Effects of industrial pollution on the reproductive biology of Squalius laietanus (Actinopterygii, Cyprinidae) in a Mediterranean stream (NE Iberian Peninsula)

Mediterranean rivers are severely affected by pollutants from industry, agriculture and urban activities. In this study, we examined how industrial pollutants, many of them known to act as endocrine disruptors (EDCs), could disturb the reproduction of the Catalan chub (Squalius laietanus). The survey was conducted throughout the reproductive period of S. laietanus (from March to July 2014) downstream an industrial WWTP located in the River Ripoll (NE Iberian Peninsula). Eighty fish (28 females and 52 males) were caught by electrofishing upstream and 77 fish (33 females and 44 males) downstream a WWTP. For both sexes, the gonadosomatic index (GSI) and gonadal histology were examined and related to water chemical analysis and fish biomarkers. Female fecundity was assessed using the gravimetric method. Fish from the polluted site showed enhanced biomarker responses involved in detoxification. Also, in the polluted site, lower GSI values were attained in both sexes and females displayed lower numbers of vitellogenic oocytes. Gonadal histology showed that all maturation stages of testicles and ovaries were present at the two study sites but fish males from the polluted site had smaller diameter seminiferous tubules. Water chemical analysis confirmed greater presence of EDCs in the river downstream the industrial WWTP. The chemicals benzotriazole and benzothiazole could be partially responsible for the observed alterations in the reproductive biology of S. laietanus.


Introduction
Conservation of freshwater fish has reached a critical point worldwide.In Europe, about 37% of freshwater fish species are included under some threat category of extinction (Freyhof and Brooks 2011).The situation is even worse in the Iberian Peninsula, where roughly 70% of its fish species are endemic (Doadrio et al. 2011).Mediterranean rivers are considered hotspots in terms of biodiversity despite being among the most endangered ecosystems worldwide (Cuttelod et al. 2008); with a long-history of anthropogenic insults, including pollution, introduction of exotic species and habitat degradation resulting in an alarming decline of fish populations (Clavero et al. 2004).In Catalonia (NE Iberian Peninsula), two out of the 29 species of native fish are regionally extinct and further 22 (75%) are under threat according to the IUCN (2012).Despite the efforts to implementing the European Water Framework Directive (2000/60/EC) (2000) with wastewater treatment practices, the quality of freshwaters remains worrying.Spillages of wastewater that have not been appropriately treated are still occurring and they invariably lead to the deterioration of water quality downstream (Maceda-Veiga et al. 2013).Moreover, as it is the case of streams in other semi-arid regions, Mediterranean-type climate streams are subjected to summer droughts that can worsen the harmful effects of pollution on the aquatic biota (Colin et al. 2016(Colin et al. , 2017)).This situation is likely to worsen due to the higher water demand for human use because of climate change (Mekonnen and Hoekstra 2016).
There is a strong body of evidence that an increasing number of chemicals, frequently found in rivers of the Iberian Peninsula, act as endocrine disruptors (EDCs) (Gorga et al. 2015;Kuster et al. 2008;Matamoros et al. 2010a;Osorio et al. 2016).Overall, among the EDCs' effects on the reproductive biology of fish are the occurrence of intersex condition (female oocytes in male testicular tissue), altered oogenesis in females, decreases in fecundity and population recruitment failure (Jobling et al. 1998;Kidd et al. 2014).Some EDCs can interact, even at very low environmental doses (ng/L), with the genesis of fish steroid and the processes of sexual maturation and differentiation (Vos et al. 2000).Fish reproductive output may be altered by the presence of EDCs in the environment (Nash et al. 2004) and this, in natural environments, may have an strong impact on the ecology and conservation status of native species.
EDCs are widespread in freshwater environments and both laboratory and field based studies have shown reproductive alterations in fish at environmentally relevant exposures.
However, it is unclear how these effects may affect fish populations in the wild (Mintram et al. 2018).The main inconvenient of most laboratory studies is that they do not fully reproduce natural conditions and/or do not consider the potential synergy of mixtures of contaminants.Indeed, mixed chemical exposure outcomes can differ significantly from those for single classes of EDCs (Mintram et al. 2018).Moreover, laboratory experiments may not account adequately for vital ecological processes and environmental variations (Galic et al. 2010).The assessment of reproductive disturbances, including histological aspects, is well documented for chemicals with well-known endocrine disruption abilities but the effects may vary depending on concentration, species and sex (Dietrich and Krieger 2009;Sumpter and Johnson 2005).There is a lack of knowledge on how pollutants discharged into Mediterranean rivers may be capable of modifying the reproductive biology of native fish, where most ecotoxicological studies have focused on exotic fish species such as carp (Fernandes et al. 2002;Solé et al. 2003;Lavado et al. 2004).The main aim of this study was to investigate how the presence of EDCs in the water composition of a Mediterranean-type climate stream could alter the reproductive output in a species of native fish.This knowledge is a necessary tool for their conservation.Current approaches for the environmental risk assessment (ERA) of chemicals, including EDCs, lack certainty for protecting wildlife because of differences in species sensitivity to pollutants (Hamilton et al. 2016).Typically, ERA relies on assessments extrapolated from animal laboratory studies at pollutant concentrations not observed in nature (Mintram et al. 2018).
The species selected was the Catalan chub, Squalius laietanus Doadrio, Kottelat & de Sostoa 2007, a freshwater fish endemic to the NE Spain and SE France.During the 1990s this species suffered an important decline in the rivers of NE Spain due to pollution, habitat alterations and the introduction of exotic species (Doadrio et al. 2007).Both sexes reach the first sexual maturity when they are 2 years old (females at 10 cm and males at 7 cm long) and the breeding period extends from April to July (Casals 2005).Since the S. laietanus population is decreasing, it has been listed as vulnerable in both the IUCN (2012) and the Spanish catalogues of threatened species (Royal Decree 139/2011 of 4th February).However, this species has currently a relatively wider distribution (Aparicio et al. 2016), so investigating the effects of pollution on its reproduction can help improve its management and conservation in the future.
In this study, the concentration of chemicals potentially acting as EDCs was evaluated through water chemical analyses, and fish stress responses were complemented using biomarkers.The gonadal cycle and the gonadal histology of both sexes, as well as female fecundity, were examined upstream and downstream from the textile dye WWTP, throughout the species breeding period.The results of this study help to clarify the role of pollutants on the freshwater fish decline in semi-arid regions.

Study area
This study was conducted in the mid stretches of the River Ripoll, a 39.5 Km tributary of the River Besòs basin located in the NE of the Iberian Peninsula.It is a river of Mediterranean-type climate with alternating periods of droughts and flooding.The polluted site was 1.6 Km downstream from the discharge of a WWTP from a textile industry near Castellar del Vallès (41°34′17.88″N,2°06′01.40″E).The control site, named Les Arenes (41º38′45.05″N,2°03′24.07″E), was 2.7 Km upstream from the WWTP of the textile dye plant (Fig. 1).In the polluted site (thereafter, downstream WWTP) the river width during the sampling period was 3-5 m and the average water depth was 20-40 cm.In the control site (thereafter, upstream WWTP) the river width was 4-6 m and the average water depth was 20-30 cm.A small dam located between the two sampling sites prevented the passage of fish from one point to another.

Site characterisation by water chemical analysis
Water chemical analysis of this study was carried out following the protocol optimized for freshwater Mediterranean watercourses by Matamoros and Bayona (2006).Water samples were collected at the two sites in March (spring sampling) and July (summer sampling) using precleaned 1 L amber glass bottles.Four water samples were, therefore, collected for chemical analysis.Samples were kept at 4°C, filtered within the next 24 h and processed as previously reported (Matamoros and Bayona 2006).A sample volume of 500 mL was percolated through a previously activated polymeric solid-phase extraction cartridge (200 mg Strata-X cartridge).
Elution was performed with 10 mL of hexane/ethyl acetate (1:1).The eluted extract was evaporated to ca. 100 µL under a gentle nitrogen stream, and 186 ng of triphenylamine was added as an internal standard.Finally, the extract was reconstituted to 300 µL with ethyl acetate.
Methylation of the acidic carboxyl group was performed in a hot gas chromatography (GC) injector (290°C) by adding 10 µL of TMSH solution (0.25 mol/L in methanol) to a 50 µL sample before injection.Derivatized samples were analysed in a Bruker 450-GC gas chromatograph coupled to a Bruker 320-MS triple quadrupole mass spectrometer (Bruker Daltonics Inc., Billerica, MA, USA) in the electron impact mode (70 eV ionization energy) fitted with a chromatographic column TRB5-MS coated with 5% diphenyl 95% dimethylpolysiloxane (20 m x 0.18 mm i.d., 0.18 µm film thickness) from Teknokroma (Sant Cugat del Vallès, Spain).A 5 µL volume of sample was injected in the PTV mode.
Chromatographic conditions, data processing and validation of the methodology have been described elsewhere (Matamoros et al. 2010a;Matamoros and Bayona 2006).

Fish sampling
Fish were sampled by electrofishing using a portable unit which generated up to 200 V and 3 A pulsed D. C throughout the breeding period (March, April, June and July 2014).Samplings took place in the middle of the month.Samplings during April (which took place at the end of the month) were considered representative of the reproductive state of the fish between April and May (Casals 2005).The number of individuals used in each section of this study is specified in Table 1.The fish (except those used in the biomarker analysis) were sacrificed with an overdose of MS-222 and frozen at -20 ° C. All fish were sized (fork length, FL ± 0.1 mm) and weighed (eviscerated mass, W E ± 0.01 g), and gonads (gonad mass, W G ± 0.01 g) and liver (liver mass, W L ± 0.01 g) were removed and weighed.All fishes used in this study were adults.Fish collection was approved by the Regional Government of Catalonia (Ref.AP/007).All applicable international, national, and/or institutional guidelines for the care and use of animals were followed.All procedures were conducted in accordance with the European Directive for animal experimentation (2010/63/EU).One of the co-authors holds a category C FELASA certificate that regulates the use of animals for experimental and other scientific purposes.

Site characterisation by the use of fish biomarkers
A subsample of 36 fish from upstream WWTP and 36 fish from downstream WWTP were transported alive in aerated tanks to the laboratory facilities (Table 1).Fish were sacrificed with an overdose of MS-222.A portion of muscle and the whole liver were frozen in dry ice and liquid nitrogen, respectively.Tissues were stored at -80°C until biochemical determinations.
A portion of muscle (≈ 0.2 g) was homogenised in ice-cold 50 mM, and the whole liver was homogenised in 100 mM buffer phosphate (pH 7.4) containing 150 mM KCl, 1 mM dithiothreitol (DTT), 0.1 mM phenanthroline, 0.1 mg/ml trypsin inhibitor and 1 mM ethylenediaminetetraacetic acid (EDTA) at a 1:4 (w:v) ratio using a polytron® blender.The homogenates were centrifuged at 10,000 g x 30' at 4°C and the supernatants obtained (S10) were used for the enzymatic determinations in muscle and liver.
The total protein contents of the samples for all of the assays were determined following Bradford's method (1976) adapted to microplate, using the Bradford Bio-Rad Protein Assay reagent and bovine serum albumin (BSA) as the standard (0.1-1 mg/mL).
Before completing the processing of the samples, the hepatosomatic index (HSI) was calculated according to the formula: HSI = W L /W E × 100.This index can provide information related to the fitness of fish, but also on its energy reserves, potential diseases, and exposure to exogenous stresses (Schmitt and Dethloff 2000).

Gonadal development and female fecundity
The gonadal cycle was described through the gonadosomatic index (GSI) calculated according to the formula: GSI = W G /W E x 100.A subsample of 36 fish (21 females and 15 males) from upstream WWTP and a subsample of 37 fish from downstream WWTP (22 females and 15 males) were analysed over the breeding period of S. laietanus (Table 1).Means and 95% CI of GSI were calculated for sex and site.
Oocyte development and female fecundity were examined by the gravimetric method (Bagenal 1978) using the same females that were used for the calculation of the GSI.A portion of one ovary from each female was weighted and transferred into Gilson's fluid and, once the gonads had been separated, oocytes were poured into a column of 18 sieves with a range of mesh-sizes (from 0.15 mm to 1.8 mm) and flushed with running water to facilitate their descent.
Based on oocyte diameter frequency distributions, developmental stages corresponding to previtellogenic (immature) and vitellogenic (ripening, or in maturation process, and ripe) oocytes were identified.Fecundity, defined here as the total number of ripening or vitellogenic oocytes in the ovaries prior to spawning (Bagenal 1978), was determined by counting all vitellogenic oocytes in gravid females (i.e. in females with ripe oocytes).Mature females could be recognized by the presentation of amber-coloured oocytes (or yolk eggs) in the ovaries.Fecundity (F) was calculated as: ln F = a + b ln FL (Bagenal 1978).Since females downstream WWTP were smaller in size, only females in the upstream site that were within a similar range of lengths (from 175 to 215 mm of FL) were taken into account to avoid an underestimation of fecundity in the polluted site.Only mature females prior to spawn (i.e. with amber-coloured oocytes in the ovaries) were considered for the calculation of fecundity.For females downstream WWTP, due to a lack of synchronization of the time of egg laying, females throughout the breeding period were considered for the calculation (N=3 in March, N=3 in April, and N=1 in June).Upstream WWTP females prior to spawn were found in June (N=6).

Histological analyses
Gonads of 20 females (five per month over the reproductive period of S. laietanus) and 15 males (five per month except in July, which was not analysed) were used in the histological analyses from upstream WWTP.In the polluted site, gonads of 18 females and 15 males (no males sample in July) were analysed (Table 1).Ovaries and testes were fixed in a 10% Bouin's solution, dehydrated with increasing ethanol, cleared in xylene and finally embedded in paraffin wax (Dietrich and Krieger 2009).Gonad sections of 5 µm were cut on a PFM Rotary 3003 microtome (pfm medical, Germany) and stained with conventional Delafield haematoxylin and eosin.Stained gonad sections were examined in an Olympus CH-2 microscope and photographed with a Nikon DS-Ri1 digital camera.Five oocyte stages were defined according to their maturation status: Stage I, primary oocytes; Stage II, large primary oocytes with Balbiani body and pleiomorphic nucleoli-like bodies bordering the nuclear envelope; Stage III, early vitellogenic stage; Stage IV, advanced vitellogenic stage with many variably sized yolk vesicles; and Stage V, mature oocyte with yolk mass occupying the entire oocyte.An additional, non-defined stage that did not fit the former criteria was scored: atretic oocyte, a set of small cells with strong staining and irregular shapes.
Frequency and size of each stage of development were calculated following Dietrich and Krieger (2009).For males, four stages of maturation were defined: spermatogonia, spermatocytes, spermatids and spermatozoa.Quantitative methods proposed by Smith (1978) and Gimeno et al. (1998) were used to determine the percentage of testicular stages and to measure the diameter of seminiferous tubules.
A maturity index for each fish was calculated according to the standardized criteria for ovaries and testes established in Baumann et al. (2013).This is an enhancement to the OECD Histopathology Guidance Document (Johnson et al. 2010).The maturity index gives a fixed value (ranging from zero to one) for each fish according to its maturity stage and it increases as fish sexual maturity progresses.

Statistical analysis
In order to test whether biomarkers and HSI showed differences between sites, a generalized lineal model (GLM) was performed for each of them.The structure of these models was the same for both: a biomarker as the dependent variable and season (2 levels: Spring and Summer) and site (2 levels: upstream and downstream) as factors together with their interaction.The gamma distribution and the identity link function were assumed for all biomarker GLMs.Due to the multiple testing, resulting p-values where adjusted by the method of Benjamini and Hochberg (1995).GSI differences between sites were tested performing a GLM for each sex.
Both male and female GLM's used GSI as the dependent variable, month (4 levels in female GLM: March, April, June and July; 3 levels in male GLM: March, April and June) and site (2 levels: upstream and downstream) as factors together with their interaction.The gamma distribution and the identity link function were used.
For the analysis of the variability of the number of oocytes between sites, a GLM was performed per development stage (vitellogenic and non vitellogenic oocytes).Each model used the number of vitellogenic and non vitellogenic oocytes as the dependent variable.Month (4 levels: March, April, June and July) and site (2 levels: upstream and downstream) were added as fixed factors together with their interaction.The negative binomial distribution was assumed (Poisson distribution was discarded due to overdispersion).The effect of site on female fecundity was tested by means of a GLM with number of vitellogenic oocytes as the dependent variable.The fork length was added as a covariate, site (2 levels: upstream and downstream) as factor as well as the interaction between them.The negative binomial distribution and the log link function for count data were assumed (Poisson distribution was discarded due to overdispersion).
In order to analyse how the degree of maturity was affected by the WWTP discharges, a generalized lineal mixed model (GLMM) was performed for each sex.Both male and female models had the same structure: maturity index was the dependent variable, month (4 levels in female GLMM: March, April, June and July; 3 levels in male GLMM: March, April and June) and site (2 levels: upstream and downstream) were added as fixed factors together with their interaction.Given that we obtained six maturity index values from each individual (each value from a different gonad section) specimen was added to the model as a random factor.
The effect of each oocyte development stage on oocyte size was tested performing a GLMM.Each model used cellular size of a stage as the dependent variable.Month and site (2 levels: upstream and downstream) were added as fixed factors together with their interaction.Levels of the month factor were not the same for all models given that the development stages were not present during the same months.Specimen was also included as a random factor as we used several measures for each female.The gamma distribution was assumed.In order to test whether the diameter of the seminiferous tubules differed between sites, a GLMM was performed.Tubule diameter was the dependent variable and month (3 levels: March, April and June) and site (2 levels: upstream and downstream) were the fixed factors together with their interaction.Tubule diameter was measured from six different gonad sections for each specimen, thus specimen was added as a random factor.The gamma distribution and the identity link function were assumed.
All analyses were conducted in R 3.4.3(R Core Team 2017).GLMs assuming a gamma distribution were performed with glm() function (package stats: R Core Team) and the negative binomial GLM (female fecundity model) with glm.nb()(packageMASS: Venables and Ripley 2002).GLMM was performed with glmer() (package lme4: Bates et al. 2015).Non-significant interactions were removed from final models.Homogeneity and normality of residuals were visually checked for all models.

Site characterisation by water chemical analysis and the use of fish biomarkers
The physico-chemical water parameters at the two sites at the spring and summer samplings are detailed in Table 2.These water quality indicators suggested poorer water conditions downstream WWTP, especially in summer.The chemical analysis of selected contaminants of emerging concern (CEC), including some EDCs, is presented in Table 3.Among the CEC detected, only benzotriazole, benzothiazole and their metabolites were found at high levels in water, particularly at the polluted site at both sampling times.
Results concerning liver and muscle biomarkers are summarized in Table 4. Results in HSI showed a significant effect of the site factor ( = 25.21,p < 0.001), being the highest values downstream WWTP.For liver biomarkers, significant effects of the season factor for CAT ( = 14.29, p < 0.001), GR ( = 32.71,p < 0.001), EROD ( = 13.70,p < 0.001) and BFCOD ( = 46.75,p < 0.001) were found, which consisted in higher values in July (summer).The site factor showed a significant effect for GR ( = 94.08,p < 0.001), GPX For LDH a significant effect of the season factor was detected ( = 57.11,p < 0.001) which consisted of a higher value in summer.In summary, biomarkers showed a general tendency to show higher values downstream than upstream WWTP on the one hand, and in summer than in spring on the other.

Gonadal development and female fecundity
Female GSI model showed a significant interaction between month and site factors ( = 12.30, p = 0.006 ) and a significant effect of the month factor ( = 24.39,p < 0.001 ).
Females upstream WWTP showed maximum GSI values in April and June while females downstream did not reach any peak (Table 5).In males, the month ( = 8.53, p = 0.014) and the site ( = 12.79, p < 0.001 ) factors were significant showing a maximum in April upstream WWTP (Table 5).In summary, both females and males displayed the highest GSI values upstream WWTP.
Size-frequency distributions of previtellogenic (≤ 0.35 mm) and vitellogenic oocytes in maturation process (> 0.35 mm) during the breeding period of S. laietanus at the two sampling sites are shown in Fig. 2. A different pattern in size-frequency distributions of vitellogenic oocytes was observed.Females upstream WWTP showed a maximum number of vitellogenic oocytes in June with a peak of mature amber-coloured oocytes (≥ 0.80 mm).By contrast, females downstream had lower amounts of vitellogenic oocytes (in maturation process) throughout the entire breeding period with no clear reproductive peak, and a small group of mature oocytes in April.The statistical analysis of vitellogenic oocytes showed a significant interaction between month and site factors ( = 18.487, p < 0.001) as well as a significant effect of the month ( = 44.256,p < 0.001) and site ( = 6.484, p = 0.011)factors.For previtellogenic oocytes a significant interaction between month and site factors was found ( = 12.167, p = 0.01) as was a significant effect of the month factor ( = 35.685,p < 0.001).Females upstream WWTP had more vitellogenic oocytes than females downstream.
As the covariate female length did not affect in the female fecundity model, it was removed from the model.The final model showed a significant effect of the site factor ( = 13.93,p < 0.001): upstream females had about 2.6 times (CI: 1.5 -4.6) more vitellogenic oocytes than downstream females.

Histological analyses
Females and males collected at both sites displayed all the defined stages of maturation.In Fig. 3, details of the different stages are presented.The female maturity index model showed a significant interaction between month and site factors ( = 8.609, p = 0.035), a significant effect of the month factor ( = 33.44,p < 0.001).Both upstream and downstream females showed a quite similar maturation pattern, but upstream females reached a maturation peak in June that was absent for downstream females (Fig. 4).For males, only the month factor was significant in the maturity index model ( = 16.98,p < 0.001).In summary, upstream and downstream males showed an identical maturation pattern which consisted of higher maturity index values in April and June than in March.
Oocyte size expressed as area units (in µm 2 ) could only be accurately measured from the Stage II of maturation onwards (Table 6).The site factor was significant for stages II ( = 9.57, p = 0.002) and IV ( = 3.89, p = 0.05), being cells of both stages larger for females downstream WWTP.Occurrence of atretic did not differ per site or season.In the tubule diameter model, only a significant effect of the site factor (χ = 6.51, p = 0.011) was found, with males upstream WWTP having a higher tubule diameter (mean = 222.99µm; CI = 184.09-261.88)than males downstream WWTP (mean = 164.80 µm;.

Discussion
Former studies carried out in the same area of the Ripoll River, revealed that the site downstream from the textile dye plant, even after wastewater treatment, was still highly polluted (Colin et al. 2016(Colin et al. , 2017;;Maceda-Veiga et al. 2013;Blanco et al. 2019).In this study, several CEC, such as pharmaceuticals, sunscreen compounds, fragrances, antiseptics, fire retardants, surfactants, pesticides and plasticizers were detected in water.Among them benzotriazole, benzothiazole and their metabolites were found at higher concentrations in the polluted site at both sampling times.Traditional fish biomarker responses considered in this study confirmed an elevation of detoxification enzyme activities and oxidative stress parameters in fish sampled downstream from the WWTP.A study carried out on two cyprinid fish species from the same locations in summer 2012, which also included some of the biomarkers considered here (EROD and BFCOD), reached similar conclusions (Blanco et al. 2019).According to these authors, males from two species (S. laietanus and Barbus meridionalis) displayed delayed sexual maturation on the polluted site, and females of B. meridionalis expressed gonadal aromatase (an enzyme that converts testosterone to estradiol) induction.Our study confirms that, even 2 years later, the endocrine disruption on fish from this area still persists and affects S. laietanus as well.The inclusion of fish liver biomarkers confirmed chemical exposures to lipophilic organic microcontaminants: from dioxin-like chemicals (reflected as EROD activity increase) to drugs of broader nature (as BFCOD activity enhancement).Both parameters, together with conjugation enzymes (GST), carboxylesterase activities (CbE) and antioxidant defences (GPX and GR), were significantly elevated downstream from the WWTP, confirming the exposure to a broad nature of chemicals.Nonetheless, the observed differences in biomarkers response between seasons (spring and summer) could be attributed to changes in water temperature and the reproductive status of fish (Van der Oost et al. 2003).
Water characteristics in the polluted site may account for the reproductive disorder observed in fish of this study.Females upstream the WWTP followed the expected reproductive pattern characteristic of S. laietanus with a maximum number of vitellogenic oocytes in June (Casals 2005).By contrast, females downstream presented vitellogenic oocytes during the entire spawning season with no clear reproductive peak and a reduced amount of mature oocytes in April, which indicates an alteration of their gonadal cycle and reproductive output.The lower fecundity rates found for females downstream the WWTP were concomitant to a reduced gonadal development relative to body weight (GSI) and to the absence of a peak for this index throughout the breeding period.Significant reductions in GSI in sewage effluent exposed fish have been reported for several species (Hecker et al. 2002;Jobling et al. 1998;Tetreault et al. 2011;Vajda et al. 2008).The histological analysis of females also confirmed the absence of a downstream peak of reproduction.Females collected at both sites showed the five oocytes stages of maturation defined in this study.However, the maturity index (calculated from the proportion of oocytes in the different stages of maturation) was higher for females upstream the WWTP in June but this was not observed in the females downstream.No major histological alterations in the morphology of female gonads were recorded in the current study and in other sewage effluent exposed fish despite the occurrence reproductive alterations (Hecker et al. 2002;Jobling et al. 2002;Solé et al. 2003).
Similar results were observed for males.Males downstream the WWTP had lower GSI values as described by Kaptaner et al. (2016) for another cyprinid fish species.Males of both sites showed the four stages of maturation defined in the histological analysis.Although for males the maturity index was not significantly different between the two sampling sites, on the other hand males in the polluted site had narrower seminiferous tubules.Reduction in diameter of seminiferous tubules has been related to a decrease in male fecundity due to less space to stock sperm (Gimeno et al. 1998;Smith 1978).The lower value of GSI in males downstream may confirm this assumption.Gimeno et al. (1998) reported that a reduction in the diameter of the seminiferous tubules is related to fish exposed to 17 β-estradiol (E2) in Cyprinus carpio.
Although a high frequency of intersex among males of fish species in wild populations exposed to wastewater discharges has been described (Jobling et al. 1998;Tetreault et al. 2011), feminization in males has not been observed in this study.
According to Scott et al. (2010), the last one can be affected by environmental contaminants such as diazinon and synthetic progestogens.An desynchronised spawning reproductive pattern, such as the one revealed for the ovarian development in females of this study, was also reported in other fish field studies due to chemicals of an oestrogenic nature (Vajda et al. 2008;Woodling et al. 2006).However, for males, the effects of EDCs are unpredictable and variable (Leino et al. 2005).Some of the CEC classified as EDCs affect females but not males of teleost fishes (Dietrich and Krieger 2009).The EDCs detected at higher concentrations in the polluted site of this study were benzotriazole (up to 1.3 µg/L), benzothiazole (up to 0.5 µg/L) and their respective metabolites.Both compounds have been detected in several rivers of the Iberian Peninsula such as the Llobregat, Ter, Ebro and Tordera (Herrero et al. 2013;Matamoros et al. 2010a).Formerly, these EDCs were clearly associated with the activity of the textile dye plant of concern (Matamoros et al. 2010b).The antiestrogenic activity of benzotriazole has not been fully demonstrated in fish under experimental conditions (Harris et al. 2007).A laboratory experiment reported a degeneration of the ovaries and a stimulation of spermatogenesis in fish females at concentrations starting from 5 mg/L, while in males an increase in GSI and alterations in seminiferous tubules was found (Liang et al. 2014).However, this concentration was higher than that found in this study downstream WWTP site.Other laboratory studies have revealed that benzotriazole was weakly oestrogenic to fish (Tangtian et al. 2012) but does not cause serious abnormalities in testes (Dietrich and Krieger 2009).Benzothiazole have effects on the structure of the gills and on the swimming of fish (Evans et al. 2000).
The highly prescribed ibuprofen was detected downstream in the summer sampling (134 ng/L).This nonsteroidal anti-inflamatory drug is responsible for causing endocrine alterations in steroidogenic enzymes in an in vitro study for male fish (Fernandes et al. 2011).
Ibuprofen has also been related to the induction of vitellogenin in male fish, a fewer broods per pair and with a delay in hatching of eggs in Oryzias latipes (Han et al. 2010).Other compounds classified as EDCs (such as bisphenol A, diazinon, oxybenzone, and methylparaben) were found in the polluted site of this study.However these and other compounds (galaxolide and tonalide) were found at concentrations 10 times lower than those that can cause effects on oocyte production (Bhandari et al. 2015;Coronado et al. 2008;Kim et al. 2014;Maxwell and reproductive peak and a lower fecundity value.Males had narrower seminiferous tubules.The chemicals benzotriazole and benzothiazole, in mixture with other compounds, could be responsible for the observed reproductive disorders in this fish species although further studies are needed in order to clarify their role as ECDs in nature.

Table 6
Oocyte size for the four stages of development and atretic oocyte of S. laietanus expressed in µm 2 .Means and 95% confidence interval (CI) are shown.In order to test statistical differences, a GLMM (generalized lineal mixed model) was performed for each sex.* Significant differences by site.All significant differences are p ≤ 0.05

(
10.31, p = 0.002).All of these biomarkers presented higher levels in the fish collected downstream WWTP.A significant interaction between season and site factors was only found for BFCOD ( = 44.88,p < 0.001), which consisted of higher values in summer only downstream WWTP.In muscle, AChE and PrChE biomarkers showed a similar pattern: a significant interaction between season and site factors (AChE: = 19.06,p < 0.001; PrChE: = 7.35, p = 0.010) as well as a significant effect of the season (AChE: = 116.11,p < 0.001; PrChE: = 126.59,p < 0.001).

Fig. 1 Fig. 2
Fig. 1 Map of the sampling sites in the Ripoll River (Besòs basin, NE of the Iberian Peninsula).The black point indicates the location of the polluted site (1.6 Km downstream from the WWTP of a textile dye plant) and the white point indicates the location of the control site (2.7 Km upstream from the WWTP).A small dam (dashed line) located just above the textile dye plant (factory symbol) prevented the passage of fish from one site to the other

Fig. 3
Fig. 3 Photomicrographs of S. laietanus gonadal sections stained with haematoxylin and eosin stain, showing ovaries and testes at different developmental phases.(a) Mature testicle, showing spermatozoa (IV) at the centre of the seminiferous tubule and all the stages of development: spermatogonia (I), spermatocytes (II) and spermatids (III) (upstream male).(b) Mature ovary, showing oocytes in phase V (vitellogenic), as well as phase III and phase I (upstream female).(c) Overview of female gonad where phase II, phase IV and phase V (vitellogenic) oocytes are shown (upstream female).(d) Detail of an atretic oocyte (AO) found in some specimens of the study and phase I and phase III oocytes (downstream female).Black and white circles indicate maturation phases with a small size.yg: yolk granule, df: follicles after spawn

Table 1 769
Number of fish used in each section of this study: site characterization by biomarkers (B), 770 gonadal development and female fecundity (G & F) and histological analysis (H).Fork length 771

Table 2
Physico-chemical water parameters upstream and downstream WWTP where fish, S. laietanus

Table 3
Occurrence of CEC in river water samples (ng/L) during spring (March) and summer (July) 2014.PPCP pharmaceuticals and personal care products.Limit of detection LOD = 3 ng/L.EDCs for which an effect on fish has been described are marked with an asterisk