DIVERGENT SELECTION FOR SECONDARY METABOLITES BETWEEN WILD POPULATIONS OF BRASSICA-OLERACEA AND ITS IMPLICATIONS FOR PLANT-HERBIVORE INTERACTIONS

Significant differences occur in the levels and types of aliphatic glu cosinolates in leaves of plants of four Brassica oleracea populations in Dorset. Plants in grassland at St Aldhelm's Head and Winspit have h igh levels of 3-butenyl glucosinolate, whereas plants of an adjacent p opulation growing on and along the top of cliffs at Kimmeridge have lo w levels of 2-hydroxy-3-butenyl, 2-propenyl and methylsulphinylalkyl g lucosinolates. Plants growing in a variable habitat at Worbarrow Tout have intermediate levels. The differences in occurrence of individual glucosinolates result from allelic variation at four loci. The level o f total aliphatic glucosinolates is under more complex genetic control , but is shown to be highly heritable. Allele frequencies at isozyme l oci indicate that genetic variation for glucosinolate production is un likely to have arisen or to be maintained by founder effects or geneti c drift. It is suggested that there is selection for high levels of bu tenyl glucosinolates at St Aldhelm's Head and Winspit because of grazi ng by generalist herbivores, whereas there is selection for low levels of 2-hydroxy-3-butenyl and other non-butenyl aliphatic glucosinolates at Kimmeridge because of two factors. First, plants effectively escap e from generalist herbivores because of physical aspects of the habita t and association with other plant species which provide physical and chemical defences. Thus there is selection for individuals which do no t carry the hypothetical metabolic costs of glucosinolate biosynthesis . Secondly, herbivory by specialist cruciferous insects at Kimmeridge, which is enhanced because of the local abundance of B. nigra, selects for individuals which have low levels of 2-hydroxy-3-butenyl glucosin olates.