Animal Behaviour 84 (2012) 1051e1059
Contents lists available at SciVerse ScienceDirect
Animal Behaviour journal homepage: www.elsevier.com/locate/anbehav
Natural versus sexual selection: predation risk in relation to body size and sexual ornaments in the green swordtail Armando Hernandez-Jimenez 1, Oscar Rios-Cardenas* Departamento de Biología Evolutiva, Instituto de Ecología, A.C. Xalapa, Mexico
a r t i c l e i n f o Article history: Received 29 November 2011 Initial acceptance 10 February 2012 Final acceptance 17 July 2012 Available online 24 August 2012 MS. number: A11-00954R Keywords: natural selection predation sexual selection swordtail trade-off Xiphophorus hellerii
In general, we assume that natural (predation-mediated) and sexual selection have opposing effects on the evolution of characters that serve as ornaments. Males of most swordtail fishes (genus Xiphophorus) have an elongation of the caudal fin known as the sword that is used to attract females, as it increases apparent body size of males. By increasing apparent body size, the sword may also attract the attention of predators or decrease the likelihood that a predator will attack. Using the green swordtail, Xiphophorus hellerii, we evaluated separately the effect of body size and the presence of the sword on the likelihood of being attacked by a predator. We conducted preference tests using a sympatric cichlid, Thorichthys ellioti, as a predator. For the effect of body size we used live pairs of male swordtails of different sizes without swords (surgically removed) as stimuli, and for the effect of the sword we used videos of male swordtails with and without their sword (digitally removed). We found no effect of body size, but the cichlid predators directed more bites towards individuals with swords. Despite their benefits in terms of attracting mates, the sword seems to increase the risk of being attacked by a predator and thus represents a trade-off between natural selection (mediated by predation) and sexual selection. Ó 2012 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Predation is a major selective force in the evolution of morphological and behavioural adaptations in animals (Sih 1985). The evolution of traits related to avoiding and reducing predation risk is favoured because individuals that exhibit these traits can invest more time and energy on other critical activities or traits (Lima 1998). Endler (1995) found that these morphological and behavioural adaptations have the effect of decreasing conspicuousness or attractiveness to predators of individuals that carry them. Therefore, many organisms experience a conflict between different ecological demands, as they must optimize foraging and reproduction, while avoiding being attacked by a predator (Johnson & Agrawal 2003). The effects of predation depend, among other things, on how the size of prey affects their risk of being caught and eaten (Juanes 1994; Juanes & Conover 1994). If the risk is greater for small individuals, selection should favour the allocation of resources to somatic growth, resulting in delayed maturation and reproduction at larger sizes; we would expect the opposite if the risk is greater for large individuals (early maturation and reproduction at smaller
* Correspondence: O. Rios-Cardenas, Departamento de Biología Evolutiva, Km 2.5 Carretera Antigua a Coatepec 351, Congregación El Haya, Instituto de Ecología A.C. Xalapa, Veracruz 91070, Mexico. E-mail address:
[email protected] (O. Rios-Cardenas). 1 E-mail address:
[email protected] (A. Hernandez-Jimenez).
sizes; Charlesworth 1980). Several studies have shown that predation can be directed to both smaller sizes (e.g. Werner et al. 1983) and larger sizes (e.g. Lafferty 1993). In predatoreprey systems of cichlids and guppies (Crenicichla saxatilisePoecilia reticulata), predation favours small prey (Reznick et al. 1990; Winemiller et al. 1990), while in cichlideplaty and cichlide swordtail systems, predation may favour larger sizes (Basolo & Wagner 2004). In those cases where natural selection favours larger sizes (less predation), there may be body structures that increase the apparent size of the fish without requiring investment of greater resources in body growth (Fuiman & Magurran 1994). Another mechanism that may favour the development of exaggerated structures is sexual selection, which promotes the evolution of structures and elaborate displays that usually occur in males (Andersson 1994). Elaborate structures of males can be used during competition for access to females, thus increasing the chances of success in contests and thus mating (intrasexual selection). In mating systems with elaborate courtship displays, elaborate structures can evolve as a result of selection by females since they may reflect the benefits that males can offer to females and their progeny (intersexual selection). These ornaments may be maintained in populations because they increase the reproductive success of males that exhibit them, and because females that choose matings based on these attributes can derive direct benefits (e.g. better territories and/or resources in the territory, care and
0003-3472/$38.00 Ó 2012 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.anbehav.2012.08.004
1052
A. Hernandez-Jimenez, O. Rios-Cardenas / Animal Behaviour 84 (2012) 1051e1059
protection to the female and her offspring, higher fertility, lower risk of predation) or indirect genetic benefits (e.g. hatchlings inherit the father’s appeal, and/or regardless of sex, the offspring could inherit the father characteristics associated with greater viability; Andersson 1994). In general, we expect the costs of producing a large trait to increase proportionally to the benefits obtained by the increment of the trait (Kotiaho et al. 1998). The costs resulting from such increment in size of a sexual trait may be direct (e.g. predation) or physiological (e.g. greater energy expenditure). Because both viability and mating success affect the fitness of a male, a trait may be expected to have a detrimental effect by decreasing viability through natural selection and to have a beneficial effect through sexual selection by increasing mating success (Kotiaho et al. 1998). Therefore, we expect the expression of a sexual trait to reflect a balance between natural and sexual selection (e.g. Reznick & Endler 1982; Andersson 1994); however, the evidence for natural selection balancing this trade-off is often not conclusive. In summary, natural and sexual selection often act in opposite directions, because the former tends to favour organisms that are less conspicuous to predators (Reznick & Endler 1982; Zuk & Kolluru 1998), while the latter promotes the evolution of conspicuous traits that tend to maximize matings (Houde 1987; Houde & Torio 1992). The goal of this study was to assess whether these selective forces are opposing or complementary in the green swordtail fish, Xiphophorus hellerii (e.g. Jennions et al. 2001). Males of most species in the genus Xiphophorus have an elongation of the lower rays of their caudal fin, forming a structure called a sword (Rauchenberger et al. 1990). Previous studies have shown that females prefer males with a sword (i.e. intersexual selection; Basolo 1990a) because it increases apparent body size (Rosenthal & Evans 1998), and larger males may be better at defending a territory (Magnhagen & Kvarnemo 1989), may suffer less predation (Scharf et al. 2000), and if it is a heritable condition, their offspring may develop larger bodies (Reynolds & Gross 1992). When food is abundant, male green swordtails invest in both body and sword growth; however, with a reduction in food availability, males shift their growth strategy and invest more heavily in sword growth (Basolo 1998). This shift in strategy may maintain a constant growth in lateral projection area (LAP; i.e. body and fin area seen from the side; MacLaren & Daniska 2008; MacLaren & Fontaine 2012) during low resource conditions. In the green swordtail, the average size of males and females is greater in populations with piscivorous fishes than in populations without these predators (Basolo & Wagner 2004). Based on this correlation, Basolo & Wagner (2004) suggested that because larger individuals might be more difficult to capture and manipulate, a large body size may avoid predation in this species. We suggest that by increasing apparent body size (by the increase in LPA due to the enlargement of the caudal fin; MacLaren & Daniska 2008; MacLaren & Fontaine 2012), the sword may also be reducing predation, a hypothesis that has not been tested. Using the green swordtail, we sought to determine the effect of body size and the presence of the sword on the likelihood of being attacked by a predator. A previous study examined the effects of UV coloration expressed on the sword of Xiphophorus nigrensis males on predator preferences (Cummings et al. 2003). However, to our knowledge this is the first experimental study looking at the effect on predation of body size and presence of the sword (two sexually selected traits) in X. hellerii. Given that body size in X. hellerii could affect the probability of being attacked by a predator, because larger individuals may be more difficult to capture and handle, we suggest that smaller individuals may be more likely to be attacked by predators. The male sword is a trait that is also likely to be related to the
probability of being attacked by a predator; if the sword affects body size perception (makes swordtails appear larger), then swordless males may be more likely to be attacked by a predator. Alternatively, if the sword attracts predators, then swordless males may be less likely to be attacked by a predator. METHODS During the rainy season, we collected X. hellerii individuals from the Pixquiac River (19 290 57.4600 N, 96 560 47.0800 W, N ¼ 28) near the town of Coatepec in the state of Veracruz, Mexico, and from a creek (19190 47.2900 N, 96 430 25.5700 W, N ¼ 42) that flowed into La Antigua river near the town of Apazapan, Veracruz, Mexico. We selected these sites based on the abundance of the species and the proximity to our laboratory. We captured animals using funnel traps and electrofishing (BADGER-1 Backpack Electrofishing Unit, Version Monocanal, 0e600 V). We only collected mature males and females. Mature males are distinguished by the presence of a sword and a well-developed gonopodium (modified anal fin used to transfer sperm). Mature females are distinguished by the presence of a brood spot (pigmentation of tissue surrounding the female reproductive organs). We transported collected fishes to the laboratory where we placed them in 54-litre tanks (60 30 30 cm) for storage and maintenance. We filled tanks with tap water treated with a chlorine-removal solution (API Stress Coat, Mars Fishcare, Inc., Chalfont, PA, U.S.A.). Fish appeared to acclimate well to this type of water, as we did not notice any unusual behaviour during the duration of the study. We fed the fish three times a day ad libitum with commercial flakes (Basic Flakes, Wardley Essentials, Hartz Mountain Corp., Secaucus, NJ, U.S.A.). Collection sites were 30 km apart; however, there was no significant difference in standard length (SL), the distance from the anterior tip of the mouth to the end of the caudal peduncle (t68 ¼ 0.922, P ¼ 0.36; mean SE: Pixquiac: 45.37 1.20 mm; Apazapan: 46.74 0.90 mm), or sword length (adjusted to SL, see Basolo & Wagner 2004 for calculations; t64 ¼ 1.93, P ¼ 0.06; mean SE: Pixquiac: 21.84 0.98 mm; Apazapan: 26.77 1.33 mm) among individuals from these localities. To determine whether the sample size used in these analyses could have influenced the probability of detecting an effect of the independent variable (if it existed), we determined the effect size and its respective confidence intervals, CI (Cumming & Finch 2001). Our analyses revealed that the effect size of body size (r ¼ 0.111, CI ¼ 0.127, 0.337) and sword length (r ¼ 0.234, CI ¼ 0.0079, 0.4509) did not differ significantly from zero, suggesting that there was no real effect given by the source population. Therefore, in subsequent analyses, we combined males from both localities without distinction. We used Thorichthys ellioti to act as the predator, because it is an omnivorous cichlid whose diet includes fish (Valtierra-Vega & Schmitter-Soto 2000) and it is sympatric with X. hellerii. In addition, among all sympatric cichlid species, T. ellioti showed a clear response to X. hellerii individuals in a preliminary study. Furthermore, in the laboratory, T. ellioti eagerly consumed small (juvenile) X. hellerii individuals. Finally, an analysis of gut content of T. ellioti (N ¼ 28) indicated that fishes made up about 38% of identifiable gut material (A. Hernandez-Jimenez, unpublished data). We used electrofishing to collect the predators (23 individuals) from the Apazapan population (see location above), where they coexist with X. hellerii. We placed predators in individual tanks (see size above) where we fed them three times daily ad libitum with commercial pellets (Cichlid Floating Pellets, Wardley Essentials) and sporadically with zebrafish, Danio rerio guppies, Poecilia reticulata, X. hellerii juveniles and other unidentified small poeciliids so they would continue to eat live food. We used a calliper to measure the
A. Hernandez-Jimenez, O. Rios-Cardenas / Animal Behaviour 84 (2012) 1051e1059
Experiment 1: Effect of Body Size Body size and stimulus fish behaviour We conducted observations of 12 stimulus fish pairs (two X. hellerii, one small and one large) under the same conditions present during the cichlid preference tests (a free-swimming predator in the central compartment, and one stimulus fish on each side of the lateral compartments of the observation tank; see below and Fig. 1). We performed these observations to assess the possibility of an effect of the difference in body size between the two stimulus fish on their behaviour. For these observations we divided the lateral compartments in two equal-sized virtual compartments (front and back) using a vertical line drawn on the glass (Fig. 1). After a 5 min acclimation period we quantified for 10 min (1) the number of times that the line was crossed (back and front movements) and (2) the frequency with which the stimulus moved up or down (fast movements with a vertical orientation from the water surface to the bottom). We randomly selected which compartment was to be observed directly. The behaviour of the fish in the other supplementary compartment was simultaneously recorded (using a Nikon Coolpix P90 camera). Later, we observed the recordings to collect the data from the stimulus fish that was not observed directly. The data met the assumptions of normality required for parametric statistics. To analyse the difference in behaviour between large and small stimuli, we performed a t test using fish size as the independent variable, and lateral and up-anddown movements as dependent variables. In addition, we determined the effect size with its respective confidence intervals (Cumming & Finch 2001). Body size and probability of being attacked by a predator We conducted an experiment in a tank divided into three compartments using one-way mirrors to separate the stimulus fish from predators (Fig. 1). Because stimulus fish could not see the
Mirror
Light
Live stimuli
total time in each preference zone and number of bites directed to a stimulus (when the cichlid hit the glass with its mouth). We specifically tested the effects of body size on the behaviour of the stimulus fish and on the probability of being attacked by a predator (experiment 1), as well as the effect of the sword on the probability of being attacked by a predator (experiment 2).
One–way mirror
Light
One–way mirror
Live stimuli
50 cm
opening and depth of the cichlid’s mouth, and estimated mouth volume based on the formula of a cone (1/3 p radius2 height), where the aperture represents the diameter of the base and the depth represents the height. We used the volume of the cichlid’s mouth as a covariate in the statistical analysis. The observation area was free of any disturbances to avoid stressing the fishes under observation. We always performed the experiments between 1000 and 1700 hours Greenwich Mean Time. The observation tank was illuminated with two 80 W white light bulbs that did not emit UV radiation (the absence of this radiation excluded signals in these wavelengths, which could affect the behaviour of the fishes in the observation tank; Kodric-Brown & Johnson 2002). There is little evidence of UV vision in African cichlids (of 33 species found in Lake Tanganyika, Lake Victoria and Lake Malawi, only 12 species can detect UV; Carleton 2009). To our knowledge, there is no study assessing UV vision in American cichlids. Even if T. ellioti has UV vision, we assumed UV did not affect its prey preference in our study because (1) we used light bulbs that lacked UV emission, (2) we placed a one-way mirror between T. ellioti and the swordtail stimulus fish (experiment 1; see below), so any UV radiation perceived by T. ellioti would be minimal, and (3) although the sword of the green swordtail reflects in the UV (Cummings et al. 2003), we either used stimulus fish with their swords removed (experiment 1), or used video recordings of stimulus fish (experiment 2), and to our knowledge, the monitors used to present these video recordings do not emit in the UV range. We starved cichlids for 48 h prior to the experiments. To begin the experiments we placed the cichlid in a central compartment (neutral zone) of the experimental tank where we conducted the tests (Fig. 1). We then placed the stimuli on each side of the experimental tank. We delimited the tank neutral zone with two clear and removable Plexiglas sheets to keep predator in the neutral zone during the 15 min of acclimatization so the predator could watch the areas of the tank where the stimuli (potential prey) were presented. Stimulus conditions are specified in each experiment (see below). After the acclimatization period, we removed the Plexiglas sheets to free the cichlid from the central compartment and began to observe its attack preferences. We measured the predator preferences based on its position (right or left preference zone) in the tank (Fig. 1). For 15 min we recorded latency (time it took for the cichlid to enter a preference zone for the first time), the
1053
50 cm Preference zone Preference zone Neutral zone Removable clear acrylic Central compartment 100 cm Figure 1. Observational tank (100 50 50 cm) divided into three compartments: two side compartments, where live stimuli were placed, and a central compartment divided in a neutral (central) zone and two preference zones. We placed one X. hellerii male in each side compartment (live stimuli) and the cichlid predator in the central compartment. Oneway mirrors divided the compartments. All aquarium walls were covered with white cardboard to avoid disturbing the fish, and fish were observed through the top mirror.
A. Hernandez-Jimenez, O. Rios-Cardenas / Animal Behaviour 84 (2012) 1051e1059
predator, they showed no signs of predator avoidance behaviour. We used this experiment to estimate the effect of body size of X. hellerii males on the probability of being attacked by a predator. The setting allowed the predator to see the stimulus fish, but prevented the stimulus fish from seeing the predator, which might have prompted an elusive behavioural response. We used 23 predators and 23 unique X. hellerii male pairs as stimuli (one pair for each predator). We surgically removed the swords of the two stimulus fish; thus, the effect of the removal was the same for both stimulus fish. Removal involved cutting fin rays only; thus, this procedure did not significantly affect fish behaviour and they eventually redeveloped their sword (Basolo 1990a, b). Using a calliper, we measured SL, height (from the anterior insertion of the dorsal fin to insertion of the pelvic fins) and width (around the gill slits) of swordtails to determine its volume based on the formula of an ovoid (p SL height width)/6. When forming the stimulus pairs (one large and one small fish), we considered all available fish to maximize the size difference between the two fish and thus promote a clear response from the predator (mean difference in volume ¼ 1403.9 mm3, difference range ¼ 641.6e3103.2 mm3; mean difference in SL ¼ 9.2 mm, difference range 1.8e22.5 mm). To determine whether the difference in size between the stimuli affected the predator’s response, we conducted a linear regression with the difference in volume between stimuli as the independent variable, and the strength of preference for large fish (cichlid response to the large fish minus the response to the small fish for each dependent variable; i.e. latency, total time and bites) as the dependent variable. In all cases we exchanged the stimulus from one compartment to another (left to right and vice versa) and repeated the observation to avoid potential side preferences by the predator independent of its preference for the stimulus fish. For data analysis we used the total number of observations in both tests (two for each pair of stimulus fish). Experiment 2: Effect of the Sword To measure the effect of the presence of the sword on the probability of being attacked by a predator, regardless of body size, we used video recordings as stimuli (to avoid possible effects on swordtail fish behaviour of manipulating the swords). To choose the fish that would serve as models in the videos (hereafter ‘models’), we performed a linear regression of SL and sword length (distance between the distal tip of the sword and its insertion on the caudal peduncle) of all available individuals (N ¼ 68); we then identified all individuals that had a completely developed sword (positive residuals; see Fig. 2) in proportion to their body size. Mean SL of fish was 48.1 mm (N ¼ 68). We chose two individuals with positive residuals and an SL closer to the mean. We chose a third individual with an SL well above the mean, because among the available fish, there were no more individuals that were close to this SL value (Fig. 2). To create the video stimuli, we carried out video recordings in a 40 12 10 cm tank divided in three compartments provided with a white background. The central compartment (17 12 10 cm) had the same length and height as the monitor used to present the videos to cichlids; thus, the size of the fish in the video did not differ from that of the live fish. During the recordings we placed a female in each side compartment to ensure that the male remained active and continued to display to the females. We illuminated this tank using white light (see above). We used a photographic camera Nikon model P90 to record 20 min of the male’s behaviour. We then selected a sequence of about 15 s that had the best display of the sword. We produced a copy of the sequence and broke it down into 24 frames/s using the software
60
50 Sword length (mm)
1054
40
30
20
10 40
50
60
70
Standard length (mm) Figure 2. Relation between body size (standard length, SL) and sword length for all captured X. hellerii males (linear regression: R2 ¼ 0.377, F1, 67 ¼ 39.938, P < 0.05). Open circles: individual X. hellerii; solid circles: model fish used as video stimuli in experiment 2; vertical line: average SL of all fish; curves: 95% CIs. Data points above the regression line represent individuals with swords that were longer than expected based on their SLs.
Sony Vegas Pro 9. Using Corel Photopaint X5, we cloned the colours of the background around the sword of each photograph and replaced the original coloration on the sword with the background coloration. We then used the edited pictures to restore the original sequence. In this manner we created a video of a swordless fish that was otherwise identical to the original (with the presence of the sword). We repeated and joined each of these 15 s sequences to form a 17 min video stimulus and added a 3 min opening sequence of the tank without the fish (only the background) to each video stimulus. Because one of the individuals used as a model to create the video stimulus was larger than the other two, we presented the three sets of video stimuli for both conditions (with and without a sword) to each predator (N ¼ 20). We used the identity of the video stimulus as a within-individual factor in the statistical analysis to determine whether there was an effect of a particular stimulus. We randomized the video stimulus presentation to avoid any order effect. Observers were blind with respect to the type of video stimulus (with or without sword) being presented. We conducted this experiment in a 56 43 30 cm tank. The tank was similar to the one used in the previous experiment (Fig. 1) in that it was divided into three equal size zones (neutral zone and lateral preference zones). However, the lateral compartments did not exist as they were replaced by two highdefinition monitors Sony Trinitron (model PVM-8040) to reproduce the video stimulus. At the beginning of each test, we placed the predator in the neutral zone delimited by removable Plexiglas sheets. We allowed the fish 10 min to acclimate to the test tank with the monitors off, followed by 3 min with the monitor on playing the sequence without fish (background) and 2 min with the stimulus sequence. After the acclimation period (15 min total), we removed the partitions to allow the cichlid to leave the neutral zone and recorded the same behaviours as in the previous experiment for 15 min. To avoid side preferences, at the end of each test we exchanged the video stimulus from one side to another and repeated the observation.
A. Hernandez-Jimenez, O. Rios-Cardenas / Animal Behaviour 84 (2012) 1051e1059
Statistical Analysis We analysed the data with the statistical software IBM SPSS v.20 using a mixed model. For both experiments (1) individuals were considered as random factors, (2) in the case of the dependent variables ‘latency’ and ‘total time’, we used a linear model, (3) in the case of the dependent variable ‘bites’, we used a generalized model with a negative binomial distribution, because the number of bites had a Poisson distribution and data were overdispersed, (4) we performed a model reduction using Akaike’s Information Criterion corrected for small sample size (AICc; Hurvich & Tsai 1991), and we report the two models with the lowest AICc values (that best fitted the data; Table 1). For experiment 1 (body size effect), we used the volume of the cichlid’s mouth and the difference in body volume between the stimulus fish as covariates, and for experiment 2, we used only the volume of the cichlid’s mouth as a covariate. For experiment 2, in addition to the factor representing the presence of the sword, we included the ‘model’ as a within-subjects factor (to account for the identity of the X. hellerii individual used to produce the video stimulus). RESULTS Experiment 1: Effect of Body Size Body size and stimulus fish behaviour Body size and behaviour of stimulus fish were not significantly correlated (t test: up-and-down movements: t11 ¼ 0.092, P ¼ 0.929; lateral movements: t11 ¼ 0.068, P ¼ 0.947). Effect sizes did not differ significantly from zero (up-and-down movements: r ¼ 0.020, CI ¼ 0.387, 0.420; lateral movements: r ¼ 0.015, CI ¼ 0.392, 0.416).
1055
the predator spent in each preference zone was not significantly affected by the volume of its mouth. However, the volume of the predator’s mouth had a significant effect on latency (Table 2), with latency increasing with the volume of the predator’s mouth (Fig. 3). To explore the possible effect of the sample size used in the previous analyses, we determined confidence intervals for the effect sizes of the three response variables. Confidence intervals for effect sizes of the three response variables did not differ significantly from zero (latency: r ¼ 0.058, CI ¼ 0.236, 0.342; total time: r ¼ 0.123, CI ¼ 0.175, 0.397; bites: r ¼ 0.191, CI ¼ 0.108, 0.453). Although this result suggests there was no significant effect of body size of stimulus fish on the probability of being attacked, increasing the sample size may reduce the size of the confidence intervals (Cumming & Finch 2001). Experiment 2: Effect of the Sword The model (the particular fish used create the video stimulus) had no significant effect on latency, total time (Table 2) or number of bites (excluded from the minimal model) directed to the stimulus. The volume of the predator’s mouth did not significantly affect its response based on latency or total time (Table 2). However, the volume of the predator’s mouth had a significant effect on the number of bites (Table 3), with number of bites increasing with the volume of the predator’s mouth (Fig. 4). The presence of the sword in the video stimulus had no significant effect on the predator’s latency to enter the stimulus zone or the total time it spent in the stimulus zone (Table 2). However, the sword had a significant effect on the number of bites that the predator directed to the video stimulus (Table 3); specifically, predators directed more bites towards fish with a sword (Fig. 5). DISCUSSION
Body size and probability of being attacked by a predator The linear regression of the difference in volume between stimulus pairs and the strength of preference for large males was not significant (latency: R ¼ 0.006, F1, 21 ¼ 0.001, P ¼ 0.979; total time: R ¼ 0.233, F1, 21 ¼ 1.274, P ¼ 0.285; bites: R ¼ 0.038, F1, 21 ¼ 0.030, P ¼ 0.864), suggesting that the difference in volume between stimulus pairs had no effect on the cichlids’ responses. Body size of stimulus fish had no significant effect on any of the predator’s response variables (Tables 2, 3). Similarly, the total time
Table 1 Models with the lowest AICc values and their associated scores for green swordtails’ probability of being attacked by cichlid predators in experiments 1 and 2 Response variable Model Experiment 1 Latency Total time Bites Experiment 2 Latency
Total time
Bites
Latency*mouth volumeþbody sizeþindividual Latency*mouth volumeþbody size Total time*mouth volumeþbody sizeþindividual Total time*mouth volumeþbody size Bites*mouth volumeþbody sizeþindividual Bites*mouth volumeþbody size Latency*mouth volumeþsword þ modelþ individual Latency*mouth volumeþswordþmodel Total time*mouth volumeþswordþmodelþ individual Total time*mouth volumeþswordþmodel Bitesþswordþmodelþindividual Bitesþswordþindividual
Previous studies have suggested that a predator’s mouth size can produce differential predation based on the size of the prey (e.g. Werner et al. 1983). In the swordtail X. hellerii, where large individuals are more common in populations with piscivorous fish predators, Basolo & Wagner (2004) suggested that predation may be less common for larger individuals. The results from our first experiment revealed no significant effect of body size of the potential prey on any of the predator response variables (latency, total time associating with the prey and number of bites directed at the prey). Thus, body size in X. hellerii does not seem to affect the probability of being attacked by a predator. Our results suggest that predation pressure may not explain the pattern between size and predation that Basolo & Wagner (2004)
AICc 576.58 574.38 639.18 636.98 165.75 170.31
Table 2 Results from the analysis of cichlids’ latency to enter and total time spent in zones associated with green swordtail stimulus fish in experiments 1 and 2 Response variable Experiment 1 Latency
1585.08 Total time 1583.02 1787.68 1785.60 407.58 403.35
AICc: Akaike’s Information Criterion corrected for small sample size; Latency: latency to enter stimulus zone; Total time: total time spent in stimulus zone; Bites: number of bites directed towards stimulus. In all cases, the same significance value was associated with the tested fixed factors in both models.
Experiment 2 Latency
Total time
Factor(s)/covariate
F
Body size Mouth volume Body size Mouth volume
0.153 4.789 0.68 0.273
Sword Model Mouth volume Sword Model Mouth volume
1.498 0.004 0.701 2.059 0.027 0.025
df 1, 1, 1, 1,
P 43 43 43 43
0.698 0.034 0.414 0.604
1, 97 2, 97 1, 18 1, 115 2, 115 1, 115
0.224 0.996 0.413 0.154 0.973 0.874
The factor(s) and covariate included in the minimal model are shown. Significant values are shown in bold.
1056
A. Hernandez-Jimenez, O. Rios-Cardenas / Animal Behaviour 84 (2012) 1051e1059 Table 3 Results from the analysis of number of bites directed by cichlid predators towards green swordtail stimulus fish in experiments 1 and 2
600
Factor/covariable(s)
Latency (s)
500
Experiment 1 Body size Experiment 2 Mouth volume Sword
400 300
F
df
P
1.51
1, 43
0.225
5.96 5.11
1, 43 1, 118
0.019 0.026
The factor and covariate included in the minimal model are shown. Significant values are shown in bold.
200 100
0
5
10
15 20 25 Mouth volume (mm3)
30
35
40
Figure 3. Linear regression of cichlid mouth volume and latency to associate with small (solid triangles, solid lines) and large (open circles, dashed lines) X. hellerii stimulus fish.
detected across populations of X. hellerii. Male and female X. hellerii that co-occur with piscivorous fishes are larger than those that do not, leading Basolo & Wagner (2004) to suggest that increased body size may result from selective pressure exerted by predation. The piscivorous fish predators detected in Basolo & Wagner’s (2004) study included large cichlid species (Cichlasoma friedrichsthalii, Cichlasoma octofasciatum, Petenia splendida) and the poeciliid Belonesox belizanus. In our study, the populations from which we obtained the study specimens (Pixquiac and Apazapan) had cichlid species that were smaller, on average, than those in Basolo & Wagner’s (2004) study, and thus could be considered to have lower predation levels (sensu Basolo & Wagner 2004). The cichlid species cohabiting with X. hellerii at our study site in Apazapan (T. ellioti) had a mean SL of 87.3 mm (N ¼ 40; maximum SL ¼ 108 mm; maximum SL reported for this species ¼ 115 mm; Miller et al. 2005) and those reported in Basolo & Wagner’s (2004) study have maximum SLs of 280 mm (C. friedrichsthalii), 217 mm
(C. octofasciatum), 350 mm (P. splendida) and 200 mm (B. belizanus) (Miller et al. 2005). Therefore, in comparison with the populations previously studied, the X. hellerii in Apazapan might be experiencing low levels of predation. However, the average SL of X. hellerii in our study populations (46.19 mm, N ¼ 68) was even higher than the average for individuals in populations with high predation from Basolo & Wagner’s (2004) study (average of reported means, 43.3 mm). One possible explanation for the discrepancy (large body sizes and apparently low predation of X. hellerii in our study) is that predation by T. ellioti is highly concentrated on small individuals. In this case, small size class adults would be scarce, because most individuals that survive would be above the threshold size preyed upon by T. ellioti. Alternatively, the existence of large individuals in Pixquiac or Apazapan populations may be due to the presence of predators other than fish (e.g. birds, snakes; Moyaho et al. 2004) that may also be exerting strong selection against small body size. Another possible explanation for this pattern is that environmental factors independent of predation risk may be affecting body size in X. hellerii. One of the environmental factors that could favour large sizes is high food availability (Basolo & Wagner 2004). Our analyses of effect size suggest that the lack of response to the size differences between potential prey was not due to small sample size or to low statistical power. We believe that individuals of T. ellioti in our study did not discriminate between large and small X. hellerii individuals because even those used as small stimuli (mean SL ¼ 41.5 mm, N ¼ 23) were large enough to protect
7
6
30
Number of bites/15 min
Number of bites
10
5
5
4
3
2
1 0
5
10
15 20 25 30 Mouth volume (mm3)
35
40
Figure 4. Linear regression of cichlid mouth volume and number of bites directed towards small (solid triangles, solid lines) and large (open circles, dashed lines) X. hellerii stimulus fish.
0
Swordless
Sworded
Figure 5. Number of bites by cichlids towards X. hellerii stimulus fish. Mean (diamond), median (line within box) and lower and upper quartiles (box) are shown.
A. Hernandez-Jimenez, O. Rios-Cardenas / Animal Behaviour 84 (2012) 1051e1059
themselves from possible attacks by predators. In fact, the individuals collected in the Pixquiac river (where T. ellioti is not found, and therefore where predation levels may be even lower than in the Apazapan population) maintain, on average, a large body size (mean SL ¼ 45.37 mm, N ¼ 28). Therefore, regardless of the factors promoting large green swordtails in Apazapan and Pixquiac, this large size could confer males protection from predators (Webb 1981; Fuiman & Magurran 1994; Christensen 1996), and once their large size allows them to escape from predation, other selective factors may be maintaining a large body size (e.g. sexual selection). To test more specifically whether the lack of discrimination between large and small sizes of green swordtails is due to low levels of predation by T. ellioti on large individuals of the Apazapan and Pixquiac populations, a further study could sample these populations exhaustively to capture smaller adult individuals and/or juveniles to be included in a similar experiment. It may then be possible to detect a preference for green swordtail individuals that are small enough to be eaten by T. ellioti. An alternative future experiment could use predator species that are large enough to consume the size range of green swordtails used as small stimuli in our study. Finally, we found that the volume of the predator’s mouth seemed to affect both the predator’s latency to move outside the neutral zone and the number of bites it directed to potential prey, and thus, influenced its decision to attack. An increase in the volume of the mouth appeared to increase the time that the predator spent evaluating the two potential prey and, once the decision was made to attack, it also seemed to increase the rate at which potential prey were attacked. Basolo & Wagner (2004) found that green swordtail males from populations where piscivorous predators were present had relatively shorter swords than males from populations where piscivorous fishes were absent. Based on this correlation, they suggested that the presence of predators causes individuals to develop smaller swords. With respect to the effect of the sword on the probability of being attacked by a predator, our results are consistent with the suggestion of these authors: sworded individuals were more conspicuous and thus were more likely to be attacked by predators. Compared to the populations studied by Basolo & Wagner (2004), predation pressures appear to be lower in the Pixquiac and Apazapan populations (see above). However, the length of the sword, adjusted for SL, in our study population (mean ¼ 25.50 mm) was similar to that of populations with predators from Basolo & Wagner’s (2004) study (average of reported means for populations with predation ¼ 24.37 mm; without predation ¼ 31.4 mm). One explanation for this pattern (of short swords) in our study population is that predation pressure exerted by T. ellioti is not only focused on small individuals (see above), but can also include biting off swords of larger individuals. The tetra fish Astyanax mexicanus is a potential predator of swordtails that takes small bites out of them. When swordtails were experimentally exposed to these tetras, approximately 50% of the males lost their sword (Rosenthal 2000), and 63e86% of male swordtails in natural populations show signs of sword damage (Rosenthal et al. 2002). Furthermore, male swordtails develop longer swords under isolated laboratory conditions than they do in the wild (O. Rios-Cardenas, personal observations), and tetras tend to bite the swords of male swordtail dummies in natural populations (O. Rios-Cardenas, A. Hernandez-Jimenez & L. Bono, unpublished data). Although the swordtail fish are not consumed, nipping of their swords may promote the development of short swords, which in turn may affect male fitness in terms of reduced mating opportunities (Johnson & Basolo 2005) and survival since injured swords may be more susceptible to infections.
1057
An alternative explanation is that other nonfish predators exert selective pressure against the development of longer swords in Apazapan and Pixquiac. To our knowledge, there are no reports of swordtail predation by snakes or birds, but there are reports of predation on other similarly sized viviparous fish by snakes (Moyaho et al. 2004). This alternative hypothesis may be likely given that the Pixquiac population has no T. ellioti predators and yet green swordtail individuals have developed short swords. Video stimuli of green swordtail individuals with an intact sword were attacked significantly more than those without a sword. The sword is a trait that makes some types of swimming activity energetically more expensive (Basolo & Alcaraz 2003), may decrease a male’s ability to escape a predator once detected, and/or by being very colourful and conspicuous, may make males more visible to predators (Godin & McDonough 2003). Even if natural selection acts against costly traits of this type, sexual selection may favour them strongly. Zahavi & Zahavi (1997) argued that such characters impose a cost for the individual (e.g. increased predation risk or increased demand for resources to development them) and are only carried by high-quality males, which have survived despite having the handicap that these characters represent. By mating with males possessing the exaggerated character, females are choosing a high-quality male (determined by its ability to survive). We suggest that this is the case in the green swordtail, where a male sword is a sexually selected trait (Basolo 1990a) that appears to be costly (Rosenthal et al. 2001; Basolo & Alcaraz 2003; Basolo & Wagner 2004) because it increases the risk of being attacked by a predator (this study) and is therefore a possible handicap (Zahavi 1977; Grafen 1990; Johnstone 1995). Besides increasing the attractiveness of males to females, the sword seems to be a reliable signal of male quality. To determine with certainty whether bright coloration (e.g. Godin & McDonough 2003) and sword length in the green swordtail have evolved as expensive traits that reflect male genetic quality, it is particularly important to demonstrate experimentally whether the cost in fitness (e.g. increased mortality risk by predation) associated with this sexual trait is lower for males in good condition than it is for males in poor condition (Zeh & Zeh 1988; Grafen 1990; Johnstone 1995). Such studies have been done only in a few fish (Moodie 1972) and bird (Møller & Nielsen 1997) species with sexual traits other than coloration and size (review in Kotiaho 2001). A recent study found that escape behaviour performance of green swordtails is highly variable, but rather than being correlated with sword length, it may be affected by variation in body shape or muscle physiology (Baumgartner et al. 2011). Therefore, according to Baumgartner et al. (2011), the sword does not seem to represent a handicap for males. In contrast, our results support the idea that the sword, which apparently arose from sexual selection by female preference (Basolo 1990a), represents a handicap by making the bearer more attractive to predators. Rosenthal et al. (2001) also suggested that sworded males were more conspicuous to predators, showing that X. multilineatus predators (A. mexicanus) oriented their bodies more often towards sworded males than towards swordless males. These findings support the results of our study for green swordtails, where predators preferentially attacked individuals with a sword. Visual preference for sworded males by T. ellioti in our study was also similar to that shown by green swordtail females (both were attracted to intact males). While predator preferences do not necessarily predict the likelihood of attack, they do indicate the degree to which predators detect and interact with different phenotypes, a necessary first step in predatoreprey encounters (Endler 1991). Because females prefer males with larger swords in the green swordtail (Basolo 1990a, 1998; Trainor & Basolo 2000), and predators share this preference, our results are consistent with the
1058
A. Hernandez-Jimenez, O. Rios-Cardenas / Animal Behaviour 84 (2012) 1051e1059
hypothesis that natural selection mediated by predation has an effect opposite to that of sexual selection. As suggested by Basolo & Wagner (2004), since the green swordtail co-occurs with piscivorous fishes in some natural populations while in others such predators are absent, the variation in sword length observed among populations and even among species of swordtail fishes may be related to different predation pressures to which these populations/species have been exposed. Depending on the degree of predation in different populations, and the balance between natural and sexual selection in these populations, individuals may develop large swords (when predation pressure is low) or small swords (when predation pressure is high). This hypothesis requires that males with short swords are less susceptible to predators while those with large swords are more susceptible (or are harassed more often). This assumption cannot be tested with our data since our study only involved the presence or absence of the sword; however, future studies could use natural variation in sword length between individuals of either the same species or different species (using comparative methods) to test this assumption. Overall, our results show that sexual and natural selection are not acting synergistically, but antagonistically, as a male trait that is used to obtain matings (the sword) also attracts predators. Sword length for body size varies across species within the genus Xiphophorus, and sword length is associated with alternative mating tactics in some species (Morris et al. 2008). Since the levels of predation among species and populations may explain variation in sword length, we suggest that future studies should also explore the possibility that levels of predation might also explain the maintenance of different morphs (with and without swords) in Xiphophorus species with alternative mating tactics. Acknowledgments We thank Antonio Acini, Jaime Camacho and Christian Rodriguez for assistance in the field and for help in maintaining the experimental fish, Denisse Maldonado for assistance in videoediting and the Mexican government for the collection permit (DGOPA.10854.061210.5440). Funding was provided by grants from the Consejo Nacional de Ciencia y Tecnología (Fellowship 234520 to A.H.-J.) and Instituto de Ecología A. C. to O.R.-C. We also thank Roberto Munguía-Steyer for assistance in statistical analyses, and Rogelio Macías-Ordoñez, Luis Mendoza-Cuenca, Molly R. Morris and Juan Rull for their comments on earlier versions of the manuscript. References Andersson, M. 1994. Sexual Selection. Princeton, New Jersey: Princeton University Press. Basolo, A. 1990a. Female preference for male sword length in the green swordtail, Xiphophorus helleri (Pisces: Poeciliidae). Animal Behaviour, 40, 332e338. Basolo, A. 1990b. Female preference predates the evolution of the sword in the swordtail fish. Science, 250, 808e810. Basolo, A. 1998. Shift in investment between sexually selected traits: tarnishing of the silver spoon. Animal Behaviour, 55, 665e671. Basolo, A. L. & Alcaraz, G. 2003. The turn of the sword: length increases male swimming costs in swordtails. Proceedings of the Royal Society B, 270, 1631e 1636. Basolo, A. L. & Wagner, W. E., Jr. 2004. Covariation between predation risk, body size and fin elaboration in the green swordtail, Xiphophorus helleri. Biological Journal of the Linnean Society, 83, 87e100. Baumgartner, A., Coleman, S. & Swanson, B. 2011. The cost of the sword: escape performance in male swordtails. PLoS ONE, 6, e15837. Carleton, K. 2009. Cichlid fish visual systems: mechanisms of spectral tuning. Integrative Zoology, 4, 75e86. Charlesworth, B. 1980. Evolution in Age-structured Populations. Cambridge: University Press. Christensen, B. 1996. Predator foraging capabilities and prey antipredator behaviours: re-versus postcapture constrains on size dependent predatoreprey interactions. Oikos, 76, 368e380.
Cumming, G. & Finch, S. 2001. A primer on the understanding, use, and calculation of confidence intervals that are based on central and noncentral distributions. Educational and Psychological Measurement, 61, 532e574. Cummings, M. E., Rosenthal, G. G. & Ryan, M. J. 2003. A private ultraviolet channel in visual communication. Proceedings of the Royal Society B, 270, 897e904. Endler, J. A. 1991. Interactions between predators and prey. In: Behavioural Ecology: an Evolutionary Approach (Ed. by J. R. Krebs & N. B. Davies), pp. 169e196. Cambridge: Blackwell Scientific. Endler, J. A. 1995. Multiple-trait coevolution and environmental gradients in guppies. Trends in Ecology & Evolution, 10, 22e29. Fuiman, L. & Magurran, A. 1994. Development of predator defenses in fishes. Reviews in Fish Biology and Fisheries, 42, 145e183. Godin, J.-G. J. & McDonough, H. E. 2003. Predator preference for brightly colored males in the guppy: a viability cost for a sexually selected trait. Behavioral Ecology, 14, 194e200. Grafen, A. 1990. Biological signals as handicaps. Journal of Theoretical Biology, 144, 517e546. Houde, A. 1987. Mate choice based upon naturally occurring color-pattern variation in a guppy population. Evolution, 41, 1e10. Houde, A. & Torio, A. 1992. Effect of parasitic infection on male color pattern and female choice in guppies. Behavioral Ecology, 3, 346. Hurvich, C. M. & Tsai, C. 1991. Bias of the corrected AIC criterion for underfitted regression and time series models. Biometrika, 78, 499e509. Jennions, M., Møller, A. & Petrie, M. 2001. Sexually selected traits and adult survival: a meta-analysis. Quarterly Review of Biology, 76, 3e36. Johnson, M. & Agrawal, A. 2003. The ecological play of predatoreprey dynamics in an evolutionary theatre. Trends in Ecology & Evolution, 18, 549e551. Johnson, J. & Basolo, A. 2005. Predator exposure alters female mate choice in the green swordtail. Behavioral Ecology, 14, 619e625. Johnstone, R. A. 1995. Sexual selection, honest advertisement and the handicap principle: reviewing the evidence. Biological Reviews, 70, 1e65. Juanes, F. 1994. What determines prey size selectivity in piscivorous fishes? Theory and application in fish feeding ecology. In: Theory and Application in Fish Feeding Ecology (Ed. by D. J. Stouder, K. L. Fresh & R. J. Feller), pp. 79e101. Columbia: University of South Carolina Press. Juanes, F. & Conover, D. 1994. Piscivory and prey size selection in young-of-theyear bluefish: predator preference or size-dependent capture success? Marine Ecology Progress Series, 114, 59e69. Kodric-Brown, A. & Johnson, S. C. 2002. Ultraviolet reflectance patterns of male guppies enhance their attractiveness to females. Animal Behaviour, 63, 391e 396. Kotiaho, J. S. 2001. Costs of sexual traits: a mismatch between theoretical considerations and empirical evidence. Biological Reviews, 76, 365e376. Kotiaho, J. S., Alatalo, R. V., Mappes, J., Parri, S. & Rivero, A. 1998. Male mating success and risk of predation in a wolf spider: a balance between sexual and natural selection? Journal of Animal Ecology, 67, 287e291. Lafferty, K. 1993. The marine snail, Cerithidea californica, matures at smaller sizes where parasitism is high. Oikos, 681, 3e11. Lima, S. 1998. Stress and decision-making under the risk of predation: recent developments from behavioral, reproductive, and ecological perspectives. Advances in the Study of Behaviour, 27, 215e290. MacLaren, R. D. & Daniska, D. 2008. Female preferences for dorsal fin and body size in Xiphophorus helleri: further investigation of the LPA bias in poeciliid fishes. Behaviour, 145, 897e913. MacLaren, R. D. & Fontaine, A. 2012. Female preference for male lateral projection area in Poecilia reticulata. Environmental Biology of Fishes, 93, 105e119. Magnhagen, C. & Kvarnemo, L. 1989. Big is better: the importance of size for reproductive success in male Pomatoschistus minutus (Pallas)(Pisces, Gobiidae). Journal of Fish Biology, 35, 755e763. Miller, R. R., Minckley, W. & Norris, S. 2005. Freshwater Fishes of Mexico. Chicago: University of Chicago Press. Møller, A. & Nielsen, J. 1997. Differential predation cost of a secondary sexual character: sparrowhawk predation on barn swallows. Animal Behaviour, 54, 1545e1551. Moodie, G. 1972. Predation, natural selection and adaptation in an unusual threespine stickleback. Heredity, 282, 155e167. Morris, M., Rios-Cardenas, O. & Darrah, A. 2008. Male mating tactics in the northern mountain swordtail fish (Xiphophorus nezahualcoyotl): coaxing and coercing females to mate. Ethology, 114, 977e988. Moyaho, A., García, C. M. & Manjarrez, J. 2004. Predation risk is associated with the geographic variation of a sexually selected trait in a viviparous fish (Xenotoca variata). Journal of Zoology, 262, 265e270. Rauchenberger, M., Kallman, K. & Morizot, D. 1990. Monophyly and geography of the Rio Panuco basin swordtails (genus Xiphophorus) with descriptions of four new species. American Museum Novitates, 29, 1e41. Reynolds, J. D. & Gross, M. R. 1992. Female mate preference enhances offspring growth and reproduction in a fish, Poecilia reticulata. Proceedings of the Royal Society B, 250, 57e62. Reznick, D. & Endler, J. A. 1982. The impact of predation on life history evolution in Trinidadian guppies (Poecilia reticulata). Evolution, 36, 160e177. Reznick, D., Bryga, H. & Endler, J. A. 1990. Experimentally induced life-history evolution in a natural population. Nature, 346, 357e359. Rosenthal, G. G. 2000. The Behavioral Ecology of Visual Signaling in Swordtails. Austin: University of Texas.
A. Hernandez-Jimenez, O. Rios-Cardenas / Animal Behaviour 84 (2012) 1051e1059 Rosenthal, G. & Evans, C. 1998. Female preference for swords in Xiphophorus helleri reflects a bias for large apparent size. Proceedings of the National Academy of Sciences, U.S.A., 95, 4431e4436. Rosenthal, G. G., Flores-Martinez, T. Y., Garcia de León, F. J. & Ryan, M. J. 2001. Shared preferences by predators and females for male ornaments in swordtails. American Naturalist, 158, 146e154. Rosenthal, G. G., Wagner, W. E., Jr & Ryan, M. J. 2002. Secondary reduction of preference for the sword ornament in the pygmy swordtail Xiphophorus nigrensis (Pisces: Poeciliidae). Animal Behaviour, 63, 37e45. Scharf, F. S., Juanes, F. & Rountree, R. A. 2000. Predator sizeeprey size relationships of marine fish predators: interspecific variation and effects of ontogeny and body size on trophic-niche breadth. Marine Ecology Progress Series, 208, 229e248. Sih, A. 1985. Predators and prey lifestyles: an evolutionary and ecological overview. In: Predation: Direct and Indirect Impacts on Aquatic Communities (Ed. by W. C. Kerfoot & A. Sih), pp. 203e224. Hanover, New Hampshire: University Press of New England. Trainor, B. C. & Basolo, A. L. 2000. An evaluation of video playback using Xiphophorus helleri. Animal Behaviour, 59, 83e89.
1059
Valtierra-Vega, M. T. & Schmitter-Soto, J. J. 2000. Hábitos alimentarios de las mojarras (Perciformes: Cichlidae) de la laguna Caobas, Quintana Roo, México. Revista de Biología Tropical, 48, 503e508. Webb, P. W. 1981. Responses of northern anchovy, Engrualis mordax, larvae to predation by a biting planktivore, Amphiprion percula. U.S. National Marine Fisheries Service, Fish Bulletin, 79, 727e735. Werner, E., Gilliam, J., Hall, D. & Mittelbach, G. 1983. An experimental test of the effects of predation risk on habitat use in fish. Ecology, 64, 1540e1548. Winemiller, K., Leslie, M. & Roche, R. 1990. Phenotypic variation in male guppies from natural inland populations: an additional test of Haskins’ sexual selection/ predation hypothesis. Environmental Biology of Fishes, 29, 179e191. Zahavi, A. 1977. The cost of honesty (further remarks on the handicap principle). Journal of Theoretical Biology, 67, 603e605. Zahavi, A. & Zahavi, A. 1997. The Handicap Principle: a Missing Piece of Darwin’s Puzzle. Oxford: Oxford University Press. Zeh, D. W. & Zeh, J. A. 1988. Condition dependent sex ornaments and field tests of sexual selection theory. American Naturalist, 132, 454e459. Zuk, M. & Kolluru, G. 1998. Exploitation of sexual signals by predators and parasitoids. Quarterly Review of Biology, 73, 415e438.
