Temporal Period Delineation
We used a clustering algorithm to define seasons ecologically (Basille
et al. 2013). To define bobcat and coyote seasons, the movement speed
and turning angle between successive locations were calculated for each
individual. Using a moving window of time, we calculated the mean speed
and tortuosity, as well as the proportion of the locations in water,
exurban, grassland and scrub, forest, agriculture and wetland landcover
areas within the moving window (USGS 2021). The DD-weighted gap method
(Yan and Ye 2007) was used to determine the optimal number of clusters
(seasons). We then used K-means clustering analysis (MacQueen 1967,
Hartigan and Wong 1979) to identify clusters of similar space use
behaviors to define seasons, adjusting bootstrap thresholds and windows
of seasonal length to ensure continuous seasons of adequate length.
Day, night, and crepuscular diel periods were also delineated
(Supporting Information). Equinox dates (NOAA 2018) were used to divide
the year into four periods, and the average sunrise and sunset time for
each period was calculated (MapLogs 2018) to account for changes in day
length between the four periods (Thornton et al. 2004). Day was
delineated as two hours after sunrise to one hour before sunset, night
as two hours after sunset to one hour before sunrise, and crepuscular as
the two lengths of time one hour before to two hours after sunrise and
sunset (Franckowiak et al. 2020).