Weed control, large seeds and deep roots: Drivers of success in direct seeding for savanna restoration

Aim We aimed to evaluate the performance of native tree species in the restoration of savanna vegetation by direct seeding, to assess whether weed control and intercropping with native grasses can contribute to the success of this method and to determine whether species performance can be explained by functional traits. Location Cerrado biome, southeastern Brazil. Old fields abandoned after decades of land use as pasture or croplands, occupied by ruderal plants and invasive grasses. Methods We established a direct seeding experiment with ten tree species native to the cerrado (Brazilian savanna) region. We used a factorial design in five blocks, with the following factors: (a) species; (b) weed control and (c) intercropping with native grasses. We evaluated the emergence, survival and growth of plants, and through multiple regressions we sought to explain the success of the species in direct seeding based on their functional traits. Results Emergence and survival in the field varied widely among species but with little or no difference between treatments. Growth was compromised by weed competition in all species. Intercropping with native grasses did not decrease weed competition. We found a functional pattern associated with species performance in direct seeding, where survival in the field is positively associated with seed mass, root depth and a greater root:shoot ratio. Conclusion Weed competition impairs seedling growth more than survival, considerably delaying restoration by direct seeding. The success of this method in the savanna will depend on the use of species that are best adapted to environments where water stress is the main obstacle to overcome and have large seeds and seedlings with a large and deep root system. Characterizing underground seedling systems is essential for predicting the success of cerrado species in direct seeding.