Spatial and temporal Bayesian approaches to predict future seabird-prey overlap and population change

• Main Authors: 3rd World Seabird Conference 2021, Scott, Beth
• Format: Article in Journal/Newspaper
• Language: unknown
• Published: Underline Science Inc. 2021
• Subjects: Ornithology; Environmental Engineering; FOS Environmental engineering; Laplace; Black-legged Kittiwake; common guillemot
• Online Access: https://dx.doi.org/10.48448/cge7-wt36; https://underline.io/lecture/34832-spatial-and-temporal-bayesian-approaches-to-predict-future-seabird-prey-overlap-and-population-change

Abstract: The marine environment is changing rapidly due to climate change and increasing anthropogenic activities. Understanding how the usage of spatial habitat by seabirds and their prey species may change with both of these pressures is essential for the predictions of population trends and current decision making on sustainable spatial management. Where both predator and prey species are highly mobile, it is important to predict how well their preferred habitats continue to overlap in future scenarios. The diversity of individually preferred habitat variables and drivers of those habitats may all be changing quite differently under the different pressures of climate change and anthropogenic activities. To predict future seabird-prey spatial habitat overlap we will present two synergistic approaches. The first is a spatial statistical Bayesian hierarchical approach called Joint Modelling with INLA (integrated Nested Laplace Approximation). Joint Modelling, as compared to typical single-species spatial distribution modelling, allows both predator and prey as a response variable, creating an output called a 'common spatial trend' that allows quantification of overlap in future predator-prey distributions using a 'business as usual' climate model for 2050. The degree of change within common spatial trends is presented between contrasting seabirds: common guillemot, black-legged kittiwake, northern gannet and two prey species: herring, sandeels. The second is a time series approach using a Bayesian Network Ecosystem Model with a range of species across all trophic levels from 1990-2014. Within both approaches, we have used the same six important bio/physical variables (2 types of primary production, stratification, temperature, vertical and horizontal speed). The outcomes presented include the startlingly amount of change in common spatial trend predicted in just 30 years and the degree to which bottom-up forces are predicted as the drivers for population change. Authors: Beth Scott¹, Neda Trifonova¹, Dinara Sadydova², Alexander Sadykov³, Michela De Dominicis⁴, Sarah Walkin⁴, Judith Wolf⁴ ¹University of Aberdeen, ²University of Queen's Belfast, ³University of Queen's Belfast, ⁴National Oceanographic Centre