Discussion
By conducting a meta-analysis of studies of 1071 plant species, our results suggest phylogenetic conservatism for the majority of plant traits tested in this study, i.e., seed mass, plant height, leaf area, genome size, and leaf N, as observed in previous studies (Moles et al., 2005; Swenson & Enquist, 2009; Davies et al., 2013; Wang et al., 2022). This is most likely due to the similar traits of phylogenetically closely related species rather than the similarity of traits at higher taxonomic levels. Apart from phylogenetic signal of these traits, we showed that growth form of the 1071 species was not randomly distributed but followed a Brownian model of trait evolution, indicating that the closely related species are more likely to have similar growth form than might be expected by chance (Kerkhoff et al., 2006). The phylogenetic signal carried by plant growth form may potentially explain the differences in seed mass, plant height, and leaf N between the woody and non-woody plant species.
Previous studies have looked at the patterns of associations of seed mass with other plant traits (Westoby, 1998; Guo et al., 2010; Santini et al., 2017), however, few studies have incorporated phylogenetic affiliations in gauging the correlations between seed mass and several attributes. By controlling phylogeny, plant height appears to be a reliable predictor of seed mass across species based on the results of PGLMM. Although dispersal mode and growth form may modify the pattern of association of plant height with seed mass (Thompson & Rabinowitz, 1989; Leishman & Westoby, 1994c; Leishman et al., 1995), we found consistent positive correlation between seed mass and plant height. The positive relationship between seed mass and plant height may facilitate long-distance seed dispersal because seed dispersal distance is more strongly correlated with plant height than with seed mass (Thomson et al., 2011).
Plant height is also closely related to leaf area (Falster & Westoby, 2003), therefore positive relationships between leaf area and plant height are likely to lead to a positive correlation between seed mass and leaf area. In our study, there were also consistent and positive correlations between seed mass and leaf area, suggesting that the correlations between leaf area and seed mass are conserved across life-forms. These patterns accord well with independently gathered data on the relationship between seed mass and leaf area both in the woody and annual species (Senn et al., 1992; Niinemets & Kull, 1994; Cornelissen, 1999; Santini et al., 2017). The positive relationship between seed mass and leaf area may reflect the great contribution of photosynthetic capacity and nitrogen content of leaves.
The role of the relationship between genome size and seed mass has gained much less attention over the two decades (Moles et al., 2005a, b; Beaulieu et al., 2007). Despite several studies that found a quadratic relationship between genome size and seed mass (Beaulieu et al., 2007; Knight & Beaulieu, 2008; Krahulcová et al. 2017), our GLM and PGLMM models showed positive association between genome size and seed mass across 1017 species. This discrepancy is largely due to the fact that previous studies failed to incorporate phylogeny into this consideration. To understand the forces shaping the evolution of seed mass, we will also need to consider other plant traits, such as leaf N and growth form. Without controlling phylogeny, seed size was associated with growth form and woody plants tended to have larger seeds than smaller herbaceous plants, possibly due to the larger height of woody plants than of herbaceous species (Jurado et al., 1991). However, incorporating phylogenetic affiliations into the model failed to detect the clear association between seed mass and growth form across the plant species, indicating that phylogeny conceals the effect of growth form on the variation in seed mass. This finding may not be in agreement with the observation that variations in seed mass are consistently associated with those in growth form (Mole et al., 2005).
Despite the strong phylogenetic signal in several plant traits, our meta-analysis successfully teased apart the relative contributions of phylogeny, plant height, leaf area, genome size, leaf N and growth form on explaining variations in seed mass across the plant species. We first showed that phylogeny had much more power to explain variations in seed mass than did other plant traits, whereas plant height, leaf area and genome size only explained the minority of these variations although the leaf-height-seed (LHS) scheme states that plant height and leaf area are closely correlated with seed mass (Westoby, 1998). Growth form and leaf N explained little variation in seed mass, reflecting the main effect of phylogeny on affecting seed mass variation. Therefore, our study suggests that divergences in seed mass have been more closely correlated with phylogeny than with divergences in other plant traits. If this pattern holds equally for plants of different taxa, investigation on the correlations between plant traits should not ignore the contribution of phylogeny.
Together, our results indicate that although there is a significant phylogenetic signal in plant traits (seed mass, plant height, leaf area, genome size, leaf N and growth form), phylogeny appears to explain variations in seed mass better than these explanatory variables. This implies that phylogeny has acted as one of the most important factors that govern the variations in seed mass across plant species. Our work is expected to open the door to further investigate the contributions of phylogeny and explanatory attributes to the variation in given plant traits.