Yield and protein in wheat

Why does the same wheat variety make fine bread on one farm and only biscuits on the next?

Two farms sow certified seed of the same variety, the same year, from the same bag. One delivers grain a miller will pay a premium for and bake into a loaf that rises tall; the other delivers grain that will only go for biscuits or feed.

Worse, it is often the better-looking crop that fails the miller — the heavier, greener, higher-yielding one. That inverts the usual expectation, that a good crop is good for everything. What is the miller measuring that the farmer's weighbridge does not?

What the miller cares about is gluten: the protein network that traps carbon dioxide and lets a dough hold its rise. So the useful quantity is grain protein, and the first thing to notice is that it is not an amount. It is a concentration — protein mass over grain mass — and a ratio has two ways to move.

Hold that and ask where each part comes from, because they arrive by different routes. The carbon is fixed after flowering, by leaves and ears working through grain fill. The nitrogen is mostly not new: the large majority of it was taken up before flowering, stored in leaf and stem, then dismantled and shipped into the grain as the canopy senesces. The nitrogen pool is largely fixed; carbon keeps arriving.

Now run a good grain-filling season — cool, moist, no disease, a long fill. Carbon accumulates well, grains are plump, yield is high, and the nitrogen pool set weeks earlier does not grow to match. Divide a fixed numerator by a growing denominator and protein concentration falls. That is the negative correlation between yield and protein, and notice that it required nothing to go wrong: the crop performed well and diluted itself. Run the opposite season — heat or drought during fill — and grains are shrivelled, yield poor, protein percentage high. That figure is not a sign of a good crop. It is small grain with the same nitrogen in it.

There is a second, deeper reason the two resist rising together, and it is an energy constraint rather than an accounting one. Building a gram of protein costs the plant substantially more photosynthate than a gram of starch — roughly twice as much, by the standard construction-cost estimates, because amino acid synthesis and peptide assembly consume more reducing power and ATP than polymerising glucose does, and nitrate must be reduced before any of it begins. So a fixed carbon supply buys fewer grams of grain if more of them are protein. The trade-off is not merely a ratio artefact; it is paid for in carbon.

Which explains the practical lever. If protein is short because the nitrogen pool is short relative to the yield achieved, nitrogen applied late — at or after flowering, when grain sites are set and cannot increase — should raise protein while barely touching yield. Applied early it does the opposite: it builds tillers and grain sites, raises yield, and can leave protein unchanged or lower. The same fertiliser, weeks apart, moves different variables.

A neighbouring walk-through on diminishing fertiliser returns treats the yield response curve: why the second hundred kilograms of nitrogen buys so much less grain than the first. This is the same input seen through a different outcome, and the fixed point of difference is worth stating — there the curve bends because some other resource becomes limiting and surplus nitrogen leaves the field; here the crop may sit at its yield optimum and still fail the miller, because the quantity in question is a ratio, and the rate that maximises tonnes is not the rate that maximises concentration.

Three refinements. Grain protein is not measured directly; it is nitrogen measured and multiplied by a conversion factor, conventionally 5.7 for wheat rather than the generic 6.25, which is why protein figures from different industries are not always comparable.

Second, quantity is not quality. Dough strength depends on which storage proteins are present — the high-molecular-weight glutenin subunits a variety carries — not only on how much protein there is. Genotype sets the character of the gluten and environment the amount, which is why "the same variety" in the question is doing real work: it holds quality constant so the site difference shows up entirely as quantity.

Third, this is where the account can be falsified. If protein is limited by nitrogen supply relative to yield, a late application at flowering should raise grain protein with little yield response, and the effect should be largest on the highest-yielding, most diluted crops. If instead late nitrogen raised yield rather than protein, or the highest-yielding sites showed the highest protein without extra nitrogen, the dilution account would be wrong.

Two honest limits. The inverse relationship is a strong tendency, not a law: breeders have found lines sitting above the usual trade-off line, and the trait is pursued precisely because it is not fixed. And I have given no threshold percentages for bread against biscuit grain, because those are contract specifications differing by country, mill and year — the milling standard, not the plant, decides where the line falls.

• How protein-based payment schedules change a farmer's nitrogen decision compared with payment by weight alone.

Every term the collection defines is gathered in the glossary.