Introduction

Karst aquifers constitute a reliable drinking water source worldwide (Stevanović 2019) and a strategic resource in mountainous regions. However, karst aquifers are highly vulnerable to contamination due to the absence of thick filtering soils and the presence of preferential flowpaths that favour fast groundwater movement due to circulation through a—very often—well-developed conduit network (Goldscheider and Drew 2007). Highly karstified carbonate aquifers are well known for the rapid response to intense rain events which result in quick variations of hydrochemical parameters and, in some cases, high turbidity peaks and elevated concentrations of suspended particles and colloids in karst springs (Pronk et al. 2007; Goldscheider 2005). The arrival of suspended sediments (which originate turbidity) and associated contamination (e.g., trace elements) also presents sanitary implications and shows health risks caused by disease transmission or heavy metal accumulation in human bodies (U.S. EPA 2018).

Studies about the presence of mobile sediments in karst aquifers have been developed throughout the last decades; and among these, three classes are traditionally differentiated (Grolimund et al. 2007): (i) colloidal particles, precipitates of inorganic pollutants; (ii) carrier colloidal particles, which bind the pollutants at their surfaces) and, (iii) biocolloids, such as bacteria or viruses. Fine grain sediments in karst systems (topics i and ii) may present two different sources: autochthonous, sediments derived from carbonate matrix dissolution, and allochthonous, sediments washed by runoff waters and introduced through swallow holes or sinking streams into the system during flooding events (Mahler et al. 1999; Feeser and O’Connell 2009; Yang et al. 2005; Stanienda 2016). Thus, considering that non-anthropogenic contamination exists in the soils’ composition represented in Fig. 9, it is possible to assume that the analyzed components in ES and AS are naturally present in such range of concentrations.

Surface water and groundwater samples

The mean concentration of the analyzed trace elements shows clear differences depending on the sampling point, both in surface water and groundwater. The geological formations crop** out in the endorheic areas of Ubrique test site are similar to that of the studies of Ruiz Cruz (1994, 1999) and present clay minerals in which chemical structures, mainly Al, Ni, and Mn—among other trace elements—can be found. Furthermore, these elements display a high correlation (p > 0.9) in P1 (Fig. 7) between them, which suggests a common origin in runoff samples without the influence of WWTP leakages. Given that mean Al, Ni, and Mn concentrations in runoff samples collected in Albarrán stream are quite similar in P1 and P2 sampling points and both show higher values compared to mean concentration in WWTP effluent (Table 3), the natural origin of such elements is determined.

Al, Mn, Ni, and turbidity display higher mean values in groundwater samples from Algarrobal spring compared to runoff and Cornicabra water samples (Table 2; Fig. 4). A similar correlation as the one described at P1 is found between these three elements at karst springs, especially evident in the Cornicabra spring (0.7–0.9) with respect to Algarrobal (0.4–0.7, Fig. 7). Karst studies dealing with diffuse recharge systems, with low sediment transfer during flooding events (Jebreen et al. 2018), reported lower Al, Mn, and Ni mean concentrations than in those affected by surface runoff leakages (Ma et al. 2011; Xu et al. 4). In addition, both trace elements show an apparent correlation with Cl− in surface samples, typically related to human activities in absence of evaporite rocks. P and Ba have been commonly associated with anthropic influence from urban sewage systems and industrial emissions. Other hydrogeological studies (Neal et al. 2006; Krishna et al. 2009) dealing with the origin and fate of trace elements in industrialized areas highlighted the potential anthropogenic source of Ba with mean concentrations ranging from 60 to 600 μg/L. The contribution of domestic wastewaters is also known to increase P concentrations (as PO43−) over natural inputs (Hardwick 1995; Liao et al. 2020). In this study, mean P concentration is one order of magnitude higher in effluent samples than in runoff waters and a significant increase is observed between P1 and P2 sampling points (Table 3), which suggests a great contribution of P coming from the WWTP. Nonetheless, mean concentration of P in waste waters (Table 3) is included within the normal range measured in municipal wastewaters (4–16 mg/L) (Metcalf and Eddy Inc. 2003).

In groundwater samples, mean concentrations of Li, Sr, and Ba (Table 3) are comparable to those found in other non-polluted karst systems, with characteristic values from carbonate rock weathering (mean concentrations between 1 and 2 μg/L of Li, 150 and 200 μg/L of Sr, and 11 and 40 μg/L of Ba; Kilchmann et al. 2004; Jebreen et al. 2018). However, the rise of Sr and Ba concentrations in Algarrobal spring together with turbidity increases (Fig. 6) is much evident than in Cornicabra (Fig. 5) and suggests a higher influence of allogenic recharge. In this research, the higher mean concentration and variability of such elements in Albarrán catchment soil (AS, Fig. 3) and runoff waters (Fig. 4), rather than ES and groundwater, may indicate that the availability of such elements is enhanced in silicate-clayey lithologies or dilution processes occur within the system (or combination of both).

The time-series of elemental P in karst groundwater did not show a clear pattern in the successive events (Figs. 5, 6) but its presence is significantly higher and more defined in Algarrobal spring (Table 3), which shows maximum appreciable peaks in Fig. 6. Despite that P is naturally present in diverse ecosystems and karst environments (Imbach 1993; Markovic et al. 2019) as PO43− with concentrations that may range between 50 and 900 µg/L for karstic springs, such ion was not detected through ion chromatography in this research. P might then be considered as a minor component in the karst springs due to its low concentration, but a major element in runoff (P1 and P2) samples in the WWTP effluent. Hence, the origin of P is probably related to human activities in Ubrique system.

Factors conditioning mobility of trace elements

The statistical analysis developed in “Statistical relationships between trace elements and transport vector” section revealed two major trends among trace elements and transport carriers (turbidity vs major ions). Some elements (Al, Mn, Ni, and Ba) presented in surface and groundwater a good correlation with turbidity and its chemical variability explained, to a major extent, by this physical parameter. The rest of the elements analyzed (Li, Sr, and P) present a higher affinity with the major ions and PCA grou** with electrical conductivity.

Transport related to suspended sediments

The statistical analysis (PCA) carried out in this study shows an association between Al, Mn, and Ni with turbidity (Fig. 8), and a Spearman correlation coefficient of p > 0.68 (Fig. 7) in both surface and groundwater. Different physico-chemical variables might influence sediment transport in groundwater, as the number of large particles and colloids appears to be determined by discharge, pH, and temperature (Atteia and Kozel 1997; Herman et al. 2012). However, in this research, the increases in the hydraulic head of the system at the beginning of each flood are responsible for the greatest turbidity variability.

Acid rain might favour chemical weathering processes such as hydrolysis that dissolves ions from the clay mineral and enhances trace elements (Al, Mn and Ni) mobility (Sposito 1996). Speciation of these trace elements in water is characterized by the combination of hydroxyl groups with Al, Mn2+ and Ni2+ (Stumm and Morgan 1981; Hem 1985; Perkins and Mason 2015). The subsequent pH and temperature conditions or surface chemical bond group will condition the occurrence of surface complexation or electrostatic attraction processes of metals onto solids. These processes constitute common phenomena in natural systems (Hart 1982; Jenne and Zachara 1987; Horowitz 1991; Cholet et al. 2019) and its enhanced mobility by sediment transport through karst aquifers may account for high concentrations at the springs. As well as previously suggested by Vesper and White (2003), Al is proposed in this research as a sediment proxy given its analogue transport with suspended sediments and other trace elements, such as Mn and Ni, because of its high correlation (R2 > 0.94) with turbidity.

Moreover, Ba also shows a high correlation with turbidity in surface sampling points and Algarrobal spring (Fig. 7). Despite that this element presents an anthropogenic origin, sorption processes onto solids have also been widely described (Eylem et al. 1990; Atun and Bascetin 2004). This would also explain the enhanced Ba concentration in Algarrobal spring due to the high turbidity records. Thus, the direct implication of suspended sediments (represented as turbidity) on Al, Mn, Ni, and Ba transport is clearly observed in surface and groundwater samples in Ubrique test site.

Transport related to dissolved ions

Despite that the processes that control Sr, P, and Li transport in runoff and groundwater must be the same, slight differences are observed. An apparent correlation between Li and Sr and most representative major ions (HCO3− and SO42−) is found (Figs. 5, 6) and emphasized in runoff samples (Fig. 7). Such elements might be present in aqueous solution as hydrated mono- or di-valent cations, without the need for aqueous complexation (Hanor 2000).

The dominant factors controlling the mobility of Li and Sr in surface appears to be solute transport mainly associated with HCO3−. A similar process is observed with P, which, despite of its complex chemical behavior, the mobility of this trace element appears to be dominated by solute migration (Fig. 8). The concentration of these elements is much lower in groundwater compared to surface water (Fig. 4), which indicates that dilution processes might occur due to the arrival of recently infiltrated rainwater through diffuse recharge and posterior mixing with waters from the saturated zone.

Both springs show an abrupt depletion in the concentration of SO42− of the first event in comparison to the following ones. In addition, a decreasing trend is also observed in the concentration of some elements such as Sr and Cl− throughout the four events analyzed (Figs. 5, 6). This circumstance can be directly related to the saturated zone thickness prior to the event and the recharge conditions. Hence, spring response defining Event 1 is especially particular due to the absence of rainfall in the previous 6 months, resulting in a greater residence time of the groundwater drained at the beginning of the event in both springs and, therefore, a greater concentration of dissolved ions such as SO42−, Cl− or Mg2+.

Assessment of relative contribution of allogenic recharge to spring flow

The different minerals and available elements found in catchment soils (Fig. 3) as well as the chemical signature of aquifer rocks might be expressed as specific ratios in water samples. Previous studies (Land et al. 1998, 2000) used specific molar ratios, as Ca/Sr, as effective tracers for analyzing aquifer–river interactions assuming differential concentrations of mineral phases. The authors successfully identified the influence of surface recharge flows in a shallow porous aquifer with mainly silicate and clay minerals using Ca/Sr jointly to Ba/Sr molar ratio. Hogan and Blum (2003) coupled this technique with water isotopes in a small (1.2 km2) carbonate–silicate river catchment and found a decrease of Ca/Sr ratio during storm events.

In this research, this approach, coupled to Ba content, is used to assess the relative contribution of allogenic recharge in terms of groundwater chemical variability. To validate the application of this method, Al concentration (interpreted as sediment proxy) is represented as symbol size. Hence, Fig. 10 shows, on the one hand, the content in Ba is notably higher (> 0.3 meq/L) in surface water samples (followed by WWTP Effluent, Table 3) rather than in groundwater samples. On the other hand, Ca/Sr ratio shows a characteristic value (between 20 and 23) in groundwater samples collected during low water conditions, while runoff samples present specific molar ratio values of Ca/Sr < 7.

Fig. 10
figure 10

Relationship between Ba (meq/L) and Ca/Sr molar ratio in runoff and groundwater samples. Al concentration (µg/L) is represented as a function of symbol size

Full size image

Groundwater samples obtained during flooding events are distributed within the Ca/Sr molar ratio range between 7 and 20 show a slightly increasing trend of Ba. Moreover, Al content apparently increases towards lower Ca/Sr ratios, which is especially notable in Algarrobal spring. Given the high correlation between Al and turbidity, it is possible to assume that allogenic recharge constitutes the main contribution of suspended sediment into the karst system. However, this spring presents much higher turbidity and Al content compared to runoff samples (Fig. 10; Table 2) which indicate that sediment storage might exist inside the system conduits and is posteriorly mobilized to the spring. This process is evidenced in the extremely high turbidity record in Event 1, which shows the most abrupt response in both springs, as reflected by hydrochemistry. The low hydrodynamic stage of the system in the first event of the hydrological year might then be determinant in the accumulation and deposition of sediments within the conduits network.

As a general trend, groundwater samples form Algarrobal spring show a higher similarity with runoff Ca/Sr molar ratios (Fig. 10) as well as higher content in the analyzed trace elements compared to Cornicabra spring. These results suggest a notably higher influence of allogenic recharge in terms of sediment input and chemical variability derived from allogenic recharge in Algarrobal spring. Despite that, in this preliminary research, it is not possible to estimate a quantitative contribution, these results are coherent with the results derived from the dye experiment previously realized in this test site (Martín-Rodríguez et al. 2023). These findings have direct implications in drinking water capture at the springs, given that the leakages from the WWTP are potentially connected to the main outlets.

Conclusions

The developed multi-criteria approach combining physical (spring discharge, turbidity, temperature, and electrical conductivity) and chemical (major ions and trace elements) data has been successfully tested to investigate recharge dynamics and vulnerability to contamination in a binary karst aquifer. The proposed framework is based on two main elements: a complete monitoring network (including climate, soil, and natural waters from permanent springs used for water supply and wastewater) and a high-periodicity water sampling strategy.

In this research, the analyzed trace elements in soil samples were measured within the normal range of clayey soils in the absence of anthropogenic influence and chemical anomalies. Chemical weathering of clayey and carbonate materials found in the main allogenic recharge catchment is responsible for the presence of Al, Mn, Ni, Li, and Sr, in water samples. In contrast, the source of Ba and P was determined to present a major anthropogenic contribution. The transport of sediment and some trace elements (Al and Mn and to a minor extent, Ni and Ba) is intrinsically related due to the chemical nature of colloids. Hence, Al has been proposed in this test site as the optimal sediment proxy given its high correlation with turbidity in groundwater samples.

The analysis of long-term time-series is essential in karst aquifers to provide a better comprehension of the data variability. In this study, the hydrodynamic conditions of the aquifer prior to individual rain events and their magnitude proved to present direct implications in trace elements transport and sediment storage and mobilization. Hence, vulnerability is enhanced after long dry periods due to the accumulation of sediments and related trace elements within the karst system. The analysis of karst spring response through time-series and the determination of specific Ca/Sr molar ratios together with Ba content revealed a higher relative contribution of allogenic recharge to Algarrobal spring compared to Cornicabra in terms of sediment transport and hydrochemical variability.

The findings achieved in this study evidence the potential benefits of combining high-periodicity monitoring and complementary laboratory analysis for a correct vulnerability assessment in water systems with a high temporal variability of chemical parameters. The methodological framework applied in this research provides a comprehensive knowledge of recharge mechanisms of the binary karst system, which is highly important for a proper management of groundwater resources. This site-specific approach is easily transferrable to different karst water systems captured for human consumption.