10. Assumptions, limitations and gaps
10.1. Soil erosion rates
There are few examples of quantified soil erosion rates in Scotland. Often these are related to specific soils, land uses and slopes, so identifying representative rates for a given combination of these variables is challenging. For the purposes of this project, a comprehensive database of erosion observations (>1,680) under a range of field conditions was used to estimate representative erosion rates for a range of land uses (Evans, pers.comm., 31/07/19). However this data was sourced from England and Wales and analysis showed that only some of the data agreed with the progression from Low to High Risk Soil Erosion Risk Classes as developed by Lilly et al. (2002) and Lilly and Baggaley (2014).
The outcome of the project was to estimate a total annual cost of soil erosion, so the erosion rates used should be on an annual basis. However, it is not known if the Evans’ data referred to one event or several events, or the duration over which the erosion (and depositional) features had developed. For the purposes of the analysis, it was assumed that the rates of erosion observed by Evans were the equivalent of a representative annual rate. This is likely to be an underestimate of actual erosion that occurs in any one year, but may be an overestimate if the feature has existed for a long time. The latter case is unlikely on arable land due to cultivations and other field operations. The Evans’ data was recorded during the 1980s and may not be representative of current (or future) erosion rates. Some researchers argue that soil erosion is being aggravated by climate change and associated extreme weather events, resulting in higher rates (and probabilities) of erosion ((Boardman et al., 1990; Boardman and Favis-Mortlock, 2001; Hough et al, 2010; IPCC, 2019; Mullan, 2013a, 2013b). Finally, it is recognised that most erosion surveys are by their nature biased, as only eroding fields are recorded. However, this limitation was not the case for the observations used in estimating the probability of Scottish soils eroding in any one year.
10.2. Probabilities of soil erosion occurring in any one year / field
Scottish observations were used to estimate the probability of soil erosion occurring in any one field in any one year. NSIS (1978-88) data was used to analyse the probability of erosion for the organo-mineral and peat soils (that are grouped under the “Organo-mineral and peat soils” in the present project). When combined, the probability of erosion was estimated to be 21% (Section 4.3). It is possible to separate these different soil types and estimate the probabilities of erosion in these classes: Peat = 31%; organo-mineral = 12%. However, should the analysis be extended, a L, M and H erosion risk for organo-mineral and peat soils could be undertaken. This would require considerable more time to run the economic model for what would become 6 Erosion Risk Classes (rather than the 4 currently modelled). Also, the evidence base of differentiating associated costs for all 6 Classes is very limited.
Given that Soil Erosion Risk Class was used to determine probability, it should be noted that this classification based on soil texture, slope and HOST class is likely to remain unchanged, even under future climate change scenarios
10.3. Land use
Land use will be a ‘snap shot’ in time and there will have been changes in land use since the data (LCM2007, IACS and NFI) were collected. However, we have assumed that any changes in land use will cancel out in space and time (i.e. areas going into arable = areas going out of arable etc.). It was also assumed that the spatial / temporal distribution of crops within the arable areas (as part of the arable rotation) would be similar for all catchments (and at the national scale). For arable crops, typical rotations (different crops) may be available from the Scottish Farm Management Handbook (2018/19) (https://www.fas.scot/downloads/farm-management-handbook-2018-19/). IACS returns over several years could identify actual rotational patterns in each catchment. This is particularly important for reseeding of ley grasses within an arable rotation, as this represents a significant increase in erosion rates for that year of reseeding.
The Countryside Information System (CIS) data, aggregated at national level was used to estimate areas of farm infrastructure (that would not be included in the analysis). It was assumed that these proportions would be applicable to all agricultural land at the catchment and national scale.
Soil erosion rates and probabilities for urban areas were excluded from the analysis, due to the paucity of data in the case study catchments and for Scotland as a whole. Soil erosion rates in urban areas are likely to be low due to the high proportion of sealed surfaces. During construction phases, exposure of bare soil and soil disturbance by heavy plant and machinery may cause erosion rates to be much higher (Blum, 1998), but data on this for Scotland was unavailable.
10.4. Assumptions and limitations of the economic model
This assessment of the annual costs of soil degradation largely confirms the difficulties, evident in previous reviews, of deriving complete and reliable estimates of the benefits provided by soils and how these change according to soil condition. There are three aspects to this challenge (i) ‘identifying’ biophysical relationships between soil properties, soil functions and ‘performance’ of soils in particular applications (ii) ‘valuing’ the diverse range of market and non-market benefits and costs attributable to soils in different applications and (iii) assessing the ‘dynamics’ of soil properties, especially under conditions of climate change, as these affect changes in the supply and value of services.
The ecosystems approach adopted provides a systematic framework for the identification and valuation of soil services, although in many respects it confirms that knowledge is insufficiently complete to allow a full assessment. Furthermore, although some information is available for soil erosion rates and probabilities on specific sites, it is difficult to aggregate this at the regional and national scale to support policy. A number of key gaps and uncertainties can be identified.
There is considerable information and knowledge pertaining to soils that has been developed over time, much of it to support provisioning services (i.e. agricultural production). While this remains a key aspect of soil management, the ecosystems service perspective requires that soil science and management adopt a broader remit to include the wide range of soils functions as they support regulating and cultural services as well as provisioning services.
An attempt was made here to link soil erosion processes to indicators of soil quality (e.g. soil depth and nutrient content) and ecosystem services. Examples include soil erosion rates, loss of soil depth and crop yield. It proved difficult to confidently predict these process/indicator/service relationships, yet this is clearly the way forward if the ecosystems approach is to be comprehensively applied.
A better understanding of the relation between soil erosion and service provision is also needed to inform ‘safe’ or ‘target’ indicator levels beyond which ecosystems services might be undesirably or irreversibly compromised. Linked to this, there is clearly a need to better understand how the use of soils drives the relationship between the stock of soil resources, defined in terms of soil quantities and qualities, and the flow of soil-based services. Pressures of more intensive land use and climate change make this a priority.
The ecosystems approach emphasises how soil erosion can reduce the capacity of the stock of soil ‘capital’ to provide soil services now and into the future. Over utilisation of soils beyond their natural ability to maintain or reinstate their inherent properties leads to degradation of stocks and service flows. In some cases this can be corrected by measures to prevent or minimise degradation, in others by substituting lost soil properties with man-made inputs such as artificial fertiliser. These interventions can be costly.
Further data and knowledge are required to adequately understand the complex relationships between soil stocks and service flows, especially under alternative management scenarios. Associated with this, as referred to earlier, is the issue of critical stocks of soil capital, of thresholds and of non linear effects, whereby soil erosion leads to non marginal, step changes in both stocks and flows. Here, uncertainty-based safe minimum standards, set within a strong regulatory regimes, are appropriate. It is recommended that critical threshold / tolerable values are explored for soil erosion rates and probabilities.
The analysis here clearly shows that there is considerable spatial variability not only in the processes of soil erosion but also in its consequences. The approach of using Soil Erosion Risk Classes and land use in determining soil erosion rates explicitly considers spatial distribution of causes and effects. An appreciation of spatial variation and scale are critical to the assessment of the consequences of soil erosion, especially in the context of diverse ecosystems services.
Variation in time is also important, especially as soil erosion processes and effects can be cumulative over time. There is incomplete understanding of cumulative rates of soil erosion associated with for example soil organic matter loss. This aspect, to develop a better understanding of the relationship between changing soil stocks and diverse service flows over time, is essential for a strategic approach to the management of soil resources. It is recommended that this is a priority for future research.
The analysis has identified a number of gaps and uncertainties in the valuation of service flows that are both specific to soils and generic to the valuation of non market ecosystem services as a whole. The main non market impacts of soil erosion appear to relate to water quality impacts and GHG emissions, all of which are subject to uncertainties in valuation. Water quality impacts will vary considerably according to the sensitivity of local ‘receptors’, while GHG valuation rests on the social price of carbon (based here on the cost of CO2 abatement). It is noted that market prices used to value soil service are also uncertain, for example for agricultural inputs and products. For this reason it is important to separate as far as possible biophysical and pricing assumptions, and to generate a range of values to reflect uncertainty.
The calculations of costs for arable and improved grassland on organo-mineral soils and peats may be an overestimate for several reasons. Firstly there is a mismatch in scale between the soil spatial data set and the land use and land management data meaning that some small areas which are unlikely to be cultivated appear on organic soils. Secondly there is an overestimation of the probability of erosion in this soil, land use combination because peat and organo-mineral soils have been grouped into one erosion risk class for the economic analysis. However most of the arable and improved grasslands will be on organo-mineral soils with relatively low soil carbon content and the probability of erosion of these soils from the analysis of the NSIS data, was 12.1% compared with the combined probability of 21% used in the economic model.
It should be noted that the erosion losses/rates do not necessarily mean that all sediment (and associated nutrients) are transported to surface water bodies and some may be retained in-field.
10.4.1. Crop yields in the economic model
When the economic model was applied in England and Wales, crop yields were assumed to vary with soil type (clay, silt, sand and peat). However, in Scotland, soil texture is less differentiated, so it was assumed that it is unlikely that yield will vary much between soil types. In the present study, soil type is taken into account in the Erosion Risk Classes, along with slope and potential to generate runoff. As such, there was no expected relationship between Risk Class and yield. Therefore, for a given crop / land use, the present study used the same crop yield for each Erosion Risk Class. Further studies could test whether representative yields for different crops are sensitive to Erosion Risk Class.
10.4.2. Drinking water extractions
Of the 5 catchments, water is extracted for drinking purposes only in the Ugie catchment. This is reflected in the greater water treatment costs for this catchment. If the other catchments are used for (drinking) water abstractions, then the costs of treating water containing eroded sediment will increase. The results are presented both with and without drinking water treatment costs.
10.4.3. Costs of carbon in water
The project team were unable to find quantified evidence of the amounts, impacts and associated costs of carbon in waterbodies (rivers, canals, lakes, lochs etc.) that is directly associated with soil erosion (and would therefore be an input to the economic model). Whilst there is generic information about DOC (Dissolved Organic Carbon), particularly in Scottish catchments with peatlands, we cannot say for sure if this DOC has been the direct result of soil erosion processes. For example, it could be the result of non-erosive runoff carrying the DOC. On the other hand, it could be the result of water running over the land which will dislodge particles, causing the organic material to dissolve and be transported (as DOC), along with particulate material.
It could also be argued that POC (Particulate Organic Carbon) is more likely to be associated with soil erosion, but we couldn’t find any data that partitioned waterborne C into the DOC v. POC fractions. In any case, according to Duan et al. (2014) “….POC is a small fraction of TOC present in most lakes (with respect to dissolved organic carbon, DOC), ….. (Dhillon and Inamdar, 2013, Son et al., 2009)”…so it is likely that POC is not as big an issue as DOC. Fraser Leith (Scottish Water) confirmed that POC may be as little as 10% of all C in watercourses.
10.4.4. Costs of N in water
The current model includes costs associated with nitrate in water as a result of erosion. This follows the methodology used in Graves et al. (2015). However, it is appreciated that N is usually carried in the water phase rather than associated with particulate material. As with carbon in watercourses, it is very difficult to disentangle N carried in the water flow and that carried by particulate matter that has been eroded. Better understanding of the partitioning of dissolved and particulate N (and C) is therefore needed.
10.4.5. P associated with soil erosion
It is accepted that the economic analysis takes no account of P in rivers / canals, only in lakes / lochs. This is because it is uncertain how much P in rivers and canals is directly attributable to soil erosion. P can result from point or diffuse sources, which may or may not relate to soil erosion processes in the catchment. The project team could not find evidence that linked levels of P in rivers directly to soil erosion events. Most P associated with soil erosion is sorbed onto eroded particles, that are likely to be deposited somewhere in the catchment - possibly in lakes / lochs. Also, the Jacobs and SAC Report (2008), used to estimate the costs of nutrients in watercourses does not give values for P in rivers and canals (although N levels are given and have been taken into account in the economic analysis).
10.4.6. GHG costs
The economic model has not accounted for methane or nitrous oxide emissions due to soil erosion (only carbon). As such, the costs given in this category will be an underestimate. Very little literature was found on the relationship between soil erosion processes and the production of methane and/or nitrous oxide.
10.4.7. Other costs
Costs of roads and properties damages are not included, due to lack of data for Scotland.
10.4.8. Mitigation measures
The economic analysis was also unable to account for the costs of soil erosion mitigation measures at the catchment and national scale. Although costs of mitigation have been reported in the Literature Review, they cannot be included in the economic analysis at the catchment or national scale, because there is no information on the numbers of measures used or their geographical location. This would require mapping these measures at a much finer resolution than used in the present study (e.g. location of field buffer strips; identification of minimum tillage use; etc.).