Modeling the effects of breeding versus winter habitat loss on the population dynamics of a Neotropical migratory songbird
Many long distance migratory songbirds that breed in North America have experienced strong declines in recent decades. Habitat loss, fragmentation or degradation have been proposed as the most likely causes of songbird declines but the relative importance of breeding versus non-breeding habitat to the decline of any species is uncertain. Optimal conservation requires predicting where habitat protection, or other measures, will have the biggest impact on population recovery. The wood thrush, Hylocichla mustelina, has become a symbol of neotropical migratory songbird declines, having declined by more than 50% since the mid-1960s. Estimates from the large scale breeding bird survey (BBS) show that these declines are not evenly distributed across the breeding range. Recently, the migratory connectivity of wood thrush has been mapped by tracking ~100 of individual birds from different wintering and breeding regions with solar geolocation devices (geolocators). We applied a novel population model framework, a migratory network to wood thrush to explain the pattern of declines and to predict future population trends. We used an integrated population model approach to estimate demographic parameters by maximizing the likelihood of the observations from geolocator data as well as observations of patterns of declines from BBS.
We divided the North American breeding range of wood thrush into four regions (breeding nodes) and the Central American non-breeding range into three regions (winter nodes). Our best-fit network model gives a close match to the observed pattern of declines between 2000 and 2010, the observed relative abundances among breeding regions which are estimated also from BBS data, and to the connectivity estimated from geolocators. We estimated forest cover and forest loss in each node from remote sensed data and show that even though proportional forest loss is greatest in the breeding nodes over the period from 2000-2010, it is the tropical deforestation, i.e. loss of winter habitat, that has been the key driver of species-level population declines since 2000. If forest loss in the winter range, particularly in the Eastern winter node, continues at the same rate, we estimate the population will continue to decline and be at 35% of its current size by 2025. We show that migratory network model is a powerful framework for exploring the causes and consequences of breeding and non-breeding ground deforestation on species-level and regional population declines.