Deciding to Grow: Agriculture and forestry in a changing environment.
PI: Dr Maria Christodoulou (Department of Statistics, University of Oxford)
Co-Is: Dr Julia Brettschneider (Department of Statistics, University of Warwick); Prof David Steinsaltz (Department of Statistics, University of Oxford)
Project Overview: Agricultural research has great potential as a testing ground for decision theory methodology, yet examples of this are rare. Within the growing season the agricultural decision problem can be reduced to the interaction between two principal players – the grower and the plant – who in combination respond to a third player, the environment.
A plant (e.g. an apple tree) invests in various “behaviours”, depending on its expectation of future conditions based on environmental cues, and the grower aims to manipulate these behaviours by intervening (e.g. irrigating, fertilising, spraying, or pruning) at crucial points to direct the growing process toward a desired outcome, such as higher yield or improved sugar/acid ratio. Some of these interventions, however, may arrive too late. Furthermore, whereas the plant responds individually to conditions, the grower manipulates outcomes by applying a one-size-fits-all approach to the whole orchard. This ignores the impact of well-established effects, such as spatial inhomogeneity of temperature due to relative location, presenting an opportunity for improving efficacy through gradient and block treatments.
In this multidisciplinary project we will explore the potential of modern decision theory for optimising production and sustainability in agriculture, using established data sets. Conversely, we seek to develop decision-theory methodology along new lines inspired by the requirements of this domain. By exploring both the available growing protocols and the resulting yields from specific agricultural decision trajectories we aim to model decision trees and the strategies followed by the different players within the growing season, and to extend these to describe optimal growing strategies based on desired outcomes.