Current trends in land-use, fire regimes, and climate change may promote savanna expansion throughout much of the Amazon Basin. Although savannas represent an alternative stable state maintained by grass-fueled fires, the sort of savanna that will replace forests and the grass species involved remains unclear.  Confounding the discourse on savanna expansion in the tropics is the existence of floristically diverse native savannas of high conservation value (e.g., the Cerrado of South America) and extensive areas dominated by exotic forage grasses. In a 22,500 km² region of eastern lowland Bolivia I used satellite image analysis to quantify the area of forest replaced by savanna between 1986 and 2005. I then field-sampled forest-replacing (derived) savannas and native savannas to compare their grass and tree species compositions. Because soil fertility can influence native savanna-forest boundaries, I analyzed soil samples from native and derived savannas for nutrient, chemical, and physical properties. To understand the consequences of different sorts of savanna expansion on ecosystem functions (i.e., flammability, maintenance of tree cover), I measured fuel loads in native and derived savannas and compared the abundances of tree species typical of savanna versus those normally found in forest environments.

Results/Conclusions

Savannas replaced 1420 km² of forest (8.4% of forest in 1986), a rate comparable to deforestation for agriculture (1200 km²). Native and derived savannas differed both floristically and in soil fertility; native savannas occurred on sandy acidic soils and derived savannas on relatively fertile soils. Whereas native savannas contained a diversity of grasses, the derived savannas were usually dominated by a single species, including: Guadua paniculata (a native bamboo), Urochloa brizantha and U. maxima (exotic forage grasses), Imperata brasiliensis (a native invader of swidden plots),  Digitaria insularis (a native ruderal), and the  native fire-adapted herb Hyptis suaveolens (Lamiaceae); just one of 17 derived savannas resembled native savannas in grass species composition. The trees in derived savannas were not thick-barked species characteristic of native savannas (e.g., Curatella americana) but forest trees (e.g., Anadenanthera colombrina) and fire-tolerant palms (Attalea spp.). Absent from low fertility sites, grasses in derived savannas on fertile soils produced fuel loads twice that of native savannas. The relationships between soil fertility, grass species composition, and fuel loads could be used to improve savanna expansion models, clarify the sort of savannas that will expand, and emphasize the different ecosystem consequences of native as opposed to derived savanna expansion.