Elk Nutrition and Habitat-Use Models for Western Oregon and Washington

Project Description

Our objective was to deliver readily applied, defensible models to inform management of elk populations and their habitat west of the Cascade crest in Oregon and Washington. Understanding nutritional resources and habitat use of large herbivores like elk, species of high cultural and economic value, can benefit their management across different land ownerships and management regimes. Distributions of elk in much of the western United States have shifted, leading to reduced hunting and viewing opportunities on public lands and increased crop damage and other undesired effects on private lands. These shifts have occurred with changes in forest management activities and increasing disturbance from factors such as traffic on open roads. Such changes can influence elk nutritional resources and their availability, which in turn affects habitat use by elk.

Purpose and Scope

We hypothesized that elk habitat use during summer (June–August) was driven by interactions among variables that influence energy balance, specifically acquisition of nutritional resources and energy expenditure, for example travel by elk on steep slopes. The Westside elk modeling project had two specific objectives:

1) Develop and evaluate a nutrition model to accurately estimate region-wide nutritional resources for elk on summer range.

2) Develop and validate a summer habitat-use model that integrated nutrition model outputs with other predictors to estimate the relative probability of use by elk.

The regional model resulting from this effort was designed to apply to all potential elk habitat across western Oregon and Washington (11 million ha), excluding a small portion of southwest Oregon south of the Coquille River (fig.1). Model predictions were applied across all land ownerships to extend model utility for a wide variety of stakeholders and applications.

Image: Figure 1: Study areas in western Oregon and western Washington, USA, where data were collected to develop or evaluate elk nutrition and habitat-use models: elk nutrition field data (shaded squares); modeling regions for elk nutrition (colored polygons); comparisons of elk nutrition model predictions with elk performance (stars); and areas used only for habitat-use modeling (circles).

Methods

We acquired elk telemetry locations and predicted elk nutritional resources in 12 study areas spanning a range of environmental conditions (fig. 1). We used some data sets for nutrition and habitat-use model development and others for model validation. Each study area served as a replicate in generating a regional habitat-use model.

To derive covariates for our models, we developed an exhaustive list of more than 50 variables, gleaned from published literature and expert opinion, that have been shown to affect habitat use by elk. We grouped the covariates into 4 categories: nutrition, human disturbance, vegetation, and physical.

• Nutrition covariates estimated dietary digestible energy (DDE) and biomass of those plant species readily consumed (accepted) by elk.
• Human disturbance covariates reflected road density and distance from open roads.
• Vegetation covariates include several canopy cover and edge-related metrics.
• Physical covariates were measures of topography.

We developed the final Westside elk nutrition and habitat-use models by using the best of the covariates that emerged from our model selection process and evaluated performance of the models in several ways. For example, we calculated selection ratios of use/availability, comparing use by elk of various categories of DDE in relation to their availability. We validated the habitat-use model by using 8 independent data sets and correlating observed elk locations to predicted use from the model.

Prior publications were foundational to development and evaluation of the Westside elk nutrition model. These include:

• Cook, J.G.; Cook, R.C.; Davis, R.W.; Irwin, L.L. 2016. Nutritional ecology of elk during summer and autumn in the Pacific Northwest. Wildlife Monographs. 195: 1–81.
• Cook, R.C.; Cook, J.G.; Vales, D.J. [et al.]. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs. 184: 1–45.
• Cook, J.G.; Johnson, B.K.; Cook, R.C. [et al.]. 2004. Effects of summer-autumn nutrition and parturition date on reproduction and survival of elk. Wildlife Monographs. 155: 1–61.

Implementation

The nutrition model, developed from field studies with captive elk in several westside locations, predicts elk nutritional resources (i.e., DDE) at scales from a single stand (about 100 ha) to multiple watersheds. This model is a “stand-alone” product, i.e., it can be applied without running the habitat-use model. Its outputs, however, are required to apply the habitat-use model.

The habitat-use model predicts the probability of elk use for landscapes of at least 10,000 ha. It can be used to assess current conditions or projected elk habitat use under a variety of management scenarios.

The modeling team shared results with stakeholders for use in land management planning and decisionmaking related to summer habitat for elk throughout the Westside region. Biologists and planners across multiple agencies, industry groups, universities, and Tribes have applied the nutrition and habitat-use models in various ways. For example:

• The Muckleshoot Indian Tribe applied the Westside elk nutrition model to calculate an elk forage index in the Tribe’s 37,000-ha Tomanamus Forest. The biologists set forage index targets within elk management areas to use in decision models to guide forest planning and evaluate management alternatives.
  • Vales, D.J.; Middleton, M.P.; McDaniel, M. 2017. A nutrition-based approach for elk habitat management on intensively managed forestlands, Journal of Forestry. 115: 406–415. https://doi.org/10.5849/jof.16-032. 
• Several biologists and students used outputs from the elk models to stratify study areas in the Westside region for sampling elk to meet a variety of objectives, such as estimating density of Roosevelt elk (C. c. roosevelti) in western Oregon and comparing habitat-use model covariate values to estimates of elk population density: Using Spatial Capture-Recapture to Estimate Density of Roosevelt Elk in Western Oregon.
• A project funded by the Rocky Mountain Elk Foundation is using the elk nutrition model to explore effects of variable density thinning on elk forage production and DDE.
• The forest biologist for the Willamette National Forest mapped nutritional resources for elk across the entire national forest at three times points (1986, 1993, 2012) to monitor trends in elk habitat conditions and guide forest planning.

Key Findings

• Predicted use by elk closely tracked increasing levels of dietary digestible energy (DDE), with the strongest predicted use at DDE levels at or above requirements for lactating females (fig. 2).
• The regional model that best predicted habitat use revealed that elk preferred habitats that were relatively high in DDE, far from roads open to motorized use, close cover-forage edges, and on gentle slopes.
• Our nutrition model, based on DDE, explained 73 to 85% of the variation in pregnancy and body fat levels of lactating females from 9 elk herds in the region, demonstrating a robust habitat-performance linkage that is often difficult to establish in wild ungulate populations (fig. 3).
• The models performed well with independent testing data (mean correlation of 0.77 across validation areas), lending credibility to their widespread application.

Figure 2: Relation of elk dietary digestible energy to predicted use by elk.

Figure 3: Understanding if nutritional resources meet requirements for survival and reproduction in elk relies on accounting for the high energetic demands for milk production and calf growth during summer. Photo by Rachel Cook.

Related Projects: 

• Blue Mountains Elk Nutrition and Habitat Models
• The Starkey Project