Direct and indirect effect of seed size on seedling survival along an experimental light availability gradient

Theoretical models of life-history strategies assume a positive relationship between seed size and subsequent offspring survival (seed size–survival relationship). There is limited empirical evidence, however, on how this relationship changes across environments or in relationship to seedling traits. Moreover, characterizing the dynamics of seedling recruitment under natural conditions is central to understanding how seedling recruitment may in turn affect large-scale ecological processes. Here, first-year seedling survival was monitored from emergence for 303 angiosperms species originated from alpine and sub-alpine meadows of the eastern Tibetan Plateau grasslands, across an experimental light gradient consisting of four treatments. We used linear models (LM)and phylogenetic generalized linear models (PGLM)to test for the seed size–survival relationship, and how this relationship differed across light treatments. We also used path analysis (PA)and phylogenetic confirmatory path analysis (PPA)to assess how seed size interacted with other seedling traits (seedling emergence time, specific leaf area, root: shoot biomass ratio, and biomass growth)to either directly or indirectly affect seedling survival. We found seed size to be positively associated with seedling survival only under low to medium light treatments only in LM (not in PGLM). PA and PPA revealed that the positive effect of seed size on seedling survival was indirect, mainly acting via biomass growth. Under low light, larger seeds exhibited greater biomass growth, which in turn increased seedling survival. In contrast, the direct effect of seed size on seedling survival is negative. In sum, the seed size–survival relationship appears to be both environmentally and phylogenetically dependent. The survival advantage of large seeds appears to be the result of their ability of higher biomass growth after emergence in low resource environments. However, our results also suggest there may be an underlying trade-off with larger seeds facing a lesser, but direct risk of increased mortality. The influence of environment on seedling traits, the interaction among traits, and phylogeny should be taking into consideration when modeling the dynamics of seedling recruitment.