Predicting which taxa are likely to establish in the wild following introduction is a key goal of invasion biology, with most risk assessments carried out at the taxonomic level of species. This overlooks the potential for intraspecific variation in invasion risk, which could be particularly important in species where significant trait variation has developed, for example due to selective breeding or geographic isolation.
We quantified how variation in invasion risk was partitioned across taxonomic levels within the genus Brassica to test the hypothesis, inherent in weed risk assessments, that most variation in plant performance resides at the species level. We selected 24 commercial varieties of annual/biennial Brassica (3 species × 2 subspecies × 4 varieties - sourced from 4 global regions) and included the widely naturalised B. rapa var. oleifera. We added seeds of each species to 80 plots randomly located at sites representative of habitat where naturalised Brassica populations occur. We monitored subsequent seedling emergence and survival through to seed set, and estimated the finite rate of population increase, λ, which we used as an overall measure of invasion potential. We used general linear mixed models to examine how variation in plant performance was partitioned among taxonomic levels.
Results/Conclusions
An initial peak in germination was followed by low survival and a rapid decline in numbers of seedlings across all varieties and sites. Most individuals died at the dicotyledon or seedling stage; observation suggests mortality was due to competition, water stress, herbivory and damage from passing traffic and stock. Few varieties (including B. rapa var. oleifera) successfully set seed, suggesting only limited potential for establishment in the wild: only 2 varieties at 1 site had λ values greater than 1. Most taxonomic variation in performance (number of seedlings recruiting) occurred at the subspecific and varietal levels, with very little variation at the species level. Even greater variation occurred among sites; the site where naturalisation did occur was highly disturbed, moist and experience a low traffic volume - these factors rarely occurred in combination, suggesting that naturalisation in Brassica is highly context dependent.
Species level assessments of invasion risk appear insufficient to assess intraspecific variation in naturalization potential. This has considerable relevance for the importation of new germplasm comprising sub-specific varieties which may pose more of a risk than expected.