The integration of airborne remote sensing data and field- based territory mapping applied to the understanding of habitat selection by woodland birds

摘要

Data derived from remote sensing has long been employed in the fields of ecology and species conservation. Optical data from satellite-borne instruments (e.g. Landsat Thematic Mapper) has commonly been used to map large-scale habitats and estimate species distributions and abundance through extrapolation. Detailed analyses of habitat structure are limited, however, by the two-dimensionality and relatively coarse resolution (commonly -30 m) of such imagery when compared to tools such as airborne lidar. High spatial resolution (< 1 m) lidar can characterise the three-dimensional structure of large areas of habitat in great detail and, as such, is ideal for the analysis of ecological patterns and processes in the spatially complex environments of woodlands. The field of bird ecology has seen the application of recent advances in the ability of lidar to characterise the multi-layered structure of woodland habitats, such as the distribution of understorey vegetation below the tree canopy, and the availability of dispersed resources such as dead trees. Previously limited to ground-based sampling methods of vegetation analysis, bird ecologists can now quantify the vegetation structure of entire bird territories, or buffers around nest sites or other areas of interest, in unparalleled detail. A limitation of lidar in these applications is the inability of the technology to discriminate between different species of tree, except for broad-scale distinctions between coniferous and broadleaved trees, which may be pertinent to the understanding of some bird habitat requirements. Integrating lidar data with that from a satellite-derived optical instrument is often inappropriate due to the disparity in spatial resolution, but a solution lies in the use of airborne instruments, such as Airborne Thematic Mapper (ATM), which can provide optical imagery at a similar resolution to lidar data. NERC'S ATM data have been used to map the tree species composition of an entire woodland canopy and, when integrated with lidar data of vegetation structure and bird territory and nest location data derived from field surveys, this can enable powerful analyses of bird-habitat interactions that are unprecedented in woodland bird research. We have applied such datasets to the problem of defining habitat selection by a declining woodland bird, the Marsh Tit (Poecilepalustris). Marsh Tits are small (10-12 g) birds that are restricted to woodland environments, occupying large (4-5 ha) year-round territories, and have suffered a population collapse of 68% between 1967 and 2004. The reasons for the decline are currently unknown, although deterioration in habitat quality has been posited as a potential factor. Studies of Marsh Tit habitat selection using ground-based field sampling have produced conflicting results, however, with some identifying woodland canopy characteristics as most important, and others the characteristics of the understorey. By integrating comprehensive, high-resolution datasets of woodland structure and composition with data of the spatio-temporal pattern of Marsh Tit occupation, we have shown that Marsh Tit habitat selection operates throughout the woodland's fall vertical profile combined with selection for certain tree species around the nest site. Such information is valuable for prescribing woodland management objectives aimed at the conservation of this species.