Female mate choice has long been recognized as a major driver for character displacement [1–3], or the evolution of novel male traits [4, 5], but can also play a vital role during speciation processes by promoting reproductive isolation through assortative mating [6, 7]. Many mating preferences are innate ; still, various extrinsic factors (ecological constraints) may affect individual mating decisions [8–10], such as altered possibilities for mate quality assessment due to increased costs of mate searching [11, 12]. Additionally, the social environment of the choosing individual is known to affect the strength [13–17] or even the direction of mating preferences [18–20].
Another decisive factor acting upon the expression of (female) mating preferences in natural systems is predation risk [21–26]. Female sand gobies (Pomatoschistus minutus), for example, normally prefer larger and more colorful males, but were found to be less choosy when exposed to a predator . Decreased choosiness due to the presence of predators was also reported for male mate choice in the sex-role reversed pipefish Syngnathus typhle . Furthermore, studies on guppies (Poecilia reticulata) revealed that females when facing a predator switch towards associating with less colorful males [29, 30], and females of the green swordtail (Xiphophorus hellerii) that usually prefer males with long swords, switch their preference towards males with short swords when exposed to videos showing successful predator attacks .
Such behavioral alterations can be interpreted as a tactic employed by the choosing individuals to reduce their own exposure to predators, as brightly colored males attract predators to the area, and by having more predators in the area females' predation risk is increased [32, 33]. For example, brightly colored males in the Trinidadian guppy are more vulnerable to predation by the predatory cichlids Aequidens pulcher and Crenicichla alta than drabber ones [34–40] and females that preferentially associate with such brightly colored males will obviously face an equally high predation risk.
Our present study was designed to investigate predator-induced changes of female mating preferences in the Atlantic molly (Poecilia mexicana, Poeciliidae). Poecilia mexicana males show a pronounced polymorphism in body size and coloration [41–43], and females prefer larger, more colorful, dominant males as mating partners [42, 44]. At the same time, large molly males are more conspicuous to predators, as exemplified by studies of avian predation on the related sailfin molly, P. latipinna , or predation by giant water bugs on P. mexicana [46, 47]. Hence, female mollies might increase their own risk of being attacked by a predator when associating with larger males.
The present study involved piscine predators of P. mexicana, and used cichlid species that regularly co-occur with Atlantic mollies in the same habitats [48–50]. Even though prey choice experiments on the cichlid species considered here (especially 'Cichlasoma' salvini) still need to be conducted, it seems straightforward to assume a preference for larger prey, as another cichlid, the aforementioned pike cichlid (Crenicichla alta), also prefers larger guppies as prey [51, 52]. Adult C. salvini and female P. mexicana show roughly comparable relationships in body size as do C. alta and female P. reticulata (ratio SL P. reticulata, 15-28 mm  v. maximum SL C. alta, about 160 mm [53, 54]: 0.10-0.18; ratio SL P. mexicana, 35-40 mm [, this study] v. maximum SL C. salvini, about 220 mm : 0.16-0.18), so it seems reasonable to argue that C. salvini would not be gape-limited when preying on average-sized P. mexicana females. Based on these considerations, we predicted that P. mexicana females should alter their mate choice behavior when facing predation risk by piscivorous cichlids; specifically, females should associate with small rather than large males when a piscine predator is around to minimize their own risk of being attacked.
Simultaneously, we asked whether females are able to recognize piscivorous predators and distinguish them from similar non-piscivorous species on the basis of visual cues. We hypothesized that only piscivorous predators would lead to a reversal of female preferences. To test our predictions, we conducted dichotomous female mate choice tests (association preference tests) and repeated the tests while either a natural molly predator ('Cichlasoma' salvini, Cichlidae) or a non-piscivorous fish (three types: two cichlids and another poeciliid, the green swordtail, Xiphophorus hellerii) were presented. This design allowed us to compare changes in the expression of female mating preferences from the 1st to the 2nd part of the tests among four different contexts (i.e. predator treatments). At least for the treatment involving green swordtails-a mainly detritivorous fish  that often occurs in the same microhabitats as P. mexicana in southern México and is of similar body size [48, 49]-we predicted that P. mexicana females should not alter their preferences. This treatment, therefore, served as a control to test whether females would be consistent in their mate choice behavior, or if any changes occurred over the course of the experiment that would not be attributable to the presence of a predator.
Finally, we asked the interrelated question of whether visual predator recognition (and the correlated specific responses of females to piscivorous as opposed to non-piscivorous fishes) is innate, or whether also experiential effects/learned predator avoidance could play a role. Comparing the responses of lab-reared (predator-naïve) and wild-caught (predator-experienced) females allowed us to disentangle innate and learned components of predator recognition [57, 58]. We hypothesized that (a) if visual predator recognition mechanisms are entirely innate, then both lab-reared and wild-caught females should respond to the presence of a molly predator during their mate choice, while (b) if experiential effects are important, then a response might not be observable in one of the two different female groups (predator-naïve or wild caught).
In summary, our study aims to answer the following questions: (1) Do P. mexicana females change their mating preferences when exposed to a visually presented piscine predator? (2) Are molly females able to distinguish between predatory species and similar-shaped non-predatory ones on the basis of visual cues alone? (3) Does predator experience affect females' responses to a predator during mate choice?