Food Security: The Challenge of Feeding 9 Billion People

The New York Times Dot Earth blog last night pointed out that Science magazine (one of the two premier publication forums for academic science) has removed the paywall from a review by Godfray et al (a set of British biologists and development studies specialists) titled Food Security: The Challenge of Feeding 9 Billion People. This is a subject near to my heart, and I somehow missed it when it first came out online last month (I subscribe to Science but am inconsistent about reading it when it comes out - I mainly have a subscription so I can always go and search the archives and read any Science papers on some subject that I'm pursuing).  So I enthusiastically headed over to read it. It turns out the current issue of Science is devoted to food security, so I'll probably be blogging about it a lot.

The point of a scientific review is to provide a survey of some field of inquiry, with references to the important papers.  The best reviews provide the reader with a real synthesis of what the state of knowledge is currently, as well as an indication of where the frontiers of research currently lie - what things are most in need of further study (but where there is some hope of actually making progress).  The target reader of reviews is generally considered to be non-specialist scientists who need to understand the field, or graduate students just coming up to speed.  Actively working specialists in the field being reviewed are assumed to be up to speed on the literature and to already have their own perspective (though it can certainly be helpful to understand how someone else views the big picture).
As a non-specialist scientist, who has nonetheless researched these issues quite a bit, I presumably fall in the target audience, but I wasn't completely happy with this review. I agree with most of what it says, but feel it misses some important elements of the situation because the authors don't have a large enough perspective on the forces acting on the food system from outside.

The historical situation from the early 1980s to about 2005 is that the productivity of the industrial food system (yields of food per acre) has been increasing faster than the demand for food, and thus real food prices have been falling, farmers have been under tremendous pressure, and small farms have been going out of business and consolidating into large farms where the economics work better because of economies of scale.  Whole regions that were formerly agricultural have been largely withdrawn from production because they are not cost competitive (eg much of New England).  Meanwhile, consumers have had ever cheaper food.  Thus feeding people retreated into the background as a major global concern and 1970s era warnings of mass starvation (eg Paul Ehrlich warning that a billion people might starve in the 1980s) came to look rather silly.  Although many very poor people continue to not have enough to eat, the fraction of the human population that is undernourished has been declining, and the problem has lain not in lack of sufficient food but rather that a fraction of people cannot pay for their share. (See here for some background on these issues.)

There are several reasons to think this situation might change, which the authors very briefly cover:

1. Population continues to increase (it's generally thought that the human population will peak somewhere around 9 billion mid century), and if yields did not continue to increase, we might start to have worsening hunger problems.
2. As developing countries get wealthier, they are likely to consume a higher fraction of their diet in the form of meat and dairy products, and these are converted from grain with only about 10% efficiency, so this causes the total demand for plant foods to rise faster than population.  The graph above is excerpted from the article and shows the increases in total production of various commodities, and as you can see the increase in production of animal products has been very strong.
3. We are converting an increasing fraction of the food supply to biofuels.
4. Climate change could reduce agricultural yields.
   

(I'm not personally very persuaded on point 4 in a global sense, though clearly there could be very significant regional problems. However, a number of scientists are concerned about this as a global issue - see here for a discussion of the issue and references).  The 2005-2008 oil price shock provided a warning that the regime of the last few decades could be changing:

In mid-2008, there was an unexpected rapid rise in food prices, the cause of which is still being debated, that subsided when the world economy went into recession (11). However, many (but not all) commentators have predicted that this spike heralds a period of rising and more volatile food prices driven primarily by increased demand from rapidly developing countries, as well as by competition for resources from first-generation biofuels production (12). Increased food prices will stimulate greater investment in food production, but the critical importance of food to human well-being and also to social and political stability makes it likely that governments and other organizations will want to encourage food production beyond that driven by simple market mechanisms.

For my taste, the introductory problem statement was too brief and lacked any quantification. For example, there was no discussion of the elasticities of food demand with income, which have been quite well studied (see here for example), and allow one, given a GDP growth estimate, to roughly quantify how much demand will increase in the future as a result of the rising-income-to-more-meat-eating channel. Similarly, there was no attempt to quantify what the factors leading to biofuel demand are, and in general the article underemphasizes this issue. To me, the fact that the global fuel system is much larger than the global food system in energetic terms means that there are considerable dangers in creating price arbitrages between them, and also means that this could easily become the largest demand side factor in the evolution of the food system. This is not discussed at all. I suspect the authors don't know very much about the pressures on global oil supply, and this limits their understanding of the potential for problems from that quarter. They then go onto discuss four possible responses to increase output of the global food system in response to these various pressures. The first is yield gaps. In a system that has been facing falling prices and an excess of supply over demand for several decades, there are many areas of slack or underinvestment that could be taken up to increase production:

There is wide geographic variation in crop and livestock productivity, even across regions that experience similar climates. The difference between realized productivity and the best that can be achieved using current genetic material and available technologies and management is termed the "yield gap." The best yields that can be obtained locally depend on the capacity of farmers to access and use, among other things, seeds, water, nutrients, pest management, soils, biodiversity, and knowledge. It has been estimated that in those parts of Southeast Asia where irrigation is available, average maximum climate-adjusted rice yields are 8.5 metric tons per hectare, yet the average actually achieved yields are 60% of this figure (19). Similar yield gaps are found in rain-fed wheat in central Asia and rain-fed cereals in Argentina and Brazil. Another way to illustrate the yield gap is to compare changes in per capita food production over the past 50 years. In Asia, this amount has increased approximately twofold (in China, by a factor of nearly 3.5), and in Latin America, it has increased 1.6-fold; in Africa, per capita production fell back from the mid-1970s and has only just reached the same level as in 1961 (2, 20). Substantially more food, as well as the income to purchase food, could be produced with current crops and livestock if methods were found to close the yield gaps.

The discussion of various examples of yield gaps in the paper seems interesting and useful. However, the first and the most obvious question to me is: how big do we think the yield gap is currently on a global basis (with error bar please)? And the next is: what is the trend in the yield gap over time? A scattering of examples is much less satisfactory than some kind of attempt to integrate the state of knowledge into a big picture quantitatively. Again, there is literature on this topic that could have been discussed.  And either the answer is known, in which case they should have discussed it, or it's not, in which case surely it should be a major focus for research?

The next area of discussion is the potential for further technological developments improving yields (which are likely to come mainly from genetic manipulations of organisms). Obviously, predicting how technology will evolve is always difficult and speculative, but from my limited state of knowledge, I didn't have much quarrel with their discussion. It generally seems like there are lots of promising ideas for further yield increase, and no obvious reason why the twentieth century trend of yield increases won't continue for a while at leat.

The third area is the discussion of waste of food, which I only had a very superficial knowledge of and so found quite informative. "Roughly 30 to 40% of food in both the developed and developing worlds is lost to waste, though the causes behind this are very different." Quantification much appreciated! The interesting point about the differences between developed and developing countries is summarized in this graph:

Basically, in the developing world food is mainly lost due to inefficiency in the technology for storage and transport, whereas in the developed world the main issue is that both stores and consumers waste a lot of it (which I think mainly speaks to the fact that food has been very cheap). Clearly there is significant potential for improvement here if food prices rise high enough to motivate behavioral changes.

Finally, the article has a brief discussion of the possibility to expand fish farming, and some of the potential issues with that. It's short but fair enough - Science papers are very restricted in overall size, and so some things have be brief. There's a reference to another paper in the special issue that is devoted to this subject, so I'll probably write about that another day.

Overall, I think the article - with its severe neglect of the biofuel issue - illustrates the problem with trying to understand the world from within the confines of particular specializations - the most important factors are very often in the interactions between different parts of the system, and thus outside one's area of knowledge. On the other hand, it's also incredibly difficult and dangerous to be a generalist - so easy to make a total fool of oneself by commenting on things that one has not studied enough to really understand. So humanity will continue to stumble into the future half-blind, no doubt.