Two Peppers

There is a sentence buried in the introduction of a 2019 paper that is doing more damage than most of the arguments people have about solar panels.

The authors name it themselves, as the thing they are writing against — a conventional understanding of land use that asserts, in their words, an inherent “zero-sum-game” of competition between solar photovoltaic installations and agricultural food production.

Land is either a field or a power station. Every panel is a hectare that stopped being a farm. It is a tidy premise, it decides planning applications and land allotments and permits, and the authors point out that this either-or discourse drives those decisions whether or not it is true.

So somebody went and checked. They put panels over a field outside Tucson, planted three crops under them, and measured what happened — not the yield alone, but the carbon going in, the water going out, the moisture left in the soil, and the temperature of the panels themselves.

What came back was not a win. It was something more useful and much harder to headline: three plants gave three different answers, and two of them were good in completely different ways.

Before any of the field results make sense, one piece of settled plant physiology has to be on the table, and it is the whole hinge of this zine.

Photosynthesis does not rise forever with light. At low light levels the rate climbs more or less in a straight line — and a plant kept down there is under stress, because it cannot reach its growth potential. Then, as light keeps rising, the climb levels off. It flattens. Past a certain intensity, more photons produce no more sugar, because the capacity of the light-harvesting reactions is finite.

That flattening has a name: the light saturation point. And the fact that makes it interesting is not that it exists. It is that it is species-specific. Every plant has one and no two are in the same place.

Which means the shape of that curve is a design parameter. As the Fraunhofer group put it, plainly, in a chapter about how to build these systems:

Notice what that sentence quietly refuses to say. It does not sort plants into the ones that can take shade and the ones that cannot. It sorts shade to plants.

A curve on a diagram is an argument. Here is the same thing as a number somebody actually harvested.

Stallknecht and colleagues grew basil, petunia and tomato under a range of semi-transparent coverings and tracked yield against the daily light integral — how many moles of usable photons land on a square metre in a day. For basil, the relationship came out sigmoidal: yield rose steadily between about 6 and 12 mol m⁻² d⁻¹, and then, in the paper's phrasing, did not increase much further once the daily light integral went past roughly 12.

Their unshaded control was sitting at about 20.

Roughly two fifths of the light falling on a basil plant is not being converted into basil. It is not being wasted, exactly — there is nothing there to waste it. It is arriving at a machine that is already running flat out.

Do not quote that 40 per cent as theirs. It is our arithmetic on their two numbers, and it is a rough figure about one crop in one experiment. The finding is the shape of the curve; the percentage is just the shape said out loud.

Back to Tucson. Three crops, all in the nightshade family, all under the same array, all through the same season.

Read the two peppers again, slowly, because the interesting thing is easy to slide past. Chiltepin and jalapeño are the same species. One is the wild variety; the other is the domesticated one. Same genus, same species, different varieties, identical treatment.

The chiltepin took the shade and tripled its fruit. The jalapeño actually took up less carbon under the panels — on a productivity chart, a loss — and still came out with the same amount of fruit, because it had become dramatically better at holding water. Same harvest, two-thirds less of it spent.

Neither of those is the failure case. One plant did more with the situation. The other did the same with far less. There is no version of this table where you can rank them without first deciding, arbitrarily, whether you are short of land or short of water.

There is an observation sitting on that table, and it needs handling carefully, because it is exactly the kind of thing that is too satisfying to be trusted without a hedge.

The chiltepin is the wild one. It is native to southern North America and northern South America, it grows under desert shrubs, and nobody bred it for anything. The jalapeño is its domesticated sister variety — widely domesticated across large biogeographic space, in the paper's words, and of considerably greater commercial value.

Under the panels, the wild one flourished. The bred one coped — well, and by a different route, but coped.

Here is the hedge, and it is load-bearing. The study did not test domestication as a variable. It picked two varieties of one species because they represent different dryland niches, and any story about what breeding did to the jalapeño is our inference, not their finding. Two plants is not a result about agriculture.

But the shape is worth sitting with, even unproven. We select crops in full sun, for full sun, and then we measure them in full sun. A variety optimised inside one narrow set of conditions is a variety whose performance outside those conditions nobody was ever paying attention to.

The chiltepin was not better. It was never asked to be uniform.

If this zine stopped at spread five it would be a nice story about solar panels. It does not, because two of the papers on this desk contradict each other, and the contradiction is better than either result on its own.

In Arizona, cherry tomato under the array doubled its fruit.

In a Michigan greenhouse, Stallknecht and colleagues found the opposite, and found it emphatically. Where basil and petunia gave those flattening sigmoidal curves, tomato yield increased linearly with the daily light integral and never approached an upper asymptote. There was no saturation point in range. Every photon removed cost fruit. Even their best-transmitting coverings came in 25 and 37 per cent below the control — and on ripe fruit at harvest the gap was 52 and 74 per cent, because the shaded plants were also slower.

The same genus. Doubling under 70–80 per cent shade in one place; losing a quarter of its yield to mild shade in another.

The temptation here is to pick a winner. Field beats greenhouse, or the newer paper beats the older one, or somebody's cultivar was different. Resist it. Both measurements are careful and both are almost certainly right about the place they were made.

The two tomatoes are not in conflict. They are answering two different questions, and neither paper ever claimed otherwise.

Ask what was scarce in each place.

In an Arizona summer, the limiting factors are heat and water. A tomato out there is losing the argument with the atmosphere long before it runs short of photons — the paper describes the crop as heat-sensitive to the point that summer flowering is accompanied by abortion from excessive temperature. Take away some light and you take away heat and evaporative demand with it. The plant gets back more than it loses. Barron-Gafford's team attribute the gain to exactly this: an alleviation of multiple stress interactions from heat and atmospheric drought.

In a Michigan greenhouse in the same season, light is the scarce thing. Remove some and you have removed the thing that was already in short supply. Nothing is given back.

This is why the Fraunhofer chapter can write, in the middle of a sober engineering discussion, that when other factors — water, for instance — are already limiting crop growth, shading is not necessarily a hindrance, and may even be beneficial.

Say the general form of it, because it is the whole zine. Whether a characteristic reads as a deficit depends on what the surrounding conditions have made scarce. Change what is scarce and the same characteristic changes sign.

Once you know the saturation point is species-specific, the engineering question inverts. It stops being which plants can cope with our panels and becomes what spacing does this plant want.

The Fraunhofer chapter states the consequence outright, and it is a striking thing to find in a technical handbook:

All of them. Not because every plant is secretly shade-tolerant, but because shading rate is a dial and somebody gets to set it. The failure mode is not an unsuitable crop. It is an unmatched ratio — one number chosen for the whole field and then applied to every plant in it regardless.

The geometry has been on the table a long time. In 1982, Adolf Goetzberger and Armin Zastrow proposed lifting the collectors about two metres off the ground and spacing the rows at roughly three times the collector height — which, they calculated, delivers nearly uniform radiation at the ground of about two thirds of what an open field receives. Forty-four years ago, in the first paper on the subject, the design question was already how much light to leave.

One borrowed word, returned. These systems are scored with the land equivalent ratio — how much separate land you would need to match what the combined plot produces. Above 1 means the pairing beats the split. It is not a solar metric: Mead and Willey defined it in 1980 for intercropping, for farmers growing two plants in one field. The tool for measuring this arrived from people who had already stopped planting one thing at a time.

Everything so far has been about what the plants got. There is a second half to the measurement, and it is the part that stops this being a story about accommodation.

Photovoltaic panels lose efficiency as they heat up. Across the growing season, the panels in the agrivoltaic installation ran about 8.9 ± 0.2 °C cooler in daylight than the same panels over bare ground — because the crop underneath was transpiring, and a transpiring plant is an evaporative cooler running on sunlight.

The soil held on to more water, too. Irrigating every two days, moisture in the agrivoltaic plot stayed about 15 per cent higher (3.2 volumetric units) than the control before the next watering. Even under daily irrigation it held 5 per cent more. The authors note something further: soil under the panels after two days was still wetter than the control's driest point after one — which suggests the watering could be cut further than they cut it.

This is the bit the zero-sum framing cannot represent at all. It has one axis and two claimants, so every outcome has to be a split. A relationship where both parties come out ahead does not fit on that axis — not as a compromise, not as a subsidy, not as one side being generous.

A hopeful piece that cannot state its own costs is just optimism. Here is what the same literature says about itself, in its own words.

The evidence base is thin. Weselek and colleagues, reviewing the field, write that although the technology is already running in commercial projects, its practicability and impact on crop production have hardly been investigated, and that only very few studies address it. The Fraunhofer chapter adds that most results lack robust long-term observation. Several of the striking numbers in this zine come from single seasons on single sites.

Nobody is sure how these things behave in a storm. Zahrawi and Aly, reviewing wind loading, report a contradiction between different versions of the American Society of Civil Engineers standards and the wind-tunnel results, and name a significant knowledge gap in wind load mitigation. Structures being sold partly as climate adaptation, with unsettled guidance on extreme weather.

The policy scaffolding barely exists. Alves, Marques da Costa and Sirnik screened 308 publications and found twelve addressing policy instruments for agrivoltaics, all published between 2021 and 2024.

And a result we are not going to use. One trial of corn — a C4 plant, with a light saturation point higher than most, so it should suffer under shade — produced a surprising outcome, with shading rate simply tracking yield reduction proportionally. The authors flag that the plot was only 100 m², which could explain it. We are printing it as an unresolved oddity rather than as evidence, because that is what it is.

We did not go looking for this. It was already in the table.

The jalapeño took up eleven per cent less carbon under the panels. Score that column alone and you have a plant that underperformed. Put the water column next to it and the same plant is the most efficient thing in the trial. The number did not change. The measure changed, and with it, which plant was failing.

The tomato is the sharper case, because it is one crop giving both answers. Nobody had to reclassify the tomato. All that changed was which resource the surroundings had made scarce — and the identical characteristic went from a liability to an advantage without the plant doing anything at all.

This is not a metaphor we are importing. It is what the measurements say. A trait becomes a deficit when it meets conditions that were arranged without it in mind, and it stops being one when the conditions change. The chiltepin did not need accommodating in Tucson because the shade was already the accommodation. The greenhouse tomato needed light because someone had built a room where light was the limit.

Which puts the design question exactly where the Fraunhofer chapter put it, and where we would put it about a classroom or an office: set the conditions to the organism, not the organism to the conditions. There is no unsuitable crop. There is a ratio nobody adjusted.

And a rhyme, offered as one. Every plant has a saturation point — a level past which more input adds nothing, and beyond which it starts to cost. A great many people reading this will recognise that curve from the inside. The mechanisms are not the same and we are not claiming they are; photosynthesis is not sensation. But the shape is the shape, and it is worth knowing that there is a point past which more is not better is a measured fact about living things and not a personal shortcoming.