Discussion
Here, we show in a field experiment that compositionally different plant communities create legacies in the soil that, in turn, alter the composition of subsequent plant communities that establish in these soils. Plant communities with different ratios of grasses and forbs created unique soil microbiomes, and these effects were most notable in the soil fungal community. These soil legacies, in turn, affected the responding plant communities. Specifically, both grass and forb abundances in the responding phase were negatively affected by their respective abundance in the previous plant community and this effect was mediated by soil processes. We show that manipulating the composition of the vegetation in grasslands alters the microbiome in the soil, and that this alters the succeeding vegetation.
Plant communities dominated by plants of a certain functional type create legacies that negatively impact plants from the same functional type. This result is very robust, as the same pattern was observed in all six plant communities that were used to condition the soil in this field experiment. This finding is also in strong agreement with previous work from pot studies (Kulmatiski et al. 2008; Petermann et al. 2008; De Kroon et al. 2012; Wubs & Bezemer, 2018). The functional type a plant belongs to also alters community structure in soil fungi (Kos et al . 2015; Heinen et al. 2018).
In the soil of plant communities that had more grasses, we found an accumulation of fungal pathogens (dominated by grass-associated fungal pathogens). Interestingly, the abundance of forb-associated pathogens was very low and there was no relationship with the abundance of forbs in the vegetation. Grasses are a phylogenetically close group of plants. Forbs, on the other hand, are a broad phylogenetic group. The specificity of grass pathogens may be one reason why grass-associated pathogens may drive grass abundance in the field, but no such phenomenon was observed for forbs. Specialized forb pathogens are unlikely to accept hosts from all forb families, given the phylogenetic breadth of the group and as a result their individual abundances may not drive abundance of forbs as a group. Specialized pathogens may often attack a range of hosts if they are closely related but some pathogens have rather broad host range (Barrett & Heil 2012) . Importantly, our results indicate that negative soil legacy effects on grasses observed in mid-successional grasslands, can be, at least partially, explained by accumulation of pathogens (Kulmatiski et al. 2008; Van der Puttenet al. 2013).
Our results further reveal that both bacteria and fungi in the soil both respond to the conditioning plant communities that grow in the soil. However, only the conditioning effects on fungal communities are longer-lasting, and have knock-on effects on the subsequent responding plant communities. We may conclude that soil bacterial communities, although responsive to conditioning treatments, play a lesser role in the community dynamics in responding plant communities. This finding is in strong agreement with recent findings that soil fungal communities are strongly shaped over time by plants, whereas bacteria are shaped less by plants, but more by varying environmental conditions over time (Hannula et al. accepted manuscript). Soil legacy effects in natural communities are likely not driven by one taxon specifically, but rather by the composition of the soil fungal community as a whole (Semchenko et al. 2018; Bennett & Klironomos 2018; Mommeret al. 2018, but see Harrison & Bardgett, 2010). Importantly, we show that conditioning effects of plant communities on soil biota, outweigh the effects on soil abiotic parameters, and are drivers of soil legacy effects on plant growth in the field.
One potential confounding factor in the results is that plant roots and seeds originating from the conditioning plant community could have been left behind in the soil after the conditioning community was removed and that these may influence responding communities via this pathway. There were some positive conspecific relationships between conditioningand responding plant species, but the observed effects were community-specific. For instance, a positive conspecific relationship was observed for Rumex acetosella . This species flowers very quickly and produces many seeds. It is possible that seeds produced during the conditioning phase, and that entered the seedbank, caused an increased local abundance of this species in the responding communities. Furthermore, we observed a positive conspecific relationship forClinopodium vulgare and Holcus lanatus . Both species regrow from root systems in pot experiments (Heinen, pers. obs.) and hence regrowth may drive the observed relationships. However, it is unlikely that these effects have had a strong effect on theresponding plant community as a whole, as the strongest relationships - observed between functional types in theconditioning versus the responding plant communities - were negative and thus cannot be explained by regrowth or seed production. We therefore conclude that soil legacy effects must be the dominant driver of these effects.
In this study we find that at the plant species level, we find very few indicators for conspecific plant-soil feedbacks and the effects are strongly conditioning plant community specific. This is an interesting finding as the field site that was used in this study has been the soil donor site for countless plant-soil feedback studies over the past decades. In the majority of these studies, plant species grown in soils from this site have negative conspecific feedback effects (e.g. REF). This indicates that individual plant-soil feedbacks as observed in individual pot studies, may be counter-balanced by other plant species that simultaneously grow in (and thus condition) the soil in natural and diverse plant communities. We may speculate that conspecific plant-soil feedbacks could play a larger role in less diverse or more disturbed systems such as dune vegetations or in very early secondary succession. However, future work is needed to investigate the role of plant diversity or ecosystem complexity in plant-soil feedbacks in the field.
In conclusion, we show here that the ratios between plants of different functional types within a plant communities mediate plant-induced microbial soil legacies and that these legacies determine the composition of plant communities in the field. Importantly, this means that by managing current plant communities in the field, we can influence the composition of future plant communities and the ecological functions they provide. This opens new avenues for optimizing nature management practices, which is vitally important in the face of global change, for instance in making nature more robust to climate change or invasions.