Independent effects of woodland loss and fragmentation on Brown Treecreeper distribution

Abstract

We examined the influence of local habitat and the surrounding landscape on the distribution of Brown Treecreepers in a matrix of woodlands and pastures. Our goals were to: (1) determine the importance and scale of the independent effects of woodland cover and fragmentation on treecreeper distribution, and (2) employ landscape variables to improve models of treecreeper distribution based on local habitat features. Woodland fragmentation was important at a large scale while both woodland cover and fragmentation were important at a smaller scale. Excluding unoccupied sites in highly fragmented landscapes improved the ability of local habitat features to explain Brown Treecreeper distribution, which appeared to be constrained by cavity density. Brown Treecreepers' response to fragmentation at the larger scale may occur because fragmentation disrupts dispersal. Alternatively, their response may be an example of a general phenomenon of fragmentation effects only arising when < 20% of woodland cover remains at a given scale. As fragmentation increases, so does the need to incorporate landscape patterns into wildlife-habitat models.

Introduction

The theory of island biogeography (MacArthur and Wilson, 1967) and principles of metapopulation dynamics (Hanski and Gilpin, 1991) suggest that the amount and spatial configuration of habitats in the landscape will influence persistence and spatial distribution of wildlife. Composition of avian communities is influenced by the amount and configuration of natural and human-dominated patches in the landscape (McGarigal and McComb, 1995, Flather and Sauer, 1996, Jokimaki and Huhta, 1996, Bolger et al., 1997, Saab, 1999). Habitat fragmentation alters the amount, configuration, and sometimes the quality, of habitat and is therefore expected to affect bird distribution. However, most studies of habitat fragmentation and birds occur at the scale of single patches, rather than landscapes, with a few exceptions (Villard et al., 1999, Trzcinski et al., 1999). It is unclear whether findings from patch studies can be extrapolated to the landscape level (Wiens et al., 1993).

Barrett et al. (1994) and Barrett (1995) documented the decline of the Brown Treecreeper (Climacteris picumnus) in fragmented habitat in the New England Tablelands of northeastern New South Wales. Walters et al. (1999) found the pattern of decline was characterized by the presence of unpaired males in fragmented habitat. Cooper and Walters (2002) demonstrated experimentally that unpaired males in fragmented woodland were not in poor quality habitat; rather, recruitment of juvenile females into isolated fragments was disrupted. They found that spatial distribution of highly fragmented habitat did not permit successful dispersal and recruitment, which occurred in more contiguous habitat. Brown Treecreepers in central New South Wales are not found in patches more than 700 m from the next nearest patch of 10 ha or larger (S. Briggs, personal communication). Thus, distribution of Brown Treecreepers appears to be influenced by landscape patterns.

Wildlife managers have focused on modelling the relationship between animal abundance and local habitat features, rather than the effects of distribution of habitats across a landscape. Using models to correctly identify suitable habitat is critical to conservation. For example, the US Endangered Species Act of 1976 is designed to protect critical habitat, even if the habitat is currently unoccupied (Schreiner, 1976). Several US governmental agencies responsible for managing wildlife use habitat models built from distribution data, such as habitat suitability index models and habitat-capacity models (Anderson and Gutzwiller, 1994). Wildlife managers also use models of wildlife–habitat relationships to make predictions about species distribution patterns at regional scales (e.g. GAP analysis project; Scott et al., 1993, Edwards et al., 1996).

A detailed understanding of the habitat requirements of Brown Treecreepers is needed to effectively conserve this locally threatened species. In this case, however, an accurate habitat model must include landscape effects. If landscape patterns strongly influence distribution patterns, the identification of critical habitat features can be confounded by a lack of individuals in good quality habitat in unsuitable landscapes. Similarly, if habitat features strongly influence distribution patterns, the identification of critical landscape patterns can be confounded by a lack of individuals in poor quality patches in suitable landscapes. Another difficulty is the time delay in response of the population to landscape changes, which can result in the continued presence of individuals in unsuitable landscapes. All species are patchily distributed at some scale and patterns of aggregation can become exaggerated by habitat loss and fragmentation (Simberloff, 1995). Brown Treecreepers are naturally patchily distributed because they are cooperative breeders that exhibit a high frequency of territorial budding, resulting in clusters of groups or clans (Noske, 1982a, Walters et al., 1999). In addition, they avoid degraded woodlands and forested areas on steep rocky hills and gorges (Noske, 1982a, Noske, 1982b).

We examined the distribution of Brown Treecreepers in relation to local habitat characteristics as well as landscape characteristics, specifically, the independent effects of woodland cover (cover=the total amount of a specified habitat) and fragmentation (fragmentation=the spatial arrangement of a specified habitat in the landscape). Measures of cover and fragmentation can vary considerably as the size of the landscape examined changes. Therefore, we analyze the effects of cover and fragmentation at multiple scales or landscape sizes. Fragmentation can also affect habitat quality through edge effects, but this study does not address edge effects, because other research showed their influence on Brown Treecreeper ecology to be minimal (Cooper and Walters, 2001). Our goals were to: (1) determine the scale and importance of the independent effects of woodland cover and fragmentation on the distribution of Brown Treecreepers, and (2) use landscape variables to improve models of Brown Treecreeper distribution based on local habitat features.