Disparity in the Distribution of Suitable Habitat: A Metric for Capturing and Tracking Responses to Urbanized Landscapes

Abstract

Global species observations from community science platforms offer an unprecedented opportunity to analyze large-scale biodiversity patterns. This holds great promise for developing data-driven metrics to measure species responses to landscape-level environmental changes, essential for ameliorating the ongoing biodiversity crisis. A primary cause of declines in local, native species is the expansion and intensification of urbanized landscapes. A species response to urbanization can remain consistent across spatial and temporal scales. However, city-specific factors, such as land-use dynamics or local biotic interactions, can profoundly influence species and cause shifts in responses. This underscores the importance of measuring responses to urbanized landscapes at the local population level in a way that facilitates spatial and temporal comparisons. I propose utilizing the receiver operating characteristic (ROC) and its area under the curve (AUC), a statistical classifier, in a novel way for this purpose. Specifically, the disparity in suitable habitat between urban and adjacent non-urban areas is summarized by a single metric, HDURBAN, which facilitates population-level comparisons across time and space. I apply it to geographic predictions of the realized niche of 1,023 species from a highly urbanized landscape in California, USA. Additionally, modeling virtual species for each of the urban response types introduced by Blairs seminal 1996 study provides place-based, expected HDURBAN scores for urban avoider, urban utilizer, and urban dweller species. Validation efforts demonstrate accurate predictions of expected urban responses of real species in three scenarios. First, following a route-of-introduction hypothesis, non-native species exhibit significantly higher HDURBAN than native species. Second, assessing the robustness of urban responses in space and time, HDURBAN is significantly correlated to the urban response of the same 38 bird species from Blairs study, conducted 450 km apart and 25 years earlier. Third, testing for sensitivity to temporal shifts within species, HDURBAN decreases significantly in a species displaced from urban areas due to negative interactions with a recently introduced, congeneric species. In summary, HDURBAN is a versatile metric that provides a data-driven way to compare species responses, supporting future research into the patterns and processes underlying spatial and temporal variation in responses to urbanized landscapes.