Speleothem Records of the Hydroclimate Variability throughout the Last Glacial Cycle from Manita pe´c Cave (Velebit Mountain, Croatia)

the Variability the pe´c Abstract: We present stable carbon ( δ 13 C) and oxygen ( δ 18 O) isotope records from two partially coeval speleothems from Manita pe´c Cave, Croatia. The cave is located close to the Adriatic coast (3.7 km) at an elevation of 570 m a.s.l. The site experienced competing Mediterranean and continental climate inﬂuences throughout the last glacial cycle and was situated close to the ice limit during the glacial phases. U-Th dating constrains the growth history from Marine Isotope Stage (MIS) 5 to MIS 3 and the transition from MIS 2 to MIS 1. 14 C dating was used to estimate the age of the youngest part of one stalagmite found to be rich in detrital thorium and thus undatable by U-Th. On a millennial scale, δ 18 O variations partly mimic the Dansgaard–Oeschger interstadials recorded in Greenland ice cores (Greenland Interstadials, GI) from GI 22 to GI 13. We interpret our δ 18 O record as a proxy for variations in precipitation amount and/or moisture sources, and the δ 13 C record is interpreted as a proxy for changes in soil bioproductivity. The latter indicates a generally reduced vegetation cover towards MIS 3–MIS 4, with shifts of ~8‰ and approaching values close to those of the host rock. However, even during the coldest phases, when a periglacial setting and enhanced aridity sustained long-residence-time groundwater, carbonic-acid dissolution remains the driving force of the karstiﬁcation processes. Speleothem morphology follows changes in environmental conditions and complements regional results of submerged speleothems ﬁndings. Speciﬁcally, narrow sections of light porous spelaean calcite precipitated during the glacial/stadial sea-level lowstands, while the warmer and wetter conditions were marked with compact calcite and hiatuses in submerged speleothems due to sea-level highstands. Presumably, the transformation of this littoral site to a continental one with somewhat higher amounts of orographic precipitation was a site-speciﬁc effect that masked regional environmental changes.


Regional Settings and Study Site
Manita peć Cave (44°18′ N, 15°28′ E) is located at 570 m a.s.l. in the side of the Paklenica canyon carved perpendicularly into the Velebit Mountain. It was formed in Upper Jurassic limestone and consists of simple descending spacious chambers with heights up to 38.5 m. Our study site, where the monitoring was conducted and in which speleothems were collected, was at the end of the cave, 150 m from and 40 m below the entrance level. Presently, the surface is mostly bare karst with sparse grass and shrubs (Figure 1b). The recharge area is limited, and the origin of the cave dripwater is restricted to the local meteoric input due to the cave position in epikarst near the summit and relatively shallow overburden (80 m). The infiltration elevation (i.e., altitude of the catchment) is just slightly above the elevation of the cave.

Regional Settings and Study Site
Manita peć Cave (44 • 18 N, 15 • 28 E) is located at 570 m a.s.l. in the side of the Paklenica canyon carved perpendicularly into the Velebit Mountain. It was formed in Upper Jurassic limestone and consists of simple descending spacious chambers with heights up to 38.5 m. Our study site, where the monitoring was conducted and in which speleothems were collected, was at the end of the cave, 150 m from and 40 m below the entrance level. Presently, the surface is mostly bare karst with sparse grass and shrubs ( Figure 1b). The recharge area is limited, and the origin of the cave dripwater is restricted to the local meteoric input due to the cave position in epikarst near the summit and relatively shallow overburden (80 m). The infiltration elevation (i.e., altitude of the catchment) is just slightly above the elevation of the cave.
Owing to the relative vicinity to the Adriatic coast (3.7 km) and the altitude (570 m a.s.l.), the climate type is Cfa-temperate humid climate with hot summers (according to the Köppen-Geiger climate classification [60]), but bordering with Cfb-temperate humid climate with warm summers. The MAAT measured in front of the cave (February 2014-February 2015) was 13.7 • C, while the cave MAAT measured in the innermost part of MP (July 2012-July 2014) was 9.04 • C [49]. Relative humidity (RH) of the cave air is constantly 100%, despite the influence of cold and dry bora winds associated with high pressure, which was recorded by small but abrupt drops of cave air temperature. With the MAAT standard deviation of 0.4 • C (1σ) and constant RH throughout the 2012-2014 period, the MP cave environment may be regarded as stable, and thus favourable for the nearequilibrium precipitation of the spelaean calcite [49]. Discrepancy between the outside and cave MAAT is ascribed to the cave morphology, which, by its descending chambers, acts as a cold trap [61]. Mean annual precipitation of the closest rain gauge station at a similar altitude (Parići village, 3.5 km to the north, 555 m a.s.l.) was 1037 mm for the 1991-2000 period, and shows a bimodal distribution with the maxima during spring and autumn [62]. Precipitation δ 18 O indicates influences from both Atlantic and Western Mediterranean vapor masses [63].

Materials and Analytical Methods
Two stalagmites, namely MP-2 and MP-3, were collected from below the active dripping sites, both from the niches elevated ca. 2 m from the cave floor. Upon collection, Stalagmate ® drip-loggers were installed to record the drip rates, which were, along with the rainfall records, used for the characterization of the karst aquifer. At the same locations, composite monthly water samples were collected for the comparison of the stable isotope composition with meteoric water samples [49]. We also collected two modern calcite samples precipitated on glass plates for the 1/2013-2/2014 period and a third sample from the 2/2014-1/2015 period.
After longitudinal cutting into the halves and polishing, solid pieces of the spelaean carbonate for U-Th dating were extracted using a hand-held dental air drill. The samples were chemically processed and analysed following the methods given in [64,65]. The samples were dissolved in nitric acid and spiked with a mixed 229 Th-233 U-236 U tracer solution. The U and Th were eluted in Eichrom TRU-spec selective ion-exchange resin. Dried samples were then dissolved in diluted nitric acid. Measurements were performed using a Nu Instruments Plasma MC-ICPMS at The University of Melbourne. An initial 230 Th/ 232 Th ratio was estimated by using the method provided in [66].
For the 14 C analyses, solid pieces of MP-3 speleothem were also used to avoid possible contamination with modern atmospheric CO 2 on the outer surface of the samples. Approx. 8-10 mg of each sample was reacted with~2 mL 85% H 3 PO 4 at 90 • C. In the first 5 min of the hydrolysis reaction, CO 2 gas evolved from the outer surface was evacuated from the reaction vessel and not used for 14 C analysis. The hydrolysis reaction was maintained at 90 • C for another 1 h. The CO 2 gas derived from the remaining carbonate material was converted to graphite using excess H 2 over an Fe catalyst [67] and the graphite was rear-pressed into an aluminium cathode for accelerator mass spectrometry (AMS) analyses. All stalagmite 14 C samples were analysed using the STAR facility at ANSTO [68]. The results were corrected for background and isotopic fractionation using measured δ 13 C, and normalized to a 95% oxalic acid (HOx-I) standard.
Sampling for stable isotope analysis was done at 1 mm resolution along the growth axis of each stalagmite and along seven growth laminae for the purpose of the Hendy test, using a tungsten carbide dental drill attached to a Taig CNC micromilling lathe. In addition, the isotopic composition of four samples of modern calcite collected on glass plates (1/2013-2/2014 and 2/2014-1/2015) were measured. A total of ca. 600 stable isotope analyses were conducted at The University of Melbourne (Australia) on an AP2003 continuous-flow mass spectrometer. The results are expressed in delta notation with respect to the VPDB standard. Long-term analytical precision of an in-house reference standard (Carrara marble), previously calibrated to international reference materials NBS-18 and NBS-19, was better than 0.05‰ and 0.1‰ (1σ) for δ 13 C and δ 18 O, respectively.

Monitoring Background
Results from the monitoring of cave microclimate, characterization of the karst aquifer and hydrological behaviour of the MP-2 and MP-3 drip sites conducted between 7/2012 and 7/2014 have been discussed in [49,63]. In short, while MP-3 showed a fracture flow response to rain events (coefficient of variation (CV) of 140%), MP-2 displayed a more stable drip regime (CV of 59%), very close to the seepage-flow class. Throughout the monitoring period, both sites showed no dripflow interruptions-even during the dry season, the highly responsive MP-3 discharged at ca. 30 drops/hour. Seasonal variations in rainwater stable isotope composition (δ 18 O between −9.4‰ and −3.4‰) were attenuated to an amplitude of 0.9‰ for MP-2 and 1.3‰ for MP-3, indicating that the residence time of the infiltrated water in the epikarst is sufficient for homogenization prior to discharge. The discharge is most likely a mixture of slow diffuse (matrix) flow from the overlying beds and fast preferential (fissure) flow [49]. Furthermore, within the relatively stable microclimate conditions in the cave and with well-homogenized dripwater, modern calcite appeared to be deposited close to isotopic equilibrium with the dripwater, when the expression of [69] is applied [49].

Speleothem Samples
Both speleothems show a relatively inhomogeneous internal structure ( Figure 2). The actively growing 15-cm long MP-2 sample consists of two parts formed by the lateral shifting of the drip site ( Figure 2). Its interior is marked with clearly visible alterations of different calcite fabrics-from light porous, which becomes laterally narrower in an upwards direction along the growth axis, to dark and compact layers, which drape laterally over the whole speleothem body. A prominent hiatus marked by an eroded layer can be observed at ca. 80 mm from the bottom of the right half of the MP-2 stalagmite.

Monitoring Background
Results from the monitoring of cave microclimate, characterization of the karst aquifer and hydrological behaviour of the MP-2 and MP-3 drip sites conducted between 7/2012 and 7/2014 have been discussed in [49,63]. In short, while MP-3 showed a fracture flow response to rain events (coefficient of variation (CV) of 140%), MP-2 displayed a more stable drip regime (CV of 59%), very close to the seepage-flow class. Throughout the monitoring period, both sites showed no dripflow interruptions-even during the dry season, the highly responsive MP-3 discharged at ca. 30 drops/hour. Seasonal variations in rainwater stable isotope composition (δ 18 O between −9.4‰ and −3.4‰) were attenuated to an amplitude of 0.9‰ for MP-2 and 1.3‰ for MP-3, indicating that the residence time of the infiltrated water in the epikarst is sufficient for homogenization prior to discharge. The discharge is most likely a mixture of slow diffuse (matrix) flow from the overlying beds and fast preferential (fissure) flow [49]. Furthermore, within the relatively stable microclimate conditions in the cave and with well-homogenized dripwater, modern calcite appeared to be deposited close to isotopic equilibrium with the dripwater, when the expression of [69] is applied [49].

Speleothem Samples
Both speleothems show a relatively inhomogeneous internal structure ( Figure 2). The actively growing 15-cm long MP-2 sample consists of two parts formed by the lateral shifting of the drip site ( Figure 2). Its interior is marked with clearly visible alterations of different calcite fabrics-from light porous, which becomes laterally narrower in an upwards direction along the growth axis, to dark and compact layers, which drape laterally over the whole speleothem body. A prominent hiatus marked by an eroded layer can be observed at ca. 80 mm from the bottom of the right half of the MP-2 stalagmite.  Stalagmite MP-3 is 25 cm long and also displays a small lateral shift in its growth axis ( Figure 2). Prior to collecting, it had been fed by fracture flow, sometimes with a drip intensity of >10,000 drops/hour, and modern calcite has not precipitated, neither on the top of the stalagmite nor on the glass plate installed after removing the speleothem [49]. It seems that, over time, the feeding fracture has widened, enabling transmission of occasionally high discharge and potentially aggressive dripwater, which are unfavourable for calcite deposition. The U-Th dating results are given in Table 1. Uranium concentrations ranged from 17 to 127 ppb with an average value of 63 ppb. At the same time, the 230 Th/ 232 Th activity ratio was very low and ranged from 1.4 to 22.5, implying a significant contribution of detrital Th in these stalagmites. Consequently, these speleothems proved to be challenging material for U-Th dating. Seven out of 28 age determinations were out of the stratigraphic order (marked with suffix * in the Table 1) and were not included in the age-depth models. Furthermore, five of these age determinations did not yield a meaningful result at all. In the case of the MP-2 speleothem, only one age determination was unsuccessful out of sixteen, while for MP-3 half of the age determinations were rejected (i.e., six out of twelve). Isotopic ratio data of the MP-3 stalagmite point to significant U loss and/or Th gain that is likely a result of post depositional alteration in the upper 70 mm of this stalagmite. We employed the stratigraphic constraint method of [66] to determine the initial 230 Th/ 232 Th activity ratio of 1.37 ± 0.26 for these speleothems. Corrected ages were calculated by using the U and Th decay constants of [70].
A Monte-Carlo finite positive growth rate model [65,71,72] was employed to construct two independent age-depth models. The first encompasses age data from the upper (younger) portion of the MP-2 stalagmite. During this time period, MP-2 experienced partial contemporaneous growth along two parallel growth axes (Figure 2a). To make use of all U-Th data collected, we transposed the data from the shorter transect to the longer one and ran a unique age-depth model. The five U-Th dates were transposed onto the longer transect based on visual similarity of the two transects and thanks to the pronounced layering of this speleothem. To account for possible extra uncertainty introduced by this procedure, we doubled the original depth uncertainty for all transposed samples and used this value (presented in Table 1) in the age-depth modelling procedure. The positions of the transposed data are indicated with white boxes in Figure 2a and the age-depth model output is presented in Figure 3a. A Monte-Carlo finite positive growth rate model [65,71,72] was employed to construct two independent age-depth models. The first encompasses age data from the upper (younger) portion of the MP-2 stalagmite. During this time period, MP-2 experienced partial contemporaneous growth along two parallel growth axes ( Figure 2a). To make use of all U-Th data collected, we transposed the data from the shorter transect to the longer one and ran a unique age-depth model. The five U-Th dates were transposed onto the longer transect based on visual similarity of the two transects and thanks to the pronounced layering of this speleothem. To account for possible extra uncertainty introduced by this procedure, we doubled the original depth uncertainty for all transposed samples and used this value (presented in Table 1) in the age-depth modelling procedure. The positions of the transposed data are indicated with white boxes in Figure 2a and the age-depth model output is presented in Figure 3a.   Figure 2b) stalagmites. Dark red and light red shaded areas represent the 68% and 95% confidence intervals, respectively. Table 1. U-Th results for the MP-2 and MP-3 stalagmites. Isotopic ratios are the activity ratios and the uncertainties are expressed as 95% confidence intervals (c.i.). Age is corrected for the initial 230 Th/ 232 Th of 1.37 ± 0.26 following the method of [66]. Results rejected from the age-depth model (outliers) are marked with a suffix *. The second age-depth model spanned the 110 ka to 45 ka time period, and it is partially presented by both speleothems. In order to produce a single age-depth model, we merged the chronology data from both speleothems. Synchronization was performed following the similar procedure published in [17,73]. In the tuning protocol, the MP-2 dataset was used as the main record because it covered the longer time span compared to the MP-3 speleothem. Eight tie points were selected in MP-3, i.e., the MP-2 δ 18 O depth profiles, to transpose δ 18 O and δ 13 C as well as the corresponding U-Th data of the MP-3 speleothem onto the MP-2 depth scale. Tuning and tie point selection was performed based on visual similarity between the two δ 18 O depth series. Linear interpolation was performed in between tie points to transpose all MP-3 data onto the MP2 depth scale. Original depth uncertainty of the MP-3 U-Th samples was increased for at least 25% to account for an extra error introduced by this tuning protocol and this increased uncertainty was then applied in the age-depth modelling procedure. All accepted ages are in stratigraphic order as presented in Figure 3b.

14 C Dating
As pointed out in the previous section, U-Th dating of the uppermost part of the MP-3 speleothem was unsuccessful due to a likely open-system behaviour [74]. Consequently, it was not possible to reliably synchronise this part of MP-3 with MP-2, so those analyses were rejected from the cross-tuning (Figure 3b). To constrain the broad time frame of the youngest deposition phase of MP-3, we employed radiocarbon dating. Its minor role in speleothem geochronology is due to the 'dead ( 14 C-free) carbon' issue [75], where unknown and variable portions of 14 C-depleted carbon derive from dissolved old limestone and aged soil-derived carbon. This is expressed as the % dead carbon fraction (DCF) and it varies not only spatially, but also reflects the environmental changes through time [75,76]. An overview of the published speleothem DCF is given in [77] and it ranges from 5% to 37% (with few special cases, such as in Corchia Cave where it is~60% [76]). Given the measured 14 C activities (Table 2), and unknown DCF, we just estimate the youngest portion of the MP-3 speleothem to the MIS-3 period.   Since the credible hydroclimate interpretation of the stable isotope records requires near-equilibrium isotopic fractionation during calcite precipitation from the dripwater [6,43], we employed three different approaches to assess the equilibrium condition. Confirmation of near-equilibrium precipitation during most of the speleothems' growth is given in Appendix A.  [43]. Given the geographical position of MP cave in the temperate zone, and lack of any palynological evidence, we could exclude alteration from a C3 to C4 plant assemblage during the Quaternary climate changes. In the Mediterranean region, the speleothem δ 13 C has been used as a proxy for the surface vegetation and soil microbial activity, as revealed in surrounding regions (e.g., [17,20,[25][26][27][28]30,[78][79][80]). Lower δ 13 C values are associated with increased soil pCO2 produced by intensive soil microbial activity and root respiration during warmer and wetter conditions and/or presence of forest cover, while higher δ 13 C values reflect reduced bioproductivity due to enhanced aridity. However, a longer residence time of water within the host rock during the arid periods leads to δ 13 C shifting towards higher bedrock-like δ 13 C values, and the same effect derives from prior calcite precipitation (PCP), which is again caused by reduced groundwater percolation within the karst aquifer [43,81]. δ 13 C variations can be discussed in relation to the speleothem fabrics and growth mode alteration, since the carbonate production, which is controlled by ecological factors Since the credible hydroclimate interpretation of the stable isotope records requires near-equilibrium isotopic fractionation during calcite precipitation from the dripwater [6,43], we employed three different approaches to assess the equilibrium condition. Confirmation of near-equilibrium precipitation during most of the speleothems' growth is given in Appendix A.

Environmental
Changes Documented in the δ 13 C Signal δ 13 C values along the MP-2 and MP-3 growth axes were within ranges of −7.8 to 0.04‰ and −7.1 to 1.4‰, respectively. Both speleothems recorded significant enrichment in calcite δ 13 C;~8.5‰ towards the end of MIS 5 in MP-3, and almost 8.0‰ during MIS 3 in MP-2 ( Figure 4), while the lowest δ 13 C values, down to −11.1‰, were measured in the modern calcite samples. Such large δ 13 C shifts (max. 1.4‰ and min. −11.1‰) in spelaean carbonate usually point to shifts between the C3 and C4 plant communities, where higher values (−6‰ to 2‰) reflect the dominance of C4 plants adapted to arid conditions and lower δ 13 C (−14‰ to −6‰) are typical for C3-dominated vegetation [43]. Given the geographical position of MP cave in the temperate zone, and lack of any palynological evidence, we could exclude alteration from a C3 to C4 plant assemblage during the Quaternary climate changes. In the Mediterranean region, the speleothem δ 13 C has been used as a proxy for the surface vegetation and soil microbial activity, as revealed in surrounding regions (e.g., [17,20,[25][26][27][28]30,[78][79][80]). Lower δ 13 C values are associated with increased soil pCO 2 produced by intensive soil microbial activity and root respiration during warmer and wetter conditions and/or presence of forest cover, while higher δ 13 C values reflect reduced bioproductivity due to enhanced aridity. However, a longer residence time of water within the host rock during the arid periods leads to δ 13 C shifting towards higher bedrock-like δ 13 C values, and the same effect derives from prior calcite precipitation (PCP), which is again caused by reduced groundwater percolation within the karst aquifer [43,81]. δ 13 C variations can be discussed in relation to the speleothem fabrics and growth mode alteration, since the carbonate production, which is controlled by ecological factors and depends on the water chemistry and water supply, determines speleothem geometry [82].
Light porous calcite associated with a retractional growth pattern, i.e., a diameter decrease, was ascribed to reduced hydrologic activity, i.e., dry periods [82][83][84], while compact layers draping over the stalagmite flanks mark more humid conditions [85]. Such a speleothem architecture is elaborated in the conceptual model in [7], describing the factors influencing stalagmite properties, and is already known from the nearby Modrič Cave [26] as well as from some Spanish [84,86] and Belgian [85] sites. In our MP-2 sample (Figure 2), alteration of calcite fabrics and speleothem geometry are also consistent with δ 13 C variations; the white porous parts have high δ 13 C values, indicating deprived plant and soil activity in response to enhanced aridity, and are reflected in the narrowing of successive layers, while the darker compact parts, which overlap the preceding layers, coupled with lower δ 13 C, apparently derive from the wetter phases. In MP-3, the major part of the stalagmite consists of white porous calcite that also gradually narrows towards the top (Figure 2), along with increasing δ 13 C values. However, in the subsequent darker layers (in alteration with the lighter ones), δ 13 C also remains high. Similarly, in Tana che Urla Cave (central Italy), brown clastic-rich calcite has high isotopic values connected with lower growth rates during drier periods [20]. Apparently, those dark and compact layers derived from phases with limited water availability and decreased soil activity, which might have enhanced soil erosion and induced a higher flux of clastic material in the cave [87]. Evidence of clastic input into MP-3 stalagmite is clearly obvious in the high concentration of detrital thorium ( 232 Th) ( Table 1), which prevents precise U-Th dating of the youngest part of MP-3. We presume that part was formed during phases of climatic deterioration within MIS 3, when reduced rainfall increased the residence time of percolating groundwater within the host rock, resulting in enriched δ 13 C due to the heavier carbon isotopic composition of the limestone [88]. The dominance of the bedrock isotopic composition (higher δ 13 C), i.e., lack of δ 13 C-lighter soil-derived organic carbon in the system, is best pronounced in LGM-speleothems formed underneath an ice cover [35] or presently in near-freezing conditions [89], where the carbonic acid dissolution (CAD) has been replaced by the sulphuric acid dissolution (SAD) [76] throughout periods when conditions are unfavourable for the production of pedogenic CO 2 . However, in accordance with the local sulphur-free lithology (Upper Jurassic limestone), this consistently high δ 13 C values of the youngest part of MP-3, roughly dated at mid-MIS 3, are presumably the result of CAD, which took place in an environment with minimal but apparently sufficient biological activity. Namely, glacial periods left some landmarks in the surroundings of MP, such as LGM terminal moraine in Rujanska kosa (6 km NW from MP cave) at the present altitude of 920 m a.s.l. [90], and tills reaching a minimum altitude of 800 m a.s.l. [91]. Given the fact that the recharge area and the source of pedogenic CO 2 that reaches MP cave is mostly at the altitude of ca. 700 m a.s.l., this was the glacier-free area, but most likely the periglacial setting maintained low plant and soil bioactivity levels. Hence, the lack of δ 13 C-lighter soil-derived organic carbon, and the apparent aridity that could sustain PCP and prolong the interaction between the percolating groundwater and the host rock, resulted in an overall δ 13 C increase.
As our time constraint based on 14 C dating is not completely confident, the MIS 3 record from adjacent Modrič Cave (9 km SE from MP) might confirm our findings; namely, that stalagmite MOD-21 deposited from 54.6 ka to 44.0 ka recorded the same increasing trend within the δ 13 C range of 8‰, associated with a dry phase, and followed by a nondepositional period [26]. Likewise, growth of MP-3, after the cessation within MIS-3, was not restored.
The other extreme, the lowermost δ 13 C values within the MP cave, were recorded in modern samples (from −11.1‰ to −9.2‰), although the terrain above the cave, in fact the whole recharge area, is presently almost bare bedrock with scarce bush patches (Figure 1b). Such a landscape is the result of anthropogenic and natural processes that led to deforestation due to the overuse of the Velebit Mountain woodlands during the 16th-17th c. by the invaders (Venetians, Otomans and Habsburgs) [92] and growing local population. The Little Ice Age also reinforced forest logging. In spite of reforestation, organized already in the late 19th century [93] and naturally intensified upon the ban of goat breeding in 1950s, the recovery of the woodland has been slow, if happening at all. At the moment, it seems unrealistic that such negative δ 13 C values of modern calcite can be produced by such modest pedologic and vegetation cover; however, this has been repeatedly confirmed. Similarly, a low modern calcite δ 13 C has been measured in adjacent Modrič Cave (32 m a.s.l; 9 km SE from MP) and Strašna peć Cave (74 m a.s.l; 48 km SW from MP). In modern calcite precipitated during 1999-2007 in Modrič Cave, δ 13 C was −8.2‰ and −9.8‰ [48], while in Strašna peć Cave, calcite deposited through 2013-2014 had values from −12.1‰ to −8.7‰ (mean −10.8‰, n = 8) ( [49] and unpublished data). Both sites are also characterized by patches of soil with sparse Mediterranean maquis, slightly more developed than the above MP Cave, which apparently gives enough biological activity to sustain such low δ 13 C values.
The δ 18 O record from the MP cave starts at GS 24, which is, unlike in our record, very well expressed by a positive shift in other regional records, such as from Croatian Mljet Island (MSM) [27], Corchia Cave in the Apuan Alps [100], Apulian Pozzo Cucú Cave (PC) [25], Israeli Soreq and Peqiin caves [33], but also in North Alpine stalagmites (NALPS) [4] and Western Mediterranean marine core GDEC-4-2 [58] (Figure 5). It was obviously an event of broad regional significance characterized by cooling of the SW and S Europe due to the renewed Northern Hemisphere ice-sheet growth after the end of the last interglacial, and a reduction in atmospheric moisture [100]. During the next 15 ka, which in NGRIP is documented by gradual climate deterioration from GI 23 throughout the relatively long GS 23, our MP-3 recorded several δ 18 O excursions, similar to the PC record [25]. Since this part of MP history is supported only by a single speleothem, we refrain from drawing any specific conclusions except that the MP-3 speleothem architecture and fabrics suggest arid conditions.
The following period, from MIS 5c to MIS 4, is instead covered by both MP stalagmites, complemented by the earlier findings from submerged speleothems [101]; namely, one of the consequences of global climate changes is sea-level changes caused by the accumulation and melting of continental ice. In the shallow northern part of the Adriatic Sea, this generated substantial palaeogeography changes throughout the Quaternary due to shifting coastlines ( Figure 6c) and exposing of the carbonate landscape to karstification, including speleothem precipitation. Two such speleothems are K-14 and K-18 from submerged U Vode Pit near Krk Island (96 km NW from MP), whose deposition was repeatedly interrupted by the MIS 5 sea-level fluctuations [101] (Figure 6a,b).
The oldest part of K-18 precipitated from >93 to 90 ka in a similarly narrowing manner as the MP samples by the end of cool and dry GS 23 ( Figure 5). The subsequent GS 22 was even more pronounced as recorded by the δ 18 O increase in MP speleothems and those from Mljet Island, Apulia and Sardinia, and the most prominent decrease in the NALPS record. It was associated with the MIS 5b sea-level fall, enabling K-18 deposition from 90 to 87 ka. Its porous white calcite (Figure 6a) resembles that of MP-3, pointing to the similar environmental setting. An abrupt shift in δ 18 O towards more negative values in Mediterranean speleothems (and towards positive in NALPS) marks the beginning of GI 21 and the MIS 5a sea-level highstand that flooded U Vode Pit and temporarily ceased K-14 and K-18 growth. Towards the MIS 4 glacial, increasing speleothem δ 18 O suggest progressive climate deterioration (cooling and drying) and resulting sea-level fall during which K-18 reactivated. Its dark and compact calcite precipitated from 82 ka to 77 ka-the short period presumably related to the mild stadial of GS 20, which is, along with GS 21, GI 20 and GI 19, relatively well expressed in the MP-3 record. Cave from Mjet Island [27]; (e) Pozzo Cucú Cave [25]; (f) Crovassa Azzurra Cave [24]; (g) Soreq and Peqiin caves in Israel [33]; and (h) marine borehole GDEC-4-2 [59]. Numbers refer to Greenland Interstadials. Dashed lines indicate the most prominent matches of recorded stadials (blue) and interstadials (red). Horizontal bars represent growth episodes of speleothems K-14 (red) and K-18 (blue) recovered from the submerged U Vode Pit (Krk Island) [101].  The oldest part of K-18 precipitated from >93 to 90 ka in a similarly narrowing manner as the MP samples by the end of cool and dry GS 23 ( Figure 5). The subsequent GS 22 was even more pronounced as recorded by the δ 18 O increase in MP speleothems and those from Mljet Island, Apulia and Sardinia, and the most prominent decrease in the NALPS record. It was associated with the MIS 5b sea-level fall, enabling K-18 deposition from 90 to 87 ka. Its porous white calcite (Figure 6a) resembles that of MP-3, pointing to the similar environmental setting. An abrupt shift in δ 18 O towards more negative values in Mediterranean speleothems (and towards positive in NALPS) marks the beginning of GI 21 and the MIS 5a sea-level highstand that flooded U Vode Pit and temporarily ceased K-14 and K-18 growth. Towards the MIS 4 glacial, increasing speleothem δ 18 O suggest progressive climate deterioration (cooling and drying) and resulting sea-level fall during which K-18 reactivated. Its dark and compact calcite precipitated from 82 ka to 77 ka-the short period presumably related to the mild stadial of GS 20, which is, along with GS 21, GI 20 and GI 19, relatively well expressed in the MP-3 record.
The most pronounced temperature drop during the studied period occurred in MIS 4, at least as far as the Greenland temperature is concerned (Figure 5). At its peak (~65 ka), with an ~80 m lower sea level [102], the coastline of the NE Adriatic nearest to the MP cave receded ~70 km away from the present position (3.7 km from the coast). Environmental conditions certainly changed and climate type most probably altered to Df (humid continental), which today dominates the highest peaks of the Velebit Mountain [49]. Nevertheless, the δ 18 O in both the MP-2 and MP-3 speleothems remains relatively low. It is apparent that the glacial climate in the eastern Adriatic, as recently shown on the western coast of Apulia [25], was milder and sufficiently moist for speleothem deposition. The growth episode of speleothem K-18 between 64 ka and 54 ka ( Figure 5) confirms this [101].
The end of MIS 5 was a period of decoupled MP δ 13 C and δ 18 O series, with a general rise in δ 18 O, and not so much increased or stable δ 13 C (Figure 4). Such a situation is regionally interpreted as a time of prevailing source effect over amount effect [24,25], which is The most pronounced temperature drop during the studied period occurred in MIS 4, at least as far as the Greenland temperature is concerned ( Figure 5). At its peak (~65 ka), with an~80 m lower sea level [102], the coastline of the NE Adriatic nearest to the MP cave receded~70 km away from the present position (3.7 km from the coast). Environmental conditions certainly changed and climate type most probably altered to Df (humid continental), which today dominates the highest peaks of the Velebit Mountain [49]. Nevertheless, the δ 18 O in both the MP-2 and MP-3 speleothems remains relatively low. It is apparent that the glacial climate in the eastern Adriatic, as recently shown on the western coast of Apulia [25], was milder and sufficiently moist for speleothem deposition. The growth episode of speleothem K-18 between 64 ka and 54 ka ( Figure 5) confirms this [101].
The end of MIS 5 was a period of decoupled MP δ 13 C and δ 18 O series, with a general rise in δ 18 O, and not so much increased or stable δ 13 C (Figure 4). Such a situation is regionally interpreted as a time of prevailing source effect over amount effect [24,25], which is attributed to the less efficient Atlantic Meridional overturning circulation during ice-sheet growth [18,103]; namely, with lower production of Atlantic moisture and changed westerlies trajectories, the relative proportion of Atlantic (more negative δ 18 O) and Mediterranean (less negative δ 18 O) moisture changes, and the region receives isotopically enriched precipitation. The opposite situation, i.e., covarying δ 13 C and δ 18 O, is ascribed to periods of similar climate control [24]; that is, the amount of precipitation that governs vegetation growth and soil bioactivity. Our record only partially follows this scheme. One of the possible causes of that discrepancy might derive from a reorganized geographical setting, since the MP site with a lowered sea level became a region with more pronounced orographic precipitation instead (or in spite) of enhanced aridity.
According to the δ 18 O variations in MP-2, the transition from MIS 4 to MIS 3 is characterized by short and rapid oscillations matching GI 17 to GS 13. It was also a depositional period of the MSM-1 stalagmite from Mljet Island and K-18 from Krk Island, both of which, however, stopped growing after the dry and cold stadial GS 15, the one also very prominent in Apulia [25] and SE Spain [12]. The same period in MP cave is marked with oscillating and finally increasing δ 13 C, leading to the highest δ 13 C values and therefore quite arid conditions. Meanwhile, in the Apulian karst, a soil bioproductivity plateau was reached (δ 13 C =~−8‰) despite the gradual δ 18 O increase; i.e., a precipitation decrease [25].

MIS 2 to MIS 1 Transition in δ 18 O Record
The transition from MIS 2 to MIS 1 is captured in the youngest part of the MP-2 stalagmite. According to the age-depth model, deposition of the youngest part of MP-2 recommenced around 13.8 ka (Figure 7), and our last time series data is at 5.3 ka (actively growing topmost part was not dated). The δ 18 O values are in range between −5.8‰ and −3.9‰ (average −4.9‰). The lowest δ 18 O value implies relatively warm and humid conditions at 12.57 +1.11 / −1.08 ka, followed by abrupt climate deterioration and high isotopic values at 11.33 +1.15 / −1.10 ka, which may, within the corresponding age uncertainties, correspond to the cool and dry Younger Dryas event (12.9-11.7 ka). However, the apparent delay from the δ 18 O excursion recorded in, e.g., the Corchia [15] and Soreq Cave speleothems [33], as well as in the NGRIP core [2], should be checked in other contemporaneous speleothems or other archives, rather than be prematurely ascribed to specific local conditions.
Geosciences 2021, 11, x FOR PEER REVIEW Figure 7. Comparison of the δ 18 O variation in the younger part of the MP-2 speleothem w contemporaneous δ 18 O speleothem records from the Corchia [15] and Soreq caves [33] and NGRIP δ 18 O data [2], with the prominent excursion ascribed to the Younger Dryas climati

Conclusions
Manita peć Cave preserves speleothem records of environmental conditions last glacial cycle at a very specific location-the littoral part of the northern Medite (eastern Adriatic coast), but at an elevation close to the glaciation limit during th phases. The main features of the changing environment obtained from the δ 18 O records from two partially coeval speleothems can be summarized as follows: • Millennial scale climate events, i.e., Dansgaard-Oeschger (DO) cycles, recon  [15] and Soreq caves [33] and with NGRIP δ 18 O data [2], with the prominent excursion ascribed to the Younger Dryas climatic event.

Conclusions
Manita peć Cave preserves speleothem records of environmental conditions over the last glacial cycle at a very specific location-the littoral part of the northern Mediterranean (eastern Adriatic coast), but at an elevation close to the glaciation limit during the glacial phases. The main features of the changing environment obtained from the δ 18 O and δ 13 C records from two partially coeval speleothems can be summarized as follows: In terms of palaeogeography, the recorded climate changes played a key role in landsea distribution, transforming the near-coastal site of MP into a continental one during the glacial/stadial stages. We assume that such settings during the MIS 4 attenuated aridity of the glacial period by promoting the MP cave into a site receiving somewhat higher amounts of orographic precipitation. • Accordingly, we also assume that switching the predominance between the amount and source effect, which was proven regionally, might be overprinted by local sitespecific features, such as in the MP site. Funding: This research was funded by the University of Zadar (Scientific Project 60200 Reconstruction of the regional palaeoclimate change-speleothem records from the North Dalmatia (Croatia).

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
Stable and U-series isotope data are archived in the SISAL database and will be freely available by the next update (SISAL_v3).

Isotopic Equilibrium
A prerequisite for credible interpretation of palaeoenvironmental changes from δ 18 O and δ 13 C records is near-equilibrium isotopic fractionation during calcite precipitation from the dripwater [6,43]. To assess this, we used three conventional approaches: the Hendy test [6], replication test [104] and a comparison of the measured cave air temperature with that calculated from the modern calcite, according to empirical relationships for watercalcite oxygen isotope fractionation [69]; namely, according to [105], if the modern calcite equilibrium conditions can be confirmed, then we can assume that ancient speleothems from the same cave should also have been precipitated at or near isotopic equilibrium.
Graphical presentations of the Hendy tests performed along three growth laminae on MP-2 and four laminae on MP-3 stalagmite are given on Figures A1 and A2. The first Hendy criterion of the δ 18 O values remaining constant along a single growth layer in support of equilibrium deposition [6] was fulfilled in all cases since the maximal δ 18 O variations were 0.53‰ and 0.49‰ within the MP-2 and MP-3 layers, respectively ( Figure A1), both below the threshold value of 0.8‰ defined by [106]. The only slight enrichment in calcite δ 18 O with distance from the growth axis was observed in the MP-3 H3 series. The second criterion of no correlation between the δ 18 O and δ 13 C values along a single lamina [6] was also satisfied ( Figure A2), with the exception of laminae H1 in MP-2 and H3 in MP-3, which returned Pearson's correlation coefficients of r = 0.80 and r = 0.83, respectively, while their p-value describing the statistical significance of the correlations were p < 0.05, pointing to significant correlations. However, possible covariation in certain growth phases is not exclusively proof of kinetic fractionation, since both δ 18 O and δ 13 C can be controlled by climate (via hydrological response and plant/soil activity) and thus might covary [104]. can be controlled by climate (via hydrological response and plant/soil activity) and thus might covary [104]. Figure A1. δ 18 O and δ 13 C values along the MP-2 and MP-3 growth layers selected for the Hendy tests [6]. For the location of the sampling tracks, see Figure 2. Equilibrium deposition is presumed due to the lack of a significant trend in the isotopic ratios towards heavier isotopic values away from the growth axis.
Since the Hendy test is not considered as explicit proof, the replication test is more reliable [104] and was used to additionally support the independence of isotopic composition on kinetic processes; namely, similarity of the isotopic profiles could have occurred either because the kinetic and/or vadose zone processes affected both speleothems equally (which is hardly possible) or, more likely, it was absent [104]. As the resemblance of δ 18 O shifts between an older part of MP-2 and MP-3 is evident ( Figure A1), we can assume that, at least during that period, precipitation of calcite was close to isotopic equilibrium with cave water, and spelaean δ 18 O was controlled exclusively by the cave temperature and water δ 18 O, in response to palaeoenvironmental changes above the MP cave.
Knowing the oxygen values of the modern calcite (δ 18 OC) and dripwater (δ 18 OW), an additional proof of near-equilibrium condition can be provided based on the matching of the measured cave air temperature (Tm) with the one calculated (Tc) after the empirical relationships for water-calcite oxygen isotope fractionation given by the equation 1000 ln α = 16.1(10 3 T −1 ) − 24.6 [69], where α = (1000 + δ 18 OC)/(1000 + δ 18 OW). As established earlier in [49] using four modern MP calcite samples, the Tc range from 9.1 ± 0.1 °C to 10.4 ± 0.1 °C matches well the Tm = 9.04 °C, giving further confidence in the isotopic record as a Figure A1. δ 18 O and δ 13 C values along the MP-2 and MP-3 growth layers selected for the Hendy tests [6]. For the location of the sampling tracks, see Figure 2. Equilibrium deposition is presumed due to the lack of a significant trend in the isotopic ratios towards heavier isotopic values away from the growth axis.
Since the Hendy test is not considered as explicit proof, the replication test is more reliable [104] and was used to additionally support the independence of isotopic composition on kinetic processes; namely, similarity of the isotopic profiles could have occurred either because the kinetic and/or vadose zone processes affected both speleothems equally (which is hardly possible) or, more likely, it was absent [104]. As the resemblance of δ 18 O shifts between an older part of MP-2 and MP-3 is evident ( Figure A1), we can assume that, at least during that period, precipitation of calcite was close to isotopic equilibrium with cave water, and spelaean δ 18 O was controlled exclusively by the cave temperature and water δ 18 O, in response to palaeoenvironmental changes above the MP cave.
Knowing the oxygen values of the modern calcite (δ 18 O C ) and dripwater (δ 18 O W ), an additional proof of near-equilibrium condition can be provided based on the matching of the measured cave air temperature (T m ) with the one calculated (T c ) after the empirical relationships for water-calcite oxygen isotope fractionation given by the equation 1000 ln α = 16.1(10 3 T −1 ) − 24.6 [69], where α = (1000 + δ 18 O C )/(1000 + δ 18 O W ). As established earlier in [49] using four modern MP calcite samples, the T c range from 9.1 ± 0.1 • C to 10.4 ± 0.1 • C matches well the T m = 9.04 • C, giving further confidence in the isotopic record as a reliable source for palaeoenvironmental reconstruction.