Utility of occupancy and N-mixture models for modelling biodiversity dynamics with ecological monitoring programs in Norway

Sandercock, B.K., Chipperfield, J.D, Cretois, B., Fossøy, F., Kleiven, E.F., Layton-Matthews, K., Nater, C.R., Töpper, J.P., Yoccoz, N.G. & Ellingsen, K.E. 2025. Utility of occupancy and N-mixture models for modelling biodiversity dynamics with ecological monitoring programs in Norway. NINA Report 2515. Norwegian Institute for Nature Research. Ecological monitoring of natural resources is increasing important in a rapidly changing world where biodiversity losses are driven by negative impacts of habitat degradation, climate change, overharvest, and pollution. Recent years have witnessed remarkable advances in new technologies to assist with monitoring changes in the distributions and abundance of plant and animal populations. New sampling techniques include environmental DNA, passive sampling with acoustic recorders and camera traps, and improved options for integration of ecological data from structured field surveys and citizen science programs. Parallel advances in quantitative ecology have included new statistical models for addressing issues of imperfect detection, misidentification, and environmental variability that are ubiquitous in field sampling. Workflows based on artificial intelligence facilitate species identification and processing large volumes of ecological data. The combination of new sampling methods and advanced quantitative tools has provided innovations for optimizing ecological monitoring programs that yield estimates of demographic parameters and population trends with greater accuracy and precision. New methods and statistics provide substantial cost-savings with more efficient sampling designs, but also more robust inferences about population and community dynamics which assist implementation of conservation and management programs. The goal of our report was to review recent advances in statistical models and software tools for occupancy models for occurrence data and N-mixture models for count data. The models are useful for estimating distributions and abundance of uncommon or rare species, and especially in situations where individuals are difficult to detect in the field because the life-stages are cryptic or behavior is secretive. We reviewed 12 different groups of statistical models for ecological monitoring programs: naïve occurrence, static, dynamic, single visit, multi-state, N-mixture, multiscale, species interactions, community, spatial, misidentification, and integrated models. Selecting a suitable model is often a trade-off between costs of increased sampling effort to meet greater demands for data collection versus benefits of reducing bias in the estimated or derived parameters. Naïve occurrence, static and single visit models require less sampling effort but are more likely to yield biased estimates of occupancy or abundance. Dynamic models relax the closure assumption and estimate site colonization and extinction. Multi-state and N-mixture models are based on state or count data which can contain more information on population status. Multi-scale models have nested levels which are useful for multiple methods or eDNA sampling. Interaction and community models estimate species interactions and biodiversity metrics for communities. Spatial and misidentification models relax the assumptions of spatial independence and certainty of species identity to account for correlations and possible errors. Integrated models combine data sources such as systematic data from structured monitoring and opportunistic data from citizen science programs. The statistical models provide a valuable foundation for producing biodiversity maps for different ecosystems. Biodiversity maps are urgently needed to reduce negative effects of habitat loss from land use conversion and to improve plans for habitat restoration and protected area networks in Norway. Our report provides a summary review of new resources available for ecological monitoring. We provide detailed descriptions of the different statistical models and illustrate applications with selected case studies from research scientists at the Norwegian Institute for Nature Research (NINA). We provide recommendations for ecological monitoring that are applicable to current and potential applications of the statistical models to ecological monitoring programs in Norway. We have also compiled an overview of the new software tools and learning resources that can be used to fit the different models to ecological data for improved inferences. Uptake of new methods and statistical models by scientists working with ecological monitoring in Norway and abroad will be essential for finding new solutions to address the crisis of biodiversity loss.