Pollinator behaviour, mate choice and the realised mating systems of Grevillea mucronulata and Grevillea sphacelata

Successful long-term conservation and management of populations of plants r equires successful management of the suite of factors that determine their 'realised' mating systems. Within the genus Grevillea, mating systems are p otentially complex. They may vary among species and among populations withi n a species, reflecting variation in pollinator behaviour and diversity ('p otential' mating systems) and in breeding system (the 'preferred' mating sy stem). We used a combination of pollinator observations, pollination experi ments and electrophoretic analysis of seed from open pollinations, to exami ne variation in potential and preferred mating systems in two populations o f each of two 'spider-flowered' Grevillea species: G. mucronulata (visited by honeyeaters and honeybees, although only birds effected pollen removal a nd transfer) and G. sphacelata (visited only by honeybees, which made frequ ent contact with pollen and stigmas). Almost all observed bird and insect f oraging bouts on either species involved movements among inflorescences wit hin plants or among inflorescences on closely neighbouring plants. On the b asis of these data, the mating system was predicted to involve a high level of selfing or inbreeding. However, the pollination experiment revealed tha t both species were highly self-incompatible and showed a clear preference for outcross pollen in mate-choice experiments. For both species, fruit set through autogamy was lower than 0.7% and fruit set from self-pollinations was always significantly lower than for outcross pollinations (0-11% cf. 25 -33% for G. mucronulata and 0% cf. 4.2-8.8% for G. sphacelata). Allozyme st udies revealed that genotypes in open-pollinated seeds on 20 G. mucronulata and 20 G. sphacelata plants were surprisingly uniform, best explained by o utcrossed matings between close neighbours. We found little between-populat ion variation in any aspect of the mating system for either species. These studies reveal that the preferred mating system of the plant and the patter n of pollinator behaviour interact to determine the mating system in a popu lation, emphasising the need for a multifaceted investigation of mating sys tems, especially in predicting the fates of populations that have pollinati on systems altered by disturbance, small size, isolation and introduced spe cies such as the honeybee.