Work Package Lead: Pete Skelsey (James Hutton Institute)

Team: Pete Skelsey (James Hutton Institute); Jim Wilson (SoilEssentials)

This WP aims to describe trends and drivers of Pba incidence in order to produce predictive models that could be used in decision support tools and will utilise a range of available landscape-scale datasets: AHDB potato crop distributions; SPUDS blackleg database, UKMO hourly weather data for GB, Hutton National Soil Archive, Countryside Survey data, Synthetic Aperture Radar SAR soil moisture data, leaf-wetness data, FLN distribution data.

Objective 4.1 - Spatiotemporal analyses using ArcGIS

Preliminary investigations with the national potato crop inspection data (SPUDS) from SASA (including blackleg incidence), and the national potato crop distribution data (IACS) have revealed statistically significant spatial clusters of Pba incidence within potato crop production areas. Initial modelling results with other landscape-scale datasets available to the project suggest an association of blackleg affected crops and potato crop area, field generation, entry class, mother crop status, soil class, and potentially other soil characteristics / meteorological variables. Fundamental to this proposed work are two unique nematode resources that comprise 1000s soil samples giving FLN abundance, location and in some cases agronomic data.  

This objective will analyse trends in blackleg incidence in both space and time using ArcGIS mapping space-time pattern mining tools to determine the locations and timing of hot- and cold-spots. These analyses will be repeated with the FLN data, and results for blackleg and FLN compared.

Objective 4.2 - Climate change risk assessment

This deliverable will use CLIMEX models to project the effect of climate change on future distributions of Pba and FLN in GB, building upon the spatiotemporal analyses of Pba and FLN distributions to quantify effect of climate on future blackleg incidence in association with FLN presence and produce a preliminary climate change risk assessment of blackleg-causing bacteria in GB.

Objective 4.3 - Machine learning - landscape-scale datasets

This objective will identify drivers of blackleg outbreaks and to develop a model for forecasting the risk of blackleg occurrence associated with any stock and planting location. A variety of models including supervised machine learning techniques will be applied to the data to determine the relative importance of soil, crop, pathogen, cultivation, climatic, landscape and FLN risk factors for blackleg incidence, and to automatically generate predictive models.

Objective 4.4 - Decisions support Tool

This objective will integrate data from the other work packages into a simple decision-support tool for predicting blackleg risk and suggesting risk management / avoidance strategies. All the drivers of incidence / severity identified in the project will be ranked as LOW / HIGH risk for blackleg, e.g. (i) presence / absence of FLN species responsible for Pba ingress into plants; (ii) irrigation scheme; (iii) previous crop history; (iv) soil, crop, pathogen, cultivation, climatic, and landscape factors. A look-up table will be created suggesting risk management strategies for each combination of drivers.