Developing a method to estimate the costs of soil erosion in high-risk Scottish catchments: final report

6. Applying the model to the case study catchments and national scale

We have updated data going into the economic costing model, including typical crop yields for Scotland (for a number of years); prices of crops (and hence impact of yield penalties due to soil erosion); costs of fertilisers; Gross Margins; forestry model (coniferous v. broadleaved); carbon values; soil nutrient values; GDP inflators; non market costs e.g. carbon sequestration etc.

6.1. Loss of crop yields through soil erosion

Typical yields for each crop in Scotland were taken from the Scottish Farm Management Handbook (2018/19) https://www.fas.scot/downloads/farm-management-handbook-2018-19 (SAC Consulting, 2018). For example, winter wheat has an indicative yield in Scotland of between 4.5 - 8.5 t ha yr.

Using Pimentel’s equation (Pimentel et al., 1995) to calculate yield penalties due to soil erosion (the rates of erosion should reflect the soil and crop type to some extent), relative to typical yield can be used to calculate costs of yield penalties.

6.2. Losses of nutrients and carbon

Soil properties are needed in the economic model. Bulk density (BD) is used to convert the estimated erosion rates into a loss of soil depth, which is associated with yield losses (Pimentel, 1995). Soil nutrients (nitrogen (N), phosphorus (P), potassium (K) and carbon (C)) are eroded with soil particles, so the costs of replacing them have to be estimated. The values used were spatially weighted averages calculated from the SSKIB dataset for each broad land use that occurred within the four Soil Erosion Risk Classes (L, M, H, and organo-mineral and peat soils) (Appendix 3; Appendix 4; Appendix 5; Appendix 6; Appendix 7).

6.3. Land use allocations in the selected case study catchments

The current methodology uses land use to determine the soil erosion rate, modified by Erosion Risk Class to determine the probability of that erosion rate occurring in any given year / on any given field. Soil erosion rates and probabilities for urban areas were excluded from the analysis, as these are likely to be low, due to the high proportion of sealed surfaces.

Also, data on this for Scotland was unavailable. Land use is also used to indicate current levels of soil nutrients (assuming more intensive land uses have higher levels of soil nutrients than the extensive land uses e.g. horticulture versus moorland). Production outputs (e.g. crop yields) are also linked to land use. For example, in grassland systems, it is assumed that ‘Improved Grassland’ is the most productive and intensively managed (e.g. around 8-10 t ha¹ yr^-1), ‘Unimproved Grassland’ is less productive and intensively managed (e.g. around 5-7 t ha^-1 yr^-1), and ‘Rough Grazing’ is low yielding with very low inputs (e.g. around 2-4 t ha^-1 yr¹). This land use category includes natural or semi-natural vegetation so it could include heather moorland, for example, whereas rough grass is just semi natural (unimproved) grass such as Molinia or Nardus or bent species. Thus accurate information on the spatial distribution of land use (and how it coincides with Erosion Risk Class) in the 5 case study catchments and for the whole of Scotland is needed as input to the economic model.

Integrated Administration and Control System (IACS) data is accurate and of high resolution (i.e. field by field). It was used to give annual snapshots of agricultural land use at the field level, allowing for high resolution spatio-temporal land use change studies at the national scale. It identifies arable, grassland and horticulture, but not forestry, woodland, wildscape or moorland. The detailed IACS land use classes then had to be linked to the crops already embedded in the economic model (and to the erosion rates assigned to each of these land uses).

Appendix 8 shows how the IACS crops found in Scotland were assigned a category relating to a) soil erosion rate (i.e. what broad land use category did the crops belong to) and b) the economic model’s range of crop types.

The National Forest Inventory (NFI) of Scotland was used to identify woodlands and forests. Where IACS or NFI data was not available, LCM 2007 was used to assign land use to the remaining areas. It is recognised that LCM 2015 would have given more up to date data but was not available to the project team within the timescales of the project.

These datasets (IACS, LCM 2007 and NFI) were combined to estimate the % of each land use in each of the five catchments. Maps of land use and Erosion Risk Class for each case study catchment were produced (Figure 4, Figure 5 and Figure 6) and the proportion of the catchment in each combination was calculated.

6.3.1. Crop rotations within arable land

Within the arable areas, the cropping patterns are needed to assign the area under each crop in any particular year. The economic model will predict yields for each of these crops. The classification of IACS classes observed into the crop types incorporated into the economic model is shown in Appendix 8. Given the broad range of crops grown under arable, some assumptions and simplifications had to be made. Set aside and fallow were assumed to be part of an ‘Extensive Arable’ rotation.

Distribution of crops over space in any one year is used as an analogy for time (i.e. ‘pseudorotation’). It was assumed that the spatial / temporal distribution of crops within arable areas would be similar in all catchments (and for the national scale). Field margins were not identified as a separate land use within arable areas.

6.3.2. Farm infrastructure

It was decided to exclude areas of farm infrastructure, such as farm buildings, yards, tracks and roads from the subsequent analyses. These can take up a proportion of the land and will have much lower soil erosion rates than the surrounding farmland, as they are often sealed by concrete, tarmac and hard standings etc.. The area involved was assumed to vary with land use, but not Erosion Risk Class (Table 14). The Countryside Information System (CIS) data was used to estimate these areas. It was assumed that these proportions would be the same for all agricultural land at the catchment and national scale.

Table 14. Proportion of land under infrastructure (and excluded from the analysis)

Land use	Proportion of land as infrastructure (%)
	Erosion Risk Class
	Low	Moderate	High	Organo-mineral soils and peats
Horticulture	15%	15%	15%	15%
Arable intensive	15%	15%	15%	15%
Arable extensive	15%	15%	15%	15%
Grassland improved	10%	10%	10%	10%
Grassland unimproved	10%	10%	10%	10%
Rough grassland	2%	2%	2%	2%
Forestry	2%	2%	2%	2%
Woodland	-	-	-	-
Wildscape	-	-	-	-

6.4. Calculating annual on-site costs

6.4.1. Erosion induced productivity loss

The effect of soil erosion on crop yields is frequently given in relation to a change in soil depth (Lal, 1998; Pimentel et al., 1995). Observed reductions show wide variation, depending on the crop rooting trait (some crops are deeper rooted than others), nutrient storage and availability, water holding capacity, soil profile characteristics, the crop grown, existing soil and crop management, and the microclimate (Lal, 1998; Pimentel et al., 1995). It could be seen that there is also a loss of opportunity in which crops can continue to grow in eroded (i.e. shallower) soils. This may require change in crop type and even land use. Since erosion induced yield declines are of considerable importance, especially in low input agriculture, tools such as EPIC (Williams, Renard and Dyke, 1983) have been developed to examine long-term effects of various components of soil erosion on crop production. Here, a simplified approach was developed, using the relationships developed by Pimentel et al. (1995). Based on a variety of measured data, Pimentel et al. (1995) found that crop yields declined by 20% over a period of 20 years (Table 15).

Table 15. Erosion induced yield penalty, based on Pimentel et al. (1995)

Yield penalty (%)
Parameter associated with soil erosion	20 year total	Average annual	per mm soil loss
Water runoff	7	0.35	0.3
Nitrogen loss Phosphorus loss Potassium loss	8	0.4	0.3
Soil depth loss	7	0.35	0.3
Organic matter loss	4	0.2	0.1
Water holding capacity loss	2	0.1	0.1
Soil biota loss	1	0.05	0.0
Total on-site losses	20	1	0.74

The following example can illustrate the approach. Assuming soil was eroded at a mean rate of 17 t ha^-1 yr^-1 and the bulk density was 1.25 Mg/m³ , the annual loss of soil depth would be estimated to be 1.36 mm yr^-1 using Equation 5. This therefore equated to a yield loss of 0.74% per mm of soil loss = 1% yield loss per annum for a soil depth loss of 1.36 mm due to erosion.

The bulk density (Mg/m³) of soils in each of the 5 case study catchments and for the national scale was determined using a set of (as yet unpublished) regression equations; one to predict the bulk density of mineral topsoils and a separate one to predict the bulk density of organic top soils, based on the proportion of carbon and clay (mineral soils) and carbon concentration (peaty soils) derived from the SSKIB dataset. As an illustration, the bulk density values for each land use and Erosion Risk Class for Scotland as a whole (national scale) are given in Table 16. The values for each of the case study catchments are given in Appendix 3. The data for each of the catchments and for the country as a whole were then used in Equation 5 to derive the depth of soil loss per annum from the rate of soil loss given for each land use in Table 8. The results for soil depth loss for Scotland as a whole are shown in Table 17.

Table 16. Soil bulk density values obtained for Scotland

Land use	Bulk density (Mg/m3)
	Erosion Risk Class
	Low	Moderate	High	Organo-mineral soils and peats
Horticulture	1.30	1.09	1.13	0.18
Arable intensive	1.44	1.29	1.13	0.18
Arable extensive	1.53	1.40	1.41	0.22
Grassland improved	1.41	1.45	1.11	0.15
Grassland unimproved	1.57	1.14	0.41	0.10
Rough grazing	1.07	1.06	0.62	0.12
Rough grass	1.10	0.98	0.62	0.05
Forestry	1.53	1.40	1.41	0.22
Woodland	1.63	1.31	0.83	0.12
Wildscape	0.70	0.49	0.33	0.05

Table 17. Soil depth loss ( mm yr^-1) for land use categories and Erosion Risk Classes for Scotland.

Land use	Soil depth loss (mm yr-1)
	Erosion Risk Class
	Low	Moderate	High	Organo-Mineral soils and peats
Horticulture	0.26	0.30	0.27	4.11
Arable intensive	0.30	0.33	0.31	5.87
Arable extensive	0.16	0.17	0.17	2.31
Grassland improved	0.21	0.21	0.27	0.68
Grassland unimproved	0.13	0.18	0.50	0.38
Rough grassland	0.07	0.07	0.12	0.35
Forestry	0.04	0.05	0.07	0.08
Woodland	0.04	0.05	0.07	0.11
Wildscape	0.09	0.12	0.18	0.26

In order to calculate the impact of erosion on the loss of yield using the calculated soil depth losses, a production model was developed based on the proportion of crops in each land use / erosion risk class category. The area of those categories that were “unknown” in the spatial data were distributed to the known categories in proportion to their occurrence. The distribution was then simplified, as appropriate, to the key crops that were used in the modelling.

Crop yield data and crop sale values for these land use/ erosion risk class categories were developed from The Farm Management Handbook 2018/19 (SAC Consulting, 2018), cofunded by the Scottish Government and EU as part of the SRDP Farm Advisory Service. Grass yields for peat were taken from Morris et al. (2010) and the annual per hectare yield of timber was developed from Forestry Commission data for timber production and values (Haw, 2017).

6.4.2. Erosion induced nutrient loss

The mean organic carbon (C; % by weight of air-dried soil), nitrogen (N), phosphorus (P) and potassium (K) concentrations (mg kg^-1 of air-dried soil) of each of the Soil Erosion Risk Classes were calculated for each catchment and at the national scale from the data held within the Scottish Soils Knowledge and Information Base (SSKIB). This dataset comprises summary data derived from the soil profile data held within the Scottish Soils Database for the main Scottish Soil Series. Organic C content (% by weight of air-dried soil) for top soils was derived primarily by CN analyser or by dichromate digestion. Total P concentration was determined by colorimetric methods following sodium hydroxide fusion and using a Discrete Analyser. K concentrations (mg kg^-1) were determined by inductively coupled plasma emission spectrometry in a ammonium acetate solution. N concentrations (mg kg^-1 ) were mainly determined by CN analyser. The data to calculate nutrient loss was derived for each of the 5 individual catchments and for Scotland as a whole for the national scale erosion costings (Appendix 4 - 7). Results for C and nutrient contents at the national scale are shown in Table 18; Table 19; Table 20; and Table 21.

Table 18. The mean carbon content values ( kg t^-1 ) for Scotland

Land Use	Soil erosion risk class
	Low	Moderate	High	Organo-mineral soils and peats
Horticulture	43.33	35.82	30.19	283.19
Arable intensive	48.54	39.12	30.19	283.19
Arable extensive	50.71	48.23	41.64	352.05
Grassland improved	58.77	70.67	86.23	334.56
Grassland unimproved	87.46	137.31	91.91	299.90
Rough grassland	76.73	127.55	102.73	361.97
Forestry	56.74	124.53	155.77	467.11
Woodland	72.87	107.35	135.37	301.59
Wildscape	71.47	86.71	59.61	166.73

Table 19. The mean N content ( kg t^-1) for Scotland

Land Use	Soil erosion risk class
	Low	Moderate	High	Organo-mineral soils and peats
Horticulture	3.44	2.60	1.83	11.06
Arable intensive	3.87	2.94	1.83	11.06
Arable extensive	4.04	3.70	2.97	14.14
Grassland improved	4.65	5.44	5.96	14.01
Grassland unimproved	6.09	8.11	4.63	12.41
Rough grassland	5.09	7.95	5.76	14.66
Forestry	4.05	8.05	9.55	18.66
Woodland	5.34	7.05	8.41	12.34
Wildscape	4.16	4.65	3.12	6.73

Table 20. The mean P content ( kg t^-1) for Scotland

Land Use	Soil erosion risk class
	Low	Moderate	High	Organo-mineral soils and peats
Horticulture	1.19	0.85	0.76	0.58
Arable intensive	1.31	1.01	0.76	0.58
Arable extensive	1.36	1.25	1.05	0.82
Grassland improved	1.19	0.85	0.76	0.58
Grassland unimproved	1.41	1.63	1.30	0.73
Rough grassland	1.01	1.22	0.71	0.64
Forestry	1.03	1.31	1.15	0.85
Woodland	1.63	1.31	0.83	0.12
Wildscape	0.70	0.49	0.33	0.05

Table 21. The mean K content ( kg t^-1) for Scotland

Land Use	Soil erosion risk class
	Low	Moderate	High	Organomineral soils and peats
Horticulture	0.04	0.08	0.07	0.35
Arable intensive	0.13	0.10	0.07	0.35
Arable extensive	0.14	0.13	0.10	0.36
Grassland improved	0.15	0.16	0.17	0.30
Grassland unimproved	0.19	0.21	0.12	0.26
Rough grassland	0.14	0.20	0.15	0.31
Forestry	0.14	0.21	0.25	0.43
Woodland	0.17	0.20	0.22	0.28
Wildscape	0.11	0.11	0.08	0.13

Whilst Table 18 to Table 21 provide data on the quantity of C, N, P and K in the soil, it is worth noting that erosion selectively takes the most important components of the soil first and eroded soils can typically contain three times more nutrients than the soil left behind (Ali, Khan and Bhatti, 2006; Lal, 1998; Sharpley, 1980). Enrichment ratios measure the relative concentrations of nutrients in deposited material and in the soil from which that eroded material came. A range of enrichment ratios and the enrichment ratios used in this study are shown in Table 22. These were used to calculate the eventual quantity of C, N, P and K lost in the eroded material per annum for each of the land use / erosion risk class combinations. It should be noted that it was not possible to differentiate between the bioavailable and total fraction of P and K. The total soil nutrient content was used in the calculations.

Table 22. A range of enrichment ratios uses to show the proportion of nutrients in eroded soils relative to the parent soils and the enrichment ratios used in this study

Nutrient	Range of enrichment ratios	Mean ER
P	1.32 - 3.04 (Zheng et al., 2005) 1.47 (Sharpley, 1985) 2.79 (Ali et al., 2006)	2.15
OC / OMC	1.08 - 1.4 (Zheng et al., 2005) 0.9 - 2.6 (Schiettecatte et al., 2008) 2.00 (Sharpley, 1985) 1.23 (Ali et al., 2006)	1.56
N	0.89 - 1.26 (Zheng et al., 2005) 1.61(Sharpley, 1985) 1.43 (Ali et al., 2006)	1.37
K	2.90 (Ali et al., 2006)	2.90

The value of the soil nutrients was given by the SAC Farm Management Handbook (SAC Consulting, 2018; Table 23) and used to calculate the value of the N, P, K and C lost in the eroded material.

Table 23. The value of soil nutrients (from SAC Farm Management Handbook (2018/19))

Fertiliser cost of N 0.67 £ / kg

Fertiliser cost of P 0.68 £ / kg

Fertiliser cost of K 0.45 £ / kg

Fertiliser cost of C 0.67 £ / kg

6.5. Calculating annual off-site costs

As well as the on-site impacts of erosion, there are a variety of off-site costs incurred each year, associated with the cost of Greenhouse Gas Emissions, and costs associated with reduced environmental and drinking water quality associated with the eroded sediment and nutrients.

6.5.1. Greenhouse gas impact of soil erosion

Greenhouse gases such as methane (CH₄) nitrous oxide (N₂O) and carbon dioxide (CO₂) contribute to climate change. Several studies have estimated the per unit cost of this. For example, Eyre et al. (1997) proposed costs of £63 t^-1 for CO₂, £263 t^-1 for CH₄ and £7,530 t¹ for N₂O. Jacobs and SAC (2008) uses a value of £25 t^-1 CO₂ based on the shadow price of carbon proposed by Defra. More recently, the EU Emissions Trading System (ETS) has created a market in GHG emissions, which provides a traded carbon price based on abatement costs. Agricultural emissions currently exist outside the ETS, but the UK Government’s Department for Business, Energy and Industrial Strategy (BEIS) (BEIS, 2019) has advised that non-traded carbon prices (£68 t^-1 CO_2e for 2019) should be used in such cases. It should be noted that these value will increase in future years as carbon trading process increase (Tom McKenna, personal communication). Jacobs and SAC (2008) has provided ratios (21:1 for CH₄ and 310:1 for N₂O) to calculate global warming potential (GWP) for other GHG gases giving a per unit cost of £1,428 t^-1 CH₄ and £20,080 t^-1 N₂O.

However, in the present study, no data on methane or nitrous oxide emissions specifically linked to soil erosion were found. Some studies refer to management practices that will reduce GHG emissions whilst also controlling soil erosion (Jarecki and Lal, 2006; Johnson et al., 2007; Smith and Olesen, 2010). However, they do not quantify the amount of these GHGs that are associated with soil erosion. Ball (2013) draws on research from Scotland, Japan and New Zealand to examine how soil structures affected by wheel compaction, animal trampling, tillage and land-use change influence GHG emissions in order to help identify key controlling properties, but soil erosion is not mentioned explicitly.

Whilst many papers (e.g. Powlson et al. (2011) outline the importance of the stock of soil C in climate regulation, data on the proportion of C in eroded material that is released to the atmosphere to form CO₂ is also lacking. Lal (2003) provides an estimate at a global scale in which he proposes that approximately 20% of the soil C that is eroded each year is emitted to the atmosphere as CO₂. The ratio of organic C in the soil to CO₂ in the atmosphere is given as 3.67 (Williams, Audsley and Sandars, 2006), and the assumed quantity of eroded C emitted to the atmosphere was multiplied by this ratio to obtain its Global Warming Potential (t CO₂e). Using these assumptions, the annual cost of GHG emissions from eroded soils in Scotland was estimated to be £10.881 million (Table 24).

Table 24. The annual cost of GHG emissions from CO₂ emitted from eroded soils in Scotland (national scale)

Land use	GHG cost of soil C loss (£ per annum)
	Soil Erosion Risk Class
	Low	Moderate	High	Organo mineral soils and peats*	Total
Horticulture	55	528	40	62	685
Arable intensive	1,391	25,110	934	11,262	38,698
Arable extensive	26,312	425,092	16,046	302,657	770,107
Grassland improved	57,709	1,700,707	247,893	959,597	2,965,907
Grassland unimproved	677	55,880	122,885	435,721	615,163
Rough grassland	2,155	230,979	342,575	4,084,078	4,659,787
Forestry	1,236	147,471	131,185	817,902	1,097,794
Woodland	1,298	87,036	79,003	118,426	285,763
Wildscape	254	23,568	73,039	350,040	446,901
Total	91,088	2,696,371	1,013,601	7,079,747	10,880,806

* The calculations of all costs for arable and improved grassland on organo-mineral soils and peats may be an overestimate as there is a mismatch in scale between the soil spatial data set and the land use (see Section 8.6 for full explanation).

6.5.2. Impact on watercourses of nutrients lost in eroded material

Whilst the loss of nutrients in eroded material represents the loss of a productive asset, the removal of the N and P into water represent a risk to the quality of the water environment, the costs of which are born by third parties. A summary of those costs associated with eroded soils are shown in Table 25. Figures specific to Scotland were not available, so Table 25 uses data derived previously for England and Wales.

Table 25. Summary of economic data used to estimate off-site cost of soil erosion in the water environment (see sections below for further explanations)

			Value	Unit
Environmental water quality	Rivers, canals	NO3-N	190	£ t-1
	Freshwater lakes	P	1,407	£ t-1
	Transitional water	NO3-N	10	£ t-1
Drinking water quality	Nitrate	NO3-N	203	£ t-1
	Sediment	Soil	18	£ t soil-1

It is noticeable that data are only available for P in freshwater lakes and NO₃-N only in rivers, canals and transitional waters. This is discussed in Section 10.4.4 below.

6.5.3. Costs of N in rivers and canals

Damage costs to rivers and canals are linked to levels of NO₃-N and the apportionment to agriculture was estimated as £47.4 million (adjusted to 2009 values) in Jacobs and SAC (2008). Since NO₃-N data are not given in Jacobs and SAC (2008), the total of 295,409 t NO₃-N for agriculture in the UK was taken from Defra project WQ0106 (Anthony et al., 2009) giving a ‘per unit’ damage cost of £161 t^-1 NO₃-N. This value was inflated to 2019 values using GDP inflators from BEIS (2019) to obtain a cost of £190 t^-1 NO₃-N.

The off-site damage cost associated with N in the eroded soil in rivers and canals at the national scale was estimated to be £1.858 million per annum (Table 26).

Table 26. The annual costs of N in rivers and canals (£) in Scotland

Land use	Cost of N in rivers and canals (£ per annum)
	Erosion Risk Classes
	Low	Moderate	High	Organomineral soils and peats	Total
Horticulture	14.79	132.10	9.60	8.01	164
Arable intensive	371.48	6,283.13	223.06	1,462.97	8,341
Arable extensive	7,055.54	109,118.05	3,830.23	40,651.08	160,655
Grassland improved	15,362.91	437,726.36	57,343.83	134,270.66	644,704
Grassland unimproved	158.41	11,034.69	20,687.11	60,255.20	92,135
Rough grassland	501.78	48,752.73	64,396.64	553,452.02	667,103
Forestry	294.55	31,887.70	26,885.45	109,197.57	168,265
Woodland	318.07	19,150.08	16,417.44	16,196.62	52,082
Wildscape	49.53	4,219.93	12,763.58	47,242.96	64,276
Total	24,127	668,305	202,557	962,737	1,857,726

6.5.4. Sediment removal by dredging rivers and canals

No figures are available specifically for Scotland. In England and Wales, Anthony et al. (2009) divided the annual costs of dredging sediment out of water courses (£9.8 million in 2009) by the total sediment load (1,906,260 t yr^-1) to give a unit cost of £5.2 t^-1 sediment. This value was inflated in the present project to 2019 values using GDP inflators from BEIS 2019 to obtain a cost of £6.1 t^-1 eroded material.

From this, the off-site damage cost associated with sediment removal from water courses in Scotland was estimated to be £4.733 million per annum (Table 27).

Table 27. The annual costs of sediment removal in Scottish rivers and canals (£)

Land use	Cost of sediment removal in rivers & canals (£ per annum) Erosion Risk Classes				Total
	Low	Moderate	High	Organo-Mineral soils and peats
Horticulture	75	882	64	14	1,036
Arable intensive	1,891	41,966	1,484	2,64	47,986
Arable extensive	34,468	582,662	25,483	56,73	699,350
Grassland improved	65,303	1,593,534	191,011	189,430	2,039,278
Grassland unimproved	515	26,923	88,380	96,06	211,885
Rough grassland	2,286	132,466	221,584	768,540	1,124,876
Forestry	1,454	78,197	55,601	115,826	251,079
Woodland	1,178	53,682	38,620	25,98	119,464
Wildscape	238	17,702	80,917	138,889	237,745
Totals	107,408	2,528,013	703,145	1,394,13	4,732,699

6.5.5. Costs of P loads in lakes / lochs

Graves et al. (2011) estimate the cost of eutrophication in lakes in the UK as £62.6 million

(adjusted to 2009 values) and linked this to an estimated P load of 45% from agriculture (Jacobs and SAC, 2008), giving an apportionment of £28.2 million to agriculture. It should be noted that the apportionment to agriculture varies significantly amongst different sources and has been reported as 29.2% in Defra Project WT0701CSF (White and Hammond, 2006) (Table 28). Since physical units for P are not given in Jacobs and SAC (2008), the total load given for the UK in Defra Project WT0701CSF (43,796 t yr^-1) is used as a proxy to develop a ‘per unit cost’ for phosphorus (£ t^-1 P) from the total cost in Jacobs and SAC (2008), giving a ‘per unit cost’ of £1,407 t^-1 P (Table 25) that can be attributed to the risk of P in lakes / lochs, lost from soils via erosion.

The off-site damage cost associated with the P in eroded soil (sediment) in lakes was estimated to be £3.708 million per annum (Table 29).

Table 28. Total phosphorus input from agricultural, household, industrial and background sources to the waters of England, Scotland, Wales and N. Ireland (source: White and Hammond, Defra project WT0701CSF)

	England	Wales	Scotland	N Ireland	UK
	t yr-1	t yr-1	t yr-1	t yr-1	t yr-1
Agriculture	8,391	2,151	1,223	1,023a	12,788
Household	21,677	1,333	2,246	988b	26,245
Industrial	645	70	1,179	40	1,934
Background	1,561	243	902	123	2,829
Total	32,274	3,797	5,551	2,174	43,796
Percentage contributions (%)
Agriculture	26	56.6	22	47.1	29.2
Household	67.2	35.1	40.5	45.4	59.9
Industrial	2	1.8	21.2	1.8	4.4
Background	4.8	6.4	16.2	5.7	6.5

a Adjusted for rural population (-107 t TP yr^-1) as described by Smith et al. (2005)

b Adjusted for P removal programme (-145 t TP yr^-1) as described by Smith et al. (2005)

Table 29. The annual costs of P in freshwater lakes (£) (Scotland)

Land use	Cost of P in fresh water lakes (£ per annum)
	Erosion Risk Class
	Low	Moderate	High	Organomineral soils and peats	Total
Horticulture	69	622	46	6	743
Arable intensive	1,731	29,572	1,075	1,059	33,437
Arable extensive	32,698	506,520	18,458	32,394	590,070
Grassland improved	64,147	1,792,235	171,749	95,635	2,123,765
Grassland unimproved	527	24,770	27,989	35,602	88,888
Rough grassland	1,619	108,677	108,871	330,570	549,738
Forestry	1,035	71,127	44,388	67,937	184,487
Woodland	1,176	49,740	26,324	9,715	86,954
Wildscape	110	6,683	18,047	25,094	49,934
Total	103,112	2,589,944	416,947	598,012	3,708,016

6.5.6. Costs of N loads in transitional waters

The costs of damage by soil erosion to transitional waters are also linked to nitrate and are estimated from Jacobs and SAC (2008) to be to be £79,147 yr^-1 (adjusted to 2019 values) for Scotland. The physical data developed by Defra project WQ0106 (Anthony et al., 2009) for nitrate losses by agriculture in England and Wales was used to obtain the ‘per unit’ value of £8.9 t^-1 NO₃-N. This value was inflated to 2019 values using GDP inflators from BEIS 2019 to obtain a cost of £10 t^-1 NO₃-N (Table 25).

The off-site damage cost associated with N in the eroded soil (sediment) in transitional waters was estimated to be £97,775 per annum (Table 30).

Table 30 The annual costs of N in transitional waters (£) (Scotland)

Land use	Cost of N in transitional waters (£ per annum)
	Erosion Risk Classes
	Low	Moderate	High	Organomineral soils and peats	Total
Horticulture	1	7	1	0	9
Arable intensive	20	331	12	77	439
Arable extensive	371	5,743	202	2,140	8,456
Grassland improved	809	23,038	3,018	7,067	33,932
Grassland unimproved	8	581	1,089	3,171	4,849
Rough grassland	26	2,566	3,389	29,129	35,111
Forestry	16	1,678	1,415	5,747	8,856
Woodland	17	1,008	864	852	2,741
Wildscape	3	222	672	2,486	3,383
Total	1,270	35,174	10,661	50,670	97,775

6.5.7. Nitrate removal in drinking water

Removal of nitrate in freshwater imposes costs on water companies. The agricultural apportionment of this is assumed to be 61% giving a total cost (adjusted to 2009 values) of £51 million (Jacobs and SAC, 2008). The agricultural load of NO₃-N (295,409 t yr^-1) estimated for England and Wales by Anthony et al. (2009) is used to derive the per unit cost (£ 172 t^-1 NO₃-N) for removal of nitrate from drinking water (in that case associated with livestock applications on grassland, but the removal costs will be the same, whatever the mechanism of getting the nitrate into the sources of drinking water). This value was inflated to 2019 values using GDP inflators from BEIS 2019 to obtain a cost of £203 t^-1 NO₃-N (Table 25).

The off-site damage cost associated with the N in eroded soil in drinking water was estimated to be £1.985 million per annum (Table 31).

Table 31. The annual costs of N removal in drinking water (£) (Scotland)

Land use	Cost of N in drinking water (£ per annum)
	Erosion Risk Class
	Low	Moderate	High	Organo mineral soils and peats	Total
Horticulture	16	141	10	9	176
Arable intensive	397	6,713	238	1,563	8,911
Arable extensive	7,538	116,584	4,092	43,432	171,647
Grassland improved	16,414	467,676	61,267	143,458	688,815
Grassland unimproved	169	11,790	22,103	64,378	98,439
Rough grassland	536	52,088	68,803	591,320	712,747
Forestry	315	34,069	28,725	116,669	179,778
Woodland	340	20,460	17,541	17,305	55,646
Wildscape	53	4,509	13,637	50,475	68,674
Total	25,778	714,031	216,416	1,028,609	1,984,833

6.5.8. Sediment removal in drinking water

The total cost of sediment removal is assumed to be 50% of OFWAT's 'other' expenditure (Jacobs and SAC, 2008). This gives a total cost of sediment removal of £38.6 million. Agriculture is assumed to contribute about 75% of the total sediment load in watercourses (Collins et al., 2012), suggesting a cost of £29 million to remove this sediment. If this number is divided by the total load of sediment from agriculture (given as 1,906 kt yr^-1 by Anthony et al. (2009) for England and Wales (in the absence of equivalent figures specifically for Scotland)), this gives a per unit cost of £15.4 t^-1 sediment. This value was inflated to 2019 values using GDP inflators from BEIS 2019 to obtain a cost of £18 t^-1 erosion.

The off-site damage cost associated with sediment removal in drinking water was estimated to be £16.542 million per annum (Table 32).

Table 32. The annual costs of sediment removal in drinking water (£) (Scotland)

Land use	Cost of sediment removal in drinking water (£ per annum)
	Erosion Risk Classes
	Low	Moderate	High	Organomineral soils and peats	Total
Horticulture	263	3,084	223	51	3,621
Arable intensive	6,610	146,677	5,187	9,245	167,719
Arable extensive	120,470	2,036,487	89,067	198,306	2,444,330
Grassland improved	228,243	5,569,634	667,612	662,085	7,127,575
Grassland unimproved	1,800	94,098	308,902	335,767	740,568
Rough grassland	7,989	462,989	774,468	2,686,159	3,931,605
Forestry	5,083	273,309	194,335	404,830	877,557
Woodland	4,118	187,626	134,984	90,816	417,543
Wildscape	831	61,872	282,815	485,436	830,955
Totals	375,407	8,835,775	2,457,594	4,872,696	16,541,473