Many studies have attempted to identify traits that predict invasiveness in plants, but with limited success. Multiple traits are often correlated, suggesting that sets of traits need to be considered together. Although this observation is not new, it is rarely assessed quantitatively for invasive species. We used principle components analysis (PCA) to characterize twenty species using seven plant traits known to be associated with invasiveness. The resulting first PC axis was interpreted as an “invasiveness scale.” To gauge the utility of this scale for predicting invasiveness we then regressed axis 1 scores against independent assessments of invasiveness from published sources.
We recorded plant traits from twenty wetland plant species grown in a full-factorial cross of nitrogen fertility (0, 8, or 16 g N m-2) and insect herbivory (present or absent) across three experimental blocks. We used twelve native species (Bidens cernua, Bromus ciliatus, Calamagrostis canadensis, Carex scoparia, Juncus effusus, Leersia oryzoides, Mimulus ringens, Penthorum sedoides, Scirpus cyperinus, Spartina pectinata, Typha latifolia, Verbena hastata) and eight introduced species (Agrostis stolonifera, Briza minor, Echinochloa crus-galli, Lythrum salicaria, Mentha spicata, Myosotis scorpioides, Phalaris arundinacea, Poa trivialis). The traits were mean aboveground and belowground biomass, seed mass, percent germination, specific leaf area, herbivory response (mean biomass with herbivores/mean biomass without herbivores), and fertility response (mean biomass at high fertility/mean biomass at low fertility).
Results/Conclusions:
Axis 1 explained 31.6% of the variance extracted and was correlated with biomass, percent germination, herbivore response, and fertility response. Using linear regression, our invasiveness scale was significantly related to scores we assigned based on species accounts in two comprehensive weed references: 1) A Global Compendium of Weeds (Randall 2002; p=0.008) and 2) A Geographical Atlas of World Weeds (Holm et al. 1979; p=0.004). A logistic regression predicting inclusion in Invasive Plant Species of the World: A Reference Guide to Environmental Weeds (Ewald 2003) based on invasiveness score was marginally significant (p=0.058); given only twenty data points, this result is also compelling. We conclude that this simple approach can yield valuable insights by aligning potential invaders along a single multivariate scale. Since plant-trait databases are becoming more widely available, this approach could also have broad applications for predicting future invasive species.
