Abstract
Seabirds and other land-breeding marine predators are considered to be useful and practical indicators of the state of marine ecosystems because of their dependence on marine prey and the accessibility of their populations at breeding colonies. Historical counts of breeding populations of these higher-order marine predators are one of few data sources available for inferring past change in marine ecosystems. However, historical abundance estimates derived from these population counts may be subject to unrecognised bias and uncertainty because of variable attendance of birds at breeding colonies and variable timing of past population surveys. We retrospectively accounted for detection bias in historical abundance estimates of the colonial, land-breeding Adélie penguin through an analysis of 222 historical abundance estimates from 81 breeding sites in east Antarctica. The published abundance estimates were de-constructed to retrieve the raw count data and then re-constructed by applying contemporary adjustment factors obtained from remotely operating time-lapse cameras. The re-construction process incorporated spatial and temporal variation in phenology and attendance by using data from cameras deployed at multiple sites over multiple years and propagating this uncertainty through to the final revised abundance estimates. Our re-constructed abundance estimates were consistently higher and more uncertain than published estimates. The re-constructed estimates alter the conclusions reached for some sites in east Antarctica in recent assessments of long-term Adélie penguin population change. Our approach is applicable to abundance data for a wide range of colonial, land-breeding marine species including other penguin species, flying seabirds and marine mammals.
Highlights
Ecologists are increasingly turning to historical abundance data to understand past changes in animal abundance and more broadly the ecosystems in which animals occur
There is widespread recognition that counts of animal populations are often subject to detection bias, and that the extent of bias may vary over time and space due to aspects of a species’ biology, the type of counting methods used, and the environmental conditions under which a population survey is undertaken [3,4]
This recognition has led to the development of a general framework for abundance estimation that explicitly accounts for detection bias and its uncertainty [3,5,6], new methods for estimating detection bias, and calls for ecologists to estimate and account for bias and uncertainty when estimating animal abundance [7]
Summary
Ecologists are increasingly turning to historical abundance data to understand past changes in animal abundance and more broadly the ecosystems in which animals occur. There is widespread recognition that counts of animal populations are often subject to detection bias, and that the extent of bias may vary over time and space due to aspects of a species’ biology, the type of counting methods used, and the environmental conditions under which a population survey is undertaken [3,4] This recognition has led to the development of a general framework for abundance estimation that explicitly accounts for detection bias and its uncertainty [3,5,6], new methods for estimating detection bias, and calls for ecologists to estimate and account for bias and uncertainty when estimating animal abundance [7]. Developing approaches to improve interpretation of historical data may allow more reliable assessment of extremely valuable long-term abundance data
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have