Knowledge library: disease models and biological control
How disease models and biological control agents work, which measurements matter, and how FairGrow brings its own models together with the knowledge of scientists.
Our own models and the knowledge of scientists
Almost all fungal and oomycete models run on the same measurements: temperature, humidity, leaf wetness and rainfall, hourly. Insect models work mainly with temperature. What that means for your fields, you can see under disease risk.
How an infection model works
A fungal spore on a leaf needs water to germinate. The warmer it is, the fewer hours of wet leaf it needs, until it gets too hot.
An infection model counts those wet hours and weighs them by temperature. Once the total passes the threshold from the literature, the disease had its chance.
That is why almost every model asks for the same measurements. What sets them apart is the threshold, and the disease and crop it was tested on.
Where models meet
FairGrow works with its own models and with models developed by scientists. Behind every model are years of research: field trials, thousands of measurements and a publication reviewed by peers.
Inside FairGrow those models meet. The work of one research group complements that of another, and together they see more than each does alone. That creates connections no single paper made, and the results can be surprisingly good.
For scientists that is rewarding too. You put time and energy into a study, it ends in a publication, and then what? When your knowledge goes on to make a real difference on a farm, for a grower who acts in time or can skip a spray, how good is that?
Two things to know
Models developed with weather station data often underestimate what happens in the crop; Philbrick et al. therefore validate Botrytis with sensors in the field. And leaf wetness is the deciding measurement in almost every model: measured leaf wetness beats calculated leaf wetness.
When a biological agent works
Whether a biological agent works depends heavily on the weather during and after application. In the model by Fedele, Bove, González-Domínguez and Rossi (2020), the agent’s temperature and moisture requirements and its survival together explain more than 90% of the differences in efficacy. The authors conclude that the choice of agent should depend on the weather and the forecast at the time of application. The article is open access (Creative Commons).
Same product, different weather
A biological agent is a living organism. Beauveria spores have to germinate on the insect, and that only works at high humidity. Strong sunlight breaks them down before they can get to work.
That explains why a product works well in one field and not in another, even with the same dose and timing. If you know the weather during and after application, you can predict that difference.
Jeger, Jeffries, Elad and Xu (2009)
Predicts the chance of success of a single biological agent per mode of action: competition, antibiosis, parasitism and induced resistance.
Why it mattersA basis for calculating a chance of success per product.
Fedele et al. (2020)
Pathogen, crop and agent each respond to temperature and humidity. Worked out for Botrytis in grapes and nine commercial strains.
Why it mattersEfficacy E = (untreated − treated) / untreated, as a function of measured weather.
Meikle et al. (2003)
How long spores stay viable after spraying, as a function of temperature and humidity. Tested with sensors in the crop.
Why it mattersPredicts how many days a Beauveria application stays effective.
Faria et al. (2022)
A meta-analysis of 59 trials. Above 27 °C and with heavy rainfall, efficacy drops.
Why it mattersShows how to break efficacy down by weather.
Beauveria bassiana
Targets: Insects (thrips, whitefly, beetles)Germination needs high humidity, around 92%. It germinates fastest between 25 and 32 °C and not at all above 37 °C (James et al. 1998; Martin et al.). UV breaks down the spores.
Metarhizium
Targets: Soil insects, grasshoppersResponds to temperature and moisture much like Beauveria; the same survival model applies (Meikle et al.).
Trichoderma
Targets: Soil fungi (Fusarium, Rhizoctonia, Sclerotinia)Efficacy depends strongly on temperature: low to absent at 10 °C (Longa and Pertot 2012), and it drops in heat and drought (Brizuela et al. 2023).
Bacillus (subtilis, amyloliquefaciens)
Targets: Leaf and soil diseasesLess sensitive because it forms spores. Efficacy drops at high temperatures and with heavy rain (Faria et al. 2022).
Entomopathogenic nematodes (Steinernema, Heterorhabditis)
Targets: Soil insects (vine weevil, leatherjackets, larvae)The moister the soil, the more virulent they are (Grant and Villani 2003). A soil temperature above 32 °C in the first 24 hours, or frost in the first days, makes the application fail (USDA ARS). S. feltiae is active from roughly 8 to 30 °C (NBAIR).
Biologicals against fire blight (defence inducers)
Targets: Apple, pearMust be applied days to weeks before the infection event. For biologicals the Maryblyt threshold is lowered to 40 to 70% (NCSU).
Application window and efficacy window
Every biological agent has a best moment to apply it, and a period in which it keeps working afterwards. Both can be calculated from the same measurements the disease models use, plus soil temperature and soil moisture.
Research that never stands still
FairGrow keeps researching biological growing. New publications, trial series and datasets appear all the time, and whatever is useful goes into our own dataset: in which weather an agent takes hold, how long it keeps working, when a natural enemy is active.
That turns biological growing into a choice more farms can back up, rather than a gamble. And the more growers see it work, the more attention biological growing gets.
Biological growing is not about swapping one product for another. It is about knowing when nature is working with you. Because we care about that ourselves, we keep looking for the knowledge that makes that moment predictable. You can see it in the field.
The FairGrow team
Where to find products, authorisations and properties
For biological products the CABI BioProtection Portal is the main source: according to CABI’s flyer (November 2025), more than 6,500 registered biological products in more than 50 countries, searchable by crop, pest and country. Which products are authorised in your country, you can see in FairGrow under crop protection.
How to look up a biological product
Choose your crop and the pest or disease you are dealing with.
Check the CABI BioProtection Portal for the biological products registered in your country.
Check the authorisation and label in the national database, or in FairGrow under crop protection.
Trial standards and assessment reports
Manufacturers’ raw trial data are not public, but the standards they were produced under and the assessment reports are. Anyone who sets up a trial to the same standards gets results that manufacturers and regulators recognise.
What makes a fair trial
A trial compares treated plots with untreated plots in the same field, in the same season. That way you see what the product did, not what the weather did.
Randomising the plots and repeating the trial a few times averages out chance. A wetter corner or a heavier patch of soil then does not count as an effect of the product.
That is exactly what EPPO PP 1/152 and PP 1/181 lay down.
A trial everyone recognises
There are international standards for how to design, assess and report a trial. Manufacturers and regulators work with them, and in the EU they are the reference point for every dossier.
Setting up a trial yourself, on a few plots or across a whole field? Follow the same standards. Then your outcome is no longer an anecdote but a result an adviser, a manufacturer or a fellow grower takes seriously.
Want to know how to approach that on your own farm? Just ask us.
What a product had to show
Behind every authorisation is an assessment report. It sets out what a product demonstrated in trials, at which dose and under which conditions. Those reports are public.
It is the sober counterpart of the product brochure: not what a product promises, but what it has shown. Which products are authorised in your country, you can see in FairGrow under crop protection.
Three examples, nothing more. These publications are a small sample of the vast body of scientific knowledge FairGrow has processed, and it keeps growing every day.
Faria et al. (2022)
59 trials with Trichoderma and Bacillus against Sclerotinia in Brazil, broken down by temperature and rainfall.
Nordic-Baltic SPOT-IT
47 field trials validating Septoria risk models.
Organic viticulture in Italy
20 farms over two years. The model gave the same protection as the grower’s spray schedule, with 38% less copper.
Calculating and proving efficacy
How do you calculate what a product did, and how do you prove it on your own farm? And how do you know whether the leaf was wet when there was no sensor?
Abbott (1925)
For separate treated and untreated groups. C and T are the share alive or healthy in the control and the treatment.
Henderson-Tilton (1955)
For counts before (b) and after (a) treatment in the same plots. The standard for mite and insect counts.
Worked example
In the untreated plots 80 of 100 aphids are still alive after a week, in the treated plots 20. According to Abbott the efficacy is (1 − 20/80) × 100 = 75%. The formula corrects for what happened without the product too: without that correction you would say 80%.
All variants (Abbott, Schneider-Orelli, Henderson-Tilton, Sun-Shepard) are in the R package ecotoxicology (GPL-3). A 2024 tutorial shows how to estimate Abbott better through a generalized linear mixed model, because the formula is statistically weak with few observations. For the course of an epidemic, AUDPC (area under the disease progress curve) is the standard measure; see epifitter.
Your whole field as a trial plot
A trial plot tells you something about that one plot. On your own farm you want to know what a product does across the whole field: in the wet spot, on the heavy soil, along the edge.
Leave a few strips untreated and compare. Satellite imagery, sensors in the field and your yield map together show where the product worked and where it worked less, separate from the differences that were already there. That turns every season into a trial, without much extra work.
Why leaves get wet at night
After sunset the leaf cools down. When the temperature drops to the dew point, dew forms, even without rain. As long as the temperature stays there, the leaf stays wet.
Estimation models mimic this with humidity and temperature, and sometimes wind and radiation. Wind dries the leaf faster, morning sun warms it up.
The more you measure, the better the estimate
There are several ways to estimate leaf wetness. The simplest looks only at humidity and misses wet hours as a result. When a model also uses temperature, wind and radiation, it gets closer to the real leaf, provided the sensors are good.
Self-learning models achieve the highest accuracy in comparative studies. Even so, measured leaf wetness in your own crop remains the benchmark.
A model is as good as its sensor
A physical model is only as good as the cheapest sensor feeding it. Leaf wetness sensors are not standardised (coating, angle, height, orientation), so calibration decides what they are worth (Montone et al.). If you have no leaf wetness sensor, FairGrow estimates leaf wetness for you; see disease risk.
Which measurements make the difference
Most models on this page rest on a handful of measurements. These six decide whether a disease model or a biological agent turns out reliable. Which sensors FairGrow supports, you can find under sensors.
Where you measure matters
A weather station measures at 2 metres, in the open air. Inside the crop it is often more humid at night and the leaves stay wet longer. For disease models and biological agents, what happens in the crop is what counts.
The same goes below ground: nematodes and Trichoderma work in the top few centimetres, so that is where the sensor belongs.
Leaf wetness
WhyThe deciding input in almost every fungal disease model.
AdviceMeasure it in the crop, or calculate it with a calibrated model such as CART or Penman-Monteith.
Humidity in the crop
WhyDrives germination of Beauveria and Metarhizium, and the late blight rules.
AdviceMeasure at crop height, not at 2 metres.
Soil temperature at 5 cm
WhyNematodes fail above 32 °C or with frost in the first days; Trichoderma is barely active around 10 °C.
AdviceMeasure at 5 cm depth, where the agents do their work.
Soil moisture
WhyNematodes become more virulent as the soil gets moister; Trichoderma works less well in drought.
AdviceMeasure in the root zone.
Radiation and UV
WhyUV breaks down the spores of fungal agents such as Beauveria and Metarhizium.
AdviceA radiation or PAR sensor gives a useful indication.
Rainfall
WhyRain wets the leaf, splashes spores around and washes products off. Bacillus works less well in heavy rain.
AdviceMeasure on the field itself, with a rain gauge clear of trees and buildings.
Open-source code and model platforms
Science is sharing more and more. Models no longer appear only as an equation in a paper, but also as code and as a service you can call. What one research group builds, another can test, improve and apply.
FairGrow follows this closely, builds its own and works together wherever that helps the grower. Are you a researcher, developer or adviser and want to know how we go about it? Or do you have a model that deserves to prove itself in practice? Let us know.
Status: 1 November 2026. Every source citation links to the original publication.
Short answers
Is this a spray recommendation?
No. This page explains how disease models and biological control agents work and what that knowledge rests on. The label, the authorisation in your country and your own observation come first. Which products are authorised, you can see under crop protection.
Which models does FairGrow use?
Its own models and validated models from science, per crop and per disease. Which ones apply to your fields, you can see in FairGrow itself, with the source alongside every result. How that works, you can read under disease risk.
Do I need a leaf wetness sensor?
Not to get started: without a sensor, FairGrow estimates leaf wetness for you. Measured leaf wetness is better than a calculated value, though; leaf wetness explains why.
Why do so many sources come from abroad?
Most models were developed and tested in the US, the UK, Italy and Norway. The thresholds hold there; on your farm the outcome can differ. That is why every result comes with its source, so you can check where it was tested.
The risk on your own fields
FairGrow calculates per field, with its own models and those from science, and with the source alongside every result.