Response of savanna gross primary productivity to interannual variability in rainfall: Results of a remote sensing based light use efficiency model

Studying the temporal pattern of savanna gross primary productivity (GPP) is essential for predicting the response of the biome to global environmental changes. In this study, MODIS satellite data coupled with eddy covariance based flux measurements were used to estimate GPP using a remote sensing based light use efficiency model across a significant rainfall gradient in the Northern Territory (NT) region of Australia. Closed forest that occurred in wet and often fireproof environments assimilated (GPP) 4-6 times more carbon than grasslands and Acacia woodlands that grow in arid environments (<600 mm annual rainfall). However, due to their small spatial extent, closed forests contributed <0.5% of the regional budget compared to savanna woodlands (86%) and grasslands (32%). Annual rainfall was found to exert a significant influence on GPP for different vegetation types except for closed forest which was less sensitive to above-average rainfall. Interannual variability in GPP showed that arid ecosystems had a higher variation (>20%) compared to woodlands and forest (?5%). This variation in GPP was correlated with that of rainfall (R2 = 0.88, p<0.05). Analysis of the impact of wettest and driest years on GPP showed a strong positive correlation between the magnitude of the relative maxima in rainfall and maxima in GPP (R2 = 0.89, p<0.05). In contrast, the relative rainfall minima exhibited an insignificant relationship with relative GPP minima (R2 = 0.45, p = 0.07). These findings provide valuable information on the carbon uptake across the savanna biome and show the sensitivity of different vegetation systems to rainfall, a variable that may change in quantity and variability with projected climate change. Such data also show regions of high levels of carbon that could be linked with savanna management to protect the resources in the Australian savannas.