Reference: Biol. Bull. 208: 120-126. (April 2005) 2005 Marine Biological Laboratory Sex Reversal in Pairs of Lythrypnus dalli: Behavioral and Morphological Changes EDMUND W. RODGERS. SHELIA DRANE, AND MATTHEW S. GROBER* Department ofBiolos>\, Georgia State Universitv, P.O. Box 4010, Atlanta, Georgia 30302-4010 Abstract. In Lythrypnus dalli. the bluebanded goby, re- inhibition by males and stimulation from other females productive success is primarily determined by functional (Robertson. 1972; Shapiro, 1979). Without sufficient so- sex, and functional sex is determined largely by rank in the cial stimulation (i.e., in isolation), fish capable of sex dominance hierarchy. In most natural social groups of L. reversal might not initiate sex change (Cole and Shapiro, dalli. one male is at the apex of the hierarchy, and 1 to 7 1995; Carlisle et al.. 2000). Size advantage (Warner et females are lower in rank. When a male exits the group, a til.. 1975) contributes to the determination of which an- female ascends to the top of the hierarchy and becomes a imal changes sex. but behavioral interactions are also male. We have examined this process in a simplified envi- critical determinants (Lutnesky. 1996; Munday. 2002). ronment a pair of females that allows us to identify Larger size often equates with increased success in ag- behavior associated with the formation of a dominance gressive encounters and therefore social dominance, pro- relationship and any other phenotypic changes associated viding a proximate mechanism for the size advantage with dominance, sex change or both. We found that pairs of hypothesis. In protogynous sex changers, the most repro- L. dalli females quickly and readily form stable dominance ductively significant resource that dominance affords is relationships, with the dominant fish changing sex into a "maleness"; thus the reproductive payoff for dominance male. This dominant animal also rapidly increases in body is extremely large, and females would be highly moti- size and length of its dorsal fin. In summary, dominant L. vated to increase their aggressive behavior in times of dalli females change sex in this simplified environment, social instability (i.e.. in the absence ofadominant male). providing excellent opportunities to examine the early be- The study of Lythrypnn.\ Jalli (Gilbert. 1890) had havioral and morphological changes associated with domi- previously dealt exclusively with larger social groups nance and sex change. (>3), and it was not known whether sex change could be induced in pairs of females. This study used pairs of fish Introduction to closely examine changes associated with the acquisi- tion of dominance and, potentially, sex change. In a Social context often determines how a given individual group environment, dominance relationships are more responds to behavioral stimuli, with individuals modify- difficult to tease apart owing to multiple interactions ing their behavior on the basis of the behavior of the between individuals in the group, whereas in pairs there individuals around them. This is especially true in several are fewer social variables contributing to an animal's species of protogynous (female to male) sex-changing behavior. We found that one of the pair quickly changes fish, where sex reversal is socially controlled. In these to male, as evidenced by male-typical behavior and male species, a unique pattern of behavior stimulates one in- reproductive morphology. dividual to change sex while inhibiting others (Robert- son, 1972: Ross el a/., 1983). Two social factors are thought to be the primary regulators of sex change: Materials and Methods We use four measures of "maleness" to determine ifany Received 1 July 2004; accepted 12 December 2004. *To whom correspondence should be addressed. ofthe fish changed sex: (1) display ofmale-typical courting [email protected] behavior; (2) male-typical papilla ratio, defined as a length- 120 SEX REVERSAL IN PAIRS OF L DALLI 121 to-width ratio greater than 1.6 (female -1.0 1/w); (3) the During sex reversal, L. dalli undergoes gonadal reorga- presence of an accessory gonadal structure (AGS); and (4) nization (St. Mary, 1994), which in a number ofsex-chang- the presence offertilized eggs, an unequivocal indication of ing gobies involves degeneration ofovarian tissue (Sadovy functional sex change. and Shapiro, 1987) and generation of AGS and testicular tissue (Cole and Shapiro, 1990). In L. dalli, the presence or Subjects absence of the AGS is the primary indicator of functional We selected 32 female specimens of Lythrypnus dalli sex; males have a highly developed AGS and females have with lengths between 23 and 30 mm for the study. Sex was none. Visual inspection was used to assess the composition d(Beetherremnitnse.d19b8y3)e.xTamhienapatpiiolnlaorfattihoeinexfteemranlaelsgoefnitthails sppaepciilelsa oAfGSre.prGooduncatdivesetxiswsuaes, iv.ee.r,ifpireedseunsciengofsetagngdsaradndpaprraefsfeinncehios-f is about 1.0, compared with about 2.0 in the males (St. tology followed by staining with hematoxylin and eosin. Mary, 1994); the largest ratio at the start ofthis experiment was 1.3. Fish with female-typical papilla have gonads con- sisting of more than 95% ovarian tissue (St. Mary, 1994). Pairing During the experiments, animals were housed in 33-1 At the start of each experiment, animals were removed aquaria, each with an individual filter system (Marineland), from the holding tank and anesthetized using MS 222. Mass cpatyacrl2ee0d.-d2Oie1nt.e(1OCgSr;IoufMpeadroiftnwfeiicsLehabwdsaa)is;lycaonuldsliekncegtpetdao(cnCoaamlmi1ef2ro-rchniialailglFhyits/hpdraer&-k asunrdeds,taannddarddilgeitnaglthiwmeargeesrewceorrdeeda.lsPomaptmialklean.ratAinosimwaelrse mweear-e Game permit # 803034-01) on Santa Catalina Island, Cali- paired by standard length (within 2 ofeach other) and fornia, in May 2002 (pre-breeding season), using an anes- body mass (within 0.1 g). The fish were individually iden- thshaaenmtdie-cnmesetotslh.uotAdiosn,seodcfuorqniudnigngatrlhoeduibpnreoefesdufilifnsaghteswea(asSsiocgnomlialnecCltahteeedm,Jiuucnsaeiln)2g0a0tn2h.de ctwihefarienedgpeblyancoetrdhehtiaorvgeebtahanendrieinfngfteocptaat3ot3ne-r1nt,hteaiwnrhkibwecihhtahvdiaiodrs.inngToltweoaPpVapnCeiamrtaulbtseo to serve as the nest (St. Mary, 1994). Prior to the experiments, the animals were kept in 180-1 holding tanks. Observations and behavior Experimental design The study is divided into two experiments. In experiment The behavior of each pair was observed twice daily, 1, morphological traits and behavioral interactions were once in the morning and once in the afternoon, for 10 min quantified between eight pairs of individuals identified as each session. Data were recorded on paper and then females. External morphological data were collected prior transferred to a computer spreadsheet. The fish were to pairing (standard length, mass, and papilla ratio, de- given one day to acclimate before observations began. scribed below). The pairs were observed until eyed eggs (a The observer recorded the number of approaches, dis- marker of fertilization), indicating functional sex change, placements, jerks, bites, nips, solicitations, and tail-wag- were seen, or until 14 days, an adequate amount oftime for gles. In addition, the observer noted which animal resided sex change in this species (Reavis and Grober. 1999), had in the nest tube, if either animal was gravid, and whether elapsed. Experiment 2 was similar to experiment 1 and was eggs were present in the tube. designed to confirm a novel observation in experiment 1, An approach is defined as a fish moving within 5 cm of that dominant fish exhibited disproportional growth of the the other fish. Movement away from the approaching fish is longest rays ofthe dorsal fin. The behavior ofthe eight pair recorded as a displacement. Jerks are a male-typical behav- of fish in experiment 2 was not observed closely. In both ior used during courtship (Behrents, 1983), and involve a experiments, the pairing and morphological measurements saltatory swim motion with movement laterally as well. (with the exception of initial dorsal fin length) were iden- Becausejerks can be directed either at the female or around tical (see below). the nest without being directed at a fish, a jerk towards a At the completion of each experiment, animals were female is scored as both an approach and a jerk. Bites are given an overdose of tricaine methanesulfate (MS 222), aggressive interactions in whichone fishbitestheotherafter and their morphological characters were measured again. an approach. In contrast, nips follow ajerk to afemale, with This included measurement ofthe longest dorsal ray (see a male nipping the tail of the female. This behavior is also Results). The animals were placed in Bouin's fixative, the part of the courting process. Solicitations are produced by gonads were dissected out to examine internal morphol- females when they move within 12 cm of the male in his ogy, and digital photographs ofall the gonads were taken line of view. Tail waggles are displayed by both sexes: the to aid in analysis. fish remains stationary while moving its tail back and forth. 122 E. W. RODGERS ET AL Statistics ences in size predicted which individual would change sex. Significance level in all cases was set at P < 0.05. Mean Morphological data were analyzed using parametric sta- values standard error of the mean (SEM) are given. All tistics. Fortraits that had "before" and "after" measures, we analyses were earned out using the Statview 5.01 (SAS used paired student's Mests (standard length and fin length Institute Inc.) unless otherwise noted. in experiment 2). For fin length in experiment 1, we used an unpaired f-test to look at differences between males and females. We also used a Mest to examine whether papillar Results morphology was significantly different between new males and females. The distributions of the frequencies of the In all cases, the pairs readily established a stable domi- various behaviors fornewly sex-changed males and females nance relationship, defined as one individual (the dominant) did not meet the criteria for parametric statistics (e.g.. nor- instigating the majority of encounters (Fig. la) and exhib- mality, homogeneity of variance); thus we used nonpara- iting male courtship behavior (Fig. Ib). while the other metric methods to examine group differences. All behav- individual (the subordinate) is displaced as a result of the ioral analyses were conducted with the Wilcoxon signed encounter (Fig. Ic). In all cases the dominant individual rank test. Values for mean and median were similar, indi- initiated spawning behavior with the subordinate individual cating a symmetrical distribution. Simple linear regressions that resulted in a clutch of eggs. Of the 8 clutches, 4 were were employed to examine the relationship between known verified as fertile; in the other 4 cases, the male consumed interdependent behaviors (e.g., approaches regressedon dis- the clutch before verification. At the completion of the placements). Though the fish were size-matched, we used experiment, all dominants had a male papillaratioas well as logistic analysis (SAS) to examine whether small differ- an ACS (Figs. 2, 3). Using our criteria, we demonstrated 16 14 - 12 - 10 - \ 23456789 10 Day I S. 15 in i I 10 u J2 & Q 5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Day Figure 1. Changesinsex-typical behavioroverthecourseofexperiment. Animalsthatare labeledmaleare individuals that changed sex i.e., those that would become male, (a) Males approach and females rarely do, demonstrating dominance, (b) Jerking behavior, a majorcomponent ofmale courting behavior, is exclusively produced by males, (c) Females were displaced at high rates compared to males, (d) Solicitations, a type of femalecourting behavior, are exhibitedexclusively by females. Notethe similarpatternofbehaviorin band d. SEX REVERSAL IN PAIRS OF L. DALL1 123 mean of5.5 0.269 mm (t = 8.475, df = 14, P < 0.0001 ). To control fordominant fish simply growing more, we used --o* - SDuobmoirndainntat/eM/alFeemale a ratio of fin length to body length. This ratio was signifi- cantly different: subordinates had a smaller fin with an average of4.648 0.137 fin lengths per body length, and dominants had aratioof2.64 0.143 (t = 10.129, df = 14, o P < 0.0001). Experiment 2 verified the marked change in fin length observed in the first experiment. Future dominants and '5. subordinates did not have different fin lengths at the begin- Or.o ning of the experiment (t = 1.669, df = 8; P > 0.05), but the dominant fish had a significantly longer fin at the end of the experiment (t = 4.091, df = 8; P = 0.0035). Dominant individuals exhibited a pronounced lengthening of their longest dorsal ray, while many of the subordinates showed a fin shortening. Initial Final We attempted to minimize size as a major contributing Figure2. Papillalength/widthratiosatthestartandconclusionforall factor; however, small differences in size did exist between Pathnai<nmas0lu.sb0o0irn0d1ei)xn.paetersim(efnetma1l.eDs)omaitntahnetesnd(moaflethse)ehxapdearismiegnnitfi(tca=ntl8y.6h4i,gdhfer=rat14i,o ethnecepsaiirnedstaannidmaarlds.leTngothaswseersse wphreedtihcetrivetheosfewshmiaclhl adniiffmearl- changed sex, we ran a logistical regression. Our results suggest that the size differences present in this experiment tsheaxtftrhoemdofemmianlaenttoinmdailvei.dual in a pair of females changed dwoemrienanontta(gWoaoldd prred=ict0o.r03o4f.whPic=h 0a.n8i5m3a2l,wdoful=d b1,ec/3om=e -0.05006). Morphological changes Behavior All fish initially had a female-typical papilla ratio (mean SEM: 1.15 0.061). There was no difference in Verification of egg fertility in experiment 1 was diffi- = = papilla ratio at the start ofthe experiment (t 0.732, df cult because of consumption of clutches by males (see 14, P > 0.05) between fish that became male (mean below). Four of eight pairs completed sex change in less SEM: 1.15 0.102) and those that remained female than 14 days, as determined by the presence of fertilized (mean SEM: 1.17 0.086). After sex change there was eggs. In the four remaining pairs, males exhibited all a significant change to a male-typical papilla in the fish that other sex-typical characteristics, but consumed at least demonstrated male-typical behavior(t = 8.64, df = 14. P < one complete clutch of eggs. Because fish that parented 0.0001). Female papilla ratio remained the same. 1.09 and fish that consumed their eggs displayed similar rates 0.052, whereas the ratio in sex changers more than doubled of behavior in all measures, we grouped them in the to 2.26 0.127 (Fig. 2). behavioral analysis. All dominant animals exhibited unambiguous male-typi- A clear dominance relationship was evident from the cal gonad morphology that included the presence of sperm first observational session and persisted unchanged and an AGS. All subordinate individuals were gravid, and throughout the experiment (Fig. 1). The dominant fish eggs were clearly visible within the ovary upon inspection. produced male-typical behavior and began courting the These results were verified histologically (Fig. 3). subordinate. The dominant fish instigated the vast major- The standard length of dominant individuals changed ity of behavioral interactions, evidenced by high rates of significantly over the course of the experiment (P < 0.05); approach behavior. Dominants displayed significantly that ofsubordinates did not (P > 0.05). When we examined higher rates of aggressive and courtship behavior (ap- the magnitude and direction of the change, we found that proaches and jerks, P < 0.05. Fig. la, b). Subordinates dominants grew significantly more relative to subordinates displayed significantly more submissive and female-typ- (paired /-test. / = 4.528, df = 15. P = 0.0003). ical behavior (displacements and solicitations, P < 0.05, In experiment 1, we noted that the dorsal fin of the Fig. Ic, d). All dominants jerked and subordinates gen- dominant fish was elongated. Although we had not mea- erally did not jerk, although there were rare individual sured initial fin length, at the conclusion of the experiment displays over the course of the experiment. Subordinate we measured the longest dorsal ray. Dominants had a mean fish were displaced at high rates (9.138 per 20 min. Fig. fin length of 10.47 0.539 mm, while subordinates had a Ic) compared to dominants (0.062 per 20 min). There 124 E. W. RODGERS ET AL ACS Figure 3. Gonad morphology ofa typical male (left) and female (right). The cross-section ofeach type is shown underneath. Note that the male gonad contains mostly testicular tissue, some ovarian tissue, and an accessory gonadul structure (AGS). The female gonad contains only ovarian tissue and is visibly full ofeggs. was a significant association between approaches and Other behavior occurred at a far lower frequency, includ- displacements, such that when a dominant approached, ing bites, nips, and tail waggles. Dominant individuals pro- the subordinate was usually displaced (R2 := 0.917; P < duced bites and nips. Tail-waggle was the one behaviorthat 0.001 ). The dominant guarded or resided in the tube for did not show an overall sex/dominance bias, but was ex- most of the time. hibited by both dominants and subordinates. SEX REVERSAL IN PAIRS OF L. DALL1 125 Discussion ior (approaches, jerks, bites, and nips). Morphological ex- amination revealed no differences between males that dem- When two female bluebanded gobies were paired, in our onstrated appropriate care and those that did not. The experimental conditions, inallcases,onefishestablisheddom- gonads of the two groups were indistinguishable from one inance over the other, with that fish instigating most of the another, and both groups developed an ACS, the hallmark encounters and winning nearly all ofthem. The dominant fish of the male sex. There were no differences in the papilla became male, exhibiting all behavioral and morphological between the two groups, and all males displayed sexually characters associated with that sex. The subordinate fish re- dimorphic elongation ofthe longest dorsal rays. In addition, mained female and exhibited all associated sex-typical behav- parenting males had fertile clutches at the same time as egg ior and morphology. consumption was occurring, suggesting that all ofthe males In Lythrypnusdalli, body size is sexually dimorphic, with were capable of fertilizing eggs. males being larger (Wiley, 1976). In our work, dominant In experiments with groups of fish greater than two individuals grew more than subordinates and changed sex. (Reavis and Grober, 1999), the amount oftime required for The observed sexual dimorphism in growth rate is consis- full sex change depended on the size of the individuals in tent with previous data from St. Mary (1994); however, we the group upon male removal. In groups where the size cannot address whether the increase in growth is a result of asymmetry between the top two females was greater than sex change or of dominance. Dominance rank has been 10%, sex change occurred more rapidly than in groups shown to affect the growth rate of individuals within a where the size asymmetry was less than 10%. The observed hierarchy of sex-changing clown fish (Buston, 2003). We delay was determined to be the result ofan increase in time alsofound that the dorsal fin was elongated in the dominant, spent resolving the dominance relationship between the top sex-changed fish. The dorsal fin is known to be sexually two individuals. Individuals of similar size take longer to dimorphic (Wiley, 1976). but we did not expect the rapidity settle a conflict than individuals of different sizes (Enquist and magnitude of the changes we observed in the first el al. 1990; Koops and Grant, 1993). In this experiment, we experiment. Experiment 2 confirmed that the fin lengthened used pairs of size-matched individuals; thus we were sur- dramatically in the dominant individual over the course of prised to find that the contestants quickly resolved the the experiment. The dorsal fin shrank in many of the sub- dominance relationship. One possible explanation is that in ordinate fish, but whether this is a natural process or the a group, individuals may receive contradictory signals. This result of aggressive encounters with the dominant is un- could result from being dominated by one individual but known. As with the change in growth rate, we cannot say also receiving positive stimulus by dominating other indi- whether the fin elongation is caused by the sex-change viduals in the group, thus prolonging conflict resolution. In process or is modulated by dominance status. Ifthe latter is dyadic contests, the signals are straightforward; one animal the case, then this trait can potentially be used as an index wins the encounters while the other loses them, thereby of rank within a hierarchy. shortening the time and probably the numberofinteractions Simplification ofthe social group to a pair ofanimals, as required to form a stable dominance relationship. in our study, revealed an interdependence of behavior be- In summary, two females is a sufficient group size to tween males and females that was not seen in previous induce sex change in L. dalli, as evidenced by one individ- studies using larger groups offish. This is best illustrated in ual exhibiting male typical gonads, external genitalia, be- Figure Ib andd, where courtship behavior(solicitations and havior patterns, and often fertilized eggs. When an L. dalli jerks) clearly shows the same pattern over time. Although female becomes dominant, the first noticeable morphologi- the precise nature of the interaction is unclear, female be- cal change is the elongation of the dorsal fin. Dominant havior apparently had an impact on the rate ofmale behav- individuals also grow more rapidly. ior, an effect that would be difficult to examine in a group with multiple females. Acknowledgments Like ReavisandGrober(1999), we used fertilizedeggsas a marker for complete sex change and allowed the experi- The authors thank Cathleen Drilling for her work in ment to run for a maximum of 14 days. All tanks had eggs collecting the experimental animals, help with behavioral in the nest within the 14 days, but only half of those nests observations, and gonad histology; Michael Black for ani- hadeggsthatremained until they couldbe verified as fertile. mal collection and behavioral observations; and William In the four other groups, the males did not display paternal Lonergan and Beth Stokes for their help with behavioral care. These males ate the eggs after day 10 of the experi- observations. We also thank Ryan Earley for his help with ment, making it impossible to use the presence ofeyed eggs statistical analysis, and Dr. Charles Derby and an anony- to verify fertility and thus terminate the experiment. The mous reviewer for their comments on an earlier version of bulk of the data indicates that these eggs were viable. this paper. This work was supported by NSF IBN 9723817 ', Clutch-consuming males exhibited all male-typical behav- to Matthew S. Grober, the Georgia Research Alliance and 126 E. W. RODGERS ET ,\L. tthhee CNeSnFteragfroereBmeehnatvi#oIraBlN-Ne9u8r7o6s7ci5e4n.ce, an STC programof lM.uuntpdnoeamsyak,yc,aPn.Mt.hLi.d129a90n60g.2e.lfiSsiBhzi.-ed-Cidereneptcetrniodopneyangltesrpeaoxttetceohrfai.npgrCeoo:tpoteegisyatnion1ug:st2h0see9x-g2rc1oh2wa.tnhgeraitne advantage model. Behav. Ecol. Sociobiol. 52: 247-254. Literature Cited Reavis, R., and M. S. Grober. 1999. An integrative approach to sex Behrents, K. C. 1983. The comparative ecology and interactions be- cJahlalnige(:Pisscoecsi)a.l,AcbteuhaEvtihonrla.l2a:nd51n-e6u0r.ochemical changes in Lythrypmis tween two xympatric gobies (Lythrypim\ iltilli and Lythiypnus zebra). Robertson.D.R. 1972. Socialcontrolofsexreversalinacoral-reeffish. BustPohn.,D.P.di2s0s1e(r3t.atioSni./eUnainvdergsriotwythofmoSdoiuftihceartnioCnaliinfcolronwian.fish.Nature424: RossS,ciRe.nceM.1,77G:.HSK.IL7o-sHeKyW,.and M. Diamond. 1983. Sex change in a CCaorlleLdoi1i,t.sn4silie5mKleC-i.,.oa1,irnn4SJan.6.tgen.heFLiden.iri.sotDgh,aSo.l.bBaiYipKin.oa.dldp.SiMglhM5aela7.arnp:xuimesSr4o.ro4rC-.G5pMor-hir1o4ol9yb5l9lpe6o0ehr..gro.y.pt2AiGne0.o0ntn0Vuha..ed(MsTees.ExltferfCcoueashcctntateuansirrg:eioifonaG,gon1bdfRi1ii.-hskdeheaFrt.eLmolyatO.tpelhshiCtrrvooyoesppditemireiitars,-a- SSahdaocpshtoivirezroyrrelao,.ml.a-oSprYfcehD.ie,r.sefoenadxYcfin.iestdrhie2:1sv29Dem1d7.r:e9si.panY5el.7nf4idiSs-Senho5h.nca7hciCp5eaeoi.lrprmoeobafi.epahhsat1rv1i9o:im8dou7i1rl.t3a,et6i-gCof1rrni5ois6htua.e.pnrdiAsadtvirfn.uohcritSbuittrtuheiedoynadniBodaenghtnahorvees.liacstoi1nvo0e-:f 43-102. Cole4Mm,:eadr9Ki.9.a,6tB-iiao1oln0n.0do3Ef.cDao.dlu.lYt1.8s6eS:xhac6ph5ia-r7no5g..ei1n9a95c.ryptSiocc,ibaelntfhaiccilmitaartiinonegaonbdy.sJe.nEsxopr.y St. Mtbheaehraybvl,iuoeCra.blaMgn.edne1dd9e9rg4o.abnydS(tehLxeytcaholrlnyospcemaqtiuiseonn</c/i/en/s/'a)fosritmrhueelptreaoffdneueccottuisoofnh.ebroBmdeayhpahvsri.ozdeEicatoneld,. EKnoqoupfs41iies97ssq9,5htu0-,e.N4MnaM7t.ni.9anA,.lAa.Oct,a.eassrtasLaeenosidfasmnmasoJeer.mnq,atuWleTan.i.t.niLaAAcjl.onuninavmGsdi.rscbaeteBnsretscgh.mi,acevhn1Y.lt9.id94g3M0a.ca:molnel1t:nW-ee1se4ftrii.s,g.ghhattCniandnag.sNiy.nJm.SmteheZgetoeorrc/yid.cahha7lln1i.d:d WWialre5an(Ay:Gnem,iderl3rb,J0se.i4ercR-txW.a)u3..an1Rl3.T1..grs9oae7Dnbl6es.i.c.etsRiS.o.LnLiR.DyfotieSbhecehigiriyoestpntsnSocouorencise.,1s9dNa0aaaln:ntldd.i63sEH(3i.yGs-sitGt6l.e.3bme81arL.8tte:)iicg1sah6n.o9df-11t9L8h7y4e5t..WhersySptemexirsnchNzaoenrbgtreha