Toward more robust plant-soil feedback research.

Understanding if and how plant-soil biota feedbacks (PSFs) shape plant communities has become a major research priority. In this paper, we draw on a recent, high-profile PSF study to illustrate that certain widely used experimental methods cannot reliably determine if PSFs occur. One problem involves gathering soil samples adjacent to multiple conditioning plants, mixing the samples and then growing phytometers in the mixtures to test for PSFs. This mixed soil approach does not establish that the conditioning plant being present caused the soil biota to be present, the first step of a PSF. Also, soil mixing approximates replacing raw data with averages prior to analysis, a move certain to generate falsely precise statistical estimates. False precision also results from sample sizes being artificially inflated when phytometers are misinterpreted as experimental units. Plant biomass ratios become another source of false precision when individual plant values contribute to multiple ratio observations. Any one of these common missteps can cause still living null hypotheses to be pronounced dead, and risks of this increase with numbers of missteps. If soil organisms truly structure plant communities, then null hypotheses indicating otherwise will not survive proper testing. We discuss conceptual, experimental and analytical refinements to facilitate accurate testing.