Agroforestry overyielding

Why can a field that gives a tenth of its ground to trees produce more food than the same field cleared?

Take a hectare of cereal, and plant rows of trees across a tenth of it. You have removed a tenth of the crop's ground, its light and its water. Arithmetic says the harvest should fall by about a tenth. Sometimes it falls by less than that, and the mixture — crop plus tree product — comes out ahead of what the same land could produce as two separate fields.

That is a strange result, and it is worth being suspicious of it before believing it. Nothing was added. So either the trees are conjuring something, or the cleared field was quietly wasting something the trees can reach.

Start with the accounting, because the claim is easy to inflate. The honest measure is the land equivalent ratio: for each component, divide its yield in the mixture by its yield when grown alone on the same land, then add the fractions. Suppose the cereal in the agroforestry field returns 0.85 of its sole-crop yield, and the trees return 0.25 of what a full tree stand would give. The ratio is 1.10 — meaning you would need 1.1 hectares of separate fields to match one hectare of the mixture. That is a pure identity, not a measurement; it simply defines what "overyielding" means. Notice what it does not say: it says nothing about labour, capital, or the years before the first tree harvest, and it counts a tonne of timber and a tonne of grain as though land were the only thing being economised.

Now the physical question. For a mixture to beat the sum of its parts, the two components together must capture more of some resource than either could alone — or must use what they capture more efficiently. Where would spare resource be hiding in a cereal field?

Three places, and they are all about timing and position rather than quantity. First, light: a cereal canopy does not close until weeks after sowing and it senesces before the year ends, so a good deal of the season's sunlight lands on bare soil. A tree with leaves outside that window intercepts light the crop was never going to use. Second, water and nutrients below the crop's rooting depth: cereals root shallowly, and nitrate leached past that depth is lost to the crop entirely. A deep-rooted tree can intercept it — the "safety net" idea, and one of the better-evidenced ones. Third, the microclimate: a windbreak does not capture more of anything, it reduces the evaporative demand on the crop beside it, so each unit of water taken up buys more growth.

So the mechanism is not generosity. It is that the sole crop leaves slack, and the tree is positioned in space or time to take what the crop was leaving.

Which immediately tells us when it will fail. If the crop already intercepts nearly all the light for nearly the whole season, already draws water from the full profile, and is fertilised past the point of limitation — an irrigated, high-input field — there is no slack. The tree can then only take what the crop wanted, and the ratio falls below one. This is the reason the same design that overyields on a semi-arid rainfed plain loses money under irrigation, with no change in the biology of either species.

There is a second, distinct mechanism worth separating from the first: facilitation. A nitrogen-fixing tree does not merely partition a resource, it adds one, and the leaf litter it drops is a nitrogen transfer to the crop. That is a real effect but a different one, and lumping it in with complementarity is how the literature's more enthusiastic claims get made.

The result is not that trees make land more productive. It is that a monoculture is a poor filter for a resource that arrives in a pattern the crop does not match, and a second species with a different pattern is a way of catching the remainder. State it that way and the conditions come with it: overyielding requires unused resource, and a partner positioned to use it without displacing the first.

Two honest qualifications. Published land equivalent ratios for agroforestry vary enormously — by species pair, spacing, rainfall, soil depth, tree age, and, decisively, by what the trial chose as its sole-crop comparison. I will not quote a headline figure, because the boundary choice moves the answer more than the practice does. And "produces more food" is often not what happens: much of the tree component is timber, fuel or fodder, so the mixture can win on land while losing on grain.

The falsification test is direct. The account says the gain comes from extra resource capture, not from something unmeasured about mixtures. So measure the capture: intercepted radiation, transpired water, nitrogen recovered below the crop's roots. If a system overyields while capturing no more of any of these than the sole crop — same light, same water, same nitrogen, more output — the complementarity account is wrong and something else is doing the work. The sharper refuting observation would be consistent overyielding in irrigated, fully fertilised, closed-canopy systems, where the account predicts there is no slack to share and the ratio should sit at or below one.

• Why the land equivalent ratio flatters agroforestry when land is cheap and labour is not.
• How nitrogen-fixing shade trees in coffee and cacao mix facilitation with competition, and how growers tell which is winning.

Every term the collection defines is gathered in the glossary.