Forest overstorey and age as habitat? Detecting the indirect and direct effects of predators in defining habitat in a harvested boreal landscape

Abstract Given the importance of wildlife habitat protection in meeting land use management objectives, criteria for “habitat” identification are surprisingly amorphous. For example, while much current habitat modeling has tended to avoid the term “niche modeling”, niche assumptions are implicit – meaning the presence of predators and competitors is essential to whether or not a species uses, or will use, an area. In this paper, we examine environmental variables associated with woodland caribou ( Rangifer tarandus caribou ) presence in the boreal forests of northwestern Ontario, Canada. Based on winter aerial surveys conducted in 2010–2013, we used logistic regression to identify important habitat characteristics (and/or their surrogates) and structural equation modeling (SEM) to explore causal and indirect caribou habitat relationships, at broad and fine scales. The best-fit regression model (>10,000 ha) to explain caribou presence, at the broad scale, included the presence of moose ( Alces alces ), wolves ( Canis lupus ), logging roads and primary roads, and all top models included wolves. In contrast, at the fine scale (