Influence of land management on soil erosion, connectivity, and sediment delivery in agricultural catchments: Closing the sediment budget

Agricultural land, and arable farming in particular, is commonly associated with increased soil erosion risk. Such systems are most vulnerable during low groundcover periods, but downstream delivery is ultimately controlled by connectivity. This study provides a catchment‐scale sediment budget integrating three discrete but complementary investigations spanning different temporal and spatial scales. The first gives details on suspended sediment fluxes at the catchment outlet (2009–2012). The second provenances sources of fluxes using quantitative sediment fingerprinting. The third sets recent data in a multidecadal (60‐year) context using radiometric (137Cs) field‐scale soil loss estimates. The catchment observatory (11 km2) is low relief with predominantly well‐drained soils and dominated by spring‐sown cereal cropping through the study period. Modelling 137Cs inventory losses across 30 fields provided a catchment‐wide mean soil loss of 2.0 Mg ha−1 yr−1. Although such rates are not atypical of intensively managed agriculture across Europe, they are considerably higher than contemporary sediment export yields of 0.12 Mg ha−1 yr−1 of which fingerprinting revealed that contemporary slope erosion contributed less than 25% (0.03 Mg ha−1 yr−1). No evidence of floodplain or in‐channel sediment storage was consistent with disconnectivity. Instead, it is hypothesised that soil loss is associated with coextraction from root crop harvesting of previously widespread sugar beet crops. Considering that the highest mass‐specific 137Cs concentration occurred during the 1960s, there appears to have been significant depletion of the cumulative 137Cs inventory where root crop harvesting occurred as compared with atmospheric fallout ‘reference sites.' The study highlights the value of multiple methodologies when seeking to understand legacy issues within agricultural catchment settings.


| INTRODUCTION
Changes in agricultural land use and management can have both positive and negative effects on soil, which in turn can impact downstream aquatic ecosystems (Mueller et al., 2012). Scaling up to Atlantic Europe, soil sustainability (incorporating soil erosion, soil organic carbon, contamination, and compaction) is identified as a key environmental challenge (Amundson et al., 2015;Creamer et al., 2010;Louwagie, Gay, Sammeth, & Ratinger, 2011). Soil erosion can remove carbon-and nutrient-rich topsoils, resulting in a subsequent need to rebuild soil fertility, structure, and carbon content (Powlson et al., 2011;Quinton, Govers, Van Oost, & Bardgett, 2010). Where sediment and associated nutrients are hydrologically connected and delivered in excessive quantities to watercourses, physical damage to aquatic habitats and species, eutrophication, and reduced light penetration can occur, resulting in a deterioration in the quality of aquatic ecosystems (Kemp, Sear, Collins, Naden, & Jones, 2011;Kjelland, Woodley, Swannack, & Smith, 2015). Successful management of soil erosion and sediment delivery must, therefore, consider the spatial variability and magnitude-frequency relationships of particle detachment, conveyance, and delivery mechanisms-a process spectrum termed 'sediment connectivity' (cf. Bracken, Turnbull, Wainwright, & Bogaart, 2015).
Sediment connectivity is generally controlled by climate, physiography, lithology, and land management, which impact the availability and transfer dynamics of sediment within a catchment sediment cascade (Bracken & Croke, 2007). Physically, catchments possess a quasi-constant set of "boundary conditions," determined by topography, geology, soil texture, and topsoil-subsoil, subsoil-geology transitions. These conditions characterise the likelihood of water and sediment storage, preferential flow paths, or impeded horizontal and/or vertical water movement. For example, following rainfall, catchments underlain by geology featuring efficient primary and/or secondary permeability (e.g., karst landscapes) support subsurface hydrological pathways thereby reducing surface sediment connectivity (Schmidt & Morche, 2006;Mellander et al., 2013). In contrast, rainfall on catchments with impeded vertical permeability may encourage sediment connectivity as water emerges at the soil surface and subsequently moves over ground .
Agricultural land management influences soil erosion rates by, for example, increased erodibility and efficiency of runoff on bare or low groundcover soils (Kirkby et al., 2004). On arable land, connectivity varies seasonally according to crop type, drilling date, and post-seeding establishment of groundcover in relation to rainfall characteristics and soil moisture conditions (Boardman & Favis-Mortlock, 2014).
Permanent crop cover in pasture-based systems is more commonly associated with reduced sediment connectivity through interception of runoff and increased infiltration and organic matter build-up (Regan et al., 2012). However, channel bank erosion and compaction of soils by grazing livestock in fluvial corridors can also contribute to sediment fluxes (Lefrançois, Grimaldi, Gascuel-Odoux, & Gilliet, 2007). On the hillslope, landscape features can both concentrate (in-field tracks and tramlines) and intercept or attenuate (hedgerows and buffer strips) flow pathways, respectively, increasing and decreasing the likelihood of sediment connectivity between field systems and the channel network (Lacoste, Michot, Viaud, Evrard, & Walter, 2014;Thomas et al., 2017). Furthermore, in agroecosystems where agricultural productivity is inhibited by saturated field soils, extensive artificial surface and/or subsurface drainage systems and the modification and maintenance of the river network aim to reduce field water storage, which may both positively and negatively impact sediment connectivity (Shore et al., 2013).
The amount of water available for routing through catchments is driven by climate. In NW European latitudes, antecedent conditions prior to rainfall events (which themselves vary in duration, magnitude, intensity, and frequency) are a strong determinant of sediment connectivity at the event scale (Perks et al., 2015;Sherriff et al., 2016).
Improving the understanding of sediment connectivity is challenged by the spatial and temporal heterogeneity of processes and high resource demands for data collection. However, the benefits of understanding and managing connectivity and disconnectivity to reduce the impacts of agroecosystems on downstream water resources make assembling an evidence-base essential to classify current behaviour and explore the influence of land management and climatic behaviour. Sediment budgeting, assembled through integration of complementary field-based techniques, quantifying losses and gains across a catchment, has been suggested as an essential framework for sediment management (Walling & Collins, 2008).
In this study, radiometric-based soil loss estimates were combined with existing downstream suspended sediment flux measurements to assemble a sediment budget to investigate temporal variations in hydrological and sediment disconnectivity. This was undertaken in a highly monitored, intensively managed, productive, and long-term predominantly arable mesoscale headwater catchment observatory with well-drained soils in Ireland . Sediment monitoring at the catchment outlet since 2009 has indicated a relatively low annual average suspended sediment yield of 0.12 Mg ha −1 yr −1 . Sediment fingerprinting of the yield showed that field topsoils are not the primary source, contributing, on average, 24% of sediment compared with 53% from subsoils and channel banks and 24% from damaged road verges (Sherriff, Rowan, Fenton, Jordan, & Ó hUallacháin, 2018). Quantification of seasonally averaged channel sediment storage indicated relatively low storage, approximately equivalent to 1 year of sediment yield (Sherriff et al., 2016). Using a LiDAR-based hydrologically sensitive area model, Thomas et al. (2016) indicated limited connectivity (7% of catchment area) during upper quartile rainfall-runoff events and highlighted the influence of microtopographically driven flow sinks, on catchment disconnectivity.
The results were attributed to high permeability soils and fractured bedrock which dominate the catchment and support subsurface hydrological pathways; these reduce surface hydrological and sediment connectivity. However, high interannual variability of annual suspended sediment yield (0.03-0.23 Mg ha −1 yr −1 ) and the high magnitude of event-scale sediment export following extreme rainfall, and likely efficient sediment connectivity, suggest periodic risks from this and similar catchments (Sherriff et al., 2016). A challenge is to quantify the extent and proportionate times of these apparent low sediment delivery rates with periods of high risk in the absence of long time series data.
The 137 Cs-based soil loss forensic approach was considered here to give a medium-term (approximately 60-year) time-integrated estimate to compare against short-term high-resolution sediment flux measurements at the catchment outlet. The basis of the technique is to use loss or gain of fallout 137 Cs inventory (Bq m −2 ) in a soil profile relative to the inventory in a noneroding reference site to infer the amount of soil lost by erosion or, conversely, gained through deposition (Lacoste et al., 2014;Porto, Walling, & Ferro, 2001;Walling & Quine, 1990). The 137 Cs methodology for soil erosion assessment is subject to ongoing debates regarding the spatial variability of 137 Cs fallout and transfer to soil and the reliability of converting sampled inventories into soil erosion rates (Parsons & Foster, 2011, 2013. Mabit, Meusburger, Fulajtar, and Alewell (2013) review the existing and future strategies to minimise the potential impact of such uncertainties which are consequently an important consideration for future study. With these considerations in mind, the aim of the study was to assemble a catchment sediment budget to identify the relative importance of soil erosion and sediment (dis)connectivity on the low sediment delivery rates. This information can help inform catchment management strategies in this and similar catchment types.
The objectives were to (a) quantify field-scale soil erosion and sediment redistribution in a highly productive arable catchment based on slope and land use using~60-year field-based 137 Cs soil loss estimates; (b) drawing on existing recent flux and channel source and storage measures, estimate the catchment sediment budget; and (c) evaluate the sediment connectivity implications for management.

| Study site
The 11.2-km 2 Castledockrell study catchment observatory is located in Co. Wexford, Ireland (52°34′N, 6°36′W). The catchment has been extensively described elsewhere (Sherriff et al., , 2016, and a summary is provided below. Hillslope soils are predominantly welldrained, shallow Brown Earths, underlain by a fine loamy drift with siliceous stone subsoil. In the stream corridor and towards the eastern parts of the catchment, smaller areas of poorly drained Groundwater Gleys are present (Figure 1; . The underlying geology is characterised by slate and silt stones of the Oaklands Formation creating a poorly productive aquifer (Tietzsch-Tyler et al., 1994); however, secondary permeability occurs through bedrock fissure flow.
The dominant land use in the catchment is arable farming (54%), primarily cultivated on well-drained soils with spring barley, and the remainder of land is predominantly pasture-based for beef/dairy cattle and sheep grazing which primarily occurs on poorly drained soils. Two tributaries comprise the stream network flowing west-east and north-south which also drain a limited proportion of artificial drainage which is located in the poorly drained soils (stream linear density 1.3 km km −2 ). The present average field size is 3.2 ha, and the areal density of hedgerows is 0.011 km 2 km −2 . A trend of increasing field size and decreasing density of hedgerows was likely over the soil redistribution decadal monitoring period as agricultural intensification was supported by large-scale adoption of machinery and the introduction of production orientated policy support, that is, the Common Agricultural Policy. Catchment rainfall, discharge, and suspended sediment yield in hydrological years (October 1 to Septem- Contributions from snow and subsequent snowmelt hydrology were minimal (less than 2 days yr −1 with snow lying as recorded at Rosslare synoptic station, Co. Wexford; Met Éireann, 2017). Field-scale land use data were collected as part of the wider monitoring programme (Wall et al., 2011). A longer term land use record (Land Parcel Information System), available from 2000 to 2013, was used to infer cultivation practices in the study catchment (DAFM, 2013).

| Laboratory analysis
Soil cores were oven dried at 105°C for greater than 48 hr, mechanically disaggregated, and sieved to 2 mm. Subsamples of 900 g, with consistent sample geometry, were analysed for 137 Cs at 661 keV on a low background ORTEC HPGe detector gamma spectrometer (ORTEC, Oak Ridge, USA) at the University of Dundee with count times over 30,000 s (mean error: 15%). Specific surface area (mg kg −1 ) was estimated using a Malvern Mastersizer 2000G with autosampler unit following organic matter removal and chemical dispersal .

| Data analysis
Caesium-137 activity concentrations (Bq kg −1 ) were converted to areal activity inventories (Bq m −2 ) using total <2-mm bulked core mass and respective core surface areas (Porto et al., 2001). The Mass Balance 2 (MB2) model (FAO/IAEA, 2013) was used to calculate gross soil erosion rates (Mg ha −1 yr −1 ) from bulked field cores and the field-scale net erosion (Mg ha −1 yr −1 ) and sediment delivery ratio (field-SDR%;  using all bulked cores within a specific field which were weighted to represent their spatial representativeness (top, 12.5%; upper-mid, 25%; middle, 25%; lower-mid, 25%; bottom, 12.5%; Figure 2). The MB2 model includes a vertical mixing component to represent cultivation; this was used for all fields with vertical 137 Cs distribution by ploughing, that is, for cultivated fields or reseeding for permanent pasture fields. The 2000-2013 land use record and sectioned cores indicated this was highly likely across the catchment (Figure 3). Cultivated fields were defined as those which supported arable crops in the available land use record (n = 24), and the remaining sampled fields were defined as permanent pasture (n = 6), whereby vertical 137 Cs mixing was possible but at a lower temporal frequency. Mechanical cultivation was assumed limited due to low permeability soils, small field sizes, or restricted accessibility for machinery. The MB2 model used the following parameters: tillage depth (115 kg m −2 ) defined by the mass depth of 137 Cs from the incremental core ( Figure 3); the proportion factor (1), accounting for the accumulation of 137 Cs during major rainfall preceding the cultivation in the catchment; and the relaxation depth (4 kg m −2 ) as defined in agricultural fields in the United Kingdom . The stronger affinity of 137 Cs to clay particles between mobilised sediments and the original soil volume in both models was corrected using the particle size factor (average ratio 0.99) based on specific surface area (Dalgleish & Foster, 1996;Walling, Zhang, & He, 2007).  Field-scale soil erosion rates showed greater gross erosion, net erosion, and sediment delivery ratios from high and medium mean slope field classes relative to low, but these were not statistically significant (Table 1). The mean estimated field-scale gross soil loss rates of 2.8 Mg ha −1 yr −1 (range 0.0-13.0 Mg ha −1 yr −1 ) and net soil erosion rate of 2.0 Mg ha −1 yr −1 resulted in an average field-SDR of 55% for all fields (Table 1). Soil losses from permanent pasture fields were greater than cultivated, but no spatial trends were evident across the catchment (Figure 6b).
To understand the influence of cultivated crop type on soil erosion mechanisms, land use data from 2000 to 2005 in the Castledockrell catchment show 57% of cultivated fields were used for root crops and the maximum proportion was 15% of the catchment area ( Figure 7a). After 2005, there was a decline overall in root crops, most notably sugar beet and greater uptake of grains. Across the wider Co.
Wexford region, Figure 7b showed the greatest areas 10,000-14,000 ha of cultivated root crops from 1954 until the early 1970s.
An overall increase in the area of sugar beet occurred from 1954 to 2000 from approximately 2,000 to 7,000 ha. Monthly rainfall data from 1954 to 2014 show annual average monthly rainfall contribution is highest in October to January at over 10% compared with February to September (Figure 8a). During the years of greatest 137 Cs fallout (1954)(1955)(1956)(1957)(1958)(1959)(1960)(1961)(1962)(1963)(1964), annual rainfall was within 15% of the 30-year average annual rainfall (1,041 mm) for 8 years. Two years, 1958 and1960, had annual rainfall of 1,366 and 1,318 mm, respectively, which were above 25% greater than the 30-year average annual rainfall.
The average net soil erosion rates of cultivated and permanent grassland fields relative to their respective catchment areas (cultivated, 874 ha; permanent grassland, 106 ha) estimated a mediumterm net soil erosion rate of 1,777 Mg yr −1 or 1.8 Mg ha −1 yr −1 for the whole catchment. This represented maximum possible sediment input to the down-catchment sediment cascade and field-scale sediment retention (731 Mg yr −1 ) and gross-soil erosion (2,508 Mg yr −1 ) assembled for a 60-year average sediment budget (Figure 9). Recent data for annual average yield combined with quantified bed sediment storage and source provenance data indicate the contemporary sediment budget (Figure 9). In this case, the annual average suspended sediment yield of 0.12 Mg ha −1 yr −1 equated to 124 Mg yr −1 . The contemporary sediment fingerprinting data suggest, on average, 24%, 53%, and 24% were derived from field topsoils, channel banks, and road verges, respectively, which corresponds to 34, 78, and 34 Mg yr −1 in Figure 9 (Sherriff et al., 2018). As measured seasonal channel bed storage was relatively consistent, this was assumed not to cause fluctuations in the catchment yield between years (Sherriff et al., 2016).

| DISCUSSION
Field core 137 Cs concentrations above and below the reference inventory showed both erosional and depositional processes were present in the Castledockrell catchment. The depth profiles of both reference locations showed vertical disturbance. In intensive catchments with 93% agricultural utilisation , identification of physically undisturbed areas of soil was problematic. The two reference sites are, however, based on low-risk soils for run-on and runoff, in marginal areas (edge of field unproductive area and a private to surface pathways being not significant in this catchment or, at least, not detectable relative to other erosional processes influencing the 137 Cs inventory ( Figure 5). These data also indicate areas of elevated erosion may occur at a single hillslope location and are often not consistent, likely due to slope geometry and anthropogenic influences such as trafficking routes across fields. This supports previous work suggesting predominantly well-drained catchment soils underlain by fractured bedrock promoting infiltration and resulting in subsurface hydrological pathways Sherriff et al., 2016Sherriff et al., , 2015Thomas et al., 2016). Thus, the likelihood for surface runoff driven erosion and sediment transport pathways is, at best, intermittent and confined to the most extreme rainfall events and/or localised areas of saturation excess or infiltration excess overland flow on poorly drained gleyic soils in the river corridor (Sherriff et al., 2016).    Verheijen et al., 2009;Cerdan et al., 2010;Regan et al., 2012;Panagos et al., 2015). Estimated soil losses from permanent pasture fields in this catchment are greater than estimated for European soils and significantly greater than cultivated fields (Cerdan et al., 2010). These permanent pasture fields are frequently located on steeper, poorly drained soils in the river corridor. Therefore, despite the interception of surface hydrological pathways by vegetation, greater surface hydrological connectivity elevates the likelihood of soil erosion and particle entrainment, that is, sediment connectivity Sherriff et al., 2018;Vannoppen, Vanmaercke, De Baets, & Poesen, 2015). These clay-dominated gleyic soils have also been shown to bind radiocaesium more strongly than coarser soil textures (Okumura et al., 2018). Greater affinity of 137 Cs to finer particles which are preferentially eroded relative to coarser fractions, particularly during surficial soil erosion, may indicate actual soil loss rates from permanent pasture fields are lower than estimated here.
From a connectivity perspective, the estimated catchment soil erosion rate highlights a major anomaly of 60-year field-scale soil loss rates with high field-SDRs and recent suspended sediment yields (2009-2013) of 0.12 Mg ha −1 yr −1 . Sediment fingerprinting completed in this catchment over 2 years indicates only 24% (0.03 Mg ha −1 yr −1 ) of the suspended sediment yield can be attributed to field topsoils. This indicates an extremely low field-tooutlet SDR of 2%. Low catchment SDRs are not unexpected in catchments dominated by subsurface hydrological pathways as areas of hydrological connectivity are spatially and temporally limited (Thomas et al., 2016). This would suggest conveyance losses occur further along the sediment cascade as particles are transported from the field, into, and through the river network to the catchment outlet (Walling, Collins, Jones, Leeks, & Old, 2006;Walling, Russell, Hodgkinson, & Zhang, 2002). Co. Wexford represents 75% of the catchment area (Lee & O'Connor, 1976;Lee & Ryan, 1966). The extent and magnitude of root crop cultivation were restricted by one in 3-year rotation of sugar beet to control nematodes which subsequently restricted the potential spatial extent of harvest associated soil loss and 137 Cs removal (Hbirkou et al., 2011). However, anecdotal evidence suggests all fields on a cultivated rotation are likely to have supported root crops in the past.
This was consistent with the trend in the area root crops over time which suggests there was previously a greater proportion of fields supporting root crops (Figure 7b). Uptake of caesium into the plants themselves must also be considered. On loam soils, a consistent texture with the cultivated catchment soils, radiocaesium uptake by cereals and root crops has been estimated at 1.4 × 10 −2 Bq kg −1 (range 4.5 × 10 −4 -4.2 × 10 −1 Bq kg −1 ) and 5.7 × 10 −2 Bq kg −1 (range 1.5 × 10 −3 -9.0 × 10 −1 Bq kg −1 ), respectively, which is a small proportion of the field core activity values (Zhu & Smoulders, 2000).
Additionally, hydrological redistribution of the recorded inventory through diffusion and convection processes is possible and have been reported elsewhere by He and Walling (1997). However, the same study and others acknowledge that cultivation and erosion dominate for vertical soil distribution due to the high affinity, up to 93%, of 137 Cs to particles (Yamashiki et al., 2014).  Ruysschaert, Poesen, Verstraeten, & Govers, 2004Tuğrul, İçöz, & Perendeci, 2012), and it follows that soil-bound 137 Cs is coextracted and removed from the catchment fallout inventory (Bq m −2 ) attached to crop roots (cf. van Oost et al., 2003;Walling et al., 2005;Golosov et al., 2017).
To estimate the maximum possible rate of harvest soil erosion, a catchment sediment budget was constructed based on the assumption that all sediment not retained in-field are lost to harvest erosion. The maximum possible rate of 1.5 Mg ha −1 yr −1 appears justifiably at the lower end of published rates, as coarser loam soil textures, such as the Brown Earths, which dominate the cultivated catchment soils, are less likely to adhere to root crops relative to clay-dominated soils (Ruysschaert et al., 2004). Harvest soil erosion will subsequently coextract adsorbed 137 Cs which was likely greatest during periods of highest 137 Cs deposition in the mid-1960s. This process is not represented in standard conversion models but is likely to artificially reduce the 137 Cs inventory in the affected fields and potentially inflate the estimated soil erosion losses reported above (van Oost et al., 2003).
The variability in sediment connectivity over time must also be considered. Cultivation of sugar beet, in terms of the agricultural calendar, likely increased the historical frequency and spatial extent of sediment connectivity and sediment yield from the catchment which is not captured in the available sediment yield data. Late autumn/early winter harvesting of sugar beet crops coincided with the main annual rainfall period (October to January) which likely increased the 'window of opportunity,' that is, efficiency of water erosion, entrainment, and transport compared with a spring cerealdominated catchment whereby post-harvest, soils are unmanaged and crop residues provide some stability to the soil profile and are drilled, on average, during drier months (Boardman & Favis-Mortlock, 2014). Significant event suspended sediment fluxes (flow-weighted mean suspended sediment concentration, 1,256 mg L −1 ) have been reported in this catchment when extreme rainfall and low groundcover combine which exceeds the equivalent metric in catchments with higher annual average suspended sediment yields (Sherriff et al., 2016). These events were likely more frequent under past land use configurations which is supported by anecdotal evidence from catchment landowners and cannot be attributed to changes in rainfall. Field amalgamation due to mechanisation over time, resulting in increased field size and reduced hedgerows, likely increased hillslope soil erosion risk and modified hydrological pathways (Thomas et al., 2016). However, the impact of cropped field positions on modelled mean suspended sediment yield versus the window of opportunity effect has been reported as 22% and 100%, respectively (Smith et al., 2018). This suggests the gradual amalgamation of fields is likely to have a minor impact on soil erosion and sediment delivery, particularly considering the low connectivity of the majority of catchment soils.
Using the time-averaged 137 Cs data, the minimum estimated hillslope sediment loss of 466 Mg yr −1 does not include sediment delivery from cultivated fields. This value was likely higher (and compensated in the sediment budget by a reduction in the maximum possible harvest soil erosion rate) but challenging to estimate and would have impacted in-field redistribution of sediment, channel sediment storage, and sediment export values but not significantly enough for observations of legacy sediment at the present day. The present estimate of field-derived sediment (indicated by sediment fingerprinting) suggests more than 69% of the minimum possible 60-year 'input to downcatchment sediment cascade' volume is lost in conveyance to the river network ( Figure 9). Large-scale sediment retention is not supported by geomorphic evidence suggesting that estimated soil erosion results from direct soil removal from the field, that is, coextraction, or efficient erosion and transport of eroded soils through the catchment.
However, using field methodologies with contrasting timescales to estimate conveyance losses (soil redistribution and sediment fingerprinting) and the catchment SDR (soil redistribution and suspended sediment yield) to infer catchment behaviour is problematic. As previously discussed, shifts in land management alter sediment connectivity and undermine the representativeness of relatively short-term datasets and corresponding metrics.
Bringing together all the evidence to explain apparent sediment disconnectivity in this catchment based on land use, associated land management, and the spatial/temporal extent of hydrologically sensitive areas, the 137 Cs data offer new insights. High average soil erosion rates and low in-field sediment retention over the last 60 years, yet the apparent lack of such processes during the present day, suggest the soil erosion and sediment delivery were historically augmented.
This matches with land use records of sugar beet which was harvested during wetter winter periods and subjected the catchment to harvest soil erosion which is responsible for loss soil from the surface of affected fields. Nevertheless, this disconnection, reduced sensitivity or function, apparently provided by current (spring crop and grassland) land cover, is likely to be undermined with reconfigurations to cropping patterns that expose soil redistributions to winter rainfallrunoff processes (Smith et al., 2018). Policies which could result in significant changes in crop type and land management windows must consider the potential impact of soil erosion, sediment connectivity, and the influence of these potentially augmented processes on soil and aquatic ecosystems (Keesstra et al., 2009). Sediment budgets, constructed using soil erosion, sediment connectivity, and sediment delivery studies, must fully consider the appropriateness of combining contrasting methodologies and complete a full appraisal of processes over space and time to support interpretations of collected data (Parsons, 2011). Combining soil erosion and sediment measurement with modelling expertise may offer a crucial future strategy to resolve discrepancies and explore the influence of anthropogenic or natural changes on catchment sediment yields across a variety of timescales (Smith et al., 2018).

| CONCLUSIONS
Soil core and field-scale soil redistribution rates of cultivated and permanent pasture fields were quantified using a forensic 137 Cs approach in an arable catchment observatory to estimate catchment soil erosion and assess sediment (dis)connectivity by constructing a sediment budget. Collating data from three discrete methodologies with contrasting timescales revealed fundamentally divergent systems of soil erosion and sediment delivery which could not be explained by sediment connectivity. The assembled sediment budget suggested that 60-year average soil loss rates contradicted contemporary basin processes, consequently raising the importance of land use (change) and land management in the catchment.
The main findings were (a) the 60-year gross soil erosion from all fields (average 2.8 Mg ha −1 yr −1 ) was higher than tolerable soil erosion rates despite dominance of well-drained soils in the catchment; (b) net soil erosion from permanent pasture fields, 4.4 Mg ha −1 yr −1 (on predominantly low permeability soils), was greater than from cultivated fields, 1.5 Mg ha −1 yr −1 (on predominantly high permeability soils), due to the greater likelihood of overland flow initiation; and (c) net soil erosion rates from cultivated fields could not be validated by sediment yield, riparian attenuation, or in-stream bed suspended sediment storage during the present day, indicating a 2% catchment sediment delivery ratio. Sediment removal from the arable catchment fields was likely due to the combined influence of coextraction from historical harvest soil erosion associated with root crops combined with increased efficiency and transport of water erosion when low groundcover periods coincided with, on average, wetter periods (compared with solely spring sown crops) and this fits with recollections from catchment farmers.
In conclusion, the reduced sensitivity of sediment export risk (i.e., disconnectivity) offered by high soil permeability and a low extent of hydrologically sensitive areas is also linked to land use. In this study, historical field-scale erosion rates were likely to have been highly influenced by winter-sown sugar beet crops and harvest erosion. This now appears reduced by spring-grown cereal crops which are harvested without removal of the root system (and associated soil). Permanent pasture land management on low permeability soils is a consistent high risk for soil erosion despite the limited extent in the catchment. This has implications for other catchments sensitive to sediment export risk and where land use configurations may mitigate or augment that risk now and under future policy driven land use change. Golosov, V., Collins, A. L., Tang, Q., Zhang, X., Zhou, P., He, X., & Wen, A. (2017). Sediment transfer at different spatial and temporal scales in the Sichuan Hilly Basin, China: Synthesizing data from multiple approaches and preliminary interpretation in the context of climatic and anthropogenic drivers. Govers, G., Quine, T. A., Desmet, P. J. J., & Walling, D. E. (1996). The relative contribution of soil tillage and overland flow erosion to soil redistribution on agricultural land. Earth Surface Processes and Landforms, 21, 929-946. https://doi.org/10.1002/(SICI)1096-9837 (199610)21:10<929::AID-ESP631>3.0.CO;2-C