As the Sword Grows: Individual Variation and Ontogenetic Effects of a Sexually Selected Trait on Locomotor Performance in <italic>Xiphophorus hellerii</italic>

Previous studies aimed at detecting costs of sexually selected traits have yielded mixed results partly because of variable methods. We present a novel approach: a repeated-measures design to examine individual variation in locomotor performance of male Xiphophorus hellerii as the sexually selected sword develops ontogenetically and to determine whether the growth of a sexually selected trait alters consistency of performance. Individual differences in sprint speed, critical swimming speed (stamina), and relative sword length were statistically repeatable over 9 wk. However, using the Akaike Information Criterion corrected for small sample sizes, the best-fit predictive models for swimming performance did not include sword length or relative sword length. Furthermore, in less supported models and within-week comparisons, there was no statistically significant effect of sword length on performance. These results suggest little effect of the sword on locomotor abilities, which is inconsistent with results from some previous experimental manipulations, possibly because compensatory traits develop ontogenetically in parallel with the sword. However, our results are consistent with correlational studies of natural variation that suggest no locomotor cost of the sword. These results do not necessarily imply a complete lack of a cost to the sword but rather lack of a functional cost for swimming performance. Introduction Sexual selection, driven by competition for mates or by mate choice, often results in the evolution of exaggerated behaviors or morphological structures (Andersson 1994). These traits evolve because they increase the reproductive success of the bearer and are assumed to incur a cost with respect to natural selection such that their evolutionary exaggeration will eventually stop. Detecting these putative costs has long been a goal of behavioral ecologists and evolutionary biologists (Kotiaho 2001). Costs have been examined in various taxa from several perspectives that have led to a variety of results (Kotiaho 2001; Oufiero and Garland 2007). The two most common ways to asses whether a putatively sexually selected trait is costly are to (1) experimentally manipulate the structure (Clark and Dudley 2009) and (2) compare individuals with naturally varying expressions of the trait (Wilson et al. 2010). Both approaches have led to conflicting results even within a single genus, such as Xiphophorus fish (Ryan 1988; Basolo and Alcaraz 2003; Royle et al. 2006; Kruesi and Alcaraz 2007; Baumgartner et al. 2011). Xiphophorus are freshwater live-bearing fish found throughout Mexico and part of Central America. Males in several species exhibit an elongation of the lower margin of the caudal fin (sword) that is favored because of a preexisting female sensory bias (Basolo 1990, 1995), and it is one of the first examples of sexual selection proposed by Darwin (1871). Because the sword is physically linked to the thrust-producing system, several previous studies have sought to determine the energetic and locomotor costs associated with the trait. For example, completely removing the sexually selected sword through experimental manipulation in adult male Xiphophorus montezumae (the species with the longest sword) has suggested costs with respect to routine oxygen consumption and critical swimming speed (a measure of endurance; Basolo and Alcaraz 2003; Kruesi and Alcaraz 2007) but not for burst speed during a Cstart in Xiphophorus hellerii (Baumgartner et al. 2011). Correlational studies of natural variation in sword length (SW) suggest no cost to endurance in Xiphophorus nigrensis (Ryan 1988) and that it may even be beneficial to burst speed during a C-start in X. hellerii (Royle et al. 2006). These discrepancies may arise for a number of reasons, including most simply a lack of consistency in species and methods used. However, the discrepancy may also be due to the fact that completely removing the trait may alter the functional relationships of various traits that affect locomotion. Ontogenetic Effects of the Sword 685 The repeatability of a trait is important for several reasons (Boake 1989; Chappell et al. 1996; Dohm 2002; Garland 2010; Careau and Garland 2012). Perhaps most notably, if individual differences in behavior, performance, or any other trait are not consistent, then that trait is a “difficult” target for natural, sexual, or artificial selection. Several studies have assessed the repeatability of swimming performance in fish (Kolok 1992, 1999; Kolok et al. 1998; Reidy et al. 2000; Claireaux et al. 2007; Oufiero and Garland 2009; Handelsman et al. 2010; Marras et al. 2010, 2011) and of sexual signals in various taxa (Kotiaho et al. 1996; Jang et al. 1997). Other studies have examined the effect of sexual ornaments on swimming performance in fish (Nicoletto 1991; Basolo and Alcaraz 2003; Langerhans et al. 2005; Kruesi and Alcaraz 2007; Wilson et al. 2010; Baumgartner et al. 2011). However, no studies to date have examined the consistency of individual performance in relation to a sexually selected trait. That is, do individual differences in performance retain any statistical consistency during or after the development of a sexual ornament? Here, we (1) present an alternative approach to detect potential costs of sexually selected traits by comparing males with and without the trait but without experimental manipulation and (2) assess the effect of the development of a sexual ornament on individual variation in locomotor performance. Using a repeated-measures design and an information-theoretic approach to statistical model selection, we determined whether sword length is an important predictor of two forms of locomotion during ontogenetic development in male green swordtails X. hellerii and whether individuals are consistent in performance as the sword develops. Furthermore, unlike many exaggerated morphological structures (e.g., eye stalks of flies; Swallow et al. 2000; Husak and Swallow 2011), the sword is directly linked to the thrust-producing system and presents an opportunity to examine functional morphological relationships in a system where the trait (sword) has evolved for purposes other than locomotion. If the sword is a functional cost and increases drag or energetic burden ontogenetically, then all else being equal, we expect males to have a reduction in locomotor performance as the sword grows, similar to experimental studies that removed the trait (Basolo and Alcaraz 2003; Kruesi and Alcaraz 2007) or correlational studies examining the effect of pregnancy on swimming (Plaut 2002).