Biological control of Scotch broom: modelling the determinants of abundance and the potential impact of introduced insect herbivores.
Abstract
Simulation and analytical models were developed for the European shrub, Scotch broom Cytisus scoparius. The simulation model is spatially explicit and allows exploration not only of changes in population size but also the proportion of ground covered by the weed. The simulation model incorporates spatially local density-dependent competition, asymmetric competition between seedlings and established plants, a seed bank, local seed dispersal and an age-structured established plant population. This model is designed to incorporate much of the known population biology of broom. The analytical models are simple approximations of the simulation. The basic model contains nine parameters: the probability a site is disturbed, Pdist; the probability a seed becomes a seedling, g; the probability a seedling survives the first year, s; the probability a seed is lost from the seed bank, d; the minimum age for reproduction, Amin; maximum plant age, Amax; seed production per site, F; the probability a seed is retained in the parental site, fh; and the probability a site becomes suitable for colonization after broom senesces, Pso. Published data on the demography of broom is reviewed from studies around the world, and some previously unpublished data is presented. These data suggest that broom in some exotic habitats can achieve higher fecundities and live longer than in its native range. Analytical approximations provide a good description of the simulation results over a wide range of biologically reasonable parameter values. Specifically, the analytical models work well when plants are long-lived or highly fecund. Analysis of the models indicates that when broom colonizes all suitable sites with probability one, that the fraction of sites occupied by broom is determined by only three parameters: the probability of disturbance, Pdist; the probability a site becomes suitable for colonization following plant senescence, Pso; and maximum longevity, Amax. In exotic habitats, where individual broom plants can produce several thousand seeds, differences in these parameters are the most likely reason why broom populations are more weedy than in the native range. The impact of insect herbivores, which reduce plant fecundity, on broom abundance is explored for several environmental scenarios. This analysis suggests that potential biological control agents are most likely to have a substantial impact if the disturbance rate is high, plant fecundity is low and seedling survival is low. Even herbivores that reduce seed production by only 75% can have a dramatic impact on broom abundance, in contrast to several published predictions. Extensions to the models to allow for arbitrary patterns of age-dependent senescence, and site-specific probabilities of disturbance, are presented.