Abstract
The study objective was to evaluate human faecal contamination impacts in the Yal-ku lagoon in the Mexican Caribbean and to estimate adenovirus infection and illness risks associated with recreational exposure during water activities. A total of 20 water samples (10 from each site × two sites) (50 L) were collected monthly over a period of 12 months from two selected sampling sites in the swimming area of the Yal-ku lagoon. The occurrence of faecal-associated viruses was explored, and human adenovirus (HAdV) and pepper mild mottle virus (PMMoV) concentrations were quantified. A quantitative microbial risk assessment (QMRA) model was used to estimate exposure and subsequent adenovirus infection and illness risk for 1 h of swimming or snorkelling. Somatic and F + -specific coliphages occurred in 100% of the samples. Both HAdV and PMMoV were detected at a 60% frequency thereby indicating persistent faecal inputs. PMMoV concentrations (44–370 GC/L) were relatively lower than the concentrations of HAdV (64–1,000 GC/L). Estimated mean adenovirus risks were greater for snorkelling than for swimming by roughly one to two orders of magnitude and estimated mean illness risks for snorkelling were >32/1,000. Human faecal contamination is frequent in the Yal-ku lagoon, which is associated with human gastrointestinal illness.
HIGHLIGHTS
Human faecal contamination detected in a Mexican Caribbean lagoon.
The rainfall seasons posed the greatest risk.
Estimated risks were greater for snorkelling than for swimming by one to two orders of magnitude.
Estimated mean illness risks for snorkelling were >32/1,000.
INTRODUCTION
Human enteric viruses have been frequently detected in aquatic environments around the world as a result of direct disposal of treated or untreated wastewater discharges into water bodies, posing a risk of illness to those who use contaminated water for drinking or recreation (Okoh et al. 2010). In recreational areas, enteric viruses such as human adenovirus (HAdV) and norovirus (NoV) have been responsible for gastrointestinal illness outbreaks (Sinclair et al. 2009; Fewtrell & Kay 2015). These outbreaks have important public health impacts and potential economic impacts for cities with large tourism economies. Travelers' diarrhea (TD) (Simons et al. 2016), for example, can affect those travelling internationally with water being one potential exposure route (Simons et al. 2016) to enteric viruses including HAdV and NoV. This is especially a challenge in the Mexican Caribbean, where there are sewage pollution challenges (e.g., failing septic tanks or disposal of sewage into the ground leading to contamination of groundwater) (Hernández-Terrones et al. 2015), in part due to the increased burden on sewage treatment systems during tourism seasons (Hernández-Terrones et al. 2015). In Cancun, Mexico, there were approximately 4.6 million tourists in 2021 and over 13.5 million tourists in the entire state of Quintana Roo, which includes other destinations like Cozumel, Playa del Carmen, and Tulum (López 2021). In addition to a large tourist population, there has been rapid population increase in the state of Quintana Roo with at least 40% of increase only in the last 10 years (Hernández-Terrones et al. 2015). A previous study epidemiologically linked swimming in coastal waters of Caribbean Mexico to enterovirus infections among tourists (Begier et al. 2008), and hepatitis A virus (HAV) and norovirus have been detected in Mexican tourist beaches in another previous research (Félix et al. 2010). Sewage contamination of recreational waters has negative effects on the economy of the area, public health, and the ecosystems of these environments (Hernández-Terrones et al. 2015).
We focus here on the public health impacts of human faecal pollution of the Mexican Caribbean waters, specifically the Yal-ku lagoon. While there are known risks from recreational water activities in the Yal-ku lagoon where human adenoviruses were reported to be the most prevalent enteric viral group (Muñoz-Cortés et al. 2023), the magnitude of these risks and how they vary by season is unknown, in part due to epidemiological challenges in identifying specific sources of individual infections that vary in whether they develop into illness and then vary in severity of illness. Because this disease burden is difficult to directly measure, quantitative microbial risk assessment (QMRA) is a powerful tool for addressing this knowledge gap in estimating individual infection and illness risks given the data about environmental concentrations of pathogens and knowledge about human behaviours and exposure factors (Haas et al. 2014).
An initial step to quantifying infection risks is to determine whether faecal pollution is present. While one approach is to directly detect and quantify pathogens, this can be challenging due to detection limits and the interpretation of left-censored data (i.e., concentration of pathogens below detectable limits). It is useful to use faecal contamination indicators to determine if the water is likely to be contaminated with faecal pathogens. For example, coliphages, viruses that infect Escherichia coli, are often used as faecal indicators, because they are naturally occurring in faeces, have little to no environmental replication, have demonstrated co-occurrence with pathogens of interest, and have similar resistance to inactivation as viral pathogens of interest (EPA Office of Water 2015). However, detection of coliphages does not specify human faecal contamination specifically which is the main water pollution source of concern in the Mexican Caribbean. Pepper mild mottle virus (PMMoV) is often used as an indicator of faecal contamination, because it is prevalent in human faeces, originating from foods consumed by humans but not that of other mammals (Rosario et al. 2009).
In Mexico, environmental virology and the study of viruses as faecal indicators in the aquatic environment is still under development; in fact, very few similar studies have been published in the Caribbean (Cordero et al. 2012; Muñoz-Cortés et al. 2023; Serville-Tertullien et al. 2022). Viruses, such as coliphages, enteric viruses, or other faecal indicator viruses, are not included in the parameters to assess water quality in Mexico. Rather, bacterial indicators are normally used in part to advise on swimability. For example, for pools, faecal coliforms should be <40 MPN (most probable number)/100 mL according to the ‘NORMA Oficial Mexicana NOM-245-SSA1-2010, Requisitos sanitarios y calidad del agua que deben cumplir las albercas’, and for seawater in recreational beaches, enterococci should be within the limit of 100 MPN/100 mL according to ‘Norma Mexicana que establece los requisitos y especificaciones de sustentabilidad de calidad de playas NMX-AA-120-SCFI-2016’.
The objectives of this study were to (1) measure concentrations of HAdV and indicators of faecal contamination (PMMoV and somatic and male F + -specific coliphages) in the Yal-ku lagoon over three seasons: dry, north, and rainy, and (2) estimate adenovirus infection risks and subsequent illness risks using measured HAdV concentrations for the three seasons for swimming and snorkelling exposure scenarios. While the use of indicator viruses is routine in some parts of the world, viral indicators are not used in the assessment of water quality in Mexico. Additionally, there are few studies on the recreational water quality in the Caribbean despite high recreational water use, especially in tourism seasons. This study will generate information that helps inform Mexican public policies regarding the generation of warnings about risks from recreational activities using new indicators.
METHODS
Water collection
Map showing the location of the sampling sites within the Yal-ku lagoon in the coastal area of the state of Quintana Roo, Mexico.
Map showing the location of the sampling sites within the Yal-ku lagoon in the coastal area of the state of Quintana Roo, Mexico.
Enumeration of somatic and male F + -specific coliphages
Somatic and F + -specific coliphage densities were evaluated using the double layer plaque assay method (EPA Office of Water 2015) as previously described by Rosiles-González et al. (2017). Escherichia coli host bacterial strains were obtained from the American Type Culture Collection (ATCC) (Manassas, VA). Somatic coliphages were enumerated using strain ATCC 700609 CN13 (Nalidixic acid-resistant mutant of ATCC 13706) as hosts. Male F + -specific coliphages were enumerated from 10 mL of water (before ultrafiltration) using E. coli strain 15597 C-300 as hosts as described in Rosiles-González et al. (2019).
Concentration of water samples and nucleic acid isolation
Water samples were subjected to ultrafiltration by using hollow fibre Fresenius hemoflow F80A polysulfone dialysis filters with 15,000–20,000 molecular weight cut-offs (MWCOs) (Fresenius Medical Care, Lexinton, MA, USA) following the protocol as described by Rosiles-González et al. (2017). Briefly, concentrated water samples of approximately 500 mL were further processed by a second concentration step using polyethylene glycol precipitation (PEG 8,000, Sigma, Saint Louis, MI, USA). The final water sample concentrates of 3.5–12.4 mL were stored at –20 °C. Nucleic acids were isolated from water concentrates using 280 and 400 μL aliquots, for RNA and DNA, respectively, using the QIAmp viral RNA and DNA MiniKits (QIAGEN, Hilden, Germany) following the manufacturer's recommendations. The final eluate volumes of 50 μL were stored at −20 °C
Viral quantification by real-time PCR
Quantification of HAdVs and PMMoV was obtained using a 96-well plate format in a CFX96 Touch™ Real-Time PCR Detection System, from Bio-Rad Laboratories, Hercules, CA, USA. Adenoviruses were quantified by qPCR using a set of primers that amplify a 132 base pair fragment of the HAdV hexon gene following the primers and procedure reported by Heim et al. (2003). HAdV quantifications were obtained using a 25 μL final reaction mix comprising the SsoAdvanced Universal SYBR Green Supermix (Bio-Rad, Hercules, CA, USA). The reaction mix consisted of 1 μL of total DNA (>5 ng μL−1), 12.5 μL 2× supermix, 200 nmol L−1 forward and reverse primers, respectively, and 9.5 μL of water. A melting curve was performed at the end of the run by 0.05 s cycles each at 0.5 °C increments between 65 and 95 °C.
PMMoV was quantified by using qPCR with primers described by Haramoto et al. (2013) that amplify a 68 bp fragment of the PMMoV replication associated protein. cDNA was obtained from 5 μL of the total RNA using the iScript Advanced cDNA Synthesis kit for RT-qPCR (Bio-Rad, Hercules, CA. USA), following the manufacturer's instructions. The reactions were carried out with a 25 μL final volume containing: 5 μL of cDNA, 2.5 μL 10× reaction buffer, 50 mmol L−1 MgCl2, 10 mmol L−1 deoxynucleotide triphosphate (dNTP), 500 nmol L−1 forward and reverse primers, respectively, 250 nmol L−1 of the probe, and 5 U μL−1 of the DNA polymerase.
Negative controls were prepared for each experiment by using polymerase chain reaction (PCR)-grade water as a template. A standard curve was obtained for each experiment by using serial 10-fold dilutions of plasmid DNA (100–105 DNA genome copies) containing each viral target gene. The limit of detection (LoD) was calculated for each target gene following the method described by Hougs et al. (2019) and determined to be <180–430 copies/L for HAdV and <76–120 copies/L for PMMoV.
To determine the occurrence of the viral genetic material under study, a sample was considered positive when at least one replicate was above the LoD, all samples below the LoD were considered negative.
Nested PCR of HAdV, cloning and sequence analysis
Nested PCR was conducted according to the primers and cycling conditions published by Allard et al. (2001). Briefly, the first round of nested PCR was conducted using 5 μL of DNA as template to amplify a 301 bp fragment of the hexon gene. For the second round, 2 μL of the first PCR product was used as a template to amplify an expected fragment of 171 bp in size. PCR-grade water was used as a template in negative control reactions. Amplifications were conducted using the GoTaq DNA polymerase (Promega, Madison, WI) in 50 μL final volume per reaction. The reaction mix consisted of 10 μL Green GoTaq buffer, 20 μmol L−1 of each forward and reverse primer, 10 nmol L−1 of each dNTP, and 1.25 U μL−1 of DNA polymerase. Amplicons of expected size were cloned into the pGEM-T Easy plasmid vector (Promega, Madison, WI) following the manufacturer's instructions. E. coli TOP 10 cells were used for transformation with recombinant plasmids by the heat shock method (Sambrook & Russell 2001). A total of five positive clones per sample were selected from EcoRI screening and subjected to sequencing at Macrogen, Seoul, Korea, using the M13 universal forward primer.
A total of 68 sequences were edited using FinchTV 1.4.0 (Geospiza, Inc.; Seattle, WA, USA; http://www.geospiza.com) to remove remaining cloning vector sequences. For viral typing, sequences were compared against available sequences at the NCBI database (http://www.ncbi.nom.nih.gov) using a BLASTn search. The percentage of nucleotide identity of each isolated partial hexon gene sequence was recorded as the result of the comparison with the first hit against the available GenBank sequences.
Statistical analyses
The correlation coefficients were calculated using R version 4.2.0 (R Core Team 2021) to evaluate the associations between somatic and F + -specific coliphage concentrations per season, between sampled sites and through sampled year, besides evaluating the associations between PMMoV and HAdV concentrations in both sites. A Shapiro–Wilk normality test was performed to evaluate the normality of the viral concentration data to determine if Pearson (parametric) or Spearman (non-parametric) coefficients would be used.
Quantitative microbial risk assessment
We utilized QMRA (Haas et al. 2014) to estimate adenovirus infection risks from 1 h of swimming or snorkelling activities in the Yal-ku lagoon for three seasons (dry, rainy, and north). A Monte Carlo approach was used to account for the variability and uncertainty in adenovirus concentrations, the fraction of genome copies assumed to represent infectious viral particles, the rate of water ingestion during recreational activities. The number of model iterations needed to stabilize the estimated central tendencies of adenovirus infection risk for each season was explored, and 100,000 iterations was considered sufficient (Table S1).
To calculate concentrations for a single sample that was processed in duplicate, the concentrations were averaged, where values below the LOD were replaced with a value randomly sampled between zero and the LOD for that sample replicate, assuming a uniform distribution. The calculations for limits of detection are in the Supplementary Material. The replacement of concentrations below the limits of detection was done 10,000 times per data set to create 10,000 separate sets of concentration data. These sets were then combined and empirically sampled with replacement in the overall exposure model. For samples that had no amplification, their concentrations were considered to be true zeroes and were not replaced. The samples and their respective limits of detection can be seen in Table 1.
HAdV and PMMoV concentrations (genome copies/L) by location, month, and seasona
. | . | . | Human adenovirus (HAdV) . | Pepper mild mottle virus (PMMoV) . | ||
---|---|---|---|---|---|---|
. | . | . | Site 1 . | Site 2 . | Site 1 . | Site 2 . |
Season and collection date . | Real-time PCR replicates . | Genome copies per litre (GC L−1) . | ||||
North season | February 2018 | R1 | 480 | 500 | 320 | 240 |
R2 | 230 | 290 | 130 | 84 | ||
Dry season | March 2018 | R1 | 580 | <180 | <92 | <100 |
R2 | 680 | <180 | <92 | <100 | ||
April 2018 | R1 | 250 | 0 | <78 | <76 | |
R2 | 180 | 0 | <78 | <76 | ||
May 2018 | R1 | 0 | 130 | <83 | <120 | |
R2 | 0 | <220 | <83 | <120 | ||
Rainy season | June 2018 | R1 | 1,000 | 960 | <100 | <100 |
R2 | 970 | 740 | <100 | <100 | ||
August 2017 | R1 | <430 | 630 | 320 | 320 | |
R2 | <430 | 980 | 220 | 300 | ||
September 2017 | R1 | <390 | 210 | 370 | 44 | |
R2 | <390 | <260 | 0 | 87 | ||
October 2017 | R1 | 0 | <200 | 75 | 150 | |
R2 | 0 | <200 | 0 | 99 | ||
North season | November 2017 | R1 | 140 | 290 | 70 | 150 |
R2 | 180 | 290 | 130 | 130 | ||
December 2017 | R1 | 64 | 0 | 88 | 63 | |
R2 | 180 | 0 | <93 | <78 |
. | . | . | Human adenovirus (HAdV) . | Pepper mild mottle virus (PMMoV) . | ||
---|---|---|---|---|---|---|
. | . | . | Site 1 . | Site 2 . | Site 1 . | Site 2 . |
Season and collection date . | Real-time PCR replicates . | Genome copies per litre (GC L−1) . | ||||
North season | February 2018 | R1 | 480 | 500 | 320 | 240 |
R2 | 230 | 290 | 130 | 84 | ||
Dry season | March 2018 | R1 | 580 | <180 | <92 | <100 |
R2 | 680 | <180 | <92 | <100 | ||
April 2018 | R1 | 250 | 0 | <78 | <76 | |
R2 | 180 | 0 | <78 | <76 | ||
May 2018 | R1 | 0 | 130 | <83 | <120 | |
R2 | 0 | <220 | <83 | <120 | ||
Rainy season | June 2018 | R1 | 1,000 | 960 | <100 | <100 |
R2 | 970 | 740 | <100 | <100 | ||
August 2017 | R1 | <430 | 630 | 320 | 320 | |
R2 | <430 | 980 | 220 | 300 | ||
September 2017 | R1 | <390 | 210 | 370 | 44 | |
R2 | <390 | <260 | 0 | 87 | ||
October 2017 | R1 | 0 | <200 | 75 | 150 | |
R2 | 0 | <200 | 0 | 99 | ||
North season | November 2017 | R1 | 140 | 290 | 70 | 150 |
R2 | 180 | 290 | 130 | 130 | ||
December 2017 | R1 | 64 | 0 | 88 | 63 | |
R2 | 180 | 0 | <93 | <78 |
aConcentrations below limits of detection are indicated by < followed by the LOD. R, replicate. Bolded values indicate values above 0 or limits of detection.
The fraction of genome copies/L assumed to be infectious was informed by two sources, one of which reports a range of 1:100– 1:50,000, another of which reports the use of 1:700 infectious particles to genome copies for adenovirus, specifically (Table 2) (Sinclair et al. 2008; Federigi et al. 2019). The rate of water ingestion for swimming was represented by a triangular distribution, informed by ingested volumes from Dorevitch et al. (2011). The mean and upper confidence limit volumes informed the mode and maximum, but due to lack of a lower bound value, a volume of 0.5 mL was assumed as the minimum (Table 2).
QMRA parameter information
Parameter . | Variable . | Units . | Distribution/point value . | Source . |
---|---|---|---|---|
Fraction of genome copies assumed to represent infectious virus | ![]() | Infectious viral particles/mL/gc/mL | Uniform (min = 2 × 10−5, max = 1 × 10−2) | Federigi et al. (2019) and Sinclair et al. (2008) |
Volume of water ingested when swimming | ![]() | mL/h | Triangular (min = 0.5, mode = 10, max = 34.8) | Dorevitch et al. (2011) |
Volume of water ingested when emptying snorkel tube/coming up for air once | ![]() | mL/emptying event | Triangular (min = 4, mode = 12.75, max = 168) | Hitchings et al. (2013) and Lawless et al. (2003) |
Number of times snorkel tube is emptied per hour of snorkelling | ![]() | Emptying event/h | Uniform (min = 1, max = 120) | Assumed |
Duration of activity | - | h | 1 | Assumed |
Dose–response parameters for probability of adenovirus infection | ![]() | – | 5.11 | Teunis et al. (2016) |
![]() | – | 2.80 | Teunis et al. (2016) | |
Dose–response parameters for probability of illness given adenovirus infection | ![]() | – | 6.53 | Teunis et al. (2016) |
![]() | – | 0.41 | Teunis et al. (2016) |
Parameter . | Variable . | Units . | Distribution/point value . | Source . |
---|---|---|---|---|
Fraction of genome copies assumed to represent infectious virus | ![]() | Infectious viral particles/mL/gc/mL | Uniform (min = 2 × 10−5, max = 1 × 10−2) | Federigi et al. (2019) and Sinclair et al. (2008) |
Volume of water ingested when swimming | ![]() | mL/h | Triangular (min = 0.5, mode = 10, max = 34.8) | Dorevitch et al. (2011) |
Volume of water ingested when emptying snorkel tube/coming up for air once | ![]() | mL/emptying event | Triangular (min = 4, mode = 12.75, max = 168) | Hitchings et al. (2013) and Lawless et al. (2003) |
Number of times snorkel tube is emptied per hour of snorkelling | ![]() | Emptying event/h | Uniform (min = 1, max = 120) | Assumed |
Duration of activity | - | h | 1 | Assumed |
Dose–response parameters for probability of adenovirus infection | ![]() | – | 5.11 | Teunis et al. (2016) |
![]() | – | 2.80 | Teunis et al. (2016) | |
Dose–response parameters for probability of illness given adenovirus infection | ![]() | – | 6.53 | Teunis et al. (2016) |
![]() | – | 0.41 | Teunis et al. (2016) |










A sensitivity analysis was conducted using Spearman correlation coefficients (Bivins et al. 2017; Hamilton et al. 2019) to quantify relationships between model inputs and adenovirus infection risk for each season, where a Spearman correlation coefficient of a larger magnitude indicated a larger influence.
RESULTS
Somatic and F + -specific coliphage densities
F + -specific and somatic coliphage concentrations determined per month and season at the Yal-ku lagoon, in sites 1 and 2. Absent bars indicate concentrations below detection limits.
F + -specific and somatic coliphage concentrations determined per month and season at the Yal-ku lagoon, in sites 1 and 2. Absent bars indicate concentrations below detection limits.
Occurrence of HAdV and PMMoV
The qPCR amplification of HAdV from at least one replicate per sample resulted in the detection of the viral genetic material in 60% (12/20) of the samples collected in both sites of the Yal-ku lagoon (Table 1). The highest concentrations of HAdV were detected in June (rainy season) in site 1 with 1,000 GC/L (R1) and 970 GC/L (R2). Overall, the concentrations of HAdV detected in the Yal-ku lagoon were variable ranging from 101 to 103 GC/L, and the genetic material of the virus was detectable in at least one site throughout all climatic seasons (Table 1). The qPCR amplification of the PMMoV from at least one replica per sample resulted in the detection of the genetic material in 60% (12/20) of the samples collected in both sites of the Yal-ku lagoon during the rainy and north seasons (Table 1). The highest concentrations of PMMoV were detected in September (rainy season) in site 1 with 370 CG/L (R1) (Table 1). Overall, the highest average concentrations of HAdV and PMMoV were detected in site 1, being 246 and 86 GC/L, respectively. There was no significant correlation between the presence of PMMoV and HAdV in the samples collected in the Yal-ku lagoon.
HAdV typing
A total of 68 HAdV partial hexon sequences of 171 bp (29 isolated from site 1 and 39 from site 2) were obtained from the Yal-ku lagoon, from which 59% (40/68) shared 96–100% of nucleotide (nt) identity with HAdV species F serotype 40 (Accession numbers MH289584, KY859455), and 41% (28/68) shared 97–99.4% of nt identity with HAdV species F serotype 41 (Accession numbers MH289551, MK420404).
QMRA
Estimated dose and infection risk summary statistics by season (n = 100,000 per season)
Season . | Median (IQR) . | Mean (SD) . | Min, Maxa . |
---|---|---|---|
Snorkelling | |||
Adenovirus dose | |||
Dry | 5.6 × 10−1 (3.1 × 100) | 3.3 × 100 (7.5 × 100) | 0.0, 1.1 × 102 |
North | 1.6 × 100 (5.0 × 100) | 4.1 × 100 (6.4 × 100) | 0.0, 7.1 × 101 |
Rainy | 2.3 × 100 (7.8 × 100) | 7.8 × 100 (1.4 × 101) | 0.0, 1.9 × 102 |
Adenovirus infection risk | |||
Dry | 3.0 × 10−2 (8.5 × 10−1) | 4.1 × 10−1 (4.0 × 10−1) | 0.0, 1.0 × 100 |
North | 6.4 × 10−1 (8.0 × 10−1) | 5.6 × 10−1 (3.9 × 10−1) | 0.0, 1.0 × 100 |
Rainy | 7.5 × 10−1 (7.6 × 10−1) | 6.1 × 10−1 (3.9 × 10−1) | 0.0, 1.0 × 100 |
Illness risk | |||
Dry | 1.0 × 10−2 (1.3 × 10−1) | 8.6 × 10−2 (1.3 × 10−1) | 0.0, 6.9 × 10−1 |
North | 5.6 × 10−2 (2.0 × 10−1) | 1.2 × 10−1 (1.4 × 10−1) | 0.0, 6.4 × 10−1 |
Rainy | 8.6 × 10−2 (2.7 × 10−1) | 1.6 × 10−1 (1.8 × 10−1) | 0.0, 7.5 × 10−1 |
Swimming | |||
Adenovirus dose | |||
Dry | 4.2 × 10−3 (1.6 × 10−2) | 1.3 × 10−2 (2.4 × 10−2) | 0.0, 2.0 × 10−1 |
North | 1.0 × 10−2 (2.2 × 10−2) | 1.7 × 10−2 (1.9 × 10−2) | 0.0, 1.3 × 10−1 |
Rainy | 1.3 × 10−2 (3.7 × 10−2) | 3.1 × 10−2 (4.4 × 10−2) | 0.0, 3.2 × 10−1 |
Adenovirus infection risk | |||
Dry | 2.7 × 10−3 (1.0 × 10−2) | 8.5 × 10−3 (1.5 × 10−2) | 0.0, 1.2 × 10−1 |
North | 6.7 × 10−3 (1.4 × 10−2) | 1.1 × 10−2 (1.2 × 10−2) | 0.0, 7.9 × 10−2 |
Rainy | 8.6 × 10−3 (2.3 × 10−2) | 2.0 × 10−2 (2.7 × 10−2) | 0.0, 1.9 × 10−1 |
Illness risk | |||
Dry | 7.1 × 10−7 (9.7 × 10−6) | 2.9 × 10−5 (1.0 × 10−4) | 0.0, 1.5 × 10−3 |
North | 4.4 × 10−6 (2.3 × 10−5) | 2.5 × 10−5 (5.3 × 10−5) | 0.0, 6.2 × 10−4 |
Rainy | 7.2 × 10−6 (6.3 × 10−5) | 1.1 × 10−4 (2.8 × 10−4) | 0.0, 3.7 × 10−3 |
Season . | Median (IQR) . | Mean (SD) . | Min, Maxa . |
---|---|---|---|
Snorkelling | |||
Adenovirus dose | |||
Dry | 5.6 × 10−1 (3.1 × 100) | 3.3 × 100 (7.5 × 100) | 0.0, 1.1 × 102 |
North | 1.6 × 100 (5.0 × 100) | 4.1 × 100 (6.4 × 100) | 0.0, 7.1 × 101 |
Rainy | 2.3 × 100 (7.8 × 100) | 7.8 × 100 (1.4 × 101) | 0.0, 1.9 × 102 |
Adenovirus infection risk | |||
Dry | 3.0 × 10−2 (8.5 × 10−1) | 4.1 × 10−1 (4.0 × 10−1) | 0.0, 1.0 × 100 |
North | 6.4 × 10−1 (8.0 × 10−1) | 5.6 × 10−1 (3.9 × 10−1) | 0.0, 1.0 × 100 |
Rainy | 7.5 × 10−1 (7.6 × 10−1) | 6.1 × 10−1 (3.9 × 10−1) | 0.0, 1.0 × 100 |
Illness risk | |||
Dry | 1.0 × 10−2 (1.3 × 10−1) | 8.6 × 10−2 (1.3 × 10−1) | 0.0, 6.9 × 10−1 |
North | 5.6 × 10−2 (2.0 × 10−1) | 1.2 × 10−1 (1.4 × 10−1) | 0.0, 6.4 × 10−1 |
Rainy | 8.6 × 10−2 (2.7 × 10−1) | 1.6 × 10−1 (1.8 × 10−1) | 0.0, 7.5 × 10−1 |
Swimming | |||
Adenovirus dose | |||
Dry | 4.2 × 10−3 (1.6 × 10−2) | 1.3 × 10−2 (2.4 × 10−2) | 0.0, 2.0 × 10−1 |
North | 1.0 × 10−2 (2.2 × 10−2) | 1.7 × 10−2 (1.9 × 10−2) | 0.0, 1.3 × 10−1 |
Rainy | 1.3 × 10−2 (3.7 × 10−2) | 3.1 × 10−2 (4.4 × 10−2) | 0.0, 3.2 × 10−1 |
Adenovirus infection risk | |||
Dry | 2.7 × 10−3 (1.0 × 10−2) | 8.5 × 10−3 (1.5 × 10−2) | 0.0, 1.2 × 10−1 |
North | 6.7 × 10−3 (1.4 × 10−2) | 1.1 × 10−2 (1.2 × 10−2) | 0.0, 7.9 × 10−2 |
Rainy | 8.6 × 10−3 (2.3 × 10−2) | 2.0 × 10−2 (2.7 × 10−2) | 0.0, 1.9 × 10−1 |
Illness risk | |||
Dry | 7.1 × 10−7 (9.7 × 10−6) | 2.9 × 10−5 (1.0 × 10−4) | 0.0, 1.5 × 10−3 |
North | 4.4 × 10−6 (2.3 × 10−5) | 2.5 × 10−5 (5.3 × 10−5) | 0.0, 6.2 × 10−4 |
Rainy | 7.2 × 10−6 (6.3 × 10−5) | 1.1 × 10−4 (2.8 × 10−4) | 0.0, 3.7 × 10−3 |
aIt is acknowledged that risk is never zero, but rather extremely low in cases where samples result in no amplification (assumed to be zero concentrations in this study). Therefore, minimums should be interpreted with caution.
Spearman correlation coefficients
Activity . | Model parameter . | Dry season . | Rainy season . | North season . |
---|---|---|---|---|
Swimming | ![]() | 0.88 | 0.78 | 0.66 |
![]() | 0.27 | 0.42 | 0.49 | |
![]() | 0.18 | 0.28 | 0.34 | |
Snorkelling | ![]() | 0.84 | 0.69 | 0.59 |
![]() | 0.22 | 0.35 | 0.37 | |
![]() | 0.30 | 0.48 | 0.51 |
Activity . | Model parameter . | Dry season . | Rainy season . | North season . |
---|---|---|---|---|
Swimming | ![]() | 0.88 | 0.78 | 0.66 |
![]() | 0.27 | 0.42 | 0.49 | |
![]() | 0.18 | 0.28 | 0.34 | |
Snorkelling | ![]() | 0.84 | 0.69 | 0.59 |
![]() | 0.22 | 0.35 | 0.37 | |
![]() | 0.30 | 0.48 | 0.51 |
Distributions of non-zero illness risks by season and activity relative to a 32/1,000 recreational illness risk benchmark*. *Horizontal lines indicate the 25th, 50th, and 75th percentiles.
Distributions of non-zero illness risks by season and activity relative to a 32/1,000 recreational illness risk benchmark*. *Horizontal lines indicate the 25th, 50th, and 75th percentiles.
DISCUSSION
Key findings
Faecal indicators (i.e., coliphages and PMMoV) were detected in 100% (coliphages) and 60% (PMMoV) of the samples, indicating that human faecal contamination is likely present in the Yal-ku lagoon throughout the different climatic seasons. Moreover, HAdV genetic material was detected in 60% of the samples of site 1, which is where the people can access the entrance of the swimming area, and in site 2, where swimming and snorkelling activities are often conducted. Recent reports show the occurrence of different enteric viruses in groundwater (Rosiles-González et al. 2019) and in coastal areas of the Yucatan peninsula (Muñoz-Cortés et al. 2023), clearly highlighting potential public health concerns. In Mexico, as in other countries, surveillance of viruses in the environment is not common nor included in water quality standards for bathing waters, and this is the first study in the area that estimates adenovirus infection risks, where adenovirus is one of the most common enteric viruses found in the aquatic environment (Rames et al. 2016).
QMRA-estimated adenovirus infection risks for snorkelling activities were approximately one to two orders of magnitude greater than that of swimming (Table 3). While there is no risk benchmark for recreational water use in Mexico, mean illness risks for snorkelling were above the US EPA benchmark of 32/1,000 while all estimate illness risks for swimming were below this threshold (Table 3). However, adenovirus is not the only potential human-sourced or animal-sourced faecal pathogen present, meaning that (1) adenovirus exposure, alone, may contribute notably high risk to snorkelling recreators and (2) adenovirus illness risk in combination with risks from other gastrointestinal illness-associated pathogens could yield overall illness risks greater than the 32/1,000 risk benchmark for swimmers. An uncertainty related to this conclusion is that little is known about the true ingestion volumes during snorkelling. More data regarding this exposure factor would instil more confidence in differences in estimated infection risk and illness risk across the two activities in our study. It should also be noted that a benchmark designed for recreational activities in the USA is likely not applicable to Mexico due to varying international approaches to recreational water quality monitoring and maintenance and the potential differences in populations and frequency of recreational activity.
Limitations
One of the limitations of the QMRA model is lack of human behaviour data to inform an estimation of adenovirus infection and illness risk beyond a single snorkelling or swimming event. Accounting for the frequency of recreational activities would allow for estimates of annual risk and more insights regarding disease burden at the population level. Additionally, there is a lack of data describing the volume of water ingested during snorkelling activities. While we utilized data to estimate this volume, data on the true volumes of water ingested and how this varies by age (children vs. adults, for example) or by skill or experience level would instil more confidence in the estimated adenovirus infection risks.
A common limitation in environmental microbiology studies is having data below detectable limits. This limitation was present in this study. While we utilized methods for addressing left-censored data in the QMRA model, there are still uncertainties regarding the true concentration of HAdV genetic material in the sampled recreational waters. We assumed a uniform distribution for concentrations below detection limits, but it is possible these concentrations abide by a lognormal or other distribution. Ten of 40 adenovirus qPCR results were below the detection limit in our study, meaning we observed 25% left-censoring, considered medium in terms of severity of left-censoring (Canales et al. 2018). The most left-censoring was observed for the rainy season (seven out of 16 samples) (Table 2), meaning the impact of decisions for handling left-censoring predominantly affected data for this season relative to others. The use of a uniform distribution as opposed to lognormal for replacing these values means larger concentrations between 0 and the LOD had a greater probability of being selected in our current approach than what is truly observed in samples. This would then indicate that estimated concentrations for those below the detection limit could be overestimates (especially for the rainy season), resulting in overestimated adenovirus infection and illness risks. However, more protective estimates of adenovirus infection and illness risk are more appropriate in the face of uncertainties about the true distribution of concentrations for virus in the Yal-ku lagoon.
In addition, the concentration of the genetic material of viruses using analytical methods such as qPCR does not reflect the actual infectious viral particles (Jiang et al. 2001). Therefore, we used a ratio to represent genome copies per infectious viral particle, as in other similar studies (Carducci et al. 2018). This is a notable area of uncertainty with large effects on estimated adenovirus infection risk (Figure S2), and, subsequently, illness risk. However, there are no data, to our knowledge, that currently fill this gap or provide a better range than what was used in this study. More data are needed in this area to not only advance this specific QMRA model but QMRA as an entire field. While the use of this ratio limits our confidence in estimates of absolute adenovirus infection or illness risk, comparative differences across seasons and activities still hold, providing important insights for seasonal or activity-specific risks for recreators in the Yal-ku lagoon.
A QMRA was conducted to estimate individual infection and illness risks for recreational events due, in part, to lack of epidemiological data describing these risks. QMRAs benefit from validation with epidemiological data, but a common challenge is that the data for validation are lacking in the first place, making a risk assessment approach a useful tool for estimating potential disease burden. This is a common limitation for QMRA models. It should be noted that the illness risks estimated in our study are especially small for swimming and would require 10,000–1,000,000 people to observe illness, making it unlikely that risks of this magnitude would be captured with epidemiological approaches. However, illness risks for snorkelling were notably higher, where a case would theoretically be observed with hundreds to thousands of individuals. More data are needed regarding cases linked to recreational water activity and whether these cases arise among local residents or visitors recreating in the Yal-ku lagoon.
CONCLUSIONS
There is likely human faecal contamination in the Yal-ku lagoon that could pose risks to recreational swimmers of adenovirus infections, especially in rainy seasons. More data from quantitative research are needed to confirm human faecal contamination sources, whether it be from the tourism industry itself (i.e., hotels/restaurants), already infected tourists (faecal contamination from submergence), or the community (septic tanks). Without any federally legal regulation of acceptable risk parameters for gastrointestinal illness in recreational areas, there is little to no oversight which will motivate intervention or recognition of public health risk. Ultimately the best solution for the enteric virus contamination observed in this site is governmental guidance and support on how to monitor recreational water bodies to protect public health.
ACKNOWLEDGEMENTS
We want to give special thanks to Dr Gabriel Sánchez Rivera and Dr Hector Antonio Lizarraga Cubedo from Centro Ecológico de Akumal (CEA) for their collaboration in water collection. Luis Jorge Negrete Alcalde was supported by a Master's degree scholarship (number 619259) from Consejo Nacional de Ciencia y Tecnología (Conacyt). This project (number 216093) was funded by Consejo Nacional de Ciencia y Tecnología (Conacyt)/‘Proyectos de Desarrollo Científico para Atender Problemas Nacionales’. A.M. Wilson was supported by a University of Arizona Health Sciences Career Development Award and is a member of the Southwest Environmental Health Sciences Center (NIEHS P30 ES006694). The publication's contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICTS OF INTEREST
The authors declare no conflict of interest.