Effects of canopy heterogeneity, seed abscission and inertia on wind-driven dispersal kernels of tree seeds

1. Understanding seed dispersal by wind and, in particular, long-distance dispersal (LDD) is needed for management of plant populations and communities, especially in response to changes in climate, land use and natural habitats. Numerical models designed to explore complex, nonlinear atmospheric processes are essential tools for understanding the fundamental mechanisms involved in seed dispersal. Yet, thus far, nearly all such models have not explicitly accounted for the spatial heterogeneity that is a typical feature of all ecosystems. 2. The recently developed Regional Atmospheric Modelling System (RAMS)-based Forest Large Eddy Simulation (RAFLES) is used here to explore how within-stand canopy heterogeneity impacts LDD. RAFLES resolves microscale canopy heterogeneity such as small gaps and variable tree heights, and it simulates their impacts on turbulence inside and above the canopy in the atmospheric boundary layer (ABL). For that purpose, an Eulerian-Lagrangian module of seed dispersal is added to RAFLES to simulate seed trajectories. 3. Particular attention is paid to the sensitivity of statistical attributes of the dispersal kernels (i.e. mean, mode, variance, tail) to key simplifications common to all seed dispersal models, such as horizontal homogeneity in the canopy and flow field, and the tight coupling between air parcel trajectories and seed trajectories (i.e. neglecting seed inertia). These attributes appear to be sensitive to various factors operating at scales ranging from the seed scale to the ABL scale. 4. Simulations with RAFLES show that LDD is characterized by a dispersal kernel with a 'tail', asymptotically approaching a power law decay of -3/2 (mainly occurring for lighter seeds at high wind speeds). This is consistent with asymptotic predictions from analytical models. The wind speed threshold at which seed abscission occurs, set-up to be twice the standard deviation of the vertical wind speed, is shown to affect short-distance dispersal, but has no significant impact on LDD. Ignoring the effects of seed inertia on the seed trajectory calculations has a minor effect on short-distance dispersal and no effect on the probability of seed uplift. Thus, it has no significant impact on LDD. 5. Synthesis. Tree-scale canopy heterogeneity affects the turbulence characteristics inside and above the canopy and, consequently, this affects dispersal kernel statistics. A key finding from this study is that ejection is enhanced above the shorter trees of the canopy. Seeds dispersed above shorter trees have a higher probability of experiencing LDD while their short-distance dispersal remains practically the same. At inter-annual time scales, such interactions could affect species composition.