Emerging organic contaminants in springs of the highly karstified Dinaric region

Abstract

Emerging organic contaminants (EOCs) have become of increasing interest due to concerns about their impact on humans and the wider environment. Karst aquifers are globally widespread, providing critical water supplies and sustaining rivers and ecosystems, and are particularly susceptible to pollution. However, EOC distributions in karst remain quite poorly understood. This study looks at the occurrence of EOCs in the Croatian karst, which is an example of the "classical" karst, a highly developed type of karst that occurs throughout the Dinaric region of Europe. Samples were collected from 17 karst springs and one karst lake used for water supply in Croatia during two sampling campaigns. From a screen of 740 compounds, a total of 65 compounds were detected. EOC compounds from the pharmaceutical (n = 26) and agrochemical groups (n = 26) were the most frequently detected, while industrials and artificial sweeteners had the highest concentrations (range 8-440 ng/L). The number of detected compounds and the frequency of detection demonstrate the vulnerability of karst to EOC pollution. Concentrations of 5 compounds (acesulfame, sucralose, perfluorobutane sulfonate, emamectin B1b, and triphenyl phosphate) exceeded EU standards and occurred at concentrations that are likely to be harmful to ecosystems. Overall, most detections were at low concentrations (50 % <1 ng/L). This may be due to high dilution within the exceptionally large springs of the Classical karst, or due to relatively few pollution sources within the catchments. Nevertheless, EOC fluxes are considerable (10 to 10⁶ ng/s) due to the high discharge of the springs. Temporal differences were observed, but without a clear pattern, reflecting the highly variable nature of karst springs that occurs over both seasonal and short-term timescales. This research is one of a handful of regional EOC investigations in karst groundwater, and the first regional study in the Dinaric karst. It demonstrates the need for more frequent and extensive sampling of EOCs in karst to protect human health and the environment.

Keywords: Dinaric karst; Drinking water resources; Emerging organic contaminants; Groundwater; Karst aquifers.

Graphical abstract

Fig. 1

A) Extent of croatian karst and sampling locations, b) Jadro Spring, discharging 7.1 m³/s (photo: Josip Kolarić, 02.03.2020). *Due to the map’s scale, the Bistrac spring’s catchment and the Zagorska Mrežnica catchment share the same location point. Hydrogeological permeability background colours are from the Hydrogeological Map of Croatia, scale 1:300,000 (Biondić et al., 1999).

Fig. 2

EOCs in Croatian karst a) The 20 most frequently detected compounds and their maximum and median concentrations. Bars show frequency (%) of detection (primary y axis), circles and crosses show concentrations (secondary y axis); b) pie chart of the overall % of detections in each of the EOC groups; c) Box-Whisker plots showing the concentrations of the 20 substances with the highest maximum concentrations; numbers inside the boxes are the number of detections.

Fig. 3

Comparison of EOCs in Croatian karst with other studies: (a) maximum concentrations for different detected compounds ranked from highest to lowest values; (b) the number of detected compounds by different compounds groups; (c) the detection frequency for compounds that were detected in both Croatian karst groundwater and other studies. Data sources used for comparisons are Lukač Reberski et al. (2022) for karst aquifers (“Global karst groundwater” in legend) and Lapworth et al. (2012) for groundwater more generally (“All types groundwater” in legend).

Fig. 4

The total number of detections versus total concentration for individual sites.

Fig. 5

Spatial distribution of: (a) total concentrations and (b) the total number of detected compounds of different EOC groups at sampling locations. The size of the pie charts corresponds to: (a) the total concentration (the sum of the concentrations of all EOC compounds detected in both campaigns at the site), and (b) the total number of compounds detected at the site. Each pie chart presents grouped data from both campaigns, and colours correspond to the proportion of different EOC groups that contribute to the total concentration (a) or the total number of compounds (b). Figure c presents proportions of major land cover categories in the spring catchments (Input data source is the European Environment Agency & Copernicus LAND Service Corine Land Cover). Numbers 1–18 correspond to sampling locations. Due to the map’s scale, the Bistrac spring’s catchment and the Zagorska Mrežnica catchment share the same location point, therefore the circles are overlaping; the bigger circles correspond to Bistrac spring (18). Hydrogeological background is from the Hydrogeological Map of Croatia, scale 1:300,000 (Biondić et al., 1999) as in Fig. 1a.

Fig. 6

Long-term spring discharge (Q) statistics and discharge during the two sampling campaigns in March and October 2019. The upper graph shows daily precipitation (P) during the calendar year 2019 for each sampling location; red and green arrows show the timing of the sampling campaigns. Sampling locations not included in the figure: Golubinka spring – long-term data unavailable, water level in March was – 2 cm, and in October – 23 cm; Čikola spring – no discharges during both sampling campaigns when samples were taken from the cave; Vransko lake – water level in March was 10.73 m, and in October 10.32 m; Koreničko vrelo - discharge data are unavailable; Bistrac – sampled only in October, discharge data are unavailable. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 7

Comparison between the two sampling campaigns (March and October) at all sampling locations: a) total concentrations and number of compounds detected in different EOC groups (Agricultural, Pharmaceutical, PCP-LS, and Industrial); b) total concentrations, discharges and mass fluxes of all detected EOCs at individual sampling sites. Bistrac spring was only sampled in October. Discharge data at Golubinka spring were estimated: March – 1 L/s, October – 5 L/s.

Fig. 8

Correlations between EOC concentrations and land cover: a) correlation between natural land cover (%) and total EOC concentration (Ʃc EOC total), b) correlation between natural land cover (%) and total number of EOCs, c) correlation between agricultural land cover (%) and concentration of agricultural compounds, d) correlation between agricultural land cover (%) and number of agricultural compounds, e) correlation between agricultural land cover (%) and concentration of PCP-LS compounds and the number of pharmaceuticals, f) correlation between urban land cover (%) and concentration of PCP-LS and industrial compounds and the number of pharmaceuticals. Each graph presents Pearson’s r and p-value; α < 0.05. Total means sum of concentration or number of all detected compounds at each location.

Supplementary figure 1