In 2005, the Amazon experienced a once-in-a-century drought. Five years later, in 2010, it was struck by a worse drought, with even lower rainfall occurring in the dry season. Both resulted in a sharp increase in tree mortality and the loss of carbon to the atmosphere.

However, the response of the Amazon forest to these two once-in-a-century droughts also showed marked differences.

The 2005 drought had a much higher rate of tree mortality and a sharper reduction in the uptake of carbon by the forest. This was even though the 2010 drought covered a wider area and was more intense. It suggested different mechanisms were involved in the response of vegetation to the two droughts.

To understand what was going on and therefore better predict how the Amazon would respond to droughts in the future and how this would impact carbon uptake, the researchers looked at satellite and on the ground measurements. They started by comparing the droughts themselves and the periods leading up to them.

They found both droughts peaked in the dry season, from August to October. Another thing they had in common was that at the beginning of the dry season there was sufficient water in the soil and more sunny days than average for photosynthesis, which led to early dry season growth spurts. But that was where the similarity ended.

The dry season drought of 2010 had less rain over the August to October period, occurred over a wider area and was generally more intense than its 2005 counterpart, yet it had less impact on Amazon vegetation. Why?

The answer was found in the lead up to the dry season, particularly the transitional period from the wet-to-dry season (May-July).

In 2005, this transition period was considerably drier than normal, whereas in 2010 the amount of rainfall was only slightly below average. This meant there was more water available in the soil during the drought period in 2010 than in 2005.

As a result, in 2005 more trees died after the initial growth spurt due to the shortage of available water. In 2010, growth slowed after the initial early season greening but enough water remained in the soil to prevent tree mortality. Hence even though the dry season drought of 2010 was worse, the tree mortality and reduction in carbon uptake was smaller than in 2005.

Interestingly, intentional burning of the Amazon forest for land clearing in both years reduced incoming radiation, suggesting that heat stress and water reduction could have been far higher.

The researchers noted in their paper that this burning combined with the forest response to drought must together be taken into account when trying to understand the impact of future droughts on the Amazon.