Decreased Sprint Speed as A Cost of Reproduction in the Lizard Sceloporus Occidentals: Variation Among Populations

Reproduction is thought to entail costs: genetic trade-offs between current versus future allocation to reproduction (see Reznick, 1985, for a review). For lizards, authors frequently use relative clutch mass (RCM; clutch mass divided by total mass of the gravid female) as an operational measure of current reproductive investment, and suggest that an increase in RCM carries associated costs of reproduction (Tinkle and Hadley, 1975; Vitt and Congdon, 1978; Shine, 1980). One potentially important cost may result from the reduced mobility of a gravid female: slower lizards may be more susceptible to predation (Vitt and Congdon, 1978; Shine, 1980, 1988; Bauwens and Thoen, 1981; Vitt and Price, 1982).

The ecological correlates of variation in RCM among species of lizards are well documented. Lizards that are active, wide-ranging foragers, or are highly arboreal, tend to have lower RCMs than those that are sedentary or are sit-and- wait predators (Vitt and Congdon, 1978; Vitt, 1981; Huey and Pianka, 1981; Vitt and Price, 1982; Magnusson et al. 1985; Anan’eva and Shammakov, 1985). This suggests that there are functional trade-offs between RCM and mobility, such that selection for increased RCM can indirectly select for decreased mobility. However, only a few studies have measured the extent to which mobility is impaired in gravid females, and whether decrements in locomotor ability are correlated with the level of reproductive investment (Shine, 1980; Bauwens and Thoen, 1981).

In this study, we measure how reproduction affects sprint speeds of females from four populations of the western fence lizard, Sceloporus occidentalis. We also investigate whether the decrement in sprint speed of a gravid female is correlated with her level of reproductive investment or with other morphological traits. In addition, we describe the time course of recovery of sprint speed following oviposition.

Because the four populations occur at different latitudes and altitudes, they may experience different predation intensities (Pianka, 1970; Tinkle and Ballinger, 1972). Increased predation pressure might cause natural selection for higher sprint speeds, possibly through decreasing reproductive investment. Therefore, we investigated whether reproductive investment and degree of impairment of sprint speed covaried among populations as well as within populations.

Because of methodological problems associated with using RCM (body mass is found in both the numerator and denominator and the allometry between clutch mass and body mass would confound any comparison of animals that differ in size; Packard and Boardman, 1987; Dunham et al. 1988), we have used multivariate methods to analyze the relationships between decreased sprint performance and the burden of a clutch of eggs. Morphological traits that potentially affect performance can be included in these analyses.

Gravid female Sceloporus occidentalis were collected in May and June 1987 and 1988 from four populations. Two populations are in the San Gabriel Mountains of southern California: one at low elevation (1300 m) on the edge of the Mojave Desert near Pearblossom and the other at high elevation (2200 m) on Table Mountain near Wrightwood. The third population (700 m) is in Oregon near Terrebonne. The fourth population (200 m) is in Washington near Lyle. Further aspects of the ecology of S. occidentalis at these study sites are given by Tsuji (1986), Sinervo (1988), and Adolph (1990).

Gravid females were transported to the laboratory (University of Washington) within 4 days of capture. The females were housed individually in plastic terraria with a moist substratum of sand and peat moss. The terraria were kept in an environmental chamber (12 h at 34°C, 12 h at 20°C, 12 h:12h L:D with fullspectrum illumination). Females were fed crickets [dusted with vitamins (Vionate™) and calcium] every day, and mealworms biweekly. Females were weighed once a week. Terraria were checked twice a day for eggs. Eggs were removed and weighed to determine clutch mass, and each female was weighed to determine her post-oviposition mass. After ovipositing, females were housed in small groups in large terraria (radiant heat, 12 h:12h L:D, supplemented with ultraviolet light), and given food and water ad libitum.

Gravid females were raced initially within 2 weeks of capture to determine their sprint speed when burdened by a clutch of eggs. On the day of racing the gravid females were held in an environmental chamber at 34°C (the thermal optimum for sprinting in this species; Adolph, 1987). Gravid females were raced on a level 2.4m×20cm racetrack to estimate maximum sprint speed. The racetrack had a rough, rubberized substratum that provided excellent traction. Speeds were determined electronically by regularly spaced photocells connected to a computer (Huey et al. 1981). Maximum sprint speed was estimated as the fastest 0.5 m interval achieved during four consecutive races held at 1-h intervals. Females were raced again between 10 and 30 days after they had oviposited. The change in sprint speed for each individual (before vs after ovipositing) was analyzed using repeated measures analysis of variance (ANOVA), with population as a factor. Further details concerning gravid female rearing and oviposition can be found in Sinervo (1990).

To determine whether sprint performance declines prior to ovipositing, some gravid females were raced twice before ovipositing, once around 14 days before ovipositing and again immediately prior to ovipositing (about 3 days). Similarly, to determine whether sprint performance increases after ovipositing, another group of females was raced shortly after ovipositing (about 7 days) and re-raced about 14 days after ovipositing. We compared the change in sprint performance in these two groups (paired t-tests). We raced an additional group of females (from the California populations) 1–40 days after they had oviposited (these females had not been raced before they oviposited). This group, which was only raced once, controls for the effect of experience on sprint speed (i.e. training that may occur between successive races by the same individual). We estimated the time course of recovery of sprint speed in this group using regression analysis.

We defined burden as the difference in the mass of each female before and after she oviposited (burden was highly correlated with the total mass of eggs in her clutch). As a measure of reproductive investment we used residuals from the regression of burden against body size, rather than using the ratio of burden to body mass (i.e. RCM). We compared reproductive investment (total clutch mass) among populations by analysis of covariance (ANCOVA) [using post-oviposition mass and snout-vent length (SVL) as covariates in two separate analyses]. We report population differences in total clutch mass (ANCOVA, using mass and SVL) expressed in terms of a difference between low-elevation California females and the other three populations [i.e. dummy variables describing difference between Washington, Oregon and high-elevation California relative to low-elevation California (Draper and Smith, 1981)]. We also measured thigh length (TL, distance from knee to knee when both femurs are held laterally and perpendicular to the body) as another possible morphological correlate of sprint speed (Garland, 1985; Losos and Sinervo, 1989; Sinervo and Huey, 1990). Unless otherwise noted, all variables were log-transformed.

Reproductive investment and morphological traits could affect sprint performance either independently or interactively. Comparing these reproductive and morphological traits among populations that differ in performance can suggest possible functional relationships among these traits, although traits could covary among populations for other reasons. Covariation that is due to functional relationships should be present within populations as well as among populations (Bennett, 1987). Because many physiological and morphological traits are correlated with body size (e.g. Calder, 1984; Schmidt-Nielsen, 1984; Sinervo and Huey, 1990), and hence could covary with one another spuriously, a proper analysis should factor out body size (Bennett, 1987).

We analyzed the relationship between the reproductive investment of a female and her sprint performance after correcting for three different measures of body size and morphology: post-oviposition body mass, SVL and TL. Burden was regressed against each of these measures individually; residuals about the regression were used as variates in subsequent regression analyses of performance. On functional grounds, one would expect a positive correlation between the burden residuals and the magnitude of a female’s decrement in sprint speed. For example, we might expect a female with a relatively large burden for her body mass (large residual) to experience a relatively large decrement in sprint speed when gravid.

Females from all three populations were slower when gravid than when raced at least 10 days after ovipositing (Fig. 1, Table 1). Sprint speed after ovipositing did not vary among populations, but the females from Oregon and Washington were slower while gravid than were females from either California population (Fig. 1, Table 1). The mean decrement in sprint speed while gravid was 1.16 ms⁻¹ for Washington females, 0.76 ms⁻¹ for Oregon females, 0.48 ms⁻¹ for females from high elevation in California and 0.53 ms⁻¹ for females from low elevation in California.

Sprint performance measured on the same individual on two occasions prior to ovipositing was not significantly lower on the second trial (immediately prior to ovipositing; mean sprint performance decrease=–0.10ms⁻¹; paired t-test, t= –1.04; N=11; P>0.32; Fig. 2A). Similarly, sprint performance of individuals from California was not correlated with time before ovipositing (Fig. 2B). Thus, the reduced sprint speed prior to ovipositing probably existed well before females were brought into the laboratory (i.e. more than 3 weeks prior to ovipositing).

However, sprint speed did increase with time after ovipositing (Fig. 2). Sprint performance increased significantly in those females (from high elevation in California) that were raced on two different dates after ovipositing (mean sprint performance increase=0.29ms⁻¹; paired t-test, t=2.79; N=16; P<0.02; Fig. 2A). Similarly, the sprint performance of individual females from California (raced for the first time) increased significantly with time after ovipositing (Fig. 2B). Within 2–3 weeks after ovipositing, their sprint speeds were comparable to the sprint speeds of males and females measured well outside the reproductive season (Fig. 2B). These increases in performance suggest that females do not recover maximum sprint performance until about 2 weeks after ovipositing.

Gravid females from Washington had significantly greater reproductive investment (burden, corrected for body mass differences) than females from the Oregon and California populations (ANCOVA, Fig. 3A). This difference is also reflected by a greater RCM (clutch mass divided by female body mass including clutch mass) for Washington lizards [0.318 (mean) ±0.005 (S.E.)] compared to lizards from Oregon (0.281 ±0.010), high-elevation California (0.270±0.079) and low-elevation California (0.277±0.077).

When corrected for SVL, burden was significantly greater in both Washington and Oregon females than in California females (ANCOVA, Fig. 3B). Thus, females from northern populations tend to carry more egg mass per unit body mass and/or per unit body length. In addition, the northern lizards had shorter legs (ANCOVA, Fig. 3C).

Populations of Sceloporus occidentalis differed in reproductive investment and sprint performance while gravid. This suggests a possible functional relationship between these traits; if so, these traits should covary among individuals within each population (Bennett, 1987). We were particularly interested in the relationship between the sprint speed decrement and the burden carried per unit of body mass and per unit of body length (SVL). We had sufficient data for females from Oregon and the two California populations to test for statistical relationships that might reflect functional relationships.

To determine whether burden might affect sprint performance, we computed residuals from the regression of burden on either measure of body size (performed separately for each population and for each measure), yielding burden residuals for each individual female. Surprisingly, we did not find a significant correlation between burden residuals (factoring out body mass) and sprint speed decrement within any population (California low, r=0.07, P>0.69, N=36; California high, r=0.15, P>0.44, N=2T, Oregon, r=0.195, P>0.37,N=22). Similarly, sprint speed decrement and burden residuals (factoring out SVL) were not significantly correlated within either of the California populations (California low, r=0.10, P>0.57, N=36; California high, r=0.12, P>0.54, N=27; Fig. 4B,C). However, in Oregon females, SVL-corrected burden was positively correlated with sprint speed decrement: females with a relatively large burden for their length experienced a greater decline in sprint speed when gravid (Fig. 4A, r= – 0.56, P=0.006, N=22).

Sprint speeds of female fence lizards while gravid were consistently lower than after they had oviposited. Sprint speeds were reduced by about 20% in gravid females from both California populations, by about 30% in gravid females from Oregon, and by about 45 % in gravid females from Washington (Fig. 1). Similarly, running speeds are reduced by 20 –30 % in several species of gravid skinks (Shine, 1980) and by 26% in gravid Lacerta vivipara (Bauwens and Thoen, 1981). Whereas previous studies determined the decrement in sprint speed of gravid females by comparing them to either non-gravid females or males, our study compared the change in sprint speed of the same individuals before and after ovipositing (Table 1). A similar study comparing the speed of gravid snakes before and after reproduction found a 20 % reduction in locomotor ability (Seigel et al. 1987).

Sprint speeds could decline because the weight of the clutch hampers locomotion (Shine, 1980; Bauwens and Thoen, 1981; Vitt and Price, 1982). On a finer scale, variation in reproductive investment among females within a population could covary with the degree of locomotor impairment. Indeed, sprint speed decreases with RCM in two species of scincid lizards (Shine, 1980). In our study, we did not find a significant correlation between the level of reproductive investment (relative to body mass) and sprint speed decrement among individuals within a population. However, the effect of reproductive investment was statistically significant in one of our populations (Oregon) if morphology -burden carried per unit of body length - was incorporated in the analysis (see below). This morphological effect could explain the larger sprint speed decrements experienced by the stockier northern lizards.

An alternative cause of the decline in sprint speed could be a temporary deterioration of the body condition of gravid females. Indeed, our results show that the sprint speed of a female immediately after ovipositing is as low as when she is burdened, and that she requires several weeks to recover her maximum ability. This suggests that reproduction not only impairs locomotion because eggs are a physical burden but that it also exacts a physiological toll on females. The poor body condition of females immediately after ovipositing supports this view (B. Sinervo and S. C. Adolph, personal observation in the laboratory and field). Thus, the disadvantage of decreased sprint speed does not end at oviposition, but persists for several more weeks.

Regardless of its causal mechanism(s), the reduced sprint speed is potentially an important cost of reproduction, particularly if it increases the risk of predation (Shine, 1980; Bauwens and Thoen, 1981; Christian and Tracy, 1981). Indeed, Shine (1980) found in laboratory trials that gravid female skinks (Leiolopisma Coventry!) were more vulnerable to predation by a snake than were non-gravid individuals (males). In addition to the potential risk of greater predation, reduced sprint speed could entail other costs for gravid female lizards. For example, foraging success (Avery et al. 1982) may depend in part on sprint speed.

Gravid lizards might be expected to compensate behaviourally for their reduced mobility. For example, gravid female Lacerta vivipara rely more on camouflage and less on flight than do males or non-reproductive females (Bauwens and Thoen, 1981). Similarly, unpublished field observations (B. Sinervo and S. C. Adolph) suggest that, during the reproductive season, gravid female S. occidentalis choose perches that are less conspicuous, and closer to shelter, than do males. Outside the reproductive season, however, males and females do not differ in microhabitat use (Adolph, 1990).

Washington females were both much slower when gravid and had greater reproductive investment (burden versus post-laying mass) compared to females from other populations. This suggests a functional trade-off between these two traits. However, differences in reproductive investment are clearly not the only cause of differences in sprint speed decrement among populations of S. occidentalis. For example, gravid females from Oregon were also slower than those from California, yet they carried the same clutch mass for their body mass.

Gravid Oregon and Washington females may be slower because they have a shorter, stockier frame (Figs 3B, 4). Indeed, we found a significant correlation between the sprint speed decrement and burden per unit body length (SVL) in females from Oregon. Quadrupedal locomotion in lizards involves both horizontal flexion of the vertebral column and rotation of the pelvis (Snyder, 1952,1962), and the presence of a clutch of eggs may interfere with this motion. Similarly, other researchers have suggested that the presence of food (Ford and Shuttlesworth, 1986) and eggs or developing young (Shine, 1988) may interfere with lateral undulation during locomotion in snakes. In S. occidentalis, this effect could be greater in Oregon and Washington females because of their shorter bodies. The effect on sprinting performance might be further exacerbated in Washington females because of their increased level of reproductive investment.

In addition to their shorter bodies, females from the more northern populations have shorter thighs (relative to SVL) compared to the faster California females (Fig. 3C). This difference could contribute to the greater sprint speed decrement experienced by gravid northern females (Fig. 1). However, we found no significant correlation between thigh length and sprint performance within each population. Thus, the possible functional relationship between thigh length and sprint performance while gravid remains unclear.

Several workers have suggested that predation pressure decreases with increasing latitude (Pianka, 1970; Tinkle and Ballinger, 1972). If so, we would expect to see more highly evolved anti-predator defenses in California populations of 5. occidentalis, compared to Oregon and Washington populations. For example, higher predation might select for faster overall sprint speeds or for lower levels of reproductive investment. We found some evidence of such a difference in the present study: the higher sprint speeds of gravid California lizards could be interpreted as an adaptation to greater predation pressure. Conversely, the larger decrement in sprinting performance of gravid females from northern populations could reflect relaxed selection ón performance which, in the case of Washington lizards, may also have permitted the evolution of increased reproductive investment and correlated changes in morphology.

Indeed, the ecological correlates of RCM among species include evolved changes in morphology. Lizard species that are active, wide-ranging foragers, or are highly arboreal, tend to have lower RCMs than those that are sedentary or sit- and-wait predators; these differences are associated with changes in body shape and limb proportions (Pianka and Parker, 1975; Vitt and Congdon, 1978; Vitt and Price, 1982). In the present study, differences in morphology and reproductive investment among northern and southern populations of S. occidentalis are also associated with the degree of arboreality: fence lizards in California are highly arboreal (especially at low elevation; Adolph, 1990) compared with lizards from Oregon and Washington (J. S. Tsuji, personal communication; Sinervo and Losos, 1991).

Our study documents variation in reproduction and morphology that may be responsible for differences in sprint performance of gravid females among populations of a single species. Interestingly, among-population variation in reproductive investment did not fully explain the differences in sprint performance of gravid females: the interaction between morphology and reproductive investment appears to be important as well. The results from our among-population comparison of a single species imply that differences in RCM among species may not be the only cause of performance differences but that morphological evolution is also likely to play a strong role in shaping the association between reproductive investment and its ecological correlates in squamates (also see Shine, 1988). Finally, we advocate a multivariate approach to the analysis of reproductive investment within and among species (see also Dunham et al. 1988). Multivariate analysis of morphology can lead to useful insights into the biomechanical basis of decrements in performance associated with reproduction.

We would like to thank R. B. Huey, J. Kingsolver, R. R. Strathmann and T. Daniel for helpful comments and suggestions and R. B. Huey, J. S. Tsuji, and F. H. van Berkum for sharing their egg data from the Washington population. Special thanks go to R. B. Huey for providing laboratory facilities and supplies. We would also like to thank B. Ruud, P. Doughty and C. Peterson for providing assistance with gravid female care and R. B. Huey and J. Tsuji for collecting the gravid females from Washington. BS was supported by the Miller Institute for Basic Research in Science, University of California, Berkeley. This research was supported by NSF grants BSR 78-12024, DEB 81-09667 and BSR 84-15855 awarded to R. B. Huey.