SEARCH

SEARCH BY CITATION

Keywords:

  • biodiversity;
  • determinants of plant community diversity and structure;
  • ecosystem functioning;
  • niche differentiation;
  • nutrient depletion;
  • root ecology;
  • root interactions;
  • species interactions

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

1. Plant diversity has profound effects on primary production. Plant diversity has been shown to correlate with increased primary production in nutrient-limited grassland ecosystems. This overyielding has been attributed to vertical niche differentiation among species below-ground, allowing for complementarity in resource capture. However, a rigorous test of this longstanding hypothesis is lacking because roots of different species could not be distinguished in diverse communities.

2. Here, we present the first application of a DNA-based technique that quantifies species abundances in multispecies root samples. We were thus able to compare root distributions in monocultures of two grasses and two forbs with root distributions in four-species mixtures. In order to investigate if vertical niche differentiation is driven by soil nutrient depletion, the topsoil layer of the communities were either nutrient-rich or -poor.

3. Immediately in the first year, 40% more root biomass was produced in mixtures than expected from the monocultures, together with significant below-ground complementarity effects, probably preceding above-ground overyielding. This below-ground overyielding appeared not to be the result of vertical niche differentiation, as rooting depth of the community tended to decrease, rather than increase in mixtures compared to monocultures. Roots thus tended to clump in the very dense topsoil layer rather than segregate over the whole profile in mixtures. The below-ground overyielding was mainly driven by enhanced root investments of one species, Anthoxanthum odoratum, in the densely rooted topsoil layer without retarding the growth of the other species.

4.Synthesis. Conventional ecological mechanisms, such as competition for nutrients, do not seem to be able to explain the increased root investments of A. odoratum in mixtures compared to monocultures, with apparently little effect on the root growth of the other species. Instead, the observed root responses are consistent with species-specific root recognition responses. From a community perspective, the observed early below-ground overyielding may initiate the recently reported increased soil organic matter, mineralization and N availability and thus may ultimately be responsible for the higher productivity at high plant species diversity.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

Anthropogenic exploitation of the natural environment causes dramatic changes in the composition of ecological communities (Chapin et al. 2000; Hooper et al. 2005). The decline of plant diversity is one of the major issues in this context, as plants are the primary producers in terrestrial ecosystems. After two decades of conducting biodiversity experiments, it is well established that plants produce more above-ground biomass when they grow in mixtures than would be expected from their monocultures (Cardinale et al. 2007). Enhanced biomass production in diverse communities has been explained by selection and/or complementarity effects (Loreau & Hector 2001). The selection effect assumes that the most productive species dominates the biomass of the species mixtures as a result of chance. Complementarity or facilitation occur when species exhibit niche differentiation allowing for resources capture that is complementary in space or time. Cardinale et al. (2007) statistically showed that above-ground overyielding is largely due to species complementarity.

To date it is unclear whether above-ground overyielding and complementarity are mirrored underground, as most experiments have focused on shoot biomass only. Only a few biodiversity studies quantified total root biomass and report mixed results, ranging from 50% below-ground overyielding (Tilman et al. 2001; Dimitrakopoulos & Schmid 2004; Reich et al. 2004) to no overyielding (Spehn et al. 2000; Gastine, Scherer-Lorenzen & Leadley 2003), or even reduced biomass partitioning to roots in mixtures compared to monocultures (Bessler et al. 2009). Until now, below-ground complementarity has never been studied in diverse communities, as relative abundance at the species level was unknown for root mixtures. This omission was largely due to a lack of technical possibilities to explore root distribution in diverse communities. We recently developed a molecular method to quantify the proportional abundance of different species in mixed-species root samples (Mommer et al. 2008), allowing for tests of below-ground complementarity.

One commonly proposed mechanism to reach functional complementarity is vertical niche differentiation below-ground. This, for example, is assumed to occur between deep-rooting dicots and shallower-rooting grasses (Parrish & Bazzaz 1976; Berendse 1981, 1983; Dimitrakopoulos & Schmid 2004; von Felten & Schmid 2008; Levine & HilleRisLambers 2009). Complementarity would increase if roots of a given species grew away from zones of intense nutrient competition with neighbours (Gersani, Abramsky & Falik 1998; Semchenko, John & Hutchings 2007b), leading to a further vertical segregation of roots of the different species, reminiscent of the two barnacle species in the classic example of realized niche differentiation (Connell 1961). Nutrient depletion would then be a driving force behind niche differentiation (Casper, Schenk & Jackson 2003). Dicots are thus expected to root even deeper and grasses even shallower in species-rich mixtures than in monocultures. Such functional complementarity in vertical root distribution of different species should in theory lead to a deeper and more even distribution of roots for the community as a whole, and may lead to below-ground overyielding. This overproduction allows a more complete exploitation of the soil column and its available nutrients and in the end a higher above-ground biomass production.

Here, we report about a biodiversity experiment with four grassland species – two grasses and two dicots – with a below-ground focus. We use the molecular technique of Mommer et al. (2008) to unravel species abundances in root mixtures to test the following specific hypotheses: (i) vertical root distribution will be different among the four species, and these differences will be more pronounced in mixtures than in monocultures, leading to vertical segregation of root systems of the different species. As it takes time for root systems to establish, the vertical segregation will increase over time. (ii) Vertical segregation of root systems results in a deeper and more even distribution of roots at community level, and leads to below-ground overyielding.

As net nitrogen mineralization typically decreases with soil depth, most of the roots are present in the topsoil layer. Therefore, competition for nutrients, and thus nutrient depletion, is expected to be most intense in the topsoil. To investigate whether rooting patterns were driven by soil nutrient depletion, we included a treatment with a nutrient-poor topsoil layer, to simulate intense nutrient competition and experimentally speed up vertical niche differentiation. The third hypothesis is, therefore, as follows: (iii) If nutrient depletion is intensifying root competition and driving vertical niche differentiation, segregation of rooting patterns of the species should be enhanced in the poor topsoil treatment compared to the nutrient-rich control.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

Experimental setup and measurements

Two grasses, Anthoxanthum odoratum L. and Festuca rubra L., and two forbs, Leucanthemum vulgare Lamk. and Plantago lanceolata L., which were also used in another biodiversity experiment (van Ruijven & Berendse 2005), were grown in replicated monocultures and 1:1:1:1 mixtures in a substitutive design (Fig. S1) in plots (50 × 50 cm; 70 cm depth; grouped by three in polyethylene containers) in the Nijmegen Phytotron (http://www.ru.nl/phytotron). Soil depth was 64 cm, divided in a 52-cm thick layer of a rich soil mixture (2:1:1 (v:v:v) sand : loamy sand : potting soil) and a 12-cm layer consisting of nutrient-poor riverine sand. The order of the layers depended on the treatment: in the nutrient-poor topsoil treatment the sand layer was on top of the nutrient-rich layer, whereas in the nutrient-rich topsoil treatment the order was reversed. To determine the difference in nutrient status in the two topsoil types, extractable nutrient content of the freshly mixed soil was determined at the start of the experiment. Soil samples (20 g, 20 mm diameter, 0–10 cm deep) were diluted in 50-mL demineralized water, shaken for 1 h gently to dissolve the nutrients in the water and analysed with an Auto Analyser 3 system (Bran+Luebbe, Norderstedt, Germany) for nitrate and phosphate. Values of total extractable nitrogen (N) and phosphorus (P) (mg kg−1 dry soil) were 12.5 ± 1.3 and 0.6 ± 0.1, respectively, for the nutrient-poor topsoil, whereas these values were 910.4 ± 63.1 and 5.8 ± 0.3, respectively, for the nutrient-rich topsoil (which was also the bulk soil; N = 5). At the bottom of all containers was a 4-cm layer with pebbles, allowing drainage via the outlet at the bottom.

Assignment of the species (mono vs. mixture) × nutrient treatments occurred randomly to a total of 36 plots (12 containers of three plots each); resulting in monocultures N = 3–4; mixtures N = 4. Different replicates of species × nutrients were always in different containers. Five-week-old seedlings, originating from seeds of natural populations occurring in the river forelands of the river Rhine, near Nijmegen, the Netherlands, were planted in June 2006. In each plot 36 seedlings were planted (plant density 144 m2), but only the area of the inner 4 × 4 plants (32 × 32 cm) was used for measurements to avoid edge effects. In the growing season of 2006, plants were watered 2 L per plot three times a week with an automatic irrigation system (PRIVA, de Lier, the Netherlands), giving constant moisture levels. In 2007 this amount was increased to 4 L per plot, three times a week. In winters, soil moisture levels were maintained by watering manually once a week.

In late September in 2006 and 2007, standing shoot biomass at the end of the growing season (living and dead biomass) was harvested by clipping 1–2 cm above soil surface, a standard practice for hay meadows in which these species occur. In order to investigate standing root mass at the end of the growing season, soil cores (20 mm diameter, two in monocultures, four in mixtures) were taken in seven soil layers (0–6, 6–12, 12–18, 18–24, 24–40, 40–55, 55–66 cm depth) at the beginning of October 2006 and 2007. Distances to surrounding individual plants were equal (see Fig. S1 in Supporting Information). Roots (<2 mm) per soil increment were collected after carefully rinsing them with tap water. Fresh weight was determined with a microbalance (Sartorius, Nieuwegein, the Netherlands). Up to 100 mg of fresh root material was stored at −80 °C for molecular analyses. From a second subsample, root length was determined from root scans (600 dpi, Epson expression 10000 XL; Regent Instruments, Quebec, Canada) with WinrhizoTM software (‘very pale roots’- option; Regent Instruments). Shoot and root dry weights were determined after drying at 70 °C for 48 h.

Quantitative real time polymerase chain reactions (RT-PCR) with primers for non-coding species-specific markers were performed on genomic DNA extracts from every mixed root sample separately (Mommer et al. 2008). Using this technique, species abundance was quantified in soil layers 1, 2, 4 and 6, comprising 60% (2006) and 80% (2007) of all root mass. Analyses were performed on the basis of 100 mg fresh root mass. Data were recalculated in terms of dry weight as the ratio between dry weight and fresh weight of roots was constant (mean ± SE: 0.18 ± 0.01, N = 50).

For the analysis of N uptake, all shoot material of individual plants (two per species per plot) was dried and pulverized. Of this shoot material, 2–2.5 mg was used in a nitrogen analyser (EA 1110; Carlo Erba – Thermo Electron, Milan, Italy) in combination with a mass spectrometer (DEltaPlus; Thermo Finnigan, Bremen, Germany.

Calculations and statistical analyses

All calculations and statistical analyses on roots were performed on average plot values: i.e. root data of two cores were pooled in monocultures, four cores in mixtures.

Mean rooting depth was calculated as the sum of the amount of roots in soil layeri multiplied by the mean depth of layeri divided by the total amount of roots in all layers. Analyses at the species level are based on root mass density as determined from the four layers using the RT-PCR technique. Coefficients of variation (CV) were determined as the standard deviation of root mass across the soil profile divided by the mean. In order to compare the species results of mean rooting depth and CV to the responses of the community as a whole, analyses at the community scale were also based on these four soil layers. However, analyses on the seven soil layers resulted in qualitatively similar results.

Net biodiversity effects, complementarity effects, selection effects of shoot and root biomass were calculated according to Loreau & Hector (2001). In this so-called additive partitioning method, the net difference in yield for a mixture (ΔY, the net effect) is equal to the observed yield YO minus the expected yield YE (i.e. the average monoculture yield of the component species):

  • image

The net effect is assumed to be the sum of the complementarity effect and the selection effect. The complementarity effect is calculated as:

  • image

in which N is the number of species, ΔRYi is the difference between the observed and expected relative yield for species I and Mi is the biomass in the monoculture of species i.

The selection effect is calculated as:

  • image

For roots, these biodiversity effects were calculated from the roots in all seven soil layers. Species abundances of roots in soil layer 3 were estimated from ratios in soil layer 4, roots in soil layers 5 and 7 from ratios in soil layer 6. To compare observed and expected values, deviation from expected values (D) were calculated as ln(observed)−ln(expected) for each species and layer. The expected value was always based on the average of the monoculture(s); i.e. ¼ of the monoculture(s). Root : shoot ratios were calculated on the basis of above- and below-ground biomass values per m2.

Statistical analyses were performed using full-factorial univariate anovas (General Linear Model, SPSS 15.0), with diversity (monoculture vs. mixture) and treatment (rich vs. poor topsoil layer) as fixed factors. When analyses were performed at the species level, species was also included as a fixed factor. Analyses were always run separately for the two different years. In cases where D values were used, significant levels of the intercepts of the predicted model were used to determine whether these values deviated significantly from zero. This was also done to evaluate net, complementarity and selection effects. Further differences in performance (of species) between monoculture and mixture were determined using t-tests (Compare Means, SPSS 15.0). If needed, variables were transformed in order to meet assumptions of anova.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

Community effects: below-ground overyielding and complementarity

We observed a marked diversity-induced overyielding below-ground in the rich topsoil treatment (Fig. 1; Table 1). Immediately after the first growing season mixtures had produced up to 35% more root biomass than expected on the basis of the monocultures (mean relative yield total RYT2006RichTop = 1.35). This below-ground overyielding, also indicated by a significant net biodiversity effect (Table 1), was mainly caused by a significant positive complementarity effect. Selection effects were small and negative, indicating that the subordinate species contributed more to the root growth stimulation than the highly productive, dominant ones. The positive effect of diversity on root biomass in the rich topsoil treatment increased over the second growing season (mean RYT2007RichTop = 1.48; Fig. 1, Table 1).

image

Figure 1.  Above- and below-ground biomass (g dry weight m−2) for the four monocultures (simple bars), expected and observed values for the mixtures (hatched and open stacked bars, respectively) in the rich and poor topsoil treatment at the end of the growing seasons of 2006 and 2007. Anthoxanthum odoratum and Festuca rubra are grasses, Leucanthemum vulgare and Plantago lanceolata are dicots. The expected values are based on the average of the monocultures. Data are mean ± SE, N = 3–4. Significant differences (P < 0.05) between expected and observed values are indicated by an asterisk, indicating significant over- or underyielding.

Download figure to PowerPoint

Table 1.   Biodiversity effects for root, shoot and total biomass (g m−2), as well as shoot nitrogen (N) accumulation (g m−2), in both treatments (rich topsoil and poor topsoil) in 2006 and 2007. Significant deviations (P < 0.05) from the intercept of the predicted model on square-root transformed values are given in bold. Data are mean ± SE, N = 3–4 plots
Rich topsoil treatmentRootShootTotalN uptakeshoot
2006Overall net effect112.0 ± 27.771.3 ± 124.3183.3 ± 112.2−3.61 ± 2.59
Complementarity effect112. 5 ± 24.894.1 ± 118.2179.9 ± 112.2−3.45 ± 2.55
Selection effect−0.5 ± 0.02−22.8 ± 7.23.4 ± 5.0−0.15 ± 0.05
2007Overall net effect232.3 ± 94.7−114.0 ± 29.9118.3 ± 102.5−1.28 ± 0.31
Complementarity effect315.0 ± 104.4−142.6 ± 32.626.1 ± 100.8−1.53 ± 0.29
Selection effect−82.7 ± 9.728.6 ± 12.692.2 ± 24.20.25 ± 0.06
Poor topsoil treatment
2006Overall net effect5.5 ± 46.6162.8 ± 69.3168.4 ± 73.1−0.21 ± 1.24
Complementarity effect7.6 ± 46.9138.0 ± 64.6171.5 ± 77.3−0.61 ± 0.96
Selection effect−2.1 ± 0.324.8 ± 6.07−3.2 ± 10.20.41 ± 0.31
2007Overall net effect342.8 ± 158.6−13.0 ± 71.6329.8 ± 150.6−0.35 ± 0.55
Complementarity effect390.8 ± 166.2−16.4 ± 68.6281.6 ± 148.8−0.48 ± 0.59
Selection effect−48.0 ± 7.63.4 ± 4.548.2 ± 25.80.13 ± 0.03

Biomass accumulation both below- and above-ground was only slightly reduced in the poor topsoil compared to the rich topsoil treatment (Fig. 1). However, overyielding and complementarity effects were retarded in the poor topsoil compared to the rich topsoil as they became only significant in the second year (mean RYT2007PoorTop = 1.62; Fig. 1, Table 1).

Above-ground, no increase of biomass production was yet apparent in the mixtures in the first 2 years (Fig. 1, Table 1), which appears to be common in the first years of biodiversity experiments, particularly in the absence of nitrogen-fixing legumes (van Ruijven & Berendse 2005). For the rich topsoil treatment, net biodiversity effects for the shoot were even negative in the second year (Table 1). The biodiversity effects for the community as a whole were not apparent, as clear significant effects in roots were counterbalanced by negative effects in the shoot.

No evidence for enhanced vertical niche differentiation in mixtures

Six weeks after planting, all species had grown roots already to the deepest soil layers (minirhizotron observations, data not shown), but most roots were allocated to the upper layers. Over both treatments, 30% and 40% of the dicot and grass roots, respectively, were present in the first 12 cm of soil after the first growing season. In this topsoil layer root densities were rather high, ranging from 132 up to 623 m dm−3 for P. lanceolata and F. rubra, respectively. These densities increased even further in the second year as all species expanded their root system mainly in this dense upper soil layer (Fig. 2): 70% of the roots were allocated to the topsoil (top 12 cm).

image

Figure 2.  Rooting patterns of different species in monocultures and mixtures. Observed (in actual four-species mixtures) and expected (based on the respective monocultures) root distributions (circles and lines; g dm−3) and mean rooting depth (diamonds; cm) of the different species in both treatments (rich or poor top soil; hatched bar indicates the position of the nutrient-poor sand layer within the container) and years. Anthoxanthum odoratum and Festuca rubra are grasses, Leucanthemum vulgare and Plantago lanceolata are dicots. Data are mean ± SE, N = 3–4. Asterisks indicate significant differences in mean rooting depth between monocultures and mixtures: *P < 0.05; **P < 0.01. Please note the different scales of root mass density among species and years.

Download figure to PowerPoint

Festuca rubra and P. lanceolata were the most selective species with regard to root placement. In monocultures, these species had 30% and 46% more roots in the rich than in the poor top soil layers, respectively. In contrast, L. vulgare had 20% less roots in the rich compared to the poor topsoil. Root densities in rich and poor topsoil became more similar after the second growing season (Fig. 2).

Fundamental niches in terms of mean rooting depth in monocultures differed significantly among species (Table 2; Fig. 2). The differences among the species were less pronounced than anticipated, as they showed a wide distribution with considerable overlap. Moreover, there was only a non-significant trend towards grasses rooting more superficial than the dicots (mean rooting depth of grasses vs. dicots in 2006 12.1 and 13.9 cm, respectively; F = 2.81; d.f. = 1; P = 0.12; Fig. 1). All species decreased their mean rooting depth in 2007 compared to 2006. In contrast to our expectation, the poor topsoil treatment did not result in significant deeper rooting of the species (Table 2; Fig. 2).

Table 2. anova results, split per year, for mean rooting depth with species, diversity (mono vs. mixtures) and treatment (i.e. rich vs. poor topsoil) as fixed factors. Overall rooting depths were significantly different among species. General diversity and treatment effects were not apparent. To explore the significant interaction between species and diversity, another anova, split per year and species, showed that diversity effects were significant for Anthoxanthum odoratum in both years (shallower rooting) and for Leucanthemum vulgare in 2006 (deeper rooting; analysis not shown). *P < 0.05; **P < 0.01; ***P < 0.001
YearSourced.f.F-value
2006Species 317.5***
Diversity 10.77
Treatment 10.41
Species × Div 35.14**
Species × Treat 34.10*
Div × Treat 16.69*
Sp × Div × Treat 30.79
Error (d.f., MS)38, 4.95 
2007Species 39.30***
Diversity 10.23
Treatment 10.14
Species × Div 33.95*
Species × Treat 31.40
Div × Treat 16.19*
Sp × Div × Treat 30.57
Error (d.f., MS)38, 2.31 

Applying the molecular tools to quantify species contributions showed that the niches shifted under interspecific competition, but the responses differed among species, treatment and year (Fig. 2; Table 2). Leucanthemum vulgare rooted significantly deeper in mixtures than in monocultures of the poor topsoil treatment in the first year of the experiment (2006), but not in the second year (2007). More pronounced was the response of A. odoratum, rooting significantly shallower in mixtures than in monocultures in the rich topsoil treatment in both years. Festuca rubra and P. lanceolata did not show significant differences in mean rooting depth in mixture vs. monoculture. While for two of the four species – A. odoratum and (partly) L. vulgare – differences in mean rooting depth were more pronounced in mixtures than in monocultures, vertical segregation of root systems seemed to occur.

At the community level, the below-ground vertical segregation of the two species did not result in a deeper and more even distribution of roots. Mean rooting depth of the total community was not increased in mixtures compared to monocultures (Table 3). Different topsoil layers did not affect the mean rooting depth of the community either (Table 3). Moreover, there was a non-significant trend for larger rather than smaller coefficients of variation of root mass over soil layers in mixtures compared to monocultures (Table 3), suggesting that root distributions tended to clump rather than be more even in the diverse community.

Table 3.   Mean rooting depth (cm, MRD) and coefficient of variation of root mass over soil layers (CV) and anova results for the community as a whole, calculated from root mass distribution over four soil layers (0–6; 6–12; 12–24; 40–55 cm depth). Treat. indicates the top soil treatment (nutrient-rich or poor); Div. indicates the diversity level of the community (monoculture vs. four-species mixture). Mean rooting depths were ln-transformed in order to meet anova assumptions. *P < 0.05; †P < 0.10; N = 3–4 plots
YearTreat.Div.MRDCVSourceD.f.MRDCV
Mean ± SEMean ± SEF-valueF-value
2006Rich topMono13.2 ± 0.70.57 ± 0.06Div.11.73.1
Mix11.7 ± 0.80.69 ± 0.07Treat.10.33.9†
Poor topMono12.3 ± 0.10.71 ± 0.05Div. × Tr.10.60.1
Mix11.9 ± 0.70.80 ± 0.04Error (MS)91.6118.6
2007Rich topMono6.4 ± 0.021.36 ± 0.02Div.10.30.9
Mix6.1 ± 0.21.47 ± 0.01Treat.12.38.7*
Poor topMono6.5 ± 0.71.32 ± 0.07Div. × Tr.11.73.3
Mix7.2 ± 0.31.29 ± 0.04Error (MS)90.446.7

Enhanced root growth of A. odoratum in mixtures

An appropriate null-hypothesis based on interspecific competitive equivalence is that all species would produce ¼ of the root mass in the mixtures compared to the monocultures (dotted line, Fig. 3). If species differ in competitive ability, some species would do better (above the dotted line of Fig. 3), and some worse (below the dotted line). Using the molecular tools to unravel the species abundance in the mixtures, we found that most root masses in mixtures were as high or higher than expected from their respective monocultures, as most points were above the dotted line, but not significantly different from the 1 : 4 expectation (Fig. 3). Only a few root masses of L. vulgare in mixtures were significantly lower than expected in the null-hypothesis (indicated with ‘–’ in Fig. 3). In contrast, most root masses of A. odoratum in mixtures, but also some root masses of other species, were significantly larger than the null hypothesis predicted (indicated with ‘*’ in Fig. 3). Moreover, the A. odoratum root masses of the topsoil were not significantly different from the 1 : 1 line (indicated with an arrow in Fig. 3), indicating that root mass in mixtures equals root mass in monocultures in the top soil layer, while there were only one-fourth of the number of individual plants in the mixtures than the monocultures.

image

Figure 3.  Root mass density in mixtures as a function of root mass density in monoculture of the different species in the different soil layers (0–6; 6–12; 12–24; 40–55 cm depth). Data are presented separately for the rich and poor topsoil treatment in both years. The dotted line represents the null-hypothesis: in the mixtures all species would produce ¼ of the root mass of the monocultures, assuming that intraspecific competition equals interspecific competition. Points that are significantly lower than expected from the null-model are indicated with – (only L. vulgare). Points that are significantly larger than the null-model are indicated with *. The solid line indicates where the root biomass in mixture is equal to those of the monocultures, despite the number of plant individuals being only one-fourth of the monocultures, and points not significantly different from this line are indicated with an arrow (Anthoxanthum odoratum in topsoil layer only). Because root biomass declined with depth (Fig. 1), both axes can also be read as the inverse of soil depth. Anthoxanthum odoratum and Festuca rubra are grasses, Leucanthemum vulgare and Plantoago lanceolata are dicots. Data are mean ± SE, N = 3–4.

Download figure to PowerPoint

In contrast to A. odoratum, the other grass species in the experiment, F. rubra, was strikingly unresponsive to growth in monoculture vs. mixture. Festuca rubra had by far the largest root system (Fig. 1), but did not take advantage of the lower root length density of the other species when growing in the mixture (Fig. 3).

The described root patterns of A. odoratum and the other species occurred in the rich top treatment immediately in the first growing season, whereas in the poor topsoil treatment this took more time. After two growing seasons, however, the patterns were largely similar in the two treatments.

Over all species, root : shoot ratios significantly increased over time (analysis on transformed value (ln + 1): F = 112.7, d.f. = 1; P < 0.001; Fig. 4), which was to be expected as nutrients were depleted over time with the removal of above-ground biomass from the plot. Root : shoot ratios were also significantly higher in the poor topsoil than in the rich topsoil (F = 9.2, d.f. = 1, P < 0.005; Fig. 4). However, the largest allocation shift was observed in A. odoratum, where the enhanced root mass production in mixtures resulted in a threefold root:shoot ratio increase compared to monocultures (Fig. 4). The other species did not show significant changes in root:shoot ratio in the rich topsoil treatment. In the poor topsoil, F. rubra and P. lanceolata even had significantly decreased root:shoot ratios in mixture compared to monoculture in the first growing season (Fig. 4).

image

Figure 4.  Allocation patterns to root and shoot given by root : shoot ratios of the different species in monoculture and mixture in both years (2006 and 2007) in the rich and poor topsoil treatment. Anthoxanthum odoratum and Festuca rubra are grasses, Leucanthemum vulgare and Plantago lanceolata are dicots. Data are mean ± SE, N = 3–4. Significant differences between monocultures and mixtures per species and year on ln+1-transformed values are indicated with asterisks: **P < 0.01; ***P < 0.001; $P < 0.1.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

While species differences in phenology of root growth (Fargione & Tilman 2005), nitrogen preference (McKane et al. 2002; von Felten et al. 2009) and specific interactions with soil organisms (Kardol et al. 2007; van der Heijden, Bardgett & van Straalen 2008) may all contribute to overyielding and complementarity in species-rich communities, species differences in rooting depth are commonly assumed to form a prime axis of niche differentiation below-ground (Parrish & Bazzaz 1976; Berendse 1981, 1983; Dimitrakopoulos & Schmid 2004; Schenk & Jackson 2005; von Felten & Schmid 2008; Levine & HilleRisLambers 2009). However, using a new molecular technique that quantifies root distributions of individual species in mixed communities, we have not been able to obtain support for this longstanding hypothesis in ecology. Species indeed differed in their fundamental and realized niches regarding mean rooting depth, but there was a wide overlap in their distribution. At the community level, overall rooting depth tended to decrease, rather than increase, in mixtures compared to monocultures, and roots tended to clump in the very dense topsoil layer rather than segregate over the whole profile in mixtures. Although the trend from the first to the second year was in the opposite direction, it may take more than the two growing seasons of our measurements before vertical niche complementarity becomes apparent below-ground.

As in other studies (Tilman et al. 2001; Dimitrakopoulos & Schmid 2004; Reich et al. 2004), we observed below-ground overyielding in our experiment, which was apparent immediately in the first year (Fig. 1, Table 1), preceding overyielding above-ground. Below-ground overyielding mainly occurred in the dense topsoil layers and increased over time (Fig. 2). This was mainly the result of enhanced root growth of A. odoratum, with little reduction in root biomass of the other species. Here, we discuss the possible underlying mechanisms and ecosystem consequences of the enhanced standing root mass with increasing diversity.

A limited role of nutrient responses in understanding the interspecific root responses

How can the increased root mass production of A. odoratum in mixtures compared to monocultures be explained by conventional mechanisms of resource foraging and biomass allocation? Root mass increase could be the prelude to overall dominance if A. odoratum would be the competitively superior species. However, root mass of the other species was only slightly impeded (Fig. 3) and selection effects were negative rather than positive (Table 1), underscoring that A. odoratum was not the most productive species in our experiment, neither above- nor below-ground (Fig. 1), and never became dominant in mixtures. Indeed, this species was always among the three least productive species out of eight common grassland species in the 9-year time series of the Wageningen biodiversity experiment (van Ruijven & Berendse 2009). In the Wageningen monocultures of A. odoratum, standing above-ground biomass was c. 200 g m−2 in the first year, and gradually decreased to c. 80 g m−2 after 9 years. Monocultures of F. rubra and other species initially yielded 250–500 g m−2, but decreased to 180–250 g m−2; which is still more than double the production of A. odoratum. Moreover, A. odoratum did not have the highest relative abundance in 4- and 8-species mixtures over the 9 years (25–20%, respectively), since F. rubra (30–15% in 4- and 8-species mixtures, respectively) and other species (i.e. P. lanceolata 40–30% in 4- and 8-species mixtures, respectively; Centaurea jacea (70–40%, respectively)) have similar or higher relative abundances. Finally, A. odoratum did not outcompete the other species in 4- and 8-species mixtures over the 9 years that the experiment was running.

The higher root mass of A. odoratum in mixture was associated with a remarkable threefold increase in root : shoot ratio of A. odoratum. A stimulation of root allocation of this magnitude seems at odds with known responses to nutrient depletion of the soil (Poorter & Nagel 2000). We observed significant effects related to nutrient depletion (significant effects of time and topsoil treatment) on root : shoot ratio, but these were not of such an extraordinary magnitude (Fig. 4). Moreover, the root : shoot ratio shift of A. odoratum occurred independent of the nutrient treatment of the topsoil, and in both years of study.

Furthermore, the lack of essential differences between the rich and poor topsoil treatment also indicates that it is not simply the availability of nutrients that determine the interspecific root responses. Based on these arguments, we suggest that competitive and allocation responses to soil nutrients alone are unlikely to drive the elevated root mass in mixtures vs. monocultures.

Alternative explanations for the species-specific root interactions

Recent advances in root ecology suggest that root growth and distribution can be tuned to the identity of neighbouring roots (Schenk 2006; de Kroon 2007) by distinguishing roots of the same physiological individual from roots of a different genotype of the same species (Maina, Brown & Gersani 2002; Gruntman & Novoplansky 2004; Dudley & File 2007) or roots of a different species (Huber-Sannwald et al. 1998; Bartelheimer, Steinlein & Beyschlag 2006; Li et al. 2007; Semchenko, Hutchings & John 2007a). Responses to neighbours range from inhibition to stimulation of root growth (Mahall & Callaway 1991; Li et al. 2006; Semchenko, John & Hutchings 2007b). The mechanisms behind root recognition responses are yet obscure, but non-nutritious cues, such as root exudates (Bais et al. 2006) and/or root rhizosphere biota (Kardol et al. 2007; van der Heijden, Bardgett & van Straalen 2008) may be involved since the root responses appear very species-specific and density-dependent.

We would argue that the root biomass responses of A. odoratum are consistent with the operation of such non-nutritious cues: growth enhancement at low densities (i.e. in mixtures) and/or inhibition of root growth at high densities (monocultures), without affecting the root growth of the other species. Density-dependent regulation of root growth in monocultures and mixtures may have occurred via accumulation of autotoxic root exudates (Bais et al. 2006) and/or host-plant specific pathogens (Van der Putten & Peters 1997; Westover & Bever 2001). Alternatively, root growth of A. odoratum in mixtures may have been stimulated by the actual presence of roots of other species, directly, due to growth-stimulating root exudates from these species and/or indirectly, due to root growth-promoting rhizobacteria (Bais et al. 2006; Berg & Smalla 2009). For our study species, A. odoratum, it has been shown that seedlings grow better on natural root growth-promoting isolates of Bacillus of other species than on its own isolates (Westover & Bever 2001), suggesting that root growth promoting bacteria may play a role in grassland systems. A more indirect explanation for the enhanced root growth of A. odoratum in diverse communities would be that high plant diversities affect soil microbial communities (Kowalchuk et al. 2002; Bartelt-Ryser et al. 2005), leading to suppression of (root) pathogens (i.e. soil suppressiveness (Weller et al. 2002); cf. studies in leaves (Mitchell, Tilman & Groth 2002)), although it usually takes several years for the microbial communities to build up (Steinbeiss et al. 2008). Such pathogens should then be more deleterious for A. odoratum than for the other species of our system in order to explain our results.

We conclude that our results are consistent with non-nutritional, density-dependent feedback mechanisms as reported for the spread of invasive species (Bais et al. 2003; Levine et al. 2006) and plant succession (Bever 1994; Kardol et al. 2007). However, although the interactions below-ground may be essentially non-nutritious and depending on species identity, the effects may improve the nutritional status in mixtures vs. monocultures (cf. Li et al. 2007) or increase competitiveness for soil nutrients. Future research is needed to demonstrate what interactions are really involved and to pinpoint the exact mechanisms in this apparent species-specific root recognition.

Ecosystem implications

The ecosystem implications of the below-ground overyielding reported here and in other studies (Tilman et al. 2001; Dimitrakopoulos & Schmid 2004; Reich et al. 2004) may be considerable, as the higher root mass in these grassland mixtures appears to persist with possibly important consequences for the soil C and N budget. In the Wageningen biodiversity experiment (van Ruijven & Berendse 2005, 2009), root masses were 40% higher in four-species mixtures than in the average monoculture after 9 years (L. Mommer & J. van Ruijven, unpubl. data). Similarly, in the Cedar Creek biodiversity experiment, total root mass was 50–200% higher in the eight-species mixtures than in the monocultures, 12 years after the start of the experiment (Fornara, Tilman & Hobbie 2009). Given the high turnover rates of roots (Van der Krift & Berendse 2002), a higher root mass production will lead to increased organic matter input into the soil. Indeed, after 4 years in the Jena Biodiversity Experiment, carbon stocks were significantly higher in plots with higher species diversity (Steinbeiss et al. 2008). In the long-term, increased organic matter input can increase mineralization and N-availability (Fornara, Tilman & Hobbie 2009), which in turn is held responsible for the higher productivity in more diverse plant communities.

Conclusions

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

Our experimental grassland communities revealed below-ground species complementarity as expected, but found no evidence for the associated vertical niche differentiation that is generally assumed. While the exact mechanisms for the observed root responses remain obscure, the results are consistent with species-specific non-nutritional root interactions. The root growth stimulation in species mixtures may be the trigger to enhanced carbon sequestration and N availability in the soil, finally leading to enhanced above-ground production in diverse plant communities. If these novel interactions indeed play a role in below-ground overyielding, they may be the drivers of fundamental ecosystem processes.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

We thank Jelle Eygensteyn for N measurements; Jan van Walsem and Frans Möller for help during harvests and Jean Knops, Wim van der Putten, Ronald Pierik and Eric Visser for comments on an earlier version of the manuscript. L.M. is supported by Netherlands Organization for Scientific Research (NWO) VENI grant 016091116.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

Figure S1. Experimental setup of a mixture plot.

Figure S2. Nutrient uptake at the community scale.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
JEC_1702_sm_FigS1.doc74KSupporting info item
JEC_1702_sm_FigS2.doc76KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.