11 January 2009

Climate change and the food supply

ResearchBlogging.orgOne of the most-cited effects of global warming is that of rising temperatures on crops - hotter average conditions should lead to warmer, drier conditions, reducing yields in the best growing areas and maybe eliminating them where conditions today are marginal. In this week's Science, a new study puts some numbers behind that speculation [$-a], and the news is not good.

Photo by Josh Sommmers.
Assembling the results of 23 climate models, authors Battisti and Naylor compare projected temperature ranges for the coming century with the ranges observed in the previous one. By the final decade of the twenty-first century, they say, summertime high temperatures in most of the continental U.S. have a 70% probability of exceeding the hottest summer temperatures ever recorded; in Saharan Africa, much of the Middle East and central Asia, the probability is 90-100%.

To put these numbers into perspective, Battisti and Naylor go to the history books, citing an array of cases in which local high temperatures have disrupted food production, creating regional shortages that eventually impacted worldwide food markets:
By comparison, extremely high summer-averaged temperature in the former Soviet Union (USSR) in 1972 contributed to disruptions in world cereal markets and food security that remain a legacy in the minds of food policy analysts to this day. ... Nominal prices for wheat -- the crop most affected by the USSR weather shock -- rose from $60 to $208 per metric ton in international markets between the first quarters of 1972 and 1974.
Battisti and Naylor end by calling for substantial investment in adaptation measures to prevent "a perpetual food crisis." Increasingly, this looks like the only practical course of action - although reducing and eliminating man-made greenhouse gas emissions is critical, turning global climate around is going to be like steering an aircraft carrier, and it's going to get pretty warm before we turn the corner.


D.S. Battisti, R.L. Naylor (2009). Historical warnings of future food insecurity with unprecedented seasonal heat Science, 323 (5911), 240-4 DOI: 10.1126/science.1164363


  1. "it's going to get pretty warm before we turn the corner."

    We've been cooling for 10 years. Which corner are you referring to?

  2. Cooling for 10 years? Got a citation for that?

  3. Unfortunately I can't access the full article but what you've posted up here makes no sense. There is a confusion between weather and climate.

    The 1972 event was an unpredictable weather event. Clearly, farmers choose their crops once a year, and are stuck with that choice. When an unexpected, unpredictable, regional, 3-month weather pattern occurs which is not compatible with the crop choice, the crop will fail.

    If climate is predictable on larger scales, a warming will be known in advance and the farmers can select crop appropriate to the modified climate. In fact, this happens almost automatically; for example, in the UK, during the medieval warm period, grapes were grown commercially as far north as Scotland. During the little ice age, no grapes were grown commercially in the UK. And now the climate has warmed again, commercial grape growing has returned. This doesn't require predictive models; due to the timescales of climate and the growing cycle, farmers do it automatically.

    Farmers adapt their crops to those suitable to the current climate. They cannot adapt to unexpected weather conditions. It is not valid to read results from the latter scenario to the former.

    As I cannot access the article, I cannot comment on whether the original article makes the mistake, but this blog post appears to.

  4. Anonymous (same one?) - it's absolutely true that weather is not climate, and that one unusually hot year is not a climactic trend. Climate consists of the range of observed and expected conditions, and this is what Battisti and Naylor are dealing with - not the question of whether it will be above freezing in Moscow, Idaho, on January 12, 2095, but the probability that temperatures in Moscow (or anywhere else, for that matter) will be unusually hot. So what they're predicting here is that the coming century will have a higher frequency of years that exceed past observed temperature ranges worldwide. I'm sorry if I hadn't made that distinction clear originally.

    It's also true that, left to their own, farmers will "adapt," I suppose. But the changes projected for the near-ish future are probably going to be faster and greater in magnitude than past climactic shifts (as with the period you cite) - and how many Scottish vitners lost their crops, and livelihoods, before they stopped trying? Predictive modeling of the sort discussed above can make adaptation quicker, with fewer catastrophic crop failures in the process.

  5. Hmm, tried posting this yesterday, didn't work for some reason. Probably my end, the browser didn't respond as I expected. I'll try again.

    I posted the second anonymous post (Jan 12 3:19) but not the first one (Jan 11 7:01). I have a high Laziness Quotient (LQ) and hitting the anonymous button was easier than trying to work out the other options.

    I found a freely available online copy of the paper here. Having read the paper itself, I am even less impressed. The article makes the exact same mistake of directly comparing a year of unexpected weather and arguing that will have the same affect as several similar years of expected climate.

    I would have argued that the dominant factor would be the regional standard deviation of weather patterns. This would be more difficult to mitigate against (although, in practice, even this can be mitigated against to some extent). You make a valid point that there is most likely a rate of change beyond which farmers may struggle to adapt. These are all, to my mind, important questions which should have been addressed in the paper as they will be far more important than the issues that were discussed.

    Several problems here of course, in that climate models have little or no credibility for regional predictions (e.g. ref 1 below). That calls into question the authors use of regional data from models as well as the predictability of such changes. Work is ongoing to address these problems (e.g. see J12 here), and perhaps there is justification in increasing these efforts.

    The paper also shows evidence of cherry picking and ignoring counter view points. The one that really stuck out was the assertion that France and Northern Italy suffered some 30,000 heat-related deaths in summer 2003. Firstly, I have seen several estimates for excess mortality (which is not the same as heat-related deaths), the highest I have seen up to now was considerably lower than that; some estimates a lot lower. That figure seems to be cherry picked. It would have been better to discuss a range of estimates to give a feel for confidence intervals.

    Furthermore, most of these deaths were in France in August 2003. This was primarily due to a failure of the way in which the French health service was run. August 2003 was not the hottest month recorded in France; July 2006 was hotter. In July 2006, despite being hotter, very few deaths were attributed to heat. This is the product of two things; firstly, the vacation problem in the French healthcare system was more important than the temperature problem; secondly, it is a clear illustration of the ability of humans to adapt - rapidly - to changing conditions.

    The failure to address these points really limit the value of the paper (IMHO).

    ref 1. Koutsoyiannis, D., A. Efstratiadis, N. Mamassis, and A. Christofides, On the credibility of climate predictions, Hydrological Sciences Journal, 53 (4), 671–684, 2008. Link