USFS
Pacific Northwest
Research Station
June 2010
Elk habitat selection in western Oregon and Washington:
Models for a new century
Why do we need new elk habitat models?
Three key reasons justify the need for new elk
habitat models:
Elk are highly sought by the public for hunting and viewing.
Elk have multi-million dollar effects on recreation and land management.
Elk management is likely to benefit a large
number of native species that depend on
early-seral (e.g., grass-forb) habitats or that
are sensitive to human disturbances.
Existing elk habitat models used in land management planning in western Oregon and Washington
are currently being updated to reflect research
conducted in the last two decades. For example,
declines in elk populations in the 1990s led researchers to pay more attention to factors such as
summer nutrition, which was previously not considered a limiting factor for elk populations.
new knowledge about elk habitat requirements,
especially nutritional conditions. After developing
and evaluating several versions, the team found
that the model that performed best in initial validation tests was based on four habitat covariates:
# distance to open public road
# dietary digestible energy (DDE)
# distance to cover-forage edge
# slope
The elk modeling team is developing this new regional model to help support current land management plan revisions and elk habitat management decision-making across most of western Oregon and Washington.
In a recent two-day workshop in Corvallis,
OR, the modeling team presented background on
model development, shared draft results, and provided example management applications to a diverse audience of potential model users. This
briefing paper summarizes the information preAn interagency team of researchers has been comsented at the workshop as well as feedback from
piling data, conducting analyses, and developing a
workshop participants.
new, regional elk habitat model that incorporates
1
Key messages from the workshop
What do we know about elk populations and habitat?
# The needs of elk are compatible with active management,
silviculture, and wildfire. However, although elk benefit
substantially from a variety of management practices
that reduce overstory cover, use of the resulting forage
base depends on managing human disturbance, for example through seasonal road closures.
# In general, the lower the canopy closure and the higher
the elevation, the greater the abundance of high quality
forage species and dietary digestible energy (DDE). Digestible protein also tends to follow this pattern.
# DDE relates directly to animal performance. Herds with
summer deficiencies in DDE may have low body fat and
reduced pregnancy rates.
# Elk habitat is relatively poor in the Coast Range and many areas in the
Cascades. Even with clearcuts, forage quality is often below maintenance level for elk in these areas.
# The model is explicitly designed to evaluate summer range conditions
because this is the time that substantially affects year-round animal
performance. For example, summer range conditions directly affect
pregnancy rates and body fat levels of elk entering winter. Consequently, evaluation and management of summer range conditions is
considered essential to year-round management of healthy elk herds.
“Elk really do know what
they’re doing out there
on the landscape,” said
Mike Wisdom. “If they’re
not spending time in the
most nutritionally dense
areas, they’re very close
to those areas.”
What can the model help us do?
# The model characterizes elk use patterns across landscapes. It can
This model is a landscape
tool that has highest utilhelp set goals for changing elk use in certain areas, and help assess
ity for large landscapes
how to get more “bang for your buck” with management prescriptions. across ownerships for
It can also show the consequences of NOT improving habitat.
multi-partner habitat
planning among all partners who manage elk
# Because the model is regional in scale, it facilitates an “all lands” appopulations and habitat.
proach that can help encourage and inform dialogue among different
ownerships and agencies.
2
Key messages from the workshop
# Users can compare different management scenarios: for example,
using the Forest Vegetation Simulator, you could remove canopy cover in units across a management
landscape and then estimate DDE at specific time intervals to make predictions of elk use. If management
speeds up or retards vegetation succession, the
model will reflect that.
What is the model NOT able to do?
# Special circumstances unique to a particular area (such
as heavy trail use) will not be reflected in model results, and will always require users to examine their
data and perhaps modify model inputs according to
their local knowledge.
# The model is not scaled to the project level, i.e., it may
not accurately reflect results of specific management
treatments across a few acres. “This is a macro tool, not a micro
tool,” explained one modeler.
# The model is not appropriate for evaluating elk habitat in seasons
other than summer. However, the summer period is considered
a limiting time period for elk nutrition, with carryover effects in
subsequent elk productivity. Consequently, the modeling focuses
on summer range habitat conditions that affect overall elk productivity and thus are considered a foundation for holistic elk
habitat management.
# The current model relies on vegetation maps for 2006 produced
with the Gradient Nearest Neighbor (GNN) approach (http://
www.fsl.orst.edu/lemma/), and will not be updated wholesale. If
needed, users can update a study area with data collected in the
year of interest or by using other remotely-sensed spatial data.
As with any spatial model, any appropriate data available for a
given landscape and modeling objectives can be used, and no
“corporate” spatial layers are fully available for all areas of the
region or for all time periods of interest.
“The model presents a picture
of the average behavior of the
average elk across a whole region,” explained Ryan Nielson.
Overall, the response of workshop participants was very
positive. The utility of the
model was ranked high, and
when breakout group members
were asked how they would
like to see model products delivered, someone said, “On my
desk first thing tomorrow
morning would be nice!”
3
Questions from workshop participants
What if active management for elk is not feasible because you have
owl habitat?
Management of public lands has become a complex balance between
sometimes competing demands. Regardless of the management objectives on a given piece of forest, this model can still provide information about impacts of different land uses on elk populations and
help practitioners weigh possible alternatives. As one workshop participant said, “Having these nutrition maps could really help us from
an administrative standpoint because they justify creating openings.”
What is the minimum-sized area for model application?
The model is designed for application at the same spatial extents at
which it was developed and validated. In general, the minimum size
of area appears to be approximately 25,000 acres (10,000 hectares).
This general guideline may change, however, with more validation
study sites to be considered in the final analyses that will be completed by fall 2010. This type of question will be addressed in more
detail in a user’s manual and related publications that are planned.
Can these models be used for winter habitat? Movement corridors?
Hunting seasons?
No. The telemetry data used to fit the models were filtered for summer (June – August) months only, when the females are trying to put
on body fat—a key component of overall productivity. These models
would not be accurate in the hunting season, nor for spring or winter
seasons. See above rationale regarding the focus on summer range
habitat evaluation for the modeling process.
Can the models be used in southwest Oregon?
No. A key covariate in the model, DDE, has not been modeled there
yet. “In the southwest Oregon region there’s just a big question
mark,” said John Cook. “We just don’t have much data for the region,
and differences in vegetation and nutrition between southwest Oregon and further north may be substantial.”
Proposed products for delivery of model tools and
information
Maps
GIS files
Stand-alone software
Peer-reviewed publications
User guide (USFS General
Technical Report)
If we want to draw elk away from private land, what forage species
should we reintroduce?
Cook stated, “We will reveal forage selection data in excruciating de4
Questions from workshop participants
tail in a monograph. It’s in table form if you want it now.” The plant species list
can help identify which species to avoid and which are palatable to elk. The
model, however, works with groups of species, not individual species, based
on their palatability.
To request forage
data, contact:
John Cook
541-962-6536
cookjg.ncasi@gmail.com
Does this model mean we should not manage for elk habitat in the Coast
Range?
There was some concern about sending the message to administrators that management for elk is a lost cause in the Coast Range. The draft regional model shows
consistently poor nutritional conditions for elk here, and body fat measurements
and pregnancy rates back this up. One concern is that nutritional conditions in
habitat managed to benefit elk through forest management may not fully meet
nutritional needs of elk in summer. This may be true in this area, but early seral
habitats are nonetheless much better than unmanaged coniferous forests. All in
all, our data clearly indicate that improvements through active management
should benefit elk, even in the Coast Range.
Can this model capture the impact of treatments on local habitat quality?
No. The model works at a broader scale than the individual treatment level. For
example, model resolution would miss the impact of management prescriptions
such as variable density thinning (“skips and gaps”) implemented on 1/2-acre
plots. What the model can be helpful for is laying out landscape-scale habitat enhancement scenarios such as determining what general areas would yield the
most promising results from silviculture activities. Another value of the model is
that it can make explicit the negative consequences of certain management constraints on elk habitat.
Extrapolating to the real world
In summary, workshop participants had questions
about what the models can tell them about their
particular areas. Again, the issue of scale comes
into play. In many places, to shift nutritional conditions into good and excellent categories and actually change elk use, active habitat management
would have to be implemented broadly. Just as
there are interacting variables that combine to
create habitat quality, there are interacting issues
practitioners must weigh, and ultimately use their
professional judgment.
The model simply illuminates the variable conditions that affect elk habitat, allowing practitioners
more food for thought. Therefore model documentation needs to clearly define limitations and
assumptions; for example, what is the minimum
size of landscape it would be appropriate to use
the model on? What are the operational definitions (such as open road)? Given this information,
it is up to practitioners to decide how to apply the
model.
5
Final requests
# There was a significant amount of interest in extending the model to include southwest Oregon. To
do this, the modeling team would need elk location data. Ten GPS-collared elk would be a start,
although even 6 or 8 might be enough if you’re getting at least 4 locations a day. In addition, it
would be ideal to collect nutritional data on elk in the diverse vegetation systems that occur here.
# There were several requests for cross-walks to other plant communities not represented in the nutri-
tion model. This is an issue that the modeling group will work on before the model is finalized for
publication and use.
# Workshop participants asked if fire regime considerations
could be added. The model is “static” in that it evaluates
landscape conditions for a given point in time. However,
any landscape conditions can be simulated based on fire or
other disturbance regimes, with the use of a variety of
commonly available software packages (e.g., VDDT models). Projections from these landscape simulation models
can then be used to evaluate future habitat conditions for
elk, by using results from the simulations as covariates in
the elk habitat model.
# In the model, canopy cover less than 40 percent is consid-
ered forage. If this threshold were lower, areas defined
as forage would be substantially more short-lived. For
example, many westside areas achieve overstory canopy cover of more than 20 percent within five years of
timber harvest. Our analyses suggest that elk are responding to a cover-forage edge defined by the 40 percent canopy cover threshold.
# There were concerns about institutional mechanisms for
user support and updating data. This is an issue that
relates to all customized modeling products developed
by a variety of management agencies and partners. Unfortunately, most management agencies do not have
the institutional structure in place for long-term formalized support. Instead, developers of customized models
typically respond as best as they can, in concert with
users, to address support issues. This is the current
situation with the proposed westside elk habitat model.
# The model input covariates and predicted levels of elk use
were created using ArcGIS software. Code for these
processes will be provided to potential users. Other delivery methods are also being explored.
Workshop summary prepared by Rachel White, USFS PNW Research Station.
How can you stay involved?
We are recruiting beta testers from the
applied world so we can continue to
improve the management utility of the
models. After we release the software,
we will have another workshop. Visit
our project web site to stay informed:
http://www.fs.fed.us/pnw/calendar/
workshop/elk/.
For more information, contact:
Mary M. Rowland
Wildlife Biologist
USDA Forest Service
1401 Gekeler Lane
La Grande, OR 97850
541-962-6582 (voice)
mrowland@fs.fed.us
Mike Wisdom
Research Wildlife Biologist
USDA Forest Service
1401 Gekeler Lane
La Grande, OR 97850
541-962-6532 (voice)
mwisdom@fs.fed.us
6