Introduction

Population growth and anthropogenic activities are associated with environmental degradation owing to the exponential increase of land use change and contaminant inputs to aquatic ecosystems (Ciszewski and Grygar 2016; Reza and Singh 2010). Contamination by metals and metalloids is one of the major quality issues in aquatic reservoirs due to their persistence and potential for bioaccumulation and biomagnification (Peters et al. 2013; Mishra et al. 2019). These elements have both natural and anthropogenic sources. Although some of them are essential for metabolic functions in organisms (e.g., Cu, Fe, Mn, Ni, and Zn; Appenroth 2010), frequently, there is a small difference between biological benefits and toxicity, depending on biotic and non-biotic parameters (Qi et al. 2010; Mishra et al. 2019).

Discharges of improperly treated effluents from urban, industrial, agriculture, and mining activities deliver high loads of suspended solids and trace elements to continental aquatic ecosystems (Ruiz-Fernández et al. 2005). Surface runoff plays an important role in the transfer of contaminants through catchments to lakes and reservoirs. These contaminants eventually accumulate in sediments (Reza and Singh 2010) which conform to an archive of the historical contaminant deposition. 210Pb dating has been used extensively to reconstruct the variation in contaminant inputs to aquatic reservoir sediments over the past 100–150 years (e.g., Ruiz-Fernández et al. 2004; Ontiveros-Cuadras et al. 2014; Méndez-García et al. 2016).

The Oviachic reservoir (OV, NW Mexico; Fig. 1) is one of the largest (~ 20,700 ha) in the region. It is at the lower reach of the Yaqui River Basin receiving discharges from the entire watershed, in which three large reservoirs (including OV) and approximately 20 smaller reservoirs are established (Muñoz-Arriola et al. 2009). The OV is of high social and economic relevance, since it secures irrigation to the Yaqui Valley, which is the second most extensive agricultural area in Mexico (232,944 ha of irrigated surface, Enríquez and Durán 2017; CONAGUA 2020a). Metal and metalloid contamination and human health issues in the Yaqui River area have been related to large-scale application of agrochemicals (Guillette et al. 1998; Adler 2005; Cantú-Soto et al. 2009; Meza-Montenegro et al. 2012).

Fig. 1
figure 1

Sampling sites at the Oviachic reservoir, Yaqui River Basin, Northwest Mexico

Full size image

Mining is an ancestral activity in NW Mexico. Sonora State is currently the main producer of Cu and Mo in Mexico, contributing to a relevant proportion of Ag, Au, and Fe (SGM 2018). The Yaqui River region has consolidated over the last 20 years as the largest producer of Au and Cu in Mexico, increasing Cu production from 301,212 ton year−1 in 2000 to 655,807 ton year−1 in 2018 (INEGI 2018). Inadequately confined mine tailings are significant sources of trace element contamination in mining zones (USEPA 1997; Reza and Singh 2010). Ore deposits in the catchment include La Caridad mine, one of the largest open-pit Cu mines in the world. Some studies reported metal and metalloid dispersion to adjacent areas derived from water and wind erosion (Meza-Figueroa et al. 2009; De la O-Villanueva et al. 2013). Moreover, much large-scale and artisanal gold mining occurs in the mid-Yaqui River Basin (SGM 2018). Miners concentrate gold using Hg as a gold amalgamator, and this activity has recently increased (SEMARNAT-INECC 2017).

The main purpose of this study was to evaluate temporal changes of sedimentation rates and trace element (As, Cr, Cu, Hg, Ni, Pb, and Zn) concentrations and fluxes, as recorded in two 210Pb-dated sediment cores from OV, under the hypothesis that changes in sedimentation and contamination are mostly driven by the recent development of major economic activities and population growth in the region.

Methods

Study area

The OV reservoir (also named Álvaro Obregón) is in Cajeme, southern Sonora State, ~ 37 km north of Obregón City (Fig. 1). The dam was built between 1947 and 1952 for flood control, irrigation of Yaqui Valley agriculture areas, power generation, and public water supply (CONAGUA 2018). It has an extension of 205 km2, 57 m mean depth (CONAGUA 2008), and a maximum ordinary water level (MOWL) of 3023 hm3, and it is located 105.5 m above sea level (SINA 2020). It is mainly fed by the Yaqui River, although it also receives the contributions of Agua Caliente River (through the Matanza stream; Fig. 1) and three aquifers (Cumuripa, Tecoripa, and Aguacaliente; CONAGUA 2015). OV is surrounded by rural areas; it supports recreational and subsistence fishing activities (carp, catfish, tilapia, and largemouth bass characterized mostly by benthic feeding habits). The Yaqui River has a natural surface runoff of 3148.2 hm3 year−1 and a basin area of 74,640 km2 (CONAGUA 2018), in which the mean precipitation is 527 mm year−1 (INEGI 2000).

North-south-oriented, low mountain ranges, separated by plains, characterize the area. The slopes are quite steep, especially in intrusive and metamorphic rocks, and the streams that drain this region promote intense erosion, producing lateral spurs that project into the plains (INEGI 2000). The Yaqui River Basin is mainly composed of rhyolite-acidic tuff (61% of the basin surface), alluvial soil (19%), and conglomerates (6%; Garrido-Hoyos 2019). The OV flood plain consists of conglomerates and agglomerates covered by alluvial terraces, outcropping basalts, and rhyolitic tuffs (SGM 2002). The main soil types in the area are cambisol, fluvisol, lithosol, regosol, vertisol, and xerosol (INAFED 2019). The climate in Cajeme is hot arid, predominantly desert (BW(h')w in 53.8% of the area) and steppe (BSo(h')w, in 34.5%), with a mean annual temperature above 18 °C and a dry winter (SEDESOL 2013). The precipitation occurs mainly from July to early September (Mexican monsoon season), and the average annual precipitation is 366.1 mm (INEGI 2017).

Sampling and analysis

Two sediment cores were collected in November 2018 (Fig. 1, Table 1) with a gravity corer (UWITEC™) using Plexiglas tubes (8.6 cm internal diameter and 1.5 m long). The sediment cores were extruded and sectioned into 1cm intervals. The core sections were freeze-dried at − 50 °C under vacuum (~ 0.1 mBar). Mass weight was recorded before and after drying. Samples were ground with a porcelain mortar and pestle (except aliquots for grain-size analysis).

Table 1 Sampling locations and dating information for sediment cores from the Oviachic reservoir, Yaqui River Basin, northwestern Mexico
Full size table

Radiometric analyses were carried out every 4 cm depth along the cores, except for the bottom segments (below 57 cm depth in OV-1 and 45 cm depth in OV-2) where measurement resolution was every 2 cm. 210Pb total (210Pbtot) activities were determined by alpha-particle spectrometry (Ortec Ametek Model 576A) through its descendant isotope 210Po, assuming secular equilibrium between both radionuclides (Ruiz-Fernández and Hillaire-Marcel 2009). 226Ra and 137Cs activities were determined by gamma-ray spectrometry (Ortec Ametek High Purity Germanium well-detector; Ruiz-Fernández et al. 2014, Díaz-Asencio et al. 2020). Counting uncertainties ranged as follows: 5–6% for 210Pb, 4–7% for 226Ra, and 6–17% for 137Cs. We assumed that supported 210Pb (210Pbsup) was in equilibrium with its parent radionuclide 226Ra. Excess 210Pb (210Pbex) corresponds to the difference between 210Pbtot and 210Pbsup (226Ra).

Bulk geochemistry of sediments was determined at each core section in order to obtain a detailed record of the composition changes. Major and trace element sediment composition was determined by X-ray fluorescence spectrometry (XRF, Spectrolab Xepos-3) under helium atmosphere, using ~ 2 g of ground sediment placed into low-density polyethylene cells (bottom covered with Prolene film) and manually compressed with a Teflon rod. Hg concentrations were quantified by atomic absorption spectrophotometry (AAS; Varian SpectrAA-240FS) coupled to a cold vapor generator (Varian VGA-110), by using ~ 0.3 g of dry and ground sediments, which were acid digested (3:1 HCl and HNO3) on a hotplate (80–100 °C, 5 h). The digestate was diluted with HPLC-grade water (Krishnamurty et al. 1976).

The sediments were also analyzed for grain-size distribution (sand, silt, and clay) by laser diffraction (Malvern Mastersizer 2000) on aliquots previously digested with 30% H2O2 to remove organic matter. For magnetic susceptibility (MS) measurements, we used a Bartington-MS2 equipment coupled to a MSG2 frequency sensor. Total carbon (Ctot) and nitrogen (Ntot), and the organic fraction of carbon (Corg) and nitrogen (Norg) concentrations, were analyzed with an Elementar Vario Micro Cube elemental analyzer. For the determination of the organic fraction, we used decalcified sediment aliquots, for which sediments were treated with 1 M HCl (drop by drop until no further effervescence was observed) and freeze-dried before analysis. The combustion temperature was 550 °C. The instrument was calibrated with the Soil Standard Silty B2182 and the High Organic Sediment Standard B2151 (Elemental Microanalysis Ltd.). Accuracy was assessed with the soil reference material Leco 502 697. The inorganic carbon fraction (Cinorg) was estimated from the difference between Ctot and Corg.

The accuracy assessment of methods included analysis of certified reference materials and control charts, in which the central value corresponds to the certified value for the analyte, and the upper and lower limits are the 95% confidence interval of the certified reference materials (CRM): IAEA-300 for 210Pb and 137Cs; IAEA-158, IAEA-405, and IAEA-433 for XRF; MESS-3 for AAS; LECO 502-309 for carbon and nitrogen concentrations; QAS3002 for grain-size distribution, and Bartington-G039 for MS. Moreover, all CRM analysis results are routinely checked through the Zeta-score approach (to compare measurements and CRM certified values; Kuselman and Fajgelj 2010). Results fell within the certified range values in the control charts, and the Zeta-score values were within the acceptable range of ± 2. Replicate analysis (n = 6) of a single sample showed that the variation coefficients were 5.8% for 210Pb and 2.5% for 226Ra and 137Cs activities; 1.7% for Al, 2.6% for Ti, 2.8% for As, 5.4% for Cr, 5.7% for Cu, 4.5% for Ni, 2.9% for Pb, and 2.7% for Zn; 5.9% for Corg, Cinorg, and N concentrations; 6% for sand and clay, and 3% for silt percentages; and 3% for MS values.

Data treatment

210Pb chronology

210Pb chronologies, mass accumulation rates (MAR, g cm−2 year−1), and sediment accumulation rates (SAR, cm year−1) were calculated by using the constant flux (CF) model (Robbins 1978; Sánchez-Cabeza and Ruiz-Fernández 2012), and dating uncertainties were estimated by Monte Carlo simulation (Sanchez-Cabeza et al. 2014). 137Cs activity profiles were used as stratigraphic markers to attempt the corroboration of the 210Pb age models, assuming that the highest values correspond to the period of maximum 137Cs fallout, caused by atmospheric thermonuclear weapon tests, during 1962–1964 (Robbins 1978).

Metal fluxes and flux ratio

Metal fluxes (μg cm−2 year−1) derive from the product of the element concentration (μg g−1) and MAR (g cm−2 year−1) at each core section (Sanchez-Cabeza et al. (2014). The flux ratio, a concise representation of historical element inputs to an ecosystem (Heyvaert et al. 2000), was calculated as the quotient of modern to pre-dam construction fluxes for each sediment core.

Enrichment factor

We evaluated the degree of trace element contamination through the enrichment factor (EF) approach (Buat-Menard and Chesselet 1979). Probably, it is the most widely used and most successful enrichment indicator presently in use (Birch 2017). For each core section, concentrations were normalized by Al, used as a reference element to compensate for variations of sediment grain size (Loring and Rantala 1992). Then, EF was calculated as the ratio of normalized element concentrations to pre-dam construction values (Table 2). Element pre-dam concentrations were estimated from the mean values in the three bottom sections of each core, nearly corresponding to the completion of the Oviachic reservoir. The EF was graded according to Essien et al. (2009): minor enrichment (1.5 < EF < 3), moderate enrichment (3 < EF < 5), and strong enrichment (EF > 5).

Table 2 Trace element concentration (μg g−1), enrichment factor (EF), flux (μg cm−2 year−1), and flux ratios in cores from Oviachic (OV) reservoir, Yaqui River Basin, northwestern Mexico
Full size table

A second approach was also used to estimate the enrichment factor (EFUCC) by considering the recommended upper continental crust values (Rudnick and Gao 2004) as background concentrations.

Sediment quality assessment

In order to assess the ecological significance of the trace element concentrations in OV, these were compared with the benchmarks threshold effect level (TEL; concentration at which adverse effects rarely occur) and probable effect level (PEL; concentration at which adverse effects occasionally occur) (Table 2; MacDonald et al. 2000).

Statistical analysis

Descriptive statistics (sample size, maximum, minimum, mean, and standard deviation) were calculated for geochemical characteristics and metal concentrations. Differences between mean values of the variables at each core were assessed through analysis of variance (ANOVA) with Tukey test at 95% of confidence (p < 0.05). Pearson’s correlation analysis was performed to identify the association among variables analyzed, and significant correlation coefficients were identified through the Student’s t test (p < 0.05).

Factor analysis (FA) was used to identify the primary sources of trace element variability in the dataset. The variables included the trace element concentrations (Al, As, Ca, Cr, Cu, Fe, Hg, Mn, Ni, Pb, Rb, Ti, Zn, and Zr) and the geochemical variables (MS, Corg, Cinorg, and grain-size fractions), for 115 observations (OV-1 and OV-2 sections). Significant variables within a factor should have at least a factor loading > |0.3| and a communality > 0.20 (Child 2006). In this study, the significant variables were those with loadings > |0.65| and communalities > 0.5.

Results

210Pb chronology

210Pbtot activities were similar between cores OV-1 and OV-2 (p > 0.05), ranging from 81.0 ± 5.2 to 183 ± 11 Bq kg−1, and 210Pbex activities from 30.4 ± 2.2 to 49.2 ± 2.6 Bq kg−1. 210Pbex activities in both cores decreased with depth, approximating exponential decay curves, with significant correlation coefficients (p < 0.05): r = 0.85 in OV-1 and r = 0.96 in OV-2. However, in some core sections, 210Pbex activities deviated from an exponential declining trend. As 210Pbex did not reach equilibrium (where 210Pbex activities approach zero), we estimated the missing 210Pbex inventories at each core (14% for OV1 and 11% for OV2) (Appleby 1998; Sánchez-Cabeza and Ruiz-Fernández 2012). The bottom section ages were 62.7 ± 0.8 years (section 63–64 cm, dated 1956) for OV-1 and 69.6 ± 0.7 years (49–50 cm, 1949) for OV-2.

137Cs activities were similar in both cores (p > 0.05), ranging from 3.9 ± 1.1 to 13.1 ± 1.2 Bq kg−1 for OV-1 and 2.9 ± 0.7 to 11.8 ± 1.0 for OV-2. 137Cs activity profiles were also similar, with values decreasing upwards and a region with the highest but fluctuating values towards the core bottoms, below the section 57–58 cm (dated 1966 ± 1 year) in OV-1 and 45–46 cm (1960 ± 1 year) in OV-2 (Fig. 2).

Fig. 2
figure 2

Profiles of total 210Pb and 137Cs activities and accumulation rates of sediment cores from the Oviachic reservoir, Yaqui River Basin, Northwest Mexico

Full size image

MAR values were comparable between cores (p > 0.05), ranging from 0.13 ± 0.03 to 0.47 ± 0.13 g cm−2 year−1 for OV-1 and 0.13 ± 0.03 to 0.38 ± 0.06 g cm−2 year−1 for OV-2. SAR ranged from 0.39 ± 0.10 to 2.67 ± 0.32 cm year−1 for OV-1 and 0.34 ± 0.06 to 2.48 ± 0.24 cm year−1 for OV-2 (Table 1). Both SAR profiles (Fig. 2) showed increasing values from the past to reach maxima by the sampling year (2018), whereas MAR values showed subsurface maxima, with the highest value in between 2001 and 2002 (Fig. 2), more clearly observed in OV-1.

Sediment characterization

Both core sediments were predominantly clayey-silty (silt 43–67%, clay 33–56%) with very low (nil to 4%) and almost constant sand content. We observed transitions of clay and silt fractions (Fig. 3) at the upper sections (2002 to 2015 for OV-1 and 1985 to 2016 for OV-2).

Fig. 3
figure 3

Profiles of geochemical variables and trace element concentrations of sediment cores from the Oviachic reservoir, Yaqui River Basin, Northwest Mexico

Full size image

Corg percentages were higher (p < 0.05) in OV-2 than OV-1, and values decreased with depth, approaching the typical exponential decay (r = 0.69 in OV-1 and r = 0.85 in OV-2) characteristic of Corg mineralization (Berner 1964). Cinorg percentages were low and also showed higher values up core. The molar ratios Corg to Norg (C:N) were highly variable (9.75–14.06 for OV-1 and 8.56–11.17 for OV-2) with comparable values between cores (p > 0.05). MS values were similar between cores (p > 0.05), with the lowest values at the core tops and the highest values in subsurface sections (10 to 25 cm, 1998–2011, in OV-1, and 10 to 32 cm, 1982–2011, in OV-2; Fig. 3). The concentration ranges of Al, Ca, Fe, Mn, Rb, Ti, and Zr were also comparable between cores. The profiles of Al, Fe, and Ti concentrations were alike in both cores, resembling those observed in the MS profiles (Fig. 3), with the highest values at subsurface sections. Rubidium and Zr concentrations increased from the core bottom up to 20–21 cm (2002) in OV-1 and from 26 to 27 cm (1991) in OV-2, and then declined upwards. In contrast, Ca and Mn concentrations decreased from the core bottom up to 23–24 (2000) in OV-1 and 23–24 cm (1995) in OV-2, and then increased steadily towards the surface.

Trace element concentrations, enrichment, and risk assessment

Trace element (As, Cr, Cu, Hg, Ni, Pb, and Zn) concentrations were similar in both cores (Table 2; Fig. 3), with almost constant distribution with depth. However, Ni, Pb, and Zn values decreased upcore, and As, Cu, and Hg maxima were observed at subsurface sections, although such increments were asynchronous (Fig. 3).

Enrichment factors (Table 2, Fig. 1S), by using pre-dam concentrations as reference values, indicated null enrichment for Cr, Cu, Ni, and Zn (EF < 1.5) along both cores; minor enrichment for As (2014–2018 in OV-1, 2015–2018 in OV-2) and Pb (1984–1996 in OV-1, 1966–1975 in OV-2). We detected a strong Hg enrichment in OV-1, and a moderate Hg enrichment in OV-2, since the early 2010s, reaching the maximum EF (5.84 for OV-1 and 3.02 for OV-2) at the core surface in both cores. The enrichment factors based on the upper continental crust concentrations (EFUCC) were lower for Cr and Ni (< 1), but higher for the rest of the elements, than the enrichment factors (EF) based on pre-dam concentrations (Table 2), accounting for null enrichment for Cr and Ni concentrations, minor enrichment for Cu and Zn, minor to strong enrichment for Pb, and moderate to strong enrichment for As and Hg.

Along both cores, Cr concentrations were below the TEL value, whereas Cu, Hg, Ni, and Pb were above TEL but below PEL. Zn concentrations were above TEL in most sections, except in the uppermost segments (0–10 cm, 2012–2018, for OV-1, and 0–18 cm, 2001–2018, for OV-2). Arsenic concentrations in all of the core sections were above TEL, and above PEL in the uppermost segments (0–10 cm, 2012–2018, for OV-1, and 0–6 cm, 2014–2018, for OV-2) (Fig. 3).

Metal fluxes and flux ratios

Trace element fluxes in both cores increased with time since dam construction (mid-1950s). In OV-1, two maxima were observed (1977 and 2001), which resemble the MAR maxima. In both cores, fluxes of Pb, Ni, Cr, and Zn decreased towards the present (mostly since ~ 2000, although in OV-2 Pb fluxes decreased since ~ 1980s). Conversely, Cu, As, and Hg fluxes showed increasing values within the past 10 years (Fig. 4). Flux ratio ranges were comparable among most elements (Fig. 5, Table 2) with maxima accounting for up to ~ 4-fold increments for As, Cu, Ni, and Zn; ~ 5-fold for Cr and Pb; and ~ 15-fold increment for Hg.

Fig. 4
figure 4

Trace element fluxes in sediment cores from the Oviachic reservoir, Yaqui River Basin, Northwest Mexico

Full size image
Fig. 5
figure 5

Trace element flux ratios in sediment cores from the Oviachic reservoir, Yaqui River Basin, Northwest Mexico

Full size image

Trace element provenance

We used factor analysis to assess which variables better explained the changes in trace element concentrations, and to attempt elucidating their primary sources. Two factors explained ~ 70% of the dataset variability (Fig. 6, Table 3). The significant variables in factor 1 (~ 45% common variance) were Zn, Zr, Fe, Rb, and Pb (positive loadings) and Corg, Ca, Mn, Cinorg, Hg, As, and Cu (negative loadings). In factor 2 (25% common variance), the significant variables were silt (positive loading) and Ti, Al, clay, MS, and Ni (negative loadings).

Fig. 6
figure 6

Factor analysis biplot for geochemical variables and trace element concentrations in sediment cores from the Oviachic reservoir, Yaqui River Basin, Northwest Mexico. a Distribution of variables and observations (core sections); b and c distribution of observations per year

Full size image
Table 3 Factor analysis results for element composition in sediment cores from Oviachic reservoir, Yaqui River Basin, northwestern Mexico
Full size table

The presence of lithophile elements in both factors, often used as indicators of terrigenous contribution (Zr, Rb, Al, and Ti; Croudace and Rothwell 2015), suggested two different terrigenous sources. Zr is an immobile element resistant to chemical weathering and commonly associated with detrital heavy minerals (Salminen et al. 2005), usually derived from igneous rocks, whereas Rb mostly associates with micas and K-feldspars in igneous and metamorphic rocks (Faure and Powell 1972). Al and Ti are present in many silicate minerals, such as mica, amphibole, and pyroxene, and can be enriched in weathered products of primary Al minerals such as secondary clay minerals (Salminen et al. 2005).

Thus, the presence of Zr and Rb in factor 1 accounts for the input of detrital primary minerals from the catchment. Al and Ti in factor 2 show the contribution of clay minerals in the finest sediment fraction. The significant correlations (p < 0.05) observed among Pb, Zn, and the lithophile elements Zr and Rb (Table 3) suggest that both trace elements have a predominant minerogenic source. The close relationship of As, Cu, and Hg with Corg, Ca, Cinorg, and Mn suggests scavenging of these trace elements by adsorption onto suspended materials (e.g., carbonates, represented by Ca and Cinorg), which serve as nucleation centers for the deposition of Mn hydroxides and organic coatings (Salomons and Förstner 1984) and can form strong complexes with trace elements. In summary, factor 1 accounts for the input of clastic debris and the scavenging of trace elements by suspended particles, and factor 2 accounts for the detrital contribution within fine particles (clays and silt). Ni and Cr showed non-significant loadings in both factors, as well as low, but significant (p < 0.05) correlations with the lithophile elements Al, Ti, Zr, and Rb, for which the main provenance of both trace elements was attributed to both detrital sources.

The distribution of the sediment sections in the loading biplot was similar for both cores (Fig. 6), showing a grouping transition: the bottom segments of the cores (between the late 1950s and 1990s) were characterized by the highest content of silt and Pb, as well as increasing concentrations of Rb and Sr, mostly grouped in quadrant I; the intermediate sections (between the 1990s and 2010s) were characterized by the highest values of Al and MS, mostly in quadrant II; and the uppermost sections (the 2010s) showed the highest concentrations of Corg, Cinorg, Ca, As, Hg, and Cu, and the lowest terrigenous contribution (Al, Ti, Zr, Rb, and MS), across quadrants III and IV (Fig. 6).

Discussion

210Pb geochronology

210Pbex activity profiles in cores OV-1 and OV-2 broadly showed a decreasing trend, although some segments showed relatively constant values, which may be caused by mixing or sediment supply variability. Post-depositional mixing was considered negligible, because this process would homogenize the sediments and, consequently, other profiles (e.g., 137Cs or trace elements, Figs. 2 and 3) would have shown similar patterns, which is not the case (Fig. 2). 210Pbex activities in sediments result from the balance between the 210Pbex flux to the sediments and sediment input (i.e., the mass accumulation rate, MAR) (Krishnaswamy et al. 1971). Consequently, 210Pbex activities would decrease due to dilution by a higher MAR (as observed in the 23–60 cm segment of core OV-1) or increase if the MAR was lower (Sánchez-Cabeza and Ruiz-Fernández 2012). Thus, assuming that post-depositional sediment mixing is negligible, 210Pbex activities deviating from an exponentially decreasing curve were interpreted as sediment accumulation changes. Factors influencing sediment inputs may include climate variability (the occurrence of drought or very wet periods), land use changes, and/or human-induced regulation of the water supply for flood control and land drainage.

137Cs activities in both cores were comparable to or slightly lower than those reported in other continental aquatic ecosystems in Mexico, e.g., Espejo de Lirios reservoir (13.5 Bq kg−1; Ruiz-Fernández et al. 2004), Zirahuén Lake (6–18 Bq kg−1; Davies et al. 2004), San Marcos reservoir (7–22 Bq kg−1; Méndez-García et al. 2016), and Valle de Bravo reservoir (5–20 Bq kg−1; Carnero-Bravo et al. 2014), but higher than in Chapala Lake (˃ 1–7 Bq kg−1; Ontiveros-Cuadras et al. 2014).

137Cs activity profiles did not show a clear 137Cs maximum that could be unequivocally attributed to the maximum 137Cs fallout period (1962–1964; Robbins 1978). The highest values were observed at the deepest segment of both cores (below sections 57–58 cm in OV-1 and 45–46 cm in OV-2), and values decreased towards the present (Fig. 2). These features are compatible with the history of 137Cs released to the environment in the northern hemisphere, with the onset occurring in the late 1940s (after the Trinity test in 1945), increasing fallout until a maximum in the early 1960s, and declining through the late 1960s and 1970s, after the nuclear test ban treaty went into effect in 1963 (UNSCEAR 2000). A single well-defined maximum of 137Cs activity would only occur in environments where sediment accumulation was constant through the period and with a small 137Cs catchment supply (Appleby et al. 2019), a condition rarely observed in aquatic ecosystems with large catchments, and impacted by land use change and soil erosion, such as the OV reservoir. Indeed, according to 210Pb dating, MAR of both cores almost doubled within the period where the highest 137Cs values were observed, diluting 137Cs sediment activities. In addition, catchment inputs can produce a delay of the 137Cs maximum, particularly in lakes with large catchments (Appleby et al. 2019), such as the OV reservoir. Furthermore, 137Cs is highly mobile and can diffuse through interstitial waters, caused by ion-exchange displacement of 137Cs by cations such as NH4+, Fe+2, and Mn+2 released under anaerobic conditions (Evans et al. 1983), as observed in other lacustrine environments (Ligero et al. 2002; Ruiz-Fernández et al. 2004; Lan et al. 2020), disrupting the shape of the 137Cs maxima. Large catchment inputs and 137Cs diffusion commonly affect 137Cs activity profiles from many lacustrine environments (e.g., Lan et al. 2020), which makes it difficult to corroborate 210Pb age models if a single clear 137Cs maximum is expected. Nonetheless, the 137Cs maxima in OV cores occurred between the onset and the period of maximum global fallout from nuclear testing, which is compatible with the 210Pb-derived chronology. According to the 210Pb-derived age models, both cores nearly reached the pre-dam soil, as the bottom age of the cores was 1955 for OV-1 and 1949 for OV-2, which are consistent with the dam construction period (between 1947 and 1952; CONAGUA 2011).

The SAR range in OV reservoir (0.34–2.67 cm year−1; Table 1, Fig. 2) was comparable to other reservoirs in Mexico, such as El Granero and San Marcos (~ 1 to ˃ 2 cm year−1; Méndez-García et al. 2016), and Intermedia and Silva (~1 cm year−1; Hansen 2012). However, MAR is a better indicator to assess sedimentation processes in aquatic environments since it accounts for sediment compaction (Sánchez-Cabeza and Ruiz-Fernández 2012). MAR in OV reservoir (0.13–0.47 g cm−2 year−1; Table 1 and Fig. 2) was similar to Espejo de Lirio Lake (0.04–0.40 g cm−2 year−1; Ruiz-Fernández et al. 2004) and Valle de Bravo reservoir (0.12–0.40 g cm−2 year−1; Carnero-Bravo et al. 2014), but higher than in Lago Verde Lake (0.03–0.10 g cm−2 year−1; Ruiz-Fernández et al. 2007), Zempoala Lagoon (0.129 g cm−2 year−1; van Afferden et al. 2005), and Miramar Lagoon (0.053 g cm−2 year−1; Hansen 2012).

The MAR steadily increased in both cores (Fig. 2), and the most recent values, although lower than those recorded in early 2000, are 2.5-fold higher than those recorded since OV construction. These MAR increments can be caused by the progressive population growth and development of anthropogenic activities in the catchment. Dam reservoirs are fed by erosion products of their watersheds, and as the dam building facilitates access to remote areas for new activities (e.g., road building, logging, farming, mining), they contribute to accelerated deforestation, soil loss, and sediment input to the reservoir. The Cajeme municipality, where OV is located, experienced a rapid population growth and socio-economic changes since the 1950s. The population increased almost fourfold within a decade (31,678 inhabitants in 1950 and 124,162 inhabitants in the 1960s; SE 1953; INEGI 1962) and about ~ 14 times between 1950 and 2015 (433,050 inhabitants; INEGI 2016). However, although population growth has continued to increase, MAR has decreased since the 2000s, suggesting the effect on MAR of additional factors (e.g., agriculture practices, climate variability).

Trace element concentrations, enrichment, and risk assessment

Concentrations of Ni and Zn in OV reservoir were comparable to the ranges reported for natural levels around the study area (7.09–59.3 μg g−1 for Ni and 66.2–107.9 μg g−1 for Zn; SGM 2015), but higher than those of Cu (13.7–27.3 μg g−1) and Pb (14.7–35.1 μg g−1) (Table 2). Besides, most element concentrations in the OV reservoir are comparable with those reported for other reservoirs in Mexico and elsewhere, affected by agriculture, urban, or industrial activities (Table 4). However, the Hg maximum is considerably higher than in all these reservoirs.

Table 4 Trace element concentrations (μg g−1) in sediment cores from reservoirs in Mexico and elsewhere
Full size table

Enrichment factors were used to assess trace element contamination in OV. The short period contained in the sediment cores (~ 70 years) precluded the estimation of pre-industrial concentration levels in the area, but were useful for evaluating element enrichment since OV reservoir completion. Results indicated null enrichment by Cr, Cu, Ni, and Zn, minor enrichment by As and Pb, and strong enrichment by Hg (Fig. 1S, supplementary information). Studies elsewhere have used average shale or continental crust levels as background concentrations (e.g., Birch 2017) to estimate enrichment factors, here referred to as EFUCC. In comparison with EF based on pre-dam values, EFUCC indicated depletion of Cr and Ni, and different degrees of enrichment for the rest of the elements, including strong enrichment for As and Hg. However, this approach has the disadvantage that background levels do not represent the specific geochemistry of the zone, which makes more uncertain the evaluation and may lead to wrong conclusions (i.e., either overestimate or underestimate enrichment), as discussed in other studies elsewhere (Rodríguez-Oroz et al. 2017; Arhin et al. 2017). This is because (1) the continental crust is composed of highly diverse lithologies, and compositional estimates may not be representative of the total continental mass (Rudnick and Gao 2004), and (2) as matter is transferred from the crust to the aquatic environment mainly via soil erosion, elements undergo natural chemical fractionation, and it is unlikely that concentrations remain constant during the biogeochemical cycle (Reimann and de Caritat 2000). Henceforward, trace element enrichment is discussed using enrichment factors based on pre-dam concentrations (EF). Despite the low enrichment, As concentrations during the last 5 years were above PEL indicating that it might cause deleterious effects on benthic biota (MacDonald et al. 2000; Buchman 2008), mainly linked to malfunction of the immune system (Duker et al. 2005).

Trace element fluxes and flux ratios

The historical metal fluxes showed increasing values since the early 1960s and high variability for most of the metals (Fig. 4). Trace element fluxes were comparable or lower than those reported for lacustrine environments, from either remote or anthropogenic-influenced areas, except for Hg, for which the maximum was considerably higher (Table 5). The long-range atmospheric transport and deposition of Hg is frequently reported around the world (Schroeder and Munthe 1998), and atmospheric deposition rates in lakes have increased during the last 20 years by a factor of 9–400 compared to background values (Biester et al. 2007). On a global scale, the major anthropogenic contribution to emissions is gold production using Hg technology (Pacyna et al. 2006). The major source of anthropogenic Hg emissions in Mexico by 1990 was nonferrous metal refining (including gold and silver refining, Pai et al. 2000). However, official data do not account for gold mining activities around Oviachic reservoir (SE 2019), and informal and unregulated mining activities (including placer-gold mining) occur at a small scale on the Yaqui River (SEMARNAT-INECC 2017).

Table 5 Trace element fluxes (μg cm−2 year−1) in sediment cores from reservoirs in Mexico and elsewhere
Full size table

Trace element provenance

Based on the factor analysis, Cr, Ni, Pb, and Zn delivery to the lake is predominantly terrigenous. Previous studies in Mexico and elsewhere show that Pb concentration changes in lacustrine sediments reflect the history of leaded gasoline consumption (e.g., Soto-Jiménez et al. 2006; Escobar et al. 2013). The use of tetraethyl lead in Mexico as an antiknock agent in gasoline began in the 1940s; government controls for Pb emissions started in 1982, and the phase-out of leaded gasoline was completed in 1997 (Soto-Jiménez et al. 2006). Only core OV-1 showed temporal profiles of Pb EF (Fig. 1S) and flux ratio (Fig. 5) that could be compatible with the history of Pb emissions in Mexico. However, if the direct atmospheric deposition of alkyl Pb were a significant source of Pb to the reservoir, the Pb enrichment and flux ratio profiles would be expected to show similar trends, which is not the case. The similarity of the flux ratio profiles of Cr, Ni, Pb, and Zn (Fig. 5) suggests that the input of these elements to the lake has a common source, most likely eroded soils from the catchment, which can be observed between the 1970s and 1990s in the flux ratio profiles (Fig. 5).

Erosion in the OV catchment is severe because of soil vulnerability and vegetation cover loss, attributed to overgrazing and land clearing (López-Reyes 2001), promoted by national government policies established in the 1970s to promote agricultural and livestock activities, including incentives to remove vegetation (Cedeño Gilardi and Pérez Salicrup 2007). In Sonora, extensive land clearing was made for grassland cultivation to feed livestock, especially Buffelgrass (Cenchrus ciliaris L.), the extension of which increased from 1000 ha in the 1970s to 400,000 ha in the late 1980s (López-Reyes 2001). Buffelgrass erosion rates are positively correlated with plant densities, owing to inefficient pasture management, i.e., overgrazing (Valdez-Zamudio and Guertin 2000). The reduction of Cr, Ni, Pb, and Zn fluxes since the 2000s may also be associated with government policies related to the implementation of control measures on land use change from forest to agricultural lands (DOF 1994), the deterioration of the agro-livestock sector as a consequence of unfavorable economic conditions, and the growth of the industrial sector in Sonora (COESPO 2010), but also likely with climate variability and water scarcity.

An increasing trend in atmospheric temperature has been reported for the lower basin of Yaqui River within the past 4 decades (Martínez-Austria et al. 2014), and since the late 1990s, its watershed has experienced strong drought conditions. Since then, the average annual runoff is below the historical mean (3061 hm3 per year; Minjares-Lugo and Salomón-Castelo 2009), causing the reservoir system, including OV, to attain the lowest historical level in 2004, with only ~ 13% of the storage capacity (Cortés-Jiménez et al. 2009). Rainfall reduction not only affects river runoff but also reduces the efficiency of the aquifer recharge from surface runoff infiltration, which can affect the height of the water table, the concentration of trace elements in groundwater, and the concentration of dissolved constituents (major ions and trace elements) in the reservoir waters.

Reduced rainfall may explain the highest concentrations observed since the early 2000s for Mn, carbonates (Cinorg and Ca), and Corg, and the fluxes and flux ratios of As and Cu, because of the concentration enhancement of these constituents, either in the aquifer waters or within the reservoir. Little is known about the three aquifers discharging into OV, except that (a) they are an unconfined type with shallow phreatic levels, formed by a sequence of coarse grain alluvial deposits with high permeability (restricted to the channels of streams with reduced thickness), and below this sequence and outside the channels of rivers and streams, the aquifer is made up of conglomerates, volcanic and intrusive rocks that present secondary permeability due to fracturing; (b) the main source of recharge is rainfall infiltration from surface runoff; (c) the area where they transit is rural (with scarce seasonal agriculture and cattle raising) without urban/industrial centers; (d) they are underexploited; and (e) the water quality is adequate for human consumption (low total dissolved solids and electric conductivity, high dissolved oxygen concentrations, and circumneutral pH), although it is not ruled out that major ions and/or heavy metals in high concentrations could result from water-rock interaction during groundwater transition associated with longer residence times and transit of water in the aquifer (CONAGUA 2015).

Despite the mining activities in the region, As and Cu in OV sediments are mostly of natural origin (as indicated by the low EF), most likely the result of weathering in the catchment and water-rock interaction within the aquifers draining to the reservoir. Strongly reducing aquifers derived from alluvium or loess, or inland aquifers in arid or semi-arid environments (such as this case), may have high As concentrations, often in combination with Mn and other trace elements, especially when groundwater movement is slow and there is element bedrock enrichment, which allows the accumulation rather than flushing of toxic metals (Mitchell et al. 2012). Although we lack information on trace element concentrations of the water delivered to OV reservoir, it is likely that lower recharge of the aquifers draining to OV reservoir caused by lower rainfall has promoted the delivery of water with higher concentrations of As, Cu, Mn, and carbonates.

According to the meteorological records in OV (CONAGUA 2020a), the mean annual temperature between 1970 and 2015 has increased by ~ 0.6 °C decade−1 and, although the rainfall record is very irregular, since 1985, most maxima have been considerably lower than those of the previous years (Fig. 7). Evaporation within the reservoir has increased as a result of higher temperatures and low rainfall, promoting higher authigenic carbonate precipitation. All trace element flux ratios presented two maxima around 1977 and 2000 (most evident in OV-2; Figs. 3 and 4), which correspond to the lowest rainfall (285 mm during 1977 and 261 mm during 2000; CONAGUA 2020b). This may be related to excessive evaporation and continuous precipitation of metals during that period.

Fig. 7
figure 7

Annual variations in precipitation, temperature, and evaporation at the Oviachic reservoir, Yaqui River Basin, Northwest Mexico

Full size image

Although Hg has a significant correlation with As and Cu (most likely owing to scavenging, common for the three elements), the timing and shape of the Hg flux profiles are different, and EFs and flux ratios are higher for Hg (Figs. 4 and 5) than for As and Cu, suggesting an additional anthropogenic source. This additional Hg source is most likely related with the relatively recent proliferation of small-scale artisanal gold mining activities, which use mercury as amalgamator, across the Yaqui River Basin (personal communications from locals; SEMARNAT-INECC 2017). Local newspapers report that Sonora is experiencing a “new gold rush” caused by the availability of gold resources and current high gold prices. In addition, there are environmental authority reports since 2010 informing about the suspension of activities and imposition of fines to irregular mining areas, some of them on the north of the OV reservoir, along the Yaqui River banks (e.g., PROFEPA 2010).

Conclusions

The reconstruction of temporal changes in sedimentation rates and trace element (As, Cr, Cu, Hg, Ni, Pb, and Zn) enrichment and fluxes, as recorded in two 210Pb-dated sediment cores from Oviachic reservoir, showed that the sediment load has increased steadily over the past 70 years (currently ~ 2.5 times higher than when the reservoir was completed) and that trace element contamination is mostly minimal, except for Hg. Nonetheless, according to international guidelines, some of the As, Cu, Pb, and Hg concentrations could be noxious for the benthic biota (and humans, through fish consumption). Highest Cr, Ni, Pb, and Zn fluxes, recorded between the 1970s and the 1990s, were attributed to soil erosion in the watershed, promoted by inadequate agriculture and husbandry practices. The recent increasing fluxes, observed within the past decade for As and Cu, are likely triggered by dryer climate conditions, enhancing concentrations either by lower rainfall and aquifer recharge, or water evaporation in the reservoir; whereas for Hg, the input of wastes from artisanal-scale gold mining activities (which use Hg as amalgamator) is suspected. Further studies on artisanal gold mining affecting the Yaqui River and the OV dam by Hg input are urgently needed and, if confirmed, regulations should be developed and implemented. Additionally, a human health risk assessment following exposure to Hg in water and fish consumption from the OV dam is strongly recommended. This study provides baseline data useful for further research on anthropogenic and climate impacts in reservoirs in the region. It highlights the importance of conducting specific risk assessments in lacustrine environments where fish production is relevant for human consumption and as a livelihood, and where water is used for public supply.