One of the major challenges for agriculture today is to manage soil properties and their spatial distribution to optimize productivity and minimize environmental impacts, such as diffuse pollution. To identify best management practices, the effects of different agricultural management practices on pollutant sources, mobilization, transfer and delivery to water bodies need to be understood. Grasslands managed for dairy and meat production, despite being widespread, have received less research attention than other agricultural land uses. Therefore, this thesis studies the effects of different grassland management practices on soil properties and their spatial distribution and the mobilization and delivery of multiple diffuse pollutants. As a grassland case study, monitoring for this thesis was conducted across three fields (6.5 – 7.5 ha) on the North Wyke Farm Platform, a grassland experimental farm in the UK. First, the effects of permanent grassland management (permanent for at least 6 years, but different grassland management > 6 years ago) were characterized as a baseline, followed by quantifying the short-term effects of ploughing and reseeding of permanent grassland fields. Throughout those management periods, i) a range of soil physical (bulk density [BD]) and chemical (soil organic matter [SOM], total N [TN], total phosphorus [TP], total carbon [TC]) soil properties and their spatial distribution were sampled and analysed by geostatistics, and ii) hydrological characteristics and multiple pollutant fluxes (suspended sediment [SS] and the macronutrients: total oxidized nitrogen-N [TONN], total phosphorus [TP], and total carbon [TC]) were monitored at high temporal resolution (monitoring up to every 15 minutes). The permanent grassland fields (or areas within fields) can be considered to be functioning differently. Past management legacy (more than 6 years ago) has affected soil properties and their distribution with subsequent effects on sediment and macronutrient delivery from the fields to surface waters. Overall, permanent grasslands were found to contribute significantly to agricultural diffuse pollution. The estimated erosion and macronutrient losses were similar to or exceeded the losses reported for other grasslands, mixed land use and even arable sites, and sediment and TP concentrations exceeded those recommended by EU / UK water quality guidelines. Ploughing and reseeding did not homogenize spatial variation and did not override past management effects. Long-term management differences affected soil properties and altered soil processes, so that the fields subsequently responded differently to ploughing and reseeding. All nutrient concentrations were significantly reduced in the older grassland field (no ploughing for 20 years), but not in the younger grassland field (no ploughing for 6 years). Ploughing and reseeding significantly accelerated the losses of sediment and macronutrients and sediment, TP and TONN exceedance frequencies of EU / UK water quality guidelines increased. Additionally, ploughing and reseeding caused a shift in the relative importance of nutrients, by increasing the relative importance of N. Such large sediment and nutrient losses from intensively managed grasslands should be acknowledged in land management guidelines and advice for future compliance with surface water quality standards. The between-field and within-field variation highlights the importance of baseline characterization and paired catchment studies. The long-term effects of management still acting on soil properties and subsequently water quality indicates how long it may take to see soil and water quality improvements after implementing mitigation measures. Therefore, long-term management history always has to be included when interpreting soil and water quality data.

This research work concentrate on developing digital soil maps to support field based plant phenotyping research. We have developed soil organic matter content (OM), cation exchange capacity (CEC), natural soil drainage class, and tile drainage line maps using topographic indices and aerial imagery. Various prediction models (universal kriging, cubist, random forest, C5.0, artificial neural network, and multinomial logistic regression) were used to estimate the soil properties of interest.