Species and experimental setup

The two species investigated commonly co-occur in hay meadows in central-northern Europe. The research was conducted in the Phytotron of the Radboud University Nijmegen (http://www.ru.nl/phytotron) in containers with separate units of 50(w)×50(l)×70(h) cm each. It is situated under a transparent rain shelter (high-quality commercial greenhouse film, 90% transparency) and open at all sides in order to allow natural weather conditions, except for some wire-netting. Plants were thus grown under near-ambient growth conditions except for watering. Monocultures of F. rubra and P. lanceolata, or mixtures of both species (1∶1 proportion), were planted in June 2008 in a replacement design. Assignment of the planting treatments occurred randomly to a total of 11 units, resulting in 3–4 replicates. Planted seedlings were raised in the greenhouse for four weeks before transplanting. Seeds from local provenance (forelands of the river Rhine, near Nijmegen, the Netherlands) were first germinated in Petri dishes and then transferred to small pots containing the same background soil until transplant. Thirty-six seedlings (6×6) were then planted in each unit giving a plant density of 144 m⁻², but only the area of the inner 4×4 plants (32×32 cm) was used for further measurements. Interplant distant was 8 cm but distance from the edge plants to the rim was 5 cm. During the growing season, plants were irrigated 2 L unit⁻¹ three times a week with tap water through an automatic irrigation system (PRIVA, de Lier, The Netherlands). In winter, watering was supplied manually once a week.

The bottom of each unit was filled with a five-cm layer of coarse gravel covered with weed cloth. Soil depth from surface to gravel stones was 55 cm, divided in a 14-cm nutrient-rich layer consisting of black soil placed at 28 cm depth, and the remaining of the profile being filled up with a mixture of the same nutrient-rich black soil and nutrient-poor riverine sand (1∶3; v∶v) resulting in a poor sandy background soil. Soil nutrients were measured with an autoanalyzer (Bran+Luebbe, Norderstedt, Germany) after nutrients were extracted by diluting 20 g of freshly mixed soil samples in 50 mL of demineralized water and shaking for 1 h. At the start of the experiment, the nutrient-rich soil contained 26.4±1.5 g kg⁻¹ organic matter, and available nutrients were as follow: 256.4±65.1 mg kg⁻¹ nitrate (NO₃⁻), 11.2±2.2 mg kg⁻¹ ammonium (NH₄⁺) and 4.1±0.3 mg kg⁻¹ phosphate (PO₄⁻³), whereas these values were 9.6±0.1, 60.8±0.1, 2.5±0.0 and 1.5±0.3, respectively, for the nutrient-poor background soil.

Each unit had separate drainage at the bottom and holes to insert a minirhizotron tube (6.4 cm inner diameter×50 cm length) horizontally with the top of the tube at 10 cm depth, and two soil suctions cups (Rhizosphere Research Products, Wageningen, The Netherlands) at 7 and 35 cm depth for collecting soil solution for analysis.

Measurements

In late August 2009, after two growing seasons, standing shoot biomass in the inner area was harvested by clipping 2 cm above soil surface. Root mass density was estimated by soil cores (20 mm diameter, four sub-replicates in each plot) in the inner area down to four soil layers (0–14, 14–28, 28–42, 42–55 cm depth). Distances to surrounding individual plants were equal. Roots per soil increment were collected after carefully rinsing them with tap water. Fresh weight was determined immediately with a microbalance (Sartorius AG, Goettingen, Germany). Up to 100 mg of fresh roots was then stored at −80°C for later molecular analyses. Shoot and root dry weights were determined after drying samples at 70°C for 48 hours. Species abundance belowground in mixtures was quantified only in the top soil layer (0–14 cm) and in the intermediate nutrient-rich layer (28–42 cm), thus processing ≈70% of the total root biomass. On these samples, genomic DNA extracts were subjected separately to quantitative real time polymerase chain reactions (RT-PCR) with primers for non-coding species-specific markers [29]. Analyses were performed on the basis of 100 mg fresh root mass, and recalculated in terms of dry weight as this was highly correlated with the fresh weight (R² = 0.89, P<0.001).

Root images from minirhizotron tubes at 10 cm depth (21.6×7.0 cm, 300 dpi; CI-600 Root Scanner, CID Inc., Camas, WA, USA) captured the rooting area of four individuals in a row (either of the same species in monocultures, or half of each species in mixtures). Images were taken every 37 days on average, except in winter (Nov–Feb). Roots were digitized and analyzed using the WinRhizoTron V. 2005a software (Regents Inc., Quebec, Canada) for root length production. In mixtures, analyses were separated by species as the different color of newly-formed roots enabled species distinction: dark red to brown roots for F. rubra, pale grey to white roots for P. lanceolata (Fig. 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Minirhizotron images at 10 cm depth taken ten weeks after the start of the experiment.

Note the abundance of P. lanceolata (Pl, white roots) and absence of F. rubra (Fr, brown roots) in mixtures images compared to the respective monocultures.

https://doi.org/10.1371/journal.pone.0055805.g002

Soluble nutrients in the soil solution at two depths (7 and 35 cm; poor and nutrient-rich soil, respectively) were monitored every 65 days on average except in winter, by sampling soil water through porous soil suction cups and analyzing the extracted water solution for available nitrate with an autoanalyzer (Bran+Luebbe, Norderstedt, Germany).

Calculations and data analysis

In our replacement design, where total plant density in the mixture was equal to the plant density used in the monoculture of each component, the competitive ability of the species was given by the Relative Yield [9], [10], [36], calculated as the ratio between the observed yield of a species in mixture and the yield of this same species in monoculture. These calculations find their origin in Lotka-Volterra competition theory [10], which is the central concept in competition and coexistence theory [37], [38]. Relative Yields of 0.5 for both species reflect the situation of competitive equivalence with intraspecific competition equal to interspecific competition. If plants compete for a finite resource, a superior competitor is expected to take a larger proportion of the shared resource resulting in a higher Relative Yield, at the expense of an inferior competitor which will develop a proportionally lower relative yield. Belowground, the root Relative Yields are expected to deviate particularly in the nutrient-rich layer with the superior competitor developing a much larger root mass than the inferior competitor. As root investments will pay-off in nutrient uptake and growth, the root Relative Yields in the nutrient-rich layer are expected to be similar to the Relative Yields aboveground.

As a finite resource is partitioned among competing species, the sum of the relative yields (Relative Yield Total, RYT) is not expected to deviate from unity. RYT>1 or overyielding is only expected in the case of niche differentiation, i.e. when both competitors have access to partly unique resources, as in the case of species with different rooting depths [13]. In such cases intraspecific competition is greater than interspecific competition for both species which is the criterion for species coexistence [38]. Significance of Relative Yields is tested by comparing the observed values with those expected from monocultures representing the null-expectation of competitive equivalence (RY = 0.5), calculated as ½ of the monoculture values.

If the species are involved in a competitive game, one of the species is expected to overinvest in roots (Relative Yield >>0.5) at the expense of the other species (Relative Yield <<0.5). Such investment will take place similarly in the nutrient-poor topsoil as in the nutrient-rich deep soil layer. Depending on the extent of dominance and suppression, RYT may appear larger than 1. As root investments are expected to be altered, Relative Yields belowground will not reflect Relative Yields aboveground.

Root and shoot masses were estimated on the basis of above and below-ground biomass values per area and core soil volume, respectively. To compare whether values of root mass observed in mixtures for each species deviated from the expected values, we ran t-test.

Significance of differences over time in soil solution nutrients were tested by ANCOVA, using diversity of species and soil depth as fixed factors, and days after plantation as covariate. Conventional tests aimed at testing temporal trends (RM-ANOVA and MANOVA) could not be applied because the sphericity assumption was not met. In ANCOVA, differences in the temporal pattern between factors were considered significant when the interaction(s) between factor(s) and ‘days after plantation’ result was significant. When a factor or interaction resulted significant, pair-wise comparisons were performed using the Sidak correction for multiple comparisons.

Differences between expected and observed root length density from minirhizotron images were explored by linear regression, by plotting expected against observed values deviating from the null 1∶1 expectation (i.e., expected = observed). Differences between observed and expected values of root lengths in mixtures on specific dates for each species were tested by t-tests, using Sidak correction for multiple comparisons. All analyses were run with PASW Statistics 18 (SPSS Inc., Chicaco, IL, USA).

Mechanisms of belowground competition

Except that we have been able to rule out a differential response to soil nutrients, we do not know what mechanisms have been driving this strong dominance and suppression early in this competition experiment. There has been a lot of attention in recent years to the effects of species-specific communities of soil pathogens affecting coexistence and production of plant communities [41]–[43]. Plantago growth is known to be sensitive to its own conditioned soils due to accumulation of self-harming fungi [44] and root pathogens [45], suggesting the presence of negative plant-soil feedback [46] in monoculture of this species. Mixtures would have been a better environment for Plantago roots to grow as self-harming biota would have been diluted. A recent plant–soil feedback experiment showed that Plantago monocultures developed 3.2-fold more biomass in the presence of Festuca soil biota compared to soil biota of its own, but the reverse was also true: Festuca monocultures grew 2.5-fold more biomass on Plantago soil than soil of its own [47]. However, it should be noted that soils in the current experiment were not conditioned purposely and, therefore, it is unlikely that species-specific soil biota solely explain the observed root responses. Moreover, differential root growth developed very early in the experiment well before species–specific soil communities were likely built up [48].

Root growth suppression, apparent already at low densities and independent of local soil nutrient concentrations, is reminiscent of allelopathy or chemical interference [49]–[51]. Release of chemical substances may be expected to quickly reduce root growth, as in the case of Festuca, but to our knowledge immediate overproduction by the superior species, as observed in Plantago, is a phenomenon that has hitherto been unassociated with allelopathy. Some studies have suggested that root exudates can stimulate root growth in response to interspecific neighbors [26], [52]–[55], which would imply that Plantago root growth was stimulated by Festuca. These phenomena have not been previously described for these species; phytotoxic effects have been attributed only to root exudates of Festuca rubra [56].

Facilitation by roots of certain species through the release of organic acids by roots of leguminous species may account for some cases of root growth stimulation in mixtures [57]. However, it seems unlikely that such facilitative mechanism can explain root growth stimulation in our non-leguminous system [20]. As discussed below, there is no sign of facilitation aboveground in our system as overyielding aboveground was not detected.

Interpreting root overproduction

Consistent with game theoretical predictions [22], Plantago won by rapid investment in roots in the presence of Festuca, at the same time suppressing Festuca and preventing it from acquiring soil resources, resulting in a much larger root investment of Plantago than predicted on the basis of classical resource competition. However, from a game theoretical perspective, pertinent questions remain. Firstly, why did Plantago win and not Festuca? With its higher root densities, fine roots and high nutrient uptake rates, the grass species Festuca had a much better starting position to compete for nutrients. For our species pair, well-known traits conferring competitive ability belowground [1], [4], [6] could not predict the competitive outcome. Further research has to unravel the root traits that have predictive power and even then, the outcome may well depend on specific combinations of species and soils.

Secondly, as competition is a process taking place among individuals, why did Plantago not overproduce to a similar extent in monoculture? Craine [22] suggested that the best solution for a plant is to alter root allocation in proportion to the root length density of competitors. This exactly seems to have occurred in our experiment: plants with lower root densities (Plantago) overproduced roots strongly and won competition from plants with already high root densities (Festuca). As root densities in Plantago monocultures are lower than in Festuca monocultures, as is generally true for forbs versus grasses, a similar competitive game between Plantago individuals may have resulted in less overproduction. In other words, making substantially more roots may have paid-off only in competition with Festuca individuals, not with other Plantago individuals.

Consequences for plant competition and coexistence

We do not know how common this belowground competitive mechanism is in plant communities. However, if it is widespread it may have easily gone unnoticed in many competition experiments. The reason is that, aboveground, competitive relationships among our species appeared to conform to the resource competition model. Similar to numerous other experiments with only aboveground information (e.g., [9]; see refs there), Relative Yields of 0.70 and 0.27 for Plantago and Festuca, respectively, would project Plantago as the winner replacing Festuca in resource competition (Fig. 1B), further confirmed by an aboveground RYT similar to unity (0.97). Due to inherent difficulties in quantifying the roots of different species, our experiment is one of the first to compare competitive interactions aboveground with those belowground. Doing so revealed that apparently classical competitive relationships aboveground were combined by unexpected responses belowground.

If our results for these two common plant species are representative for a wider group of plants, the implications for long-term competitive superiority and coexistence may be profound. Competitive games are predicted to generate a “Tragedy of the Commons” where plants invest more to the acquisition of a limiting recourse than is optimal in the absence of competition [22]. Likewise, Plantago individuals invested 65% more biomass in their roots in mixtures than in monoculture (percentage total biomass increased from 32.4 to 53.5; Fig. 1E). If Plantago roots had not overproduced but only replaced the roots of Festuca in mixture (belowground RY 0.87 rather than 1.66), this increase in root investment would only have been 16% (percentage total biomass increase from 32.4 to 37.7). Although the investment pays-off in terms of immediate competitive gain, such major root investment may compromise biomass production in the long run, reminiscent of a Tragedy of the Commons. Interestingly, in a two-species Plantago-Festuca mixture within a long-term biodiversity experiment [58], Plantago initially dominated the mixture aboveground as in our experiment. But over the 11 years of study Plantago never outcompeted Festuca and after eight years Festuca even gained in abundance (J. van Ruijven, pers. comm.). This trajectory suggests that the overinvestment of Plantago in roots may have compromised its competitive ability in the long run.

Consistent with our results, there are indications from biodiversity studies that root mass is increased in species mixtures [53], [57], [59] and that this higher root biomass may already develop prior to positive effects of biodiversity on aboveground production [59]. Moreover, evidence is increasing that interactions in multi-species communities are driven by species-specific soil biota giving opportunities for local coexistence [41]–[43], [47], [60], [61], whereas opportunities for niche partitioning for nutrients seem to be limited [28], [62]. Future work should demonstrate to what extent the root responses seen in our experiment also play a role in more diverse plant communities.