bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: May 6, 2025

Abstract Bayesian hierarchical models are ubiquitous in ecology. Random effect model structures often employed that treat individual effects as deviations from larger population-level effects. In this way individuals assumed to be ‘exchangeable’ samples. Ecologists may address exchangeability assumption intuitively, but might certain modeling contexts ignore it altogether, including situations where have large implications for study design. Multispecies occupancy based on detection/non-detection data an approach can utilized by those tasked with monitoring rare and endangered species because most literature suggests that, compared single models, improved parameter estimates assured. Yet, we illustrate through a power analysis how sampling requirements detect experimental treatment vary tremendously depending whether the is met. The degree which community respond similarly covariates governs ability accurately estimate parameters using multispecies models. Detecting small or moderate changes resulting habitat restoration treatments impossible datasets (e.g., < 36 locations, each surveyed 8 times) even paired treatment-control design if violated. By contrast, when met, confidently estimated few 12 locations (6 pairs) 6-8 survey events. Often, know statistical needed species-specific depends unknown values of responses diverge. When violated, at lower levels effort, provide worse inference than

Language: Английский