Flight initiation distance analysis
We first examined the effects of social and environmental factors on FIDs. Specifically, we ran a univariate mixed model with the following predictors: year, life-stage (juvenile or adult), sex, individual trappability, trial number, number of conspecific present, vegetation cover, starting distance, home site human activity score, and distance to shelter. We included home site identity, walker identity, and squirrel identity nested within home site identity as random intercepts. Home site identity was included as a random intercept because multiple squirrels were tested from the same home site and thus would more likely be similar to each other. Walker identity was included to control for potential experimental variation arising from differences among walkers. Finally, we recognized that in scenarios where individuals were tested in the presence of other conspecifics simultaneously, the FID of these individuals were not independent. Thus, we re-ran our model after filtering the data to only include the FID of the first individual to flee within these group testing situations but found that the same predictor effects remained as in our full dataset model (Table S1).
Individual FID repeatability (RFID) was calculated by extracting the variance components from our full FID model and using the following formula:
\begin{equation} R_{\text{FID}}=\ \frac{V_{\text{ind}}}{(V_{\text{ind}}+V_{\text{home}}+V_{\text{obs}}+\ V_{\text{res}})}\nonumber \\ \end{equation}
Where Vind is the among-individual variance, Vhome is the among-home site variance, Vobs is the among-observer variance, and Vres is the residual variance.