JOURNAL OF EXPERIMENTAL ZOOLOGY 283:81–90 (1999)

© 1999 WILEY-LISS, INC.

Sperm Competition and Mode of Fertilization inthe Grass Goby Zosterisessor ophiocephalus(Teleostei: Gobiidae)

M. SCAGGIANTE,1 C. MAZZOLDI,1 C.W. PETERSEN,2 ANDM.B. RASOTTO1*1Dipartimento di Biologia, Università di Padova, 35121 Padova, Italy2College of the Atlantic, Bar Harbor, Maine 04609

ABSTRACT Accessory structures to the male reproductive system are known in several fami-lies of teleost, but their role in sperm production patterns and fertilization dynamics is still un-clear. The intraspecific variability in seminal vesicles, shown by the grass goby Zosterisessorophiocephalus, presents an opportunity to examine both its possible correlation to alternativemale mating tactics and the function of these accessory structures. In this species, males areknown to release sperm in the form of sperm trails, bands of mucosubstances in which sperm areembedded.

Surgical removal of seminal vesicles and histochemistry demonstrate that mucins, involved inthe production of sperm trails, are secreted by the seminal vesicles. Gametes show a high longev-ity; sperm motility lasts on average 80 minutes. Eggs can be fertilized for several hours and do notneed to be laid over trails because sperm are able to reach them via the surrounding water.

Gonosomatic and seminal vesicle somatic indices, histology and histochemistry of gonads andseminal vesicles, sperm counts, and sperm trail longevity, suggest the presence of alternativemale mating tactics in this species. Larger males have smaller testes and larger seminal vesiclescompared to those of smaller males. The major role of seminal vesicles is mucin secretion inlarger males, sperm storage in smaller ones. Trails of larger males last longer and releasesperm more constantly over time than those of smaller males. Overall differences betweenmales support the presence of differences in the intensity of sperm competition, with largermales performing nesting behaviour while smaller ones sneak spawnings. J. Exp. Zool. 283:81�90, 1999. © 1999 Wiley-Liss, Inc.

Sperm competition, defined as the competitionbetween the sperm of two or more males to fertil-ize the eggs of a female (Parker, ’70), is well knownto strongly influence not only the behaviour ofmales but also their reproductive anatomy andphysiology (Smith, ’84). In fish, it has been dem-onstrated that males exposed to higher levels ofsperm competition have relatively larger testesand release an absolute higher number of spermper spawn or ejaculate (Robertson and Warner,’78; Warner and Robertson, ’78; Shapiro et al., ’94;Stockley et al., ’97; Petersen and Warner, ’98).

Testes and the associated structures of the malereproductive system have roles that go beyondsperm production. The production of substancesthat may serve hormonal or pheromonal roles, fa-cilitating sperm delivery and sperm storage, areimportant roles of the testis and its accessorystructures (Fishelson, ’91; Van den Hurk andResink, ’92). These accessory structures often areembedded within the testes, or lay adjacent to

them. It is unclear how these structures arerelated to sperm production patterns, both evolu-tionarily and functionally. Current sperm compe-tition models that use gonosomatic indices toestimate sperm production within or between spe-cies assume that functions other than sperm pro-duction are either not important or are constantacross comparisons.

The family Gobiidae is a large family of small,freshwater and marine species with demersalspawning and male parental care. Male spermduct glands, usually called seminal vesicles, areaccessory structures of the reproductive appara-tus of all species of the suborder Gobioidea (Miller,’92). However, neither the significance of their

Grant sponsors: Italian Ministero dell’Università e della RicercaScientifica e Tecnologica; U.S. National Science Foundation.

*Correspondence to: M.B. Rasotto, Dip. Biologia, Viale G. Colombo3, 35121 Padova, Italy. E-mail: rasotto@civ.bio.unipd.it

Received 1 December 1997; Accepted 28 April 1998

82 M. SCAGGIANTE ET AL.

high variability in the development and secretoryactivity, both across and within species (Miller,’84; Fishelson, ’91; Joyeux et al., ’91), nor theirroles in reproduction, are well understood. Semi-nal vesicles are known to secrete mucosubstancesin many gobiid species (Miller, ’84; Lahnsteineret al., ’90; Fishelson, ’91). In several other teleostfamilies of demersal spawners with male pa-rental care, such as Blenniidae, Labrisomidae,Tripterygiidae, Cottidae, Gasterosteidae, Heter-opneustidae, etc., males also release mucinsproduced either by structures accessory to thegonads (Lahnsteiner et al., ’90; Rasotto, ’95) orby an hypertrophied kidney (Bucher and Hofer,’93). One hypothesized role of these mucosub-stances is the production of sperm trails: bandsof viscous materials in which sperm are embed-ded when released (Marconato et al., ’96; Otaet al., ’96). Many fishes, including some withthese accessory structures, are known to havealternative male mating tactics that often re-sult in different levels of sperm competition andmating success for different individuals in apopulation (Taborsky, ’94). Intraspecific vari-ability in seminal vesicles has been describedin some gobiid species, but its possible correla-tion with different male mating opportunitiesor different reproductive tactics has never beeninvestigated.

To examine the hypotheses of seminal vesiclesfunction(s) and their possible relationship to malemating tactics, we studied the Mediterraneangrass goby, Zosterisessor ophiocephalus. The spe-cies has seminal vesicles known to produce mu-cins (Lahnsteiner et al., ’90) and presents aconspicuous variability in the size of mature males(De Girolamo, ’94). A recent study demonstratedthat in Z. ophiocephalus and two other goby spe-cies, sperm trails are released by males before fe-males start to lay eggs, and the trails continue torelease active sperm over prolonged periods oftime (Marconato et al., ’96). In these species,sperm are released directly onto the nest surface.This asynchronous mechanism of gamete releaseis atypical for teleosts, where successful externalfertilization is assumed to require the synchro-nous release of gametes by both partners (Brederand Rosen, ’66; Jamieson, ’91).

Two questions arise from these observations inZ. ophiocephalus. The first concerns the glandu-lar production of sperm trails. It has been sug-gested that seminal vesicles play a major role insperm trail production, but whether this is theirprimary function and whether the function of

seminal vesicles varies intraspecifically are notknown. The second question regards the evolu-tionary significance of sperm trail production andthe asynchronous gamete release between malesand females. If sperm trails or similar sperm re-lease mechanisms are a function of accessorystructures, then by studying intraspecific and in-terspecific variation in its development we maygain insight into the selective forces that shapedthe evolution of these structures and their prod-ucts. Marconato et al. (’96) suggested one pos-sible advantage of sperm trails: by releasingsperm in trails, territorial males could defendterritories while active sperm were still beingreleased in a nest.

The specific aims of this study were (1) to evalu-ate the investment in both gonads and seminalvesicles in males of different size; (2) to clarifythe function of seminal vesicles; (3) to comparesperm trail longevity and sperm viability of malesof different size; and (4) to study fertilizationdynamics.

MATERIALS AND METHODSThe grass goby, Zosterisessor ophiocephalus

(Pallas, 1822), is a large coastal gobiid inhabitingthe seagrass meadows of Zostera (Z. marina andZ. nolti) in shallow brackish water. During thereproductive season, lasting from March to May,territorial males dig a burrow under the seagrassrhizomes. Females lay the eggs on the ceiling ofthe burrow, attaching them to overhanging Zosterarhizomes. The eggs diameters are 0.88–1.21 mm,and hatching occurs after 10 days at a water tem-perature of 15°C. Males defend the nest and takecare of the eggs laid by one to several females,cleaning and fanning them until hatching (Gan-dolfi et al., ’91; De Girolamo, ’94). In the VenetianLagoon (Northern Adriatic Sea, Italy), the grassgoby reaches sexual maturity after one year andlives for up to two and a half years, reaching amaximum total length of 25 cm (De Girolamo, ’94).

Mature males and females of Z. ophiocephaluswere collected during the reproductive seasons1995 and 1996, in the Venetian Lagoon using per-manent fish traps. Males and females were rec-ognized by the sexually dimorphic genital papillae(Gandolfi et al., ’91) and were kept separate inlarge stock tanks. Each aquarium was providedwith a sandy bottom and with artificial shelters.Seawater was renewed daily, and temperaturerange was kept at 17–18°C. Light regime followednatural conditions, and fishes were fed once a daywith fresh black mussel meat.

SPERM COMPETITION AND FERTILIZATION IN A GOBY 83

Gonosomatic and seminal vesiclesomatic indices

In 1995 and 1996, respectively, 129 (size range:8.5–22.0 cm total length) and 84 (size range: 8.1–22.6 cm total length) freshly caught males weresampled. If not already dead, they were sacrificedwith an excess of anaesthetic (MS222). Totallength (TL) and weight of body, gonads and semi-nal vesicles of every specimen were measured (tothe nearest mm and 0.01 g). Gonosomatic [GSI =(gonad weight/body weight) × 100] and seminalvesicle somatic indices [SVSI = (seminal vesiclesweight/body weight) × 100] were calculated.

Histology, histochemistry, and seminalvesicle surgical removal

Testes and seminal vesicles were processed his-tologically to establish gonad maturation, to studythe morphology of seminal vesicles, and to deter-mine their function(s). Histochemistry was used toidentify the chemical nature of the seminal vesiclesecretion. Surgical removal of seminal vesicles inlive males was performed to determine the role oftheir secretion in sperm trail composition.

The entire reproductive apparatus of 25 males,ranging in size from 10.4 to 20.2 cm TL, was re-moved, fixed in Dietrich’s solution (900 ml dis-tilled water, 450 ml 95% ethanol, 150 ml 40%formaldehyde, 30 ml acetic acid), embedded inparaplast, sectioned transversally and serially at5–6 µm, and mounted on slides. Sections fromeach specimen were stained with haematoxylinand eosin, as well as histochemical stains. Forpolysaccharide detection, sections were stained bythe reaction of periodic acid-Schiff (PAS) (Pearse,’85), and for the differentiation of sulphated andnon-sulphated mucins by the methods of AlcianBlue at pH 1.0 and pH 2.5 (Pearse, ’85). Proteinswere stained by the mercury Bromophenol Bluemethod (Pearse, ’85). The same histochemicalstainings were applied on sperm trails obtainedfrom squeezed males.

Testes were classified as one of three stages ofsexual maturation: “resting”, “ripe”, and “post-spawning” on the basis of the predominance ofdifferent stages of spermatogenesis (Nagahama,’83). Classification of seminal vesicle developmentwas made using two criteria: the thickness of theirchamber walls and the presence of secretion orsperm into the chamber lumina.

From seven males, ranging from 12.4 to 20 cmTL, we surgically removed the seminal vesicles.Males were first gently squeezed. The extruded

sperm trail, laid on a clean slide, was stained byPAS; then the males were deeply anaesthetizedin an aerated seawater solution of MS222 for 6–8min. A 1.5 cm ventral incision was made close tothe genital papilla, the seminal vesicles were re-moved, and the incision was closed with 2.0 sur-gical silk. The gills periodically were irrigated withfresh seawater during surgery. Males recoveredrapidly, were immediately returned to their hometanks, and were fed 12 hr later. Following six daysof recovery in the tanks, males were again gentlysqueezed and their sperm trails collected on cleanslides and stained by PAS.

Sperm trail longevity, sperm release andfertilization dynamics

To measure the longevity of sperm trails, 37males (size range: 8.0–22.0 cm TL), in 1995, and16 (size range: 10.3–21.5 cm TL), in 1996, wereanaesthetized with a solution of MS222 andsqueezed, pressing the genital papilla gently overan acetate sheet. The 62 trails obtained were mea-sured (length and width to the nearest mm) andkept in beakers at controlled temperatures (15–17°C). The trails were periodically checked forsperm movement under a light microscope (mag-nification of 100×). Trail longevity (minutes)was measured from initial squeezing to the endof sperm movement. For males with multiplesamples, means for length, width, and trail lon-gevity were used in the analyses. To check if trailsobtained by squeezing had different longevity fromnatural ones, six trails obtained from malesspawning in the laboratory were compared withsix trails squeezed from same-sized males.

As an alternative measure of sperm trail lon-gevity, we tested whether trails could fertilizeeggs for extended periods following squeezing.A squeezed sperm trail was put in a beaker withnewly squeezed eggs. After 30 min the eggs wereremoved and put in another beaker to develop.After three hours, another egg mass was addedto the initial beaker with the trail, left for 30 min,then removed. These trials were repeated untilthe trail was completely dissolved, and the finalegg sample was added three hours after it. Tenhours after immersion with trail, eggs were ob-served under a stereomicroscope to assess thenumber of fertilized eggs. This experiment wasrepeated five times.

We also tested if eggs could be fertilized bysperm trails not laid down in close proximity toeggs. Gametes were obtained by squeezing ma-ture males and females, anaesthetized with a so-

84 M. SCAGGIANTE ET AL.

lution of MS222, on different acetate sheets.Sheets with eggs and sperm were put in a beakerwith seawater and maintained at a distance of 11cm (N = 8) or 20.5 cm (N = 2). After 10 hr thenumber of fertilized eggs were counted under astereomicroscope.

To quantify sperm release from trails, 7 males,in 1995, and 11 males, in 1996, were squeezed asabove. According to the size categories recognizedby the histological analyses, 9 males were defined“large” (larger or equal to 16.0 cm TL) and 9“small” (smaller or equal to 12.6 cm TL). Eachtrail was first kept in 80 ml of seawater for 5 min(first sample) and then removed to another bea-ker with seawater. The sperm in the 80 ml samplewere stained with five drops of Rose Bengal for20 min and then fixed with 16 ml of formalde-hyde (Leong, ’88; Shapiro et al., ’94). The samplewas then passed through a millipore filter (0.22mm pore size) under vacuum and the filter papercontaining sperm was dried on a slide warmer,placed on a glass slide and cleared with immer-sion oil. Using a light microscope (magnificationof 400×) we counted the number of sperm presentin an area measuring 0.31 × 0.31 mm (=0.0961mm2). The count was repeated ten times in ran-domly selected portions of the filter. The meanvalue of these readings was used to backcalculatethe total number of sperm released from the trailin the first sample. Subsequently, the trail peri-odically was checked for sperm movement andsimilar samples were collected for processing af-ter 30, 60, 120, 210, 390, 520, 720, and 1560 minThe ratios between the number of sperm releasedin each sample and that released in the first onewere computed for the statistical analyses.

Sperm longevity was estimated as the durationof sperm motility. Thirty-five males, ranging insize from 8.0 to 22.0 cm, were anaesthetized witha solution of MS222. While gently squeezing themale, sperm samples were collected directly fromthe genital papilla with a pipette and mixed withseawater at controlled temperature (15–17°C). Adrop of sperm was put under a light microscope(magnification of 100×) and sperm movement wasperiodically checked until 95% of sperm stoppedmoving. To control for changes in slide tempera-ture affecting sperm longevity while under illu-mination, a sperm drop, squeezed with the firstone and kept at controlled temperature, was thenchecked for sperm movement. The duration ofsperm motility (seconds) was recorded from theactivation of the sperm when mixed with seawa-ter until 95% of sperm stopped moving in the last

control. Means for sperm longevity obtained fromthe same males were used for analyses.

To test the duration of egg viability, a batch ob-tained from a female was divided and put in dif-ferent petri dishes. In the first dish, sperm wereadded simultaneously with eggs, while in the otherdishes fresh sperm were added at increasing timeintervals. The maximum time from egg squeez-ing to sperm addition was 40 hr. Ten hours fol-lowing sperm addition, the number of fertilizedeggs was estimated under a stereomicroscope.

Data analysesData were tested for normality using the Kolmo-

gorov–Smirnov test and, if they were non-normallydistributed, they were square-root or log trans-formed. If the transformed variables were stillnon-normally distributed, non-parametric testswere performed. Data from the two years werecompared. When significant differences werefound in variables between years, separate testswere done with data from each year.

RESULTSGonosomatic and seminal vesicle

somatic indicesGonosomatic Index decreased with male size

(Fig. 1 and Table 1; years 1995: r Pearson = –0.67,P < 0.001, N = 129; 1996: r Pearson = –0.26, P =0.018, N = 84). This negative correlation appearsto be due to a sharp decline in GSI from 8-16cm; when considering only males larger then 16cm TL, both in 1995 and 1996, no significantcorrelations were present between size and GSI

Fig. 1. Relationship between GSI and male TL (1995: r =-0.67; 1996: r = -0.26).

Abbreviationsc connective tissuee epitheliumm secretions sperm

SPERM COMPETITION AND FERTILIZATION IN A GOBY 85

(1995: r Spearman = –0.34, P > 0.05, N = 21; 1996:r Spearman = –0.275, P > 0.05, N = 12). Therewere significant differences in gonad sizes (GSI,ANOVA: F = 18.42, P < 0.001) and in male TL(Mann-Whitney U test: U = 4539.0, P = 0.045)between years.

Seminal Vesicle Somatic Index showed the op-posite pattern to that of the GSI, increasing withbody size in both years (Fig. 2 and Tab. 1, 1995: rPearson = +0.77, P < 0.001, N = 129; 1996: rPearson = +0.58, P < 0.001, N = 84). Between 1995and 1996 there were no significant differences inthe relative size of the seminal vesicles amongmales (SVSI, ANOVA: F = 0.39, P > 0.05).

Histology, histochemistry and seminalvesicle surgical removal

All males collected were sexually mature. Tes-tes were classified as ripe because of the pres-ence of many sperm in the lobule lumina (Fig. 3)and of all stages of spermatogenesis in the semi-niferous epithelium lining the seminiferous lob-ule walls. Seminal vesicles are multichambered(Fig. 4), paired organs opening into the terminalportion of the sperm duct. The wall of the cham-ber consists, from inside to outside, of a single

layer of epithelial cells, a basal lamina, and a thinlayer of connective tissue containing capillaries.Differences in the thickness of the monolayeredepithelium and the content of chamber luminawere observed. Three categories were recognized:

Type a: seminal vesicles with highly ex-tended chambers, thin chamber walls of flatepithelial cells, and chamber lumina com-pletely filled with secretion (Fig. 5). Fewsperm were observed and these were seenexclusively in the proximal portions to thesperm duct. All males larger than 16 cm TL,and one male of 14.1 cm, showed this type ofvesicles.

Type b: seminal vesicles with thick cham-ber walls, the epithelial wall cells cylindri-cal in shape, the chamber lumina mainlyfilled with sperm, and only a small amountof secretion present among sperm (Fig. 6).The largest specimen showing this type ofvesicle was 12.6 cm TL.

Type c: seminal vesicles with chamber wallsof intermediate thickness, epithelial cells cu-bic in shape, and chamber lumina showing amixture of sperm and secretion (Fig. 7). Thesevesicles were present in two males of 11 and11.6 cm TL and in all males ranging from 12.7and 16 cm TL except for a single 14.1 cm TLspecimen showing Type a vesicle.The apical portion of epithelial cells and the se-

cretion present in the lumina reacted positivelyto PAS (Figs. 4, 6), to Alcian Blue at pH 2.5 (Fig.8), and to mercury Bromophenol Blue (Fig. 9) in-dicating that seminal vesicle secretions weresialoglycoproteins. Sperm trails too gave strongreactions to PAS, to Alcian Blue at pH 2.5, and toBromophenol Blue, indicating that sialoglico-proteins were the main component of trails. Com-parison of sperm trails obtained from the samemales before and after surgical removal of semi-nal vesicles showed a drastic decrease of mucins,as evidenced by a reduced PAS staining (Fig. 10)in the latter.

Sperm trail longevity, sperm release andfertilization dynamics

Between 1995 and 1996, there were no signifi-cant differences in the TL of squeezed males(ANOVA: F = 0.01, P > 0.05), trail length (ANOVA:F = 2.44, P > 0.05), trail width (Mann-Whitney Utest: U = 219.5, P > 0.05), or trail longevity(ANOVA: F = 2.69, P > 0.05), so the data for bothyears were pooled (Table 2).

TABLE 1. Values of male TL, GSI and SVSI

Variable Mean ± SE Range1995 N = 129

Male TL (cm) 13.36 ± 0.24 8.50–22.00GSI 3.27 ± 0.18 0.09–8.60SVSI 0.43 ± 0.03 0.00–1.92

1996 N = 84

Male TL (cm) 12.91 ± 0.33 8.10–22.60GSI 2.19 ± 0.22 0.08–8.15SVSI 0.40 ± 0.04 0.00–1.76

Fig. 2. Relationship between SVSI and male TL (1995: r= +0.77; 1996: r = +0.58).

86 M. SCAGGIANTE ET AL.

Fig. 3. Cross section through testis of a 19 cm TL grassgoby, Zosterisessor ophiocephalus, male. Spermatogonia andcysts of primary spermatocytes, secondary spermatocytes, andspermatids are present in the lobule walls whereas sperm (s)fill lobule lumina. Hematoxylin and eosin staining. ×100.

Fig. 4. Seminal vesicles of a 17.7 cm TL grass goby male.Seminal vesicles are organized in chambers whose walls con-sist of a monolayered epithelium (e) and connective tissue(c). Chamber lumina are filled with a homogeneous secretion(m). Both epithelial cells and secretion positively react to PASstaining. PAS staining. ×190.

Fig. 5. Seminal vesicles of a 19 cm TL grass goby male.Chambers are highly extended and their lumina completelyfilled with secretion (m). Haematoxylin and eosin staining.×100.

Fig. 6. Seminal vesicles of an 11.2 cm TL grass goby male.Chambers are filled with sperm (s) and only a low amount ofsecretion (m) is evidentiated by PAS reaction. ×230.

Fig. 7. Seminal vesicles of a 13.9 cm TL grass goby male.Both secretion and sperm (arrow) are equally present in cham-ber lumina. Haematoxylin and eosin staining. ×190.

Fig. 8. Seminal vesicles of a 20.2 cm TL grass goby male.

SPERM COMPETITION AND FERTILIZATION IN A GOBY 87

Trail longevity was positively correlated withtrail width (r Pearson = +0.48, P < 0.001, N = 53)and with male TL (r Pearson = +0.70, P < 0.001,N = 53), but was not correlated with trail length(r Pearson = +0.23, P > 0.05, N = 53). Male TLwas positively correlated with trail width TL (rPearson = +0.60, P < 0.001, N = 53), but was notcorrelated with trail length (r = +0.15, P > 0.05,N = 53). In a stepwise multiple regression, onlymale TL was significantly correlated with traillongevity (Fig. 11, adjusted r2 = 0.48; TL: T = 6.96,P < 0.001; trail width: T = 0.725, P > 0.05).

No significant differences were found in lon-gevities between trails released spontaneously bymales and trails from squeezed males (Wilcoxontest: Z = –0.40, P > 0.05, N = 6 pairs).

Sperm from trails always fertilized eggs, untilthe entire trail dissolved. The mean trail longev-ity was 19 hr 55 min. The mean fertilization ratein the last sample, obtained just before the spermtrail dissolution, was 45%. Eggs added after threehours from the complete dissolution of the trailswere never fertilized.

Sperm trails kept at a distance of 10 or 20.5 cmfrom eggs successfully fertilized them in all tri-als, with a mean fertilization rate of 59%.

The number of sperm released from trails inthe first sample (time 0 min) was not significantlydifferent between trails squeezed from large andsmall individuals (Mann-Whitney U test: U = 21.0,P > 0.05) and the means were 0.56 ± 0.15 × 106,for large males, and 4.20 ± 2.04 × 106, for small

ones. There were significant differences in the ra-tios of sperm (see materials and methods) betweenlarge and small males (test for combining prob-abilities: c2 = 51.30, P < 0.001, N = 9; computedfrom Mann-Whitney U tests for each time sample)(Sokal and Rohlf, ’81). The trails squeezed fromlarge males released, in the times following thefirst one, larger proportions of sperm and for alonger period of time than the ones released bytrails from small males (Fig. 12).

Sperm motility for free sperm outside of spermtrails averaged 83 min (Tab. 3) and was not cor-related with male TL (r Pearson = +0.11, P > 0.05,N = 35).

Adding sperm to eggs after different time in-tervals, fertilization rate decreased from 55% withnew eggs to a fertilization rate of 37% after 30 hr.The maximum time after egg squeezing at whicheggs were still fertilized was 40 hr, and the fer-tilization rate was 22%.

Both epithelial cell of the chambers and secretion filling cham-ber lumina positively react to Alcian Blue at pH 2.5. ×210.

Fig. 9. Seminal vesicles of a 19 cm TL grass goby male.Secretion of seminal vesicles react positively to Mercury Bro-mophenol Blue. ×210.

Fig. 10. Sperm trails artificially obtained from a 20.2 cmTL grass goby male before (left) and after (right) surgicalremoval of seminal vesicles. PAS staining evidentiates a dras-tic reduction of the mucous amount in sperm trails releasedafter surgery. ×3.

TABLE 2. Values of male TL and characteristicsof trails

Mean ± SE Range N

MaleTL (cm) 14.53 ± 0.53 8.00–22.00 53

Traillength (cm) 0.82 ± 0.07 0.30–3.50 53

Trailwidth (cm) 0.22 ± 0.02 0.10–0.65 53

Trail longevity 15 hr 32 min ± 1 hr 15 min – 53 1 hr 4 min 29 h 45 min

Fig. 11. Relationship between trail longevity and male TL(r = +0.70).

Fig. 12. Sperm released from trails by males of differ-ent size categories. The y values are the ratios betweenthe number of sperm released in sample and that releasedin the first one.

88 M. SCAGGIANTE ET AL.

DISCUSSIONGrass gobies exhibit a striking change in both

the relative size of the gonads and in the size andfunction of the seminal vesicles with male bodysize. Small males have relatively large gonads,implying large amounts of sperm production. Inmature males smaller than 12.6 cm TL, the ma-jor function of seminal vesicles is sperm storageand mucin production is minimal. At the otherextreme, large males (>16 cm TL) have relativelysmall gonads, implying lower relative amounts ofsperm production, and the seminal vesicles pro-duce large amount of mucins that fill seminalvesicle chambers, causing their conspicuous en-largement. Males of intermediate body size haveseminal vesicles showing a variable mixture ofsperm and mucins.

The histochemical stainings we performed con-firm that seminal vesicles secrete mucins in theform of sialoglycoproteins, as was demonstratedin this species by Lahnsteiner et al. (’92). Thedrastic reduction of mucins in the sperm trailsobtained from males after the surgical removal ofseminal vesicles also reflects the importance ofseminal vesicle secretions in sperm trail produc-tion, as suggested by Marconato et al. (’96). Thesmall amount of mucins still present in the trailsof operated males could be due to remnants ofthese substances in the sperm duct since the lasttrail release. Our histological examinations re-vealed no likely secondary sites for mucin produc-tion in the gonads.

The production of mucins, a viscous materialthat dilutes in seawater only slowly, accounts forsperm trail longevity. Larger males, which havea larger relative investment in seminal vesiclesand mucin production, lay down wider sperm trailswith greater longevity, thus prolonging the fertili-zation period of a male spawn. These changes inallocation to male reproductive structures and thefunctional differences caused in both sperm pro-duction and longevity, are best explained by theexistence of different male mating tactics by dif-ferent sized males. Nesting males of this speciesin the Venetian Lagoon are typically large, with48 out of 59 grass goby nesting males larger than

16 cm TL (Scaggiante, ’96; Mazzoldi, personal ob-servation). Nesting males, usually individualslarger than 16 cm TL, with their production oflong-lasting sperm trails, do not need to stay closeto the female over the entire spawning period inorder to secure fertilization, and can devote timeand energies to defending the nest from intrud-ers. Small males, which may be incapable of dig-ging and defending the large multichambered nestcharacteristic of this species, mate using a differ-ent tactic. Their high investment in sperm pro-duction equips them to engage in non-territorialreproductive tactics. Alternative male mating tac-tics, typically employed by smaller males, are com-mon in fishes and often involve sneaking intonests and releasing sperm in competition with thesperm of the territorial male (reviewed in Tabor-sky, ’94). These small males typically have a largeinvestment in sperm production, with muchhigher GSI than territorial males who spend lessenergy on sperm and more energy on defendingthe nest site and attracting mates (reviewed inStockley et al., ’97; Petersen and Warner, ’98). Inthe grass goby, multiple ripe males, most of themsmaller than 12.6 cm TL, are sometimes observedin nest with female (Mazzoldi, personal observa-tion). These grass goby small males produce notjust more but more quickly active sperm, whilethe territorial male sperm release tactic is to re-lease a lower but steadier supply of sperm in thenest. Aquarium experiments show a three-fold in-crease in sperm number when a small male is in-troduced in the tank where a female and a largemale are spawning (Scaggiante, ’96). The trade-off occurs through a change in function of theseminal vesicles from sperm-storage organ to amucin-producing organ.

While large and small males show anatomical,physiological, and behavioral characteristics con-sistent with alternative mating tactics, malesranging in size between 12.6 to 16 cm TL presentintermediate investment both in testes and semi-nal vesicles, the latter used both for sperm stor-age and mucin production. Probably these males,while too small to dig a nest, are sufficiently largeto compete opportunistically for vacated territoriesand, depending on opportunity, can become nestingmales or adopt a sneaking tactic of mating.

Sperm and egg longevity play important rolesin the fertilization dynamics of the grass goby andhave important implications for the evolution ofits mating system. Eggs retain the capability tobe fertilized for several hours and sperm motilitylasts on average about 80 min. Our experiments

TABLE 3. Values of male TL and sperm motility

Mean ± SE Range N

Male TL (cm) 14.16 ± 0.65 8.00–22.00 35Sperm motility 83 min 28 sec ± 50 min 58 sec – 35

3 min 37 sec 130 min 40 sec

SPERM COMPETITION AND FERTILIZATION IN A GOBY 89

of artificial fertilization demonstrated that eggsdo not need to be laid directly over sperm trailsto be fertilized and, indeed, sperm trails have beenreported to be laid by the nesting male in a shortertime compared to female egg laying, and not nec-essarily in contact with eggs (Marconato et al.,’96; Scaggiante, ’96). While information on teleostegg viability is scant, sperm are known to be quiteshort-lived in freshwater fishes and ranging from30 to 120 sec in most marine fishes (Ginzburg,’68; Billard, ’86; Petersen et al., ’92). The oneknown exception is the herring, in which spermare active for several hours following release(Dushkina, ’73).

The synchronous release of short-lived gametescharacterize spawning in most externally fertiliz-ing fishes, but grass gobies are an exception tothis pattern. Instead, long-lived gametes are at-tached to the substrate within a burrow, reduc-ing the dispersive effects of the seawater andprolonging the association between viable eggsand sperm. These factors influence the potentialstrategies open to both territorial males and tomales employing alternative male mating tactics.Marconato et al. (’96) suggested that by layingsperm in trails, territorial males could in ef-fect defend their territories from potentialsperm competitors while still fertilizing eggs inabsentia through the sperm released from theirsperm trails.

When sperm competition occurs, differences inmale GSI have been suggested to indicate differ-ences in the number of sperm released by males(Parker, ’90; Parker and Begon, ’93). Differencesin male ejaculate size are well known for mam-mals and birds (e.g. Møller, ’88, ’89) but only re-cently have been demonstrated in some marineteleost pelagic spawners in which alternative mat-ing tactics are present (Shapiro et al., ’94; Marco-nato and Shapiro, ’96). In Z. ophiocephalus, thereare differences not only in the total amount ofsperm produced but in the functional longevity ofthe sperm. Trails of larger males release spermmore constantly over time and last longer thanthose of smaller males, which release most of thesperm immediately. Thus, despite identical dura-tions of sperm motility for free sperm, male grassgobies have functionally polymorphic spawns, withlarger males having fewer but longer-functioningsperm per spawn.

While intraspecific variations in seminal vesicledevelopment have been observed in other gobiidspecies (Joyeux et al., ’91; Scaggiante, ’96), a cor-relation with male body size and the performance

of different functions has never before been re-ported in a species. Anatomy, histology and his-tochemistry of seminal vesicles, investigation oftheir organization, their secretion, and inferenceof function have mainly been conducted over a lim-ited range of body size of males for each species.This approach is unlikely to detect the possiblediverse uses of these glands in males of differentsizes. Our results on the grass goby Z. ophio-cephalus show that seminal vesicles can exhibitdifferent functional roles for different individualsin the population, and suggests that the majordifferences between males are differences in theintensity of sperm competition and the presenceof a territorial male mating tactic. By expandingour comparative work of accessory organs to otherspecies, we believe that the interspecific and ad-ditional intraspecific differences that may be foundin histology, size, and function of gonads and semi-nal vesicles will allow us to better understand therole of these structures in the evolution of the re-productive and mating system of these fishes.

ACKNOWLEDGMENTSThe idea of this work came out during some ex-

citing discussions with A. Marconato a few monthsbefore his death. We dedicate this paper to him: aperson whose scientific competence, enthusiasm,and friendship we miss. The research was sup-ported by grants from the Italian Ministerodell’Università e della Ricerca Scientifica e Tec-nologica (M.S., C.M., M.B.R.) and from the U.S.National Science Foundation (C.W.P.). The manu-script was greatly improved by the helpful reviewsof Y. Sadovy and H. Hess.

LITERATURE CITEDBillard R. 1986. Spermatogenesis and spermatology of some

teleost fish species. Reprod Nutr Dév 26:877–920.Breder CM Jr, Rosen DE. 1966. Modes of reproduction in

fishes. Garden City: Natural History Press.Bucher F, Hofer R. 1993. Histological and enzyme histochemi-

cal changes in the kidney of a male bullhead (Cottus gobio)during the spawning period. J Fish Biol 42:403–409.

De Girolamo M. 1994. Osservazioni sulla pressione di pescanelle acque della laguna di Venezia e sulla struttura dipopolazione e biologia riproduttiva di tre gobidi (Pisces,Teleostei). Thesis. Faculty of Science, University of Padova,Padova, Italy.

Dushkina LA. 1973. Influence of salinity on eggs, sperm andlarvae of low-vertebral herring reproducing in the coastalwaters of the Soviet Union. Mar Biol 19:210–223.

Fishelson L. 1991. Comparative cytology and morphology ofseminal vesicles in male gobiid fishes. Jpn J Ichthyol38:17–30.

Gandolfi G, Zerunian S, Torricelli P, Marconato A. 1991. Ipesci delle acque interne italiane. Roma: Istituto Poligraficoe Zecca dello Stato.

90 M. SCAGGIANTE ET AL.

Ginzburg AS. 1968. Fertilization in fishes and the problemof polyspermy. Akad. Nauk SSSR, Institute Biologii Ratzv-tiya, Moscov. Translated from Russian by Israel Programfor Scientific Translations, Jerusalem 1972.

Jamieson BGM. 1991. Fish evolution and systematics: evi-dence from spermatozoa. Cambridge: Cambridge Univer-sity Press.

Joyeux JC, Bouchereau JL, Tomasini JA. 1991. La repro-duction de Gobius niger (Pisces, Gobiidae) dans la lagunede Mauguio - France. Rapports gonosomatiques, fécondité,ponte, oeuf et larves. Vie Milieu 41:97–106.

Lahnsteiner F, Richtarski U, Patzner RA. 1990. Functionsof the testicular gland in two blenniid fishes, Salaria(=Blennius) pavo and Lipophrys (=Blennius) dalmatinus(Bleniidae, Teleostei) as revealed by electron microscopyand enzyme histochemistry. J Fish Biol 37:85–97.

Lahnsteiner F, Seiwald M, Patzner RA, Ferrero EA. 1992.The seminal vesicles of the male grass goby Zosterisessorophiocephalus (Teleostei, Gobiidae). Zoomorpholology111:239–248.

Leong RJH. 1988. Sperm concentrations and egg fertiliza-tion rates during spawning of captive anchovy, Engraulismordax. Calif Coop Oceanic Fish Invest Rep 30:136–139.

Marconato A, Rasotto MB, Mazzoldi C. 1996. On the mecha-nism of sperm release in three gobiid fishes (Teleostei:Gobiidae). Environ Biol Fishes 46:321–327.

Marconato A, Shapiro DY. 1996. Sperm allocation, spermproduction and fertilization rate in the bucktooth parrot-fish. Anim Behav 52:971–980.

Miller PJ. 1984. The tokology of gobioid fishes. In: Potts GW,Wooton JR, editors. Fish reproduction: strategies and tac-tics. London: Academic Press. p 119–153.

Miller PJ. 1992. The sperm duct gland: a visceral synapo-morphy for gobioid fishes. Copeia 1992:253–256.

Møller AP. 1988. Testes size, ejaculate quality, and spermcompetition in birds. Biol J Linn Soc 33:273–283.

Møller AP. 1989. Ejaculate quality, testis size and spermproduction in mammals. Funct Ecol 3:91–96.

Nagahama Y. 1983. The functional morphology of teleost go-nads. In: Haor WS, Randall DJ, Donaldson EM, editors.Fish physiology. Vol IX A. New York: Academic Press. p223–275.

Ota D, Marchesan M, Ferrero EA. 1996. Sperm releasebehaviour and fertilization in the grass goby. J Fish Biol49:246–256.

Parker GA. 1970. Sperm competition and its evolutionaryconsequences in the insects. Biol Rev (Camb) 45:525–567.

Parker GA. 1990. Sperm competition games: sneaks and ex-tra-pair copulations. Proc R Soc Lond B, Biol Sci 242:127–133.

Parker GA, Begon ME. 1993. Sperm conpetition games:Sperm size and number under gametic control. Proc R SocLond B, Biol Sci 253:255–262.

Pearse AGE. 1985. Histochemistry, theoretical and appliedanalytical technology. Edinburgh, Scotland: ChurchillLivingstone.

Petersen CW, Warner RR. 1998. Sperm competition in fishes.In: Birkhead TR, Møller AP, editors. Sperm competitionand sexual selection. London: Academic Press. (in press).

Petersen CW, Warner RR, Cohen S, Hess HC, Sewell AT.1992. Variable pelagic fertilization success: implicationsfor mate choice and spatial patterns of mating. Ecology73:391–401.

Rasotto MB. 1995. The male reproductive apparatus of someblennioidei (Pisces: Teleostei). Copeia 1995:907–914.

Robertson DR, Warner RR. 1978. Sexual patterns in thelabroid fishes of the western Caribbean II: the parrotfishes(Scaridae). Smithson Contrib Zool 255:1–26.

Scaggiante M. 1996. Dinamica di accoppiamento e difecondazione in due gobidi mediterranei. Thesis. Facultyof Science. Padova, Italy: University of Padova.

Shapiro YD, Marconato A, Yoshikawa T. 1994. Spermeconomy in a coral reef fish, Thalassoma bifasciatum. Ecol-ogy 73:391–401.

Smith RL. 1984. Sperm competition and the evolution ofanimal mating systems. Orlando: Academic Press.

Sokal RR, Rohlf FJ. 1981. Biometry. San Francisco: W.H.Freeman and Company.

Stockley P, Gage MJG, Parker GA, Møller AP. 1997. Spermcompetition in fishes: the evolution of testis size and ejacu-late characteristics. Am Nat 149:933–954.

Taborsky M. 1994. Sneakers, satellites, and helpers: para-sitic and cooperative behavior in fish reproduction. AdvStudy Behav 23:1–100.

Van den Hurk R, Resink JW. 1992. Male reproductive sys-tem as sex pheromone producer in teleost fish. J Exp Zool261:204–213.

Warner RR, Robertson DR. 1978. Sexual patterns in thelabroid fishes of the western Caribbean I: the wrasses(Labridae). Smithson Contrib Zool 254:1–27.