Science1 publisher2 min readPublished
USDA is placing sterile screwworm flies where its predictive models point
USDA's screwworm response now takes its placement orders from habitat and weather models built by its own livestock insects lab in Kerrville, Texas. The agency did not report how accurately those models forecast.
The Scientist · Science desk
What happened
- New cases of New World screwworm have recently surfaced in the United States, according to USDA's Agricultural Research Service, which is running predictive models to forecast where the fly could establish next.
- USDA says model output now sets the locations, the timing and the number of sterile flies released under the Sterile Insect Technique.
- The models take in habitat suitability, temperature and precipitation, and the record of past outbreaks along with past control efforts, including the ones that did not slow the pest.
- The same models are being used to judge how well current and planned prevention strategies slow the fly's movement, so USDA can adjust them.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- decision Release sites are a forecast output now, so a wrong forecast costs flight hours and a missed release window.
- constraint Model-chosen surveillance narrows where new detections can come from, so part of the evidence that would test the map comes from the areas the map picked.
- exposure Ranchers in country the models rank low get fewer traps and fewer sterile flies, so they will find their own cases first.
- precedent The same platform that chose the prevention strategies is now grading them.
Sterile flies work only when they are in the same place, at the same time, as the wild ones. USDA's own account puts that condition first. The success of the Sterile Insect Technique depends on knowing when and where to release the flies, and releases go into areas where the pest is present or likely to spread [17][9]. Scott Hutchins, USDA undersecretary for research, education and economics, said, "Modeling its movement and using this information for optimal placement maximizes our impact and hastens eradication" [10]. The agency describes the whole approach as one that lets it act earlier and put resources where they are most needed to protect livestock, pets and wildlife [2].
The model class is familiar to any ecologist. Habitat suitability, weather, and a record of where the animal has already been found are the standard ingredients of a range map. USDA says the analysis runs through advanced analytical tools and artificial intelligence [8]. Detection records carry a known bias: an insect is recorded where somebody looked for it. Once the same platform picks the trap lines and the surveillance areas, the next round of records is concentrated where the current map already points [16]. The modeling should still happen. Validation has to come from cases the program did not choose to look for, such as detections reported by producers in low-ranked country.
The announcement does not include a count of confirmed US cases or the number of sterile flies going out [15]. A suitability map can rank counties correctly and still lose on lead time, meaning the days between a fly establishing in a herd and a case being confirmed. Lead time is the capability a forewarning system is meant to deliver. Testing it takes dull bookkeeping: release sites and dates on one side, confirmed detections and their dates on the other.
Kim Lohmeyer directs the ARS Knipling-Bushland Livestock Insects Research Laboratory in Kerrville, Texas. Lohmeyer said the models "are already being used to select areas for increased livestock and wildlife surveillance, adult fly trapping and implementing SIT" [11]. Lohmeyer also called a forewarning system "a game-changer in our fight to respond quickly to this urgent pest" [12]. In the same statement, Lohmeyer said the models "cannot predict every scenario," a sentence the published text breaks off in the middle of [13].
What to watch
- Whether USDA publishes how many confirmed detections fell inside high-suitability areas before anyone found them there.
- Whether release sites, dates and trap-catch records become available at county resolution, so placement can be checked against later cases.
- A confirmed detection in country the models ranked low. That would push sterile releases outside the maps now guiding them.