UC Berkeley

One of the most fundamental questions facing ecologists is: why do animals live where they do? Patch occupancy depends on a myriad of biotic and abiotic factors, any of which may encourage or discourage the presence of a species. Understanding the relationship between occupancy and environmental characteristics is integral to managing and conserving species in a dynamic environment. This dissertation studies avian occupancy in widely dispersed emergent wetlands in the Sierra Nevada foothills of California and relates occupancy to the characteristics of wetland patches, behaviors of the birds using those wetlands, and landscape composition. The first chapter of this dissertation focuses on violations of the assumption of closure in occupancy models for two secretive marsh birds, Black Rails (Laterallus jamaicensis) and Virginia Rails (Rallus limicola). For occupancy models, a key assumption is that there is no immigration and emigration between survey periods. Violating this assumption could overestimate occupancy and lead to an improper understanding of the characteristics that influence site occupancy. I found that there were significant closure violations for both Black and Virginia Rails, although the characteristics of those violations differed. Black Rails were more likely to colonize wetlands between surveys, and the wetlands colonized were those that were occupied in the previous year. Virginia Rails were more sensitive to environmental changes and would abandon drying wetlands more readily than Black Rails. The second chapter of this dissertation uses a multispecies occupancy model to understand the importance of spillover effects on occupancy across the entire avian assemblage using wetlands in the Sierra Nevada Foothills. The presence of animals in a patch of habitat is dependent not only on the characteristics of that patch, but also the landscape surrounding it. I investigated whether there was a spillover effect from matrix habitats such as grassland and forest on wetland bird species or the reverse, a spillover effect from the wetlands on species inhabiting the matrix habitat surrounding each wetland. I observed spillover effects in both directions, with matrix species assemblages depending on wetland water source and wetland species assemblages depending on the landscape composition around the wetland. The final chapter of this dissertation uses aerial remote sensing to assess Black and Virginia Rail habitat, compares the predictive power of remote sensing to ground-truthed data, and assess the ability of occupancy models to predict rail occupancy at novel sites using only aerial imagery. For this chapter, I differentiated occupied habitat from unoccupied habitat using known locations and occupancy status at wetlands. I classified sites using a maximum likelihood classifier and high resolution imagery from the National Agriculture Imagery Program. I found that raw spectral reflectance accurately predicted wetland occupancy for both Black and Virginia Rails, although the effectiveness of characterizing a wetland varied between years. For Black Rails, spectral reflectance was most similar to the wetland structure, whereas for Virginia rails, spectral reflectance was most similar to wetland wetness. However, in both cases, spectral reflectance was informative when included alongside ground-collected data. Although the data collected in this dissertation are focused on a very specific habitat type and location, my results clearly demonstrate the importance of biological context on understanding animal occupancy. My results are broadly applicable in other study systems and help inform conservation strategies for multiple species. By understanding landscape composition, the drivers of animal movement, and the biotic and abiotic factors correlated with occupancy, I can better predict changes in animal populations in an increasingly changing environment.