Data from: Shift in relative importance of complementarity and selection drives different effects of community evenness on richness-invasibility relationships
Data files
Apr 11, 2025 version files 46.86 KB
-
README.md
16.32 KB
-
Supplementary_data-0311.xlsx
30.54 KB
Abstract
How biodiversity affects ecosystem functioning is context-dependent. Competition for different resources (e.g., light vs. soil water) may lead to differential effects of species richness and their relative abundance (evenness) on ecosystem functioning. We constructed experimental communities with different richness and evenness levels and let them be invaded by an exotic annual Conyza canadensis or not at both xeric and mesic sites. Community invasibility negatively correlated with species richness at both high and low evenness in the xeric site, but only at high evenness in the mesic site. In the xeric site, soil water competition determined community invasibility, and complementarity effects and changes in niche overlap between invader and native species were the main mechanism for diversity effect on invasibility. In contrast, in the mesic site, light competition determined community invasibility, and selection effects due to presence of competitively superior native species played key roles. Changing evenness altered height differences between competitively superior native species and invader, leading to the different richness-invasibility relationships. Therefore, aboveground vs. belowground resource competition drives different effects of community evenness on richness-invasibility relationships. Our study highlights the roles of resource competition and functional traits in regulating the effects of species diversity on community invasibility.
https://doi.org/10.5061/dryad.3ffbg79t7
Description of the data and file structure
Linhai experiment
In the mesic Linhai site, we used 16 common native species (electronic supplementary material, table S1) to construct experimental communities with five levels of species richness (1, 2, 4, 8 and 12 species) and two levels of evenness (low and high). In October 2010, we built 352 plots (1 m × 1 m × 1 m) aboveground with bricks in an abandoned farmland in Linhai, and each plot was filled with a loamy soil collected in the Yunfeng mountain area near Linhai (organic matter: 1.44 ± 0.19 g kg-1; total P: 0.13 ± 0.03 g kg-1; total N: 0.62 ± 0.17 g kg-1). We planted two plots with monocultures of each of the 16 species (totaling 32 plots) and four plots with each of the 20 mixtures containing different species combinations for each of the other four species richness levels (i.e., 2, 4, 8 and 12 species, totaling 320 plots). The species assigned to each mixture were chosen by a random draw from the 16 species. For each of the 2-, 4-, 8- and 12-species mixture, two plots were randomly selected and used to create two high-evenness communities (plots), and the other two were used to create two low-evenness communities (plots).
For the high-evenness communities, all species had the same abundance, and for the low-evenness communities one species was exclusively dominant as grassland communities in the mountain grasslands around Linhai are usually dominated by a single species. For the low-evenness communities, the dominant species was randomly chosen; the relative abundance of component species was 3:1 for the two-species mixtures, 8:2:1:1 for the four-species mixtures, 12:2:2:2:2:2:1:1 for the eight-species mixtures, and 12:2:1:1:1:1:1:1:1:1:1:1 for the 12-species mixtures. We used such low-evenness levels following the dominance of natural grassland communities in the mesic site of Linhai. In the low evenness communities, the evenness values were 0.795, 0.632, 0.589 and 0.478 for the two-, four-, eight- and 12-species mixtures, respectively. In the high-evenness communities, all evenness values were 1. The evenness values were calculated according to Alatalo (1981).
For the two plots with exactly the same species composition, one plot was randomly chosen for the invasion treatment (i.e., invaded by the exotic annual forb C. canadensis), and the other plot was used as the control (i.e., not invaded by C. canadensis). In December 2012, 100 seeds of C. canadensis were evenly sown in the central area (0.6 m × 0.6 m) of each plot for the invasion treatment. Five monocultures of C. canadensis were also established with 100 seeds evenly sown in the central area (0.6 m × 0.6 m) of five 1 m × 1 m plots without any native vegetation.
In October 2013, we harvested all living plants in each plot. The plants were sorted to species, dried at 80°C for 48 h and weighed. In this experiment, the plots were constructed aboveground with bricks, which made roots much easier to harvest. After removal of the bricks, most soils were carefully taken away, and plants (with roots) and the soils adhering to them were placed on gauze. After the soils were washed away, the aboveground parts and roots of each plant were carefully separated. Phytolacca americana, Macleaya cordata, A. bidentata and R. japonica had deep taproot but little fibrous roots, and other species had mainly shallow fibrous roots (< 20 cm deep). These differences made it possible to sort the roots into species. Some fine roots were lost during cleaning. However, dry mass of these fine roots represented only 1.2–2.9% of dry mass of all roots in a plot (data were calculated from 40 randomly selected plots during biomass harvest).
Shenmu experiment
In the xeric Shenmu site, we used eight common native species (electronic supplementary material, table S2) to construct experimental communities with four levels of species richness (1, 2, 4 and 8 species) and two levels of evenness (low and high). In August 2010, we established 68 plots (1 m × 2 m) in an abandoned farmland after ploughing and harrowing it three times. The soil in the region is mainly composed of silt loam and sandy loam (Zhang et al., 2018). Eight species were planted in monoculture in one of the plots (totaling 8 plots), and 10 mixtures containing different species combinations were each established in two plots for each of the other three species richness levels (i.e., 2, 4 and 8 species, totaling 60 plots). The species assigned to each of the two- and four-species mixtures were chosen by a random draw from the eight species. For each mixture, we created a high-evenness community (plot) by assigning equal relative abundance to all species, and a low-evenness community (plot) by randomly assigned relative abundance levels to the component species (3:1 for two-species mixtures, 48:5:5:6 for four-species mixtures, and 48:3:3:2:2:2:2:2 for eight-species mixtures). We used such low-evenness levels as natural grassland communities in the xeric site of Shenmu are usually dominated by a single species with a relative density of 41.2% to 70.2% based on the survey of natural grassland communities in June 2010. Consequently, in the low-evenness treatments, the evenness values were assigned to 0.795, 0.546, and 0.407 for the two-, four- and eight- mixtures, respectively. In the high-evenness treatments, all evenness values were 1. The plots were separated by 2-m-wide walkways.
Seeds of the eight species were collected in the natural grassland communities around Shenmu. In September 2010, seeds were sown within an area of 20 m × 20 m (each species with 5 m × 8 m) in a farmland near the experimental site. In May 2011, seedlings were transplanted into the 1 m × 2 m plots. Each plot was planted with 64 seedlings, and the density was similar to the density (30–40 plants/m2) of the natural grassland communities in Shenmu. In each plot, seedlings of the same species were not placed next to each other, if possible, and the 64 seedlings were evenly distributed. Ten days after transplantation, we checked the status of each seedling and dead seedlings were replaced. The plots were regularly weeded.
In November 2012, 100 seeds of the invasive species C. canadensis were evenly sown in the central area (0.6 m × 0.6 m) of one half (1 m × 1 m) of each plot (thereafter referred to as the invasion plot). The other half (1 m × 1 m) of each plot was not sown with seeds of the invader and used as the control. Five monocultures of C. canadensis were also established in 1 m × 1 m plots, with 100 seeds evenly sown in the central areas (0.6 m × 0.6 m).
In September 2013, biomass of plants in each control and invasion plot (1 m × 1 m) was harvested. All aboveground and belowground parts of living plants were sorted to species, dried to constant mass at 80°C and weighed. The soil in Shenmu is very porous (Zhang et al., 2018). This allowed us to easily dig out the whole plants with roots (0 ~ 40 cm). Plants with some soils in each plot were placed on gauze. After soils were washed away, the roots of each plant were carefully separated. Some fine roots were lost during cleaning. However, dry mass of these fine roots represented only 2.5 – 3.3% of total dry mass of all roots in a plot (data were calculated from 20 randomly selected plots).
Files and variables
File: Supplementary_data.xlsx
Description:
Linhai experiment
In the mesic Linhai site, we used 16 common native species (electronic supplementary material, table S1) to construct experimental communities with five levels of species richness (1, 2, 4, 8 and 12 species) and two levels of evenness (low and high). In October 2010, we built 352 plots (1 m × 1 m × 1 m) aboveground with bricks in an abandoned farmland in Linhai, and each plot was filled with a loamy soil collected in the Yunfeng mountain area near Linhai (organic matter: 1.44 ± 0.19 g kg-1; total P: 0.13 ± 0.03 g kg-1; total N: 0.62 ± 0.17 g kg-1). We planted two plots with monocultures of each of the 16 species (totaling 32 plots) and four plots with each of the 20 mixtures containing different species combinations for each of the other four species richness levels (i.e., 2, 4, 8 and 12 species, totaling 320 plots). The species assigned to each mixture were chosen by a random draw from the 16 species. For each of the 2-, 4-, 8- and 12-species mixture, two plots were randomly selected and used to create two high-evenness communities (plots), and the other two were used to create two low-evenness communities (plots).
For the high-evenness communities, all species had the same abundance, and for the low-evenness communities one species was exclusively dominant as grassland communities in the mountain grasslands around Linhai are usually dominated by a single species. For the low-evenness communities, the dominant species was randomly chosen; the relative abundance of component species was 3:1 for the two-species mixtures, 8:2:1:1 for the four-species mixtures, 12:2:2:2:2:2:1:1 for the eight-species mixtures, and 12:2:1:1:1:1:1:1:1:1:1:1 for the 12-species mixtures. We used such low-evenness levels following the dominance of natural grassland communities in the mesic site of Linhai. In the low evenness communities, the evenness values were 0.795, 0.632, 0.589 and 0.478 for the two-, four-, eight- and 12-species mixtures, respectively. In the high-evenness communities, all evenness values were 1. The evenness values were calculated according to Alatalo (1981).
For the two plots with exactly the same species composition, one plot was randomly chosen for the invasion treatment (i.e., invaded by the exotic annual forb C. canadensis), and the other plot was used as the control (i.e., not invaded by C. canadensis). In December 2012, 100 seeds of C. canadensis were evenly sown in the central area (0.6 m × 0.6 m) of each plot for the invasion treatment. Five monocultures of C. canadensis were also established with 100 seeds evenly sown in the central area (0.6 m × 0.6 m) of five 1 m × 1 m plots without any native vegetation.
In October 2013, we harvested all living plants in each plot. The plants were sorted to species, dried at 80°C for 48 h and weighed. In this experiment, the plots were constructed aboveground with bricks, which made roots much easier to harvest. After removal of the bricks, most soils were carefully taken away, and plants (with roots) and the soils adhering to them were placed on gauze. After the soils were washed away, the aboveground parts and roots of each plant were carefully separated. Phytolacca americana, Macleaya cordata, A. bidentata and R. japonica had deep taproot but little fibrous roots, and other species had mainly shallow fibrous roots (< 20 cm deep). These differences made it possible to sort the roots into species. Some fine roots were lost during cleaning. However, dry mass of these fine roots represented only 1.2–2.9% of dry mass of all roots in a plot (data were calculated from 40 randomly selected plots during biomass harvest).
Shenmu experiment
In the xeric Shenmu site, we used eight common native species (electronic supplementary material, table S2) to construct experimental communities with four levels of species richness (1, 2, 4 and 8 species) and two levels of evenness (low and high). In August 2010, we established 68 plots (1 m × 2 m) in an abandoned farmland after ploughing and harrowing it three times. The soil in the region is mainly composed of silt loam and sandy loam (Zhang et al., 2018). Eight species were planted in monoculture in one of the plots (totaling 8 plots), and 10 mixtures containing different species combinations were each established in two plots for each of the other three species richness levels (i.e., 2, 4 and 8 species, totaling 60 plots). The species assigned to each of the two- and four-species mixtures were chosen by a random draw from the eight species. For each mixture, we created a high-evenness community (plot) by assigning equal relative abundance to all species, and a low-evenness community (plot) by randomly assigned relative abundance levels to the component species (3:1 for two-species mixtures, 48:5:5:6 for four-species mixtures, and 48:3:3:2:2:2:2:2 for eight-species mixtures). We used such low-evenness levels as natural grassland communities in the xeric site of Shenmu are usually dominated by a single species with a relative density of 41.2% to 70.2% based on the survey of natural grassland communities in June 2010. Consequently, in the low-evenness treatments, the evenness values were assigned to 0.795, 0.546, and 0.407 for the two-, four- and eight- mixtures, respectively. In the high-evenness treatments, all evenness values were 1. The plots were separated by 2-m-wide walkways.
Seeds of the eight species were collected in the natural grassland communities around Shenmu. In September 2010, seeds were sown within an area of 20 m × 20 m (each species with 5 m × 8 m) in a farmland near the experimental site. In May 2011, seedlings were transplanted into the 1 m × 2 m plots. Each plot was planted with 64 seedlings, and the density was similar to the density (30–40 plants/m2) of the natural grassland communities in Shenmu. In each plot, seedlings of the same species were not placed next to each other, if possible, and the 64 seedlings were evenly distributed. Ten days after transplantation, we checked the status of each seedling and dead seedlings were replaced. The plots were regularly weeded.
In November 2012, 100 seeds of the invasive species C. canadensis were evenly sown in the central area (0.6 m × 0.6 m) of one half (1 m × 1 m) of each plot (thereafter referred to as the invasion plot). The other half (1 m × 1 m) of each plot was not sown with seeds of the invader and used as the control. Five monocultures of C. canadensis were also established in 1 m × 1 m plots, with 100 seeds evenly sown in the central areas (0.6 m × 0.6 m).
In September 2013, biomass of plants in each control and invasion plot (1 m × 1 m) was harvested. All aboveground and belowground parts of living plants were sorted to species, dried to constant mass at 80°C and weighed. The soil in Shenmu is very porous (Zhang et al., 2018). This allowed us to easily dig out the whole plants with roots (0 ~ 40 cm). Plants with some soils in each plot were placed on gauze. After soils were washed away, the roots of each plant were carefully separated. Some fine roots were lost during cleaning. However, dry mass of these fine roots represented only 2.5 – 3.3% of total dry mass of all roots in a plot (data were calculated from 20 randomly selected plots).
Variables
Linhai experiment (sheet 1)
1. Number of variables: 8
2. Number of cases/rows: 160
3. Variable List:
plot code: Plot name.
log richness: log species number in plot.
evenness: 1 low evenness; 2 high evenness.
log invader biomass: biomass invader in each plot.
complementarity effect: Complementarity effect among native species.
selection effect: Selection effect among native species.
CHD: the community-weighted mean height difference between the native communities and invader
FDQ: the functional diversity Q index of root traits。
4. Missing data codes:
N/A: not applicable
Shenmu experiment (sheet 2)
1. Number of variables: 9
2. Number of cases/rows: 60
3. Variable List:
plot code: Plot name
log richness: log species number in plot.
evenness: 1 low evenness; 2 high evenness.
log invader biomass: biomass invader in each plot.
complementarity effect: Complementarity effect among native species.
selection effect: Selection effect among native species.
CHD: the community-weighted mean height difference between the native communities and invader
FDQ: the functional diversity Q index of root traits
CRD: the community-weighted mean root difference between the native communities and invader
Study sites
We conducted two experiments in two sites with contrasting environmental conditions. One site was located in Linhai (28°09′N, 121°53′E), Zhejiang Province, in southeastern China, and the other site was located in Shenmu (38°83′N, 110°51′E), Shanxi Province, in northwestern China. Linhai has a subtropical monsoon climate with a mean annual temperature of 19.5°C and an annual mean rainfall of 1800 mm, and Shenmu has a semi-arid continental climate with a mean annual temperature of 8.9°C and an annual mean rainfall of 423 mm.
We carried out a survey of the natural grassland communities in both Linhai and Shenmu. We selected 10 plots of 1 m × 1 m in each site. In Linhai, a total of 34 plant species were recorded, and species richness in 1 m × 1 m plots was 10 ± 2 (mean ± SE). Of the 34 species, 16 were common and used for the Linhai experiment described below; Achyranthes bidentata, Macleaya cordata, and Reynoutria japonica were the predominant species (electronic supplementary material, table S1). Light interception efficiency [i.e., (light intensity above the canopy – light intensity at the ground level)/light intensity above the canopy)] was 0.702 ± 0.079, and soil water content was 22.1 ± 2.9%. In Shenmu, a total of 19 species were found and species richness in 1 m × 1 m plots was 7 ± 2 (mean ± SE). Of the 19 species, 8 were common and selected for the Shenmu experiment; Artemisia annua and Stipa capillata were the dominant species (electronic supplementary material, table S2). Light interception efficiency was 0.547 ± 0.048, and soil water content was 8.3 ± 1.7%.
Linhai experiment
At the Linhai site, we used the 16 common native species (electronic supplementary material, table S1) to construct experimental communities with five levels of species richness (1, 2, 4, 8 and 12 species) and two levels of evenness (low and high). In October 2010, we built 352 plots (raised beds; 1 m × 1 m × 1 m) in an abandoned farmland, and each plot was filled with a loamy soil collected in the Yunfeng mountain area near Linhai (organic matter: 1.44 ± 0.19 g kg-1; total P: 0.13 ± 0.03 g kg-1; total N: 0.62 ± 0.17 g kg-1). We planted two plots with monocultures of each of the 16 species (totaling 32 plots). For mixtures, species were chosen by a random draw from the 16-species pool. For each of the other four species richness levels (i.e., 2, 4, 8 and 12 species), 20 mixtures containing different species combinations were selected and each was planted in four plots (totaling 320 plots). Of the four plots, two were randomly selected and used to create two high-evenness communities (plots), and the other two were used to create two low-evenness communities (plots).
In the high-evenness communities, all species shared equal abundance. In the low-evenness communities, one species had great abundance, mirroring the dominance of a single species in the mountain grasslands near Linhai. The relative abundance of component species was 3:1 for the two-species mixtures, 8:2:1:1 for the four-species mixtures, 12:2:2:2:2:2:1:1 for the eight-species mixtures, and 12:2:1:1:1:1:1:1:1:1:1:1 for the 12-species mixtures. The dominant species was randomly chosen, but the two low-evenness plots of the mixture containing the same species had the same dominant species. We based low-evenness levels on patterns of evenness in natural grassland communities at the site. The evenness values were 0.795, 0.632, 0.589 and 0.478 for the two-, four-, eight- and 12-species mixtures, respectively. In the high-evenness communities, all evenness values were 1. The evenness values were calculated according to Alatalo (1981).
Seeds of the 16 native species were collected in Linhai. In November 2010, seeds were sown in containers (64 cm × 42 cm × 27 cm); in May 2011, seedlings were transplanted into the plots. Each plot was planted with 48 seedlings, and the density was similar to the natural density (40–60 plants/m2) of plant communities in the mountain grasslands around Linhai. In each plot, seedlings of the same species were not placed next to each other if possible, and the 48 seedlings were evenly distributed. We checked all seedlings ten days after transplantation, and dead seedlings were replaced. All plots were regularly weeded.
For the two plots with exactly the same species composition, one plot was randomly chosen for the invasion treatment (i.e., invaded by the exotic annual forb Conyza canadensis), and the other plot was used as the control (i.e., not invaded by C. canadensis). Seeds of C. canadensis were collected in Linhai. In December 2012, 100 seeds of C. canadensis were first mixed with 50 g sand and then the seed-sand mixture was evenly sown in the central area (0.6 m × 0.6 m) of each plot for the invasion treatment. Five monocultures of C. canadensis were also established with 100 seeds evenly sown in the central area (0.6 m × 0.6 m) of five 1 m × 1 m plots without any native vegetation.
In October 2013, we harvested all living plants in each plot. The plants were sorted to species, dried at 80°C for 48 h, and weighed. In this experiment, the plots were constructed aboveground with bricks, which made roots much easier to harvest. After removal of the bricks, most of the soil was carefully removed, and plants including roots and the soils adhering to them were placed on gauze. After the remaining soil was washed away, the aboveground parts and roots of each plant were carefully separated. Phytolacca americana, Macleaya cordata, Achyranthes bidentata and Reynoutria japonica had a deep taproot but small fibrous roots, and other species had mainly shallow fibrous roots (< 20 cm deep). These differences made it possible to sort the roots into species. Some fine roots were lost during cleaning. However, dry mass of these fine roots represented only 1.2–2.9% of dry mass of all roots in a plot (data were calculated from 40 randomly selected plots during biomass harvest).
Shenmu experiment
At the Shenmu site, we used eight common native species (electronic supplementary material, table S2) to construct experimental communities with four levels of species richness (1, 2, 4 and 8 species) and two levels of evenness (low and high). In August 2010, we established 68 plots (1 m × 2 m) in an abandoned farmland after ploughing and harrowing it three times. The soil in the region is mainly composed of silty loam and sandy loam (Zhang et al., 2018). Eight species were planted in monoculture in one of the plots (totaling 8 plots), and 10 mixtures containing different species combinations were each established in two plots for each of the other three species richness levels (i.e., 2, 4 and 8 species, totaling 60 plots). The species assigned to each of the two- and four-species mixtures were chosen by a random draw from the eight species. For each mixture, we created a high-evenness community (plot) by assigning equal relative abundance to all species, and a low-evenness community (plot) by randomly assigned relative abundance levels to the component species (3:1 for two-species mixtures, 48:5:5:6 for four-species mixtures, and 48:3:3:2:2:2:2:2 for eight-species mixtures). We used these levels to approximate those in natural grassland communities in the Shenmu site, which are usually dominated by a single species with a relative density of 41.2% to 70.2% based on a survey of natural grassland communities in June 2010. Evenness values were 0.795, 0.546, and 0.407 for the two-, four- and eight-species mixtures, respectively. In the high-evenness treatments, all evenness values were 1. The plots were separated by 2-m-wide walkways.
Seeds of the eight species were collected in the natural grassland communities around Shenmu. In September 2010, seeds were sown within an area of 20 m × 20 m (each species with 5 m × 8 m) in a farmland near the experimental site. In May 2011, seedlings were transplanted into the 1 m × 2 m plots. Each plot was planted with 64 seedlings, and the density was similar to the density (30–40 plants/m2) of natural grassland communities in Shenmu. In each plot, seedlings of the same species were not placed next to each other, if possible, and the 64 seedlings were evenly distributed. Ten days after transplantation, we checked the status of each seedling and dead seedlings were replaced. The plots were regularly weeded.
In November 2012, a seed-sand mixture containing 100 seeds of C. canadensis and 50 g sand was evenly sown in the central area (0.6 m × 0.6 m) of one half (1 m × 1 m) of each plot (thereafter referred to as the invasion plot). The other half (1 m × 1 m) of each plot was not sown with seeds of the invader and used as the control. Five monocultures of C. canadensis were also established in 1 m × 1 m plots, with 100 seeds evenly sown in the central areas (0.6 m × 0.6 m). Seeds of C. canadensis used here were collected in Linhai. The seeds of C. canadensis collected in the Linhai site showed no significant differences in germination rate and relative growth rate (from 10 days to 70 days after germination) when introduced to the Shenmu site. Additionally, the plant height of C. canadensis developed from seeds collected in the Linhai site showed no significant difference from that observed in natural communities at Shenmu (electronic supplementary material, figure S1).
In September 2013, plants in each control and invasion plot (1 m × 1 m) were harvested. All aboveground and belowground parts of living plants were sorted to species, dried to constant mass at 80°C and weighed. The soil in Shenmu is very porous (Zhang et al., 2018). This allowed us to easily dig out the whole plants with roots (0–40 cm). Plants with some soils in each plot were placed on gauze. After soils particles were washed away, the roots of each plant were carefully separated. Some fine roots were lost during cleaning. However, dry mass of these fine roots represented only 2.5–3.3% of total dry mass of all roots in a plot (data were calculated from 20 randomly selected plots).
Environmental factors and functional traits
Photosynthetically active radiation (PAR) was measured using a PAR ceptometer (GLZ-C, Zhejiang Top Instrument Co., Ltd, China). Three points were randomly selected in the central area (0.5 m × 0.5 m) of each control and invasion plot. We took measurements on cloud-free days (October 7–10, 2013, for the Linhai experiment and September 2–4, 2013, for the Shenmu experiment) between 11:00 and 14:00, when the sun was close to directly overhead. In each plot, PAR both above the community canopy and at the ground level were measured at each of three points. Light interception efficiency was calculated as (PAR above the canopy - PAR at the ground)/ PAR above the canopy.
In each plot, three soil cores (6.4 cm in diameter and 20 cm in depth) were sampled in the central area (0.5 m × 0.5 m) and thoroughly mixed as one composite sample. Soil water content was determined with the gravimetric method after drying soils at 105°C for 24 h. In each plot, soil water content was measured three times, on May 16, August 11 and October 9, 2013 at Linhai and on May 21, July 18 and September 4, 2013 at Shenmu. The mean values of the three measurements were used as the measure of water content. Soil samples from the third date at each site were used to measure total nitrogen concentration. For each sample, a subsample was air-dried for two weeks, sieved through 2 mm mesh, ground into powder, and run through as autoanalyzer (Autoanalyzer 3, BRAN+LUEBBE, Germany).
We randomly selected 10 plants from the monocultures of each species at each site, and measured their height, root length, root volume, root surface area and average root diameter. Root traits were measured with WinRHIZO (Regent of Canada). Roots of each plant were dried to constant mass at 80°C and weighed.