COBISS: 1.01 A MULTIPARAMETER ANALYSIS OF ENVIRONMENTAL GRADIENTS RELATED TO HYDROLOGICAL CONDITIONS IN A BINARY KARST SYSTEM (UNDERGROUND COURSE OF THE PIVKA RIVER, SLOVENIA) MULTIPARAMETRSKA ANALIZA OKOLJSKIH GRADIENTOV, POVEZANIH S HIDROLOŠKIMI RAZMERAMI V BINARNEM KRAŠKEM SISTEMU (PODZEMNI TOK REKE PIVKE, SLOVENIJA) Janez MULEC1,2, *, Metka PETRIČ1,2, Alenka KOŽELJ3, Clarissa BRUN3, Erika BATAGELJ3, Aleš HLADNIK4 & Ladislav HOLKO5 Abstract UDC 551.444:556.114(497.471) Janez Mulec, Metka Petrič, Alenka Koželj, Clarissa Brun, Eri­ka Batagelj, Aleš Hladnik & Ladislav Holko: A multiparam­eter analysis of environmental gradients related to hydrologi­cal conditions in a binary karst system (underground course of the Pivka River, Slovenia) Chemical and bacterial gradients under different hydrologi­cal conditions were studied in a well-developed underground karst system. Water samples were collected from the main un­derground drainage conduit of the Pivka River from October 2013 until June 2016. The system responds quickly to external pulses (precipitation events), and is also impacted by human interventions, as is demonstrated mainly by fluctuations of sul­phates, chlorides, and occasionally elevated concentrations of organic and faecal pollutants. Chemical and bacterial param­eters showed a monotonous trend of decreasing concentrations from the ponor towards the interior of the karst massif during stable hydrological conditions, and a significant change dur­ing high water conditions. High flow events tend to equilibrate chemical and bacterial parameters in the underground river. Concentrations of chlorides, TOC (total organic carbon) and nitrates were the most indicative parameters describing the for­mation of the gradient. Stable isotopes of hydrogen and oxygen in water indicated that the main karst conduit collects isotopi­cally different waters from the aquifer. The river water collected after nine kilometres of underground flow was always isotopi­cally lighter than the waters collected from the upstream sites. Multiparameter analysis proved to be a useful tool for provid­ing a more comprehensive understanding of the dynamics of the underground water, which influence both the underground environment and the ecology of the biome. Key words: karst, hydrology, water chemistry, nutrients, stable isotopes, PCA, bacteria. Izvleček UDK 551.444:556.114(497.471) Janez Mulec, Metka Petrič, Alenka Koželj, Clarissa Brun, Erika Batagelj, Aleš Hladnik & Ladislav Holko: Multiparametrska analiza okoljskih gradientov, povezanih s hidrološkimi razmerami v binarnem kraškem sistemu (podzemni tok reke Pivke, Slovenija) V dobro razvitem podzemnem kraškem sistemu smo pri raz­ličnih hidroloških pogojih preučevali kemijske in bakterijske gradiente. Vzorce vode smo odvzeli iz glavnega podzemnega toka reke Pivke med oktobrom 2013 in junijem 2016. Sistem se hitro odziva na zunanje impulze (padavinski dogodki) in je tudi podvržen človekovim posegom, kar dokazujejo predvsem nihanja v koncentraciji sulfatov in kloridov ter občasno poviša­ne koncentracije organskih in fekalnih onesnaževal. Spremljan­je kemijskih in bakterijskih parametrov v stabilnih hidroloških razmerah je pokazalo monotoni trend zniževanja koncentracij od ponora proti notranjosti kraškega masiva. Razmere se izra­zito spremenijo v času visokih vod, ko pride v podzemnem vo­dotoku do izenačenja tako kemijskih kot bakterijskih paramet­rov. Kloridi, TOC (skupni organski ogljik) in nitrati so bili naj­bolj indikativni parametri za opis nastanka gradienta. Stabilni izotopi vodika in kisika v vodi so pokazali, da vodotok glavnega kraškega kanala zbira izotopsko različne vode iz vodonosnika. Voda podzemne reke po devetih kilometrih toka v podzemlju je bila vedno izotopsko lažja kot vode iz gorvodno vzorčevanih mest. Multiparametrska analiza se je izkazala kot uporabno orodje za celovitejše razumevanje dinamike podzemnih voda, ki vpliva tako na podzemno okolje kot ekologijo bioma. Ključne besede: kras, hidrologija, kemija vode, hranila, stabil­ni izotopi, PCA, bakterije. 1 Research Centre of the Slovenian Academy of Sciences and Arts, Karst Research Institute, Titov trg 2, SI-6230 Postojna, Slovenia. University of Nova Gorica, UNESCO Chair on Karst Education, Glavni trg 8, SI-5271 Vipava, Slovenia 2 UNESCO Chair on Karst Education, University of Nova Gorica, Glavni trg 8, SI-5271 Vipava, Slovenia. 3 National Laboratory of Health, Environment and Food, Verdijeva 11, 6000 Koper, Slovenia. 4 Faculty of Natural Sciences and Engineering, University of Ljubljana, Snežniška ulica 5, SI-1000 Ljubljana, Slovenia. 5 Slovak Academy of Sciences, Institute of Hydrology, Ondrašovská 16, 03105 Liptovský Mikuláš, Slovakia. * Corresponding author: janez.mulec@zrc-sazu.si Received/Prejeto: 30.11.2018 DOI: https://doi.org/10.3986/ac.v48i3.7145 ACTA CARSOLOGICA 48/3, 313-327, POSTOJNA 2019 JANEZ MULEC, METKA PETRIČ, ALENKA KOŽELJ, CLARISSA BRUN, ERIKA BATAGELJ, ALEŠ HLADNIK & LADISLAV HOLKO INTRODUCTION For management of water resources for human use and for the well-being of dependent aquatic ecosystems the understanding of groundwater and surface water interac­tions at all scales is necessary. This is especially important for binary karst systems that are fed by both autogenic (diffuse in.ltration into a karst aquifer) and allogenic recharge (sinking water from surrounding non-karst ar­eas). Streams coming from non-karst watersheds com­monly show large discharge fluctuations, have low min­eral content and carry dissolved or particulate organic matter, especially during flood events (Bailly-Comte et al. 2009). At the contact with karst or a short distance beyond it, they sink into the ground, travel through con­duits in the aquifer and eventually discharge downstream through caves and springs. As a result, groundwater and surface water constitute a single hydrodynamic system (Katz 2002; Koit et al. 2017). The defining characteris­tics of karst aquifers are rapid throughput times, localiza­tion of flow along essentially one-dimensional flow paths within the conduit system, and the presence of deposits of clastic sediments in many of the conduits. Peak flows in karst systems may be up to 100 times the low flows, and flow velocities are also far higher during high flows. The ability of the conduit system to store and release con­taminants is dependent both upon the nature of the con­taminants and upon the storm flow characteristics of the system (Vesper et al. 2001). Karst waters are significant sources of water supply in many regions of the world. In the context of a global shortage of good quality water, they are particularly vul­nerable to eutrophication resulting either from natural processes, e.g., toxic algal blooms, human intervention, such as intense agriculture production, or water pollu­tion from industry and urbanization (Jianhua et al. 2016; Long et al. 2012; Shi et al. 2009; Vallejos et al. 2015; Wang et al. 2018). Impact of pollution on underground ecosystems is particularly detrimental in well-karstified aquifers, because they generally react rapidly to changes in hydrological conditions (Ender et al. 2018; Hartmann et al. 2014). Once in the light-deprived underground environment, different materials, particulate matter and partly and/or completely dissolved compounds of natural and anthropogenic origin form concentration gradients towards the interior of the karst massif. These gradients are useful in providing a better understand­ing of the processes affecting the hydrological cycle in the karst. Formation of a gradient in the underground is generally challenged by dilution and/or concentra­tion of compounds, and details of abiotic and/or biotic conversions of the material should be further explored. Many aquatic organisms adapted to such environments rely upon nutrients that are either introduced from the surface or result from in situ chemolithoautotrophy-based primary production (Hutchins et al. 2016). Con­tamination of groundwater with pathogenic organisms is associated with an introduction of faecal material of human and animal origin (Heinz et al. 2009). Notewor­thy contributors to potentially hazardous faecal pollution are point sources such as failed septic systems, leaking sewer lines and cesspools, animal-fattening areas, dairy farms and other intensive animal-husbandry operations (Macler & Merkle 2000). Water quality in karst systems can be monitored using several physical, chemical and microbial parameters. Bacterial indicator groups that re­spond to organic pollution represent an additional esti­mator of the health of underground systems (Pronk et al. 2006). Monitoring the fate of natural tracers, e.g., stable isotopes that are directly related to the water cycle, and compounds subjected to chemical and biochemical con­versions can reveal the complexity of underground water quality, particularly in cases with an evident impact from eutrophication. The objective of the present study was to analyse and correlate the geochemical and bacterial gradients in a well-developed underground karst system during dif­ferent hydrological conditions. Such a study of transport properties that demonstrate well-defined flow character­istics within the system can add important information to the results obtained at a broader scale, when only the input and output points are known, with limited data on water chemistry and artificial tracers (Gabrovšek et al. 2010). Study was undertaken in a binary karst system comprising 9 km-long underground course of the Piv­ka River, which is occasionally affected by the pollution events in the catchment area. The watercourse within the system is readily accessible at various locations by way of natural cave entrances, and the cave system harbours a diverse stygobitic (groundwater-adapted) fauna (Pipan & Culver 2007). Physical, chemical and bacterial param­eters were used simultaneously to assess the water quality during different hydrological (flow) conditions relative to the input values in a complex karst aquifer, and to relate dilution/concentration effects and (bio)chemical conver­sions. Along with the multiparameter approach, moni­toring was carried out at several points along the under­ground watercourse, adding additional information to the studies where only the input and output environmen­tal values of the karst system were considered. MATERIAL AND METHODS SITE DESCRIPTION The Pivka Basin with the Pivka River (southwestern Slo­venia) is slightly to moderately impacted by agriculture, industry and urbanization. In its upper part the Pivka River is recharged from a karst aquifer within the Ja­vorniki Mountains, which are composed mostly of Cre­taceous limestone (Buser et al. 1967). In the lower part of the river course, limestone is overlain by low-perme­ability Eocene flysch deposits, upon which the surface drainage networks of the Pivka River and its tributary the Nanoščica River have developed (Fig. 1A). Discharges of the Pivka River range from 0.001 to 66 m3/s, and the mean discharge is 5.3 m3/s (Archive hydrological data as­sessed 26 January 2015, http://vode.arso.gov.si/hidarhiv/pov_arhiv_tab.php). The Pivka River provides the main allogenic recharge to the famous Postojna Cave system (Postojnska jama), which comprises several caves: from Postojna Cave at the ponor site (E 14.2037° N 45.7827°, 529 m a.s.l.) to the Pivka Cave (Pivka jama), which pro­vides the ultimate access to the underground Pivka River course within this system (Fig. 1A). Some of the system’s dry passages are open for tourist visits, attracting approx­imately 700,000 visitors per year. The known underground course of the Pivka River is a continuous and almost completely accessible water passage with different channel geometries. Upstream from the Pivka Cave, the flow is obstructed by several collapses that cause significant ponding of water, giving rise to flow cascades along a series of lakes impounded upstream of each collapse zone. Evidently braided chan­nels are present and which flow paths are followed by the river flow depends upon the flow rates. The channel slope in this part of the system is about 0.01, i.e., about twice the slope in the first part of the cave system (Gabrovšek et al. 2010). Underground flow continues through unex­plored channels towards Planina Cave (Planinska jama) at the southern edge of the Planina polje (Planinsko polje). Here the Pivka cave stream flows along the west­ern branch of the cave and converges with the Rak cave stream, which occupies the eastern branch of the cave. The latter passage carries most of the water from the Rak River, which sinks into Tkalca Cave (Tkalca jama). Downstream from the underground confluence, the water flows out of Planina Cave as the Unica spring (E 14.2457° N 45.8199°, 453 m a.s.l.) (Fig. 1A). A subsurface connection between the Pivka River in Postojna Cave and the Pivka cave stream in Planina Cave has been con­firmed by several tracer tests. Tracers injected into the Ja­vorniki karst plateau were also detected in the Pivka cave stream, which confirms that there is additional autogenic recharge from the karst aquifer (Gabrovšek et al. 2010; Kogovšek & Petrič 2004; Ravbar et al. 2012). SAMPLING The underground Pivka River was sampled at four sites: (Site 1) immediately downstream of the ponor in Posto­jna Cave (Veliki Dom, 0.1 km inside the cave), (Site 2) 0.9 km from the ponor (Spodnji Tartar), (Site 3) at the si­phon in Pivka Cave (4.1 km from the ponor), and (Site 4) just upstream of the confluence with the Rak cave stream in Planina Cave, at a distance of 9.0 km from the ponor (Figs. 1A and 1B). Hydrological conditions are the most important fac­tor that determine the underground water flow and water chemistry. Seven sampling campaigns were carried out during the period from October 2013 to June 2016 under different hydrological conditions (27 June 2013, 21 Octo­ber 2013, 14 July 2014, 25 November 2014, 1 September 2015, 15 December 2015, and 28 June 2016). Because of the extensive nature of the cave system, sampling was time consuming and lasted for up to 5 hours (between the first and the last sampling sites). This time difference should be taken into account when interpreting the data collected during changeable hydrological conditions. On-site measurements of water temperature (T), electrical conductivity (EC), pH and dissolved oxygen (DO) were performed using a WTW Multiline 3420 por­table meter (WTW, Germany) during collection of wa­ter samples for chemical and microbiological analyses. Water samples were subdivided for selected analyses in the caves at each individual site: water from two 1.5 li­tre plastic bottles was analysed for true colour, turbidity, ammonium, chloride, fluoride, nitrite, nitrate, sulphate, orthophosphate and total phosphorus. Water from two 1 litre glass bottles was analysed for anionic surfactants, and water from 250 millilitre glass bottles without air was analysed for TOC. A 20 ml glass bottle was filled directly from the source for analysis of isotopes of oxygen and hy­drogen. For microbiological analyses, water samples were collected aseptically in 0.5 litre sterile plastic bottles. All samples were transferred to the laboratory in a cool box. High-frequency monitoring of physico-chemical parameters contributes to better understanding of the hydrogeological functioning of karst systems (e.g., Pronk et al. 2006; Tissier et al. 2013). In the periods from Au­gust 2013 to August 2014 and from November 2015 to August 2016, T and EC were measured every 30-minutes by Onset HOBO Conductivity data loggers at the ponor site (150 m upstream from Site 1) and in the Pivka cave stream in Planina Cave (Site 4). Precipitation data collected at the Postojna me­teorological station (E 14.1928° N 45.7661°, 533 m a.s.l) were obtained from the Slovenian Environment Agency. Discharges of the Pivka River at the ponor in Postojna Cave (E 14.2037° N 45.7827°, 529 m a.s.l.) were defined as sums of the discharges of the Nanoščica and Pivka riv­ers, measured by the Slovenian Environment Agency at the two hydrological stations (E 14.1812° N 45.7800°, 517 m a.s.l and E 14.1861° N 45.7299°, 529 m a.s.l, respec­tively) upstream of their confluence. The confluence is 2.1 km upstream the ponor in Postojna Cave. ANALYSES OF PHYSICAL AND CHEMICAL PARAMETERS Laboratory analyses were performed at the National Lab­oratory of Health, Environment and Food, Koper, and Karst Research Institute ZRC SAZU, Postojna, accord­ing to the following methods: true colour (SIST EN ISO 7887-2012), turbidity (SIST EN ISO 7027:2000), ammo­nium (ISO 7150-1:1984), nitrite (SIST EN 26777:1996), nitrate (HM075-HPLC), anionic surfactants (SIST ISO 7875-1:2004), total organic carbon – TOC (ISO 8245:1999), chloride (ISO 10304-1- 2007), sulphate (ISO 10304-1- 2007), fluoride (ISO 10304-1- 2007, HM052/UV-VIS), orthophosphates (SIST EN ISO 6878:2004), total phosphorus (SIST ISO 6878-7:2004). Water samples were analysed for their stable iso­topes of oxygen and hydrogen using a Thermo Delta Plus isotope-ratio mass spectrometer (Thermo Fischer Scientific). The samples were placed in 10 ml screw-top vials, and for D/H, a platinum catalyst was added. The vials were sealed with septa and all air was removed from the sample vials by an automated, autosampler-assisted flushing procedure that uses a mixture of either H2 or CO2 in He. After the required equilibration time (D: 40 min, 18O: 20 h), the whole batch of samples was analysed. The results were elaborated by a five points calibration curve linear regression algorithm. For the upper and lower points of the calibration curve the following ref­erence materials were used: VSMOW (Vienna Standard Mean Ocean Water, .18O 0 ‰; .2H 0 ‰); SLAP (Stan­dard Light Antarctic Precipitation, .18O -55.5 ‰; .2H -428 ‰); for the intermediate points a mix of the two was made, while the control for the accuracy of the cali­bration was the GISP (Greenland Ice Sheet Precipitation, .18O -24.78 ‰; .2H -189.7 ‰). The data were expressed in the conventional . notation in per mille with respect to Vienna Standard Mean Ocean Water. Analytical accu­racy was ±0.1‰ for .18O and ±1‰ for .2H. Evolution of the isotopic composition of water along the river course was used to identify possible contribution of different waters to the river. BACTERIAL INDICATOR GROUPS Several indicator bacterial groups were taken to trace the external impact in the cave system. The total con­centration of heterotrophic aerobic bacteria, coliforms, Escherichia coli and enterobacteria were estimated us­ing RIDA®COUNT plates (Mulec et al. 2012). The com­mercially available RIDA®COUNT test plates contain standard nutrients and a specific chromogenic detection system to detect cultivable microorganisms or a selected group. The specific microbial enzymes will change the originally colourless substrate to a distinctively coloured colony (Morita et al. 2003). Plates were inoculated with one millilitre of the water sample and cultivated for 48 hours at 37°C, and grown colonies were expressed as Colony-Forming Units (CFU) per ml. Enterococci were determined following the ISO 7899-2:2000 membrane filtration method. Bacterial analyses were performed at the National laboratory of health, environment and food, Koper, and Karst research institute ZRC SAZU, Postojna. For statistical analyses the following bacterial in­dicator groups were used: total bacterial counts (BAC), concentration of enterococci (ENCOC), E. coli (ECO), coliforms (COL), non-coliform bacteria (NCOBA – rep­resented bacterial group that excludes coliform bacteria), non-enteric bacteria (NENBA – represented bacterial group which excludes enteric bacteria), non-E. coli coli­forms (NECCO – calculated as the number of E. coli col­onies subtracted from the total coliform counts), non-E.coli enterobacteria (NECEN – calculated as the number of E. coli colonies subtracted from the total enterobacte­rial counts (Oarga et al. 2012). PRINCIPAL COMPONENT ANALYSIS Results of bacterial and physico-chemical analyses were interpreted by means of a multivariate statistical tool, Principal Component Analysis (PCA). PCA is widely used for data exploration and/or reduction purposes: by plotting data in the coordinate system of two PCs one can visualize trends and patterns present in the sam­ples. These trends are deduced from the scores diagram. Similarities and differences among the original variables are investigated by means of the loadings plot (Golež & Hladnik 2013; Hladnik & Muck 2002). Mathematically, PCA can be performed either by eigenvalue decompo­sition of a data covariance (or correlation) matrix or by singular value decomposition of a data matrix, usually af­ter mean centring the data matrix for each original vari­able (Abdi & Williams 2010). PCA was performed using singular value decomposition of the data matrix after standardization - subtraction of the mean and division by the standard deviation of each attribute (parameter). RESULTS AND DISCUSSION HYDROLOGICAL CONDITIONS Sampling campaigns to investigate water chemistry and bacterial load were carried out under various hydrologi­cal conditions. On 27 June 2013, 1 September 2015, and 15 December 2015, during a period of low flows, the hydrological conditions were stable, which means that during the sampling campaign no significant changes in the discharge and physical parameters were detected at the sampling points along the underground watercourse. Hydrological conditions were more unstable during the sampling on 25 November 2014, in high flow conditions at the end of the strongest flood pulse of the monitored period. Three samplings were carried out during flood pulses following lesser precipitation events: on 21 Octo­ber 2013 during an increase of discharge, and on 14 July 2014 and 28 June 2016 during recessions (Fig. 2). Time series of measured EC values at sampling sites 1 and 4 were compared to provide a more detailed as­sessment of hydrological characteristics at the time of sampling. Fig. 3 presents an example from the sampling campaign in October 2013. At Site 1, the sample was col­lected 13 hours after the peak of the precipitation event, at the time when the EC decreased at the ponor indicat­ing the beginning of water-quality changes induced by the event. The sample at Site 4 was taken 4 hours earlier, when stable EC values at this site were indicative of the pre-event conditions. Thus the two samples are not di­rectly comparable. The October 2013 sampling campaign represents the most extreme difference in flow conditions during the sampling period. Interpretation of chemical and microbiological data during unstable hydrologi­cal conditions must be based upon an understanding of what the individual samples represent. STABLE ISOTOPES Stable isotopes have long been used to elucidate the sources of water (Clark & Fritz 1997; Kendall & Cald­well 1998). Isotopic compositions of the samples at Sites 1 and 2 were quite similar, i.e., the differences were close to the limits of analytical accuracy (Fig. 4). This indicates the same source of water at both sampling sites. Water at Site 3 was lighter isotopically than at Sites 1 and 2, but its isotopic composition in July and November 2014 differed significantly from those at Sites 1 and 2. The difference in isotopic composition of river waters between Sites 1, 2 and 3 indicated contributions of isotopically-lighter wa­ter along the river course. A reverse pattern was observed in July 2014, when the water at Site 3 was isotopically very heavy compared to samples from all other sites. The July 2014 sampling was conducted during flow recession fol­lowing a minor rainfall–runoff event. It is therefore pos­sible that other flow paths delivering isotopically-heavier water were activated. Water at Site 4 is noticeably depleted in heavy iso­topes compared to the upstream sites. Samples collected in September 2015 are plotted below the Global Mete­oric Water Line, which indicates evaporated water (Fig. 5). However, the pattern was similar to that observed on other sampling days, i.e., isotopically-similar waters at Sites 1 and 2, different (usually lighter) water at Site 3 and noticeably lighter water at Site 4. The analyses showed that along the course of the Pivka River water becomes isotopically lighter, which is the opposite of the gradient known from surface streams in non-karst areas (Holko et al. 2015). Change in the iso­topic composition of the river waters occurs mainly be­tween Sites 3 and 4. The isotopic gradient existing in the river between the ponor and Site 4 indicates an increas­ing contribution of water percolating through the rocks of the massif. A very clear gradient was also observed during the flood conditions on 25 November 2014, when the isotopic composition of water at Site 3 was signifi­cantly lighter than at Sites 1 and 2. This indicates the mo­bilization of flow paths that are less active between the two river sections during low flow conditions. This is in accordance with the results of another study, which showed activation of additional flow paths during high flow conditions (Gabrovšek et al. 2010). CHEMICAL AND MICROBIOLOGICAL PARAMETERS AND THEIR INTERRELATIONSHIPS Stable isotopes indicated activation of other flow paths during high flows. In addition, the chemical composition of the water provided information on interactions with the surroundings, e.g., rocks and sediments, and even the impact of (bio)chemical conversions. The measured pa­rameters showed differences between sites, some of them being directly related to changes in hydrological condi­tions. Levels of dissolved oxygen were usually lower at the ponor (Fig. 6A); this can be attributed partly to en­hanced microbial consumption and higher temperature at this site. Levels of nitrogen (maximum concentration of ammonium at 0.15 mg/l and nitrates at 6.76 mg/l) and phosphorus (< 0.5 mg/l) compounds were surprisingly low considering that the upstream area is subjected to ag­ricultural and industrial pressure and occasional pollu­tion impact. High concentrations of cultivable microbial indicators (Tab. 1) indicated that at least some of the ni­trogen, phosphorus and sulphur compounds were fixed in microbial biomass. Sulphate and chloride (Fig. 6B) exhibited greater fluctuations in concentration (Tab. 1). The presence of anionic surfactants, even in low concentrations, indi­cated pollution related to human activities upstream of the ponor. Water samples from 27 June 2013, 21 October 2013, 14 July 2014, 25 November 2014 and 1 September 2015 were also screened for the presence of chloroform, trichloroethane, tetrachloromethane, trichloroethane, bromodichloromethane, tetrachloroethene, dibromoch­loromethane, and bromoform. Quantities of these com­pounds were below detection limits. Concentrations of bacterial indicators varied among sites. They were higher in the upper part of the under­ground water flow (Sites 1 and 2, Fig. 6C), and particu­larly in the samples from 21 October 2013, 14 July 2014 and 28 June 2016. The proportions of different bacterial indicators within the community did not remain con­stant across the sampling sites. The lowest concentrations of bacterial indicators were encountered during the high discharge of the Pivka River on 25 November 2014. They were accompanied by the lowest recorded values of TOC and other chemical parameters, which were caused by water dilution due to precipitation (Tab. 1). PCA analyses were run to investigate further the re­lationships between the measured parameters across the sampling campaigns and to explain the contributions of individual original variables (Fig. 7). The first three ex­tracted PCs account for more than 83% of data variance, i.e., of parameters’ variability: PC1 – 52.7 %, PC2 – 19.2 % and PC3 – 11.5 %. The remaining 18 PCs (designated as Residuals in Fig. 7) therefore capture only a tiny per­centage of the data variability. Several of the investigated parameters showed simi­lar behaviour with respect to the monitored sampling sites. Each of the eleven variables that are situated close to each other on the right-hand side of the PC1–PC2 load­ings diagram (Fig. 8A) – NECCO, COL, BAC, NCOBA, NECEN, ENCOC, TURBID, ENBAC, NH4, ECO and NENBA (see also Fig. 7) – exhibits a rather monotonous trend of decreasing values in the direction of river flow, i.e., between sites 1 > 2 > 3 > 4 (or in some rare cases 2 > 1 > 3 > 4). This is true for four of the five sampling campaigns. Sampling performed on 25 November 2014 is evidently an exception to this rule, because here most of the measurements for all four sites show only a minor fluctuation; this fact is also evident when looking at the PC1–PC2 scores diagram (Fig. 8B), where all four cor­responding data points, i.e., sampling sites, are located in close proximity in the bottom-left part of the plot. This can be explained by specific hydrological conditions. The samples were taken during high water conditions at the end of the strongest flood pulse in the observed period, when a high dilution by rainwater and fast transport within the cave system balanced the chemical and bacte­rial composition of the groundwater. On the other hand, variables located on the left-hand side of the loadings plot are generally characterized by an increase in their values along the river course 1< 2< 3< 4. This is true for DO in particular, and to some extent also for nitrate. In addition, pronounced negative projec­tion of the variable DO onto the horizontal axis – PC1 – revealed that this parameter was high during sampling in November 2014, because all four corresponding data points (1–4) in the scores diagram are also marked by significant negative projections onto PC1. Chloride and to some extent TOC and nitrate are, according to Fig. 7 and 8A, the variables that are mainly responsible for the variability between the four sampling dates. Measured chloride values were highest in September 2015 and lowest in July and November 2014 (see locations of the corresponding data points in Fig. 8B). In contrast to TOC and nitrate, which are sub­jected to dilution and biochemical processes, chloride does not suffer microbial conversions. The underground Pivka River is an example of a well karstified and dynamic underground karst system. During periods of high discharge there are no great dif­ferences in physical, chemical and bacterial parameters between the sampling sites, whereas during times of low discharge clear gradients are apparent. Similar values of chemical and bacterial parameters during the high flow in November 2014 confirmed the expected behav­iour, which can be attributed to a very high dilution by rainwater and a homogenous, well-mixed and fast pulse of water with no distinguished underground gradient. However, stable isotope analyses showed a clear gradient during that event (Fig. 4), which indicates the activation of flow paths delivering isotopically lighter percolation waters from the karst massif. Chloride, TOC and nitrate represented the most in­dicative parameters describing the formation of a chemi­cal gradient. Not only these parameters, but also oth­ers are subjected to dilution effects in the underground (Tab. 1, Fig. 8A). Some parameters also reflect microbial metabolism, e.g., TOC and heterotrophic decomposition of organic matter, and nitrates and nitrification. The im­pact of nitrification, indicated by an increase in nitrate concentrations downstream from the ponor, should not be underestimated in the Pivka River and similar karst systems. The bacterial groups used in the study, which reflect mostly human impact, are not crucial for biogeo­chemical cycling of elements and matter. They represent only a minor part of the underground microbiome, but their importance lies rather in their potential pathogenic nature for humans and animals. In comparison to the conditions of surface aquatic ecosystems, karst ground waters are exposed to more stable environmental conditions. Hydrological condi­tions and the microbiome direct the fate of organic loads and pollution. Continuous measurements of physical pa­rameters and regular monitoring of chemical parameters and bacterial indicators, in this case particularly E. coli and enterococci, are crucial to monitor the health of the ecosystem particularly during varying hydrological con­ditions. CONCLUSIONS Underground karst systems with natural accessibility, such as the underground Pivka River, provide excellent study sites to monitor the quantitative changes of chemical and biological factors. Concentration gradients of physico-chemical and bacterial parameters towards the interior of the karst massif were common during stable hydrological conditions, with chloride, TOC and nitrate concentrations being the most indicative parameters. Impact of rainwa­ter and fast transport are responsible for equilibrium of chemical and bacterial parameters in the underground. Environmental pollution reflected in faecal bacteria can be used as an additional natural tracer in order to provide better understanding of underground water dynamics and formation of concentration gradients. Simultaneous analysis of stable isotopes gives a more comprehensive prospect on underground karst hydrology. Analyses of sta­ble isotopes of hydrogen and oxygen in the underground Pivka River indicated mixing with isotopically lighter per­colation water along the underground river, which can be explained by an important share of autogenic recharge in the observed binary karst system. Additionally, the results indicated the mobilization of otherwise less active flow paths during high-flow conditions. More data and long-term monitoring are needed to confirm whether this is a common mechanism for the studied karst system. In order to understand water dynamics and health of the under­ground ecosystem, the chemical compounds that enter microbial metabolism, e.g., nitrate, ammonium, sulphates and organic compounds, should be carefully interpreted and correlated with microbial biomass and activity. ACKNOWLEDGEMENTS The authors acknowledge the monitoring plan for Pos­tojna Cave, and financial support from the Slovenian Research Agency (research core funding No. P6-0119) and project (“Natural resources of karst show caves: a balance among protection, exploitation, and promotion”, No. J7-7100). The authors acknowledge David Lowe for language editing assistance. REFERENCES Abdi, H. & L.J. Williams, 2010: Principal component analysis.- Wiley Interdisciplinary Reviews: Com­putational Statistics, 2, 433-459. DOI: http://dx.doi.org/10.1002/wics.101. Bailly-Comte, V., Jourde, H. & S. Pistre, 2009: Conceptu­alization and classification of groundwater-surface water hydrodynamic interactions in karst water­sheds: Case of the karst watershed of the Coulazou River (Southern France).- Journal of Hydrology, 376, 456-462. DOI: https://doi.org/10.1016/j.jhy­drol.2009.07.053. Buser, S., Grad, K. & M. Pleničar, 1967: Basic geologi­cal map of SFRY 1:100.000, sheet Postojna (in Slo­vene).- Federal Geological Survey, Belgrade. Clark, I.D. & P. Fritz, 1997: Environmental isotopes in hy­drogeology.- Lewis, pp. 352, Boca Raton. Ender, A., Goeppert, N. & N. Goldscheider, 2018: Spatial resolution of transport parameters in a subtropical karst conduit system during dry and wet seasons.- Hydrogeology Journal, 26, 2241-2255. DOI: https://doi.org/10.1007/s10040-018-1746-x. Gabrovšek, F., Kogovšek, J., Kovačič, G., Petrič, M., Ravbar, N. & J. Turk, 2010: Recent results of tracer tests in the catchment of the Unica River (SW Slove­nia).- Acta Carsologica, 39, 27-37. DOI: https://doi.org/10.3986/ac.v39i1.110. Golež, M. & A. Hladnik, 2013: Interpreting the age of the ruins of St. John the Baptist's church with mul­tivariate analysis.- Journal of Cultural Heritage, 14, 354-358. DOI: https://doi.org/10.1016/j.cul­her.2012.09.005. Hartmann, A., Goldscheider, N., Wagener, T., Lange, J. & M. Weiler, 2014: Karst water resources in a changing world: Review of hydrological modeling approach­es.- Reviews of Geophysics, 52, 218-242. DOI: htt­ps://doi.org/10.1002/2013rg000443. Heinz, B., Birk, S., Liedl, R., Geyer, T., Straub, K., Andres­en, J., Bester, K. & A. Kappler, 2009: Water quality deterioration at a karst spring (Gallusquelle, Ger­many) due to combined sewer overflow: evidence of bacterial and micro-pollutant contamination.- En­vironmental Geology, 57, 797-808. DOI: https://doi.org/10.1007/s00254-008-1359-0. Hladnik, A. & T. Muck, 2002: Characterization of pig­ments in coating formulations for high-end ink-jet papers.- Dyes and Pigments, 54, 253-263. DOI: htt­ps://doi.org/10.1016/s0143-7208(02)00050-5. Holko, L., Danko, M., Hlavčo, J., Kostka, Z. & P. Malík, 2015: Isotopic and chemical composition monitor­ing of surface and ground waters in theg Jalovecký creek catchment (Western Tatra Mts., Liptovská kotlina valley) (in Slovak).- Podzemná voda, 21, 83-100. Hutchins, B., Engel, A., Nowlin, W. & B. Schwartz, 2016: Chemolithoautotrophy supports macroinverte­brate food webs and affects diversity and stability in groundwater communities.- Ecology, 97, 1530-1542. DOI: https://doi.org/10.1890/15-1129.1. Jianhua, M., Mingming, H. & Z. Xi, 2016: Present situa­tion and control of outbreaks of algal blooms in the three gorges reservoir.- Meteorological and Envi­ronmental Research, 7, 33-37. Katz, B.G., 2002: Demystifying groundwater flow and contaminant movement in karst systems using chemical and isotopic tracers.- In: Kuniansky, E.L. (ed.) US Geological Survey Karst Interest Group, Proceedings, 20th-22nd August 2002, Shepherd­stown, West Virginia, U.S. Geological Survey, 13-19, Atlanta. Kendall, C. & E.A. Caldwell, 1998: Fundamentals of iso­tope geochemistry.- In: Kendall, C. & J.J. McDon­nell (eds.) Isotope tracers in catchment hydrology. Elsevier, pp 51-86, Amsterdam. DOI: https://doi.org/10.1016/b978-0-444-81546-0.50009-4. Kogovšek, J. & M. Petrič, 2004: Advantages of longer-term tracing - three case studies from Slovenia.- En­vironmental Geology, 47, 76-83. DOI: https://doi.org/10.1007/s00254-004-1135-8. Koit, O., Ravbar, N., Marandi, A. & J. Terasmaa, 2017: Treshold-controlled three-stage hydraulic be­haviour of a mantled shallow carbonate aquifer (Tuhala karst area, North Estonia).- Acta Carso­logica, 46, 265-282. DOI: https://doi.org/10.3986/ac.v46i2-3.4951. Long, D., Voice, T., Niagolova, N. & S. McElmurry, 2012: Effects of human activities on karst groundwater geochemistry in a rural area in the Balkans.- Ap­plied Geochemistry, 27, 1920-1931. Macler, B. & J. Merkle, 2000: Current knowledge on groundwater microbial pathogens and their con­trol.- Hydrogeology Journal, 8, 29-40. DOI: https://doi.org/10.1007/pl00010972. Morita, H., Ushiyama, M., Aoyama, S. & M. Iwasaki, 2003: Sensitivity and specificity of the Sanita-kun Aerobic Count: Internal validation and indepen­dent laboratory study.- Journal of AOAC Interna­tional, 86, 355-366. Mulec, J., Krištůfek, V. & A. Chroňáková, 2012: Com­parative microbial sampling from eutrophic caves in Slovenia and Slovakia using RIDA®COUNT test kits.- International Journal of Speleology, 41,1-8. DOI: https://doi.org/10.5038/1827-806x.41.1.1. Oarga, A., Bulc, T.G., Jenssen, P.D. & J. Mulec, 2012: Monitoring of microbial indicator groups in organi­cally heavily loaded wastewater treatment systems by using RIDA®COUNT kits.- Fresenius Environ­mental Bulletin, 21, 3886-3893. Pipan, T. & D. Culver, 2007: Epikarst communities: bio­diversity hotspots and potential water tracers.- En­vironmental Geology, 53, 265-269. DOI: https://doi.org/10.1007/s00254-007-0640-y. Pronk, M., Goldscheider, N. & J. Zopfi, 2006: Dynam­ics and interaction of organic carbon, turbidity and bacteria in a karst aquifer system.- Hydrogeology Journal, 14, 473-484. DOI: https://doi.org/10.1007/s10040-005-0454-5. Ravbar, N., Barbera, J., Petrič, M., Kogovšek, J. & B. An­dreo, 2012: The study of hydrodynamic behaviour of a complex karst system under low-flow conditions using natural and artificial tracers (the catchment of the Unica River, SW Slovenia).- Environmental Earth Sciences, 65, 2259-2272. DOI: https://doi.org/10.1007/s12665-012-1523-4. Shi, Z., Liu, X., Liu, Y., Huang, Y. & H. Peng, 2009: Cata­strophic groundwater pollution in a karst environ­ment: a study of phosphorus sludge waste liquid pollution at the Penshuidong Cave in Yunnan, China.- Environmental Earth Sciences, 59,757-763. DOI: https://doi.org/10.1007/s12665-009-0071-z. Tissier, G., Perrette, Y., Dzikowski, M., Poulenard, J., Ho­blea, F., Malet, E. & B. Fanget, 2013: Seasonal chang­es of organic matter quality and quantity at the out­let of a forested karst system (La Roche Saint Alban, French Alps).- Journal of Hydrology, 482, 139-148. DOI: https://doi.org/10.1016/j.jhydrol.2012.12.045. Vallejos, A., Andreu, J., Sola, F. & A. Pulido-Bosch, 2015: The anthropogenic impact on Mediterranean karst aquifers: cases of some Spanish aquifers.- Environ­mental Earth Sciences, 74, 185-198. DOI: https://doi.org/10.1007/s12665-014-3994-y. Vesper, D.J., Loop, C.M. & W.B. White, 2001: Contami­nant transport in karst aquifers.- Theoretical and Applied Karstology, 13, 101-111. Wang, B., Qiu, X., Peng, X. & F. Wang, 2018: Phytoplank­ton community structure and succession in karst cascade reservoirs, SW China.- Inland Waters, 8, 229-238. DOI: https://doi.org/10.1080/20442041.2018.1443550. A MULTIPARAMETER ANALYSIS OF ENVIRONMENTAL GRADIENTS RELATED TO HYDROLOGICAL CONDITIONS IN A BINARY KARST SYSTEM (UNDERGROUND COURSE OF THE PIVKA RIVER, SLOVENIA) JANEZ MULEC, METKA PETRIČ, ALENKA KOŽELJ, CLARISSA BRUN, ERIKA BATAGELJ, ALEŠ HLADNIK & LADISLAV HOLKO Fig. 1: Study site: A – hydro­geological map of the area with designated sampling sites along the underground Pivka River, B – schematic representation (not to scale) of sampling sites (1-4) along the underground Pivka River, with distances from the ponor in Postojna Cave; sources: Lidar (Slovenian Environment Agency); Hydrogeological map (Geological Survey of Slovenia); Cave cadastre (Karst Research Institute ZRC SAZU; Speleologi­cal Association of Slovenia). A MULTIPARAMETER ANALYSIS OF ENVIRONMENTAL GRADIENTS RELATED TO HYDROLOGICAL CONDITIONS IN A BINARY KARST SYSTEM (UNDERGROUND COURSE OF THE PIVKA RIVER, SLOVENIA) JANEZ MULEC, METKA PETRIČ, ALENKA KOŽELJ, CLARISSA BRUN, ERIKA BATAGELJ, ALEŠ HLADNIK & LADISLAV HOLKO Fig. 2: (A) Daily precipitation in Postojna and (B) discharges of the Pivka River at the ponor during the sampling campaigns. Fig. 3: An example of unstable hy­drological conditions in October 2013: A – precipitation, B – on-site measured EC at sampling sites 1 and 4, with designated sampling times at four sites along the ground­water course of the Pivka River. A MULTIPARAMETER ANALYSIS OF ENVIRONMENTAL GRADIENTS RELATED TO HYDROLOGICAL CONDITIONS IN A BINARY KARST SYSTEM (UNDERGROUND COURSE OF THE PIVKA RIVER, SLOVENIA) Fig. 4: .18O (‰) at sampling sites (1–4) during the study period; the blue and green symbols display summer values, orange and red symbols show the autumn values and white symbols indicate the evaporated water sampled in Sep­tember 2015. Fig. 5: Comparison of the waters from different sampling campaigns with indicated sampling sites (1–4) relative to the Global Meteoric Wa­ter Line (GMWL, .2H=8.18O+10). JANEZ MULEC, METKA PETRIČ, ALENKA KOŽELJ, CLARISSA BRUN, ERIKA BATAGELJ, ALEŠ HLADNIK & LADISLAV HOLKO Tab. 1: Ranges of physico-chemical and bacterial parameters of the underground Pivka River at sampling sites, with designated codes for PCA analysis. SAMPLING CAMPAIGN Parameter Code (PCA) 27-Jun-13 21-Oct-13 14-Jul-14 25-Nov-14 1-Sep-15 15-Dec-15 28-Jun-16 T (°C) TEMP 11.9–16.6 10.2–11.3 13.0–15.6 6.7–7.8 13.6–18.9 2.3–7.2 14.8–15.6 EC (µS/cm) COND 407–458 357–393 317–337 393–394 366–405 427–454 353–420 pH 7.75–8.16 7.76–8.06 7.55–7.72 7.99–8.08 7.56–7.75 7.91–8.26 7.68–7.88 DO (mg/l) DO 5.15–9.66 7.80–10.54 7.93–10.27 10.67–11.38 5.90–9.87 11.78–12.77 8.12–9.93 Colour (m–1) COLOR 0.20–0.50 0.10–0.20 1.00–1.40 0.20–0.23 0.59–1.22 nd nd Turbidity (NTU) TURBID 0.49–3.40 2.40–6.00 3.80–15.10 1.10–1.40 1.20–7.00 nd nd NH4 (mg/l) NH4 0.007–0.087 0.022–0.072 0.031–0.153 0.009–0.015 0.011–0.063 nd 0.030–0.074 NO2 (mg/l) <0.001–0.130 0.010–0.055 0.011–0.056 0.004–0.008 0.008–0.061 nd 0.010–0.061 NO3 (mg/l) NO3 3.04–5.88 3.45–6.37 2.60–4.18 2.82–3.12 4.20–6.57 2.05–6.76 nd F (mg/l) <0.100 <0.100–0.120 0.112–0.127 0.022–0.051 0.046–0.111 nd 0.037–0.172 Cl (mg/l) Cl 8.0–20.7 10.4–17.0 4.9–8.6 3.4–3.7 24.0–28.3 9.9–16.8 9.8–13.0 SO4 (mg/l) SO4 4.3–6.5 11.6–13.3 3.5–7.0 2.7–3.1 10.3–10.9 8.4–12.5 4.3–4.8 o-PO4 (mg/l) o-PO4 0.075–0.186 0.270–0.310 0.114–0.171 0.016–0.021 0.095–0.109 0.090–0.120 0.037–0.172 Tot-PO4 (mg/l) TOT-PO4 0.105–0.342 0.370–0.460 0.215–0.327 0.041–0.057 0.111–0.184 nd 0.117–0.228 Anionic surfactants (mg/l) <0.010–0.015 0.011–0.017 <0.010–0.015 0.002–0.007 0.005–0.012 nd nd TOC (mg/l) TOC 4.57–7.21 3.27–5.00 3.56–4.69 1.24–1.34 3.96–6.79 nd 1.21–3.32 Bacteria (CFU/ml) BAC 84–835 163–2520 398–2020 106–172 191–531 102–274 274–7030 Coliforms (CFU/ml) COL 29–206 40–389 109–443 35–60 54–262 26–67 147–1020 Enterococci (CFU/100 ml) ENCOC 0–20 7–530 80–820 37–72 12–57 nd nd Enterobacteria (CFU/ml) ENBAC 40–294 40–627 138–795 34–72 57–332 29–83 125–1300 E. coli (CFU/ml) ECO 0–6 1–24 3–48 1–4 1–4 0–1 1–78 Non-coliforms (CFU/ml) NCOBA 55–633 121–2131 289–1657 71–122 137–391 75–207 167–3520 Non-enterics (CFU/ml) NENBA 44–557 116–2385 216–1335 62–110 133–365 73–204 149–5730 Non-E. coli coliforms (CFU/ml) NECCO 29–200 40–366 106–395 34–57 52–258 26–66 146–942 Non-E. coli enterics (CFU/ml) NECEN 40–289 39–609 135–747 33–69 56–328 29–83 124–1231 A MULTIPARAMETER ANALYSIS OF ENVIRONMENTAL GRADIENTS RELATED TO HYDROLOGICAL CONDITIONS IN A BINARY KARST SYSTEM (UNDERGROUND COURSE OF THE PIVKA RIVER, SLOVENIA) Fig. 6: Gradients of (A) oxygen, (B) chlorides and (B) bacteria between sites 1–4, with respect to the dis­tance from the ponor of the under­ground Pivka River during differ­ent hydrological conditions. JANEZ MULEC, METKA PETRIČ, ALENKA KOŽELJ, CLARISSA BRUN, ERIKA BATAGELJ, ALEŠ HLADNIK & LADISLAV HOLKO Fig. 7: PCA results show the contri­bution of measured parameters to the first three PCs. Fig. 8: PCA results: A – PC1-PC2 loadings plot, B – PC1-PC2 scores plot. A MULTIPARAMETER ANALYSIS OF ENVIRONMENTAL GRADIENTS RELATED TO HYDROLOGICAL CONDITIONS IN A BINARY KARST SYSTEM (UNDERGROUND COURSE OF THE PIVKA RIVER, SLOVENIA) JANEZ MULEC, METKA PETRIČ, ALENKA KOŽELJ, CLARISSA BRUN, ERIKA BATAGELJ, ALEŠ HLADNIK & LADISLAV HOLKO A MULTIPARAMETER ANALYSIS OF ENVIRONMENTAL GRADIENTS RELATED TO HYDROLOGICAL CONDITIONS IN A BINARY KARST SYSTEM (UNDERGROUND COURSE OF THE PIVKA RIVER, SLOVENIA) JANEZ MULEC, METKA PETRIČ, ALENKA KOŽELJ, CLARISSA BRUN, ERIKA BATAGELJ, ALEŠ HLADNIK & LADISLAV HOLKO Appendix: Measured parameters of the underground Pivka River at sampling sites, part 1. Sampling date Sampling site T (°C) EC (µS/cm) DO (mg/l) Colour (m-1) Turbidity (NTU) NH4 (mg/l) NO2 (mg/l) NO3 (mg/l) F (mg/l) Cl (mg/l) SO4 (mg/l) o-PO4 (mg/l) TOT-PO4 (mg/l) Anionic surfactants (mg/l) 27-Jun-13 1 16.6 458 5.15 0.50 3.40 0.087 0.130 3.04 < 0.1 20.7 6.5 0.186 0.314 0.013 27-Jun-13 2 16.4 456 7.88 0.40 3.20 0.036 0.032 4.08 < 0.1 19.5 6.2 0.171 0.342 0.015 27-Jun-13 3 15.0 436 8.00 0.30 1.30 0.011 <0.001 5.88 < 0.1 14.1 5.5 0.186 0.226 <0.01 27-Jun-13 4 11.9 407 9.66 0.20 0.49 0.007 0.004 4.78 < 0.1 8.0 4.3 0.075 0.105 <0.01 21-Oct-13 1 11.2 393 7.80 0.20 4.50 0.072 0.055 3.56 0.120 17.0 13.2 0.290 0.440 0.017 21-Oct-14 2 11.3 390 9.14 0.20 6.00 0.034 0.035 3.45 0.120 16.0 13.3 0.270 0.400 0.012 21-Oct-15 3 10.2 383 10.54 0.20 2.40 0.024 < 0.010 5.46 < 0.1 12.8 12.1 0.310 0.370 0.012 21-Oct-16 4 10.3 357 10.49 0.10 2.50 0.022 0.019 6.37 < 0.1 10.4 11.6 0.290 0.460 0.011 14-Jul-14 1 15.5 317 7.93 1.40 15.10 0.153 0.043 2.60 0.126 6.8 7.0 0.155 0.321 0.015 14-Jul-14 2 15.6 319 8.79 1.40 14.60 0.084 0.040 2.77 0.127 7.1 6.7 0.155 0.327 0.012 14-Jul-14 3 15.5 336 9.40 1.20 8.30 0.070 0.056 3.21 0.113 8.6 5.8 0.171 0.313 0.009 14-Jul-14 4 13.0 337 10.27 1.00 3.80 0.031 0.011 4.18 0.112 4.9 3.5 0.114 0.215 nd 25-Nov-14 1 6.7 393 10.84 0.20 1.40 0.010 0.008 2.86 0.022 3.4 2.7 0.021 0.041 0.004 25-Nov-14 2 6.7 394 10.67 0.20 1.40 0.009 0.008 3.04 0.039 3.6 2.7 0.016 0.048 0.007 25-Nov-14 3 7.0 394 11.04 0.20 1.40 0.012 0.006 2.82 0.044 3.6 3.1 0.017 0.047 0.004 25-Nov-14 4 7.8 394 11.38 0.23 1.10 0.015 0.004 3.12 0.051 3.7 3.0 0.016 0.057 0.002 1-Sep-15 1 18.9 389 5.90 1.04 7.00 0.063 0.061 4.20 0.111 27.6 10.9 0.098 0.153 0.008 1-Sep-15 2 18.7 393 7.52 0.99 4.80 0.023 0.015 4.74 0.065 28.3 10.4 0.109 0.184 0.005 1-Sep-15 3 15.8 366 9.19 1.22 4.00 0.014 0.010 5.62 0.072 24.0 10.3 0.098 0.123 0.007 1-Sep-15 4 13.6 405 9.87 0.59 1.20 0.011 0.008 6.57 0.046 28.3 10.3 0.095 0.111 0.012 15-Dec-15 1 2.3 454 12.49 nd nd nd nd nd nd 16.8 12.5 0.098 nd nd 15-Dec-15 2 3.1 454 12.77 nd nd nd nd nd nd 15.9 11.9 0.094 nd nd 15-Dec-15 3 5.5 444 12.32 nd nd nd nd nd nd 14.5 12.0 0.110 nd nd 15-Dec-15 4 7.2 427 11.78 nd nd nd nd nd nd 9.9 8.4 0.124 nd nd 28-Jun-16 1 14.8 357 8.12 nd nd 0.074 0.061 2.80 nd 11.0 4.5 0.142 0.209 nd 28-Jun-16 2 14.9 354 8.85 nd nd 0.059 0.055 2.90 nd 9.8 4.3 0.145 0.222 nd 28-Jun-16 3 15.6 353 9.50 nd nd 0.057 0.043 3.30 nd 10.0 4.5 0.172 0.228 nd 28-Jun-16 4 15.0 420 9.93 nd nd 0.030 0.010 4.60 nd 13.0 4.8 0.037 0.117 nd A MULTIPARAMETER ANALYSIS OF ENVIRONMENTAL GRADIENTS RELATED TO HYDROLOGICAL CONDITIONS IN A BINARY KARST SYSTEM (UNDERGROUND COURSE OF THE PIVKA RIVER, SLOVENIA) Appendix: Measured parameters of the underground Pivka River at sampling sites, part 2. Sampling date Sampling site TOC Bacteria (CFU/ml) Coliforms (CFU/ml) Enterococci (CFU/ml) Enterobacteria (CFU/ml) E. coli (CFU/ml) Non-coliforms (CFU/ml) Non-enterics (CFU/ml) Non-E. coli coliforms (CFU/ml) Non-E. coli enterics (CFU/ml) 27-Jun-13 1 6.75 835 202 20 278 5 633 557 197 274 27-Jun-13 2 7.21 630 206 7 294 6 424 336 200 289 27-Jun-13 3 5.04 381 148 3 221 0 234 160 148 221 27-Jun-13 4 4.57 84 29 0 40 0 55 44 29 40 21-Oct-13 1 5.00 2520 389 530 136 24 2131 2385 366 112 21-Oct-14 2 4.71 1860 332 380 627 18 1529 1233 314 609 21-Oct-15 3 3.55 166 40 10 40 1 126 127 40 39 21-Oct-16 4 3.27 163 43 7 48 1 121 116 42 47 14-Jul-14 1 4.69 2020 363 680 685 27 1657 1335 337 659 14-Jul-14 2 4.55 1590 443 820 795 48 1147 795 395 747 14-Jul-14 3 4.12 1510 320 570 558 10 1190 953 310 548 14-Jul-14 4 3.56 398 109 80 138 3 289 261 106 135 25-Nov-14 1 1.34 133 59 72 72 3 74 62 56 69 25-Nov-14 2 1.34 172 50 74 72 4 122 100 46 69 25-Nov-14 3 1.24 168 60 61 58 3 108 110 57 55 25-Nov-14 4 1.26 106 35 37 34 1 71 72 34 33 1-Sep-15 1 6.79 531 262 57 332 4 269 199 258 328 1-Sep-15 2 5.93 520 129 12 155 1 391 365 128 154 1-Sep-15 3 6.26 353 120 16 152 2 233 201 118 150 1-Sep-15 4 3.96 191 54 22 58 2 137 133 52 56 15-Dec-15 1 nd 208 65 nd 83 0 144 126 65 83 15-Dec-15 2 nd 274 67 nd 71 1 207 204 66 70 15-Dec-15 3 nd 102 27 nd 29 0 75 73 27 29 15-Dec-15 4 nd 147 26 nd 42 0 121 105 26 42 28-Jun-16 1 3.32 7030 760 nd 1300 69 6270 5730 691 1231 28-Jun-16 2 3.07 6750 1020 nd 1180 78 5730 5570 942 1102 28-Jun-16 3 2.68 2050 860 nd 1180 39 1190 870 821 1141 28-Jun-16 4 1.21 274 147 nd 125 1 127 149 146 124 nd- no data