Shading in winter barley variety trials
Plant breeding produces a steady stream of new candidate varieties of the major crops in order primarily to increase yield but also to manage threats posed by factors such as disease. They are evaluated firstly for regulatory approval to market, and then subsequently for possible inclusion on the Recommended List of varieties available to growers. Validity of decision making relies on the underlying biological and statistical assumptions in the analysis. The increasing prevalence of 6-row hybrid varieties led to a review in the specific case of winter barley.
Field trials comprise up to 90 varieties sown side-by-side in replicated, small plots (approx. 10m x 2m). Growing varieties alongside each other means varieties are grown under similar field and environment conditions but this approach differs from standard agricultural practice in which only a single variety of a crop tends to be sown in a field.
Variety trial yields are, therefore, potentially susceptible to the impact of inter-plot competition, which can enhance or depress individual plot yields. Shading of shorter varieties by taller neighbours is a known source of potential competition effects in crop trials.
For winter barley, not only is there a wide height range across varieties but varietal height tends to be strongly associated with three genetic groups: two-row conventional; six-row conventional; and six-row hybrid varieties. The key challenges were to investigate the presence and magnitude of shading effects and then to consider ways to mitigate its impact.
Starting from the standard mixed model approach for analysing individual trials laid out as incomplete block (α-lattice) designs, we modified it to fit a series of models, each reflecting a specific, postulated biological model driving shading effects.
In all cases we assumed that the impact of shading on an individual plot’s yield is the outcome of a competition solely between the shading effect exerted by the variety in the plot itself and those exerted by the varieties in immediately neighbouring plots. In addition to comparing relative goodness of fit of the models for each trial, we were able to compare over-trials estimates of variety means from each of these models with those obtained from the standard approach ignoring competition.
The models provided strong evidence of shading effects, whose magnitude meant that some form of mitigation was needed. We proposed mitigation by statistical design as mitigation by statistical analysis was not feasible given the imprecision of shading effects estimates and uncertainty over the most appropriate underlying shading model. The other option of extensive use of buffer plots or larger plots would prove too costly.
From Autumn 2023 composite designs will be adopted for winter barley trials. Each replicate will comprise three distinct buffered sections, one for each of the three genetic groupings. BioSS is currently modifying trial plan generation software to meet the specific requirements of these trials.
This work was carried out at BioSS by Ian Nevison in collaboration with Adrian Roberts and Tess Vernon. It was funded by the Agriculture and Horticulture Development Board. We are grateful to the British Society of Plant Breeders Limited for supplying data used in this evaluation.