Mule deer fawn recruitment dynamics in an energy disturbed landscape

Abstract

Wildlife population dynamics are modulated by abiotic and biotic factors, typically climate, resource availability, density-dependent effects, and predator-prey interactions. Understanding whether and how human-caused disturbances shape these ecological processes is helpful for the conservation and management of wildlife and their habitats within increasingly human-dominated landscapes. However, many jurisdictions lack either long-term longitudinal data on wildlife populations or measures of the interplay between human-mediated disturbance, climate, and predator density. Here, we use a 50-year time series (1962-2012) on mule deer (Odocoileus hemionus) demographics, seasonal weather, predator density, and oil and gas development patterns from the North Dakota Badlands, USA, to investigate long-term effects of landscape-level disturbance on mule deer fawn fall recruitment, which has declined precipitously over the last number of decades. Mule deer fawn fall recruitment in this study represents the number of fawns per female (fawn:female ratio) that survive through the summer to October. We used this fawn recruitment index to evaluate the composite effects of interannual extreme weather conditions, energy development, and predator density. We found that density-dependent effects and harsh seasonal weather were the main drivers of fawn fall recruitment in the North Dakota Badlands. These effects were further shaped by the interaction between harsh seasonal weather and predator density (i.e., lower fawn fall recruitment when harsh weather was combined with higher predator density). Additionally, we found that fawn fall recruitment was modulated by interactions between seasonal weather and energy development (i.e., lower fawn fall recruitment when harsh weather was combined with higher density of active oil and gas wells). Interestingly, we found that the combined effect of predator density and energy development was not interactive but rather additive. Our analysis demonstrates how energy development may modulate fluctuations in mule deer fawn fall recruitment concurrent with biotic (density-dependency, habitat, predation, woody vegetation encroachment) and abiotic (harsh seasonal weather) drivers. Density-dependent patterns emerge, presumably due to limited quality habitat, being the primary factor influencing fall fawn recruitment in mule deer. Secondarily, stochastic weather events periodically cause dramatic declines in recruitment. And finally, the additive effects of human disturbance and predation can induce fluctuations in fawn fall recruitment. Here we make the case for using long-term datasets for setting long-term wildlife management goals that decision makers and the public can understand and support.

Keywords: disturbance; energy development; game management; harsh weather; mule deer; mule deer fawn survival; predator‐prey; recruitment; wildlife management.

FIGURE 1

Selected photos from our study area depicting (a) typical landscape features of the North Dakota badlands, (b) an operational well pad, (c) mule deer in proximity to an industrial facility, and (d) a mule deer doe with fawn. Photos: Ashley Salwey, Jesse Kolar and Craig Bihrle, North Dakota Game and Fish.

FIGURE 2

Map of our study area in North Dakota, showing (a) the location of mule deer study sites (in yellow) in North Dakota in relation to the Little Missouri river and the location of the Medora weather station (red triangle, middle of image, and bordering river) and (b) the total cumulative location of individual active well sites between 1962 and 2012 (black dots) and major highways (orange network) in relation to our study sites.

FIGURE 3

Long‐term population metrics for mule deer over the study period (1962–2012). Mule deer fawn recruitment, calculated as the ratio of fawns: female in fall (a), boxplots representing the variability among the 26 study sites (b) and mule deer density calculated as the total number of animals in spring (c) across 26 study sites in south‐west North Dakota.

FIGURE 4

Biplot for the principal component analysis for year of study and mule deer density. When PC1 values increase, mule deer density increases over time, whereas when PC2 values increase mule density declines along the time series 1962–2012. Including PC1 and PC2 in our recruitment model allows us to distinguish between density‐dependent effects (PC1, 87.9% variability explained) and recruitment decline due to other factors (PC2, 12.1% variability explained).

FIGURE 5

Biplot for the principle component analysis of landscape predictors included in our models. Increasing values of PC1 represents study sites in the north‐east that have experienced encroachment of Rocky Mountain juniper, whereas increasing values of PC2 accounts for study sites in the north‐east that have rugged terrain and less encroachment of woody vegetation, PC3 is shown in Appendix S1: supplementary material 1.

FIGURE 6

Fan chart of percentiles of oil and gas active well density (in shaded 10‐percentile intervals) across 26 study areas in the North Dakota badlands for the years of record (1962–2012). The median density is shown as a white line.

FIGURE 7

Effect of year and mule deer density principal components on mule deer fawn recruitment. PC1 increases (and recruitment decreases) when mule deer density increases with year of the study (a). PC2 increases (and recruitment decreases accordingly) when mule deer density decreases, accounting for years of decline not attributable to density‐dependent effects (b) and the effect of landscape increasing values of PC2 on mule deer fawn recruitment, where recruitment increases in their core range (north‐east) that is unaffected by the encroachment of woody vegetation such as Rocky Mountain juniper (c).

FIGURE 8

Interactive effects of active well density and average winter minimum temperature (a) and active well density and average snow depth in spring (b) on fall mule deer fawn recruitment (fawn/female ratio) in the North Dakota badlands (1962–2012) as predicted by the top ranked generalized additive mixed effect model.

FIGURE 9

Interactive effects of coyote density and average winter minimum temperature (a), coyote density and average snow depth in spring (b), and coyote density and average summer minimum temperature (c) on fall mule deer fawn recruitment (fawn/female ratio) in the North Dakota badlands (1962–2012) as predicted by the top ranked generalized additive mixed effect model.

FIGURE 10

Interactive effect of coyote density and well density on fall mule deer fawn recruitment (fawn/female ratio) in the North Dakota badlands (1962–2012) as predicted by the top ranked generalized additive mixed effect model. Note that this interaction was not significant in the top‐ranked model.