Rain, Bark, and Residual Variability in Stemflow From Three Dominant Tree Species of a Southern Great Lakes Forest

ABSTRACT Stemflow redistributes rainfall entrained on canopy surfaces, creating spatially concentrated chemical and biological fluxes, yet the relative roles of its principal meteorological drivers remain uncertain. We monitored stemflow for 15 mature trees (5 each of Fagus grandifolia , Acer saccharum , Liriodendron tulipifera ) across 40 storms (April–November 2024) in a closed‐canopy forest (Holden Arboretum, Ohio, USA) and normalised responses by LiDAR‐derived stemflow drainage areas obtained with a pruning algorithm (CanoPyHydro). A linear mixed‐effects model tested rain amount, species (as a proxy for bark roughness within the context of branch architecture derived from CanoPyHydro), and their interaction; the relationships between residual structure (derived from separate OLS regressions) and other meteorological drivers were investigated using PCA including eight pre−/in‐storm meteorological descriptors. Rainfall amount dominated variability and interacted with species: rain ( p < 0.001), species ( p = 0.045), and rain × species ( p < 0.001) were significant. Species‐wise OLS fits showed high explanatory power of rain alone ( R 2 : beech = 0.84, maple = 0.79, poplar = 0.75) and a steeper rain‐stemflow slope for beech (0.072 mm stemflow per mm rain) than for maple or poplar (0.03 mm mm −1 ). PCA summarised non‐rain structure along a temperature/radiation axis and a pressure‐change axis; only beech exhibited a weak residual association with these axes (adj. R 2 = 0.137, p = 0.025). Occult (non‐rain) inputs were detected for all species and were largest for beech. Results indicate that, in closed‐canopy temperate forests, event‐scale stemflow is primarily set by rain amount and bark class, with fine‐scale architectural effects muted; modest, species‐specific non‐rain influences likely act through bark storage and near‐bark vapour processes. Incorporating explicit bark storage and non‐rain inputs into interception models should improve predictions of stemflow‐mediated water (and solute) delivery to near‐stem soils.