Online Resource 4. Methods and Results of the first seed addition experiment conducted on experimental grass plots.
Materials and methods
In October 2003 in California and March 2004 in Argentina half of the annual and perennial plots were treated with 200 locally collected pappus seeds of C. solstitialis, leaving the other half of the plots without seeds to assess the effects of the seedbank and dispersal and be able to measure light and soil moisture (see below) without the influence of the invader. In all cases, C. solstitialis seeds were added shortly before a rain, and in California, that rain was the first one of the season. Establishment of annual grasses in Argentina was poorer than in California (Table 1); for this reason, at the time of C. solstitialis invasion, plots with annuals in Argentinawere re-seeded using the same seed density as in the previous year (after Dukes 2001). After invading plots with C. solstitialis seeds, we measured photosynthetically active radiation (PAR) as described in the main text. In addition, we measuredvolumetric water content from 0.10 m to 1.20 m-deep in California and from 0.10 m to 1.10 m-deep in Argentinaas described in the main text in late spring (June and November), summer (July and January), and late summer (August and March). At the peak of C. solstitialis flowering, during the summer following the addition of seeds, we determined levels of invasion in plots as C. solstitialis proportional establishment (plants in the summer/seeds in the fall), dry aboveground biomass (mean of a maximum of five randomly selected individuals), and number of capitula (mean of the same individuals used for biomass).
Statistical analyses
Differences in proportional PAR and soil water dynamics between annual and perennial grass plots were assessed as described in the main text of this manuscript. Because no C. solstitialis established in the California plots and no differences in proportional establishment and size were detected between invaded and non-invaded annual and perennial plots in Argentina (p>0.05 for all cases; see Results), we used both C. solstitialis seeded and non-seeded plots in our soil moisture analyses, and thus we were able to work with a larger sample size (n=20 per functional type and region). Differences in proportional establishment, aboveground biomass, and number of capitula of C. solstitialis between annual and perennial experimental plots for a given region (Argentina only, see Results) were assessed with generalized linear models fit to a Binomial, Normal, and Poisson distribution, respectively.
Results
Experimental grass plots and seed additions
The proportion of PAR reaching the soil under constructed grass plots in California was very low and virtually identical for annual and perennial grasses. In contrast, proportional PAR under the canopy of annual grasses in Argentina was approximately four times higher than that under perennial grasses (Fig. 1; see Results in main text for statistical output). In contrast to general predictions, moisture in the soil profile was higher under perennial grasses than under annual grasses throughout the dry season in California (Fig. 2; Table 2 in the main text). More precisely, moisture was around 10% higher under perennial than annual grasses at 0.10 m to 0.75 m-deep, similar between functional types at 1 m, and became slightly but significantly higher under annuals at the greatest depth throughout the season (see Online Resource 8). In contrast to California, plant functional types exhibited remarkably similar soil water dynamics in central Argentina (Fig. 2).
In California, no C. solstitialis plant established in our experimental plots by the end of the dry 2003-04 season. Lack of recruitment was most likely due to a combination of competition from experimental grass communities and low rainfall, as 95% of seeds germinated after seven days in a trial conducted with a sample of 300 seeds. In Argentina, there were no differences in C. solstitialisestablishment and size between plots with annuals and those with perennials, but the invader was much more fecund in plots with annual grasses (Table 2).
Online Resource 4, Table 1. Percent cover of plants, litter, and bare ground at the time of invading experimental plots with C. solstitialis seeds in northern California and central Argentina. Values are means ± 1 SE of 20 plots per functional type in each region.
Region / Plot type / Plants / Litter / Bare groundNorthern California / Annual grasses / 92 ±2 / 8 ± 2 / 0
Perennial grasses / 95 ± 2 / 1 ± 0.4 / 5 ± 2
Central Argentina / Annual grasses / 31 ± 4 / 54 ± 3 / 15 ± 3
Perennial grasses / 86 ± 4 / 3 ± 0.5 / 12 ± 3
Online Resource 4, Table 2. Data and statistical results for dependent variables estimating invasion success of C. solstitialis in plots with annual versus perennial grasses in Argentina.
Community type / Variable / Mean ± SE / Statistics / pAnnuals / Proportional establishment / 0.0085 ± 0.004 / χ21=0.013 / 0.911
Perennials / 0.0045 ±0.003
Annuals / Aboveground biomass (g) / 28.084 ± 12.588 / χ21=1.555 / 0.212
Perennials / 4.938 ± 2.270
Annuals / Fecundity (no. capitula/plant) / 234.381 ± 145.901 / χ21=753.043 / <0.001
Perennials / 23.444 ± 10.989
Online Resource 4, Fig. 1. Proportional photosynthetic active radiation (PAR) during C. solstitialis seedling stage in plots with annual and perennial grasses in a parallel seed addition experiment in (a) northern California and (b) central Argentina. Circles are means ± 1 SE of 20 plots (except for perennials in California where two plots were disturbed by pocket gophers and excluded from analyses). Different letters indicate statistical differences (p<0.05) between treatments.
Online Resource 4, Fig. 2. Soil moisture dynamics in annual vs. perennial grass plots in a parallel seed addition experiment in (a) northern California and (b) central Argentina. Circles show means ± 1 SE of 20 plots; SE are within the diameter of the circles. Different letters indicate statistical differences (p<0.05) between treatments.