Changes in the composition and behaviour of a pollinator guild with plant population size and the consequences for plant fecundity

Summary Small population size in plants is often associated with decreased fruit set through lower pollinator visitation and reduced offspring fitness as a result of inbreeding. Whether the potentially negative impacts of small population size are realized may be potentially influenced by the behaviour and ecology of pollinators. Here, we investigate changes in both guild composition and pollinator behaviour with plant population size in the bird-pollinated Anigozanthos flavidus (Haemodoraceae) and the consequences for fruit set and seed germinability. We used a germination stimulant to reduce the potentially confounding affects of interpopulation variation in dormancy when quantifying seed germinability. All populations were visited in small numbers by western spinebills, while the behaviourally dominant New Holland honeyeaters only visited large populations. Fruit set was extremely high (mean = 96·4 ± 0·6%) regardless of population size. The percentage of foraging bouts interrupted by aggressive interactions tended to be less in small plant populations. The percentage of interrupted foraging bouts was as high as 31% in some populations, suggesting that aggression may frequently promote pollen movement. Western spinebills tended to visit fewer stems per plant than New Holland honeyeaters, a behaviour that potentially favours greater outcrossing. There was no relationship between seed germinability or seeds produced per capsule and plant population size, suggesting that the foraging behaviour of western spinebills provides sufficient pollinator services in small populations to mitigate the potential reduction in seed production and germinability. Our findings highlight that, even within a pollinator guild, differences in behaviour can have important implications for the fitness of plants in small populations, with the potential for some pollinators to provide unexpected levels of resilience.