Within-sample variation in snowshoe hare faecal glucocorticoid metabolite measurements

Faecal glucocorticoid metabolite (FGM) concentrations are used increasingly as a non-invasive measure to index physiological stress experienced by diverse taxa. However, FGM may not be evenly distributed throughout a faecal mass or faecal pellet group. Moreover, within-sample variation in FGM measurements associated with different sampling and/or processing techniques is rarely reported despite potentially having important implications for inferring stress levels in free-ranging wildlife. Using a captive collection of snowshoe hares (Lepus americanus), we (i) assessed repeatability of FGM measurements (i.e. precision) from two processing techniques (measurements derived from dividing whole pellet groups into equal proportions prior to processing [G1], measurements from subsamples derived from thoroughly homogenized whole pellet groups [G2]) and (ii) conducted a power analysis to estimate sample-size requirements for detecting statistically significant differences in FGM concentrations at a population level. Our results indicate that the mean percent coefficient of variation (%CV) for within-sample FGM variation was slightly higher for G1 (%CV = 35, range 13.45-65.37) than for G2 (%CV = 23, range 7.26-47.94), though not statistically significant (two sample t-test, n = 8, t = 1.57, P = 0.16). Thus, FGM is relatively evenly distributed within snowshoe hare faecal pellet groups. However, subsampling from homogenized whole pellet groups may be more appropriate when the sampling time frame is less controlled (e.g. multiple defecation events) because a subsample derived from a homogenized whole pellet group might be more representative of the animal's 'average' physiological state compared to FGM concentrations derived from a few haphazardly selected faecal pellets. Power analysis results demonstrated the importance of a priori consideration of sample sizes. Relatively small effect sizes (e.g. <20%) may require sampling that is logistically and/or cost prohibitive. Yet for many situations of ecological or conservation interest, treatment effects may be substantial (>25%) and thus moderate sample sizes may be sufficient for testing research hypotheses regarding changes FGM concentrations.