Survey frequency affects the estimates and conclusions of long-term changes in waterbird populations: a case study from a South African wetland

Abstract

Accurately detecting the magnitude and direction of population trends provides essential information for conservation management. There are numerous local and international long-term waterbird monitoring programmes with substantial variation in how frequently surveys are conducted. We aimed to assess how temporal frequency of waterbird surveys influence three aspects of population trends: (1) level of uncertainty around estimates of long-term trends; (2) frequency, duration and overlap of significant short-term increases or declines in trend trajectory; and (3) differences between estimated change in population size measured over four different time periods. Data from 28 years of surveys at a waste water treatment facility in South Africa were used to model the population trend of 41 waterbird species. We used generalised additive mixed-models to compare significance and magnitude of population change trends between monthly and biannual surveys. Direction and shape of species trends were similar between protocols although uncertainty around trends was generally lower when monthly data were used. The number of significant trend trajectories, the majority of which were declines, was higher in monthly models. Temporal overlap of periods of significant change of trends of the two protocols averaged 30%. However, both sampling protocols offered similar conclusions about long-term change in population size of species. General patterns differed conspicuously among species, although groups of species showed similar trajectories of population change. We conclude that biannual surveys are appropriate if the goal is to assess general patterns of population change. However, if management action relies on understanding the significance of population trajectories, surveys at a finer temporal scale are required, especially for species showing a pattern of unpredictable movement. Trade-offs between logistical constraints and statistical rigour as well as species identity need to be carefully considered for the design of long-term monitoring programmes of waterbirds.