With a population estimate of 99 pairs, the Madagascar fish-eagle (Haliaeetus vociferoides) is one of the rarest birds of prey in the world. I investigated the ecological requirements of the Madagascar fish-eagle in 1994 and 1995 to help determine management action to prevent its extinction. I investigated fish-eagle foraging ecology in 1996 to determine its prey preference and whether fish abundance and availability affect fish-eagle foraging rates and foraging success. Madagascar fish-eagle nest and perch trees were taller, broader, had more unobstructed branches, and had a greater arc of accessibility than unused trees. Perch trees also were deciduous more often and had a narrower growth form than unused trees. Nest sites had more shoreline perch trees than unused sites. Lakes occupied by fish-eagles were deeper and clearer, and had more shoreline perch trees, more fish, a greater total fish weight, and more fish species than unoccupied lakes. I developed logistic regression models to predict the probability of Madagascar fish-eagle use based on the measured habitat variables. Nest and perch tree models included tree height. The nest site model included number of shoreline perches. Lake models included number of shoreline perches and either number of fish, total fish weight, or number of fish species. These models can be used to predict fish-eagle habitat use with > 70% accuracy. Introduced tilapia, Oreochromis spp. and Tilapia spp., made up the majority of both the gill net (66.3%) and fish-eagle catch (64.7%) in similar proportion, which suggests that the fish-eagle is an opportunistic predator. Replacement of native fish species by exotics, thus, probably has not been detrimental to the island's fish-eagle population. Male fish-eagle foraging success was positively correlated with number of fish, total fish weight, and number of fish species, which suggests that declines in the fish population could adversely affect the fish-eagle population. The results of this study indicate that Madagascar fish-eagles require bodies of water with large shoreline trees and an ample fish population. I recommend greater protection of aquatic habitats, monitoring and management of freshwater fish populations, and education of local people in sustainable tree harvesting practices. / Ph. D.

Northern Goshawk breeding habitat was conceptualized and analyzed as a system composed of different sized components organized within a hierarchy. The organization of the hierarchy was asymmetric with smaller components nested within and subordinate to larger components. Components considered in the breeding habitat hierarchy of goshawks were nest tree, nest site, nest territory, foraging area, and landscape. Choice at any level within the hierarchy was contrasted with the next level in the hierarchy. Thus it was found that goshawks chose nest trees with a largerdbh (79.0 ± 4.4 cm) than the nest site average (43 ± 2.7 cm) (p=0.000036); the averagedbh of the nest site (43.2± 3.1 cm) was significantly larger than the territory (33.2± 1.9cm) (p=0.018). The nest territory when contrasted with random unoccupied locations in the landscape demonstrated a composition of a central site of trees with largedbh and fewer stems when compared with the surrounding forest. I hypothesize that this pattern reduces thepredation risk at the nest site. I investigated the nest site relative to the assumed foraging area (radius of 3000m; approximately 3000ha). Goshawks selected nest sites which were significantly further from young clear cuts (>10 ha). The average distance from a nest site to a recentclearcut greater than 10 ha was 1,350 m of the nest while for unoccupied sites the average distance was 770 m (p = 0.011). Finally, at the landscape level, areas of approximately 10,000 ha, goshawks on Vancouver Island were more abundant, had more nest sites, and had higher productivity per active nest in old growth forests (>120 years) than in either second growth (majority less than 80 years) or fragmented landscapes. The majority of goshawk prey, as determined though pellet analysis was red squirrel (69%). Red squirrels were equally abundant in second growth and old growth yet less abundant in fragmented landscapes. Despite similar abundance of their main prey item goshawks were not as abundant in second growth forests as in old growth forests. I conclude that the structure of second growth and fragmented landscape reduces the availability of key prey species and therefor the suitability of these landscape fpr goshawks is lower. / Graduate