Discussion

Because most PSF research continues to be performed on plant monocultures in greenhouse conditions, the extent to which PSFs affect plant communities in the field remains unclear (Crawford et al., 2019; Forero et al., 2019; Ke & Wan, 2020; Reinhart et al. 2021). Our factorial experiment provided unusually comprehensive information about PSFs in the field. We measured all possible PSFs for nine species and found that plants, on average, created soils that changed subsequent plant growth by 36%. However, because plants realized both positive and negative PSFs, the net effect was that plants grew 14% less on home than away soils. While most PSF studies simply measure PSFs, we also tested the effect of these PSFs in plant communities. Despite causing 36% changes in plant biomass, PSFs had little effect on Null model predictions of plant community biomass across a range of species richness. While somewhat surprising, a lack of a PSF effect was appropriate in this site because species richness effects in this study were caused by selection effects and not complementarity effects (PSFs would appear as complementarity effects).
PSFs had little effect on Null model predictions for several reasons. First, even though the absolute value of PSFs was reasonably large, the net PSF effect was small because some PSFs were positive while others were negative. Second, PSFs for the two dominant plant species were small (-0.14 to 0.12). Third, because PSFs were, on average, smaller than differences in intrinsic growth rates (36% versus 193%), they were unlikely to change competitive outcomes between species (Kulmatiski 2016; Lekberg et al. 2018). Finally, A. elatius dominated across all species-richness levels so ‘away’ soils had little effect onA. elatius growth regardless of species richness. Broadly, our results demonstrated that large PSF values alone are not sufficient to explain plant species coexistence or the diversity productivity relationship at this site. In fact, overyielding at the site was caused primarily by selection effects, so complementarity effects of any kind (e.g., niche partitioning or PSF) were unimportant. Results do not exclude a role for PSF as a mechanism of species coexistence and productivity, particularly at other sites with larger complementarity effects, rather results highlight that PSF effects must be considered in the context of other factors affecting plant growth such as intrinsic growth rates (Crawford et al. 2019; Lekberg et al. 2019).

The Curious Case of Plant-Soil Feedbacks and the Dominant Species

We predicted that negative PSFs would cause overyielding because soil pathogens would be ‘diluted’ in diverse communities relative to monocultures (Kulmatiski et al., 2012; Maron et al., 2011; Schnitzer et al., 2011). However, A. elatius was such a dominant species that it maintained at least 75% relative biomass across all species-richness levels in the current diversity-productivity experiment. From a PSF perspective, an important consequence of this dominance is that A. elatius effectively only grew on ‘home’ soils. Therefore, A. elatius never benefited from pathogen dilution on ‘away’ soils. Research examining potential PSF effects ‘in vitro’ often assume that species are competitively equivalent (Bever et al 2003; Kulmatiski et al. 2011). Performing this experiment in field conditions helps refocus the role of PSF in the context of strong competitive imbalances among species which are common in field conditions (Crawford et al. 2019; Lekberg et al. 2018).

Are Neutral PSFs a Successful Strategy?

In addition to primarily growing on ‘self’ soils, the dominant speciesA. elatius realized a small PSF with little variability within or across soil treatments (Fig. 2). It is possible that small PSFs and small variability covary. It is reasonable to expect that, for a plant species to dominate in many communities, it will grow well across soil treatments and, therefore, demonstrate small and consistent PSFs. In contrast, plant species with large positive PSFs may have difficulty establishing in ‘away’ soils, while species with a large negative PSF may have difficulty attaining large growth on ‘home’ soils (Levine et al., 2006). Our results suggest that there may be a selective pressure to maintain neutral PSFs with low variability to dominate plant communities. Consistent with this idea, we found that competitive species were associated with small PSF values (Fig. 3) while sub-dominant species demonstrated large positive and large negative PSF. This perspective may help explain why PSFs often show weak correlations with landscape abundance (Reinhart et al. 2021, but see Mangan et al. 2010; Kulmatiski et al. 2017).
There is also a statistical reason that dominant species may demonstrate small PSFs. It is more likely that plant species with small growth will realize large proportional changes in growth (Pfisterer & Schmid, 2002). For example, a plant species that can grow to 50 g m-2 on ‘home’ soils can easily be imagined growing to 0 or 200 g m-2 on ‘away’ soils, resulting in PSFs of 1.0 and -0.75, respectively. However, it is essentially impossible for plant species to grow to 1,000 g m-2 on ‘home’ soils and 4,000 g m-2 on ‘away’ soils because 4,000 g m-2 is beyond carrying capacity in grasslands. As a result, subdominant species are more likely to have large PSFs than dominant species. We are not aware of other studies suggesting these ideas and this is likely because PSF experiments rarely perform the types of large factorial experiments needed to examine PSFs for many species across soil types (Rinella and Reinhart 2018).

Diversity-Productivity Relationships

Species richness effects were similar to other biodiversity experiments in more mesic sites (Cardinale et al., 2011; Hector et al., 1999). However, the mechanisms driving this response differed between the current and pre-existing experiments. In the pre-existing experiment, polyculture biomass was driven by selection (21% of monoculture biomass) and complementarity (14%) effects. In the current experiment, overyielding was largely explained by selection effects (43%) and countered by negative complementarity effects (-20%). A. elatiuswas more dominant in the current than the pre-existing experiment (Fig. 5; Clark et al. 2020). Community productivity in the Jena Experiment varies widely among years due to different environmental conditions (Weisser et al., 2017), so it is likely that climate or other environmental conditions that differed between the two studies also caused greater dominance effects in the current experiment (Marquard et al., 2009; Guimarães-Steinicke et al., 2019). For example, a large flooding event in 2013 may have increased A. elatius growth by increasing nutrient availability (Wright et al., 2015). A. elatius is strongly competitive for light and nitrogen, so greater seeding rates in the current experiment may have exaggerated asymmetric competitive effects (Lorentzen et al. 2008; Roscher et al. 2008). It is interesting to note, that even though the mechanisms differed, the net biodiversity effect was similar in the new and old experiments.

Species-Level vs. Soil-Level PSFs

Because sample sizes increase exponentially as species are added to factorial PSF-experiments, most studies measure PSFs for one to a few target species (Smith-Ramesh & Reynolds, 2017; Van der Putten et al., 2013). By measuring all 72 potential PSFs for nine species, this study provided unusually comprehensive insights into how PSFs vary among soil conditioned by different species. For the most part, PSFs were consistent among soil treatments. It is not unreasonable to expect PSFs to vary widely across differently conditioned soils (Bezemer et al., 2006; Rinella & Reinhart, 2018; Smith-Ramesh & Reynolds, 2017). For example, a plant species may grow well on a soil conditioned by a N-fixing species and poorly on a soil conditioned by an early-successional species that accumulated a large pool of generalist soil pathogens (Chapin et al., 1994; Van der Putten et al., 2013). However, we observed only one species that had a positive PSF on one soil treatment and a negative PSF on another soil treatment (P. pratense ). The fact that PSF values were consistent across soil treatments suggests that PSFs in this system are determined primarily by growth on ‘home’ soil.

Site Differences

It has been suggested that PSFs will intensify competitive effects in nutrient-rich conditions and strengthen facilitative effects in nutrient-poor conditions (Bever, 2003; Lekberg et al., 2018). Consistent with this idea, we found that PSFs were more negative, and competitive effects (selection effects) were larger in the current experiment, performed at a mesic, nutrient-rich site relative to a similar recent study performed at a drier and nutrient-poor site (Forero 2021). Both absolute (0.36 vs 0.27) and net (-0.14 vs. 0.10) PSFs were larger at the nutrient-rich vs. nutrient-poor site, respectively (Forero 2020). Further, overyielding was smaller at the nutrient-rich site than the nutrient-poor site (Craven et al. 2016; Forero 2021). Larger PSFs and competitive effects in nutrient-rich conditions provide a potential explanation for why the strength and trajectory of biodiversity-ecosystem functioning relationships over time differ between more and less fertile soils (Eisenhauer et al., 2019; Guerrero‐Ramírez et al., 2019; Ratcliffe et al., 2017).