Diagnosing the cause of population changes: localized habitat change and the decline of the endangered St Helena wirebird.
Abstract
Statistical models of population density and/or change in relation to habitat could aid the management of endangered species and help diagnose causes of population decline. Such models seek 'global' explanations for any decline, but endangered species are often vulnerable to highly localized habitat change that can have a significant impact on overall population. In such cases, statistical models can aid in interpreting the impact of localized habitat change on population density and so have a role in diagnosis. We illustrate this approach using data from the endangered St Helena wirebird, a small sand plover endemic to the island of St Helena in the South Atlantic Ocean. The wirebird inhabits pasture habitats and semi-desert areas. Census data suggest that wirebird abundance declined during the early 1990s, from 425 adult birds in 1989 to 335 adult birds by 1998-99. Abundance declined by 34.4% in pasture habitats, whereas abundance in semi-desert increased by 18% over this period. We tested the hypothesis that the decline in abundance in pasture habitats was driven by vegetation change. Multiple regression modelling revealed that wirebird density in pasture was positively correlated with the cover of broad-leaved herbs, but negatively correlated with vegetation height and gorse cover prior to the decline in 1989. A performance test of this model showed that predictions were reliable. The cover of broad-leaved herbs declined significantly between 1989 and 1998-99. However, there was no correlation between the change in wirebird density and the change in broad-leaved herb cover, indicating that this vegetation change was unlikely to have caused the decline in wirebird abundance. The decline in the abundance of wirebirds in pasture habitats was primarily due to changes in 3 sites. Each experienced site-specific habitat changes, which were likely to cause a decrease in wirebird abundance on the basis of the multiple regression modelling of wirebird density in relation to habitat. We conclude that highly localized vegetation change, rather than general herb cover, plays an important role in the overall population decline. Our analysis provides clear habitat management recommendations: pasture management for livestock needs to be compatible with the production of herb-rich grassland, consisting of short vegetation and few woody plants, in order to benefit wirebird conservation. Management decisions for endangered species are often required urgently, and as a result are invariably based on incomplete information, particularly concerning the demographic mechanisms involved in decline. Our approach illustrates a role for simple statistical models under these circumstances. Such models help diagnose the cause of a decline and promote initial, remedial, management actions. This work also shows how habitat effects on endangered species can be highly localized.