Age-related sex differences in body condition and telomere dynamics of red-sided garter snakes

Abstract

Life-history strategies vary dramatically between the sexes, which may drive divergence in sex-specific senescence and mortality rates. Telomeres are tandem nucleotide repeats that protect the ends of chromosomes from erosion during cell division. Telomeres have been implicated in senescence and mortality because they tend to shorten with stress, growth and age. We investigated age-specific telomere length in female and male red-sided garter snakes, Thamnophis sirtalis parietalis. We hypothesized that age-specific telomere length would differ between males and females given their divergent reproductive strategies. Male garter snakes emerge from hibernation with high levels of corticosterone, which facilitates energy mobilization to fuel mate-searching, courtship and mating behaviours during a two to four week aphagous breeding period at the den site. Conversely, females remain at the dens for only about 4 days and seem to invest more energy in growth and cellular maintenance, as they usually reproduce biennially. As male investment in reproduction involves a yearly bout of physiologically stressful activities, while females prioritize self-maintenance, we predicted male snakes would experience more age-specific telomere loss than females. We investigated this prediction using skeletochronology to determine the ages of individuals and qPCR to determine telomere length in a cross-sectional study. For both sexes, telomere length was positively related to body condition. Telomere length decreased with age in male garter snakes, but remained stable in female snakes. There was no correlation between telomere length and growth in either sex, suggesting that our results are a consequence of divergent selection on life histories of males and females. Different selection on the sexes may be the physiological consequence of the sexual dimorphism and mating system dynamics displayed by this species.

1. Introduction

Life-history strategies vary widely both between and within species. Such strategies describe how limited resources are used and prioritized [1,2], generating trade-offs between different physiological processes that mediate growth, reproduction and survival [3–5]. For example, organisms that ‘live fast’ are characterized by rapid growth and maturation, and high reproductive output, but age more quickly and have short lifespans [6,7]. Conversely, organisms that ‘live slow’ grow and mature more gradually and have lower reproductive output, but age more slowly and have longer lifespans [6,7]. Reproduction–longevity trade-offs are often difficult to detect within a population due to condition-mediated positive correlations between natural history traits [4,8]. However, there should be a link between condition, cellular maintenance and ageing. Body condition reflects the efficient collection, assimilation and deployment of resources and depends on the individual's capacity to cope with handicaps like infection, injury, parasitism and environmental stress throughout ontogeny [9–14].

As long-lived organisms age, they tend to experience reduced survival and reproductive output that may be mediated by condition ([15], but see, [16]). One mechanism linking differences in life histories, lifespans and ageing appears to be variation in telomere dynamics [17–19]. Telomeres are hexameric tandem repeat sequences of 5′-TTAGGG-3′ at the ends of chromosomes that typically shorten over the life of an organism due to repeated cellular divisions and damage caused by reactive oxygen species (ROS; [20,21]). Among species, telomere dynamics may covary with life-history strategies [22,23], and the rate of telomere attrition correlates with lifespan [17,24]. However, it is unclear whether short telomeres cause death or whether they are correlated with some other mechanism of senescence [18,25,26]. Body condition indices (BCI: body mass controlled for structural length) may be a useful measurement of somatic maintenance that is associated with longer telomeres (e.g. [18]).

Interspecific differences in telomere attrition are probably due to prioritizing cellular maintenance (e.g. DNA-repair) over other cellular functions [27,28], as autosomal mutations are negatively correlated with lifespan (mammals, [29]). DNA damage can lead to mutations, telomere loss and cellular senescence; thus, the maintenance of the genome probably explains telomere length stability in longer lived organisms [30,31]. To date, most studies of telomere dynamics and life-history strategies have focused on interspecific comparisons [17,19,22,25]. While these studies have yielded insight into telomere dynamics, elucidating the mechanisms underlying the observed trends is complicated by genetic variation between species. Studying organisms that exhibit intraspecific differences in reproductive tactics and/or life-history strategies provides a natural experimental scenario to study telomere dynamics while minimizing the noise of interspecific genetic variation. For example, females and males often exhibit sex-related differences in reproductive strategies and sexual selection [32–36], which may result in sex-specific telomere dynamics [24,37–39]. Thus, we sought to investigate telomere dynamics in a highly dimorphic species with well-characterized life history and reproductive strategies: the red-sided garter snake, Thamnophis sirtalis parietalis (a non-venomous colubrid).

Red-sided garter snakes are sexually dimorphic with respect to body size, with females growing approximately 30% longer, on average, than males [40]. In the Interlake populations of Manitoba, Canada, red-sided garter snakes hibernate for eight months in communal dens and emerge en masse in spring, to form large aggregations where males scramble to locate and mate with females [41,42]. Mating activity at the dens lasts approximately six weeks from late April through May [43] with some males mate-searching and courting for two to four weeks [41,44,45].

During the spring breeding season, male garter snakes are aphagous and have relatively high levels of corticosterone [46–50]. Courtship and copulatory plug production are energetically expensive [42,51,52], and males may lose 10% of body mass during two weeks of mate-searching, courtship and mating [44,50]. In other species, physiological stress and fasting lead to increased ROS production, the depletion of endogenous antioxidants, and increased cellular damage and senescence [53–61]. One of the hallmarks of male ageing is poor sperm performance, which is strongly influenced by oxidative and other physiological stressors (reviewed in [62]). Indeed, larger (and therefore older, [63]) male red-sided garter snakes have poorer sperm performance than smaller males [64], suggesting that these males undergo senescence in the wild.

In contrast with males, female garter snakes seem to prioritize growth and maintenance over short-term reproductive success. Females reach sexual maturity at 3 years of age, while males are sexually mature at 1 or 2 years [65]. Most females mate every year before migrating to feeding grounds [66], but they reproduce only when they have acquired sufficient body mass or ‘capital’, which is typically every other year [67,68]. Like most snakes, female garter snakes do not provide post-natal parental care [69]. Furthermore, female fecundity increases with body length [70–72] and, presumably, also with age because snakes exhibit indeterminate growth [63]. Biennial reproduction and increasing reproductive fitness with age may generate selection on increased cellular maintenance, body condition and growth in females. In this species, body condition is positively correlated with fat mass (Uhrig et al. 2012, unpublished data). With such life-history variation between the sexes, the red-sided garter snake is an exceptional model for investigating how different reproductive strategies and telomere dynamics interact, while minimizing the genetic variance that makes interspecific studies difficult to interpret.

We hypothesized that the sex-specific reproductive strategies of red-sided garter snakes would be associated with differences in age-related declines in telomere lengths. This study aims to determine: (i) the relationship between body condition, telomere length and age in garter snakes, and (ii) whether this relationship differs with sex. We predict that male garter snakes will experience greater telomeric attrition with age than females, due to the much more intense reproductive investment in males. Furthermore, if females are investing more in somatic maintenance than males, we expect females will maintain better body condition.

2. Material and methods

At the peak of breeding season (10 May 2015), we collected an excess of snakes by hand from mating aggregations with the aim to collect the full range of body lengths found at our Inwood, Manitoba, study site (males: N = 100; females: N = 50). We transported snakes to Chatfield research station, 16 km away, where they were weighed (±0.01 g) and measured for snout–vent length (SVL: ±1 mm) where we culled our sample to ensure an equal distribution of sizes for each sex. We selected the four longest and four shortest animals of each sex and an even distribution of intermediate sizes, obtaining a final sample of 42 males and 30 females (figure 1a), the remaining 78 animals were returned to the point of capture the next day. All animals were adults; juveniles are only rarely found at den sites (R.T.M. > 25 years of personal observation; [41,70]). Blood (less than 0.1 ml) for telomere analysis was taken from the caudal vein, added to 300 µl of RNAlater and frozen (−30°C) until DNA extraction. Approximately 1 cm of tail tissue was collected for skeletochronological ageing; see expanded methods in electronic supplementary material, document 1 for more details.

Figure 1.

(a) Age (years) and sex predicted body size (ln(snout-to-vent length): ln(SVL)): older animals were longer and females were significantly longer than males of the same age. Open circles indicate males and solid triangles indicate females (note: for clarity with overlapping data points, male data are offset slightly to the right). The least-squares regression lines were calculated separately for females (solid line, r = 0.400) and males (dashed line, r = 0.445). (b) Body condition (BCI) differed with age and sex and there was a significant sex × age interaction (p = 0.004). Females had higher BCI than males and BCI decreased with male age but not females. Open circles indicate males and solid triangles indicate females. The least-squares regression lines were calculated separately for females (solid line, r = 0.015) and males (dashed line, r = −0.479). The diagonal hatched box (age = 0.00 to 2.03), is the age-range through which BCI did not differ between females and males as determined by the Johnson–Neyman procedure [73]. (Online version in colour.)

(a). Skeletochronology/histology

Individual age was estimated by a modified version of the technique described by Waye & Gregory ([74,75] 1999) and Clesson, Bautista [76]. Vertebrae were examined microscopically and the number of growth rings was identified for each animal; see electronic supplementary material, for more details.

(b). Quantifying telomere length

Telomere length was measured using real-time quantitative PCR (qPCR) as we have done previously [77] using the 18S ribosomal RNA (18S) gene as the non-variable in copy number reference gene [77–79]; see electronic supplementary material more details.

(c). Statistical analyses

We calculated two measures of BCI. In both cases BCI is the standardized residuals (mean = 0; s.d. = 1) from linear regressions of ln(body mass) as a function of ln(SVL) [80]. We ran this linear regression model once with males and females pooled, and it was clear that females had much higher BCI than males. Therefore, it was more biologically relevant to generate BCI for each sex separately using a separate regression model for each sex, thus creating BCI specific for each sex (ssBCI) to account for differences in allometry [80]. Growth was calculated as size (SVL)/age. Visual inspection of regression plots for male telomere length given age suggested a curvilinear relationship as has been described in many taxa, including squamate reptiles [25,38,81–84], and F-tests we used to formally test the goodness of fit for first-order versus quadratic regressions. We used ANCOVA to test for age-specific sex differences in telomere length and body condition. When we found a significant sex by age interaction we used the Johnson–Neyman (J-N) procedure to determine ages where the sexes differed in condition [73]. All analyses were conducted in SigmaPlot 13.0, except the J-N procedure which was conducted in MS Excel on the spreadsheet provided as electronic supplementary material in White [73]. See electronic supplementary material, for more details.

3. Results

(a). Skeletochronology, size and body condition

Age and sex predicted body size (SVL): older animals were longer and females were significantly longer than males of the same age (ANCOVA: sex × age p = 0.487 (dropped from model): R² = 0.366; age: F_1,69 = 14.636, p < 0.001; sex: F_1,69 = 16.569, p < 0.001; figure 1a). The shape of the age distributions was not different between the sexes (Kolmogorov–Smirnov test: D = 0.205, p = 0.412) and females in our sample were significantly older than males (F_1,70 = 6.384, p = 0.014; mean (range), females: 4.3 years (2–9 years); males: 3.5 years (2–6 years)). There was a significant sex × age interaction on BCI (ANCOVA: R² = 0.542; age: F_1,69 = 5.403, p = 0.023; sex: F_1,69 = 0.003, p = 0.953; sex × age F_1,69 = 8.695, p = 0.004), suggesting that females and males differentially maintain body condition as they age. Because of the significant sex × age interaction, we computed the region of non-significance for the age-effect on BCI between the sexes (−8.454 to 2.029 yrs) using the J-N procedure [73]. This approach demonstrates that BCI differed between the sexes at ages greater than 2.03 years, which included most of the snakes in this sample (figure 1b; note age values < 0 are meaningless and omitted from the figure). Given the profound sex-differences in body condition, we recalculated BCI for each sex with separate regressions (i.e. ‘sex-specific’ BCI) and reran the analysis. We still found a significant sex × age interaction (p = 0.023), which revealed that sex-specific body condition tends to increase with age in females, but decreases with age in males (electronic supplementary material, figure S2). We used this sex-specific BCI (ssBCI) to explore the relationship between body condition and telomere length in further analyses.

(b). Telomere length and age

Telomere length was shorter in males than females (F_1,70 = 7.288, p = 0.009). The relationship between telomere length and age was different for males and females. Age did not predict telomere length in females (females: simple linear regression R² = 0.000, F_1,29 = 0.005, p = 0.945: quadratic regression; R² = 0.000, F_2,28 = 0.050, p = 0.951; figure 2a). However, in males, telomeres shorten with age, a relationship better fit by quadratic regression than linear regression (test of first-order = null hypothesis versus quadratic: F_2,41 = 5.538, p = 0.024: simple linear regression: R² = 0.108, F_1,41 = 4.856, p = 0.033; quadratic regression: R² = 0.219, F_2,39 = 5.472, p = 0.008; figure 2b).

Figure 2.

The relationship between natural log of blood telomere length and age in years was different for females (a) and males (b). (a) Age did not predict telomere length in females (females: simple linear regression; r = 0.013, F_1,29 = 0.005, p = 0.945: quadratic regression; r = 0.067, F_2,29 = 0.050, p = 0.951). (b) However, in males, telomeres shorten with age, which is better fit by quadratic regression than a linear regression (test of first-order = null hypothesis versus quadratic: F_2,41 = 5.538, p = 0.024: quadratic regression: r = 0.468, F_2,41 = 5.472, p = 0.008).

(c). Telomere length, body size and growth

Although age and SVL were directly related in both sexes (see above), SVL and telomere length were not related (ANCOVA sex × SVL p = 0.538 (dropped interaction): R² = 0.095; sex: F_1,69 = 5.900, p = 0.018; SVL: F_1,69 = 0.057, p = 0.813). Separate analyses to test for a quadratic relationship, as was found in males for age and telomere length, showed no evidence for a relationship between SVL and telomere length in either sex (females p = 0.200; males p = 0.229). Finally, growth (size/age) was not significantly associated with telomere length (either SVL/age: R² = 0.052, p = 0.085; residual SVL given age: R² = 0.031, p = 0.137; or sex-specific residual SVL given age: R² = 0.001, p = 0.766).

(d). Telomere length and body condition

Sex-specific body condition (ssBCI) and blood telomere length were positively correlated (R² = 0.131, F_1,70 = 10.564, p = 0.002), and, although females had higher ssBCI than males, the relationship between ssBCI and telomere length was the same for both sexes (ANCOVA sex × ssBCI: p = 0.510 (dropped interaction): R² = 0.145; sex: F_1,69 = 7.601, p = 0.007; ssBCI: F_1,69 = 4.005, p = 0.049; figure 3).

Figure 3.

Combined sex-specific body condition (standardized residuals from separate regressions of body mass given snout-to-vent length for each sex) and natural log of blood telomere length were positively correlated (r = 0.602). Females had higher BCI than males, but the relationship between BCI and telomere length was the same. Open circles indicate males, and solid triangles indicate females. The least-squares regression lines were calculated separately for females (solid line, r = 0.362) and males (dashed line, r = 0.506).

4. Discussion

Sex-differences in aging may result from sex-specific optimization of investment to reproduction and somatic maintenance in response to the challenges of different life-history strategies between the sexes. We have shown that body condition positively correlates with telomere length in both sexes of red-sided garter snakes, which supports our assertion that body condition is an intuitive measure of somatic investment. However, the relationship between body condition and age differed strikingly between sexes, with females maintaining their body condition with age, while condition decreased with age in males. Likewise, telomeres were exponentially shorter in older male garter snakes, while the telomere lengths of females were independent of age. Nonlinear relationships between telomere length and age have been shown in several taxa (e.g. [81,84]), and is consistent with an exacerbating cycle of cellular damage and increased dysfunction seen in ageing humans [85]. Females had the longest telomeres and were the oldest individuals in our sample, suggesting they live longer than males in this population. These results support our prediction that males experience greater telomere loss with age due to prioritization of current reproduction over cellular maintenance and longevity. Overall, the decrease in both body condition and telomere length in males with age suggests that they senesce at an earlier age than females.

Telomere shortening has been implicated as a cost of reproduction in several species. For example, in blue tits (Cyanistes caeruleus), when brood size was experimentally increased, parents experienced a decrease in blood telomere length, with males suffering from greater telomere loss than females [86]. Relative reproductive success seems to result in greater telomere attrition in common terns (Sterna hirundo) [87]. For both male and female Atlantic silversides (Menidia menidia), gonadal somatic index (GSI: gonad mass relative to total body mass) was negatively correlated with telomere length and lifespan [88]. These studies suggest increased reproductive investment comes at a cost of telomere attrition.

Studies of telomere dynamics are rare in reptiles and there are only two reports on snakes. Bronikowski [89] reported telomere lengths for male wandering garter snakes (Thamnophis elegans). Wandering garter snakes are an interesting species for studying telomere dynamics because in the mountains of Northern California there are two eco-types with very different life histories: one short lived ‘meadow’ eco-type and a long-lived ‘lakeshore’ eco-type [90–92]. As in our study, Bronikowski [89] showed declining telomere length with male age (up to 12 years of age, based on skeletochronology), but was unable to find among eco-type differences, and did not report telomere lengths for females. In water pythons (Liasis fuscus) of Northern Territory, Australia, females have longer telomeres than males [82], similar to our study. Furthermore, telomere length increased from hatching to 4 years of age, but declined very slightly with age in both sexes up to 18 years of age [82].

(a). Why might selection on telomere dynamics differ between male and female garter snakes?

Our study is observational and cross-sectional, so our causal interpretation of the sex-specific differences in the relationship between age and telomere length is necessarily tentative. In Manitoba's Interlake region, winter temperatures often hover around −40°C for weeks and, because snow provides insulation from the cold, there are likely cryptic mass fatalities deep within dens during years of light snowfall [70]. The snakes' brief three to four month active season begins and ends with chance freezes and floods that lead to mass mortality events that are likely to generate selection on rapid growth and early maturity in both sexes [70,93]. Mortality due to these stochastic events is usually not consistently biased toward either sex and the adult sex-ratio is 1 : 1 [41,70,93]. Predation and road kills are not sex biased either [93]. However, a mass mortality event could differentially affect size classes among sexes. For example, a winterkill event in 1998–1999 shifted the size distribution toward smaller animals in subsequent years in both sexes, but the largest females were most strongly affected [93]. Small males, and to a lesser extent large females, are more likely to be trapped and suffocate in large mating aggregations (more than 500 animals) [94]. Such events could cull a size class or spare only old females with the longest telomeres, generating results similar to ours. Nevertheless, we have not witnessed similar events in our yearly visits since 1999, thus other explanations may better fit our results.

Males engage in energetically expensive reproductive behaviour annually, while most females generally reproduce biennially. Although male size affects mating success when a single pair of males competes for copulation, the effect is small to non-existent in the largest aggregations at the den sites, reducing selection for increased male size [95,96]. The largest females, however, are able to reproduce annually, leading to greater fecundity and generating higher selection on female growth and longevity [70–72,97]. Females in Interlake populations seem to have higher reproductive output given female size than populations farther south in less harsh climates with longer feeding/growth seasons [70,98]. Therefore, selection on cellular maintenance and longevity are likely to be stronger in females than males because the costly mechanisms that prevent telomere loss are balanced by increasing fecundity with age and size in females, but have fewer benefits for males.

(b). What physiological mechanisms might explain sex-specific telomere attrition?

We do not know the specific mechanisms that lead to sex-differences in telomere length, but there are several non-mutually exclusive hypotheses to explain our results. For these ectotherms, body temperature and metabolic rate are very low during winter brumation (approx. 1°C [99]) and only rise in late April when the ground warms. Both sexes enter winter hibernacula at the same time [70], but males, on average, emerge earlier than most females. Therefore, body temperature and metabolic rate will be lower, for slightly longer, in females than males. Lower body temperature associated with torpor is correlated with positive effects on telomere length and somatic maintenance in some mammals (e.g. [100]).

High levels of corticosterone experienced by males during the mating season [49] may increase metabolism, but also may increase mitochondrial ROS production, DNA damage and telomere erosion [31,56,101,102]. The high energetic demands of courtship and mating of aphagous males [51] probably limits the resources that can be allocated to DNA-repair mechanisms, limiting the chance for telomere repair [103]. For example, the increased male–male competition among male rhesus macaques (Macaca mulatta) is correlated with DNA oxidative damage (8-OHdG) and shorter lifespan [57]. In red-sided garter snakes, the energy for antioxidant synthesis, DNA-repair and telomere maintenance is limited by male fasting [27,54,55,59,104]. Fasting itself may increase oxidative stress [53,58,60,61]. Fasting increases the generation of mitochondrial ROS and lipid peroxidation in rats (Rattus norvegicus) [61]. Fasting male northern elephant seals (Mirounga angustirostris) exhibit increased oxidative damage to DNA and lipids [58]. Given the stochastic mortality, weak sexual selection on male size, and oxidative stress induced by during energetically costly courtship and mating while fasting, selection to mitigate damage by ROS via investment in cellular maintenance and growth may be weak in male red-sided garter snakes. Weak selection for enhanced cellular maintenance might explain both the reduction of body condition and telomere length with age. This may be the consequence of selection for a live fast, die young strategy in males.

Females were in better body condition than males in our study, which generally indicates they have larger energy stores than males [105]. In brown tree snakes, Boiga irregularis, and in female red-sided garter snakes, this additional energy reserve correlates with lower levels of corticosterone [106,107], potentially leading to lower stress overall and more stable telomere length [28,108]. Furthermore, having greater energy reserves may allow for greater expenditure on antioxidants and cellular repair. Species of snakes that live longer are capable of producing a stronger response to DNA damage by activating repair mechanisms and experience lower levels of mitochondrial ROS, which presumably generates less oxidative damage to DNA [89,109]. We show that female T. sirtalis parietalis have a greater lifespan than males and may potentially use mechanisms similar to those of other snakes to maintain genome stability and telomere length. The underlying mechanisms causing the sexual dimorphism may provide explanations for sex-specific differences in telomere length.

Sexual size dimorphism varies greatly across taxa, and trends associated with dimorphism, lifespan and telomere attrition are not consistent [24,110]. For garter snakes, the difference in size between males and females seems to be controlled by testicular androgens suppressing growth in males [111]. Testosterone can reduce cellular resistance to free radicals [112], leading to increased DNA damage and telomeric attrition [19,113]. In the closely related red-spotted garter snake, Thamnophis sirtalis concinnus, females treated with an oestrogen receptor antagonist, tamoxifen, experienced a decrease in growth rate [114], suggesting that oestrogen plays a role in the sexual size dimorphism observed in T. sirtalis parietalis. Oestrogens act as antioxidants and/or stimulate endogenous antioxidant and cellular repair mechanisms [115–117] potentially reducing ROS and leading to the telomeric stability observed in this study and in females across other taxa [24,38]. The most energetically demanding component of reproduction for female garter snakes is the production of yolk proteins (i.e. vitellogenesis) [118]. There is evidence that the yolk protein, vitellogenin, may act as an antioxidant, [119–123] reducing DNA damage, telomere attrition and cellular senescence at a time when cellular respiration and ROS production are highest. Thus, selection acting on the mechanisms that increase female growth and provisioning of offspring seem to also favour antioxidant production, a reduction in oxidative stress and cellular repair involved in slowing the ageing process.

In this cross-sectional study, we investigated differences in telomeres within a single species. We found that telomere dynamics is strongly linked with sex and therefore life-history strategies. Sex-specific telomere dynamics may be tightly linked to selection on males for early reproduction and costs associated with yearly energetic investment in courtship and mating while fasting. By contrast, females have biennial reproduction and investment in somatic maintenance has a fitness pay-off of greater fecundity with increasing size later in life. Future studies should include longitudinal data, increased sampling of the largest size classes, the measurement of telomerase activity, general DNA damage, and antioxidant production throughout the entire active season, to assess our hypothesis that females live longer by investing more in cellular maintenance and repair than males.

Ethics

Procedures performed on animals were approved by Oregon State University (IACUC ACUP-4317) and the research was conducted under permit from Manitoba Conservation (WB16264).

Data accessibility

Data have been uploaded to the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.jv463 [124].

Authors' contributions

All authors made significant intellectual and material contributions to this paper.

Competing interests

The authors declare they have no competing interests.

Funding

This material is based in part upon work supported by the National Science Foundation (DBI-1308394 to C.R.F.) and University of Sydney (Animal and Veterinary Biosciences Fellowship to C.M.W.) and University of Minnesota Morris Division of Science and Mathematics (to H.L.W.).