Atlas Journal of Biology 3 (1): 212–217, 2014 doi: 10.5147/ajb.2014.0135 A tla s Jo ur na l o f Bi ol og y - IS SN 2 15 8- 91 51 . P ub lis he d By A tla s Pu bl ish in g, L P (w w w .a tla s- pu bl ish in g. or g) Frequency-Dependent Mate Selection in the Guppy (Poeciliidae: Poecilia reticulata) Andre N. Porter and Jack S. Frankel* Department of Biology, Howard University, Washington, DC 20059, USA. Received: July 16, 2014, 2014 / Accepted: August 16, 2014 __________________________________________________ * Corresponding author: jfrankel@Howard.edu 212 Abstract Heterogeneity within a population enhances its long-term survival. A fundamental method of maintaining population heterogeneity is the retention of rare or uncommon pheno- types by selective mating strategies. Employing two color morphs, red tuxedo and red, of the guppy (Poecilia reticulata, Poeciliidae), this study was designed to investigate whether P. reticulata females would preferentially seek out heteroge- neous groupings of males exhibiting two color morphs. Adult female guppies were exposed simultaneously to two groups of males (n=10); one comprised of only the tuxedo color morph (n=5) and the other of both color morphs (n=5). For the latter group, the ratio of males exhibiting the two color- ations was changed incrementally over multiple trials, begin- ning and ending with homogeneous groupings (5:0, 4:1, 3:2, 2:3, 1:4, 0:5). Experiments were conducted employing a 76 L aquarium partitioned at both ends to house the male groups. Individual females were placed into an acclimation, holding column located in the center compartment of the experimen- tal aquarium. Subsequent to a five minute acclimation period in the holding column, each female was released. Courting behaviors and population affinities (i.e. female location rela- tive to the male compartments) were recorded every five sec- onds over a five minute observation period. Females did not show a preference for male groupings exhibiting either the 4:1 or 1:4 color ratios. However, females clearly showed af- finity for the male groupings of 3:2 and 2:3 (p <0.05). This investigation on female mate choice in P. reticulata clearly re- veals that a mating strategy is occurring in this species and that it is frequency-dependent. Key Words: Poecilia reticulata, guppy, mate choice, frequency- dependent selection, color morphs. Introduction Genetic variation and phenotypic diversity are central con- cepts in evolutionary biology and closely linked to the complex- ity of organisms, ecosystem recovery, and the ability of species to respond to environmental changes (Bazin et al., 2006). Mate choice is known to be a powerful evolutionary mechanism influ- encing the stability and maintenance of polymorphisms within a species (Pryke and Griffith, 2007). In general, females of a species are more selective than males in discriminating among prospective mates, as they evaluate genetic quality and re- source-holding potential based on secondary sexual character- istics, such as body size, coloration, ornamentation, and aggres- siveness (Basolo, 1990; Clotfelter et al., 2006; Bierbach et al., 2013). Elucidating mate choice strategies can assist in ascertaining and accessing possible cause and effect relationships in mating preference. Within non-random mating systems, selective mech- anisms must be in place to maintain population heterogeneity. The retention of rare or uncommon phenotypes is a fundamental method of maintaining such heterogeneity. Thus, in species where mating is non-random, females are often found to be particu- larly selective when the parental male investment is low (Baldauf et al., 2009). Here, females would be expected to select and court with males that exhibit traits present in low phenotypic frequency (i.e. uncommon phenotypes), thereby ensuring that those phenotypes remain within the population. Utilizing such a frequency-dependent selection mating strategy, populations would ultimately exhibit an increase in the frequency of those rarer phenotypes, thereby insuring their genetic diversity (Gross, 1991; Pilastro et al., 1997; Punzalan et al., 2005). Coloration has been shown to affect both species recognition and mate choice in poeciliid fishes (Endler, 1983). In fact, the maintenance and benefits of color variation in numerous spe- cies of poeciliids by selective mating strategies has been inves- tigated in both natural and artificial populations (Borowsky and This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecom- mons.org/licenses/by/3.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the origi- nal work is properly cited. A tla s Jo ur na l o f Bi ol og y - IS SN 2 15 8- 91 51 . P ub lis he d By A tla s Pu bl ish in g, L P (w w w .a tla s- pu bl ish in g. or g) 213 Kallman, 1976; Houde and Endler, 1990; Baer et al., 1994; Royle et al., 2008; Culumber and Rosenthal et al., 2013). Ad- ditional studies have illustrated the importance of coloration in mate selection by poeciliid females, where they exhibit a natural affinity towards certain colorations when shown a population of males containing phenotypically variable color morphs (Houde, 1997; Kingston et al., 2003). The study presented here was designed to investigate the va- lidity of frequency-dependent selection based on a body color polymorphism in the freshwater guppy Poecilia reticulata (Poe- ciliidae). The Poeciliidae include over 40 species from North, South and Central America, which occupy a wide range of habi- tats within these regions and exhibit both morphological and behavioral differentiation within and between species (Breden et al., 1999). As a group, they are relatively small fishes, rang- ing from 1.5 – 6.0 cm in length, with highly developed sexual dimorphisms. The guppy, P. reticulata, one of the more common poeciliids, is distributed world-wide. This species ranges in size from 1.5 to 3.5 cm for adult males and from 3.0 to 6.0 cm for adult females. While native to Antigua, Barbados, Jamaica, Trinidad, Tobago, Brazil, Venezuela, and the U.S. Virgin Islands, populations of P. reticulata have been introduced to many other habitats and have readily occupied bodies of freshwater made accessible to them. The guppy is extremely popular in the aquarium trade, due to ease of upkeep and the colorfully ornamented caudal and dorsal fin morphs found in males through artificial selection. This species is not only sexual dimorphic with regard to both col- oration and color intensity, but exhibits a wide array of shape and fin morphology polymorphisms as well. As with other poeciliids, P. reticulata serves as an excellent model organism to employ in a study on frequency-dependent mate selection, since it exhibits both distinct and uniform color morphs. Utilizing red and red tuxedo color morphs, this study investigated whether P. reticulata females would preferentially seek out heterogeneous groupings of males as opposed to ho- mogeneous ones. It is hypothesized that phenotypic heterogene- ity amongst potential male mates is more attractive to females than a homogeneous grouping, as determined by observing the courting behaviors of females when confronted with these di- verse male groupings. Materials and Methods To investigate female preference for male groups based on the phenotypic frequencies of two color morphs, a series of ex- perimental trials were conducted employing similar constructs as those described for examining female color preference in the pygmy swordtail, Xiphophorus pygmaeus (Baer et al., 1994). In the work presented here, two phenotypes were chosen; a red tuxedo color morph exhibiting a red trunk with a dark blue cau- dal peduncle, red dorsal and caudal fins with dark blue margins, and a red color morph which displays a red trunk with red dor- sal and caudal fins. These phenotypes were selected for their uniformity in overall coloration and color pattern from individual to individual within these two color morphs. Adult guppies were obtained from G & G Aquatics Inc., Lor- ton, VA., USA. Male and female guppies were housed in 76 L capacity, high configuration (61cm x 32cm x 42cm) holding aquaria during the course of this investigation. Females were chosen at random from an all-female stock population (n = 120) maintained in a dedicated aquarium. Males of each color morph were selected from initial stock populations (n = 80) for uniformity in total body length, dorsal and caudal fin lengths, and coloration. Males selected from the stock populations were maintained with individuals of like phenotypic coloration, i.e. red males with red males, tuxedo males with tuxedo males. Water temperature was maintained at 23 ± 2 °C by the laboratory HVAC system. Fish were fed TetraMin flakes twice daily. Water quality (ammonia, nitrite and nitrate) was assessed weekly. The pH level in all aquaria was maintained between 6.8 and 7.2. Experimental trials were conducted in a 76 L capacity, long configuration (74cm x 20cm x 30cm) aquarium (Fig. 1). Groups consisting of five males (all-tuxedo, all-red, or tuxedo/red com- binations) were held on either side of the experimental aquari- um in 11cm x 20cm x 30cm male compartments, designated as Sections A and B. Male compartments were constructed with solid Plexiglas sheets secured to the aquarium with a continu- ous strip of aquarium silicon to physically separate males from females during the trials. Lighting consisted of fluorescent lights directly above the experimental aquarium. Additionally, three sides of the experimental aquarium were covered with white paper, leaving only the front and top uncovered to allow for ob- servation of female location and courting behavior while elimi- nating distractions from the sides and back. To begin each trial, two groups of five males were collected and placed in each of the male compartments. Initially, one sec- tion was selected for placement of all-tuxedo males and the oth- er for all-tuxedo males, combinations of tuxedo and red males, or all-red males. Male groups were then subsequently designat- ed as A or B; correlating with the corresponding partitioned side as seen in Figure 1. One female was then randomly chosen and placed in a Plexiglas acclimation column (11cm x 11cm x 50cm) located in the center of the experimental aquarium, designated as “no preference” Section D. Once a female was placed in the acclimation column, she was given a five minute acclimation period. When the acclimation period expired, the column was removed, releasing a female within section D and allowing her five minutes to move freely within that section and into areas A and B which are juxtaposed to the male compartments. The relative position of each female was designated by her location in A, B, or D and was recorded at five second intervals through- out the duration of each five minute trial period. Time intervals spent in section D was interpreted as “no preference,” while time intervals spent in preference areas A or B as “preferentially courting” males in sections A or B. After each trial, females were captured and placed in a separate holding aquarium to prevent reuse for a given trial set. To avoid side bias, all trial sets were repeated with male groupings switched from section A to B. Eleven trials were conducted per paired male groupings. Adult female guppies were exposed simultaneously to two groups of males (n = 10); one comprised of only the tuxedo color morph (n = 5) and the other of the two color morphs (n = 5). For the latter group, the ratio of males exhibiting the two A tla s Jo ur na l o f Bi ol og y - IS SN 2 15 8- 91 51 . P ub lis he d By A tla s Pu bl ish in g, L P (w w w .a tla s- pu bl ish in g. or g) A tla s Jo ur na l o f Bi ol og y - IS SN 2 15 8- 91 51 . P ub lis he d By A tla s Pu bl ish in g, L P (w w w .a tla s- pu bl ish in g. or g) 214 colorations was changed incrementally over multiple trials, be- ginning and ending with homogeneous groupings (5:0, 4:1, 3:2, 2:3, 1:4, 0:5). For this study, two complete trial sets were undertaken. For each individual trial (e.g. 1, 2a, 2b, 3a, 3b, etc.), 11 obser- vations were recorded for 11 different females. Designations “a” and “b” indicate replicate trial observations. A total of 60 observations were recorded (one every five seconds for five minutes) per trial female, giving a total of 660 observations for each group of 11 females/trial. At the completion of all trial sets, a total of 12,540 data points were recorded. The data was then subjected to chi-square analysis for the probability of females deviating from the expected time interval allocation of 1:1 for movement into preference areas A or B due to chance alone. All trials were conducted under an approved Howard University IACUC protocol (IACUC-GSAS-11-03). Results and Discussion Table 1 presents the total five-second time allotments for 11 female guppies/trial occupying preference areas A and B when exposed to (i) all-tuxedo groupings in both male compart- ments (Trial 1), (ii) all-tuxedo groupings opposed to tuxedo/ red combinations (Trials 2a – 9b), and (iii) all-tuxedo group- ings opposed to all-red groupings (Trials 10a and 10b). When confronted with paired homogeneous tuxedo groupings (Trial 1), females clearly showed no preference towards males in either compartment (p = 0.491). Further, females did not favor either all-tuxedo or all-red male groupings (Trials 10a and 10b), nor did they show an affinity towards 4 tuxedo: 1 red or 1 tuxedo: 4 red combinations (Trials 2a, 2b, 3a, 3b, 8a, 8b, 9a and 9b). In fact, females preferentially exhibited courting behaviors di- rected toward the homogeneous, all-tuxedo groupings when the 1 tuxedo: 4 red combination occupied the opposing compart- ment (Trials 8b and 9a). Females clearly exhibited a particular preference for male grouping of both 3 tuxedo: 2 red and 2 tuxedo: 3 red (p < 0.001; Trials 4b, 5b, 6a, 6b, 7a and 7b). Total five-second time allotments for female guppies in pref- erence areas A and B, when confronted with groupings of five tuxedo males and combinations of tuxedo and red males (Trials 3 – 8), are presented in Table 2. With the exception of trial sets 8b and 9a (1 tuxedo: 4 red occupying A and 5 tuxedo occupy- ing B; 5 tuxedo occupying A and 1 tuxedo: 4 red occupying B, respectively), where females exhibited an affinity towards the Front view of the experimental aquarium Aerial view of the experimental aquarium Figure 1. Front and top aerial views of the experimental aquarium.Overall dimensions were 74 cm long x 20 cm wide x 30 cm tall. Male compartments are labeled Sections A and B. The female preference areas, A and B, are juxtaposed to the male compartments. The female no preference area, labeled Section D, is in the center of the aquar ium and contains the holding column. A tla s Jo ur na l o f Bi ol og y - IS SN 2 15 8- 91 51 . P ub lis he d By A tla s Pu bl ish in g, L P (w w w .a tla s- pu bl ish in g. or g) 215 Table 2. Total time allocations per trial (in 5 second intervals) by females (n = 11) courting heterogeneous male groupings of red tuxedo and red color morphs and all-red tuxedo groupings, expected ratios, degrees of freedom (df), chi-square values, and probabilities. Trial Time Allocations Courting Red Tuxedo/ Red Time Allocations Courting All-Red Tuxedo Expected Ratio df Chi-square Value Probability 3a. 4:1/5 184 222 1:1 1 3.557 0.059 3b. 4:1/5 170 190 1:1 1 1.111 0.292 5a. 3:2/5 204 224 1:1 1 0.935 0.334 5b. 3:2/5* 270 110 1:1 1 67.368 <0.001 7a. 2:3/5* 299 199 1:1 1 20.080 <0.001 7b. 2:3/5* 305 220 1:1 1 13.762 <0.001 9a. 1:4/5* 153 215 1:1 1 10.446 0.001 9b. 1:4/5 160 195 1:1 1 3.451 0.063 2a. 4:1/5 237 227 1:1 1 0.216 0.643 2b. 4:1/5 205 185 1:1 1 1.026 0.311 4a. 3:2/5* 224 258 1:1 1 2.938 0.123 4b. 3:2/5* 225 145 1:1 1 17.297 <0.001 6a. 2:3/5* 262 88 1:1 1 86.503 <0.001 6b. 2:3/5* 200 95 1:1 1 37.373 <0.001 8a. 1:4/5 193 191 1:1 1 0.010 0.919 8b. 1:4/5* 205 255 1:1 1 5.435 0.002 Total 16 270.508 <0.001 Pooled 3496 3019 1:1 1 34.924 <0.001 Heterogeneity 15 235.584 <0.001 *Trials with significant chi-square values (p < 0.05). Numbers in red and blue denote the number of red and red tuxedo morphs used within the compartment. Compartment Trial A/B Time Allocated in A Time Allocated in B Expected Ratio df Chi-square Value Probability 1 5/5 230 245 1:1 1 0.474 0.491 2a. 5/4:1 227 237 1:1 1 0.216 0.643 2b. 5/4:1 185 205 1:1 1 1.026 0.311 3a. 4:1/5 184 222 1:1 1 3.557 0.059 3b. 4:1/5 170 190 1:1 1 1.111 0.292 4a. 5/3:2 258 224 1:1 1 2.938 0.122 4b. 5/3:2* 145 225 1:1 1 17.297 <0.001 5a. 3:2/5 204 224 1:1 1 0.935 0.334 5b. 3:2/5* 270 110 1:1 1 67.368 <0.001 6a. 5/2:3* 88 262 1:1 1 86.503 <0.001 6b. 5/2:3* 95 200 1:1 1 37.373 <0.001 7a. 2:3/5* 299 199 1:1 1 20.080 <0.001 7b. 2:3/5* 305 220 1:1 1 13.762 <0.001 8a. 5/1:4 191 193 1:1 1 0.010 0.919 8b. 5/1:4 255 205 1:1 1 5.435 0.020 9a. 1:4/5 153 215 1:1 1 10.446 0.001 9b. 1:4/5 160 195 1:1 1 3.451 0.063 10a. 5/5 246 253 1:1 1 0.098 0.754 10b. 5/5 210 205 1:1 1 0.060 0.806 *Females preferentially courting male groups of 3 red: 2 red tuxedo or 2 red: 3 red tuxedo over all-red tuxedo groupings. Numbers in red and blue denote the number of red and red tuxedo morphs used within the compartment. Table 1. Total time allocations per trial (in 5 second intervals) by females (n = 11) courting male groupings of red tuxedo and/or red color morphs, expected ratios, degrees of freedom (df), chi-square values, and probabilities. A tla s Jo ur na l o f Bi ol og y - IS SN 2 15 8- 91 51 . P ub lis he d By A tla s Pu bl ish in g, L P (w w w .a tla s- pu bl ish in g. or g) A tla s Jo ur na l o f Bi ol og y - IS SN 2 15 8- 91 51 . P ub lis he d By A tla s Pu bl ish in g, L P (w w w .a tla s- pu bl ish in g. or g) all-tuxedo grouping, they did not show a preference for either all-tuxedo or 1:4 or 4:1 groupings. Further, while females were clearly attracted towards those male groupings with combina- tions of 3:2 and 2:3, they overwhelmingly selected heteroge- neous groupings in general, regardless of compartment location (p < 0.001). Table 3 illustrates the comparison of total five sec- ond allotments spent within preference areas A or B as com- pared to those observed for females in the no-preference area D. Here, clear indications of choice were apparent, with females spending significantly more time outside of the no preference section D in 16 of 19 trial sets. There are many potential explanations for the coexistence of color polymorphisms within populations. In practice, however, it has proven difficult to determine which mechanisms explain the presence of multiple color variants within any particular popula- tion (Munday et al., 2003). Female mate choice has been pro- posed as a means of increasing diversity where a strong prefer- ence for rare or novel color patterns is indicated. However, this “rare male” advantage will be detectable only when phenotype frequencies are perturbed from their equilibrium values (Hughes et al., 2013). The results obtained from this investigation of mate choice in P. reticulata provide strong evidence towards female mating preference for groupings of males exhibiting multiple pheno- types. Data compiled from this study clearly indicates that when female guppies are presented with a choice between males that exhibit a single phenotype or multiple phenotypes, they exhibit some preference towards the latter. Further, in 16 of the 19 com- plete trials, female time allocations towards preference groups (i.e. making a courting decision) exceeded the time spent for making no courting choice (i.e. remaining in section D) and is consistent with findings in other mate selection studies (Kodric- Brown, 1985; Baer et al., 1994; MacLaren and Fontaine, 2012). As trials extended beyond the initial all-tuxedo groupings, fe- males showed no preference for male phenotypic combinations of 4:1 (Trials 2a, 2b, 3a, 3b) or 1:4 (Trials 8a, 8b, 9a, 9b). In fact, females exhibited an affinity towards the all-tuxedo male grouping when confronted with that homogeneous grouping and the 1 tuxedo: 4 red combination (Trials 9a and 8b; p < 0.05) and occurred whether the all-tuxedo grouping occupied section A or B. In effect, for trials where male groupings had a phe- notypic frequency of 0.20 for the “rare” color morph, females showed no statistically significant preference for those groups. As the phenotypic frequency of the initially rare color morph increased to 0.40, the affinity of female guppies for those male groupings became evident. Of the eight trial sets comprised of color ratios of 2:3 or 3:2, six resulted in statistical significance of female mate choice for these combinations. This clear affin- ity of females towards male groupings exhibiting 40 % rarity for either tuxedo or red color morphs may, in part, be due to a phenotypic threshold for frequency-dependent selection for this trait. Roff (1998) has provided evidence that certain levels of phenotypic variation may be maintained in a population at some “threshold level” and suggests that frequency-dependent selec- tion can give rise to, and maintain, polymorphisms in a popula- tion. This may speak to an equilibrium state where male groups 216 Compartment Trial A/B Time Allocated To A or B Time Allocated To D Expected Ratio df Chi-square Value Probability 1. 5/5 475 185 1:1 1 127.424 <0.001 2a. 5/4:1 464 196 1:1 1 108.824 <0.001 2b. 5/4:1 390 275 1:1 1 21.818 <0.001 3a. 4:1/5 406 254 1:1 1 35.006 <0.001 3b. 4:1/5 360 300 1:1 1 5.455 0.020 4a. 5/3:2 482 178 1:1 1 140.024 <0.001 4b. 5/3:2 370 290 1:1 1 9.697 0.002 5a. 3:2/5 428 232 1:1 1 58.206 <0.001 5b. 3:2/5 380 280 1:1 1 15.152 <0.001 6a 5/2:3 350 310 1:1 1 2.424 0.120 6b. 5/2:3 295 365 1:1 1 7.424 0.006 7a. 2:3/5 498 162 1:1 1 171.055 <0.001 7b. 2:3/5 525 135 1:1 1 230.455 <0.001 8a. 5/1:4 384 276 1:1 1 17.673 <0.001 8b. 5/1:4 460 200 1:1 1 102.424 <0.001 9a. 1:4/5 368 292 1:1 1 8.752 0.003 9b. 1:4/5 355 305 1:1 1 3.788 0.052 10a. 5/5 499 161 1:1 1 173.097 <0.001 10b. 5/5 415 245 1:1 1 43.788 <0.001 Numbers in red and blue denote the number of red and red tuxedo morphs used within the compartment. Table 3. Time allocations per trial (in 5 second intervals) by females (n = 11) courting males in prefer- ence areas A or B, or making no decision and remaining in no preference section D, expected ratios, degrees of freedom (df), chi-square values, and probabilities. 217 A tla s Jo ur na l o f Bi ol og y - IS SN 2 15 8- 91 51 . P ub lis he d By A tla s Pu bl ish in g, L P (w w w .a tla s- pu bl ish in g. or g) of ~20% for a novel phenotype represent a threshold level with the common phenotype (Hughes et al., 2013). Therefore, evidence exists that traits being maintained through frequency- dependent mating strategies may be not only driven by these disruptive mating strategies, but also maintained by them. In the research presented here, the affinity of females towards 2:3 and 3:2 heterogeneous male groupings when the color morphs reach frequencies between 0.40 and 0.60 may flag a threshold for female mate choice for these particular color morphs in P. reticulata. Acknowledgements The authors wish to extend their appreciation to George Mid- dendorf and Hemayet Ullah, Department of Biology, Howard University, Washington, DC, and to Tyler Frankel, Department of Animal and Avian Sciences, University of Maryland, College Park, MD, for their suggestions during the course of this investi- gation. References Baer FC, M Danitzker, and MJ Ryan (1994) A test for preference of association in a color polymorphic poeciliid fish: laboratory study. Environmental Biology of Fishes 43: 207-212. Baldauf SA, H Kullmann, SH Schrotth, T Thunken, and TC Bakker (2009) You can’t always get what you want: size assortative mating by mu- tual mate choice as a resolution of sexual conflict. BMC Evolutionary Biology 9: 129-138. Basolo A (1990) Female preference for male sword length in the green swordtail, Xiphophorus helleri (Pisces: Poeciliidae). Animal Behavior 40: 332-338. Bazin E, G Sylvain, and N Galtier (2006) Population size does not influence mitochondrial genetic diversity in animals. Science 312: 570-572. Bierbach D, V Sassmannshausen, B Streit, L Arias-Rodriguez, and M Plath (2013) Females prefer males with superior fighting abilities but avoid sexually harassing winners when eavesdropping on male fights. Behavioral Ecology Sociobiology 67: 675-683. Borowsky R, and K Kallman (1976) Patterns of mating in natural popu- lations of Xiphophorus (Pisces: Poeciliidae). I: X. maculatus from Be- lize and Mexico. Evolution 30: 693-706. Breden F, MB Ptacek, M Rashed, D Taphorn, and CA Figueiredo (1999). Molecular phylogeny of the live-bearing fish genus Poecilia (Cyprin- odontiformes: Poeciliidae). Molecular Phylogenetics and Evolution 12 (2): 95-104. Clotfelter ED, LK Curren, and CE Murphy (2006) Mate choice and spawning success in the fighting fish Betta splendens: the importance of body size, display behavior and nest size. Ethology 112: 1170- 1178. Culumber ZW, and GG Rosenthal (2013) Mating preference does not maintain tail spot polymorphism in the platyfish Xiphophorus varia- tus. Behavioral Ecology, 24 (6): 1287-1291. Endler JA (1983) Natural and sexual selection on color patterns in poe- ciliid fishes. Environmental Biology of Fishes 9: 173-190. Gross M (1991) Evolution of alternative reproductive strategies: fre- quency-dependent sexual selection in male bluegill sunfish. Philo- sophical Transactions of the Royal Society Biological Sciences 332: 59-66. Houde AE (1997) Sex, color and mate choice in guppies. Princeton University Press, 224 pp. Houde AE, and JA Endler (1990) Correlated evolution of female mat- ing preferences and male color patterns in the guppy Poecilia reticu- lata. Science 248: 1405-1408. Hughes KA, AE Houde, AC Price, and FH Rodd (2013) Mating advan- tage for rare males in wild guppy populations. Nature 503: 108- 109. Kingston J, G Rosenthal, and MJ Ryan (2003) The role of sexual selec- tion in maintaining a colour polymorphism in the pygmy swordtail, Xiphophorus pygmaeus. Animal Behavior 65: 735-743. Kodric-Brown A (1985) Female preference and sexual selection for male coloration in the guppy (Poecilia reticulata). Behavioral Ecol- ogy and Sociobiology 17: 199-205. MacLaren RD, and A Fontaine (2012) Female bias for male lateral projection area in Poecilia reticulata. Environmental Biology of Fishes 93 (1): 105-119. Munday PL, PJ Eyre, and GP Jones (2003) Ecological mechanisms for coexistence of colour polymorphism in coral-reef fish: an experimen- tal evaluation. Oecologia 137: 529-526. Pilastro A, E Giacomello, and A Bisazza (1997) Sexual selection for small size in male mosquito fish (Gambusia holbrooki). Proceedings of the Royal Society Biological Sciences 264: 1125-1129. Punzalan D, FH Rodd, and KA Hughes (2005) Perceptual processes and the maintenance of polymorphism through frequency-dependent predation. Evolutionary Ecology 19: 303-320. Pryke SR, and SC Griffith (2007) The relative role of male vs. female mate choice in maintaining assortative pairing among discrete co- lour morphs. Journal of Evolutionary Biology 20 (4): 1512-1521. Roff DA (1998) The maintenance of phenotypic and genetic variation in threshold traits by frequency-dependent selection. Journal of Evolu- tionary Biology 11: 513-529. Royle NJ, J Lindström, and NB Metcalfe (2008) Context-dependent mate choice in relation to social composition in green swordtails, Xiphophorus helleri. Behavioral Ecology 19 (5): 998-1005.