LB 13-161 - Past landscapes best predict current distribution of South American foxes

Background/Question/Methods:

Species distributions are expectedly sensitive to landscape changes at different spatial and temporal scales. While time-lagged responses have been found across multiple taxa, these studies examine mostly species richness and are geographically restricted. While biodiversity aggregates are useful for assessing and mitigating future losses, they ignore the variation in individual species responses. Instead, we test for time-lagged responses in occupancy of two mammal species that are ecologically similar, but differ in geographic range and body size. We built single-species occupancy models from detections obtained through a systematic camera trap survey in Los Lagos Chile in 2019 to determine if present-day landscapes or past landscapes better explained current occupancy of two generalist mesocarnivores (Lycalopex griseus and L. culpaeus). Specifically, we extracted habitat amount and configuration metrics at three scales (250m, 500m, 1000m) from satellite imagery spanning a 33 yr time-series to incorporate into models

Results/Conclusions:

Across 3500 trapnights, we obtained 39 culpeo detections and 225 chilla detections. Culpeo and chilla occupancy both exhibited a time-lagged response to habitat configuration, with models with covariates from past years outperforming models with covariates from the same year the presence-absence data was collected. The scale and magnitude of the time lag varied by species; patch isolation (+) and edge/area (-) from 2015 at the 500m scale best described culpeo occupancy, while patch isolation (+) from 2011 best described chilla occupancy. We demonstrated that past landscapes better explained the present-day occupancy of two generalist carnivores, and that the scale at which past habitat metrics were measured resulted in opposite effects expected from range size. Past landscape configuration may be an important hidden driver of occupancy even for generalist species residing in regions with extensive forest loss and fragmentation. Therefore, given the wide use of occupancy modeling in conservation efforts, we encourage incorporating metrics to test for time lags in single-species models.