Proximate causes of altitudinal differences in body size in an agamid lizard

Abstract Body size is directly linked to key life history traits such as growth, fecundity, and survivorship. Identifying the causes of body size variation is a critical task in ecological and evolutionary research. Body size variation along altitudinal gradients has received considerable attention; however, the underlying mechanisms are poorly understood. Here, we compared the growth rate and age structure of toad‐headed lizards (Phrynocephalus vlangalii) from two populations found at different elevations in the Qinghai‐Tibetan Plateau. We used mark‐recapture and skeletochronological analysis to identify the potential proximate causes of altitudinal variation in body size. Lizards from the high‐elevation site had higher growth rates and attained slightly larger adult body sizes than lizards from the low‐elevation site. However, newborns produced by high‐elevation females were smaller than those by low‐elevation females. Von Bertalanffy growth estimates predicted high‐elevation individuals would reach sexual maturity at an earlier age and have a lower mean age than low‐elevation individuals. Relatively lower mean age for the high‐elevation population was confirmed using the skeletochronological analysis. These results support the prediction that a larger adult body size of high‐elevation P. vlangalii results from higher growth rates, associated with higher resource availability.

Compared with the animals living at relatively warm low-elevation sites, high-elevation ectotherms are expected to be larger at sexual maturity following a negative temperature-size relationship (ectotherms usually grow slower but mature at a larger size at lower rearing temperatures, e.g., Angilletta & Dunham, 2003), which is the case in some species (Angilletta, Steury et al., 2004;Morrison & Hero, 2003;Pincheira-Donoso, Hodgson, & Tregenza, 2008;Walters & Hassall, 2006). However, body size has been found to decrease with increased elevation in other species, which has been linked to resource limitations, and thus restricted growth (Chown & Klok, 2003;Hodkinson, 2005).
The Qinghai toad-headed lizard (Phrynocephalus vlangalii), which is widely distributed in the Qinghai-Tibetan Plateau with an elevation range from 2,000 m to 4,500 m, is an excellent model species for studying altitudinal patterns of life history traits in reptiles.
Altitudinal variation in body size of this species has been reported in previous studies; however, the results of these studies remain controversial. Jin, Liu, and Li (2007) found a negative relationship between body size of adult P. vlangalii and elevation, but a recent study showed that the females from high-elevation sites were significantly larger than those from low-elevation sites (Li, Zhou, & Liu, 2014). Despite the controversy in altitudinal pattern in body size, the underlying proximate causes of this altitudinal variation in body size remain elusive.
Here, we first compared the body size of P. vlangalii from two sites at different elevations (Maqu, 2,930 m elevation, hereafter the low-elevation site; Maduo, 4,250 m elevation, hereafter the highelevation site) in the northeast part of Qinghai-Tibetan Plateau.
Then, we measured the size of newborns produced by wild-caught females, measured growth rate and determined age structure via mark-recapture experiments, and identified adult age by skeletochronological analysis in these two populations. We found that body size was slightly larger in the high-elevation population than in the low-elevation population. In order to identify the proximate causes of altitudinal variation in body size in this species, we tested the following predictions derived from the aforementioned hypotheses using data on neonate size, growth rate, estimated age at maturity, and age structure of adults, respectively. First, if neonate size determines adult body size, newborns would be larger in the highelevation population than the low-elevation population. Second, if a faster growth rate leads to larger adult body size, juvenile lizards would grow more rapidly in the high-elevation population than the low-elevation population. Third, if lizards grow to larger body size through delayed maturity, lizards from the high-elevation population would have a longer duration of growth than the low-elevation population. Lastly, if increased survivorship (~longevity) leads to a larger adult body size, adult lizards would survive longer and would be older in the high-elevation population than the low-elevation population. and parturition occurs between mid-July and late August (Wu, Fu, Yue, & Qi, 2015). Females at low-elevation sites give birth earlier than those at high-elevation sites (Li et al., 2014). Offspring mass increases, but litter size and adult body size decrease with increasing elevation .

| Body size of newborns
In mid-July of 2011, 59 gravid females captured in the field (29 from the low-elevation and 30 from the high-elevation) were transferred to a laboratory in Hangzhou Normal University. The females were randomly housed in twelve 60 × 40 × 30 cm terraria (4-5 females per terrarium) filled with a 20 cm depth of moist sand. Terraria were housed in an AAPS (artificial atmospheric phenomena simulator) room at 20 ± 2°C on a natural light cycle. A 60-W light bulb was suspended above one end of each terrarium (20 cm above the terrarium floor) to provide supplementary heating from 0900 to 1700 hr. Food, that is, mealworms (larvae of Tenebrio molitor) and house crickets (Achetus domesticus), and water enriched with vitamins and minerals were provided ad libitum. Parturition occurred between late July and mid-August. All newborns were weighed and measured for SVL, and individually marked using the toe-clipping method. Fifty-four females (26 from the low-elevation and 28 from the high-elevation) gave birth to young that were all well developed, whereas the remaining females produced litters with various numbers of stillborns. The mean values for body size (SVL and mass) of newborns from a single litter were used in the following analyses to avoid pseudoreplication, and the abnormal litters were excluded from analyses. All females and juveniles were released at the site of maternal capture.

| Data analysis
We calculated an index of body condition using the residuals from a linear regression of ln-transformed body mass against ln-transformed SVL, and size-specific and mass-specific growth rates during the markrecapture experiments using the formula ln(measurement 2 /meas- where L t is the SVL at age t, L ∞ is the theoretical maximum length at infinite age, L 0 is the initial SVL, and K is a growth constant that describes the rate at which L ∞ is attained] is assumed to be appropriate to describe lizard growth (El Mouden, Znari, & Brown, 1999;Schoener & Schoener, 1978). We calculated von Bertalanffy growth parameters from Ford-Walford plots for each F I G U R E 1 Monthly mean air temperature and rainfall for the two sites where Phrynocephalus vlangalii were collected (data from http:// data.cma.cn/) population, which were constructed by regressing final (August 2012) SVL against initial (August 2011) SVL. Data from the mark-recapture study included individuals of 26.9-55.5 mm and 24.5-57.5 mm for the low-elevation and high-elevation populations, representing the upper 76% and 80% of the full-length ranges in the two populations, respectively. The values of L ∞ and K were calculated from the following equations: L ∞ = α/(1-β) and K = −ln β, where α is the y-intercept and β is the slope of Ford-Walford plots. Then, we used the standard and transformed [R = dS/dt = K(L ∞ -L t )] von Bertalanffy growth equations using the mean SVL of newborns of each population as the body size at age of 0 year (L 0 ) to estimate SVLs and growth rates at each age, respectively. Additionally, we used the transformed equation, t = ln[(L ∞ -L t )/(L ∞ -L 0 )]/-K, to estimate the age for each individual captured in the field and at sexual maturity, using the recorded SVLs of the smallest reproductive females of each population.
Between-sex differences in body size and growth rate of juvenile lizards were ignored because the gender of juveniles was difficult to identify using an uninjured technique. Accordingly, one-factor analysis of variance (ANOVA) was used to determine between-site difference in body size of field-captured juveniles and newborns, or age estimated by skeletochronology. Two-factor ANOVA was used to determine the differences in body size and estimated age of field-captured adults between populations and sexes, or differences in growth rate of recaptured juveniles between populations and years. Three-factor ANOVA was used to determine the differences in growth rate of recaptured adults between populations, sexes, and years. One-(or two-) factor analysis of covariance (ANCOVA) was used to determine the differences in mass between populations (and between sexes) after removing the effect of SVL. The values of Cohen's d for t test and partial eta-square (η p 2 ) for ANOVA (or ANCOVA) were presented as the measures of effect size to indicate the standardized difference and the proportion of effect variance in the total variance, respectively. A value of Cohen's d below 0.2 or η p 2 below 0.01 is considered as negligible, between 0.2 and 0.5 or between 0.01 and 0.06 as small, between 0.5 and 0.8 or between 0.06 and 0.14 as medium, and >0.8 or 0.14 as large (Cohen, 1988). Prior to parametric analyses, the normality of distributions and homogeneity of variances in the data were tested using the Kolmogorov-Smirnov test and Bartlett's test, respectively.
Throughout this article, values are presented as mean ± standard error (SE), and the significance level is set at α = .05.
Body mass of adults was affected by the interaction of site of origin × sex, with greatest mean value for high-elevation females but smallest for low-elevation males (F 1, 733 = 10.36, p < .01, η p 2 = 0.014).
Adult SVL was not affected by the interaction of site of origin × sex

| Growth of lizards in the field
Overall, juveniles at the high-elevation site grew faster than those at the low-elevation site (Table 1, Figure 3). Juveniles increased their SVL (but not mass) more rapidly in 2011 than in 2012 (Table 1).
The year × site of origin interaction had a significant effect on sizespecific growth rate, but not on mass-specific growth rate of juveniles (Table 1). The mean value of juvenile size-specific growth rate in 2011 was highest at the high-elevation site, but lowest at the low-elevation site ( Figure 3). Juvenile lizards increased their SVL (F 1, 124 = 74.94, p < .001, η p 2 = 0.377) and mass (F 1, 124 = 21.26, p < .001, η p 2 = 0.146) more rapidly than adult lizards (Figure 3). Both size-specific and massspecific growth rates of recaptured adults did not differ between years, between sites, and between sexes and were not affected by the interactions of these factors (Table 1).

| Body size of reproductive females and newborns
Females in the laboratory produced 54 clutches (26 from lowelevation site, 28 from high-elevation site) between late July and mid-August. The smallest reproductive females were 51.7 mm and 49.9 mm SVL for the low-elevation and high-elevation populations, respectively. Female SVL did not differ between the two popula-

| Age and growth rate estimation using the von Bertalanffy growth equation
The von Bertalanffy growth parameters, theoretical maximum length (L ∞ ) and the growth constant (K), derived from simulated mark-recapture datasets were 64.0 and 0.53, 70.2, and 0.45 for the low-elevation and high-elevation populations, respectively. Estimated age at sexual maturity for the low-elevation and high-elevation populations was 2.1 and 1.8 years, respectively. Estimated SVLs of high-elevation lizards were larger than those of low-elevation ones over the age of 2 years, and estimated growth rates of high-elevation lizards were higher than those of low-elevation ones at each age ( Figure 4).
Mean estimated age of field-captured adults was greater at the lowelevation site (3.13 ± 0.07 years) than that at the high-elevation site T A B L E 1 Results of two-factor (with year and site of origin as the factors for juveniles) or three-factor (with year, site of origin, and sex as the factors for adults) ANOVAs on specific growth rate of Phrynocephalus vlangalii at different elevations in the field mark-recapture experiments

Juvenile Adult
Size-specific growth rate Mass-specific growth rate Size-specific growth rate

| Age estimation by skeletochronology
Age determination by bone layers showed that the mean age for individuals from the low-elevation site (3.68 ± 0.24 years, ranging from 2 to 6 years) was greater than that from the high-elevation site (3.00 ± 0.22 years, ranging from 2 to 5 years; one-factor ANOVA with site of origin as the factor: F 1, 42 = 4.24, p = .046, η p 2 = 0.092).
In earlier studies of P. vlangalii, offspring size was believed to increase with increasing altitude Li et al., 2014). This difference between the results of these studies may be due to differences in measurement methods for reproductive traits or population sampling. In the  study, offspring size of P. vlangalii was assessed using the mass of scaled embryos that removed from pregnant females, rather than the mass of newborns. Scaled embryos continued to grow before parturition, consequently, offspring size might be underestimated. In fact, mean mass of embryos from the low-elevation population (0.89 g) was slightly larger than that from the high-elevation population (0.85 g), if others were excluded.
F I G U R E 4 The estimated snout-vent lengths and annual growth rates at each age using the von Bertalanffy growth equation F I G U R E 5 The estimated age of adult lizards captured from different elevation sites using the von Bertalanffy growth equation The Li et al. (2014) study included the high-elevation population that studied here, but not the low-elevation population. Therefore, these results from different studies may not be contradictory to each other and indicate that life history responses to local environments may be more complicated than expected and vary with site and year of the study (Angilletta, Steury et al., 2004;Stearns, 1992). That larger adult size for high-elevation lizards than low-elevation ones was also observed in the Li et al. (2014) study.
Newborns from high-elevation population were smaller than those from low-elevation population, which was inconsistent with our first prediction. If between-site difference in adult body size of P. vlangalii resulted from newborn size variation, the opposite pattern should be produced. Offspring size is assumed to affect animal growth, survival, and size at sexual maturity (Marshall & Keough, 2008;Räsänen, Söderman, Laurila, & Merilä, 2008). Large offspring are favored in poor-growing (such as low temperature, food scarcity) environments (Sinervo, 1990;Warner & Andrews, 2003). Why did not lager newborns occur in relatively colder, high-elevation environment in P. vlangalii? One possibility is that environmental condition at the highelevation site is not as disadvantageous as expected for lizard growth.
In fact, another study showed that potential prey availability at highelevation site was more abundant and led to higher growth rates for juveniles, than at low-elevation site (Lu et al., in review). Accordingly, offspring size is not likely to be an important source of variation in adult body size of P. vlangalii.
Consistent with our second prediction, high-elevation lizards grew more rapidly than low-elevation ones. The growth rate of lizards may be immediately affected by food availability (Iraeta et al., 2013). The high-elevation site could provide more food resources and allow P. vlangalii individuals to grow faster and surpass the adverse effects of smaller neonate body size, and finally reach a similar or even larger adult body size compared with the low-elevation site.
The estimated growth curve ( Figure 4) suggests that, despite a smaller size at birth, high-elevation juvenile P. vlangalii would reach a similar or larger size than low-elevation juveniles during the third active season. Presumably, parturition occurred naturally in late July and mid-August at the low-elevation and high-elevation sites, respectively (Li et al., 2014); newborns from both populations would become sexual maturity in that season (September and June for the low-elevation and high-elevation populations, respectively) according to estimated ages at sexual maturity. Therefore, larger body size of high-elevation adult P. vlangalii is likely to be due to faster growth during the juvenile stages. Similar results have been observed in other lizard species.
For example, hatchling Psammodromus algirus and Sceloporus graciosus from high-elevation populations is smaller, but grow faster over the active season to reach the same or larger size by the following years compared with those from low-elevation populations (Iraeta et al., 2006;Sears, 2005).
An individual animal can achieve a relatively large adult size by delaying maturation or prolonging growth period (Angilletta, Niewiarowski et al., 2004;Iraeta et al., 2006). Delayed maturation at a large body size occurs in some other species of reptiles and amphibians (Liao & Lu, 2011;Wapstra, Swain, & O'Reilly, 2001). In this study, however, the estimated age at sexual maturity was younger for the high-elevation population than for the low-elevation population, which was inconsistent with our third prediction. If delayed maturation occurred in the high-elevation population, an older age at sexual maturity should be observed. Accordingly, large adult body size for P. vlangalii at the high-elevation site might not be caused by the delayed maturation. Delayed maturation at a larger body size is favored in colder environments where juvenile lizards tend to have higher survivorship Shine & Charnov, 1992;Stearns, 1992).
Unfortunately, we were currently unable to determine between-site difference in juvenile survival rate due to limited mark-recapture data.
However, no obvious difference in juvenile return rate (the proportion of recaptured individuals in total marked lizards) was found between the two study sites (low-elevation vs. high-elevation: 22.6% vs. 21.5%), probably implying a similar survival probability in the two populations.
Larger body size can also result from the increased longevity (Morrison et al., 2004;Speakman, 2005). Longevity of lizards from the two study populations (the mean values for estimated age of fieldcaptured adults both based on the von Bertalanffy growth equation and skeletochronology) suggested that on average, adults were older in the low-elevation population than in the high-elevation population.
Individuals in a population that live longer should result in greater mean population age (Leclair & Laurin, 2006). Despite having a larger adult body size, high-elevation individuals did have a shorter longevity than low-elevation ones, which was inconsistent with our final prediction. Therefore, the longevity is unlikely to be an important factor leading to a larger adult body size for the high-elevation population.
No direct correlation between adult body size and longevity was also found in other reptile and amphibian species (Leclair & Laurin, 2006;Oromi, Sanuy, & Sinsch, 2012;Roitberg & Smirina, 2006). For example, the mountain populations of sand lizards (Lacerta agilis) live longer, but have similar adult body size, compared with the lowland populations (Roitberg & Smirina, 2006). Furthermore, environmental oxygen concentration is also considered as a potential factor influencing adult body size of ectothermic animals (Callier & Nijhout, 2014). However, reduced oxygen concentration at the high-elevation site should produce smaller adult body sizes, which is contrary to our results.
In summary, high-elevation P. vlangalii attained a larger adult body size than low-elevation ones, which was primarily due to fast individual growth rates that are likely to be induced by local environmental resource. Our results possibly reflected divergent life history strategies between the low-and the high-elevation populations of P. vlangalii under different environmental conditions. Higher food availability at the high-elevation site allowed lizards to be born at a smaller size, grow faster, and attain a similar or even larger size after sexual maturity. Contrarily, larger neonates were produced at the low-elevation site with less food resources, lived longer, but grew more slowly to a smaller adult size.

ACKNOWLEDGMENTS
The National Natural Science Foundation of China (Grant no. 31670399) and China Scholarship Council supported this study.