Why we built an environmental-conditions monitor for Hemileia vastatrix across seven coffee origins — what it measures, why we also watch fertilizer affordability, where it falls short, and why a deliberately imperfect leading signal still earns its place.
Coffee leaf rust — Hemileia vastatrix, "la roya" — is the most economically important disease in the history of coffee. When it took hold in Ceylon in the 1870s it effectively ended the island's coffee industry; that is why Sri Lanka grows tea today. It reached the Americas in 1970 and has been endemic in arabica ever since.
It is an obligate fungus: it can only live on a living coffee leaf. Spores landing on the underside of a leaf germinate in a film of water, push a germ tube to a stomatal pore, and enter. Weeks later the tell-tale orange pustules erupt, the leaf drops, and a defoliated tree cannot fill its cherries. The damage shows up not as dramatic die-off but as a quiet haircut to yield and quality the following season.
The reason a trader should care is the 2008–2013 Central American and Colombian epidemic. Production fell roughly 31% in Colombia across the epidemic years versus 2007, and about 16% in Central America in 2013 versus 2011–12. The International Coffee Organization estimated losses on the order of half a billion dollars and hundreds of thousands of jobs gone in a single crop year, with more than two million people's incomes touched (Avelino et al., 2015). It was a genuine, multi-year arabica supply shock — and, as we will see, it was not caused by weather alone.
Rust has a narrow set of conditions it likes, and those conditions are forecastable. Our Coffee Leaf-Rust environment index turns that into a dekadal (10-day) risk calendar across seven origins — Brazil arabica and conilon, Colombia, Central America, Ethiopia, Vietnam, and Indonesia — built from hourly temperature and dew point (Open-Meteo, recent actuals plus a 14-day forecast).
The biology is more subtle than "warm and wet," and getting it right is the point. Infection is a two-stage process with two different temperature optima:
The practical consequence is counter-intuitive and easy to get wrong: cool highland nights are not protective. Once a leaf has been wet long enough — our engine looks for the longest continuous nightly wet spell, with roughly six hours as one infection opportunity — a mild-to-cool night is more conducive, not less. Our matrix scores each night from these two inputs (leaf-wetness duration × temperature) into a Low/Medium/High/Extreme class, then averages across the dekad. The result is a single cross-origin calendar you can read at a glance, with the two underlying inputs shown beneath it so you can see why a cell is hot.
This is the trigger. It answers one question well: is the weather loading up for an infection wave? It is deliberately a leading, weather-only signal — and on its own, it is only half the story.
Here is the part most weather-driven crop monitors miss. Whether conducive weather actually becomes disease depends on how well-defended the tree is — and that is set by husbandry: nutrition and fungicide. A well-fed, well-sprayed plantation can sit through a conducive spell with limited damage. A neglected one ignites.
And husbandry is economic, with a crucial twist: it leads by about a season, and it is inverted relative to the price you would expect. Fertilizer and preventive sprays are bought when a grower can afford them. When coffee prices are strong relative to input costs, growers invest, and the next crop is resilient. When prices sour or fertilizer spikes, inputs get deferred — and vulnerability builds into the crop a season or two later. Inputs applied yesterday set resilience today; what a grower can afford today is a leading indicator of next season's vulnerability.
This is not a theory we invented. The 2012–13 epidemic was not purely a climate event: it followed the 2011 price decline that squeezed Central American growers into cutting fungicide and fertilizer, and the peer-reviewed post-mortems name economic constraint as a driver of the outbreak's severity (Avelino et al. 2015; field studies on economic constraints and rust in Nicaragua). The same dynamic recently played out in reverse and in another crop: cocoa's 2024 price spike funded a wave of husbandry investment.
So alongside the weather we track a simple, transparent input-affordability indicator — coffee price divided by a fertilizer basket (urea + potash), as a percentile versus its own long history, lagged to the husbandry window and read by direction. It is not fused into the rust score. It sits beside it, because the honest reading is weather × vulnerability, and we want you to weigh them yourself.
The current readings illustrate the point better than any abstraction. The husbandry window behind the 2025–26 crop sat at the top percentile of affordability — strong prices, normalized fertilizer — which is exactly why this year's Brazilian crop is robust (a high-load "on" year, +23% arabica). But affordability is now souring hard as fertilizer rebounds and prices ease off their peak. That is not a statement about today's disease; it is a flag about the next crop's resilience.
We would rather you trust this signal for what it is than oversell it.
None of those caveats are fatal, because of what the signal is for. It is not a disease forecast. It is situational awareness — a way to see, weeks ahead and across every major origin at once, when the conditions and the vulnerabilities are lining up. Its value comes precisely from the discipline above:
For the coming weeks, three things are worth watching together — individually each is noise, but together they are a story:
The same series that drive this monitor — the dekadal rust-risk classes, the leaf-wetness and night-temperature inputs, the input-affordability trend, and the ENSO state — are the data we publish, queryable directly and through the MCP layer for AI-assisted analysis. If you would rather run the question yourself than take our read, that is exactly what the data is for.
Explore the Data