Combining potential and realized distribution modeling of telemetry data for a bycatch risk assessment.

Establishing marine species distributions is essential for guiding management and can be estimated by identifying potential favorable habitat at a population level and incorporating individual-level information (e.g., movement constraints) to inform realized space use. In this research, we applied a combined modeling approach to tracking data of adult female and juvenile South American sea lions (;  = 9) from July to November 2011 to make habitat predictions for populations in northern Chile. We incorporated topographic and oceanographic predictors with sea lion locations and environmentally based pseudo-absences in a generalized linear model for estimating population-level distribution. For the individual approach, we used a generalized linear mixed-effects model with a negative exponential kernel variable to quantify distance-dependent movement from the colony. Spatial predictions from both approaches were combined in a bivariate color map to identify areas of agreement. We then used a GIS-based risk model to characterize bycatch risk in industrial and artisanal purse-seine fisheries based on fishing set data from scientific observers and artisanal fleet logs (2010-2015), the bivariate sea lion distribution map, and criteria ratings of interaction characteristics. Our results indicate population-level associations with productive, shallow, low slope waters, near to river-mouths, and with high eddy activity. Individual distribution was restricted to shallow slopes and cool waters. Variation between approaches may reflect intrinsic factors restricting use of otherwise favorable habitat; however, sample size was limited, and additional data are needed to establish the full range of individual-level distributions. Our bycatch risk outputs identified highest risk from industrial fisheries operating nearshore (within 5 NM) and risk was lower, overall, for the artisanal fleet. This research demonstrates the potential for integrating potential and realized distribution models within a spatial risk assessment and fills a gap in knowledge on this species' distribution, providing a basis for targeting bycatch mitigation outreach and interventions.