The scale of landscape effect on seed dispersal depends on both response variables and landscape predictor

摘要

ContextLandscape structure can affect seed dispersal, but the spatial scale at which such effect is maximized (scale of effect, SoE) is unknown.ObjectivesWe assessed patterns and predictors of SoE on the seed rain in two Mexican regions: the relatively well-preserved Lacandona rainforest, and the more deforested Los Tuxtlas rainforest. We hypothesized that source limitation at Los Tuxtlas makes seed dispersal more reliant on landscape patterns measured across larger spatial scales, especially when considering connectedness-related landscape metrics and dispersal-dependent responses.MethodsWe recorded the abundance and diversity of tree seeds in 20 forest sites per region, separately assessing local (dropping from neighboring trees) and dispersed (immigrant) seeds. We measured forest cover, fragmentation, and matrix openness in 11 concentric landscapes surrounding each site and tested for differences in SoE among regions, landscape metrics, response variables, and seed origins.ResultsContrary to expectations, SoE did not differ between regions and seed origins. Yet, as expected, forest cover tended to have larger SoE than matrix openness, with fragmentation showing intermediate values. Response variables also followed the predicted SoE pattern (abundancediversityspecies richness).ConclusionsForest cover has larger SoE than matrix openness, possibly because forest cover is related to large-scale processes (e.g. long-distance dispersal) and matrix openness may drive small-scale processes (e.g. edge effects). Species richness may have larger SoE because of its dependence on long-distance dispersal. Therefore, to accurately assess the effect of landscape structure on seed dispersal, the optimal scale of analysis depends on predictor and response variables.