Canadian Journal of Microbiology

Groundwater microbial diversity and antibiotic resistance linked to human population density in Yucatan Peninsula, Mexico

Journal: Canadian Journal of Microbiology

Manuscript ID cjm-2019-0173.R2

Manuscript Type: Article

Date Submitted by the 24-Sep-2019 Author:

Complete List of Authors: Moore, Anni; Morningside College, Biology and Chemistry Lenczewski, Melissa; Northern Illinois University, Geology and Environmental Geosciences Leal-Bautista,Draft Rosa Maria; Centro de Investigación Científica de Yucatán Duvall, Melvin; Northern Illinois University, Biological Sciences

Keyword: antibiotic resistance, karst, groundwater, microbial communities

Is the invited manuscript for consideration in a Special Not applicable (regular submission) Issue? :

https://mc06.manuscriptcentral.com/cjm-pubs Page 1 of 50 Canadian Journal of Microbiology

1

1 Groundwater microbial diversity and antibiotic resistance linked to human population

2 density in Yucatan Peninsula, Mexico

3

4

5 Anni Moore,a* Melissa Lenczewskib#, Rosa Maria Leal-Bautistac, and Melvin Duvalla,d

6

7 Department of Biological Science, Northern Illinois University, DeKalb, IL USAa

8 Department of Geology and Environmental Geosciences, Northern Illinois University,

9 DeKalb, IL, USAb

10 Water Science Unit, Centro de Investigacion Cientifica de Yucatan, Cancun, QR, Mexicoc

11 Institute for the Study of the Environment,Draft Sustainability, and Energy, Northern Illinois

12 University, DeKalb, IL USAd

13 # Corresponding Author

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 2 of 50

2

14 ABSTRACT

15 Microbial community composition in selected karst groundwater sites in the Yucatan

16 Peninsula, Mexico was assessed to determine the environmental variables influencing

17 groundwater microbial diversity. The karst aquifer system is a groundwater dependent

18 ecosystem and is the world’s second largest underwater karst cave system. The area’s

19 geology allows precipitation to infiltrate into the groundwater system and prevents

20 accumulation of surface water; as such, groundwater is the only source of fresh water on the

21 peninsula. The sampling locations consisted of three karst sinkholes that extend through the

22 freshwater zone into the saline water, and an abandoned drinking water well of an ocean-side

23 resort during the dry and rainy seasons. The analysis showed that highly diverse microbial

24 communities are present in the YucatanDraft groundwater, sustained by permanently warm

25 temperatures and high nutrient input from human activity. Proximity to densely populated

26 areas, such as tourist resorts, is the most important factor influencing both diversity, presence

27 of fecal , and antibiotic resistance profile.

28

29 Key Words: microbial communities, karst, groundwater, antibiotic resistance

30

31

https://mc06.manuscriptcentral.com/cjm-pubs Page 3 of 50 Canadian Journal of Microbiology

3

32 INTRODUCTION

33

34 Yucatan Peninsula, Mexico, lacks surface water, and is completely dependent on groundwater

35 for its freshwater needs. At the same time, the Yucatan aquifer is also extremely vulnerable to

36 pollution due to the karst geology and the increased population and tourism that often lacks

37 adequate sewage treatment and disposal options. As a result, the fresh groundwater in most

38 parts of the peninsula is now unsuitable for human consumption.

39

40 The geological features that contribute to the vulnerability of the aquifer are the highly

41 permeable bedrock and inadequate soil covering. The bedrock consists of karstified soft and

42 porous marine carbonate rocks, formingDraft a highly cavernous terrain (Perry et al. 2009). Its

43 permeability is enhanced further by extensive fracturing of the terrain, allowing saltwater flow

44 into the cave system and resulting in subsurface erosion, collapse of the cave roofs, and

45 formation of an extensive network of caves and karst sinkholes (Perry et al. 2009; Gines and

46 Gines 2007). The mean depth of the soil cover of the peninsula is 7.2 cm (White and Hood

47 2004), which allows water at the surface to infiltrate quickly and directly into the aquifer

48 (Perry et al. 2003). The lack of sufficient soil cover together with the karst cannot provide

49 sufficient filtering of contaminants (chemical or biological) into the aquifer.

50

51 The aquifer of the peninsula is characterized by a thin freshwater lens overlaying the

52 seawater-derived saline layer (Lefticaru et al. 2006). This permanent stratification is sustained

53 due to the density difference between the fresh and saline waters, and a narrow mixing zone

54 (interface) exists between the two. The thickness of the freshwater layer varies throughout the

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 4 of 50

4

55 peninsula from less than 10 m near the coast to over 60 m in the center of the peninsula. The

56 aquifer can easily be accessed through the karst sinkholes locally called cenotes that penetrate

57 deep into the saline layer of the aquifer.

58

59 The main threat to the aquifer comes from human activities. While the lack of large-scale

60 industrial development has kept the aquifer free from various industrial pollutants, the aquifer

61 of Yucatan Peninsula is threatened mainly by municipal human waste. This has become an

62 especially acute problem since the planned development of major tourist destinations on the

63 Caribbean coast caused the population of eastern state of Quintana Roo to explode from

64 88,150 in 1970 (Instituto Nacional de Estadística, Geografía e Informática 2001) to 1.326

65 million in 2010 (Instituto Nacional de DraftEstadística y Geografía 2013) with 17 million tourists

66 in 2017 (Instituto Nacional de Estadística y Geografía 2019). While this kind of planned

67 development has economically benefited the local population, the development of adequate

68 wastewater treatment has not kept up with the development of tourist resorts.

69

70 As a result, the freshwater aquifer, which is the only source of drinking water for the

71 peninsula, is most notably contaminated with fecal bacteria, which has great health and

72 economic impacts on the population (Pacheco et al. 2000). While the aquifer is well

73 characterized geochemically, no studies have looked at the native groundwater microbial

74 communities, their response to pollution, and their potential to deal with the high nutrient load

75 introduced into the aquifer. This information will provide both an overview of the

76 biogeochemical interactions in the aquifer and should help improve groundwater

77 management.

https://mc06.manuscriptcentral.com/cjm-pubs Page 5 of 50 Canadian Journal of Microbiology

5

78

79 This study aims to investigate the overall microbial diversity and genomic potential in the

80 groundwater, to determine which environmental variables drive the overall diversity, how

81 these factors influence the presence of fecal indicator bacteria, pathogens, cyanobacteria, and,

82 at the genomic level, antimicrobial and heavy metal resistance, and bacterial degradation of

83 aromatic hydrocarbons.

84

85

86 Draft

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 6 of 50

6

87 MATERIALS AND METHODS:

88

89 Sampling sites

90

91 Three cenotes (Cenote X, U, and C) and an abandoned drinking water well were sampled in

92 the Yucatan Peninsula (Figure 1 and Table 1). Cenote X is located in a remote area,

93 surrounded by woods and agricultural fields. The freshwater layer of Cenote X is

94 characterized by eutrophic growth during the hot, rainy season, and turnover event in October

95 or November when the warm surface layer of water begins to cool, becomes denser, and

96 sinks, bringing the clear freshwater up from deeper layers. The dry season is characterized by

97 relatively clear freshwater layer. The edgeDraft of the cenote is lined by trees, allowing vegetation

98 input. Cenote U is also located in a remote area, surrounded by woods and agricultural fields,

99 and it serves as a water source for the surrounding farms. Cenote U is in a partially collapsed

100 cave, limiting the vegetation input. Cenote U is characterized by clear water throughout the

101 year. Cenote C is located at the closed territory of a limestone quarry without public access to

102 the water. The nearest resort and the coast are about 3 km distant, and the nearest town about

103 8 km east of the cenote. The groundwater flow is towards the sea, and there are no towns

104 located within the 20 km ‘upstream’ from the cenote, although there are a few single houses

105 about 10-15 km west and northwest of the site. Cenote C is partially covered with a limestone

106 ledge (uncollapsed part of the former cave roof), limiting the vegetation input. Cenote C is

107 characterized by exceptionally clear water during both dry and rainy seasons.

108

https://mc06.manuscriptcentral.com/cjm-pubs Page 7 of 50 Canadian Journal of Microbiology

7

109 The abandoned well is located on the campus of a busy tourist resort and was abandoned

110 because of saline water intrusion. The well is located about 0.5 km from the coast. It is

111 completely covered with no light or vegetation input, and no public access. The resort

112 operates its own moving bed biofilm reactor type wastewater treatment facility. This

113 technology is compact and convenient to operate, although its efficiency of nutrient removal

114 (COD, BOD, phosphate and nitrate) is half of the fixed-bed bioreactor (Choi, Lee, and Lee

115 2012). The volume of sewage handled by this specific facility and the output volume are not

116 available. The effluent of these treatment facilities (owned by the hotel group in multiple

117 tropical locations globally) has traditionally been used by that hotel chain for irrigation

118 purposes (50%) and deposited into injection wells (50%) (National Environment and Planning

119 Agency 2008). Draft

120

121 Water Sampling

122

123 Water samples were obtained from the four sites during two dry seasons (December 2008 or

124 March 2009), and one rainy season (July 2009). Cenote U was sampled only during the one

125 rainy season, once. Prior to each sampling, the geochemical profiles of the cenotes were

126 obtained using the Hydrolab Multiparameter Sonde MS5 45922 (Pedersen 2007). There is

127 very little seasonal variation in the profiles.

128

129 The water samples were obtained either by using peristaltic pump (cenotes X and U, and the

130 hotel well), or a Niskin Bottle (Cenote C). The samples were obtained from three depths

131 (freshwater, interface and saline layers) from the cenotes, and from freshwater and interface

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 8 of 50

8

132 layers in the hotel well that does not have an established saline layer (Table 1; Figure 2). The

133 sediment samples were obtained from the cenotes X and C; the hotel well and Cenote U did

134 not yield any sediment. The water was filtered through 0.2 µm pore size and 47 mm diameter

135 Millipore IsoporeTM polycarbonate membrane filters. The amount of water filtered varied

136 between the samples, ranging from 0.2 to 10 liters, depending on the amount of particulate

137 matter in the water. Because of the heavy algal growth, the rainy season fresh water sample

138 from Cenote X was pre-filtered using glass fiber pre-filters (SartoriusTM). The filters were

139 placed in sterile Petri dishes and frozen until DNA extraction.

140

 141 DNA was extracted from the filters usingDraft MoBIO UltraClean Soil DNA Isolation Kit 142 (currently Qiagen, Germantown, MD). A quarter of each filter was used for an initial

143 extraction; for most sites additional extractions were necessary to obtain enough DNA for

144 metagenomic analysis. DNA from the sediment samples was also extracted using the MoBIO

145 UltraClean Soil DNA Isolation Kit, using 0.5 g of the sediment.

146

147 Sequencing and data assembly

148

149 For the taxonomic classification of the communities, the 16S ribosomal rRNA V4 variable

150 region was sequenced at Argonne National Laboratories, Lemont, IL, USA in collaboration

151 with the Earth Microbiome Project (www.earthmicrobiome.org). The region was amplified

152 using 515F (GTG CCA GCM GCC GCG GTA A) and 806R (GGACTACHVGGGTWTCTAAT)

153 primers. All twenty-one samples were sequenced on the paired-end Illumina (San Diego, CA,

154 USA) MiSeq platform. The 16S rRNA V4 data were mapped to a reference sequence database

https://mc06.manuscriptcentral.com/cjm-pubs Page 9 of 50 Canadian Journal of Microbiology

9

155 using QIIME pipeline (Caporaso et al. 2010). The operational taxonomic units (OTUs) were

156 picked with a closed-reference OTU picking protocol against the Greengenes database

157 (greengenes.lbl.gov/), and the taxonomic identity was determined at 97% sequence similarity

158 level.

159

160 The metagenome sequencing was performed using the shotgun metagenomics method at

161 Argonne National laboratories on Illumina HiSeq 2000, 2x100 cycle run. The libraries of

162 eighteen samples (all sites, except Cenote U, which had low DNA yield and only one season

163 of samples) were prepared using TruSeq DNA sequence library preparation kit (Illumina, San

164 Diego, CA, USA). The metagenomic data were quality control filtered and annotated using

165 MG-RAST; the protein coding genes wereDraft annotated using the SEED Subsystems level 3

166 annotations (Meyer et al. 2008).

167

168

169 Metagenomic Data Deposition

170

171 All metagenomic data generated and used in this study are freely available on MG-RAST

172 server under project name ‘Yucatan groundwater’ as X10-mar09 (ID 4536385.3), X60-dec08

173 (ID 4536387.3), X83-dec08 (ID 4536389.3), XS-dec08 (ID 4536390.3), X10-jul09 (ID

174 4536384.3), X52-jul09 (ID 4536386.3), X70-jul09 (ID 4536388.3), C4-mar09 (ID

175 4536378.3), C15-dec08 (ID 4536374.3), C18-dec08 (ID 4536376.3), CS-dec08 (ID

176 4536379.3), C4-jul09 (ID 4536377.3), C13.5-jul09 (ID 4536373.3), C17-jul09 (ID

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 10 of 50

10

177 4536375.3), H4-mar09 (ID 4536383.3), H10-dec08 (ID 4536380.3), H4-jul09 (ID

178 4536382.3), and H10.4-jul09 (ID 4536381.3).

179

180

181 Statistical analysis

182

183 Taxonomic analysis of the microbial communities was performed looking at the frequency of

184 individual groups of microbes in the groundwater, as well as alpha diversity and beta diversity

185 values. The diversity values were obtained from the QIIME analysis.

186

187 The alpha diversity was assessed usingDraft Chao1 index, phylogenetic diversity (PD) index, and

188 Shannon entropy. Chao1 index is used as a non-parametric estimator for microbial

189 richness. This metric estimates the true number of species from the observed number of

190 species based on the probability of the particular species being observed repeatedly upon

2 푛1 191 resampling. Chao1 index is calculated as SChao = Sobs+ , where n1 is the number of species 2푛2

192 observed once, and n2 is the number of species observed during both sampling events (Hughes

193 et al. 2001).

194

195 Phylogenetic diversity (PD) considers not just the number of different species observed, but

196 also how phylogenetically diverse the community is. This measure is defined here as the

197 minimum total of branch lengths connecting all the organisms of the community in the

198 phylogenetic tree (Faith and Baker 2006).

199

https://mc06.manuscriptcentral.com/cjm-pubs Page 11 of 50 Canadian Journal of Microbiology

11

200 Shannon entropy is a diversity measure that considers both the species richness as well as

201 their abundance. This assures that neither abundant nor rare species are favored in

202 calculations, making it a fair measure (Tuomisto 2010). Shannon entropy is calculated as H =

203 - ∑pi log pi, where pi is the observed relative abundance of fragment i, which is calculated as

204 Ni /N (Yang et al. 2004).

205

206 The statistical analysis of the results was performed using multivariate regression method on

207 SAS® Enterprise Guide® ver 4.3. The level of significance α was set to 0.05, the probability

208 value calculated, and, compared to two-tailed probability (α/2). Eight variables were

209 evaluated to see what influenced the alpha diversity measures and the microbial composition

210 of the communities (phylum, genus, orDraft specific group level) (Table 2).

211

212

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 12 of 50

12

213 RESULTS

214

215 Overall, the metagenomic sequencing of 18 samples produced approximately 286,728,222

216 raw sequence reads with total of 33,591,513,568 bp. After quality control and removal of

217 artificial duplicate reads, 239,157,337 sequences (27,743,268,124 bp) with an average length

218 of 130 bp were used for analysis. From the amplicon sequencing, the 21 samples yielded

219 1,729,428 sequences and 86,076 OTUs with 97% sequence identity. The number of sequences

220 per sample can be seen in Supplemental Table S1.

221

222

223 DiversityDraft at phylum level

224

225 Overall, 79 different phyla were present in the samples. In addition, three unclassified

226 categories were recognized: the unclassified bacteria, unclassified archaea, and the sequences

227 that could not be classified as either bacteria or archaea. Thirteen of the identified phyla were

228 present in the samples over (or near) 1% and were used for distribution analysis (Figure 3).

229 The summary of the significant trends as response to environmental variables on the phylum

230 level is shown in Table 3.

231

232 Genus level abundance

233

234 Overall, 1502 different OTUs were recovered from the samples, and 525 of these (35%) were

235 identified at the genus level according to the Greengenes ribosomal database (Desantis et al.

https://mc06.manuscriptcentral.com/cjm-pubs Page 13 of 50 Canadian Journal of Microbiology

13

236 2006). Eight genera each had the relative abundance over 1%: Flavobacterium (relative

237 abundance 1.46%), Prochlorococcus (1.58%), Brevundimonas (1.46%), Rhodobacter

238 (1.73%), Novosphingobium (1.81%), Desulfobacterium (1.42%), Acinetobacter (17.10%), and

239 Pseudomonas (5.45%). Of these, Acinetobacter was significantly more prevalent in Cenote X

240 and the inland sites. Pseudomonas was significantly more prevalent in Cenote U (Table 4).

241

242

243 Fecal bacteria

244

245 Five groups of fecal indicator organisms – enterobacteria (total coliforms), Escherichia coli

246 (fecal coliform), enterococci, BacteroidesDraft, and Bifidobacterium – were detected in the cenotes

247 and the hotel well. Fecal indicator bacteria were significantly more prevalent during the rainy

248 season (p = 0.002); the prevalence of individual indicator organisms varied (Figure 4 and

249 Figure 5). Enterobacteriaceae are present at each site in both seasons, although they are

250 almost absent from the Cenote X sediment. Enterobacteriaceae are also the most prevalent

251 indicator organisms for all the sites except the sediment of Cenote X, although they are

252 present in significantly higher numbers during the rainy season (p = 0.0009).

253

254 The exclusively fecal coliform bacteria (represented here by genus Escherichia) were absent

255 during both seasons from the freshwater and interface layers at Cenote X and the saline layers

256 of Cenote C and Cenote U. The hotel well site had significantly more E. coli than Cenote X (p

257 = 0.047). Looking at the human impact, the tourist impact site (hotel well) had marginally

258 more E. coli than the minimal impact site (Cenote C; p = 0.058). Enterococci were absent

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 14 of 50

14

259 from many samples; the highest amount of enterococci were found in the interface sample of

260 the hotel well in rainy season. Both Cenote X and C had marginally less (p = 0.053 and p =

261 0.054, respectively) enterococci than the hotel well, and the tourist-impacted site (hotel well)

262 had significantly more (p = 0.047) than minimal impact sites.

263

264 The genus Bacteroides was similarly sparsely distributed among the sites, and absent from

265 Cenote X except for traces in the dry season fresh water sample. However, it made up

266 magnitudes higher fractions in the rainy season samples of the hotel well, where this tourist-

267 impacted site had significantly higher amounts (p = 0.034) than other sites.

268

269 Bifidobacterium was absent from half Draftof the sites, but like E. coli, it was more evenly

270 distributed. Bifidobacterium was completely absent from Cenote U and the dry season water

271 samples at Cenote X. Bifidobacterium followed the same trends as other indicator organisms

272 – the tourist-impacted hotel well had significantly higher amounts (p = 0.025) than the other

273 sites.

274

275 Prevalence of antibiotic and heavy metal resistance genes

276

277 The metal resistance pathways observed in the cenotes included resistance to arsenic, copper,

278 chromium, mercury and cobalt, zinc, and cadmium. Overall, Cenote C had significantly fewer

279 resistance genes associated with copper (p = 0.008), cobalt, zinc and cadmium (p = 0.013),

280 chromium (p = 0.0037) and mercury (p = 0.018) than the other sites.

281

https://mc06.manuscriptcentral.com/cjm-pubs Page 15 of 50 Canadian Journal of Microbiology

15

282 A similar trend was observed in the case of antibiotic resistance, although here the trend was

283 not as clear-cut. The antibiotic resistance pathways detected included many multidrug

284 resistance operons and associated structures, such as the mtdABCD multidrug resistance

285 cluster, multiple antibiotic resistance mar-locus, and various other multidrug resistance efflux

286 pumps and operons. Of the individual antibiotic resistance pathways, fluoroquinolone, beta-

287 lactamase, erythromycin, fosfomycin, streptothricin, vancomycin, methicillin and

288 aminoglycoside resistance genes were detected. Cenote C had significantly lower amount of

289 fluoroquinolone resistance (p = 0.044) together with fewer sequences associated with

290 multidrug resistance pathways, while the hotel well had higher number of sequences

291 associated with vancomycin (p = 0.0051) and fluoroquinolone (p = 0.056) resistance.

292 Draft

293 Aromatic hydrocarbon degrading bacteria

294

295 While the ability to degrade various hydrocarbons and aromatic compounds can be assessed

296 better by looking at the genomic information of the communities, 25 genera that exclusively

297 contain the known hydrocarbon or aromatic hydrocarbon degrading bacteria were identified

298 and analyzed for their abundance across the sites. These genera include Rhodococcus

299 (Actinomycetales), Sporotomaculum and Syntrophomonas (Clostridiales), Asticcaulis,

300 Brevundimonas, Caulobacter, Mycoplana, and Phenylobacterium (Caulobacterales),

301 Thalassospira (Kiloniellales), Aminobacter and Xanthobacter (Rhizobiales), Oleomonas

302 (Rhodospirilliales), Novosphingobium and Lutibacterium (Sphingomonadales), Methylibium,

303 Variovorax, and Xenophilus (Burkholderiales), Dechloromonas and Propionivibrio

304 (Rhodocyclales), Marinobacter (Alteromonadales), Alcanivorax and Oleibacter

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 16 of 50

16

305 (Oceanospirillales), Acinetobacter and Alkalindiges (Pseudomonadales), and

306 Hydrocarboniphaga (). These bacteria make up a surprisingly large fraction

307 (20%) of the total bacteria due to the prevalence of Acinetobacter, Novosphingobium and

308 Brevundimonas.

309

310 Acinetobacter dominates all the water samples from Cenote X (except the dry season

311 freshwater sample), all the samples from Cenote U, both the interface samples from Cenote C,

312 and except for the dry season interface sample from the hotel well. Novosphingobium is

313 prevalent only in the rainy season fresh water sample from Cenote X, and Brevundimonas

314 composes a large fraction only in the rainy season interface sample from Cenote C. Overall,

315 both the inland cenotes (X and U) hadDraft more hydrocarbon and aromatic compound degraders

316 than the coastal sites (p = 0.0084).

317

318 The focus of functional genomic comparison was on the degradation pathways that deal with

319 aromatic compounds that are exclusively derived from coal tar or petroleum. The hotel well

320 had more sequences associated with anaerobic toluene and ethylbenzene degradation (p =

321 0.034) and toluene-4-monooxygenase (p = 0.0004). Cenote X had more sequences associated

322 with biphenyl (p = 0.043) and chlorobenzoate (p = 0.044) degradation. Cenote C had

323 significantly fewer sequences associated with naphthalene (p = 0.017), and chlorobenzoate (p

324 = 0.026) degradation. Additionally, freshwater layer had more cresol degradation (p = 0.043).

325

326

327

https://mc06.manuscriptcentral.com/cjm-pubs Page 17 of 50 Canadian Journal of Microbiology

17

328 Special groups of interest: Cyanobacteria

329

330 Cyanobacteria constituted a total of 1.76% of the organisms present with Prochlorococcus

331 (Synechococcales) being the most abundant cyanobacterium. While cyanobacteria were

332 present in every sample, they were more prevalent in the freshwater sites with sunlight

333 exposure, although the statistical significance was only marginal (p = 0.06). The relative

334 abundance of cyanobacteria was highest at Cenote C (Figure 6). The fresh and saline layers in

335 Cenote U were the only sites showing the presence of Microcoleus (Chroococcales), which is

336 one of the cyanobacteria that can switch to sulfur reduction in dark conditions (Stal and

337 Moezelaar 1997). The overall presence of the order Chroococcales (that include other

338 common genera, such as ChroococcusDraft and Microcystis) was significantly higher in the inland

339 cenotes (X and U, p = 0.045) and in the sediment samples (p = 0.0029). Pseudoanabaenales

340 had significantly higher presence in the sediment than in the water column (p = 0.0035).

341

342 Other relatively abundant cyanobacteria include the uncultured early diverging, possibly non-

343 photosynthetic lineages belonging to the 4c0d-2 clade, notably the MLE1-12, that has been

344 observed in drinking water (Poitelon et al. 2009), and YS2 that has been detected in the

345 human and animal gut (Lin et al. 2013; Chen et al. 2011). These uncultured lineages were

346 significantly more prevalent (p<0.001) in the hotel well in both depths and seasons.

347

348

349

350

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 18 of 50

18

351 Diversity

352

353 Alpha diversity was assessed using three indices: Shannon entropy, Chao1 index, and

354 phylogenetic diversity (PD). Overall, the human-impacted hotel well had significantly higher

355 diversity values for all three indices than the sites with limited or no human impact. The

356 diversity was also significantly higher during the dry season (October-April). Sediment

357 samples (which yielded overall greater amounts of DNA) also had greater diversity values

358 than water samples for all three indices used. Higher sunlight exposure resulted in reduced

359 phylogenetic diversity, but not overall species richness. All the diversity metrics can be found

360 in supplemental data. The summary of the significant (and marginal) results is displayed in

361 Table 5. Draft

362

https://mc06.manuscriptcentral.com/cjm-pubs Page 19 of 50 Canadian Journal of Microbiology

19

363 DISCUSSION

364

365 Overall, the results indicate that the samples from human-impacted sites (hotel well) are

366 significantly different from samples from sites where the human impact is lesser or virtually

367 absent. The DNA evidence indicates that a rich community of microbes is present in the

368 cenotes. Overall, 1502 OTUs were detected in Yucatan groundwater, making the diversity

369 similar to the higher estimates from other karst aquifers (Gray and Engel 2013). It should be

370 noted that although the overall diversity is on the high end for this type of an environment, the

371 true diversity (including novel diversity) is likely even higher. Since the closed OTU picking

372 only maps the sequences to the reference sequence collection (in our case Greengenes),

373 sequences with no reference match areDraft discarded, resulting in underestimation of the true

374 microbial diversity in the environment.

375

376

377 Taxonomic composition and functional diversity

378

379 When looking at the composition of the microbial communities, our results indicate that the

380 cenotes differ from most groundwater and surface freshwater sites, which are mainly

381 dominated by Betaproteobacteria (Garrido et al. 2014; Röske et al. 2012). In contrast, marine

382 environments are dominated by Alphaproteobacteria (Gilbert et al. 2012), and where

383 Betaproteobacteria make up only a minute fraction of the total bacteria (Glöckner, Fuchs, and

384 Amann 1999). Additionally, except for the families Enterobacteraceae and Xantomonadaceae,

385 and the orders Pseudomonadales and Legionellales, are also more

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 20 of 50

20

386 dominant in the saline environments than freshwater (Newton et al. 2011). Considering the

387 constant marine water intrusion, it is likely that the cenotes and the whole groundwater

388 microbial communities may resemble the ocean microbiome, especially considering the lack

389 of surface water influence.

390

391 Dominance of Acinetobacter has been documented mainly in sewage and wastewater

392 treatment influent (Shanks et al. 2013; Karadaget al. 2013), where the phosphate

393 accumulating Acinetobacter is expected to reduce the phosphate load in the water as well as

394 degrade various hydrocarbons. However, these samples are usually characterized by the co-

395 dominance of other groups, such as Lachnospiraceae (Shanks et al. 2013), or Trichococcus

396 and Aeromonas (VandeWalle et al. 2012),Draft none of which are seen in the cenotes.

397

398 The sediment communities of both Cenote X and C are dominated by Deltaproteobacteria.

399 Sediment domination by Gamma- and Deltaproteobacteria has been shown in the anoxic

400 sediments with high rates of sulfate reduction (Wang et al. 2012; Zinger et al. 2011). Because

401 this phylum includes majority of the sulfate reducing microbes, the prevalence of

402 Deltaproteobacteria is expected, although considering the highly sulfidic water of the cenotes,

403 it is surprising that Deltaproteobacteria are not more prevalent in the interface and saline

404 layers of other sites.

405

406 Euryarcheota were also most prevalent in the sediment. This phylum was dominated by the

407 WCHD3-30 cluster of archaea, an uncultured group that has been found in methanogenic

408 zones of sediments, and are related to type II methanogens, such as Methanosaeta (Dojka et

https://mc06.manuscriptcentral.com/cjm-pubs Page 21 of 50 Canadian Journal of Microbiology

21

409 al. 1998). Although the number of known methanogens in the cenotes is very low, it is likely

410 that methanogenesis is more prevalent in the sediment considering the uncultured and

411 uncharacterized methanogenic archaea.

412

413 Actinobacteria, which have been shown to be abundant in freshwater (Holmfeldt et al. 2009;

414 Newton et al. 2007), are expectedly prevalent in the freshwater layers of Cenotes X, C, and U,

415 but not in the hotel well. The abundance of Actinobacteria is mainly due to two families, the

416 uncultured C111 and ACK-M1, both of which are commonly found in (and often dominating)

417 the freshwater habitats around the world (Urbachet al. 2001; Gucht et al. 2005).

418

419 Phylum Bacteroidetes is present in all Draftsites, although it is more numerous in the coastal sites

420 than in the inland cenotes. This phylum is often prevalent in freshwater lakes (Newton et al.

421 2011). Because many members of Bacteroidetes have been associated with degradation of

422 complex biopolymers, many studies have noted an increased level of Bacteroidetes in the

423 water during cyanobacterial blooms and in the presence of high amounts of dissolved organic

424 carbon (Eiler and Bertilsson 2007). One Bacteroidetes lineage often found elevated during

425 cyanobacterial blooms is Flavobacterium (Newton et al. 2011). While most of the

426 Bacteroidetes in the cenotes are present in low amounts, there was a significant

427 Flavobacterium presence in Cenote C, which also has a significant fraction of cyanobacteria.

428

429 The only phylum with significantly higher presence in the saline layer was Planctomycetes.

430 While Planctomycetes are found in freshwater habitats and even soil, abundance and high

431 diversity of these bacteria is characteristic of anoxic marine water and sediment (Kirkpatrick

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 22 of 50

22

432 et al. 2006), and therefore it is not surprising to find it in high abundance in saline

433 groundwater with constant marine water intrusion.

434

435 Fecal bacteria

436 The presence of fecal bacteria in the Yucatan aquifer is an established problem, owing to the

437 geological setting and the lack of water treatment infrastructure, adequate sewage treatment

438 and disposal options in the rural towns and villages, and the questionable wastewater effluent

439 disposal practices in the bigger towns and tourist resorts. While the previous studies have

440 evaluated the presence of fecal bacteria by traditional culture-based and biochemical methods,

441 this study assessed fecal indicator bacteria in the context of the whole community microbial

442 composition. Draft

443

444 Overall, the amount of fecal indicator bacteria in the cenotes makes up less than 1% of the

445 total microbes. Only fecal coliforms (Escherichia), fecal streptococci, Bacteroides, and

446 Bifidobacterium were used for the trend analysis. While total coliforms (Enterobacteriaceae)

447 are most prevalent of all the indicator bacteria in all the samples, they do not follow the same

448 trends as the other fecal indicators, confirming once again that total coliforms are not the best

449 indicators of fecal contamination, since some coliforms are not of fecal origin, and because of

450 their ability to persist in the environment (Carrero-Colon et al. 2011).

451

452 The removal of total coliforms from the analysis shows that most fecal bacteria are present in

453 the hotel well and in the freshwater zone of Cenote U. The increased amounts of fecal bacteria

454 in the hotel well is not surprising, considering the population density in the area, the number

https://mc06.manuscriptcentral.com/cjm-pubs Page 23 of 50 Canadian Journal of Microbiology

23

455 of tourists, the shallowness of the well, and the sewage effluent disposal practices. In Cenote

456 U, which is a popular swimming place, the fecal bacteria are significantly more prevalent in

457 the freshwater sample.

458

459 In addition to human-impacted sites, the fecal bacteria are also more abundant during the

460 rainy season, when there is an increased surface runoff, accelerating the transport of fecal

461 contaminants into the groundwater. It may also be that increased recreational use of the

462 cenotes during the hot summer months contributes to this problem. Even in the hotel well, the

463 rainy season has considerably more fecal bacteria than the dry season, even though the

464 occupancy rates of the coastal resorts are actually lower during the summer months.

465 Draft

466

467 Microbial defense mechanisms – antibiotic and heavy metal resistance

468

469 Antibiotic resistance is a serious problem worldwide, and the sub-therapeutic concentrations

470 of various antibiotics in the environment contribute to this issue by putting a selective

471 pressure on the microbes in natural environments to acquire antibiotic resistance genes

472 (Andersson and Hughes 2014). Antibiotics and antibiotic resistance genes end up in the

473 groundwater through improper disposal of pharmaceuticals, through human waste from the

474 antibiotic-consuming population, agricultural runoffs, and wastewater spills (Sui et al. 2015).

475 The latter is especially worrisome in freshwater-limited areas, such as Yucatan Peninsula,

476 where the fresh groundwater serves both as recreational and drinking water reservoirs, and

477 where the fast groundwater flow can distribute these resistance genes over long distances.

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 24 of 50

24

478

479 When looking at the distribution and prevalence of antibiotic resistance genes and pathways,

480 the most common features are various multidrug resistance efflux pumps (and similar

481 operons), followed by fluoroquinolone and methicillin resistance, and beta lactamases, all of

482 which are relatively uniformly distributed throughout the sites. The hotel well that has the

483 highest human impact also has significantly higher levels of vancomycin and fluoroquinolone

484 resistance. This prevalence of fluoroquinolone resistance in Yucatan groundwater is

485 worrisome, although not uncommon, since the rapid development of fluoroquinolone

486 resistance is a global trend (Dalhoff 2012).

487

488 In addition to antibiotic resistance, thereDraft is also relatively high level of metal resistance genes

489 present in the cenotes, which is somewhat surprising considering the virtual absence of heavy

490 and transitional metals in these environments. Genes and pathways encoding for the resistance

491 to arsenic, cobalt, zinc copper and cadmium are prevalent in all the sites, while the resistance

492 pathways to mercury and chromium are present in some sites. It is possible that an

493 unknown/undetected factor is responsible for the selective pressure keeping these genes in the

494 microbial community, or that the prevalence of these genes in the virtually metal free cenotes

495 is an additional evidence of how the absence of environmental selection does not mean the

496 genes will be lost.

497

498

499

500

https://mc06.manuscriptcentral.com/cjm-pubs Page 25 of 50 Canadian Journal of Microbiology

25

501 Degradation of aromatic compounds

502

503 While the bacteria known to degrade aromatic hydrocarbons were more prevalent in Cenote

504 X, the genomic data showed significantly increased amounts of degradation pathways both in

505 the Cenote X and the hotel well. The overall most prevalent pathways were the

506 chloroaromatic degradation and biphenyl degradation. However, these pathways are present

507 naturally in various bacteria in the environment (Bugg et al. 2011), and do not necessarily

508 indicate the presence of the aromatic hydrocarbons.

509

510 The degradation pathway is the same for naphthalene and anthracene and encoded on the

511 NAH plasmid that is relatively commonDraft in various Pseudomonas strains as well as in

512 Aeromonas, Flavobacterium, Bacillus and others (Sanservino et al. 1993; Cerniglia and

513 Heitkamp 1989). Because naphthalene and anthracene are relatively soluble in water, the

514 degradation of these compounds is fast, and the ability to degrade them is widespread in both

515 pristine and contaminated environments (Cerniglia and Heitkamp 1989). Since Yucatan

516 Peninsula lacks industries where naphthalene or anthracene are used or produced, the

517 fluctuation of naphthalene degrading pathways in the water is not tied to industrial

518 contamination.

519

520 The pathways for toluene and ethylbenzene degradation are encoded on the TOL-plasmid that

521 is relatively common in Pseudomonas and various other organisms. The prevalence of both

522 toluene monooxygenase and anaerobic toluene and ethylbenzene degradation pathways

523 identified in this study only in the hotel well microbial community may suggest the presence

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 26 of 50

26

524 of these compounds in the water. As both of these chemicals are found in gasoline, and as the

525 resort has more motorized traffic than either Cenote C or X, there are various sources that can

526 leak low levels of toluene and ethylbenzene to the ground.

527

528 Two trends were specific to the redox conditions: cresol degradation was more prevalent in

529 the aerobic freshwater zone and the anaerobic degradation of toluene and ethylbenzene was

530 more prevalent at the interface. Anaerobic toluene and ethylbenzene degradation are common

531 to some sulfate reducing bacteria, such as the Desulfobacteraceae family (Koizumi et al.

532 2002) as well as various denitrifying and iron reducing bacteria (Heider et al. 1998).

533 Considering the prevalence of Desulfobacteraceae throughout the sampling sites, the

534 prevalence of these pathways in the reducingDraft environments is not surprising, regardless of the

535 presence or absence of the chemicals.

536

537 Cyanobacteria

538

539 Although cyanobacteria, which are present in most aquatic environments, are not generally

540 used as primary water quality indicators, the prevalence of cyanobacteria can indicate specific

541 pollutant problems, such as high levels of phosphate (Conley et al. 2009). Based on the 16S

542 analysis, cyanobacterial taxa are not prevalent in the cenotes, with the exception of Cenote C

543 that receives the most sunlight. The cyanobacteria are not numerous in the freshwater zones of

544 Cenote X and Cenote U where the light conditions are favorable. Since iron is one of the

545 limiting nutrients for cyanobacterial growth (Mills et al. 2004), the lack of metals is likely one

546 of the major contributing factors to the lack of these organisms.

https://mc06.manuscriptcentral.com/cjm-pubs Page 27 of 50 Canadian Journal of Microbiology

27

547

548 Diversity within the communities

549

550 Besides looking at the taxonomic composition of the microbial communities, an equally

551 important indicator in ecosystem health is the overall diversity. Most of the factors

552 influencing microbial diversity are predictable. The biggest impact to all the three diversity

553 metrics came from the substrate where the microbes were growing (sediment had more than

554 water), and from the tourist impacted site (hotel well had more than other sites).

555

556 The increased diversity in the sediment is not surprising, because the sediment harbors higher

557 number of microbes than water. The impactDraft that a busy resort with a wastewater injection

558 well and effluent usage for irrigation purposes have on the groundwater and its microbial

559 content and diversity is also predictable. Location of the well less than half a kilometer from

560 the coast can also enhance the diversity, as the tides can push marine water into the well

561 (hence the saline water intrusion). The impact of marine water intrusion can also be a factor in

562 the composition of the microbial communities – the hotel well is the only site with the

563 presence of exclusively marine microbes, such as Mariprofundus (Zetaproteobacteria).

564

565 However, in this study only the phylogenetic diversity was affected by the sunlight, while

566 Shannon entropy and the Chao1 index were not. This indicates that while Cenote C (the only

567 site that received sunlight throughout the water column) may not necessarily have an overall

568 lower number of microbes present, but rather, as the data indicated, these microbial

569 communities consist of closely related photosynthetic organisms. Cenote C had the highest

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 28 of 50

28

570 number of photosynthetic bacteria (both cyanobacteria, and purple and green sulfur bacteria)

571 of all the sampled sites.

572

573 Overall, the microbial diversity in Yucatan groundwater is high similar to other tropical karst

574 environments that have some human impact. The biggest factors that influence the microbial

575 community composition and the diversity in Yucatan groundwater are anthropogenic

576 influence and sunlight. While there are also minor seasonal differences between the dry and

577 rainy season communities that are likely influenced by increased runoff during rainy season,

578 and the coastal sites are influenced by the seawater intrusion, these effects are minimal.

579 Additionally, the sediment community is more diverse than the planktonic community, with

580 the sediment community having a significantlyDraft higher proportion of sequences from unknown

581 organisms. While the taxonomic difference was mainly influenced by human impact, the

582 metagenomic difference had much clearer differences between the sites, light input and

583 salinity/redox potential.

584

https://mc06.manuscriptcentral.com/cjm-pubs Page 29 of 50 Canadian Journal of Microbiology

29

585 ACKNOWLEDGEMENT

586

587 The authors would like to thank the Northern Illinois University Center for Latino and Latin

588 American Studies for the financial support. We would like to thank the Earth Microbiome

589 Project, especially Jack Gilbert and Sarah Owens, for allowing us to participate and for their

590 great insights. We would like to thank the following people who have contributed to this

591 research: Andrew Thompson, Dave Breed, Cheyenne Johnson, Rich Becker, and Bianca

592 Pedersen. We also like to thank the two anonymous reviewers for their thoughtful comments.

593

594

595 Draft

596

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 30 of 50

30

597 References

598

599 Andersson, D.I., and Hughes, D. 2014. Microbiological effects of sublethal levels of

600 antibiotics. Nature Reviews Microbiology, 12(7): 465-478.

601

602 Bugg, T.D., Ahmad, M., Hardiman, E.M., and Rahmanpour, R. 2011. Pathways for

603 degradation of lignin in bacteria and fungi. Natural Product Reports, 28(12): 1883-1896.

604

605 Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K.,

606 Fierer, N., Gonzalez Pentildea, A., Goodrich, J.K., Gordon, J.I., Huttley, G.A., Kelley, S.T.,

607 Knights, D., Koenig, J.E., Ley, R.E., Lozupone,Draft C.A., McDonald, D., Muegge, B.D., Pirrung,

608 M., Reeder, J., Sevinsky, J.R., Turnbaugh, P.J., Walters, W.A., Widmann, J., Yatsunenko, T.,

609 Zaneveld, J., and Knight, R. 2010. QIIME allows analysis of high-throughput community

610 sequencing data. Nature Methods, 7(5): 335–336.

611

612 Carrero-Colon, M., Wickham, G.S., Turco, and R.F. 2011. , phylogeny, and

613 physiology of fecal indicator bacteria. In The Fecal Bacteria. Edited by M.J. Sadowski and

614 R.L. Whitman. American Society for Microbiology Press, Washington DC, pp 23-38.

615

616 Cerniglia, C.E., and Heitkamp, M.A. 1989. Microbial degradation of PAH in aquatic

617 environment. In Metabolism of Polycyclic Aromatic Hydrocarbons in the Aquatic

618 Environment. Edited by U. Varanasi. CRC Press, Boca Raton, FL, pp 41-68.

619

https://mc06.manuscriptcentral.com/cjm-pubs Page 31 of 50 Canadian Journal of Microbiology

31

620 Chen, Y., Penner, G.B., Li, M., Oba, M., and Guan, L.L. 2011. Changes in bacterial diversity

621 associated with epithelial tissue in the beef cow rumen during the transition to a high-grain

622 diet. Applied and Environmental Microbiology, 77(16): 5770–5781.

623

624 Choi, H.J., Lee, A.H., and Lee, S.M. 2012. Comparison between a moving bed bioreactor and

625 a fixed bed bioreactor for biological phosphate removal and denitrification. Water Science

626 and Technology, 65(10): 1834–1838.

627

628 Conley, D.J., Paerl, H.W., Howarth, R.W., Boesch, D.F., Seitzinger, S.P., Havens, K.E.,

629 Lancelot, C., and Likens, G.E. 2009. Controlling eutrophication: nitrogen and phosphorus.

630 Science, 323(5917): 1014-1015. Draft

631

632 Dalhoff, A. 2012. Global fluoroquinolone resistance epidemiology and implications for

633 clinical use [online]. Interdisciplinary Perspectives in Infectious Diseases, Oct 14, 2012. doi:

634 10.1155/2012/976273.

635

636 Dojka, M.A., Hugenholtz, P., Haack, S.K., and Pace, N.R. 1998. Microbial diversity in a

637 hydrocarbon- and chlorinated solvent-contaminated aquifer undergoing intrinsic

638 bioremediation. Applied and Environmental Microbiology, 64(10): 3869-3877.

639

640 Eiler, A., and Bertilsson, S. 2007. Flavobacteria blooms in four eutrophic lakes: linking

641 population dynamics of freshwater bacterioplankton to resource availability. Applied and

642 Environmental Microbiology, 73(11): 3511-3518.

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 32 of 50

32

643

644 Faith, D.P., and Baker, A.M. 2006. Phylogenetic diversity (PD) and biodiversity conservation:

645 some bioinformatics challenges [online]. Evolutionary Bioinformatics, 2:121-128.

646 https://doi.org/10.1177/117693430600200007

647

648 Garrido, L., Sanchez, O., Ferrera, I., Tomas, N., and Mas, J. 2014. Dynamics of microbial

649 diversity profiles in waters of different qualities. Approximation to an ecological quality

650 indicator. Science of the Total Environment, 468: 1154-1161.

651

652 Gines, A., and Gines, J. 2007. Eogenetic karst, glacioeustatic cave pools and anchialine

653 environments on Mallorca Island: a discussionDraft of coastal speleogenesis. International Journal

654 of Speleology, 36(2): 57-67.

655

656 Glöckner, F.O., Fuchs, B.M., and Amann, R. 1999. Bacterioplankton composition if lakes and

657 oceans: a first comparison based on fluorescence in situ hybridization. Applied and

658 Environmental Microbiology, 65(8): 3721-3726.

659

660 Gray, C.J., and Engel, A.S. 2013. Microbial diversity and impact on carbonate geochemistry

661 across a changing geochemical gradient in a karst aquifer. ISME Journal, 7(2): 325-337.

662

663 Gucht, K., Vandekerckhove, T., Vloemans, N., Cousin, S., Muylaert, K., Sabbe, K., Gills, M.,

664 Declerk, S., Meester, L., and Vyverman, W. 2005. Characterization of bacterial communities

https://mc06.manuscriptcentral.com/cjm-pubs Page 33 of 50 Canadian Journal of Microbiology

33

665 in four freshwater lakes differing in nutrient load and food web structure. FEMS

666 Microbiology Ecology, 53(2): 205-220.

667

668 Heider, J., Spormann, A.M., Beller, H.R., and Widdel, F. 1998. Anaerobic bacterial

669 metabolism of hydrocarbons. FEMS Microbiology Reviews, 22(5): 459-473.

670

671 Holmfeldt, K., Dziallas, C., Titelman, J., Pohlmann, K., Grossart, H.P., and Riemann, L.

672 2009. Diversity and abundance of freshwater Actinobacteria along environmental gradients in

673 the brackish northern Baltic Sea. Environmental Microbiology, 11(8):2042-54.

674

675 Hughes, J.B., Hellmann, J.J., Ricketts,Draft T.H., and Bohannan, B.J.M. 2001. Counting the

676 uncountable: Statistical approaches to estimating microbial diversity. Applied and

677 Environmental Microbiology, 67(10): 4399-4406.

678

679 Instituto Nacional de Estadísticay Geografía e Informática, Mexico. 2019. Perspectiva

680 estadistica Yucatan Marzo 2019. Retrieved from: https://www.inegi.org.mx/default.html

681

682 Instituto Nacional de Estadísticay Geografía e Informática, Mexico. 2013. Perspectiva

683 estadistica Yucatan Marzo 2013.Retrieved from: https://www.inegi.org.mx/default.html

684

685 Instituto Nacional de Estadística, Geografía e Informática, Mexico. 2001. Indicadores

686 Sociodemográficos de México (1930-2000). Retrieved from:

687 https://www.inegi.org.mx/default.html

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 34 of 50

34

688

689 Karadag, D., Özkaya, B., Ölmez, E., Nissilä, M.E., Çakmakçı, M., Yıldız, Ş., and Puhakka,

690 J.A. 2013. Profiling of bacterial community in a full-scale aerobic composting plant.

691 International Biodeterioration and Biodegradation, 77: 85-90.

692

693 Kirkpatrick, J., Oakley, B., Fuchsman, C., Srinivasan, S., Staley, J.T., and Murray, J.W. 2006.

694 Diversity and distribution of Planctomycetes and related bacteria in the suboxic zone of the

695 Black Sea. Applied and Environmental Microbiology, 80(14): 3079-3083.

696

697 Koizumi, Y., Kelly, J.J., Nakagawa, T., Urakawa, H., El-Fantroussi, S., Al-Muzaini, S.,

698 Fukui, M., Urushigawa, Y., and Stahl,Draft D.A. 2002. Parallel characterization of anaerobic

699 toluene-and ethylbenzene-degrading microbial consortia by PCR-denaturing gradient gel

700 electrophoresis, RNA-DNA membrane hybridization, and DNA microarray technology.

701 Applied and Environmental Microbiology, 68(7): 3215-3225.

702

703 Lefticariu, M., Perry, E.C., Ward, W.C., and Lefticariu, L. 2006. Post-Chicxulub depositional

704 and diagenetic history of the northwestern Yucatan Peninsula, Mexico. Sedimentary Geology,

705 183(1-2): 51-69.

706

707 Lin, A., Bik, E.M., Costello, E.K., Dethlefsen, L., Haque, R., Rekman, D.A., and Singh, U.

708 2013. Distinct Distal Gut Microbiome Diversity and Composition in Healthy Children from

709 Bangladesh and the United States [online]. PloS ONE, 8(1): e53838.

710 doi:10.1371/journal.pone.0053838

https://mc06.manuscriptcentral.com/cjm-pubs Page 35 of 50 Canadian Journal of Microbiology

35

711

712 Meyer, F., Paarmann, D., D‘Souza, M., Olson, R., Glass, E.M., Kubal, M., Paczian, T.,

713 Rodriguez, A., Stevens, R., Wilke, A., Wilkening, J., and Edwards, R.A. 2008. The

714 metagenomics RAST server – a public resource for the automatic phylogenetic and functional

715 analysis of metagenomes [online]. BMC Bioinformatics, 9(1):386. doi:10.1186/1471-2105-9-

716 386.

717

718 Mills, M.M., Ridame, C., Davey, M., La Roche, J., and Geider, R.J. 2004. Iron and

719 phosphorus co-limit nitrogen fixation in the eastern tropical North Atlantic. Nature,

720 429(6989): 292-294.

721 Draft

722 National Environment and Planning Agency, Government of Jamaica. 2008. Main Issues

723 Raised and the Agency’s Response to Water Resource Authority. Accessed from

724 http://www.nepa.gov.jm/eias/StAnn/PearTree/Hojapi%20Issues.pdf

725

726 Newton, R.J., Jones, S.E., Eiler, A., McMahon, K.D., and Bertillson, S. 2011. A guide to the

727 natural history of freshwater lake bacteria. Microbiology and Molecular Biology Reviews,

728 75(1): 14-49.

729

730 Newton, R.J., Jones, S.E., Helmus, M.R., and McMahon, K.D. 2007. Phylogenetic ecology of

731 the freshwater Actinobacterial lineage. Applied and Environmental Microbiology, 73(22):

732 7169-7176.

733

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 36 of 50

36

734 Pacheco, J.A., Cabera, A.S., and Marin, L.E. 2000. Bacteriological contamination in the

735 karstic aquifer of Yucatan, Mexico. Geofisica Internacional, 39(3): 285-291.

736

737 Pedersen, B. 2007. Water geochemistry and geomicrobiology of two sulfidic cenotes, Yucatan

738 Peninsula. PhD dissertation, Department of Geology and Environmental Geoscience,

739 Northern Illinois University, DeKalb, IL.

740

741 Perry, E., Paytan, A., Pedersen, B., and Velazquez-Oliman, G. 2009. Groundwater

742 geochemistry of the Yucatan Peninsula, Mexico: Constraints on stratigraphy and

743 hydrogeology. Journal of Hydrology, 367 (1-2): 27-40.

744 Draft

745 Perry, E., Velazquez-Oliman, G., and Socki, R.A. 2003. Hydrogeology of the Yucatan

746 Peninsula. In The Lowland Maya Area: Three Millennia at the Human-Wildland Interface.

747 Edited by A. Gomez-Pompa, M.F. Allen, S.L. Fedick, and J.J. Jimenez-Osrnio. Food Products

748 Press, New York, pp 115-138.

749

750 Poitelon, J.B., Joyeuxa, M., Weltéa, B., Dugueta, J.P., Prestelb, E., Lespinetb, O., and

751 DuBow, M.S. 2009. Assessment of phylogenetic diversity of bacterial microflora in drinking

752 water using serial analysis of ribosomal sequence tags. Water Research, 43(17): 4197-4206.

753

754 Röske, K., Sachse, R., Scheerer, C.,and Röske, I. 2012. Microbial diversity and composition

755 of the sediment in the drinking water reservoir Saidenbach (Saxonia, Germany). Systematic

756 and Applied Microbiology, 35(1): 35-44.

https://mc06.manuscriptcentral.com/cjm-pubs Page 37 of 50 Canadian Journal of Microbiology

37

757

758 Sanseverino, J., Applegate, B.M., King, J.M., and Sayler, G.S. 1993. Plasmid-mediated

759 mineralization of naphthalene, phenanthrene, and anthracene. Applied and Environmental

760 Microbiology, 59(6): 1931-1937.

761

762 Shanks, O.C., Newton, R.J., Kelty, C.A., Huse, S.M., Sogin, M.L., and McLellan, S.L. 2013.

763 Comparison of the microbial community structures of untreated wastewaters from different

764 geographic locales. Applied and Environmental Microbiology, 79(9): 2906-2913.

765

766 Sui, Q., Cao, X., Lu, S., Zhao, W., Qiu, Z., and Yu, G. 2015. Occurrence, sources and fate of

767 pharmaceuticals and personal care productsDraft in the groundwater: a review. Emerging

768 Contaminants, 1(1): 14-24.

769

770 Urbach, E.K., Vergin, L., Young, L., and Morse, A. 2001. Unusual bacterioplankton

771 community tructure in ultra-oligtrophic Crater Lake. Limnology and Oceanography, 46(3):

772 557-572.

773

774 VandeWalle, J., Goetz, G., Huse, S., Morrison, H., Sogin, M., Hoffman, R., Yan, K., and

775 McLellan, S. 2012. Acinetobacter, Aeromonas and Trichococcus populations dominate the

776 microbial community within urban sewer infrastructure. Environmental Microbiology ,14 (9):

777 2538-2552.

778

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 38 of 50

38

779 Wang, Y., Sheng, H.F., He, Y., Wu, J.-Y., Jiang, Y.-X., Fung-Yee Tam, N., and Zhou, H.-W.

780 2012. Comparison of the levels of bacterial diversity in freshwater, intertidal wetland, and

781 marine sediments by using millions of Illumina tags. Applied and Environmental

782 Microbiology, 78(23): 8264-8271.

783

784 White, D.A., and Hood, C.S. 2004. Vegetation patterns and environmental gradients in

785 tropical dry forests of the northern Yucatan Peninsula. Journal of Vegetation Science, 15(2):

786 151–161.

787

788 Yang, Y.-H., Yao, J., and Wang, M.-C. 2004. RAPD marker and substrate utilization pattern

789 applied to study microbial communityDraft diversity in the soil affected by agricultural chemicals.

790 Journal of Environmental Science and Health, 39(10): 125-138.

791

792 Zinger, L., Amaral-Zettler, L.A., Fuhrman, J.A., Horner-Devine, M.C., Huse, S.M., Welch,

793 D.B.M., Martiny, J.B.H., Sogin, M., Boetius, A., Ramette, A. 2011. Global patterns of

794 bacterial beta-diversity in seafloor and seawater ecosystem [online]. PloS one, 6(9): e24570.

795 https://doi.org/10.1371/journal.pone.0024570

796 797

https://mc06.manuscriptcentral.com/cjm-pubs Page 39 of 50 Canadian Journal of Microbiology

39

798 799 800 801 802 803 Table 1: Comparison of the four sampling sites. 804 Cenote X Cenote U Cenote C Hotel Geographical coordinates 20.909567 20.989455 20.585669 20.370214 (Latitude/Longitude) -88.866947 -88.601856 -87.174442 -87.333573

Location Inland Inland Coast Coast

Type of cenote Pit cenote Pit cenote Cavernous Water well cenote Depth 120 m 118 m 20 m 20 m

Interface 50-60 m 68-75 m 13-15 m 9 m and below

Light input Low Low High None

Vegetation carbon input High DraftLow Low None

Human access to water Recreational Recreational None None 805 806

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 40 of 50

40

807 Table 2: Environmental variables evaluated in the study. 808

Site Cenote X Cenote C Cenote U Hotel Well

Season dry vs. rainy dry vs. rainy rainy only dry vs. rainy

Substrate water vs. sediment water vs. sediment water only water only

Location inland coastal inland coastal

Salinity freshwater freshwater freshwater freshwater interface interface interface interface saline saline saline

Light high light (fresh water) high light low light no light low light (interface and saline)

Type of human agricultural impact Draftminimal impact agricultural impact tourist impact impact 809 810

https://mc06.manuscriptcentral.com/cjm-pubs Page 41 of 50 Canadian Journal of Microbiology

41

811 Table 3: Summary of significant trends observed on the phylum level. 812 Phylum Fraction of Significant trend total microbes Actinobacteria 4.13 Freshwater has more than interface (p = 0.028) and saline layer (p = 0.015)

Bacteroides 7.66 Coast has more than inland (p = 0.009)

Chlorobi 1.3 Coast has marginally more than inland (p = 0.0528)

Chloroflexi 5.7 None

Cyanobacteria 2.13 None

Firmicutes 2.77 Coast has more than inland (p = 0.0228)

OP3 1.07 None

Planctomycetes 2.54 Saline layer has more than freshwater and interface (p = 0.034), and sediment has more than water (p = 0022) * 55.7 DraftTotal Proteobacteria Alpha 11.49 Rainy season has more than dry season (p = 0.0085) Proteobacteria Freshwater has more than interface (p = 0.005) and saline 5.28 layer (p = 0.0034); Cenote X has less than hotel well (p = Beta 0.0311) Proteobacteria

Gamma 31.2 Coast has less than inland (p = 0.0045) Proteobacteria

Delta Rainy season has less than dry season (p = 0.0116); Proteobacteria 5.47 sediment has more than water (p = 0.0083)

Epsilon 1.3 None Proteobacteria

SAR406 1.13 None

Verrumicrobia 1.5 None

Bacteria, Phylum 6.7 Cenote U has less than hotel well (p = 0.0164); dry season unidentified has more than rainy season (p = 0.0247); and sediment has more than water (p = 0.0026)

Euryarcheota 2.37 Sediment has more than water (p < 0.0001)

Unclassified 0.64 Sediment has more than water (p = 0.0003)

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 42 of 50

42

813 * The phylum Proteobacteria was broken down to classes Alpha, Beta, Gamma, Delta and Epsilon 814 Proteobacteria. 815 816 Table 4: Relative abundance of the dominant genera in specific samples. 817 Genus Relative Site Season Salinity abundance Novosphingobium 29.8% Cenote X Rainy Fresh

Prochlorococcus 22% Cenote C Rainy Fresh

Brevundimonas 21.7% Cenote C Rainy Interface

Rhodobacter 11.8% Cenote C Dry Saline 16.4% Hotel well Rainy Fresh

Flavobacterium 12.8% Cenote C Dry Saline

Desulfobacterium 9.8% Cenote C Rainy Interface 818 Draft

https://mc06.manuscriptcentral.com/cjm-pubs Page 43 of 50 Canadian Journal of Microbiology

43

820 Table 5: The change in alpha diversity as a response to environmental variables. 821 Chao1 PD Shannon entropy Site (compared to hotel well): Cenote X less (p = 0.045) None Cenote C less (p = 0.013) Cenote U less (p = 0.02) less (p = 0.052)

Season Dry (compared to rainy season): more (p = 0.053) more (p = None 0.056) Substrate Water (compared to sediment): less (p = 0.0024) less (p = less (p= 0.01) 0.0055) Location Coast (compared to inland): None None more (p = 0.058) Salinity (compared to saline layer): Fresh None None None Interface Draft Light (compared to no light): High light Low Light None Less (p = 0.057) None

Human impact Tourist (compared to minimal impact): More (p = 0.048) More (p= More (p= 0.007) 0.042) 822 823 824

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 44 of 50

44

825

826 Figure Captions

827

828

829 Figure 1: Sampling sites in Yucatan Peninsula, Mexico.

830 Figure 2: Sampling plan with details of location, time and depth.

831 Figure 3: Distribution of the thirteen most prevalent phyla together with unidentified

832 organisms and unclassified bacteria.

833 Figure 4: Relative abundance of the fecal indicator bacteria in the cenotes.

834 Figure 5: Relative abundance of fecal indicator bacteria other than coliforms.

835 Figure 6: Relative abundance of prevalentDraft cyanobacteria

836

https://mc06.manuscriptcentral.com/cjm-pubs Page 45 of 50 Canadian Journal of Microbiology

45

837

838

839 Figure 1: Sampling sites in Yucatan Peninsula, Mexico.

840 Draft

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 46 of 50

46

841

842

843 844 Figure 2: Sampling plan with details ofDraft location, time and depth. 845

846

https://mc06.manuscriptcentral.com/cjm-pubs Page 47 of 50 Canadian Journal of Microbiology

47

1.00E+00

9.00E-01

8.00E-01

7.00E-01

6.00E-01

5.00E-01

4.00E-01

3.00E-01 Relative frequency (%) 2.00E-01

1.00E-01

0.00E+00

C4jul09 H4jul09 X10jul09X52jul09X70jul09 XSdec08CSdec08 X60dec09X83dec08 C4mar09C15dec08C17dec08 C18jul09U10jul09U70jul09U80jul09H4mar09 X10mar09 DraftC13.5jul09 H10mar09 H10.4jul09 Unclassified Euryarchaeota Bacteria;Other Actinobacteria Bacteroidetes Chlorobi Chloroflexi Cyanobacteria Firmicutes OP3 Planctomycetes Proteobacteria SAR406 Spirochaetes Verrucomicrobia

X – Cenote X; C – Cenote C; U – Cenote U, H – hotel well. The number indicates sampling depth (m), S: 847 sediment. December-March: dry season, July: rainy season

848

849 Figure 3: Distribution of the thirteen most prevalent phyla together with unidentified

850 organisms and unclassified bacteria.

851

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 48 of 50

48

852

100%

90%

80%

70%

60%

50%

40%

30% Relative frequency (%) 20%

10%

0% Draft C4jul09 H4jul09 X10jul09X52jul09X70jul09C4mar09 C18jul09U10jul09U70jul09U80jul09H4mar09 XSdec08CSdec08 X10mar09X60dec09X83dec08 C15dec08C17dec08 C13.5jul09 H10mar09 H10.4jul09 Enterobacteraceae (Coliforms) Escherichia (Fecal coliform) Enterococcus Bacteroides X – Cenote X; C – Cenote C; U – Cenote U, H – hotel well. The number indicates sampling depth (m), S: 853 sediment. December-March: dry season, July: rainy season

854 Figure 4: Relative abundance of the fecal indicator bacteria in the cenotes.

https://mc06.manuscriptcentral.com/cjm-pubs Page 49 of 50 Canadian Journal of Microbiology

49

855

100% 90% 80% 70% 60% 50% 40%

Relative frequency (%) 30% 20% 10% 0%

C4jul09 H4jul09 X10jul09X52jul09X70jul09C4mar09 C18jul09U10jul09U70jul09U80jul09H4mar09 XSdec08CSdec08 X10mar09X60dec09X83dec08 C15dec08DraftC17dec08 C13.5jul09 H10mar09 H10.4jul09 Escherichia (Fecal coliform) Enterococcus Bacteroides Bifidobacterium X – Cenote X; C – Cenote C; U – Cenote U, H – hotel well. The number indicates sampling depth (m), S: 856 sediment. December-March: dry season, July: rainy season

857 Figure 5: Relative abundance of fecal indicator bacteria other than coliforms.

https://mc06.manuscriptcentral.com/cjm-pubs Canadian Journal of Microbiology Page 50 of 50

50

858

859

100%

90%

80%

70%

60%

50%

40%

Relative frequency (%) 30%

20% 10% Draft 0%

C4jul09 H4jul09 X10jul09X52jul09X70jul09C4mar09 C18jul09U10jul09U70jul09U80jul09H4mar09 XSdec08CSdec08 X10mar09X60dec09X83dec08 C15dec08C17dec08 C13.5jul09 H10mar09 H10.4jul09 Uncultured 4C0d-2 Uncultured S15B-MN24 Nostocales and Stigonematales Croococcales X – Cenote X; C – Cenote C; U – Cenote U, H – hotel well. The number indicates sampling depth (m), S: 860 sediment. December-March: dry season, July: rainy season

861 Figure 6: Relative abundance of prevalent cyanobacteria.

862

863

864

865

866

867

868 869 870

https://mc06.manuscriptcentral.com/cjm-pubs