Appendix 1.
Characterisation of phosphorus sources in rural catchments
1. Introduction
Inputs of phosphorus (P) to surface waters are a major concern for the developed countries due to the increased incidence of eutrophication leading to a range of undesirable impacts including impairment of human health (algal toxins), reduced biodiversity of aquatic species, reduction in amenity value and increased costs of water treatment for drinking. This widespread concern has resulted in the establishment of nutrient criteria and targets to help define the extent of water quality impairment and quantify the necessary reductions in P inputs required for ecologically diverse and healthy waterbodies (U.S. Environmental Protection Agency, 2001; Granlund et al., 2005; Mainstone et al., 2008). The targeting of measures to reduce P inputs to water requires accurate identification of the main contributing sources, knowledge of their distribution in the catchment and an understanding of their mode of delivery to the watercourse. Historically, P controls have been primarily targeted at large municipal and industrial discharges but experience from a number of countries with eutrophication problems suggests that reductions in other anthropogenic sources of P derived from the wider catchment area are also required (Foy et al., 1995; Carpenter et al., 1998; Kronvang et al., 2005; Environment Agency, 2007).
Sources of P entering surface waters have traditionally been grouped into point and non-point (or diffuse) (Novotny and Olem, 1994). This is a convenient grouping that allows non-point sources to be quantified as the difference between a measured total P load at a catchment outlet minus the sum of the larger point sources. The latter are usually storm independent, flowing continuously or semi-continuously from single points, require some form of consent or permit, and are therefore routinely monitored and quantifiable. Some definitions of point sources also include runoff from farmyards or animal feedlots (e.g. Carpenter et al., 1998), but these are storm dependent and not routinely monitored. Non-point sources of P originate from a number of different upstream areas and scattered delivery points (critical source areas). They are more episodic in nature and therefore temporally and spatially very variable. Carpenter et al. (1998) included runoff from agricultural (forested, cultivated and pasture) land, urban runoff, continuous discharges from small rural sewage treatment works (STW), leakages from septic tanks and atmospheric deposition over a water surface within the non-point group. In reality, discharges and runoff entering surface waters have varied and overlapping hydrological and compositional characteristics which often makes their simple grouping difficult (Edwards and Withers, 2008), although attempts have been made (e.g. OSPAR Harmonised Reporting Procedures for Nutrients (HARP-NUT) Guidelines (OSPAR, 2004). Additional sources of P also arise due to a range of biotic and abiotic processes which re-mobilize particulate and dissolved P from within the stream channel and connected open ditches (Svendsen et al., 1995; Withers and Jarvie, 2008).
Numerous national and specific catchment studies have quantified the relative importance of point and non-point P sources in terms of their contribution to the annual P load using a variety of different approaches (Johnes et al., 1996; Pieterse et al., 2003; European Environment Agency, 2005; Nemery et al., 2005; Smith et al., 2005; Kronvang et al., 2007; Bowes et al., 2008). Due to the problems of quantifying sources that are not routinely monitored, many of these studies adopt simple groupings of sources (e.g. background, point, diffuse), often using derived data that can be related back to specific catchment characteristics (e.g. population, forestry, crops and livestock). Whilst providing some tentative strategic direction to P loss reduction, these approaches do not identify the specific contributing sources or source areas responsible for P delivery in catchments, or provide information on the localized ecological relevance of the delivered P, and as such are an oversimplification. For example, non-point source area contributions from agricultural land are spatially variable and highly seasonal and will therefore have varying ecological impacts depending on the timing and concentrations of the P delivered and within-channel retention, especially for localized impacts in flowing waters with low residence times (Svendsen et al., 1995; Edwards et al., 2000; Jarvie et al., 2006).
Current approaches to source apportionment therefore naturally bias P loss reduction strategies towards the wastewater and agricultural sectors, which are viewed as the main contributors of point and non-point P loads, respectively. Consequently, less attention has been paid to P derived from sources/source areas other than large STWs and farmed land, such as septic tanks and impervious surfaces (farmyards, farm tracks and highways), which have hydrological and chemical properties that are ‘intermediate’ between point and non-point (Edwards and Withers, 2008). Discharges and runoff from these ‘intermediate’ source/source areas can contribute P under both low and high flow conditions more continuously during the year and are therefore potentially more ecologically damaging than runoff from farmed land, even though they may not contribute as large a proportion of the annual P load (Hively et al., 2005; Arnscheidt et al., 2007, Jarvie et al., 2008). For example, water quality impairment associated with increased nutrient (and other contaminant) loadings can occur when as little as 10-15% of the catchment area is occupied by impervious surfaces, even in small rural catchments dominated by agriculture (Wang et al., 2000; Roy et al., 2003). Dudley and May (2007) concluded that rural septic tanks can contribute up to 20% of TP loads in some catchments.
Water quality in lowland rivers is heavily dependent on the diluting potential of headwater streams in rural areas dominated by agricultural land, yet there is very little comparative information on the abundance and importance of the multiple sources of P entering headwaters. For example, ignoring the contribution from impervious surfaces and septic tanks might result in little or no water quality improvement downstream if only farming practices are considered responsible for eutrophication in rural areas. For example, using high frequency sampling, Jordan et al. (2007) showed that whilst storm runoff from managed pasture was the main source of large episodic P exports in a rural catchment in Ireland, ambient stream P concentrations between events were being sustained by much smaller chronic and acute inputs of P which were storm independent. From studies on a Danish river system, Svendsen et al. (1995) concluded that P loading estimates based only on knowledge of agricultural land area and soil type could be misleading because of the contribution of P from scattered dwellings not connected to the main sewerage system.
Some preliminary investigations on the range in nutrient concentrations in discharges and runoff associated with various individual sources and source areas contributing P directly to streams draining rural micro (7-10 km2) catchments are reported. These catchments are dominated by agricultural land and have no major (>10,000 per capita population) municipal or industrial discharges, but nevertheless have some degree of habitation associated with rural (village) communities. The aim of the study was to provide some information on the complexity and interaction of the different sources/source areas contributing P to these catchments and consider their potential relevance for local stream ecology.
2. Materials and methods
2.1. Study catchments
Three sets of near-adjacent catchments were located in lowland river basins: the Herefordshire Wye, Hampshire Avon and Leicestershire Welland (Fig. 1). These river basins were chosen because they are representative of the major geology, soil types and agricultural and rural land use of large areas of lowland Britain. Each set of near-adjacent streams included both a stream draining low agricultural intensity land-use and one or more streams draining higher intensity agricultural land use. Near-adjacent streams were chosen to ensure similar baseline catchment characteristics, including catchment area, soils, underlying geology and rainfall patterns. Streams draining low agricultural intensity land uses were generally characterised by grassland and woodland without major human settlement to provide a ‘background’ or ‘control’ against which to assess impacts of more intensive agricultural activity. Streams draining high intensity agricultural land uses included a range of arable and livestock farming systems and differed in the degree of agricultural intensification, either in respect of the proportion of cultivated land, the extent of field underdrainage and/or in P inputs and average soil P fertility (Table 1). Each catchment was visited on a number of occasions to collect farm and catchment information and take soil samples from fields representative of soil types, land use history and P inputs for analysis of soil physical and chemical characteristics and soil P fertility (Olsen-extractable P, OP). Where farmers would not co-operate, land use information was obtained by visual inspection of the catchment. Full land use information was therefore obtained in each catchment to identify the proportion of cultivated land, grassland, woodland and other uses (farm buildings and urban areas), Table 1.
2.1.1. Welland
Digby Farm and Belton Bridge catchments are located on steeply sloping soils developed over Lias clay with ironstone in Leicestershire (Denchworth Association and related soils). The catchment tributaries feed into the headwaters of the Eye Brook, which feeds into the Eye Brook Reservoir and on to the R. Welland. The clayey and fine loamy over clayey soils are slowly permeable and seasonally waterlogged and runoff water is often finely turbid. Most fields have underdrainage systems and ditches are common.
· Digby Farm (DF) is a 0.44 km2 catchment with low intensity agriculture. The majority of fields are in permanent pasture used for dairy/sheep grazing and silage production receiving ca. 11-20 kg P ha-1 y-1 of compound fertiliser each year and small amounts of cattle and sheep FYM (ca. 15-20 kg ha-1) in either spring or autumn. Soil fertility is generally deficient or low with concentrations of OP ranging from 3-23 mg L-1. The highest OP concentration represents one winter cereal field receiving larger manure inputs. The land is owned by three farmers and there are 4 residents on a septic tank system within the catchment.
· Belton Bridge (BB) is a 1.5 km2 catchment with higher intensity mixed agriculture. Fields in the south/west of the catchment are in continuous arable production (winter cereals/oilseed rape) but receive no P fertilizer. Much of the cereal land is minimally cultivated. Fields in the north-east of the catchment are in ley grassland grazed by sheep, spring cereals and stubble turnips which receive fertilizer (5-11 kg P ha-1) and sheep manure (15-20 kg P ha-1) annually. A small area in the north-east of the catchment which was not surveyed has low quality permanent pasture grazed by horses. Soil OP concentrations are relatively low (6-26 mg L-1). There are three farms and ca. 15 residents who are all on septic tank systems. Just downstream of Belton Bridge is the village of Loddington.
Figure 1. Map showing the location of study sites and catchments.
1
Table 1. General characteristics of the study catchments including annual P inputs.
River / Area / Rainfall1 / Dominant / Farming system / Land use (%) / P inputs5 / Soil P6Catchment / km2 / mm / soil types2 / Wood / Arable3 / Pasture / Other4 / kg ha-1 / mg L-1
Welland
Digby Farm (DF) / 0.44 / 689 / Irondown / Low intensity
dairy and sheep / 3 / 4 / 90 / 3 / f, 11-20
m, 15-20 / 10
(3-23)
Belton Bridge
(BB) / 1.48 / 671 / Denchworth / Intensive arable
and sheep / 20 / 47 / 32 / 0 / f, 5-11
m, 15-20 / 11
(6-26)
Avon
Cools Cottage
(CC) / 1.57 / 905 / Bromsgrove Whickham / Low intensity beef with some arable / 42 / 15 / 41 / 2 / f, 0
m, 10-15 / 19
(8-30)
Priors Farm
(PF) / 4.65 / 864 / Whickham Denchworth / Intensive dairy,
beef and sheep / 9 / 10 / 70 / 10 / f, 5-44
m, 8-76 / 21
(5-46)
Wye
Whitchurch
(WC) / 6.46 / 812 / Eardiston / Low intensity arable,
beef and sheep / 9 / 23 / 60 / 8 / f, <5
m, 15 / 14
(3-33)
Dinedor
(DD) / 8.69 / 694 / Bromyard Eardiston / Intensive arable, beef and sheep / 19 / 53 / 23 / 5 / f, 11-75
m, 15-75 / 15
(3-70)
Kivernoll
(KN) / 9.87 / 737 / Bromyard Eardiston / Intensive arable, and poultry + STW / 13 / 68 / 11 / 7 / f, 26-92
m, 75 / 35
(11-57)
1Long-term (30 yr) annual rainfall. 2Soil Association (Soil Survey of England and Wales, 1984). 3Including ley-arable. 4Urban areas and farmyards. 5Fertilizers, f; manures, m; when applied. 6Average (range) of Olsen-P concentrations in fields sampled for fertilizer recommendations (MAFF, 2000).
1
2.1.2. Hampshire Avon
Cools Cottage and Priors Farm catchments are predominantly located on seasonally waterlogged fine loamy over clay, or clayey, soil developed in Kimmeridge clay (Whickham and Denchworth Associations) in Wiltshire and used for livestock farming. The catchments are located in very close proximity to one another in the extreme west of the Hampshire Avon river basin and are part of the Sem tributary that feeds into the R. Nadder near Salisbury (Fig. 1). Most fields are either flat or only gently sloping and the heavy clay soils are under-drained to varying degrees of efficiency with the result that the land lies wet during winter.
· Cools Cottage (CC) is a 1.6 km2 catchment containing largely woodland and low intensity permanent pasture grazed by beef cattle and calves in summer. There is only a small area of cereal production on some lighter sandstone soils (Bromsgrove Association) that form the more steeply sloping upper perimeter land in the catchment. However, much of the upper part of the catchment drains into a lake within the woodland that provides the source of the main tributary and is therefore isolated from the flatter land in the lower part of the catchment. Little or no P fertiliser is currently applied and only small amounts of winter-stored manure P are recycled annually (10-15 kg P ha-1). However, soil P fertility is not always low (OP, 8-30 mg L-1) suggesting greater P inputs have been applied in the past. There are 2 farms and the estimated 10 residents are all on septic tank systems.
· Priors Farm (PF) is a 4.7 km2 catchment which has historically been in high intensity grassland for dairy production. In recent years there has been a reversion to slightly less intensive dairy and beef farming but with inclusion of forage maize and winter cereals to provide winter feed. There are now only two large dairy farms in the catchment. Livestock have direct access to the stream in a number of places, there are a number of cattle crossing points and farmsteads are located in close to the stream. Nutrient inputs in cow slurry and farmyard manure (FYM) (8-76 kg ha-1 y-1) and P fertiliser (5-44 kg P ha-1 y-1) are relatively high and soil OP concentrations range from 5-46 mg L-1 in sampled fields. There are 10 farms and the resident catchment population is estimated at 515 of which only ca. 115 (22%) have septic tank systems. The remainder of the population is on mains drainage linked to a STW at East Knoyle in the northern part of the catchment. In the north of the catchment there is a wetland that collects road runoff and possibly some sporadic discharge from the STW.