Skip to main content Skip to navigation

Rapid, continuous monitoring for farming and ecological diversity best practice for hedgerows and field margins: an artificial intelligence approach

Primary Supervisor: Dr Ed Harris, Department of Agriculture and Environment

Secondary supervisor: Dr Tom Pope

PhD project title: Rapid, continuous monitoring for farming and ecological diversity best practice for hedgerows and field margins: an artificial intelligence approach

University of Registration: Harper Adams University

Project outline:

    The aim of this project is to overcome the challenges of monitoring and measuring agriculture boundary habitat quality (e.g. hedgerow, field margins) and associated ecological diversity by leveraging artificial intelligence and remote sensing data rapidly and at different scales. The approach will be to capture artificial intelligence training data for boundary habitats across a range of representative samples associated with a range of practice and management. Sampling will be conducted according to established standards and will be geolocated and ground-truthed using unmanned aerial vehicle photographs. First, hedgerow and field boundaries will be classified on the ground according to quality or management in order to create a training data set. The labelled photos comprising the training dataset will be used to create a predictive model to classify quality based on remote sensing satellite data features. Thus, boundary habitat will be identified and classified according to an existing categorical scheme to facilitate ongoing continuous monitoring. Second, a statistical model between boundary habitat and ecological diversity for indicator taxa will be made. The resulting artificial intelligence will rapidly and continuously identify density and variation in boundary habitat quality at different spatial scales, from single fields to large regions, identifying areas of good practice and areas with an opportunity for improvement. Critically, this will also generate testable predictions for the association between farm practice and ecological diversity and will also be useful to inform, e.g., the Environmental Land Management scheme for the Sustainable Farming incentive in the UK as well as regional and local conservation effort.

    Human impacts to natural habitats must be minimized to achieve sustainability and resilience to environmental change. Field margins and hedgerows are linear features associated with farming practice that comprise an important cultural and aesthetic tradition, while also providing practical benefits such as being boundaries for livestock and land ownership. However, there is a growing body of evidence that such boundary habitat features can also function to bolster biodiversity and ecological resilience by providing wildlife habitat and dispersal corridors.  The challenge to reduce the ecological impact of farming practice must incorporate a greater understanding of the function of adjacent boundary habitat in order to achieve sustainable practice.

    Hedgerow habitat is recognized to provide carbon sequestration, while being associated with pollinator abundance and diversity, and with biodiversity and dispersal in birds and mammals. Likewise, there is evidence that many organisms exploit field boundary habitat for movement, for example hedgehogs, birds and pollinators. Yet, while the animal conservation and ecosystem service benefits of boundary habitat is widely recognized as important, the ecological quality and management of this habitat is extremely variable, highly localized and notoriously hard to measure. Because of this, measuring the functional connection between boundary habitat management and associated biodiversity have proven a difficult challenge. The primary reasons for this are practical and logistic constraints.

    To measure the association between boundary habitat classification and ecological diversity, established methods will be used for training sample sites to capture data for "indicator" taxa for habitat quality and ecological function. Here we will focus on moths and birds, known as bioindicators for other taxa. Moths are a useful group as well because they can be relatively easily sampled, and they have a known function as pollinators; birds also have a known close association with farmland and boundary habitat and can be readily sampled. There are good, long-term biodiversity datasets available to provide context for this study for both taxa. All biodiversity data will be deposited in an established, open data framework. Data will be analysed in a spatial context according to hedgerow quality, land use and in the context of comparable and relevant spatio-temporal data (e.g. using the Global Biodiversity Information Facility).


    1. Coulthard, E., Norrey, J., Shortall, C., Harris, E., 2019. Ecological traits predict population changes in moths. Biological Conservation 233, 213–219.
    2. Kussul, N., Lavreniuk, M., Skakun, S., Shelestov, A., 2017. Deep Learning Classification of Land Cover and Crop Types Using Remote Sensing Data. IEEE Geoscience and Remote Sensing Letters 14, 778–782.
    3. Vickery, J., Feber, R., Fuller, R., 2009. Arable field margins managed for biodiversity conservation: A review of food resource provision for farmland birds. Agriculture, Ecosystems & Environment 133, 1–13.

    BBSRC Strategic Research Priority: Sustainable Agriculture and Food:Plant and Crop Science

    Techniques that will be undertaken during the project:

    Conservation, ecology, farming practice, artificial intelligence data science, crop science, land management, biodiversity, deep learning, remote sensing, GIS, Environmental Land Management scheme

    The techniques involved in this project will include ecological survey of field margin and hedgerow habitat, and biodiversity sampling, analysis of spatial data using GIS tools, and using deep learning neural networks to classify computer vision data for boundary habitat state and quality. The project would be suitable for someone interested or with a background in environmental science, ecology, practical land management, agriculture or in data science. Training will be provided as required in complementary methods.

    Contact: Dr Ed Harris, Harper Adams University