Identifying important at-sea areas for seabirds using species distribution models and hotspot mapping

Ian R Cleasby,Ellie Owen,Linda Wilson,Ewan D Wakefield,Peadar O'Connell,Mark Bolton

doi:10.1016/j.biocon.2019.108375

Abstract

Marine protected areas (MPAs) underpin the sustainable management of marine ecosystems but require accurate knowledge of species distributions. Recently, advances in tracking technology and habitat modelling have enabled the production of large-scale species distribution models (SDM), which provide the basis for hotspot mapping. In the UK, hotspot mapping to inform seabird MPA identification has involved converting observed or predicted distributions to polygons using either Maximum Curvature or Getis-Ord (Gi*) analysis. Here, we apply both mapping techniques to UK-wide, breeding season SDM predictions for four seabird species (Black-legged Kittiwakes Rissa tridactyla, Common Guillemots Uria aalge, Razorbills Alca torda and European Shags Phalacrocorax aristotelis) in order to compare their performance and inform seabird MPA. When using Maximum Curvature, grid cells within the identified maximum curvature boundaries were defined as hotspots. For Getis-Ord analysis, we defined hotspots as either (1) grid cells containing the top 1% or (2) the top 5% Gi* scores or (3) cells in which Gi* scores were statistically significant. Hotspots based upon Maximum Curvature or statistically significant Gi* scores covered the greatest area and were generally larger than current marine Special Protection Areas. Hotspots based on the top 1% or top 5% of Gi* scores were smaller and were concentrated around the largest breeding colonies. All hotspot methods consistently identified several high-density areas that should be prioritised for seabird conservation. Ultimately, the choice of hotspot identification method should be informed by considering species ecology alongside conservation goals to ensure hotspots are of sufficient size to protect target populations.

Highlights

Accurate species distribution estimates are key to effective wildlife management and conservation, enabling researchers to identify and prioritize geographical areas for environmental protection
In total the sample sizes used in the species distribution models of Wakefield et al (2017) were: Black-legged Kittiwake – 464 birds tracked from 20 sites, median tracking duration per individual = 42 h; Common Guillemot – 178 birds tracked from 12 sites, median tracking duration per individual = 54 h; Razorbill – 281 birds tracked from 14 sites, median tracking duration per individual = 70 h; European Shags – 230 birds tracked from 13 sites, median tracking duration per individual = 75 h
The larger areas identified by statistically significant Gi* or Maximum Curvature hotspots covered these regions as well, and encompassed almost the entirety of Scottish inshore waters

Summary

Introduction

Accurate species distribution estimates are key to effective wildlife management and conservation, enabling researchers to identify and prioritize geographical areas for environmental protection. The identification and protection of areas for conservation in the marine environment has generally lagged behind that in the terrestrial environment (Perrow et al, 2015) This is despite wide-spread recognition that effective conservation of seabirds and other higher marine predators requires protecting important at-sea areas (Game et al, 2009). To address this shortfall, there is a pressing need to develop effective means of mapping the at-sea distributions of seabirds and identify priority areas for conservation (Lascelles et al, 2012)

Objectives

Methods

Results

Conclusion