Elk Nutrition & Habitat Use Models for Management Overview of Westside Modeling Work

advertisement
Elk Nutrition & Habitat Use
Models for Management
Overview of
Westside Modeling Work
Westside Elk Models Completed

Nutrition Model

Habitat Use Model
Elk Modeling Workshop


Day 1:
Modeling Objectives, Methods, Results,
Management Applications, Discussion
Day 2:
“Hands On” Computer Applications of
Models on Example Westside Landscapes
Elk Modeling Workshop
For additional information, model programs, or data:
1. Visit PNW Research Station website:
http://www.fs.fed.us/pnw/calendar/workshop/elk/
Web search “Elk Models PNW Station”
2. Visit Elk Modeling FTP site:
ftp://ftp2.fs.fed.us/incoming/pnw/ElkHabModeling/
ElkWorkshop_Day2Files/
Today’s Presentations

Morning: Modeling Work

Afternoon: Application Methods & Examples

End of Day: Q & A/Discussion Period
Introductory Presentation

Background & Need for New Models

Modeling Objectives

Overview of Work Accomplished
Partners (Staffing, Funding, or Data)








American Forest Resource Council
Boone and Crockett Club
Hancock Timber Company
Lower Elwha Klallam Tribe
Makah Nation
Muckleshoot Indian Tribe
National Council for Air and Stream Improvement
National Fish and Wildlife Foundation
Partners (Staffing, Funding, or Data)








Oregon Department of Fish and Wildlife
Oregon Forest Industries Council
Oregon State University
Plum Creek Timber Company
Quileute Tribe
Rayonier, Inc.
Rocky Mountain Elk Foundation
Sauk-Suiattle Indian Tribe
Partners (Staffing, Funding, or Data)







Sporting Conservation Council
USDA Forest Service (WO, R6, PNW)
USDI Bureau of Land Management (WO, OR-WA)
Washington Department of Fish and Wildlife
Washington Forest Protection Association
WEST, Inc.
Weyerhaeuser
Background and Need
R. Cook
Elk Habitat Models—Who Cares?

Elk compose a multimillion dollar economy
in the western U.S.
(hunting, viewing,
agricultural damage).

Elk distribution across
landscapes, and thus
the species’ economic
impacts across land
ownerships, can be
directly controlled
through management.
Elk Habitat Models—Who Cares?

Elk are an “early-seral”
species whose needs
are compatible with
active silviculture.

Other “early-seral”
species may benefit
from habitat
management for elk.

Public land use plans
continue to feature elk
and elk management.
Background and Need

Models developed in 1970s & 80s with
limited empirical data.

Empirical data restricted to single variable
habitat relations with elk.

Contributions of multiple variables to
predict relative probability of elk use not
empirically-based (e.g., geometric mean);
resource selection modeling not yet
developed.
Why New Elk Habitat Models?

“Old” models still in use
but structure and some
components outdated
(e. g., thermal cover)

Data from many radiotelemetry studies now
available for new
modeling approaches.

Data on elk nutritional
resources now available
for modeling at
landscape scales.
Modeling Objectives
R. Cook
Modeling Work:
• Summer range (elk productivity)
• Large (regional) landscapes,
multiple land ownerships, integrated
management strategies
• Local landscapes,
smaller projects within large
landscapes
Modeling Work:
• Summer range (June 1-Aug. 31)
• Regional landscapes (≥25,000 ac)
• Non-hunting periods
(but populations
are hunted)
• Female elk
Objectives
1. Build a model that predicts nutritional
resources for elk across all landscape
conditions on summer range
(Nutrition Model).
2. Construct & rank a set of plausible,
competing models that predict the relative
probability of elk use across landscape
conditions on summer range, & validate
“best” model with data independent of
model construction (Habitat Use Model).
Objectives
3. Include nutrition model predictions as a
covariate in the habitat use model.
4. Include additional human disturbance,
vegetation, and physical covariates that
potentially affect or account for elk use of
nutritional resources.
Objectives
5. Use data from
multiple study
areas (diverse
environments &
land ownerships)
to construct,
select, and
validate habitat
use model.
Objectives
Two predictions of primary interest:
1. Composition (percent area)
of a landscape by nutritional
condition (e.g., excellent,
good, marginal, poor)
—Nutrition Model
Objectives
Two predictions of primary interest:
2. Level of elk use
(relative probability of use)
across a landscape, given all
covariates that affect use
—Habitat Use Model.
Objectives
Two management options of primary interest:
1. Degree to which management can change
landscape area by nutritional condition
(Nutrition Model).
2. Degree to which management can change
level of elk use on a landscape, given all
factors that affect use (Habitat Use
Model).
Modeling Objectives:
• Habitat Use Model: Designed as a
tool to evaluate and manage elk
distributions within and across land
ownerships on regional landscapes,
and on smaller (local) areas within
the large landscapes.
• Elk management problems usually
are related to where elk occur rather
than elk numbers per se.
Work Accomplished
R. Cook
Nutrition Model

Regression model that predicts elk dietary
digestible energy based on data from
grazing trials conducted with tame elk
across representative environments.
Nooksack S.A.
Willapa Hills S.A.
Springfield S.A.
Nutrition Model
• % Canopy Cover
• Prop. Hardwoods
Dietary Digestible Energy (DDE)
Predictor Variables within a
given Modeling Region &
Potential Natural Vegetation Zone
Nutrition Model Data Sources
Predictor Variables
1. % Canopy Cover
2. Prop. Hardwoods
Data Source
GNN1
GNN
1 Spatial data estimated for time period 2006 based on Gradient
Nearest Neighbor Analysis (Ohmann & Gregory 2002)
Additional nutrition
field work with tame elk,
summer 2011
Habitat Use Model

Use available telemetry data for model
development (selection) & validation.
Two Analytical Steps

Model Selection of Top-Ranked Model
Among Many Plausible, Competing Models

Validation of Top-Ranked Model
HABITAT USE MODELING
No. Study Areas & Years:
7 Study Areas, 21 Years
Study Area Sizes:
˜ 33,000 ac to ˜ 243,000 ac
Average ˜ 140,000 ac
Study Areas encompass a
wide spectrum of land
ownerships & associated
management conditions.
Radio Telemetry Data
VHF – triangulation, direct observation
 GPS – automated
 High variability across study areas in
location accuracy, re-location frequency,
re-location time of day and no. of animals

Radio Telemetry Data

Model Selection: 6 to17 elk per area; 908
to 2208 locations per animal

Model Validation: 3 to 36 elk; <10 to 6298
locations per animal
>50 Covariates Considered in the
Model Selection Process
Human
Disturbance
Nutrition
DDE
(continuous/categorical)
Accepted Biomass (AB)
Distance to:
GE DDE
MGE DDE
Percent area in MGE
Mean DDE in MGE
Density of & Distance to:
Open Roads
Closed Road
High Traffic Roads
Low Traffic Roads
Public Use Roads
Administrative Use Only
Motorized Use Trails
Vegetation
Overstory Canopy Cover
Physical/
Other
Slope (continuous/categori
Percent Area in:
Flat to Gentle Slope
Cover-Forage Ratio
Mod to Steep Slope
Size & Spacing of Cov & For Very Steep Slope
Dominant CC
Proportion of Optimal CoverAspect
Distance to:
Forage
Cover
Cover-Forage Edge
Optimal Cover
Thermal Cover
Hiding Cover
Convexity
Curvature
Soil Depth
Solar Radiation
Distance to Water
Land Ownership
Reduced Set of Covariates in Competing Models
Which Model Best Supports the Telemetry Data?
“Best Model” Contained 4 Covariates (Model Selection):
1.
Dietary Digestible Energy of Forage (Nutrition).
2.
Distance to Roads Open to Public Motorized Use
(Human Disturbance).
3.
Slope (Nutrition, Energy Efficiency).
4.
Distance to Cover-Forage Edge (Nutrition,
Human Disturbance, or both).
Habitat Use Model Data Sources
Model Covariates
Data Sources
1. Diet. Digestible Energy
Nutrition Model
2. Dist. to Open Roads
Available Rds Layers
3. Slope
Nat. Elev. Dataset
4. Dist. to Cov-For Edge
GNN
Model Validation – remarkable agreement between predicted and
observed elk use – correlation coefficients from 0.32 to 0.99, most >0.80
Nutrition Model
Elk Habitat Use Model
• % Canopy Cover
• DDE
• Prop. Hardwoods
• Dist. to Public Roads
• % Slope
• Dist. to Edge
Dietary Digestible Energy (DDE)
Predicted Level of Elk Use
Morning Presentations

Nutrition Modeling: John Cook

Habitat Use Modeling & Interpretations: Ryan Nielson

Habitat Use Modeling Results: Mary Rowland
Afternoon Presentations

Beta-Testing Process: Jennifer Boyd

Summarizing Model Results: Mike Wisdom

Umpqua NF Model Application: Justin Hadwen

BLM Model Application: Lisa Renan

Muckleshoot Tribe Model Application: Dave Vales,
Mary Rowland

Wrap-Up: Mike Wisdom
Question and Answer/Discussion Session at End of Day
Download