Plant community dynamics of a tropical semi-arid system following experimental removals of an exotic grass

Abstract

Question Does the presence of the exotic grass Megathyrsus maximus influence the abundance and richness of plant communities on Mona Island Reserve? Location Semi-arid shrubland, Mona Island Reserve in the Caribbean Region, Puerto Rico. Methods We used a large-scale experiment and established plots with different management treatments of Megathyrsus (complete and partial grass removal) to evaluate the effects of grass removal on plant community composition, coupled with measurements of microclimatic characteristics at this site. To complement this experiment, we selected areas not invaded by this grass with similar geomorphological features. We performed vegetation surveys of plant species richness and abundance in all plots and compared temporal changes in vegetation among treatments during a period of 31 mo (from May 2007 to December 2009). We also analysed the associations between changes in species richness and plant abundance with changes in environmental conditions at the plot level and at the island level (precipitation and temperature). Results Plant abundance increased following grass removal but changes in abundance were not necessarily linear over time. Species richness was significantly higher in grass removal plots but differences resulted from a consistent decrease in richness in plots with Megathyrsus. Both abundance and richness were highly variable over time regardless of treatment. The evaluation of temporal community dynamics suggested that these may be driven by temporal changes in climate, and in the case of manipulated plots, by changes in microclimatic conditions. Even when new species arrived into plots following grass removal these failed to establish. An evaluation of successional trajectories showed that removing the invasive grass does not lead to communities that are more similar to non-invaded sites, most likely due to changes in precipitation dynamics. Conclusions Grass removal does not necessarily result in predictable successional changes in plant communities where natural regeneration is allowed to occur. Restoring grass-invaded areas in this semi-arid reserve to conditions that simulate current non-invaded sites would likely require restoration strategies that combine grass management with activities that facilitate the establishment of native species. This may be a general occurrence in grass-invaded areas in semi-arid regions.