A recent paper in Science (Kromdijk et al., 2016) reports 15% faster early growth of tobacco plants genetically engineered to handle fluctuating light better.  Although the plants were genetically modified with genes from bacteria, the basis for the improvement was apparently an increase in activity of genes the plants already had.

I found this result surprising, because increased activity of existing genes is something I would have expected to have evolved naturally,  unless several independent and simultaneous mutations are only way to get this phenotypic change (so the phenotype may have arisen rarely, if at all) or unless benefits of greater activity of these genes are outweighed by costs or risks. The hypothesis that most “simple” improvements have already been tested by natural selection (and incorporated, unless constrained by tradeoffs) is a major theme of my book.

The relevant costs and risks are those that would have limited Darwinian fitness (survival and reproduction) of a plant’s ancestors in past environments, which may not always imply costs or risks to crops today.  But increased growth of the transgenic plants in one field experiment is not enough to convince me that this is a tradeoff-free improvement.

I also troubled by an apparent discrepancy between their photosynthesis data and their crop-growth data. Figure 3 in the paper shows that the modified plants have a greater photosynthetic rate than unmodified plants (about 13 vs. 12 umolCO2/m2s) for several minutes after a sharp decrease in light level, from 2000 to 200 uE.  How much yield increase would we expect from this increase in photosynthesis?

As I read the paper, photosynthesis by the unmodified plants will match the modified ones, if low light continues long enough, because the proposed benefit comes from faster adaptation to a decrease in light, not a consistent increase in low-light photosynthesis.  But let’s assume that light levels alternate between high and low often enough that the modified plants have the full low-light advantage in their Fig. 3.  For simplicity, let’s further assume that plants spend half their time in high light (photosynthesis about 28 umolCO2/m2s for both genotypes) and half in low light (12 or 13 umolCO2/m2s).  Then the average photosynthesis for the modified plants would be 20.5 vs. 20.0 umolCO2/m2s for the unmodified plants, a 2.5% benefit.

How can we reconcile a 2.5% photosynthetic benefit with the 15% faster growth reported in the field?  A good place to start would be to have someone not associated with this paper repeat the field experiment.  If the 15% faster growth is real, maybe this genetic modification has additional benefits, beyond those shown in the paper’s Figure 3.

However, crop plants can have a five-fold or greater difference in photosynthesis between 2000 vs. 200 uE light levels, much greater than the twofold difference in the paper’s Figure 3.  So, in many crops, photosynthesis at low light would make less relative contribution to daily photosynthesis and crop yield, decreasing the average benefit below 2.5%.

My tentative conclusion is that yield benefits from applying this genetic modification to major crops are likely to be much less than 15%.  However, even a 1% increase in worldwide yields of a major crop could have significant benefits.  The first farmers to adopt an innovation would increase their incomes.  As the innovation spreads, a 1% increase in food supply would drive down prices, reducing benefits to late-adopting farmers, but benefiting consumers.