Review for DEFRA: Milestone 1a of Project PE0203, 18th March 2002
Theoretical and Practical Effectiveness of Phosphorus and Associated Nutrient/Sediment Mitigation Measures in England and Wales
P. M. Haygarth
Institute of Grassland and Environmental Research, North Wyke Research Station, Okehampton EX20 2SB
P. J. A. Withers
ADAS Rosemaund, Preston Wynne, Hereford HR1 3PG
M. Hutchins
ADAS Woodthorne, Wergs Road, Wolverhampton, WV6 8TQ
1. Introduction and Rationale
This review is approached in four sections. Firstly, it undertakes a ‘bottom up’ theoretical overview of the key processes that pertain to the three types of phosphorus (P) mobilisation, and thus its mitigation. This is followed by a theoretical discussion of management ‘timetables’, because we believe that mitigation options must be couched within a short and long-term time frame. Subsequently, a ‘top down’ review is presented that considers practical management measures for the integrated control of P in conjunction with nitrogen (N) and sediment/silt. Finally, some simple scenario modelling examples are undertaken to assess the actual potential impact of various mitigation strategies in real catchments.
2. Theoretical Assessment of Phosphorus Mobilization from Agricultural Land
Conceptually, there are three separate ways by which P may be mobilized, called solubilisation, physical detachment and incidental transfers, each of which represents completely separate processes (Haygarth and Jarvis, 1999). Thus the potential options for mitigation are also separate. The timetables of their effectiveness will also vary sonsiderably.
Solubilisation
The process
Solubilisation is traditionally akin to leaching, and we use the operational definition in terms of <0.45 mm. For P, the process of solubilisation may be either chemical, resulting from an excess of P in relation to the soils buffer capacity (Breeuwsma et al., 1995; Heckrath et al., 1995) or biological, resulting in rapid release P from organic matter and soil biomass following perturbation (Turner and Haygarth, 2001). The former, chemical solubilisation is well understood, whereas the latter is a recent ly identified phenomenon and little is known about it’s role.
Mitigation potential
Solubilisation is only effectively controllable in the long term (see section 3) and it is well established that soil P levels (and thus management history and inputs) affect chemical solubilisation and thus concentrations of P in drainage (Heckrath et al., 1995). Thus, assuming 300 mm of drainage per year and a soil solution total P concentration of 100 μg L-1 (reflecting the soil P level and the P fertilizer/input history), this will yield an export of 0.3 kg total P per hectare per year. This is a low export but is sufficient to contribute to eutrophication. Conversely, any increase or decrease in soil P level will, broadly, be reflected in the export coefficient, since there is a generally logarithmic relationship between soil P status and solubilisation. Hence reducing inputs and/or running down of soil P reserves are mitigation options are slow (tens of years) yet in the long term they are unarguably ways where we must focus for long term soils and catchment sustainability. These measures are long term, but, relative to detachment, has a higher (albeit slower) likelihood of achieving a reduction. One option for reducing soil P levels may be to ‘plough in’ surface soils that are high in P, but this is an approach is speculative and has no experimental support.
Physical detachment
The process
Detachment is the physical release of soil particles and colloids >0.45 mm, often with P attached. Often, but not exclusively, this refers to soil erosion per se (Burnham and Pitman, 1986; Elliot et al., 1991; Evans, 1990; Heathwaite and Burt, 1992; MAFF, 1997; Morgan, 1980; Quinton, 1997) and has attracted considerable interest, as has the role of particle transfer in P loss (Catt et al., 1994; Kronvang, 1990; Sharpley and Smith, 1990; Zobisch et al., 1994). More recently, the role of small colloids in detachment and transport of P have also been described (Haygarth et al., 1997).
Mitigation potential
Theoretically, mitigation of detachment can be focussed on two tiers. The first is often overlooked, but involves discreet and ongoing sheet washing of silts and colloids throughout ‘normal’ winter rain. This occurs when rivers become turbid, yet land remains, at least, visibly, uneroded. It is speculated, albeit unproven, that ‘discreet’ solid losses can be largely controlled by similar mechanisms, as solubilisation i.e. that soils with high P status will yield ‘solids’ with high P status. Further, to a certain extent, poaching and presence or absence of livestock and bare soils are all speculated to contribute to these losses. However, because of their nature they remain little studied. Conversely, the second tier involves catastrophic (‘erosion’) losses and these are classically more easy to conceptualize, study and thus prevent. Classical low tillage and contour cultivation options can be used and, in one recent example (MAFF NT1032) we observed 3.75 kg transfer of total P from a Rowden plot in only 16 days, following bare soils during a spring reseed. This is similar annual exports during ‘normal’ circumstances and thus clearly there is potential to avoid such events if tillage can be better timed to avoid runoff. The mitigation potential is high and effective in the short term, but still involves a high degree of uncertainty in effectiveness.
Incidental
The process
The work of Haygarth and Jarvis (1999) also argued the existence of a ‘incidental’ mobilization, where anthropogenic sources (manure, fertiliser) coincided with high water flows, but this should be viewed as a secondary process and can be subsumed into the two primary processes.
Mitigation potential
Incidental transfers are undesirable to both the farming community and to catchments and are, theoretically, easily prevented if manure or fertilizer is not added at times or in places where runoff is likely. Incidental transfers have been shown by Preedy et al. (2001) to generate exports in the order of 3 kg TP per week and if such events can be avoided there is the capability to reduce losses by ca. 50%. In practice however the farming community faces many problems with manure storage in the ‘wet and mild’ English and Welsh climates and runoff predication and thus P loss avoidance may be, in practice, difficult to achieve. Incidental transfers are revisited in Section 3 under accelerated circumstances. Like soil erosion associated losses, the mitigation potential is high and effective in the short term, but still involves a high degree of uncertainty in effectiveness.
3. A Theoretical Consideration of approaching Short and Long Term Management Priorities
Our conceptual rationale is that managing mitigation relies on a management of risk that should be approached in both the long and the short term. ‘Short term’ time resolution is sensitive to identification of events or storms (identified from time series data such as a P concentration (chemograph) or discharge (hydrograph)), but cannot consider trends beyond this time frame. In contrast, ‘long term’ time resolution does not identify individual events, but it integrates storms to extend ad infinitum, theoretically over tens of years, if necessary.
Within this framework, we have identified two rates of transfer, ‘background’ and ‘elevated’. In the short term, two additional types of transfer can be identified as ‘systematic’ or ‘accelerated’. Here, the transfer rate, which is the mass of export per unit time (i.e. load), is the basis upon which the objective classification is made. As load is the product of P supply (as determined in the short term in a chemograph by P concentration (Pc)) and water discharge (Q), the varying contribution of these two factors will determine the extent to which transfer is limited by supply of P (i.e. Pc) or the hydrological transport (i.e. Q). This concept is critical for characterising the circumstances of P transfer.
Long term
Background transfers
Background transfers are based on the assumption that some transfer of P from soil to water will occur as a natural geological process. The quantitative basis and extent of this hypothesis remains largely untested, although some studies are available where information can be inferred and extrapolated (Letkeman et al., 1996; Walker and Syers, 1976). It is essential that we devise a means for determining, at least theoretically, the background transfer rate from any given soil, so this can be established as a theoretical baseline with which to compare the elevated rates of transfer. There are many avenues to explore here and there are considerable opportunities for studies of (i) contemporary pristine environments, for example those conducted at Glacier Bay, Alaska (Chapin et al., 1994) and (ii) historically (long term) pristine environments that have since been perturbed, using retrospective palaeo-ecological techniques (Foster and Lees, 1999; Foster and Walling, 1994). Theoretically, background transfers will depend on a combination of factors including the nature of the (i) parent material and (ii) the nature of climax vegetation. We postulate that, in the long term, background transfers will attain quasi-equilibrium, with changes only occurring as a result of the progression of ecological succession.
Elevated transfers
Elevated transfers will exceed rates of background transfers as imported P is added to the system and thus the ability of soils to buffer and retain P is exceeded. The term ‘elevated’ is partially derived out of underlying agricultural transfers, described previously by Haygarth et al. (2000). The most obvious examples include anthropogenic additions of P via fertiliser or feed concentrate–animal–excreta/manure to agricultural soils (Frossard et al., 2000; Haygarth et al., 1998b). These could be called agriculturally elevated transfers, but in other circumstances may be of forestry, urban or industrial in nature. Most contemporary studies of P transfer have been carried out on elevated rates of transfer and classic studies for agricultural soils have emerged. Particularly noteworthy are the findings of studies on chemical solubilisation, such as those reported by Heckrath et al. (1995) and Pote et al. (1996), which demonstrated a correlation between the soil P test values and the concentrations (and hence inferred loads) of P in drainage water. In Figure 1 we illustrate the hypothesis for background and elevated rates transfers, showing them in context with one another over the long-term time-scale.
Short term
Systematic transfers
Whilst background and elevated rates of transfers integrate and dominate the long term trends, high resolution analysis of hydro/chemographs reveals two additional modes of response in the short term. These can be separated by analysis of the time series of the P breakthrough chemograph in relation to the water flow hydrograph. We call the first of these ‘systematic’ transfers (the term was first used by Nash et al. (2001) and is partially derived from their studies and partially relates to ‘underlying agricultural transfers’ (Haygarth et al., 2000), approximating ‘base flow’ as described in standard hydrological texts). Systematic transfers exemplify an approximately ‘flat’ or ‘smooth’ Pc breakthrough, in relation to the relatively varied response of Q (Figure 2). Therefore, for systematic transfers over the short term, P transfer from soil is well buffered: This is logical as it is in keeping with traditional view about the ability of soils to hold and buffer P against leaching (Russell, 1957). In these circumstances, Q limits systematic P transfers.
Accelerated transfers
Accelerated transfers may occur where the P concentration responds positively to changes in discharge (Figure 3), with Pc responding to a rise in Q. We can subdivide this and hypothesise three circumstances by which this may occur:
I. Scenario 1. Where there is a high source of easily mobile P that is only available in the short term. For example, following soil rewetting and rapid release of organic P to soil solution (Turner and Haygarth, 2001), or in agricultural systems the recent addition of fertiliser manure or excreta.
II. Scenario 2. Where fast water flows are so energetic, perhaps due to intense rainfall or preferential/overland flow, that the physical energy detaches or solubilises P pools that would be otherwise immobile in systematic circumstances.
III. Scenario 3. Occurring as a combination of 1 and 2, where a pool of easily mobile P is combined with fast flows. This exemplifies circumstances of ‘incidental transfers’, where manure/ fertiliser amendment coincides with rapid hydrological transport from the soil (Haygarth and Jarvis, 1999).
An ad hoc list of circumstances which may fit to these scenarios could be geological perturbations such as landslides, the detachment of soil particles and colloids as a consequence of heavy trafficking or poaching by livestock and the cracking and rewetting of soils at the end of summer drought (Heathwaite and Dils, 2000; Hooda et al., 2000; Nash et al., 2000; Turner and Haygarth, 2000). In contrast to systematic transfers, Pc limits accelerated P transfers, not Q.
Synthesis
This work has provided a new conceptual framework for approaching P transfer and a summary of the concepts presented is given in Table 1. Most critical is the importance of the time-scale, knowledge of which may highlight priorities for risk and mitigation.
We have to accept that some P transfers will be inevitable, thus embracing the concept of background transfers. There is definitely the need for more fundamental studies here, as these transfers seem to have been overlooked amongst the largely agronomic orientated research conducted to date. Issues such as the natural buffering capacities of different parent materials need to be examined, and data from existing literature re-examined. Theoretically, background P transfer rates will be primarily influenced by a combination of (i) parent material, (ii) effective rainfall, and (iii) drainage. Soil age and disturbance may also be issues that require focus. Until we have established a model framework for background transfers, we have no effective reference point against which to calibrate the extent of systematic or accelerated P transfers from agricultural soils.
It is fortunate that the greatest elevated accelerated P transfers also occur during the shortest time scale and therefore present the best opportunities for reducing loads. Recent work by Preedy et al. (2001) identified ‘incidental’ (N.B. this term can now be replaced by ‘accelerated’) transfers of manure and fertiliser derived P that can result in rates of transfer where > equivalent of 2 kg P ha-1 have been exported in only a few days. This contrasts with systematic circumstances or P transfer that may be of this order, but occurring over a year (Haygarth et al., 1998a; Haygarth and Jarvis, 1997).