Determinants of Geographic Patterns in Seed Dispersal Modes in Continental U.S. Plant Communities

Abstract

Seed dispersal is a critical ecological process that influences plant community structure, promotes regeneration, and determines species’ abilities to track shifting climatic conditions. As climate change accelerates, understanding what drives variation in dispersal strategies is increasingly important for predicting biodiversity responses and informing conservation strategies. While broad patterns of dispersal modes—such as the dominance of animal-dispersed (zoochorous) plants in tropical forests and wind-dispersed (anemochorous) plants in higher latitudes—are well documented, these patterns have been less studied in temperate regions. Moreover, little is known about how dispersal modes covary with other plant functional traits and how their distribution is shaped by both abiotic conditions and biotic dispersal agents like animals. This study addresses key knowledge gaps by evaluating the biogeographic and environmental structuring of plant dispersal modes across temperate and boreal forests in the continental United States. Specifically, it investigates:

Whether dispersal mode distributions vary predictably with climate, topography, and disperser availability; Whether these patterns support the resource-availability hypothesis (which links dispersal strategies to environmental resource gradients) or the disperser-availability hypothesis (which links them to the abundance of dispersal agents); and How dispersal modes are integrated with broader suites of plant functional traits. To answer these questions, I applied a Predictive Trait Model (PTM) using the Generalized Joint Attribute Modeling (GJAM) framework. This Bayesian multivariate model jointly estimates how dispersal modes respond to environmental and biotic variables, including annual precipitation, temperature, wind speed, topographic position index (TPI), and a novel metric of disperser availability. Disperser demand was estimated by integrating species-level diet fractions, body mass, and density data for frugivorous birds and granivorous mammals using distance sampling from NEON data. Trait and plot data were obtained from a continental-scale forest inventory database (Qiu et al., 2022). Visualizations included distribution maps, climate space plots, model coefficient plots, and a dendrogram of trait responses.

Key Findings:

Geographic Trends: Zoochory is more common in warm, humid, low-latitude forests, while anemochory dominates in cooler, drier, and more open habitats in higher latitudes and elevations. This spatial partitioning aligns with expectations from tropical ecosystems. Climatic Associations: In environmental space, zoochory is concentrated in areas of low water deficit and high temperature, whereas anemochory prevails in cooler and more water-limited conditions. These patterns suggest that environmental constraints shape dispersal mode prevalence across regions. PTM Results: Despite directional associations, none of the environmental predictors had statistically significant effects. For zoochory, wind speed showed a negative association, while for anemochory, temperature and wind speed had weakly positive effects. Disperser demand showed little influence, possibly due to limitations in the input dataset. Trait-Derived Clusters: A dendrogram of predictor effects showed that dispersal modes cluster with relevant traits—zoochory with fleshy fruits and broad leaves; anemochory with winged fruits and needle-like foliage—highlighting their integration into broader ecological syndromes. Species-Level Patterns: While plant height and seed mass were poor predictors of wind response, taxonomic patterns emerged. For example, Acer species consistently showed positive wind speed effects (some significant), reflecting adaptations for wind dispersal via samaras. Conversely, some conifers (e.g., Pinus sabiniana) had unexpectedly negative or variable responses.

Together, these findings highlight that while plant dispersal modes are clearly structured across geographic and climatic gradients, their underlying drivers are multifaceted and context-dependent. The co-occurrence of dispersal strategies with specific leaf, fruit, and seed traits suggests that dispersal is part of broader ecological syndromes shaped by environmental filtering. However, the limited predictive power of individual environmental variables underscores the challenges of modeling dispersal using coarse-scale data. This study also reveals important methodological limitations, including simplifications in the disperser demand metric, assumptions inherent in distance sampling, and spatial mismatches between plot-level and animal survey data. Moreover, the exclusion of multi-vector dispersal and long-distance dispersal events may overlook critical processes influencing range shifts. Future research should focus on improving the temporal and spatial resolution of both wind and animal data, incorporating more detailed dispersal traits (e.g., seed release height, aerodynamic morphology), and expanding taxonomic and functional coverage of dispersers. By integrating these improvements, future models can move toward more mechanistic, trait-based predictions of plant movement and community assembly under global change.