Introduction
To combat eutrophication of surface waters, the EU Water Framework Directive (Directive 2000/60/EC) prescribes restoration or enhancement of the chemical and ecological status of water bodies. Because agriculture is currently considered a major contributor to phosphorus (P) loading of surface waters (e.g., HELCOM
2009), agricultural sources need to be included in P mitigation plans.
The nutrient loads from agriculture are affected by enterprise type and intensity. A site with high soil P concentrations and a hydrological connection to a stream through a ditch network or tile drainage system will contribute to the total P loads in a body of water to much greater extent than reflected by its proportion of catchment area. Relatively small (hot-spot) areas develop, for example, in places that receive continuous inputs of manure, such as feedlots, around dairy houses and around water stations in pastures (e.g., Weld et al.
2001; Page et al.
2005).
Dissolved P concentrations in runoff tend to be particularly elevated as a result of long-term P accumulation in soil. The dissolved P pool is generally considered to be totally available for biological utilization (Ekholm and Krogerus
2003), whereas much of the P present as particulate P (PP) may be practically inert in the short term (Sonzogni et al.
1982). Particulate P can partly enter the biological cycle under special circumstances (Lehtoranta et al.
2015), but when aiming to limit the amount of P readily available to aquatic algae, dissolved P is logically the first priority for remedial actions because it can immediately trigger algal growth in P-limited waters.
Management of soil P stocks, limiting mineral fertiliser and manure/slurry inputs and fencing off streams from livestock have been identified as the most cost-effective ways to reduce losses of dissolved P, whereas the use of soil amendments and edge-of-field measures is associated with distinctly higher abatement costs (McDowell and Nash
2012). However, when treating runoff from hot-spot areas the use of higher-cost measures can be justified, because P input management has a long lag time between implementation and effect. Potential options for mitigating hot-spot P losses include decreasing P mobilization or intercepting transported P using chemicals or a reactive medium as a soil amendment, as an edge-of-field P-trap, or in ditches that connect to larger waterways.
In particular, the use of solid by-products as P filter materials has attracted strong interest in recent years (Brooks et al.
2000; Vohla et al.
2011; Buda et al.
2012). Although most published studies involving P sorbents are restricted to P retention measurements in the laboratory, some field data on the use of different types of materials are also available (McDowell et al.
2007; Penn et al.
2007; Shipitalo et al.
2012). At best, P sorbents can substantially decrease P concentrations and loads to adjacent waterways. However, in regions with a long winter period associated with soil frost, snow accumulation and rapid snowmelt that carries a substantial proportion of the annual P load to watercourses within a short period (Rekolainen
1989), such filters would be still frozen during the time of the high P losses in early spring.
Another option could be to decrease bioavailability of P by adding chemicals that react with dissolved P to runoff water, making the P less readily available for freshwater algae and bacteria (Neufeld and Thodos
1969; Närvänen et al.
2008). Metal salts such as ferric or aluminum sulfates or chlorides, which are widely used in water and wastewater purification, could be used for this purpose. They can be regarded as a proven option in engineered treatment facilities that allow optimization of their use efficiency, but when used to treat agricultural runoff they would be applied in very different environmental settings.
We examined the use of a simple type of ferric sulfate dispenser for reducing the concentrations of dissolved P in stream runoff, using as test sites 15 agricultural ditches in SW Finland. The objective was to evaluate the P reduction effect in variable field conditions, the practicality of the method and the cost of converting dissolved P to a sparingly available form. The starting hypothesis was that the method is suitable for use in early spring, when the soil is frozen and natural P retention processes are inactive.
Discussion
This study explored the potential of a relatively easily adoptable method for dispensing precipitation chemical to stream water in order to decrease the availability to water organisms of dissolved P in agricultural runoff. Precipitation chemicals can be administered to water directly, which makes them a potential way of targeting snowmelt runoff, which is a major carrier of P to watercourses in agricultural regions at northerly latitudes (Rekolainen
1989). The early phases of snowmelt in particular may contain very high dissolved P concentrations as a result of P release from grass vegetation and surface soil (Rekolainen
1989; Uusi-Kämppa
2005). These intensive runoff peaks usually take place before the soil thaws, at which time methods such as vegetated buffers, wetlands, buried or edge-of-field P traps and traditional soil management-based P mitigation measures can be expected to perform poorly.
Ferric sulfate was applied at doses designed to bring about a notable decrease in dissolved P concentration, with the actual amount being based on previous tests (Närvänen et al.
2008). Addition of the chemical did not remove P from the water, but converted it to an iron-associated form that is presumably sparingly available to microbes and algae (Li and Brett
2013). This could suppress the growth of algae and other biomass in the receiving waters, and enhance the settling of P bound in Fe–P associations to sediments in ditches, streams or in lakes. However, these Fe–P associations remain in the aquatic system and may be subjected to re-dissolution over time, for example as a result of sediment anoxia.
There were no significant changes in total P concentrations in the paired samples taken from water upstream and downstream of the dispenser units. This is because floc formation of the solid matter (and the associated particulate P) did not effectively take place with the relatively small ferric sulfate additions used. Moreover, there was no substantial accumulation of iron sludge below the sampling points downstream of the dispenser units at any of the study sites. If removal of particulate P is desired the dose should be higher, and the system would need to be optimized for that by adjusting pH, mixing and settling, sludge removal, etc. All this is relatively easily arranged in waterworks and wastewater treatment plants, but it would be a challenge to achieve in ditches at field margins. Sedimentation ponds could be used to collect sludge, but utilization of the precipitated P would then be a challenge, because P stripped by metal salts is considered uneconomic to recover for recycling (de-Bashan and Bashan
2004).
From an environmental point of view, any P entering waters should be bound in associations that have a low bioavailability, rather than being present in dissolved form that triggers growth of algae and bacteria. Overall, the results obtained at most of the test sites used in this study indicated a marked reduction, of on average about 60–70 %, in dissolved P by the ferric sulfate dispenser method. Compared with the ability of established P mitigation measures (e.g., conservation tillage, buffer zones, constructed wetlands) to decrease dissolved P concentrations, this is an impressive reduction and occurs immediately. Effects of similar magnitude have been reported in studies on P retention media (McDowell et al.
2007; Penn et al.
2007; Kirkkala et al.
2012; Penn et al.
2012), but such P barriers would still be frozen during the start of spring flow peak in the conditions of the present study.
Despite the high effectiveness displayed, the ferric sulfate dispenser method should be selectively used, in the first instance at sites identified by previous surveys as having high dissolved P concentrations. The P concentrations occurring in the water have a direct effect on the economics of chemical use. In the present study, the estimated cost of conversion of 1 kg dissolved P to a less bioavailable form ranged from approximately EUR 20 to EUR 400. These cost estimates can be taken as indicative, as the assessment of P fluxes was uncertain. If correct, the lower end of these estimates is similar to the average P abatement cost for wastewater treatment plants reported by Hautakangas et al. (
2014) in their analysis of 182 treatment plants in the Baltic Sea region (EUR 15–20 kg
−1 for an abatement target of 40–80 % of total P load). Another P mitigation measure that is comparatively low cost is gypsum application. Iho and Laukkanen (
2012) calculated the annual costs of applying about 4 Mg gypsum per hectare to a field in south-west Finland based on data in a catchment study by Ekholm et al. (
2012) and concluded that at an annual price of EUR 73, the field hectare would achieve the full P mitigation potential from gypsum use, i.e., about 65 % reduction in total P loss. A total P load of 2 kg ha
−1 can be expected from a high-P soil and thus gypsum amendment would cost about EUR 50–60 kg
−1 total P. Iho and Laukkanen (
2012) related their results to the estimated marginal damage from P losses (150 EUR kg
−1 P; Kosenius
2010), which is close to the average abatement cost estimated e.g., for Nautela. The upper end of the cost estimates in the present study (~400 EUR kg
−1) agree with those reported by McDowell and Nash (
2012) for in-field and edge-of-field measures targeting dissolved P in Australia and New Zealand.
With regard to practical applicability, issues with ferric sulfate dosing were apparent at some test sites in this study. The effect of the platform failure at the Nautela site, resulting in a very acidic pulse of water, was potentially serious. Dissolution of ferric sulfate acidifies water, i.e., leads to a drop in pH, because its reaction with water results in formation of ferric hydroxides and the liberation of protons from water molecules in the reaction: Fe2SO4 + 4H2O → 2FeOOH + 6H+ + SO4
2−. The target pH change in the present study was 0.5–1.0 pH unit, and this or a smaller change was generally achieved most of the time, but some exceptions were recorded. At Nautela, the chemical hopper was rather small in relation to peak flow volumes and heavy rain occasionally emptied the dispenser. Having a small chemical hopper may be regarded as a safety precaution, but the downside is the burden of refilling the unit more frequently.
Apart from the collapse of the dispenser station at Nautela, at some other sites snow and ice hindered water flow over the v-notch weir and caused any ferric sulfate left in the dispenser to dissolve in the small volume of ponded water created during the rise in water level above the weir, resulting in a distinct drop in pH. These incidents also indicate that small ditches should be preferred over larger streams, because any possible overdosing will only have a local impact on the small ditch and the water would be diluted when it joined a larger stream.
Other difficulties in the operation of the units arose from the hygroscopic nature of the granulated chemical during periods with high air moisture, typically in autumn. A combination of low flow and damp air caused ferric sulfate clumping that in some cases blocked the doser outlet. This problem was discussed with the manufacturer of Ferix-3, which has tested different types of coating materials to overcome problems associated with hygroscopicity, but currently there seems to be no appropriate coating materials available. Coating of the chemical grains would also increase the price, which is an important criterion in chemical procurement by the main users, i.e., water and wastewater treatment plants.