Male genitalia present an extraordinary pattern of rapid divergence in animals with internal fertilization, which is usually attributed to sexual selection. However, the effect of ecological factors on genitalia divergence could also be important, especially so in animals with non-retractable genitalia because of their stronger interaction with the surrounding environment in comparison to animals with retractable genitalia. Here we examine the potential of a pervasive ecological factor (predation) to influence the length and allometry of the male genitalia in guppies. We sampled guppies from pairs of low-predation (LP) and high-predation (HP) populations in seven rivers in Trinidad, and measured their body and gonopodium length. A key finding was that that HP adult males do not have consistently longer gonopodia than do LP adult males, as had been described in previous work. However, we did find such divergence for juvenile males: HP juveniles have longer gonopodia than do LP juveniles. We therefore suggest that an evolutionary trend toward the development of longer gonopodia in HP males (as seen in the juveniles) is erased after maturity owing to the higher mortality of mature males with longer gonopodia. Beyond these generalities, gonopodium length and gonopodium allometry were remarkably variable among populations even within a predation regime, thus indicating strong context dependence to their development/evolution. Our findings highlight the complex dynamics of genitalia evolution in Trinidadian guppies.

Sampling and fish care

We sampled juvenile and adult male guppies in one low and one high predation locality in each of seven different streams in the Northern Mountain Range, Trinidad. These localities were classified as low and high predation localities based on the absence or presence of piscivorous fish, respectively (Endler, 1978; Gotanda & Hendry, 2014; Kelly, Godin, & Abdallah, 2000; Reznick & Endler, 1982). All fish were transported to our laboratory at the William Beebe Tropical Research Station in Trinidad, acclimatized for 30 min, transferred to 20 L aquariums, and immediately treated for bacterial, fungal, and parasitic infections with Polyguard^TM (Seachem Laboratories, Inc.). Fish were fed live brine shrimp or flake food if they remained more than 24 hours in the laboratory, but most fish were released back to their original site the day after processing (details of processing below). All fish were kept at 20 – 24°C and on a natural 12:12 (light:dark) photoperiod. All fish handling was in accordance with McGill Animal Use Protocol No. 4570.

Measurements and maturity status

Body length (from snout to caudal peduncle) and gonopodium length (from base of gonopodium to distal tip, excluding the hood – see Kelly, Godin, & Abdallah, 2000) were obtained from digital photographs using the software ImageJ (Abràmoff, Magalhães, & Ram, 2004). We first anesthetized the fish with an aqueous solution of tricaine methanesulfonate (MS-222) and NaHCO₃ and then placed them on their right side on a white background containing a ruler. We then photographed the left side of each fish with a Nikon D300 digital camera equipped with a 60mm macro lens, with illumination provided by two full spectrum fluorescent lights and a Nikon speedlight commander Kit R1C1 flash.

            The development stage of males was determined based on the stage of development of the hood, a sensory protuberance in the gonopodium (Houde, 1997), which was visualized under a Leica ES2 stereomicroscope before the photographs were taken. Males were categorized as mature when the hood extended beyond the distal tip of the gonopodium (the hook), and immature when the hood was shorter than the gonopodium (Houde, 1997). Furthermore, we visually classified the development of the gonopodium into three different stages (Fig. 1): (1) early-stage, when the differentiation of the anal fin into the gonopodium is ongoing and it bears a wide base, forming a triangular shape – not shown in Figure 1; (2) Advanced stage, including the sub-stages “Hood not developed”, when the gonopodium has developed a thinner base – what remains henceforth – and has acquired an appearance of a fully developed gonopodium, similar to the “Hood developing” sub-stage and the “Final stage” of development, as described below; and “Hood developing”, representing the phase in which the hood has just started to develop until the phase in which it has acquired a filament-like shape but is still shorter than the gonopodium; and, finally (3) Final stage, when the hood is fully developed and longer than the hook – typical from adult males. Although we initially distinguished the sub-stages “Hood not developed” from “Hood developing”, we subsequently grouped them together as “Advanced stage” because there was no apparent difference in the allometric growth between these stages. We removed early-stage juveniles from the statistical analysis due to small sample sizes; therefore, we refer to advanced stage juveniles simply as juveniles henceforth.