Gonadal histology in the self-fertilizing hermaphroditic ...

Gonadal histology in the self-fertilizing hermaphroditic fish Rivulus marmoratus (Pisces, Cyprinodontidae) CRISTINA G. SOTO,JOHNF. LEATHERLAND, A N D DAVID L. G. NOAKES~ Institute of Ichthyology, Department of Zoology, University of Guelph, Guelph, Ont., Canada N l G 2 W l Received June 17, 1992 Accepted July 2, 1992

Can. J. Zool. Downloaded from www.nrcresearchpress.com by OREGON STATE UNIVERSITY on 02/02/12 For personal use only.

SOTO,C. G., LEATHERLAND, J. F., and NOAKES,D. L. G. 1992. Gonadal histology in the self-fertilizing hermaphroditic fish

Rivulus marmoratus (Pisces, Cyprinodontidae) . Can. J. Zool . 70: 2338 - 2347. In nature, most individuals of Rivulus marmoratus are reportedly self-fertilizing hermaphrodites. A few rare individuals are males, but females have never been found in the field or laboratory. We describe the gonadal histology of this unique fish, mainly on the basis of light-microscopic studies with some elaboration by electron microscopy. Analysis of gonadal structure and the characterization of the stages of spermatogenesis and oogenesis allowed us to categorize functional gender and to construct the probable sequence of gonadal development in mature individuals. In our laboratory, some hermaphrodites underwent ovarian regression and testicular proliferation to become secondary males. One individual also developed as a primary male. However, the type of male could generally not be distinguished. Testicular tissue composed only a small portion (less than 10%) of the gonad of most hermaphrodites.

SOTO,C. G., LEATHERLAND, J. F., et NOAKES,D. L. G. 1992. Gonadal histology in the self-fertilizing hermaphroditic fish Rivulus marmoratus (Pisces, Cyprinodontidae). Can. J. Zool. 70 : 2338 - 2347. En nature, la plupart des individus de Rivulus marmoratus sont hermaphrodites et pratiquent I'auto-fkcondation. Quelques rares individus sont des miles, mais aucune femelle n'a jamais kt6 recontree en nature ou en laboratoire. L'histologie des gonades de ce poisson trks particulier a Cti etudiee au microscope photonique et certains details ont 6te examines au microscope Clectronique. L'analyse de la structure des gonades et la determination des stades de spermatogenkse et d'ovogenkse nous ont permis de definir le sexe fonctionnel des poissons et de reconstruire la sequence probable du dkveloppement gonadique chez des individus a maturiti. En laboratoire, certains hermaphrodites ont subi une regression des ovaires et une recrudescence testiculaire et sont devenus des miles secondaires. Un individu est Cgalement devenu un mile primaire. Cependant, de f a ~ o ngCnCrale, le type de mile pouvait rarement &re distinguk. Le tissu testiculaire constituait seulement 'une petite fraction (moins de 10%) de la gonade chez la plupart des hermaphrodites. [Traduit par la redaction]

Introduction The cyprinodontid fish Rivulus marmoratus is the only known naturally occurring, self-fertilizing hermaphroditic vertebrate (Harrington and Rivas 1958; Harrington 1961; Atz 1964). Three sexual phenotypes can occur in fish bred in the laboratory: hermaphrodite, primary male, and secondary male (Harrington 1967, 1968, 1971, 1975). Primary males are reportedly produced by incubation of embryos at lower temperature ( < 20°C) (Harrington 1967). Secondary males originate from hermaphrodites by regression of the ovarian component of the ovotestes and proliferation of the testicular component (Harrington 1967, 1971). Almost all fish captured in the field have been hermaphrodites. Until recently, only two males had been caught in the wild in Florida (Harrington 1975). The third wild-caught male is reported here. Several presumptive males, based on their bright orange body colour, were caught in Florida in 1987 (S. Ritchie, personal communication), and a significant proportion (10 -25 %) of samples collected in Belize were males (Davis et al. 1990). Secondary males have not been confirmed in nature (Harrington 1971). Although some aspects of the biology of R. marmoratus have been described, there has been no report on the histology of the gonads of the species. This is essential both for an understanding of the biology of the species and for comparison with other hermaphrodites. We studied the functional histology of the gonad to identify gender and to assess the process of change from hermaphrodite to secondary male. We define gender following Lloyd (1980), according to two gonadal indicators of gametogenic tissue of the gonads of 'Author to whom correspondence should be addressed. Printed in Canada 1 lmprimk au Canada

individuals. We also consider the importance of functional tissue types, as has been done with sex-changing species. Specifically, gonadal histology has often been used to determine reproductive function, with maturity criteria from gonochores: ovaries are mature if they contain evidence of present or prior yolk deposition; testes are mature if they contain spermatids or spermatozoa (Sadovy and Shapiro 1987). Gonadal transition in R. marmoratus (i.e., ovarian regression and testicular proliferation) might be thought to be comparable to sex change in protogynous fishes. However, although there are parallels between sex change and the process of testicular proliferation and ovarian regression, there are important differences. The gonads of protogynous hermaphrodites function first as ovaries and later as testes. Rivulus marmoratus functions first as a self-fertilizing hermaphrodite and then becomes a secondary male by losing female structure and function (Harrington 1971).

Materials and methods Source of $sh Most of the R. marmoratus used in this study were obtained from W. P. Davis (Environmental Protection Agency, Gulf Breeze, Fla.) or were descendants of these fish raised at the University of Guelph. They consisted of three genetic strains defined by histocompatibility (Kallman and Harrington 1964; Harrington and Kallman 1968), together with descendants of fish caught in Naples, Florida, in 1980, plus fish caught on Marco Island, Lower Matecumbe Key, and Capri Island, Florida, in 1986. Representative specimens have been deposited at the Royal Ontario Museum (ROM), Toronto (ROM Accession No. 5406, catalogue Nos. 55293 -55296). Fish ranged from 14 to more than 48 months old (laboratory fish), from 16.0 to 42.0 mm standard length, and from 50 to 1230 mg wet weight.

2339

SOT0 ET AL.


Pretreatment of fish Fish were held individually at 26 f 1OC and 12 h light : 12 h dark photoperiod in 14 parts per thousand synthetic seawater (Instant Ocean, Aquarium Systems, Mentor, Ohio), and fed Artemia salina nauplii (approximately 15 mg dry weight) every other day. Histology The description of gonads is a composite of 140 fish. Several were sectioned in serial longitudinal section, several were processed for electron microscopy, and most (125) were sectioned in serial transverse section (with a subset of these for the detailed study of gametogenesis and gonadal structure). The gonad of R. marmoratus is a cream-coloured, bilobed structure between the intestine and the swim bladder. The anterior tips of the lobes are attached to the body wall by connective tissue. The lobes fuse posteriorly and the genital duct exits between the rectum and urinary ducts. Four equi-spaced sections in the lobes of the gonads anterior to the bifurcation were examined in 102 fish as part of a larger study (Soto 1988). Since R. marmoratus is a simultaneous hermaphrodite, we have adapted the terminology describing gonads accordingly: ,thus, gonadal lumen instead of ovarian lumen and common genital sinus instead of oviduct (Cole and Robertson 1988; K. Cole, personal communication). All fish were killed with excess quinaldine sulfate (Argent Chemical Laboratories, Inc., Redmond, Wash.). For routine light microscopy, the gonads were fixed in Bouin's solution, embedded in paraffin wax, sectioned at 7 pm, and stained with haematoxylin and eosin by means of standard procedures. For electron microscopy, tissues were fixed in ice-cold cacodylate-buffered (0.1 M, pH 7.4) glutaraldehydeparaformaldehyde (3 % :2%) for up to 18 h, minced in fresh fixative, washed in buffer, postfixed in cacodylate-buffered 1% osmium tetroxide for I h at 4OC, washed again in buffer, dehydrated in acetone, and embedded in Epon or Spurr resin. "Thick" sections (0.5 pm) were prepared for orientation of the blocks, heat mounted on glass slides, and stained with 1% azure I1 in 1% sodium borate. "Thin" sections (gold) were mounted on uncoated copper grids, stained with uranyl acetate (saturated solution in methanol) and Reynold's lead citrate. The sections were examined using a Joel transmission electron microscope. A Zeiss interactive digital analysis system (Z~DAS) in conjunction with a Zeiss compound microscope was used to measure the diameters of the stages of spermatogenesisand oocyte growth. Only the diameters of oocytes with obvious nuclei were measured, with the exception of oocytes at yolk stage 11, which lacked an obvious nucleus. ZIDASsoftware gave maximum and minimum oocyte diameters, from which we used the mean for each oocyte. The Z ~ D A Swas also used to quantify areas of ovarian and testicular tissue and total cross-sectional area of the gonads (lobes within fish were summed for these measures). Gonadal indicators Gonadal indicators, testicular ratio and ovarian status, were chosen to reflect both gender and the process of transition from hermaphrodite to secondary male (Soto 1988). They were based on histological evidence of sex change in several protogynous species (Moe 1969; Shapiro 1981; Nakamura et al. 1989), maturity scales of ovaries of several gonochoristic species (Moe 1969; Forberg 1982), and on our own observations of the gonads of R. marmoratus. The ratio (%) of cross-sectional testicular area to total lobular cross-sectional area (TXA) was calculated from four sections for each fish. The mean of the four is testicular ratio (TR). Measurement of TXA omitted atretic follicles and recognizable oocytes greater than 300 pm in diameter, since they depressed the ratio excessively. For ovarian status the total number of recognizable oocytes of each oocyte stage ( 0 s ) was counted for the four sections. Scores were made on an arbitrary 0-4 scale where 0 = no oocytes, 1 = early peri-nucleolus stage, 2 = late perinucleolus stage and early yolk vesicle stage (YVS), 3 = > l mid or late YVS, 4 = mid or late YVS and > 1 yolk stages I and 11. Fish with gonads of ovarian status 3 or 4 were considered to be vitellogenic sensu Wallace and Selman (198 1). PKM, a k-means cluster analysis program (Dixon et al. 1981), was used to organize the data of ovarian status and TR into seven groups of similar fish.

Results Gonadal structure in hermaphrodites The lobes of hermaphrodites were oval in cross section. Within them, the gonadal lumina were lined with a fingerlike epithelium (Fig. 1A). The gonadal lumina began at the upper central part of each lobe, continued posteriorly (Fig. 2C), and united in the common genital sinus (CGS). The gonadal ducts exited posteriorly between the rectum and urinary ducts (Fig. 2B). Oocytes and spermatogenic tissue were intimately associated with no apparent physical boundary between them. The spermatogenic tissue of all fish, regardless of gender, was composed of cysts of cells each of a single stage of spermatogenesis developing synchronously (Fig. 2A). The CGS had a thick outer wall of eosinophilic fibres. Its lumen was lined with simple, columnar, basophilic epithelium (Fig. 2D). Gametogenic stages found in this region were restricted to spermatozoa in ducts in the walls of the CGS (Figs. 2D, 2F) or the occasional mature oocyte (Fig. 2E). Some ducts containing loose spermatozoa (not in cysts) were fused with adjacent ones, possibly permitting the movement of spermatozoa (Fig. 2F). No well-defined, typical sperm ducts, other than those mentioned, were seen in the hermaphroditic gonad.

Stages of oocyte maturation in hermaphrodites Oogonia Oogonia occurred in clusters or pairs, embedded in the fingerlike folds of the epithelium (Fig. 1A). They were present in small numbers and difficult to distinguish from large spermatogonia. Both the cytoplasm and nucleus were pale and translucent, and cell boundaries were difficult to distinguish. A single, dark pink nucleolus was prominent. Mean cell 0.65 pm (range 13 -22), with a mean diameter was 15 nuclear diameter of 10 pm (range 8 - 13).

+

Early peri-nucleolus stage (EPNS) Early oocytes overlapped somewhat in size with larger oogonia but had cytoplasm that stained markedly with haematoxylin (Fig. 1B). The cell nuclei contained two or three nucleoli, usually at their periphery. Cell and nuclear diameters were 19 pm (range 15 -25) and 12 pm (range 8 - 13), respectively. Mean cell and nuclear diameters of the more haematoxylinpositive cells were 64 pm (range 15 - 124) and 26 pm (range 8 -58), respectively. In these cells, the nucleoli were mainly located in the periphery of the nucleus (Fig. 1B).

Late peri-nucleolus stage (LPNs) (Fig. 1 B) The cytoplasm of these cells was stained pale purple. The nucleoli (up to nine in number) were still peripherally located and variable in size (1.5 -4.1 pm). The nucleus was smaller relative to oocyte diameter (40 - 57 %). Cell and nuclear diameters were 1 16 pm (range 88 - 154) and 4 8 pm (range 35 - 79), respectively. Throughout this stage the oocyte was surrounded by a primary theca of spindle-shaped cells, 0.7 - 1.0 pm thick.

Yolk vesicle stage In the early YVS, the cytoplasm was more granular in appearance and there were spaces on the oocyte periphery (Fig. 1B) that sometimes were larger and formed a peripheral zone of clear spheres. These early YVS oocytes were less frequently stained dark purple with tiny vesicles. Cell and nuclear diameters were 133 pm (range 88- 165) and 51 pm (range 36 - 69), respectively. In larger oocytes, the mid and late YVS, the cytoplasm was


CAN. J . ZOOL. VOL. 70, 1992

FIG. 1. Gonadal histology of ovarian tissue of hermaphroditic Rivulus marmoratus. (A) A cluster of oogonia ( 0 ) in the epithelium (E) surrounding an early oocyte (EO). Scale bar = 20 pm. (B) Several stages of oocyte growth are indicated: early peri-nucleolar stage (EP), late peri-nucleolar stage (LP), and yolk vesicle stage (YV). Scale bar = 200 pm. (C) The yolk stage I oocyte shown (Y 1) has spheres of red cytoplasm (C) that apparently coalesce in yolk stage I1 (Y2). Scale bar = 100 pm. (D) A very large yolk stage I1 oocyte (Y2) is shown. The acidophilic cytoplasm has coalesced (C). Scale bar = 400 pm. (E) An electron micrograph shows the follicular layers and inner membrane of a yolk stage I1 oocyte: the outermost layer, the theca (TH), the inner layer, the granulosa (G), and the zona radiata (Z), respectively. Scale bar = 5 pm. (F) An electron micrograph shows spheres of very different size but similar texture and appearance (indicated by arrows). They occupy most of the cytoplasm in this yolk stage I1 oocyte. Scale bar = 10 pm.

SOT0 ET AL.

paler than in the early YVS, with greater numbers of larger vesicles (Fig. IB). some vesicles- contained pale pink- -or purple-stained inclusions. Concomitant with yolk vesicle formation, the cytoplasm was greater in size relative to the spherical nucleus. The latter was eosinophilic and granular in appearance. The nucleoli were large (3.1 - 5.1 pm) and still located on the periphery of the cell. Cell and nuclear diameters were 193 7 pm (range 143-270) and 63 pm (range 44 - loo), respectively. There was a thin thecal layer of between 0.6 and 1.1 pm associated with all YVS follicles. A four-layered membrane structure (two basophilic follicular layers, the theca (1.1 pm) and the granulosa (1.5 pm), and two inner layers, a thin eosinophilic granular layer and an inner smooth dark eosinophilic layer with a combined thickness of 0.6-0.9 pm) was evident in the late YVS follicles. The eosinophilic granular layer was variable in occurrence and seemed to form after the zona radiata. Since the process of accumulation of vesicles is continuous, an arbitrary distinction was made between this stage and the following. Thus, an oocyte with vesicles comprising up to 50% of the cytoplasmic area was considered a late YVS oocyte.


+

Yolk stage I Yolk vesicles occupied at least 50% of the area in yolk stage I oocytes (Fig. 1C) . The nucleus was irregular in outline and smoother in texture, and contained up to six nucleoli. The spherical nuclei were up to 4.3 pm in diameter. The follicular layers and oocyte membranes were well defined: a thin outer layer or theca of spindle-shaped cells (1.0 pm thick, range 0.6 - 2. l), a thicker, inner granulosa layer composed of large oval cells (2.1 - 3.2 pm thick), a thin layer irregular in appearance, the tunica propria, (1.6 pm, range 1.1 -2.1, not always visible), and an inner striated, eosinophilic layer, the zona radiata (ZR) (1.6-6.3 pm thick). Oocytes on the "borderline" between yolk stage I and I1 had several small acidophilic spheres in the cytoplasm (Fig. 1C). Mean cell and nuclear diameters were 339 pm (range 221 -610) and 72 pm (range 41 - 106), respectively. Yolk stage 11 In larger or more mature oocytes, eosinophilic cytoplasm comprised the majority of the oocyte. The cytoplasm of the largest oocytes appeared dark red and uniform with a peripheral band of yolk vesicles, some of which contained pale purplestained inclusions (Fig. 1D) . Thicknesses were as follows: thecal layer 1.1 -3.6 pm, granulosa layer 7.4 - 10.2 pm, zona radiata 3.2 -28.1 pm (Fig. 1E). The tunica propria, described above, was no longer distinguishable. The mean oocyte diameter was 722 pm (range 535 - 1000). In electron micrographs, some vesicles were electron translucent, while others contained minute granules and (or) electron-dense coalesced yolk globules (Fig. 1F) .

Gonadal structure in males Seven of the eight bright orange fish examined in detail, including a bright orange fish caught in the wild in 1986, contained no evidence of oocytes or ovarian tissue and so were categorized as males. Gonadal structure and the arrangement of testicular tissue differed from that of hermaphrodites. The eighth fish differed greatly and will be discussed later. Males were characterized by paired gonads with wedge-shaped lobes in transverse section, containing solidly packed spermatogenic

234 1

tissue (Fig. 3A). Histologically, males lacked a common genital sinus. The posterior region was often composed of one major and several minor ducts and thin basophilic walls. A discrete large sperm duct was evident only near the posterior section of the testicular lobe (Fig. 3A). In sections of Epon-embedded tissue the gametogenic cells appeared to be contained in compartments of different sizes (as long as 374 pm and as wide as 2 1 pm), some void or partially filled with spermatozoa (Fig. 3B). The compartment walls were composed of large, ill-defined Sertoli cells that had cytoplasmic projections amongst the gametogenic tissue; these cells contained large nuclei with the heads of spermatozoa in close association with the cytoplasm in some cases (Figs. 3B, 3C). Along the entire length of the gonad, ripe testes had a large collection of sperm in the central region of the lobes, sometimes occupying 85 % of the gonad in cross-sectional area. The sperm pool or sinus was formed by the coalescence of the ripe nests at the mouths of the tubules where minor sperm ducts led into the common duct. The testicular lobes of the seven males had a thin external layer of epithelial cells surrounding spermatogenic cysts. The outer nests contained spermatogonia, and the central nests usually contained spermatozoa (Fig. 3D). Thin acidophilic walls of tubules could sometimes be seen enclosing a progression of cysts (Fig. 3D). The exceptional (eighth) male was considered a secondary male, since it contained evidence of previous hermaphroditism. It differed from the others in several ways. The gonadal lobes were more oval than wedge shaped in transverse section. In the anterior region of the gonad, one lobe was large, with compact testicular tissue. The other lobe was small, with several early peri-nucleolus stage oocytes surrounding a large, obvious gonadal lumen. Spermatogenic tissue appeared to be expanding within the ovarian lumen, a condition also seen in some hermaphrodites (Fig. 3E). Presumed atretic oocytes, brownish-yellow clumped material with basophilic nuclei and pale yellow translucent material, were seen (Figs. 3E, 3F). The infolding layer of basophilic epithelium that surrounded the gametogenic cells resembled the outer wall of the common genital sinus and the gonadal lumen, respectively (Figs. 3E, 3F). One male was considered to be a primary male, since it was 4 months old, the age at which fish become sexually mature in the laboratory.

Stages of spermatogenesis in males and hermaphrodites Spermatogonia Spermatogonia resembled oogonia of hermaphrodites, since both were translucent cells with pale nuclei and generally prominent nucleoli. The functional terminology was based solely on their locations within the ovotestis: adjacent to spermatogenic cysts or other oocytes, respectively. Fully mature oogonia (20 pm) tended to be larger than the largest spermatogonia (15 pm). Large spermatogonia were generally present in small cysts (1 -3 cells) (Fig. 2A). Smaller spermatogonia were grouped together in cysts of 6- 18 cells. The cytoplasmic boundaries were usually difficult to distinguish but cell sizes ranged between approximately 7 and 15 pm in diameter. Nuclei, which were more or less spherical, ranged between 2.6 and 7.3 pm in diameter. Sperrnatocytes Primary spermatocytes were found in cysts of 11 -60 cells. The cytoplasmic boundaries were indistinct, but the nuclei

Can. J. Zool. Downloaded from www.nrcresearchpress.com by OREGON STATE UNIVERSITY on 02/02/12 For personal use only. 2342 CAN. J . ZOOL. VOL. 70. 1992

S


were more uniformly oval and haematoxylin positive than spermatogonia (2.7-5.2 ,um in diameter) (Fig. 2A). In the Epon-embedded preparation, cell and nuclear diameters were 4.2-6.3 and 2.7-5.7 ,urn, respectively. Secondary spermatocytes These were infrequently found and when present closely resembled primary spermatocytes. However, their nuclei tended to stain more intensely with haematoxylin and they were more irregular in size and more tightly packed than primary spermatocytes. In paraffin-embedded preparations the nuclei ranged from 2.1 to 3.7 ,urn, whereas in Epon-embedded sections the cell and nuclear diameters were 3.2 -4.7 and 2.1 - 3.2 ,um, respectively (Fig. 2A). Spermatids Spermatid nuclei were more haematoxylin positive and uniform in size within a cyst than those of spermatocytes; larger numbers of spermatids comprised a tightly packed cyst. They were spherical with very high nucleus:cytoplasm ratios (Fig. 2A). Spermatid nuclei ranged in diameter from 1.0 to 2.7 ,um and from 1.5 to 2.3 ,um in paraffin- and Epon-embedded preparations, respectively. Cell diameters, measured in Epon sections, were 2.6 - 3.2 ,urn. Spermatozoa These were the smallest cells with the most darkly staining, or most electron-dense, nuclei (0.8 - 1.4 and 0.8 -2.1 ,um (Fig. 2A) in paraffin-embedded and electron micrographs, respectively). Cluster analysis The cluster analysis provided convenient groupings to facilitate description of 102 gonads (Fig. 4). Most fish in clusters 2, 5, 6, and 7 had low (less than 20%) TRs and oocytes of various stages. All fish in these clusters were considered to be hermaphrodites. The fish of cluster 1 were designated males, since they had large amounts of testicular tissue and lacked oocytes. The two males with early peri-nucleolus stage oocytes ( 0 s = 1) were secondary males (cluster 4). Clusters 3 and 4 are fish with intermediate quantities of testicular tissue and were categorized as transitional. The justification for these groupings is in the Discussion.

Discussion Various stages of oogonial maturation have been described in several teleosts (Yamamoto and Yamazaki 1961; Forberg 1982; Begovac and Wallace 1987; Down and Leatherland 1989). We preferred to use the general term "early oocyte," distinguished from "oogonium" by its opaque cytoplasm, to identify the early stages of oogenesis. The EPNS stage identified in R. marmoratus was characterized by an increase in basophilic properties of the cytoplasm and the appearance of

O ET AL.

2343

multiple nucleoli comparable to that reported in other teleosts (Forberg 1982; Wallace et al. 1987). In some species, nesting oocytes have been identified (Treasurer and Holliday 1981; Forberg 1982; Robb 1982). These remain in meiotic prophase for an extended period of time. For R. murmoratus, there are no data to confirm resting oocytes, but descriptions of developing oocytes (Treasurer and Holliday 1981) coincide with the increase in size and paling of the cytoplasm seen in LPNs oocytes. Only one type of yolk vesicle was identified in R. murmoratus. It was either a spherical clear space, or contained pale pink or purple contents. More accurately termed cortical alveoli (Wallace et al. 1987), these yolk vesicles are difficult to preserve, frequently lose their staining characteristics, and assume a vacuolar appearance. In many fish species, droplets or inclusions begin to form during the YVS (Wallace et al. 1987; Yamamoto and Yamazaki 1961; Moe 1969; Khoo 1979; Forberg 1982). The acidophilic spheres in our study were probably globules of yolk (vitellogenin). In stage I1 oocytes these yolk globules accumulated in a manner similar to ,thatdescribed in goldfish (Carassius auratus), carp (Cyprinus carpio), and mummichog (Fundulus heteroclitus), with minute, spherical bodies coalescing in the inner part of the ooplasm, displacing the yolk vesicles to the periphery (Yamamoto and Yamazaki 1961; Selman and Wallace 1986; Down and Leatherland 1989). Yolk globules of the flounder (Liopsetta obscura) (Yamamoto 1957), red grouper (Epinephelus morio) (Moe 1969), and capelin (Mallotus villosus) (Forberg 1982) coalesce to make a continuous mass of acidophilic yolk. The presence of oocytes in yolk stage I or I1 in R. murmoratus indicates true vitellogenesis (sensu Wallace and Selman 1981). Specific changes in the nucleus and the micropyle that identify several further stages of yolk proliferation and maturation in some species (Yamamoto and Yamazaki 1961; Treasurer and Holliday 1981) were not seen in R. murmoratus. Instead, our categorization was based on a combination of oocyte size and the appearance of the cytoplasm and nucleus. Two follicular layers, the theca and granulosa, are apparent at the close of the YVS in R. marmoratus as in other species (Yamamoto and Yamazaki 1961; Robb 1982). Treasurer and Holliday (1981) described a third layer between the other two, at the late developing stage, but we did not observe such a layer in R. murmoratus. The tunica propria and zona radiata of R. murmoratus became evident at the same time as the granulosa layer, during the late YVS, and began to widen throughout yolk stage I. Between yolk stages I and 11, the zona radiata increased rapidly in size. The order of this process is not surprising, since the two inner membranes derive from the granulosa (Treasurer and Holliday 1981). Perhaps the tunica propria disappears during yolk stage 11, since only the theca, granulosa,

FIG. 2. Histology of the gonads of males and testicular tissue of hermaphroditic Rivulus marmoratus. (A) Cysts of spermatogenic stages in the testis of a male: spermatogonia (SG), primary spermatocytes (SCl), secondary spermatocytes (SC2), spermatids (SD), and spermatozoa (SZ). Scale bar = 20 pm. (B) The exit of the testis in relation to the rectum (R) and the urinary duct (U). The arrows indicate two smaller ducts, one containing spermatozoa and one empty, and a large central duct containing spermatozoa (SZ). Scale bar = 150 pm. (C) A longitudinal section shows the general configuration of testicular tissue (T) and oocytes in a hermaphrodite. The gonadal lumen (L) runs along the length of the lobe. Scale bar = 200 pm. (D) The common genital sinus of a hermaphrodite is shown in transverse section. The outer wall (W) is thick and comprised of pink fibres. Spermatozoa (SZ) are found in ducts. The lumen is lined with infolding epithelium (E). Scale bar = 200 pm. (E) A more anterior part of the common genital sinus of the same fish as in Fig. 3D. Empty ducts are indicated by unlabelled arrows while others contain spermatozoa (SZ). The infolded epithelium is still apparent, lining the gonadal lumen (L), although a mature oocyte (M) occupies most of it. Scale bar = 150 pm. (F) Arrows indicate the points of fusion of three ducts containing spermatozoa. Scale bar = 50 pm.


C A N . J . ZOOL. VOL. 70, 1992


SOTO ET A L .

Ovarian

status

FIG. 4. Cluster groups of adult Rivulus marrnoratus to show the relationship between testicular ratio and ovarian status, as determined histologically. Details of cluster analysis are given in the text. Numbers of fish in each group are as follows: cluster 1, 27; 2, 21; 3, 6; 4, 2; 5, 19; 6, 23; 7, 5.

and zona radiata were evident in electron micrographs (Fig. 1E) . Our descriptions of the stages of spermatogenesis correspond closely to those of Hyder (1969) for Tilapia leucosticta, Moe (1969) for red grouper, Burke and Leatherland (1984) for Ictalurus nebulosus, and Down and Leatherland (1989) for carp. It was not possible to categorize spermatogonial stages, but the increased stainability of the spermatocytes is consistent with the disappearance of the nucleolus and the granular appearance of the cytoplasm in Fundulus species (Grier et al. 1980). Hermaphrodites generally had relatively small amounts of testicular tissue (relative to ovarian tissue) (TR < 20%) and oocytes of various stages. A gonad containing testicular tissue and vitellogei~icoocytes must be considered hermaphroditic (Sadovy and Shapiro 1987). Hermaphrodites of R. marmoratus generally had relatively small amounts of testicular tissue and oocytes of various stages. Ovotestes with exclusively early peri-nucleolus stage oocytes and small amounts of testicular tissue were not considered to be males. Since they were known to be chronologically mature, we assume that they were either in a mature resting stage (Moe 1969) and could potentially

2345

spawn again, or were aging as hermaphrodites and would not undergo transition to secondary males. Smith (1975) also noted that in functional hermaphrodites of R. marmoratus the testicular portions of the gonad remain small and variously located. The germinal tissue is therefore undelineated (Sadovy and Shapiro 1987), in contrast to delineated gonads in other hermaphroditic species that have male and female tissues separated by connective tissue. The quantity of testicular tissue in young hermaphrodites (Harrington 1971) may be as small as in the functional female phase of the protogynous Sacura margaritacea, a species with spermatogenic stages in the testicular zone during the female phase (Reinboth 1963). Hypoplectrus nigricans, the black hamlet, devotes approximately 9 % of its gonadal volume to testis (Fischer 1981). The small relative size of the testis in R. marmoratus is likely related to the fact that the species is self-fertilizing and thus needs only sufficient sperm to fertilize its own eggs internally as they are produced, compared with the large numbers of sperm necessary in species with external fertilization. The category "male" specifically describes those fish with large quantities of testicular tissue (TR > 70%), and either lacking oocytes or possessing small numbers of early perinucleolus stage oocytes. The former, including the seven males described in detail in the Results, may be primary males. The latter fish and transitionals were considered to be functional secondary males at different stages of ovarian regression and testicular proliferation. Early peri-nucleolus stage oocytes were the only oocytes present and evidence from protogynous species suggests that these are remnant oocytes and will not develop further (Moe 1969; Warner and Robertson 1978; Shapiro 1981; Nakamura et al. 1989; Cole and Robertson 1988). Many of the R. marmoratus males examined appeared to be of the restricted type (Grier et al. 1980; Grier 1981), since most of the spermatogonia were peripheral, with a progression of stages within regularly oriented tubules. However, spermatogonia in more central locations do not fit this definition. Several other cyprinodontid species are of the restricted type (Grier et al. 1980). Well-defined sperm ducts, as found in most gonochoristic teleost species (Harder 1975), were not seen. There appeared to be efferent duct-like compartments in the posterior region of some testes resembling the efferent ducts of the mummichog (Selman and Wallace 1986). In restricted testes, the Sertoli cells transform into efferent duct cells (van den Hurk et al. 1974; Grier et al. 1978, 1980; Selman and Wallace 1986). The appearance of the Sertoli cells of R. marmoratus comprising the compartment walls (Fig. 3B) agrees with observations of hypertrophy (Grier et al. 1980) of efferent duct cells of Poecilia latipinna. However, the appearance of large central sinuses filled with spermatozoa along the gonadal length is similar to that described by Hyder (1969) for T. leucosticta. In that species, the central lobules increasingly

FIG. 3. Various aspects of the testes of male Rivulus marmoratus. (A) General configuration of the lobe of the testis of a male. A large compartment with a loose thick wall of cells (indicated by the arrow) contains spermatozoa. Scale bar = 100 pm. (B) A thin section shows that compartments containing spermatozoa in a male are lined with loosely defined cells (indicated by the arrow). The difference in density between the "cysts" (CY) and the compartments should be noted. Scale bar = 30 pm. (C) An electron micrograph shows one of the cells containing the compartment walls of the same male as in Fig. 3B. The nucleus (N) and the head of a spermatozoon (SZ) are indicated. Scale bar = 2 pm. (D) The tubular structure of the testis of a male is shown, with spermatozoa (SG) and spermatocytes (SC) more peripherally located, and spermatozoa (SZ) located centrally. A tubule wall is marked with an unlabelled arrow. Scale bar = 50 pm. (E) General configuration of the testis of a male with evidence of previous hermaphroditism is shown: infolding epithelium (E), thick outer wall (W), gonadal lumina (L), and presumed atretic oocytes (A). Scale bar = 200 pm. (F) The same fish as in Fig. 3E, with pale yellow translucent material (YT) composing part of the presumed atretic follicle (A). The epithelial lining (E) and gonadal lumen (L) are also labelled. Scale bar = 150 pm.


2346

C A N . J . ZOOL. VOL. 70. 1992

serve as repositories (formed by the breakdown and confluence of a number of "lobules") and cease to assume a major spermatogenic role. Eventually large parts of the organ can be converted into a tube filled with motile sperm, as in Anthias squamipinnis (Shapiro 1981). Both of these species have unrestricted testes according to Grier (1981), yet the testes of R. marmoratus males have much in common with them. In general, primary and secondary males were indistinguishable histologically. The testes of secondary males have been said to retain the oviduct (common genital sinus) from the hermaphroditic gonad (Harrington 1967). In Coryphopterus personatus, a protogynous hermaphrodite, the testes obliterate any evidence of the ovarian lumen (Cole and Robertson 1988). This also seemed to be the case in R. marmoratus. Harrington (197 1) reported that gonadal lumina became filled by testicular tissue but were still discernible. Early oocytes and atretic follicles in secondary males, including brown atretic bodies (Sadovy and Shapiro 1987), were also described in R. marmoratus (Harrington 1967). However, some gonochoristic species have yellow -brown bodies in the testes, and some sex-changing species lack these bodies (Sadovy and Shapiro 1987; Grier 1987). Oocytes and atretic material may eventually disappear in secondary males of R. marmoratus as in mature males of the sex-changing red grouper (Moe 1969). Our failure to distinguish primary and secondary males may result from our limited sample, but we cannot support the assertion that this distinction is easy (Harrington 1967). Unequivocal evidence of secondary males requires proof of previous ovarian function, an unattainable measure for wild-caught fish. Age may be involved in testicular proliferation and ovarian regression, since most of our transitional fish were older (i.e., > 30 months). However, many of our older hermaphrodites did not show elevated testicular ratios and some younger fish were transitional or secondary males, so the relationship is neither necessary nor just a biproduct of senescence. Fish maintained in our laboratory at constant temperature (26°C) and photoperiod can become secondary males (Soto 1988). The conditions related to the production of secondary males remain unclear, as does the role of males in reproduction in the species (Soto 1988).

Acknowledgements We thank W. P. Davis for supplying fish and for providing invaluable assistance in the field, suggestions and constructive criticisms on the care and culture of this species, and comments on the manuscript. Thanks are extended to D. W . Stanley and H. Swatland for the use of the microscope image analysis system, E. D. Bailey, A. L. A. Middleton, and V. G. Thomas for the use of environmental chambers, J. C. George for the use of the electron microscope facility, and L. Lin for the preparation of the material for electron microscopy. C .G. S . was supported by a postgraduate fellowship from the Natural Sciences and Engineering Research Council of Canada (NSERC). Support to J.F.L. and D.L.G.N. was provided by operating grants (A6974 and A698 1, respectively) from NSERC. J. P. Bogart, K. S. Cole, J. C. Roff, and Y. Sadovy provided advice and constructive criticism. Atz, J. W. 1964. Intersexuality in fishes. In Intersexuality in vertebrates including man. Edited by C. N. Armstrong and A. J. Marshall. Academic Press, New York. pp. 145-232. Begovac, P. C., and Wallace, R. A. 1987. Ovary of the pipefish, Sygnathus scovelli. J. Morphol. 193: 1 17 - 133. Burke, M. A., and Leatherland, J. F. 1984. Seasonal changes in

testicular histology of brown bullheads, Ictalurus nebulosus LeSuer. Can. J. Zool. 62: 1185-1194. Cole, K. S., and Robertson, D. R. 1988. Protogyny in the Caribbean reef goby, Coryphopterus personatus: gonad ontogeny and social influences on sex-change. Bull. Mar. Sci. 42: 317-333. Davis, W. P., Taylor, D. S., and Turner, B. J. 1990. Field observations of the ecology of mangrove rivulus (Rivulus marmoratus) in Belize and Florida (Teleostei: Cyprinodontiformes: Rivulidae). Ichthyol. Explor. Freshwaters, 1: 123- 134. Dixon, W. J., Brown, M. B., Engleman, L., Frane, J. W., Hill, M. A., Jennrich, R. I., and Toporek, J. D. 1981. BMDP Statistical Software 1981. University of California Press, Berkeley. Down, N. E., and Leatherland, J. F. 1989. Histopathology of gonadal neoplasms in cyprinid fish from the lower Great Lakes of North America. J. Fish Dis. 12: 415-438. Fischer, E. A. 1981. Sexual allocation in a simultaneously hermaphrodite coral reef fish. Am. Nat. 117: 64-82. Forberg, K. G. 1982. A histological study of development of oocytes in capelin, Mallotus villosus villosus (Muller). J. Fish Biol. 20: 143 - 154. Grier, H. J. 1981. Cellular organization of the testis and spermatogenesis in fish. Am. Zool. 21: 345 -357. Grier, H. J. 1987. Brown bodies in the gonads of the black sea bass, Centropistes striatus. In Proceedings of the Third International Symposium on Reproductive Physiology of Fish, St. John's, Nfld., August 2-7, 1987. Edited by D. R. Idler, L. W. Crim, and J. M. Walsh. Marine Sciences Research Laboratory, St. John's, Nfld. p. 199. Grier, H. J. Fitzsimmons, J. M., and Linton, J. R. 1978. Structure and ultrastructure of the testis and sperm formation in goodeid teleosts. J. Morphol. 156: 419 -438. Grier, H. J., Linton, J. R., Leatherland, J. F., and De Vlaming, V. L. 1980. Structural evidence for two different testicular types in teleost fishes. Am. J. Anat. 159: 331. Harder, W. 1975. Anatomy of fishes. Schweizerbartsche Verlagsbuchhandlung, Stuttgart. Harrington, R. W., Jr. 1961. Oviparous hermaphroditic fish with internal self-fertilization. Science (Washington, D.C.), 134: 1749 - 1750. Harrington, R. W., Jr. 1967. Environmentally controlled induction of primary male gonochorists from eggs of the self-fertilizing fish, Rivulus murmoratus Poey . Biol. Bull. (Woods Hole), 132: 174 - 199. Harrington, R. W., Jr. 1968. Delimitation of the thermolabile phenocritical period of sex determination and differentiation in the ontogeny of the normally hermaphroditic fish, Rivulus marmoratus Poey. Physiol. Zool. 41: 447 -460. Harrington, R. W., Jr. 1971. How ecological and genetic factors interact to determine when self-fertilizing hermaphrodites of Rivulus murmoratus change into functional secondary males, with a reappraisal of the modes of intersexuality among fishes. Copeia, 1971: 389-432. Harrington, R. W., Jr. 1975. Sex determination and differentiation among uniparental homozygotes of the hermaphroditic fish Rivulus marmoratus (Cyprinodontidae: Atheriniformes). In Intersexuality in the animal kingdom. Edited by R. Reinboth. Springer-Verlag, New York. pp. 249-262. Harrington, R. W., Jr., and Kallman, K. D. 1968. The homozygosity of clones of the self-fertilizing hermaphroditic fish, Rivulus murmoratus Poey (Cyprinodontidae, Atheriniformes). Am. Nat. 102: 337 -343. Harrington, R. W., Jr., and Rivas, L. R. 1958. The discovery in Florida of the cyprinodont fish, Rivulus marmoratus, with a redescription and ecological notes. Copeia, 1958: 125 - 130. Hyder, M. 1969. Histological studies on the testis of Tilapia leucosticta and other species of the genus Tilapia (Pisces: Teleostei). Trans. Am. Microsc. Soc. 88: 21 1-231. Kallman, K. D., and Harrington, R. W. 1964. Evidence for the existence of homozygous clones in the self-fertilizing hermaphroditic fish, Rivulus marmoratus Poey. Biol. Bull. (Woods Hole), 126: 101 - 114.


SOTO ET AL.

Khoo, K. H. 1979. The histochemistry and endocrine control of vitellogenesis in goldfish ovaries. Can. J . Zool . 57: 6 17 -626. Lloyd, D. G. 1980. The distribution of gender in four angiosperm species illustrating two evolutionary pathways in dioecy. Evolution, 34: 123- 134. Moe, M . A., Jr. 1969. Biology of the red grouper Epinephelus morio (Valenciennes) from the eastern Gulf of Mexico. Prof. Pap. Ser. Fla. Dep. Nat. Resour. Mar. Res. Lab. No. 10. pp. 1-95. Nakamura, M., Hourigan, T. F., Yamauchi, K., and Grau, E. G. 1989. Histological and ultrastructural evidence for the role of gonadal steroid hormones in sex change in the protogynous wrasse Thalassoma duperrey. Environ. Biol. Fishes, 24: 117 - 136. Reinboth, R. 1963. Naturlicher geschlechtswechsel bei Sacura margaritacea (Hilgendorf) (Serranidae). Annot. Zool . Jpn. 36: 173 - 178. Robb, A. P. 1982. Histological observations on the reproductive biology of the haddock, Melanogrammus aeglefinus (L.) . J . Fish Biol. 20: 397-408. Sadovy , Y., and Shapiro, D. Y. 1987. Criteria for the diagnosis of hermaphroditism in fishes. Copeia, 1987: 136 - 156. Selman, K., and Wallace, R. A. 1986. Gametogenesis of Fundulus heteroclitus. Am. Zool. 26: 173 - 192. Shapiro, D. Y. 1981. Size, maturation and the social control of sex reversal in the coral reef fish Anthias squamipinnis. J. Zool. (1965 - 1984), 193: 105 - 128. Smith, C. L. 1975. The evolution of hermaphroditism in fishes. In Intersexuality in the animal kingdom. Edited by R. Reinboth. Springer-Verlag, New York. pp. 295 - 3 10. Soto, C. G. 1988. Gonadal histology and external appearance of the

2347

self-fertilizing hermaphrodite Rivulus ocellatus. M. Sc. thesis, Department of Zoology, University of Guelph, Guelph, Ont. Treasurer, J. W., and Holliday, F. G. T. 198 1. Some aspects of the reproductive biology of perch Perca fluviatilis L. A histological description of the reproductive cycle. J. Fish Biol. 18: 359-376. van den Hurk, R., Meek, J., and Peute, J. 1974. Ultrastructural study of the testis of the black molly (Mollienisia latipinna). 11. Sertoli cells and Leydig cells. Proc. K. Ned. Akad. Wet. Ser. C , 77: 470 -476. Wallace, R., and Selman, K. 1981. Cellular and dynamic aspects of oocyte growth in teleosts. Am. Zool. 21: 325 -343. Wallace, R. A., Selman, K., Greeley, M. S., Jr., Begovac, P. C., Lin, Y.-W. P., McPherson, R., and Petrino, T. R. 1987. Current status of oocyte growth. In Proceedings of the Third International Symposium on Reproductive Physiology of Fish, St. John's, Nfld., August 2-7, 1987. Edited by D. R. Idler, L. W. Crim, and J. M. Walsh. Marine Sciences Research Laboratory, St. John's, Nfld. pp. 167- 177. Warner, R. R., and Robertson, D. R. 1978. Sexual patterns in the labroid fishes of the western Caribbean. I. The wrasses (Labridae). Smithson. Contrib. Zool. 254: 1 -27. Yamamoto, K. 1957. Studies on the formation of fish eggs. XI. The formation of a continuous mass of yolk and the chemical nature of lipids contained in it in the oocyte of the flounder, Liopsetta obscura. J. Fac. Sci. Hokkaido Univ. Ser. VI Zool. 13: 344-351. Yamamoto, K., and Yamazaki, F. 196 1. Rhythm of development in the oocyte of the goldfish, Carassius auratus. Bull. Fac. Fish. Hokkaido Univ. 12: 93 - 114.