An Integrated Time-Varying Ornstein-Uhlenbeck Process for Jointly Modeling Individual and Population-Level Movement of Golden Eagles

TL;DR

A full-year stochastic differential equation model for jointly modeling both individual movement and species distribution data is proposed and shows that this joint model results in efficient computation of the spatio-temporal dynamics of the entire population, and thus provides straightforward inference on the species distribution data.

Abstract

With technological advancements, the quantity and quality of animal movement data have increased greatly. Currently, no movement model can be used to describe full-year data from migratory species by leveraging both individual movement and species distribution data. Herein we propose a full-year stochastic differential equation model for jointly modeling both individual movement and species distribution data. We show that this joint model, under certain assumptions, results in efficient computation of the spatio-temporal dynamics of the entire population, and thus provides straightforward inference on the species distribution data. We illustrate this model by analyzing 215 bird-years of golden eagle movement in western North America jointly with relative abundance data from eBird. We use the results to estimate wind project risk for these eagles and predict where they came from earlier in the year based on a single telemetry observation from later in the year. Our joint model enables additional inference and greater predictive power than afforded by sole use of eBird relative abundance.

Figures (9)

• Figure 1: Movement tracks based on satellite telemetry data for golden eagles in western North America, separated by subpopulation. The eagles were placed in subpopulations based on a clustering of their normalized (displacement from January 1 location) telemetry data off via k-means algorithm. This highlights the different migratory behaviors that individual golden eagles exhibit. These four subpopulations could be described as moderate-distance migrants, long-distance migrants, short-distance migrants, and non-migratory (full-year resident) eagles respectively.
• Figure 2: Selected weeks from the 2018 eBird adaSTEM output for golden eagles in western North America, which is used as our species distribution data. This highlights the range of migratory behaviors exhibited by individual golden eagles in the population.
• Figure 3: (a). Locations of the 18,456 wind turbines within the range of our species distribution data that are within the contiguous U.S. (b). Locations of the 396 wind projects within the range of our species distribution data that are within the contiguous U.S. (c). Locations of Utah County - Utah (top left), Boulder County - Colorado (top right), and Santa Fe County - New Mexico (bottom).
• Figure 4: Posterior mean of the relative abundance of golden eagles in western North America for selected weeks.
• Figure 5: Posterior mean of the density of the stationary distribution of the OU process of the winter and summer attractive points.