SCIENCE@home

Transferring species distribution models to understudied systems: Does data integration help?

Charlotte R. Patterson,Xiaotian Zheng,Scott D. Foster,Kate J. Helmstedt

Published 2025 in Methods in Ecology and Evolution

ABSTRACT

The rapid expansion of biological data collection is uneven globally and many remote sites remain understudied. Species distribution models fitted in data‐rich survey areas can be transferred to predict distributions in understudied areas. However, transfers are challenging and often fail due to factors such as differences in conditions between training and projection sites. Integrated species distribution models that harness benefits from multiple data types can improve distribution predictions relative to single‐dataset approaches, but whether data integration improves spatial transferability remains unclear. Here, we evaluate the performance of integrated (PA‐PO), presence‐only (PO) and presence–absence (PA) models in simulated spatial model transfer scenarios that emulate two reasons that transfers fail: environmental dissimilarity of training and projection sites and a hidden spatial process that induces spatial autocorrelation of the species distribution. We found that PA‐PO integrated models outperformed PO and PA models on average during spatial transfer, but the magnitude of the improvement varied among transfer scenarios. Integrated models were better able to predict species distributions at new, environmentally dissimilar, sites than single‐dataset approaches due to lower requirements for extrapolation into novel environments and reduced effects from sampling bias. Spatial autocorrelation from a hidden spatial process was accounted for well by integrated models, leading to modest but notable gains in predictive performance during transfer. Our findings highlight the potential for diverse data sources to be integrated to improve predictions of species distributions in understudied regions. We outline how factors like data quality and transfer conditions can indicate situations when integration could be beneficial for spatial transfer.

PUBLICATION RECORD

CITATION MAP

EXTRACTION MAP

CLAIMS

  • No claims are published for this paper.

CONCEPTS

  • No concepts are published for this paper.

REFERENCES

Showing 1-47 of 47 references · Page 1 of 1

CITED BY

  • No citing papers are available for this paper.

Showing 0-0 of 0 citing papers · Page 1 of 1