Habitat selection by American martens (Martes americana) in coastal northwestern California

Abstract

The Humboldt marten, Martes americana humboldtensis, has undergone a dramatic decline throughout its historical distribution in coastal Northwestern California. There is currently only one known population occupying an area occurring in <5% of the historical distribution of the subspecies. Conservation and management efforts to benefit this population are hampered by lack of information on the habitat ecology of martens in the coastal forest of northwestern California. Furthermore there have been no investigations of the habitat ecology of marten populations anywhere in the coastal forests of the Pacific States. I investigated habitat relationships of the only known population of martens within the historical distribution of M a. humboldtensis at three spatial scales (microhabitat, stand, and home range) and in relation to four forest management regimes (industrial timberlands, and U. S. Forest Service (USFS) matrix lands, late-successional reserves, and wilderness). Over 12 months of fieldwork during 2000 and 2001, I detected martens at 26 of 159 track plate sample units distributed on a systematic grid located over the region known to be occupied by the population. I used an information-theoretic approach to rank 56 a priori candidate models that described hypothesized habitat relationships at each spatial scale. Marten detections occurred in two distinct habitat types, those with forests on serpentine soils and forests associated with more productive soil types, which are more common in the region. At the microhabitat scale in serpentine habitats, martens were detected at sites with dense shrub cover, sparse tree cover, and abundant surface rocks. Dense shrub cover and abundant surface rocks may provide key overhead and escape cover for martens in serpentine habitats. At the microhabitat scale in non-serpentine habitats martens were detected at sites having the most mesic aspects, with dense tree and shrub cover, and with a higher abundance of large diameter snags. At the stand scale martens selected conifer-dominated stands with dense shrub cover in the latest seral stages (old growth and late-mature) in non-serpentine habitats and variable seral stages in serpentine habitats. At the home-range scale the probability of detecting a marten decreased with increasing amounts of logging within 1-km of the sample unit and increased with increasing maximum patch size of old growth, old growth plus late-mature, or serpentine habitat within 1-km of the sample unit. Martens were detected significantly more frequently in USFS lands than in private industrial timberlands. Within USFS lands, martens were detected most frequently in matrix and late-successional reserves, and least frequently in the wilderness area. This study provides new information on the habitat ecology of martens in the coastal forests of northwestern California. It demonstrates the importance of investigating marten habitat at multiple spatial scales and provides insights to linkages among scales and how martens respond to forest management. It also provides information to aid conservation and restoration of martens in northwestern California through identification of areas currently occupied or with suitable habitat, information to identify suitable habitat in areas outside the study area, and information to guide conservation planning for martens and site-specific habitat restoration.