The combined effect of fertiliser nitrogen and phosphorus on herbageyield and changes in soil nutrients of a grass/clover and grass-only sward

René Schils* and Paul SnijdersResearch Institute for Animal Husbandry, P.O. Box 2176, 8203 AD Lelystad, The Netherlands; *Author forcorrespondence (fax: +31-320-242584; e-mail: rene.schils@wur.nl)

Received 17 October 2002; accepted in revised form 28 August 2003

Key words: Nitrogen, Nutrient efficiency, Perennial ryegrass, Phosphorus, Soil, White clover

Abstract

The combined effect of reduced nitrogen �N� and phosphorus �P� application on the production of grass-only andgrass/clover swards was studied in a five-year cutting experiment on a marine clay soil, established on newlysown swards. Furthermore, changes in soil N, P and carbon �C� were measured. Treatments included four P �0,35, 70 and 105 kg P ha–1 year–1� and three N levels �0, 190 and 380 N kg ha–1 year–1� and two sward types�grass-only and grass/clover�. Nitrogen was the main factor determining the yield and quality of the harvestedherbage. On the grass-only swards, N application increased the DM yield with 28 or 22 kg DM kg N–1, at 190or 380 kg N ha–1 year–1, respectively. The average apparent N recovery was 0.78 kg kg–1. On the grass/cloverswards, N application of 190 ha–1 year–1 increased grass production at the cost of white clover, which decreasedfrom 41 to 16%. Phosphorus application increased grass yields, but did not increase clover yields. A positiveinteraction between N and P applications was observed. However, the consequences of this interaction for theoptimal N application were only minor, and of little practical relevance. Both the P-AL-value and total soil Pshowed a positive response to P application and a negative response to N application. Furthermore, the positiveeffect of P application decreased with increasing N application. The annual changes in P-AL-value and total soilP were closely related to the soil surface surplus, which in turn was determined by the level of N and P appli-cation and their interaction. The accumulation of soil N was similar on both sward types, but within the grass-only swards soil N was positively affected by N application. The accumulation of organic C was unaffected by Nor P application, but was lower under grass/clover than under grass-only.

Introduction

Over the last 40 years, production grasslands in theNetherlands have been fertilised with increasingamounts of nitrogen �N� and phosphorus �P�. In 1990,the fertiliser application on specialised dairy farmswas at a level of 304 kg N and 19 kg P ha–1 year–1.Next to fertiliser, grasslands have received N and Pfrom animal excreta, either directly during grazing orthrough application of slurry or farmyard manure. Fora typical dairy farm on sandy soil in the early 1980s,Aarts et al. �1992� estimated an additional input of164 kg N and 25 kg P ha–1 year–1 through animal ex-

creta during grazing and 120 kg N and 30 kg P ha–1

year–1 through slurry application.The amounts of N and P applied to grasslands ex-

ceed the uptake, causing environmental problems likenitrate leaching to groundwater �Fraters et al. 1997;Oenema et al. 1998� and N and P eutrophication ofsurface water �Oenema and Roest 1998�. Therefore,from 1985 onwards, the Dutch government has intro-duced a series of measures that aim to reduce the Nand P losses from farming �VROM 1989�. In 1998,the Mineral Accounting System �Minas� was intro-duced, which is an N and P accounting system onfarm level �MANMF 1997�. These measures will lead

© 2004 Kluwer Academic Publishers. Printed in the Netherlands.165Nutrient Cycling in Agroecosystems 68: 165–179, 2004.

to a reduction in the use of fertiliser N and P ongrassland and a substitution of fertiliser N by biologi-cally fixed N �Schils et al. 2000a, b�, the latter mainlythrough the increased use of mixtures of perennialryegrass �Lolium perenne L.� and white clover �Tri-folium repens L.�.

Although much is known about the individual ef-fects of N and P on grassland production �e.g., Agter-berg and Henkens 1995; Vellinga and André 1999�,there has been little attention for the combined effectsof N and P, especially in field trials. The majority offertiliser N experiments has been carried out withample fertiliser P and vice versa. Furthermore, P ap-plication on grass/clover swards and its interactionwith N application has not received any attention inthe Netherlands. Furthermore, there is a lack ofknowledge regarding the changes in soil P followingreduced P application. Amongst farmers there is con-cern that reduced P surpluses on grassland will leadto sub-optimal P levels in the soil, which in turn leadsto a reduced herbage yield and quality. In an earlierstudy, using data from dairy farm monitoring projects,it was estimated that a P surplus of 20 to 25 kg ha–1

year–1 was necessary to maintain soil P levels at anagriculturally optimal value �Oenema and Van Dijk1994�. On the other hand, the same study indicatedthat environmentally acceptable P surpluses are aslow as 0.5 kg ha–1 year–1.

The objectives of the present experiment are to �i�quantify the combined effect of fertiliser N and P onherbage production, herbage quality and soil P ofgrass and grass/clover swards, �ii� increase the under-standing of P utilisation in grass/clover swards, and�iii� provide a basis for P recommendations on grass/clover swards.

Materials and methods

Site

The experiment was established in Lelystad �52 °N�on a well drained sedimentary calcareous light marineclay soil, reclaimed from the IJssel Lake in 1957. Thesite has been used for dairy farming since 1973, firstwith amply fertilised perennial ryegrass dominatedswards, later with moderately fertilised perennialryegrass/white clover mixtures.

In January 1994, the experimental site wasploughed to a depth of 25 cm. In April, all plots weresown with 20 kg ha–1 of perennial ryegrass cvs. Her-bie and Exito �50/50�. Additionally, the grass/cloverplots were sown with 5 kg ha–1 of white clover cvs.Alice and Retor �50/50�.

As a result of ploughing, the nutrient availabilityin the soil tended to increase with depth, up to 25 cm�Table 1�. According to the Dutch fertiliser recom-mendations �PR 1998�, the potassium values werevery high and the P-AL values low. To prevent anynegative effect of the low P-AL values on plant es-tablishment, the whole field received a P starterdressing of 4.4 kg P ha–1 as triple superphosphate.

Weather data were obtained from the nearestweather stations �KNMI 1993–1999�. The precipita-tion surplus was calculated as the difference betweenprecipitation and reference evaporation, according toMakkink �Hooghart and Lablans 1988�. The temper-ature sum, at 1 m above soil level, was calculatedfrom the first of January onwards, as the sum of theaverage of daily maximum and minimum tempera-tures, if the average was above 0 °C. During the ex-periment, all growing seasons were warmer than

Table 1. Soil characteristics at the start of the experiment �April 1994�.

Sampling depth �cm�

0–5 5–10 10–15 15–20 20–25 25–30

Density �kg l–1� 0.97 1.17 1.23 1.26 1.15 1.29Particles � 16 �m �%� 37 38 40 38 37 38Organic matter �%� 3.4 3.4 3.6 4.0 4.6 3.6pH-KCl 7.3 7.2 7.2 7.2 7.2 7.2K-HCl �mg K 100 g–1� 22 24 27 31 31 24P-AL �mg P2O5 100 g–1� 12 11 14 15 19 13Total P �mg P 100 g–1� 54 55 58 61 66 58Total N �mg N 100 g–1� 151 155 168 187 203 165Organic C �mg C 100 g–1� 1857 2077 2103 2167 2430 1953Corg/N ratio 12.3 13.4 12.5 11.6 12.0 11.8

166

normal and, with the exception of the last season, alsodrier than normal �Table 2�. The autumn and winterperiods were generally warmer and equally wet orwetter than normal. Only the autumn/winter period of1995/1996 was drier and colder than normal. The ex-perimental field was not irrigated.

Treatments

The experiment was a randomised block trial withthree replicates of combinations of two sward types�grass and grass/clover�, three N levels �0, 200 and400 kg ha–1 year–1, designated as N0, N1 and N2, re-spectively� and four P levels �0, 35, 70 and 105 kgha–1 year–1, designated as P0, P1, P2 and P3, respec-tively�. Grass swards were combined with all three Nlevels, while grass/clover swards were only combinedwith 0 and 200 kg N ha–1 year–1. The five resultingsward type�N level treatments were combined withthe four P levels, resulting in 20 treatment combina-tions altogether.

Nitrogen was applied as calcium ammonium nitrate�27% N� and P as triple superphosphate �20% P�. Thedistribution of the annual application over the first 6cuts was 25, 20, 20, 15, 10 and 10% for N, and 25,15, 15, 15, 15 and 15% for P. In the establishmentyear, the first fertiliser was applied at sowing and inthe following years between the 16th and 24th ofMarch at a temperature sum of 514, 253, 521 and578 °C in the consecutive years. Throughout the ex-periment, the whole field was fertilised with 42 kg Kha–1 cut–1 as potassium chloride �50% K�.

Harvests were planned to take place when the fast-est growing plots yielded approximately 3.5 t DMha–1 for the first cut, and approximately 2.5 t DM ha–1

for later cuts. All treatments were harvested on thesame day. In each year, five cuts were taken, exceptin 1995, when seven cuts were taken. As a result, theaverage annual application rates were somewhatlower than the planned application rates. Excludingthe establishment year, the first cut was harvested be-tween 20 April and 21 May. The following cuts wereharvested after an average growing period of 34 days,with a range of 20 to 56 days. In 1998, some harvestswere delayed due to wet weather, which increased theDM yield at cutting.

Soil samples were taken at the time of sowing to adepth of 30 cm, in layers of 5 cm each. Samples werebulked per replicate and analysed for pH-KCl, texture�particles � 16 �m�, K-HCl, P-AL �Egnér et al.1960; Tunney et al. 1997�, total P �destruction withFleischmann acid�, total N �oxidation at 1050 °C�, or-ganic matter �loss on ignition at 550 °C, corrected forclay particles and CaCO3� and organic C �oxidationat 600 °C�. The methods are conform those describedby Houba et al. �1997�. Additionally, undisturbed soilsamples were taken from the same layers to calculatesoil density.

To obtain information about soil compaction aftersowing, metal plates �20 cm�20 cm� were placed ata depth of 30 cm on each corner of the trial field. Toprevent disturbance of the soil profile, the plates wereinserted laterally from a hole, dug at the side of themarked place. The depth of the metal plates wasmeasured prior to each soil sampling. In March 1995,the depth of the metal plates was 28 cm, and there-fore the sampling depths for the following years wereadjusted to 0–5, 5–10, 10–23 and 23–28 cm.

In the following years, soil samples were taken inMarch, prior to the first fertiliser applications.

Table 2. Mean values of daily minimal, mean and maximal temperatures and total precipitation surplus �precipitation – reference evapora-tion� per period of six months �KNMI 1993–1999�.

April – September October – March

Temperature �°C� Surplus Temperature �°C� Surplus

Min Mean Max �mm� Min Mean Max �mm�

30 Year mean 9.0 13.6 18.6 � 35 1.7 4.9 8.1 2821993/1994 2.1 5.1 8.0 3631994/1995 9.9 14.7 19.4 � 83 3.5 6.8 10.0 5211995/1996 10.1 15.0 20.1 � 98 0.3 3.7 6.9 401996/1997 8.5 13.8 18.5 � 115 1.8 5.0 8.2 2781997/1998 9.6 14.8 19.7 � 154 3.4 6.7 9.7 2731998/1999 10.3 14.7 19.1 131 2.6 5.7 8.6 480

167

Samples were bulked per treatment and analysed forP-AL, total P, total N, organic matter and organic C.Only at the final sampling date, March 1999, soilsamples were analysed per plot. Soil density was alsomeasured at the sampling dates in 1996, 1997 and1999.

Measurements and data analysis

Herbage was harvested with a Haldrup forageharvester from an area of 6 m�1.5 m, leaving astubble of 4–5 cm. Yields were recorded and a sample�approximately 300 g fresh weight� was taken foranalyses of DM, total N �oxidation at 1050 °C� and P�destruction with Fleischmann acid�. On grass/cloverplots, an additional sample was taken for manualseparation of grass and white clover. The grass frac-tion also contained small proportions of not sownspecies like annual meadowgrass �Poa annua L.�,smooth stalked meadowgrass �Poa pratensis L.� anddandelion �Taraxacum offıcinale Web. s.l.�. Grass andwhite clover fractions were analysed for DM, andfrom 1996 onwards, also for total N and P.

The apparent N and P recoveries were calculatedas: ��N or P yield of fertilised plot – N or P yield ofunfertilised plot� / fertiliser N or P application�. Theapparent N and P efficiencies were calculated as:��DM yield of fertilised plot – DM yield of unfertil-

ised plot� / fertiliser N or P application�. The N and Puse efficiencies were calculated as the DM yield perkg of N or P uptake. Apparent biological N fixationby white clover was calculated as the difference be-tween the N yield of grass/clover plots and grass plotswith a similar N application. The apparent N transferfrom white clover to perennial ryegrass was calcu-lated as the difference in N yield of the grass compo-nent in the mixture and the unfertilised grass plots.

Data were analysed with standard procedures foranalyses of variance and multiple regression analyses,using GENSTAT 5 �GENSTAT 1998�.

Results

Dry matter yield

Each year, the annual DM yield was significantly af-fected by sward type and N level �Table 3�. On thegrass-only swards, the average N efficiency was 28and 22 kg DM kg N–1 after application of 190 or 380kg N ha–1 year–1, respectively. The N-efficiency wasconsiderably lower in the establishment year than inthe following four years. The effect of N applicationon the grass/clover plots was only 4 kg DM kg N–1

on average, but varied from 13 kg DM kg N–1 in theestablishment year to � 4 kg DM kg N–1 in 1995.

Table 3. Annual DM yield �t ha–1� in relation to sward type, N level and P level.

Sward type N level 1994 1995 1996 1997 1998

Grass N0 6.82 5.63 3.88 5.28 10.04N1 10.93 11.56 9.28 10.78 15.46N2 12.48 16.14 12.89 14.61 17.34

Grass/clover N0 8.59 13.97 10.83 12.84 14.11N1 11.18 13.12 10.95 13.31 15.57

Mean 9.98 12.08 9.57 11.36 14.50

Significance/ LSD �P � 0.05�Sward type * 0.16 *** 0.29 *** 0.28 *** 0.35 *** 0.27Sward type�N level *** 0.26 *** 0.44 *** 0.44 *** 0.54 *** 0.42

P level 1994 1995 1996 1997 1998

P0 9.61 11.77 8.91 10.99 13.90P1 9.91 11.91 9.45 11.44 14.46P2 10.03 12.19 9.84 11.35 14.70P3 10.44 12.46 10.06 11.68 14.94

Significance *** ** *** NS ***LSD �P � 0.05� 0.23 0.40 0.39 0.48 0.37

NS � not significant; * P � 0.05; ** P � 0.01; *** P � 0.001.

168

Phosphorus application had a significant positiveeffect on the DM yield in four out of five years �Table3�. The average apparent P efficiency after applicationof 34, 68 or 100 kg P ha–1 year–1 was 13.0, 9.0, and9.1 kg DM kg P–1, respectively.

The average response of DM yield to P applicationincreased considerably with increasing N application�Figure 1, quadrant II�. The average P efficiency was1.5, 11.3 and 21.3 kg DM �kg P�–1 after applicationof 0, 190 and 380 kg N ha–1 year–1, respectively. Onthe other hand, the average N efficiency increasedfrom 23 to 26 kg DM �kg N�–1 from the lowest to thehighest P application. However, this interactionbetween N and P application was never statisticallysignificant.

White clover

Nitrogen application had a consistent significantnegative effect on white clover content � Table 4�. Theoverall mean white clover content was reduced from41 to 16% following N application. On average, ap-plication of 190 kg N ha–1 year–1 increased the grass

DM yield with 3.64 t ha–1 year–1, but decreased thewhite clover DM yield with 3.06 t ha–1 year–1.

Phosphorus application had no significant effect oneither grass DM yield, white clover DM yield orwhite clover content. In 1995, the white clover con-tent and white clover DM yield were significantlylower with increasing P application, but only in com-bination with N application.

The average annual apparent N fixation was 176 kgha–1, but ranged from 67 to 253 kg ha–1. Applicationof N reduced the average apparent N fixation from257 to 95 kg ha–1 year–1. Without N application, Papplication tended to increase apparent N fixation,whereas with N application, P application tended toreduce apparent N fixation. The average apparent Ntransfer �1996–1998� was 63 kg N kg ha–1 year–1

without N application and nil with N application.

Nitrogen yield and N use effıciency

Each year, the annual N yield was significantlyaffected by sward type and N level �Table 5�. On thegrass-only swards the average N recovery was 0.75

Figure 1. Relationships between annual DM yield, P yield, apparent P recovery �APR� and P application for grass-only �G� and grass/clover�GC�; mean of five years.

169

or 0.81 kg kg–1, after application of 190 or 380 kg Nha–1 year–1, respectively. In the establishment year, Napplication increased the N yield of the grass/cloverplots, resulting in an N recovery of 0.53 kg kg–1. Inthe next three years, N application had a significantnegative effect on the N yield of the grass/cloverplots, while in the fifth year the N yield was not af-fected by N application.

Phosphorus application had a positive effect on theN yield, although this was only significant in the firsttwo years �Table 5�. The average response of N yield

to P application increased with increasing N applica-tion, but this interaction was only statistically signif-icant in the establishment year.

On the grass-only plots, the N use efficiencydecreased with increasing N application, on averagefrom 48 to 32 kg DM kg N–1. On the grass/cloverplots, the N use efficiency was 31 kg DM kg N–1

without N application and 35 kg DM kg N–1 with Napplication. The N use efficiency was not affected byP application.

Table 4. Annual white clover content �%� of grass/clover swards in relation to N level and P level.

1994 1995 1996 1997 1998

P level Low High Mean Low High Mean Low High Mean Low High Mean Low High Mean

P0 16 7 11 60 15 37 46 22 34 53 32 42 38 20 29P1 17 6 12 61 11 36 47 18 32 55 30 43 34 14 24P2 18 7 13 62 13 37 43 17 30 48 30 39 25 17 21P3 20 8 14 63 6 34 48 11 30 52 27 39 24 15 20

Mean 18 7 12 61 11 36 46 17 32 52 30 41 30 17 23

Significance/ LSD �P � 0.05�N level *** 1.7 *** 2.6 *** 4.2 *** 5.4 *** 5.2P level NS 2.4 NS 3.7 NS 6.2 NS 7.6 NS 7.3N level � P level NS 3.4 * 3.2 NS 8.8 NS 10.8 NS 10.3

Table 5. Annual N yield �kg ha–1� in relation to sward type, N level and P level.

Sward type N level 1994 1995 1996 1997 1998

Grass N0 183 117 76 102 182N1 334 273 203 228 329N2 451 476 365 411 480

Grass/clover N0 263 453 358 445 426N1 368 345 305 391 435

Mean 320 333 262 316 370

Significance/ LSD �P � 0.05�Sward type * 6 *** 12 *** 11 *** 16 *** 12Sward type�N level *** 9 *** 18 *** 17 *** 24 *** 18

P level 1994 1995 1996 1997 1998

P0 310 321 252 301 357P1 318 329 258 316 373P2 324 337 268 321 379P3 328 344 268 325 373

Significance *** * NS NS NSLSD �P � 0.05� 8 16 15 21 16

170

Phosphorus yield and P use effıciency

The P yield increased significantly with increasing Plevel �Table 6�. The average apparent P recoveries af-ter application of 34, 68 or 100 kg P ha–1 year–1 were0.19, 0.14 and 0.12 kg kg–1, respectively. The fertil-iser P effect consistently increased during consecutiveyears. The average apparent P recovery at the lowestP level increased from 0.08 kg kg–1 in 1994 to 0.35kg kg–1 in 1998.

Nitrogen application increased the P yield signifi-cantly in all years. On the grass-only plots the aver-age P yield was 25, 43 and 50 kg ha–1 year–1 afterapplication of 0, 190 and 380 kg N ha–1 year–1, re-spectively. On the grass/clover plots, N applicationincreased the P yield only in two out of five years�Table 6�.

On the grass-only plots, N level interacted signifi-cantly with the effect of P level �Table 6 and Figure1, quadrant IV�. The response of the annual P yieldto P application increased with increasing N level.The average P recoveries at the lowest P level were0.05, 0.20 and 0.33 kg kg–1 for N0, N1 and N2, re-spectively �Figure 1, quadrant III�.

Grass/clover plots showed a response to fertiliser Pthat was different from the response on grass-onlyplots. Without P fertilisation, the average P yield ofthe grass/clover plots was equal to that of grass-N2plots �Figure 1, quadrant IV�. However, P fertilisationincreased the P yield of grass/clover plots not as muchas the P yield of the grass-only plots. The lower re-sponse to P fertilisation of the grass/clover plots canbe attributed to the clover component of the mixture.White clover P yield showed no response to fertiliserP, while the grass component showed a responsesimilar to the one in the grass-N1 plots �Figure 2,quadrant IV�.

The P use efficiency decreased with increasing Papplication, on average from 310 to 252 kg DM kgP–1. However, on the grass-N0 plots the P use effi-ciency was almost unaffected by P application, withan average value of 255 kg DM kg P–1 �Figure 1,quadrant I�.

On the grass-only plots, N application increased theP use efficiency to 269 kg DM kg P–1 �N1� and 292kg DM kg P–1 �N2�. Nitrogen application on grass/clover plots only increased the P use efficiency of thegrass component, whilst the P use efficiency of theclover component remained unchanged.

Tabl

e6.

Ann

ual

Pyi

eld

�kg

ha–

1�

inre

latio

nto

swar

dty

pe,

Nle

vel

and

Ple

vel.

1994

1995

1996

1997

1998

Swar

dty

peN

leve

lP0

P1P2

P3M

ean

P0P1

P2P3

Mea

nP0

P1P2

P3M

ean

P0P1

P2P3

Mea

nP0

P1P2

P3M

ean

Gra

ssN

023

2524

2524

2225

2427

2414

1414

1414

2021

2122

2138

4041

4441

N1

3336

3840

3739

4448

4945

2530

3437

3234

4042

4741

4659

6970

61N

236

4041

4340

4957

6166

5831

3845

4841

3752

5557

5045

6269

7663

Gra

ss/c

love

rN

029

3031

3531

4851

5457

5234

4042

4641

4449

5155

5046

5461

5955

N1

3537

3840

3848

4952

5451

3138

4242

3841

5154

5550

4658

6371

60

Mea

n31

3434

3734

4145

4850

4627

3235

3833

3543

4547

4244

5461

6456

Sign

ifica

nce/

LSD

�P�

0.05

�P

leve

l**

*0.

9**

*1.

5**

*2.

0**

*2.

5**

*2.

1Sw

ard

type

*0.

6**

*1.

1**

*1.

4**

*1.

8**

1.5

Swar

dty

pe�

Ple

vel

NS

1.4

NS

2.3

NS

3.2

NS

4.0

NS

3.3

Swar

dty

pe�

Nle

vel

***

1.0

***

1.7

***

2.2

***

2.8

***

2.3

Swar

dty

pe�

Nle

vel�

Ple

vel

**2.

0**

1.4

***

4.5

**5.

6**

*4.

6

171

Soil phosphorus

The P-AL value and total soil P showed a positive re-sponse to P application, and a negative response to Napplication. The positive effect of P application onsoil P contents decreased with increasing N applica-tion. The responses of the P-AL value and total soil Pdecreased with increasing depth. Phosphorus applica-tion had the largest effect on the P-AL value, evendown to a depth of 23–28 cm �Figure 3�. Within thegrass/clover plots, N application had no effect on theP-AL value.

The development of total soil P was in line withthe development of the P-AL value, but was more re-stricted to the upper soil layers �Table 7�. A good re-lationship �R2 � 97.7%� was found between P-ALvalue and total soil P, namely: P-AL � � 1.65 �0.3669��total P� � 0.011017 � �total P�2.

Soil nitrogen and carbon

During the experiment, total soil N in the top layer�0–5 cm� increased from 151 to 245 mg 100 g–1 dry

soil. The N accumulation in the top soil of the grass-only plots was significantly increased by N applica-tion �Table 8�, but was not affected by P application.The soil N content of the grass/clover plots was simi-lar to that of the grass-only plots with fertiliser N. Inthe deeper soil layers the N content was not affectedby N application, P application or sward type.

Similar to soil N, organic C in the top layer �0–5cm� increased during the experiment, from 1857 to2581 mg 100 g–1 dry soil. The C accumulation in thetop soil of the grass-only plots was not affected by Nor P application, but was significantly higher in thegrass-only plots than in the grass/clover plots �Table8�. In the deeper soil layers, the C content was notaffected by N application, P application or swardtype.

As the C content increased less than the N content,the C/N ratio decreased during the experiment. In thetop soil, the average C/N ratio decreased from 12.3to 10.5. At the end of the experiment, the C/N ratiowas significantly lower in the grass/clover plots thanin the grass-only plots. Within the grass-only plots,

Figure 2. Relationships between annual DM yield, P yield and P application, for grass �G-GC� or clover �C-GC� from grass/clover plots, andgrass-only �G� as a reference; mean of three years �1996–1998�.

172

Fig

ure

3.D

evel

opm

ent

ofP-

AL

valu

e�m

gP 2

O5

100

g–1

dry

soil �

inre

latio

nto

Pap

plic

atio

n,fo

ral

lco

mbi

natio

nsof

swar

dty

pean

dN

appl

icat

ion,

and

for

all

soil

laye

rs.

Top–

dow

n:0–

5,5–

10,

10–2

3an

d23

–28

cm;

left

–rig

ht:

gras

sN

0,N

1,N

2...

gras

s/cl

over

N0,

N1.

173

the C/N ratio decreased significantly with increasingN application.

Phosphorus surplus

The average soil surface surplus �fertiliser input mi-nus removed herbage� for P was � 35, � 8, �23 or�53 kg ha–1 year–1 after a fertiliser P application of1, 34, 68 or 100 kg P ha–1 year–1, respectively. Al-though the soil surface surplus varied between years,this positive effect of application rate on the soil sur-face surplus was consistent throughout the experi-ment �Table 9�. The soil surface surplus for total Pwas also affected by N application rate. Increasing Nrates increased the P uptake and hence decreased thesoil surface surplus.

Each year, the accumulation of total soil P, mea-sured to a depth of 28 cm, increased with increasingP application, with large variations between years�Table 9�. The net difference between soil accumula-tion and the soil surface surplus, which varied from� 75 to �138 kg P ha–1 year–1, was not affected byany of the experimental treatments. Negative valuesindicate an unaccounted loss of P from the studiedsoil–crop system, and positive values indicate an un-accounted input. Considering the whole five-year ex-perimental period, an unaccounted accumulation of138 kg P ha–1 occurred. After exclusion of the estab-lishment year �1994�, the average soil surface surplusand soil accumulation matched better.

Regression analysis showed that the soil surfacesurplus had a significant effect on the changes in theamount of soil P in the 0–5 and 5–10 cm soil layers,whereas changes in the deeper soil layers could notbe related to the soil surface surplus. Changes in theP-AL value were affected to a depth of 10 cm by thesoil surface surplus. The following relationships, withstandard errors between brackets, were found for the0–5 cm and 5–10 cm soil layers:

��P-AL value�0–5 cm=

6.43�0.345� � 0.179�0.0114� × Psoil surface surplus,

R2 � 66.1%

��P-AL value�5–10 cm=

1.75�0.116� � 0.0581�0.00548� × Psoil surface surplus,

R2 � 44.7%

Tabl

e7.

Tota

lP

�mg

P10

0g–

1dr

yso

il �at

the

end

ofth

eex

peri

men

tin

rela

tion

tosw

ard

type

,N

leve

lan

dP

leve

l,m

easu

red

info

urso

illa

yers

.

0–5

cm5–

10cm

10–2

3cm

23–2

8cm

Swar

dty

peN le

vel

P0P1

P2P3

Mea

nP0

P1P2

P3M

ean

P0P1

P2P3

Mea

nP0

P1P2

P3M

ean

Gra

ssN

066

8497

115

9057

6366

7565

6063

6565

6355

5555

5655

N1

6175

9110

583

5559

6466

6156

5962

5959

5355

5555

54N

259

7284

100

7956

6061

6661

5862

6464

6255

5658

5857

Gra

ss/c

love

rN

062

7385

100

8055

6169

6962

5862

6162

6158

5656

5556

N1

6071

8510

480

5658

6868

6158

6262

6562

5456

5759

56

Mea

n61

7588

105

8256

6063

6962

5862

6363

6155

5656

5756

Sign

ifica

nce/

LSD

�P�

0.05

�P

leve

l**

*1.

5**

*1.

4**

*2.

3N

S1.

6Sw

ard

type

***

1.1

NS

1.0

NS

1.6

NS

1.1

Swar

dty

pe�

Ple

vel

*2.

3N

S2.

3N

S3.

6N

S2.

5Sw

ard

type

�N

leve

l**

*1.

6**

*1.

6*

2.5

*1.

7Sw

ard

type

�N

leve

l�P

leve

l**

3.3

*3.

2N

S5.

1N

S3.

5

174

Nitrogen surplus

The average soil surface surplus for N was � 132,� 84 or � 57 kg ha–1 year–1 on grass-only plots af-ter a fertiliser N application of 0, 190 or 380 kg ha–1

year–1, respectively. The soil surface surplus of thegrass/clover plots is exactly equal to those of thegrass-only plots with corresponding N levels, due tothe method used to calculate the biological apparentN fixation. Similar to the findings with P, the soil sur-face surplus for N showed some variations betweenyears � Table 10�. The soil surface surplus for total Nwas also affected by P application rate. Increasing Prates increased the N uptake and hence decreased thesoil surface surplus.

There was no consistent relationship between Napplication rate or sward type and the accumulationof total soil N, measured to a depth of 28 cm �Table

10�. The fluctuation between years followed the samepattern as that of the accumulation of soil P.

The net difference between soil surface surplus andsoil accumulation showed a wide variation, but wasnot affected by any of the experimental treatments.Considering the whole five-year experimental period,an unaccounted accumulation of 681 kg N ha–1 oc-curred. Exclusion of the establishment year �1994�resulted in an unaccounted loss of 190 kg N ha–1.

Discussion

Nitrogen response

Nitrogen was the driving factor with respect to theperformance of the grass-only swards. The effect ofnitrogen on the performance of perennial ryegrass is

Table 8. Organic C and total N �mg 100 g–1 dry soil� at the end of the experiment in relation to sward type and N level, measured in four soillayers.

Organic C Total N

Sward type N level 0–5 cm 5–10 cm 10–23 cm 23–28 cm 0–5 cm 5–10 cm 10–23 cm 23–28 cm

Grass N0 2573 1907 2144 1960 235 178 202 173N1 2682 1935 1967 1888 247 176 183 169N2 2625 1937 2107 2018 253 186 203 185

Grass/clover N0 2473 1936 2030 1940 244 183 193 177N1 2554 1889 2085 1975 248 184 201 174

Mean 2581 1921 2067 1956 245 181 196 176Significance/ LSD �P � 0.05�Sward type ** 65 NS 54 NS 92 NS 80 NS 4.4 NS 5.6 NS 9.0 NS 7.5Sward type�N level NS 101 NS 83 NS 142 NS 123 *** 6.9 NS 8.7 NS 13.9 NS 11.7

Table 9. Soil surface surplus �fertiliser input – herbage removal� and soil accumulation of P �kg ha–1 year–1� in relation to P level, measuredto a depth of 28 cm.

Soil surface surplus Change in soil P �0–28 cm�

Year P0 P1 P2 P3 Mean P0 P1 P2 P3 Mean

1994 � 27 3 35 65 19 79 176 160 213 1571995 � 41 � 2 44 83 21 � 105 � 108 � 29 24 � 541996 � 27 0 30 60 16 79 78 165 158 1191997 � 35 � 16 9 32 � 3 � 49 � 29 � 79 � 28 � 461998 � 44 � 24 � 1 25 � 11 � 56 8 34 33 5

Total � 174 � 40 116 266 42 � 52 126 251 399 181

1995–1998 � 147 � 37 81 261 23 � 131 � 50 91 186 24

175

well documented �Vellinga and André 1999�. In TheNetherlands, the annual N requirements are based ona marginal response of 7.5 kg DM �kg N�–1 �Unwinand Vellinga 1994�. The marginal N response of thegrass-only treatments was derived from a simple re-gression of DM yield on N level: Y � � 0.02874X2

� 33.0X � 6348 �R2�94.2%�, where Y � DM yield�kg ha–1 year–1� and X � N application �kg ha–1

year–1�. From this response curve it was calculatedthat the marginal N response at the highest N appli-cation of 380 kg ha–1 year–1 was 11.2 kg DM �kgN�–1. This means that the optimal N application, ac-cording to the current recommendations, for this ex-periment was higher than 380 kg ha–1 year–1.Therefore it has to be realised that the range of Nrates studied in this experiment is possibly lower thanthe currently applied N rates in practical farming onthis soil type �Reijneveld et al. 2000�.

The N recovery and especially the N efficiency ofthe grass-only plots was markedly lower in the estab-lishment year than in the other years. Although acomparison with an unploughed sward in the sameyear is not present, it may be hypothesised that min-eralisation from the old stubble and roots increasedthe soil N supply �Davies et al. 2001�.

On the grass/clover plots, N application led to anincreased grass production at the cost of white clo-ver, as reported earlier by Frame and Newbould�1986�. Nitrogen application benefits grass more thanclover due to its greater competitive ability �Dunlopand Hart 1987�, and the faster and greater N accumu-lation �Boller and Nösberger 1988�. In relation to thetarget marginal N response of 7.5 kg DM �kg N�–1, Napplication was only advisable in the establishment

year. In that year the increased grass DM yield was3.10 t ha–1 year–1, while the clover DM yield wasonly reduced by 0.74 t ha–1 year–1. In the followingfour years, the increased grass DM yield was nearlycompletely offset by the decreased clover DM yield,resulting in an average N efficiency of only 1.6 kgDM �kg N�–1.

Phosphorus response

The P response, measured in this experiment, is ajoint effect of fertiliser and soil P. During the experi-mental years the P-AL value increased with increas-ing fertiliser P level. Therefore, the calculatedapparent P efficiencies and P recoveries were increas-ingly overestimated as the experiment progressed.The highest apparent P recovery observed in the fifthyear of the experiment was 0.56 kg kg–1 in the grass-N2-P1 treatment, indicating that a substantial propor-tion of the applied P was still accumulating in thesoil–root–stubble system.

The observed P response of the grass/clover swardswas different from that of the grass-only swards. Thegrass component of the mixed swards showed a posi-tive response of DM yield, P concentration and Pyield to P application, similar to the responsesobserved in the grass-only swards. In contrast, theclover component of the mixed sward showed no re-sponse in P yield. However, a lower P use efficiencywith increased P application meant that the DM yielddecreased and the P concentration increased. Thiscorresponds with the view that white clover is gener-ally a weaker competitor for nutrients when grown inassociation with grass, due to its thicker and less

Table 10. Soil surface surplus �fertiliser input – herbage removal� and soil accumulation of N �kg ha–1 year–1� in relation to sward type andN level, measured to a depth of 28 cm.

Soil surface surplus Change in soil N �0–28 cm�

Grass Gr/Cl Grass Gr/Cl

Year N0 N1 N2 N0 N1 Mean N0 N1 N2 N0 N1 Mean

1994 � 183 � 135 � 51 � 183 � 135 � 137 426 639 709 845 1226 7681995 � 117 � 56 � 36 � 117 � 56 � 76 � 184 � 436 � 187 � 450 � 570 � 3641996 � 76 � 22 � 5 � 76 � 22 � 40 565 787 706 852 231 6281997 � 102 � 57 � 72 � 102 � 57 � 78 106 � 106 � 240 � 133 416 91998 � 182 � 148 � 120 � 182 � 148 � 156 � 215 � 433 � 35 � 430 � 485 � 320

Total � 660 � 418 � 284 � 660 � 418 � 488 698 451 953 684 818 721

1995–1998 � 477 � 283 � 233 � 477 � 283 � 351 272 � 188 244 � 161 � 408 � 47

176

branched root system �Dunlop and Hart 1987�. On theother hand, others reported a stronger P response forwhite clover than for perennial ryegrass �Sinclair etal. 1996�. The observed P concentrations measured inthe white clover component on the P0-plots werelower than the lower end of the range of reportedcritical values of 3.0 to 4.0 g kg–1 DM �Dunlop andHart 1987; Morton et al. 1998; Whitehead 2000�. Themore vigorous growth of grass, especially on the plotsreceiving fertiliser-N, may have enhanced the lowerP uptake by clover. Taking the optimal ratio betweenP and N concentration of 0.074 �Morton et al. 1998�as an indicator for balanced nutrient supply, no sub-optimal P/N-ratios were observed in white clover. Theminimum P/N-ratio observed in white clover variedfrom 0.071 to 0.081.

Nitrogen×phosphorus response

Interactions between N and P application have onlybeen observed to a limited extent. Especially if apractical range of applications is considered �approx-imately 150 to 400 kg N ha–1 year–1 and 40 to 100 kgP ha–1 year–1� the implications for management prac-tices are relatively small. This can be illustrated bythe calculation of the marginal N efficiencies at dif-ferent P application levels. Regression of DM yield,restricted to grass-only plots, on N and P applicationresulted in the following model: Z � � 0.02944X2

� 31.85X � 3.66Y � 0.0272XY � 6159�R2�95.1%�, where Z � DM yield �kg ha–1 year–1�,X � N application �kg ha–1 year–1� and Y � P appli-cation �kg ha–1 year–1�. With a P application of 40 and100 kg ha–1 year–1, a marginal N efficiency of 12.5kg DM �kg N�–1 is attained at an N application of 347and 374 kg kg ha–1 year–1, respectively.

The negative effect of P application on white clo-ver yield appeared to increase with N application.This suggests that the competitive ability of grassesintensifies after N application, which was also notedin several other studies �Dunlop and Hart 1987�. Fur-thermore, an overview of Harris �1987� indicated thatthe competitive effect of grasses for P can also be de-veloped by an increased availability of fixed N. Theobservations on the grass/clover-N0 plots are inagreement with that hypothesis. The response of thewhite clover content from the lowest to the highest Papplication was �4, �3, �2, � 1 and � 14% in theconsecutive years.

Soil phosphorus

The P-AL-value of the soil showed a large responseto the experimental treatments. At the end of the ex-periment the P-AL-values in the top soil layer �0–5cm� ranged from 13 to 97. At the low end of therange, the P-AL-values, in all soil layers, seemed tostabilise around a level of 10.

As expected, the responses of the P-AL value andtotal soil P were related to the soil surface surplus forP, which in turn was determined by N and P applica-tion rates and their interaction. Although a significanteffect of sward type on P-AL-value and total P wasmeasured, it is more likely that this is an indirect ef-fect related to the level of N, either through fixationor fertiliser application. This hypothesis is supportedby the fact that sward type had no significant effecton the relationship between soil surface surplus andchange in either P-AL-value or total P.

The changes in soil P values were closely relatedto the soil surface surplus for P. A general concern offarmers, that balanced P application will lead to a re-duction of P-AL-values, was not justified by the re-sults of this experiment. Even with negative surplus-ses, the P-AL value was not reduced. Similar resultswere obtained for the relationship between changes intotal soil P and the soil surface surplus, indicating anenrichment of the topsoil. Upward P transport occurseither through capillary rise or through root uptake indeeper soil layers �De Willigen and Van Noordwijk1987�. On the other hand, the observed enrichmentcould be a measurement inaccuracy that falls withinthe margin of error. The observed enrichment of 28kg P ha–1 year–1, in the soil layer of 0–28 cm, is only1% of the total amount of P found in that layer. Fur-thermore, the observed enrichment occurred mainlyin the establishment year. This is unexpected, as theamount of P taken up for the development of rootsand stubbles is not accounted for in the totalsoil–plant balance. Assuming a total root and stubbleweight of 6 to 8 t DM ha–1 �Sibma and Ennik 1988�and a P content between 0.2 and 0.5% �De Willigenand Van Noordwijk 1987; Whitehead 2000�, the totalamount of P in roots and stubbles would be between12 and 40 kg ha–1.

The present experiment only studied the effect offertiliser inputs, whilst in farming practice P ongrassland will be almost completely applied throughshallowly injected cattle slurry. Furthermore, a con-siderable amount of P is not removed by cutting, butheterogeneously returned to the sward by the faecal

177

excreta of the grazing animals. Preliminary data ofcurrently ongoing experiments �Van Middelkoop etal. 2004�, in which the aspects of cattle slurry andgrazing are accounted for, indicate a slightly higherrequired soil surface surplus to maintain a constantP-AL-value.

Soil nitrogen and carbon

The accumulation of total soil N was similar underboth sward types. After five years, only the unfertil-ised grass-only plot had a significantly lower soil Ncontent in the layer of 0–5 cm. Similar to the findingswith P, an unaccounted accumulation of soil Noccurred.

Including the input of atmospheric deposition of 35kg N ha–1 year–1 �IKC 1993� leaves an unaccountedN accumulation of 101 kg ha–1 year–1, which is only1.5% of the total amount of soil N in the layer of 0–28cm.

The difference method for estimating the biologi-cal N fixation is based on the assumption that grass/clover and grass-only swards take up the sameamount of soil N. Considering the present experi-ment, this assumption only holds for the swards re-ceiving no fertiliser N, as it is likely that for bothsward types almost all available soil N will be takenup �Whitehead 1995�. Furthermore, the differencemethod does not take into account differences in ac-cumulation of N in the soil, roots and stubbles. As wefound a significantly lower N accumulation in the un-fertilised grass-only plots, it is likely that the appar-ent N fixation underestimated the real N fixation.

The accumulation of organic C, in the top soil layerof 5 cm, was significantly lower on the grass/cloverswards than on the grass-only swards. At comparablelevels of aboveground DM production, 1030 kg Cha–1 year–1 accumulated in the layer of 0–5 cm of thegrass-only-N1 sward, compared to 761 kg C ha–1

year–1 for the grass/clover-N0 sward. It may be as-sumed that the lower C/N-ratio under grass/clover in-creases the N mineralisation. This is confirmed byearlier work of Elgersma and Hassink �1997�, inwhich the potential N mineralisation under unfertil-ised grass/clover swards was higher than underunfertilised grass-only swards. Alvarez et al. �1998�suggested that the higher N mineralisation underwhite clover could be attributed to the high organic Ncontents and low C/N-ratio of the organic inputs.

Conclusions

Nitrogen was the main factor determining the yieldand nutrient content of the harvested herbage. A posi-tive interaction between N and P application on herb-age P yield was observed. In the mixed sward, Papplication increased the grass yield, but did not in-crease the white clover yield.

There were no significant differences in accumula-tion of soil N and P under grass-only and grass/clo-ver swards. Due to a lower C accumulation, the C/N-ratio was lower under grass/clover swards. Changesin total soil P and P-AL value were closely related tothe soil surface surplus, which in turn was determinedby the level of N and P application and its interac-tion.

Acknowledgements

We thank the staff at the Waiboerhoeve for their tech-nical assistance in the fieldwork. Furthermore wewould like to thank Professor Dr. P.C. Struik, Dr. A.Elgersma, Dr. B.H. Janssen, P.A.I. Ehlert and theanonymous referees for their suggestions.

References

Aarts H.F.M., Biewinga E.E. and Van Keulen H. 1992. Dairyfarming systems based on efficient nutrient management. Neth.J. Agric. Sci. 40: 285–299.

Agterberg G.C. and Henkens P.L.C.M. 1995. Basis for phosphatefertilizer recommendations on grassland �in Dutch�. Meststoffen1995: 12–23.

Alvarez G., Chaussod R., Loiseau P. and Delpy R. 1998. Soil in-dicators of C and N transformations under pure and mixedgrass/clover swards. Eur. J. Agron. 9: 157–172.

Boller B.C. and Nösberger J. 1988. Influence of dissimilarities intemporal and spatial N-uptake patterns on 15N-based estimatesof fixation and transfer of N in ryegrass–clover mixtures. PlantSoil. 112: 167–175.

Davies M.G., Smith K.A. and Vinten A.J.A. 2001. The mineralisa-tion and fate of nitrogen following ploughing of grass and grass-clover swards. Biol. Fert. Soils 33: 423–434.

De Willigen P. and Van Noordwijk M. 1987. Roots, plant produc-tion and nutrient use efficiency. Ph.D. Thesis, Agricultural Uni-versity Wageningen, the Netherlands, 282 pp.

Dunlop J. and Hart A.L. 1987. Mineral nutrition. In: Baker M.J.and Williams W.M. �eds�, White Clover. CAB International,Wallingford, UK, pp. 153–183.

Egnér H., Riehm H. and Domingo W.R. 1960. Untersuchungenüber die chemische Bodenanalyse als Grundlage für die Beurtei-lung des Nährstoffzustandes der Boden II. Chemische Extrac-

178

tionsmethoden zur Phosphor- und Kaliumbestimmung. KunligaLandbrukshögskolans Annaler 26: 199–215.

Elgersma A. and Hassink J. 1997. Effects of white clover �Trifo-lium repens L.� on plant soil nitrogen and soil organic matter inmixtures with perennial ryegrass �Lolium perenne L.�. Plant Soil197: 177–186.

Frame J. and Newbould P. 1986. Agronomy of white clover. Adv.Agron. 40: 1–88.

Fraters B., Vissenberg H.A., Boumans L.J.M., De Haan T. and DeHoop D.W. 1997. Results of the monitoring programme: qualityof upper ground water on farms in the sandy areas of the Neth-erlands 1992–1995 �in Dutch�. RIVM rapport 714801014,RIVM, Bilthoven, the Netherlands, 181 pp.

GENSTAT 1998. GENSTAT 5 Release 4.1. Lawes AgriculturalTrust, Rothamsted Experimental Station, Harpenden, UK.

Harris W. 1987. Population dynamics and competition. In: BakerM.J. and Williams W.M. �eds�, White Clover. CAB International,Wallingford, UK, pp. 203–297.

Hooghart J.C. and Lablans W.N. 1988. From Penman to Makkink:a new calculation method for climatological evaporation data �inDutch�. Technische rapporten TR-111, Koninklijk NederlandsMeteorologisch Instituut, De Bilt, the Netherlands.

Houba V.J.G., Van der Lee J.J. and Novozamsky I. 1997. Soilanalysis procedures: other procedures. Department of Soil Sci-ence and Plant Nutrition, Landbouwuniversiteit Wageningen,Wageningen, the Netherlands.

IKC 1993. Handbook Dairy Husbandry �in Dutch�. Publication No.35, Information and Knowledge Centre �IKC�, Lelystad, theNetherlands, 629 pp.

KNMI 1993–1999. Monthly Weather Surveys of the Netherlands�in Dutch�. Koninklijk Nederlands Meteorologisch Instituut, DeBilt, the Netherlands.

MANMF 1997. Policy document on manure and ammonia. Minis-try of Agriculture, Nature Management and Fisheries, TheHague, the Netherlands, 38 pp.

Morton J.D., Smith L.C. and Morrison J.D. 1998. Clover nutrientratios to monitor pasture yield responses to fertiliser nutrients.In: Baker M. �ed.�, Nelson, A Pastoral Palette. Proceedings ofthe New Zealand Grassland Association. New Zealand GrasslandAssociation, Palmerston North, New Zealand, pp. 141–148.

Oenema O. and Van Dijk T.A. 1994. Phosphate Losses and Phos-phate Surpluses in Dutch Agriculture �in Dutch�. Report of thetechnical project groep P-deskstudy. Ministery of Agriculture,Nature Management and Fisheries, The Hague, the Netherlands,102 pp.

Oenema O., Boers P.C.M., Van Eerdt M.M., Fraters B., Van DerMeer H.G., Roest C.W.J., Schröder J.J. and Willems W.J. 1998.Leaching of nitrate from agriculture to groundwater: the effectof policies and measures in the Netherlands. Environ. Pollut.102�S1�: 471–478.

Oenema O. and Roest C.W.J. 1998. Nitrogen and phosphoruslosses from agriculture into surface waters; the effects of poli-

cies and measures in the Netherlands. Wat. Sci. Tech. 37�2�: 19–30.

PR 1998. Fertiliser recommendations for grassland and foddercrops �in Dutch�. Praktijkonderzoek Rundvee, Schapen enPaarden, Lelystad, the Netherlands, 53 pp.

Reijneveld J.A., Habekotté B., Aarts H.F.M. and Oenema J. 2000.Typical Dutch �in Dutch�. Praktijkonderzoek Veehouderij,Lelystad, the Netherlands. 83 p.

Schils R.L.M., Boxem T., Sikkema K. and André G. 2000a. Theperformance of a white clover-based dairy system in compari-son with a grass/fertiliser-N system. I. Botanical compositionand sward utilisation. Neth. J. Agric. Sci. 48: 291–303.

Schils R.L.M., Boxem T., Jagtenberg C.J. and Verboon M.C.2000b. The performance of a white clover-based dairy system incomparison with a grass/fertiliser-N system. II. Animal perform-ance, economics and environment. Neth. J. Agric. Sci. 48: 305–318.

Sibma L. and Ennik G.C. 1988. Development and growth of pro-duction grass under Dutch conditions �in Dutch�. Pudoc,Wageningen, the Netherlands, 53 pp.

Sinclair A.G., Smith L.C., Morrison J.D. and Dodds K.G. 1996.Effects and interactions of phosphorus and sulphur on a mownwhite clover/ryegrass sward 1. Herbage dry matter productionand balanced nutrition. New Zealand J. Agric. Res. 39: 412–433.

Tunney H., Breeuwsma A., Withers P. and Ehlert P.A.I. 1997.Phosphorus fertiliser strategies: present and futures. In: TunneyH., Carton O.T. and Brookes P.C. �eds�, Phosphorus Loss to Wa-ter from Agriculture. CAB International, Wallingford, UK, pp.177–203.

Unwin R.J. and Vellinga Th.V. 1994. Fertilizer recommendationsfor intensively managed grassland. In: ‘t Mannetje L. and FrameJ. �eds�, Grassland and Society. Proceedings of the 15th GeneralMeeting of the European Grassland Federation. WageningenPers, Wageningen, the Netherlands, pp. 599–602.

Van Middelkoop J.A., Schils R.L.M., Den Boer D.J., SchoumansO.F., Van der Salm C., Bakker R.F. and Chardon W.J. 2004. Ef-fecten van stikstof- en fosfaatverliesnormen op grasland �inDutch�. PV-rapport Research Station for Animal Husbandry,Lelystad, the Netherlands �in press�.

Vellinga Th.V. and André G. 1999. Sixty years of Dutch nitrogenfertiliser experiments, an overview of the effects of soil type,fertiliser input, management and developments in time. Neth. J.Agric. Sci. 47: 215–241.

VROM 1989. National Environmental Policy Plan �in Dutch�.Ministerie van Volkshuisvesting, Ruimtelijke Ordening en Mi-lieu. Tweede Kamer, vergaderjaar 1989–1990, 21137, nrs. 1–2.VROM, The Hague, the Netherlands, 258 pp.

Whitehead D.C. 1995. Grassland nitrogen. CAB International,Wallingford, UK, 397 pp.

Whitehead D.C. 2000. Nutrient elements in grassland: Soil-Plant-Animal relationships. CAB International, Wallingford, UK, 384pp.

179