Local Differentiation in the Loess Deposition as a Function of Dust Source: Key Study Novo Orahovo Loess Paleosol Sequence (Vojvodina, Serbia)

: Typical patterns of the Late Pleistocene loess–paleosol units are preserved in the Novo Orahovo brickyard, Northern Serbia. Presented preliminary luminescence chronology supports the chronostratigraphic interpretations of global isotopic marine climate reconstructions. Magnetic susceptibility and sedimentological evidence exhibit general similarities with the marine oxygen-isotope stratigraphy. These records provide new insights into the dust accumulation regimes over the eastern side of the Baˇcka loess plateau and offer new paleoenvironmental information for the region. They represent an important step forward towards the establishment of a catena from the thin loess-like sediments of the Banat foothills in the east towards the thicker and seemingly more complete loess sections of the southeastern and central Carpathian Basin. Grain-size data from the loess record of Nova Orahovo explain the regional differentiation in dust deposition.

Despite the recent extensive investigations in the southeastern part of the Carpathian Basin [1][2][3]6,9], some local-to-regional elements of the Late Pleistocene environmental dynamics still remain unresolved. To fill these gaps related to the last glacial environmental reconstructions of that region, we investigated the Novo Orahovo (Vojvodina, Serbia) loess section. We present the preliminary absolute luminescence chronology, magnetic, and sedimentological records of the Late Pleistocene climate and environment variations from the Novo Orahovo loess-paleosol sequence, located on the interfluve area between the Danube and Tisa (Tisza) rivers at the Bačka loess plateau (LP) in Northern Serbia ( Figure  1).

Material and Methods
The Novo Orahovo brickyard is situated in Northern Serbia (latitude N 45°51′55″ and longitude Е 19°48′01″) in the Bačka LP, between the alluvial plains of the Danube and Tisa rivers. The investigations of the loess-paleosol sequences at the Novo Orahovo quarry began in 2017. After careful cleaning and a detailed description of the investigated sections, samples were collected for color determination, magnetic susceptibility (MS), and grain-size analysis and luminescence dating. Dry and moist colors of the loess and paleosol units were characterized using Munsell Soil Color Charts. Bulk sediment samples were collected at 5 cm intervals for texture analysis.
The MS was measured using a Bartington MS2 B Dual Frequency Sensor instrument-36 mm sample cavity diameter, at the Laboratory of Physical Geography, Faculty of Sciences, University of Novi Sad, Serbia. The samples were first placed in non-magnetic plastic boxes, after which they were carefully compressed with a non-magnetic plastic pusher. Before closing the box, cotton wool was placed to prevent any movement of the material during the measurement.
The grain-size (GS) distribution of samples was carried out on a Fritsch Analysette 22 MicroTec laser diffraction grain-sizer at the Department of Geoinformatics, Physical and Environmental Geography, University of Szeged. The device is equipped with a green (λ = 532 nm, p = 7 mW) and an infrared (λ = 940 nm, p = 9 mW) laser, and has a measurement

Material and Methods
The Novo Orahovo brickyard is situated in Northern Serbia (latitude N 45 • 51 55 and longitude E 19 • 48 01 ) in the Bačka LP, between the alluvial plains of the Danube and Tisa rivers. The investigations of the loess-paleosol sequences at the Novo Orahovo quarry began in 2017. After careful cleaning and a detailed description of the investigated sections, samples were collected for color determination, magnetic susceptibility (MS), and grain-size analysis and luminescence dating. Dry and moist colors of the loess and paleosol units were characterized using Munsell Soil Color Charts. Bulk sediment samples were collected at 5 cm intervals for texture analysis.
The MS was measured using a Bartington MS2 B Dual Frequency Sensor instrument-36 mm sample cavity diameter, at the Laboratory of Physical Geography, Faculty of Sciences, University of Novi Sad, Serbia. The samples were first placed in non-magnetic plastic boxes, after which they were carefully compressed with a non-magnetic plastic pusher. Before closing the box, cotton wool was placed to prevent any movement of the material during the measurement.
The grain-size (GS) distribution of samples was carried out on a Fritsch Analysette 22 MicroTec laser diffraction grain-sizer at the Department of Geoinformatics, Physical and Environmental Geography, University of Szeged. The device is equipped with a green (λ = 532 nm, p = 7 mW) and an infrared (λ = 940 nm, p = 9 mW) laser, and has a measurement range of 0.08-2000 µm. Sample preparation followed the procedures detailed in Kun et al. (2013) [32] and Serban et al. (2015) [33]. Furthermore, no chemical dispersion was applied to avoid the modifying effect of the dispersant on the measurements. A longer, 180 s ultrasonic pretreatment (f = 36 kHz, p = 60 W) was applied instead [30]. To generate Quaternary 2023, 6, 23 3 of 15 grain-size distribution curves, the laser diffraction data were processed according to the Mie optical theory, using the following parameters: refraction index of 1.52 and absorption index of 0.1 for the dispersed sample and refraction index of 1.33 for water [34]. Apart from specific characteristic grain-size ranges, we also calculated the U-ratio. The latter sedimentary proxy expresses the proportion of medium-and coarse-grained silt (44-16 µm) versus fine-sized silt (16-5.5 µm) [35]. It especially enables the elimination of potentially present pedogenic (mostly submicron) clay.
The samples for luminescence dating were obtained by hammering stainless steel tubes into the freshly cleaned profile. In total, 20 samples were collected. Here, we present the initial dating results of 5 quartz samples from the upper 610 cm of the profile. The inner material of the cylinders was used for equivalent dose measurements (conducted at the Luminescence Dating Laboratory, University of Szeged, Hungary) and the outer material (~2 cm) for the present water content and radionuclide activity determination. The preparation of samples was conducted according to standard laboratory techniques [36]. All procedures were tested in subdued yellow light conditions. Selected samples were wet sieved using 90, 63, and 40 µm sieves. The <40 µm fraction was placed into Atterberg settling cylinders to separate the 4-11 µm silt fraction. However, none of the samples contained a sufficient amount of fine silt for further processing; therefore, in a second run, the 11-20 µm medium silt fraction was separated. Carbonate and organics were removed by repeated treatment in 10% HCl and 10% H 2 O 2, respectively. To isolate the quartz grains, a portion of the 11-20 µm fraction was immersed in 35% H 2 SIF 6 for 10 days. Subsequently, the quartz grains were treated with 30% HCl to remove any remaining fluorides.
The equivalent dose (De) of samples was determined on a Risø DA-20 TL/OSL-type luminescence reader equipped with a 90SR/90Y β-source and an EMI ET9107-type photomultiplier [37]. Stimulation was carried out using a blue (470 ± 30 nm) LEDs set for 90% power release, while detection was made through a Hoya U-340 filter. Throughout the measurements, the single aliquot regeneration (SAR) protocol was applied [38,39]. Subsequently, preheat plateau and dose recovery tests were performed to identify the optimal measurement parameters and assess the reproducibility of measurements. Appropriate stimulation temperatures were determined using preheat plateau and dose recovery tests on sample 1141. The initial 0.4 s minus a late background subtraction from the last 30 s was used when plotting dose points.
Natural aliquots were first bleached using blue LEDs at room temperature for 100 s, followed by a 3000 s pause, and subsequently further bleached at room temperature for 100 s [38]. Finally, a laboratory β-dose of 50 Gy was administered to the bleached aliquots and measured using a SAR protocol in which the preheating temperature varied between 180 and 300 • C, while the cut-heat temperature was fixed at 160 • C.
During OSL measurements, optical stimulation was performed at 125 • C (heating rate: 5 • C/s) for 40 s. Standard rejection criteria were used to select aliquots that performed well during the SAR measurements [38]. Potential feldspar contamination of the quartz extracts was monitored by the IR/OSL depletion ratio, as proposed by Duller (2003) [40]. In the case of each sample, at least 24 aliquots were analyzed. Dose-response curves were fitted in each case using the single saturating exponential function. The α-value was directly determined by recovering known α doses with β regeneration measurements on aliquots that previously went through a SAR procedure [41]. A representative dose-response curve is presented in Figure 2b.
The environmental dose rate (D*) was determined using high-resolution, extendedrange gamma-ray spectrometry (Canberra XtRaCoaxial Ge detector). Dry dose rates were calculated using the conversion factors of Liritzis et al. (2013) [42]. The alpha and beta attenuation factors were estimated using the calculations of Brennan et al. (1999) [43] and Brennan (1999) [44], respectively, with dose rate conversion factors obtained from Guérin et al. (2012) [45]. Uncertainties are based on the propagation, in quadrature, of individual errors for all measured quantities, which, if unknown, are recorded as 10%. The cosmic ray contribution to the total dose rate was calculated according to Prescott and Hutton (1994) [46] for each sample as a function of depth, altitude, and geomagnetic latitude, assuming an uncertainty of 10%. The total dose rate was calculated using DRAC version 1.2 [47]. The environmental dose rate (D*) was determined using high-resolution, extendedrange gamma-ray spectrometry (Canberra XtRaCoaxial Ge detector). Dry dose rates were calculated using the conversion factors of Liritzis et al. (2013) [42]. The alpha and beta attenuation factors were estimated using the calculations of Brennan et al. (1999) [43] and Brennan (1999) [44], respectively, with dose rate conversion factors obtained from Guérin et al. (2012) [45]. Uncertainties are based on the propagation, in quadrature, of individual errors for all measured quantities, which, if unknown, are recorded as 10%. The cosmic ray contribution to the total dose rate was calculated according to Prescott and Hutton (1994) [46] for each sample as a function of depth, altitude, and geomagnetic latitude,

Lithostratigraphy and Pedostratigraphy
The stratigraphic framework of loess-paleosol sequences from the Vojvodina region, Northern Serbia, is generally simple [2]. Aeolian dust deposition happened on a nearly horizontal platform of the Bačka LP, which exhibits a geomorphologically similar setting as that of the Chinese Loess Plateau [48][49][50][51]. Following the criteria presented by  [3], we describe, stratigraphically characterize, and label the loess and paleosol units at the Novo Orahovo loess section (see Figure 3 and Table 1).  The uppermost part of the penultimate loess unit L2 is just 75 cm exposed in the lower part of its profile and correlates to the latest MIS 6. Many small carbonate spherical concretions (1-2 cm diameter) and humic infiltration in the former root channels were observed close to the contact with the overlaying paleosol S1 (developed during MIS 5). The last interglacial pedocomplex, S1, is 155 cm thick and shows a gradual decrease in pedogenic intensity over time. The 35 cm thick basal transitional AB horizon (10YR6/2-4) is strongly developed and characterized by a granular and partly blocky structure. Above this pedohorizon, two humic granular horizons Ah1 and Ah2 have been developed at a profile depth between 890 cm and 820 cm. Finally, the relatively weakly developed uppermost A horizon is 50 cm thick and strongly affected by numerous krotovinas. This pedogetic horizon represents a gradual weakening of pedogenesis and gently transforms into the overlaying thick loess unit L1. This last glacial composite loess unit is 680 cm thick in total. The lower loess subunit L1LL2 is 265 cm thick, porous, and represents a typical loess. It is interbedded by sandy layers, containing numerous krotovinas, and is in some parts bioturbated. The middle paleosol L1SS1 is developed at a profile depth between 515 cm and 450 cm. This is a slightly developed fossil chernozem-like paleosol (A horizon) with intensive granular structure and krotovinas. The uppermost loess subunit L1LL1 is 360 cm thick. This typical porous pale-yellow loess is intercalated by two embryonic weakly developed paleosols at approximate depths of 3.75 cm (L1LL1SSS2) and 1.75 cm The uppermost part of the penultimate loess unit L2 is just 75 cm exposed in the lower part of its profile and correlates to the latest MIS 6. Many small carbonate spherical concretions (1-2 cm diameter) and humic infiltration in the former root channels were observed close to the contact with the overlaying paleosol S1 (developed during MIS 5). The last interglacial pedocomplex, S1, is 155 cm thick and shows a gradual decrease in pedogenic intensity over time. The 35 cm thick basal transitional AB horizon (10YR6/2-4) is strongly developed and characterized by a granular and partly blocky structure. Above this pedohorizon, two humic granular horizons Ah1 and Ah2 have been developed at a profile depth between 890 cm and 820 cm. Finally, the relatively weakly developed uppermost A horizon is 50 cm thick and strongly affected by numerous krotovinas. This pedogetic horizon represents a gradual weakening of pedogenesis and gently transforms into the overlaying thick loess unit L1. This last glacial composite loess unit is 680 cm thick in total. The lower loess subunit L1LL2 is 265 cm thick, porous, and represents a typical loess. It is interbedded by sandy layers, containing numerous krotovinas, and is in some parts bioturbated. The middle paleosol L1SS1 is developed at a profile depth between 515 cm and 450 cm. This is a slightly developed fossil chernozem-like paleosol (A horizon) Quaternary 2023, 6, 23 6 of 15 with intensive granular structure and krotovinas. The uppermost loess subunit L1LL1 is 360 cm thick. This typical porous pale-yellow loess is intercalated by two embryonic weakly developed paleosols at approximate depths of 3.75 cm (L1LL1SSS2) and 1.75 cm (L1LL1SSS1), respectively. The top of the loess subunit L1LL1 is heavily bioturbated at the contact with the modern pedocomplex S0. This Holocene soil has a thickness of approximately 90 cm. The lower Ck horizon contains many CaCO 3 nodules of 1-5 cm in diameter and former root channels were infiltrated by upper humic material. The transitional AC horizon (10YR 5/1 3/3) is 15 cm thick, consisting of very porous silt loam with a granular structure. At the top, typical granular Ah horizon (10YR 6/3 4/4) remarkable krotovinas are present. Table 1 shows the detailed morphological description of the Novo Orahovo loess and paleosol units.

Magnetic Susceptibility (MS)
MS records within the Novo Orahovo loess-paleosol sequence mostly coincide with lithologic and pedologic variations reflecting the changes in past climate and environmental conditions. The magnetic record supports the SPECMAP marine oxygen-isotope record of the past 150,000 years [52,53]. Figure 3 shows MS records at the Novo Orahovo section. The values of MS are related to pedogenetic processes and reflect differences in composition, concentration, and particle size of magnetic minerals between interglacial, interstadial, and stadial loess and paleosol units [54,55]. MS variations in the interglacial pedocomplexes S1 (average 60.5 SI units) and S0 (average 34.2 SI units) are higher than those measured in loess layers L2 (average 15.6 SI units), L1LL2 (average 188 SI units), and L1LL1 (average 20.8 SI units). The weakly developed interstadial fossil soil L1SS1 has relatively low MS values (average 25.6 SI units) compared to the interglacial paleosols; they are only slightly higher than the values measured in loess strata. Two initial pedogenic layers within loess unit L1LL1 are characterized by similar MS values as observed in loess deposits. The MS variations recorded at the Novo Orahovo loess-paleosol sequence generally display a great similarity to other investigated loess sections in the Vojvodina region [2,3]. This type of magnetic signal enhancement via soil formation is also similar to those observed in Chinese and Central Asian loess deposits [54][55][56][57].

Grain-Size (GS) Variations
Variations in the GS distribution in comparison to the pedostratigraphy of the Novo Orahovo loess-paleosol sequence are presented in Figure 4. Generally, the GS record at the Novo Orahovo section corresponds less to pedostratigraphy than other previously investigated sections [8, 17,18]. The loess grain-size distribution shows a minor clay subfraction at all levels (e.g., samples in Figure 4 at 415 cm (L1LL1) 6.9% and 725 cm (L1LL2) 6.8% depth), while in the paleosols, a slight but distinctive increase of the clay subfraction is present: a slight increase in the weak subsoil at 565 cm in L1SS1 (e.g., Figure 5 at 565 cm depth) reaching 11%. Further along, L2 (below 950 cm) shows a high clay content of 22% ( Figure 4).   In contrast, U-values remain relatively similar in all loess units and paleosols (only a slightly smaller modal value at 965 cm). The U-ratio of the Holocene soil at the Novo Orahovo loess profile has an average value of 1.9, which was also an average value during the Early glacial. The maximum U-value of S0 is present in the first 10 cm of the profile (2.2). The average Late Pleniglacial U-ratio is slightly higher in the depth intervals of 235 cm to 265 cm and in between 600 cm and 700 cm. A slight drop of the U-ratio is present in the Middle Pleniglacial (average value of 2.2-2.5). The last interglacial phase has, on average, slightly lower values in paleosol S1. Values of the U-ratio are increasingly moving from the top of the profile towards approximately 250 cm depth, after which they decrease continuously until the base of the L1LL1 loess unit. From there, the U-ratio is steadily increasing towards the middle part of L1LL2 and decreases again until a depth of approximately 850 cm in pedocomplex S1. The last interglacial pedocomplex, S1, is characterized by relatively high U-ratio values compared with the transition to the penultimate glacial loess unit, L2 (Figure 4).  The lowest values of grain sizes < 2 µm occur in the loess layers with a mi value of 6% in L1LL1 and L1LL2. The percentage increases with pedogenesis, as L1SS1, and S1 have maximum values of 11%, 10%, and 22%, respectively. The lates corresponds to the AB horizon of the Early glacial. The <10 µm curve is in good agr with the <2 µm fraction, as lower values follow the Early and Late Pleniglacial w and 19%, respectively. This granulation is dominant at a depth of 9.35 m, where prises up to 50% of the sample (this single sample might also be considered as an tional outlier). The Middle Pleniglacial is more pronounced with the maximum v 32%, reaching almost the value of the Holocene soil. The next three curves of >15 µ µm, and >63 µm content display inversed values to the previous two. The maximu ues of the grain-size fraction >15 µm is seen in the loess horizon L1LL2 (73%) a The lowest values of grain sizes <2 µm occur in the loess layers with a minimum value of 6% in L1LL1 and L1LL2. The percentage increases with pedogenesis, as the S0, L1SS1, and S1 have maximum values of 11%, 10%, and 22%, respectively. The latest value corresponds to the AB horizon of the Early glacial. The <10 µm curve is in good agreement with the <2 µm fraction, as lower values follow the Early and Late Pleniglacial with 18% and 19%, respectively. This granulation is dominant at a depth of 9.35 m, where it comprises up to 50% of the sample (this single sample might also be considered as an exceptional outlier). The Middle Pleniglacial is more pronounced with the maximum value of 32%, reaching almost the value of the Holocene soil. The next three curves of >15 µm, >31 µm, and >63 µm content display inversed values to the previous two. The maximum values of the grain-size fraction >15 µm is seen in the loess horizon L1LL2 (73%) and the minimum is at paleosol S1 (36%). The values of >31 µm fractions drop to 9%, 19%, and 17% in the soil units S1, L1SS1, and S0, respectively. Higher values are present in the glacial periods. The minimum values of the grain sizes >63 µm reach 0% in soils, while the maximum value of 18% is present in the Early Pleniglacial, indicating fine sand bands.

Luminescence Geochronology
The De value did not show any significant sensitivity to preheat temperature up to 300 • C, with the recycling ratios close to unity and the recuperation close to 1% over the 160-300 • C first preheat interval. The dose recoveries given various preheats are summarized in Figure 2a. The overall average measured-to-given-dose ratio is 1.03 ± 001 (n = 21). The results show that at any given temperature over the 160-300 • C first preheat interval, the SAR protocol was able to recover a given dose within 10% of unity. However, no clear plateau could be observed over the entire 160-300 • C preheat temperature interval. For this reason, a preheat temperature of 240 • C, which gave the best measured dose ratio, is chosen for all consecutive measurements. Table 2 presents results for the U, Th, and K concentrations, water content, calculated total dose rates, and ages. The calculated OSL ages range from 15.9 ± 0.5 ka to 47.8 ± 1.6 ka, indicating that the upper 610 cm of the Novo Orahovo loess section formed during Marine Isotope Stages MIS 2-MIS 4. The ages show a consistent increase and display no jumps nor inversions outside error margins. Sample 1151, taken below the boundary between S0 and L1 (110 cm depth), was dated at 15.9 ± 0.5 ka, suggesting accumulation during the late phase of MIS 2. This is consistent with other dated loess sections in the region, as similar ages of 13.6 ± 1.0 ka and 10.2 ± 0.7 ka for the upper part of L1 were reported for the Titel LP and the Irig loess site, respectively [9,58].  [60] also reported comparative ages of 23.3 ± 1.5 and 23.6 ± 1.4 ka, respectively, for the L1L1 loess at the Stari Slankamen section. The lowermost sample 1138 taken at the L1S1L1LL2 boundary yielded an age of 47.8 ± 1.6 ka. This suggests that the formation of the L1SS1 paleosol started somewhat later in MIS 3 and fits with the expected age for this paleosol layer. Most recently, similar ages for the L1S1SS1 paleosol (27-38 ka) were observed at the Irig loess-paleosol sequence [57]. Comparable ages for the upper paleosols in the Vojvodina region have also been reported for numerous loess sites: Surduk: 31.8 ± 3.7 ka [61], Stari Slankamen: 34.4 ± 2.2 ka [59], Crvenka: 38 ± 4 ka [6], and the Titel LP: 34.2 ± 2.4 ka [9].
As demonstrated above, the calculated OSL ages for the upper 610 cm for the Novo Orahovo loess profile show a remarkable agreement with the ages from several investigated loess-paleosol sequences in the Vojvodina region. This implies that the loess accumulation and paleosol formation had a similar trend in this region.

Discussion
The Novo Orahovo loess site is located further to the northeast compared to other similar previously investigated Northern Serbian loess sections. Figure 6 shows the relationship between low-field MS records of investigated loess sites in Vojvodina and their correlation with the marine oxygen-isotope stratigraphy [62]. Variations in MS correlate closely with the pedostratigraphy of the analyzed exposures, providing a similar pattern of variation, and the main MS peaks can be easily correlated between the sections. The general pattern of magnetic records of Irig, Ruma, Batajnica, Stari Slankamen, Mošorin (Titel loess plateau), Crvenka, and Novo Orahovo sections is characterized by high MS values in strongly developed interglacial soils S0 and S1, low MS values in loess layers L1LL1, L1LL2, and L2, and slightly higher values in weakly developed and morphologically variable L1SS1 pedocomplex.
closely with the pedostratigraphy of the analyzed exposures, providing a similar pattern of variation, and the main MS peaks can be easily correlated between the sections. The general pattern of magnetic records of Irig, Ruma, Batajnica, Stari Slankamen, Mošorin (Titel loess plateau), Crvenka, and Novo Orahovo sections is characterized by high MS values in strongly developed interglacial soils S0 and S1, low MS values in loess layers L1LL1, L1LL2, and L2, and slightly higher values in weakly developed and morphologically variable L1SS1 pedocomplex. The geographical position of the Novo Orahovo site provides a unique opportunity to reconstruct climate and environmental evolution in the transitional area between the southeastern limit and central part of the Carpathian (Pannonian) Basin. While the paleoclimatic and paleoenvironmental reconstruction derived from the loess-paleosol sequences at the Novo Orahovo brickyard generally displays a similar stratigraphic pattern and magnetic variations observed in other investigated sites in the Vojvodina region [2,3] (Figure 6), there are also some distinct differences in glacial-interglacial GS distributions compared to other similar sites in the region. For example, at the Novo Orahovo section, the Middle Pleniglacial warm climatic conditions are reflected in an increase of MS values in weakly developed pedocomplex L1SS1. Compared to this, higher MS values in the upper part of pedocomplex L1SS1 are observed at the Irig, Ruma, Batajnica, and Stari Slankamen sections situated at the Srem LP. The MS pattern of the Veliki Surduk section at Titel LP and Crvenka at the southern limit of Bačka LP, located in between the Srem loess plateau and the Novo Orahovo section, show more uniform values. Further investigations will need to be tailored to address the question: do these changes in MS variations represent regional paleoclimatic gradient or is this just a consequence of some local environmental exception related to the Novo Orahovo loess-paleosol sequence? The mean grain size at Novo Orahovo is slightly coarser than in other sections in Vojvodina. The modal values at Novo Orahovo are 29-37 µm from the top to the bottom of S1. At Irig, the mode is 22-32 µm (except two samples in the L1LL1 loess unit) [58], while at Ruma [8], the mode is 24-29 µm. It is not likely that the wind strength increases would be equal in both glacial and interglacial periods at Novo Orahovo in contrast to other places in the Vojvodina region and also in China (e.g., the Luochuan type loess site [8]). Therefore, the grain size of the source region may be the reason for these conditions. Additionally, loess and paleosol units between Irig and Ruma show variable morphological characteristics of recent and two younger fossil soils (L1SS1 and S1). Vandenberghe et al. (2014) [8] interpreted these The geographical position of the Novo Orahovo site provides a unique opportunity to reconstruct climate and environmental evolution in the transitional area between the southeastern limit and central part of the Carpathian (Pannonian) Basin. While the paleoclimatic and paleoenvironmental reconstruction derived from the loess-paleosol sequences at the Novo Orahovo brickyard generally displays a similar stratigraphic pattern and magnetic variations observed in other investigated sites in the Vojvodina region [2,3] (Figure 6), there are also some distinct differences in glacial-interglacial GS distributions compared to other similar sites in the region. For example, at the Novo Orahovo section, the Middle Pleniglacial warm climatic conditions are reflected in an increase of MS values in weakly developed pedocomplex L1SS1. Compared to this, higher MS values in the upper part of pedocomplex L1SS1 are observed at the Irig, Ruma, Batajnica, and Stari Slankamen sections situated at the Srem LP. The MS pattern of the Veliki Surduk section at Titel LP and Crvenka at the southern limit of Bačka LP, located in between the Srem loess plateau and the Novo Orahovo section, show more uniform values. Further investigations will need to be tailored to address the question: do these changes in MS variations represent regional paleoclimatic gradient or is this just a consequence of some local environmental exception related to the Novo Orahovo loess-paleosol sequence? The mean grain size at Novo Orahovo is slightly coarser than in other sections in Vojvodina. The modal values at Novo Orahovo are 29-37 µm from the top to the bottom of S1. At Irig, the mode is 22-32 µm (except two samples in the L1LL1 loess unit) [58], while at Ruma [8], the mode is 24-29 µm. It is not likely that the wind strength increases would be equal in both glacial and interglacial periods at Novo Orahovo in contrast to other places in the Vojvodina region and also in China (e.g., the Luochuan type loess site [8]). Therefore, the grain size of the source region may be the reason for these conditions. Additionally, loess and paleosol units between Irig and Ruma show variable morphological characteristics of recent and two younger fossil soils (L1SS1 and S1). Vandenberghe et al. (2014) [8] interpreted these geomorphological forms as relicts of dry valleys (dels) which potentially served as local sources for loess transport. Ludwig et al. (2021) [61] found that in the Carpathian Basin during the Last Glacial Maximum peak, dust emissions in spring coincide with the highest wind speeds and relatively low precipitation amounts.
The similarity between the modal values and U-ratios of loess and paleosols at Novo Orahovo supports the idea that loess deposition was restricted or absent during interglacials, as was also demonstrated at the Ruma loess site [8]. Since U-values, displaying grain sizes >5 µm, reflect the dust properties, the general stability of U-values in both loess and paleosols point to a similar source area and transport process in loess as well as in paleosols.
The U-ratios in the Novo Orahovo section are between 1.1 in paleosol layers and 2.5 in loess units. These values are generally similar to other Serbian loess sites such as Ruma (U-ratio varying between 1.6 and 2.2) [8], Irig (U-ratio varying between 1.4 and 2.0) [10], and Dupljaja (U-ratio varying between 1.4 and 2.1), as well as in the famous Chinese type loess section in Luochuan, with a U-ratio that varies between 1.0 and 2.2 [8,51]. The stability and relatively high values of the modal size at Novo Orahovo contrast with many sites in plateau position in the Vojvodina region [2,5,49,60,67] and China [51].
Novo Orahovo's geomorphological position provides a reasonable explanation for the described regional variability in grain-size records. The site is located near theČik River valley, and also in the vicinity of recent and paleo-floodplains of the Tisza and Mures rivers to the east, as well as the dune fields to the north, as potential sources of aeolian material (Figure 1). We conclude that the regional geomorphological position of the site-more specifically, the relative distance to source areas for loess supply-provides a good explanation for the regional differences of dustiness, as expressed by loess accumulation rate and grain-size variability (especially the modal size) of the loess cover. This is in agreement with the conclusions by Stevens et al. (2020) in England [68].
Finally, Novo Orahovois is located in the modern, most arid, part of the Carpathian Basin ( Figure 7A). Gavrilov et al. (2021) [69] provided a map of spatial distribution of aridity (the De Martonne Aridity Index) based on data from 78 meteorological stations in Hungary and Vojvodina during the period between 1949 and 2017. Aridity is most pronounced during the summer season ( Figure 7B). Several previous studies have suggested that current climatic gradients in the region are similar to those during the Late Pleistocene period [2,3,[15][16][17]66]. These arid local climate conditions associated with sparse vegetation make the surroundings of Novo Orahovo sensitive to the emission of material available for further aeolian action. Biotic proxy records from other Bačka LP sites, such as Madaras [70][71][72], Katymár [70][71][72][73][74], and Crvenka [6,21,22], also support previous interpretations. A novel n-alkenes paleoenvironmental record provides evidence for the continued dominance of grasslands during the Late Pleistocene at the Crvenka loess site, indicating more arid climate during the interglacial than glacial periods [6,21,22]. In spite of the general predomination of different grassland environments at Bačka LP, analyzed biotic proxies also show that stadials were more humid than interstadials, leading to the expansion of arboreal plants influencing reduced erosion and aeolian input in cooler phases [6,21,22,[71][72][73][74].

Conclusions
Investigations of the loess-paleosol sequence at the Novo Orahovo brickyard have established the importance of this site as a record of Late Pleistocene climate and environmental dynamics in the southeastern part of the Carpathian Basin.
The loess record at Novo Orahovo provides an opportunity to reconstruct local and regional environmental processes and conditions during the past ∼150,000 years. Sedimentological, pedological, and magnetic evidence recorded in Novo Orahovo show many

Conclusions
Investigations of the loess-paleosol sequence at the Novo Orahovo brickyard have established the importance of this site as a record of Late Pleistocene climate and environmental dynamics in the southeastern part of the Carpathian Basin.
The loess record at Novo Orahovo provides an opportunity to reconstruct local and regional environmental processes and conditions during the past ∼150,000 years. Sedimentological, pedological, and magnetic evidence recorded in Novo Orahovo show many similarities with previously investigated loess sites in the Vojvodina region, indicating relatively stable Late Pleistocene environmental conditions at these discontinuous loess plateaus. However, the observed specific grain-size composition of the Novo Orahovo loess-paleosol sequence is a consequence of its specific geomorphological position and most likely existence of arid climate conditions. These environmental conditions promote higher availability of the source material for further aeolian action. Additionally, existing biotic proxies from other investigated sites at Bačka LP indicate the standing of favorable local vegetation conditions for dust uptake during the whole Late Pleistocene.