For more information on species & ecosystem science:

Wildlife Science

Fish Science

Habitat Science


Lead Scientist: Jim Watson

Ecoregions: Columbia Plateau

Ecological Systems: Inter-Mountain Basins Big Sagebrush Steppe, Inter-Mountains Basins Big Sagebrush Shrubland, Columbia Plateau Steppe and Grassland

Closeup photo of a Washington ground squirrel
Photo by Bob Davies

Washington ground squirrel

Map of study areas
Click on map to enlarge

Four study areas where Washington ground squirrels were surveyed in eastern Washington in 2008 and 2009. The inset shows a hypothetical survey grid with random survey quadrants (blue squares) and squirrel detections (red dots).

WDFW personnel conducting survey
Photo by Steve Germaine

Detection surveys were conducted along line transects in areas with previously documented use by Washington ground squirrels


Shrubsteppe Ecology

Occupancy Modeling of Washington Ground Squirrels Populations

Project Description

Small mammals are important to all ecosystems, and the shrubsteppe ecosystem is no exception.  Ground squirrels and other burrowing small mammals serve important ecologic functions such as reducing soil compaction and improving seed dispersal.  They are also important in the food chain and are a dietary staple for a host of avian and mammalian predators.

Many native ground squirrels in Washington have declined dramatically in recent years largely due to agricultural conversion and extermination.  One of those species, the Washington ground squirrel (WAGS), is a state candidate for listing in Washington and listed as endangered in Oregon.  The WAGS has been the subject of considerable research in both states, including habitat and distribution studies conducted by the Washington Department of Fish and Wildlife (WDFW).  Locating and cataloguing the small remnant populations of WAGS is vital to maintaining the species, and this information is often gathered by conducting surveys along transects to look for squirrel sign and listen for squirrel calls.  The fossorial lifestyle and behavioral mechanisms for predator avoidance make detection of squirrels and inexact science: were squirrels actually present but undetected? What is the probability of detecting squirrels in a population on a given visit if the site is occupied? What environmental factors or survey methods might affect detection? Answers to these questions improve our ability to detect squirrels and help us to interpret the results of our surveys.

In 2008 we initiated a project to model occupancy and detection of WAGS in eastern Washington.  We identified four study areas where use by WAGS had been documented in recent years.   In each study area we established a 1 km x 1 km grid divided into 100 quadrants, half of which we randomly selected and surveyed for ground squirrels.   Squirrels were detected from the presence of scat, vocalizations, or observation.  We simultaneously collected information related to survey conditions, such as weather and time of day, as well as quadrant characteristics including visibility, and presence of rock cover.  The survey was repeated three times during the season in order to model detection. 

 We used program PRESENCE to model squirrel occupancy and detection and identify important variables related to these parameters.  The entire study was conducted again 2009 to collect information on new variables and refine data collection methods for variables we identified as important during 2008 surveys. 

Key Findings

  • 1,008 surveys were conducted during combined seasons with 366 squirrel detections.  The proportion of sites where squirrels were detected at least once (naïve estimate) was in 2008 was 0.52, and 0.63 in 2009. Squirrel detectability was a function of pre-survey counts of recently-used burrows, and surveyor effects, in both 2008 and 2009. Neither relationship was unexpected, because we presumed active burrow counts would provide an index for detectability.  In spite of pre-survey training, different abilities of surveyors to detect squirrels still was a factor influencing detection rates.
  • The most influential site characteristics related to squirrel occupancy in the 2009 analysis were the degree of shrub cover and protective rock cover, both positive relationships. Badger sign and site visibility, as influenced primarily by ground vegetation, were also positively related to squirrel occupancy.
  • Weather conditions were the most influential survey conditions on detections, with higher detection rates a function of lower wind speeds and lower barometric pressure.
  • Through the model ranking process we have identified key sampling and site covariates that allow estimation of squirrel occupancy using the most efficient survey methods for detecting squirrels in different environmental conditions.   

What’s New

We plan to conduct a third season of surveys in 2010 to further refine occupancy and detection models.


  • U. S. Fish and Wildlife Service, Section 6 Program


  • Scott Anderson
  • Todd McCollough
  • Robert Davies
  • Tiffany Baker
  • John Greenfield