Science of the Total Environment 369 (2006) 280 – 294 www.elsevier.com/locate/scitotenv
Impacts of agricultural phosphorus use in catchments on shallow lake water quality: About buffers, time delays and equilibria Peter Schippers a,b,⁎, Hendrika van de Weerd a , Jeroen de Klein a , Barend de Jong a , Marten Scheffer a a
b
Aquatic Ecology and Water Quality Management Group, Wageningen University and Research Centre, P.O. Box 8080, 6700 DD Wageningen, The Netherlands Wageningen IMARES, Wageningen University and Research Centre, P.O. Box 68, 1979 AB IJmuiden, The Netherlands Received 9 January 2006; received in revised form 19 April 2006; accepted 19 April 2006 Available online 15 June 2006
Abstract Phosphorus (P) losses caused by intensive agriculture are known to have potentially large negative effects on the water quality of lakes. However, due to the buffering capacity of soils and lake ecosystems, such effects may appear long after intensive agriculture started. Here we present the study of a coupled shallow lake catchment model, which allows a glimpse of the magnitude of these buffer-related time delays. Results show that the buffering capacity of the lake water was negligible whereas buffering in the lake sediment postponed the final lake equilibrium for several decades. The surface soil layer in contact with runoff water was accountable for a delay of 5– 50 years. The most important buffer, however, was the percolation soil layer that may cause a delay of 150–1700 years depending on agricultural P surplus levels. Although the buffers could postpone final lake equilibria for a considerable time, current and target agricultural surplus levels eventually led to very turbid conditions with total P concentrations of 2.0 and 0.6 mg L− 1 respectively. To secure permanent clear water states the current agricultural P surplus of 15 kg P ha− 1 yr− 1 should drop to 0.7 kg P ha− 1 yr− 1. We present several simple equations that can be used to estimate the sustainable P surplus levels, buffer related time delays and equilibrium P concentrations in other catchment–lake systems. © 2006 Elsevier B.V. All rights reserved. Keywords: Macrophytes; Response time; Runoff; Soil; Watershed; Water Framework Directive
1. Introduction It is generally acknowledged that phosphorus (P) is often the limiting nutrient determining algal growth in shallow lakes (Schindler, 1977; Sondergaard et al., 2001; Moss et al., 2003). In many catchments P loads that ⁎ Corresponding author. Wageningen IMARES, Wageningen University and Research Centre, P.O. Box 68, 1979 AB IJmuiden, The Netherlands. Tel.: +31 255564691; fax +31 255564644. E-mail address:
[email protected] (P. Schippers). 0048-9697/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2006.04.028
originate from human activities exceed natural loading by several orders of magnitude (Falkowski et al., 2000). Two major artificial sources of P can be distinguished: (1) Point sources, especially effluents from waste water treatment plants and (2) diffuse losses by agricultural activity (Chapman et al., 2003; Merseburger et al., 2005; Wood et al., 2005). The latter has become increasingly important because more and more wastewater treatment plants have become equipped with phosphate removal devices, whereas agricultural losses have risen strongly with the introduction of artificial fertilizer and the
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
dumping of manure on agricultural soils (Oenema and Roest, 1998; Gillingham and Thorrold, 2000). This enrichment has large ecological consequences for terrestrial (Schippers and Joenje, 2002; Wassen et al., 2005) and aquatic ecosystems (Scheffer, 1998; Moss et al., 2003; Schippers et al., 2004). Reducing diffuse P losses has become a major issue, since in 2015 the EC Water Framework Directive demands good ecological and chemical states of lake and river systems. To ensure optimal crop growth or grassland production it is common practice in agriculture to provide plants with amounts of P fertiliser that exceed the removal of P by crop and grass harvesting (Withers et al., 2001; Oborn et al., 2003; Djodjic et al., 2005). The resulting P surplus ends up either attached to soil particles or flushed away by water. With respect to the flushing two pathways can be distinguished: (1) P is flushed from the field with the surface run-off water or (2) P percolates through the soil with the percolation water (Oenema and Roest, 1998). These pathways differ strongly with respect to the residence time of P in the field. The surface runoff pathway is relatively fast whereas the percolation pathway can be very slow (Arheimer and Liden, 2000). The residence times of the latter pathway might range from a single decade up to centuries (Behrendt et al., 1996; Oenema and Roest, 1998). These long residence times exist in soils because P strongly adheres to Fe and Al minerals in the soil (Vanderzee and Vanriemsdijk, 1988; Vanderzee et al., 1989). The soil permeability of water, the steepness of the slope and rainfall intensity determine which pathway dominates (Kleinman et al., 2004; Syversen, 2005; Ulen and Jakobsson, 2005). Another major factor that determines P export from the soil is the precipitation surplus that is defined as rainfall minus evapo-transpiration (McIntosh and Thom, 1981). A high precipitation surplus often results in a high amount of runoff, which consequently causes high P loads (Cooper et al., 2002). On the other hand the water surplus also dilutes P that in turn may lead to lower P concentrations in the water that runs into the lake. The P concentration in a lake is determined by the P concentration of the inflowing water, the residence time and the internal P dynamics of the lake. Generally the P concentration in the lake (Pl) is lower than the P concentration of the inflowing water (Pi) because there is a net loss of P to the sediment (Vollenweider, 1976). If a lake has received P-rich water over a longer time-span, the release from the sediment P buffer might prevent lake restoration for long time (Sondergaard et al., 1999; Sondergaard et al., 2001). Macrophytes play an important role in shallow lake dynamics (Scheffer et al., 1993; Van Donk et al., 1993; Van den Berg et al., 1998). They enhance sedimentation, reduce the resuspension, compete with phytoplankton for
281
nutrients, (Scheffer, 1998; Van Nes et al., 2002) and therefore generally induce lower Pl levels. In addition to this, macrophytes are hiding places for zooplankton, which control phytoplankton populations. Furthermore, vegetation facilitates ambush predators like pike, which control species (e.g. Bream) that strongly enhance the resuspension of the sediment (Scheffer, 1998). Until now, most studies estimating catchment P losses are based on land use data of specific catchments (Johnes, 1996; Johnes and Hodgkinson, 1998; Jordan et al., 2000; Cooper et al., 2002). Here we present and analyze a simple model in which we explore the mechanisms and future dynamics of a catchment–lake system. We specifically ask: (1) what are the maximum agricultural P surpluses allowed to preserve a clear lake dominated by macrophytes? (2) what are the time delays of the various P buffers in the catchment–lake system?, and (3) can we derive simple equations to predict response times, equilibrium concentrations and critical agricultural loads? 2. Model We developed a generic model that comprises two sub models: a catchment model and a lake model. The catchment model is a soil column model that accounts for the heterogeneity of soils in the catchment and the surface runoff process. The lake model describes the dynamics of total P in the lake water and sediment as a result of in and outflow processes, water sediment interactions and the presence of macrophytes. 2.1. Catchment model The modelled watershed consists of an agricultural area and natural area. We assume that soils in the natural area of the catchment are in equilibrium with the natural P and water surplus. Hence the P concentration of water (Pn in mg L− 1) that originates from the natural area is: Pn ¼
Snd 100 W
ð1Þ
where Sn is the P surplus of the natural area (kg P ha− 1 yr− 1) and the W is the yearly precipitation surplus (mm yr− 1). Analogously we can calculate the P concentration of the water at the soil surface of the agricultural area. This water, contaminated with the agricultural P surplus infiltrates into the soil or runs off along the soil surface. This is simply modelled by a factor R that determines the runoff fraction of the water surplus. The P concentration in the run-off water is assumed to be in equilibrium with the adsorbed P of the upper soil layer. The percolation
282
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
pathway of the agricultural part of the catchment is simulated as a soil column model (Fig. 1). The amount of layers and the average depth of the column can be chosen. The soil in the column has a certain maximum P storage capacity (Qm in g m− 3) and an affinity for P (K in m3 g− 1). When water with a certain P concentration infiltrates into the soil, P is partly adsorbed by soil particles and partly dissolved in the water phase. The amount of P attached to soil particles (Pp in g m− 3) can be calculated with the Langmuir equation (Vanderzee et al., 1989; Groenendijk and Kroes, 1999; Koopmans et al., 2002): Pp ¼ Qm
Kd Pw Kd Pw þ 1
ð2Þ
where Pw is the concentration in the water phase of a certain layer (mg L− 1). To account for the soil and catchment variability we introduce a standard deviation over the column length (Fig. 1). This standard deviation accounts for the variability in residence times of P in the soil caused by variations in depth, macro pore flows
(Kirkby et al., 1997) and diffusion (McGechan and Lewis, 2002). Therefore, three flows from the catchments can be distinguished: (1) the flow from the natural area with a constant (low) concentration, (2) the percolation flow from the agricultural area and (3) the surface runoff from the agricultural area. These flows merge into the river system that subsequently runs into the lake. To keep the model simple we assume no absorption and time delays during horizontal transport through the soil and river system. 2.2. Lake model We started from a simple mass balance approach in which the in lake P concentration (Pl mg L− 1) is the result of in- and outflow, losses to the sediment and internal loading: dPl Pi −Pl iPs −sPl ¼ þ dt s D
where Pi is the P concentration of the water that comes from the catchment and flows into the lake, τ is the residence time of the lake water (day), i is the internal loading coefficient (day− 1), s is the sinking rate of particle P in the water indicating the lumped loss rate to the sediment (m day− 1), Ps is the amount of P in the upper sediment layer (g P m− 2) and D is the mean depth of the lake (m). Only the bio-perturbed upper sediment layer is in close contact with the lake water and is generally believed to be the source of the internal loading (Lijklema and Hieltjes, 1982; Van der Molen et al., 1998; Sondergaard et al., 1999, 2001). This bio-perturbed layer is assumed to have a constant thickness of about 100 mm (Lijklema and Hieltjes, 1982). As a result of sedimentation, a flux of P from the lake water adds P to the sediment. P losses from this layer are: (1) the internal loading (i in day− 1) which is assumed to be proportional to the P concentration of the upper sediment layer and (2) the burial rate (b in day− 1) that is determined by the sedimentation rate (mm d− 1) and the thickness of the upper layer. Under these conditions the changes in the P concentration of the upper sediment layer (Ps g P m− 2) can be described as: dPs ¼ sPl −iPs −bPs dt
Fig. 1. Schematic diagram of the catchment model. Water and dissolved phosphorus (P) infiltrates in a soil column. While water percolates through, P is divided between the water and the solid phase according to the Langmuir isotherm. The standard deviation accounts for the catchment heterogeneity with respect to the residence time for P.
ð3Þ
ð4Þ
2.3. Macrophytes Macrophytes play an important role in the lake dynamics (Van den Berg et al., 1998; Horppila and
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
Nurminen, 2003) because they reduce sediment resuspension, enhance sedimentation and compete with phytoplankton for resources (Scheffer, 1998; Van Nes et al., 2002). This in turn affects the total water P concentration because the largest fraction of P in the water is usually associated with phytoplankton and suspended matter. Therefore, when macrophytes are present we expect lower P concentrations in the water phase. The effect of macrophytes is represented by reducing the internal loading i by 78% (Horppila and Nurminen, 2003), which results in a reduction of the equilibrium P concentration by 50% (Scheffer, 1998). On the other hand, macrophytes are very sensitive to turbidity caused by suspended particles and phytoplankton. High P concentrations generally induce turbid conditions because phytoplankton growth, in turn reduces the light availability for the macrophytes. In shallow lakes, a P concentration of 0.1 mg P L− 1 is often
283
considered to be the critical threshold for macrophyte survival (Van Nes et al., 2002). Therefore, in the model we assume macrophytes are always present when the lake P is lower than 0.1 mg P L− 1 and vice versa. 2.4. Parameterisation We use an average Dutch catchment shallow lake system as an example to demonstrate our approach (Table 1). Since the parameter i is rarely measured independently we estimated this parameter with Eq. (3) using median Dutch lake parameters (Portielje and Van der Molen, 1999): Pl = 0.16, Pi = 0.26, residence time of 146 days and a depth of 1.95 m, assuming a P density in the active sediment layer of 50 g m− 2 (Van der Molen et al., 1998). In turn b can be estimated from the Vollenweider (1976) equation to estimate Pl / Pi of a lake in equilibrium.
Table 1 Parameters and variables used in the paper Par. Description
Standard value (range) Unit
Lake parameters and variables s Sinking velocity of particle-P D Depth t Residence time b Burial rate i Internal loading coefficient Pc Critical Pl where macrophytes just can survive M P reduction factor by macrophytes Pl P concentration in lake water Pi P concentration of lake inflow Ps Active sediment P density Fp Concentration in lake relative to inflow (Pl / Pi) Lw Response time of the lake water Ls Response time of the lake sediment C Tolerated fraction to calculate Lw and Ls
0.1 1.95 146 2.8 * 10− 5 3 * 10− 4 0.1 0.23 – – – – – – 0.20
m day− 1 m day day− 1 day− 1 mg P L− 1 – mg P L− 1 mg P L− 1 g P m− 2 – day day –
Jorgensen et al., 1991; Van der Molen et al., 1994 Median value of 22 Dutch lakes Median value of 22 Dutch lakes Calculated see text Calculated see text Van Nes et al. (2002) Calibrated see text Variable Variable Variable Variable Variable, Eq. (10) Variable, Eq. (11) Parameter
Soil parameters and variables V Pore fraction of soil Qm Max. P storage of the soil K Absorption constant of the soil Pp P attached to soil particles
0.5 500 1 –
– g m− 3 m3 g− 1 g m− 3
Vanderzee et al. (1989) Behrendt et al., 1996; Koopmans et al., 2002 Vanderzee et al., 1989; Koopmans et al., 2002 Variable
226 (50–800) 0.125 (0.01–0.5) 0.0225 0.01 1 0.03 0.5 (0.05–1) 0.25 (0.02–0.4) –
mm yr− 1 – kg P ha− 1 yr− 1 mg L− 1 m m – m yr yr
Heijboer and Nellestijn, 2002 Meinardi et al. (1994) Oenema and Roest (1998) Oenema and Roest (1998) Behrendt et al., 1996 Sharpley et al., 1994, Sharpley, 1995 Catchment dependent Catchment dependent Variable, Eq. (7) Variable, Eq. (7) modified (see text)
15 (1–25)
kg P ha− 1 yr− 1 Oenema and Roest, 1998; Smith et al., 1995
Catchment parameters and variables W Water surplus Dutch conditions R Surface runoff fraction of water agriculture Sn P surplus nature area Pn [P] of water released from the nature area Average thickness of soil until ground water Tt Tr Surface soil layer in contact with runoff water A Fraction of catchment area used for agriculture σ Standard deviation of T Lp Response time of the percolation pathway Lr Response time of the surface runoff pathway Agricultural characteristics Sa P surplus agricultural area ‘normal’
Source/comment
284
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
3. Simulations and results
3.3. The lake model
3.1. Initial and standard conditions
To study the performance of the lake model we conducted a simulation considering the lake model only. Here we increased the P concentration of the inflowing water (Pi) after 50 years, from pristine (Pi = 0.01 mg L− 1) to polluted conditions (Pi = 0.5 mg P L− 1) conditions. Additionally we calculated the long term equilibrium values of this simulation of both lake water and sediment and finally, we explored the effect of different Pi concentrations on the long term equilibria of lake and sediment.
Initially the whole catchment and lake system was assumed to be in equilibrium with the natural P background surplus of 0.023 kg ha− 1 yr− 1 and a water surplus of 226 mm per year, which gave an equilibrium Pi concentration of 0.01 mg L− 1, a value that is based on measurements in deeper percolation water under Dutch agri- cultural sandy soils (Oenema and Roest, 1998, Table 1). This resulted in a macrophyte rich lake with an initial lake water concentration of 0.0032 mg P L− 1. Our standard catchment and lake were parameterized according to Table 1, which represents Dutch catchment and shallow lake conditions. If not mentioned otherwise, agriculture was started at the beginning of the simulation with a P surplus level of 15 kg P ha− 1 yr− 1 and covered 50% of the catchment area. 3.2. The catchment model To explore the properties of the catchment model we performed two simulations with two contrasting catchment types that differ with respect to the catchment variability σ and the surface runoff fraction R. In these catchments we explored the response of normal agriculture practise with a P surplus of 15 kg P ha− 1 yr− 1 (Edwards and Withers, 1998; Oenema and Roest, 1998; Smith et al., 1995) which lasted 700 years. After this period the natural background P surplus of 0.0225 kg ha− 1 yr− 1 was restored in the whole catchment. The P concentration rose rapidly in the first 10 years due to fast saturation of the surface runoff layer (Fig. 2ab, A), followed by a second increase due to the breakthrough of the percolation pathway (Fig. 2ab, B). At point C (Fig. 2ab) all soil was saturated. After 700 years, when the natural background surplus was restored, a fast drop in the P concentration occurred, due to the cleaning of the surface runoff soil layer (Fig. 2ab, D). After that, P levels in the percolation pathway slowly dropped, but a considerable concentration still remained after 1000 years (Fig. 2ab, E). The catchment type with low soil variability was characterized by a very steep breakthrough of the percolation pathway (Fig. 2a) whereas the catchment with a high runoff fraction was characterized by a high initial concentration (Fig. 2ab, A). Note that the saturating concentrations of both catchments were identical. Furthermore, the percolation breakthrough was postponed at higher runoff values.
Fig. 2. Simulated response of the total P concentration leaving the catchment and flowing into the lake of four contrasting catchment types, with respect to catchment variability (σ) and runoff (R). Both catchments were loaded by agriculture with a P surplus of 15 kg P ha− 1 yr− 1 which lasted 700 years. Fig. a. = catchment variability σ is 0.1 m, fig. b. = catchment variability σ is 0.4 m. Bold or green line represents high runoff fraction (R = 0.2) and thin or blue line R = 0.05. Capitals: A: breakthrough of the surface runoff pathway, B: breakthrough of the percolation pathway, C: the catchment is completely saturated with P, D: the surface runoff pathway cleans up, E: slow recovery of the percolation water.
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
285
forward attraction basin that is separated by an unstable equilibrium line, connecting stable equilibrium lines, as seen in the work of Scheffer et al. (1993) and Scheffer (1998). This was caused by the fact that we used two differential equations to determine the nutrient state of the lake whereas Scheffer used just a single parameter to determine this state. 3.4. The coupled catchment–lake system The next step is to explore the dynamics of the coupled catchment–shallow lake system in two scenarios. In the first scenario agriculture was started in the first year with a P surplus level of 15 kg P ha− 1 yr− 1 and continued during the whole simulation, while in the second scenario the natural background P surplus was restored after 50 years. We followed the important state variables of this coupled system during 10,000 years. In the first scenario, after the start of the agricultural practise (Fig. 5a, A) we observed an increase of the Pi up
Fig. 3. Simulated response of the total P in lake water (a) and sediment concentrations (b) due to a sudden increase of the inflowing phosphorus concentration (Pi) from 0.01 to 0.5 mg P L− 1 at time is 50 years. The hair line indicates the long-term equilibrium concentration. 39 years after the P change the macrophytes disappear causing a second increase in the lake concentration.
The lake responded to the sudden change in the Pi with a quick jump of the lake concentration (0.0032 mg L− 1) to 0.061 mg L− 1 (Fig. 3a). This happened in a period of about one month. Subsequently the lake and the sediment concentrations gradually increased. 39 years after the Pi change the macrophytes disappeared, which caused a switch to the turbid state in which the lake P concentrations have risen within one month from 0.1 to 0.23 mg L− 1, resulting in a concentration that is close to equilibrium. Note that macrophytes enhance sedimentation and increase of P concentration of the sediment (Fig. 3b). Finally, the long term equilibrium analysis revealed the alternative equilibria of the lake and sediment state in response to the lake input concentration Pi (Fig. 4ab for a review see Scheffer et al., 2001). Here we see that the macrophytes reduce the lake P concentration but that they also enhance the sediment P state. Surprisingly our system does not show a straight-
Fig. 4. Calculated long term alternative equilibria of a shallow lake system. fig. a: equilibria of the total P in lake water, fig. b: equilibria of P in the sediment. Note that the presence of macrophytes reduces the lake phosphorus concentration but increases the phosphorus concentration of the top sediment layer.
286
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
to a relatively steady level. This was caused by the saturation of the top soil layer responsible for the surface run-off P, despite the fact the percolating water was still clean. After this, the lake P concentration gradually rose due to the increase in sediment P concentration, and after 75 years the macrophytes disappeared causing a sudden jump in the P lake concentration (Fig. 5a, B). After 380 years the percolation pathway was saturated causing another increase (Fig. 5a, D) to the final state, in which all compartments are saturated (Fig. 5a, E). Note that the final values of P inflow and the P lake were very high: Pi = 3.3 mg L− 1 and Pl = 2.1 mg L− 1 indicating that common agriculture in 50% of the catchment leads eventually to very high P concentrations. In the second scenario the cessation of the agricultural P input after 50 years induced a fast drop of the Pi concentration (Fig. 5b, C). This was caused by the fact that the cessation occurred in the stage when the runoff pathway was still dominant. Here the soil layer in
Fig. 5. Simulated catchment and lake dynamics of a coupled catchment shallow lake system. Fig. a: a pristine system in which agriculture was started with a phosphorus (P) surplus of 15 kg P ha− 1 yr− 1 in 50% of the catchment area which continues through time, fig. b: the same as before, however, here the agricultural practice was stopped after 50 years and the natural background P surplus was restored. Capitals: A: runoff breakthrough, B: macrophytes disappear from the lake in scenario a, C the agriculture is stopped in scenario b. D: percolation pathway breakthrough. E: system is in equilibrium with the current P surplus of the catchment. Different lines represent the total P concentrations in the soil, lake inflow, lake water and lake sediment.
contact with runoff water cleaned up quickly while the infiltrated phosphorus continued its pathway through the soil causing a second peak, 1500 years after the simulation started (Fig. 5b, D). The percolation peak was, however, not high enough to shift the lake to the turbid state. The initial conditions were restored after 4000 years (Fig. 5b, E) indicating that it might take about 4000 years to wash away 50 years worth of agricultural P. In this scenario the P concentration in the lake was always low enough to allow macrophytes. 3.5. Effects of catchment properties We conducted a series of simulations to explore the effects of catchment properties: the catchment variability (σ), the surface runoff fraction (R) and the water surplus of the catchment. When parameters were not explored we used standard values according to Table 1. There was no effect of the catchment variability on the runoff process. Therefore the graphs were identical during the first 50 years (Fig. 6a). After 62 years the macrophytes disappeared in the treatment with the highest catchment variability (0.4). In this simulation the P concentration in the lake already increased due to percolation processes after 30 years. This can be understood from the fact that in catchments with a high variability a larger part of the catchment has a relatively low buffer capacity for P. In the other simulations the macrophytes were lost after 76 years and the breakthrough of the percolation pathway occurred after 350 years. In all simulations the final P concentration in the lake was 2.04 mg L− 1, a value far above the level where macrophytes can survive (0.1 mg L− 1), indicating that an agricultural surplus of 15 kg ha yr− 1 in 50% of the catchment area led to a severely polluted lake system. These simulations indicate that catchment variability mainly affected the lake concentration in the long run. Differences in the surface runoff fraction had large effects on the P concentration in the lake (Fig. 6b). When the runoff fraction was high (R N 0.2) the macrophytes were lost within 20 years, whereas when the runoff fraction was relatively low (R b 0.1) macrophytes survived for more than 131 years. Finally, however, the macrophytes were lost after 213 years in all simulations. The final P level of all simulations in the lake was as in the preceding calculations 2.04 mg L− 1. These simulations indicate that high surface runoff leads to a fast increase of the P concentrations in the lake but does not effect the concentrations in the long run. Water surplus appeared to be an important factor (Fig. 6c). When the surplus was lower than 100 mm yr− 1 the
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
287
macrophytes disappeared within 11 years whereas relatively high water surpluses of 400 and 800 mm yr− 1 led to a survival of the macrophytes of more than 200 years. Eventually however, the macrophytes were lost after 230 years in all simulations. In contrast to both other simulation series the treatment did affect the final concentrations: High water surplus led to low concentrations due to dilution and vice versa. Note that the residence time of the lake was kept constant despite the higher water load. 3.6. Agricultural P surplus scenarios Here we try to answer the following question: what happens when we are able to reduce the agricultural P surplus in the agricultural area of our standard catchment? To explore this we varied the agricultural P surplus between 1 and 25 kg ha− 1 yr− 1. When the agricultural P surplus was 25 kg P ha− 1 yr− 1 the macrophytes had already disappeared after 15 years (Fig. 7a). Lower surplus values led to increased macrophyte longevity, at a surplus concentration of 1 kg P ha− 1 yr− 1 the P lake level even stayed below the critical level for the macrophytes during the whole simulation period. Next we explore the effect of cleaning a saturated catchment–lake system, which is initially in equilibrium with a certain agricultural P surplus and in which a pristine P surplus of 0.0225 kg P ha− 1 yr− 1 was restored at the start of the simulation. Results of this exploration are displayed in Fig. 7b. The numbers in the Fig. 7b do not refer to agricultural P load but to initial saturation conditions of the catchment. The recovery of a saturated catchment–lake system lasted a very long time (Fig. 7b). In systems saturated with an agricultural P load above 2.5 kg P ha− 1 yr− 1 the macrophytes did not recover the first 1000 years whereas at a surplus of 1.0 kg P ha− 1 yr− 1 the critical macrophyte concentration of 0.1 mg L− 1 was not reached. In none of the simulations the equilibrium concentration with natural P surplus of 0.0225 was reached within 1000 years. It took 4400 years to approach this equilibrium. Subsequently we evaluated the pollution and recovery of our pristine standard system that receives a range of agricultural P surpluses (1–25 kg P ha− 1 yr− 1) during Fig. 6. Calculated effects of catchment variability (fig. a), the surface runoff fraction (fig. b), and the water surplus of the catchment (fig. c), on the total phosphorus (P) concentration of the lake. In the initial conditions the catchment–lake system were pristine. At time zero in 50% of the area, agriculture was started with a P surplus of 15 kg ha− 1 yr− 1. The catchment variability parameters were 0.02, 0.05, 0.1. 0.2 and 0.4 m respectively (fig. a). The surface runoff fraction was varied between 0.01 and 0.5 (fig. b), and the water surplus levels were between 50 and 800 mm yr− 1 (fig. c).
288
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
Fig. 7. Agricultural phosphorus (P) surplus scenarios: fig. a: initial conditions: a catchment–lake system in equilibrium with a natural P surplus of 0.0225 kg P ha− 1 yr− 1 followed by continuous agriculture in 50% of the area with an agricultural P surplus of 1, 2.5, 5, 10 and 25 kg P ha− 1 yr− 1 (different lines), fig. b: the catchment–lake system was initially in equilibrium with an agricultural P surplus of 0.025, 1, 2.5, 5, 10, 25 kg P ha− 1 yr− 1 (different lines), at time zero the natural background P surplus was restored to 0.0255 kg P ha− 1 yr− 1, fig. c: the natural background surplus was restored after 50 years. fig. d: the agricultural P surplus was reduced to the target load of 5 kg P ha− 1 yr− 1 after 50 years.
the first 50 years where after the pristine P surplus was restored (Fig. 7c). Only in the 25 P surplus treatment macrophytes were lost. These macrophytes reappeared again 48 years after the pristine P surplus was restored. In all simulations the lake P concentrations dropped very quickly after the pristine P surplus was restored, but in the long run however, percolating P induced a steady rise of the concentration causing a new peak. The pristine situation was reached again after 4400 years in all the simulations. Finally, we studied the recovery of a system in which the agriculture was able to reduce its P surplus level to the current target of 5 kg P ha− 1 yr− 1 (Smith et al., 1995) after 50 years. In the treatments where the initial P
surplus was 20 or 25 kg ha− 1 yr− 1 the macrophytes disappeared within 33 years and did not reappear within the simulation (Fig. 7d). At P surpluses less than 15 kg ha − 1 yr − 1 the macrophytes survived more than 531 years. The final lake P concentration in all simulations was 0.6 g mg L− 1, which can be still considered to be a strongly polluted system. 4. Useful equations Several equations can be derived from our model that might be useful to estimate equilibrium values, time scales and critical values of lake and catchment processes for other catchment–lake systems.
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
4.1. Estimating long term lake concentrations from agricultural P surplus data Simulations performed with the same water and P surplus levels tend to have the same final lake concentration (e.g. Fig. 6ab). This can be understood from the fact that when the absorption capacity of the catchment–lake system is saturated, the net agricultural P surplus equals the net P loss. In this case we can calculate the saturated concentration in the lake from the agricultural P surplus and the water surplus of the catchment when we know the ratio (Fp) between the lake inflow concentration and lake water concentration. Given our model, the equilibrium P concentration in the lake (Pl⁎ mg L− 1) of a saturated catchment–lake system is: Pl⁎ ¼ Fp d
100 ðSa d A þ Sn ð1−AÞÞ W
ð5Þ
where: W is the precipitation surplus (mm yr− 1), A is the fraction of the catchment area that is used for agriculture. Sa is the phosphorus surplus of the agricultural area. Sn is the phosphorus surplus of other areas in the catchment e.g. nature. Fp can be estimated with Eq. (9) or with the Vollenweider equation.
289
4.2. Estimation of the critical agricultural P surplus The next question is which agricultural P surplus is allowed to preserve a sustainable clear lake with macrophytes in the long run. For this purpose we rearrange Eq. (5) in which Sac is now the critical agricultural P surplus (kg ha− 1 yr− 1) and Plc is the critical lake concentration where macrophytes just can survive: WPlc 1 þ Sn 1− Sac ¼ A 100Fp A
ð6Þ
If the critical P concentration (Plc) in the lake is 0.1 mg L− 1 and if Fp = 0.62 (average Fp in turbid Dutch lakes), the maximum acceptable agricultural P surplus (Sac) for our standard catchment is 0.7 kg ha yr− 1. This is far below common practise and still 7 times lower than the desired agricultural surplus according to Smith et al. (1995). If we explore this equation for several values of the precipitation surplus W and the fraction of the standard catchment that is in use for agriculture, A, we see that normal agricultural practise (15 kg P ha yr− 1) is only sustainable at very low catchment fractions and very high precipitation surpluses (Fig. 8).
Fig. 8. Contour plot of the critical P surplus values (kg P ha yr− 1) needed to sustain a permanently clear lake of the standard catchment–lake system, affected by the precipitation surplus and the fraction of agriculture in the catchment.
290
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
4.3. Estimation of time scales of catchment saturation Here we want to predict the time it takes for a catchment in equilibrium with certain agricultural P surplus (Sa1), to approach a new equilibrium determined by a higher agricultural P surplus Sa2. This can be done because we can calculate the storage capacity of the soil (Eq. (2)) at both P surplus levels and we know that the net input of the soil approximately equals (Sa2 − Sa1). The time span of the percolation pathway (Lp in years) from the change in the agricultural surplus level to the centre of the responding breakthrough curve is: Lp ¼
10Qm ðTt −Tr Þ d ðSa2 −Sa1 Þð1−RÞ
! !! Sa2 Sa1 − W W Sa2 þ 100K Sa1 þ 100K ð7Þ
Here Tt is the thickness of the soil profile and Tr is the thickness of the surface soil layer in contact with runoff water, K is the affinity of the soil for P (m3 g− 1) and R is the runoff fraction. For the duration of the runoff process (Lr in years) nearly the same equation can be used, only (Tt − Tr) should be replaced by Tr because only the surface layer (Tr) is involved, and R should be zero since all P is passing the runoff soil layer. This equation can be used to estimate the catchment response time due to an increase in the P surplus level, however, it cannot be used to estimate response times due to a decrease in P surplus levels!
constant. This assumption can be made since the sediment response time is much slower than the water response time. lnC Lw ¼ 1 s − sþD
ð10Þ
C represents the proportional distance from the new equilibrium, relative to the intercept between the old and the new equilibrium, that is tolerated for the calculation of Lw, e.g. when C = 0.2 Lw gives the time until the lake water is 20% from the new equilibrium. For our lake, Lw is 27.7 days. Note that Lw is independent of both old and new equilibrium levels. 4.6. Sediment response time Finally, the sediment response time (Ls in days) affected by a change in Pi can be derived from Eqs. (3) and (4), since the lake water can be regarded to be in equilibrium because of its relatively fast response time. lnC Ls ¼ ð11Þ iss Dþss−i−b For our standard lake Ls is 25,404 days which is 69.6 years. 5. Discussion
4.4. Lake water and sediment equilibrium concentrations
5.1. Rainfall and surface runoff
The long-term equilibrium values of the sediment P (Ps in g m− 2) and lake P concentration (Pl in mg L− 1) of the whole lake system as described by Eqs. (3) and (4) can be derived from following equations: sPi Ps⁎ ¼ ð8Þ sbs ði þ bÞ þ D Pi ði þ bÞ Pl⁎ ¼ ð9Þ sbs ði þ bÞ þ D
It is common knowledge that high rainfall is associated with high P concentrations because intensive rainfall promotes surface runoff which contains larger amounts of P (see e.g. Gardner et al., 2002). In our model, water surplus on a yearly basis and the run fraction are independent parameters that cause an inverse relationship, because the large water surplus dilutes the P surpluses, which results in lower concentrations. This was done because high runoff is not directly dependent on the total water surplus on a yearly basis, but also relates to how individual showers and rainfall is distributed over the year, the permeability of soils for water, the topography and hydrology of the watershed and the rainfall–evaporation ratio. These properties are typically associated with the geography and topography of a catchment. Therefore we regard the runoff fraction to be a catchment property in our analysis and do not relate it to the water surplus so potential users can define their catchment by these two parameters which make it possible to describe a specific catchment more accurately.
These equations can be used to directly assess the final lake water and sediment equilibria at a given lake inflow concentration (Pi). 4.5. Lake water response time The next step is to estimate the lake water response time (Lw in days) to a fast shift in Pi. This can be derived from Eq. (3) under the assumption that the sediment P is
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
5.2. Agricultural P surplus and long term equilibria Several conclusions can be drawn from our results regarding agricultural P surpluses. Given our standard catchment, current agricultural P surpluses of about 15– 25 kg P ha yr− 1 will eventually lead to a very turbid lake conditions with concentrations of 2.0–3.4 mg L− 1, which is far above the critical level where macrophytes can survive. But even the desired P surplus of 5 kg P ha− 1 yr− 1 (Smith et al., 1995) will in the long run lead to concentrations of 0.6 mg L− 1, which will still induce very turbid lake conditions. To allow macrophyte growth, P lake concentrations should be about 0.1 mg L− 1 (Van Nes et al., 2002). To secure this concentration in our standard catchment–lake system the agricultural P surplus should be lower than 0.7 kg P ha− 1 yr− 1. Van der Molen et al. (1998) reported a value of 0.42 kg P ha yr− 1 to secure permanent clear lakes. These values are 5–7 times lower than the current target agricultural P surplus level indicating that even target agricultural surplus values are not sustainable in the long run. Clearly in catchments with a relatively small agricultural area and a high water surplus, current agricultural practice might be less critical because higher water loads and larger natural areas will dilute the agricultural P surpluses (Eq. (6)). On the other hand, in many areas of Europe the water surplus is much lower than in Dutch conditions (Meinardi et al., 1994) and in this situation the effect of the P surplus is even more critical.
291
after 4400 years. Surprisingly the initial P surplus level that determines the equilibrium soil P level did not affect the recovery time (Fig. 7b). Systems that received P loads from temporary agriculture over 50 years recovered faster, mainly because the runoff pathway cleaned up quickly. In addition to this, the percolation breakthrough was postponed and diluted. Nevertheless, our work shows a clear noticeable peak from the percolation pathway long after the agriculture has ceased, which might cause a threat in the far future. 5.4. Buffers and time delays Four buffers can be distinguished in our model that cause temporal delay with respect to the final equilibria of the lake P concentration: (1) the surface soil layer in contact with runoff water, (2) the percolation soil layer, (3) the lake water and (4) the lake sediment. As stated in the preceding sections the delay induced by the percolation and runoff buffer depends strongly on the agricultural P surplus levels (Fig. 9). In contrast, the lake and sediment response times were independent of the P load (Fig. 9, Eqs. (10) and (11)). This can be understood from the fact that high Pi values induce a fast increase of the P concentration in the lake but at the same time are also creating new higher lake equilibria. The four distinguished buffers are responsible for contrasting time delays (Fig. 9). The smallest buffer is the lake water, which reaches equilibrium in about one month.
5.3. Agricultural P surplus levels and time delays Another aspect of the agricultural P surplus is the time scale elapsing until a critical concentration is reached. High P surpluses of 25 kg ha− 1 yr− 1 lead to a fast saturation of the upper soil layer in close contact with surface runoff water. In this case macrophytes were lost in 15 years whereas the breakthrough of the percolation pathway occurred after 203 years. At the target P surplus level of 5 kg P ha− 1 yr− 1 the macrophytes disappear after 631 years in our standard catchment. This slow response is due to the fact that the P flux from the surface runoff pathway alone is not large enough to elevate the lake P concentration above the critical level of macrophyte survival. In this case the breakthrough of the percolation pathway determines the disappearance of the macrophytes. These results suggest that limited agricultural P surplus, although not sustainable on the long run, may delay turbid lake conditions for many years. The recovery of saturated systems takes a long time; it takes about 1300 years for macrophytes to reappear in systems in equilibrium with an agricultural P surplus of 2–25 kg P ha yr− 1. The final equilibrium was restored
Fig. 9. Response times of various buffers in a lake–catchment system. Dark or green bars represent response times due to an agricultural surplus of 25 kg P ha− 1 yr− 1 whereas light or blue bars are response times due to an agricultural surplus of 1 kg P ha− 1 yr− 1. Note that P load does not affect the response times of the lake sediment and lake water.
292
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
The next buffer is the surface soil layer in contact with runoff water. Depending on the agricultural P surplus, a delay of 5–50 years can be expected. The lake sediment is responsible for a delay of 69 years in our calculations. This value is determined by the lake properties and is independent of the P load (Eq. (11)). The most important buffer is the percolation soil layer, which causes an estimated delay of 150–1700 years. These time scales are well in line with literature (Marsden, 1989; Behrendt et al., 1996; Oenema and Roest, 1998; Sondergaard et al., 2001; Carpenter, 2005). Additionally these buffers differ with respect to how their final new equilibrium is reached. Water from soil buffers initially shows no sign of response to the change of the agricultural P surplus. Subsequently, the concentration rises suddenly when the soil is near saturation, resulting in a typically sigmoid shape (Vanderzee et al., 1989). The lake processes, on the other hand, show the opposite characteristic. In response to a change in Pi, the lake P concentration is initially changing fast, but this change is slowing down when the lake water concentration is approaching equilibrium, resulting in a typically negative exponential curve (see e.g. Marsden, 1989). The responses of the percolation pathway are especially problematic because of the large time delays and the fact that when a change in the concentration is measured, the soil profile is already substantially loaded. Therefore means that measures will only be effective in the long run.
To secure low concentrations in the long run, sustainable measures are necessary. These are: (a) reducing agricultural surplus to sustainable levels and (b) reducing the agricultural area of a catchment to sustainable dimensions (see Fig. 8 and Eq. (6)). If we do not want to burden future generations with our waste, we should choose sustainable measures. These measures, however, will put large constraints on agricultural practise in terms of P surplus levels (Withers and Jarvis, 1998) and the quantity and distribution of agricultural areas in and around catchments. Furthermore, our work shows that soil and lake buffers might prevent lake recovery for many years, even when pristine P surplus levels are restored. This indicates that the targets of the EC Water Frame Work directive, to have a good ecological and chemical status in 2015, will not be met in many occasions.
5.5. Postponing and sustainable measures
Acknowledgements
We can distinguish two kinds of measures to improve lake water quality: postponing measures and sustainable measures. Postponing measures are measures that will eventually lead to a polluted lake system when the catchment–lake system reaches final equilibrium, and will increase the time span until critical lake P concentrations are reached. Several postponing measures can be distinguished:
This research was funded by the EC program BUFFER: “Key nutrient transport mechanisms important for the prediction of nutrient and phytoplankton concentrations in European standing waters”, contract number EVK1-CT-1999-00094. We thank: the Buffer scientist for their stimulating discussion and great hospitality, two anonymous reviewers for their helpful comments and Jayne E. Rattray for improving the English of this paper.
(a) Reducing surface runoff by increasing infiltration into the soils. This measure will increase the percolation pathway, which is principally much slower and will postpone the eventually high P concentrations in open water. (b) Reducing agricultural surplus but only to nonsustainable levels e.g. the target level of 5 kg ha− 1 yr− 1. A reduction in the agricultural surplus will not only cause a reduced final concentration at equilibrium but also delays the breakthrough of the surface runoff and percolation pathway. (c) Selection of deep good P absorbing soils in the catchment for agricultural practise. The amount of
P absorbed by the soil is determined by the thickness of the soil layer and the soil properties like Q and K. When the absorption capacity is high the residence time of P in the soil is also high which will lead to a postponed breakthrough. (d) Selection of low variability soils for agriculture. When soil variability is high, flow through and adsorbing properties will make the breakthrough curve of the percolation pathway less steep allowing an earlier rise of the P concentration of the percolation water.
References Arheimer B, Liden R. Nitrogen and phosphorus concentrations from agricultural catchments — influence of spatial and temporal variables. J Hydrol 2000;227:140–59. Behrendt H, Lademann L, Pagenkopf WG, Pothig R. Vulnerable areas of phosphorus leaching — detection by GIS-analysis and measurements of phosphorus sorption capacity. Water Sci Technol 1996;33:175–81. Carpenter SR. Eutrophication of aquatic ecosystems: bistability and soil phosphorus. Proc Natl Acad Sci U S A 2005;102:10002–5. Chapman AS, Foster IDL, Lees JA, Hodgkinson RJ, Jackson RH. Sediment and phosphorus delivery from field to river via land drains in England and Wales. A risk assessment using field and national databases. Soil Use Manage 2003;19:347–55.
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294 Cooper DM, House WA, Reynolds B, Hughes S, May L, Gannon B. The phosphorus budget of the Thame catchment, Oxfordshire: 2. Modelling. Sci Total Environ 2002;282:435–57. Djodjic F, Bergstrom L, Grant C. Phosphorus management in balanced agricultural systems. Soil Use Manage 2005;21: 94–101. Edwards AC, Withers PJA. Soil phosphorus management and water quality: a UK perspective. Soil Use Manage 1998;14:124–30. Falkowski P, Scholes RJ, Boyle E, Canadell J, Canfield D, Elser J, et al. The global carbon cycle: a test of our knowledge of earth as a system. Science 2000;290:291–6. Gardner CMK, Cooper DM, Hughes S. Phosphorus in soils and field drainage water in the Thame catchment, UK. Sci Total Environ 2002;282:253–62. Gillingham AG, Thorrold BS. A review of New Zealand research measuring phosphorus in runoff from pasture. J Environ Qual 2000;29:88–96. Groenendijk P, Kroes JG. Modelling the nitrogen and phosphorus leaching to ground water and surface water with ANIMO 3.5. Wageningen: Alterra; 1999. p. 138. Heijboer D, Nellestijn J. Klimaat atlas van Nederland, de normaal periode 1971–2000. Rijswijk: Elmar; 2002. p. 182. Horppila J, Nurminen L. Effects of submerged macrophytes on sediment resuspension and internal phosphorus loading in Lake Hiidenvesi (southern Finland). Water Res 2003;37:4468–74. Johnes PJ. Evaluation and management of the impact of land use change on the nitrogen and phosphorus load delivered to surface waters: the export coefficient modelling approach. J Hydrol 1996;183: 323–49. Johnes PJ, Hodgkinson RA. Phosphorus loss from agricultural catchments: pathways and implications for management. Soil Use Manage 1998;14:175–85. Jordan C, McGuckin SO, Smith RV. Increased predicted losses of phosphorus to surface waters from soils with high Olsen-P concentrations. Soil Use Manage 2000;16:27–35. Jorgensen SE, Nielsen SN, Jorgensen LA. Handbook of ecogical parameters and ecotoxicology. Amsterdam: Elsevier; 1991. p. 1263. Kirkby CA, Smythe LJ, Cox JW, Chittleborough DJ. Phosphorus movement down a toposequence from a landscape with texture contrast soils. Aust J Soil Res 1997;35:399–417. Kleinman PJA, Sharpley AN, Veith TL, Maguire RO, Vadas PA. Evaluation of phosphorus transport in surface runoff from packed soil boxes. J Environ Qual 2004;33:1413–23. Koopmans GF, McDowell RW, Chardon WJ, Oenema O, Dolfing J. Soil phosphorus quantity–intensity relationships to predict increased soil phosphorus loss to overland and subsurface flow. Chemosphere 2002;48:679–87. Lijklema L, Hieltjes AHM. A dynamic phosphate budget model for a eutrophic lake. Hydrobiologia 1982;91-2:227–33. Marsden MW. Lake restoration by reducing external phosphorus loading — the influence of sediment phosphorus release. Freshw Biol 1989;21:139–62. McGechan MB, Lewis DR. Sorption of phosphorus by soil, part 1: Principles, equations and models. Biosyst Eng 2002;82: 1–24. McIntosh DH, Thom AS. Essentials of Meteorology. London: Taylor & Francis; 1981. p. 239. Meinardi CR, Beusen AHW, Bollen MJS, Klepper O. Vulnerability to diffuse pollution of European soils and ground water. Bilthoven: RIVM; 1994. p. 21. Merseburger GC, Marti E, Sabater F. Net changes in nutrient concentrations below a point source input in two streams draining catchments with contrasting land uses. Sci Total Environ 2005;347: 217–29.
293
Moss B, Stephen D, Alvarez C, Becares E, Van de Bund W, Collings SE, et al. The determination of ecological status in shallow lakes — a tested system (ECOFRAME) for implementation of the European Water Framework Directive. Aquat Conserv—Mar Freshw Ecosyst 2003;13:507–49. Oborn I, Edwards AC, Witter E, Oenema O, Ivarsson K, Withers PJA, et al. Element balances as a tool for sustainable nutrient management: a critical appraisal of their merits and limitations within an agronomic and environmental context. Eur J Agron 2003;20:211–25. Oenema O, Roest CWJ. Nitrogen and phosphorus losses from agriculture into surface waters. The effects of policies and measures in The Netherlands. Water Sci Technol 1998;37:19–30. Portielje R, Van der Molen DT. Relationships between eutrophication variables: from nutrient loading to transparency. Hydrobiologia 1999;409:375–87. Scheffer M. Ecology of shallow lakes. London: Chapman and Hall; 1998. p. 357. Scheffer M, Carpenter S, Foley JA, Folke C, Walker B. Catastrophic shifts in ecosystems. Nature 2001;413:591–6. Scheffer M, Hosper SH, Meijer ML, Moss B, Jeppesen E. Alternative equilibria in shallow lakes. Trends Ecol Evol 1993:275–9. Schindler DW. Evolution of phosphorus limitation in lakes. Science 1977;195:260–2. Schippers P, Joenje W. Modeling the effect of fertilizer, mowing, disturbance and width on the biodiversity of plant communities of field boundaries. Agric Ecosyst Environ 2002;93:351–65. Schippers P, Vermaat JE, de Klein J, Mooij WM. The effect of atmospheric carbon dioxide elevation on plant growth in freshwater ecosystems. Ecosystems 2004;7:63–74. Sharpley AN. Dependence of runoff phosphorus on extractable soil— phosphorus. J Environ Qual 1995;24:920–6. Sharpley AN, Chapra SC, Wedepohl R, Sims JT, Daniel TC, Reddy KR. Managing agricultural phosphorus for protection of surface waters — issues and options. J Environ Qual 1994;23:437–51. Smith RV, Lennox SD, Jordan C, Foy RH, McHale E. Increase in soluble phosphorus transported in drain flow from a grassland catchment in response to soil phosphorus accumulation. Soil Use Manage 1995;11:204–9. Sondergaard M, Jensen JP, Jeppesen E. Internal phosphorus loading in shallow Danish lakes. Hydrobiologia 1999;409:145–52. Sondergaard M, Jensen JP, Jeppesen E. Retention and internal loading of phosphorus in shallow eutrophic lakes. Sci World 2001;1:427–42. Syversen N. Effect and design of buffer zones in the Nordic climate: the influence of width, amount of surface runoff, seasonal variation and vegetation type on retention efficiency for nutrient and particle runoff. Ecol Eng 2005;24:483–90. Ulen B, Jakobsson C. Critical evaluation of measures to mitigate phosphorus losses from agricultural land to surface waters in Sweden. Sci Total Environ 2005;344:37–50. Van den Berg MS, Scheffer M, Coops H, Simons J. The role of characean algae in the management of eutrophic shallow lakes. J Phycol 1998;34:750–6. Van der Molen DT, Los FJ, Vanballegooijen L, Vandervat MP. Mathematical-modeling as a tool for management in eutrophication control of shallow lakes. Hydrobiologia 1994;276:479–92. Van der Molen DT, Portielje R, Boers PCM, Lijklema L. Changes in sediment phosphorus as a result of eutrophication and oligotrophication in Lake Veluwe, The Netherlands. Water Res 1998;32:3281–8. Van Donk E, Gulati RD, Iedema A, Meulemans JT. Macrophyte-related shifts in the nitrogen and phosphorus contents of the different trophic levels in a biomanipulated shallow lake. Hydrobiologia 1993;251: 19–26.
294
P. Schippers et al. / Science of the Total Environment 369 (2006) 280–294
Van Nes EH, Scheffer M, van den Berg MS, Coops H. Dominance of charophytes in eutrophic shallow lakes — when should we expect it to be an alternative stable state? Aquat Bot 2002;72:275–96. Vanderzee S, Vanriemsdijk WH. Model for long-term phosphate reaction-kinetics in soil. J Environ Qual 1988;17:35–41. Vanderzee S, Leus F, Louer M. Prediction of phosphate-transport in small columns with an approximate sorption kinetics model. Water Resour Res 1989;25:1353–65. Vollenweider RA. Advances in defining critical loading levels for phosphorus in lake eutrophication. Mem Inst Ital Idrobiol 1976;33: 53–83.
Wassen MJ, Venterink HO, Lapshina ED, Tanneberger F. Endangered plants persist under phosphorus limitation. Nature 2005;437:547–50. Withers PJA, Jarvis SC. Mitigation options for diffuse phosphorus loss to water. Soil Use Manage 1998;14:186–92. Withers PJA, Edwards AC, Foy RH. Phosphorus cycling in UK agriculture and implications for phosphorus loss from soil. Soil Use Manage 2001;17:139–49. Wood FL, Heathwaite AL, Haygarth PM. Evaluating diffuse and point phosphorus contributions to river transfers at different scales in the Taw catchment, Devon, UK. J Hydrol 2005;304:118–38.