Distance, weather, and forage conditions drive timing of autumn migration in female mule deer

Abstract

Background: Seasonal migration is a behavioral strategy that animals evolved to exploit seasonally changing resources. Ungulates in northern temperate landscapes often seasonally migrate between low-elevation winter ranges and higher-elevation summer ranges, allowing individuals to exploit a diversity of forage resources during summer while avoiding extreme conditions during winter. In autumn, the timing of this behavior often overlaps with hunting seasons for managed ungulate populations. Migration presents challenges for managing ungulates when the timing of autumn migrations varies across years and migrations cross management jurisdictions.

Methods: We evaluated the spatial and temporal patterns of autumn migration using GPS collar data collected during 2017-2019 from 68 female mule deer (Odocoileus hemionus) that migrated seasonally within three study areas in northwest Montana. We related the timing of autumn migration to environmental variables including precipitation, snow depth and density, temperature, plant phenology, migration distance, and estimates of relative hunting intensity. We summarized variables across multiple temporal scales (2-day, and 1 week) to identify possible lagged or cumulative effects of conditions on mule deer behavior. We incorporated these variables into a time-to-event modeling framework to estimate their relative impacts on the timing of initiation of autumn migration.

Results: The collective annual space use of deer in each study area spanned up to 9 hunting districts, and individual deer used an average of 2.1, 2.8, and 2.0 hunting districts per year (range 1-4) in the Cabinet-Fisher, Rocky Mountain Front, and Whitefish study areas, respectively. Furthermore, the return of deer to winter ranges occurred over a 3-month timeframe spanning archery, rifle, and closed hunting periods. While some deer returned to winter range relatively early during archery season in September, others remained in summer range into December, after the general rifle season concluded. Declines in daily minimum temperatures and increased weekly precipitation provided the strongest cues for mule deer to begin their autumn migration. Mule deer with longer migration distances were more likely to initiate their migration sooner, and declining forage conditions also showed a modest effect on timing. Mule deer migrations occurred during times of lower hunting activity prior to its peak during rifle season.

Conclusions: Our study demonstrates changing weather conditions were the primary driver of the initiation of autumn migration for mule deer. Given most migrations spanned more than one hunting district, the boundaries of management units were mismatched with the scale of ecological processes, implying that management actions in certain districts may have unintended consequences for populations in nearby districts.

Keywords: Odocoileus hemionus; Climate; Hunting; Migration; NDVI; Precipitation; Snow water equivalent; Temperature.

Fig. 1

Map displaying mule deer summer (yellow) and winter (blue) home range centroids in each of the three study areas in northwest Montana: (A) Cabinet-Fisher, (B) Whitefish, and (C) Rocky Mountain Front. Black lines connect each summer and winter home range for a mule deer during an individual year

Fig. 2

Proportion of migratory female mule deer that had returned to the hunting district encompassing their winter capture locations, using a 3-day average to smooth lines, averaged across years, and within 3 study areas of western Montana, 2017–2020

Fig. 3

Estimated hazard function (black line) for the initiation of autumn migration for mule deer in 3 study areas of western Montana from discrete-time survival models, 2017–2020. Most migrations, daily counts shown with black histogram, were initiated in late September and early October during the archery hunting season (purple shaded area) and prior to the beginning of the rifle hunt season (red shaded area)

Fig. 4

Risk ratio estimates for variables included in the top overall model using discrete-time survival models. Asterisks indicate the level of significance for each variable