Print PageAdvances in computing power and quantitative methods increasingly allow for estimation of occupancy and structural connectivity on the basis of diverse data inputs, including data that are not collected in the field. Occupancy is the expected probability that a given location is occupied by a species. Structural connectivity refers to contiguity of a species’ habitat. Models of occupancy and connectivity that are informed by detailed understanding of a species’ natural history and include measures of habitat quality and configuration typically have lower uncertainty, thus are more relevant to management and policy, than models that rely on coarse-grain metrics of land cover or climate. Additionally, the outputs of some types of models of occupancy and connectivity have less uncertainty than bioclimatic envelope models or so-called niche models. The fit and predictive capacity of models of occupancy further are improved by including time-series data reflecting dynamic processes such as local colonization and extirpation, whereas models of connectivity are improved by data on a species’ movements and genetics.
We applied multiseason occupancy models to examine probabilities of detection, occupancy, colonization, and local extinction in multiple species of breeding birds in the Great Basin and in giant garter snakes (Thamnophis gigas) in California’s Central Valley. Input data were collected over 6 to 10 years by individuals who were highly familiar with regional ecosystems. We modeled occupancy of all taxa as functions of abiotic and biotic covariates, most of which were measured in the field and some of which were derived from remote sensing data. Reliable outputs from occupancy models serve as inputs to models of connectivity. We used the computer program Zonation to examine current connectivity for breeding birds and potential connectivity given multiple, spatially explicit scenarios of changes in climate and land cover.
Results/Conclusions
Occupancy outputs allowed inference to phenomena such as different responses to the same environmental variables by difference species in putative guilds. Such inferences would not have been possible without knowledge of natural history and collection of ground data. Strong inferences about current drivers of species distribution improve the realism of projections of responses to environmental change. Grounding the connectivity models and interpretations in natural history minimized uncertainty and increased potential for practical application. Collaborations with state and federal resource agencies and private landowners who are end-users for the work suggest that incorporation of natural history into cutting-edge models increases the potential that science will inform decision-making.
