Spotted wing drosophila is a pest species, recently introduced to the UK, impacting on soft fruit production (raspberries, strawberries, and blackberries). An emergence experiment was designed to collect genetic data from individuals that feed on either wild or cultivated fruits to understand whether genetic differentiation is present within the pest populations, associated with different feeding behaviour. A machine learning approach to analysing the genetic data collected from this panel of individuals has successfully been piloted to explore the presence of genetic differentiation. At this stage, the genetic data collected from this appropriately designed experiment turned out to lack power to detect differentiation. However, we obtained a robust pipeline combining a well-designed experiment with a cutting-edge statistical approach, and we aim to collect more in-depth genetic markers from the same panel of individuals to increase our detection power. 

Fruit flies are a very common group of insects belonging to the melanogaster species group. They usually lay eggs in rotten fruits of vegetables and are very common in households. Among those, the spotted wing drosophila (a species called Drosophila suzukii) shows a slightly different behaviour as it lay eggs in ripening, undamaged fruits of a wide range of cultivated and wild soft fruits, including raspberries, strawberries, and blackberries. This species is native of the southern-east Asian continent but has spread to most of the rest of the world over the past two decades. It was firstly recorded in the UK in 2012, and it was probably introduced as a result of worldwide trade. The spotted wing drosophila can cause serious damage to cultivars as the larvae feed on the pulp of the infested fruits. As the species is adapted to survive both on wild and cultivated variates, there is much interest in understanding whether there are differences in the genetics of individuals preferentially feeding on either wild or cultivated fruits. This understanding could be used to develop targeted techniques to seek to eradicate subpopulations adapted to the cultivars while preserving the naturally occurring individuals feeding on wild fruits. 

In order to assess genetic differences between spotted wing drosophila individuals feeding on the different fruit cohorts, researchers in the Ecological Sciences group at JHI have collected infested fruits from both the wild and from cultivars. They used these in an emergence experiment, where they collected flies emerging from each type of fruit. The origin of the collected flies is known, and the researchers genotyped both cohorts of D. suzukii individuals using a panel of 8 genetic markers. Data from this carefully designed experiment were analysed using a machine learning approach, training the algorithm to learn which genotypes belong to wild-feeding or cultivar-feeding individuals. If sufficient differences actually exist in the training set, the algorithm should be able to assign each individual to its correct group of origin. In this analysis, we failed to find any evidence of a genetic difference between these two cohorts of individuals. However, the explanation most likely lies in the limited number of loci collected and the (also limited) number of individual flies sequenced (only 24 per category). The design of the experiment described above is well suited to the needs of a machine learning approach with supervised learning (e.g., when a training set of data with a known characteristic is used to teach the algorithm what to look for in other data sets of the same kind), so on the basis of this experience, we have strongly recommended that SEFARI scientists collect much more extensive genetic data and/or data from more individuals. We anticipate that increasing the power of the study in this way will support future successful detection of differences in genetic background between wild associated and cultivar associated individuals of D. suzukii. 

This work was done in collaboration with Ali Karley and Gaynor Malloch at The James Hutton Institute and was funded under the Underpinning National Capacity element of the Scottish Government's Strategic Research Programme for environment, agriculture and food.