Correspondence: Sandra L. McLellan, Great Lakes WATER Institute, University of Wisconsin-Milwaukee, 600 E. Greenfield Avenue, Milwaukee, WI 53204, USA. Tel.: +1 414 382 1700; fax: +1 414 382 1705; e-mail: email@example.com
The spatial and temporal variability of bacterial communities were determined for the nearshore waters of Lake Michigan, an oligotrophic freshwater inland sea. A freshwater estuary and nearshore sites were compared six times during 2006 using denaturing gradient gel electrophoresis (DGGE). Bacterial composition clustered by individual site and date rather than by depth. Seven 16S rRNA gene clone libraries were constructed, yielding 2717 bacterial sequences. Spatial variability was detected among the DGGE banding patterns and supported by clone library composition. The clone libraries from deep waters and the estuary environment revealed highest overall bacterial diversity. Betaproteobacteria sequence types were the most dominant taxa, comprising 40.2–67.7% of the clone libraries. BAL 47 was the most abundant freshwater cluster of Betaproteobacteria, indicating widespread distribution of this cluster in the nearshore waters of Lake Michigan. Incertae sedis 5 and Oxalobacteraceae sequence types were prevalent in each clone library, displaying more diversity than previously described in other freshwater environments. Among the Oxalobacteraceae sequences, a globally distributed freshwater cluster was determined. The nearshore waters of Lake Michigan are a dynamic environment that experience forces similar to the coastal ocean environment and share common bacterial diversity with other freshwater habitats.
The Laurentian Great Lakes contain c. 17% of the world's surface freshwater and are classified as inland seas due to their inherent physical properties of wave formation, strong currents, and the ability to drive weather patterns (Mortimer, 2004; Rao & Schwab, 2007). Lake Michigan is the second largest of these lakes. Lake Michigan, along with Lake Huron and Lake Superior, is generally considered oligotrophic, based on nutrient status (Munawar & Munawar, 2001; Barbiero et al., 2002). Lake Michigan is a relatively young (c. 10 000 years) and isolated system that has resulted in limited biodiversity at each trophic level (Jude et al., 2005). Currently, the bacterial communities of the Great Lakes are virtually unexplored. Large-scale similarity among the microbial biota of the Great Lakes ranges between 85% and 97% based on DNA reassociation kinetics (Pascoe & Hicks, 2004). Lake Superior was the exception, possessing 44–47% similarity to the other lakes (Pascoe & Hicks, 2004). However, DNA reassociation kinetics only demonstrates the degree of similarity between environmental DNA extracted from the Great Lakes without revealing information about bacterial assemblage identity. The picoplankton community (0.2–1.2 μm) of Lake Michigan is dominated by bacteria, where 91–97.7% of extracted nucleic acids were of bacterial origin (Pascoe & Hicks, 2004). More specifically, the levels of Bacteria and Archaea have been shown to fluctuate seasonally in the water column at an off-shore location (MacGregor et al., 2001; Van Mooy et al., 2001).
To our knowledge, no published studies have explored the taxonomic assemblages in the water column of Lake Michigan, leaving this environment uncharacterized in terms of BCC. It is difficult to link ecosystem processes to bacterial communities. However, understanding the diversity and identity will facilitate the association of specific community members with function and, subsequently, ecosystem processes (Bell et al., 2005). The purpose of this study was to examine bacterial community diversity in the coastal zone of Lake Michigan, creating a basis for further analysis of bacterial community function. Coastal zones of the Great Lakes are dynamic ecological systems influenced by currents, wind patterns, and wave formation processes (Rao & Schwab, 2007). The study sites were located on the western side of Lake Michigan near Milwaukee. Two nearshore open water environments with similar proximity to the Milwaukee harbor were chosen for comparison with intermediate locations inside the harbor. The harbor environment is influenced by discharge from three Milwaukee area rivers that drain 2175 km2 of mixed land use and flow into Lake Michigan. We examined BCC in relation to depth in the water column and across geographically distant locations to gain insight into nearshore temporal and spatial variability during the spring and summer seasons.
Materials and methods
Four sites were chosen for comparison of the microbial community in the nearshore environments of Lake Michigan and within the outer harbor of Milwaukee (Fig. 1). The Linnwood site (43°04.588N, 87°50.311W) is located 2.7 km from the shore on a 20 m contour, northeast of the harbor. The second nearshore site, Green Can (42°59.416N, 87°49.668W) is located southeast of the Milwaukee harbor and is 2.5 km from the shore inside the 20-m contour. Linnwood and Green Can sites are separated by 7.5 km. The third location was the Gap, which is located just inside the Milwaukee harbor (43°01.573N, 87°52.903W). The Junction site is the confluence of the Milwaukee, Kinnickinnic, and Menomonee Rivers. Water parameters were collected with YSI 600XL (YSI Incorporated, Yellow Springs, OH). Profiles of the water column along the 20-m contour (42°59.357N 87°48.954W) were measured using Seabird model 25-01 CTD (Seabird Electronics, Bellevue, WA). Water samples were collected in 2006 on 4 April, 16 May, 12 June, 19 June, 7 August, and 6 September. For the two nearshore locations and the Gap, 50-L water samples were collected at three depths: 1 m below surface, mid-depths, and near bottom using a submersible pump. These samples correspond to Linnwood at 1, 9, and 18 m; Green Can at 1, 6, and 12 m, and Gap at 1 m. Each water sample was sequentially filtered through a 10-, 0.8- (Supor® 800), and 0.2-μm (Supor® 200) filter (293 mm diameter) (Pall Life Sciences, Ann Arbor, MI). Highly turbid samples were also filtered though a 3-μm (Versapor®) filter (Pall Life Sciences). For the Junction site, 200 mL of water was filtered through a 0.45-μm filter (47 mm diameter; Millipore, Billerica, MA). Filters were stored at −80 °C until processed.
Nucleic acids were extracted separately from one-eighth of each 293-mm diameter filter using the DNA isolation method presented by Zhou et al. (1996). The method was adjusted for filtered water samples by cutting and crushing the frozen filter before extraction, serving as the mechanical lysis step. DNA from the Junction samples was isolated using the PowerSoil DNA Isolation kit (MO BIO Laboratories Inc., Carlsbad, CA). Nucleic acids were further purified using either the Wizard DNA clean-up kit (Promega, Madison, WI) or the PowerClean DNA clean-up kit (MO BIO Laboratories Inc.).
Denaturing gradient gel electrophoresis (DGGE) analysis
DGGE was used for examining microbial community composition by location, date, and depth. The DGGE procedures followed those described by Muyzer et al. (1993). DGGE-PCR amplification was carried out using V3-specific primers 341F (40-bp G+C clamp) and universal primer 518R. The template DNA was a combination of equal volumes of nucleic acids isolated from the different filter types in order not to favor bacteria associated with the different sizes. Template DNA used in the PCR reactions ranged from 10 to 20 ng μL−1. PCR was conducted using the Eppendorf Master Taq kit (Eppendorf North America, Westbury, NY) with 1 × buffer, 0.2 mM dNTP mix, 0.15 μM of each primer, 1 ×TaqMaster Enhancer, and 1.5 U Taq polymerase. The PCR conditions were as follows: an initial denaturing step of 5 min at 94 °C, 35 cycles of 45 s at 94 °C, 1 min at 55 °C, 1 min at 72 °C, and an elongation step of 5 min at 72 °C. DGGE gels were cast with denaturing gradients ranging from 25% to 60% (100% concentration of the denaturant is 7 M urea and 40% deionized formamide) using a 7.5% acrylamide solution (acrylamide–bis-acrylamide: 37.5 : 1). Samples were run on large gels of 16 × 24 cm, allowing for greater separation of individual DGGE bands. A bacterial ladder created from five bacteria species, Iodobacteria sp., Cryseobacterium sp., Pseudomonas libanesis, Aeromonas sp., and Clostridium perfringens, was included three to four times in each gel, which allowed for standardization of samples across gels. Gels were run at 200 V for 1 h and at 150 V for 5 h. Gels were stained with 1 : 10 000 dilution of SYBR Green (Cambrex Bio Science, Rockland, ME) for 30 min. Images of DGGE gels were captured with Epi Chemi II Darkroom (UVP, LLC, Upland, CA). The DGGE gel images were analyzed with Bionumerics (Applied Maths, Kortijk, Belgium). DGGE banding pattern similarities were based on the presence or absence of bands used to generate a binary data set from which the Dice coefficients were calculated. Dendrograms were generated using unweighted pair group method with arithmetic means.
16S rRNA gene clone library construction
DNA extracted from the various nearshore environments was amplified with universal primers targeting the entire16S rRNA gene (8F and 1492R) (Crump et al., 1999). PCR products were isolated by gel electrophoresis and then purified with QIAquick PCR purification kit (Qiagen Inc., Valencia, CA). The cloning reaction was carried out with the TOPO TA 2.1 Cloning Kit (Invitrogen Corporation, Carlsbad, CA). Plasmid DNA was isolated using a manual method adapted to a 96-well format (e.g. M. Rise, pers. commun.). Cells were grown overnight in 1.2 mL Luria–Bertani medium with ampicillin (100 mg mL−1). Cells were lysed with a 0.2 N sodium hydroxide and 1% sodium dodecyl sulfate solution. The cell lysates were transferred to an AcroPrep™ 96-well filter plate (3.0 μm GF/0.2 μm BioInert; Pall Life Sciences), which was used to remove cellular debris. The purified plasmid DNA was precipitated with 100% isopropanol and then resuspended in Tris-EDTA buffer (5 mm Tris-HCl, 0.05 mM EDTA, pH 8.5). Sequencing reactions were carried out with the ABI Big Dye Terminator kit (Applied Biosystems, Foster City, CA) using the 8F primer. Clones were sequenced on an ABI Prism 3700xi (Applied Biosystems). Single sequence reads were trimmed for quality using phred (Ewing & Green, 1998), which provided c. 800-bp reads for further analysis.
16S rRNA gene sequence analysis
Sequences were examined using blast algorithm (Altschul et al., 1990) and the Ribosomal Database Project II Classifier (Wang et al., 2007) to identify similar sequences and assess 16S rRNA gene taxonomy. Each clone library was analyzed using dotur to calculate an estimate of bacterial diversity using multiple diversity indices (Schloss & Handelsman, 2005). Sequences with 99% similarity were detected through the use of the cap3 sequence assembly program (Huang & Madan, 1999). This level of stringency was used because nearly identical sequences potentially represent ecotypes or different species (Ferris & Ward, 1997; Eckburg et al., 2005). Sequence alignments were carried out using vector nti (Invitrogen Corporation) and clustal w (Thompson et al., 1994). A neighbor-joining tree for all Betaproteobacteria was used for analysis by unifrac program to characterize bacterial community similarity (Thompson et al., 1994; Lozupone & Knight, 2005; Lozupone et al., 2006). A neighbor-joining tree was created for the dominant Betaproteobacteria operational taxonomic units (OTU) types using clustal w (Thompson et al., 1994) and paup* (Swofford, 2003). A general time-reversible model was chosen for the mrbayes analysis with a proportion of the sites being invariable and the remaining sites gamma-distributed. Four independent Markov chain Monte Carlo analyses, each starting with random trees, were performed for 350 000–2 000 000 generations with sampling every 100 generations. Trees recovered before chain stabilization were discarded with appropriate burn-in values and a 50% majority rule tree was calculated (Huelsenbeck & Ronquist, 2001).
Nucleotide sequence accession numbers
The 16S rRNA gene sequences from this study were deposited in the NCBI GenBank database under accession numbers EU639689–EU642405.
The nearshore zone of Lake Michigan is extremely dynamic, with temperature being one of the most variable parameter measured. The 2006 surface temperatures at the nearshore sites ranged from 3.4 to 22.6 °C (Table 1). The largest temperature shifts were detected between the May and June samplings, with an increase of 9 °C. Water column profiles illustrate that temperature and light transmission varied with depth (Fig. 2). Other parameters were relatively constant in the water column (data not shown). For example, on these dates, at Green Can, dissolved oxygen changed slightly from 10.16 to 10.26 mg L−1 throughout the water column. Specific conductivity ranged from 290.9 to 291.1 μS cm−1 and from 291.4 to 291.9 μS cm−1 during May and June profiles, respectively. Slight fluctuations in pH were detected over the depth of the water column ranging from 8.41 to 8.37 for May and 8.5 to 8.54 for June.
Table 1. Surface water temperatures for 2006 sampling dates
Data from 12 June is incomplete.
Seasonal and depth variation in community composition
Community composition was determined for six dates spanning April to September 2006. DGGE banding patterns were most similar on individual sampling dates within each site based on an overlap in bacterial phylotypes across all depths (Fig. 3), which suggests that the dominant phylotypes are not limited by depth. Stability of BCC was different between the nearshore sites. The Linnwood site had higher similarity (>78%) compared with the Green Can site (c. 58%) over the 2006 sampling events (Fig. 3a and b). The short-term variation (i.e. 7 days between sampling events) in community composition was less for the Linnwood site at all depths than the Green Can site (Fig. 3c), further supporting the stability of dominant phylotypes over time that occurs at Linnwood. The individual open water locations clustered with themselves regardless of depth, suggesting that each site has a distinct distribution of dominant bacterial populations. The bacterial community at the harbor–lake interface (i.e. Gap) appears to reflect elements of both nearshore environments with higher similarity to Linnwood BCC. The Junction site (freshwater estuary) was highly distinct from all other locations analyzed (Fig. 3c), which may be due to the input of riverine bacteria. Slight differences in the DGGE banding pattern of the Junction site may also be associated with the different extraction technique used to isolate DNA from these samples.
Composition of clone libraries
To examine spatial dynamics, we created seven clone libraries from the two nearshore locations at three depths and one from surface water of the Milwaukee harbor from water collected on a single day (19 June 2006). These clone libraries yielded a total of 2717 partial 16S rRNA gene sequences. Clone libraries were assessed using dotur to estimate the total OTUs and evaluate the completeness of each library. The expected number of OTUs at 99% identity was greatest in the Milwaukee harbor and decreased for open water environments of Lake Michigan (Fig. 4). In the open water environments, the highest estimated numbers of OTUs were observed at the greatest depths (e.g. Green Can 12 m and Linnwood 18 m). In the shallower sampling depths, the rarefaction curves began to plateau at 350 clones signifying that the majority of dominant sequences types were captured in these clone libraries. For the harbor location, additional clones are needed to fully describe diversity of dominant bacteria.
The overall composition of dominant taxa was analogous across each clone library regardless of spatial scale. Betaproteobacteria accounted for the largest percentage of sequences ranging from 42.0% to 64.7% (Fig. 5). There was one exception from Green Can 12 m environment where Gammaproteobacteria sequences comprised the majority of sequence types (43.4%) and Betaproteobacteria was second most dominant taxa (40.2%). Bacteroidetes (5.4% to 30.7%) and Alphaproteobacteria (5.1–17.1%) were the next most abundant taxa detected. Multiple minor populations were identified from the following: Firmicutes, OP-10, Epsilonproteobacteria, Fusobacteria, Chloroflexi, Acidobacteria, and Cyanobacteria with little similarity across sites.
A total of 1360 Betaproteobacteria sequences were observed among the seven clone libraries, which yielded 335 different OTUs at 99% identity. The Milwaukee harbor had the greatest richness of 135 Betaproteobacteria OTUs. Richness of OTUs among Betaproteobacteria was reduced by 47–64% in the nearshore clone libraries as compared with the harbor. Similarly, clone libraries created from the deepest waters had the highest number of Betaproteobacteria OTUs, as compared with clone libraries with closer proximity to the surface waters, which supports the estimate of OTUs determined with dotur analysis.
The diversity and distribution of Betaproteobacteria sequences was used in a statistical analysis to compare the seven environments. The relatedness of the various environments was assessed using the unifrac program (Lozupone et al., 2006). A neighbor-joining tree was created from the Betaproteobacteria sequences, which was used for the principal component analysis (Fig. 6). The first three principal components account for 63% of the variation among the samples, suggesting that richness and evenness of Betaproteobacteria is a good proxy for revealing differences between BCC from different environments. The Gap and Green Can 12 m sites were the most distinct environments. The Linnwood 9 and 18 m environments clustered together indicating high similarity in the Betaproteobacteria sequence types detected.
The Lake Michigan sequences were shown to be highly related to others previously identified from freshwater environments. Seven bacterial families associated with Betaproteobacteria were repeatedly detected (Fig. 7). The most numerically abundant sequence types were associated with Comamonadaceae, which accounted for 68.8% of the total Betaproteobacteria sequences. These sequence types were mostly associated with the freshwater clusters of Rhodoferax sp. BAL 47 and GKS-16. Sequences related to the P. necessaries cluster were the next most commonly identified sequence type. The families of Incertae sedis 5 (66 sequences) and Oxalobacteraceae (65 sequences) were detected at a high frequency. However, neither of these families are associated with the eight Betaproteobacteria freshwater clusters, even though both bacterial families are represented in many clone libraries from other freshwater environments (Cottrell et al., 2005; Pearce, 2005; Tamaki et al., 2005; Newton et al., 2006; Eiler & Bertilsson, 2004). Further analyses were carried out to determine if any additional Betaproteobacteria freshwater populations could be determined based on clone abundance in the Lake Michigan libraries in comparison with other freshwater libraries. Clones associated with Incertae sedis 5 did not yield any freshwater-specific clusters that may be a result of conducting the analysis on partial-length 16S rRNA gene sequences. One freshwater-specific cluster, LMich_GC12m, was identified associated within the family of Oxalobacteraceae (Fig. 8). Sequences associated with the LMich_GC12m cluster share 96% identity, and it is composed of sequences from five freshwater environments and one of unknown origin, thus representing a new globally distributed Betaproteobacteria cluster.
This represents the first report characterizing bacterial taxa present in the nearshore coastal environment of Lake Michigan. Our study offers insight into both the spatial and temporal variability in community composition across spring and summer and further identifies the dominance of Betaproteobacteria that may relate to whole-lake circulation processes. The bacterial communities of the nearshore waters are complex assemblages that appear to have similar overall composition, suggesting that community dynamics vary temporally within this seasonal time frame. Other studies have shown rapid changes in terminal-restriction fragment length polymorphism patterns over short periods of time (c. 14–21 days) (Boucher et al., 2006) and high seasonal variability in composition (Yannarell et al., 2003; Moss et al., 2006). Other freshwater bacterial communities have been shown to contain populations that were commonly present over long periods of time and some populations that could only be detected on short time-scales (Yannarell et al., 2003; Lindström et al., 2005; Boucher et al., 2006), which is similar to the consistent presence of dominant phylotypes regardless of sampling date.
Water temperature has been repeatedly suggested as a strong predictor of seasonal fluctuations in BCC (Crump & Hobbie, 2005; Lindström et al., 2005). The seasonal temperature variation in the nearshore environment may be an important factor associated with bacterial community structure over the 2006 sample dates. However, temperature changes alone cannot explain the short-term variation detected in the bacterial community structure across sampling dates. Thermal stratification of Lake Michigan in the nearshore environment generally begins in late April and continues into October (Mortimer, 2004). Temperature in the nearshore environment of Lake Michigan is constantly changing over spring and summer in response to wind-driven currents, whole-lake circulation, and solar radiation (Rao & Schwab, 2007). Real time measurements acquired from the Great Lakes Urban Coastal Observing System confirmed the occurrence of significant vertical stratification with a 10 °C change occurring over a 20-m depth on short time-scales (http://www.glwi.uwm.edu/glucos). Fluctuations in water temperature can occur quickly in the nearshore environment (Consi et al., 2007). Dramatic alterations in nearshore water temperatures are a reflection of downwelling or upwelling of colder offshore waters; both types of events make this environment spatially complex (Mortimer, 2004).
A reoccurring theme, derived from other freshwater bacterial communities, is the distribution of dominant populations not limited by depth. These large-scale clone libraries provided greater information on total bacterial diversity and a way to match community structure data generated from DGGE to actual diversity when characterizing the community. In our study, there was greater spatial variation between sites rather than depth. However, some potential depth-specific taxa (e.g. Acidovorax-, Delftia-, Comamonas-, and Paucibacter-like sequences) were identified in the clone libraries (data not shown). The DGGE technique has been criticized for reducing bacterial diversity to only dominant phylotypes (von Wintzingerode et al., 1997), which may conceal differences among the depth environments due to the dominance by common Betaproteobacteria populations. Crump et al. (2003) demonstrated highest similarity temporally for dominant members of the bacterial community. More importantly, these populations were detected at multiple depths on the same day. Haukka et al. (2005) also demonstrated bacterial communities from different depths in the water column (epilimnion, thermocline, and hypolimnion) clustered together. Konopka et al. (1999) demonstrated that about 60% of dominant DGGE bands were shared across more than one depth. These results suggest that some populations are ubiquitous throughout the water column, similar to the dominance of Betaproteobacteria sequences in the Lake Michigan clone libraries.
In the clone libraries, there was higher total bacterial richness detected with increasing depth at the nearshore sites. Increased number of OTUs may relate to deeper depths containing populations that are specific to those environments. The hypersaline Soda Lake in California had higher expected OTUs at deeper depths (23 and 35 m) than shallower sites (Humayoun et al., 2003), which supports our results. The freshwater estuary environment at the Gap had the greatest bacterial richness, which was based upon the prevalence of singlet sequences. This environment has greater nutrient load than the oligotrophic waters of Lake Michigan and has input from area rivers, urban runoff, and sewage, which may account for the increased level of diversity. These findings are consistent with elevated diversity in other estuarine/coastal systems (Crump et al., 1999; Bernhard et al., 2005).
The Milwaukee harbor has greater bacterial richness at the phylum and subphylum levels than the two nearshore environments. The Junction site is located at the confluence of three rivers that drain c. 2175 km2 area of agricultural, suburban, and urban land uses. This water flows into the Milwaukee harbor and out to the open waters of Lake Michigan. The constantly changing nutrient and sediment load of these waters may play a role in the distinct bacterial community detected at the Junction. As the river water moves into the Milwaukee harbor, it begins to mix with lake water causing the bacterial community to be a mixture of both upstream river water and lake water. This dynamic is reflected in DGGE results (Fig. 3c) and clone libraries (Fig. 5). The transition between rivers and larger bodies of water provides a unique environment where the bacterial community will be reflective of both environments (Crump et al., 1999, 2007). In the Columbia River estuary, half of the free-living bacteria originated from both the ocean and river communities (Crump et al., 1999; Bernhard et al., 2005; Lozupone & Knight, 2007). Gradients of physical or chemical characteristics have been shown to be a major factor in structuring aquatic bacterial communities (Bernhard et al., 2005; Lozupone & Knight, 2007).
Betaproteobacteria comprised 50% of all 16S rRNA gene sequences from the nearshore lake and harbor environments, generating 335 different OTUs (≥99% identity). The Lake Michigan data contained a predominance of sequences that are associated with beta I cluster. Dominance by Betaproteobacteria is not limited to surface waters but also continues with increasing depth in the Lake Michigan water column. Betaproteobacteria are commonly detected in freshwater lakes worldwide as the most numerically dominant cell and clone type (Glöckner et al., 1999; Methé & Zehr, 1999; Zwart et al., 2002) and are the principal group detected among total bacterial cells increasing with depth (Glöckner et al., 1999; Schweitzer et al., 2001; Pearce, 2003). Both biotic and abiotic factors (e.g. dissolved organic carbon, iron, and Chl a concentration) have been statistically correlated to their presence (Methé & Zehr, 1999). The ability of Betaproteobacteria to quickly respond to changes in available nutrients may be linked to their dominance (Newton et al., 2006). The interfaces between riverine and marine systems have shown similar dominance by Betaproteobacteria (Crump et al., 1999; Cottrell et al., 2005). The nutrient-rich waters of the Milwaukee estuary may be important in maintaining higher Betaproteobacteria dominance and diversity.
Currently, freshwater Betaproteobacteria sequences have been classified into eight clusters. Six of the freshwater clusters were determined from a total of 114 sequences from 12 different environments globally distributed (Zwart et al., 2002), which may not represent a complete description of freshwater Betaproteobacteria populations. Following the initial publication, only two additional freshwater Betaproteobacteria clusters associated with Rhodocyclales have been described (Crump & Hobbie, 2005). Crump & Hobbie (2005) suggested that even though other freshwater environments contained similar sequence types, the sequence length and the small number of other sequences associated with these groups prevented their characterization as distinct freshwater clades. Further examination of freshwater Incertae sedis 5 and Oxalobacteraceae sequence types resulted in a single new freshwater-specific population associated with Oxalobacteraceae. The new freshwater cluster, LMich_GC12m, is composed of globally distributed freshwater 16S rRNA gene sequences.
Lake Michigan is a freshwater inland sea possessing properties of the ocean associated with physical transport being a dominant factor in mediating geochemical and biological processes (Rao & Schwab, 2007). Distribution of bacteria in large lakes in relationship to these processes is uncharacterized. In the nearshore environment, average currents transport water along the western coastline in a S/SW direction (Beletsky et al., 1999; Mortimer, 2004; Rao & Schwab, 2007). Across nearshore sites, similarity between bacterial communities was c. 80% for two closely spaced sampling events. Our data corresponded with the levels of variability within an inland lake detected by Yannarell & Triplett (2004) where bacterial community similarity ranged between 82% and 87%. In our study, the most common and dominant sequence types corresponded to the BAL 47 cluster. However, the function of this population is unknown. The distribution of this population may relate to the along shore current that exists for the western coast of Lake Michigan. Circulation patterns result in mixing across the southern basin in a counterclockwise direction (Beletsky et al., 1999). Water movement across the southern basin of Lake Michigan occurs on the order of weeks to months based on previously presented models (Beletsky et al., 2006). Upwelling events have been shown to transport detached benthic algae during an upwelling event to be detected upwards of 15 km offshore (Dettmers et al., 2005). This suggests that it is feasible that dominant bacterial phylotypes could be similarly transported throughout the southern basin of Lake Michigan. Overall, this work represents one of the first characterizations of the bacterioplankton populations in the oligotrophic environment of Lake Michigan. This data will serve as the basis for investigating the biogeochemical role and geographic distribution of these organisms.
We thank Kim Weckerly for providing graphics for the study area and Andrea Zimmerman for laboratory technical support. We gratefully acknowledge the crew of the R/V Neeskay for field support. Data acquisition was partially funded by NOAA Oceans and Human Health Initiative (extramural grant number NA05NOS4781243) and the Department of Defense's Defense Advanced Research Projects Agency (grant number NBCH1050024). Water column profile data were supplied by J.T. Waples, and funded by NSF grant OCE-0351824. Historical data for the Milwaukee harbor and nearshore waters of Lake Michigan were provided by the Great Lakes WATER Institute's Waterbase (http://www.waterbase.uwm.edu).