Past and future of predictions in plant invasions: a field test by time

• Published in: Diversity & distributions Vol. 7; no. 3; pp. 145 - 151
• Main Author: Pysek, Petr
• Format: Journal Article
• Language: English
• Published: Oxford, UK Blackwell Science Ltd 01.05.2001; Blackwell Science; Blackwell
• Subjects: agricultural weeds; Animal, plant and microbial ecology; Applied ecology; Biodiversity Research; Biological and medical sciences; Biological invasions; Biological taxonomies; Central Europe; Conservation, protection and management of environment and wildlife; Czech Republic; Datasets; Ecological invasion; Flora; Fundamental and applied biological sciences. Psychology; Introduced species; Invasive species; Parks, reserves, wildlife conservation. Endangered species: population survey and restocking; Plant invasions; Plants; prediction; screening systems; Species diversity; Czech Republic; Europe; Czech Rep; Introduced species; Weed; Forecasting; Arable land; Invasion
• ISSN: 1366-9516; 1472-4642
• DOI: 10.1046/j.1366-9516.2001.00107.x

Intensive interest in the alien flora of the Czech Republic stimulated one of the earliest attempts (in the early 1970s) to predict potential invaders of arable land. Another paper published recently by the same research team using the same methodology provides a relatively unique possibility to assess the success of these predictions after a quarter of a century. The predictions were successful for 39.3% of the 28 species included, while 60.7% of invasion cases must be considered as failures. Prediction was rather unsuccessful for the members of Asteraceae (14.3%), and more correct for annuals/biennials (45.5%) than for perennials (16.7%). No pattern was found with respect to the area of origin. The results indicate that past predictions based largely on intuition were less successful than modern prediction systems using the knowledge of a large number of characters and carried out using advanced computation methods. The correct identification of invaders using such systems reaches values between 61% and 91%.