Data fusion and integrated species distribution models for three endangered ferns (Culcita macrocarpa, Diplazium caudatum, and Pteris incompleta) in a Mediterranean biodiversity hotspot.

Abstract

Introduction: Accurately modeling the distribution and abundance of rare and threatened species is considered critical for informing conservation strategies under increasing environmental pressures. Three threatened paleomediterranean relict ferns, Culcita macrocarpa, Diplazium caudatum, and Pteris incompleta, are restricted to climatically stable microhabitats within Los Alcornocales Natural Park (southern Spain), rendering them particularly vulnerable to environmental change. Methods: A joint-likelihood framework was employed within Integrated Species Distribution Models (ISDMs) to estimate spatiotemporal abundance of the three fern species. Structured abundance data (2014–2023) from the Andalusian Fern Recovery Plan were integrated with opportunistic presence-only records obtained from Global Biodiversity Information Facility (GBIF). Twenty-two model configurations were tested to evaluate the benefits of multi-species modeling and data-fusion strategies. Results: Predictive performance was improved by multi-species modeling, with shared ecological and spatial structures being captured more effectively. Spatiotemporal random effects were found to be more influential than fixed effects, reflecting local-scale heterogeneity in fern distributions. Spatiotemporal patterns were captured most effectively by the model excluding GBIFdata fusion. Signs of overfitting were observed in the model incorporating data fusion, with GBIF inclusion failing to consistently improve predictive performance due to limited observations and spatial biases. Population trends were indicated to be generally stable, with localized increases and limited declines documented in two C. macrocarpa populations.Discussion: The value of ISDMs in leveraging complementary data sources is demonstrated by these findings, providing an effective framework for conservation planning in data-limited systems facing environmental change.