INTRODUCTION
Research of aeolian dunes and coversands, which are a common constituent of the landscape of Central Europe, contributes significantly to the acquisition of knowledge on Late Pleistocene and Holocene environmental change. Radiocarbon (14C) dating of paleosols buried in aeolian sandy deposits remains the centerpiece of such studies in Poland and adjacent countries for more than 40 years (Sokołowski et al. Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022 and references therein). However, as new outcrops become available, the need for conducting further 14C age measurements persists (Tolksdorf and Kaiser Reference Tolksdorf and Kaiser2012; Buró et al. Reference Buró, Lóki, Győri, Nagy, Molnár and Négyesi2019, Reference Buró, Négyesi, Varga, Sipos, Filyó, Jull and Molnár2022; Kappler et al. Reference Kappler, Kaiser, Küster, Nicolay, Fülling, Bens and Raab2019; Moska et al. Reference Moska, Jary, Sokołowski, Poręba, Raczyk, Krawczyk, Skurzyński, Zieliński, Michczyński, Tudyka, Adamiec, Piotrowska, Pawełczyk, Łopuch, Szymak and Ryzner2020, Reference Moska, Sokołowski, Jary, Zieliński, Raczyk, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022; Sokołowski et al. Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022). New 14C dates can help corroborate or reevaluate existing knowledge on the chronology of dune activity and stabilization. The present contribution also aims to shed light on the timeframes of anthropogenic impact on such processes.
The majority of pre-Holocene paleosols that occur in the European inland dune deposits are to date ascribed to relatively warmer pedogenic periods of the latest Pleistocene, Bølling (ca. 14.7–14.1 ka BP), and Allerød (ca. 13.9–12.9 ka BP) Interstadials. These periods were preceded by cooler and drier phases of dune formation (Kappler et al. Reference Kappler, Kaiser, Küster, Nicolay, Fülling, Bens and Raab2019): at the end of the Late Pleniglacial (ca. 20–14.7 ka BP), the Older Dryas (ca. 14 ka BP), and the Younger Dryas (ca. 12.9–11.7 ka BP). Starting with the Early Holocene climate amelioration, remobilization of dunes interrupting the development of soils tends to be controlled by local factors, counting in human activity, rather than by climate (De Keyzer and Bateman Reference De Keyzer and Bateman2018; Kasse and Aalbersberg Reference Kasse and Aalbersberg2019; Sokołowski et al. Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022). The record of Late Holocene (Subatlantic/Anno Domini) dune-forming phases in Central and Eastern Europe has long been overlooked. Only recently has the prevalence and scale of dune reactivation during the Late Holocene so-called “anthropogenic dune formation phase” (Twardy Reference Twardy, Święchowicz and Michno2016) been recognized (Pierik et al. Reference Pierik, van Lanen, Gouw-Bouman, Groenewoudt, Wallinga and Hoek2018; Hsieh et al. Reference Hsieh, Ninard, Uchman, Łaska and Łapcik2023). However, due to the limited scope of previous studies and the complexity of spatial and temporal Late Holocene dune activity patterns, more research is needed.
Here, we compile 23 new accelerator mass spectrometry (AMS) 14C dates of charcoals from paleosols that occur in 13 inland dunes located in Central and Eastern Poland. Sample collection and dating were conducted during the period 2019–2022. Sixteen of the new dates indicate that the paleosols are not older than 1600 cal BP/350 AD, demonstrating the prevalence of human-induced dune remobilization during historical times. The remaining seven samples are dated to the Late Glacial–Early Holocene period and can be treated as reference markers. Several of the new dates (from sites 3, 6, 7, 9, and 12) were already referred to and put into paleogeographic context by Ninard et al. (Reference Ninard, Łapcik, Uchman and Stachacz2022) and Hsieh et al. (Reference Hsieh, Ninard, Uchman, Łaska and Łapcik2023).
SAMPLE COLLECTION AND METHODS
Charcoal samples were collected from 23 distinct paleosol horizons exposed in 13 sand pits located in inland dune landforms (Figure 1; Table 1). The number of paleosols exposed varied from one (sites 4–8 and 12), through two (sites 1 and 2), three (site 11), four (site 13), five (site 10) up to six (sites 3 and 9). In case there is more than one sample dated from a particular site, each was collected from a separate paleosol horizon (Table 1; Figure 2). All investigated paleosols contain abundant charred plant remains ranging in diameter from less than 1 mm up to 4 mm. At least 20 charcoal grains of 2–4 mm in diameter per sample were collected and stored in polyethylene bags.
Figure 1 Location of sampling sites.
Table 1 List of sampling sites with background information.
AMS 14C dating of 23 samples was conducted entirely by Poznań Radiocarbon Laboratory (laboratory code designation Poz). The preparation of samples generally followed the procedure used at the Oxford Radiocarbon Accelerator Unit. The sample material was sequentially treated with 1M HCl (80°C), 0.025–0.2M NaOH (80°C), 0.25M HCl (80°C) and rinsed with deionized water after each reagent. The first HCl treatment lasted until gas bubble emanation of the sample ceased, no less than 20 min, and the second HCl treatment was 1 hr long. NaOH treatment was repeated until no coloration of the solution was observed, each time for no less than 5 min. Combustion of the samples (900°C for 10 hr) was carried out in sealed quartz tubes in the presence of CuO and Ag. The subsequent graphitization process consisted of reduction of the CO2 obtained in a reaction with H2, in the presence of Fe powder as a catalyst. The resulting graphite target was pressed into an aluminum cathode holder.
The 14C content in the samples was measured using a Compact Carbon AMS spectrometer produced by National Electrostatics Corporation. Measurement was based on a comparison between the intensities of 14C, 13C, and 12C ionic beams, measured for each sample and for standard samples (modern standard: Oxalic Acid II and standard of 14C-free carbon—coal). The conventional 14C age calculation was corrected for isotopic fractionation based on δ13C values measured using the AMS spectrometer. Conventional 14C ages were reported with 1σ uncertainty, being the best estimate of total measurement uncertainties. The calibration of 14C ages reported by the laboratory (Table 2) was performed using the OxCal v4.4.4 package (Bronk Ramsey Reference Bronk Ramsey2009), with respect to the IntCal20 calibration curve (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey, Butzin, Cheng, Edwards and Friedrich2020). Calibration results are reported at the 95.4% confidence level as BP and BC/AD age ranges (Table 2; Figure 3).
Table 2 Radiocarbon dating and calibration results.
Figure 3 Calibrated 95.4% range probability distributions of the Late Holocene (cal AD; upper plot) and Late Glacial/Early Holocene (cal BP; lower plot) dates.
RESULTS OF THE 14C DATING
Site 1: Laszczyny
The section is located in the central part of an up to 13 m high parabolic dune landform. Two dated paleosol horizons are inclined accordingly to the cross-stratification in sand. Charcoals from the upper paleosol (Rusty soil; Table 2: 1; Poz-133092) are collected ca. 2.1 m below the ground surface. The lower paleosol (incipient Arenosol; Table 2: 2; Poz-133094; dated also by Sokołowski et al. Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022; GdA-6327) is separated from the upper one by structureless 1.6-m-thick sand.
Site 2: Osiny
The section is located within the S arm of an up to 8 m high parabolic dune landform. The upper paleosol (Podzol; Table 2: 3; Poz-116770) occurs at a depth of 0.5–1 m. The lower paleosol (incipient Arenosol; Table 2: 4; Poz-116718; also dated by Sokołowski et al. Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022; GdA-6317) occurs 2–4 m below in cross-stratified sand, inclined according to stratification.
Site 3: Ochotnik
The section is located in the S part of an up to 5 m high parabolic dune landform. The topmost 0.8 m thick interval contains tetrapod and human footprints. The scattered charcoals in the lower part of this interval gave the youngest date (Table 2: 5; Poz-116758). Below, a Podzol (Table 2: 6; Poz-116769) with abundant cattle hoofprints occurs (Hsieh et al. Reference Hsieh, Ninard, Uchman, Łaska and Łapcik2023). In the lower part of the sedimentary succession, a suite of several laterally splitting Arenosols occurs. Two lowermost horizons, separated by a 0.6 m thick sand interval, were chosen for dating (organic matter-rich Arenosols; Table 2: 7; Poz-116757 and Table 2: 8; Poz-116771, respectively).
Site 4: Zwierzyniec
The section is located within the N arm of an up to 7 m high parabolic dune landform. A paleosol (Podzol; Table 2: 9; Poz-153782) occurs below ca. 0.5–1.2 m thick sand interval that contains abundant cattle hoofprints. A cemented orstein horizon underlies the paleosol.
Site 5: Grębociny
The section is located in the central part of a 10 m high longitudinal dune landform. Similarly to the nearby Zwierzyniec site, a paleosol (Podzol; Table 2: 10; Poz-153801) occurs below an interval of thickness varying from 0.2 m to 1.5 m, which contains abundant cattle hoofprints. A cemented orstein horizon underlies the paleosol.
Site 6: Mnin
The section is located within the N arm of an up to 6 m high parabolic dune landform. Strongly bioturbated, organic matter-rich paleosol (Podzol; Table 2: 11; Poz-112287) is covered by 1–1.2 m sand layer with very abundant cattle and occasional human footprints, documented by Hsieh et al. (Reference Hsieh, Ninard, Uchman, Łaska and Łapcik2023).
Site 7: Oddziały
The section is located within the S arm of an up to 5 m high parabolic dune landform. Similarly to the nearby Mnin site, the paleosol (weakly developed Podzol; Table 2: 12; Poz-122285) is covered by a sand layer up to 1 m thick with cattle and human footprints. A cattle metal chain link was found directly above the paleosol (Hsieh et al. Reference Hsieh, Ninard, Uchman, Łaska and Łapcik2023).
Site 8: Chlewice
The section is located in the central part of an up to 6 m high parabolic dune landform. A paleosol (Podzol; Table 2: 13; Poz-153802) occurs below an interval of thickness varying from 0.4 to 1 m that contains few tetrapod footprints. A cemented orstein horizon underlies the paleosol.
Site 9: Gołąb
The section, documented in detail by Ninard et al. (Reference Ninard, Łapcik, Uchman and Stachacz2022), is located in the N margin of up to 6 m high transverse dune landform. Late Glacial paleosol horizons, dated by Zieliński et al. (Reference Zieliński, Sokołowski, Jankowski, Standzikowski and Fedorowicz2019) and Sokołowski et al. (Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022) were exposed several hundred meters to the NE. The uppermost part of the section displays undisturbed stratification. Below 0.08 m thick recent paleosol (Brunic Arenosol), seven distinct paleosols (uppermost, Podzol, the rest, strongly bioturbated Arenosols) occur within the 3.5 m thick top part of the section. Paleosols are separated from each other by 0.79, 0.12, 0.18, 0.1, 0.12, and 0.25 m massive sand intervals, respectively. Radiocarbon dates are obtained from the first (Podzol; Table 2: 14; Poz-112281), third and sixth paleosols (Arenosols; Table 2: 15; Poz-112282 and Table 2: 16; Poz-112283, respectively).
Site 10: Ostrowik
The section is located in an up to 12 m high complex dune landform. Five paleosol horizons were excavated at the N margin of the dune slope. The upper three (incipient Arenosols; the uppermost one, Table 2: 17; Poz-133097) occur at depths of 0.45 m, 0.55 m, and 0.7 m, respectively. The fourth paleosol (Podzol; Table 2: 18; Poz-133098) occurs at a depth of 1.1 m. The lowermost paleosol, a strongly bioturbated Arenosol, occurs within the spodic horizon of overlying Podzol at a depth of 1.4 m.
Site 11: Przyłęk
The section, located in the central part of the up to 6 m high complex dune landform, displays three paleosol horizons. The uppermost one (weakly developed Podzol, Table 2: 19; Poz-133218) occurs at a depth of 0.7 m. Two lower paleosols (incipient Arenosols; the lower one, Table 2: 20; Poz-133099), occur at depths of 1.5 m and 1.9 m. Both Arenosols display a concave-up geometry, reflecting the morphology of an interdune depression.
Site 12: Unin
The section is located in the central part of up to 14 m high parabolic dune landform. A paleosol (weakly developed Podzol; Table 2: 21; Poz-112286) occurs at depths ranging from 0.8 to 2.5 m. The topmost part of the paleosol and the laminated sand overlying contain abundant tetrapod footprints (Hsieh et al. Reference Hsieh, Ninard, Uchman, Łaska and Łapcik2023). The paleosol is erosionally truncated to the east.
Site 13: Rytele-Olechny
The section is located within the S arm of an up to 6 m high parabolic dune landform and contains four paleosol layers in total. Two uppermost (thin incipient Arenosols) occur at depths of 0.5 m and 0.7 m, respectively. The third paleosol (well-developed Arenosol; Table 2: 22; Poz-133095), occurs at a depth of 0.9 m and the lowest one (weakly developed Podzol; Table 2: 23; Poz-133096) at a depth of 1.1 m.
DISCUSSION
A widespread paleosol, the so-called Usselo-type soil, is found in dunes and coversands throughout Western and Central Europe. It is typically developed as Arenosol with possible signs of podzolization. The Usselo-type soil serves as a marker horizon—in principle for Allerød, with several cases dated to the Younger Dryas and the earliest Holocene (Kaiser et al. Reference Kaiser, Hilgers, Schlaak, Jankowski, Kuehn, Bussemer and Przegiętka2009). Among the sites described in the present paper, paleosols corresponding to pedological properties and the age of the Usselo-type soil occur in sites 2 (Osiny; the lowermost paleosol; sample Poz-116718), 3 (Ochotnik; two lowermost paleosols; Poz-116771 and Poz-116757), 9 (Gołąb; five lowermost paleosols of which two have been dated; Poz-112282 and Poz-112283) and 11 (Przyłęk; two lowermost paleosols of which one has been dated; Poz-133099). The lower paleosol at site 1 (Laszczyny; Poz-133094) visually resembles the Usselo-type soil. However, dating indicates its Bølling age, in contrast to previously known Usselo-type soils, which are younger (Kaiser et al. Reference Kaiser, Hilgers, Schlaak, Jankowski, Kuehn, Bussemer and Przegiętka2009).
Buried Podzols occur at sites 2–10 (samples Poz-116770, Poz-116769, Poz-112287, Poz-122285, Poz-112281, Poz-133098, Poz-133218, Poz-112286, and Poz-133096) and are distinguished by ages ranging from 7th–8th century to 18th–early 20th century. It should be noted that paleosols can, in fact, be younger than dated charcoals, as the redeposition of charcoals from older soils cannot be excluded. In all the sites that display more than one paleosol horizon, only one is a Podzol. An explanation for the better development of Medieval paleosols compared to Late Glacial–Early Holocene and Modern period–recent soils possibly lies in warmer climate or longer, anthropogenically induced periods of surface stabilization during the Middle Ages (Sevink et al. Reference Sevink, van Geel, Jansen and Wallinga2018; Kappler et al. Reference Kappler, Kaiser, Küster, Nicolay, Fülling, Bens and Raab2019; Moska et al. Reference Moska, Jary, Sokołowski, Poręba, Raczyk, Krawczyk, Skurzyński, Zieliński, Michczyński, Tudyka, Adamiec, Piotrowska, Pawełczyk, Łopuch, Szymak and Ryzner2020).
14C dates reported in the present contribution, except for those falling in the Late Glacial–Early Holocene period, are clustered in the 5th–6th to 18th–19th century interval. Notwithstanding the fact that paleosols ranging in age from Boreal through Atlantic to Subboreal were recorded in multiple sites spread throughout Central and Western Europe (Konecka-Betley and Janowska Reference Konecka-Betley and Janowska2005; Tolksdorf and Kaiser Reference Tolksdorf and Kaiser2012; Twardy Reference Twardy, Święchowicz and Michno2016; Lungershausen et al. Reference Lungershausen, Larsen, Bork and Duttmann2018; Pierik et al. Reference Pierik, van Lanen, Gouw-Bouman, Groenewoudt, Wallinga and Hoek2018; Kappler et al. Reference Kappler, Kaiser, Küster, Nicolay, Fülling, Bens and Raab2019), such dates are not present at all among our new results.
Soils develop on the dunes during the conditions of nondeposition and stabilization of the land surface by vegetation, which remained the case throughout most of the Holocene (Kappler et al. Reference Kappler, Kaiser, Küster, Nicolay, Fülling, Bens and Raab2019). Conversely, the burial of soils by dune sands occurs due to surface destabilization caused by vegetation depletion. Human-induced intensification of aeolian activity could also result in soil material being winnowed away and not preserved in the sedimentary record. Assuming purely anthropogenic determinants of dune remobilization that locally interrupted Holocene soil development (Tolksdorf and Kaiser Reference Tolksdorf and Kaiser2012; Lungershausen et al. Reference Lungershausen, Larsen, Bork and Duttmann2018; Pierik et al. Reference Pierik, van Lanen, Gouw-Bouman, Groenewoudt, Wallinga and Hoek2018), lack of such interruptions, recorded as sand layers covering the paleosols, could simply stem from human activity not being intensive enough to force surface destabilization.
Deforestation and agricultural activities have occurred in Europe with intensity varying in both time and space since the Bronze Age. While the Roman Period marked an increase in anthropogenic impact, the Early Middle Ages were characterized by their relative decrease, which, in turn, resulted in partial restoration of primeval forest vegetation (Tolksdorf and Kaiser Reference Tolksdorf and Kaiser2012). Compared to Western Europe, its Central-Eastern part was subjected to limited human habitation and land use throughout the Bronze Age, Iron Age, and Roman Period (Tolksdorf and Kaiser Reference Tolksdorf and Kaiser2012; Twardy Reference Twardy, Święchowicz and Michno2016; Kappler et al. Reference Kappler, Kaiser, Küster, Nicolay, Fülling, Bens and Raab2019; Kasse and Aalbersberg Reference Kasse and Aalbersberg2019). Especially in the barren areas covered by aeolian sands in Central and Eastern Poland, the development of the rural settlement network remained delayed prior to the High Middle Ages. A significant expansion of deforestation and agriculture in these areas began no earlier than during the 10th to 12th centuries, still occurring at a slower rate compared to western and southern Poland (Buko Reference Buko2007). Furthermore, it cannot be excluded that older Holocene paleosols might have been eroded during the Late Holocene intensification of dune mobility, while the pre-Holocene deposits building the lower part of the dunes remained in place. However, such erosion events would expectedly be recorded as erosional truncation surfaces in the intervals underlying the Late Holocene paleosols, which were not noticed in the studied sections.
Apparently, soil development and burial occurred almost synchronously and, at least in some cases, were not restricted to certain localities or dune fields. A comparable age of paleosols is observed between sites located within the same region, but several dozen kilometers apart: site 2 (Osiny, Poz-116770) to site 3 (Ochotnik, Poz-116758)—31 km apart; site 3 (Ochotnik, Poz-116769) to site 8 (Chlewice, Poz-153802)—52 km apart; site 6 (Mnin, Poz-112287) to site 7 (Oddziały, Poz-122285)—27 km apart; site 10 (Ostrowik, Poz-133098) to site 11 (Przyłęk, Poz-133218)—40 km apart; site 9 (Gołąb; Poz-112281) to site 13 (Rytele-Olechny; Poz-133096)—130 km apart. Presumedly, such “dune formation sub-phases” of regional extent occurred in response to spatiotemporal variation in settlement and land use.
The dating of samples from Laszczyny and Osiny sites was independently conducted by Sokołowski et al. (Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022) with results similar to those presented in this paper. Laszczyny La2C (Poz-133094) sample dated to 12,180 ± 60 BP corresponds to 12,200 ± 70 BP (GdA-6327) obtained by Sokołowski et al. (Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022), which confirms the Bølling Interstadial age of the paleosol. Likewise, the Osiny Osi6C (Poz-116718) sample age of 11,310 ± 50 BP virtually equals the result of 11,310 ± 70 BP (GdA-6317) of Sokołowski et al. (Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022), which falls in the middle part of the Allerød Interstadial. On the contrary, new dates from the Gołąb site do not coincide with previous results, from which only Late Glacial paleosols have been reported to date (Zieliński et al. Reference Zieliński, Sokołowski, Jankowski, Standzikowski and Fedorowicz2019; Poz-52907; Sokołowski et al. Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022; GdA-6319, GdA-6321, GdA-6322, GdA-6323). A suite of Holocene (Preboreal and Boreal to Subatlantic/Medieval) paleosols recognized in this study was accessible only temporarily in the marginal part of the dune.
A decrease in the concentration of dates in the Early and Middle Holocene noticed by Sokołowski et al. (Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022) also coincides with the scarcity of radiocarbon ages in the same time interval among those reported in the present contribution. However, an underrepresentation of Late Holocene dates among those reported by Sokołowski et al. (Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022) is hardly explicable. The above applies, in particular, to the younger paleosols at the Laszczyny and Osiny sites, which Sokołowski et al. (Reference Sokołowski, Moska, Zieliński, Jary, Piotrowska, Raczyk, Mroczek, Szymak, Krawczyk, Skurzyński, Poręba, Łopuch and Tudyka2022) did not account for. The notability of the Laszczyny site for studies of anthropogenic impact on dune activity is derived from the presence of human and ungulate footprints in the Early Medieval paleosol (Hsieh et al. Reference Hsieh, Ninard, Uchman, Łaska and Łapcik2023).
CONCLUSION
Most of the radiocarbon dates acquired in Poland from inland dune paleosols fall in the Late Glacial to the earliest Holocene period. Consequently, it could be implied that dune activity was insignificant throughout the remaining part of the Holocene. On the contrary, the present contribution shows that much younger paleosols occur commonly over vast areas. Out of 23 dated buried paleosols, 7 incipient horizons documented at 5 of the investigated sites were dated to the Late Glacial–Early Holocene period. The remaining majority, 16 dates obtained from relatively well-developed buried paleosols, indicate soil formation during the last two millennia. As the remobilization of dunes in Europe during Holocene is assumed to have occurred independently of climatic conditions, burial of the soils by aeolian sand may be ascribed to anthropogenic factors: deforestation and land cultivation in areas adjacent to investigated sites. Therefore, the prevalent occurrence of paleosols buried during the Medieval to Modern periods marks the peak of the anthropogenic dune formation phase in Central and Eastern Poland.
ACKNOWLEDGMENTS
The research is financed by the National Science Centre, Poland, from Daina 1, Grant Agreement No. 2017/27/L/ST10/03370. The publication has been supported by a grant from the Faculty Geography and Geology under the Strategic Programme Excellence Initiative at Jagiellonian University. Two anonymous reviewers are thanked for valuable remarks.
Open access
