Introduction
Plants modify the soil they grow in, be it through taking up nutrients (Fireman & Hayward, 1952; Zinke, 1962), returning litter of varying chemical quality thereby impacting nutrient cycling (e.g., Wedin & Tilman, 1990; Hobbie, 1992), or accumulating specific soil organisms in their rhizosphere (Bever et al. , 2010). This, in turn, influences their success through plant-soil feedbacks (PSF) (van der Puttenet al. , 2013). In the 1980’s and the 1990’s, most attention has been paid to PSF driven by abiotic nutrient dynamics (e.g., Chapin, 1980, 1991; Tilman, 1980; Chapin et al. , 1986; Horner et al. , 1988; Kuiters, 1990). Correlations have been observed between nutrient use efficiency and litter chemical quality, such that plants from nutrient-poor environments tend to favor their own success (i.e., experience positive PSF) by producing a recalcitrant litter (Berendse, 1994; Bryant et al. , 1991), which slows down nutrient cycling and increase their competitive ability (Wedin & Tilman, 1990). However, seminal work showing implications for soil-borne disease in sand dune succession (van der Putten et al. , 1993) has prompted a shift of focus towards a role for pathogens in driving plant community structure (e.g., Dobson & Crawley, 1994; Thrall et al. , 1997). This still transpires in more recent views of PSF not being driven by litter chemistry, but by the accumulation of specific soil biota, emphasizing pathogens (e.g., Klironomos, 2002; Bauer et al. , 2015; Beveret al. , 2015; Albornoz et al. , 2017).
Plants accumulate a variety of organisms in their rhizosphere (e.g., bacteria, fungi, nematodes) ranging from harmful to beneficial. Because the bulk of measured PSF tend to be negative (Kulmatiski et al. , 2008), many have proposed PSF as a mechanism promoting plant coexistence through enemy accumulation by dominants (Olff et al. , 2000; Mangan et al. , 2010; Reinhart, 2012). As a result, a disproportionate amount of research has focused around pathogen accumulation in the rhizosphere (e.g., Mills & Bever, 1998; Manganet al. , 2010; Bagchi et al. , 2010,2014; Sarmiento et al. , 2017). Pathogens have been suggested to explain the underyielding of low-diversity plots in plant diversity experiments (Schnitzeret al. , 2011; de Kroon et al. , 2012; van Ruijven et al. , 2020), and the negative density dependence of plant growth in a variety of environments (e.g., old fields: Klironomos, 2002; shrublands: Laliberté et al. , 2015; temperate forests: Packer & Clay, 2000; tropical forests: Mangan et al. , 2010). However, negative feedbacks need not be caused by enemies (Bever et al. , 1997; Bever, 1999). Indeed, the accumulation of less efficient microbial mutualists (Bever, 2002) or of microbial competitors (Hodge et al. , 2000) can also make a plant grow better in heterospecific vs conspecific soil, leading to negative PSF even in the absence of pathogens. Likewise, positive PSF have often been ascribed to mutualistic symbionts (e.g., Klironomos, 2002; Revillini et al. , 2016), ignoring the variety of soil biota guilds that can generate positive PSFs. For example, although plants can grow better in a soil with specific symbiotic partners (e.g., mycorrhizal fungi), they can also positively respond to the presence of specific microorganisms that (1) better cycle/solubilize nutrients (Marschner et al. , 2011), (2) are less efficient competitors for mineral nutrients (Hodge et al. , 2000), or (3) are less virulent pathogens (Agrios, 2005; Dominguez-Begines et al. , 2020). Yet, very few studies have specifically addressed the role for non-symbiotic and non-mutualistic organisms in driving positive PSF (Yang et al. , 2013).
While there is a need to identify the full spectrum of microbial taxa/guilds that drive PSF, this could be hampered by the tremendous diversity of the soil microbiome. The rhizosphere of a single plant individual can easily be colonized by thousands of microbial taxa (Wanget al. , 2018), which could make it hard to identify individual microbial taxa driving PSF. One way forward could be to directly correlate PSF to soil microbial α-diversity or community structure, regardless of individual microbial taxa and their guild affiliation. However, while many have argued that plant performance could be tied to the diversity of their microbiome (van der Heijden et al. , 1998; Laforest-Lapointe et al. , 2017), a higher diversity also opens the way to more complex indirect interactions among microorganisms (Hector et al. , 2010), with PSF potentially being simply more complex and unpredictable, not more positive (e.g., Tardif & Shipley, 2015). It thus remains unsure whether community-level properties of the microbiome can be of any help to simplify PSF predictions.
Another key knowledge gap in PSF theory is to determine how PSF-driving microorganisms respond to environmental filters. A pivotal assumption of PSF as key drivers of plant and microbial community assembly is that the microorganisms influencing plant performance must be responsive to plant community structure (Bever et al. , 1997). Yet, many studies have identified soil abiotic properties as the core drivers of soil microbiome (e.g., Dumbrell et al. , 2010). If this is true for microorganisms influencing plant growth, this means that pot-based PSF, where soil chemistry is kept homogeneous while plant identity is manipulated, may have limited ecological relevance. Also, if the wide array of microorganisms causing PSF all respond differently to environmental filters, predicting PSF along ecological gradients (e.g., Smith-Ramesh & Reynolds, 2017; Bennett & Klironomos, 2018) will be challenging. For example, fertile soils may promote higher plant density and dominance by a few competitive species (Fraser et al. , 2015), conditions that would promote pathogen accumulation and negative PSF (Thrall et al. , 2007). On the other hand, high soil fertility could also relieve plants from microbial competitors for nutrients (e.g., Liu et al. , 2016), which may rather lead to positive PSF. The high soil microbial diversity appears again as an obstacle towards a mechanistic understanding of PSF.
In this study, we aimed at providing a more comprehensive explanation for PSF observed in a grassland ecosystem, focusing on the growth responses of two abundant C3 grasses: Koeleria macrantha (Ledeb.) Schult and Bromus inermis Leyss. Building on a previous greenhouse experiment where we measured Bromus andKoeleria growth responses to 490 soil inocula collected in the field (Chagnon et al. , 2018), here we characterized the soil microbiome from a large subset of these inocula. This allowed us to identify the microbial taxa that best correlate with altered plant performance (i.e., non-neutral PSF), and answer the following questions:
  1. Are negative PSF driven largely by microbial pathogens or due to other microbial guilds?
  2. Can plant growth response to soil inocula be predicted based on microbial α-diversity or community structure?
  3. Are microorganisms causing PSF more correlated with gradients in plant community structure than with proximate soil properties?