Fig 1.
Diagram showing predictions of variation in network structure within wildlife populations of co-occurring individuals. The full network structure is unobserved without known locations of all individuals but with an adequate sample of the population metrics of aggregation can identify the degree of connectivity. Network structure can range from a single highly mixed homogeneous group (a) to heterogeneous with multiple unconnected groups (c). As network structure changes different numbers of sampled individuals may be necessary for accurate evaluation.
Fig 2.
Map showing the study region and locations of the National Elk Refuge and Fossil Butte National Monument in the state of Wyoming, home to the wintering grounds of the Jackson elk herd (JKSN) and West Green River elk herd (WGR), respectively. The background relief is based on the SRTM 30-m digital elevation model provided by NASA JPL [50]. Inset map indicating location of Wyoming derived from U.S. Census Bureau boundary data [51].
Fig 3.
Seasonal Variation in Herd Structure.
Map showing 90% contour from kernel density estimates for elk from the Jackson (JKSN, purple) and West Green River (WGR, blue) herds during low (a,c) and high density (b,d) periods in Wyoming, USA. The location of the National Elk Refuge and Fossil Butte National Monument are indicated by black bordered polygons. The background relief is based on the SRTM 30-m digital elevation model provided by NASA JPL [50].
Table 1.
Minimum sample size required for aggregation metrics.
Fig 4.
Sample Size Accumulation Curves for Each Aggregation Metric.
The proportion of simulations at each sample size that are within 20% of the baseline value calculated at maximum available sample size for elk from the Jackson and West Green River herds in Wyoming, USA. Each aggregation metric is calculated during high- and low-density periods for each herd (a-d, e-h, respectively). Aggregation metrics include the daily 1st quartile inter-elk distance (a,e), mean daily proximity rates (b,f), and the area of 90% (c,g) and 50% contours from kernel density estimates (d,h). Dashed lines indicate when the proportion of simulations within 20% of the baseline value is ≥ 90% (numerical values shown in Table 1). Accumulation curves are fitted using loess smoothing methods for visualization.