*2.8. Immunoassay Methods*

For the analysis of FCMs two different EIAs were used. The immunoassays were a corticosterone EIA (commercial kit #K014-H1 or H5, provided by Arbor Assays, Ann Arbor, MI, USA) [12] and a 5α-pregnane-3β,11β,21-triol-20-one EIA (group-specific EIA, measuring FCMs with a 5α-3β,11β-diol configuration [15]). Details of the EIAs including cross-reactions are given by Arbor Assays and Touma et al. [15], respectively. All fecal extracts were diluted with EIA buffer prior to being analyzed. The dilution factor was determined by running pooled sample extract at different dilutions against the standard curve, to identify the one that resulted in ~50% binding for the corticosterone EIA (Figure 1A) and the group-specific EIA (Figure 1B). Through parallelism tests, we showed that serial dilutions of pooled extract tracked the EIA's standard curve (Figure 1). These findings demonstrated that methanol residue in sample extracts did not interfere with assay performance. Consequently, sample extracts were diluted 1:10 for the corticosterone EIA, and 1:200 for the group-specific EIA. However, extracts from field study 2 analyzed with the corticosterone EIA had to be diluted 1:80 instead of 1:10 (change discussed later).

We followed manufacturer's instructions for the corticosterone EIA. However, when we analyzed fecal samples with this EIA for field study 2, we also used a wavelength of 650 nm (reference wavelength) in addition to 450 nm. We followed the protocol described by Touma et al. [15] for the group-specific EIA. All samples were assayed in duplicate. Intra-assay coefficients of variation (CVs) were calculated by averaging all sample CVs and inter-assay CVs were calculated by averaging CVs of low and high concentration controls for all plates (except for field study 2, we could only calculate average of low concentration controls). Intra-assay and inter-assay CVs for the corticosterone EIA were (*n* = 9) 5.7% and 12.7% (field study 2: 12.8%) and for the group-specific EIA (*n* = 6) 8.6% and 8.6%, respectively.

**Figure 1.** Parallelism curves for pooled fecal extract from deermice measured with (**A**) corticosterone enzyme immunoassay (EIA) and (**B**) group-specific EIA. Standard curves for each EIA are shown in black with each concentration as an open diamond. Parallelism is shown in red with serial dilutions of pooled fecal extract as open red circles.

## *2.9. Statistical Analyses*

All data analyses were done in R [30] within RStudio [31]. We used linear mixed effect models (LMMs) from R packages "lme4" [32] and "lmerTest" [33] to test for diurnal patterns and how each treatment (dexamethasone suppression and ACTH stimulation) affected FCM levels of laboratory-bred deermice compared to baseline. We considered FCMs collected on the last day of acclimation as baseline FCMs. Because deermice were sampled repeatedly, individual identification was included as a random effect, where the model structure was sex + treatment × sampling time, except for when testing for diurnal patterns, which was sex + sampling time. *p*-values were calculated using Satterthwaite's method [33]. We used the R package "emmeans" [34] to perform post hoc pairwise comparisons where *p*-values were adjusted accordingly.

For field study 1, we used a one-way ANOVA with post hoc Tukey's Honest Significant Differences to compare FCM levels across trap confinement treatments (i.e., 0–2, 4–6 and 8–10 h), sex and reproductive status, where the model structure was treatment + sex + reproductive status. For field study 2, we used a LMM to examine the difference between FCMs across two sampling times (i.e., 0–4 h vs. overnight), sex, and reproductive status, where the model structure was treatment + sex + reproductive status. Because each deermouse was sampled twice, individual identification was included as a random effect.

FCM data were ln-transformed prior to all analyses to meet the assumptions of normality and homoscedasticity. Below, we present ln-transformed means with standard errors (ln ng/g of dry feces) where we considered results statistically significant at α = 0.05. However, in the figures we present non-transformed data to ease comparisons with other studies.
