Communities differences between surface and deep soils change through time
The beta-diversity between surface and deep layers was particularly high soon after the retreat of glaciers, and then decreased with time. As seen for the alpha diversity, this pattern was consistent across nearly all taxa. Collembola and Oligochaeta are the only taxa for which this was not evident, but these animals were nearly absent from recently deglaciated soils (and particularly from the deep layers; Fig. 2), probably because many of them require well developed soils, with abundant organic matter to find resources (Phillips et al., 2019). Therefore, for Collembola and Oligochaeta, many plots at early development stages were excluded from this analysis, reducing statistical power. In principle, the variation of beta diversity between surface and deep layer can be attributable to both species gain/loss (nestedness) and replacement (Baselga, 2010). Turnover was more important than nestedness for invertebrates (Collembola, Insecta, Oligochaeta), while for microorganisms (Bacteria and Mycota) turnover and nestedness showed a similar importance (Supplementary fig. S2), and the relevance of these two components of beta-diversity remained similar through time (Supplementary Fig. S2; Table S5).
The decrease in beta-diversity between surface and deep layers through time confirms the hypothesis of homogenization of communities (Rime et al., 2015), and extends it to the whole soil biota, as bacteria, microeukaryotes and animals responded the same way (Fig. 3). Community homogenization is probably related to the structural modifications observed during the development of soil horizons (e.g. Schaetzl & Anderson, 2005). The study of sites at different stages of soil formation has shown a differentiation of organic horizon immediately after glacial retreat (O), followed by the development of an organo-mineral horizon (A) during the first 150 years (Crocker & Major, 1955; Mavris, Egli, Plotze, Blum, Mirabella and Giaccai, 2010). The strong vertical variation of physical, chemical and structural features (e.g. light, temperature, pH; Moradi et al., 2020; Mundra et al., 2021) clearly affects communities, which show a particularly strong response to fine-scale environmental heterogeneity (Rime et al., 2015; Moradi et al., 2020; Mundra et al., 2021). For example, immediately after glacier retreat, the amount of fine sediments is the highest at the surface (Rime et al., 2015). This can determine differences in humidity between the surface and the deeper layers, that in turn affect communities (Rime et al., 2015). The decrease of beta-diversity can be explained by the progressive deepening of the organo-mineral horizon (Mavris et al., 2010), where abundant resources favor the establishment of complex communities. Plant richness and cover quickly increase during the first decades after glacier retreat, and 40 years after glacier retreat plants cover generally rises above 50% (Rime et al., 2015). Plant roots generally influence the first 20 cm of soils and more, and could have determined the homogenization of superficial and deep samples of our study (Rime et al., 2015). Differences between surface and deep layers would probably be stronger if a larger vertical gradient is analyzed (e.g. from surface to 50 cm deep; Moradi et al., 2020), and this is certainly an important aspect that deserves future studies. However, in glacier forelands the study of deep layers by eDNA analysis is sometimes problematic because rock outcrops are frequent a few centimeters below the surface. In any case, for all the taxa considered, time since glacier retreat remained the main determinant of community variation, as it explained much more variation in community composition compared to depth (Table 1). This confirms the idea that, even though fine-scale heterogeneity certainly has a role, time since glacier retreat remains the main determinant of community evolution after glacier retreat (Ficetola et al., 2021; Rime et al., 2015).
For microorganisms, the significant community differences between soil layers (Table 1) likely are determined by taxa that are specialists of given environmental features. This idea is supported by the observation that all MOTUs identified as indicators of surface or deep soils are bacteria or fungi (supplementary Table S6). Conversely, for invertebrates, soil depth explained a very limited amount of variation in community composition (Table 1). This could be due to the lower richness of these taxa (which limits statistical power), or to the fact that a broader vertical profile would be required to identify specialists (Moradi et al., 2020). Several taxa identified as indicators in Rime et al. (2015) showed similar patterns across the different locations of our study, confirming the strong functional variation of communities through time. Taxa identified as clear indicators both here and by Rime et al. (2015) include the Clostridium bacteria, known as anaerobic, which were consistently found as indicators of the earliest stage of soil development. Similarly, several fungal saprotrophs were indicators across the different stages of soil development, while Lachnum was a microfungus consistently associated with the most developed soils (Nguyen et al., 2016).Gemmatimonas tend to be copiotrophic bacteria (Ho, Di Lonardo, & Bodelier, 2017) and include multiple MOTUs that were found as indicators of different stages. Interestingly, fungi such as Laccaria orHygrophorus (i.e., potential ectomycorrhizal taxa; Nguyen et al., 2016) were also indicators of later stages of development. Contrary to Rime et al. (2015), arbuscular mycorrhizal fungi (i.e., Glomeromycetes) tended to be associated with oldest forelands, confirming the growing importance of plant-associated fungi along community development (Davey et al., 2015).