SEARCH

SEARCH BY CITATION

Keywords:

  • costs of reproduction;
  • locomotor impairment;
  • Niveoscincus microlepidotus;
  • physiological exhaustion

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Pregnancy is associated with reduced locomotor performance in several reptile species, but the reasons for this reduction remain unclear. Previous authors generally have assumed that the decreased maternal mobility is due to the physical burden of the clutch, but our data on a viviparous Tasmanian scincid lizard (Niveoscincus microlepidotum) suggest a different interpretation. Running speeds of gravid female skinks decrease during gestation (as litter mass increases), but this locomotor impairment is due to physiological changes associated with pregnancy, rather than simple physical burdening. Maternal running speeds are unrelated to litter masses, and do not increase in the week after parturition. Females with very large abdominal fat-bodies (due to ad libitum feeding in the laboratory), equivalent in mass to the litter, nonetheless run rapidly. If the locomotor ‘costs’ of reproduction reflect all-or-none physiological changes associated with pregnancy, then the magnitude of such costs may correlate only weakly with the actual level of reproductive investment. Because life-history models predict that the relationship between fecundity and ‘cost’ has important evolutionary consequences, our results highlight the need to clarify the causal basis for locomotor impairment in gravid reptiles.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Costs of reproduction have attracted considerable scientific interest over the last two decades, because mathematical models predict that such costs should strongly influence the evolution of a species’ life history strategies (e.g. Stearns, 1989; Roff, 1992). However, the actual measurement and interpretation of reproductive costs have proven difficult in the field. For example, individuals in good phenotypic condition tend both to survive well and to have a high reproductive output (e.g. van Noordwijk & de Jong, 1986). Such positive correlations between life history traits have provided a strong incentive for the adoption of experimental techniques to demonstrate reproductive costs. For example, ‘egg swapping’ among nests has often been used to manipulate clutch size and, hence, reproductive effort in birds (reviewed in Bell & Koufopanou, 1986; Lindén & Møller, 1989). In lizards, ablation of ovarian follicles has been used for similar purposes ( Schwartzkopf, 1994; Sinervo, 1994).

Costs of reproduction can take several forms, and be measured in various currencies (e.g. energy, risk). One of the most direct such costs involves locomotor impairment during reproductive activities. For example, decreases in running speeds, manouvrability, and associated locomotor traits have been documented in a variety of taxa ( Cuthill & Houston, 1997). The patterns seem to be particularly strong in reptiles, although not always consistent among taxa. For example, previous studies have demonstrated that in some reptile species, pregnant females run more slowly than nonpregnant ones ( Shine, 1980; Van Damme et al., 1989 ). In others, however, an increase in the clutch mass relative to the female mass does not seem to incur a locomotor impairment ( Brodie, 1989; Sinervo et al., 1991 ).

Locomotor impairment of the female has primarily been attributed to the physical burden that she is carrying (e.g. Cuthill & Houston, 1997). However, this is not the only reason that a pregnant female might run more slowly than a nonpregnant animal. Particularly in viviparous species, the endocrinological and physiological changes associated with gestation may modify other aspects that influence locomotor speed. For example, gravid females might display shifts in muscular strength, metabolic capacity or in motivation to run ( Bauwens & Thoen, 1981).

In principle, it should be relatively easy to test between these two interpretations. Both hypotheses predict that maternal running speeds should decrease during gestation (as litter mass increases), but they generate different predictions about running speeds of postparturient animals. The ‘physical burden’ hypothesis suggests that running speeds should rapidly return to prereproductive levels, whereas the ‘physiological modifications’ hypothesis predicts a slower recovery in speeds. The two hypotheses also make different predictions with respect to the effects of other kinds of physical burden on running speeds. If speed is determined primarily by physiological changes associated with pregnancy, adding an additional burden to the female should not substantially modify her locomotor performance. On the other hand, a burden of the same mass as the litter, carried in approximately the same place as the litter, should reduce maternal speeds to ‘pregnant’ levels if physical burdening is the prime reason for the slower running speeds of gravid animals.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The Tasmanian snow skink, Niveoscincus microlepidotus, is a small (to 5 g) terrestrial lizard that occurs in high-elevation localities throughout the island of Tasmania, south of the Australian mainland. Females reproduce biennially to triennially, and any reproductive female captured in spring (September–October) can thus be allocated to one of two categories: (i) near ovulation, or with recently ovulated eggs in her oviducts (i.e. with more than one year of pregnancy ahead of her), and (ii) near term (i.e. she will give birth within the next month). We took advantage of the simultaneous availability of females of these two different reproductive stages to explore locomotor costs of reproduction.

The lizards were captured by hand or by noose and were brought back to the laboratory during spring in 1993–95. Upon arrival, we weighed each lizard to the nearest 0.1 g, and measured its snout–vent length (SVL), total length and the regenerated part of its tail. To quantify body condition, we used residual scores from the general linear regression between mass and snout–vent length.

Upon arrival at the laboratory, females were assigned to individual cages, 25 × 15 × 10 cm, with fine gravel substrate and an inverted oyster shell to mimic the thermal characteristics of a rock while providing shelter. The cages were then placed in a heated rack to provide each animal with the opportunity to select a temperature between ambient and approximately 35 °C. Mean selected body temperatures of this species average around 30 °C (Jane Melville, personal communication). The cages of pregnant females were checked for neonates at least three times daily, which were counted and separated from their mother. The lizards were fed mealworms (Tenebrio) and watered ad libitum.

Performance trials were conducted on four categories of lizards. In addition to the two categories of females in the population described above (i.e. females with developing follicles and those close to parturition), we also measured sprint speeds of a sample of males and of postpartum females. The near-term females were kept in captivity until they had given birth, whereafter they were re-tested for sprint performance. Thus, most females that gave birth in captivity were run twice, once before and once after parturition. However, in a given trial category, females were only run once.

Immediately before a sprint trial, the lizards were placed in a thermal chamber and heated to 30 °C. The lizards spent 30 min in the chamber, which elevated cloacal temperatures to 30 ± 1 °C before the trials commenced. The trials were performed on a 2-m racetrack with infrared sensors every 0.5 m, and with the accumulated time automatically scored in milliseconds on a digital reader. The average running speed (m s−1) was calculated from the accumulated time taken to travel the entire length of the track (i.e. using the reading of the last infrared sensor only and considering the length of the racetrack). A lizard that stopped running before it completed the 2-m track was encouraged to continue by being tapped on its tail with an artist’s paint brush. Thus, running speed is not independent of motivation to run in this analysis (or in any other analysis we are aware of). However, we incorporate the tail tap score into our analysis and present these results below.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Sprint performance relative to reproductive category

The fastest runners in our experiment were the males, with an average running speed of 0.38 m s−1. Males did not differ in running speed from follicle-developing females, but ran significantly faster than both pre- and postparturient females (Table 1).

Table 1.   Running speeds (m s−1) of males and three categories of females and statistical differences between all categories based on Tukey’s Studentized Range (HSD) test, which controls the type I experimental error rate. Alpha = 0.05, Confidence interval = 0.95, d.f. = 303, Critical value of studentized range = 3.653. Comparisons significant at the 0.05 level are indicated by an asterisk. Thumbnail image of

The only significant difference in running speeds among the categories of female lizards was between follicle-developing females and postparturient females: the postparturient animals ran at 81% the speed of follicle-developing ones (Table 1). Locomotor performance did not shift appreciably at parturition, with pre- and postparturient females running at virtually identical speeds (Table 1).

Determinants of running speed within lizard categories

Running speed was not correlated with any morphological characteristic, such as body mass, SVL or tail length in any of the lizard categories (P > 0.30 for correlations between speeds and all morphological traits, apart for a weak trend for body mass in preparturient females, P = 0.10, r = 0.18, N = 82). In males, running speed was lower in males with regenerated tails (correlation between proportion of the tail that was regrown, vs. speed: r = −0.44, P = 0.02, N = 28). However, this correlation was not significant in any of the female categories (−0.07 < r < 0.16; 0.17 < P < 0.77; 61 < N < 134). Because our analyses primarily concern female traits, we will not consider tail regeneration in subsequent analyses.

The number of times a lizard stopped (and thus, was subsequently tapped with the paintbrush on its tail) was negatively correlated with running speed in all four lizard categories (−0.43 < r < −0.83; P < 0.0001 for all categories). Thus, lizards that were less inclined to keep on running received proportionally more ‘encouragement’ to continue. Hence, any differences in running speed between categories of lizards in the preceding analyses are likely to be conservative. Nevertheless, the inclination to continue vs. interrupt a running trial can be informative per se with respect to a lizard’s endurance and/or escape tactics. We therefore looked for differences in this trait between the different categories. In keeping with overall running speeds, the fastest group (males) received the least tail taps (on average 10.8 per run) and the slowest group (postparturient females) the most taps (22.2 per run on average; Table 2). Again, there was no significant difference in trait means between pre- and postparturient females, but both follicle-developing females and males received significantly less taps than did postparturient females (Table 2). Thus, postparturient females ran slower than any other category in spite of receiving the most tail taps during a run.

Table 2.   Number of taps with a paint brush on a lizard’s tail tip upon interruption of its sprint trial. A number of trials were run without monitoring the number of taps on tails and therefore sample sizes differ somewhat to those given for sprint speeds in Table 1. Statistical differences between groups are tested by Tukey’s studentized range (HSD) test, which controls the experimental error rate. Alpha = 0.05, Critical value = 3.652, d.f. = 323, Confidence intervals = 0.95. Comparisons significant at the 0.05 level are indicated by an asterisk. Thumbnail image of

Might overall differences in speed simply reflect differences in the number of times a lizard stops (was tapped), or, rather, the mean distance that it continues running in response to each tap (ranging from 18.5 cm in males, 10.8 taps/2 m, to 9.0 cm in postparturient females, 22.2 taps/2 m)? Among females, there was a significant difference in motivational response to tail taps between follicle-developing and postparturient females (Table 2). Could this motivational effect explain much of the difference in running speed between female lizards in these two categories? Further analysis shows that this is a contributing factor, but cannot explain the entire difference. A heterogeneity of slopes test with running speed as dependent variable revealed a significant difference in the (negative) slope between running speed and number of tail taps between the two female categories (F = 29.64, P = 0.0001, Dferror = 191). In other words, there was a difference in running speeds between females in the two categories, even after removing the effect of ‘number of taps’. Again, this result suggests that postparturient females are less highly motivated to run.

Unsurprisingly, postparturient females were lighter-bodied than preparturient animals (Table 3). However, follicle-developing females were more heavy-bodied than either of these groups, and males were more heavy-bodied still (Table 3). Thus, there was a completely reversed rank order of body condition vs. running speed among the groups (compare Tables 1 and 3). Therefore, variation in running speeds cannot be attributed to differences in physical burden – that is, in the mass that a lizard carries relative to its body length.

Table 3.   Differences in mean condition index between the four categories of snow skinks used in the sprint trials. The condition indices are the residuals from the mass – SVL regression. Differences between groups were tested statistically using Tukey’s studentized Range test (HSD), which controls the type I experimenal error rate. Alpha = 0.05, Confidence limit = 0.95, d.f. = 321, Critical value of Studentized range = 3.652. Comparisons significant at the 0.05 level are indicated by an asterisk. Thumbnail image of

Post-parturient females thus had the slowest running speed on average, in spite of carrying the lightest burden. However, it remains possible that although physiological aspects of pregnancy reduce running speeds (thus generating the differences among groups), there is also an effect of physical burden on running speeds within the group of reproductive females. If this were the case, we would expect to see correlated changes in body condition and sprint speed, within the subset of females for which we have data on sprint performance both before and after parturition. That is, females that lost the most mass at parturition (i.e. had been carrying a heavier burden) would then increase their running speeds after parturition to a greater extent than other females. For practical reasons, we could not run each female on an exact date prior to and subsequent to parturition. Therefore there was variation in both time to parturition in preparturient females (mean 8.6 ± 5.6 SD) and time since parturition in the postparturient category (mean ± SD 8.0 ± 5.6 days). This variation also allowed us to look for recovery effects from gestation/parturition in the postparturient females.

There were significant individual differences in willingness to run without interruption, as indicated by a significant correlation between a femal’s tail tap scores pre- vs. postparturient (r = 0.57, P = 0.0001, N = 55). That is, some females ran more willingly without interruption both before and after they had given birth. This temporal consistency was also reflected in an almost-significant correlation coefficient between running speeds pre- and postparturient (r = 0.26, P = 0.06, N = 55). However, there was no correlation between the change in condition index associated with parturition and the corresponding change in a femal’s running speed (1st reading – 2nd reading; rs = 0.09, P = 0.52, N = 50). Thus, females that deposited a heavier litter did not thereby regain additional sprint speed, compared with females that produced smaller litters. Furthermore, there was no correlation between a femal’s size-corrected clutch size (residuals from the clutch size – SVL regression) and her running speed before or after parturition (mean clutch size 2.2 ± 0.7; rs = 0.01, P = 0.95; rs = 0.14, P = 0.34, N = 48, respectively) or the difference in her running speeds before vs. after parturition (rs = −0.02, P = 0.89, N = 46).

Post-parturient females not only ran as rapidly (on average) as preparturient animals, but also did not increase in locomotor speed with increasing time since parturition. There was no correlation between days since parturition and running speed (r = −0.09, P = 0.31, N = 135). A femal’s condition index did not correlate with her running speed either before or after parturition (rs = 0.16, P = 0.25, N = 50, and rs = −0.12, P = 0.38, N = 51, respectively).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Our laboratory trials confirm that pregnancy entails a reduction in locomotor performance in Niveoscincus microlepidotum, as it does in many (but not all) other reptiles that have been investigated in this respect (e.g. Shine, 1980; Seigel et al., 1987 ; Van Damme et al., 1989 ). However, the causal basis for this impairment may be more complex, or more diverse, than might appear at first sight.

Previous authors have generally attributed locomotor impairment in gravid animals to the physical burden ( Cuthill & Houston, 1997). Indeed, mathematical models have been developed that specifically incorporate quantitative relationships between relative clutch mass (RCM), running speeds and maternal vulnerability to predation in reptiles ( Vitt & Price, 1982). We do not doubt that direct physical effects may sometimes be important in this respect. The primary evidence for direct ‘physical burden’ effects involves correlations between running speeds and RCM (e.g. Shine, 1980; Seigel et al., 1987 ). Although this evidence can be challenged on the grounds that RCM might also correlate with the degree of maternal physiological modification, a physical effect is intuitively plausible. Experimental work with birds provides more direct evidence for a physical effect: flight performance is impaired by injection of sterile water mimicking the burden of developing eggs ( Jones, 1986).

In the case of Niveoscincus, our data strongly suggest that the significant locomotor impairment of gravid females is not a direct effect of physical burden. The evidence for this conclusion is as follows:

1 running speeds were no higher in postpartum than in prepartum females;

2 running speeds did not decrease with a decreasing time until parturition;

3 running speeds were not correlated with the magnitude of physical burden (i.e. mass relative to body length), either in an overall comparison among groups, or when analysis was restricted to within groups;

4 females that lost a greater burden at parturition did not show any greater increase in locomotor speed after that time.

Several aspects of the Niveoscincus system allow us to obtain a clearer picture of these comparisons than perhaps would be the case for most other species. For example, the prolonged gestation of this species results in seasonal co-occurrence of females at different stages of gestation. Hence, we can compare their locomotor performance without having to allow for confounding effects of seasonal changes in running speeds ( Qualls & Shine, 1998). Also, the high feeding rates of captive females during early pregnancy meant that these animals were actually heavier-bodied than preparturient females (Table 3), a circumstance unlikely to be encountered in most reptiles. The additional mass is laid down as abdominal fat bodies, which (because of their similarity to the developing embryos in mass and position within the body) should have similar physical effects as an increase in litter mass. The high running speeds of these females, despite their heavy burdens, indicate that female Niveoscincus are capable of carrying heavy burdens without substantial locomotor impairment.

Previous studies on running speeds of gravid reptiles provide further grounds for rejecting the notion that lower speeds during pregnancy are simply due to the mass or volume of the litter. In our study, running speeds were reduced during pregnancy, but essentially unaffected by the magnitude of the physical burden. In other studies, pregnancy had diverse effects on maternal locomotor performance. In some cases gravid females were slower, in approximate proportion to the magnitude of the physical burden imposed by the litter (e.g. Shine, 1980). In another case pregnancy did not alter running speeds, but changed the femal’s ‘tactics’ when running (i.e. fewer stops and turns) with the result that gravid females actually traversed a raceway faster than nongravid animals ( Qualls & Shine, 1998). Even within a single species, there may be significant geographical variation in the effects of pregnancy, and its various stages, on locomotor performance (e.g. Sinervo et al., 1991 ). For example, Sinervo et al. also came to the conclusion that exhaustion from gestation contributed to the variation in running speed in their trials. Furthermore, pregnancy reduced running speeds in female skinks in one population, but increased speeds of females in an adjacent population ( Qualls & Shine, 1997).

This kind of diversity is not unexpected, given the extraordinary diversity in reptilian reproduction. For example, it may well be true that simple physical burdening is the primary effect of pregnancy on locomotor performance in many oviparous species, because the eggs are carried only briefly, and do not have intimate physiological connection with the maternal system. In contrast, Niveoscincus has prolonged gestation and intimate maternal–fetal communication ( Stewart & Thompson, 1994). The obvious analogy is with reproduction in humans. For several reasons, such as the hormonal breakdown of pelvic cartilage near birth, pregnant women are likely to be slower runners than nonpregnant women. However, we would not attribute that difference entirely to the physical burden of pregnancy, or expect a postpartum woman to regain optimal locomotor performance as soon as she has given birth. Although we do not understand such endocrinological and physiological modifications associated with pregnancy in lizards, the physiological changes involved may well be profound. In addition, a 15-month gestation period is likely to have considerable exhausting effects on the organism resulting not only in depletion of resources but also result in regression of muscle tissue as females become more sedentary through gestation ( Bauwens & Thoen, 1981; Olsson, 1992).

In conclusion, our experiments demonstrate that female snow skinks suffer costs of reproduction that translate into reduced locomotor performance at the end of the 15-month gestation period. This performance impairment cannot be explained solely by the increased physical burden of the offspring. Instead, reproduction involves physiological modifications to the organism, either via changes to systems important in locomotion, or through general ‘wear and tear’. In either case, the female may take considerable time to recuperate from pregnancy. This conclusion has implications for other areas of reptilian evolutionary ecology. For example, the lower reproductive frequency of viviparous reptiles than of oviparous species has generally been interpreted in terms of the time taken for gestation (e.g. Tinkle & Gibbons, 1977), but plausibly may also reflect these longer-term effects. The notion that such costs may incorporate an important component of ‘physiological exhaustion’ has implications for life-history evolution, and in particular for models that seek to explain the evolution of reproductive investment (e.g. Bull & Shine, 1979; Stearns, 1989). For example, a female is released of a mere ‘back pack’ cost at parturition. Thus, the cost of carrying a burden should primarily be related to risks of predation during a given gestation period. However, costs of ‘wear and tear’ on the organism may influence future reproductive investments and senescence and accumulate through life. Thus, these physiological costs of reproduction may be more directly related to trade-offs between life history parameters and, hence, evolution of life history patterns. Therefore, before we can understand the relationship between reproductive output and costs of reproduction, we need to determine the proximate mechanisms that link these two life history traits.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

M.O. and R.S. thank the Australian Research Council for financial support. M.O. also thanks the Swedish Natural Science Research Council, the Wennegren Foundation, and the Helmut Hertz foundation for financial support, the Wilkes families for their continued help throughout this project, and Hobart City Council for access to the State Park within which the field work was conducted. We are grateful to two anonymous reviewers for insightful comments that improved the quality of a previous draft of this paper.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  • 1
    Bauwens, D. & Thoen, C. 1981. Escape tactics and vulnerability to predation associated with reproduction in the lizard (Lacerta vivipara) . J. Anim. Ecol. 50: 733 743.
  • 2
    Bell, G. & Koufopanou, V. 1986. The costs of reproduction. In: Oxford Surveys in Evolutionary Biology, Vol. 3 (R. Dawkins & M. Ridley, eds), pp. 83–131. Oxford University Press, Oxford.
  • 3
    Brodie, E.D. 1989. Behavioral modification as a means of reducing the cost of reproduction. Am. Nat. 109: 225 238.
  • 4
    Bull, J.J. & Shine, R. 1979. Iteroparous animals that skip opportunities for reproduction. Am. Nat. 114: 296 316.
  • 5
    Cuthill, I. & Houston, A. 1997. Managing time and energy. In: Behavioral Ecology – an Evolutionary Approach (J. R. Krebs & N. B. Davies, eds), pp. 97–120. Blackwell Science, Cambridge.
  • 6
    Jones, G. 1986. Sexual chases in sand martins (Riparia riparia): cues for males to increase their reproductive success . Behav. Ecol. Sociobiol. 19: 179 185.
  • 7
    Lindén, M. & Møller, A.P. 1989. Cost of reproduction and covariation of life history traits in birds. Trend. Ecol. Evol. 4: 367 371.
  • 8
    Nordwijk, A.J. & Van De Jong, G. 1986. Acquisition and allocation of resources: their influence on variation in life history tactics. Am. Nat. 128: 127 142.
  • 9
    Olsson, M. 1992. Sexual selection and reproductive strategies in the sand lizard, Lacerta agilis. PhD-thesis, The University of Gothenburg, Sweden.
  • 10
    Qualls, C.P. & Shine, R. 1998. Costs of reproduction in conspecific oviparous and viviparous lizards, Lerista bougainvillii. Oikos 82: 539 551.
  • 11
    Qualls, F.J. & Shine, R. 1997. Geographic variation in ‘costs of reproduction’ in the scincid lizard Lampropholis guichenoti. Funct. Ecol. 11: 757 763.
  • 12
    Roff, D. 1992. The Evolution of Life Histories – Theory and Analysis. Chapman & Hall, London.
  • 13
    Schwartzkopf, L. 1994. Measuring trade-offs: a review of studies of costs of reproduction in lizards. In: Lizard Ecology – Historical and Experimental Perspectives (L. J. Vitt & E. R. Pianka, eds), pp. 7–30. Princeton University Press, New Jersey.
  • 14
    Seigel, R.A.., Huggins, M.M., Ford, N.B. 1987. Reduction in locomotor ability as a cost of reproduction in gravid snakes. Oecologia 73: 481 485.
  • 15
    Shine, R. 1980. ‘Costs’ of reproduction in reptiles. Oecologia 46: 92 100.
  • 16
    Sinervo, B. 1994. Experimental tests of reproductive allocation paradigms. In: Lizard Ecology – Historical and Experimental Perspectives (L. J. Vitt & E. R. Pianka, eds), pp. 73–90. Princeton University Press, New Jersey.
  • 17
    Sinervo, B.., Hedges, R., Adolph, S.C. 1991. Decreased sprint speed as a cost of reproduction in the lizard Sceloporus occidentalis: variation among populations . J. Exp. Biol. 155: 323 336.
  • 18
    Stearns, S.C. 1989. Trade-offs in life-history evolution. Funct. Ecol. 3: 259 268.
  • 19
    Stewart, J.R. & Thompson, M.B. 1994. Placental structure of the Australian lizard, Niveoscincus metallicus. J. Morphol. 220: 223 236.
  • 20
    Tinkle, D.W. & Gibbons, J.W. 1977. The distribution and evolution of viviparity in reptiles. Misc. Publ. Mus. Zool., Univ. Michigan 154: 1 55.
  • 21
    Van Damme, R.., Bauwens, D., Verheyen, R.F. 1989. Effect of relative clutch mass on sprint speed in the lizard Lacerta vivipara. J. Herp. 23: 459 461.
  • 22
    Vitt, L.J. & Price, H.J. 1982. Ecological and evolutionary determinants of relative clutch mass in lizards. Herpetologica 38: 237 255.