Terrestrial Inputs Shape Coastal Bacterial and Archaeal Communities in a High Arctic Fjord (Isfjorden, Svalbard)

Abstract

The Arctic is experiencing dramatic changes including increases in precipitation, glacial melt, and permafrost thaw, resulting in increasing freshwater runoff to coastal waters. During the melt season, terrestrial runoff delivers carbon- and nutrient-rich freshwater to Arctic coastal waters, with unknown consequences for the microbial communities that play a key role in determining the cycling and fate of terrestrial matter at the land-ocean interface. To determine the impacts of runoff on coastal microbial (bacteria and archaea) communities, we investigated changes in pelagic microbial community structure between the early (June) and late (August) melt season in 2018 in the Isfjorden system (Svalbard). Amplicon sequences of the 16S rRNA gene were generated from water column, river and sediment samples collected in Isfjorden along fjord transects from shallow river estuaries and glacier fronts to the outer fjord. Community shifts were investigated in relation to environmental gradients, and compared to river and marine sediment microbial communities. We identified strong temporal and spatial reorganizations in the structure and composition of microbial communities during the summer months in relation to environmental conditions. Microbial diversity patterns highlighted a reorganization from rich communities in June toward more even and less rich communities in August. In June, waters enriched in dissolved organic carbon (DOC) provided a niche for copiotrophic taxa including Sulfitobacter and Octadecabacter. In August, lower DOC concentrations and Atlantic water inflow coincided with a shift toward more cosmopolitan taxa usually associated with summer stratified periods (e.g., SAR11 Clade Ia), and prevalent oligotrophic marine clades (OM60, SAR92). Higher riverine inputs of dissolved inorganic nutrients and suspended particulate matter also contributed to spatial reorganizations of communities in August. Sentinel taxa of this late summer fjord environment included taxa from the class Verrucomicrobiae (Roseibacillus, Luteolibacter), potentially indicative of a higher fraction of particle-attached bacteria. This study highlights the ecological relevance of terrestrial runoff for Arctic coastal microbial communities and how its impacts on biogeochemical conditions may make these communities susceptible to climate change.

Keywords: Arctic; biogeochemical cycles; climate change; freshwater runoff; land-ocean connectivity; melt season; pelagic microbial communities; rivers and sediments.

FIGURE 1

Station map of Isfjorden. Water column samples were collected from two depths in June and August 2018. Color of the symbols indicates the location in the fjord. An asterisk at the station name indicates sites where only sediments were collected. Other stations include both water column samples and sediment samples. Glaciers are represented in light-gray and land in dark-gray. The insert marks the location of the fjord in Svalbard. Details of the sample names, coordinates, location and type are available in Supplementary Table 1. The map was made using PlotSvalbard (v0.8.11) in R (Vihtakari, 2019). The Svalbard map originates from the Norwegian Polar Institute (2020, CC BY 4.0 license) and the bathymetry shapefile from the Norwegian Mapping Authority (2020, CC BY 4.0 license).

FIGURE 2

Boxplots showing alpha diversity indices according to water type and seasonal groupings. Individual data are given by black points, boxplots show the median values (line), interquartile range (box), and range of the data (whiskers). Alpha diversity indices (indicated on the left axis) were calculated as Chao1 (A), Pielou’s index for evenness (B), Shannon’s diversity index (C), and rare biosphere as singletons (SSO) generated after rarefaction (number of OTUs occurring only once in the sample after rarefaction) (D). nOTU and ACE (not shown) followed the same trend as Chao1, and inverse Simpson’s followed the same trend as Shannon’s. The bottom axis indicates the habitat (sediment, river) or water type (water column). A post hoc Dunn’s test was performed on the Kruskal-Wallis test output to test for differences between seasonal groups and water type groups (Supplementary Figures 2, 3 and Supplementary Table 4).

FIGURE 3

Non-metric multidimensional scaling (NMDS) ordinations showing bacterial beta diversity based on Bray-Curtis dissimilarities of (A) microbial community structure in the fjord (stress value = 0.15) and (B) functional composition based on predictions of the KEGG metabolic pathways based on OTU data (stress value = 0.06). The same legend applies to the two plots.

FIGURE 4

Taxonomic affiliation of abundant taxa. (A) Stacked barplot showing the bacterial relative abundance of the most abundant classes. For water column samples, the sampling month is indicated above the plot, and samples are further separated between surface waters (SW) and advected waters (AdW). A table summarizing relative abundances of these classes according to month and water type is provided in the Supplement (Supplementary Table 5). (B) Heatmap showing the mean relative abundances of the most abundant genera for each habitat and month in the water column. Relative abundances are shown on a log-scale for higher resolution. A high relative abundance is indicated by a blue color, a lower abundance is indicated by a red color, and a gray color indicates the absence of the taxon.

FIGURE 5

Heatmap showing differential abundances of the principal predicted functions in element cycling. Functions were predicted based on taxonomy using Tax4Fun (R). Relative abundances were z scaled to be comparable (differential abundance). For water column samples, the sampling month is indicated above the plot, and samples are further separated between surface waters (SW) and advected waters (AdW). Blue indicates a high abundance, and red a low abundance. Metabolic functions were determined with KEGG Orthologs genes or enzymes related to the KEGG reaction or pathway (Supplementary Table 2).

FIGURE 6

Redundancy analysis (RDA) with environmental drivers of community structure (A) highlighting seasonal gradients in the fjord water column, and (B) highlighting the spatial gradient in the fjord water column in August. Significant constraining variables are indicated in red. RDAs were run on OTUs, and only OTUs with an RDA score > 0.07 are represented and are scaled by 3. The name of the closest related genus or family (highest specified taxonomic resolution) is given for these OTUs. Percentages indicate the amount of variance explained by each axis.

FIGURE 7

Heatmap showing significant Spearman correlations (adjusted Spearman rho) between relative abundance of seasonal indicator taxa and Isfjorden physico-chemical variables. The indicators for the two months were statistically retrieved using the indicspecies package in R (IV ≥ 0.7, p ≤ 0.001). A regression was done between the relative abundance of the indicator taxa and the environmental variables using individual data. The sampling month is indicated above the plot and the taxonomic affiliations are indicated beneath the plot. Relevant indicators were chosen among highly abundant indicators and are ordered by decreasing abundance for each month. For a full overview of indicator species, see Supplementary Table 6. Only significant correlations are shown. A blue color indicates a positive correlation and a red color indicates a negative correlation. p-values were FDR-corrected with the BH correction.

FIGURE 8

Conceptual figure depicting major seasonal changes in the microbial communities in relation to changing physical and chemical variables between the early (June) and late (August) melt season. This figure includes components of conceptual figures presented by Kellogg et al. (2019); McGovern et al. (2020), and some symbols are modified after the Integration and Application Network (ian.umces.edu/symbols/). For detailed information on biogeochemistry of the system, see McGovern et al. (2020).