Introduction

Plant-soil feedbacks (PSFs) have gained attention over the past 25 years as a potential mechanism of plant growth and coexistence (Bever 1994; van der Putten et al. 2013). Yet, most PSF research has been performed using plant monocultures in greenhouse conditions (Kulmatiski et al. 2008; Forero et al. 2019). Recent work suggests that these greenhouse experiments provide little insight into plant growth in field communities (Forero et al. 2019; Reinhart et al. 2021). There remains, therefore, a need to better understand the role of PSF in plant communities in the field (Kulmatiski and Kardol 2008; Lekberg et al. 2018).
One robust trait of plant communities that may be, at least in part, explained by PSF is that productivity tends to increase with diversity (Kulmatiski et al. 2012; Tilman et al., 2001; Weisser et al., 2017). It has long been thought that the positive diversity-productivity relationship can be explained because species extract resources in different times or places (i.e., niche partitioning or complementarity; Hector et al., 1999; Loreau & Hector, 2001; Tilman et al., 1997). This mechanism can explain both species coexistence and why more diverse communities are more productive, because they more fully exploit resource space (Barry et al., 2019; Loreau, Naeem, Inchausti, Bengtsson, Grime, et al., 2001). However, resource complementarity has been found to be insufficient to explain either the extent of, or variation in, diversity-productivity relationships (Barry et al., 2019; Hector et al., 1999; Schnitzer et al., 2011). For example, despite overall positive diversity effects, some species and communities often underyield in diversity-productivity experiments (Hector et al., 2002). As a result, there has been interest in discovering additional mechanisms behind diversity-productivity relationships (Eisenhauer, Reich, & Scheu, 2012; Loreau et al., 2001).
Selection effects and disease accumulation have been suggested as additional mechanisms underlying positive diversity-productivity relationships (Loreau & Hector 2001; Maron et al. 2011; Schnitzer et al. 2011). Selection effects occur if species with ‘selected’ traits disproportionately affect mixtures at the expense of other species (Fox et al., 2005, Loreau & Hector, 2001, Roscher et al. 2009). Disease accumulation can cause overyielding if species-specific diseases accumulate and suppress plant growth more in low-diversity communities than high-diversity communities (i.e., pathogen dilution; Maron et al., 2011; Mommer et al., 2018; Schnitzer et al., 2011). However, neither selection effects nor disease accumulation are likely to explain the wide range of overyielding and underyielding observed in diversity-productivity experiments (Hector et al. 2002, Kulmatiski et al., 2012).
PSFs have been suggested as a mechanism that can explain both underyielding and overyielding (Kulmatiski et al. 2012). PSF describes a process in which plants change soil conditions, which can then affect further plant growth (conspecific or heterospecific; Bever, 2003; Hamilton & Frank, 2001; Wardle et al., 2004). These effects are often attributed to soil microbial communities (Ehrenfeld et al., 2005, Ke & Wan, 2020, Reynolds et al. 2003), but they can also result from changes to soil chemistry (Ehrenfeld et al., 2005, Smith-Ramesh & Reynolds, 2017), soil structure (Kyle et al., 2007), and soil animals (Eisenhauer et al., 2012).
Disease accumulation is one component of PSF that results in negative PSFs and can be expected to cause overyielding (Maron et al., 2011; Mommer et al., 2018; Schnitzer et al., 2011). Conversely, symbiont accumulation is another component of PSF, potentially resulting in a positive PSF. For example, a plant that accumulates species-specific symbionts can be expected to benefit more from those symbionts in a dense monoculture than in a diverse community (Kulmatiski et al. 2012). The role of plant mutualists in soil has been reported to affect plant community performance (Latz et al., 2012; Wagg et al., 2011) and suggested to co-determine selection and complementarity effects (Eisenhauer, 2011; Eisenhauer, Reich, & Isbell, 2012). However, positive PSF can also occur when a species’ growth is suppressed by soils cultivated by a different species (e.g., allelopathy; van der Putten et al., 2016). In either case, species with positive PSFs can be expected to be more productive in monoculture than in polyculture (i.e., they underyield; Kulmatiski et al., 2012).
While conceptually appealing, the magnitude of PSF effects in plant communities remains poorly understood for several reasons. Across the literature, roughly two-thirds of plants create negative PSFs, and one-third create positive PSFs (Cortois et al., 2016; Kulmatiski et al., 2008; Lekberg et al., 2018; van der Putten et al., 2016). However, most PSF research has been performed in the greenhouse and greenhouse-derived PSFs have been found to be larger than and uncorrelated with field-derived PSFs (Forero et al., 2019; Kulmatiski et al., 2008; Schittko et al., 2016). Further, most PSF research has measured PSFs without explicitly testing the role of the PSFs in plant mixtures (Ke & Wan, 2020; Kulmatiski et al., 2012; van der Putten et al., 2013). As a result, it is not known if PSFs affect species coexistence or community productivity or if PSFs are overwhelmed by other factors related to plant growth such as competitive interactions, herbivory, or intrinsic growth rates (Heinze & Joshi, 2018; Kulmatiski et al., 2011; Lekberg et al., 2018; Reinhart et al. 20218).
The overarching goal of this study was to test the role of PSFs in the diversity-productivity relationship. We established paired PSF and diversity-productivity experiments with mesic grassland species in Jena, Germany. Working in this site allowed us to test PSF effects in current and pre-existing diversity-productivity experiments (Roscher et al., 2016). We report PSF values and their relationship to competitive ability, but the emphasis of this paper was to test PSF effects during plant community establishment (i.e., plant growth during diversity-productivity experiments). To do this, a suite of plant community growth models was parameterized with (PSF) or without (Null) PSF data, and model predictions were compared to plant biomass in two and three-year-old plant communities. Consistent with modeling and greenhouse experiments, we predicted that PSFs effects would improve Null model predictions of plant community productivity because a) PSFs would be predominantly negative and explain overyielding, and b) positive PSFs would occur and contribute to underyielding (Maron et al. 2011; Schnitzer et al. 2011; Kulmatiski et al. 2012).