Jacob Weiner makes a compelling argument for human-imposed group selection as the most-promising route to higher-yielding crops, in his recent opinion paper, “Looking in the Wrong Direction for Higher-Yielding Crop Genotypes.”  His main point is one that he and I have made before: past natural selection is unlikely to have rejected genetic changes that are simple enough to have arisen repeatedly and that consistently improved individual fitness across the whole range past environments.

Those past environments would usually have included a variety of pests and pathogens, so there may not be much room for improving generic defenses.  But the specific pests and pathogens attacking a crop today are often different from those it saw in the past, and plant breeders can give us resistant varieties faster than natural selection would. So I would have revised Weiner’s paper slightly, to make it clear it’s focused on improving yield potential (yield in the absence of pests and pathogens).

Many of the changes proposed by biotechnologists meet the criteria of simplicity (increasing activity of an existing gene, for example), so the “improved” phenotypes have presumably arisen repeatedly and been repeatedly rejected by natural selection because they reduced individual fitness.  Some of natural selection’s rejects involve tradeoffs between individual fitness and the collective performance of plant communities, including seed yield.  Those have not usually been the phenotypes pursued by biotechnologists, but they should be.

Are we smart enough to identify opportunities linked to individual-versus-community tradeoffs, beyond those involving shorter plants?  I’ve made some suggestions, such as reducing solar tracking and limiting transpiration at times when water-use efficiency is low.  An alternative approach is to select for group, rather than individual performance.

Of course, plant breeders already do this, when they compare yields of single-genotype plots.  Weiner asks, “How can plant breeders practice group selection without using thousands of plots?”  Good question!  I’m not sure how many plots we’d need to have a good chance of finding additional useful traits, beyond shorter stature, more-erect leaves, smaller tassels on maize, and whatever other group-benefiting traits have resulted from the group selection plant breeders have already done.  With recent improvements in robotics, maybe comparing thousands of plots would be cheaper than most biotech projects.  While, we’re at it, let’s make sure plots are big enough to avoid edge effects, which can undermine group selection.