POPULATION-DYNAMIC CONSEQUENCES OF DIRECT AND INDIRECT INTERACTIONS INVOLVING A LARGE BLUE BUTTERFLY AND ITS PLANT AND RED ANT HOSTS

1. Caterpillars of the butterfly, Maculinea rebeli, feed sequentially in flower-buds of Gentiana cruciata and in nests of Myrmica schencki a nts, with which they have a sort of 'cuckoo bird' relationship. Ants f eed them in preference to their own larvae, reducing the production of new workers and hence the colony size the following year. 2. Other My rmica compete with M. schencki for nest-sites. They adopt caterpillars with equal facility, but fail to rear them to maturity and are conseq uently less damaged. 3. Using a mathematical model we explore how spec ies interactions can influence the populations of the butterfly, M. sc hencki, and other Myrmica species. The model assumes a 1 ha square of 900 cells with a fixed population of gentians. A cell can contain at m ost a single Myrmica colony which can adopt a proportion of the caterp illars leaving the gentians. Both density-dependent and density-indepe ndent butterfly mortalities occur on the plant and within the nest. 4. M. schencki is more successful in hotter, drier habitat patches than other Myrmica. This is simulated by a one-dimensional gradient in the maximum yearly reproductive rate (R) of M. schencki, the gradient for other Myrmica species being approximately reciprocal to that of M. sch encki. M. schencki and other Myrmica compete pre-emptively for vacant nest sites. 5. The model is seeded with nests and run for 50 generatio ns, to stabilize the ant distributions, before being 'colonized' by a single female of Maculinea rebeli. Addition of the butterfly to the sy stem reduces dramatically both the number and average size of M. schen cki nests. As a result, the number of nests of other Myrmica increases as they encroach onto the drier areas. In intermediate areas, where m ost pre-emptive competition occurs, the nest-size of other Myrmica inc reases, but overall, their average size also falls because of the dire ct effect of the caterpillars. 6. The system stabilizes at about 268 n ests of M. schencki, 73% being in cells containing G. cruciata and so able to adopt caterpillars and 27% being in plant-free cells, avoiding parasitization. Only 16% of cells are plant-free, demonstrating the n egative effect of the butterfly on the spatial distribution of M. sche ncki and its ability to mediate apparent competition between plants an d ants. 7. The model predicts that: (a) the largest populations of but terflies should occur in systems of intermediate plant density (c. 100 0-1500 plants ha-1); (b) the greatest turnover in ant nests should occ ur in dryer areas, where nests of M. schencki are relatively most dama ged by the butterfly, despite a pronounced impact on the ant populatio ns in the first years following colonization; (c) only c. 0.2% of nest s should go extinct each subsequent year as a consequence of butterfly parasitization; and (d) only about half of the potential rearing capa city of the M. schencki population should be exploited by the butterfl y in any one year. 8. We suggest that a subtle combination of changes in the environment can either benefit or endanger the butterfly popula tion depending on how each of the ant species are affected. 9. From a conservation stand-point, population size is not necessarily an accura te predictor of the fragility of the butterfly population; as long as the butterfly population is safe from stochastic mechanisms of local e xtinction, the intrinsic and resource-based components of the butterfl ies' basic reproductive rate are better indicators of persistence.