Industries / Food, farming and forestry / Agriculture and farming / Grain elevators and storage
Pests and spoilage
This is the work of keeping insects and moulds from eating or poisoning stored grain: finding them early, killing insects with fumigant gas or airtight storage, and keeping grain dry enough that moulds cannot grow.
5
Milestones achieved, from 2000 to 2023.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
2000
Electronic trap counts go astray
A probe trap is a perforated tube pushed into the grain; insects crawl in and fall to the bottom, where they are counted. The Electronic Grain Probe Insect Counter (EGPIC) added an infrared beam that counted insects as they fell and sent the counts to a computer. R. T. Arbogast, P. E. Kendra, D. K. Weaver and D. Shuman tested eight of them in stored oats at a seed plant in north-central Florida against weekly hand counts. Errors ran from 79.4% too few to 82.4% too many, and the average error, ignoring its direction, was 31.7%.
LessonAn automatic insect counter has to be checked against hand counts in a working store before its numbers are trusted.
Credit
R. T. Arbogast, P. E. Kendra, D. K. Weaver and D. Shuman
2004
Damp maize poisons Kenya
Between January and July 2004, 317 people in Kenya's Eastern and Central provinces fell ill from aflatoxin, a poison made by moulds, and 125 died, by the count reported by the US Centers for Disease Control and Prevention (CDC). The source was home-grown maize stored damp. People who kept maize inside the home rather than in an outside granary had 12 times the odds of illness, and those who stored damp maize 3.5 times. Household samples held up to 8,000 parts per billion (ppb) of aflatoxin; the Kenyan and US limit is 20 ppb.
LessonDrying grain fully and keeping it dry in store protects lives, not only the harvest.
Credit
Kenya Ministry of Health and CDC investigators
2007
Airtight bags stop storage insects
From 1987 a Purdue University team led by entomologist Larry Murdock worked with farmers in northern Cameroon on storing cowpeas in three plastic bags, one inside another. The sealed bags are hermetic, meaning airtight, which limits how long insects inside can survive. In 2007 the Bill & Melinda Gates Foundation funded Purdue Improved Cowpea Storage (PICS), a five-year, US$11.4 million project designed to reach 3.7 million farm families in ten West and Central African countries. In a 2023 test of maize stored for six months at three poultry farms in Ghana, Opoku and colleagues counted 161 insects in samples from ordinary polypropylene sacks against 7 in PICS bags.
Credit
Larry Murdock and Purdue University; Bill & Melinda Gates Foundation
2007
Beetles outlast phosphine
Phosphine is a poisonous gas released from tablets or cylinders and used worldwide to kill insects in stored grain; since methyl bromide was phased out, it is, in the words of the paper below, the only economically and environmentally viable fumigant for routine use. In 2007 a population of rusty grain beetle (Cryptolestes ferrugineus) collected at a central store at Edgeroi, New South Wales, proved about 1,450 times more resistant than susceptible beetles over a 72-hour fumigation. Ramandeep Kaur and Manoj Nayak of Agri-Science Queensland call this the highest resistance detected in any stored-product insect. They worked out new doses: at 25 °C, 1.0 milligram per litre (mg/L) of phosphine must be held for 20 days, or 2.0 mg/L for 10 days, to kill 99.9% of the resistant strain, and they proposed these as changes to Australia's registered rates.
LessonRelying on one fumigant breeds resistance, and higher doses only work in storage sealed well enough to hold the gas, which large bunkers and old leaky silos are not.
Credit
Manoj Nayak and Ramandeep Kaur, Agri-Science Queensland; Plant Biosecurity Cooperative Research Centre
2023
Resistance counted worldwide again
The only worldwide survey of phosphine resistance was made in the 1970s. Laura Machuca-Mesa, Leonardo Turchen and Raul Guedes gathered the 46 studies published from 1975 to 2021, covering 980 populations of 13 stored-product insect species. They found 72.96% of the populations resistant, and lesser grain borers on average 73 times more resistant than susceptible ones. Reports came mostly from four countries: Australia, Brazil, Greece and the United States. They published their data with the article.
LessonMost grain-growing countries do not publish resistance tests, so how well phosphine still works there is unknown.
Credit
Laura M. Machuca-Mesa, Leonardo M. Turchen and Raul Narciso C. Guedes
Watch for
Camera probes count insects
AIVision Food Inc., founded by Zhongli Pan, an adjunct professor at the University of California, Davis, says its patented SmartProbe, a probe trap with a camera, a collecting chamber and temperature and humidity sensors, photographs its catch on a schedule, counts the insects with AI and sends alerts over WiFi. Pan showed case studies from California rice stores, including the Butte County Rice Growers Association, and claimed a 74% cut in chemical cost for disinfestation through early detection.
Achieved whena university or store operator outside the company publishes SmartProbe counts and first-detection times against hand-counted probe traps and grain samples in the same stores.
Claimed by
Zhongli Pan, AIVision Food, at the IAOM Southeast Asia Conference and Expo for the flour-milling trade in Manila, October 2025
Open
01
Count catches from a photo
A free phone tool that counts and names the insects in one photo of a probe-trap catch tipped onto a white tray, tested against hand counts at working grain stores. The tester only photographs catches that store staff collect in their usual rounds and never enters a bin.
The briefCredit · unclaimed02
A living resistance map
An open dataset and map of every published phosphine resistance test on stored-grain insects, one row per tested population with place, year, species, test method and result, updated every quarter, with the countries that have no tests shown as gaps.
The briefCredit · unclaimed
Open 01 · the brief
Count catches from a photo
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A free phone tool that counts and names the insects in one photo of a probe-trap catch tipped onto a white tray, tested against hand counts at working grain stores. The tester only photographs catches that store staff collect in their usual rounds and never enters a bin.
Why it is still open
The US Agricultural Research Service (ARS) has published, free of copyright, images of the five main stored-grain beetles with a model that named the species correctly 96.57% of the time in testing, but not counts of real trap catches. From September 2025 to September 2027 an ARS cooperative project in Manhattan, Kansas, is adapting traps to take pictures and comparing them with checks by a trained technician in semi-field tests, not with phone photos at working stores. The best open image set from outside ARS, by Song-Quan Ong and Toke Thomas Høye in 2024, covers two species on sticky traps photographed in a studio box with one phone model. SmartProbe counts insects inside a commercial trap, with no independent accuracy figures found, and the EGPIC's infrared counts missed hand counts by 31.7% on average. A search found no public test of phone-photo counts against hand counts for probe-trap catches.
catalog.data.govscientificdiscoveries.ars.usda.govars.usda.govpmc.ncbi.nlm.nih.goviaom.orgebi.ac.uk
Built with what exists
A phone camera; a white tray and a printed scale bar; the ARS five-beetle images and Ong and Høye's images as starting sets; Label Studio for labelling; an Ultralytics YOLO detection model; standard grain probe traps.
First step
Ask one store to hand over its probe-trap catches for a month, and photograph and hand-count 50 of them.
Done looks like
300 catches from at least three stores and two grains, each photographed and hand-counted by a trained person; the tool's count within 10% of the hand count on 90% of catches; the species named correctly for 90% of insects of the five main species (lesser grain borer, rusty grain beetle, red flour beetle, maize weevil and saw-toothed grain beetle); images and labels published.
Open 02 · the brief
A living resistance map
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open dataset and map of every published phosphine resistance test on stored-grain insects, one row per tested population with place, year, species, test method and result, updated every quarter, with the countries that have no tests shown as gaps.
Why it is still open
Machuca-Mesa and colleagues' 2023 meta-analysis stops at 2021 and comes as a spreadsheet attached to the article; it found testing confined mostly to four countries. New surveys keep appearing, such as tests of 28 lesser grain borer and 34 red flour beetle populations from the United States and Canada published in 2025. A search found no public map of resistance that is kept up to date.
Built with what exists
The 2023 spreadsheet as the base; the Europe PMC and OpenAlex search APIs to find new papers; a web agent such as Browser Use to gather national reports and theses (see the Note "Deleting 80% to start over"); a language model to draft rows for a person to check; QGIS or Leaflet for the map; Zenodo for versioned releases.
First step
Download the 2023 spreadsheet, geocode its 980 populations, and list the papers from 2022 on that a Europe PMC search for "phosphine resistance" returns.
Done looks like
All papers from 1975 to mid-2026 entered, each row linked to its source; the 1975 to 2021 rows reproduce the meta-analysis's 72.96% resistant share within one percentage point; four quarterly updates published.