Simulation and statistical modelling approaches to investigate hydrologic regime transformations following Eastern hemlock decline

Ecohydrological processes occurring at or near the Earth's surface are strongly influenced by Eastern hemlock [EH; Tsuga canadensis (L.) Carriere], a foundation tree species of eastern North American forests. EH populations are currently threatened by the invasive hemlock woolly adelgid (HWA; Adelges tsugae Annand). HWA HWA populations have been expanding rapidly throughout the EH's range. Catchment-scale research examining the hydrological consequences of HWA infestation is lacking, and plot-scale studies remain conflicted in their findings. Given the complex relationships between canopy interception, unsaturated and saturated groundwater storage, and root water uptake, it is not immediately clear how EH loss will affect the hydrologic cycle. We investigated the impact of EH mortality on stream discharge characteristics across a regional sample of catchments utilizing both simulation and statistical modelling approaches. We first examined the relationship between various catchment characteristics, including EH health, and three hydrological variables through regression analysis. We then employed a non-parametric statistical test to evaluate differences in hydrologic regime trends between non-infested and infested catchments. Finally, we calibrated a physically based hydrologic model and considered differences in optimal model parameter values and simulated overland runoff between non-infested and infested catchments. HWA presence modified several ecohydrological characteristics and precipitation partitioning between groundwater flows and surface runoff, potentially driving higher stream flashiness and overland flow, lower baseflow contributions and catchment storage, shorter flow-path lengths, and variable source area dilation at infested sites. Our results suggest that EH decline will augment flooding potential associated with the increasing frequency and intensity of Atlantic Basin tropical cyclone events. Further, our physically based simulation provides more determinate results than regression analysis, indicating that a purely statistical methodology, commonly utilized in studying the relationship between landcover characteristics and hydrologic regime, neglects dynamic physical ecohydrologic relationships.