Interspecific mate choice in sailfin and shortfin species of mollies

ANIMAL BEHAVIOUR, 1998, 56, 1145–1154 Artcle No. ar980909

Interspecific mate choice in sailfin and shortfin species of mollies MARGARET B. PTACEK

Department of Biological Science, Florida State University, Tallahassee (Received 16 October 1997; initial acceptance 22 December 1997; final acceptance 24 April 1998; MS. number: A8045)

ABSTRACT I examined interspecific patterns of female choice for three species of mollies: the ‘sailfin’ species, P. latipinna and two ‘shortfin’ species, P. orri and P. mexicana. Females of the sailfin species consistently preferred conspecific males across all male treatment combinations containing conspecifics. Females of the two shortfin species had lower levels of female preference and spent less time interacting with males during trials. The only consistent pattern of female choice in either shortfin species was a preference for the sailfin P. latipinna males over heterospecific shortfin males, but not over conspecifics. This finding suggests a potential bias by shortfin females for the sailfin male phenotype. Males of the three species differed in both behaviour and morphology. Length and relative position of the dorsal fin reliably separated P. latipinna from the two shortfin species. Body depth and shape distinguished males of the two shortfin species. Dorsal fin height was not a reliable cue for distinguishing sailfin males from males of either shortfin species. Males from different populations of the sailfin species P. latipinna differ in both size-specific behavioural rates and size-specific morphometry, particularly in the overall size-adjusted area of the dorsal fin. Increased rates of courtship displays and increased area of the dorsal fin separate males of P. latipinna from males of the two shortfin species as well. If the same traits used by females of P. latipinna for intraspecific discrimination are also important species-recognition signals, this suggests a potential role for sexual selection in contributing to the divergence of male phenotype during speciation of sailfin mollies. 

intraspecific and interspecific mate choice. One such study in the butterfly Pieris occidentalis (Weirnasz 1989; Weirnasz & Kingsolver 1992) has experimentally documented that females do use the same male character (dorsal forewing colour pattern) in both intraspecific and interspecific mate choice. This result implies that species recognition is just an extension of intraspecific mating preferences (e.g. Verrell 1988; Ryan & Rand 1993; Endler & Houde 1995). How often such a pattern of female choice occurs in other taxa remains unclear. This study examines interspecific mate choice in three species of mollies (Poeciliidae: Poecilia). Mollies are an ideal group in which to examine the role of sexual selection in generating interspecific patterns of phenotypic diversity; differences in the two major species complexes of mollies are found primarily in males and are strongly associated with divergence in their mating systems. Taxonomists (Hubbs 1933; Miller 1975) divide mollies into two groups: the P. sphenops species complex contains 10 species of ‘shortfin’ mollies and the P. latipinna species complex contains three species of ‘sailfin’ mollies. Males of these two species complexes differ dramatically in both morphological and behavioural traits. Sailfin species are characterized by a sexual

The role of sexual selection in promoting speciation is one of the most important but least understood problems in evolutionary biology. This is in part because studies of species recognition and studies of sexual selection have been done mostly in isolation (Endler & Houde 1995). However, as Ryan & Rand (1993) point out, both species recognition and sexual selection are essentially problems in animal communication, and both can result from how females respond to variation in male traits. Differences in the pattern of female preferences among conspecific populations for male traits that act as mating signals can result in population-specific values of these traits in males (Ryan & Wilczynski 1991; Ryan et al. 1992; Endler & Houde 1995; Ptacek & Travis 1997). Such divergence in mate recognition signals among populations can lead ultimately to reproductive isolation and speciation (e.g. Darwin 1871; Lande 1981; Thornhill & Alcock 1983; West-Eberhard 1983, 1984; Kaneshiro & Boake 1987). Few studies, however, have examined whether females use the same phenotypically varying male traits in both Correspondence: M. B. Ptacek, Department of Biological Sciences, Idaho State University, Pocatello, ID 83209-8007, U.S.A. (e-mail: [email protected]). 0003–3472/98/111145+10 $30.00/0

1998 The Association for the Study of Animal Behaviour

1145



1998 The Association for the Study of Animal Behaviour

1146 ANIMAL BEHAVIOUR, 56, 5

dimorphism in which males possess a greatly enlarged dorsal fin (Regan 1913; Hubbs 1933; Parzefall 1969) that is erected and presented to the female in a courtship display (Parzefall 1969, 1979; Farr et al. 1986). Receptive females respond to this display by remaining stationary, folding the median fins, and sometimes twisting the abdomen to accept a copulation (Parzefall 1969; personal observation). Males of the sailfin species P. latipinna show low levels of intermale aggression, both in the field and the laboratory (Travis 1994), and are not known to form permanent male dominance hierarchies (Farr 1989). Thus, reproductive success in sailfin species appears to be more a function of female choice than strong male–male competition. Males of shortfin species do not show sexual dimorphism in fin morphology and in three species in which behaviour has been examined, P. sphenops, P. chica and P. mexicana (Parzefall 1969, 1979; Brett & Grosse 1982; Balsano et al. 1985; Woodhead & Armstrong 1985), only gonopodial thrusting has been observed. Gonopodial thrusting is an attempt at insemination without female cooperation; a male orients himself behind a female, brings his gonopodium (the modified anal fin that serves as an intromittent organ during internal fertilization) to a forward position, and attempts to insert it forcefully into the female’s gonopore. Gonopodial thrusting has no apparent signal function; there are no countersignals directed from the female towards the male (Farr 1989). Males of these shortfin species do not rely on female cooperation for mating and have adopted a completely different mating system than males of sailfin species. Reproductive success is determined primarily by a social structure based on male–male aggression (Farr 1989). Males of shortfin species, including P. chica (Miller 1975), P. sphenops (Parzefall 1969), P. mexicana and P. orri (personal observation) form permanent dominance hierarchies, with dominant males assuming a dark coloration strikingly different from that of females or subordinate males. Dominant males aggressively keep other males from gaining access to females and females are forcibly inseminated by these males through gonopodial thrusting. Thus, male–male competition appears to play a much larger role in determining male reproductive success than does female choice in shortfin molly species. Farr (1989) contends that female choice, the direct influence on male mating success, is limited to those species of poeciliids with a courtship display; for mollies, this would include the three sailfin species. Experimental studies have demonstrated female choice for at least one of the sailfin species, P. latipinna (Ptacek & Travis 1997). Females of P. latipinna prefer native males over foreign males in some population combinations. This result indicates that females from different populations exercise divergent preferences for certain male phenotypic traits (i.e. dorsal fin size and courtship display rates) that vary among conspecific populations. This result implies an active role for sexual selection in contributing to the maintenance of the behavioural or morphological distinctions among populations of males. With sufficiently divergent female preferences, such male traits might

ultimately function in premating isolation between species. To test this hypothesis, I examined patterns of interspecific female preferences for three species, one sailfin species, P. latipinna, and two shortfin species, P. orri and P. mexicana. Specifically, I addressed the following questions. First, do females of P. latipinna prefer conspecifics over males of shortfin species? If so, this suggests that females are potentially using the same traits that distinguish conspecific males from different populations for species recognition as well. Second, do females of shortfin species in which males lack courtship behaviours and associated morphological differences show any pattern of female choice? If shortfin females do show preferences, do they prefer conspecific males or the sailfin male phenotype? Third, because P. latipinna differs dramatically from either shortfin species in both behaviour and morphology, I asked which specific phenotypic traits of males distinguish shortfin species from sailfin species and could potentially be reliable cues for females to use in species recognition? METHODS

Origin and Maintenance of Experimental Fish Individuals of P. orri were collected from a population in Belize and P. mexicana were collected from a population in the Rio Tigre near Tampico, Mexico. Stock populations of both species were established and have been maintained as randomly breeding populations in large tanks (750-litre) in the greenhouse facility at Florida State University since 1989. Individuals of P. latipinna were collected from the Mounds Pond population (Wakulla County, Florida) and have been maintained in J. Travis’ laboratory at Florida State University since 1995. All three populations used in this study are allopatric with respect to each other. However, P. latipinna and P. mexicana do occur sympatrically in portions of their range, and P. orri is sympatric with another sailfin species, P. velifera, in portions of its geographic distribution, but not in the population used in this study. Males and females of all three species were housed under similar conditions in the laboratory for at least 6 weeks prior to testing. I maintained fish in the laboratory in 75-litre aquaria, approximately 10 males and 10 females per aquarium, on a 14:10 h light:dark cycle at 25C. Males used as ‘choices’ were never housed in the same aquarium as the ‘tested’ females.

Observations of Behaviour I conducted female choice tests in a 75-litre aquarium (1223252 cm) that was divided into five sections of equal size. The two outer sections were separated by panels of clear Plexiglas with slits at the top of each to allow water flow between all three sections. This design allowed for visual cues and diffuse water-borne chemical cues, providing females with access to certain potential signals used in species recognition. Females could not

PTACEK: INTERSPECIFIC MATE CHOICE IN MOLLIES 1147

directly contact males in this design, limiting tactile cues, but only by restraining access of males to females can female choice be decoupled from male choice. A single male was placed in each of the end compartments of the choice tank. The three central sections were delineated only by markings on the front side of the tank so that I could accurately record the movements of the female among these three central sections. I minimized disturbance to the fish by the presence of the observer by covering three sides and the bottom of the observation aquarium with an inner layer of black paper (to minimize reflections from the glass) and an outer layer of aluminium foil. The front glass of the aquarium was covered with one-way film. The tank was illuminated by a single 20-W Gro-lux fluorescent light bulb suspended 15 cm directly above the tank. After each trial, water was drained and the test aquarium washed with fresh water. I made all behavioural observations between 0800 and 1400 hours. Males were isolated approximately 24 h prior to testing to increase the probability that they would be sexually active during trials. This protocol produces rates of behaviours in males of P. latipinna that are typical of those observed in natural populations (Travis 1994). I controlled for receptivity of females by maintaining individual gravid females in 3.75-litre aquaria until they gave birth and testing them within 24 h of releasing their broods. Females of P. latipinna are receptive for a 2–3-day period upon initial maturation, or immediately after producing a brood, after which they are unreceptive to fertilization for 28–35 days (Farr & Travis 1986; Snelson et al. 1986; Travis 1989). By using the same protocol for females of all three species, I attempted to standardize receptivity for females of P. orri and P. mexicana as well. During each trial, the two object males were first placed in the choice tank, one in each of the end compartments, and given a 15-min acclimation period. The males were matched for size by using males that were within1 mm standard length of each other. The males were from one of three male treatment combinations (P. latipinna versus P. mexicana, P. orri versus P. latipinna, P. mexicana versus P. orri). The treatment combinations were designed to control for any biases by females towards one or the other side of the choice tank by placing a male of a particular species on the left side of the tank in one treatment combination, but on the right side of the tank in the next treatment combination. The test female was then placed in the centre compartment and given a further 15-min acclimation period during which she was allowed to move freely about the central compartment. The behavioural observations were then conducted for 10 min. I recorded the time in seconds that the female spent on the left or right side of the tank and in the neutral central section of the tank. I also recorded the number of any type of courtship displays by either male when the female was on his side of the tank.

Analysis of Female Choice Experiments I tested 12 females from each of the three species (P. latipinna, P. orri and P. mexicana), 36 females total,

once with each of the three male treatment combinations. I used a repeated measures ‘crossover’ design (Cochran & Cox 1957; Travis & Woodward 1989; Ptacek & Travis 1997), which allowed me to test for any effects of order of observation (first, second, or third combination experienced by the female) or sequence in which the combinations were offered. I measured a female’s preference as the difference between the time spent near the conspecific male and the time spent near the heterospecific male. In those combinations where neither male was a conspecific, the time spent near the male on the right side of the tank was subtracted from the time spent near the male on the left side of the tank. The amount of time spent in the central portion of the compartment was used to calculate a female’s level of ‘apathy’. I analysed the preference data in three steps. First, I performed an analysis of variance on the preference scores of the 108 trials to test for the effects of order and sequence. I examined residuals from the raw preference scores to verify that assumptions of parametric tests were met. Second, after verifying that there were no significant effects of order or sequence, I used analysis of variance to test preference scores for effects of species of female, female identity within species, the treatment combination of males, and the interaction of the combination of males with the species of female. Third, after finding a significant interaction of female species with male combination, I examined the average preference scores from each of the nine male treatment combinations (three combinations for each species of female) for significant deviations from zero using twotailed t tests to diagnose which combinations induced females to exercise a preference. This procedure follows the spirit of Fisher’s protected least significant difference test. I examined a female’s level of apathy by analysing the number of seconds that each female spent in the centre of her compartment using analyses of variance in the same manner as described above for preference scores. I examined a female’s level of ‘indecision’ by determining for each female the number of times that she crossed the central portion of the tank swimming from one test male to the other. Because this measure is based on count data, I first transformed each measure using a square root transformation (square root (N) + square root (N + 1); Freeman & Tukey 1950). After verifying with an analysis of variance that there was no significant effect of order of the trial or sequence of the treatment combinations, I used an analysis of variance to test the transformed values of the number of crosses for effects of species of female, female identity within species, the treatment combination of males, and the interaction of the combination of males with the species of female. Because the species of female was the only significant main effect, I used Tukey’s honestly significant difference test as an a posteriori comparison of all pairwise means for the three species of females. I examined residuals from the raw values of time in the centre and the transformed values of the number of crosses to verify that assumptions of parametric tests were met.

1148 ANIMAL BEHAVIOUR, 56, 5

LFFR

LLFR

PDD HCF

LDF DCP

DMB

LCF

LG PAD Figure 2. Locations of the 10 morphological measurements made on males of three species of mollies: LDF=length of the dorsal fin at the base, LFFR=length of the first dorsal fin ray, LLFR=length of the last dorsal fin ray, LG=length of the gonopodium, DCP=body depth at the caudal peduncle, DMB=body depth at mid-body, LCF=length of the caudal fin, HCF=height of the caudal fin, PDD=predorsal fin distance, PAD=preanal fin distance.

the linear combinations of the 10 morphological traits that best distinguished the three species.

RESULTS

Patterns of Interspecific Female Choice

Figure 1. Line drawings showing males of the three species of mollies used in female choice experiments: Poecilia latipinna (top), P. mexicana (middle), P. orri (bottom).

Measurements and Analysis of Interspecific Male Morphology Males of the three species used in this study differ dramatically from one another in certain morphological features (Fig. 1). To determine which morphological traits of males reliably distinguish the three species, I used an image analysis system to measure 10 morphological characters (Fig. 2) for 30 males of each of the three species. To determine whether interspecific differences existed in morphological shapes, independently of the effects of body size, I first transformed morphological character values into shape variables of the form [ln(trait) – ln(standard length)], and then performed a multivariate analysis of variance (MANOVA) on the 10 shape variables. By using this transformation, I could obtain tests of ‘shape’ differences among species that were independent of body size differences among species, and of any sizeassociated shape differences (Mosimann & James 1979). Univariate ANOVAs were then run to examine the patterns of variation in each variable separately. Significant tests of the species effect indicated which morphological traits were not the same in all species. I then performed a discriminant analysis using all 10 shape variables to determine which traits contributed to species distinctions. This discriminant analysis revealed

Female preference scores showed no effect of trial order (F8,60 =1.06, NS) or the sequence in which the male treatment combinations were presented (F2,60 =2.15, NS) for any of the three species of females tested. Therefore, the variance in preference scores could be partitioned among four sources. There were no significant effects of female identity within species (F33,66 =0.90, NS), but both main effects of female species (F2,33 =5.67, P<0.01) and male combination (F2,66 =11.65, P<0.001) significantly affected preference scores. The interaction between female species and male combination was also significant (F4,66 =5.18, P<0.01). Females of the sailfin species, P. latipinna, preferred conspecific males in both male treatment combinations that included them (Fig. 3). The average time spent with the conspecific male was greater in the P. latipinna versus P. mexicana male combination (T11 =7.96, P<0.001) than in the P. orri versus P. latipinna male combination (T11 =3.69, P<0.005). No preference was shown by females in the male treatment combination where both males were heterospecific shortfin species (T11 =0.08, NS). Thus, it appears that females of P. latipinna can reliably distinguish conspecifics and actively discriminate against heterospecific shortfin species. Females of the two shortfin species did show preferences in some male species combinations, but the patterns were not consistent across all combinations (Fig. 3). Females of P. orri spent significantly more time with conspecific males compared with the heterospecific shortfin species, P. mexicana (T11 =3.33, P<0.01). Females of this species also showed a preference for the sailfin phenotype compared with the shortfin heterospecific species, spending significantly more time with the P. latipinna male than the P. mexicana male (T11 =4.29, P<0.005). However, when given a choice between

PTACEK: INTERSPECIFIC MATE CHOICE IN MOLLIES 1149

300

600 (a) P. latipinna 500

Apathy level (s)

400 300 200 100

200

100

0

Average time (s) spent with male

600

(b) P. orri

0

500 400

200 100 0

(c) P. mexicana

500 400

P. latipinna P. orri P. mexicana

300 200 100 0

P. orri versus P. latipinna

P. mexicana versus P. orri

Figure 4. Responses of females to size-matched males presented in three different species combinations. The average±SE time (in seconds) that females spent in the centre of the choice tank for each combination during the 10-min observation period is plotted for P. latipinna ( ), P. orri ( ) and P. mexicana ( ).

300

600

P. latipinna versus P. mexicana

P. latipinna versus P. mexicana

P. orri versus P. latipinna

P. mexicana versus P. orri

Figure 3. Responses of females to size-matched males presented in three different species combinations. The average±SE time (in seconds) that females spent with each male for each combination during the 10-min observation period is plotted for P. latipinna (a) P. orri (b) and P. mexicana (c).

P. latipinna males and conspecifics, these females showed no significant preference (T11 =
1.22, NS). Females of the other shortfin species, P. mexicana, showed little preference for males of any species in any treatment combination (Fig. 3). The only significant result was in the P. orri versus P. latipinna male combination where females spent significantly more time with the sailfin males than with males of the heterospecific shortfin species (T11 =
2.65, P<0.05). No preference was

shown by these females in the P. latipinna versus P. mexicana male treatment combination (T11 =1.07, NS) or the P. mexicana versus P. orri male treatment combination (T11 =1.76, NS). Apathy levels, measured as time spent in the central section of the choice tank, showed no effect of the order of the observation (F8,60 =0.59, NS) or of the sequence of male treatment combinations (F2,60 =0.77, NS). The variance in apathy levels was then partitioned among the remaining four sources. Species did not differ in their level of apathy (F2,33 =1.56, NS); however, females within a species were heterogeneous (F33,66 =2.76, P<0.001). There was a significant main effect of male combination (F2,66 =3.42, P<0.05) and the interaction between female species and male combination was significant (F4,66 = 3.16, P<0.05). Females of all three species spent, on average, over 30% of the 600-s observation period in the neutral central section of the choice tank for two of the three male treatment combinations (Fig. 4). Females of P. latipinna had much lower levels of apathy in the P. latipinna versus P. mexicana combination than in the other two male treatment combinations. This difference in apathy levels between male combinations for females of P. latipinna accounts for the significant interaction between female species and male combination that was observed in the ANOVA. Females of the two shortfin species spent, on average, twice as much time in the central section of the tank during trials than did females of P. latipinna in this male combination. Apathy levels were similar for females of all three species in the P. orri versus P. latipinna male combination and the P. mexicana versus P. orri male combination. Females in a choice-tank design can have high apathy levels for one of two reasons. First apathy level will be

1150 ANIMAL BEHAVIOUR, 56, 5

Interspecific Differences among Males

Number of crosses

30

20

10

0

P. latipinna versus P. mexicana

P. orri versus P. latipinna

P. mexicana versus P. orri

Figure 5. Responses of females to size-matched males presented in three different species combinations. The average±SE number of times that females crossed the centre section of the choice tank for each combination during the 10-min observation period is plotted for P. latipinna ( ), P. orri ( ) and P. mexicana ( ).

high if a female spends most of the trial in the centre of the choice tank not interacting with males at all. Second, a female can have a high apathy score if she spends much of the trial swimming back and forth across the centre compartment, alternating the time spent with each male. To distinguish between these two influences on apathy level, I counted the number of crosses that a female made between the two males during a trial, and used this as a measure of a female’s level of indecision. Levels of indecision showed no effect of trial order (F8,60 =1.17, NS) nor the order in which the male combinations were presented (F2,60 =0.09, NS). There was significant heterogeneity among females within a species (F33,66 =2.76, P<0.001) and there were also significant differences among female species in their levels of indecision (F2,33 =5.44, P<0.01). The average level of indecision for females of P. orri was significantly higher than that of females of P. mexicana or P. latipinna (Tukey’s HSD: q=5.37, v=66, k=3, P<0.001; P. orri versus P. mexicana: q=30.03; P. orri versus P. latipinna: q=28.94) across all male combinations (Fig. 5). Females of P. latipinna and P. mexicana did not differ significantly in their average level of indecision (q=1.09). Both species had high levels of apathy in all treatment combinations (Fig. 4), but for different reasons. Females of P. mexicana spent a large portion of the trial in the centre of the compartment not interacting with males (high apathy level, but low level of indecision). Females of P. orri had high apathy levels and high levels of indecision, crossing the centre of the tank more than twice as often as females of the other two species (meanSE, P. orri=10.71.60; P. mexicana=4.60.50; P. latipinna=4.80.57). The main effect of male combination was not significant (F2,66 =2.11, NS), nor was the interaction between female species and male combination (F4,66 =0.57, NS).

Males of P. latipinna performed courtship displays, by raising and fanning the dorsal fin towards conspecific females (averageSE number of displays/min= 1.270.21) at a significantly higher rate (F2,69 =7.23, P<0.001) than they displayed to heterospecific females. They did display to heterospecific females, but at significantly lower rates (averageSE number of displays/ min=0.650.14 towards P. orri females (Tukey’s HSD, q=0.62, v=66, k=3, P<0.02); averagenumber of displays/min=0.460.10 towards P. mexicana females (Tukey’s HSD, q=0.81, v=66, k=3, P<0.002)), and in some trials males did not display at all to females of either shortfin species. Males of P. latipinna would swim alongside heterospecific females keeping the dorsal fin erect, but not fanning it in the characteristic display behaviour. Males of P. mexicana swam alongside females at the partition with their dorsal fin erect and pelvic fins extended away from the body. Some males would shimmy their bodies and bend them towards the females in a manner similar to the sigmoid behaviour of courting male guppies (Liley 1966; Farr 1975; personal observation). Males displayed to conspecific females at slightly higher rates (averageSE number of displays/min= 0.250.13) than to females of P. latipinna (0.180.07 displays/min) and females of P. orri (0.130.07 displays/ min), but differences in display rates in response to females of different species were not significant (F2,69 = 0.69, NS). Males of P. orri also swam alongside females at the partition with their dorsal fin erect. They did not, however, perform any behaviours that could be interpreted as courtship displays, such as sigmoid motion with their bodies or extension of pelvic fins. Males of P. orri keep their dorsal fin erect while swimming even in the absence of females (personal observation). While males of the three species differed in their levels of behaviour, all males kept the dorsal fin erect during interactions with females. Thus, I focused analyses on morphological traits and examined their potential as reliable indicators of species differences available to females of all three species during choice trials. The overall MANOVA revealed a highly significant species effect (Wilk’s lambda=0.011, F20,156 =67.34, P<0.001). Thus, morphology is not the same across the three species. Univariate ANOVAs showed that there were species differences in each of the 10 morphological shape variables measured (F2,87 value range 6.33–444.25; all Ps<0.0001; data and 10 analyses of variance available upon request). The discriminant function analysis used two canonical factors to correctly classify 89 of the 90 males to species (Table 1). Canonical factor 1 of the discriminant analysis was most highly correlated with length of the dorsal fin at the base and predorsal fin distance (Table 1). These two traits best describe the length and relative position of the dorsal fin of males. This factor distinguishes the sailfin P. latipinna from the two shortfin species (Fig. 6). Males of P. latipinna have longer dorsal fins; the origin is more anterior and there are more fin rays. Canonical factor 2

PTACEK: INTERSPECIFIC MATE CHOICE IN MOLLIES 1151

Table 1. Canonical loadings (correlations between morphological shape variables and the canonical factors) for the first two canonical factors from a discriminant analysis of morphological traits from three species of mollies

Length of dorsal fin at base Length of first dorsal fin ray Length of last dorsal fin ray Length of gonopodium Body depth at caudal peduncle Body depth at mid-body Length of caudal fin Height of caudal fin Preanal fin distance Predorsal fin distance

Factor 1

Factor 2

0.634* 0.076 0.104 0.272** 0.118 0.078 0.500 −0.123 −0.215*** −0.610*

0.188 −0.018 0.363* −0.093 0.586* 0.783* 0.579* 0.322** −0.102 0.127

–1.5 Shape of last fin ray

Variable

–1.0

–2.0

–2.5

–3.0

*P<0.001; **P<0.01; ***P<0.05.

–3.5 –3.0

4.0 3.0

–2.0 –1.5 Shape of first fin ray

–1.0

Figure 7. Differences among males of three species of mollies in dorsal fin height are plotted as the association between shape of the last dorsal fin ray and shape of the first dorsal fin ray. L=P. latipinna, O=P. orri and M=P. mexicana. Note the degree of overlap among the three species, especially for males of P. latipinna and P. orri.

2.0 1.0 Factor 2

–2.5

variables for males of each species clearly shows overlap among species for these traits (Fig. 7). This is especially true for males of P. latipinna and males of P. orri, where considerable overlap of dorsal fin height exists. Thus dorsal fin height, per se, would not be a reliable cue for females of these two species to use when distinguishing conspecific males from heterospecifics.

0.0 –0.1 –0.2 –0.3 –4.0 –10.0

DISCUSSION –5.0

0.0 Factor 1

5.0

10.0

Figure 6. Species scores on the first two canonical axes from a discriminant analysis of 10 morphological shape variables from three species of mollies. L=P. latipinna, O=P. orri and M=P. mexicana. The morphological traits with the highest correlations with each canonical axis are discussed in the text.

was most highly correlated with mid-body depth, body depth at the caudal peduncle and caudal fin length (Table 1). These traits best describe the overall body shape of males. This factor easily distinguishes the two shortfin species (Fig. 6), with males of P. mexicana having narrower bodies and shorter caudal fins. An interesting result from this analysis was that dorsal fin height, as measured by relative length of the first dorsal fin ray, did not load significantly with either canonical factor from the discriminant analysis (Table 1). Relative length of the last dorsal fin ray was not significantly correlated with canonical factor 1, the discriminant function that clearly separated the sailfin species P. latipinna from the two species of shortfins (Table 1, Fig. 6). A plot of the association between these two

Interspecific Patterns of Female Choice Females of the sailfin species P. latipinna consistently preferred conspecific males in both male treatment combinations that contained conspecifics. They had higher levels of apathy and no preference for one species over the other in the combination where both males were heterospecifics. Previous work has demonstrated that females of P. latipinna can distinguish between males from their own populations and males from non-native populations (Ptacek & Travis 1997). Although the exact set of cues that females use in making these intraspecific discriminations is unknown, males certainly differ among populations in size-associated rates of courtship displays (Ptacek & Travis 1996) and in the overall size-adjusted area of the dorsal fin (J. Travis, M. B. Ptacek & N. B. Martin, unpublished data). Sailfin males can also be distinguished from shortfin males on the basis of these same behavioural and morphological traits. This implies that the same suite of phenotypic traits that differ among intraspecific populations of males can also distinguish conspecific males from heterospecifics. It is likely that females of P. latipinna are using these same traits as cues for species recognition. A definitive test of which traits are

1152 ANIMAL BEHAVIOUR, 56, 5

essential for species recognition can only be conducted by experiments where traits are systematically varied, one at a time, holding all other traits constant, such as with video playbacks (e.g. McKinnon 1995; McClintock & Uetz 1996; Rosenthal et al. 1996). Females of both shortfin species showed a preference for the sailfin males over heterospecific shortfin males, but showed no preference between sailfin and conspecific males. This suggests that shortfin females recognize and are attracted to the sailfin phenotype, but that speciesspecific cues of conspecific males override this attraction. This also suggests that females of shortfin species may show a bias towards characteristics of the sailfin male phenotype. Preferences for heterospecific traits have been documented in other species as well (poeciliid fish genera Xiphophorus and Priapella: Basolo 1990, 1995a, b; the bush cricket, Conocephalus nigropleurum: Morris et al. 1978; the Physalaemus pustulosus species complex of frogs: Ryan & Rand 1990, 1993; and the wolf spider, Schizocosa rovneri: McClintock & Uetz 1996). In fact, in the wolf spider S. roverni, video images of heterospecific males performing leg-waving courtship displays were equally as attractive to females as video images of conspecific males (that lack a courtship display) (McClintock & Uetz 1996). This response by females of S. roverni is similar to my finding that shortfin females showed no discernible preference for conspecifics over sailfin males. Most of these studies have suggested that sensory exploitation (Ryan 1990; Ryan & Rand 1990) or preexisting biases (Basolo 1990) explain these patterns of female preference. Both mechanisms suggest that female preferences are determined by the properties of the signalling environment and the female’s sensory system and brain. Biases in the female sensory system favour the evolution of male traits that most effectively stimulate females (Endler & McLellan 1988; Endler 1992a, b; Ryan 1994). Phylogenetic studies have been used to demonstrate (Basolo 1990, 1995a, 1996; Ryan & Rand 1990; Proctor 1992; McClintock & Uetz 1996) and falsify (Hill 1994; Gray & Hagelin 1996) the pre-existence of a female bias. Without knowledge of the phylogenetic relationships of shortfin species of mollies to the sailfin species, it is impossible to determine whether the preference for sailfin males is indicative of a preference for a novel trait (having nothing to do with shared ancestry) or a preexisting bias (due to common ancestry of female sensory systems) for the sailfin phenotype. Taxonomic divisions of the group argue that the sailfin species are monophyletic and derived from shortfins (Rosen & Bailey 1963; Parenti & Rauchenberger 1989), but to distinguish between pre-existing ancestral bias and preference for novelty will require an adequate phylogeny for the group and testing shortfin species that are known sister taxa to sailfins for patterns of preference. An alternative explanation for the preference for the sailfin phenotype could simply be that P. latipinna males have longer dorsal fins resulting in larger fin area (see Results from male morphology analyses above) and present a larger physiological stimulus to shortfin females. Rowland (1989) suggested that female stickle-

backs, Gasterosteus aculeatus, preferred supernormal dummy males, 1.5 times longer than normal-sized males, because these larger dummy males formed a larger image on the female’s retina and elicited a higher level of physiological excitation. Doebler et al. (1997) found that females of P. formosa (the all-female gynogenetic species of molly), after undergoing spontaneous masculinization (‘pseudomales’), preferred females of P. latipinna over females of P. mexicana. Females of P. latipinna are larger in overall body depth and Doebler et al. (1997) suggest that these females may represent a stronger visual stimulus. Overall, levels of female preferences were weaker in shortfin species than in sailfin, P. latipinna females. Shortfin females had consistently high levels of apathy in all three male treatment combinations (Fig. 4) and females of P. orri had extremely high levels of indecision across all male combinations (Fig. 5). The existence of higher preference scores, and lower levels of apathy and indecision for females of P. latipinna is consistent with Farr’s (1989) observation that species of poeciliids in which males perform courtship behaviours also have female choice. However, females of the two shortfin species tested did show preferences for particular male phenotypes in certain male treatment combinations. This suggests that female choice may have evolved in species of poeciliids lacking courtship as well. However, whether or not these shortfin females would show intraspecific patterns of female choice remains to be tested.

Interspecific Differences in Male Signals Males of the three species differed in their degree of courtship behaviour. Males of P. latipinna performed courtship displays to conspecific females, and at lower rates, to heterospecific females during choice tests. Some males of P. mexicana also performed a type of courtship display that involved keeping the dorsal fin erect, the pelvic fins extended away from their bodies and bending their bodies towards females in a sigmoid display. Females of P. mexicana have been observed to cooperate during copulation, but previous studies were unclear whether males have a courtship display (Balsano et al. 1985). Males of P. orri showed no evidence of courtship behaviour, but did interact with females along the partition and kept their dorsal fin erect while swimming. Males of the three species also differed in morphology, but different traits contributed more to distinctions between the sailfin species and the two shortfin species than between the two shortfin species. Length and relative position of the dorsal fin contributed most to the separation of P. latipinna from either shortfin species. Overall body depth and shape distinguished the two shortfin species from one another. Dorsal fin height did not reliably separate sailfin males from shortfin males (Fig. 7), thus differences in overall dorsal fin area between males of P. latipinna and the two shortfin species are due to length differences rather than to differences in height. Overall, sailfin males differed from males of the two shortfin species in the same behavioural and morphological traits that distinguish males from different populations of P. latipinna (Ptacek & Travis 1996; J. Travis,

PTACEK: INTERSPECIFIC MATE CHOICE IN MOLLIES 1153

M. B. Ptacek & N. B. Martin, unpublished data). This implies that the same traits used by females in intraspecific comparisons (Ptacek & Travis 1997) may also be important in species recognition for females of P. latipinna. Such a pattern of female preference has been recently demonstrated in butterflies (Weirnasz 1989; Weirnasz & Kingsolver 1992). If intraspecific variation in female mating preferences can produce divergence of male phenotypes and subsequent reproductive isolation, then sexual selection resulting from female choice can lead to speciation. The pattern of female preference in P. latipinna argues for the potential importance of sexual selection through female choice during speciation of sailfin mollies. Acknowledgments Special thanks to J. Travis for helpful discussions in design and analysis of the experiments. Comments from C. Baer, A. Basolo, F. Breden, J. Richardson, I. Schlupp and J. Travis greatly improved the manuscript. Figure 1 was drawn by L. Nelson and Fig. 2 was modified by J. Lee. This research was sponsored in part by National Science Foundation grant DEB 92-20849 to J. Travis, Florida State University. The research presented here was described in Animal Research Protocol No. 9115 approved on 18 July 1994 by the Institutional Animal Care and Use Committee of Florida State University. References Balsano, J. S., Randle, E. J., Rausch, E. M. & Monaco, P. J. 1985. Reproductive behavior and maintenance of all-female Poecilia. Environmental Biology of the Fishes, 12, 251–263. Basolo, A. L. 1990. Female preference predates the evolution of the sword in swordtail fish. Science, 250, 808–810. Basolo, A. L. 1995a. A further examination of a pre-existing bias favouring a sword in the genus Xiphophorus. Animal Behaviour, 50, 365–375. Basolo, A. L. 1995b. Phylogenetic evidence for the role of a pre-existing bias in sexual selection. Proceedings of the Royal Society of London, Series B, 259, 307–311. Basolo, A. L. 1996. The phylogenetic distribution of a female preference. Systematic Biology, 45, 290–307. Brett, B. L. H. & Grosse, D. J. 1982. A reproductive pheromone in the Mexican poeciliid fish Poecilia chica. Copeia, 1982, 219–223. Cochran, W. G. & Cox, G. M. 1957. Experimental Designs. New York: J. Wiley. Darwin, C. 1871. The Descent of Man and Selection in Relation to Sex. London: J. Murray. Doebler, M., Schlupp, I. & Parzefall, J. 1997. Changes in mate choice with spontaneous masculinisation in Poecilia formosa. Ethology Supplement, 32, 204. Endler, J. A. 1992a. Introduction to the symposium. American Naturalist Supplement, 139, 1–3. Endler, J. A. 1992b. Signals, signal conditions, and the direction of evolution. American Naturalist Supplement, 139, 125–153. Endler, J. A. & Houde, A. E. 1995. Geographic variation in female preferences for male traits in Poecilia reticulata. Evolution, 49, 456–468. Endler, J. A. & McLellan, D. A. 1988. The processes of evolution: towards a newer synthesis. Annual Review of Ecology and Systematics, 19, 395–421.

Farr, J. A. 1975. The role of predation in the evolution of social behavior of natural populations of the guppy, Poecilia reticulata (Pisces: Poeciliidae). Evolution, 29, 151–158. Farr, J. A. 1989. Sexual selection and secondary sexual differentiation in poeciliids: determinants of male mating success and evolution of female choice. In: Ecology and Evolution of Livebearing Fishes (Poeciliidae) (Ed. by G. K. Meffe & F. F. Snelson, Jr), pp. 91–123. Englewood Cliffs, New Jersey: Prentice Hall. Farr, J. A. & Travis, J. 1986. Fertility advertisement by female sailfin mollies, Poecilia latipinna (Pisces: Poeciliidae). Copeia, 1986, 467– 472. Farr, J. A., Travis, J. & Trexler, J. C. 1986. Behavioural allometry and interdemic variation in sexual behaviour of the sailfin molly, Poecilia latipinna (Pisces: Poeciliidae). Animal Behaviour, 34, 597–509. Freeman, M. F. & Tukey, J. W. 1950. Transformations related to the angular and the square root. Annals of Mathematical Statistics, 21, 607–611. Gray, D. A. & Hagelin, J. C. 1996. Song repertoires and sensory exploitation: reconsidering the case of the common grackle. Animal Behaviour, 52, 795–800. Hill, G. E. 1994. Geographic variation in male ornamentation and female preference in the house finch: a comparative test of models of sexual selection. Behavioral Ecology, 5, 64–73. Hubbs, C. L. 1933. Species and hybrids of Mollienesia. Aquarium, 1, 263–268, 277. Kaneshiro, K. Y. & Boake, C. R. B. 1987. Sexual selection and speciation: issues raised by Hawaiian Drosophila. Trends in Ecology and Evolution, 2, 207–212. Lande, R. 1981. Models of speciation by sexual selection on polygenic traits. Proceedings of the National Academy of Science of the U.S.A., 78, 3721–3725. Liley, N. R. 1966. Ethological isolating mechanisms in four sympatric species of poeciliid fishes. Behaviour, Supplement, 13, 1–197. McClintock, W. J. & Uetz, G. W. 1996. Female choice and preexisting bias: visual cues during courtship in two Schizocosa wolf spiders (Araneae: Lycosidae). Animal Behaviour, 52, 167–181. McKinnon, J. S. 1995. Video mate preferences of female threespined sticklebacks from populations with divergent male coloration. Animal Behaviour, 50, 1645–1655. Miller, R. R. 1975. Five new species of Mexican poeciliid fishes of the genera Poecilia, Gambusia, and Poeciliopsis. Occasional Papers of the Museum of Zoology, University of Michigan, 672, 1–44. Morris, G. K., Kerr, G. E. & Fullard, J. H. 1978. Phonotactic preference of female meadow katydids (Orthoptera: Tettigoniidae). Canadian Journal of Zoology, 56, 1479–1487. Mosimann, J. E. & James, F. C. 1979. New statistical methods for allometry with application to Florida red-winged blackbirds. Evolution, 33, 444–459. Parenti, L. R. & Rauchenberger, M. 1989. Systematic overview of the poeciliines. In: Ecology and Evolution of Livebearing Fishes (Poeciliidae) (Ed. by G. K. Meffe & F. F. Snelson, Jr), pp. 3–32. Englewood Cliffs, New Jersey: Prentice Hall. Parzefall, J. 1969. Zur vergleichenden Ethologie verschiedener Mollienesia-Arten eino¨schliesslich einer Hohlenform von M. sphenops. Behaviour, 33, 1–37. Parzefall, J. 1979. Zur genetik und biologischen bedeutung des aggressionsueshaltens von Poecilia sphenops (Pisces, Poeciliidae). Zeitschrift fu¨r Tierpsychologie, 50, 399–422. Proctor, H. C. 1992. Sensory exploitation and the evolution of male mating behaviour: a cladistic test using water mites (Acari: Parasitengona). Animal Behaviour, 44, 745–752. Ptacek, M. B. & Travis, J. 1996. Interpopulation variation in male mating behaviours in the sailfin molly, Poecilia latipinna. Animal Behaviour, 52, 59–71. Ptacek, M. B. & Travis, J. 1997. Mate choice in the sailfin molly, Poecilia latipinna. Evolution, 51, 1217–1231.

1154 ANIMAL BEHAVIOUR, 56, 5

Regan, C. T. 1913. A revision of the cyprinodont fishes of the subfamily Poeciliinae. Proceedings of the Zoological Society of London, 11, 977–1018. Rosen, D. E. & Bailey, R. M. 1963. The poeciliid fishes (Cyprinodontiformes), their structure, zoogeography and systematics. Bulletin of the American Museum of Natural History, 126, 1–176. Rosenthal, G. G., Evans, C. S. & Miller, W. L. 1996. Female preference for dynamic traits in the green swordtail, Xiphophorus helleri. Animal Behaviour, 51, 811–820. Rowland, W. J. 1989. Mate choice and the supernormality effect in female sticklebacks (Gasterosteus aculeatus). Behavioral Ecology and Sociobiology, 24, 433–438. Ryan, M. J. 1990. Sexual selection, sensory systems, and sensory exploitation. Oxford Surveys in Evolutionary Biology, 7, 157–195. Ryan, M. J. 1994. Mechanisms underlying sexual selection. In: Behavioral Mechanisms in Evolutionary Ecology (Ed. by L. A. Real), pp. 190–215. Chicago: University of Chicago Press. Ryan, M. J. & Rand, A. S. 1990. The sensory basis of sexual selection for complex calls in the tungara frog, Physalaemus pustulosus (sexual selection for sensory exploitation). Evolution, 44, 305–314. Ryan, M. J. & Rand, A. S. 1993. Species recognition and sexual selection as a unitary problem in animal communication. Evolution, 47, 647–657. Ryan, M. J. & Wilczynski, W. 1991. Evolution of intraspecific variation in the advertisement call of a cricket frog (Acris crepitans, Hylidae). Biological Journal of the Linnaean Society, 44, 249–271. Ryan, M. J., Perrill, S. A. & Wilczynski, W. 1992. Auditory tuning and call frequency predict population-based mating preferences in the cricket frog, Acris crepitans. American Naturalist, 139, 1370–1383. Snelson, F. F., Jr, Wetherington, J. D. & Large, H. L. 1986. The relationship between interbrood interval and yolk loading in

a generalized poeciliid fish, Poecilia latipinna. Copeia, 1986, 295–304. Thornhill, R. & Alcock, J. 1983. The Evolution of Insect Mating Systems. Cambridge, Massachusetts: Harvard University Press. Travis, J. 1989. Ecological genetics of life-history traits in poeciliid fishes. In: Ecology and Evolution of Livebearing Fishes (Poeciliidae) (Ed. by G. K. Meffe & F. F. Snelson, Jr), pp. 185–200. Englewood Cliffs, New Jersey: Prentice Hall. Travis, J. 1994. Size-dependent behavioral variation and its genetic control within and among populations. In: Quantitative Genetic Approaches to Animal Behavior (Ed. by C. R. B. Boake), pp. 165–187. Chicago: University of Chicago Press. Travis, J. & Woodward, B. D. 1989. Social context and courtship flexibility in male sailfin mollies, Poecilia latipinna (Pisces: Poeciliidae). Animal Behaviour, 38, 1001–1011. Verrell, P. 1988. Stabilizing selection, sexual selection and speciation: a view of specific-mate recognition systems. Systematic Zoology, 37, 209–215. West-Eberhard, M. J. 1983. Sexual selection, social competition, and speciation. Quarterly Review of Biology, 58, 155–183. West-Eberhard, M. J. 1984. Sexual selection, competitive communication and species-specific signals in insects. In: Insect Communication (Ed. by T. Lewis), pp. 283–324. New York: Academic Press. Wiernasz, D. C. 1989. Female choice and sexual selection on male wing melanin pattern in Pieris occidentalis (Lepidoptera). Evolution, 43, 1672–1682. Wiernasz, D. C. & Kingsolver, J. G. 1992. Wing melanin pattern mediates species recognition in Pieris occidentalis. Animal Behaviour, 43, 89–94. Woodhead, A. D. & Armstrong, N. 1985. Aspects of mating behaviour of male mollies (Poecilia spp.). Journal of Fish Biology, 27, 593–601.