The role of seed limitation in Serengeti savanna tree recruitment
Previous work in Serengeti has demonstrated compositional turnover in adult trees across a gradient in mean annual precipitation (MAP) with Acacia tortilis dominating the dry end and Acacia robusta the wet end. Most recruitment limitation studies in savannas focus on effects of bottlenecks on seedlings or mature trees, with fewer trying to understand how seed limitation may contribute to recruitment and compositional turnover. To understand the role of seeds in tree recruitment, we quantified seed production, infestation and germination for A. tortilis and A. robusta species at a series of sites spanning the Serengeti MAP gradient in 2013 and 2014. Our main question was: what is the role of recruitment limitation at the seed stage and could variation with MAP serve as a mechanism for species turnover across Serengeti? We used mixed models fit by maximum likelihood to test for field-measured seed production, infestation rates and laboratory –measured germination rates as a function of MAP and tree height (for seed production). We used AIC - based model selection to test for effects of MAP and species differences for 2013 and 2014 data.
Model selection results showed that the proportion of trees producing seeds increased with tree height, but did not differ between species. Infestation rates differed between species within years, but reversed in order from 2013 to 2014. Germination rates under laboratory conditions were higher in A. tortilis than in A. robusta and were strongly reduced by insect infestation for both species. Contrary to expectations, infestation did not completely inhibit germination. In general, none of the processes investigated (seed production, infestation or germination) varied across the MAP gradient, with one exception: laboratory germination rates in A. robusta were positively related to MAP at collection site, suggesting that water availability limits the germination potential for this species across its range. Our results provide only weak support for the notion that rainfall-mediated effects on seed limitation explain the observed turnover of dominant tree species across the Serengeti MAP gradient, suggesting that post-germination filters may be more important.