Carbonate sedimentation and effects of eutrophication observed at the Kaliˇsta subaquatic springs in Lake Ohrid (Macedonia)

. To date, little is known about the role of spring waters with respect to authigenic carbonate precipitation in a shallow lacustrine setting. Lake Ohrid, located in South-eastern Europe, is a large lake fed to over 50% by karstic springs of which half enter subaquatically and inﬂuence signiﬁcantly its ecology and species distribution. In order to evaluate how sedimentological processes are inﬂuenced by such shallow-water springs, the Kaliˇsta subaquatic spring area in the north west of Lake Ohrid was investigated by a sidescan sonar survey and with sediment traps and three transects of gravity short cores. Results indicate that sedimentation in the spring area is dominated by authigenic carbonate precipitation. High sedimentation rates and evidences for bio-induced precipitation processes were observed in the water column and in the sediments. Two distinct stratigraphic units characterize the shallow subsurface, both composed of carbonate silts with high carbonate contents of up to 96%, but differing in color, carbonate content and diatom content. A chronological correlation of the cores by radiocarbon dates and 137 Cs activities places the transition between the two stratigraphic units after (cid:24) 1955 AD. At that time, coastal sedimentation changed drastically to signiﬁcantly darker sediments with higher contents of organic matter and more abundant diatoms. This change coincides with the recent human impact of littoral eutrophication.


Introduction
Subaquatic springs are often a substantial source of ionrich waters that may affect authigenic mineral-forming processes in lakes. In particular calcium-rich waters from karstic carbonate-dominated catchments may provide Ca 2+ and HCO − 3 ions to significantly enhance lacustrine carbonate precipitation. With this contribution, we highlight the role of the subaquatic springs in Lake Ohrid for its sedimentbuilding processes and recent sedimentation pattern.
More than for most other lakes, the specific ecology of Lake Ohrid is particularly a mirror of its catchment and its eco-climatological setting and thus likely is strongly dependent on subaquatic springs. Lake Ohrid, shared by Macedonia and Albania ( Fig. 1), was probably formed during the Tertiary (Meybeck, 1995). Therefore it is one of the most ancient and long-lived permanent lakes in Europe (Wagner et al., 2008). This long timescale of existence allowed the independent development of an extraordinary ecosystem, characterized by more than 200 endemic species (Stankovic, 1960).
The establishment of this spectacular biodiversity was supported by the subaquatic inflow of cool, clean and oxygenrich water into the lake (Matzinger et al., 2007). Particularly the subaquatic springs seem to have a large potential to support specific habitats for endemic species as they supply oxygen, nutrients, and ions such as calcium at shallow as well as at greater depths. Moreover, they create distinctly different, but constant boundary conditions for the organisms near the springs. Several endemic organisms are found exclusively close to the springs and indicate that the spring water may have been important for the evolution of the unique aquatic Published by Copernicus Publications on behalf of the European Geosciences Union. Table 1. Chemical and physical parameters of Kališta spring waters in April, May and September 2007. The samples were taken at the outlets of surface and subaquatic springs. Barrel, flask and plastic bag indicate different sampling techniques. Lake water samples were taken at different depths above the springs. Saturation indices for carbonate were calculated according to Plummer and Busenberg (1982 ecosystem of Lake Ohrid (Stankovic, 1960;Matzinger et al., 2006b;Albrecht and Wilke, 2008;Hauffe et al., 2010;Trajanovski et al., 2010). Because the hydraulic through-flow is only weak in Lake Ohrid, with a long water residence time of ∼70 yr, those subaquatic springs contribute a significant portion of the water budget, as Lake Ohrid is fed to ∼54 % by spring waters. The water balance of these springs indicates that half of the spring water enters the lake at its surface, while the other half intrudes subaquatically (Matzinger et al., 2006b). Furthermore, the dilution of regular lake waters with pure spring waters (Matzinger et al., 2006a) is important for the oligotrophic status of Lake Ohrid and its subsequent low algae productivity. However, knowledge on the activity and the hydrological conditions of these springs including their sedimentological environment is very limited or entirely absent. First studies focused on regular analyses of the water chemistry (Table 1) since 2004 so that now information is available on how these waters potentially affect limnologic processes in the lake. The sediment deposition nearby the subaquatic springs is an interesting archive reflecting the role of those springs. Sediments act as recorders of chemical, physical and biological boundary conditions, so that changes in lake and/or spring waters should be reflected in the sediment composition. Especially, the high Ca 2+ and HCO − 3 content of the spring waters (Table 1) suggest that authigenic calcite precipitation could take place particularly near the springs affect-ing the local sedimentation processes. Previous studies from deep basin sediment cores indicate that carbonate precipitation seems to have been absent or reduced in Lake Ohrid during certain periods of the Holocene, which could have been caused by lower ion concentrations and/or less spring activity (Shapley et al., 2005;Wagner et al., 2009Wagner et al., , 2010Vogel et al., 2010b). As the subsurface of Lake Ohrid is targeted for a deep drilling campaign in the frame of the International Continental Scientific Deep Drilling Program (ICDP; Wagner et al., 2008), knowledge of the sedimentologic role of the springs is needed in order to establish the paleoenvironmental implications of the future deep drill cores.
This study thus aims at characterizing the sedimentation pattern at the Kališta spring area. We address the question, whether the subaquatic springs substantially influence the in-situ carbonate precipitation as it is stored in the sediments. We will investigate, whether spring sedimentation changed through time. One concern is the impact of humaninduced eutrophication since the 1950s due to increasing nutrient input following the growing population and its anthropogenic activity. Due to its large volume and slow response time (Matzinger et al., 2007), Lake Ohrid is still oligotrophic and eutrophication could be delayed. A potential effect on the shallow water realm has never been investigated and the record of the near-spring sediments provides important integrative information.  right: Sidescan sonar image of the spring area and surroundings with mooring locations (green crosses) and coring locations (blue dots). The vertical lines mimic the ship tracks and are sidescan artifacts. Note surface roughness in spring area, indicating partly exposed bedrock and macrophyte patches. a This data is omitted for discussion. The mass accumulation rate of 135 g m −2 d −1 is an artifact: Probably not the entire sediment trapped in 2e actually sedimented, but part of it was flushed into this trap through turbulence from the surrounding lake floor.
Carbonate sedimentation at springs in Lake Ohrid 1 Fig. 3: Schematics of the two moorings with pairs of cylinder sediment traps (with 2 indicated distance from ground) deployed in two local depressions in the spring zone. 3 4 Fig. 3. Schematics of the two moorings with pairs of cylinder sediment traps (with indicated distance from ground) deployed in two local depressions in the spring zone.

Study site
Lake Ohrid (40 • 54 to 41 • 10 N and 20 • 38 to 20 • 48 E) is situated at 693 m a.s.l. in a mountainous and partly karstic region. Lake Ohrid is large (358 km 2 ) and deep (289 m) and is one of the most voluminous (55 km 3 ) lakes in Europe. A particularity of Lake Ohrid is the ∼27% contribution of the subaquatic springs to the overall water input of ∼38 m 3 s −1 (Watzin et al., 2002;Matzinger et al., 2007). Together with another ∼27% inflow by surface springs, the karstic springs provide altogether the major part of the Lake Ohrid water input. The karstic aquifers are fed either by infiltrated precipitation or by nearby Lake Prespa (Matzinger et al., 2006a), which is situated ∼160 m higher towards south-east (Fig. 1). There are in total about five known major zones of subaquatic inflows (Fig. 1), as identified and located in Albrecht and Wilke (2008).
In this paper, we focus on the Kališta spring area, situated in the north-western part of the lake (Fig. 1), which is fed by infiltrated atmospheric water. The site's particular lake floor can be observed by eye through the clear waters. The spring area is characterized by water depths of 4 to 6 m, but there are depressions with maximum water depths of 8 m. The most active subaquatic springs are concentrated in a ∼8 m deep depression concentrated in a 40 to 60 m wide zone located ∼100 m from the shore (Fig. 2). Saturation indices for calcium carbonate (Plummer and Busenberg, 1982) are, Observations at these underwater springs show outflowing "bubbling" water with 10 to 30 cm-high white plumes sometimes forming small craters.

Methods
Two moorings with sediment traps were positioned in depressions identified by echosounder profiles at 7.6 and 8.0 m water depths within the subaquatic spring zone during two weeks in May 2007 (Figs. 2, 3). Divers tied the moorings to anchors after the disturbed sediment had settled for 10 to 15 min. The sediment traps consisted each of two paired cylindrical open tubes 50 to 70 cm in length, with an inner diameter of 5.8 cm. Both moorings had two and five traps, respectively, at various water depths ( Fig. 2; Table 2). The vertical position of the mooring during the 14-days sampling period was guaranteed by buoys. The protection caps on the two lower sediment traps were removed 10 min after mooring deployment, when disturbed sediment due to diver's movements had settled, thus preventing an impact on the collected sediment.
After recovery of the sediment traps, organic matter was removed with a 2% H 2 O 2 solution for four days, the homogeneous dry material of the sediment traps was weighed, and mass accumulation rates k (g m −2 d −1 ) of dry material were calculated using: where m dry (g) is the dry mass collected, a (0.0052 m 2 ) is the open trap area, and t the time interval of exposure (14 d).
Twenty gravity cores of up to 82 cm length were recovered in May 2007 along three transects (Fig. 2). Cores were taken along the shore every ∼100 m starting from the springs towards north and south in water depths of up to ∼10 m, as well as towards the center of the lake, reaching a water depth of ∼50 m. Additional push cores were recovered in September 2007 by divers within the depression of the active spring zone (core 46) and at a distance of 1250 m from the spring on the northern transect (core 49; Fig. 2).
Before opening, cores were scanned petrophysically with a Geotek multisensor core-logger (MSCL; wet bulk-density, p-wave velocity, magnetic susceptibility). Then they were opened, photographed and described macroscopically. Total carbon (TC) and total inorganic carbon (TIC) were analyzed using a coulometer (5012 Coulometer, UIC Inc. Coulometrics) measuring the CO 2 content derived from beforehand freeze-dried samples by burning (TC) and acidification (TIC). Carbonate contents were calculated based on the assumption that TIC is completely formed by CaCO 3 . Total organic carbon (TOC) content was calculated as difference between TC and TIC. Grain size was measured through laser diffraction with a Malvern Mastersizer 2000 Particle Analyzer. X-ray fluorescence (XRF) core scanning at 2 mm resolution with a COX ITRAX core scanner provided a non-destructive and fast way of analyzing the elemental composition of the downcore chemical profiles. Smear slides were prepared from selected depths in the cores and from trap material to analyze with a polarization microscope the composition and crystal morphology. Furthermore, particles were also analyzed with a Scanning Electron Microscope (SEM) including energy-dispersive X-ray spectroscopy (EDX) chemical analysis, in order to investigate mineralogy, morphology and origin of particles.
Chronostratigraphy of the sediment cores was determined on the basis of (i) radiocarbon analysis (measured at the Leibniz Laboratory for Radiometric Dating and Isotope Research in Kiel, Germany; Table 3) on terrestrial organic matter and (ii) using 210 Pb and 137 Cs activities measured with Germanium gamma detectors. Radiocarbon data were calibrated using the OxCal4.1 program and the IntCal09 data set (Bronk Ramsey, 2009;Reimer et al., 2009).
In spring 2008, a sidescan sonar survey was carried out covering the larger spring zone with a Klein 3000 dual frequency sidescan sonar operated at frequencies of 100 and 500 kHz. Individual transect covered 75 m at each side, with an overlap of about 50%. Mosaicking of profiles and target management was done using the software package Sonar Wiz.

Sediment traps
The sediment traps were deployed on 15 May and retrieved on 29 May 2007, and covered a period characterized by the absence of strong winds. After retrieval, the inside of the liners and the surface of the trapped sediment were covered by algae, indicating rapid growth. All traps contained cream-colored and flocculent sediment of variable amounts (Table 2). In the analyzed three trap samples (1b, 2d, 2e; Table 2; Fig. 3), carbonate content varies between 70.5 and 86.4%, and TOC (after 4 d of H 2 O 2 treatment for removing algal growth) between 1.4 and 6.6% (Table 2). The highest mass accumulation rate occurred in the deepest trap of mooring 2 (2e; Table 2) with 135 g m 2 d −1 . However, as a dead fish and two leeches were collected, this value is falsified. The lowest amount of particles was accumulated in the shallowest trap of mooring 2 (trap 2a) with 0.3 g m 2 d −1 (Table 2). In general, our data indicate that particle flux increases with water depth (Fig. 4).
Samples analyzed under the optical microscope in crosspolarized light and with the SEM/EDX allowed the characterization of the particles. Most particles were confirmed to be calcite crystals with different morphologies, ranging from perfect authigenic idiomorphic crystals (Fig. 5a) to corroded crystals or larger irregular clusters (Fig. 5b). The largest observed calcite crystals in the traps reach lengths of up to 20 µm (Fig. 5a). Large clusters of idiomorphic calcite crystals in sediment trap material can be up to 150 µm long. Prominent ∼1 µm large holes occur frequently in numerous calcite crystals from the sediment traps (Fig. 5b).

Lithologies
The sidescan sonar image confirmed the large depression forming the more active spring zone, which is separated by sharp edges from the surrounding area (Fig. 2). Inside the depression, dense patches of macrophytes were identified. At some locations directly in the spring area, coring failed due to absence of sediment so that PVC-liners broke upon hitting the exposed bedrock. Upon opening, many cores had a strong sewage smell, in particular from the topmost 5 cm.
The magnetic susceptibility of most cores is rather constant over depth and ranges between 50 and 150 × 10 −5 (SI units). All cores, except core 20, which originates from a deeper water depth of 53 m (Fig. 2), are composed only of two main lithostratigraphic (Units T and L) units that are characterized by a distinct change in lithology (Figs. 6,7,8) but by similar grain size distribution (56% silt on average; mean grain size ranges between 35 and 65 µm). On the basis of grain composition, grain size, color and geochemical signatures, Matter (2007) subdivided the lithologic succession of the recovered cores in seven and four lithotypes for Units T and L, respectively. Here, we present the range of lithologies within these units. The top part of the cores (Unit T; "Top"-unit) consists of dark-colored (brown to dark gray) carbonate silt with abundant diatoms frustules, which occur in smear slides of 19 out of 27 analyzed samples. Ostracods (>120 µm) and gastropod-shell fragments occur throughout the unit. Dominant grain size is in the silt-range with only few sand layers. Bioturbating chironomid larvae were found in core 05 just above the base of Unit T. No laminations occur and the sediments appear rather homogenous, except for core 05, which has a diffuse lamination in Unit T. The lower part of the cores comprises Unit L ('Lower' unit), which is composed of brighter-colored (light gray to gray) carbonate silt that shows in smear slides in contrast to Unit T only minor diatom contents. Ostracods and gastropods occur frequently and sand-sized layers become more abundant. Two particular coarser layers (>60 to 70% sand) are found at depths around 18 ± 5 cm and 33 ± 5 cm in most cores, respectively. They are characterized by lower wet bulk density values than in the fine-grained lithologies. Gravel of 0.5 to 2 cm diameter was found in the lower part of cores 01, 02, 03 and 46. Except for these coarser beds, the sediments of Unit L are also rather homogeneous and do not show a significant layering. Within the spring zone, Unit L gets bright-colored (bright beige to almost white) below 30 to 45 cm core depth, resembling lacustrine chalk.
Carbonate content varies between 62 to 90% in Unit T, and is higher in Unit L with values between 89 and 96% (Fig. 7), except in few coarser layers, in which carbonate content drops to 63 to 85%. The two units display different contents in organic matter as TOC in Unit L is below 1.4%, whereas Unit T, darker in color, has higher TOC values ranging from 0.3 to 3.5%.
Core 20, taken at 53 m depth on the eastern transect, resembles basinal cores from other studies (e.g. Matzinger et al., 2007;Wagner et al., 2008;Vogel et al., 2010b). It is composed of a finer grained lithology when compared to the shallow-water cores and furthermore does not show the two lithostratigraphic units as described for the shallow-water cores. Its lithology is characterized by a lower range in carbonate content (9 to 67%) and TOC values between 1 and 2%. Color ranges from dark beige to dark grey and black diffuse laminations occur. The fine lithology is dominated by a silty grain-size contributing on average 87% of the sediment with an average range between 15 and 30 µm.

Geochemical signatures
XRF data provide downcore elemental chemical signatures, of which Ca, Fe and Si are shown in Fig. 7, as their profiles show prominent changes in sediment composition and properties. The Ca profile is clearly correlated to the measured carbonate content. As dolomite is not formed under these in situ conditions, calcite forms the dominant sediment component, which was also confirmed in SEM images. Ca counts usually are lower in Unit T compared to Unit L, coinciding with the lower carbonate content in the upper Unit. The Si and Fe curves correlate among each other and anticorrelate with Ca. They both show a significant upcore increase at the Unit L-Unit T boundary. This increase towards higher Fe and Si values in Unit T is rather sharp and coincides with the change towards darker color with the exception of core 12, where the change occurs below the boundary.

Calcite particles
Idiomorphic calcite crystals and their clusters vary in length between 20 and 100 µm (Fig. 5c, d). As in trap samples, many ∼1 µm large holes occur in the calcite crystals (Fig. 5c). They are very often covered by an organic coating. Large clusters were found in most samples from Unit T, but are very rare and smaller in Unit L samples. In core 20, the deep water core, the idiomorphic crystals are smaller than 20 µm. In most samples from Unit T, 20 to 100 µm large calcite crystals with irregular edges were found, most probably particles that were transported by downslope currents. Indications for such downslope transport processes were also found on sidescan sonar images. This kind of calcite crystal was very rare in Unit L samples of the spring-zone cores but occurred frequently in the spring-distal cores at the edges of the transect.  6. Photographs of the cores from the northern and southern transects, as well as the subaquatic spring zone (middle). For location see corresponding core number in Fig. 2. Color bars to the right of core photographs indicate lithologic units T (green) and L (blue). Insert map shows location of displayed cores with red dots along the three transects (Fig. 2). Core photographs of the cores from the eastern transect are shown in Fig. 7.   Fig. 7. Geochemical profiles (carbonate content, total organic carbon (TOC), XRF-counts of Si, Ca, and Fe) of cores 13, 12, 05, 17 and 20 together with core photographs. Green shades indicate Unit T. Note geochemical changes across Unit L-Unit T boundary. Unit assignment in core 12 is not straightforward as indicated by a question mark. Insert map shows location of displayed cores with red dots along the three transects (Fig. 2).  (Table 3). Unit assignment in core 12 is not straightforward as indicated by a question mark.

Core correlations
The lithologies of the cores can be correlated along the various transects (Figs. 6, 7). Cores 15 and 16 (in the South) as well as cores 10, 12, 13, and 49 (in the North) are from locations in shallow water that are more than 200 m away from the spring zone; they all show similar downcore sequence with a thin Unit T and an underlying Unit L containing mostly two coarse layers (Fig. 6).
Cores 08 and 14, taken at 100 m distance from the spring area, are somehow transitional between the more distal cores and the core from the spring zone ( Fig. 6; Unit L was not reached in those cores). They have in the uppermost parts a similar lithology as the cores from within the spring area, but their lower part of Unit T differs from both the spring zone cores and the other more distant cores as it is coarser and with a lower carbonate content.
Cores 17 and 20 represent the eastern transect towards the lake center and greater water depths. Their lithologies show major changes compared to all shallow-water cores. In the eastern transect, Units L and T can only be defined to core 17 (Fig. 7). Core 20 from a water depth of 50 m shows no clear boundary, however, the TOC curve also indicates an upcore transition to organic-rich lithologies (Fig. 7). The cores within the spring zone (cores 01, 02, 03, 05, 46) show the general pattern of the two lithologic units. Unit T can be much thicker than outside the spring zone (Fig. 6), but lithologies are more irregular and thickness variations among the cores are larger when compared to the cores outside the zone that show a more regular succession.

Age control of the cores
137 Cs activities were measured partly at 5 mm intervals in the upper parts of five cores (05, 12, 13, 17 and 20). The measured downcore curves (Fig. 8) do not display clearly the two expected peaks in radionuclide activity, which would correspond to the climax in atmospheric nuclear bomb testing in AD 1963 and the Chernobyl accident in AD 1986 (Appleby, 2001). However, a clear downcore drop in 137 Cs values to basically zero occurs systematically at the Unit L-Unit T boundary, indicating that the maximum age of this boundary coincides with the onset of nuclear bomb testing in the 1955s (Appleby, 2001). The absence of two successive pulses in 137 Cs activity can be a function of bioturbation and/or current and storm actions that prevent the shallow-water sedimentary succession to represent a continuous period. 210 Pb values varied significantly, lacking clear downcore trends, so that they were not used to modify the age models.
Three organic macro remains of terrestrial origin from three different cores were used for radiocarbon analysis (Table 3, Fig. 6). Resulting calibrated ages (2-sigma) range from 3487 ± 74 cal. yr BP at 21 cm in core 12, to 3532 ± 88 cal. yr BP at 73 cm depth in core 03, and to 6350 ± 51 cal. yr BP at 46 cm in core 13. These ages, all taken within Unit L, indicate either reworking of particles in Unit L sediments, or a very dynamic sedimentary environment with transport/erosion phases also in the pre-1955's section. These are probably caused by currents and/or storm activity, which both are very likely to occur for the shallow-water zone in Lake Ohrid.
The Unit L-Unit T boundary marks thus a chronostratigraphic sequence boundary separating pre-1955 AD sediments (Unit L) from post-1955 AD sediments (Unit T). Within both units, however, we expect incomplete stratigraphy as neither the 137 Cs and 210 Pb profiles, nor the 'old' radiocarbon ages indicate regular and conformable sedimentation so that both units cannot be stratigraphically subdivided.

Carbonate precipitation in Kališta spring area
Seasonal biologically induced calcite precipitation is a common effect in low to mid-latitudes and has already been observed in the deep basin of Lake Ohrid with a previous sediment-trap study (Matzinger et al., 2006b), which established that carbonate precipitation occurs mostly during summer. Our study documents that sediment trap material from the coastal zones is characterized by higher TIC values (between 8.5 and 10.4%) compared to those measured in the deep water in the open basin (mean value of 7.8% in summer period at 50 m water depth; Matzinger et al., 2006). These patterns are supported by (i) higher Ca2 + ion contents in the shallow water spring zone (21-34 mg l −1 ) when compared to the averaged lake-wide values (13 mg l −1 ) and also by (ii) lower sedimentation rates in the deep basin (∼0.9 mm yr −1 Matzinger et al 2006b). Assuming that no or only little carbonate dissolution occurred during the Holocene in the water column in deeper water, unlike during the Glacial epoch (Vogel et al., 2010b), the sediments deposited in the shallow Kališta site are less affected by dilution from the noncarbonate fraction, i.e. from siliciclastic constituents or from diatoms. Furthermore, the large and characteristic idiomorphic calcite crystals and their clusters observed on SEM images (Fig. 5) confirmed the efficiency of authigenic carbonate precipitation in the spring area and indicate that carbonate dissolution is negligible at least for the period recovered. Water analysis show (Table 1) that the Kališta spring water are richer in dissolved ions and in nutrients than the waters in the central lake, so that enhanced precipitation of CaCO 3 due to photosynthetic activities of microorganisms and algae is likely. However, the constant cold water of the springs (colder than lake surface water in summer) might attenuate to some extent the precipitation of CaCO 3 .
An indication of biological enhancements in calcite precipitation are the regular ∼1 µm diameter holes that are observed in SEM images of the collected calcite crystals (Fig. 5b, c). Similar holes were identified as being related to the effect of picocyanobacteria . According to studies conducted in oligotrophic lakes Dittrich and Obst, 2004), picocyanobacteria (or picoplankton) with cell size of 0.2 to 2 µm act as nuclei for the formation of calcite crystals. The crystals nucleate and grow around organic matrices reaching out from the cell membrane. Eventually, holes are left when the organisms dissolve. This assumption is supported by the fluorescence microscope analysis of a water sample collected on 29 September 2007 at 1 m water depth over the spring zone, which resulted in an average concentration of 824 × 10 3 picocyanobacteria cells ml −1 . This cell count is very high, as for a comparison only ∼100 to 300 × 10 3 cells ml −1 were counted in the epilimnion of Lake Lucerne during August and September, respectively . Similar holes in calcite crystals were also observed in authigenic precipitates of a Lake Ohrid core from the deep basin (core Lz1084; Wagner et al., 2008) as shown by SEM images of bulk sediment, indicating similar processes also occurring in the open water even though at lower intensities.
The presence and shape of large authigenic particles in most samples from Unit T suggest that the particles are deposited in the sediment without major dissolution. The frequent occurrence of calcite particles with irregular edges also documents that some transport occurs in the shallow-water area, for instance during storms. These particles are however less frequent or absent in Unit L from the spring-area cores, suggesting that immediate authigenic calcite precipitation is in the active spring area more significant than reworking and redistribution of previously precipitated crystals.

Comparison with basinal cores and extent of the spring's influence
Assuming an age of post 1955 AD for all Unit L-Unit T transitions encountered in the cores, allows to compare relative sedimentation rates for Unit T along the transects (Figs. 2,  6). It has however to be considered, that the unit boundary does not necessarily represent a fixed time (e.g. 1955) but is rather a dynamic boundary that is influenced by reworking processes. As the area lies in the range of the wave base for most extreme storm events, sediment reworking by waves and long-scale shore currents must be considered for both lithostratigraphic units. Consequently, sedimentation within Unit T likely reflects the interplay between particle deposition (i.e. production through authigenic precipitation) and particle removal, as the depositional environment is shallow and highly dynamic.
Comparing the thickness of Unit T reveals large variations in between the cores from the spring, the littoral zone, and the deeper water (Figs. 6, 8). In the spring zone, thickness of Unit T amounts to more than 10 cm and reaches its maximum with 28 cm in core 05. Only core 02 is characterized by the absence of Unit T sediments, which could be caused by an extreme proximity to an active spring outlet that physically prevents sedimentation through strong currents. In a distance more than 300 m away from the springs and also towards deeper water, Unit T is always thinner than 10 cm (Figs. 6, 7), reflecting the increased sedimentation rate in vicinity to the springs. All these values document significantly enhanced apparent sedimentation rates in the active spring area, also  03,13,16,17,20,49 from this study with data from basinal cores (OHR02-1, OHR03-1, OHR03-2 and Lz1083 to Lz1086; Matzinger et al., 2006a) and of core OH05-15 recovered in 35 m water depth in the southern littoral basin, at ∼1 km distance from the shore. The two minima of core OH07-03 correspond to coarser sediment lithologies. Note the general trend of upcore decreasing carbonate values in most cores independent on their water depth. when compared to deeper water areas but also when compared to other non-spring shallow water zones. Even if no absolute sedimentation rates should be calculated, due to the non-continuous episodic nature of deposition, the clear trend with thicker Unit T sediments in the spring zone points towards proximity of an active particle source, i.e. spring-water enhanced carbonate authigenesis. Figure 9 compares the % carbonate downcore trends in a series of cores, including those from this study and those from Matzinger et al. (2006b), which were collected in the deeper basin and towards the southern shore. It can be recognized that the carbonate content in the four cores within or close to the active spring zone is 2 to 5-times higher than the values in basinal cores. Core 20, from the eastern transect forms the transition between the shallow water spring signature and the basinal carbonate contents. The pattern observed in core 20 matches well the general pattern of basinal cores, i.e. that, unlike in the shallow-water cores, now sharp lithologic change can be observed (Matzinger et al., 2006b). Two effects may explain the lower carbonate content in the basin: the first effect is the proximity to the shore, as surface water temperature and higher biologic activity favor carbonate precipitation. The second effect is the mineral-rich water input from sub-aquatic and littoral springs. Although no quantitative discharge values are available, the relatively large area with numerous sub-aquatic springs is likely to have an effect on local water chemistry.
The carbonate-increasing effects of the Ca 2+ and HCO − 3rich spring waters seems, however, to be rather local, as a lake wide sediment surface survey in Lake Ohrid shows a general trend towards higher TIC values in the northern, northeastern, and southwestern parts of the lake basin coin-ciding with areas with enhanced productivity (Vogel et al., 2010a). This study also documents the difference in carbonate content of the surface sediments in the vicinity of river inflows with high and low Ca 2+ and HCO − 3 contents from carbonate or non-carbonate catchments, resulting in strong or weak rates of carbonate precipitation, respectively (Vogel et al, 2010a).
The similarity in carbonate signatures of core 20 (50 m water depth in front of springs) and core OH05-15 (35 m depth in the southern basin lacking any spring influence) indicates that the springs have hardly any direct influence on the sedimentation at greater water depths. This can be explained by the existence of counter-clockwise surface currents, which apparently transport local influxes mainly parallel to the shore (Vogel et al., 2010a). The gradual reduction in thickness of Unit T sediments at larger distances from the springs is likely an indication for such a greater springinfluence on sedimentation along the northern and the southern transects. Part of the increased thickness of Unit T in the spring area, however, may also be caused by the bathymetric depression of the main spring zone offering more accomodation space. However, elevated thicknesses are also detectable in the entire spring area, which is not characterized by greater water depths when compared to the spring-distant coring sites where Unit T is thinnest.

Paleoenvironmental history
The transition from Unit L to Unit T observed in all cores, except core 20, is an indicator for major change in sedimentary conditions that likely occurred after AD 1955. This change is expressed with a darker sediment color, an increase in TOC, a slight decrease in carbonate, an increase in Fe and Si content as well as in diatom content, and a sewage smell upon core opening. The lower carbonate content in the few coarse layers in Unit L is interpreted as a supply also from noncarbonate detrital material, that originates from strong runoff events. The prominent lithologic change across the Unit T-Unit L boundary is likely in concert with a significant upcore increase of anthropogenically-induced nutrients that affect the sedimentary conditions, as it has been observed in sediment cores of many eutrophied lakes. In particular TOC is an indicator for elevated productivity, which is also confirmed by sudden appearance of large amounts of diatoms in Unit T sediments and by concurrent elevated Si counts. The rotten smell of Unit T sediments also matches water samples in the Kališta spring zone during 2006 and 2007 that indicated high values of E. coli and phosphomineralizing bacteria, which are indicators for anthropogenic waste waters entering the lake. In fact, expanding touristic activities in the area during the last decades and the construction of various lakefront resorts increased the potential of pollution through untreated waters (pers. comm. Lence Lokoska, HBI). The general increase in Fe in Unit T is difficult to interpret as Fe is in Lake Ohrid not only of terrigeneous but also of diagenetic origin, as concretions may occur, in particular in lithologies with high TOC (Wagner et al., 2010;Vogel et al., 2010a).
The general upcore decrease in carbonate content is not only observed in the cores from this study, but can also be seen in the cores from the basin and the southern shore (Matzinger et al., 2006b), which show also a similar pattern of reduced carbonate values towards the core top (Fig. 9). These patterns indicate a lake wide change in sedimentation patterns related to the same forcing mechanisms.
To which degree sedimentation processes have been influenced by a lake level regulation in 1962 (Popovska and Bonacci, 2007) remains unclear. At that time, the River Sateska, previously a direct tributary of the River Crn Drim, was deviated into Lake Ohrid and a gate on the outflow River Crn Drim was constructed in Struga (Popovska and Bonacci, 2007). This increased the lake inflow by 5.5 m 3 s −1 (∼70%), however, the resulting nutrient and sediment contributions is considered to be small (Matzinger et al., 2007) and most likely not directly responsible for the observed eutrophication.

Conclusions
This study investigates and quantifies the effect of the Kališta subaquatic springs on coastal sedimentation in Lake Ohrid. The high accumulation rates of CaCO 3 within the spring zone and the large idiomorphic calcite crystals and clusters found in the sediment trap material and cores indicate that the ion-rich spring water entering the lake water at the Kališta subaquatic springs significantly enhances local carbonate precipitation. These springs provide special crystal-lization conditions in the spring area so that sedimentation rates are higher in the vicinity of the springs. Picocyanobacteria are assumed to play a crucial role in the crystallization process, as typical holes were found in the crystals and high cell concentrations were observed in the water above the springs. The different morphologies of the carbonate particles analyzed at different depths in the cores offered further clues to identify the origin of the calcite. The amount and size of authigenic calcite crystals found in cores within 200 to 300 m north and south (shore parallel) to the spring area showed the possible extent of the spring's waters influence.
Two distinct lithostratigraphic units revealed a drastic change in coastal sedimentation that occurred after AD∼1955. The increase of anthropogenic nutrient input might explain the enhanced biomass production as witnessed by an increased diatom content and elevated Si counts as well as by slightly reduced (diluted) carbonate contents, dark color and sewage smell of the freshly-opened cores. Even though similar changes also occur in deeper waters, Lake Ohrid might show a faster and more drastic response to eutrophication and human impact in its shallow waters.