Abstract: Mate-choice copying is a form of social learning in which an individual’s choice of mate is influenced by the apparent choices of other individuals of the same sex and has been observed in more than 20 species across a broad taxonomic range. Though fitness benefits of copying have proven difficult to measure, theory suggests that copying should not be beneficial for all species or contexts. However, the factors influencing the evolution and expression of copying have proven difficult to resolve. We systematically searched the literature for studies of mate-choice copying in nonhuman animals and, then, performed a phylogenetically controlled meta-analysis to explore which factors influence the expression of copying across species. Across 58 published studies in 23 species, we find strong evidence that animals copy the mate choice of others. The strength of copying was significantly influenced by taxonomic group; however, sample size limitations mean it is difficult to draw firm conclusions regarding copying in mammals and arthropods. The strength of copying was also influenced by experimental design: copying was stronger when choosers were tested before and after witnessing a conspecific’s mate choice compared to when choosers with social information were compared to choosers without. Importantly, we did not detect any difference in the strength of copying between males and females or in relation to the rate of multiple mating. Our search also highlights that more empirical work is needed to investigate copying in a broader range of species, especially those with differing mating systems and levels of reproductive investment.
DISCUSSION
We performed a meta-analysis of 158 effect sizes from 58 published studies of mate-choice copying in nonhuman animals. Overall, we find strong evidence that animals copy the mating preferences of others. The overall standardized mean difference was 0.57 (95% CI = 0.34–0.81), which is considered a medium effect (Cohen 1992). This converts to a mean odds ratio of 2.83 so that the observers in this sample are almost three times as likely to choose a partner or phenotype that they previously observed in a positive sexual interaction with a rival compared to individuals with no social information. This result suggests that social information can have a significant influence on animal mating preferences across a wide range of species. Further, the strength of copying observed is similar to that reported in the recent meta-analysis of female copying by Jones and DuVal (2019: a mean odds ratio of 2.71), of which our data set overlaps. We also detected significant variation in the degree of copying across species, which was partly explained by taxonomic group: copying was strongest for mammals and weakest for arthropods. The strength of copying was also influenced by the type of experiment used to test for copying: copying was stronger when studies compared mating behavior before and after the presentation of social information in contrast to studies in which separate control and social treatment groups were compared. Interestingly, we did not detect any difference in copying between males and females or in relation to the rate of multiple mating. We also detected little evidence for publication bias against nonsignificant results.
Whereas evidence suggests that mate-choice copying is taxonomically widespread, a small number of model species are overrepresented in our data set and in the literature more generally: of the 89 studies identified in our search, 20 focus on guppies, 9 on Japanese quail, and 9 on D. melanogaster. Despite this, species identity explained a small proportion of the variation in effect size in our analysis. This confirms that copying is not driven by a few influential species and that within-species variability in effect size is high. Phylogenetic history explained only a small amount of variation in effect sizes, indicating that any similarities between species are not due to them being closely related evolutionarily. This is unsurprising, given the large evolutionary distances between the species in the data set (with the exception of species in the Poeciliidae) and the evolutionary lability of behavioral traits (Blomberg et al. 2003). Nevertheless, we did detect differences in the degree of copying when we sorted the species in our data set into four broad taxonomic groups: mammals, birds, fishes, and arthropods (incorporating insects, arachnids, and crustaceans). The degree of copying was highest for mammals, followed by birds and fishes, and then lowest for arthropods. This finding could be explained by broad-scale differences in cognitive ability, ecological conditions, or social group size. However, these results should be interpreted with caution: first, because the number of effect sizes for mammals and arthropods was small and, second, because the small number of species tested across the entire data set means it is unclear how generalizable these results are. Importantly, we need more tests of copying in species groups other than those highlighted here. For example, to our knowledge, there have been no experimental tests of copying in amphibians, reptiles, or nonhuman primates.
We found that the type of experimental design significantly influenced the degree of copying. Copying was stronger for studies with a “before-and-after” design compared to studies with separate control and social treatment groups (no pretest). This is the opposite of what we predicted based on our hypothesis that no pretest studies may often underestimate preferences in the absence of social information. We note, however, that our categories of “before-and-after” and “no pretest” designs mostly align with the “random” and “unattractive” categories used by Jones and Duval (2019), which relate to whether the study used a reversal paradigm to test for copying (the unattractive category). The alignment comes from the fact that the reversal paradigm is a subset of the “before-and-after” test. Accordingly, Jones and DuVal (2019) found that copying was stronger when using a reversal paradigm, which is consistent with our result shown here. One potential explanation for the observed effect could be related to the fact that before-and-after designs test each individual twice, potentially in quick succession. This provides the opportunity for observers to gain experience that could influence their choice in a nonrandom way. For example, individuals may become choosier with successive encounters because they have more information on the quality of mates in the population, they perceive mate density to be higher, or because they are attempting to “trade-up” from previous mates (e.g., Pitcher et al. 2003).
There were several moderators which did not significantly influence the degree of copying. For example, copying was performed to a similar extent in animals born in the wild or in captivity. This, first, confirms that mate-choice copying is not an artifact of captivity—it is a real behavior that exists in wild populations. However, the recent meta-analysis by Jones and DuVal (2019) found that females copied more when tested in the wild compared to when tested in captivity. This discrepancy likely arises because our moderator focuses on observer history prior to, but not during, the behavioral tests and suggests that the environment during the test is more important for influencing copying behavior. Copying was not influenced by the mating rate of the species tested. This null result could be explained partly by the lack of data on species with low rates of multiple mating, and we suggest that more studies should be carried out on such species. Importantly, any conclusions relating to this moderator may also be further limited by the fact that species with low rates of multiple mating are taxonomically limited in our sample, occurring only in birds and arthropods. Nevertheless, mate-choice copying has been shown in at least two bird species that show some form of social monogamy and biparental care (zebra finches and Japanese quail), even though theory suggests that copying should not be favored because of the diminishing returns associated with sharing a mate that provides direct benefits or parental care (Vakirtzis and Roberts 2009). We also found no difference in the strength of generalized versus individual copying. These two forms of copying may require different cognitive abilities, but whether this leads to differences in the strength of copying is not clear. Crucially, the relative prevalence, and strength, of these two forms of copying has important evolutionary consequences. Both forms of copying can widen the gap between attractive and unattractive males (Leadbeater 2009), create frequency-dependent bias in mate choice (Santos et al. 2014) and support the invasion of new traits into a population (Santos et al. 2017). However, because generalized copying allows learned preferences to be applied to multiple mates, it can lead to the cultural transmission of preferences, and so has the potential to be a much stronger evolutionary force than individual copying.
Most surprisingly, we found no evidence that males copy less than females. This result is unexpected as males may be substantially increasing their exposure to sperm competition by copying the mate choices of their rivals (Simmons 2001; Vakirtzis and Roberts 2009). In support of this, there is widespread evidence that males often prefer to mate with virgin females if given the choice (Bonduriansky 2001; Simmons 2001). Again, this could be partly explained by the relatively small number of studies testing male copying. Another possibility is that when faced with previously mated females, males may ejaculate more sperm as a defense against combat sperm competition, instead of rejecting a mating opportunity (Simmons 2001; Kelly and Jennions 2011). The ability of males to strategically allocate sperm in this way has the potential to reduce the costs of copying a rival’s choice. Copying could also be beneficial to males if they can reduce the risk of being copied themselves, perhaps by mating or courting out of view of rival males (Simmons 2001; Brown et al. 2012; le Roux et al. 2013) or by mating with less-preferred females when rivals are present as a form of deception (also known as the “audience effect”: Plath et al. 2008; Castellano et al. 2016; Witte et al. 2018). Alternatively, male mate-choice copying may be likely when males face high costs of reproduction as appears to be the case for the fantail darter Etheostoma flabellare, the three-spined stickleback Gasterosteus aculeatus, and the broadnosed pipefish Syngnathus typhle (Witte et al. 2015). Indeed, mate-choice copying studies have often been performed using species with paternal care (fish and zebra finches) and, therefore, high male reproductive investment. It will, therefore, be useful to test for copying in species that exhibit a broader range of mating systems and reproductive strategies than that shown by the current available sample.
Several other aspects of experimental design did not influence the degree of mate-choice copying. For example, copying was observed regardless of the type of social information available to the observer (whether the demonstrator mated, courted, avoided, or affiliated with a partner), how the observer preference was measured, or whether the demonstrator was allowed to “choose” a target individual or was artificially placed with one. Nevertheless, variations in other aspects of experimental design do have the potential to contribute to some of the unexplained variability in effect sizes seen across studies. For example, for indirect measures of mate choice (Rosenthal 2017), the criteria used to determine mating preferences differ between studies. Vukomanovic and Rodd (2007) conducted mate-choice copying experiments with the sailfin molly P. latipinna using different criteria to investigate how this could influence the measurement of copying. Using the criteria of Dugatkin (1992), where females had to spend over 50% of their time within a 13-cm “preference zone” in front of the male, significant copying was detected. However, using the criteria of Lafleur et al. (1997), where females had to spend 15 consecutive seconds in a 2.5-cm preference zone, no copying was detected. This demonstrates that even when recording the same behaviors, results can differ depending on exactly which methodology is used.
Though several moderators did influence the strength of copying in our sample, most of the heterogeneity in effect size detected (both here and in Jones and DuVal 2019) remains unexplained. This suggests that there are other factors that influence the degree of copying, which we were unable to include in our meta-analysis. In many cases, this is because of the difficulty associated with obtaining good estimates across all species sampled. For example, a potentially important driver of copying, and mate choice in general, is the cost of mate sampling. However, we have very few estimates of the costs of mate sampling and mate choice in any species so that broad-scale comparisons are difficult (Rosenthal 2017). Interestingly, empirical studies that have attempted to indirectly measure this effect, by testing copying in high-cost environments, have failed to find any significant change in copying (Briggs et al. 1996; Dugatkin and Godin 1998). Another key driver of copying is expected to be the ability to assess mate quality, which again will be difficult to accurately estimate for all species. However, empirical tests have found some support for this effect: for example, copying is more likely when mates are more similar in guppies and mollies (Dugatkin 1996; Witte and Ryan 1998). Related to this, a valuable topic for future research would be the nature of memory in the context of mate-choice copying. How long are demonstrations remembered for? How easily are these memories forgotten or overwritten with conflicting experiences?
In summary, this meta-analysis of 58 experimental studies suggests that mate-choice copying is a widespread and robust phenomenon across the animal kingdom. Despite different methodological approaches, the average effect size is very similar to that found in the recent meta-analysis of mate-choice copying by Jones and DuVal (2019), of which our data set overlaps. We found evidence that the amount of copying differs significantly across taxonomic groups and between the two main types of experimental design found in the literature but no evidence that copying is influenced by sex or the rate of multiple mating. Importantly, our systematic review has also revealed areas where more empirical work is needed. Unfortunately, though mate-choice copying has been observed across a wide taxonomic range, most studies are still only performed on a few model species. In addition to uncovering the details of how mate-choice copying operates within a species, time should be taken to investigate how widespread the phenomenon is across the animal kingdom and which aspects of a species’ ecology or biology predispose the evolution of copying.
Posts
No comments:
Post a Comment