What drives seed dispersal effectiveness?

Abstract

Seed dispersal is a critical phase in plant reproduction and forest regeneration. In many systems, the vast majority of woody species rely on seed dispersal by fruit-eating animals. Animals differ in their size, movement patterns, seed handling, gut physiology, and many other factors that affect the number of seeds they disperse, the quality of treatment each individual seed receives, and consequently their relative contribution to plant fitness. The seed dispersal effectiveness framework (SDE) was developed to allow systematic and standardized quantification of these processes, offering a potential for understanding the large-scale dynamics of animal-plant interactions and the ecological and evolutionary consequences of animal behavior for plant reproductive success. Yet, despite its wide acceptance, the SDE framework has primarily been employed descriptively, almost always in the context of local systems. As such, the drivers of variation in SDE across systems and the relationship between its components remain unknown. We systematically searched studies that quantified endozoochorous SDE for multiple animal species dispersing one or more plant species in a given system and offered an integrative examination of the factors driving variation in SDE. Specifically, we addressed three main questions: (a) Is there a tradeoff between high dispersal quality and quantity? (b) Does animal body mass affect SDE or its main components? and (c) What drives more variation in SDE, seed dispersal quality, or quantity? We found that: (a) the relationship between quality and quantity is mediated by body size; (b) this is the result of differential relationships between body mass and the two components, while total SDE is unaffected by body mass; (c)neither quality nor quantity explain more variance in SDE globally. Our results also highlight the need for more standardized data to assess large-scale patterns in SDE.

Keywords: animal–plant interactions; forest regeneration; frugivory; seed dispersal.

FIGURE 1

Distribution of study species and data points used in the final analyses. Numbers outside the brackets represent the number of data points, that is, a single SDE or component value for an animal dispersing a plant or a pooled group of plant species. Numbers in brackets represent the number of plant species (or pooled species) for which SDE or its components were quantified. The figure notes only studies that were used at least once in one of the analyses and excludes studies that were filtered out (see Section 2).

FIGURE 2

The relationship between seed dispersal effectiveness, quality, and quantity is affected by body mass in birds. The curved surface shows model predictions of the bGLM, averaging the random intercept across studies. The model indicates that among large‐bodied birds, quality and quantity are negatively correlated, while in small‐bodied birds, quality and quantity are positively correlated. A second version of the plot, including data points, is available as Figure S1.

FIGURE 3

The relationship between body mass and seed dispersal effectiveness (SDE), quality, and quantity in birds. X axis—body mass, z‐transformed. Y axes—SDE, SDE quality, SDE quantity, all z‐transformed. p Values are from Bayesian mixed effects model (see Section 3). For quality and quantity, they derive from a single model that includes both, thus providing the effect of each independently of the other. Each dot represents a single interaction (a bird species with a plant species or a pool of species). Colors represent plant species (or a pool of species). Note that the almost full convergence of the regression line intercepts, indicating low variance of the random effect, is likely enhanced by the fact that the data for each study was centered by setting the mean to 0 (see Section 2). Note that the choice of axes does not represent typical causality (where x is “independent” and y “dependent”). Axes were selected based on the models where, in order to assess independent relationships, body mass was a response variable. Also note that these are type 1 models (where residuals are parallel to the y axis), which may, in cases where the predictors are measured rather than experimentally controlled, inflate the slope but not the direction or its statistical significance. Regression lines should therefore be seen as an indication of the directionality of the relationship between variables, not an exact estimate of the effect size.

FIGURE 4

The relationship between the correlation between seed dispersal effectiveness (SDE) quality and quantity (x axis) and the relative contribution of each component to total SDE (y axis). Each data point is a single study‐species for which at least three data points were available to calculate the correlation between the components and their relative contribution. X axis: correlation coefficient between SDE quality and quantity (measured as R ² of a linear regression model). Y axis—the ratio of proportions of variance explained by quality and quantity (quality/quantity), log transformed. A vertical line separates studies with positive and negative correlation between SDE quality and quantity. The horizontal line separates studies in which quality explained more or less than quantity (since log(1) = 0). Upper left square: negative correlation, higher contribution of SDE quality. Upper right square: positive correlation, higher contribution of SDE quality. Bottom right square: positive correlation, lower contribution of SDE quality. Bottom left square: negative correlation, lower contribution of SDE quality. The numbers to the right of each data point are the sample size for the corresponding plant‐study species.