Ecological Archives E087-166-A1

Eric J. Baack, Nancy C. Emery, and Maureen L. Stanton. 2006. Ecological factors limiting the distribution of Gilia tricolor in a California grassland mosaic. Ecology 87:2736–2745.

Appendix A. Environmental correlates of core and edge habitat in five Gilia tricolor habitats.

During peak flowering in the spring of 2000, transects and sampling plots were established in five distinct G. tricolor populations to characterize variation in soil and vegetative conditions associated with population boundaries. Spot-checks in the spring of 2002 indicated that these boundaries were consistent between years. In each population we established 2-3 transects, spaced at 15 m intervals, that ran from central areas within the population to at least one boundary. At 5-m intervals along each transect we sampled 0.5 × 0.5 m plots (N = 136 plots across the five populations). Each plot was characterized as “near-edge” or “core,” based on whether it was less than 5 m from the nearest G. tricolor population boundary.

Soil and vegetative characteristics were examined in a subset of plots. We collected surface soil samples (to a depth of approximately 5 cm) which were subsequently dried, sieved and sent to the UC Davis DANR Analytical Laboratory for nutrient analysis. Extractable phosphorous and nitrate were measured in ppm, and potassium, calcium, and magnesium were measured in meq per 100 g of soil. In these same sampling plots, we counted the number of plant species native to California (“native plant richness”) and the number of non-native plant species (“alien plant richness”). Many soil and vegetative variables were highly inter-correlated, so we used principal components analysis (PROC PRINCOMP; SAS Institute, 2000) to summarize variation among sampled plots, and used the first three principal components in subsequent statistical analyses. Mixed model analysis of variance was used to determine whether principal components for soil and vegetation variables differed significantly between near-edge sites and core sites. Edge status was a fixed factor in the model, whereas population and the population-by-edge interaction were considered random (PROC MIXED; SAS Institute, 2000).

We characterized spatial structure in plant performance by examining above-ground plant biomass and seed production as a function of distance from the population boundary in each of the 5 populations. We randomly selected four G. tricolor plants from each of 120 plots. Plot means for above-ground dry biomass and seed number were analyzed to test for differences in average plant size between near-edge sites and core plot locations.


Soil attributes and vegetation characteristics show consistent patterns of covariation within our five G. tricolor study populations. Principal components analysis yielded three important components that, taken together, account for 76.3% of the variation in soil chemistry and vegetation variables among sample locations (Table 1). Calcium, phosphorous, potassium, and the Ca:Mg ratio all load heavily onto the first principal component (PC1), whereas soil moisture, magnesium, and nitrate load most strongly onto the second principal component (PC2). Higher scores for the third principal component (PC3) indicate locations with greater numbers of plant species native to California. The number of non-native plant species per plot increases significantly with Ca:Mg ratio (Pearson’s ρ = 0.327; df = 44; P = 0.0219) and shows non-significant, positive associations with other variables indicative of more fertile soils (N, P, and K).

Soil factors associated with greater numbers of non-native plant species are negatively associated with the performance of G. tricolor. Across all populations, the average dry weight of individual G. tricolor plants is negatively correlated with both the Ca:Mg ratio (Pearson’s ρ = -0.282; df = 44; P = 0.0602) and soil potassium (Pearson’s ρ = -0.340; df = 44; P = 0.0224).

Mixed model analysis of principal components revealed that several soil and vegetation characteristics change within 5 m of G. tricolor population boundaries. The mean PC1 score is significantly greater in near-edge samples (least squared mean ± SE for PC1: 0.83 ± 0.90) compared to more central samples (-0.37 ± 0.87; F1,4= 9.92; P = 0.0345), indicating that edge sites tend to be less severely serpentine (i.e., have a higher Ca:Mg ratio) than core sites. In contrast, the principal component associated with greater numbers of plant species native to California (PC3) tends to be greater in core samples (least squared mean ± SE for PC3: 0.18 ± 0.50 vs. - 0.24 ± 0.51; F1,4= 5.90; P = 0.0720). Site scores for PC2 do not change significantly near G. tricolor population boundaries (P = 0.3283).

We found negligible changes in G. tricolor density and per-plant performance between core sample locations and those close to population boundaries. Mixed model analysis across all five study populations demonstrated that average biomass per G. tricolor plant does not differ significantly between core and near-edge sites (P = 0.1664). Similarly, in the Central population, G. tricolor density does not show a statistically significant change within 5 m of the population boundary (P = 0.4942).

TABLE A1. The three most important principal components describing site-to-site variation in soil and vegetation characteristics within five populations of G. tricolor. Analysis of ten site variables yielded 9 principal components, of which only these three (PC1 – PC3) explained greater-than-average variance in site characteristics. Loadings of each variable onto the three principal components are shown, as is the portion of overall site-to-site variance explained by each principal component. The calcium:magnesium ratio is included, since this is often used as an indicator of the degree to which soils are serpentinized.





Percentage of site variation explained




Site variables:


Calcium:Magnesium ratio




Calcium (meq /100 g dry soil)




Magnesium (meq / 100 g dry soil)




Soil moisture (proportion of fresh weight)




Nitrate (NO3 , ppm)




Phosphorous (meq / 100 g dry soil)




Potassium (meq / 100 g dry soil)




Native species richness




Alien species richness




[Back to E087-166]