Seed dispersal is a pivotal process in seed plant life cycle, owing to its effects on seed germination, seedling survival, population recruitment, and diversity maintenance in the entire community (Howe and Smallwood, 1982; Rogers et al., 2021). There are diverse dispersal modes, such as anemochory (wind-driven dispersal), hydrochory (water-mediated dispersal), autochory (self-dispersal), and zoochory, which relies on a diverse array of animals for seed dispersal (Howe and Smallwood, 1982). It is widely known that these varying dispersal modes impose selective pressures on many seed and fruit traits, especially the seed size, a key trait which is associated with multiple stages of the life cycle of plants, such as dispersal, germination, and establishment, particularly during early development (Leishman et al., 2000).

Most current studies have focused on the average value of seed size at the species level (Galetti et al., 2013; Chen and Moles, 2015). A recent study indicated that the selection pressure of dispersal mode affected not only average value but also intraspecific variation, that is, the coefficient of variation (CV) values of seed size. The results showed that biotic-dispersed species had a larger CV than abiotic-dispersed species, and species dispersed by birds had a larger CV than those dispersed by mammals (Zhang and Wang, 2024). However, the seed size distribution of many plant species is not normally distributed (Fig. 1A), that is, many species of seeds have a right-skewed distribution, with more smaller seeds and fewer larger ones (Willis and Hulme, 2004; Wang and Ives, 2017). Generally speaking, right-skewed seed size can be partially explained by the idea that a given trait value has a lower bound and the upper bound is limitless (Kattge et al., 2011). However, the intensity of right-skewness differs among species, and some species show an approximately normal distribution or even a left-skewed distribution (Hendrix and Sun, 1989; Shimada et al., 2015). Then our question is: Given that dispersal mode has a selective effect on intraspecific variation of seed size, that is, the coefficient of variation (CV), does it also have a selective effect on skewness?

Plant fruits are often categorised into two groups: fleshy and dry fruits. Fleshy fruits, such as drupes and berries, primarily rely on frugivores for seed dispersal because they offer edible pulps to attract animals to consume and transport seeds (i.e., endozoochory). In most cases, each fleshy-fruited species has a diverse group of frugivory seed dispersers, ranging from small birds to large mammals (Rogers et al., 2021; Ong et al., 2022). Small seeds can be dispersed by all frugivores, whereas large seeds can only be swallowed and dispersed by large frugivores (Galetti et al., 2013; Chen and Moles, 2015). Furthermore, compared with small frugivores, large frugivores usually disperse seeds further away and provide a better seed dispersal service (Wotton and Kelly, 2012; Ong et al., 2022). Typically, small-bodied frugivores exhibit greater species diversity and abundance than large frugivores (Carreira et al., 2020), indicating that plants may encounter small frugivores more frequently than large frugivores. To ensure that a sufficient number of seeds are dispersed in this situation, plants may produce a large number of small seeds for small frugivores; however, they may engage in bet-hedging by producing a small number of larger seeds to wager on a longer dispersal distance and a better dispersal service by large frugivores. Therefore, fleshy-fruited species may exhibit a significant right-skewed seed size distribution.

Dry-fruited species show diverse dispersal modes, depending on both biotic and abiotic mechanisms. Synzoochorous species primarily rely on scatter-hoarding behaviour carried out by many rodent and several bird species for seed dispersal, and these animals generally have body sizes that are not particularly large (Howe and Smallwood, 1982; Muñoz and Bonal, 2008). Scatter hoarders often prefer to remove and cache larger seeds (Wang and Ives, 2017), but usually within a threshold, because the handling ability is positively related to body mass, and scatter hoarders cannot handle very large seeds (Muñoz and Bonal, 2008). Therefore, synzoochorous species may not show a less right-skewed seed size distribution. The same reason may also apply to myrmecochorous species that depend on small ants for seed dispersal. Epizoochorous species may also show a less right-skewed seed size distribution because seeds that are too large may not adhere well to animal fur (Hovstad et al., 2009). For species dispersed by wind or self-explosion, a short dispersal distance for large seeds (Greene and Johnson, 1993; Narbona et al., 2005) may also lead to a less right-skewed seed size distribution. Hence, most of the dispersal modes of dry-fruited species may prevent plants from producing very large seeds. Therefore, we hypothesise that the seed size distribution of fleshy-fruited species is more skewed to the right than that of dry-fruited species.

To test the above hypothesis, we compiled a dataset consisting of 161 skewed seed size distributions belonging to 103 species, with 50 fleshy-fruited and 111 dry-fruited species. Our aim was to address the following question: Is the mean value of skewness across fleshy-fruited species larger than that across dry-fruited species?

We first collected data on skewness of intraspecific seed size distribution (hereafter referred to as skewness) from the literature published until 15 September 2023. We conducted a search on papers published in English using the search terms (seed* size* skew*), (seed* mass* skew*), or (seed* weight* skew*) in the ISI Web of Science, Google Scholar, and other online databases such as the Figshare repository and Dryad database. We only included studies that described the names of plant species in our dataset. In addition, to ensure data reliability, we excluded skewed records that were calculated with less than 50 seeds. Moreover, to further increase the sample size of our database, we incorporated skewed data on the sizes of the 58 species of seeds that we measured, which were all collected within Yunnan Province in southwest China. All the species names were checked against the U.Taxonstand Online database (Zhang and Qian, 2024; Zhang et al., 2025). Finally, our dataset included 161 skewness records for 103 species from 72 genera and 33 families, which were categorised as either fleshy fruits (i.e., holding noticeable fleshy pulp and arils, such as baccate, berry, drupe, and pome) or dry fruits (including achene, capsule, follicle, nut, pod, cone, mericarps, and samaras) (Wu et al., 2024).

All statistical analyses were performed using the R v.4.2.3 (R Core Team, 2024). A phylogenetic tree containing all the studied species were constructed using "phylo.maker" function in package "V.PhyloMaker" (Jin and Qian, 2019). The differences in the mean values of skewness between fleshy- and dry-fruited species were tested using phylogenetic generalized linear mixed models (PGLMMs) with a Gaussian distribution. In the model, species names and the phylogenetic covariance matrix were treated as random effects.

Overall, our dataset included 161 skewness records for 103 species from 72 genera and 33 families (Fig. 1). The skewness varied significantly across species, ranging from −1.40 (Erythrophleum fordii) to 4.84 (Cryptotaenia canadensis), with a mean value of 0.22 ± 0.71 (±SD) (Fig. 1D). Fleshy-fruited species (0.45 ± 0.60) had a larger skewness value than dry-fruited species (0.12 ± 0.74) (z = 2.030, p = 0.042; Fig. 1E).

Our study showed that the seed size distribution of fleshy-fruited species was more right-skewed than that of dry-fruited species, that is, many small seeds were accompanied by a few large seeds. Compared to smaller seeds, larger seeds are usually dispersed further (Thomson et al., 2011) and are more likely to germinate and establish seedlings successfully because of their stronger tolerance to stressful environments and stronger competition with both homogeneous and heterogeneous individuals (Tilman, 1994; Muller-Landau, 2010). Therefore, a relatively small proportion of large seeds may play an important role on individual fitness and population dynamics, especially for fleshy-fruited species that have an obvious right-skewed seed size distribution. In the tropics, many species coexist and more than half of them produce fleshy fruits (Wang et al., 2022), raising the question of whether the right-skewed distribution of seed size contributes to species coexistence and biodiversity maintenance. If this is the case, the abundance and number of frugivore species that are decreasing due to current human disturbance and climate change, particularly the decreasing trend and intensity of large animals (Rogers et al., 2021; Fricke et al., 2022), may reduce the probability of large seeds being dispersed. Will this cause the distribution of seed size of these plants to shift to the left, influence the seedling regeneration of these species, and then ultimately affect the species composition of the community?

Previous studies on interactions between seed size and seed dispersers mainly focused on mean values, no matter the impact of dispersal on seed size evolution or the effects of seed size on disperser foraging preferences (Moles et al., 2005; Galetti et al., 2013). However, recent studies have indicated that overall seed size distribution may be an important trait that should be considered. For example, Zhang and Wang (2024) found that biotic-dispersed species have a larger coefficient of variation (CV) than abiotic-dispersed species, and species dispersed by birds have a larger CV than those dispersed by mammals. Shimada et al. (2015) reported that trees that produced seeds with higher mean and CV values of seed size, and lower skewness and CV values of tannin content, had a higher frequency of seed removal by rodents. Therefore, we suggest that in addition to the mean value, the skewness and CV of the seed size should also be carefully considered, because all the three seed size parameters jointly influence foraging preferences of dispersers. Moreover, the influence of dispersal on the evolution of seed size may also simultaneously affect these three parameters, and occasionally, trade-offs may arise among them. Therefore, focusing on a single parameter to investigate the interactions between seed size and animal dispersers may present biased results.

Here, we proposed that the significance of the right-skewed distribution of seed size in fleshy-fruited species is attributable to the similarly right-skewed distribution of disperser body sizes. This is because: (1) small seeds can be dispersed by all frugivores, whereas large seeds can only be swallowed and dispersed by large frugivores and (2) small-bodied frugivores exhibit greater species diversity and abundance than large frugivores. However, this relationship may depend on the size of the seeds. Specifically, if the seeds are sufficiently large relative to the size of the dispersers, smaller dispersers may utilize smaller seeds, while larger dispersers may prefer larger seeds. Conversely, if the seeds are small, this relationship may not hold, as smaller dispersers might also utilize the "larger" seeds within their size range. Therefore, this hypothesis may be contingent upon seed size. However, in our dataset, we did not detect a significant correlation between the skewness and the average seed size among fleshy-fruited species (PGLMM: z = 0.449, p = 0.653), this might be due to the narrow range of the average seed size (i.e., from 0.01 g to 5.10 g). Furthermore, the skewness data were compiled from published papers, which may introduce sample heterogeneity. Therefore, it is crucial to acknowledge that the scale of sample collection can influence the skewness, such as the breadth of the sample area and the number of sampled mother trees.

Despite the prevalence of right-skewed seed size distributions, a significant number of species exhibited a left-skewed distribution, even among those that produced fleshy fruits and depended on frugivores for seed dispersal. This indicates that other processes may contribute to seed size distribution, e.g., seed germination and seedling establishment processes (Cao et al., 2016; Cui et al., 2023); however, further investigation should be conducted in the future to corroborate this hypothesis. In this study, although we only focused on the skewness of seed size distribution, many other important plant traits, such as tannin content in seeds, flower numbers, and diameter at breast height, also showed a certain degree of right skewness (Schmitt, 1983; Wright et al., 2003; Shimada et al., 2015), suggesting that right skewness may be common in nature. However, compared to the mean and coefficient of variation of traits, available skewness data are currently scarce, and most studies rarely provide skewness values. Therefore, given the importance of skewness coefficient, which has long been overlooked, we urge future studies to report skewness values to facilitate subsequent data compilation and integration analyses.

Acknowledgements

This study was funded by National Natural Science Foundation of China (32171533 and 31971444), and Anhui Provincial Natural Science Foundation (2208085J28).

CRediT authorship contribution statement

Xiaoyan Hu: Writing – original draft, Visualization, Software, Methodology, Investigation, Formal analysis, Data curation. Jinyu Zhang: Writing – original draft, Visualization, Software, Methodology, Investigation, Formal analysis, Data curation. Bo Wang: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Funding acquisition, Formal analysis, Conceptualization.

Data availability

The data that support the findings of this study are openly available in Figshare Repository (https://doi.org/10.6084/m9.figshare.28468718.v3).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.