1Plants sharing pots have been observed to show more allocation to roots than do solitary plants, even when resources per plant are constant, indicating plasticity to root neighbours. Lowered red to far-red ratio (R : FR), a cue of above-ground competition, is known to cause increased stem elongation and decreased allocation to roots. In nature, however, cues for above- and below-ground competition are likely to be experienced simultaneously, and may elicit contradictory responses.
2We investigated whether the presence of the above-ground competition cue (low R : FR) affected the responses to the below-ground competition cue (presence of neighbouring roots) in soybean and whether these responses depended on the presence or absence of symbiotic bacteria or on the availability of nutrients.
3In a fully factorial study in a growth room, light quality, root neighbours, nutrient availability and Bradyrhizobium inoculation were manipulated. Stem elongation and biomass allocation were measured.
4As predicted, plants allocated more to roots in the presence of root neighbours, regardless of R : FR, Bradyrhizobium and nutrient treatment. Plants elongated under low R : FR, although the degree of elongation was affected by Bradyrhizobium and nutrients. R : FR did not affect allocation to roots. Surprisingly, we found no reduction in reproductive allocation and therefore no ‘tragedy of the commons’ cost; instead, growth was increased in the presence of root neighbours.
5We conclude that soybean responds to both competition cues independently, but that below-ground resources such as Bradyrhizobium and nutrient availability moderate the morphological responses to above- and below-ground competition.
6To our knowledge, this is the first study that tests responses to above- and below-ground competition cues simultaneously and that shows the independence of these responses. This study provides supporting evidence that plants can sense neighbouring roots as a cue of impending competition and demonstrates clearly that the root allocation response is not dependent on the presence of Bradyrhizobium or an artefact of inadvertent R : FR manipulation.
In nature, plants compete both below-ground and above-ground. Once plants have detected the presence of neighbours, phenotypic plasticity allows them to increase their competitiveness (Callaway 2002). Below-ground, plants have been shown to use increased root proliferation in areas of high nutrient concentrations as a mechanism for rapid exploitation of resource patches under competitive situations (Hodge 2004; de Kroon & Mommer 2006). They have also been shown to respond to competition by increasing root allocation when resources are being depleted (Marschner 1995). However, studies have recently shown that the behaviour of a plant below-ground is considerably more complex than a simple response to resource depletion (Schenk 2006). Several recent studies have demonstrated that plants can anticipate root competition and increase root proliferation in the presence of neighbouring roots even before resources are depleted (Gersani et al. 2001; Maina et al. 2002; O’Brien et al. 2005). These studies have been used as the basis for drawing predictions on plant behaviour under competitive situations (Brown 2001). However, although novel and revealing, these studies did not consider that plants may simultaneously experience both above- and below-ground competition, with the former often mediated by light quality.
The nature of the cue for below-ground competition has been inferred from studies of responses to the presence of neighbouring roots (Gersani et al. 2001; Maina et al. 2002; O’Brien et al. 2005), hereafter referred to as ‘root neighbour’ studies. These studies differ from ecological competition studies in that average soil resources per plant are maintained at a constant level by growing either one plant in a pot or two plants sharing two pots instead of adding competitors into constant conditions. These root neighbour studies, performed in soybean (Glycine max) (Gersani et al. 2001), bean (Phaseolus vulgaris) (Maina et al. 2002), and pea (Vigna unguiculata) (O’Brien et al. 2005) showed that plants that had root neighbours had increased root biomass and increased root allocation even though nutrients were controlled, indicating plasticity to root neighbours rather than a response to resource depletion. They argue that this greater root allocation increases a plant's below-ground competitive ability (although see Cahill 2003 and Kembel & Cahill 2005); however, plants grown with root neighbours showed a reduction in reproductive yield (Gersani et al. 2001; Maina et al. 2002; O’Brien et al. 2005). This ultimate reduction in fitness gives rise to the proposed ‘tragedy of the commons’ (Hardin 1968) model as a result of root competition (Gersani et al. 2001). According to this model, the benefits of increased root allocation are perceived individually, but the costs are shared among all members utilizing the common resource. Therefore, the overall outcome of the response is eventually detrimental to all individuals. It has been argued that the observed results in the root neighbour studies could be the result of changes in pot size (absolute soil volume) regardless of the presence of neighbours (Schenk 2006; Hess & de Kroon 2007). However, McConnaughay & Bazzaz (1991) found that in five species, while root biomass and other size components increased in larger pots, root allocation was not affected by pot size differences.
Taken together, these studies of leguminous species that look at above- and below-ground competition cues separately thus predict contradictory shifts in allocation in response to the presence of competitors. Studies of non-leguminous species, however, have yielded different results suggesting between-species variation in response to root neighbours (Cahill 2003; S. A. Dudley & L. A. Donovan, unpublished data; Semchenko et al. 2007). The possibility exists therefore that responses to below-ground competition cues can either contradict, and therefore antagonize the shifts in allocation prompted by above-ground competition cues, or complement them, therefore showing synergistic effects. Thus, in order to understand plastic shifts in allocation in response to competition cues, we tested the effects of both competition cues applied simultaneously. We used two different levels of nutrient availability in order to manipulate the intensity of below-ground competition. We predicted a greater response to the below-ground cue relative to the above-ground cue in low nutrient availability. We also expected plants grown in low nutrients to have reduced biomass and increased root allocation (e.g. see McConnaughay & Bazzaz 1991).
Much of the pioneering work in below-ground competition cues has been done in leguminous species, which form symbiotic relationships with nitrogen-fixing bacteria (Bradyrhizobium japonicum in the case of soybean). However, the potential role of the symbiotic bacteria in relation to responses to competition cues has not yet been explored. Symbiotic nitrogen fixation occurs in specialized structures (nodules) that are initiated in the outer root cortex following root hair infection by the bacterium (Evans 1975), and so nodule mass may influence the root mass. Under low R : FR, as a proxy for above-ground competition, plants produce fewer nodules and less nodule mass (Kasperbauer et al. 1984; Hunt et al. 1987; Kasperbauer & Hunt 1994). However, root neighbour experiments with leguminous species (Gersani et al. 2001; Maina et al. 2002) have not manipulated the presence or absence of this symbiosis, and so the potential dependence of the response on the bacterial partner is unknown.
In this experiment, we grew soybeans in a factorial design that included high and low R : FR light, root neighbours present or absent, nutrient availability medium or scarce, and inoculation or non-inoculation with Bradyrhizobium. Density, irradiance and average soil volume per plant were kept constant. We measured internode length, leaf area and biomass components. We tested the following hypotheses: (i) the presence of the above-ground competition cue (low R : FR) affects the responses to the below-ground competition cue (presence of neighbouring roots); (ii) the responses to the below-ground competition cue are dependent on the presence or absence of Bradyrhizobium bacteria; and (iii) the responses to the above- and below-ground competition cues are dependent on the availability of nutrients.
Materials and methods
Soybean (Glycine max) was selected for this experiment because it has been shown that it will respond to inter-plant root competition by over-producing roots (Gersani et al. 2001). The cultivar ‘maple arrow’ was utilized. Maple arrow is an indeterminate variety, as is Williams, the cultivar used in previous studies (Gersani 2001). Maple arrow, however, is a short-season variety and is currently the standard variety for eastern Ontario (Agriculture Canada 2004).
The experiment was a totally factorial design (24) with treatments that consisted of: high R : FR or low R : FR light, root neighbours present or absent, nutrient availability medium or scarce, and Bradyrhizobium bacteria inoculation or non-inoculation. The factorial nature of the experiment gave 16 treatment combinations, each replicated eight times, including four plants per replicate for a total of 32 plants per treatment combination and 512 plants in total. Seeds were planted directly into their pots in a mix of 1 : 1 sand and turface (Profile Products LLC, Buffalo, Grove, IL, USA) and grown from 25 May to 6 July (2004) in a growth room at a temperature of 22 °C and with a 13-h photoperiod for a total of 42 days. Pots were arranged in trays containing a total of 64 plants each. Tray places in the bench were randomly switched on a weekly basis to avoid any local effects from the room. Plants from a different experiment were used as a border row surrounding the trays to avoid any border effects. Seeds that did not germinate within 1 week of the first emergence were replaced the 8th day after the first germination. The number of seeds replaced was 96 out of the total 512.
In order to create the root neighbour treatment soybean seeds were planted in board pots (Zipset Plant Bands, bleached board, light weight, from Stuewe & Sons, Inc., Corvallis, Oregon, USA) in the shape of rectangular cuboids with a constant cross-sectional area. The dimensions depended on root neighbour treatments. In the root neighbours absent treatment, one plant per pot was grown in a pot 3.81 × 3.81 cm, 35.56 cm deep. In the root neighbours present treatment, four plants per pot were grown in a pot with four times the cross sectional area, i.e. 7.62 × 7.62 cm, 35.56 cm deep. Each pot in the root neighbour present treatment was matched with four pots in the root neighbour absent, with all other conditions the same, so that the sample size of plants was kept constant in each treatment. The pots were packed into 30.48 × 30.48 cm square trays that held 64 plants total. Groups of four plants with and without root neighbours were arrayed in a checkerboard fashion. Utilizing these dimensions allowed us to maintain a constant density of 689 plants per square metre, and ensured that a solitary plant had the same soil depth and average soil volume as a plant with root neighbours. Therefore, we manipulated the presence or absence of roots from a neighbouring plant while presumably keeping resources constant, to test for responses to the below-ground competition cue instead of below-ground competition (resource depletion). Pot size can also be considered a resource for plants. However, in our experimental design, like that of Gersani et al. (2001), absolute soil volume (pot size) varies with presence of competitors, so that we are unable to separate out pot size effects.
R : FR was manipulated by growing high density stands in high R : FR and low R : FR conditions (Ballaréet al. 1991; Dudley & Schmitt 1996; Maliakal et al. 1999). Each tray of 64 plants was placed into one of the light treatments (four trays per light treatment). Light was provided by a mix of fluorescent lights (fluorescent cool white Philips lamps), which provided light rich in red relative to far-red, and incandescent lights that provided further far-red light. Plants were placed under either a high R : FR shade (Mitsui Chemicals, Inc., Tokyo, Japan) (R : FR c. 1.9) or a low R : FR shade (R : FR c. 0.5) and transmittance was matched to within 3–8% so all plants were under similar light levels (PAR of approximately 100 µmol m−2 s−1). The low R : FR shade was constructed as indicated by Lee (1985) and consisted of one part Solvaperm Yellow G dye and four parts of Hostperm Violet RL pigments (both obtained from Hoescht Celanese, Guelph, Canada) suspended in a solution in clear finish varnish.
For the Bradyrhizobium manipulation, seeds were either left alone (uninoculated) or were dusted prior to planting (inoculated) with 0.82 g of soybean inoculant (‘Soy select’ (Bradyrhizobium japonicum) from Nitragin (Milwaukee, WI), minimum of 2 × 109 viable cells g−1) in 238 g of seeds as indicated by the company. This treatment was applied on a per tray basis.
The nutrient treatments were applied within trays. Fertilization commenced 10 days after planting and was maintained in a weekly basis. A low nitrogen fertilizer (Awesome Blossom, Technaflora Plant Products Ltd, BC, Canada, containing the following elements: ammonical nitrogen (0.82%), nitrate nitrogen (1.18%), P2O2 (11%), K2O (11%), B (0.00011%), Cu (0.00001%), Fe (0.0008%), Mn (0.00024%), Mo (0.00001%), Zn (0.00005%)) was used for plants in the inoculation treatment (NPK = 2-11-11) in order to promote the symbiotic relationship between the plants and bacteria. The same fertilizer with an addition of ammonium nitrate (as indicated by Plaster 1997) was applied to plants in the non-inoculation treatment (NPK = 11-11-11). Two micronutrients (Ca and Mg) were added by utilizing dolomitic limestone as indicated by Bailey & Nelson 2004) in proportions of 42.5 g in 1.9 L of water per tray per week. The first nutrient application, at the medium nutrient level, was administered to all plants, while in the subsequent applications the scarce nutrient treatment was given one-tenth of the medium nutrient treatment.
Above-ground harvesting commenced on 6 July 2004 with plants ranging between R4-R6 stages of maturity (from small 2-cm pods to pods containing green seeds) (Kansas State University Agricultural Experiment Station and Cooperative Extension Service 1997). Measurements of node height and total plant height (height at the apical meristem) were taken and leaf area was measured (ADC AM100 Leaf Area Meter, Houston, TX, USA). Above-ground biomass was determined by collecting plant material (leaves, stem and pods) in paper bags and drying it in an oven for 72 h (leaves and stems) and 96 h (pods) at 65 °C. During above-ground harvesting, pots with roots were stored in a cold room at 5 °C. One week later roots were washed, collected in paper bags, and dried for 96 h at 65 °C. Some of the pots were contaminated by an unknown fungus. This fungus degraded the pots and compacted the soil making the roots more adhesive. However, no phenotypic effects of this contamination were observed on above-ground traits. For plants in the root neighbour treatments, all roots in a single pot were weighed together because of the difficulty of assigning roots to individual plants in the root neighbour present treatment. Pots that showed no fungal contamination were sectioned in four segments, each 8.9 cm long, to create a profile of root allocation. This profile showed vertical root growth with the most proliferation occurring in the first and last portions of the root system (Fig. 1). These profiles were not affected by any of the treatments (results not shown) and provided no indication of root restriction.
The data were analysed using SAS statistical software (version 8.02; SAS, Cary, NC, USA). We used proc glm to carry out analyses of variance and covariance. We used analysis of covariance to test for differences in allocation and elongation (Coleman et al. 1994; McConnaughay & Coleman 1998). Elongation was measured as the least square mean (lsmean) from an analysis of covariance with plant total height as the dependent variable and stem weight as the covariate (lsmeans option, proc glm). Similarly, reproductive allocation was measured as the lsmean from an analysis of covariance with pod mass as the dependent variable and vegetative biomass (sum of stem mass, leaf mass and root mass) as the covariate.
Because trays were assigned to different Bradyrhizobium and light treatments, F-ratios for light, Bradyrhizobium and light × Bradyrhizobium effects were tested over the mean square for trays, nested within the light × rhizobium effect, in the denominator. Nutrient and root neighbour main effects and interactions, because these treatments were applied within trays, were tested over the mean square error. For below-ground traits, one tray that was completely infected by the unknown fungus was not included in the analyses and for the remaining trays, which showed only partial contamination, presence of fungus was included in the analyses. Plants from replaced seeds were not included in the analyses of above-ground traits because they emerged under a plant canopy as a consequence of germinating later. However, a post hoc test was performed on these replaced plants with the objective of looking for changes in root allocation at greater levels of elongation. Because of the difficulty of separating the roots of plants in the root neighbours present treatment, below-ground traits and total biomass were analysed as groups of four plants, either the four in a large pot, or the four adjacent plants in single pots. In order to check that the assumption of pooling plants into groups of four does not create spurious results (Laird & Aarssen 2005), we checked for differences in population asymmetry between root neighbour treatment by comparing the average Gini coefficients of the groups of four (proc univariate). No differences were found between any of the treatments (data not shown). Root allocation was measured as the lsmean from an analysis of covariance with root mass as the dependent variable and shoot biomass (sum of stem mass and leaf mass) as the covariate. To check for a pooling effect on this parameter, we compared the slopes of the root to shoot regression equations of solitary plants as individuals and as groups of four and found no differences (data not shown).
responses to above and below-ground cues
Overall, different traits were affected by the different competition cues. The analyses of variance for above-ground traits, with the individual plant as the observation (Table 1), demonstrated effects of the low R : FR cue but not the root neighbours cue. Changes in above-ground morphology and allocation with the individual plant as the observation (Table 2) showed a similar pattern, though elongation was affected by root neighbours when Bradyrhizobium was present. In contrast, analyses of below-ground biomass and allocation, using the group of four plants as the observation (Table 3), primarily showed effects of root neighbours and not R : FR. Root mass, however, was affected by low R : FR but only in the presence of Bradyrhizobium. No trait demonstrated any interactions between the light and root neighbour treatments (Tables 1–3) indicating that the presence of one competition cue did not affect the responses to the other competition cue.
Table 1. Analyses of variance for above-ground traits for growth room grown soybean (Glycine max)
Table 3. Analyses of variance and covariance for below-ground and overall plant traits for growth room grown soybean (Glycine max)
Root to shoot ratio (root mass/shoot mass)
Reproductive allocation (pod mass/veg. biomass)
Treatment tested over tray nested with light × Rhizobium interaction.
Degrees of freedom for the error terms were: total biomass 91, root mass 91, root mass/shoot mass 90 and reproduction 90. Bold numbers indicate significant values.
Light × neighbours
Light × nutrient
Neighbours × nutrient
Light × Rhizobium× neighbours
Light × Rhizobium× nutrient
Light × neighbours × nutrient
Rhiz × neighbours × nutrient
Light × Rhizobium× neighbours × nutrient
Tray (light ×Rhizobium)
responses to low r : fr
Total plant biomass was not affected by the light treatments (Table 3). This result was expected because density was constant and transmittance did not differ between the low and high R : FR treatments. Soybean plants grown under low R : FR were significantly taller than plants grown under high R : FR (low R : FR = 33.2 cm, high R : FR = 28.9 cm; SE = 0.6; Table 1). In contrast, the number of pods produced by plants under low R : FR was significantly reduced (Table 1). Moreover, as expected, the low R : FR treatment affected the plants by inducing the characteristic elongation response, such that, for any given stem mass, plants grown under low R : FR were significantly taller than under high R : FR (low R : FR = 33.5 cm, high R : FR = 28.9 cm; SE = 0.3; Table 2).
The low R : FR treatment also affected specific leaf area (leaf area per unit leaf mass) as well as allocation to stem over leaves. Specific leaf area was significantly greater under low R : FR (low R : FR = 127.6, high R : FR = 110.7; SE = 1.0; Table 2). Allocation to stem relative to leaves was increased in low R : FR (low R : FR = 0.170 g, high R : FR = 0.141 g; SE = 0.001; Table 2).
Root to shoot ratio was unaffected by the light treatments (Table 3). However, in our experimental plants, elongation induced by the light treatment was relatively low. There was a greater difference in elongation between the experimental and replacement plants, which were planted 8 days later (ancova: lsmeans for height, experimental plants = 29 cm, replacement plants = 44.5 cm) presumably because of growing in the presence of taller plants. We therefore did a post hoc test for the root neighbours absent treatment, in which individual root allocation could be assessed, comparing root allocation of replacement plants with experimental plants for the root neighbours absent treatments. There was decreased allocation to roots for the replacement plants (ancova: lsmeans for root mass, experimental = 0.095 g, replacement = 0.076 g, F = 43.20, P < 0.0001, n experimental = 175, n replacement = 68), suggesting that higher levels of elongation reduced root allocation.
No direct effect of light was observed on root mass. However, an effect was observed that was moderated by the presence or absence of Bradyrhizobium. Under the inoculated treatment, plants in high R : FR produced significantly more root mass than plants in low R : FR (inoculated: high R : FR = 0.51 g, low R : FR = 0.41 g; SE = 0.03) while under the non-inoculated treatment the trend was reversed with plants under low R : FR producing more root mass (non-inoculated: high R : FR = 0.29 g, low R : FR = 0.33 g; SE = 0.02; Table 3).
presence of root neighbours
Although plant density and soil resources were kept constant for the root neighbour present and absent treatments, it was found that, contrary to our expectations, total plant biomass was significantly increased by the presence of root neighbours (root neighbours present = 1.99 g, root neighbours absent = 1.76 g; SE = 0.08; Table 3).
The root neighbours present treatment affected root production (root mass) and root allocation (root to shoot ratio) as well as nodulation. It was found that, in the presence of root neighbours, plants produced significantly more root mass (root neighbours present = 0.42 g, root neighbours absent = 0.35 g; SE = 0.01; Table 3), and also allocated more to roots than to shoots (root : shoot; root neighbours present = 0.38 g, root neighbours absent = 0.35 g; SE = 0.005; Table 3, Fig. 2). These results are in accordance with results found in previous single factor studies. In the root neighbour present treatment, nodule number was also increased (anova root neighbours present = 7.53, root neighbours absent = 6.08, F-ratio 7.69, P < 0.006, SE = 0.36).
No direct effects of root neighbours were observed on any above-ground trait, although there was an effect of the combination of root neighbours and the presence or absence of Bradyrhizobium. Plants that had been both inoculated and grown in the absence of root neighbours elongated more than plants in the other treatment combinations (Table 2, Fig. 3).
effects of bradyrhizobium and nutrient
Unexpectedly, total biomass was similar in the medium and scarce nutrient treatments (Table 3). Total plant biomass was affected, however, by the Bradyrhizobium treatment, with inoculated plants having greater mass than non-inoculated plants (inoculated = 2.33 g; non-inoculated = 1.42 g; SE = 0.1; Table 3).
Nutrient availability had significant effects on allocation, both on stem relative to leaves and root relative to shoot. Plants allocated significantly more to stem relative to leaf mass in scarce nutrients than they did in medium nutrients (scarce = 0.160 g, medium = 0.151 g; SE = 0.001; Table 2), and significantly more to roots relative to shoots (scarce = 0.38 g, medium = 0.35 g; SE = 0.005; Table 3).
It was also found that inoculated plants produced more root mass (inoculated = 0.46 g; non-inoculated = 0.31 g; SE = 0.02; Table 3), as did plants in scarce nutrients (scarce = 0.41 g, medium = 0.36 g; SE = 0.01; Table 3).
interactions between bradyrhizobium, nutrients and light
The effects of light on total height and elongation were moderated by the presence of Bradyrhizobium and the availability of nutrients under a three-way interaction that followed the same trend for both traits (Tables 1 and 2). In the non-inoculated treatment, the difference in height of plants of a given stem mass between low R : FR and high R : FR was greater in scarce nutrients. In the inoculated treatment, however, no significant difference in elongation was observed in scarce nutrients while in medium nutrients plants in low R : FR were significantly more elongated (Fig. 4).
Reproduction showed a similar three-way interaction between light, Bradyrhizobium, and nutrients. Reproductive allocation (pod mass relative to vegetative biomass) and pod mass showed the same trends (Tables 1 and 3). Plants in high R : FR always produced more pod mass relative to vegetative biomass than plants in low R : FR. However, for non-inoculated plants, those in high R : FR allocated more to reproduction in scarce nutrients than they did in medium nutrients, while plants in low R : FR allocated more to reproduction in medium nutrients than they did when nutrients were scarce. In the inoculated treatment, however, a change in both trends was observed; plants in high R : FR allocated more to reproduction in medium nutrients and plants in low R : FR allocated approximately the same on both nutrient treatments (Table 1; Fig. 5).
The presence of fungus was included in the analysis of below-ground traits and it was found to be significant for root mass. The observed effect was that plants that were in contaminated pots had slightly greater root mass than those in unaffected pots (Table 3).
Previous studies in crop legumes examined responses to above- and below-ground competition cues separately, even though their results indicate that these cues elicit contradictory responses (Kasperbauer et al. 1984; Hunt et al. 1987, 1989; Britz 1990; Pausch et al. 1991; Gersani et al. 2001; Maina et al. 2002; O’Brien et al. 2005). In this study we examined soybean responses to above- and below-ground competition cues simultaneously for the first time. We found that, in this soybean cultivar, responses to each competition cue occurred independently from one another, and we can thus reject the hypothesis that the presence of one competition cue affects the plant responses to another cue.
In the presence of root neighbours, plants responded by increasing root mass and root to shoot ratio, a result obtained by previous studies in legumes (Gersani et al. 2001; Maina et al. 2002; O’Brien et al. 2005). The observed increase in allocation to roots was found both in the presence and absence of Bradyrhizobium bacteria, and in high and low nutrients. These results indicate that the below-ground competition response is not dependent on the presence or absence of the symbiont. The lack of an interaction between nutrient levels and presence of root neighbours indicates that the below-ground competition response is also not dependent on nutrient availability.
Soybean responded to low R : FR, the above-ground competition cue, as expected, with increased elongation (a competitive trait) and increased allocation to stem over leaves. These responses allow the plant to better forage for light and to shade its neighbours (Smith 1995). Increased specific leaf area (a resource acquisition trait) and a reduced number of pods (a measure of fitness) were also effects of reduced R : FR, indicating wider leaves to better acquire light, and reduced allocation to reproduction under the stress of imminent above-ground competition. However, low R : FR did not change the allocation to roots relative to shoots in this study. This could be attributed to the relatively slight elongation shown by the plants, because plants that were replaced, and therefore emerged later and grew under an existing plant canopy and so further reduced R : FR conditions, elongated more and showed a significant shift in allocation towards shoots relative to roots. R : FR is not the only cue of above-ground competition. Blue light, light intensity and ethylene have been shown to serve as above-ground competition cues, or to interact with R : FR (Ballaré 1999; Pierik et al. 2003, 2004; Franklin & Whitelam 2005; Vandenbussche et al. 2005). Although we only manipulated R : FR (because this can be done without changing light resources for photosynthesis), irradiance and ethylene should have been similar in the two light treatments as plants were always in high density.
When the above-ground and below-ground competition cues were applied simultaneously, soybean responded to both cues independently. It is argued that a trade-off in the ability to compete above- and below-ground should exist if the amount of resources available to the plant is kept constant (e.g. Shipley & Meziane 2002), because either light acquisition can be improved by allocating more biomass to leaves and stems, or nutrient uptake can be improved by allocating more biomass to roots (Shipley & Meziane 2002). Although soil resources and density were kept constant for plants in the root neighbours treatment, and transmittance was matched for plants in high and low R : FR, no such trade-off was observed. This lack of trade-off between the ability to compete above- and below-ground is consistent with results found by previous studies (Cahill et al. 2005). However, in our study, plants were harvested at a relatively early stage, and elongation differences induced by light quality, although significant, were relatively slight. Greater costs could develop later in growth, and with greater elongation.
Although we found an increase in root allocation in response to root neighbours similar to that reported by Gersani et al. (2001), Maina et al. (2002) and O’Brien et al. (2005), our results did not show a ‘tragedy of the commons’ situation (reduction in allocation to reproduction), which would have indicated a cost of increased root allocation. Evidence suggests that different soybean cultivars can respond differently to competition (Pfeiffer et al. 2001). Therefore, the differences observed with the study performed by Gersani (2001), namely the lack of costs arising from the overproduction of roots in the presence of root neighbours (tragedy of the commons), could possibly be attributed to this factor. However, maple arrow and Williams, the cultivars used in this study and in Gersani (2001), respectively, are similar varieties (both indeterminate) differing mainly in that maple arrow is a shorter-season variety. Because we harvested plants at the R6 stage of maturity, so that we could examine changes in allocation, we were unable to assess changes in reproduction to their full extent, since indeterminate cultivars will keep producing pods until the plant senesces. For this reason we not only looked for costs in terms of reduction in reproduction but also in total plant biomass. The increase of total plant biomass in the presence of root neighbours, rather than a reduction, provides further evidence of a lack of a ‘tragedy of the commons’ situation. It is possible that costs may arise in later stages of development. However, it is also possible that costs may be dependent on density conditions, with plants growing in low-density stands showing costs (as in the studies by Gersani et al. 2001; Maina et al. 2002; O’Brien et al. 2005) and plants growing in high-density stands, such as those in this study, showing a competitive advantage.
In recent reviews, Schenk (2006) and Hess & de Kroon (2007) argue that the experimental design utilized in the root neighbour studies could be flawed. They contend that the observed increase in root biomass could be due to changes in pot size (Schenk 2006; Hess & de Kroon 2007), and demonstrate that much of the published data is consistent with this hypothesis. Unfortunately, because pot size and presence of neighbours are confounded in our study, as in that of Gersani et al. (2001), we are unable to test this intriguing hypothesis. However, the focus of this study is on allocation to roots, which have been shown not to be affected by pot size (McConnaughay & Bazzaz 1991). McConnaughay & Bazzaz (1991) also showed that root restriction caused by reduced pot size was accompanied by changes in root architecture. We found no changes in the profile of root allocation between the plants in small pots (solitary treatment) and large pots (neighbours treatment) (Fig. 1). In both pots, root growth was vertical, with proliferation of fine roots most noticeable in the topmost and lowermost segments, with little lateral growth in the middle segments. No indications of root restriction were seen.
Because, in nature, a nearby competitor will most likely provide both types of competition cues simultaneously, possessing an independent response mechanism to each may seem redundant. However, responses to the low R : FR cue, which occur above-ground and are phytochrome mediated (Smith 1995), allow the plant to perceive neighbours at an early stage and from a distance (Ballaréet al. 1990), but they lack specificity (Dudley & Schmitt 1995). In contrast, the responses to root neighbours, which occur below-ground, appear to provide further information on the identity of the competitor, such as self vs. non-self, and same species vs. different species (Huber-Sannwald et al. 1996; Falik et al. 2003; Holzapfel & Alpert 2003; Gruntman & Novoplansky 2004). What is surprising is how no interaction, either in cost or synergism, was found between these two cues, leaving open the question how plants ‘decide’ whether to compete more strongly above- or below-ground.
The answer to this question may lie in the role of below-ground resource levels. We found that above-ground traits were mostly affected by the above-ground cue and that below-ground traits were mostly affected by the below-ground cue. However, below-ground resource levels, as determined by nutrients and Bradyrhizobium bacteria, affected morphology and moderated the responses to above- and below-ground competition cues. We observed that elongation was increased when plants were in the absence of root neighbours but only in the presence of Bradyrhizobium. The elongation response to low R : FR, however, was reduced in scarce nutrients, but again only in the presence of Bradyrhizobium. The below-ground trait of root mass was increased by plants in low R : FR in the presence of Bradyrhizobium but decreased in its absence. These results indicate that below-ground resources may play a key role on how plants respond to competition cues, and whether they compete more above- or below-ground.
Reproductive allocation, the ‘tragedy of the commons’ measure, showed the most complicated response to the environmental variables. In a three-way interaction, nutrient levels and Bradyrhizobium bacteria jointly affected the increased reproductive allocation response to R : FR by moderating the intensity of the response and even reversing some trends. Notably, R : FR and the presence of Bradyrhizobium are environmental variables that were not directly controlled in previous studies. These results show the complexity underlying fitness-related traits, and the difficulty in resolving causality on fitness.
The direct effects of nutrient levels on phenotype were as expected except for total plant biomass. We observed decreased allocation to shoot over roots and to leaves over stems, as well as increased root mass production in the scarce nutrient treatment, which was one tenth of the medium nutrient treatment. Total biomass, however, was not affected by nutrients. Since plants in both nutrient treatments received an equal first dosage of fertilizer at the medium level, this might have provided enough nutrients to mask any effects of the scarce nutrient treatment in total biomass; it is possible that growth may become more nutrient limited over time and effects of low nutrient levels may be observed in later stages of plant development.
The presence of Bradyrhizobium bacteria induced an increase in root mass as well as total biomass, suggesting that plants that are able to form a successful symbiosis with the bacteria may possess an advantage in acquisition of nutrients. It is also possible that some of the increase in both root mass and total biomass under this treatment was due to the production of nodules, structures that were absent on uninoculated plants.
This study provides evidence in support of previous studies that found that plants can sense neighbouring roots as a cue of impending competition. The results clearly show that this is not dependent on the presence or absence of Bradyrhizobium or induced by inadvertent manipulation of density (and therefore R : FR). We reject the hypothesis that responses to the above- and below-ground competition cues will affect one another because in this soybean cultivar and under the experimental conditions of this study, responses to the below-ground competition cue occurred independently from responses to the above-ground competition cue. Moreover, we found that below-ground resources such as nutrient availability and the presence of Bradyrhizobium bacteria moderated the relative allocation to competition responses. While in our results we saw little evidence of conflict between responses to above- and below-ground cues, our results suggest that resources may be important in determining how plants resolve conflicting responses to cues of competition.
We thank A. Yeas and L. Beaton for their help in preparing the experimental set up, S. Irazuzta and M. Sloan for their assistance in collecting data; and Lisa Donovan for helpful feedback. N. Rajapakse generously provided the high R : FR shading. The research was supported by an NSERC Discovery grant to S.A.D.