This work was supported by the French Environment and Energy Management Agency (ADEME) and the Alsace region through the financing of three years of research for the thesis of Lucie Froehlicher. Without their participation, this work could not have been achieved. Field work and soil analyses were funded by the Agence de l’Eau Rhin-Meuse, the leading funder of the GERIHCO 3 Program. Hubert Ortlieb (Société Hubert Ortlieb Géomètre Expert) carried out the topographical survey of the site. Messrs. Munch, Wolf, and Eckert, farmers in Habsheim, gave permission for access to their plots and approval for the digging of pits.

1 - Introduction

1Relative to other French regions, Alsace is not a region where erosion risks are particularly elevated (Le Bissonnais et al., 2002; Van Dijk et al., 2009; Van Dijk & Koller, 2015). The notable exceptions, however, are the three areas (fig. 1A) of Sundgau in the Haut-Rhin, Kochersberg, and to a lesser extent the Outre-Forêt in the Bas-Rhin (Auzet, 1987; Auzet et al., 1995, 2006; Van Dijk & Kwaad, 1996; Van Dijk et al., 2010). The greater sensitivity to erosion in these locations is directly linked to the loess nature of surficial deposits. The depth of loess often exceeds several metres (Théobald, 1958, 1976; Vogt, 1992) to occasionally more than 40 m in thickness (Ménillet, 1980). As a general rule, the physical characteristics of loess deposits – very pure silts having a low structural stability – are responsible for this high sensitivity to erosion (Leplat, 1968).

2As observed in other temperate regions (Evrard et al., 2007), erosional activity is demonstrated by numerous mudslides originating within agricultural fields (Ritsema et al., 1996). This phenomenon is common within the two areas of Kochersberg and Sundgau (fig. 1B). Each year, many municipalities (fr. communes) in these specific regions are affected by mudslides and are subject to declarations of natural disaster emergencies. In 2009, 54 communes were affected by these situations (Auzet et al., 2005).

• Agricultural practices (tillage, types of crops, etc.) generally play an important role (Govers et al., 1994; Lindstrom, 2002; De Alba et al., 2004; Wang et al., 2015). In Sundgau and Kochersberg the dominant crop is maize, a crop that favours the generation of mudslides due to its limited groundcover;

• Topography of the sites is characterized by hilly terrain marked by relatively steep slopes (Vogt, 1992);

• Regional weather conditions (Bryan, 2000; Comino et al., 2015) favour mudslide development as most violent storm events in the Alsace occur in May through June, precisely during the period when crop plants do not adequately cover the soil (Armand, 2009);

• Physical characteristics of the solum are important as soils are developed on loess materials that are rich in silt and, as a consequence, have a very low structural stability (Mussy & Soutter, 1991; Smalley et al., 2011);

• Soil types are another factor as generally Luvisols (leached brown soils) are found on loess plateaus. Luvisols are often partially truncated due to agricultural erosion. In extreme situations, the soil is fully truncated and the loess is exposed directly under the tillage horizon. These calcaric Cambisols are even more sensitive to erosion than Luvisols.

3The consequences of this intensive erosion are both economic and environmental. The resultant mudflows impact agricultural plots (destruction of crops, loss of surficial soil horizons rich in organic matter, creation of physical barriers for seed germination and plant growth in downslope areas, coverage of fields with sediment crusts, etc.), buildings (degradation of infrastructure; basement flooding), and the environment (soil destruction, pollution of waterways by nitrates and pesticides, etc.).

4Numerous studies have been undertaken in order to quantify the loss of agricultural soils due to erosion (Auzet, 1987; Armand et al., 2009). The compiled works highlight the crucial role of living hedges in reducing the erosion of agricultural soils (Ouvry et al., 2012). These vegetative barriers reduce soil loss and result in the accumulation of colluvium upslope of the hedges. These studies, however, focus on the erosional processes linked with major crops. Unlike other regions such as Brittany or Normandy (Carnet, 1979; Carnet et al., 1979; Follain, 2005), the study of erosional history remains relatively rare in Alsace (Schwartz et al., 2009; Schneider et al., 2015). This information is critical, however, as the temporal dimension of large erosive phases is an important element for the understanding of the hillslope dynamics (Carcaud & Arnaud-Fassetta, 2015). This is particularly true for regions such as Sundgau and Kochersberg, where multiple archaeological studies, often a result of preventive excavations, have shown these areas to be characterized by a relatively high density of human occupation since the beginning of the Neolithic period (5,400-5,300 cal. yr BC) in the Rhine Valley). These estimated densities would require a fairly intense agricultural activity, supported by the abundance of related structures such as storage pits (Zehner, 1998; Châtelet, 2009; Bernard, 2014). As such, anthropogenic pressure on these territories has been very strong since the origin of agriculture; not surprising given the well-known fertility of loess terrain (Ertlen et al., 2014).

5The objective of this article is to develop a historical view of erosion in a loess covered area of Alsace. We present the results obtained along a reference toposequence located at Habsheim (Alsace/Haut-Rhin). Our results are based in large part on the description of soil profiles that highlight periods of soil truncation or soil accumulation. The toposequence is marked by embankments or lynchets associated with the hedge vegetation, planted between the cultivated fields in order to limit erosion (Amoreux, 1787; Liagre, 2006; Ouvry et al., 2012). These hedges have collected eroded colluvium for many millennia, allowing for a long-term view of erosion in the region. Our chronological framework is supported by OSL dating of this toposequence.

2 - Materials and methods

2.1 - Lynchets and their potential as archives of erosion history

6The term lynchet (fr. rideau de culture) designates ledges that run more or less parallel with the contour lines of hillsides (Aufrère, 1929) and resemble agricultural terraces (fig. 2). However, unlike terraces, they do not arise due to conscious human planning.

7They are formed by an upslope accumulation of colluvium mobilised by the erosion of agricultural soils on a portion of the slope. The result is a progressive “backwater effect” due to the hedge causing eroded sediments to accumulate on the upslope side of the hedge. All the while, immediately downslope from the hedge, ablation is occurring at the base of the foliage of the hedge causing an overdeepening. These topographic strips are therefore the unintended consequence of agricultural practices in these areas. The landscape induced by these agrarian structures are, according to Allix (1931), "anthropogeographic landscapes" (fr. paysages anthropogéographiques).

8Lynchet height and the rate of sediment accumulation behind the hedge are proportional to the quantity of colluvium produced, an amount dependent on many natural and human factors: slope orientation (north- or south-facing); climatic factors such as the intensity of precipitation and its seasonal distribution (Bryan, 2000; Comino et al., 2015); the proportion of silt in the soil (Musy & Soutter, 1991; Duchaufour, 1997); the sensitivity of the soil to erosion (Bryan, 2000; Van Dijk et al., 2010); the percentage of slope of the terrain upslope of the hedge and the distance between two hedges (Follain, 2005); the type of agricultural practices, in particular tillage (T) and its orientation (perpendicular or parallel to the slope) and depth, or the use of no-tillage techniques (NT) (Govers et al., 1994; Lindstrom, 2002; De Alba et al., 2004; Wang et al., 2015); the type of crop (cereals, pulses, etc.); and, the type of hedge downslope of the field (with or without ledges, ditches, weedy strips, etc.) (Baudry et al., 2003).

9Lynchet height can be impressive (fig. 3). In the Sundgau region, some lynchets may exceed 2 m. Height is proportional to the quantity of colluvium trapped upslope of the hedge structure. As such, the age of a hedge is an important factor given that the total accumulation of colluvium is proportional to the length of time of accumulation. Therefore, colluvium stored behind the lynchets represents a potential archive of erosional history at a plot-scale. When coupled with OSL dating, the morphological and analytical characteristics of lynchets as well as discontinuities within the deposits should shed light on a site’s erosional history.

2.2 - Site selection

10Strict criteria and a rigorous protocol were established for selecting a study site. First, the study site was to be located in one of the two large loess regions (Kochersberg and Sundgau) where the highly erodible nature of the loess soils has already been well-studied (Auzet, 1987; Armand, 2009; Van Dijk et al., 2010; Van Dijk & Koller, 2015). Second, the site must also incorporate the complete topographic configuration to permit a full understanding of slope processes. The toposequence must therefore be complete, from the ablation zone at the hilltop to the accumulation zone downslope. The chosen site was required to have one or more lynchets perpendicular to the slope and having living hedges. Finally, the site must also have been in quasi-continual agricultural use for the longest possible time. Historical information was obtained through documents including territorial maps (fr. plans de finage) showing landownership in Alsace (1760-1762 AD), a Napoleonic land register for ca. 1825 AD, German ordnance/topographical maps from ca. 1880 AD, and aerial photos from the IGN (since 1950 AD).

11The initial selection phase was made at the regional scale (fig. 4A) based on an airborne LIDAR (Light Detection and Ranging) analysis of microtopographies. The LIDAR coverage, however, was unevenly distributed; the Haut-Rhin was almost fully covered, whereas the Bas-Rhin had coverage for only a limited region. The objective of this first phase of exploration was to detect all areas characterized by linear ridges running perpendicular to the main slope and perfectly visible on the LIDAR (fig. 5). However, LIDAR data in of itself does not distinguish lynchets on the imagery from similar features. Ridges of accumulated, ploughed material (ackerbergs), rounded fields, and terraces may all be characterized by raised linear forms and are sometimes difficult to distinguish from lynchets using LIDAR. As such, LIDAR analysis performed on the Alsace allows a systematic detection of all of the previously described landforms. Ground-truthing in the field (in phase 3) permits an evaluation of the effectiveness of LIDAR for finding and identifying lynchets (and similar landforms).

12In the second phase, potential study sites identified during the first phase were matched with geological, pedological, topographical, geohistorical/archaeological data (fig. 4B), reducing a preselection to 30 sites. The final phase (fig. 4C) involved field verification of the 30 sites in order to confirm or reject the selections – in particular eliminating agricultural terraces – and to ensure that all necessary criteria were respected. Of the few remaining candidates, a hillside near Habsheim was selected as the most promising study site.

2.3 - Site description

13The village of Habsheim is located in Sundgau. The toposequence of interest is located to the west of the village, on a small hill covered with loess deposits which can exceed 5 m in thickness (Théobald, 1976) (fig. 6). The toposequence (centred at 47°43’39.20"N, 7°24’36.42"E; 280 m a.s.l.) is approximately 140 m long and oriented North-South. The toposequence passes through three agricultural plots and has two lynchets associated with hedges (fig. 7).

14Geohistorical documents confirm that all three plots crossed by the transect have been in use as cropland since 1760 AD (ADHR C1164/7). The plots in question do not seem to have been used as vineyards during this period. The precision is important as a number of wine parcels of land dedicated to wine production remain in the immediate vicinity; remnants of a relatively large-scale attempt to establish vineyards in the Sundgau region during the eighteenth and nineteenth centuries. The presence of vineyards on these plots would suggest that the landforms were more likely to be terraces rather than lynchets.

2.4 - Field and laboratory analyses

15Following use of a soil auger to survey the entire hillside, nine pits were dug along the transect, three pits per section (fig. 7). The base of a soil pit was systematically set at 20 cm below the top of the loess. Pit depth, therefore, varied between 0.5 and 2 m. For each pit, detailed soil descriptions were made, carefully noting indices of reworking (presence of archaeological artifacts, charcoal, pebbles from the Rhine River brought in with manure, discontinuities, etc.), so as to identify phases of ablation or accumulation.

16Optically simulated luminescence (OSL) techniques determine "the last exposure to light of quartz and feldspar mineral grains and thus the time a sediment was laid down" (Fuchs & Lang, 2009). Samples for OSL dating were collected using an opaque PVC tube at the base of each colluvial horizon in order to determine the age of the colluvium deposit (Preusser et al., 2008). These tubes, having a length of approximately 30 cm, were fully inserted horizontally into the sediment. After extraction of the tubes, the ends were sawed off and sealed so as to ensure the sediment was not disturbed during transport. When the colluvium deposit was thick, additional complementary samples were obtained within the deposit.

17OSL determination was conducted in the Luminescence Dating Laboratory of the Physics Institute, Silesian University of Technology at Gliwice (Poland). Six OSL dates on quartz (via multigrain analysis) were obtained along the toposequence. The multigrain approach is effective even for deposits that have only been transported over a short distance (Lang & Hönscheidt, 1999) as is the case for our study site. However, it should be noted that a portion of the grains used in sample may have been brought to the surface after deposition and burial through bioturbation and/or soil mechanics (Krull et al., 2006). Their final burial was therefore not necessarily contemporaneous with the colluvium deposition particularly if deposition of the colluvium was diffuse. As such, in this type of material, it is wise to consider the obtained OSL ages as being minimal rather than absolute ages.

18Samples were oven dried at 40°C for 7 days and sieved to pass a 2 mm sieve for further analysis. The particle size distribution was measured by laser granulometry (Beckmann-Coulter LS230). Before the particle size measurement, the samples were treated with H₂O₂ to destroy soil organic matter aggregated in the particles. The samples were also washed by KCl, distilled water and sodium hexametaphosphate to deflocculate the aggregates without destructing the carbonates.

3 - Results

3.1 - Soil descriptions

19The studied toposequence is divided into three sections, each corresponding to an agricultural plot (fig. 7 & 8). Soil descriptions are detailed following this organization of sections.

3.1.1 - Upslope section (soil profiles: HABS 01, HABS 02, HABS 03; Section 1)

20The average slope for this section (between HABS 01 and HABS 03) is 2.8 % although 4.3 % between HABS 02 and HABS 03. In the first two soil profiles (HABS 01 and HABS 02) located in the uppermost portion of the toposequence on the flat summit of the hill, the tilled horizon (L = Ap) is 39-cm-thick and lies directly on loess. The profile is typical of soils having the upper portion of the soil profile truncated by erosion. Evidence for the initial existence of a Luvisol on the slope is provided by the presence of a Bt horizon in profile HABS 06 (cf. infra). The sharpness of the planar contact and the absence of a transition horizon suggests that erosion is still active. The profile HABS 03 is located at approximately 20 m downslope from HABS 02 and immediately adjacent to the lynchet that limits this section. Under a 38-cm-thick tilling horizon, there is an accumulation of pedogenetised carbonate colluvium (i.e. SC horizon) having a thickness of 57 cm. At 95 cm depth, the loess (IIC) appears after a transition of only a few centimetres.

3.1.2 - Middle section (soil profiles HABS 05, HABS 06, HABS 10; Section 2)

21The average slope of this section, not including the lynchet face, is 9.09 %. The pit upslope of the hedge, HABS 05, is very similar to HABS 01 and HABS 02. Under a 28-cm-thick tilled horizon, the profile shows a transition within a few centimetres to the C horizon of loess. This again represents a soil profile completely truncated by agricultural erosion. HABS 10, located approximately 6 m downslope of HABS 05, is very similar to HABS 03 (sequence of a tillage horizon (L), carbonate colluvium (SC), and a loess horizon (IIC)). HABS 06 is located 12.5 m downslope of HABS 05, just above the second hedge. Having a depth of more than 1.75 m above the top of the loess, this profile presents three well-defined horizons: L (45 cm), SC (55 cm), and Bt (75 cm). The SC horizon is rich in carbonate derived from the colluvium. The Bt horizon, non-carbonate and orange colored, is divided into two sub-horizons. The upper portion, between 90 and 140 cm depth, is comprised of colluvium material. The base is marked by a discontinuity, highlighted by the presence of allochthonous pebbles between 120 and 140 cm depth (alluvial pebbles from the Rhine), that were likely deposited here at the same time as the imported manure. Indeed, when the farmer established in the Rhine Valley scraped the manure from the barn, pebbles from the geological substrate were involuntarily mixed with manure (J.-J. Wolf, personal communication). The lower portion, located between 140 and 175 cm depth, corresponds to an in situ Bt horizon. The difference in thickness of the horizon L between the upslope profile (HABS 05 at 28 cm) and the downslope profile (HABS 06 at 45 cm) suggests that active erosion is minimal (since 2011, the parcel has been covered by grass).

3.1.3 - Downslope section (soil profiles, HABS 07, HABS 08, HABS 09; Section 3)

22The average slope of the section is 10.11 %. The upslope pit, HABS 07, is characterized by a till horizon of 32 cm in thickness that is in direct contact with a Bt horizon of 80 cm in thickness. For its part, the profile HABS 08 is more complete since it presents a 25-cm-thick E horizon below the L  horizon (35 cm in thickness), and then a 60 cm-thick Bt horizon. The soil profile of HABS 09 is very different. This very deep profile has five horizons: L (32 cm), S1 (43 cm), S2 (35 cm), SC (24 cm), and the IIC horizon of loess below 1.34 m depth. It is developed entirely within carbonate colluvium. The Bt horizon that is observed in the three pits located immediately upslope (HABS 06 to 08) has completely disappeared in HABS 09.

3.2 - Particle size analyses

23Granulometry of the soil samples shows that clay percentages are relatively stable along the entire toposequence having values around 15 % except in the Bt horizons (about 20 %). No general trend can be given for sands or silts, it show stable values within a given profile. At the scale of the toposequence, no correlations are observed between the granulometry and the position of a profile along the toposequence. A certain variability does exist, as silt ranges from 60 to 80 % between profiles and sands range from 8 to 25 %. Note that although morphological, textural, and chemical differentiation (no reaction to the HCl test) of the Bt horizons is quite distinct in the field, this difference is not observed for particle size analyses (fig. 9).

3.3 - Osl dating

24OSL ages (tab. 1 & fig. 8) fall between a maximum of 5,760 ± 500 yr (HABS 09 at 127 cm depth, immediately above the loess) and a minimum of 996 ± 83 yr (HABS 08 at 60 cm depth).

4 - Interpretation and discussion

25Morphological analysis of the soils along the toposequence confirms that these features are lynchets rather than agricultural terraces. The perfect continuity of the loess surface, regardless of the relative position of a given profile and the degree of soil horizon development reflects lynchets. The building of terraces, through the excavation of soils and embankment, would normally produce clear breaks in the surface topography of the loess, and a mixture of soil and sediment introduced from other sites.

26The toposequence showed strong variability within soil profiles, direct related to erosion: soils that were fully or partially truncated were observed in the upper portion of the three sections while soils were thickened by accumulated colluvium in the downslope portions of the sections. In the lowermost portion of the toposequence, the initial Luvisol has been partially preserved (see profiles HABS 06 to HABS 08), although not everywhere as was evidenced by its absence in the downslope profile HABS 09. It is clear that erosion has played an important role along the entire transect and continues through to the present. A portion of the erosive phase preceded the erection of hedges, evidenced by analysis of the upslope section. We do not find any material that was likely to have belonged to an eroded Bt horizon in the colluvium stored upslope of the upper lynchet, contrary to that observed in Section 2; the Luvisol that had occupied the top of the slope has been completely truncated prior to the erection of the first hedge. The organic and carbonate colluvium stored upslope of this hedge was derived solely from erosion of the now-truncated soil surface horizon. Subsequent to the establishment of hedges, each section operated independently. It is therefore appropriate to distinguish between an initial phase of erosion that affected the entire catchment area prior to the planting of hedges from subsequent phases following their establishment.

27OSL dates can offer a more specific picture of events. The first phase of erosion began around 5,800 yr, i.e., at the beginning of the Upper Neolithic (Lefranc, 2009). Although it is difficult to constrain the beginning of this erosive phase, the end of this period is more certain. The base of the colluvium stored upslope of the first hedge is dated to 2,700 BP (the beginning of the Iron Age) and it is clear that ablation of the Luvisol was already complete in the upslope portion of the transect prior to this date. If one considers that colluvium had begun to accumulate behind the first hedge after its planting, the date of about 2,700 yr constrains the age of the upper hedge and would have been in place since the beginning of the Early Iron Age (Hallstatt). To our knowledge, this is the first time that a hedge and the related lynchet have been directly dated, at least in the Alsace region. In Brittany, Gebhardt et al. (2007) estimated that the hedges on slopes had been present since at least the later portion of the Iron Age, but these ages were deduced from palynological and anthracological analyses. Although our date that we obtained is relatively old, it is consistent with regional archaeological discoveries. The Habsheim region holds many clues regarding Hallstatt populations during that time period (Wolf, 1973a,b, 1979a,b, 1985, 1996a,b; Zehner, 1998).

28OSL dates suggest that the other hedges and lynchets are medieval in age. However, caution must be used. Given that the most recent dates were obtained at relatively shallow depths (60 cm), it remains possible that a portion of the soil had been returned periodically to the surface through bioturbation and as such reset the OSL age (i.e. make it younger). As such, these ages should be considered as representing minimum ages for hedges 1 and 2. In addition, the burial of colluvium has not necessarily been rapid. If the development of the lynchet has been progressive and not abrupt, the colluvium would have been buried gradually, another potential cause of young OSL ages. These points would need to be clarified in the future, performing dating at a grain by grain scale (Krull et al., 2006). The complete package of field and laboratory observations as well as OSL dating allow us to propose a scenario of the succession of different erosional phases (fig. 10).

4.1 - An initial phase of erosion and deposition affecting the entire hillslope during the prehistoric period

29The initial phase is structured by a series of dates on colluvium samples located immediately above the loess (C): HABS 03 (85-88 cm), HABS 06 (130-133 cm), and HABS 09 (127-130 cm). This phase began at 5,760 ± 500 yr or earlier manifested by the complete truncation (uppermost portion of the transect and downslope) or partial truncation (elsewhere) of the Luvisol. In the middle portion of the transect, the Bt horizon of a Luvisol has been partially retained. A portion of the Bt horizon, eroded upslope, was deposited onto these partially truncated Luvisols. This phase was completed prior to 2,700 BP. No relics of the Bt horizon were retained within the colluvium of the upper hedge.

30In terms of an archaeological chronology, these ages lie between 3,810 BC and 760 BC, or from the Early Neolithic until the end of the Bronze Age or the beginning of the Early Iron Age (Lefranc, 2009). This first major phase of colluviation is very likely linked to the cultivation of the territory at the beginning of the Neolithic (Mäckel et al., 2003). The large-scale land clearing at the time would have led to significant erosion of loess material on slopes. The beginning of this phase is consistent with the patterns of land use in the Neolithic. Although the type of land use during the Early Neolithic (Rubané) is still very much under debate (Bogaard, 2004, 2005; Ebersbach & Schade, 2004; Bogaard et al., 2005; Kreuz et al., 2005; Kreuz, 2008; Van der Zanden, 2013), it seems that livestock was common (perhaps within a forest setting) and that agriculture, practised without iron plows and with very rudimentary means, caused little erosion. From the Middle Neolithic, when land clearing was more systematic and use of plows was more common, erosion would have increased.

4.2 - Planting of a hedge about 2,700 yr ago and the subsequent capture of colluvium

31Towards the end of the Bronze Age or the beginning of the Hallstatt (a minimum age around 2,700 BP), the installation of a hedge changed the erosive regime. Erosional processes were confined to the sections or plots between hedges. At an initial glance, two areas can be distinguished: an upslope area, corresponding to section 1 and a downslope area (corresponding to sections 2 and 3) (fig. 10B & 10C). The burial depth of colluvium dated at 2 710 ± 200 yr (at 88 cm depth) suggests that there is currently little bioturbation. Due to the possible processes discussed above, it is difficult to confirm whether the planting of the hedge occurred at the end of the Bronze Age (due to a younger age) or at the Early Iron Age (an older age). Be that as it may, this period represents a change in the occupation of the land, with the establishment of the first parcelling of land marked by defined and permanent limits; this pattern is maintained through to the present. It is clear that the creation of lynchets has contributed to defining the spatial limitations of the agricultural plots in this region.

4.3 - Implantation of the second hedge and earthen constructions since about 1 400 yr ago

32The OSL dates within HABS 06 allow us to further develop the erosional history of the toposequence (fig. 10D). If, as has been seen, the date obtained at the base of the Bt colluvium characterized the first phase, the date obtained at the base of the E horizon (perhaps rather an S horizon developed in the carbonate colluvium) seems to indicate that a second hedge planting at ca. 1,370 ± 100 yr, which is approximately 545-745 AD, therefore the Early Middle Ages. The other two ages obtained from HABS 08 (996 ± 83 yr or 936-1102 AD) and HABS 09 (1,060 ± 160 yr or 795-1115 AD) also fall into the Early Middle Ages. It is difficult to say whether the age differences between the first date and the two younger OSL dates comes from a spread of this erosional phase over several centuries or because of bioturbation of the soil. The selected samples were located at a lesser depth (60 cm) than the first, older date (80 cm) and a greater proportion of grains is therefore likely to have been mixed by the meso- and macrofauna of the soil. Regardless, these three dates are compatible with an extension or a resumption of agricultural activities related to the establishment of the village of Habsheim in 758 AD under the name of Habuhinasheim (Stoffel, 1868).

33The results from the analysis of this toposequence lead to a number of important finding. First, the total ablation of the Luvisols in the upper part of the toposequence occurred prior to 2,710 ± 200 yr. The partial truncation of the Luvisols in the middle section and their subsequent burial by colluvium in the Middle Neolithic shows that erosion was intense at the beginning of agriculture in the region. Events of this type are identified early in the Rhone basin and cause removal of Luvisols located downslope and in marl depressions (Berger 1996, 2003, 2015; Notebaert & Berger, 2014) and in many areas from Germany (Houben, 2012). In Alsace, the intensity of erosion in loess covered regions has often demonstrated by the taphonomy of archaeological finds; many structures are partially truncated (Zehner, 1998; Ertlen et al., 2014). However, with truncation being necessarily subsequent to the buried human structures, these dates are relative, in the form of a terminus post quem. A few studies are exceptions to this pattern. At Gottesheim, north of Kochersberg (fig. 1A), Schwartz et al. (2009) provide evidence of tillage within loess evidenced by ordered loess carbonated concretions. The covering colluvial material at the Gottesheim site has been transformed into a Luvisol, which attests to its age. A minimum radiocarbon age of 5,700 BP (Schwartz et al., unpublished) has been obtained. The micromorphological analysis of the colluvial material above the tillage zone shows the deposits to contain, in addition to the clay coatings related to the development of the present-day Luvisol, disorganized fragments of coatings, vestiges of the eroded portion of Luvisols (Schwartz et al., unpublished). The results obtained at Habsheim and Gottesheim are thus two facets of the same reality. Furthermore, the work of the LGV Est Européenne in the sector of Kochersberg (fig. 1A) also detailed an early erosive phase dated to the Early Neolithic (Ertlen et al., 2012, 2014). The Bronze Age, which has been the subject of many archaeological investigations in Alsace, is there marked by slope stability (Ertlen et al., 2014; Schneider et al., 2015), relative to the Sundgau area. In the Early Iron Age, a high demographic increase, the massive colonization of the Kochersberg (Bernard, 2014), and the degrading climate in Western Europe (Van Geel et al., 1996) all led to significant changes in erosion processes at the catchments scale in Alsace and in Europe more generally (Lespez et al., 2008). The major changes in agrarian practices described by Lespez et al. (2012) for Normandy, seem to translate at our site with the establishment of the first system of hedges. The resumption of agriculturally-related erosion processes in the Middle Ages after a period of relative stability appears to be synchronous between the regions of Kochersberg (Ertlen et al. 2014) and Sundgau. In addition, the complete set of OSL ages obtained along our toposequence coincides with archaeological periods that are well-represented in the vicinity of Habsheim (Wolf, 1973a,b, 1985, 1996a,b), a region rich in Neolithic and Late Iron Age (Hallstatt) sites.

5 - Conclusion

34Finally, our study shows that many of the existing agricultural plot limits in the Alsace Plain are indeed very old as they date from the end of the Bronze Age or the beginning of the Iron Age. This is the first time that it is possible to assert ages this old (or at least this old) in the Alsace. These ages may seem surprising for the region where most of the agrarian limits are often deemed to be more recent. As well, the piecemeal nature of plots on the Alsace Plain-layed out in the form of hardwood parquet flooring or piano keys are posterior to the Thirty Years War (Boehler, 1995). Lynchets, by their physical characteristics namely the height of the embankments and their significant slope, contribute to the long-term fixing plot limits and their removal is very difficult. In some regions, such as the Kochersberg, many of these embankments have been eliminated since the reparcelling of land in year 1955, a prelude to the recent intensification of agriculture. In other sub-regions such as the Sundgau, being more conservative and marked by an agriculture where farmers hold second occupations, these landscape features have been preserved. As such, agrarian history is much easier to decipher in these preserved agricultural landscapes.