Implementing agricultural phosphorus science and management to combat eutrophication

The challenges of mitigating diffuse phosphorus (P) pollution are manifold, but no more complex than in the arena of implementing P-based management in agricultural watersheds. Phosphorus-based practices and the strategies that guide the implementation of these practices, once considered novel in the 1990s, have now been tried across North America and Europe, providing an ever-growing wealth of experience. These experiences highlight the iterative interaction of applied science and social experimentation that comes from trying to modify fundamental aspects of our food production and conservation systems.

The science and practice of implementing nutrient management strategies are often disconnected in watershed management, even though both are key to the success or failure of watershed remediation. It is clear that the science of understanding how P management affects water quality and the implementation of management practices via voluntary and coercive means are mutually dependent. Science provides justifications and narratives to underscore or drive implementation processes. Implementation guides or constrains the range of options considered by applied science. At its best, this tautology, or feedback loop, represents the process of adaptive management. However, one must recognize that this reinforcing process invariably absorbs assumptions and biases that are unrecognized by those involved.

Both “sacred cows” and their converse, “sacrificial lambs” are regularly encountered in the science and implementation of P management. Sacred cows can be found in the assumptions that certain environmental processes and management practices are so established that they are left unquestioned or perennially advocated. Sacrificial lambs, on the other hand, may occur as phenomena that are underestimated in their importance to the fate and transport of P or as practices that are readily condemned without due assessment. Recognizing the presence of these biases is required for objective and sustainable management solutions.

Even without sacred cows and sacrificial lambs, the challenges facing P management can vary dramatically, dependent upon local production systems, physiography, history, culture and politics, and, of course, economics and policies. Although there is a long history of watershed programs grappling with P management, efforts to confront diffuse sources of P, i.e., non-point sources of P, are often placed on the periphery, or margin, of other agricultural and conservation initiatives. When traditional conservation programs are insufficient to control diffuse P losses, watershed P problems are often described as novel (e.g., dissolved P loadings via tile drains) or unforeseen (legacy sources of P). In part this marginalization reflects the secondary nature of P as a plant nutrient, when compared with nitrogen (N). However, as we will seek to illustrate, the ubiquity of P sources and the degree of P management required for successful watershed outcomes are often underestimated (willfully or inadvertently).

To understand successes and challenges in P-based management, we review case studies from North America and Europe, delving into the challenges and opportunities associated with voluntary and regulated approaches to implementation. In general, North American experiences have emphasized voluntary adoption, whereas European experiences have followed regulations. Even so, all shades of coercion can be found on both continents. Beginning with the recent, highly publicized case of Lake Erie, whose resurgent water quality problems point to the vexing nature of P-based management, we review case studies with obvious, and less obvious, sacred cows and sacrificial lambs, highlighting uncertainty, successes, and factors affecting strategic and tactical outcomes.

Agricultural P management has re-emerged as a priority concern in Lake Erie (Fig. 1), one of the Great Lakes bordering the USA and Canada and the site of historical successes in P mitigation. In 2014, prevailing winds directed a cyanobacterial bloom from Western Lake Erie into the drinking water intake for the City of Toledo, causing a spike in the toxin microcystin that overwhelmed the treatment facility. Toledo had to halt water supply to 400 000 users, prompting calls for strict regulations on agricultural P, which has so far been subject to voluntary management (White 2014).

Today’s dissolved P concerns in Western Lake Erie contrast with the historical success of point and non-point source P control programs in helping to lower P loads to the lake (Richards et al. 2009). From 1975 to 1995, loads of total P from the two largest watershed inputs, the Maumee and Sandusky Rivers, declined by 75 % and loads of dissolved P declined by 50 % (Sharpley et al. 2012). Agricultural conservation efforts targeting highly erodible lands contributed to substantial reductions in sediment and particulate P in runoff and stream flow. Nutrient management planning, particularly the prescription of fertilizer P rates based upon soil test P levels and crop needs, resulted in overall reductions of P applied as fertilizer of more than 30 % and as manure of 25 % (Baker and Richards 2002; Richards et al. 2002).

On the heels of these tremendous improvements in Lake Erie water quality, an uptick in dissolved P loads occurred in the mid-1990s, despite the persistence of historically low total P loads. Since 1995, dissolved P loads from Western Lake Erie Watersheds have increased (Fig. 2; Baker et al. 2014), triggering harmful algal blooms (Stumpf et al. 2012; Michalak et al. 2013). Identifying and controlling agricultural sources of dissolved P to Western Lake Erie has been difficult. Even climate change is a factor. Recent shifts in annual rainfall distribution have resulted in more intense rains in spring months during the 5 years, compared with the previous 10 years (Joosse and Baker, 2011; Smith et al. 2014), increasing the potential for P runoff at a vulnerable time for agriculture and a sensitive time for lake response (Chaffin et al. 2011). Furthermore, loads to Western Lake Erie from agricultural fields are low, on the order of 1–2 kg P ha⁻¹ year⁻¹ (Smith et al. 2014), complicating the identification of culprits and making room for competing narratives.

While many of the P management concerns for agriculture in the Western Lake Erie region are directed toward crop production, across the USA, intensive livestock production has received the majority of the attention related to diffuse P losses from agriculture (U.S. Environmental Protection Agency 2013). The Illinois River and Eucha-Spavinaw watersheds span the states of Arkansas and Oklahoma and have been the focus of litigation aimed at a burgeoning poultry industry and expanding urbanization (Fig. 1). A rapid, five-fold increase in the human population in Northwest Arkansas over the last 20 years has coincided with the expansion of confined poultry broiler operations, which now produce over 2000 million birds annually, nearly 25 % of the total broiler production in the USA. Under the U.S. Clean Water Act of 1972, the U.S. Environmental Protection Agency has ruled that upstream watershed users are responsible for downstream water quality. As a result, in 2001, the City of Tulsa, Oklahoma and in 2004 the Attorney General of Oklahoma filed law suits to mitigate the accelerated eutrophication of municipal water supplies, Eucha-Spavinaw reservoirs, and Lake Tahlequah.

As expected, the litigation had an initial effect of polarizing communities—rural and urban, Arkansas and Oklahoma—with competing narratives reflecting an array of long-standing biases and perspectives that had little to do with the sources and causes of watershed P loadings (Sharpley et al. 2012). As with Lake Erie, dissolved P was highlighted as a primary concern, making traditional soil conservation practices inadequate for mitigation, particularly as most productive agricultural lands were pastures (Sharpley et al. 2007). To move litigation toward settlement, the presiding Judge required that site assessment must be carried out to prevent application of poultry litter to fields in the watershed that were at high risk of P loss in runoff. The Judge mandated the use of a P Index that was developed specifically for the unique pastures, topography, and climate of the area (DeLaune et al. 2006).

A competing set of site assessment tools was proposed by experts from Arkansas, which accounts for the majority of the land area of the watersheds, and Oklahoma. Ultimately, Arkansas’ site assessment tool was adopted by the court. During this process, the Judge imposed various requirements on site assessment that, while ostensibly representing compromise, actually complicating standard procedures. Most notably, the Judge imposed a soil P ceiling for litter application, preventing application to soils with Mehlich-3 P > 300 mg kg⁻¹ (DeLaune et al. 2006). In 2013, as part of a court settlement agreement, this threshold was made much more restrictive: now P cannot be applied to soils with Mehlich-3 P > 150 mg kg⁻¹.

The litigation-derived P management standards have led to a marked decrease in the rates of litter applied in the Eucha-Spavinaw and Illinois River watersheds (from ~100 to 40 kg P ha⁻¹ year⁻¹) (Sharpley et al. 2012). Additionally, the Judge required at least 33 % of all litter produced in the Eucha-Spavinaw Watershed be exported, leading to the development of one of the only viable manure export programs in the USA. The success of the manure export program stems from the nature of the manure (dry poultry litter, which lends itself to transport due to low moisture content, high nutrient content, and a positive image as an organic fertilizer amendment), and from repeated adaptation of export tactics based upon the court-mandated requirement for balancing P application on agricultural soils (Fig. 3). The export of poultry litter out of northwest Arkansas and northeast Oklahoma to non-litigated watershed amounts to over 85 % of the litter produced in the Eucha-Spavinaw Watershed (Fig. 3; Herron et al. 2012; Sharpley et al. 2012). In the Illinois River Watershed, where the lawsuit has not reached a settlement phase, it is more difficult to determine amounts exported, but they are estimated at 100 000 tons, or roughly 37 % of litter produced in the watershed (approximately 275 000 tons in 2010; Herron et al. 2012). Nitrogen exported in litter from the Eucha-Spavinaw Watershed increased from a value of $1 094 000 USD in 2003 to $2 153 000 USD in 2010 (Fig. 3). Assuming that no P would be needed on the pastures when litter is exported, this equates to a significant cost to local beef farmers to buy commercial fertilizer N to maintain pasture productivity.

Key lessons learned include the imperative that export programs are tied to the logistics of importation so that litter can substitute with commercial fertilizer. In the Arkansas watersheds, key ties have been made between the fertilizer industry, particularly distributors, and the livestock industry so that mutual goals are achieved. Increasingly, farmers in the USA derive their nutrient recommendations from private sources, making them a key partner. These ties need to extend to areas where manure is exported, a process that is recognized in Arkansas but ongoing. In Lake Erie, the tie to the fertilizer distributors has been made through voluntary certification to promote environmental stewardship within the industry and, potentially, to provide a marketing edge for certified nutrient management planners.

Despite initial concerns that the restrictions placed by the court case would force poultry growers out of the litigated watersheds, poultry farmers have adapted to the P-based regulations, in part through subsidies supporting manure export. As a result, this case study represents an important example of the potential for farmers to overcome the impacts of mandated manure application restrictions. However, beef farmers (primarily cow-calf) in the area have suffered from the loss of a cheap and plentiful source of N in poultry litter that has enabled profitable cattle production on local pastures. The export of poultry litter under the litigated settlement has produced a slow decline in beef herd size and pasture productivity, coupled with an increased potential for erosion due to worsening pasture conditions with declining fertility. In order to maintain the economic viability of all farming enterprises, not just the poultry farms, it has become clear that the nutrient management planning process must go beyond addressing poultry litter application rates and environmental risk and include educational efforts to help farmers develop sustainable whole-farm operations.

In the time following implementation of court-mandated nutrient management changes, there has been a slow but constant decrease in the concentration (mg L⁻¹) of total P in baseflow of the Illinois River as it flows from Arkansas into Oklahoma (Fig. 4). Since 2003, required P-risk nutrient management has decreased litter applications to area pastures (Fig. 3) and water treatment plant upgrades have reduced point source inputs of P, making it impossible to isolate the impact of litter export on P loadings to the Illinois River (Haggard 2010). Annual variations in flow have served to hide the effect of lower concentrations (mg L⁻¹) of P in baseflow on watershed P losses (kg ha⁻¹). However, concentrations have decreased a third compared with pre-2003 (from 0.29 mg L⁻¹ in 2002 to 0.07 mg L⁻¹ in 2010; Fig. 4), leading to directional trends that provide hope to agricultural and conservation communities alike.

Phosphorus loadings to freshwaters of the United Kingdom (UK, comprising England, Wales, Scotland, and Northern Ireland) are a major water quality concern; for example in England and Wales, P is the largest single contributor to poor ecological status in rivers and lakes (EA 2013, 2014). Across the UK, under the European Union (EU) Water Framework Directive, targets of annual average concentrations of reactive P in rivers and total P in lakes and reservoirs have been set to help reduce eutrophication and achieve good ecological status (e.g., Ryder and Bennett 2010). Agriculture is a major contributor and is currently targeted for P mitigation (McGonigle et al. 2012; EA 2014; Zhang et al. 2014). However, very different approaches to achieving agricultural P loss controls are used in Great Britain (England, Wales and Scotland), where implementation programs are largely voluntary, versus Northern Ireland, where national regulations are applied across all of agriculture, but have a specific focus on P with regard to farm P budgets and fertilizer use.

Of the case studies reviewed here, Sweden represents the most tightly regulated setting for agricultural P-based management, with a great portion of the costs related to P mitigation measures covered by subsidies. Agricultural P management in Sweden coupled to eutrophication of the Baltic Sea is today, to a large extent, driven by the Baltic Sea Action Plan (BSAP), an international accord that was devised in 2007 after P was implicated as the main cause of cyanobacterial blooms in the Baltic Sea (Boesch et al. 2006). Despite generally low intensity of land use, Swedish agriculture is estimated to account for 40–50 % of the total anthropogenic P loads from the nation’s Baltic watersheds (Brandt et al. 2009). To meet the targets prescribed in BSAP and achieve the P load reductions required, national regulations related to e.g., spreading of animal manure (limited to 110 kg P ha⁻¹ over a 5-year period) and the EU Water Framework directive need to be followed. In addition, a voluntary advisory program (‘Greppa näringen’), which was introduced in 2000 to give farmers in sensitive areas free individual support, has helped to reduce agricultural P losses at a cost of about $4 million USD year⁻¹.

A central theme of Swedish P mitigation programs is to pay farmers for conservation and nutrient management measures they adopt. Since 2000, subsidies have been available through the Swedish Rural Development Program, which is partly funded by the EU, to compensate farmers for carrying out certain practices to reduce both N and P losses. Practices qualifying for such subsidies especially related to P include conservation buffer zones for highly erodible soils, constructed wetlands, and, perhaps most notably, organic crop production, which is a subsidy to help reduce environmental disturbances in general. In recent years additional mitigation strategies have been included, such as the installation of drainage management practices on tile drains and ditches, i.e., ‘controlled drainage’. Subsidy of water quality mitigation practices has helped to spur adoption of practices aimed at preventing P losses from agriculture, but has not removed some of the profound obstacles encountered under less-regulated, less-subsidized settings.

Although many practices promoted for water quality protection have widespread support in the agricultural community, artificial drainage represents as much of a sacred cow in Sweden as it does in Lake Erie. About 50 % of arable land in Sweden is tile drained, especially those soils with high clay content. As with Lake Erie, artificial drainage is imperative to allow field management operations to be performed as early as possible in spring and to protect crops from flooding. However, because artificial drainage is seen as such an essential part of agricultural infrastructure, few in the agricultural community have to date been willing to open discussion on options for limiting new drainage or removing drainage to control non-point source pollution. Instead, other measures to improve soil structure such as liming of clay soils and tile-drainage backfills to increase P adsorption to the soil matrix have been tested with good results (Ulén and Etana 2014). Also, grass buffers along rivers and open ditches have been emphasized and perhaps even over-implemented in Sweden, especially when viewed through the lens of P mitigation. Grass buffers are used as a multifunctional tool in agricultural landscapes around the world, providing many ecosystem services other than the regulation of nutrients and sediments (Stutter et al. 2012). For P, buffer zones are thought to be effective in promoting sedimentation and retention of sediment-bound P, but their efficacy in preventing dissolved P loss has been widely questioned (Dorioz et al. 2006). Indeed, when buffer zones are bypassed with concentrated flow pathways or when the P-binding capacity of the soils is largely saturated, they can range from ineffective to a source of P (Uusi-Kämppä and Jauhiainen 2010). Even so, from 2000 to 2006, Swedish farmers received about $3 million USD year⁻¹ in subsidies to install and maintain riparian buffer zones, with an estimated reduction in watershed P loads of 6 tons year⁻¹, equivalent to a total P removal efficiency of $500 kg⁻¹ year⁻¹.

National debate over the cost effectiveness of buffer zones as P mitigation tool is largely stifled by the amount spent on agricultural subsidies in Sweden for organic agriculture, effectively turning buffer zones (or any other P mitigation efforts) into a sacred cow for Swedish taxpayers. Recently, an evaluation of the Swedish Rural Development Program concluded that programs to reduce agricultural non-point source pollution with specific practices such as buffer zones were much more cost-effective than national subsidies for organic crop production, for which Swedish farmers received $75 million USD year⁻¹ from 2000 to 2006. In fact, it can be argued that subsidies for P mitigation practices, regardless of their cost effectiveness, help to offset the P losses from organic farms, which are sometimes greater than from conventional systems (Aronsson et al. 2007).

When Sweden’s P mitigation programs are evaluated on the basis of water quality alone, without considering cost, there is cause for cautious optimism. Downward trends in P concentrations in large rivers in agricultural areas of southwest Sweden have been noticed, although, the picture is somewhat diverse with increasing trends in some rivers (Fölster et al. 2012). Thus, implementation of mitigation strategies in Sweden has had a slight beneficial effect on reducing agricultural P losses, but it is too early to draw any general conclusions.

Recent experience with the implementation of P mitigation strategies under regulatory and volunteer settings points to a core set of issues confronting success. Dissolved P losses from agriculture appear to be particularly difficult to control. Uncertainty over the outcome of implementation strategies, an inherent aspect of diffuse P pollution creates room for competing perspectives on the best practices to mitigate P loss and the most appropriate means for implementing these practices. Unintended consequences of past management efforts or current initiatives appear to be the rule, rather than the exception, from conflicts between soil conservation and dissolved P management to adverse impacts on producers who have benefited from cheap, local sources of livestock manure. Since P-based management often requires activities that are separate from standard practice and incur additional costs, subsidies are a natural outcome of government-led mitigation efforts. These subsidies can be important to the success of programs, but they are by no means a universal requirement for success. In all cases, local empirical data, particularly in the form of water quality monitoring, are the best means of convincing skeptics and compelling local responsibility to act. Indeed, without adequate local information on practice adoption and effectiveness, whether imposed by regulation or adopted on a voluntary basis, it is impossible to judge the success and failure of mitigation programs.

The commonalities observed between case studies should not obfuscate the imperative for local, site-specific consideration of practice potential, compatibility and effectiveness. In the Western Lake Erie, a relatively recent reversal of historical water quality improvement recently culminated in the temporary closure of public drinking water supplies for the City of Toledo in 2014. This event has elevated concerns even higher and has created a politically charged environment in which ready accusations from certain sectors have placed a heavy onus on waiting for locally derived empirical findings. In this environment, uncertainty and competing interests within the farm and conservation community have introduced certain practices (especially those involving tillage), and, more importantly, unnecessary conflict.

In the Illinois River and Eucha–Spavinaw watersheds, USA, litigation amplified differences between stakeholders in the states of Arkansas and Oklahoma. Most notably, regulations derived from the lawsuit were initially feared as sufficiently draconian to drive away many poultry farmers. However, a successful litter export program and subsidies were key to success and to minimizing adverse impacts to poultry farmers. An unanticipated casualty of the P-based regulations, however, has been the beef farmers, who have been dependent upon local sources of poultry litter to improve pasture conditions and sustain beef herds.

In the UK and Sweden, multinational EC and Baltic State agreements set a framework in which P-based management strategies have been promulgated. These strategies differ widely across partner states, even within the UK. In some cases, the implementation of management programs has coincided with separate, industry-led initiatives, resulting in positive outcomes that are not necessarily the product of the government program. For most programs, the key metric of success for most management programs is the extent to which a practice is implemented, rather than the effectiveness of practice adoption in mitigating water quality degradation. Better ties between practice implementation and water quality benefit are required to ensure cost-effective implementation programs.