Soil sampling
A visual vegetation shift marks the soil moisture transition from the wet area to the mesic-dry area further inland. This transition border falls close to plot 16 on transect 1 and plot 68 on transect 4 (red line in Fig. 1). Soil samples were collected on June 7th, 2016, from five locations 30 m apart on each side of the soil moisture transition border separated by 30 m across the transition border. Soil sampling locations intersected transect 1 and 4 between plots 14 and 67 in the mesic-dry and plots 17 and 69 in wet area (Fig. 1). At each location, two soil samples were taken using a soil corer when possible (5 cm diameter × 10 cm depth) or with a hand shovel when soils were too wet for using a soil corer (as was the case for most of the samples on the wet side). When using a shovel, sampling depth and soil volume was estimated to correspond to that of the soil cores. The first sample at each location was collected around a F. meleagris plant and the second at 0.5 m distance from the sampled F. meleagris plant. For the second sample (referred to as non-Fritillaria soil), we also ensured that no other F. meleagris plant was within 0.5 m of the sample. The corer/shovel was wiped with 70% ethanol-soaked tissue paper between each sampling. In total, 20 soil samples were collected: fiveFritillaria /non-Fritillaria soil sample pairs from the wet side and five pairs from the mesic-dry side. The most common plant species were recorded at each sampling location for cross reference to the more complete plant community data recorded for plots along the transects (Table S1).
All samples were individually placed in plastic bags and kept on ice during sampling before storage at 4°C overnight. The following day, soils were homogenized in the plastic bags and subsamples of soil were transferred to 15 ml conical centrifuge tubes and frozen at -20°C, followed by freeze drying. Another subsample was weighed before drying at +80°C for 48 h to estimate gravimetric soil moisture (Holliday, 1990) (Table S2).