Monday, April 11, 2022

Why Global Temperature Doesn’t Matter for Global Crop Yield

Looking at a little of the literature on the effect of climate change on agriculture, I noticed something that seems to be a mistake — perhaps someone here can explain why it isn’t.

Crop yields depend, among other things, on temperature, with an optimal average temperature for each crop — about 15°C for wheat, for example (Lobell 2012). If temperature goes up by a degree, yield in an area that used to be 15° and is now 16° goes down a little. This seems to be one of the effects that goes into estimates of reduced yield as a result of climate change.

But it shouldn’t. The same warming that shifts 15° up to 16° also, somewhere a little farther north in the northern hemisphere or south in the southern, warms 14° to 15°, 13° to 14°, and so on. If wheat was being grown between, say, 13° and 17°, the area of cultivation can shift by one degree towards the pole and continue to have a temperature range of 13°-17° and the same temperature-related yield as before.

I can see two possible objections to this argument. The first is that the land a little closer to the pole may be less well suited to growing wheat in respects other than temperature. That is obviously possible but why would you expect it? Is there any reason why land that happens to have the ideal temperature for growing wheat is also more likely than other land to have the ideal soil or the ideal amount of rain? If not, then on average the shift is to land about as well suited in other ways and now ideally suited in temperature. A more careful analysis might find a deviation from that in either direction, land a little closer to the poles a little better or a little worse, but why should we expect either?

The second objection is that shifting the area of cultivation is costly — wheat farms have irrigation systems suitable for growing wheat, appropriate farm machinery, are owned or managed by people experienced in growing wheat. You can’t just pick all that up and shift it a hundred miles further north.

How serious an issue this is depends in part on how fast the shift happens. Looking at maps showing average temperature, it seems to go down as you move towards the poles by about a degree every hundred miles, with a good deal of variation. At current rates of climate change, global temperature should be going up by about a degree every thirty years. So shifting the area of cultivation to keep the temperature at which wheat is being grown constant should require moving it by about three miles a year, with farms at the warm edge of the zone shifting to crops with a higher optimal temperature such as maize (18°) while farms at the cold edge are shifting from barley or vegetables to wheat.

The real pattern would, of course, be more complicated than this, but why isn’t it the right first approximation? If so, then reduced yield with warming should not be included in the effect of climate change on agriculture. What should be included is the large increase in total arable land as temperature contours shift towards the pole, since it is cold, not heat, that restricts the area of land suitable for agriculture.

Am I missing something? Alternatively, is all of this already being included in models of the effect of climate change on agricultural output? If so, perhaps someone can point me at examples.


Anonymous said...

Actually, land closer to the equator should be better for growing crops because it gets more light.

back40 said...

The main determinant may be rainfall rather than temperature, though they are related. Patterns will change in less than straight forward ways.

Markets also matter in crop choice. Each year the acres planted of each crop are a matter of profit potential. If production costs are too high or demand is too low for good prices then it might not be possible to produce a crop that sells for more than it cost to produce.

This year drought is a main concern rather than temperature. Crop inputs are also very expensive, so it is a magnified risk to even plant a crop that might drought kill.

It's hard to take crop yield predictions seriously. If it was possible a lot of speculative profits could be made.

SB said...

On your first point, there's an evolutionary argument that since each crop originally evolved in a particular place, it's most likely best adapted to the combination of rainfall, soil, temperature, and light in that place. A different place that now has the same temperature as the crop's original home averaged during its evolution as a species is more likely to be worse than better (from the crop's perspective) in other ways. I don't know how strong an effect that is.

And as Anonymous points out, sunlight makes places closer to the equator objectively better for most plants than places closer to the poles. I suspect that's a stronger effect than the one in my first paragraph above, but I don't know for certain.

The question of what it costs to shift crops or move in response to climate change, and how quickly it can be done, might be substantially different depending on whether mountain ranges, deserts, and other uncultivable places are arranged predominantly east-west (as in Europe and Africa) or north-south (as in the Americas). The sort of gradual 3-miles-per-year change you describe is probably easier (both for wild species and for human farmers) in the latter than in the former.

Anonymous said...

Does the suitability deteriorate at the same rate going hotter or colder? Because if there are significantly different slopes (say, +10 degrees from ideal has the same yield as -5 degrees) then that'll imply dramatically different globally ideal temperatures.

Simon said...

The light aspect may not be as important as it might seem at first, since while the sun is at a lower angle further north, the days there are also longer in the summer (i.e. in the local growing season) than places closer to the equator.

Regarding rainfall, this will probably increase on average with global warming, since what goes up (in terms of evaporation) must come down. Rain will probably also be more variable, but this should be mitigatable via reservoirs and irrigation.

Most plants are grown now well outside of where they were first domesticated, and in any case they would likely have been originally adapted to ice age climates (which are much longer than interglacial periods typically) making it probably not correct that they would originally be adapted to some currently existing combination of climate factors. Modern breeding might make them currently more adapted in this way, but this should be fixable by further breeding.

On the other hand, I have the impression that high air temperatures are non-optimal for even relatively hot-adapted plants. So, temperature increases may reduce tropical agricultural productivity even though right now even the equator does support plant life.

Also, even if a movement of agricultural land is physically possible, it may be prevented by political factors (including for example attempts to preserve the existing ecosystems in places that are currently wilderness).

chofland said...

A new job has inspired me to expand my agronomic knowledge from zero to... very little. But let me make some naïve points.

Land use change is a big problem with greenhouse gas emissions. Cutting down trees releases a lot of carbon. Tilling previously untilled land releases a lot of carbon. If you're talking about moving to "virgin" land for agriculture, this would accelerate the process. Not true if you're shifting around where crops are grown, but this sounds like a game of musical chairs where something has to get left out eventually.

Crop breeding is huge business and does amazing things. I think it would be very doable to make crops tolerant the equivalent of 3 miles south every year -- I mean, until we hit Venus-like temperatures.

Farmers rotate crops, so this is very much a marginal process. They would grow less of "cool" crops in the rotation, and more warmth-tolerant crops; over time the crop mix would just migrate north, without farmers packing up their equipment and moving north -- if that's what anyone is envisioning.

I am curious why higher temperatures (within reasonable ranges) would not INCREASE yields. Crops require a certain number of "degree growing days". All else equal, warmer days make for shorter seasons. I guess that's not higher yield, but rather the same yield in less time. But you should be able to leverage that by getting in a second crop or something.

When the temperature goes up above the band of ideal temperature, the plant doesn't just fry. There is a band where the additional degree days just don't contribute any more. Then above that you hit a point where the heat is damaging. Those ranges are usually like 5 or 10 degrees (Fahrenheit!), I think, so a few degrees average change isn't likely to push the crops over the limit.

If scientists say yields are going to go down when temps go up, I believe them. But there has to be some chain of effects. Plants are energy storage devices, so the first order effect should be that increased temps fill them up with energy faster.

Matteo Pastrello said...

It should also be noted that not only the average temperature matters for crop yields, but also the standard deviation from that average (usually increased by climate change). If for example the medium temperature doesn't change from 20°C but many more days in a year you have extreme temperatures like 35°C and -5°C, the productivity will decrease.

Simon said...

> the standard deviation from that average (usually increased by climate change)

Actually, while global warming increases the deviation in rainfall, it generally reduces the deviation in temperature, both theoretically and in the current observed data.

Matteo Pastrello said...

> it generally reduces the deviation in temperature, both theoretically and in the current observed data.
I'm not a climate expert so that assumption that I've made was based on the popular view and understanding that other friends of mine involved in the field gave me, and since I never heard any argument of the contrary I assumed it was true. If you have any reference on that point I would love to read them, thank you.

David Friedman said...

Expanding on Simon's point...

Greenhouse gas warming is generally greater in cold times and places than in hot. The arctic, according to the IPCC projections, warms about three degrees for every degree of global warming. So winters become milder faster than summers become hotter, reducing the deviation in temperature.

That is also the reason that I suspect climate change reduces temperature-related mortality, contrary to what most of the talk claims. Deaths from cold are much more common than from heat at present, cold temperatures in cold places will increase more than hot temperatures in hot places, so one would expect the reduction in cold-related mortality to be greater than the increase in heat-related.

David Friedman said...


The current IPCC report is webbed, although finding things in it is a pain since it is very long. It has a set of graphs showing increase in minimum temperature per degree of warming in various regions, so you can see that in relatively cool area it is two to three degrees per degree. It has a table showing the effect on high temperatures in hot areas, and it's significantly less.

One of the repeated claims is that climate change makes extreme weather events more common. That ought to make you suspicious because it isn't clear what it means. How do you sum the increase in unusually high temperatures and the decrease in unusually cold, the increase in high end tropical cyclones and the decrease in low end ones, which are still extreme events? There isn't a natural metric, so whether the conclusion is a net increase or decrease depends on how you choose to define an "extreme" in each case.

Matteo Pastrello said...

Thank you very much for the counter-arguments, I will look into them in detail. Even if it my practice to always allow every claim that an "opponent" make in a discussion for every field that I'm not an expert in, it is nice to have something to add to my arguments.

Simon said...

David makes some good points on the practical effect and pole-equator variability reduction.

I had thought that there was a stronger case for the actual variability in any one spot (not just the effect of the variability or variability between locations) having lowered, but it doesn't seem so clearcut on review now. Here's an article with mixed results:

In theory, temperature variability depends on the net rate at which energy enters or leaves the system during times of heating or cooling (divided by heat capacity). Both heating and cooling are slowed down by clouds, and warming should lead to more evaporation and thus more clouds, so should lead to less variability outside of desert areas. Very hot temperatures tend to lead to thunderstorms which provides a particularly strong negative feedback in hot (as long as wet) areas.

Anonymous said...

The temperature zone that is conducive for wheat farming may indeed move north, but soil composition at those northern latitudes may be not be as hospitable for the growing of wheat as soil composition at more southern latitudes.

Furthermore, the profile of extant plant pathogens may not be equivalent at different latitudes.

These factors, and others, must be taken into account to project agricultural yield.


David Friedman said...

The soil composition might be less hospitable, it might be more. My point is not that we know yield would stay the same but that we have no reason to expect it to decrease (from warming alone — other elements of climate change might increase or decrease it). It might decrease, it might increase.

"If scientists say yields are going to go down when temps go up, I believe them."
I don't know all the reasons that some scientists say that, but at least one reason is a mistake. The simplest way of calculating the effect of temperature on yield is by looking at yields in the same place in different years, some of which are warmer than average. But that is measuring the effect on yield of a temperature change without allowing for changes made by the farmer to deal with it, since by the time the farmer knows that this will be a warm year most of the relevant decisions have already been made. If average temperatures go gradually up, farmers will adjust what crop varieties they plant, how they irrigate them, and lots of other decisions accordingly.

So that calculation, at least, will overstate the decrease in yield, perhaps even convert an increase into a decrease. I don't know enough about the subject to know if there are other reasons to think yield typically decreases with warming.

This is, unfortunately, an area where a lot of people know what conclusion they want to reach, and act accordingly. Some of what I have been writing on the subject is in chapter drafts webbed at