Effects of global change on inflorescence production : a Bayesian hierarchical analysis
暂无分享,去创建一个
The effects of global change on seed production may dramatically impact plant community composition, because species-specific recruitment rates influence species diversity, successional trajectories and invasion rates. We developed a Bayesian hierarchical model to quantify the effects of three global change factors (elevated CO2, nitrogen deposition, and declining diversity) on allocation to inflorescence production of 12 grassland species. We used the results from these analyses to consider (1) how seed production might be affected by global change and (2) whether species within functional groups respond similarly to global change. We found that all three global change factors affected allocation to inflorescence production in different ways. Elevated CO2 decreased the number of inflorescences per unit biomass for all species, although 95% credible intervals overlapped zero for seven of twelve species. Increased nitrogen had both positive (five species) and negative effects (two species) on the number of inflorescences per unit biomass. There were also positive (two species) and negative (three species) effects of declining diversity on allocation to inflorescence production. Only the effects of nitrogen on inflorescence allocation could be generalized to functional groups: C3 grasses generally decreased allocation to inflorescence production with increased nitrogen, while C4 grasses increased allocation to inflorescence production under elevated nitrogen. The cause of this response is unclear, as other traits besides photosynthetic pathway differentiate C3 grasses from C4 grasses in this system (e.g. clonality, seasonality). Overall, our results suggest that global change will strongly affect seed production of grassland species, and that categorizing those responses by ecophysiological traits is probably not desirable. We also discuss the advantages a Bayesian hierarchical framework has over classical statistical models in analyzing these data.