Austral Ecology (2011) 36, 68–75 Sexual dimorphism and sexual selection in a montane scincid lizard (Eulamprus leuraensis)aec_2119 68..75 SYLVAIN DUBEY,* MAYA CHEVALLEY AND RICHARD SHINE Biological Sciences A08, University of Sydney, Sydney, NSW 2006, Australia (Email: [email protected]) Abstract Sex-based divergences in body sizes and/or shapes within a species imply that selective forces act differently on morphological features in males versus females. That prediction can be tested with data on the relationship between morphology and reproductive output in females, and between morphology and realized paternity (based on genetic assignment tests) in males. In a sample of 81 field-collected adult Blue Mountains water skinks (Eulamprus leuraensis), males and females averaged similar overall body sizes (snout–vent lengths (SVLs)). Reproductive success (based on 105 progeny produced by the females) increased with SVL at similar rates in both sexes (as expected from the lack of sexual size dimorphism). Multiple paternity was common. Males had larger heads than females of the same body size, and (as predicted) reproductive success increased with relative head size in males but not in females. Males also had relatively longer limbs and shorter trunks than females, but we did not detect significant sex differences in selection on those traits. Reproductive success in both sexes was increased by relatively longer hind limbs. Our data clarify mating systems in this endangered species, and suggest that mating systems are diverse within the genus Eulamprus. Key words: fecundity selection, mating system, multiple paternity, reptile, Scincidae. INTRODUCTION determined trait, only a limited set of selective forces can be invoked to explain that divergence. For Darwinian theory predicts that patterns of phenotypic example, larger body size might enhance female repro- variation within a population are strongly influenced ductive success because it allows greater fecundity, or by the ways in which those variations translate into might enhance male reproductive success if it underlying fitness differentials among individuals. increased a male’s ability to win battles with rival Thus, for example, geographic variation in the fre- males (e.g. Vitt & Cooper 1985; Lebas 2001; Olsson quency of alternative morphotypes (such as coloura- et al. 2002; Du et al. 2005; Stuart-smith et al. 2008; tion or body shape) is thought to reflect corresponding Dubey et al. 2009). Hence, studies on sexual dimor- geographic variation in the selective advantages or dis- phism provide an excellent opportunity to examine the advantages of those alternative trait values (e.g. Endler putative selective basis for divergence in morphologi- 1995; Endler & Houde 1995). Unfortunately, spatial cal traits (Vitt & Cooper 1985; Andersson 1994; variation of this kind also can be generated by factors Madsen & Shine 1995; Shine et al. 1998). Some of the such as variable local conditions (inducing phenotypi- data needed for such analyses are relatively straight- cally plastic trait expression, e.g. Downes 1999; Shine forward to obtain. It is easy to measure morphological & Elphick 2001), and by phylogenetic history (differ- traits in the two sexes, and to monitor reproductive ent lineages in different areas, e.g. Dubey et al. 2007). output of females (at least on a per-litter basis). Quan- Even greater confounding factors plague attempts to tifying male reproductive success poses greater logis- interpret the selective basis of interspecific variation in tical challenges, but the development of genetic phenotypic traits (e.g. Harvey & Pagel 1991). Charles methods for paternity assignment now enables us to Darwin (1871) suggested that comparisons of selective make direct comparisons between the sexes in terms of forces and responses between males and females reproductive output (e.g. Stapley & Keogh 2005; within a single species might provide a way to over- Dubey et al. 2009). Ideally, such comparisons should come some of these logistical problems. measure lifetime reproductive success, but few studies Conspecific males and females experience similar have achieved that ambitious aim (e.g. Clutton-Brock environmental conditions, and have the same genetic 1988; Madsen & Shine 1994; Krüger 2002). If selec- heritage. Thus, if the sexes differ in some genetically- tive forces diverge strongly between the sexes, comparisons based on data from a single reproduc- *Corresponding author. tive season can be used to look for the predicted Accepted for publication December 2009. correlations between sexually dimorphic traits and © 2010 The Authors doi:10.1111/j.1442-9993.2010.02119.x Journal compilation © 2010 Ecological Society of Australia SEXUAL SELECTION IN A SCINCID LIZARD 69 sex-specific selective forces. We have conducted such Six microsatellite loci isolated and characterized from an analysis for a montane lizard species from south- Eulamprus kosciuskoi (Scott et al. 2001: EK8, EK23, EK37, eastern Australia. EK100, EK107) and Gnypetoscincus queenslandiae (Sumner et al. 2001: GQ20/21) were amplified and scored. Amplified products were genotyped with a 3130 xl genetic analyser (Applied Biosystems) using Genemapper software V3.7 METHODS (Applied Biosystems). Polymerase chain reaction amplifica- tions were performed in a 9800 Fast thermal cycler (Applied Study species, area and sampling protocol Biosystems) as 5 mL reactions containing 0.075 U Ta q T i DNA polymerase (Biotech), 0.1 mmol dNTPs, 0.4 mmol of The Blue Mountains water skink, Eulamprus leuraensis,isa each primer, 20 mmol Tris-HCl, pH 8.5, and 50 mmol KCl, medium-sized (to 85-mm snout–vent length (SVL), 15 g) 1.25 mmol MgCl2. Cycling conditions included a hot start viviparous lizard restricted to a scarce and highly fragmented denaturation of 95°C for 3 min, followed by 35 cycles of 95°C habitat type (‘hanging swamps’) along the Great Dividing for 30 s, 60°C (55°C for EK23, GQ20/21 and EK37) anneal- Range west of Sydney. Known from less than 40 populations, ing temperature for 30 s, 72°C for 30 s (1 min for EK23, the species is classified as endangered under both state and GQ20/21 and EK37), and a final extension of 72°C for federal legislation (Threatened Species Conservation Act – 30 min. NSW 1995; Environmental Protection and Biodiversity Conservation Act – Commonwealth 1999; Dubey & Shine 2010). We collected tissue samples (tail clips) from 81 adult Statistical analyses of morphology and E. leuraensis during fieldwork from November 2008 to April reproductive success 2009, from six small populations (Table 1). Our sample com- prised 40 males and 45 adult (and thus, potentially gravid) We compared mean SVLs between the sexes with anova.To females, and the latter were retained in captivity until compare body proportions between the sexes, we first con- parturition. We housed lizards individually in plastic boxes ducted a Principal Components (PC) analysis on all mor- (320 ¥ 220 ¥ 100 mm) in a room maintained at 18°C (day- phological variables (SVL, head length, limb lengths, light period: 07.00–19.00 hours). Underfloor heating cables interlimb distance, body mass) and took the first PC axis allowed each female to control her body temperature over the (which explained 78.3% of the total variance: loading factors range 20–33°C for part of the day. Each female was fed five for SVL, head, front limbs, back limbs, interlimb and mass crickets twice weekly.The 41 females produced a total of 105 were 0.433, 0.425, 0.420, 0.392, 0.346 and 0.427, respec- neonates, all of which were weighed and measured <24 h tively) as our index of overall body size.We used this PC axis after birth. asacovariateinancovas with sex as the factor and morpho- logical traits as dependent variables. To explore links between morphology and reproductive DNA extraction and microsatellite analysis success, we conducted two types of analyses. In each case, the morphological variables were SVL, and residual scores from Tissues were placed in 200 mL of 5% Chelex containing the general linear regression of each trait (e.g. head length) 0.2 mg mL-1 of proteinase K, incubated overnight at 56°C, against our overall measure of body size (i.e. PC axis 1).The boiled at 100°C for 10 min and centrifuged at 12 000 rpm for first analysis was based on whether or not an individual was 10 min.Then the supernatant (containing purified DNA) was known to have reproduced in the year of collection (based on removed and stored at -20°C. maternity and paternity of the captive-born offspring); this category plus sex were included as factors in a two-factor anova with morphological variables as dependent variables. Table 1. Sample sizes of free-ranging Blue Mountains We cannot exclude the possibility that some males classified water skinks for which genetic data were gathered as ‘non-reproductive’ did indeed sire offspring with females other than the ones we sampled. However, this potential Adult Potential error in classification should be randomly distributed among females fathers individuals of different body sizes, and consequently should Site sampled Litters Neonates sampled not create any directional bias in patterns of correlation between morphology and mating success. BH3 8 7 20 5 The second set of analyses treated reproductive success as BH4 13 13 32 5 a continuous variable (based on the number of offspring KT1 5 5 12 7 produced by each adult lizard), using the morphological MH4 3 3 7 3 MRP1 10 8 19 9 measures as covariates, sex as a factor, and reproductive WF7 6 5 15 11 success as the dependent variable. To generate a measure of Total 45 41 105 40 reproductive success that can be compared between the sexes, we calculated the number of offspring per parent The table shows the number of adult (and thus, potentially divided by the mean number of offspring for adults of that gravid) females sampled, litters, neonates and potential sex.The standardization was necessary because our raw data fathers (number of adult males tested) sampled for each site.