Introduction
Plants and soil organisms are interdependent and the microbiome in the soil is shaped by the plant that grows in the soil (Phillipot et al . 2013; Bardgett & Van der Putten 2014). This microbial signature can remain as a legacy in the soil after the plant is gone, and the soil legacy can affect another plant growing later in the same soil (Kulmatiski et al. 2008; Van der Putten et al. 2013; Testeet al. 2017; Eppinga et al. 2018). It is often speculated that soil legacy effects created by plants play an important role in regulating plant community dynamics. Recent studies suggest that inoculation of soils with biotic legacies can change plant community development under natural conditions (Wubs et al., 2017; Wubs et al., 2019). However, experimental evidence for soil legacy effects of plant communities with different characteristics on responding plant communities in natural systems is lacking (Reynolds et al. 2003; Ehrenfeld et al. 2005; Van der Putten et al. 2013).
Herbaceous grassland plant species such as grasses (monocots) and forbs (dicots) differ fundamentally in functional traits such as root architecture (Craine et al. 2001; 2002; Ravenek et al.2016), water and nutrient acquisition (Tjoelker et al. 2005; Ravenek et al. 2016), and in chemistry and defense (Latz et al. 2015; 2016; Zhang, Van der Putten & Veen 2016). These differences between plant functional types can modulate soil communities (Koset al. 2015; Heinen et al. 2018) and, consequently, create soil legacy effects that affect following plant growth (Latz et al. 2015; Zhang, Van der Putten & Veen 2016; Wubs & Bezemer 2018; Heinen et al. 2018; Heinen, Biere & Bezemer, 2019). Generally, grass and forb species exhibit negative conspecific soil legacy effects (Kulmatiski et al . 2008), which is often explained by the accumulation of specialized pathogens (Van der Putten et al . 2013; 2016). However, growing in conspecific soil can also lead to positive effects through the accumulation of suitable mutualists in the soil (Morrien et al., 2017; Hannula et al. 2017; Teste et al . 2017). Experiments with individually potted plants show that grasses often have increased performance on soils conditioned by forb species and vice versa , where specialized pathogens may not have accumulated (Petermann et al. 2008; De Kroon et al. 2012; Wubs & Bezemer 2018). As species-specific communities of soil organisms develop around the roots of plants, soil legacies may become stronger over time (Diez et al. 2010). While it has been shown that individual plants in the field influence their local soil community (De Rooij-Van der Goes, Peters & Van der Putten 1998; Bezemer et al.2006; Casper & Castelli 2007; Van de Voorde et al. 2011; Hannulaet al. 2019), how plant communities with different characteristics drive soil legacies in the field and what the consequences of these legacies are on the establishment of responding mixed plant communities in these soils is not known (Van der Puttenet al. 2013; Ehrenfeld et al. 2005; Kardol et al.2007).
We grew six different plant communities in a temperate grassland. Each plant community consisted of a combination of grass and/or non-leguminous forb species (hereafter: forbs) which were grown in different ratios (0:100; 25:75; 75:25 or 100:0% forb:grass, respectively). The plots were exposed to different durations of conditioning by starting the treatments in different years (two-year legacy was started in 2015; one-year legacy was started in 2016). In total, there were 192 plots. After the conditioning phase (in 2017), all plant communities were removed from the soil, and an identical seed mixture of 33 grassland species was sown in each treatment plot as a responding plant community. In both phases we recorded the abundance of all plant species, soil abiotic characteristics, and soil fungal and bacterial community composition.
We hypothesized that manipulation of the composition of the conditioning plant communities will result in different microbial soil legacies, and specifically in the accumulation of specialized soil pathogens and mutualists. Second, we hypothesized that in the response phase, grasses and forbs would be less abundant in soils that had been dominated by their own functional type in the conditioning phase, due to the accumulation of soil pathogens. Third, we hypothesized that these effects would be stronger in soils with a two-year legacy than in one-year legacy soils, due to the gradual development of specific soil microbiomes over time. Lastly, we hypothesized that soil legacy effects would be mediated by microbial changes in the soil, rather than by soil abiotic characteristics.