Integrating local pastoral knowledge, participatory mapping, and species distribution modeling for risk assessment of invasive rubber vine (Cryptostegia grandiflora) in Ethiopia's Afar region
The threats posed by invasive plants span ecosystems and economies worldwide. Local knowledge of biological invasions has proven beneficial for invasive species science and management, but to date no work has explored the importance of integrating this knowledge with species distribution modeling for invasion risk assessments. In this study, we integrated local pastoral knowledge with Maxent modeling to assess the suitable habitat and potential impacts of invasive Cryptostegia grandifloraRobx. Ex R.Br. (rubber vine) in the Afar region of Ethiopia. We conducted in-depth focus groups across seven villages within the Amibara and Awash-Fentale districts. Rubber vine occurrence points were collected in the field with focus group participants and processed in Maxent with moderate resolution (250m) MODIS-derived vegetation indices, topographic data and anthropogenic variables. We tested initial model fit using a jackknife procedure and validated the final model with an independent occurrence data set collected through participatory mapping activities with the same pastoralists. A multivariate environmental similarity surface analysis revealed areas with novel environmental conditions for future targeted surveys.
Model performance was evaluated using area under the receiver-operating characteristic curve (AUC) and showed good model fit across the 18 jackknife models (average AUC = 0.797) and the final model (test AUC = O.959). Our results reveal the growing threat rubber vine poses to Afar, with suitable habitat extending downstream of its current known location in the middle Awash River basin. Furthermore, local pastoral knowledge provided important context for its rapid expansion due to acute changes in seasonality and extensive habitat alteration, in addition to immediate threats posed to a number of native tree species that provide critical ecosystem services to local communities. This work demonstrates the important benefits of integrating local ecological knowledge with species distribution modeling for early detection and targeted surveying of recently established invasive species.