A random thought on adaptation

There have been some interesting stories lately about how the new drought tolerant seeds are performing. Anecdotes of farmers marveling over how their crops fare are not exactly evidence, since there may be many more who were disappointed. But it does at least suggest that new seeds are better adapted. Similarly, one often hears stories about how agriculture is expanding into new areas, which is another way that agriculture could adapt to global warming.

I have little doubt that new varieties and migration of crop areas will help with climate change, but the key question is how much. Will it be a 1% or 50% type of effect? A lot of our research is about trying to find and analyze datasets that can answer this question. Typically any one dataset can only tell us so much, so it’s really about trying to piece together a picture from multiple different analyses. Not all of these analyses have to be very sophisticated. For example, a simple plot of average country yields vs. average growing season temperature is shown below (I made this based on the methods and data in this paper from last year. The figure is part of a review that is coming out in Plant Physiology later this year.)

The green blobs each represent a country, with the size proportional to total production. The vertical gray line shows an independent estimate of the optimal season temperature for yields, based on a recent review by Hatfield et al. that was based on experimental studies. It’s not exactly a pretty figure – lots of factors differ between countries other than temperature, which results in a lot of scatter. But I think it illustrates three important points that are sometimes missed:

1. The highest yielding countries are fairly strongly clustered around the gray line, except for barley where they are significantly cooler. (This is a surprisingly good match given that the gray lines were completely independent of this dataset.) Although many countries grow each crop above its optimum, the maximum yields are clearly lower at high temperatures. This casts some doubt on the notion that yields can be maintained as temperatures rise, since the warmer countries should already have incentive to better adapt to their conditions.

2. Large producers span a pretty wide range of temperatures. This is sometimes cited as evidence that agriculture is well adapted to a wide range of climates, but I think it’s more accurate to say that farming is profitable across a wide range of climates. For example, people sometimes like to point out that if we grow corn in Alabama, how can global warming be a concern? The answer is that we grow corn in Alabama, but not nearly as well as if it had the climate of Illinois. The lack of a tight relationship between yields and crop areas indicates that agriculture is not greatly optimized to current climate. To me, this casts doubt on any argument that migration of agriculture will be a major source of adaptation. There are clearly a lot of factors other than climate that enter into a decision about which crop to grow.

3. For maize and wheat, a lot of the bigger producers tend to fall to the right of the optimum. This helps to illustrate why the global production of these crops are typically predicted to be hurt by warming, even if some countries gain.

None of these points are proven by the figure. It’s pretty rare that a simple cross-section like this can prove anything. But sometimes simple plots can really challenge a strong prior belief. If it was common to match crops to their temperature optimum – either by relocating where the crops grow or by changing the crop’s optimum temperature – then I would expect to see a much flatter cross-sectional relationships between yields and temperature, or a much tighter concentration of area around the optimum.