Statistical models for the persistence of threatened birds using citizen science data: A systematic review

摘要

Background Due to risk factors such as climate change, habitat destruction, overhunting and pollution, bird extinctions are now occurring at a rate that far exceeds their speciation rate. There are no robust indicators of biodiversity conservation that can be used to complement existing national indicators of economic and social health. The statistical methods which are used to model and evaluate the persistence or extinction risk of threatened bird species using citizen science data are reviewed in this study. Citizen science data helps to increase the number of records, thereby improving our understanding of the dynamics in declining bird species populations. Methods Adhering to the PRISMA guidelines a comprehensive systematic review was performed using three databases: ProQuest Central , Scopus and Web of Science from January 1900 to January 2019. Only journal articles which analysed the persistence or extinction risk of threatened bird species using a statistical model, predictive model or a trend analysis, developed using citizen science data were included in this study. Bird species in near threatened or least concern categories that are declining in population/range were also included, since these may be the next wave of species to be added to the endangered species lists. Results Most of the 39 unique studies describing statistical models for this purpose used generalized linear models, followed by hierarchical/linear mixed models, machine learning models and persistence probability models respectively. A quality assessment tool was created in order to evaluate these articles. The review suggested several methods for measuring the persistence of threatened bird species, but there was no attempt to identify critical tipping points using methods such as change-point analysis. Conclusion The findings suggest that the persistence of threatened bird species varies depending on various risk factors which need to be addressed in order to produce better outcomes for the conservation of threatened birds. This review reveals the most suitable statistical methods for this purpose.