UC Davis

Riparian vegetation planting and management are vital to river engineering projects. To inform these activities, a better understanding of what influences riparian vegetation recruitment and identifying where vegetation will most likely establish and survive is needed. This study investigated whether the recruitment of Populus fremontii (Fremont cottonwoods), a dominant riparian species in the western USA, could be predicted at 0.91-m2 resolution deterministically and statistically throughout a dynamic, alluvial regulated river. The testbed was a ~ 34-km stretch of the Yuba River in California, USA, mapped in 2017 after a large flood reset the terrain. Five years later from August through November 2022, a field campaign characterized juvenile cottonwoods recruitment. For the deterministic test, a riparian seedling recruitment model was used with expert-estimated parameters. Bioverification analysis, or comparison of model predictions to observed organism locations, found that the model did not accurately differentiate the predicted optimal locations from the lethal locations. Hydrophysical and topographic variables were then used as predictor rasters in a data-driven, supervised classification Random Forest (RF) model with cottonwood presence and absence data to statistically test whether recruitment locations could be predicted. The RF model performed well, having an overall accuracy of 87% and reached an AUC-ROC value of 94% with only a few predictors. When the statistical model was coupled with biophysical interpretations of model behavior, topographic variables were much more significant drivers for recruitment locations than hydrophysical variables. Further mechanistic developments to understand underlying governing equations and parameters are feasible drawing on the lessons from bioverification and the RF model. Both deterministic and statistical methods are recommended to advise stakeholders about suitable locations for riparian vegetation restoration or topographic characteristics needed to promote restoration efforts, as each yield unique insights.