HIERARCHICAL ECO-ZONE DELINEATION IN DATA-SCARCE AFROTROPICAL RIVER LANDSCAPES USING OPEN DATA MODELLING FRAMEWORKS

Understanding spatial ecological heterogeneity is critical for modelling and managing aquatic ecosystems, yet most Afrotropical environmental classifications remain limited to drainage divides or coarse expert-based schemes. These approaches overlook key spatial processes that control aquatic conditions across landscapes, especially in data-scarce regions Here, we integrated global remote sensing datasets, open environmental data, and multiple machine-learning algorithms within an open-source workflow to delineate hierarchical eco-zones for Uganda’s lake, riverine, and wetland landscapes. Climate variables, topography, and hydrologically relevant terrain metrics were used to delineate template ecological regions, which were validated by examining patterns of potential natural vegetation (PNV) through zonal statistics to assess ecological heterogeneity across scales. The workflow produced four nested eco-regional maps, with five major ecoregions at the broadest level and 3 finer bioregional subdivisions. 10-fold cross-validation using PNV distributions demonstrated robust predictive performance despite regional data limitations. This framework demonstrates how scalable, cloud-ready modelling tools and open data can provide spatially coherent ecological stratification for hydrological modelling, bioassessment, water quality prediction, and biodiversity monitoring. The approach is transferable to other data-scarce tropical countries, enhancing the ability to design region-specific ecological assessments, identify vulnerability hot

Presentation Preference: Standard Oral (12 Minutes)

Primary Presenter: Amon Aine, University of Natural resources and life Sciences (aine3amon@gmail.com)

Authors:

Wolfram Graf, University of Natural Resources and Life Science, Vienna  (wolfram.graf@boku.ac.at)

Gabriele Stecher, University of Natural Resources and Life Science, Vienna  (gabriel.stecher@students.boku.ac.at)

Theresa Scharl, University of Natural Resources and Life Science, Vienna  (theresa.scharl@boku.ac.at)

Grace Ssanyu, Kyambogo University  (assanyu@kyu.ac.ug)

Mathew Mathew, Herrnegger  (mathew.herrnegger@boku.ac.at)