Study site description, sampling design and plant growth response measurement
During summer 2015 we conducted a field study at the University of Alberta’s Mattheis Ranch, located in southern Alberta and at the northern edge of the Great Plains (50°53′29.9″N 111°55′07.4″W). The vegetation was a semi‐arid mixed grassland dominated by Koeleria macrantha (Ledeb.) Schult, Hesperostipa comata (Trin. & Rupr.) Barkworth, Pascopyron smithii (Rydb.) Á Löve and Bromus inermis Leyss. (Shorthouse, 2010).
To evaluate plant-soil feedbacks in this system, we sampled 49 10-cm-diameter soil cores in 10 different plots, for a total of 490 soil samples. Details on the soil sampling methods can be found in Chagnonet al. (2018). These soils were used as living inocula in a greenhouse experiment, where Bromus or Koeleria were grown in a 9:1 mixture of sterile soil and living inoculum. The sterile soil consisted of a 1:1 mixture of a sandy soil collected at our study site, and fine sand. Using such a design, any growth response by Bromusor Koeleria was expected to be caused largely by soil biota (Brinkman et al. , 2010). After 15 weeks of growth, plants were harvested and total (shoot + root) dry mass was used as a measure of plant growth response to soil inocula (i.e., PSF). The remaining methods described below are new to this paper, and were not part of the original publication (Chagnon et al. , 2018).
To identify soil microorganisms potentially generating plant growth responses in our experiment, we combined the data generated from the previously described greenhouse experiment with targeted soil DNA sequencing. Of our 490 initial soil living inocula, we selected 230 inocula that would also be analyzed for microbial community structure, representing 23 samples in each of our 10 plots. The 23 inocula were chosen in a way that maximized their spatial dispersion in all four quadrants of the plot (Fig. S1). The field soil subsamples had been kept at -20C until DNA extraction, which was within 2 months after soil collection.