We sampled western columbine leaves (Aquilegia formosa) from four major meadow complexes (spatial clusters of meadows) within the H.J. Andrews Experimental Forest, Oregon, USA: two complexes along Frissel Ridge (M1 and M2), one on Carpenter Mountain (CM), and one on Lookout Mountain (LOM). Meadows within complexes were selected based on a two-step approach. First, we focused on meadows from which we had hummingbird RFID data (see project SA028) and those from which we had A. formosa seed set and/or pollen flow data. Because we aim to test the degree to which contemporary hummingbird movement predicts plant population genetic structure, we sought to sample plants in locations that should best reflect movement estimates. Divergence between the two is therefore less likely to be caused simply by sampling different meadows with different conditions, and more likely related to the biology of gene flow among plants on the landscape.
We then calculated meadow area (m2) and connectivity based on the amount of forest in a 100m-radius buffer around the meadow centroid (meadows with high proportions of forest cells surrounding them were assumed less connected). We stratified meadows into four categories: large-connected, large-isolated, small-connected, and small-isolated meadows based on natural but arbitrary breaks in the distributions of meadow size and connectivity (breaks were placed at 3.2 log10(m2) and 63% forest cover for meadow size and connectivity, respectively). Meadows from which we had complementary data (RFID or seed set data) were categorized into respective strata. We then filled the list in by randomly sampling within strata up to 32 meadows such that the final list included at least two meadows belonging to each stratum (large-connected, large-isolated, small-connected, small-isolated) within each of the four meadow complexes (M1, M2, CM, LOM).
