An enemy's enemy is an ally: Competitive indirect interactions mediate coexistence of trees, grasses, and subshrubs in Neotropical Savanna

Savannas are complex ecosystems where multiple growth forms, including grasses, trees, and subshrubs, coexist through intricate ecological interactions. Understanding the mechanisms that promote this coexistence is key to conserving savanna biodiversity. This study investigated the effects of grass competition on the survival and growth of trees and subshrubs, as well as the interactions between these two growth forms. Using three species of trees and three species of subshrubs, we conducted a greenhouse experiment to assess how competition between trees and subshrubs is influenced by the presence of grass. We found that grass competition significantly reduced the survival of tree seedlings, while subshrubs were unaffected. When trees and subshrubs competed directly, subshrub biomass was reduced, but only in the absence of grass. Tree seedling biomass was not affected by subshrubs, but was negatively impacted by grass in the absence of subshrubs. Both root and stem biomass of trees and subshrubs were reduced by grass competition; however, this effect was mitigated when grasses, subshrubs, and trees competed simultaneously. These results indicate that when grasses, subshrubs, and trees compete together, the intensity of competition is reduced, promoting coexistence and contributing to the balance of growth forms in savanna ecosystems. While fire is an important factor in savanna dynamics, our study emphasizes the critical role of competition in maintaining this balance. Future studies should explore how fire and competition interact to further our understanding of biodiversity and ecosystem function in Neotropical savannas.

Methods
Species selection, experimental design, and establishment

To investigate the effects of a native grass on the growth of subshrubs and trees in Brazilian savanna, we selected three congeneric pairs representing these growth forms. We obtained seeds of the grass Paspalum atratum from the Brazilian Agricultural Research Corporation (Embrapa) and collected seeds for most congeneric pairs in Brasília, Distrito Federal, Brazil, with the exception of Copaifera oblongifolia and Anacardium humile, which were collected in Chapada Gaúcha and Grão Mogol, Minas Gerais, respectively, between November 2013 and July 2014. The factorial experiment involved the presence and absence of grass competition (G+; G-), congeneric competitor and without it (C+; C-) (Figure 1), and two growth forms, trees and subshrubs (T; S). For each species, there were 48 replicates for each combination of grass and congener treatments. We sowed the seeds in polyethylene sacks with a diameter of 20 cm and a depth of 40 cm. The grass was sown five months before the other species in February 2014 to ensure its establishment (Figure 1). To avoid shading by grasses on individuals assigned to the grass-free treatment, plants in the G+ and G- treatments were arranged into separate groups. We broke the physical dormancy of Stryphnodendron seeds by immersing them in sulfuric acid for 5 minutes. The substrate for all plants was 70% oxisol subsoil from a site previously occupied by savanna (40 cm deep or more) mixed with 30% washed sand, without any added nutrients. The experiment was conducted in a greenhouse with an automatic irrigation system that provided approximately 7 mm of water per day.

Data collection and statistical analysis

After one year, we collected the biomass from six replicate plants from each combination of factors. The subshrubs and trees were separated into shoots and roots, dried in a drying chamber at 70°C for 72 hours, and weighed on a precision scale (0.0001 g). The remaining plants were used to quantify survival; after 1.5 years, we counted all living plants in each sack to determine if competition from grass and/or congeners affected the survival of trees and subshrubs.