Obtaining high-quality species interaction data is essential for understanding broad ecological processes. However, collecting these data at large spatiotemporal scales is often an expensive and labour-intensive process. Long-term accumulated observations by multiple people or devices across different localities — referred to as regional ecological knowledge (REK) — are a valuable source of indirect information on species interactions at a regional scale. As such, REK represents a cost-effective approach for estimating regional ecological networks without direct sampling. Writing in Proceedings of the Royal Society B, Zhu and colleagues examined how closely interaction networks inferred from REK correspond with those derived from field-collected camera trap data. As a high-quality baseline, the authors first derived empirical ecological networks based on 17,572 observed interactions between fruit-producing plants and frugivorous birds who disperse their seeds. These data were recorded using camera traps on 22 islands in the Thousand Island Lake region in China. To infer REK-based interaction networks, they defined ‘regionally observed interactions’ as assumed interactions between co-occurring plant and bird species that had been observed to interact on other islands, but not on the focal island. The authors compared the empirical and REK-inferred networks for each island and found that most network metrics were biased when using a REK-based approach, especially owing to local differences in species richness and network size. Nevertheless, REK-based and empirical networks showed similar responses to changes in island area and degree of habitat fragmentation. The authors suggest that actual patterns of species interaction can be inferred using relatively coarse interaction data such as REK-based networks, but that these inferences should be made with caution.
Original references: Proc. R. Soc. B 290, 20231221 (2023)
This is a preview of subscription content, access via your institution