Better Ecological Inference — Even When You Don’t Detect Every Animal Analysing Ecological Data with Detection Error
Learn to analyse ecological field data with detection error using R. Work with point counts, ARU data, N-mixture models, distance sampling and time-removal methods. https://prstats.org/course/analysing-ecological-data-with-detection-error-aedd01/ *Analysing Ecological Data with Detection Error (AEED01)* is a live online course designed for ecologists, conservation biologists, and wildlife researchers who collect field data where *imperfect detection*, observation error, or non-detection are part of the reality. The course teaches you how to recognise and properly model detection error, so your inferences about presence, abundance, occupancy, or distribution are robust, defensible, and scientifically sound. What You Will Learn - Why ignoring detection error often leads to *biased or misleading conclusions* — especially important when working with mobile, elusive, or rare species. - How to apply statistical methods and models that account for *imperfect detection*, such as occupancy models, detection-adjusted count or abundance models, and other approaches implemented in R (e.g., using packages inspired by detection-error frameworks). - When and how to use *single-visit or repeat-survey designs, distance sampling, removal methods, or hierarchical models* to properly estimate real population parameters rather than naive indices. - How to incorporate *environmental covariates, observer variation, habitat heterogeneity, and detection heterogeneity* to improve model realism and ecological inference. - Best practices for *designing surveys and data collection* to make detection-error modelling feasible and effective — including guidance on repeated visits, survey protocols, sampling effort, and data structure. Why This Course Matters Many animals — especially marine mammals, large mammals, cryptic species, or rare taxa — are difficult to detect reliably. Observations often miss individuals, or detectability varies with environment, behavior, observer effort, or time. Without accounting for detection error, analyses of presence/absence, occupancy, abundance, distribution, habitat use, or population trends can be seriously flawed. Recent ecological research warns that ignoring complex observation processes can lead to “biased inference and poor predictions.” By learning detection-error-aware methods, you can: - Estimate *true occupancy or density* rather than undercounted indices. - Quantify *uncertainty around presence or abundance*, essential for conservation, impact assessment, and monitoring. - Better understand patterns of habitat use, distribution, and behaviour — even when detection is imperfect. - Make more defensible claims in reports, publications, or management recommendations. Course Format & Who Should Attend - Live-online course (dates & duration published on course website) — delivered in clear, accessible sessions combining theory, case studies, and hands-on coding. - Suitable for researchers, field ecologists, conservation biologists, postgraduate students — anyone working with observational ecological data in R or planning to. - Participants should ideally have some experience with ecological data, survey work or field studies — but the course will guide you through the statistical and practical challenges from first principles. Please email [email protected] with any questions. -- Oliver Hooker PhD. PR stats [[alternative HTML version deleted]] _______________________________________________ R-sig-ecology mailing list [email protected] https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
