Abstracts

Presented by Cory Overton - Email: coverton@usgs.gov

Advancements in technology are rapidly increasing the capacity for wildlife researchers to obtain high frequency data on ever smaller organisms and at less expense. This capability results in comparatively large datasets and include more individuals that provides counterpoints to relatively sparse data from decades past. Undoubtedly, these newer methods can provide more accurate estimation of important ecological processes such as space use, movements, and resource selection patterns along with better representation of within population variability. However, many approaches used to quantify animal movement or space use are not robust to differences in the cadence of relocations and comparisons to historical data may not account for methodological biases even when using the same technique. Additionally, data available for analyses can represent a trade-off between accurately estimating space or movements of an individual and inclusion of more individuals to adequately represent population variability. We leveraged a database containing over 925,000 locations on 451 individual mallard, pintail, and cinnamon teal to evaluate difference space use estimates under multiple scenarios of data collection cadence or interval. We quantified the area of space used across 20-day periods using 5 analytical techniques on waterfowl data collected at a 30-minute cadence and the amount of bias present in that data subset to 6 different data cadences (1-, 2-, 3-, 6-, 12-, and 24-hourly intervals). Relative bias in space use estimates tended to increase with larger data cadence through the 12-hourly interval before decreasing for most methods, reflecting both the impact of fewer relocations in higher cadences and the lack of circadian patterns of space use in the 24-hourly cadence dataset. Average absolute relative bias across all cadences ranged from 12% to 80% indicating that some space estimation methods, e.g. adaptive radius local convex hulls, can be robust to data collection interval when appropriately parameterized.