Linking telemetry data with forage availability maps advances information for moose management

Evidence-based forest and wildlife management requires precise estimates of forage availability for spatial planning and conflict mitigation. Recently, novel datasets and associated maps have been developed for use in Swedish forest and wildlife management. These foodscape maps show the estimated nationwide abundance of six tree species, within a browsing height window, that are important forage sources for Swedish ungulates. In this short communication, we tested their applicability to infer on movement and habitat selection using GPS data of 54 adult moose in two study sites in southern and central Sweden. We model habitat selection using a combination of a three-state hidden Markov model (HMM) and integrated Step Selection Analysis. We further tested their performance in models where habitat selection was not accounted for (behavior-unspecific models). Models containing information on both the foodscape and landcover performed best in the foraging HMM state and in behavior-unspecific models, across all seasons. In the encamped and travelling HMM state, forage + landcover models performed similarly to forage models. In behavioral states associated with foraging, habitat selection models containing data from the foodscape maps were consistently selected as the most parsimonious models compared to models that only included landcover information as covariates. Further, moose showed nuanced habitat selection patterns based on their behavioral state, and selection for young forest shifted once forage abundance was accounted for, from a generally preferred habitat to a generally avoided habitat. We showcase the predictive capability of the newly developed maps to draw nuanced inference on habitat selection of moose across several seasons in several behavioral states.