Investigator:

Martin Bouda

Advisor: Peter Raymond

Description:

I will attempt to use satellite imagery to investigate the causes of the counter-theoretical green-up observed in the Amazon region during the 2005 growing season (see Saleska et al. "Amazon Forests Green-Up During 2005 Drought" in Science Vol. 318, 26 Oct. 2006). In particular, the hypothesis I will be attempting to test is that the increase of canopy chlorophyll concentration during the drought is due to increased productivity of sub-dominant species with greater water use efficiency (WUE), following an initial wilting/drought deciduous behaviour of the dominant species. The initial wilt should show up as a slump in a vegetation index over a broad area, followed by a new (and anomalous) increase in the index as the sub-dominant species take over. Should this pattern be observed in areas that showed up as anomalously green in the Saleska paper and not in other regions of the amazon, I would wish to pick one or more smaller areas within each of these categories and try to use spectral analysis to see if a difference in the composition vegetation cover can be detected this way. The leaves of the plants with greater WUE should have different characteristics than those of the usual dominants, which I hope will show up in their spectral signature.

The project should help me with establishing a region for my Master's research, which will deal with changing hydrology and its influence on species composition of plant communities. The Amazon is one region I am considering to work in and if the hypothesis outlined above is borne out by the evidence, I would in all probabiliy choose to go there.

The broader significance of this research is that it should inform the way that amazon rainforests are modeled in coupled atmoshpere-vegetation models. The current models predict that the rainforest will collapse to savannah as a result of the influence of climate change on its hydrological cycle. There are very good reasons to believe this model prediction is incorrect (because of the 2005 observations and also because the models predict this collapse of the forest when fed atmospheric data from the past several decades, when they it did not in fact turn to savannah). One way of improving the predictive power of the model would be to include in the models sub-dominant components of the forest community with diverse strategies of obtaining and conserving water. If it can be shown that such sub-dominant species give the community greater resilience in the face of drier conditions, they would no doubt be incorporated into the model as a separate component of the vegetation. This may lead the models to accurately predict the actual fate of the amazon forest in the near future.