Stabilising sparse-data trends in the National Plant Monitoring Scheme: a scalable Bayesian model for plant abundance

The UK National Plant Monitoring Scheme produces around 600 annual species within-habitat trends that combine information on occupancy and positive cover. These trends underlie developing government indicators. We give a full specification of the Bayesian model and priors, and document changes introduced in the 2025 run (covering 2015-2024) relative to the 2024 run (2015-2023) to stabilise trends for sparsely recorded species. The principal change replaces first-order random-walk temporal priors with stationary autoregressive (AR(1)) priors; we also tighten several detection and cover priors. Future work will address non-response bias in NPMS survey returns.