Human vs machine: The quest for BYDV monitoring accuracy in cereals

Henny Lowth provides an update from Strategic Cereal Farm East, where learning from successes and failures underpins the management direction.

The second Strategic Cereal Farm East (2023–2029) is a couple of seasons in and its trials are in full swing.

At the start of the programme, host farmer David Jones (Morley Farms Ltd), put the reduction of inputs at the top of the farm’s wish list.

The £45/ha annual payment to avoid insecticide use on arable crops under the Sustainable Farming Incentive (SFI) certainly adds incentive to take more risks.

One trial area examines the use of a Barley yellow dwarf virus (BYDV) resistant winter wheat variety and decision support systems (DSS) to make sure aphid-vector sprays only go on when it pays.

David believes it is possible to omit aphid sprays in many years, with the right evidence to hand.

Although pyrethroid sprays are often perceived as cheap, they need to deliver a return on investment and be limited to avoid negative impacts (e.g. on non-target organisms and insecticide resistance risks).

Recently, I caught up with our research partners – Duncan Coston (ADAS), Jack Poulton (Niab) and Tom Pope (Harper Adams University, HAU) – to get an update on the BYDV trials.

The research includes tests of ways to quickly and accurately assess BYDV symptoms, which is critical to gauge treatment impacts, as yield responses are often tricky to measure.

Classic BYDV symptoms include crop stunting and leaf yellowing (sometimes with red tipping of upper leaves).

But symptoms can be subtle, often confined to patches and mistaken with other symptoms of stress. It can make assessment challenging, even for seasoned crop scouts.

This is why the team is trialling light-capture technologies, which have a broader spectral range than the naked eye, to spot disease:

• Normalized Difference Vegetation Index (NDVI), which combines visible red and near-infrared (NIR) light to quantify vegetation
• Normalized Difference Red Edge (NDRE), which uses a band that sits on the transition region between visual red and NIR light to monitor changes to leaf colouration (chlorophyll levels), which can be associated with BYDV symptoms

Such technologies may even be able to detect stress indicators long before symptoms are apparent to the naked eye.

Certainly, they will help address a general challenge associated with traditional disease assessments: the variability introduced by humans.

Even highly trained assessors can disagree about disease scores. Machines have the potential to truly standardise disease assessments.

ADAS oversees the traditional crop scouting. The team systematically walks the fields (in a grid pattern) that pit a winter wheat variety marketed as exhibiting resistance to BYDV (RGT Grouse) against a susceptible variety (KWS Dawsum). It takes about a day in the spring to quantify BYDV symptoms.

Recent autumn conditions have seen relatively low aphid pressures in the region. Based on BYDV monitoring results from the nearest suction trap (Broom’s Barn, near Bury St Edmonds), we know about a fifth of aphids carry BYDV. However, no visible BYDV symptoms were seen this spring (in either treatment).

Niab has operated the drones to capture high-quality, overlapping images that are being used to build comprehensive maps of crop conditions at the plot- and whole-field scales.

One of the biggest barriers has been the substantial computational power required to process large volumes of data.

The drones have also battled with the weather (rain and wind gusts) and even solar flare activity, which can disrupt GPS signals.

Additionally, battery life has limited flight duration to about two hours.

We plan to release the results of the drone assessments later this year.

Tom Pope (HAU) believes it is essential to improve the disease assessment process, whether by conventional or innovative means.

Not only for research purposes but to provide post-season evaluation tools to gauge what does and does not work – information that can help guide decisions for future crops.

Whether by human or machine, accurate BYDV identification remains a challenge due to symptom overlap with other crop health issues.

Currently, it is important to confirm the presence or absence of BYDV in laboratory tests of plant tissue samples.

While crop assessment technology will become more mainstream, all three experts agree that a combination of experienced advisors and technology will provide the most powerful disease-management approach for the foreseeable future.

Since 2018, we have funded virus screens on aphid samples (bird cherry-oat and grain aphids) caught each week during the autumn from selected suction traps across England.

Results from previous autumns show that most aphids do not carry viruses associated with BYDV symptoms. For example, data from the last four autumns (2021–24) reveals that the mean proportion of viruliferous aphids never exceeded 30% (some weekly results exceeded this figure).

Virus monitoring is due to commence again in early autumn.

Our BYDV management tool can guide spray timings (it is one of the decision support systems used at Strategic Cereal Farm East).

It estimates when the second aphid generation (most associated with BYDV spread) is likely to be present, based on accumulated daily air temperatures.

The temperature sum starts at crop emergence or after a spray. The tool will include 2025 weather data from early autumn.