Abstract/Summary

Soil is a complex, variable, living medium essential to supporting life on earth through the provision of a range of ecosystem services, yet it is a non-renewable resource. Soil can be identified as the basis for food production, providing us with clean water, hosting biodiversity, cycling nutrients, and buffering against climate change. Anthropogenic pressures to increase productivity to enable food security is damaging these soil systems. The current environmental boundaries for soil systems are being transgressed, leading to the degradation of soil systems globally. This is particularly apparent in peatlands drained for agriculture, where departure from their natural state has caused intense degradation. Peatlands are essential to UK natural capital, food production, and ecosystem service provision. Previous attempts to create metrics to assess soil health and functioning have focused primarily on mineral soils and are not appropriate for assessing the soil health of drained agricultural peatlands. Specific tools to assess peat health were lacking and thus needed to be developed. Here I describe the development of two tools to enable assessment of the health of lowland peat systems using simple indicators that allow farmers to benchmark, compare, and sustainably manage peat health. The first tool created a minimum indicator set to classify peat health though Principal Component Analysis, leading to the development of an Additive and Weighted Peat Health Index approach. These indices were able to effectively distinguish deep peat from wasted peat, as identified by farmers. Additionally, the Peat Health Indices revealed that healthier fields required less farm inputs, indicating a better functioning system. The second tool developed was a Bayesian network. This tool incorporates probability distributions in assessing peat health, enabling the direct assessment of ecosystem uncertainty, and providing an estimated distribution of peat health given the observation of simple on-farm indicators. The network was developed through expert opinion and use of the ECOSSE biogeochemical model. The network was validated through k-fold cross validation, scenario analysis and expert evaluation. This thesis demonstrates the development and application of health assessment tools for drained agricultural lowland peat using easily measurable soil properties. We anticipate these tools to be a starting point for the assessment of peat health across the East Anglian fenland region and lead to the development of a national monitoring network using the Bayesian network approach.