Ten years of directing seeding restoration in the Brazilian savanna: Lessons learned and the way forward.

Savannas and grasslands have lost almost 50% of their original cover worldwide. Therefore, the development of methods and information on open-canopy ecosystem restoration is urgent for the inclusion of these ecosystems into global and regional priorities. In the Brazilian savanna, the most diverse savanna in the world, restoration efforts focused on open ecosystems have been virtually absent, but have increased in the last 10 years. Such efforts are frequently threatened by invasive exotic grasses (IEG) that invade and dominate areas excluding native species, oftentimes aided by altered soil conditions. Long-term studies of savanna restoration trajectories are rare. In this study, we surveyed 22 savanna restoration areas established two to ten years before the study with similar restoration methods to assess their current status. We show that the current restoration methods are successful in establishing native species and allowing species turnover but they are threatened by IEG. Restoration success varies and is affected by soil conditions, IEG landscape cover and post-sowing weeding. Despite that, the simultaneous introduction of different plant functional groups allows turnover from fast to slow-growing plants. Establishing savanna native species is possible at an operational scale with current knowledge and techniques. However, native species establishment fails to prevent IEG reinfestation, which needs to be managed in restoration efforts in the Brazilian savanna.

Factors affecting savanna and forest regeneration in pastures across the cerrado.

The Cerrado region comprises the world's most biodiverse savanna and the largest cultivated pastures for cattle in Brazil. Forty percent of these pastures are unproductive or degraded, with bare soil and native vegetation increasingly replacing exotic forage grasses. This study sought to investigate the regeneration of native vegetation in the pastures of the Cerrado and to evaluate the contribution of biophysical, land management, and landscape attributes to this process. Across the Cerrado, we analyzed pasture plant communities and the attributes of pasture management intensification, fire events, landscape native vegetation cover, and climate and soil types of 93 active pastures and 15 abandoned pastures. For the abandoned pastures, time since abandonment was an additional variable. On actively cultivated pastures, savanna regeneration varied from 0 to 70%, with a diversity of herbs and woody species. Pasture management was the main predictor of savanna regeneration on cultivated pastures. On abandoned pastures, time since abandonment was the main predictor. Exotic grass cover had a strong negative relationship with savanna regeneration and they were present even in pastures abandoned for 44 years. Our study reveals the potential of natural regeneration of the Cerrado and its particular predictors. The occurrence of pastures with high natural regeneration indicates that national policies can promote native vegetation restoration and silvopastoral systems with predictable, low cost implementation.

Feasible or foolish: Attempting restoration of a Parthenium hysterophorus invaded savanna using perennial grass seed.

The annual herb Parthenium hysterophorus L. (Asteraceae), remains one of Southern Africa's most significant invasive weeds, commonly invading savannas, and their rangelands, causing severe losses to agriculture, livestock production and native biodiversity. Previous studies have suggested that perennial grasses may act as useful competitive species, capable of suppressing the growth and invasion of P. hysterophorus. To explore this, a total of 48 plots were established within an invaded savanna, using a randomised block design, and included treatments with and without the clearing of P. hysterophorus, as well as with and without the sowing of native perennial grass seed (Anthephora pubescens, Chloris gayana, Cynodon dactylon, Digitaria eriantha, Eragrostis curvula, Panicum maximum and Themeda triandra). Plots were assessed yearly in terms of P. hysterophorus density and growth as well as grass species composition, basal cover, and biomass over a three-year period. Clearing alone was found to exacerbate invasion, increasing P. hysterophorus density by 40%. Whereas the sowing of grass seed, in both the cleared and uncleared plots, increased the abundance of perennial grass species by 28%, subsequently reducing the size, reproductive output and density of P. hysterophorus over the three years. In addition, these sowing efforts contributed towards partial restoration of the plots, enhancing grass basal cover by ~15% and biomass production by 17%. Overall, this research suggests that sowing of native grass species, with or without clearing, may be a useful supplementary control or restoration tool towards the long-term management of P. hysterophorus invasions in managed savannas and rangelands in Southern Africa.