Evolutionary trade-offs: how are soybeans like salmon?
Answer: they're both semelparous (reproduce once, then die), so evolutionary trade-offs between number and size of offspring are expected to be similar.
This week's paper is "Evolutionary aspects of the trade-off between seed size and number in crops" (Field Crops Research 100:125-138) by Victor Sadras. You can read the abstract on the web for free. For the full version, you can pay $30 to download, visit your nearest agricultural research library (in the U.S., often at a state university), or email the author at: firstname.lastname@example.org. My discussion is mostly based on a shorter version presented at the Australian Agronomy Conference.
Demand for grain is increasing, to feed growing human and livestock populations and more recently for ethanol production. Unless those trends are reversed, we will either need to expand the land area used for agriculture or increase grain yields per unit area. Grain yield is the product of plants per area, seeds per plant, and weight per seed. Unfortunately, increasing any one of these (by increasing seeding rate, or through plant breeding) tends to decrease the others.
This paper looks at how natural selection (in the wild ancestors of crop plants and in fish) and plant breeding (especially in maize and sunflower) shape trade-offs between seed number per plant and seed size. The similar patterns in plants and fish show that, as predicted by the relevant aspects of evolutionary theory, we are dealing with fundamental constraints that we are unlikely to change.
The genomes of plants and fish (and other semelparous species) are the legacy of past evolution in which the variants that survived were those that maximized the number of offspring surviving to maturity, subject to two constraints:
1) resources are more abundant some years than others, and
2) for any given level of resources, there is some optimal trade-off between size and number of offspring.
In theory, plants could deal with variation among years by making more seeds in good years, by making bigger seeds in good years, or both. A major point of this paper is that most crop plants respond to good conditions mainly by making more seeds, not bigger seeds. Figure 4 (3 in full version) shows 291% variation in seed number of soybean, but only 43% variation in seed size. Trout showed even greater tendency to vary number of eggs rather than their size. More extensive data for a variety of crops confirm this point in the full paper (Fig. 4).
Why vary number rather than size? Sadras refers to a paper by Smith and Fretwell in American Naturalist, a journal read by more evolutionary biologists than crop scientists. This paper showed that, if the chance of a seed surviving to become an adult plant increases with its size, but with diminishing returns (twice as big is only 50% more likely to survive, for example), then there will be some optimum seed size. Beyond that size, the plant would increase its number of descendants more by making more seeds, rather than larger seeds. Colin Donald (whose papers have greatly influenced my own thinking) came to similar conclusions, based on a clever experiment involving competition between seeds of the same genotype but different size.
Sadras concludes that there has been stabilizing selection for seed size. Does this mean that, once seeds have reached their target size, natural selection would favor cessation of seed growth, even if the plant still had resources available that could be used for additional growth? I don't think so, unless seed survival actually decreases with seed size beyond some limit (e.g., due to pod splitting). Rather, it means that there is strong selection for getting seed number "right", i.e., starting the number of seeds that will use all the available resources in growing to the target weight per seed. This is tricky, because some plant processes that determine eventual seed number occur relatively early in the season. What if the rest of the season turns out to be unusually wet or dry?
Sadras shows that there is indeed a strong relationship between resource availability (as indicated by growth rate) and seed (or fish egg) number (Fig. 1). For barley, wheat, and "prolific maize", seed number is predicted with reasonable accuracy by assuming that seed size can't vary (Fig. 2).
On the other hand, maize limited to one ear per plant was unable to increase seed number enough to use all resources available under ideal conditions, so it departed from theoretical predictions. Sadras points out that, in breeding for maize with only one or two ears, or sunflower with a single head, we have decreased the ability of plants to respond to their environment by varying seed number. (Wild sunflower, and teosinte, the wild ancestor of maize, both have more flowers per plant than their cultivated descendants.) They appear to compensate somewhat, with more variation in seed size, relative to seed number, than crops like wheat or soybean (Fig. 4b in full paper). He also mentions work by Harper showing that variability in seed size increases when variation in seed number is limited surgically.
How will crop scientists use the evolutionary insights in Sadras' paper? One possible follow-up would be to think about how variation in resource availability differs between today's agricultural fields and the past environments that shaped the evolution of crops and their wild ancestors. If water supply, for example, is consistently better in irrigated agriculture, then bet-hedging mechanisms that control seed number might be too conservative. If plants make too few seeds, they may have "extra" resources available during seed-fill. Are current crop genotypes able to use those extra resources to make each seed bigger, or will some resources be wasted?
A more general answer is that any crop scientist with some spare time (often a problem, as decreased funding for agricultural research in many countries has increased workloads) could benefit from reading more about evolutionary trade-offs, especially in plants, but also in other species. Salmon and soybeans are different in many ways, but this paper shows that the similarities can be illuminating.