Passive sampling as an alternative strategy to monitor metals and PAHs trends at an upstream and rural catchment: a French case study

The study presents an alternative strategy to conventional spot sampling for monitoring metals and polycyclic aromatic hydrocarbons (PAHs) at an upstream, rural and karstic catchment in the north eastern part of France, in order to get insight into their spatial and temporal variability. Passive samplers, as diffusive gradients in thin films (DGT) and semipermeable membrane device (SPMD), are monthly deployed from August 2012 to March 2016 at five of the catchment monitoring stations located on the Saulx and Ornain Rivers. An improvement of the frequency of quantification (by a factor 2 to 8, depending on the targeted compound) is observed allowing us to better identify spatial and temporal variability. For instance, the upstream monitoring station on the Saulx River is characterized by high concentrations of Ni and Mn whereas the upstream monitoring station on the Ornain River is enriched in Cu and Zn. Furthermore, five metals (Al, Fe, Mn, Ni, Zn) and three PAHs (fluoranthene, pyrene and chrysene) show significant variations with water levels when grouped in three categories (low, medium and high water levels) in relation with hydrological and climatic patterns. This study leads to a more accurate assessment of the background pollution in metals and PAHs within surface waters than when based on spot sampling data. Optimisation of surface water quality monitoring program to take into account variability of pollutants. Benefit of using passive samplers for monitoring metals and PAHs in river waters. Assessment of background pollution as well as temporal and spatial variations more accurately. Optimisation of surface water quality monitoring program to take into account variability of pollutants. Benefit of using passive samplers for monitoring metals and PAHs in river waters. Assessment of background pollution as well as temporal and spatial variations more accurately.


Introduction
Anthropogenic sources (urban, industrial and agricultural activities) as well as natural processes (precipitation inputs, erosion and chemical weathering) greatly influence water quality of surface and ground waters. Due to spatial and temporal variations in water chemistry, a monitoring program providing a representative and reliable estimation of the quality of surface and ground waters is necessary to support decision making, in particular whether to implement a cost-effective set of measures to achieve load reduction of targeted substances.
In that view, water quality monitoring programs have been designed and regularly updated to take into account regulatory requirements (e.g. Water Framework Directive, WFD [1]). Most water quality monitoring programs are classically based on spot samples collection at a stated frequency, followed by laboratory analysis. To set the sampling frequency, the expected fluctuation with time of each parameter to be monitored should be considered. For practical reasons, as well as cost issues, a sampling frequency of 4 to 12 per year, equally distributed over the year, is often chosen [2][3][4]. In a similar way, the number of monitoring stations and their location should reflect the spatial variability of water quality parameters of interest [2,5].
In order to optimise on a regular basis the monitoring programs, data on the variability with time and space are usually insufficient and thus are strongly needed.
Typical drawbacks of spot sample based monitoring have been identified, such as a lack of representativeness for water bodies characterised by highly fluctuating concentration with time, the probability to miss an episodic pollution event, the need to collect large volumes of water to perform the analysis required at the low concentration level [6][7][8][9].
Alternative approaches have been developed since the last decades to overcome these drawbacks [6,7,10]. Among these alternative approaches are continuous measurements and passive sampling. Although continuous measurement is well adapted and often used for water quality parameters such as temperature, pH, electrical conductivity, dissolved oxygen and turbidity, it is either very costly for metals or not available for pollutants for PAHs, and pesticides. On the opposite, passive samplers have been found suitable for measuring pollutants, either in water or in sediments over time periods [6].
Accumulation over exposure period may give a more representative view of the contamination level of a water body. Moreover, the analytical challenge of analysing very low concentrations in bottle sample is reduced as passive samplers may serve as a pre-concentration tool [16,18,19].
A long-term environmental programme (OPE) was set up in 2007 by ANDRA (French National Radioactive Waste Management Agency) as a research tool for observing simultaneously various environmental media such as water, air, soil, flora, fauna and human activities within an upstream and rural catchment, in the north-eastern part of France (approx. 250 km from Paris) [20].
The OPE water quality monitoring program was optimised in 2011 in order to better take into account the variability in time and space of the water quality of the rivers within the OPE [21]. Seventeen monitoring stations were then selected to address the spatial variability of major components (constituting ions and nutrients) and the sampling frequency was set to six per year for these parameters. However, the optimising process didn't involve chemical substances such as metals and polycyclic aromatic hydrocarbons (PAHs) due to some difficulty encountered to assess their variability over time and space because of their poor quantification in spot samples.
In consequence, complementary monitoring was strongly recommended to investigate more thoroughly the water quality variability with time and space for metals and PAHs within the OPE surface waters using passive samplers to overcome this problem in an upstream catchment with low contamination background [21].
The aim of this work is to assess the ability of passive samplers to give insights on the variability of metals and PAHs with time and space in order to first evaluate their potential as an alternative way of monitoring these substances and second, if relevant, validate their use within the long term monitoring program of surface water within the OPE.
In Sect. 2, the study area as well as long-term monitoring design, including the selection of passive samplers, the targeted pollutants, protocols for handling, deploying, transporting and analysing passive samplers are described. Specific sub-sections are dedicated to quality controls (QCs), and time weighted average (TWA) concentration calculations.
Section 3 presents the results obtained from different point of view: first the frequencies of quantification in passive samplers versus spot samples are investigated, second the level of contamination in the OPE surface waters, with comparison with data from other European river basins, third the spatial and temporal variabilities are explored, in relation to local sources (agricultural and industrial), water level, turbidity and temperature.

Study area
The OPE covers a total area of 900 km 2 (Fig. 1). The site is located along the eastern Paris Basin (approximately 100 km west of Nancy, at the Haute-Marne and Meuse district), a region characterised by limestone plateau, with altitudes ranging between 300 and 400 m. The study area is mostly rural and composed of vast agricultural land dotted with wooded areas (roughly 12% meadows, 35% forests, 40% field crops). Human activity is predominantly agricultural with a few small industrial units (mainly carpentry, foundry and cheese). Finally, the climate is temperate oceanic with continental influence.
The most important streams are the Saulx River and the Ornain River, the latter being a tributary of the Marne River. All secondary tributaries to the Ornain and the Saulx (including Orge, Barboure, Ormançon, Ognon, Maldite) are intermittently flowing. Moreover, as theses rivers are part of a karstic hydrological system in which the subsystems of surface and underground waters are connected, the underground water flow being generally dominant, highly fluctuating water quality and quantity have been observed with time and space [21][22][23].
Passive samplers for monitoring metals and PAHs were deployed at five monitoring stations located on the Saulx and Ornain (Fig. 1). These five stations were selected to take into account spatial variability in water composition and flow rates as highlighted in a previous study [21]. Moreover, these monitoring stations have been instrumented since 2012 to measure continuously key water quality parameters such as temperature, electrical conductivity, dissolved oxygen, pH and turbidity, by means of a Hydrolab DS5 multiparameter probe and water level using an OTT CBS-compact bubbler sensor).

Objective and design
The main objective was to assess the ability of passive samplers to give insights on the variability of metals and PAHs with time and space, or in other word to evaluate their potential as an alternative way of monitoring these substances. For that purpose, passive samplers were deployed during 3 years to include inter-annual hydrological and climatic variations. Balancing quality of data and cost requirements was the main driving force for designing the on-going monitoring using passive samplers in OPE surface waters.
The selection of the passive samplers was driven by practical, operational and cost issues: (i) selected passive samplers should be fully commercially available, (ii) standard operation procedures should be available for deployment, handling and extraction, (iii) easily transferable to routine accredited laboratories.
As getting reliable measurements is a key issue for monitoring water quality, a study was conducted in 2011 in order to optimise the deployment period, identify if there was any impact of biofouling on measurements, estimate the measurement precision and define the necessary quality assurance and quality control (QA/QC) procedures and tools (blanks, performance reference compound (PRC) stability etc.) ( [30] and Supporting Information, Part A, for a summary of the results obtained).
The design consisted of specifying the protocols for passive sampler deployment and retrieval, transport conditions, laboratory extraction and analysis as well as QCs.
Moreover, one of the condition of using SPMD and DGT within the OPE water quality monitoring network was that it could be carried out by routine accredited laboratories and not expert ones. For that purpose, both selected laboratories (ASPECT S.E. and Synlab) were involved from the beginning of the project and were thoroughly trained for field deployment, extraction protocols for chemical analyses and PRCs spiking of SPMD.

Outline
DGT (Chelex 100, open pore purchased at DGT Research, Lancaster, UK) and SPMD (LPDE semi permeable membrane 91.4 cm × 2.5 cm containing 1 mL of triolein, purchased at Exposmeter AB, SE) were deployed at three monitoring stations on the Saulx River (from upstream to downstream: Saulx 1, Saulx 2 and Saulx 3) and at two monitoring stations on the Ornain River (from upstream to downstream: Ornain 1 and Ornain 2) from August 2012 to March 2016. One SPMD and two DGTs were deployed continuously at each monitoring station for periods of 4 weeks.

Handling, deployment, retrieval and transport protocols
Procedures for handling, deployment, retrieval, transport and storage were adapted from DGT Research [31] and USGS recommendations [32][33][34] as well as guidelines from Aquaref (French Reference Laboratory in support of aquatic monitoring) [35,36].
To sum up, DGTs were stored in a refrigerator (4 ± 3 °C) in the laboratory and transported at cool temperature using blue ice packs (5 ± 3 °C). Precautions for handling included removing the DGT from the sealed bag at the last minute, rinsing DGT on retrieval with deionized water and placing it in the sealed bag rapidly afterward.
The SPMD were stored in a freezer (− 20 ± 5 °C) before being spiked with 5 PRCs compounds selected to be representative of the variety of 15 PAHs targeted in this study [acenaphtene D8, fluorene D10, anthracene D10, fluoranthene D10 and benzo(g, h, i)perylene D12], 24 h before field deployment. SPMD were transported in the field in airtight metal cans at a temperature below 0 °C using dry ice. Precautions for handling included removing the SPMD from the metal can at the last minute, rinsing SPMD on retrieval with deionized water and placing it in the metal can rapidly afterward.

Analysis
After deployment, the chelex resin was removed from the DGT and extracted in 1.5 mL of nitric acid 1 M for 24 h. The acid extract was diluted to 10 mL in ultrapure water and the analysis of metals was performed by ICP-AES using the standard method EN ISO 11885 [37]. The mass of metal was then calculated taking into account the volume of the gel (0.16 mL) as recommended by [38]. The limits of quantification of metals in DGT were 2.5 ng for Cd, 5 ng for Cr, Mn and Pb, and 25 ng for Al, Cu, Fe, Ni and Zn. The analytical expanded uncertainty (with a coverage factor k = 2) was ranging from 15 to 50%, depending on the metal and the level of concentration.
The SPMD was extracted with heptane for 48 h. The extract was then purified on a silica column and then analysed by GC-MS following a standard method derived from the EPA 610 method, Appendix A [39]. The limit of quantification for PAHs in SPMD was 0.1 µg. The analytical expanded uncertainty (with a coverage factor k = 2) was 50%.
The analytical results were expressed as mass measured in passive sampler.

Quality controls
The following QCs were implemented for both DGT and SPMD, taking into account recommendations from [32,33]: -For assessing any contamination from laboratory and field conditions 1 laboratory blank per laboratory extraction and 1 field blank per field deployment period. -For assessing PRC spiking process and stability of PRC over the deployment period 1 laboratory blank spiked with PRC analysed after 24 h (PRC spiking process) and 1 laboratory blank spiked with PRC analysed after 4 weeks (PRC stability during 28 days stored at − 20 °C) -For estimating precision of the overall process (sampling and analysis) on a regular basis triplicate were deployed once a year at each of the five monitoring stations. In order to take into account hydrological and climatic conditions, these triplicates were carried out at different time during the year.
The results of these QC are summarised in Supporting Information (Part B).

Estimation of time weighted average concentrations
TWA (µg/L) concentrations of labile metals (corresponding to free metal ions and labile complexes) were calculated using formulas from [36,38], considering a 80% elution rate: where Δg is the diffusive gel thickness (0.08 cm), A is the exposure area (3.14 cm 2 ), D is the diffusive coefficient of the metal (in cm 2 /s, provided by DGT Research at 25 °C), M is the mass of metal accumulated in the resin and measured by analytical method (ng), t the deployment time (s).
The diffusion coefficients D are temperature dependent and can be calculated: The mean temperature for each time-period was estimated using continuous data at each of the five monitoring stations.
Moreover, the mass of metals measured in DGT was corrected taking into account the level of contamination in the DGT (intrinsic contamination of the chelex resin) and the contribution of the analytical process. Indeed, blanks subtraction might be performed providing that limited variation between devices is demonstrated, otherwise, it should be integrated in the LOQ estimation [40]. In this study, blank subtraction could be applied using mean values (disregarding outliers) of both laboratory and field blanks for each DGT batch (see Supporting Information, Part B).
To estimate the TWA concentrations of freely dissolved PAHs, the template provided by the United States Geological Survey (USGS) with PRC data was used [32,40]. In this study, only the PRCs with a concentration in the SPMD at the end of the deployment period of 20% to 80% of the initial spiked concentration were selected for the calculation. Indeed, a PRC loss higher than 80% indicates that this PRC as well as other compounds with similar physicochemical properties (e.g. similar octanol water partition coefficient K OW ) are in a nearly equilibrium state, whereas PRC loss lower than 80% signifies that the uptake is in a linear stage [29,33].
C water is the concentration in water available for accumulation in SPMD (ng/L), C SPMD is the concentration in the SPMD (ng/g), k u the accumulation rate constant (L/kg/day) and k e the elimination/exchange rate in 1/day. During exposure, PRC that are present at the beginning of deployment (C PRC 0 ), diffuse to the water. Their concentration in SPMD C PRC, SPMD is expressed as: Using the following empirical equation to estimate the equilibrium constant K SPMD The TWA LOQ were estimated as following: the TWA concentrations for each substance was calculated from the mass measured at the LOQ level in passive sampler (DGT and SPMD), at various climatic and hydrologic conditions. The average of the TWA concentrations obtained was then considered as the TWA LOQ for each substance.

Results and discussion
The results presented in this paper corresponded to monthly deployments of passives samplers for 3 years and a half, starting in August 2012 and ending in March 2016. Due to some difficulty encountered in purchasing DGT, resulting in a lack of data in 2013 (3 months) and 2014 (10 months) for metals.

Frequencies of quantification in passive samplers versus spot samples
Comparing TWA concentration of labile compounds with dissolved (< 0.45 µm) for metals or total concentration for PAHs in spot samples is rather inappropriate, as underlined by several authors [12,18], considering that the measurements do not reflect the same fraction of matrix and the same time scale. However, a comparison of frequencies of quantification highlighted the impact of lower limits of quantification reached through passive sampling. A direct consequence of lowering the limit of detection is to better investigate the temporal and spatial variability, with an increased number of quantified data available. If compared with the frequency of quantification in spot samples collected every two months from 2012 to 2016 at the five monitoring stations, it was obvious that metals such as Zn, Ni, Cr and Pb were more quantified in DGT, with 40% to 98% of quantification in DGT versus 5% to 27% of quantification in spot samples (Table 1). Concerning Cu, no improvement was observed using DGT. This may be linked to the speciation of Cu and its affinity with humic substances that can only partially diffuse through the gel due to their large size. The comparison for Al, Fe, and Mn was not possible, as these metals were not analysed in spot samples as they are not part of the OPE monitoring program which is based on WFD requirements [1]. Improvement of LOQ has also been reported by others, with a decrease of a factor 40 between the LOQ in spot samples and the LOQ in passive samplers [19].
In the same way, PAHs were more frequently quantified in SPMD than in spot samples, except for the 5-6 rings PAHs [benzo(a)pyrene, benzo(k)fluorenthene, indeno (1, 2, 3 cd)pyrene, dibenzo(a, h)anthracene, benzo(g, h, i) perilene] for which the PRC loss was less than 20%. For instance, phenanthrene, fluoranthene and pyrene were quantified in more than 89% of SPMD, whereas their frequency of quantification in spot samples was only 41-43% (Table 2).
Concerning PAHs containing 5-6 rings, the very low uptake could be the results of their speciation in water, in particular their link to dissolved organic matter that hindered them from being accumulated in the semi-permeable membranes, or a not enough deployment time period [11].
These results clearly demonstrated the great improvement in frequency of quantification when using passive samplers to monitor metals and PAHs at the OPE surface waters.
Similar conclusions with regards to improved frequency of quantification using passive samplers have been drawn by several authors for monitoring PAHs underlying the benefit of using passive samplers to have a better understanding of the periodic occurrence and fate of metals and PAHs with 3 to 4 rings in surface waters [13,18,26].

Level of contamination
The TWA concentrations measured in the Saulx and Ornain Rivers (Fig. 2) were found to be lower than TWA concentrations measured downstream on the Marne and Seine Rivers [11]. Using the same passive samplers as in this study, these authors observed significant concentration gradients in both metals and PAHs from up to downstream monitoring stations, due to accumulation of point source discharges and diffuse pollution.  For instance, TWA concentrations in Ni were generally found lower than 0.2 µg/L in the Saulx River and the Ornain River, whereas it ranged from roughly 0.3 to 0.7 µg/L from Theil on the Marne River to Andresy on the Seine River. Similarly, TWA concentrations in 3-4 rings PAHs, such as fluoranthene and pyrene were found to be mainly lower than 8 and 10 ng/L in this study whereas at the most downstream monitoring station on the Seine River (Andresy) their TWA concentrations could reach as high as 20-25 ng/L.
If these TWA concentrations are compared with other European rivers, similar observations can be drawn. The sum of PAHs ranged from 13 to 72 ng/L on the Danube River Basin [15], from 15 to up to 100 ng/L in the Seine River Basin [11,13], 55 ng/L in the Rhône River [18].
Concerning Ni, similar concentrations were measured in these rivers, ranging from 0.2 to 0.5 µg/L.

Spatial and temporal variability
A significant spatial variability was observed for four metals (Kruskal and Wallis test, p < 0.05): Mn, Ni, Cu and Zn. The monitoring station upstream the Ornain River (Ornain 1) was found to be more enriched in Cu and Zn than the other monitoring stations, probably due to metallic roof leaching from houses and farms at Gondrecourt village located just upstream the station (Fig. 3).
The upstream station on the Saulx River (Saulx 1) was characterized by higher concentrations in Ni and Mn. This monitoring station, although located upstream  the Saulx River, was singularly of a poor water quality because of agricultural activities upstream as well as the presence of ponds in potential eutrophication state. A significant (Kruskal and Wallis test, p < 0.05) spatial variability was observed for five PAHs: fluorene, phenanthrene, fluoranthene, pyrene and chrysene (Fig. 3).
In general, the monitoring station in the middle section of the Saulx River (Saulx 2) had the lowest concentration in PAHs (in particular in phenanthrene, pyrene and chrysene). This station was characterized by a higher water level than the other two stations: spring water inputs are diluting the pollution from upstream (monitoring station Saulx 1). The downstream monitoring station (Saulx 3) was found to be enriched in fluorene, phenanthrene and pyrene. This may be due to a local industry (Carbo France, see Fig. 1) located just a few kilometres upstream, and producing charcoal for industries, restaurants and barbecues.
Looking at the variations in water level over the 3 years and a half deployment periods, important inter-annual variations could be observed (Fig. 4): low water level conditions occurred generally during summer for 4 to 6 months, starting as early as April in 2014 and lasting until as late as November 2015, except in 2013 when low water level conditions was only observed during 2 months (in July and August). Moreover, exceptionally high water level conditions were measured in September and October 2013, whereas they generally occurred from November to March.
Several factors may affect the accumulation rates of metals and PAHs in passive samplers, such as flow rate, temperature, turbidity (due to their affinity with suspended particles) and pH. As shown in Fig. 4, a linear relationship between water level and turbidity (the highest turbidity values were obtained at high water levels) and a quadratic relationship with temperature (high water levels occurred mainly during cooler months),were observed. Moreover, pH variations were very small over the 3 years (less than 0.4 pH) as expected for rivers flowing on carbonaceous rocks. Consequently, TWA concentrations measured on a monthly basis for 3 years and a half were grouped according to water level conditions (low, medium and high water level) to assess the impact of hydrological and climatic cycles on metals and PAHs concentrations in the OPE river waters.
For five metals out of eight, a significant difference was identified between the median of TWA concentrations at low, medium and high water level conditions (Kruskal and Wallis test, p < 0.05): Al, Fe, Mn, Ni and Zn (Fig. 5). Moreover, the highest concentrations were observed in medium andhigh water level conditions and the lowest concentrations in low water level conditions. This could be explained by the hydrological pattern of the OPE catchment: in winter inputs to the rivers are dominated by runoff and soil weathering, thus potentially enriched in suspended matter and some time in organic matter, whereas in summer inputs come from groundwater and spring water mainly.
Only three PAHs out of eight showed a significant water level relationship (Kruskal and Wallis test, p < 0.05): fluoranthene, pyrene and chrysene. As for metals, the highest concentrations were found to be in medium and high water level conditions. Even though the population density is rather low (approx. 10-15 inhabitants/km 2 ), this could be attributed to atmospheric depositions from domestic heating (principally wood burning) during winter.
Indeed, higher concentrations in PAHs were observed in winter 2010 and 2011 probably due to domestic heating and lower concentration in summer 2010 and 2011 within the Danube River Basin between the Cities of Vienna and Bratislava [15]. However, at one monitoring station (Cunovo, Slovakia) high concentrations in pyrene and chrysene were also observed in summer.
In the same way, a seasonal variation of concentration of PAHs in SPMD was observed in the Seine River Basin [17], with concentrations higher in winter due to domestic heating using fossil fuels or wood. The highest mean concentrations in winter in raw water at a drinking water plant located in a large urban area in Poland that can be influenced by combustion products in the atmosphere due to the heating season [41]. Inter-annual variations of TWA concentration of the sum of PAHs (FLU, PHE, ANT, FLR, PYR, BaA, CHY, BbF) showed that the level of contamination in PAHs is not easily predictable from 1 year to another (Fig. 6). In particular at downstream monitoring stations (Saulx 3 and Ornain 2), variations during the winter months (January and February) characterised by high water levels and low temperature, the TWA concentration in PAHs are highly fluctuating from 2013 to 2016 with concentrations as low as 10 ng/L to nearly 70 ng/L.
When looking at both spatial and temporal variability in the sum of metals TWA (for metals other than Al, Fe, Mn) during 2015, no evident trend could be drawn (Fig. 7). During summer months, the TWA concentration in metals was rather low, whereas in winter (November, December, January and February) TWA concentrations are higher together

Conclusion
This study allowed us to validate the benefit of using passive samplers to monitor metals and PAHs in the surface waters at an upstream and rural catchment. Using passive samplers improved greatly the number of quantified data (by a factor 2 to 8, depending on the targeted compound) as the quantification limits were much lowered due to the accumulation of metals and 3 to 4 rings PAHs during 4 weeks. Moreover, spot sampling collected every 2 months gave data often below the quantification limits and thus were not found suitable to assess correctly temporal variability and trends in the long run for metals and PAHs. As a consequence of the improved limit of quantification, it was possible to assess the background pollution in metals and PAHs more accurately at an upstream and rural catchment. Indeed, the low level of contamination measured on the Saulx River and Ornain River when compared with downstream monitoring stations on the Marne and Seine River Basins strongly support the use of passive samplers.  It was then possible to get insight in the spatial and temporal variability of metals and PAHs in the Saulx River and Ornain River over 3 years and a half. Moreover, the results obtained on the QCs were sound enough to ensure sufficient data quality (as detailed in Supporting Information). However, the cost associated with QCs was found rather high, with the number of QC higher than 20-30% of the total number of passive samplers deployed. Therefore, optimising the number of QCs (blanks) will be kept in mind to reduce their ratio to 10%, for the next monitoring period.
Passive samplers, together with continuous monitoring, are necessary complementary tools to be implemented at upstream, rural and karstic catchment in order to better understand the occurrence and variability of pollutants such as metals and PAHs, but also to get a better information on the water quality integrating its fluctuation with time. Currently, studies are undergoing to monitor other contaminants, such as pesticides and pharmaceuticals products, using passive samplers at the study sites.
Author contributions All authors contributed to the study conception and design. Data collection and analysis as well as validation and environmental interpretation were performed by SL-F, AV and BL. PM and SW were involved in conducting experimentations (deployment and analysis of passive samplers). The first draft of the manuscript was written by NG and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Funding The research was co-financially supported by ANDRA and LNE. The authors declare that no other funds, grants and financial support were received during the preparation of this manuscript.

Data and material availability
The datasets generated during the current study on passive samplers are not publicly available due to the need to create new entry fields to implement them in ANDRA database (issue still ongoing in 2022) to make them compatible. Spot sampling data are however available from ANDRA on reasonable request.

Conflict of interest
The authors have no relevant financial or nonfinancial interests to disclose. The authors have no competing interests to declare that are relevant to the content of this article. No potential conflicts of interest were identified.

Ethical approval
The research involved in this study did not deal with either human participants and/or animals. Indeed, the analytical methods used are based on physical or chemical measuring principles and the targeted compounds were physical parameters (e.g. temperature, turbidity) or chemical pollutants (metals and PAHs).
Informed consent Consequently, no informed consent was necessary.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.