Integrated catchment management for reducing pesticide levels in water: Engaging with stakeholders in East Anglia to tackle metaldehyde.

In the agriculture intensive eastern region of England, plant protection products are widely applied to protect crops such as wheat and oilseed rape from pests and diseases, thus creating a risk of reaching nearby water courses through surface runoff. The EU Drinking Water Directive sets a stringent limit of 0.1 μg/l and 0.5 μg/l for individual and total pesticides respectively in treated potable water. However, peak metaldehyde levels have been persistently detected in raw water and reducing them to these limits has proven challenging and costly, in particular when using conventional treatment. In line with the EU Water Framework Directive, a more suitable approach and one adopted by the local water company, Anglian Water Services Ltd., would require moving towards mitigating pollution at source, preferably through participative action with multiple stakeholders in the agricultural industry. Initial findings demonstrate the potential of product substitution for reducing metaldehyde levels in surface waters. Reviewing Anglian Water's "Slug it Out" trial, we discuss key learnings derived from their experiences and make recommendations about the potential of the catchment approach to address the wider pesticide challenge.


H I G H L I G H T S
• Regulatory risk of pesticides reaching waters through surface runoff • Metaldehyde detected in raw waters and difficult to remove for potable use • Engaging farmers to reduce pollution at source preferable to end-of-pipe treatment • Product substitution trial effective in reducing metaldehyde levels in water • Further investigation for upscaling to address wider pesticide challenge

G R A P H I C A L A B S T R A C T
a b s t r a c t a r t i c l e i n f o

Introduction
The past decade has shown increasing pressures in England's surface waters intended for potable use due to the presence of pesticides. The coupled with ozonation (Davey et al., 2011). Although water treatment works have been effective in the removal of pollutants from raw water, ensuring their capacity to address new or emerging contaminants has been technically and economically challenging (Dillon et al., 2011). Furthermore, upgrading treatment to remove these contaminants would only protect people from exposure as an end-of-pipe solution, with their sources and pathways in the environment almost entirely neglected.
The need for a more holistic approach to water resources management triggered a reform in water policies and the EU Council Directive 2000/60/EC Water Framework Directive (WFD), the embodiment of the integrated catchment management approach (ICM), was introduced in 2000 (Giakoumis and Voulvoulis, 2018). Article 7 of the WFD obligates Member States to depart from end-of-pipe treatment and adopt a prevention-led approach to achieve DWD compliance (Dolan, 2013;Dolan et al., 2013). As a preliminary mitigation measure, surface and ground waters that are intended for potable consumption were established as Drinking Water Protected Areas (DrWPAs) as required by Article 6 of the WFD. Field level scale pesticide risks maps were also produced, to identify areas of high risk. In England, 486 surface water bodies are classified as DrWPAs, with 122 of these being identified as being 'at risk' of pesticide contamination (Bell, 2015;Gulson, 2015). In addition, 'Safeguard Zones' have also been established at the upstream parts of DrWPAs catchments where raw water at these sites is deemed to be 'at risk' of being polluted. Out of the 118 Safeguard Zones established due to pesticide pressures, 96 were surface water and 22 groundwater (Bell, 2015;The Pesticides Forum, 2015). Metaldehyde presents the most significant risk on non-compliance for pesticides in 102 DrWPAs (Castle et al., 2017;The Pesticides Forum, 2015), with approximately 1893 t of metaldehyde used in Great Britain between (Fera, 2018. Metaldehyde is a molluscicide that is widely applied by farmers to land in the autumn and winter months when molluscs thrive in the wet weather conditions, to protect crops from slug damages (Castle et al., 2017). It was first introduced in 1936 and used in slug baits in the early 1940s (Edwards et al., 2009). Metaldehyde is typically applied onto agricultural land using spinning disc applicators that are mounted on quad bikes (Castle et al., 2017). The guideline provided by the Metaldehyde Stewardship Group (2018) is a maximum application for an individual dose rate of 210 g of metaldehyde/ha (particularly from August to the end of December every year), a recommended reduced rate of 160 g for additional protection of water. Metaldehyde pellets are also not allowed to fall within 10 m of any field boundaries or watercourse (Metaldehyde Stewardship Group, 2018). A maximum total dose rate of 700 g/ha in a calendar year was also set as the statutory legal requirement for metaldehyde application (Environment Agency, 2016).
Metaldehyde was first detected in raw water in 2007 (Bristol Water, 2008;Water Briefing, 2018a) and across the UK the following year, resulting from the development of new analytical techniques, it was included in most water companies' routine pesticide monitoring programmes of raw water abstractions and treated water. Since then it has been identified intermittently in rivers and reservoirs in England, particularly those that traverse intensively farmed arable land (Castle et al., 2017). In the wetter autumn and winter months, elevated levels have been frequently observed in surface waters and reservoirs exceeding the DWD limits (Dillon et al., 2013;Lu et al., 2017).
The compound is soluble and highly mobile in soil, making it likely to move via surface runoff towards nearby water bodies (Castle et al., 2017;University of Hertfordshire, 2018). Once it reaches water bodies, it stabilises and will persist for a long time (Davey et al., 2011). Metaldehyde removal at conventional WTW that utilises granular activated carbon filtration is relatively poor, especially at higher concentrations due to its low tendency to attach to organic carbon (Castle et al., 2017;Davey et al., 2013;Dillon et al., 2011;Tao and Fletcher, 2016;University of Hertfordshire, 2018). As a result, its removal from potable water at peak concentrations has been challenging for water companies (Dillon et al., 2011;Dolan, 2013). Furthermore, water companies in England and Wales were issued with 'Undertakings' (Section 19, Water Industry Act 1991) by the Drinking Water Inspectorate to take an ICM approach to meet the DWD standards and report the findings to this regulatory body by 31 March 2020. Currently, 16 water companies have been issued with undertakings for metaldehyde and 15 for total pesticides (Drinking Water Inspectorate, 2018), to ensure the protection of water quality at DrWPAs and therefore reduce the level of purification treatment required in the production of drinking water.
To understand the role of ICM in addressing the pesticide challenge, here we investigate the potential of mitigating pollution at source through ICM and focus on the Anglian region as a case study. The strategies and measures to deal with metaldehyde undertaken by the local water company, Anglian Water Services Ltd., are summarised, and key learnings derived from their experiences are discussed to make recommendations for their wider application to address the pesticide challenge.

ICM to address the pesticide challenge
Based on the overall principles of the WFD, ICM provides a favourable opportunity to address the pesticide challenge (Dolan, 2013). Fenemor et al. (2011) define ICM as "A process that recognises the catchment as the appropriate organising unit for understanding and managing ecosystem processes in a context that includes social, economic and political considerations, and guides communities towards an agreed vision of sustainable natural resource management in their catchment." Adding on to this definition, Fenemor and Bowden (2001) also stated that the implementation of ICM approaches requires collaboration between an interdisciplinary team comprising of scientists, resource managers, policy makers and community representatives. There is no 'onesize-fits-all' approach to ICM, as even the WFD is not a prescriptive piece of legislation (Baffoe-Bonnie et al., 2012). This enables flexibility in the development of strategies tailored for specific catchments. In England and Wales, the application of ICM to meet the DWD standards is driven by legislative requirements through 'improvement programmes' issued by the Drinking Water Inspectorate. Developing ICM strategies to tackle diffuse pollution requires a combination of economic incentives, legal arrangements and voluntary behaviour (Blackstock et al., 2010). This approach requires shifting from a top-down, technocratic and exclusionary style of water governance (Benson et al., 2014;Fritsch and Benson, 2013;Smith et al., 2015;Surridge et al., 2010), and moving towards mitigating pollution at source through participative action by multiple stakeholders at all levels (Voulvoulis et al., 2017). This is also sanctioned by Article 14 of the WFD, which outlines consultation, access to information and active involvement as the requirements for public participation in water resources management (European Commission, 2000;Van der Heijden and Heuvelhof, 2012).
For the water sector, engaging with stakeholders in the farming community within their respective catchments offered the opportunity to prompt behavioural changes to ultimately reduce diffuse water pollution from agriculture at its source. The evidence for this had also come from some initial catchment-based and voluntary programmes launched in the UK prior to 2008 for the protection of water resources such as the Voluntary Initiative and Catchment Sensitive Farming. The Voluntary Initiative was introduced in 2001 as a delivery mechanism for the UK's National Action Plan for the Sustainable Use of Pesticides Directive (Directive 2009/128/EC). It applies Integrated Pest Management to protect crops, wildlife, pollinators and reduce the risks of diffuse water pollution from agriculture. The Voluntary Initiative also partnered with other initiatives such as the Catchment Sensitive Farming, which was launched in December 2005 and delivered by the Environment Agency and Natural England. The aim of Catchment Sensitive Farming was to engage with the agricultural community and promote the voluntary uptake of good farming practices; including good practices for pesticide management and handling. The delivery of the Catchment Sensitive Farming was focused in DrWPAs within 50 Priority Catchments in England. Following the detection of metaldehyde peaks in 2008, the Metaldehyde Stewardship Group (an industry led voluntary initiative) was also formed to promote best practices to minimise the environmental risks of metaldehyde. These existing initiatives provided further support for the regulatory drive to promote the application of ICM to address the pesticide challenge. They resulted in a reduced application of metaldehyde between 2008 and 2012, shown as evidence of a downward trend in metaldehyde levels in water, with seasonal peaks in raw and treated waters (Davey et al., 2013;Water UK, 2013). However, conditions of high slug pressures have continued to influence farmers to resort to applying slug pellets. For example, such an incident in 2012 following a mild and wet summer resulted in peak concentrations of metaldehyde in rivers and reservoirs in the autumn of that year, exceeding the initial peaks detected till then (Lu et al., 2017;Water UK, 2013).
Furthermore, the significant costs of removing metaldehyde from water and the technical challenges for meeting the DWD standards for potable water (Fig. 1) meant that a new way of addressing the metaldehyde challenge is necessary. Engaging with stakeholders in the farming community within defined catchment boundaries could potentially lead to behavioural changes that can ultimately reduce diffuse water pollution from agriculture. The principle is simple: reducing metaldehyde use at its source in agriculture would result to improvements in water quality downstream of a catchment. While this implies that total regulatory restriction on the availability of metaldehyde would maximise such improvements, this option has been estimated to cause significant losses to farming income (The Andersons Centre, 2014;Vrain, 2015). Approximately 22% and 59% of the total area of wheat and oilseed rape in the UK are affected by slugs, which is projected into an estimated yield loss of £43.5 million (Nicholls, 2014) if there are absolutely no means of controlling slug pressures.
Following the withdrawal of approval for methiocarb in December 2013 (Jess et al., 2014), the only approved alternative for metaldehyde is ferric phosphate (Castle et al., 2017). Five main ferric phosphate products are available in the UK market; Sluxx, Sluxx HP and Derrex from Certis, Ironmax Pro from DeSangosse Ltd. and Slugger from Goldengrass Ltd. Ferric Phosphate is less soluble in water and therefore less likely to end up in watercourses when it rains (University of Hertfordshire, 2018), significantly reducing the risk of WFD and DWD noncompliance.
Therefore, getting farmers to replace metaldehyde with ferric phosphate would be a good outcome for the catchment approach. However, this is not without its complexities and uncertainties, especially when taking into consideration matters of scalability (spatial, temporal and financial), feasibility and sustainability of the schemes; among others (Bratt, 2002;Castle et al., 2017;Collins et al., 2016;Davey et al., 2013;Dolan et al., 2014;Smith et al., 2015;Tippett et al., 2005;Vrain, 2015). For example, farmers are not entirely convinced on ferric phosphate's efficacy and its relatively higher cost hinders their preference for this alternative molluscicide (Davey et al., 2013;Dolan, 2013). Ferric Phosphate causes slugs to die underground (AHDB, 2017), leading to the perception that the product is less effective than metaldehyde due to the lack of evidence of dead slugs on the surface (Davey et al., 2013). Table 1 shows the main differences between metaldehyde and ferric phosphate.
Following a similar product substitution approach 1 trialled in 2008 by Wessex Water Services Ltd. based in the southwest region of England, that resulted in an almost complete removal of metaldehyde from the catchment, with only one exceedance since October 2014 (Wessex Water, 2015, the opportunity of metaldehyde substitution by ferric phosphate was investigated by Anglian Water Services Ltd. In 2015, 'Slug It Out' was launched to address the metaldehyde challenge in the East Anglia region.

Methods and materials
We follow a systematic process for examining this complex public policy challenge, reviewing available evidence from a case study Hydrogen Peroxide (H2O2) to produce highly reactive hydroxyl radicals. This has been shown to be effective in the degradation of metaldehyde (Li et al., 2017;Autin et al., 2013;Drogui and Lafrance, 2012), however, it is a costly and energy intensive process (Tao and Fletcher, 2016;Anglian Water Services Ltd., 2016a). Applying this technology across East Anglia is estimated at £600 million to build with an additional £17 million to operate per year, reflecting into a 21% increase in water customers' bills (Anglian Water Services Ltd., 2016a; Price, 2016). Photo source: BBC News, 2014 (with permission from Anglian Water Services Ltd.) where the implementation of a particular option was adopted by the water industry. Policy analysis uses a variety of tools to develop relevant information and present it to the parties involved in the policymaking process in a manner that helps them come to a decision. It is a problem-oriented approach that does not presume a model structure for assessing the consequences of a policy or ranking the alternatives, but offers valuable insight from available evidence using case studies. Following Yin's (1994) three categories of case studies, this research follows the 'explanatory method' 2 whereby ICM intervention at the source of pollution is theorised to demonstrate a reduction in metaldehyde levels in raw surface water. This theory is subsequently tested at seven natural catchments in the Anglian region (see Section 3). The data presented in this research are secondary data from Anglian Water to demonstrate the impacts of ICM interventions on metaldehyde levels in surface waters. Findings and key learnings derived from the case study shall provide a baseline to inform future research towards developing a framework for the application of ICM in the water sector.

Determination of metaldehyde risks for the water sector
Water companies in England and Wales are required to monitor pesticide levels in raw water intended for human consumption as sanctioned by Article 7 of the DWD (1998) and Part 4 of the Water Supply (Water Quality) Regulations 2016. Metaldehyde concentration in raw and treated water samples were determined at Anglian Water's laboratories by high performance liquid chromatography triple quadrupole tandem mass spectrometry (LC-QqQ) (Anglian Water, 2018a). The Company's procedures were tailored based on the Standing Committee of Analysts' Blue Books which outlines the methods for the examination of waters and associated materials (Environment Agency, 2009a).
The frequency of metaldehyde sampling and monitoring increases during the high-risk period (Price, 2018) between 1st August to 31st January each year, defined by Anglian Water's procedural document (Olsson, 2018). This represents site specific conditions of higher metaldehyde application rates on agricultural land during this period. Additional metaldehyde monitoring may be carried out where extra vigilance is required during lower risk periods. For example, in events of high slug pressures, on-going application or high rainfall (Olsson, 2018). Slug risk on agricultural land can be determined through slug trapping, ideally before cultivation, during mild weather conditions (5-25°C) and when the surface is visibly moist (AHDB, 2017). Slug traps are left overnight with bait underneath, and the number of slugs is counted the following morning while the soil surface is still moist (AHDB, 2017). Based on AHDB's (2017) guideline for slug trapping, 3 a catch of 4 or more slugs per trap for winter cereal crops and oilseed rape that are established in standing cereals are indicative of a possible slug burden risk. For oilseed rape in cereal stubbles, potatoes and field vegetables, the threshold for slug risk is 1 or more slugs per trap (AHDB, 2017).

Study area
The Anglian region is located on the east of England and Anglian Water Services Ltd. (herewith Anglian Water) is the primary provider of potable water and water recycling services in the region (Fig. 2). Their service catchment excluding the town of Hartlepool (Anglian Water, 2018b), lies within the Anglian River Basin District (Article 3; European Commission, 2000). They supply 1.2 billion l of potable water per day to more than six million customers in their service area (Anglian Water, 2018b).
Predominantly rural with heavy clay soils underlying its surface, ideal for the cultivation of winter wheat and oilseed rape (Dolan et al., 2014), the region is one of the most productive agricultural landscapes in the world (Environment Agency, 2009b) with 1.38 million ha of total farmed areas in the East of England. 4 Arable crops make up 78% of the farmed areas with cereals (54%) and oilseed rape (9%) being the main crops grown in this region (Defra, 2018).
The Anglian region receives b700 mm of rainfall per year, making it one of the driest regions in the UK (Leathes et al., 2008;Met Office, 2016;Wynn et al., 2015). Adding to the pressure on the availability of water resources is population growth. Corke and Wood's (2009) 20year projection showed that the population and number of households 2 Through the explanatory case studies, researchers explain a phenomenon in both surface and deep level by first forming a theory and subsequently testing it (Yin, 1994;Hossieni et al., 2012).   in the Anglian region are expected to increase to 6.8 million and 3 million respectively by 2026. This will create higher demands for potable water in an already water-stressed region. Furthermore, implementing alternative water resources management options such as abstraction management, dilution or closure of water treatment works during peak metaldehyde incidents is less desirable as it can add more pressure to available water resources. Therefore, addressing the metaldehyde challenge is not only a regulatory obligation, but it is also to ensure the security of water resources for current and future consumers in the Anglian region.

Slug it out: ICM for metaldehyde in the Anglian region
In England, metaldehyde use is the highest in the eastern region 5 (Fera, 2018). Fig. 3 shows the distribution of metaldehyde application in the region by area and by total weight over time. Despite the prominent drop in metaldehyde application by total weight in 2010, potentially due to nation-wide campaigns such as the Voluntary Initiative, Catchment Sensitive Farming and Metaldehyde Stewardship Group, the total areas treated with metaldehyde in this region remain relatively high. This meant that the presence of metaldehyde in raw surface water prior to any form of catchment intervention is inevitable, some of which may persist in soils and drains until resurfaced due to soil preparation of rainfall events.
Due to the vast scale of the Anglian region, detailed catchment modelling was carried out by Anglian Water to identify areas of high and low priority for catchment interventions to address the peak metaldehyde levels in their catchments. The modelling process considered three environmental characteristics: soil type (high risk for clay or drained soils), steep sloped areas (high risk for gradients N 3°) and proximity to watercourses (high risk for areas b 200 m). Ground-truthing on site was also carried out to validate the models, where Catchment Advisors would carry out a catchment walkover and surveys on site to determine any other sources of runoff (e.g. unmarked ditches and surface or sub-surface drains). Nineham et al. (2015) also modelled several scenarios for ICM intervention to determine their effectiveness on reducing metaldehyde concentrations. Their model indicated the most effective ICM measure is product substitution in areas of heavy, drained soils and areas that are within 500 m of water courses (Nineham et al., 2015). As a result, seven natural hydrological catchments feeding into major water abstraction reservoirs were identified as the high priority areas. These natural catchments are the Alton, Ardleigh, Grafham, Pitsford, Ravensthorpe and Hollowell, and Rutland Catchments (Fig. 4). The local landscapes across these priority areas are variable, ranging from rural catchments such as at Ravensthorpe and Hollowell, to those containing large villages near Alton, Grafham, Pitsford and Rutland, and townships at Ardleigh. The characteristics of these natural catchments are summarised in Table 2.
Seasonal metaldehyde peaks during the autumn and winter months were observed at all catchments in this region exceeding the regulatory standards of 0.1 μg/l, typically following rainfall with wet soil conditions and higher slug burdens. The metaldehyde levels in surface water in this region between 2008 and 2015 are shown in the supporting materials attached in Appendix A (see Figs. A1 to A5). The surface water presented in these figures is the rivers and streams that either flow naturally or pumped into major reservoirs in the Anglian region.
To oversee the execution of the Slug it Out product substitution trial, six Catchment Advisors with qualifications, knowledge and experience in the agricultural sector were employed by Anglian Water. Blackstock et al. (2010) highlighted that there is a higher persuasion factor if key messages are delivered through a credible and trusted source such as advisors with farming background. Engaging with stakeholders in a common language and communication style that can be understood and accepted by both parties Webler and Tuler, 2006) enabled the company to break the communication barriers between the water sector and the farming community and establish an understanding of the needs and priorities of both sectors. Furthermore, the Catchment Advisors' local knowledge of their respective catchments and ability to empathise with the farmers' needs and problems (Ingram, 2008) also presents their credibility to gain the farmers' confidence and trust (Blackstock et al., 2010). The Catchment Advisors must take into account that the farmers' decisions to participate in such schemes are driven by financial considerations, climatic factors, practicality of the mitigation measures, potential risks to their crops and the efficacy of  the measures or alternative products to protect their crops (Castle et al., 2017;Dolan et al., 2014). In addition to this, there is limited evidence to show the success of product substitution or local voluntary bans in reducing metaldehyde concentrations in surface waters (Davey et al., 2013).
Although several farmers voiced out their reservations towards the efficacy of ferric phosphate, they were willing to trial it in the first year if compensated for the cost difference. Therefore, farmers were incentivised for their commitment to not use any form of metaldehyde on the areas of their land which lie within the targeted catchment areas. Three separate payments comprising of a hosting payment, payment for the cost difference between ferric phosphate and 3% dry metaldehyde pellets, and a water quality bonus were offered to participating farmers. The hosting payment of £2.25/ha with a guaranteed minimum of £250 was provided as compensation for the farmers' time as well as payment for farmers to provide Catchment Advisors access to information on the farm, their slug management records and access to the fields for monitoring slug risks and water quality. The payment for the cost difference of £15/ha mitigated the full price difference between 3% dry metaldehyde and ferric phosphate. Finally, a water quality bonus of £2.25/ha was provided as an incentive to the farmers in natural catchments. To receive this bonus, the surface waters that flow naturally into the reservoirs must successfully achieve metaldehyde levels of 0.1 μg/l throughout the duration of the trial. The scheme's payment structure is shown in Fig. 5 while the planning and implementation structure of Slug it Out is shown in Fig. 6. The metaldehyde levels in the natural catchment reservoirs were not accounted for the bonus. They are not representative of the effects of ICM interventions as some reservoirs also receive raw water through pumped catchments. Metaldehyde peaks at pumped catchments are presently managed through abstraction management (turning off the pumps at the intakes). Furthermore, the reservoirs already retain historic metaldehyde which takes a long time to degrade.

Preliminary implementation results and policy findings
In the first year of Slug it Out, 89 farmers operating in 7642 ha of arable farms participated in the scheme. The addition of the Rutland catchment in the second year increased the total number of participating farmers to 123, covering 10,050 ha of arable land. It is estimated that over 2000 t of metaldehyde were offset in the second year (Anglian . During the trial period, the Catchment Advisors ensured regular engagement with the farmers in their catchments. This was to maintain their working relationships and to keep the farmers informed and updated on the trial's progress, water quality, and other activities carried out on their land such as water sampling and slug trapping. Communication took place through one-to-one conversations, e-mails, telephone calls, text messages, newsletters, events, social media and websites (Anglian Water, 2016b). While there were a few farmers that were initially apprehensive about participating in Slug it Out, the relationships established with the farming community, word-of-mouth and 'peer pressure' from neighbouring farmers resulted in them deciding to participate as well. Farmer uptake of Slug it Out was reached 100% in all target catchments in for the first two years of the trial. A network of 167 agronomists was also engaged by Anglian during the second year (Anglian . Having the support and endorsement of local agronomists for Slug it Out increased farmers' confidence towards the trial and ferric phosphate use, as farmers place a high level of trust in their agronomists' advice (Anglian . This was particularly evident in cases where agronomists voluntarily applied ferric phosphate onto their land to show the efficacy of the product. Peak metaldehyde levels were still observed in the natural catchments, especially during the high-risk season between 1st August and 31st January. However, the second year of the trial has shown vast improvements in the reduction of metaldehyde levels. Figs. B1 to B6 in Appendix B show the metaldehyde levels in Slug it Out catchments before and after the implementation of the trial. The pumped catchments were included into the graphs as a baseline to show metaldehyde levels in areas with no ICM interventions in place. In the case of the Grafham catchment (Fig. B3, Appendix B), the Diddington Brook natural catchment monitoring point was introduced following the identification of the catchment as high priority during the catchment modelling phase. Peak incidents of metaldehyde levels exceeding the DWD standards occurring during the trial were investigated by the Catchment Advisors. These investigations have shown that the peaks were not always to be related to farmer use. For example, the Ardleigh catchment is significantly connected to the city of Colchester with local industries operating in the area. Investigations were carried out following elevated levels of metaldehyde in the Northern Salary Brook and Western Salary  Brook natural catchments between August 2015 and January 2016 (Fig. B2, Appendix B). These investigations pointed towards the potential of non-agricultural sources of metaldehyde such as those from domestic users and effluents from industrial areas that may contain acetaldehyde. The latter is a base material in the synthesis of both metaldehyde and products such as ethyl acetate, perfumes, flavourings, aniline dyes, plastics, synthetic rubber, and other chemical compounds (Anglian Water, 2016b). A full investigation on these external sources of metaldehyde, the extent of their effects in surface water and how to mitigate them were proposed in the later stages of the trial. In other catchments, there were strong indications of legacy metaldehyde remaining in drains or soils from previous applications and reactivated following rainfall or resurfaced from soil management. Rainfall data were also observed at all Slug it Out catchments to observe the correlation between rainfall events, the likelihood of surface runoff in fields and metaldehyde risk in surface water (Anglian Water, 2017). Sources of metaldehyde from storage areas and spreading equipment can also act as a secondary pollution source if it is brought into land in sensitive land areas for storage and handling. Feedback from participating farmers was collected at the beginning and the end of each trial year. Overall, Slug it Out received strong support due to its simple design during the initial engagement and implementation phase. The Catchment Advisor's support, regular communication and ability to provide technical advice were appreciated by the farmers. The farmers' general perception towards the efficacy of ferric phosphate was positive. All farmers in the Slug It Out catchments were satisfied that ferric phosphate is as equally effective as metaldehyde. In some instances, their positive experiences with the product and knowledge gained from the trial has influenced the farmers to use ferric phosphate on land outside of the natural catchments. Several farmers with fields that cut through catchment boundaries opted to use ferric phosphate across all their lands to simplify application logistics. In Grafham, some farmers chose to apply the funding provided to adopt cultural controls. Enhanced efforts were made to produce a fine, consolidated seed bed such as double rolling and surface cultivations such as harrowing. This resulted in one farmer not needing to apply any slug pellets to fields that would previously have received applications. The organic credentials of ferric phosphate also appealed to farmers involved in fruit production in the Ardleigh catchment. However, despite the above, the price difference remained a major factor in their choice of slug pellets. Several farmers have indicated that they would revert to metaldehyde in the absence of subsidies.

Discussion
Although ICM is not a new concept, its regulatory push with the implementation of the WFD has brought a new way of thinking for both EU water regulators and industry. The development of an ICM strategy requires a combination of economic incentives, legal arrangements and, in most cases, the adoption of mitigation measures and best management practices calling for voluntary behaviour (Blackstock et al., 2010). The Slug it Out product substitution trial led by Anglian Water, demonstrated that water quality in raw surface water can be improved, and in some instances, it has the potential to achieve DWD compliance as observed at the Alton, Ravensthorpe and Hollowell catchments, with the complete removal of agricultural sources of metaldehyde. Nineham et al. (2015) 6 reported that priority areas across the subcatchments are variable and need to be selected based on crop type, soil type, source of water intake (pumped vs. natural catchments), and metaldehyde concentrations in the tributaries feeding into the reservoirs. For example, the River Colne pumped catchment is a dominant source of metaldehyde that feeds water into the Ardleigh reservoir (Anglian Water, 2016b;Nineham et al., 2015). This catchment is also predominantly underlain by low permeability soils, contributing a rapid pathway for metaldehyde through the catchment. Measures targeting pumped catchments are therefore a priority.
ICM opens up opportunities to tackle diffuse water pollution, however planning and implementing its measures can be complex and challenging, particularly where stakeholders are involved. A critical step for the success of ICM was shown to be establishing a connection and trust between the Catchment Advisors and farmers. By being able to have an open discussion, a wider range of water quality matters related to agriculture were also communicated positively. This supported the view that the success of an ICM scheme requires catchment managers to have a good understanding of local issues (McGonigle et al., 2014). Having Catchment Advisors who are already familiar with the catchments and can speak the agricultural language is key to effective communication and knowledge exchange with the farmers. The role of agronomists towards shaping the farmers' decisions must also be incorporated into a stakeholder engagement strategy as their opinion of ferric phosphate will be reflected on the field. In addition, the local knowledge and expert opinions from agronomists would provide valuable insight to identify appropriate catchment interventions measures for farmers and to inform development of future ICM strategies to address other Diffuse Water Pollution from Agriculture beyond metaldehyde (Anglian Dolan et al., 2014). Overall, Catchment Advisors must carefully strategise how they would approach farmers and encourage the uptake of a product substitution trial. A detailed framework outlining the requirements for effective stakeholder engagement can serve as a reference point by catchment practitioners to address diffuse water pollution.
Although the likelihood of behaviour change among farmers is difficult to quantify, the parameters that influenced their decisions, predominantly cost and efficacy, were identified and utilised as motivators to facilitate behaviour change. To gain a better understanding of the farmers' perceptions on ferric phosphate and the trial, each participating farmer was visited before and after the trial to gather their feedback and opinions. Strong support was received for the simple design and implementation of the Slug it Out scheme, making it "easy to be involved". Technical advice and support on the use of ferric phosphate was appreciated particularly where farmers had historic experiences of using metaldehyde pellets. Regular communication about local metaldehyde levels during the season through emails and newsletters was valued by farmers and made it easy to engage with local success.
Overall, while the potential of the approach is clear, as land use surrounding the catchments increases in scale and complexity, regulatory compliance cannot be guaranteed. The catchment modelling process indicated that over 80% of the natural catchments would require product substitution to achieve regulatory compliance (Nineham et al., 2015). However, the Slug It Out Trial demonstrated that metaldehyde can still be present in surface waters above the regulatory limits even at 100% uptake across the natural catchments, due to external or nonagricultural sources of metaldehyde. Sources of metaldehyde from storage areas and spreading equipment can also act as a secondary pollution source if it is brought into a sensitive land area for storage or loading. A complete picture of these less conventional sources of metaldehyde is not yet fully understood but as ICM advances this is expected to change.
Slug pressures during the trial period were lower compared to previous years due to the relatively dry and mild weather conditions. The warm weather also enabled crops to establish rapidly, therefore reducing the need for secondary product application. While the trial showed long term climatic factors and slug burden not to be easily forecasted, again as ICM advances better communication between the company and farmers can also deliver further benefits. Early warnings of diffuse run-off from fields during excessive rainfall events, and other Catchment Sensitive Farming initiatives such as rural sustainable drainage systems that slow down or prevent the transport of pollutants to watercourses and better pesticide storage and management are examples of such benefits.
In 2015, the number of reported peak metaldehyde incidents significantly dropped to five compared to twenty-one reported incidents in the previous year (The Pesticides Forum, 2016). The incidents were mitigated through precautionary closures of the water intakes at each respective water treatment works. These incidents further demonstrate the potential of ICM in effective source management. But as it is the price difference of the products that ultimately determines most of the farmers' choice of pellets, the economic aspects are also critical. According to Davey et al. (2013), financial incentives may be necessary to overcome cost barriers as reliance on voluntary actions alone to reduce metaldehyde use is unlikely to be sufficient. Although the financial model for Slug it Out provided an extra incentive for farmers to participate in this scheme, and there are clear benefits with initiating such schemes, the question is who bares the costs of mitigating metaldehyde in the long run. Who should be ultimately accountable for the protection of water resourcesthe regulators, water sector, agricultural sector, or consumers?
There is clearly a competing interest between the provisions of public goods and natural capital. End users of drinking water and food supplies will be affected by the higher cost of implementing ICM measures by the water sector or the higher cost of agricultural practices. Furthermore, current ICM measures may be feasible in smaller sub-catchments, but scaling-up across the wider catchments may not be economically sustainable for water companies. On top of the cost of recruiting catchment advisors, Anglian Water has been paying £15.00/ha to mitigate the cost difference between 3% dry metaldehyde and ferric phosphate, while similarly, Severn Trent's Farm to Tap scheme launched in July 2018, is offering farmers up to £8/ha for reducing metaldehyde levels found in their water sources (NFU, 2017;Severn Trent Water, 2016;Swire, 2018;Water Briefing, 2018b). At present, there is an economic argument for the water sector to cover the cost of the mitigation measures, but it may be economically unfeasible to scale up these across the whole region.
Incentivisation at the beginning of product substitution trials may be necessary to gauge interest in participation and build farmers' confidence in the efficacy of ferric phosphate. But, a progressive transition towards a voluntary switch to ferric phosphate should be the end goal of a product substitution trial. Ideally, the implementation of cultural controls to reduce slug burden should be prioritised, and ferric phosphate could be applied when necessary. Demonstrating the benefits of this strategy could increase farmers' interest in making long-term behavioural changes. Some of these benefits include reducing cost, improving soil health, reducing farmers' reliance on chemicals for pest control (Drogui and Lafrance, 2012) and building the evidence base needed to show the efficacy of ferric phosphate (Ricardo-AEA, 2014). However, there is insufficient evidence to show the economic benefits of product substitution combined with cultural controls to mitigate slug burden which could lead to farmers fully take on the cost of the desired change. Therefore, there is a need to build this evidence base to redesign the financial model for product substitution trials implemented by Anglian Water, Wessex Water, Severn Trent Water and others in England and Wales to minimise the economic impacts of pesticide management on the water companies and end users. This should also include costs relative to scale (spatial, temporal, and socio-economic) and other ICM interventions required to address the wider pesticide challenge.
With herbicides such as clopyralid and propyzamide similarly difficult to remove via conventional treatment and without any alternatives in the market, this could be an even more difficult challenge to address. In addition, considering that metaldehyde as an active substance has been used since the 1930s, while efforts to address its levels in water have been in place due its detection since 2007, this leads to a question about what other chemicals are present in drinking water resources and might be undetected because we do not look for them.
And this is where ICM delivers. Transparency in agricultural practices and engaging the agricultural community into ICM initiatives can create a more holistic appreciation of land (SuRCaSE, 2009) which recognises the interconnectedness between land and water systems as well as links between the social and environmental systems (Surridge et al., 2010). In addition, involving local stakeholders into ICM initiatives develops their social commitment towards improving their practices (Surridge et al., 2010), creates a mutual understanding of the challenges within the catchment (European Communities, 2003), as well as build the stakeholders' capacity to collaborate effectively . Although environmental improvements might entail additional costs, they often deliver 'win-win' outcomes. Where there are perceived costs associated with environmental improvements, it is often because the costs fall disproportionately on one sector or party while the benefits are shared. Through ICM environmental and economic benefits can be achieved at the same time, and this approach can deliver multiple benefits.
Although product substitution trials do no fully embody an ICM approach, they are the beginning of a learning process towards working together with others to build on opportunities and reduce risks. Anglian Water engaging with the farming community has opened up conversations on the wider pesticide challenge. This gave rise to the potential of mitigating diffuse water pollution from agriculture more holistically and enabling the company to gain more benefits for water quality. Ultimately, catchment managers as well as regulators must also recognise that ICM is often not a short-term solution that provides immediate results. Rather, it is a continuous learning process between all stakeholder groups. As Fenemor et al. (2011) highlighted, the successful implementation of ICM requires ongoing stakeholder participation in an "adaptive management process". This process requires extensive monitoring data to provide an evidence-base on decision making and inform improvements on mitigation measures to ensure long term benefits for water quality. Surridge et al. (2010) highlighted that a key challenge in the implementation of the ICM is the disconnection between governments and authorities at the regional and national scales with local stakeholders operating at the smaller scale; thus widening the gap on the necessary evidence required to shape the decisions made for the future of water policies. Bridging this evidence gap is a requirement for policy-makers to design targeted and effective catchment interventions (McGonigle et al., 2012). Such evidence should include the priorities of the stakeholders, experiences and key learnings from ICM interventions and trials, and more importantly, the required logistics, funding mechanism and human capital requirements to implement ICM schemes.
ICM ultimately requires for land and water to be managed in an integrated way, by identifying the pressures on the water environment, recognising the potential for conflict between the interests of users and working together to agree common objectives and implement solutions. It takes time, so it is critical through initiatives such as the ones discussed here to establish structures and processes that support collaborative working and help to build trust. Over time, stronger collaboration to improve the management of the catchment should mean that people are more willing to share power, risks and ownership of the process which does pay off to all involved in the end.