The results of the present study represent the first field measurements of metallothionein-like proteins (MT) and cholinesterase activity (ChE) in oysters from the Colombian Caribbean coast as biomarkers of environmental health. Furthermore, the simultaneous measurement of 22 environmental co-variables (4 physical-chemical water parameters and concentrations of 9 metals in tissue and sediments), at 5 stations along a gradient of approximately 250 km of coastline, during three time points, permitted the estimation of the relationship between biomarkers (as criterion variables) and environmental co-variables, to identify relevant “predictor” variables and to test the hypothesis that MT concentrations and ChE activity in oyster tissues are sensitive and informative indicators of contaminant exposure and/or effect at the different sampling locations.
This study initially set out to compare environmental contamination patterns and biomarker responses in shellfish along a presumed two-dimensional contamination gradient, defined by organophosphates (OP), on the one hand, and metals on the other hand. However, whereas we expected to encounter high pesticide concentrations in Barú (Barbacoas Bay), due to presumed agricultural inputs via the Dique Channel, the sediment and tissue analyses, surprisingly, returned concentrations below detection limits for virtually all organochlorine (OCP = DDTs and HCHs) and OP compounds, for all five sampling stations (Aguirre-Rubi et al., 2017). Although these pesticide species used to be common contaminants in environmental matrices in the past, their non-detection in the present study is in line with recent national trends indicating a marked recent reduction of pesticide presence in coastal waters of Colombia, with current detection rates of 16% (Vivas-Aguas et al., 2012; 2015; Garcés-Ordoñez et al., 2016; INVEMAR, 2017). Nevertheless, low tissue concentrations of OPs, whose use is still permitted, might not as much indicate diminished or discontinued use of these pesticides, but rather, low accumulation in tissues and sediments, due to their inherently short environmental half-lives. Lastly, it cannot be ruled out that OCPs and OPs are still present in traces in the coastal environment, but below current limits of detection (which ranged between 2–50 ng/kg for individual compounds).
The results of the metal analyses in oyster tissues (previously reported by Aguirre-Rubi et al., 2017) and in sediments (current study) indicate ubiquity of metal contamination above background at every one of the five sampling stations and underline the disconcerting conclusion that there are no “pristine” (i.e., low metal-concentration) regions in the Colombian Caribbean. The combinations of metals detected at different stations and for sampling dates suggest diverse and variable natural and anthropogenic sources, including industry, domestic discharges, and agriculture (Alonso et al., 2000; Morillo et al., 2004; Fernandez et al., 2007; UNEP; 2008; Olivero-Verbel et al. 2008). The measured sediment concentrations of As, Cd, Cr, Cu, Hg, Ni, Pb, Sn and Zn are consistent with other studies from Colombia and the greater Caribbean that document locally-variable, elevated concentrations of these metals in fluvial, estuarine and coastal sediments, sufficiently high to potentially cause toxic effects in macroinvertebrates and fishes in some cases (Alonso et al., 2000; Vallejo Toro et al., 2016; Barros- Barrios et al., 2016; Tejeda- Benitez et al., 2016; Moncaleano et al., 2017 and 2018, Fernandez-Maestre et al. 2018; Carranza-Lopez et al. 2020). Metals in the present study that had sediment concentrations above the TEL, as defined by Buchman (2008), were As, Cd, Cr, Cu, Hg and Ni.
In contrast to sediments, tissue metal concentrations were typically in the “low” range reported for the sentinel oyster Crassostrea virginica of the US NOAA Mussel Watch Program (MWP, Kimbrough et al., 2008). This was the case for Cu, Pb, Zn (all stations), Sn (exception: Santa Marta Marina Dec 2012 and Mar 2013, and Ctg-2 Oct 2012), As (exception: Taganga Oct 2012), Ni (exception: Taganga-Oct 2012) and Hg (exception: Taganga Oct 2012 and Santa Marta Marina Oct 2012), with noted exceptions having tissue concentrations in the “medium” range. The notable exception was tissue Cd, which exceeded the MWP “high” concentration threshold (15 µg/g) on two occasions (Barú-Oct 2013, Ctg-1-March 2013) and the “medium” threshold on two other occasions (Ctg-1-Dec 2012 and Ctg-2-Oct 2012). Cd tissue concentrations also exceeded the Colombian guideline for human consumption (1 µg/g) of National Resolution 122 of January 26, 2012.
The pattern of apparent metal accumulation in tissues with respect to surrounding sediments is consistent with the biota-sediment accumulation factor (BSAF) profile described for C. virginica by Thomann et al. (1995), which tends to decrease in the order: Zn > Cd > Cu > Hg > As > Sn > Pb > Cr > Ni, with highest BSAFs (10–100) for Zn and Cd, intermediate values (BSAF 1–10) for Cu, As and Hg, and lowest BSAF (typically < 1) for Sn, Pb, Cr and Ni. This suggests that the tropical cup oyster species monitored in the present study (Crassostrea rhizophorae and Saccostrea sp.) accumulate metals in much of the same manner as C. virginica, indicative of similar physiological mechanisms at play. Nevertheless, in the present study, despite having low apparent BSAF, tissue concentrations of Sn, Pb, Cr and Ni were found to correlate significantly with sediment metal concentrations, suggesting uptake via sediment-contact, whereas Hg, As and Cd, despite having much higher apparent BSAF values, had rather weak correlations between tissue and sediment concentrations, suggesting uptake via other routes (e.g., food or water). The high Cd tissue concentrations in the present study, compared to reference values, are consistent with laboratory observations, such as those of Géret et al. (2002), that found significant Cd accumulation in Crassostrea gigas after 21 days of exposure in digestive gland (78 µg/g) and gills (38 µg/g), and Moncaleano et al. (2017) who showed rapid and nearly 130-fold Cd accumulation in Saccostrea sp. over a 96 h exposure period, reaching final tissue concentrations as high as 297 µg/g, as well as Gueguen et al. (2017) for Pinctada margaritifera, and Benali et al. (2017) for Mytilus galloprovincialis.
Biomarker and metal concentrations in tissues and sediments showed considerable variation among stations and sampling occasions, precluding the definition of simple and station-specific or date-specific patterns. Moreover, biomarker measurements for a given sampling location were often characterized by high internal variance, evidenced by large standard errors of the mean, resulting in often non-significant differences in ANOVA, despite considerable procedural efforts to reduce measurement variability (e.g., pooling 5 individuals per sample and analyzing triplicate samples). Two-way ANOVA and post hoc analysis (Tukey HSD, p < 0.05) confirmed significant spatial variance in biomarker responses among the five sampling stations (Santa Marta Marina, Taganga, Ctg-1, Ctg-2 and Barú) for each of the sampling dates (October 2012 and March and October 2013), confirming the absence of a station-specific and temporally- invariant biomarker “profile”.
The overall 5-6-fold range of variation of MT concentrations observed in the present study in oyster digestive gland (3.36–15.73 µg/mg protein) and gills (1.65–9.30 µg/mg protein) is in line with previously reported MT concentrations for other ostreid species. Moncaleano et al. (2017) reported mean MT concentrations of 8.6 ± 2.5 µg/mg protein in digestive gland and 9.3 ± 1.6 µg/mg protein in gills of field-collected Saccostrea sp. exposed to seawater in the laboratory for 96 h (= controls). Likewise, Gold-Bouchot et al. (2007) reported MT concentrations ranging between 4–32 µg/mg in digestive gland and gills of C. virginica collected in the Laguna de Terminos, Mexico; and Bernal- Hernandez et al. 2010 reported MT concentrations in gills between 2.1–21.3 µg/mg in Crassostrea corteziensis from the Boca de Camichín estuary in Mexico. The higher MT concentrations in digestive gland compared to gills are consistent with previous studies (Amiard et al., 2006; Jenny et al., 2004; Moncaleano et al., 2017).
The enzyme activities of the three different cholinesterase fractions analyzed, T-ChE (digestive gland: 15.1–49.1; gills: 14.5–69.1 nmol/min/mg protein), Es-ChE (digestive gland 6.2–58.6; gills 4.3–40 nmol/min/mg protein) and Er-ChE (digestive gland 3.6–19.4; gills 6.6–46.8 nmol/min/mg protein), were very variable and somewhat higher when compared to the range reported by Moncaleano et al. (2018) for field-collected Saccostrea sp., exposed to seawater (= controls) in the laboratory for 96 h (T-ChE in digestive gland: 5.2–28.1 and gills: 9.3–16.4; Es-ChE in digestive gland: 3.5–16.9 and gills: 2.3–11.5; Er-ChE in digestive gland: 5.5–19.5 and gills: 5.4–12; activity expressed in nmol/min/mg protein). Similar ChE activities have also been observed for several other bivalve species, with Bocquene et al. (1997) reporting ChE activity of 28.1 ± 2.8 nmol/min/mg protein in gills of C. gigas; Mora et al. (1999) reporting 25.3 ± 7.7 nmol/min/mg protein in gills of M. galloprovincialis activities; Valbonesi et al. (2013) reporting 5.8 ± 1.9 nmol/min/mg protein ChE in gills of Ostrea edulis and 16.7 ± 4.8 nmol/min/mg protein in gills of M. galloprovincialis; Peric et al. (2013) reporting activities less than 10 nmol/min/mg protein in gills of Arca noae; and Bautista- Covarrubias et al. (2017) reporting 0.4–2.5 nmol/min/mg protein in gills of Crassostrea sp. during dry season and 8.3–34.4 nmol/min/mg protein during rainy season.
None of the four biomarkers measured displayed evidence of marked induction (in the case of MT) or marked inhibition (in the case of ChE activity) at any station, with values generally varying by less than a factor of 5 among stations or sampling occasions. This might be the consequence of an inherently low dynamic range of these biomarkers, reflecting tight metabolic control by oysters, or might be due to the fact that contaminant concentrations in sediments and tissues were generally non-detectable (for OPs) or below toxicological thresholds (for the majority of metals) likely to cause significant up- or down-regulation of these biomarkers, which often tend to be 10 times higher in laboratory exposures (Mourgaud et al. 2002; Moncaleano et al., 2017 and 2018).
Water parameters (temperature, salinity, dissolved oxygen, and pH) and the 9 metal species, analyzed in tissues and sediments, showed a complex, multi-dimensional concentration pattern among the five sampling stations. To reduce the number of environmental variables and to test the hypothesis that tissue biomarker correlated with broader contamination gradients, two separate PCAs were applied to distinguish the variance contribution of contaminant co-variables from physical-chemical co-variables, resulting in a total of six principal environmental components. Two distinct physical-chemical gradients were identified, represented by two principal component axes, named FQ1 (correlating with salinity and dissolved oxygen), and FQ2 (correlating with pH and temperature). These two main components were found to summarize in a more holistic way the hydrographic environmental gradients experienced by oysters at the sampling stations (rather than the four individual variables temperature, salinity, dissolved oxygen, and pH by themselves), bearing in mind that oysters are exposed to more than one environmental variable at the same time.
In the case of the metals, four compositionally distinct gradients were identified by PCA, one (M1) whose composition resembled an “urban/industrial”-type contamination profile (defined by tissue Hg, Ni, Pb and Zn and sediment Cd, Cr and Ni), typical of Taganga Bay on one extreme, which receives input from an underwater sewer outfall from Santa Marta, a second one (M2) suggestive of maritime antifouling paint inputs (high tissue Cu, Sn), as exemplified by Santa Marta Marina), a third one (M3) characterized by likely riverine and agricultural inputs (high tissue Cd and sediment As), characteristic of the three Cartagena/ Barbacoas Bay stations (Barú, Ctg-1 and Ctg-2), all in the proximity of the Dique Channel, and a fourth (M4), represented by elevated sediment Sn concentrations, for which Barú, Taganga and Santa Marta Marina (October 2012) had substantial associations.
Stepwise multiple regression between the biomarkers and the 6 PCA components confirmed significant correlations, between some of the biomarkers and the two physical-chemical parameter gradients: For example, MT in digestive gland and T-ChE activity in gills correlated significantly with FQ2 (and individually with pH and temperature), whereas Es-ChE activity in gills correlated significantly with FQ1 (and individually with salinity and DO). On the other hand, MT in gills and T-ChE and Es-Che activity in digestive gland did not correlate significantly with any of the two principal FQs, nor did Es-ChE activity in either of the tissues, even though they showed weak correlations with individual water parameters (e.g., MT in gills with temperature, pH and DO). The observed correlations between MT and ChEs with physical-chemical variables (or the consolidated gradient axes) are plausible and consistent with other field studies that have shown these two biomarkers to be modulated significantly due to seasonal, oceanographic (i.e. temperature and salinity), nutritional and reproductive variation (Escartin and Porte, 1997; Doran et al., 2001; Choi et al., 2011, Leiniö and Lehtonen, 2005), explaining why the amplitude of biomarker responses in field studies generally tends to be greater than in controlled laboratory studies.
An important finding of this study was that next to the physical-chemical parameters, tissue biomarkers also correlated significantly with three of the four metal component axes, with best predictor variables varying depending on biomarker: In the case of MT in digestive gland, the best contaminant predictor variables were principal components M1 and M2, whereas for MT in gills, they were M3 and M1. For the three ChE enzyme fractions, the most frequent predictor variable was M1 (for T-ChE gills, and Es-ChE in digestive gland and gills), followed by M3 (for T-ChE digestive gland, Es-ChE gills), with only one significant correlation found with component M2 (T-ChE in gills). No significant environmental predictor variables were found for Er-ChE activity in gills. Conversely, principal component M4 had no significant correlation with any of the biomarkers studied. Recalling that principal component M1 had strong loadings from sediment Cd, Cr and Ni, as well as from tissue Hg, Ni, Pb and Zn, it appears that these proxies of urban-contamination are responsible for the modulation of five of the eight biomarkers (MT-dg, MT-gi, T-ChE-gi, Es-Che-dg and Es-ChE-dg, with dg denoting “digestive gland” and gi “gill”, respectively). Environmental component M2, represented by a gradient of sediment Cu, Hg, Pb and Zn and elevated tissue Cu and Sn concentration, best exemplified by Santa Marta Marina (and inversely, by Taganga station), was a significant predictor for only two biomarkers (MT-dg and T-ChE-gi). Lastly, component M3, indicative of elevated sediment As and tissue Cd, was a significant predictor of three biomarkers (MT-gi, T-ChE-dg and Es-ChE-gi), which were elevated in Barú and Ctg-2. The results of the SWMR using the consolidated environmental gradients (summarized in Table 12), thus, clearly show that MT and ChE biomarkers are modulated by a combination of physical-chemical and metal variables, of which the metal gradients more frequently were selected as significant predictor variables in SWMR (absolute frequency of inclusion in best model: 10) compared to the physical-chemical gradients (absolute frequency of inclusion in best model: 3).
More specific analyses of the relationships between biomarkers and individual environmental variables using SWMR and Spearman correlation helped to identify significant individual environmental variables that might be directly (i.e., mechanistically) or indirectly (i.e., as proxies of direct causes) involved in modulating the different biomarker responses. In the majority of SWMR models, positive correlations were observed between biomarkers and metal variables, with tissue metals more frequently displaying positive partial regression coefficients (12 positive vs. 5 negative significant standard partial regression coefficients, Table 11) than sediment metals (7 positive vs. 5 negative regression coefficients). Combining the results of SWMR-A and B, the sum of positive significant partial regression coefficients was: 19 positives vs. 10 negatives, indicating a marked prevalence of positive correlations between biomarkers and metal concentrations. In contrast, physical-chemical co-variables showed no preferred correlation pattern with biomarkers, with 7 positive and 7 negative regression coefficients for the totality of SWMR models examined.
The current finding of apparent induction of MT in gills by Cd is consistent with results from controlled laboratory studies, such as Viarengo et al. (1989) in M. galloprovincialis, who showed upregulation of MT in gills after 2 days of exposure to Cd. Similarly, Roesijadi et al. (1997) demonstrated significantly increased expression of MT in C. virginica after 2 days of Cd exposure (at 73-fold ambient concentrations), as did Géret et al. (2002) in C. gigas after Cd exposure (200 µg/L) over 21 days. Likewise, Jenny et al. (2004) observed significant upregulation of MT gene transcription in gills and digestive glands of C. virginica exposed for 96 h to Cd (0.25–0.44 M, approx. 28–50 g/L), similar to Aceto et al. (2011), who reported increased gene expression (MT-10, MT-10 Intronless, MT-20) in field-exposed M. galloprovincialis. Finally, Yingprasertchai et al (2019) also reported an increased expression of sgMT due to Cd stress at 200 µg/L in the Sydney rock oyster Saccostrea glomerata. Moncaleano et al. (2017) also observed up to two-fold increase of the MT concentrations in digestive gland and gills of Saccostrea sp. exposed to Cd (100 µg/L, 96 h); Gueguen et al. (2017) reported significant up-regulation of MT by Cd in Pinctada margaritifera, noting furthermore that Cd stimulated significant changes in the transcription of individual genes and Chan & Wang (2019) reported elevated MT concentrations (20–70 µg/g wet wt.) in Crassostrea hongkongensis, related with a strongly detoxification response for Cd in whole oyster tissues. Nevertheless, the Cd concentrations measured in whole tissues in the present study (0.73–22.86 µg/g) were considerably below those reported by previous authors to acutely induce MTs (e.g. Moncaleano et al. 2017 found that MT concentrations in digestive gland and gills of Saccostrea sp. were invariant to acute Cd exposure < 100 µg/L over 96 h, corresponding to Cd whole-tissue concentrations of 30 µg/g), suggesting that Cd exposure in the field was not quite high enough to produce marked site-specific differences in MT levels.
The apparent induction of MT in digestive gland by Ni is supported by observations by Mourgaud et al. (2002), who demonstrated that Ni induces MT in M. galloprovincialis (as well as Cd, Cu, and Zn) and is also able to bind to MT proteins. Similarly, Attig et al. (2010) reported a significant increase of MT in digestive gland of M. galloprovincialis exposed to Ni at 2.5 µM and 13 µM, consistent with Costa et al. (1994) who noted high affinity of Ni for sulfhydryl groups of proteins, including MTs. Peric et al. (2012) also reported a strongly positive, albeit statistically non-significant correlation, between MT content and tissue Ni concentrations in M. galloprovincialis (Ni concentrations between 0.47 to 2.78 mg/kg). MT induction by Ni has been reported not only in bivalves but also in the cod Eleginus nawaga (Eriksen et al., 1990) and in the copepod Tigriopus brevicornis (Barka et al. 2001). On the other hand, Amiard et al. (2008) did not find correlation between MT and Ni concentrations in mussels exposed to Ni in field and laboratory conditions. For Zn, Peric et al. (2012) reported a positive correlation between MT content and tissue Zn in M. galloprovincialis (Zn concentration 117–271 mg/kg). Nevertheless, other authors failed to observe increases of tissue MT concentrations by Zn and Cu exposure (Géret et al., 2002; Jenny et al., 2004; Amiard et al.2008; Moncaleano et al, 2017). Interestingly, Liu and Wang (2016) reported a significant negative relationship between Cu, Zn, Cr and Ni with MT concentration for C. hongkongensis and Crassostrea angulata, a finding also reported for Crassostrea sikamea by Weng and Wang (2014) after Cu and Zn exposure. This situation shows that there is still a lack of consensus regarding the effects of metals on metal regulation in bivalve species. Undoubtedly, protein regulation responses vary among species, due to differences in metal sequestration mechanisms, such as scavenging by metallothionein and other metal complexes or differences in metallothionein turnover rates that are still unknown and metals (Wang and Rainbow, 2010).
MT induction by As has been observed in Corbicula fluminea for aqueous As concentrations as low as 100 µg/L (Santos et al. 2007), corroborating the strong correlations observed between As and MT in gills in the present study. For Hg, Géret et al. (2002) reported induction of MT in gills in C. gigas, corresponding to Hg tissue concentrations of 22.6 ±3.9 µg/g in digestive gland and 74.8 ± 14 µg/g in gills, which is very high compared with the tissue concentration of Hg in the present study (< 0.18 µg/g). Nevertheless, Mourgaud et al. (2002) reported that Hg was not a significant inducer of MT in M. galloprovincialis and neither was As, Cr and Pb. On the other hand, Gueguen et al. (2017) reported significantly up-regulated expression of MT in in P. margaritifera upon Cr exposure concentrations between 1–10 µg/L, even though Cr accumulation was low, with Cr tissue concentrations < 10 µg/g dry weight, which is comparable to the Cr concentrations in the present study (0.23–9.14 µg/g). Despite Cr not being selected in the SWMR models as a significant predictor variable of MT, single Pearson correlations between MT and Cr did show significant correlations. However, due to the strong covariance between Cr and Ni, SWMR models favored the inclusion of Ni over Cr, due to a higher partial regression coefficient.
Finally, some authors have shown that variation in MT content is not due exclusively to metal concentrations but related to other environmental and biological factors as well. For example, Geffard et al. (2001) reported that MT levels in digestive glands of C. virginica in the Gironde estuary (France) were strongly affected by seasonal variations and body size, while Raspor et al. (2004) observed that variations in MT content of M. galloprovincialis (digestive gland) were partially explained by inherent biological co-variables, such as sexual maturation, gland size and food availability. Similarly, Leiniö and Lehtonen (2005) observed significant seasonal variations in MT content in Mytilus edulis and Macoma balthica, whereas Amiard et al. (2006) noted a low correlation between MT concentrations and environmental metal concentrations in aquatic invertebrates, which they attributed to differential responses of different isoforms of MT.
Ivanina et al. (2009) reported a significant temperature effect on MT expression in C. virginica, whereas Riba et al. (2003) observed an inverse effect of pH and salinity on the uptake of Zn and on MT concentrations in Ruditapes philippinarum. Other studies, likewise, have reported considerable negative correlation between salinity and MT levels in the digestive gland of mussels M. galloprovincialis (Ivanković et al 2005; Hamer et al. 2008, Sun et al. 2018). Considering these results, it, therefore, comes as no surprise that temporal and spatial differences in MT concentrations in the present study, as confirmed by ANOVA, were partially explained by environmental co-variables, such as pH (MT-dg and MT-gi) and temperature (MT-gills) in the SWMR, consistent with high regression coefficients of the consolidated principal component FQ2 (carrying strong factor loadings from pH and temperature).
The failure to observe pronounced inhibition of ChE activity at any of the locations in the present study is consistent with the overall finding of non-detectable OP concentrations in sediments and oyster tissues. Whereas ChE activity can be modulated by subtle differences OP and carbamate exposure concentrations (Montserrat et al., 2002, 2007, Bernal- Hernandez et al., 2010), the consistent failure off detecting OPs in sediments and tissue argues against using variations in OP concentrations to explain the spatial and temporal variations in the activity of the three cholinesterases observed, and, instead, suggests influences of other environmental and/or biological variables. Next to OPs and carbamates, ChE activity is known to be affected by a variety of metals, as well as PAHs, hydrocarbons, detergents, phytotoxins and other classes of pesticides such as neonicotinoids (Magni et al. 2006; Senger et al. 2006; Choi et al. 2011; Moncaleano et al., 2018), only a fraction of which were measured in in the present study. In this respect, it is peculiar that enzyme activity of the three kinds of ChEs tended to correlate positively with many the analyzed metal species in sediments (Cr, Cd, Cu, Ni and Sn) and tissues (As, Cd, Cr, Hg, Pb, Sn, and Zn). We conjecture that the generally positive correlation between ChE activity and metal concentrations might be due to a stimulatory effect of metals on ChE expression of oysters, perhaps by a global up-regulation of stress responses, resulting in increased production of ChE protein (among others), even though each enzyme molecule by itself might be susceptible to kinetic inhibition by metals. Since sediment metal concentrations (and therefore water concentrations) were generally low (i.e., below TEL for most metals), exposure concentrations might simply not have been high enough to elicit kinetic inhibition of ChEs. Therefore, it is quite possible that oysters with elevated ChE activities were "up-regulating" ChE protein concentrations, as part of a non-specific stress response to metals and other, unmeasured, environmental contaminants. Interestingly, the few metal species whose tissue concentrations did correlate negatively with ChE activity, namely As (standard partial regression coefficient rs = -0.69 and − 0.83, for Er-ChE-dg and Er-ChE-gi), Cd (rs = -0.23, for Es-ChE-dg) and Sn (rs = -0.11, for Es-ChE-dg) were those for which tissue concentrations exceeded the “medium” or “high” thresholds relative to MWP reference values, as was characteristic of oysters from Barú and Ctg-2, where tissue Cd and tissue As were high but ChE activity low. An inhibitory effect of Cd on cholinesterases has been confirmed for various bivalve species, including M. galloprovincialis (Magni et al. 2006, Tsangaris et al., 2010), Adamussium colbecki (Bonacci et al., 2006, Cravo et al., 2012), Ruditapes decussatus (Cravo et al. 2012), Meretrix casta (D’Costa et al. 2017) and Saccostrea sp. (Moncaleano et al., 2018), supporting the possibility that oysters with elevated tissue Cd and relatively low ChE activities, as was the case for oysters from Barú and Ctg-2, might have been inhibited by Cd. Regarding As, laboratory studies have shown both significant inhibition of ChE activity by As in the freshwater clam Lamellidens marginalis (Chakraborty et al., 2012), as well as induction of ChE expression by As in Perna indica (Rajkumar et al. 2013). Negative correlations between ChE activity and Cu and Cr (as observed in C. virginica by Gold Bouchot et al., 2007) and Ni in M. galloprovincialis (Attig et al., 2010) are not corroborated by our field results.
Next to metals, variations in ChE activity in gills were also partially explained by physical-chemical variables (salinity, temperature, and pH) with negative correlations prevailing over positive regressions in SWMR models (6 negative regression coefficients vs. 2 positive regression coefficients). Among the four physical-chemical variables, salinity returned strongly negative partial regression coefficients (rs < -0.5, with T-ChE-dg, Es-ChE-dg and Es-ChE-gi) whereas temperature showed predominantly positive regression coefficients (rs > 0.5, with T-ChE-gi and Es-ChE-gi), whereas regression coefficients for pH as a predictor of ChE activity where negligible (rs = -0.008, with Es-ChE-dg).
The three-way ANOVA showed that next to spatial and temporal effects (due to spatial- temporal variations in environmental predictor variables), the concentration of MT and the activity of the three different types of cholinesterases (T-ChE, Es-ChE and Er-ChE) were also influenced by the type of tissue (digestive gland or gills). This biological source of variability is well-known, and Lowe & Day (2002), Gold- Bouchot et al. (2007), Bernal- Hernandez et al. (2010) and Andrade-Brito (2012) have all recommended gills as the most suitable tissue for the measurement of biomarkers, first because gills represent the largest surface area in contact between the individual and the environmental, and second, because gill tissues are highly permeable, which favors rapid uptake of waterborne contaminants, as the filter-feeding process not only retains food, but also a wide variety of other particulate and dissolved substances, including metals and pesticides. For MT, the use of the gill as well as digestive gland is common, since next to being sensitive to MT exposure they also tend to display highest MT concentrations (Bebianno et al., 1993; Mackay et al., 1993; Baudrimont et al., 1997; Mouneyrac et al., 1998; Geffard et al., 2001; Ceratto et al., 2002), compared to other tissues, such as muscle. For cholinesterases, previous studies have reported high activity and sensitivity in the gill, mantle, and adductor muscle (Monserrat et al., 2002; Damiens et al., 2004; Bernal-Hernández et al., 2010), yet also in digestive gland (Bocchetti et al., 2008; Moncaleano et al. 2018). The sensitivity of MT and ChE activity in both gill and digestive gland tissue, as observed in the present study, shows that both tissues are suitable for the measurement of these biomarkers in future monitoring programs in Colombia, even though each tissue responds in a different manner to environmental variables.
This study demonstrated that MT and ChE activity in cup oysters varied significantly between stations and sampling dates, which was explained to a large extent by variations in environmental parameters, including physical-chemical water variables (T, S, pH and DO) and sediment and tissue metal concentrations, which differed among stations and sampling dates. The consolidation and re-partitioning of environmental co-variables, from 22 dimensions (i.e., 4 physical-chemical water parameters, 9 tissue metals and 9 sediment metals) to only 6 main components greatly helped in data reduction and for defining major orthogonal environmental gradients and for estimating relative weight of each. The integrated analysis (excluding Ctg-1 station, due to data deficiency and untypically high ChE activities), showed that MT and cholinesterase biomarkers responded differentially to the principal environmental gradients, indicating non-redundancy of the two biomarkers. The inclusion of metal gradients as significant explanatory variables in stepwise multiple regression models for all but one of the 7 biomarkers variables, highlights their probable mechanistic involvement in modulating the observed biomarker responses. Nevertheless, bearing in mind that the environmental gradients were derived from two separate PCAs, it is entirely possible that the 2 physical-chemical gradients (FQ1 and FQ2) and the 4 metal gradients (M1, M2, M3, M4) were not strictly orthogonal (i.e., statistically independent), but interrelated, resulting in the favoring of one over the other in the SWMR models, even though both might be part of the same gradient.