Some plant species are able to distinguish between neighbours of different genetic identity and attempt to pre-empt resources through root proliferation in the presence of unrelated competitors, but avoid competition with kin. However, studies on neighbour recognition have met with some scepticism because the mechanisms by which plants identify their neighbours have remained unclear.
In order to test whether root exudates could mediate neighbour recognition in plants, we performed a glasshouse experiment in which plants of Deschampsia caespitosa were subjected to root exudates collected from potential neighbours of different genetic identities, including siblings and individuals belonging to the same or a different population or species.
Our results show that root exudates can carry specific information about the genetic relatedness, population origin and species identity of neighbours, and trigger different responses at the whole root system level and at the level of individual roots in direct contact with locally applied exudates. Increased root density was mainly achieved through changes in morphology rather than biomass allocation, suggesting that plants are able to limit the energetic cost of selfish behaviour.
This study reveals a new level of complexity in the ability of plants to interpret and react to their surroundings.
Plant roots do not merely respond to nutrient concentrations in the soil, but are able to detect the presence and even genetic identity of neighbours, and react to them in ways consistent with behavioural theory. The presence of potential competitors for belowground resources has been shown to trigger increased biomass allocation to root growth (Gersani et al., 2001; Maina et al., 2002; Falik et al., 2003; O'Brien et al., 2005; Padilla et al., 2013). Such behaviour is expected to improve competitive ability but results in reduced resource acquisition efficiency and, consequently, lower fitness than would be attained if all individuals cooperated by restraining root growth – a phenomenon known as tragedy of the commons (Zhang et al., 1999; Gersani et al., 2001; Craine, 2006; O'Brien et al., 2007). It has also been demonstrated that some plant species are able to recognize closely related individuals (kin) and limit selfish proliferation of competitive organs in their presence (Dudley & File, 2007; Biedrzycki et al., 2010; Biernaskie, 2011; Lepik et al., 2012). Responses to neighbours can also differ depending on whether the neighbour belongs to the same or a different population or species (Krannitz & Caldwell, 1995; Mahall & Callaway, 1996; Semchenko et al., 2007a).
How plants determine the presence and identity of their neighbours remains largely unknown (Callaway, 2002; Callaway & Mahall, 2007). Understanding the mechanisms of neighbour recognition is important for the study of ecosystem functioning and plant evolution, but also for nature conservation and the development of techniques to reduce wasteful competition in agricultural crops (Schenk, 2006; Weiner et al., 2010; Inderjit et al., 2011; Thorpe et al., 2011). However, empirical studies have been plagued by methodological shortcomings. Most studies attempting to demonstrate the tragedy of the commons have simultaneously manipulated neighbour presence and soil volume (Gersani et al., 2001; Maina et al., 2002; O'Brien et al., 2005). In such an experimental design, changes in root production could be triggered by changes in space availability or nutrient distribution rather than neighbour recognition (Schenk, 2006; Hess & De Kroon, 2007; Semchenko et al., 2007b). However, some studies on root interactions were not compromised by issues related to soil volume and have recorded increased root growth in the presence of neighbours (Falik et al., 2003; Gruntman & Novoplansky, 2004; Semchenko et al., 2007a). Responses ascribed to kin recognition have also been subject to criticism. Differences observed between groups of related and unrelated individuals in biomass allocation to roots and seed production could be caused by numerous processes other than kin recognition, such as asymmetric competition, niche partitioning, genotypic variation in competitive ability and frequency-dependent selection imposed by natural enemies (Klemens, 2008; Masclaux et al., 2010; File et al., 2012). These may either cancel out the effects of kin recognition or produce outcomes that are indistinguishable from those caused by kin recognition.
A rigorous test of the mechanisms mediating responses to neighbour presence and identity is critical to determining whether tragedy of the commons in plant root competition and the ability of plants to recognize kin are genuine phenomena. Plant roots exude a wide range of chemical compounds that are known to alter soil properties and microbial communities (reviewed in Bais et al., 2006; Li et al., 2007; Broeckling et al., 2008). The direct inhibitory effect of certain root exudates on root growth has been demonstrated in several study systems (reviewed in Schenk et al., 1999; Bais et al., 2006). Some of the nontoxic exudates could be involved in regulating root proliferation in response to neighbour presence and identity, but this possibility remains largely unexplored. It was found that the chemical compounds produced by roots belonging to the same or a different genetic lineage can have a different effect on root branching in target seedlings grown in sterile liquid solution in a week-long experiment (Biedrzycki et al., 2010). However, it is unknown whether the effects of root exudates persist in the longer term and in the presence of soil microbiota, which can quickly degrade root exudates and neutralise or modify effects seen in sterile conditions (Kaur et al., 2009; Lankau, 2010; Ehlers, 2011).
The specificity and complexity of information that can be conveyed by root exudates also remains to be examined. It has been shown that in heterogeneous environments plants respond locally to nutrient or light patches at the level of individual roots and leaves, but these responses can be enhanced or suppressed by systemic signals reflecting the nutritional status of other parts of the plant (reviewed in de Kroon et al., 2009). Responses to heterogeneous soil conditions not only involve changes in biomass allocation and growth, but also adjustments in morphological traits such as specific root length and branching (Farley & Fitter, 1999; Semchenko et al., 2008; Mommer et al., 2011). It can be hypothesised that neighbours’ exudates are likely to create a heterogeneous environment that may elicit both local as well as systemic responses, and not only changes in root growth, as has been shown so far, but also changes in root morphology. The function of nontoxic root exudates as cues of neighbour proximity may also require local adaptation similar to that shown at a larger scale by plants exposed to the toxic exudates produced by invasive species (reviewed in Bais et al., 2006; He et al., 2009; Thorpe et al., 2009).
The aim of this study was to test whether root exudates can carry specific information about the genetic relatedness, species identity and population origin of neighbours, and trigger responses indicative of selfish or cooperative behaviour. We examined the impact of root exudates in the absence of neighbours, as neighbours are likely to affect nutrient availability and may emit other signals. We designed an experiment in which: soil properties (including natural microbial communities) mimicked naturally occurring conditions; the concentrations of exudates did not exceed those occurring under natural conditions; and plants were exposed to root exudates in a spatially explicit way – exudate solution was applied to a soil patch on one side of the recipient plant, mimicking the presence of a neighbour at that location, and a control solution was added to a soil patch on the opposite side. Root exudates were collected from plants of different identity relative to the recipient plant, including siblings, unrelated conspecifics and heterospecifics from the same and a distant community. Deschampsia caespitosa, a common grass species of nutrient-rich moist meadows, was used as the focal species. By comparing root mass and morphology in soil patches treated with control solution and exudates of different origin, we show that root exudates can act as cues of neighbour identity and trigger complex responses in the recipient plants.
Materials and Methods
Root exudate collection
Root exudates were collected as water solution that leached through soil occupied by plant groups of different identity relative to the recipient plants used in experiments involving the following: (T1) siblings – offspring of the same mother plant as the recipient plant (probably half-siblings as the pollen source was not controlled); (T2) unrelated conspecifics from the same community as the recipient plant (originating from a mixture of different mother plants; location 58°25′32″N; 26°30′40″E); (T3) representatives of a different species co-occurring in the same community as the recipient plant (location 58°25′32″N; 26°30′40″E); (T4) conspecifics from a different population (location 58°27′51″N; 25°12′19″E); and (T5) representatives of a different species inhabiting a different community (location 58°27′51″N; 25°12′19″E). Both study sites are moist, nutrient-rich meadows. Deschampsia caespitosa (L.) was used as the focal species (i.e. species receiving exudates) and Lychnis flos-cuculi (L.), a forb species that frequently occurs with D. caespitosa, was used in heterospecific treatments. The choice of the focal species was guided by a previous study, in which plants of D. caespitosa grew larger in groups of siblings than nonsiblings, implying the possible involvement of kin recognition (Lepik et al., 2012).
In order to create plant groups for exudate collection, seeds of appropriate identity and origin were sown on the surface of 5-l pots filled with a 1 : 1 mixture of fine sand and commercially available soil (pH 6.5, water-soluble N 100 mg l−1, P 80 mg l−1, K 400 mg l−1). Natural soil from the home site of recipient plants was added at the rate of 25 g l−1 of soil mixture to provide plants with natural soil biota. The bottoms of the pots contained a 1-cm-thick layer of gravel to improve drainage. Six replicate pots were used per treatment, except for the sibling treatment (T1), which was represented by 10 pots sown with seeds originating from 10 different mother plants. Three weeks after seed sowing, seedlings were thinned to leave 15 plants in each pot. All pots were placed randomly on a bench in a glasshouse with a 16 h : 8 h, day : night illumination cycle. Exudate collection started two months after seed sowing. Each pot was watered until 120 ml of solution leached from the bottom of the pot. Collected solutions were immediately frozen at −18°C. Exudates were collected twice a week, for 5 wk, and each new solution was added to the corresponding previous collection (i.e. 1200 ml per pot was collected in total).
Root exudate application
Before the start of the experiment, exudate solutions were filter-sterilised (pore size 0.2 μm, Minisart, Sartorius Stedim biotech, Göttingen, Germany) and their nutrient ion concentrations (K+, NO3− and PO43−) determined using ion chromatography (ICS-1000, Dionex Corp., Sunnyvale, CA, USA). To minimise differences in nutrient concentrations between the control solution and different exudate solutions, concentrated liquid fertiliser (6% N, 2.18% P, 4.15% K) was added at the rate of 1.8 ml to 1 l of either water or exudate solution. The final N, P and K concentrations varied among the control and different exudate solutions between 58.2 and 59.5 mg l−1, 39.7 and 42.8 mg l−1, 88.4 and 270.1 mg l−1, respectively. The results of the experiment showed that the remaining differences in nutrient concentrations between different solutions had little effect on root growth (R2 of the relationship between solution N, P or K concentration and root length densities at the site of exudate application were 0.04, 0.02 and 0.04, respectively; P-values 0.1300, 0.3153 and 0.1371, respectively; n = 59, one missing value).
Seeds collected from ten mother plants of D. caespitosa from the focal study population (58°25′32″N; 26°30′40″E) were germinated on moist sand. Seven days later, six seedlings from each mother plant were transplanted singly into the centre of 3.5-l pots containing soil mixture prepared in the same way as for the root exudate collection. Pots were distributed equally between the five exudate treatments and a control so that each mother plant was represented in each treatment, resulting in ten replicates per treatment. All plants were placed randomly on benches in a glasshouse with a 16 h : 8 h, day : night illumination cycle. The position of the pots was re-randomised weekly.
Exudate application started on the fourth day after seedling transplantation and was repeated twice a week for 10 wk in total. Exudates were applied to the pots in a consistent, spatially explicit pattern. Two plastic cylinders (2.5 cm in diameter and 7 cm in height) were installed on the soil surface (pressed into the soil to a depth of 0.5 cm) on each side of the recipient plant at a distance of 4 cm from the rooting point. The function of the cylinders was to hold solution as it was applied to the soil. Control solution was applied within one of the cylinders and exudate solution within the other cylinder; control solution was added to both cylinders in the control treatment. Every 2 wk, the quantity of control and exudate solution was increased as plant size increased as follows: 25 ml → 30 ml → 35 ml → 45 ml → 55 ml.
After 83 d of growth, the shoots of each focal plant were harvested and soil cores of 3.8 cm diameter and 14 cm depth were collected from directly beneath the sites of exudate and control solution application. Roots contained in the soil cores were washed out and 5–7 adventitious roots with higher order roots attached were selected for morphological analysis. The selected roots were scanned (Epson Perfection V700 Photo, Epson, Japan) and the total length of main adventitious roots (diameter always > 0.4 mm) and more distal, finer branches (diameter always < 0.4 mm) were calculated using WinRhizo Pro 2008a (Regent Instruments Inc., Quebec, Canada). Scanned roots and all other roots were dried at 70°C for 48 h and weighed separately for each soil core. Root branching intensity was calculated as the ratio of the root length of higher order branches to the length of main adventitious roots. Specific root length was calculated as the ratio of the root length to the dry mass of the scanned root sample. Based on specific root length, total root dry mass retrieved from each soil core and the volume of each soil core (159 cm3), root length density per unit of soil volume was calculated. The number of shoots produced by each focal plant was counted and shoot mass was determined after drying at 70°C for 48 h.
In order to test for exudate-mediated kin recognition, linear mixed models were fitted to data collected from treatments T1 and T2, in which exudate origin (sibling or unrelated conspecifics from the same population), root sample location (site of control solution application or exudate application) and their interaction were included as fixed factors. To test for population- and species-specific effects, linear mixed models were fitted to data collected from treatments T2–T5, in which exudate community origin (same community or a different community), exudate species identity (conspecific or heterospecific), root sample location (site of control solution application or exudate application) and their interactions were included as fixed factors. In all models, the identity of the mother lineage and pot identity nested within mother lineage were included as random factors. Loge-transformed total root mass, root branching intensity, specific root length and root length density per unit of soil volume were used as response variables. Linear models including the same factors except for root sample location were fitted to test for the effects of exudates on shoot traits. Data analyses were performed using R 2.15.0 (R Development Core Team, 2012). Mixed models were implemented using R package nlme (Pinheiro et al., 2012).
At the level of the whole root system (i.e. averaged across the control and exudate patches within a pot), plants subjected to the exudates collected from unrelated individuals from the same population produced roots that had higher specific root length and were more branched than plants receiving sibling exudates (main effect of exudate origin in Table 1, Fig. 1a,b). As a result of these morphological changes and a moderate but nonsignificant increase in root mass, root length density averaged across the exudate and control soil patches was 77% greater in the unrelated than in the sibling treatment (Fig. 1c,d).
Table 1. The effects of exudate origin (sibling or nonsiblings from the same population) and root sample location (exudate or control patch) on four root traits in Deschampsia caespitosa
Specific root length (cm mg−1)
Root branching intensity
Root dry mass (g)
Root length density (cm cm−3)
Analysis was performed on loge-transformed data and models included mother plant identity and pot as random factors. *, P < 0.05; **, P < 0.01; ***, P < 0.001.
Although sibling exudates suppressed root proliferation at the level of the whole root system, they did not eliminate the localised response in the part of the root system subjected to exudates: independent of whether exudates came from siblings or unrelated individuals, greater root mass was recovered from the soil patch treated with exudates, and the roots had higher specific root length and were more branched than in the soil patch treated with the control solution (main effect of sample location in Table 1, Fig. 1a–c). As a result, root length density in the soil patch treated with exudates, either sibling or nonsibling, was on average 51% greater than in the patch treated with the control solution (Fig. 1d). However, root length density in the sibling treatment never exceeded that of the control plants (shown with a dashed line in Fig 1). By contrast, root length density in patches treated with the exudates of unrelated conspecifics was considerably higher than that of control plants.
Besides their involvement in kin recognition, root exudates had species- and population-specific effects on root growth. At the level of the whole root system, application of conspecific exudates resulted in greater root length density when obtained from individuals from the same population than when obtained from individuals from a different population, whereas exudates obtained from a different species (L. flos-cuculi) did not trigger any significant response independent of community origin (Fspecies × community = 4.9, df = 1, 26, P = 0.0363, Fig. 2). Although root length density at the level of the whole root system was regulated by a population-specific cue, a spatially localised response was mediated by a species-specific cue that was common to both examined populations: greater root length density was observed in the soil patch treated with conspecific exudates than in the patch treated with control solution independent of population origin, whereas heterospecific exudates did not trigger spatial changes in root distribution (Fspecies × sample location = 6.5, df = 1, 35, P = 0.0150, Fig. 2).
Exudate application did not have a significant effect on either shoot mass or the number of shoots produced by focal plants (P > 0.05 for all effects; Supporting Information Tables S1–S3).
Our results identify root exudates as a mechanism for highly specific neighbour recognition in plants. We found that the localised application of root exudates to a soil patch at one side of the recipient plant initiated clear changes in root growth both locally in the treated soil patch and at the level of the whole root system. First, root exudates could apparently mediate kin recognition and elicit responses to conspecifics from the same population that were consistent with the predictions of behavioural theory. Plants subjected to exudates collected from unrelated individuals from the same population showed a sharp increase in root length production at the level of the whole root system, whereas such a response was suppressed in plants receiving exudates collected from siblings (Fig. 1). Increased root length can be interpreted as an effort to enhance resource uptake and is, therefore, in accordance with the prediction that the presence of competitors should trigger the selfish behaviour of attempting to pre-empt resources from neighbours (Zhang et al., 1999; Gersani et al., 2001; Craine, 2006; O'Brien et al., 2007). A mechanism of genotypic recognition has been identified in Artemisia tridentata: plants receiving volatiles from genetically identical cuttings were more resistant to herbivory than plants receiving volatiles from nonself cuttings (Karban & Shiojiri, 2009). It has also been shown that immersing Arabidopsis seedlings in solutions containing exudates produced by seedlings of either the same or a different genetic lineage had a significant impact on root branching (Biedrzycki et al., 2010). Our results show that exudate-mediated kin recognition is possible even among siblings of out-breeding plants, and that the consequences of such recognition go beyond the seedling stage and involve multiple root traits. Many studies have examined the toxic properties of root exudates in plant–plant interactions known as allelopathy (reviewed in Schenk et al., 1999; Bais et al., 2006; Inderjit et al., 2011). This study shows that root exudates can also play a signalling role in neighbour recognition and elicit competitive root proliferation in nonsibling interactions, while suppressing root growth in sibling interactions.
Second, besides their involvement in kin recognition, root exudates had a population-specific effect on root growth. At the level of the whole root system, plants subjected to exudates obtained from individuals from the same population produced significantly greater root length than plants receiving exudates obtained from individuals from a different population (Fig. 2). The population specificity of root interactions has been documented in two other studies, but the mechanisms involved were not identified (Krannitz & Caldwell, 1995; Mahall & Callaway, 1996). Local adaptation to toxic root exudates has been shown in studies on invasive plants (He et al., 2009; Thorpe et al., 2009) and adaptation to the leaf chemicals of aromatic plants has been shown for associated native species (Grøndahl & Ehlers, 2008). Our results suggest that local adaptation to the chemical profile of root exudates may also be required for exudates to function as cues about the presence and identity of potential competitors.
Although we detected kin recognition and population specificity in plant responses to exudates at the level of the whole root system, the spatially localised response to exudate application was mediated by a species-specific cue. Root length density was significantly higher in the soil patch treated with conspecific exudates than in the soil patch treated with the control solution independent of population origin or whether exudates were collected from siblings or unrelated individuals. However, root growth was not stimulated in the soil patches treated with exudates collected from a different species, Lychnis flos-cuculi. These results indicate that root exudates contain multiple cues: some allow kin recognition and trigger changes in root growth at the level of the whole root system, whereas others are not genotype-specific and trigger spatially localised responses. Further research is required to identify the processes responsible for this pattern. Chemical compounds mediating kin and conspecific recognition may differ in structure and consequently also in diffusion rates or persistence in the soil. Exudates triggering a localised response may be compositionally less variable, precluding kin recognition. The tendency to maintain high root density in the patch containing sibling exudates may also be due to the low cost of root production in our experiment and could change in the presence of actual competition or a reduction in nutrient concentrations (Cahill et al., 2010; McNickle & Brown, 2012).
We recorded no responses to the exudates collected from a heterospecific, L. flos-cuculi. It is possible that the lack of a response is adaptive, for example if the two species do not strongly compete due to niche partitioning, or if sharing soil with L. flos-cuculi is not beneficial for the focal species (e.g. a negative impact on mutualist microbes). However, it is also possible that recognition of L. flos-cuculi occurs but is mediated by compounds that could not be collected or preserved using our experimental approach (e.g. volatiles).
Previous studies have generally examined allocation to root mass, rather than changes in root morphology, in order to evaluate the occurrence of tragedy of the commons or kin recognition in root interactions (Gersani et al., 2001; Maina et al., 2002; O'Brien et al., 2005; Dudley & File, 2007; Murphy & Dudley, 2009; Bhatt et al., 2010). In our study, greater root length density in patches containing exudates was primarily achieved through an increase in specific root length. The cost of trying to pre-empt resources from neighbours in this way may not be as high as it would be if achieved by increasing root mass. Higher specific root length ensures greater resource capture capacity without the need for additional root mass investment (Farley & Fitter, 1999; Mommer et al., 2011). Additionally, fine roots are known to be short-lived (Van der Krift & Berendse, 2002; McCormack et al., 2012). Although short root lifespan would usually be considered detrimental, it can be advantageous when resources are contested. Such roots may maximise their net benefit to a plant if they die soon after a soil patch shared with competitors is depleted of resources. However, if the nutrient patch persists, the production of fine roots requiring frequent replacement would still result in a less efficient strategy over time.
In summary, the results of this experiment show that root exudates can convey highly specific cues about the genetic identity of neighbours, and trigger different responses at the levels of entire root systems and of individual roots that come into direct contact with the exudates. The finding that root proliferation in response to conspecific exudates was achieved mainly through morphological adjustments, rather than biomass investments, suggests that apparently selfish root proliferation in response to neighbour presence may not have ‘tragic’ consequences for fitness. Root morphological plasticity could be included in future game-theoretic models as a mode of adjusting the costs and benefits of root production in the presence of competitors. Lastly, our study has identified root exudates as the general mechanism of neighbour recognition but further research is needed to identify which chemical compound or combination of compounds in the root exudates triggers the observed responses.
We thank John Davison, Mike Hutchings, Mari Moora, Kristjan Zobel, James Cahill and two anonymous reviewers for providing discussions and comments on the manuscript. Martin Liira provided technical support with the chemical analysis of root exudates. This study was supported by the University of Tartu (0119) and Estonian Science foundation (grant 9332).