The authors express their gratitude to the Centre National de la Recherche Scientifique (Zone Atelier CNRS "Bassin de la Loire") for its financial participation. We also wish to thank Arnaud Tourman for his help during fieldwork.

1 - Introduction

1Studies on the origin of bogs and their expansion during the Holocene have revealed the close relationship between the paludification dynamics and climatic changes as well as atmospheric CO₂ and CH₄ levels (Blunier et al., 1995; Klinger et al., 1996; Campbell et al., 2000; Smith et al., 2004; MacDonald et al., 2006; Korhola et al., 2010). However, these studies were mainly performed within boreal and arctic bioclimatic areas on the Northern hemisphere. The development of bogs in mid-latitudes was much less studied, and the according data rely mainly from England, Ireland, The Netherlands, Spain and France. In the north of England a huge number of bogs was built up between 8,000 and 5,000 cal. yr BP (Hughes et al., 2000; Barber et al., 2003). Among the 57 radiocarbon dates obtained for bog initialization by Tallis (1991) in England, 15 dated between 10,000 and 6,800 cal. yr BP, whereas others are spread between 6,800 and 1,400 cal. yr BP. Concerning the palynological results from 54 bogs in Northern Ireland, Reffay (1972) noted that half of these sites started peat accumulation during the Boreal and Atlantic chronozones. Finally, Charman (2002) stated that initialization of blanket bogs dated to 8,300-5,800 cal. yr BP (i.e. ca. 7,500-5,000 BP). In The Netherlands studies on Phragmites-reed and Alder peatlands in relation to Holocene phases of marine transgressions in coastal lowlands showed that these peatlands were installed since the Boreal chronozone (Pons, 1992). Hydro-geomorphological features drove the subsequent peatland expansion, so that these coastal peatlands reached their maximal expansion around 3,300 cal. yr BP. In Spain, a series of 30 radiocarbon dates from Galicia showed that the peat development started in the early phase of the Holocene around 11,000-10,000 cal. yr BP, and was followed by an intensive phase of ombrotrophication between 6,000 and 5,000 cal. yr BP, whereas a huge number of mires originated from the two periods 6,000-5,000 and 3,000-2,000 cal. yr BP (Pontevedra-Pombal et al., 2006). In France, on the other hand, studies on 83 well-dated peatlands in the eastern part of the Massif Central mountain range showed that the peat initiation history is divided into two periods: the first covers the interval 10,200-5,300 cal. yr BP, and the second covers the last 4,500 cal. yr BP (Cubizolle et al., 2012).

2In a general view, the reasons for the initiation and subsequent expansion of peatlands are not only climatically and geomorphologically driven, but are also related to the prehistorical and historical anthropogenic and morphogenetic activities (Cubizolle et al., 2004). This debate on the impact of human societies on the development of bogs started in the 1960^th in Great Britain when numerous authors proposed - based on archaeological results - Mesolithic societies to have possibly been a decisive factor in changing groundwater levels (Rankine et al., 1960; Keef et al., 1965; Chambers, 1988). Later, the combined role of fire and pasture did heavily affect vegetation cover and soil quality, so to change the hydrological budgets of lowlands (Simmons & Innes, 1985; Moore, 1988, 1993; Caseldine & Hatton, 1993; Robichaud, 2000). As this debate was not followed up, it took until the end of the 1990^th for recent studies on peatlands of the French Massif Central to bring to light that the first generation of bogs (10,200-5,300 cal. yr BP) was well determined by climatic reasons, whereas the second period of peat initialization (since 4,500 cal. yr BP) was linked to human action on their environments and their hydro-geomorphological consequences (Cubizolle et al., 2003, 2004, 2012). Nevertheless, some reasons for the appearance of a certain number of peatlands during the first period are not completely understood, especially those appearing between 6,200 and 5,300 cal. yr BP. This period is in fact marked by the progressive installation of first Neolithic agro-communities within all of the Massif Central (de Beaulieu & Goeury, 1987; de Beaulieu et al., 1988; Reille, 1989; Miras, 2004). In our research area in the northeast of the Massif Central, numerous studies exist from the lowlands (Cubizolle, 2005; Georges, 2007), as well as from the mountains above 800 m a.s.l. So, pollen from pastoral plants such as Plantago lanceolata were found at ca. 6,300 cal. yr BP at the site of Braveix (1,370 m) in the Monts du Forez (Francez, 1990), and Reille (1989) noted cereal type pollen and elevated grass pollen amounts (Poaceae) around 5,800 cal. yr BP at ‘La Pigne’ at 1,335 m a.s.l. (also in the Monts du Forez).

3The present work therefore aims at clarifications on the role of past anthropogenic activities on the initialization of peat growth around 7,000-5,500 cal. yr BP. In addition, the question on how such an impact may have affected the lateral growth of peatlands within bottom valleys will be discussed based on the study of the radiocarbon dated, minerotrophic mire of La Prenarde-Pifoy within a granitic geological setting in the Eastern part of the Massif Central and the interdisciplinary study of geomorphology, micromorphology, pollen, macrofossils, and diatoms.

2 - Environmental context

2.1 - Study area

4The La Prenarde-Pifoy mire (N 45°30’11’’ / E 3°58’51’’) is located in a small head catchment within the upper catchment of the River Loire. The mire is situated in the southeast of the Monts du Forez mountain range, which is part of the Hercynian basement of the French Massif Central (Etlicher, 2005). The highest uplands orientated NNW / ESE, the Pierre-sur-Haute Massif, culminates at the altitude of 1,634 m a.s.l. (fig. 1). The low altitudes, 280 - 650 m a.s.l., are located at the bottom of tectonic basins like the Forez basin and Limagne basin, which are related to the Oligocene rifting. In a geological point of view, the Monts du Forez are included in the north part of the anatectic and granodirites incorporated in a metamorphic complex basement of various rocks.

5The regional is of oceanic type and varies depending on the altitude: at 500 m a.s.l., the mean annual temperature is around 10.5°C, whereas it drops to around 3°C at 1,630 m a.s.l.; mean annual precipitation ranges from 700 mm at 500 m a.s.l. to 1,500-1,600 mm at 1,630 m a.s.l.; a nival mantle is present in winter, whereas important precipitation exists in summer. At 1,125 m a.s.l. the site of La Prenarde-Pifoy registers mean annual precipitation of 1,040 mm (for the period 2002-2011), and a mean annual temperature of 6.3°C.

2.2 - The la Prenarde-Pifoy mire and its catchment

6The La Prenarde-Pifoy mire is situated in the centre of an acidophilic fir forest (Abies alba) with blueberry in the understory (Vaccinium myrtillus). The site lies therefore directly at the altitudinal limit of the lower mountains in use historically for cultivation and of the higher mountains with their forests. The mire covers about 2.5 ha of a granitic depression where running waters from three different valleys merge (fig. 2). Nevertheless, the zone where peat develops with a thickness of more than 50 cm is only about 1.6 ha wide. The longitudinal slope is less than 3 %, and the catchment area – covered by 80 % with forest - is about 0.85 km² with altitudes varying between 1,121 m and 1,228 m (fig. 2). From a geological point of view the catchment area lies within a monzogranite complex. An escarpment along a crack increases drastically the slope just below the mire (fig. 2).

7The toposequences show the following succession (fig. 3):

• the gruss is observed only at the top of the hill. No periglacial deposits or soil covers the gruss except a lithosol very locally;

• in the upper part of the slope, a leptosol is present with a cambic podzol covering the head, this latter being a typical periglacial slope deposit of middle latitudes;

• in the middle part of the slope, the toposequence continues with a dystric cambisol, which changes to an anthroposol covered by bracken (Pteridium aquilinum);

• the beginning of the humid mire zone is characterized by a gleysol that is rapidly replaced by a histosol.

8The La Prenarde-Pifoy mire and peat bog can be characterized as a minerotrophic mire with woods growing on it since the 1940s. Birch (Betula pubescens) and pine (Pinus sylvestris) dominate and are associated with peat mosses (Sphagnum palustre) and sedges (Carex rostrata). The central part of the mire only is overgrown by woods on 4,000 m². However, Betula pubescens invades successively the Juncus acutiflorus prairie with Sphagnum teres. The pH-values vary between 5 within the Sphagnum cover and 7 within the Juncus dominated plant society. The peat thickness reaches a maximum of 1.70 m, with about 22 % of the mire showing peat thicknesses of more than 1.5 m. In the south-western part of the mire the peat thickness does not reach more than 1.20 m, probably because of the peat compaction due to a draining initiative for peat exploitation in the 1970s. The stratigraphy therefore shows a decomposed peat facies within the upper 40 cm.

9The peatland of La Prenarde-Pifoy was established during paludification processes as shown by a succession of sapric, mesic and fibrous peat starting from the base, and through the absence of gyttja layers or other lacustrine facies (Payette & Rochefort, 2001; Porteret, 2008). The La Prenarde-Pifoy is therefore originally a minerogenic peatland supplied by running water from the slopes.

3 - Methods

3.1 - Peat coring

10The description of the stratigraphies before extraction of sediment cores for palaeoecological analyses was performed by topographic land survey (Leica 403 L total station combined with a GPS Trimble XT) as well as by sediment cores extracted with a Russian hand core sampler (with a diameter of 55 mm, and a length of 600 mm; see also Cubizolle, 2009). The mean frequency and distance between the land survey and cores was 5 m. Stratigraphies were described using the Munsell Soil Color Charts and the von Post test for the estimation of the decomposition degree and of the peat type (Payette & Rochefort, 2001). Peat initiation sampling and dating methods are detailed in Korhola (1995), Korhola et al. (2010), Weckström et al. (2010) and Cubizolle et al. (2007, 2012). The palaeoecological analyses were performed at two different localities on the mire (fig. 4).

11We thereby had to sample several cores at 15 to 20 cm distance from each other in order to obtain enough material to work with. At the point 28000 one core - the core 28000-B - was used for the palynological and diatom studies, and another core - the core 28000-C - was used for the macrofossil studies. A third core - 28000-A - was taken at the point 28000 in the context of research on the lateral development of the mire. Radiocarbon dates obtained on this core 28000-A were also used in order to verify the chronology of the sediment sequences used for the palaeoecological studies. At the point 28027 several cores were taken, one core - the core 28027-A - was used for additional palynological studies, whereas another core - 28027-B - was used for the micromorphological study. The upper 50 cm of peat from the 28027-A core was palynologically not analysed due to the highly decomposed peat. Finally, we checked that all the stratigraphies were well-correlated one towards the others, and performed radiocarbon dates on all of them.

3.2 - Palynological analysis

12Palynological research was performed on two core stratigraphies, one - the core 28027-A - situated in the western part of the mire, and the other core - 28000-B - situated in the central part of the mire. The extraction of all pollen and spores was done using the Thoulet liquid concentration method with a density of 2 (Argant, 1990). A minimum of 300 pollen and spore grains per sample were determined and counted, using a 3-cm sampling step, even narrowed to 2 cm for certain levels. Percentage values were calculated according to International standards and by using a 100 % reference sum made out of the total of arboreal (AP) and non-arboreal (NAP) pollen, but by excluding local taxa such as sedges (Cyperaceae) and all spore producing taxa (e.g. Sphagnum). Please note that damaged saccate pollen was counted as one, and that agglutinations of pollen from one and the same taxa were also counted as one. Palynological determination followed Reille (1999), and the definition of human impact indicators is based on Behre (1986), Richard (1997), and de Beaulieu et al. (1988). The palynological diagrams were plotted using the graphical computer programme TILIA (Grimm, 2011). Statistical zonation of the according pollen and spore finds into local pollen assemblage zones (LPAZ) was performed by using the computer programme CONISS (Grimm 1987).

13In the subsequent discussion chapter, we used classical pollen zones employed in the Massif Central in France (de Beaulieu et al., 1988) and in Central Europe in general (Haas et al., 1998), but which were also calibrated to calendar ages in order to allow wider chronological comparisons:

• beginning of the Subatlantic chronozone, 2,600 BP = about 2,700 cal. yr BP;

• beginning of the Subboreal chronozone, 5,000 BP = about 5,800 cal. yr BP;

• beginning of the Younger Atlantic chronozone, 6,000 BP = about 6900 cal. yr BP;

• beginning of the Older Atlantic chronozone, 8,000 BP = about 8,900 cal. yr BP;

• beginning of the Boreal chronozone, 9,000 BP = about 10,100 cal. yr BP;

• beginning of the Preboreal chronozone, 10,000 BP = about 11,700 cal. yr BP.

3.3 - Plant and Animal Macrofossil Analysis

14Plant and animal macrofossils analyses were performed at the Institute of Botany, University of Innsbruck (Austria) on the 158 cm long PRE-F 07-stratigraphy (point 28000-C, see above and fig. 4) of the La Prenarde-Pifoy mire sediments. Fourteen sediment samples of 2 cm sediment thickness each were extracted by hand at 10 cm intervals. After determination of the samples weight, the sediments were sieved under running water using 1,000, 500, 250 and 125 µm mesh sieves in order to pick out all determinable plant and macrofossils. All macrofossils > 250 µm were determined and counted under a stereo-microscope and by using according determination literature as well as the University of Innsbruck macrofossil reference collection. Abundant macrofossil taxa and species groups in some samples (such as bryophyte stems, undetermined vegetative parts or root remains) were counted for a determined sample volume and thereafter extrapolated to the entire, according sieve fraction. Finally, all macrofossil counts were transformed to a common sample weight of 20 g fresh-weight in order to make results comparable. Plotting of the results was done using the computer graphics programme TILIA (Grimm, 2011). Statistical zonation of the according macrofossil finds into local macrofossil assemblage zones (LMAZ) was performed by using the computer programme CONISS (Grimm, 1987).

3.4 - Diatoms analysis

15Diatoms are very abundant in minerotrophic mires like the La Prenarde-Pifoy mire. Their frustules are generally destroyed by pollen extraction techniques so a specific protocol exists for extraction, preparation for peat samples and determination (Serieyssol et al., 2011). In the laboratory, the cores are cut into 1-2 cm sections depending on the resolution needed. For removing organic material, 30 % hydrogen peroxide (H₂O₂) was added. After digestion of the organic matter, the sample is rinsed with distilled or demineralised water three or four times or until all traces of chemicals have been removed. Thereafter, diatoms can be separated from sediment particles by settling times. Diatoms are most readily observed using a microscope with differential interference contrast. The minimum number of valves needed in order to get an excellent representation of the diatom community is 300-350. Taxonomic references used were Germain (1981), Hustedt (1930, 1927-1966), Krammer and Lange-Bertalot (1988, 1991a,b, 2000, 2008), and Patrick and Reimer (1975a,b). Only diatom species with abundance > 2 % were included in the analysis (Brugam & Swain, 2000). Rare species were down-weighted.

3.5 - Micromorphological analysis

16Micromorphological analyses were performed by means of thin plates made by the Soil Sciences Laboratory of the University of Gand (Belgium) and followed the protocol by Fitzpatrick (1993). Such analyses are generally used to reconstruct local palaeoenvironmental settings in conjunction with other methods in palaeoecology (Puffe & Grosse-Brauckman, 1963; Babel, 1975; Lee, 1983; Davis & Wickham, 1987; Bouma et al., 1990). Microscopic structures are most readily observed using a polarized light microscope. Specific techniques were described in details by Castel (1991), Bullock et al. (1995), Korhola (1995), Bhiry (2001) and Prone (2003). The micromorphological approach combined several markers: organic matter, charcoal, sclerotia, mineral particles, as well as plants remains.

3.6 - Radiocarbon dates

17Estimation of the timing of peat initiation was on one hand obtained by radiocarbon dating of the basal peat layer, and by dating the facies layers above and below the peat (Cubizolle et al., 2007). This work was only performed on the 1.6 ha part of the mire presenting a peat thickness of more than 0.5 m. In these sapric levels of the basal peat accumulation it is impossible to clearly distinguish the different peat compounds with the exception of wood fragments sometimes found, which than were eliminated and, sometimes, dated separately (tab. 1). Consequently, the radiocarbon dates were performed on undifferentiated peat. For the non-peaty facies the wood was equally put apart, and the dates were performed on the undifferentiated organic matrix. This radiocarbon dating work was done on 20 different sediment cores, which equals in one per 750 m². Nevertheless, these cores are irregularly placed on the 1.6 ha area of the mire. For each of the dated sediment cores, the most ancient radiocarbon dates obtained on basal peat were further used for interpolation with ArcMap and by using the interpolation module.

18The radiocarbon dates obtained from the stratigraphies analysed with palaeoecological methods were performed on a series of material: undifferentiated peat, wood, and organic sediments (tab. 2; fig. 5 & 6). Several of the dates obtained from organic sediments were not usable, as they provided approximately the same age at different depth (gray-shaded on tab. 2).

19Radiocarbon dating was mainly performed at the “Centre de Datation par le Radiocarbone” of the Claude Bernard University (Lyon I, France), in collaboration with the Universities of Oxford and Groningen. Two additional dates were provided by the University of Utrecht laboratory (R.J. Van de Graaff Laboratorium, The Netherlands). Radiocarbon dates were calibrated using the computer program OxCal 2010 v. 4.1.7, which used the IntCal 2009 calibration curve (Reimer et al., 2009).

4 - Results

4.1 - Peat initiation and lateral expansion of the mire

20The stratigraphical analysis of the La Prenarde-Pifoy mire sediments clearly shows that the mire was build up due to paludification. The spatial distribution of the oldest radiocarbon dates obtained in the basal layers of the mire shows that peat accumulation began in the western part of the mire, at the end of the Younger Atlantic, between 6,200 and 5,895 cal. yr BP (tab. 1 & fig. 4). Thereafter the lateral expansion of the mire was as follows:

21- from 6,200 to 4,500 cal. yr BP the peat spread towards the east while peat initiation also started in the eastern part of the mire from 5,271-4,838 cal. yr BP. Around 4,500 cal. yr BP peat covered 2,900 m² (18,13 % of the present mire area). This oldest parts of the peat accumulation took place within the sector where peat thickness is more than 1.5 m today (fig. 4);

• from 4,500 to 3,000 cal. yr BP the lateral expansion of the peat is weak and very difficult to quantify as only two radiocarbon dates were obtained for this interval. On the other hand, we witness that most of the dates are in the 3,000-1,500 cal. yr BP interval (tab. 1). Indeed, from the end of Subboreal to the first part of Subatlantic, peat quickly spread and reached the slope in the eastern and southern parts of the mire. Around 1,500 cal. yr BP, 78.75 % of the present mire area (12,600 m²) was formed,

• finally, the last 3,400 m² of the present mire area has been covered by peat over the course of the last 1,500 years.

4.2 - Soil study

22In the eastern part of the mire the histosol is replaced on the slope by an anthroposol, which is now overgrown with Pteridium aquilinum and Poa chaixii (fig. 7).

23This anthroposol, which has been lain fallow since around two hundred years, shows the following profile (fig. 7):

• from 3 to 7 cm of Mul or Moder humus;

• a Ah horizon, from 5 to 30 cm thick;

• a B horizon, which was very silty in the upper part and sandier in the lower part, with more or less organic matter;

• a C horizon, hydromorphic, sandy-gravelous in the upper part of the plot and silty-argillaceous downwards the plot, with a lot of charcoal. At the depth of -62 and -68 cm an accumulation of charcoal was dated to 1,555-1,385 cal. yr BP (1585 ± 40 BP, Ly-14098).

4.3 - Palynology

4.3.1 - Core 28000 of La Prenarde-Pifoy

24Palynological analyses were performed on core 28000-B and rely to one of the sediment cores taken from the central part of the La Prenarde-Pifoy mire. The 158 cm long section stretches the last 6,000 years of sedimentation and has to be considered as a standard profile for the La Prenarde area by showing a classical stratigraphy with a sandy-clayey facies until 138 cm (with abundant wood fragments at 152 cm), which is topped by an organo-minerogenic sediment rich in organics from 138 to 125 cm depth (fig. 8). The sedimentological change at 138 cm is dated to about 4,300 cal. yr BP. From 125 to 98 cm depth a sapric peat rich in sand is present, which is followed around ca. 2,300 cal. yr BP by a layer rich in wood at 98 cm and which is than superposed by a mesic peat until 58 cm depth. Above 58 cm a fibric peat is dominant which also characterizes the today’s local mire sedimentation.

25According to the visual and CONISS-zonation into local pollen assemblage zones (LPAZ) the stratigraphy 28000-B from La Prenarde-Pifoy shows a palynological record for the last 6,000 years and can be subdivided into 3 main zones with 7 subzones (fig. 8):

• LPAZ-1 (core 28000-B; depth 158-141 cm; age 6,000-4,500 cal. yr BP; archaeological period: Neolithic) is subdivided into two subzones: LPAZ-1a lasts from 6,000 to 5,200 cal. yr BP and is characterized by relatively high pollen values of pine (Pinus), willow (Salix, up to 5 %), elm (Ulmus) and hazel (Corylus, > 25 %), rising values of oak (Quercus), as well as unusually high values of lime (Tilia) of more than 20 % (fig. 8). As this latter genus is insect pollinated and is normally only represented in low amounts in mid-Holocene peat profiles, dense lime woodland must therefore have surrounded the La Prenarde mire area at that time. Pollen values of classical human impact indicators are rare during this subzone, even if some (Artemisia) do show partly higher values. High values of Ranunculaceae and fern spores (monolete spores) do point to rather humid local mire conditions at that time. The subzone LPAZ-1b (5,200-4,500 cal. yr BP) shows a rather similar vegetation composition as in subzone LPAZ-1a, however, hazel value remain on a lower level (around 20 %), whereas birch (Betula) pollen values rise to nearly 10 %. Oak and lime pollen values do also rise, the latter showing maximal values of more than 20 % for oak and of more than 35 % for lime. Rising pollen values of grasses (Poaceae), Apiaceae and Rumex do imply - together with a general rise in herb pollen - some landscape opening due possibly to pastoral land use. Eventually some early form of Neolithic plant cultivation took also place around the mire as evidenced by the unusually high Fabaceae (legume) pollen values during this period.

• LPAZ-2 (core 28000-B; depth 141-78 cm; age 4,500-1,600 cal. yr BP; archaeological periods: Neolithic to Roman Age) is subdivided into three subzones: LPAZ-2a lasts from 4,500 to 3,800 cal. yr BP and is characterized by rapidly decreasing oak, hazel and lime values, and rapidly rising alder (Alnus), beech (Fagus) and fir (Abies) values. The establishment of a classical Abieti-Fagetum around 4,500 cal. yr BP can therefore be recognized in the La Prenarde area. During this major vegetation transition typical human impact indicators are absent or show very low values only. However, first continuous pollen values of heather (Calluna) are present, which might point to some abandoned pastoral land or some slight drought on the mire surface at that time. The latter might be more probable as Ranunculaceae and Cyperaceae values do also show some reduction during this subzone. The subzone LPAZ-2b lasts from 3,800 to about 2,200 cal. yr BP and shows massively reduced values of birch, hazel, and lime, whereas alder, beech and fir do now reach their maximal values (Alnus > 40 %, Fagus > 20 %, Abies > 25 %). Apart from an alluvial alder forest on and around the mire a general upland reforestation can be stated for this time period, and even is some anthropogenic pollen indicators and one cereal type pollen grain are present on the herb side, there is probably only modest land use around the La Prenarde mire until 2,200 cal. yr BP according to the results from core 28000-B. However, during the last 100-200 years of this subzone the mire might have become much dryer than beforehand, as shown by the massively rising blueberry (Vaccinium) pollen values together with the sharply decreasing Ranunculaceae and fern spore values. Finally, subzone LPAZ-2c (lasting from 2,200 to 1,600 cal. yr BP) is characterized by the very rapid local demise of the Abieti-Fagetum (Abies & Fagus), which points to clear-cutting of the forests around the La Prenarde mire during the Late Iron Age and Roman Period. This forest opening is paralleled by a massive rise in grass (Poaceae) pollen values, as well as in a general increase of plantain (Plantago) values (towards more than 2 %) which clearly indicates pastoral activities and vegetation related to footpaths or trodden areas as produced by huge numbers of livestock. The presence of nettle (Urtica) in higher values might also point to some eutrophication of the mire area by livestock excrements or nutrient rich in-running waters.

• LPAZ-3 (core 28000-B; above 78 cm depth; age: last 1,660 years; archaeological periods: Medieval Period and Modern Times) is subdivided into two subzones: LPAZ-3a lasts from 1,600 to 1,200 cal. yr BP and is characterized by massively reduced alder (Alnus) values (probably originating in local hydrological changes of the mire) together with a massive rise in pine (Pinus) values, all pointing to a major microclimatic shift towards dryer conditions. Together with this pine expansion, a general reforestation of the area can be recorded by the newly rise of beech and fir values, which might represent the reduced anthropogenic and livestock pressure on the area during the so-called Migration Period. The subzone LPAZ-3b (representing the vegetation of the last 1,260 years) shows a very similar vegetation composition as described in LPAZ-3a, however, some new arboreal species are now present throughout this zone, as for example hornbeam (Carpinus), long-distance transported olive tree pollen (Olea) and walnut (Juglans). In addition, high cereal pollen type values starting about 1,200 cal. yr BP do show the importance of the area for cereal cultivation during the Medieval Times. Possibly some legume cultivation took also place around the La Prenarde mire as evidenced by the high Fabaceae pollen values for the Late Medieval Period. High upland pollen values and abundant Poaceae, Plantago, Rumex, as well as Calluna pollen values do highlight the intensive pastoral use of the area since Medieval Times. Interestingly, some vine (Vitis) pollen are present in the topmost sediments, which are related to Modern Times, possibly - as this plant is insect pollinated - revealing the historical cultivation of Vitis for wine-making in the close valleys (which are still in use as vineyards today). Finally, the massive expansion of sedges during the last few centuries might be due to hydrological changes related to local anthropogenic activities impacting the mire surface itself.

4.3.2 - Core 28027-A of La Prenarde-Pifoy

26Palynological analyses were also performed on core 28027-A, and rely to one of the sediment cores taken from the western part of the La Prenarde-Pifoy mire. The 120 cm long stratigraphy shows an organo-minerogenic facies rich in organic compounds until 114 cm depth, followed by a sapric peat rich in wood remains until 93 cm depth (fig. 9). Please note, that the change from the organo-minerogenic sediments to the peat formation is dated to about 4,900 cal. yr BP. From 93-50 cm the stratigraphy is characterized by mesic peat, which is finally followed by decomposed peat to the top of the sequence.

27According to the visual and CONISS-zonation into local pollen assemblage zones (LPAZ) the stratigraphy 28027-A from La Prenarde-Pifoy shows a palynological record for the last ca. 5,260 years and can be subdivided into 4 main zones (fig. 9):

• LPAZ-1 (core 28027-A; depth 120-105 cm; age 5,200-4,400 cal. yr BP; archaeological period: Neolithic): This LPAZ-1 is characterized by relatively high pollen values of hazel (Corylus, ca. 20 %), oak (Quercus, ca. 15 %), beech (Fagus, ca. 15 %), and lime (Tilia, > 5 %), whereas alder (Alnus) values remain low (fig. 9). In the lower part of this zone, some cereal type pollen as well as some pollen from herbs (Apiaceae, Brassicaceae, Valerianaceae, showing continuous pollen values above 1 %) may point to some open land around the La Prenarde mire at that time. In addition, reasonably high pollen values of meadowsweet (Filipendula), as well as Cyperaceae and ferns (the latter two taxa were excluded from the 100 % pollen sum), are representation of relatively wet local mire conditions;

• LPAZ-2 (core 28027-A; depth 105-77 cm; age 4,400-3,200 cal. yr BP; archaeological periods: final Neolithic & Bronze Age): This LPAZ-2 is characterized by a successive decline of Corylus and Tilia, which is paralleled to a massive rise in Alnus (with values > 30 %) and a regular rise of fir (Abies) towards pollen values of 15 % (fig. 9). Some typical human impact indicators such as plantain (Plantago) and Rumex are present, together with high values of Fabaceae and some Liliaceae, which may possibly point to a local cultivation of legumes in the surroundings of La Prenarde, as these taxa are normally rarely found in mire sediments. The values of Ranunculaceae, Filipendula and ferns remain high and do show a constantly high humidity of the mire, which is also confirmed by the first presence of Sphagnum moss spores;

• LPAZ-3 (core 28027-A; depth 77-60 cm; age 3,200-2,500 cal. yr BP; archaeological periods: Late Bronze Age and Early Iron Age): this LPAZ-3 is characterized by an increase in the pollen values of sedges (Cyperaceae) and by a slight general reforestation (fig. 9) lasting until 2,700 cal. yr BP (to values of nearly 85 % arboreal pollen) and which might be related to known climatic deterioration at that time (Haas et al., 1998). This phase is also paralleled with a decrease in grass pollen presence (Poaceae), which might point to a lower pastoral pressure in the region. Towards the last two hundred years of this LPAZ the Cerealia type pollen values increase again, together with the grass values, as well as the Chicoriaceae and heather (Calluna) values, all pointing to a massive rise in human and pastoral impact around La Prenarde during the Early Iron Age;

• finally, LPAZ-4 (core 28027-A; depth 60-51 cm; age 2,500-2,100 cal. yr BP; archaeological period: Iron Age) is characterized by higher values of pine (Pinus), which are paralleled with massively reduced alder, beech (Fagus), fir (Abies), Cyperaceae, and fern values (fig. 9), all pointing to locally much dryer climatic conditions during the Iron Age. The first presence of hornbeam (Carpinus) as well as constantly high values of several human impact indicators (e.g. Cerealia, Plantago, Rumex, Calluna) do point - together with the general rise of total herb pollen values towards nearly 40 % - to a regular use of the areas surrounding La Prenarde for cultivation. The upper 50 cm of the core were not analysed palynologically due to the poor pollen and spore preservation within the highly decomposed peat.

4.4 - Plant and animal macrofossil analysis

28Macrofossil analyses were performed on core 28000-C only and rely to the sediment core taken from the central part of the La Prenarde-Pifoy mire. The 158 cm long section already studied for pollen and spores stretches the last 6,000 years of sedimentation (described above) and can be subdivided into five local macrofossil assemblage zones (LMAZ) according to visual and statistical CONISS zonation (fig. 10).

• LMAZ-1 (core 28000-C; depth 158-141 cm; age 6,000-4,500 cal. yr BP; archaeological periods: Neolithic): This LMAZ-1 fits astonishingly well to the local pollen zone LPAZ-1 from the same core and is characterized by a high number of wood remains (substantia lignosa) pointing to an alluvial forest (probably made out of willow bushes as evidenced by the pollen content; fig. 8) in the vicinity of the coring location on the central part of the mire. High values of rather large minerogenic particles (sand > 1 mm) and some few charcoal parts in the lowermost sample (fig. 10) may be seen correlated to the mire formation and erosional processes around 6,000 years ago;

• LMAZ-2 (core 28000-C; depth 141-123 cm; age 4,500-3,200 cal. yr BP; archaeological periods: Final Neolithic and Bronze Age) shows a very similar macrofossil content as in LMAZ-1 with the exception of lower amounts of wood remains, somehow higher amounts of vegetative and root remains as well as the first abundant presence of insect chitin remains, possibly reflecting some more open biotopes (fig. 10). Minerogenic particles are still abundantly found, showing continuing erosional phenomena;

• LMAZ-3 (core 28000-C; depth 123-86 cm; age 3,200-2,200 cal. yr BP; archaeological periods: Late Bronze Age and Iron Age) shows again higher values of wood and root remains, as well as quite high values of Coenococcum geophilum sclerotia (fungi), which are good indicators of soil disturbance and erosional processes (fig. 10). In addition, abundant minerogenic particles and some macroscopic charcoal are regularly present during the Iron Age part of this zone (i.e. ca. 2,500-2,200 cal. yr BP) pointing to some erosion and fire events on or around the mire for that time period. The latter might be more reasonable to think of as this Iron Age phase is also characterized by remains of aquatic animals (e.g. Trichoptera) pointing to the presence of stagnant or slightly running water at the centre of the mire. Some Cyperaceae fruits are also present at the end of this zone at around 2,200 cal. yr BP, which parallels well with this more humid phase and with the synchronously rising sedge pollen values (fig. 8);

• LMAZ-4 (core 28000-C; depth 86-60 cm; age 2,200-1,000 cal. yr BP; archaeological periods: Late Iron Age to Medieval Times) shows a similar macrofossil content as the zone before, however, a huge peak in macroscopic charcoal is revealed for the time span around 1200 cal. yr BP (fig. 10), possibly related to Medieval human activity on the mire or in the near forest surroundings;

• finally, LMAZ-5 (core 28000-C; depth 60-1 cm; age: last 1,060 years (= since ca. 1,000 cal. yr BP); archaeological periods: Medieval & Modern Times) is characterized by some single finds of tree remains (e.g. Betula and Pinus) as well as high amounts of Cyperaceae remains (fig. 10). Lots of different herb macrofossil (especially for the early Modern Times) do either represent an increase in overall plant diversity due to disturbance factors or a more open landscape, or may be a result of a better preservation within these relatively young samples. The find of the aquatic vascular plant Najas marina has thereby also to be highlighted, as it only grows in oligotrophic to eutrophic water, showing thereby a possible change in water quality about 400 years ago, even if in parallel bryophyte values increase massively (fig. 10).

4.5 - Diatoms analysis

29Diatoms analyses results were performed on core 28000-B, parallel to palynology and to macrofossils analyses on core 28000-C. The 158 cm long section can be subdivided into five local diatoms assemblage zone (LDAZ) according to visual and statistical CONISS zonation (fig. 11):

• LDAZ-1 (core 28000-B; depth 158-139 cm; age 6,000-4,300 cal. yr BP; archaeological periods: Neolithic). There are very few frustules in these mineral basal layers. Only remains of Pinnularia viridis are present, a benthic, aerophilous and ubiquist species;

• LDAZ-2 (core 28000-B; depth 139-126 cm; age 4,500-3,200 cal. yr BP; archaeological periods: Final Neolithic and Bronze Age). A few more species were present in this zone. The main taxa are from the genera Eunotia, Pinnularia and Cocconeis, while Pinnularia divergens is very abundant revealing an acidiobiontic mire. Planktonic diatoms are almost totally absent. The appearance of Cocconeis placentula shows the eutrophic character of the mire. Oligo-mesotrophic species are almost totally absent;

• LDAZ-3 (core 28000-B; depth 126-84 cm; age 3,200-2,200 cal. yr BP; archaeological periods: Late Bronze Age and Iron Age). The greater diversity in the diatoms assemblages is observed. This change correlates with peat initiation (fig. 11). Benthic diatoms are always very abundant with a great number of Pinnularia but also Cocconeis placentula and several species of Fragilaria and Gomphonema appeared (fig. 11). Nevertheless, five planktonic species appeared, especially Aulacoseira distans and Aulacoseira alpigena. Eutrophic diatoms are more abundant than oligo-mesotrophic diatoms in particular between 2,400 and 2,000 cal. yr BP (fig. 11);

• LDAZ-4 (core 28000-B; depth 84-61 cm; age 2,200-1,000 cal. yr BP; archaeological periods: Late Iron Age to Medieval Times). An important change occurs with a pronounced increase in planktonic diatoms, which are almost as abundant as the benthic diatoms. However, the diversity of planktonic species is small with essentially Aulacoseira alpigena, Aulacoseira distans and Fragilaria ulna that indicate greater humidity and the development of open, still pools or ponds in the mire. Secondly, eutrophic diatoms are as numerous as the oligo-mesotrophic diatoms. The proportion of Pinnularia viridis the most ubiquist species decreases significantly;

• finally, LDAZ-5 (core 28000-B; depth 61-1 cm; age: last 1,060 years (=since ca. 1,000 cal. yr BP); archaeological periods: Medieval to Modern Times). Very little difference from the last zone is recorded. However, a decrease in eutrophic diatoms abundance was registered. Even so Aulacoseira alpigena, Aulacoseira distans and Fragilaria ulna remain abundant. Oligo-mesotrophic diatoms are more numerous in particular Gomphonema angustatum, Fragilaria construens var. venter and Tabellaria ventricosa. Fragilaria constricta, an oligo-saprobic species, appears sporadically and Eunotia hexaglyphis, an oligotrophic species, increase in abundance. Indeed some species as Eunotia minor and Pinnularia subcapitata reveal the acidity of the mire.

4.6 - Micromorphological analysis

30The micromorphological study of core 28027-B gives useful information essentially related to charcoal and sclerote abundances (fig. 12):

• charcoal is very numerous in the basal layers of the core from 120 cm to 114 cm in the organic or very organic sandy gravelous sediments and sapric peat. Two radiocarbon dates were obtained in these layers: 6,174-5,906 cal. yr BP at the depth of 118 cm and 4,523-4,410 cal. yr BP at the depth of 115 cm;

• above, between 115 cm (4,523-4,410 cal. yr BP) and 63 cm (2,744-2,491 cal. yr BP), very few charcoals appeared;

• charcoal becomes numerous again above 63 cm (2,744-2,491 cal. yr BP);

• the distribution of sclerotes is almost the exact opposite with a lot of remains between -96 cm and -69 cm.

5 - Discussion

31The chronology of cultural innovation may highly differ for the Neolithic Period according to different authors. In the eastern part of the Massif Central, Georges (2007) used the one based on regional archaeological data and distinguished a first Neolithic (7,500-5,800 cal. yr BP) from a second Neolithic Period (5,800-4,300 cal. yr BP). But as this is not a common chronology, we prefer to use the chronology proposed by Demoule (2007) for France: Early Neolithic (7,500-6,900 cal. yr BP), Middle Neolithic I (6,900-6,200 cal. yr BP), Middle Neolithic II (6,200-5,500 cal. yr BP), Final Neolithic (5,500-4,500 cal. yr BP), Final Neolithic/Chalcolithic (4,500-4,100 cal. yr BP), and the Bronze Age starting 4,100 cal. yr BP.

32In general terms, the statistical CONISS zonation into local assemblage zones of the 28000 stratigraphy resulted in high similarities between pollen, diatom and macrofossil data. The same accounts for satisfactory correlations related to the results obtained on the stratigraphies 28027-A and -B. Nevertheless, it is the information gained by the palynological analyses, which traces back the palaeoenvironment around the La Prenarde mire since the Atlantic Period within three major phases.

5.1 - Phase 1: The end of the Younger Atlantic and the beginning of the Subboreal (Middle Neolithic II)

33The first key period correlates with the change from the Late Atlantic to the Subboreal Period dated to 5,800 cal. yr BP within the 28000-B stratigraphy. Note that several radiocarbon dates obtained within the organo-minerogenic layers of the 28000-C stratigraphy were not usable independently from their sample depth as they all keyed out at 2,500 BP, even if the pollen stratigraphy clearly showed a real age around the beginning of the Subboreal. The proximity of a palaeo-channel of the Chantereine rivulet was probably responsible for this problematic, given that three dates obtained from the same levels within the adjacent 28000-A stratigraphy showed perfectly fitting dates. These dates were therefore used for chronological interpolation of the according layers of the 28000-B and C stratigraphies. In this central part of the mire the peat had not yet accumulated at the end of the Younger Atlantic Period. However, the palynological data as well as the presence of wood remains indicate that the milieu was wet and wooded. Around the mire, the broadleaved forest was characterized by the dominance of lime (Tilia). Within the LPAZ-1 of the 28000-B core no real indication of anthropisation is perceptible except for an unusual presence of high Fabaceae pollen values. In contrast, some cereal pollen do show up at the base of the LPAZ-1 of the 28027 core between 5,200-4,800 cal. yr BP, over the course of the Final Neolithic, together with a whole row of herbaceous species such as Apiaceae, Brassicaceae, and Valerianaceae. On the other hand charcoal is very abundant around the same period in the basal layers of peat core 28027-C between 5,200 and 4,400 cal. yr BP (fig. 12.), a situation which was found in all cores in this peatland area. In addition, two archaeological sites at 3 and 4 km distance from La Prenarde-Pifoy mire respectively (Georges, 2007; fig. 1) and dating to the Final Neolithic Period (ca. 5,200-5,000 cal. yr BP) do corroborate this human and livestock action. In a wider context the Forez basin lowlands do incorporate archaeological signs of human presence since at least the Final Neolithic. Nevertheless, even if the oldest archaeologically excavated settlements date to around 5,700 cal. yr BP only (Georges, 2007), some additional evidence for numerous Middle and Final Neolithic settlements on the highest areas of the alluvial Loire River plain in the Forez lowlands exists (studies in progress, unpublished results, F. Delrieu, Service Régional de l’Archéologie, DRAC Rhône-Alpes, France, pers. com. 2012). In the context of the eastern parts of the Massif Central it is notable that the first palynological indication for anthropisation was recorded around 5,700 and 5,400 cal. yr BP, at the end of the Middle Neolithic II and the Final Neolithic (Argant & Cubizolle, 2005; Cubizolle et al., 2012). At one of the recently studied mire sites - the Etui raised bog within the Bois Noirs north of our study area (fig. 1) - these first evidences of human action are dated to the Final Neolithic (5,446-5,082 cal. yr BP). Finally, the end of the Younger Atlantic Period marks also the beginning of peat accumulation at La Prenarde-Pifoy, at first in the area of the core 28027, around 6,204-5,895 cal. yr BP (fig. 4) and subsequently around 5,271-4,836 cal. yr BP about 30 metres south-east of the 28000 core point. On one hand the date of 6,204-5,895 cal. yr BP fits well the known period of worldwide humidity and high lake levels as described by Magny (2004), on the other hand the date of 5,271-4,836 cal. yr BP (representing the start of peat accumulation in the south-eastern part of the peat) falls into a period generally known for low lake levels and drier climatic conditions in eastern France. Consequently, this absence of good correlation between climatic control and the chronology of lateral peat expansion leads us to consider an anthropogenic origin of the local peat accumulation processes. Based on models developed by our English colleagues (Moore, 1975, 1988, 1993; Caseldine & Hatton, 1993) it is well possible, that cutting forests combined with anthropogenic fires and pastures in the lowland valley for the alimentations of agro-pastoral societies may have changed the hydrological balance from negative towards zero or even positive (leading to the mire development), especially if we keep in mind, that such an opening may have favoured herb dominated prairies for livestock.

34Our results therefore sustain the hypothesis already proposed in the past that first agro-pastoral activities started during the end of the Middle Neolithic in some areas of the eastern, granitic Massif Central (Argant & Cubizolle, 2005). However, even if this early impact is regularly found, the nature and intensity of this human occupation remains unclear. Nevertheless, the human/livestock impact at La Prenarde-Pifoy remains much less pronounced than the one described for the Neolithic (from 6,500-5,000 cal. yr BP) within the northerly lying Morvan Mountain range (Jouffroy-Bapicot, 2010). The La Prenarde-Pifoy results would therefore much better fit the results from the Millevaches Plateau in the west of the Massif Central. There, first hints of human impact appear between 6,500 and 6,000 cal. yr BP through palynological indicators (Miras et al., 2011), which are followed by some reinforced signals during the Middle Neolithic and which than are culminating into a strong impact at the change from the Neolithic to the Bronze Age.

5.2 - Phase 2: Subboreal (the end of the Final Neolithic and Bronze Age)

35At La Prenarde-Pifoy mire the next strong period of human impact is recorded palynologically during the Subboreal (from 4,800-2,800 cal. yr BP in the 28027 stratigraphy, and from 4,500-2,700 cal. yr BP in the 28000 stratigraphy). This period corresponds with the end of the Final Neolithic and the Bronze Age. This period is also marked by a slowing down of the lateral expansion of the mire between 4,500 and 3,000 cal. yr BP (fig. 4), and seems to have happened without any climatic correlation (Barber et al., 2004; Magny, 2004; Cubizolle et al., 2012). This reduced peat accumulation is also pronounced through the significant reduction of the importance of sedges (Cyperaceae) and of Sphagnum peat mosses (fig. 8). The surroundings of the mire became highly wooded with percentage values for trees and shrubs of more than 80 % in both stratigraphies (fig. 8 & 9), culminating even at 95 % in the 28000-B core around 3,300 cal. yr BP with a dominance of alder, beech and fir.

36In addition, a major forest composition change started around 4,600-4,500 cal. yr BP with a rapid succession towards a beech-fir forest, which is a classical scenario for this time period in Western Europe. Indeed, this development of beech and fir shows some singularities at La Prenarde-Pifoy in terms of a regional context: First, the beech expansion happens much later than previously known, as normally the mean age for a Fagetum development is recorded at ca. 5,400 cal. yr BP in the Monts du Forez and Monts de la Madeleine Mountains, and at ca. 4,950 cal. yr BP in the Pilat Massif (Argant & Cubizolle, 2005). On the other hand the presence of Fagus and Abies is recorded synchronously and with a similar importance of the two species at the La Prenarde-Pifoy mire, which is very different from all other palaeoecological sites studied in the area where fir trees expanded much more progressively and later than beech (Argant & Cubizolle, 2005). However, some dating insecurities might exist as fir and beech are present at the base of the 28027 stratigraphy already around 5,200 cal. yr BP with values up to 15 % for beech and up to 5 % for fir. These relatively small values increase somehow after 4,400 cal. yr BP but do never reach values higher than 20 % for Fagus and 10% for Abies (fig. 8 & 9).

37Concerning palynological indicators for agro-pastoral activities they are more or less absent within both stratigraphies. Two singular evidences of cereal type pollen and of Rumex around 3,200 cal. yr BP in core 28000-B are opposed to a decrease of Poaceae and the absence of other pastoral indicators (fig. 8). The same accounts for the 28027 stratigraphy even if some few Rumex and Plantago pollen are present (fig. 9), which fits the absence of charcoal particles in the same core between 4,500 and 3,000 cal. yr BP (fig. 12). Together with the palynological evidence of a forest extension this implies, that human action was minimal during this time period, i.e. the end of the Neolithic and whole of the Bronze Age. However, this is in some contrast to the archaeological evidences obtained in the lowland basin of the Forez region, where a pronounced human presence is recorded for the Bronze Age (Georges, 2007), situation which is also confirmed by the synchronous activities in the Cantal area (Miras et al., 2011). All this discussion leads us to consider that the development of human action around the La Prenard-Pifoy mire may represent a somehow more local than regional phenomenon, more based on the use of local resources in the lowland valleys than at higher altitudes.

5.3 - Phase 3: Since the beginnings of the Iron Age and Subatlantic Period

38Contrary to the Bronze Age, huge agro-pastoral action can be recorded in the Monts du Forez and around La Prenarde-Pifoy miremire since 2,600 cal. yr BP. This Iron Age action must be seen as a major turnover in the local history of nature-society interactions, as this impact on mountain vegetation and landscape remains present on a high level until the end of the 19^th century AD. Over the course of these last 2,600 years we can highlight four periods.

5.3.1 - Period 1: the First Iron Age

39First major impact occurred during the first part of the Iron Age, best visible from the results obtained on the 28000-B stratigraphy. A rapid reduction of the arboreal pollen total together with a peak of Poeaceae characterizes this phase between 2,600 and 2,400 cal. yr BP, which is also characterized by the inverse evolution of Fagus and Abies during these times. On one hand beech values are slightly increasing and on the other hand fir values show a remarkable peak together with alder (fig. 8). This favourable evolution for Abies is frequently found in palynological studies of the Massif Central (sometimes already since the Bronze Age) and has often been discussed as an indication of human action on beech-fir forests (de Beaulieu et al., 1988; Argant & Cubizolle, 2005; Cubizolle, 2005). Additionally, a certain abundance of Coenococcum geophilum is found between 2,900 and 2,300 cal. yr BP (850-300 BC) possibly indicating (due to the ecology of this fungi) massive soil erosion probably related to the agricultural use of former forest soils. However, no direct sign of cultural plant growth is recordable during this rather short phase, given that after 2,400 cal. yr BP the forest is obviously recovering somehow. The comparison with other sites palaeoecologically studied in the Massif Central is difficult, as high resolution data for the first Iron Age are often missing, except for the Limousin area where some strong anthropological signals were recorded (Allée et al., 1997; Allée, 2003; Allée & Lespez, 2006). From an archaeological point of view the First Iron Age is well documented in the more western regions as the Limagne lowlands where an intensification of Hallstatt settlement activities was recorded (Trément et al., 2007).

5.3.2 - Period 2: The Second Iron Age and the Gallo‑RomanTimes

40The 28027 stratigraphy shows a very similar vegetation and landscape development, even if a pronounced human impact remains visible until 2,100 cal. yr BP through a notable reduction of tree pollen percentages, high abundance of charcoal particles, as well as the well-developed presence of cereal type pollen, and pastoral indicators such as Poaceae, Calluna and Rumex. After 2100 cal. yr BP, from the end of the Second Iron Age, the stratigraphy 28000 shows an important reduction of the forest cover again, mainly due to the demise of Fagus and Abies lasting until ca. 1600 cal. yr BP (the end of the Gallo-Roman Empire). During this period the values of Poaceae and Plantago do also rise, even if the cereal type pollen and Calluna are absent. Nevertheless, the Second Iron Age and Gallo-Roman Times have to be seen as key period for the area around the La Prenarde-Pifoy mire as the agro-pastoral society became sedentary at long-term, and implying a substantial subsequent impact on the local landscape. Archaeological data for this time period are scarce, so a further discussion on the evidence of former settlement structures is not possible at the moment, even if some Gallo-Roman activities are recorded in the Ambert basin nearby (Fassion et al., in press). Other elements do reinforce our interpretation presented here, as for example through the major lateral expansion of the La Prenarde-Pifoy miremire dated to 2,600-1,900 cal. yr BP (fig. 4) and through the growth of a huge additional number of peatlands due to anthropogenic origin, which seems to happen completely independent from all known climatic phases of higher precipitation or humidity (Barber et al., 2004; Magny, 2004; Cubizolle et al., 2012). Here again, the most probable explanations relies to major modifications of the vegetation cover by humans, which may have reduced evapotranspiration, and a change in the hydrological balance from negative towards zero or positive. In this context it is notable that sedges (Cyperaceae) took over on the mire surface together with a reasonable growth of Sphagnum peat mosses (fig. 8). The high percentages of planktonic diatoms between 1,900-1,800 cal. yr BP (fig. 11) corroborate somehow these results even if they fall into a slightly latter phase.

41Palaeoecological evidence for significant agro-pastoral developments since the Late Iron Age exists for several areas of the Massif Central since the 1970^th (de Beaulieu et al., 1988). Recently, similar results were obtained using a synthesis made from nine different palynological sequences from the eastern granitic Massif Central and showing a pronounced and permanent intensification of agro-pastoral activities since 2,400 cal. yr BP (Cubizolle et al., 2012) similar to the one synchronously recorded on the Millevaches Plateau in the Limousin area and for the Morvan area north of the Monts du Forez area (Jouffroy-Bapicot, 2010; Miras et al., 2011). Other evidence comes from Late Iron Age intentional burning and forest clearings (Cubizolle et al., 2001) as well as from erosional features related to agriculture in the Limousin area (Allée, 2003). At least, a lot of mires with an anthropogenic origin appeared between 2,400 and 2,150 cal. yr BP in the eastern part of the Massif Central (Cubizolle et al., 2012).

5.3.3 - Period 3: The Medieval Times

42From around 1,600 cal. yr BP onwards the total arboreal cover (as represented by the total tree and shrub pollen percentages) increases again mainly due to the rise in importance of pine, beech and subsequently fir (fig. 8) during the migration period/Early Medieval Times. Around 1,250 cal. yr BP (8th century AD) this arboreal total reaches nearly 90 % of the pollen sum, even if some cereal type pollen, as well as high Plantago pollen values do show that some human action must have existed within a highly forested landscape. This is confirmed by the presence of an anthroposol between 1,555 and 1,385 cal. yr BP (between the 4th and 6th century AD) (fig. 7) and by a first presence of hornbeam (Carpinus) pollen, a quickly growing tree known to have been introduced and used for the production of pole wood, for example for vine growing during the Medieval and Modern Times. A much clearer agro-pastoral impact is visible palynologically after the 8th century AD during the High Medieval Period and through the nearly complete demise of the beech-fir forest and an absolute low-stand (for all of the stratigraphy) of the total arboreal values to less than 60 % around 1,050 cal. yr BP (10th century AD), which more or less means that an open, treeless landscape existed at that time. This landscape opening is clearly related to the installation of medieval towns and villages in the region, which highly impacted on the forest composition because of the timber used for building constructions. It is therefore of special interest, that fir values did completely disappear at La Prenarde-Pifoy towards the end of the Medieval Period probably related to the excellent quality of Abies as construction timber. Extremely high values of Cerealia type pollen, as well as of Rumex, Calluna, Poaceae and of Plantago do well represent the massive use of former forest soils for agriculture and pasture. Rising Carpinus values, as well as the appearance of walnut pollen (Juglans) around 800 cal. yr BP (12^th century AD), which was then used as a fruit tree in the lowland valleys, do complete this aspect of permanent, large-scale anthropisation of the Forez basin and uplands. This picture of a rather low medieval human impact until the 8^th century AD at altitudes above 1,100 m a.s.l., followed by a massive agro-pastoral action (including upland cereal cultivation) after the 8^th century AD is a common feature described by palynological studies in the eastern, granitic part of the Massif Central (Argant & Cubizolle, 2005; Fassion et al., in press). Other evidence for this massive human action and for the clearing of the beech-fir forest exists from the small Jonziecq valley just 3 km south of the La Prenarde-Pifoy site (Cubizolle, 2005; Cubizolle et al., 2012). In addition, it is during the 4^th and 10^th century AD where most of the regional peatlands are build up considering all of the Holocene, even if fewer seems to build up around the year 1,000 AD (Cubizolle et al., 2012). Interestingly, a huge rise in heather (Calluna) values is recordable between the 10^th and 15^th century AD at the La Prenarde-Pifoy mire, suggesting the use of slash and burn agriculture for this time span, as the massive presence of Calluna pollen values might be seen as a reaction to fire events. This strong agro-pastoral activity during the High and Late Medieval Period is also recorded in the northerly lying Morvan Mountains (Jouffroy-Bapicot, 2010) as well as in other uplands such as the western part of the Bretagne (Barbier, 1999).

5.3.4 - Period 3: Modern Times

43A remarkable palaeoecological change is also recorded for the first part of the Modern Times (500-150 cal. yr BP), where the forest cover increases again according to the palynological results from the 28000-B stratigraphy, especially due to rising importance of beech, fir and increasing pine values. An interesting peak of Fabaceae pollen in the 28000-B core around the change from the late Medieval Period to the Modern Times may be related to the cultivation of legumes around the La Prenarde-Pifoy site, as such high pollen values are rarely found due to Fabaceae species growing on bogs, but mostly because of according field cultivation. However, such a cultivation of Fabaceae in the area would need direct evidence by archaeobotanical studies from archaeological sites nearby or from written sources, and remain therefore a hypothesis for the moment. Starting with the 15^th century AD, the values of cereals, Calluna and Poaceae are reduced in parallel, situation which fits well the Lac Pavin record in the northern part of the Massif Central (Stebich et al., 2005) and which fits the well-known morpho-climatic evolution in the north-eastern part of the Massif Central (Cubizolle, 1997). The explanation to be forwarded here is that this reduced human activity corresponds to the Little Ice Age cooling. The second half of the 18^th century AD was thereby especially difficult in terms of climatic conditions, as this period has been revealed as pretty humid and cold, with freezing temperature even by the end of July, which accounted for a large fail in cultural plant yield in the Livradois-Forez region at that time. This climatic deterioration is visible at the La Prenarde-Pifoy site through a rise in sedges (Cyperaceae) and through the development of planktonic diatoms on the other hand (fig. 11).

6 - Conclusions

44The interpretation of the geomorphological and palaeoecological data obtained from the La Prenarde-Pifoy mire leads us to focus on three major and decisive environmental periods of environmental and climatic change. First major agro-pastoral activities were recorded at the change from the Younger Atlantic to the Subboreal chronozone, i.e. between 5,200 and 4,800 cal. yr BP (which corresponds to the Final Neolithic Period), where these activities seem to be huge enough so they can be recorded palaeoecologically even above 1,000 m a.s.l. in the eastern, granitic Massif Central. However, concerning the dynamics of peatland formation it is impossible to definitely decide between an anthropogenic or climatic origin. The second major phase of interest covers a long period between 4,800 and 2,700 cal. yr BP (second part of the Subboreal and the end of the Final Neolithic and Bronze Age) and is characterized by the dominance of a fir-beech forest with nearly no recordable human action and only modest lateral expansion of the La Prenarde-Pifoy mire. Finally, the last major phase starts around 2,700-2,600 cal. yr BP (Subatlantic chronozone) with the beginning of the Iron Age and is characterized by massif agro-pastoral activities of different nature and intensity. It is also a phase of lateral extension of the mire, with a maximum of peat formation between 2,600 and 1,900 cal. yr BP. During the first part of the Iron Age some modest, local human activities are recorded in the upland mountains of the Monts du Forez and around the La Prenarde-Pifoy mire. The fir-beech forest was then locally reduced in space, however the cultural fields probably occupied only a limited area as shown by the palynological evidence. During the second part of the Iron Age and during the Gallo-Roman period a strong accentuation and durability of the agro-pastoral impact is recorded. During the Medieval Times and starting with the 8^th century AD a cereal based agricultural system is installed in the Monts du Forez mountain area on the most favourable soils, and possibly using slash-and-burn agricultural techniques, a cereal based agricultural system, which is also known from other upland areas of the Massif Central. Finally, the climatic pejoration of the so-called Little Ice Age Period was very probably responsible for the clear, sometimes local, reduction of agro-pastoral activities between the 15^th and 18^th century AD in the mountainous areas of the north-eastern Massif Central above 1,000 m a.s.l.