Patterns and potential drivers of intraspecific variability in the body elemental composition of a terrestrial consumer, the snowshoe hare (Lepus americanus)

Intraspecific variability in ecological traits is widespread in nature. Recent evidence, mostly from aquatic ecosystems, shows individuals differing at the most fundamental level, that of their chemical composition. Age, sex, or body size may be key drivers of intraspecific variability in the body concentrations of carbon (C), nitrogen (N), and phosphorus (P). However, we still have a rudimentary understanding of the patterns and drivers of intraspecific variability in chemical composition of terrestrial consumers, particularly vertebrates. Here, we investigate the whole-body chemical composition of snowshoe hare Lepus americanus, providing one of the few studies of patterns of stoichiometric variability and its potential drivers for a terrestrial vertebrate. Based on snowshoe hare ecology, we expected higher P and N concentrations in females, as well as in larger and older individuals. We obtained whole-body C, N, and P concentrations and C:N, C:P, N:P ratios from a sample of 50 snowshoe hares. We then used general linear models to test for evidence of a relationship between age, sex, or body size and stoichiometric variability in hares. We found considerable variation in the C, N, and P concentrations and elemental ratios within our sample. Contrary to our predictions, we found evidence of N content decreasing with age. As expected, we found evidence of P content increasing with body size. As well, we found no support for a relationship between sex and N or P content, nor for variability in C content and any of our predictor variables. Despite finding considerable stoichiometric variability in our sample, we found no substantial support for age, sex, or body size to relate to this variation. The weak relationship between body N concentration and age may suggest varying nutritional requirements of individuals at different ages. Conversely, P’s weak relationship to body size appears in line with recent evidence of the potential importance of P in terrestrial systems. Snowshoe hares are a keystone herbivore in the boreal forest of North America. The substantial stoichiometric variability we find in our sample could have important implications for nutrient dynamics in both boreal and adjacent ecosystems.


Introduction
consumers' elemental composition varying as a function of an individual's age, sex, or body size with body size could influence the overall condition of an individual -which ultimately determines its fitness and nutritional value for its predators (Stevenson & Woods, 2006). In a strongly P-limited 137 environment like the boreal forest, larger individuals could indeed show higher concentrations of P. 138 From all of the above it follows that, during an individual's ontogenic development, its content content of snowshoe hares increases with increasing body size and as individuals grow older. We 143 also expect (2) female hares to have higher content of limiting nutrients, N and P, than males, 144 due to the higher reproductive costs. At the same time, we investigate the relationship between 145 organismal concentration of limiting nutrients, such as N or P, and an individual's body condition. 146 In this case, we expect (3) snowshoe hares in better condition to have higher concentrations of 147 N, P or both, at all life stages. We present one of the first assessments of whole-body elemental 148 composition of a small terrestrial mammal and discuss how intraspecific stoichiometric variability 149 might influence trophic dynamics and ecosystem processes.  Arm (NL, 47°31 ′ 00 ′′ N, 53°40 ′ 00 ′′ W) and Long Harbour (NL, 47°25 ′ 46 ′′ N, 53°51 ′ 30 ′′ W). In the 179 laboratory, we thawed and weighed each specimen to the closest 0.1 g. We collected data on total 180 body length, left hind foot length, and skull length and width for each hare to the closest mm, 181 repeating each measurement 3 times and using their arithmetic mean in all subsequent analyses 182 (see Supplementary Information section S1.2).

183
Like rodents, the teeth of lagomorphs grow continuously during their life, making conventional 184 aging techniques based on dentine and cement inapplicable (Morris, 1972). To account for this, we 185 aged our specimens using a mixed approach involving counting bone tissue growth lines deposited 186 after each winter in the mandibular bone. We used an ageing method developed for mountain hares 187 Lepus timidus to select the area of the bone from which to count the growth lines (Iason, 1988).

188
For all 50 snowshoe hares in our sample, we extracted the complete mandibular bone, cleaned it of 189 all soft tissues, and shipped the clean bones to Matson's Laboratory (Manhattan, MT, USA) for 190 age determination (see SI section S1.3). 191 We determined specimen sex using a DNA-based approach (Shaw et al., 2003; see SI section S1.4). As the snowshoe hare genome is not yet completely sequenced, we used published 193 primers for the European rabbit Oryctolagus cuniculus to amplify the genetic material extracted 194 from our specimens and from two control snowshoe hares of known sex (Fontanesi et al., 2008). In  Axis regression (Peig & Green, 2009; see SI section S1.5). The SMI formula is: whereM i is the SMI of individual i, M i is its body weight, L i is the linear measure of body size of was, compared to what it should be.

216
As the SMI is sensitive to the length measurement used to calculate it, we ran a separate set of 217 models using a SMI produced using skull length, which also showed a strong relationship with body weight (see SI section S1.5). Furthermore, we considered average body length as a separate estimate 219 of the effect of body size on the C:N:P stoichiometry of snowshoe hares. We calculated average body 220 length of individual snowshoe hares by taking the arithmetic mean of the three measurements of 221 total body length we collected from each specimen, and used this value in all subsequent analyses. homogenized material to produce 10 g of dry material for elemental composition determination. 234 We transferred all ground samples to glass vials and stored them in desiccators to prevent moisture 235 accumulation and mold formation. 236 We sent the 50 dried, whole-body samples to the Agriculture and Food Laboratory (AFL) at were run in duplicate to assess within-sample variability in %P (see SI section S1.6). In addition, 254 to capture variability within individuals due to our homogenization protocol, we selected 5 random 255 specimens for which we sent 2 additional samples (n=10) of the homogenized paste to AFL (see SI 256 section S1.6). Upon receiving the results back from AFL, to obtain C:N:P stoichiometry and molar 257 ratios for each hare, we calculated each hare's dry body weight and converted the concentration of 258 each element to molar mass using atomic weights.

277
Snowshoe hares in our sample varied in age between 0 ("young-of-the-year") and 6 years old, the 278 majority (74%) being between 0 and 1 years old. Only one individual, a female, was 6 years old.

279
Males were more common (31 out of 50) than females (19 intercept-only model was within 2 ∆AICc of the top ranked model (Table 1). For %P, the two top 292 ranked models included relative body condition and average body length, respectively (Table 1).

293
%P was positively related to relative body condition (R 2 =0.073; Fig. 2) and average body length 294 (R 2 =0.047). Again, evidence for these relationships is weak as the intercept-only model was the 295 third-best performing model and within 2 ∆AICc of the top ranked models (Table 1). We also 296 observed a qualitative pattern of higher %P among older males (Fig. 3), but found no statistical 297 support for it (Table 1). For %C, the top ranked model was the intercept-only model, which 298 provides no evidence of a relationship between variation in %C and age, sex, or body size of 299 individuals (Table 1).

300
For the stoichiometric ratios, the top ranked model for C:N included only age, which had a 301 positive relationship with C:N ratio (R 2 =0.074; Table 2). For this relationship too, evidence is 302 weak as the intercept-only model was within 2 ∆AICc of the best-performing one. We found no The positive trend for P is evident, and is weakly supported by the results of our modeling. Conversely, there is no visual evidence of a relationship between %C or %N and relative body condition, which is further confirmed by the results of our modeling (Table 1). Solid lines are ordinary least square regression lines, shaded areas represent 95% confidence intervals around them.  Figure 3: Variability in C, N, P concentrations and their stoichiometric ratios with increasing age among 50 snowshoe hares. Upper panels: while concentrations of P appear largely invariant as age increases, we notice a negative trend for N concentration for both sexes. This is further supported by the weak relationship found between age and %N through our modeling approach. Conversely, our modeling does not provide any support for the seemingly increasing trend we observe for %C. Lower panels: values of C:N appear to increase with age, for both males and females, as would be expected given the negative relationship between %N and age. Conversely, the values of N:P seem to decrease as males get older, which might mean that %N is more strongly influencing the variability of this ratio than %P is. No trend appears evident for C:P, which is in line with the lack of pattern in the variability of %C. We added a jitter to the data to improve readability of the graphs. All other specifications as in Fig. 2. evidence for a relationship between age, sex, body size, and either C:P or N:P as the top ranked 304 model for both these ratios was the intercept-only model (Table 2). Using skull length instead 305 of left hind foot length to calculate K n did not qualitatively change our results (see SI Tables S1 306 and S2). 307 Table 1: Top ranking GLMs for %C, %N, and %P based on ∆AICc values. We report only models that scored better than the null model, together with the null model. k, number of parameters in the model, LL, log-likelihood, K n , relative body condition, ABL, average body length. We provide coefficient values as estimate (±SE).  308 We provide one of few assessments of the body elemental composition of a terrestrial vertebrate 309 and investigate potential drivers of this fundamental ecological trait. Overall, we find considerable 310 variation in the concentrations of C, N, P, and their ratios within our sample of snowshoe hares.

311
However, age, sex, and body size appear to explain little of this variation. Our models highlight 312 a weak and negative relationship between an individual's age and its N concentration, and a sym-313 metrically weak and positive trend between age and C:N. Likewise, we find weak support for a 314 relationship between an individual's body size and its P content. Together, these results provide 315 some of the first evidence for intraspecific variability in the stoichiometry of a terrestrial vertebrate 316 but raise the need to consider a broader suite of potential drivers of the variability we observed.

317
Based on our analyses, we found weak evidence in support of our prediction that age might 318 drive variability in body elemental composition of snowshoe hares. In particular, we observe a 319 negative trend in N concentration: young individuals (0-1 years old) have seemingly higher N 320 concentrations than older ones -with a more pronounced decrease among males than among 321 females (Fig. 3). As would be expected from this pattern, C:N values show an opposite, positive 322 trend with age (Fig. 3) (Krebs et al., 2018). The higher N content among 328 leverets we observe, then, could be a sign of early investments in production of N-rich protein to 329 develop the muscle mass necessary for their hide and run anti-predator response. We also observed a qualitative pattern of increasing %P with age among males. While we lack quantitative support 331 for this trend (Table 1)  Counter to our prediction, we find no evidence for a relationship between hare stoichiometry 340 and sex. Male individuals did show larger variability in their N concentration than females (Figs 1-341 3), but our models provide no quantitative support for this observation. This lack of evidence for 342 differences between sexes may not be surprising. Several studies that investigated the relation-