Fecundity, the Germinal Epithelium, and Folliculogenesis in Viviparous Fishes

Harry J. Grier, Mari Carmen Uribe, Lynne Parenti, and Gabino De la Rosa •Gonadas and Gametogenesis of Viviparous Fishes 193

Fecundity,the Germinal

Epithelium, andFolliculogenesis inViviparous Fishes

Harry J. Grier1,2,Mari Carmen Uribe3,

Lynne R. Parenti1,Gabino De la Rosa-Cruz3

1Division of Fishes, National Museum of Natural History,Smithsonian Institution, Washington, DC 20560;

2Tropical Aquaculture Laboratory, Ruskin, Florida;3Lab. Biología de la Reproducción, Facultad de Ciencias,

Universidad Nacional Autónoma de México, México, DF

• Viviparous Fishes Genetics, Ecology, and Conservation194

Abstract

Reproduction in viviparous fish, and fish thatpractice internal fertilization but lay fertilizedeggs, is shown to have evolved independentlyin several taxa based on the morphology ofsperm packages, either spermatozeugmata orspermatophores. However, the origin of germcells, both male and female, is from germinalepithelia that are apparently unchanged inspite of great diversity in gonad morphology.An exception is the Atherinomorpha wherefemales have a typical ovarian germinalepithelium, but in males an epithelioidassociation between Sertoli cells and germ cellsoccurs because all of the criteria that define anepithelium are not met. Adaptations forinternal fertilization and viviparity arise fromnew functions of the cells that are common inall fish, not the participation of new cell types.Based upon the restricted distribution ofspermatogonia within testicular lobules in theAtherinomorpha, there is no support forrelationships between them and the viviparoussurfperches or mullet where spermatogoniamay occur along the walls of the lobules. As areproductive mode, viviparity results in greatlyreduced fecundity compared to fish whosereproductive mode is external fertilization andscattering of eggs. It appears that conceptsused in fisheries biology can be adapted tobetter understand viviparity.

Resumen

Se ha demostrado, con base en la morfologíadel empaquetamiento de los espermatozoides,tanto en espermatozeugmata como enespermatóforos, que la reproducción en pecesvivíparos, y en peces que presentanfertilización interna pero que depositan loshuevos fertilizados, ha ocurridoindependientemente en diversos taxa, Sinembargo, el origen de las células germinalesmasculinas y femeninas, es el epitelio germinalque, aparentemente, no ha cambiado a pesarde la gran diversidad de la morfología de lasgónadas. Una excepción son losAtherinomorpha, en los cuales las hembrasmuestran un epitelio germinal ovárico típico,pero en los machos ocurre una asociaciónepitelioide entre las células de Sertoli y lasgerminales, debido a que no se reúnen todoslos caracteres que definen a un epitelio. Lasadaptaciones para la fertilización interna y laviviparidad se originaron, no por la ocurrenciade nuevos tipos celulares, sino por nuevasfunciones de las células presentes en todos lospeces. Con base en la distribución restringidade las espermatogonias dentro de lóbulostesticulares en Atherinomorpha, no se puedeestablecer relación entre ellos y embiotócidos olisas en los cuales las espermatogonias puedenlocalizarse a lo largo de las paredes de loslóbulos. En la viviparidad, como modoreproductivo, ocurre una gran reducción de lafertilidad, comparada con la que presentan lospeces que muestran modo reproductivo confertilización y oviposición externas. Parecieraque los conceptos utilizados en biología depeces deben adecuarse para una mejorcomprensión de la viviparidad.

Fecundity,the Germinal

Epithelium, andFolliculogenesis inViviparous Fishes

Harry J. Grier1,2,Mari Carmen Uribe3,

Lynne R. Parenti1,Gabino De la Rosa-

Cruz3

• Viviparous FishesHarry J. Grier and Mari Carmen Uribe, book editors.New Life Publications, Homestead, Florida, 2005. p 193-217.


Introduction

The Teleostei is a diverse clade of aquaticvertebrates that has evolved variedmodes of reproduction. Most teleost fish

scatter gametes into the external, aqueous envi-ronment. The embryo develops within an egg insurroundings that may be considered hostile to-wards its survival. Few eggs, larvae, or juvenilessurvive to become reproductive adults.

The number of eggs shed at a single spawningis termed “batch fecundity.” In their classic re-view, “Modes of Reproduction in Fishes,” Brederand Rosen (1966) reported batch fecundity invarious fish species in an anecdotal fashion, com-mon at that time. Since then, the measurementof batch fecundity has become an integral part oflife history data used in fisheries science, inspiredby the pivotal work of Hunter and colleagues(Hunter et al., 1985; Hunter and Macewicz,1985). Egg-scattering fish produce enormousnumbers of eggs at a single spawning event (batchfecundity), spawn multiple times within a spawn-ing season (indeterminate annual batch fecun-dity), or spawn once per year (determinate annualbatch fecundity). The estimation of batch fecun-dity, its relationship to indeterminate or deter-minate batch fecundities annually and within aspawning season, establishes research imperativesin fisheries science and documents a successfulmode of reproduction (the scattering of enor-mous numbers of eggs into the environment withextremely low survivorship) that is typical for te-leosts (viz. Breder and Rosen, 1966).

Viviparity, or internal fertilization coupledwith giving birth to live young, is the mode of

reproduction for a relatively small number ofteleosts (viz. Breder and Rosen, 1966). Approxi-mately 3% of all teleost species are known to beviviparous (Callard and Ho, 1987; Wourms,2004). Viviparity has evolved independently inseveral teleost clades (viz. Parenti, 1981;Lydeard, 1993). The concepts of fecundity asused in fisheries science have never been appliedto the study of reproduction in viviparous fishes.As fisheries concepts are applied to commercialfisheries, so too might they be applied to themanagement and protection of viviparous fishpopulations, providing the means for docu-menting population status and developing man-agement plans for species conservation.

Viviparous fishes have a much lower fecun-dity, by orders of magnitude, than fish that pro-duce eggs (Table 1A,B). The enormousfecundity in fish that scatter eggs, however, isdiminished when compared to viviparous spe-cies that have evolved unique adaptations toensure internal fertilization, gestation, and birth.Our principal goal is to compare these two re-productive modes in fishes, egg-scattering andviviparity, within a framework of homology andanalogy to understand the basis of reproductivediversity and to gain further insight into theevolution of morphological diversity amongfishes. We also document some of the similari-ties and differences in gonad morphology be-tween viviparous fishes and those with externalfertilization and assimilate this information intoa structural framework to interpret currentknowledge and to guide future research.


Why Viviparity?

Estimated batch fecundity (BF) is presented forfour species of marine, perciform fish and, by actualegg count, one species of freshwater cyprinid,popular as an aquarium fish (Table 1A). Theseexamples of fecundity are contrasted against themuch lower brood fecundity of viviparous fishes inthe cyprinodontiform atherinomorph familiesPoeciliidae and Goodeidae (Table 1B). The associa-tion between a brood of young, born to a vivipa-rous fish, and batch fecundity, as the concept wasdeveloped for fish that scatter eggs (Hunter et al.,

1985; Hunter and Macewicsz, 1985), has not beenestablished previously. Herein, we establish theconcept that viviparity is a special kind of BF, spe-cifically “indeterminate batch fecundity” becauseviviparous fish give birth to multiple broods ofyoung. In viviparous fishes, BF is the actual num-ber of intra-ovarian young that are counted upondissection of, or the number of young born to, a fe-male. BF in a viviparous fish can be determinedwith accuracy in contrast to estimating the largenumbers of eggs that are produced during a spawn-ing event in most externally-fertilizing species(Table 1A).

Table 1.

Fecundity in Fish: A. Saltwater Fish that Scatter Eggs.B. Danio rerio and Viviparous Fish.

Table 1A.

Family Total Spawning Batch Annual Egg Location ReferenceSpecies Length (cm) Frequency Fecundity Production

LabridaeTautoga onitis 26-52 1.2 days 2800 to 290,000 to Chesapeake Bay White et al.

181,200 10,510,000 Virginia, USA 2002Mugilidae

Mugil cephalus 33-60 1 per annum 213,000 to 213,000 to South Carolina McDonough3,890,000 3,890,000 et al. 2003

SciaenidaeCynoscion nebulosus 4.4 days 145,452 to 3.2 to 17.6 South Carolina Roumillat &

529,976 million USA Brouwer, 2002Micropogonias 40 3-4 days 216,000 in Nov 3.3 to 7.3 Rio de la Plata Macchi et al.furnieri 96,900 in Feb. million Argentina 2002

TABLE 1B.

Family Standard Average AverageSpecies Length (mm) # Oocytes # Embryos Range Location Reference

CyprinidaeDanio rerio 33-34 668 +/– 277SE NA 257 to 1115 Aquarium fish unpublished

PoeciliidaeGambusia affinis 50 94.3 +/– 17 NA 56 to 113 Aquarium fish unpublished

PoeciliidaePoecilia reticulata 30-31 NA 17.2+/–7.02 10 to 33 Aquarium fish unpublished

GoodeidaeXenotoca eiseni 42-44 NA 25.5+/–3.43 19 to 32 Aquarium fish unpublished


Both fisheries and captive spawning data ofcommon snook, Centropomus undecimalis (fam-ily Centropomidae, order Perciformes; Nelson,1994), an inshore, marine fish that is commonin passes and around mangrove islands in thetropics and subtropics (Taylor et al., 1998), haveestablished that the spawning frequency is ev-ery 1.6 days during a protracted spawning sea-son that lasts nearly four months, beginning inMay and ending in September in Florida, USA(Taylor et al, 1998; Grier and Taylor, 1998).Moderate-sized females average an estimated280,000 eggs per spawn when induced to ovu-late using gonadotropin-releasing hormones(Neidig et al., 2000). There are no data on theBF of large females that might be expected toproduce over a million eggs per spawning event.As a rough estimate, a single common snookfemale could produce 7 to 8 x 109 eggs duringher reproductive lifetime. If larvae from only twoof the eggs grow to become reproductive, spawn-ing adults, then steady-state population dynam-ics should be maintained. The startlingconclusion from this simple example is that sur-vival from egg to adult is abysmally low. Thenumber of eggs that survive to grow into repro-ductive adults approximates 0.000,000,004%over the lifetime egg production of a femalecommon snook.

Given the estimated annual egg productionof marine fish, and BF of one freshwater fishspecies (Table 1A,B), an inference is the ex-tremely low survival of eggs, larvae that hatch,and growing juveniles for egg-scattering fishes.The probability of a “scattered fish egg” hatch-ing and the larva growing to a spawning adult isprobably not significantly different from zero,but that low probability succeeds at the popula-tion level, guaranteeing the survival of numer-ous species because of their enormous fecundity.

In contrast to the scattering of eggs into theenvironment or carrying embryos to birth, thereproductive mode of a small number of teleosttaxa involves internal fertilization and the layingof fertilized eggs, as in the atherinomorphsHoraichthys setnai (see Kulkarni, 1940; Grier,1984), Labidesthes sicculus (see Grier et al., 1990),and in select ostariophysans (Burns andWeitzman, 2004). A fine distinction betweeninternal fertilization and internal gametic asso-ciation needs to be recognized. Internal gameticassociation is when sperm and eggs associate inthe ovarian lumen, but eggs are not fertilized untilthey are laid and sperm are activated by water(Burns and Weitzman, 2004). Internal gametic

association may be suspected when it is knownthat an isolated female may lay eggs that subse-quently develop, yet developing eggs are notfound in the ovarian lumen where sperm arepresent. In the atherinopsid, Labidesthes sicculus,internal fertilization is known to occur becauseeggs with developing embryos were observed inthe ovarian lumen (Grier, et al., 1990). BF esti-mates and spawning frequencies are not availablefor fishes with these modes of reproduction. BFmay not be much different from those of fishthat scatter eggs, after accounting for size varia-tion in spawning fish, however.

In viviparous fish, BF might be expressed asthe number of young in relationship to either thestandard length (SL), the total length (TL) (ex-pressed in mm or cm), or the weight (gms) of thefemale times 100, as with the gonadosomatic in-dex (GSI). Insofar as the gonad might represent35% or more of the total weight of a viviparousfemale prior to giving birth, but a much lowerpercentage after birth or early in embryonic de-velopment (Grier, unreported), gonad weightshould be subtracted from total weight if BF isbased upon female weight. The measurement ofBF as a ratio (based on length or weight) mightallow intraspecific fecundity estimates within apopulation or interspecific fecundity estimatesbased on real species differences:

BF = Number of young x 100or BF = Number of young x 100

SL or TL Total weight-Gonad weight

There are no data on BF in viviparous fishfrom natural populations as they may relate tochanges during an annual reproductive cycle, bysize, by age, or by way of comparisons to eggproduction in fish that have external fertiliza-tion. The data presented in Table 1B come fromdata on 25 captive fish of each species, at thespecific sizes indicated. The poeciliids, Poeciliareticulata, the Guppy, and Gambusia affinis, theWestern Mosquitofish, were maintained inaquaria on a long photoperiod (16 hours of lightper day) at 26 to 28 oC. The goodeid, Xenotocaeiseni, was maintained in an outdoor, plastic-lined pool under conditions of natural daylength and temperatures prevailing during thespring and summer in Florida. The Xenotocadata are derived from a group of one-year oldfish in a size range of 42-44 mm SL. Fecundityinformation for other goodeids includeChapalichthys encaustus in which there were sel-


tween 400 and 800. BF in the smaller Guppy iseven lower when compared to the productioneggs in fish species with external fertilization(Table 1A, B). Nevertheless, this does not pre-clude high fecundity in guppies, long known byaquarists as the “millions fish” (Sterba, 1959;Hoedeman, 1974).

Many of the Goodeidae are listed as threat-ened or endangered (Domínguez-Domínguez etal. 2004). Similar information on brood inter-vals and BF in natural populations of goodeids,or data generated from captive stocks, might beuseful for developing species management plansfor threatened or endangered viviparous fishes.Viviparity confers some obvious survival advan-tages over external fertilization and egg scatter-ing, even though there is a greatly reduced batchfecundity and annual young production.

Gonad Morphology

Males:To introduce sperm into the female reproductivetract, with or without internal fertilization fol-lowed by the laying of fertilized eggs or vivipar-ity, it is not efficient for males to release freesperm into the water, as there is little expectationthat sperm will enter the female reproductivetract. Once exposed to an external milieu, spermbecome activated and have a rather short lifespan, as demonstrated for trout (Billard et al.,1987). Without exception, fish that practiceinternal fertilization have evolved intricate mor-phological adaptations for the transfer of spermto the female reproductive tract where spermmay remain viable for an extended time period.The array of morphological adaptations include:[1] sperm transfer to the female reproductivetract by highly modified male anal fins (Greven,2004; Meisner, 2004), [2] via a genital papilla(Grier et al., 1990), [3] neuronal modificationcoupled with evolution of a coordinated behav-ior (Greven, 2004; Rosa-Molinar, 2004), [4]sperm packaging in an acellular capsule or sper-matophore, as in Horaichthys setnai (Kulkarni,1940; Grier, 1993), and [5] production of nakedsperm bundles or spermatozeugma (singular) orspermatozeugmata (plural) (vide infra; Grier,1993). In Anableps anableps and Jenynsia lineataspermatozeugmata are not formed, but freesperm are transferred into the female reproduc-tive tract through tubular gonopodia (Grier etal., 1981). In the halfbeak genus Zenarchopterus,free sperm are released into the testicular efferentducts during spermiation where they associate

Fig. 1.A lobule terminus in the testis ofthe striped mullet, Mugil cephalus,is outlined by pink-staininginterstitial cells. There is acontinuous GE, illustrated by acontinuous distribution ofspermatogonia (SG)and spermatocysts, eachcontaining spermatocytes (SC),spermatids (ST), or sperm (SP),gamete development issynchronous within spermatocysts.Sperm are also present in thelobule lumen (SP*). The clear linesbetween neighboringspermatocysts are Sertoli cellprocesses (SE), defining theborders of the spermatocysts.Bar = 50 µm

Fig. 2.During late maturation, adiscontinuous GE is observed at theterminus of the lobule in the testis ofthe striped mullet, Mugil cephalus,and spermatocysts are absent(arrows) from part of the GE; thelumen of the lobule is engorged withsperm (SP*) in this fish that spawnsonce per year. Bar = 100 µm

dom more than ten embryos, and Goodeaatripinnis in which a range of 15-60, approxi-mately 2 cm long, embryos were noted (Meyeret al., 1985), but sample size was not reported.BF from Table 1A can be estimated based onstandard length. The annual production ofyoung would be analogous to indeterminate BFin egg-scattering fish, provided that the averageinterval between broods is known. Birth fre-quency in viviparous fish is probably signifi-cantly correlated with temperature, and mayvary during a reproductive season. By the ap-plication of common parameters used in fish-eries science, important aspects of life historycould be ascertained, and meaningful compari-sons might be made pertaining to viviparousfish populations and species.

The data for large female Western Mosqui-tofish (Table 1B) can be used as a rough indica-tor to estimate BF and annual production ofyoung for this size class maintained in aquaria.Reproduction in wild G. affinis extends over sixto seven months in Florida, from February orMarch through September, and the interval be-tween broods is 21-24 days at 26 to 28 oC (Grier,unreported). A female would be expected to pro-duce approximately nine broods during a spawn-ing season. In these large female G. affinis (Table1B), the number of young produced during aspawning season would range approximately be-

SP*

SGSP*

SESC

SPST

SG

SG


and form elongated spermatozeugmata (Fig. 11;Grier and Collette, 1987). In Zenarchopterusdispar, the morphology of spermatozeugmata isnot determined by the associations of Sertolicells and germ cells that form within the sperm-atocysts, as in the Poeciliidae and the Goodeidae(vide infra). It is to be expected that fish practic-ing internal fertilization will use different mech-anisms to package sperm for transfer to thefemale reproductive system, more of whichlikely remain to be discovered.

Compartmentalization characterizes the ver-tebrate testis, whether or not the reproductivemode is viviparity. Gametes are produced withinthe germinal compartment. Within the inter-stitial compartment are found hormone-secret-ing Leydig cells, myoid cells primarily coursingover the surface of the germinal compartmentsand separated from them by a basement mem-brane (Grier 1993), blood supply, and immunesystem cells, most notably eosinophilic granu-locytes in fishes. Initial attempts to identify ho-

mologous cell types in some fish testes led toerroneous interpretations and confusion. For atime, it was believed that some fish (Esox luciusand Salmo salar [see Marshall and Lofts, 1956])had lobule boundary cells as Leydig cell ho-mologs. This misinterpretation resulted fromuse of techniques with inherent low resolution–Sudan staining to demonstrate lipids in frozensections. Upon ultrastructural examination, itwas revealed that the lipids in the testes of Esoxlucius were not within the cytoplasm of a Leydigcell homolog in the interstitial compartment ofthe testis. The lipids were in the cytoplasm ofSertoli cells located within the germinal com-partment. In Esox niger, however, lipids were notobserved within Sertoli cells. The presence orabsence of lipids in the testes of Esox speciesunderscores reproductive differences within thegenus, and also reflects on reproductive diversityof other fish. It is not possible for an interstitialcell type, the Leydig cell, to be homologous witha cell type, the Sertoli cell, in the germinal com-

Fig. 3.The periphery of a spermatocyst inthe testis of the poeciliid,Xiphophorus helleri. Spermatids,their nuclei with condensedchromatin (n), are associated withSertoli cells (SE). Each spermatid isforming a residual body (fRB) inwhich the trailing cytoplasm has aGolgi (g) apparatus, mitochondria(m), in what will be the midpiece,and trailing flagella (f). Sertoli cellshave an oval nucleus with a well-developed nucleolus (nu), and aGolgi apparatus (g). Thespermatocyst, within the germinalcompartment of the testis, isseparated from the interstitium by abasement membrane (BM).Bar = XX µm

Fig. 4.The periphery of a spermatocyst inthe testis of the poeciliid,Xiphophorus helleri. Residual bodies(RB) are in the lumen of thespermatocyst. Sperm (SP) nuclei (n)are embedded in a region of Sertolicell (SE) cytoplasm that is devoid oforganelles such as granularendoplasmic reticulum (ger) andlysosomes (ly). The sperm midpieceof composed of a flagellum (f)surrounded by mitochondria (m). TheSertoli cell nucleus has a well-developed nucleolus (nu). At thisstage of sperm maturation, thebasement membrane is becomingmulti-layered. Bar = XX µm

partment of the testis. These compartments arealways separated by a basement membrane asthey are in the ovary (vide infra).

The two-compartment arrangement in ver-tebrate testis morphology is maintained in thewidely divergent “shapes” of germinal compart-ments between taxa, but not the compositionof cells within them. Between vertebrate taxa,the shapes of the germinal compartments vary.From the most primitive vertebrates to the most

I

g

f

fRB

f

gm

n

nuSE

BM

advanced, testis structure is hypothesized tohave evolved through polyspermatocystic (hag-fish, lampreys, and elasmobranchs), to anasto-mosing tubular (lower fishes such as gar, Amia,and sturgeons), and lobular polyspermatocystic(higher fishes and extant amphibians) types(Grier, 1993). It is only in reptiles, birds, andmammals that the spermatocyst is not the unitof testicular function in which sperm mature

l ySE

BM

n u

ger

SP

m

f

RB

n

*


(Callard, 1991). Anastomosing tubular testesoccur in basal teleosts. Higher teleosts possesslobular testes (Grier, 1993; Parenti and Grier,2004) in which the germinal compartments endblindly at the testis surface (Figs. 1, 2). The phy-logenetic transition between anastomosing tu-bular and lobular testes has not yet been clearlyidentified, although Parenti and Grier (2004)proposed the lobular testis as diagnostic of theNeoteleostei. So far as is known, viviparity oc-curs in an array of neoteleost taxa, all of whichhave lobular testes (Lydeard, 1993; Parenti andGrier, 2004).

A second character of the vertebrate testis isphagocytosis of residual bodies that are cast offby maturing spermatids, as in the poeciliidsXiphophorus helleri (Figs. 3-5) and Poecilialatipinna (Fig. 6), and the goodeid, Xenotocaeiseni (Fig. 7). Phagocytosis of residual bodiesdesignates Sertoli cell homology between verte-brate taxa (Grier, 1993). Other characters thatmight denote homology of synthesis are andro-gen binding protein and formation of a blood

testis barrier, but sufficient comparative infor-mation is lacking.

The presence of a basement membrane is oneof the defining characters of an epithelium, agerminal epithelium (GE) with both a somaticcell type(s) and a germ cell type(s). Both malesand females have germinal epithelia that areseparated from interstitial or stromal tissues,respectively, by the basement membrane. It isimportant to recognize basic, homologous struc-tures and similar cell functions in gonad mor-phology as these can be used to establish acomparative nomenclature (Grier, 1993) thatunderlies the “unifying concept,” (Grier, 2000)wherein the nomenclature and definitions ap-plied to homologous structures in vertebrategonads are identical.

Spermatozeugmata and spermatophores ininternally fertilizing fishes form because ofmodified associations between Sertoli cells andgerm cells that do not occur in externally fertil-izing fishes. Their formation does not includethe evolution of new cell types, but altered func-

Fig. 5.The periphery of a spermatocyst inthe testis of the poeciliid,Xiphophorus helleri, in whichphagocytosis and digestion ofresidual bodies (RB) are illustrated.Sertoli cell processes are observedto be engulfing one residual body(asterisk) while a second residualbody (RB) is within the Sertoli cellcytoplasm. Other phagocytizedresidual bodies have becomedigestive vacuoles (DV), all withelectron dense breakdown products(arrow heads). The Sertoli cell has awell-developed nucleus with a singlenucleolus (nu) and cytoplasmiclysosomes (ly). One sperm is alsoobserved to be in the process ofdegradation, surrounded bymembranous lamellae (l). Spermnuclei (n) are embedded in thecytoplasm of the Sertoli cell (SE).Their midpieces (mp) and flagella (f)extend into and occupy the center ofthe spermatocyst. At this stage ofdevelopment, just beforespermiation, the basementmembrane has become multi-layered. Bar = XX µm

Fig. 6.The periphery of a spermatocystfrom the testis of the poeciliid,Poecilia reticulata. Sertoli cellprocesses (SE) are observed as theyenvelope sperm nuclei (n). Theseprocesses are those that are devoidof organelles in the previousmicrographs. Sperm midpieces (mp)and flagella (f) extend into thespermatocyst. Spherical residualbodies (RB) are observed among theflagella. Bar = 2 µm

BM

DV

nuSE

n

mp ly

I

RB

f

*

➤

➤

➤

➤

➤

n

f

RB

mp

SEn


tion and morphology of the same cell types thatare found in all fish testes. The associations be-tween Sertoli and germ cells, correlated withsperm packaging, have evolved independentlyin different taxa. For example, in the Poeciliidae,elongating spermatid nuclei become associatedwith Sertoli cells (Figs. 3-5). Within the effer-ent and sperm ducts, spermatozeugmata areobserved where their surface is composed ofsperm nuclei (Fig. 8). In the Goodeidae (Fig. 7;Grier et al., 1981), spermatid flagella becomeassociated with the Sertoli cells, and sperm fla-gella course over and form the surface ofspermatozeugmata (Fig. 9). Both of thesecyprinodontiform atherinomorph families(sensu Parenti, 1981) have evolved viviparityindependently as reflected in the formation andmorphology of spermatozeugmata, and as alsoreflected by intrafollicular (Poeciliidae) versusextrafollicular gestation (Goodeidae, Fig. 10),embryonic adaptations (Turner, 1933; 1937;1940), and geographic isolation (Poeser, 2004;Contreras-McBeath, 2004).

Fig. 7.The periphery of a spermatocystfrom the goodeid, Xenotoca eiseni. Aphagocytized residual body (RB) anddigestive vacuoles (DV) are visiblewithin Sertoli cell (SE) cytoplasm. Abasement membrane (BM)separates the Sertoli cell from theinterstitium. Within thespermatocyst, a sperm (SP) is visibleand their flagella (f) associate, evenat the distal ends (asterisks), withthe Sertoli cell plasmalemma. Spermnucleus (n), mitochondria (m) in thesperm midpiece. Bar = 2 µm

Fig. 8.Spermatozeugmata from the testis ofthe poeciliid, Poecilia latipinna, havea surface composed of spermnuclei. Bar = 10 µm

Fig. 9.Spermatozeugma from the testis ofthe goodeid, Xenotoca eiseni, have asurface composed of sperm flagella(f). Bar = 4 µm

f

SE

f DV

RB

BM

f

*

*

n

mSP


Spermatozeugmata are not formed at spermia-tion, however, but individual sperm are releasedinto the efferent ducts. Spermatozeugmata, simi-lar to those observed in other poeciliids, are alsoobserved in the testes of Tomeurus gracilis (Grierand Parenti, unreported).

Independent evolution of internal fertiliza-tion within the Atherinomorpha is also re-flected in the halfbeak Zenarchopterus species(Grier and Collette, 1987) and the ricefishHoraichthys setnai (Kulkarni, 1940; Grier,1984). In Zenarchopterus dispar, sperm arevoided into the efferent ducts after which theybegin to associate into, and form, elongatedspermatozeugmata (Fig. 11) whose surface con-sists primarily of sperm flagella (Fig. 12). Themorphology of spermatozeugmata varies some-what between species within the genus, butthey always form in a similar fashion, hypoth-esized to be a synapomorphy for the genus(Grier and Collette, 1987). In Zenarchopterusdispar, a secretory product is observed withinthe testis ducts (Fig. 11). The secretion appar-

Within the Atherinomorpha, internal fertili-zation, with or without viviparity, is inferred tohave evolved independently within several otherfamilial groups including: Hemiramphidae(Grier and Collette, 1987; Meisner, 2004),Adrianichthyidae (Kulkarni, 1940; Grier, 1984;Rosen and Parenti, 1981), Anablepidae (Grieret al., 1981), Poeciliidae genus Tomeurus (Rosenand Bailey, 1963), and Phallostethidae (Grierand Parenti, 1994), as reflected by dissimilarmorphology. Within the developing spermat-ocysts of Anableps anableps, Anableps dowi, andJenynsia lineata (Anablepidae), sperm associatewith Sertoli cells in similar fashion to that ob-served in the Poeciliidae. Anablepids possesstubular gonopodia, however, which apparentlyobviates the need to package sperm. Anablepsdowi is the only one of these three species notedabove that forms “partial spermatozeugmata”(Grier et al., 1981), whereas the other two spe-cies release free spermatozoa into the efferentducts. Associations between germ cells and Ser-toli cells form, as in the poeciliids (Figs. 3-6).

Fig. 10.The testis from the hemiramphid,Zenarchopterus dispar, showing thestructure of lobules (L) inrelationship to the efferent ducts(ED). As with all atherinomorphs,spermatogonia (SG) are restricted tothe distal termini of the lobules thatterminate blindly at the testisperiphery. Extending proximally,spermatocysts containingspermatocytes (SC), and spermatids(ST) are observed. Free sperm arevoided into the efferent ductswhereupon they coalesce to formspermatozeugmata (SZ), observed inthe efferent and main (MD) ducts.Modified from Grier and Collette,1987, used with permission fromCopeia. Bar = 100 µm

Fig. 11.Spermatozeugma from thehemiramphid, Zenarchopterus disparhave a surface composed of parallelarrays of sperm flagella (f). Wherethese arrays separate, sperm nuclei(n) are observed. Bar = 10 µm

SG

SC

LST

S

SZ

ED

MD

SZ

f

f

n


ently originates from the efferent duct cells thatdevelop from hypertrophied Sertoli cells atspermiation as in other atherinomorphs (Grier,1993). Again, histology indicates that Sertolicells in the Atherinomorpha transform intoefferent duct cells, becoming columnar andsecretory. In the poeciliids, the efferent ductsare filled with a periodic acid Schiff-positivesecretory product (Fig. 14) that permeates thespermatozeugmata. This secretion originatesfrom efferent duct cells, as in Poecilia latipinna(van den Hurk and Barends, 1974) and Brotulamultibarbata (Fig. 18b). In the Atherino-morpha, a special case of sperm packaging oc-curs in H. setnai, the only species in which truespermatophores are formed (Kulkarni, 1940).The spermatophore capsule is formed by thesecretion that is produced by hypertrophiedSertoli cells (Figs. 15a, b, c; Grier, 1984). InH. setnai, the Sertoli cell secretion condensesaround the sperm within a spermatocyst toform a spermatophore. The hypertrophy ofSertoli cells may be one more character that

supports atherinomorph monophyly, the mostcogent being the restricted distribution of sper-matogonia to the distal termini of the lobules(Parenti, 2004; Parenti and Grier, 2004).

The Sertoli and germ cells compose an epi-thelium, specifically a germinal epithelium(GE), in most taxa (Grier, 2000; Grier and LoNostro 2000). Association of the cells meets thecriteria of an epithelium. Cells that compose theGE: 1) are connected laterally by cytoplasmicinterdigitations, desmosomes, and tight junc-tions (Grier, 1993, 2000), 2) rest upon a base-ment membrane by histological definition (Figs.3-5, 7), 3) border a body cavity (a lumen), and4) are avascular.

A testicular GE occurs in most fish, includ-ing internally fertilizing fish (Figs. 17a, 18a),the exception being the Atherinomorpha wherethere is no lumen within lobules (Fig. 11,12;Parenti and Grier, 2004). Rather than a lumen,Sertoli cell processes, forming the borders ofspermatocysts, “bridge” the width of lobules.Therefore, the third defining character of an

Fig. 12.In the “epithelioid” arrangement ofSertoli cells (SE) and germ cells,here spermatids (ST) and sperm(SP), spermatocysts extend acrossthe width of a lobule. There is nolumen. The structure ofspermatocysts, as being composedof synchronously-developing germcells surrounded by thin Sertoli cellprocesses, is clearly evident in thetestis of the hemiramphid,Zenarchopterus dispar. Boundarycells (B), sperm nuclei (n), spermflagella (f), Bar = 10 µm

Fig. 13.In the fundulid, Fundulus grandis,freeze fracture admirably illustratesSertoli cell processes (arrow heads),encompassing single spermatocysts,which span the width of the lobule.Between the curved arrowheads, aSertoli cell process does not quitspan the width of the lobule, andspermiation, release of sperm intothe efferent ducts (ED), is beginning.As in all atherinomorph testes,spermatogonia (SG) are restricted tothe testis periphery, here being ahomogeneous mass as the fracturetechnique only differentiatesbetween cells or tissues and theintercellular space around them.Spermatocytes (SC), spermatids(ST), sperm (SP). Bar = 20 µm

B

f

ST

n

SP

SE

B

SC

SG

SP

ST

ED

➤➤

➤

➤

➤


i.e., the cells involved, the Sertoli and germ cellsthat will produce sperm, do not border upon alumen. Prior to the initiation of meiosis in co-bia, a lumen begins to develop within thesecords and, thus, the initial epithelioid associa-tion of Sertoli and germ cells becomes a GE.An epithelioid arrangement of Sertoli and germcells does not transform into a GE in the Ather-inomorpha.

Spermatozeugmata are formed in some non-atherinomorph fish taxa. The Embiotocidae,or surfperches, comprise a family of some two-dozen species of viviparous fishes inhabitingthe north Pacific coast of western NorthAmerica, Japan and Korea (Nelson, 1994). Itwas suggested that surfperches share a restrictedspermatogonial testis type (Grier et al., 1980)with the Atherinomorpha, a testis type notfound in putative atherinomorph sister groups(Parenti and Grier, 2004). The purported ho-mology of testis structure between theembiotocids and atherinomorphs is rejectedupon examination of the maturation process,below. Unfortunately, specimens available tous precluded sufficient histology to describetestis maturation from a single species throughan annual reproductive cycle. It was necessaryto “reconstruct” a process of testicular matura-tion by examination of fixed material from dif-ferent species in which males were in differentreproductive condition.

During early maturation in the embiotocidCymatogaster aggregata (Fig. 16a), the testis isatherinomorph-like: 1) spermatogonia are re-stricted to the distal termini of the lobules. 2)There is a distal to proximal distribution of ear-liest and progressively later stages of developingsperm. 3) Sertoli cell processes bridge the widthsof lobules; there is an epithelioid associationbetween Sertoli and germ cells. As testis matu-ration progresses, as in Embiotoca jacksoni, 1)spermatogonia are not observed at the distal ter-mini of the lobules (Fig. 16b). 2) The progres-sion of developing sperm, from earlier to laterstages of maturation, also breaks down. Earlierstages of maturing sperm may occur both dis-tally and proximally within lobules: the walls oflobules are lined with spermatocysts with later-developing stages randomly distributed withearlier stages of maturation. The depicted dis-tribution of developing sperm (Fig. 16b) canonly occur when spermatogonia are locatedalong the walls of the lobules and do not entermeiosis simultaneously, quite unlike any knownatherinomorph but typical of other neoteleosts

epithelium, above, is not met. Without a cen-tral lumen, although satisfying the other crite-ria that define an epithelium, the Sertoli andgerm cells in atherinomorph fish must be con-sidered to have an epithelioid cellular associa-tion (Parenti and Grier, 2004). Hypothetically,a simple rearrangement in the association ofSertoli cells and germ cells, at the time whenspermatocysts form, probably restricts sper-matogonia to the distal termini of lobules inthe Atherinomorpha as in both Zenarchopterus(Fig. 11) and in Fundulus (Fig. 12), Sertoli cellcytoplasm extends across the lobules even atthe testis periphery, and no lumen forms.

The term “epithelioid” was first applied tothe fish testis to describe annual gonad matura-tion classes in cobia, Rachycentron canadum(Brown-Peterson et al., 2002), a large, pelagicperciform fish in the family Rachycentridae(Nelson, 1994). During the regression class incobia, cords of Sertoli and germ cells (sper-matogonia) grow from the distal ends of the lob-ules, extending the lengths of lobules. A centrallumen exists within the lobules, but not thesedeveloping cords of tissue. All of the criteria nec-essary to define an epithelium are not present,

Fig. 14.A spermatozeugma (SZ) from thepoeciliid, Cnesterodondecemmaculatus, is surrounded byperiodic acid Schiff-positive(magenta) secretion that permeatesinto its center. Its circumference iscomposed of sperm nuclei (n).Bar = 10 µm

SZ

n


(following Parenti and Grier, 2004). 3) Sertolicell processes are no longer observed to extendacross the diameters of lobules. An epithelioidSertoli and germ cell association has vanishedfrom the lobules. 4) During spermiogenesis,sperm become organized as a spermatozeugmawithin a spermatocyst prior to being releasedinto a developing central lumen within the lob-ule. In all atherinomorphs examined, spermato-zeugmata are released into the testis ducts, andthere is no continuous lumen within the lobulethat connects to the efferent ducts. These aredistinctive differences in testis morphology be-tween the Atherinomorpha and other fishes(Figs. 17a,b, 18a) that achieve internal fertiliza-tion, that have only been described recently(Parenti and Grier, 2004).

By analogy, and as in cobia testis cyclingduring the annual reproductive season (Brown-Peterson et al., 2002), an epithelioid arrange-ment of Sertoli and germ cells becomes a GEas in the surfperches, a proposed example ofconvergent evolution or homoplasy. Our his-tology indicates that Sertoli cells in surfpercheshypertrophy after spermiation and secrete a pe-riodic acid Schiff-positive product, another in-ferred homoplasy as this also characterizes theAtherinomorpha. The periodic acid Schiff-positive secretion observed between sper-matozeugmata in the efferent ducts insurfperches (Figs. 16d, e) also occurs in thePoeciliidae (Fig. 14), and in the ophidiiformsDinematichthys sp. (USNM 338466; Fig.17a,c) and Brotula multibarbata (USNM214124; Fig. 18a). At the end of the surfperchreproductive season, there is a nearly total lossof germ cells from within lobules, as in theembiotocid Phanerodon furcatus. Occasionally,a spermatozeugma is observed within a sper-matocyst, and the germ cells are scattered, asspermatogonia (Fig. 16c), along the wall of thelobules being interspersed with columnar,secretory Sertoli cells. The interpretation of theGE in P. furcatus is that it has become discon-tinuous, as originally observed and defined inCentropomus undecimalis during mid and par-ticularly late maturation (Grier and Taylor,1998). A discontinuous GE is a special casewhere, during maturation, a Sertoli cell epi-thelium occurs in which there are scatteredgerm cells (Grier and Taylor, 1998; Taylor etal., 1998). The surfperch testis has character-istics of the restricted type of the ather-inomorphs. By examination of the maturationprocess through the year, however, its non-

Fig. 15a-c.[a] Two spermatocysts containingspermatids from the testis of theadrianichthyid, Horaichthys setnai.The Sertoli cells have squamousnuclei (arrows). [b] During latespermiogenesis, the spermatidshave condensed chromatin, andSertoli cells have transformed intocuboidal cells with spherical nuclei.[c] Upon completion ofspermiogenesis, Sertoli cells aretransformed into columnar cells andsecrete a spermatophore capsulesurrounding the sperm.Spermatophore capsule (C), Sertolicell nuclei (SEn), spermatids (ST),sperm (SP). Bar = 10 µm

SEn

C

SP

ST

SEn

SEn

ST


atherinomorph characteristics are clear. Theprocess of testis maturation is not well docu-mented here, indicating that further workneeds to be done with multiple species to de-tail the process.

Armed with a new definition –epithelioid–to describe the distribution of germ and Sertolicells, the surfperch lobule appears to be an epi-thelioid assemblage of germ and Sertoli cellsearly in the maturation process, similar to theAtherinomorpha. As in other neoteleosts, how-ever, (viz., Grier, 1993; Parenti and Grier, 2004),there seems to be a continuous loss of sper-matogonia within testis lobules during matura-tion, leaving them without germ cells, asidefrom scattered spermatogonia. A discontinuousGE develops as in Centropomus undecimalis(Grier, 1993; Grier and Taylor, 1998) and inthe striped mullet, Mugil cephalus (compare Figs.1,2). Surfperch Sertoli cells, lining the lobulesafter germ cells are released as sperm, form asecretory, hypertrophied epithelium –as in theatherinomorphs (Figs. 10,13,15). Hypertrophyof Sertoli cells does not occur in other non-atherinomorph neoteleosts that scatter eggs.This is yet another example in which the func-tion of Sertoli cells becomes modified coinci-

Fig. 16a-c.Lobule maturation in embiotocidtestes. [a] Early maturation inCymatogaster aggregata,spermatogonia reside at the distaltermini of the lobules; spermatocytesand spermatids are more proximal,while sperm are located closest tothe efferent ducts. Spermatocystsstretch across the lobule width, andthere is no lumen. Bar = 50 µm.[b] Later in development, in the testisof Embiotoca jacksoni,spermatogonia are no longerobserved at the distal termini oflobules, spermatocysts withdeveloping sperm in different stagesof maturation are randomly observedalong the lobule wall, and sperm,organized in spermatozeugmata,within a lumen that has developedafter spermiation begins. Bar = 50 µm.[c] In late maturation of the testis inPhanerodon furcatus, the lobulesare devoid of developing sperm. Allthat exists of germ cells arescattered spermatozeugmata withinspermatocysts or within the lobulelumina. Occasional spermatogoniaare among columnar Sertoli cells.Bar = 50 µm.

SG

SC

ST

SP

SZ

SG

ST

SCSZ


dent with the evolution of internal fertilizationand sperm packaging. Within the testis ductsof embiotocids, spermatozeugmata are sur-rounded periodic acid Schiff-positive secretion(Figs. 16d,e), but this does not penetrate intotheir centers as it does in the poeciliid,Cnesterodon decemmaculatus (Fig. 14).

Ophidiiformes comprise a group of largelymarine fishes that have truncated bodies and,in some families, internal fertilization. Spermat-ozeugmata are formed within the spermatocystsand released into the lobule lumina where theyare surrounded by a periodic acid Schiff-posi-tive secretion, as in Dinematichthys sp. (Figs.17a-c) and Brotula multibarbata (Figs. 18a-c).The secretion is formed by Sertoli cells (Fig.18b) and is observed within their cytoplasmwhen surrounding a spermatozeugma within aspermatocyst. The periodic acid Schiff-positiveglycoprotein was misinterpreted as the coveringof a spermatophore (Nielsen et al., 1968), but itobviously becomes a secretion surrounding allof the spermatozeugmata within the testis ducts.The periodic acid Schiff-positive secretion is ob-served in the testis ducts of these internally-fer-tilizing fishes. It is analogous to the secretionsobserved in certain atherinomorphs and in sur-fperches. No putative atherinomorph relativehas the atherinomorph testis type (see Parenti,2004), although we consider the survey of testistypes among teleost fishes to be preliminary(Parenti and Grier, 2004).

Females: As in males, there are modificationsin the female reproductive tract of viviparousfishes. As noted above, both intrafollicular(Poeciliidae) and extrafollicular (Goodeidae, Fig.10) embryo gestation occurs. In the goodeidSkiffia bilineata, there are marked changes in theovarian GE during gestation (Mendoza, 1943).The epithelial cells become secretory, and a newfunction associated with viviparity has evolved.These changes in the epithelium are accompa-nied by an increase in vascularity and swelling inthe stroma (ovigerous folds) during gestation. Aswith other morphological changes associated withviviparity, new function(s) become part of thenormal cellular activities within tissues. Cell andtissue types are not created de novo. In this in-stance, the GE remains unchanged relative to itscellular composition and primary function ofproducing follicles, but the epithelial cells changedramatically during gestation. Here, we reinter-pret Mendoza’s (1940) research on the gestationalcycle in S. bilineata (as Neotoca bilineata), con-clusions that are yet to be corroborated in other

species of viviparous goodeids. The reinterpreta-tion is possible now because of the establishedrole of the GE in producing follicles.

The initial description of a GE in a femalefish, in Centropomus undecimalis (Grier,2000), adopted the same criteria that wereused to define the male GE and are the crite-ria that define an epithelium: 1) Epithelialcells are interconnected. As in males, therewere observed interdigitated, juxtaposed epi-thelial cells, tight junctions, and desmosomes.2) Epithelial cells rest upon a basement mem-brane. 3) Epithelial cells border a body sur-face, either the coelom in basal teleosts withgymnovarian ovaries or the ovarian lumen inteleosts with cystovarian ovaries. 4) The GEis avascular in females as in males. Germinalepithelia differ from all other epithelia in thatpossess both somatic and germ cells. Thesecriteria are repeated here (vide supra) to em-phasize homology and analogy between male

Fig. 16d-e.Lobule maturation in embiotocidtestes. [d] Spermatozeugmata in thetestis ducts of Cymatogasteraggregata are surrounded byperiodic acid Schiff-positivesecretion. Bar = 10 µm. [e]Spermatozeugmata in the testis ofPhanerodon furcatus aresurrounded by period acid Schiff-positive secretion. Bar = 10 µm.Spermatogonia (SG), Spermatocytes(SC), spermatids (ST), sperm (SP),spermatozeugma (SZ)

SZ

SZ


and female reproductive systems despite theirobvious morphological differences. They alsounderlie the development of the unifying con-cept (Grier and Lo Nostro, 2000) and the in-terpretation of a follicle as an epithelialderivative that is always separated from theovarian stroma by a basement membrane.

In teleosts, the ovarian GE is that epitheliumlining the surface of ovarian lamellae or “oviger-ous folds,” as they have also been called(Mendoza, 1943). In adult female fish, oogoniaare scattered within the GE. Therefore, the ova-rian GE is a “discontinuous GE,” a term firstdefined for establishing male annual reproduc-tive classes in common snook (Taylor et al.,1998; Grier and Taylor, 1998). This distinctionbetween different forms of a GE can be applied

to classify the ovarian GE, but not to establishannual reproductive classes in females as can bedone in males. This is owing to the initiation ofmeiosis and folliculogenesis as a continuous pro-cess, at least in female common snook (Grier,2000), throughout the year. Oocyte stages mustused to establish reproductive classes in femalefish. The criteria for establishing annual repro-ductive classes, based on changes in the maleGE, also cannot be applied to atherinomorphmales because the testis is an epithelioid arrange-ment of Sertoli and germ cells.

Through mitosis, oogonia maintain theirpopulation within the GE. When entering meio-sis and initiating the process of folliculogenesis,oogonia become oocytes and a continual sourceof new follicles. Each follicle in the fish ovary iscomposed of an oocyte surrounded by a layer offollicle (granulosa) cells (Grier, 2000), both celltypes being derived from the GE. Follicles areseparated from the stroma (including the thecawhich is derived from the stroma) by a basementmembrane, as are the cell nests that develop fromthe GE (Fig. 20b-d). As indicated previously(Grier, 2000; Grier and Lo Nostro, 2000), thisdefinition of a follicle in the fish ovary differs fromthose that prevail in the fish literature, and isadapted from definitions in numerous histologytextbooks. This definition of a follicle preciselyreflects its origin from an epithelium, the GE,and reflects homology throughout the vertebrates(Parenti and Grier, 2004).

Ovarian GE occur in ovaries of externally-fertilizing atherinomorphs, the killifishes Fun-dulus grandis (Fundulidae) and Gnatholebiashoignei (Rivulidae; Fig. 19a, b; Parenti and Grier,2004), and the halfbeak Hemiramphus brasi-liensis (Hemiramphidae; Fig. 19c) and in theovaries of viviparous atherinomorphs: cuatroojos, Anablep anableps (Fig. 19d), the poeciliidsPoecilia latipinna (Fig. 19e), Cnesterodondecemmaculatus (Fig. 19f ), and Poeciliopsis gra-cilis (Uribe et al. 2004), and the goodeid Ilyodonwhitei (Fig. 20, Uribe et al., 2004). At the levelof light microscopy, the ovarian GE in anotherviviparous fish, Dinematichthys sp. (Fig. 19g), isidentical to that observed in the atherino-morphs, and GE in these species are morpho-logically identical to those described in theperciform, Centropomus undecimalis by Grier(2000) and the synbranchiform Synbranchusmarmoratus by Ravaglia and Maggese (2003).

The “emerging constant” of an ovarian GEin fishes (Parenti and Grier, 2004), and an iden-tical process of folliculogenesis, is supported by

Fig. 17a-c.Testis of the ophidiiform,Dinematichthys sp. [a] Lobuleshowing the development of aspermatozeugma within aspermatocyst. Spermatozeugmataoccur within the lobule lumen, eachsurrounded by periodic acid Schiff-positive secretion. Bar = 10 µm.[b] Several spermatozeugmata areready to be released into the lobulelumen. Bar = 50 µm.[c] Spermatozeugmata within thelumen of ducts in the testis areembedded in periodic acid Schiff-positive secretion. Interstitium (I),lobule lumen (L), spermatids (ST),spermatocytes (SC), spermatozeugma(SZA), spermatozeugmata (SZ).Bar = 10 µm

SZSZ

SZA

SZ

*

L

SZ

ST


the observance of primary growth oocytes in theGE of the cyprinodontiform, Gnatholebiashoignei (Fig. 19b) and in the perciform,Centropomus undecimalis (Grier, 2000). Indeed,it is self evident that activity of the GE, withregard to mitotic activity of oogonia, transfor-mation of oogonia into oocytes upon the initia-tion of meiosis, and coincident generation offollicles during reproductive cycles in viviparousfish underlies the sequential production ofyoung and superfetation. The association and

Fig. 18a-c.Early maturation in the ophidiiform,Brotula multibarbata. [a]Spermatocysts with different stagesof sperm development, extendingfrom the periphery of the testis tothe efferent ducts.Spermatozeugmata are observedforming within spermatocysts(yellow arrows), and are within thelobule lumina and the main testisduct. [b] Periphery of a lobuleshowing periodic acid Schiff-positive secretion in Sertoli cellcytoplasm (thick arrow) andsurrounding spermatozeugmata inthe lobule lumen (*). Within theinterstitium, a capillary with redblood cells is observed. Bar = 10 µm.[c] Spermatozeugmata within theducts are surrounded by periodicacid Schiff-positive secretion. Redblood cells (RBC), sperm flagella (f),sperm nuclei (n), spermatogonia(SG), spermatocytes (SC),spermatids (ST), sperm (SP),spermatozeugmata in lobules (SZL)and in ducts (SZD)

origin of follicles from the epithelium of theovarian lamellae was reported by Mendoza(1943), describing the germinal tissue in agoodeid, and by Turner (1938) in examiningadaptations for viviparity in Anableps anableps.In their time and with the limitations of theirtechniques, the origin of germ cells in the fishovary became a phenomenon of importance toMendoza and Turner. The origin of germ cellsis likewise now becoming a topic of investiga-tion and appears to be quite similar across taxa.

SZD

SZL

RBCn

f

SZ

SZ


The pioneering observations of Mendoza(1943) on the germinal tissue in the goodeid S.bilineata, have been corroborated in Ilyodonwhitei in an extraordinary series of micrographs.The process of folliculogenesis in I. whitei is acoordinated progression in the development of afollicle (Figs. 20a-20d), identical to that describedin common snook (Grier, 2000). Oogonia residewithin the GE (Fig. 20a) and divide mitoticallyto produce cell nests that remain attached to theGE via a thin cord of cells (Fig. 20a-d). At somepoint, oogonia enter meiosis, becoming oocytes.A cell nest is composed of either a cluster of oo-gonia or of early diplotene oocytes and prefolliclecells (derived from the epithelial cells in the GE).Subsequently, primary oocyte growth beginswithin cell nests (Fig. 20b). Upon initiation ofprimary growth, an oocyte and its associatedprefollicle cells begin to be segregated from thecell nest. Separation is accomplished by the syn-thesis of an intervening basement membrane (Fig.20b) between the primary growth oocyte and theremainder of the cell nest. Eventually, the base-ment membrane will completely encompass theoocyte, forming a follicle that remains attachedto the GE (Fig. 20b, c, d). The process by whichfolliculogenesis begins from a cell nest, and thefollicle is later observed to be attached to the GE,is unknown. Prefollicle cells that are associatedwith oocytes will become follicle cells whenfolliculogenesis is completed. Mendoza (1943)noted the attachment of “full grown oocytes tothe ovarian delle,” an invagination of the GE,seen in Fig. 20c as an infolding of the GE base-ment membrane.

In Ilyodon whitei (Fig. 20a-f ), cell nests pro-trude from the GE towards the stroma but areseparated from it by an encompassing extensionof the basement membrane of the GE, i.e., acell nest is encompassed completely by the GEbasement membrane as particularly well de-picted in Fig. 20b, d. Cells within a cell nest areall derived from and form a continuum withthose in the GE. They are not separated fromthe GE by a basement membrane. In examin-ing dozens of cell nests and follicles in I. whitei,a constant morphological feature is observed:the basement membrane that subtends and sup-ports the GE always separates cell nests and fol-licles, both epithelial derivatives, from theovarian stroma, a mesodermal derivative. Theprocess of folliculogenesis must involve a con-tinued synthesis of basement membranes toencompass and maintain the separation betweencells of different origins.

Fig. 19a-d.The ovarian GE borders the lumen insix atherinomorphs (first three haveexternal fertilization and the latterthree are viviparous) and aviviparous ophidiiform fish: [a] Asingle oogonium in the GE in theovary of Fundulus grandis is flankedby prefollicle cells. The GE basementmembrane is barely visible. Beneaththe GE, a primary growth oocyte,with basophilic cytoplasm andforming cortical alveoli, is observed.Bar = 10 µm. [b] A primary growthoocyte is observed within the GE ofthe rivulid, Gnatholebias hoignei(USNM 245947). The GE rests on abasement membrane that bordersthecal cells surrounding a follicle.Columnar follicle cells and a well-developed zona pellucida areobserved along with a few corticalalveoli at the oocyte surface.Bar = 10 µm. [c] The GE in thehemiramphid, Hemiramphusbrasiliensis contains an earlydiplotene oocyte, and is subtendedby a basement membrane. A primarygrowth oocyte, with forming corticalalveoli, is observed beneath the GE.[d] A single oogonium, flanked byprefollicle cells, resides within theGE of the anablepid, Anablepsanableps, as well as a cell nestcontaining oogonia.

ca

BM

ZPOC

OL

SRBM

PF OGOL

OL

ca

PG

BM

PF

OG

CA

FCT

PGBMOL


Here, we demonstrate for the first time thatin the fish ovary, follicles maintain an attach-ment to the GE via their sharing a single base-ment membrane over a short distance of thefollicle surface (Fig. 20b-d). The shared base-ment membrane both attaches a follicle beneaththe GE and separates it from the cells compos-ing the GE. To observe this attachment, appro-priate histological sections through a folliclemust be observed. The basement membrane ofboth the follicle and the GE must be cut in crosssection, not obliquely, and well stained. Thisobservation of morphology permits an entirelynew perspective on follicle development in fish,namely, the continuum between oogonia andprocess of folliculogenesis beginning within aGE, followed by oocyte maturation within afollicle, ovulation, and formation of the postovulary follicle with the follicle and post ovula-tory follicle being attached to the GE.

In goodeids, ovulation occurs when a devel-oping embryo is voided into the ovarian lumenfrom a follicle. The single basement membrane,shared by the follicle and the GE, breaks at thetime of ovulation, a separation of the follicle cellsand the epithelial cells of the germinal epithe-lium occurs, and the egg is voided into the ova-rian lumen. The follicle cells are left behind.These, and the theca, become a postovulatoryfollicle (POF) (Fig. 21). Notably, the formerfollicle cells in a POF continue to be separatedfrom the stroma after ovulation; the POF base-ment membrane remains intact after ovulationand is continuous with that of the GE (Fig. 21).Furthermore, the epithelial cells of the GE alsoform a continuum with the former follicle cellsof the POF which were derived from the GE.Thus, and until the POF degenerates, the epi-thelial cells of the GE become reunited with thefollicle cells that were once derived from them.Throughout its development, the follicle that isproduced by the GE remains sequestered fromthe stroma, a different tissue layer, by a base-ment membrane. The separation is maintainedeven after ovulation. This observation is not dis-similar to that in males where the GE is alwaysseparated from the interstitial tissue by a base-ment membrane.

Far ahead of his time, Mendoza (1943) notedthat the flattened cells at the bottom of a “delle”offer “an attenuated and weakened place in theovigerous fold epithelium through which the eggescapes into the ovarian cavity.” With additionalknowledge, we identify that “weakened place”as the location of a single basement membrane

that is shared between the GE and the follicle.The histological techniques of Mendoza (1943)did not include a stain for basement membranes,and the concept of a GE, including the impor-tance of a basement membrane to separate epi-thelial derivatives, the follicle, from the stroma,was not arrived at until over a half century later(Grier, 2000). Mendoza (1943) incorrectly sur-mised that follicle cells were derived from the“subepithelial connective tissue.” Nevertheless,his contribution should be remembered as theorigin of follicles emerges as a new area for re-search in fish and vertebrate reproduction. In

Fig. 19e-g.[e] An oogonium is observed withinthe GE of the poeciliid, Poecilialatipinna, and primary growthoocytes reside beneath the GE. [f] Inthe poeciliid, Cnesterodondecemmaculatus, a prefollicle cellsand a single oocyte, just beginningprimary growth signified by a thinrim of basophilic cytoplasm aroundthe nucleus, protrudes into thestroma between two follicles withoocytes in different stages ofdevelopment. The basementmembrane is barely visible. Bar = 10µm. [g] The GE of the ophidiiform,Dinematichthys sp., is supported bya basement membrane. Within theGE is an early diplotene oocyte,interpreted as such because of thethin rim of basophilic material (RNA)around the inner nuclear membrane.Beneath the GE, primary growthoocytes are observed, two having awell-developed zona pellucida. Bar =10 µm. Basement membrane (BM),cortical alveoli (ca), early diploteneoocyte (DO), follicle cells (FC),ovarian lumen (OL), oogonium (OG),primary growth oocyte (PG),prefollicle cells (PF), stroma (SR),thecal cells (T), zona pellucida (ZP)

GE

OL

PG

OC

OL

BM

OC

PG

GE

PG

OG

OL


retrospect, the morphology reflects on how anoocyte always ovulates into the ovarian lumenand never into the stroma.

The results of our recent investigations arenot limited to the germinal epithelium, but in-clude also oocyte development. In all ather-inomorphs surveyed to date, yolk is fluidthroughout vitellogenesis, yet another ather-inomorph synapomorphy (Parenti and Grier,2004). In non-atherinomorphs, yolk is granu-lar and becomes fluid only during the cytoplas-mic events leading to ovulation.

Conclusions

Morphological similarity, and the uniform no-menclature that can be applied to its description,supports homology. As proposed originally(Grier and Lo Nostro, 2000), the unifying con-cept stressed unified definitions of morphologicalstructure between taxa. As such, it is recognizedthat the origin of ovarian follicles from a GE isidentical between fish and mammals (Grier,2000; Parenti and Grier, 2004). The ovarian GEis composed of both somatic cells and germ cells.In fish, the latter may be oogonia and diploteneoocytes that may advance in development as faras the beginning of primary growth. Germ cellsand follicle cells in a rabbit were described byZamboni (1972) as “breaking through” the base-

Fig. 20a-e.Folliculogenesis in the goodeid, Ilyodon whitei: [a] a singleoogonium is observed within the GE, and cell nests containingearly diplotene oocytes extend from the GE into the stroma;primary growth oocytes are also observed. Bar = 10 µm. [b]The basement membrane subtending the GE extends arounda cell nest, separating it from the stroma. Within the cell nest,one early diplotene oocyte has an enlarged nucleus. [c] Acell nest is observed extending into the stroma and attachedto the GE. The cell nest is surrounded by extension of thebasement membrane of the GE. Within the cell nest,basement membrane synthesis (large red arrow) isseparating a primary growth oocyte from the remaining earlydiplotene oocytes. An oocyte within a follicle is observed nextto the cell nest, connected to the GE, but a single basementmembrane (small red arrow) is resolved between the GE andthe follicle cells. Bar = 10 µm. [d] An oocyte and cell nest: thebasement membranes are well resolved. The follicle has aperiodic acid Schiff-positive basement membrane separatingthe cells in the GE from the follicle cells. Bar = 10 µm.[e] Magnified section from the previous micrograph showingprefollicle cells and early diplotene oocytes within a cell nest.These form a continuum with the cells in the GE. Next to thecell nest, the GE basement membrane extends to the surfaceof the follicle. A single basement membrane is resolvedbetween the follicle cells and the cells of the GE (yellowarrow). Bar = 10 µm. Basement membrane (BM), cell nest(CN), follicle cells (FC), early diplotene oocytes (DO), oocytes(OC), oogonium (OG), ovarian lumen (OL), prefollicle cells (PC),primary growth oocyte (PG), stroma (SR)

PG

SR

PC

CN

PG

OG

GE

SRBM

PG

OL

CN

PG

SG

OL

M

SC

PF

EBM S


ment membrane of the GE to gain access to andreside in the ovarian stroma. In Centropomusundecimalis (Grier 2000) and Ilyodon whitei, theprocess is organized as epithelial cells, oogonia,and oocytes partake in a highly coordinated pro-cess of folliculogenesis. The origin of follicle(granulosa) cells from the somatic cell in the GEis documented between vertebrates (Tokarz,1978) and in mammals (Zamboni, 1972). Asmore data become available for fishes and othervertebrates, the emerging constant of the GEwill become recognized as uniting homologousand essential processes (folliculogenesis, oocytedevelopment, and ovulation) into a continuumthat essentially remains unchanged throughoutvertebrate evolution. Changes in these processesmay be understood as specific modes of repro-duction in fishes and other vertebrates.

After ovulation in Ilyodon whitei, the base-ment membrane that surrounded the follicleremains attached to the basement membranesubtending and supporting the GE. Even afterovulation, the follicle cell component of POFremains separated from the stromal tissue by abasement membrane. This new observationstresses the separation between germ layers: abasement membrane separates epithelial (thefollicle is derived from an epithelium) deriva-tives from tissue in other body compartments.It has been suggested that the term “folliclecomplex” be used to describe the follicle andits surrounding basement membrane and th-eca (Grier and Lo Nostro, 2000). The defini-tion of a follicle, as being composed only ofthe oocyte and encompassing follicle cells,agrees with the origin of the follicle from theGE, establishes its homology to follicles inother vertebrate taxa, and also agrees with his-tology text book definitions (Grier, 2000; Grierand Lo Nostro, 2000).

We describe the GE in male and female teleo-sts using the same definition, giving examples thatgerminal epithelia in viviparous fish are essentiallythe same as those in fish that have an egg-scatter-

Fig. 21.A post ovulatory follicle (POF) in the ovary of the goodeid,Ilyodon whitei: the former follicle cells are vacuolated (v),where lipids presumably dissolved out of the section duringhistological processing, and cell nuclei (n) are indistinct.There is either extensive synthesis of basement membranesurrounding the POF, or there may also be collagen that stainspositively with the periodic acid Schiff reaction. Clearly, acontinuous basement membrane underlies the POF and theGE (arrows). The section includes part of a second POF.Stroma (SR), a developing oocyte (OC) surrounded by aperiodic acid Schiff-positive zone pellucida. Bar = 10 µm

CNPG

OL

GE

BM

F

ZP

PC

S

OCE

➤

➤

n

v

POFSR

O C

GEOL

➤

➤➤

➤

➤

➤


ing mode of reproduction, in an array of clades.In all vertebrates, males and females, the GE iscomposed of somatic cells and germ cells. Inmales, the somatic cells are the Sertoli cells. Infemales, they are the epithelial cells that becomeprefollicle cells associated with a meiotic oocytesor cell nests of oocytes. Finally, prefollicle cellsbecome the follicle cells, surrounding an oocyte,at the completion of folliculogenesis. Of course,the germ cell series begins with gonia, either sper-matogonia or oogonia. A major difference be-tween males and females is the existence of aprolonged diplotene stage of development in fe-males. All oocytes are in diplotene of the firstmeiotic division until just before ovulation whenthe first meiotic division is completed. Germinalepithelia are the source of gametes in both malesand females in spite of morphological diversityamong vertebrate gonads, unless an epithelioidassociation of Sertoli and germ cells exists as inthe Atherinomorpha.

Internal fertilization and viviparity arose in-dependently among fish taxa (Parenti, 1981;Lydeard, 1993). Internal fertilization involvesthe evolution of modified functions of Sertolicells, which become columnar and secretory, andnot the intervention of unusual cell types forthe ultimate packaging of sperm. Primarily, Ser-toli cell secretions are glycoproteins (periodicacid Schiff –positive), and are seen in numer-ous taxa that achieved internal fertilization in-dependently of each other.

The distribution of germ cells within thelobular, germinal compartments of male teleo-sts informs reproductive mode, evolution, andphylogeny. The restricted distribution of sper-matogonia within lobules of Atherinomorphasupports their monophyly, but not their phylo-genetic relationship to any other taxon withlobular testes (Parenti, 2004; Parenti and Grier,2004). Similarly, the type of yolk that is formedduring vitellogenesis, fluid or globular, is alsoof significance and may also reflect on mono-phyly. Morphological diversity, superimposedupon the constancy of fundamental processes(the origin of germ cells from germinal epithe-lia across taxa and between sexes, for example),reflects underlying genetic diversity.

Morphological diversity, involving conservedevolution of derived functions in cell typespresent in the ovary, is common in viviparousfemale fish. In poeciliids, fertilization and thewhole of embryonic development is intra-follicular (Turner, 1940), stretching the mean-ing of ovulation. In goodeids, ovulation

apparently occurs shortly after intrafollicular fer-tilization, and the embiotocid Cymatogaster andthe atherinomorphs Xiphophorus, Anableps, andJenynsia, all have intrafollicular fertilization(Mendoza, 1943). When embryos developwithin the ovarian lumen, ovulation must be ofdeveloping embryos, not eggs. A cycle in theepithelial cells lining the ovarian lumen (the GE)has been documented by Mendoza (1940) inSkiffia bilineata. Hypertrophied cells are in-volved in producing nourishment for the em-bryos during the gestational cycle, when theincrease in size of individuals may be 1,400times that of the egg (Wourms, 2004).

We present specific examples of morpho-logical similarity in ovarian germinal epitheliain fish with external fertilization and fish thathave evolved a viviparous mode of reproduc-tion independently of each other. The GE ismorphologically the same although the ovarianmorphology may be quite different across ver-tebrate taxa. In at least Skiffia bilineata, the epi-thelial cells of the GE change during thegestational cycle (Mendoza, 1940), yet the epi-thelium remains intact. Morphology changesprimarily in the stroma and in eggs. Althoughthe stroma in mammals appears different fromthat in fish, in both, the stroma forms the th-eca surrounding follicles (Van Blerkom andMotta, 1979). Morphology has evolved to pro-duce eggs that must develop in a changing en-vironment (evolution of pelagic eggs [eggs thatfloat in saltwater], demersal eggs [attached toa substrate], eggs with attachment filaments,large eggs as in salmonids that are buried instream beds during a prolonged larval devel-opment, eggs surrounded by gelatinous coat-ings or by calcium in amphibians and reptiles,respectively, and so on. Although they followdifferent developmental sequences, germinalepithelia in females and males are both derivedfrom somatic cells and germ cells, all derivedfrom a common source. The germ cells arederived from primordial germ cells originat-ing in the hindgut and migrating to the ger-minal ridges of the developing embryo. Follicleand Sertoli cells are derived from the mesoder-mal cells in the germinal ridge. Germinal epi-thelia in males and females are defined thesame, except for the epithelioid arrangementobserved in male atherinomorphs—an evolu-tionary modification of the Atherinomorpha.

By application of the periodic acid Schiffstaining technique that renders the basementmembrane and intercellular connective tissue


bright magenta, a new interpretation of follicledevelopment in the Goodeidae has been estab-lished. As a result of the dispersed nature of stro-mal cells in the goodeid ovary, follicle formationis particularly easy to visualize. Quite unexpect-edly, we discovered that follicles remain attachedto the underside of the GE throughout theirdevelopment; each follicle shares a short lengthof the basement membrane underlying the GE.This is the point where ovulation occurs, at thebase of the “delle” (Mendoza, 1943) where thereis but a single basement membrane. Duringovulation, the shared basement membranebreaks, the follicle enters the ovarian lumen,leaving behind a post ovulatory follicle. Thebasement membrane, originally surrounding thefollicle, remains continuous with that of the GE.During ovulation, a new continuum is formedby the follicle cells that were originally derivedfrom the epithelial cells of the GE, and the epi-thelial cells composing the GE. Two epithelialcell layers become one until the post ovulatoryfollicle degenerates. During oocyte develop-ment, follicle cells rest upon the basement mem-brane separation them from the theca, areattached laterally, are avascular, but do not bor-der a lumen as the follicle is filled with the de-veloping oocyte. They should be considered anepithelioid layer of cells.

The significance of the new observations isthat the epithelial derivative, the follicle (theoocyte and surrounding layer of follicle cells) isalways separated from the stroma (a meso-dermally derived tissue) by a basement mem-brane that is derived from that subtending theGE. Folliculogenesis, oocyte development, in-cluding the formation of yolk, and the processesleading to ovulation, are continuous. Textbookdefinitions of a follicle (Grier, 2000) preciselyreflect its origin from an epithelium. The fol-licle extends into but never truly becomes partof the theca, always being separated from it by abasement membrane.

The “unifying concept” is herein expandedto include the origin of gametes in both sexesfrom a GE and the homology of follicles of vi-viparous fish, externally fertilizing fish, and allother vertebrates. Mitosis and the initiation ofmeiosis within a GE underscores fecundity and,most likely, modes of reproduction in fish. Ac-tivity of the GE, as it relates to reproductivemodes in fish is a new, unexplored area in fishreproduction, particularly in viviparous fishwhere viviparity is superimposed upon folliculo-genesis and oocyte development.

Acknowledgements

The senior author acknowledges the Florida Fishand Wildlife Research Institute (formerly theFlorida Marine Research Institute) where many ofthe ideas presented herein were developed. Inparticular, numerous discussions with Ron Taylorhelped to develop the concepts presented on fe-cundity and how approaches in fishery sciencecould underlie studies in viviparous fish reproduc-tion. Many thanks go to those who, over theyears, developed histological staining techniquesand helped with microscopy: Pam Nagle, NorettaPerry, Ruth Reese and Yvonne Waters (FWRI),Marcela Aguilar Morales (Facultad de Ciencias,UNAM), and Helen Wimer (National Museumof Natural History). The skills of Llyn French arereflected in the artwork. We thank Mike Horn forcontribution of surfperch gonads. Partial supportfor this research came from the Tropical Aquacul-ture Laboratory and a contract from the Smith-sonian Institution to H. J. Grier.

Voucher specimens have been or will be de-posited in the collections of the National Mu-seum of Natural History, Smithsonian Institution(USNM). Some catalogued specimens are re-ferred to in the text with a USNM number; oth-ers are cited in Parenti and Grier (2004:table 1).Jeffrey Clayton (USNM) provided valuable tech-nical assistance.


References

Billard R, Cosson MP, Christen R. 1987. Some recentdata on the biology of trout spermatozoa. In: IdlerDR, Crim LW, Walsh JM, editors. Proceedings of theThird International Symposium on the ReproductivePhysiology of Fish. St John’s, Newfoundland: Memo-rial University of Newfoundland. p 187-190.

Breder CM, Rosen DE. 1966. Modes of Reproduction inFishes. New York, NY Natural History Press. p 941.

Brown-Peterson NJ, Grier HJ, Overstreet RM. 2002.Annual changes in the germinal epithelium determinemale reproductive classes of cobia, Rachycentroncanadum. J Fish Biol. 60:178-202.

Burns JR, Weitzman SH. 2004. Insemination in ostario-physan fishes. This volume.

Callard GV. 1991. Spermatogenesis. In: Pang P, Schreib-man M, editors. Vertebrate endocrinology: fundamen-tals and biomedical implications. New York: AcademicPress. p 399-437.

Callard IP, Ho SM. 1984. Vitellogenesis and viviparity. In:Chester-Jones I, Ingleton PM, Phillips JG, editors.Fundamentals of Comparative Vertebrate Endocrinol-ogy. Plenum Press, p 255-282.

Contreras-McBeath T. 2004. Fish conservation in Mexi-co, with emphasis in livebearing species. This volume.

Domínguez-Domínguez O, Mercado-Silva N, Lyons J.2004. Conservation status of Mexican goodeids: prob-lems, perspectives, and solutions. This volume

Greven H. 2004. Structural and behavioral traits associ-ated with sperm transfer in Poeciliinae. This volume.

Grier HJ. 1981. Cellular organization of the testis andspermatogenesis in fishes. Amer Zool 21:345-357.

Grier HJ. 1984. Testis structure and formation of sper-matophores in the atherinomorph teleost Horaichthyssetnai. Copeia 1984(4):833-839.

Grier HJ. 2000. Ovarian germinal epithelium and folli-culogenesis in the common snook, Centropomus un-decimalis (Teleostei, Centropomidae). J Morphol243:265-281.

Grier HJ, Burns JR, Flores JA. 1981. Testis structure inthree species of teleosts with tubular gonopodia.Copeia 1981:797-801.

Grier HJ, Collette BB. 1987. Unique spermatozeugmata intestes of halfbeaks of the genus Zenarchopterus (Tele-ostei: Hemiramphidae). Copeia 1987(2):300-311.

Grier HJ, Linton JR, Leatherland JF, De Vlaming VL.1980. Structural evidence for two different testiculartypes in teleost fishes. Amer J Anat 159:331-345.

Grier HJ, Lo Nostro F. 2000. The germinal epithelium inthe fish gonads: the unifying concept. In: Norberg B,Kjesbu OS, Taranger GL, Anderson E, Stefansson SO,editors. Proceedings of the Sixth International Sympo-sium on the Reproductive Physiology of Fish. TheUniversity of Bergen, Norway, p 233-236.

Grier HJ, Moody DP, Cowell BC. 1990. Internal fertili-zation and sperm morphology in the brook silverside,Labidesthes sicculus. Copeia 1990: 221-226.

Grier HJ, Parenti LR. 1994. Reproductive biology andsystematics of phallostethid fishes as revealed by gonadstructure. Env Biol Fish 41:287-299.

Grier HJ, Taylor RG. 1998. Testicular maturation and re-gression in the common snook. J Fish Biol. 53:512-542.

Hoedeman JJ. 1974. Naturalists’ Guide to Fresh-waterAquarium Fish. London & Sydney. Sterling Publish-ing Co. p 245.

Hunter JR, Macewicz BJ. 1985. Measurement of spawn-ing frequency in multiple spawning fishes. In LaskerR, editor. NOAA Technical Report NMFS 36. USDept of Commerce, NOAA, National Marine Fisher-ies Service, p 79-94.

Hunter JR, Lo NCH, Leong RJH. 1985. Batch fecundityin multiple spawning fishes. In Lasker R., editor.NOAA Technical Report NMFS 36. US Dept of Com-merce, NOAA, National Marine Fisheries Service.

Kulkarni CV. 1940. On the systematic position, structuralmodifications, bionomics and development of a re-markable new family of cyprinodont fishes from theprovince of Bombay. Rec Indian Mus 42:379-423.

Lydeard C. 1993. Phylogenetic analysis of species richness:has viviparity increased the diversification of actino-pterygian fishes? Copeia 1993:514-518.

Macchi GJ, Acha EM, Militelli MI. 2002. Seasonal eggproduction of white mouth croaker (Micropogoniasfurnieri) in the Río de la Plata estuary, Argentina-Uru-guay. Fish Bull 101:332-342.

McDonough CJ, Roumillat WA, Wenner CA. 2003. Fe-cundity and spawning season of striped mullet (Mugilcephalus) in South Carolina estuaries. Fish Bull 101:822-834.

Meisner A. 2004. Male modifications associated with in-semination in teleosts. This volume.

Mendoza G. 1940. The reproductive cycle of the vivipa-rous teleost, Neotoca bilineata, a member of the fam-ily Goodeidae. II. The cyclic changes in the ovariansoma during gestation. Biol Bull 78:349-365.

Mendoza G. 1941. The reproductive cycle of the vivipa-rous teleost, Neotoca bilineata, a member of the fam-ily Goodeidae. Biol Bull 81:70-79.

Mendoza G. 1943. The reproductive cycle of the vivipa-rous teleosts, Neotoca bilineata, a member of the fam-ily Goodeidae. IV. The germinal tissue. Biol Bull84:87-97.

Meyer MK, Wischnath L, Foerster W. 1985. Lebend-gebärende Zierfische: Arten der Welt. W. Germany.MERGUS-Verlag. p 496.

Neidig CL, Skapura DP, Grier HJ, Dennis CW. 2000.Techniques for spawning common snook: broodstockhandling, oocytes staging, and egg quality. North AmJ Aquacult 62:103-113.



Nelson JS. 1994. Fishes of the World, Third Edition. JohnWiley and Sons.

Nielsen, JG, Jespersen A, Munk O. 1968. Spermato-phores in Ophidioidea (Pisces, Percomorphi). Gala-thea Report, Scientific Results of the Danish Deep-SeaExpedition round the world 1950-52. 9:239-254, plus24 plates.

Parenti LR. 1981. A phylogenetic and biogeographicanalysis of cyprinodontiform fishes (Teleostei, Ather-inomorpha). Bull Amer Mus Nat Hist 168:335-557.

Parenti LR. 2004. The phylogeny of atherinomorphs: evo-lution of a novel fish reproductive system. This volume.

Parenti LR, Grier HJ. 2004. Evolution and phylogeny ofgonad morphology in bony fishes. Integr Comp Biol44:333-348.

Poeser F. 2004. Phylogeny and biogeography of Poecilia.This volume.

Ravaglia MA, Maggese MC. 2003. Ovarian follicle ul-trastructure in the teleost Synbranchus marmoratus(Bloch, 1795), with special reference to the vitellineenvelope development. Tissue and Cell 35:9-17.

Rosa-Molinar E. 2004. The shift to a sexually dimorphicbody plan and nervous system in a Poeciliine fish. Thisvolume.

Rosen DE, Bailey RM. 1963. The poeciliid fishes (Cy-prinodontiformes), their structure, zoogeography, andsystematics. Bull Amer Mus Nat Hist 126:1-176.

Rosen DE, Parenti LR. 1981. Relationships of Oryzias,and the groups of atherinomorph fishes. Amer MusNovit No. 2719. 25 p.

Roumillat WA, Brouwer MC. 2004. Reproductive dy-namics of female spotted seatrout (Cynoscion nebu-losus) in South Carolina. Fish Bull 102:473-487.

Sterba G. 1959. Freshwater Fishes of the World. NewYork. Viking Press p 568.

Taylor RT, Grier HJ, Whittington JA. 1998. Spawningrhythms of common snook in Florida. J Fish Biol53:502-520.

Tokarz RR. 1978. Oogonial proliferation, oogenesis andfolliculogenesis in non-mammalian vertebrates. In:

Jones RE, editor. The Vertebrate Ovary—Compara-tive Biology and Evolution. New York, Plenum Press,p 145-179.

Turner CL. 1933. Viviparity superimposed upon ovo-viparity in the Goodeidae, a family of cyprinodontteleosts fishes of the Mexican Plateau. J Morphol55:207-251.

Turner CL. 1937. The trophotaeniae of the Goodeidae, afamily of viviparous cyprinodont fishes. J Morphol61:495-523.

Turner CL. 1938. Adaptations for viviparity in embryosand ovary of Anableps anableps. J Morph 62:323-349.

Turner CL. 1940. Pseudoamnion, pseudochorion and fol-licular pseudoplacenta in poeciliid fishes. J Morphol67:59-89.

Uribe MC, De la Rosa G, Guerrero SM, García AlarcónA, Aguilar ME. 2004. Estructura del ovario de tele-ósteos vivíparous. Gestación intraovárica: intraluminalen Ilyodon whitei (Goodeidae), e intrafolicular enPoeciliopsis gracilis (Poeciliidae). In Lozano Vilano ML,Contreras Balderas AJ, editors. Homenaje al DoctorAndrés Reséndez Medina, UANL México p 31-45.

Van Blerkom J, Motta P. 1979. The cellular basis of mam-malian reproduction. Baltimore/Munich: Urban andSchwarzenberg. p 5-65.

Van den Hurk R, Barends P. 1974. Cytochemistry of theintratesticular efferent duct system of the black molly(Mollienisia latipinna). Kon Ned Wettenschappen-Amsterdam 77C:205-214.

White GG, Munroe TA, Austin HM. 2003. Reproduc-tive seasonality, fecundity, and spawning frequency oftautog (Tautoga onistis) in the lower Chesapeake Bayand coastal waters of Virginia. Fish Bull Get vol-ume424-442.

Wourms J. 2004. Functional morphology, development,and evolution of trophotaenia. This volume

Zamboni L. 1972. Comparative studies on the ultrastruc-ture of mammalian oocytes. In: Biggers JD, SchuetzAW, editors. Oogenesis. Baltimore: University ParkPress. p 2-45.



Fecundity, the Germinal Epithelium, and Folliculogenesis in Viviparous Fishes

Documents