REGULAR ARTICLE

Effects of nitrogen and sulphur gradients on plantcompetition, N and S use efficiencies and species abundancein a grassland plant mixture

T. Tallec & S. Diquélou & C. Fauveau &

M. P. Bataillé & A. Ourry

Received: 1 February 2008 /Accepted: 17 June 2008 / Published online: 2 August 2008# Springer Science + Business Media B.V. 2008

Abstract Sulphur (S) depletion of grassland soils hasoccurred in Europe for many decades. This is knownto promote a decrease in ecosystem productivity andis suspected to alter plant community structure.Considering the strong links between nitrogen (N)and S metabolism in plants, these effects shoulddepend on N availability. We tested this hypothesis ina pot experiment, considering a four grassland speciesplant mixture (three Poaceæs: Lolium perenne, Agro-stis capillaris and Poa pratensis and one Fabaceæ:Trifolium repens), and submitted it to a double N andS gradient. We used labelled 15N-fertilizer and 34S-fertilizer in order to determine both nutrient useefficiencies by each species and to analyze theinfluence of competition for these nutrients on plantmixture dynamics. We compared species relativephysiological performance (RPP) in the monocultureand their relative ecological performance (REP) in the

mixture of the four species. We analysed gradienteffects at establishment and at regrowth after cutting.At establishment, grass production and S useefficiency increased along the N gradient. The Sgradient slightly favoured the dominance of L.perenne, increased A. capillaris production andenhanced N use efficiency of both species. Atregrowth, increased S promoted more significanteffects, enhancing T. repens performance in increas-ing its N2 fixation ability and maintaining this at highN. It also induced a change in grass species relativeperformance (dry matter production and N useefficiency) at high N, enhancing that of L. perenneand decreasing that of A. capillaris. At both estab-lishment and regrowth, RPP did not reflect REP,meaning that species behave differently along thegradient when grown in mixture. Finally, the Sgradient and the N gradient modulated relative plantspecies abundance. It appears that modulation of Savailability could be used as a tool to drive grasslandcommunity structure.

Keywords Competition . Nitrogen/sulphuravailability . Plant mixture . Nutrients use efficiency

Introduction

Sulphur (S) is an important nutrient for plant growthand development (Droux 2004). During recent

Plant Soil (2008) 313:267–282DOI 10.1007/s11104-008-9699-9

Responsible Editor: Herbert Johannes Kronzucker.

T. Tallec : S. Diquélou : C. Fauveau :M. P. Bataillé :A. OurryINRA, UMR950,14000 Caen, France

T. Tallec : S. Diquélou (*) :C. Fauveau :M. P. Bataillé :A. OurryUniversité de Caen Basse-Normandie, UMR950Ecophysiologie Végétale Agronomie et nutrition N, C, S,14032 Caen cedex, Francee-mail: sylvain.diquelou@unicaen.fr

decades, increasing frequency of S deficiency hasbeen reported in northern European croplands andgrasslands, mainly due to the reduction in SO2

emissions from industrial air-borne pollution (Murphyand Boggan 1988). S deficiency reduces yields andquality of both crops, e.g. oilseed rape (McGrath andZhao 1995), wheat (Withers et al. 1995), barley (Zhaoet al. 1999), and grasslands (Murphy et al. 2002), andis reported to affect grassland plant diversity (Cullen1970; Murphy and Boggan 1988).

Grassland species are affected by defoliation(herbivory or cutting), nutrient availability andcompetition from neighbouring plants. These threefactors influence the growth, survival and reproduc-tion of individual plants and subsequently theabundance of plant populations (Harper 1977).According to Tilman (1987), competition occurswhen species share at least one limited resource.The ability of a single plant to capture nutrientresources from the soil more rapidly and/or moreefficiently than another should confer to this plant ahigher competitive ability than its neighbours.When nutrient availability changes, these differ-ences between species may change biodiversity (DeVries and Kruijne 1960; Olde Venterink et al. 2001).Numerous field studies and mesocosm competitionexperiments tested nutrient gradient effects onspecies composition or functional group diversityand on productivity of grassland or old-fields(Austin 1982; Campbell and Grime 1989; Aertsand Van der Peijl 1993; Wedin and Tilman 1993;Aerts 1999; Güsewell and Bollens 2003). Most ofthem considered nitrogen (N), phosphorus (P), orpotassium (K) gradients. Only few considered S orN!S gradients, and mainly dealt with annualryegrass/clover mixture, like Trifolium subterraneumand Lolium rigidum. They proved that morphologi-cal variations and shifts in competition processesbetween species could be induced by differentrelative and total N/S availabilities (Gilbert andRobson 1984a,b; Sinclair et al. 1996; Morton et al.1999). They showed that S supply enhances thegrowth of clover independently of N supply and thatof ryegrass under high N supply only. Walker et al.(1956) showed that S enhances higher dry matterproduction and plant N content in a grass/cloverpasture and hypothesized that these effects may bedue to an increase in N2-fixation by clovers andunderground transference from the latter to grasses.

However, as for most studies (Ceccotti and Messick1997), they did not consider species balance andfertilizer use efficiency was not determined.

As species have different S requirements (Scherer2001), we state that, like a N gradient, a S gradientmodifies the balance of matrix species and plantcommunity dynamics in grasslands. As suggested byprevious studies on annual species (Gilbert andRobson 1984a,b), we hypothesize that S effect shoulddepend on N availability, since N and S assimilationand metabolism in plants are linked (constant N: Sratio; Clarkson et al. 1999). The implied mechanismssuch as competitive abilities to capture N and S, arepoorly documented, especially in herbaceous peren-nial plant communities, subjected to defoliationregime (grazing or cutting).

Selecting four of the most common co-occurringperennial species of western Europe grasslands: threegrasses, Lolium perenne L. (Perennial Rye-grass),Poa pratensis L. (Kentucky Bluegrass), Agrostiscapillaris L. (Common Bent-grass) and a legume,Trifolium repens L. (White Clover), we built up amodel community and subjected this mixture tovarious initial N!S availabilities. Our aims were (1)to investigate short term changes in plant mixturestructure (dry matter production and relative abun-dance of species) along a N!S gradient, followingboth establishment phase, where N was potentiallyample available, and regrowth phase after cutting,where N limitation was exacerbated, (2) to establish ifthese changes result from alterations of speciesphysiological aptitude to exploit environment and/orof their competitive abilities in mixture, (3) to analysethe implication of N–S uptake abilities on theseprocesses. To reach the first two objectives, weperformed a mesocosm experiment in which N: Savailabilities were modulated at species establishmentand compared species performance in mixture to thatin monoculture. We considered three local N fertil-ization rates and crossed them with three S fertiliza-tion rates. Plants were submitted plant to twosuccessive cuts, spaced out of 6 weeks, like in localhay fields for spring and summer cuts. We assessedthe third objective by the simultaneous supply of 15Nand 34S isotope when setting up the nutrient gradient.This allowed us to determine the fate of thesenutrients in the plant mixture and to assess thealteration in the ability of each species to exploit Nand S along this gradient.

268 Plant Soil (2008) 313:267–282

Materials and methods

Experimental set up

The experiment was conducted in a greenhousebetween February and April 2006. Plants were grownfrom seed and sown 3 weeks prior to the start ofexperiment. When seedlings had their first welldeveloped leaf (=day 0), they were transplanted into2 L polyvinyl-chloride (PVC) (monoculture; pot toparea: 154 cm2, soil height 15.5 cm) or 3.5 L pots(mixture; pot top area: 254 cm2, soil height 16.5 cm).Pots were filled with a sieved (2 mm mesh),homogeneous soil (silty-clay soil: 36.7% clay,41.3% silt, 4.2% organic matter, 3.18‰ total N, 1‰total S, pH: 6.1). Monocultures consisted of eightplants per pot (520 plants m!2) and mixtures of 12plants per pot (480 plants m!2), established accordingto their relative natural abundance in Normandygrasslands (France): five plants of L. perenne (41%),three of T. repens (25%), two of P. pratensis (17%)and two of A. capillaris (17%). Their spatial patternwas assessed to ensure that each plant had neighbour-ing plants of all other species (Fig. 1).

Nine treatment combinations (three N!three S) wereapplied on day 2, each combination being replicated fourtimes (180 pots). S was applied as calcium sulphate (1%atom 34S excess) at a rate of 0 (low S), 15 (intermediateS) and 30 (high S) kg S ha!1. N was applied asammonium nitrate (5% atom 15N excess) at 0 (low N,corresponding to extensive grasslands), 50 (intermediate

N, corresponding to mean fertilization rate of grasslandsin Normandy) and 180 (high N, corresponding tofertilization rate of intensive grasslands in Normandy)kg N ha!1. A basal mix of K (150 kg K ha!1) and P(60 kg P ha!1) as K2HPO4 was applied to each pot, atthe same time. Pots receiving no or moderate calciumsulphate were supplemented with CaCl2, so that eachpot received the same amount of Ca. T. repens plantswere infected with a standard mixture of Rhizobiumtrifolii T354 known to support N fixation.

Throughout the experiment, the air temperature waskept at 20/16±2°C (day/night). Plants were grownunder natural light conditions until day 5, and thensupplemented by artificial illumination (400 W high-pressure sodium lamps, Philips SON T-PIA Agro)providing 400 !mol m!2 s!1 PAR (photosyntheticallyactive radiation) at plant height with a 16/8 h photo-period. The soil water content was kept at ±25%(relative to dry weight) by weighing the pots once aweek and by adding the required quantity of water andwatering them four times a week. The containers weremoved twice a week to minimize positional effects.

The species response to the N!S gradient wasanalyzed during the establishment phase and, follow-ing cutting, during the regrowth phase in establishedmixture and monoculture.

Measurement

Biomass was harvested by species with a cutting heightof 50 mm above the root–shoot junction on days 42and 84, allowing a period of 6 weeks for both firstgrowth and regrowth phases (almost corresponding totime interval between spring and summer cuts in localhay fields). All samples were dried (48 h at 65°C) toconstant weight, and dry mass (DM) per plant wasdetermined for each species.

The N and S analyses were made for each speciesgrown in mixture, to assess the fate of nutrients. Asindividual number per species differed, results perplant are considered in order to allow comparisonsbetween species. Plant material was finely powdered.N and S contents and 15N and 34S abundances wereestimated using an isotope ratio mass spectrometer(IRMS, Isoprime, GV Instrument). Natural 15Nabundance (0.3663%±0.0004) of atmospheric N2

was used as a reference for 15N analysis. Natural 34Sabundance (4.21%) of Canyon Diablo Troïlite mete-orite was used as a reference for 34S.

L.p.

L.p. L.p.

L.p.

L.p.

T.r.T.r.

T.r.

A.c .

A.c.

P.p.

P.p.

Fig. 1 Relative positioning of the transplants (L.p.: Loliumperenne; A. c.: Agrostis capillaris; P. p.: Poa pratensis; T. r.:Trifolium repens)

Plant Soil (2008) 313:267–282 269

Table 1 N and S effect on Lolium perenne, Agrostis capillaris, Poa pratensis and Trifolium repens dry mass in monoculture and inmixture, N and S yields, %, N and S derived from fertilizer and use efficiencies in mixture

Variable Establishment phase Regrowth phase

N S N!S N S N!S

DM in monoculture Lolium perenne 0.000 0.018 0.033 0.000 0.703 0.010Agrostis capillaris 0.000 0.885 0.872 0.000 0.693 0.000Poa pratensis 0.035 0.702 0.696 0.000 0.681 0.000Trifolium repens 0.011 0.007 nt 0.000 0.000 0.000

DM in mixture Lolium perenne 0.000 0.037 0.600 0.000 0.000 0.000Agrostis capillaris 0.000 0.291 0.036 0.000 0.197 0.530Poa pratensis 0.009 0.484 0.984 0.000 0.433 0.955Trifolium repens 0.018 0.661 0.010 0.013 0.000 nt

S yield Lolium perenne 0.004 0.000 0.124 0.988 0.000 ntAgrostis capillaris 0.001 0.033 0.209 0.110 0.317 0.213Poa pratensis 0.643 0.005 0.737 0.399 0.000 0.598Trifolium repens 0.008 0.000 0.056 0.000 0.000 0.056

S% DM Lolium perenne 0.020 0.000 0.326 0.000 0.000 0.220Agrostis capillaris 0.027 0.009 0.306 0.000 0.000 0.696Poa pratensis 0.004 0.001 0.261 0.000 0.000 0.004Trifolium repens 0.000 0.000 0.146 0.005 0.000 0.001

SDFF Lolium perenne 0.000 0.010 0.750 0.719 0.000 0.041Agrostis capillaris 0.000 0.015 0.178 0.000 0.002 0.023Poa pratensis 0.015 0.148 0.381 0.710 0.003 0.775Trifolium repens 0.450 0.422 0.316 0.162 0.000 0.015

RSUE Lolium perenne 0.000 0.016 0.581 0.512 0.238 0.992Agrostis capillaris 0.000 0.356 nt 0.087 0.168 0.641Poa pratensis 0.017 0.299 0.540 0.558 0.865 0.643Trifolium repens 0.000 0.000 0.032 0.000 0.000 0.294

N yield Lolium perenne 0.000 0.963 0.704 0.000 0.000 0.000Agrostis capillaris 0.000 0.751 0.295 0.000 0.416 0.016Poa pratensis 0.000 0.677 0.980 0.008 0.682 0.535Trifolium repens 0.320 0.021 0.042 0.038 0.000 nt

N% DM Lolium perenne 0.000 0.000 0.838 0.000 0.001 0.000Agrostis capillaris 0.000 0.433 0.643 0.008 0.000 0.103Poa pratensis 0.000 0.041 0.593 0.009 0.000 0.004Trifolium repens 0.000 0.000 0.038 0.000 0.000 0.048

NDFF Lolium perenne 0.001 0.706 nt 0.000 0.740 ntAgrostis capillaris 0.000 0.827 nt 0.000 0.811 ntPoa pratensis 0.000 0.512 nt 0.000 0.371 ntTrifolium repens 0.000 0.000 0.034 0.000 0.000 0.088

RNUE Lolium perenne 0.001 0.492 0.999 0.000 0.001 0.474Agrostis capillaris 0.006 0.740 0.025 0.603 0.160 0.318Poa pratensis 0.328 0.728 0.580 0.516 0.162 0.852Trifolium repens 0.025 0.094 0.759 0.318 0.019 0.023

N: S ratio Lolium perenne 0.000 0.000 0.464 0.000 0.000 0.019Agrostis capillaris 0.000 0.000 0.341 0.000 0.061 0.808Poa pratensis 0.000 0.000 0.979 0.000 0.000 0.382Trifolium repens 0.000 0.000 0.487 0.096 0.000 0.003

P values resulting from ANOVA or Kruskal–Wallis (in italic) tests. In bold, P values "0.05; nt=non tested

DM: Dry mass (mg per plant); S and N yields (mg per plant); SDFF: S derived from fertilizer (%); RSUE: Real S use efficiency (%);NDFF: N derived from fertilizer (%); RNUE: Real N use efficiency (%)

270 Plant Soil (2008) 313:267–282

S derived from 34S-fertilizer (SDFF) and N derivedfrom 15N-fertilizer (NDFF) were calculated per plantby isotope mass balance as:

SDFF %! " # E34SplantE34Sfert

$ 100

NDFF %! " #E15Nplant

E15Nfert$ 100

where E34S and E15N are the 34S and 15N atom %excess of plant (Splant and Nplant) and of fertilizers(Sfert and Nfert).

The real S and N fertilizers use efficiencies (RSUEand RNUE, respectively), indicating the proportion offertilizer taken up per plant for each species, werecalculated as:

RSUE #Splant $ SDFF

Sfert

RNUE #Nplant $ NDFF

Nfert

where Splant and Nplant are the shoot S and N contentsin the above-ground DM per plant and, Sfert and Nfert

are the amount of S and N applied per pot.

Relative plant performances

In order to analyse and compare species’ behaviour inmonoculture then in mixture along the nutrientgradient, we selected two indices defined by Austin(1982): the relative physiological performance (RPP)and the relative ecological performance (REP). Theyare considered as good indices of performance abilityand competitiveness because the response of eachspecies is evaluated, taking into account the relativeability of that species and others to tolerate or exploitthe environment (Navas et al. 2002). Austin (1982)defined the relative physiological performance (RPP)of a species i in a monoculture at nutrient level j as:

RPPij # BmijMmj

where Bmij is the DM per unit area of species i atnutrient level j and Mmj is the DM per unit area of themost productive species at nutrient level j.

Relative ecological performance (REP) of a speciesi in mixture at nutrient level j was defined as:

REPij # BxijMxj

where Bxij is the DM per unit area of species i atnutrient level j and Mxj is the DM per unit area of themost productive species at nutrient level j.

RPP and REP vary between 0 and 1 and the valuesincrease with enhanced performance of species irelative to other species at nutrient level j.

Statistical analyses

The effects of S availability, N availability and theirinteraction on individual plant variables in mixturewere tested for each species by a two-way ANOVA(Minitab 13). When necessary, data were transformedto meet statistical assumptions. Not all variablessatisfied the required assumptions, even after trans-formation, and the non-parametric Kruskall–Wallistest was used to test N effect, S effect and treatmenteffect (N+S) as it does not test the N!S interaction.Comparison of means for the individual treatmentswas done at the 5% probability level based on theTukey test or on the Mood median test to determinewhich means differed significantly.

Results

Above-ground DM production and species relativeperformance

The four species differed considerably in theirresponses to nutrient availabilities (Table 1; Figs. 2and 3). Trends recorded at the establishment phase(Fig. 2) strongly contrasted with those at the regrowthphase (Fig. 3).

At the establishment phase, N significantly in-creased the DM production of grass when grown eitherin monocultures (Fig. 2a) or as a mixture (Fig. 2b). Inthe monoculture, A. capillaris was the most respon-sive species, followed by L. perenne, then P.pratensis. As a result, A. capillaris kept the highestRPP in all treatments, whereas that of P. pratensisdecreased along the N gradient. The effect of Sdepended on species, and was higher in the mixture.In monocultures, the DM production of grass did not

Plant Soil (2008) 313:267–282 271

vary (Fig. 2a, c, e) along the S gradient, whereas theDM production and the RPP of T. repens increased athigh N (Fig. 2g). In the mixture, L. perenne DMproduction was higher at intermediate S compared toother S levels, especially at low N (Fig. 2b; P"0.05).This species had almost twofold more DM productionthan that of other species, emerging as the bestcompetitor regardless of N and S levels. The additionof S significantly increased A. capillaris DM produc-tion at high N (Fig. 2d; P"0.05), increasing its REPand demonstrating that this species was a bettercompetitor than T. repens and P. pratensis in nutrientrich treatments. Indeed, the S gradient had no effecton P. pratensis (Fig. 2e, f) and, at high N, decreasedT. repens DM production, this species profiting fromN supply at low S only (Fig. 2h). In contrast, at low

and intermediate N, increased S significantly in-creased T. repens DM production (Fig. 2h). In alltreatments P. pratensis had the lowest REP andemerged as the subordinate species. As a result ofthe responses of individual species, the DM produc-tion of the mixture significantly increased with N(+55% between low and high N), intermediate Senhancing it at low N.

At the regrowth phase, the N gradient stillsignificantly increased the DM production of grass(Table 1), but also their relative performances (RPPand REP), in both monoculture (Fig. 3a) and mixture(Fig. 3b), although their production was lower than atthe establishment phase (Fig. 2 versus Fig. 3).Increased N slightly increased T. repens DM produc-tion in monoculture only (Fig. 3g), and decreased it

0

400

800

1200

1 2 3

Low NIntermediate NHigh N

0

400

800

1200

1 2 30

400

800

1200

1 2 30

400

800

1200

1 2 30

400

800

1200

1 2 30

400

800

1200

1 20

400

800

1200

1 2

0.890.880.80

0.800.800.79

0.810.780.72

0.890.880.80

0.800.800.79

0.810.780.72

I HLS level

RPP

(A) MonocultureLolium perenne Trifolium repensPoa pratensisAgrostis capillaris

1.001.001.00

1.001.001.00

1.001.000.96

1.001.001.00

1.001.001.00

1.001.000.96

I HL

0.680.700.68

0.810.870.80

0.990.991.00

0.680.700.68

0.810.870.80

0.990.991.00

I HL

0.740.680.53

0.590.650.60

0.850.890.75

0.740.680.53

0.590.650.60

0.850.890.75

I HL

0

400

800

1200

1600

0 15 300

1000

2000

3000

4000

0 15 300

400

800

1200

1600

0 15 300

400

800

1200

1600

0 15 300

400

800

1200

1600

0 150

400

800

1200

1600

0 15

1.001.001.00

1.001.001.00

1.001.001.00

1.001.001.00

1.001.001.00

1.001.001.00

I HLS level

REP0.590.510.37

0.420.300.52

0.460.340.44

0.590.510.37

0.420.300.52

0.460.340.44

I HL

0.210.170.18

0.270.220.24

0.280.180.26

0.210.170.18

0.270.220.24

0.280.180.26

I HL

0.250.200.36

0.300.280.19

0.430.400.41

0.250.200.36

0.300.280.19

0.430.400.41

I HL

(B) MixtureLolium perenne Trifolium repensPoa pratensisAgrostis capillaris

DM

in m

g pl

ant-1

DM

in m

g pl

ant-1 (a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

Establishment

Fig. 2 Above-ground dry matter per plant and relative speciesperformances (RPP and REP) of Lolium perenne, Agrostiscapillaris, Poa pratensis and Trifolium repens grown A in

mixture and B in mixture, at establishment, under low,intermediate and high N level along the S gradient. Note thedifferent scales for the y axis

272 Plant Soil (2008) 313:267–282

regardless of S level in the mixture (Fig. 3h).However, T. repens was the most productive speciesin all monoculture treatments except at high N–low S.In this treatment, L. perenne clearly produced moreDM than T repens. T. repens was also the dominantspecies in mixture at low and intermediate N, with thehighest REP, except in the low S–intermediate Ntreatment (Fig. 3h). The latter treatment was charac-terized by the enhancement of the DM production inall grass species with N and the dominance of L.perenne. The S gradient had greater and more variedeffects than at the establishment phase. At low andintermediate N, it mostly acted on T. repens,increasing its DM production in both monoculture(Fig. 3g) and mixture (Fig. 3h), and decreasing RPPand REP of grasses. At high N, S enhanced L.

perenne DM production in both culture types (Table 1and Fig. 3a, b), whereas, in contrast to trends noticedin monoculture (Fig. 3c), it decreased significantlythat of A. capillaris grown in mixture (Fig. 3d). Theonly S effect on P. pratensis was an enhancement ofits DM production at high N when grown as amonoculture (Fig. 3e), where it emerged as the mostproductive grass. These contrasting effects of the Sgradient on the different species modified specieshierarchy in high N mixtures (Fig. 3b). It promotedthe transition from a A. capillaris and L. perennedominated mixture to one dominated by L. perenneand T. repens at intermediate S, and one dominated byT. repens and L. perenne at high S. It also constrainedthe DM production of the mixture, which significantlyincreased with S addition (+200% between low and

1 2 3

0.700.731.00

0.320.360.72

0.310.330.40

0.700.731.00

0.320.360.72

0.310.330.40

I HLS level

RPP

(A) MonocultureLolium perenne Trifolium repensPoa pratensisAgrostis capillaris

0.700.790.82

0.300.380.57

0.280.350.38

0.700.790.82

0.300.380.57

0.280.350.38

I HL

0.870.920.70

0.340.450.67

0.290.310.43

0.870.920.70

0.340.450.67

0.290.310.43

I HL

1.001.000.79

1.001.001.00

1.001.001.00

1.001.000.79

1.001.001.00

1.001.001.00

I HL

0.391.000.77

0.120.221.00

0.090.100.33

0.391.000.77

0.120.221.00

0.090.100.33

I HLS level

REP0.170.431.00

0.060.110.61

0.040.060.30

0.170.431.00

0.060.110.61

0.040.060.30

I HL

0.080.230.37

0.030.050.31

0.020.030.15

0.080.230.37

0.030.050.31

0.020.030.15

I HL

1.000.760.52

1.001.000.84

1.001.001.00

1.000.760.52

1.001.000.84

1.001.001.00

I HL

(B) MixtureLolium perenne Trifolium repensPoa pratensisAgrostis capillaris

DM

in m

g pl

ant-1

DM

in m

g pl

ant-1 (a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

0

400

800

1200Low NIntermediate NHigh N

0

400

800

1200

0

2000

4000

6000

Regrowth

Fig. 3 Above-ground dry matter per plant and relative speciesperformances (RPP and REP) of Lolium perenne, Agrostiscapillaris, Poa pratensis and Trifolium repens grown A in

mixture and B in mixture, at regrowth, under low, intermediateand high N level along the S gradient. Note the different scalesfor the y axis

Plant Soil (2008) 313:267–282 273

L I H L I H L I H

Low Intermediate High

0

2

4

6

8

0

1

2

0

1

2

3

0

2

4

6

8

10

S yi

eld(

mg

plan

t-1)a

b b b b b

aab bb

bb

a

bab ab

bb

a

b

bcdcd

cd

(B) Regrowth

0

20

40

60

80

100

0

20

40

60

80

100

L I H L I H L I H

Low Intermediate High

0

20

40

60

80

100

0

20

40

60

80

100

a aa

a

a

a

abc c c

d d

a b b b

c c

b b b

c c

a a

b

ababab

SDFF

(%)

(A) Establishment

Trifolium repens

Poa pratensis

Agrostis capillaris

Lolium perenne

(a)

(c)

(b)

(d)

(e) (f)

(g) (h)

SDFFS yield

SDFFS yield

SDFFS yield

SDFFS yield

Fig. 4 S yield and S derived from fertilizer (SDFF) in (a, b)Lolium perenne, (c, d) Agrostis capillaris, (e, f) Poa pratensisand (g, h) Trifolium repens grown in mixture at A establishmentand at B regrowth along N and S gradients. Lines: S yield (rightvertical axis; note the different scales for this axis); Histograms:

SDFF (S derived from fertilizer expressed as percentage of Syield; left vertical axis). Bars are mean±SE (n=4). Twodifferent letters indicate a significant difference within eachspecies and phase; P"0.05

274 Plant Soil (2008) 313:267–282

high S), while in contrast to the establishment phase,N decreased it (!20% between low and high N).

S yield and uptake

The S yield of all species increased along the S gradientin both establishment and regrowth phases (Fig. 4;Table 1). The N gradient modulated the significanceof this increase in different ways according to bothspecies and growth phases (Fig. 4a versus b).

At the establishment phase, the S% DM in theshoot of all species increased along the S gradient,while N input generally decreased it (Table 1). L.perenne and A. capillaris were characterized byhigher S yields than those of other species (Fig. 4a).The S yields increased with S addition, especially athigh N for grasses (Fig. 4a, c, e) and at low N for T.repens (Fig. 4g). The N gradient had no effect ongrasses at low S, but it significantly increased the Syield of L. perenne at high S (Fig. 4a) and that of A.capillaris at intermediate and high S (Fig. 4c). No Neffect was detected on P. pratensis. In contrast tothose of grass, the S yield of T. repens decreasedalong the N gradient (Fig. 4g). Accordingly, theamount of S derived from fertilizer (SDFF; Fig. 4a)by grass progressively increased along the N gradient,from 20% at low N up to 60% at high N, but wasalmost unaffected by the S level. Their relative S useefficiencies (RSUE; Table 2) decreased along the Sgradient, but very significantly increased along the N

gradient. As a result, the S recovered by the plants inmixture increased from 10% at low N to 40% at highN. T. repens derived most of its S content from Sfertilizer (between 60 and 80%) regardless of N level(Fig. 4a). Its RSUE significantly decreased along theN gradient (Table 2).

At the regrowth phase, like at the establishmentphase, the S% DM in the shoot of each speciesincreased with S supply, while N decreased it(Table 1), this only at high S level for T. repens.The S yield of T. repens was higher than at theestablishment phase (Fig. 4h versus g), whereas thoseof grass were lower (Fig. 4b, d, f versus a, c, e). The Syield of grasses remained similar regardless of N leveland slightly increased along the S gradient, except forA. capillaris (Fig. 4b, d, f). In contrast, the S yield ofT. repens decreased along the N gradient (Fig. 4h),whereas S addition very significantly increased it. TheSDFF was in the range of 20–40% in the grassspecies (Fig. 4b, d, f), increasing along the S gradientat low N and along the N gradient at intermediate S.The SDFF of T. repens increased along the S gradientregardless of N level. T. repens clearly had the highestRSUE, followed by that of L. perenne (Table 2). TheRSUE of grasses remained similar regardless of bothS and N gradients, whereas that of T. repens increasedwith S and decreased with N as at the establishmentphase. Thus, T. repens emerged as the best competitorfor S fertilizer uptake, mainly under low andintermediate N availability.

Table 2 N and S effect on the mean RSUE (real S use efficiency as % of 34S-fertilizer applied) per plant of Lolium perenne, Agrostiscapillaris, Poa pratensis and Trifolium repens grown in mixture

Establishment phase Regrowth phase

S level Intermediate High Intermediate High

Lolium perenne N level Low 1.61ab 1.01a 0.56NS 0.64NS

Intermediate 2.65bc 1.91a 0.61NS 0.71NS

High 5.93c 5.24c 0.65NS 0.73NS

Agrostis capillaris Low 0.28a 0.20a 0.17NS 0.17NS

Intermediate 0.66a 0.37a 0.23NS 0.16NS

High 3.34b 1.66b 0.31NS 0.21NS

Poa pratensis Low 0.21ab 0.10a 0.12NS 0.1NS

Intermediate 0.26ab 0.22b 0.1NS 0.11NS

High 0.38b 0.37b 0.07NS 0.1Trifolium repens Low 2.59c 1.18ab 1.91c 3.2d

Intermediate 1.86bc 1.83ab 1.14b 2.61cd

High 0.97ab 0.81a 0.17a 0.94b

Results followed by different letters indicate a significant difference between treatment within each species and phase; P"0.05

Plant Soil (2008) 313:267–282 275

N yield and uptake; N: S ratio

The S gradient had lower effects on N yield than on Syield, except for T. repens at the regrowth phase. TheN gradient increased N yields of grass, this moresignificantly at the establishment phase, while itdecreased that of T. repens.

At the establishment phase, the N% DM in theshoot of grass strongly increased along the Ngradient, while that of T. repens decreased, andgenerally decreased along the S gradient, while thatof T. repens increased (Table 1). L. perenne (Fig. 5a)was characterized by the highest N yield, followed bythose of T. repens (Fig. 5g) and A. capillaris (Fig. 5c).The N yields of grass increased with N addition, butwere independent of S level, except for A. capillarisat high N (Fig. 5c; Table 1). In contrast, that of T.repens remained almost the same regardless of Nlevel and increased along the S gradient (Fig. 5g;Table 1). The NDFF increased along the N gradient ineach species (Fig. 5a). The S gradient did not affectNDFF in grass, whereas it decreased it in T. repens,suggesting that S input promoted N2 fixation. L.perenne then A. capillaris possessed much higherRNUE values than the two other species (Table 3).Apart from that of P. pratensis, RNUE of otherspecies improved along the N gradient but not alongthe S gradient. Considering the whole mixture, mostof the applied N was recovered at the establishmentphase, from 30% at low N to 50% at high N.

At the regrowth phase, T. repens clearly had thehighest N yield and N% DM in the shoot. Only the N% DM of L. perenne increased along the N gradient,while that of T. repens still decreased (Table 1). The Sgradient still increased the N% DM of T. repens. Italso increased that of the three grass species, thisbeing not observed at the establishment phase. Forgrass, high N promoted higher N yields (Fig. 5b, d, f),this especially for L. perenne, suggesting that thisspecies was the best competitor for N capture. Inparallel, it decreased the N yield of T. repens(Fig. 5h). The S gradient promoted an increase inthe N yield of L. perenne (Fig. 5h). In grass, theNDFF increased along the N gradient, while it wasunaffected by the S gradient (Fig. 5b, d, f). In T.repens, NDFF also increased along the N gradient,but strongly decreased along the S gradient (Fig. 5h).Just as at the establishment phase, L. perenne and A.capillaris had the highest RNUE (Table 3). That of L.

perenne increased along the S and N gradients whilethat of T. repens decreased along the S gradient athigh N.

As a result of the differences in N and S yields, theN: S ratio strongly varied among treatments, espe-cially for grass species (Table 1). At both establish-ment and regrowth phases, it increased with Naddition (from 12 to 45 at low S for L. perenne) anddecreased with S addition (from 45 to 11 at high N forL. perenne). The N: S ratio of T. repens was lessaltered by both N and S gradients suggesting that thisspecies was more able to maintain its internal N: Sequilibrium. At the regrowth phase, T. repens showedmuch higher N: S ratios than those of grass except athigh N–low S.

Discussion

Effect of N and S availabilities on speciesperformance at establishment phase

In the establishment phase, as expected (e.g. Jones etal. 1973; Marschner 1986; Harris and Clark 1996;McKenzie 1996; Zemenchik and Albrecht 2002),grass yield in mixture strongly increased withincreasing N availability. RPP and REP profiles ofspecies indicated that their physiological performanceassessed in monoculture did not fully reflect theirability to compete with other species and was not theonly factor explaining their success or subordinationin mixture. Interspecific competition for both soilnutrients, then light, mostly modulated species hier-archy and their relative performance in mixture.

L. perenne performance, in terms of S yield, S useefficiency (RSUE) and N allocation to growth (Nyield per plant remained stable whereas N% DMdecreased), was increased by intermediate S level inmixture, especially at low N. That suggests that Slimitation may constrain L. perenne growth inunfertilized grassland. However, whatever N and Slevels, its dominance persisted, and it emerged as thebest performer in mixture. This dominance was notbased on its physiological performance, it could evenbe the worst performer in monoculture (see RPP inlow N treatments), but on its competitiveness againstother species. That confirms the findings of Aguiar etal. (2001), who showed this species to be moreaffected by intraspecific competition than by inter-

276 Plant Soil (2008) 313:267–282

0

20

40

60

80

100

0

10

20

30

40

0

2

4

6

8

10

L I H L I H L I H

Low Intermediate High

0

50

100

150

200

0

20

40

60

80

100

L I H L I H L I H

Low Intermediate High

0

20

40

60

80

100

0

20

40

60

80

100

0

20

40

60

80

100N

DFF

(%)

N y

ield

(mg

plan

t-1)

a a a

b b b

a a a

b b

abc d

ef

b ab a

c c c

b ab a

c c c

bc ab a

dd

a a

b bb

aa a

b bb

a a a

b bb

a

b

N level

S level

a ab

c c

d

(B) Regrowth(A) Establishment

d

Lolium perenne

Agrostis capillaris

Poa pratensis

Trifolium repens

(a)

(c)

(b)

(d)

(e) (f)

(g) (h)

NDFFN yield

NDFFN yield

NDFFN yield

NDFFN yield

Fig. 5 N yield and N derived from fertilizer (NDFF) in (a, b)Lolium perenne, (c, d) Agrostis capillaris, (e, f) Poa pratensisand (g, h) Trifolium repens grown in mixture at A establishmentand at B regrowth along N and S gradients. Lines: N yield(right vertical axis; note the different scales for this axis);

Histograms: NDFF (N derived from fertilizer expressed aspercentage of N yield; left vertical axis). Bars are mean±SE(n=4). Two different letters indicate a significant differencewithin each species and phase; P"0.05

Plant Soil (2008) 313:267–282 277

specific competition. This competitiveness has beenexplained by its larger root system, described asdeeper and more prospective (Boot and Mensink1990) in mixture than in monoculture (Marriott andZuazua 1996). These authors suggested that thisallowed L. perenne to exploit soil more efficientlythan others and to supplant them, even at low N.Indeed, we proved that L. perenne possessed thehighest RNUE and RSUE per plant, projecting thisspecies as the best soil forager and exploiter.Accordingly, L. perenne possessed the highest capac-ity to allocate N to aboveground growth and grewfaster. This allowed it to exert a negative pressure onadjacent species by altering the quality and quantityof light that infiltrates through the canopy as showedby Laidlaw and Withers (1998). Our results also agreewith those of Hodge et al. (1999, 2000) on L.perenne–P. pratensis mixture, who showed that theformer although performing less than P. pratensis inmonoculture, produced much higher root length anddensity, taking up more soil N in mixture. Indeed, weproved that P. pratensis, with the lowest RNUE andRSUE whatever treatment in mixture, had the lowestcapacity to exploit soil. It appeared as the poorestperformer in mixture and was the subordinate specieswhatever the N and S conditions.

A. capillaris was the best performer in monocul-ture, confirming the results of Crush et al. (2005),who, comparing eleven temperate forage grass speciesin a single growth phase experiment, found that thisspecies benefited from the highest nitrate uptake perunit root weight. However, when grown in a mixtureit was out-competed by L. perenne. In grassland, suchsubordination of this species is well known, and hasbeen attributed to its lower canopy height than other

grass species (especially L. perenne) resulting inshading (Barthram et al. 2005). However, in high Ntreatments, S clearly enhanced its biomass productionand its performance and competitiveness in mixtureclose to that of L. perenne in mixture. In contrary toother species its N yield increased, demonstrating thatthis species is a strong competitor for nutrient uptake.This emphasizes the conclusions of Newbery andWolfenden (1996) suggesting that to perform in highN soils, A. capillaris needs a high availability of othermajor nutrients. Our results explain this fact by theenhancement of N use efficiency (RNUE) with anincreasing S availability.

N clearly enhanced dry matter production of grasswhatever the S level, but this effect was restricted tolow S for T. repens grown in mixture. N had astronger effect on its N budget. At low S, along the Ngradient, we observed a decrease in N% DM and anincrease in N derived from 15N-fertilizer (NDFF)applied at experiment start. These findings underlinethat N2 fixation was altered by N, as suggested byMacduff et al. (1996) and Soussana et al. (2002).Moreover, at high N, the S gradient altered its drymatter production. This can be explained by alowering of its relative competitive ability to captureS in N-rich soil mixture, as N decreased both its Syield and S use efficiency (RSUE). In a mixture ofperennials grasses and clover, this finding expands onresults of Gilbert and Robson (1984a) on annualTrifolium subterraneum/Lolium rigidum mixture,which showed that grass absorbs more S at theexpense of the legume especially under high Navailability. In both monoculture and mixture, thenegative N effect, especially at low S, resulted in adecrease in relative physiological and ecological

Table 3 N and S effect on the mean RNUE (real N use efficiency as % of 15N-fertilizer applied) per plant of Lolium perenne, Agrostiscapillaris, Poa pratensis and Trifolium repens grown in mixture

Establishment phase Regrowth phase

S level Low Intermediate High Low Intermediate High

Lolium perenne N level Intermediate 4.60ab 3.88a 4.29ab 0.63a 0.81ab 1.1bc

High 5.92b 5.96b 6.83b 0.93ab 1.27c 1.34Agrostis capillaries Intermediate 2.78ab 1.96ab 1.99a 0.43NS 0.37NS 0.54NS

High 2.40a 3.82b 3.71b 0.52NS 0.32NS 0.4NS

Poa pratensis Intermediate 337NS 0.73NS 0.60NS 0.18NS 0.15NS 0.17NS

High 0.84NS 0.77NS 0.86NS 0.19NS 0.12NS 0.14NS

Trifolium repens Intermediate 0.97ab 0.74a 0.74a 0.12abc 0.09a 0.14bc

High 2.01b 1.07ab 1.02ab 0.22c 0.1ab 0.11ab

278 Plant Soil (2008) 313:267–282

performances of T. repens, in agreement with numer-ous studies in grasslands (Simon et al. 1997).

Effect of N and S availabilities on speciesperformance at regrowth phase

As for the establishment phase, N clearly enhancedthe DM production of grass and their physiologicaland ecological performances (RPP and REP), thisespecially at low S. L. perenne and A. capillarisperformances in monoculture and mixture followedalmost the same pattern at low and intermediate N:their performances in mixture are governed by theirspecific aptitude to take up N and S at these N levels.Accordingly, L. perenne performed better than othergrasses as it did at establishment phase. However athigh N, REP reflected RPP only at high S. Thatmeans that competitive ability for S or relative Srequirement dominated physiological ability to ex-ploit the environment.

At high N–low S, both L. perenne and A. capillarisdominated the mixture, the latter being the bestcompetitor. Such a pattern could be attributed,according to Crush et al. (2005), to a higher nitrateinterception ability of A. capillaris, this being crucialin a poorer N soil than at establishment. But ourresults also suggest other mechanisms implying that Srequirement is different according to species. L.perenne, the most N and S demanding grass, asproved by its highest shoot N and S contents, N and Suse efficiency, was in such condition probably S-limited. This species was indeed characterized by avery high N% DM, more than twofold higher than atother S availabilities, indicating that N was not dilutedinto biomass but accumulated without contributing togrowth. This emphasises the findings of Freney et al.(1978) and Gilbert et al. (1997) on wheat and barleythat growth rate was reduced and amides accumulatedwhen a lack of S for protein synthesis occurs. Incontrast, A. capillaris, which requires less S (lowershoot S % DM and RSUE) than L. perenne,performed better. This shows that the metabolism ofS and N are also closely co-ordinated in grasslandspecies and that grass species relative performanceclearly depends on both absolute N and S availabil-ities and their ratio. This also implies that the internalN: S ratio greatly varied, as when one nutrient wasdeficient, the other accumulated without being use forgrowth.

At high N–intermediate S, L. perenne and T.repens replaced A. capillaris. The increase in Savailability benefited L. perenne, which, in a lessdeficient S soil condition, emerged as the bestcompetitor. As discussed for establishment phase, thisprobably may lay on a more optimal root foragingthan other species in mixture, allowing it to mobilizenutrients more efficiently than other species, whereasin monoculture each L. perenne individual clearlysuffered from intraspecific competition.

As observed in the establishment phase, P. praten-sis was the subordinate species and appeared to be apoor competitor even at high N–high S supply whereit however showed a high performance in monocul-ture (RPP). This result illustrated that of Bender et al.(2006) who showed that in mixture P. pratensisproduced 44% less dry matter than in monoculture,other species being stronger competitors. As in theestablishment phase, the gap between the potentialperformance suggested by its RPP and the realizedperformance in mixture may be explain by therepression of root development in the presence ofother grass species such as L. perenne (Hodge et al.1999, 2000). Indeed, in mixture, P. pratensis wascharacterized by the worst N and S use efficiencies,indicating that it was the worst competitor for nutrientcapture.

T. repens was generally the best physiologicalperformer in monoculture and the best competitor inmost mixture treatments. Under low N availability, T.repens showed the highest performances. In thisoligotrophic soil, atmospheric N2 fixation was notaltered by high soil N availability and it performedwell, whereas grass species did not meet their optimalN requirement. At intermediate N, T. repens stilldominated even if REP of grasses increased, espe-cially at low S. At both these N levels, the dominanceof T. repens was amplified along the S gradient. Athigh N–low S, T. repens had lower physiological andecological performances than L. perenne and A.capillaris. As previously discussed for establishmentphase, in such condition grasses dominated because(1) N2 fixation was altered and (2) grasses competedbetter for soil N. Indeed, even if T. repens showed ahigh shoot N content, high N–low S treatment wascharacterized by the highest absolute NDFF andRNUE proving that the contribution of N2 to Nnutrition decreased. The most relevant result is that atthe same high N level, high S availability allowed T.

Plant Soil (2008) 313:267–282 279

repens to suppress grasses. Its DM productionstrongly increased along the S gradient. Its shoot N% DM also increased, but its N derived from 15N-fertilizer (NDFF) and its N use efficiency (RNUE)significantly decreased. That proves that T. repensexploited more efficiently another N source thanfertilizer when more S is available. As soil wasalmost depleted in native mineral N at the beginningof the experiment, as poor organic N mineralizationoccurred (data not shown) and as T. repens poorlycompeted with grasses for soil N, this other sourceshould be considered as mainly N derived from N2

fixation processes. Therefore a higher S availabilityenhanced N2 fixation or suppressed inhibition causedby high N availability provided by high N fertilizationrate. This strong relationship between S nutrition andN acquisition by T. repens was emphasized by theparallel dynamics of RSUE values and shoot Ncontent. DeBoer and Duke (1982), dealing withMedicago sativa L. and Habtemichial et al. (2007)with Viscia faba L., have already shown that S supplyincreased N2 fixation in N-rich environment and thatsubsequent effects of S deficiency may be due to lossof fixed N. Our study suggests that a similarmechanism may occurred in plant mixtures integrat-ing T. repens, giving the latter a competitive advan-tage, and allowing its persistence in high N fertilizedconditions.

In general terms, even if nutrient capture fell forgrass at regrowth phase, increasing S availabilitygenerally resulted in an increase in N% DM,especially at low and intermediate N availabilities,and even in a significant increase in N yield for L.perenne. In that case, the S effect may be indirect andlinked with the high performance of T. repens whichcould then be of benefit to grass species from higherN transfer from T. repens.

Conclusion

Our results show that a S gradient clearly alters bothphysiological and ecological performances of species,their competitive abilities and plant mixture structure.It modulates the N gradient effects in both monocul-ture and mixture. S effects emerged in the establish-ment phase, but strongly increased at regrowth aftercutting of mature plants, when competition fornutrients increased. These effects depended both on

the relative N: S requirements of species and on theircompetitive ability to take up N and S. They werealso based on the enhancement of N use efficiency byS availability. Indeed, along the S gradient, the N useefficiency of grasses was promoted while N2 fixationby T. repens strongly increased, or remained high inhigh N soils. This conferred to the latter a strongcompetitive advantage in high S compared to low Ssoils, this being essential in high N environment inallowing it to compete and avoid competitive exclu-sion. Based on the positive effect on the dominantspecies L. perenne and T. repens, increasing Savailability promoted increased yield of whole plantmixture whatever N availability. As a result, Sdepletion of soils should affect both plant mixturestructure and productivity. Conversely, S fertilisationof grasslands, especially in high N fertilized grass-lands, emerged as a tool of plant diversity modulationand potentially of leguminous species management.

Acknowledgments We wish to thank R Ségura and AFAmeline for assistance in the maintenance of the experiment.We also thank P Beauclair and J Bonnefoy for help duringharvests. T. Gordon and anonymous referees improved themanuscript through helpful comments. Their assistance isgratefully acknowledged.

References

Aerts R (1999) Interspecific competition in natural plantcommunities: mechanisms, trade-offs and plant–soil feed-backs. J Exp Bot 50:29–37

Aerts R, Van der Peijl MJ (1993) A simple model to explain thedominance of low-productive perennials in nutrient-poorhabitats. Oikos 66:144–147

Aguiar MR, Lauenroth WK, Peters DP (2001) Intensity ofintra- and interspecific competition in coexisting short-grass species. J Ecol 89:40–47

Austin MP (1982) Use of a relative physiological performancevalue in the prediction of performance in multispeciesmixtures from monoculture performance. J Ecol 70:559–570

Barthram GT, Elston DA, Mullins CE (2005) The physicalresistance of grass patches to invasion. Plant Ecol 176:79–85

Bender J, Muntifering RB, Lin JC, Weigel HJ (2006) Growthand nutritive quality of Poa pratensis as influenced byozone and competition. Environ Pollut 142:109–115

Boot RGA, Mensink M (1990) Size and morphology of rootsystems of perennial grasses from contrasting habitatsas affected by nitrogen supply. Plant Soil 129:305–307

Campbell BD, Grime JP (1989) A comparative study of plantresponsiveness to the duration of episodes of mineralnutrient enrichment. New Phytol 112:261–267

280 Plant Soil (2008) 313:267–282

Ceccotti SP, Messick DL (1997) A global review of croprequirements, supply, and environmental impact on nutri-ent sulphur balance. In: Cram WJ, De Kok LJ, Stulen I,Brunold C, Rennenberg H (eds) Sulphur metabolism inhigher plants: molecular, ecophysiological and nutritionalaspects. Backhuys, Leiden, pp 155–163

Clarkson DT, Diogo E, Amancio S (1999) Uptake andassimilation of sulphate by sulphur deficient Zea mayscells: the role of O-acetyl-L-serine in the interactionbetween nitrogen and sulphur assimilatory pathways. PlantPhysiol Biochem 37:283–290

Crush R, Waller JE, Care DA (2005) Root distribution andnitrate interception in eleven temperate forage grasses.Grass Forage Sci 60:385–392

Cullen NA (1970) Establishment of pasture on yellow-brownloams near Te Anau. VI. Pasture response to rates ofphosphorus, sulphur and molybdenum. New Zeal J AgricRes 14:10–17

De Vries DM, Kruijne A 1960 The influence of nitrogenfertilization on the botanical composition of permanentgrassland. Medeling I.B.S. 103 Wageningen.

DeBoer DL, Duke SH (1982) Effects of sulphur nutrition onnitrogen and carbon metabolism in Lucerne (Medicagosativa L.). Physiol Plant 54:343–350

Droux M (2004) Sulfur assimilation and the role of sulfur inplant metabolism: a survey. Photosynth Res 79:331–348

Freney JR, Spencer K, Jones MB (1978) The diagnosis ofsulphur deficiency in wheat. Aust J Agric Res 29:727–738

Gilbert MA, Robson AD (1984a) The effect of sulfur supply onthe root characteristics of subterranean clover and annualryegrass. Plant Soil 77:377–380

Gilbert MA, Robson AD (1984b) Studies on competition forsulfur between subterranean clover and annual ryegrass. I.Effect of nitrogen and sulfur supply. Aust J Agric Res35:53–64

Gilbert SM, Clarkson DT, CambridgeM, Lambers H, HawkesfordMJ (1997) SO4

2! deprivation has an early effect on thecontent of ribulose-1,5-bisphosphate carboxylase/oxygenaseand photosynthesis in young leaves of wheat. Plant Physiol115:1231–1239

Güsewell S, Bollens U (2003) Composition of plant speciesmixtures grown at various N:P ratios and levels of nutrientsupply. Basic Appl Ecol 4:453–466

Habtemichial KH, Singh BR, Aune JB (2007) Wheat responseto N2 fixed by faba bean (Vicia faba L.) as affected bysulfur fertilization and rhizobial inoculation in semi-aridNorthern Ethiopia. J Plant Nutr Soil Sci 170:412–418

Harper JL (1977) Population biology of plants. Academic,London

Harris SL, Clark DA (1996) Effect of high rates of nitrogenfertiliser on white clover growth, morphology, andnitrogen fixation activity in grazed dairy pasture innorthern New Zealand. New Zeal J Agric Res 39:149–158

Hodge A, Robinson D, Griffiths BS, Fitter AH (1999) Whyplants bother: root proliferation results in increasednitrogen capture from an organic patch when two grassescompete. Plant Cell Environ 22:811–820

Hodge A, Stewart J, Robinson D, Griffiths BS, Fitter AH(2000) Spatial and physical heterogeneity of N supply

from soil does not influence N capture by two grassspecies. Funct Ecol 14:645–653

Jones LHP, Jarvis SC, Cowling DW (1973) Lead uptake fromsoils by perennial ryegrass and its relation to the supply ofan essential element (sulphur). Plant Soil 38:605–619

Laidlaw AS, Withers JA (1998) Changes in contribution ofwhite clover to canopy structure in perennial ryegrass/white clover swards in response to N fertilizer. GrassForage Sci 53:287–291

Macduff JH, Jarvis SC, Davidson IA (1996) Inhibition of N2fixation by white clover (Trifolium repens L.) at lowconcentrations of NO3! in flowing solution culture. PlantSoil 180:287–295

Marriott CA, Zuazua MT (1996) Tillering and partitioning ofdry matter and nutrients in Lolium perenne growing withneighbours of different species: effects of nutrient supplyand defoliation. New Phytol 132:87–95

Marschner H (1986) Mineral nutrition of higher plant.Academic, London

McGrath SP, Zhao FJ (1995) A risk assessment of sulphurdeficiency in cereals using soil and atmospheric depositiondata. Soil Use Manage 11:110–114

McKenzie FR (1996) The influence of applied nitrogen onherbage yield and quality of Lolium perenne L. pasturesduring the establishment year under subtropical condi-tions. S Afr J Plant Soil 13:22–26

Morton JD, Smith LC, Metherell AK (1999) Pasture yieldresponses to phosphorus, sulphur, and potassium applica-tions on North Otago soils, New Zealand. New Zeal. JAgric Res 42:133–146

Murphy MD, Boggan JM (1988) Sulphur deficiency in herbagein Ireland. Ir J Agric Food Res 27:83–90

Murphy MD, Coulter BS, Noonan DG, Connolly J (2002) Theeffect of sulphur fertilisation on grass growth and animalperformance. Ir J Agric Food Res 41:1–15

Navas ML, Garnier E, Austin MP, Viaud A, Gifford RM(2002) Seeking a sound index of competitive intensity:application to the study of biomass production underelevated CO2 along a nitrogen gradient. Austral Ecol27:463–473

Newbery RM, Wolfenden J (1996) Effects of elevated CO2 andnutrient supply on the seasonal growth and morphology ofAgrostis capillaris. New Phytol 132:403–411

Olde Venterink HO, Wassen MJ, Belgers JDM, Verhoeven JTA(2001) Control of environmental variables on speciesdensity in fens and meadows: importance of direct effectsand effects through community biomass. J Ecol 89:1033–1040

Scherer HW (2001) Sulphur in crop production. Eur J Agron14:81–111

Simon JC, Leconte D, Vertes F, Meur D (1997) Management ofpersistence of white clover in mixtures. Fourrages128:483–498

Sinclair AG, Smith LC, Morrison JD, Dodds KG (1996) Effectsand interactions of phosphorus and sulphur on a mownwhite clover/ryegrass sward. 1. Herbage dry matterproduction and balanced nutrition. New Zeal J Agric Res39:421–433

Soussana JF, Minchin FR, Macduff JH, Raistrick N, AbbertonMT, Michaelson-Yeates TPT (2002) A simple model offeedback regulation for nitrate uptake and N2 fixation in

Plant Soil (2008) 313:267–282 281

contrasting phenotypes of white clover. Ann Bot-London90:139–147

Tilman D (1987) On the meaning of competition and themechanisms of competitive superiority. Funct Ecol 1:304–315

Walker TW, Adams AFR, Orchiston HD (1956) The effect oflevels of calcium sulphate on the yield and composition ofa grass and clover pasture. Plant Soil 7:290–300

Wedin DA, Tilman D (1993) Competition among grasses alonga nitrogen gradient—initial conditions and mechanisms ofcompetition. Ecol Monogr 63:199–229

Withers PJA, Tytherleigh ARJ, O’Donnell FM (1995) Effect ofsulphur fertilizers on the grain yield and sulphur content ofcereals. J Agric Sci 125:317–324

Zemenchik RA, Albrecht KA (2002) Nitrogen use efficiencyand apparent nitrogen recovery of Kentucky bluegrass,smooth bromegrass, and orchardgrass. Agron J 94:421–428

Zhao FJ, Hawkesford MJ, McGrath SP (1999) Sulphurassimilation and effects on yield and quality of wheat. JCereal Sci 30:1–17

282 Plant Soil (2008) 313:267–282