How many buffalo does it take to change a savanna? A response to Bowman et al. (2008)

Bowman et al. (Journal of Biogeography, 2008, 35, 1976-1988) aimed to explain observed increases in woody cover on floodplains and savannas of Kakadu National Park using estimates of buffalo (Bubalus bubalis) density as a causal variable. They found that buffalo were a minor model variable and concluded that buffalo are 'not a major driver of floodplain and eucalypt dynamics'. However, the authors mislabelled the historical density of buffalo on their site, citing a period as high density instead of low density. Further, their results were not contextualized within the substantial body of scientific and historical evidence of the buffalo's strong influence on vegetation in Kakadu. The authors instead postulated three unanalysed drivers of observed patterns of change: fire regime, rainfall and atmospheric CO2. We suggest that further analyses of change in woody vegetation should make use of accurate historical records of grazers as well as available data sets on fire history.