Environmental variation has stronger effects than plant genotype on competition among plant species


  • Marc T. J. Johnson,

    1. Department of Ecology and Evolutionary Biology, University of Toronto, Toronto, ON, M5S 3B2, Canada;
    2. Department of Biology, Duke University, Durham, NC 27708-0338, USA;
    Search for more papers by this author
  • Russell Dinnage,

    1. Department of Ecology and Evolutionary Biology, University of Toronto, Toronto, ON, M5S 3B2, Canada;
    Search for more papers by this author
  • Alice Y. Zhou,

    1. Department of Ecology and Evolutionary Biology, University of Toronto, Toronto, ON, M5S 3B2, Canada;
    2. School of Medicine, University of Pennsylvania, Philadelphia, PA 19104-6087, USA; and
    Search for more papers by this author
  • Mark D. Hunter

    1. Ecology and Evolutionary Biology and Natural Resources and Environment, University of Michigan, Ann Arbor, MI 48109-1048, USA
    Search for more papers by this author

Correspondence author. E-mail: mtj5@duke.edu


  • 1Competition is a key factor affecting the performance and co-existence of species. Most ecological research on competition treats species’ populations as phenotypically homogenous. However, plant populations typically contain genetic variation for multiple traits and have the potential to rapidly adapt to changing environments.
  • 2Recent theoretical and empirical research suggests that such variation and evolution may affect the ecological outcome of competitive interactions. We conducted a series of experiments to test the hypothesis whether genetic variation for competitive traits in a native plant (Oenothera biennis) affects the performance and diversity of competing plant species.
  • 3In greenhouse and field experiments, the presence of O. biennis reduced the performance and diversity of neighbouring plant species.
  • 4In greenhouse experiments, we detected heritable variation in O. biennis for above-ground and below-ground growth, and O. biennis varied genetically in response to competition, indicating the potential for adaptive evolution in response to selection by competitors.
  • 5Variation among O. biennis genotypes also affected the performance of neighbouring plants in the greenhouse, whereby genetic variation in O. biennis shoot : root ratio explained up to 41% of the variation in the performance of an exotic grass (Bromus inermis).
  • 6Despite effects of O. biennis genotype on B. inermis in the greenhouse, variable soil fertility had a much stronger effect on the grass's performance, and there were no effects of O. biennis genotype on neighbouring plants in the field.
  • 7Synthesis. Our results show that interspecific competition affected the biomass and diversity of plants, but heritable variation in competitive ability of O. biennis is not expected to affect short-term ecological dynamics in this system. Nevertheless, O. biennis has the potential to influence co-existence over longer timescales by adapting to competitors.


Interspecific competition frequently affects the performance and co-existence of species. Most models of competition assume that species’ populations are phenotypically homogenous, such that species have fixed values for traits that influence competitive ability (Tilman 1982; Holt et al. 1994; Chesson 2000). In reality, single populations often exhibit variation for multiple traits, and a combination of conceptual and mathematical theory suggests that intraspecific variation for competitive ability can affect species co-existence (Aarssen 1983; Abrams 2006; Urban 2006; Vellend 2006). Therefore, predicting the effects of competition on species interactions may rely on knowing the distribution of traits within species and the ecological effects of these traits on competitors. Here we describe a series of experiments that examined whether genetic variation in a native plant can affect competition with neighbouring plant species.

Recent research at the intersection of community ecology and evolutionary biology hypothesizes that heritable variation and evolution of ecologically important traits can affect ecological processes and patterns within communities (Whitham et al. 2003; Urban & Skelly 2006; Johnson & Stinchcombe 2007). Most research has focused on the ‘bottom-up’ effects of genetic variation in basal resource populations on consumer communities (Maddox & Root 1987; Dungey et al. 2000; Johnson & Agrawal 2005; Wimp et al. 2005), and how evolution in resource (Yoshida et al. 2003) and predator populations (Fussmann et al. 2003) shape predator–prey dynamics. Outside of extensive research on the evolution of character displacement (Schluter 2000), the effects of genetic variation and evolution on the ecology of competing species has received little attention compared to the study of multitrophic effects.

Genetic variation is predicted to affect the ecological outcome of competition when: (i) a population genetically varies in response to competition, indicating the potential for an evolutionary response to selection by competitors (Shaw & Platenkamp 1993; Cahill et al. 2005); or, (ii) genetic variation in a species’ traits affects the fitness of competing species by depleting or preventing access to shared resources (Aarssen 1989). Theoretical models indicate that these mechanisms can lead to greater or lesser co-existence among species compared to models that ignore genetic variation and evolution (Urban 2006; Vellend 2006). Most empirical evidence for the role of genetic variation and evolution in competitive interactions comes from plants. For example, Trifolium repens exhibits fine-scale adaptation to competition with specific plant species (Turkington & Harper 1979), and even to specific genotypes within competing species (Aarssen & Turkington 1985). Recent studies also show that variation among plant genotypes can influence the growth, biomass and density of competing plant species (Taylor & Aarssen 1990; Proffitt et al. 2005; Fridley et al. 2007; Lankau & Strauss 2007). Moreover, genetic variation in traits can lead to intransitive competitive interactions due to overlap in the competitive ability of co-existing plant species (Taylor & Aarssen 1990; Whitlock et al. 2007), which can increase co-existence within plant communities (Aarssen 1989; Laird & Schamp 2006; Lankau & Strauss 2007). Together, these results show that genetic variation in competitive ability can influence the performance of individual plants and potentially regulate diversity.

Further research is needed to understand whether genetic variation in competitive ability is ecologically important in nature. Most experiments have been conducted in pots where environmental variance is kept to a minimum to assess the potential effects of genetic variation on competitive interactions (Aarssen & Turkington 1985; Taylor & Aarssen 1990; Fridley et al. 2007). However, both the genotype and the environment determine a plant's phenotype, and environmental variation in the field may dampen the ecological effects of genetic variation. A combination of experiments from controlled growth environments and the field would provide the strongest test for the ecological importance of genetic variation. Studies that additionally manipulate ecological factors of known importance to competition (e.g. resource availability) provide further insight into the relative importance of genetic variation.

Here, we examine the hypothesis that genotype identity of a native plant (common evening primrose, Oenothera biennis) is an important factor affecting the performance and diversity of neighbouring plant species. Our first objective was to determine whether O. biennis competes with an exotic grass species Bromus inermis, with which it commonly co-occurs in nature. Once we established the nature of competitive interactions, we used greenhouse experiments to address three specific questions: (i) is there genetic variation in O. biennis for traits that can influence interspecific competitive interactions? (ii) Does genetic variation in plant traits of O. biennis affect the performance of B. inermis? and, (iii) what is the relative importance of plant genotype of O. biennis vs. variation in soil fertility in affecting the performance of B. inermis? Based on our results from pairwise competition greenhouse experiments, we then tested whether O. biennis genotype affects the community-wide performance and diversity of naturally occurring plants in the field.


study system

Our greenhouse experiments focused on interactions between two plant species common in open habitats of eastern North America, Common Evening Primrose (O. biennis L., Onagraceae) and Smooth Brome Grass (Bromus inermis Leyss., Poaceae). Oenothera biennis (hereafter ‘Oenothera’) is a native facultative biennial that forms a single basal rosette prior to bolting into a 0.5–2.5-m flowering stalk. It occurs commonly in open habitats, such as old fields, mowed lawns, and disturbed soil. Oenothera has a reproductive system (permanent translocation heterozygosity) which renders it functionally asexual (Cleland 1972). This genetic system makes for a powerful experimental tool because single clonal genotypes can be grown from seed and replicated across treatments and experiments. Bromus inermis (hereafter Bromus) is an exotic perennial grass that grows up to 1.5 m tall and produces tillers that form large clonal stands. Bromus and Oenothera often co-occur as they both colonize recently disturbed soil.

We also conducted a field experiment that examined competition between Oenothera and the diversity of plants that naturally colonize disturbed fields at University of Toronto's Koffler Scientific Reserve at Jokers Hill. All of the Oenothera genotypes used in experiments were collected as seed on or within 10 km of Jokers Hill.

greenhouse experiments

We conducted three greenhouse experiments that shared the same methods for planting, manipulation of competition, manipulation of soil fertility and measures of plant performance. Therefore, we provide a detailed description of the methods for ‘experiment 1’ and then highlight modifications to these methods for ‘experiment 2’ and ‘experiment 3’.

Greenhouse experiment 1

Our first objective was to assess whether Oenothera and Bromus compete when growing in close proximity, and whether variation in soil fertility alters the strength of competition. Oenothera and Bromus plants were grown in the presence and absence of one another, with and without fertilizer, in a fully factorial 2 × 2 design with 10 replicate plants per treatment combination. To examine the average competitive interactions between Oenothera and Bromus, we used a single genotype of Oenothera that exhibited intermediate above-ground and below-ground growth when compared to 21 other genotypes used in subsequent experiments. Seeds of Bromus came from a monospecific stand at Jokers Hill and because of the clonality of Bromus, the seed stock was likely derived from a single maternal family. The same seed stock was used in all experiments.

Seeds of both species were germinated in petri dishes on wet filter paper placed in a windowsill. Following germination, we transplanted seedlings to 250 mL pots containing a 1 : 2 mixture of potting soil (Promix General Purpose BX soil, Premier Horticulture, Dorval, Canada) and coarse sand, respectively. In the absence of competition, a single Oenothera or Bromus plant grew alone. In the presence of a competitor, one Oenothera plant and one Bromus plant were planted 3 cm apart in a single pot. Fertilized plants received 5 mL of 200 p.p.m. 20 : 20 : 20, N : P : K, liquid fertilizer (Plant Prod, Brampton, ON, Canada) at the time of planting, and the same dose 1 week after planting. High soil fertility was maintained throughout the experiment by adding 0.23 g of slow release Nutricote pellets (13 : 13 : 13, N : P : K; Vicksburg Chemical, Vicksburg, MS, USA) to the soil surface. Control plants received an equal volume of water in place of fertilizer. The experiment was conducted in a single greenhouse set to 25 °C during the day and 20 °C at night. Natural light was supplemented with 400 W pressure sodium lamps set to a 16 : 8 h (day : night) cycle; plants were watered as needed. We completely randomized all pots at the beginning of the experiment and spaced plants to reduce shading between pots. The experiment ended after plants’ roots filled the pot and growth stopped, which ranged from 6 to 9 weeks.

We determined the response of both Oenothera and Bromus to the factorial manipulation of competition and fertilizer by measuring six variables that depict plant performance: total plant biomass, above-ground biomass (shoot), below-ground biomass (root), shoot : root ratio, maximum plant growth rate and maximum plant size. The shoot : root ratio reflects the relative allocation plants make to their shoots vs. roots. Variation in this ratio need not be adaptive (Cahill 2003), but in conjunction with data on absolute allocation to shoots, roots, and total biomass, the shoot : root ratio can provide insight into how variation in allocation patterns within the plant influence neighbouring plant performance, as well as the response of a focal plant to competition. Biomass was determined by harvesting plants at the end of the experiment. We removed all soil from roots by gently shaking away loose soil followed by submersion in water where remaining soil was massaged free of the roots. Plants were separated into above-ground and below-ground portions and dried for 1 week at 60 °C and weighed to the nearest 0.001 g.

We measured plant size weekly as maximum rosette diameter (Oenothera) or maximum stem height (Bromus). These measures accurately predicted total plant biomass after 8 weeks of growth in both Oenothera (r = 0.89, P < 0.001, N = 28) and Bromus (r = 0.72, P < 0.001, N = 32). These data were used to estimate maximum plant growth rate and maximum plant size by fitting the weekly measures of plant size from each plant to the logistic growth equation in Proc NLIN of sas 9.1 (SAS Institute, Cary, NC, USA), which typically explained over 95% of the variation in the growth of individual plants from both species. The logistic growth equation was depicted as Sizet = Sizeeq/(1 + ze−gt), where Sizet represents plant size at time t, g represents the maximum rate of plant growth, and z is a plant-specific constant. Sizeeq is the asymptotic size that individual plants reached under our experimental conditions. Biologically, growth rate depicts the maximum rate (cm day−1) of growth experienced by a plant during its development, while maximum plant size estimates the maximum total biomass achieved by individual plants and also the potential competitive effect that a plant can have on its neighbours via shading.

We used Analysis of Variance (anova) to assess the effects of interspecific competition, fertilizer and their interaction, on each of the six variables described above for Oenothera and Bromus, separately. Analyses were performed using Proc glm in sas. A square-root or log-transformation was used to improve normality and reduce heteroscedasticity when needed.

Greenhouse experiment 2

The objectives of this experiment were to determine whether Oenothera exhibited genetic variation for traits that could influence interactions with neighbouring plant species, and whether variation among Oenothera genotypes affected the performance of Bromus. We used 22 genetic families (hereafter genotypes) of Oenothera, which were grown in a common environment for one generation to reduce maternal effects. Seeds were collected from these plants and used in the present experiment.

The experiment consisted of three treatments: (i) Oenothera grown alone, (ii) Bromus grown alone, and (iii) Oenothera and Bromus grown together. Each Oenothera genotype was replicated with 20 plants except one genotype with poor germination which had nine plants. Eight plants were randomly assigned to compete with Bromus and the remaining plants were grown in the absence of competition. Forty-two Bromus plants were also grown in the absence of competition. In total, there were 472 replicate pots containing 430 Oenothera and 210 Bromus plants.

To examine whether Oenothera genotype, competition with Bromus, and their interaction, affected the performance of Oenothera, we used restricted maximum likelihood (REML) in Proc Mixed of sas. The significance of random effects (genotype and genotype X competition) was assessed using the log-likelihood ratio test (Littell et al. 1996). The significance of the fixed factor (competition) was calculated using an F-test where d.f. were adjusted according to Kenward & Roger (1997). Broad-sense heritabilities were calculated as H2 = Vg/VT, where Vg and VT are the genetic and total components of variance, respectively. This equation is appropriate for O. biennis because it produces clonal progeny (Lynch & Walsh 1998). P-values for heritabilities were determined as the significance of plant genotype using untransformed data with the genotype X competition interaction excluded from the model. We assessed the effect of Oenothera genotype on Bromus performance using Proc Mixed, where only the effect of genotype was included in the model. One-way anova was used to test how the presence/absence of Oenothera (irrespective of genotype) influenced Bromus performance.

We employed stepwise multiple regression to determine how genetic variation in the six traits measured from Oenothera predicted variation in the response of Bromus. To do this, we calculated the best-linear unbiased predictors (BLUPs; similar to genotype means) of trait values for each Oenothera genotype (Littell et al. 1996). We also calculated BLUPs for Bromus traits according to the Oenothera genotype against which the grass competed. Thus, there were 22 values for each Bromus and Oenothera trait, one for each Oenothera genotype. We then regressed each Bromus trait individually against all Oenothera traits; we did not include total Oenothera biomass into the model because it was highly correlated with above-ground and below-ground biomass. We then chose the best fitting model using stepwise regression with an entry/exit value of P = 0.15. All analyses were performed in Proc REG of SAS.

Greenhouse experiment 3

This experiment evaluated the relative importance of plant genotype, soil fertility, and their interaction on Bromus performance. All plants were grown in competition with a heterospecific neighbour and we used 21 of the 22 Oenothera genotypes from experiment 2. Each genotype was replicated with 14–16 plants, divided equally into the two fertilizer treatments. In total, there were 334 pots containing 668 plants. We again used reml in Proc Mixed for statistical analyses. The model was a fully factorial design, involving fertilizer as a fixed effect, plant genotype and genotype X fertilizer as random effects.

field experiment

We conducted a field experiment at Jokers Hill during summer-2004 to investigate whether our results from greenhouse experiments translate to community-wide impacts on the performance and diversity of plants under more natural conditions. Our objective was to simulate the conditions plants experience when recruiting to freshly disturbed fields, which is the primary way in which Oenothera establishes new populations (Gross & Werner 1982), and to measure the effects of Oenothera genotype on the plant community that naturally establishes itself following disturbance. To do this we replicated the same experiment in two old fields (hereafter ‘gardens’) with contrasting productivity and plant cover. The ‘Forest’ garden was situated on a small hill where dry soil and gaps in the vegetation were common. The ‘North 30’ garden was an old field with dense vegetation and high productivity. At each garden, we ploughed a 700 m2 area and delineated to two equal-sized blocks to account for environmental gradients.

This experiment used a larger number of genotypes (29), where thirteen of the genotypes overlapped with the greenhouse experiments. In early May, we germinated seeds and transplanted seedlings into 250 mL pots containing soil collected from the garden into which a given plant was later to be transplanted. Plants were then placed into outdoor meshed tents free of insects for 2 weeks, and during this time they were watered every second day but received no fertilizer. Plants were subsequently transferred to the field, removed from their pots and planted directly into the soil. All plants were planted into rows and columns separated by 1.5 m. In total, there were 638 Oenothera plants with 22 replicates per genotype, randomized and equally replicated within each block and garden. Plants received no additional water in the field and plants that died were not included in analyses.

To mimic natural processes of community establishment following disturbance, we allowed plants to recruit into the experimental fields throughout the summer. Plants rapidly colonized the fields and at the time of harvest, 19 species (5 native and 14 exotic) were growing in close proximity with Oenothera. We harvested both gardens in late-August by removing the above-ground portion of Oenothera plants; heavy soil and deep taproots precluded harvesting roots. We also collected above-ground portions of all plants within a 10 × 10 cm quadrat immediately adjacent to, but not including, the base of the focal Oenothera plant (‘near quadrat’). A second 10 × 10 cm quadrat sample of the plant community was taken 30 cm away from the center of the focal Oenothera plant (‘far quadrat’); this quadrat was beyond the longest leaves of the focal plant. The location of quadrats was standardized by always sampling from the East side of focal plants. The strong competitive effects detected in this experiment (see Results) indicate that the size of quadrats was adequate to detect ecologically significant variation. Plants taken from each quadrat were sorted to species, dried at 60 °C and weighed to the nearest 0.1 g; plants were larger in this experiment and fine-scale measurements of biomass were not needed.

To assess the community-wide performance and diversity of plants growing close to focal Oenothera plants, we calculated total biomass, monocot biomass, dicot biomass, species richness, and the Shannon Index of diversity within each quadrat. Although many plants recruited to the gardens, no one species was common enough to measure individual responses.

We compared the vegetation between near and far quadrats to assess how proximity with Oenothera affected the performance and diversity of neighbouring plant species. This comparison reflects the local effects of competition by Oenothera because this species forms a taproot, so that most shading and below-ground competitive effects occurred within the limits of the outermost leaves of Oenothera and were likely negligible at a distance of 30 cm.

Statistical Analysis

We used reml in Proc mixed for all analyses. Oenothera biomass was analysed with garden as a fixed effect, plant genotype, block (garden), genotypeXgarden, and genotypeXblock (garden) as random effects. Analyses of plant community variables used a repeated measures design, where the full model was: Community variable = meanoverall + garden + quadrat + gardenXquadrat + genotype + block(garden)+ genotypeXgarden + genotypeXquadrat + genotypeXquadratX garden + error. Samples taken from the paired near/far quadrats were given unique identifiers and treated as repeated subjects in analyses. Garden, quadrat and their interaction were fixed factors and all other factors were random effects. We deleted non-significant interactions and blocking factors using a backward selection procedure.


greenhouse experiment 1

Overall, competition was asymmetric as Oenothera suppressed total biomass of Bromus by 57% (F1,30 = 106.8, P < 0.001), while Bromus did not significantly reduce total biomass of Oenothera (F1,24 = 1.9, P = 0.18). Oenothera caused a reduction in nearly all measures of Bromus performance (except growth rate) and caused a significant decrease in the shoot : root ratio (F1,30 = 23.2, P < 0.001). Bromus caused a significant reduction in Oenothera below-ground biomass (F1,24 = 11.3, P = 0.003) and maximum plant size (F1,33 = 50.2, P < 0.001), which resulted in an increase in the shoot : root ratio (F1,24 = 7.0, P = 0.01).

Increased soil fertility led to a significant increase in all measures of plant biomass for both species (P < 0.001 for all biomass variables). Competition and fertilizer never interacted to affect significantly the performance of Oenothera (P ≥ 0.07 for all measures). In contrast, these factors interacted to affect significantly all measures of Bromus biomass (P < 0.01 for all measures), which was due to stronger competitive effects in the presence of fertilizer.

greenhouse experiment 2

Oenothera exhibited significant heritable variation for all measures of plant performance with broad-sense heritability values ranging from 0.04 (shoot : root ratio) to 0.35 (growth rate) (specific heritabilities available from MTJJ). Therefore, there was variation among Oenothera genotypes for traits that could influence the performance of competing plants. Plant genotype and competition also interacted to affect below-ground biomass and the shoot : root ratio of Oenothera plants (Fig. 1, Table S1 in Supplementary Material), indicating the presence of genetic variation in response to competition with Bromus. Bromus also caused a reduction in all measures of Oenothera biomass and maximum plant size, as well as an increase in the shoot : root ratio (Supplementary Table S1).

Figure 1.

Genetic variation in Oenothera in response to competition. (a) Below-ground biomass and (b) shoot : root ratio exhibited a significant genotype X competition interaction (see Table S1).

Plant genotype of Oenothera significantly affected several measures of Bromus performance (Table S2). This effect is clearly seen for maximum plant size (Fig. 2), where the maximum height obtained by Bromus varied 33% between the most competitive Oenothera genotype and the least competitive genotype. In the absence of competition, maximum plant size of Bromus was only 17% greater than plants that had competed against the least competitive Oenothera genotype, while maximum size was 55% greater in the absence of competition compared to plants competed against the most competitive genotype (Fig. 2). Below-ground biomass and the shoot : root ratio of Bromus were also significantly affected by the identity of competing genotypes (Supplementary Table S2), as these variables varied by as much as 41% and 70% among Oenothera genotypes, respectively. Total biomass of Bromus showed a nearly significant response to plant genotype (P = 0.06; Supplementary Table S2).

Figure 2.

The effect of Oenothera genotype on the growth of Bromus plants. The figure shows the mean height and best fitting growth curve of Bromus in the absence of competitors (triangles fitted by solid curve), and in the presence of the most competitive (solid circles fitted by dash-dot curve) and least competitive (open circles fitted by dotted curve) Oenothera genotypes. Growth curves of Bromus competing against other plant genotypes fell in between these extremes. Maximum growth rate and the maximum size of plants were estimated by fitting the logistic growth equation (best fitting curve shown) to the height of each Bromus plant measured over 5 weeks.

Multiple regression showed that genetic variation in the shoot : root ratio of Oenothera explained 35–41% of the variation in Bromus performance. As the shoot : root ratio of Oenothera increased, Bromus plants allocated more biomass to roots (slope = 0.072 ± 0.018, P = 0.001, partial-R2 = 0.41), which led to an increase in total plant biomass (slope = 0.054 ± 0.016, P < 0.001, partial-R2 = 0.35) and a decrease in the shoot : root ratio (slope = –0.177 ± 0.050, P = 0.001, partial-R2 = 0.34) (Fig. 3). Although Oenothera genotype affected the maximum size of Bromus, none of the measured Oenothera traits were significant predictors, indicating that additional unmeasured traits also played a role.

Figure 3.

The response of Bromus performance to genetic variation in shoot : root ratio of Oenothera biomass. Oenothera genotypes with higher shoot : root ratios caused Bromus plants to increase total biomass, primarily by increasing allocation to roots, which led to a decrease in the shoot : root ratio of Bromus. Each point represents the genotype mean of shoot : root ratio for an Oenothera genotype (x-axis) and the mean response of Bromus to that genotype (y-axis). Variables were back-transformed to their original units.

greenhouse experiment 3

As observed in experiment 2, Oenothera exhibited genetic variation for all components of performance, except shoot : root ratio (Supplementary Table S3). Broad-sense heritability ranged from 0.08 (below-ground biomass, no fertilizer) to 0.27 (maximum plant size, fertilizer), and the mean across all variables was 0.15 without fertilizer and 0.20 with fertilizer. Plant genotype and fertilizer interacted to affect total biomass, above-ground biomass and growth rate, indicating that Oenothera exhibited genetic variation in its response to soil fertility (Supplementary Table S3).

Surprisingly, there was no main effect of Oenothera genotype on Bromus performance and genotype did not interact with fertilizer to affect Bromus (Supplementary Table S4). When we analysed the influence of plant genotype for fertilized and unfertilized plants separately, we detected a significant effect of Oenothera genotype on the shoot : root ratio of Bromus in the presence of fertilizer (inline image = 3.8, P = 0.026). In contrast to the weak effect of plant genotype, soil fertility had a large effect on Bromus performance. Fertilized plants showed significant increases for all measures of biomass, shoot : root ratio and maximum plant size (Table S4).

field experiment

Oenothera exhibited marginally non-significant genetic variation for above-ground biomass (Genotype effect: North 30: inline image = 2.0, P = 0.079; Forest: inline image = 2.3, P = 0.065). Heritability values were lower in both the North 30 (H2 = 0.018) and Forest (H2 = 0.062) gardens compared to heritabilities measured in the greenhouse (Above-ground biomass: Experiment 2, H2 = 0.26; Experiment 3, inline image = 0.23, inline image = 0.26).

In contrast to the effects of Oenothera genotype in the greenhouse, genotype had no detectable effects on either the diversity or biomass of neighbouring plants in the field (Table S5). Proximity to Oenothera did suppress the species richness, Shannon diversity, and biomass of neighbouring plants, but these competitive effects varied between gardens (Fig. 4, Table S5). At the Forest garden, plant species richness and Shannon diversity were unaffected by proximity to Oenothera, while dicot and monocot biomass were reduced by 37% and 23%, respectively (Fig. 4). At the North 30 garden, the richness and Shannon diversity of plants were reduced by 17% and 18% by Oenothera, respectively (Fig. 4). Likewise, Oenothera reduced dicot biomass by 47%, while monocot biomass was unaffected by Oenothera (Fig. 4).

Figure 4.

The effect of experimental garden and competition on the diversity and biomass of plants competing with Oenothera in the field. (a) Plant species richness, (b) Shannon diversity index, (c) dicot biomass, and (d) monocot (grass) biomass, were measured from 10 × 10 cm quadrats placed immediately adjacent to every Oenothera plant (near) and 30 cm away from the centre of Oenothera plants (far), in each of two experimental gardens (Forest and North 30). Asterisks indicate whether the community response significantly differed (P < 0.05) between near and far quadrats at a given garden.


The major findings of this study are that: (i) genetic variation in Oenothera traits affected the performance of competitors in controlled growth experiments, but (ii) this variation does not appear to be biologically important for the community-wide response of plants to Oenothera genotype in more natural environments. Several results support these conclusions. First, greenhouse and field experiments showed that Oenothera negatively affected the growth, biomass and diversity of neighbouring plant species, indicating that Oenothera competes with neighbouring species (Figs 2, 4). Second, Oenothera exhibited genetic variation for multiple traits (Appendices A, C; Fig. 1) and variation in the shoot : root ratio among Oenothera genotypes explained up to 41% of the variation in Bromus performance (Fig. 3). Third, the effects of plant genotype on competitors were inconsistent between experiments and variation in soil fertility had the largest effect on Bromus performance (Supplementary Table S4). Finally, our field experiment showed that heritable variation in Oenothera biomass was greatly reduced from that of the greenhouse, and there was no longer an effect of genotype on the performance of neighbouring plants. Therefore, our results suggest that it may not be necessary to investigate the effects of Oenothera genotype on its competitors in order to understand the short-term ecological consequences of competitive interactions in this community.

genetic effects on the performance and co-existence of competitors

For genetic variation in one plant species to affect the performance of another plant species there must be genetic variation in traits that influence competitive interactions. Consistent with this, we detected genetic variation in Oenothera for multiple traits that may play a role in above-ground and below-ground competition (Appendices A, C). Genetic variation in one of these traits, shoot : root ratio, strongly predicted variation in below-ground biomass of Bromus (Fig. 3), which also resulted in effects on the shoot : root ratio and total biomass of Bromus. Thus, although Oenothera competitively suppressed most aspects of Bromus performance (Appendices B, D), and there was genetic variation in above-ground and below-ground growth of Oenothera (Table S1), the effects of Oenothera genotype on Bromus performance were largely mediated by genetic variation in below-ground competitive ability.

Our results contribute to several recent studies that suggest genetic variation in plant competitive ability may be an important factor affecting the abundance and co-existence of competing plant species (Taylor & Aarssen 1990; Booth & Grime 2003; Proffitt et al. 2005; Fridley et al. 2007; Lankau & Strauss 2007; Whitlock et al. 2007). As with our experiment, greenhouse or outdoor competition experiments performed in pots typically find evidence for an effect of plant genotype on the performance of neighbouring plants. These results on their own do not provide compelling evidence for the importance of plant genotype in affecting the performance of competitors in natural communities. In the field, Oenothera genotype did not significantly affect the community-wide biomass or diversity of co-occurring plants (Supplementary Table S5). Therefore conclusions derived from pot competition experiments might not translate to field experiments, where environmental variance is greater. This is not to say that Oenothera is not an important player in plant communities. On the contrary, Oenothera negatively affected the biomass and diversity of monocots and dicots that naturally colonized the gardens (Fig. 4).

It is possible that differences among Oenothera genotypes had undetected effects on the surrounding community in the field. A limitation of our field results was that we were unable to harvest the roots of plants, whereas our greenhouse results suggest that below-ground biomass are most strongly affected by genetic variation in plant traits (Fig. 3). It is also possible that Oenothera genotype influenced the performance of individual plant species in the field. However, any such effects were likely weak and inconsistent because the heritability of Oenothera biomass was much lower in the field and plant genotype typically explained zero variance in community-wide biomass and diversity (Table S5).

We found inconsistent effects of Oenothera genotype on Bromus performance between greenhouse experiments. This inconsistency was unexpected because our methods were identical between experiments in almost everyway, including the same clonal genotypes of Oenothera, seed stock of Bromus, soil mixture, pot sizes, greenhouse space and experimenters. Furthermore, Oenothera genotypes exhibited significant heritable variation for most plant traits in both experiments. A likely explanation for the discrepancy is that although genotype means of shoot : root ratio were positively correlated between experiments (r = 0.53, P = 0.01), the correlation co-efficient was much less than 1, indicating a genotype X experiment interaction. The main difference between experiments was time of year (fall vs. winter), which resulted in a different intensity and length of natural sunlight (despite light supplementation) and we believe this difference caused the interaction. Indeed, previous experiments demonstrated that genotype X environment interactions frequently affect the expression of multiple traits in Oenothera (Johnson & Agrawal 2005; Johnson 2007). Thus, our results indicate that the ecological effects of Oenothera genotype on competitors are inconsistent and plant genotype explains much less variation than environmental factors such as soil fertility.

evolution in competitive ability and the co-existence of species

Theoretical models predict that genetic variation in competitive ability facilitates evolution in response to selection by competitors, which can influence the probability of co-existence among species. For example, evolution can increase co-existence between two populations when they evolve to use different resources (Abrams 2006; Vellend 2006). By contrast, evolution can decrease the probability of co-existence when two populations utilize the same resource and one population evolves to persist on lower concentrations of the shared resource than its competitors (Levin 1971). For this to occur, populations must contain genetic variation in competitive ability. Consistent with this, we detected a significant genotype X competitor interaction that affected below-ground biomass and the shoot : root ratio of Oenothera (Fig. 1, Table S1), indicating the presence of genetic variation on which selection by competitors can act. Although the short-term effects of Oenothera genotype on neighbouring plants is expected to be negligible in nature, genetic variation within Oenothera populations may enable populations to adapt to competitors over multiple generations. Therefore, the ecological and evolutionary consequences of genetic variation in competitive ability may differ.

effects of plant genotype within and across trophic levels

This study expands on recent research that examines how genetic variation in traits of Oenothera shapes the ecological structure of multitrophic communities. We previously showed that genetic variation in Oenothera predicts variation in resistance to herbivory (McGuire & Johnson 2006), herbivore population dynamics (Johnson 2008), and the composition and diversity of over 100 arthropod species (Johnson & Agrawal 2005; Johnson & Agrawal 2007). In light of the present study, genetic variation in Oenothera appears to be more important in affecting species interactions between trophic-levels than within trophic-levels. The generality of this observation has not been explored, but the pattern is expected whenever herbivores impose weak and/or inconsistent selection on plant defense, and competitors impose strong selection towards an optimum, which erodes genetic variation in competitive ability.

when to expect effects of plant genotype on competitors

Based on the results from field studies, we propose that preliminary predictions are emerging about when genotype identity is expected to affect the performance, abundance and diversity of competing plants. In one study, genotype identity of a coastal grass affected the growth and fine-scale abundance of two other plant species (Proffitt et al. 2005). Unlike Oenothera, the coastal grass formed clonal stands and exhibited genetic variation for stand size and senescence within stands. This difference in life-history likely contributed to the contrasting effects of plant genotype between studies, because genetic variation in the growth, biomass or senescence of vegetatively reproducing species affects the total area occupied by individuals, and therefore potential competitive effects. Genetic variation in secondary metabolites may also affect the structure of plant communities via allelopathic effects on neighbouring plants (Lankau & Strauss 2007). Evidence for this prediction comes from observational studies that show the diversity and composition of monoterpenes in the leaf litter of pine trees correlate with changes in soil chemistry and plant community structure (Iason et al. 2005; Pakeman et al. 2006).

There are still too few studies to perform meta-analyses on the effects of genetic variation in plant traits on competitive interactions. In Oenothera, genetic variation is unlikely to have strong effects on short-term ecological dynamics with competitors, but this is clearly not the case for all systems. Additional experiments on a diversity of species and systems will help us understand when intraspecific variation will be important in affecting the performance and co-existence of species.


Authors thank B. Hall, A. Petrie and L. Vingilis-Jaremko for technical assistance with greenhouse experiments. J. Ancheta and J. Jensen assisted with field experiments. A. Agrawal, R. Laird, M. Vellend, J.C. Cahill and four anonymous reviewers provided constructive comments on earlier drafts of the paper. Our research is funded by the NSERC Canada (MTJJ, RD), Mountain Equipment Co-op (MTJJ) and the National Science Foundation (MDH).