RETHINKING HABITAT AND HOW WE STUDY HUMAN-WILDLIFE RELATIONSHIPS

Abstract

Current methods to model species habitat use through space and diel time are limited. Development of such models is critical when considering rapidly changing habitats where species are forced to adapt to anthropogenic change, often by shifting their diel activity across space. The first chapter of this manuscript focuses on redeveloping occupancy models to incorporate hypotheses on species diel habitat use. This alternative occupancy framework, called the multi-state diel occupancy model (MSDOM), can evaluate species diel activity against continuous response variables which may impact diel activity within and across seasons or years. We used two case studies on fosa, a mesocarnivore endemic to Madagascar, and coyote in Chicago, USA, to conceptualize the application of this model and to quantify the impacts of human activity on species’ spatial use in diel time. We found support that both species altered their diel activity across intensity of human disturbance—in and across years, and by degree of human disturbance. Our results exemplify the importance of understanding animal diel activity patterns and how human disturbance can lead to temporal habitat loss. This adapted model will allow future studies to answer explicit questions in regards to species diel habitat use and direct conservation efforts to protecting habitats over shorter, diel, periods. Chapter two of this manuscript focuses on incorporating human dimension research to understand relationships between people and wildlife. Human dimension research in ecology is especially needed in urban landscapes where more wildlife are living among and adapting to human dominated landscapes. Thus, we focus on understanding the complex drivers of human-wildlife relationships that have become increasingly important for managing both people and wildlife. A common approach to researching these drivers is via survey questionaries and the use of Likert items and scales, which require analytical techniques that handle their unique structure. Here, we apply a hierarchical Bayesian modeling framework to conduct ordinal regression that is well suited to Likert response data and allows the evaluation and comparison of model hypotheses. Our case study focuses on two objectives, understanding how people value coyotes and the frequency in which people interact with coyotes. We measured how people value coyotes with a Likert scale on peoples perceived risks and benefits of having coyotes on a landscape and measure frequencies of interactions with two Likert items on people’s sightings and incidents (growling, stalking attacking people or owned animals) with coyotes. We investigated how people’s demographics, knowledge of coyotes, and