How we decide which RL trials make the cut

Paul Gosling, who manages the Recommended Lists for cereals and oilseeds (RL), explains how we decide which trial plots make the grade and provides an update on our winter wheat on trial site in Norfolk (which was harvested last week).

The Terrington St Clement winter wheat on trial site hosted one of 31 dedicated UK winter wheat fungicide-treated yield trials sown for harvest 2025 (and a fungicide-untreated trial).

On this site, which is close to the Wash, we recently harvested 35 recommended and 15 candidate varieties, replicated three times (there were 150 fungicide-treated variety plots).

The Norfolk trials are part of a UK network that comprises several hundred trials each year (350 across all crops), which are nurtured through many seasonal challenges by our RL trial operators.

It requires fine-tuned logistics to ensure that all trials plots (about 25,000 of them) are harvested in a timely manner, with planning being key – especially in a challenging year like 2025.

In early summer, members of the AHDB field team visit all sites to score the quality of every plot. They start in Cornwall and Devon and finish in Northern Ireland and Scotland. It takes about three weeks to complete the official inspections for the RL cereal trials. They also inspect the statutory Variety Lists (VL) trials, on behalf of Defra, and stop off to demonstrate at summer variety open events along the way. It is a busy time.

The quality scores, which underpin the trial validations, are based on plot condition and use a 1 to 4 scale:

1. Exclude
2. Of concern
3. Slight concern
4. Acceptable quality

We exclude yield data from plots that do not meet the RL’s strict quality standards. There are many reasons why we issue a score of 1, such as drought or waterlogging resulting in poor growth.

The RL data team carefully scrutinises all yield data and checks for anomalies, paying very close attention to data that did not record acceptable quality (scores 2 and 3).

Initially, we conduct statistical analyses on each trial separately to detect trends that are unlikely to be associated with a varietal effect.

These trends can be across the trial or confined to patches of variable yield that may indicate environmental effects that bias the trial.

The team always cross-references findings with RL trial operator observations to help fully understand any variation or anomalies.

Sometimes, we do additional statistical analyses to look for correlations between yield and some measurable factor, such as bird damage scores or combine-loss data.

We also look at how the trial compares with the UK and regional data sets.

These analyses and observations may result in the exclusion of more plots or even the rejection of whole trials for yield.

Ultimately, the RL field team decides whether a trial is valid.

The RL team also performs an over-year analysis of yield data, validates lodging and quality trials, analyses agronomic data (such ripening, grain quality and height) and calculates the disease ratings – there is a lot to do.

This year, the dry conditions impacted many commercial crops and trials across the UK. However, the Norfolk trial site is on deep, silty clay, which protected the winter wheat, with plot scores generally good and just a few small gaps recorded in some plots.

The good pre-harvest outlook was backed up by the yield data for the site – the control varieties yielded an average of 13.86 t/ha, with some plots yielding above 15 t/ha. This is far higher than the over-trial, over-year (2021–25) mean, which is about 11.2 t/ha.

This site has a high yield potential, but small-plot trial yields are often higher than whole-field commercial yields.

Fortunes at some other sites have not been as good. For example, the dry conditions at a winter wheat trial site in Suffolk (second wheat) led to poor tillering, gappy and uneven plots across the site, with six plots excluded – mainly recording scores of 2 or 3.

This year, we may need to exclude more trials than usual at a late stage due to the dry conditions. This is because small differences in soil moisture can introduce unacceptable levels of variability across trial sites.

The validated data, which is published on the RL pages (as harvest results), is used by the RL Crop Committees to determine which varieties should make it onto the next RL edition (which will be the subject of a future blog post).