Spatial and Temporal Variation in Enterococcal Abundance and Its Relationship to the Microbial Community in Hawaii Beach Sand and Water

Abstract

Recent studies have reported high levels of fecal indicator enterococci in marine beach sand. This study aimed to determine the spatial and temporal variation of enterococcal abundance and to evaluate its relationships with microbial community parameters in Hawaii beach sand and water. Sampling at 23 beaches on the Island of Oahu detected higher levels of enterococci in beach foreshore sand than in beach water on a mass unit basis. Subsequent 8-week consecutive samplings at two selected beaches (Waialae and Kualoa) consistently detected significantly higher levels of enterococci in backshore sand than in foreshore/nearshore sand and beach water. Comparison between the abundance of enterococci and the microbial communities showed that enterococci correlated significantly with total Vibrio in all beach zones but less significantly with total bacterial density and Escherichia coli. Samples from the different zones of Waialae beach were sequenced by 16S rRNA gene pyrosequencing to determine the microbial community structure and diversity. The backshore sand had a significantly more diverse community and contained different major bacterial populations than the other beach zones, which corresponded to the spatial distribution pattern of enterococcal abundance. Taken together, multiple lines of evidence support the possibility of enterococci as autochthonous members of the microbial community in Hawaii beach sand.

INTRODUCTION

Water quality monitoring in marine environments relies primarily on the fecal indicator bacteria (FIB) enterococci, assuming that enterococci in water are mainly of fecal origin and that higher enterococcal concentrations correspond to higher risks of fecal contamination (1, 2). The U.S. Environmental Protection Agency's (EPA's) recreational water quality criteria of 2012 reaffirmed enterococci as the primary FIB for marine environments and recommended a quantitative real-time PCR (qPCR) method for Enterococcus sp. as a rapid alternative approach (3). However, recent studies have detected high levels of enterococci in marine beach sand, with concentrations often being 10- to 100-fold higher than in the corresponding beach waters on a mass unit basis (4–7). Enterococci in beach sand also exhibit a typical spatial distribution pattern across different zones of a beach transect, with the backshore sand (beyond the high-tide mark) containing significantly higher numbers of enterococci (approximately 10-fold more) than the foreshore sand (between the low-tide mark and the high-tide mark) and nearshore sand (submerged) (6–8). The enterococcus-laden beach sands can serve as potential sources of enterococci in beach water, and different levels of interaction between different beach sand zones and beach water may also occur (9). This not only can confound water quality-monitoring efforts (6), but may also have direct public health effects, as shown by the correlation between exposure to enterococcus-laden beach sands and enteric illness (10, 11).

The potential impacts of enterococcus-laden beach sand on beach water quality monitoring warrant further investigation to determine the sources of enterococci in marine beach sand. Fecal contamination, either directly deposited on-site or transported from remote locations through sewage spills and stream runoffs, is conventionally assumed to be the most important source because of the high abundance of enterococci in human and warm-blooded-animal feces. However, recent experimental evidence has supported an alternative scenario where enterococci may also exist as natural members of the indigenous beach sand microbial community. Enterococcus species are found in various natural environments (12–17), and the growth of enterococci in beach sand can be readily stimulated (18). This alternative source of enterococci in beach sand can have significantly different consequences for beach water quality monitoring and public health protection than enterococci from actual fecal contamination.

Previous studies of beach sand have predominantly focused on the abundance and distribution of enterococci (4–7), with limited attention to their relationships with the overall microbial communities (8). Examining the relationships between enterococci and indigenous microbial communities is a promising approach to determine whether autochthonous enterococci represent a significant alternative source in beach sand. The two different sources of enterococci in beach sand (i.e., exogenous fecal sources versus autochthonous populations) can be expected to exhibit different spatial and temporal relationships with the indigenous microbial communities. Understanding the microbial community structure and biodiversity of beach sand is also important to assess the ecological health of the beach sand, as higher biodiversity usually confers better resistance and resilience against environmental perturbation and contamination (19, 20). Indigenous sand microbial communities can affect the growth and die-off of enterococcal cells that are transported and deposited onto beach sand from actual fecal sources (21, 22), while the indigenous enterococcus populations would need to coexist with other members of beach sand microbial communities via various positive and negative interactions.

In this study, we aimed to determine the spatial and temporal variations of enterococcal abundance and its relationships with indigenous microbial communities in different beach sand zones and beach water. The enterococcal abundance in beach sand and water was first investigated through a one-time sampling of 23 different beaches, followed by a weekly sampling schedule over 8 weeks at two selected beaches on the Island of Oahu in Hawaii. Microbial communities of the beach sand and water were characterized using both culture-based enumeration methods and culture-independent 16S rRNA gene pyrosequencing. Several specific bacterial populations (Escherichia coli, Clostridium perfringens, and total Vibrio) and total bacterial density (i.e., heterotrophic plate counts [HPC]) were enumerated. Unique bacterial species and their relative abundances were determined using the bar-coded pyrosequencing of 16S rRNA gene amplicons. The spatial and temporal variation of enterococcal abundance and various microbial community parameters in the different beach sand zones and beach water were compared.

MATERIALS AND METHODS

Sampling sites and sample collection.

Twenty-three beaches located in different geographic regions on the island of O'ahu in Hawai'i were sampled (Fig. 1). Beach sand was collected from the foreshore within the swash zone using a trowel; multiple sand samples were collect from a half-meter-radius circle, pooled, and placed into sterile Whirlpak bags. Beach water was collected in sterile 1-liter bottles at knee depth. Triplicate foreshore sand samples and triplicate beach water samples were collected from three beach transects approximately 30 m apart. Sampling at the 23 beaches was performed in the early morning (7 to 10 a.m.) on four different outings (Table 1). The samples were stored and transported to the laboratory at 4°C in the dark for immediate processing and analysis. The moisture contents of sand samples were determined based on the weight difference before and after oven drying at 105°C.

Fig 1.

Enterococcus concentrations in beach foreshore sand and beach water at 23 marine beaches on Oahu in Hawaii. The color gradient (red to blue) within the symbols reflects enterococcus concentrations in beach sand, while white represents concentrations below the detection limit, The relative abundances of enterococci in beach sand and water on a unit mass basis are represented by a normal slice and an exploded slice of the pie chart, respectively. The beach names are provided in Table 1. (Map reprinted from reference 41.)

Table 1.

Sampling dates and beach names and numbers for the multibeach sampling efforts

Sampling date (November 2011)	Beach name (no.)
9	Waialae (1), Hawaii Kai (2)
15	Kailua (3), Lanikai (4), Waimanalo (5), Kualoa south (8), Kualoa north (9), Kahala (10)
21	Magic Island (6), Sand Island (7), Swarsky (11), Punaluu (12), Hauula (13), Laie (14), Hukilau (15), Pahipahi'ālua (16)
30	Sunset (17), Haleiwa (18), Nanakuli (19), Manners (20), Maili (21), Waimea (22), Mokuleia (23).

Subsequently, two beaches that contained a high incidence of enterococci in beach sand, Waialae beach (Fig. 1, number 1) and Kualoa beach (Fig. 1, number 8), were selected for more extended sampling over eight consecutive weeks. The sampling procedures for beach sand and beach water were the same as described above, except that sand samples were collected from three different zones of each sampling transect. The three different beach sand zones were backshore sand (dry sand beyond the high-tide mark not normally impacted by wave action), foreshore sand (wet sand in the swash zone during sampling), and nearshore sand (submerged sand at knee depth). Waialae beach is impacted by urban land use and stream runoff, while Kualoa beach is affected by rural land use. The eight consecutive weekly samplings occurred between 19 March and 7 May 2012 in the early morning hours (7 to 10 a.m.). All samples were stored and transported to the laboratory at 4°C in the dark for immediate processing and analysis.

Bacterial enumeration.

All beach sand and water samples were analyzed for enterococci. Sand samples were first extracted with deionized (DI) water using a procedure described by Boehm et al. (23) to release bacterial cells from the sand matrix. Briefly, 10 g of sand was mixed with 100 ml of sterile distilled water and shaken by hand, and after the sand settled, the supernatants were removed and used in a standard membrane filtration method for enumeration of enterococci (24). The sand extracts and beach water samples (100 ml) were filtered through sterile GN-6 0.45-μm membranes (Pall Life Science, Port Washington, NY) and incubated at 42°C for 24 h on membrane-Enterococcus indoxyl-β-d-glucoside (mEI) agar for the selective cultivation of enterococci.

The beach sand and water samples collected during the 8-week sampling period were also analyzed for E. coli, C. perfringens, HPC, and total Vibrio. The modified mTEC method was used to enumerate E. coli from beach sand extracts and beach water samples following the standard protocol (25). The mCP agar method was used to enumerate C. perfringens following Hawaii Department of Health standard procedures (26). HPC was determined using tryptic soy agar (TSA) plates; 10-fold serial dilutions of the beach sand extracts or beach water samples were plated, and visible colonies after 24 h of incubation at 37°C were counted as HPC. Total Vibrio was enumerated by incubating membrane filters on thiosulfate-citrate-bile salts-sucrose (TCBS) agar (Difco), and yellow/green colonies were counted as total Vibrio after incubation for 24 h at 37°C (27).

DNA extraction.

Beach sand and water samples collected from Waialae beach on four sampling dates (2, 23, and 30 April 2012 and 7 May 2012) were subjected to total genomic DNA extraction for subsequent pyrosequencing. Bacteria from sand samples (100 g) were first extracted with 100 ml of DI water as described above, and the sand extracts were concentrated by centrifugation at 13,000 × g for 5 min at 4°C. The supernatants were filtered onto sterile GN-6 0.45-μm membranes to capture the remaining bacterial cells. For each sample, the cell pellet from centrifugation and the cell-bearing membrane were pooled for total genomic DNA extraction. For beach water samples, 1 liter of water was filtered directly onto the membrane filters, and the cell-bearing membranes were used in DNA extraction. DNA extraction was conducted using the UltraClean Soil DNA Isolation kit (Mo Bio, Carlsbad, CA) according to the manufacturer's instructions. For each sampling date, DNA extracts from the same type of samples (i.e., backshore sand, foreshore sand, nearshore sand, and beach water) from the three sampling transects were pooled to reduce the pyrosequencing workload, which gave a total of 16 composite DNA samples for the four sampling dates.

Pyrosequencing.

Pyrosequencing of the bar-coded 16S rRNA gene amplicons of the 16 composite DNA samples was performed using bacterial tag-encoded FLX amplicon pyrosequencing (bTEFAP) with 16S rRNA gene primers 104F and 530R for bacteria. The bTEFAP pyrosequencing procedure and subsequent sequence processing were conducted by the Research and Testing Laboratory (Lubbock, TX) using previously described protocols (28). Briefly, the sequence data were first checked to remove sequencing noise using USEARCH and to remove chimeras using UCHIME. To obtain the taxonomic information, the sequence reads were queried against a high-quality 16S rRNA gene database derived from NCBI using BLASTN+ (KrakenBLAST). The BLAST queries generated identity scores to well-characterized 16S rRNA gene sequences, and different threshold identity scores were used for different taxonomic levels (species level, >97% identity; genus level, 95% to 97% identity; family level, 90% to 95% identity; order level, 85% to 90% identity; class level, 80% to 85% identity; and phylum level, 77% to 80% identity).

Data analysis.

Geometric means were calculated to represent bacterial average concentrations. For the plate-counting methods, the absence of one discernible colony was considered below the detection limit, which was mathematically represented by 0.9 for subsequent calculation (such as in log transformation). Microbial abundance in beach sand and water was expressed using CFU/100 g (wet weight) and CFU/100 ml, respectively, which were directly compared on a unit mass basis. Abundance differences among different beach zones were analyzed by analysis of variance (ANOVA), followed by Tukey's post hoc test. Rarefaction curves were calculated using Analytical Rarefaction software version 1.3, available from the UGA Stratigraphy Lab (http://www.uga.edu/strata/software/index.html). For the microbial community diversity indices, the species richness (Chao 1) and species diversity (Shannon-Weaver index) were calculated using the EstimateS software package (http://purl.oclc.org/estimates), and species evenness (Pielou's evenness) was calculated using Primer 6 (29). All diversity indices were plotted in Sigma Plot (the error bars represent standard errors for samples collected at different dates), while statistical significance was determined using paired t tests in Microsoft Excel with a statistical add-in (Statistixl, Australia). Heat maps were drawn in Tableau (Tableau Software, Seattle, WA) using log-transformed species data to prevent the most dominant species from skewing the data. Nonmetric multidimensional scaling (nMDS) plots were calculated from log-transformed species and phylum data. Similarity matrices were then generated using the Bray-Curtis similarity index in Primer 6 and were used to calculate the distances between samples in two-dimensional space (30). The nMDS was performed using 100 random starting configurations of sample points; the accuracy of the nMDS representation was determined by calculating the Kruskal stress (31).

RESULTS

Enterococcal abundances in beach sand and water.

The abundances of enterococci in beach foreshore sand and seawater were first investigated by sampling 23 beaches in various geographic regions of O'ahu in Hawai'i (Fig. 1). Considerable geographic variation was observed; enterococci were detected in 14 of the 23 beach sands (61%) and in 13 of the 23 beach water samples (57%). Among the 11 beaches where enterococci were detected in both beach sand and water, the enterococcal concentrations in sand (2 to 302 CFU/100 g sand) were significantly higher (on a unit mass basis) than those in the corresponding beach water (1 to 98 CFU/100 ml), with enterococcal abundance ratios between sand and water averaging 6.5. The abundance of enterococci appeared to be affected by land use. Out of six beaches with enterococci exceeding 10 CFU/100 g sand, five are located near population centers, and the highest enterococcus level in beach sand and water was observed at Waialae beach, which is impacted by urban land use, while the majority of the beaches (7/9) with no enterococci detected in the beach sand are located in rural regions.

The abundances of enterococci from different beach sand zones (backshore, foreshore, and nearshore sand) and beach water and their temporal variations were determined through sampling over eight consecutive weeks at Waialae beach and Kualoa beach (Fig. 2). Different beach zones contained significantly different levels of enterococci (ANOVA, P < 0.001), and overall, the enterococcal abundance decreased from the backshore sand to the beach water (i.e., backshore sand > foreshore sand ≈ nearshore sand > beach water). At Waialae beach (Fig. 2A), enterococci were detected in 100% of the backshore beach sand samples, 87.5% of the foreshore sand samples, 83.3% of the nearshore sand samples, and 91.7% of the beach water samples over 8 weeks. The average concentration of enterococci in the backshore sand within the 8-week sampling period was 10^2.42 CFU/100 g, which was significantly higher than the enterococcus levels in the foreshore sand (10^1.70 CFU/100 g), the nearshore sand (10^1.67 CFU/100 g), and the beach water (10^1.08 CFU/100 ml) on a unit mass basis. No significant difference in enterococcal abundance was observed between the foreshore sand and the nearshore sand. A similar enterococcal distribution pattern was also observed at Kualoa beach (Fig. 2B), although the enterococcal detection frequencies and the overall enterococcal abundances in the different beach zones were much lower than those observed at Waialae beach.

Fig 2.

Concentrations of enterococci in beach backshore sand, foreshore sand, nearshore sand, and beach water at Waialae beach (A) and Kualoa beach (B) over the 8-week sampling period. The error bars represent the standard deviations of the samples from the triplicate sampling transects.

The enterococcal concentrations in the different types of beach sand and beach water were compared using the data obtained from the multibeach sampling and the 8-week consecutive samplings at the two selected beaches (Table 2). During the 8-week sampling, the backshore sand, despite its high enterococcal concentration, did not show a significant correlation in enterococcal concentration with the beach water. The nearshore sand exhibited a strong correlation in enterococcal concentration with the beach water (r = 0.46; P = 0.07), while the foreshore sand and beach water exhibited the strongest and most significant correlation (r = 0.55; P = 0.03). Strong and significant correlation in enterococcal abundance between beach foreshore sand and beach water was also observed in the multibeach sampling data (r = 0.88; P < 0.001).

Table 2.

Correlation of enterococcus concentrations between different beach sands and beach water^a

Comparison	Pearson's r (P value)
Multibeach sampling (n = 23)
Foreshore sand vs water	0.88 (<0.001)
8-wk sampling (n = 16)
Backshore sand vs water	0.34 (0.19)
Foreshore sand vs water	0.55 (0.03)
Nearshore sand vs water	0.46 (0.07)

^aBacterial concentrations were log transformed.

Enterococci and microbial communities.

The abundances of enterococci in the different beach sand zones and beach water at Waialae beach and Kualoa beach were compared with several microbial community parameters, including two alternative FIB (E. coli and C. perfringens), total bacterial density (HPC), and a group of marine bacteria (total Vibrio) (Fig. 3). The abundances of E. coli in beach sand and water were significantly lower than those of enterococci, which corresponds well to their higher death rates in high-salinity marine environments. Like enterococci, significantly more E. coli bacteria were detected in the backshore sand than in the foreshore sand and nearshore sand (Fig. 3A and B). C. perfringens was only sporadically detected, and its distribution patterns in the different beach zones were difficult to recognize. As expected, the beach sand and water samples contained significantly higher levels of HPC and total Vibrio than enterococci. For example in the Waialae backshore sand, HPC and total Vibrio concentrations were 10^5.4 CFU/100 g and 10^4.5 CFU/100 g, which are 933 and 117 times higher than the concentration of enterococci, respectively (Fig. 3A).

Fig 3.

Box plots showing the concentration range and fluctuation of enterococci, E. coli, C. perfringens, HPC, and total Vibrio in the beach backshore sand, foreshore sand, nearshore sand, and beach water at Waialae beach (A) and Kualoa beach (B) over the 8-week sampling period.

The abundances of enterococci and the microbial community parameters in the backshore sand, foreshore sand, nearshore sand, and beach water were correlated to identify potential relationships (Table 3). The enterococcal concentrations in all types of beach sand showed a strong positive and significant correlation with that of total Vibrio, suggesting possible sharing of habitats or other links between enterococci and total Vibrio in beach sand. Enterococcal concentrations also showed positive correlation with HPC, although a statistically significant correlation (P < 0.05) was observed only in the foreshore sand and the nearshore sand, but not in the backshore sand. The correlation between enterococci and E. coli was significant only in the nearshore sand samples, while the correlation between enterococci and C. perfringens was not significant in all beach sand zones. No significant positive correlation between enterococci and the microbial community parameters was observed in the beach water.

Table 3.

Pearson's correlation coefficients and P values between enterococci and other microbial-community members^a

Sample (n = 48)	Pearson's correlation coefficient (P value)
	Backshore sand	Foreshore sand	Nearshore sand	Beach water
E. coli	0.19 (0.19)	0.16 (0.29)	0.34 (0.02)	0.26 (0.08)
C. perfringens	0.11 (0.45)	−0.24 (0.09)	0.10 (0.50)	−0.08 (0.61)
HPC	0.24 (0.10)	0.40 (0.005)	0.56 (<0.001)	0.23 (0.12)
Total Vibrio	0.56 (<0.001)	0.31 (0.04)	0.69 (<0.001)	0.25 (0.08)

^aBacterial concentrations were log transformed.

Beach microbial community diversity.

The bar-coded 16S rRNA gene pyrosequencing recovered the majority of the microbial community biodiversity in the Waialae beach backshore sand, foreshore sand, nearshore sand, and beach water, as indicated by the rarefaction curves of unique bacterial species approaching an asymptotic stage (Fig. 4). The estimated total species richness in the Waialae backshore sand was 196, while the estimated total species richness in the foreshore sand, the nearshore sand, and the beach water were significantly lower at 135, 143, and 103, respectively. Diversity indices, including Chao 1 (Fig. 5A), the Shannon-Weaver index (Fig. 5B), and Pielou's index (Fig. 5C), were also calculated for the different beach zones based on the four samples collected on different sampling dates. The backshore sand consistently exhibited significantly higher community diversity than the foreshore sand, nearshore sand, and beach water, while no significant difference was observed among the last three, which corresponds well to the enterococcal abundance distribution pattern among the different beach zones.

Fig 4.

Rarefaction curves of the pyrosequencing effort for the Waialae backshore sand, foreshore sand, nearshore sand, and beach water.

Fig 5.

Species richness (A), species diversity (B), and species evenness (C) in samples from the backshore, foreshore, nearshore, and beach water of Waialae beach. The error bars indicate the standard errors of the indices among the four samples collected on different dates.

Pyrosequencing detected a total of 662 unique bacterial species within 24 phyla. Twenty-six of the unique species were identified as major populations, which had relative abundance levels of more than 5% in any of the 16 beach sand/water samples. The remaining species in each of the samples were categorically represented as minor species of that sample. The major and minor species exhibited different distribution patterns among the different beach zones, which is particularly obvious between the backshore sand and other beach zones (Fig. 6). The four most abundant bacterial species in the backshore sand were Nitrilirupter, Acidobacterium, Pseudomonas, and Paracoccus species, which on average accounted for 10.0%, 10.2%, 4.4%, and 3.8% of the microbial communities, respectively, but were present in the other beach zones at much lower abundances. Conversely, the most abundant bacterial species in the foreshore sand, nearshore sand, and beach water (i.e., Pseudoalteromonas species), which on average accounted for 41.8%, 21.4%, and 24.7% of the microbial communities, respectively, was present in the backshore sand at a much lower level (0.83%). The minor species in the backshore sand accounted for 78.8% of the community composition, while the minor species in the foreshore sand (40.8%), nearshore sand (40.8%), and beach water (54.0%) represented much smaller portions of the communities, which agrees with the higher bacterial diversity observed in the backshore sand than in other beach zones (Fig. 4 and 5).

Fig 6.

Heat map of the major bacterial populations and the minor populations (grouped as one entry) detected in the different beach zones (BS, backshore; FS, foreshore; NS, nearshore; BW, beach water) on the different sampling dates. Dark blue (−4.0 on the color scale bar) indicates no detection.

The pyrosequencing effort detected Enterococcus species in 4 of the 16 samples, while neither E. coli nor C. perfringens was detected, corroborating the comparative abundances of the three FIB determined by the membrane filtration methods. The Enterococcus species detected included E. pallens, E. hermanniensis, and E. malodoratus, but not E. faecalis or E. facium, which are usually expected from fecal contamination due to their predominant presence in warm-blooded-animal feces. The highest relative abundance of Enterococcus species detected by pyrosequencing was 0.51%, and the relative abundance of Enterococcus species detected through pyrosequencing showed no correlation with the enterococcal abundances by cultivation.

The unique bacterial species and their relative abundances in the individual samples detected by pyrosequencing were compared by nMDS (Fig. 7). Samples obtained over time showed different levels of clustering according to the sampling locations and showed different levels of temporal variation. All four backshore samples consistently clustered close together, indicating good temporal stability in the backshore sand microbial community. Microbial communities in the foreshore sand, the nearshore sand, and the beach water were more scattered. The microbial communities of the beach water appeared to cluster more with the microbial communities in the foreshore sand and the nearshore sand samples than those in the backshore sand samples.

Fig 7.

nMDS plot based on unique bacterial species and their relative abundances. The symbol shapes indicate different beach zones, and the same shading indicates the same sampling date.

DISCUSSION

Cases of higher levels of enterococci in marine beach sand than in beach water have been widely reported in temperate and subtropical regions (4–7), including two beaches in Hawaii (5, 7). The present study confirmed these observations through a more extensive sampling effort involving 23 beaches from different geographic regions on the island of O'ahu in Hawai'i. Generally, beach sands containing high numbers of enterococci were more likely to be located in regions impacted by urban land uses, while the beaches in rural areas were less likely to have enterococci in either beach sand or water. Urban beaches are more likely to receive fecal contamination from municipal sewage spills, urban storm runoffs, and on-site animal defecation. In addition, urban beaches tend to receive more organic- and inorganic-nutrient discharges than beaches in rural areas (32, 33), which could support higher densities of microbial communities and hence larger enterococcal populations if they exist as autochthonous members of the microbial communities.

Since enterococci in beach sand can affect water quality monitoring and public health (6, 10, 11), it is necessary to adequately understand the source(s) of enterococci in beach sand. The most obvious and best-recognized source of enterococci is point and/or nonpoint fecal pollution, which can be transported through wastewater effluent discharge and storm runoffs and deposited in beach sand by wave action and tidal wetting (34). Another potential source of enterococci could be the autochthonous enterococcal populations in beach sand, as numerous previous studies have reported the persistence and even growth of enterococci in beach sand (18, 35, 36).

In this study, we observed a gradual decrease in enterococcal abundance along the beach transect at Waialae and Kualoa beaches (i.e., backshore > foreshore ≈ nearshore > beach water), which provided a good opportunity to determine whether autochthonous enterococcal populations are significant contributors to the enterococci detected in beach sand. Similar local distribution pattern of enterococci in beach sand were observed in earlier studies of Hawaii beaches (5, 7) and in other subtropical beaches in California and Florida (6, 8). First, the relatively small variation of enterococcal concentrations observed in the same types of sand samples from the different sampling transects precludes direct fecal deposition as a potential source. Although the deposition of enterococci from fecal sources by wave action or tidal washing could not be ruled out by the distribution pattern, as the deposited enterococcus cells could have experienced differential die-offs to result in the observed abundance differences in the different beach zones (21), this scenario is challenged by the relatively stable enterococcal population sizes over the 8-week sampling period. On the other hand, the detected enterococci being largely autochthonous would more reasonably explain the observed distribution pattern. For example, in the absence of fecal pollution, autochthonous enterococcal populations in the beach foreshore and nearshore sand were transported into and diluted by the beach water, which left a larger enterococcal population in the backshore sand.

The lack of a strong correlation in enterococcal concentrations between the beach water and the beach backshore sand (Table 2) confirms that the enterococci in the two beach zones lacked significant exchange, reflecting the fact that the backshore sand was not normally directly impacted by beach water. This lack of impact from beach water coupled with the temporal stability and high abundance of enterococci in beach backshore sand further supports the possibility of autochthonous enterococcal populations. Since beach foreshore sand and nearshore sand are physically closer to and therefore undergo more frequent interactions with beach water (6), it is not surprising that strong and significant correlations in enterococcal concentration were observed between the foreshore/nearshore sands and beach water (9).

Additional evidence supporting this alternative, autochthonous source of enterococci in beach sand were obtained from various microbial community parameters and their relationship with enterococcal abundance. The high enterococcal abundance in the backshore sand did not appear to be simply the outcome of high overall bacterial density. The backshore sand, despite its higher enterococcal abundance, did not contain a significantly higher number of total bacteria than the other beach zones (Fig. 3). This is also illustrated by the different correlation coefficients between enterococci and HPC in the different types of beach sand; significant correlations were observed only in the foreshore sand and the nearshore sand, but not in the backshore sand (Table 3). The lack of correlation between enterococci and HPC in the backshore sand suggests complex ecological interactions within the microbial community. This is consistent with a 2012 study by Piggot et al. that observed a nonlinear, unimodal relationship between enterococci and extracellular polymeric substances from biofilms in South Florida backshore beach sand (8). The strong and significant correlation between enterococci and total Vibrio abundances in beach sand in all beach zones suggests that enterococci may share habitats with Vibrio species, which are known autochthonous populations in marine environments (37, 38). Although the highest concentrations of E. coli bacteria were also observed in the beach backshore sand (Fig. 3), no significant correlation between enterococci and E. coli was observed (Table 3). This could be attributed to the different survivabilities of E. coli and enterococci, especially in the high-salinity marine environments (39). Furthermore, E. coli itself has also been recognized as a member of indigenous microbial communities in various environments (40–42).

Sand microbial community structures of the different beach zones of Waialae beach showed considerable differences based on pyrosequencing of bar-coded 16S rRNA gene amplicons. Pyrosequencing has recently emerged as a cost-effective approach to resolving microbial community structure, with adequate coverage of the community biodiversity (43, 44). Due to its ability to generate a large number of sequence reads, both the major and the minor microbial populations can be detected, and its use has been demonstrated in studies on the human gut microbiota (43) and soil microbiomes (44). The sequence reads obtained for samples from the different beach zones in this study were large enough for the rarefaction curves to approach asymptote (Fig. 4), enabling reasonable estimation of overall species richness. However, it should be noted that the sequencing depths achieved here were still not adequate to detect minor populations in the microbial communities with high frequencies. For example, although the 16 samples used for pyrosequencing all contained enterococci based on cultivation, Enterococcus was detected in only four of them by pyrosequencing. This can be explained by the relatively small number of enterococci in the community; for example, in the Waialae backshore sand, the ratio of HPC over enterococci was 933, and the total bacterial density (culturable plus unculturable) can be expected to be significantly higher.

The backshore sand of Waialae beach exhibited significantly higher species richness (Chao 1), species diversity (Shannon-Weaver index), and species evenness (Pielou's index) than the other beach zones, which corresponds to the observed enterococcal distribution pattern. The other beach zones exhibited no significant differences in terms of species richness, diversity, or evenness (Fig. 5), which also coincides with the enterococcal abundance pattern. Like other ecosystems, the higher biodiversity in the beach backshore sand can be expected to confer better resistance and resilience against environmental perturbation and contamination (19, 20), which was supported by the high stability of microbial communities in the beach backshore sand (Fig. 6 and 7).

The backshore sand at Waialae beach also differed significantly from the other beach zones in respect to dominant bacterial species (Fig. 6). A previous study at South Florida beaches reported a similar difference using terminal restriction fragment length polymorphism (T-RFLP) analysis of small cloned 16S rRNA gene libraries without sequence information (8). In this study, the pyrosequencing of 16S rRNA gene amplicons not only identified unique bacterial species, but also determined their relative abundances based on the detection frequencies (45). The major bacterial species detected in the backshore beach sand of Waialae beach were species commonly found in soil environments, including Acidobacterium (46), Pseudomonas (47), and Paracoccus (48) species, while the most dominant bacterial species detected in the foreshore sand, nearshore sand, and beach water were Pseudoalteromonas spp., which are primarily marine bacteria (49, 50). The enterococci detected by pyrosequencing in the beach sands were not E. faecalis or E. faecium, which are dominant in warm-blooded-animal feces (51, 52).

The significant difference in microbial communities among the different zones of Waialae beach was also clearly indicated by the community clustering (nMDS plot) based on the detected microbial species and their relative abundances (Fig. 7). The beach backshore sand samples that were collected over a period spanning 4 weeks were tightly clustered together and clustered away from the samples from other beach zones. Microbial communities in the foreshore sand, nearshore sand, and beach water samples were much more scattered, reflecting higher spatial and temporal fluctuation. The foreshore and nearshore sand samples were connected and were clustered away from the microbial communities in beach water samples. One exception was the samples collected on 30 April 2012, when the microbial community of the foreshore/nearshore sand samples clustered with those of backshore sand samples. Significant wave action was recorded on that sampling date, which might have promoted the interactions among the different beach sand zones.

Taken together, the present study presents multiple lines of evidence supporting the possibility of enterococci as autochthonous populations of the beach sand microbial community, which could represent a significant nonfecal source of enterococci in beach sand. This highlights the importance of beach sand in water quality monitoring and protection, as the presence of a high abundance of autochthonous enterococcal populations in beach sand can potentially invalidate enterococcus-based monitoring efforts at marine beaches. Future research is needed to elucidate the environmental and ecological factors controlling the size of the autochthonous enterococcal population in beach sand, which is also important for an understanding of the survival behavior of fecal enterococci and fecal pathogens that are introduced into beach sand during actual fecal pollution.