Hormones and Behavior 58 (2010) 660–668

Contents lists available at ScienceDirect

Hormones and Behavior

journal homepage: www.elsevier.com/locate/yhbeh

Hormonal and behavioral correlates of morphological variation in an Amazonian electric fish ( nattereri: Apteronotidae)

Cristina Cox Fernandes a,b,⁎, G. Troy Smith c, Jeffrey Podos a, Adília Nogueira b, Luis Inoue d, Alberto Akama e, Winnie W. Ho c, José Alves-Gomes b a Department of Biology, University of Massachusetts, Amherst MA, USA b Coodenação de Pesquisa em Biologia Aquática, Laboratório de Fisiologia Comportametal e Evolução, Instituto Nacional de Pesquisas da Amazônia, Manaus, Brazil c Department of Biology, Indiana University, Bloomington IN, USA d Embrapa, Manaus, Brazil e Universidade Federal do Tocantins, Porto Nacional, Brazil article info abstract

Article history: The weakly electric fish from the main channel of the Amazon river, Sternarchogiton nattereri, offers a striking Received 9 February 2010 case of morphological variation. Females and most males are toothless, or present only few minute teeth on Revised 21 May 2010 the mandible, whereas some males exhibit exaggerated, spike-like teeth that project externally from the Accepted 6 June 2010 snout and chin. Androgens are known to influence the expression of sexually dimorphic traits, and might be Available online 16 June 2010 involved in tooth emergence. In this study we assess the relationship in S. nattereri between morphological variation, 11 ketotestosterone (11-KT) and testosterone (T). We also examine relationships of morphology Keywords: Electric fish and androgen levels with electric organ discharge (EOD) frequency, reproductive condition, and seasonality. Androgens Our main finding is that male morph categories differed significantly in plasma concentrations of 11-KT, with Testosterone toothed males showing higher levels of 11-KT than toothless males. By contrast, we did not detect statistical 11-ketotestosterone differences in T levels among male morph categories. Reproductive condition, as measured by Sexual dimorphism gonadosomatic indexes (GSI), differed across two sample years, increased as the season progressed, and Electric organ discharge was higher in toothed males than in non-toothed males. EOD frequency was higher in toothed males than in either toothless males or females. Taken together, our findings suggest that S. nattereri male sexual characters are regulated by 11-KT levels, and that both morphology and androgens interact with reproductive condition and EOD frequency in ways that vary within and across reproductive seasons. © 2010 Elsevier Inc. All rights reserved.

Introduction In teleost fishes, androgens mediate the expression of a broad array of male secondary sex characters including swords in swordtails Xipho- Androgens regulate secondary sex traits in many vertebrate taxa phorus (Baldwin and Goldin, 1939),nuptialcolorationinmaleFundulus (e.g., vom Saal, 1983; Hews and Moore, 1995; Peters et al., 2000; Cox (Lofts et al., 1966), sonic muscle mass in midshipman Porichthys (Brantley et al., 2009). Classical studies on the relationship between androgens et al., 1993), head crest in peacock blennies Salaria (Oliveira et al., 2001; and sexual traits often focused on behavioral differences between the Saraiva et al., 2010), elongated spines of the dorsal fininfilefish sexes (e.g. Beach, 1945). Androgens, however, can also mediate Monacanthus (Gorbman et al., 1983), and indentation of the dorsal fin intrasexual variation in traits, particularly in males. For example, and electric organ discharge (EOD) in various of mormyrids (Bass exogenous testosterone treatment induces young male tree lizards et al., 1986; Landsman and Moller, 1988; Herfeld and Moller, 1998). (Urosaurus ornatus) to develop characteristics of territorial males, Accumulating evidence suggests a particularly central role in sexual trait including territorial behavior and orange/blue coloration (Hews et al., development for two types of androgens, testosterone (T) and especially 1994). Androgens presumably mediate secondary sexual traits through 11-ketotestosterone (11-KT, Kime, 1993). organizational and activational effects, thus providing with The primary aim of our study is to describe relationships between mechanisms by which they can adjust their phenotypes to varying 11-KT, T, and the expression of intrasexually dimorphic teeth in an environmental circumstances (Moore and Thompson, 1990). Amazonian electric fish, Sternarchogiton nattereri. In this species, females and most males are toothless, or present only few minute teeth on the mandible. Some males, however, exhibit exaggerated, spike-like teeth that project externally from the snout and chin (Cox ⁎ Corresponding author. Department of Biology, University of Massachusetts, Amherst MA, USA. Fernandes et al., 2009;andFig. 1). Male dentition projecting entirely E-mail address: [email protected] (C. Cox Fernandes). outside the mouth is the only morphological trait that distinguishes

0018-506X/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.yhbeh.2010.06.006 C. Cox Fernandes et al. / Hormones and Behavior 58 (2010) 660–668 661

Fig. 1. Representative specimens of Sternarchogiton nattereri. A. Full body view showing the supraoccipital bone (front arrow) and the base of the dorsal filament (rear arrow). The distance between these two points, ST, provided a measure of body size not influenced by head or tail morphology. Note the regenerated tail. B and C. Category I fishes, with prominent external teeth. Scars and scratches are visible on the nape. D. Category II fish, with incipient teeth (visible only via x-ray). E. Category III fish, no teeth on the premaxilla. F. Female.

these male morphs. Indeed, morphological sexual dimorphism in this fishes are mediated by androgens (e.g., Taborsky, 1997; Oliveira et al., species is so extreme that toothed males had been previously classified 2001; Saraiva et al., 2010). as a different and species (Oedemognathus exodon, Myers, 1936, Secondary aims of our work are to describe specific relationships of see Cox Fernandes et al., 2002 for a similarly pronounced example of male morph and androgens with electrical communication signals, sexual dimorphism). reproductive condition, and seasonality. Like other electric fishes of the Organizational and activational effects of androgens on tooth family Apteronotidae, S. nattereri produces weak, wave-type electric development were first identified in salamanders, in studies involving organ discharges (EODs), which are used for electrolocation and testicular transplants and castration. Noble and Pope (1929) found that communication. In some apteronotid species, EOD frequency has been the typical dentition of male Desmognathus fuscus (large distinctive shown to vary with sex and/or social status, both of which are regulated premaxillary monocuspid teeth) was expressed in adult females who in part by androgens (Hagedorn and Heiligenberg, 1985; Dunlap et al., received testicular transplants, and that castration caused male 1998; Dunlap and Oliveri, 2002; Kolodziejski et al., 2005; Tallarovic and dentition to develop with a more typically female pattern (short Zakon, 2005). Exogenous application of androgenic hormones has been premaxillary bicuspids). More recent work in salamanders has shown found to raise EOD frequency in some species of electric fishes, and to that androgen levels influence tooth size and form (Ehmcke et al., 2003). lower EOD frequency in other species (Dunlap et al., 1997; Zakon and In fishes, to the best of our knowledge only one prior study has Dunlap, 1999). The relationship between androgens and EOD frequency examined the effects of androgens on tooth formation. Female medakas in S. nattereri is presently unknown. (Oryzias latipes) given exogenous methyl testosterone were found to In natural populations of electric fishes, little is known about the develop enlarged distal teeth typically found only in adult males seasonality of reproduction, hormones, and the development of secondary (Takeuchi, 1967). As a sexually dimorphic trait, enlarged distal teeth sexual traits. Reproductive condition in most tropical freshwater fishes seem likely to be used by male fishes as weapons, in intrasexual varies in synchrony with flood pulses of rainy seasons. Environmental competition for mates (Emlen, 2008). In our study species, scars and conditions that may trigger seasonal changes in reproductive physiology scratches are often found on the heads and napes of toothed males, but include food availability, water conductivity, water speed, turbulence, and never on toothless males or females (Cox Fernandes et al., 2009). These oxygen concentration levels (Schwassmann, 1978; Junk, 1984; Fernandes, marks likely result from direct fights, or perhaps from aggressive social 1997). Manipulation of some environmental factors, such as conductivity, interactions during courtship. Toothless males, on the other hand, more water level and simulated rain, have been shown in several gymnotiform likely use other kinds of non-aggressive behavioral tactics to access species to influence gonadal recrudescence and regression (Kirschbaum, females. A broad literature indicates that multiple reproductive tactics in 1979). In central Amazonia, fish typically reproduce for several months, 662 C. Cox Fernandes et al. / Hormones and Behavior 58 (2010) 660–668 beginning as river levels start to rise in December with the onset of the rainy season (Junk et al., 1989; Araujo-Lima and Oliveira, 1998). Many fish species spawn in the Amazon main channel, as food supplies become more abundant for offspring along the floodplains (Schwassmann, 1978; Junk, 1984; Marrero and Taphorn, 1991). In a study of the electric fish Sternopygus macrurus, in natural populations in the Venezuala llanos, Zakon et al. (1991) documented hormone levels, EODs, and maturation stages as they varied from the early to late dry season, when these fish breed. Plasma steroid levels (T, 11-KT and estradiol) were found to correlate with gonadal recrudescence, and 11-KT concentrations in particular were found to correlate inversely with male EOD frequencies (Zakon et al., 1991). It is not known whether these patterns apply to other Fig. 2. Water level of the Rio Negro during 1997/98 (line with black circles) and 2006/ species of electric fishes (such as S. nattereri) that breed in other regions 07 (line with triangles), based on daily averages provided by the Administration of the (such as the Amazon). Port of Manaus. A main logistic challenge in our study was to capture these fishes alive and in good condition. S. nattereri occurs in the white muddy water indices, see below). These fishes were also collected with trawl nets of the Amazon main channel and its tributaries, at depths between few from the vicinity of Manaus. All other analyses focused only the 2006/07 and more than 20 m deep (Cox Fernandes, 1995). Large inventory specimens. samples of S. nattereri were previously collected with benthic trawl tows of 15–20 min on the main river channels of the Amazon and Orinoco Housing conditions rivers (Fernandes et al., 2004). Most fishes in these trawl tows, however, either arrived dead or were not in condition to be kept alive. We found After collection, fish were transferred to 28 individual tanks with here that quick trawl tows, on the order of 2–5mineach,allowedfishes biological filters filled with oxygenated white water from the river (pH to be brought to the surface with minimal damage and stress to them. of 5.6 and a conductivity of 94 μs/cm). The individual tanks have a 45 L capacity (50×30×30 cm) and were filled to the 40 L mark. Tempera- Methods ture was maintained between 23.6 and 25.6° C, and photoperiod held at 12:12 L:D, following the natural light conditions in the tropics. In these Fish collection tanks specimens acclimated for about 24 h before blood collection and EOD recordings. Blood collection and EOD recordings were always done Fish were captured using a 3×1 m try net trawl designed to operate during the day, between 10 AM and 6 PM. in dugout canoes (see Lopez-Rojas et al., 1984, for details on equipment). Once collected, fish were quickly placed in plastic bags Morphology filled halfway with water from the collection site, and transported to the Laboratório de Fisiologia Comportamental e Evolução at the Instituto Males used in this study were divided into 3 categories based on Cox Nacional de Pesquisas da Amazônia, Manaus, Brazil. Many small Fernandes et al. (2009), as illustrated in Fig. 1 (see also Table 1): (I) individuals were returned immediately to the river. Fish collected in Teeth present, prominent; individuals with large and numerous visible the same trawl tows were considered to co-occur, given the short tow teeth on the premaxillary bone, (II) Teeth present, incipient; premaxilla durations. The proximity of the laboratory to the field site allowed us to with incipient fine, conical teeth, visible by x-ray, and (III) Teeth absent; study hormone levels and EODs across the breeding season, and with premaxilla toothless. Representatives of the three male morphs are minimal artifacts introduced by laboratory conditions. illustrated in Fig. 1. To differentiate males with incipient teeth and males All fish used in this study were collected in the Amazon river, without teeth, preserved specimens were radiographed, at the Academy approximately 3–5 km above the confluence of the Negro river, of Natural Science of Philadelphia. Males without teeth and females between 03° 15′ 15.60″ S and 59° 59′ 29.45″ W, at Ilha da Marchantaria were differentiated by gonadal examination. (Ilha dos Mouras), an island of about 15 km2, near Manaus, Brazil. All Total length (TL) of fish was taken during blood collection (Table 1), procedures associated with collection, handling and holding in as a reference for their individual identification. In apteronotids, TL is captivity were in accordance with guides for research with animals of often not reliable as a measure of body size, because of tail loss (and the Instituto Nacional de Pesquisas da Amazônia. subsequent regeneration) due to predation (Lundberg et al., 1996). Also, Fifty-one individuals (31 males, 20 females) were captured during 4 in the focal species, the presence of external teeth may interfere with TL months (Nov. and Dec. 2006, and Feb. and Apr. 2007), encompassing a measurements. We therefore used a size measurement that does not broad interval of the rainy season and rising of the river level (Table 1, include fish tail and snout: ST, i.e. the distance from posterior edge of the Fig. 2). In addition to the 2006/07 samples, 43 males collected by one of supraoccipital bone to the anterior point on the base of the dorsal us (CCF) during the reproductive season of 1997/98 were used for a filament (Fig. 1 of Cox Fernandes, 1998, Table 1). ST was not measured supplementary analysis of reproductive condition (gonadosomatic from three category I males and one category II male (Table 1).

Table 1 Morphological categories, sample sizes by month, and two measures of body size for the 2006/07 specimens. N=November, D=December, F=February, A=April. TL=total length, ST=distance from posterior edge of the supraoccipital bone to the anterior point on the base of the dorsal filament (Fig. 1A).

Category Sample size TL (mm) ST (mm)

ND F A Range Mean SD Range Mean SD

Male I. 622163–225 201.2 22.5 96.3–112.6 106.5 5.5 Teeth present, prominent Male II. 7 199–234 214 13.9 86.2–107.1 98.4 5.7 Teeth present, incipient Male III. 4 9 1 168–234 198.6 19.8 73.7–105.7 92.9 9.8 Teeth absent Female 6 14 212–138 188.7 19.3 86.1–107.8 96.7 8.5 C. Cox Fernandes et al. / Hormones and Behavior 58 (2010) 660–668 663

Reproductive condition buffer1: 50-1:100 (11-KT) or 1:20-1:50 (T). Assays were performed following kit instructions. Intra-assay variations were 27.2–28.8% We assessed reproductive condition of males only, using gonado- (11-KT) and 11.6% (T). Inter-assay variation for 11-KT was 18.3%. We somatic indices (GSIs), as measured from (1) males captured in were unable to quantify 11-KT levels for one category I male, or T levels December 2006 and (2) the 1997/98 specimens. For the December for two category I males and one category III male. Eight of 20 females 2006 sample, we were unable to successfully measure GSI from one had T measured (Table 1). category I specimen and three category II specimens (Table 1). The 1997/98 specimens were not x-rayed and thus could be categorized Statistics only as either teeth present (Category I) or absent (Category III). For consistency in GSI calculations, body mass, testes and sperm The effect of category on body size for males was assessed in two ducts were taken from preserved specimens and not during the blood comparisons: males with prominent versus incipient teeth (Table 1, sampling while specimens were fresh. Male reproductive organs can be category I versus II, using a Mann–Whitney U test), and males with teeth small and easily damaged while fresh, yet are firm after preservation versus without teeth (Table 1, categories I and II combined versus III, and can be dissected with greater reliability. Testes and sperm ducts using a t-test). The same category comparisons were then performed for were removed from preserved specimens via an incision on the right GSI using the same statistical tests. We also used t-tests to analyze GSI side of the lateral body cavity. Testes were weighed and GSI was variation between categories I and III for our 1997/98 sample, calculated using the following formula: GSI=(mg/mb )*100,where independently for the 2 months (Nov. ‘97 and Jan. ’98) with the largest mg =the mass of the gonads and mb =bodymass. sample sizes. Relationships between androgen levels (both 11-KT and T) and body EOD recordings size were assessed using simple linear regression analyses. We also calculated linear regressions for T by body size for males and females Electric organ discharges (EODs) were recorded from individual fish alone. We quantified the relationship between 11-KT and T levels using placed in an 18 L (15×30×40 cm) tank filled to about 2/3 of its volume a Pearson Product moment correlation. Variation in androgen levels by with water from the Amazon river (Temp=25.8° C, pH=5.69, morphological categories was assessed using analyses of covariance Conductivity=94 μs/cm , Oxygen=70.8% and Total solids=0.06 g/ (ANCOVA), with morphological category as a dependent factor and L). An aerator was kept in the recording tank during recordings and the capture week as a covariate, to control for seasonal effects. Differences water was replaced by new water, from the same collection site, after among morphological categories were assessed post-hoc using Fisher's every third or fourth fish recorded. The EOD signal was differentially protected least significant difference (PLSD) tests. The same ANCOVA/ captured through a pair of gold electrodes positioned in front of the head PLSD approach was then applied to test for male category differences in and behind the tail of the fish, approximately equidistant of the EODf. pb0.05 was considered statistically significant. All statistical tests, reference electrode placed in the mid distance between them. The signal including tests of residuals (normality, independence, and constancy of was then differentially amplified (A-M Systems Sequim, WA Model variances) were performed using Statistica (StatSoft, version 7), Systat 3000, 100-10000X), digitized (Digidata 1440, Axon Instruments, (SigmaPlot, version 10.2), or Excel (Microsoft, version 12.1.0). Sunnyvale, CA), and recorded on a computer running Axoscope (version 10.0, Axon Instruments, Sunnyvale, CA). All recordings were done Results between 10 AM and 6 PM. The fundamental EOD frequency (EODf) was estimated from the digitized signal from the peak of a Fast Fourier Body size Transform (Blackman window, 65536 points) generated in Clampfit (version 9.0, Axon). Males with prominent teeth were significantly larger than males with incipient teeth (category I versus II, Table 1,ST,Mann–Whitney U Blood collection test, U=39.5, p =0.055). Furthermore, males with teeth were significantly larger than males without teeth (categories I and II Fish were anesthetized with clove oil (0.2 ml/5 l of water) for about combined vs. category III, t-test, df=24, t=2.71, p=0.01). 3 min until all fin movements stopped. A 1 ml heparinized syringe with a 30G needle was used to extract blood from the vertebral sinus. About Gonadosomatic indices (GSI) 0.7–1.0 ml of blood was stored in a heparinized 1.5 ml MaxyClear Axygen microtube and maintained on ice for about two hours until In our 2006/07 sample, GSI was highly consistent across categories centrifugation. The blood samples were centrifuged at 6000 rpm for [category I, 0.147±0.060 (x±SD), category II, 0.138±0.051, category 15 min, and the plasma was removed and stored in 0.6 ml MaxyClear III, 0.130±0.053]. There were no statistical differences in GSI across Axygen microtube tube at −40° C until hormone analyses were categories (category I vs. II, Mann–Whitney U test, U=10, p=1.0; performed. categories I and II combined vs. category III, t-test, df=16, t=0.46, After blood collection, specimens were: (1) further anesthetized p=0.65). In our 1997/98 sample (Fig. 3), however, GSI was significantly with an overdose of clove oil, (2) measured for total length, which was higher in fish with teeth (category I) than in fish without teeth (category taken from the tip of the snout to the end of the caudal finor,when III), during both the months of November (t-test, df=13, t=4.52, fishes had no caudal fin, the end of the body (3) labeled, (4) fixed in 10% pb0.001) and January (t-test, df=13, t=5.29, pb 0.001). Comparing formalin for at least a week and transferred to 70% ethanol, and (5) the two samples, we find that the GSIs were substantially higher in sexed by examination of gonads through incisions in the abdominal 1997/98 than in 2006/07. During 1997/98, both male categories cavity. All fish were deposited at the Fish Collection at the Instituto exhibited a steady increase of GSI as the months progressed (Fig. 3). Nacional de Pesquisas da Amazônia. Androgens, body size, and morph categories Androgen assays 11-KT levels did not vary significantly with body size, either for the Plasma concentrations of 11-ketotestosterone (11-KT) and testos- entire pooled sample (simple linear regression, n=27 males, df=26, terone (T) were measured using commercially available kits for enzyme F=3.18,p=0.09,Fig. 4A) or for each male category separately (simple immunoassay [Cayman Chemical Co., Ann Arbor, MI (11-KT), Assay linear regression, pN0.08 for all). Similarly, T levels did not vary Designs, Ann Arbor, MI (T)]. Plasma samples were diluted in assay significantly with body size across the entire pooled sample (simple 664 C. Cox Fernandes et al. / Hormones and Behavior 58 (2010) 660–668

Fig. 3. Gonadosomatic index (GSI) values of male S. nattereri with teeth (category I) (gray bars) and without teeth (category III) (black bars), during 1997 and 1998. Means and standard deviations shown.

linear regression, n=25 males plus 8 females, df=32, F=3.11, categories combined, T varied negatively with body size (simple linear p=0.09, Fig. 4B), or for the three individual male categories separately regression, df =24, F=4.02, p=0.05). Females alone showed no such (simple linear regression, pN0.18 for all). But for the three male relationship (simple linear regression, df=7, F=1.57, p=0.25). T concentrations did not differ significantly by sex (t-test, n=28 males and 8 females, df=34, t=-1.21, p=0.23). 11-KT and T concentrations were only weakly correlated with each other (Pearson product moment correlation 0.062). Males of different morphological categories varied significantly in levels of 11-KT (ANCOVA, main effect of teeth,F(2, 28)=3.35, p=0.002; Fig. 5A). Males from category I had significantly higher 11- KT levels than did males from category III (PLSD, p=0.03), whereas males from category II had intermediate 11-KT levels that did not differ significantly from those of either category I or III (PLSD, pN0.1). Male morphological categories did not vary significantly in levels of T (main effect of teeth, F (2, 26)=−1.23, p=0.23; Fig. 5B).

Sex and morph differences in EOD frequency

Males and females did not differ significantly in EOD frequency (EODf) (t-test, df=37, t=1.58, p=0.12, Fig. 6). Within males, however, EODf was found to differ between morphological categories (ANCOVA, main effect of teeth, F(2,35)=3.91, p=0.03; Fig. 6). Fish from category I showed significantly higher EODf than did fish from categories II (PLSD, p=0.002) or III (PLSD, p=0.025). Neither T nor 11-KT concentrations correlated significantly with EOD frequency, in either males with teeth (categories I and II) or males without teeth (category III) (pN0.15 for all).

Seasonality in androgens

Seasonal variation in 11-KT and T concentrations, in relation to morph category and month of capture, is reported in Fig. 7.11-KTlevels are seen to peak for all categories during December and then to drop off (Fig. 7A). The very highest level of 11-KT is seen in a category III fish, in spite of the trend for fish in this category to show low 11-KT levels (Figs. 5A and 7A). This individual specimen is unusual for category III males in having an enlarged facial profile typical of a category II male. T levels are seen to peak earlier in the season, in November (Fig. 7B), particularly in males of category III.

Discussion

Fig. 4. A. The relationship between 11-ketotestosterone plasma concentration and ST (males n=27). B. The relationship between testosterone plasma concentration and ST Our main result is that male morphs of S. nattereri differed (28 males and 8 females). significantly in plasma concentrations of 11-KT, with toothed males C. Cox Fernandes et al. / Hormones and Behavior 58 (2010) 660–668 665

Fig. 6. EODs frequency of males of the three categories (I, II and III) and females. Means and standard errors shown.

the dentary bulb of this fish. The distinct morphology and elevated 11-KT levels in this fish suggests a possible lag time between 11-KT levels and tooth eruption. In this case tooth presence might depend on circulating hormone levels but may take a few weeks for the teeth to form. If this fish were toothed when captured, the pattern shown in Fig. 5A would have been even stronger. On the other hand, T levels did not vary significantly among male categories, although there was a trend for category III males to show the highest T levels (Figs. 5B and 7B). In general, morphological and behavioral correlates of T are less consistent across species, with some species showing greater T levels in courting males, but other species showing greater T levels in noncourting males (Kindler et al., 1989; Brantley et al., 1993; Knapp and Neff, 2007). Sneaker males often have higher levels of T because they have very large testes and engage in sperm competition. High T levels in sneakers may promote spermato-

Fig. 5. A. Concentration of 11-KT in males of the three categories (I, II and III). genesis without triggering 11-KT-dependent traits that are present in B. Concentration of T in females and males of the three categories (I, II and III). Means territorial males. and standard errors shown. Our data also provide insight into relationships between male morphology and reproductive condition. Males with teeth were found showing higher levels of 11-KT than did toothless males (Fig. 5A). This to be larger (Table 1) and also, for our 1997/98 sample, to have larger finding is consistent with results in other species that have associated gonadosomatic indices (GSI, Fig. 3). Both results suggest that toothed morphology, male androgen levels, and reproductive tactics. In Atlantic males are in or approaching reproductive condition. More specific salmon, for example, skeleton growth and hook-nose (kype) formation inferences about reproductive tactics may be drawn using forensics. Cox in males is under control of androgens (Witten and Hall, 2003), and 11- Fernandes et al. (2009) reported that many toothed males (Category I KT levels are high in males with well-developed testes (Crim et al., only) show scars and scratches on their heads and dorsolateral body 1992). In the plainfin midshipman, Porichthys notatus, large territorial areas, whereas toothless males do not show such marks. These “type 1” males court females directly with vocalizations, compete with observations strongly suggest that teeth are used in aggressive contexts other males, and have significantly elevated 11-KT levels, as compared among toothed males (Fig. 1BandC,Cox Fernandes et al., 2009). to “type 2” sneaker males (Bass and Marchaterre, 1989; Brantley et al., Because teeth of S. nattereri (like kype of Atlantic salmon) are likely to be 1993). Similar elevation of 11-KT in parental/territorial male types used in aggressive contexts, and given that androgens can regulate occurs in bluegill sunfishes and several species of scarids, labrids and aggression, we suspect that 11-KT levels increases aggression in fish that blennids (Brantley et al., 1993; Oliveira et al., 2001; Knapp and Neff, have teeth. But our data do not allow direct examination of this 2007). hypothesis, given that we measured 11-KT mainly at one time point, The finding that 11-KT concentrations are significantly higher in during the month of December. We speculate that other androgens may toothed than non-toothed male S. nattereri (Fig. 5A) holds in spite of a be involved in tooth growth before the reproductive season, but single category III male who shows the highest 11-KT levels in our entire evidence towards this end would require further observational and sample (4.7 ng/ml, Fig. 7A). This individual was classified as Category III experimental work. because it lacks obvious teeth in x-ray. But this individual is distinct from The absence of scars and scratches in non-toothed males implies that other Category III fish in being relatively large, in having an enlarged they do not engage in aggressive combat while seeking mates. During head and face, and a swollen bulb on its dentary—traits more typical of dissection we noted that the testes of non-toothed males were mature. If Category I fish. We speculate that teeth would have soon emerged from non-toothed males had shown immature testes, one might have 666 C. Cox Fernandes et al. / Hormones and Behavior 58 (2010) 660–668

Non-competitive males might also occur in other species of this family (Apteronotidae). Two examples are hasemani and bonapartii.InP. hasemani, two male morphs used to be considered different species. One of these morphs is characterized by extremely long snouts, while the other morph shows a range of variation in snout length, from short to intermediate (Cox Fernandes et al., 2002). In A. bonapartii, males also vary in snout length, although not as much as in P. hasemani (Hilton and Cox Fernandes, 2006). In both species, males with longer snouts tend to show more advanced stages of sexual maturity. As we suggested for S. nattereri, smaller- snouted males of these species might engage in alternative repro- ductive tactics. However, it is premature to exclude the hypothesis that longer-snouted males are simply older individuals who have had enough time to develop the long snout. To learn if short-snouted males would develop long snouts would require lab studies that track growth and maturation. Another factor that influences reproduction is seasonality. Repro- duction of Amazonian fish is known to vary with flood cycles, which themselves can vary considerably from year to year. Our GSI data shows a notable difference in fish reproductive condition between 1997/98 and 2006/07. More specifically, GSIs were uniformly low in 2006/07, but were larger and showed significant variation by category in 1997/98. During 1997/98, GSI was observed to increase steadily from November through March (Fig. 3). We suppose that fish in the earlier sample were reproducing, whereas fish in the later sample were in an earlier maturational stage. These kinds of year-to- year differences might be related to inter-annual variations in the Amazon flood pulse (Fig. 2). During November 1997, the river reached a much lower level than it did during November 2006. In fact, the low- water level of 1997 was the very lowest recorded over a 51-year time span (1953–2003), whereas the low-water level of 2006 was considered typical (Bitterncourt and Amadio, 2007). This variation may influence environmental cues that trigger the onset and intensity of reproduction, leading to earlier or later breeding seasons for fish. A related topic worth exploring will be if and how different male morphs accumulate reserves for future reproduction. If males without teeth do not spend time and energy fighting for females, they may in principle continue to feed during the months of reproduction, improving their ability to breed at a later date (“capital vs. income breeders” sensu Schutz et al., 2009). Competitive males might also face constraints in their ability to feed efficiently, given the large number and unusual position of their teeth, and also given that these fishes eat small food items that do not necessarily require teeth. Additionally, competitive males may face energetic limits in their ability to engage in lateral migrations, in which fish leave and later re-enter the Fig. 7. A. Seasonal variation in plasma concentration of 11-KT among S. nattereri males. B. Seasonal variation in plasma concentration of T among males. Open squares Amazon floodplain (Fernandes, 1997). represent males with prominent teeth (category I), black circles represent males with Our work also showed differences among male categories in electric incipient teeth (category II), and open triangles represent males without teeth organ discharges, with toothed males showing significantly higher EOD (category III). frequencies than either toothless males or females. Sexual dimorphism in EOD frequency varies substantially across apteronotid species. In brown ghost knifefish (Apteronotus leptorhynchus),malesproduceEODsat speculated that they were juvenile fish that eventually would develop frequencies that are 100–200 Hz higher than those of females, with little into category II and then category I fish, such that the categories could be overlap between the sexes (Meyer et al., 1987; Dunlap et al., 1998; considered consecutive developmental stages. However, the non- Kolodziejski et al., 2005). In black ghost knifefish (Apteronotus albifrons), toothed males appear to be sexually mature. This suggests that they male EODs are lower than those of females (Dunlap et al., 1998; probably engage in some alternative reproductive tactic. Non-territorial Kolodziejski et al., 2005). In other apteronotids (A. bonapartii and (e.g., sneaker or parasitic) males often have GSIs and T levels that are devenanzii), EOD frequency is sexually monomorphic higher than those of territorial males, because they produce large (Zhou and Smith, 2006; Ho et al., 2007). EOD frequency varies within sex quantities of sperm, allowing them to engage successfully in sperm in some species. In A. leptorhynchus, dominant males have higher EOD competition (Taborsky, 1998; Oliveira et al., 2001). Dominant male frequency than subordinate males, and similar relationships have been morphs, when without rivals, usually are positioned closer to females reported between EOD frequency and dominance in females (Hagedorn during spawning, thus requiring less sperm to fertilize eggs (Taborsky, and Heiligenberg, 1985; Dunlap, 2002; Dunlap and Oliveri, 2002; 1998). Moreover, non-competitive sneaker males may experience Tallarovic and Zakon, 2005). Thus, in some other apteronotid species greater selection for sperm competition, to better enable matings, EOD frequency is a reliable signal of sex and social status. It is likely that which could drive an increase in their relative testes size. In our fish, EOD frequency serves as a badge that identifies males with teeth. This though, GSI and T were not higher in non-toothed males. signal could function in intrasexual interactions by deterring physical C. Cox Fernandes et al. / Hormones and Behavior 58 (2010) 660–668 667 contests with toothed males or in courtship if females preferred to mate Dunlap, K.D., 2002. Hormonal and body size correlates of electrocommunication behavior during dyadic interactions in a weakly electric fish, Apteronotus with toothed males. Playback studies to test the responses of S. nattereri leptorhynchus. Horm. Behav. 41, 187–194. males and females to different EODs are needed to test these hypotheses. Dunlap, K.D., Oliveri, L.M., 2002. Retreat site selection and social organization in captive In conclusion, much remains to be learned about the influence of electric fish, Apteronotus leptorhynchus. J. Comp. Physiol. A 188, 469–477. Dunlap, K.D., McAnelly, M.L., Zakon, H.H., 1997. Estrogen modifies an electrocommunica- environmental, seasonal, and hormonal factors on the morphology and tion signal by altering the electrocyte sodium current in an electric fish, Sternopygus.J. behavior of S. nattereri. Our study suggests that 11-KT likely induces and Neurosci. 17, 2869–2875. regulates tooth development, and as such may influence male–male Dunlap, K.D., Thomas, P., Zakon, H.H., 1998. Diversity of sexual dimorphism in contests during the breeding season. Our study also suggests that 11-KT electrocommunication signals and its androgen regulation in a genus of electric fish, Apteronotus. J. Comp. Physiol. A. 183, 77–86. and male morph correlate with body size, testes growth, and electric Ehmcke, J., Wistuba, J., Clemen, G., Schlatt, S., 2003. Targeted expression of androgen organ discharge frequencies, additional factors that may impact repro- receptors in tooth-forming tissues of a neotropical salamander (Bolitoglossa fi ductive effort and competition. Finally, our study suggests that these schizodactyla) enables region-speci c formation of dimorphic types of teeth in the males. Gen. Comp. Endocrinol. 134, 26–35. factors may all vary in their expression within and across years, in concert Emlen, D.J., 2008. The evolution of animal weapons. Annu. Rev. Ecol. Syst. Evol. 39, with the highly variable environmental conditions of the Amazon river. A 387–413. comprehensive understanding of androgen variation and function in Fernandes, C.C., 1997. Lateral migrations of fishes in Amazon floodplains. Ecol. Freshw. – fi Fish 6, 36 44. these shes will thus require not just attention to hormonal mechanisms Fernandes, C.C., Podos, J., Lundberg, J.G., 2004. Amazonian Ecology: tributaries enhance but also continued field work over multi-year time scales. the diversity of electric fishes. Science 305, 1960–1962. Gorbman, A., Dickhoff, W.W., Vigna, S.R., Clark, N.B., Ralph, C.L., 1983. Comparative Endocrinology. John Wiley & Sons, New York. Acknowledgments Hagedorn, M., Heiligenberg, W., 1985. Court and sparks: electric signals in the courship and mating of gymnotoid electric fish. Anim. Behav. 33, 254–265. Our research benefited greatly from the following people and their Hews, D.K., Moore, M.C., 1995. Influence of androgens on differentiation of secondary sex characters in tree lizards, Urosaurus ornatus. Gen. Comp. Endocrinol. 97, 86–102. institutions for lab facilities, equipment, expertise, helping on field Hews, D.K., Knapp, R., Moore, M.C., 1994. Early exposure to androgens affects adult trips and loan of specimens. At INPA: Maria Mercedes Bittencourt, expression of alternative male types in tree lizards. Horm. Behav. 28, 96–115. Vera Ferreira da Silva, technicians and fishermen, Coordenação de Hilton, E.J., Cox Fernandes, C., 2006. Sexual Dimorphism in Apteronotus bonapartii – Pesquisas em Biologia Aquática; Maristerra R. Lemes, Laboratório de (Castelnau, 1855) (: Apteronotidae). Copeia 4, 826 833. Herfeld, S., Moller, P., 1998. Effects of 17 Methyltestosterone on sexually dimorphic Genética e Biologia Reprodutiva de Plantas; Jaqueline da Silva Batista characters in the weakly discharging electric fish, Brienomyrus niger (Günter, 1866) and Kyara Formiga-Aquino, Laboratório de Biologia Molecular; Vera (Mormyridae): electric organ discharge, ventral body wall indentation and anal-fin – M. F. Almeida-Val and Adalberto L. Val, Maria de Nazaré Paula da Silva, ray bone expansion. Horm. Behav. 34, 303 319. fi Ho, W.W., Cox Fernandes, C., Alves-Gomes, J.A., and Smith, G.T., 2007. Electrocommu- Laboratório Eco siologia e Evolução Molecular; Lucia Rapp Py-Daniel, nication signals in the weakly electric fish "Apteronotus" bonapartii are sexually Coleção de Peixes; and students at the Laboratório de Fisiologia monomorphic and uncorrelated with androgens. Proceedings of the International Comportametal e Evolução, especially Augusto Bücher and Gilberto Congress of Neuroethology (vol. 8, p. 89) Vancouver, BC, (July, 2007). Junk, W.J., 1984. Ecology of the várzea, floodplain of Amazonian white-water rivers. In: Schwertmer. At the Academy of Sciences in Philadelphia, John G. Dumont, H.J. (Ed.), The Amazon Limnology and landscape ecology of a mighty tropical Lundberg and Kyle Luckenbill. At UMass Amherst we thank Larry river and its basin. Dr. W. Junk Publishers, Dordrecht, Netherland, pp. 215–243. Schwartz, Tom Zoeller, Duncan Irschick and Matt Gaucher. Special Junk, W.J., Bayley, P.B., Sparks, R.E., 1989. The flood pulse concept in river-floodplain systems. In: Dodge, D.P. (Ed.), Proceedings of the International Large River thanks to two anonymous reviewers who provided wonderfully Symposium: Can. Spec. Publ. Fish. Aquat. Sci., 106, pp. 110–127. constructive comments on a previous draft. This work was supported Kime, D.E., 1993. “Classical” and “non-classical” reproductive steroids in fish. Rev. Fish by a CNPq grant (CNPq#481691/2004-9 to J.A.G.) and the UMass Biol. Fish. 3, 160–180. Kindler, P.M., Philipp, D.P., Gross, M.R., Bahr, J.M., 1989. Serum 11-ketotestosterone and Natural History Collections (C.C.F.). testosterone concentrations associated with reproduction in male bluegill (Lepomis macrochirus: Centrarchidae). Gen. Comp. Endocrinol. 75, 446–453. References Kirschbaum, F., 1979. Reproduction of the weakly electric fish Eigenmannia virescens in captivity. Behav. Ecol. Sociobiol. 4, 331–355. Araujo-Lima, C.A.R.M., Oliveira, E.C., 1998. Transport of larval fish in the Amazon. J. Fish Knapp, R., Neff, B.D., 2007. Steroid hormones in bluegill, a species with males Biol. 53, 297–306. alternative reproductive tactics including female mimicry. Biol. Lett. 3, 628–631. Baldwin, F.M., Goldin, H.S., 1939. Effects of testosterone propionate on the female Kolodziejski, J.A., Nelson, B.S., Smith, G.T., 2005. Sex and species differences in viviparous teleost, Xiphophorus helleri Heckel. Proc. Soc. Exp. Biol. Med. 42, 813–819. neuromodulatory input to a premotor nucleus: a comparative study of substance P Bass, A.H., Marchaterre, M.A., 1989. Sound-generating (sonic) motor system in a teleost and communication behavior in weakly electric fish. J. Neurobiol. 62, 299–315. fish (Porichthys notatus): sexual polymorphism in the ultrastructure of myofibrils. J. Landsman, R.E., Moller, P., 1988. Testosterone changes the lectric organ discharge and Comp. Neurol. 286, 141–153. external morphology of the mormyrid fish, Gnathonemus petersii (Mormyriformes). Bass, A.H., Denizot, J.P., Marchaterre, M.A., 1986. Ultrastructural features and hormone- Experientia 44, 900–903. dependent sex differences of mormyrid electric organs. J. Comp. Neurol. 254, 5110528. Lofts, B., Pickford, G.E., Atz, J.W., 1966. Effects of methyl testosterone on the testes of a Beach, F.A., 1945. Bisexual mating behavior in the male rat: effects of castration and hypophysectomized cyprinodont fish, Fundulus heteroclitus. Gen. Comp. Endocrinol. 6, hormone administration. Physiol. Zool. 18, 390–402. 74–88. Bitterncourt, M.M., Amadio, S.A., 2007. Proposta para identificação rápida dos períodos Lopez-Rojas, H., Lundberg, J.G., Marsh, E., 1984. Design and operation of a small trawling hidrológicos em areas de várzea do rio Solimões-Amazonas nas proximidades de apparatus for use with dugout canoes. N. Am. J. Fish. Manage. 4, 331–334. Manaus. Acta Amazon. 37, 303–308. Lundberg, J.G., Cox Fernandes, C., Albert, J.A., Garcia, M., 1996. Magosternarchus, a new Brantley, R.K., Wingfield, J.C., Bass, A.H., 1993. Sex steroid levels in Porichthys notatus,a genus with two new species of electric fishes (Gymnotiformes: Apteronotidae) fish with alternative reproductive tactics, and a review of the hormonal bases for from the Amazon River Basin, South America. Copeia 3, 657–670. male dimorphism among teleost fishes. Horm. Behav. 27, 332–347. Marrero, C., Taphorn, D.C., 1991. Notas sobre la historia natural y la distribucion de los Cox, R.M., Stenquist, D.S., Henningsen, J.P., Calsbeek, R., 2009. Manipulating testoster- peces Gymnotiformes en la cuenca del rio Apure y otros rios de la Orinoquia. one to assess links between behavior, morphology, and performance in the brown Biollania 8, 123–142. anole Anolis sagrei. Physiol. Biochem. Zool. 82, 686–698. Moore, M.C., Thompson, C.W., 1990. Field endocrinology of reptiles: hormonal control Cox Fernandes, C., 1995. Diversity, distribution and community structure of electric fishes of alternative male reproductive tactics. In: Epple, A., Scanes, C.G., Stetson, M.H. (Gymnotiformes) in the channels of the Amazon river system, Brazil. Ph. D. (Eds.), Progress in Comparative Endocrinology. Wiley-Liss, New York, pp. 685–690. Dissertation, Duke University, Durham. Meyer, J.H., Leong, M., Keller, C.H., 1987. Hormone-induced and maturational changes Cox Fernandes, C., 1998. Sex-related morphological variation in two species of apteronotid in electric organ discharges and electroreceptor tuning in the weakly electric fish fishes (Gymnotiformes) from the Amazon river basin. Copeia 3, 730–735. Apteronotus. J. Comp. Physiol. A 160, 385–394. Cox Fernandes, C., Lundberg, J.G., Riginos, C., 2002. The largest of all electric-fish snouts: Myers, G.S., 1936. A new genus of gymnotid eels from the Peruvian Amazon. Proc. Biol. hypermorphic facial growth in male Apteronotus hasemani, and comments on the Soc. Wash. 49, 115–116. nominal species A. anas (Gymnotiformes: Apteronotidae). Copeia 1, 52–61. Noble, G.K., Pope, S.H., 1929. The modification of the cloaca and teeth of the adult Cox Fernandes, C., Lundberg, J.G., Sullivan, J.P., 2009. Oedemognathus exodon and salamander, Desmognathus, by testicular transplants and by castration. Br. J. Exp. Sternarchogiton nattereri (Apteronotidae, Gymnotiformes): the case for sexual Biol. 6, 399–411. dimorphism and conspecificity. Proc. Acad. Nat. Sci. Phila. 158, 193–207. Oliveira, R.F., Canario, A.V.M., Grober, M.S., Serrão Santos, R., 2001. Endocrine correlates Crim, L.W., Wilson, C.E., So, Y.P., Idler, D.R., Johnston, C.E., 1992. Feeding, reconditioning, of male polymorphism and alternative reproductive tactics in the Azorean rockpool and rematuration responses of captive Atlantic Salmon (Salmo salar) kelt. Can. J. blenny, Parablennius sanguinolentus parvicornis. Gen. Comp. Endocrinol. 121, Fish. Aquat. Sci. 49, 1835–1842. 278–288. 668 C. Cox Fernandes et al. / Hormones and Behavior 58 (2010) 660–668

Peters, A., Astheimer, L.B., Boland, C.R.J., Cockburn, A., 2000. Testosterone is involved in Tallarovic, S.K., Zakon, H.H., 2005. Electric organ discharge frequency jamming during acquisition and maintenance of sexually selected male plumage in superb fairy- social interactions in brown ghost knifefish, Apteronotus leptorhynchus. Anim. wrens, Malurus cyaneus. Behav. Ecol. Sociobiol. 47, 438–445. Behav. 70, 1355–1365. Saraiva, J.L., Gonçalves, D.M., Oliveira, R.F., 2010. Environmental modulation of vom Saal, F.S., 1983. Models of early hormone effects on intrasex aggression in mice. In: androgen levels and secondary sex characters in two populations of the peacock Svare, B.B. (Ed.), Hormones and Aggressive Behavior, pp. 197–222. Springer, New blenny Salaria pavo. Horm. Behav. 57, 192–197. York. Schutz, D., Pachler, G., Ripmeester, E., Goffinet, O., Taborsky, M., 2009. Reproductive Witten, P.E., Hall, B.K., 2003. Seasonal changes in male Atlantic salmon (Salmo salar investment of giants and dwarfs: specialized tactics in a cichlid fish with alternative L.): remodelling and regression of the kype after spawning. J. Anat. 203, male morphs. Funct. Ecol. 1–10. 435–450. Schwassmann, H.O.W., 1978. Time of annual spawning and reproductive strategies in Zakon, H.H., Dunlap, K.D., 1999. Sex steroids and communication signals in electri fish: a Amazonian fishes. In: Thorpe, J.E. (Ed.), Rhythmic Activity of Fishes, London. tale of two species. Brain Behav. Evol. 54, 61–69. Academic Press, pp. 187–200. Zakon, H.H., Thomas, P., Yan, Hong-Young, 1991. Electric organ dischargen frequency Taborsky, M., 1997. Bourgeois and parasitic tactics: do we need collective, functional and plasma sex steroid levels during gonadal recrudescence in a natural terms for alternative reproductive behaviours? Behav. Ecol. Sociobiol. 41, 361–362. population of the weakly electric fish Sternopygus macrurus.J.Comp.Physiol.A Taborsky, M., 1998. Sperm competition in fish: “bourgeois” males and parasitic 169, 493–499. spawning. Trends Ecol. Evol. 13, 222–227. Zhou, M.C., Smith, G.G., 2006. Structure and sexual dimorphism of the electrocommu- Takeuchi, K., 1967. Large tooth formation in female medaka, Oryzias latipes, given nication signals of the weakly electric fish, Adontosternarchus devenanzii. J. Exp. methyl testosterone. J. Dent. Res. 46, 50. Biol. 209, 4809–4818.