The Impact of Climate Change on Citrus Nutrition, Yield, and Fruit Quality
A climate review of citrus, and why "we found no effect" and "we could not have found one" are very different sentences
In 1987, researchers in Phoenix planted eight sour orange seedlings, surrounded them with clear plastic chambers, and began pumping extra carbon dioxide into half of them. They kept it up for seventeen years. It remains the longest CO2 experiment ever run on a fruit tree, and almost everything we know about how oranges will respond to a richer atmosphere traces back to those eight trees.
The headline results were encouraging. The trees given extra CO2 produced roughly 70 percent more fruit and wood. Their juice carried more vitamin C. Their leaves, after an early dip, settled back to normal mineral levels.
It is tempting to stop there and conclude that citrus will do fine. I spent several months writing a review of climate change and citrus, and the further I went, the less comfortable I became with that conclusion. Not because the data are wrong, but because of what they were never able to show.
The crops we already know about
Start with what happens to other plants. When CO2 rises, most crops photosynthesize faster and pack more carbohydrate into their tissues. The minerals do not keep pace, so every gram ends up a little more diluted. Across C3 plants, a meta-analysis put the average mineral decline at about 8 percent. In open-field enrichment trials, wheat grain lost 9.3 percent of its zinc, 5.1 percent of its iron, and 6.3 percent of its protein.
Those are small numbers per bite, and enormous ones per population. Billions of people get most of their zinc and iron from exactly these crops.
So the natural question is whether an orange behaves like a wheat kernel. And here is the part that surprised me: nobody has checked. There is no experiment, anywhere, that has measured minerals, protein, folate, or carotenoids in the fruit of a citrus tree grown under elevated CO2. We have leaves. We have whole-tree biomass. We have vitamin C, which is built from carbon and tends to move in the opposite direction. The part you actually eat has never been tested.
Why the Phoenix trees could not settle it
A missing measurement is a gap. What turned it into the center of the review was a second problem, and it is a statistical one.
The Phoenix study ran four trees per treatment, housed two to a chamber. For a seventeen-year commitment, that is a reasonable design, and the researchers deserve real credit: they published their standard errors, which is the only reason the next step was possible.
From those errors, I estimated how much juice vitamin C varied from tree to tree. The typical coefficient of variation was 2.9 percent, which is impressively tight. Then I asked what the original paper did not: with four trees and that little noise, what is the smallest change this experiment could reliably detect?
The answer is about 7 percent.
And vitamin C is the easy case. Fruit minerals are usually noisier. At a coefficient of variation of 10 percent, four trees can only detect a change of about 24 percent.
Put that next to the wheat numbers. Declines of 5 to 9 percent would have slipped straight through. The Phoenix experiment did not find that citrus escapes nutrient dilution. It was never in a position to find out.
This distinction, between evidence of absence and absence of evidence, is one of the most common quiet errors in science writing. The citrus data do contain a genuinely interesting hint: well-fertilized trees may scale up nutrient uptake as they grow, avoiding dilution entirely. That would be a meaningful result if it holds. But it is a hypothesis, and the review treats it as one.
It also does the constructive thing and specifies the experiment that would answer it: CO2 crossed with nitrogen supply, because the two competing explanations predict different outcomes when nutrients run short. Chambers or field rings as the unit of replication instead of individual trees. Eight replicates per treatment if variability is tight, twenty-six if it is typical. A full nutrient panel on juice and pulp, measured alongside fruit size and fruit number so that dilution can be separated from uptake. It fits on a single table, and it may be the most useful page in the manuscript.
Going back to the source data
The nutrition question anchors the review, but it was not the only place where I rebuilt someone else's analysis instead of quoting their summary.
Citrus greening. Huanglongbing, carried by a tiny insect called the Asian citrus psyllid, has done more damage to Florida citrus than any storm. Under the disease and a string of hurricanes, Florida's orange crop fell from 242 million boxes in 2003–04 to 12.2 million in 2024–25. A 2019 study modeled how temperature governs the disease's transmission, and its authors made the work reproducible by publishing their code. I used that code, refit it to the original laboratory insect data, and ran 2,000 bootstrap simulations to put honest error bars on the result.
The reconstruction matches the published peak, near 24 degrees, and the upper cutoff at 33. It does not match the lower cutoff, and the reason is worth knowing: depending on which of two lab studies you trust for egg-laying rates, the lower limit moves by about 6 degrees. The review reports that disagreement openly.
More importantly, the model reveals something the original curve only implies. Because transmission climbs steeply up to its peak and falls off above it, the same amount of warming pushes risk in opposite directions in different places. Regions whose cooler months sit below about 24 degrees (the Mediterranean, California, northern Florida) gain months when the disease can spread. The hottest tropical months lose them.
Lemon color. A research group in Spain tracked lemon peel color over six seasons and published both their model and their raw measurements. Their model predicts color from recent nighttime lows and the previous reading, and it reproduces their field data closely, with an R-squared of 0.95.
So I warmed it up. One extra degree of autumn warmth pushes the date when lemons turn properly yellow back by about 5 days. Two degrees, 14 days. Three degrees, 41 days, and at that point the fruit never reaches full yellow on the tree at all. The delay grows faster with each degree because color change slows sharply as nights approach the 15-degree threshold where it begins. For growers, that means more fruit needing artificial degreening after harvest, with the cost that comes with it. These projections come from a single site pushed beyond its original range, and the manuscript is clear about that, but the direction and the acceleration are hard to argue with.
What's in the full manuscript
The complete review runs 11 pages in a two-column journal layout: roughly 7,000 words, six figures, three tables, and 44 references, opening with a structured abstract and highlights.
It starts with physiology: why warm winters weaken the cold signal that tells subtropical citrus to flower, why heat and drought together cause damage neither causes alone, and how salinity and flooding choke nutrient uptake at the root. It then follows three outcomes, nutrition, yield, and fruit quality, back to their mechanisms. It covers the indirect threats that may outweigh the direct ones, from shifting growing regions in Japan to hurricanes and disease. It closes with adaptation options at the scale of the tree, the orchard, and the region, and a ranked research agenda.
Every number traces to a cited source, including official USDA production statistics, and the code behind all three original analyses is available.
This is for you if you work in tree crops, plant nutrition, climate adaptation, or food policy and want mechanisms instead of headlines. It is also for anyone who wants a worked example of reading a literature critically: what a power analysis does to a null result, and what you learn by refitting a published model from its own data.
This is not for you if you are looking for new field or lab data. It is a synthesis with three computational analyses, and it is explicit about where the evidence ends. That candor is the point.
Citrus growers will spend the coming decades planting trees for a climate no orchard has yet experienced. Being precise about what we know, and what we only assume, is the least we owe them.
[Get the full manuscript]
Citrus is grown mainly in subtropical and Mediterranean-type climates that are warming, drying, and seeing more frequent extremes. Because orchards stay in the ground for decades, trees planted now will mature under conditions that no current orchard has experienced.