Corticosterone manipulation reveals differences in hierarchical organization of multidimensional reproductive trade-offs in r-strategist and K-strategist females

Authors


Lesley T. Lancaster, Department of Ecology and Evolutionary Biology, EMS A316, University of California, Santa Cruz, 1156 High St., Santa Cruz, CA, USA.
Tel.: 831 459 4022; fax: 831 459 5353;
e-mail: lesley@biology.ucsc.edu

Abstract

Life history trade-offs are often hierarchical with decisions at one level affecting lower level trade-offs. We investigated trade-off structure in female side-blotched lizards (Uta stansburiana), which exhibit two evolved strategies: yellow-throated females are K-strategists and orange-throated are r-strategists. Corticosterone treatment was predicted to differentially organize these females’ reproductive decisions. Corticosterone-treated yellow females suppressed reproduction but survived well, and augmented egg mass without decreasing clutch size. Conversely, corticosterone enhanced mortality and reproductive rates in orange females, and increased egg mass only after lengthy exposure. Corticosterone did not affect post-laying condition, suggesting that corticosterone increased egg mass through enhanced energy acquisition (income breeding). Corticosterone enhanced survival of lightweight females, but decreased survival of heavy females, introducing a foraging vs. predation trade-off. We conclude that rather than being a direct, functional relationship, observed trade-offs between offspring size and number represent evolved differences in hierarchical organization of multidimensional trade-offs, particularly in response to stress.

Introduction

Life histories are strongly influenced by trade-offs, or negative associations between traits that each strongly influences fitness (Stearns, 1989; Roff, 1992; Clobert et al., 1998). This is primarily because structural and resource limitations prevent organisms from simultaneously maximizing each component of fitness, and therefore they must allocate resources into one fitness-enhancing trait or another. Resource limitations, biotic interactions (trade-offs imposed by predators or competitors) or genetic causes such as linkage, antagonistic pleiotropy or epistasis can affect trade-off structure (Zera & Harshman, 2001; Roff & Fairbairn, 2007; Sinervo & Clobert, 2008). Trade-offs imposed by resource limitation, biotic interactions and structural limitations of the organism represent ultimate constraints that impose the need for trade-offs. The genetic aspects of trade-offs (such as pleiotropy) either arise as adaptations to cope with imposed ultimate constraints or, when maladaptive, cause temporary trade-offs that can later be resolved by new adaptations altering the genetic architecture (Houle, 1991). These two distinct types of contributions to observed trade-offs can be difficult to unravel in the absence of experimental manipulations (Charlesworth, 1990).

When considered from the point of view of resource allocation, an organism’s organization of trade-offs is often hierarchical (de Jong, 1993), with allocation decisions at one level influencing whether or how the organism reacts to trade-offs at the next level. As more resources are allocated along one fitness axis (e.g. reproduction), trade-offs hierarchically nested within that axis are reduced or eliminated (Pease & Bull, 1988; de Jong, 1993; Worley et al., 2003). For example, if the survival vs. reproduction trade-off is higher in the hierarchy than are trade-offs among parentally influenced offspring characteristics, increasing the allocation of resources to reproduction reduces or eliminates trade-offs between offspring characteristics (because of a surplus of resources dedicated to all offspring functions). Therefore, resource allocation trade-offs are best considered on multiple levels and are not captured by simple negative associations between two traits (de Jong, 1993; Worley et al., 2003; Roff & Fairbairn, 2007). This predicts that the more resources are available, the less trade-offs will be observed on any level. This occurs over an individual’s lifespan for plastic resource allocation traits (e.g. Tatar & Carey, 1995) and over evolutionary time for those mediated by antagonistic pleiotropy or other genetic mechanisms (e.g. experimental evolution in Caenorhabditis elegans; Barnes & Partridge, 2003). Organisms with plastic allocation strategies at the level of individual trade-off decisions are also predicted to exhibit phenotypic plasticity in their entire trade-off structures, with shifts in relative allocation to different functions in different environments (Ernande et al., 2004). Less well-explored, populations may also express genetic as well as environmental variation in the hierarchical organization of trade-offs (e.g. whether survival vs. reproduction takes precedence over offspring quality vs. quantity, or vice versa). In addition, trade-offs might be multidimensional with a more complicated organizational structure not captured by simple, two-dimensional hierarchical models (Gaillard et al., 1989; Ferrière & Clobert, 1992; Clobert et al., 1998; Sinervo & Clobert, 2008, Mills et al., in press).

Adding to the complexity, resource acquisition rates are not usually the same among individuals within a population. Acquisition rate may not depend on overall resource availability, a general environmental effect (Lynch & Walsh, 1998), but may be a special environmental effect as a result of microhabitat differences in resource availability. Because resource acquisition is often behavioural, it can also have a strong genetic component resulting from individual differences in foraging behaviour, which may interact with microhabitat differences, resulting in a G × E (genetic by environmental) interaction (Hughes & Taylor, 1997). Furthermore, resource acquisition itself is involved in trade-offs with biotic interactions with predators or conspecifics that limit feeding rates (Brown & Kotler, 2004). Finally, genetic/physiological trade-offs, such as antagonistic pleiotropy of endocrine effects on multiple life history traits, are predicted to strongly affect the entire trade-off structure (Ketterson & Nolan, 1999), and genetic or environmental variation in endocrine activity is predicted to increase the dimensionality of observed trade-offs (Sinervo & Clobert, 2008).

In order to unravel the hierarchical organization of multidimensional trade-offs that underlie the allocation decisions of different life history phenotypes, experiments are needed to perturb the current pattern of life history allocation (Sinervo & Clobert, 2008). We chose to perform this by experimentally altering levels of the stress hormone corticosterone in the model species, the common side-blotched lizard (Uta stansburiana), which exhibits an intra-populational female reproductive strategy polymorphism. This provides a unique opportunity to compare the hierarchical organization of the trade-off structure across different evolved strategies but against the same genetic background.

Glucocorticoids, including corticosterone, are a key proximate factor involved in evolved plasticity in life history allocation decisions. Glucocorticoids are elevated in response to environmental cues commonly thought of as stressors, and have a variety of physiological and behavioural effects, all of which strongly affect life history allocations. Glucocorticoids affect energy acquisition and utilization by enhancing gluconeogenesis and mobilizing energy stores from fat reserves (Porterfield, 2001; Moore & Jessop, 2003). They can also either increase or decrease feeding rates depending on whether the animal is under chronic or tonic stress (Wingfield et al., 1998). Glucocorticoids also increase restlessness and movement behaviours, inducing dispersal or abandonment of territory or offspring (de Fraipont et al., 2000; Cote et al., 2006). Glucocorticoids can either increase or decrease growth, survival and reproduction depending on the species or social strategy that is studied, and on the duration of exposure (Wingfield et al., 1998; Comendant et al., 2003; Meylan & Clobert, 2005). Furthermore, differing condition or different life stages cause animals to either elevate or suppress glucocorticoid secretion in response to environmental stressors (Romero, 2002). These categories of effects suggest that glucocorticoids represent an endocrine mechanism translating environmental cues into life history allocations. Striking differences in observed physiological effects of glucocorticoids across different species with different evolved life history strategies suggests that glucocorticoids do not simply mediate organisms’ responses to two-dimensional trade-offs. Instead, glucocorticoids (and other hormones) may be proximately involved in shaping the entire hierarchical pattern of multidimensional life history trade-offs, and glucocorticoids likely induce different patterns of resource allocation against different genotypes and different environmental or developmental conditions (Dufty et al., 2002).

We experimentally altered corticosterone titres in females of U. stansburiana, which are well characterized for their offspring quantity – size trade-off (Sinervo & Licht, 1991a,b). This annual lizard species exhibits six alternative throat colours (orange, blue-orange, blue, blue-yellow, yellow and yellow-orange). Linkage mapping (Sinervo et al., 2006b), theory (Sinervo, 2001) and laboratory crosses (Sinervo et al., 2001) indicate that colour is controlled by three codominant colour alleles at the OBY locus: o, b and y. In males, alternative throat colours correspond to alternative mating strategies (Sinervo & Lively, 1996). In females, the throat colours correspond to alternative reproductive strategies: females carrying orange alleles (oo, bo and yo), hereafter, orange females, also known as r-strategists (Sinervo et al., 2000), evolved to lay larger clutches of smaller offspring. In contrast, yellow females, who lack orange alleles (yy, by and bb), are K-strategists that lay smaller clutches of larger eggs, which produce larger hatchlings (Sinervo et al., 2000). Clutch size and egg mass are both heritable, reflecting evolved reproductive strategies (Sinervo & Doughty, 1996; Sinervo et al., 2000). Females of this species experience elevated baseline corticosterone (their primary glucocorticoid) during the breeding season (Wilson & Wingfield, 1992). They also experience relatively higher baseline corticosterone levels during the breeding seasons when crowded by female conspecifics (Comendant et al., 2003). Endogenous corticosterone titres do not differ by female strategy type when statistically controlling for the effects of the social environment, and endogenous titres are consistent across years (Comendant et al., 2003). Previous studies in this system have shown no linear effects of corticosterone on clutch size, egg size or clutch mass, i.e. total reproductive investment per reproductive episode (Sinervo & DeNardo, 1996), and varying results for the effects of corticosterone on female survival, depending on year and season (experimentally elevated corticosterone: Sinervo & DeNardo, 1996; endogenous corticosterone: Comendant et al., 2003). Here, we experimentally manipulated corticosterone in females of this species via silastic implants, varying both the timing of implantation and the rate of release of corticosterone from the implant, to investigate linear and higher-order effects of corticosterone on reproductive strategy.

We investigated corticosterone’s effects on all the following reproductive characteristics: survival into the breeding season and whether or not females successfully produced a clutch of fertilized eggs (possibly reflecting a survival vs. reproduction trade-off if the decision to allocate resources towards reproduction increases mortality rates); timing of reproduction (i.e. does corticosterone accelerate or delay reproduction?); clutch mass, number of eggs and average egg size; female pre- vs. post-laying mass (input of the female’s fat reserves into eggs); survival of the female after laying; and survival of the progeny to adulthood.

We predicted that corticosterone’s effects would differ for orange and yellow females (Svensson et al., 2002), reflecting different evolved life history plasticities for r-strategist vs. K-strategist female types. These differences are predicted to go beyond allocation to offspring quality vs. quantity, and r- and K-strategist females are predicted to express evolved differences in the entire hierarchical structure of trade-off organization. This prediction rests on the hypothesis that different organization of physiologically or genetically based trade-offs may function as adaptations to the same underlying resource limitations and structural constraints.

Methods

Fieldwork was conducted on a wild population of U. stansburiana on Billy Wright Road, Merced Co., CA, USA in 2003 and 2004. Females were captured during the course of normal population censuses and were brought into the laboratory for surgery in regular intervals starting March 1 through April 15, before they reproduced. Females were scored for throat colour genotype, weighed to the nearest 0.05 g, measured to the nearest 0.5 mm for snout-to-vent length (SVL) and randomized among the following treatment groups by flipping a coin. Treatments fell into three categories. In one group, we manipulated corticosterone levels using implants. In this group, some females (n = 111) received a low-dose silastic implant made of 3 mm lengths of 1.57 mm inside diameter and 3.18 mm outside diameter silastic brand laboratory tubing (Dow Corning; Midland, MI, USA) filled with 1 mm of crystalline corticosterone (Sigma, St Louis, MO, USA), and 1 mm plugs of silicone on each end, designed to imitate high-end basal physiological levels (DeNardo & Licht, 1993; Sinervo & DeNardo, 1996). Other females (n = 162) received a high-dose silastic implant made of 1.98 mm inside diameter and 3.18 mm outside diameter silastic tubing, designed to imitate stress-related levels, which are double or more than double basal levels (Comendant et al., 2003). Other females (n = 141) received a sham implant, which was identical to the corticosterone implants but did not contain corticosterone and was filled with silicone gel. In the second treatment group, we removed yolk from developing follicles in order to directly manipulate reproductive characteristics and experimentally decouple the effects of reproductive characteristics on female fitness from the direct effects of corticosterone treatment on female fitness. In this second group of manipulated females, some females (n = 70) received a unilateral yolkectomy surgery referred to as follicle ablation (Sinervo & Licht, 1991a; b), which experimentally gigantizes eggs. Other females (n = 69) received a sham surgery with an incision but no implant or follicle ablation. Another treatment group of females (n = 51) received both a high-dose corticosterone implant and follicle ablation. Other females (n = 41) received low-dose corticosterone plus follicle ablation. Eight females received a sham implant and follicle ablation. The third and final treatment group included females (n = 107) who were brought into laboratory as controls but received no surgery. Treatments were randomly dispersed over 2 years to account for variation in frequencies of female genotypes across years (frequencies follow a 2-year cycle, Sinervo et al., 2000). For all females, follicular stage at time of treatment was determined by abdominal palpation and recorded as estimated size in millimetre. For follicle-ablated females, follicle size was directly measured in order to confirm our estimates from palpation. Only females with follicles of 4–6 mm were considered for inclusion in the follicle ablation group as that is the developmental stage at which the treatment is effective in gigantizing eggs (Sinervo & Licht, 1991a). Females receiving surgery were anesthetized with a local injection of lidocaine and their body temperature was reduced to approximately 4 °C to limit movement during surgery.

Following treatment, females were released to their site of capture and monitored for survival. When gravid (following ovulation), females were brought in to the laboratory to lay their eggs. Upon laying, two eggs per clutch were miniaturized by aspirating 20% of the yolk with a syringe (methods from Sinervo et al., 1992). This results in smaller hatchling size at birth, which when progeny are released into nature, provides a causal assessment of the effects of yolk volume on progeny survival. For females who had received follicle ablation, this procedure reduces egg size approximately back to normal. Therefore, in this case, clutch size and egg size are further experimentally decoupled. Eggs were incubated at 28 °C in individual cups of moistened vermiculite, and upon birth, hatchlings were weighed to the nearest 0.01 g, measured for SVL to the nearest 0.5 mm, toe clipped for individual identification and released to the site of their mother’s capture. The following spring, progeny were recaptured as adults to determine survival to the start of the next breeding season.

Statistical methods

For analysis, levels of corticosterone treatment were scored as a continuous variable with the values 0, 1 and 2, based on the fact that our two experimental levels of corticosterone treatment were chosen from a continuous distribution of possible treatment levels. For female survival and reproductive traits, all analyses were limited to females not receiving follicle ablation. We wanted to determine effects of corticosterone on reproductive characteristics without confounding effects of experimental manipulation of clutch and egg size. However, follicle-ablated females were included in analyses of offspring survival in order to experimentally decouple effects of offspring size and number from each other and from other genetically correlated traits affecting fitness. In all analyses reported, the sham females did not differ from the control females; therefore, we included the sham females and excluded the controls from analyses in the interest of balanced sample sizes.

Survival to the first clutch

Females were scored as survivors if they were recaptured 15 days post-treatment or later. Survival was scored as 1 = lived, 0 = died. A nominal logistic regression was performed, regressing survival on level of corticosterone, level of corticosterone2, treatment date, follicular stage at time of treatment, female mass and SVL at time of treatment, and number of o and y alleles (e.g. an oo female receives a score of 2 on the orangeness axis, whereas bo and yo females are scored as 1 and bb, by and yy females are scored as 0, following Sinervo, 2001). All interactions up to three-way interactions between these variables were included, and year was included as a random factor to control for differences between 2003 and 2004. Interactions of the other factors with year were not considered. All nonsignificant effects were subsequently trimmed from the model using a backwards stepwise procedure (eliminating higher-order terms first) such that all tested effects not reported in the results did not affect survival.

Reproduction

Reproduction was assessed for females who survived into the breeding season (15 days post-treatment). These females were scored as having reproduced if they laid a clutch of fertilized eggs. Reproduction was scored as 1 or 0 for yes vs. no, and a nominal logistic regression was run with the same factors as the analysis for survival. The interval between treatment date and laying date was assessed for all females who reproduced, in order to determine potential effects of corticosterone on reproductive timing (de Fraipont et al., 2000). This continuous variable was regressed on the same factors as those putatively predicting survival and reproduction (corticosterone, corticosterone2, treatment date, follicular stage at treatment, mass and SVL at treatment, o and y alleles, and all interactions). A backwards, stepwise procedure was performed to generate final models for both reproduction and timing of reproduction.

Clutch characteristics and post-laying female mass

Clutch mass, number of eggs and average egg mass were assessed for the first clutch following treatment for all females who reproduced, and the female’s post-laying mass was recorded. We regressed clutch mass on level of corticosterone, level of corticosterone2, mass and SVL at the time of treatment, treatment date, the treatment to laying date interval, follicular stage at time of treatment, the number of o and y alleles of the dam, all interactions up to three-way interactions and a term for year (2003 vs. 2004), removing nonsignificant effects in a stepwise manner. Upon obtaining the final model for clutch mass, we regressed clutch size and average egg mass against those effects that were significant in the clutch-mass model (these included all initial effects and interactions except follicle stage at treatment, SVL and treatment date; see Results). This was performed to determine how corticosterone affected offspring quantity vs. quality in its effects on clutch mass. We compared each female’s post-laying mass to her mass at the time of treatment with a repeated measures manova, including all the same factors as in the model for clutch mass (with the exception of female mass at treatment, now a response variable).

Post-laying survival

Survival of females after laying was assessed at 20 days (or later) following their first laying date. We performed a nominal logistic regression examining the effects on survival after laying of corticosterone, corticosterone2, clutch mass, female post-laying mass, the treatment to laying date interval, and laying date, number of o and y alleles, and all interactions among these variables. We also included year as a random factor. Nonsignificant effects were removed in a backwards stepwise manner. Recapture rates exceed 98% (Sinervo et al., 2006a), so there was no need to use complex capture–recapture models (Clobert et al., 1987).

Offspring survival

We approximated fitness of dams in terms of contribution to the gene pool of the next generation. Because the distribution of offspring survival was heavily skewed in favour of females who produced either zero or one surviving offspring, and because we could not make the assumption that producing two surviving offspring doubled a female’s fitness compared with one surviving offspring (as the effects of offspring quality may extend into adulthood), we scored females on whether or not they produced at least one offspring (from their first clutch) that survived until the start of the breeding season as adults the following year. For this analysis, we included all the females from the study, including those treated with corticosterone and those receiving follicle ablation surgery. Seventy-five percent of females producing progeny had at least one hatched offspring arising from eggs experimentally miniaturized at laying, further manipulating hatchling size. We initially included average hatchling mass, number of hatchlings, hatch date, year, corticosterone, corticosterone2, maternal o and y alleles and all interactions as predictors of hatchling survival in a logistic regression model. Nonsignificant effects were pruned to produce the final model.

All variables were checked for normality and noncollinearity, and alpha was set to 0.05 for statistical significance.

Results

Survival to the first clutch

Female survival was positively influenced by the number of o and y alleles, and a significant orange × yellow interaction revealed that yo females survived better than by or bo females (Table 1a). Treatment date also influenced survival such that females who received treatment later in the season survived less well than those treated early in the season. Corticosterone interacted with the number of o alleles in its influence on survival: increasing corticosterone treatment (levels 0, 1 and 2) linearly decreased chances of survival for females carrying one or more orange allele, but corticosterone treatment increased survival of non-orange females (Table 1a, Fig. 1a).

Table 1.   Corticosterone mediates current vs. future reproduction allocation differentially by female strategy type: (a) Corticosterone enhances survival of yellow females and decreases survival of orange females; (b) Enhanced rates of reproduction among orange females experiencing corticosterone, and suppression of reproduction by corticosterone in yellow females.
EffectEffect directionχ2P-value
a. Female survival following treatment (n = 444)
 Level of corticosterone+0.040.85
 Orange alleles+9.780.002
 Corticosterone × orange alleles10.360.001
 Treatment date19.10< 0.0001
 Yellow alleles+5.240.02
 Orange alleles × yellow alleles+4.470.03
b. Did the female reproduce, given survival? (n = 234)
 Level of corticosterone+0.000.97
 Yellow alleles+0.130.72
 Corticosterone × yellow alleles5.010.03
 Female mass at treatment+5.140.02
 Follicle stage at treatment+5.410.02
 Year 35.61< 0.0001
Figure 1.

 The effect of corticosterone on survival and reproduction differed by female strategy type (Table 1). (a) Female survival to 15 days following treatment. (b) Female reproduction (production of at least one clutch), assessed for females who survived. Solid/orange line: females carrying one or more o alleles; dashed/blue line: females lacking o alleles (yellow females). [Colour version of figure available online.]

Reproduction

Whether a female reproduced was positively influenced by both female mass and follicular stage at time of treatment. It also differed significantly by year, with more successful reproduction occurring in 2004. Corticosterone interacted with the number of yellow alleles, increasing the probability of reproduction for non-yellow females and decreasing chance of successful reproduction for yellow females (Table 1b, Fig. 1b).

The interval between treatment date and laying date was not affected by corticosterone. Instead, mass and follicular stage at treatment both negatively affected the interval such that larger females and those with larger follicles took less time to reproduce following treatment (response variable = treatment to laying date interval. effects: follicular stage: β = −3.21 ± 0.98, F1,131 = 10.81, P = 0.001; mass: β = −6.24 ± 2.07, F1,131 = 9.07, P = 0.003; year: F1,131 = 32.03, P < 0.0001, n = 135).

Clutch characteristics

Clutch mass was affected by an interaction between level of corticosterone2 × treatment to laying date (F1,101 = 5.87, P = 0.02, n = 117; Fig. 2a) and also by female mass, measured at the time of treatment (β = 0.21 ± 0.06, F1,101 = 10.54, P = 0.002, n = 117). In addition, significant in this model is an interaction between level of corticosterone2 × number of yellow alleles (F1,101 = 6.99, P = 0.01, n = 117; Fig. 3a) such that low-dose corticosterone was more likely to increase clutch mass for yellow females. In addition, significant is the interaction: corticosterone2 × treatment to laying date × number of orange alleles (F1,101 = 5.13, P = 0.03, n = 117; Fig. 3b, c), indicating that the covariate treatment to laying date is a more important mediator of the effects of low-dose corticosterone on clutch mass for orange females than for yellow females. When the response variable is number of eggs, corticosterone and treatment to laying date interval are nonsignificant (corticosterone2 × treatment to laying date F1,128 = 2.21, P = 0.14, n = 134) and only female mass significantly affects egg number (β = 0.38 ± 0.17, F1,128 = 5.10, P = 0.03, n = 134). However, corticosterone acts on average egg size (corticosterone2 × treatment to laying date: F1,110 = 9.57, P = 0.003, n = 117; Fig. 2b. Effect of female mass on average egg size in the same model is β = 0.03 ± 0.01, F1,110 = 5.22, P = 0.02, n = 117).

Figure 2.

 (a) Moderate elevation of corticosterone levels, experienced early with respect to egg laying, increased total clutch mass. Higher levels of corticosterone, or corticosterone elevation imminent to laying, did not produce this effect. (b) Corticosterone-induced clutch mass increase was primarily because of an increase in egg mass, but without decreasing clutch size.

Figure 3.

 (a) K-strategist yellow females increased clutch mass in response to corticosterone without respect to timing of corticosterone increase. (b, c) However, r-strategist orange females had a higher threshold for duration of corticosterone exposure before increasing egg mass (reflecting differing patterns of selection on progeny of r- and K-strategist females).

Female post-laying mass and survival

The difference in mass experienced by females from the time of treatment to their post-laying mass was not affected by corticosterone or female throat colour, and was associated with the female’s SVL (F1,109 = 8.81, P = 0.004, n = 113), follicle stage at treatment (F1,109 = 19.72, P < 0.0001, n = 113), and the interval between treatment date and laying date (F1,109 = 19.51, P < 0.0001, n = 113). Advanced follicular development at the time of treatment and longer SVL were associated with greater weight loss, whereas females with longer treatment to laying date intervals experienced weight gain. Female survival to 20 days after laying was affected by an interaction between corticosterone × post-laying female mass (χ2 = 4.87, P = 0.03, n = 120). In the absence of corticosterone treatment, females with a heavier post-laying mass survived best, and heavier females who did not receive corticosterone treatment survived best overall. However, the lightest females experienced a linear increase in post-laying survival with increasing levels of corticosterone treatment.

Offspring survival

Corticosterone did not directly affect whether females produced at least one offspring that survived to adulthood the next year; however, it increased female fitness indirectly through its effect on clutch mass. The number of offspring hatched was positively correlated with the number of eggs in the clutch (β = 0.56 ± 0.09, F1,169 = 36.68, P < 0.0001, n = 171), and hatchling size was positively correlated with incubated egg size (β = 0.95 ± 0.05, F1,532 = 432.92, P < 0.0001, n = 534). Both number of progeny and size of progeny significantly increased the chance of producing surviving offspring (effect of average offspring size: χ2 = 11.06, P = 0.0009, effect of progeny size × maternal orange alleles: χ2 = 4.69, P = 0.03 (Fig. 4a), effect of number of offspring hatched: χ2 = 19.57, P < 0.0001 (Fig. 4b), effect of year: χ2 = 8.75, P = 0.003; n = 176) with 27% of females producing surviving hatchlings. The effect of offspring size on offspring survival was less positive for progeny of orange dams. Progeny size was uncoupled from progeny number because of our experimental manipulations of follicle ablation and egg miniaturization, and therefore these variables were not confounded with each other in the analysis (effect of progeny number on within-clutch average progeny size: β = 0.0045 ± 0.0035, F1,174 = 1.65, P = 0.20).

Figure 4.

 Selection on females adaptively reflected the effects of corticosterone. (a) Selection on offspring size was positive for females of both strategy types, but was stronger for yellow females. Solid/orange line: orange females; dashed/blue line: non-orange (yellow) females. (b) Selection on offspring number was also positive for all females, and did not differ by female strategy type. [Colour version of figure available online.]

Discussion

Corticosterone is naturally increased in both orange and yellow females in response to crowding by conspecifics (Comendant et al., 2003), suggesting that its role is at least in part to enhance the competitive ability of each female strategy. Corticosterone interacted with both female reproductive strategy types (r-strategists vs. K-strategists), and with timing of reproduction and female condition in affecting reproductive characters in U. stansburiana (Table 1). Corticosterone linearly affected survival into the reproductive period, but differentially by female strategy type. Orange females experienced decreasing chances of survival with increasing levels of corticosterone, whereas yellow females experienced increasing chances of survival with increasing levels of corticosterone. However, once having survived, the effects of corticosterone on reproduction showed an opposite trend, with yellow females being less likely to reproduce with increasing levels of corticosterone and orange females being more likely to reproduce. These results indicate that corticosterone mediates a current vs. future reproductive trade-off in this system such that females carrying y alleles favour survival over reproduction when their corticosterone levels are elevated, and females carrying o alleles favour reproduction over survival. This reflects a benefit to r-strategist orange females from laying earlier in the season so that their smaller hatchlings emerge under less dense conditions; therefore, they do not benefit from suppressing reproduction in favour of survival to later in the season as K-strategist yellow females do. Furthermore, r- and K-strategist females experienced different costs of elevated glucocorticoids early in reproduction: r-strategist females translated glucocorticoid elevation into increased rates of reproduction, thus incurring survival costs (see below). K-strategists did not experience a survival cost as corticosterone caused them to suppress reproduction (reproduction cost).

Corticosterone did not act to accelerate or repress timing of reproduction in this study, as has been shown in some species (Salvante & Williams, 2003). Instead, corticosterone interacted with reproductive timing to affect clutch mass. Females of both strategies put more materials into their first clutch if treated with low-dose corticosterone at least several weeks before laying. Materials were invested primarily into egg size, and corticosterone did not affect egg number. Although follicular development at the time of treatment did influence the interval between treatment and laying date, follicle stage at treatment did not directly influence the effect of corticosterone on clutch mass/egg mass. This suggests that females are likely primed by threshold duration of chronically elevated corticosterone in order to initiate egg mass increase, but that the actual inclusion of additional yolk into follicles occurs late in follicular development. Yellow females were more sensitive to the effects of any duration of corticosterone exposure on clutch mass, whereas orange-throated females only increased clutch mass in response to a longer duration of exposure. Females of both strategy types benefit from producing larger hatchlings under the competitive (crowded) conditions promoting endogenous corticosterone increase in this species. However, progeny of r-strategist, orange females benefited less by increasing egg size, perhaps because they pass on genes benefiting hatchlings from smaller eggs (maternal–offspring co-evolution; Wade, 1998). Thus, orange females’ higher threshold for duration of corticosterone treatment before initiating a larger egg mass may be adaptive.

It is noteworthy that a genetic polymorphism for egg size vs. number trade-off in this species also involves a genetic difference in response to stress (corticosterone) along the survival vs. reproduction trade-off, a result that may be general. For K-strategist (yellow) females, it appears that the offspring quantity vs. quality trade-off takes precedence over the survival vs. reproduction trade-off as all life history decisions are biased in favour of offspring quality (for K-strategists, corticosterone either suppressed reproduction or increased offspring size, suggesting that enhanced survival is a by-product of suppressed reproduction when females lacked the resources to increase egg size). For r-strategists (orange) females, the current survival vs. reproduction trade-off appears to take precedence, indicated by the fact that orange females translated elevated corticosterone into current reproduction at the expense of survival, and offspring quantity vs. quality allocations were affected by corticosterone secondarily, after a longer duration of corticosterone exposure.

Changes to female post-laying mass were unaffected by corticosterone, indicating that corticosterone treatment caused females of both strategy types to gather energy for reproduction by increasing their rate of feeding during oogenesis (income breeding) rather than by mobilizing fat reserves into progeny (capital breeding; Jonsson, 1997). This suggests that the normally experienced egg size vs. number trade-off in this species is not mediated by resource availability per se, but instead, reproductive investment is limited by a third trait, energy acquisition rate, which is not maximized by females in the absence of chronically elevated corticosterone.

Feeding comes with a cost of increasing predation risk (Houston et al., 1993), likely reflected in the results for post-laying female survival, where corticosterone increased survival of lighter females and decreased survival of heavier females. Survival of females in good condition may have depended more heavily on antipredator behaviours such as remaining close to their refuge and high on rocks with good vantage points for approaching predators. However, survival of females in poor condition may have depended on bringing up their energy reserves by foraging. If corticosterone acts to increase foraging rate, which increases chance of predation, the survival benefits of any effect of corticosterone on energy gain through enhanced foraging rates may outweigh the associated survival detriments only for females in poorer condition. Because corticosterone acts to increase the rate of resource acquisition, it supplants the egg size vs. number trade-off with another trade-off; increased mortality during foraging vs. staying close to the refuge but not acquiring resources. Predator avoidance vs. energy acquisition may commonly be the primary trade-off in prey animals’ hierarchy of life history decisions, although long-term studies are required to infer evolutionary causation.

Corticosterone’s effects on reproductive decisions are strategy-specific, primarily enhancing offspring quality in K-strategists and favouring increased rates of reproduction in r-strategists. Orange females’ higher mortality and increased rate of reproduction in response to corticosterone may reflect an immediate effect of corticosterone treatment on enhanced feeding and reproduction rate, whereas orange females’ slower response to increase egg mass in response to corticosterone suggests that they only invested in offspring quality if reproduction was delayed by poor condition prior to treatment. Yellow females, in turn, may have delayed intensifying their energy acquisition rates in response to corticosterone until the final stages of oogenesis, when the extra input of energy could be directly translated into offspring quality. This delay would also minimize prelaying mortality as a result of predation exposure during foraging. This hypothesis is supported by observations that orange females are more likely to be spotted out foraging early in the breeding season than are yellow females, perhaps in response to endogenous corticosterone (BS, LL and LH; personal observations). Further studies on strategy-specific timing (with respect to reproductive stage) of daily foraging rates in response to corticosterone will determine the relative importance of this effect in directing or enhancing r- vs. K-resource allocation strategies in this and other systems. Further studies could also elucidate specific genetic, possibly endocrine, factors differing between r- and K-strategists that interact with corticosterone to direct strategy-specific effects as well as potential influences of variation in territory or mate quality on female allocation strategies in this system.

Results of this study demonstrate functional integration between the offspring quantity vs. quality trade-off, the survival vs. reproduction trade-off and the energy acquisition vs. predator avoidance trade-off. What appears to be a trade-off along a single dimension (the differences between r-strategist, orange and K-strategist yellow females in clutch size and egg size in this case) can be shown with experimental perturbation to actually represent a multidimensional trade-off involving behavioural interactions with environmental stressors, downstream effects of corticosterone on foraging strategies and reproductive suppression, female condition and predation risk. Under chronic corticosterone exposure for long periods and minimal predation risk, both types of females increased egg mass without decreasing clutch size, suggesting that they are not constrained to face a direct, functional trade-off on this level. Instead, orange and yellow females differ in how their hierarchy of multidimensional trade-offs is organized in response to stress, and these differences in hierarchical organization result in differential mortality, reproductive suppression and offspring quality. The existence of a polymorphism for overall trade-off organization suggests that physiological and genetic trade-offs may be more evolvable (in response to changes in constraints such as resource distribution or structural changes to morphology) than previously supposed.

Acknowledgments

The authors thank April Brand, Myra Brown, Bea Janez, Jacob Martinez, Andrew McAdam, Nicole Sandoval, Daniel Strain, Ryan Taylor and Elodie Vercken for assistance with lizards and eggs, and the Arbeilbedee family for permission to work on their land. The authors also thank Ammon Corl, Bruce Lyon, Mitchell Mulks and Dhanashree Paranjpe for comments on the manuscript. The present work was funded by NSF grants to BS and LH.

Ancillary