Growing GMO maize is no game

A couple of days ago I wrote a bit about how confused I am by the arguments around low agricultural prices and high food prices. I cited, as I often do under these circumstances, the Shakesperean observation that what farmers decide to do depends crucially on what their neighbour farmers have decided.

Biologists call that density-dependent selection, and it's a well-known version of the Prisoner's Dilemma game. A new paper in PLOS Computational Biology by Alice Milne and her colleagues shows that it applies beautifully to pest management, at least in a model of how maize farmers in the US might respond to the threat of European corn borer (Ostrinia nubilalis). ¹ It's a great story.

The most effective defence against European corn borer is one of the genetically engineered Bt varieties that kill caterpillars, but GM seed carries a hefty price tag. If there's going to be an outbreak of the pest, then the investment is worthwhile, but outbreaks are unpredictable. So each season, farmers need to decide whether to pay the price or risk losing some of their harvest. What makes it interesting is that the best strategy depends to some extent on what everybody else is doing. If everybody else is investing in GMO maize, then even if there is an outbreak, their pest control will reduce the threat to your crops. And there's good evidence that many farmers do indeed hitch a free ride.

Hutchison et al. showed that Bt maize generated an estimated $230 million annual benefit to maize growers in Illinois, Minnesota and Wisconsin. Much of this economic benefit (75%) accrued to farmers who did not plant Bt maize; these farmers did not pay technology fees but still benefitted from the area-wide suppression provided by those farmers who cooperated to use Bt to reduce pest densities.

How, then, do farmers decide whether to pay extra for Bt maize? They talk to seed dealers and crop consultants, who have considerable influence, their neighbours, who have less influence, and extension agents, who have little influence. And of course what matters most is not the actual risk but the farmers' perceived risk. Milne and her colleagues first built a model of the corn borer's population biology and then married it to a sociological model of each farmer's decision and the forces that determine it.

The crucial result -- and there's a lot more to chew over in the full paper -- is that

[F]armers in a region are often influenced by similar circumstances, and this can create a coordinated response to a pest. This coordinated response, although not intentional, can affect ecological systems at the landscape scale.

That's what makes it worthwhile for some farmers to decide not to invest in Bt maize, which is a good thing as it tends to limit the opportunities for the pest to evolve resistance. Importantly, although farmers in the model may consult their immediate neighbours, they do not come to collective decisions. As a result, the model has nothing to say about how to arrive at a more equitable division of the costs -- both of using Bt maize and of losing potential yield by not doing so. It does, however, point the way.

By understanding the dynamics of farmer decisions we can determine how to manage better the system, through improved communication, subsidy or taxation, to achieve robust and cost effective area-wide control, while minimizing the risk of the evolution of resistance to control strategies.