Plant-soil feedbacks (PSFs) have been suggested to explain species coexistence and productivity, but support for the role of PSFs is largely derived from theoretical models and greenhouse experiments. To provide a more direct test of the role of PSF in plant communities, here we parameterize a suite of plant growth models with or without field-based PSF data from two sites in the USA and Germany. We compare model predictions to plant growth in separate diversity-productivity experiments. More specifically, 16 species at Cedar Creek, USA and nine species at Jena, Germany were grown in factorial, two-phase PSF experiments. At each site, plants were grown in a common garden for a two-year culturing phase, removed, then grown for a two-year response phase. In a separate diversity-productivity experiment, species from the PSF experiment were grown for three years in 1, 2, 4, 8, 9 or 16-species communities. To develop a prioritized list of soil organisms that were most likely to determine observed PSFs, archael, bacterial and fungal community compositions were correlated with PSFs.
Results/Conclusions
PSF values were predominantly negative, but varied from positive to negative at each site. PSF values helped explain plant community productivity both between and within sites. Between sites, PSFs were more negative at the drier, sandier Cedar Creek site than at Jena. Consistent with the presence of more negative PSFs, overyielding in the diversity-productivity experiment was greater at Cedar Creek than Jena. Diverse communities were roughly three and two times more productive than monocultures at Cedar Creek and Jena, respectively and PSFs could explain much of this effect. Within sites, species with the most negative PSFs tended to overyield the most in communities and species with the most positive PSFs tended to underyield the most in communities. PSF-informed plant growth models outperformed null models in which PSF effects were removed. Finally, by correlating plant growth responses with the microbial ‘species’ associated with each soil type, we were able to produce a prioritized list of soil organisms that were most likely to increase or decrease each plant species’ growth. Results provide field-based evidence that by predicting both overyielding and underyielding, PSFs can help explain how plants grow in communities in field conditions.