Developing and assessing methods to census and monitor burrow-nesting seabirds in Ireland

• Main Author: Arneill, Gavin E.
• Other Authors: Quinn, John, Jessopp, Mark John
• Format: Doctoral or Postdoctoral Thesis
• Language: English
• Published: University College Cork 2018
• Subjects: Atlantic puffin; Manx shearwater; European storm petrel; Seabird; Zoology; Census methods; Monitoring methods; fratercula; Fratercula arctica
• Online Access: http://hdl.handle.net/10468/7358

Censusing and monitoring populations are key priorities in conservation. This is particularly challenging for seabirds, where several life history characteristics and the remote nature of breeding colonies of many species make them difficult to study. Burrow-nesting species are the most difficult of all seabird groups to census due to their cryptic breeding habits, nocturnal behaviours within breeding colonies, and coexistence with other burrowing species. Historically estimates of population size in these species were obtained subjectively from the activity within colonies on a given day/night, though the relatively recent development of methodologies such as tape-playbacks have made it possible to generate population estimates using quantitative data. Nevertheless, gaps remain in our knowledge, such as the appropriate sampling approaches to take, the efficacy of some recently established automated methods, and the use of predictive species distribution modelling that could guide these time consuming efforts. In my thesis, we address some of these issues for three key burrow-nesting species in the northern hemisphere: the Manx shearwater (Puffinus puffinus), the European storm petrel (Hydrobates pelagicus) and the Atlantic puffin (Fratercula arctica). In the first paper, we explore a range of sampling approaches to estimate and detect changes in population size, using data from Manx shearwater censuses as a case study. This demonstrated that a priori knowledge of the density and distribution in a colony allows multi-stage stratification that dramatically improves the accuracy of population estimates at low levels of sampling. Power analyses found that many existing monitoring efforts are likely to fail to detect population trends due to the enormous effect of high variation of densities between randomly selected plots. However, subjectively sampling within areas of highest density significantly increases the power to detect declines. My thesis also shows that these breeding distributions can be predicted a ...