A chronology of landsliding based on archaeological and documentary data: Pavlovské vrchy Hills, Western Carpathian Flysch Belt

Pavlovské vrchy Hills represent a distinctive elevation near the Czech-Austrian border where the active, dormant and relict landslides cover 12% of the area. Here we focused on the chronology of landsliding in this area using geological, archaeological and historical evidence. The earliest records of landsliding were determined in locations underlying the dated archaeological settlements. The Upper Paleolithic settlement complex dated between 37–24 ka cal BP, was originally deposited over these landslides. It was consequently destroyed in certain places by additional landslides preceding the last (Upper Pleniglacial) loess deposition (22 ka cal BP). These landslides took place before and after the Upper Paleolithic occupation of this area. This Pleistocene landslide event ranks among the oldest (albeit indirectly) dated landslide within the Czech part of the Western Carpathian Flysch Belt. The chronology of later, historical, landsliding was determined using written records (chronicles, official reports, archival evidence, etc.). Continuous records of landsliding were available as of the middle of the seventeenth century. The major concentration of landslides occurred at the beginning of the twentieth century (1910–1915). The 1663 landslide is currently the oldest landslide, in the Czech part of the Western Carpathian Flysch Belt, which was dated on the basis of documentary data.

Historical landslides. The general lack of suitable materials for landslide dating using geochronological methods, even among the youngest landslides, lead to utilization of other methods in WCFB, e.g. dendrochronological 16 . The existence of written documents (and related references to the current weather therein) has already been utilized by historical climatologists. Research focusing on landsliding which were active during historical times can also utilize this data. Špůrek 23 provided the first comprehensive overview of dated landslide events with this method which covered the area of Central Europe, but also contained records from other parts of the world. The primary sources of his work were newspapers and historical documents. Several episodes of intensification of mass movements were identified in this work in WCFB (e.g. 23 ). The most recent (1939,1941,1970,1997,2006 and 2010) were identified from the village of Halenkovice and its vicinity (approx. 70 km NE from Pavlovské vrchy Hills 7 );. An example of a landslide episode with a large spatial extent (e. g. the entire area of the CR and its vicinity) is that which took place in the spring of 1941 7,[23][24][25] . Other historical documentation of landslide events, over more than the last 200 years, was conducted by Raška et al. 26,27 .
In this paper, which combines geological, archaeological, and historical records, we present evidence from the Pavlovské vrchy Hills for repeated landsliding which affected human settlements. We also synthetized data and built a chronology of landsliding for this part of the Czech WCFB from prehistory to the recent times. This area (located in Moravia, i.e., eastern part of CR; Fig. 1) has been systematically surveyed by archaeologists as of the 1920s. Landslides were mentioned a number of times along with the abundant archaeological findings [28][29][30] . It was also inhabited up until the end of World War II by a German population which carried out detailed reports, often mentioning natural processes, in their chronicles. Such systematic observations, including reports on landsliding, are not available for the rest of the Czech WCFB until the first quarter of the twentieth century.
Geological and geomorphological setting. The topography of this area is strongly affected by the presence of relatively hard limestone which overhangs the neighboring weak flysch rocks forming a rather flat relief (Fig. 2). The Pavlovské vrchy Hills exceed the neighboring landscape up to 350 m 31 . The Hills are mainly built up by the rootless, tectonically transported, sediments of the lowermost unit of the WCFB nappe stack. They consist of numerous imbricates, duplexes, and partial thrust sheets of the Upper Jurassic to Lower Miocene strata. Tectonic klippen of Jurassic marls and carbonates and Upper Cretaceous (Turonian to Campanian) clastic rocks are tectonically incorporated into the thrust sheets of the younger, Paleogene to Lower Miocene sequences (Fig. 3).
The Pavlovské vrchy Hills make up part of the exterior Menilithic-Krosno Group of nappes of the WCFB. The area around Dolní Věstonice, built up of pre-Quaternary calcareous claystones, sandstones, and rare conglomerates, belongs to the Pouzdřany unit and mainly to the Waschberg -Ždánice units 32,33 . Waschberg -Ždánice units are the major frontal thrust system of the WCFB.
The area was the subject of erosion and denudation in the Late Miocene after the WCFB moved to its present position. Limestone bodies form an anticlinal nappe structure protruding from the surrounding, mostly claystone strata. It is a geological structure, typical for the Pavlovské vrchy Hills and disturbed by vertical fault tectonics, which resulted in the sliding of large block along the claystone surface. Pleistocene denudation and accumulation processes on the NW slopes of the Pavlovské vrchy Hills led to a complicated lithostratigraphic sequence, often disturbed by sliding. The block-stony rubble colluvium with the soil-clay matrix created large landslides that slid down the slopes. These landslides resulted in the next disintegration and a mixing of the material on the slopes arose and arises under appropriate climatic conditions, earthflows with devastating effects to both prehistoric settlements and modern villages.
Evidence of this process are the limestone blocks found in the riverbed of the Dyje River 28 . The role of bedrock and its structure in WCFB as a factor of landslide origin was investigated by Rybář 34 who defined the typical case situations. The common pattern in the WCFB is the alteration of inclined rigid and plastic beds. It has been demonstrated that flysch slides commonly originate on slopes determined by downslope bedding.
The Quaternary cover of the Pavlovské vrchy Hills consists of fluvial sandy gravels of the Dyje River in several terraces (fluvial accumulations), colluvial slope sediments and aeolian sediments. During the warm interglacial and shorter interstadial periods of the Pleistocene, the fossil soils and their derivates (soil sediments) were formed in the aeolian sediments. Later, in the Holocene, the subfossil soils were developed in fluvial sandy loams (flood loam). This evidence illustrates the complicated geological evolution over the entire Quaternary in the last 2.588 Ma (www.stratigraphy.com; and 35,36 ).
Prehistoric and historical settlement of the pavlovské vrchy hills. Pleistocene. Although the earliest dated archaeological artifacts in the Pavlovské vrchy Hills belong to the Early Upper Paleolithic (probably Aurignacian), here we focus on the subsequent Middle Upper Paleolithic (Gravettian), the most extensive and the best dated archaeological unit.
Within the key regional section DV 09, found at the bottom of site DV II, (known as "brickyard" or "Calendar of Ages") covering the complex sequence of Upper Pleistocene loess, paleosols, and colluvia, the archaeological layer was dated by C 14 between 30,553-29,442 cal BP 37 . This layer also corresponds to an extensive Upper Paleolithic archaeological complex excavated over the last century at the Dolní Věstonice and Pavlov sites by Karel Absolon, Assien Bohmers, Bohuslav Klíma and Jiří Svoboda (see 28,38,39 ).
These archaeological excavations provided 6 radiocarbon dates from the site of DV I 40 , 37 dates from DV II 41 , and 30 dates from Pavlov I 42 . The majority of these dates were measured at Groningen and Oxford Radiocarbon Laboratories and calibrated after OxCal v.4.2., using IntCal13 calibration curve 43 . These dates are also supported by a series of OSL dates from DV II 41,44 and, most recently, by direct radiocarbon dates obtained from human fossils found at the same site and measured at the Aix-en-Provence Laboratory 45 . All these dates provide a consistent record of human occupation during the MIS3 period. In terms of microchronology, the dates place a scarce Early Upper Paleolithic (probably Aurignacian) occupation at several discrete spots at DV II and Pavlov I dated between 37-33 ka cal BP, followed by the dominant early Gravettian occupation at all sites between 33-29.5 ka cal BP, and by scarce evidence of the late Gravettian between 29.5-24 ka cal BP. At the sites of DV I, II, and Pavlov I, evidence of landsliding is documented both below and above the archaeological deposits. During the Last Glacial Maximum (a period culminating around 22 ka cal BP; see 46,47 ), evidence of human settlements disappeared. The cultural layers and the evidence of landsliding were superposed by archaeologically sterile "Upper Pleniglacial Loess" 48 . The thickness of the Upper Loess layer in the Dolní Věstonice section DV 09 is about 5.8 m 37,49 . According to the optically stimulated luminescence (OSL) method of dating, the "Upper Pleniglacial Loess" sedimentation in DV 09 took place between 27-20 ka cal BP 37,44 . This loess can be used for determining the age of the earlier slide movements that disturbed underlying Gravettian cultural layer.

Data and Methods
Documentary data. We used available historical data such as written records (chronicles, official reports, archival evidence etc.), newspapers, maps and photographs. This data was complemented by secondary sources (early scientific papers, the history of towns and villages, etc.). This kind of data allowed us to localize landsliding in terms of spatial and temporal accuracy. A number of documentary records were hand-written in Neo-Gothic italic script in German (in more recent times also in Czech). The terminology of landslides included a rich variety of terms: "Rutschung" [slide], "Erdrutschung" [landslide], "Erdrutsch" [sliding], "Einsturz" [slump] or "Senkung" [decline]. The information related to landslides was carefully transliterated and excerpted. Where possible, individual records were checked by cross-referencing to other evidence. Individual events, dates of occurrence (if they were available), the type of affected property, and the extent of damage and source of data became the basic information for further critical analyses and interpretations. The primary limitation when dating historical landslides in WCFB is the lack of available resources. Documentary data (chronicles, official reports, archival evidence, etc.) are only seldom available before the nineteenth century.
LiDAR and digital elevation model data. Current geological maps, reports of geophysical investigation and actual LiDAR data (digital elevation model) were used in order to delineate the actual extent of landsliding in the Pavlovské vrchy Hills (Fig. 4). Slope deformations known in the study area are mainly typical landslides, earth or debris flows 56 and vast, complex slope deformations up to many hundreds of metres in length 12,57 . Engineering-geological survey data. An engineering-geological survey related to the construction of Nové Mlýny water reservoirs was completed in 1967. These works 58 provided data on the extent and depth of landslides related to the archaeological sites. Construction of the water reservoir took place between 1981 and 1988. The water in the reservoir has been in direct contact from that time with the frontal part of the slope instabilities (Fig. 3).

Results
Landslides identified on the basis of archaeological data. An extensive complex landslide with the archaeological site DV I 28,29,38,40 was a key locality for the reconstruction of the Upper Pleistocene slope movements. The development of mapping and knowledge about this landslide and individual archaeological excavations are shown in Fig. 5. The existence of several isolated settlements, located within the DV I archaeological site, determines that this area was inhabited and abandoned several times. Klíma 29 suggested that this landslide area was selected to establish a Gravettian settlement as the wavy terrain included water springs, platforms suitable for building shelters, and a depression for bone and waste disposal.
Further local migrations of prehistoric hunters within the Pavlovské vrchy Hills could have been caused by partial activation of slope movements. Knor et al. 28 have drawn valuable profiles through the partial walls of older archaeological excavations (already destroyed today), which unequivocally demonstrated the destruction of cultural layers by sliding movements (see Fig. 6). Moreover, in the same publication, it was stated that loess layers were stacked by sliding processes to 5-6 slices, which apparently caused superposition of 5 to 6 culture layers, one above another. This profile demonstrated that the slope movements that had affected the cultural layer were gradually completed during sedimentation of the Upper Pleniglacial loess. Determination of age of this last loess deposition was crucial for dating of the landsliding which affected and consequently destroyed Gravettian settlements. Excavation at the same site later yielded additional sections documenting a landsliding stage between the deposition of the cultural layer and the uppermost loess 40 .
The above-mentioned Upper Pleistocene landslides in all probability only rejuvenated much older rotationally-translational rupture zones. Flysch sediments of the Ždánice unit (WCFB) were transported in   www.nature.com/scientificreports www.nature.com/scientificreports/ landsliding based on the documentary data in WCFB in CR. There are only 3 older records, identified using the same technique, all located in the western part of CR. They took place during the sixteenth century 23 .
The database on landsliding (Table 1), which took place between 1663 and 1941, at Pavlovské vrchy Hills currently contains 30 events which occurred in 19 years in six villages. Many individual landslides were reactivated several times, but it was not always possible to distinguish reactivation from the emergence of new landsliding and identification of their precise positions was not always possible.
Impacts of landsliding on local society. It was clear, from the available archaeological evidence, that the settlements of mammoth hunters were destroyed several times by landsliding. They therefore had to change their home for a number of times as also documented from archaeological data.
Concerning the historical landslides, they were not usually reported, if they did not cause any economic losses. As many as 30 landslides were identified between 1663 and 1941 in this area (Table 1) At least eight houses, ten farm buildings and seven cellars were completely destroyed over the 1663-1941 period. Another 17 buildings and 12 cellars were spoiled by landsliding. The oldest detailed description can be determined for spring 1774 59,62,63 . Other reports come from 1830 64 . Part of a clay quarry close to Dolní Věstonice was affected by landsliding in 1906 65 (Fig. 8). Exceptionally wet weather in 1910 caused many landslides in January 1911. Two houses were completely destroyed in Pavlov in 1911 and many gardens deformed and moved up to 50 m downslope 63 . Another phase of landsliding took place between 1915 and 1917 when cracks were registered in almost all the buildings in Pavlov village, including the church and 15 houses and 11 cellars were destroyed 62 . The last intensification of landsliding was registered in 1941 when 3 cellars were destroyed, and many vineyards were damaged 59 . The above-mentioned landslide chronology can be compared to those previously published for WCFB 7 . www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
We presented the chronology of landsliding, for an area in the Czech WCFB, which encompasses the last five centuries. We also showed unique evidence of Pleistocene landsliding which took place after the termination of the Upper Paleolithic occupation. Finally, we demonstrated the earliest landsliding at the base of the Pleistocene sections that predate considerably the Upper Paleolithic period. No information on exact dates on landsliding between 22 ka cal BP and 1663 AD was available from this area.  66 . Data were further checked using the following works 67,68 and information in local chronicles. **Destruction of pseudokarst cave roofs on Liščí kopec hill. www.nature.com/scientificreports www.nature.com/scientificreports/ The proven age of the active Pleistocene landsliding ranks among the oldest reactivations within the Czech WCFB. Despite the fact that some of the above-mentioned findings were already reported in older works 28,29 , no modern dating methods were available at that time to precisely date both quaternary sediments and remnants of bones. New approaches allowed for the identification of the above-mentioned findings as contemporary with the Gravettian settlements and the age of the Upper Pleniglacial loess which was deposited on the already destroyed Gravettian settlements. While some old landslides in arid and semi-arid regions of the world are still in morphologically distinctive forms, low resistance rocks to denudation together with humid climate of WCFB resulted that the oldest landslides can only be identified using other than direct morphological approaches, e.g. geophysical methods. The Upper Pleniglacial loess fortunately conserved the Gravettian settlements which could therefore be discovered in the twentieth century.
Pleistocene chronology of landsliding. A chronology of landsliding at the Pavlovské vrchy Hills during the Pleistocene can be reconstructed (see Fig. 9) as follows: 1. Basal large-scale landslides of an unknown age, formed by Jurassic limestone scree and redeposited Tertiary silts and marls. They created a versatile paleo-relief of elevations and side-gorges on the slopes of the Pavlovské vrchy Hills. The basal landslides are partially covered by loess and palaeosols. 2. These elevations and plateaus were consequently inhabited by Upper Palaeolithic (particularly Pre-Gravettian and consequently Gravettian) hunters, documented at several sites between 37-24 ka cal BP. The relevant stratigraphy demonstrated that basal landslides (no. 1 in Fig. 9) clearly predate the Upper Palaeolithic period. Details about these findings can be found at the following sites: DV I and II, Pavlov I 28,29,38,40,42 . 3. One of the cases, where a landslide of Tertiary silts and marl created a water-fan, subsequently filled by water molluscs and mammoth bones, was documented in a gorge below the DV II site. The dating and archaeological context are consistent with the large series of dates from the adjacent settlement. A charcoal sample taken from the location of DV II provided a date of 26,100 ± 200 cal BP 30 which can be used for identification of the age of the landsliding. 4. Local landslides within the Gravettian settlements. These events partially destroyed the Gravettian layers and were subsequently covered by later Upper Pleniglacial loess sedimentation. These landslides cannot be dated precisely because they only disintegrated the already deposited cultural layer. They occurred between the end of the settlement and the formation of the cultural beds and the end of the sedimentation of Upper Pleniglacial loess. Related site DV I was described in 38,42 .
Historical records and landsliding. Despite the importance of documentary data for the study of landslides in the Pavlovské vrchy Hills, several uncertainties have to be considered. Some of the original chronicles and other narrative sources were carried away by their authors or their offspring after 1945. A great deal of documentary data was also damaged (both deliberately and unintentional) over the centuries (wars, natural disaster, inappropriate storing, etc.) and their references are known only via early scientific papers and the histories of towns and villages. The landslides usually did not attract the attention of chroniclers if they did not cause any economic losses. Most of the reports described the occurrence of landslides only briefly and more detailed accounts appeared as of the second half of the eighteenth century. The oldest accounts therefore made the exact localization of landslides impossible in some cases and the same uncertainty was related to the determination of the exact date when the landslide occurred. More recent documentary data did not have to be more accurate, however, than the older one and had to be checked carefully. Finally, except for the different informational value of documentary data, the repeating damages at the same localities and same buildings, which were continuously repaired, made it more difficult to determine the accurate extent of the damage. Despite all these uncertainties, documentary evidence is valuable and an indispensable source of data describing the occurrence as well as the consequences of landslides in the Pavlovské vrchy Hills during the last four centuries. The last stage of the sliding activity in the Pavlovské vrchy Hills was documented in 2014, when there was a great deal of damage to the road infrastructure. The cause of this activity was the high rainfall value in September 2014. Engineering-geological works provided www.nature.com/scientificreports www.nature.com/scientificreports/ data on the actual extent and depth of slope deformations (Fig. 3), but the activity was recorded outside archaeological sites. No landslides were recorded here between 1941 and 2014. The current state of landsliding in the area of Pavlovské vrchy Hills was identified using detailed field mapping and LIDAR data analysis.

conclusions
We studied the chronology of landsliding at the Pavlovské vrchy Hills (Czech part of the WCFB). Leaving aside the evidence of earlier landsliding in the subsoil, the findings about repeated landsliding, presented in this work, can be used as an example of landslide recurrence persisting from the Upper Pleistocene to the present. This work thereby provided new information about landsliding in the Czech WCFB. The main contribution of this work can be summarized as follows: • We analyzed available geochronological data, related to the archaeological excavations, which helped us identify several previously unknown phases of landsliding from the Czech part of the WCFB. • An indirect dating of landslides which destroyed the cultural Gravettian beds was further based on the known age at the base of the Upper Pleniglacial loess which conserved both the landslides and the settlements. • The identification of the 1663 landslide in the village of Pavlov is thus the oldest landslide in Czech WCFB from the historical period, and the fourth oldest landslide in CR, dated using documentary data. • Compilation of the chronology of landsliding (see Table 1) for the Pavlovské vrchy Hills between as the seventeenth century and information about related losses from landsliding.