In a recent issue of PNAS, Carson et al. (1) show evidence for forest expansion over human-occupied savannas ∼2,000 y ago in northeast Bolivia. This study fundamentally changes our understanding of the magnitude and nature of pre-Columbian land use in the Amazon region. The authors conclude that previous theories of land management, as well as its expected impact on global biogeochemical cycles, are “potentially flawed” because they did not consider climate-driven forest–savanna biome shifts. One important aspect not explored by the authors is that such forest–savanna shifts and, by extension, their ecological and anthropological implications are by no means unique to that region.

Analyses of fossil pollen near Carson et al.’s site revealed that forests have been expanding since the mid-to-late Holocene. During that period, almost 1,000 miles away, riparian forests began expanding over savannas in central Brazil (2). This process was followed by increased tree density and abundance of forest species further south ∼2,700 y ago (3), and expansion of araucaria and Atlantic forests over ancient grasslands of southern Brazil ∼3,500 y ago (4). Similar shifts in C₃/C₄ species abundance and increased woody cover ∼3,900 y ago have also been documented in many ecosystems of central Argentina (5). These findings show a predictable (latitude-dependent) pattern of vegetation dynamics following the end of the Holocene climatic optimum (Fig. 1).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Beginning of forest expansion during the mid-to-late Holocene in ecotonal regions of South America, based on average radiocarbon dates reported in palynological (sites 1 and 3) and isotopic studies of soil organic matter (sites 2, 4, and 5). Site numbers match reference numbers.

Although there certainly exist cases in which forests and savannas have been stable over time, consideration of these dynamic environments is relevant for two reasons. First, they represent ecotones or transition zones, where vegetation responses to climate fluctuations should be more pronounced. Second, these are regions where forests have expanded despite significant increases in human occupation, evidenced by charcoal particles in sedimentary records, which imply frequent fires during the late Holocene (1, 3, 4). Combined, these observations offer further support to Carson et al.’s (1) interpretation that forest expansion has occurred in response to changes in climate despite increased anthropogenic disturbance. Moreover, the findings suggest a regional to continental process defined by limits of modern biomes, which will improve future studies of human–environment interactions.

Notably, changes in vegetation structure are typically accompanied by changes in floristic composition, soil fertility, and faunal distribution (2, 4, 5), which could have affected the ability of human populations to persist. There is, nonetheless, a key piece still missing from the puzzle: direct evidence for past changes in climate. Among other proxies, Carson et al. (1) mention changes in dust concentration and oxygen isotope ratios in lacustrine calcite, arguing that shifts toward wetter conditions promoted forest expansion. However, these climatic records come from the high Andes and it is unclear whether they can be used to attribute causality in climate-vegetation shifts. Differences between human and climatic impacts can be elucidated through reconstructions of lowland water regimes and should be a research priority in these and other areas of dynamic anthropological and ecological progressions.