PREDICTING THE AUSTRALIAN WEED STATUS OF SOUTHERN AFRICAN PLANTS

A method of predicting weed status was developed for southern African plants naturalized in Australia, based upon information on extra-Austr alian weed status, distribution and taxonomy. Weed status in Australia was associated with being geographically widespread in southern Afric a, being found in a wide range of climates in southern Africa, being d escribed as a weed or targeted by herbicides in southern Africa, with early introduction and establishment in Australia, and with weediness in regions other than southern Africa. Multiple logistic regressions w ere used to identify the variables that best predicted weed status. Th e best fitting regressions were for weeds present for a long time in A ustralia (more than 140 years). They utilized three variables, namely weed status, climatic range in southern Africa and the existence of co ngeneric weeds in southern Africa. The highest level of variation expl ained (43%) was obtained for agricultural weeds using a single variabl e, weed status in southern Africa. Being recorded as a weed in Austral ia was related to climatic range and the existence of congeneric weeds in southern Africa (40% of variation explained). No variables were su itable predictors of non-agricultural (environmental) weeds. The regre ssions were used to predict future weed status of plants either not in troduced or recently arrived in Australia. Recently-arrived species wh ich were predicted to become weeds are Acacia karroo Hayne (Mimosaceae ), Arctotis venustra T. Norl. (Asteraceae), Sisymbrium thellungii O.E. Schulz (Brassicaceae) and Solanum retroflexum Dun. (Solanaceae). Twen ty species not yet arrived in Australia were predicted to have a high likelihood of becoming weeds. Analysis of the residuals of the regress ions indicated two long-established species which might prove to be go od targets for biological control: Mesembryanthemum crystallinum L. (A izoaceae) and Watsonia meriana (L.) Mill. (Iridaceae).