Efficacy of remote sensing technologies for burrow count estimates of a rare kangaroo rat

Abstract

Effective management of rare species requires an understanding of spatial variation in abundance, which is challenging to estimate. We tested the efficacy of high-resolution imagery to detect burrows of the Texas kangaroo rat (TKR; Dipodomys elator) as a means of estimating abundance across its geographic range. Specifically, we estimated burrow counts using an Unmanned Aerial System (UAS) to collect data from very high-resolution Red–Green–Blue (RGB) imagery and estimate digital elevation (2.5-mm pixel resolution) over active and inactive burrows located on mesquite mounds and anthropogenic features (roadsides, fences, etc.). In 2018, we identified 26 burrow locations on a private ranch in Wichita County, Texas, USA, and characterized burrows based on topography and vegetation density. We found that TKR burrows can only be identified with data of <5 cm pixel resolution, thus eliminating the possibility of using high-resolution imagery data currently available for Texas. Alternatively, we propose that the use of National Agriculture Imagery Program (NAIP) imagery at 0.5- and 0.6-m pixel resolution, in combination with resampled digital elevation data, can provide an effective means for identifying potential TKR burrow locations at the county level. We present 3 different approaches at the county and local scale that combine topographic and vegetation fractional cover information using a weighted overlay approach. The modeling approaches have strong predictive capabilities and can be integrated with UAS data for visual confirmation of active and inactive burrows. We concluded that very high-resolution imagery and topographic information at pixel resolutions <5 cm collected by airborne systems can effectively help locate active TKR burrows. However, to remain cost effective, upscaling to the county level will require reducing the sampling area to the most suitable habitat. Modeling approaches, such as those proposed in this study, can help effectively locate these sampling areas.