Microbial communities of the Laurentian Great Lakes reflect connectivity and local biogeochemistry

Abstract

The Laurentian Great Lakes are a vast, interconnected freshwater system spanning strong physicochemical gradients, thus constituting a powerful natural laboratory for addressing fundamental questions about microbial ecology and evolution. We present a comparative analysis of pelagic microbial communities across all five Laurentian Great Lakes, focusing on Bacterial and Archaeal picoplankton characterized via 16S rRNA amplicon sequencing. We collected samples throughout the water column from the major basins of each lake in spring and summer over 2 years. Two oligotypes, classified as LD12 (Alphaproteobacteria) and acI-B1 (Actinobacteria), were among the most abundant in every sample. At the same time, microbial communities showed distinct patterns with depth during summer stratification. Deep hypolimnion samples were frequently dominated by a Chloroflexi oligotype that reached up to 19% relative abundance. Stratified surface communities differed between the colder, less productive upper lakes (Superior, Michigan, Huron) and warmer, more productive lower lakes (Erie, Ontario), in part due to an Actinobacteria oligotype (acI-C2) that averaged 7.7% of sequences in the lower lakes but <0.2% in the upper lakes. Together, our findings suggest that both hydrologic connectivity and local selective pressures shape microbial communities in the Great Lakes and establish a framework for future investigations.

Figure 1

The Laurentian Great Lakes study system. A. Map of the lakes with grayscale illustrating depths from <100 m to >300 m. Points coloured by lake indicate the locations of sampling stations. B. Physicochemical parameters and bacterial cell counts across lakes. Temperature, chlorophyll a (chl a), dissolved SiO₃/SiO₄ as Si, nitrite plus nitrate (NO₂ ⁻ + NO₃ ⁻) and total dissolved phosphorus (TDP) are shown for all surface samples in 2012 and 2013 (average ± SE; data from US EPA Great Lakes Environmental Database). For N and P, black dots near the x‐axis indicate concentrations below the detection limit, which is indicated by a grey dashed line. These values were set to the detection limit for calculating mean/SE. Microbial cell concentrations as a function of depth are provided for one master station per lake (indicated in Table S1) in 2012 and 2013. Box plots and scatter points are colour coded by the lake.

Figure 2

Microbial communities were strongly structured by depth during summer stratification. A. Principal coordinate analysis of pairwise Bray–Curtis similarities calculated between samples collected at several depths from each of the five Great Lakes during spring and summer surveys. Variation explained by each axis is indicated in parentheses. B. Boxplots of principal coordinate axis 1 values (median ± quartiles) for surface (SRF), deep chlorophyll layer (DCL), shallow hypolimnion (<83 m) and deep hypolimnion (>83 m) samples, highlighting differences observed across depths.

Figure 3

Depth‐stratified oligotypes exhibited a range of distribution patterns and included many deep‐water specialists. A. Heat map showing relative abundance of abundant oligotypes with significant depth enrichment, based on a Wald test comparing log2 fold change with p < 10⁻⁶. Dendrograms depict complete‐linkage clustering among samples and oligotypes, grouping together samples with similar oligotype composition and oligotypes with similar distributions across samples. B. Example depth distributions of four oligotypes during summer stratification: Chloroflexi Anaerolineaceae node62729, Ca. Methylopumilus LD28 node66881, Cyanobacteria Synechococcus node63671 and Methylophilaceae betIV node38603.

Figure 4

Surface communities across the Great Lakes are comprised of both ubiquitous and region‐specific taxa. Heat map shows the 50 most abundant oligotypes during summer 2013, as log2 transformed counts normalized for sequencing depth. Dendrograms depict complete‐linkage clustering among samples (top) and oligotypes (left). Samples, colour coded by lake, formed two groups: upper lakes (Superior, Michigan and Huron) and lower lakes (Erie and Ontario). Oligotype clusters generally reflect ubiquitous taxa or taxa enriched in upper or lower lakes. The leftmost colour bar indicates oligotypes that were significantly enriched in upper or lower lakes based on a Wald test comparing log2 fold change with p < 10⁻⁶. The second colour bar identifies the phylum or proteobacterial class of each oligotype. Relative abundance by lake (mean ± interquartile range) is summarized for select oligotypes with region‐specific distributions.

Figure 5

Maximum relative abundance of each oligotype as a function of the percent of samples where that oligotype was detected. Oligotypes are colour‐coded by phylum. Included in this analysis are samples collected in spring and summer from surface, deep chlorophyll layer (summer, if present), mid‐hypolimnion (summer) and bottom depths. See Fig. S7 for this analysis carried out separately for specific sets of samples. Symbol size corresponds to the median proportion of sequences for each oligotype (when detected).

Figure 6

Manual oligotyping of four tribes revealed differences in composition with depth and across lakes in summer 2013. A–D. Stacked bar plots with the 10 most abundant oligotypes collected at three depths for each station. E–H. Bubble plots depict the distribution of the 15 most abundant oligotypes within each tribe across surface samples, ordered by rank abundance (most abundant at bottom). Relative abundances (proportion of total tribe sequences) were normalized within each oligotype (i.e., 1 corresponds to the highest relative abundance observed for that oligotype across samples, 0.25 corresponds to a relative abundance that is 25% of the highest observed relative abundance). Stations are colour coded by lake and ordered from north to south in Michigan and west to east in the other lakes.