By Warren Alphonso —

Stumbling across yield

I started out trying to concretely answer: how can we make food cheaper? One clear way to do this is to increase the amount of food each seed can produce, or some finite set of inputs (eg fertilizer, water) can produce, or some area of land can produce. I wanted to understand how crop yield has changed over time, and what the trends and future outlooks of changes in yield can look like. Thankfully, I stumbled across "Crop yields and global food security: will yield increase continue to feed the world?" by Fischer, Byerlee, and Edmeades, which is an amazingly detailed exploration of this topic. There's so much detail here that I find it overwhelming to ask and answer a finite question, which is how I prefer to write and publish posts. But I don't want to get stuck. So here's a less structured post where I try to organize the stuff I've come across so far.

What is yield? Mass of food per area of cropland. (Does it make sense to consider mass? Optimizing for that over calories or macronutrients or micronutrients seems like it might not be great. Though maybe individual crop genes don't differ a lot in densities of these things though.)

We can also consider the potential yield (yield under best realistic conditions, proper agronomy) as opposed to the actual yield (the report calls this "farm yield" which I'll use over "actual yield"). The yield gap of a farm for some crop is the difference between these two.

We can also classify yield change across two axes:

  1. was this change to potential yield or farm yield?
  2. was this change a genetic or non-genetic change?

For example, I've come across people talking about making cereals fix their own nitrogen, like legumes do. I think this would be a genetic change to farm yield. (It's not potential yield unless it changes the amount of nitrogen the plant can usefully absorb.)

Okay but literally what is yield made up of? How can we make number go up?

What is potential yield? Units of PY is mass/area.

PY = (above-ground plant dry matter mass/area) * (harvest index: ratio of grain yield to total dry matter)

We can break up dry matter (DM) into:
DM = (photosynthetically active radiation intercepted/time/area) * (radiation use efficiency: mass dry matter per joule)

We can constrain PAR and RUE also:

  • PAR:
    • Above Earth's atmostphere, we get an average of 1360 W/m^2 . At ground level, this peaks at about 1000 W/m^2 . About half of this is in photons energetic enough (<700nm) to be useful to photosynthesis, so we're at a peak of 500 W/m^2 = 40 MJ/m^2 / day max. Obviously we don't have peak sunlight all day, maybe cut this by 10x to approximate average, so we're at 4 MJ/m^2 / day.
  • RUE:
    • some example values of RUE that I haven't checked:
      • maize: 3.3 g DM / MJ
      • wheat: 2.7 g DM / MJ
      • rice: 2.2 g DM / MJ
      • soybean: 1.9g DM / MJ

Another perspective is that PY = (number of grains) * (mass per grain). Apparently most of the increase in yield recently has been due to increase in number of grains, while mass per grain has stayed constant or even decreased.