-430. 12. Underwood, A. J. 1972. Spawning, larval development and settle- 260 C. S. HICKMAN AND M. G. HADFIELD ment behaviour of Gihhula cineraria (Gastropoda: Prosobranchia) with a reappraisal of torsion in gastropods. Mar. Biol. 17: 341-349. 13. Stanley, S. M. 1982. Opercula and predators: The evolutionary significance of gastropod torsion. Neites Jahrb. Geol. Palaontol. Abh. 164: 95-107. 14. Pennington, J. T.. and F.-S. Chia. 1985. Gastropod torsion: a test of Garstang's hypothesis. Biol. Bull 169: 391-396. 15. Goodhart, C. B. 1987. Garstang's hypothesis and gastropod torsion. J. Molluscan Stud. 53: 33-36. 16. Edlinger, K. 1988. Torsion in gastropods: a phylogenetic model. Malacol. Rev. Suppl. 4: 21 1-250. 17. ll.is/pi niiai . G. 1988. On the origin and evolution of major gastro- pod groups, with special reference to the Streptoneura. J. Molluscan Stud. 54: 367-411. 18. Page, L. R. 1997. Ontogenetic torsion and protoconch form in the archaeogastropod Haliotis kamtschatkana: evolutionary implications. Acta Zool. 78: 227-245. 19. Wanninger, A., B. Ruthensteiner, and G. Haszprunar. 1999. Tor- sion in Patella caerulea (Mollusca, Patellogastropoda): ontogenetic process, timing, and mechanisms. Invertebr. Biol. 119: 177-187. 20. Wagner, G. 1994. Homology and the mechanisms of development. Pp. 273-200 in Homology: The Hierarchical Basis of Comparative Biologv. B. K. Hall, ed. Academic Press, San Diego, CA. 2 1 Hickman. C. S. 1999. Larvae in invertebrate development and evo- lution. Pp. 21-59 in The Origin and Evolution of Larval Forms. B. K. Hall and M. H. Wake. eds. Academic Press, San Diego, CA. 22 Boutan, L. 1898. Sur le developpement de VAcmaea virginea. C. R. Hebd. Seances Acad. Sci. Paris 126: 1882-1889. 23. Bandel, K. 1982. Morphologic und Bildung der fruhontogenetischen Gehause bei conchiferin Mollusken. Fades 7: 1-198. 24. Bandel, K. 1986. The ammonitella: a model of formation with the aid of the embryonic shell of archaeogastropods. Lethaia 19: 171-180. 25. Morita, R. 1993. Development mechanics of retractor muscles and the "Dead Spiral Model" in gastropod shell morphogenesis. Neues Jahrb. Geol. Palaontol. Abh. 190: 191-217. 26. Collin, R., and J. Voltzow. 1998. Initiation, calcification, and form of larval "archaeogastropod" shells. J. Morphol. 235: 77-89. 27. Hickman, C. S. 1996. Phylogeny and patterns of evolutionary radi- ation in trochoidean gastropods. Pp. 177-198 in Origin and Evolu- tionary Radiation of the Mollusca, }. D. Taylor, ed. Oxford University Press. Oxford. Reference: Bio/. Bull. 200: 2dl-267. (June 2001) Multiple Paternity in Littorina obtusata (Gastropoda, Littorinidae) Revealed by Microsatellite Analyses IAN G. PATERSON1 *, VANESSA PARTRIDGE2, AND JOHN BUCKLAND-NICKS2 ^Department of Biology, Acadia University, Wolfi'ille, Nova Scotia. Canada BOP 1X0; and ~ Department of Biology, St. Francis Xavier University. Antigonish, Nova Scotia. Canada B2G 2W5 Abstract. Parental identity for juvenile Littorina obntsata was determined from three egg masses by means of micro- satellite DNA markers. Results confirm that the attendant adult female in each case was the dam of the offspring and that at least 4-6 males contributed to each brood. This correlates with our behavioral observations that indicated multiple copulations between the female and several males in each experimental aquarium. A significant number of offspring from each brood were sired by non-sampled males (males that had copulated with females before capture) whose sperm had been stored by the female. This is the first direct evidence of multiple paternity in the Littorinidae. Results are discussed in reference to current theories of sperm competition, male precedence, and cryptic female choice. Introduction Littorinids, as well as most other gastropods, fertilize internally. Sperm are stored temporarily in the bursa copu- latrix and for longer periods (3 months or more) in a second storage organ, the seminal receptacle, which usually is located deep within the oviduct (Buckland-Nicks et al.. 1999). In other animals where multiple paternity occurs, this scenario of two sperm storage organs may be correlated with cryptic female choice, in which the female selects among the sperm of several rival males (see review by Birkhead and Moeller. 1998). Some gastropods are capable of using stored sperm for more than one year (Triib, 1990: Baur. 1998). Furthermore, in pulmonates, one copulation per reproductive season is sufficient to fertilize all the eggs produced by an individual in one year, which suggests that females control sperm release from the seminal receptacle, using only a fraction of stored sperm for each batch of eggs (Chen and Baur, 1993; Haase and Baur, 1995). Molecular genetic markers have been used very effec- tively in the past to create DNA "fingerprints" that can genetically determine paternity. Previous studies of gastro- pod relationships have suggested that starch gel electro- phoresis is inadequate to "elucidate the ecological dynamics of fertilization and paternity" (Gaffney and McGee, 1992), whereas the high variability at microsatellite DNA loci enables a more powerful analysis of parental relationships. The intertidal snail Littorina obtusata is gonochoric. Fe- males lay benthic egg masses containing 50-150 eggs (Ban- del, 1974) in which embryos undergo direct development and emerge as crawling juveniles. In the present study, single locus microsatellite DNA markers were used to de- termine paternity of L. obtusata juveniles hatched from three egg masses kept in isolation in aquaria with several potential sires and the attendant dam in each case. Materials and Methods Samples Received 2 November 2000; accepted 14 February 2001. * To whom correspondence should be addressed. ian.paterson@acadiau.ca. E-mail: Adult specimens of Littorina obtusata were collected in early May 1999 from sites along the northeastern shore of Nova Scotia, Canada. Each shell was marked for gender and site number with nail varnish. All female snails used in the breeding experiment were collected from the same site (Port Bickerton) to ensure that all were in a relatively synchro- nous breeding cycle. Males were collected from five sites (Table 1 ) to increase the chance of genetic variability among the competing sires and thus increase the ability to assign paternity to the resulting offspring. Individual females were isolated with 4-5 males and kept in 20-1 aquaria at 8°C with a 12-h light/dark cycle. Mating 261 262 I. G. PATERSON ET AL. Table 1 Collection sires in Nova Scoria of male Littonna obtusata used in breeding and paternity analysis Location Male ID code Position Port Bickerton Fisherman's Harbour Wharf Isaacs Harbour Seal Harbour Wharf Tor Bay Provincial Park PB FH IH SH TB 45° 06'N, 61° 43'W 45°07'N. 61° 40' W 45° 12'N. 61° 39'W 45° 09'N, 61° 34'W 45° 14'N, 61° 19'W behavior was observed opportunistically throughout the ex- periment. Males that were observed in a stationary position, forward, and on the right side of the female's shell with the shell-lip and mantle touching, were recorded as "copulat- ing." Males that were observed similarly positioned but slightly distal and without contact between the male mantle and female shell-lip were recorded as "possible mate-guard- ing" (see Discussion). Once egg masses were laid, the water temperature was increased to 10°C, which was reported to be a suitable temperature for egg development in the con- gener Littorina sitkami (Buckland-Nicks, 1974). Three ad- ditional males were included in the analyses for one brood (brood C). These males had escaped into the study tank and were then considered as potential sires. When the juveniles reached about 1.4 mg in weight, they were frozen in a physiological buffer ( 10 mM Tris, pH 6.8) at — 86°C. Adults were removed from their shells and frozen at — 86°C in buffer following successful oviposition by the female. DNA extraction For each adult snail, a section of the foot was excised, rinsed with double-distilled H2O and chopped, using a fresh razor blade, on a clean glass plate. This tissue was placed in 100 /xl of TE ( 10 mM Tris-HCl pH 8, 1 mM EDTA) plus 0.5% Tween 20, and incubated at 60°C with 3 ju.1 proteinase K (20 mg/ml). Following complete digestion, the DNA was isolated using a standard phenohchloroform method (Sam- brook et ai, 1989). The DNA was further purified by passing it through polyvinylpolypyrrolidone spin columns (Berthelet et ai, 1996; Paterson and Snyder, 1999) to ensure the removal of compounds that are known to co-purify with molluscan DNA and inhibit the polymerase chain reaction (Winnepenninckx et a!., 1993; Mikhailova and Johannes- son, 1998). DNA concentration was estimated for each individual against Lambda Hindlll molecular weight marker using agarose gel electrophoresis (Maniatis et al., 1982). DNA from individual juveniles ( — 1.4 mg wet weight) was isolated using the following protocol. Each juvenile was rinsed with double-distilled H2O and placed in an Eppendorf tube with 40 ju.1 of TE plus 0.5% Tween 20, then crushed against the side of the tube with a micropipetter tip. Two microlitres of proteinase K (20 mg/ml) was added, and the tube was incubated at 60°C for 20 min, followed by a 5-min incubation at 95°C to denature the proteinase K. The tubes were spun briefly to pellet the shell debris. To avoid any loss of DNA during subsequent purification steps, no further purification was conducted. DNA amplification DNA was screened for microsatellite repeats at 12 loci using primers designed on the congener Littorina subrotun- ilntci (Tie et al., 2000). Of the 12 primer pairs, 8 did not produce scorable amplifications, and one locus was invari- ant (Lsub 6). Three loci were used for the analyses (Table 2). Approximately one-third of the offspring of each brood was typed at these three loci (40-46 juveniles per brood). Amplifications were carried out by polymerase chain reaction (PCR) in 5-/il volumes containing —0.2 ng DNA, 0.08 juJlf y33P-ATP end-labeled forward primer. 0.42 juA/ forward primer, 0.5 n.M reverse primer, 20 /iM each dNTP. 0.2-0.5 units Taq polymerase. and 1 X buffer containing 20 mM Tris pH 8, 50 mM KC1, 1.5 mM MgCl2, and 0.1% Tween 20. Reactions were performed in an MJ Research PTC 100 thermocycler using an initial denaturation of 3 min at 94°C, followed by 25 cycles of denaturing at 93°C for 30 s per cycle, followed by optimal annealing (Lsub 8: 60°C; Lsub 32: stepdown from 55°C to 48°C; Lsub 62: Table 2 Microsatellite loci used for paternity assessment in Littorina obtusata Locus Repeat motif* PCR product size range (bp) Number of alleles Primer sequence (5'-3'l Lsub 8 (ATA),, 178-211 10 ccagtgaccagatcatagcg ggaattgtaaagtgcttggagc Lsub 32 (CAA),,TG(AAC), 210-343 14 atcacatcgcacacgcttac TG(AAC),n acggtgtgtcatcatcaacg Lsub 62 (AAC)I4 160-187 8 cgctttcccgttataccaac caccgtaaaaccttgtgagc : From Tie et al. 2000. MULTIPLE PATERNITY IN L1TTOR1NA 263 64°C) for 40 s + Is per cycle. Products were separated using 6% denaturing polyacrylamide gel electrophoresis and visualized using autoradiography (Hillis et ai, 1996). All indi viduuls were scored at each locus except for one offspring of female A which did not amplify and was discarded, and two offspring of female B which were not typed at one locus each. Maternity was confirmed by direct comparison to the maternal genotype. Statistical analyses Paternity was initially assessed using exclusion, wherein putative fathers were excluded if there were any mismatches between their alleles and those of the offspring. Subse- quently, paternity inference was conducted using CERVUS software (Marshall et ai, 1998; Slate et ai. 2000), which is a Windows-based maximum likelihood program designed for use with co-dominant markers. Paternity assignment using likelihood techniques was determined at an 80% confidence threshold. Marshall et al. (1998) suggested that 80% confidence in paternity is more accurate than can be achieved using direct observation, and better than can be achieved using exclusionary approaches because the degree of confidence in non-excluded males is unknown in the exclusion technique. Paternity is inferred from the log- likelihood ratios (LOD score) based on the genotypes of the offspring, candidate sire, and dam. Confidence levels are determined through simulation (10.000 iterations) and de- fined by the statistic delta (A) where A is the difference between the LOD scores of the two most likely candidates. Input parameters for the CERVUS software have impor- tant effects on the resulting paternity assignments. The "error rate" parameter of the simulation program allows an estimate of scoring error to be included in the calculations. Allowing for error prevents exclusion of true paternal can- didates due to errors causing single locus mismatches, while potentially including (erroneously) sires that would other- wise be excluded. As the number of loci sampled increases, this error estimator becomes more important due to the increased probability of mis-scoring or mutation (Slate et ai, 2000). Our data set is small and all scores were double- checked for accuracy, so the error rate is probably lower than the 1% adopted by Marshall et al. (1998). Another important simulation parameter for this data set is the pro- portion of males sampled. We assumed a minimum number of non-sampled males (9, see Discussion). Results Behavioral obsen'ations Mating behavior in Littorina obtusata involves pursuit of a female by a male during which he crawls over the right side of her shell and inserts the penis, engorged with he- molymph. into her gonopore, which opens into the oviduct and bursa copulatrix. A series of ventral mammiliform penial glands are brought into contact with the shell and mantle during this process. Male ejaculate exits the prostate gland in a dorsal penial groove and passes along the groove with the aid of ciliary action. On several occasions when copulating pairs were separated, ejaculate was seen moving along the penial groove toward the tip of the penis. A total of 19 copulatory pairings and 8 incidences of suspected mate-guarding were observed. Males remained in copulatory position with females for 1-4.5 h and occasion- ally remained on the female, without genital contact, for up to 2.5 h after copulation. Each female was observed copu- lating with more than one male, and repeat copulations by individual males occurred in a number of cases. Eggs were laid from 7 to 21 days after introduction of females to the aquaria. Microsatellite analyses Paternity of each brood involved not only sampled males from aquaria but also non-sampled males that had copulated with females prior to capture. Assessment of paternity through exclusion (putative parents are excluded as true parents if a mismatch occurs at any allele that is present in the offspring) suggested that a minimum of 2 of the sampled males contributed to each brood. Additionally, the presence and distribution of the alleles that were not contributed by the sampled males requires a minimum of 2 non-sampled sires of brood A, 4 non-sampled sires of brood B, and 3 non-sampled sires of brood C to explain the genotypes found in the offspring. Based on exclusion, these non- sampled males contributed from 22.5% (9 of 40, family C) to 70.7% (29 of 41, family A) of the offspring per brood. Thus, the alleles present in the offspring are accounted for by a minimum of 4 sires of brood A, a minimum of 6 sires of brood B, and a minimum of 5 sires of brood C. Determination of paternity based solely on exclusionary methods is simplistic and, in this case, inadequate to accu- rately resolve paternity. Paternity assessment was therefore conducted using CERVUS software to enable paternity assignments with 80% confidence based on maximum like- lihood calculations. An important aspect of the CERVUS likelihood analyses is the estimate of scoring error (Mar- shall et al, 1998). We determined paternity assignment for a range of scoring-error estimates (from 0% to 1.0%); however, results were similar and only 0.5% assignments are presented here (Table 3). Paternity was assigned with 80% confidence in 19.5% of the offspring of brood A, 40.8%' of brood B, and 20.0% of brood C. Paternity assign- ment was recalculated for each brood by using a modified data set that included the two most likely non-sampled male genotypes inferred from the genotypes of the non-assigned offspring. With inclusion of the inferred male genotypes. 264 I. G. PATERSON ET AL. Table 3 Summary of paternity inference at 807c confidence computed using the CEKVUS program Female Candidate sire (number of obs. matings) Known paternal genotypes Offspring sired Percent sired Including inferred paternal genotypes Offspring sired Percent sired A PB1 (1) 2 4.9 2 4.9 FH1 (1) 0 — 0 — 1H1 (0) 0 — 0 — SHI (0) 6 14.6 0 — TB1 (3) 0 0 3 7.3 No assignation 33 80.5 6 14.6 Unknown Al n/a n/a 25 61.0 Unknown A2 n/a n/a 5 12.2 B PB2 (3) 11 23.9 17 37.0 FH2 (2) 8 17.4 14 30.4 SH2 (2) 0 0 0 0 No assignation 27 58.7 8 17.4 Unknown Bl n/a n/a 2 4.3 Unknown B2 n/a n/a 5 10.9 C PB4 (2) 7 17.5 7 17.5 FH4U) 0 0 1 2.5 SH4 (0) 0 0 0 0 TB4 (5) 0 0 0 0 IH3 (0) 1 2.5 1 2.5 SH3 (0) 0 0 0 0 PB3 (0) 0 0 3 7.5 No assignation 32 80.0 24 60.0 Unknown Cl n/a n/a 1 2.5 Unknown C2 n/a n/a 3 7.5 Total assigned 36 28.0 89 70.0 Paternity assignment is indicated for a scoring-error estimate of 0.5%. Unknowns indicate the two most likely non-sampled sires inferred from the observed genotypes. paternity was assigned for 85.4%, 82.6%, and 40.0%- of the offspring in the three respective broods (Table 3). Multiple mating was observed in all three females. Mul- tiple paternity is clearly evident within the three broods, but the degree of paternal success varies throughout. Results produced using the modified data set (inferred male geno- types included) indicate that brood A was dominated by a single sire (61.0%. unknown Al); brood B had two rela- tively equally successful sires (37% for PB2, 30.4% for FH2) with at least four other sires contributing less than 11% each; and brood C was sired by at least five males, with no clearly dominant contributors. Discussion Duration of copulation and incite guarding The duration of copulations we observed for Littorina obtusata in the laboratory varied from 1 to 4.5 h, which is in agreement with Ankel's ( 1936, cited in Baur, 1998) observations for this species and similar to times reported from some other prosobranchs (Martel et al., 1986). !n contrast, Struhsaker ( 1966) reported copulatory pairing to last 30-45 min in L. picm and only 5-10 min in L. pintado. These differences in duration of mating may reflect real differences in reproductive behavior or mor- phology among different littorinids, including the pres- ence or absence of penial glands, which are thought to secure the penis in place during copulation (Reid, 1989; Buckland-Nicks and Worthen, 1992). Parker (1970a) suggested that in some animal species, such prolonged copulations could be the male's way of reducing the probability of female remating by acting as a mechanical plug. Males of many animal species exhibit "passive phase behavior" in which they guard reproductive fe- males after copulating (e.g., Parker and Smith. 1975; Parker, 1984). The males in this study exhibited such behavior, remaining on the right side of the female's shell without genital contact for up to 2.5 h after copulation. An extended mating time plus guarding behavior may allow sufficient time for sperm to be transferred from the bursa copulatrix to the seminal receptacle, thereby in- creasing the male's chance of paternity. The time taken to transfer sperm between storage organs is not known for the Littorinidae. but Martel et al. (1986) reported that MULTIPLE PATERNITY IN LITTORINA 265 sperm are transferred quickly from the bursa to the sem- inal receptacle in Buccinum undatum, taking at the most 3 days. Multiple paternity Multiple paternity has been demonstrated in a number of molluscan species. Among gastropods, studies of multiple paternity have been largely restricted to pulmonates (e.g., Mulvey and Vrijenhoek. 1981; Rollinson et al.. 1989; Wethington and Dillon, 1991; Baur. 1994). with fewer investigations conducted on marine prosobranchs (e.g., Gaffney and McGee, 1992). However, multiple copulations by females with different males have been reported in a wide range of gastropod families, suggesting that multiple paternity is common (see Martel et al.. 1986). Evidence of multiple paternity in L. obtusata was expected from both field and laboratory observations of multiple copulations by individual females. In this study, the number of males contributing to each brood is relatively high, with an abso- lute minimum of 4-6 sires per brood. This may be an underestimate of the number of true sires since only a fraction of each brood was tested, and a number of the alleles examined were fairly common. The candidate sire that had immediate access to the females prior to spawning (i.e., sampled males) sired fewer offspring (average of 37.8%) than the previously mated males (non-sampled), providing no evidence of last-male precedence. This is in agreement with a study of pseudoscorpions by Zeh and Zeh (1994). who found last- male precedence to be typical of two male systems (see also Parker. 1970b; Parker and Smith. 1975; Gwynne. 1984). but found no evidence of last-male precedence in broods from more promiscuous females. The success rate of individual males appears highly vari- able— the dominant sire contributed from 61% (non-sam- pled Al, brood A) to 17.5% (PB4. brood C) of the offspring. Although the dominance of brood A by a single male is suggestive of first-male precedence, this is not upheld in the other broods. Indeed, brood B had greater than half of the offspring sired by "later" males (67.4%), and no previously mated males contributed to more than 1 1% of the offspring (Table 3). Thus, first-male precedence is not the prevalent reproductive mode in L. obtusata. Copulation does not necessarily indicate successful sperm transfer to the female, because the male may not have inserted his penis or may not have delivered ejaculate (Gib- son, 1964). However, when Buckland-Nicks et al. (1999) examined sperm transfer in the penis of Littorina littorea specimens that had been flash-frozen during confirmed cop- ulation, they determined that sperm transfer often had been occurring at the time of disruption. In the present study, incidence of copulation was inferred based on male position and duration of coupling. Pairs invariably separated when picked up and, as in observations made by Gibson ( 1964) on L. planaxis, ejaculate was observed moving along the penis in some instances. It i-- probable that sperm transfer was not occurring during some observations of "copulation"; for example, male TB4 was observed mating 5 times and male SH2 observed 2 times (Table 3), yet neither was successful in siring offspring. In L. obtusata the high degree of multiple paternity in the resulting offspring indicates that sperm must be mixing within the female's reproductive tract. Although release of sperm from the seminal receptacle may merely be the result of the female contracting the muscle layer surrounding the seminal receptacle (Buckland-Nicks and Chia, 1990), post- copulatory selection is also a possibility. This could occur either through sperm competition within the seminal recep- tacle or bursa copulatrix, or through post-copulatory female choice. Post-copulatory female choice may be accom- plished in a number of ways, such as by dumping unwanted sperm, as observed in L. scabra (reported in Buckland- Nicks, 1998), by sperm digestion in the bursa copulatrix (see reviews by Buckland-Nicks, 1974; Fretter and Graham, 1994), or by sperm sorting and differential use within the reproductive tract (Haase and Baur, 1995). On sperm storage organs How does such short-term mate guarding increase the male's chances of successful paternity? Species with a single blind-ended sperm storage organ usually show a pattern of last-male precedence in siring offspring, because rival males have the opportunity to actively disrupt or remove any sperm already stored there (e.g., Waage, 1979. 1984) or to passively displace the sperm deeper, rendering it ineffectual (e.g., Lefevre and Jonsson, 1962; Parker, 1970b). However, last-male precedence is not prevalent in species with two sperm storage organs, like littorinid snails, where there is an opportunity for the first male's sperm to be transferred to the second storage organ before physical interference is possible by a subsequent male. Besides the number of storage organs present, the shape of the sperm storage organs also has an effect on sperm precedence. Species which exhibit last-male precedence often have a relatively long, tubular, blind-ended storage organ, which is thought to promote stratification of sperm, with older sperm being overlaid by sperm from successive matings (e.g., Walker, 1980; Briskie and Montgomerie. 1993). However. Zeh and Zeh (1994) suggest that though tubular morphology may favor sperm stratification, this relationship breaks down if the storage organ becomes completely filled, which causes increased pressure and in- duces sperm mixing. Spherical storage organs are thought to promote sperm mixing (Walker. 1980). In L. obtusata, the bursa copulatrix is the primary sperm storage organ. The bursa is a blind- 266 I. G. PATERSON ET AL ended sac, providing the potential for sperm stored there to be disturbed by the penial filaments of rival males. Although there is no evidence of physical sperm displacement from the bursa copulatrix in Linorinu, this remains a possibility because penial filaments are extensible and very mobile (Buckland-Nicks, 1974; Buckland-Nicks et ai, 1999). The secondary sperm storage organ, the seminal receptacle, is much smaller than the bursa and is situated deep within the pallial oviduct, beyond the reach of the penial filament (Buckland-Nicks and Chia, 1990). The longer a male guards a female, preventing her from participating in further mat- ing, the greater the likelihood of his sperm leaving the bursa copulatrix and becoming stored in the seminal receptacle, to which male genitalia have no access. Whether female littorinids can selectively store and use sperm from the seminal receptacle is not known. Our results for two of the three broods (A and C) indicate that the males that mated prior to capture had more success than the last males to mate, even though these later males were isolated with the female for up to 3 weeks. Moreover, males from places other than the home locale of the females (Port Bickerton) had little to no success. It is thus possible that females have some ability to stratify sperm in the seminal receptacle or to dump sperm from the bursa copulatrix. In this context, Haase and Baur (1995) suggested that females may be able to manipulate fertilization of eggs by storing sperm in different areas of the seminal receptacle. In L. obntsata, one might expect that the first sperm to reach the seminal receptacle would be stored close to its exit, favoring first-male precedence. However, our results sup- port neither first nor last-sperm precedence and suggest there is mixing of sperm within the female's reproductive tract. The results of paternity from the three egg masses of L. obntsata raise some intriguing questions about sperm selection. To fully understand the occurrence of sperm competition or cryptic female choice in these snails, exper- iments must be designed that track the sperm of individual males inside the female, as well as in the resulting offspring. This might be accomplished by using sperm-specific fluo- rophores, genetic markers, or the tried and tested method of radioautography (Beeman, 1970). Acknowledgments This study would not have been possible without the generous contribution of primers by E. Boulding. Thanks are also due to M. Snyder and D. Stewart for helpful advice and the use of lab space and equipment. We are grateful to G. Gibson for comments that improved the manuscript. This study was supported by a NSERC of Canada research grant to J.B-N. Literature Cited Ankel, VV. E. 1936. Prosobranchia. Pp. 1-240 in Die Tierwelt der Nord- nml O.ttsce, G. Gnmpe and E. Wagler, eds. Akademische Verlagsge- meinschaft. Leipzig. Bandel. K. 1974. Studies on Littorinidae from the Atlantic. Veliger 17: 92-114. Baur. B. 1994. Multiple paternity and individual variation in sperm precedence in the simultaneously hermaphroditic land snail Arianta arbustorum. Behav. Ecol. Sociobiol. 35: 413-421. Baur, B. 1998. Sperm competition in molluscs. Pp. 255-305 in Sperm Competition and Sexual Selection, T. R. Birkhead and A. P. M0eller, eds. Academic Press. New York. Beeman, R. D. 1970. The anatomy and functional morphology of the reproductive system in the opisthobranch mollusk Phyllaplysia taylori Dull. 1950. Veliger 13: 1-31. Berthelet, M., L. G. Whyte, and C. W. Greer. 1996. Rapid, direct extraction of DNA from soils for PCR analysis using polyvinylpoly- pyrrolidone spin columns. FEMS Microbiol. Lett. 138: 1 7-22. Birkhead. T. R., and A. P. Maeller. 1998. Sperm Competition and Sexual Selection. Academic Press. London. 826 pp. Briskie. J. V.. and R. Montgomerie. 1993. Patterns of sperm storage in relation to sperm competition in passerine birds. Condor 95: 442-454. Buckland-Nicks, J. 1974. The functional morphology of the reproduc- tive systems and the larval development of two intertidal snails. Litto- rina sitkana (Philippi 1845) and Linorina scutii/ata (Gould 1849) with special reference to the spermatozoa. Ph.D. thesis. University of Al- berta. Canada. 258 pp. Buckland-Nicks, J. 1998. Prosobranch parasperm: sterile germ cells that promote paternity? Micron 29: 267-280. Buckland-Nicks, J., and F. S. Chia. 1990. Egg capsule formation and hatching in the marine snail Linorina sitkana. Philos. Trans. R. Soc. Loml. B 326: 159-176. Buckland-Nicks. J., and G. T. Worthen. 1992. Functional morphology of the mammiliform penial glands of Linorina saxatilis (Gastropoda). Zoomorphology 112: 217-225. Buckland-Nicks, J., I. Bryson, L. Hart, and V. Partridge. 1999. Sex and a snail's sperm: on the transport, storage and fate ot dimorphic sperm in Littorinidae. Int. J. Invertebr. Reprod. De\. 36: 145-152. Chen, X.. and B. Baur. 1993. The effect of multiple mating on female reproductive success in the simultaneously hermaphroditic land snail Arianta arbustorum. Can. J. Zoo/. 71: 2431-2436. Fretter, V., and A. Graham. 1994. British Prosobranch Molluscs. Their Functional Anatomy and Ecology, vol. 161. The Ray Society of Lon- don, London. 820 pp. Gafi'ney, P. M.. and B. McGee. 1992. Multiple paternity in Crepidula fornicata. Veliger 35: 12-15. Gihson, D. G. 1964. Mating behaviour in Linorina planaxis Philippi (Gastropoda: Prosobranchia). Veliger 7: 134-139. Gwvnne, D. T. 1984. Male mating effort, confidence of paternity, and insect sperm competition. Pp. 1 17-149 in Sperm Competition and the Evolution of Animal Mating Systems, R. L. Smith, ed. Academic Press. New York. Haase. M., and B. Baur. 1995. Variation in spermathecal morphology and storage of spermatozoa in the simultaneously hermaphroditic land snail Arianta arbustorum (Gastropoda: Pulmonata: Stylommatophora). Imcrlchr. Reprod. De\: 28: 33-41. Hillis, D. M., B. K. Mable. A. Larson, S. K. Davis, and E. A. Zimmer. 1996. Nucleic acids IV: sequencing and cloning. Pp. 321-384 in Molecular Systematic^. 2nd ed. D. M. Hillis, C. Moritz, and B. K. Mable. eds. Sinauer. Sunderland. MA. Lefevre. G., and U. B. Jonsson. 1962. Sperm transfer, storage, displace- ment and utilization in Drosophila melanogaster. Genetics 47: 1719- 1736. M. .minis. T., E. F. Kritsch. and J. Sambrook. 1982. Molecular Clou- MULTIPLE PATERNITY IN UTTORINA 267 ing, n Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor. NY. 545 pp. Marshall. T. C., J. Slate, L. E. B. Kruuk, and J. Pcmberton. 1998. Statistical confidence for likelihood-based paternity inference in natural populations. Mol. Ecol. 7: 639-655. Martel. A., D. H. Larrivee, and J. H. Himmelman. 1986. Behaviour and timing of copulation and egg-laying in the neogastropod Buccinum undatum L. J. Exp. Mar. Biol. Ecol. 96: 27-42. Mikhailova. N., and K. Johannesson. 1998. A comparison of different protocols for RAPD analysis of Littorina. Hydrobiologia 378: 33-42. Mulvey. M., and R. C. Vrijenhoek. 1981. Multiple paternity in the hermaphroditic snail Biomphalaria obstrucla. J. Hered. 72: 308-312. Parker, G. A. 1970a. Sperm competition and its evolutionary conse- quences in insects. Biol. Rev. Camb. Philos. Soc. 45: 525-567. Parker, G. A. 1970b. Sperm competition and its evolutionary effect on copula duration in the fly Scalophaga stercoraria. J. Insect Physiol. 16: 1301-1328. Parker, G. A. 1984. Sperm competition and the evolution of animal mating strategies. Pp. 2-60 in Sperm Competition and Sexual Selec- tion. T. R. Birkhead and A. P. Moeller. eds. Academic Press. New York. Parker, G. A., and J. L. Smith. 1975. Sperm competition and the evolution of the precopulatory passive phase behaviour in Lucusta tnigratona migrcttorioides. J. Entomol. Ser. A 49: 155-171. Paterson, I. G., and M. Snyder. 1999. Genetic evidence supporting the taxonomy of Geuni peckii (Rosaceael and G. radiatiim as separate species. Rhodora 101: 325-340. Reid, D. G. 1989. The comparative morphology, phylogeny and evolu- tion of the gastropod family Littorinidae. Philos. Trans. R. Soc. Ltmd. B324: 1-110. Rollinson, D., R. A. Kane, and J. R. L. Linnes. 1989. An analysis of fertilization in Biiliiin.i cemicu.i (Gastropoda: Planorbidae). J. Zoo/. Lund. 217: 295-310. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular Clon- ing, a Laboratory Manna/. Cold Spring Harbor Laboratory Press. Cold Spring Harbor. NY. Slate, J., T. Marshall, and J. Pemberton. 2000. A retrospective assess- ment of the accuracy of the paternity inference program CERVUS. Mol. Ecol. 9: 801-808. Struhsaker, J. W. 1966. Breeding, spawning periodicity and early de- velopment in the Hawaiian Liitorina: L pintado (Wood). L. picta (Philippi) and L. scabra (Linne). Proc. Malacol. Soc. Land. 37: 137— 166. Tie, A. D., E. G. Boulding. and K. A. Naish. 2000. Polymorphic microsatellite markers for the marine gastropod Littorina subrotundata. Mol. Ecol. 9: 108-1 10. Triib, H. 1990. Zuchtung von Hybriden zwischen Viviparus ater und V. contecttis (Mollusca: Prosobranchia) im Zurichsee und okologische Untersuchungen in einer gemischten Population im Gardasee. Ph.D. Thesis, University of Zurich. Switzerland. 88 pp. Waage, J. K. 1979. Dual function of the damselfly penis: sperm removal and transfer. Science 203: 916-918. Waage, J. K. 1984. Sperm competition and the evolution of odonate mating systems. Pp. 251-290 in Sperm Competition and the Evolution of Animal Mating Systems. R. L. Smith, ed. Academic Press, New York. Walker. W. F. 1980. Sperm utilization strategies in nonsocial insects. Am. Nat. 115: 780-799. Wethington, A. R., and R. T. Dillon, Jr. 1991. Sperm storage and evidence for multiple paternity in a natural population of the freshwater snail Phvsa americana. Am. Malacol. Bull. 9: 99-102. W innepenninckx, B., T. Backeljau. and R. De Wachter. 1993. Extrac- tion of high molecular weight DNA from molluscs. Trends Genet. 9: 407. Zeh. J. A., and D. Zeh. 1994. Last-male sperm precedence breaks down when females mate with three males. Proc. R. Soc. Loud. B 257: 287-292. Reference: Bio/. Bull. 200: 268-280. (June 2001) Development of Serotonin-like and SALMFamide-like Immunoreactivity in the Nervous System of the Sea Urchin Psammechinus miliaris AMY-JANE BEER, CLAIRE MOSS*. AND MICHAEL THORNDYKEt School of Biological Sciences, Royal Holloway, University of London. Egham, Surrey TW20 OEX. UK Abstract. The present immunocytochemical study uti- lizes serotonin and SALMFamide antisera, together with confocal laser scanning microscopy, to provide new infor- mation about the development of the nervous system in the sea urchin Psammechinus miliaris (Echinodermata: Echi- noidea). Special attention is paid to the extent of the nervous system in later larval stages (6-armed pluteus to metamor- phic competency), a characteristic that has not been well described in this and other species of sea urchin. An exten- sive apical ganglion appears by the 6-armed pluteus stage, forming a complex of 10-20 cells and fibers, including discrete populations of both serotonin-like and SALMF- amide-like immunoreactive cells. At metamorphosis this complex is large, comprising at least 40 cells in distinct arrays. Serotonin-like immunoreactivity is also particularly apparent in the lower lip ganglion of 6- to 8-armed plutei; this ganglion consists of 15-18 cells that are distributed around the mouth. The ciliary nerves that lie beneath the ciliary bands in the larval arms, the esophagus, and a hith- erto undescribed network associated with the pylorus all show SALMFamide-like immunoreactivity. The network of cells and fibers in the pyloric area develops later in larval life. It first appears as one cell body and fiber, then increases in size and complexity through the 8-armed pluteus stage to form a complex of cells that encircles the pylorus. SALMF- amide-like, but not serotonin-like, immunoreactivity is seen in the vestibule wall, tube feet, and developing radial nerve fibers of the sea urchin adult rudiment as the larva gains metamorphic competency. Received 21 March 2000; accepted 9 March 2001. * Current address: Scottish Association for Marine Science, Oban, Ar- gyll. PA34 4AD, UK. t To whom correspondence should he addressed. E-mail: m.thorndyke@rhbnc.ac.uk. Introduction The presence of a nervous system in echinoid larvae was first suggested by observations of a putative thickening in the apical area (MacBride. 1914) and the presence of neu- ron-like cells in the adoral ciliary band (Mortensen, 1921). Further evidence for such a system was circumstantial, derived from observations of feeding, locomotion, settle ment, and orientation behavior. These behaviors, although relatively simple, suggested the coordinating influence of a nervous system, an idea finally confirmed by studies on ciliary activity (Mackie et al.. 1969). More recent histolog- ical and histochemical studies have provided greater details of the arrangement and fine structure of neurons in the ciliated bands and esophageal musculature of echinoderms (Burke, 1978; Nakajima, 1986, 1988; Lacalli et aL 1990). A rich quota of neurochemicals is associated with the larval nervous system, including acetylcholine, serotonin, y-aminobutyric acid (GABA), noradrenaline, and dopamine (Bisgrove and Burke, 1987; Nakajima, 1988). Such a vari- ety of agents hints at the organizational complexity of this system. Histochemical and immunocytochemical methods have been especially productive in revealing previously unknown neural structures in larvae (Nakajima, 1986; Kroll and Voronezhskaya. 1996). The discovery of the native echinoderm neuropeptides, SI (GFNSALMFamide amino acid notation) and S2 (SGPYSFNSGLTFamide). and the subsequent availability of highly specific antibody probes have provided a further means to investigate the larval and adult echinoderm nervous system. SI was originally isolated from starfish and shown to have a ubiquitous distribution in the nervous system of Asterias nibens (Elphick et al.. 199 la; Newman et al., 1995a, b). Similar peptides that seem to be members of the same family were also identified in holothurians (Diaz- 268 NEURAL DEVELOPMENT IN ECHINOIDS 269 Miranda et til., 1992). and Si-like molecules are also present in crinoids and ophiuroids (Ghyoot et til.. 1994; Candia Canievali et al., 1998). This family of peptides has also been localized in the larval nervous systems of starfish (Moss et til., 1994; Byrne et al.. 1999) and sand dollars (Thorndyke et til.. 1992), revealing subsets of neurons not previously identified with other probes. Novel neurochemical markers (such as peptide antisera) can help us complete our picture of the larval and adult nervous systems and, perhaps, add supporting data to other studies of the physiological function of the molecules. That is, information about the location of neurochemicals within a cell, within a cell type, or within a tissue type can some- times help to indicate whether these molecules may be involved in motor or sensory activities or may be neuro- transmitters or neuromodulators. In echinoderms, neuro- chemical markers, when expressed at various developmen- tal stages, can help us visualize the origin, growth, and fate of particular neural structures. The aim of the present study was to map the development of the larval echinoid nervous system using immunocyto- chemical techniques, with established commercial (seroto- nin) and novel (SALMFamide) antisera serving as probes. We used Psammechinus miliaris, a small regular sea urchin of the class Echinoidea, which is common along the north- east Atlantic and North Sea coasts of the United Kingdom and Scandinavia. The development of P. miliaris has never been fully described in the literature, although we and other researchers (e.g., Kelly et al., 2000) have found it to be similar to that of other species of indirect developing sea urchins, passing through embryo, prism, and 4-, 6-. and 8-armed plutei stages to metamorphic competency (MacBride, 1913, 1914, 1918). In the current work, empha- sis has been placed on the later larval stages, which are less well described by immunocytochemical studies. Confocal microscopy has been used to facilitate the accurate local- ization of larval neurons, to identify previously unknown components of the nervous system, and to improve the resolution of the cells and fibers. Materials and Methods Animal maintenance and lan'til culture Specimens of Psammechinus miliaris were obtained from the University Marine Biological Laboratory at Millport, Isle of Cumbrae, and from Dunstaffnage Marine Labora- tory, Oban, on the west coast of Scotland. They were maintained in circulating seawater systems at Royal Hollo- way or in Oban and fed on Viva lactnca. Mytilus etlulis, and an artificial food consisting of soya meal, fish oil, and spinach. Males and females could generally be distin- guished by slight differences in the appearance of their gonopores. Animals were spawned by the injection of 0.5 M KC1 in filtered seawater (FSW) into the area of the gonads. with the needle inserted through the peristomial membrane (Iwate and Fukase, 1964). Animals were placed in glass beakers for about 30 min. or until they spawned. Gametes were collected and eggs fertilized by standard methods (Strathmann, 1987). After hatching, larvae were maintained in autoclaved seawater in glass jars and gently stirred to prevent them from gathering at the bottom of the vessel. Once the gut had developed, the larvae were fed Duntiliella tertiolecta so that algae could always be seen in the gut. The larval cultures were kept clean by replacing one-third of the water three times a week. Cultured algae were separated from the growth medium and resuspended in FSW before being fed to the larvae. To prevent overcrowding, the con- centration of larvae in each jar was kept at about 1 per 5 milliliters of FSW. Animals were thus maintained through- out larval development, until competency. Competent lar- vae with a large, well-developed rudiment were induced to settle by the introduction of algal-fouled glass plates from the aquarium. Similarly, larvae would also often settle and metamorphose on the bottom of the culture vessels, if these were allowed to become fouled with algal debris. Fixation Samples of larvae were removed from culture and fixed at regular intervals, ensuring that each stage (4-armed. 6-armed. and 8-armed plutei) was adequately represented. At least 100 larvae per sample were fixed at earlier stages. This reduced to 20 per sample for 8-armed animals with developed rudiments. Larvae were anesthetized in 4% MgCU in FSW or killed by the addition of a few drops of 4% paraformaldehyde (PFA) in FSW. Excess seawater was then pipetted off. and the samples were fixed in 4% PFA in FSW for 2 h. For storage, animals were rinsed in FSW and stored in FSW containing 0.01 % NaN3 at 4°C. Immunocytochemistry Whole-mount incubations were carried out in 2-ml glass tubes at room temperature. Larvae were rinsed in phosphate buffered saline (PBS) and permeabilized by a 30-min incu- bation in 0.5% Triton-X 100 in PBS. followed by 30 min in 5% normal goat serum in PBS. To improve antiserum penetration, and thus the staining of the rudiment in older larvae, a further permeabilization step was carried out. After fixation, larvae were rapidly dehydrated to 70% ethanol, rehydrated, and incubated in 0.5% Triton-X 100 in PBS for 1-2 h. Primary antibodies were diluted in PBS, and the larvae were incubated in this solution for 16-18 h. The serotonin antiserum (SeraLabs, UK) was raised in rabbit against synthetic serotonin and used at 1:200 in PBS. All results with this antibody are referred to as serotonin-like immunoreactivity (5HT-Li IR). The peptide antisera were raised in rabbits and produced in this laboratory; they are all polyclonals. Anti-Si (BLIV/anti-KYSALMFa) or anti-Si 270 A.-J. BEER ET AL. (BLIIIa/anti-KYSALMFa) (Elphick et ui, 1991a, b) were used at 1:200 in PBS, which gave clear staining with low background. These antisera are two consecutive bleeds from one animal and will be referred to as Si-like immunoreac- tivity (SI -Li IR). Some 6-armed plutei were also stained for S2 with a new polyclonal raised in this laboratory (Nl/anti- KYSGLTFa) (Potton. 1997), which produced preparations of high specificity and low background levels. Results with this antiserum are referred to as S2-like immunoreactivity (S2-L1 IR). After overnight incubation in primary antisera, larvae were washed in PBS, three times each for 5 min. Visual- ization was by a 1-h incubation in biotinylated anti-rabbit IgG (1:200 in PBS) (Vector Labs, Peterborough, UK), rins- ing in PBS, and then a final incubation for 1 h in avidin- conjugated FITC or Texas red (1:100 in PBS) (Vector Labs). Larvae were rinsed in PBS and mounted in Vecta- shield antifade mountain (Vector Labs) and viewed on a Zeiss Axioplan with fluorescent attachments, or on a Leica TCS-4D confocal laser scanning microscope (CLSM). Controls were produced by preabsorption of twice the working concentration of primary antiserum with an equal volume of 2 X 10 ^M KYSALMFa, KYSGLTFa, or 5HT (Polak and Van Noorden, 1986). This reduced staining to background levels in all cases, although it was not com- pletely removed. Preadsorption controls revealed weak cross-reactivity between the peptide antisera where pread- sorption of anti-Si with S2 did not affect staining as much as did preadsorption of anti-S2 with S 1 . Further controls for secondary antibodies were carried out by replacing the primary with PBS, or by using direct visualization with FITC- or Texas red-conjugated secondary antibodies, which controlled for reactions with endogenous biotins. Results For the purposes of this study, the development of Psam- mechinus miliaris was divided into five stages: prism (2 to 3 days post-fertilization), 4-armed, 6-armed, and 8-armed plutei (from 3 to 20 days), and competent plutei (from 20 to 30 days). The adult rudiment began to form at the 6-armed stage, and was well developed in the late 8-armed pluteus. In cultures of P. miliaris the total development period, from egg to new juvenile, was between 20 and 30 days. Figure 1 shows the key features of the pluteus nervous system and summary diagrams of the results. Figure 1A depicts the extent .of the larval nervous system as it is known from previous immunocytochemical studies. Figure IB shows the more extensive system revealed by the present immunocy- tochemical study. Observations of living larvae revealed distinctive cell types associated with the sea urchin larval nervous system. I 'se observations, together with previous descriptions of neuronal cell types in these areas (Burke, 1978; Nakajima, 1986; Lacalli et ai, 1990), assisted us in identifying those cells in P. miliaris that were stained by immunocytochem- ical methods. In late prism larvae (2-3 days post-fertiliza- tion), the ciliated band could be clearly observed. The cells of the blastocoelar network were also visible (Fig. 2 A). These cells were previously thought to be part of the ner- vous system, but subsequent work has shown that they are not neural but have some other function, possibly associated with muscle activity (Burke, 1978). Neuron-like cells within the ciliary bands were visible; their characteristic shape, with a thick apical process, suggested they might be sensory cells (Fig. 2B, C). Fine basal processes contribute to a tract of fibers beneath the ciliary band, forming part of the ciliary nerve (Fig. 2C), which then putatively connects to other larval structures. The ciliary band and its associated neurons and fibers elongates as the larval arms develop and form the main neural pathways of the larva. Along with the apical ganglion, lower lip ganglion, and neural components of the gut, these structures constitute the main neuronal elements of the sea urchin larval nervous system (Fig. 1 ). The immunocytochemical studies focused on later pluteal stages, which have been less thoroughly described in pre- vious investigations (Bisgrove and Burke, 1987; Thorndyke ct ai, 1992). These studies enabled a detailed examination of the fully developed apical ganglion and adult rudiment. Development of immunoreactivity in plutei to the 8-armed stage The earliest cellular labeling observed in young plutei was a single cell within the thickened lower lip of the stomatodeum, which was strongly immunoreactive to the serotonin antibody (Fig. 3A). This cell was seen in the most precocious larvae at about 72 h post-fertilization. The cell has a basal unlabeled nucleus and faintly immunoreactive basal processes that extended beneath the lip. There was also a suggestion of labeled cells in the ciliated band of the anterolateral arm buds. There was no comparable labeling with the other antisera in larvae of the same age or in control specimens that received only primary antibody or an inap- propriate secondary antibody. At the fully developed 4-armed stage (4 or 5 days post-fertilization) the lower lip contains several cells with 5HT-Li IR in a symmetrical arrangement (Fig. 3B). A similar pattern of labeled lower lip cells is also apparent in 4-armed specimens stained with anti-Si, although these cells are comparatively large (15 jurn), and less numerous. At this stage, no SI -Li IR cells were seen in the apical thickening. The larval stomach contains algae and is autofluorescent. By the 6- to 8-armed pluteus stage, the number of labeled cells in the lower lip complex is increased, with up to 18 neurons showing 5HT- and SI -Li IR. When viewed from the side, these cells have a characteristic shape, with broad bases against the esophageal epithelium and narrower api- NEURAL DEVELOPMENT IN ECHINOIDS 271 Figure 1. Diagrams of pluteus larvae. (A) Previously demonstrated sites of innervation (solid fill) in a sea urchin larva (Nakajima. 1986; Bisgrove and Burke, 1987; Thorndyke et ai. 1992). (B) The extent of the nervous system, as revealed in this study. (C) The main areas of S 1-Li IR (gray) in the early 4-armed. early 6-armed. and 8-armed pluteus. (D) The mam areas of S2-Li IR (gray) in the 6-armed pluteus. (E) The main areas of 5HT-L1 IR (gray) in the early 4-armed. early 6-armed. and 8-armed pluteus. clearly showing its predominantly ventral distribution, a-apical ganglion, cb-ciliary band, ce-ciliary epaulette, en-ciliary nerve. 11-lower lip. m-mouth. pp-pyloric plexus, r-rudiment. s-stomach. ul-upper lip. Not to scale. ces at the ectoderm of the inner edge of the oral hood. From the 6-armed stage onwards the apical complex begins to increase dramatically in both size and complexity, whereas the lower lip complex expands no further. Indeed, the 5HT- Li IR appears to weaken at later stages. The row of small cells first seen in the upper lip at the 4-armed stage alters little during subsequent development. In later plutei, 5HT-Li IR cells in the upper lip could be seen contributing processes to the expanding apical gan- glion (Fig. 4A-D). This association is close enough at this stage for the apical and upper lip complexes to be consid- ered part of the same ganglion. The apical complex becomes the predominant site of both 5HT and SI -Li IR neurons, as the 4- and 6-armed plutei progresses to a full 8-armed stage. 272 A.-J. BEER ET AL. Figure 2. Observations of live larvae. (A) Multipolar cells of the blastocoelar network in the oral hood of an early 4-armed pluteus (arrows). (B) Higher magnification of (A) showing a flask-shaped bipolar neuron with a tapering apical process (arrowhead) and basal varicose axon contributing to the ciliary nerve (arrows). (C) Diagram of (B) with the neuron shown in black (no! to scale). Bars: A = 25 /xm, B = 10 ju.ni. NEURAL DEVELOPMENT IN ECH1NOIDS 273 Figure 3. Immunoreactivity in the lower lip ganglion. (A) Late prism larva with a single serotonergic cell in the lower lip (arrowl. (B) Lower lip ganglion with 10-12 cells showing 5HT-Li IR (arrow). Bars = 50 /nm. with cells being added rapidly through this period (Fig. 4A, B. D). In fully developed 6- and 8-armed plutei, the differences in distribution between peptide and serotonin immunoreac- tivity become more apparent. The ciliary nerves show S2-Li IR, rather than SI -Li or 5HT-L1 IR (Fig. 5 A). These axons possess varicosities and some immunoreactive cell bodies associated with the descending loops of the ciliary bands (Fig. 5B, C). The neuronal cell bodies are generally multi- polar, rounded or elliptical, and with large nuclei (Fig. 5B). Unlike the cells of the apical ganglion, these cells have no obvious sensory processes or contacts with ectodermal sur- faces. No 5HT-Li IR was found in these areas. Development of immunoreactivity in late plutei (8-anned stage to metamorphosis) As the plutei grow and the rudiment forms, changes in the nervous system or in the distribution of neurotransmitters become more difficult to observe with traditional epifluo- rescence microscopy. However, scanning confocal laser mi- croscopy allowed us to examine these changes in detail, especially in relation to the apical ganglion complex, which became very large and convoluted in older larvae. The apical ganglion continues to increase in size and complexity until the mature 8-armed stage. In these older larvae, it is as much as 80 /nm in length (left to right side of the larva) and 60 /u,m deep, extending from the thickened apical ciliated band down to the upper lip. To resolve this cluster of perikarya and dense neuropile, rotations were calculated for three-dimensional confocal images (Fig. 6). Some double-labeling studies were attempted, but due to the large size of the specimens and the large number of cells in the ganglion, the results obtained were not clear and have therefore not been presented here. However, confocal im- ages of separately stained preparations were informative and suggested that the distributions of S 1 and 5HT-LJ IR within the apical ganglion differed significantly. Some cells may have expressed both substances, and there may have been other cell types present that expressed neither, but overall the rotational images provided the most comprehen- sive view of this structure to date. In mature larvae, about 24 cell bodies in the apical ganglion show 5HT-Li IR. These are arranged symmetri- cally on either side of a dense central plexus, with all perikarya appearing to contribute fibers to the plexus or the subciliary axon tract (Fig. 6A-C). Specific pairs of cells are distinct and therefore can be recognized from one larva to the next. Three major nerves emerge on each side of the plexus. The first forms a descending loop beneath the cili- ated band before leading up into the antero-lateral arm, the second extends around the larval mouth, and the third ap- pears to extend into the larval body on either side of the esophagus (Fig. 4D). The smaller 5HT-Li IR nerves in the larval arms may also feed into the apical ganglion, although the precise termination of these fibers was not determined. The overall shape of the serotonergic elements in the apical ganglion was characterized by cell bodies arranged anterior and posterior to the central plexus, but never very far ventral or dorsal to it (Fig. 6A-C). Therefore, it appears as a rather two-dimensional structure in confocal rotations. The bilateral symmetry of the SI -Li IR components of the apical ganglion is less pronounced than that of the 5HT-Li IR components. Cell bodies are arranged along the length of the ganglion, rather than in clusters at either end (Fig. 6D-F). All views of the ganglion appear to show a variety of cell shapes. Some are flask-shaped, with tapering apical processes typical of sensory cells, whereas others are more rounded. Upon rotation, the rounded cells appear as i .; - - - - : , . . -• - " - ' -. - -. -: il-Li IR was die posterior end of the stomach at the pylorus (Fig. A i and is associated with a distinctive elongate cell body (Fig. ~B I. The fiber is about 1 /tin in diameter with 3 /im varicosities. and the bipolar cell body is 10-15 /im long. .An optical section through the pylorus shows SI -Li IR on either side, although the fiber does not appear to be a ring. This cir- cumpy loric nerve proves to be the origin of a nerve complex that increases in size throughout the rest of larval life. NEURAL DEVELOPMENT IN ECfflNOIDS Figure 5. s"-? i IB accnrij«pd wilh riv rifany hxarl in a ft-amic-d tarra TV-gyri girti onnftical bar «£»•» microscoc;. -I •_. IR varicose axoos oodoiyiiE ifae cffiaed bad (arrow I aad u •muling H> larse T_T.. -:_-• . . - ' '_ -- " : :- - - - - ' : . : " - : " - "" •-" : _r.:e: „ . . _ - . ; - _ _- . > 1-1 - - ; ; : _- _ - - . . - : anns. as well as inn die cu:_- ; ^rmc) of die pest-oral aons. Bas: A ad C = 50 urn. B = 10 foa. forming a loose w eb of fibers around much of the distal end of the stomach. Several days after the appearance of the first pyloric nerve, a second SI -Li IR neuron appears with a similar form to the first • F-.:: ~C N: :re neurons are grad- ually added throughout development, and this complex con- tinued to grow even in mature larvae where metamorphosis K Delayed, apparently due to nutritional limitations For example, in 50-day-old larvae, the - ric naoral . mplex consists of numerous cells and a plexus of fibers iFig. 7Di. No 5HT-Li IR was observed in this area. : •-...: : ' . •'.. : :. In early attempts to label die adult rudiment, the larval complexes were clearly immunoreacrive. bat nothing could be seen in the rudiment itself. Initial interpretations were that nMmpm? in die rudiment either had not differentiated or did not express neuropeptide. However, semimin tohridme blue sections demonstrated die presence of developing ra- dial nerve cords and die circumoral nerve ring. Therefore. die penneabilizarion procedure was Hindifiral in a 276 A.-J. BEER ET AL. Figure 6. Representative drawings of 5HT and SI -Li IR neurons in the apical complex, andconfocal images of the apical ganglion showing selected rotations from a full 360° series. (A) to (C) show the serotonergic cells of the apical ganglion. (A) Diagram of the apical ganglion viewed from the anterior/ventral surface of the larva, as deduced from the three-dimensional rotations. Not to scale. (B) 0° rotation. (C) 65° rotation. (D) to (F) show the SALMFamidergic cells of the apical ganglion. (D) Diagram of the SALMFamidergic cells of the apical ganglion as deduced from three-dimensional rotations. Not to scale. The ganglion is bilaterally symmetrical, apparently with pairs of cells (arrows). (E) 0° rotation. (F) 50° rotation; indicates the nerve plexus. Bars = 25 to confirm unequivocally that the absence of immunoreac- tivity is due to the lack of antigen. This test still revealed positive labeling in several structures of the rudiment. The first appearance of SALMFamide-like immunoreac- tivity in the rudiment of the larvae was in the ectodennally derived lining of the vestibular cavity (Fig. 8 A). Some SI -Li IR appeared in the covering of the primary podia, which is also ectodermal in origin. The earliest SI -Li IR in the circumoral and radial nerve cords was rather diffuse. In later specimens competent to metamorphose (the primary podia were observed to move independently), the immuno- reactivity was more concentrated, occurring in well-defined fiber tracts in the rays of the water vascular system, with the nerve ring in the tube feet clearly visible (Fig. 8B, C). No 5HT-Li IR was observed in the rudiment. Discussion The larval nen'ous system This study describes the neural development of Psammechi- nus miliaris, from the 4-armed pluteus stage through to a metamorphically competent larvae with 8 arms and a devel- oped adult rudiment. Most important, it extends the previously available descriptions of the echinoid larval nervous system to the late pluteus. and includes such previously unknown fea- NEURAL DEVELOPMENT IN ECHINOIDS 277 Figure 7. Si-Li IR >n the pyloric neural complex. (A) 8-armed larva showing the pylorus. (B) Confocal image of the first immunoreactive cell (c) and fiber (arrow) in the complex in a 6-armed larva (10 days post- fertilization). (C) Two immunoreactive cells are present in the pyloric area by 16 days post-fertilization. (D) A plexus of cells develops in mature larvae, especially those which have delayed metamorphosis. Bars: A = 100 /j,m. B to D = 50 jam. Figure 8. SI -Li IR in the adult rudiment. (A) Imrmi loreactivity can be seen in the apical ganglion of the larva (arrowhead) and in the vestibule wall of the rudiment (arrow) of a 27-day-old larva. (B) The earliest evidence of S 1 -Li IR in the developing radial nerves. Fibers in the neuro- epithelium underlying the epineural space lead into the primary podia (arrows). (C) Si-Li IR shows the radial nerve cords (arrow) terminate in the podial rings of the primary tube feet (arrowhead). Bars = 100 /xm. 278 A.-J. BEER ET AL tures as the pyloric plexus and details of the nervous system within the adult rudiment. It also provides the fullest descrip- tion to date of the apical ganglion in mature larvae. This ganglion is a complex structure containing numerous cell types, with connecting fibers to other organs within the larval body, including the adult rudiment. The structure of certain elements of the echinoderm larval nervous system has been previously determined by ultra- structural, histological, and histochemical studies, revealing a network of cell bodies and fibers that are regionally concentrated to form ganglia (Burke, 1978; Nakajima, 1986; Bisgrove and Burke, 1987; Lacalli et at.. 1990; Moss etal., 1994; Byrne et al., 1999; Chee and Byrne, 1999). The ciliary bands are a major component of the larval nervous system and contain numerous nerve cells of several types. Other components include localized specializations such as in the lower lip, the apical (or dorsal) ganglion, and an underlying major nerve tract, the ciliary nerve. Immunocy- tochemical studies have been especially useful in providing details of the overall structure of this system; in the absence of a single unambiguous marker, previous authors have combined information from several immunological markers to construct a neural map. Our latest neural map of echinoid larvae is one further step towards a better understanding of the structure and function of the echinoderm nervous sys- tem. The development of the nervous system of P. miliaris correlates with changes in larval form, function, and behav- ior. Additional neurons often appear in association with newly formed or enlarged larval organs. For example, the ciliated band continues to grow in length until the fourth and final pair of arms, the pre-oral arms, are formed. According to Nakajima (1986), this band also increases in width, the number of rows of ciliated cells increasing with age, espe- cially in the epaulettes. The nerves underlying the ciliated band elongate as it grows, and additional immunoreactive cell bodies appear. The apical ganglion continues to grow as larvae approach competency, with the addition of immuno- reactive neurons and an associated increase in complexity in the neuropile and ciliary nerves surrounding the area. We show that neural structures associated with the pylorus continue to develop, even in advanced larvae. This contin- ual addition to the larval nervous system for the greater part of larval life suggests an increased behavioral repertoire, possibly specifically in association with the development of the adult rudiment and the timing of metamorphosis. The development of 5 HT and SALMFamide immunoreactivity The early development of 5HT-L1 IR in P. miliaris fol- lows a pattern seen in other studies on echinoids (Bisgrove and Burke. 1987). Immunoreactive cells appear, and the number of stained cells increases as the larva begins to swim and feed. 5HT-Li IR is restricted to those components of the nervous system in the ventral half of the larva that are considered to be sensory — in particular, the apical ganglion and flask-shaped cells in the ciliary bands. The suggestion of a predominantly sensory function for 5HT in P. miliaris is in keeping with other studies on larval echinoderms (Bisgrove and Burke, 1987; Nakajima, 1988; Moss et al., 1994), molluscs (Kempf et al., 1997; Marois and Carew, 1997), and phoronids (Hay-Schmidt, 1990). The role of 5HT may be associated with the detection of chemical cues, possibly those required to induce feeding or settlement and metamorphosis. However, the absence of 5HT in the rudi- ment and juvenile may also be significant. The podia of the rudiment have been implicated in substrate choice by meta- morphosing sea urchins (Cameron and Hinegarder, 1978), and no 5HT was detected in these structures in the current study. This suggests that if 5HT is involved in settlement behavior, it may be as part of a pathway that integrates both the larval and juvenile nervous systems. The development of SI -Li IR follows a pattern similar to that of 5HT, and its temporal appearance can also be seen to coincide with increased locomotory activity and feeding. Its final distribution is wider than that observed for 5HT, with numerous cells and fibers throughout the larval body, as well as in the newly described pyloric plexus. The shape and location of cells in the pyloric plexus and subciliary com- plexes is highly suggestive of a motor function, whereas the tapering apical processes that bristle around the apical gan- glion may represent sensory receptors. This mixed function has also been suggested by other studies in which SI has been localized (Thorndyke et at.. 1992; Moss et a!.. 1994), and it is possibly an early indicator of their ubiquitous distribution in adult echinoderms (Newman. 1995a, b). It is interesting that the preliminary studies with anti-S2 revealed staining only in the ciliary bands and a few neuronal cell bodies, but none in the ganglia or gut. This may suggest a purely motor function for S2 in coordinating ciliary activity and is again in keeping with the more limited distribution of this peptide in adult echinoderms (Newman, 1995a, b). As the work on S2 was only carried out on one developmental stage, we cannot draw definitive conclusions about its over- all distribution. However, it does provide a clear compari- son to the SI distribution and reinforces the likelihood that these two related peptides have very distinct functions in both larvae and adults. Co-localization of SALMFamide peptides and 5HT was not examined. The distributions revealed in this study clearly point towards discrete populations of cells express- ing a single neurotransmitter. However, comparison of the two staining patterns also suggests that co-localization within specific cells of the ganglia is possible. Indeed, SI has an extensive distribution in larval echinoderms and is probably of equal importance to larval function as 5HT. Equivalent patterns of neuropeptide localization in other NEURAL DEVELOPMENT IN ECHINOIDS 279 marine larvae have been reported tor the molluscan neu- ropeptide FMRFarnide, which is also found in central larval ganglia and peripheral nerves in larval phoronids (Hay- Schmidt, 1990). Otherwise the reporting of neuropeptide localization in the ganglia of marine invertebrate larvae is rare, and we are not yet able to draw any conclusions about their putative roles. The apical ganglion The apical ganglion is the major neural complex in the pluteus and probably has a central coordinating role. An apical neural structure, with varying degrees of complexity, has been implicated in the coordination of swimming, feed- ing, and settlement behavior in many marine invertebrate larvae, including molluscs (Chia and Koss. 1984; Marois andCarew, 1997: Kempt' et al., 1997), polychaetes (Lacalli. 1981), and echinoderms (Burke. 1983; Chia et al.. 1986; Nakajima et al., 1993). This structure appears to be con- served in these larvae, and a consistent feature is the pres- ence of 5HT-Li IR neurons, usually in a bilaterally sym- metrical arrangement. In the present study, CSLM has been used to trace the development of the apical ganglion and to analyze its struc- tural complexity. At the 8-armed larvae stage it is by far the most complex neural structure in the larva, with an array of immunopositive cells and possibly other as yet undetected neurons. Although it appears highly developed, it presum- ably degenerates completely at metamorphosis, as it appears to coordinate only larval functions. In sea urchins, therefore, the entire larva seems to be under the control of a neces- sarily complex apical ganglion, which supports its functions in totality. In other words, the larval and juvenile nervous systems have been thought of as being almost totally dis- tinct. The apical ganglion may have a wider significance as a precursor of the chordate nervous system (Lacalli, 1994). Cell types within the larval nervous system have been compared with those in other phyla in an attempt to find homologous structures, implicating the larval nervous sys- tem, and especially the apical ganglion, as phylogenetically significant structures. The extensive ganglion revealed in 8-armed P. miliaris larvae may help shed further light on this possibility. Echinoderms, as deuterostome inverte- brates, are useful for tracing such cells in chordates to their putative invertebrate ancestors. For example, Lacalli and West ( 1993) propose that the distinctive multipolar cells of echinoderm circumoral ciliary bands are homologous with similar neurons in the ciliary bands of the hemichordate tornaria, as well as with those in the dorsal nerve cord of ammocoete larvae of the lancelet Branchiostoma. Development of immunoreactivity in the rudiment, and the fate of the larval nervous system SALMFamide probes have revealed the presence of neu- ropeptides in the rudiment of the P. miliaris pluteus, indi- cating that adult neural development and differentiation is well under way during larval stages. This is the first immu- nocytochemical visualization of nerve cells in the develop- ing rudiment and shows the clear separation of the larval and juvenile systems. The discovery of the pyloric plexus in the older larvae is of particular interest since it has not been previously de- scribed. The structure continues to grow in mature larvae (and may continue to grow after metamorphosis), whereas other larval neural components have stopped growing, or may even be diminishing. These observations suggest that the pyloric plexus neurons may be incorporated into the adult basi-epithelial gut plexus. This is perhaps not surpris- ing, since this area of the gut is incorporated into the rudiment along with associated nerves. Such a finding would indicate that some larval neural tissue is carried into the juvenile, with the major part of this system being resorbed at metamorphosis. However, the larval pyloric plexus and that of the juvenile may also be distinct struc- tures that simply show expression of the same neuropeptide. Further studies focusing on this specific area may reveal more details of the metamorphic process and the relation- ship between larval and adult forms. Acknowledgments This work was supported by a BBSRC studentship No. 93309321 to A.-J.B. Thanks to Zyg Podhorodecki and An- ton Page for photographic and confocal microscopy assis- tance. Also thanks to Maeve Kelly, Scottish Association for Marine Science, for help with culturing larvae. Literature Cited Bisgrove, B. W., and R. D. Burke. 1987. Development of the nervous system of the pluteus larva of Strongylocentrotus droebachiensis. Cell Tissue Res. 248: 335-343. Burke. R. D. 1978. The structure of the larval nervous system of the pluteus larva of Strongylocentrotus purpuratus. Cell Tissue Res. 191: 233-247. Burke. R. D. 1983. Neural control of metamorphosis in Dendraster excentricus. Biol. Bull. 164: 176-188. Byrne, M., F. Chee, P. Cisternas, and M. C. Thorndyke. 1999. Lo- calisation of the neuropeptide SI in the larval and adult nervous system of the sea star Patiriella regularis. Pp. 187-191 in Echinoderm Re- search 1998, M. D. Candia Carnevali and F. Bonasoro. eds. Balkema. Rotterdam. Cameron, R. A., and R. T. Hinegarder. 1978. Early events in sea urchin metamorphosis, description and analysis. J. Morphol. 157: 21-31. Candia Carnevali. M. D., F. Bonasoro, U. Welsch, and M. C. Thorndyke. 1998. Arm regeneration and growth factors in crinoids. 280 A.-J. BEER ET AL. Pp. 145-150 in Echinoderms: San Francisco. R. Mooi and M. Telford, eds. Balkema. Rotterdam. Chee, F., and M. Byrne. 1999. Development of the larval serotonergic nervous system in the sea star Patiriella regularis as revealed by confocal imaging. Biol. Bull. 197: 123-131. Chia, F.-S., and R. Koss. 1984. Fine structure of the cephalic sensory organ in the larva of the nudibranch Roslanga pulchra. Zoomorphology 104: 131-139. Chia, F.-S., R. D. Burke, R. Koss, P. V. Mladenov, and S. S. Rumrill. 1986. Fine structure of the doliolaria larva of feather star, Florometra serratissima, with special emphasis on the nervous system. J. Morphol. 189: 99-120. Diaz-Miranda, L., D. A. Price, M. J. Greenberg, T. D. Lee, K. E. Doble, and J. E. Garcia- A rraras. 1992. Characterization of two novel neuropeptides from the sea cucumber, Holotlntria glaberrina. Biol. Bull. 182: 241-247. Elphick, M. R., R. H. Emson, and M. C. Thorndyke. 1991a. Isolation of the neuropeptide SALMFamide-1 from starfish using a new anti- serum. Pepiides 12: 455-459. Elphick, M. R., D. A. Price, T. D. Lee, and M. C. Thorndyke. 1991b. The SALMFamides: a new family of neuropeptides isolated from an echinoderm. Proc. R. Soc. Loiul. B 243: 121-127. Ghyoot, M., J. L. Cobb, and M. C. Thorndyke. 1994. Localisation of neuropeptides in the nervous system of the brittlestar Ophiura ophiura. Philos. Trans. R. Soc. Land. B 346: 433-444. Hay-Schmidt, A. 1990. Distribution of catecholamine-containing, 5HT- like and FMRFamide-like 1R neurons and processes in the nervous system of the actinotroch larvae of Phoronis miielleri. Cell Tissue Res. 259: 105-118. Iwata, K. S., and H. Fukase. 1964. Comparison of discharge of the gametes by three artificial means in sea urchins. Biol. J. Okayama Univ. 10: 57-64. Kelly, M. S., A. J. Hunter, C. L. Scholheld, and J. D. McKenzie. 2000. Morphology and survivorship of larval Psammechinus miliaris (Gme- lin) (Echinodermata: Echinoidea) in response to varying food quantity and quality. Aquaculture 183: 223-240. Kempf, S. C., L. R. Page, and A. Pires. 1997. Development of seroto- nin-like immunoreactivity in the embryos and larvae of nudibranch molluscs with emphasis on the structure and possible function of the apical sensory organ. J. Comp. Neural. 386: 507-528. Kroll, R. P., and E. E. Voronezhskaya. 1996. Early elements in gas- tropod neurogenesis. Dev. Biol. 173: 344-347. Lacalli, T. C. 1981. Structure and development of the apical organ in trochophores of Spirobranchus polycerus. Phy/Iodoce maculata and Phyllodoce mucosa (Polychaeta). Proc. R. Soc. Land. B 212: 381-402. Lacalli, T. C. 1994. Apical organs, epithelial domains and the origins of the chordate central nervous system. Am. Zool. 34: 533-541. Lacalli, T. C., and J. E. West. 1993. A distinctive nerve cell type common to diverse deuterostome larvae: comparative data from echi- noderms, hemichordates and amphioxus. Ada Zool. 74: 1-8. Lacalli, T. C., T. H. Gilmour, and J. E. West. 1990. Ciliary band innervation in the bipinnarian larva of Pisaster ochraceus. Philos. Trans. R. Soc. Land. B 330: 371-390. MacBride, E. W. 1913. The development of Echinocardium cordatum I. External features of development. Q. J. Microsc. Sci. 59. MacBride, E. W. 1914. ,4 Textbook of Embryology. Macmillan Press, London. MacBride, E. W. 1918. The development of Echinocardium cordatum II. Q. J. Microsc. Sci. 63. Mackie, G. O., A. N. Spencer, and R. Strathmann. 1969. Electrical activity associated with ciliary reversal in an echinoderm larva. Nature 223: 1384-1385. Marois, R., and T. J. Carew. 1997. Fine structure of the apical ganglion and its serotonergic cells in the larva of Aplysia califomica. Biol. Bull. 192: 388-398. Mortensen, T. H. 1921. Studies on the development and larval forms of echinoderms. G.E.C. Gad, Copenhagen. Moss, C., R. D. Burke, and M. C. Thorndyke. 1994. Immunocyto- chemical localisation of the neuropeptides SI and serotonin in larvae of the starfish Pisaster ochraceus and Asterias rubens. J. Mar. Biol. Assoc. UK 74: 61-71. Nakajima, V. 1986. Development of the nervous system of sea urchin embryos: Formation of the ciliary bands and the appearance of two types of ectoneural cells in the pluteus. Dev. Growth Differ. 28: 531-542. Nakajima, V. 1988. Serotonergic nerve cells of starfish larvae. Pp. 235-239 in Echinoderm Biology. R. D. Burke. P. V. Mladenov, P. Lambert, and R. L. Parsley, eds. Balkema, Rotterdam. Nakajima, V., R. D. Burke, and Y. Noda. 1993. The structure and development of the apical ganglion in the sea urchin pluteus larvae of Strongylocentrotus droebachiensis and Mespilia globulin. Dev. Growth Differ. 35: 531-538. Newman, S. J., M. R. Elphick, and M. C. Thorndyke. 1995a. Tissue distribution of the SALMFamide neuropeptides SI and S2 in the starfish Asterias rubens using novel monoclonal and polyclonal anti- bodies. I. Nervous and locomotory systems. Proc. R. Soc. Land. B 261: 139-145. Newman, S. J., M. R. Elphick, and M. C. Thorndyke. 1995b. Tissue distribution of the SALMFamide neuropeptides SI and S2 in the starfish Asterias rubens using novel monoclonal and polyclonal anti- bodies. II. Digestive system. Proc. R. Soc. Loud. B 261: 187-192. Polak, J. M., and S. Van Noorden. 1986. Immunocylochemistry. Mod- ern Methods and Applications. John Wright and Sons, Bristol, United Kingdom. Potton, D. 1997. Neuroendocrine control of feeding in the European starfish, Asterias rubens. Ph.D. thesis. University of London, United Kingdom. Strathmann, M. F. 1987. Reproduction and development of marine invertebrates from the northern Pacific coast: data and methods for the study of eggs, embryos and larvae. University of Washington Press, Seattle. Thorndyke, M. C., B. D. Crawford, and R. D. Burke. 1992. Local- isation of a SALMFamide neuropeptide in the larval nervous system of the sand dollar. Dendraster excentricus. Achi Zool. 73: 207-212. Reference: Bio/. Bull. 200: 281-297. (June 2001) Function and Functional Groupings of the Complex Mouth Apparatus of the Squat Lobsters Munida sarsi Huus and M. tenuimana G.O. Sars (Crustacea: Decapoda) A. GARM* AND J. T. H0EG Department of Zoomorpkology, Zoological Institute, University of Copenhagen. Universitetsparken 15, 2100 Copenhagen, Denmark Abstract. Like all other decapods, the anomuran squat lob- sters Munida sarsi and M. tenuimana have a mouth apparatus composed of six pairs of mouthparts plus labrum and para- gnaths (upper and lower lips). To study the functional signif- icance of this complexity, we examined the mouthparts with scanning electron microscopy and also observed their function directly, under laboratory conditions, using macro-video equipment. No differences were found between the two spe- cies. The movement patterns of the mouthparts are described in detail and illustrated as serial drawings. Proceeding from max- illipeds 3 towards the mandibles, the movement pattern gets increasingly stereotypical, with the mandibles performing but a single movement in a medio-lateral plane. From morphology, the mouthparts are subdivided into 20 parts, but from the functional analyses the 20 parts form 8 functional groups: /, transporting mouthparts (maxilliped 2 endopod and maxil- liped 3 endopod); 2, transporting— aligning mouthparts (max- illiped 1 basis); 3, sorting-aligning mouthparts (maxilla 1 basis and maxilla 2 basis); 4, current- generating mouthparts (flagella of maxilliped 2 and maxilliped 3 exopods); 5, cutting- crushing mouthparts (incisor and molar processes, labium, and mandibular palp); 6, ingesting mouthparts (maxilla 1 coxa, maxilla 2 coxa, and maxilliped 1 coxa); 7. respiratory nunith- Received 2 February 2000; accepted 12 January 2001. * To whom correspondence should be addressed. E-mail: Algarm@zi.ku.dk Abbreviations: Lb. labrum; Inc. incisor process, Mp, molar process; Mdp, mandibular palp: MX I cox, maxilla 1 coxa; Mxl bas. maxilla 1 basis; Mxl endo, maxilla 1 endopod; Mx2 cox. maxilla 2 coxa; Mx2 bas, maxilla 2 basis; Mx2 endo, maxilla 2 endopod; Scapho. scaphognathite; Mxpl cox, maxilliped 1 coxa; Mxpl bas, maxilliped 1 basis; Mxpl endo, maxilliped 1 endopod; Mxpl exo, maxilliped 1 exopod; Mxpl epi, maxilliped 1 epipod; Mxp2 endo, maxilliped 2 endopod; Mxp2 exo. maxilliped 2 exopod; Mxp3 endo, maxil- liped 3 endopod; Mxp3 exo. maxilliped 3 exopod. pans (scaphognathite, maxilliped 1 epipod, and maxilliped 2 and maxilliped 3 exopods); and 8, dorso-ventral mouthparts (maxilla 1 endopod, maxilla 2 endopod, maxilliped 1 endopod, and maxilliped 1 exopod). These groupings apply mostly to the processes of food handling and have little significance with respect to grooming. When comparing our results to the liter- ature on other decapods, we found much resemblance to con- ditions in other anomurans. Introduction One of the most interesting features among crustaceans is their very complex mouth apparatus. The basic limb-pattern for Eucrustacea (the condition in the stem species to all recent Crustacea) was a labrum, paired paragnaths, and two pairs of mouthparts (mandibles and maxillae 1), followed by a large number of more-or-less similar limbs ( Walossek, 1998). Extant members of the Cephalocarida retain this system, but almost all other recent species have at least three pairs of mouthparts, i.e., mandibles (Md), maxillae 1 (Mxl ), and maxillae 2 (Mx2). Further specialization of the mouth apparatus is a very significant event in most crustacean lineages and often involves the specialization of thoracic limbs for food manipulation. This is especially so in the Decapoda, which has the first three pairs of thoracic ap- pendages, the maxillipeds (Mxpl -3). specialized into feed- ing appendages. In many of these decapods, the feeding apparatus is even more advanced since one or more of the pairs of pereiopods, especially if chelate, take part in food manipulation. This complexity has without doubt played an important role in the success of the Decapoda, since it enables the members of the order to feed on such a great diversity of food objects (Schembri, 1982c; Cartes, 1993). 281 282 A. GARM AND J. T. HOEG Unfortunately, we have but little understanding of the functional significance of this very complex feeding appa- ratus found in decapods, which on each side consists of at least 20 parts, each with their separate functions. Earlier functional studies of the decapod mouth apparatus have resulted in a division into inner mouthparts (Md, Mxl, Mx2. and Mxpl ) and outer mouthparts (Mxp2 and Mxp3) (Nicol. 1932; Kunze and Anderson, 1979; Schembri, 1982a). This division is based on morphology, but is often used in ways that indicate similar functions within the groupings. Many of the studies have concentrated mostly on Mxp3, the larg- est of the mouthparts. but again many of the functional interpretations are based on morphology alone (e.g., Green- wood, 1972; Farmer, 1974; Slithers and Anderson. 1981; Suthers, 1984; Lavalli and Factor, 1992). Grooming of the anterior part of the body using Mxp3 is well documented, and functional similarities are found throughout the Deca- poda (Bauer. 1981. 1989). We know much less about the functions of Mxl, Mx2, and Mxpl. The small size and fast movements of these appendages impede any detailed obser- vation, and in many decapods they are also hidden behind the larger Mxp2 and Mxp3. Recording mouthpart move- ments with macro-video equipment overcomes the first two problems (Stamhuis et ui, 1998); studying anomurans solves the last, since most are bottom dwellers with a rather open mouth apparatus, and pagurids in particular have re- ceived a lot of attention. At the gross level, anumurans have quite similar mouthparts, and the similarity with respect to Md, Mxl. Mx2, and Mxpl is particularly striking (Pike. 1947; Roberts. 1968; Kunze and Anderson, 1979; Schem- bri, 1982a). The Anomura is accordingly an especially suitable taxon for studying mouthpart function, both from an experimental point of view and because it offers a chance to evaluate to what extent similar morphology implies sim- ilar function. In a scanning electron microscopy (SEM) study of the morphology of mouthparts in the anomuran species Miinida sarsi, we offered some provisional interpretations of their function (Garm and H0eg. 2000). In the present study we present detailed video-based evidence of mouthpart func- tions in this species and in M. tenuimana, with particular attention to the roles of Mxl, Mx2, and Mxpl. Materials and Methods Video-recordings Seventy specimens of Miinida sarsi with carapace lengths between 7 mm and 28 mm were caught with a Sneli dredge at the Faeroe Bank <08°44'08" E, 61°24'46" N) at depths between 330 and 345 m in water with a temperature of 7°C. Six specimens of M. tenuimanu were caught with a trian- gular dredge northeast of the Faeroe Bank (09°28'95" E, 61°25'52" N) at depths between 690 and 715 m in water with a temperature of 4°C. Both species were caught in August 1998. The animals were kept alive at the Kaldbak Laboratory in 1000-1 tanks with running natural seawater. keeping the temperature at 10°C. The animals were fed with other animals from the catch areas, but also with animal and algal tissue from shallow waters. Video-recording took place in 50-1 tanks under the same conditions with four kinds of sediment: two kinds of mud- gravel from the areas of the catches (many foraminiferans at 330-345 m); mud from shallow water (very rich in organic material); and shell-gravel (low in organic material). The recordings were made from outside the tank by a color (Y/C) CCD camera with a Micronikkor 105-mm lens, which enabled us to record structures 5 /urn wide. Record- ings were made on super VMS, and light was obtained from a 120-W spotlight. Comparison of our videos with in situ recordings of M. sarsi made by a submersible shows that M. sarsi behaved naturally in the tanks. To trace the movement patterns of the mouthparts, we analyzed video sequences on computer using the MS DOS version of Sigmascan. Repre- sentative shots of the mouthpart movements were grabbed with a time resolution of 0.02 s (50 fields/s) and imported into Corel PhotoPaint 8.0, with a resolution of 786X564 pixels (Fig. 1). We outlined the involved mouthparts and used the outlines for serial drawings. These drawings there- fore accurately reflect the positions of mouthparts in the video sequences. For the analyses of the respiratory move- ments, the right branchiostegite of four individuals was dissected to get a clear view of the structures involved. All four animals behaved normally afterwards, although they only survived for a further 3-4 days. Light and scanning electron microscopy Specimens of Miinida sarsi and M. tenuimana that had been fixed in 29r formalin were obtained from The BIOFAR I project at the Faeroe Islands (Station 070, 61°24'69" N, 08°43'97" E and Station 314, 60°51'8X" N, 10°14'OX" E, Norrevang et ai, 1994). Adult males and females with a carapace length between 17 mm and 28 mm were used. The mouthparts were cleaned by ultrasound and manually with a beaverhair brush. A standard dissection microscope was used for the drawings (Fig. 2). SEM preparation followed Felgenhauer (1987), except that osmium was not used. The photographs were taken on a JEOL 840 scanning electron microscope and were stored electronically using the JEOL program SemAfore 3.0. They were processed and manipu- lated in CorelDraw 8.0. Abbreviations of mouthpart subdivisions Labrum (Lb), mandibular incisor (Inc) and molar process (Mp), mandibular palp (Mdp), maxilla 1 coxa (Mxl cox), maxilla 1 basis (Mxl bas), maxilla 1 endopod (Mxl endo), maxilla 2 coxa (Mx2 cox), maxilla 2 basis (Mx2 bas), maxilla 2 endopod (Mx2 endo), scaphognathite (Scapho), FUNCTIONS OF MUNIDA MOUTHPARTS 283 ^ Mx'l bas *- M.\2 bas Figure 1. Frame-grabbed video images of MuniJa sarsi. ( A ) Overview of buccal area in resting position. ( B ) Flagellum of Mxp2 exopod in resting position. Dotted line outlines flagellum. (C) Flagellum from B spread out when moving anterio-dorsally. Dotted line outlines flagellum. (D) Sediment sorting by Mx2 basis and Mxpl basis. (E) M. sarsi on muddy sediment. Arrows indicate flagella currents. Size of arrows indicate strength of current measured as relative velocity of suspended particles. (F) Holding a small prey item with Mxl basis and Mx2 basis. (G) Biting by incisor processes. Mandibular palps and Lb are raised and incisor processes overlap with left on dorsal side. (H) Mxp3 endopods hold large prey; setal screen prevents prey items from escaping anteriorly. (I) Flagellum of Antl is groomed by setae distally on carpus of Mxp3 endopod. Arrowheads indicate direction of movements. Antl = antenna 1, Bas = basis. Car = carpus. Dae = dactylus, Endo = endopod. Exo = exopod. Fla = flagellum. Inc = incisor process. Lb = labrum. Mer = merus. Mdp = mandibular palp. Mxl = maxilla 1, Mx2 = maxilla 2. Mxpl = maxilliped 1. Mxp2 = maxilliped 2, Mxp3 = maxilliped 3. P = prey item. Pro = propodus, Sed = sediment, SS = setal screen. maxilliped 1 coxa (Mxpl cox), maxilliped 1 basis (Mxpl bas). maxilliped I endopod (Mxpl endo). maxilliped 1 exopod (Mxpl exo), maxilliped 1 epipod (Mxpl epi), maxilliped 2 endopod (Mxp2 endo), maxilliped 2 exopod (Mxp2 exo), maxilliped 3 endopod (Mxp3 endo), maxilliped 3 exopod (Mxp3 exo). Results Munida sarsi and M. tenuimana have very similar mouth- parts, which differ only in details of setation (unpubl. re- sults). Similarly, from the video-recordings we found no differences in feeding behavior between the two species, and the results presented below therefore apply to both species referred to hereafter as Munida. The mouthparts have a wide variety of functions, includ- ing food manipulation, ingestion, water current generation, and grooming (Table 1 ), and the patterns of movement are often complex. When the endopods of Mxp2 and Mxp3 handle large prey items, the movements and structures 284 A. GARM AND J. T. H0EG involved largely depend on the size and shape of the prey. This changes with every situation, though some generaliza- tions can still be made. Other movement patterns are much more stereotypical, especially those performed by Md, Mxl, Mx2. and Mxpl when handling small prey items, and we will principally focus on these latter functions. Duration or frequency for the stereotypical movements are listed in Table 1 and illustrated in Figures 3 to 13. EP, FUNCTIONS OF MUNIDA MOUTHPARTS 285 Table 1 Behavioral processes and the structures involved Behavioral process Structures involved1 Duration (s)~ Frequency (Hz)- 1. Current generation 2. Prey gathering 3. Sediment gathering 4. Transfer sediment to inner mouthparts 5. Sediment sorting/Particle rejection' 6. Rejection of large prey 7. Rotating particle (transverse plan)3 8. Rotating particle (medial plan)3 9. Put prey between mandibles 10. Biting soft prey 1 1. Biting calcified prey 12. Crushing very calcified prey 13. Transfer prey to mouth 14. Ingestion6 15. Grooming of Ant 1 16. Grooming of Ant2 17. Respiration7 Mxp2 fla, Mxp3 fla Mxp3 endo, PI Mxp3 endo, P1-P4 Mxp2 endo. Mxp3 endo Mxl bas. Mx2 bas, Mxpl has Mxp2 endo, Mxp3 endo Mxl bas, Mx2 bas (Mxpl bas4) Mxl bas, Mx2 bas (Mxpl bas4) Mxpl bas, Mxp2 endo5 Inc, Mxl bas Inc. Mxl bas Mp Lb. Mdp Mxl cox, Mx2 cox, Mxpl cox Mxp3 endo (car) Mxp3 endo (pro) Mx2 scapho Variable Variable Variable Variable Variable 1.20 ± 0.15 Variable Variable Variable Not observed 1.77 ±0.48 2.28 ± 0.34 15.2 ± 1.6 3.26 ± 0.28 3.21 ± 0.08 3.18 + 0.27 2.6 ±0.1 ' Bas = basis. Car = carpus. Cox = coxa, Endo = endopod. Fla = flagellum. Inc = incisor process, Lb = labrum, Mdp = mandibular palp, Mp molar process. Mxl = maxilla 1, Mx2 = maxilla 2, Mxpl = maxilliped 1, Mxp2 = maxilliped 2, Mxp3 = maxilliped 3, Pl-4 = pereiopod 1-4. Pro propodus, Scapho = scaphognathite. 2 Average ± standard deviation. 3 The movements are circular, and the frequency is for one round of movements. 4 Mxpl bas are not directly involved, see text for details. 5 Mxp2 endo are used only for large prey items. 6 See Garm and H0eg (2000). 7 When not handling any prey. When handling prey, the frequency is higher, see 6, 8, 9. Resting position and current generation Figure 1A shows the position of the mouthparts when they are not handling food. Lb and mandibular palps are in ventro-posterior positions, and Md, Mx2, and Mxpl are opened to only about one-third of their maximum. Mxp2 and Mxp3 are held laterally, with their endopods bent into a U-shape and the flagella on their exopods (Mxp2 exo and Mxp3 exo) beating almost continuously in a dorso-ventral plane. This high-frequency beating generates a unidirec- tional current around the anterior end of the animal; the current proceeds from the sediment up through the mouth- parts (Fig. IE). During beating, the right Mxp2 flagellum is synchronized with the left Mxp3 flagellum and vice versa. The high frequency is correlated with the special morphol- Figure 2. Morphology of the mouthparts. B, F. H, N. Q from Munida tenuimana, rest from M. sarsi. (A) Left Md seen posterio-medially. (B) Setae on mandibular palp bending ventrally. (C) Mp with tubercles (arrow). Note large posterior tooth on incisor process (arrowhead). (D) Dorsal view of right Mxl. (E) Closeup of spines (arrow) and robust setae (arrowhead) on medial edge of Mxl basis. (F) Cluster of setae on tip of Mxl endopod. (G) Dorsal view of right Mx2. Scapho inserts asymmetrically on rest of Mx2. (H) Dorsal view of dissected Mx2 endopod. Long slender setae project medially. (I) Medial edge of Mx2 basisl composed of rather fragile setae. (J) Dorsal view of right Mxpl. Note close contact between endopod and exopod. (K) Medial view of Mxpl. Mxpl basis is curved blade shaped. (L) Medial view of Mxp2. Fla is in resting position. (M) Dorsal view of Mxp2 fla cuticle in resting position. (N) Medial view of Mxp2 fla cuticle in resting position. (O) Posterior view of shovel-shaped distal end of Mxp2 endopod. (P) Medial view of Mxp3. Fla is in upper most position. (Q) Posterior view of shovel-shaped distal end of Mxp3 endopod. Arrowhead indicate setal cluster on Pro used for grooming antenna 2. (R) Setal cluster on Car used for grooming antenna 1. (S) Weakly serrate seta from cluster used for grooming antenna 1. (T) Strongly serrate setae from cluster used for grooming antenna 2. Scalebars: A, D. H, G, J. K. L. O, P, Q = 1 mm; B. C. E. F. I. M. T = 100 /urn; N = 50 /j.m; R = 200 /im; S = 20 |u.m. Bas = basis, Car = carpus, Cox = Coxa, Endo = endopod. Epi = epipod. Exo = exopod. Fla = flagellum, Inc = incisor process. Md = mandible, Mdp = mandibular palp, Mp = molar process, MX 1 = maxilla 1 , Mx2 = maxilla 2, Mxpl = maxilliped 1. Mxp2 = maxilliped 2. Mxp3 = maxilliped 3. Pro = propodus, Scapho = scaphognathite. 286 A Mxp3 exo A. GARM AND J. T. H0EG C / D t=0.08 t=0.16 t=0.3 t=0.58 t=0.64 Figure 3. Sediment gathering by the endopods of Mxp3 (Mxp3 endo, dark gray), seen anterio-laterally. (A) Both dactyli of M\p3 endopod are in the sediment (Sed). (B-D) Left Mxp3 endopod moves dorso-posterially. lifting up a lump of Sed sticking to the setae, and endopod of left maxilliped 2 (Mxp2 endo, light gray) moves dorso-anteriorly to get in front of the lump. Arrowhead in B indicates flexible merus-carpus joint. (E) Left Mxp2 endopod moves posteriorly, combing through setae of Mxp3 endopod. picking up Sed with distal segments. (F) Mxp2 endopod moves further posteriorly, transferring Sed to Mxpl bas (see Fig. 4). Note that now right Mxp3 endopod has picked up a lump of Sed. Mxp3 exo = exopod of maxilliped 3, P2 = pereiopod 2. t = time in seconds (specific for series of pictures, not generalized). ogy of the flagellar cuticle (Fig. 2M. N). When the flagellum is moving dorso-anteriorly. the water pressure causes the folded and flexible cuticle on the dorsal side of the flagellum to unfold. This raises the plumose setae into a fan, and this movement thus becomes the power stroke (Fig. 1C). During the ventro-posteriorly directed recovery stroke, the flagellar cuticle folds again, causing the setal fan to close (Fig. IB). The flagellar current serves to reject small particles and gives the animal an opportunity to detect the composition of the sediment, since water is pumped up from close to the sediment and past the setae on the mouthparts (for details, see Garm and H0eg. 2000). Food gathering Most larger food particles are picked up by the long chelipeds and passed to the Mxp3 endopods, both of which extend to grasp the food between their dactyli and propodi. In some cases, the dactyli of the Mxp3 endopods can also pick up food objects directly from the sediment. Sediment is gathered either with the dactyli of pereiopods 2-4 (P2-4). the chelipeds (PI), or the dactyli of the Mxp3 endopods. In the first two cases, the sediment is passed on to the dactyli of the Mxp3 endopods as they comb through the setae that hold the sediment on the cheliped or P2-4. When the Mxp3 endopods collect sediment, they start by pressing the dactylus and sometimes also the propodus into it (Fig. 3A). Thereafter the distal end of the endopod bends posterio-dorso-medially (towards the mouth) and shovels up a lump of sediment (Fig. 3B-D). The great flexibility of the endopod, especially in the merus-carpus joint (Fig. 3B. arrowhead; Fig. 4), greatly facilitates this process. The efficacy of the Mxp3 endopods in handling sediment is improved by a rim of strong, serrate setae along the distal end (Fig. 2P. Q): this enlarges the shovel, and the setae serve as hooks. The Mxp3 endopods usually pick up sedi- ment from the area right under the flagella. The two Mxp3 endopods may move synchronously when gathering sedi- ment, but they are normally used one at a time, as seen in Figure 3. Handling large prey items With respect to larger food items, the animals were not at all choosy and ate all kinds of animal tissue presented to them, even sponges (Porifera) and gorgonians (Gorgona- cea). Such large food objects are grasped by the endopods of Mxp2 and Mxp3 (Fig. 1H) and pushed directly towards the mandibles. The Mxp3 endopods are bent into a U-shape and hold the prey between the meri and the propodi. The dactyli are held under the prey if it is not too large. The bases of Mxl and Mx2 are fairly inactive and mostly press against the prey, while possibly making fine adjustments of the food item for the mandibles. When the mandibles cut into a large food item, the bases of Mxpl often scrape very actively along the object in circles, as shown in Figure 5. If the prey is soft, such as a lump of fish meat, the movements of the Mxpl bases tend to squeeze it so it will fit more easily between the incisor processes of the mandibles (Inc). On a few occasions, the animals were also observed attempting to FUNCTIONS OF MUN1DA MOUTHPARTS 287 t = 0 t = 0.48 Figure 4. Transfer of panicles (P) from endopod of maxilliped 3 (Mxp3 endo. dark gray) to the basis of maxilliped 1 (Mxpl bas, light gray) seen ventro-anteriorly. Corresponds to D-F in Fig. 3. (A. B) Left Mxp3 endopod bends posterio-medially. Endopod of left maxilliped 2 (Mxp2 endo, middle gray) moves ventrally towards Mxp3 endopod. (C. D) Mxp2 endopod moves posteriorly, with the setae on dactylus combing through the setae on the distal part of Mxp3 endopod. Arrowhead in C indicates flexible merus-carpus joint. (E, F) Mxp3 endopod moves anterio-laterally; Mxp2 endopod moves posteriorly, with the distal part reaching the inner mouthparts. Left Mxpl basis moves laterally to be able to collect P from Mxp2 endopod (see Fig. 5). Inc = incisor. Mdp = mandibular palp, t = time in seconds (specific for series of pictures, not generalized). use the crista dentata on the ischium of Mxp3 endopods in a crablike manner. This would involve holding the prey between the incisor processes while both Mxp3 endopods move dorso-anteriorly, grasping the prey with the crista dentata while in the dorsal position, then lowering the Mxp3 endopods and thereby tearing the prey. However, this be- havior w:as never observed to be successful in Munida. because the crista dentata never got hold of the prey items. Large objects are not rejected by the flagellar current, but are simply dropped or removed by Mxp2 endopods. Handling small prey items When carrying a small food item, the Mxp3 endopods bend posterio-medially to meet the endopod of maxilliped 2 (Mxp2 endo) (Figs. 3. 4), which passes sediment and par- ticles further towards the mouth, that is. to the bases of Mxpl. When the Mxp3 endopod approaches the mouth, the collateral Mxp2 endopod moves first dorso-anteriorly to get in front of the food item held distally on the Mxp3 endopod (Fig. 3C) and thereafter ventro-posteriorly to shovel up the item (Figs. 3D-F, 4A-D). Holding the food with its distal end, the Mxp2 endopod bends further towards the mouth and reaches the area of the bases of Mxpl. Meanwhile the collateral Mxpl basis moves aside to make room for the food, and the Mxp3 endopod extends again (Fig. 4E, F). Most of the flexure of the Mxp2 endopod takes place in the merus-carpus joint (Fig. 4C. arrowhead), and the distal segments also form a shovel that enables them to carry the food (Fig. 2L, O). The serrate setae found distally on the Mxp2 endopod are less robust than those on the Mxp3 endopod, and therefore they can pass between the latter to collect the potential food. Like the Mxp3 endopods, the endopods of Mxp2 can be used at the same time — when one is moving anteriorly the other is moving posteriorly — but this is not the normal pattern. The Mxpl bases collect the particles from the Mxp2 endopod by making circular movements as shown in Figure 5, but in the opposite direction. Moving the curved and blade-shaped bases (Fig. 2J, K) medially when they reach the Mxp2 endopod and laterally when they meet the Mx2 bases ensures the direction of particle transport. Mx2 moves medially and grasps the prey. Sediment sorting and panicle rejection Almost all collected particles eventually reach the Mx2 and Mxl bases, which are the mouthparts responsible for rejecting or retaining food items. Figure 6 shows how the 2SX A. GARM AND J. T. H0EG D t=0.18 t=0.24 t=0.32 Figure 5. Movements of the bases of maxilliped 1 (Mxpl bas. gray) seen anterio-laterally. To simplify drawings, maxilla 1 and maxilla 2 are left out and setae are shown only when in contact with prey. The appendages make circular movements in an angle to plane of view indicated by ellipse; thin part of ellipse is towards the animal. Black dots indicate the positions of the limbs in the movement. This movement can shred small and soft food objects (Sfo) held by mandibles or move particles away from mandibles. Movement can be reversed, which will move particles from endopod of maxilliped 2 (Mxp2 endo) towards basis of Mx2. Inc = incisor process, Lb = labrum. Mdp = mandibular palp, t = time in seconds (specific for series of pictures, not generalized). Mxl and Mx2 bases move in circles parallel to the man- dibular incisor (Inc) when sorting sediment. The direction of movement is dorso-anteriorly when they are in the medial position and ventro-posteriorly when they are in the lateral position. Thus, from the animal's point of view, the distal tips of the right-side appendages rotate clockwise, while those on the left side rotate counterclockwise. This ensures that the cuspidate setae on their medial edges hit the parti- cles from the ventro-posterior and push them in a dorso- anterior direction. The Mxl bases and the Mx2 bases move out of phase, ensuring that one of these appendage pairs remains in contact with the particle most of the time. The circular movements continue until the particle enters the flagellar current and is rejected anteriorly (Figs. ID, 6), or until the animal decides to retain the particle. The Mxl bases are responsible for most of the particle movement. The medial rims of the Mxl bases have much more robust cuspidate setae than the Mx2 bases (Fig. 2E, I), and this allows them to press harder against the particles, obtaining a more firm hold. Within a specific pair (Mxl bases or Mx2 bases), left and right sides are synchronized so they are in the medial position at the same time. The Mxpl bases move as described under "Handling small prey items," providing more particles, pushing them towards the Mx2 bases, and ensuring that nothing is lost ventro-anteriorly (Fig. 6). Dur- ing the sorting process, the Mxp2 endopods are held medi- ally where they form a setal screen which also prevents particles from escaping ventro-anteriorly. When the animal is handling large prey, the Mxp3 endopods provide a similar setal screen (Fig. 1H). Panicle rotation When the animal finds a small particle worth eating, the Mxl and Mx2 bases stop making circular movements and begin moving to and fro in the medio-lateral plane while keeping the particle in front of the incisor process (Fig. IF). Again, the Mxl and Mx2 bases move out of phase. If the particle is not orientated correctly for the Md. it will sub- sequently be rotated by the Mxl and Mx2 bases (Figs. 7, 8). During rotation of a particle in the transverse plane, the Mx2 and Mxl bases again move out of phase and make circular movements parallel to the incisor process (Fig. 7). Unlike the situation during sediment sorting, all four mouth- parts concerned (i.e., both left- and right-side Mx2 bases and MX 1 bases) move either clockwise or counterclockwise. To turn a particle clockwise (as seen from the animal's point of view), the left-side Mxl and Mx2 bases move dorso- anteriorly when in their medial position and ventro-poste- riorly when in their lateral position, thereby pushing the left side of the particle dorso-anteriorly. In the medial position, the bases of the right-side Mxl and Mx2 move in the opposite direction as the left-side bases: thus they push the right side of the particle ventro-posteriorly, and it is rotated in the transverse plane. The animal rotates the particle counterclockwise by reversing these movements. Rotating a particle in the medial plane (Fig. 8) is always FUNCTIONS OF MUNIDA MOUTHPARTS B C 289 I O.IIS t=0.12 t=0.17 t=0.23 t=0.3 t=0.35 Figure 6. Rejection of unwanted particle (PI by bases of maxilla 1 (Mxl bas. dark gray), maxilla 2 (Mx2 bas, middle gray), and maxilliped 1 (Mxpl bas. light gray) seen ventro-anteriorly: setae are shown only when in contact with P. Arrows indicate direction of movements. (A) Mxl basis in contact with P pushing dorso-anteriorly. (B. C) Mx2 basis gets in contact with P and pushes it dorso-anteriorly. Mxl basis moves laterally and releases P. (D, E) Mx2 basis releases P. (F) P is pushed above mouthparts by Mxl basis and enters the flagellar current. (G) P is rejected. During the rejection, Mxpl basis is not in contact with P, but makes posterio-anteriorly circles (see Fig. 5). probably ensuring that P is not lost anteriorly. Inc = incisor process, Mdp = mandibular palp, t = time in seconds (specific for series of pictures, not generalized). done in one direction, with the part of the particle closest to the animal moving dorso-anteriorly and the part away from the animal moving ventro-posteriorly, that is, clockwise when seen from the animal's right side. The mouthparts of the left and right sides move in synchrony, as when rotating particles in the transverse plane, but the distal tips of the Mx2 bases and the Mxl bases now circle in opposite direc- tions. The Mxl bases move dorso-anteriorly when they are in the medial position and ventro-posteriorly when in the lateral position. They thereby hit the particle from a ventro- Mxl bas x2 bas B 1) t=0.06 t=0.12 t=0.16 H t=0.22 t=0.28 t=0.34 t=o; Figure 7. Rotation of a small particle (P) in a transverse plan by bases of maxilla 1 (Mxl bas, dark gray) and maxilla 2 (Mx2 bas. middle gray). The view is ventro-ameriorly and setae are only shown when in contact with P. (A. B) P is held by Mxl basis. Left Mxl basis moves dorso-anteriorly; right Mxl basis moves a little laterally. This starts the rotation. (C) Mxl basis releases P; Mx2 basis gets in contact with P. (D-F) Mx2 basis in contact with P. Left moves dorso-anteriorly and right ventro-posteriorly. which rotates P further. (G) P is hit by left Mxl basis from ventro-posterior and laterally by right Mxl. (H) Back in start position. During rotation the bases of maxilliped 1 (Mxpl bas, light gray) are not in contact but tend to make anterio-postenorly circles (see Fig. 5). probably ensuring P is not lost anteriorly. Inc = incisor process. Mdp = mandibular palp. Mxp2 endo = endopod of maxilliped 2. t = time in seconds (specific for series of pictures, not generalized). 290 Left t=0.06 A. GARM AND J. T. H0EG BCD t=0.33 Figure 8. Rotation of a small particle (P) in a medial plan by the bases of maxilla 1 and maxilla 2 (Mxl bas, dark gray and Mx2 bas, light gray). View is ventro-anterior. Only left side is shown in B-G, and setae are shown only when in contact with P. Right and left side move in synchrony and in phase. Arrows indicate direction of movements. Dotted line indicates hidden structure. Mx2 basis makes circular movements; black dots indicate the positions of the appendages in the circle. ( A. B ) MX 1 basis releases P. ( C ) Mx2 basis hits anterior edge of P from dorso-anterior, pushing it ventro-posteriorly. (D-F) Mxl basis hits posterior edge of P from ventro-posterior, pushing it dorso-anteriorly. Mx2 basis releases P. (G) Mx2 basis moves laterally and up; Mxl basis releases P. Process is ready to start again. Inc = incisorprocess, Mdp = mandibular palp, t = time in seconds (specific for series of pictures, not generalized). posterior direction and push it dorso-anteriorly. When cir- cling in the opposite direction, the Mx2 bases hit the particle from a dorso-anterior direction and push it ventro-posteri- orly. This causes the particle to be rotated around a point between the bases of Mxl and Mx2. In both cases, the Mxl bases are responsible for most of the rotation, and the animal has serrate and cuspidate setae in contact with the particle at all times. The Mxpl bases tend to make circular movements that push the particle toward the MX 2 basis. Both during sediment sorting and when rotating a small particle, the Mxl, Mx2, and Mxpl bases normally circle with the same frequency. The normal order of contact with the particle is Mxl basis, Mx2 basis, Mxpl basis, Mxl basis, and so forth in a repeated way, as seen in Figure 6. However, the Mxpl bases occasionally per- form more complex movement patterns independent of the positions of the Mxl and Mx2 bases, as indicated in Figure 7. Cutting-crushing Having orientated the particle correctly for maceration, the animal puts it between the mandibles (Fig. 9). At first, the mandibular incisors (Inc) move laterally while the labrum (Lb) and mandibular palps (Mdp) retract dorso-anteriorly to make room for the item (Fig. 9B). Subsequently, the Mxpl bases (and possibly also the Mxl and Mx2 bases) push the particle between the incisor processes, which move medially until they overlap about one-fifth, always with the right incisor process on the anterior side (Fig. 9C, D). A large posterior tooth on the left incisor process assures a good grip on the item (Fig. 2C, arrowhead). Meanwhile the food particle is held tight by the robust spines and cuspidate setae on the medial rim of the Mxl bases (Fig. 2E). If the particle is slim, the incisor process will not move laterally first, but will move medially directly from resting position, thereby performing the cut. After the cut, the incisor processes move laterally and the mandibular palps and Lb move ventro-posteriorly (Fig. 9E, F), pushing the cut-off piece of food towards the mouth, an action facilitated by serrate setae on the mandibular palps (Fig. 2B) and setule-like outgrowths on Lb. The rest of the particle is pushed ventro- anteriorly to be further processed by the Mxl and Mx2 bases. The ingestion is handled by the coxae of Mxl and Mx2 and perhaps Mxp 1 . This is not seen on the videos but extrapolated from the organization of the mouth apparatus. The three pairs of coxae are situated just ventral to the mouth and. along with their setae, they curve into the mouth opening ( for details, see Garm and H0eg. 2000). Prey too hard to cut with the incisor process, such as calciferous polychaete tubes, are instead crushed by the molar processes of the mandibles (Mp). The flattened shape and their rim of tubercles ensure that even very smooth objects such as mollusc shells will not easily slip away (Fig. 2C). The movements are identical to the cutting action (Fig. 9). except that the prey is placed between the molar processes posteriorly to the incisor processes and the process lasts longer. Grooming the antennae Grooming of both antenna 1 ( Antl ) and antenna 2 ( Ant2) is performed by the Mxp3 endopods. but by two different FUNCTIONS OF MUN1DA MOUTHPARTS 291 I) Right Left t=0 t=0.16 t=0.54 t=1.36 t=1.76 t t=2 Figure 9. Cutting by mandibles seen ventro-anteriorly. (A) Resting position. Incisor processes (Inc) are a little opened; labrum (Lb) and mandibular palps (Mdp) are lowered. Prey (P) is held by bases of maxilla 1 (Mxl bas). (B) Incisor processes move laterally; Lb and mandibular palps move anteriorly to make room for P. (C) Cut starts. Incisor processes move medially; Lb and mandibular palps move further anteriorly. (D) Incisor processes with maximum overlap, always left on dorsal side. Lb and mandibular palps in antenormost position. (E) Cut ends. Incisor processes move laterally; Lb and mandibular palps move posteriorly and push the cut-off piece towards mouth. (F) Cut has ended; limbs back in resting position. During the cut. Mxl are held medially to hold P. If P is small, stage B is skipped, t = time in seconds (specific for series of pictures, not generalized). clusters of setae (Figs. 2P-R, 10, 11). The antennae are always groomed one at a time, and Antl are groomed much more frequently than Ant2. Grooming of an Antl starts with the appendage in question bending ventrally and both Mxp3 endopods moving dorsally (Fig. 10A). In the next step, Mxp3 endopods move medially and catch the peduncle of Antl with serrate setae on the carpi (Fig. 2R). With a loose grip, they move ventrally until they reach the flagellum of Right Antl ^T Fla*? \ t=0.68 Right Mxp3 endo D t=0.9 Figure 10. Grooming of antenna 1 (Ant 1) by endopods of maxilliped 3 (Mxp3 endo), seen anterio-laterally. (A) Mxp3 endopods reach up while Antl bends down. (B) Mxp3 endopods move ventro-medially and Antl bends posteriorly. (C) Antl is caught by long setae distally on carpus (Car), which move along shaft of Antl. (D) When the flagellum (Fla) is reached. Mxp3 endopods move further medially. (E-G) Mxp3 squeeze tight around Fla and aesthetascs are pulled through setae on Car. t = time in seconds (specific for series of pictures, not generalized). 292 A. GARM AND J. T. H0EG Right M\p3 endo Figure 11. Grooming of right antenna 2 (Ant2) by endopod of maxilliped 3 (Mxp3 endo) seen ventro- anteriorly. The antenna is caught on right side and released on left. Arrows indicate direction of movements. Black dots indicate contact between Ant2 and propodus (Pro). (A) Mxp3 endopod has moved dorsally and right Ant2 has bent ventrally. Mxp3 endopod moves medially and catches Ant2 with long setae on distal part of Pro. (B) Mxp3 endopod moves downwards and Pro setae run along Ant2, grooming it. (C-E) Mxp3 endopod moves further down and Ant2 pulls itself up through setae distally on Pro. t = time in seconds (specific for series of pictures, not generalized). t=1.14 Antl (Fig. 10B-D). Now the Mxp3 endopods move further medially, squeezing the Antl flagellum tightly between the carpi, and the flagellum is pulled through the serrate setae as the Antl moves dorsally (Figs. 11, 10E-G). Occasionally, the Mxp3 endopods move farther dorsally in the beginning of the process, reaching the eye situated just dorsal to Antl. The eye is then groomed by setae on the dactyli and the propodi. The same setal clusters on the carpi also groom the Mxp2 and Mxp3 flagella, and the movements of the Mxp3 endopods are much the same. Ant2 is also groomed by the Mxp3 endopods, but with setal clusters on the propodi instead of on the carpi (Figs. 2Q, T; 11). One of the Ant2 bends ventrally, and both Mxp3 endopods move dorsally and grasp the long flagellum of Ant2 between setal clusters distally on the propodi (Fig. 1 1A). When in contact with the flagellum, the Mxp3 endo- pods move ventro-anteriorly and Ant2 moves dorsally, pull- ing itself through the robust serrate setae (Fig. 1 1B-E). Ant2 grooming is an asymmetrical process: the appendage is grasped on the side of the animal where it is attached and subsequently released on the other side. Due to the twist of the movement, the two Mxp3 endopods move apart during the process and groom two different sites of Ant2 at a given time (Fig 1 1C, D). This means that Mxp3 endopods cannot squeeze the Ant2 flagellum as tightly as when grooming the Antl flagellum. The setal clusters grooming Antl and Ant2 are situated differently on the Mxp3 endopod and are composed of different types of serrate setae (Fig. 2S, T). Those groom- ing Ant2 are more stout and have much more robust denticles. One curious observation concerned a specimen of M. sarsi that had lost both Mxp3 during sampling. It neither ate nor groomed Antl. but it did groom Ant2 with the chelae. Respiration Beating of the scaphognathites (Scapho), also called gill bailers, produces the respiratory currents. The Scapho are situated laterally on Mx2 (Fig. 2G), which places them in the anterior part of the gill chambers. When an Mx2 basis executes medio-lateral movements, its Scapho moves ven- tro-dorsally (Fig. 12). The movements of the Scapho are comparable to swimming with flippers. Because it inserts on the Mx2 coxa with the less flexible posteriormost part, the flexible anterior part moves with a delay when compared to the posterior part. When the posterior part starts moving dorsally, the anterior part initially remains at rest, but after a while it follows until the entire Scapho reaches its dorsal- most position (Fig. 12A-E). During the ventral stroke, the movement is again initiated by the posterior part, followed, with some delay, by the anterior part (Fig. 12F-I). The delayed movements of the anterior part ensure that the respiratory current is unidirectional, with water entering the posterior part of the gill chamber and exiting anteriorly. We never observed the reversal of the respiratory current that Bauer (1981) described for other decapods. When moving ventrally, the anterior part of the Scapho sweeps across the epipod of Mxp 1 and the exopods of Mxp2 and Mxp3, and these structures help direct the exhalant current (Fig. 12G-I). The posterior part sweeps the two anteriormost pairs of gills. Dorsoventral mouthparts Four structures of the mouth apparatus have not yet been dealt with, since their activities are not clear. These struc- tures, which constitute the dorso-lateral part of the mouth apparatus, are the endopods of Mxl, Mx2, and Mxpl, and the exopod of Mxpl. When the Mxl and Mx2 bases move t = 0 t = 0.28 t = 0.04 FUNCTIONS OF MUNIDA MOUTHPARTS t = 0.08 t = 0.12 t = 0.32 t = 0.4 t = 0.44 293 t = 0.16 Figure 12. Respiratory movements of right scaphognathite (Scapho, gray) seen laterally; branchiostegite is removed. Arrows indicate moving part and direction of movement. Dotted arrows indicate respiratory current. Junction between Scapho and rest of maxilla 2 (arrowhead in A) is situated posteriorly on Scapho, which make posterior part follow movements of Mx2, dragging anterior part behind. Delayed movement of anterior part ensures that respiratory current is directed anteriorly. Note that Scapho sweeps across epipod of maxilliped 1 (Mxpl epi). Mxp2 exo = exopod of maxilliped 2. Mxp3 exo = exopod of maxilliped 3. t = time in seconds (specific for series of pictures, not generalized). in the medio-lateral plane, the endopods of Mxl and Mx2 perform small semicircular movements (Fig. 13). This causes the distal tip of the Mx2 endopod to rub against the medial part of the base of the mandibular palp and the Mxl endopod to sweep across a small part of the dorsal side of the Mx2 endopod. Both Mxl and Mx2 endopods have a well-defined cluster of setae (Fig. 2F, H). The long serrulate setae of the Mx2 endopod lie in the space between the Figure 13. Placement and movements of dorso-ventral mouthparts (endopods of right maxilla 1 and maxilla 2 [Mxl endo and Mx2 endo] and exopod of right maxilliped 1 [Mxpl exo]), seen latero-anteriorly. Arrows indicate direction of movements. Dotted arrow indicates respiratory cur- rent. Open arrowhead indicates Mx2 setae lying in the space between Mxl basis and incisor process (Inc) (see also Fig. 2H); black arrowhead indi- cates Mxl setae (see also Fig. 2F). When Mx2 basis make medio-laterally movements, tip of Mx2 endopod rubs against base of the mandibular palp (Mdp). Setae on Mxl endopod are apparently not in any contact. Flagellum of Mxpl exopod is situated in respiratory current. incisor process and the Mxl basis and are in contact with the incisor process. The smaller serrate setae of the Mxl endopod seem not to contact anything. The flagellum of the exopod of Mxpl lies in the exhalant current and moves very little, even when the rest of the Mxpl is actively processing food. Water from the flagellar current also passes the fla- gellum of the Mxpl exopod. The Mxpl endopod lies very close and dorso-ventral to the Mxpl exopod (Fig. 2J, K), and is therefore also placed in the currents, but in Fig. 13 the Mxpl exopod obscures it from view. Discussion Functional grouping of mouthparts The six pairs of mouthparts (Md, Mxl, Mx2, and Mxpl -3) and the labrum are very different in both morphol- ogy (Fig. 2) and function (Figs. 3-13. Table 1). To some degree, their pronounced differences in size and shape cor- relate with function, since the large and pediform Mxp2 and Mxp3 handle large prey items, whereas the smaller and flattened Mxl, Mx2, and Mxpl enable the animal to accu- rately manipulate even very small food objects. The dorso- ventrally flattened form of these small mouthparts enables several independently moving structures to operate close together, a prerequisite to accurate manipulation of small food. The freedom of movement decreases in the mouth- parts that are nearer to the mouth and arranged very close together, and this explains why the Lb, Mxl, Mx2, and Md perform rather stereotypical movements in a two-dimen- 294 A. GARM AND J. T. HOEG sional plane. The maxillipeds have much more space in which to operate, and the Mxp2 and Mxp3 endopods can move in all directions anterior to the animal due to their flexible joints. They can therefore manipulate a great variety of food items. Although morphology and organization could support the division of the mouth apparatus into inner and outer mouth- parts, this is a much too simplified view when it comes to functions of the mouthparts. As mentioned earlier, each side of the mouth apparatus can be divided into at least 20 separate parts, but these parts seldom work independently; rather they operate in functional groups. The Mxp2 and Mxp3 endopods have fairly similar functions and often work jointly, at least when processing potential food ob- jects. The Mxl and Mx2 bases form another group with very high correlation of movements and functions when processing food particles. Both of these two groups relate to the Mxpl bases, which do not clearly belong to a single functional group, but can work closely together with either the Mxp2 and Mxp3 endopods or with the Mxl and Mx2 bases (Fig. 4 and Fig. 7. respectively). Therefore, the Mxpl bases constitute a group of their own. The flagella of the Mxp2 and Mxp3 exopods form a well-defined functional group having exactly the same function, and they almost always work together. The labium, the incisor processes, the molar processes, and the mandibular palps form another functional group, where the elements rarely move indepen- dently. The functional group composed of the coxae of Mxl, M.\2, and Mxpl cannot be seen in the videos, but their role in ingestion can be interpreted from their position and morphology (for details, see Garm and H0eg. 2000). The scaphognathite, the Mxpl epipod. and the Mxp2-3 exopods form a functional group because they cooperate in creating and directing the respiratory currents. The Mxl. Mx2, and Mxpl endopods and the Mxpl exopod seem to form the last functional group. This follows solely from their position, since their functions are but poorly understood. The distal tip of the Mx2 endopod might groom the base of the mandibular palp, and the setae might groom the ventro- lateral side of the incisor process. Situated in the respiratory current, sensory setae on the Mxpl endopod and exopod could allow them to be used to sense the need for gill grooming. The Mxpl endopod in the brachyuran crab Eba- lia tuberosa forms an exhalant canal, as do the Mxpl endopod and exopod in the hermit crab Pagurus rubricatus (Schembri, 1982a. b), but this is unlikely to be the case in Munida because those structures are so slender. We never observed any contact between the setal cluster on the distal tip of the Mxl endopod and any other object or mouthpart. and this could imply that these may be remote chemosen- sory setae. At least for Munida, our groupings seem to have more explanatory power than the conventional and superficial separation into inner and outer mouthparts. We suggest the following terms based on the observed functions: 1. transporting mouthparts for Mxp2 endopod and Mxp3 endopod; 2. transporting— aligning mouthpart for Mxpl basis; 3. sorting-aligning mouthparts for Mxl basis and Mx2 basis; 4. current- generating mouthparts for flagella of Mxp2 and Mxp3 exopods; 5. cutting-crushing mouthparts for incisor and molar processess, mandibular palp, and labrum; 6. ingesting mouthparts for Mxl coxa, Mx2 coxa, and Mxpl coxa; 7. respiratory mouthparts for scaphognathite. Mxpl epipod, and Mxp2 and Mxp3 exopods; 8. dorso-lateral mouthparts (with uncertain functions) for Mxl endopod. Mx2 endopod, Mxpl endopod, and Mxpl exopod. These groupings are not entirely strict, as illustrated by the grooming behavior, in which each mouthpart (except Inc and Mp) partakes in grooming adjacent structures. The Mx2 bases can also move independently of the Mxl bases, as evidenced by the respiratory movements, when the ani- mal is not handling any food. Moreover, there is a slight division of functions between Mxl and Mx2 when handling food items. The robust spines and serrate setae arming the Mxl bases indicate their main function to be mechano- effectory. whereas the several types of more delicate setae on the Mx2 basis suggest a mechanosensory function, a chemosensory function, or a combination of the two. If true, this gives the Mx2 bases a key role in sensing the quality of the potential food particles. Comparison with other decapods For most decapod taxa there is a lack of behavioral data, but a few earlier studies do describe some functional mor- phology accompanied by movement patterns (Anomura: Nicol, 1932; Roberts, 1968; Kunze and Anderson, 1979; Schembri. 1982a; Zainal. 1990. Thalassinidea: Stamhuis et al., 1998. Palinura: Suthers and Anderson, 1981. Brachyura: Schembri. 1982b. Astacidea: Barker and Gibson, 1977; Lavalli and Factor, 1995. Caridea: Moore et al., 1993. Penaeidea: Hunt et al., 1992). Most of these data concern anomurans, and it seems as if morphology and organization of the mouth apparatus are very similar for pagurids and most galatheids. One of the more detailed observations on mouthpart function comes from Schembri's ( 1982a) study on the shallow-water hermit crab Pagurus rubricatus, which grooms the anterior body (antennae, eyes, and mouthparts), gathers sediment and other prey items, and processes potential food by the Mxp2 FUNCTIONS OF MUNIDA MOUTHPARTS and Mxp3 endopods in a way similar to that reported here for Munida. The resemblance is especially pronounced in the handling and maceration of large, soft prey items. The cutting- crunching process of the mandible and associated structures proceeds in much the same way, although P. rubricatus seems to hold the prey with the mandibular palp and the labium. The palp and labium were also observed to push the food directly into the mouth, but we find this unlikely, because their movements must be perpendicular to the mouth opening (Schembri. 1982a, fig. 8). From his figure 8 it seems more likely that these structures deliver the food to the coxae of Mxl, Mx2, and Mxpl, which control ingestion. We agree with Schembri (1982a) in grouping together the Mxp2 and Mxp3 exopods, and the functions are identical: creating currents around the buccal field. The pattern is a little dissimilar, since in P. rubricatus a flagel- lum on the Mxpl exopod also contributes to the currents around the buccal field, and the flagella beat on one side at a time producing an asymmetrical flow, which shifts when the animal changes the side of beating. Schembri (1982a) also observed that the flagellar currents merely place the particles in the respiratory flow, which thereafter causes the rejection of unwanted particles. This is clearly not the case for Munida, since the flagellar currents are much stronger than the respiratory ones. Schembri (1982a) made some very interesting obser- vations on the function of the bases of Mxl and Mx2. P. rubricatus sorts the sediment in a different way from what we observed for Munida. The Mx2 bases seemed to be the most active, collecting the sediment from the Mxp2 endopods (Mxpl bases are not involved) and pressing it through cuspidate setae on the medial edge of the Mxl bases, which then serve as a passive filter. We emphasize this because it could indicate that very similar structures (even at the setation level) can have different movement patterns and thereby serve very different func- tions. One should notice that Schembri (1982a) did not use video-recordings, without which 3-4 Hz movements (observed for Munida) can be difficult to follow. Similar movement patterns for Mxl and Mx2 are not mentioned in other studies on pagurids (Roberts, 1968; Kunze and Anderson, 1979). In general, hermit crabs possess additional feeding mechanisms that we did not observe in Munida. These mechanisms are antennulary filter feeding (Kunze and Anderson. 1979: Schembri. 1982c: Manjulatha and Babu, 1991), suspension feeding (Gerlach el ai. 1976; Schem- bri. 1982c). and gravel scrubbing (Orton, 1927; Roberts. 1968; Greenwood, 1972; Schembri, 1982a, c). The latter is the most significant and involves picking up pieces of gravel with the chelipeds and handing it over to the mouthparts, where the biofilm is scrubbed off. In P. rubricatus, the Mxp2 and Mxp3 endopods hold and turn the gravel, while the Mx2 and Mxpl bases scrub it off with "vigorous" movements. Compared to Munida, where the chelae are primarily used in aggressive behav- ior (Berril, 1970; pers. obs.), pagurids generally seem to make more extensive use of their chelae and cristae dentatae to macerate food items. Nicol (1932) studied feeding in five species of gala- theids (Galathea dispersa, G. squamifera, G. strigosa. Munida rondeletii (=rugosa), and Porcellana longicor- nis). Her description of the gross morphology of the mouthparts closely resembles what we found for Munida sarsi and M. tenuimana. The functions she ascribed to the Mxp2 endopods. Mxp3 endopods, and mandibles are also very similar to those reported here. The Mxp3 endopods of Porcellana longicornis are somewhat different than those of other galatheids, since they have long plumose setae used to entrap particles in suspension. The general function, however, is the same — to collect prey and trans- fer it toward the mouth. Our observations also agree with those of Zainal ( 1990), who reports on both morphology and function of the mouthparts of Munida rugosa, but with limited detail. Outside the Anomura, the most detailed data comes from mud shrimps, Thalassinidea. Stamhuis et ai (1998) give video-based information on most of the mouthpart functions of the thalassinid Callianassa subterranea during sediment sorting. The animal lives as a selective deposit feeder in mud burrows; due to this specialized way of living, the mouthparts of C. subterranea and especially their setation are somewhat different from that described for Munida. Still. Mxp3 endopod is used for collecting and transporting the food items, whereas the Mxp2 endopod is used for sorting the sediment by size. The sediment is not further sorted, and the bases of Mxpl and Mx2 are merely used for transporting the particles towards the mouth. However, these appendages move with the same frequency as we report for Munida, 3-4 Hz. Very unlike Munida, C. subter- ranea does not use its Mxl during deposit feeding and uses Md only to open or close the mouth (Stamhuis et al., 1998). The movements of the maxillipeds of C. subterranea are also much more stereotypical than what we find for Munida. It has to be kept in mind that Stamhuis et al. analyzed only one type of feeding (deposit feeding), which of course will reduce the functional scheme of the mouthparts. There is morphological evidence (e.g.. well-developed mandibles, spines on maxilla 1. and crista dentata) that C. subterranea possesses other feeding strategies; it is most likely also carnivorous (Stamhuis et al., 1998). Hunt et al. ( 1992) studied the role of "the anterior mouth- parts" (Md, Mxl, Mx2. Mxpl, Lb, and paragnaths) in the penaeid prawn Penaeus merguiensis. Much of the morphol- ogy is similar to Munida, but the mandibular palp is differ- ent and does not take part in feeding. When eating large prey items, Md, paragnaths, Mx2, and Mxpl move laterally and Lb moves anteriorly to make room, while Mxl basa put 296 A. GARM AND J. T. H0EG the prey between the incisor processes for cutting (Hunt et al. 1992). Afterwards the Lb alone pushes the prey towards Mxl coxae for ingesting. When feeding on small particles, P. mergiiiensis does not sort the sediment as Munida does, but filters them out of the water column by using pappose setae on Mxpl and on the bases of Mxl and Mx2 (Hunt et al., 1992). The currents around the buccal area from which the particles are filtered are created by the respiratory beat- ing of Mx2, since P. mergiiiensis lacks flagella on the maxillipeds. What these comparisons show is that within hermit crabs and squat lobsters there are great similarities in mouthpart morphology, even at the setation level. De- spite some differences, the functions reported for other squat lobsters and for hermit crabs are more or less comparable to our findings for Munida. Therefore, there is reason to believe that the functional groupings we suggest could apply to other anomurans; it goes without saying that more data are needed from other species before any final conclusions can be drawn. The additional feeding strategies described for some hermit crabs (i.e., Schembri, 1982c) indicate that more functional groups must be added. It would be of major interest to gain information from the Lithodidae and Porcellanidae, the two other groups of anomurans that have either a differ- ent size range than the Galatheidae and Paguridae (most lithodid crabs are very large) or a different way of eating (porcellanid crabs are filter-feeders). Very little informa- tion is available for decapod taxa other than the Ano- mura. Not surprisingly, the two comparable studies (Hunt et al., 1992; Stamhuis et al.. 1998) show a great diversity of mouthpart morphology and function within the Deca- poda, and the functional groupings suggested here do not seem appropriate for all taxa. Acknowledgments We thank the Coast Guard of the Faeroe Islands for the use of their ship Tjaldur. We also thank Grethe Bruntse for assistance with the field work and the Kaldbak Lab- oratory for generously allowing us to use their facilities. We are grateful for the material and data supplied by the BIOFAR I collection and by Dr. Andre Freiwald, Bre- men. We appreciate the financial support given by The Hede Nielsen Family Foundation and The Danish Society of Natural History (to AG). JTH also gratefully acknowl- edges grant nos. 94-01636 and 96-01405 from the Danish Natural Science Research Council and nos. 970381/40- 1228 and 950260/40-1 190 from The Carlsberg Founda- tion. We thank Nikolai Konow for never-ceasing interest in our study and Dr. G. Walker for some constructive comments. Literature Cited Barker, P. L., and R. Gibson. 1977. Observations on the feeding mechanism, structure of the gut, and digestive physiology of the European lobster Homarus gammarus (L.) (Decapoda: Nephropidae). J. Exp. Mar. Biol. Ecol. 26: 297-324. Bauer, R. 1981. Grooming behavior and morphology in the decapod Crustacea. J. Crustacean Biol. 1: 153-174. Bauer, R. 1989. Decapod crustacean grooming: functional morphology, adaptive value, and phylogenetic significance. Pp. 49-73 in Crusta- cean Issues. Vol. 6. B. E. Felgenhauer, L. Watling, and A. B. Thistle, eds. A. A. Balkema, Rotterdam. Berril, M. 1970. The aggressive behavior of Munida sarsi (Crustacea: Galatheidae). Sarsia 43: 1-1 1. Cartes, J. E. 1993. Diets of two deep-sea decapods: Nematocarcinus exilis (Caridea: Nematocarcinidae) and Munida tenuimana (Anomura: Galatheidae) on the western Mediterranean slope. Ophelia 37: 213- 229. Farmer, A. S. 1974. The functional morphology of the mouthparts and pereiopods of Nephrops noniegicns (L.) (Decapoda: Nephropidea). J. Nat. Hist. 8: 121-142. Felgenhauer, B. E. 1987. Techniques for preparing crustaceans for scanning electron microscopy. J. Crustac. Biol. 7: 71-76. Garm, A., and J. T. Hoeg. 2000. Functional mouthpart morphology of the squat lobster Munida sarsi Huus with comparison to other anomu- rans. Mar. Binl. 137: 123-138. Gerlach, S. A., D. K. Ekstram, and P. B. Eckardt. 1976. Filter feeding in the hermit crab, Pagurus bernhardus. Oecologia 24: 257-264. Greenwood, J. G. 1972. The mouthparts and feeding behaviour of two species of hermit crabs. J. Nat. Hist. 6: 325-337. Hunt, M. J., H. Winsor, and C. G. Alexander. 1992. Feeding in penaeid prawns: the role of the anterior mouthparts. J. Exp. Mar. Binl. Ecol. 160: 33-46. Kunze, J., and D. T. Anderson. 1979. Functional morphology of the mouthparts and gastric mill in the hermit crabs Clibanarius taenia- tus. Clibanarius virescens, Paguristes squamosus and Dardanus setifer (Anomura: Paguridae). Aust. J. Mar. Freshwater Res. 30: 683-72 1 . Lavalli, K. L.. and J. R. Factor. 1992. Functional morphology of the mouthparts of juvenile lobsters, Homarus americanus (Decapoda: Ne- phropidae), and comparison with the larval stages. J. Crustac. Biol. 12: 467-510 Lavalli, K. L., and J. R. Factor. 1995. The feeding appendages. Pp. 349-393 in Biology of the Lobster Homarus americanus. J. R. Factor, ed. Academic Press. San Diego, CA. Manjulatha, C., and D. E. Babu. 1991. Functional organisation of mouth parts, and filter feeding, in Clibanarius longitarsus (Crustacea: Anomura). Mar. Biol. 109: 121-127. Moore, P. G., P. S. Rainbow, and R. J. Larson. 1993. The mesopelagic shrimp Notosiomus robustus Smith (Decapoda: Oplophoridae) ob- served in situ feeding on the medusan Atolla wyvillei Haeckel in the Northwest Atlantic, with notes on gut contents and mouthpart mor- phology. J. Crustac. Biol. 13: 690-696. Nicol, E. A. T. 1932. The feeding habits of the Galatheidea. J. Mar. Biol. Assoc. UK 18: 87-105. Nerrevang, A., T. Brattegard, A. B. Josefson. J. A. Sneli, and O. S. Tendal. 1994. List of Biofar stations. Sarsia 79: 165-180. Orion, J. H. 1927. On the mode of feeding of the hermit crab Enpagurus bemhardus, and some other Decapoda. J. Mar. Biol. Assoc. UK 14: 909-921. Pike, R. B. 1947. Galathea. L. M. B. C. Memoirs XXXIV. Proc. Trans. Liverpool Biol. Soc. 55: 1-179. Roberts, M. H. 1968. Functional morphology of mouth parts of the FUNCTIONS OF MUNIDA MOUTHPARTS 297 hermit crabs, Pagurus longicarpus and Pagitrus pol/icaris. Chesapeake Sci. 9: 9-20. Schembri, P. J. 1982a. Functional morphology of the mouth pans and associated structures of Pagurus rubricates (Crustacea: Decapoda: Anomura) with special reference to feeding and grooming. Zoomor- phology (Berlin) 101(1): 17-38. Schembri, P. J. 1982b. The functional morphology of the feeding and grooming appendages of Ebalia tuberosa (Crustacea: Decapoda: Leu- cosiidae). J. Nat. Hist. 16: 467-480. Schembri, P. J. 1982c. Feeding behavior of 15 species of hermit crabs (Crustacea: Decapoda: Anomura) from the Otago region, southeastern New Zealand. J. Nat. Hist. 16: 859-878. Stamhuis, E. J., B. Dauwe, and J. J. Videler. 1998. How to bite the dust: morphology, motion pattern and function of the feeding append- ages of the deposit-feeding thalassinid shrimp Callianassa subterra- nea. Mar. Bio/. 132: 43-58. Suthers, I. M. 1984. Functional morphology of the mouthparts and gastric mill in Penaeus plehejus (Decapoda: Penaeidea). Aust. J. Mar. Freshwater Res. 35: 785-792. Suthers, I. M., and D. T. Anderson. 1981. Functional morphology of mouth parts and gastric mill of Ibacus peronii (Palinura: Scyllaridae). Aust. J. Mur. Freshwater Res. 32: 931-944. Walossek. D. 1998. On the Cambrian diversity of Crustacea. Pp. 3-29 in Crustaceans and the Biodiversity Crisis, Vol. I, F. Schram and J. C. v. Vaupel Klein, eds. Brill, Leiden. Zainal, K. A. Y. 1990. Aspects of the biology of the squat lobster, Munida rugosa (Fabricius. 1775). Ph.D. dissertation. University of Glasgow, 180 pp. Reference: Biol. Bull. 200: 29H-304. (June 2001) Phylogeny of Hydrothermal-Vent-Endemic Gastropods Alviniconcha spp. from the Western Pacific Revealed by Mitochondrial DNA Sequences S. KOJIMA1 •*, R. SEGAWA2, Y. FIJIWARA3. K. FUJIKURA3, S. OHTA1, AND J. HASHIMOTO3 lOcean Research Institute, University of Tokyo, 1-15-1 Minamidai, Nakano, Tokyo 164-8639. Japan; 'Department of Biolog\. Faculty of Science, Tokyo Metropolitan University, 1-1 Minami-Osawa, Hachioji, Tokyo 192-0364, Japan: and ~ 'Japan Marine Science and Technology Center, 2-15 Natsushiina, Yokosuka, Kanagawa 237-0061, Japan Abstract. Mitochondrial genes for cytochrome oxidase I (COI) from hydrothermal- vent- endemic gastropods of the genus Alviniconcha were sequenced to determine the phy- logenetic relationships among specimens from three areas in the western Pacific. Individuals of Alviniconcha hessleri were collected at two vent fields (depths 1470 m and 3600 m) in the Mariana Trough. Specimens collected in the North Fiji Basin could be divided into two genetically distinct groups, both of which also differed from A. hessleri from the Mariana Trough. None of the specimens of the genus Alviniconcha collected in the Manus Basin differed genetically from the dominant group from the North Fiji Basin. We suggest that the specimens of the genus Alviniconcha analyzed in the present study can be tenta- tively classified into A. hessleri and two undescribed spe- cies. Introduction Since the late 1970s, various biological communities supported by chemosynthetic energy have been discovered in deep-sea reducing environments, such as hydrothermal vent fields and seep areas (Sibuet and Olu, 1998; Van Dover, 2000). The elucidation of dispersion, isolation, and speciation of endemic species is one of the most important issues of deep-sea biology (Vrijenhoek. 1997). Comparative studies of the genetic structure of their populations will Received 16 June 2000; accepted 12 February 2001. * To whom correspondence should be addressed. E-mail: kojima@ori. u-tokyo. ac.jp. provide useful information about such evolutionary pro- cesses. The population structures of some dominant endemic species in deep-sea reducing environments have been ana- lyzed by electrophoretic examination of allozymes (Vrijen- hoek, 1997). Analyses based on DNA sequences would reveal more details of the structures of such populations. To date, population analyses using DNA markers have been reported for just a few species (for review, see Tyler and Young. 1999). In addition, most population genetic studies of organisms in deep-sea reducing environments have been performed on specimens collected in the eastern Pacific and the Atlantic. The accumulation of genetic information about endemic species in deep-sea reducing environments in the western Pacific, where the chemoautosynthesis-based com- munities have unique species composition (Tunnicliffe et ai, 1998). should provide more clues to the processes and mechanisms of evolution of fauna in deep-sea reducing environments. Gastropods of the genus Alviniconcha, which live in symbiosis with chemosynthetic bacteria (Stein et ai, 1988; Endow and Ohta. 1989), are one of the most dominant groups in the chemoautosynthesis-based communities at some hydrothermal vent sites in the western Pacific back- arc basins, namely, the Mariana Trough (Craig et ai, 1987; Hessler and Lonsdale. 1991; Fujikura et ai. 1997). the Manus Basin (Both et ai. 1986; Tufer, 1990; Auzende et ai. 1997; Hashimoto et ai. 1999). the North Fiji Basin (KAIYO 87 Shipboard Party, 1988; Desbruyeres et ai. 1994), and the Lau Basin (NAUTILAU Group, 1990; Des- bruyeres et ai, 1994) (Fig. 1). 298 PHYLOGENY OF ALVINICONCHA SPP 299 O Figure 1. Collection sites: CMT. Central Mariana Trough: SMR. Southern Mariana Ridge; MB, Manus Basin: NFB. North Fiji Basin. An open circle denotes the location of the Lau Basin. Okutani and Ohta ( 1988) described the species Alvinicon- cha hessleri on the basis of specimens collected in the hydrothermal fields along the spreading axis of the central Mariana Trough (3650 m depth). Further populations of A. hessleri were discovered in a vent area at a shallower site on the southern Mariana Ridge, at a depth of 1470 m (Johnson el ai. 1993). No distinct morphological differ- ences were apparent in the specimens from these two vent sites (Hasegawa et al., 1997). Populations of Alviniconcha gastropods have also been discovered at the hydrothermal vents in the Manus Basin, the North Fiji Basin, and the Lau Basin. On the basis of slight differences in the radula and the shell. Beck (1991) proposed that Alviniconcha in the Manus Basin might be a subspecies of A. hessleri. Because distinctive morphological characteristics are scarce, information at the molecular level is necessary to establish exact taxonomic relationships among species of the genus Alviniconcha. To date, only one molecular analysis of this genus has been published (Denis et ill.. 1993). In that study, an electrophoretic analysis of 12 enzymes showed that the populations of Alviniconcha gas- tropods in the North Fiji Basin and the Lau Basin are genetically distinct from one another. Molecular analysis of populations in the other hydrothermal vent sites will provide a useful basis for judgments about the taxonomic status of each population. In the present study, we used nucleotide sequences of mitochondria! DNA to analyze the phyloge- netic relationships among populations of the genus Alviniconcha in the Mariana Trough, the North Fiji Basin, and the Manus Basin. Materials and Methods During dives of the submersibles Shinkai 2000 and Shinkai 6500 of the Japan Marine Science and Technol- ogy Center (JAMSTEC) and Ncnitile of Institut Francais de Recherche pour 1' Exploitation de la Mer (IFREMER), 40 specimens of Alviniconcha hessleri and 35 of the genus Alviniconcha were collected, as summarized in Table 1. Mitochondrial DNA (mtDNA) was extracted from the deep-frozen head-foot region of each individual by a mod- ified version of the method of Komm et al. (1982). In the case of some damaged samples, which had been collected in the North Fiji Basin in 1990 and stored at -20°C, total DNA was extracted by grinding the tissue, digestion with sodium dodecyl sulfate (SDS) and proteinase K, and extrac- tion with phenol and chloroform. A fragment (about 450 bp) of the mitochondria! gene for cytochrome oxidase I (COI) was amplified by the polymer- Table 1 Summary of collection information for Alviniconcha specimens Species Sampling site Depth (m) Submersible Dive # Sample # Alviniconcha hessleri Alice Springs. Central Mariana Trough 3600 Shinkai 6500 D15? CMT-92-1-20 Forecast Vent, Southern Mariana Ridge 1470 Shinkai 6500 D186 SMR-93-1-11 D350 SMR-96-1-9 Alviniconcha spp. PACMANUS site, Manus Basin 1630 Shinkai 6500 D914 MB-96-1-20 White Lady site. North Fiji Basin 1970 Nanlilc D12 NFB-90-1-4 Shinkai 6500 D77 NFB-91-l-t D80 NFB-91-5. 6 STARMER II site. North Fiji Basin 1980 Nautile D15 NFB-90-5, 6 D20 NFB-90-7-9 300 S. KOJIMA ET AL ase chain reaction (PCR) with universal metozoan primers. COI-3, 5'-GTNTGRGCNCAYCAYATRTTYACNGT-3', and COI-6, 5'-GGRTARTCNSWRTANCGNCGNG- GYAT-3' (Shimayama et al.. 1990). The conditions for PCR were as follows: 94°C for 60 s; then 30 to 40 cycles at 92°C for 40 s, 40°C for 60 s, and 72°C for 90 s. Genere- leaser (BioVenture Inc., Murfreesboro, TN) was used to sequester products of cell lysis that might have inhibited the polymerase. The nucleotide sequences (306 bp) were determined for both strands of PCR products by the dideoxynucleotide chain-termination method using a Seque- nase PCR product sequencing kit (United States Bio- chemical Coop.. Cleveland, OH) and primers COI-3 and COI-6. A longer fragment (about 960 bp) of COI, which contains the region mentioned above, was amplified by PCR using Genereleaser and primers COI-B. 5'-GGATGAACNGT- NTAYCCNCC-3' (Hasegawa et al., 1996) and COI-6. The conditions for PCR were as follows: 94°C for 60 s; then 30 to 40 cycles at 92°C for 40 s, 50°C for 60 s. and 72°C for 90 s. The nucleotide sequences within the upper region of this fragment were determined bidirectionally by a se- quencer DSQ-2000L (Shimazu Corp., Kyoto, Japan) using primers Gastro-3, 5'-TTAGCTGGTGCTTCNTCNATYY- TNGG-3' (Kojima et al.. 2000) and TW-2, 5'-ACTACR- TARTANGTRTCRTG-3 (Kojima et al.. 1997b). A nucle- otide sequence of a single specimen of A. hessleri from the South Marina Ridge (No. CMT-92-1) was reported in a previous paper (Kojima et al.. 2000). Amino acid sequences of COI were deduced by reference to the modified genetic code of molluscan mtDNA (Shimayama et al.. 1990; Hoff- mann et al., 1992). The genetic distances between haplotypes were calcu- lated by Kimura's two-parameter method (Kimura, 1980). Phylogenetic trees were constructed by the neighbor-joining method (Saitoh and Nei. 1987) using MEGA (Kumar et al.. 1993) and the maximum parsimony method using the heu- ristic search approach of the computer program PARSI- MONY, which was provided by Dr. K. Tamura of Tokyo Metropolitan University. Ifremeria nautilei, a hydrother- mal-vent-endemic gastropod species closely related to Alviniconcha (Beck, 1991; Waren and Bouchet, 1993), was used as an outgroup for phylogenetic analysis (Kojima et al.. 2000). Differences in frequencies of haplotypes between popu- lations were examined by the exact test of population dif- ferentiation (Raymond and Rousset, 1995) using ARLE- QUIN (Schnider et al., 1996) and the randomized chi- squared test of independence (Raff and Bentzen, 1989). Chi-squared values were generated from 1000 simulated random samplings of the data. An unbiased fixation index, FST (Weir and Cockerham, 1984). was estimated, and the significance of the indices was tested by a nonparametric permutation approach using ARLEQUIN (Schnider et al., 1996). Results Partial sequences (696 bp) of mitochondria! genes for COI were determined from 40 specimens of Alviniconcha hessleri and 35 specimens of Alviniconcha spp. Sequences of all individuals of the genus Alviniconcha collected in the North Fiji Basin and the Manus Basin were distinct from those of A. hessleri from the Mariana Trough. In addition, the sequences of three specimens of the genus Alviniconcha collected during a single dive (Dive 20 of Nautile) at the STARMER II site in the North Fiji Basin were very differ- ent from those of other specimens from the same area. These other North Fiji Basin specimens included two from the STARMER II site and 10 from the White Lady site. The STARMER II site is only 120 m southwest of the White Lady site in the North Fiji Basin. Individuals with mito- chondrial DNA of the dominant type and those with DNA of the less frequent type are referred to hereafter as Alviniconcha spp. type 1 and Alviniconcha spp. type 2, respectively. Figure 2 shows nucleotide sequences from Alviniconcha hessleri. Alviniconcha spp. type 1, and Alviniconcha spp. type 2. A single amino acid substitution was detected between Alviniconcha spp. type 1 and the others (Fig. 2). Within the sequences from each of A. hessleri. Alviniconcha spp. type 1, and Alviniconcha spp. type 2. no amino acid substitutions were detected. Twenty-seven, thirteen, and three haplotypes were ob- tained for A. hessleri. Alviniconcha spp. type 1, and Alviniconcha spp. type 2, respectively. Figure 3 shows the phylogenetic relationships among the specimens of the ge- nus Alviniconcha. Specimens of A. hessleri. Alviniconcha spp. type 1, and Alviniconcha spp. type 2 formed separate monophyletic clusters. There were greater genetic varia- tions among the three clusters of Alviniconcha than within each cluster. The monophyly of A. hessleri, Alviniconcha spp. type 1, and type 2 was supported by high bootstrap probabilities (93%, 100%, and 100%, respectively). Alviniconcha hessleri, Alviniconcha spp. type 1, and type 2 formed monophyletic groups in all maximum parsimony trees as well (data not shown). Three of 27 haplotypes of A. hessleri were discovered in the specimens from both the central Mariana Trough and the southern Mariana Ridge (Fig. 3). Statistical anal- ysis, by the exact test of population differentiation (Ray- mond and Rousset. 1995) and the randomized chi- squared test of independence (Raff and Bentzen. 1989), showed a genetic difference between the populations in the central Mariana Trough and on the southern Mariana Ridge (P = 0.035 and P = 0.023, respectively). In the case of Alviniconcha spp. type 1. four haplotypes were discovered among the specimens from both the North Fiji PHYLOGENY OF ALVINICONCHA SPP 301 Ah ATT ACT GCA GTA TTA TTA TTA CTT TCT TTA OCA GTT TTA GCA GGT GCT ATT ACA ATG OT TTA ACA GAT OGA Al G R A2 YC C C... C. Ah AAT TTY AAT ACT GCT TTC TTT GAC CCA GCT GGA GGT GGT GAT CCA ATT TTA TAY CAA CAT TTA TTT TGA TTC Al C T..R..Y..R C..G A2 C G C..T C C Ah TTT GGG CAC OCA GAA GTT TAT ATT TTA ATT CTT OCT GGC TTT GGA ATG ATT TCT CAT ATT GTT AGA CAT TAT A1..Y R C... A2 T T A C C... Ah TCA GCY AAG AAA GAA ACY TTT GGG ACA CTT GGA ATA ATY TAT GCA ATG ATA GCA ATT GGT TTA TTA GGT TTT Al T T G..Y A C A2 T T A G..T A..G C Ah ATT GTA TGA GCT CAY CAT ATG TTT ACA GTA GGT ATG GAT GTA GAT ACT CGR GCT TAY TTT ACA GCA GCC ACT Al T A T..C T... A2 C A T Ah ATA ATT ATT GCY GTT CCA ACT GGA ATT AAG GTY TTC AGY TGA CTY GCC ACA ATT CAT GGS TCA AAA ATC AAG Al Y T T T..Y C..A T..R A2 T T T T..T Y A T... Ah TAT GAA ACT CCT ATG CTT TGA GCT HA GGR TTT ATT TTC TTG TTT ACT GTA GGG GGT TTA ACT GGA ATT GTT Al G YC A2 A..C A A. .A Ah CTT TCT AAT TGY TCY YTA GAT ATT ATG ATG CAY GAY ACT TAG TAT GTT GTA GCY CAT TTC CAT TAC GTR CTA Al C..TC T^ ..C ..C T ..C G ..R A2 T..TT T..T T C T..TT. Ah TCA ATR GGR GCA GTC TTY GCC YTA TTT GCA GCT TTT AAC TAT TGA TTY CCA TTA ATR AGA GGG GTA ACA CTT Al G..A C..YT Y C G..R A2 A..A..G T..TC T..C C G..G Ah CAC TCT CGT TGA ACA AAA GCT CAT TTT TAT ATY ATR TTT ATY GGR GTA Al Y..G T..R... A2 ..T A C T ..G T ..A ... Figure 2. Nucleoude sequences of mitochondnal genes for cytochrome oxidase I from Alviniamclui lu-sxkn (Ah). AMniconcha spp. type 1 (Al ), and Alviniconc/ia spp. type 2 (A2). Dots indicate nucleotides identical to those in Ah. An underlined codon encodes an amino acid different from that in Ah. R denotes G or A: Y is T or C: S is G or C. Basin and the Manus Basin (Fig. 3), and there was no P = 0.705 for the randomized chi-squared test of inde- genetic difference between the two populations (P - pendence). Although the FST value between the two 0.793 for the exact test of population differentiation, and populations of A. hessleri was significant (P < 0.05) and 302 S. KOJIMA ET AL. 100 0.01 100 ~ 100 I— SMR-93-9 (AB051768) — SMR-96-2{AB051770) — SMR-93-2(AB051772) — CMT-92-3(AB051774) — CMT-92-2(AB051766) CMT-92-14, SMR-93-10, SMR-96-9 (AB051764) I— SMR-96-5(AB051775) — CMT-92-5 (AB051773) — SMR-96-1 (AB051769) I CMT-92-1 8 (AB051765) T— CMT-92-7 (AB051777) S~MR-93-7 (AB051780) SMR-96-8 (AB051785) I— CMT-92-13 (AB051778) -|B1 [-CMT-92-8, 10 (AB051788) ^H CMT-92-4, 6, 15, 17 (AB051783) 93l-SMR-93-3 (AB051787) 94j- CMT-92-9 (AB051789) ~^ CMT-92-1 6 (AB051784) 85 1— CMT-92-1 1, SMR-93-11 (AB051790) [I SMR-93-4, SMR-96 4 (AB051771) "— SMR-93-8 (AB051781) CMT-92-1, 19, SMR-96-3 (AB051767) >— SMR-96-7(AB051779) iSMR-93-1,4,6, SMR-96-6 (AB051776) "— CMT-92-1 2 (AB051 782) -CMT-92-20 (AB051786) I MB-96-3, 12, 17, NFB-90-6 (AB051800) 75 f1— MB-96-1 1 (AB051801) -MB-96-20 (AB051803) 10o f~~- NFB-90-1 (AB051802) -MB-96-1 3, 18 (AB051796) 85 h- NFB-90-2 (AB051795) -MB-96-10, 14, NFB-91-1 (AB051792) NFB-90-5 (AB051797) MB-96-1, 2, 5, 6, 8, 9, 16, NFB-90-3, 4, NFB-91-1, 5, 6 (AB051791) jMB-96-4, 15, NFB-91-2 (AB051793) ""-MB-96-7 (AB051798) r-~B-91-3(AB051794) MB-96-1 9 (AB051799) NFB-90-9 (AB051806) NFB-90-7 (AB051804) NFB-90-8 (AB051805) Ah A1 A2 Figure 3. Phylogenetic tree for specimens of Alviniconcha spp. The tree was constructed by the neighbor- joining method with Ifremeria naiitilei (the most dominant haplotypes in the populations of the Manus Basin and the North Fiji Basin) as an outgroup. Numbers refer to specimens indicated in Table 1. Bootstrap values are shown above or below branches of clades that are supported by bootstrap values of more than 70%. Ah, A 1 . and A2 denote Alviniconcha hessleri. Alviniconcha spp. type 1, and Alviniconcha spp. type 2, respectively. The nucleotide sequence of each specimen will appear in the GSDB. DDBJ. EMBL, and NCBI nucleotide sequence databases under the accession number in parentheses. estimated to be 0.047, that of Alviniconcha spp. type 1 was not significant (P > 0.05). Discussion In the present study, we analyzed the phylogenetic rela- tionships among populations of the Alviniconcha gastropods in hydrothermal areas in the western Pacific. Both the neigh- bor-joining (NJ) method and the maximum parsimonious (MPl method showed that the 43 haplotypes identified among 80 specimens of the genus Alviniconcha formed three monophyletic clusters (Fig. 3). In the NJ tree, the monophyly of each of the three clusters was supported by a high bootstrap probability. The three clusters of Alvinicon- cha were distinct from one another, with a smaller range of sequence variations within clusters than among clusters. Therefore, we suggest tentatively that the specimens of the genus Alviniconcha analyzed in the present study should be classified as A. hessleri and two undescribed species. Although the habitat on the southern Mariana Ridge is much shallower than the type locality of A. hessleri (the central Mariana Trough), all haplotypes identified from specimens collected at two vent sites formed a single cluster (Fig. 3). Although the exact test of population differentia- tion, a randomized chi-squared test, and the test of signifi- cance of the FST value showed that a genetic difference between the two populations was significant at the 5% level, they shared three haplotypes. From the pairwise FST value, the absolute number of migrants exchanged between these PHYLOGENY OF ALVINICONCHA SPP. 303 two populations was estimated to be 10.2. These results suggest that these two populations are conspecific and might be connected by significant gene flow. The present results are consistent with the absence of distinct morphological differences between these populations (Hasegawa et til.. 1997). The distribution of conspecific populations in the bathyal zone and the abyssal zone is in marked contrast with the strict depth zonation exhibited by endemic groups that inhabit the seep areas off central Japan, namely, bivalves of the genus Calyptogena (Kojima and Ohta. 1997) and ves- timentiferans (Kojima et al. 1997b). This difference might be due to environmental factors, such as water currents and the topography of the sea floor, or to differences in biolog- ical characteristics among groups. The absence of genetic differences between the popula- tions of Alviniconcha spp. type 1 in the Manus Basin and in the North Fiji Basin provides a typical example of the active interchange of fauna between western Pacific back-arc ba- sins, which was first noted by Hessler and Lonsdale ( 1991 ). For Ifremeria ntnitilei. which is related to Alviniconcha spp.. Kojima et al. (2000) reported that no haplotype was shared between the Manus Basin and the North Fiji Basin and concluded that the two populations should be consid- ered to be conspecific populations that are isolated geo- graphically from one another. The morphology of the larval shell of Alviniconcha is suggestive of planktonic develop- ment, whereas the larval type of Ifremeria remains unclear (Waren and Bouchet. 1993). The difference in degree of genetic differentiation between /. nautilei and Alviniconcha spp. type 1 might be attributed to differences in the larval type (Kojima et al.. 2000). Denis et al. (1993) reported significant genetic differ- ences between a population of Alviniconcha in the North Fiji Basin and one in the Lau Basin, and they proposed that individuals in these two populations should be treated as separate species. The North Fiji Basin is situated near the Lau Basin (Fig. 1 ): thus Alviniconcha spp. type 2 might be conspecific with the Lau Basin species proposed by Denis et al. (1993). Unfortunately, samples from the Lau Basin are not available, so we have been unable to test this hypothesis. To clarify the taxonomic status of these putative species, detailed morphological examination and further molecular analysis are required. The dominance of gastropods of the family Provannidae offers one of the prominent features in some chemoauto- synthesis-based communities in the back-arc basins in the western Pacific. This study and our previous ones (Kojima et a/.. 2000) revealed the population structures of some species in the two representative genera of this family, namely Alviniconcha and Ifremeria. Although many other endemic taxa have been reported from the western Pacific (Desbruyeres and Segonzac 1997: Miura et al., 1997; Oku- tani et al.. 1999; Fujikura et al.. 2000), information about their population structures is still limited (Kojima et al.. 1997a. b). Information about the population structures of many endemic species in deep-sea reducing environments will be needed to reveal details of the evolution in such environments in the western Pacific and to compare them with those in other oceans. Acknowledgments The authors thank the shipboard parties of the STARMER cruises, the Y96-13 cruise, and the BIOAC- CESS-Manus '96 cruise, the operation teams of the sub- mersibles Shinkai 2000, Shinkai 6000, and Nautile, and the officers and crew of the tender ships Natsitshima, Yokosuka, and Nadier for their help in collecting samples. Dr. U. Tsunogai graciously provided specimens of A. hessleri col- lected during Dive 355 of the submersible Shinkai 6500. Dr. K. Tamura kindly provided the computer program PARSI- MONY. Thanks are also extended to Dr. I. Hayashi and two anonymous reviewers for comments that improved manu- scripts. Part of this study was supported by grants from the Science Technology Agency of Japan and from the Ministry of Education. Science, Sports and Culture of Japan (no. 12NP0201). Literature Cited Auzende, J.-M., J. Hashimoto, A. Fiala-Medioni, S. Ohta, and I'equipe Bioaccess. 1997. Etude geologique et biologique in situ de deux zones hydrothermales du bassin de Manus (Papouasie Nouvelle- Guinee). C. K. Acini. Sci. Serie 7/325: 585-591 (in French with English abstract and abridgment). Beck, L. A. 1991. Olgaconcha titfari n. gen. et n. sp. — a new mesogas- tropod (Gastropoda: Prosobranchia) from hydrothermal vents in the Manus back-arc basin (Bismarck Sea. Papua New Guinea). Ann. Naturhisl. Miis. Wien. 92B: 277-287. Both, R., K. Crook. B. Taylor, S. Brogan, B. Chappell, E. Frankel, L. Lui, J. Sinton, and D. Tiffin. 1986. Hydrothermai chimneys and associated fauna in the Manus back-arc basin, Papua New Guinea. EOS Trans. AGU 67: 489-490. Craig, H., Y. Horibe, K. A. Farley, J. A. Welham, K.-R. Kim, and R. N. Hey. 1987. Hydrothermal vents in the Mariana Trough: results of the first Alvin dives. EOS Trans. AGU 68: 1531. Denis, F., D. Jollivet, and D. Moraga. 1993. Genetic separation of two allopatric populations of hydrothermal snail Alviniconcha spp. (Gas- tropoda) from two South Western Pacific back-arc basins. Biochem. Syst. Ecol. 21: 431-440. Desbruyeres, D.. and M. Segonzac ( eds. ). 1997. Handbook of Deep-Sea Hydrothfnnal Vent Fauna. IFREMER. Brest. 279 pp. Desbruyeres, D., A.-M. Alayse-Danet, S. Ohta. and Scientific parties of BIOLAU and STARMER Cruises. 1994. Deep-sea hydrothermal communities in Southwestern Pacific back-arc basins (the North Fiji and Lau Basins): composition, microdistribution and food web. Mar. Geol. 116: 227-242. Endow, K., and S. Ohta. 1989. The symbiotic relationship between bacteria and a mesogastropod snail. Alviniconcha hessleri. collected from hydrothermal vents of the Mariana back-arc Basin. Bull. Jpn. Soc. Microb. Ecol. 3: 73-82. Fujikura, K.. T. Yamazaki, K. Hasegawa, U. Tsunogai. R. J. Stern, H. Ueno, H. Yamamoto. Y. Maki, S. Tsuchida, T. Kodera, H. Yamamoto, C. Sun, and T. Okutani. 1997. Biology and earth 304 S. KOJIMA ET AL. scientific investigation by the submersible "Shinkai 6500" system of deep-sea hydrothermalism and lithosphere in the Mariana back-arc Basin. JAMSTEC J. Deep Sea Res. 13: 1-20 (in Japanese with English abstract). Fujikura, K., S. Kojima, Y. Fujiwara, J. Hashimoto, and T. Okutani. 2000. New distribution records of vesicomyid bivalves from deep-sea chemosynthesis-based communities in Japanese waters. Venus Jpn. J. Malacol. 59: 103-121. Hasegawa, T., T. Yamaguchi, S. Kojima, and S. Ohta. 1996. Phylo- genetic analysis among three species of intertidal barnacles of the genus Tetradita (Cimpedia: Balanomorpha) by nucleotide sequences of a mitochondria! gene. Benthos Res. 51: 33-39 (in Japanese with English abstract). Hasegawa, K., K. Fujikura, and T. Okutani. 1997. Gastropod fauna associated with hydrothermal vents in the Mariana back-arc Basin: summary of the results of 1996 "Shinkai 6500" dives. JAMSTEC J. Deep Sea Res. 13: 69-83. Hashimoto, J., S. Ohta, A. Fiala-Medione, J.-M. Auzende, S. Kojima, M. Segonzac, Y. Fujiwara, J. C. Hunt, K. Gena, T. Miura, T. Kikuchi, T. Yamaguchi, T. Toda, H. China. S. Tsuchida, J. Ishi- bashi, K. Henry, M. Zbinden, A. Pruski, A. Inoue. H. Kobayashi, J.-L. Birrien, J. Naka, T. Yamanaka, C. Laporte, K. Nishimura, C. Yeats. S. Malagun, P. Kia, M. Oyaizu. and T. Katayama. 1999. Hydrothermal vent communities in the Manus Basin. Papua New Guinea: Results of the B1OACCESS cruises '96 and '98. InterRidge News 8(2): 12-18. Hessler, R. R., and P. Lonsdale. 1991. Biogeography of Mariana Trough hydrothermal vent communities. Deep-Sea Res. 38: 185-199. Hoffmann, R. J., J. L. Boore, and W. M. Brown. 1992. A novel mitochondria! genome organization for the blue mussel, Mytilus edu- 1ns. Genetics 131: 397-412. Johnson, L., H. Fryer, H. Matsuda, T. Ishii, and T. Gamo. 1993. Hydrothermal deposits and two magma sources for volcanoes near 13°20'N in the Mariana backarc: a view from Shinkai 6500. EOS Trans. AGU Fall Meet. Suppl. 74: 681. KAIYO 87 Shipboard Party. 1988. Rift system in the North Fiji Basin: Results of Japan-France cooperative research on board KAIYO 87. La Mer 26: 36 — 16 (in Japanese with English abstract). Kimura, M. 1980. A simple method for estimating evolutionary rate of base substitutions through comparative studies of nucleotide sequence. J. Mol. Evol. 16: 111-120. Kojima. S., and S. Ohta. 1997. Bathymetrical distribution of the species of the genus Calyptogena in the Nankai Trough. Venus Jpn. J. Malacol. 56: 293-297. Kojima. S., K. Fujikura, J. Hashimoto, and S. Ohta. 1997a. Distribu- tion and genetic structure of Calyptogena soyoae complex. JAMSTEC J. Deep Sea Res. 13: 691-696 (in Japanese with English abstract). Kojima. S., R. Segawa, J. Hashimoto, and S. Ohta. 1997b. Molecular phytogeny of vestimentiferans collected around Japan revealed by the nucleotide sequences of mitochondria] DNA. Mar. Rid. 127: 505-513. Kojima, S., R. Segawa, Y. Fujiwara, J. Hashimoto, and S. Ohta. 2000. Genetic differentiation of populations of a hydrothermal vent-endemic gastropod. Ifremeria mmtilei. between the North Fiji Basin and the Manus Basin revealed by nucleotide sequences of mitochondria! DNA. Zoo/. Sci. 17: 1167-1174. Komm, B., A. Michaels, and J. l.inton. 1982. Isolation and character- ization of the mitochondria! DNA from the Florida spiny lobster. Panulirus argus. Comp. Bwchem. Physiol. 73B: 923-929. Kumar, S., K. Tamura, and M. Nei. 1993. MEGA: Molecular Evolu- tionary Genetics Analysis Ver. 1.0. The Pennsylvania State University, University Park, PA. Miura. T., J. Tsukahara, and J. Hashimoto. 1997. Lamellibraclua satsiima, a new species of vestimentiferan worms (Annelida: Pogono- phora) from a shallow hydrothermal vent in Kagoshima Bay. Japan. Proc. Biol. Soc. Wash. 110: 447-456. NAUTILAU Group. 1990. Hydrothermal activity in the Lau Basin. First results from the NAUTILAU Cruise. EOS Trans. AGU 71: 678-679. Okutani, T., and S. Ohta. 1988. A new gastropod mollusk associated with hydrothermal vents in the Mariana back-arc basin. Western Pa- cific. Venus Jpn. J. Malacol. 47: 1-9. Okutani, T., K. Fujikura, and S. Kojima. 1999. Two new hadal bi- valves of the family Thyasiridae from the plate convergent area of the Japan Trench. Venus Jpn. J. Malacol. 58: 49-54. Raff, D. A., and P. Bentzen. 1989. The statistical analysis of mitochon- dria! DNA polymorphisms and the problem of small samples. Mol. Biol. Evol. 6: 539-545. Raymond, M., and F. Rousset. 1995. An exact test for population differentiation. Evolution 49: 1280-1283. Saitoh, N., and M. Nei. 1987. The neighbour-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 10: 471-483. Schnider, S., J.-M. Kueffer, D. Roessli, and L. Excoffer. 1996. AR- LEQUIN. a Sojhmre Package fur Population Genetics. Genetics and Biometry Lab., Dept. of Anthropology. University of Geneva. Geneva. Shimayama, T., H. Himeno. J. Sasuga, S. Yokobori, T. Ueda, and K. Watanabe. 1990. The genetic code of a squid mitochondria] gene. Nucleic Acids Symp. Ser. 22: 77-78. Sihuet, M., and K. Olu. 1998. Biogeography. biodiversity and fluid dependence of deep-sea cold-seep communities at active and passive margins. Deep-Sea Res. II 45: 517-567. Stein, J. L., S. C. Cary, R. R. Hessler, S. Ohta, R. D. Vetter, J. J. Childress, and H. Felbeck. 1988. Chemoautotrophic symbiosis in a hydrothermal vent gastropod. Biol. Bull. 174: 373-378. Tufar, VV. 1990. Modern hydrothermal activity, formation of complex massive sulfide deposits and associated vent communities in the Manus Back-Arc Basin (Bismarck Sea. Papua New Guinea). Mitt. Osterr. Geol. Ges. 82: 183-210. Tunnicliffe, V., A. G. McArthur, and D. McHugh. 1998. A biogeo- graphical perspective of the deep-sea hydrothermal vent fauna. Adv. Mar. Biol 34: 351-442. Tyler, P. A., and C. M. Young. 1999. Reproduction and dispersal at vents and cold seeps. / Mar. Biol. Assoc. UK 79: 193-208. Van Dover, C. L. 2000. The Ecology of Deep-Sea Hydrothermal Vents. Princeton University Press, Princeton, NJ. 424 pp. Vrijenhoek, R. C. 1997. Gene flow and genetic diversity in naturally fragmented metapopulations of deep-sea hydrothermal vent animals. J. Hered. 88: 285-293. Waren, A., and P. Bouchet. 1993. New records, species, genera, and a new family of gastropods from hydrothermal vents and hydrocarbon seeps. Zoo/. Scr. 22: 1-90. Weir, B. S., and C. C. Cockerham. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 1358-1370. Reference: Bio/. Bull. 200: 305-320. (June 2001) Ecology and Behavior of Gecarcoidea natalis, the Christmas Island Red Crab, During the Annual Breeding Migration AGNIESZKA M. ADAMCZEWSKA1'* AND STEPHEN MORRIS2 1 Ocean Research Institute, University of Tokyo, 1-15-1 Minamidai, Nakano-ku, Tokyo 164-8639, Japan; and 'Morlab, School of Biological Sciences. Woodland Road, University of Bristol, Bristol BS8 1UG, UK Abstract. The terrestrial crab Gecarcoidea natalis is en- demic to the forests of Christmas Island but must migrate each year to the coast to breed. During 1993 and 1995. radio-tracking, mark and recapture, and counting methods were used to establish the routes, walking speeds, direction of travel, and destinations of migrating crabs, as well as crab numbers and distribution. The density of crabs ranged from 0.09 to 0.57 crabs per square meter, which gave a popula- tion estimate of 43.7 million adult crabs on the island. During the dry season the crabs were relatively inactive but on arrival of the wet season immediately began their mi- gration. The crabs generally walked in straight lines, and most crabs from around the Island traveled toward the northwest shore instead of simply walking toward the near- est shore. The maximum recorded distance walked by a red crab in one day was 1460 m, but the mean was 680 m per day in 1993 and 330 m in 1995. Comparing the 1993 and 1995 study seasons, there was a 3-week difference in the timing of the start of the migration, but the spawning date was fixed by the lunar phase and took place 17 to 18 days after mating. In 1993. late rain prompted a "rushed" migra- tion and crabs walked directly to their shore destinations; in contrast, in 1995 most crabs made stops of 1 to 7 days during the downward migration. By giving the crabs a chance to feed along the way and minimizing the time that the population was concentrated near the shore, these stops may be important in ensuring that the animals have enough food after the long dry season. Furthermore, this behavior Received 23 April 1999; accepted 21 February 2001. * To whom correspondence should be addressed. E-mail: islika@ uri.u.lokyo.ac.jp implies that the crabs are able to judge how far away they are from the shore during the migration. Introduction The red crab Gecarcoidea natalis is endemic to Christ- mas Island and is distributed throughout the entire island (Hicks, 1985: O'Dowd and Lake. 1989; Hicks et al, 1990; Green, 1993). For most of the year the red crabs are found within primary and mature regrowth forest (Hicks et al., 1990; Green, 1997; pers. obs.), but each year these crabs must migrate to the coast to breed. The migration of red crabs is a well-documented phenomenon described by an- ecdotal observations (Andrews, 1900; Gibson-Hill, 1947) but by only limited quantitative data, including some infor- mation on crab densities and walking speeds across cleared areas (Hicks. 1985; O'Dowd and Lake, 1989; Green, 1993, 1997). There are no data on migratory routes, destinations or distances, nor on walking speeds of red crabs within the rainforest. The arrival of the monsoonal rains allows increased ac- tivity of red crabs and stimulates the annual migration (Hicks, 1985; Green. 1993). During this breeding migration red crabs, like other terrestrial gecarcinids, must abandon their home ranges and travel down to the coast to mate and spawn (Garth, 1948; Gifford, 1962; Bliss et al., 1978; Lake and O'Dowd, 1991; Finder and Smits, 1993; Green, 1997). The downward migration normally requires at least a week, and the crabs migrate mainly during the first few hours of the morning and in the late afternoon (Hicks, 1985; Green, 1993). The males excavate burrows, which they must de- fend from other males, on the lowest shore terraces; mating occurs in or near the burrows. Soon after mating the males 305 306 A. M. ADAMCZEWSKA AND S. MORRIS start the journey back inland to the forest, while the females lay their eggs and remain in the burrows for 2 weeks (Hicks. 1985; Hicks et al., 1990). At the end of the incubation period the females vacate their burrows and make their way to the coastal cliffs, which almost completely surround the island, to cast their eggs into the ocean. The females then return to the forest while the crab larvae spend 3-4 weeks at sea before returning to land as juvenile crabs (Gibson-Hill, 1947; Hicks. 1985; Hicks et al., 1990). Since gecarcinid crabs must migrate to the shore to spawn, they must have well-developed navigational mech- anisms. The means of navigation during the migrations are unknown, but visual cues, polarized light, magneto-recep- tion, and learning are thought to be involved (Daumer et al., 1963; Herrnkind, 1968; DeWilde, 1973; Bliss et al., 1978; Lohmann et al.. 1995; Vannini and Cannicci. 1995; Deutschlander et al., 1999). Some data are available on the activity levels of terrestrial crustaceans in the natural environment (e.g., Wolcott and Wolcott, 1985; Gherardi et al., 1988; Gherardi and Vannini, 1989; Micheli et at.. 1990; Weinstein, 1995). but data comparing the activity patterns of gecarcinid crabs during normal foraging activities and during the migration are sparse (see DeWilde, 1973). The ecology of red crabs is characterized by a period of very low activity during the dry season, followed immediately by a breeding migration (Hicks, 1985; Green, 1997) that may require crabs to travel more than 4 km in less than a week. Such abrupt changes in activity levels impose specific demands on the physiology and metabolism patterns of red crabs. The availability of food and water changes on a seasonal basis. Particularly during the migration, this could be a major determinant of the strategy, performance, and diel capacity for exercise. Limitations in food have implications for the importance and use of stored fuel reserves, both during the dry season and during the migratory activities. Thus, determining the distances traveled and the duration of the migration as well as the routes and destinations of the crabs was crucial in appreciating the physiological de- mands. The aim of this study was to examine activity levels of red crabs in the field during the migration compared to the non-migratory season. The emphasis was to characterize the migration of G. natalis in detail, by determining the direc- tion, speed, and distance traveled during the migration, as well the relationship between the point of origin of crabs in the forest and their shore destinations, thus providing a context for the physiological demands during their annual migration. Materials and Methods Christmas Island was formed by a series of geological uplifts. Thus, the structure of the island is reminiscent of a I05°35'E CHRISTMAS ISLAND Indian Ocean - Road I I National Park 0 4km Figure 1. Christmas Island, Indian Ocean, showing areas of National Park and the location of the two study areas. Study Area 1 was located in the relatively undisturbed part of the island; Study Area 2 included the residential and industrial areas. wedding cake, with a central plateau (>200 m above sea level) and a number of cliffs and terraces descending to a rugged shore line with few beaches (Fig. 1; Gray, 1995). Field studies on Christmas Island were carried out in two study areas. The primary study site (Study Area 1) was located on the central plateau, within the Christmas Island National Park (Fig. 1 ). This pail of the Island is covered by primary rainforest and well-established secondary regrowth and has received minimum human disturbance over the past 20 years. Study Area 2, located in the northeast corner of the Island (Fig. 1), included the existing human settlement and airport, an area crisscrossed by many roads. The annual activity cycle of red crabs was divided into a migratory and a non-migratory season. "The migratory sea- son" (initiated by the arrival of the wet season, usually in November or December) is used in this study to collectively refer to the breeding behavior of the crabs, including the annual breeding migration, the related breeding activities, and the return journey from the shore terraces to the forest. The "non-migratory season" includes all other times of the year when the red crabs were not involved in breeding activities. Non-migratory season Population density. To obtain an index of density for red crabs, twelve stations (A to L) were selected; these included sites on the central plateau and on the upper terraces in both study areas (Fig. 2). All stations were located in primary or MIGRATION BEHAVIOR OF RED CRABS 307 D / 2 •••* c/ANCA —• /•* Field Figure 2. Locations of the sites for determining the density of red crabs during February and March 1995 (non-migratory season). Sites A-E were within Study area 1: sites F-L were in Study Area 2. mature regrowth forest, but some were very close to cleared areas. At each station the visible red crabs were counted in each of six grids (9X9 m). All counting was done either in the 2 hours after dawn, a period of foraging activity for red crabs, or during overcast conditions that normally stimulate a similar degree of activity (Hicks. 1985; Green, 1993, 1997). There is very little understory vegetation in the rainforest of Christmas Island, so foraging red crabs are easily visible and provide relatively accurate estimates of the crabs present in the area. Additionally, stations that were populated predominantly by large male crabs were discrim- inated from those with a more mixed population. These data were confirmed as having homogenous variance by using Bartlett's )f test and were then analyzed by analysis of variance (ANOVA: Systat 5.03). Post hoc testing was by Tukey's HSD, and P < 0.05 was taken as significant in all cases. Activity- patterns. Two separate experiments were con- ducted to determine the activity patterns of red crabs during the non-migrator)' season (March 1993; wet season) at station B within Study Area 1 (for location, see Fig. 2). Firstly, crabs were selected at random and observed for a period of 2 min during their normal foraging activities. The distance that each crab walked during the 2 min was re- corded (n = 15). During the second experiment, the actual walking speed of foraging crabs was recorded. A crab was chosen at random and observed until it started walking; the crab was then timed for as long as it continued to walk, and the distance it traveled was recorded (n = 14). All crabs used in these two experiments were subsequently captured, sexed. and measured (carapace width). These observations were made when the red crabs are most active, i.e., when the humidity was over 85% (Green, 1993). Migratory activities Three methods were used to obtain information about the migration of red crabs: radio tracking to obtain precise speeds, routes, and direction of travel; spray-painting large numbers of crabs to gain information on en masse direction of travel; and counting of crabs at designated stations to provide quantitative snapshots of the numbers of crabs present in various parts of the island. In Study Area 1, crabs were studied during the migratory season (November/December) of 1993 and 1995, but in Study Area 2 crabs were studied during the 1 995 migration season only. The behavior of migrating red crabs in a relatively undisturbed area (Study Area 1) was compared to that of crabs in the inhabited part of the island (Study Area 2). Since red crabs have rather small home ranges during most of the year (Lake and O'Dowd. 1991; Green, 1993; pers obs.), the areas within the rainforest selected for color 308 A. M. ADAMCZEWSKA AND S. MORRIS Figure 3. for density c (where Y = Blue. Study Area 1 showing locations of numbered grids (•) used ounts during the migration and sites (•) of color coding crabs yellow) and of radio-tracking "sites of origin" at Silver and coding of the red crabs prior to the start of the migration were assumed to be the sites of origin for those crabs. Radio-tracking of crabs. At the commencement of the migration in 1993, four large male red crabs (mass 400-503 g) were captured at each of two chosen study sites (Blue and Silver sites; for locations see Fig. 3) and fitted with radio transmitters (see below). During the 1995 migration this radio-tracking experiment was repeated by fitting male crabs (mass 415-480 g) originating at those same sites with radio transmitters on 8 and 9 November. Additionally, in 1995, radio transmitters were attached to three large male crabs (mass 460-570 g) at each of two more sites of origin (Yellow and Orange sites; Fig. 4) within Study Area 2, on 9 and 10 November respectively. To attach the radio transmitters to the crabs a plastic sawn-off 20-ml syringe was first glued with an epoxy glue to the top of the carapace of each crab and allowed to set for a minimum of 6 hours. The radio transmitters (Microlite transmitters, model GP1: Titley Electronic, Ballina, NSW, Australia) were then inserted into the syringes with the antennae trailing behind the animals. The transmitter and the plastic syringe housing weighed 20-30 g. The posterio- dorsal surface of each crab was spray-painted with acrylic paint corresponding to the color associated with its site of origin. Each crab was released in exactly the same spot where it was collected. The crabs were tracked daily with a collapsible, direc- tional antenna (A. H. Systems, Chatsworth, CA) connected to a Regal 2000 telemetry receiver (Titley Electronics). Crabs were firstly located by triangulation and, where pos- sible, the location of each crab was confirmed visually. The triangulation technique proved quite reliable, pinpointing the location to within 30 m. The position and time of location for each crab fitted with a radio transmitter was recorded on a topographical map every day. Color-coding of crabs. The posteriodorsal part of the carapace of crabs was spray-painted with nontoxic acrylic spray-paint, avoiding the eyes and mouthparts. In Study Area 1. at the commencement of the migration in 1993 (on the 17 Nov.), 530 crabs (male and female) were spray- painted yellow and 285 crabs were spray-painted blue (for location of sites, see Fig. 3). During the 1995 migration, 684 crabs were spray-painted blue and 750 were spray-painted silver at their respective sites of origin within Study Area 1 (Fig. 3). In Study Area 2 (1995), 5622 red crabs were spray- painted different colors at eight locations within primary and mature regrowth forest in the plateau area (see Fig. 4 and Table 1 for location and the number of crabs painted at Study Area 2 Figure 4. Study Area 2 showing locations of numbered grids (•) used for density counts during the migration and sites (•) of color coding crabs (where Pu = Purple. Gr = Green, W = White, Pi = Pink. B = Brown, Go = Gold) and of radio-tracking "sites of origin" at Orange and Yellow. MIGRATION BEHAVIOR OF RED CRABS 309 Table 1 Summary of 1995 color-coding experiment in Slutlv Area 2 Crabs Site code Date painted* Painted (no.) Sightedt (no.) (%) Purple (Pu) 5 556 70 13 Green (Or) 5-8 999 310 31 Yellow (Y) 5-8 886 236 27 White (W) 5-8 701 269 38 Pink (Pi) 5-8 611 164 27 Brown (B) 7 169 15 9 Orange (O) 6-8 621 204 33 Gold (Go) 6-8 1079 70 6 TOTAL 5622 1338 23 * Day or range of days in November 1995. t Between 8 November and 18 December 1995. each site of origin). In Study Area 2, volunteers searched for painted crabs from 0630 to 1100 hours and from 1500 to 1730 between 8 November and 18 December 1995. To maximize sightings, the surveying took place mainly on and alongside roads and tracks, where color-coded crabs could be easily spotted as they crossed. All sightings reported by ANCA (Parks Australia) staff and local residents were also recorded. Walking speed and density- of crabs during the migration Walking speeds of red crabs were measured for crabs crossing Murray Road (a limestone surface about 10 m wide; Fig. 2). during the downward and return male and female migrations. Observations on the migration patterns, mating activities, and spawning of the red crabs were re- corded daily. During the 1995 migration season, counting stations were set up on roads in Study Area 1 (2 stations; Fig. 3) and in Study Area 2 (9 stations; Fig. 4). At each counting station three counting grids (each 5 X 5 m) were spray-painted on the road. Counts were carried out at each station at 0800 and 1700 hours every day between 5 November and 18 Decem- ber (a total of 87 counts at each station). The density of red crabs per square meter ± SEM was calculated from the mean of the three grids at each site. The data were analyzed by analysis of variance (ANOVA), using Systat for Win- dows version 5.03. Results Non-migratory season Population density: The density of crabs ranged from 0.09 to 0.57/m2. Sites located at the central plateau were occupied predominantly by large male crabs at a mean density of 0.14 ± 0.03/nr. This density was significantly lower than that of mixed populations located mostly at lower altitudes (Fig. 5). The exception was station K, which was on the plateau but inhabited by a mixed population. However, at that part of the plateau the island falls almost directly to the ocean, making it the first accessible habitat for the smaller and juvenile crabs in this part of the island. There was evidence of decreasing crab density with increas- ing elevation in both study areas (Fig. 5); the relationship could be described as follows; crab density (number/m2) = 1.047 - 0.003 (elevation in m); r = 0.47, simple linear regression). Activity patterns. In March 1993, red crabs had finished their migratory activities for the season but, with relative humidity over 90%, were still actively foraging above ground. Foraging red crabs walked intermittently, traversing an average distance of only 0.52 m in 2 min. During locomotion, the walking speeds ranged from 0.62 to 2.0 m • min"1 (mean =: 1.11 ± 0.50 m • min"1). The average carapace width of these crabs was 8.25 cm (range 6.5 - 11.5 cm). Migratory activities The monsoonal rains arrived at different times in the two study seasons, resulting in very different migratory behavior of red crabs. In 1993, the rains were late, migration on the plateau was initiated by heavy local rainfall at noon on 17 November. Within an hour, large numbers of red crabs began accumulating on vehicle tracks and open areas within Study Area 1 and commenced their annual migration. The migration started 1 week after the optimal date predicted on the basis of the lunar cycle, and it lasted only about 6 days (Fig. 6). 0.7 -i 0.6 - ' B 0.5 - G" III E 0.4 - J3 Q 0.3 - u L I I F 0.2 - I I ^ L 0.1 - o - 1 D5°c i 5lE ['iK 1 1 1 1 50 200 220 240 260 280 300 320 Elevation (m) Figure 5. Density/elevation distribution of red crab population, mean number of crabs per nr ± SEM. Letters designate counting site (see Fig. 2 for location map). Closed symbols indicate a crab population of mixed sexes and different sizes. Open symbols indicate crab populations com- posed mostly of large male crabs. Sites from Study Area 1 are shown as circles (O, •); sites from Study Area 2 are shown as squares (D. •). 310 A. M. ADAMCZEWSKA AND S. MORRIS Full c Moon New • < Moon Full Moon New • Moon November December Predicted 1993 migration 1 995 migration Downward Return spawning male 17-23 28-30 12 Nov Nov Dec 5-26 Nov 3-8 Dec 17 Dec Figure 6. Graphical representation of the synchronization of breeding activities of Gecarcoidea natalis with the lunar phase. Predicted migratory activities were based on the descriptions presented by Hicks (1985) and Hicks el al. (1990). The actual timing of the downward migration, the return migration of male crabs, and the spawning recorded in 1993 and 1995 are also presented. The migration of 1995 did not start abruptly on the plateau. A few light showers in October increased the hu- midity, and the red crabs became more active. A small percentage of red crabs were seen on their downward mi- gration during the first week of November, before the main migration. The main migration on the plateau started 12 days before the date predicted by the lunar phase (Fig. 6), and consequently the red crabs were engaged in the down- ward migration for up to 3 weeks during the 1995 season. Migration activity occurred almost exclusively during daylight. On rainy, overcast days the crabs were seen cross- ing roads throughout the day, although the numbers de- creased during the middle of the day. However, on sunny days there was a distinct period of low activity during the middle of the day (1130-1500). During this time large numbers of red crabs accumulated in the bushes by the sides of the roads and resumed their migration later in the after- noon. A few individuals persisted in migrating for several hours after sunset on some days, but there was no migratory activity at night. Migratory routes (Study Area 1) Radio-tracking data showed that red crabs do not neces- sarily migrate to the nearest coast (Fig. 7a, b; Fig. 8a, b). All crabs with radio transmitters released in the Blue site in 1993 covered about 4.13 km to the northwestern shoreline within 6 days (Fig. 7), rather than walking only 1.8 km to the southeastern shore. Furthermore, the crabs did not need to follow contour lines to direct themselves towards the shoreline. Importantly, crabs radio-tracked from the same point of origin followed remarkably similar routes in both study seasons, and they traveled in surprisingly straight lines (Fig. 7a, b; Fig. 8a, b). The silver-coded crabs from the center of the Island also headed northwest, traveling approx- imately 2.4 km in as little as 3-4 days (Fig. 8a, b). The Jedda Cave road, where 500 crabs were painted yellow in 1993 (Fig. 3), was used as a travel route by some individ- uals. Yellow-colored crabs were noted on the road up to 3 days after being painted (Fig. 9a). Other crabs simply crossed the road without taking advantage of easier routes along cleared areas. Sightings of color-coded red crabs (Fig. 9a-c) provided a larger data set confirming that crabs orig- inating from a particular area on the island tend to migrate to similar areas on the shore to carry out their breeding activities. The distances traveled by the red crabs with radio trans- mitters in Study Area 1 varied greatly, from 120 to 1460 m/day. During the "rushed" migration in 1993, seven of the radio-tracked crabs traveled more than 900 m/day, and the average distance traveled in a day was 680 m (Fig. 7a, b). During the 1995 migration, the longest distance traveled in a day was 1000 m, and the average daily distance was 330 m — less than half that recorded during the 1993 migration. During the 1993 migration season, the crabs made progress towards their shore destinations each day (Fig. 7a, b). In contrast, during the 1995 season, 5 of the 7 radio- tracked crabs in Study Area 1 paused in their downward migration to the ocean and remained in one place (within - 20 m diameter) for 2 to 5 days before they resumed their journey to the coast (Fig. 8a, b). Using the 1993 data and assuming that red crabs walk continuously for 12 h per day (i.e., during the daylight hours), the calculated average walking speed during the migration of the crabs with radio transmitters was 1.1 m • min"1. This was considerably slower than the speed of 4.7 ± 0.4 m mm (/; = 17) recorded for red crabs crossing Murray Road ( 10 m wide) during both of the study seasons. During the return migration, walking speeds for returning males (5.1 ± 0.3 m • min~'; n = 13) and returning females (6.2 ± 0.5 m • min~'; n = 7) crossing Murray Road were significantly faster than those of crabs on their down- ward migration (P = 0.016). Migratory routes (Study Area 2) In Study Area 2 the crabs also traveled predominantly northwest (Fig. lOa, b). Radio-tracking showed that crabs traveled along the sides of a road (thus becoming tempo- rarily diverted from their course towards the shore) before eventually crossing the road to reach the terraces (Fig. lOa. b). The crabs with radio transmitters from the Yellow site took between 10 and 13 days to travel 1.2 km to the shore terraces (Fig. lOb). The shore closest to the Orange site was 2.1 km due east, but most of the orange-coded crabs trav- eled more than 3.4 km in the northwest direction, walking for 17-21 days to reach the shore terraces at a point very close to the crabs from the Yellow site (Fig. lOa, b. Fig. 1 1 ). Three crabs traveled over 1 km in a day, but the average MIGRATION BEHAVIOR OF RED CRABS 311 A B Figure 7. Migratory paths determined for radio-tracked crabs originating from the Blue site in Study Area 1 in 1993 (A) and in 1995 (B). The B symbol shows site of origin. The location of the enlarged section is indicated by the black rectangle. 312 A. M. ADAMCZEWSKA AND S. MORRIS A B Figure 8. Migratory paths determined for radio-tracked crabs originating from the Silver site in Study Area 1 in 1993 (A) and in 1995 (B). The 3 symbol shows site of origin. The location of the enlarged section is indicated by the black rectangle. MIGRATION BEHAVIOR OF RED CRABS 313 A B C Figure 9. Locations of sightings of spray-painted crabs. (A) Yellow-painted crabs (•) and blue-painted crabs (•) during the 1993 migration. Yellow-coded crabs were seen for 3 days after painting on the Jeda Cave Road, and thus individual sightings are not shown. (B) Silver-painted crabs (•) during the 1995 migration. (C) Blue-painted crabs (•) during the 1995 migration. Open symbols show site of origin. distance covered per day was only 200 m. In Study Area 2, 5 of the 6 radio-tracked crabs also made stops and remained within a small area (5-30 m diameter) for periods ranging from 1 to 7 days before continuing towards the coastal terraces (Fig. lOa. b). Of the 5622 red crabs color-coded in eight sites within Study Area 2, 1338 sightings of color-coded crabs, on their downward or return migrations, were recorded. For every 100 color-coded crabs released, the subsequent sightings ranged from 38% for the white-coded crabs to only 67c for 314 A. M. ADAMCZEWSKA AND S. MORRIS A - Road IB Release site km km Figure 10. Migratory paths as determined by radio-tracking in Study Area 2 during the 1995 migration. (Al Three crabs from the Orange site. (B) Three crabs from the Yellow site. Open symbols show site of origin. the gold-coded crabs (Table 1 ). The sightings indicated that painted crabs traveled predominantly towards the north- western coast regardless of their site of origin (Fig. 11). In 4 of the 8 color-coded groups the greatest number of sight- ings were on Murray Road between grid 1 and grid 8 (for location, see Fig. 4). Crabs originating from the Brown site were an exception as the majority of crabs from this site headed due north. Terrace activities At least half of the crabs fitted with radio transmitters descended to the lowest shore terrace, immediately above the ocean. Males on the shore terraces were engaged in digging burrows; in places, burrow density was 3/rrr. In the 1993 season, the first matings were sighted on 25 November, and the return migration of the males began on 28 November and lasted 3 days (Fig. 6). The female red crabs spawned pre-dawn (0300-0400 hours) on 12 and 13 December, and their return migration lasted only 2 days. During the return migration very small individuals (~2 cm carapace width) that had not been observed on the down- ward migration were seen traveling inland with the return- ing females. In 1995, the first matings were sighted on 29 November, and the return migration of the males began in earnest on 3 December (Fig. 6). The main return migration of male crabs lasted 5 days, with a few late crabs still walking for 2 more days. The female red crabs spawned pre-dawn on 17 De- cember, with two small-scale spawnings occurring in a few places on the Island on 1 6 and 18 December. The returning females reached Murray Road on 19 and 20 December in very large numbers. Density of crabs during the migration The downward migration (in 1995) lasted from 5 to 26 of November. Prior to 9 November, only a few crabs were seen crossing through the counting grids (see Fig. 4 for locations) and only in certain areas (Fig. 12). The main downward migration showed a wavelike increase and decrease in the number of crabs moving across roads on different days — for example, grid Ib (Fig. 12a), grid 6 (Fig. 12e). and grid 8 (Fig. 12d). The density of crabs was greatest on Murray Road, in Study Area 2, between grids 1 and 8 (Fig. 12a-d). The maximum density of crabs recorded was 1.35/nr on grid 2 during the downward migration (Fig. 12c). but in Study Area 1. the maximum densities reached only 0.4-0.5/irr (Fig. 12i, j; see Fig. 3 for locations). The lowest densities of crabs. <0.2/m2, were recorded on Phosphate Hill (grids 5. 6, and 7; Fig. 12e, f) and on the Base Line Road (grids 9 and 10; Fig. 12g, h respectively). No crabs were ever recorded in grid 5 (Fig. 4). The number of days during which at least one crab was present in the MIGRATION BEHAVIOR OF RED CRABS 315 Figure 11. Composite showing general migration routes, directions, and destinations for the red crab migration. Derived from color-coding and radio-tracking data. Density of stipples indicates frequency of marked crab recaptures. counting grids ranged from 2 at grid 7 to 1 1 at grid 6. In contrast, in the grids located on Murray Road, the crab migration lasted from 26 to 34 days. Discussion The distribution of Gecarcoidea natalis is not homoge- neous throughout Christmas Island, and the activity levels and behavior of the crabs show extreme seasonal dichot- omy. The largest male crabs live farthest from the coast; because they have the greatest distance to travel, they must start their migration before other crabs if they are to reach the shore together. Crabs from all locations walked in sur- prisingly straight lines to definite shore destinations to which they navigated together with their neighbors. Radio- tracking revealed that given adequate time to complete the migration, the male red crabs paused for several days en route to feed instead of crowding early into the limited terrace area adjacent to the shore. The migratory behavior, strategy, and potential stress (Adamczewska and Morris. 2001 ) are influenced by the distance an individual crab must travel. However, since the breeding migration is triggered by the arrival of the monsoonal rains and is synchronized with the lunar cycle (Hicks. 1985; Grey. 1995). the timing of the arrival of the seasonal rains will be fundamental in determining the speed of migration and the time spent walking each day. The red crab population The population of red crabs on Christmas Island was estimated by subdividing the area of the entire Island into three sections: (i) the terraces, with an area of 57.7 km2 and an average density of 0.6 crabs per square meter; (ii) the plateau, with an area of 59.6 km2 and a crab density of 0. 15/nr; and (iii) mined or cleared areas, with an area of 16.7 km2 and a density of 0.01/m2 (Lake and O'Dowd, 1991). The estimated population of adult red crabs on Christmas Island was 43.7 million, substantially less than the 120-157 million crabs suggested previously (Hicks, 1985; Hicks el ai. 1990). Importantly, the density of red crabs was similar in the mature regrowth forest within Study Area 2 and the relatively undisturbed forest of the terraces in Study Area 1. The densities of red crabs recorded in this study were very similar to the values of 0.04 to 0.4/nr obtained by Lake and O'Dowd (1991 ) in transects. The population den- sity of red crabs is very high; although several studies report the densities of other terrestrial gecarcinids in the range of 0.4 to 6/m2 (Green, 1993, and references therein), these are generally the maximum densities recorded. The densities of red crabs can reach up to 40/m2 on the ocean cliffs and beaches during spawning (Hicks. 1985; pers. obs.). As a consequence of the high biomass, these crabs have a very 316 Crabs.m 1.2 0.9 -- 0.6 -- 0.3 -- A. M. ADAMCZEWSKA AND S. MORRIS 2 511 8/11 12/1115/1119/1122/1126/1129/11 3/12 6/12 10/1213/1217/12 Date (day/month) B C 5/11 8/11 12/11 15/11 19/11 22/11 26/11 29/11 3/12 6/12 10/12 13/12 17/12 Date (day/month) Grid 2 1.8 1.5 -- 1.2 -- 0.9 -- 0.6 -- 0.3 -- 5/11 8/11 12/11 15/11 19/11 22/11 26/11 29/11 3/12 6/12 10/12 13/12 17/12 Date (day /month) Figure 12. Density of crabs crossing roads during the downward migration (solid bars) and during the return migration of male crabs (bars with diagonal fill) and female crabs (bars with stippled fill) during the 1995 migratory season. Grids 1-10 were in Study Area 2; grids 1 1 and 12 were in Study Area 1. For location of grids, see Figures 3 and 4. large impact on forest dynamics (Lake and O'Dowd, 1991; related to population structure and altitude. The top pla- Green, 1993). teau is populated predominantly by large males and, The density of red crabs on Christmas Island was progressing towards the shore, the coastal terraces are D H Crabs. m" 5/11 8/11 12/1115/1119/1122/1126/1129/113/12 6/1210/1213/1217/12 Date (day /month) 0.3 0 0.3 0 0.3 Grid 6 5/11 9/11 12/1116/1119/1123/1126/1130/11 3/12 7/12 10/1214/1217/12 Date (day/month) Grid/ 5/11 8/11 12/11 15/11 19/11 22/11 26/11 29/11 3/12 6/12 10/12 13/12 17/12 Date (day /month) 0 -Ur, Grid 9 5/11 8/11 12/1115/1119/1122/1126/1129/11 3/12 6/12 10/1213/1217/12 Date (day/month) 0.3 Grid 10 0-LrrAr 5/11 8/11 12/11 15/11 19/11 22/11 26/11 29/11 3/12 6/12 10/12 14/12 17/12 Date (day/month) 0.6 0.3 Grid 11 I 5/11 9/11 12/11 16/11 19/11 23/11 26/11 30/11 3/12 7/12 11/12 16/12 Date (day/month) Grid 12 n 5/11 9/11 12/11 16/11 19/11 23/11 26/11 30/11 3/12 7/12 11/12 14/12 18/12 Date (day /month) Figure 12. Continued. 318 A. M. ADAMCZEWSKA AND S. MORRIS populated by smaller individuals of both sexes (Fig. 5; Green, 1993, 1997). Crab densities decrease sharply in just 8-10 m when moving from the forest into cleared areas (Lake and O'Dowd, 1991). Walking speeds are slow, and the crabs seem quite sedentary during the non-migratory season; one marked crab was sighted within 15 m of where it was released 8 months earlier. Red crabs appear to be unaffected in their distribution and activities by conditions or events only a few meters removed (Lake and O'Dowd, 1991; pers. obs.), and they tend to have small foraging ranges during the non-migra- tory phase of the year. Routes and destinations of migrating crabs A combination of radio-tracking, color-coding, and counting stations located in various places around the island has revealed valuable information about the mi- gratory patterns of red crabs. Three points have became apparent: firstly, crabs traveled predominantly towards the northwestern shore, which for some of the crabs was twice as far away as the southeastern shore; secondly, the crabs have specific shore destinations to which they ap- parently return each year; and thirdly, crabs originating from any one area on the plateau tend to travel in a similar direction and thereby reach the shore terraces in proximity to one another. The northwestern shore is much calmer, so the larvae probably have a better chance of landing ashore there after 3 weeks in the ocean. If the chances of recruitment on the northwestern shore are indeed substantially higher then elsewhere, then selec- tive pressure would act in favor of the crabs that return to breed to the same shore where they emerged as lar- vae, as proposed by Gibson-Hill (1947) and DeWilde (1973). The mode of navigation used by the red crabs during their migration remains speculative, but this study clearly shows that the crabs do not simply walk "down" toward the ocean as was suggested by Gibson-Hill ( 1947). Mag- netic orientation could certainly account for straight paths taken by the red crabs (Vannini and Chelazzi, 1981; Lohmann et al., 1995). Some gecarcinid crabs are thought to orient themselves towards the bright horizon (Klaassen. 1975; Bliss et al., 1978; Wolcott and Wolcott, 1982); however, although red crabs in the jungle are unable to see the horizon, they do nonetheless avoid entering darkened areas. If the young crabs can benefit from, or even need to, follow the older crabs to find their way to the shore to breed, then the idea of the "inexperienced" crabs following the "older, more experienced" crabs, proposed by DeWilde ( 1973), would certainly increase the breeding success of the young individuals. Additionally, traveling in groups would be advantageous in synchronizing the breeding activities. Crabs originating from the Green site and from the Gold site had very similar degrees of dispersal at the coast despite having walked different distances ( 1 .4 km and nearly 3 km respectively); such similarity could only be expected if they traveled as cohorts. Experiments to relocate crabs before and during the migration would provide useful information about the navigation mechanisms used by the red crabs. A comparison of the two study seasons of 1993 and 1995 in Study Area 1 reveals a 3-week difference in the timing of the start of the downward migration in relation to the lunar phase. In contrast, the spawning date appeared to be very consistent and most predictable from the lunar phase (Fig. 6). The synchronization of the spawning activities of red crabs may be important in maximizing the chances of sur- vival for the newly hatched larvae (Morgan and Christy, 1995). In 1995, the wavelike patterns in densities of crabs crossing through the density estimation grids may be par- tially attributed to the different distances that crabs may have to travel to reach any given point, but they were at least partially the result of different timing of the initial rainfall in the different parts of the Island (pers. obs.). The counting grids provide snapshot images of the numbers of crabs moving through various parts of the island and further support the conclusion, which was based on sightings of color-coded crabs, that red crabs preferentially migrate to- wards the northwestern shore. The radio-tracking data revealed an important differ- ence in the behavior of migrating crabs between the two study seasons. In 1993, the red crabs walked towards their shore destinations every day and walked relatively longer daily distances. In 1995, however, 10 out of 13 radio-tracked crabs made stops of 1 to 7 days duration at various locations before eventually arriving at the shore terraces. A consequence of this behavior is to minimize the duration of time that the entire island's red crab population is crowded on the shore terraces and the subsequent depletion of local food resources. Since the red crabs embark on their annual breeding migration immediately after the end of the dry season during which their feeding opportunities were greatly reduced (Green, 1993), early arrival of the monsoonal rains provides an opportunity for these animals to stop and feed during their downward migration. Furthermore, if the red crabs were "avoiding" reaching the terraces too early, this implies they "know" how long it will take to reach the shore and do not need to actually come in contact with the seawater to know when they have reached the coast (Hicks, 1985; Hicks et al.. 1990; Greenaway, 1994). The red crabs quite clearly avoid open areas: most crabs would travel for some distance along the side of the road under the cover of vegetation before eventually crossing the road. This observation could explain the more erratic mi- gratory routes of red crabs in the more disturbed part of the island (Fig. lOa, b). Although the red crabs do not have any MIGRATION BEHAVIOR OF RED CRABS 319 predators that might take advantage of their exposure in the open, desiccation and elevation in body temperature can reduce endurance and maximum speed of travel (Weinstein and Full, 1994; Weinstein et al.. 1994); this behavior min- imizes exposure to lower humidity and higher temperatures in the clearings. The walking speed of red crabs crossing the road in this study was very similar to the 5.4 m • min ' reported previously (Hicks. 1985), but was about 4 times faster than the overall migration speed estimated from the radio-tracking data. After G. natalis completes its annual breeding activities, most crabs leave the lowest shore terraces and return to the upper terraces and the plateau. At least some crabs appear to follow a route similar to the one they took during their downward migration and to move farther inland than their place of origin (e.g.. Fig. lOb). The environmental conditions restricting activity, cou- pled to the breeding biology of G. nalalis, require the crabs to undertake a long migration after an extended period of relative inactivity. The breeding season, which can last from 3 to 6 weeks, represents the most active period of the year for the red crabs. Such varied activity levels impose quite different physiological demands on the animals. To assess the demands of the migration, we have also examined metabolic fuel stores, metabolic sta- tus, and capacity to maintain locomotion in red crabs during their migratory activities (Adamczewska and Morris, 2001). Acknowledgments This work is dedicated to the memory of Dr. Holger Rumpff, without whose enthusiasm, energy, and friend- ship these studies would not have been possible. We would like to thank the Government Conservator, Gra- hame Beech, as well as the staff of ANCA (Parks Aus- tralia) for their assistance and hospitality. Special thanks go to Max Orchard and Roger Hart. Part of this work was carried out and reported as an ANCA consultancy. We are especially indebted to all the volunteer "crab painters and spotters" for their assistance. We also thank Pros- pectors Supplies Pty. Ltd. for the donation of compasses. King Gee Clothing Pty. Ltd for work clothes, and Zeus Chemical Products Pty. Ltd for the nontoxic marking paints. Literature Cited Adamczewska, A. M.. and S. Morris. 2001. Metabolic status and re- spiratory physiology of Gecarcoidea natalis, the Christmas Island red crab, during the annual breeding migration. Bio/. Bull. 200: 321-335. Andrews, C. \V. 1900. Monograph of Christmas Island {Indian Oi't'tint. British Museum (Natural History). London. Bliss, D. E., J. Van Montfrans, M. Van Montfrans, and J. R. Boyer. 1978. Behavior and growth of the land crab Gecarcinus lateralis (Freminville) in southern Florida. Bull. Am. Mus. Nat. Hist. 160: 111-152. Daumer, K., R. Jander, and T. H. Waterman. 1963. Orientation of the ghost-crab Ocvpode in polarized light. Z. Vgl. Physiol. 47: 56-76. Deutschlander, M. E., J. B. Phillips, and S. C. Borland. 1999. The case for light-dependent magnetic orientation in animals. J. Exp. Biol. 202: 891-908. DeWilde, P. A. W. J. 1973. On the ecology of Coenobita clypeatus in Curasao. Stud. Fauna Curasao Other Caribb. Is/. 144: 1-138. Garth, J. S. 1948. The brachyura of the "Askoy" expedition. Bull. Am. Mus. Nat. Hist. 92: 1-66. Gherardi, F., and M. Vannini. 1989. Spatial behaviour of the freshwa- ter crab, Potamon fluviatile: a radio-telemetric study. Biol. Behav. 14: 28-45. Gherardi, F., F. Tarducci, and M. Vannini. 1988. Locomotor activity in the freshwater crab Potamon fluviatile: The analysis of temporal patterns by radio-telemetry. Ethology 77: 300-316. Gibson-Hill, C. A. 1947. Field notes on the terrestrial crabs. Bull. Raffles Mus. 18: 43-52. Gifford, C. A. 1962. Some observations on the general biology of the land crab. Cardisioma guanhiimi (Latreille) in South Florida. Biol. Bull. 123: 207-223. Gray, H. S. 1995. Christmas Island Naturally. Scott Four Colour Print, Perth, Western Australia. Pp. 11-145. Green, P. T. 1993. The role of Red Land Crabs in structuring rain forest on Christmas Island, Indian Ocean. Ph.D. thesis, Monash University, Clayton, Victoria. Green, P. T. 1997. Red crabs in rain forest on Christmas Island, Indian Ocean: activity patterns, density and biomass. J. Trap. Ecol. 13: 17-38. Greenaway, P. 1994. Salt and water balance in field populations of the terrestrial crab Gecarcoidea natalis. J. Crustac. Biol. 14: 435-453. Herrnkind, W. F. 1968. Adaptive visually-directed orientation in Uca pugilator. Am. Zool. 8: 585-598. Hicks, J. W. 1985. The breeding behaviour and migrations of the ter- restrial crab Gecarcoidea natalis. (Decapoda: Brachyura). Aust. J. Zool. 33: 127-142. Hicks, J., H. Rumpff, and H. Yorkslon. 1990. Christmas Crabs. Golden Earth Design and Printing. Singapore. Klassen, F. 1975. Ecological and ethological studies on the reproductive biology of Gecarcinus lateralis (Decapoda. Brachyura). Forma Func- tioS: 101-174. Lake, P. S., and D. J. O'Dowd. 1991. Red crabs in rain forest, Christ- mas Island: biotic resistance to invasion by an exotic snail. Oikos 62: 25-29. Lohmann, K. J., N. D. Pentucheff, G. A. Newitt, G. D. Stetten, R. K. Zimmer-Faust, H. E. Jarrard, and L. C. Boles. 1995. Magnetic orientation of spiny lobsters in the ocean: Experiments with undersea coil systems. J. Exp. Biol. 198: 2041-2048. Micheli, F., F. Gherardi, and M. Vannini. 1990. Growth and repro- duction in the freshwater crab, Potamon fluviatile (Decapoda, Brachyura). Freshwater Biol. 23: 491-503. Morgan, S. G., and J. H. Christy. 1995. Adaptive significance of the timing of larval release by crabs. Am. Nat. 145: 457-479. O'Dowd, D. J., and P. S. Lake. 1989. Red crabs in rain forest. Christ- mas Island: Removal and relocation of leaf-full. J. Trap. Ecol. S: 337-348. Pinder, A. W., and A. W. Smits. 1993. The burrow microhabitat of the land crab Cardisoma guanhiimi: Respiratory/ionic conditions and phys- iological responses of crabs to hypercapnia. Physiol. Zool. 66: 216- 236. Vannini, M., and S. Cannicci. 1995. Homing behaviour and possible congitive maps in crustacean decapods. J. Exp. Mar. Biol. Ecol. 193: 67-91. 320 A. M. ADAMCZEWSKA AND S. MORRIS Vannini, M., and G. Chelazzi. 1981. Orientation of Coenobita mgosiis Weinstein R. B., R. J. Full, and A. N. Ahn. 1994. Moderate dehydra- (Crustacea: Anomura): A field study on Aldabra. Mar. Biol. 64: 135- tion decreases locomotor performance of the ghost crab Ocypode 140. qiiatlruta. Physiol. Zool. 67: 873-891. Weinstein, R. B. 1995. Locomotor behaviour of nocturnal ghost crabs on Wolcott, T. G., and D. L. Wolcott. 1982. Larval loss and spawning the beach: Focal animal sampling and instantaneous velocity from behavior in the landcrab Gecarcimis lateralis (Freminville). J. Crustac. three-dimensional motion analysis. J. Exp. Biol. 198: 989-999. Biol. 24: 477-485. Weinstein R. B., and R. J. Full. 1994. Thermal dependence of locomo- Wolcott, T. G., and D. L. Wolcott. 1985. Factors influencing the limits tor energetics and endurance capacity in the ghost crab, Ocvpode of migratory movements in terrestrial crustaceans. Conlrib. Mar. Sci. quadrata. Physiol. Zool. 67: 855-872. Suppl. 27: 257-273. Reference: Bio/. Bull. 200: 321-335. (June 2001) Metabolic Status and Respiratory Physiology of Gecarcoidea natalis, the Christmas Island Red Crab, During the Annual Breeding Migration AGNffiSZKA M. ADAMCZEWSKA1 * AND STEPHEN MORRIS2 xOcean Research Institute, University of Tokyo, 1-15-1 Minamidai, Nakano-ku, Tokyo 164-8639, Japan: and'Morlab, School of Biological Sciences, Woodland Road. University of Bristol, Bristol BS8 1UG, UK Abstract. With the arrival of the monsoonal rains and after months of inactivity during the dry season, the terres- trial crab Gecarcoidea natalis embarks on its annual breed- ing migration to the coast. The physiological demands of the migration were assessed by determining respiratory gases in the hemolymph, key metabolites, and energy stores in G. natalis during two migratory seasons. At the end of each day of migration the pulmonary hemolymph Po2 decreased by 1-2.5 kPa, but the hemocyanin remained saturated with O, and the venous reserve was largely unchanged (O: > 0.4 mmol • 1~'). The breeding migration of red crabs was accomplished without recourse to anaerobiosis, even though at times walking speeds (up to 6.2 ± 0.5 m • min"1) exceeded those that promoted anaerobiosis in non-migrating crabs and in crabs exercised in the laboratory. In contrast to all previous studies, at the end of each day of migrating, red crabs experienced an alkalosis (up to 0.1 pH units) rather than any acidosis. This alkalosis was removed overnight when the crabs were inactive. Although there were seasonal fluctuations in the glyco- gen, glucose, and triglyceride stores, crabs engaging in the migration did not draw on these stores and must have fed along the way. In contrast, crabs returning from breeding activities on the shore terraces had significantly depleted glycogen stores. Additionally, in 1993, the male crabs re- turning from the breeding activities on the terraces were dehydrated and experienced a decrease in muscle tissue water of 11%. In contrast to the breeding migration per se, fighting for burrows in which breeding occurs produced Received 23 April 1999; accepted 20 February 2001. * To whom correspondence should be addressed. E-mail: ishka@ ori.u.tokyo.ac.jp severe anaerobiosis in males, especially the victors: after 1 35 s of combat, the maximum L-lactate concentration in the hemolymph was 35 mmol • 1~'. It appears that burrowing, courtship, and mating are more demanding than the migra- tion itself. Furthermore, the data provide evidence that the metabolic responses of migrating individuals of G. natalis might be different from those at other times of the year. Introduction All terrestrial gecarcinid crabs must migrate to the coast to release their eggs into the ocean (Gibson-Hill, 1947; Gifford, 1962; Klaassen, 1975; Wolcott. 1988; Adamcze- wska and Morris, 2001). However, the annual breeding migration of Gecarcoidea natalis, the Christmas Island red crab, is unique among the gecarcinid crabs for a number of reasons. Red crabs live further inland than most other ge- carcinids (e.g., Bliss et al., 1978; Wolcott and Wolcott, 1985) and must thus travel farther to reach the coast. Fur- thermore, the red crabs undertake their migration during daytime, immediately after the arrival of the monsoonal rains and following extended periods of inactivity during the dry season. Since the migration is closely synchronized with the lunar cycle, when the rains in some years arrive late, the red crabs may have to make their downward migration in a relatively short period of time if they are to complete their burrowing and courtship activities. The red crabs may have to travel more than 4-5 km in 5-6 days during "rushed migrations" (Adamczewska and Morris, 2001). The severity of exercise is often determined by measure- ments of metabolic rate as well as acid-base status and key metabolites in the hemolymph and muscle tissue. An exten- 321 322 A. M. ADAMCZEWSKA AND S. MORRIS sive body of literature demonstrates that most crustaceans, including red crabs, tend to have low aerobic scope and rely on anaerobiosis to partially support locomotor activity (e.g., review: McMahon. 1981; Booth and McMahon, 1985; Head and Baldwin. 1986; Herreid and Full, 1988; Greenaway et al.. 1988; Forster et al., 1989; van Aardt, 1990; Henry et a!.. 1994; Adamczewska and Morris, 1994b, 1998a). During their migration, red crabs must sustain locomotion for up to 6 consecutive days to reach the shore terraces where the males construct and defend burrows and engage in mating activities (Hicks et al., 1990; Green, 1993, 1997; Adam- czewska and Morris, 2001). Laboratory studies of exhausting locomotion revealed information about the respiratory physiology of the crabs and indicated that they have no exceptional exercise ability (Adamczewska and Morris. 1994a, b). The gills and lungs of G. natalis are involved in gas exchange to different degrees, depending on the severity of exercise (Adamczew- ska and Morris. 1994a. 1998a; Farrelly and Greenaway. 1994). During exercise, G. natalis experienced the classical respiratory and metabolic perturbations found in other crus- taceans during exercise; that is, hemolymph acidosis, in- creased ventilation and heart rate, and elevated glucose and L-lactate concentrations (Wood and Randall. 198 Ib; Booth etal., 1984; Herreid and Full, 1988: van Aardt 1990; Henry et al., 1994; Thorpe et al., 1995). The aerobic scope of red crabs was surprisingly poor, implying little scope for in- creased locomotion; despite the fact that the hemocyanin remains saturated with O2 when leaving the gas exchange surfaces. G. natalis accumulated very high levels of lactate during exercise in the laboratory (Adamczewska and Mor- ris, 1994b). Intermittent locomotion can increase the total distance walked before the onset of fatigue in ghost crabs (Weinstein and Full. 1992) and thus could present an important loco- motor strategy for migrating crabs. However, red crabs in the laboratory incurred greater respiratory and metabolic perturbations during enforced intermittent locomotion than during continuous exercise (Adamczewska and Morris, 1998a). The crabs became anaerobic even though their walking speeds were within the range recorded for red crabs during the migration (Adamczewska and Morris, 2001). On Christmas Island, seasonal differences in respiratory status and metabolite concentrations in free-ranging red crabs were related primarily to a higher level of basal activity in the crabs during the wet season (Adamczewska and Morris. 2000). When required to walk for 5 min. non-migrating crabs responded to exercise in essentially the same way during both the wet and the dry seasons. The responses of the crabs to non-exhausting exercise in the field during either season were comparable to those in the laboratory studies (Adamczewska and Morris. 1998a; 2000) and did not help explain how this species completes the annual breeding migration. The conditions and circum- stances of the migration cannot be simulated in the labora- tory nor replicated outside of the migration season. Thus, understanding the physiological processes involved and the consequences for migrating red crabs could only be ad- dressed by making determinations in situ as the crabs jour- neyed to the ocean to breed. The migration is the most intense period of activity during the year. Its physical demands immediately follow prolonged periods of inactivity during the dry season, when the crabs may spend up to 2 months in their burrows (Green, 1993). This abrupt change in activity levels demands a sudden change in energy metabolism and is compounded by the fact that the red crabs have reduced foraging opportu- nities because of the unfavorable environmental conditions during the dry season. Because the migratory activities are energetically demanding, they may require the use of stored metabolic fuels. Indeed, crustaceans are known to have substantial lipid and glycogen reserves in the midgut gland and in muscle tissues, which are utilized during starvation (e.g., Parvathy, 1971; Pillay and Nair. 1973; Gibson and Barker. 1979: review: Chang and O'Connor, 1983; Nishida etal.. 1995). G. natalis appears to have no superior locomotor abilities compared to other crabs, and this raises a number of ques- tions. Primarily, is there a significant change in the respi- ratory physiology and metabolism at the onset of the mi- gration, such that a different exercise regime can be supported? This could include, for example, tolerating an exceptional diurnal anaerobiosis and O, debt. Is prolonged locomotion facilitated by a reliance on stored fuels? Do the migrating crabs tolerate physiological perturbations such as dehydration and acidosis that limit their behavior at other times? These factors could circumscribe the potential mi- gratory strategy of this species. Do the male crabs attempt to reach the terraces as fast as possible to obtain the best places for burrowing, arriving in less than peak condition? Alter- natively, do they take more time to complete the downward migration — thus arriving at the shore terraces fit for combat, with ample energy stores — and then attempt to displace already resident crabs from their burrows? The comprehensive description of the red crab migration (Adamczewska and Morris, 2001 ) provided a context for the first field investigation of the respiratory physiology and metabolism of migrating terrestrial crabs. The field investi- gations were carried out on Christmas Island during two migrations as well as in the inter-migratory period to assess annual as well as seasonal differences. Metabolites were measured in the hemolymph and muscle of migrating crabs during breeding activities and after the migration. The func- tion and functioning of the respiratory and circulatory sys- tems in gas exchange and transport were also determined. Stores of metabolic energy were quantified at various stages of the migration, and their importance in facilitating the migration and the associated breeding activities was as- MIGRATION PHYSIOLOGY OF RED CRABS 323 IU5°35'E Christmas Island Indian Ocean Pre-mlg sample site Figure 1. Map of Christmas Island showing the location of the ANCA (Parks Australia) field station and the locations of sampling sites prior to the start of the migration and during the migration. sessed. The data were interpreted together with the behav- ioral and ecological information to provide a model for migrating individuals of G. natalis. Materials and Methods Respiratory gas and acid-base status in the hemolymph Red crabs (Gecarcoidea natalis) were sampled in the field during two migration seasons (November and Decem- ber) of 1993 and 1995. To assess the functioning of the respiratory surfaces and the circulatory system in gas ex- change and transport to and from the tissues, hemolymph was sampled at various times and analyzed for respiratory gases and pH. Hemolymph was sampled from three groups of crabs in 1993 and from two groups in 1995. In 1993 the first group of crabs was sampled at the end of the dry season prior to the start of the migration, during the limited amount of time when they were found outside of their burrows (/; = 12: 6 males and 6 females). The study site for this group was a 5-min drive from the ANCA (Parks Australia) Field Station ('Pre-mig study site': Fig. 1). In 1995 the migration of the red crabs commenced earlier than expected ( Adam- czewska and Morris, 2001 ). and thus it was not possible to obtain samples from resident red crabs in the immediately pre-migratory condition. The remaining sample groups were composed of crabs migrating down to the coast. Two groups were sampled during both the 1993 and 1995 downward migration at the same site located on Murray Road near the junction of Jedda Cave Road (Fig. 1). Crabs sampled before sunrise, after "sleeping" the whole night, were designated as the morning, or "AM", group. Red crabs become immobile and unrespon- sive at night and "sleep" with their eyestalks folded down. The second group, referred to as the "PM" group, consisted of crabs sampled in the afternoon, at about 1630, after they had been walking for most of the day (Adamczewska and Morris, 2001) and had crossed Murray Road towards the shore terraces and the coast. In 1993 the AM and PM samples were taken on each of 5 days, and thus n = 5 for each group: in the slower migration of 1995. n = 9 for each of AM and PM sample groups. Hemolymph was sampled from large individuals of both sexes, but most often from male crabs (Adamczewska and Morris, 2001). Three types of hemolymph were sampled from each crab: pulmonary (300 ju,l from the post-pulmonary vessel, hemo- lymph leaving the lungs): arterial (500 /id from the pericar- dial cavity, a mixture of post-pulmonary and post-branchial hemolymph): and venous hemolymph (700 /u,l from the venous sinus at the base of the last walking leg). The samples were analyzed for O2 content ([O2]) and partial pressure (Po,), CO2 content ([CO2]), and pH at the ANCA Field Station as described previously by Adamczewska and Morris (1996). An aliquot of the remaining hemolymph was deproteinized with HC1O., (0.6 mol • 1~', ratio 1:1) and 324 A. M. ADAMCZEWSKA AND S. MORRIS neutralized with 2.5 mol • I ' K2CO, for analysis of L- lactate. The deproteinized sample and the remainder of whole hemolymph were stored frozen until further analysis of metabolites, calcium, and osmotic pressure (see below). The percentage of Hc-O^ saturation was determined as described previously by Adamczewska and Morris (1996). Metabolites, metabolic fuels, and hydration state of the hemolymph and tissues Glucose, L-lactate, and urate were measured in the hemo- lymph and leg muscle tissue to assess the extent of anaer- obiosis and accelerated glycolysis in the crabs during the migration. Additionally, glycogen and lipids (triglycerides) were measured in the midgut gland (MGG) and the leg muscle as indices of metabolic fuel reserves. Hemolymph [Ca] was measured as an index of exoskeleton decalcifica- tion, which can be important in providing HCO, base from CaCO, in response to acidosis (e.g., Henry et ai, 1981). Sampling protocol. Samples of leg muscle, MGG. and venous hemolymph were collected from free-ranging male and female red crabs prior to the start of the migration (n = 6 males and 6 females) during the 1993 migration season. Leg muscle tissue and venous hemolymph were collected each day of the downward migration in both 1993 and 1995: in the morning before sunrise (AM, /; == 4) and in the afternoon at about 1630 (PM, n -- 4). To minimize the number of sacrificed animals, samples of the MGG were taken only from the PM group. During the 1995 season, no samples could be taken from crabs prior to the start of the migration; however, to provide a comparison, samples were taken from non-migrating crabs on the 10th day of the downward migration (color- coded crabs that did not leave their home ranges; Adam- czewska and Morris, 2001 ). Muscle, MGG, and hemolymph samples were also taken from crabs that had arrived at the lowest shore terraces but had not yet begun breeding activities (terrace: n = 8). Male (/; = 8) and female (n = 8) red crabs were sampled during their respective return migrations from the shore terraces to the plateau, during both migration seasons. The site of sampling for the crabs returning from the terraces was the same as for the downward migration (Fig. 1). Samples of the MGG and leg muscle were also taken from crabs during the times of the year when the crabs were not involved in breeding activities: in June 1994 (the middle of the dry season) and in February 1995 (the wet season — 2 months after breeding activities ended). These samples were analyzed for energy stores in the form of glycogen and triglycerides (see below). Hemolymph analysis. In the laboratory, hemolymph sam- ples were analyzed for L-lactate (Boehringer Mannheim test kit #138 084). glucose (Sigma Diagnostics test-kit #510), and urate (Sigma Diagnostics test kit #685). Osmotic pres- sure was determined using a vapor-pressure osmometer (Wescor 5IOOC) calibrated with two precision standards, 290 and 1000 mOsm, containing NaCl. The concentration of Ca in the hemolymph was measured with an atomic absorption spectrophotometer (GBC 906, Melbourne). To suppress interference, samples and standards were diluted with 7.2 mmol • 1 ' LaCl3. Tissue extraction and analysis. Leg muscle tissue was obtained by encouraging a crab to autotomize the second walking leg. The muscle tissue from the leg was immedi- ately removed (—0.3 g) and deposited into a preweighed tube with 2 ml of ice-cold HC1O4 (0.6 mol • 1~ ') to depro- teinize the sample. The vials with the muscle tissue were weighed and then homogenized with an OMNI 1000 ho- mogenizer (6-mm generator) and frozen until further pro- cessing and analysis (see below). Muscle tissue was also removed from another leg, weighed, and dried in an oven at 60 °C; the dry weight was recorded and the percentage of tissue water calculated. Crabs were weighed and anesthetized by cold narcosis before removal of the MGG. The carapace was lifted and all of the MGG removed and weighed. One portion of the MGG was deproteinized in a preweighed tube with 2 ml of ice-cold HC1O4 (0.6 mol • 1~ ' ) and homogenized in the field for later analysis. A second portion of MGG was set aside for triglyceride analysis. The remainder of the MGG was weighed and dried in an oven at 60 °C; the dry weight was recorded and the percentage of tissue water calculated. All samples to be analyzed for metabolites, lipids, and ions were kept frozen and transported on dry ice. In the laboratory, muscle and MGG samples previously deproteinized with HC1O4 were homogenized with a glass homogenizer (Wheaton type) to obtain a finer homogenate. The homogenates were centrifuged in a refrigerated centri- fuge for 15 min at 5300 X g, the supernatant was removed, and the pellet was resuspended in 0.7 ml of 0.4 mol • 1 ' HC1O4 and centrifuged again. The second supernatant was removed and pooled with the first; the combined superna- tant was neutralized with 0.8 ml of K:CO3 (3.75 mol • 1~ ') and kept in an ice bath for 1 h. After another centrifugation, the final supernatant was removed and used for analysis. The concentrations of metabolites in the muscle tissue were expressed as millimoles per kilogram of wet tissue mass. The deproteinized and neutralized leg muscle samples were analyzed for L-lactate (Boehringer Mannheim test kit #138 084). Additionally, the samples from both leg muscle and MGG samples were analyzed for glucose and glycogen by using a modified method described in Bergmeyer ( 1985). Briefly, amyloglucosidase (Boehringer Mannheim Cat. #208 469) made up to > 10 kU • I"1 in an acetic acid buffer at pH 4.8 was used to hydrolyze glycogen into glucose during a 2-h incubation, with shaking, at 40 °C. At the end of the incubation the enzyme was denatured by addition of 0.6 mol • 1 ' HC1O4. The solution was centrifuged at MIGRATION PHYSIOLOGY OF RED CRABS 325 10,000 X g for 10 min and the supernatant used for glucose analysis. For each sample, free glucose was also measured and subtracted from the glucose content obtained from the glycogen digestion. Frozen MGG samples were analyzed for triglycerides by first homogenizing a sample with an equal mass of gum arabic (Sigma Catalog #G-9752) and 15 parts of bovine serum albumin (6% w/v Boehringer Mannheim Catalog #100-030) in a fine glass homogenizer (Wheaton type) at 55 °C. This homogenate was then diluted 1:10 with 6% w/v bovine serum albumin and used for triglyceride analysis (Sigma test kit #334-A). Triglyceride levels in the muscle tissue of red crabs were not detectable by this method. Terrace activities During the 1995 migration, concentrations of lactate, glucose, and urate were measured in the hemolymph as indicators of anaerobiosis and the intensity of male combat. Only male-male interactions involving physical contact were considered. Male crabs on the shore terraces were observed and if a fight was noticed, its duration was re- corded. Fights lasting less than 10 s were not considered. The fight was considered to have ended when the victor remained in possession of the burrow and the loser retreated or was chased away by the other opponent. At this point both crabs were promptly captured (within 5 s) and hemo- lymph was withdrawn from the venous sinus by puncturing the arthrodial membrane. The design of the experiment dictated that fights could be used only if they occurred near the observer and only if they were observed from start to finish. A total of 14 fighting pairs were sampled over 4 days of breeding activities on the terraces. Hemolymph samples were also taken from six male crabs resting at the entrance of their burrows; these were used to determine the basal metabolite status of male crabs on the terraces during the breeding activities. The hemolymph samples were treated and stored as described above and analyzed in Sydney for glucose, L-lactate, and urate. Statistical analysis The data were tested for homogeneity of variances using Bartlett's ^ test, and any heterogenous data were log- or square-root transformed. Respiratory gas data were ana- lyzed using two-factor ANOVA for the three hemolymph types (pulmonary, arterial, and venous) and the different sampling groups in each season. Seasonal differences be- tween the AM and PM groups were also examined by ANOVA. Statistical analyses on tissue and hemolymph metabolites and on water and ion measurements between the AM and PM samples for the days of downward migration were carried out by two-factor ANOVA. If no difference between the AM and PM groups was detected, then the data were pooled for each day for further analysis by ANOVA. Analyses between tissue types and between the two sampling seasons were also carried out where appropriate. Post hoc testing was done using the Tukey HSD multiple comparison test. All analyses were performed using the Systat for Windows 5.0 statistical package. The samples from fighting males were analyzed by analysis of co-variance (ANCOVA). Signifi- cance level was taken as P = 0.05 in all cases. All data are presented as mean ± SEM. Results There were no differences in hemolymph respiratory gases, acid-base balance, metabolite concentrations, or en- ergy stores of male compared with female crabs at the end of the dry season in 1993. Therefore, these data were pooled and from here on are referred to as the "resident pre-mig" group. Oxygen in the hemolymph The Ppo2 [efferent pulmonary oxygen partial pressure (Po2)] was consistently higher than the Pao2 (arterial Po2) in both 1993 and 1995 (Fig. 2). The Pvo2 (venous Po2) was always lower than the Pao2 and the P po2. During the 1995 migration, the hemolymph Po2 of the PM group was statis- tically lower than in the AM group (Fig. 2). In 1993, although the Po2 in the pulmonary hemolymph (8.26 ± 2.20 kPa) of the AM group was 3.5 kPa higher than in the PM group (Fig. 2), there was no overall statistical difference between the crabs in these two groups. The O2 content ([O-,]) of the hemolymph was similar in both migration seasons (Fig. 3). The O2 content of pulmo- nary hemolymph ([O2]p = 1.13 and 0.96 mmol • F1 in 1993 and 1995 respectively) was very similar to the arterial values, but [O2] in venous hemolymph decreased signifi- cantly, to about 0.50 mmol • 1~'. in both seasons. The AM group sampled during the 1993 migration season was an exception because the arterial [O2] was significantly lower than the pulmonary hemolymph concentration. Addition- ally, in 1993, the [O2]v in the AM group was significantly greater than in the pre-mig group (Fig. 3). The saturation of hemocyanin (He) with O2 varied be- tween samples in a manner similar to that of the [O2], The O-, saturation of the pulmonary hemolymph was high (mean = 102 ± 6%) and similar to the He saturation of arterial hemolymph except in the AM group of 1993, in which the arterial hemolymph was only 74% saturated. In comparison, the mean Hc-O2 saturation of venous hemo- lymph decreased significantly in both seasons, to 43% in 1993 and 63% in 1995 (P < 0.001 ). Carbon dioxide content and pH of the hemolymph The pH of pulmonary, arterial, and venous hemolymph was similar within each group of red crabs sampled (Fig. 4); 326 A. M. ADAMCZEWSKA AND S. MORRIS , 1993 D - Pre-mig - am - pm 1995 10 - 0 -am • -pm* Pulmonary Arterial Hemolymph pa Venous Figure 2. Oxygen partial pressure (Po2, kPa) in the pulmonary, arte- rial, and venous hemolymph of Gecarcoidea natalis during two migration seasons, 1993 (top panel) and 1995 (bottom panel). Morning and afternoon groups were sampled during both migration seasons (n = 5 in 1993: n = 9 in 1 995 ). Top panel ( 1993 season I includes measurements of Po2 in crabs sampled 1 week prior to the start of the migration (pre-mig n = 121. The letters "p" and "a" indicate that the mean is significantly different from that of pulmonary or arterial hemolymph respectively, within a sample group. The * symbol next to the legend indicates a significant difference between the AM and the PM groups. thus, for further analysis between groups, the pH for the three hemolymph types was pooled for each sample group. The resident pre-mig red crabs in 1993 had a mean hemo- lymph pH of 7.61 ± 0.03, which decreased significantly after the onset of the migration to near 7.4 (Fig. 4). Most interestingly, the hemolymph was more acidic in crabs sampled in the morning (AM pH = 7.35 ± 0.07) than in crabs sampled in the afternoon (PM pH = 7.46 ± 0.06, P < 0.001 ). A similar alkalosis at the end of the day of migration was seen in 1995 (AM-PM difference of 0.042 pH units) but was less significant than during 1993 (P = 0.068; Fig. 4). There was no difference in hemolymph CO2 content ([CO2]) between pulmonary, arterial, and venous samples within any one group of red crabs sampled (Fig. 5). During 1993, the [CO2] in the hemolymph of migrating crabs was significantly greater than in the resident pre-mig crabs ( mean [CO2] = 9.8 ± 0.8 mmol -1 ' ). In addition, the [CO2] in the hemolymph of the PM group was significantly greater than that of the AM group; the concentration was 3.3 mmol • r1 higher in 1993 and 1.7 mmol • T1 higher in 1995 (Fig. 5). Hydration state of red crabs during migratory activities There were no differences in the hydration state of crabs at the start and the end of a day walking; the AM and PM samples were thus pooled for each day. In the 1993 season the percentage of water in crab leg muscle was very stable during the entire downward migration (range 77.2%- 78.3%) and, importantly, did not differ from that of the pre-migratory crabs (76.1% ± 0.4%). However, the male crabs returning from the shore terraces were dehydrated: the percentage of water in their muscle tissue was 11% lower 1993 D - Pre-mig - am - pm pa 1995 Pulmonary Arterial Hemolymph Venous Figure 3. Oxygen content ([O2], mmol -1 ') in the pulmonary, arte- rial, and venous hemolymph of Gecarcoidea natalis during two migration seasons, 1993 (top panel) and 1995 (bottom panel). Morning and afternoon groups were sampled during both migration seasons (n = 5 in 1993; n = 9 in 1995). Top panel (1993 season) also includes measurements of [O,] in crabs sampled 1 week prior to the start of the migration (pre-mig n = 12). The letters "p" and "a" indicate that the mean is significantly different from that of pulmonary or arterial hemolymph within a sample group. The pair of A symbols denote a significant difference in venous O2 between the resident pre-mig and AM crabs. MIGRATION PHYSIOLOGY OF RED CRABS 327 7.8 -i 1993 1995 Pulmonary Arterial Hemolymph Venous Figure 4. The hemolymph pH of pulmonary, arterial, and venous samples taken from Gecarcoidea natalis during two migration seasons. 1993 (top panel) and 1995 (bottom panel). Morning and afternoon groups were sampled during both migration seasons (n = 5 in 1993; n = 9 in 1995). Top panel ( 1993 season) also includes measurements of hemolymph pH in crabs sampled 1 week prior to the start of the migration (pre-mig n = 12). The * symbol next to the legend indicates a significant difference between the AM and the PM groups. The "R" denotes a significant difference from the crabs at rest — that is, from the resident pre-mig crabs. than in all the other groups sampled in that season. In contrast, during the 1995 migration season, both the return- ing male and female crabs had a significantly higher amount of water in muscle tissue (79.1%) than the crabs sampled during the downward migration (days 5-7; range 75.2%- 76.9%). The percentage of water in the MGG was more variable and significantly lower than in the muscle tissues, with a mean of 57.0% ± 1.5% during the 1993 migration and 61.7% ± 2.4% in 1995. Crabs that did not participate in the 1995 migration were unusual in that water constituted only 45.4% ± 1.8% of their midgut gland. The osmotic pressure of the hemolymph was more vari- able than the percentage of water in the muscle tissue. The osmotic pressure in the hemolymph of the AM and PM groups on any one day was the same, and thus the data for each day were pooled for further analysis. The hemolymph osmotic pressure was largely constant during the downward migra- tion of both the 1993 (771 ± 3 mOsm) and 1995 (803 ± 4) seasons; but in 1993 the osmotic pressure of crabs during the downward migration was significantly lower than that of the pre-mig crabs (815 ±7 mOsm) and the returning fe- males (869 ± 14 mOsm). Additionally, in both 1993 and 1995 the returning males had lower osmotic pressure (mean 699 ± 16 mOsm) than did the crabs sampled during their downward migration. Calcium and urate concentration in the hemolymph The calcium concentration in hemolymph of resident pre-mig crabs (13.74 ± 0.42 mmol • F1) was significantly higher than in all the other groups sampled in 1993 (mean [Ca] = 9.29 ± 0.87 mmol -F1). During the 1995 migration 1993 D - Pre-mig 0 -am • - pmg: _ ., 1995 Pulmonary Arterial Hemolymph Venous Figures. Carbon dioxide content ([CO,], mmo! • 1 ') in the pulmo- nary, arterial, and venous haemolymph of Gecarcoidea natalis during two migration seasons, 1993 (top panel) and 1995 (bottom panel). Morning and afternoon groups were sampled during both migration seasons (n = 5 in 1993; „ = 9 in 1995). Top panel ( 1993 season) also includes measurements of [CO,] in crabs 1 week prior to the start of the migration (pre-mig n = 12). The * symbol next to the legend indicates a significant difference between the AM and the PM groups. The "R" denotes a significant difference of a sample group from the crabs at rest — that is, from the resident pre-mig crabs. 328 A. M. ADAMCZEWSKA AND S. MORRIS o r"i ri DC DC DC u a n ^c o -r -r O O rj O O E | O O O O O £IJ +1 +1 +1 +1 +1 *j rj- in m CM o ™— fN r«"l IT, fN CN o o o o o «~. -r r- n CN rf o o o — o o S E 0 0 C 0 0' 0 1 X +1 +1 +1 +1 +1 +1 £ TT rf m oo m TJ- o o o o o" o £ S; O IT1, — . — — S u CJ r) DC DC DC ? 3 O ^C l/"i DC DC VO — -t DC DC C £ d — o o" o 1? Ct> +1 +1 +1 +1 +1 'C _J r*", •— \o f^i ^ _^ 00 — 00 -rf CJ — g C — S ^ C-3 J^ -± *T OC DC 3C U •o 'S O. 1 — • DC \C O> in - - 0 0; 0 I *o | o o o o o cd +1 +1 -H +1 +1 u o X -r -f -t m o £ o1 j. o 00 C .3 — 2 CH- *o E -^ uJ o r-i u-i m — . ti^'i- "^ — — rN DC OC DO — £ = 3 ri -r O u~. sC O CN — DO m in «•*". 1 ? — o d o o — • — 73 •S of + 1 +1 +l +1 +1 +i S 1 1 m rf\ ^ rj 5C, p-( n-, OJ S -C C — -3 "5. 1 1 = . , Q+ *Q 1 1 "c. ri m m -— •. - — - - — > — — ri oc oc DC li ^ •- U -5 ? £ — rt — DC -rf r*-, 73 rS ^ Wj i. "• "o — •> — — £ • p u ,SJ ,U GO a> ^Z ^- w ~ * 111 O Q Q£ ™ -J fa a. < a. H o£ z£ = = season, the mean hemolymph [Ca] varied by up to 4.5 mmol • 1~' between the sample groups — that is. the AM and PM groups, males engaged in combat and breeding, and male crabs returning from the terraces — but these differences were not statistically significant. Hemolymph urate was similar in both seasons, with an average concentration of 0.037 ± 0.005 mmol • 1~ ' in 1993 and 0.041 ± 0.006 mmol • 1~' in 1995. There were no statistical differences between any of the groups of red crabs sampled. L-lactate and glucose in the hemolymph and muscle tissue The concentration of L-lactate in the hemolymph gener- ally remained low. less than 2 mmol • 1~' in 1993 and less than 1 mmol • 1~' in the 1995 season (Table 1), except in fighting males (see below). The L-lactate concentration in the muscle tissue of mi- grating red crabs in 1993 was significantly lower than the 6.3 ± 1.2 mmol • kg" ' in the muscle of the pre-mig crabs (Table 1). Furthermore, in that year migrating crabs had a mean muscle L-lactate concentration that was low ( 1 .0 ± 0.1 mmol • kg~') in the morning but increased slightly (to 3.3 ± 0.8 mmol -kg"1) in the afternoon (P < 0.001). No such diel differences occurred in the 1995 migration season: the AM and PM crabs both contained lactate at close to 3 mmol • kg~' (Table 1). The glucose concentration in the hemolymph was below 1 mmol • r ' in both seasons (Table 1 ). Red crabs descend- ing to the terraces in 1993 had higher hemolymph glucose levels (0.85 ±0.12 mmol • 1~') than both the pre-migratory crabs (0.44 ± 0.04 mmol • T1) and the crabs sampled during their return migration (0.51 ± 0.06 mmol •!"',/'< 0.001; Table 1). During the 1995 migration (mean [glucose] = 0.80 ± 0.09 mmol -I"1) the variances of the data were severely heterogenous, and no significant differences could be detected between any of the sample groups (Table 1 ). In the muscle tissue, the concentration of glucose was also less than 1 mmol • kg"1 during both seasons (Table 1). However, during 1993, the concentration in the leg muscle of the crabs engaged in the downward migration (0.29 ± 0.06 mmol • kg"1) was lower (P < 0.001) than that in the resident pre-mig crabs (0.83 ± 0.09 mmol • kg"1). The variances of the data collected during the 1995 migration were heterogenous, and it was not possible to detect any differences in glucose levels (Table 1 ). Energy stores in the midgiit gland and muscle tissues Glycogen stores were measured as glucose units and expressed as concentration per wet weight of tissue. During the downward migration of 1995, leg muscle glycogen concentrations (69 mmol glucose • kg"1) were about twice those of the crabs in 1 993. The concentration of glycogen in the muscle of pre-migratory crabs in 1993 (31.6 ± 2.4 MIGRATION PHYSIOLOGY OF RED CRABS 329 Table 2 Glucose and glycogen in the midgut gland of Gecarcoidea natalis (values given as mean ± SEM; sample size provided in parenthesis) 1993 migration! 1995 migration Non-migrating Day in relation to migration Glucoset* Glycogent Glucose* Glycogen Glucose Glycogen (mmol • kg~'| (mmol • kg~') (mmol • kg"') (mmol • kg"3) (mmol • kg"1) (mmol • kg"') Non-mig Jun. 1994 9.65 ± 1.48 (8)# 0.95 ± 0.21 (8)t Non-mig Feb. 1995 — — — — 3.83 ± 1.01 (8)# 12.14 ± 3.4 (8)t Pre-mig 17.81 ± 2.18(12) 1.58 ±0.35(12) — — — — 5th day of mig — — 32.13 ± 7.11 (4) 18.36 ± 3.51 (4) — — Non-mig (10th) — — 19.03 ± 1.82 (8)D 4.31 ± 0.34(8) — — 1 1th day of mig — — 29.18 ± 7.95(4) 19.02 ± 5.78 (4) — — 1 2th day of mig — — 25.68 ± 1.64(4) 14.81 ± 2.55(4) — — Return male 8.92 ± 1.14(8) 3.55 ± 0.99 (8) 13.44 ± 2.53(8) 5.49 ± 0.96 (8)D — — Return female 5.74 ± 0.88 (8) 0.55 ±0.21 (8) 6.14 ±0.84(8) 8.22 ± 1.50(8)D — — D Different from crabs engaged in the downward migration. t Difference between the two migration seasons. * Difference between glucose and glycogen within a season. # Two like symbols indicate a difference between the two groups. t Two like symbols indicate a difference between the two groups. mmol glucose • kg"1) was significantly lower than that of red crabs during the downward migration (41.6 ± 5.7 mmol glucose -kg"'). In both years, glycogen concentration in the muscle of both males and females had decreased signifi- cantly (to 14.2 ± 2.5 mmol glucose • kg"1 in 1993 and 24.9 ± 5.5 mmol glucose • kg"1 in 1995) by the time the crabs were sampled during their return migration. The mean muscle glycogen concentration in crabs sam- pled during the dry season of June 1994 (17.6 ± mmol glucose • kg~ ') was just one-third of that measured in crabs during the wet season of February 1995. The concentration of glycogen in the MGG was much lower than in the leg muscle, whereas the concentration of free glucose in the MGG was greater at all times (Table 2). The "total glucose" stores ([glucose] + [glycogen]) in the MGG during the 1995 migration season were consistently greater than in 1993 (Table 2). The concentration of total glucose in the MGG of return- ing males (12.5 mmol • kg"1) and returning females (5.7 ± 0.9 mmol • kg"1) in 1993 was lower than that of crabs n = 52; 1995 mean = 234.8 ± 36.7 mmol glycerol • kg ', n = 44). There were seasonal differences in the triglyceride concentration of the MGG. The lowest levels were in sam- ples collected in February 1995 (mean = 71.4 ± 6.6 mmol glycerol • kg"1, n = 8), but triglycerides were considerably higher during the dry season in June 1994 (mean = 162.09 ± 22.6 mmol glycerol • kg"1, n = 8). Nonetheless, the MGG triglycerides were lower during both of these seasons than during the actual migration. Metabolite status in the hemolymph of male crabs after combat Fights on the terraces occurred only between male crabs and were always for the possession of a burrow. Most of the fights on the terraces were very brief, less than 10 s, and thus were not used in this study. The duration of recorded fights varied between 18 and 135 s, with the majority lasting between 40 and 60 s. The mean concentration of urate in the hemolymph was before the migration ( 19.4 mmol -kg ; Table 2). Similarly, 0.043 mmol • 1 ' and it did not increase in fighting crabs kg ') and in 1995 the returning males (19 ± 3 mmol females ( 14 ± 2 mmol -kg"1) also had greatly reduced total glucose stores in the MGG compared with the crabs sam- pled during the downward migration (Table 2). There were otherwise no obvious seasonal trends in the total glucose stores in the MGG (Table 2). The triglyceride content in the MGG of red crabs, mea- sured as glycerol. did not change during the course of the migratory activities, and the mean concentration of triglyc- erides in the MGG was similar in both of the migration seasons (1993 mean = 247.3 ± 38.7 mmol glycerol • kg"1, (Fig. 6A). Similarly, there was no difference in the hemo- lymph glucose concentration in crabs sampled at burrow entrances and immediately after a fight (Fig. 6B). The mean concentration of L-lactate in the hemolymph of male crabs sampled at burrow entrances was high (7.4 ± 3.9 mmol -I"1; Fig. 6C). Importantly, there was a significant increase in L-lactate in the hemolymph (P < 0.01) with increase in fight time, but only for the victorious crabs (Fig. 6C; ANCOVA and regression analysis). This increase can be described by the following equation [Lac] = time (s) X 0.25 - 3.20 (r = 0.71). The maximum [L-lactate] in the 330 A. M. ADAMCZEWSKA AND S. MORRIS o.iu - o * - victor O - loser •j 0.08 - * • - at burrow o £ 0.06 - • victors £ » Q 0.04 - * • ? »o$ o * losers cd •=- 0.02 - 0 t i i i i i B _'- 1.5 - I 1.0- 40 -i 40 60 80 100 fight duration (s) 120 140 Figure 6. The concentration of (A) uric acid, (B) glucose, and (C) L-lactate (mmol • 1~') in the hemolymph of male Gecarcoidea natalis guarding burrows during the breeding season (•) and immediately after combat, for losers (O) and victors (*). Samples were taken from crabs engaged in physical fighting for possession of the burrows, and hemolymph metabolite concentration was plotted against the duration of the fight. The [L-lactate] was statistically dependent on fight duration in the victors but not the losers. hemolymph of 35 mmol • 1 ' was recorded after 135 s of combat for the crab that won. Discussion Christmas Island is composed largely of highly porous coral limestone with very little surface water outside of the wet season. Like other terrestrial crustaceans, the Christmas Island red crabs are prone to dessication and retreat into their burrows when humidity falls below 85% (Bliss el al., 1978; Wolcott. 1992; Green. 1993, 1997). Dehydration has been shown to decrease locomotor capacity of some crus- taceans (Weinstein et al., 1994) and to interfere with respi- ratory gas transport in the hemolymph (Burggren and Mc- Mahon, 1981 ). The arrival of the monsoonal rains increases humidity, thereby creating more suitable conditions for ac- tivity above ground, and provides a cue for the start of the annual breeding migration of Gecarcoidea natalis (Hicks, 1985; Green, 1997; Adamczewska and Morris, 2001). Dur- ing some years, when the monsoonal rains are late and an isolated shower triggers the migration, hemolymph osmo- larity of 1 100 mOsm has been recorded and the red crabs experience high mortality rates (Greenaway, 1994). The maintenance of hydration may be very important in deter- mining the ability of the red crabs to undertake the breeding migration. During the two migrations studied, the red crabs re- mained well hydrated. Although hemolymph osmolarity varied by up to 150 mOsm, the hydration state of leg muscle tissue generally varied by only 1%. The only sign of dehy- dration (11% decrease in muscle tissue water) was in the males during their return migration of 1993. These crabs had engaged in the strenuous activities of digging and defending their burrows during a 4-day period with no rain. The return migration of male crabs coincided with heavy rainfall, and although the male crabs were clearly drinking rain water, as seen by the hemodilution, they did not appear to rehydrate their muscle tissue before leaving the coastal terraces. A preferential defense of the extracellular over the intracellular fluid volume, previously reported for two ge- carcinid species (Harris and Kormanik, 1981), may be a strategy limited to true land crabs (Taylor and Greenaway. 1994). Field studies using markers of extracellular fluid volume will be necessary to determine exact changes in hemolymph volume in migrating crabs (Harris and An- drews, 1982; Taylor and Greenaway. 1994). Respiratory gas exchange in migrating crabs The different durations of the migration in the two study years (5 days in 1993 vs. 21 days in 1995; Adamczewska and Morris, 2001) imposed very different physiological demands on the red crabs, but the respiratory changes the crabs experienced were remarkably similar. Previous infer- ences of the relative contribution of the gills and lungs in O2 uptake by G. natalis have suggested both that the lungs are more important (Adamczewska and Morris, 1994a) and conversely that the gills and lungs play a comparable role (Adamczewska and Morris, 1998a; Farrelly and Green- away, 1994). In both years studied here, the PPo2 of red crabs was consistently higher than the fao2 but the pulmo- nary-arterial Po2 difference decreased at the end of each day of walking. This decrease was brought about by the lower PpO, (by 1- 2.5 kPa) after each day walking and indicates some diffusion limitation across the lungs. Crabs forced to exercise in the field outside of the migration season showed no such loss of pulmonary oxygenation (Adamczewska and Morris, 2000). Thus, although the lungs routinely play an important role in O-, uptake, the gills are nonetheless re- cruited to a greater extent during the migration. It is clear MIGRATION PHYSIOLOGY OF RED CRABS 331 2.67 Figure 7. The pH/Bicarbonate relationship for Gecarcoidea natalis during two migration seasons. (•) 1993 and (O) 1995. The pre-migratory values in 1993 were assumed to be applicable to the 1995 season. The />co, isopleths were constructed using the Henderson-Hasselbalch equa- tion using aCO2 = 0.03962 and pK = 6. 1 . The non-bicarbonate buffer line was adopted from Adamczewska and Morris (1994a). that hemolymph shunting occurs between the gills and lungs during locomotion in red crabs. Investigation of hemolymph flow is now essential to determine the role of these two gas exchange structures in O2 uptake. During the migration, the Po2 of hemolymph leaving the gas exchange surfaces varied between the groups of crabs sampled (mean 3-7.7 kPa) but was within the range of values recorded previously for this species (3.7-10.7 kPa: Farrelly and Greenaway, 1994; Adamczewska and Morris, 1994a). Despite the differences in Po2, the high affinity of the He for O2 in red crabs (P50 = 1.77 kPa: Adamczewska and Morris, 1998b) ensured He saturation with O2 at the gas exchange surfaces. Excretion of CO2 was also maintained by walking crabs during the migration, but without any internal hypercapnia. Red crabs exercised in the field either before or after the migration season exhibited significantly decreased hemo- lymph CO2 content due to a metabolic acidosis imposed on a respiratory acidosis (Adamczewska and Morris, 2000). Conversely, each day of migration resulted in an increase in hemolymph CO2 of 2-3 mmol • 1~ ', which was of metabolic rather than respiratory origin. Since the Pco, remained constant and the diffusion gradient across the respiratory surfaces was unchanged, the increase in CO2 did not indi- cate impairment of CO-, excretion. Acid-base perturbations More detailed analysis using the Henderson-Hasselbalch plot showed that the AM crabs experienced a relative respi- ratory acidosis compared with the pre-mig crabs (Fig. 7). This acidosis (—0.2 pH units) persists during the wet season even after the migration is completed and appears to be a simple consequence of the higher activity levels made pos- sible by high relative humidity. (Adamczewska and Morris, 2000). Active red crabs must increase O^ uptake, and con- sequently CO2 excretion is facilitated by an internal hyper- capnic acidosis that improves the outward Pco^ gradient (e.g., O'Mahoney and Full, 1984). Thus, the daily alkalosis observed in migrating red crabs in both 1993 and 1995 was superimposed on a seasonal acidosis brought about by in- creased basal activity. Importantly, there was no evidence of any additional respiratory acidosis as a result of walking each day during the migration (Fig. 7). The obvious metabolic alkalosis experienced by the red crabs at the end of each day of migration contradicts all previous studies of acid-base perturbations during locomo- tion in crustaceans (e.g., McMahon el al., 1979; Smatresk et ai, 1979; Wood and Randall, 1981b; review: McMahon, 1 98 1; Booth etal., 1984; Greenaway et al.. 1988), including exercising red crabs (Adamczewska and Morris, 1998a, 2000). The alkalosis, which must originate either from the removal of H+ from the hemolymph or the addition of metabolic base, was removed overnight, only to reappear at the end of the next day of migration. A diel de-calcification of the exoskeleton to supply metabolic base in the form of HCO,~ seems unlikely because since there was no evidence of any change in hemolymph [Ca] between AM and PM crabs; such a change would normally accompany this process (e.g., Henry et al., 1981). A small increase in lactate concentra- tion in the muscle of PM crabs in 1993 was not reflected in any increase the hemolymph. Similarly, a large increase in muscle lactate (values routinely near 6 mmol • kg"1) seen in relatively inactive free-ranging crabs in the dry season of June 1994 was not reflected in the hemolymph (values only 0.72 mmol •!"'). Furthermore, the muscle L-lactate concen- trations of the relatively inactive dry-season crabs were nearly 4 mmol • kg~' greater than those of more active wet-season crabs (Adamczewska and Morris, 2000). Clearly there are some aspects of the dynamics of lactate and possibly H+ distribution that are different in G. natalis and require further investigation. The daily alkalosis experi- enced by red crabs during their downward migration would partially offset the wet-season acidosis and assist O2 loading by increasing the oxygen affinity of the hemocyanin (Ad- amczewska and Morris, 1998b). Metabolism during breeding activities Crabs generally do not possess high aerobic capacity and tend to become anaerobic during enforced exercise (Smatresk and Cameron, 1981; Booth et al., 1982; Herreid and Full. 1988; Forster et al., 1989; Henry et al., 1994). Hemolymph and tissue L-lactate levels may increase in excess of 30 mmol • 1~' in exercising crabs (e.g.. Green- away et al., 1988: Milligan et al., 1989; Adamczewska and 332 A. M. ADAMCZEWSKA AND S. MORRIS Morris, 1994b). Since anaerobiosis accumulates acid and an O2 debt that must be repaid later (Wood and Randall, 198 la; Ellington, 1983; Herreid, 1980), prolonged locomotion can- not depend on the anaerobic provision of energy. In migrat- ing red crabs the L-lactate concentrations in the hemolymph and muscle tissue were generally low and comparable to those measured in free-ranging crabs during the non-migra- tory seasons (Adamczewska and Morris, 2000). Further- more, L-lactate levels were similar during both of the mi- gratory seasons, despite the fact that red crabs migrating in 1993 walked twice as far each day as those migrating in 1995 (Adamczewska and Morris, 2001). If the small diel change in muscle concentration of lactate during the 1993 migration represented anaerobiosis, this would be equiva- lent to only approximately 0.4 mmol ATP, or to 3-4 min of resting O2 uptake (Adamczewska and Morris, 1994b). Laboratory studies indicated that to cover a set distance in a fixed amount of time, red crabs are better suited to continuous travel than to periods of rapid walking inter- rupted by periodic rests (Adamczewska and Morris, 1998a). The mean daily walking speeds of red crabs (1.1 m • min~ ' , estimated from radio-tracking during the 1993 migration) are within the range that this species is capable of support- ing aerobically (Adamczewska and Morris, 1994b; 2001). However, walking speeds measured for red crabs crossing roads (e.g., 6.2 m • min"1 for returning females: Adamcze- wska and Morris, 2001) show that sometimes during the migration the crabs walk more than 4 times faster than can be sustained aerobically under laboratory conditions. There was an approximate doubling of muscle lactate in the re- turning female crabs in 1993, to 6.34 mmol • kg"1, but this was again confined to the muscle, and lactate concentrations remained low in the hemolymph (Table 1 ). This change in muscle lactate represents only about 5 min of resting oxy- gen uptake. The possession of a burrow has direct consequences for the breeding success of a male red crab. For male crabs, defending a burrow was more strenuous than prolonged locomotion during the downward migration. The basal he- molymph [L-lactate] of crabs guarding their burrows was elevated by more than 6 mmol • 1"' compared with migrat- ing crabs. Winning fights is energetically demanding, and the levels of L-lactate (up to 35 mmol • P1) measured after combat were comparable to the concentrations recorded after 45 min of intense exercise (Adamczewska and Morris. 1994b) or after many hours of hypoxia (Gade. 1984; Albert and Ellington, 1985; van Aardt, 1988). G. natalis expends much more energy in defending burrows than the aquatic crab Necora puber expends in a fight under laboratory conditions (Thorpe etal., 1995). The high levels of L-lactate in red crabs either indicate an extreme effort put into win- ning a fight or represent an accumulated sum of attempting to win many fights. Since the losers of conflicts did not show the same accumulation of L-lactate as the winners, the decision to give up the fight is apparently determined early on, so that the eventual loser minimizes his investment in the conflict. The metabolism of red crabs appears fundamentally al- tered during the migration. The crabs exhibit a unique metabolic alkalosis of the hemolymph that occurs only during the migration: crabs exercised under field conditions (Adamczewska and Morris, 2000) and in the laboratory (Adamczewska and Morris, 1994b) show the classical mixed respiratory/metabolic acidosis. The walking speed of red crabs crossing roads and clearings was estimated to be 380% of the speed that can be sustained aerobically (Ad- amczewska and Morris. 1994b, 2001), yet no significant anaerobic metabolism was detected during the downward migration. One possible conclusion is that the aerobic scope of red crabs, which is otherwise unusually low (Adamcze- wska and Morris, 1994b), is specifically increased for the purpose of the migration. If so, the signals and processes by which this is achieved are yet to be elucidated. Storage and usage of metabolic fuels The midgut gland and muscle tissue compose a large portion of the body mass in crustaceans and are considered to be major organs for the storage of metabolic fuel (Vonk, 1960; Chang and O'Connor, 1983; England and Baldwin, 1983;Tsaiefa/., 1984; Schirf etal., 1987). In G. natalis, the muscle tissue stores glycogen, whereas the MGG is a major site of lipid storage (Adamczewska and Morris, 1994b; Morris, unpub. data). In animals, fatty acids are primarily stored in the form of triglycerides, and in G. natalis tri- glycerides were estimated to constitute about 70% of total lipids in the MGG (380 mg lipid • g"1 dry tissue mass: S. Morris, unpub. data). The lipid stores in brachyuran crusta- ceans vary considerably between species, but total lipid stores in the MGG can reach 900 mg • g ' dry mass (Giese, 1966; Heath and Barnes, 1970; Parvathy, 1971; Kucharski and DaSilva, 1991). The glycogen stores in the muscle tissue of G. natalis sampled in the field (1 1-64 mg glucose • g~' dry mass) were only a fraction of what this species is capable of storing (250 mg glucose • g"1 dry mass, Adamczewska and Morris, 1994b) but were within the range found in other brachyuran species (8-71 mg • g"1 dry tissue mass: Giese, 1966; Pillay and Nair, 1973; Hilmy et al., 1986; Kucharski and Da Silva, 1991 ). Since the stores of metabolic fuels in red crabs in the field were not exceptional, red crabs either lack special requirements for large energy stores under natural condi- tions or have no opportunity to accumulate such stores (Greenaway and Linton, 1995; Greenaway and Raghaven, 1998). Leaf litter is the major component of the diet of red crabs. During the dry season, when the red crabs are largely confined to burrows, a thick layer of leaf litter accumulates MIGRATION PHYSIOLOGY OF RED CRABS 333 on the forest floor (Greenaway and Raghaven, 1998: Green et al., 1999). Although glycogen stores in the muscle tissue of red crabs appeared to decrease during the dry seasons (June 1994 and pre-mig 1993), there was no corresponding decrease in triglyceride levels nor in muscle protein (S. Morris, unpub. data). More frequent sampling throughout the year is now required to determine the seasonal impor- tance of energy storage in this species. During the two migratory seasons red crabs apparently utilized neither triglycerides nor stores of muscle glycogen during the downward migration. However, both male and female crabs sampled on their return migration had substan- tially reduced glycogen reserves in both the muscle and the MGG. Interestingly, although the females stayed on the terraces 2 weeks longer than the male crabs (Hicks et al., 1990; Adamczewska and Morris, 2001), the glycogen re- serves were depleted to a similar extent in both sexes. Assuming that 20<7r of the body mass is muscle tissue (Giese. 1966; Cameron and Wood, 1985) and using pub- lished values of MO2 (212 jimo! O2 • kg'1 • min~': Adam- czewska and Morris, 1994b), the rate of glucose utilization in red crabs during walking would be 0.636 mmol of glu- cose per hour. The glycogen used by the red crabs during either the breeding activities on the terraces or the return migration was 26 and 53 mmol glucose • kg"1 in the two seasons and would be sufficient to provide fuel for about 2.5 h and 5 h of activity in 1993 and 1995 respectively. The glucose and glycogen stores utilized from the MGG would provide metabolic fuel for a further 0.5-2 h of walking. Therefore, considering that the red crabs spend a week or more engaged in breeding activities on the terraces, the glycogen stores used would only be sufficient to fuel a small portion of the active metabolism; the remainder must be supported by food intake (Greenaway and Linton, 1995; Green, 1997). Sampling of G. natalis during the migration has provided unique information about the energy metabolism of these crabs and their capacity to undertake the annual migration. Perhaps surprisingly, it has become clear that the red crabs undertake the annual migration with minimal reliance on energy stores and support prolonged locomotion aerobi- cally. In contrast, the recruitment of anaerobiosis by male red crabs while defending and maintaining their burrows indicates that these breeding activities are more energeti- cally demanding than the actual migration. Above all, it seems clear that migrating red crabs possess a physiological capacity and exhibit physiological responses different from those of the crabs held in the laboratory and from those of non-migrating crabs in the field. Laboratory and further field-based investigations of the regulation of their metab- olism are now required, especially of possible seasonal hormonal signals that might alter metabolic state. Acknowledgments This work is dedicated to the memory of Dr. Holger Rumpff, without whose commitment, enthusiasm, and es- pecially friendship these studies would not have been pos- sible. We thank the Government Conservators and staff of Parks Australia, Christmas Island, for their assistance and hospitality. We also express our appreciation to the two anonymous reviewers who put so much time into helping us to improve the manuscript. Thanks also to Ansett Airlines for provision and shipment of the dry ice and samples. Literature Cited Adamczewska, A. M., and S. Morris. 1994a. Exercise in the terrestrial Christmas Island red crab Gecarcoidea natalis. I. Blood gas transport. J. Exp. Biol. 188: 235-256. Adamczewska, A. M., and S. Morris. 1994b. Exercise in the terrestrial Christmas Island red crab Gecarcoidea natalis. II. Energetics of loco- motion. J. Exp. Bini 188: 257-274. Adamczewska, A. M., and S. Morris. 1996. The respiratory gas trans- port, acid-base state, ion and metabolite status of the Christmas Island blue crab. Cardisoma hirtipes (Dana) assessed in situ with respect to immersion. Physiol. Zoo/. 69: 67-92. Adamczewska, A. M., and S. Morris. 1998a. Strategies for migration in the terrestrial Christmas Island red crab Gecarcoidea natalis: intermit- tent versus continuous locomotion. J. Exp. Biol. 201: 3221-3231. Adamczewska, A. M., and S. Morris. 1998b. The functioning of the haemocyanin of the terrestrial Christmas Island red crab Gecarcoidea natalis and roles for organic modulators. J. Exp. Biol. 201: 3233-3244. Adamczewska, A. M., and S. Morris. 2000. Respiratory gas transport, metabolic status and locomotor capacity of the Christmas Island red crab Gecarcoidea natalis assessed in the field with respect to dichot- omous seasonal activity levels. J. Exp. Zoo/. 286: 552-562. Adamczewska, A. M., and S. Morris. 2001. Ecology and behavior of Gecarcoidea natalis, the Christmas Island red crab, during the annual breeding migration Biol. Bull. 200: 305-320. Albert, J. L., and VV. R. Ellington. 1985. Patterns of energy metabolism in the shore crab, Menippe mercenaria. during severe hypoxia and subsequent recovery. J. E.\p. Zoo/. 234: 175-183. Bergmeyer, H. U. 1985. Methods of Enzymatic Analysis. 3rd ed. Vol. VI. Metabolites 1 : Carbohydrates. VCH. Weinheim, Germany. Bliss. D. E., J. Van Montfrans, M. Van Montfrans, and J. R. Boyer. 1978. Behavior and growth of the land crab Gecarcinus lateralis (Freminvillel in southern Florida. Bull. Am. Mus. Nat. Hist. 160: 111-152. Booth. C. E., and B. R. McMahon. 1985. Short communications. Lac- tate dynamics during locomotor activity in the blue crab. Callinectes sapidus. J. Exp. Biol. 118: 461-465. Booth, C. E., B. R. McMahon, and A. VV. Finder. 1982. Oxygen uptake and the potentiating effects of increased hemolymph lactate on oxygen transport during exercise in the blue crab. Callinectes sapidus. J. Comp. Physiol. 148: 111-121. Booth, C. E., B. R. McMahon. P. L. DeFur, and P. R. H. Wilkes. 1984. Acid-base regulation during exercise and recovery in the blue crab, Callinectes sapidus. Respir. Physiol. 58: 395-376. Burggren, W. W., and B. R. McMahon. 1981. Hemolymph oxygen transport, acid-base status, and hydrommeral legulation during dehy- dration in three terrestrial crabs, Cardisoma, Birgus and Coenobita. J. Exp. Zoo/. 218: 53-64. Cameron, J. N., and C. M. Wood. 1985. Apparent H* excretion and CO, dynamics accompanying carapace mineralization in the blue crab (Callinectes sapidus) following moulting. J. Exp. Biol. 114: 181-196. 334 A. M. ADAMCZEWSKA AND S. MORRIS Chang, E. S., and J. D. O'Connor. 1983. Metabolism and transport of carbohydrates and lipids. Pp. 263-287 in The Biology oflhe Crustacea, Vol. 5. L. H. Mantel, ed.. Academic Press. New York. Ellington, W. R. 1983. The recovery from anaerobic metabolism in invertebrates. J. Exp. Zool. 228: 431-444. England, W. R., and J. Baldwin. 1983. Anaerobic energy metabolism in the tail musculature of the Australian yabby Cherax destructor (Crustacea, Decapoda, Parastacidae): role of phosphagens and anaero- bic glycolysis during escape behavior. Ph\siol. Zool. 56: 614-622. Farrelly C. A., and P. Greenaway. 1994. Gas exchange through the lungs and gills in air breathing crabs. J. Exp. Biol. 187: 113-130. Forster, M. E., F. M. Waldron, and H. H. Taylor. 1989. Recovery from exhausting exercise in a bimodally breathing crab. Leptograpsus variegatus (Decapoda: Grapsidae). J. Exp. Mar. Biol. Ecol. 127: 165- 173. Gade, G. 1984. Effects of oxygen deprivation during anoxia and mus- cular work on the energy metabolism of the crayfish, Orconectes limosus. Comp. Biochem. Physiol. 77A: 495-502. Gibson, R., and P. L. Barker. 1979. The decapod hepatopancreas. Oceanogr. Mar. Biol Anna. Rev. 17: 285-346. Gibson-Hill, C. A. 1947. Field notes on the terrestrial crabs. Bull. Raffles Mus. 18: 43-52. Giese, A. C. 1966. Lipids in the economy of marine invertebrates. Physiol. Rev. 46: 244-298. Gifford, C. A. 1962. Some observations on the general biology of the land crab, Cardisoma guanhumi (Latreille) in south Florida. Biol. Bull. 123: 207-223. Green, P. T. 1993. The role of Red Land Crabs in structuring rain forest on Christmas Island, Indian Ocean. Ph.D. thesis, Monash University, Clayton, Victoria, Australia. Green, P. T. 1997. Red crabs in rain forest on Christmas Island, Indian Ocean: activity patterns, density and biomass. / Trap. Ecol. 13: 17-38. Green P. T., P. S. Lake, and D. J. O'Dowd. 1999 Monopolization of litter processing by a dominant land crab on a tropical oceanic island. Oecologia. 119: 435-444. Greenaway, P. 1994. Salt and water balance in field populations of the terrestrial crab Gecarcoidea nalalis. J. Crustac. Biol. 14: 435-453. Greenaway, P., and S. M. Linton. 1995. Dietary assimilation and food retention time in the herbivorous terrestrial crab Gecarcoidea natalis. Physiol. Zool. 68(6): 1006-1028. Greenaway, P., and S. Raghaven, 1998. Digestive strategies in two species of leaf-eating land crabs (Brachyura: Gecarcimdae) in a rain forest. Physiol. Zool. 71: 36-44. Greenaway P., S. Morris, and B. R. McMahon. 1998. Adaptations to a terrestrial existence by the Robber Crab Birgus latro. II. In vivo respiratory gas exchange and transport. J. Exp. Biol. 140: 493-509. Harris, R. R., and M. B. Andrews. 1982. Extracellular fluid volume changes in Carcinus maenas during acclimation to high and low environmental salinities. J. Exp. Biol. 99: 161-173. Harris, R. R., and G. A. Kormanik. 1981. Salt and water balance and antennal gland function in three Pacific species of terrestrial crab (Gecarcoidea lalandii, Cardisoma camifex, Birgus latro). II. The ef- fects of desiccation. J. Exp. Zool. 218: 107-116. Head, G., and J. Baldwin. 1986. Energy metabolism and the fate of lactate during recovery from exercise in the Australian freshwater crayfish Cherax destructor. Aust. J. Mar. Freshw. Res. 37: 641-646. Heath, J. R., and H. Barnes. 1970. Some changes in biochemical composition with season and during the moulting cycle of the common crab Carcinus maenas (L.). J. Exp. Mar. Biol. Ecol. 5: 199-233. Henry, R. P., G. A. Kormanik, N. J. Smatresk, and J. N. Cameron. 1981. The role of shell dissolution as a source of HCO,~ for buff- ering hypercapnic acidosis in aquatic and terrestrial decapod crusta- ceans. J. Exp. Biol 94: 269-274. Henry, R. P., C. E. Booth, F. H. Lallier, and P. J. Walsh. 1994. Post-exercise lactate production and metabolism in three species of aquatic and terrestrial decapod crustaceans. / Exp. Biol. 186: 21 5-234. Herreid, C. F. 1980. Review: Hypoxia in invertebrates. Comp. Biochem. Physiol. 67 A: 311-320. Herreid, C. F., and R. J. Full. 1988. Energetics and locomotion. Pp. 333-377 in Biology of the Land Crabs, W. W. Burggren and B. R. McMahon, eds. Cambridge University Press, New York. Hicks, J. W. 1985. The breeding behaviour and migrations of the ter- restrial crab Gecarcoidea natalis. (Decapoda: Brachyura). Aust. J. Zool 33: 127-142. Hicks, J., H. Rumpff, and H. Yorkston. 1990. Christmas Crabs. Golden Earth Design and Printing. Singapore. Hilmy, A. M., N. F. Abd EI-Hamid, and K. H. G. Adham. 1986. Glycogen content of some organs of two marine crustaceans. Bull. Inst. Oceanogr. Fish. 12: 215-229. Klaassen, F. 1975. Ecological and ethological studies on the reproduc- tive biology of Gecarcinus lateralis (Decapoda, Brachyura). Forma Function 8: 101-174. Kucharski, L. C. R., and R. S. M. Da Silva. 1991. Effect of diet composition on the carbohydrate and lipid metabolism in an estuanne crab, Chasmagnathus granulata (Dana, 1851). Comp. Biochem. Physiol. 99A: 215-218. McMahon, B. R. 1981. Oxygen uptake and acid-base balance during activity in decapod crustaceans. Pp. 299-335 in Locomotion and En- ergetics in Arthropods. C. F. Herreid and C. R. Fourtner. eds.. Plenum Press, New York. McMahon, B. R., D. G. McDonald, and C. M. Wood. 1979. Ventila- tion, oxygen uptake and haemolymph oxygen transport, following enforced exhausting activity in the Dungeness crab Cancer magister. J. Exp. Biol. 80: 271-285. Milligan, C. L., P. J. Walsh, C. E. Booth, and D. G. McDonald. 1989. Intracellular acid-base regulation during recovery from locomotor ac- tivity in the blue crab Callinectes sapidus. Physiol. Zool. 62: 621- 638. Nishida, S., Y. Takahashi, and J. Kittaka. 1995. Structural changes in the hepatopancreas of the rock lobster. Jasus edwardsii (Crustacea, Palmudidae) during development from the puerulus to post-puerulus. Mar. Biol 123: 837-844. O'Mahoney, P. M., and R. J. Full. 1984. Respiration of crabs in air and water. Comp. Biochem. Physiol 79A: 275-282. Parvathy, K. 1971. Glycogen storage in relation to the moult cycle in two crustaceans Emerita asiatica and Ligia exotica. Mar. Biol. 10: 82-86. Pillay, K. K., and N. B. Nair. 1973. Observations on the biochemical changes in gonads and other organs of Uca annulipes. Ponunus pe- lagicus and Metapenaeus affinis (Decapoda: Crustacea) during the reproductive cycle. Mar. Biol 18: 167-198. Schirf, V. R., T. Turner, L. Selby, C. Hannapel, P. Dela Cruz, and P. F. Dehn. 1987. Nutritional status and energy metabolism of crayfish (Procambarus clarkii. Girard) muscle and hepatopancreas. Comp. Bio- chem. Physiol. 88A: 383-386. Smatresk, N. J., and J. N. Cameron. 1981. Post-exercise acid-base balance and ventilatory control in Birgus latro, the coconut crab. J. Exp. Zool. 218: 75-82. Smatresk, N. J., A. J. Preslar, and J. N. Cameron. 1979. Post-exercise acid-base disturbance in Gecarcinus lateralis. a terrestrial crab. J. Exp. Zool. 210: 205-210. Taylor. H. H., and P. Greenaway. 1994. The partitioning of water and solutes during evaporative water loss in three terrestrial crabs. Physiol. Zool. 67: 539-565. Thorpe, E., A. C. Taylor, and F. A. Huntingford. 1995. How costly is fighting? Physiological effects of sustained exercise and fighting in swimming crabs, Necora puber (L.) (Brachyura, Portunidae). Anim. Behav. 50: 1657-1666. MIGRATION PHYSIOLOGY OF RED CRABS 335 Tsai, D. E., H. Chen, and C. Tsai. 1984. Total lipid and cholesterol content in the blue crab, Cullinectes sapitlus Rathhun. Comp. Biochem. Physiol. 78B: 27-31. van Aardt, W. J. 1988. Lactate metabolism and glucose patterns in the river crab, Potamonautes warreni Caiman, during anoxia and subse- quent recovery. Comp. Biochem. Phvsiol. 91A: 299-304. van Aardt, W. J. 1990. Oxygen uptake and haemocyanin oxygen affin- ity of Potamonautes warreni Caman after exercise. S. Afr. J. Zoo/. 25: 11-17. Vonk, H. J. 1960. Digestion and metabolism. Pp. 291-316 in Physiol- ogy of Crustacea. Vol. 1 . T. H. Waterman, ed. Academic Press, New York. Weinslein, R. B., and R. J. Full. 1992. Intermittent exercise alters endurance in an eight-legged ectotherm. Am. J. Physiol. 262: R852- R859. Weinstein R. B., R. J. Full, and A. N. Ahn. 1994. Moderate dehydra- tion decreases locomotor performance of the ghost crab Ocypode quadrata. Physiol. Zool. 67: 873-891. Wolcott, T. G. 1998. Ecology. Pp. 55-96 in Biology of the Land Crabs. W. W. Burggren and B. R. McMahon, eds. Cambridge University Press, Cambridge. Wolcott, T. G. 1992. Water and solute balance in the transition to land. Am. Zool. 32: 428-437. Wolcott, T. G., and D. L. Wolcott. 1985. Factors influencing the limits of migratory movements in terrestrial crustaceans. Contrib. Mar. Sci. Suppl. 27: 257-273. Wood, C. M., and D. J. Randall. 1981a. Oxygen and carbon dioxide exchange during exercise in the land crab Cardisoma carnifex. J. Exp. Zool. 218: 7-22. Wood, C. M., and D. J. Randall. 1981b. Haemolymph gas transport, acid-base regulation, and anaerobic metabolism during exercise in the land crab Cardisoma camifex. J. Exp. Zool. 218: 23-35. Reference: Bio/. Bull. 200: 336-343. (June 2001) Morphology of the Symbiosis Between Corculum cardissa (Mollusca: Bivalvia) and Symbiodinium corculorum (Dinophyceae) MARK A. FARMER,1 WILLIAM K. FITT2 *, AND ROBERT K. TRENCH3 ^Department of Cellular Biolog\, and 2 Institute of Ecology, University of Georgia, Athens, GA 30602 USA, ^Department of Ecology, Evolution and Marine Biology, University of California at Santa Barbara, Santa Barbara, CA 93106 USA Abstract. Light and transmission electron microscopy of tissues of the symbiotic clam Corculum cardissa (L) showed that a symbiotic dinoflagellate. Symbiodinium cor- culorum (Trench), is found predominantly in the mantle and the gills. The data suggest that in C. cardissa the algae are located in a zooxanthellal tubular system that is associated with the hemocoel and is similar to that seen in tridacnine ("giant") clams. The algae occur within the lumen of the tertiary tubules and are thus separated from the hemolymph by a tissue that is one cell layer thick. Under a light microscope the tertiary tubules appear as rows of symbionts originating from the digestive diverticulum, presumably branching from the primary tubules that are also seen in symbiotic tridacnine clams. This morphological arrange- ment is discussed with regard to the ontogeny and the evolution of the tubular system within symbiotic bivalves. Introduction Several species of marine bivalves in the family Cardi- idae harbor symbiotic dinoflagellates that belong to the genus Symbiodinium. These bivalves include all of the spe- cies in the subfamily Tridacninae, including the well-known genera of larger clams, Tridacna and Hippopus, as well as less well-known genera of much smaller clams in the sub- family Fraginae, such as Corculum and Fragum (Kawaguti, 1950, 1983; Schneider, 1998). For many years, the symbi- otic algae in tridacnines were depicted as being located in the hemal spaces, whence they were culled by wandering Received 10 June 1999; accepted 14 February 2001. * To whom correspondence should be addressed. E-mail: fitt@ sparrow.ecology.uga.edu amoebocytes and digested in the digestive gland; the indi- gestible remains were thought to reside in the kidneys (e.g., Yonge, 1936, 1953, 1975, 1980; Goreau et al., 1973). How- ever, a system of tubules, arising from one of the divertic- ular ducts of the stomach of Tridacna and ramifying through much of the clam and containing the symbionts, was described by K. Mansour (1946a, b), but forgotten. Finally, 46 years later, the "zooxanthellal tubular system" was redescribed by Norton et al. (1992). who proposed that the primary, secondary, and blind-ended tertiary tubes of the tubular system do not connect with the hemocoel; therefore the algae are not found in the hemolymph compartment (see Fitt, 1993, for a review). In the heart cockle, Corculum cardissa. as in the giant tridacnine clams, symbiotic dinoflagellates are located in the mantle tissue; but unlike the tridacnine. C. cardissa has many algae located in the gills as well (Kawaguti, 1968). Early electron microscopic images, in both instances, were interpreted as indicating that the algae are within the hemal system (Kawaguti, 1966, 1968). This interpretation was consistent with the author's observations that blood cells are apparently in contact with the algae in C. cardissa and Tridacna (Kawaguti, 1966, 1968). The occurrence of symbiotic algae in a tubular system in Tridacna (Mansour, 1946a, b; Norton et al., 1992) raises the question of whether a similar zooxanthellal tubular system also occurs in the cockle Corculum car- dissa. or any other related species (i.e., Fragum spp.). The goal of this study was to document evidence of a tubular system in C. cardissa, and to determine whether the tubules would penetrate the gill tissue, a conceptually difficult morphology. In the current study, ultrastructural observations indicate that a tubular system also exists in 336 ALGAL SYMBIOSIS IN THE HEART COCKLE CORCULUM 337 *«v»; Figure 1. Light micrograph of a paraffin-embedded section through the mantle tissue of Tridacna maxima showing rows of symbiotic algae (arrowheads) in tertiary tubules. Iridophores (ir), animal cells with crystalline proteins that refract light, are found in the mantles of all species of clams containing symbiotic algae. Scale bar, 20 /im. Figure 2. Light micrograph of a paraffin-embedded section through the gill tissue of Corculum cardissa showing rows of symbiotic algae in tertiary tubules, (a) Overview; scale bar = 12 fj.m. (b) Higher magnification, scale bar = 10 ^im. Corculum; that the symbiotic dinoflagellates occur within the lumina of the tubes, which themselves are located within the hemocoel: and that the algae within the tertiary tubules are separated from the hemolymph by a tissue that is mostly only one cell layer thick. Materials and Methods Corculum cardissa (Linne) was collected from the sandy reef flat at about 0.5 m depth in Belau (Palau), Western Caroline Islands. Animals were fixed in 6% glutaraldehyde. 338 M. A. FARMER ET AL. postfixed in 3% osmium tetraoxide, dehydrated, and embed- ded in Spurr's medium as previously described (Trench et al., 1981). The tissues of Tridacna spp. were fixed, embed- ded, and observed as described in Trench et al. (1981). Thick sections (1 /im) were prepared for examination by light microscopy on an LKM Ultratome V. These were photographed with an Olympus Vanox microscope and a PM-10 camera. Ultrathin sections for electron microscopic examination were prepared on an RMC-6000 ultrami- crotome, stained with uranyl acetate and lead citrate in the standard manner, and observed and photographed with a Philips 400 transmission electron microscope (TEM). Results and Discussion Light microscopic examination of Corculitm cardissa revealed that, similar to observations made on symbionts living in mantle tissues of Tridacna (Fig. 1; Mansour, 1946a; Fitt and Trench, 1981; Norton et al., 1992; Norton and Jones, 1992), algal cells are arranged in rows in both the mantle (e.g., Kawaguti, 1968; Figs. 1, 2) and the gills (Fig. 2). However light microscope observations could not re- solve the tertiary tubule structure in either genus of clam. Electron microscopic examination of the gills of Corcu- lum (Fig. 3) shows that the algae are indeed juxtaposed to animal blood cells. Kawaguti's (1968) early descriptions from C. cardissa note that algae are sometimes accompa- nied by "wandering cells," but he includes no figures. In contrast, TEM pictures of symbionts in Tridacna crocea and T. maxima clearly show nearby animal cells (Kawaguti, 1966; Fitt and Trench, 1981). Algal symbionts in C. car- dissa are not in direct contact with the animal's blood cells, but are separated from the hemolymph and the blood cells Figure 3. Transmission electron micrograph of a portion of the gill of Corculum cardissa showing a portion of the tertiary tubule (t), and a blood cell (BC) close to cells of the alga Symbiodiniwn corculorum (Sc) in the tubule. Scale bar. 1 /urn. ALGAL SYMBIOSIS IN THE HEART COCKLE CORCULUM 339 Figure 4. Transmission electron micrograph of the relation between the symbiotic algae, the tubule cells, and the blood cells in gill tissue of Corculum cardissa. (a) An algal cell closely appressed to a tubule cell. The algal cell wall is juxtaposed to the tubule cell plasmalemma, which can be followed around the enclosed cell nucleus, which it encloses, (b) Two algal cells in adjacent tertiary tubules, separated by the cytoplasm of the two tubule cells (tc). A blood cell is close by (be). Scale bars. 1 /j.m. by the cells of the tubules, which at the tertiary level are about one cell layer thick (Figs. 3-5). Evidence for this comes from closer examination of high-magnification TEM images of the structural relations between the algae, the cells of the tubules, and the blood cells (Fig. 4). First, algae in the lumina of the tubules are often pressed against the inner plasmalemma of the tubular cells; when nuclei of the appressed tubular cells are apparent (as in Fig. 4a), this could lead to the interpretation that the algae are intracel- lular (e.g., Kawaguti. 1968). Second, the algae are clearly separated from the molluscan blood cells by the cells com- posing the tubules (Fig. 4b). Overlapping cell processes form the tertiary tubules (Fig. 5). Two or more unseparated adjacent symbionts (Fig. 6) also indicate that the algae are in tubules and not living intracellularly within host cells. In bivalves that harbor symbiotic dinoflagellates. the structure of the tubular system, in which the tubules arise from the digestive system and are contiguous with it, sug- gests that the morphological and functional relation between host and symbionts in bivalves is directly analogous to that found in symbiotic cnidarians (Fitt, 1993). In both cases, the algae enter the digestive system via the mouth. In bivalves, symbiotic dinoflagellates enter via the mouth, and exit via the anus (Ricard and Salvat, 1977: Trench et al., 1981; Maruyama and Heslinga, 1997); their entire residence in the clam is in association with the digestive system. In contrast, symbionts in cnidarians enter and exit via the mouth and eventually take up residence inside of host digestive cells. The location of the symbionts in bivalves and cnidarians is also analogous with respect to metabolite flux between host 340 M. A. FARMER ET AL. Figure 5. Transmission electron micrograph of an alga in the gill of Corculum cardissa surrounded by overlapping processes of tertiary tubule cells (to. Scale bar. 1 /j.m. and symbiont (Fitt et al., 1985). In bivalves, where the algae are intercellular, they are separated from the hemolymph (circulating nutrients) by the proximal and distal plasma- lemma of the tubule cells (Fig. 7). Hence, nutrient exchange between the algae and the hemolymph of bivalves may potentially be regulated by the tubule cells. In cnidarians. where the algae are intracellular, the symbionts are sepa- rated from their nutrient source, the gastrovascular system, by two membranes, the host cell plasmalemma and the symbiosome membrane. The presence of the tubular system in symbiotic bivalves is also significant from ontogenetic and evolutionary per- spectives. Studies of algal symbioses in tridacnines (Fitt and Trench, 1981) clearly show that the tubules develop only in the presence of dinoflagellate symbionts; the algae in the tubules are observed as "rows extending from the region of the stomach and digestive gland toward the developing siphonal tissue" (Fitt et al., 1981 ). Juvenile clams that were allowed to develop in the absence of Symbiodinium did not show evidence of this feature. From the report of Norton et u/. ( 1995). it is also apparent that, when the algae are lost from Tridacna during events of thermal "stress," the tubules atrophy. Whether the tubules are reformed should the sym- biosis recover, or whether lack of recovery of bleached clams is the result of the inability of the tubular system to regenerate, is unknown. For marine symbioses, only two other instances have been recorded in which the symbionts appear to play a significant role in the ontogenetic event in the host. One example is the process of strobilation in symbiotic scypho- zoans such as Mastigias (Sugiura, 1964) and Cassiopeia (Colley and Trench, 1985); these jellyfish produce ephyrae only in the presence of Symbiodinium. Another example is the influence that Vibrio fischeri. a symbiotic luminous bacteria, has on morphogenesis of certain parts of the light organ in the squid Eprymna scolopes (Claes and Dunlap. ALGAL SYMBIOSIS IN THE HEART COCKLE CORCULUM 341 Figure 6. Symbiotic algae in tertiary tubules in the gills of Corculum cardissa, showing (a) several symbionts in a tubule and (b) high magnification of the two cells in 6a. Scale bar = 1 /j.m. Figure 7. Schematic representation of the relation between the alga (A), the tubule cells of the tertiary tubules (TC), and the hemolymph containing blood cells (be), n, nucleus; cp. chloroplast. 2000). In neither example are the "signals" that elicit the developmental response in the host known. In the case of the dinoflagellate associations, the exopolysaccharides ex- uded by Symbiontinium (Markell el al., 1992; Markell and Trench, 1993) may be a source of the signals. For instance, in the process of root nodulation in leguminous plants, the initiation of root hair curling and infection thread formation are dependent on chemical signals from the bacterial sym- bionts (Brewin. 1991). A system of tubules originating in the stomach and ramifying through the hemolymph is uncommon in bi- valve molluscs and appears to be directly related to symbiosis with dinoflagellates. As far as is known, no nonsymbiotic bivalves demonstrate this feature. In addi- tion to the tridacnine clams and Corculum cardissa de- scribed here, the bivalve Fraguin fragum has also been reported to harbor symbiotic dinoflagellates (Kawaguti, 1983). and recent TEM images (Kempf, unpubl.) show morphological features similar to those presented here for C. cardissa. This finding supports the interpretation 342 M. A. FARMER ET AL that all algal symbionts in molluscs occur in tubule extensions of the digestive system. Cladistic analyses based on morphological characters (Schneider, 1992, 1998) and phylogenetic relationships based on analysis of small subunit ribosomal RNA gene sequences (Ma- ruyama er al.. 1998) both indicate that the known bi- valves with symbiotic dinoflagellates are closely related, all belonging to Cardiidae. In addition, the available molecular genetic evidence (NcNally et al., 1994) reveals that the symbiotic algae associated with Corculum and Tridacna are also very closely related (LaJeunesse, 2000), but not identical. We also suppose that freshwater bivalves, such as Anodonta, that are symbiotic with the green alga Chlorella sp. (Pardy, 1980) may demonstrate a tubular structure in which to house the algae, as these symbionts probably also enter their hosts through the digestive system. The only other molluscan group that shows an analogous morphology is the opisthobranch gastropods: some sacco- glossan opisthobranchs temporarily harbor derived chloro- plasts from feeding (Trench, 1975), and eolid nudibranchs often maintain dinoflagellates for a short time after feeding on symbiotic cnidarians (Kempf, 1984). Most significantly in relation to symbiotic bivalves, in eolidacean nudibranchs "branches of the posterior aorta . . . accompany the branches of the midgut gland (digestive diverticulum) into the cerata . . ." (Hyman, 1967, p. 477), suggesting development of tubules in conjunction with development of blood vessels, or vice versa. We speculate that bivalves, like their gastropod relatives, possess a suite of genes that encode the expression of the tubular system, but these genes are expressed only after activation by some "signal" produced by dinoflagellate symbionts as they enter the host digestive tract. These situations would be analogous to the production of various galls in plants following infection by bacteria, fungi, in- sects, or other parasitic plants (Bidwell, 1979), all of which produce chemical signals. Acknowledgments We thank Professor Stephen Kempf for providing us with unpublished transmission electron micrographs of tissues of Fragum fragum, and two anonymous reviewers for sugges- tions on improving the text. WKF acknowledges support from NSF and the NOAA National Undersea Research Program (UNCW and CMRC). Literature Cited Bidwell, R. G. S. 1979. Plant Physiology. 2nd ed. MacMillan. New York. Brewin, N. J. 1991. Development of the legume root nodule. Annu. Rev. Cell Biol. 7: 191-226. Claes, M. F., and P. V. Dunlap. 2000. Aposymbiotic culture of the sepiolid squid Euprymna scohpes: role of the symbiotic bacterium Vibrio fischeri in host animal growth development, and light organ morphogenesis. / Exp. Zool 286: 280-296. Colley, N. J., and R. K. Trench. 1985. Cellular events in the re- establishment of a symbiosis between a marine dinoflagellate and a coelenterate. Cell Tissue Res. 239: 93-103. Fitt, W. K. 1993. Nutrition of giant clams. Pp. 31-40 in Biology and Mariculture of Giant Clams., W. K. Fitt, ed. ACIAR. Canberra, Aus- tralia. Fitt, W. K., and R. K. Trench. 1981. Spawning, development, and acquisition of zooxanthellae by Tridacna squamosa (Mollusca: Bi- valvia). Biol. Bull. 161: 213-235. Fitt, W. K., C. R. Risher, and R. K. Trench. 1981. Larval biology of tridacnid clams. Aquaculture 39: 181-195. Fitt, W. K., T. A. V. Rees, and D. Yellowlees. 1985. The relation- ship between pH and the availability of dissolved inorganic nitrogen in the zooxanthella-giant clam symbiosis. Limnol. Oceanogr. 40: 976- 982. Goreau, T. F., N. I. Goreau, and C. M. Yonge. 1973. On the utilization of photosynthetic products from zooxanthellae and of a dissolved amino acid in Tridacna maxima (Mollusca: Bivalvia). J. Zool. Land. 169: 417-454. Hyman, L. H. 1967. The Invertebrates. Vol. VI, Mollusca I. McGraw Hill. New York. Kawaguti, S. 1950. Observations on the heart cockle. Corculum cardissa (L). and its associated zooxanthellae. Pac. Sci. 4: 43-49. Kawaguti, S. 1966. Electron microscopy on the mantle of the giant clam with special reference to zooxanthellae and iridophores. Biol. J. Okayama Univ. 12: 81-92. Kawaguti, S. 1968. Electron microscopy on zooxanthellae in the mantle and gill of the heart shell. Biol. J. Okayama Univ. 14: 1-11. Kawaguti, S. 1983. The third record of an association between bivalve mollusks and zooxanthellae. Proc. Jpn. Acad. Ser. B 59: 17-20. Kempf, S. C. 1984. Symbiosis between the zooxanthella Symbiodinium (=Gymnodinium) microadriaticum. (Freudenthal) and four species of nudibranchs. Biol. Bull. 166: 110-126. LaJeunesse, T. C. 2000. Diversity, distribution and host specificity of algal symbionts of the dinoflagellate genus Symbiodinium. Ph.D dis- sertation. University of California Santa Barbara. Mansour, K. 1946a. Communication between the dorsal edge of the mantle and the stomach of Tridacna. Nature I Land.) 157: 844. Mansour, K. 1946b. Source and fate of the zooxanthellae of the visceral mass of Tridacna elongata. Nature (Land.) 158: 130. Markell, D. A., and R. K. Trench. 1993. Macromolecules exuded by symbiotic dinoflagellates in culture: Amino acid and sugar composi- tion. J. Phycol. 29: 64-68. Markell, D. A., R. K. Trench, and R. Iglesias-Prieto. 1992. Macro- molecules associated with the cell walls of symbiotic dinoflagellates. Symbiosis 12: 19-31. Maruyama, T., and G. A. Heslinga. 1997. Fecal discharge of zooxan- thellae in the giant clam Tridacna derasa. with reference to their growth rate. Mar. Biol. 127: 473-477. Maruyama, T., M. Ishikura, S. Yamazaki, and S. Kanai. 1998. Molecular phylogeny of zooxanthellate bivalves. Biol. Bull. 195: 70- 77. McNally, K. L., N. S. Govind, P. E. Thome, and R. K. Trench. 1994. Small subunit ribosomal DNA sequence analyses and a reconstruction of the inferred phylogeny among symbiotic dinoflagellates (Pyrrho- phyta). J. Phycol. 30: 316-329. Norton, J. H., and G. W. Jones. 1992. The giant clam: an anatomical and histological atlas. Australian Center for International Agricultural Research (ACIAR) Monograph Series, Canberra. 142 pp. Norton, J. H., M. A. Shepherd, H. M. Long, and W. K. Fitt. 1992. The ALGAL SYMBIOSIS IN THE HEART COCKLE CORCULUM 343 zooxanthellal tubular system in the giant clam. Biol. Bui/. 183: 503- 506. Norton, J. H., H. C. Prior, B. Baillie, and D. Yellowlees. 1995. Atro- phy of the zooxanthellal tubular system in bleached giant clams Tri- dacna gigas. J. Invertebr. Pathoi 66: 307-310. Pardy, R. L. 1980. Symbiotic algae and IJC incorporation in the fresh- water clam Anodonta. Biol. Bull. 158: 349-355. Ricard, M., and B. Salvat. 1977. Faeces of Tridacna maxima (Mollus- ca-Bivalvia), composition and coral reef importance. Proc. Third Int. Coral ReefSymp., Miami: 495-502. Schneider, J. A. 1992. Preliminary cladistic analysis of the bivalve family Cardiidae. Am. Malacol. Bull. 9: 145-155. Schneider, J. A. 1998. Phylogeny of the Cardiidae (Bivalvia): Phyloge- netic relationships and morphological evolution within the subfamilies Clinocardiinae. Lymnocardiinae. Fragmae and Tridacninae. Malacolo- gia 40: 321-373. Sugiura, Y. 1964. On the life history of rhizostome medusae. II. Indis- pensability of zooxanthellae for strobilation in Masligias papua. Em- bryologia 8: 223-233. Trench, R. K., 1975. Of "leaves that crawl": functional chloroplasts in animal cells. Soc. Exp. Biol. Symp. XXIX: 229-265. Trench, R. K., D. S. Wethey, and J. W. Porter. 1981. Observations on the symbiosis with zooxanthellae among the tndacnidae (Mollusca, Bivalvia). Biol. Bull. 161: 180-198. Yonge, C. M. 1936. Mode of life, feeding, digestion and symbiosis with zooxanthellae in the Tridacndae. Scientific Report, Great Barrier Reef Expedition 1: 283-331. Yonge, C. M. 1953. Mantle chambers and water circulation in the Tridacnidae. Proc. Zoo/. Soc. Land. 123: 551-561. Yonge, C. M. 1975. Giant clams. Sci. Am. 232: 96-105. Yonge, C. M. 1980. Functional morphology and evolution in the Tri- dacnidae. Rec. Aust. Mus. 33: 735-777. Reference: Bio/. Bull. 200: 344-350. (June 2001) 5-Hydroxytryptamine Stimulates Net Ca2+ Flux in the Ventricular Muscle of a Mollusc (Busycon canaliculatum} During Cardioexcitation C. LEAH DEVLIN Department of Biology, Penn State University, Abington College, Abington, Pennsylvania 19001 Abstract. Noninvasive, self-referencing calcium (Ca"+) electrodes were used to study the mechanisms by which 5-hydroxytryptamine (5-HT) affects net Ca2+ flux across the sarcolemma of myocytes from ventricular trabeculae (from a marine gastropod, Busycon canaliculatum). Treat- ment of isolated trabeculae with 5-HT causes a net Ca2 + efflux, which is 30% blocked by verapamil. These findings suggest that the efflux is in part the result of a previous Ca2+ influx through L-type Ca2+ channels and is due to a rapid Ca2+ extrusion mechanism inherent to the sarcolemma of these myocytes. 5-HT-induced net Ca2+ efflux is also re- duced by about 40% by treatment with a sodium (Na+ )-free, lithium ( Li + (-substituted saline, which shuts down the Na-Ca exchanger during Ca2 + extrusion. Cyclopiazonic acid (CPA). an inhibitor of the sarcoplasmic reticulum (SR) Ca2+ ATPase, almost completely abolishes the 5-HT-in- duced net Ca2 + efflux, suggesting that the SR rather than the extracellular pool is the primary Ca2+ reservoir serving 5-HT-induced excitation. Introduction The rhythmicity of molluscan cardiac muscle is regulated by numerous neurotransmitters and neuromodulatory agents, the most widely studied of these being the biogenic amine 5-hydroxytryptamine (5-HT), the tetrapeptide FMRF- amide. and acetylcholine (ACh). Pharmacological applica- tion of 5-HT onto the isolated hearts of the gastropods Busycon canaliculatum (Hill, 1958; Huddart and Hill, 1996), Lymnaea stagnalis (Bucket! et til., 1990), and Aply- sia californica (Liebeswar et al., 1975) causes a dose- dependent increase in inotropic and chronotropic activity Received 18 January 2000; accepted 17 February 2001. E-mail: cld5@psu.edu with a primary effect on long-duration chronotropic regu- lation. Underlying this potentiation in cardiac rhythmicity is an increase in the amplitude and frequency of cardiac action potentials as well as the coupled systolic force (Hill et al., 1992). The combined anatomical, pharmacological, and physiological studies on the gastropods Aplysia, Busycon, and Lymnaea have provided overwhelming evidence that 5-HT acts as a cardioexcitatory neurotransmitter in this class of mollusc. 45Ca2+ efflux studies have been conducted on a variety of molluscan cardiac and smooth muscles to determine how Ca2 + ions may be mobilized by 5-HT during the process of contraction or relaxation (Bloomquist and Curtis, 1972, 1975; Koch and Greenberg, 1981; Sawada et al., 1984; Ishii et til., 1989). The present study uses a newer technique to study Ca2+ flux, noninvasive self-referencing Ca2 + elec- trodes, to determine possible mechanisms by which 5-HT affects trans-sarcolemmal net Ca2+ flux in the Busycon ventricle as a clue to understanding its role during the excitation-contraction (E-C) coupling in the gastropod heart. An earlier study focused on the action of FMRFamide on net Ca2+ flux in the ventricle of Busycon canaliculatum (Devlin, 1997), so we may be able to compare mechanisms working at putative Ca2+ pools accessed by 5-HT and FMRFamide to achieve the same end — enhanced cardiac performance. Materials and Methods Specimens of Busycon canaliculatum, the channeled whelk, were obtained on the day of each experiment from the Marine Resources Center of the Marine Biological Lab- oratory (MBL). Woods Hole, Massachusetts. All experi- ments were conducted at the National Institutes of Health BioCurrents Research Center located at the MBL. 344 S-HT-INDUCED CA:+ FLUX IN MOLI.USCAN VENTRICLE 345 Preparation The shell was completely cut away from the animal with bone forceps, exposing the heart and its enveloping opaque pericardium. The pericardium was opened, and the ventricle was removed by severing its connections to the aorta and atrium. The ventricle was opened and pinned down in a dish of natural seawater to expose the inner latticework of tra- beculae. An individual trabecula was isolated with fine scissors from the inner wall of the ventricle and secured with minuten pins to a Sylgard well in a small recording chamber. The trabecula was bathed in a nominally Ca"+ free artificial seawater (ASW) prepared according to the MBL formula (423 mM Na + , 9.7 mM K + , 9.9 mM Ca2 + , 51.2 mM Mg2 + , 538.6 mM Cl", 27 mM SO4~2, 2.3 mM HCO3 ). Magnesium was used to replace the 9.9 mM Ca2 + omitted from the Ca2+ -free ASW. The nominal (or back- ground) amounts of Ca2+ ions still present in the Ca2+-free ASW were typically 100 micromolar (/o,M) or less and were monitored continuously with a Ca2+ electrode throughout the experiment. Ion flux measurement The self-referencing ion electrode technique is used to detect net ion flux generated from a biological source, in this case, the sarcolemma of myocytes from a Busycon trabec- ula. The net ion flux across the membrane is the sum of both inward and outward ion movements. In the present experi- ments, electrodes loaded with a Ca2+-specific ionophore (Fluka Chemika Ca2 + ionophore — cocktail A with the neu- tral carrier ETH 1001 ) were used to detect net Ca2+ flux (in pmol cm"2 s"1) across the sarcolemma of the myocytes. The electrode was programmed with PC-based software (lonprobe) to oscillate with an excursion of 10 jam and a slow frequency of 0.3 Hz; this minimized mixing of the bathing saline. The difference in voltage (;u,V) between the two ends of the excursion was measured and could be taken as corresponding to a Ca2+ concentration gradient, since the excursion was constant. To construct the electrodes, borosilicate micropipettes were pulled from 1.5-mm-diameter glass capillaries (World Precision Instruments. Inc.), then back-filled with 100 mM CaCl2 in 0.1% agar. The pipette was front-filled with cal- cium ionophore — cocktail A containing neutral carrier ETH 1001 (Fluka Chemika) to produce an ionophore column of 40 n.m. A Ag/AgCl wire inserted into the back of the micropipette served as the coupling to the headstage. To calibrate the electrode prior to the experiment, the electrode was tested in 0.1 mM, 1 mM, and 10 mM Ca24 solutions to check its Nernstian properties — that is, an approximate 28 mV difference per decade change in Ca2+ concentration. The return electrode was a Ag/AgCl wire inserted into a glass capillary containing 3 M KC1 agar. The electrode oscillated at right angles to the long axis of the trabecula and was positioned at two distances from the muscle surface. One electrode position was 500 ju-m away from the muscle surface. At this relatively great distance, Ca2 + flux from the ion source (i.e., the muscle) cannot be detected. Thus, only the control, background levels of Ca2 + in the saline are recorded at this position. The second electrode position was only 5 /nm from the muscle surface, so net Ca2+ flux could be measured directly at the sarco- lemma. From the background level of Ca2+ ions in the bathing saline, the Ca2+ ion concentration gradient at the muscle surface, and the diffusion constant forCa2+ ions, net Ca2+ ion flux was calculated using a modification of the Pick equation. The mathematical formulas used in the con- version of voltage to flux units, and other technical aspects and applications of the self-referencing electrode technique, are described by Smith et al. (1999). Pharmacological agents The neurotransmitter, 5-HT, and the L-type channel blocker, verapamil, were obtained from Sigma Chemical Company (St. Louis, MO). Verapamil was selected over other Ca2"1" blockers because of its consistent antagonistic action on other invertebrate cardiac and smooth muscle types (Devlin 1993a, b, 1997, Devlin and Smith, 1996). Diltiazem or nifedipine. which often act as agonists in molluscan muscle preparations, were not used in this study. The Ca2+ channel agonist, Bay K 8644, and the sarco/ endoplasmic reticulum Ca2+ ATPase (SERCA) inhibitor, cyclopiazonic acid (CPA), were obtained from Research Biochemicals International (Natick, MA). At the beginning of each experiment each drug was diluted in the nominal Ca2H"-free ASW to achieve the designated experimental concentration. Experimental protocol The following general protocol was used throughout the present experiments. An initial net Ca2+ efflux generated by 10 7 M 5-HT (the control) was recorded during a 3-min treatment; this net Ca2+ efflux immediately returns to basal levels when the 5-HT is washed from the trabecula by flushing with ASW for 20 to 30 min. Next, the trabecula was pretreated for 10 min with verapamil or Bay K 8644 (10 " M) or other inhibitors (CPA or Li* ASW). 5-HT ( K)"7 M) dissolved in solutions of verapamil or Bay K 8644 (10~5 M) or other inhibitors (CPA or Li+ ASW) was rcapplied to the trabecula for a 3-min treatment period. The control, the initial response to 5-HT alone, was then com- pared to the response to 5-HT in the presence of the agonist, antagonist, or inhibitor. A r-test analysis was then per- formed. 346 C. L. DEVLIN Figure 1. An experimental record showing a concentration-dependent effect of 5-HT on net Ca2 + efflux (in pmol cm"2 s~'). ASW = sponta- neous net Ca2 + flux measurement taken when the electrode was placed directly at the ion source, the trabecula, at a distance of less than 5 /^m during treatment with an artificial Ca2 + -free seawater (ASW). 5-HT (10~lc M to 10~5 M) = net Ca2+ efflux measurement taken when the electrode was still placed directly at the trabecula at a distance of less than 5 (xm during treatment with a range of 5-HT concentrations. Maximum net efflux was recorded between 10~s M and 10~7 M 5-HT. Background (BG) = small oscillations around the baseline that were recorded when the elec- trode was placed 500 /xni from the ion source, the trabecula. Results With the electrode placed 5 ju,m from the muscle surface, a basal net Ca2 + efflux of 1 .2 1 ± 0.42 pmol cm"2 s~ ' (n = 10) is recorded from the trabecula at rest in Ca"+-free ASW. A range of 5-HT concentrations (10~12 M to 10~5 M) was then tested. Above a threshold of about 10" I0 M, all con- centrations of 5-HT enhance basal net Ca~+ efflux. The maximal efflux is induced at 5-HT between 10~s-10~7 M; above 10~7 M, the net Ca2 + efflux is actually smaller than those induced by lower 5-HT concentrations (Fig. 1 ). Thus, the 5-HT receptor was desensitized by exposures to higher doses of its ligand. Because the effect of 10~7 M 5-HT does not desensitize the receptor and is completely reversible, it was chosen as the concentration to be challenged by various Ca2+ channel antagonists or agonists. A stable net Ca2 + efflux induced by 10~7 M 5-HT is on the order of 2.63 ± 1.01 pmol cm~2 s~' (n = 8). Effect of Ca2+ -channel antagonists or agonists on 5-HT- induced net CcT + efflux To test the hypothesis that the Ca2+ ions mobilized by 5-HT during E-C coupling are from the extracellular saline. I studied the effects of two L-type Ca2+ channel drugs, an antagonist (verapamil) and an agonist (Bay K 8644), on the 5-HT response. In concentration-response experiments con- ducted with verapamil or Bay K 8644 alone, neither drug has a significant effect on net Ca2+ flux over the concen- tration range tested (10~12 to 10~5 M} (not shown). How- ever, verapamil inhibits the action of 5-HT. reducing the response to 67% ± 16% (n = 3, P < 0.025) of the control (5-HT-induced net efflux measured prior to verapamil treat- ment) (Fig. 2). After verapamil was washed from the prep- aration, subsequent exposures to 5-HT are often enhanced. Bay K 8644 has no significant effect on the 5-HT response. The effect of a Nti + -free, Li+ -substituted ASW on 5-HT responses To determine whether the net Ca2+ efflux was mediated by the Na+-Ca2+ exchanger, sodium (Na+) ions were omit- ted from the bathing saline and replaced instead with an equivalent concentration (423 mM) of lithium (Li"1") ions; the rationale is that Li+ transverses the Na+ channel but cannot be substituted for Na+ in the Na+-Ca2 + exchanger (Lipp and Niggli, 1994). The 5-HT-induced net Ca2+ efflux is reduced by the presence of Li+ ions to 57% ± 9% (n = 3, P < 0.01) that of control (the 5-HT response in normal ASW) (Fig. 3). implicating the Na+-Ca2+ exchanger as a mechanism of Ca2"1" extrusion during excitation by 5-HT. Li"1" treatment alone reduced the basal net Ca"+ efflux to 69% ± 11% (H = 3. P <0.01). The effect of cyclopiazonic acid (CPA) on 5-HT responses Figure 4 shows the effect of SERCA inhibitor CPA ( 10~5 M) on 5-HT-induced net Ca2+ efflux. The 5-HT response is Figure 2. An experimental record showing that the 5-HT-induced net Ca: + efflux signal (in pmol cm"2 s~') is inhibited by the L-type channel antagonist verapamil (10~5 M). ASW = spontaneous net Ca2* efflux measurement taken when the electrode was placed directly at the ion source, the trabecula, at a distance of less than 5 ;um. This signal was recorded prior to any drug treatment when the trabecula was bathed only in an artificial Ca2""-free seawater (ASW). 5-HT (1st) = net Ca2+ efflux measurement taken when the electrode was placed directly at the trabecula during treatment with 10~7 M 5-HT. Verapamil = net Ca~+ efflux mea- surement taken when the electrode was placed directly at the trabecula after a 10-min treatment with verapamil (10~5 M). 5-HT in ver = net Ca2+ efflux measurement taken when the electrode was placed directly at the trabecula when treated with 5-HT ( 10~7 M) in verapamil (10~5 M). After verapamil was washed from the muscle, the final response to 5-HT (2nd) was greatly enhanced. All above recordings at the trabecula were taken at distance of 5 jim or less. Background (BG) = small oscillations around the baseline that were recorded when the electrode was placed 500 /j.m from the ion source, the trabecula. 5-HT-INDUCED CA:" FLUX IN MOLLUSCAN VENTRICLE 347 ^-^ 5 - o o. U Figure 3. The inhibitory effect of LiASW on 5-HT-induced net Ca2* efflux (in pmol cm~2 s~'l. First bar. background (BG): Net Ca2+ efflux measurement taken when the electrode was placed 500 j^m from the ion source, the trabecula. Second bar. spontaneous flux: Net Ca2* efflux measurement taken when the electrode was placed directly at the ion source, the trabecula. at a distance of less than 5 ^m during treatment with an artificial seawater. Third bar, 5-HT control: Net Ca:+ efflux measure- ment taken when the electrode was still placed directly at the trabecula during treatment with 5-HT (10~7 Ml Fourth bar: After washing the muscle for 30 min with artificial seawater. net Ca2* efflux measurement was taken when the electrode was placed directly at the ion source, the trabecula. at a distance of less than 5 /nm during treatment with a Na-free. Li-substituted artificial seawater (LiASW). Fifth bar. 5-HT in LiASW: Net Ca2~ efflux measurement taken when the electrode was still placed directly at the muscle during treatment with 5-HT (10~7 M) in LiASW. The data are means pooled from three muscles ± SD. inhibited by CPA to 18.5% ± 10% (;; = 3, P < 0.005) of the control (5-HT-induced flux prior to CPA treatment). CPA ( 10~5 M) alone also reduces basal net efflux to 76% ± 117r (n = 3, P < 0.025) that of the control. These data indicate that excitation by 5-HT relies more on Ca2 + ions from the sarcoplasmic reticulum (SR) than does spontane- ous myogenicity. The inhibitory effect of CPA on 5-HT responses is completely reversible. Discussion The present experiments showed that 5-HT stimulates a net efflux of Ca2 + in the ventricular muscle of the whelk Busvcon canaliculatitm. The Ca~* signal is sensitive to partial block by verapamil, which suggests that some of the Ca2 + movement is through sarcolemmal L-type Ca2 + chan- nels. These L-type channels account for about 20% of the Ca2 + mobilized into the internal compartment during E-C coupling and may provide the source for Ca"* -induced Ca2* release from the SR. resulting in increased systolic force. The remaining 80% of Ca2+ ions are probably re- leased from the SR. as shown by experiments with SERCA inhibitor CPA. which almost completely blocks the 5-HT response. Li+ ions, which act to block the Na-Ca exchange process, partially inhibit the 5-HT response, suggesting some reliance on the Na+-Ca2+ exchanger during the Ca2+ extrusion process. Verapamil. an L-type Ca2+ channel blocker. inhibits 5-HT-induced net Ca2+ efflux by about 30% in myocytes of the Busycon trabeculae. Similarly, in both gastropod (Hud- dart and Hill, 1996) and bivalve ventricles (Devlin. 1993)— where 5-HT enhances both inotropic and chronotropic re- sponses— the effect of 5-HT is dependent on the movement of extracellular Ca2+ ions through an L-type channel, since 5-HT responses are blocked by verapamil (Devlin. 1993b). Although verapamil has no significant effect on basal net Ca2+ efflux in the Busycon ventricle, it does inhibit the 5-HT response, suggesting a use-dependent mechanism of block. This same use-dependent block by verapamil also occurs in many other invertebrate and vertebrate muscle preparations that were either chemically or electrically stim- ulated (Lee and Tsien, 1983: Zahradnik and Zachar, 1983; Vaghy et al., 1988; Nanasi el «/.. 1990). Whereas verapamil inhibits the 5-HT-induced net Ca2+ efflux from the Busycon cardiac myocytes by about 30%, it Figure 4. The inhibitory effect of SERCA inhibitor, cyclopiazonic acid (CPA) on 5-HT-induced net Ca-* efflux (in pmol cm ). First bar. spontaneous flux: Net Ca2 + efflux measurement taken when electrode was placed directly at the ion source, the trabecula, at a distance of less than 5 p.m during treatment with an artificial seawater. Second bar. 5-HT control: Net Ca2* efflux measurement taken when the electrode was still placed directly at the trabecula dunng treatment with 5-HT (10~7 M). Third bar. 5-HT in CPA: Net Ca2 + efflux measurement taken when the electrode was still placed directly at the muscle during treatment with 5-HT (10~7 Af) in CPA ( 10~5 M). The data are means pooled from three experiments on three ventricles ± SD. 348 C. L. DEVLIN reduces the FMRFamide response by 60% in the same preparation (Devlin, 1997). Therefore, 5-HT relies less on extracellular Ca2+ ions than does FMRFamide during the process of E-C coupling in the Busycon ventricle. The results also point to two distinct Ca2 + release mechanisms used by these respective neurotransmitters. However, both 5-HT and FMRFamide activate L-type Ca2 + channels (whether directly or indirectly remains to be determined) during cardioexcitation, since verapamil partially inhibits net Ca2 + efflux induced by either chemical. This finding also indicates that the net Ca2+ efflux is in part an effect of an earlier Ca2+ influx through L-type channels, possibly carried by the HVA (high-voltage activated) current de- scribed by Yeoman el al. (1999). 5-HT (from 1(T9 to 10~6 M) depolarizes the heart cells from Helix pomatia (Kiss and S.-Rosza, 1978), Aplysia dactyiomela (Sawada et al., 1984), and Dolabella auricu- laria (Hill, 1974). This is the same concentration range in which 5-HT stimulates a net Ca2+ efflux from the ventric- ular myocytes of B. canalicitlatum. Two depolarizing Ca2 + currents, designated the LVA (low-voltage activated), a T-type current, and the HVA (high- voltage activated), an L-type current, have been identified in the ventricular cells of the gastropod Lvmnaea (Yeoman et al.. 1999). The sequential activation of the LVA and HVA currents pro- vides the mechanism for pacemaking, AP generation, and a Ca2+ source for E-C coupling in the gastropod ventricle (Yeoman et al., 1999). Since 5-HT increases the Ca2 + - dependent (and a Na + -dependent) component of cardiac APs and coupled systolic force in bivalve ventricles (Dev- lin, 1993b), potentiation of an HVA-like current may be involved given that the Ca2+-dependent component is sen- sitive to verapamil, diltiazem, and Bay K 8644. 5-HT also induces a Ca2+ current in Aplysia RB neurons (Pellmar. 1984) and mammalian neurons (Burnashev, 1998). The relationship between the gastropod 5-HT receptor and coupled ion channels (such as the L-type channel) remains unclear because the subtypes that mediate excita- tory responses in the gastropod heart are still being phar- macologically identified. However, a large body of evidence suggests that the 5-HT receptor from both gastropod (S.- Rozsa and Kiss, 1976: Kebabian et al.. 1979; Mandelbaum et al., 1979; S.-Rozsa, 1984; Sawada et al., 1984; Drum- mond et al., 1985; Huddart and Hill, 1996) and bivalve hearts (Higgins, 1974, 1977; Higgins and Greenberg, 1974; Paciotti and Higgins, 1985) is associated with the adenylate cyclase-cAMP signaling pathway, which when stimulated produces an intracellular rise in cAMP. Higgins and Green- berg ( 1974) found that cAMP increases the phosphorylation of SR proteins that mediate Ca2 + sequestration into micro- somes prepared from bivalve hearts. This then decreases the length of the diastolic phase of the cardiac cycle and primes the SR to release more Ca2+ ions during the next contrac- tion. cAMP was reported to stimulate Ca2 + release from intracellular stores in molluscan neurons as well (Kononecko et al., 1983). In contrast, FMRFamide does not appear to work through the adenylate cyclase-cAMP signal- ing pathway in gastropod ventricles (Drummond et al., 1985; Huddart and Hill, 1996) but instead may mediate phosphoinositide hydrolysis, the process that has been im- plicated in some bivalve molluscs (Bayakly and Deaton. 1992). Both 5-HT and some cAMP analogs stimulate 45Ca2+ efflux from Aplysia ventricular myocytes during enhanced chronotropic and inotropic activities (Sawada et al., 1984). In the present experiments, 5-HT also induces a large net Ca2+ efflux (on the order of 2.6 pmol cm~2 s~') from Busycon trabeculae that is recorded even when slow con- tractions of the trabecula are visible under the microscope. This net Ca2+ efflux is sustained over many hours. Sawada et al. (1984) also reported that excitatory drugs that stimu- late contraction and cause simultaneous 45Ca2+ efflux do not deplete cytoplasmic Ca2+ in the ventricular myocytes of Aplysia. These combined Ca24" efflux data from Busycon and Aplysia ventricles suggest a substantial internal Ca2+ reserve such as the well-developed SR in the gastropod cardiac muscle described by Sanger (1979). The 5-HT-induced Ca2+ efflux recorded from the Busy- con ventricle is a large, stable signal that immediately returns to control levels upon washing with seawater. In contrast, in bivalve ventricles (from mussel. Geukensia de- missa) that are excited by 5-HT. a transient 45Ca2+ efflux corresponds to the onset of a 5-HT-induced contracture but is not sustained throughout the contracture (Koch and Greenberg, 1981 ). When 5-HT is washed from the Geuken- sia ventricle, a second large 45Ca2+ efflux occurs, during which total tissue Ca2+ increases. Sawada et al. (1984) similarly noted that Ca2+ efflux does not necessarily reflect a reduction in cytoplasmic Ca2+, because internal reserves may be available. A second application of 5-HT to the mussel heart does not induce a subsequent 45Ca2 + efflux (Koch and Greenberg, 1981), whereas repeated efflux sig- nals are stimulated by 5-HT in the gastropod heart. These findings reflect the large difference in complexity between the gastropod SR and the bivalve SR as sustainable Ca2+ pools. The SERCA inhibitor, CPA. blocks 5-HT responses by 80% in the Busycon heart, suggesting that the SR is the major Ca2+ reservoir used during cardioexcitation by 5-HT. CPA acts by inhibiting the binding of Ca2+ ions at high- affinity binding sites on the SERCA; this shuts down both Ca2+ uptake and subsequent Ca2+ release at the SR of skeletal, cardiac (Balke et al., 1994), and smooth muscle (Suzuki et al., 1992). An indirect effect of inhibiting Ca2+sequesteration at the SR is a reduction in Ca2+-in- duced Ca2+ release; such a reduction is typically activated by the entry of extracellular Ca2+ through voltage-gated Ca2+ channels. In short, this would serve to limit the Ca2 + 5-HT-INDUCED CA2+ FLUX IN MOLLUSCAN VENTRICLE needed to activate the contractile proteins. On the other hand, a reduction in Ca2 + -induced Ca2 + release also de- creases the Ca2"1" -dependent K+ current (IK.Ca>- tnus um~ mutely prolonging the excitability of vertebrate smooth muscle (Suzuki et al.. 1992). How CPA is affecting excit- ability in the mollusc ventricle warrants further investiga- tion, especially considering its effectiveness in blocking Ca2+ efflux and the complete reversibility of its action. The net Ca2 + efflux that was recorded during 5-HT treatment was composed primarily of Ca2+ released from the SR and secondarily of Ca2+, from the extracellular fluid, that had previously entered the myocytes via an L-type current. This net Ca2+ efflux reflects the ability of the sarcolemma to rapidly redistribute Ca~+ ions outward, therefore preventing the toxic effects of Ca~+ overload, and reestablishing intracellular and extracellular Ca2+ gradients before the next successive depolarization and contraction. To determine if the net efflux was in part due to activity of the Na+-Ca2 + exchanger, a Na+-free, Li + -substituted sa- line (Li+ ASW) was used. The Li+ ASW reduces the basal net Ca2+ efflux during normal, autorhythmic activity of the Busycon trabeculae by only 20%-30%, whereas the 5-HT and FMRFamide responses are inhibited during Li+ treat- ment by 407c and 73%, respectively (Devlin, 1997). This difference in response suggests a greater reliance on the Na+-Ca2+ exchanger during chemically mediated excita- tion than during spontaneous myogenic activity. These Li + substitution experiments on the gastropod trabeculae also show that Na+ ions are a necessary stimulus for Ca~ + mobilization through voltage-gate channels or from an in- tracellular pool, perhaps during Na+-induced Ca~+ release (Lipp and Niggli, 1994), and that they are also involved in the process of Ca2+ extrusion. The data from the present study are in agreement with studies on the Na+-Ca~+ ex- changer in other myocytes where the removal of extracel- lular Na+ reduces Ca2+ efflux in guinea atrial cells (Reuter and Seitz, 1968) and in internally dialyzed myocytes (Miura and Kimura, 1989). Acknowledgments This work was supported by the Marine Biological Lab- oratory, M. G. F. Fuortes and Lucy B. Lemann Fellowships, the NASA Life Sciences Program, and a Pennsylvania State University Faculty Development Grant. Thanks are also extended to the staff of the N. I. H. BioCurrents Research Center at the MBL for their support in this research. Literature Cited Balke, C. \V., T. M Egan, and W. G. Wier. 1994. Processes that remove calcium from the cytoplasm during excitation-contraction cou- pling in intact rat heart cells. J. Physiol. 474: 447-462. Bayakly, N. A., and L. E. Deaton. 1992. The effects of FMRFamide, 5-hydroxytryptamine and phorbol esters on the heart of the mussel Geukensia dernissa. J. Comp. Physiol. B. 162: 463-468. Bloomquist, E., and B. A. Curtis. 1972. The action I calcium-45 efflux from the anterior byssal retractor muscle of /Wyimo edulis. J. Gen. Physiol. 59: 476-485. Bloomquist, E., and B. A. Curtis. 1975. 45Ca efflux from anterior byssus retractor muscle in phasic and catch contraction. Am. J. Physiol. 229: 1237-1243. Bucket!, K. J., G. J. Dockray, N. N. Osborne, and P. R. Benjamin. 1990. Pharmacology of the myogenic heart of the pond snail Lym- naea stagnalis. J. Neurophysiol. 63: 1413-1425. Burnashev, N. 1998. Calcium permeability of ligand-gated channels. Cell Calcium 24: 325-332. Devlin, C. L. 1993a. Acetylcholine-induced contractions in a holothu- rian (Isosticlmpus badionotux) smooth muscle are blocked by the calcium antagonists, diltiazem and verapamil. Comp. Biochem. Physiol. 106C: 573-577. Devlin, C. L. 1993b. An analysis of control of the ventricle of the mollusc Mercenaria mercenaria. II. Ionic mechanisms involved in excitation by 5-hydroxytryptamine. J. Exp. Biol. 179: 63-75. Devlin, C. L. 1997. A vibrating Ca2+-selective electrode measures Ca:+ flux induced by the neuropeptide FMRFamide in a gastropod ventricle. Comp. Biochem. Phvsiol. 116A: 93-100. Devlin, C. L., and P. J. S. Smith. 1996. A non-invasive vibrating calcium-selective electrode measures acetylcholine-induced calcium flux across the sarcolemma of a smooth muscle. J. Comp. Physiol. B 166: 270-277. Drummond, G. I., S. Wernham, and K. Lukowiak. 1985. Stimulation of adenylate cyclase in the heart of Aplysia californica by biogenic amines. Comp. Biochem. Physiol. C. 80: 129-133. Higgins, W. J. 1974. Intracellular action of 5-hydroxytryptamine on the bivalve myocardium. I. Adenylate and guanylate cyclases. J. Exp. Biol. 190: 99-110. Higgins, W. J. 1977. 5-hydroxytryptamine-induced tachyphylaxis of the molluscan heart and concomitant desensitization of adenylate cyclase. J. Cyclic Nudeotide Res. 3: 293-302. Higgins, W. J., and M. J. Greenberg. 1974. Intracellular action of 5-hydroxytryptamine on the bivalve myocardium. II. Cyclic nucleo- tide-dependent protein kinases and microsomal calcium uptake. J. Exp. Zool. 190: 305-316. Hill, R. B. 1958. The effects of certain neurohumors and of other drugs on the ventricle and radula protractor of Busycon canaliculatum and on the ventricle of Strombiis gigas. Biol. Bull. 115: 471-482. Hill, R. B. 1974. Effects of 5-hydroxytryptamine on action potentials and on contractile force in the ventricle of Dolabella auricularia. J. Exp. Biol. 61: 529-539. Hill, R. B., H. Huddart, and C. L. Devlin. 1992. Activation of a molluscan heart. Pp. 149-165 in Phylogenetic Models in Functional Coupling of the CNS and the Cardiovascular System, R. B. Hill and K. Kuwasawa, eds. S. Karger Medical and Scientific Publishers, Basel, Switzerland. Huddart, H., and R. B. Hill. 1996. Modulatory mechanisms in the isolated internally perfused ventricle of the whelk. Busycon canalicu- latum. Gen. Pharmacol. 21: 809- 818. Ishii, N., A. W. M. Simpson, and C. C. Ashley. 1989. Effects of 5-hydroxytryptamine (serotonin) and forskolin on intracellular free calcium in isolated and tura-2 loaded smooth-muscle cells from the anterior byssus retractor (catch) muscle of Mytilus edulis. Pflugers Arch. 414: 162-170. Kebabian. P. R., J. W. Kebabian, and D. O. Carpenter. 1979. Reg- ulation of cyclic AMP in heart and gill of Aplysia by the putative neurotransmitters dopamine and serotonin. Life Sci. 24: 1757-1764. Kiss, T., and K. S.-Rozsa. 1978. Pharmacological properties of 5-HT receptors of the Helix pomatia L. (Gastropoda) heart muscle cells. Comp. Biochem. Physio/. C. 61: 41-46. Koch, R. A., and M. J. Greenberg. 1981. Calcium fluxes accompanying 350 C. L. DEVLIN the action of 5-hydroxytryptamine on mussel hearts. Comp. Biochem. Physiol. C. 70: 229-239. Kononecko, N. I., P. G. Kostyuk, and A. D. Sherbatko. 1983. The effect of intracellular cAMP injections on stationary membrane con- ductance and voltage and time dependent ionic currents in identified snail neurons. Brain Res. 268: 321-338. Lee, K. S., and R. W. Tsien. 1983. Mechanism of calcium channel blockade by verapamil, D600, diltiazem and nitrendipine in single dialyzed heart cells. Nature 302: 790-794. Liebeswar. G., J. E. Goldman, J. Koester, and E. Mayeri. 1975. Neural control of the circulation in Aplysia. III. Neurotransmitters. J. Neurophysiol. 38: 767-779. Lipp, P., and E. Niggli. 1994. Sodium current-induced calcium signals in isolated guinea-pig ventricular myocytes. J. Physiol. 474: 439-446. Mandelbaum, D. E., J. Koester, M. Schonberg, and K. R. Weiss. 1979. Cyclic AMP mediation of the excitatory effect of serotonin in the heart ofAptysia. Brain Res. 177: 388-394. Miura, Y., and J. Kimura. 1989. Dependence on internal Ca and Na and competitive binding of external Na and Ca. J. Gen. Physiol. 93: 1129-1145. Nanasi, P. P., A. Varro, D. A. Lathrop, and M. Danko. 1990. Use dependent action of antiarrhythmic drugs in frog skeletal muscle and canine cardiac Purkinje fiber. Gen. Pharmacol. 21: 747-751. Paciotti, G. F., and W. J. Higgins. 1985. Potentiation of the 5-hydroxy- tryptamine-induced increases in myocardial contractility in Mercenaria mercenaria ventricle by forskolin. Comp. Biochem. Physiol. C 80: 325-329. Pellmar, T. C. 1984. Enhancement of inward current by serotonin in neurons of Aplysia. J. Neurophysiol. 15: 13-25. Reuter, H., and N. Seitz. 1968. The dependence of calcium efflux from cardiac muscle on temperature and external ion composition. J. Phvsiol. 195: 451-470. S.-Rozsa, K. 1984. The pharmacology of molluscan neurons. Prog. Neurobiol. 23: 79-150. S.-Rozsa, K., and T. Kiss. 1976. Role of cyclic nucleotides in the effect of transmitters on the heart of Helix pomatia L. Comp. Biochem. Physiol. C 53: 13-16. Sanger, J. W. 1979. Cardiac fine structure in selected arthropods and molluscs. Am. Zool. 19: 9-27. Sawada, M., M. Ichinose, I. Ito, T. Maeno, and D. J. McAdoo. 1984. Effects of 5-hydroxytryptamine on membrane potential, contractility, accumulation of cAMP, and Ca2+ movements in anterior aorta and ventricle of Aplysia. J. Neurophysiol. 51: 361-374. Smith, P. J. S., K. Hammar, D. M. Porterfield, R. H. Sanger, and J. R. Trimarchi. 1999. A self-referencing, non-invasive, ion selective electrode for single cell detection of trans-plasma membrane calcium flux. Microsc. Res. Tech. 46: 398-417. Suzuki, M., K. Muraki, Y. Imaizumi, and M. Watanabe. 1992. Cy- clopiazonic acid, an inhibitor of the sarcoplasmic reticulum Ca2+- pump, reduces Ca2 + -dependent K+ currents in guinea-pig smooth muscle cells. Br. J. Pharmacol. 107: 134-140. Vaghy, P. L., K. Itagaki, K. Miwa, E. McKenna, and A. Schwartz. 1988. Mechamsmof action of calcium modulator drugs. Pp. 176-186 in Calcium Antagonists: Pharmacology and Clinical Research, Vol. 522. P. M. Vanhoutte. R. Paoletti. and S. Govoni, eds. New York Academy of Sciences, New York. Yeoman, M. S., B. L. Brezden, and P. R. Benjamin. 1999. LVA and HVA Ca"+ currents in ventricular muscle cells of the Lymnaea heart. J. Neurophysiol. 82: 2428-2440. Zahradnik, I., and J. Zachar. 1983. Inhibitory effect of verapamil upon calcium and potassium currents in crayfish muscle membrane. Gen. Physiol. Biophys. 2: 181-192. INDEX ACHITUV. Y., see O. Levy, 118 ADAMCZEWSKA, AGNIESZKA M.. AND STEPHEN MORRIS. Ecology and be- havior of Gecarcoidea natalis, the Christmas Island red crab, during the annual breeding migration. 305 ADAMCZEWSKA. AGNIESZKA M., AND STEPHEN MORRIS, Metabolic status and respiratory physiology of Gecarcoidea natalis, the Christmas Island red crab, during the annual breeding migration, 321 The adaptive bleaching hypothesis: experimental tests of critical assump- tions. 51 Aerobic. 59 Aiptasia pallida, 127 Alviniconcha, 298 AMANO, SHIGETOYO, AND ISAO HORI, Metamorphosis of coeloblastula per- formed by multipotential larval flagellated cells in the calcareous sponge Leucosolenia laxa. 20 Anaerobic. 59 Analog VLSI. 243 Anemotaxis, 222 Annelida. 107 Antennae. 1 Artificial neural networks. 195 Asexual reproduction, 107 Audition. 227 Aurelia labiata, 92 limbala. 92 Auditory micromechanics. 190 Axon, 252 B Balanus amphitrite, 87 BARLOW, ROBERT B.. JAMES M. HITT. AND FREDERICK A. DODGE, Limulus vision in the marine environment. 169 Barnacle. 87 BARTOL. IAN K.. Role of aerobic and anaerobic circular mantle muscle fibers in swimming squid: electromyography, 59 BEER, AMY-JANE. CLAIRE Moss, AND MICHAEL THORNDYKE, Development of serotonin-like and SALMFamide-like immunoreactivity in the ner- vous system of the sea urchin Psammechinus miliaris, 268 Behavior, 118, 160 mating. 169 Biological sensor, 190 A biologically inspired controller for hexapod walking: simple solutions by exploiting physical properties, 195 Biomechanics, 1 Biomimetic, 227, 243 robotics. 160 BIRMINGHAM. J. T.. Increasing sensor flexibility through neuromodulation. 206 Bivalve pump, 77 BLAZIS. DIANA E. J.. AND FRANK W. GRASSO. Introduction. Invertebrate Sensory Information Processing Workshop, 147 Body plan, 257 BOWMAN. JOHN A. L.. see Bernard B. Rees, 247 Brain function. 160 BREITHAUPT, THOMAS. Fan organs of crayfish enhance chemical informa- tion flow. 150 BROWNE. KENNETH A.. AND RICHARD K. ZIMMER, Controlled field release of a waterborne chemical signal stimulates planktonic larvae to settle, 87 BUCKLAND-NICKS, JOHN, see Ian G. Paterson, 261 Budding. 107 Ca2* flux, 344 Calcification, 257 Carausius morosus, 195 Cardiac muscle. 344 Cephalopod, Antarctic. 67 CHADWICK-FURMAN. N. E.. see 0. Levy. 1 18 CHAHL, JAVAAN S., see Mandyam V. Srinivasan. 216 Chemical communication, 150 cue, 87 orientation, 150 senses, 160. 211 sensor, 222 Chemoreception, 2 1 1 Chemotaxis. 222 Christmas Island, 305, 321 Circadian rhythm, 169 Circular muscle, 59 Clam, giant. 336 Coeloblastula, 20 COFFROTH, MARY ALICE, see Robert A. Kinzie III, 51 COL 92 Cold adaptation, 67 Communication, chemical. 150 Compound eye. 177 Computer simulation. 195 Constraint. 257 Controlled field release of a waterborne chemical signal stimulates plank- tonic larvae to settle. 87 Coral. 118 bleaching. 51 Crab, land, 305, 321 Crayfish, 150, 201 Cricket. 184 Crinoids, 1 CRONIN, THOMAS W., AND JUSTIN MARSHALL, Parallel processing and image analysis in the eyes of mantis shrimps, 177 CRUSE. HOLK. see Josef Schmitz, 195 Crustacean, 211, 305, 321 Cryptic female choice. 261 Cytochrome oxidase c subunit I, 92 D DAWSON, MICHAEL N., AND DAVID K. JACOBS. Molecular evidence for cryptic species of Aurelia aitrita (Cnidaria, Scyphozoa), 92 Dean. Jeffrey, see Josef Schmitz. 195 Decapod, 281 Dendrite, 252 DERBY. CHARLES D.. AND PASCAL STEULLET, Why do animals have so many receptors? The role of multiple chemosensors in animal perception. 211 Development, 257, 268 Development of host- and symbiont-specitic monoclonal antibodies and confirmation of the origin of the symbiosome membrane in a cnidar- ian-dinoflagellate symbiosis, 127 351 352 INDEX TO VOLUME 200 Development of serotonin-like and SALMFamide-like immunoreactivity in the nervous system of the sea urchin Psammechinus miliaris. 268 DEVLIN. C. LEAH, 5-Hydroxytryptamme stimulates net Ca2+ flux in the ventricular muscle of a mollusc (Busycon canaliculatum) during car- dioexcitation, 344 DODGE, FREDERICK A., see Robert B. Barlow, 169 E Echinoderm. 33, 268 Ecology, 305 Ecology and behavior of Gecarcoidea natalis, the Christmas Island red crab, during the annual breeding migration, 305 Egg size, 33 Electromyography, 59 Energy content, 33 store, 321 Equilibrium pathways in Crustacea, 155 Escape hatches for the clonal offspring of serpulid polychaetes, 107 Estuary, 87 ETNIER, SHELLEY, Flexural and torsional stiffness in multi-jointed biolog- ical beams, 1 Evolution of development, 257 Evolutionary novelty, 257 Exercise, 305, 321 Eye, compound. 177 Factors controlling the expansion behavior of Faviafavus (Cnidaria: Seler- actinia): effects of light, flow, and planktonic prey, 1 18 Fan organs of crayfish enhance chemical information flow, 150 FARMER, MARK A., WILLIAM K. FITT, AND ROBERT K. TRENCH. Morphology of the symbiosis between Corciilum cardissa (Mollusca:Bivalvia) and Symbiodinium corciilorum (Dinophyceae). 336 Favia, 118 FIJIWARA, Y., see S. Kojima , 298 Finlets, 9 FITT, WILLIAM K., see Mark A. Farmer, 336 Flagellated cell, 20 Flexibility. 1 Flexural and torsional stiffness in multi-jointed biological beams, 1 Flow field, 150 Fluid mechanics, 160 FRASER, PETER J., Statocysts in crabs: short-term control of locomotion and long-term monitoring of hydrostatic pressure. 155 FUIIKURA, K., see S. Kojima. 298 Function and functional groupings of the complex mouth apparatus of the squat lobsters Munida sarsi Huus and M. lenuimana G. O. Sars (Crustacea: Decapoda). 281 Functional morphology, 281 G GARM, A., AND J. T. HOEG. Function and functional groupings of the complex mouth apparatus of the squat lobsters Munida sarsi Huus and M. tenuimana G. O. Sars (Crustacea: Decapoda). 281 Gastropod. 257 Gecarcoidea, 305. 321 Gene expression, 247 GFP, 252 Giant clam. 336 Gill geometry. 77 muscle, 77 GRASSO, FRANK W., Invertebrate-inspired sensory-motor systems and au- tonomous, olfactory-guided exploration, 160 GRASSO, FRANK W., see Diana E. J. Blazis, 147 Green fluorescent protein, 252 H HADFIELD. MICHAEL G., see Carole S. Hickman, 257 HASHIMOTO, J., see S. Kojima, 298 Hearing. 190 Hemocyanin, cephalopod, 67 HICKMAN, CAROLE S., AND MICHAEL G. HADFIELD, Larval muscle contrac- tion fails to produce torsion in a trochoidean gastropod, 257 HIF-1. 247 HIGGINS. CHARLES M., Sensory architectures for biologically inspired au- tonomous robotics, 243 HITT. JAMES M., see Robert B. Barlow, 169 HOEG, J. T, see A. Garm, 281 Honeybee. 216 HORI, ISAO, see Shigetoyo Amano, 20 HUBBARD, ALLYN E., see David C. Mountain, 227 Hybridization, 97 Hybrids of two closely related tropical sea urchins (Genus Echinometra): evidence against postzygotic isolating mechanisms. 97 Hydrostatic pressure sensing, 155 Hydrothermal vent, 298 5-Hydroxytryptamine, 344 5-Hydroxytryptamine stimulates net Ca: + flux in the ventricular muscle of a mollusc (Biisycon canaliculatum) during cardioexcitation, 344 Hypoxia, 247 Hypoxia-inducible factor, 247 I Immunocytochemistry. 127 Increasing sensor flexibility through neuromodulation, 206 Innovative biomechanics for directional hearing in small flies, 190 Insect flight, 216 Insights for robotic design from studies of the control of abdominal position in crayfish. 201 Internal transcribed spacer one. 92 Interspecific relationships between egg size and the level of parental investment per offspring in echmoderms, 33 Invertebrate sensory information processing: implications for biologically inspired autonomous systems, a workshop, 145 Invertebrate-inspired sensory-motor systems and autonomous, olfactory- guided exploration. 160 ISHIDA. H., T. NAKAMOTO. T. MORIIZLIMI. T. KIKAS, AND J. JANATA, Plume- tracking robots: a new application of chemical sensors, 222 ISIP-BIAS Symposium. 145 ITS-1, 92 JACOBS. DAVID K.. see Michael N. Dawson, 92 JANATA, J., see H. Ishida. 222 Jellyfish, 92 K KATSUYAMA, Y., see T. Okada. 252 KEMPF. STEPHEN C., see Timothy S. Wakefield. 127 KIKAS, T., see H, Ishida, 222 Killifish, 247 KINDERMANN. THOMAS, see Josef Schmitz, 195 Kinematics, 9 KINZIE, ROBERT A., Ill, MICHELLE TAKAYAMA, SCOTT R. SANTOS, AND MARY ALICE COFFROTH, The adaptive bleaching hypothesis: experi- mental tests of critical assumptions, 51 KOJIMA, S., R. SEGAWA. Y. FIJIWARA, K. FUJIKURA, S. OHTA. AND J. HASHIMOTO. Phylogeny of hydrothermal-vent-endemic gastropods. Alviniconcha spp., from the Western Pacific revealed by mitochon- drial DNA sequences, 298 INDEX TO VOLUME 200 353 Lactate dehydrogenase. 247 Land crab. 305, 321 Landing, 216 Landing strategies in honeybees and possible applications to autonomous airborne vehicles, 216 Larva. 20, 87, 257 Larval collector, 87 Larval muscle contraclion fails to produce torsion in a trochoidean gastro- pod. 257 LAUDER, GEORGE V.. see Jennifer C. Nauen. 9 LEVY, O.. L. MIZRAHI, N. E. CHADWICK-FURMAN. AND Y. ACHITUV, Factors controlling the expansion behavior of Favia favus (Cnidaria: Sclerac- tinia): effects of light, flow, and planktonic prey. 1 18 Life history, 33 Limulus, 169 Limn/its vision in the marine environment. 169 Load compensation feedback control loops, 201 Lobster, 2 1 1 Locomotion. 9. 195 central control of. 155 Lolliguncula brevis, 59 Muscle circular, 59 gill. 77 receptor organ, 201 smooth, 77 Muscular alteration of gill geometry in vitro: implications for bivalve pumping processes, 77 N NAKAMOTO. T., see H. Ishida. 222 NAUEN, JENNIFER C.. AND GEORGE V. LAUDER. Three-dimensional analysis of finlet kinematics in the chub mackerel (Scomber japonicus), 9 Navigation, 216 Nervous system, 160 stomatogastric, 206 Neural code, 169 Neurobiology. 268 Neuromodulation. 206 Neuromorphic. 243 Neuronal form in the central nervous system of the tadpole larva of the ascidian dona intestinalis, 252 M MAClsAAC. S. STANLEY, see T. Okada. 252 MACMILLAN, DAVID L.. AND BLAIR W. PATULLO. Insights for robotic design from studies of the control of abdominal position in crayfish, 201 Macro-video, 281 MARSHALL, JUSTIN, see Thomas W. Cronin, 1 77 Mating behavior. 169 McEowARD, L. R.. AND K. H. MORGAN, Interspecific relationships between egg size and the level of parental investment per offspring in echino- derms. 33 MEDLER. SCOTT, AND HAROLD SILVERMAN, Muscular alteration of gill geometry in vitro: implications for bivalve pumping processes. 77 MEINERTZHAGEN, I. A., see T. Okada, 252 Metabolic status and respiratory physiology of Gecarcoidea natalis, the Christmas Island red crab, during the annual breeding migration, 321 Metamorphosis, 20, 87 Metamorphosis of coeloblastula performed by multipotential larval flagel- lated cells in the calcareous sponge Leucosolenia laxa. 20 Microphone technology. 190 Microsatellite DNA, 261 Migration. 305. 32 1 Mitochondria! DNA. 298 MIZRAHI, L., see O. Levy, 1 18 Modeling. 195 Molecular evidence for cryptic species of Aurelia aurita (Cnidaria. Scy- phozoal. 92 Molecular phylogeny, 298 Mollusca. 261. 336, 344 MORGAN, K. H., see L. R. McEdward. 33 MORIIZUMI, T., see H. Ishida. 222 Morphogenesis, 257 Morphology, functional, 281 Morphology of the symbiosis between Corculum cardissa (Mollusca: Bivalvia) and Symbiodinium corculorum (Dinophyceae), 336 MORRIS. STEPHEN, see Agnieszka M. Adamczewska, 305. 321 Moss. CLAJRE, see Amy-Jane Beer, 268 Motor control, 195 neuron. 252 MOUNTAIN. DAVID C., AND ALLYN E. HUBBARD. Sensing scenes with silicon. 227 Mouthparts. 281 Multiple paternity in Littorina ohtusata (Gastropoda. Littorinidae) revealed by microsatellite analyses. 261 Munida sarsi, 281 tenuimana. 281 O OHTA. S., see S. Kojima, 298 OKADA. T., S. STANLEY MAC!SAAC, Y. KATSUYAMA. Y. OKAMURA. AND I. A. MEINERTZHAGEN. Neuronal form in the central nervous system of the tadpole larva of the ascidian Ciona intestinalis. 252 Okamura, Y., see T. Okada. 252 Olfaction, 150. 160, 211. 227 Optic flow. 216 Optomotor. 184 Orientation, chemical, 150 Oxygen. 247 transport, 67 Panulirns. 2 1 1 Parallel processing and image analysis in the eyes of mantis shrimps. 177 PARTRIDGE. VANESSA, see Ian G. Paterson. 261 PATERSON. IAN G.. VANESSA PARTRIDGE. AND JOHN BUCKLAND-NICKS. Multiple paternity in Littorina obtusata (Gastropoda. Littorinidae) revealed by microsatellite analyses, 261 PATULLO, BLAIR W., see David L. Macmillan. 201 PEARSE, JOHN S.. see M. Aminur Rahman, 97 Peptide, 87 Perception, 2 1 1 FERNET, BRUNO, Escape hatches for the clonal offspring of serpulid polychaetes, 107 Phonotaxis, 184 Phylogeny of hydrothermal-vent-endemic gastropods, Alviniconcha spp., from the Western Pacific revealed by mitochondrial DNA sequences, 298 Physiology. 321 Plume-tracking robots: a new application of chemical sensors. 222 Polyandry. 261 Polychaeta. 107 PORTNER. H. O.. see S. Ziehnski. 67 Position control. 201 Procambarus c/arkii. 150 R RAHMAN, M. AMINUR. TSUYOSHI UEHARA. AND JOHN S. PEARSE. Hybrids of two closely related tropical sea urchins (Genus Echinometrd): evi- dence against postzygotic isolating mechanisms, 97 Redundancy. 21 1 354 INDEX TO VOLUME 200 REES. BERNARD B.. JOHN A. L. BOWMAN. AND PATRICIA M. SCHLILTE, Structure and sequence conservation of a putative hypoxia response element in the lactate dehydrogenase-B gene of Fitndit/its, 247 Representation. 184 Reproduction, asexual. 107 ROBERT. DANIEL, Innovative biomechanics for directional hearing in small flies, 190 Robot, 184, 206. 222 Robotics, 160, 216, 243 Role of aerobic and anaerobic circular mantle muscle fibers in swimming squid: electromyography, 59 SANTOS, SCOTT R., see Robert A. Kinzie III. 51 SARTORIS, F. J., see S. Zielinski. 67 Scene analysis, 227 SCHMITZ, JOSEF. JEFFREY DEAN. THOMAS KINDERMANN. MICHAEL SCHUMM, AND HOLK CRUSE, A biologically inspired controller for hexapod walking: simple solutions by exploiting physical properties, 195 SCHLILTE, PATRJCIA M.. see Bernard B. Rees, 247 SCHUMM. MICHAEL, see Josef Schmitz. 195 Scombrid fishes, 9 SEGAWA, R.. see S. Kojima, 298 Senses, chemical. 21 1 Sensing scenes with silicon, 227 Sensor, 206 artificial. 177 chemical. 222 Sensory. 211 Sensory architectures for biologically inspired autonomous robotics, 243 Sensory-motor feedback, 201 Serial EM. 252 Setae, 281 Settlement, 87 Sibling species. 92 Silicon. 227 SILVERMAN, HAROLD, see Scott Medler. 77 Smooth muscle. 77 Sperm competition. 261 storage. 261 Sponge, calcareous, 20 Squid. 59 SRINIVASAN. MANDYAM V., SHAOWU ZHANG, AND JAVAAN S. CHAHL. Land- ing strategies in honeybees and possible applications to autonomous airborne vehicles, 216 Statocysts in crabs: short-term control of locomotion and long-term mon- itoring of hydrostatic pressure, 155 STEULLET, PASCAL, see Charles D. Derby, 21 I Stick insect. 195 Stiffness flexural. 1 torsional. 1 Stomatogastric nervous system. 206 Stomatopod crustacean. 177 Stretch receptor. 206 Structure and sequence conservation of a putative hypoxia response ele- ment in the lactate dehydrogenase-B gene of Fundulus, 247 Symbiodinium bermudense. 127 corculorum, 336 Symbiosis, 127. 336 TAKAYAMA. MICHELLE, see Robert A. Kinzie III. 51 Temperature. 67 Temperature effects on hemocyanin oxygen binding in an Antarctic cepha- lopod, 67 THORNDYKE. MICHAEL, see Amy-Jane Beer. 268 Three-dimensional analysis of finlet kinematics in the chub mackerel (Scomber japonicus), 9 Torsion. 257 TRENCH. ROBERT K., see Mark A. Farmer, 336 Tube growth. 107 Turbulence, 160 u UEHARA. TSUYOSHI. see M. Aminur Rahman, 97 Ultrastructure. 336 Urchin. 97 Urochordate, 252 Vector coding in the nervous system, 155 Veliger. 257 View from the boundary. 184 Vision. 169. 216. 227 chip, 243 color, 177 Visual ecology. 177 VLSI. 227. 243 w WAKEFIELD, TIMOTHY S., AND STEPHEN C. KEMPF, Development of host- and symbiont-specific monoclonal antibodies and confirmation of the origin of the symbiosome membrane in a cnidarian-dinoflagellate symbiosis, 127 Water pumping. 77 WEBB. BARBARA, View from the boundary, 184 Western Pacific, 298 Why do animals have so many receptors? The role of multiple chemosen- sors in animal perception. 211 ZHANG. SHAOWU, see Mandyam V. Srinivasan. 216 ZIELINSKI. S.. F. J. SARTORIS. AND H. O. PORTNER, Temperature effects on hemocyanin oxygen binding in an Antarctic cephalopod. 67 ZIMMER. RICHARD K.. see Kenneth A. Browne, 87 Zooxanthellae. 51. 118, 336 The Ultimate Deconvolution Machine. Original Widefield lii When image quality is essential and capturing lost information is a must, Carl Zeiss 3D Deconvolution system is the answer. Our fast, intuitive software allows reassignment of non-focal haze, noise and blur to provide image restoration to its brilliant best. Choose between theoretically calculated or measured PSF, iterative or non-iterative method, and the Zeiss 3D Deconvolution will do the rest. • , ! If - ;--^y. / Iterative DCV (measured PSF): Ma.\imum Likelihood: 15 CG Iterations: Regularised. Whether conventional widefield or confocal fluorescence imaging, your own 3D deconvolution system from Zeiss takes you to new heights of image fidelity. Let us show you how! For immediate demonstration, call 800-233-2343. Carl Zeiss, Inc. Microscopy & Imaging Systems One Zeiss Drive Thornwood, NY 10594 800.233.2343 Fax 914.681.7446 micro@zeiss.com www.zeiss.com/micro ZEISS MBL WHOI LIBRARY UH IBBfl T