Elk resource selection patterns in a semiarid riparian corridor

Elk (Cervus elaphus) have depredated corn at Bosque del Apache National Wildlife Refuge (BDANWR), New Mexico, USA, which has interfered with the refuge's ability to provide supplemental nutrition to overwintering sandhill cranes (Grus canadensis) and other waterbirds. To identify management options for minimizing cropland depredation, we examined elk resource selection patterns using negative binomial generalized linear mixed models. We used 8,244 global positioning system (GPS) locations collected from 9 adult female elk to model fine-scale resource use (sampling units were 100 × 100-m cells; n = 3,646) and corn field use (sampling units were corn fields; n = 18) by a resident herd along the Rio Grande River in central New Mexico, USA. The fine-scale model suggested that elk use in cropland areas increased when alfalfa and corn were present and elk use was greatest 0.14 km from uncultivated areas. Elk use in uncultivated areas increased as canopy cover increased. Elk use exhibited a quadratic relationship with hiding cover, which varied with distance to cropland. We validated the fine-scale model with an independent sample of radio-marked adult female elk (n = 12; 1,106 locations). The fine-scale model was successful in predicting elk use; 84.1% (SE = 1.1) of radio-marked elk locations fell within high or medium-high use cells. Corn field use models indicated that elk use increased as the proportion of the corn field perimeter adjacent to alfalfa increased. Elk use of corn fields declined as distance to uncultivated areas and the proportion of other corn fields at the same growth stage increased. Probability of elk use peaked when corn reached heights of 1.4 m to 1.7 m and use varied with distance to uncultivated areas. Corn fields at these heights were in the late vegetative or tassel-milk growth stage, which are the stages at which damage to corn plants is most detrimental to yield. The average distance each elk moved per day during the corn growing season was 5,013 m (SD = 957) and varied among individuals (3,251-6,317 m). This is relatively large in relation to the size of the managed floodplain at BDANWR. Our results, couched in elk daily movements, can help direct crop management, vegetation manipulation, and timing of hazing efforts aimed at reducing elk use of crops.