Estimation of plant species richness exploiting probabilistic sampling and purposive lists: Empirical evidence and practical proposal for forest inventories

Forest surveys, especially national forest inventories, are evolving towards multipurpose resource strategies, expanding their scope in several directions including biodiversity assessment. This article focuses on the estimation of plant species richness that constitutes one of the most relevant biodiversity indicators in forest ecosystems. In forest inventories, surveys are usually performed by locating plots in the forest area of interest according to a probabilistic scheme, so that the estimation of plant species richness can be approached from a probabilistic perspective by (i) recording a matrix of species presence/absence called incidence data, and then (ii) adopting a class of widely used, automated estimators called nonparametric estimators. The purpose of this article is to raise awareness within the forestry community of the recent findings of Di Biase et al. (2025) who show, both theoretically and through simulations, the inadequacy of nonparametric estimators affected by massive negative bias due to the difficulty of sampling rare species, at the same time highlighting the appeal of a data integration approach that consists in exploiting lists of rare species compiled by purposive surveys to improve the sample-based estimates. A case study performed in the Nature Reserve of Poggio all'Olmo (Central Italy) is considered to confirm the failure of nonparametric estimators and the suitability of the data integration in estimating plant species richness. This approach paves the way for the practical integration of species richness assessment into forest inventories, thus meeting the technical requests to support modern multipurpose forestry.