ABSTRACT
Forced by tides and waves, large volumes of seawater are flushed through the beach daily. Organic material and nutrients in seawater are remineralized and cycled as they pass through the beach. Microorganisms are responsible for most of the biogeochemical cycling in the beach; however, few studies have characterized their diversity in intertidal sands, and little work has characterized the extent to which microbes are transported between different compartments of the beach. The present study uses next-generation massively parallel sequencing to characterize the microbial community present at 49 beaches along the coast of California. In addition, we characterize the transport of microorganisms within intertidal sands using laboratory column experiments. We identified extensive diversity in the beach sands. Nearly 1,000 unique taxa were identified in sands from 10 or more unique beaches, suggesting the existence of a group of “cosmopolitan” sand microorganisms. A biogeographical analysis identified a taxon-distance relationship among the beaches. In addition, sands with similar grain size, organic carbon content, exposed to a similar wave climate, and having the same degree of anthropogenic influence tended to have similar microbial communities. Column experiments identified microbes readily mobilized by seawater infiltrating through unsaturated intertidal sands. The ease with which microbes were mobilized suggests that intertidal sands may represent a reservoir of bacteria that seed the beach aquifer where they may partake in biogeochemical cycling.
INTRODUCTION
Sandy beaches encompass 75% of the world's unfrozen shorelines (1). The permeable nature of sandy beach sediments allows for large quantities of seawater to pass through them over relatively short time scales. As a result, beaches provide important ecosystem services, including seawater filtration and purification (2, 3). Dissolved and particulate organic materials are mineralized as seawater passes through the sands; thus, beaches also play an important role in nutrient cycling (2, 4). Microorganisms present in the lacunar environment between sand grains provide these ecosystem services. However, few studies (described below) have characterized the microbial community in beach sands.
A cross section through the beach reveals regions with diverse chemical and physical characteristics (Fig. 1). Intertidal sands oscillate between being submarine and subaerial and thus experience intermittent tidally driven and wave-driven wetting, drying, heating, and cooling. Subtidal sands, located at the base of the water column, are constantly bathed by seawater and experience relatively constant temperatures compared to intertidal sands. An unsaturated, vadose zone is present beneath subaerial sands. The saturated zone of the beach aquifer, located beneath the vadose zone, consists of three distinct water types: the intertidal saltwater cell, a region of meteoric fresh groundwater, and a deep saltwater wedge (5) (Fig. 1). The interface between these waters has been coined the subterranean estuary due to its steep chemical gradients (6). The intertidal saltwater cell is the region of the beach where seawater filtration occurs. The residence time of water in the cell can vary from days to weeks (5). Depending on whether a beach is dissipative or reflective, between 1 and 100 m3 of seawater/m/day can be flushed through the intertidal saltwater cell via wave and tidal pumping, and wave runup (7). The intertidal saltwater cell can be large in volume at high tide (as depicted in Fig. 1) or small to nonexistent at low tide (5). The transport or dispersal of natural populations of bacteria between these different compartments within the beach has not been studied.
Schematic of beach system. Intertidal sands near the high-tide line were sampled during the 49-beach survey. The intertidal saltwater cell is transient in location and size and is drawn here at its largest extent (5). The inset on the left illustrates the column transport experiments. Filter-sterilized seawater was infiltrated through unsaturated intertidal sand, and effluent was collected to test the potential for bacteria in intertidal sands to be transported into the intertidal saltwater cell.
In the present study, we are interested in one potential bacterial transport pathway within the beach system: between subaerial, intertidal sands and the intertidal saltwater cell (Fig. 1). Transport between these compartments may occur via infiltrating seawater. High tides or wave runup inundate previously unsaturated, intertidal sands and an infiltration front seeps through the vadose zone toward the saturated zone. The infiltration front may scour bacteria from the sands and transport them through the vadose zone to the intertidal saltwater cell (8). Once present in the intertidal saltwater cell, bacteria may become resident there and participate in nutrient cycling or other ecosystem services, or they may fail to thrive. A single study has investigated this pathway previously. Russell et al. (9) focused exclusively on one bacterial genus—Enterococcus—and found that it was transported from subaerial intertidal sands toward the intertidal saltwater cell via infiltrating seawater.
The majority of work characterizing microbial communities at beaches has been carried out in subtidal sands, or in the saturated zone of the beach aquifer. The microbial community present in subtidal sands has been characterized using a variety of techniques, including massively parallel next-generation sequencing, 16S rRNA clone libraries, terminal restriction fragment length polymorphism, automated ribosomal intergenic spacer analysis, denaturing gradient gel electrophoresis, and characterization of enzymatic activity (10–17). These studies identified resident and rare taxa present in diverse geographical locations and in some cases related microbial community diversity to biogeochemical or physical gradients. Several studies have characterized the microbial community present in the saturated zone of the beach aquifer, but these have focused on the presence of specific functional groups of microbes. For example, Santoro et al. (18, 19) investigated the diversity denitrifiers and nitrifiers in the subterranean estuary of a Californian beach aquifer.
Compared to subtidal sands, little research has characterized the microbial community of intertidal sands. The microbial diversity and community structure in intertidal sands near the island of Sylt in the Wadden Sea were examined by using comparative 16S rRNA sequencing to identify the most common bacterial classes (20). Newton et al. (21) and Kostka et al. (22) profiled the microbial communities present in intertidal sands of beaches in the Gulf of Mexico after the Deepwater Horizon oil spill using massively parallel sequencing of the 16S rRNA gene. Cui et al. (23) and Halliday et al. (24) explored bacterial diversity of intertidal sands on the island of Oahu, Hawaii, and a Californian and Massachusetts beach, respectively, also using massively parallel sequencing.
The present study uses massively parallel next generation sequencing to investigate the diversity of microorganisms present in intertidal beach sands at 49 beaches along the California Coast. A biogeographical analysis investigates whether a taxon-distance relationship exists, and whether key beach-specific and sand-specific characteristics can explain the observed differences in taxon assemblages. In addition, we investigated the potential for the infiltration of seawater through intertidal sands to transport bacteria to the intertidal saltwater cell. We investigated whether specific taxa have a propensity to be dispersed via this mechanism.
MATERIALS AND METHODS
Beach sand survey.Sand was collected at 49 California beaches between the Mexico and Oregon borders (see Table S1 in the supplemental material) on four separate outings between 16 and 29 October 2009. The climate in California is Mediterranean, with distinct dry and wet seasons, and sampling was conducted prior to the onset of the rainy season. In the 3 days prior to sampling, coastal counties reported rainfall of <2.5 cm (data not shown and http://cdec.water.ca.gov). Beaches represented a wide range of natural and anthropogenic conditions, including sand grain size, sand organic carbon content, presence of a putative pollution point source (river, creek, or storm drain), surrounding land cover, and degree of shelter from waves (see Table S1 in the supplemental material). At each beach, a sample of dry, exposed, unshaded sand was collected 1 m above the high-tide line. The samples were not within the reach of the tide during collection, but these sites presumably could be inundated during spring tides or large-wave events. Each sand sample was collected by compositing 10 25-cm3 subsamples to obtain a total volume of 250 ml. Sand was homogenized in the lab using sterile spatulas and 50-cm3 aliquots were stored at −80°C until DNA extraction in 2013.
These same sand samples were previously analyzed by Yamahara et al. (25) for bacterial pathogens, including Enterococcus using culture dependent and independent methods. Other measurements, including the moisture content (θm), organic carbon content by mass (Cm), and percent fine grains in the sand by mass, were previously reported by Yamahara et al. (25) using the methods reported therein. The 25th, 50th, and 75th percentiles of percent fines, organic carbon content, and moisture content of the 49 sands were calculated, and then each sample was assigned to its corresponding quartile.
Land cover within a 10-km-radius circular buffer around each beach was determined using the 2001 National Land Cover Data set (U.S. Geological Survey [http://seamless.usgs.gov]) and ARCMAP software (ESRI, Redlands, CA) using methods described by Yamahara et al. (25). A site was defined as “developed” if >50% of the land cover was developed and “undeveloped” otherwise.
Bacterial transport experimental setup.We tested whether bacteria present in sand could be mobilized by a seawater infiltration front, as what may occur when wave uprush and/or the rising tides inundate unsaturated, intertidal sands. Seawater and beach sand were collected from Lovers Point Beach and Cowell Beach on 19 March 2013 using sterile techniques and stored on ice during transport to the laboratory. Seawater from each site was filter sterilized by passing through a 0.1-μm-pore-size filter (Millipore, Billerica, MA) and used immediately. Then, 100 ml of the filtrates for both beaches were then filtered though 0.22-μm-pore-size Durapore filters (Millipore) to collect any microorganisms that may remain in the filter sterilized water for downstream processing and comparative analyses and stored at −80°C. The remaining filtrate was set aside for immediate use in the column infiltration experiments. The sand samples were homogenized in sterile 1-liter glass beakers with sterile spatulas, and a 30-cm3 subsample was placed at −80°C. The homogenized sands were then packed into 30-cm-long, 2.5-cm-diameter polyvinyl chloride (PVC) infiltration columns described previously by Russell et al. (9). Briefly, these columns have a funnel at the top so that seawater can be added to the column and then drain by gravity through the sand and collected at the base of the column. Estimated by volumetric displacement, the pore volumes for Lovers Point sand and Cowell Beach sand columns were 90 and 65 ml, respectively. The porosities (ϕ) were 0.61 (Lovers Point) and 0.44 (Cowell). Three pore volumes of filtered seawater from Lovers Point or Cowell were passed through the packed sand column for the corresponding beach. The effluent was collected: 250 ml for Lovers Point Beach and 127.5 ml for Cowell Beach. The effluent was then filtered through 0.22-μm-pore-size Durapore filters to collect microorganisms, and the filters were stored at −80°C. Sands were removed from their columns after the infiltration experiments and homogenized in sterile 1-liter glass beakers with sterile spatulas. A 30-cm3 subsample was taken from each homogenized sand column and stored at −80°C. A single replicate column experiment was conducted for Lovers Point and Cowell, respectively.
DNA extraction.DNA was extracted from filters by using a MoBio PowerWater DNA isolation kit (MO BIO Laboratories, Inc., Carlsbad, CA) according to the manufacturer's instructions. DNA was eluted in 100 μl of warmed PW6 solution, the elution buffer provided in the extraction kit, and stored at −20°C until use.
DNA was extracted from 10 g of sand using the MoBio PowerMax soil DNA kit. DNA was eluted in 5 ml of Solution C6 and concentrated using the ethanol precipitation recommended in the kit protocol. Briefly, 0.2 ml of 5 M sodium chloride (NaCl) was added to the eluted DNA, followed by 10.4 ml of 100% cold molecular-grade ethanol and inverted three to five times to mix. The mixture was centrifuged at 2,500 × g for 30 min at room temperature. The liquid was decanted, and the pellet was washed in 70% (vol/vol) molecular-grade ethanol. Washed DNA pellets were allowed to air dry to evaporate residual ethanol, and precipitated DNA was resuspended in 50 μl of TE buffer (10 mM Tris, 1 mM EDTA; Ambion/Life Technologies, Grand Island, NY) and stored at −20°C until use. The median concentration of DNA in TE buffer was 28.6 ng/μl (interquartile range, 14.4 to 48.8 ng/μl).
PCR amplification of 16S rRNA gene sequences and pyrosequencing.Bacterial communities from water and sand samples were characterized with pyrosequencing. Indexed amplicon libraries were constructed using fusion PCR primers targeting the V6-V4 hypervariable regions of the bacterial 16S rRNA gene. All sample DNA was diluted 1:10 or 1:50 before being added to the PCRs. Fifty-microliter reactions consisted of 1× Platinum PCR SuperMix high-fidelity master mix (Invitrogen/Life Technologies, Carlsbad, CA), 50 nM concentrations of forward and reverse primers, and 5 μl of template DNA. PCRs were done in triplicate and pooled. The fusion primers contained either A (forward) and B (reverse) 454 Titanium adapters. A unique 11-nucleotide multiplex identifier was included in each forward primer between the adapter sequence and the 16S-specific sequence. The 16S rRNA sequences used to target the V6-V4 region were 518F-CCAGCAGCYGCGGTAAN and 1064R-CGACRRCCATGCANCACCT (21). PCR cycling parameters included an initial denaturation step at 94°C for 2 min, followed by 40 cycles of 94°C for 30 s, 57°C for 45 s, and 68°C for 1 min and a final extension at 68°C for 2 min. Amplicons were purified by using the QIAquick PCR purification kit (Qiagen, Valencia, CA) and pooled in equal molar concentrations for sequencing. Purified amplicon libraries were visualized by using 1.5% agarose gels to ensure the correct amplicon size. Pyrosequencing was performed at the Genome Sequencing and Analysis Core Resource at Duke University using three one-half picotiter plates on a Roche 454 GS-FLX sequencing instrument.
Data analysis.Sequencing reads were processed using the QIIME pipeline version 1.70 (26). High-quality reads with no ambiguous bases, maximum homopolymer runs of 5 bp, no primer mismatches, a minimum average quality score of 25, and minimum/maximum lengths of 200 and 1,000 were retained for analysis. Reverse primers and subsequent nucleotides were removed by using the “truncate_only” option in the “split_library.py” command. Operational taxonomic units (OTU) were assigned by using the QIIME implementation of UCLUST with a cutoff of 97% identity. Based on the most abundant sequences with a cluster, representative OTU sequences were assigned taxonomy using the RDP classifier (27). Sequences were aligned to the GreenGenes core gene set using PyNAST. Reads from the 49 sand samples from along the California coast were rarified to 4,000 reads, which is less than the lowest number of reads obtained for a sample (4,371) to eliminate bias from unequal sample depth (28). Samples from the Lovers Point column experiment were rarefied to 6,000 reads per sample, and the Cowell Beach column experiment samples were rarified to 12,400 reads following the same logic. All analyses that follow were completed with the rarified data sets.
The OTU abundance table for the 49 beach survey samples was imported into Matlab R2009b (Natick, MA) to investigate relationships among samples. In the present study, the term “taxa” is used to describe unique OTU. Taxon abundance was square root transformed prior to analysis. A taxon-distance relationship was investigated by comparing the Bray-Curtis taxon dissimilarity and geographic distance matrices using a nonparametric mantel test. The geographic distance between the samples was calculated using the spherical law of cosines and the latitude and longitude coordinates of each site. An analysis of similarity (ANOSIM) was used to explore the extent to which primary land cover (developed versus undeveloped), the presence of waves, the presence of a source, the presence of high versus low enterococci, and the quartile of percent fines, organic carbon content, and moisture content explained sample taxon similarity. Chao1 and Shannon-Weaver diversity indices were calculated for each beach sample by using OTU level data with PRIMER-E (Ivybridge, United Kingdom) and R Bioconductor packages (29).
Lexical analysis was used to analyze the data from the sand transport column experiment. Words from the taxonomic assignments for sand taxa that were over-represented in taxonomic assignments for water taxa after it passed through the sand were identified using LACK version 4.3 (30) by using the binomial approach with a cutoff of P < 0.05. The LACK program compares a list of taxonomic assignments from a sample to a master list of possible taxonomic assignments (both input by the user). The program statistically determines whether any taxonomic assignments in the master list appear more frequently in the sample list than they would at random. The results of all statistical testing were deemed statistically significant if P was <0.05; some results for P < 0.1 are also presented.
Nucleotide sequence accession number.Sequences from the present study have been deposited in the NCBI Sequence Research Archive (SRA) under accession number SRP037547.
RESULTS
Sand survey.The number of reads that passed quality control varied from 4,371 to 21,935 for each of the 49 beaches (see Table S1 in the supplemental material). In the rarified data set (with 4,000 reads per beach sand sample), 48,543 unique OTU were identified. Of these, 35,625 (73%) were singletons (meaning they appeared one time in the data set) and 5,423 (11%) were doubletons (meaning they appeared twice in the data set), while the remaining OTU appeared more than two times in the data set. Chao1 richness ranged from 2,025 to 20,543 for each of the 49 beaches, while the Shannon-Weaver diversity index varied from 4.1 to 8.1 (see Table S1 in the supplemental material). Henceforth, unique OTU will be referred to as taxa.
Bacteria from 42 phyla were identified in the sand samples (see Table S2 in the supplemental material). Ninety percent of the sequences were classified into six bacterial phyla: Proteobacteria, Bacteroidetes, Firmicutes, Actinobacteria, Planctomycetes, and Acidobacteria. A total of 0.2% of the sequences (n = 375) were classified as archaeal phyla, indicating some cross-reactivity between the bacterial primers and archaeal 16S rRNA genes.
We identified “cosmopolitan taxa” that were present at 10 or more unique beaches. Nearly 1,000 taxa (993 [2%] of the total 48,543 unique taxa) were shared among 10 or more beaches. Two taxa, designated Anaerospora spp. and Gramella spp., were found at each of the 49 beaches. The taxonomic designation of the “cosmopolitan taxa” can be found in Table S2 in the supplemental material to the class level and in Table S3 in the supplemental material to the lowest level provided by the software. A total of 77% of the 48,543 unique taxa were present at one single beach at a frequency of 1 to 72 (out of 4,000) reads at the beach where detected.
The distribution of the 15 most abundant taxa (see Table S4 in the supplemental material) across all of the 49 beach sand samples and their corresponding taxonomic designations, are shown in Fig. 2. These “abundant taxa” are part of the cosmopolitan group described above and together represent 16.4% of the total reads from the 49 beaches. At MA02 (South Beach), abundant taxa represented ca. 50% of the 4,000 reads, while at a few beaches the abundant taxa represented <10% of the 4,000 reads (HUM01[Dry Lagoon], SO02 and SLO04 [San Simeon], MC02 [Andrew Molera], and MC03 [Carmel River]).
Locations of 49 beaches (left panel [see Table S1 in the supplemental material]) and distribution of “abundant taxa” at these 49 beaches (right panel [see Table S2 in the supplemental material for the full taxonomic designation from the software]). The number of reads out of 4,000 is shown on the bottom axis, and the percentage of reads out of 4,000 is given on the top axis. The beaches are listed in order from north (top) to south (bottom). The beach codes can be linked to beach names using Table S1 in the supplemental material. The legend in the lower left gives the taxonomic assignment for each OTU; they are in order from most to least abundant. Note that Xanthomonadales is listed twice in the legend since it was assigned to two unique OTU. The taxonomic designation is the lowest assignment provided by the software. In some cases, the read could be only be assigned to a class, while others could be assigned to a genus. The dashed lines show the relative location of several Californian cities also shown in the map.
Taxon dissimilarity and geographic distances among beaches were correlated (Mantel test, r = 0.25, P < 0.001), indicating that sands collected from beaches close together in space tended to have similar taxa present (Fig. 2). There were no significant correlations between taxon dissimilarity and environmental distances in organic carbon content (Corg), percent fines, and moisture content. However, when the environmental variables were broken into quartiles, the percent fine quartile explained a small, but significant amount of taxon similarity between sites (r = 0.08, P = 0.02, ANOSIM). Similarly, Corg quartile explained a significant amount of taxon similarity in the sand OTU (r = 0.15, P = 0.005, ANOSIM). Note that the percent fines and Corg were not correlated among sand samples.
Each sand sample was designated as having enterococcal densities above and below the median of the distribution from all of the sand samples (2.3 CFU/g). Sands with enterococcal densities above the median tended to have more similar taxa than those with enterococcal densities below the median (r = 0.076, P = 0.02, ANOSIM).
Land cover in a 10-km-radius circle surrounding the beaches where sands were collected was categorized as developed or undeveloped. Beaches that had the same land cover classification tended to have more similar taxa than sites that had different land use classifications (r = 0.22, P = 0.002, ANOSIM). The taxa in sands from beaches with the same wave conditions (i.e., exposed versus sheltered beaches) were more similar to each other than to sites with different wave conditions (r = 0.17, P = 0.1, ANOSIM). The presence of a putative pollution source on the beach was not significant in the ANOSIM analysis.
We tested whether the Chao1 and Shannon-Weaver indices (see Table S1 in the supplemental material) covaried with beach factors. They were not significantly rank correlated to moisture content, organic carbon, or enterococci concentrations. In addition, the indices were not significantly different at beaches with or without waves, putative sources, or surrounded by developed land cover.
Transport of bacteria from intertidal sands.Sand from Lovers Point was collected and placed in a column unaltered. Filter-sterilized seawater from Lovers Point was run through the column, and the effluent was collected at the base of the column. There were a total of 6,108 unique OTU present in the sand (based on the 12,000 rarified reads obtained from sand both before and after water was run through the column).
The filter-sterilized water did not yield any 16S rRNA PCR amplicons (data not shown), and thus no sequences were obtained for the filter-sterilized water. After the filter-sterilized seawater passed through the sand column, it was found to contain 1,467 unique OTU, 537 of which were also present in the sand. The remaining 930 OTU that were found in the water and not in the sand were present primarily as singletons (84% of the 930 taxa). These taxa could represent rare sand taxa that were not detected during the sequencing of the sand microorganisms.
A lexical analysis using LACK identified Lovers Point sand taxa that were over-represented in the list of taxonomic designations of water reads (a list of taxonomic designations and associated P values is provided in Table S5 in the supplemental material). Notably, archaeal reads and bacterial reads from the Firmicutes and Proteobacteria phyla and the Bacilli, BD1m, Chloracidobacteria, Flavobacteria, Gammaproteobacteria, Nitriliruptoria, and Thermoplasmata classes were overrepresented in the list of taxa present in water relative to those present in the sand before water was applied, suggesting that they may have been preferentially mobilized by the wetting front.
The same flowthrough column experiment was repeated with sand and filter-sterilized seawater from Cowell Beach. A total of 7,599 unique OTU were present among the 24,800 rarified reads obtained from sand (considering reads before and after water was applied).
As for Lovers Point, the Cowell Beach filter-sterilized water did not yield any 16S rRNA PCR amplicons. After the filter-sterilized seawater was passed through the sand column, it was found to contain 1,247 unique OTU, 491 of which were also present in the sand. The remaining 756 OTU that were found in the water and not in the sand were present primarily as singletons (70% of the 756 OTU).
A lexical analysis identified Cowell sand taxa that were over-represented in the list of taxonomic designations of water reads (see Table S5 in the supplemental material). Bacterial taxa from the Firmicutes phylum and the Bacilli, Alphaproteobacteria, and Nitriliruptoria classes were overrepresented in the list of taxa present in water (P < 0.05).
Interestingly, some taxonomic designations were over-represented in both Cowell and Lovers Point waters (see Table S5 in the supplemental material) after they passed through their respective sands. These included (i) Firmicutes (phylum), (ii) Bacilli and Nitriliruptoria (classes), (iii) Bacillales, Nitriliruptorales, Oceanospirillales, Pseudomonadales, and Rhodobacterales (orders), (iv) Alteromonadaceae, Bacillaceae, Halomonadaceae, Moraxellaceae, Nitriliruptoraceae, and Rhodobacteraceae (families), and (v) Alteromonas, Gramella, Phaeobacter, and Psychrobacter (genera).
The complete list of taxonomic designations that were over-represented in water relative to sand (see Table S5 in the supplemental material) was compared to the taxonomic designations of the “cosmopolitan taxa” (see Table S3 in the supplemental material) and “abundant taxa” (see Table S4 in the supplemental material) from the California beach sand survey. A total of 49 (83%) of the 59 designations in Table S5 in the supplemental material were shared by the cosmopolitan taxa, and 25 (42%) of the 59 designations in Table S5 in the supplemental material were shared by the abundant taxa.
DISCUSSION
Previous work has explored the diversity of microorganisms primarily in subtidal or submerged permeable marine sediment (10–17), while little work has explored their diversity in intertidal sands (20–24). The present study uncovered extensive diversity in intertidal sands at 49 beaches spanning 1,350 km of the California coast using next-generation 454 sequencing of the V6-V4 hypervariable region of the 16S rRNA gene. Nearly 1,000 OTU (here referred to as taxa) were observed in sands from at least 10 unique beaches, indicating the existence of a group of “cosmopolitan” sand taxa. Several of these taxa were associated with nitrogen cycling, such as Nitrosopumilus spp., ammonia-oxidizing archaea (31); Planctomyces spp., likely capable of anammox (32); Devosia spp., potentially capable of fixing N2 (33); and numerous organisms capable of denitrification, including Pseudomonas stutzeri, Vibrio spp. (34), and Halomonas spp. (34). Other taxa in the cosmopolitan group include salt-tolerant organisms (e.g., Halomonadaceae) (35), common marine (e.g., Flavobacteriaceae) (36) and soil (e.g., Xanthomonadaceae and Verrucomicrobiaceae) (37, 38) bacteria, anoxygenic phototrophs (Erythrobacter spp.) (39), sulfate reducers (e.g., Thermodesulfovibrionaceae) (40), hydrocarbon-degrading bacteria (e.g., Methylibium spp., Planomicrobium spp., and Lutibacterium spp.) (41), stress-tolerant bacteria (e.g., Bacillus spp., Octadecabacter antarcticus, Psychrobacter pacificensis, and Deinococcus spp.) (42–45), and bacteria potentially capable of precipitating calcite to solidify sand (Sporosarcina spp.) (46).
We compared the taxa identified in California beach sands to those identified in other studies of intertidal sands that used next generation sequencing methods. Nitrilirupter, Acidobacterium, Pseudomonas, Pseudoalteromonas, and Paracoccus spp. were found to be the four most abundant bacterial species in intertidal sands of Hawaii (23). Pseudoalteromonas, Pseudomonas, and Paracoccus spp. were among the Californian cosmopolitan taxon group detected in sand from at least 10 unique beaches. Although Nitrilirupter and Acidobacterium spp. were not identified in our cosmopolitan data set, taxa belonging to Nitriliruptoraceae family and Acidobacteria phylum were included in the cosmopolitan data set. In fact, taxa designated as being members of the Nitriliruptoraceae family were detected at 48 of the 49 Californian beaches. Newton et al. (21) report the most common bacterial families in exposed, intertidal sands in the Gulf of Mexico to be Saprospiraceae, Flavobacteriaceae, Planctomycetaceae, and Sinobacteraceae, while the most common genera to be Zeaxanthinibacter (Bacteroidetes), Haliscomenobacter (Bacteroidetes), and Erythrobacter (Alphaproteobacteria). Taxa from the Saprospiraceae, Flavobacteriaceae, and Planctomycetaceae families and the Erythrobacter genus were members of our cosmopolitan taxa, indicating overlap in sand bacterial taxa between the two studies. Kostka et al. (22) report Gammaproteobacteria (class), Alphaproteobacteria (class), and Bacteroidetes (phylum) to be the most abundant bacterial groups in “clean” intertidal sands of St. George Island, FL. Organisms in these groups were characterized as “abundant” and “cosmopolitan” taxa identified in our study. Similarly, abundant phyla reported by Halliday et al. (24) were members of our “cosmopolitan” taxon group.
The number of unique OTU found in each sand sample (between 639 and 2,750) and their diversity (Shannon-Weaver index between 4.1 and 8.1) were similar to those found by Kostka et al. (22) in Florida beach sands. Comparison to other studies employing next-generation sequencing to study microbial ecology of beach sands was difficult because their analyses of richness and diversity do not appear to be done at the OTU level.
A biogeographical analysis suggests a deterministic relationship between the taxa present in beach sands and specific attributes of the beach. In particular, sands having similar grain size, organic carbon content, and exposed to a similar wave climate tended to have similar taxa present. These beach attributes are known to shape the diversity of meiofauna at beaches (7), so it is not surprising that they may also shape microbial diversity. These analyses were completed with a square root transformed abundance matrix. However, the analysis was repeated with just a presence/absence taxon matrix, and the results remained unchanged (data not shown), indicating abundant taxa did not skew results.
A distance-taxon relationship was observed, indicating sand taxon similarity decayed with increasing distance between beaches. Distance-taxon relationships are frequently reported in studies of microbial biogeography and arise from a combination of selection, drift, dispersal, and mutation (47). In our case, a distance-taxon relationship remained even when controlling for variation in environmental characteristics, including percent fines, organic carbon content, and surrounding land cover using partial Mantel tests (data not shown). However, it is possible that an unaccounted-for environmental factor, such as temperature, may account for the observed relationship. Newton et al. (21) investigated a distance-taxon relationship among beach sand taxa at oiled and unoiled beaches in the Gulf of Mexico but did not find a significant relationship between taxon similarity and geographic distance.
The degree of anthropogenic influence on a beach can be approximated by the percentage of surrounding land use that is developed and the concentration of enterococci, a fecal indicator organism, present in the sand. Microbial taxa were more similar among beaches with developed land cover and high concentrations of enterococci compared to those with undeveloped land cover and low concentrations of enterococci, respectively. This may suggest that the beach microbial community is affected to some extent by anthropogenic activities via either dispersal of organisms from anthropogenic activities or selection. Pyrosequencing identified Enterococcus spp. at only two beaches (LA03 and SD02), despite the fact that it was cultured from nearly all of the beach sands. Cui et al. (23) detected Enterococcus spp. in 4 of 16 Hawaiian sand samples using pyrosequencing and in 16 of 16 samples using standard cultivation practices. This indicates that Enterococcus represents a very low proportion of the bacteria present in intertidal sands, and greater sequencing depth may be needed to detect it in each sample (24). Assuming even a low-end estimate of 106 bacteria per g of sand and enterococcal concentrations between 1 and 10 per g, we would need greater sequencing depth than 4,000 to definitively detect these organisms.
The flowthrough column experiments investigated the potential for dispersal of sand taxa from intertidal sands through the vadose zone to the intertidal saltwater cell via infiltrating seawater. The lexical analysis identified words that were over-represented in the taxonomic designations of taxa that were readily dispersed in the flowing seawater relative to those in the sand. Over-representation may indicate that specific types of organisms tend to enter the water phase readily and have high propensity for dispersal. Of the over-represented genera, Alteromonas, Gramella, Phaeobacter, and Psychrobacter were over-represented in the water in experiments conducted with both Lovers Point and Cowell Beach sands. These bacteria have been described as marine, heterotrophic bacteria (Alteromonas and Gramella) (48, 49), surface colonizers from the Roseobacter clade (Phaeobacter) (50), and an extremophile capable of living at cold temperatures (Psychrobacter) (51). Many of the overrepresented designations were shared by “cosmopolitan” and “abundant” taxa from the California beach sand survey, suggesting that many of the common and abundant sand taxa are readily mobilized from and transported through intertidal sands. This suggests the intertidal sands represent a reservoir of bacteria that seed the intertidal saltwater cell, potentially fueling its role as a seawater purifier (2). Additional insight into the transport of different bacteria within beach sands and the beach aquifer, and the propensity for these organisms to form biofilms and not be readily dispersed, may be gained through additional experiments such as these that include biological and technical replicates.
The mechanisms whereby taxa in intertidal sands were mobilized in the column experiments are unknown. Bacteria in intertidal sands can be attached directly to sand grains, potentially in a biofilm, or attached to air-water interfaces or at air-water-grain interfaces. Dispersion by chemical perturbation, thin film expansion during wetting events, air-water interface scouring, and shear mobilization can potentially mobilize bacteria (8). A previous study showed that indigenous Enterococcus present in intertidal beach sands were readily mobilized by seawater infiltrating through the sand toward the intertidal saltwater cell (9). Numerical modeling suggested that air-water interface scouring and thin film expansion were likely responsible for mobilizing Enterococcus spp. within intertidal sands, so this mechanism may be important for mobilizing other organisms as well.
In the present study we used massively parallel 454 sequencing of tagged 16S amplicons to assess the diversity and transport of microbes in intertidal beach sands. We used the V6-V4 region of the 16S rRNA gene since it has been recently used to study bacterial communities in intertidal beach sands (21). However, there are different regions of the 16S rRNA gene that have been targeted by investigators with the goal of improving community coverage and alleviating errors associated with primer biases (52–55). Pyrosequencing errors and artifacts of sequencing inaccuracies could influence the outcome of our analysis (56, 57). However, we attempted to minimize biases and artifacts in our analysis through the quality assurance process in QIIME. By utilizing a stringent sequence quality control, clustering identity assignments and rarefaction of the data sets, we minimized the likelihood of carrying through artifact OTU into our analyses.
Analysis of single samples at single time points during the 49-beach survey only provides a snapshot of the prokaryotic community present and does not allow insight into temporal variability over seasonal or other time scales. Increased sampling and replication may provide a more robust characterization of the microbial community in California beach sands. By increasing sample replication and accounting for biological variability, the use of bioinformatics workflows such as metagenomeSeq (58), edgeR (59), and DESeq (60) will alleviate the need for rarefaction. In addition, future work should explore the possibility of sequencing nucleic acids from active cells (61, 62). The DNA pool that we sequenced for the present study could contain DNA of dead or metabolically inactive cells (63).
This is one of the most extensive studies of bacterial communities present in intertidal beach sands because it covers sands from more than 1,000 km of shoreline. We identified a group of cosmopolitan and abundant taxa that inhabit the unique environment in these sands, where moisture content and temperature vary immensely and organic carbon content is low. Intertidal sands represent one component of the beach system. They are occasionally bathed by seawater, which percolates through the sands before entering the intertidal saltwater cell. As seawater percolates through the intertidal sands, it can both deposit seawater microbes, as well as mobilize microbes already present in the intertidal sands. Through our laboratory column experiments, we found a subset of cosmopolitan and abundant beach sand taxa are readily mobilized from intertidal sands by infiltration seawater potentially scoured from the sand by a propagating air-water interface. These taxa have the potential to be transported to the intertidal saltwater cell where they can either become part of the resident population, participating in nutrient cycling or other ecosystem services, or fail to thrive. Further work examining the microbial communities in all of the different compartments of the beach system will provide more insight into the role of dispersal and selection in shaping the beach microbiome.
ACKNOWLEDGMENTS
We thank Jesse Port and Christina Preston for their comments on the manuscript and Sanjay Mohanty for assistance with art work.
This research was funded by the National Science Foundation (NSF OCE-1129270 and NSF OCE-0910491).
FOOTNOTES
- Received 12 February 2014.
- Accepted 14 April 2014.
- Accepted manuscript posted online 18 April 2014.
- Address correspondence to Alexandria B. Boehm, aboehm{at}stanford.edu.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/AEM.00513-14.
REFERENCES
- Copyright © 2014, American Society for Microbiology. All Rights Reserved.
