Search

Coastal habitats have experienced significant degradation and fragmentation in recent decades under the strain of interacting ecosystem stressors. To maintain biodiversity and ecosystem functioning, coastal managers and restoration practitioners face the urgent tasks of identifying priority areas for protection and developing innovative, scalable approaches to habitat restoration. Facilitating these efforts are models of seascape connectivity, which represent ecological linkages across heterogeneous marine environments by predicting species-specific dispersal between suitable habitat patches. However, defining the suitable habitat patches and migratory pathways required to construct ecologically realistic connectivity models remains challenging. Focusing on two reef-associated fish species of the Florida Keys, United States of America (USA), we compared two methods for constructing species- and life stage-specific spatial models of habitat suitability—penalized logistic regression and maximum entropy (MaxEnt). The goal of the model comparison was to identify the modeling algorithm that produced the most realistic and detailed products for use in subsequent connectivity assessments. Regardless of species, MaxEnt's ability to distinguish between suitable and unsuitable locations exceeded that of the penalized regressions. Furthermore, MaxEnt's habitat suitability predictions more closely aligned with the known ecology of the study species, revealing the environmental conditions and spatial patterns that best support each species across the seascape, with implications for predicting connectivity pathways and the distribution of key ecological processes. Our research demonstrates MaxEnt's promise as a scalable, species-specific, and spatially explicit tool for informing models of seascape connectivity and guiding coastal conservation efforts. [ABSTRACT FROM AUTHOR]