Intraspecific genetic diversity modulates plant–soil feedback and nutrient cycling
Summary
- Plant genetic diversity can affect ecosystem functioning by enhancing productivity, litter decomposition and resistance to natural enemies. However, the mechanisms underlying these effects remain poorly understood. We hypothesized that genetic diversity may influence ecosystem processes by eliciting functional plasticity among individuals encountering kin or genetically diverse neighbourhoods.
- We used soil conditioned by groups of closely related (siblings) and diverse genotypes of Deschampsia cespitosa – a species known to exhibit kin recognition via root exudation – to investigate the consequences of kin interactions for root litter decomposition and negative feedback between plants and soil biota.
- Genetically diverse groups produced root litter that had higher nitrogen (N) content, decomposed faster and resulted in greater N uptake by the next generation of seedlings compared with litter produced by sibling groups. However, a similar degree of negative soil feedback on plant productivity was observed in soil conditioned by siblings and genetically diverse groups. This suggests that characteristics of roots produced by sibling groups slow down N cycling but moderate the expected negative impact of soil pathogens in low-diversity stands.
- These findings highlight interactions between neighbouring genotypes as an overlooked mechanism by which genetic diversity can affect biotic soil feedback and nutrient cycling.
Introduction
Plant species diversity and associated variation in functional traits are central to the maintenance of ecosystem services, such as primary production, carbon storage and nutrient cycling (Knops et al., 2002; Steinbeiss et al., 2008; Cong et al., 2014). Species-rich plant communities also experience reduced pathogen load and specialist herbivore damage compared with species monocultures (Mitchell et al., 2002; Lau et al., 2008; Schnitzer et al., 2011). Although the importance of species diversity is well established, there is increasing evidence that intraspecific genetic diversity plays an important role in maintaining species co-existence (Booth & Grime, 2003; Whitlock, 2014; Schoeb et al., 2015) and enhancing net primary productivity (Crutsinger et al., 2006), litter decomposition (Schweitzer et al., 2005), and resistance to damage by herbivores and foliar pathogens (Zhu et al., 2000; Parker et al., 2010; Tooker & Frank, 2012; Barton et al., 2015). The mechanisms linking intraspecific genetic diversity with ecosystem functioning are still poorly understood, not least because genetically diverse populations may enhance or inhibit ecosystem processes to a greater degree than predicted based on the effects of constituent genotypes in isolation (Hughes et al., 2008).
Nonadditive effects of genetic diversity on ecological processes may be explained by a number of mechanisms. Primary productivity might be enhanced by resource niche partitioning among diverse plant genotypes, although empirical evidence for this is lacking (Crutsinger et al., 2006; Atwater & Callaway, 2015). The effect of plant genetic diversity on ecosystem processes may also be mediated by microbial and invertebrate composition and activity in the soil. Mixed-genotype plant litter has been shown to decompose faster than expected from the decomposition rates of single-genotype litters (Schweitzer et al., 2005; Wang et al., 2014). This effect has been attributed to resource complementarity whereby mixed litter provides a more complete range of substrates to a more diverse set of decomposers. How the increased rates of decomposition in genotypic mixtures affect the balance between nitrogen mineralization and immobilization is not known but significant feedback effects on plant nutrition could be expected.
In addition to effects mediated by nutrient cycling, genotypic diversity within species could reduce negative plant–soil feedback by slowing down the accumulation of soil pathogens, as has been shown for diversity at the species level (Maron et al., 2011; Schnitzer et al., 2011). This prediction is supported by studies that demonstrate genotypic and population specificity of plant interactions with antagonistic soil biota (Liu et al., 2015; Luo et al., 2016), and reduced levels of aboveground herbivory and leaf pathogen damage in genotypic mixtures compared with monocultures (Parker et al., 2010; Tooker & Frank, 2012; Barton et al., 2015). The latter finding has been attributed to the difficulty for natural enemies in spreading to susceptible genotypes and/or induced defence in susceptible genotypes when growing among resistant genotypes (associational resistance; Barbosa et al., 2009).
In addition, plant genotypes may express different phenotypes when grown in mixtures compared with genotypic monocultures, and such plasticity may have consequences for ecosystem processes. In particular, several studies have demonstrated the ability of plants to recognize kin via root exudates and volatiles (Biedrzycki et al., 2010; Karban et al., 2013; Semchenko et al., 2014) and modify their growth, biomass allocation and morphology in response to the genetic relatedness of neighbours (Dudley & File, 2007; Bhatt et al., 2010; Biernaskie, 2011; Lepik et al., 2012). Such plasticity represents a special case of indirect genetic effects, which are defined as the effect of a genotype on the phenotype of neighbouring conspecifics (Mutic & Wolf, 2007; Costa e Silva et al., 2013). Interactions with kin are likely to increase selection for a resource-conservative strategy whereby resources are diverted away from wasteful competition with closely related individuals towards investment into tissue longevity and defence against natural enemies, which are known to spread faster in genetically homogenous compared with diverse stands (File et al., 2012a,b; Semchenko et al., 2013). Improved tissue longevity and protection can be achieved by the production of thicker leaves and roots with higher tissue density and lower concentration of nitrogen relative to carbon compounds, such as lignin (Tjoelker et al., 2005; Roumet et al., 2016). Such phenotypic changes can have cascading effects on soil microbial activity and composition, which can, in turn, feed back to influence plant phenotype, nutrition and fitness (Whitham et al., 2006; Madritch & Lindroth, 2011; Bray et al., 2012). Thus, plant responses to the genetic relatedness of neighbours represent a process with strong potential to influence ecosystem functioning. However, its role has not been explicitly tested.
In the present study, we test the hypothesis that plant responses to the genetic relatedness of surrounding individuals and associated changes in functional traits can have significant consequences for nutrient cycling and feedback between plants and soil biota. The study builds on our previous discovery that root exudates of genetically diverse individuals of Deschampsia cespitosa trigger the production of longer, more branched roots with higher specific root length (SRL), whereas sibling exudates suppress such a response (Semchenko et al., 2014). As low SRL is associated with slow growth and high tissue longevity, this finding likely reflects a switch to a more resource-conservative strategy: low SRL in the presence of closely related neighbours raises construction costs of resource-capturing organs but may reduce susceptibility to pathogen attack and decomposition (Tjoelker et al., 2005; Reich, 2014; Lemmermeyer et al., 2015; Roumet et al., 2016). We therefore predicted that (1) plants grown in groups of closely related individuals produce low quality root litter that decomposes more slowly compared with groups of genetically diverse plants; and (2) changes in root properties in response to interactions with kin mitigate the stronger negative plant–soil feedback expected in stands with low genetic diversity. This was tested by growing D. cespitosa plants in either groups of siblings or genetically diverse individuals and following the decomposition of produced root litter, and the growth of and N uptake by the next generation of seedlings in soils conditioned by plants of different identities.
Materials and Methods
Experimental design
Soil conditioning phase
Seeds from multiple mother plants of Deschampsia cespitosa (L.) P. Beauv. were collected from a flooded meadow in Estonia with a total area of c. 4 ha (58°25′32″N; 26°30′40″E). Plants were grown as either sibling groups (offspring of the same mother plant) or genetically diverse groups composed of individuals from multiple mother plants from the same population (Fig. 1). In the sibling treatment, 10 pots were sown with seeds from 10 seed heads (a single seed head per pot), each collected from a different mother plant separated from all others by at least 20 m. Pots with genetically diverse plant groups were replicated six times and were assembled using seed mixture from the same ten locations as those used in the sibling treatment – 10 seed heads used in the sibling treatment along with additional seed heads from the immediate vicinity (< c. 2 m). Additional seed heads were used in the genetically diverse groups because the ten seed heads used to generate the sibling groups did not provide sufficient seeds for both conditioning and feedback stages of the experiment. Deschampsia cespitosa is a self-incompatible grass species that forms dense tussocks of genetically identical shoots but does not exhibit long-distance clonal dispersal (Halevy, 2000). It is therefore highly likely that seeds collected from shoots growing > 20 m apart represent different genotypes. Seeds of appropriate identity 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 was added at the rate of 25 g l−1 of soil mixture to provide plants with natural soil biota. Three weeks after seed sowing, seedlings were thinned to leave 15 plants in each pot. In addition, six pots were filled with the same soil mixture but no seedlings were planted in these pots to create a control treatment. All pots were placed randomly on a bench in a glasshouse with a 16 h : 8 h, day : night illumination cycle.
All pots were harvested after 3 months. Shoots were cut at soil level and removed but roots remained in the conditioned soil. Soil from each pot was broken up, mixed and split into two equal parts, and one part was sterilized using gamma radiation (dose 15 kGy). Soil from each sibling group was kept separately to maintain genetic relatedness between plants in the conditioning and feedback stages, whereas control soil and soil conditioned by genetically diverse groups were each pooled within each sterilization treatment and split between 10 replicate pots for the feedback experiment. Further details on soil handling are presented in Supporting Information Methods S1.
Three soil samples were taken at random from each conditioning (siblings, genetically diverse, control) and sterilization (sterile and nonsterile) treatment. Within the sibling treatment, the same three randomly selected sibling groups were sampled from sterile and nonsterile treatments. Roots were separated from the soil, dried, and root carbon (C) and nitrogen (N) content was determined using a CN elemental analyser (Elementar vario EL cube; Elementar Analysensysteme GmbH, Langenselbold, Germany). Soil pHKCl and N (Kjeldahl method), available phosphorus (P), available potassium (K) and organic matter content were determined according to methods described in Moore & Chapman (1986).
Soil feedback experiment
Seeds collected from the same 10 mother plants as those used to create sibling groups in the conditioning phase were germinated on moist sand. Seven days later, six seedlings from each mother plant were transplanted (a single seedling per pot) into the centre of 1.3-l pots filled with soil (0.57 kg DW) that was (1) conditioned by siblings, genetically diverse groups or no plants (control), and (2) sterilized or left unsterilized (Fig. 1). Soil conditioned by sibling groups was occupied in the feedback stage by a seedling from the same maternal lineage. This resulted in a factorial design of three conditioning and two sterilization treatments, with 10 replicates for each treatment combination (60 pots in total). All plants were placed randomly on benches in a glasshouse with a 16 h : 8 h, day : night illumination cycle. The position of pots was re-randomized every 2 wk. As the soil had been extensively exploited by dense vegetation during the conditioning phase, plants were fertilized four times during the feedback stage with 20 ml of 0.7% solution of liquid fertiliser (6% N, 2.18% P, 4.15% K) in the feedback stage. This partially replenished nutrients that were removed from the soil during the conditioning stage. In the natural habitat of the study species, nutrient deposition occurs in the form of yearly flooding.
Two weeks after transplantation, the length of the longest leaf was registered for each seedling to examine the effect of soil feedback at early stages of seedling establishment. Plants were harvested after 58 d of growth. Roots were carefully washed out of the soil and separated into roots belonging to focal plants (attached to aboveground shoots and different in colour and texture from dead roots) and dead roots that were left undecomposed in the conditioned soil. In our experiment, we followed the approach for in situ decomposition estimation as proposed by Dornbush et al. (2002). Namely, the dead roots from the conditioning stage were left to decompose in the soil undisturbed, rather than being washed out, dried, weighed and placed in a litter bag. Although not as precise as estimates obtained using litter bags, this approach retains realistic conditions of root litter density and spatial distribution, and allows close contact between dead roots, soil and the live roots of newly planted seedlings. Soil from each pot in the conditioning stage was divided into two equal parts, and one part was sterilized with gamma radiation. As decomposition was largely suppressed in sterilized soil (although some microbial colonization probably occurred in the course of experiment), a comparison of the remaining dead root mass in nonsterile and sterile soil at the end of the feedback stage provided an estimate of root decomposition. Initial root litter mass for each conditioned soil was estimated using shoot biomass data from the conditioning stage and the mean root-to-shoot biomass ratio from a previous study that used the same soil and maternal lines as in the current study, as well as similar growth conditions and duration (Semchenko et al., 2014). Using predicted initial root mass and litter mass remaining in soil at harvest, we calculated the proportion of litter mass remaining after 58 d of incubation. All plant material was dried at 70°C for 48 h and weighed separately. Root and shoot C and N content of focal plants were determined using an Elementar vario EL cube analyser (Elementar Analysensysteme GmbH). An adventitious root branch with all attached higher order branches was selected from live roots and scanned (Epson Perfection V700 Photo; Epson, Suwa, Japan) and the total root length was calculated using WinRhizo Pro 2008a (Regent Instruments Inc., Quebec City, QC, Canada). Specific root length (SRL) was calculated as the ratio of the root length to the dry mass of the scanned root sample. Trait data are presented in Table S1.
Statistical analysis
Soil conditioning phase
Linear fixed effects models were used to assess the effects of conditioning treatment (siblings vs genetically diverse groups) on root C and N content and C : N ratio. Similarly, linear fixed effects models were used to estimate the effects of soil conditioning (siblings, genetically diverse groups, control) and sterilization on soil pH, N, P, K and organic matter content at the end of the conditioning phase.
Soil feedback experiment
Linear mixed models were used to estimate the effects of soil conditioning (siblings, genetically diverse groups) and sterilization (sterile, nonsterile) on each of the following response variables in the feedback experiment: (1) proportion of root litter remaining after 8 wk of incubation, (2) initial seedling growth at 2 wk (maximum leaf length), (3) total dry mass at harvest, (4) root : shoot biomass ratio, (5) specific root length and (6) C and N content and C : N ratio of focal plants (root and shoot). Mother plant lineage was included in the analysis as a random factor. Data were loge-transformed (or square root-transformed for the proportion of litter mass remaining) before analysis if necessary to satisfy model assumptions. Distributional assumptions were checked by examining normal probability plots, and variances were compared using Levene's test.
The effect of the soil sterilization treatment reflects the result of eliminating soil biota but also a potential flush of nutrients from dead microbial cells (McNamara et al., 2003). To obtain information on the role of soil biota associated with plants without the confounding effect of a nutrient flush, we calculated a soil feedback index that compared plant traits in soil conditioned by different plant groups to soil that was unoccupied by plants during the conditioning stage. The soil feedback index was calculated as ln(Ttreat/Tcontrol), where Ttreat is the trait mean for plants in each soil conditioning × sterilization treatment combination (soil conditioned by siblings or genetically diverse groups and either sterile or nonsterile soil) and Tcontrol is the trait mean for plants grown in the corresponding control soil (no plants in the conditioning stage and either sterilized or left unsterilized). Negative values indicate reduced plant performance in soil conditioned by plants compared to control soil that was not occupied by any plants in the conditioning stage. Means and standard errors for the soil feedback index were obtained as parameter estimates from linear mixed models containing the three soil conditioning treatments (fixed factor: siblings, genetically diverse groups and control as the reference level) and mother plant lineage as a random factor. Models were performed separately for sterile and nonsterile soil to allow comparisons to be made with the appropriate control soil. For seedling length and root C : N ratio, the index was calculated on untransformed data (Ttreat − Tcontrol) as loge-transformation resulted in a skewed distribution.
All analyses were performed using R v.3.3.2 (R Core Team, 2016). Mixed models were performed using the lme4 package (Bates et al., 2015).
Results
Root decomposition and C/N content
The proportion of root litter remaining in the soil at the end of the feedback stage was similar between plant diversity treatments when soil was sterilized. In unsterilized soil, a significantly greater proportion of root litter was lost from soil conditioned by genetically diverse groups compared with soil conditioned by sibling plants (79% vs 48% litter mass loss, respectively; a significant interaction between the effects of soil conditioning and sterilization in Table 1; Fig. 2).
| df | Prop. litter remaining | Root N% | Total N (g) | Seedling length (cm) | Plant mass (g) | |
|---|---|---|---|---|---|---|
| Conditioning (C) | 1 | 4.9* | 8.4** | 1.2 | 0.1 | 0.6 |
| Sterilization (S) | 1 | 33.8*** | 54.0*** | 49.5*** | 25.8*** | 101.6*** |
| C × S | 1 | 6.5* | 8.9** | 13.9*** | 2.0 | 0.1 |
| Residuals | 27 |
- F-values and their significance are shown: *, P < 0.05; **, P < 0.01; ***, P < 0.001.
Groups of siblings and genetically diverse plants produced similar total aboveground biomass by the end of the conditioning stage (mean mass 25.3 g and 24.7 g, respectively; P = 0.7329, t-test). The roots of plants grown in groups of genetically diverse individuals in the conditioning stage had 28% higher N content than roots produced when grown among siblings, resulting in a considerably lower root C : N ratio in genetically diverse groups compared with sibling plants (Table 2). These differences in root N content carried over to the next generation of plants in the feedback stage, but were only detectable in the unsterilized soil treatment (interaction between the effects of soil conditioning and sterilization, N% F1,27 = 8.9, P = 0.0060; C : N F1,27 = 8.0, P = 0.0088; C% F1,27 = 2.4, P = 0.1353; total plant N F1,27 = 13.9, P = 0.0009; Table 1). Plants grown in unsterilized soil conditioned by groups of genetically diverse individuals produced roots with significantly higher N content and a significantly lower C : N ratio compared with plants grown in soil conditioned by siblings (both differences P < 0.05, Tukey test; Figs 3a, S1). The main cause for differences in C : N ratios was the higher root N content of plants grown in soil conditioned by genetically diverse groups and a slight increase in C content in the roots of plants grown in soil conditioned by siblings relative to those grown in the unconditioned control soil (i.e. soil that was not occupied by any plants in the conditioning stage; both differences P < 0.05, t-test; Figs 3e, S1). Total N uptake of plants (shoots and roots combined) grown in unsterilized soil conditioned by genetically diverse groups was similar to uptake by plants grown in unconditioned control soil, whereas plants in soil conditioned by siblings exhibited a significantly lower N uptake compared with the unconditioned control and genetically diverse treatments (P < 0.05; Fig. 3b,f).
| Treatment | Siblings | Diverse | |
|---|---|---|---|
| C% | F1,10 = 3.8$ | 45.45 (0.14) | 44.77 (0.32) |
| N% | F1,10 = 6.0* | 0.57 (0.02) | 0.74 (0.06) |
| C : N | F1,10 = 8.7* | 80.31 (3.31) | 62.76 (4.94) |
- F-values, the significance of the treatment ($, P < 0.1; *, P < 0.05), raw means and standard errors (in parentheses) are shown.
Soil feedback effects on plant growth
Initial seedling growth was not significantly affected by soil conditioning (Table 1; Fig 3c). A marginally nonsignificant suppression of growth in soil conditioned by siblings compared to unconditioned control was observed in the unsterilized treatment (Fig. 3g). After 8 wk of growth, plant growth was significantly improved by soil sterilization (Table 1; Fig. 3d). Compared with plant biomass in soil that was not occupied by plants in the conditioning stage, plant growth was strongly suppressed in unsterilized soil and not significantly affected in sterile soil (Fig. 3h). This suggests that plant presence in the conditioning stage modified soil biota with a negative impact on plant growth in the feedback stage of the experiment. The strength of growth suppression was similar in soils conditioned by siblings and genetically diverse groups.
Effects of soil conditioning on biomass allocation and root morphology
The root : shoot biomass ratios of plants in the feedback stage of the experiment were not significantly affected by soil sterilization and plant diversity in the conditioning stage (all effects P > 0.1; Table 3). There was a marginally nonsignificant interactive effect of soil sterilization and plant diversity treatments on SRL (F1,27 = 3.9, P = 0.0598; Table 3). When soil was sterilized, plants produced roots with lower SRL when grown in soil conditioned by siblings compared with soil conditioned by diverse plant groups. In unsterilized soil, SRL was similar in soil conditioned by siblings and genetically diverse groups.
| Root : shoot ratio | SRL (cm mg−1) | ||
|---|---|---|---|
| Nonsterile | Siblings | 0.45 (0.03) | 47.5 (4.8) |
| Diverse | 0.34 (0.02) | 44.0 (5.3) | |
| Sterile | Siblings | 0.41 (0.03) | 32.6 (5.1) |
| Diverse | 0.42 (0.05) | 47.0 (5.6) | |
- Raw means and standard errors (in parentheses) are shown.
Effects of soil conditioning and sterilization on soil properties
Soil conditioned by groups of genetically diverse individuals and siblings both resulted in significantly lower soil organic matter content and higher pH compared with soil that remained unoccupied by plants during the conditioning stage (significant effect of soil conditioning F2,14 = 11.7, P = 0.0010 and F2,14 = 128.9, P < 0.0001, respectively; Tables S2, S3). Soil conditioned by siblings exhibited higher K content than soil conditioned by genetically diverse groups (significant effect of soil conditioning F2,14 = 16.7, P = 0.0002) but did not differ significantly in N and P availability, organic matter content, or pH (P > 0.05; Tables S2, S3).
Discussion
In the present study, we tested whether phenotypic plasticity to neighbour identity can mediate the effects of plant genetic diversity on ecosystem processes by triggering changes in root chemical properties, with consequences for nutrient cycling and plant–soil interactions. We found that plants produced roots with higher nitrogen (N) content when grown among genetically diverse individuals compared with groups of closely related siblings. Also, root litter produced in genetically diverse groups decomposed faster and resulted in greater N uptake by the next generation of seedlings compared with litter produced in sibling groups. However, plant biomass accumulation was equally suppressed in soils conditioned by genetically diverse and sibling groups compared with soils that were unoccupied by any plants in the conditioning stage. This suggests that the slower N cycling observed in low- compared with high-diversity stands may have been offset by a superior ability of plants to resist negative feedback from soil biota, which is reported frequently in soils conditioned by conspecifics (Maron et al., 2011; Schnitzer et al., 2011).
Consequences for nutrient cycling
Several studies have now demonstrated that plants respond to the genetic composition of their neighbourhoods, including genetic relatedness within species (Murphy & Dudley, 2009; Biedrzycki et al., 2010; Lepik et al., 2012) and neighbour species identity (Mahall & Callaway, 1992; Semchenko et al., 2007; Abakumova et al., 2016). If these responses lead to changes in leaf or root litter quality, modified rates of decomposition and nutrient cycling can be expected (Genung et al., 2013). However, experiments investigating the effects of genetic diversity on litter decomposition and nutrient cycling have almost exclusively used artificially mixed litter from genotypes that were grown in isolation (e.g. Madritch & Hunter, 2005; Schweitzer et al., 2005; Madritch et al., 2006; Crutsinger et al., 2009). Our results show that such an approach overlooks the role of biotic interactions and indirect genetic effects in determining litter quality and mediating the relationship between genetic diversity and ecosystem functioning. We found that root litter produced in groups of genetically diverse individuals was characterized by higher N content and lower carbon (C) : N ratio compared with litter produced in groups of siblings, which translated into faster rates of litter decomposition and higher N uptake by seedlings in the next generation. Notably, differences in N uptake were evident only in unsterilized soil. This suggests that additional N in seedlings grown in soil conditioned by genetically diverse plants was derived from decomposing root litter. Deschampia cespitosa is a dominant species in its home community and frequently interacts with conspecifics (Lepik et al., 2012; Semchenko et al., 2013). The pronounced effects of genotypic diversity in this species may thus be expected to have important consequences throughout the ecosystem. It is likely that plant genetic diversity left a positive legacy effect on nutrient cycling not only via higher litter quality, but also by supporting a more diverse decomposer community. Plant genotypes can support distinct soil microbial communities due to differences in the chemical composition of root exudates and litter (Micallef et al., 2009; Madritch & Lindroth, 2011), which may be further modified by interactions with multiple genotypes occupying the same soil space. Further manipulative experiments are needed to ascertain the relative contribution of litter properties and decomposer communities as mediators of plant diversity effects on nutrient cycling.
To date, the vast majority of studies have reported kin recognition based on the observation that plants modified their phenotype in response to the genetic relatedness of neighbours (reviewed in File et al., 2012b; Lepik et al., 2012). However, it is often unclear if the response to kin was a result of neighbour perception via specific cues or was mediated by resource competition. According to niche partitioning theory, closely related plants are expected to have more similar resource niches and experience stronger competition compared to genetically diverse groups (MacArthur & Levins, 1967; Silvertown, 2004). Greater niche overlap and competition for limiting nutrients among siblings should result in lower population-level productivity and greater allocation to fine root production. By contrast, we found no differences in biomass production in this study and have recorded slightly greater biomass accumulation and decreased fine root production in sibling groups in earlier studies (Lepik et al., 2012; Semchenko et al., 2014). We also showed for the same population of D. cespitosa that fine root production was suppressed even in the absence of direct competition for resources when exposed to the root exudates of siblings compared with nonsiblings (Semchenko et al., 2014). However, it is possible that resource uptake dynamics and interactions with soil organisms (File et al., 2012a; Pickles et al., 2017) interact with responses to neighbour identity and jointly contribute to the effects of genetic diversity on ecosystem functioning.
Consequences for biotic soil feedback
It has been shown previously that higher genetic diversity of plant populations results in lower natural enemy spread and damage aboveground (associational resistance; Zhu et al., 2000; Tooker & Frank, 2012; Barton et al., 2015). It could therefore be expected that soil pathogens also accumulate faster and cause greater damage when plants are grown in closely related compared with genetically diverse groups (Luo et al., 2016). Greater negative impact of soil biota on plant growth in stands with low genetic diversity may also be due to stronger competition with soil microbes for nutrients. We found that the growth of D. cespitosa seedlings was equally suppressed in soils conditioned by siblings and genetically diverse individuals relative to control soil that remained unvegetated during the conditioning stage. When soils were sterilized by gamma radiation, plant growth was no longer suppressed, suggesting that the observed negative soil feedback was caused by plant-associated soil biota. The absence of differences in plant biomass accumulation between soils conditioned by siblings and genetically diverse groups is intriguing given that the latter soil enabled higher plant N uptake. Therefore, despite the positive effect of litter decomposers on plant nutrition, other soil organisms seemingly prevented plants from achieving greater biomass in soil previously occupied by genetically diverse plants compared with soil conditioned by siblings. In a previous study, we found evidence for kin recognition in the study species, D. cespitosa, which triggered production of roots with lower specific length when subjected to sibling root exudates (Semchenko et al., 2014). The current study shows that plants grown together with siblings also produce roots with higher C : N ratio. This change in root properties reflects a shift from maximizing growth rates towards greater longevity (Tjoelker et al., 2005; Reich, 2014) and may make roots less susceptible (and attractive) to soil pathogens (Herms & Mattson, 1992; Lemmermeyer et al., 2015). Therefore, it is possible that the expected pattern of greater natural enemy damage in locations with low genetic diversity may be partially offset in species capable of recognizing kin and shifting available resources from competition to other functions, such as defence and longevity. However, detailed analysis of root chemical composition and abundance of soil pathogens is necessary to evaluate this hypothesis.
In the present study, soil conditioned by sibling groups was occupied in the feedback stage by a seedling from the same maternal lineage to represent a scenario where low genetic diversity at the population level and/or poor seed dispersal result in closely related individuals occupying the same soil patch in consecutive generations. Besides the perception of their current neighbours, plants also may be affected by identity cues left by previous generations (Mazzoleni et al., 2015). We found a marginally nonsignificant interactive effect of plant genetic diversity and soil sterilization on the root morphology of seedlings in the next generation. Plants had very similar root morphology in unsterilized soil but had lower specific root length when grown in sterilized soil conditioned by siblings. A similar response was observed in a previous study when plants were exposed to root exudates produced by siblings (Semchenko et al., 2014). This finding suggests that neighbour identity cues may have remained in the soil after plant death but were modified or weakened by soil microbes in unsterilized soil.
Conclusions
This study proposes a novel mechanism by which intraspecific genetic diversity may affect ecosystem functioning: some plant species exhibit an ability to shift their resource allocation between growth- and longevity-related functions depending on the genetic diversity of their neighbourhoods, with consequences for litter decomposition, nutrient cycling and plant–soil feedback. This finding highlights the need to account for functional variability within plant genotypes when linking diversity to ecosystem processes, and to work towards integrating research on phenotypic plasticity with community and ecosystem ecology.
Acknowledgements
We thank John Davison, Richard Bardgett and Kristjan Zobel for providing valuable comments on the manuscript. Jaan Aruväli and Deborah Ashworth provided technical support with the chemical analysis of plant samples. This study was supported by the University of Manchester, Estonian Science foundation (grant 9332) and Institutional Research Funding (IUT 20-31) of the Estonian Ministry of Education and Research.
Author contributions
M.S. and A.L. designed and carried out the experiment; M.S. and S.S. analysed the data; and M.S. and S.S. wrote the paper with input from A.L.
