LiM

N

RA

A

sssgIcp

ilad

c

Z

dhuiGjVvM

1h

Basic and Applied Ecology 14 (2013) 506–514

ocal variation in conspecific plant density influences plant–soil feedbackn a natural grassland

artine Kos∗, Johan Veendrick, T. Martijn Bezemer

etherlands Institute of Ecology (NIOO-KNAW), Department of Terrestrial Ecology, P.O. Box 50, 6700 AB Wageningen, The Netherlands

eceived 19 November 2012; accepted 12 July 2013vailable online 22 August 2013

bstract

Several studies have argued that under field conditions plant–soil feedback may be related to the local density of a plantpecies, but plant–soil feedback is often studied by comparing conspecific and heterospecific soils or by using mixed soilamples collected from different locations and plant densities. We examined whether the growth of the early successionalpecies Jacobaea vulgaris in soil collected from the field is related to the local variation in plant density of this species. In arassland restoration site, we selected eight 8 m × 8 m plots, four with high and four with low densities of J. vulgaris plants.n 16 subplots in each plot we recorded the density and size of J. vulgaris, and characteristics of the vegetation and the soilhemistry. Soil collected from each subplot was used in a greenhouse pot-experiment to study the growth of J. vulgaris, both inure field soil and in sterile soil inoculated with a small part of field soil.In the field, flowering J. vulgaris plants were taller, the percentage of rosette plants was higher and seed density was larger

n High- than in Low-density plots. In the pot experiment, J. vulgaris had a negative plant–soil feedback, but biomass was alsoower in soil collected from High- than from Low-density plots, although only when growing in inoculated soil. Regressionnalyses showed that J. vulgaris biomass of plants growing in pure soil was related to soil nutrients, but also to J. vulgarisensity in the field.We conclude that in the field there is local variation in the negative plant–soil feedback of J. vulgaris and that this variation

an be explained by the local density of J. vulgaris, but also by other factors such as nutrient availability.

usammenfassung

Verschiedene Studien haben dargelegt, dass unter Feldbedingungen das Pflanze-Boden-Feedback mit der lokalen Dichteer Pflanze in Verbindung stehen könnte, aber das Pflanze-Boden-Feedback wurde oft untersucht, indem konspezifische undeterospezifische Böden miteinander verglichen wurden oder indem gemischte Bodenproben von verschiedenen Standortennd Pflanzendichten benutzt wurden. Wir untersuchten, ob das Wachstum der frühen Sukzessionspflanze Jacobaea vulgaris inm Freiland gesammelten Böden mit der Variation in der lokalen Dichte dieser Art in Beziehung steht. In einem rekultivierten

rasland wählten wir acht 8 m × 8 m große Probeflächen aus, vier mit hoher und vier mit geringer Dichte von J. vulgaris mit

eweils 16 Unterprobeflächen. Hier registrierten wir die Dichte und Größe der J. vulgaris-Pflanzen, sowie Eigenschaften deregetation und Daten zur Bodenchemie. Boden von jeder Unterprobefläche wurde in einem Topf-Experiment im Gewächshauserwendet, um das Wachstum von J. vulgaris sowohl in reiner Freilanderde als auch in steriler Erde, die mit einer geringenenge Boden aus dem Freiland inokuliert war, zu untersuchen.

∗Corresponding author. Tel.: +31 0 317 473632; fax: +31 0 317 473675.E-mail address: M.Kos@nioo.knaw.nl (M. Kos).

439-1791/$ – see front matter © 2013 Gesellschaft für Ökologie. Published by Elsevier GmbH. All rights reserved.ttp://dx.doi.org/10.1016/j.baae.2013.07.002

ItnaBB

dw©

Kp

I

ttpW2PdcecoPiKsNls2

e2esbssmResbBdsla&

M. Kos et al. / Basic and Applied Ecology 14 (2013) 506–514 507

m Freiland waren auf Flächen mit hoher Pflanzendichte die blühenden J. vulgaris-Pflanzen größer und der Anteil von Roset-enpflanzen sowie die Samendichte höher als auf Flächen mit geringer Pflanzendichte. Im Topf-Experiment zeigte J. vulgaris einegatives Pflanze-Boden-Feedback, aber die Biomasse war ebenfalls geringer in Erde von Flächen mit dichtem Pflanzenbestandls in Erde von Flächen mit geringer Pflanzendichte – allerdings nur bei inokulierter Erde. In Regressionsanalysen zeigte dieiomasse der J. vulgaris-Pflanzen eine Beziehung zu den Nährstoffgehalten, wenn sie in reiner Erde wuchsen, aber auch eineeziehung zur Pflanzendichte im Freiland.Wir schließen hieraus, dass es im Freiland eine räumliche Variation des negativen Pflanze-Boden-Feedbacks gibt und dass

iese Variation durch die lokale Dichte von J. vulgaris aber auch durch andere Faktoren wie die Nährstoffverfügbarkeit erklärterden kann.

2013 Gesellschaft für Ökologie. Published by Elsevier GmbH. All rights reserved.

eywords: Calluna vulgaris; Inoculated soil; Jacobaea vulgaris; Nutrients; Pure soil; Ragwort; Senecio jacobaea; Soil pathogens; Spatial

siFvs1aswsaaKssaa2ftfiooJ

lKrlsK2snpif

atterning

ntroduction

Plants influence the biology, chemistry and structure ofhe soil they grow in, which in turn, can lead to changes inhe performance of plants that grow later in the soil. Thisrocess is referred to as plant–soil feedback (PSF) (Bever,estover, & Antonovics 1997; Ehrenfeld, Ravit, & Elgersma

005; Kulmatiski, Beard, Stevens, & Cobbold 2008; van derutten et al. 2013). PSF can be mediated by abiotic soil con-itions, such as availability of nutrients, as well as bioticonditions, such as presence of microorganisms (Bezemert al. 2006b; Kulmatiski et al. 2008). Biotic PSF effectsan be highly species-specific, while abiotic PSF effects areften less species-specific (Aerts and Chapin 2000; Reynolds,acker, Bever, & Clay 2003). PSF can be positive due to

ncreased nutrient availability (Chapman, Langley, Hart, &och 2006) or accumulation of mutualistic microorganisms

uch as arbuscular mycorrhizal fungi (Klironomos 2002).egative PSFs can arise from nutrient depletion or immobi-

isation (Ehrenfeld et al. 2005) or the build-up of pathogenicoil microorganisms (Klironomos 2002; Packer and Clay000; van der Putten, van Dijk, & Peters 1993).Several studies have argued that the PSF of a species can

xplain the abundance of this species in the field (Klironomos002; Kulmatiski et al. 2008; Mangan et al. 2010; Reynoldst al. 2003). These studies typically have examined howpecies-specific PSFs can explain interspecific variation, e.g.etween dominant and rare plant species. However, severaltudies have also argued that for a single plant species, thetrength of the PSF may be density-dependent, and henceay depend on the local density of the species in the field.ecently, a large database analysis of more than 200,000 for-st plots and including 151 tree species revealed that for mostpecies the establishment of seedlings was negatively affectedy the local abundance of this species (Johnson, Beaulieu,ever, & Clay 2012). These effects are likely mediated byensity-dependent effects on soil pathogens. For example,eedling mortality of the tropical tree species Sebastiana

ongicuspis increases at higher local densities of conspecificsnd this is caused by soil fungal pathogens (Bell, Freckleton,

Lewis 2006).

(di

Although most of this work has been done with treepecies, similar density-dependent PSF may be expectedn grasslands (Bever 1994; Klironomos 2002; Petermann,ergus, Turnbull, & Schmid 2008). Recently, van de Voorde,an der Putten, & Bezemer (2012) compared PSF of the earlyuccessional plant Jacobaea vulgaris in soil collected from0 fields where this species occurred in different densities,nd showed that indeed PSF was negatively related to the den-ity of this species in the field. Similarly, in a field experimentith sown and unsown plots, J. vulgaris grew much better in

oil collected from the sown plots where it was rare or evenbsent, than in soil from the unsown plots where it was presentt high densities in the field (Bezemer, Harvey, Kowalchuk,orpershoek, & van der Putten 2006a). However, in the latter

tudy not only the density of J. vulgaris, but also the den-ities of other plant species differed greatly between sownnd unsown plots, and several studies have shown that therere also strong heterospecific effects on PSF (Mangan et al.010; van de Voorde, van der Putten, & Bezemer 2011). Soar, studies that examined PSF of J. vulgaris have used mix-ures of soil samples taken from different locations within aeld or an experimental plot. The local density of J. vulgarisften varies greatly within a single field (Bezemer personalbservation). Whether PSF is related to the local density of. vulgaris within a single field has not been tested.

Most studies test biotic PSF in sterile bulk soil inocu-ated with a small quantity of live field soil (e.g. Bever 1994;ardol, Bezemer, & van der Putten 2006; Reinhart 2012). The

ationale is that soil biota are introduced through the inocu-um of live field soil, while the large amount of sterile bulkoil standardises soil nutrient levels across treatments (e.g.ardol et al. 2006; Troelstra, Wagenaar, Smant, & Peters001). Inoculation with field soil, rather than using pure fieldoil, prevents confounding effects of local differences in soilutrient availability that may exist in field soil, for exam-le due to local disturbance of the soil. This is particularlymportant in bioassays where the soil effects on plant per-ormance are tested directly without a conditioning phase

e.g. Bezemer et al. 2006a). However, using inoculated soiloes not represent the natural field condition, where plantsnteract with biotic as well as abiotic components of the soil

5 plied E

(aJen(2

gdceiviohtvsdcwftd

lJsdsgwbooawwimp

M

P

oAirdl

fivrcKWb(

S

Tnccu5Vwaoi(abvD(tno

l(dWodedsppgvpgwew

08 M. Kos et al. / Basic and Ap

Wardle et al. 2004a; Wardle, Walker, & Bardgett 2004b)nd soil resources are not homogenously distributed (Bliss,ones, Mitchell, & Mou 2002). Furthermore, the sterile soilnvironment may also facilitate the rapid increase of a limitedumber of fast-growing species of microorganisms in all soilsde Boer, Verheggen, Gunnewiek, Kowalchuk, & van Veen003).In this study we carry out a bioassay to examine whether

rowth of J. vulgaris is related to local variation in plantensity within a single field. Jacobaea vulgaris is a poorompetitor and seedlings require bare or disturbed soil tostablish (McEvoy, Rudd, Cox, & Huso 1993). The variationn local density of this species may therefore be related toariation in the openness of the vegetation or the seed densityn the seed bank, but could also be related to the availabilityf soil nutrients or to PSF. A number of greenhouse studiesave shown that J. vulgaris quickly develops a strong nega-ive PSF (Bezemer et al. 2006a; van de Voorde et al. 2011;an de Voorde et al. 2012), probably due to the build-up ofpecies-specific pathogenic fungi (Bezemer et al. 2006a; vane Voorde et al. 2012). If this negative PSF plays a signifi-ant role in the spatial dynamics of this species in the field,e may expect that plants will grow less well in soil collected

rom locations with high J. vulgaris densities than from loca-ions where the plant is not growing or is present only at lowensities.

In an area of 400 m × 400 m within a restoration grass-and, we selected four 8 m × 8 m plots with high density of. vulgaris and four with lower density. Furthermore, weelected four plots within an adjacent heathland that bor-ered the grassland but where J. vulgaris is absent. In 16ubplots in each plot we recorded characteristics of J. vul-aris, of the vegetation, and the chemistry of the soil. Soilas collected from these subplots and used in a greenhouseioassay to examine the effects of plant density on the growthf J. vulgaris. Plants were grown either in pure field soil,r in sterilised bulk soil inoculated with a relatively smallmount of live field soil to standardise soil nutrient levelshile introducing soil biota. We hypothesised (i) that thereould be a negative relationship between J. vulgaris density

n the field and plant growth in bioassay and (ii) that the soil-ediated density effects would be less strong when using

ure field soil due to confounding effects of soil nutrients.

aterials and methods

lant species

Ragwort, Jacobaea vulgaris Gaertn. spp. vulgaris (syn-nym Senecio jacobaea L.), is a plant species in thesteraceae family that is native to Europe, but is invasive

n other continents (Bain 1991; Wardle 1987). In its nativeange, it can become very abundant in disturbed areas (vane Voorde et al. 2012). In The Netherlands, it is particu-arly abundant in nature restoration areas on former arable

peps

cology 14 (2013) 506–514

elds, and along roadsides (Bezemer et al. 2006a). Jacobaeaulgaris is a biannual plant that spends its first year as aosette, and flowers during the second year, but floweringan be delayed for several years by herbivory Prins, Vrieling,linkhamer, & de Jong 1990; van der Meijden and van deraals-Kooi 1979. The plant reproduces mostly by seeds,

ut can also reproduce vegetatively via root or crown budsHarper and Wood 1957; Wardle 1987).

tudy area

The study was carried out at the Planken Wambuis, Veluwe,he Netherlands (52◦04′ N, 5◦44′ E). This area consists ofatural grassland, heathland, forest and small patches of agri-ultural land. The natural areas are grazed by free roamingattle and horses. For this study two adjacent fields weresed with different vegetation types. Both fields were roughly0 ha in size (700 m × 750 m). The first field, Mosselscheeld, is a restoration grassland where agricultural practicesere ceased in 1985. Species richness in this area was on aver-

ge 15 plant species per m2. Species with an average coverf 5% or more were: Holcus lanatus (21%), Agrostis cap-llaris (13%), Plantago lanceolata (13%), Lolium perrenne10%), Achillea millefolium (7%), Jacobaea vulgaris (6%)nd Moehringia trinervia (5%). The adjacent field, Valen-erg, is a traditional heathland that is dominated by Callunaulgaris (Scotch heather) and grasses (Molinia caerulea andeschampsia flexuosa). Both sites had a sandy loam soil

60% coarse sand, 28% fine sand, 4% silt, 8% clay). Evenhough the heathland borders the grassland, J. vulgaris wasot present in the heathland. Within the grassland the densityf J. vulgaris varied spatially.

In July 2010, within an area of 400 m × 400 m of the grass-and, four areas with a high density of J. vulgaris plantsHigh-density; 11.3 ± 1.4 plants m−2) and four with a lowerensity (Low-density; 2.7 ± 0.7 plants m−2) were selected.ithin each of these eight areas a plot of 8 m × 8 m was laid

ut. The distance between two plots was at least 80 m. Theseimensions were selected so that the plot fitted well withinach local High- or Low-density area. Each plot was sub-ivided into 16 sub-plots of 2 m × 2 m. In each of the 16ub-plots, the number of rosettes and flowering J. vulgarislants was recorded. In the central 0.5 m × 0.5 m of each sub-lot, % cover of J. vulgaris, % cover of grasses, and % bareround was estimated, and height of the tallest flowering J.ulgaris plant was recorded. There were also four 8 m × 8 mlots in the heathland where % cover of grasses and % bareround was recorded in the 16 subplots. Jacobaea vulgarisas absent in the heathland. In the central 0.5 m × 0.5 m of

ach sub-plot, four soil cores (5 cm diameter, 15 cm depth)ere collected. The soil samples from each sub-plot were

ooled so that there was about 1.2 kg of soil collected fromach sub-plot. In total, there were 192 soil samples (12lots × 16 sub-plots). In the laboratory each soil sample wasieved (0.5 cm mesh size), homogenised and soil samples

plied E

wf

S

baps1sscg(blofe

S

esslestm(pwtsstgsiddwawPw

S

p

(p1ncaatcmi

S

WegeMtwstdtptoaiidpgwHposaABsfiaewbys

M. Kos et al. / Basic and Ap

ere then used for a seed bank experiment, a bioassay andor chemical analysis.

eed bank experiment

To determine the density of J. vulgaris seeds in the seedank in each of the 12 plots, 100 g of soil from each of fourdjacent sub-plots (making up one quarter of the plot), wasooled and homogenised so that there were four replicateoil samples per plot. Plastic containers (12.5 cm × 12.5 cm,3 cm height) with a layer of 2 cm (65 g) sterilised pottingoil to improve drainage, were filled with 250 g pooled fieldoil. In total, there were 48 containers (12 plots × 4 repli-ate soil samples). The containers were placed randomly in areenhouse at 70% RH and a 16 h light (21 ◦C) and 8 h dark16 ◦C) photo regime. Natural day–light was supplementedy 400 W metal halide lamps (225 �mol m−2 s−1 PAR, 1amp per 1.5 m2). For a period of eight weeks, the numberf J. vulgaris seedlings and the number of seedlings of otherorbs and grasses were recorded once every two weeks. Afterach recording, all seedlings were removed.

oil bioassay

In a greenhouse pot-experiment we grew J. vulgaris inach of the 192 soils. Pots (0.8 l) were filled with 700 goil. For each sub-plot, there was one pot with pure fieldoil and one pot with a mixture of sterilised bulk soil andive field soil (6:1). Plants growing in inoculated soil werexposed to comparable levels of available nutrients. Bulkoil was collected from a nearby restoration grassland inhe Planken Wambuis area. The bulk soil was sieved (0.5 cm

esh size), homogenised and sterilised by gamma irradiation>25 K Gray, Isotron, Ede, The Netherlands). Five additionalots were filled with 700 g sterilised bulk soil so that thereere in total 389 pots (192 sub-plots × 2 soil-types + 5 con-

rol pots). Jacobaea vulgaris seeds were collected from aingle population at the Planken Wambuis. The seeds wereurface sterilised (1 min in 0.5% sodium hypochlorite solu-ion and rinsed with water afterwards) and germinated onlass beads in a climate chamber at 20 ◦C. One one-week-oldeedling was planted in each pot. Pots were placed randomlyn the greenhouse and plants were grown under the same con-itions as described above. Seedlings that died within sevenays were replaced once. Plants were watered three times pereek and the soil was kept at 17% moisture. Seven weeks

fter transplanting, the aboveground biomass of each plantas clipped and roots were carefully washed from the soil.lant material was oven-dried (70 ◦C) and root and shoot dryeight of each plant was determined.

oil chemistry

A sub-sample of 100 g soil of each of the four corner sub-lots of each plot was sieved (0.5 cm mesh size) and dried

afgp

cology 14 (2013) 506–514 509

72 h at 40 ◦C). In this sub-sample, pH, phosphorous (P),otassium (K+) and magnesium (Mg2+) were analysed in:10 (w/v) 0.01 M CaCl2. Concentrations of available ammo-ium (NH4

+–N) and nitrate (NO3−–N) were determined

olorimetrically in the CaCl2-extract using a Traacs 800utoanalyzer (TechniCon Systems Inc.). Total extractablemounts of phosphorous (Olsen-P) were determined usinghe method of Olsen, Cole, Watanabe, & Dean (1954) andolour intensity was measured at 720 nm. The % organicatter was determined as weight loss of a soil sample after

gnition at 430 ◦C for 24 h.

tatistical analysis

All analyses were performed in IBM SPSS Statistics forindows (19th edition, SPSS Inc., Chicago, Illinois, USA),

xcept when indicated otherwise. Soil chemistry, J. vul-aris and vegetation characteristics and number of seedlingsmerging form the seed bank were analysed using Mixedodels with density (High- or Low-density) as a fixed fac-

or and plot identity as a random factor. In this way subplotsithin each plot were included in the analyses but were con-

idered as pseudoreplicates. Similar Mixed Models were usedo analyse the difference in soil chemistry between High-ensity, Low-density and C. vulgaris plots. Prior to analysis,he number of rosettes, the number of emerged seedlings,H, and concentrations of P, K+, NO3

− and NH4+ were log-

ransformed; % J. vulgaris, % grass cover, % bare soil, and %rganic matter were arcsin-square root transformed to fulfilssumptions of normality. Plant biomass and root/shoot ration the bioassay was analysed seperately for pure soil and fornoculated soil using Mixed Models with soil origin (High-ensity, Low-density or C. vulgaris) as a fixed factor andlot identity as a random factor. Because none of the plantsrown in pure C. vulgaris soil survived, the Mixed Modelith data from plants grown in pure soil was carried out forigh- and Low-density soils only. Individual pair-wise com-arisons were based on LSD tests. To test the “pure” effectf plant density in pure soil, a second Mixed Model analy-is was performed in which nutrient effects were includeds a covariable. First, a multivariate Principal Componentnalysis (PCA) was performed in Canoco version 4.55 (Terraak & Smilauer 2002) using the soil chemistry variables as

pecies data. Second, the values of the sample scores on therst PCA axis (explaining 46% of the variation) were used as

fixed covariate in the Mixed Model. The means and standardrrors in High- and Low-density soil predicted by the modelere obtained. One-way ANOVA was used to compare plantiomass in pure, inoculated and sterile soil. For this anal-sis we used mean values per plot for pure and inoculatedoil. The average of pure soil was based on data from High-

nd Low-density plots only. A post hoc Tukey test was usedor pair-wise comparisons. The relationship between J. vul-aris density in each subplot and soil chemistry, and betweenlant biomass in the bioassay and J. vulgaris density or soil

5 plied E

cBLad

R

F

sFrppdhdcwnbd(

btavd

bpt%dJbe

S

piPh

T(

C

V

S

S

M

10 M. Kos et al. / Basic and Ap

hemistry in the field was tested using linear regression.ecause of the experimental set-up, in which the High- andow-density plots were selected a priori, linear regressionnalyses were performed separately for High- and Low-ensity plots.

esults

ield observations and seed bank experiment

The percentage of J. vulgaris plants that was in the rosettetage was higher in High- than in Low-density plots (Table 1).or both High- and Low-density plots, there was a positiveelationship between the number of rosettes and floweringlants, but this relationship was stronger in Low-densitylots (High-density: F1,62 = 5.33, P = 0.024, R2 = 0.08; Low-ensity: F1,62 = 24.56, P < 0.001, R2 = 0.28). The maximumeight of flowering plants was larger in High- than in Low-ensity plots (Table 1), even when only the sub-plots thatontained a single flowering plant in the central 0.5 m × 0.5 mere included in the analysis (F1,7 = 6.44, P = 0.040). The

umber of J. vulgaris seedlings emerging from the seedank was also higher in High-density plots. High- and Low-ensity plots did not differ in % of grass cover or bare soilTable 1). Overall, soil abiotic characteristics did not differ

(

Cf

able 1. Field vegetation characteristics, the number of seedlings per comean ± SE) in field plots containing high- and low-densities of J. vulgari

haracteristic High-density

egetationJ. vulgaris plants (no. m−2) 11.3 ± 1.4

Rosettes (no. m−2) 6.6 ± 1.3

Flowering plants (no. m−2) 4.6 ± 0.4

% Rosettes 55.1 ± 5.1

Max. J. vulgaris height (cm) 60.6 ± 4.5

J. vulgaris cover (%) 11.6 ± 1.5

Grass cover (%) 60.4 ± 3.2

Bare soil (%) 28.7 ± 2.2

eed bankJ. vulgaris seedlings (no. per cont.)b 1.0 ± 0.2

Other forb seedlings (no. per cont.) 16.2 ± 2.6

Grass seedlings (no. per cont.) 2.9 ± 1.4

oil chemistrypH CaCl2 4.60 ± 0.04

P CaCl2 (mg kg−1) 1.53 ± 0.24

K+ CaCl2 (mg kg−1) 50.74 ± 6.85

Mg2+ CaCl2 (mg kg−1) 43.49 ± 1.82

NH4+ CaCl2 (mg kg−1) 3.25 ± 0.31

NO3− CaCl2 (mg kg−1) 3.07 ± 0.81

Olsen-P (mg kg−1) 82.30 ± 6.01Organic matter (%) 4.63 ± 0.33

eans are shown based on the average value of each plot. Statistical analyses wereaShown in subscript after the F-value are the numerator-df and the denominator-bStatistical analysis performed by one-way ANOVA instead of Mixed Model.

cology 14 (2013) 506–514

etween High- and Low-density plots (Table 1). However,here was a high level of heterogeneity in soil chemical char-cteristics within the experimental field, and soil chemistryaried greatly among the four High-density and the four Low-ensity plots (see Appendix A: Table 1).

In High-density plots, there was a positive relationshipetween the density of J. vulgaris plants in a subplot andH (F1,14 = 14.32, P = 0.002, R2 = 0.51), but a negative rela-ionship with Olsen-P (F1,14 = 6.18, P = 0.026, R2 = 0.31) and

organic matter (F1,14 = 7.60, P = 0.015, R2 = 0.35). In Low-ensity plots, there were no significant relationships between. vulgaris density and soil chemistry. There was no relationetween bare soil or grass cover and J. vulgaris density inither High- or Low-density plots.

oil bioassay

On average, biomass of J. vulgaris was highest whenlants were grown in sterilised control soil, intermediate innoculated soil, and lowest in pure field soil (F2,22 = 48.31,

< 0.001, Fig. 1A). Root/shoot ratios were significantlyigher in pure soil than in inoculated or in sterile soil

F2,22 = 20.21, P < 0.001, Fig. 1B).

In pure soil, none of the J. vulgaris plants planted in. vulgaris soil survived, and plant biomass did not dif-

er between plants grown in soil collected from High- and

ntainer (cont.) germinating from the seed bank and soil chemistrys plants.

Low-density Fa P

2.7 ± 0.7 F1,6 = 31.09 0.0011.1 ± 0.3 F1,6 = 29.23 0.0021.7 ± 0.3 F1,6 = 31.71 0.001

29.5 ± 4.5 F1,6 = 12.77 0.01242.6 ± 2.7 F1,8 = 8.82 0.019

4.7 ± 1.1 F1,6 = 19.78 0.00464.2 ± 3.8 F1,6 = 0.48 0.51728.2 ± 3.6 F1,6 < 0.01 0.998

0.1 ± 0.1 F1,30 = 19.94 <0.00114.8 ± 2.4 F1,6 = 0.19 0.677

3.6 ± 0.8 F1,6 = 0.48 0.513

4.57 ± 0.03 F1,6 = 0.24 0.6421.46 ± 0.33 F1,6 = 0.03 0.867

32.51 ± 5.33 F1,6 = 5.75 0.05436.06 ± 3.22 F1,6 = 4.05 0.091

3.02 ± 0.27 F1,6 = 0.32 0.5944.60 ± 1.10 F1,6 = 1.33 0.293

84.56 ± 5.89 F1,6 = 0.07 0.7963.98 ± 0.43 F1,6 = 1.50 0.267

performed by Mixed Models.df of the Mixed Model, rounded off to integers.

M. Kos et al. / Basic and Applied Ecology 14 (2013) 506–514 511

Fig. 1. Mean (A) total plant biomass and (B) root/shoot ratio (±SE,based on the average values per plot) of Jacobaea vulgaris plantsin the bioassay grown in pure field soil (‘Pure soil’), in sterilisedbulk soil inoculated with live field soil (‘Inoculated soil’) or in ster-ile soil. The field soil was collected from four plots with high J.vulgaris density (‘High-density plots’), four with low J. vulgarisdensity (‘Low-density plots’) or from four plots dominated by Cal-luna vulgaris from which J. vulgaris was absent (‘C. vulgaris plots’).Jacobaea vulgaris did not grow in pure C. vulgaris soil. Signifi-cant differences are indicated by different letters, with capital lettersdenoting pair-wise differences among the three soil types (pure soil,inoculated soil and sterile soil) based on one-way ANOVA, and low-ercase letters denoting pair-wise differences among the three soiloM

Ltitisdpw(

Fig. 2. Relationship between Jacobaea vulgaris density in the fieldand its growth in the greenhouse in field-collected soil. (A) Relation-ship between the number of plants per m2 and J. vulgaris biomassin pure field soil collected from plots with high J. vulgaris den-sity (F1,61 = 10.83, P = 0.002, R2 = 0.15). (B) Relationship betweenthe number of plants per m2 and J. vulgaris biomass in pure fieldsoil collected from plots with low J. vulgaris density (F1,61 = 7.90,P 2

o

uiwRv(

wair

rigins (High-density, Low-density and C. vulgaris plots) based onixed Models (P < 0.05).

ow-density plots (F1,6 = 0.02, P = 0.904, Fig. 1A). Whenhe effect of soil nutrients in pure soil was removed byncluding it as a covariable in the analysis, plant biomassended to be lower in High-density soil (0.71 ± 0.07 g) thann Low-density soil (0.81 ± 0.06 g), but the difference was notignificant (F1,6 = 4.12, P = 0.089). Root/shoot ratios did notiffer significantly between plants grown in pure High- and

ure Low-density soil (F1,6 = 4.25, P = 0.085, Fig. 1B), alsohen the effects of soil nutrients were removed statistically

F1,5 = 0.10, P = 0.763).

bpm

= 0.007, R = 0.11). The line is the estimated relationship basedn linear regression analysis.

In inoculated soil, plants grew best when soil was inoc-lated with C. vulgaris soil, significantly less well in soilnoculated with Low-density soil, and worst in soil inoculatedith High-density soil (F2,9 = 13.87, P = 0.002, Fig. 1A).oot/shoot ratios were significantly higher in inoculated C.ulgaris soil than in inoculated High- or Low-density soilF2,9 = 4.90, P = 0.008, Fig. 1B).

The biomass of plants grown in pure High-density soilas negatively related to the density of J. vulgaris plants in

subplot (Fig. 2A). In contrast, the biomass of plants grownn pure soil collected from Low-density plots was positivelyelated to the J. vulgaris density in the field (Fig. 2B). Plant

iomass in pure soil collected from High-density plots wasositively related to P, K+, NO3

−, Olsen-P and % organicatter (see Appendix A: Table 2). In pure soil collected from

5 plied E

LNuvbbsA

D

itawJwtougocmpmbJtKtif(pOeiw

io2iiwddetlipd

dastepensvnst(b(vt

gabpiba

tciislh

Co1ppsrftp

dgnwtL

12 M. Kos et al. / Basic and Ap

ow-density plots, plant biomass was positively related toO3

−. Neither in High-density nor in Low-density inoc-lated soil, there was a relationship between density of J.ulgaris in a subplot and plant biomass in the bioassay. Plantiomass in inoculated soil was not related to soil nutrients,ut for High-density plots there was a negative relation-hip between biomass and the pH of the inoculated soil (seeppendix A: Table 2).

iscussion

Our study shows that there is local variation in plant–soilnteractions for J. vulgaris and that this variation is relatedo plant density in the field. In the bioassay, J. vulgaris had

negative PSF. Biomass of plants grown in inoculated soilas lower when field soil originated from plots with a high

. vulgaris density than when soil was collected from plotsith a low density. Recently, Bezemer et al. (2013) showed

hat fungal pathogens such as Phoma exigua and Fusariumxysporum, which occur at our field site as well (Bezemer,npublished results), accumulate around the roots of J. vul-aris. These pathogens may subsequently inhibit the growthf the plant. Although we did not analyse the microbial soilommunity, our results suggest that a higher density of plantsay lead to a larger build-up of soil pathogens, and that these

athogens subsequently negatively affect plant growth. Thisechanism of conspecific negative density-dependence has

een shown mostly for tree species (e.g. Bell et al. 2006;ohnson et al. 2012), although several studies have suggestedhat this may also play a role in grasslands (e.g. Bever 1994;lironomos 2002; Petermann et al. 2008). Our results show

hat, indeed, this mechanism also occurs in grasslands. Sim-lar results have been shown when densities of J. vulgarisrom different fields that were located several kilometres apartvan de Voorde et al. 2012) or from different experimentallant communities (Bezemer et al. 2006a) were compared.ur study now shows that such negative density-dependent

ffects can also be detected on a much smaller spatial scale,.e. within a single grassland area of only 400 m × 400 m,here J. vulgaris densities varied spatially.We studied PSF in J. vulgaris not only in sterile bulk soil

noculated with a small part of live field, similar to manyther PSF studies (e.g. Bezemer et al. 2006a; Kardol et al.006; Reinhart 2012; van de Voorde et al. 2012), but alson pure field soil to examine the strength of these plant–soilnteractions under more natural conditions. When J. vulgarisas grown in pure field soil in the pot-experiment, biomassid not differ between soils collected from High- and Low-ensity plots. This is in line with our expectation that densityffects mediated by soil biota are less strong in pure field soilhan in inoculated soil because of the confounding effects of

ocal differences in nutrient availability in the soil. Interest-ngly, when the effects of nutrients were statistically removed,lant biomass tended to be lower in pure soil from High-ensity plots compared to Low-density plots, although this

ooie

cology 14 (2013) 506–514

ifference was not significant. Moreover, in the regressionnalysis, biomass of plants grown in the greenhouse in pureoil collected from High-density plots was negatively relatedo the local density of J. vulgaris plants in the field. Hence,ven though soil nutrients greatly affected plant growth inure soil, we were still able to detect a density-dependentffect of J. vulgaris on plant growth. We did not find such aegative relationship between plant biomass and plant den-ity in Low-density soils. We propose that the number of J.ulgaris plants in Low-density plots was too low to cause aegative PSF. It is important to note that the identity and den-ity of other plant species may also have differed spatially inhe field. Heterospecific plants can cause strong PSF effectsMangan et al. 2010). The growth of J. vulgaris can alsoe greatly affected by soil-mediated effects of other speciesvan de Voorde et al. 2011) and the growth responses of J.ulgaris in the greenhouse may have been partly caused byhe soil-mediated effects of other species.

It is well known that soil nutrients positively affect therowth of J. vulgaris (Hol, Vrieling, & van Veen 2003)nd we also observed a positive relationship between plantiomass and soil nutrients in both High- and Low-densityure soil. However, soil nutrient levels did not differ signif-cantly between the High- and Low-density plots, probablyecause soil nutrient levels varied greatly within and amongll plots.

Jacobaea vulgaris biomass was smaller in pure field soilhan in inoculated soil. This may suggest that pure soilontains higher levels of pathogenic microorganisms thannoculated soil, but may also result from the higher availabil-ty of soil nutrients in inoculated soil due to the process ofterilisation by gamma irradiation (Troelstra et al. 2001). Theower root/shoot ratio in inoculated soil supports the latterypothesis.

None of the plants survived in pure soil collected from. vulgaris plots, but this is not unexpected as J. vulgarisnly occurs in soils with a pH above 3.95 (Harper and Wood957). Indeed, when a small amount of soil from C. vulgarislots was inoculated in sterilised grassland soil and hence soilH was higher, plants grew very well and biomass was evenimilar to that of plants grown in entirely sterilised soil. Thisesult emphasizes the important role of soil abiotic conditionsor the growth of this plant species. Moreover, it suggestshat soil organisms that are pathogenic to J. vulgaris are notresent in C. vulgaris soil.

Although we found a negative relationship between plantensity in the field and plant growth in the bioassay, J. vul-aris plants in the field did not yet appear to suffer from theegative density-dependent effect. Instead, flowering plantsere significantly taller in High-density plots, suggesting that

hese plots were more suitable for J. vulgaris growth thanow-density plots. It is possible that, by measuring the height

f only the tallest flowering plant in the central 0.5 m × 0.5 mf a sub-plot, our data is biased towards finding taller plantsn plots with more plants, i.e. the High-density plots. How-ver, this is not likely, as there was no relation between the

plied E

neott

wdetWtHAv1or

vavafpfisOvsp

A

mfWptto

A

i2

R

A

B

B

B

B

B

B

B

B

B

C

d

E

H

H

J

K

M. Kos et al. / Basic and Ap

umber of flowering plants and the height of the tallest flow-ring plant (data not shown). Furthermore, when we includednly the sub-plots that contained a single flowering plant inhe analysis, flowering plants were still significantly taller inhe High-density plots.

The density and the percentage of first-year rosette plantsere higher in High- than in Low-density plots. This could beirectly related to the larger number of J. vulgaris seedlingsmerging from the seed bank, which in turn, could be relatedo the higher density of flowering plants that produce seeds.

e do not know whether seedling establishment was nega-ively affected by plant density, nor whether the rosettes inigh-density plots were smaller than in Low-density plots.lthough the spatial patterning of J. vulgaris is related toariation in the openness of the vegetation (McEvoy et al.993), we found no difference in the percentage of bare soilr grass cover between High- and Low-density plots, nor aelation between bare soil or grass cover and plant density.

In conclusion, we show that plant–soil interactions of J.ulgaris vary on a local scale, i.e. within a single grassland,nd that this variation is related to the density of plants in theegetation. However, other factors such as soil nutrient avail-bility and seed density in the seed bank are also importantor the spatial patterning of J. vulgaris, and these multiplerocesses should be considered simultaneously. Plant–soileedback is a rapidly developing research field, and it isncreasingly acknowledged that PSFs may operate at smallpatial scales (Brandt, de Kroon, Reynolds, & Burns 2013).ur results show that it is important to include local spatialariation in plant–soil interactions to increase our under-tanding of the role of these interactions in affecting planterformance and community dynamics.

cknowledgements

We thank two anonymous reviewers for constructive com-ents on an earlier version of the manuscript; Wiecher Smant

or soil chemical analyses and Natuurmonumenten (Plankenambuis) for giving permission to perform this study on their

roperty. This work was funded by the Netherlands Organiza-ion of Scientific Research (NWO, VIDI grant no. 864.07.009o TMB). This is publication 5477 of the Netherlands Institutef Ecology (NIOO-KNAW).

ppendix A. Supplementary data

Supplementary material related to this article can be found,n the online version, at http://dx.doi.org/10.1016/j.baae.013.07.002.

eferences

erts, R., & Chapin, F. S. (2000). The mineral nutrition of wildplants revisited: A re-evaluation of processes and patterns. In

K

cology 14 (2013) 506–514 513

A. H. Fitter, & D. G. Raffaelli (Eds.), Advances in ecologicalresearch (30) (pp. 1–67). San Diego: Elsevier Academic PressInc.

ain, J. F. (1991). The biology of Canadian weeds 96:Senecio jacobaea L. Canadian Journal of Plant Science, 71,127–140.

ell, T., Freckleton, R. P., & Lewis, O. T. (2006). Plant pathogensdrive density-dependent seedling mortality in a tropical tree.Ecology Letters, 9, 569–574.

ever, J. D. (1994). Feedback between plants and their soil com-munities in an old field community. Ecology, 75, 1965–1977.

ever, J. D., Westover, K. M., & Antonovics, J. (1997). Incor-porating the soil community into plant population dynamics:The utility of the feedback approach. Journal of Ecology, 85,561–573.

ezemer, T. M., Harvey, J. A., Kowalchuk, G. A., Korpershoek,H., & van der Putten, W. H. (2006). Interplay between Seneciojacobaea and plant, soil, and aboveground insect communitycomposition. Ecology, 87, 2002–2013.

ezemer, T. M., Lawson, C. S., Hedlund, K., Edwards, A. R., Brook,A. J., Igual, J. M., et al. (2006). Plant species and functional groupeffects on abiotic and microbial soil properties and plant–soilfeedback responses in two grasslands. Journal of Ecology, 94,893–904.

ezemer, T. M., van der Putten, W. H., Martens, H., van de Voorde,T. F. J., Mulder, P. P. J., & Kostenko, O. (2013). Above- andbelow-ground herbivory effects on below-ground plant-fungusinteractions and plant–soil feedback responses. Journal of Ecol-ogy, 101, 325–333.

liss, K. M., Jones, R. H., Mitchell, R. J., & Mou, P. P. (2002).Are competitive interactions influenced by spatial nutrient het-erogeneity and root foraging behavior? New Phytologist, 154,409–417.

randt, A. J., de Kroon, H., Reynolds, H. L., & Burns, J. H.(2013). Soil heterogeneity generated by plant–soil feedbacks hasimplications for species recruitment and coexistence. Journal ofEcology, 101, 277–286.

hapman, S. K., Langley, J. A., Hart, S. C., & Koch, G. W. (2006).Plants actively control nitrogen cycling: Uncorking the microbialbottleneck. New Phytologist, 169, 27–34.

e Boer, W., Verheggen, P., Gunnewiek, P., Kowalchuk, G. A., & vanVeen, J. A. (2003). Microbial community composition affectssoil fungistasis. Applied and Environmental Microbiology, 69,835–844.

hrenfeld, J. G., Ravit, B., & Elgersma, K. (2005). Feedbackin the plant–soil system. Annual Review of Environment andResources, 30, 75–115.

arper, J. L., & Wood, W. A. (1957). Senecio Jacobaea L. Journalof Ecology, 45, 617–637.

ol, W. H. G., Vrieling, K., & van Veen, J. A. (2003). Nutri-ents decrease pyrrolizidine alkaloid concentrations in Seneciojacobaea. New Phytologist, 158, 175–181.

ohnson, D. J., Beaulieu, W. T., Bever, J. D., & Clay, K. (2012).Conspecific negative density dependence and forest diversity.Science, 336, 904–907.

ardol, P., Bezemer, T. M., & van der Putten, W. H. (2006). Tempo-ral variation in plant–soil feedback controls succession. EcologyLetters, 9, 1080–1088.

lironomos, J. N. (2002). Feedback with soil biota contributesto plant rarity and invasiveness in communities. Nature, 417,67–70.

5 plied E

K

M

M

O

P

P

P

R

R

T

T

v

v

v

v

v

W

W

14 M. Kos et al. / Basic and Ap

ulmatiski, A., Beard, K. H., Stevens, J. R., & Cobbold, S. M.(2008). Plant–soil feedbacks: A meta-analytical review. EcologyLetters, 11, 980–992.

angan, S. A., Schnitzer, S. A., Herre, E. A., Mack, K. M. L.,Valencia, M. C., Sanchez, E. I., et al. (2010). Negative plant–soilfeedback predicts tree-species relative abundance in a tropicalforest. Nature, 466 [752-U710]

cEvoy, P. B., Rudd, N. T., Cox, C. S., & Huso, M. (1993). Dis-turbance, competition, and herbivory effects on ragwort Seneciojacobaea populations. Ecological Monographs, 63, 55–75.

lsen, S. R., Cole, C. V., Watanabe, F. S., & Dean, L. A. (1954).Estimation of available phosphorus in soils by extraction withsodium bicarbonate. USDA Circular, 139, 1–19.

acker, A., & Clay, K. (2000). Soil pathogens and spatial patternsof seedling mortality in a temperate tree. Nature, 404, 278–281.

etermann, J. S., Fergus, A. J. F., Turnbull, L. A., & Schmid,B. (2008). Janzen–Connell effects are widespread and strongenough to maintain diversity in grasslands. Ecology, 89,2399–2406.

rins, A. H., Vrieling, K., Klinkhamer, P. G. L., & de Jong, T.J. (1990). Flowering behavior of Senecio jacobaea: Effects ofnutrient availability and size-dependant vernalization. Oikos, 59,248–252.

einhart, K. O. (2012). The organization of plant communities: Neg-ative plant–soil feedbacks and semiarid grasslands. Ecology, 93,2377–2385.

eynolds, H. L., Packer, A., Bever, J. D., & Clay, K. (2003). Grass-roots ecology: Plant–microbe–soil interactions as drivers of plantcommunity structure and dynamics. Ecology, 84, 2281–2291.

er Braak, C. J. F., & Smilauer, P. (2002). CANOCO ReferenceManual and CanoDraw for Windows user’s Guide. Softwarefor Canonical Community Ordination, Version 4.5. Ithaca, NewYork: Microcomputer Power.

W

Available online at www.s

ScienceD

cology 14 (2013) 506–514

roelstra, S. R., Wagenaar, R., Smant, W., & Peters, B. A. M. (2001).Interpretation of bioassays in the study of interactions betweensoil organisms and plants: Involvement of nutrient factors. NewPhytologist, 150, 697–706.

an de Voorde, T. F. J., van der Putten, W. H., & Bezemer, T. M.(2011). Intra- and interspecific plant–soil interactions, soil lega-cies and priority effects during old-field succession. Journal ofEcology, 99, 945–953.

an de Voorde, T. F. J., van der Putten, W. H., & Bezemer, T. M.(2012). The importance of plant–soil interactions, soil nutri-ents, and plant life history traits for the temporal dynamics ofJacobaea vulgaris in a chronosequence of old-fields. Oikos, 121,1251–1262.

an der Meijden, E., & van der Waals-Kooi, R. E. (1979). Populationecology of Senecio jacobaea in a sand dune system. Journal ofEcology, 67, 131–153.

an der Putten, W. H., van Dijk, C., & Peters, B. A. M. (1993). Plant-specific soil-borne diseases contribute to succession in foredunevegetation. Nature, 362, 53–56.

an der Putten, W. H., Bardgett, R. D., Bever, J. D., Bezemer, T. M.,Casper, B. B., Fukami, T., et al. (2013). Plant–soil feedbacks:the past, the present and future challenges. Journal of Ecology,101, 265–276.

ardle, D. A. (1987). The ecology of ragwort (Senecio jacobaeaL) – A review. New Zealand Journal of Ecology, 10,67–76.

ardle, D. A., Bardgett, R. D., Klironomos, J. N., Setala, H., vander Putten, W. H., & Wall, D. H. (2004). Ecological linkagesbetween aboveground and belowground biota. Science, 304,1629–1633.

ardle, D. A., Walker, L. R., & Bardgett, R. D. (2004). Ecosystemproperties and forest decline in contrasting long-term chronose-quences. Science, 305, 509–513.

ciencedirect.com

irect