California grasslands are often considered “non-equilibrium” because the combination of climate variability and contingent propagule availability make species composition in any given year difficult to predict. Connectivity of seed sources over space and continuity of seed availability over time may therefore control spatial and temporal variability of species. Here we explore the consequences of connectivity for invasion dynamics in two California grasslands. In the first, we explore a field pattern in which Medusahead, a patch-forming invasive annual grass, and Avena, the dominant annual grass, appear to reciprocally invade and replace one another. We hypothesize that this pattern is due high heterospecific seed rain overcoming a resident competitive advantage; without high adjacent seed rain, each patch-type would be relatively stable. We use field data to parameterize competition models and simulate the degree of outside seed rain needed to generate a swap between patch-types. In the second, we explore a pattern in which Bromus, an annual grass, intermittently invades and recedes from a native-forb dominated serpentine grassland. Using spatially-explicit long-term data, we test the hypothesis that invasions are facilitated by ant mounds that provide refugia for Bromus during drought years, thereby increasing connectivity over time.
Results/Conclusions
Spatial and temporal variability in seed source strongly shaped the invasion dynamics in both grasslands. In the first grassland, population models indicated that Medusahead and Avena both experience strong conspecific density dependence and heterospecific facilitation. Without patchy and external seed rain, these dynamics led to stable coexistence among the species within a patch. However, when a locally rare species experienced high external conspecific seed input, its population increased rapidly due to heterospecific facilitation. When locally common, each species’ populations were rapidly curtailed by strong density dependence. Consequently, initially patchy landscapes with highly connected seed dispersal remained highly patchy, but the identity of the dominant species in any given patch varied over time. At the second site, we found that Bromus populations across the site were highly variable over time, but that invasions typically radiated from patches, generally on old ant mounds, in which Bromus abundance was more stable over time. Consequently, temporal connectivity in these patches facilitated continuous invasions, and the spatial distribution of these patches shaped the spatial patterning of invasions over time. We suggest that it is important to consider both connectivity in space and connectivity in time in invasion dynamics.