REGIONAL VARIATION IN HOME‐RANGE‐SCALE HABITAT MODELS FOR FISHER (MARTES PENNANTI) IN CALIFORNIA

Abstract

Abstract. We analyzed recent survey data and mapped environmental variables integrated over a home range scale of 10 km 2 to model the distribution of fisher (Martes pennanti) habitat in California, USA. Our goal was to identify habitat factors associated with the current distribution of fishers in California, and to test whether those factors differ for widely disjunct northern and southern populations. Our analyses were designed to probe whether poor habitat quality can explain the current absence of fishers in the historically occupied central and northern Sierra Nevada region that separates these two populations. Fishers were detected at 64/433 (14.8%) sample units, including 35/111 (32%) of sample units in the Klamath/Shasta region and 28/88 (32%) of sample units in the southern Sierra Nevada. Generalized additive models (GAM) that included mean annual precipitation, topographic relief, forest structure, and a spatial autocovariate term best predicted fisher detections over the species' recent historical range in California. Models derived using forest structure data from ground plots were comparable to models derived from Landsat Thematic Mapper imagery. Models for the disjunct Klamath/Cascades and southern Sierra Nevada populations selected different environmental factors and showed low agreement in the spatial pattern of model predictions. Including a spatial autocovariate term significantly improved model fits for all models except the southern Sierra Nevada. We cannot rule out dispersal or habitat in explaining the absence of fishers in the northern and central Sierra Nevada, but mapped habitat quality is low over much of the region. Landscapes with good fisher habitat may exist in rugged forested canyons of the currently unoccupied northern Sierra Nevada, but these areas are fragmented and at least 60 km from the nearest recent fisher detections.

Characterizing Habitat Elements and Their Distribution over Several Spatial Scales: The Case of the Fisher

“…The fact that our study area has one on-the-ground inventory plot every 1.619 ha (4 acres) means that we can model fisher habitat at a fine scale as well as a range of scales which is not possible when inventory plots are infrequent (e.g., 1 FIA plot every 6000 acres) or when fine scale high resolution light detection and ranging (LIDAR) and hyperspectral imagery is not available. This is especially true when home range territories of the fisher range from~527 ha (1302 ac) and~5806 ha (14,347 acres) [23].…”

Estimating habitat value using forest inventory data: The fisher (Martes pennanti) in northwestern California

“…This is a different ecoregion than the southern Sierra, with different climatic and topographic influences. Fisher habitat, diet and genetics differ substantially between the two populations (Zielinski et al, 1999;Wisely et al, 2004;Zielinski et al, 2004a,b;Golightly et al, 2006;Davis et al, 2007). Because habitat features important to fishers differ between the regions, methods to evaluate and monitor resting habitat will likely differ between the regions as well.…”

Developing and testing a landscape-scale habitat suitability model for fisher (Martes pennanti) in forests of interior northern California

“…It is easily detected using noninvasive survey methods (remote cameras: Kays andSlauson, 2008, track-plate stations: Ray and and these methods have been used in standard protocols Zielinski et al, 2005) to generate systematically collected, independently verifiable (McKelvey et al, 2008) and spatially precise survey data. Resulting data on the detections and non-detections have been used to build statistical landscape-scale habitat models for the fisher in northwestern California (Carroll et al, 1999;Davis et al, 2007), portions of the Sierra Nevada (Campbell, 2004;Davis et al, 2007;Spencer et al, in press) and statewide (Davis et al, 2007).…”

“…Unlike other regions of California (e.g., Zielinski et al, 2005), prior to our work there was no systematically collected survey data for the majority of our study region. Nonetheless, predictions for the western portion of our study region were derived from a habitat modeling exercise in northwestern California (Carroll et al, 1999) and predictions for the entire study region were derived from characteristics associated with fisher detections elsewhere (Davis et al, 2007). However, characteristics of fisher habitat, diet and genetics, vary substantially in different parts of the state (Zielinski et al, , 2004aWisely et al, 2004;Golightly et al, 2006;Davis et al, 2007).…”

“…Nonetheless, predictions for the western portion of our study region were derived from a habitat modeling exercise in northwestern California (Carroll et al, 1999) and predictions for the entire study region were derived from characteristics associated with fisher detections elsewhere (Davis et al, 2007). However, characteristics of fisher habitat, diet and genetics, vary substantially in different parts of the state (Zielinski et al, , 2004aWisely et al, 2004;Golightly et al, 2006;Davis et al, 2007). The poor performance of a landscape habitat model for the southern Sierra Nevada at predicting fisher detections in northern California and vice versa (Davis et al, 2007) demonstrates the need to develop and test specific models for the Klamath region.…”