Ecology, 94(8), 2013, pp. 1878–1885� 2013 by the Ecological Society of America

Functionally and phylogenetically diverse plant communitieskey to soil biota

ALEXANDRU MILCU,1,2,18 ERIC ALLAN,3 CHRISTIANE ROSCHER,4,5 TANIA JENKINS,6 SEBASTIAN T. MEYER,7 DAN FLYNN,8

HOLGER BESSLER,9 FRANCOIS BUSCOT,10,11 CHRISTOF ENGELS,9 MARLEN GUBSCH,12 STEPHAN KONIG,10

ANNETT LIPOWSKY,8 JESSY LORANGER,13 CARSTEN RENKER,11 CHRISTOPH SCHERBER,14 BERNHARD SCHMID,8

ELISA THEBAULT,15 TESFAYE WUBET,10 WOLFGANG W. WEISSER,7 STEFAN SCHEU,16 AND NICO EISENHAUER17

1CNRS, Ecotron–UPS 3248, Campus Baillarguet, 34980, Montferrier-sur-Lez, France2Imperial College London, Division of Biology, Ascot SL5 7PY United Kingdom

3Institute of Plant Sciences, University of Bern, Altenbergrain 21, 3013 Bern, Switzerland4UFZ, Helmholtz Centre for Environmental Research, Department of Community Ecology, Halle, Germany

5Max Planck Institute for Biogeochemistry, Hans-Knoll Strasse 10, 07745 Jena, Germany6Department of Ecology and Evolution, University of Lausanne, Biophore-Quartier Sorge CH-1015 Lausanne, Switzerland7Technische Universitat Munchen, Department of Ecology and Ecosystem Management, Hans-Carl-von-Carlowitz-Platz 2,

85350 Freising, Germany8Institute of Evolutionary Biology and Environmental Studies, University of Zurich, Winterthurerstrasse 190,

CH-8057 Zurich, Switzerland9Humboldt-Universitat zu Berlin, Institute of Plant Nutrition, Albrecht-Thaer-Weg 4, 14195 Berlin, Germany

10UFZ, Helmholtz Centre for Environmental Research, Department of Soil Ecology, Theodor-Lieser-Straße 4, 06102 Halle, Germany11University of Leipzig, Institute of Biology, Johannis-Allee 21-23, 04103 Leipzig, Germany12Institute of Plant Sciences, ETH Zurich, Universitatsstrasse 2, 8092 Zurich, Switzerland

13CNRS CEFE, UMR 5175, 1919 Route de Mende, 34293, Montpellier, France14Georg-August-University Gottingen, Agroecology, Department of Crop Sciences, Grisebachstrasse 6, 37077 Gottingen, Germany

15CNRS, UMR 7618, Laboratoire ‘‘Biogeochimie et ecologie des milieux continentaux,’’ 46 Rue d’Ulm, 75005 Paris, France16Georg-August-University, J.F. Blumenbach Institute of Zoology and Anthropology, Berliner Strasse 28, 37073 Gottingen, Germany

17Friedrich-Schiller-University, Institute of Ecology, Dornburger Strasse 159, 07743 Jena, Germany

Abstract. Recent studies assessing the role of biological diversity for ecosystemfunctioning indicate that the diversity of functional traits and the evolutionary history ofspecies in a community, not the number of taxonomic units, ultimately drives the biodiversity–ecosystem-function relationship. Here, we simultaneously assessed the importance of plantfunctional trait and phylogenetic diversity as predictors of major trophic groups of soil biota(abundance and diversity), six years from the onset of a grassland biodiversity experiment.Plant functional and phylogenetic diversity were generally better predictors of soil biota thanthe traditionally used species or functional group richness. Functional diversity was a reliablepredictor for most biota, with the exception of soil microorganisms, which were betterpredicted by phylogenetic diversity. These results provide empirical support for the idea thatthe diversity of plant functional traits and the diversity of evolutionary lineages in acommunity are important for maintaining higher abundances and diversity of soilcommunities.

Key words: above–belowground interactions; biodiversity; functional diversity; functional traits; JenaExperiment; phylogenetic diversity; plant species richness; soil fauna.

INTRODUCTION

Linking changes in community composition and

diversity between trophic levels presents a major

challenge for community and ecosystem ecology (Van

der Putten et al. 2001, Wardle et al. 2004, Haddad et al.

2009). Particularly, understanding the links between

above- and belowground communities has emerged as

an important challenge given that soil biota are not just

a ‘‘black box’’ of highly redundant species and that they

drive a range of ecosystem functions (Scheu and Setala

2002, Wardle et al. 2004). We are increasingly learning

that soil biota are closely interlinked with aboveground

communities and that there is a greater degree of

specificity between plants and soil organisms than was

previously assumed (Scheu 2001, Wardle et al. 2004,

Manuscript received 5 November 2012; revised 30 January2013; accepted 25 February 2013. Corresponding Editor: B. A.Wardle.

18 E-mail: a.milcu@imperial.ac.uk

NOTES1878 Ecology, Vol. 94, No. 8NOTES1878 Ecology, Vol. 94, No. 8

Bardgett and Wardle 2010). There is compelling

evidence that soil biota are responsive to the quality

and quantity of organic matter inputs as well as to

changes in micro-environmental conditions associated

with changes in plant diversity (Wardle et al. 2004,

Bardgett and Wardle 2010). Despite this, a large body of

literature suggests that soil biota may be less sensitive to

changes in plant diversity than aboveground biota

(Gastine et al. 2003, Scherber et al. 2010). However,

this conclusion was mainly based on short-term studies

that investigated only two facets of plant diversity, i.e.,

species and functional group richness (Eisenhauer et al.

2012).

Recent studies indicate that the diversity of functional

traits or the evolutionary history of a community, not

the number of taxonomic units, ultimately drives

biodiversity–ecosystem-functioning relationships (Ca-

dotte et al. 2009, Flynn et al. 2011). A trait is any

morphological, biochemical, behavioral, and phenolog-

ical characteristic of an individual that potentially

affects its performance and fitness (Petchey and Gaston

2002). Identifying the most relevant functional traits

underpinning the biodiversity–ecosystem-functioning

relationship can be challenging, with the results sensitive

to the number and choice of traits included in the

analyses (Petchey and Gaston 2006). Given these

potential limitations of trait-based approaches, phylo-

genetic diversity, the sum of the shared evolutionary

history in a community, has been proposed as a useful

proxy to describe the true functional diversity of a

community (Cadotte et al. 2009). Phylogenetic diversity

should affect ecosystem functioning if ecological dissim-

ilarity is correlated with evolutionary divergence,

meaning that the more phylogenetically divergent

species are present, the more likely it is that they have

dissimilar functional traits and occupy different niches,

thereby differentially impacting ecosystem functioning

(Felsenstein 1985, Maherali and Klironomos 2007).

Additionally, plant phylogenetic diversity may be

particularly important for higher trophic levels, and

more than just a proxy for functional diversity if plant

phylogeny reflects coevolutionary interactions between

plants and other organism groups (Dinnage et al. 2012).

Lately, functional and phylogenetic diversity have

been shown to be better predictors of primary produc-

tivity (Cadotte et al. 2009, Clark et al. 2012) and

arthropod diversity and abundances than were plant

species or functional group richness (Dinnage et al.

2012). However, we have remarkably little empirical

evidence on whether these indices are superior predictors

of soil biota than species and functional group richness.

In addition, information on the relevance of different

plant diversity indices may contribute to a better

understanding of how plant diversity effects cascade

into belowground food webs. Here we assess the

performance of several plant functional and phyloge-

netic diversity metrics, alongside more conventional

metrics of plant community composition such as

realized plant species richness (Rdiv), functional group

richness (FG) and functional group biomass as deter-

minants of soil biota using data from one of the most

comprehensive biodiversity experiments so far, the Jena

Experiment (Roscher et al. 2004). We focused on the

abundances and diversity of soil biota collected from 82

grassland plots with experimentally manipulated plant

species (1, 2, 4, 8, 16, and 60) and functional group

richness (1, 2, 3, and 4), measured on the sixth year from

the onset of the diversity treatments.

METHODS

Study site and experimental design.—The experimental

site (508550 N, 118350 E, 130 m above sea level; mean

annual temperature 9.38C, mean annual precipitation

587 mm) was a former arable field located on the

floodplain of the Saale River, Jena, Germany. The

number of plant species, plant functional groups and

plant identity is controlled, in a randomized four block

design comprising 82 plots of 20 3 20 m. Plots were

established in May 2002 with 1, 2, 4, 8, 16, or 60

perennial grassland plant species typical for local

Arrhenatherum grasslands, with 16, 16, 16, 16, 14, and

4 replicates, respectively (see Roscher et al. [2004] for

details on experimental design). Plant compositions in

the plots were randomly chosen from a pool of 60

species and maintained by a combination of biannual

mowing, weeding, and herbicide applications.

Functional diversity estimated from traits.—We select-

ed 12 plant functional traits, based on literature-

informed knowledge, that affect soil biota and processes

through the quality and consistency of plant-derived

organic inputs as well as through changes of microhab-

itat environmental conditions. For each of the 60 plant

species, the traits were derived from in situ measure-

ments (shoot biomass dry mass [mg], biomass-to-N ratio

[mg N/g], shoot lignin and hemicelluloses content [%],

shoot C-to-N ratio, seed mass [mg], leaf area ratio [mm2/

mg dry mass], ability to fix atmospheric N2 [binary]) and

literature surveys (seasonality of foliage [ordinal; 1,

summer green; 2, partly evergreen; 3, evergreen], number

of known secondary compounds, rooting type [ordinal;

1, long-living primary root system; 2, secondary fibrous

roots in addition to the primary root system; 3, short-

living primary root system, extensive secondary root

system] and rooting depth [cm] as used by Roscher et al.

[2004]; see Appendix A for details on plant traits). The

traits were scaled to have a mean of zero and variance of

one. The resulting trait matrix was converted into a

Euclidean distance matrix and used to calculate the

distance based functional diversity metrics; for calculat-

ing functional diversity (FD) the distance matrix was

converted into a functional dendrogram by a UPGM

clustering analysis (Petchey and Gaston 2002; see also

August 2013 1879NOTES

Appendix B). For each plant community, FD was

calculated as the branch length connecting the member

species of the respective community. In addition, Rao’s

quadratic diversity (Qr) (Rao 1982) was estimated as

done by Botta-Dukat (2005) using species percentage

cover (averaged for May and August 2008) to weight

branch lengths between species to generate an abun-

dance-weighted functional index that incorporates

information about functional richness as well as

functional evenness of a community.

Phylogenetic diversity.—A phylogeny of all 60 species

in the Jena Experiment species pool was constructed

based on four genes, using Bayesian methods (for

details, see Allan et al. [2013] and Appendix C). Two

measures of phylogenetic diversity were calculated from

this phylogeny: mean pairwise distance (MPD) and

mean nearest neighbor distance (MNND). MPD mea-

sures the mean phylogenetic distance between all pairs of

species (close and distant relatives) and is affected by the

number of deeper splits in the phylogeny. MNND

measures the mean distance between each species and its

closest relative and therefore measures diversity only at

the tips of the phylogeny. However, several communities

had high MPD but low MNND, providing the

justification for including both measures in our analyses.

Soil organism and plant sampling.—Five soil cores (5

cm diameter, 5 cm depth) were taken from each plot for

determining microbial biomass (Cmic). The pooled five

samples were homogenized, sieved (2 mm) to remove

larger roots, animals, and Cmic was measured using an

O2-microcompensation apparatus (Scheu 1992). Glu-

cose was added to saturate the catabolic enzymes of the

microorganisms (4 mg/g dry mass added as solution to

increase the water content to the water holding capacity

of the soil). The mean of the lowest three hourly

readings within the first 10 h was taken as maximum

initial respiratory response (MIRR; lL O2�h�1�g soil dry

mass�1) and microbial biomass (lg C/g soil dry mass)

was calculated as 38 3 MIRR (Beck et al. 1997).

Belowground macro- and mesofauna were Tullgren

extracted from one large (20 cm diameter, 20 cm depth)

and one small (5 cm diameter, 10 cm depth) soil cores

per plot. Diversity of arbuscular mycorrhizal fungi

(AMF) was analyzed using a molecular TaqMan qRT-

PCR assay designed after initial ITS PCR based

inventory of the AMF species list present at the site

(Konig et al. 2010). Plant species cover was estimated

from a 9-m2 subplot in each plot, whereas the

aboveground plant community biomass was the average

of the late May and late August (year 2008) harvests by

clipping the vegetation at 3 cm above ground in four

rectangles of 0.23 0.5 m per plot (see Weigelt [2010] for

methodological details on measurements of plant

biomass and cover estimates). Data on soil biota have

been used previously to explore the effects of plant

species and functional groups (Eisenhauer et al. 2010,

Scherber et al. 2010), but have not been used to test

functional or phylogenetic diversity indices.

Statistical analyses.—We used path analysis, a partic-

ular case of structural equation modeling involving only

measured variables, to test the support for multiple

potential drivers while accounting for the unavoidable

colinearities among the explanatory variables (Grace

2006; see also Appendix D). As the calculation of the

abundance weighted functional and phylogenetic diversi-

ty indices was based on the realized species richness

(Rdiv) from the same year as the soil biota samplings

(year 2008), we therefore preferred to use the year 2008

Rdiv in the analyses (note that Rdiv is highly correlated

with sown species richness [Pearson’s r (realized, sown)¼0.99]). For each response variable (see Appendix E) a full

model of causal relationships was created including

simultaneously several hypothetical pathways through

which log-transformed (to reduce leverage and linearize

relationships) Rdiv could affect soil biota. The full model

(see Appendix F) included a direct pathway between Rdiv

and soil biota abundance/diversity and several indirect

pathways via functional diversity indices (FD and Qr),

phylogenetic diversity indices (MPD and MNND),

functional group richness (FG) and functional group

biomass (i.e., the biomass of legumes, grasses, small

herbs, and tall herbs). We included the path between Rdiv

and the other diversity measures because the Jena

Experiment manipulated species richness and therefore

variation in functional/phylogenetic diversity is caused by

variation in species richness and composition between

plots. Functional and phylogenetic diversity were in turn

hypothesized to affect soil biota directly or indirectly via

several measures of plant biomass (shoot, root, and total)

as plant productivity has been shown to be an important

driver of belowground communities (Spehn et al. 2000).

A continuous variable (percent clay content in the upper

10 cm soil layer ranging from 13.7% to 25.6% was

preferred to the block variable to account for the

variability in soil texture. The full models were simplified

by step-wise exclusion of variables with nonsignificant

regression weights and nonsignificant covariances as

estimated by AIC (Akaike information criterion) scores

until a minimal adequate model was achieved. Minimal

adequate models were indicated by non-significant

differences when comparing the predicted and observed

covariance matrices (v2 tests with P . 0.05), by lower

AIC, lower root mean squared error approximation

(RMSEA ,0.05) and higher comparative fit index (CFI

. 0.90) (Grace 2006, Arbuckle 2009). Path analysis was

performed using the SPPS Amos 18 statistical package

(Arbuckle 2009). Scatter plots with univariate linear

regression line were produced in R 2.15.0 (R Develop-

ment Core Team 2012) for visualizing the direction of the

relationship between the soil biota and the different

diversity indices (Appendix G). Data on abundance and

diversity of the soil biota was square-root transformed to

NOTES1880 Ecology, Vol. 94, No. 8

reduce heteroscedasticity of error variances, whereas

microbial biomass was log-transformed to improve

normality. Although variables such as total mesofauna

and macrofauna abundances and diversity are not

independent from the comprising subordinate groups,

we consider them informative because independent data

sets with abundances and diversity of multiple trophic

levels of soil biota as well as in-situ collected plant

functional traits are rarely available. In addition, it is

unknown whether diversity and abundance of these

groups respond similarly to functional and phylogenetic

plant diversity.

RESULTS

Plant functional or phylogenetic diversity indices were

retained as significant predictors of soil biota diversity

and abundance in 10 out of the 11 minimal adequate

path analysis models, while direct effects of realized

species richness (Rdiv) were retained in just two models

(Fig. 1, Appendixes H and I). Rdiv is correlated with

FIG. 1. Minimal adequate models for the effects of multiple plant community predictors on the abundance and diversity ofvarious groups of belowground organisms (see Appendix F for the maximal model). Solid arrows show significant relationships(pathways) between variables, dotted arrows indicate a nonsignificant relationship, and numbers next to arrows show standardizedparameter estimates (i.e., standardized regression weights). Circles (e1–e6) indicate error terms, and double-headed arrows indicatesignificant correlations between the error terms. Squared multiple correlations (R2) for the predicted/dependent group of soil biotais given on the box of the dependent variable. Abbreviations are: Rdiv, realized plant species richness; Cmic, microbial biomass;AMFdiv, arbuscular mycorrhizal fungal diversity; CollembolaAB, abundance of Collembola; OrbatidaAB, abundance ofOribatida; TotalmesofaunaAB, total mesofauna invertebrate abundance; MacrosaprotrophAB, macrofauna decomposerinvertebrate abundance; MacrosaprotrophDIV, macrofauna decomposer invertebrate species richness; MacroherbivoreAB,macrofauna herbivore invertebrate; MacroherbivoreDIV, macrofauna herbivore invertebrate species richness; Totalmacrofau-naAB, total macrofauna abundance; TotalmacrofaunaDIV, total macrofauna species richness; FD, functional diversity; Qr, Rao’squadratic diversity; MPD, mean pairwise phylogenetic distance; MNND, mean nearest neighbor phylogenetic distance; LegBM,biomass of legumes; RootBM, root biomass; ShootBM, shoot biomass. No adequate models with significant regression weightscould be obtained for abundance of Gamasida, abundance of Symphyla, macrofauna predator abundance, and macrofaunapredator species richness (see Appendices H and I for full results of the path analysis).

August 2013 1881NOTES

most of the other diversity measures (Appendix D),

therefore the path between Rdiv and the other diversity

measures was retained in almost all models; however the

lack of direct paths between Rdiv and soil biota diversity

and abundance indicates that Rdiv is a less important

predictor for soil biota than functional/phylogenetic

diversity. No minimal adequate models could be

achieved for the abundances of gamasid mites and

Symphyla or for the diversity and abundance of

macropredators. Rao’s quadratic diversity (Qr) was

retained as the sole predictor in the models predicting

the abundance and diversity of herbivores (Fig. 1) and

the diversity of macrofauna. Functional diversity (FD)

was retained in six of the eleven adequate models and

affected soil biota directly but also indirectly (via plant

shoot or root biomass). Note that direct paths indicate

effect pathways unrelated to plant biomass. Plant shoot

biomass was retained as a significant predictor in the

models for microbial biomass and the total abundance

of macrofauna (Fig. 1), which was in turn affected by

the amount of legume biomass and FD. Root biomass

was retained in the models for predicting the abundance

of Collembola, mesofauna and macrosaprotrophs and

the diversity of macrosaprotrophs. Phylogenetic diver-

sity was retained in the models predicting microbial

biomass (retained MPD) and the diversity of arbuscular

mycorrhizae (retained MNND). In all minimal adequate

models, the standardized and unstandardized regression

weights (Fig. 1, Appendixes H and I) indicate that

higher functional or phylogenetic diversity led to higher

abundances and diversity of the analyzed taxonomic

groups (Appendix G) and increased plant root and

shoot biomass.

DISCUSSION

Functional and phylogenetic diversity indices have

been proposed as a pragmatic and more accurate way of

capturing potential niche complementarity in a commu-

nity (Cadotte et al. 2009, Clark et al. 2012). While it

often has been shown that plant functional and

phylogenetic diversity drives aboveground communities

and processes (Cadotte et al. 2009, 2012, Flynn et al.

2011), little empirical evidence is available showing that

it is also a key determinant of soil communities. The

findings of this study provide strong evidence that

belowground communities increase in complexity (abun-

dances and species richness) in response to increased

plant functional diversity. The results also show that

functional and phylogenetic diversity metrics outweigh

the traditionally used species and functional group

richness as predictors of soil taxa abundance and

diversity. Measuring multiple functional traits should

provide a higher resolution picture of potential niche

complementarity in a community beyond what species

or functional group richness can explain since not every

species increases functional diversity by an identical

amount (Petchey and Gaston 2002). Furthermore,

additional mechanisms emerging from niche comple-

mentarity such as increased microhabitat heterogeneity

(Eisenhauer et al. 2011), substrate diversity (Spehn et al.

2000), and asynchronous population fluctuations

(Roscher et al. 2011), all of which have been shown to

contribute to increased ecosystem stability (Naeem and

Li 1997, Milcu et al. 2010), can be better captured by

functional diversity (Flynn et al. 2011, Cadotte et al.

2012). This is in contrast to previous studies showing

that soil organisms mainly respond to the presence of

certain plant functional groups (e.g., N2 fixers [Spehn et

al. 2000, Milcu et al. 2008]) and underlines the

importance of considering multiple plant traits in

functional metrics aiming to predict belowground

communities.

Recent studies have emphasized the importance of the

diversity of plant evolutionary lineages for ecosystem

functioning and diversity of arthropods (Cadotte et al.

2012, Dinnage et al. 2012), and here we show that the

evolutionary history of the plant community (measured

as phylogenetic diversity) also drives mycorrhizal diver-

sity and soil microbial biomass. This supports the

existence of strong coevolutionary links between soil

microorganisms and plants and the existence of specific

associations between plants and microorganisms (Rey-

nolds et al. 2003, Eisenhauer et al. 2010). Arbuscular

mycorrhizae are obligate biotroph symbionts that form

tight associations with their host plants, and recent

studies using network theory to link the diversity of host

plants and of arbuscular mycorrhizae found a nested

relationship pattern suggesting strong specialization

(Montesinos-Navarro et al. 2012). As these groups are

important for plant nutrient uptake and decomposition,

and a high mycorrhizal diversity has been shown to be

associated with higher plant biomass production (Mahe-

rali and Klironomos 2007), their greater abundance in

phylogenetically diverse communities might also partially

explain the increase in plant biomass with plant

phylogenetic diversity (Cadotte et al. 2009).

In our study, FD and Qr were superior predictors of

soil biota for meso- and macrofauna, which suggests that

these groups instead respond to the diversity in traits

affecting plant resource quality and micro-environmental

changes. This suggests that different soil organisms are

specialized on different types of plant resources and plant

communities with higher diversity of resources support

more abundant and diverse meso- and macrofauna.

Plant traits are not stable plant characteristics, but vary

with growing seasons, environmental (Ackerly and

Cornwell 2007) and diversity gradients (Gubsch et al.

2011), or even at a given time point within populations

(Albert et al. 2010, Clark 2010). In spite of efforts to

incorporate intraspecific trait variation in measures of

functional diversity (Albert et al. 2010) and the relative

amount of variation in different traits (Clark et al. 2012), it

NOTES1882 Ecology, Vol. 94, No. 8

is unclear how this variation could best be included in

modeling relationships between functional trait diversity

and processes related to other trophic levels, which also

are not constant temporally. It is usually assumed in trait-

based approaches that within-species trait variation is

smaller than between-species trait variation (McGill et al.

2006). Nevertheless, we cannot exclude the possibility that

missing or weak relationships between plant functional

diversity and soil biota were due to not incorporating

intraspecific trait variation. However, the many significant

effects of functional trait diversity that we did find suggest

that the majority of belowground organism groups

respond to interspecific variation in functional traits.

Predator trophic levels (gamasid mites and macro-

predators), did not respond to any of our measures of

plant diversity. This is in line with previous studies

indicating that effects of species diversity of one trophic

level become weaker with the trophic distance (De Deyn

et al. 2004, Scherber et al. 2010). Moreover, it suggests

that the failure to detect bottom-up effects of plant

species richness on predatory trophic levels is not simply

due to overlooking indirect effects via functional and

phylogenetic diversity. Instead, top-down (Haddad et al.

2009) or environmental drivers probably shaped the

abundance and diversity of predatory communities.

The multiple pathways through which species richness

can affect the diversity and abundance of soil fauna

emphasizes the multidimensionality of factors involved in

plant–soil interactions. While many of the plant func-

tional and phylogenetic effects where direct, in some cases

they were mediated by plant productivity, e.g., soil

microbial biomass increased with shoot biomass (which

in turn was affected by the presence of legumes and

functional diversity), but also with phylogenetic diversity.

Shoot biomass has previously been suggested to affect

microbial biomass through the amount of root exudates

entering the soil (Eisenhauer et al. 2010). Furthermore,

root biomass was also retained in several models as an

indirect pathway through which functional diversity

affected abundances and diversity of saprotrophic fauna,

in line with previous studies underlining the role of the

quantity of belowground inputs for decomposer food

webs (Pollierer et al. 2007, Bardgett and Wardle 2010).

Overall, the relationship between functional diversity

and belowground communities may still seem somewhat

weaker when compared with aboveground food webs

(Scherber et al. 2010). However, it is highly unlikely that

there will ever be a ‘‘perfect’’ index of functional diversity

able to equally predict the response of all trophic levels

(Petchey and Gaston 2006). Moreover, the heterogeneous

nature of the soil environment at different spatial and

temporal scales poses a major challenge for ecologists

trying to quantify the importance of biotic determinants

of soil communities and very likely explains the low effect

sizes detected in some studies so far (Gastine et al. 2003,

Viketoft et al. 2009). Despite some limitation of the used

metrics (e.g., unknown effects of plant functional trait

variation), the results provide strong evidence that soil

biota are responsive to facets of plant diversity such as

functional and phylogenetic diversity and add to the

mounting evidence that plant diversity is a key driver of

belowground communities and ecosystem functioning

(Zak et al. 2003, Milcu et al. 2008, Eisenhauer et al. 2012).

ACKNOWLEDGMENTS

Deutsche Forschungsgemeinschaft (DFG) is acknowledgedfor funding the Jena Experiment (FOR 456) and NicoEisenhauer (Ei 862/1 and Ei 862/2). A. Milcu is grateful toGeorgina Mace and Jacques Roy for their support.

LITERATURE CITED

Ackerly, D. D., and W. Cornwell. 2007. A trait-based approachto community assembly: partitioning of species trait valuesinto within-and among-community components. EcologyLetters 10:135–145.

Albert, C. H., W. Thuiller, N. G. Yoccoz, A. Soudant, F.Boucher, P. Saccone, and S. Lavorel. 2010. Intraspecificfunctional variability: extent, structure and sources ofvariation. Journal of Ecology 98:604–613.

Allan, E., T. Jenkins, A. J. F. Fergus, C. Roscher, M. Fischer,J. Petermann, W. W. Weisser, and B. Schmid. 2013.Experimental plant communities develop phylogeneticallyoverdispersed abundance distributions during assembly.Ecology 94:465–477.

Arbuckle, J. 2009. Amos 18 user’s guide. SPSS Incorporated,Armonk, New York, USA.

Bardgett, R. D., and D. A. Wardle. 2010. Aboveground-belowground linkages: biotic interactions, ecosystem process-es, and global change. Oxford University Press, Oxford, UK.

Beck, T., R. Joergensen, E. Kandeler, F. Makeschin, E. Nuss,H. Oberholzer, and S. Scheu. 1997. An inter-laboratorycomparison of ten different ways of measuring soil microbialbiomass C. Soil Biology and Biochemistry 29:1023–1032.

Botta Dukat, Z. 2005. Rao’s quadratic entropy as a measure offunctional diversity based on multiple traits. Journal ofVegetation Science 16:533–540.

Cadotte, M., J. Cavender-Bares, D. Tilman, and T. Oakley.2009. Using phylogenetic, functional and trait diversity tounderstand patterns of plant community productivity. PLoSONE 4:843–845.

Cadotte, M., R. Dinnage, and G. D. Tilman. 2012. Phyloge-netic diversity promotes ecosystem stability. Ecology 93:223–233.

Clark, C. M., D. F. B. Flynn, B. J. Butterfield, and P. B. Reich.2012. Testing the link between functional diversity andecosystem functioning in a Minnesota grassland experiment.PLoS ONE 7:e52821.

Clark, J. S. 2010. Individuals and the variation needed for highspecies diversity in forest trees. Science 327:1129–1132.

De Deyn, G. B., C. E. Raaijmakers, J. Van Ruijven, F.Berendse, and W. H. Van Der Putten. 2004. Plant speciesidentity and diversity effects on different trophic levels ofnematodes in the soil food web. Oikos 106:576–586.

Dinnage, R., M. W. Cadotte, N. M. Haddad, G. M. Crutsinger,and D. Tilman. 2012. Diversity of plant evolutionary lineagespromotes arthropod diversity. Ecology Letters 15:1308–1317.

Eisenhauer, N., H. Beßler, C. Engels, G. Gleixner, M.Habekost, A. Milcu, S. Partsch, A. Sabais, C. Scherber,and S. Steinbeiss. 2010. Plant diversity effects on soilmicroorganisms support the singular hypothesis. Ecology91:485–496.

Eisenhauer, N., A. Milcu, E. Allan, N. Nitschke, C. Scherber,V. Temperton, A. Weigelt, W. W. Weisser, and S. Scheu.

August 2013 1883NOTES

2011. Impact of above and below ground invertebrates ontemporal and spatial stability of grassland of differentdiversity. Journal of Ecology 99:572–582.

Eisenhauer, N., P. B. Reich, and S. Scheu. 2012. Increasingplant diversity effects on productivity with time due todelayed soil biota effects on plants. Basic and AppliedEcology 13:571–578.

Felsenstein, J. 1985. Phylogenies and the comparative method.American Naturalist 125:1–15.

Flynn, D., N. Mirotchnick, M. Jain, M. Palmer, and S. Naeem.2011. Functional and phylogenetic diversity as predictors ofbiodiversity-ecosystem function relationships. Ecology 92:1573–1581.

Gastine, A., M. Scherer-Lorenzen, and P. Leadley. 2003. Noconsistent effects of plant diversity on root biomass, soilbiota and soil abiotic conditions in temperate grasslandcommunities. Applied Soil Ecology 24:101–111.

Grace, J. B. 2006. Structural equation modeling and naturalsystems. Cambridge University Press, Cambridge, UK.

Gubsch, M., N. Buchmann, B. Schmid, E.-D. Schulze, A.Lipowsky, and C. Roscher. 2011. Differential effects of plantdiversity on functional trait variation of grass species. Annalsof Botany 107:157–169.

Haddad, N. M., G. M. Crutsinger, K. Gross, J. Haarstad,J. M. H. Knops, and D. Tilman. 2009. Plant species lossdecreases arthropod diversity and shifts trophic structure.Ecology Letters 12:1029–1039.

Konig, S., T. Wubet, C. F. Dormann, S. Hempel, C. Renker,and F. Buscot. 2010. TaqMan real-time PCR assays to assessarbuscular mycorrhizal responses to field manipulation ofgrassland biodiversity: effects of soil characteristics, plantspecies richness, and functional traits. Applied and Environ-mental Microbiology 76:3765.

Maherali, H., and J. N. Klironomos. 2007. Influence ofphylogeny on fungal community assembly and ecosystemfunctioning. Science 316:1746.

McGill, B. J., B. J. Enquist, E. Weiher, and M. Westoby. 2006.Rebuilding community ecology from functional traits.Trends in Ecology and Evolution 21:178–185.

Milcu, A., S. Partsch, C. Scherber, W. W. Weisser, and S.Scheu. 2008. Earthworms and legumes control litter decom-position in a plant diversity gradient. Ecology 89:1872–1882.

Milcu, A., E. Thebault, S. Scheu, and N. Eisenhauer. 2010.Plant diversity enhances the reliability of belowgroundprocesses. Soil Biology and Biochemistry 42:2102–2110.

Montesinos-Navarro, A., J. G. Segarra-Moragues, A. Valiente-Banuet, and M. Verdu. 2012. The network structure of plant–arbuscular mycorrhizal fungi. New Phytologist 194:536–547.

Naeem, S., and S. Li. 1997. Biodiversity enhances ecosystemreliability. Nature 390:507–509.

Petchey, O. L., and K. J. Gaston. 2002. Functional diversity(FD), species richness and community composition. EcologyLetters 5:402–411.

Petchey, O. L., and K. J. Gaston. 2006. Functional diversity:back to basics and looking forward. Ecology Letters 9:741–758.

Pollierer, M., R. Langel, C. Korner, M. Maraun, and S. Scheu.2007. The underestimated importance of belowgroundcarbon input for forest soil animal food webs. EcologyLetters 10:729–736.

R Development Core Team. 2012. R: a language andenvironment for statistical computing. R Foundation forStatistical Computing, Vienna, Austria. www.r-project.org

Rao, C. R. 1982. Diversity and dissimilarity coefficients—aunified approach. Theoretical Population Biology 21:24–43.

Reynolds, H. L., A. Packer, J. D. Bever, and K. Clay. 2003.Grassroots ecology: plant–microbe–soil interactions as driv-ers of plant community structure and dynamics. Ecology 84:2281–2291.

Roscher, C., J. Schumacher, J. Baade, W. Wilcke, G. Gleixner,W. W. Weisser, B. Schmid, and E. D. Schulze. 2004. The roleof biodiversity for element cycling and trophic interactions:an experimental approach in a grassland community. Basicand Applied Ecology 5:107–121.

Roscher, C., A. Weigelt, R. Proulx, E. Marquard, J.Schumacher, W. W. Weisser, and B. Schmid. 2011. Identi-fying population- and community-level mechanisms ofdiversity–stability relationships in experimental grasslands.Journal of Ecology 99:1460–1469.

Scherber, C., N. Eisenhauer, W. W. Weisser, B. Schmid, W.Voigt, M. Fischer, E. D. Schulze, C. Roscher, A. Weigelt,and E. Allan. 2010. Bottom-up effects of plant diversity onmultitrophic interactions in a biodiversity experiment.Nature 440:553–556.

Scheu, S. 1992. Automated measurement of the respiratoryresponse of soil microcompartments: active microbial bio-mass in earthworm feaces. Soil Biology and Biochemistry 24:1113–1118.

Scheu, S. 2001. Plants and generalist predators as mediatorsbetween the below-ground and the above-ground system.Basic and Applied Ecology 2:3–13.

Scheu, S., and H. Setala. 2002. Multitrophic interactions indecomposer food webs. Pages 223–264 in B. A. Hawkins andT. Tscharntke, editors. Multitrophic level interactions. Cam-bridge University Press, Cambridge, UK.

Spehn, E. M., J. Joshi, B. Schmid, J. Alphei, and C. Korner.2000. Plant diversity effects on soil heterotrophic activity inexperimental grassland ecosystems. Plant and Soil 224:217–230.

Van der Putten, W. H., L. E. M. Vet, J. A. Harvey, and F. L.Wackers. 2001. Linking above-and belowground multitro-phic interactions of plants, herbivores, pathogens, and theirantagonists. Trends in Ecology and Evolution 16:547–554.

Viketoft, M., J. Bengtsson, B. Sohlenius, M. P. Berg, O.Petchey, C. Palmborg, and K. Huss-Danell. 2009. Long-termeffects of plant diversity and composition on soil nematodecommunities in model grasslands. Ecology 90:90–99.

Wardle, D. A., R. D. Bardgett, J. N. Klironomos, H. Setala,W. H. Van Der Putten, and D. H. Wall. 2004. Ecologicallinkages between aboveground and belowground biota.Science 304:1629.

Weigelt, A., E. Marquard, V. M. Temperton, C. Roscher, C.Scherber, P. N. Mwangi, S. Felten, N. Buchmann, B. Schmid,and E. D. Schulze. 2010. The Jena Experiment: six years ofdata from a grassland biodiversity experiment. Ecology 91:930–931.

Zak, D. R., W. E. Holmes, D. C. White, A. D. Peacock, and D.Tilman. 2003. Plant diversity, soil microbial communities,and ecosystem function: are there any links? Ecology 84:2042–2050.

SUPPLEMENTAL MATERIAL

Appendix A

A table with the means and standard deviations of the 12 functional traits derived from the 60 species present in the JenaExperiment (Ecological Archives E094-170-A1).

NOTES1884 Ecology, Vol. 94, No. 8

Appendix B

Dendrogram representing the relationships between the 60 species present in the Jena Experiment based on functional traits(Ecological Archives E094-170-A2).

Appendix C

Maximum clade-credibility phylogeny of the 60 species in the Jena Experiment (Ecological Archives E094-170-A3).

Appendix D

Correlation matrixes of predictors (Ecological Archives E094-170-A4).

Appendix E

A table presenting the different groups of soil biota sampled in year 2008 (Ecological Archives E094-170-A5).

Appendix F

A schematic of the maximal model used in structural equation modeling (Ecological Archives E094-170-A6).

Appendix G

Scatter plots with linear regression line for visualizing the direction of the relationship between the soil biota and the differentdiversity indices (Ecological Archives E094-170-A7).

Appendix H

Model fit estimates for the minimal adequate models (Ecological Archives E094-170-A8).

Appendix I

Standardized and unstandardized maximum-likelihood estimates for the minimal adequate structural equation models(Ecological Archives E094-170-A9).

August 2013 1885NOTES

Milcu et al. 2013

1/21

Appendices

Appendix A

Table with the means and standard deviatoins (SD) of the 12 functional traits derived from

the 60 species present in the Jena Experiment.

Appendix B

Dendrogram representing the relationships between the 60 species present in the Jena

Experiment based on functional traits.

Appendix C

Maximum clade-credibility phylogeny of the 60 species in the Jena Experiment.

Appendix D

Correlation matrixes of predictors.

Appendix E

Table with the different groups of soil biota sampled in year 2008.

Appendix F

Schematic of the maximal model used in structural equation modelling.

Appendix G

Scatter plots with linear regression line for visualising the direction of the relationship

between the soil biota and the different diversity indices

Appendix H

Model fit estimates for the minimal adequate models.

Appendix I

Standardised and unstandardized maximum-likelihood estimates for the minimal adequate

structural equation models.

Milcu et al. 2013

2/21

Appendix A

Means and standard deviations (SD) of the 12 functional traits derived from the 60 species

present in the Jena Experiment. Trait data were collected in monocultures and averaged for

the two sampling periods (May and August 2006, corresponding to the time of biomass

harvest) with the exception of six species, which were sampled in 2008. For details on the in-

situ collected plant trait data see Roscher et al. (2012). Literature sourced data was presented

in Roscher et al. (2004) and the data on secondary compounds in Loranger et al. (2013).

Functional trait Type (Unit) Mean SD Source

shoot biomass continous (mg dry weight) 600.75 805.49 in-situ

biomass-to-N ratio Continous (mg N g-1) 48.03 21.60 in-situ

leaf area ratio continous (mm2 mg-1 dry weight) 12.69 6.15 in-situ

seed mass continous (mg) 2.30 3.61 in-situ

hemicellulose content percentage 16.36 5.14 in-situ

lignin content percentage 8.16 4.73 in-situ

C-to-N ratio ratio 26.52 11.19 in-situ

sesonality of foliage ordinal 2.22 0.87 literature

rooting type ordinal 2.33 0.75 literature

ability to fix nitrogen binary (presence / absence) 0.20 0.40 literature

known secondary comounds counts 6.72 8.57 literature

rooting depth continous (cm) 0.70 0.27 literature

Additional literature cited

Loranger, J., S. Meyer, B. Shipley, J. Kattge, H. Kern, C. Roscher, and W. Weisser. 2013.

Predicting invertebrate herbivory from plant traits: evidence from 51 grassland species in

experimental monocultures. Ecology 93, 2674–2682.

Milcu et al. 2013

3/21

Appendix B

Dendrogram representing the relationships between the 60 species present in the Jena

Experiment based on functional traits. Different functional groups are colored: graminoids in

green, legumes in yellow, small herbs in red and tall herbs in blue.

Milcu et al. 2013

4/21

Achillea millefolium

Ajuga reptans

Alopecurus pratensis

Anthoxanthum odoratum

Anthriscus sylvestris

Arrhenatherum elatiusAvenula pubescens

Bellis perennis

Bromus erectusBromus hordeaceus

Campanula patula

Cardamine pratensis

Carum carvi

Centaurea jaceaCirsium oleraceum

Crepis biennis

Cynosurus cristatusDactylis glomerata

Daucus carota

Festuca pratensisFestuca rubra

Galium mollugo

Geranium pratense

Glechoma hederacea

Heracleum sphondylium

Holcus lanatus

Knautia arvensis

Lathyrus pratensis

Leontodon autumnalis

Leontodon hispidus

Leucanthemum vulgare

Lotus corniculatus

Luzula campestris

Medicago lupulinaMedicago varia

Onobrychis viciifolia

Pastinaca sativa

Phleum pratense

Pimpinella major

Plantago lanceolataPlantago media

Poa pratensisPoa trivialis

Primula veris

Prunella vulgaris

Ranunculus acrisRanunculus repens

Rumex acetosaSanguisorba officinalis

Taraxacum officinale

Tragopogon pratensis

Trifolium campestreTrifolium dubium

Trifolium fragiferumTrifolium hybridum

Trifolium pratenseTrifolium repens

Trisetum flavescens

Veronica chamaedrys

Vicia cracca

120 100 80 60 40 20 0

Branch lengths in millions of yearsBranch lengths in millions of years

Appendix C

Maximum clade-credibility phylogeny of the 60 species in the Jena Experiment as published

in Allan et al. (2012). Different functional groups are colored: graminoids in green, legumes

in yellow, small herbs in red and tall herbs in blue. 95% confidence intervals for node ages

are shown. Congeners were used for Onobrychis viciifolia (O. montana) and for Pimpinella

major (P. saxifraga). Node support was high: 66% of nodes gave a posterior probability of 1

and a further 23% a posterior probability >0.97. Only 7 nodes were less well-supported, 4 in

the Poaceae, plus the placement of Bellis perennis (0.64), Rumex acetosa (0.68) and the node

between Cardamine pratensis and Geranium pretense (0.82).

Milcu et al. 2013

5/21

Appendix D

Correlation matrix of the predictors. Acronyms: realized plant species richness (Rdiv),

biomass of legumes (legBM), biomass of grasses (grBM), biomass of small herbs (shBM),

biomass of tall herbs (thBM), functional diversity (FD), Rao’s quadratic diversity (Qr), Mean

Pairwise phylogenetic Distance (MPD) and Mean Nearest Neighbour phylogenetic Distance

(MNND).

Rdiv grBM shBM thBM legBM FD Qr MPD MNND

Rdiv 1.00 0.35 0.41 0.25 0.17 0.90 0.78 0.55 0.05

grBM 0.35 1.00 -0.11 -0.15 -0.12 0.39 0.22 -0.05 -0.26

shBM 0.41 -0.11 1.00 -0.18 0.17 0.37 0.37 0.35 0.12

thBM 0.25 -0.15 -0.18 1.00 -0.02 0.25 0.17 0.27 0.20

legBM 0.17 -0.12 0.17 -0.02 1.00 0.21 0.17 0.18 0.13

FD 0.90 0.39 0.37 0.25 0.21 1.00 0.64 0.41 -0.05

Qr 0.78 0.22 0.37 0.17 0.17 0.64 1.00 0.45 0.05

MPD 0.55 -0.05 0.35 0.27 0.18 0.41 0.45 1.00 0.82

MNND 0.05 -0.26 0.12 0.20 0.13 -0.05 0.05 0.82 1.00

Correlations of predictors marked red are significant at p < 0.05

Milcu et al. 2013

6/21

Appendix E

List of acronyms for the different groups of soil biota sampled in year 2008. AB and DIV refer to abundances and species richness, respectively.

Data on species richness was not available for the mesofauna groups.

Acronym Description

Cmic Microbial biomass

AMFDIV Arbuscular mycorrhizal fungal species richness

CollembolaAB Abundance of Collembola

OribatidaAB Abundance of Oribatida

GamasidaAB Abundance of Gamasida

SymphylaAB Abundance of Symphyla

TotalmesofaunaAB Total mesofauna invertebrate abundance including Collembola, Oribatida, Gamasida and Symphyla

MacrosaprotrophAB Macrofauna decomposer invertebrate abundance, including Oligocheta, Isopoda, Diplopoda, Diplura and Protura

MacrosaprotrophDIV Macrofauna decomposer invertebrate species richness

MacroherbivoreAB Macrofauna herbivore invertebrate abundance including Stylomatophora, Thysanoptera, Coleoptera, Lepidoptera,

Heteroptera, Sternorrhyncha and Auchenorrhyncha

MacroherbivoreDIV Macrofauna herbivore invertebrate species richness

MacropredatorsAB Macrofauna predator abundance including Linyphiidae, Hahniidae, Lycosidae, Tetragnathidae, Thomisidae,

Theridiidae, Dictynidae, Gnaphosidae, Chilopoda and Colleoptera

MacropredatorsDIV Macrofauna predator species richness

TotalmacrofaunaAB Total macrofauna abundance including Macrosaprotroph, Macroherbivores and Macropredators

TotalmacrofaunaDIV Total macrofauna species richness

Milcu et al. 2012

7/21

Appendix F

Schematic of the maximal model used in the path analysis. The maximal model included

simultaneously all measures of functional (FD and Qr) and phylogenetic (MPD and MNND)

diversity as well as any additional explanatory indices (i.e. legBM, grBM, shBM, thBM and

FG). Measurements of plant net primarily productivity (shoot, root and total plant biomass)

and soil texture have also been introduced in the model and retained in the minimal adequate

models if significant.

Milcu et al. 2012

8/21

Appendix G. Scatterplots with linear regression line for visualising the direction of the relationship between the soil biota and the different

diversity indices. A localy-eighted regression procedure (loess) has been also plotted for estimating a regression surface (the grey surface

alongside the regressin line) by a multivariage smooting procedure fitting a linear function of the independent variables in a moving fashion that

is analogous to how a moving average is computed for a time series. For brevity we chose to present the log species richness (Rdiv) and

functional group richness (FG) alongside the best functional diversity (FD or Qr) and phylogenetic diversity (MPD or MNND) predictors.

Several groups of soi biota where no diversity predictors are significant are not presented. * P , 0.05; ** P , 0.01; *** P , 0.001; ns, not

significant. The graphs and associated statistics were performed in R 2.15 ( R Development Core Team, 2012).

**, r2= 0.10 ***, r

2= 0.17 *, r

2= 0.07 ***, r

2= 0.15

ns, r2= 0.00 ns, r

2= 0.03 ns, r

2= 0.00 *, r

2= 0.05

Milcu et al. 2012

9/21

**, r2= 0.11 ns, r

2= 0.02 *, r

2= 0.07 ns, r

2= 0.05

ns, r2= 0.03 ns, r

2= 0.02 *, r

2= 0.05 ns, r

2= 0.03

*, r2= 0.08 ns, r

2= 0.00 *, r

2= 0.06 ns, r

2= 0.01

Milcu et al. 2012

10/21

ns, r2= 0.03 ns, r

2= 0.03 *, r

2= 0.06

*, r2= 0.07

*, r2= 0.07 *, r

2= 0.06 ***, r

2= 0.16 **, r

2= 0.09

ns, r2= 0.02 ns, r

2= 0.02 *, r

2= 0.06 ns, r

2= 0.03

Milcu et al. 2012

11/21

ns, r2= 0.01 +, r

2= 0.04

ns, r2= 0.03

ns, r2= 0.00 *, r

2= 0.06

+, r2= 0.04 +, r

2= 0.04

ns, r2= 0.02

ns, r2= 0.02 ns, r

2= 0.02 ns, r2

= 0.01 ns, r2= 0.02

Milcu et al. 2012

12/21

Appendix H

Results of structural equation modelling outputs for the most parsimonious adequate models. See Fig. 1 for the schematics of the adequate

models and Appendix F in Supporting Information for the schematic of the maximal model; AIC: Akaike information criterion; RMSEA: root

mean squared error approximation; CFI: comparative fit index; NA: not applicable. No adequate models with significant regression weights

could be obtained for GamasidAB, SymphylaAB, MacropredatorAB and MacropredatorDIV (see Appendix E for the list of acronyms). Note

that here P-values >0.05 imply that the departure of the data from the model is not significant hence, the models are considered adequate. For

OribatidaAB probability levels could not be computed as the model was saturated.

* The AIC of the maximal models varied between 250.64 and 247.49 depending on the dependent variable.

Model χ2 df P

AIC*

RMSEA

CFI

Cmic 6.47 9 0.840 54.42 0 1

AMFDIV 0.40 3 0.939 22.40 0 1

CollembolaAB 2.52 2 0.284 26.51 0.04 1

OribatidaeAB NA 0 NA 10.00 0.10 1

TotalmesofaunaAB 0.85 1 0.357 28.00 0 1

MacrosaprotrophAB 1.74 1 0.187 27.74 0.06 0.99

MacrosaprotrophDIV 0.01 1 0.935 26.01 0 1

MacroherbivoreAB 0.24 1 0.620 16.24 0 1

MacroherbivoreDIV 0.40 1 0.525 16.40 0 1

TotalmacrofaunaAB 5.51 9 0.788 41.51 0 1

TotalmacrofaunaDIV 0.36 1 0.548 16.36 0 1

Milcu et al. 2012

13/21

Appendix I. Table outputs with maximum likelihood estimates of the minimum adequate

structural equation models.

Microbial biomass (Cmic)

Regression Weights:

Estimate S.E. C.R. P

FD <--- Rdiv 28.454 1.559 18.256 ***

leg <--- Rdiv .377 .129 2.924 .003

MPD <--- Rdiv 116.201 20.385 5.700 ***

shootBM <--- leg 39.867 12.304 3.240 .001

shootBM <--- FD 1.949 .466 4.185 ***

Cmic <--- MPD .000 .000 3.786 ***

Cmic <--- clay .016 .002 7.004 ***

Cmic <--- shootBM .000 .000 2.208 .027

Standardized Regression Weights:

Estimate

FD <--- Rdiv .901

leg <--- Rdiv .315

MPD <--- Rdiv .543

shootBM <--- leg .313

shootBM <--- FD .405

Cmic <--- MPD .320

Cmic <--- clay .567

Cmic <--- shootBM .186

Intercepts:

Estimate S.E. C.R. P

FD -5.301 1.169 -4.534 ***

leg .291 .097 3.009 .003

MPD 77.888 15.291 5.094 ***

shootBM 43.918 9.422 4.661 ***

clay 21.581 .397 54.391 ***

Cmic 2.527 .051 49.590 ***

Variances:

Estimate S.E. C.R. P

e1 <--> e3 10.679 4.053 2.635 .008

e1 <--> e2 -122.552 48.586 -2.522 .012

Milcu et al. 2012

14/21

Arbuscular mycorrhizal fungal diversity (AMFDIV)

Regression Weights:

Estimate S.E. C.R. P

MNND <--- Rdiv -4.045 21.328 -.190 .850

AMFdiv <--- clay .027 .011 2.359 .018

AMFdiv <--- MNND .001 .001 1.948 .051

Standardized Regression Weights:

Estimate

MNND <--- Rdiv -.022

AMFDIV <--- clay .258

AMFDIV <--- MNND .213

Intercepts:

Estimate S.E. C.R. P

MNND 100.757 16.296 6.183 ***

clay 21.455 .406 52.868 ***

AMFDIV 1.924 .254 7.566 ***

Variances:

Estimate S.E. C.R. P

Rdiv .171 .028 6.105 ***

e2 12.275 2.011 6.105 ***

e1 5791.766 948.761 6.105 ***

e3 .120 .020 6.105 ***

Abundance of Collembola (CollembolaAB)

Regression Weights:

Estimate S.E. C.R. P

FD <--- Rdiv 28.629 1.578 18.138 ***

rootBM <--- Rdiv 324.351 98.496 3.293 ***

rootBM <--- FD -7.174 3.103 -2.312 .021

CollembolaAB <--- rootBM .063 .024 2.671 .008

Standardized Regression Weights:

Estimate

FD <--- Rdiv .902

Milcu et al. 2012

15/21

Estimate

rootBM <--- Rdiv .809

rootBM <--- FD -.568

CollembolaAB <--- rootBM .294

Intercepts:

Estimate S.E. C.R. P

FD -5.308 1.179 -4.501 ***

rootBM 300.159 35.811 8.382 ***

CollembolaAB 53.573 10.515 5.095 ***

Variances:

Estimate S.E. C.R. P

Rdiv .177 .029 6.145 ***

e1 33.247 5.410 6.145 ***

e2 24172.051 3933.459 6.145 ***

e3 1197.611 194.884 6.145 ***

Abundace of Oribatid mites (OribatidaAB)

Regression Weights:

Standardized Regression Weights:

Intercepts:

Estimate S.E. C.R. P

OribatidaAB 53.720 8.524 6.302 ***

Variances:

Estimate S.E. C.R. P Label

Rdiv .177 .029 6.145 ***

Estimate S.E. C.R. P

OribatidaAB <--- Rdiv 34.033 11.409 2.983 .003

Estimate

OribatidaAB <--- Rdiv .325

Milcu et al. 2012

16/21

Total mesofauna invertebrate abundance (TotalmesofaunaAB)

Regression Weights:

Estimate S.E. C.R. P

FD <--- Rdiv 28.453 1.567 18.153 ***

rootBM <--- Rdiv 303.236 98.056 3.092 .002

rootBM <--- FD -6.680 3.105 -2.152 .031

TotalmesofaunaAB <--- rootBM .065 .031 2.071 .038

TotalmesofaunaAB <--- FD .744 .398 1.870 .062

Standardized Regression Weights:

Estimate

FD <--- Rdiv .901

rootBM <--- Rdiv .758

rootBM <--- FD -.528

TotalmesofaunaAB <--- rootBM .226

TotalmesofaunaAB <--- FD .204

Intercepts:

Estimate S.E. C.R. P

FD -5.272 1.178 -4.475 ***

rootBM 304.207 35.944 8.463 ***

TotalmesofaunaAB 77.756 14.065 5.528 ***

Variances:

Estimate S.E. C.R. P

Rdiv .177 .029 6.186 ***

e1 33.225 5.371 6.186 ***

e2 24506.447 3961.818 6.186 ***

e3 2082.479 336.663 6.186 ***

Total saprotroph mesofauna abundance (MacrosaprotrophAB)

Regression Weights:

Estimate S.E. C.R. P

FD <--- Rdiv 28.637 1.583 18.086 ***

rootBM <--- Rdiv 295.812 98.880 2.992 .003

rootBM <--- FD -6.535 3.108 -2.102 .036

MacrosaprotrophAB <--- Rdiv 4.794 2.075 2.311 .021

MacrosaprotrophAB <--- rootBM -.010 .005 -1.874 .061

Milcu et al. 2012

17/21

Standardized Regression Weights:

Estimate

FD <--- Rdiv .900

rootBM <--- Rdiv .735

rootBM <--- FD -.517

MacrosaprotrophAB <--- Rdiv .263

MacrosaprotrophAB <--- rootBM -.213

Intercepts:

Estimate S.E. C.R. P

FD -5.470 1.195 -4.576 ***

rootBM 307.712 36.684 8.388 ***

MacrosaprotrophAB 10.127 2.326 4.353 ***

Variances:

Estimate S.E. C.R. P

Rdiv .172 .028 6.186 ***

e1 33.055 5.344 6.186 ***

e2 24439.434 3950.985 6.186 ***

e3 52.609 8.505 6.186 ***

Macrosaprprotroph diversity (MacrosaprotrophDIV)

Regression Weights:

Estimate S.E. C.R. P

FD <--- Rdiv 28.637 1.583 18.086 ***

rootBM <--- Rdiv 295.812 98.880 2.992 .003

rootBM <--- FD -6.535 3.108 -2.102 .036

MacrosaprotrophDIV <--- FD .025 .008 3.088 .002

MacrosaprotrophDIV <--- rootBM -.001 .001 -2.194 .028

Standardized Regression Weights:

Estimate

FD <--- Rdiv .900

rootBM <--- Rdiv .735

rootBM <--- FD -.517

MacrosaprotrophDIV <--- FD .330

MacrosaprotrophDIV <--- rootBM -.235

Milcu et al. 2012

18/21

Intercepts:

Estimate S.E. C.R. P

FD -5.470 1.195 -4.576 ***

rootBM 307.712 36.684 8.388 ***

MacrosaprotrophDIV 2.236 .287 7.791 ***

Variances:

Estimate S.E. C.R. P

Rdiv .172 .028 6.186 ***

e1 33.055 5.344 6.186 ***

e2 24439.434 3950.985 6.186 ***

e3 .846 .137 6.186 ***

Macroherbivore abundance (MacroherbivoreAB)

Regression Weights:

Estimate S.E. C.R. P

Qr <--- Rdiv 8.477 .800 10.595 ***

MacroherbivoreAB <--- Qr .055 .022 2.537 .011

Standardized Regression Weights:

Estimate

Qr <--- Rdiv .771

MacroherbivoreAB <--- Qr .279

Intercepts:

Estimate S.E. C.R. P

Qr .255 .604 .423 .673

MacroherbivoreAB 1.710 .157 10.915 ***

Variances:

Estimate S.E. C.R. P

Rdiv .172 .028 6.186 ***

e1 8.438 1.364 6.186 ***

e2 .749 .121 6.186 ***

Milcu et al. 2012

19/21

Macroherbivore diversity (MacroherbivoreDIV)

Regression Weights:

Estimate S.E. C.R. P

Qr <--- Rdiv 8.477 .800 10.595 ***

MacroherbivoreDIV <--- Qr .047 .012 3.827 ***

Standardized Regression Weights:

Estimate

Qr <--- Rdiv .771

MacroherbivoreDIV <--- Qr .401

Intercepts:

Estimate S.E. C.R. P

Qr .255 .604 .423 .673

MacroherbivoreDIV 1.262 .089 14.112 ***

Variances:

Estimate S.E. C.R. P

Rdiv .172 .028 6.186 ***

e1 8.438 1.364 6.186 ***

e2 .244 .039 6.186 ***

Total macrofauna abundance (TotalmacrofaunaAB)

Regression Weights:

Estimate S.E. C.R. P

leg <--- Rdiv .408 .129 3.148 .002

FD <--- Rdiv 28.637 1.583 18.086 ***

shootBM <--- FD 1.945 .475 4.098 ***

shootBM <--- leg 39.982 12.542 3.188 .001

TotalmacrofaunaAB <--- shootBM .004 .003 1.651 .099

TotalmacrofaunaAB <--- clay -.127 .050 -2.537 .011

Standardized Regression Weights:

Estimate

leg <--- Rdiv .339

FD <--- Rdiv .900

shootBM <--- FD .401

Milcu et al. 2012

20/21

Estimate

shootBM <--- leg .312

TotalmacrofaunaAB <--- shootBM .178

TotalmacrofaunaAB <--- clay -.274

Intercepts:

Estimate S.E. C.R. P

FD -5.470 1.195 -4.576 ***

leg .262 .098 2.685 .007

shootBM 43.949 9.488 4.632 ***

clay 21.528 .398 54.029 ***

TotalmacrofaunaAB 7.223 1.115 6.475 ***

Variances:

Estimate S.E. C.R. P

Rdiv .172 .028 6.186 ***

e1 33.055 5.344 6.186 ***

e2 .221 .036 6.186 ***

e5 12.150 1.964 6.186 ***

e3 2725.403 440.600 6.186 ***

e4 2.317 .375 6.186 ***

Total macrofauna diversity (TotalmacrofaunaDIV)

Regression Weights:

Estimate S.E. C.R. P

Qr <--- Rdiv 8.477 .800 10.595 ***

TotalmacrofaunaDIV <--- Qr .033 .015 2.172 .030

Standardized Regression Weights:

Estimate

Qr <--- Rdiv .771

TotalmacrofaunaDIV <--- Qr .241

Intercepts:

Estimate S.E. C.R. P

Qr .255 .604 .423 .673

TotalmacrofaunaDIV 2.638 .111 23.772 ***

Milcu et al. 2012

21/21

Variances:

Estimate S.E. C.R. P

Rdiv .172 .028 6.186 ***

e1 8.438 1.364 6.186 ***

e2 .376 .061 6.186 ***