« Can a selfish gene stop malaria? | Main | Evolutionary sound bites »

Darwinian agriculture I

Next week, I'm speaking at a one-day symposium on "Darwinian Agriculture: the evolutionary ecology of agricultural symbiosis", in Wageningen, The Netherlands. So, instead of reviewing a recent paper, this week I'm going to discuss some of the not-quite-so-recent papers on which my talk will be based. The following week, I plan to summarize some of the talks I hear at the meeting.

I may do the same thing in August, when my grad students and I speak at the much larger Ecological Society of America meetings in San Jose, California. Feel free to comment if you feel cheated of your weekly paper review, and I might reconsider. The Evolution meetings are in Christchurch, New Zealand, this year, but my grant won't stretch that far.

"Darwinian Agriculture: when can humans find solutions beyond the reach of natural selection?" was the title of a paper that Toby Kiers, Stuart West, and I published in 2003. Our answers to the title question suggested how increased understanding of past and ongoing evolution could improve: 1) breeding of crops and livestock, and 2) design of agricultural ecosystems.

With respect to genetic improvement of crop plants, we wrote:

"most simple, tradeoff-free options to increase competitiveness (e.g., increased gene expression, or minor modifications of existing plant genes) have already been tested by natural selection. Further genetic improvement of crop yield potential over the next decade will mainly involve tradeoffs, either between fitness in past versus present environments, or between individual competitiveness and the collective performance of plant communities."

Since then, every time I give a talk on this subject, I look for papers that might disprove this tradeoff hypothesis. I also look for examples of tradeoffs that were rejected by natural selection, but which might be acceptable in agriculture. For example, many people are working on improving drought tolerance of crops. Is it possible to improve on natural selection for this trait?

In 2004, Capell et al. reported that "Modulation of the polyamine biosynthetic pathway in transgenic rice confers tolerance to drought stress" (Proc. Nat. Acad. Sci. 101: 9909). Sounds good, but what about the data? They reported "drought-induced rolling of leaves" in control plants, but not in their transgenic plants. Leaf rolling reduces solar heat gain by leaves and may help plants survive until it rains, so it's not clear that losing this trait is an improvement. The transgenic plants apparently "had a slightly decreased growth rate and flowered 4-5 days later" although there was enough variability that they couldn't be 95% sure of this. How did the combination of reduced leaf rolling, slower growth, and later flowering affect grain yield under drought in the field? What about yields in wetter years? Apparently they didn't measure yield, even in a greenhouse. I predict that, under field conditions, the transgenic plants will have lower yield than their nontransgenic parents, at least in normal years. (Otherwise, plants with a modified polyamine pathway trait would already have displaced "normal" competitors.) I don't want to guess, either way, about yields under drought.

In 2005, Park et al. reported (Proc. Nat. Acad. Sci. 102:18830) that transgenic tomato plants, which were genetically engineered to over-express an ion pump gene, grew more roots. This let them take up more water (from soil in pots in a greenhouse), so that they wilted less than nontransgenic tomatoes when the pots weren't watered for 5 days. OK, but is there likely to be any water in the soil during drought, under field conditions, at the depths where the extra roots form? And what about the carbon cost of making more roots? Would investing that carbon in more leaves (which bring in still more carbon, through photosynthesis), rather than more root, perhaps contribute more to tomato fruit yield? The answer to this question might depend on whether the tomatoes are irrigated and, if not, whether the year is wet or dry. I would be surprised if these transgenic tomatoes turn out to yield more than existing cultivars, especially with irrigation or normal rainfall.

So far, I don't see any reason to believe that further increases in the ability of individual plants to survive drought (a trait that natural selection has been working on for millions of years) will lead to much improvement in crop yield over a useful range of field conditions. But are there drought-related traits that would have been eliminated by past natural selection, but which would nonetheless be useful in agriculture?

Maybe. Mathematical modeling (Field Crops Research 61:179) suggests that natural selection results in plants that make too much root for their own (collective) good. From a farmer's point of view, crops should make just enough root to take up the available water, because any resources used to make extra root are resources not available to make tomatoes or beans. But natural selection has favored aggressive rooting, in which each plant "steals" as much water as possible from neighboring plants.

How could we modify this behavior? We could simply breed for less root, not more, but maybe we can do better than that. Surprising recent research suggests that plant roots respond to nearby roots in a way that discriminates between a plant's own roots and those of a neighbor (Proc. Natl. Acad. Sci. 101:3863). The resulting below-ground aggression appears to have a significant cost in reduced seed production (Plant Ecol. 160:235). Modifying root-root interaction genes so that plants "love their neighbors as themselves" might therefore optimize root growth, at the level of the whole crop, over a wide range of conditions. [Update: this might not be the best approach. See Darwinian Agriculture II.]

In effect, this would be selecting for group-level benefits, reversing the effects of past natural selection for individual competitiveness. A similar approach has already been successful in breeding laying hens. Selecting which hens to use for breeding, based on the egg production of a group of hens rather than individuals, led to less aggression among hens and greater collective egg production (Poultry Sci. 75:294).

I guess I will leave an evolutionary perspective on the design of agricultural ecosystems for another post. Any curious and impatient readers can look up our 2003 paper.


I appreciate this post and those you suggest may follow. Out in the field in the evolution/creationism battles, as many good examples of practical applications of evolutionary principles as we can find are very useful. I myself can't imagine why any farmer could be a creationist, since every plant and animal he grows is the production of evolutionary processes, be it via natural selection or artificial selection.

I greatly enjoyed reading your 2003 paper. I was reminded of Lerner's explanation for the phenomenon of genetic homeostasis: man's intense selection for traits that were not highly selected for (or may even have been strongly selected against) in nature had the potential of disrupting coadapted gene complexes that had been assembled over long periods of time in the wild. Disrupting those complexes could actually reduce the fitness of the organisms being bred.

I was particularly interested in your discussion of what we can learn from natural ecosystems. It is fashionable to assume that the natural ecosystem--particular community structures, for example--is somehow 'best', and that we should mimic it. I agree with you we can learn quite a bit from the study of natural communities, and then use that knowledge to model new agricultural ecosystems which can take advantage of what we have learned instead of slavishly duplicating nature. We have to accept that what we are looking for in crops is not necessarily what those particular organisms were looking for to survive in the wild.

Some evolutionary changes are fast enough for farmers to notice -- resistance to some herbicides took only a few years in weeds of rice -- but the process may be described as "herbicide effectiveness broke down", as if it were a change in the chemical rather than the weed population.

Regarding coevolved gene complexes, one thing I've been noticing is how much more complex naturally evolved control systems (plant responses to pathogens, for example) are, relative to those designed by humans trying to improve crops. I'm pretty sure that turning on a plant's defenses all the time is too simple, but are the natural systems ever too complex? That is, has natural selection failed to find simpler solutions that would work better, but which can't easily evolve in a series of steps?


That's a great question. I think the answer is yes, considering how contingent and short-sighted natural selection has to be. I imagine the biggest problem in designing new solutions will be untangling, bypassing, or turning off the original process without producing a cascade of additional required fixes because of interdependency.

Post a comment

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)

Type the characters you see in the picture above.