In plant communities, a better use of nitrogen for growth and access to additional nitrogen by N2 fixing legumes have been suggested as mechanisms that contribute to the positive relationship between species richness and above-ground biomass production. We studied above-ground biomass production, and plant and soil nitrogen in a biodiversity experiment using a pool of nine potentially dominant grassland species (five grasses: Alopecurus pratensis, Arrhenatherum elatius, Dactylis glomerata, Phleum pratense, Poa trivialis; two legumes: Trifolium pratense, T. repens; two non-legume herbs: Anthriscus sylvestris, Geranium pratense) in the second and third year after establishment.
Total above-ground biomass and nitrogen pools increased with species richness in the second year. In the third year, the positive effect of species richness on above-ground biomass was less pronounced but still significant, while nitrogen pools did not increase with species richness. Above-ground nitrogen pools and biomass production were significantly lower in the third year.
Above-ground nitrogen pools of mixtures were higher than expected from monocultures in 94% of cases and increased with the species richness of mixtures in the second year. In the third year, only 55% of the mixtures had higher nitrogen yields than expected. A significant increase in the amount of biomass produced per gram nitrogen, a decrease of soil nitrate pools and the inability of mixtures to use soil nitrate more completely than expected from monocultures indicated lower nitrogen availability in the third year. Complementary resource use was not solely dependent on the occurrence of N2 fixing legumes in mixtures, but also played a role in mixtures without legumes.
Synthesis. Our study shows that biodiversity effects through complementary nitrogen use can occur among potentially dominant grassland species. The degree of resource partitioning depends on growing conditions, with stronger complementarity effects under nutrient-rich conditions.
Several experimental studies that artificially assembled plant communities of different species richness have shown positive relationships between biodiversity and processes measured at ecosystem level. Two general mechanisms are assumed to contribute to the positive relationship between species richness and ecosystem processes: (i) selection for species with particular traits (the sampling or selection effect; Aarssen 1997; Huston 1997; Tilman 1997); and (ii) niche complementarity in resource use, facilitation or mutualistic interactions (the complementarity effect; Tilman et al. 1997; Loreau 1998). Physiological differences among species are a prerequisite for complementary or facilitation of resource use, which may result in an increased total resource capture at the community level. Many studies in experimental grasslands have shown that, in particular, N2 fixing legume species and their interactions with non-legume species are key contributors to overyielding with increasing species richness (Hooper & Vitousek 1997; Tilman et al. 1997; Symstad et al. 1998; Mulder et al. 2002; Spehn et al. 2005). However, the impact of legumes on community biomass production and nitrogen pools was variable (Spehn et al. 2002), and positive effects of species richness on community productivity were also found in experimental communities without legumes (van Ruijven & Berendse 2003).
Increased species richness does not necessarily result in increased above-ground biomass production compared with highly productive monocultures, as shown in agricultural studies of intercropping (Vandermeer 1989), semi-natural vegetation (Grime 2001; Thompson et al. 2005) and in ecological experiments (Hector et al. 1999, 2002; Troumbis et al. 2000; Cardinale et al. 2006). The occurrence of a positive relationship between species richness and biomass production has been attributed to species combinations with particular functional characteristics of the species involved or certain environmental conditions (Hooper & Dukes 2004). Previous studies have shown that a higher nutrient availability may increase biodiversity effects (Reich et al. 2001; He et al. 2002; Fridley 2002, 2003). Dukes (2001) presented a conceptual model that describes the potential importance of resource availability for overyielding. The model hypothesizes that overyielding only occurs in resource-rich conditions where dominant species are not able to utilize available resources completely and allow other species access to remaining resources. In contrast, under resource-poor conditions species may compete more strongly for resources and reduce resource partitioning according to the conceptual model.
Here we present results from a biodiversity experiment assembled from a pool of nine potentially dominant grassland species (‘dominance experiment’ of The Jena Experiment, Roscher et al. 2004). Species selection was restricted according to the criterion that these species are known to contribute substantially to biomass production and to grow almost as monocultures under high nutrient availability in semi-natural grasslands. The experimental set-up was determined by criticism directed towards previous experiments that the positive effects of species richness on biomass production were attributable to the greater chance of both highly competitive species and N2 fixing species being included in more diverse mixtures (Huston & McBride 2002; Thompson et al. 2005). We have shown that above-ground biomass production increases with species richness in our experiment with only dominant species (Roscher et al. 2005). However, resource availability in this experiment is not constant. Mowing without fertilizer addition resulted in the continuous export of nutrients with mown biomass. Symbiotic N2 fixation by legumes and N-inputs by airborne deposition (2.3 ± 0.1 g m−2 year−1; Oelmann et al. 2007a) may compensate for at least part of this nitrogen loss. Here, we compare the second and third year of this experiment to test the following hypotheses: (i) mixtures of potentially dominant species are able to exploit nutrients more completely than expected from monocultures; and (ii) changes in resource availability affect resource partitioning among species and alter the relationship between species richness and above-ground nitrogen pools, and between species richness and biomass production, but these effects should be less pronounced in mixtures with legumes.
site description and experimental design
The Jena Experiment is located in the floodplain of the River Saale in Jena (Thuringia, Germany, 50°55′ N, 11°35′ E, 130 m a.s.l.). The area around Jena has a mean annual temperature of 9.3 °C and a mean annual precipitation is 587 mm (Kluge & Müller-Westermeier 2000). The soil of the experimental area is a Eutric Fluvisol developed from up to 2 m thick fluvial sediments and almost free of stones. Due to the fluvial dynamics of the Saale River, soil texture varies between sandy loam and silty clay with increasing distance from the river. The site was converted from grassland to an arable field around 1960 and received high fertilizer inputs for the decades prior to the establishment of the experiment in 2002.
Semi-natural Central European mesophilic grasslands (Arrhenatherion community; Ellenberg 1988) served as a ‘target’ community to create a species pool for the experiment. These originally species-rich grasslands were traditionally used for haymaking. Agricultural intensification during the last century – with fast rotation, high fertilizer inputs and the use of high-yielding cultivars – led to a decline in species richness in these plant communities. Whereas the main experiment of the Jena Experiment is based on 60 species, the present study was carried out in a sub-experiment with nine species known for their potential to become dominant in highly productive grassland communities. These species include five grasses (Alopecurus pratensis L., Arrhenatherum elatius (L.) P. Beauv. ex J. Presl et C. Presl, Dactylis glomerata L., Phleum pratense L. and Poa trivialis L.), two legumes (Trifolium pratense L., Trifolium repens L.) and two non-legume herbs (Anthriscus sylvestris (L.) Hoffm., Geranium pratense L.).
In total, the dominance experiment comprises 206 plots measuring 3.5 × 3.5 m. Species richness ranges from one in monocultures to two, three, four, six and the complete set of nine species in mixtures. Individual species occur eight times at each species richness level in mixtures. All two-species combinations are present with the same frequency at each species richness level of mixtures. Each mixture was set up with an identical replicate. In total, the design contains 2 × 9 monoculture plots, 2 × 36 two-species mixtures (each possible species combination), 2 × 24 three-species mixtures, 2 × 18 four-species mixtures, 2 × 12 six-species mixtures and 8 nine-species mixtures. The experimental site was divided into four blocks perpendicular to the riverside following the gradient of soil characteristics. Mixtures were randomly allocated to the blocks, ensuring that each block contained an equal number of mixtures of a given species richness level and that identical replicates of a mixture did not occur in the same block. Plots were sown with a constant sowing density of 1000 viable seeds per m2 (adjusted for germination rates from laboratory tests) which were equally distributed among the species in multi-species mixtures. Although the non-leguminous herbs A. sylvestris and G. pratense initially grew slowly, 1 year after sowing all species were established successfully in all experimental plots to which they were added.
Plots were weeded regularly to prevent the establishment of species not sown with the original plant assemblages. The management was adapted to extensive meadows used for haymaking, and plots were mown twice a year to 5 cm height. Mown plant material was removed. The plots were not fertilized during the period of this study. A detailed description of the design and the establishment of the experiment can be found in Roscher et al. (2004).
Above-ground biomass was harvested twice in both years at estimated peak standing crop at the end of May and August before mowing of the plots. Two randomly placed samples of 20 × 50 cm were harvested 3 cm above soil surface leaving out the outer margin of 70 cm of the plots to avoid edge effects. Biomass samples were sorted to species, litter (dead plant material) and weeds, dried (70 °C, 48 h) to constant weight and weighed. Biomass values from both samples per plot were averaged to obtain species and community biomass of sown species (gdw m−2). Biomass samples per plot were pooled to analyse plant nitrogen. The samples of one replicate per mixture (growing on two different plots) of the May 2004 harvest were analysed per species. Dry plant material was ground to a fine powder. Approximately 20 mg were used for analysis with an elemental analyser (Vario EL Element Analyzer, Elementar, Hanau, Germany). Plant community nitrogen pools were calculated by multiplying nitrogen concentrations by above-ground biomass. Tissue nitrogen concentrations per plot and species (for the May harvest 2004) were used to assess the amount of biomass produced per gram nitrogen (gdw (g N)−1).
During the May harvest, material of fully expanded sun leaves of the legumes T. pratense and T. repens was sampled in monoculture and in one replicate per mixture where the legumes grew to determine N-isotope ratios and leaf-nitrogen concentrations. Isotope ratios were measured from approximately 3 mg of dried and finely ground leaf material with an isotope-ratio mass spectrometer (Delta C prototype IRMS, Finnigan MAT, Bremen, Germany). The δ15N values were calculated relative to the atmospheric nitrogen isotope ratio:
where R represents the molar ratio of 15N : 14N in a sample (Högberg 1997). These values were used to assess whether legumes were actively fixing nitrogen, because plant species which form symbiosis with N2 fixing bacteria usually have δ15N values closer to 0 than plants without such symbiosis. Nitrogen concentrations in leaf material were determined as described above.
To analyse inorganic nitrogen pools (ammonium, nitrate), soil samples were taken in March, June (after mowing) and October (after mowing). Three soil cores of 1 cm diameter were sampled from 0 to 15 cm depth in each plot and pooled. Fresh soil samples were stored at 4 °C and processed within 3 days after sampling. The samples were sieved over a 2-mm mesh sieve, and visible root parts were removed. Immediately after sieving, soil samples were extracted with 1 m KCl (50 mL on 5 g soil material). The filtered extracts were frozen until analysis for nitrate and ammonium with a Continuous Flow Analyzer (SAN Plus, Skalar, Erkelenz, Germany). Ammonium concentrations were below the detection limits (0.04 mg L−1 N) for almost all samples. Nitrate concentrations were expressed as µg g−1 dry soil after determining the water content of the soil samples gravimetrically (72 h, 105 °C).
complementarity of nitrogen use
OT is the observed nitrogen mass in the above-ground biomass in mixture, and ET is the sum of monoculture nitrogen masses in the above-ground biomass (Ei) of the component species divided by the number of species (ni) in the mixture. Total plant nitrogen in species mixtures would be higher than in monocultures (DT > 0) if species are complementary in their nitrogen use, while competition for nitrogen would either result in DT ≈ 0 if there are compensatory trade-offs among species or even in DT < 0 with interference competition (Hooper 1998).
Similar to the index of complementarity in above-ground nitrogen pools, we calculated the expected values of soil nitrate pools for each plot to test if mixed communities were depleting nitrate more efficiently than their corresponding monocultures. The calculation of DT NO3 is identical to that of DT (eqn 2), with OT as the observed mean soil nitrate pool per year, and ET as the expected nitrate pools based on monoculture nitrate pools (mean of two plots) of each species grown in mixture. Species were weighted by the proportion of each species in above-ground biomass in each plot. A value of DT NO3 < 0 indicates that mixtures are reducing soil nitrate pools to lower levels than would be expected from the corresponding monocultures, suggesting complementary resource use (Palmborg et al. 2005).
Apart from the objective to analyse effects of species richness, our experiment was designed to also test effects of individual species on ecosystem functioning (Roscher et al. 2004). However, because species richness and species identity are correlated in such experiments, they cannot be fully separated in statistical analyses. At least, the size of this correlation was exactly the same for all nine species (see also design used by Bell et al. 2005). The data were therefore analysed by two types of general linear models. In both types, blocks and the linear contrast of species richness were fitted first. Block effects accounted for the gradient in soil conditions at the field site and block-wise weeding and mowing management and data collection. In the first type of model, community composition (different species compositions within richness levels; random-effects factor) directly followed the linear contrast of species richness. In the second type of model the presence of each species or the presence of legumes as a functional group were tested before the remainder of community composition (the degree of this random term now decreased by one because of the additionally fitted identity contrasts). All analyses were performed as repeated-measures mixed-model anovas to account for differences between years and to test whether effects of species richness and community composition differed between years. Separate analyses for each year were performed to get detailed information about within-year effect sizes and significances. Data were log-transformed if necessary. All data analysis was done with S-Plus® 7.0 software (Insightful Corp., 2005, Seattle, WA).
above-ground nitrogen pools
Above-ground nitrogen pools averaged across all plots reached 14.3 (± 5.5) g N m−2 in the second year (2003) of the experiment and declined significantly by more than 30% to 10.0 (± 4.2) g N m−2 in the third year (2004; Table 1a, Fig. 1a–c). We found a positive effect of species richness on above-ground nitrogen pools that accounted for 10% of variation in 2003 (Table 1b). In contrast, species richness did not affect above-ground nitrogen pools in 2004 (Table 1c). Species composition within species richness levels explained a large percentage of variation in both years. In particular, the presence of legumes resulted in significantly higher above-ground nitrogen pools. This positive effect of legume presence was considerably larger in 2004 than in 2003 (explaining 31% vs. 11% of variation). Arrhenatherum elatius was the only non-legume species which affected above-ground nitrogen pools in both years, positively in 2003 and negatively in 2004 (Table 1).
|Source||d.f.||Above-ground nitrogen pool||Above-ground biomass||Biomass : N ratio||DT in nitrogen use|
|MS||% SS||F||P||MS||% SS||F||P||MS||% SS||F||P||MS||% SS||F||P|
|Block||3||58||1||4.18||0.008||220 317||2||6.91||< 0.001||222||1||4.72||0.004||1.09||2||6.77||< 0.001|
|Species richness (SR)||1||372||3||7.76||0.006||↑||2 949 659||10||30.23||< 0.001||↑||1 302||1||1.96||0.164||1.66||1||3.56||0.062|
|Species composition||98||48||40||3.45||< 0.001||97 580||33||3.06||< 0.001||663||72||14.11||< 0.001||0.47||31||2.88||< 0.001|
|Year (YR)||1||1973||17||93.65||< 0.001||3 705 714||13||64.79||< 0.001||5 809||6||71.30||< 0.001||38.14||28||146.90||< 0.001|
|Species richness × YR||1||267||2||12.65||0.001||755 031||3||13.20||< 0.001||161||< 1||1.98||0.163||1.80||1||6.92||0.010|
|Species composition × YR||98/89||21||18||2.83||0.001||57 195||19||2.32||< 0.001||81||9||1.79||0.002||0.26||17||2.62||< 0.001|
|Block||3||24||1||1.89||0.136||96 530||2||3.47||0.019||415||3||13.35||< 0.001||0.39||2||2.20||0.093|
|Species richness (SR)||1||633||10||14.77||< 0.001||↑||3 335 310||19||29.34||< 0.001||↑||271||1||0.84||0.363||3.45||6||7.44||0.008||↑|
|Arrhenatherum elatius||1||1000||16||30.32||< 0.001||↑||5 100 598||29||81.89||< 0.001||↑||233||1||0.72||0.399||3.52||6||8.21||0.005||↑|
|Dactylis glomerata||1||189||3||4.57||0.035||↓||66 605||< 1||0.58||0.447||6 344||17||24.16||< 0.001||↑||2.52||4||5.72||0.019||↓|
|Phleum pratense||1||63||1||1.48||0.226||610||< 1||0.01||0.942||1 523||4||4.88||0.030||↑||3.20||5||7.40||0.008||↓|
|Alopecurus pratensis||1||187||3||4.52||0.036||↓||280 180||2||2.50||0.117||200||1||0.61||0.435||0.68||1||1.48||0.228|
|Poa trivialis||1||180||3||4.35||0.040||↓||253 667||1||2.26||0.136||820||2||2.57||0.112||1.04||2||2.26||0.136|
|Geranium pratense||1||28||< 1||0.65||0.423||264 827||2||2.36||0.128||290||1||0.89||0.347||0.24||< 1||0.52||0.471|
|Anthriscus sylvestris||1||71||1||1.66||0.201||566 514||3||5.20||0.025||↓||427||1||1.32||0.253||6.08||10||15.21||< 0.001||↑|
|Trifolium pratense||1||353||6||8.89||0.004||↑||9 719||< 1||0.08||0.772||5 008||14||18.12||< 0.001||↓||0.13||< 1||0.27||0.602|
|Trifolium repens||1||142||2||3.40||0.068||122 083||1||1.07||0.302||4 668||13||16.67||< 0.001||↓||3.92||6||9.23||0.003||↓|
|Legumes||1||667||11||18.31||< 0.001||↑||1 750||< 1||0.02||0.902||11 715||32||56.53||< 0.001||↓||0.47||1||1.02||0.316|
|Species composition||98/89||43||67||3.32||< 0.001||113 697||63||4.09||< 0.001||325||87||10.44||< 0.001||0.46||66||2.65||< 0.001|
|Block||3||40||3||4.78||0.004||238 875||9||8.42||< 0.001||254||2||5.62||0.001||0.79||7||9.04||< 0.001|
|Species richness (SR)||1||5||< 1||0.17||0.679||363 088||5||9.37||0.003||↑||1 195||3||2.82||0.097||< 0.01||< 1||< 0.01||0.942|
|Arrhenatherum elatius||1||182||5||7.38||0.008||↓||617||< 1||0.02||0.900||7 360||15||20.86||< 0.001||↑||1.53||5||6.23||0.014||↓|
|Dactylis glomerata||1||91||3||3.57||0.062||21 828||< 1||0.56||0.456||4 989||10||13.23||< 0.001||↑||0.13||< 1||0.48||0.490|
|Phleum pratense||1||29||1||1.10||0.298||< 1||< 1||< 0.01||1.000||713||2||1.69||0.196||0.13||< 1||0.51||0.476|
|Alopecurus pratensis||1||< 1||< 1||0.01||0.925||129 802||2||3.43||0.067||491||1||1.16||0.285||0.11||< 1||0.42||0.521|
|Poa trivialis||1||31||1||1.17||0.281||209 855||3||5.68||0.019||↓||9||< 1||0.02||0.888||2.01||6||8.39||0.005||↑|
|Geranium pratense||1||8||< 1||0.31||0.580||68 139||1||1.77||0.186||484||1||1.14||0.288||< 0.01||< 1||< 0.01||0.999|
|Anthriscus sylvestris||1||23||1||0.88||0.352||255 978||3||7.01||0.009||↓||395||1||0.93||0.337||0.08||< 1||0.31||0.580|
|Trifolium pratense||1||799||22||43.68||< 0.001||↑||576 510||7||17.37||< 0.001||↑||9 223||19||27.65||< 0.001||↓||5.09||15||24.83||< 0.001||↑|
|Trifolium repens||1||167||5||6.72||0.011||↑||432||< 1||0.11||0.740||4 236||9||11.00||0.001||↓||6.07||18||31.28||< 0.001||↓|
|Legumes||1||1104||31||72.85||< 0.001||↑||416 628||5||11.96||< 0.001||↑||16 039||33||60.91||< 0.001||↓||0.07||< 1||0.27||0.607||↑|
|Species composition||98/89||26||72||3.12||< 0.001||38 738||49||1.37||0.060||424||86||9.41||< 0.001||0.26||69||2.97||< 0.001|
Above-ground nitrogen pools are an integrated measure that may be decomposed in biomass and plant nitrogen concentrations to get further insight into the mechanisms that drive their dynamics.
above-ground biomass production
Annual above-ground biomass production increased linearly with species richness in both years, but the slope of this relationship was lower in 2004 than in 2003. Averaged across all plots, biomass decreased from 817 (± 293) gdw m−2 in 2003 to 627 (± 195) gdw m−2 in 2004 (Table 1, Fig. 1d–f). Species composition within species richness levels explained a large percentage of variation in biomass production between plots in both years. In 2003, the presence of A. elatius had a large positive effect on biomass production (explaining 29% of the total variation). In contrast, the presence of legumes, in particular of T. pratense, had a larger positive effect in 2004 (explaining 7% of the total variation).
biomass produced per gram nitrogen (= biomass : n ratio)
Community biomass per gram nitrogen was not significantly related to species richness in both years (Fig. 1g–i, Table 1). However, biomass : N ratios increased by about 11% from 2003 to 2004 (59 ± 13 gdw (g N)−1 vs. 66 ± 15 gdw (g N)−1). Legume presence in mixtures led to significantly lower biomass : N ratios. In contrast, the presence of grasses, in particular of D. glomerata (in 2003 and 2004) and A. elatius (in 2004), increased the amount of biomass produced per gram nitrogen (see Table 1).
Analysis of nitrogen concentrations in tissue from individual species harvested in May 2004 showed that A. pratensis was the only species where biomass : N ratios increased with species richness. All other species did not show such variation in biomass : N ratios, either in response to species richness nor in response to the presence of legumes (Table 2). In general, grass species had the largest biomass : N ratios in comparison to herbaceous species with intermediate biomass : N ratios and legume species with the lowest biomass : N ratios (Fig. 2a). The biomass : N ratio of monocultures increased significantly from the 2003 to 2004 (F1,9 = 7.61, P = 0.022), and this effect varied only marginally among species (interaction term: F1,9 = 2.76, P = 0.076; Fig. 2b).
|Species||Block||Species richness||Legume presence||N|
|Arrhenatherum elatius||2.49||0.04||< 0.01||38|
|Dactylis glomerata||2.53||< 0.01||3.58||38|
complementarity in nitrogen use (dt)
The difference between above-ground nitrogen pools in mixtures and expected values from monocultures (DT) was positive in both years averaged across all species richness levels (test for overall mean ≠ 0: F1,89 = 221.96, P < 0.001 (2003); F1,89 = 7.66, P = 0.007 (2004); Fig. 1k–m). Nevertheless, complementarity in nitrogen use was significantly lower in 2004 than in 2003. In 2003, only 6% of mixtures had DT < 0. In 2004, 45 % of all mixtures had DT < 0, which indicates that competition for nitrogen was stronger in 2004 compared to 2003. Paralleling this result, DT was positively related to species richness in 2003, but not in 2004 (Table 1).
evidence for n2 fixation of legumes
δ15N values measured in leaf material of T. pratense and T. repens decreased significantly from 2003 to 2004 suggesting a higher reliance of legumes on symbiotic N2 fixation in the third year (paired t-tests, P < 0.001). While average δ15N values in leaf material of T. repens were significantly lower than zero in both years (2003: T37 = −4.56, P < 0.001; 2004: T37 = −19.93, P < 0.001) indicating symbiotic nitrogen fixation, average δ15N values in leaf material of T. pratense were significantly larger than zero in 2003 (T36 = 3.13, P = 0.003), but decreased below zero in 2004 (T36 = −3.61, P < 0.001). However, average nitrogen concentration in leaf material of both legume species declined from 2003 to 2004 (paired t-tests, P < 0.001).
biomass production of individual species
To test whether observed differences in biomass : N ratios at the community level between years were related to different biomass proportions of individual species, we compared the proportion of biomass of each species in mixtures between years and with initial values at the start of the experiment (sown proportion of each species). The grass species D. glomerata, A. pratensis and the herbaceous species A. sylvestris increased their biomass proportions significantly from 2003 to 2004, while we found significant decreases in biomass proportions in the grass species P. pratense, P. trivialis and the legume species T. repens (Fig. 3). However, the amount of these year-to-year changes was marginal in comparison to the differences among species. The grass species with the highest biomass : N ratio, A. elatius and D. glomerata, achieved higher biomass proportions than expected from sowing (P < 0.001) in both years. Biomass proportions of A. pratensis did not differ significantly from expected values in 2003 (P = 0.558) but were higher than expected in 2004 (P = 0.009), whereas biomass proportions of P. pratense were higher than expected in 2003 (P = 0.027), but not in 2004 (P = 0.612). Among the non-leguminous species with a low biomass : N ratio, the grass P. trivialis and the herbs A. sylvestris and G. pratense also achieved lower biomass proportions than expected in both years (P < 0.001). Biomass proportions of the legume T. pratense did not differ significantly from expected values (2003: P = 0.391; 2004: P = 0.110); biomass proportions of T. repens were lower than expected in both years (P < 0.001).
soil nitrate concentrations
Nitrate concentrations in the top-soil (0–15 cm) decreased with increasing species richness (Fig. 4a–c, Table 3). In particular, the presence of the grass species D. glomerata (in 2003), P. pratense (in 2003), A. pratensis (in 2003 and 2004) and A. elatius (in 2004) led to a decrease in soil nitrate concentration, while the presence of legumes caused an increase. In addition, we found significantly lower soil nitrate concentrations in 2004 compared to 2003 (Table 3, log-transformed data), although the magnitude of the change was rather small (2003: 1.43 (± 1.00) µgN ; 2004: 1.19 (± 0.65) µgN ). While observed over expected soil nitrate concentrations (DT NO3) decreased significantly with species richness in 2003, we found no effect of species richness in 2004. Averaged across all species richness levels the values were significantly lower than zero in 2003 (test for overall mean ≠ 0: F1,89 = 32.81, P < 0.001). In contrast, values were not significantly different from zero in 2004 (test for overall mean ≠ 0: F1,89 = 3.62, P = 0.060; Fig. 4d–f, Table 3).
|Source||d.f.||Soil nitrate||DT NO3 in soil nitrate|
|MS||% SS||F||P||MS||% SS||F||P|
|Species richness (SR)||1||0.271||4||6.60||0.012||↓||0.270||3||10.06||0.002||↓|
|Species composition||98/89||0.041||62||7.61||< 0.001||0.027||22||1.12||0.287|
|Year (YR)||1||0.116||2||14.07||< 0.001||0.243||2||7.15||0.009|
|Species richness × YR||1||0.030||1||3.63||0.060||0.260||2||7.65||0.007|
|Species composition × YR||98/89||0.008||12||1.39||0.486||0.034||28||1.54||0.018|
|Species richness (SR)||1||0.242||6||7.39||0.008||↓||0.530||9||17.14||< 0.001||↓|
|Arrhenatherum elatius||1||0.014||< 1||0.41||0.522||0.070||1||2.29||0.133|
|Dactylis glomerata||1||0.185||5||5.95||0.016||↓||0.021||< 1||0.68||0.413|
|Alopecurus pratensis||1||0.468||11||16.62||< 0.001||↓||0.003||< 1||0.10||0.751|
|Geranium pratense||1||0.016||< 1||0.50||0.480||0.006||< 1||0.18||0.672|
|Trifolium repens||1||0.776||19||31.00||< 0.001||↑||0.301||5||10.83||0.001||↑|
|Species composition||98/89||0.033||78||6.81||< 0.001||0.031||49||1.45||0.037|
|Block||3||0.037||5||7.22||< 0.001||0.182||11||8.90||< 0.001|
|Species richness (SR)||1||0.060||3||3.70||0.058||< 0.001||< 1||< 0.01||0.966|
|Arrhenatherum elatius||1||0.088||4||5.70||0.019||↓||0.003||< 1||0.11||0.744|
|Phleum pratense||1||0.040||2||2.51||0.116||0.498||10||23.47||< 0.001||↓|
|Alopecurus pratensis||1||0.172||7||11.77||< 0.001||↓||0.124||3||4.87||0.030||↓|
|Poa trivialis||1||0.004||< 1||0.27||0.607||0.007||< 1||0.25||0.617|
|Geranium pratense||1||0.023||1||1.41||0.238||< 0.001||< 1||< 0.01||0.995|
|Trifolium pratense||1||0.092||4||5.94||0.017||↑||0.004||< 1||0.14||0.711|
|Trifolium repens||1||0.244||11||17.61||< 0.001||↑||0.071||2||2.73||0.102|
|Species composition||98/89||0.016||69||3.13||< 0.001||0.027||49||1.30||0.107|
observational vs. experimental studies
A number of theoretical and observational studies in grassland systems have suggested that high nutrient availability leads to competitive exclusion of subdominant species and increased community biomass, resulting in a ‘humped-backed’ relationship between productivity and species richness (Grime 1973; DiTommaso & Aarssen 1989; Grace 1999). In contrast, experimental studies in which species richness was an explanatory, rather than response, variable usually found a monotonically increasing relationship between species richness and productivity which was even stronger on fertilized as opposed to unfertilized plots (Reich et al. 2001; He et al. 2002; Fridley 2003; Balvanera et al. 2006). This difference between observational and experimental studies can be explained by the different direction of causality in the two relationships (Schmid 2002; Schmid & Hector 2004). Because in the present experiment species richness was the cause and productivity the response, it is not surprising that we also found a stronger relationship in the year with presumably higher soil nutrient availability, corresponding to the results obtained in the previously mentioned biodiversity experiments where nutrient availability was deliberately manipulated.
evidence for complementarity
The set-up of our experiment with potentially dominant species was based on the assumption that these species should have strongly overlapping resource-use niches and thus a low potential for complementarity. However, our results indicate that this potential still existed, at least at the low levels of richness tested in this experiment. First, soil nitrate pools decreased with increasing species richness, suggesting that more diverse communities were more efficient in exploiting the available resource pool, facilitating an increased accumulation of biomass as compared with low-diversity communities. Decreasing soil nitrate pools with increasing species richness have often been found in studies with temperate grassland species (e.g. Tilman et al. 1996, 1997; Niklaus et al. 2001; Scherer-Lorenzen et al. 2003; Palmborg et al. 2005; Oelmann et al. 2007b). Only a few studies did not report such a relationship (e.g. Symstad et al. 1998). Second, our measures of complementarity (DT) for both below-ground nitrate pools and above-ground N pools showed that, on average, mixtures were depleting soil resources to lower levels, and accumulating N in biomass to higher levels, than expected based on monoculture performance. Furthermore, in 2003, the year with presumably higher soil nutrient availability, the strength of both measures of complementarity increased with species richness. Although with our statistical methods we could not distinguish between several potential mechanisms leading to increased uptake of nitrogen and higher N pools in biomass (for examples with isotopic tracers see McKane et al. 2002; Kahmen et al. 2006), the conclusion that potentially dominant species are able to use resources in a complementary way is consistent with our findings.
nutrient availability and other potential causes of the year-to-year variation
Although it is well known from semi-natural grassland systems that mowing and removal of above-ground biomass without fertilization may cause resource depletion in grassland systems (e.g. Olff & Bakker 1991) our 2-year study is not sufficient to prove a temporal trend. The amount of plant available mineral nitrogen in soils depends on a number of processes that are known to be highly variable in space and time, including plant uptake, ammonification and nitrification, and microbial immobilization (Hartwig 1998; Schimel & Bennett 2004). Because these processes are also dependent on soil water content, the extremely dry and hot summer in July and August 2003 in mid-Europe (Ciais et al. 2005) might have negatively influenced regrowth after the first harvest in May through both reduced water availability and reduced N availability as soil N mineralization rapidly slows down when soils dries. Consequently, soil nitrate levels were similar in October and June, while in the climatically ‘normal’ year 2004, N availability increased again towards the end of the growing season due to reduced plant uptake and continued mineralization (Oelmann et al. 2007b). The lower nitrate availability in early 2004 could thus be a lag-effect. The drought could also have affected N availability through effects on legumes: high summer temperatures and low water supply may reduce nodule formation, provoke shedding of nodules from the roots and finally negatively influence rates of N2 fixation (Sprent 1972).
Nevertheless, there is good evidence that soil resource levels were lower in the third compared with the second year. First, a full budgeting of both N and P pools during the same years on other plots of The Jena Experiment revealed particularly strong N losses due to harvest exports (Oelmann et al. 2007a). Second, increased amounts of biomass produced per gram nitrogen, both in mixed communities and in monocultures (Fig. 2b), and decreased biomass production and above-ground nitrogen pools in 2004 also suggested lower nitrogen availability. Although there is growing evidence that organic nitrogen may also play a role in plant nitrogen nutrition (e.g. Weigelt et al. 2005), its significance for plant nutrition under natural field conditions in fertile grassland soils seems to be rather low because these species strongly prefer inorganic N forms (Jones et al. 2005; Harrison et al. 2007).
effects of legumes and other species
A number of biodiversity experiments have shown that the strength of overyielding and complementarity is not constant over time (e.g. Tilman et al. 2001; Mulder et al. 2002; Pacala & Tilman 2002; Hooper & Dukes 2004; van Ruijven & Berendse 2005; Cardinale et al. 2007; Fargione et al. 2007). Mulder et al. (2002) and Fargione et al. (2007) suggested that a greater dependence on nitrogen fixed by legumes increases complementarity over time, while van Ruijven & Berendse (2005) supposed that complementarity in nutrient uptake among non-leguminous species and increased biomass produced per gram nitrogen of individual species contributed to this effect. In contrast to these reports, we found that complementarity in nitrogen use based on above-ground biomass pools (DT) was lower in the third year despite increased biomass : N ratios at the community level and also despite an apparently stronger dependence of legumes on symbiotic N2 fixation. Whereas high soil N availability often suppresses N2 fixation, legumes usually rely more on fixed nitrogen under low resource availability (Carlsson & Huss-Danell 2003). Neighbouring plants may benefit from this additional nitrogen source after mineralization of legume exudates, roots or above-ground plant tissue (Hartwig 1998). In contrast to other studies (Mulder et al. 2002; Spehn et al. 2002; Lee et al. 2003; Temperton et al. 2007), the presence of legumes in our experiment did not affect nitrogen concentrations in plant material of non-fixing plants growing in plots with legumes. The lack of such an increase does not necessarily contradict the view that legumes had a positive effect on non-legume species in these mixtures. Biomass fractions of T. pratense did not change during the study period (Fig. 3), but we found positive effects of this species on above-ground biomass production and complementarity in nitrogen yield only in 2004. These findings suggest that non-leguminous species benefited from additional nitrogen in plots with T. pratense and produced more biomass, while effects of lower resource availability became particularly evident in plots without legumes in 2004 (Fig. 1).
However, our results also provide evidence that complementarity among potentially dominant species can occur in mixtures without legumes. For example, the negative effects of the grasses A. pratensis, D. glomerata and P. pratense on observed over expected soil nitrate concentrations (DT NO3) suggested that these co-dominant grass species might be particularly efficient in acquiring soil resources and contribute to complementarity in nutrient uptake (Palmborg et al. 2005). In addition, A. pratensis was the only species in our experiment that increased biomass : N ratios with increasing community species richness. Biomass : N ratios are often used as estimate of nitrogen use efficiency (e.g. van Ruijven & Berendse 2005; Fargione & Tilman 2006) although tissue longevity and retranslocation efficiency also affect nitrogen use efficiency (Vitousek 1982; Berendse & Aerts 1987; Aerts & Chapin 2000). The higher biomass proportions of species with larger biomass : N ratios are consistent with resource-competition theory that predicts the dominance of species which are able to efficiently acquire and use limiting resources (Tilman 1990; Fargione & Tilman 2006; Harpole & Tilman 2006).
We conclude that diversity effects through complementary resource use or facilitation may occur even among highly competitive and potentially dominant species. Complementary resource use not only occurred in mixtures with N2 fixing legumes, but also played a role in mixtures without legumes. Our analysis of two consecutive years gave evidence that this complementarity is strongly dependent on growing conditions, with soil resource availability presumably playing a major role. Lower nitrogen availability probably decreased the potential for resource partitioning among these species and flattened the relationship between species richness and above-ground nitrogen pools and biomass production.
We are grateful to many people, who helped to manage the experiment and assisted during the biomass harvests, soil sampling and sample preparation for chemical analyses, in particular C. Bergmann, S. Eismann, E. Frohberg, S. Hengelhaupt, S. Junghans, E. Machalett, H. Scheffler, U. Wehmeier, D. Weigel and several student helpers. I. Hilke, S. Matthaei and H. Geilmann conducted chemical and stable isotope analyses. We gratefully acknowledge E.-D. Schulze for critical advice and R. Aerts, and two anonymous referees for useful comments on the manuscript. The Jena Experiment is funded by the German Science Foundation (FOR 456) with additional support from the Friedrich Schiller University of Jena, the Max Planck Society, and the Swiss Science Foundation (grant no. 31-65224-01).
1997) High productivity in grassland ecosystems: effected by species diversity or productive species?
Oikos, 80, 183–184.
2000) The mineral nutrition of wild plants revisited: a re-evaluation of processes and pattern.
Advances in Ecological Research, 30, 1–67.
2006) Quantifying the evidence for biodiversity effects on ecosystem functioning and services.
Ecology Letters, 9, 1146–1156.
2005) The contribution of species richness and composition to bacterial services.
Nature, 436, 1157–1160.
1987) Nitrogen-use-efficiency: a biologically meaningful definition?
Functional Ecology, 1, 293–296.
2006) Effects of biodiversity on the functioning of trophic groups and ecosystems.
Nature, 443, 989–992.
2007) Impacts of plant diversity on biomass production increase through time because of species complementarity.
Proceedings of the National Academy of Sciences, USA, 104, 18123–18128.
2003) Nitrogen fixation in perennial forage legumes in the field.
Plant and Soil, 253, 353–372.
2005) Europe-wide reduction in primary productivity caused by the heat and drought in 2003.
Nature, 437, 529–533.
1989) Resource manipulations in natural vegetation: a review.
Vegetation, 84, 9–29.
2001) Productivity and complementarity in grassland microcosms of varying diversity.
Oikos, 94, 468–480.
1988) Vegetation Ecology of Central Europe. Cambridge University Press, Cambridge.
2006) Plant species traits and capacity for resource reduction predict yield and abundance under competition in nitrogen-limited grassland.
Functional Ecology, 20, 533–540.
2007) From selection to complementarity: shifts in the causes of biodiversity-productivity relationships in a long-term biodiversity experiment.
Proceedings of the Royal Society London, Series B, 274, 871–876.
2002) Resource availability dominates and alters the relationship between species diversity and ecosystem productivity in experimental plant communities.
Oecologia, 132, 271–277.
2003) Diversity effects on production in different light and fertility environments: an experiment with communities of annual plants.
Journal of Ecology, 91, 396–406.
1999) The factors controlling species density in herbaceous plant communities: an assessment.
Perspectives in Plant Ecology, Evolution and Systematics, 2, 1–28.
1973) Competitive exclusion in herbaceous communities.
Nature, 242, 344–347.
2001) Plant Strategies, Vegetation Processes and Ecosystem Properties. John Wiley, Chichester.
2006) Non-neutral patterns of species abundance in grassland communities.
Ecology Letters, 9, 15–23.
2007) Preferences for different nitrogen forms by coexisting plant species and soil microbes.
Ecology, 88, 989–999.
1998) The regulation of symbiotic N2 fixation: a conceptual model of N feedback from the ecosystem to the gene expression level.
Perspectives in Plant Ecology, Evolution and Systematics, 1, 92–120.
2002) Interactive effects of diversity, nutrients and elevated CO2 on experimental plant communities.
Oikos, 97, 337–348.
2002) Overyielding in grassland communities: testing the sampling effect hypothesis with replicated biodiversity experiments.
Ecology Letters, 5, 502–511.
1999) Plant diversity and productivity experiments in European grasslands.
Science, 286, 1123–1127.
1997) 15N natural abundance in soil–plant systems.
New Phytologist, 137, 179–203.
1998) The role of complementarity and competition in ecosystem responses to variation in plant diversity.
Ecology, 79, 704–719.
2004) Overyielding among plant functional groups in a long-term experiment.
Ecology Letters, 7, 95–105.
1997) The effects of plant composition and diversity on ecosystem processes.
Science, 277, 1302–1305.
1997) Hidden treatments in ecological experiments: re-evaluating the ecosystem function of biodiversity.
Oecologia, 110, 449–460.
2002) Evaluating the relative strength of biotic versus abiotic controls on ecosystem processes.
Biodiversity and Ecosystem Functioning (eds
S. Naeem &
P. Inchausti), pp. 47–60. Oxford University Press, Oxford.
2005) Dissolved organic uptake by plants-an important N uptake pathway?
Soil Biology and Biochemistry, 37, 413–423.
2006) Niche complementarity for nitrogen – an explanation for the biodiversity and ecosystem functioning relationship?
Ecology, 87, 1244–1255.
2000) Das Klima ausgewählter Orte der Bundesrepublik Deutschland: Jena.
Berichte des Deutschen Wetterdienstes, 213, 1–290.
2003) Legume presence increases photosynthesis and N concentrations of co-occurring non-fixers but does not modulate their responsiveness to carbon dioxide enrichment.
Oecologia, 137, 22–31.
1998) Separating sampling and other effects in biodiversity experiments.
Oikos, 82, 600–602.
2002) Resource-based niches provide a basis for plant species diversity and dominance in arctic tundra.
Nature, 415, 68–70.
2002) How plant diversity and legumes affect nitrogen dynamics in experimental grassland communities.
Oecologia, 133, 412–421.
2001) A link between plant diversity, elevated CO2 and soil nitrate.
Oecologia, 127, 540–548.
2007a) Nitrogen and phosphorous budgets in experimental grasslands of variable diversity.
Journal of Environmental Quality, 36, 396–704.
2007b) Soil and plant nitrogen pools as related to plant diversity in an experimental grassland.
Soil Science Society of America Journal, 71, 720–729.
1991) Long-term dynamics of standing crop and species composition after the cessation of fertilizer application to mown grassland.
Journal of Applied Ecology, 28, 1040–1052.
2002) The transition from sampling to complementarity.
The Functional Consequences of Biodiversity: Empirical Progress and Theoretical Extensions (eds
S. Pacala &
D. Tilman), pp. 151–166. Princeton University Press, Princeton, NJ.
2005) Inorganic soil nitrogen under grassland plant communities of different species composition and diversity.
Oikos, 110, 271–281.
2001) Plant diversity enhances ecosystem responses to elevated CO2 and nitrogen deposition.
Nature, 410, 809–810.
2004) The role of biodiversity for element cycling and trophic interactions: an experimental approach in a grassland community.
Basic and Applied Ecology, 5, 107–121.
2005) Overyielding in experimental grassland communities – irrespective of species pool or experimental design.
Ecology Letters, 8, 419–429.
2003) Positive effects of plant species diversity on productivity in the absence of legumes.
Ecology Letters, 6, 170–175.
2005) Diversity–productivity relationships: initial effects, long-term patterns, and underlying mechanisms.
Proceedings of the National Academy of Sciences, USA, 102, 695–700.
2003) The role of plant diversity and composition for nitrate leaching in grasslands.
Ecology, 84, 1539–1552.
2004) Nitrogen mineralization: challenges of a changing paradigm.
Ecology, 85, 591–602.
2002) The species richness–productivity controversy.
Trends in Ecology and Evolution, 17, 113–114.
2004) The value of biodiversity experiments.
Basic and Applied Ecology, 5, 535–542.
2005) Ecosystem effects of biodiversity manipulations in European grasslands.
Ecological Monographs, 75, 37–63.
2002) The role of legumes as a component of biodiversity in a cross-European study of grassland biomass nitrogen.
Oikos, 98, 205–218.
1972) The effects of water stress on nitrogen-fixing nodules. IV. Effects of whole plant Vicia faba and Glycine max.
New Phytologist, 71, 603–611.
1998) Species loss and ecosystem functioning: effects of species identity and community composition.
Oikos, 81, 389–397.
2007) Positive interactions between nitrogen-fixing legumes and four different neighbouring species in a biodiversity experiment.
Oecologia, 151, 190–205.
2005) Biodiversity, ecosystem function and plant traits in mature and immature plant communities.
Functional Ecology, 19, 355–358.
1990) Mechanisms of plant competition for nutrients: the elements of a predictive theory of competition.
Perspectives on Plant Competition (eds
J.B. Grace &
D. Tilman), pp. 117–141. Academic Press, San Diego, CA.
1997) Distinguishing between the effects of species diversity and species composition.
Oikos, 80, 185.
1997) The influence of functional diversity and composition on ecosystem processes.
Science, 277, 1300–1302.
2001) Diversity and productivity in a long-term grassland experiment.
Science, 294, 483–485.
1996) Productivity and sustainability influenced by biodiversity in grassland ecosystems.
Nature, 379, 718–720.
2000) Hidden diversity and productivity patterns in mixed Mediterranean grasslands.
Oikos, 90, 549–559.
1989) The Ecology of Intercropping. Cambridge University Press, Cambridge.
1982) Nutrient cycling and nutrient use efficiency.
American Naturalist, 119, 553–572.
1997) Biodiversity and plant litter: experimental evidence which does not support the view that enhanced species richness improves ecosystem function.
Oikos, 79, 247–258.
2005) Preferential uptake of soil nitrogen forms by grassland plant species.
Oecologia, 142, 627–635.