A couple of days ago, I posted on the news that Dr Simon Lewis, a rainforest expert from the University of Leeds, has filed a complaint about an article written by Jonathan Leake at the Sunday Times. Leake's article concerned the IPCC's use of "grey" literature to support a claim that the Amazon is very sensitive to drops in rainfall and that as much as 40% was in danger of being wiped out by small reductions in precipitation.
The use of grey literature is not disputed, but Lewis has complained that the article's headline, presumably written by an editor rather than Leake himself, implied that the sensitivity claim was false. This, he says, is not the case, "there is a wealth of scientific evidence suggesting that the Amazon is vulnerable to reductions in rainfall". It is just that the citation was "bizarre".
In this second piece on the complaint, I'm going to take a look at the science supporting Dr Lewis's position.
Dr Lewis helpfully expands on his understanding here:
It is very well known that in Amazonia tropical forests exist when there is more than about 1.5 meters of rain a year, below that the system tends to ‘flip’ to savanna, so reductions in rainfall towards this threshold could lead to rapid shifts in vegetation. Indeed, some leading models of future climate change impacts show a die-off of more than 40% Amazon forests, due to projected decreases in rainfall. The most extreme die-back model predicted that a new type of drought should begin to impact Amazonia, and in 2005 it happened for the first time: a drought associated with Atlantic, not Pacific sea-surface temperatures. The effect on the forest was massive tree mortality, and the remaining Amazon forests changed from absorbing nearly 2 billion tonnes of CO2 from the atmosphere a year, to being a massive source of over 3 billion tonnes.
In support of his position, Lewis cites a number of authorities and I will look at these in turn.
When the Amazongate story first broke, Daniel Nepstad was one of the first to defend the accuracy of the claim (while again acknowledging the incorrect citation). Nepstad works at the Woods Hole Research Center, a green campaigning organisation and issued a statement declaring that 40% was the correct figure. Nepstad is slightly unclear about the true source of the figure, saying that the authors of the grey report used by the IPCC had got the information from "IPAM", which turns out to be another green NGO. He then talks about having found that 15% of the Amazon was severely drought stressed in 1998 and that in a later article he had shown that "half of the forests of the Amazon depleted large portions of their available soil moisture during seasonal or episodic drought".
Astute readers will notice that this is not the same thing as saying that 40% would be wiped out by small reductions in rainfall. Indeed, as Nepstad himself notes in a report published in 2007 (and one that was not cited by Dr Lewis)
One of the great ecological puzzles of the Amazon forests is their ability to withstand severe seasonal drought with no visible signs of drought stress.
As far as I can tell, it appears that the Amazon is quite able to deal with drought, even severe drought. For dieback to take place, it requires repeated drought.
Philips et al 2009 (PDF)
Oliver Philips is a colleague of Simon Lewis at Leeds University and indeed Lewis himself is among the 66 (!) authors. As far as I can tell, Lewis reckons this is among the key pieces of evidence to support his claim, but it is hard to see why. The study looks at the severe Amazonian drought in 2005 and found that in these circumstances the Amazon turns from a carbon sink into a net emitter of carbon. Philips et al seem to concur with Nepstad that it is prolonged drought that is the problem:
Prolonged tropical droughts can kill trees...and some models predict climate-induced Amazon dieback this century.
I see nothing in the paper that could support a claim that 40% of the Amazon could be wiped out by small changes in rainfall.
In fact this is a position that Dr Lewis himself seems to agree with. In a recent post at RealClimate he trenchantly sets forward his views on what matters regarding Amazon dieback. Discussing the findings in Philips et al he points out that trees were affected by the major drought, but that this is not the key point:
The evidence for the possibility of a major die-back of the Amazon rainforest is due to two factors,
1. That climate change induced decreases in rainfall in the dry season occur, and
2. The trees cannot tolerate these reductions in rainfall.
The critical question is how these forests respond to repeated droughts, not merely single-year droughts. The forests are of course able to withstand these single droughts (otherwise there would be no rainforest!) — it is their ability to survive an increased frequency of the most severe droughts that is critical to answer.
So, everyone agrees that dieback of the Amazon is predicated on repeated droughts. In order to justify the 40% claim of the IPCC we therefore need to see that small but repeated reductions in rainfall cause dieback.
The evidence that repeated droughts will take place appear to be largely based on application of models. In his PCC complaint, Dr Lewis cites Huntingford et al and Nepstad et al, but in his RealClimate article he points to a paper by Bett et al as well. Before looking at these I want to discuss a couple of points made in the RC article.
Firstly, there is considerable doubt over the model output, in particular with respect to the idea that Amazonian rainfall will decrease in a warming world:
I should add that there is considerable uncertainty associated with the models suggesting decreases in rainfall, and uncertainty as to how Amazon forests may react (especially when one considers the impacts of deforestation, logging, and fire combined with climate change impacts). But this uncertainty is being chipped away at by scientists, a task in which the Samanta et al. paper assists.
But he also sets out the IPCC view on these models. Mostly, he says, the modelling work has been ignored by the IPCC, who cite Betts and Huntingford and make the infamous 40% claim. But, he reiterates, this claim is correct:
Rainforest persists above a threshold of rainfall, below which one finds savanna. If this threshold is crossed a landscape dominated by rainforest can ‘flip’ to savanna. Therefore a ’slight’ reduction can lead to a ‘dramatic’ reaction. Of course, evidence of a shift to a new lower rainfall climate regime is needed, and evidence of large areas of forest close to that rainfall threshold would be required for the IPCC statement to be reasonable; there is ample published evidence for both.
You will note that what he states as the preconditions for Amazon dieback here are slightly different to those he gives in the complaint. Putting these together we really have three conditions
- that the Amazon will move to a new, lower rainfall regime (repeated drought)
- that trees are sensitive to repeated drought
- that 40% of the Amazon is close to this threshhold.
The second point seems undeniable, so I don't propose to look into it any further. I will accept Dr Lewis's figure of 1.5m of rainfall per annum as the correct one. However the question of whether the Amazon will move to a regime of repeated droughts looks more interesting.
Will the Amazon dry up?
I am grateful to Professor Hector Maletta of the Universidad del Salvador in Buenos Aires, who left a link to his draft paper on the science of Amazon deforestation. I would urge readers to get hold of this paper themselves, because it is remarkable stuff, much of which bears directly on the question of the IPCC's 40% statement.
The dieback sections of Professor Maletta's paper examine four different modelling studies - Cox, Nepstad, Betts and Lenton all of whom predict rapid large-scale dieback and replacement of rainforest by savannah. I think we can say with some confidence that these papers are the ample evidence referred to by Dr Lewis.
However, the evidence seems highly questionable at best. For example Nepstad suggests that more than half of the rainforest could be gone by 2030. As Maletta puts it:
These dire predictions (mostly inspired by the Amazon 2005 drought and recent El Niño episodes) emerge not from a global or regional model, but as a possible result of the hypothetical persistence of then-recent events (up to the early 2000s) plus the purely theoretical hypothesis of a 'tipping point' to be hypothetically reached if deforestation advances past some unknown percentage of tree cover, thus possibly triggering an 'abrupt change' process of unknown duration. The critical percentage of tree cover that would trigger the dieback process, if it exists, is unknown, though hypothesized to be 30%.
Nepstad's findings, says Prof Maletta, should be treated with caution in view of...
- the lower rates of deforestation observed in recent years
- improved protection of fragile areas
- increasing establishment of protected areas in the frontier of deforestation
- lack of data on crucial issues such as the very existence of a tipping point, and parameters such as the critical value for the tipping point in case it exists, the time it would take for the subsequent process to complete, and the reversibility or irreversibility of the process.
And as Maletta point out, even those who have been the main proponents of the idea of Amazon dieback say that "it remains just a theoretical possibility without much in the way of empirical evidence". This is a long way from Dr Lewis's "ample published evidence".
Both the Cox and Lenton studies predicate the idea of Amazon dieback on the idea of the development of persistent El Nino conditions, another point discussed by Maletta. The problem with this hypothesis is that according to the IPCC,
there there is no consistent indication at this time of discernible future changes in ENSO amplitude or frequency"
As readers will note, this is a serious blow to the idea of a persistent El Nino if this is really required to bring about persistent drought and therefore dieback in the Amazon. However, Lenton at least, choose to differ from this finding, saying, without providing supporting evidence, that a transition towards a higher ENSO amplitude will happen within a millennium. So even Lenton, who is out of kilter with the IPCC consensus, thinks that global warming will only affect amplitude (which will not make El Ninos more persistent or permanent, as required by the dieback hypothesis) and moreover that it will only do this on rather long timescales. This makes Nepstad's claims of Amazonian dieback in the next twenty years look, well, interesting.
A few days reading of the literature is not enough to draw definitive conclusions, but it appears to me that Dr Lewis's position - that there is "ample evidence" that the Amazon will move to a lower rainfall regime is one that is at best debatable and may in fact not be a fair representation of the literature. If any readers want to investigate this claim further, or indeed if anyone wants to dig into the other aspects of Lewis's claims - that large areas of the Amazon receive rainfall close to the 1.5m threshold and that there is the possibility of Atlantic-driven drought as well as El Ninos - I'd be interested to hear from you.