HOT WAR: Researchers found a link between El Nino years--when the climate is hotter and drier in the tropics--and the outbreak of civil wars in tropical countries.
Image: © iStockphoto.com / Matt Theilen
-
The Best Science Writing Online 2012
Showcasing more than fifty of the most provocative, original, and significant online essays from 2011, The Best Science Writing Online 2012 will change the way...
Read More »
In 1957, civil war broke out in then Burma (now Myanmar) and Oman. In 1965, similar conflicts erupted in Burundi, Chad, the Dominican Republic, Indonesia and Peru. In 1997, the Comoros, Congo, Eritrea, Niger and Rwanda also saw battles over national power that each killed more than 25 people. What do all these years and countries have in common? First, they are all tropical countries, strongly affected by the weather pattern known as El Niño, a warming of equatorial waters in the eastern Pacific Ocean that occurs cyclically every three to seven years and generally makes the climate hotter and drier throughout the tropics. Secondly, these were all years in which the global climate was enduring an El Niño.
In fact, research published August 24 in Nature now demonstrates a link between El Niño years and civil war in 93 tropical countries. (Scientific American is part of Nature Publishing Group.) "Since 1950, one out of five civil conflicts have been influenced by El Niño," says economist Solomon Hsiang of Columbia University, lead author of the research. "This represents the first major evidence that global climate is a major factor in organized patterns of violence."
Starting with data on conflicts that killed more than 25 people, as compiled by the Center for the Study of Civil War to include 175 countries and 234 civil wars in the last six decades or so, the researchers mapped out how many of these disputes occurred in years with an El Niño weather pattern. They found that the risks of civil war breaking out in a tropical country during an El Niño doubled. Then, running a comparative simulation in which such El Niño weather patterns did not occur, the researchers determined that the hotter, drier conditions helped stoke 48 civil wars that did not occur in the modeled El Nino-free world. "Even in this modern world, climate variability has an impact on the propensity of people to fight," says climate modeler Mark Cane of Columbia University. "When people get warmer than comfortable they get irritable and they are more prone to fight."
That is one hypothesis, at least, for why this effect would be seen, along with the fact that El Niño conditions are associated with declines in crop yields across the globe, which may contribute to conflict. It is also clear that El Niño alone does not inspire civil war; it just ups the risk. And it is evident that rich tropical countries, such as Australia, do not see this linkage between warmer and drier conditions and civil war. Hsiang likens the finding to ice patches on roads in winter: the ice patches increase the likelihood of car crashes, just as El Niño conditions exacerbate underlying tensions in a given society. "The state of global climate can determine whether conflict is more or less likely," he says.
That could be bad news as the global climate is changing in a generally warmer direction thanks to increasing concentrations of greenhouse gases, like carbon dioxide, in the atmosphere. "It is frankly difficult to see why that won't carry over to a world that is disrupted by global warming," Cane says. "If these smaller, shorter-lasting and, by and large, less serious kinds of changes associated with El Niño have this effect, it is hard to imagine that the more pervasive changes that come with anthropogenic climate change are not also going to have negative effects on civil conflict.”
Certainly, history has been influenced by climate changes, ranging from the drought-driven collapse of Mayan society to the partially climate-induced rising food prices that helped to touch off the more recent "Arab spring." In fact, diminished agricultural production as a result of climate variability matches up with fluctuations in the frequency of war, mass migrations and even regime change in the historical record. "It seems to us that the climate-war association is not only valid in the past, but also valid in present days," says geographer David Zhang of the University of Hong Kong, who was not involved in this study but has analyzed world history and found linkages between war and temperature change in previous published research.
Ultimately, because El Niño patterns can be predicted as much as two years in advance, the world might have forewarning of the potential for increased conflict. "At a minimum, national governments and international institutions should be ready for such events," says economist-in-training Kyle Meng of Columbia University, who also worked on the latest study.
But Meng and Hsiang's work cannot explain the root causes of such conflicts, merely showing a correlation between the outbreak of civil war and El Niño events, so it may not prove a useful guide. "Correlation without explanation can only lead to speculation," says political scientist Halvard Buhaug of the Peace Research Institute of Oslo.
And, of course, forewarned does not mean forearmed. Researchers predicted the present famine in the Horn of Africa more than two years ago and yet nothing was done. "It was not until the famine and violence actually began that donors began supplying the resources needed," Hsiang notes.
Nonetheless, Hsiang's research does seem to show that, for the more impoverished countries of the tropics, the linkage between climate and conflict has yet to be broken—even if it may have been in richer countries that are more technologically insulated from such shocks, thanks to air conditioning, food surpluses and other signs of wealth. As Hsiang says, "In the modern world, we still depend on climate to a very large extent."
Rights & Permissions
See what we're tweeting about
27 Comments
Add CommentThis is antique environmental determinism, reshod. Where to begin...?
Reply | Report Abuse | Link to thisIt has been established that El Nino events take place about once every 5 years. The authors consider 234 conflicts in the past 60 years. A period of 60 years will tend to have about 12 El Nino events, which is 20% of the years in question. 20% of 234 conflicts is 47 which about equals the number they give (48) of conflicts supposedly motivated by El Nino. This is simply the expected number of conflicts which would happen to fall on El Nino years and IS NOT evidence of causality.
Furthermore, comparing skirmishes where about 30 people died with wars involving hundreds of thousands dead is, by any criteria, to compare apples and oranges.
And finally, there has never been any mystery regarding the real causal factors of the major conflicts of the past 60 years in the tropics. We might begin with noting the dissolution of colonial regimes, ethnic conflicts because of borders drawn under those regimes, the Cold War, and so on.
What I do find mysterious is that Scientific American is helping give a voice to this new brand of environmental determinism.
You are spot on. But did we expect anything else from these environmentalists. They use similar methods to "prove" global warming is human caused, it exists at all and what the supposed cause is.
Reply | Report Abuse | Link to thisOf course, they also use similar proof to predict that once we have a globe warmed to a so called average of 2 degrees more than today, humans will just throw in the towel and start WWIII. It is no wonder they think El Ninos cause civil wars, they already think warming of any kind causes wars.
I started to chuckle when 'Arab spring' and 'collapse of Mayan society' are in the same sentence.
Reply | Report Abuse | Link to thisUnfortunately environmentalism as a positive force is continuously besmirched by 'the cult of environmentalism'.
Connecting war with global warming is actually nothing new. The US military has already recognized that global warming represents an increasing threat to national security and are already preparing for the kinds of conflicts they expect to see. A comprehensive summary of concerns that we should be aware of are detailed in this 2007 report:
Reply | Report Abuse | Link to thishttp://securityandclimate.cna.org/report/National%20Security%20and%20the%20Threat%20of%20Climate%20Change.pdf
It has been compiled through extensive research and draws on consultation from several retired military three- and four-star generals and admirals. Military preparation is in progress:
http://www.nytimes.com/2007/04/15/us/15warm.html
http://www.time.com/time/specials/2007/article/0,28804,1730759_1731383_1731632,00.html
http://www.nytimes.com/2009/08/09/science/earth/09climate.html
http://www.npr.org/templates/story/story.php?storyId=121352495
@Kayrock:
I'm not going to subscribe to Nature just to read the article (not by itself at least, and not today), but anyone worth their salt can click the link, read the abstract, and see these researchers aren't operating on the same junior high math you apparently are. Just because you've found a way to randomly punch numbers to get something that is ONLY in the viscinity of (not even equal to) another number they gave doesn't mean that's how it was arrived at. This should be pretty obvious just reading the author's description.
Unsurprisingly, SA did an article about precisely that military preparation I am talking about:
Reply | Report Abuse | Link to thishttp://www.scientificamerican.com/blog/post.cfm?id=beware-the-military-industrial-comp
Junior high math my dear and most rude "da bashtid"? No, it's just straightforward numbers - deal with it, and while you're at it, learn grammar school spelling. By "viscinity" I assume you meant to write "vicinity". And in the world most of us live in, 47 is extraordinarily close to 48. Have a good day and please try to be more courteous in your exchanges.
Reply | Report Abuse | Link to this@"da bashtid"
Reply | Report Abuse | Link to thisAnd moreover, this topic is not about global warming, as you evidently misunderstood in your visceral reaction. Please read carefully. It is about El Nino. Not the same.
Yes, Global Warming has a mind of its own, controls everyone and decides entire nations will attack each other, that is reasonable.
Reply | Report Abuse | Link to thisThe reality da bahstid is there is simply no way to get the right kind of information to determine if El Ninos cause war. Global Warming is no different. You do not know enough to prove human caused global warming even exists, let alone claim it will cause wars.
There are as many reasons humans go to war as there are wars. I doubt you will find any evidence, such as a diary of a general that says, one day he woke up and decided it is just too damn hot out today, so lets go invade someone.
Sure humans will war over space and resources. Yes El Nino can reduce those resources, making other space more valuable. But the real cause of this is some humans being unable to find solutions to their problems and instead choose to invade someone else because it is easier than actually working.
I agree with Kayrock completely. Correlation in a smallish data-set does not imply cause-effect relationship. And, to put the Rwandan genocide as "war" due to famine? Looks like the researchers were just trigger happy to report their analysis without giving the conclusion second thought. Or, they probably thought why not publish it and people can discuss and debunk it. Hey we get a paper out of it, heck of a publicity and tons of citations, though perhaps negative citations. My adviser once said, the easiest way to get citations is to put out a very bad theory and then everybody can refer to the paper and refute the theory.
Reply | Report Abuse | Link to thisWell Kayrock, let's take a look at what you said versus what the researchers said. The researchers said that during El Nino, the risk of conflict approximately doubled, and the additional total count was 48. That means that if you model a baseline using non-El Nino years, the total count of conflicts above the baseline that occurred during El Ninos was 48.
Reply | Report Abuse | Link to thisYou have observed that 20% of the total conflicts is about 47, and that is very close to 48. Your math is correct, but it has little to do with the observations of the researchers. They did not say that the total conflicts during El Nino years was 48; they said the total count above a background level was 48. Let's see, if you expect a mean value of 47, and you get 48 more than that, you get 95. 95 is about double 48. (That might not be what the researchers did, but I'm making a point with _simple_ math.) If I were vindictive, I'd say that all of you agreeing with Kayrock just failed the word problem part of their math test.
Look, it should not be a surprise that hungry people tend to be more restless than well fed people. Most of Russia's wheat exports go to the Middle East, and Egypt is (or maybe was) their largest customer. In the Arab Spring, Egypt went first. That is just one example, but there seems to be a pattern between people who had to pay more for food and conflicts that have happened.
Why did "Let them eat cake." really piss off the Parisians? (Although, that may have been wrongly attributed, but still, it stuck.)
Here, have a look at the supplemental material, but unless you are better at math than Kayrock, it won't mean much to you.
http://www.nature.com/nature/journal/v476/n7361/extref/nature10311-s1.pdf
Just skimming this material shows that the researcher's work is a lot more nuanced than what can be effectively presented in an short news article. So, Kayrock et al have made an error of oversimplification of material that has already been greatly simplified.
What was Kayrock saying about "...it's just straightforward numbers.."?
BTW, priddseren, you never did answer my question about which temperature data set you think has problems, and here you are still spewing that assertion that the record of warming is somehow not to be trusted.
Reply | Report Abuse | Link to this@ChrisG -- Having re-read the story after reading your comment, I see that you may be correct. If so, the story is very poorly written. Why did they not give the number for the baseline? Why did they not directly state the number of conflicts which occurred in El Nino years? In computer programming we don't call that simplifying, we call it obfuscating. The relevant supporting numbers are not displayed, they are implied.
Reply | Report Abuse | Link to thisThis is actually the second story I read on the subject, and the first directly reported Kayrock's apparent misunderstanding, stating that 50 out of 250 conflicts occurred in El Nino years, no mention of "over the baseline". I'm not about to pay for a subscription to Nature just to see if the explanation is more clear there, especially after seeing that others who have read the material have come away with bad numbers. Someone should do a study on how global warming causes poor journalism.
Assuming that you are right, and the study actually found that 40 percent of the civil conflicts started in the 20 percent of years which were El Nino, the real question becomes "How do I get a job doing such easy work?" Drought and famine make war more likely, and even moreso in underdeveloped countries? Wow! I've been collecting data for over 50 years, noting that the sun comes up every day, and I feel I can predict, with a high level of confidence, that it will continue to do so, unabated, for AT LEAST another 50 years! Gotta go...have to start writing a paper to submit to Nature!
I just read the whole article and its supplementary files. I would suggest you do the same.
Reply | Report Abuse | Link to thisThe researchers do not say that the total count of conflicts about a background level was 48; rather, they state that 48.2 conflicts were associated with El Nino events, and only on the order of about a 3% tendency, and with that we’re back to my previous comment.
What they discuss at length is the ACR (annual conflict risk) which they calculate through consideration of a complex series of variables. The short version is that they believe that upward variations of ACR are statistically associated with El Nino events. The maximum average ACRs discussed are only on the order of about 3%, and as jsaaymi mentions, this is all based on a very small data set within which spurious correlations would not be unexpected.
Even within that very limited data set, the authors undertook some heavy-handed data selection. If I may paraphrase to avoid possibly objectionable excessive quoting, they restricted their sample universe to 1950-2004 because they recognized extreme outlying data sets for 1946 and 1948. The data set for 1989 was also rejected. As they state, “…[T]hese outliers exert a large and probably unreasonable influence on our model, so they remain omitted….”
I make no claim that my comments are profound. I am merely trying to comment on this environmental deterministic study.
Discussion is invigorating but it should be well founded. Once again, I point out that in no way do the authors propose that there were 48 conflicts above a background level.
Excuse me, I meant "baseline level" rather than "background level".
Reply | Report Abuse | Link to thisAny temperature taken before 1970 is likely suspect simply because measuring temperature in the easy to reach places of America and Europe and what is less than 1000th of a percent of the total atmosphere. At least when trying to make claims of human caused global warming. Prior to 1920 or so, we only a few measures from major cities again primarily in places easy to get too and fewer of them.
Reply | Report Abuse | Link to thisThen the sparse measures written earlier than 1900 are hardly accurate and this stuff was cherry picked to support the claim.
Plus creating a so called global average has no meaning, especially trying to take less than 1000th of a percent of the total atmosphere and calling that some sort of global average, then trying it with hardly any data from inaccurate tools in the preindustrial ages.
Then when you get to reasonable data like ice cores going back through ice ages, we find not only no global warming but CO2 levels or temperature are not even close to what it has been in the past.
Even Muller and Watts disagree with you. Let us know when you have published your own research.
Reply | Report Abuse | Link to this"But to the dismay of skeptic bloggers, his preliminary analysis supports that canonical view. "
http://news.sciencemag.org/scienceinsider/2011/04/qa-with-richard-muller-a-physicist.html
"...the overall mean temperature trends are nearly identical across site classifications."
http://scienceblogs.com/deltoid/2011/05/anthony_watts_contradicted_by.php
When you read the whole article you find that the number of countries in the groups affected by El Ninos is not constant across the time period, which is why they use a measurement they call annual conflict risk (ACR). It is independent of the variation of counts within groups either affected by or not much, if at all affected, by an ENSO event.
Reply | Report Abuse | Link to thisThey find that the baseline ACR is 3% and the ACR of an affected country during an NINO3 event is 6%. In the counts as the play out, the difference between what is modeled at 3% and the actual events is 48.2.
Your math assumes the groups are of a constant size and is invalid when they are not. So, we are back to you not comprehending what you are reading.
This bit : "Then, running a comparative simulation in which such El Niño weather patterns did not occur, the researchers determined that the hotter, drier conditions helped stoke 48 civil wars that did not occur in the modeled El Nino-free world," makes me wonder if any of their numbers are valid.
Reply | Report Abuse | Link to thisJust because you can write a computer program that comes up with numbers you like doesn't make those numbers meaningful, it just means you are clever enough to reverse engineer your conclusions and represent the process in a model.
Did this model include the personalities of every resident of every country and all of their interactions? Of course not! You can bark at me until your hair turns grey and your teeth fall out about how scientific the model is, but I'll still file this one under "Figures don't lie, but liars can figure!"
And people wonder why some folks don't believe in human contribution to climate change! It's tripe like this that hurts the credibility of science.
Right, like I said, they state that the putative ACR during El Nino events is about 3% above the baseline. Since the baseline value is placed at 3%, this totals 6%.
Reply | Report Abuse | Link to thisThis is why they propose that the risk of tropical conflicts "doubles" under the supposed influence of El Nino events, but actually the "doubling" is only a 3% increase, and is only perceptible via their extensively manipulated statistical analysis (including the brazen omission of unfavorable data sets).
In any case, they do not state that that the 3% increase in ACR CAUSED 48.2 conflicts, but rather that it AFFECTED those cases. They do say they find that 48.2 conflicts (21% of the 234 total) were associated with El Nino events (I refer to their concluding paragraph). Well, yes, this exactly what would be expected simply given the number of conflicts they consider (234) and the number of El Nino events in the time period in question (60 years). In my first comment I used the rough average figure of a El Nino event about every five years, and their chart tends to confirm this, showing about 12 or 13 El Nino events for the period. Sometimes number-crunching studies meant to be overarching are actually closely tied to event facts.
Sorry, I meant to write "evident facts" on the last line of my comment.
Reply | Report Abuse | Link to this@Kayrock -- I suspect you are arguing semantics. They do express themselves very poorly, but I suspect the data may be more supportive than the explanation they give.
Reply | Report Abuse | Link to thisCrappy data and even worse writing aside, do you agree or dispute that drought and the associated lack of food and income could increase the chance of civil war?
@SkepticalKen ---
Reply | Report Abuse | Link to thisI absolutely agree that drought can be a contributing factor in civil strife. Haile Salassie found that out in 1974, after a catastrophic drought had killed hundreds of thousands of Ethiopians.
But in the polemic study we're commenting on, the authors are not directly proposing that any of the El Nino events during the period in question caused devastating droughts, despite the fact that's what they insinuate in their introduction It's funny but true that they do not directly mention agriculture, crops, etc. in their article, but they do state at the onset that (quote) "Historians have argued that [El Nino events] may have driven global patterns of civil conflict in the distant past, a hypothesis that we extend to the modern era and test quantitatively." They give 3 references for this, all of which are to studies about catastrophic crop failures attributed to El Nino events in historic times (preceeding the past 60 years of their own study). So that's their hypothesis as stated, but they seem to forget this further on in their article, when they attribute quasi-metaphysical properties to El Nino's effects on conflict [quote]: "Precipitation, temperature, sunlight, humidity and ecological extremes can adversely influence both agrarian and non-agrarian economies. In addition, [El Nino event] variations affect natural disasters, such as tropical cyclones, and trigger disease outbreaks. All of these have adverse economic effects, such as loss of income or increasing food prices, and it is thought that economic shocks can generate civil conflict through a variety of pathways. Furthermore, altered environmental conditions stress the human psyche, sometimes leading to aggressive behaviour. We hypothesize that El Niño can simultaneously lead to any of these adverse economic and psychological effects, increasing the likelihood of conflict."
So a major question we should ask is whether droughts have caused crop failures (or at least poor yields) leading to any of the civil conflicts of the past 60 years in this study. Another funny thing: the authors do not bother to include this vital data about crop failure (or poor yields), which I'm sure can be easily found for all of regions of the conflicts in question.
I can venture opinions about the area with which I am familiar, Latin America, which "contributed" several of the conflicts included in the data set. To the best of my knowledge, drought has not been noted as a contributing factor in any of this region's conflicts over the past 60 years.
Qualifying what I mentioned about the authors' omission of crop failure (or at least poor yields), they do factor in grouped data on crop yield, but do not directly address the critical linkage of crop failure and actual cases of conflict which is the stated basis of their hypothesis.
Reply | Report Abuse | Link to thisYou still are not comprehending what you are reading.
Reply | Report Abuse | Link to this"We estimate the number of conflicts associated
with ENSO by assuming all conflicts in the weakly affected group were unaffected and a baseline ACR of 3% for the teleconnected group would have remained unchanged in the absence of ENSO variations. We then project the observed sequence of NINO3 realizations onto our linear conflict model (dACR/dNINO350.0081) and find 48.2 conflicts (21%) were associated with ENSO."
Yes, they did not claim 'caused'. So? Theirs is an observation of a correlation, not an explanation of cause. You are continuing to confuse all conflicts which occurred during NINO3 events, with those above the modeled baseline, as they state.
This article left a lot of gaps I don't much care for but the following dicussion has been fascinating. One point in support of Kayrock - Many of the conflicts listed owe far more to the break up of colonial powers and particularly in Central America, the direct empire building of the USA and USSR where both funded opposing armed groups.
Reply | Report Abuse | Link to this
Reply | Report Abuse | Link to this@Chris G: Checking back. In response to your latest comment, I do indeed comprehend the article, but I would note that you do not address my comments about the authors cooking the books by their egregiously deliberate omission of whole data sets unfavorable to their results. Yes, I think we all understand that the authors merely propose a 3% increase in conflicts correlated with El Nino events. And as previously mentioned, this is quite unconvincing given the tiny data sets, the deliberate rejection of unfavorable data, and a morass of assumptions which are encapsulated in their background data.
But upon reflection, the discourse in these posts has been quibbling over what is at best a very weak statistical correlation. Any way you cut it, a putative 3% spike in a tiny data set would be expected to fall within the noisy range of likely (high probability) error. Nevertheless, this unremarkable result is reported under the sensationalistic headline, “El Nino Found to Drive Tropical Wars”. Would any of us agree with that? Would you agree with that? And if not, why do you not criticize it - the public is being misled.
But actually, the colossal and fatal omission in this study about historic events is, ironically, of history itself. As bucketofsquid and other posters have pointed out, the lion's share of civil conflicts in the period in question have causes which have been clearly identified by historians, including the breakup of colonial empires, the cold war, and others which have been cited. What is most peculiar about this study is that it is completely blind to history; the authors seem to believe the only salient historic factoids from tropical civil conflicts are their date and number of fatalities (yes, I recognize that they also use economic data, but history is absent from their study). Once again, I would observe that this study seems to be an attempt to revive antique environmental determinism.
I am compelled to add that this omission of history is not really a surprising result because none of the authors have that background. Two are economists and one is a climate scientist. They would have done well to have gotten a couple of historians on board, but that would have been most inconvenient since it is very difficult to see how historians could have ever agreed with anything close to their ahistorical and deterministic conclusions. @Chris G: Checking back, and in response to your latest comment, I do indeed understand the article, but I would note that you do not address my comments about the authors cooking the books by their egregiously deliberate omission of whole data sets unfavorable to their results. Yes, I think we all understand that the authors merely propose a 3% increase in conflicts correlated with El Nino events. And as previously mentioned, this is quite unconvincing given the tiny data sets, the deliberate rejection of unfavorable data, and a morass of assumptions which are encapsulated in their background data.
But upon reflection, the discourse in these posts has been quibbling over what is at best a very weak statistical correlation. Any way you cut it, a putative 3% spike in a tiny data set would be expected to fall within the noisy range of likely (high probability) error. Nevertheless, this unremarkable result is reported under the sensationalistic headline, “El Nino Found to Drive Tropical Wars”. Would any of us agree with that?
But actually, the colossal and fatal omission in this study about historic events is, ironically, of history itself. As several posters have pointed out, the lion's share of civil conflicts in the period in question have causes which have been clearly identified by historians, including the breakup of colonial empires, the cold war, and others which have been cited. What is most peculiar about this study is that it is completely blind to history; the authors seem to believe the only salient historic factoids from civil conflicts are their date and number of fatalities (yes, I recognize that they also use economic data, but history is absent from their study). Once again, I would observe that this study is an attempt to revive antique environmental determinism.
I am compelled to add that this omission of history is not really a surprising result because none of the authors have that background. Two are economists and one is a climate scientist. They would have done well to have gotten a couple of historians on board, but that would have been most inconvenient since it is very difficult to see how historians could have agreed with anything close to their ahistorical and deterministic conclusions.
We might ask how this paper managed to get through peer review. A peer review process worth its salt would have strongly recommended giving at least a nod to history.
Sorry, the comment section seems a bit buggy and my previous post doubled up. If you are interested, please start reading from the 4th paragraph, beginning with "@Chris G."
Reply | Report Abuse | Link to this