A critical time in the life of a male occurs at reproduction when his behavior, physiology, and resources must be brought to bear for the central purpose of his life. We ask whether reproduction results in dysfunction of the stress axis, is linked to life history, and causes senescence. We assessed if deterioration in the axis underlies variation in reproductive lifespan in males of 5 species of North American ground squirrels whose life history varies from near semelparity to iteroparity. The most stressful and energy-demanding time occurs in spring during the intense 2–3 week breeding competition just after arousal from hibernation. We compared their stress axis functioning before and after the mating period using a hormonal challenge protocol. We found no evidence of stress axis dysfunction nor was there a relationship between reproductive lifespan and stress axis functional deterioration.  Moreover, there was no consistent relationship between free cortisol levels and downstream measures. Thus, stress axis function was not traded off to promote reproduction and conclude that it is a pre-requisite for life. Hence, it functions as a constraint and does not undergo senescence.

Study areas and field procedures 

We trapped prebreeding (within a few days of first emerging from hibernation) and postbreeding males (2–3 weeks later, same site). For the 3 western species (AGS, CGS, and RGS) logistics (i.e. ~2500 km between the AGS and RGS sites), asynchronous timing of emergence and reproduction, and a narrow window of reproduction within a species permitted us to trap only them in 2007 and the 2 eastern species (FGS, TLGS) in 2008.  In 2007, we trapped RGS at Kinsella, Alberta (53°N 111°33’W) on March 25–27 and April 15–16, AGS at the Pelly River Ranch, Yukon (62°50’N 137°18’W) on April 9–12 and April 27–30, and CGS at Kananaskis, Alberta (51°2’N 115°2’W) on April 22–24 and May 7–9, respectively. In 2008, we trapped TLGS on April 15–19 and May 5–11 and FGS on May 6–11 and May 28–30 near Portage La Prairie, Manitoba (TLGS at 50°7’N 98°23’W, FGS at 50°12’N 98°14’W). 

Animals were trapped by placing burrow traps or cage traps (Tomahawk #102, Tomahawk Live Trap Company, Tomahawk, WI, USA) in or next to burrows that animals were seen to have entered. Traps were monitored at least every 30 minutes and captured males were then kept in cage traps covered with a pillowcase in a shaded, quiet location until trapping was completed. We started trapping soon after animals emerged (07h00–09h00) and continued trapping for at most 4 hours or until we had as many animals as we could process at one time (6–9 animals). 

Field Hormonal and Sampling Procedures

Trapped animals were brought to a field laboratory, placed in a cool, quiet location, covered with pillowcases, and left for at least 1 hour to habituate before initiating the hormonal challenge procedure. One animal at a time was removed from its trap, weighed, and anesthetized with isoflurane (IsoFlo, Abbott Laboratories, Saint Laurent, QC, Canada).  We took an initial 0.6 mL blood sample (BASE) and injected 0.1 or 0.4 mg/kg dexamethasone (DEX) (Sabex, Montreal, Canada), then returned the animal to the covered trap and processed the next animal.  From the BASE sample, we obtained 4 subsamples: 0.3 ml for a complete blood count (Vita-tech Veterinary Laboratory Services, Markham, ON, Canada); 75 µl microhematocrit tubes for hematocrit (packed red blood cell volume, %); a drop for glucose levels (mg/dL, FreeStyle glucose meter, Abbott Laboratories, Alameda, CA, USA); and the rest for hormone and free fatty acid measurements. The latter sample was centrifuged at 8800g for 5 min, and the plasma from the hematocrit tubes was pooled with it and frozen at -20°C.

The DEX tests the negative feedback to the pituitary and inhibits ACTH release. As we found no difference in DEX response to the different doses in 2007 (unpublished data), we administered 0.1 mg/kg to all animals in 2008. Two hours after the DEX injection, we anesthetized the animal again and took a 0.2-0.3 mL blood sample (“DEX” bleed) and then injected 4 IU/kg of adrenocorticotropic hormone (ACTH; Synacthen Depot, Novartis Pharmaceuticals Canada Inc, Dorval, Quebec) into the thigh muscle to directly stimulate adrenal GC production.  Post-ACTH blood samples of 0.2 mL each were taken at 30, 60 and 120 minutes (“P30”, “P60” and “P120” bleeds). Glucose was measured at each sampling point. Blood samples were centrifuged and plasma collected and frozen at -20°C and, on return to the university, stored at -80°C. After the P120 bleed, the animal was euthanized by anesthetic overdose and decapitated. 

On necropsy, we dissected out and weighed both adrenals (nearest mg), abdominal fat (fat that was easily removed from the abdominal cavity and did not include fat in the mesentery: nearest 0.1g), both testes (nearest 0.01g), and eye lens. Adrenal mass was a rough indicator of adrenal capacity based on the expectation that adrenal hypertrophy would occur from the prebreeding to the postbreeding period reflecting greater capacity to produce GCs.  Abdominal fat was a measure of a key energy reserve and expected to decline markedly. Testes mass was a measure of the capacity to produce sperm and testosterone and expected to decline markedly from the prebreeding to the postbreeding period. Age may be a key determinant in how individuals respond to stress and we attempted to age our squirrels by measuring their eye lens weights and by sectioning their femurs and jaws.

To quantify the intensity of inter-male aggression during the breeding season, we examined the skin externally and internally for evidence of wounding. The degree of wounding was scored according to the number of punctures (bite marks that penetrated the skin, usually only visible from the inside of the hide) and externally visible wounds (tears in the skin that exposed the muscle below or had scabbed over). Our scoring categories were:  “None” for no wounds, “Minimal” for <10 punctures and <2 cm2 of externally visible wounding, “Moderate” for ≥ 10 punctures and/or >2 cm² but <7 cm² of externally visible wounding, and “Severe” for ≥ 7 cm² of externally visible wounding and/or significant damage to muscle, testes, or eyes. Because FGS were not abundant, we did not euthanize the prebreeding animals. As a result, their wound scores were based solely on an external examination. Thus, we also do not have prebreeding adrenal, abdominal fat, or testes mass for FGS.

Laboratory procedures

We measured total cortisol levels by radioimmunoassay (RIA) in all species. For FGS and TLGS we switched to a commercial cortisol kit (Diasorin GammaCoat Cortisol Kit, Stillwater, MN, USA) using kit instructions. To cross-validate levels between the two RIA methods, we used 18 plasma samples (levels ranging from 22 ng/mL to 452 ng/mL) and obtained the cortisol levels measured by our initial protocol and by the kit protocol. These were highly correlated (r² = 0.97), though the Diasorin kit tended to give values about 7 ng/mL higher than our initial protocol. This difference might be attributable to loss from the dichloromethane extraction procedure in our initial protocol, which was not used with the Diasorin kit.

Initial tests revealed that cortisol is the dominant GC in RGS and CGS, with corticosterone being undetectable. A similar pattern was found in AGS. However, we found that TLGS had corticosterone levels that were typically 10–20% those of their cortisol levels, and that FGS had corticosterone levels around 30–40% of their cortisol levels. Thus, in both of these species, we measured corticosterone levels. Because plasma samples were limited for some individuals, we only measured corticosterone levels in 5 prebreeding and 5 postbreeding individuals from both species, for all 5 hormone challenge plasma samples (BASE, DEX, P30, P60, and P120). We calculated corticosterone as a percentage of total GC at each bleed in these animals. We used these means to estimate amount of corticosterone in the remaining animals.

To estimate free GC concentrations (i.e. hormone not bound to CBG) we measured maximum corticosteroid binding activity (MCBC). The equilibrium dissociation constants (±SE) are needed for calculating MCBC and for AGS, RGS, CGS, TLGS and FGS these were 4.04 ± 0.42 nM, 5.4 ± 0.56 nM, 4.73 ± 0.29 nM, 17.8 ± 2.3 nM, and 26.9 ± 10.0 nM, respectively (unpublished data). For TLGS and FGS, we assumed that the dissociation constants of CBG for cortisol and corticosterone were equal and that the two GCs could be treated as a single GC pool.

To assess the overall adrenal capacity to respond to the ACTH injection, we calculated the area under the free GC response curve (the AUC). AUC was measured from the DEX bleed through to the last bleed, using the DEX bleed GC concentration as a baseline. It provided an integrated measure of the animals’ sensitivity and capacity to produce GC. To measure the “intrinsic restraint” of the HPA axis, we used the ratio of the maximum free GC level in response to ACTH to the BASE bleed free GC. Because ACTH directly stimulates the adrenals, the maximum free GC levels are not constrained by lack of drive from the hypothalamus or pituitary or by negative feedback. In contrast, the free GC level at BASE was the result of ongoing stimulation of the HPA axis as a result of capture stress but modulated by negative feedback of the GCs at the level of the pituitary. The ratio of these two levels (i.e. maximum/BASE) represents a measure of the degree to which the HPA axis produces a less-than-maximal GC response at the BASE bleed (i.e. “intrinsic restraint”). The higher the ratio, the stronger the intrinsic restraint.

Testosterone levels (ng/mL) were measured. The assay had a mean recovery of [1,2,6,7-3H] testosterone added to plasma of 96.5 ± 0.7% (range 92–102%), with a detection limit of 10 pg/25 mL. It had an intra- and inter-assay coefficient of variation of 5% and 7.1%, respectively. Free fatty acids (FFA) were measured; they had an intra- and inter-assay coefficient of variation of 6.6 and 10.0%, respectively.

Statistical analysis

All data were analyzed using SAS 9.2 (SAS Institute Inc., Cary, NC, USA). For simple comparisons among species, we used one-way ANOVAs and the conservative Tukey-Kramer multiple comparison post-hoc test. For our regressions of postbreeding GC levels against mean reproductive lifespan and probability of surviving more than one breeding season (PROC GLM), we did not test for or control for phylogenetic non-independence.  In part, our sample size precluded using phylogenetic contrasts:  tests for phylogenetic independence perform poorly on small datasets and if our species were not independent, our sample size would be reduced by one in order to do the contrast.  Our decision is further justified by the fact that we selected very closely related species, and because previous comparative studies of GC levels have found that much more diverse species groupings are nonetheless phylogenetically independent.

For the prebreeding and postbreeding comparisons, data were first examined for normality. Where significantly non-normal data could not be normalized with a transformation, we used the Wilcoxon-Mann-Whitney non-parametric two-sample test using the EXACT option to generate a Monte Carlo-based exact P-value. Because we released prebreeding Franklin’s ground squirrels and some of our postbreeding samples came from recaptured individuals, we analyzed differences between pre- and postbreeding periods using a mixed model in SAS PROC MIXED, with subjects as a repeated measure. All other species comparisons used PROC TTEST; where pre- and postbreeding data had non-homogeneous variances based on the folded-F test, we use Satterthwaite-adjusted degrees of freedom for the t-tests (which can result in degrees of freedom that are not whole numbers). Unless otherwise specified, we present means with 95% confidence intervals.