Introduction

As the warming climate pushes the southern border of permafrost northwards, carbon that has been trapped in the permafrost for thousands of years becomes available (Tranvik et al., 2009; Vonk et al., 2012). Thaw ponds emerge as a characteristic feature of thawing permafrost, creating microbial hotspots of carbon cycling (Roiha et al., 2015) and comprising significant sources of greenhouse gas emissions (Abnizova et al., 2012; Negandhi et al., 2013).

The environmental importance of these habitats together with the impact of microbial community composition on carbon degradation (Logue et al., 2016) make it essential to understand the microbial processes and communities in these ponds. Several studies have assessed the underlying microbial communities and their assembly processes (Crevecoeur et al., 2015; Comte et al., 2016; Przytulska et al., 2016). These studies found a bacterial community that resembled the bacterioplankton communities of temperate and boreal lakes but enriched with methanotrophic lineages (Crevecoeur et al., 2015). Still, while the taxonomic composition of thaw pond communities is characterized by these functional groups, it varies significantly across individual thaw ponds, potentially driven by stochastic processes and environmental filtering (Comte et al., 2016). As thaw ponds increase in size with age, often growing from a puddle to a pond within a decade, the age of the pond is likely to be an additional deterministic factor associated with the microbial community composition.

We hypothesize that emerging thaw ponds are initially colonized by a pioneer community, which then undergoes a progressive succession as the pond ages. In this study, we investigated thaw pond communities across a gradient of age classes. To extend previous thaw pond characterization focused on bacterial communities, we used three ribosomal markers (universal SSU, bacterial 16S rRNA gene, and fungal ITS2) to characterize a greater spectrum of the microbial community. We were specifically interested in fungi as they are known to be superior degraders of high molecular weight carbon (Harms et al., 2011), which is the main carbon pool in thaw ponds (Roiha et al., 2015).

Study implementation

The 12 thaw ponds targeted represent three different age categories (emerging, middle-aged, and old) and are located in the Canadian sporadic permafrost area (Figure 1a, Supplementary Methods). The ponds were also sampled vertically according to their thermal stratification (oxic epilimnion, transition zone (metalimnion), and suboxic/anoxic hypolimnion). We extracted DNA from water samples concentrated on 0.22 μm filters and PCR amplified and sequenced three ribosomal markers (the universal V6-8 region, the bacterial V4 region, and the fungal ITS region; Supplementary Methods). We also measured a set of indicators for carbon quantity and quality; and methane and carbon dioxide concentration (Supplementary Figure S1, Supplementary Methods).

Figure 1
figure 1

Study location and characteristics of the microbial communities. (a) Map of the thaw ponds. The emerging and middle-aged ponds are located around still existing palsa formations while the old ponds have replaced their respective palsas completely. Palsas are frost heaves containing permanently frozen ice lenses, consisting of an ice core with overlying soil. At the study site, the height of palsas is up to 3 m. Colors depict different age groups, and asterisks indicate ponds that we sampled also vertically. (b) Two-dimensional NMDS (stress =0.11) showing the beta-diversity within the ponds with shapes illustrating pond age and colors sampling depth. Both age and depth were found to influence the universal community assembly (Adonis, R2=0.25, P<0.001 (for age, excluding hypolimnion samples) and R2=0.20, P<0.001 (for depth)). (c) Universal, (d) bacterial and (e) fungal community composition in the epilimnia of ponds representing different stages of succession. The composition is shown at the phylum level for bacteria and fungi and at the domain level for the universal community, except that fungi have been separated from Eukaryota due to their central role in the study. For bacteria, the group ‘Other’ is the sum of phyla containing less than 0.1% of the total community. For the bacterial and universal communities also the taxonomic composition in the hypolimnion of the old ponds is shown. The hypolimnion is the anoxic bottom layer present only in the old ponds. A more detailed taxonomic overview can be found in Supplementary Figure S3.

Full size image

Results and Discussions

The emerging thaw ponds were significantly different from the older ponds in their environmental parameters. In particular, they differed in their lower pH and higher concentration of dissolved C02, dissolved organic carbon (DOC), refractory carbon, and total nitrogen. Although the methane concentrations were high in hypolimnia of the old ponds, epilimnetic methane was not significantly different between emerging and old ponds. (Supplementary Table S1, Supplementary Figure S1). The organism beta-diversity was significantly correlated with pond age (Figure 1b; Adonis: R2=0.25, P<0.001). The successional progression of thaw ponds was also visible in the relative abundance of the taxonomic groups (Figures 1c–e, Supplementary Figure S2; for a detailed inspection of taxonomic groups, see Supplementary Figure S3). Overall, bacteria constituted the most abundant microbial group followed by Eukarya and Archaea (Figure 1d), both of which can be expected to account for important functional aspects at different depths in the ponds (for example, archaeal methane production at the bottom (Supplementary Figure S4); or algal primary production at the surface). The fungal community ranged between 1 and 3% of the total community and harbored typical freshwater lineages such as Chytridiomycota (Figure 1e; Monchy et al., 2011), but especially in the emerging ponds, which possessed the highest carbon loads, a large part of the community was composed of fungal lineages whose taxonomic resolution could not be resolved with certainty (Figure 1e, Supplementary Table S1). In fact, around 25% of all fungi could only be assigned to kingdom level, and 9% of the fungal reads matched the enigmatic, undescribed fungal lineages identified by Nilsson et al. (2016). In all the pond stages, both the universal and the bacterial markers identified members of candidate phylum Omnitrophica (candidate division OP3) and taxa from the ‘candidate phyla radiation’ (CPR, Brown et al., 2015), in particular Parcubacteria (candidate division OD1) as dominant taxa. These phyla also exhibited the greatest alpha-diversity (Table 1). Typically, these candidate phyla are found in anoxic environments (Peura et al., 2012; Wrighton et al., 2012; Probst et al., 2016), and we can only speculate on their ecological roles in aerobic layers of thaw ponds. The previous thaw pond studies did not recover these poorly known lineages, but instead reported typical freshwater lineages as the dominant bacterioplankton (Crevecoeur et al., 2015, Supplementary Figure S2). Some of the ponds studied here are the same as those examined in the earlier studies (here assigned SAS2 A-B, also sampled in the same season; for example, Crevecoeur et al., 2015). This excludes geographical and seasonal variation as the primary cause for this discrepancy—leaving annual variation or methodological biases as explanations. For example, Rossi et al. (2013) have speculated that previously used primers may have had mismatches within the candidate phyla. We analyzed primer matches in silico and could confirm mismatches of previous primers for at least Omnitrophica and Parcubacteria (Supplementary Table S2).

Table 1 Diversity measures of the total bacterial community in the ponds, excluding hypolimnion samples, representing different phases of the succession (in italics, ±standard deviation), median of all phyla, and the median diversity of each individual phylum
Full size table

Although our study is somewhat restricted geographically and may not represent the complete circumpolar permafrost border, it shows that permafrost thaw ponds undergo a progressive succession over time with respect to both microbial community composition and the composition of carbon compounds. Given the observation that the carbon pool especially in the emerging and middle stages of pond development is dominated by carbon compounds of terrestrial origin (see HI- and FI-indices and CDOM concentration in Supplementary Figure 1), we hypothesize that the cryptic microorganisms are likely to be involved in the processes associated with thawing permafrost soils.

Conclusions

Our study site, located at the edge of the permafrost line, represents an area central to the global warming discussions due to its sensitivity to temperature change (Abnizova et al., 2012). The microbial diversity in permafrost thaw ponds is to a large extent composed of microbes with enigmatic ecological roles and taxonomic affiliations. Although these ponds share some ecosystem features with lakes, they form a distinct ecosystem type in terms of environmental parameters, organism composition, and slow pond progression. Our findings render thaw pond microbiology unique among surface waters and may establish permafrost thaw ponds as a new ecosystem category. The fact that the ecology of their microbiome cannot be assessed at present only further hints at our incomplete understanding of the processes underlying the carbon and nutrient cycling, both of which lie at the very heart of climate change discussions.