An integrated interdisciplinary approach to evaluate potentially toxic element sources in a mountainous watershed

Abstract

Potentially toxic elements (PTEs, i.e., Cd, Ni, Cr) and their source apportionment in waters are of major environmental concern. Different approaches can be used to evaluate PTEs sources in environment, but single-way approaches are often limited and can easily fail. PTEs sources apportionment should include the evaluation of geochemical background and spatiotemporal trends analyses. We propose an integrated approach, and we apply it to a mountain catchment in the Italian central Alps, where ultramafic terranes crop out. We collected water and glacial sediment samples during the melting season. Then, we analyzed major ions and PTEs in waters, and we quantified the total PTEs load in sediments through acid digestion. Data were then processed through spatial and temporal trends analysis, clustering of variables and the evaluation of partition between the different compartments. We found a high geochemical background of part of the PTEs, consistently with results from other areas worldwide on mafic and ultramafic terranes (high concentrations of Ni, Cr and Fe), while we identified an additional atmospheric deposition source for Zn, Cd and Ag. Also, redundant observations on Cu, As and Pb indicated a possible mixed source. This study elucidates the need for an integrated approach to avoid unnecessary or misleading assumptions in the PTE’s source appointment. A single-way approach application, in fact, can fail in understanding element source in a complicated and dynamic compartment like surface water.

Appendix: Detailed methods

Detailed water analyses

Physicochemical parameters (pH, temperature and electrical conductivity) were evaluated in situ using specific field probes: pH and temperature were measured with a HANNA instruments HI 9025 pH meter; electrical conductivity was measured with a HANNA Instruments HI 9033 conductivity meter. Water samples for laboratory analyses were filtered through 0.45-µm filters and then collected in LPDE bottles.

Carbonates, as ${\text{HCO}}_{3}^{ - }$_, were estimated by colorimetric titration using 0,01 M HCl and Bromocresol Green as indicator. Other major anions (Cl⁻, ${\text{NO}}_{3}^{ - }$, ${\text{SO}}_{4}^{2 - }$) and cations (Ca²⁺, Mg²⁺, Na⁺, NH₄⁺, K⁺) were analyzed using an ionic chromatography Metrohm Eco IC (Swiss Confederation).

Samples for trace elements analyses, once collected, were acidified adding 2% volume ultrapure HNO₃ and analyzed using an iCAP-Q ICP-MS instrument from Thermo Fisher Scientific (USA). All samples were spiked with In as internal standard, and instrumental drift was beneath the 10% for all samples.

Limit of detection (LOD) for major ions, as referenced from the instrument, is 0.05 ppm. For trace elements, LOD was estimated as three times the standard deviation of blank samples (Long and Winefordner 1983).

Detailed sediment samples analysis

Once in laboratory, sediment samples were air-dried in oven at 105 °C for 3 h (Quevauviller 1998) and then < 2 mm fraction was sieved and selected for analysis (Chabukdhara and Nema 2012). Then, 500 mg of sample was inserted in Teflon vessels, and 3 ml of solution (pure hydrochloric and nitric acid solution in proportion 1: 2) was added (Filgueiras et al. 2002; Pueyo et al. 2008). The digestion was performed in MLS-1200 Mega, Milestone (USA) microwave digester. After cooling, the solutions were diluted with ultrapure water and then analyzed using a Thermo Fisher Scientific (USA) Icap-Q ICP-MS instrument. Samples were run in triplicate and present less than 5% of relative standard deviation.

Analysis solutions

All the solutions used in laboratory for this study were made using ultrapure water from a Millipore MilliQ system (18.8 MΩcm resistivity).

Acid solutions were obtained from a Carlo Erba® reagents (Italy) 65% volume solution and then were purified through sub-boiling distillation using a Milestone (USA) DuoPUR system.

Standard solutions for major ions and trace element analysis were obtained from dilution of MERCK (Germany) multi-elemental standards.

Detailed statistical methods

ANOVA

ANOVA test compares the averages and the variances of two different datasets following a categorical variable. The null hypothesis is that these datasets are the same, and the variance among samples is, therefore, the same as the difference between the datasets. F value is calculated as the ratio of variance inside the group and among the groups, and also the test p value is calculated (Ross and Willson 2017).

Cluster analysis

Ward’s method starts from a singleton (single-point clusters) and aims to create clusters with the lowest possible sum of squares increment. We decided to use this method because it creates small clusters (Ward 1963).

To avoid interferences in the cluster analysis due to different measure units of variables, all the measures were scaled and centered on average, using Eq. A1:

$$x_{i}^{\prime } = \frac{{x_{i} - \mu }}{s}$$

(A1)

where µ is the average value, s is the standard deviation, x_i is the original value and $x_{i}^{\prime }$, is the standardized value (Sahariah et al. 2015).

Statistical analysis was performed using R version 3 (R Core Team 2014) and the package “dendextend” to perform cluster analyses (Galili 2015).