Long-term population monitoring of northern Idaho ground squirrel: 2018 implementation and population estimates

Call Number:

Keywords:

Abstract:

We implemented the fifth full year of the new northern Idaho ground squirrel (NIDGS) long-term population monitoring strategy in 2018. The sampling frame consisted of the original 1,757 100-m x 100-m grid cells across occupied habitat, plus an additional 883 grid cells from a newly created stratum of cells. We conducted line-transect distance surveys in 1,107 cells and recorded 1,692 NIDGS at 564 cells (51%). From these data, program DISTANCE estimated a density of 0.84 squirrels/ha and a total population size of 2,173 squirrels (95% CI: 1,923–2,434). We post-stratified data based on relative density (higher, lower, or unknown), with resulting densities of 0.98 squirrels/ha in stratum 1, 0.55 squirrels/ha in stratum 2, and 0.71 squirrels/ha in the newly created stratum 3. Corresponding unadjusted population sizes were 1,340, 213, and 595, respectively. Our adjusted index to overall abundance was 2,933 NIDGS. We compared the 1-year change in population estimates between 2018 and 2017 in 3 ways: from the DISTANCE analyses of survey data from strata 1 and 2, from DISTANCE analysis of 500 core grid cells intended to be surveyed every year, and from a pair-wise comparison of the 500 core cells. While the point estimate for population size was 22% lower in 2018 compared with 2017, NIDGS abundance did not differ significantly. Similarly, average detections per cell from the core 500 grid cells were not significantly lower in 2018. We explored several environmental variables, including tree canopy cover, aspect, heat load index, soil properties, and proximity to nearest squirrel as site covariates in occupancy modeling with program PRESENCE. The most parsimonious model included proximity to nearest squirrel and tree canopy cover, with variable detection across visits. Probability of detection, given a cell was occupied, was estimated at 0.65 for visit 1, 0.57 for visit 2, and 0.82 for visit 3. We applied this model to the full 2,590-cell sampling frame to generate a probability of occupancy map. Roughly one third of the cells in our expanded sampling frame had >75% probability of being occupied, approximately half of the cells had <50% probability of being occupied, and 116 cells (5%) had 0% probability. We conducted presence–absence surveys at 7 sites where no grid cells were selected for surveys, and conducted exploratory surveys across 690 ha outside of known sites to determine if sites were still extant and document dispersal into new or treated areas.

Notes:

ELECTRONIC FILE - Zoology, Mammals

Endangered Species Section 6 Grant F17AF01082
and Cooperative Agreement No. F17AC00437