Cold climates and the evolution of viviparity. produce poor-quality offspring in the lizard, in reptiles: cold incubation temperatures

Transcription

1 BiologicalJoumal of the Linriean Socieiv (l999), 67: With 4 figures Article ID: bijl , available online at on ID E bl 8 c Cold climates and the evolution of viviparity in reptiles: cold incubation temperatures produce poor-quality offspring in the lizard, Sceloporus virgatus CARL P. QUALLS" AND ROBIN M. ANDREWS Department of Bioloa, Virginia Pohtechnic Institute and State Universip, Blacksburg, VA 24061, U.S.A. Received 29 June 1998; accepted for publication 20 December 1998 Evolutionary origins of viviparity among the squamate reptiles are strongly associated with cold climates, and cold environmental tempcraturcs are thought to be an important selective force behind the transition from egg-laying to live-bearing. In particular, the low nest temperatures associated with cold climate habitats are thought to be detrimental to the developing embryos or hatchlings of oviparous squamates, providing a selective advantage for the retention of developing eggs in utero, where the mother can provide warmer incubation temperatures for her eggs (by actively thermoregulating) than they would experience in a nest. However, it is not entirely clear what detrimental effects cold incubation temperatures may have on eggs and hatchlings, and what role these effects may play in favouring the evolution of viviparity. Previous workers have suggested that viviparity may be favoured in cold climates because cold incubation temperatures slow embryogenesis and delay hatching of the eggs, or because cold nest temperatures are lethal to developing esgs and reduce hatching success. However, incubation temperature has also been shown to have other, potentially long-term, effects on hatchling phenotypes, suggesting that cold climates may favour viviparity because cold incubation temperatures produce offspring of poor quality or low fitness. We experimentally incubated eggs of the oviparous phrynosomatid lizard, Sceloporus vilgatus, at temperatures simulating nests in a warm (low elevation) habitat, as is typical for this species, and nests in a colder (high elevation) habitat, to determine the effects of cold incubation temperatures on embryonic development and hatchling phenotypes. Incubation at cold nest temperatures slowed embryonic development and reduced hatching success, but also affected many aspects of the hatchlings' phenotypes. Overall, the directions of these plastic responses indicated that cold-incubated hatchlings did indeed exhibit poorer quality phenotypes; they were smaller at hatching (in body length) and at 20 days of age (in length and mass), grew more slowly (in length and mass), had lower survival rates, and showed greater fluctuating asymmetry than their conspecifics that were incubated at warmer temperatures. Our findings suggest that cold nest temperatures are detrimental to S. vilgatus, by delaying hatching of their eggs, reducing their hatching success, and by producing poorer quality offspring. These negative effects would likely provide a selective advantage for any mechanism through which these lizards could maintain warmer incubation temperatures in cold climates, including the evolution of prolonged egg retention and viviparity The Liiineari Society of Lolidoil * Corresponding author. Present address: Ohio State University/Brown Treesnake Project, P.O. Box 8255, MOU-3, Dededo, Guam , U.S.A. qualls. 14@osu.edu /99/ $30.00/ The Linnean Society of London

2 fluctuating morphology growth survival fitness 354 C. 1. QUALLS AND R. M. ANDREWS ADDITIONAL KEY WORDS:-embryonic development - stress - phenotypic plasticity ~ developmental instability ~ asymmetry ~ ~ ~ ~. CONTENTS Introduction Detrimental effects of cold nest temperatures Mechanisms for coping with cold nest temperatures..... Hypotheses and experimental design Material and methods Collection and husbandry Incubation treatments Hatchling housing conditions Incubation period. morphology. and growth rate Fluctuating asymmetry I hermoregulatory behaviour Escape behaviour Data manipulation and analysis Results Embryo developmental ratc Hatching snccess and hatchling survival Body size Growth rate Activity levels and thermoregulatory behaviour Levels of fluctuating asymmetry Body shape Escape behaviour and performance Discussion Effects of cold incubation temperatures Coping with or avoiding the effects of cold incubation temperatures Conclusion Acknowledgements References INTRODUCTION Most reptile species are oviparous (egg.laying). including all turtles and crocodilians. but viviparity (live birth) has evolved independently within many different lineages of snakes and lizards (Blackburn ; Shine. 1985). The frequency with which viviparity has airsen within the Squamata has led to much research and speculation as to the adaptive nature of this major life history shift. The transition from oviparous to viviparous reproduction is thought to occur through a gradual increase in the length of time that eggs are retained within the mother s oviducts prior to oviposition. proceeding through intermediate stages of oviparity with prolonged egg retention. and culminating in the retention of developing eggs in utero throughout embryogenesis and the birth of fully developed offspring (Packard. Tracy & Roth. 1977; Shine & Bull. 1979). Comparative evidence reveals that most origins of reptilian viviparity are associated with cold climates. implicating cold environmental temperatures as a probable selective force behind the transition from egg-laying to live-bearing (Blackburn. 1982; Shine ). In particular. the negative effects of cold nest temperatures on developing eggs. and/or the benefits of warmer incubation temperatures to eggs retained in utero. are considered the primary mechanism through which cold climates favour the evolution of viviparity.

3 INCUBATION TEhIPERATURE AND OFFSPRING QUALITY 355 Cold incubation temperatures are thought to be detrimental because they slow embryonic development or because they increase egg mortality, and prolonged retention of eggs in utero is considered beneficial because females can provide warmer incubation temperatures (through their own thermoregulation) than those in a nest, thereby accelerating the development of their offspring or protecting their eggs from lethally low temperatures (Packard et al., 1977; Tinkle & Gibbons, 1977; Shine & Bull, 1979; Shine, 1983). Comparative and phylogenetic analyses suggest a strong association between cold climates and origins of squamate viviparity (see Blackburn, 1982; Shine, 1985; Heulin et al., 1993; Qualls & Shine, 1998b), and hence, the basic premise of the cold climate model (i.e. that cold climates favour the evolution of viviparity) has become widely accepted. However, the model also postulates specific mechanisms through which cold temperatures would favour the transition from oviparity to viviparity. Specifically, it posits: ( 1) that slowed embryonic development (and hence, delayed hatching) or increased egg mortality are the primary detrimental effects of cold incubation temperatures, and (2) that prolonging egg retention is the primary mechanism by which oviparous squamates can avoid these negative consequences of cold nest temperatures. These aspects of the cold climate model have received relatively little empirical scrutiny (but see Shine, 1983; Heulin, Osenegg & Lebouvier, ; Qualls & Shine, 1996), yet most recent discussions of the evolution of viviparity (including our own) implicitly assume them to be true, by ignoring other plausible alternatives. The purpose of this study is to explicitly evaluate these two components of the cold climate model, along with other plausible scenarios that are intuitively reasonable, yet have not received much attention. Detrimental efects of cold nest temperatures With respect to the first assumption (1 above), we suggest that cold climates may be detrimental to oviparous reproduction for reasons other than slowed embryonic development and/or increased egg mortality, and that these other effects can also play an important, and perhaps greater, role in favouring the evolution of viviparity. Among snakes and lizards, the temperatures that eggs experience during incubation can affect many phenotypic traits of the developing embryos and the resulting hatchlings, including morphology, behaviour, growth rate, survival, locomotor performance, and thermoregulatory preferences (e.g. Gutzke & Packard, 1987; Phillips et al., 1990; Burger, 1991; Van Damme et al., 1992; Shine & Harlow, 1993; Elphick & Shine, 1998; Qualls & Shine, 1998~). Thus, oviparous reproduction may not be well suited to cold climates because cold nest temperatures produce hatchlings with poor quality phenotypes (Shine, 1995; Qualls & Shine, 1996). Cold nest temperatures could affect phenotype quality through two different (though not mutually exclusive) mechanisms, one direct and one more subtle. Under the most direct mechanism, reaction norms for plastic traits may be such that cold incubation temperatures produce phenotypes with particular attributes that are associated with low fitness (such as small body size, poor body condition, slow growth rates, etc.; see Shine, 1995; Qualls & Shine, 1996). Alternatively, incubation at low temperatures may impair overall phenotype quality in a more subtle manner, by adversely affecting the fidelity of the developmental process. We know that embryogenesis is not a perfect process, that developmental errors occur, and stress

4 356 C. P. QUAI,I,S AND R. M. ANDREWS during development can cause developmental instability (Parsons, 1990; Polak & Trivers, 1994; Mdler, 1997). Fluctuating asymmetry (minor, random deviations from perfect left-right symmetry) is thought to be the result of, and hence, an indicator of developmental instability. Embryonic development of most animals occurs in a bilaterally symmetrical manner, and it is generally assumed that the goal of an organism s developmental program is a perfectly symmetrical body. If so, random departures from perfect left-right symmetry should reflect the occurrence of developmental errors (i.e. developmental instability), and thus, the degree of fluctuating asymmetry should be inversely related to developmental stability (Palmer & Strobeck, 1986; Polak & Trivers, 1994). Further, developmental instability is, generally, negatively related to fitness components such as growth, fecundity, and survival (Mdler, 1997). Thus, if cold incubation temperatures are more stressful than warmer temperatures for developing reptile embryos, cold nest temperatures (in cold climate habitats) could cause developmental instability, which is associated with low fitness. Mechanismsfor coping with cold nest temperatures Thus, the low nest temperatures associated with cold climates can have unfavourable consequences for oviparous reproduction, because of a mismatch between the incubation temperatures that eggs experience in nests and the thermal optima for embryonic development. If so, when they encounter cold environments, what evolutionary options do squamate reptiles have for overcoming these problems? In general terms, there are only two solutions; incubation temperatures must be modified to better suit the developmental physiology of embryos, or embryo physiology must be modified to better suit colder incubation temperatures. The cold climate model for the evolution of viviparity assumes the former solution, that eggs are retained in utero for longer periods of time, during which the mother s thermoregulatory activity can provide incubation temperatures that are warmer than those in a nest, and are more suitable for embryogenesis. However, there is no reason to assume, a priori, that the latter solution, modifying embryo physiology to better suit cold nest temperatures, is not also possible. Physiological and developmental processes are heritable traits that are also subject to evolutionary forces (Arnold, 1987; Garland & Carter, 1994), and thus, the reaction norms of embryonic development can be modified to achieve more optimal phenotypes when eggs experience cold nest temperatures. Hypotheses and experimental design Thus, we can explicitly formulate alternative or complementary hypotheses to two main components of the cold climate model for the evolution of viviparity: (1) in addition to slowing embryonic development and/or increasing egg mortality, cold incubation temperatures can be detrimental to oviparous squamates (and hence, favour the evolution of viviparity) because they produce hatchlings with poor quality phenotypes; and (2) rather than adapting incubation temperatures to better suit embryo physiology (i.e. by retaining their eggs), oviparous squamates can adapt to life in cold climates by altering their embryo physiology to better suit cold incubation

5 INCUBATION 'I'EhIPEKATURE AND OFFSPRING QUALITY 3.57 temperatures. To evaluate these hypotheses, we need more detailed knowledge of how cold incubation temperatures affect embryos and hatchlings of oviparous squamates, and what mechanisms (if any) extant oviparous taxa from cold climates employ to cope with these effects. We conducted an experimental incubation study on the phrynosomatid lizard, Scelopoms virgatus, to provide some of these missing details. This species is oviparous and occurs in montane habitats of southern Arizona and New Mexico (U.S.A.) and of adjacent Mexico (Behler & King, 1979). Populations of this species are mostly restricted to relatively warm climates in the lower elevations of these mountains, but isolated populations also occur in colder climates at higher elevations. The occurrence of S. virgatus populations in both relatively low elevation and high elevation habitats (hereafter referred to as LE and HE populations, respectively) provides a valuable opportunity to determine what effects cold incubation temperatures have on embryos and hatchlings, and to assess whether embryo developmental physiology is adapted to local environmental (i.e. nest temperature) conditions. We experimentally incubated eggs from both the HE and LE populations of Sceloporus virgatus at biologically realistic temperatures simulating nest conditions in both the low elevation (warm) and high elevation (cold) habitats. Eggs from each population were incubated under both temperature conditions, in a two factor hierarchical design (population of origin by incubation temperature treatment). We recorded the incubation periods and hatching success of the eggs, and also the size, shape, growth rates, thermoregulatory behaviour, activity levels, escape behaviour, levels of fluctuating asymmetry, and survival of the hatchlings. This experimental design allowed us to make comparisons of eggs and hatchlings between the two incubation treatments, and between the two populations, to address two main questions: (1) What are the effects of cold incubation temperature on embryonic development and hatchling phenotypes of S. virgatus? Are slowed development and/or increased egg mortality of primary importance, or could more subtle effects on phenotype quality be more important for the evolution of viviparity? (2) What mechanisms, if any, do HE S. virgatus exhibit that may allow them to better cope with cold nest temperatures? Do the reaction norms for embryonic development and hatchling phenotype differ between the two populations (i.e. population x treatment or genotype x environment interaction), such that HE lizards appear physiologically adapted to incubation at cold nest temperatures? ILIA'TERIAL AND hiethods Collection and husbandy Gravid female Scelopoms virgatus were collected between 23 June and 4 July 1995, from two populations in the Chiricahua Mountains of southeastern Arizona. The two populations are situated at a low elevation (1 800 m, N= 29 females) and a high elevation (2400 m, N= 19 females), near the American Museum of Natural History's Southwestern Research Station (SWRS). After capture, the lizards were taken to SWRS, where they were housed in glass terraria, in an open-air animal holding facility. Each terrarium contained a substrate of soil, with rocks or tree bark provided as cover. During daylight hours, thermoregulatory opportunities were provided by a

6 358 C. P. QUALLS AND R. R.1. ANDREWS 60 w incandescent lamp suspended over one end of each cage; overnight temperatures followed outdoor ambient temperatures. Food (crickets) and fresh water were provided daily. Eggs were obtained from the S. uirgatus at SWRS, after which the females were released at their site of capture. On 9 July 1995, oviposition was induced in all lizards, by intraperitoneal injection of oxytocin. Unlike many other reptiles, S. uirgatus do not oviposit when their embryos reach any particular developmental stage(s); rather, they are able to facultatively prolong egg retention, allowing them to delay oviposition until the summer monsoonal rains begin (Rose, 1981; Andrews & Rose, 1994). In the Chiricahua Mountains, these rains normally begin in early July; and because of their geographic proximity, both populations probably receive their first heavy rains from the same storm system, at or near the same time. Thus, we believe that our protocol of inducing oviposition on 9 July, and at the same time for lizards from both populations, was biologically realistic. During oviposition, the soil in the cages was kept moist to prevent the eggs from desiccating, and all eggs were removed to containers of moist vermiculite immediately after they were laid. Any lizards that did not lay their entire clutch of eggs were injected again on the following day. We recorded the number of eggs in each female s clutch and all eggs were packed in plastic containers filled with moist vermiculite, which were kept in an air-conditioned room until 13 July, when they were transported to the laboratory at Virginia Tech (Blacksburg, Virginia). Once in the laboratory, one or two eggs from each clutch were sampled to determine the developmental stages of the embryos and the dry masses of the eggs. The sampled eggs were dissected to determine the developmental stages of the embryos (following Dufaure & Hubert, 196 1); where embryo traits were intermediate between two defined stages, half-stage designations were assigned. After staging, the entire eggs (including the shells) were dried to a constant mass at 50 C and reweighed to determine their dry masses (to ). Further samples of eggs (either four, six, or eight eggs per clutch) were selected randomly, from 17 of the high elevation (HE) clutches and 18 of the low elevation (LE) clutches, to be experimentally incubated as part of this study. All remaining eggs were assigned to other experiments (see Andrews, Qualls & Rose, 1997; Andrews et al., 1999). All eggs were assigned to experimental treatments or sampled on 14 July, four to five days after oviposition was induced. Incubation treatments Two controlled temperature chambers (Percival model no. I-30BL with B 1 option) were used to produce two experimental incubation temperature treatments. The chambers were programmed to follow fluctuating temperature regimes (as is typical of natural nests) with die1 temperature ranges of or 2O-3O0C, simulating nest temperatures in the habitats of the high elevation and low elevation populations respectively (Fig. 1). Our temperature regime is based on actual nest temperatures recorded for LE S. virgutus (from Andrews & Rose, 1994). The treatment is based on estimates of nest temperatures in the HE habitat, assuming that nests in this habitat would be buried at depths and exposures similar to those of low elevation S. uiyatus (see Andrews et al., 1997). Our intent here was not to accurately simulate nest temperatures encountered by the HE population, which

7 INCUBATION TERIPEMI'URE AND OFFSPRING QUALITY 32 c h 26 m 3 24 U g G Time of day Figure 1. Die1 temperature regimes under which eggs of high elevation and low elevation Sceloporus vi7patu~ were incubated. The solid circles represent the warmer (20-30) treatment, simulating nest temperatures in a low elevation habitat. The open circles represent the colder (15-25) treatment, simulating nest temperatures in a high elevation habitat. we may or may not have done. Rather, our intent was to simulate the nest temperatures that would be experienced by low elevation lizards, that are (presumably) adapted to warm climates, when they first encounter the cold climatic conditions typical of high elevation habitats (either due to range expansion or climatic change). Specifically, we wanted to simulate the incubation temperature that eggs would experience if our LE population of S. uirgatus expanded its range into higher elevation habitats, but still exhibited the same nesting behaviour. An equal number of eggs from each clutch were assigned randomly to each of the and incubation temperature treatments. Eggs were placed individually in 65 ml glass jars containing moistened vermiculite, which were sealed with plastic kitchen wrap and secured with rubber bands. According to their assigned treatments, the jars containing eggs were placed in one of two environmental chambers, representing either the or the treatment, where they remained until hatching. Twice weekly, the jars were rotated within shelves, and the shelves were rotated among positions within each chamber, to minimize any position effects. The initial ratio of distilled water to dry vermiculite in the jars was 0.7:l.O by mass, giving a water potential of approximately -200 kpa (Packard et al., 1987), which is within the range of water potentials observed in the field for S. viyatus (-300 to > kpa; B. R. Rose, unpublished data). The eggs in both treatments took up large amounts of water during development, roughly tripling in mass, and showed no indications of a water shortage (Andrews et al., 1997). Hatchling housing conditions After hatching, the lizards were transferred to cages in an animal holding facility. The cages were 28 cm long by 13 wide and 14 tall, and contained a substrate of sand, loosely-folded paper towel as cover, and a shallow water dish formed from aluminium foil. One end of each cage was placed over an electric heating cable to

8 360 C. P. QUAI,I,S AND K. M. ANDREWS provide basking opportunities, and fluorescent vita-lites provided ultraviolet radiation. Light from outside windows provided a natural photoperiod, and timers controlled the fluorescent lights and heating cable. The fluorescent lights switched on at 0800 and off at 1630, and the heating cable turned on at 1000 and off at During the day, the heating cable produced a temperature gradient of approximately 38 to 23 C within each cage, and overnight, cage temperatures fell to room temperatures (mean daily minimum = 18.2 C, SD = 1.3). Water was provided several times daily, by spraying a fine mist into the cages, which formed droplets on the cage sides and small puddles in the foil dish, both of which the lizards drank from readily. The lizards were fed daily, with a mixture of small crickets, wax moth larvae, flour beetle larvae, and meal worms, which were dusted with a mineral supplement. Four lizards were housed in each cage, and the cage assignments were maintained throughout the study. Hatchlings were assigned to cages based on the order in which they hatched, and four lizards were placed into each cage before any were assigned to the next cage. This system assured that housing densities did not vary among cages or through time, and that all lizards were housed with individuals of approximately the same age. All lizards were maintained under these conditions until they were at least 20 days old. Incubation period, morphology, and growth rate The chambers were checked twice daily for hatchlings, and each lizard s date of hatching was recorded to calculate their incubation period. On the day they hatched, we measured the mass, snout-vent length (SVL), and tail length of each lizard, and gave each a unique toe-clip for identification purposes. Measurements of body mass, SVL, and tail length were repeated when each lizard was 20 days old. Size adjusted growth rates (based on both SVL and mass) were calculated for each individual over this 20 day period, using the following formulae: In SVL at 20 days old -In SVL at hatching = SVL growth in mm/mm/20 d; In mass at 20 days old -In mass at hatching=mass growth in g/g/20d. The morphology data were also used to examine the body shape of the lizards. We used analysis of covariance, to assess how fat the lizards were (their body mass relative to their SVL). Fluctuating asymmet9 We quantified the level of fluctuating asymmetry exhibited by each individual, by examining the level of bilateral asymmetry for 12 scale-count characters. We chose scale characters (following the terminology of Smith, 1967) for which distinct left and right-side counts could be made, and which one would expect to be bilaterally symmetrical (assuming perfect development). The 12 characters examined were: the number of subdigital lamellae beneath the first digit of the fore-limb; the number of subdigital lamellae beneath the first digit of the hind-limb; the number of postrostral scales; the number of scales in contact with the nasal scale (excluding the postrostrals); the number of circumorbital scales; the number of supraocular scales in contact with the superciliaries; the number of remaining supraoculars (those not counted in previous character); the number of supralabial scales; the number

9 INCUBATION TEMPERATURE AND OFFSPRING QUALITY 36 1 of lorilabial scales; the number of infralabial scales; the number of sublabial scales in contact with the infralabials; and the number of femoral pores. For each lizard, we recorded a left-hand and a right-hand count for each character and calculated the difference between the two. We calculated two overall indices of fluctuating asymmetry for each individual: total asymmetry = the sum of the absolute values of the left-right differences for all 12 characters; and the number of asymmetrical characters =how many of the 12 characters had left-right differences not equal to zero. 7hermoregulatoy behauiour On the day after hatching, the lizards thermoregulatory behaviour was observed by placing them in a thermal gradient. The thermal gradient was divided into 12 separate runways, each 4 cm wide and 89 cm long (see Andrews, 1994 for a more detailed description of the thermal gradient design). Lizards were placed individually into one of the runways at approximately 1000, where they remained for 2 h. On occasions when more than 12 lizards had to be observed, a second set of lizards was placed in the gradient immediately after the first group was removed. The heating and cooling elements of the thermal gradient were adjusted so that air temperatures ranged from approximately 20 to 40 C, and increased approximately linearly over its length. The positons and movements of lizards were monitored with a video camera, throughout the 120 min that each set remained in the thermal gradient. Singleframe video snapshots of the gradient were captured every 5min, using a video camera attached to an AV-equipped Power Macintosh computer. These images were later scored on the computer, to determine each lizard s position along the length of the gradient, at every 5 min interval. We calculated the standard deviation of each lizard s first seven locations (during the first 30 min) on the gradient, as an index of their activity levels (i.e. how much they moved) when exposed to a novel environment; but we did not collect thermoregulatory data for this period of acclimation to the gradient. The remaining position data (1 7 locations per individual) were converted to selected air temperatures. The lanes of the thermal gradient contained nine thermocouples, which were located 5 mm above the floor of the gradient and spaced at intervals of 10 cm along its length. These thermocouples were used to record air temperatures along the length of the gradient, for each time period over which a group of lizards was observed. These data were used to generate, for each group of lizards, calibration curves of air temperature versus position along the length of the gradient (third order regressions, all with &-0.99). Position data were then converted to selected air temperatures by substituting position into the appropriate regression equation, and correcting for minor differences among lanes. From these data, we calculated each individual s mean selected temperature and the standard deviation of their 17 temperature observations (as an index of the precision with which they maintained this mean temperature). Data for temperatures selected in a thermal gradient may not be normally distributed (especially if the preferred temperature range is closer to one end of the available range than the other), in which case the use ofmeans and standard deviations would be inappropriate; however, frequency distributions of our temperature data showed no obvious truncation or departures from normality. We also calculated median selected

10 362 C. P. QUALLS AND R. M. ANDREWS temperatures for each individual, and comparison of each lizard s mean and median selected temperatures revealed no significant difference between the two measures (two-tailed, paired t-test: mean difference = C; df = 2 16; t = 1.44; P= 0.15). Thus, we concluded that the use of means and standard deviations were appropriate for our data. We used this indirect method of assessing thermoregulatory behaviour because of difficulties inherent in the actual measurement of body temperatures. Measuring body temperatures directly would have required frequent disturbance and stressing of the lizards, and ectotherms as small as our hatchlings (body mass <0.5 g) have very low thermal inertia, and their body temperatures would likely have changed substantially during the capturing and handling necessary for the direct measurement of body temperatures (see Andrews, 1994; Qualls & Shine, 1998a). While this method could only provide estimates of the air temperature immediately surrounding the hatchlings, we expected a strong positive correlation between air temperatures and body temperatures for our hatchling lizards. Because of its importance, we tested this assumption by measuring air and body temperatures for hatchlings that were placed into the thermal gradient and physically restrained to a small area (by fencing off a short section of the gradient). After being allowed to equilibrate for approximately 30 min, the lizard s body temperatures were measured by quickly capturing the lizards by hand, and inserting a thermocouple into their cloaca. Linear regression of these body temperature data over estimates of air temperature at each lizard s location (estimated by the methods described above) showed a strong positive relationship between body temperature and air temperature (a= 14 hatchlings, slope = 0.62, 12 = 0.75, P<O.OOOl). Our measured body temperatures were, on average, two degrees cooler than air temperatures, and the slope of the regression was substantially less than one, both of which are the expected results of rapid loss of body heat once the lizards were removed from the gradient and exposed to cooler, room-temperature air. Thus, we concluded that our method of estimating selected microhabitat temperatures was an effective method of indirectly quantifying the thermoregulatory behaviour of the hatchling lizards. We also note that our methodology should work equally well for assessing the thermoregulatory behaviour other small ectothermic animals, and a more detailed description of our method is available from C. Qualls upon request. Escape behaviour We also assessed the hatchlings escape behaviour by chasing them down an electronically timed racetrack to record their running speeds (i.e. the time taken to cover a set distance) and the frequency with which they stopped and/or turned around. The racetrack was 1 m long by 5 cm wide and made of wood with sand glued to the surface to provide traction. Five sets of infrared photocells (connected to an electronic stopwatch) divided the track into four 25 cm intervals. Running speeds were measured in a controlled temperature room at five different temperatures (see below). For the escape behaviour trials, lizards were placed individually into small glass jars, then placed in the controlled temperature room and given at least 30 min to equilibrate to room temperature before running. Lizards were placed into an enclosed starting box for 30s, then chased down the track with a paintbrush. To avoid altering their body temperatures, the lizards were not touched by hand

11 INCUBATION TEMPERATURE AND OFFSPRING QUALITY 363 or with any other object that was not at room temperature. They were placed into the starting box either directly from their jars or from a plastic tub that collected the lizards at the end of the track. Each hatchling s escape behaviour was quantified at five different temperatures (28, 30, 32, 34, and 36 C), on the second through sixth days after they hatched. All lizards were run at the same temperature on any given day, and the room temperature was reset each day, rotating the five temperatures. Thus, all lizards were run at all five temperatures, each temperature on a different day, with the order of temperatures experienced by each individual depending on the day it hatched. Each hatchling was chased down the track twice at each temperature, with the second run immediately following the first. We recorded the time taken to cover the distance, and whether the lizard stopped and/or changed directions, for each 25cm interval. Thus, for each hatchling at each temperature, we recorded their running speeds over eight 25 cm sprints, and whether they stopped and/or turned around during each sprint. From these data, we extracted each individual s maximum 25 cm sprint speed (fastest of the eight segments), average sprint speed (average of the eight segment speeds), and the total number of segments in which they stopped/ turned. We also calculated the change in each individual s average sprint speeds from the first to the second run, as an index of their stamina. Any fatigue experienced by the lizards should result in a decrease in running speeds from the first to the second run, and the magnitude of this decrease should be an indicator of stamina. Data manipulation and anabsis Our full dataset was pseudoreplicated, because multiple individuals from a single female s clutch were placed into each treatment. To increase the independence of our data, we calculated mean values of each variable for all offspring from the same clutch in each treatment. Thus, for each variable, all individuals from each clutch of eggs were represented by only two data points, one for each incubation treatment. All data and analyses reported in this paper used these clutch mean data, except for the analyses of hatching success and hatchling survival, which used data for each individual. We analysed the data primarily using two factor analysis of variance (ANOVA) and analysis of covariance (ANCOVA), with population of origin and incubation treatment as the factors. For ANCOVAs, we compared intercepts only when slopes tests (factor by covariate interactions) were not significant (i.e. had -0.05). As is customary, we removed non-significant slopes terms from our ANCOVA models prior to evaluating the intercepts tests. However, we did make one departure from common procedure, in an effort to be more conservative. When slopes terms were not significant but had K0.25, we presumed that these terms might explain an important amount of the variance in the total model. Thus, we retained these nearsignificant interaction terms in the final models from which we computed intercepts tests.

12 364 C. P. QUALLS AND R. M. ANDREW S RESULTS Emblyo developmental rate Developmental rates of the embryos (as evidenced by incubation periods) did not differ between the two populations, but were markedly slower at the colder incubation temperatures. Incubation periods were substantially shorter for eggs in the treatment than for those in the treatment, and were also slightly shorter for eggs from the HE population than for the LE population (Table 1). However, the difference between the two populations did not reflect any difference in developmental rates (see also Andrews et al., 1999), but was an artifact of the fact that, at the time the eggs were collected, eggs from the HE lizards (N= 15 clutches, mean stage= 33.42, SE = 0.22) contained embryos at more advanced developmental stages than those from the LE population (N= 20 clutches, mean stage = 31.95, SE = 0.30). When incubation periods were compared using a two-factor analysis of covariance, with population and treatment as the factors and embryo stage at oviposition as the covariate, population was found to have no effect, but the difference between treatments was still significant (Table 2). Further, the difference in embryo stages between the two populations was associated with different female body sizes in the two populations (HE females: N= 15, mean SVL = 62.3 mm, SD = 3.1; LE females: N= 20, mean SVL= 58.9 mm, SD = 3.2; F,,33 = 9.75, P= 0.004). Embryo stage increased with maternal SVL, and when embryo stage was analysed using onefactor analysis of covariance, with population as the factor and female SVL as the covariate, embryo stage did not differ between the two populations (population: F,, 44 = 0.56, P= 0.46; SVL:F,,44 = 30.6, P= ; interaction B0.44). Thus, it is likely that the more advanced stages of HE eggs were due to the ability of larger females to begin reproduction earlier in the season (see Olsson & Shine, 1997), and not due to any intrinsic difference in reproductive timing between LE and HE lizards. Hatching success and hatchling survival Incubation at cold nest temperatures had detrimental effects on both the hatching success of eggs, and on the survival of hatchlings. For eggs from both populations, hatching success was significantly lower in the treatment than in the warmer treatment (Table 3). Further, survival rates (to the age of 20 days) of the lizards that hatched successfully was also lower for those from the treatment, but this difference was only significant for the LE population. Body size Body sizes of the hatchlings were significantly affected by both incubation treatment and population of origin (Tables 1 and 2). Hatchlings from treatment had shorter SVLs and shorter tails than did those from the treatment, but there was no difference in hatchling mass between the two treatments. At 20 days of age, the lizards from the treatment still had shorter SVLs and tail lengths, and had lower body masses as well. In addition to these treatment effects, hatchlings from the HE population had longer SVLs and heavier body masses than those from

13 INCUBATION TEhIPERATUKE AND OFFSPRING QUALITY 365 TABLE 1. Summary of data on the effects of incubation temperature treatments on incubation period, morphology, growth, levels of fluctuating asymmetry, and bchaviour in offspring of high elevation and low elevation Sceloporus uiyatus. The treatments simulated nest temperatures in a coldcr high elevation habitat (15-25) and in a warmer low elevation habitat (20-30). Sample sizes are given in parentheses, followed by means plus or minus one standard error. See text for explanation of variables. See Table 2 and the text for statistical tests Incubation period (days) SVL at hatching (my) Tail length at hatching (mm) Mass at hatching (g) SVL at 20 days (mm) Tail length at 20 days (mm) Mass at 20 days (g) SVL growth rate (mm/mm/20 d) Mass growth rate (dd20 d) Number of asymmetric characters Sum of asymmetries Initial activity Mean selected temperature ( C) Thermoregulatory precision High elevation treatment treatment (15) 92.2f 1.4 (15) 44.7f0.6 (15) 23.06i0.18 (15) 24.38k0.17 (15) 22.46k0.27 (15) 26.75k0.21 (15) 0.447k0.008 (13) 0.452f0.007 (15) 23.93i0.304 (15) 25.70k0.176 (15) 24.47i0.324 (14) 28.30k0.330 (15) 0,482~0,027 (13) 0.533f0.018 (15) (15) f0.006 (15) (15) 0.185f0.030 (15) 6.12f0.252 (15) 5.23k0.167 (15) (15) 6.57ko.219 (13) 8.54k0.596 (15) l0.17f0.979 (13) 37.07f0.223 (15) 34.73k0.357 (I 5) f (15) k0.100 Low elevation treatment treatment (18) f 1.9 (20) 48.5f0.8 (18) 22.72k0.22 (20) 23.88f0.21 (18) 22.55f0.33 (20) 26.55f0.27 (18) 0.417f0.011 (20) 0.426f0.012 (17) 23.39f0.305 (20) 25.23f0.223 (1 7) k (20) f (I 7) & (20) f (17) O.024 f (20) f (17) f0.048 (20) (18) 6.07k0.308 (20) 4.76k0.160 (18) 7.98f0.387 (20) 5.82f0.275 (18) 7.40f0.596 (20) 11.92kO.861 (18) 37.39f0.249 (20) f0.337 (18) 0.229k0.052 (20) 0.876k0.076 the LE population. By 20 days old, lizards from the two populations no longer differed in SVL, but the HE lizards still had greater body masses. It is possible that some or all of the differences in offspring size between the two populations could be an artifact of population-specific differences in the mass of the eggs (or more correctly, the reproductive investment in each egg). To assess this possibility, we repeated all of the analyses of offspring size using analysis of covariance to correct for variation in egg dry mass: i.e. two-factor ANCOVAs with population and treatment as the factors and egg dry mass (average dry mass of the one or two eggs sampled and dried from each clutch) as the covariate. We have not included all of the results from these analyses, because analysing the data in this manner yielded the same conclusions as the analyses listed in Table 2, except that SVL at hatching was no longer different between the two populations (F,,G3 = 2.10, P= 0.152). Growth rate Growth rates were also affected by incubation temperatures. Hatchlings from the treatment had significantly slower size-adjusted growth rates than those from

14 366 C. P. QUALLS AND R. M. ANDREWS TABLE 2. Statistical tests on the effects of incubation temperature treatments and population of origin on incubation period, morphology, growth, levels of fluctuating asymmetry, and behaviour in offspring of high elevation and low elevation Sceloporus viyutus. The treatments simulated nest temperatures in a colder high elevation habitat (15-25) and in a warmer low elevation habitat (20-30). For all variables except incubation period, two-factor analysis of variance was used to test the null hypotheses that population of origin or the incubation temperature treatments had no effect on the variables in question. For incubation period, two-factor analysis of covariance was used, with embryo stage at oviposition as the covariate; slopes terms were all non-significant (-0.25). SVL = snout-vent length. See text for explanations of variables and tests, and Table 1 for a summary of the data Statistical test: Statistical test: Statistical test: effect of population effert of incubation population x treatment of origin treatment interaction Incubation period (days) FI,,,j=l.91, P=0.172 Fl,,,l=11770, P=O.OOOl FJ,,3=l,28, P=0.263 SVL at hatching (mm) Fl,1,,=4.41, P=0.040 Fl,,,,=38.16, RO.0001 Fl,l,,=0.165, P=O.686 Tail length at hatching (mm) Fl,,,=0.042, P=0.839 E.;,,j, = 214.6, RO.0001 F1,,,,=0.262, P=0.610 Mass at hatching (g) Fl,,j,=7.41, P=0.008 Fl,,,,=0.465, P=0.498 Fl,,,=O.O24, P=0.878 SVL at 20 days (mm) Fl,l,l = 3.83, P= Fl,,,,j = 48.07, P<O.OOOl Fl,l,,l = 0.019, P= length at 20 days (mm) Fl,1,,=0.594, P=0.444 Fl,,,,= 119.5, P<O.OOOI F1,,,2=0.054, P=0.817 Mass at 20 days (g) F1,,,,4=6.80, P=O.O11 Fl,,,,= 10.34, P=0.002 Fl,b,=0.087, P=0.767 SVL growth rate (mm/mm/20d) FI,,,,=O.817, P=0.370 Fl,,,s=11.72, P=O.OOl FI,,,,=2.18, P=0.145 Mass growth rate (g/g/20 d) Fl.,,,l=2.35, P=0.130 Fl,l,.j=15.88, P=O.OOOZ FlJ,,=O.I58, P=0.692 Number of asymmetric characters PI,,,,= 1.29, P=0.261 FI,,+=22.41, P<O.OOOI FI,,,,=0.785, P=0.379 Sum of asymmetries Fl,l,,=0.662, P=0.419 Fl,l,4=27.90, RO.0001 Fl,,,,=2.40, P=O.126 Initial activity Fi,,,i=O.I48, P=0.702 Fl,,,+= 15.22, P= Fl,t,,=3.37, P=0.071 Mean selected temperature Fl,,,,=0.660, P=0.420 FI,,,,=65.59, P<O.OOOI Fl,,,,=0.055, P=O.815 'Ihermoregulatory precision FI,~,,= I.49, P=0.226 F1,~,+=69.14, P<O.OOO 1 Fl,,,i =0.448, Pz0.506 TABLE 3. Effects of incubation temperature on hatching success and hatchling survival rates of lizards, Sceloporus viyutus, from a high elevation and a low elevation population. Eggs from both populations were experimentally incubated at temperatures simulating nests in a warmer low elevation habitat (20-30 treatment) and in a colder high elevation habitat (15-25 treatment). The percentages of all eggs that hatched successfully, and the percentages of all hatchlings that survived until 20 days of age, are listed for eggs from each population that were assigned to each incubation treatment. The percentage survival values are followed by sample sizes (the total number of eggs or hatchlings in each group) in parentheses. Fisher's Exact P-values are for contingency table tests of the null hypothesis that survival did not differ between the two incubation temperature treatments Hatching success Hatchling survival High Low High Low elevation elevation elevation elevation treatment 82.1% (56) 82.6% (69) 89.1% (46) 71.9% (57) treatment 94.6% (56) 94.4% (71) 96.2% (53) 92.5% (67) Fisher's Exact P the treatment, in terms of both SVL and mass (Tables 1 and 2). There were no differences in growth rates between lizards from the two populations. Activip levels and themoregulatoly behaviour Incubation temperature had a significant effect on the activity levels and thermoregulatory behaviour of hatchlings, but there were no differences between the two

15 INCUBATION TEMPERATURE AND OFFSPRING QUALITY 367 populations (Tables 1 and 2). Hatchlings from the treatment were less active in response to a novel environment (i.e. they moved less during their first 30 min in the thermal gradient), they selected warmer temperatures in the thermal gradient, and they maintained these temperatures more precisely than their conspecifics from the 2&30 treatment. Levels djluctuating asymmetly All 12 of the scale characters we quantified exhibited asymmetry in at least some of the hatchlings. For each character, the differences between left and right-side counts were normally distributed and had modal values of zero, conforming to the definition of fluctuating asymmetry. The amount of fluctuating asymmetry in scdation did not differ between the two populations, but hatchlings from the treatment exhibited significantly higher fluctuating asymmetry than those from the warmer incubation treatment (Tables 1 and 2). Both the number of scale characters for which the hatchlings were asymmetrical, and the sum of their asymmetries (i.e. the sum of the absolute values of the differences between left and right-side counts for all characters) were significantly higher for the treatment. Body shape The body shape of hatchlings was not affected by their population of origin, but incubation temperature had a significant effect on the mass of hatchlings relative to their SVL. Two-factor analysis of covariance, with population and treatment as the factors, SVL as the covariate, and mass (In transformed) as the dependent variable, revealed that lizards from the treatment were fatter (i.e. had greater mass relative to their SVL) at hatching than those from the treatment (population: Fl,62=2.76, P=O.102; treatment: Fl,62=35.06, P=O.OOOl; SVL: F,,G2= 116, P= ; all interaction terms had -0.05). This difference in stockiness had disappeared by the age of 20 days (population: Fl,58 = 1.54, P= 0.220; treatment: F,,,, = 1.35, P= 0.250; SVL: Fl,58 = 2 12, P= ; all interaction terms had -0.05). Escape behaviour and peformance Data from the escape behaviour trials were analysed using repeated measures ANOVA, with two between factors (population and treatment) and one within factor (the temperature at which the measurements were taken: 28, 30, 32, 34, and 36 C). Incubation temperature significantly affected both the maximum and average sprint speeds of the hatchlings, as well as their stopping frequency (the number of segments in which they stopped); hatchlings from the two populations also differed in their average sprint speeds and their stopping frequency. Overall, maximum and average sprint speeds increased, and stopping frequency decreased, with increasing temperature (Fig. 2). Lizards from the treatment had significantly faster maximum sprint speeds, over 25 cm, than did those from the treatment (Fig. 2), but there was no difference in maximum speeds between the two populations (treatment: Fl,64 = 40.60, P= ; population: Fl,G4 = 0.280, P= 0.599; temperature: F4,256 =

16 368 C. P. QUALIS AND R. M. ANDREWS 16.53, P= ; all interaction terms had DO. 13). Lizards from the HE population had faster average sprint speeds than did those from the lower elevations (Fig. 2); average speeds were also higher for hatchlings incubated in the treatment than for those from the treatment, but a significant temperature by treatment interaction showed that average sprint speeds increased more rapidly with warmer temperatures for lizards from the treatment (treatment: = 63.69, P= ; population: F,,,, = 4.96, P= 0.030; temperature: F4,256 = , P= ; temperature by treatment interaction: F4,256 = 9.69, P= ; all other interaction terms had -0.1). Hatchlings from the treatment and those from the HE population stopped less often (Fig. 2) than did their counterparts from the warmer incubation treatment and the LE population (treatment: F],64 = 53.54, P= ; population: Fl,64 = 6.17, P= ; temperature: F4,256 = 4.50, P= 0.002; all interaction terms had -0.17). Running speeds were faster for hatchlings from the incubation treatment and for lizards from the HE population (average speeds only), illustrating that these lizards covered the distance of 25 cm more quickly, but this does not necessarily mean that these lizards ran faster. Hatchlings from both of these groups (15-25 treatment and HE population) also stopped less often during the escape behaviour trials, which may account for some or all of their faster running speeds. To test for this possibility, we regressed running speeds on the number of segments in which the lizards stopped. Separate linear regressions were performed for maximum speeds and average speeds at each of the five temperatures, and these tests all showed that running speeds decreased with increases in stopping frequency (N= 10 regressions; all had negative slopes with P<O.OOOl; r' ranged from 0.48 to 0.84). Therefore, we corrected the running speed data (both maximum and average sprint speeds) for stopping frequency. For each variable, we pooled the data from all five temperatures and regressed these data on stopping frequency. Both of these relationships were negative and highly significant (maximum sprint speed: N= 340 observations, slope = 0.045, P<O.OOOl, r' = 0.496; average sprint speed: N= 340 observations, slope = 0.024, P<O.OOOl, r'=0.735). Residual running speeds were calculated from each of these regressions, to generate measures of maximum and average sprint speeds that were corrected for the frequency with which the lizards stopped; that is, how fast the lizards actually ran when they were moving. These data, residual maximum speeds and residual average speeds (Fig. 3), were analysed using repeated measures ANOVA, as for the uncorrected running speed data. When corrected for stopping frequency, there were no significant differences in maximum sprint speeds between the treatment groups or populations, but speeds still increased with increasing temperature (treatment: F],64= 2.68, P= 0.107; population: FI,64= 3.30, P= 0.074; temperature: F4,256 = 9.46, P= ; all interaction terms had DO. 17). Average sprint speeds that were corrected for stopping frequency also did not differ between the two populations, but a significant temperature by treatment interaction showed that the relationship between residual average sprint speeds and temperature differed between the two incubation treatments (treatment: Fl,64 = 6.94, P= ; population: F,,64 = 0.005, P= 0.942; temperature: F4,25(, = 15.7, P= ; temperature by treatment interaction: = 12.1, P= 0.001; all other interaction terms had P2 0.1). Corrected average speeds increased noticeably with increasing temperatures for lizards from the treatment, but were relatively constant across temperatures for lizards from the treatment (see Fig. 3).

17 INCUBAI'ION TEMPERATURE AN11 OFFSPRING QUALITY O T - T, ' 0.3 Ir ri li 0.8 I $ " HE LE HE LE HE LE HE LE HE LE Figure 2. Summary of escape behaviour data for hatchling Sceloporus vigatus from high elevation and low elevation populations, that were incubated at temperatures simulating nests in low elevation (warm) and high elevation (cold) habitats. Mean values are given for (A) maximum sprint speed over 25 cm, (B) average sprint speed over 25 cm, and (C) frequency with which the lizards stopped and/or turned around. Escape behaviour trials were repeated at each of five ambient room temperatures (28, 30, 32, 34, and 36"C), and the data are summarized separately for each temperature. Within each temperature, the two bars on the left represent data for lizards from the high elevation (HE) population, and the two on the right represent lizards from the low elevation (LE) population. (0) hatchlings from the colder (15-25) incubation treatment; (m) hatchlings from the warmer (2C30) treatment. Error bars indicate one standard error. See text for statistical tests. Overall, the lizards' running speeds slowed from the first to the second run, presumably indicating fatigue, but comparison of these changes in running speed revealed no differences in stamina between populations or incubation treatments. For all but three of the twenty combinations of population, incubation treatment, and temperature groups, average sprint speeds decreased from the first to the second

18 370 C. P. QUALIS AND R. M. ANDREWS h UI 0.15 I 3 p a B T! i $ O.O8 HE LE HE LE HE LE HE LE HE LE Figure 3. Summary of residual running speed data for hatchling Sceloporus vigutu from high elevation and low elevation populations, that were incubated at temperatures simulating nests in low elevation (warm) and high elevation (cold) habitats. Mean values are given for maximum sprint speed over 25 cm (A) and average sprint speed over 25 cm (B), after each variable was corrected for frequency with which the lizards stopped and/or turned around. Escape behaviour trials were repeated at each of five ambient room temperatures (28, 30, 32, 34, and 36OC), and the data arc summarized separately for each temperature. Within each temperature, the two bars on the left represent data for lizards from the high elevation (HE) population, and the two on the right represent lizards from the low elevation (LE) population. (0) hatchlings from the colder (15-25) incubation treatmcnt; (m) hatchlings from the warmer (20-30) treatment. Error bars indicate one standard error. See text for statistical tests. run (Fig. 4). This ratio of 17 negative changes in speed to only three positive, is significantly different from the 10:10 ratio we would expect assuming an equal probability of positive or negative shifts (x'=9.8, df= 1, WO.01). However, the change in speeds did not differ among the two incubation treatments or the two populations, nor did it vary with temperature (treatment: F,,64 = 0.426, P= 0.517; population: F,,64 = 0.784, P= 0.379; temperature: F4,256 = 1.33, P=0.260; all interaction terms had -0.09). DISCUSSION Effects of cold incubation temperatures Our data suggest that cold nest temperatures (associated with high elevation or high latitude habitats) may be detrimental to oviparous squamates by slowing

19 INCUBATION TEMPERATURE AND OFFSPRING QUALITY I t m c: 2.A u I I I I HE LE HE LE HE LE HE LE HE LE Figure 4. Summary of stamina data for hatchling Sceloporus virgutus from high elevation and low elevation populations, that were incubated at temperatures simulating nests in low elevation (warm) and high elevation (cold) habitats. Mean values are given for the change in the hatchlings' average sprint speeds (over 25 cm) from the first to the second of two consecutive runs; this variable was calculated as an estimator of stamina. Escape behaviour trials were repeated at each of five ambient room temperatures (28, 30, 32, 34, and 36"C), and the data are summarized separately for each temperature. Within each temperature, the two bars on the left represent data for lizards from the high elevation (HE) population, and the two on the right represent lizards from the low elevation (LE) population. (0) hatchlings from the colder (15-25) incubation treatment; (m) hatchlings from the warmer (20-30) treatment. Error bars indicate one standard error. See text for statistical tests. embryonic development and increasing egg mortality, as is traditionally assumed; in comparison with conspecifics from warmer incubation temperatures, eggs in the cold (1 5-25) incubation temperature treatment had longer incubation periods and lower hatching success. However, cold-incubated hatchlings also were smaller at hatching (in length only) and at 20 days old (in length and mass), grew more slowly (in length and mass), had lower survival rates, had greater fluctuating asymmetry, were less active, had higher thermal preferenda, maintained their body temperatures more precisely, had faster running speeds, stopped less frequently when escaping, and were 'fatter' (had greater mass relative to SVL) than their conspecifics from the warmer (20-30) treatment. Thus, cold incubation temperatures affected many aspects of hatchling phenotypes; but how do these traits relate to offspring fitness? Can we conclude that the coldincubated hatchlings have poorer quality phenotypes? For several of these phenotypic traits, we can reasonably infer (based on empirical and/or logical grounds) that this may, indeed, be the case. The connection between survival and fitness is, perhaps, most obvious. Field studies on juvenile lizards have also shown that small body size if often associated with lower survival, although this is not always true (Ferguson & Fox, 1984; Sinervo et al., 1992). Slow growth rates may exacerbate the negative effects of small body size (by prolonging the time over which they are incurred), and can also be detrimental to reproductive success, by delaying the onset of sexual maturity, or by resulting in smaller body size (and lower fecundity) at the time of reproduction (Shine, 1980; Schwarzkopf, 1994). Thus, on the basis of these traits, we would conclude that cold incubation temperatures exert a negative influence on offspring phenotype quality in Sceloporus virgatus. However, this was not the case with all of the traits we measured; for some, our results seem counter-intuitive, and for others we cannot confidently predict their

20 372 C. P. QUALLS AND R. M. ANDKEWS effects on fitness. For example, hatchlings from the treatment actually had faster running speeds than those from the treatment. Much of this was due to the tendency of hatchlings from the treatment to stop less often while escaping, but these lizards had faster average sprint speeds (at warmer temperatures) even after they were corrected for stopping frequency. Intuitively, the ability to sprint faster when escaping would seem beneficial to fitness, or at least not detrimental. However, escape behaviour is more complex than simply running fast (see Bauwens & Thoen, 1981; Vitt & Price, 1982; Brodie, 1989, 1992; Schwarzkopf & Shine, 1992). For some squamates, frequent stops and starts may be a more effective escape strategy than simply running fast. If this is the case with juvenile S. vitutus, then the tendency of hatchlings from the treatment to stop less often could actually have a negative effect on their fitness. Similarly, cold-incubated hatchlings were fatter (i.e. had greater body mass relative to body length) at hatching, which has been associated with higher survival in juvenile sand lizards, Lucertu ugilis, but only in some years (Olsson, 1992). Hatchlings from the treatment were also less active in a novel environment, chose warmer microhabitat temperatures, and maintained these temperatures more precisely, but it is not clear what effect, if any, these traits would have on offspring fitness. Thus, cold-incubated hatchlings may accrue some fitness advantage as a result of their body condition, escape behaviour, activity levels, or thermoregulatory behaviour. However, it is likely that any resulting benefit would be outweighed by the costs associated with their small body size, slow growth, and low survival. Cold-incubated eggs also produced hatchlings with higher levels of fluctuating asymmetry, further supporting the conclusion that cold incubation temperatures are detrimental to offspring quality. Environmental stress can cause developmental instability (i.e. lower the fidelity of the developmental process), resulting in fluctuating asymmetry (Polak & Trivers, 1994; Mdler, 1997). Thus, the higher levels of fluctuating asymmetry associated with the colder incubation temperature suggest that these temperatures were more stressful for the developing embryos of both HE and LE lizards, leading to increased developmental instability. This conclusion is intuitive if we assume that the incubation treatment (which is indicative of the lower elevation habitats where S. virgatus normally occur) more closely represents ancestral conditions for S. virp.atus and that the colder treatment represents a recently encountered novel environment. Further, fluctuating asymmetry and other phenotypic traits differed among incubation temperature treatments in a correlated manner, such that more asymmetrical individuals from the treatment also exhibited poorer quality phenotypes for other traits as well. Such a negative correlation between fluctuating asymmetry and fitness has also been shown in many other taxa (see Moller, 1997 and references therein), and may indicate a causal relationship between developmental instability and phenotype quality or fitness. In this study, exposure of developing embryos (i.e. eggs) to cold temperatures induced significant developmental instability, as evidenced by fluctuating asymmetry of the resulting hatchlings. If mistakes during embryonic development are more often detrimental to an organism s physical and physiological functions than they are beneficial (as is the case with genetic mistakes ), then we would expect greater developmental instability during embryogenesis to result in offspring with more impaired abilities. Such impaired abilities could explain why our cold-incubated hatchlings exhibited relatively slow growth rates and low survival, despite the fact that they were reared (after

21 INCUBATION TEMPEK4TURE AND OFFSPRING QUAIJTY hatching) under identical conditions to their warm-incubated conspecifics. Thus, the cold incubation temperatures used in this study may not have directly influenced phenotypic traits such as hatchling growth rate or survival, but may have influenced these traits only indirectly, by increasing developmental instability. Coping with or avoiding the efects of cold incubation temperature., The second objective of this study was to determine whether the HE lizards exhibited any evidence of physiological adaptation to development under cold nest conditions. Specifically, we assessed whether the reaction norms for embryonic development and hatchling phenotype differed between the two populations (i.e. population x treatment or genotype x environment interaction), such that HE lizards developed more successfully, or had higher quality phenotypes when incubated at cold nest temperatures. We found, essentially, no differences in reaction norms to incubation temperature between the two populations. Thus, we find no evidence that embryo physiology of the HE population is better adapted to development at cold temperatures than that of the LE lizards. This raises a potential paradox. If the HE lizards are not physiologically adapted to development in cold climate habitats, how do they cope with the apparent detrimental effects of cold nest temperatures? One possibility is that HE S. virgatus may retain their eggs in utero for extended periods of time (an intermediate step in the evolution of viviparity), keeping them warmer than they would be in a nest (through maternal thermoregulation). The unusual reproductive characteristics of this species precluded a simple test of this hypothesis (i.e. by comparing embryo developmental stages at oviposition for the two populations). Female S. virgatus are able to facultatively prolong retention of their eggs, delaying oviposition until the summer monsoonal rains begin; and although embryonic development is retarded, embryogenesis does continue while the eggs are retained in utero (Andrews & Rose, 1994; Andrews, 1997). Thus, the developmental stage(s) at which eggs are oviposited is partially dependent on weather patterns, and likely varies from year to year. Further, all lizards in this study were artificially induced to oviposit at the same time (because Sceloporus virgatus do not oviposit readily in captivity), and we were not able to observe oviposition in the field. Thus, we were not able to assess whether the HE lizards typically retain their eggs in utero throughout more of development than the LE lizards do. Another possibility is that HE lizards may select nest sites on the basis of different criteria than their LE counterparts, allowing them to avoid (at least in part) cold incubation temperatures. For example, HE lizards may oviposit their eggs in more exposed (warmer) locations, or may use more superficial nests. If eggs are buried less deeply, they will experience greater die1 temperature ranges, which may result in warmer effective mean nest temperatures (Packard & Packard, 1988; Andrews, submitted). Finally, HE lizards may suffer detrimental effects of cold incubation temperatures, yet still be able to persist in the high elevation habitat. The biotic and/or abiotic characters of the high elevation habitat may simply be relatively undemanding (e.g. low predation pressure, abundant food, abundant cover, lack of competition, etc.), allowing a population of relatively unfit individuals to persist. All of these scenarios are possible solutions to this apparent paradox, and represent interesting avenues for future research.

22 374 C. P. QUALLS AND R. M. ANDREWS Conclusion Our data suggest that the invasion of a cold environment by S. virgatus would lead to delayed hatching of their eggs, as well as decreases in hatching success and in the quality of their hatchlings' phenotypes, all as a result of lower nest temperatures. All else being equal, these detrimental effects would provide a strong selective advantage for any mechanisms by which warmer incubation temperatures could be maintained, including prolonged egg retention. Thus, our study supports the cold climate model for the evolution of viviparity; our results suggest that cold climates are detrimental to oviparous reproduction by S. virgatus and that these negative effects would selectively favour increased egg retention. Further, this study suggests that the negative effects of cold incubation temperatures are more varied than has been traditionally assumed (i.e. delayed hatching, higher egg mortality), and can include widespread effects on developmental stability and the quality of hatchling phenotypes. ACKNOWLEDGEMENTS We are indebted to T. Mathies, F. R. Mtndez-de la Cruz, M. Villagran-Santa Cruz, and F. J. Qualls for providing valuable comments on this manuscript, to B. R. Rose for providing assistance in the field, and to T. Matthews, B. Horne, and C. Brewer for their assistance in the laboratory. We also wish to thank the staff of the Southwestern Research Station for their hospitality. Financial support for this work was provided by a National Science Foundation grant (BSR ) to R. M. Andrews. REFERENCES Andrews RM Activity and thermal biology of the sand-swimming skink NeosepJ rqnoldsi: die1 and seasonal patterns. Copeia 1994: Andrews RM Evolution of viviparity: variation between two sceloporine lizards in the ability to extend egg retention. Journal of zoology (London) 243: Andrews RM. submitted. Evolution of viviparity in squamate reptiles: a variant of the cold-climate model. Journal of zoology (London). Andrews RM, Mathies T, Qualls CP, Qualls FJ Rates of embryonic development of Sceloporus lizards: do cold climates favor rapid development? Copeia 1999: Andrews RM, Qualls CP, Rose BR Effects of low temperature on embryonic development of Sceloporus lizards. Copeia 1997: Andrews RM, Rose BR Evolution of viviparity: constraints on egg retention. Physiological,ZOO~OQ 67: Arnold SJ Genetic correlation and the evolution of physiology. In: Feder ME, Bennett AF, Burggren WW, Huey RB, eds. New Directions in Ecological Physiology. Cambridge: Cambridge University Press, Bauwens D, Thoen C Escape tactics and vulnerability to predation associated with reproduction in the lizard Lacerta uivipara. Journal of'ilnimal Ecology 50: Behler JL, King FW ?he Audobon Socieo Field Guide to North American Reptile., and Amphibians. New York: Alfred A. Knopf. Blackburn DG Evolutionary origins of viviparity in the Reptilia. I. Sauria. Amphibia-Reptilia 3:

23 INCUBATION TEhlPERATURE AND OFFSPRING QUALITY 375 Blackburn DG Evolutionary origins of viviparity in the Reptilia. 11. Serpentes, Amphisbaenia, and Ichthyosauria. Amphibia-Reptilia 6: Brodie ED Behavioral modification as a means of reducing the cost of reproduction. Amm can Naturalist 134: Brodie ED Correlational selection for color pattcrn and antipredator behavior in the garter snake Thamnophis ordinoides. Evolution 46: Burger J Effects of incubation temperature on behavior of hatchling pine snakes: implications for reptilian distribution. Behavioral Ecology and Sociobiology 28: Dufaure JP, Hubert J Table de dtveloppement du lkzard vivipare: Lacerta viuipara. Archiues Anatomie Microscopic Morphologie Experimental 50: Elphick MJ, Shine R Longterm effects of incubation temperatures on the morphology and locomotor performance of hatchling lizards (Bassiana duperreyi, Scincidae). Biological Journal of the Linnean Society 63: Ferguson GW, Fox SF Annual variation of survival advantage of large juvenile side-blotched lizards, Uta stansburiana: its causes and evolutionary significance. Evolution 38: Garland T Jr, Carter PA Evolutionary physiology. Annual Review ofpiysiology 56: Gutzke WHN, Packard GC Influence of the hydric and thermal environments on eggs and hatchlings of bull snakes Pituophis melanoleucus. Physiological ~oology 60: Heulin B, Guillaume C, Bea A, Arrayago MJ Interpretation biogkographique de la bimodalit6 de reproduction du lezard Lacerta vivipara - Jacquin - (Sauria, Lacertidae): un modtle pour I ttude de l kvolution de la viviparite. Biogeographica 69: Heulin By Osenegg K, Lebouvier M Timing of embryonic development and birth dates in oviparous and viviparous strains of Lucerta vivipara: testing thc predictions of an evolutionary hypothesis. Acta Oecologica 12: Meller AP Developmental stability and fitness: a review. American Naturalist 149: Olsson M Sexual selection and reproductive strategies in the sand lizard (Lacerta a,&.\). Ph.D. Thesis, University of Goteborg. Olsson M, Shine R The seasonal timing of oviposition in sand lizards (Lacerta agilis): why early clutches are better. Journal ofeuolutionary Biology 10: Packard GC, Packard MJ The physiological ecolocgy of reptilian eggs and embryos. In: Gans C, Huey RB, eds. Biology ofthe Reptilia (Volume 16). New York: Alan R. Liss, Packard GC, Packard MJ, Miller K, Boardman TJ Influence of moisture, temperature, and substrate on snapping turtle eggs and embryos. Ecology 68: Packard GC, Tracy CR, RothJ The physiological ecology of reptilian eggs and embryos, and the evolution of viviparity with the class Reptilia. Biological Reviews 52: Palmer AR, Strobeck C Fluctuating asymmetry: measurement, analysis, patterns. Annual Reuiea of Ecology and Systematics 17: Parsons PA Fluctuating asymmetry: an epigenetic measure of stress. Biologcal Reviews 65: Phillips JAY Garel A, Packard GC, Packard MJ Influence of moisture and temperature on eggs and embryos of green iguanas (Iguana kuana). Herpetologica 46: Pol& My Trivers R The science of symmetry in biology. Trends in Ecology and Evolution 9: Qualls CP, Shine R Reconstructing ancestral reaction norms: an example using the evolution of reptilian viviparity. Functional Ecology 10: Qualls CP, Shine R. 1998a. Costs of reproduction in conspecific oviparous and viviparous lizards, LeRrista bougainvillii. Oikos 82: Qualls CP, Shine R. 1998b. Lerista bougainvillii, a case study for the evolution of viviparity in reptiles. Journal of Evolutionary Biology 11: Qualls FJ, Shine R. 1998c. Geographic variation in lizard phenotypes: importance of the incubation environment. Biological Journal of the Linnean Society 64: Rose BR Factors affecting acitivity in Scelopom virgatus. Ecology 62: Schwarzkopf L Measuring trade-offs: a review of studies of costs of reproduction in lizards. In: Vitt LJ, Pianka ER, eds. Lizard Ecology: Historical and Experimental Perspectives. Princeton: Princeton University Press, Schwarzkopf L, Shine R Costs of reproduction in lizards: escape tactics and susceptibility to predation. Behavioral Ecology and Sociobiology 31: Shine R Costs of reproduction in reptiles. Oecologia 46:

24 376 C. P. QUAL.LS AND R. R.1. ANDREWS Shine R Reptilian viviparity in cold climates: testing the assumptions of an evolutionary hypothesis. Oecologia 57: Shine R The evolution of viviparity in reptiles: an ecological analysis. In: Gans C, Billett F, eds. Biology ofthe Reptilia (Volume 15). New York John Wiley and Sons, Shine R A new hypothesis for the evolution of viviparity in reptiles. American Naturalid 145: Shine R, Bull a The evolution of live-bearing in lizards and snakes. American Naturalist 113: Shine R, Harlow P Maternal thermoregulation influences offspring viability in a viviparous lizard. Oecologia 96: Sinervo B, Doughty P, Huey RB, Zamudio K Allometric engineering: a causal analysis of natural selection on offspring size. Science 258: Smith HM Handbook oflizards: Lizards ofthe United States and Canada. Ithaca, New York: Cornell University Press. Tinkle DW, Gibbons JW The distribution and evolution of viviparity in reptiles. Miscellaneous Publications Museum of ~oology, Universig ojmichigan 154: Van Damme R, Bauwens D, Brana F, Verheyen RF Incubation temperature differentially affects hatching time, egg survival and sprint speed in the lizard Podarcis muralis. Herpetologca 48: Vitt LJ, Price HJ Ecological and evolutionary determinants of relative clutch mass in lizards. Herpetologica 38: