Late Glacial and Holocene Paleoenvironmental Reconstruction of the Submerged Karst Basin Pirovac Bay on the Eastern Adriatic Coast

Abstract

This study focuses on the analysis of sediment core retrieved from the deepest part (25 m) of Pirovac Bay. A long sedimentary sequence (7.45 m) supplemented by a shorter sediment core (1.45 m) from a shallower part of the bay was analyzed for sedimentological, mineralogical, geochemical, and micropaleontological (ostracod) parameters. The sediment thickness above the underlying karst paleorelief (karstic bedrock) is up to 12 m. Sediments recorded a transition from a freshwater to a marine environment starting from post-Neapolitan Yellow Tuff tephra sedimentation. First, the floodplain developed in Pirovac Bay, with intermittent pools and ponds, followed by wetland environment. The formation of a shallow freshwater paleolake during the Middle Holocene at 10 cal kyr BP was enabled by the rising sea level and high freshwater input from the karstified underground from the adjacent Lake Vrana (Biograd na Moru). The onset of marine intrusions through the karstified underground is evident with formation of a brackish lake in the Pirovac Bay basin. Marine transgression and flooding of the bay occurred at 7.3 cal kyr BP, evidenced by the geochemical and ostracod parameters, providing crucial insights into the dynamics of coastal inundation under past climate change. Intriguingly, freshwater ostracod species were still present in the marine sediments, brought into the bay from Lake Vrana through surficial canal Prosika and groundwater discharge (numerous estavelles) along the northeastern shores of the bay, proving their mutual influence. This submerged Holocene freshwater paleolake, reported here for the first time, underlines the sensitivity of coastal karst systems to the rise in sea level and serves to stress how important understanding of these processes is for effective management in coastal zone and climate change adaptation strategies. The findings provided evidence supporting the existence of coastal marine basins as freshwater lakes prior to being flooded by seawater as a consequence of the Holocene post-glacial sea level rise.

1. Introduction

The Eastern Adriatic Coast is highly indented and comprises numerous islands, bays, and coves, with restricted, shallow marine depositional environments, i.e., the Dalmatian type of coast [1]. The Croatian Dinaric karst region on the Eastern Adriatic Coast is characterized by various karstic forms, including karst basins (dolines and poljes), as well as submerged counterparts [2,3]. Karst basins often contain a thick succession of sediments. Pirovac Bay is an example of such a type of environment. It is a relatively large submerged shallow karst basin located in the central part of the Eastern Adriatic Shelf.

Coastal marine sediments have strong similarities to lake sediments, as they accumulate in relatively shallow water, often receive large proportions of land-derived organic matter, and are deposited relatively rapidly compared to deeper marine sediments. Coastline changes during the Late Glacial and Holocene are related to the global sea level rise [4,5]. Sea level changes caused the formation of different depositional environments in isolated basins with sills during the Late Pleistocene and Holocene [6]. Similar silled environments have been described in the Mediterranean region as well [7,8]. Nevertheless, the karst geomorphological system behaves differently than other areas in the dominant role of dissolution or carbonate rocks, which results in developed underground circulation rather than surface channels [9]. Karst landscapes are prone to sediment run-off and erosional processes—that is, a lack of preserved sediment successions. The combination of these factors strongly affected past environmental changes in the eastern part of the Adriatic Sea. The water circulation in porous karst aquifers has an important role in sea-water intrusion and transgression [10]. In coastal zones, underground mixing of fresh groundwater and seawater occurs, as evidenced by the hydrogeological research performed on the Croatian Islands [11,12].

The Northern and Central Adriatic Late Pleistocene-Holocene alluvial plain was progressively drowned due to the sea level rise [13,14,15]. Submerged landscapes that existed along the karstified Eastern Adriatic include caves [16,17], lakes and karst poljes formed in isolation basins with sills [18,19,20,21,22,23], and river canyons [24,25]. Coastal karst inundation reconstruction using submerged caves along the Eastern Adriatic Coast is provided during the Late Pleistocene and Holocene [16,17]. The freshwater lakes along the Eastern Adriatic Coast were recorded only during the MIS3 period [19,20], while, due to the post-glacial sea level rise, Holocene lakes were exclusively brackish [21,22,23]. Marine intrusion in coastal areas through karst underground was recorded on the Islands of Cres [18] and Pag [26], while marine influence was also observed in karst polje in the Dalmatian hinterland [27] and the river estuary in Istria [28]. The submerged river canyons of Zrmanja and Krka River estuaries and Prokljan Lake were flooded during the Early to Middle Holocene [24,25]. Thus, the coastal sediment archives present along the Eastern Adriatic can provide a reconstruction of past climate and sea level variability in a karst setting [21,26,29]. Though there have been many studies that focused on the Holocene sea level rise in different kinds of coastal settings, relatively less is known regarding the exact implications on submerged karst basins, especially the interaction between freshwater and saltwater intrusion and its long-term ecological implications.

Pirovac Bay is a silled karst basin (i.e., isolation basin) filled with Quaternary sediments, which compelled us to investigate its sediment succession to reconstruct the evolution of the bay. The geomorphic setting (i.e., the existence of karst depressions) was investigated using acoustic methods. The key constituents of the sediments in the karstic area are carbonate versus siliciclastic and organic material, which are used to differentiate depositional environments in Pirovac Bay. The concentration of total organic carbon (TOC) represents the amount of organic matter (OM) preserved after sedimentation and therefore depends both on the initial production and the degree of degradation [30]. The isotopic composition and C/N ratios of sedimentary organic matter differ between marine and terrestrial sources [31,32]. Terrestrial organic matter generally exhibits depleted δ¹³C and δ¹⁵N values compared to marine organic matter. This proxy is particularly valuable in isolation basins for interpreting marine transgression and basin isolation associated with changes in the relative sea level. Calcitic microfossils, such as ostracods, have a long fossil record because of the preservation of their calcite carapace in sedimentary deposits. Fossil and subfossil ostracods play important roles in paleoclimate and paleoenvironmental reconstructions in the presence or absence of specific taxa [33,34,35]. Thus, our multiproxy study (sedimentology, mineralogy, elemental geochemistry, organic carbon and its stable isotopes, and microfossil analysis of ostracods) aims to reconstruct the paleoenvironmental history preserved in Pirovac Bay, a submerged karst basin situated on the Eastern Adriatic Coast, as it would offer advanced abilities for the determination and prediction of such systems with regard to sea level rise in the future to achieve effective protection and natural resource management of coasts and similar environments that are easily vulnerable.

2. Study Site

Pirovac Bay is one of the numerous bays along the indented Eastern Adriatic Coast, located in Central Dalmatia, Croatia. The Eastern Adriatic Coast belongs to the Adriatic carbonate platform, one of the largest Mesozoic carbonate platforms of the Mediterranean [36,37], which covers more than 40% of the Croatian territory and is known as the Dinaric karst. Pirovac Bay is a small, semi-enclosed bay (~12 km long, max 2.2 km wide) with a maximum depth of 25 m in its central part, elongated in the Dinaric direction (NW–SE) and parallel to the Adriatic coastline (Figure 1). It is surrounded by the mainland on the northeastern side and a few islands on the northwestern side, the largest being Murter in the southwestern part of the bay. The southeastern part of Pirovac Bay is known as Makirina Cove, a shallow prolongation of the bay up to 2 m deep, previously investigated for sediment thickness and geochemical characterization of organic-rich sediments [38,39]. Makirina Cove was a salt pan site during Roman and Medieval times [40]. The total surface catchment area is 90.8 km². The largest settlement is the Town of Pirovac, located on the northeastern side of the bay, and a smaller settlement is the Town of Murter on the Island of Murter on the southwestern side of the bay.

Pirovac Bay is mostly surrounded by Cretaceous carbonate rocks, which include dolomites and limestones (Figure 1). The wider area of Pirovac Bay is mostly composed of Early and Late Cretaceous carbonate rocks [41,42]. A detailed geological map of the region surrounding Makirina Cove on the scale of 1:10,000 was made previously [38]. The oldest lithostratigraphic units in the area are Lower Cretaceous dolomites (K_1,2; Albian-Cenomanian in age, Ivinj dolomites), composed of dolomites, limestones, and dolomitic breccia, and they are the core of the anticline structures of the wider Makirina Cove. Towards the NW, these dolomites dip beneath the sea in Pirovac Bay [38]. The eastern coast of the bay is composed of Upper Cretaceous limestones and dolomites (K₂^1,2; Cenomanian-Turonian, Makirina limestones and dolomites). The Upper Cretaceous rudist limestone unit (K₂³; Early Senonian, Kamena rudist limestones) is present continuously in the SW part of Pirovac Bay and inland, in the Lake Vrana catchment [38]. The Cretaceous rocks are transgressively overlain by Paleogene deposits, which consist of foraminiferal limestone of the Lower and Middle Eocene ages (E_1,2). They gradually transition into the flysch unit (E_2,3), which contains marl layers alternating with thin layers of siliciclastic sandstones of the Middle Eocene age [38]. Quaternary deposits (terra rossa, colluvial–deluvial deposits) accumulate in the coastal area in the SE part of the bay (Ivinj draga). The smaller karst depressions and sinkholes were filled with terra rossa soil [39]. These landforms have been converted to agricultural land, covered with mostly vineyards and olive groves. The thickness of the Quaternary sediments of Makirina Cove, being accumulated on a dolomite basement, gradually increases (0–3.5 m) from the coast towards the central part of the cove [38]. In this shallow SE part of Pirovac Bay, organic-rich sediments accumulated, which are among the largest accumulations of peloids in Croatia [38,39]. The climate in the region is a typical Mediterranean type [39]. The winters are mild, rainy, and windy, and the summers are warm and dry, with high insulation. The mean annual temperature was 16 °C, in January, 6–8 °C, and 25–26 °C in July, and the average annual precipitation was 770 mm (1971–2000), with the maximum rainfall rates recorded during the fall [39].

Lake Vrana near Biograd na Moru, the largest freshwater lake in Croatia with over 30 km² of surface area, is in the vicinity of Pirovac Bay (Figure 1). It is very shallow, with a depth between 2 m and 4 m, depending on the hydrological season. This leads to frequent changes in the lake level in the range 0.03–2.25 m above sea level (asl), while the average water level is 0.82 m asl [43,44]. Lake Vrana is a cryptodepression, with the deepest part of the lake at 3.5 m below sea level (bsl) [45]. The intermittent submarine spring (vrulja) appear along the northeastern shore of the lake [43]. The lake and the bay are connected by the artificial canal Prosika (800 m long, 4 m wide and 8 m deep) in the southeastern part of the lake. It was constructed in the 18th century, as part of the first amelioration works, to lower the lake level and reduce flooding for agricultural development in Vrana Polje [44]. During the wet seasons and high lake level in Lake Vrana, the water discharges into the bay via canal Prosika [43,44,45,46]. Regardless of the present surface water connection, they are connected through well-developed karst underground waterflows and karst conduits through a narrow karstified carbonate ridge (Cretaceous limestones) (800–2500 m wide [43]), which proves their interconnection today and in the past. Based on the well-known hydraulic and hydrogeological mechanism of marine intrusion through karst conduits and fractures in Croatian karst islands [11,12,47], brackish springs and the direction of seawater intrusion can be tracked on the shores of Lake Vrana and Pirovac Bay and its surrounding catchment [39,43]. Along the southeastern shore of Lake Vrana and the northeastern shore of Pirovac Bay, the intermittent coastal freshwater and brackish springs appear. Freshwater springs act as estavelles [39,43], with a dual function depending on the groundwater conditions. It acts as a coastal spring and discharges freshwater during the wet season and high lake levels in Lake Vrana and Pirovac Bay or as a ponor (sink or swallow hole) during the dry season and low lake levels in Lake Vrana and Pirovac Bay [39,43,44,45,46].

3. Materials and Methods

3.1. Study Area Description

Pirovac Bay, located in coastal area of the Eastern Adriatic Sea, is a relatively shallow bay (Figure 1), prone to influences of the Quaternary changes in sea level. Studies along the Eastern Adriatic Coast (Figure 1) point out the vulnerability of coastal karst systems and Pirovac Bay, among other sites, represents a solid basis for research of paleoenvironmental and paleoclimate changes. Its proximity and connection to the protected Lake Vrana (Biograd Na Moru), an ornithological reserve, makes it important site to study past changes to predict possible future implication on freshwater ecosystem and biodiversity. Geology of the Dalmatian-type of coast, where the study site is located, served as a unique factor for paleohydrology and the sea level rise that can be tracked in the sediments in the investigated basins.

3.2. Hydroacoustic Survey

A dense grid of high-resolution seismic profiles throughout Pirovac Bay was acquired to obtain detailed information about submarine geomorphology and sediment distribution/thickness throughout the basin. The hydroacoustic survey was performed in June 2015 using a parametric subbottom profiler system Innomar SES-2000 light plus at both high (100 kHz) and low (6 and 8 kHz) frequencies, which allowed good resolution, as well as penetration up to 12 m into the subsurface. A survey was conducted on a small vessel using a Differential Global Positioning System (DGPS) navigation. The data were processed using Innomar ISE post-processing software.

3.3. Sediment Cores Collection

In 2011, a 7.45-m-long sediment core PROSIKA-1 was recovered from the central part of the bay at its deepest part (25 m depth) (WGS coordinates: 15°36′59.126′′ E, 43°50′8.015′′ N; Figure 1B) using a piston corer from a platform (UWITEC, Austria). The core diameter was 60 mm, whereas the length of each core section was 3 m. Core catcher (cc) samples were taken at the end of each section at 5- and 10-cm intervals and packed into plastic bags. The complete PROSIKA-1 sediment record was composed of consecutive 3-m-long sections cored at one location. The parallel core (PROSIKA-1A) was taken a few meters away and was used to supplement the main sediment core starting from 1.5 m below the seafloor (bsf). Sediment core ARTA-1, with an overall core length of 1.45 m (1.33 m sediment core plus 12 cm cc samples in 8 cm and 4 cm intervals), was recovered in a small and shallow depression in the northwestern part of the bay (WGS coordinates: 15°34′24.735′′ E, 43°51′57.692′′ N; Figure 1B) at a depth of 13 m bsl southeast of the Island of Arta Mala.

3.4. Laboratory Analyses

In the laboratory of the Croatian Geological Survey, the cores were split into two halves. After splitting the cores, magnetic susceptibility (MS) using a Bartington (Bartington Instruments Ltd., Oxon, United Kingdom) point surface sensor (MS2E; volumetric low-field MS–κ_LF) and color reflectance analysis using a X-RITE DTP-22 Color Digital Swatchbook spectrophotometer in L*a*b* color space were measured at 1 cm resolution on a wet sediment, covered with thin transparent plastic film to avoid sensor contamination.

Both halves of the cores were sliced in 1-cm-thick samples, and the working half was sliced and dried at 40 °C in a dryer, while the wet archived half was used for subsampling of 1 cm³ volume of the sediment at 5-cm resolution in core ARTA-1 and 10-cm resolution in core PROSIKA-1 to determine density and the water content. The samples were then weighed and dried overnight at 105 °C. The dry samples were weighed again to calculate the dry bulk density (DBD). Wet sediments were also used for ostracod analysis. Sediment color was determined using the Munsell color chart. The selected dry samples were used for grain size analysis and milled into a fine powder using a mortar and pestle for geochemical and mineralogical analyses.

The grain size analysis was performed using the laser diffractometer Shimadzu Sald-2000 (Shimadzu Crorporation, Kyoto, Japan) in Croatian Geological Survey in the range of particle sizes between 0.17 µm and 2 mm. A total of 83 samples from core PROSIKA-1 and 26 samples from core ARTA-1 were analyzed. The samples (0.2 g) were pretreated with 30% H₂O₂ on a hot block at 80 °C to remove organic matter. Sodium hexametaphosphate (Na₄P₂O₇) was added (2 mL) as a dispersant; then, the samples were sonicated before being analyzed. The result for every sample was the average of three measurements. The data were processed using the software GRADISTAT v8 [48]. Samples were not treated to dissolve the carbonate component considering that, due to a karst environment surroundings, carbonates are an important factor in the paleoenvironment assessment.

The mineral composition was determined on 20 samples from the sediment core PROSIKA-1 and 21 samples from sediment core ARTA-1 using X-ray powder diffraction (XRD) on a PANalytical X’Pert Powder X-ray diffractometer (Malvern Panalytical, Almelo, the Netherlands) using Ni-filtered CuKα radiation in Croatian Geological Survey. Mineral phases were identified using HighScore X’Pert Plus and the ICDD (International Centre for Diffraction Data) database (PDF-4/Minerals). The powder samples were pressed into a back-loading holder and were run using a spinner stage with the following parameters: 4–66°2θ, 0.02° step size, 45 kV, 40 mA, 0.04 rad Soller slits, 10 mm mask, 1/4° incident divergence slit, and anti-scatter slits. Clay minerals were identified after the removal of carbonates using a buffered sodium acetate solution (NaOAc). The clay-sized sediment suspensions were centrifuged and transferred to glass slides and left to settle with the preferred orientation. The samples were treated with a series of diagnostic tests: air-dried (AD), ethylene-glycolated (EG; overnight), heated to 400 °C and 550 °C for at least half an hour, treated with dimethyl sulfoxide (DMSO), heated to 350 °C of K-saturated samples for an hour, and treating with glycerol the Mg-saturated samples. Subsequently, XRD measurements were performed after each treatment in the range of 4–30° 2θ for 7 samples from both cores, PROSIKA-1 and ARTA-1. The samples were taken in 5-cm intervals for every sample to ensure enough material after carbonate removal. The identification of clay minerals was done based on XRD patterns from the literature [49,50].

The geochemical composition was determined on 83 samples from core PROSIKA-1, taken at 5-cm intervals from 0 to 200 cm of the core and every 10-cm interval from 200 cm of the core up to 730 cm. A total of 31 samples from core ARTA-1 were analyzed at a resolution of 5 cm. Geochemical analyses were performed as quantitative elemental assays of forty-five major and trace elements using inductively coupled plasma–mass spectrometry (ICP-MS) and inductively coupled plasma–emission spectrometry (ICP-ES) at the commercial laboratory of Bureau Veritas in Vancouver (Canada). The samples (1 g) were dissolved in multiple acids (HCl-HNO₃-HClO₄-HF) at 200 °C. The reference standard materials used were STD DS11, STD OREAS262, STD BVGEO01, STD OREAS25A-4A, and STD OREAS45H.

Total carbon (TC), total nitrogen (TN), total organic carbon (TOC), and total inorganic carbon (TIC) were analyzed using a CN elemental analyzer Thermo Fisher Flash 2000 (Thermo Fisher Scientific, Waltham, MA, USA) in Croatian Geological Survey. In total, 96 samples from core PROSIKA-1 and in 33 samples from core ARTA-1 were analyzed. Total carbon and nitrogen were measured directly, whereas total organic carbon was measured after the removal of carbonates using 4 M HCl and heating for 2 h. The total inorganic carbon was calculated as the difference between the TC and TOC concentrations. The C/N ratio was calculated from the TOC and TN concentrations and given as atom/atom ratios. Carbonate-free samples were weighed in tin capsules to obtain stable isotopes of carbon and nitrogen (δ¹³C_org and δ¹⁵N_org). The analyses were performed in the Stable Isotope Facility (SIF) at the University of California (Davis, CA, USA) using an isotope ratio mass spectrometer (IRMS). The final delta values were expressed relative to the international standard Vienna Pee Dee Belemnite (VPDB) and air for carbon and nitrogen, respectively.

For the ostracod analyses, subsamples averaging 13 g (dry weight) in 2 cm intervals were taken successively from the sediment cores PROSIKA-1 and ARTA-1. Two core-catcher samples were taken at the base of core ARTA-1 (8 and 4 cm intervals). Ostracod valves were recovered from 43 sediment samples from core PROSIKA-1 and from 11 samples from core ARTA-1. The sediment was soaked in water for 48 h; sieved through a mesh, washed through sieves (˃0.9; 0.466, 0.263, 0.122, and 0.063 mm); transferred to glass boards; and dried at 60 °C. Ostracods and other remains (foraminifera, gastropods, bivalves, otoliths, plant fragments/seeds, and characean gyrogonites) were hand-picked under a Zeiss stereomicroscope and stored dry on micropaleontological slides. For ostracods, the size fraction > 122 μm was qualitatively analyzed under stereo binoculars at magnifications between 10× and 80×. Valves and carapaces were counted separately, so that each carapace was counted as two valves. Valves of juveniles and valves damaged beyond recognition were excluded from counting and analysis, except damaged adult valves of large species, which are usually saved with broken valves. The identification of ostracods followed the identification keys and taxonomic reports from the literature [51,52,53,54,55,56]. An ostracod percentage diagram was generated using the absolute abundance of all species. Scanning electron microscope (SEM) microphotographs were obtained using gold-coated ostracod samples in a JEOL JSM-35CF system in Croatian Geological Survey.

In total, nine AMS radiocarbon measurements and one correlated cryptotephra from the two sediment cores provided ages for the Pirovac Bay sediment cores (

Table 1. Results of the radiocarbon dating of the sediment cores PROSIKA-1 and ARTA-1 from Pirovac Bay (GdA—Silesian University, Gliwice, Poland; Beta-No.—Beta Analytic Inc., Miami, FL, USA; ND—not defined).

Sample (cm)	Corrected Depth (cm)	Lab. No.	Material	δ13C (‰)	Radiocarbon Age (14C BP)	Calendar Age–Calibrated (cal yr BP)
PROSIKA-1
185–186	185–186	Beta-375528	Marine shell	+2.8	5170 ± 40	5300
264–265	248–249	Beta-328263	Marine shell	+0.3	6590 ± 40	6900
279–280	rejected	Beta-375529	Marine shell	−0.3	6470 ± 30	/
470–471	470–471	Beta-347640	Shell	−5.0	9550 ± 40	9500
627–628	627–628	Beta-375530	Wood	−31.1	9860 ± 40	11,600
741–754	722–735; not used	Beta-343644	Shell fragments	−9.7	11,960 ± 50	/
ARTA-1
22–23	rejected	GdA-4356	Marine shell	ND	7818 ± 36	/
56–57	56–57	GdA-4357	Marine shell	ND	6377 ± 29	6600
68–69	68–69	GdA-4358	Plant	ND	7712 ± 39	8000

). The last sample for radiocarbon dating was collected from the sediment core PROSIKA-1 722 at a 735 cm interval due to the lack of suitable material (shell fragments). Accelerator mass spectrometry radiocarbon dating was performed at Beta Analytic Inc. (Miami, FL, USA) and in the Radiocarbon Laboratory, Institute of Physics, Centre for Science and Education, Silesian University of Technology (Gliwice, Poland). The calibration curve Marine20 and associated marine reservoir age correction [57] were used for the marine shell samples, while, for the lower section of the core PROSIKA-1 (shells and wood), the IntCal20 calibration curve was used [58]. Limitations could arise from the freshwater reservoir effect, which was not determined in the lower part of the sediment core PROSIKA-1, where the shell fragments were analyzed (sample 470–471 cm).

3.5. Statistical Methods

Statistical analyses, including the correlation analysis, were performed using STATISTICA7 [59] to assess the significance of differences between various parameters and to identify trends in the data. All the diagrams for the sedimentological, geochemical, and ostracod parameters were constructed using C2 software (version 1.7.7) [60]. The age depth of the PROSIKA-1 and ARTA-1 sediment cores was modeled using the package rBacon in the R environment [61].

4. Results

4.1. Acoustic Survey

The submerged karst geomorphology of Pirovac Bay was identified in high-resolution acoustic profiles. We differentiated karst paleorelief and the smooth geomorphology of marine and other sedimentary surfaces (Figure 1C). Acoustic data provided a detailed geomorphological setting of the investigated area and enabled the estimation of the sediment thickness throughout the bay and the recognition of different acoustic units (Figure 2A,B).

The karst bedrock of the Pirovac Bay depression is covered with up to 12 m of sediments in the central (deepest) part of the bay. Water depths in this area range from approximately 18 to 25 m. The depositional basin becomes shallower to the northwest and southeast. Southeast of the Island of Arta Mala, we recognized a shallower depression (max 13 m), which is also infilled with sediments. The water depth is in contrast to marine sediment thickness; that is, in the deepest part of the bay, the sediments are not the thickest. The acoustic survey also enabled the recognition of Arta sill (6 m bsl) and Pirovac sill (8 m bsl) (Figure 1C), which separate the Pirovac Bay depressions from the rest of the Adriatic Sea.

Furthermore, we differentiated four main acoustic units in the surveyed area (Figure 2). Acoustic Unit 1 is the lowermost unit in the sediment infill of Pirovac Bay that directly overlies the top of the karstic bedrock. It is composed of moderate-to-high amplitude subparallel reflectors. Sediment core data are unavailable for Acoustic Unit 1, and therefore, it cannot be interpreted with certainty. The top three (3) acoustic units can be correlated with the succession of an existing 7.4-m-long, dated sediment core PROSIKA-1 (Figure 2). Acoustic Unit 2 is acoustically semi-transparent with few moderate-amplitude internal reflectors. The maximum thickness of this unit (1.6 m) was estimated in the northern part of the central basin. The sediments from Acoustic Unit 3 show low-amplitude subparallel reflector configurations. Unit 3 ranges in thickness from less than 2 m to over 4 m. The topmost unit (Acoustic Unit 4) is acoustically semi-transparent (Figure 2). The maximum thickness of this unit was determined in the southern part of the central basin (5 m). The thickness of Acoustic Unit 4 decreases northward and westward, possibly as a consequence of the significant input of material from the southeastern part of the catchment area.

4.2. Sediment Cores Chronologies

The age model for core PROSIKA-1 is based on five radiocarbon ages and one precisely determined tephra correlation (Figure 3A). Two age reversals appeared, one in each sediment core (PROSIKA-1 and ARTA-1). Those dates were rejected, as indicated in Table 1. The volcanic glass shards were identified near the base of the core PROSIKA-1 at a depth of 721 cm (PROS 721), which was geochemically correlated to eruption of the Neapolitan Yellow Tuff (NYT) [62] (Figure 3C). The NYT tephra of the Campi Flegrei eruption was dated to 14.5 ± 0.4 kyr BP [63], so this tephrochronology improved the age in PROSIKA-1 sediment core. The age depth model of the core PROSIKA-1 gave a maximum age of 14.4 cal kyr BP at 7.45 m, with an averaged sedimentation rate of ca. 0.055 cm yr⁻¹. The final accumulation rate (Figure 3B) clearly demonstrated lower accumulation rates in the period between ca. 10,000 cal kyr BP and 6000 cal kyr BP.

In core ARTA-1, three datable materials were found: marine shells and one plant remain. The uppermost date showed age reversal; therefore, it was rejected from further interpretation (Table 1). The lower section of the core was not dated, due to the limited material for radiocarbon analysis, but the ages were linearly interpolated to the base of the core using a depth age model (Figure 3B). The ARTA-1 sediment core has a maximum age of 15.5 cal kyr BP at 1.45 m sediment depth, with an average sedimentation rate of ca. 0.009 cm yr⁻¹. In sediment core ARTA-1, the final accumulation rate is significantly higher compared to the core PROSIKA-1 (Figure 3E).

4.3. Lithology and Geochemistry

4.3.1. Sediment Core PROSIKA-1

Four distinct lithological zones were identified in core PROSIKA-1 (Figure 4). The lowermost Zone 1 (745–660 cm; post-NYT age–12.4 cal kyr BP) is characterized by the brighter gray color sediments and medium to coarse silts. Silt is the dominant fraction in the entire sediment core (ranges from 52.99% to 88.95%). In Zone 1, silt predominates with a median (med.) value of 73.48%, but the sand fraction appears in greater amounts compared to the rest of the core (15.31%), while the clay content is the lowest (9.52%). The concentrations of nitrogen and TOC are relatively low throughout this zone (med. 0.04% and 0.45%, respectively), but the C/N ratio is quite high (med. 14.0), with a decreasing trend towards the end of the zone. The carbon stable isotope composition is low (δ¹³C_org = −26.57‰ med.), while the nitrogen isotopes (δ¹⁵N_org) are the highest (med. 4.84‰). Magnetic susceptibility (κ_LF) is low throughout the whole sediment core (med. 0.4 × 10⁻⁵ SI), with a few peaks > 5 × 10⁻⁵ SI (Figure 4). The peak of κ_LF at 721 cm (11.7 × 10⁻⁵ SI) correlates well with geochemical data (Figure 5), with high concentrations of siliciclastic elements (Al, Fe, K, Na, Zr, and Ba) (Table S1), and was therefore examined for glass shards using smear slides. This layer was determined as cryptotephra, which was correlated to the NYT tephra of the Campi Flegrei eruption based on the geochemical data of separated glass shards [62]. In general, siliciclastic elements (Al and Fe) are higher in Zone 1, although Ca and Mg are also present in significant concentrations, indicating a dominant carbonate composition of the sediments, with calcite as the major mineral phase, followed by quartz and dolomite, and traces of muscovite/illite (Figure 5). After carbonate removal, the determined clay minerals were smectite, chlorite, illite, and kaolinite. The concentration of Mg is associated with dolomite and clay minerals (mainly chlorite) as detrital, land-derived mineral phases. Halite is present as the result of sediment drying.

The second lithozone is divided into two subzones: Subzone 2/1 (660–510 cm; 12.4–10.0 cal kyr BP) and Subzone 2/2 (510–375 cm; 10.0–8.3 cal kyr BP) (Figure 4). Silt predominates in grain size (med. 83.08%), with sand and clay that vary in the range between 6.36% and 10.03% (med. values), respectively. The peaks of magnetic susceptibility (κ_LF) appear at 642 cm (14.1 × 10⁻⁵ SI) and between 535 and 545 cm (11.4–37.2 × 10⁻⁵ SI) but without clear correlation to higher concentrations of siliciclastic elements (Figure 5); therefore, no further analysis was performed for tephra identification. Both subzones are marked by the increase of N (med. 0.13%) and TOC (med. 1.30%), whereas the TOC is slightly higher in Subzone 2/1 (med. 1.31%) (Figure 4). TIC values are also high, suggesting carbonate sedimentation, confirmed with mineral composition. Calcite and dolomite are dominant mineral phases, with minor quartz and traces of muscovite/illite in Subzone 2/1, supported by elevated concentrations of the siliciclastic elements (Figure 5). In Subzone 2/2, with the highest TIC values, calcite predominates, with minor quartz and the absence of dolomite. This is supported by the low Mg and siliciclastic elements (Al and Fe) (Table S1). There is also a difference in clay mineral composition; in Subzone 2/1, it is the same as in previous Zone 1 (smectite, chlorite, illite, and kaolinite), but in Subzone 2/2, they are present in smaller amounts after carbonate removal, and hydroxy-interlayered vermiculite, chlorite, illite, and kaolinite appear. The C/N ratios show little variation in both subzones (med. 11.8), as well as δ¹³C_org values (med. −23.99‰), after an increasing trend in the lower part of the Subzone 2/1 upwards, while the δ¹⁵N_org values are high in this part (med. 3.82‰) and decrease upwards (med. 2.23‰ in Zone 2/2), with the lowest peak at 440 cm (0.13‰) (Figure 4).

A subtle change in siliciclastic and organic matter composition allowed us to distinguish the third lithozone, which starts with Subzone 3/1 (375–320 cm; 8.3–7.6 cal kyr BP) (Figure 4). It is characterized by the lowest content of siliciclastic elements (Al and Fe), and the higher Ca content and sand fraction (med. 15.86%). TOC and Mo concentrations are very low (Figure 5). In Subzone 3/2 (320–300 cm; 7.9–7.6 cal kyr BP), the content of the siliciclastic elements (Al and Fe) increases, whereas Ca decreases (Figure 5). The sand fraction is almost absent (med. 0.13%), and clay increases (med. 15.52%). This subzone characterizes elevated N and TOC values (med. 0.17% and 1.68%), with a peak at 305 cm (0.34% and 4.15%, respectively) (Figure 4). Here, P and S are also higher, while Mo peaks at 300 cm after the TOC peak. C/N ratios remain relatively constant, with slightly higher values in Subzone 3/1 (med. 11.7) compared to Subzone 3/2 (med. 11.9). The values of stable carbon (med. −25.69‰) and nitrogen (med. 2.28‰) isotopes are low; after which, there is an increase in their values in the next zone.

In lithological Zone 4 (300–0 cm; 7.3–0 cal kyr BP), the clay fraction increases to med. 16.43%, with almost no presence of sand (med. 0.58%). The peak of κ_LF at 278 cm (60.7 × 10⁻⁵ SI) did not show correlation to siliciclastic elements; therefore, we excluded it as a tephra-correlated peak. In this zone, siliciclastic elements (Al and Fe) increase, as well as Mg, and a low Mg/Al ratio reflects a stronger influence of detrital materials (Figure 5). The concentration of Ca decreases, but Sr values, which are incorporated in calcite and aragonite minerals, increase. Low Mg-calcite is present in the upper 200 cm of the core. Quartz, dolomite, and muscovite/illite are also present, and clay minerals are the same as in the previous zones 2/2, 3/1, and 3/2 (hydroxy-interlayered vermiculite, chlorite, illite, and kaolinite). Both N (med. 0.11%) and TOC (med. 1.19%) concentrations are relatively constant throughout Zone 4 (Figure 4). TOC peaks at 180 cm, with a value of 2.31%, whereas the C/N ratio peaks at 26.1. Besides three more C/N peaks with values between 15.8 and 17.5, at 225 cm, 120 cm, and 90 cm, the med. value in zone 4 is 12.5, with slightly lower values of ~10 in the upper 40 cm of the core. Stable carbon and nitrogen values increase after 300 cm and remain constant upwards (−22.26‰ and 4.10‰, respectively). Slightly elevated δ¹⁵N_org values are recorded in the topmost 50 cm of the core (~5‰), where the δ¹³C_org value shows high variability in the upper 30 cm. In the topmost 50 cm, elements Pb and Cu increase, indicating an anthropogenic influence (Figure 5).

4.3.2. Sediment Core ARTA-1

The sediment core ARTA-1 is 145 cm long in total. The lowest lithological Zone 1 (145–70 cm) is characterized by gray sediments, with high values of magnetic susceptibility (κ_LF) (Figure 6) and siliciclastic elements (Al and Fe) (Figure 7) (Table S1). Their decreasing trend allowed division into two subzones, where the highest κ_LF (med. 29.1 × 10⁻⁵ SI) is in Subzone 1/1 (145–95 cm; 15.5–10.6 cal kyr BP), followed by a decreasing trend (med. 20.6 × 10⁻⁵ SI) in Subzone 1/2 (95–85 cm; 10.6–9.6 cal kyr BP). Insoluble residue is high in Zone 1 and slightly higher in Subzone 1/2. The sand fraction is not present or only in very low amounts, but the sediments from subzones 1/1 and 1/2 contain a lot of clasts of various sizes (from 0.5 to 5 mm). Silts are the dominant grain size fraction (med. 83.80%), while the clay fraction is present with the median value of 16.18%. The nitrogen (0.06%) and TOC (0.50%) concentrations are relatively low in Subzone 1/1, with an increasing trend; therefore, in Subzone 1/2, the values are 0.12% and 1.68%, respectively. Similarly, C/N ratios are 11.5 (med.) in Subzone 1/1 and increase to 16.0 (med.) in Subzone 1/2. The carbon isotope record in the whole of Zone 1 has relatively low values (δ¹³C_org = −26.48‰ med.), while the nitrogen isotopes (δ¹⁵N_org) are the highest (med. 5.45‰) (Figure 6). In the mineral composition in Subzone 1/1, quartz and calcite are dominant, while muscovite/illite and dolomite are also present, as well as feldspars and clay minerals. Clay minerals are dominated by smectite and illite, followed by kaolinite and chlorite. In Subzone 1/2, calcite, as well as dolomite, are absent, which is also evident in very low TIC median values (0.19%). Clay minerals are present in low amounts in Subzone 1/2, presented by hydroxy-interlayered vermiculite, chlorite, illite, and kaolinite (Figure 7). In the subzones 1/1 and 1/2, no datable material was found, and the sediments did not contain any shell fragments or plant remains, which appear from the next zone upwards.

Sediments become darker in Zone 2, as evidenced by the lower values of color parameter L*. We distinguish Subzone 2/1 (85–70 cm; 9.6–8.2 cal kyr BP) and Subzone 2/2 (70–58 cm; 8.2–6.8 cal kyr BP) (Figure 6). The grain size composition is similar in both zones, dominated by silt (med. 81.70%), while the sand is absent, and clay appears in the highest amount in the core (med. 18.29%). The decreasing magnetic susceptibility and insoluble residue and reappearance of calcite characterize Subzone 2/1. Quartz and muscovite/illite are also present. Both N and TOC increase (med. 0.21% and 2.62%, respectively), with high C/N values (med. 14.9). In Subzone 2/2, magnetic susceptibility is low (med. 5.4 × 10⁻⁵ SI), as well as the dry bulk density, being the lowest in this zone. The TOC content is the highest (med. 3.91%), as well as nitrogen (med. 0.32%), while the C/N ratio remains similar to the values as in the previous Subzone 2/1 (Figure 6). Here, S and Mo are the highest in the core. Carbonate elements (Ca and Sr) are increasing upwards. In mineral composition, calcite and quartz are present, followed by dolomite, muscovite/illite, and halite (which results from sediments drying from this zone upwards) (Figure 7). The carbon isotope values slightly increase to med. −25.54‰, while the nitrogen isotope values slightly decrease (med. 3.25‰).

Subzone 3/1 (58–37 cm; 6.8–4.4 cal kyr BP) is characterized by low magnetic susceptibility (med. 1.1 × 10⁻⁵ SI), and the insoluble residue is in a decreasing trend, as well as siliciclastic elements (Figure 6). Silt is continuously high (med. 87.27%), and the clay fraction is decreasing to median values of 8.61%, while the sand fraction increases to med. 4.34%. The TOC and nitrogen concentrations decrease to med. 2.37% and med. 0.21%. C/N ratios and show little variation from the previous zone (med. 13.1), while the δ¹³C_org record increases to −21.40‰, in contrast to the decrease in the δ¹⁵N_org values (1.29‰). Carbonate elements are the highest in the whole Zone 3. This is supported by calcite, quartz, and dolomite as the main mineral constituents, followed by aragonite and halite. The clay minerals are the same as in previous zones 2/2, 2/1, and 1/2 (Figure 7).

Subzone 3/2 (37–0 cm; 4.4–0 cal kyr BP) marks the lowest magnetic susceptibility (med. 0.8 × 10⁻⁵ SI), with a slight increase (~2.4 × 10⁻⁵ SI) in the upper 8 cm (Figure 6). It coincides with the increasing clay fraction (med. 21.4%) in the topmost 5 cm. The sand fraction (med. 26.28%) increases, while the silt (med. 66.88%) decreases. TOC concentrations continued to decrease from the previous zone (med. 0.88%), as well as N (med. 0.09%) and C/N ratios (8.9). The carbon and nitrogen isotopes are relatively similar to the previous Subzone 3/1 (−20.47‰ and 1.42‰, respectively). The siliciclastic elements are the lowest in this zone, opposite to the carbonate elements, which are the highest. Calcite and dolomite are the dominant mineral phases, followed by quartz and aragonite in the lower part of the zone, while, in the upper part of the zone, in the topmost 10 cm, dolomite predominates, and low magnesium calcite appears. In clay minerals, there is a shift to the presence of swelling clay minerals (smectite), alongside chlorite, illite, and kaolinite, similar to Subzone 1/1 (Figure 7).

4.4. Ostracods

Forty-seven ostracod species from the freshwater and marine families Darwinulidae, Candonidae, Ilyocyprididae, Cyprididae, Cytherideidae, Limnocytherideidae, Leptocytheridae, Loxoconchidae, Cytheruridae, Trachyleberidae, Cytheromatidae, Pontocypididae, Hemicytheridae, and Xestoleberididae were recognized. Among the ostracod species, 34 were identified to the species level (Table S2).

Well-preserved ostracods are identified within samples in the PROSIKA-1 core and in the upper 72 cm of core ARTA-1. Only one sample (89–91 cm) in ARTA-1 was barren of ostracods. Species richness in PROSIKA-1 varied between three (630–632 cm) and twenty (282–284 cm) per sample and generally slightly increased from a depth of 314 cm to the top of the core.

In ARTA-1, species richness oscillated between 13 and 18 species per sample in the upper part of the core (from 72 cm to the top of the core). The lowest species richness values were recorded in the lower part of the core (from 145 to 72 cm) when ostracod abundances were minimal and varied between one and five species per sample.

The total number of freshwater species was twenty in the PROSIKA-1 core and fourteen in the ARTA-1 core. The total number of marine species in PROSIKA-1 was eighteen and fourteen in ARTA-1. There were nine shared freshwater species between both cores and eleven marine species (Figures S1 and S2; Table S2).

In addition to the ostracods, the samples also contained additional marine and freshwater micropaleontological remains, including benthic foraminifera, bivalves, gastropods, gastropod operculae, plant remains, seeds, and calcified gyrogonites.

4.4.1. Ostracod Assemblage in Sediment Core PROSIKA-1

In the PROSIKA-1 core, we distinguished the freshwater and the marine ostracod assemblages (Figure 8A). The changes in ostracod assemblages are visible through the detected lithological zones and subzones. Within the zones, there is a difference in the number of ostracod valves and the composition of the ostracod assemblage. The number of ostracod valves increases in the younger samples.

The ostracod fauna of the first and second zones is freshwater in character. The ostracod assemblage of Zone 1 (745–660 cm) is dominated by Hungarocypris madaraszi, Juxilyocypris kemphi, Trayanocypris cf. leavis, and Cypris pubera. Other species that appear in this zone are Candona sp., Pseudocandona marchica, Herpetocypris reptans, Cypridopsis vidua, Limnocythere inopinata, Pseudocandona sp., Potamocypris sp., and Heterocypris sp. In the 2/1 Subzone (660–510 cm), Juxilyocypris kemphi is the dominant species, as, in the first zone, Hungarocypris madaraszi, Cypris pubera, Cyclocypris sp., Herpetocypris reptans, Potamocypris sp. Hungarocypris madaraszi, Cypridopsis vidua, and Limnocythere inopinata are also present. Ilyocipris cf. grabschuetzi, Darwinula stevensoni, and Cyprideis torosa occur at the end of the subzone. Four species: Cypris pubera, Potamocypris sp. Cyclocypris sp., and Ilyocipris cf. grabschuetzi disappear in the 2/1 Subzone.

The dominant species in the Subzone 2/2 (510–375 cm) are Pseudocandona marchica and Cypridopsis vidua. Other species in this subzone are Darwinula stevensoni, Candona sp., and Herpetocypris reptans. Species Candona neglecta and Pseudocandona cf. compressa are registered for the first time in the upper half of the Subzone 2/2. Three species, Juxilyocypris kemphi, Hungarocypris madaraszi, and Cypridopsis vidua, occur last. A small number of Cyprideis torosa are present in the subzone. The presence of charophytes, seeds, plant fragments, and freshwater gastropods corroborates the freshwater conditions.

At the beginning of Subzone 3/1 (375–320 cm), we registered the first valve of marine ostracod Carinocythereis cf. carinata and a slight increasing number of Cyprideis torosa, and freshwater ostracods Candona angulata, Candona sp., Pseudocandona cf. compressa, Pseudocandona marchica, Pseudocandona sp., Herpetocypris reptans, and Limnocythere inopinata. In the subzone, the last occurrence of Trayanocypris cf. leavis and the first occurrence of Ilyocypris bradyi and Ilyocypris cf. monstrifica are registered. At the top of Subzone 3/1, other marine species, Semicytherura cf. incongruens, Hiltermanicythere turbida, Pterygocythereis cf. P. ceratoptera, and Cytheroma variabilis, occur. In the following Subzone 3/2 (320–300 cm), the dominant are Cyprideis torosa, Candona angulata, and Ilyocypris bradyi. Species Candona sp., Pseudocandona marchica, Pseudocandona sp., Ilyocypris cf. monstrifica, Heterocypris sp., and Limnocythere inopinata were also found. Marine ostracods Semicytherura cf. incongruens, Pterygocythereis jonesii, Cytheroma variabilis, and Carinocythereis whitei were found in Subzone 3/2.

The uppermost Zone 4 (300–0 cm) contains marine and freshwater ostracods. Marine ostracods are dominant. Most numerous are Cyprideis torosa, Semicytherura cf. incongruens, Hiltermannicythere turbida, Pterygocythereis jonesii, Cytheroma variabilis, Carinocythereis whitei, Leptocythere cf. rara, and Leptocythere ramose. In smaller numbers than the species already mentioned, we identified Semycytherura sp., Callistocythere sp. Hiltermannicythere rubra, Pterygocythereis cf. P. ceratoptera, Carinocythereis cf. carinata, Propontocypris cf. intermedia, Aurila convexa, Leptocythere sp., Xestoleberis sp., and Loxoconcha sp. With marine ostracods, we recorded numerous freshwater species, Candona angulata, Candona sp., Pseudocandona marchica, Pseudocandona sp., and rare Ilyocypris bradyi, Ilyocypris cf. monstrifica, Darwinula stevensoni, Candona neglecta, Heterocypris sp., Cypridopsis vidua, and Limnocythere inopinata. The number of freshwater ostracod valves varies through the zone. Generally, freshwater species decrease from the sample at 240–242 cm towards the top of the core.

4.4.2. Ostracod Assemblage in Sediment Core ARTA-1

In the ten samples from the ARTA-1 core, we determined mixed marine and freshwater ostracods (Figure 8B). The changes in ostracod assemblages are visible through the detected lithological zones and subzones. In the core ARTA-1, minor discrepancies are visible in the changes of ostracods with the determined lithological zones and subzones, which are most likely caused by the small number of samples analyzed.

A few freshwater species, such as Candona sp., Mixtacandona sp., Bradleystrandesia? sp., and Limnocythere inopinata, and marine Cyprideis torosa, Hiltermanythere turbida, and Leptocythere sp., are found at the base of the first zone (145–132 cm). Within Zone 1 (145–85 cm), ostracods were not found, except for one valve, Semicytherura cf. incongruens (sample 99–101 cm). Subzone 2/1 (85–70 cm) contains a few Cyprideis torosa, Semicytherura cf. incongruens, and Semycytherura sp. In Subzone 2/2 (72–56 cm), ostracod assemblages change, and we found freshwater and marine species for the first time together. Most numerous are freshwater Bradleystrandesia? sp., and Ilyocypris bradyi. Other freshwater species Darwinula stevensoni, Candona angulata, Candona sp., Pseudocandona cf. marchica, Pseudocandona sp., Cypris bispinosa, Cypridopsis vidua, and Limnocythere inopinata are found rarely. The number of marine species Cyprideis torosa, Semicytherura cf. incongruens, Hiltermanythere rubra, H. turbida, Carinocythereis whitei, Leptocythere ramosa, Xestoleberis dispar, Xestoleberis glaberscens, Loxoconcha sp., and Loxoconcha cf. agilis in Subzone 2/2 is small, but they are slowly increasing. The composition of the ostracod assemblages gradually changes in Zone 3 (56–0 cm). Marine species are increasingly numerous, and freshwater can still be found. More numerous are Cyprideis torosa, Semicytherura cf. incongruens, Hiltermanythere turbida, Xestoleberis dispar, Xestoleberis glaberscens, Loxoconcha sp., and Loxoconcha cf. agilis. Species Hiltermannicythere rubra, Pterygocythereis jonesii, Carinocythereis cf. carinata, Carinocythereis whitei, Leptocythere ramosa, and Leptocythere sp. followed in lower abundance. Together with marine species, fourteen freshwater species were determined: Candona angulata, Candona sp., Pseudocandona cf. marchica, Pseudocandona sp., Ilyocypris bradyi, Bradleystrandesia? sp., and Limnocythere inopinata. In low numbers Heterocypris sp., Herpetocypris reptans, Cypris bispinosa, Cypridopsis vidua, and Metacypris cordata were found. Pterygocythereis jonesii was found only in Subzone 3/2, and freshwater Ilyocypris bradyi and Metacypris cordata were limited to Subzone 3/1.

5. Discussion

In the following discussion, we consider the full range of proxy evidence for paleoenvironmental changes and karst landscape evolution recorded in Pirovac Bay from two cores, PROSIKA-1 (7.45 m bsf) and ARTA-1 (1.45 m bsf). The objective was to evaluate the potential of geochemical analyses of sediments and elemental and isotopic sedimentary organic matter, combined with ostracods in reconstructing past environments. This approach is especially valuable in areas where microfossil preservation is poor or where dating constraints are limited.

5.1. Geomorphic Setting of Pirovac Bay

Our results have proven that, geomorphologically, the Pirovac Bay area, where PROSIKA-1 was cored, can be defined as a karst depression. The additional sediment core ARTA-1 was taken in a small depression in the NW part of the bay, southeast of the Island of Arta Mala (Figure 1C). This geomorphic setting (i.e., the existence of depressions) enabled the accumulation of up to 12-m-thick Quaternary sediment successions in our study area.

Pirovac Bay depressions can further be defined as isolation basins, in which the sill depth controls the development of depositional environments and marine flooding or isolation of the basin e.g., [6,64]. The silled basins are a common occurrence along the Eastern Adriatic [18,19,20,21,22,23].

Furthermore, the determination of the sill depth is of immense importance for understanding the paleoenvironmental evolution of the area. We estimated that the shallowest sill is in the NW part of the bay, between the Island of Mala Arta and land, at a 6 m water depth (Arta sill) (Figure 1C). Thus, Arta sill had a crucial role in the development of environmental conditions in Pirovac Bay, which will be discussed in more detail in the following sections.

5.2. Late Glacial to Middle Holocene Freshwater Environment (Post-NYT Age–8.3 cal kyr BP)

Cryptotephra found at a depth of 721 cm in core PROSIKA-1 was correlated to the Neapolitan Yellow Tuff (NYT) tephra [62]. It is a common tephra layer on the Eastern Adriatic Coast, also found relatively close to Pirovac Bay but towards the open sea in the eastern part of the Mid-Adriatic deep (MAD) [65]. Additional findings include locations that are located more south of our study site in the paleolake sediments in Lake Veliko Jezero on the Island of Mljet [22] and in cave sediments on the Island of Korčula [66]. This tephra served as an excellent anchor for the chronological framework of the PROSIKA-1 core. Multiproxy analyses of sediments from PROSIKA-1 Zone 1 (post-NYT age–12.4 cal ka BP; 7.4–6.6 m bsf) suggest deposition in a freshwater environment, without marine influence (Figure 9). The mineralogical and geochemical composition reveal a higher sediment input of siliciclastic material and dolomite from the catchment during this period. Clay minerals are detrital and could originate from the flysch and/or loess-like deposits present in the surrounding area, which show similar composition (smectite, chlorite, illite, and kaolinite) [67].

Higher C/N values (~14) reflect mixed aquatic (authigenic) and terrestrial (allochthonous) land-derived organic matter of higher-order plant material [30]. δ¹³C_org versus C/N values of the PROSIKA-1 Zone 1 fell into the freshwater dissolved organic carbon (DOC) and terrestrial plants fields (Figure 10), evident from a range of organic sources for various coastal environments [68]. Terrestrial organic matter, often originating from C3 plants, generally has lower δ¹³C values [68]. A high groundwater discharge in the Pirovac Bay depression during this period is evident, considering that the bay has an underground connection with the nearby Lake Vrana near Biograd na Moru [43,44]. Enhanced water input during the wet and warm Bølling-Allerød period (BA; 14.5–12.8 cal kyr BP; [69]) is recorded in Lake Vrana sediments on the Island of Cres [29] and Mid-Adriatic Deep [65]. During the Late Glacial, intermittent freshwater bodies like pools and ponds could have been formed in the floodplain environment in Lake Vrana [70]. A similar paleoenvironment is expected in the Pirovac Bay depression during this period. Water drained through the permeable karstified bedrock into the basin in Pirovac Bay and further into the sea. Ostracod assemblage supports other analysis and suggests the deposition in a freshwater floodplain environment. Numerous are species Juxilyocypris kempfi, Hungarocypris madaraszi, and Trayanocypris cf. leavis, typical for this paleoenvironment, as well as species Candona sp., Pseudocandona marchica, Herpetocypris reptans, Cypridopsis vidua, Limnocythere inopinata, Pseudocandona sp., Potamocypris sp., and Heterocypris sp., with very wide ecological ranges [53,70,71]. Reflector configurations in Acoustic Unit 2 in high-resolution seismic profiles also support this interpretation (Figure 2). During meltwater pulse (MWP) 1A, at the beginning of the BA period, the sea level rose rapidly from 120 to 80 m bsl, and at 13 cal kyr BP, the sea was approximately 75 m lower than it is today [4,5]. This means the sea and the paleoshoreline were far away from the present shoreline, which enabled the formation of the floodplain environment (Figure 11).

The further evolution to freshwater wetland, a more stagnant water environment, probably occurred from 12.4 to 10.0 cal kyr BP (6.6–5.1 m bsf; PROSIKA-1 Subzone 2/1). This phase corresponds to the end of the cold Younger Dryas period [73]. The wetland phase is inferred based on the increasing trend of Ca and endogenic calcite sedimentation in Zone 2/1, accompanied by a simultaneous decrease in the input of siliciclastic material and dolomite from the surrounding catchment (Figure 9). Both nitrogen and TOC concentrations increased in that interval, indicating a more productive environment. The C/N ratio is lower (~11) compared to the previous zone and reveals a mixed source of organic matter, only partly from terrestrial sources and partly from freshwater DOC, which moves towards marine DOC (Figure 10). Carbon stable isotopes tend to have the less negative values, indicating more aquatic freshwater to marine sources of organic matter. Nitrogen isotope values are higher in the lower part and decrease in the upper part of the zone. The freshwater species from Zone 1 are still present in Subzone 2/1. In the uppermost samples, the first appearance of Darwinula stevensoni and Ilyocypris cf. grabschuetzi occur. These two species possibly indicate the beginning of a warmer period [53,55]. In Subzone 2/1, more numerous are Herpetocypris reptans and Cypridopsis vidua in comparison to the previous zone.

In the period from 10.0 to 8.3 cal kyr BP (5.1–3.7 m bsf; PROSIKA-1 Subzone 2/2), the permanent lake environment was established in Pirovac Bay. Ca and endogenic calcite predominated, without almost no sediment delivery and detrital input into the lake. Clay minerals (hidroxy-interlayered vermiculite, chlorite, illite, and kaolinite) shifted to a composition similar to terra rossa soils [74], indicating different source material during this period. The number of different freshwater species of the genus Pseudocandona gradually increased in this period, in association with Candona sp., Herpetocypris reptans, and Cypridopsis vidua. A small number of Cyprideis torosa are present in the zone. A similar freshwater paleoenvironment was described in nearby Lake Vrana near Biograd na Moru [70,71]. The development of a lacustrine environment could also be inferred from the acoustic data—Acoustic Unit 3 with subparallel reflectors (e.g., [20]).

The paleolake in Pirovac Bay developed due to the rising sea level and thus a rise in groundwater levels in Pirovac Bay and towards Lake Vrana, which existed as wetland during that period (Figure 11 and Figure 12). They were hydrologically connected only through permeable karst, through which Lake Vrana fed with freshwater the lake in Pirovac Bay. The freshwater lake sediments in the Pirovac basin are the first documented indicators of freshwater paleolake in the Eastern Adriatic karstic region during the Middle Holocene period. Along the Eastern Adriatic Coast, brackish paleolakes were formed earlier—during the Late Glacial; in the Lošinj Channel [19], in Lake Veliko Jezero, and Stupa Bay on the Island of Mljet [21,22]; and in the Koločep Channel [23]. The freshwater sediments in Pirovac Bay were identified at depths of 6.6–3.7 m bsf at the total water depth of the sediment core PROSIKA-1 of 25 m equivalent to approximately 32–28 m bsl. They were deposited in a freshwater environment for approximately 3500 years. The position of the Prosika sill, located at a depth of 8 m bsl, provided insights into the maximum paleolake water level. Based on the sill’s depth, the paleolake’s water level could have reached between 18 m and 22 m during its maximum extent, since there is no evidence for paleolake formation in the Arta depression, described in next paragraph. On the geophysical profile (Figure 1C), a paleoterrace is visible that could indicate a possible lake paleoshore at a depth of 10 m bsl. Taking this into account, the maximum probable water depth of the paleolake could be approximately 20 m (Figure 13). We believe that the geomorphology of the Pirovac Bay basin governed this landscape evolution (~20 m deep during the lake sediments deposition), as well as the geology of the Pirovac Bay catchment, which is composed of dolomitic carbonate rocks of low permeability. These factors are also considered to be crucial for the development of the lake in Lake Vrana on the Island of Cres during the periods of low sea levels [29].

Late Glacial and Early Holocene sediments from core ARTA-1 Subzone 1/1 (15.5–10.6 cal kyr BP, 1.45–0.95 m bsf) and ARTA-1 Subzone 1/2 (10.6–9.6 cal kyr BP, 0.95–0.85 m bsf) were characterized by high sediment input of siliciclastic material and low carbonate content (Figure 9). Both subzones showed a freshwater algae signature: freshwater DOC with lower C/N values in ARTA-1 Subzone 1/1 and freshwater DOC with higher C/N ratios in ARTA-1 Subzone 1/2 (Figure 10). Ostracods are very rare or absent throughout most of the whole ARTA-1 Zone 1 (Figure 9). At the base of the ARTA-1 core, the freshwater species Candona sp., Mixtacandona sp., Bradleystrandesia? sp., and Limnocythere inopinata were observed. Occurrence of the genus Mixtacandona is indicative of interstitial waters [75], while species of the genus Bradleystrandesia prefer temporary pools but is also found in permanent ponds [53]. Alongside the higher siliciclastic input from the catchment, this environment is interpreted as a floodplain with intermittent pools and ponds (Figure 9 and Figure 11). There is no evidence for freshwater paleolake formation in the Arta depression.

5.3. Middle Holocene Brackish Lake (8.3–7.3 cal kyr BP)

The Middle Holocene period includes PROSIKA-1 Subzone 3/1 (375–320 cm; 8.3–7.6 cal kyr BP), which is characterized by the lowest content of siliciclastic material and the higher carbonate content. Organic carbon concentrations were relatively higher than in the previous Subzone 2/2, while the stable nitrogen and carbon isotope values remain continuous compared to the previous period and fall into a similar field of source of organic matter (terrestrial mixed freshwater and marine) (Figure 10). In the PROSIKA-1 Subzone 3/1, freshwater species dominated, but a small number of marine ostracods were observed. In the interval between 3.40 and 3.30 m bsf, marine ostracods appeared in higher abundancies and therefore indicated a higher seawater influence.

In PROSIKA-1 Subzone 3/2 (320–300 cm; 7.6–7.3 cal kyr BP), the content of siliciclastic elements (Al and Fe) starts to increase, while Ca is lower. The Mo concentrations slightly increased within this zone, suggesting the intrusion of seawater and the development of suboxic to anoxic conditions, which coincide with elevated TOC and S concentrations. The carbon isotope values decreased, while the stable nitrogen isotopes slightly increased. This is followed by the increased C/N ratio. These correlations could indicate the decomposition of organic matter [76] in this interval. Isotopically light δ¹³C_org values with relatively high C/N values (>10) fall within the freshwater DOC and terrestrial plant fields (Figure 10). This effect could be the result of the preferred decomposition of authigenic marine organic matter due to the increased marine influence and thus probably higher oxygen content as the result of increased TOC. Under such conditions, marine OM decomposes more easily. According to the increasing number of Cyprideis torosa in this PROSIKA-1 Subzone 3/2 and the presence of marine ostracods, the formation of a brackish lake likely began, with a stronger marine influence. C. torosa is a crucial species for understanding the paleoenvironmental history of the investigated area and the broader region due to its ability to adapt to a wide range of salinities (0.4‰ to 150‰), and its occurrence in both coastal and inland water bodies makes it a valuable species for studying past salinity variations and hydrodynamic conditions [56,77]. The numerous C. torosa indicates seawater intrusion through permeable karstified bedrock. A higher abundance of the C. torosa, especially in the PROSIKA-1 core Zones 3 and 4 (Figure 8A), indicates a gradual increase in salinity. This species is, therefore, a clear indicator of brackish water and so has been widely utilized as an index fossil in paleosalinity reconstructions, as ecophenotypic modifications of its shell surface (i.e., different ornamentations and percentages of sieve pore shapes) depend on salinity [78,79]. In our samples, the shell surface of C. torosa is smooth and pitted. Nodosities have been noticed only in a few juvenile instars in PROSIKA-1 core subzone 3/1 (sample 374–375 cm). The shape of lateral sieve pores of C. torosa in the analyzed samples varies in outline from oval and irregular. Generally, increased percentages of irregular sieve pore surfaces are characteristic of forms that live in more saline waters [78].

Our results showed that a brackish lake in Pirovac Bay existed between 8.3 and 7.3 cal kyr BP, with the marine influence through the karstic underground (Figure 11). The sea level at 8 cal kyr BP was ~12 m lower and, at 7.6 cal kyr BP, ~9 to 10 m lower than today [5] (Figure 12 and Figure 13). Marine intrusion occurred through the permeable karstic underground and resulted in the formation of a brackish lake.

Furthermore, the marine influence is evident in ARTA-1 Subzone 2/1 (9.6–8.2 cal kyr BP, 0.85–0.70 m bsf), supported by the appearance of a few marine ostracods. The higher C/N ratio (~12–13) points to terrestrial organic material, mixed with freshwater DOC (Figure 10). In ARTA-1 Subzone 2/2 (8.2–6.8 cal kyr BP, 0.70–0.58 m bsf), marine species are present, and sediments transit to the marine DOC field, according to the organic matter source material (Figure 10).

Figure 12. Global sea level rise (blue line, MWP—meltwater pulse) [5] and the paleoenvironmental evolution of Pirovac Bay, with the Sr/Ca sediment record from core PROSIKA-1. The red dot represents the sea level limiting data point for Pirovac Bay, and other dots are from different studies on the Eastern Adriatic Coast [16,19,20,21,22,26,80].

Figure 12. Global sea level rise (blue line, MWP—meltwater pulse) [5] and the paleoenvironmental evolution of Pirovac Bay, with the Sr/Ca sediment record from core PROSIKA-1. The red dot represents the sea level limiting data point for Pirovac Bay, and other dots are from different studies on the Eastern Adriatic Coast [16,19,20,21,22,26,80].

Figure 13. Schematic diagrams illustrate the depositional environments and the transgression process in Pirovac Bay that occurred due to the Holocene sea level rise [5]: from the freshwater environments (A) and the development of freshwater paleolake (B) to the brackish paleolake (C) and marine environment (D). Block diagrams support the paleoenvironmental evolution. The present sea level and the maximum lake level are marked, according to the acoustic profile (Figure 1C), where the lake paleoshoreline can be recognized (B,C). Dark green and blue arrows mark the subsurface waterflows. Deposited sediments are showed in gray colors.

Figure 13. Schematic diagrams illustrate the depositional environments and the transgression process in Pirovac Bay that occurred due to the Holocene sea level rise [5]: from the freshwater environments (A) and the development of freshwater paleolake (B) to the brackish paleolake (C) and marine environment (D). Block diagrams support the paleoenvironmental evolution. The present sea level and the maximum lake level are marked, according to the acoustic profile (Figure 1C), where the lake paleoshoreline can be recognized (B,C). Dark green and blue arrows mark the subsurface waterflows. Deposited sediments are showed in gray colors.

5.4. Middle to Late Holocene Marine Environment (7.3 cal kyr BP to the Present)

The marine transgression over the Arta sill, at the depth of 6 m, could have occurred at 7.3 cal kyr BP at 3.0 m bsf of core PROSIKA-1 (Figure 12 and Figure 13). The establishment of a marine environment was indicated by increased Sr/Ca concentrations (Figure 9 and Figure 12). The isotopically higher δ¹³C_org values versus C/N ratios correspond to marine DOC and marine particulate organic carbon (POC) (Figure 10). This supports the interpretation of a marine depositional setting with mixed marine algal (<10 C/N value) and terrestrial plants (>10 C/N values) as a source of organic material. In the last 1300 years (uppermost 50 cm), the samples had the lowest C/N values (~8) and transitioned to more marine algae organic carbon production. The marine environment was further confirmed with the autochthonous marine ostracod fauna. Most numerous were Cyprideis torosa, Semicytherura cf. incongruens, Hiltermannicythere turbida, Pterygocythereis jonesii, Cytheroma variabilis, Carinocythereis whitei, Leptocythere cf. rara, and Leptocythere ramose. Alongside the marine fauna, we also identified allochthonous freshwater ostracod fauna. Freshwater ostracods are introduced into the shallow marine environment of Pirovac Bay through groundwater discharge along the eastern shores of the bay. Those coastal springs are connected to the nearby freshwater Lake Vrana. As a result, starting from 7.3 cal ka BP, both marine and freshwater ostracods with a tolerance to elevated salinity began to appear in the bay, suggesting a significant freshwater influx from Lake Vrana through the surficial canal Prosika and groundwater springs (estavelles) along the bay’s northeastern shores. Acoustic data (Acoustic Unit 4) also suggest the deposition of homogenous marine sediments with a semi-transparent acoustic pattern (Figure 2) (e.g., [20,81].

At the onset of the marine transgression, dated to approximately 7.3 cal kyr BP, the relative sea level must have been lower than the –6 m depth of the Arta sill (Figure 12).

In the ARTA-1 core, Subzone 3/1 (6.8–4.4 cal kyr BP; 0.58–0.37 m bsf) and Subzone 3/2 (4.4–0 cal kyr BP; 0.37–0 m bsf) support the development of marine environment based on stable isotope composition ¹³C_org versus C/N ratios. This plots the sediments from Subzone 3/1 into marine DOC and terrestrial organic matter, with >10 C/N values, while sediments from Subzone 3/2 trend towards marine algae and POC, with <10 C/N values (Figure 10). In both subzones, shallow marine species are dominant: Cyprideis torosa, Semicytherura cf. incongruens, Hiltermanythere turbida, Xestoleberis dispar, Xestoleberis glaberscens, Loxoconcha sp., and Loxoconcha cf. agilis. The detected freshwater ostracod species in Subzones 3/1 and 3/2 are introduced in the marine environment.

6. Conclusions

Two new sediment records from a shallow coastal bay from the Eastern Adriatic Coast in Central Dalmatia were used to investigate environmental evolution and sediment organic matter source variability through the Late Glacial and Holocene, influenced by the sea level rise. The karst depression in Pirovac Bay contains a ca. 14,500-year-long record of paleoenvironmental changes, dated according to the NYT tephra at the base of the core PROSIKA-1 [62]. With the results of the multiproxy investigation (sedimentological, mineralogical, geochemical, paleontological, and geophysical data), it is possible to differentiate the freshwater and brackish to marine environments.

Late Glacial to Early Holocene sediments were deposited in floodplains and intermittent pools and ponds, with typical freshwater ostracods for such environments. Freshwater ostracod taxa from this floodplain and intermittent pools and ponds are abundant in the Late Pleistocene to Early Holocene transition sediments in nearby Lake Vrana [70,71]. Two freshwater environments coexisted as floodplains to wetlands during the Early Holocene. During the Middle Holocene, sediments accumulated in a shallow freshwater paleolake with carbonate-rich waters and a rich ostracod fauna. The sediments were deposited in a freshwater environment during a period of ~3500 years. The limiting factors that controlled the formation of a freshwater paleolake in Pirovac Bay, except the rising sea level which raised the groundwater levels, were geomorphological setting and geological characteristics. The relatively small depth of the bay, governed by the shallow sill, compared to other studied isolated karst basins (Lošinj Channel, Veliko Jezero, and Stupa Bay on the Island of Mljet) in the Eastern Adriatic Coast [18,19,20,21,22,23] allowed the Holocene freshwater paleolake development. Abundant carbonate dolomite rocks in the bay catchment retained the water during the Early and Middle Holocene. Therefore, the sediment core data from the submerged karst basin Pirovac Bay represents the first evidence of a submerged Holocene freshwater paleolake on the Eastern Adriatic Coast. The formation of a brackish lake, under the marine influence due to the sea level rise, was determined from 8.3 to 7.3 cal kyr BP, followed by marine transgression at 7.3 cal kyr BP. From the depth of Arta sill, we suggest that the relative sea level must have been lower than the sea level limiting point (–6 m) prior to marine flooding. The changes in sea level had an influence on the organic matter isotopic signal during the Late Glacial and Holocene and thus on the source of organic carbon in Pirovac Bay. The results further demonstrated that the ostracods are indicative of water column freshening since the onset of marine transgression at 7.3 cal kyr BP. The freshwater discharge occurred due to the connection of the bay to adjacent freshwater Lake Vrana through the surficial canal Prosika and groundwater discharge (numerous estavelles) along the northeastern shores of the bay. This was observed in the reliable record of freshwater fauna, with continuous mixing of Middle and Late Holocene freshwater and marine fauna due to the existence of the karst terrain in Pirovac Bay. The results obtained significantly improved our understanding of the evolution of karst landscape in response to sea level changes. Reconstruction of the coastal geomorphology will also aid in the efforts to unravel the submerged karst along the Eastern Adriatic.

Future research could focus on high-resolution dating; expanded paleoecological analysis; and numerical modeling that refine our understanding of the complex interactions between sea level rise, groundwater dynamics, and ecosystem change in submerged karst systems.