Skip to main content

The Exhibitor Blog

Subpage Hero

05 Feb 2021

Seeking a better model for wheat bulb fly prediction

As featured in Arable Farming Magazine

Seeking a better model for wheat bulb fly prediction

by Arable Farming Feb 2021 Issue

Research aimed at forecasting wheat bulb fly attacks by modelling rather than egg count surveys is showing encouraging results. Andrew Blake reports.

With spray control options withdrawn and seed treatment best confined to later sowings, preventing wheat bulb fly larvae from hitting yields is becoming more of a challenge.

The pest can affect all cereals, except oats, and the currently accepted way to assess the threat and the need to apply the sole approved seed treatment, Signal 300 ES (cypermethrin), is by counting eggs which the fly lays in bare soil.

Based on those numbers AHDB offers treatment advice (see panel).

The latest research project*, led by ADAS entomologist Steve Ellis, continues to assess bulb fly risk by egg-counting but is also investigating modelling as an alternative to that laborious and expensive exercise.

Dr Ellis says: The basic reason for the work is to provide information on egg numbers so farmers can decide whether to apply a seed treatment if they plan to sow from November onwards.

Less effective

Treatments are less effective for earlier sowings because they run out of steam before the main egg hatch period which is usually in January/February. This seasons egg counts suggest a low risk year, as have been the past several years (see graph); but the impact of previous crops in the sampled fields (15 each in the East and North where the pest is most prevalent) is worth noting, says Dr Ellis.

Over all sites, the highest risk was after French beans with a mean of 535 eggs/square metre, although only two sites were sampled, one of which had a low count.

The next highest risk was after sugar beet with a mean of 175 eggs/sq.m across eight sites sampled.

Average egg numbers in the East were higher than in the North with 173 and 111 eggs/sq.m respectively.

The counts in the East were heavily influenced by two very high counts after sugar beet and French beans, as well as by a high count also after sugar beet. Overall, the results showed 7%, 3%, 13% and 77% of fields in the very high, high, moderate, and low risk infestation categories respectively identical to the breakdown in 2019, he adds.

Project details

• AHDB project 21120003 Autumn survey of wheat bulb fly incidence

• September 1, 2019 October 31, 2021

• Funding: £32,000 (all AHDB)

Treatment thresholds

• Early-sown crops (before November) are unlikely to benefit from seed treatments and excess shoot production makes these crops more resilient.

However, bulb fly pressure of 250 eggs/sq.m may impact yield

• For late-sown crops (November-December), consider seed treatment where populations exceed 100 eggs/sq.m

• For very late-sown crops (January), consider seed treatment irrespective of population size (unless no eggs are present) Note: Treated seed must not be sown after January 31

Modelling update

In 1993 a risk prediction model (Young and Cochrane) using data from 1952 and based on July temperature, August rain days and October rainfall in the previous year was found to be only 59% accurate.

A key objective of the latest project has been to test that model and update it by combining 2005-2019 data with the original and incorporating a wider range of weather influences, says Dr Ellis.

These include preceding September sun days, preceding October rain days, January mean temperature, January frost, April and May maximum temperatures, April rainfall and July minimum temperature.


Briefly, we took a range of meteorological parameters and put them all into a predictive model, says his ADAS colleague Dr Daniel Leybourne.

We then refined it through several stages until we arrived at the one with the best predictive power.

At each stage we removed the met parameter that was least significant until further refinement no longer improved the models predictive accuracy.

This updated model has a predictive power of 70%, an 11% increase when compared with the original Young and Cochrane model, he says.

Dr Ellis adds: Improving the predictive power of decision support models is likely to increase confidence in their findings and boost uptake by farmers and agronomists. One surprise from this seasons field sampling has been the average 306 eggs/ sq.m from organic land but only 21 eggs/sq.m from mineral soils, says Dr Ellis.

Although the mean in organic soils was heavily influenced by one very high and one high count after sugar beet plus one very high count after French beans, the results suggest that the potential for bulb fly damage in eastern England in organic soils is greater than in mineral ones.

Reliable model should help manage the pest

Reliable model should help manage the pest

Yorkshire-based agronomist and former chairman of the Association of Independent Crop Consultants Patrick Stephenson is looking forward to the projects completion.

He says: If we can have a reliable bulb fly model that produces a guideline early in the season, it should play a key role in planning and managing.

Growers and advisers are only now starting to come to grips with the loss of effective chemistry against this pest.

Weve been lucky because the weather in recent years hasnt favoured it.

Fields most likely to be at high risk, for example late lifted beet or potatoes, havent been drilled until spring.

In many cases the solution is simple dont grow wheat in a high-risk situation.


Natural predation accounts for about 70% of all eggs, he adds.

So, egg numbers have always been a strong indicator of economic risk.

But gathering soil samples is labour intensive and slow and now with limited chemical control you have to ask is it worth it? The most common cultural control method is to sow early, which in the North is before the first week of October, he says.

Ensure good fertility for rapid emergence and aim to produce sufficient tillers to withstand losing some.

If you must sow late, consider using a seed dressing which will give partial protection, be prepared to apply nitrogen early in spring and roll the crop. Ensuring that soil is not left bare from mid-June to September, avoiding deep drilling and increasing the seed rate can all help, he adds.

Biocontrols may yet form part of the counter-measures, with beetle species Aleochara bipustulata and A.laevigata having been identified as bulb fly parasites; but as is the case with many such solutions, deploying them may prove impractical.

View all The Exhibitor Blog