Environmental Biology of Fishes 41: 301-309, 1994. ? 1994 Kluwer Academic Publishers. Printed in the Netherlands. Does gonad structure reflect sexual pattern in all gobiid fishes? Kathleen S. Cole\ D. Ross Robertson2 & Alcibiades A. Ceden02 I Department of Biology, Bishops University, Lennoxville, Quebec JIM lZ7, Canada 2 Smithsonian Tropical Research Institution (Panama), Unit 0948, APO AA 34002-0948, U.S.A. Received 9.2.1993 Accepted 1.3.1994 Key words: Ontogeny, Protogyny, Gobiosoma, Gobiidae Synopsis In immature and adult females of protogynous gobies, small distinctive masses of cells associated with the ovarian wall develop into testis-associated glandular structures during sex change. These precursive accessory gonadal structures, or pAGS, have been found in females of known protogynous goby species, but not among gonochoric goby species, suggesting that their presence can be used as a species-specific indicator of prot? ogyny within the family. However, a detailed examination of a developmental series of ovaries in two go? nochoric species, Gobiosoma illecebrosum and G. saucrum, revealed the presence of a gonadal feature previ? ously thought to be restricted to protogynous gobies. Among immature females of both species, pAGS-like structures having a similar appearance and placement as functional pAGS of protogynous gobies were found. In female G. illecebrosum, the size of these structures among immatures progressively decreased with matura? tion and were absent in all but the smallest adult females. A similar pattern was evident in a small sample of G. saucrum. Population demography based on field collections showed that G. illecebrosum exhibits sex ratios and male and female size-frequency distributions typical of gonochores and laboratory experiments indicated that final sexual identity was unaffected by social environment during the juvenile period. Thus, the presence of pAGS in juvenile female G. illecebrosum is not related to an ability to change sex at that ontogenic interval. Whether the transient pAGS observed here are vestiges of an ancestral protogynous condition is unknown. Based on their presence among immatures in two gonochore gobies, however, only the presence of pAGS in adult females should be used to predict protogyny among gobies. Introduction As far as is known, males of all fishes in the sub? order Gobioidei share unique glandular structures typically associated with the sperm duct of the tes? tis. These have variously been referred to as semi? nal vesicles (Egami 1960, Arai 1964), sperm-duct glands (Miller 1984) and, in some protogynous go? bies, where they derive from cell masses located in the ovarian wall, accessory gonadal structures (AGS - Cole & Robertson 1988). The pre cursive cell masses (pAGS; illustrated in Fig.1a) which de- velop into AGS during sex change have been found in all females examined (both immature and ma? ture) among 11 experimentally confirmed proto? gynous goby species within five genera (Cole 1983, 1988, 1990, Cole & Robertson 1988, Cole & Shapiro 1990, Cole unpublished data) but not among a small number of similarly examined females of four go? nochore goby species (Cole 1988). Consequently, Cole (1988) postulated that among gobies the pres? ence of pAGS associated with the ovary should be considered a reliable indicator of protogyny. Hermaphroditism has now been reported for 13 302 gobiid genera, including Gobiodon and Paragobio? don (Lassig 1977), Coryphopterus and Gobiosoma (Robertson & Justines 1982), Lythrypnus (Cole 1988), Pleurosicya, Bryaninops, Luposicya (Fishel? son 1989) and Lophogobius, Fusigobius, Eviota, Trimma and Priolepis (Cole 1990). In seven genera in which more than one species has been examined (Gobiodon, Paragobiodon, Lophogobius, Lythryp? nus, Fusigobius, Trimma, Priolepis and Coryphop? terus), hermaphroditism has been found in all spe? cies (n = 33) examined to date (Cole unpublished data, Cole & Hoese unpublished data). This in? cludes Coryphopterus, in which hermaphroditism has been found in all 10 of the 11 described species that have been examined (one rare species remains unexamined). Universal protogyny among examin? ed congeners within the above genera suggests that hermaphroditism in gobiids may be a shared trait among closely related species and, possibly, an an? cestral condition (Cole & Shapiro 1990). One gobiid genus, however, does not follow this pattern. In Gobiosoma, protogyny has been report? ed for one species, G. multifasciatum (Robertson & Justines 1982) but is absent in three others, G. sau? crum, G. illecebrosum (Robertson & Justines 1982) and G. evelynae (Cole 1988). If hermaphroditism is ancestral in Gobiosoma, species such as G. sau? crum, illecebrosum and evelynae which demon? strate no functional hermaphroditism may never? theless reveal some features of gonad structure typ? ically restricted to protogynous species. Given that the only consistent gonadal feature of protogyny in gobiids is the presence of pAGS associated with the ovary of females prior to sex change (Cole & Ro? bertson 1988, Cole & Shapiro 1990), it seems likely that any structural anomalies in gonochore species that are closely related to hermaphroditic species will be found associated with the ovary. Previous work (Robertson & Justines 1982) indi? cated that G. illecebrosum is a gonochore with a 1:1 sex ratio and no sex-change potential among adult females. However, with the occurrence ofprotogy? ny in a closely-related species and the proposition that hermaphroditism may be an ancestral condi? tion in Gobiosoma, the gonochore designation for G. illecebrosum deserves closer examination. Previ? ous experiments that tested for sex change ability in gonochoric species of Gobiosoma were performed only with adult females. Since in some other proto? gynous fishes (i.e. Scaridae - Robertson & Warner 1978) females appear capable of transforming the ovary to a testis prior to maturity, it is possible that immature females of some 'gonochore' gobies could have similar precocious sexual lability. If so, one might expect such lability in apparently gono? choric species that have protogynous congeners. This paper describes an investigation of several aspects of ovarian morphology and sexual develop? ment in G. illecebrosum and, to a lesser extent, G. saucrum and G. multifasciatum. The first part ex? amines the histostructure of ovaries of immature and mature females of these three species for the possible presence in G. illecebrosum and G. sau? crum of structural features typically associated with ovaries of protogynous goby species, including G. multifasciatum. The second part presents informa? tion on population sexual demography based on field collections and results of a series of laboratory experiments with G. illecebrosum that were de? signed to test for: (a) sex change potential in adult and subadult females and adult males, and (b) labil? ity in the development of sexual identity by undif? ferentiated immature fish. We predicted that if G. illecebrosum is a strict gonochore, both immatures and adults from field collections and laboratory ex? periments would exhibit a 1:1 primary sex ratio, irre? spective of the natural or experimental social envi? ronment. Methods and materials Field collections A large field collection of G. illecebrosum was car? ried out in August 1993 to obtain baseline data on sex-ratios and size-frequency distributions. Individ? ual G. illecebrosum were collected from small patch reefs in the vicinity of the Smithsonian Tropical Re? search Institute's (S.T.R.I.) field station in the San BIas Islands, Panama (Lat. 9 34' N, Long. 78 58"W) using a fish anesthetic quinaldine sulfate and a dip net. Gobiosoma illecebrosum live in small groups, or aggregations, on isolated coral heads. As the so- cial structure and mating systems of these groups and whether, in fact, they comprise cohesive social groups, is unknown, all individuals collected from a single coral head henceforth will simply be referred to as an aggregation. Collected individuals were killed by over-anesthetization immediately after collection, preserved in Dietrich's fixative, then ex? amined with a dissecting microscope to establish sex based on genital papilla structure. In this spe? cies, the male genital papilla is elongate with a pointed terminus, has a small genital pore at the apex, and often has melanocytes scattered along the length; the female genital papilla is shorter, square to rectangular in outline with a blunt terminus, has a broad genital pore at the apex and exhibits few, or no, melanocytes (Cole & Robertson unpublished data). Individuals were classified as male or female accordingly, then measured (mm, standard length). The resulting data were then combined with infor? mation taken from similarly collected individuals previously collected from the same locale in 1980 (Robertson & Justines 1982) to provide information on sex ratio and size-frequency distributions for males and females. Histological examination In separate collections from those described above, individual G. illecebrosum were collected from small patch reefs in the vicinity of the S.T.R.1. San Bias field station with quinaldine sulfate. In the first collection, made in October 1988, all fish (n = 93) were killed immediately after collection, preserved in Dietrich's fixative, decalcified in Fisher's Cal-Ex solution and embedded in toto in paraplast. The posterior portion of the body was then serially sec? tioned at 10 /lm and all sections were mounted, stained with Harris' haematoxylin and eosin and viewed with a light microscope. This series was used to examine gonad structure in sexually undifferen? tiated fish, immatures, adult females and adult males. Some additional immature fish (n = 28) ob? tained from subsequent collections were treated similarly and added to the histological sample for a total of 121 histologically examined specimens. Gobiosoma multifasciatum and G. saucrum of a 303 range of sizes including both immatures and adults (see Results) were collected at the same site and treated in the same manner as the G. illecebrosum prepared for histological examination. Rearing experiments In separate collections from the above, additional G. illecebrosum, including adults, sexually distinct immatures and smaller, sexually indistinct imma? tures, were collected between October 1988 and June 1989, transported in insulated containers to S.T.R.I.'s Naos Marine Laboratory in Panama City and within 24 h of initial capture were placed in aquaria for the different experimental treatments. In each experimental replicate one or more fish, depending on the treatment (see below) were main? tained in a visually isolated 50 I aquarium provided with rocks and coral skeletons for shelter, flow? through sea water, aeration and a natural photope? riod. A superabundance of freshly hatched brine shrimp nauplii was added to each aquarium daily as food for the test fish. Each experiment ran between 1-2 months until all immatures had surpassed the minimum size at which maturity occurs in the wild (21 mm SL, see Results). To establish experimental groups, individuals were first divided into juveniles and adults accord? ing to size (based on size of first gamete production according to histological information described above and presented in Results). The sex of individ? uals having sexually differentiated papillae were further identified as male or female, as described above. Most individuals up to 12 mm standard length (SL) had sexually undifferentiated papillae and could not be sexed externally. Five series of experiments were run: 1. Groups of adult or subadult females: To sub? stantiate previous reports of an absence of protogy? ny in this species, we established three experimen? tal groups each comprising three adult-sized fe? males and six groups each comprising three imma? ture-sized fish that had female-shaped genital papillae, for 3 weeks. 2. Group of immature males: Up to now, protogy? ny is the only known pattern of sequential sex 304 change among gobies but, as a preliminary test of sexual lability among adult fish with male papillae, we kept one group of five immature males together for two months. 3. Small immatures in varying social environ? ments: In these experiments, test fish were all less than 10 mm standard length (i.e. individuals of a size that typically do not have sexually distinct germ cells within the gonad, as described in Results) and had undifferentiated genital papillae. (a) Solitary immatures: 27 randomly chosen fish were reared singly. (b) Pairs of immatures: 44 randomly constructed pairs were reared to maturity. (c) Single immature and single adult: single, ran? domly chosen immatures were reared either with an adult male (n = 16) or an adult female (n = 18). Each of these experiments ran for 5-6 weeks. Results Population demography ofG. illecebrosum Thirteen aggregations collected in 1980 by Robert? son & Justines (1982) and an additional 21 aggrega? tions collected in August 1993 were examined for adult (i.e. ~ 21 mm SL) sex ratios. The sex ratios for 1980 (18 female, 20 male) did not differ significantly from that of 1993 (41 female, 34 male) (Yates ad? justed X2 = 0.29, P = 0.59, Sokal & Rohlf 1981). In the combined sample of 113 adults, the adult sex ra? tio (59 female, 54 male) did not differ significantly from 1:1 (G = 0.105, P = 0.75, G-test for goodness of fit, Sokal & Rohlf 1981), and the two sexes did not differ significantly from one another in terms of size-frequency distributions (Kolmogorov-Smir? nov two sample test, p = 0.77). Similarly, female:male sex ratios among juveniles measuring 17-20 mm SL collected in 1993 (35 fe? male, 25 male) did not differ significantly from those collected in the 1980 sample (17 female, 12 male; X2 = 0.15, P = 0.70) and combined, did not dif? fer significantly from a 1:1 sex ratio (G = 1.67, P = 0.20). In terms of sexual composition of aggregations, seven aggregations collected in August 1993 and seven more collected in 1980 (Robertson & Justines 1982) included four or more adults (Le. individuals ~ 21 mm SL). In the former, 22 were male and 22 were female while in the latter 16 were male and 22 were female. Gonad structure of field-collected immatures and adults of both sexes ofG. illecebrosum Of 93 fish collected in October 1988, 28 exhibited a short, blunt papilla typical of female gobies. Thirty? three other fish had an elongate, pointed papilla characteristic of males in other goby species (Miller 1984). In all cases, histological examination of the gonad verified these sexual designations. This sex ratio was not significantly different from 1:1 (Yates corrected X2 = 0.072, P = 0.79). The remaining 32 fish had a small, immature papilla having no sexual? ly distinct features. Of these 32 immature fish, plus 28 other individu? als with undifferentiated papillae obtained from subsequent collections which were all examined histologically, eight ranging in size from 9-11 mm SL had clearly differentiated gonadal lobes but germ cell identity was indistinct and sex could not be assigned. Ten individuals ranging from 13-18 mm SL were immature males. Their gonads consisted of two small testicular lobes having seminiferous tu? bules lined with crypts of spermatogonia and sper? matocytes, but no spermatozoa. Associated with each testicular lobe was a smaller body having anas? tomosing, cell-lined lumina but little other structur? al differentiation. These bodies joined with their re? spective testicular lobes posteriorly at the point of union of the two testicular lobes and were identical in appearance to typical early-stage accessory gona? dal structures (i.e. sperm duct glands sensu Miller 1984) found in immature males of other gonochoric goby species (Cole unpublished data). No oocytes or other ovarian features were evident in any of the immature testes observed in this sample. 1Wenty-one individuals ranging from 10-22 mm SL were immature females and their bilobed ova? ries contained previtellogenic oocytes in various stages of development, as well as localized protrub? erances of the ventral ovarian wall in the region of 305 Fig. I. Ovarian structure in a protogynous goby and in immature and early adult Gobiosoma illecebrosum. a) - cross-sectional view ofthe posterior portion of the ovary, including associated pAGS (indicated by arrows), in the protogynous goby, Coryphopterus hyalin us. b-f: G. illecebrosum. b) -ovary of immature female (11.8 mm SL) showing (arrow) large pAGS-like structures associated with the ventral wall in the posterior region close to the point of union of the two lobes; c) - pAGS-like structures of immature female, 19.1 mm SL; d) - reduced pAGS-like structures of slightly larger immature female, 20.4 mm SL; e) - remnant pAGS-like structures of small adult female. 23.6 mm SL; f) - remnant pAGS-like structures of slightly larger adult female. 24.2 mm SL. Bar is 100 Ilm. the union of the two ovarian lobes (Fig. 1b-d). These structures, which were quite large in the smallest immature females (Fig.1~), were similar in both appearance and location to pre cursive cell masses found in females of protogynous goby spe? cies (Fig. 1a) and described elsewhere (Cole 1988, 1990) which, upon sex change, develop into accesso? ry gonadal structures. The remaining fish (17 females and 4 males) were all adult. While localized enlargements of the ven? tral ovarian wall were large in small immature fe? males (Fig. 1b), these cell masses became both rela? tively and absolutely smaller (Fig. 1c,d) as females approached the size of maturity. Based on the pres- ence of vitellogenic oocytes in the ovary of the smallest adult female present in this sample, we esti? mated the size for first maturity for females to be 21 mm SL. Remnants of these cell masses were still visible in three adult females (21, 23 and 24 mm SL, Fig. Ie, f) but were absent in all of the remaining 14 adult females whose size ranged from 21-32 mm SL (Fig. 2). Neither immature nor adult females exhib? ited any testicular tissue, either in the form of semi? niferous tubules or unorganized crypts of sperma? tocytes, within the body of the ovary, or accessory gonadal structures. No oocytes were evident within the testes proper 306 ? Ul ~ I .... > .... '0 4 .S '"0 i : z 1 ? ! I ~ .S 4 \0.0 o ... a i Z 2 1 Mature females tI. , .. , .. n. , .. I' ....... If. II. It ... Standard length in mm Immature females tt. ,I. , .. t7. 11. 11. II. II. 17. II. 't. II. Standard length in mm Fig. 2. Bar histogram illustrating the occurrence of pAGS?like structures among (i) immature (n == 21); and (ii) adult (n == 17), female G. illecebrosum. Horizontal scale represents increasing size (standard length) in 2 mm increments. Open bar, females with no pAGS?like structures; black bar, females with vestigal pAGS?like structures; hatched bar, females with pAGS?like structures. or the associated AGS of the four adult males (18- 23 mm SL) examined in this sample. Gonad structure ofG. saucrum and G. multifasciatum Ofthe nine histologically examined adult female G. saucrum (13-17.5 mm SL), none had pAGS associ? ated with the ovary. Six of 10 immature females (9- 10 mm SL) had distinct pAGS and the remaining four had smaller pAGS-like structures associated with the caudo-ventral ovarian wall. Three adult male G. saucrum (15-19 mm SL) and four immature males (9-11 mm SL) all exhibited fully differentiated AGS associated with the testis. Among all of the mature males and one immature male (11 mm SL) the lumina of the AGS were filled with an acellular, acidophillic (i.e. eosin-staining) secretion. The AGS of the remaining three imma? tures were fully formed but inactive. No ovarian tis? sue was visible in the testes or AGS of any of the seven males examined. Among three immature female G. multifascia? tum of 12-14 mm SL, all had pAGS. No adult fe? males were present in this sample, although pAGS have been reported elsewhere in adult females of G. multifasciatum (Cole 1988). All three males of 19-21 mm SL were adult with large, well-developed AGS, the lumina of which were filled with sperma? tozoa. No ovarian tissue was visible. Experimental rearings ofG. illecebrosum Robertson & Justines (1982) previously ran experi? ments in which groups of 4-6 adult females were maintained together for 3-5 weeks. Since the num? ber of replicates of those experiments was small (n = 4) and they found one male among the females at the end of the experiments, we set up some addi? tional all-female groups to verify gonochorism. Up? on histological examination of the six groups each consisting of three immature females, and the three groups each made up of three adult females, no in? dividuals showed any indication of any partial or complete sex change, ovarian degeneration or early-stage testicular tissue. In the single group of five immature males, no individual showed any signs of testicular degeneration or the appearance of early-stage ovarian tissue in their gonads. Of 27 immatures reared in isolation, 12 were fe? male and 15 male. This ratio is not significantly dif? ferent from 1:1 (Yates corrected X2 = 0.143, P = 0.71). Among the 44 pairs of immatures reared to maturity,20% (n = 9) were made up of two females, 18% (n = 8) of two males and 62% (n = 27) of a fe? male and male, for a total of 45 females and 43 males. This distribution was not significantly differ? ent from that expected by chance (IF:2MF:1M) from random assemblages of pairs from a popula? tion with a 1:1 sex ratio (G = 1.19, P = 0.60). Of 18 immatures reared with an adult female, 33% were female and 67% were male; of 16 immatures reared with an adult male, 31 % were female and 69% were male. Neither of these sex ratios was significantly different from 1:1 (with adult female, G = 2.039, P = 0.15; with adult male, G = 2.306, P = 0.13) and did not differ from each other (G test for independence, G = 0.017, P = 1.0, Sokal & Rohlf 1981). Discussion In various species of sequentially hermaphroditic fishes, including gobies, sex-change can be induced in some adult individuals by a change in the social environment (Fishelson 1970, 1975, Robertson 1972, Fricke & Fricke 1977, Warner 1978,1988, Sha? piro 1979,1981,1987, Shapiro & Lubbock 1980, Ro? bertson & Justines 1982, Fricke 1983, Ross 1983, Cole & Robertson 1988). By the same token, one might expect that variation in social conditions dur? ing development from an undifferentiated juvenile state to adulthood should expose lability in gonadal development that may be present in a species, par? ticularly one with close relatives that do show adult sexual lability. In our experiments we varied the so? cial conditions under which differentiated adult G. illecebrosum were maintained and under which both differentiated and undifferentiated juveniles developed to adulthood. We predicted that if G. i/? lecebrosum is a strict gonochore with a 1:1 primary sex ratio that lacks lability in sexual potential, then no sex-change should occur among either groups of mature females, immature females or mature males; and that the adult sex-ratio offish that devel? oped from an undifferentiated state to maturity 307 should not differfrom 1:1 regardless of the social sit? uation in which they were reared. The results of the present study support and ex? tend those of Robertson & Justines (1982) which in? dicated a lack of sexual lability in adult G. illecebro? sum. Comparisons of adult sex-ratios and size-fre? quency distributions from field collections present? ed here demonstrate that G. illecebrosum exhibits male and female size-frequency distributions and a 1:1 sex-ratio typical of gonochore fish species. In ad? dition, the failure of sexually undifferentiated juve? niles that were reared to maturity under similar, and varying, social conditions, and of experimental groups of adults, to show signs of any developmen? tal plasticity indicates that G. illecebrosum is an illa? bile gonochore with a 1:1 sex ratio. Since sample sizes in each of the experiments we ran were not large, the power of each statistical test to detect a deviation from a 1:1 sex ratio is not high. In particular, it should be noted that the sex ratio of single juveniles reared with a female was (non-sig? nificantly) male-biased. Considering the experi? mental social situation, this bias is in the direction one would expect if sexual differentiation was plas? tic. However, we do not give much weight to this result since the experiment in which juveniles were reared with a male also produced a (non-signifi? cant) male bias. Further, it is possible to combine the results of a series of statistical tests that aim to test the same hypothesis. We used this 'combination of probabilities' (Sokal & Rohlf 1981) technique to test whether the four experiments in aggregate pro? duced sex ratios different from 1:1 and obtained a probability value of 0.75, indicating that the non? significant sex ratio biases in each individual experi? ment do not indicate an overall sex ratio different from 1:1. Immature and mature females of all protogynous gobies so far studied bear distinctive cell masses as? sociated with the ovary, termed precursive accesso? ry gonadal structures (pAGS) which, during sex? change, develop into accessory glandular structures similar to sperm duct glands (sensu Miller 1984) de? scribed for gonochoric gobies (Cole & Robertson 1988, Cole 1988, 1990, Cole & Shapiro 1990, Cole unpublished data). As we expected, pAGS were present in the ovaries of immature. protogynous G. 308 multifasciatum and no such structures were present among adult females of the two examined gono? chore species, G. illecebrosum and G. saucrum. However, among immature females of these latter two species, ovary-associated structures that seem to be similar, both in appearance and location on the ovary, to pAGS, were found. This indicates that both G. illecebrosum and G. saucrum, while func? tional gonochores, possess some hermaphroditic features in the immature ovary. The presence of pAGS-like structures among im? mature females of G. illecebrosum evidently is not related to an ability to change sex at that stage, since our experiments do not indicate that there is any lability in the sexual development of G. illecebro? sum. This suggests that the pAGS-like structures found in immature G. illecebrosum and G. saucrum are either unresponsive, or not competent to re? spond, to changes that activate pAGS in sex-chang? ing gobies. The progressive diminution of the pAGS-like structures as females of G. illecebrosum approach the size of maturity, and their absence in all but the smallest mature females, is striking. Were these structures merely extensions of the ovarian wall, maturation should have had a neutral or posi? tive effect on their development. Their gradual re? duction in absolute size with approaching maturity and ultimate disappearance among adults suggests that endogenous changes associated with matura? tion in females are antagonistic to their retention. This would be consistent with the fate of testis-asso? ciated tissue which presumably requires a certain level of circulating androgens for continued main? tenance or further development. According to Cole (1988), the presence of pAGS? like structures associated with the ovary in gobiid species is a reliable indicator for protogyny. How? ever, based on our findings here, it is evident that this postulation requires modification. Clearly, go? nad structure, particularly among immatures, does not necessarily reflect functional sexual pattern, at least in some Gobiosoma and possibly in other go? bies. Consequently, only the presence of pAGS in adult females may reliably predict protogyny among gobies. The development and subsequent disappearance of pAGS-like structures in two gonochore Gobio- soma species is intriguing. It is tempting to specu? late that these species, like present-day G. multifas? ciatum, were once protogynous and that this sexual pattern has been secondarily lost. If so, the transient pAGS observed here are simply ontogenetic rem? nants of a former labile sexual pattern and, as such, their ephemeral presence during development, if observed in other gonochore goby species, may be useful indicators for ancestral protogyny. However, inferring ancestral states and directionality for evo? lutionary transitions on the basis of a single morph? ological feature is risky at best, particularly in this case since little is known about gonad development in fishes in general, and gobies in particular. In order to make educated guesses regarding an? cestral sexual patterns in G. illecebrosum and G. saucrum, we first need to establish generalities for patterns of gonad deVelopment in Gobiosoma, in? cluding examinations of gonad development at the cellular level for both gonochoric and protogynous species. Comparative work on other gonochoric and protogynous gobies, and other teleosts (see al? so Fishelson 1992), is also essential. As a family, go? bies exhibit a variety of sexual patterns including gonochorism, simultaneous hermaphroditism (St. Mary 1994) and protogyny (i.e. Cole 1990). In order to understand the evolution of this diversity, the ex? istence of patterns in the distribution of gonocho? rism and protogyny among gobiids, and the role of phylogeny in that distribution, need to be deter? mined. Unfortunately, phylogenetic relationships within this most speciose family of marine fishes (Nelson 1984) are still poorly understood (Hoese 1984, Birdsong et al.1988). Finally, details of the life histories and mating systems of gonochoric and protogynous gobies will need to be elucidated in or? der to understand the role of selective forces that have favored the development of protogyny in other fishes (e.g. Charnov 1982, Warner 1978, 1988), in producing this diversity of sexual patterns among the gobiids. Acknowledgements We thank the Kuna General Congress and the Gov? ernment of the Republic of Panama for permission to carry out this study in the San Bias Islands, Mar? valee Wake for helpful conversations regarding the difficulties in determining the evolutionary status of morphological features in vertebrates and two anonymous reviewers. Photographic assistance was generously provided by C. Coleby, Royal Ontario Museum and the Photographic Department of the Smithsonian Tropical Research Institute. J. Porter and R. van Hulst provided assistance with manu? script preparation, and statistical analyses and graphics, respectively. This research was supported by a Smithsonian Short-term Fellowship, a Bishop's University Senate Research Grant awarded to KSC and general STRI research funds. Portions of this manuscript were written while the senior author was supported by a National Science and Engineer? ing Research Council of Canada Operating Grant. References cited Arai, R 1964. Sex characters of Japanese gobioid fishes (I). Bull. Nat. Sci. Mus. Tokyo 7: 295-306. Birdsong, R$., E. 0. Murdy & E Pezold.1988. A study of vertebral column and median fin osteology in gobioid fishes with com? ments on gobioid relationships. BUll. Mar. Sci. 42: 174-214. Charnov, E.1982. Alternative life-histories in protogynous fishes: a general evolutionary theory. Mar. Ecol. Prog. Ser. 9: 305-309. Cole, K.S. 1983. Protogynous hermaphroditism in a temperate territorial marine goby, Coryphopterus nicholsi. Copeia 1983: 809-812. Cole, K.S. 1988. Predicting the potential for sex-change on the basis of ovarian structure, in gobiid fishes. Copeia 1988: 1082- 1086. Cole, K.S. 1990. Patterns of gonad structure in hermaphroditic gobies (Teleostei: Gobiidae). Env. BioI. Fish. 28: 125-142. Cole, K.S. & D.R Robertson. 1988. Protogyny in a Caribbean reef goby, Coryphopterus personatus: gonad ontogeny and so? cial influences on sex change. Bull. Mar. Sci. 42: 317-333. Cole, K.S. & D.Y. Shapiro. 1990. Gonad structure and hermaph? roditism in the gobiid genus Coryphopterus (Teleostei: Gobii? dae). Copeia 1990: 966-973. Egami, N. 1960. Comparative morphology of the sex characters in several species of Japanese gobies, with reference to the ef? fects of sex steroids on the characters. J. Fac. Sci. Vnlv. Tokyo, Ser. IV 9: 67-100. Fishelson, L. 1970. Protogynous sex reversal in the fish Anthias squamipinnis (Teleostei, Anthiidae) regulated by the pres? ence or absence of a male fish. Nature 227: 90-91. Fishelson, L. 1975. Ecology and physiology of sex reversal in An? thias squamipinnis (Peters), (Teleostei, Anthiidae). pp. 284- 309 294. Tn: R Reinboth (ed.) Intersexuality in the Animal King? dom, Springer Verlag, New York. Fishelson, L.1989. Bisexuality and pedogenesis in gobies (Gobii? dae: Teleostei) and other fish, or why so many little fish in trop? ical seas? Senckenbergia marit. 20: 147-169. Fishelson, L. 1992. Comparative gonad morphology and sexuality of the Muraenidae (Pisces, Teleostei). Copeia 1992: 197-209. Fricke, H. W. 1983. Social control of sex: field experiments with the anemonefish Amphiprion bicinctus. Z. TIerpsychol. 61: 71-77. Fricke, H.W. & S. Fricke. 1977. Monogamy and sex change by aggressive dominance in a coral reef fish. Nature 266: 830-832. Hoese, D.E 1984. Gobioidei: relationships. pp. 588-591. Tn: H.G. Moser et al. (ed.) Ontogeny and Systematics of Fishes, Spec. Pub. No.1, Am. Soc. of Ichthy. and Herpet., Allen Press, Law? rence. Lassig, B.R.1977. Socioecological strategies adopted by obligate coral-dwelling fishes. pp. 56S-S70.In: D.L. Taylor (ed.) Proc. Third Intern. Coral Reef Symp. I, Biology, Miami. Miller, P.J. 1984. The tokology of gobioid fishes. pp. 119-153. In: G.w. Potts & RJ. Wootton (ed.) Fish Reproduction: Strate? gies and Tactics, Academic Press, New York. Nelson, J.S. 1984. Fishes of the world, 2nd ed. John Wiley and Sons, New York. 416 pp. Robertson, D.R. 1972. Social control of sex reversal in a coral reef fish. Science 177: 1007-1009. Robertson, D.R & G. Justines.1982. Protogynous hermaphrod? itism and gonochorism in four Caribbean reef gobies. Env. BioI. Fish. 7: 137-142. Robertson, D.R & RR Warner. 1978. Sexual patterns in the la? broid fishes of the western Caribbean, II. The parrot fishes (Scaridae). Smithsonian Contributions to Zoology 255: 1-26. Ross, RM., G.S. Losey & M. Diamond. 1983. Sex change in a coral reef fish: dependence of stimulation and inhibition on relative size. Science 221: 574-575. St. Mary, C.M.I994. Novel sexual patterns in two simultaneously hermaphroditic gobies, Lythrypnus dalli and Lythrypnus ze? bra. Copeia (in press). Shapiro, D.y'1979. Social behavior, group structure and the con? trol of sex reversal in hermaphroditic fish. Advances in the Study of Behavior 10: 43-102. Shapiro, D.Y. 1981. Size, maturation and the social control of sex reversal in the coral reeffishAnthias squamipinnis (Peters). J. Zool. (Lond.) 193: 105-128. Shapiro, D.Y. 1987. Differentiation and evolution of sex change in fishes. BioScience 37: 490-497. Shapiro, D.Y. & R. Lubbock. 1980. Group sex ratio and sex re? versal. J. Theor. BioI. 82: 411-426. Sokal, RR & EJ. Rohlf. 1981. Biometry. The principles and practice of statistics in biological research, 2nd ed. W.H. Free? man and Company, San Francisco. 859 pp. Warner, RR 1978. The evolution of hermaphroditism and uni? sexuality in aquatic and terrestrial animals. pp. 78-101. In: E.S. Reese & FJ. Lighter (ed.) Contrasts in Behavior. John Wiley & Sons, New York. Warner. RR 1988. Sex change in fishes: hypotheses. evidence and objectives. Env. Bio/. Fish. 22: 81-90. Kassi Cole and her typical smile. Photograph by M. Keenleyside.