Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2025 Oct 3;27(10):e70183. doi: 10.1111/1462-2920.70183

Spatio‐Temporal Resolution of Microbial Functions and Taxa Associated With Cyanobacterial Harmful Algae Blooms Along a 500‐Km Aquatic Continuum in the Lake Erie Watershed

Sophie Crevecoeur 1,✉, Lori Phillips 2, Arthur Zastepa 3, Jérôme Comte 4,5, Ngan Diep 6, Alice Dove 7, Thomas Edge 8, Thijs Frenken 9, R Michael McKay 9,10, Susan B Waston 11
PMCID: PMC12491988  PMID: 41039986

ABSTRACT

Biogeochemical processes rendered by the aquatic microbiome could influence the development of cyanobacterial harmful algal blooms (cHABs), but those biotic factors are poorly understood and rarely considered. We focused on the link between microbial functions, community composition and environmental gradients along the Thames River–Lake St. Clair–Detroit River–Lake Erie corridor across different seasons. We measured the abundance and expression (transcripts) of genes involved in nutrient cycling and microcystin toxin production with qPCR and determined microbial community composition with high‐throughput sequencing of the 16S rRNA gene. Throughout the year, genes and transcripts involved in P acquisition, denitrification and N fixation were in higher abundance upstream in the Thames River and Lake St. Clair. Gene abundance, rather than expression, correlated with environmental variables, but functional changes were linked to changes in the aquatic microbiome and did not respond directly to larger environmental gradients. Network analysis revealed tighter connections between gene expression and biotic variables than gene presence, with ubiquitous and streamlined‐genomes microbes associated with the dominant bloom‐causing cyanobacteria, highlighting the cooperative dynamic of these associations. Overall, the results highlight the link between the changing microbiome, microbial processes and the watershed influence in the presence of cHABs.

Keywords: aquatic microbiome, cyanobacteria, harmful algal blooms, Lake Erie watershed, microbial processes

The aquatic microbiome can influence cyanobacterial harmful algal blooms. Abundances of genes and transcripts for phosphorus acquisition, denitrification and nitrogen fixation were higher upstream, whereas in Lake Erie genes and transcripts for microcystin production were higher. The environmental gradient changed the microbiome's composition, which was then linked to functional changes.

graphic file with name EMI-27-e70183-g005.jpg

1. Introduction

Western Lake Erie has shown signs of eutrophication since the mid‐late 1900s due to nutrient (nitrogen and phosphorus) loading from a primarily agricultural watershed. Actions associated with the Great Lakes Water Quality Agreement (GLWQA) currently target P loads by implementing best management practices (BMP) across the watershed. Despite remediation actions, the lake still suffers from recurrent harmful algae blooms, dominated by potentially toxic cyanobacteria (phylum Cyanobacteriota), notably Microcystis, especially in the western basin (Watson et al. 2016). While the role of hydrology and nutrient loading as contributing factors to cyanobacteria growth and toxin production has been the focus of numerous studies (Schindler et al. 2008; Kane et al. 2014; Newell et al. 2019; Chaffin et al. 2023), less attention has been given to the role of biotic interactions when studying cHABs (Woodhouse et al. 2016; Pound et al. 2021; Bramburger et al. 2023). Yet, a better understanding of the involvement of heterotrophic microbes and the entire microbiome in carbon and nutrient cycling could give new insight on how to approach lake management.

The aquatic microbiome mediates nutrient cycling through a variety of processes. Microbes can solubilise and mineralise inorganic P, incidentally contributing to lake eutrophication (Li et al. 2022), while N can be cycled through fixation, nitrification, or denitrification, a multi‐step and multi‐organism process resulting in gaseous N loss (Knowles 1982; Jetten 2008; Riemann et al. 2022). Additionally, several microbial taxa have the ability to degrade cyanobacterial toxins (Kormas and Lymperopoulou 2013). Microbes can be directly attached to the mucilaginous micro‐environment around the cyanobacterial aggregates (Cai et al. 2013), which is what is historically called the phycosphere or, more recently, the interactome (Cook et al. 2020) and provide a direct nutrient supply to the cyanobacteria (Jiang et al. 2007; Parveen et al. 2013; Louati et al. 2015). Additionally, the free‐living co‐occurring microbes in the surrounding environment also occupy an important role in supporting cHABs, for example by recycling leaked organic matter (Christie‐Oleza et al. 2017) or regenerating N supplies (Wang et al. 2021). The role of those biogeochemically relevant co‐occurring microbes is seldom investigated, even though they are part of the same aquatic microbiome as the bloom‐forming cyanobacteria and provide a better indicator of how the ecosystem responds to disturbance as they are more sensitive to environmental changes (Yang et al. 2017). Here, we considered both the interactome and the free‐living co‐occurring microbes as part of the microbiome associated with cyanobacterial blooms.

Our previous study on Lake Erie aquatic microbiome highlighted that its structure was influenced by a gradient of decreasing nutrient concentrations from the upstream Thames River to the downstream Western Basin (Crevecoeur et al. 2023). This type of community change could induce a shift in microbial processes, unless there is a high level of functional redundancy (co‐existence of taxa sharing the same functions), in which case a change in microbial community structure would not translate into a change in functions (Allison and Martiny 2008). In watersheds like Lake Erie, which is greatly impacted by human disturbance, we do not know how the gradients in environmental parameters and the changes in microbial community composition influence microbial functions. However, we expect the cyanobacterial community to be dependent on those functions because of their reduced genome size compared to eukaryotic algae (Cook et al. 2020; Cheng et al. 2023), an evolutionary strategy which allows for rapid reproduction and evolution and that can be especially efficient if the missing functions are compensated with resources contributed by the rest of the microbiome, a scenario known as the ‘Black Queen Hypothesis’ (Morris et al. 2012). Indeed, several studies found functional potential in the Microcystis‐associated microbiome that were not found in the Microcystis genome, such as genes involved in carbon recycling, N fixation and decomposition of hydrogen peroxide (Cook et al. 2020; Smith, Berry, et al. 2022). Cyanobacterial blooms have also been associated with Betaproteobacteria which are involved in degrading dissolved organic matter (DOM) and nitrification (Louati et al. 2015) and with Acidobacteria for exchange of metabolites (Smith, Kharbush, et al. 2022). Studying the role of these associated microbes could offer a new insight on the persistence of cyanobacteria in an ecosystem where worsening of eutrophication symptoms are difficult to explain with P loading alone (Bocaniov et al. 2023).

To better understand how changes in the aquatic microbiome can be associated with the formation of cHABs in the western basin of Lake Erie and to link microbial community function and composition, we combined high throughput sequencing of the 16S rRNA gene with quantitative PCR (qPCR) of several functional genes and transcripts involved in N and P cycling and microcystin toxin production. In different seasons, we used a watershed scale approach and sampled a 500‐km long aquatic continuum (Thames River, Lake St. Clair, Detroit River, western basin of Lake Erie) representing gradients of nutrient input from upstream to downstream. We hypothesised that (1) presence and expression of genes involved in N and P cycling and microcystin toxin production correlate with N, P and microcystin toxin concentrations; (2) the environmental gradient induces less change in microbial function compared to community composition because of functional redundancy and (3) cyanobacterial taxa co‐occur with certain microbial taxa and functions due to their dependence on the rest of the aquatic microbiome. Ultimately, the objective of this study was to assess how the spatial and temporal changes in the aquatic microbiome structure and functions relate to the conditions favouring cyanobacterial blooms.

2. Methods

2.1. Study Sites and Sampling

Four stations in the upper Thames River (NISS, TF, ST, LO), three in Lake St. Clair (RCM, 135, 134), one in the Detroit River (1159) and four in the western basin of Lake Erie (1 K, 885, 966, 970), were selected along the aquatic corridor and sampled monthly from May to October 2019 (one sample collected on 30 April was lumped with the ‘May’ samples). Only one station was sampled in Lake Erie in June, July and September (1 K); whereas in May, August and October, ship support (access to the CCGS Limnos) enabled the sampling of three additional stations (966, 885, 970) (Figure 1A). In total, 60 genomic samples were collected for DNA and RNA analysis (Table S1). All DNA samples were analysed by metabarcoding of the 16S rRNA gene and by qPCR for several genes involved in nutrient cycling and microcystin toxin production. Depending on sample availability and RNA yield, a subset of 25 RNA samples from four stations was selected for the same qPCR analyses on the RNA transcripts converted to cDNA. To increase the coverage for the detection of a gene involved in microcystin production, an additional five stations collected in the western basin of Lake Erie during September 2021 (215, 885, 966, 972, 1163) were included in the analysis (Figure 1A). Sample collection and processing are detailed in Crevecoeur et al. (2023). In brief, surface water samples were collected either with a Niskin bottle (Lake Erie), a Van Dorn bottle (Lake St. Clair and Detroit River) or from the shore with a pole holding sterile bottles (on the Thames River). No surface scum was observed at any site. Temperature (Temp), specific conductivity (Cond) and pH were measured at each site with an SBE 25plus Sealogger CTD (Seabird) on board the CCGS Limnos (Lake Erie) or with a Quatro Pro (YSI Inc.) (Lake St. Clair, Detroit and Thames Rivers). Water samples for total phosphorus (TP), total dissolved phosphorus (TDP), soluble reactive phosphorus (SRP), total nitrogen (TN), total dissolved nitrogen (TDN), ammonium (NH4) and nitrate/nitrite (NO x ) were submitted to Environment and Climate Change Canada's National Laboratory for Analytical Testing (NLET, Burlington, Ontario) and prepared following their Standard Operating Procedures for filtration and preservation (NLET 1997).

FIGURE 1.

FIGURE 1

Open in a new tab

Location of the sampling sites in the Thames River (NISS, TF, ST, LO), Lake St. Clair (RCM, 135, 134), Detroit River (1159) and Lake Erie (1 K, 215, 885, 966, 970, 972, 1163). Map created with R using the open‐access databases ‘worldHires’ (https://www.evl.uic.edu/pape/data/WDB/) and river line downloaded from the Government of Canada Open Data Portal (http://open.canada.ca/en/open‐data) (A). Total phosphorus (TP) and total nitrogen (TN) concentrations as a function of the sampling location. The line in each box plot indicates the median, the box delimits the 25th and 75th percentile and the whisker is the range (B).

2.2. DNA Processing and Metabarcoding

Between 150 and 300 mL of water was filtered through 0.2 μm polyethersulfone filters (Fisher Scientific) in triplicate for DNA and through a 0.22 μm Sterivex cartridge filters (Millipore) for RNA, which were then filled with RNA later (Thermo Fisher Scientific) and stored at −80°C until extraction.

DNA extraction and sequencing followed a standard approach for the government of Canada previously described in Edge et al. (2020) and Crevecoeur et al. (2023). In brief, DNA was extracted using the DNeasy PowerSoil DNA isolation kit (Qiagen) and the V4–V5 hypervariable regions of the 16S rRNA gene were amplified with primers 515F (GTGCCAGCMGCCGCGGTAA) and 926R (CCGYCAATTYMTTTRAGTTT). Library preparation was performed at the Energy, Mining and Environment Biotechnology Research Center of the National Research Council of Canada (Montreal, Quebec) and samples were sequenced on an Illumina MiSeq platform at the National Research Council (Saskatoon, Saskatchewan).

RNA extractions were performed with the Power Water RNA isolation kit (Qiagen) and verified for DNA contamination using a test PCR targeting the 16S rRNA gene with primers 515F and 806R (no amplification expected). Pure RNA was converted into cDNA using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems–Ambion).

2.3. Bioinformatic Analysis

16S rRNA sequences were analysed with the Dada2 pipeline (Callahan et al. 2016) in R (R Core Team 2018) and using the high‐performance computing environment of Shared Services Canada in Dorval, Quebec (Edge et al. 2020). Details of bioinformatics analysis are provided in Crevecoeur et al. (2023). In brief, non‐biological sequences of the primers were removed with cutadapt (Martin 2012), low‐quality bases at the end of the read were trimmed based on the quality profile and sequences with a maximum expected error (maxEE) greater than 2 were removed. The remaining high‐quality sequences were merged in amplicon sequence variants (ASVs) and displayed in an ASVs table. Chimeras were removed de novo and taxonomy was assigned with the Silva database version 128 (Yilmaz et al. 2013), with the exception of sequences corresponding to Cyanobacteriota that were classified with a Barcode of Life Data System (BOLD) reference database (Ivanova et al. 2019). The ASV table was filtered of sequences corresponding to Archaea, Chloroplasts and Eukaryotes and rarefied a hundred times at 10,000 reads per sample with the ‘rarefy_even_depth’ command available in the R package phyloseq version 1.36.0 (McMurdie and Holmes 2013). The average read counts of the 100 tables were used for downstream analyses. One replicate from two samples was lost during the rarefaction process and all ASVs were kept due to the 100 iterations of the rarefaction.

2.4. qPCR

qPCR was used to assess the abundance of genes and transcripts associated with different N and P cycling functions, as well as the abundance of different taxonomic groups. Before qPCR analysis, DNA and cDNA samples were normalised to a working concentration of approximately 1–2 ng μL−1 and re‐quantified in duplicate using Quant‐iT (Life Technologies) dsDNA high sensitivity and OliGreen ssDNA assays, respectively, to determine the final qPCR concentration for each individual sample. Potential carry‐over of DNA in the cDNA was assessed by qPCR of cleaned RNA extracts. All qPCR assays were run on Bio‐Rad CFX384 Real‐Time PCR Detection Systems (Bio‐Rad, USA) in 5 μL reaction volumes with 1 μL of DNA/cDNA and 0.12 μg of UltraPure BSA (Life Technologies, USA).

Quantified functions associated with N cycling included: bacterial ammonium monooxygenase (B‐amoA, Rotthauwe et al. 1997), archaeal ammonium monooxygenase (A‐amoA; Pester et al. 2011), nitrite oxidoreductase (nxrA; Poly et al. 2008—not detected for the transcripts), dissimilatory nitrite reductase (nrfA; Welsh et al. 2014), nitrate reductase (narG; Gregory et al. 2000), bacterial nitrite reductase (B‐nirK; Liu et al. 2003), archaeal nitrite reductase (A‐nirK; Lund et al. 2012), Clade I nitrous oxide reductase (nosZ; Henry et al. 2006), Clade II nitrous oxide reductase (nosZII; Jones et al. 2012) and nitrogenase (nifH; Rösch et al. 2002). Functions associated with P cycling included alkaline phosphatase (phoD; Ragot et al. 2015), acid phosphatase (phoC; Fraser et al. 2017), phosphono‐acetaldehyde hydrolase (phnX; Bergkemper et al. 2016) and pyrroloquinoline quinone (pqqC; Zheng et al. 2017). The full list of master mixes, primer concentrations, run conditions and references is provided in Table S2. The reaction efficiency associated with plasmid standards that were run with every assay ranged from 81% to 110%. The identity of the plasmids, which were derived from soil and sediment DNA, was determined by comparison to the GenBank database.

Assays for qPCR targeting a gene marker or transcript involved in microcystin production (mcyE) were performed using the Phytoxigene kit (Diagnostic Technology, Australia) following the manufacturer's instructions. Five microlitre of template DNA was added to 20 μL of the provided rehydrated master mix. After an initial denaturation at 95°C for 2 min, amplification was carried out by 40 cycles of denaturation at 95°C for 15 s and annealing‐extension at 60°C for 40 s. A standard curve was generated using standards provided by the manufacturer and was used to calculate gene or transcript copy numbers (target gene copy number vs. C t) with a reaction efficiency ranging from 92% to 94%.

2.5. Microcystin Cyanotoxin Extraction and Analysis

Particulate microcystins were collected from whole water samples by filtering onto GF/C filters and frozen at −20°C until analysis. Subsequently, these samples were thawed, tip‐sonicated and syringe filtered through 0.45 μm nitrocellulose filters for measurement by enzyme‐linked immunosorbent assay (ELISA). ELISA was performed on these extracts according to Zastepa et al. (2017) and following manufacturer directions, which specified a method detection limit of 0.05 μg MC‐LR equivalents/L (Abraxis, PA, US, PN 520011SAES).

2.6. Statistical Analysis

The average value for the microbial and cyanobacterial community composition and for the abundance of genes and transcripts was calculated for each system and plotted on stacked bar plots with R using package ggplot2 version 3.3.6 (Wickham 2016). The ratio for each gene presence and expression was calculated between corresponding samples and averaged by process type.

Multiple Spearman's correlations between genes or transcripts concentration and environmental variables were calculated with the rcorr function in the Hmisc package (Harrell 2023). The p‐values were corrected for false discovery rate with the Benjamini and Hochberg (BH) method and only correlations associated with a significant p‐value (< 0.05) were kept. Heatmaps of the correlation plots were drawn with ggplot2 version 3.3.6 (Wickham 2016). Correlations between mcyE gene and transcript abundances and the concentration of total particulate microcystin in the water were tested with Spearman's correlations and also plotted with ggplot2 version 3.3.6 (Wickham 2016). Variation in functions or taxonomy was assessed with Bray–Curtis dissimilarities and in environmental variables with Euclidean distances using the vegdist function in the vegan package (Oksanen et al. 2019). Mean and standard error of the data averaged in 15 bins were plotted with ggplot2. Correlation between distance and dissimilarity matrices was tested with the Mantel test, using the function mantel.rtest in the package ade4 (Dray and Dufour 2007).

Links between genes or transcripts and microbial taxa were assessed with co‐occurrence networks based on Sparse Correlations for Compositional data (sparCC), which removes the bias associated with compositional data such as sequencing data (Friedman and Alm 2012). Pairwise correlations and associated bootstrapped p‐values (1000 repetitions) were calculated with the sparcc command in the package SpiecEasi version 1.1.2 (Kurtz et al. 2022). Gene abundances and expressions were transformed into compositional data (%) as suggested in (Cobo‐Díaz et al. 2019). To reduce the complexity of the network, the ASV table was restricted to ASVs reaching a minimum of 50 reads in at least one sample (equivalent to 0.5% of the community) and present in more than 30% of the samples. Only significant (p < 0.05) and strong (ρ > |0.25|) correlations were kept. Visualisation of the network was done with the software Gephi (Bastian et al. 2009). Each selected ASV was additionally classified to the tribe level using the FreshTrain database (Newton et al. 2011) as well as submitted to a BLASTn search to the nr database in NCBI GenBank (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

3. Results

3.1. Physico‐Chemical Parameters

The pH level ranged from 8.01 on average in the Detroit River in September to 9.02 in Lake St. Clair in August (Table S3). Surface water temperature was higher during July and August in all systems and reached a maximum of 27.1°C at one station in Lake St. Clair during July. Each month, the concentrations of the different fractions of N and P were higher in the Thames River and decreased along our water sampling continuum (Figure 1B, Table S3). The highest values for TP and TDP were found in the Thames River, reaching 179 and 118 μgL−1 during May and September, respectively. Lake St. Clair recorded average TP between 15.1 and 24.3 μgL−1, which corresponds to a mesotrophic state, while Lake Erie had higher TP values on average, ranging from 32.2 μgL−1 in September to 46.3 μgL−1 in June, which correspond to a eutrophic state. SRP values were also consistently higher in the Thames River but remained below limits of detection in the rest of the continuum during June, July and September. TN, TDN and NOx concentrations were on average one order of magnitude higher in the Thames River compared to the rest of the aquatic continuum, except during the months of May and June when TN concentrations were higher in Lake Erie and reached an average of 3.87 and 2.25 mgL−1, respectively. TN, TDN and NOx concentrations then dropped below 1 mgL−1 in Lake St. Clair, the Detroit River and Lake Erie for the consecutive months. The trend of decreasing concentration from upstream to downstream was not observed for NH4 +, which values stayed relatively low in all systems and peaked in Lake Erie and St. Clair during June and July, respectively. TN:TP and TDN:TDP mass ratios also decreased from upstream to downstream.

3.2. Microbial Community Composition

The microbial community composition showed little spatial and temporal variation. In general, the phyla Pseudomonadota and Bacteroidota decreased in relative abundance from the Thames River to Lake Erie (Figure 2A), while the phyla Actinomycetota and Cyanobacteriota increased in relative abundance from upstream to downstream. Cyanobacteriota relative abundance was very low in all samples during the months of May and June (0.5% and 0.7% on average, respectively) but increased later in the season, reaching an average relative abundance of 22.8% ± 2.5% of the microbial community in Lake Erie during August. During the spring months of May and June, the cyanobacterial community was mostly composed of the genera Planktothrix, Synechococcus and Aphanothece (Figure 2B). The genus Synechococcus stayed dominant in Lake St. Clair, Detroit River and Lake Erie during the whole sampling season and peaked in Lake St. Clair in September, composing 9.5% ± 0.5% of the microbial community on average or 60% ± 10% of the cyanobacterial community, while Microcystis started to become more abundant from August to October and peaked in Lake Erie during August, composing 5.4% ± 4.8% of the microbial community on average or 23% ± 20% of the cyanobacterial community. The cyanobacterial community in the Thames River during September and October was strongly dominated by the genus Planktothrix.

FIGURE 2.

FIGURE 2

Open in a new tab

Microbial taxonomy and functions. Relative abundances of the main microbial phyla expressed as % of the microbial community (A) and cyanobacterial genera (B) (legend on the left‐hand side), concentration of gene copies (C), transcripts (D) and transript:gene ratio (E) coloured based on their function (legend on the right‐hand side). Note the changing y axis.

3.3. Link Between Microbial Functions and Environmental Variables

The presence and expression of genes involved in N and P‐cycling, as well as microcystin production, were quantified through measurement of gene (qPCR) (Figure 2C) and transcript (RT‐qPCR) (Figure 2D) abundances. The total abundance of functional genes and transcripts was higher in the Thames River or in Lake St. Clair than in Lake Erie and decreased from upstream to downstream (Figure 2C,D). Genes and transcripts involved in P mineralisation, solubilisation, denitrification (nirK, nosZ and narG) and N fixation (nifH) were consistently detected across space and time. On the other hand, genes and transcripts involved in nitrification (amoA and nxrA), DNRA (nrfA) and microcystin production (mcyE) were under the detection limit during May, June and July. For example, genes and transcripts involved in microcystin production (mcyE) were undetectable or below 103 copies mL−1 in May and June for all samples but became detectable in August, September and October, with maximum values in Lake St. Clair and Lake Erie. The ratio of transcript to gene copy numbers (transcript: gene ratio) could be calculated for 30 samples and gave an indication of the changes in the level of expression, as an increased ratio suggests a higher level of activity for a particular gene. In May and June, the transcript: gene ratio for most of the processes tended to increase from upstream to downstream, while the trend was the opposite in July (Figure 2E). In August and September, there was a marked increase in the transcript: gene ratio of microcystin production from Lake St. Clair and the Detroit River to Lake Erie. The transcript:gene ratio of P mineralisation also tended to increase from upstream to downstream, except in July and September. The transcript:gene ratio related to genes involved in P solubilisation was in general stable across the water continuum, but slightly increased in Lake Erie during August. The average transcript:gene ratio of N fixation tended to decrease from upstream to downstream, except in May and June.

Several significant correlations were observed between gene presence and environmental variables such as Cond, pH and several N and P fractions (Figure 3A), in contrast to gene expression (Figure 3B) or with transcript:gene ratio, where there were fewer correlations (Figure 3C). For example, several genes involved in P mineralisation (phoD, phoC and phnX) and P solubilisation (pqqC) showed positive correlation with the several P and N fractions (Figure 3A), while the transcripts and transcript:gene ratio of phoC only correlated with NH4 + (Figure 3B) and the transcript:gene ratio of phnX negatively correlated with TP, SRP and TDN (Figure 3C). The same trend was observed for the genes involved in N cycling, where those genes correlated with several N and P fractions, while only the transcripts of the gene coding for the archaeal ammonia‐monooxygenase (A‐amoA) and the nitrous oxide reductase (nosZII) were positively correlated with NH4 and TP, respectively. As for the gene involved in microcystin production, mcyE gene presence was only negatively correlated to TN, NOx and TN:TP ratio. Additionally, there was a strong and significant correlation between the concentration of total particulate microcystin in the water and mcyE gene copies (Spearman's ρ = 0.96, p < 0.01), but not significant for mcyE transcripts (Spearman's ρ = 0.44, p = 0.2) (Figure 4).

FIGURE 3.

FIGURE 3

Open in a new tab

Heatmap of the correlations between gene copies (A), transcripts (B), transcript:gene ratio (C) and environmental variables conductivity (Cond), pH, temperature (Temp), total phosphorus (TP), total dissolved phosphorus (TDP), soluble reactive phosphorus (SRP), total nitrogen (TN), total dissolved nitrogen (TDN), nitrite/nitrate (NO x ) and TN:TP ratio. Grey X indicate assay below detection limit. Only significant correlation are displayed (BH corrected p < 0.05).

FIGURE 4.

FIGURE 4

Open in a new tab

Relationship between concentration of total particulate microcystin in the water and the abundance of the mcyE genes (round) and transcripts (triangle).

3.4. Influence of the Changes in Environmental Variables on Community and Functional Dissimilarities

The changes in community composition across the watershed were positively and significantly correlated with the changes in environmental variables (Figure 5A; Mantel's r = 0.3, p < 0.01), but this trend was not significant for the changes in microbial functions which stay relatively stable as the distance in environmental variables increased (Figure 5B; Mantel's r = 0.1, p > 0.05 for both gene presence and expression). On the other hand, the change in microbial community composition was significantly correlated with the change in function and the relationship seemed stronger for gene expression compared to gene presence, although the value of the Mantel test did not differ between the two (Figure 5C; Mantel's r = 0.2, p < 0.05 for both gene presence and expression).

FIGURE 5.

FIGURE 5

Open in a new tab

Relationship between taxonomical dissimilarities and environmental distances (A), functional dissimilarities and environmental distances (B), and functional and taxonomic dissimilarities (C). Data were binned into 15 groups, dots and error bar represent the means and the standard error of the binned data, respectively.

3.5. Co‐Occurrence Networks

Co‐occurrence networks were used to decipher links between functional genes and microbial taxa. Both networks based on gene presence (Figure 6A) and expression (Figure 6B) showed the same percentage of positive correlation (69%), particularly, for correlations among genes and transcripts involved in nutrient cycling. The network based on gene presence had more vertices (42) than the network based on gene expression (39) but fewer edges (138 vs. 161, respectively). Consequently, the network based on gene expression was more clustered and connected, with a clustering coefficient (the degree to which nodes in a network tend to cluster together) of 0.58 compared to 0.52 for the network based on gene presence and an edge density (proxy for the network's connectedness) of 0.22 compared to 0.16 for the network based on gene presence. Both networks selected ASVs belonging to dominant phyla already identified in the community as Actinomycetota, Bacteroidota, Chloroflexota, Cyanobacteriota and Pseudomonadota for which genera were further classified into tribes with FreshTrain and BLASTed on GenBank (Table S4). Half of the taxa were shared across the two networks, which include the two highly connected ASV 1 (Pseudomonadota—tribe LD12, Fonsibacter) and 2 (Actinomycetota—tribe acI‐B1, Nanopelagicus) with more than 20 edges in both networks. Those two ASVs were negatively correlated with the presence and expression of several genes involved in denitrification, DNRA, nitrification, P mineralisation and solubilisation. One gene involved in nitrification (Archaeal amoA) was also positively correlated with ASVs 15 and 16 (both Bacteroidota—tribe Flavo‐A3, Flavobacterium), but otherwise there were not many correlations between functional gene presence and bacterial taxa. On the other hand, there were more connections between functional gene expression and bacterial taxa. For example, the expression of the pqqC gene, indicative of P solubilisation, was one of the most connected nodes correlating with 11 different ASVs (1, 2, 7, 10, 11, 13, 15, 16, 19, 34 and 81), while the expression of genes involved in denitrification (nosZ, nosZII and narG) correlated to 7 ASVs (3, 4, 11, 13, 15, 16 and 26) and nitrification (Archaeal amoA) to 5 ASVs (1, 10, 15, 16 and 19).

FIGURE 6.

FIGURE 6

Open in a new tab

Co‐occurrence networks of microbial ASVs and functional genes (A) and transcripts (B). Equivalent network only highlights vertices corresponding to Cyanobacteriota ASVs and their connection for gene presence (C) and expression (D).

The network based on gene presence selected five cyanobacterial ASVs (Figure 6C), two belonging to the genus Microcystis (ASVs 31 and 52), one to Synechococcus (ASV 64) and two to Planktothrix (ASVs 5 and 17), but the latter two were only linked to each other and not included in the rest of the network. Only one of those ASVs was kept in the network based on gene expression (ASV 31, Figure 6D). In both networks, this ASV was connected to at least one of the two most connected ASVs in the network, ASVs 1 and 2. The presence and expression of the gene involved in microcystin production (mcyE) were not correlated with the rest of the functional genes but were positively correlated to one Microcystis ASV (ASV 31) present in both networks.

4. Discussion

In Canada, the Thames River is responsible for the highest P and SRP load to the western basin of Lake Erie via the Detroit River (Maccoux et al. 2016). Even though the TP load in Lake Erie has declined since 1975 coincident with the implementation of BMPs, the fractions of dissolved P and SRP loads have followed the opposite trend (Richards et al. 2009; Singh et al. 2023), calling for further management action and understanding of P dynamics. Here we observed the highest level of total and dissolved nutrients (N, P) concentrations in the Thames River, reflecting the impact of land use dominated by row‐crop agriculture in this catchment (Kao et al. 2022). This nutrient load becomes diluted when it reaches the Lake St. Clair and the Detroit River which, despite contributing 41% of the TP load and being the largest TP source on the Canadian side, is dominated by flow from oligotrophic Lake Huron. In total, the Detroit River has an enormous volume and contributes 80% of the water load into the western basin of Lake Erie (Scavia et al. 2019). These hydrological factors contribute to the creation of an environmental gradient along our sampling corridor, where total and dissolved nutrient concentrations were generally lower at downstream sampling locations towards Lake Erie. The concomitant decrease of N:P mass ratio for total and dissolved fractions, from upstream to downstream, also suggests that N is decreasing at a higher rate than P across the water continuum. Indeed, N concentration in Lake Erie can sometimes reach limitations (Stow et al. 2015; Chaffin et al. 2013, 2014). However, the status and scarcity of nutrients, especially for the dissolved fraction, can be difficult to measure because they are cycled too quickly (Burford et al. 2023) or rely on transformations involving organic pools that are rarely measured (Kharbush et al. 2023). Focusing on the presence and expression of genes involved in nutrient cycling offers a new perspective for assessing the nutrient status in Lake Erie.

The results presented here span 2 years of sampling (2019 and 2021) and include several stations across rivers and lakes that were eventually grouped by ecosystem in order to observe broad trends in nutrient gradients and environmental impacts on microbial community structures and functions. Even though interannual variation in community composition is observed in Lake Erie, our sampling design spanning 2 years of sampling would be minimally affected since, based on historical data, fluctuations of phytoplankton group biomass follow an approximate 5‐year cycle (O'Donnell et al. 2023). However, potential interannual variability as well as spatial heterogeneity within the same basin (Wynne and Stumpf 2015), notably because of the influence of the Maumee and Sandusky River plumes (Jankowiak et al. 2019), should be considered in further studies.

4.1. Functional Gene Abundance More Strongly Correlates With Environmental Variables Than Transcripts

Our first hypothesis, that functional gene abundances correlated with nutrient and microcystin toxin concentrations, was supported for gene presence, but fewer significant relationships were found with gene expression patterns. This outcome reflects the inherent influence of the environmental gradient on microbial functional capacity and also highlights the difficulty of linking short‐lived transcripts (seconds to minutes) to environmental changes that occur over longer time frames (Moran et al. 2012). In a meta‐analysis of qPCR‐based microbial functions (Rocca et al. 2015), significant correlations were only found between microbial processes and gene presence and not with transcripts. This decoupling was again observed recently for microbial organic matter cleavage in the microbiomes of different environments (Zhao et al. 2024).

In general, P mineralisation and solubilisation processes facilitate microbial P acquisition and uptake (Mackey and Paytan 2009) and the activity and upregulation of genes like phoD and phoC have been linked to P scarcity in several cyanobacterial groups (Gebhard and Cook 2008; Mateo et al. 2010; Harke et al. 2016). On the other hand, genes associated with cleaving C–P bonds to mineralise complex organophosphorus compounds (phnX) and P solubilisation (pqqC) processes are usually found in constant abundance through sub‐surface soil (Hussain et al. 2021) and can stimulate plant growth (Meyer et al. 2011). PhnX was also found prevalent in P‐limited regions (Sosa et al. 2019). A recent study on a transect from a small tributary flowing in an agriculturally dominated catchment in the Lake Erie watershed found increased abundance of phoD and phoC towards the mouth of the creek, where P concentration decreased, while phnX and pqqC were in higher abundance upstream (Knorr et al. 2023). Those results highlight the co‐loading of nutrients and competent microbial communities from smaller tributaries. Here, on the other hand, the abundance of P cycling genes was positively correlated with several P fractions and only the transcript:gene ratio of phnX was negatively correlated with total and soluble P, indicating this gene might be preferentially expressed in conditions of P scarcity.

Genes involved in nitrification have been shown to correlate with NH4 and NOx in aquatic ecosystems (Knorr et al. 2023), but that does not always translate to a correlation with nitrification rates (Hoffman et al. 2024). Our data showed a positive correlation between the number of archaeal amoA transcripts and NH4 + concentration, suggesting gene expression might be a better predictor of this process. NO x can be a substrate for N loss through denitrification (Kuypers et al. 2018), which can substantially reduce (up to 100%) the N load (Grantz et al. 2014). While denitrification typically occurs in anoxic conditions, at the water/sediment interface (Finlay et al. 2013), aerobic denitrifying bacteria have been isolated from lakes and reservoirs, where this process may be taking place in micro‐environments, such as anoxic niches inside a microbial floc (Hao et al. 2022). Here, only the presence of some of the genes involved in denitrification correlated with NO x concentration, but the high detectable level of transcript for denitrification across the watershed suggests the microbial community could be primed for that process even in surface water.

It is still unclear whether N fixation can compensate for events of N limitation in Lake Erie (Chaffin et al. 2014). Nitrogen fixation also happens in anoxic conditions or within specialised cells (heterocysts) developed by diazotrophic cyanobacteria (Kumar et al. 2010). This process seems to be important irrespective of N concentrations, as N fixation rates were consistently measured in the Great Lakes despite elevated N concentrations (Salk et al. 2018; Natwora and Sheik 2021) and the presence and expression of the nifH gene involved in N fixation have been detected in N‐repleted waters of Sandusky Bay (Hampel et al. 2019). Here, the presence and expression of the nifH gene were not associated with N concentrations but were correlated with the different P species. Incidentally, high‐P conditions and experimental P enrichment in Lake Erie waters promoted the expression of nifH in the genera Anabaena and Planktothrix (Harke et al. 2016). In our dataset, diazotrophic cyanobacteria (Anabaena/Aphanizomenon/Dolichospermum and Cylindrospermopsis) composed less than 1% of the whole microbial community. Despite their low relative abundance, the expression and presence of genes involved in N fixation have been observed in members of the AAD complex in Lake Erie during a Microcystis bloom (Yancey et al. 2023) and N‐fixing Dolichospermum presence was detected in Lake Erie samples where the N fixation rate was more than 10 times above the average rate of Lake Erie (Natwora and Sheik 2021). N fixation by Aphanizomenon has been shown to provide N supplies to toxigenic Microcystis during N limitation in hypereutrophic Lake Mendota (Beversdorf et al. 2013). N supply could also be delivered by heterotrophic bacterial diazotrophs (HBD), which can colonise planktonic aggregates to access anoxic micro‐niches (Riemann et al. 2022), as observed during Microcystis and Planktothrix blooms (Song et al. 2017; Cook et al. 2020; Wang et al. 2021).

Here the presence of the mcyE genes involved in microcystin production was still a better predictor of the toxin concentration in the water than gene expression. This could indicate post‐transcriptional regulation of the gene expression and a temporal decoupling between transcription and translation into protein (Ngwa et al. 2014). Elevated concentrations of N have been linked to a higher presence or expression of the mcyE gene (Harke et al. 2016; Jankowiak et al. 2019). However, our data showed that the mcyE gene presence was negatively correlated with two N fractions (TN, NOx) and the TN:TP ratio. This could be explained by the fact that the relative abundance of potential microcystin toxin producers like Microcystis also increased in Lake St. Clair, Detroit River and Lake Erie, where the N concentrations and TN:TP ratio tended to drop compared to the Thames River.

4.2. Changes in Microbial Functions Are Linked to Changes in Microbial Community Structure

The positive and significant correlation between environmental distances and taxonomic dissimilarities confirms the influence of the environmental gradient on the microbial community, as was also found in Crevecoeur et al. (2023). In agreement with our second hypothesis that the environmental gradient will induce fewer changes in microbial function due to functional redundancy, here the environmental variation did not directly induce a change in microbial functional potential. Instead, that change was more strongly linked to the change in community composition. The concept of complete functional redundancy implies that co‐existing taxa share similar functions to the extent that a change in community does not induce a change in function. This has been observed across the global ocean microbiome (Sunagawa et al. 2015), during cyanobacterial blooms (Steffen et al. 2012) and in soils and sediments of water catchments (Frossard et al. 2012). However, here we did not observe such disconnection because functional changes were correlated with taxonomical changes and this relationship appears even stronger with gene expression, suggesting a lower redundancy (Figure 6C). From this observation, also reported in soil (Allison and Martiny 2008) and marine ecosystems (Galand et al. 2018), it is concluded that changes in function might be mediated by changes in microbial community composition and serves to reinforce the importance of considering changes in microbial community composition to understand how microbial processes will be altered by environmental disturbances.

4.3. Cyanobacteriota Co‐Occur With a Gene Involved in Microcystin Toxin Production, but With Few Microbial Taxa

Network analysis can be used as a way to infer co‐occurrence and potential interaction between two biological features, which can be microbial taxa but also genes and/or transcripts (Jiang et al. 2019). Here, the network based on gene expression showed more connections between the selected ASVs and functional genes than the network based on gene presence. The ASVs selected in both networks corresponded to ubiquitous genera with broad‐ranging nutrient cycling capacities typically found in freshwater systems, notably tribe LD12 (Fonsibacter), tribe PnecB (Polynucleobacter), tribes Lhab‐A1 and Lhab‐A3 (Limnohabitans), tribes Flavo‐A3 and bacII‐A (Flavobacterium) and tribes acI‐A6 and acI‐A7 (Planktophila) (Table S4) (Newton et al. 2011; Wood 2011). Limnohabitans has a high substrate uptake rate and has been demonstrated to grow on algal‐derived materials (Kasalický et al. 2013), whereas Flavobacterium can become dominant after periods of algal blooms and is favoured by high heterotrophic activity (Newton et al. 2011). Several of these ubiquitous genera were positively correlated with the expression of genes involved in P‐solubilisation, nitrification and denitrification, which supports their role in the cycling of N and P within the system, potentially providing N and P to the cyanobacterial community and contributing to an increase of dissolved P across the Lake Erie watershed observed elsewhere (Richards et al. 2009; Singh et al. 2023). Several ASVs selected in both networks, including the two most connected ASVs (1 and 2), are also known to have reduced streamlined genomes (Fonsibacter, Nanopelagicus, Planktophila, Methylopumilus) (Salcher et al. 2015; Neuenschwander et al. 2018; Mondav et al. 2020). This feature makes those genera auxotrophic and reliant on the resources supplied by the rest of the microbial community (Mondav et al. 2020), which could be one of the explanations for their high prevalence and connection within the network analysis.

Several cyanobacterial ASVs were included in the co‐occurrence network and correlated with ASVs that were central to the network, such as ASV1 and ASV2, as well as with the mcyE gene presence and expression. However, only one Microcystis ASV (ASV 31) was included in the gene expression network. This ASV was the most abundant in Lake St. Clair during the month of August, representing more than 3% of the microbial community and became abundant in the western basin of Lake Erie at the same time, representing more than 2.5% of the microbial community. Interestingly, this ASV was also 100% similar to the Microcystis ASV identified by Crevecoeur et al. (2023) as ASV1, which was dominant in Lake St. Clair, the Detroit River and the western basin of Lake Erie during the summer from 2016 to 2019, but decreased in relative abundance during the autumn. ASV31 could therefore be more tightly linked to microcystin toxin production. Future work re‐constructing the Microcystis pangenome could help reveal which ASV has the potential to produce microcystin toxin.

5. Conclusion

Here we investigated the functional patterns of the aquatic microbiome in the Lake Erie watershed, a system periodically impacted by cHABs. The patterns of presence and expression of genes involved in N and P cycling followed the environmental gradients of nutrients and decreased in abundance from upstream to downstream. In contrast, genes and transcripts involved in microcystin toxin production increased in downstream Lake Erie during the summer and fall. There were many more significant correlations between gene presence and environmental variables compared to gene expression, reflecting the influence of the watershed in shaping microbial community functional potential. However, the variation in function did not respond directly to changes in environmental variables and was linked to changes in aquatic microbiome structure, highlighting the importance of considering change in microbial community composition when studying microbial processes. Even though gene expression showed fewer significant relationships with environmental variables, network analyses of associations with biotic variables revealed more connections between microbial ASVs and functional genes expression than with gene presence. Within these networks, cyanobacterial ASVs corresponding to Microcystis and Synechococcus co‐occurred with microbes linked to functional gene presence and expression that were mostly ubiquitous and had streamlined genomes, which is consistent with the Black Queen Hypothesis stating that associated microbes share ‘leaky’ functions. Overall, assessing the connection between microbial processes and taxa provided new insights into the role of the aquatic microbiome in nutrient cycling and toxin production in Lake Erie and could ultimately improve our ability to predict the occurrence and toxicity of future cHABs.

Author Contributions

S.C., T.E., S.B.W. and J.C. designed the study. S.C., A.Z., N.D., A.D., T.F. and R.M.M. collected samples. S.C., L.P. and A.Z. provided and analysed data. S.C. supervised the project, performed bio‐informatic and statistical analyses and wrote the manuscript with help in editing and correcting from all the other authors. All authors contributed to the article and approved the submitted version.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Number of samples collected in each system.

Table S2: Quantitative PCR assay run information.

Table S3: Average (SD in brackets) of the environmental variables for each system across seasons: pH, temperature, total phosphorus (TP), total dissolved phosphorus (TDP), soluble reactive phosphorus (SRP), total nitrogen (TN), total dissolved nitrogen (TDN), ammonia (NH4), nitrite/nitrate (NO x ), total particulate microcystin (TPM), TN:TP and TDN:TDP ratios.

Table S4: Identity of ASVs selected in networks analysis and corresponding to FreshTrain and GenBank sequences (following a BLASTn search), at the lowest taxonomical level.

EMI-27-e70183-s001.docx (79.1KB, docx)

Acknowledgements

We would like to thank L. Cynthia Watson, Brent Seuradge, Savana Knorr, Celine Siong, Mariane Racine, Anqi Liang, Elis Damasceno, Maria Molina, Xenia Boyko, Hannah Singh, Matthew Pineda and Julianne Radford for their precious help on the field and in the lab, Charles Greer and Sylvie Sanschagrin for their implication in the library preparation and Ken Drouillard for assistance in the field. We thank the two anonymous reviewers for their valuable comments. This study was funded under the Canada's Genomics Research and Development Initiative (GRDI) and Great Lake Protection Initiative (GLPI), also supported by grants from the Natural Sciences and Engineering Research Council of Canada (NSERC) (RGPIN‐2019‐03943), the Great Lakes Center for Fresh Waters and Human Health supported by NIEHS (1P01ES02328939‐01) and NSF (OCE‐1840715).

Crevecoeur, S. , Phillips L., Zastepa A., et al. 2025. “Spatio‐Temporal Resolution of Microbial Functions and Taxa Associated With Cyanobacterial Harmful Algae Blooms Along a 500‐Km Aquatic Continuum in the Lake Erie Watershed.” Environmental Microbiology 27, no. 10: e70183. 10.1111/1462-2920.70183.

Funding: This work was supported by the Canada's Genomics Research and Development Initiative, Canada's Great Lake Protection Initiative (GLPI), Natural Sciences and Engineering Research Council of Canada (NSERC) (RGPIN‐2019‐03943), Great Lakes Center for Fresh Waters and Human Health supported by the National Institute of Environmental Health Sciences (NIEHS) (1P01ES02328939‐01) and National Science Foundation (OCE‐1840715).

Data Availability Statement

The data that support the findings of this study are openly available in NCBI at https://www.ncbi.nlm.nih.gov/sra/PRJNA877648, reference number PRJNA877648.

References

  1. Allison, S. D. , and Martiny J. B.. 2008. “Resistance, Resilience, and Redundancy in Microbial Communities.” Proceedings of the National Academy of Sciences 105: 11512–11519. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bastian, M. , Heymann S., and Jacomy M.. 2009. “Gephi: An Open Source Software for Exploring and Manipulating Networks.” Proceedings of the International AAAI Conference on Web and Social Media 3: 361–362. [Google Scholar]
  3. Bergkemper, F. , Kublik S., Lang F., et al. 2016. “Novel Oligonucleotide Primers Reveal a High Diversity of Microbes which Drive Phosphorous Turnover in Soil.” Journal of Mirobiological Methods 125: 91–97. 10.1016/j.mimet.2016.04.011. [DOI] [PubMed] [Google Scholar]
  4. Beversdorf, L. J. , Miller T. R., and McMahon K. D.. 2013. “The Role of Nitrogen Fixation in Cyanobacterial Bloom Toxicity in a Temperate, Eutrophic Lake.” PLoS One 8: e56103. 10.1371/journal.pone.0056103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bocaniov, S. A. , Scavia D., and Van Cappellen P.. 2023. “Long‐Term Phosphorus Mass‐Balance of Lake Erie (Canada‐USA) Reveals a Major Contribution of In‐Lake Phosphorus Loading.” Ecological Informatics 77: 102131. 10.1016/j.ecoinf.2023.102131. [DOI] [Google Scholar]
  6. Bramburger, A. J. , Filstrup C. T., Reavie E. D., et al. 2023. “Paradox Versus Paradigm: A Disconnect Between Understanding and Management of Freshwater Cyanobacterial Harmful Algae Blooms.” Freshwater Biology 68: 191–201. 10.1111/fwb.14019. [DOI] [Google Scholar]
  7. Burford, M. A. , Willis A., Xiao M., Prentice M. J., and Hamilton D. P.. 2023. “Understanding the Relationship Between Nutrient Availability and Freshwater Cyanobacterial Growth and Abundance.” Inland Waters 13: 1–152. 10.1080/20442041.2023.2204050. [DOI] [Google Scholar]
  8. Cai, H. Y. , Yan Z. S., Wang A. J., Krumholz L. R., and Jiang H. L.. 2013. “Analysis of the Attached Microbial Community on Mucilaginous Cyanobacterial Aggregates in the Eutrophic Lake Taihu Reveals the Importance of Planctomycetes.” Microbial Ecology 66: 73–83. 10.1007/s00248-013-0224-1. [DOI] [PubMed] [Google Scholar]
  9. Callahan, B. J. , McMurdie P. J., Rosen M. J., Han A. W., Johnson A. J., and Holmes S. P.. 2016. “DADA2: High‐Resolution Sample Inference From Illumina Amplicon Data.” Nature Methods 13: 581–583. 10.1038/nmeth.3869. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Chaffin, J. D. , Bridgeman T. B., and Bade D. L. 2013. “Nitrogen Constrains the Growth of Late Summer Cyanobacterial Blooms in Lake Erie.” Advances in Microbiology 03: 16–26. 10.4236/aim.2013.36a003. [DOI] [Google Scholar]
  11. Chaffin, J. D. , Bridgeman T. B., Bade D. L., and Mobilian C. N.. 2014. “Summer Phytoplankton Nutrient Limitation in Maumee Bay of Lake Erie During High‐Flow and Low‐Flow Years.” Journal of Great Lakes Research 40: 524–531. 10.1016/j.jglr.2014.04.009. [DOI] [Google Scholar]
  12. Chaffin, J. D. , Westrick J. A., Reitz L. A., and Bridgeman T. B.. 2023. “Microcystin Congeners in Lake Erie Follow the Seasonal Pattern of Nitrogen Availability.” Harmful Algae 127: 102466. 10.1016/j.hal.2023.102466. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Cheng, W. , Hwang S., Guo Q., et al. 2023. “The Special and General Mechanism of Cyanobacterial Harmful Algal Blooms.” Microorganisms 11: 987. 10.3390/microorganisms11040987. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Christie‐Oleza, J. A. , Sousoni D., Lloyd M., Armengaud J., and Scanlan D. J.. 2017. “Nutrient Recycling Facilitates Long‐Term Stability of Marine Microbial Phototroph–Heterotroph Interactions.” Nature Microbiology 2: 17100. 10.1038/nmicrobiol.2017.100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Cobo‐Díaz, J. F. , Baroncelli R., Le Floch G., and Picot A.. 2019. “Combined Metabarcoding and Co‐Occurrence Network Analysis to Profile the Bacterial, Fungal and Fusarium Communities and Their Interactions in Maize Stalks.” Frontiers in Microbiology 10: 261. 10.3389/fmicb.2019.00261. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Cook, K. V. , Li C., Cai H., et al. 2020. “The Global Microcystis Interactome.” Limnology and Oceanography 65: 194–207. 10.1002/lno.11361. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Crevecoeur, S. , Edge T. A., Watson L. C., et al. 2023. “Spatio‐Temporal Connectivity of the Aquatic Microbiome Associated With Cyanobacterial Blooms Along a Great Lake Riverine‐Lacustrine Continuum.” Frontiers in Microbiology 14: 1073753. 10.3389/fmicb.2023.1073753. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Dray, S. , and Dufour A.‐B.. 2007. “The ade4 Package: Implementing the Duality Diagram for Ecologists.” Journal of Statistical Software 22: 1–20. 10.18637/jss.v022.i04. [DOI] [Google Scholar]
  19. Edge, T. A. , Baird D. J., Bilodeau G., et al. 2020. “The Ecobiomics Project: Advancing Metagenomics Assessment of Soil Health and Freshwater Quality in Canada.” Science of the Total Environment 710: 135906. 10.1016/j.scitotenv.2019.135906. [DOI] [PubMed] [Google Scholar]
  20. Finlay, J. C. , Small G. E., and Sterner R. W.. 2013. “Human Influences on Nitrogen Removal in Lakes.” Science 342: 247–250. 10.1126/science.1242575. [DOI] [PubMed] [Google Scholar]
  21. Fraser, T. D. , Lynch D. H., Gaiero J., Khosla K., and Dunfield K. E.. 2017. “Quantification of Bacterial Non‐Specific Acid (phoC) and Alkaline (phoD) Phosphatase Genes in Bulk and Rhizosphere Soil From Organically Managed Soybean Fields.” Applied Soil Ecology 111: 48–56. 10.1016/j.apsoil.2016.11.013. [DOI] [Google Scholar]
  22. Friedman, J. , and Alm E. J.. 2012. “Inferring Correlation Networks From Genomic Survey Data.” PLoS Computational Biology 8: e1002687. 10.1371/journal.pcbi.1002687. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Frossard, A. , Gerull L., Mutz M., and Gessner M. O.. 2012. “Disconnect of Microbial Structure and Function: Enzyme Activities and Bacterial Communities in Nascent Stream Corridors.” ISME Journal 6: 680–691. 10.1038/ismej.2011.134. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Galand, P. E. , Pereira O., Hochart C., Auguet J. C., and Debroas D.. 2018. “A Strong Link Between Marine Microbial Community Composition and Function Challenges the Idea of Functional Redundancy.” ISME Journal 12: 2470–2478. 10.1038/s41396-018-0158-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Gebhard, S. , and Cook G. M.. 2008. “Differential Regulation of High‐Affinity Phosphate Transport Systems of Mycobacterium smegmatis : Identification of PhnF, a Repressor of the phnDCE Operon.” Journal of Bacteriology 190: 1335–1343. 10.1128/jb.01764-07. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Grantz, E. M. , Haggard B. E., and Scott J. T.. 2014. “Stoichiometric Imbalance in Rates of Nitrogen and Phosphorus Retention, Storage, and Recycling Can Perpetuate Nitrogen Deficiency in Highly‐Productive Reservoirs.” Limnology and Oceanography 59: 2203–2216. 10.4319/lo.2014.59.6.2203. [DOI] [Google Scholar]
  27. Gregory, L. G. , Karakas‐Sen A., Richardson D. J., and Spiro S.. 2000. “Detection of Genes for Membrane‐Bound Nitrate Reductase in Nitrate‐Respiring Bacteria and in Community DNA.” FEMS Microbiology Letters 183: 275–279. 10.1111/j.1574-6968.2000.tb08971.x. [DOI] [PubMed] [Google Scholar]
  28. Hampel, J. J. , McCarthy M. J., Neudeck M., Bullerjahn G. S., McKay R. M. L., and Newell S. E.. 2019. “Ammonium Recycling Supports Toxic Planktothrix Blooms in Sandusky Bay, Lake Erie: Evidence From Stable Isotope and Metatranscriptome Data.” Harmful Algae 81: 42–52. 10.1016/j.hal.2018.11.011. [DOI] [PubMed] [Google Scholar]
  29. Hao, Z.‐L. , Ali A., Ren Y., Su J.‐F., and Wang Z.. 2022. “A Mechanistic Review on Aerobic Denitrification for Nitrogen Removal in Water Treatment.” Science of the Total Environment 847: 157452. 10.1016/j.scitotenv.2022.157452. [DOI] [PubMed] [Google Scholar]
  30. Harke, M. J. , Davis T. W., Watson S. B., and Gobler C. J.. 2016. “Nutrient‐Controlled Niche Differentiation of Western Lake Erie Cyanobacterial Populations Revealed via Metatranscriptomic Surveys.” Environmental Science & Technology 50: 604–615. 10.1021/acs.est.5b03931. [DOI] [PubMed] [Google Scholar]
  31. Harrell, F. E. 2023. “Hmisc: Harrell Miscellaneous.” R Package Version 5.0‐1. https://github.com/harrelfe/hmisc.
  32. Henry, S. , Bru D., Stres B., Hallet S., and Philippot L.. 2006. “Quantitative Detection of the nosZ Gene, Encoding Nitrous Oxide Reductase, and Comparison of the Abundances of 16S rRNA, narG, nirK, and nosZ Genes in Soils.” Applied and Environmental Microbiology 72: 5181–5189. 10.1128/aem.00231-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Hoffman, D. K. , McCarthy M. J., Zastepa A., Boedecker A. R., Myers J. A., and Newell S. E.. 2024. “Nitrification in the Water Column of Lake Erie: Seasonal Patterns, Community Dynamics, and Competition With Cyanobacterial Harmful Algal Blooms.” Aquatic Ecosystem Health & Management 26: 43–62. [Google Scholar]
  34. Hussain, S. I. , Phillips L. A., Hu Y., et al. 2021. “Differences in Phosphorus Biogeochemistry and Mediating Microorganisms in the Matrix and Macropores of an Agricultural Clay Loam Soil.” Soil Biology and Biochemistry 161: 108365. 10.1016/j.soilbio.2021.108365. [DOI] [Google Scholar]
  35. Ivanova, N. V. , Watson L. C., Comte J., et al. 2019. “Rapid Assessment of Phytoplankton Assemblages Using Next Generation Sequencing—Barcode of Life Database: A Widely Applicable Toolkit to Monitor Biodiversity and Harmful Algal Blooms (HABs).” Preprint, bioRxiv. 10.1101/2019.12.11.873034. [DOI]
  36. Jankowiak, J. , Hattenrath‐Lehmann T., Kramer B. J., Ladds M., and Gobler C. J.. 2019. “Deciphering the Effects of Nitrogen, Phosphorus, and Temperature on Cyanobacterial Bloom Intensification, Diversity, and Toxicity in Western Lake Erie.” Limnology and Oceanography 64: 1347–1370. 10.1002/lno.11120. [DOI] [Google Scholar]
  37. Jetten, M. S. M. 2008. “The Microbial Nitrogen Cycle.” Environmental Microbiology 10: 2903–2909. 10.1111/j.1462-2920.2008.01786.x. [DOI] [PubMed] [Google Scholar]
  38. Jiang, D. , Armour C. R., Hu C., et al. 2019. “Microbiome Multi‐Omics Network Analysis: Statistical Considerations, Limitations, and Opportunities.” Frontiers in Genetics 10: 995. 10.3389/fgene.2019.00995. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Jiang, L. , Yang L., Xiao L., Shi X., Gao G., and Qin B.. 2007. “Quantitative Studies on Phosphorus Transference Occuring Between Microcystis Aeruginosa and Its Attached Bacterium (Pseudomonas sp.).” Hydrobiologia 581: 161–165. 10.1007/s10750-006-0518-0. [DOI] [Google Scholar]
  40. Jones, C. M. , Graf D. R. H., Bru D., Philippot L., and Hallin S.. 2012. “The Unaccounted Yet Abundant Nitrous Oxide‐Reducing Microbial Community: A Potential Nitrous Oxide Sink.” ISME Journal 7: 417–426. 10.1038/ismej.2012.125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Kane, D. D. , Conroy J. D., Peter Richards R., Baker D. B., and Culver D. A.. 2014. “Re‐Eutrophication of Lake Erie: Correlations Between Tributary Nutrient Loads and Phytoplankton Biomass.” Journal of Great Lakes Research 40: 496–501. 10.1016/j.jglr.2014.04.004. [DOI] [Google Scholar]
  42. Kao, N. , Mohamed M., Sorichetti R. J., Niederkorn A., Van Cappellen P., and Parsons C. T.. 2022. “Phosphorus Retention and Transformation in a Dammed Reservoir of the Thames River, Ontario: Impacts on Phosphorus Load and Speciation.” Journal of Great Lakes Research 48: 84–96. 10.1016/j.jglr.2021.11.008. [DOI] [Google Scholar]
  43. Kasalický, V. , Jezbera J., Hahn M. W., and Šimek K.. 2013. “The Diversity of the Limnohabitans Genus, an Important Group of Freshwater Bacterioplankton, by Characterization of 35 Isolated Strains.” PLoS One 8: e58209. 10.1371/journal.pone.0058209. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Kharbush, J. J. , Robinson R. S., and Carter S. J.. 2023. “Patterns in Sources and Forms of Nitrogen in a Large Eutrophic Lake During a Cyanobacterial Harmful Algal Bloom.” Limnology and Oceanography 68: 803–815. 10.1002/lno.12311. [DOI] [Google Scholar]
  45. Knorr, S. , Weisener C. G., and Phillips L. A.. 2023. “The Role of Agricultural Drainage, Storm‐Events, and Natural Filtration on the Biogeochemical Cycling Capacity of Aquatic and Sediment Environments in Lake Erie's Drainage Basin.” Science of the Total Environment 905: 167102. 10.1016/j.scitotenv.2023.167102. [DOI] [PubMed] [Google Scholar]
  46. Knowles, R. 1982. “Denitrification.” Microbiological Reviews 46: 43–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Kormas, K. A. , and Lymperopoulou D. S.. 2013. “Cyanobacterial Toxin Degrading Bacteria: Who Are They?” BioMed Research International 2013: 463894. 10.1155/2013/463894. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Kumar, K. , Mella‐Herrera R. A., and Golden J. W.. 2010. “Cyanobacterial Heterocysts.” Cold Spring Harbor Perspectives in Biology 2: a000315. 10.1101/cshperspect.a000315. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Kurtz, Z. M. , Müller C., Miraldi E., and Bonneau R.. 2022. “SpiecEasi: Sparse Inverse Covariance for Ecological Statistical Inference.” R Package Version 1.1.2.
  50. Kuypers, M. M. M. , Marchant H. K., and Kartal B.. 2018. “The Microbial Nitrogen‐Cycling Network.” Nature Reviews. Microbiology 16: 263–276. 10.1038/nrmicro.2018.9. [DOI] [PubMed] [Google Scholar]
  51. Li, Y. , Yu X., Zheng J., Gong Z., and Xu W.. 2022. “Diversity and Phosphate Solubilizing Characteristics of Cultivable Organophosphorus‐Mineralizing Bacteria in the Sediments of Sancha Lake.” International Journal of Environmental Research and Public Health 19: 2320. 10.3390/ijerph19042320. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Liu, X. , Tiquia S. M., Holguin G., et al. 2003. “Molecular Diversity of Denitrifying Genes in Continental Margin Sediments Within the Oxygen‐Deficient Zone off the Pacific Coast of Mexico.” Applied and Environmental Microbiology 69: 3549–3560. 10.1128/aem.69.6.3549-3560.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Louati, I. , Pascault N., Debroas D., Bernard C., Humbert J. F., and Leloup J.. 2015. “Structural Diversity of Bacterial Communities Associated With Bloom‐Forming Freshwater Cyanobacteria Differs According to the Cyanobacterial Genus.” PLoS One 10: e0140614. 10.1371/journal.pone.0140614. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Lund, M. B. , Smith J. M., and Francis C. A.. 2012. “Diversity, Abundance and Expression of Nitrite Reductase (nirK)‐Like Genes in Marine Thaumarchaea.” ISME Journal 6: 1966–1977. 10.1038/ismej.2012.40. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Maccoux, M. J. , Dove A., Backus S. M., and Dolan D. M.. 2016. “Total and Soluble Reactive Phosphorus Loadings to Lake Erie: A Detailed Accounting by Year, Basin, Country, and Tributary.” Journal of Great Lakes Research 42: 1151–1165. 10.1016/j.jglr.2016.08.005. [DOI] [Google Scholar]
  56. Mackey, K. R. M. , and Paytan A.. 2009. “Phosphorus Cycle.” In Encyclopedia of Microbiology, edited by Schaechter M., 3rd ed., 322–334. Academic Press. [Google Scholar]
  57. Martin, M. 2012. “Cutadapt Removes Adapter Sequences From High‐Throughput Sequencing Reads.” Bioinformatics in Action 17: 10–12. 10.14806/ej.17.1.200. [DOI] [Google Scholar]
  58. Mateo, P. , Berrendero E., Perona E., Loza V., and Whitton B. A.. 2010. “Phosphatase Activities of Cyanobacteria as Indicators of Nutrient Status in a Pyrenees River.” Hydrobiologia 652: 255–268. [Google Scholar]
  59. McMurdie, P. J. , and Holmes S.. 2013. “Phyloseq: An R Package for Reproducible Interactive Analysis and Graphics of Microbiome Census Data.” PLoS One 8: e61217. 10.1371/journal.pone.0061217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Meyer, J. B. , Frapolli M., Keel C., and Maurhofer M.. 2011. “Pyrroloquinoline Quinone Biosynthesis Gene pqqC, a Novel Molecular Marker for Studying the Phylogeny and Diversity of Phosphate‐Solubilizing Pseudomonads.” Applied and Environmental Microbiology 77: 7345–7354. 10.1128/aem.05434-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Mondav, R. , Bertilsson S., Buck M., Langenheder S., Lindström E. S., and Garcia S. L.. 2020. “Streamlined and Abundant Bacterioplankton Thrive in Functional Cohorts.” MSystems 5: e00316‐00320. 10.1128/mSystems.00316-20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Moran, M. A. , Satinsky B., Gifford S. M., et al. 2012. “Sizing up Metatranscriptomics.” ISME Journal 7: 237–243. 10.1038/ismej.2012.94. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Morris, J. J. , Lenski R. E., and Zinser E. R.. 2012. “The Black Queen Hypothesis: Evolution of Dependencies Through Adaptive Gene Loss.” MBio 3: e00036‐12. 10.1128/mbio.00036-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Natwora, K. E. , and Sheik C. S.. 2021. “Assessment of Nitrogen Fixation Rates in the Laurentian Great Lakes.” Journal of Great Lakes Research 47: 1288–1295. 10.1016/j.jglr.2021.07.005. [DOI] [Google Scholar]
  65. Neuenschwander, S. M. , Ghai R., Pernthaler J., and Salcher M. M.. 2018. “Microdiversification in Genome‐Streamlined Ubiquitous Freshwater Actinobacteria.” ISME Journal 12: 185–198. 10.1038/ismej.2017.156. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Newell, S. E. , Davis T. W., Johengen T. H., et al. 2019. “Reduced Forms of Nitrogen Are a Driver of Non‐Nitrogen‐Fixing Harmful Cyanobacterial Blooms and Toxicity in Lake Erie.” Harmful Algae 81: 86–93. 10.1016/j.hal.2018.11.003. [DOI] [PubMed] [Google Scholar]
  67. Newton, R. J. , Jones S. E., Eiler A., McMahon K. D., and Bertilsson S.. 2011. “A Guide to the Natural History of Freshwater Lake Bacteria.” Microbiology and Molecular Biology Reviews 75: 14–49. 10.1128/mmbr.00028-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Ngwa, F. F. , Madramootoo C. A., and Jabaji S.. 2014. “Comparison of Cyanobacterial Microcystin Synthetase (Mcy) E Gene Transcript Levels, mcy E Gene Copies, and Biomass as Indicators of Microcystin Risk Under Laboratory and Field Conditions.” Microbiology 3: 411–425. 10.1002/mbo3.173. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. NLET (National Laboratory for Environmental Testing) . 1997. Schedule of Services. Environment and Climate Change Canada, C.C.f.I.W. NLET (National Laboratory for Environmental Testing). [Google Scholar]
  70. O'Donnell, D. R. , Briland R., Budnik R. R., Ludsin S. A., and Hood J. M.. 2023. “Trends in Lake Erie Phytoplankton Biomass and Community Structure During a 20‐Year Period of Rapid Environmental Change.” Journal of Great Lakes Research 49: 672–684. 10.1016/j.jglr.2022.12.011. [DOI] [Google Scholar]
  71. Oksanen, J. , Blanchet F. G., Friendly M., Kindt R., Legendre P., and McGlinn D.. 2019. “vegan: Community Ecology Package.” R Package Version 2.5‐6. https://CRAN.R‐project.org/package=vegan.
  72. Parveen, B. , Ravet V., Djediat C., et al. 2013. “Bacterial Communities Associated With Microcystis Colonies Differ From Free‐Living Communities Living in the Same Ecosystem.” Environmental Microbiology Reports 5: 716–724. 10.1111/1758-2229.12071. [DOI] [PubMed] [Google Scholar]
  73. Pester, M. , Rattei T., Flechl S., et al. 2011. “amoA‐Based Consensus Phylogeny of Ammonia‐Oxidizing Archaea and Deep Sequencing of amoA Genes from Soils of Four Different Geographic Regions.” Environmental Microbiology 14: 525–539. Portico. 10.1111/j.1462-2920.2011.02666.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Poly, F. , Wertz S., Brothier E., and Degrange V.. 2008. “First Exploration of Nitrobacter Diversity in Soils by a PCR Cloning‐Sequencing Approach Targeting Functional Gene nxrA.” FEMS Microbiology Ecology 63: 132–140. 10.1111/j.1574-6941.2007.00404.x. [DOI] [PubMed] [Google Scholar]
  75. Pound, H. L. , Martin R. M., Sheik C. S., et al. 2021. “Environmental Studies of Cyanobacterial Harmful Algal Blooms Should Include Interactions With the Dynamic Microbiome.” Environmental Science & Technology 55: 12776–12779. 10.1021/acs.est.1c04207. [DOI] [PMC free article] [PubMed] [Google Scholar]
  76. R Core Team . 2018. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. [Google Scholar]
  77. Ragot, S. A. , Kertesz M. A., and Bünemann E. K.. 2015. “phoD Alkaline Phosphatase Gene Diversity in Soil.” Applied and Environmental Microbiology 81: 7281–7289. 10.1128/aem.01823-15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  78. Richards, R. P. , Baker D. B., and Crumrine J. P.. 2009. “Improved Water Quality in Ohio Tributaries to Lake Erie: A Consequence of Conservation Practices.” Journal of Soil and Water Conservation 64: 200–211. [Google Scholar]
  79. Riemann, L. , Rahav E., Passow U., et al. 2022. “Planktonic Aggregates as Hotspots for Heterotrophic Diazotrophy: The Plot Thickens.” Frontiers in Microbiology 13: 875050. 10.3389/fmicb.2022.875050. [DOI] [PMC free article] [PubMed] [Google Scholar]
  80. Rocca, J. D. , Hall E. K., Lennon J. T., et al. 2015. “Relationships Between Protein‐Encoding Gene Abundance and Corresponding Process Are Commonly Assumed Yet Rarely Observed.” ISME Journal 9: 1693–1699. 10.1038/ismej.2014.252. [DOI] [PMC free article] [PubMed] [Google Scholar]
  81. Rösch, C. , Mergel A., and Bothe H.. 2002. “Biodiversity of Denitrifying and Dinitrogen‐Fixing Bacteria in an Acid Forest Soil.” Applied and Environmental Microbiology 68: 3818–3829. 10.1128/aem.68.8.3818-3829.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  82. Rotthauwe, J. H. , Witzel K. P., and Liesack W.. 1997. “The Ammonia Monooxygenase Structural Gene amoA as a Functional Marker: Molecular Fine‐Scale Analysis of Natural Ammonia‐Oxidizing Populations.” Applied and Environmental Microbiology 63, no. 12: 4704–4712. 10.1128/aem.63.12.4704-4712.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
  83. Salcher, M. M. , Neuenschwander S. M., Posch T., and Pernthaler J.. 2015. “The Ecology of Pelagic Freshwater Methylotrophs Assessed by a High‐Resolution Monitoring and Isolation Campaign.” ISME Journal 9: 2442–2453. 10.1038/ismej.2015.55. [DOI] [PMC free article] [PubMed] [Google Scholar]
  84. Salk, K. R. , Bullerjahn G. S., McKay R. M. L., Chaffin J. D., and Ostrom N. E.. 2018. “Nitrogen Cycling in Sandusky Bay, Lake Erie: Oscillations Between Strong and Weak Export and Implications for Harmful Algal Blooms.” Biogeosciences 15: 2891–2907. 10.5194/bg-15-2891-2018. [DOI] [Google Scholar]
  85. Scavia, D. , Bocaniov S. A., Dagnew A., Long C., and Wang Y.‐C.. 2019. “St. Clair‐Detroit River System: Phosphorus Mass Balance and Implications for Lake Erie Load Reduction, Monitoring, and Climate Change.” Journal of Great Lakes Research 45: 40–49. 10.1016/j.jglr.2018.11.008. [DOI] [Google Scholar]
  86. Schindler, D. W. , Hecky R. E., Findlay D. L., et al. 2008. “Eutrophication of Lakes Cannot Be Controlled by Reducing Nitrogen Input: Results of a 37‐Year Whole‐Ecosystem Experiment.” Proceedings of the National Academy of Sciences 105: 11254–11258. 10.1073/pnas.0805108105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  87. Singh, N. K. , Van Meter K. J., and Basu N. B.. 2023. “Widespread Increases in Soluble Phosphorus Concentrations in Streams Across the Transboundary Great Lakes Basin.” Nature Geoscience 16: 893–900. 10.1038/s41561-023-01257-5. [DOI] [Google Scholar]
  88. Smith, D. J. , Berry M. A., Cory R. M., et al. 2022. “Heterotrophic Bacteria Dominate Catalase Expression During Microcystis Blooms.” Applied and Environmental Microbiology 88: e02544‐02521. 10.1128/aem.02544-21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  89. Smith, D. J. , Kharbush J. J., Kersten R. D., and Dick G. J.. 2022. “Uptake of Phytoplankton‐Derived Carbon and Cobalamins by Novel Acidobacteria Genera in Microcystis Blooms Inferred From Metagenomic and Metatranscriptomic Evidence.” Applied and Environmental Microbiology 88: e01803‐01821. 10.1128/aem.01803-21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  90. Song, H. , Xu J., Lavoie M., et al. 2017. “Biological and Chemical Factors Driving the Temporal Distribution of Cyanobacteria and Heterotrophic Bacteria in a Eutrophic Lake (West Lake, China).” Applied Microbiology and Biotechnology 101: 1685–1696. 10.1007/s00253-016-7968-8. [DOI] [PubMed] [Google Scholar]
  91. Sosa, O. A. , Repeta D. J., DeLong E. F., Ashkezari M. D., and Karl D. M.. 2019. “Phosphate‐Limited Ocean Regions Select for Bacterial Populations Enriched in the Carbon‐Phosphorus Lyase Pathway for Phosphonate Degradation.” Environmental Microbiology 21: 2402–2414. 10.1111/1462-2920.14628. [DOI] [PMC free article] [PubMed] [Google Scholar]
  92. Steffen, M. M. , Li Z., Effler T. C., Hauser L. J., Boyer G. L., and Wilhelm S. W.. 2012. “Comparative Metagenomics of Toxic Freshwater Cyanobacteria Bloom Communities on Two Continents.” PLoS One 7: e44002. 10.1371/journal.pone.0044002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  93. Stow, C. A. , Cha Y., Johnson L. T., Confesor R., and Richards R. P.. 2015. “Long‐Term and Seasonal Trend Decomposition of Maumee River Nutrient Inputs to Western Lake Erie.” Environmental Science & Technology 49: 3392–3400. 10.1021/es5062648. [DOI] [PubMed] [Google Scholar]
  94. Sunagawa, S. , Coelho L. P., Chaffron S., et al. 2015. “Structure and Function of the Global Ocean Microbiome.” Science 348: 1261359. 10.1126/science.1261359. [DOI] [PubMed] [Google Scholar]
  95. Wang, K. , Mou X., Cao H., Struewing I., Allen J., and Lu J.. 2021. “Co‐Occurring Microorganisms Regulate the Succession of Cyanobacterial Harmful Algal Blooms.” Environmental Pollution 288: 117682. 10.1016/j.envpol.2021.117682. [DOI] [PMC free article] [PubMed] [Google Scholar]
  96. Watson, S. B. , Miller C., Arhonditsis G., et al. 2016. “The Re‐Eutrophication of Lake Erie: Harmful Algal Blooms and Hypoxia.” Harmful Algae 56: 44–66. 10.1016/j.hal.2016.04.010. [DOI] [PubMed] [Google Scholar]
  97. Welsh, A. , Chee‐Sanford J. C., Connor L. M., Löffler F. E., and Sanford R. A.. 2014. “Refined NrfA Phylogeny Improves PCR‐Based nrfA Gene Detection.” Applied and Environmental Microbiology 80: 2110–2119. 10.1128/aem.03443-13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  98. Wickham, H. 2016. ggplot2: Elegant Graphics for Data Analysis. Springer‐Verlag. [Google Scholar]
  99. Wood, S. N. 2011. “Fast Stable Restricted Maximum Likelihood and Marginal Likelihood Estimation of Semiparametric Generalized Linear Models.” Journal of the Royal Statistical Society, Series B: Statistical Methodology 73: 3–36. 10.1111/j.1467-9868.2010.00749.x. [DOI] [Google Scholar]
  100. Woodhouse, J. N. , Kinsela A. S., Collins R. N., et al. 2016. “Microbial Communities Reflect Temporal Changes in Cyanobacterial Composition in a Shallow Ephemeral Freshwater Lake.” ISME Journal 10: 1337–1351. 10.1038/ismej.2015.218. [DOI] [PMC free article] [PubMed] [Google Scholar]
  101. Wynne, T. T. , and Stumpf R. P.. 2015. “Spatial and Temporal Patterns in the Seasonal Distribution of Toxic Cyanobacteria in Western Lake Erie From 2002–2014.” Toxins 7: 1649–1663. [DOI] [PMC free article] [PubMed] [Google Scholar]
  102. Yancey, C. E. , Mathiesen O., and Dick G. J.. 2023. “Transcriptionally Active Nitrogen Fixation and Biosynthesis of Diverse Secondary Metabolites by Dolichospermum and Aphanizomenon‐Like Cyanobacteria in Western Lake Erie Microcystis Blooms.” Harmful Algae 124: 102408. 10.1016/j.hal.2023.102408. [DOI] [PubMed] [Google Scholar]
  103. Yang, C. , Wang Q., Simon P. N., et al. 2017. “Distinct Network Interactions in Particle‐Associated and Free‐Living Bacterial Communities During a Microcystis aeruginosa Bloom in a Plateau Lake.” Frontiers in Microbiology 8: 1202. 10.3389/fmicb.2017.01202. [DOI] [PMC free article] [PubMed] [Google Scholar]
  104. Yilmaz, P. , Parfrey L. W., Yarza P., et al. 2013. “The SILVA and “All‐Species Living Tree Project (LTP)” Taxonomic Frameworks.” Nucleic Acids Research 42: 643–648. 10.1093/nar/gkt1209. [DOI] [PMC free article] [PubMed] [Google Scholar]
  105. Zastepa, A. , Watson S. B., Kling H., and Kotak B.. 2017. “Spatial and Temporal Patterns in Microcystin Toxins in Lake of the Woods Surface Waters.” Lake and Reservoir Management 33: 433–443. 10.1080/10402381.2017.1384415. [DOI] [Google Scholar]
  106. Zhao, Z. , Baltar F., and Herndl G. J.. 2024. “Decoupling Between the Genetic Potential and the Metabolic Regulation and Expression in Microbial Organic Matter Cleavage Across Microbiomes.” Microbiology Spectrum 12: e0303623. 10.1128/spectrum.03036-23. [DOI] [PMC free article] [PubMed] [Google Scholar]
  107. Zheng, B.‐X. , Hao X.‐L., Ding K., et al. 2017. “Long‐Term Nitrogen Fertilization Decreased the Abundance of Inorganic Phosphate Solubilizing Bacteria in an Alkaline Soil.” Scientific Reports 7: 42284. 10.1038/srep42284. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Table S1: Number of samples collected in each system.

Table S2: Quantitative PCR assay run information.

Table S3: Average (SD in brackets) of the environmental variables for each system across seasons: pH, temperature, total phosphorus (TP), total dissolved phosphorus (TDP), soluble reactive phosphorus (SRP), total nitrogen (TN), total dissolved nitrogen (TDN), ammonia (NH4), nitrite/nitrate (NO x ), total particulate microcystin (TPM), TN:TP and TDN:TDP ratios.

Table S4: Identity of ASVs selected in networks analysis and corresponding to FreshTrain and GenBank sequences (following a BLASTn search), at the lowest taxonomical level.

EMI-27-e70183-s001.docx (79.1KB, docx)

Data Availability Statement

The data that support the findings of this study are openly available in NCBI at https://www.ncbi.nlm.nih.gov/sra/PRJNA877648, reference number PRJNA877648.

Articles from Environmental Microbiology are provided here courtesy of Wiley

RESOURCES