R. Ford Denison’s Darwinian Agriculture

I am a fan of Ford Denison.  It was his lab that reconciled the fact of the legume-rhizobium symbiosis with the evolutionary theory predicting that such a mutualistic relationship should fall apart because of cheaters.  When I saw he had a book out on agriculture I snapped it up.

Only Denison could have written Darwinian Agriculture.  As both an evolutionary biologist and a past director of UC Davis’ LTRAS, the Long-Term Research on Agricultural Systems project, he is well versed in the two fields and can evaluate agriculture trends using an analysis that is rare among agricultural scientists.  For instance, the Green Revolution was initiated by grain crop varieties that were higher yielding not mainly because they were more responsive to higher inputs, but rather because they were bred for cooperation over competition.

A plant, say, rice, growing in a field of other rice plants with different genotypes, will compete for sunlight by attempting to grow taller than its neighbors.  If the plant can capture more sunlight, it will be rewarded by having a larger share of the next generation made up of its own offspring, offspring that will carry the parent’s genes that made it competitive.  This competition for sunlight comes with a cost, however.  By expending so much of its resources on all that stem tissue, it has less left over for seed production than it would if it had no neighbors with which to compete.  Meanwhile, the neighbors are also growing taller in an attempt to improve their own reproductive success.

What if all the plants in the field had identical genes?  Competition would then be a zero-sum game, whereby the amount of extra seeds that a plant could produce by out-competing its neighbors would not increase the representation of its genes in the next generation at all, considering that the seeds they would supplant would have had those same genes.  This shared genetic fate allows for the possibility of cooperation.  If a breeder comes up with a variety that does not grow tall, and if it is grown in monoculture, it can give a higher yield than a related variety that expends valuable resources on extra stem production.

Of course growing a single variety in monoculture is problematic for other reasons.  A pest or pathogen that is able to overcome the genetic resistance contained in that variety will be able to overcome the entire field in short order.  Without a diversity of genotypes for natural selection to operate on, there can be no survival of the fittest among equals.  Also, a plant variety without a competitive drive will do poorly against weeds.  Thus, Green Revolution seeds are sold with a package of manufactured inputs to compensate.  There are other undesirable outcomes of Green Revolution technology as well, but the evolutionarily sound principle of selection for cooperation remains useful.

Perhaps the best example of selection for cooperation that Denison presents involves chicken breeding.  Chickens are evolved to be individually competitive as much as plants are, and they will peck at each other to the point of reduced overall egg output.  One investigator, however, had the idea of selecting not the best individual chickens for breeding, but rather the best groups.  Chickens were kept in pens of four individuals, and the ones allowed to breed were those with the greatest number of eggs per pen.  In six generations fighting was way down and egg production was up.  Through group selection chickens were bred to be cooperative.  The key was to measure the output of groups rather than individuals, a principle that could be applied to small plots of crops as well.

Denison also returns to the example of cooperation in the legume-rhizobium symbiosis.  Legumes produce nodules on their roots to host bacteria that can convert atmospheric nitrogen into nitrogen fertilizer.  The plant rewards the nodule-dwelling bacteria, collectively known as rhizobia, with food that it produces from photosynthesis.  It would seem like a beneficial relationship to both partners, but any bacteria that could harness the plant’s feeding system without going through the trouble of producing fertilizer would do even better.  These “cheater” rhizobia are common, and evolutionary theory predicts that they should have long ago displaced beneficial rhizobia unless the plant could somehow pick and choose among rhizobia strains.  Denison and his students showed that legume plants can indeed punish a poor-performing strain enclosed in a single nodule, causing that strain’s reproduction to suffer.  However, because of various considerations, detailed in Professor Denison’s explanation of the Hamilton r coefficient, the legume’s sanction system is only just good enough to get the fertilizer it needs for the current season, leaving plenty of poor performers in the population that is released back into the soil.  With this in mind, Denison comes up with a scheme for breeding legumes for stronger sanctions, involving the use of a test crop to evaluate the legacy rhizobial benefit left by the previous season’s plant, and then going back to breed from the seeds that had been collected from the plants that were subsequently shown to have left the most beneficial rhizobia behind.

Denison has an opinion on genetic engineering of crop plants that is different from most.  He is not opposed to genetic engineering in principle, but maintains that most such research will have little to show for the vast sums of money spent on it.  The reason is that the kinds of intervention that are possible by tweaking genes have already been tried over millions of years of evolution and thousands of years of coevolution, and any unequivocally beneficial simple modifications have already been adopted.  Additional modifications involve tradeoffs and are not likely to make a significant difference.  He does leave open the possibility of modifications that are too complex for evolution to have stumbled upon, such as re-engineering the chloroplast to recycle the byproducts of photorespiration, but most modifications of that level of complexity are beyond the scope of genetic engineering.

My biggest complaint about the book is the straw man that Denison sets up to provide a false balance for his critique of genetic engineering.  He selects some of the philosophers of alternative agriculture such as Wes Jackson and uses them to stand in for the science of agroecology.  He then provides a very reasoned scientific critique of some philosophical exhortations to farm using nature as a model to mimic, using his arguments to attack a certain conception labeled “agroecology” while ignoring the science of agroecology as actually practiced.  Agroecological scientists do indeed look to natural systems − as well as to traditional agriculture − for principles that can be used for improving modern agriculture, but then they test any ideas that come from such an examination.  That is what makes agroecology a science, and Denison even advocates for this kind of examination and testing.

Throughout the book I find many other points that I want to respond to, but on the whole the work is valuable, and a blog post cannot fully do it justice.  Denison is a very accessible writer, and part of the book is background on agriculture and on evolution for the uninitiated.  The text is peppered with interesting tidbits, such as the wasp that can smell when a butterfly is about to lay eggs and pursues her or hitches a ride on her to parasitize those eggs after she has left them.  Most importantly, though, agriculture cannot escape the principles of evolution, and Denison has provided plenty of food for thought.