Plant diversity effects on soil heterotrophic activity in experimental grassland ecosystems

Plant and Soil 224: 217–230, 2000. © 2000 Kluwer Academic Publishers. Printed in the Netherlands. 217 Plant diversity effects on soil heterotrophic activity in experimental grassland ecosystems Eva M. Spehn1,∗ , Jasmin Joshi2, Bernhard Schmid2 , Jörn Alphei3 and Christian Körner1 1 Botanisches Institut der Universität Basel, Schönbeinstr. 6, CH-4056 Basel, Switzerland; 2 Institut für Umweltwissenschaften der Universität Zürich, Winterthurerstr. 190, CH-8057 Zürich, Switzerland and 3 Abteilung Oekologie der Universität Göttingen, Berliner Str. 28, D-37073 Göttingen, Germany Received 21 September 1999. Accepted in revised form 11 May 2000. Key words: BIODEPTH, decomposition, earthworms, legumes, microbial biomass, plant species richness Abstract The loss of plant species from terrestrial ecosystems may cause changes in soil decomposer communities and in decomposition of organic material with potential further consequences for other ecosystem processes. This was tested in experimental communities of 1, 2, 4, 8, 32 plant species and of 1, 2 or 3 functional groups (grasses, legumes and non-leguminous forbs). As plant species richness was reduced from the highest species richness to monocultures, mean aboveground plant biomass decreased by 150%, but microbial biomass (measured by substrate induced respiration) decreased by only 15% (P = 0.05). Irrespective of plant species richness, the absence of legumes (across diversity levels) caused microbial biomass to decrease by 15% (P = 0.02). No effect of plant species richness or composition was detected on the microbial metabolic quotient (qCO2 ) and no plant species richness effect was found on feeding activity of the mesofauna (assessed with a bait-lamina-test). Decomposition of cellulose and birchwood sticks was also not affected by plant species richness, but when legumes were absent, cellulose samples were decomposed more slowly (16% in 1996, 27% in 1997, P = 0.006). A significant decrease in earthworm population density of 63% and in total earthworm biomass by 84% was the single most prominent response to the reduction of plant species richness, largely due to a 50% reduction in biomass of the dominant ‘anecic’ earthworms. Voles (Arvicola terrestris L.) also had a clear preference for high-diversity plots. Soil moisture during the growing season was unaffected by plant species richness or the number of functional groups present. In contrast, soil temperature was 2 K higher in monocultures compared with the most diverse mixtures on a bright day at peak season. We conclude that the lower abundance and activity of decomposers with reduced plant species richness was related to altered substrate quantity, a signal which is not reflected in rates of decomposition of standard test material. The presence of nitrogen fixers seemed to be the most important component of the plant diversity manipulation for soil heterotrophs. The reduction in plant biomass due to the simulated loss of plant species had more pronounced effects on voles and earthworms than on microbes, suggesting that higher trophic levels are more strongly affected than lower trophic levels. Introduction In terrestrial ecosystems, soil heterotrophic organisms regulate soil processes such as nutrient cycling and carbon flow. In grasslands, voles are often important herbivores (Rychnovska, 1993). Earthworms are decomposers of dead plant material and their key role in nutrient cycling and physical soil improvement, ∗ FAX No: 41-61-267-3504. E-mail: Eva.Spehn@unibas.ch especially in grasslands, has long been recognized (Darwin, 1881, Edwards and Lofty, 1972; Lee, 1985). Earthworms and other macrofauna mix organic material in the soil, reduce the size of the detritus particles and make them available to microbes (Swift et al., 1979). Micro- and mesofauna feed on microbes and thereby increase microbial turnover and plant nutrient availability. Most soil heterotrophs depend on plants as primary producers and on quantity (litter production, root turnover and root exudation), quality and availab- 218 ility of organic substrate over time. Therefore, organic matter mineralisation is largely determined by interactions between the soil food web and plants (Clarholm, 1985). Recent field and microcosm studies showed that plant species loss often leads to a decline in plant biomass (Ewel et al., 1991; Naeem et al., 1994; Tilman et al., 1996). In the BIODEPTH experiment (BlODiversity and Ecological Processes in Terrestrial Herbaceous ecosystems), where experimental grassland ecosystems were set up with five levels of plant species richness at eight field sites across Europe, an overall linear reduction of plant biomass with a logarithmic decrease of species number was found (Hector et al., 1999). In the current study, which was part of the BIODEPTH experiment, monocultures contained only 40% of the biomass of the most diverse species mixtures in the third year of the experiment (Spehn et al., 2000). A reduction in plant biomass caused by a loss in plant diversity is expected to have strong effects on the decomposer community: microbial biomass is likely to decrease, because organic carbon sources limit soil microbial activity in most terrestrial ecosystems (Smith and Paul, 1990; Van de Geijn and van Veen, 1993). Consequently, mesofauna that depend directly or indirectly on bacteria and macrofauna may suffer from plant diversity loss because of this indirect negative effect on food supply. A 30% decline in earthworm biomass occurred in experimental grassland communities when plant species richness was reduced from 31 to 5 species, and this decline was most likely driven by an accompanying decrease in fineroot biomass production (Zaller and Arnone, 1999). When changes in biodiversity affect primary production, there may be associated subsequent effects on soil heterotrophic organisms and decomposition due to this change in biomass. However, there could also be other additional biodiversity effects. Tilman (1996) addressed this issues, distinguishing effects of decreased plant diversity on soil NO− 3 beyond simple changes in root biomass. Additionally, studies manipulating the diversity of litter (e.g. Chapman et al., 1988; Blair et al., 1990; Wardle et al., 1997; Bardgett and Shine, 1999; Hector et al., in press) are able to hold total organic matter input constant but change litter quality, demonstrating that the presence of individual plant species sometimes has considerable influence on decomposer communities. Differences in the allocation of carbon in grass species significantly affects bacteria and bacteria-feeding nematodes in the rhizosphere (Griffiths et al., 1992). Large differences in nitrogen mineralization occurred in initially identical soils after 3 years of growth of monocultures of different perennial grass species (Wedin and Tilman, 1990) due to differences in the quality and quantity of plant litter (Wedin and Pastor, 1993). A diverse mix of litter can affect decay rates, although measurements of mixed-species litter decomposition (Staaf, 1980; Jenkinson et al., 1985) have yielded inconsistent results. In some studies, responses of decomposition rates are idiosyncratic, with both synergistic and antagonistic effects of litter diversity (Chapman et al., 1988; Wardle et al., 1997; Hector et al., in press; Wardle and Nicholson, 1996). Other studies showed no significant effects in decay rates when litter of species was mixed together as compared to monospecific litter (Blair et al., 1990; Rustad, 1994). Finally, food supply over the season is more variable in communities with low plant species richness because different phenologies of species overlap less in time (McNaughton, 1993). A reduction in plant species richness can therefore have considerable effects on the decomposer subsystem, via (a) effects on plant biomass production, (b) effects of differences in chemical composition of plant debris, or (c), via the timing of debris provision. The objective of this paper is to quantify the effects of plant species loss on some key groups of the decomposer community in the Swiss part of the BIODEPTH experiment. We directly measured the size and diversity of earthworm communities, and indirectly assessed the biomass and activity of microbes and the abundance of voles. To assess diversity effects on the cycling of organic material we measured the decomposition of standard test material and monitored soil moisture and soil temperature as abiotic drivers of decomposition. The following specific hypotheses were tested: (1) A decrease in plant species richness reduces the biomass of soil heterotrophic communities through a reduction in food quality and less constant supply of food over time and due to a reduction in the amount of available food sources (litter, roots). (2) As a consequence of reduced soil heterotrophic biomass or activity, rates of decomposition of organic test material might also be reduced. (3) That reduced plant biomass and canopy complexity enhance transmission of radiation and hence stimulate soil heat flux and topsoil dehydration in low diversity communities, affecting decomposition. 219 Methods ecosystem processes than among mixtures at lower diversity levels (Doak et al., 1998). Study site The experimental site is situated at Lupsingen (47◦ 27′ N, 7◦ 41′ E, 440 m a.s.l.) south of Basel in the Swiss Jura Mountains. The mean air temperature in January is 0.7 ◦ C and in July 18.3 ◦ C (annual mean temperature for 1986–1995: 9.0 ◦ C). Precipitation is 1046 ± 38 mm per year (data from the nearest meteorological station). The calcareous loamy soil is close to neutral (pH 6–7) and overlies Jurassic bedrock. The site was formerly cropped with corn (1993) and oilseed rape (1994) and was ploughed late in 1994, left bare over winter and harrowed twice before the experimental plots were established in spring 1995. During the test period the experimental field was surrounded by hay fields. All experimental plots were mowed in June and September, following the traditional management of meadows in this area. All plant species that had not been sown with the original species assemblage were regularly removed to maintain the diversity treatments. Experimental design We created 5 diversity levels with a species richness of 32, 8, 4, 2, and 1 species per plot. In each of two replicate blocks, each plot within a diversity level had a different species composition to avoid confounding of the effects of diversity with species identity. We replicated species richness – 2 different mixtures with 32 species, 5 mixtures with 8 species, 8 mixtures with 4 species, 7 mixtures with 2 species and 10 different monocultures and replicated composition – all of the previous 32 mixtures occurred in one plot in two replicated blocks, making 64 plots in total (Table 1). The overall species pool consisted of 48 species (Table 1), representative of semi-natural species-rich grassland communities typical for this area (Arrhenaterion alliance according to the vegetation classification of Ellenberg, 1988). We distinguished three ‘functional groups’ of species, grasses (G), legumes (L) and nonleguminous forbs (F), and we varied the number of functional groups in such a way that this factor was as orthogonal as possible with the factor species richness. Nevertheless, all combinations of diversity treatments between the two factors cannot be established, e.g. monocultures can only contain one functional group and the 32-species mixtures always contained all three functional groups. At the higher diversity levels we used fewer mixtures to reduce overlap in species composition and because we expected less variability in Establishment of the experimental plant communities The species assemblages were sown into 2 × 2 m plots at a density of 2000 viable seeds per m−2 (equally divided between the species in each plant assemblage). Seedling density 75 days after sowing was independent of the diversity level (see Diemer et al., 1997). During the experiment, all species that were sown were still present, except for the 32-species mixtures, where the real species richness varied between 24 × 1.8 (n=4) in the second year (1996) and 27 × 1.7 (n=4) species in the third. Species richness was measured as presence of species in the plots at the time of the harvests. The plant communities significantly differed in primary production. Aboveground plant biomass increased linearly with the logarithm of plant species number in all three years of the experiment. The most species-rich communities produced 143% more biomass than the mean of all monocultures and 25% more biomass than the most productive monoculture in the third year of the experiment (Spehn et al. 2000). Total below-ground plant biomass showed no significant diversity effects, but biomass of fineroots (root diameter < 1 mm) decreased from 200 g m−2 in high diverse mixtures to 120 g m−2 in monocultures (P = 0.03). We used these biomass data as covariates in the statistical analyses to separate different diversity effects (see statistical analysis in this section). Voles The few data for voles (Arvicola terrestris L.) presented here did not emerge as part of the experimental design, but resulted from a spontaneous invasion at the end of the first year of the experiment. Thereafter, the abundance of voles was controlled by frequent trapping and occasional fumigation with carbon monoxide, so no data from the 2nd and 3rd year are available. However, these animals showed quite astonishing plot preferences, so we report these observations. Vole abundance was estimated by the number of exit holes from burrows counted on 2 × 8 m plots. Earthworm sampling Earthworms were sampled in April 1998 when soils were wet and earthworms were highly active (i.e. high surface cast production) using the electro-sampling method (Thielemann, 1986). Eight steel rods (65 cm 220 Table 1. Details of the 32 different species mixtures including identity and numbers of the species sown and functional group composition (Grasses (G) Herbs (H) and Legumes (L)). The whole set of species combinations is replicated once. Failure of species establishment is shown in brackets (- = failure in one and - - = failure in both replicates). Species abbreviations in alphabetical order are : Ae = Arrhenatherum elatius (L.) J. et C. P RESL, Am= Achillea millefolium L., Ao= Anthoxanthum odoratum L., Ap= Alopecurus pratensis L., Cc= Cynosurus cristatus L., Dg= Dactylis glomerata L., Dc= Daucus carota L., Fp= Festuca pratensis H UDSON, Fr= Festuca rubra L., Hl= Holcus lanatus L., Ka= Knautia arvensis (L.) COULTER, LAp= Lathyrus pratensis L., Lc= Lotus corniculatus L., Lp= Lolium perenne L., Pl= Plantago lanceolata L., Pp= Poa pratensis L., Ra= Ranunculus acris L., Tf= Trisetum flavescens (L.) P.B., To= Taraxacum officinale W EBER, Tp= Trifolium pratense L., Tr= Trifolium repens L., ∗ = additional species were: Agropyron repens (L.) P.B., Agrostis stolonifer L. (- -) Anthyllis vulneraria L. (-) Onobruchis viciifolia S COP ., Ajuga reptans L., Anthriscus sylvestris (L.) H OFFM. (- -), Bellis perennis L., Centaurea scabiosa L., Crepis biennis L., Galium verum L., Salvia pratensis L. (-), Sanguisorba officinalis L. (-), Scabiosa columbaria L., Silene vulgaris (MOENCH) G ARCKE. ∗∗ = additional species were: Agrostis tenuis S IBTH. (-), Festuca ovina L., Phleum pratense L., Medicago sativa L., Vicia cracca L., Ajuga reptans L., Campanula patula L., Centaurea jacea L., Leucanthemum vulgare L AM., Geranium pratense L., Heracleum sphondylium L. (- -), Pimpinella major (L.) H UDSON (- -), Potentilla erecta (L.) RÄUSCHEL (- -), Prunella vulgaris L., Sanguisorba officinalis L. Species Number of Functional Species code mixture species group sown composition 1–10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 1 2 2 2 2 2 2 2 4 4 4 4 4 4 4 4 8 8 8 8 8 32 32 6 G, 2H, 2 L G G G GL GL GH GH G G GL GL GH GH GHL GHL G GL GH GHL GHL GHL GHL one monoculture each of Dg, Lp, Pp, Ae, Tf, Fp, TP, Tr, Pl and To Dg Lp Ae Fp Pp Tf Lp Tr Ae Tp Fp Pl Pp To Fp Hl Pp Tf Cc Dg Fp Lp Ae Fr Tf Tr Hl Lp Pp Tp Dg Fr Tf Ra Ae Cc Pp Am Ae Lp LAp To Dg Fp Lc Pl Ao Ae Cc Dg Fp Hl Pp Tf Ap Ao Cc Fr Lp Pp LAp Tp Ap Ae Dg Fp Hl Tf Pl Ra Ae Dg Fr Lp Lc Tr Am Dc Fp Lp Pp Tf Tp Tr Ka To Ap Ae Cd Dg Fr Lp Pp Tf LAp Lc Tp Tr Am Dc Ka Pl Ra To ∗ Ao Ae Dg Fp Hl(- - ) Lp Pp LAp(-) Lc Tp Tr Am Dc Pl Ra To ∗∗ long × 6 mm diameter) were pushed into the ground down to a depth of 35–40 cm in a circle 60 cm in diameter. Electric voltage (300–500 V DC) was applied for 40 minutes in 4-second pulses supplied pairwise to opposing rods in the circle. Each successive electrical pulse was applied clockwise to the next pair of opposing rods. The portable transformer (DEKA 4000, Deka Gerätebau, Marsberg, Germany) and the pulse generator were powered by a 12 V car battery. All earthworms emerging at the soil surface were collected and stored in cool water. Earthworms were sorted by species and assigned to one of the three ecological groups, following Bouché (1977): epigeics (worms living on the soil surface); anecics (worms forming vertical burrows in the soil); endogeics (worms inhabiting the mineral soil horizon). Earthworm numbers and fresh weight per ecological group were determined about one hour after collection, then earthworms were released in their plots of origin to avoid disturbance of the experiment. Microbial activity and microbial biomass In June 1997 and October 1997 soil microbial activity and soil microbial biomass were determined. Soil samples were taken from each plot to a depth of 10 cm. Roots and macrofauna were removed by hand 221 and the soil samples were sieved (<2 mm) before pre-incubating the samples (June: 3 days at 22 ◦ C, October 5 days at 20 ◦ C). Respiration was measured by slightly different methods for the two sample dates. In June, CO2 evolution was used to measure respiration of soil samples in continuous flow incubators connected to a differential infrared gas analyser (details in Niklaus, 1998). The October respiration of soil samples was measured by oxygen consumption, using an automated system based on electrolytic O2 microcompensation (details in Scheu, 1992). Therefore only aerobe microbial respiration was monitored in October, whereas in June both, aerobe and anaerobe respiration were measured. Soil basal respiration (Rmic µg CO2 -C g−1 soil h−1 ) was calculated as the 10 hour mean of all 10 minute-readings beginning 7.5 hours after the start of the measurements for the June samples (October: mean O2 consumption rates between 15 and 20 h). The same samples were thereafter amended with glucose (0.8 g g−1 dry weight) and substrate-induced respiration (SIR) was measured as above for another 6 (October: 10) hours to calculate microbial biomass carbon (Cmic , µg C g−1 soil), using the equation Cmic = 0.37 + 3.52 ∗ SIR (Anderson and Domsch, 1978). The metabolic quotient for CO2 (qCO2 , h−1 ), an index of activity per biomass, was calculated by dividing basal respiration by the microbial biomass carbon (Cmic ). Soil organic carbon was determined from October samples by dry combustion using an automated CHN analyser (LECO, CHN-1000, LECO Corporation, St. Joseph, Michigan, USA). Soil biota activity We used three types of test material for a standardized assessment of biological activity of soil organisms. (1) As a trial of soil fauna feeding activity (mostly mesofauna) we used a ‘bait-lamina-test’ (Thörne, 1990). The plastic bait sticks were 15 cm long, 6 mm wide and had 16 holes in a row of 8 cm length. The holes had a diameter of 1.5 mm and were filled with bait composed of 65% cellulose, 15% agar, 10% bran, 10% bentonit(C) (binding material). Ten sticks per plot were placed vertically in the soil down to a depth of 10 cm and left there for 10 days in May 1996. After removal from the soil all holes that were empty and at least half-empty were counted. To get a better spatial resolution we divided the stick into an upper and lower half and distinguished two soil horizons (2–6 cm and 6–10 cm depth). (2) As an indicator for shorter-term ‘decomposition’ we measured the dry weight loss of cotton, a standard organic substrate containing ca. 95% cellulose, with an initial N concentration of 0.09%. Strips of cotton (12.2 × 12.5 cm) were buried vertically in the topsoil (0–10 cm) over 3–4 months of the main growing period (7 May–24 July 1996; 9 June–4 September 1997). Cotton strips were protected against voles with a nylon mesh with a mesh size of 1 mm. (3) Longer-term decomposition was measured by dry weight loss of birch-sticks. Sets of three flat birch sticks (11.5 × 0.9 × 0.2 cm) were bundled using polyester thread, and buried vertically 1 cm below the surface for 18 months (March 1997–August 1998). Only the central stick in the bundle was monitored. Initial N concentration was 0.08%. Soil moisture and soil temperature Soil moisture was determined at several sampling dates in the second and third year of the experiment and at different moisture conditions by using a portable soil moisture sensor (ThetaProbe ML1, Delta-T Devices, Cambridge, UK; description of operation principle in Gaskin, 1996). We determined the mean of at least 5 measurements in each plot in the top layer of the soil (0–6 cm depth). A soil-specific calibration was made with gravimetric samples for different soil water contents. Soil temperature was determined with an electric thermometer (2 mm steel rod, Testo 110, Testo, Lenzkirch, Germany) in the upper soil at 5 cm depth at the end of May 1997, when the canopy was fully developed. The measurements were taken at midday of two successive cloudless days, with four readings per plot taken in a random sequence. Statistical analysis The data were analysed with analysis of variance (ANOVA) using multiple regression approaches (Neter and Wasserman, 1974) as implemented in Genstat 5 (Payne et al., 1993). The hierarchical experimental design included the treatment factors number of species (S), number of functional groups (FG), the identity of species mixture (M) and the presence of legumes (L) (Table 2). To avoid the problem of confounding effects of species identity with effects of species number or functional group number, the mean squares of S and FG were divided by the mean square of M and this variance ratio was then tested against tabulated F-values. To test for linearity of regression 222 Table 2. Skeleton analysis of variance for the measured characters Source of variation d.f. Mean square Variance ratio Block (B) Subblock (SB) [nested within B] Species richness (S) Species richness (log-linear) Species richness (deviation from log-linearity No. of functional groups (F) Functional groups (linear) Functional groups (deviation from linearity) Mixture (M) [nested within S,F,L] Legume presence contrast (L) Remainder (all other c.) (R) Residual = Plot [P[B]] Total 1 1 4 1 3 2 1 1 24 1 23 17 63 msB msSB msS msSL msSD msF msFL msFD msM msL msR msP msT msB /msSB msSB /msP msS /msM msSL / msM msSD / msM msF / msM msFL / msM msFD / msM msM / msR msL / msR msR / msP msT / msP Notes: Random effects (error model) are in italics, fixed effects (treatment model) are in roman letters; effects are always adjusted for the previous effects. functions S and FG, we split the two factors into a test for linearity (for the logarithm of species number or for the functional group number) and a test for the deviation from linearity (Neter and Wasserman, 1974). The spatial factor block (B), and a covariate ‘subblock’ (SB) were added to account for small-scale spatial variation (Diemer et al., 1997). To separate effects of diversity on heterotrophic soil organisms driven by changes in primary production from other diversity effects, we performed analysis of covariance (ANCOVA) using the same model as above, except introducing above-and below-ground biomass as covariables, which were tested first in the model. We then looked for differences in the amount of variance explained by the factor S in the ANCOVA- compared to the ANOVA-model (see also Joshi et al., 2000). Results Voles Voles significantly preferred highly diverse plant species mixtures (P = 0.03; Figure 1) and the presence of legumes irrespective of species number (P < 0.001), whereas the number of functional groups had no significant effect (P = 0.9). Earthworms Total biomass of earthworms decreased with decreasing plant species diversity from 140±13 g m−2 in Figure 1. Effects of plant species richness on voles (P = 0.03), measured as number of exit holes from burrows per plot. Shown are the means ± S.E. of vole disturbances per species richness level. the most diverse mixtures to 76±6 g fresh mass m−2 in monocultures (linear contrast P = 0.0025; Figure 2). Earthworm density also decreased from 266±9 in high-diverse mixtures to 163±14 individuals m−2 in monocultures (linear contrast P =0.016). For both measures the decrease was log-linear (deviation from log-linearity P = 0.34 and P = 0.09, respectively). A significant decrease in earthworm biomass and density was also observed with decreasing number of plant functional groups if functional group richness was tested before plant species richness (P = 0.0025 and P 223 0.1 and P = 0.07, respectively). As a consequence, the effect of the number of species was no longer significant if it was fitted after the aboveground plant biomass covariate (S:P = 0.4) but it remained significant with belowground plant biomass or fineroot biomass as a covariate (S:P = 0.003 and P = 0.006). Hence, plant species richness is likely to affect earthworms mostly via effects on aboveground plant biomass production. The occurrence of mixtures with high earthworm biomass and low aboveground plant biomass within each diversity level (Figure 2b) suggests that there could also be additional effects of plant diversity. Plant species composition and the presence of legumes in the mixtures both showed no significant influence on biomass or density of earthworms. We found a total of six species of earthworms, all of them occurred in at least 80% of the 64 plots, except Octolasion cyaneum Sav. which was found in 30% of the plots only. Anecic species (Lumbricus terrestris L., Nicodrilus longus Ude) accounted for 79% of total earthworm biomass. Endogeic species (Allolobophora rosea Say., A. chloroitica Sav., Octolasion cyaneum Sav.) accounted for 18% of total earthworm biomass, while the epigeic species (Lumbricus castaneus Sav.) only made up 3% of total earthworm biomass. The species richness of earthworms did not differ significantly between plant diversity treatments (P = 0.25). On average, plots contained five different earthworm species. The significant decrease in earthworm biomass and density with species or functional group reduction was mainly due to the decrease in anecic species (Figure 2a), but the relative abundances of the different ecological groups of earthworms was not affected. Microbial responses Figure 2. (a) Effects of plant species number on earthworm biomass (g m−2 , P < 0.001), divided for anecic (P =0.003),  endogeic (P =0.9) and epigeic earthworms (P =0.5) (Bouché, 1977). Means ± S.E. per species richness level. (b) The two effects of plant species number and biomass production on total earthworm biomass, illustrated by diameters of circles (means per plot). =0.04 respectively, data not shown). An obvious effect of plant species richness reduction was a reduction in plant biomass production (Spehn et al., 2000). When aboveground plant biomass was fitted in the model as a covariate it explained a considerable part of the variation in earthworm biomass (P = 0.03, belowground plant biomass or fineroot biomass as covariate: P = Microbial biomass decreased log-linearly with plant species loss and linearly with loss of functional groups both producing 15% reductions (32-species mixtures compared with monocultures P =0.046, deviation from linearity P =0.8, Figure 3, three compared with one functional group P =0.007) in October 1997, but not in June 1997. The absence of legumes reduced microbial biomass significantly on both sampling dates by 15% in June and in October 1997 (P = 0.008 and P = 0.024, respectively) irrespective of species diversity. A covariance analysis with aboveground plant biomass (harvested in September 1997) explained a considerable part of the variation in microbial biomass of October 1997 (P < 0.001), but belowground plant biomass or fineroot biomass (harvested in September 224 Figure 4. Effect of plant species richness on Cmic / Corg ratio, measured in October 1997 (P = 0.08). Means ± S.E. cies on microbial biomass was no longer significant if it was fitted after the aboveground plant biomass covariate (S:P = 0.6), but remained significant with belowground plant biomass or fineroot biomass as a covariate (S:P = 0.02 and P = 0.03). Basal respiration was not significantly influenced by plant diversity in June or October 1997, respectively. The presence of legumes stimulated microbial respiration only in June 1997 (P =0.047). The metabolic quotient was neither significantly affected by plant species richness nor by plant functional group richness at both sampling dates. The metabolic quotients did also not differ between both dates (and methods). The Cmic /Corg ratio significantly decreased with plant species loss (P = 0.003; Figure 4) and with decreasing number of plant functional groups (P < 0.001). Soil activity Figure 3. Effect of plant species number on microbial respiration (Rmic ), microbial biomass C (Cmic ) and microbial metabolic quotient for CO2 (qCO2 ) as function of plant species richness. Black symbols are samples from October 1997 (O2 -consumption), light symbols those of June 1997 (CO2 -production), both measured with SIR-method (Anderson and Domsch, 1978). Small figures show the effect of presence of legumes in the mixtures across all diversity levels, n.s. indicates ‘not significant’ (P > 0.05), ∗ = ‘P ≤ 0.05’, and ∗∗ = ‘P ≤ 0.01’. Means ± S.E. 1997) did not (P = 0.7 and P = 0.07, respectively). As a consequence, the effect of the number of spe- Mesofauna feeding activity as assessed by bait-lamina tests was not consistently affected by plant species richness in May 1996 (Figure 5). Feeding activity in the upper (2–6 cm depth) and the lower soil layer (6–10 cm depth) showed the same patterns across diversity levels, with a generally higher feeding activity in the upper soil layer (two thirds of total feeding activity). Decomposition of cellulose measured by dry weight loss of cotton strips was significantly affected by plant community composition in 1996 (M: P < 0.001), but no significant effects of plant species num- 225 Figure 5. Effect of plant diversity on soil fauna feeding activity, assessed with ‘bait-lamina-test’. The mean ± S.E of eaten bait is shown ( 0–5 cm depth,  5–10 cm depth). ber (S) or number of plant functional groups (FG) were found either in 1996 or in 1997 (Figure 6, left and middle graph). In both years we observed a significantly higher rate of weight loss of cotton strips when legumes were present in mixtures (L: P = 0.006 and 0.005 for 1996 and 1997, respectively). Decay of birch wood (overall mean loss of weight 37±2%) was not affected by any of the treatment factors (S, FG, M, L) during the 18 months of exposure (Figure 6, right graph). Soil moisture and soil temperature Soil moisture did not vary significantly with plant species richness (Figure 7, left graph), but there was a significant effect of species composition (M: P = 0.02 at 3 of 5 sampling dates). Generally, the variation in soil moisture within each sampling date is small (coefficients of variation are between 0.07 and 0.16). In contrast, soil temperature increased with loss of species number by 2 K (monocultures compared with most diverse mixtures), which was marginally significant (S: P = 0.08, Figure 7, right graph). It significantly increased with the decrease of number of functional groups (FG: P = 0.03), especially with absence of legumes (L: P = 0.007) and was significantly affected by species composition (M: P = 0.03). Discussion The decline in plant diversity negatively affected soil heterotrophic organisms like voles, earthworms and microbes, whereas negative consequences for soil activity and decomposition of standard test material only occurred when legumes were missing in the grassland species assemblages. The decline in plant diversity in general affects soil heterotrophic organisms in two ways: (1) by decreasing plant biomass production (this seems to be the more important pathway in our study), (2) less diverse mixtures probably provided a less balanced diet in terms of food quality and a less constant supply in time. Soil moisture, an important abiotic factor which could co-determine changes in soil activity, was not affected. Effects on soilfauna Voles feed on roots and rhizomes (Corbet and Ovenden, 1982), therefore their significant preference for highly diverse plant mixtures and mixtures with legumes may reflect a preference for a mixed diet or for a species not included in the lower diversity levels. It also may relate to the greater quantity of belowground biomass as quality in terms of N and P concentration did not differ among diversity levels (unpublished data). The pronounced response of earthworms (especially the strong reduction of the ‘anecic’ species) as organic matter input decreased with reduction of species number is in line with known earthworm 226 Figure 6. Effect of plant species richness on dry weight loss (%) of cotton and birchwood sticks used as test material for soil decomposing activity. Means ± S.E. responses. A gradual decrease of organic matter was assumed to be the most influential factor on earthworm communities in studies where permanent grasslands were changed into arable cropping fields (Edwards and Lofty, 1978; Edwards, 1983). The fact that anecic earthworms were affected most can be explained by their predominant foraging on fresh organic matter (Edwards and Bohlen, 1996). A decrease in biomass production should affect these anecic worms faster and more strongly than endogeic earthworms, which feed on humified organic matter. Zaller and Arnone (1999) on the other hand, mainly found a decrease in endogeic earthworm abundance with decreasing plant species richness, but still explain this with an accompanying decrease in fineroot production. In addition, the more diverse diet and more stable supply of biomass may have enhanced the response. Figure 2b illustrates that plant species richness might also have other effects on earthworms, as plant biomass could not entirely explain the size of earthworm communities. Given that earthworm activity has a strong influence on soil fertility, soil water-holding capacity, soil aeration, and soil drainage (Edwards et al., 1990; Roth and Joschko, 1991; Zaller and Arnone, 1999) longer-term feedbacks of these earthworm responses are to be expected. Effects on microbes Microbial biomass also decreased with decreasing plant diversity. A study assessing potential C-source utilisation by bacterial communities on the same field site found a log-linear decrease of bacterial abundance with decreasing plant species richness, and even a decrease in bacterial ‘functional’ diversity (Stephan et al., 2000). In most agro-ecosystems microbial growth is limited by input of fresh organic carbon (Smith and Paul, 1990) and hence plant-biomass production (Zak et al., 1994). In our experiment, reduced plant species richness was indeed associated with a decrease of plant-biomass production, the most likely reason for the decline in microbial biomass. A field experiment by Groffman et al. (1996), with monocultures of ten different grasses found different plant-induced changes in microbial biomass and activity, caused by variation in labile C release among these species. In a further study, grass species differed in their support of rhizosphere bacteria, again most likely due to interspecific differences in rhizodeposition (Griffiths et al., 1992). A sensitive measure for short-term responses to C inputs into soil is the microbial biomass supported per unit of total organic soil carbon Cmic (Corg ratio) (Insam and Domsch, 1988). This ratio also significantly decreased with a decrease in plant species richness or plant biomass in our experiment (Figure 4), which may be an early indicator for a longer-term decline in soil organic matter of low-diversity mixtures. This is in accordance with the assumption that primary production largely controls soil organic matter as is embodied in ecosystem carbon models (Schlesinger, 1977). Wardle and Nicholson (1996) showed that soil microbial biomass is primarily set by plant 227 Figure 7. Effect of plant species richness on soil moisture (vol%) at different census dates (left graph) and on soil temperature (right graph, P = 0.08). Means ± S.E. biomass production, but in accordance with some of our observations, is also influenced by the identity and number of plant species present. In our experiment, the effect of plant species richness on microbial biomass remained statistically significant, even if corrected for effects of belowground (not aboveground) plant biomass production. A decrease in plant species richness may therefore affect microbes directly. One reason could be a decrease in resource heterogeneity. Christie et al. (1974, 1978) found less microbial biomass when Lolium perenne and Plantago lanceolata, two grassland species which were included in our experiment, were grown separately as compared to growing together. Decomposer communities changed when mixed species litter assemblages were provided (irrespective of total N inputs in foliar litter), although the direction of changes remained unpredictable (Blair et al., 1990). Another way in which diversity could directly affect decomposers is through the spatial and temporal provision of substrate, which may have been much more patchy in low compared with high species diversity and therefore could have contributed to the lower microbial biomass per unit of soil (Holland and Coleman, 1987). We conclude that, in addition to effects mediated by greater plant biomass with increasing plant species richness the mixed diet was also beneficial for microbes. In our experiment legumes play a key role, probably because of the high P abundance in the soil, which permits high rates of nitrogen fixation and legume growth (Spehn et al., 2000). Quantity and the quality of fresh organic input might affect microbial biomass and this probably explains part of the decrease in microbial biomass and activity in plant community assemblages without legumes. In our experiment N and P-concentration in total aboveground community biomass decreased when legumes were lacking in a mixture (unpublished data), but the quantity of the substrate decreased as well, therefore it is not possible to distinguish between these two potential plant-driven effects. No effects of species richness on decay of different test substrates could be detected, but plant community composition did have an effect. In particular mixtures without legumes had a lower decomposition rate. Our results are in agreement with earlier studies that found no relationship between decomposition rates of standardized material and plant species diversity (Naeem et al., 1995; Mulder et al., 1999). Decomposition of standard test materials is an index of the environmental contribution to decomposition, and does not measure the decomposition of actual litter, although their low substrate quality might reflect later stages of litter decomposition (Vossbrinck et al., 1979). However, decomposition rates do reflect effects on microbes caused by legume presence, probably due to higher nitrogen availability in the soil. 228 Effects on soil climate The reduction in species richness changed abiotic soil conditions only slightly. We found an increase of 2 K in soil temperature in low-diversity mixtures on a bright day, where reduced vegetation cover and leaf area enhanced transmission of radiation (Spehn et al., 2000). This is in agreement with Dugas et al. (1996) who found greater midday soil heat flux in desert grassland and shrub communities that had a smaller leaf area. Theoretically, soil warming can have large effects on microbial activity (Paul and Clark, 1996). However, we expect effects to be less pronounced for the whole season as compared with bright day effects only, significant influences on microbes were thus unlikely. The increase of canopy transmittance in low diversity mixtures may also have enhanced soil evaporation (Kelliher et al., 1993). Below wheat canopies (Leuning et al., 1994), where light climate was probably comparable to some of our low diversity grassland assemblages, 37% of total evaporation was recorded. In contrast, only 10–20% was recorded in grasslands with a small amount of radiation reaching the soil surface (Ripley and Saugier, 1978). This potential enhancement of soil evaporation as species richness decreases seemed to have been counterbalanced by reduced canopy transpiration (reduced leaf area index) in low diversity mixtures, because we measured no differences in moisture in the upper soil horizons across diversity levels. Conclusion Our hypothesis that heterotrophic soil organisms are negatively affected by a decline in plant species diversity, mediated primarily by a reduction in plant production, was confirmed. However, trophic levels within the decomposer community were affected differently. Large effects occurred on voles and earthworms, whereas changes on microbial biomass were small or not detectable. It seems that the negative effects of reduced plant species richness on larger soil heterotrophs reflects the combined influence of smaller and more variable organic matter input as well as the decreased variety and quality of plant products. These effects apparently do not translate into significantly altered overall soil activity in the short-term, since the decomposition of standard material was not affected. Acknowledgements We are grateful to Matthias Diemer for coordination of this project, Pascal Niklaus for advice with microbial measurements, Johann Zaller for advice with earthworm sampling and species identification, Jean-Luc Delli for soil temperature measurements, and numerous helpers in the field. We also thank all members of the BIODEPTH project for suggestions concerning the ideas and methods of this paper and Pascal Niklaus, Andy Hector and two anonymous reviewers for helpful comments on the manuscript. The Swiss Federal Office for Education and Science funded this project (Project EU-131 1 to B. Schmid). References Anderson J P E and Domsch K H 1978 A physiological method for the quantitative measurement of microbial biomass in soils. Soil Biol. Biochem. 10, 215–221. Bardgett R D and Shine A 1999 Linkages between plant litter diversity, soil microbial biomass and ecosystem function in temperate grasslands. Soil Biol. Biochem. 31, 3 17–321. Blair J M, Parmelee R W and Beare M H 1990 Decay rates, nitrogen fluxes, and decomposer communities of single- and mixed-species foliar litter. Ecology 71, 1976–1985. Bouché M B 1977 Ecologie et Paraécologie: Peut-on apprecier le role de la faune dans les cycles biogeochimiques. In Soil organisms as components of ecosystems. Eds U Lohm and T Persson. pp. 157–163. Ecological Bulletin, Stockholm. Chapman K, Whittaker J B and Heal O W 1988 Metabolic and faunal activity in litters of tree mixtures compared with pure stands. Agr. Ecosyst. Environ. 24, 33–40. Christie P, Newman EI and Campbell R, 1974 Grassland species can influence the abundance of microbes on each other’s roots. Nature 250, 570–571. Christie P, Newman EI and Campbell R, 1978 The influence of neighbouring grassland plants on each other’s endomycorrhizas and root surface microorganisms. Soil Biol. Biochem. 10, 521– 527. Clarholm M 1985 Interactions of bacteria, protozoa and plants leading to mineralization of soil nitrogen. Soil Biol. Biochem. 17, 181–187. Corbet G B and Ovenden D 1982 Mammals of Britain and Europe. Paul Parey, Hamburg, Berlin. Darwin C 1881 The formation of vegetable mould through the action of worms. John Murray, London. Diemer M, Joshi J, Korner C, Schmid B and Spehn E 1997 An experimental protocol to assess the effects of plant diversity on ecosystem functioning utilized in a European research network. Bulletin of the Geobotanical Institute ETH 63, 95–107. Doak D F, Bigger D, Harding E K, Marvier M A, O’Malley R E and Thomson D 1998 The statistical inevitability of stabilitydiversity relationships in community ecology. Amer. Naturalist 151, 264–276. Dugas W A, Hicks R A and Gibbens R P 1996 Structure and function of C-3 and C-4 Chihuaihuan Desert plant communities: Energy balance components. J. Arid Environ. 34, 63–79. 229 Edwards C A 1983 Earthworm ecology in cultivated soils. In Earthworm Ecology. From Darwin to vermiculture. Ed. J E Satchell. pp 123–138. Chapman & Hall, London. Edwards C A and Bohlen P J 1996 Biology and ecology of earthworms. Chapman & Hall, London. Edwards C A and Lofty J R 1972 Biology of Earthworms. Chapman & Hall. London. Edwards C A and Lofty J R 1978 The influence of arthropods and earthworms upon root growth of direct drilled cereals. J. Appl. Ecol. 17, 533–534. Edwards W M, Shipitalo M J, Owens L B and Norton L D 1990 Effect of Lumbricus terrestris L. burrows on hydrology of continuous no-till corn fields. Geoderma 46, 73–84. Ellenberg H 1988 Vegetation ecology of Central Europe. Cambridge University Press, Cambridge. Ewel J J, Mazzarino M J and Berish C W 1991 Tropical soil fertility changes under monocultures and successional communities of different structure. Ecol. Appl. 1, 289–302. Gaskin G J and Miller J D 1996 Measurement of soil water content using a simplified impedance measuring technique. J. Agric. Res. 63, 153–160. Griffiths B S, Welschen R, von Arendonk J J C M and Lambers H 1992 The effect of nitrate-nitrogen on bacteria and the bacterial feeding fauna in the rhizosphere of different grass species. Oecologia 91, 253–259. Groffman P M, Eagan P, Sullivan W M and Lemunyon J L 1996 Grass species and soil type effects on microbial biomass and activity. Plant Soil 183, 61–67. Hector A, Beale A J, Minns A, Otway S J and Lawton J H 2000 Consequences for the reduction of plant diversity for litter decomposition: effects through litter quality and microenvironment. Oikos, 90. (In press). Hector A, Schmid B, Beierkuhnlein C, Caldeira M C, Diemer M, Dimitrakopoulos P G, Finn J, Freitas H, Giller P S, Good J, Harris R, Högberg P, Huss-Danell K, Joshi J, Jumpponen A, Körner C, Leadley P W, Loreau M, Minns A, Mulder C P H, O’Donovan G, Otway S J, Pereira J S, Prinz A, Read D J, Scherer-Lorenzen M, Schulze E-D, Siamantziouras A-S D, Spehn E M, Terry A C, Troumbis A Y, Woodward F I, Yachi S and Lawton J H 1999 Plant diversity and productivity experiments in European grasslands. Science 286, 1123–1127. Holland E A and Coleman D C 1987 Litter placement effects on microbial and organic matter dynamics in an agroecosystem. Ecology 68, 425–433. Insam H and Domsch K H 1988 Relationship between soil organic carbon and microbial biomass on chronosequences of reclamation sites. Microbial Ecol. 15, 177-188. Jenkinson D S, Fox R H and Rayner J H 1985 Interactions between fertilizer nitrogen and soil nitrogen-the so-called ‘priming’ effect. J. Soil Sci. 36, 425–444. Joshi J, Matthies D and Schmid B 2000 Root hemiparasites and plant diversity in experimental grassland communities. J. Ecol. 88. (In press). Kelliher F M, Leuning R and Schulze E D 1993 Evaporation and canopy characteristics of coniferous forests and grasslands. Oecologia 95, 153–163. Lee K B 1985 Earthworms. Their ecology and relationships with soils and land use. Academic Press, Sydney. Leuning R, Condon A G, Dunin F X, Zegelin S and Denmead O T 1994 Rainfall interception and evaporation from soil below a wheat canopy. Agr. Forest Meteorol. 67, 221–238. McNaughton S J 1993 Biodiversity and function of grazing ecosystems. In Biodiversity and ecosystem function. Eds. E-D Schulze and H A Mooney. pp 361–383. Springer, Berlin. Mulder C P H, Koricheva J, Huss-Danell K, Hogberg P and Joshi J 1999 Insects affect relationships between plant species richness and ecosystem processes. Ecology Letters 2, 237–246. Naeem S, Thompson L J, Lawler S P, Lawton J H and Woodfin R M 1994 Declining biodiversity can alter the performance of ecosystems. Nature 368, 734–737. Naeem S, Thompson L J, Lawlers S P, Lawton J H and Woodfin R M 1995 Empirical evidence that declining species diversity may alter the performance of terrestrial ecosystems. Phil. Trans. Roy. Soc. London B 347, 249–262. Neter J and Wasserman W 1974 Applied Linear Statistical Models. Richard D. Irwin Inc., Homewood, Illinois. Niklaus P A 1998 Effects of elevated atmospheric CO2 on soil microbiota in calcareous grassland. Glob. Change Biol. 4, 451–458. Paul E A and Clark F E 1996 Soil microbiology and biochemistry. Academic Press, San Diego. Payne R W, Lane P W, Digby P G N, Harding S A, Leech P K, Morgan G W, Todd A D, Thompson R, Tunicliffe Wilson G, Welham S J and White R P 1993 GENSTAT 5 Reference Manual. Clarendon Press, Oxford. Ripley E A and Saugier B 1978 Biophysics of a natural grassland. J. Appl. Ecol. 15, 459–479. Roth C H and Joschko M 1991 A note on the reduction of runoff from crusted soils by earthworm burrows and artificial channels. Z. Pflanz. Bodenk. 154, 101–105. Rustad L 1994 Element dynamics along a decay continuum in a red spruce ecosystem in Maine, USA. Ecology 75, 867–879. Rychnovska M 1993 Structure and functioning of seminatural meadows. Developments in Agricultural and managed-forest Ecology 27, Elsevier, Amsterdam, London, NY. Scheu S 1992 Automated measurement of the respiratory response of soil microcompartments: Active microbial biomass in earthworm faeces. Soil Biol. Biochem. 24, 1113–1118. Schlesinger W H 1977 Carbon balance in terrestrial detritus. Annu. Rev. Ecol. Syst. 8, 51–81. Smith J and Paul E 1990 The significance of soil microbial biomass estimations. In Soil Biochemistry. Eds J Bollag and G Stotzky. pp 357–393. Marcel Dekker, New York. Spehn E M, Joshi J, Schmid B, Diemer M and Körner C 2000 Aboveground resource use increases with plant species richness in experimental grassland ecosystems. Funct. Ecol. 14. (In press). Staaf H 1980 lnfluence of chemical composition, addition of raspberry leaves, and nitrogen supply on decomposition rate and dynamics of nitrogen and phosphorus in beech leaf litter. Oikos 35, 55–62. Stephan A, Meyer AH and Schmid B 2000 Plant diversity affects culturable soil bacteria in experimental grassland communities. J. Ecol. (In press) Swift M J, Heal 0 W and Anderson J M 1979 Decomposition in terrestrial ecosystems. University of California Press, Berkeley, California, USA. Thielemann U 1986 Elektrischer Regenwurmfang mit der OktettMethode. Pedobiologia 29, 296–3O2. Thörne E 1990 Assessing feeding activities of soil-living animals. Bait-lamina-tests. Pedobiologia 34, 89–101. Tilman D, Wedin D and Knops J 1996 Productivity and sustainability influenced by biodiversity in grassland ecosystems. Nature 379, 7 18–720. Van de Geijn S C and van Veen J A 1993 Implications of increased carbon dioxide levels for carbon input and turnover in soils. Vegetatio 104/105, 283–292. 230 Vossbrinck C R, Coleman D C and Woolley T A 1979 Abiotic and biotic factors in litter decomposition in a semi-arid grassland. Ecology 60, 265–271. Wardle D A, Bonner K I and Nicholson K 5 1997 Biodiversity and plant litter: Experimental evidence which does not support the view that enhanced species richness improves ecosystem function. Oikos 79, 247–258. Wardle D A and Nicholson K S 1996 Synergistic effects of grassland plant species on soil microbial biomass and activity: implications for ecosystem-level effects of enriched plant diversity. Funct. Ecol. 10, 410–416. Wedin D and Pastor J 1993 Nitrogen mineralisation dynamics in grass monocultures. Oecologia 96, 186–192. Wedin D A and Tilman D 1990 Species effects on nitrogen cycling: a test with perennial grasses. Oecologia 84, 433–441. Zak D R, Tilman D, Parmenter R R, Rice C W, Fisher F M, Vose J, Milchunas D and Martin C W 1994 Plant production and soil microorganisms in late-successional ecosystems: a continentalscale study. Ecology 75, 2333–2347. Zaller J G and Arnone J A 1999 Earthworm responses to plant species loss and elevated CO2 in calcareous grassland. Plant Soil 208, 1–8. Zaller J G and Arnone J A 1999 Interactions between plant species and earthworm casts in a calcareous grassland under elevated CO2 . Ecology 80, 873–881. Section editor: R Merckx