Host-plant availability drives the spatiotemporal dynamics of interacting metapopulations across a fragmented landscape

Abstract

The dynamics of ecological communities depend partly on species interactions within and among trophic levels. Experimental work has demonstrated the impact of species interactions on the species involved, but it remains unclear whether these effects can also be detected in long-term time series across heterogeneous landscapes. We analyzed a 19-year time series of patch occupancy by the Glanville fritillary butterflyMelitaea cinxia, its specialist parasitoid waspCotesia melitaearum, and the specialist fungal pathogenPodosphaera plantaginisinfectingPlantago lanceolata,a host plant of the Glanville fritillary. These species share a network of more than 4,000 habitat patches in the angstrom land islands, providing a metacommunity data set of unique spatial and temporal resolution. To assess the influence of interactions among the butterfly, parasitoid, and mildew on metacommunity dynamics, we modeled local colonization and extinction rates of each species while including or excluding the presence of potentially interacting species in the previous year as predictors. The metapopulation dynamics of all focal species varied both along a gradient in host plant abundance, and spatially as indicated by strong effects of local connectivity. Colonization and to a lesser extent extinction rates depended also on the presence of interacting species within patches. However, the directions of most effects differed from expectations based on previous experimental and modeling work, and the inferred influence of species interactions on observed metacommunity dynamics was limited. These results suggest that although local interactions among the butterfly, parasitoid, and mildew occur, their roles in metacommunity spatiotemporal dynamics are relatively weak. Instead, all species respond to variation in plant abundance, which may in turn fluctuate in response to variation in climate, land use, or other environmental factors.