Multi-scale habitat models for reintroduced bird populations: a case study of South Island saddlebacks on Motuara Island

Understanding resource selection by animals is important when considering habitat suitability at proposed release sites within threatened species recovery programmes. Multi-scale investigatory approaches are increasingly encouraged, as the patchy distribution of suitable habitats in fragmented landscapes often determines species presence and survival. Habitat models applied to a threatened New Zealand forest passerine, the South Island saddleback (Philesturnus carunculatus carunculatus), reintroduced to Ulva Island (Stewart Island) found that at landscape scale breeding pairs’ preferences for sites near the coast were driven by micro-scale vegetation structure. We tested these results by examining models of breeding site selection by a reintroduced saddleback population on Motuara Island (Marlborough Sounds) at two scales: (1) micro-scale, for habitat characteristics that may drive breeding site selection, and (2) landscape scale, for variations in micro-scale habitat characteristics that may influence site colonisation in breeding pairs. Results indicated that birds on Motuara Island responded similarly to those on Ulva Island, i.e. birds primarily settled at the margins of coastal scrub and forest and later cohorts moved into larger stands of coastal forest where they established breeding territories. Plant species composition was also important in providing breeding saddleback pairs with adequate food supply and nesting support. However, Motuara Island birds differed in their partitioning of habitat use: preferred habitats were used for nesting while birds were foraging outside territorial boundaries or in shared sites. These differences may be explained because Motuara has a more homogeneous distribution of microscale habitats throughout the landscape and a highly bird-populated environment. These results show that resource distribution and abundance across the landscape needs to be accounted for in the modelling of density–bird–habitat relationships. In the search for future release sites, food (invertebrates and fruiting tree species) should be abundant close to available nesting sites, or evenly spread and available throughout the landscape.