A 14,000 year peatland record of environmental change in the southern Gutland region,Luxembourg

Geographical setting, regional climate, and vegetation

The peatland is located near the village of Niederpallen in the commune of Redange (5°54′E, 49°45′N) (Figure 1). It covers a small area of about 1 ha and is surrounded by extensive and intensive grassland and agricultural fields.

OPEN IN VIEWER

Figure 1.

Regional setting of the study area in the southwest of Luxembourg.

The Rouer peatland is part of the hilly Gutland landscape, which is typical for the south of Luxembourg. The Gutland is characterized by weathered calcareous geological substrates. Trias formations like red sandstone, shell limestone, Keuper, and also Lias formations like Luxembourger sandstone, overlaid by loess, clay, and limestone, are typical for this area (Hollenbach et al., 2014; Schneider, 2019). The geological substrate at the study site is Keuper.

The climate in this region is characterized by precipitation that totals about 750–800 mm per year and a mean annual temperature of between 8.5 and 9°C. The elevation is 270 m a.s.l. Hollenbach et al. (2014) measured pH values of 5.8–6.3 in the uppermost peat layers.

The peatland’s vegetation is mainly composed of Carex acutiformis and Scirpus sylvaticus communities. Additionally, there are high abundances of Schoenoplectus tabernaemontani and the endangered species Triglochin palustre in the central area of the study site. The surrounding area is dominated by a high coverage of Phragmites australis and Juncus inflexus (Hollenbach et al., 2014; Schneider, 2019). The study site is extensively used by horses for grazing (in summer) and has been mowed in alternating sections every year in autumn/winter with a Caterpillar mower since 2016. Detailed species lists and vegetation descriptions can be found in Schneider (2019). The nearby streams are characterized as streams of colline type of the Gutland landscape (Administration du Cadastre et de la Topographie, 2018).

Nature conservation aspects

Because of the very small number and area of peatlands in Luxembourg and their poor habitat condition (European Environment Agency, 2019; MDDI, 2017), it is particularly important to preserve the few remaining peatlands and improve their condition. Within the framework of SICONA’s LIFE-Grassland project, the land adjacent to the previously investigated area was purchased. It is currently undergoing restoration in order to be able to better preserve the study area and to enable its extension. The sealing of drainage ditches to restore water saturation and the removal of topsoil (according to geoelectrical investigations) was carried out in 2019. A transfer of species-rich mowed material is planned for this summer, as well as the reintroduction of endangered species which will take place in autumn. This study area is particularly important for Luxembourg, not only as it is the only such well-preserved climate archive, but also due to its good base cation supply, in contrast to otherwise rather acidic peatlands in Luxembourg, and the occurrence of one of only five populations of the critically endangered species Triglochin palustre in Luxembourg (Schneider, 2019).

Peat characteristics and chronology

The sedimentary deposits of the Ro1A sequence consist of four lithological units (Figure 2). Unit A is characterized by a Pleistocene solifluction soil composed of marl conglomerates. The accumulation of organic sediments starts in unit B, which is characterized by a silty-clayey gyttja with variable levels of embedded sand and organic matter. The initial deposition of peat is observed with the accumulation of a well-decomposed sedge peat in unit C1. Homogeneous alder forest peat with interspersed Cyperaceae rhizomes and wood fragments dominate unit C2. Wood fragments diminish rapidly at the transition to unit C3, which is characterized by sedge peat. Unit D, at the top of the sequence, is a strongly decomposed organic soil with a higher content of fine sand.

OPEN IN VIEWER

Figure 2.

Stratigraphy, stratigraphic zones and chronology of core Ro1A of Rouer peatland.

The developed age-depth model (Figure 2) is based on five AMS radiocarbon dates and shows a constant peat accumulation rate within units B, C1, and C2 with a sample resolution of 20–65 years per centimeter. In units C3 and D, the accumulation rate is significantly higher (9–20 years per centimeter) with peak values at around 1100–1300 cal. yr BP.

Inorganic geochemistry

The results of the XRF-measured Ti/coh and Fe/Ti ratios are shown in Figure 3. As titanium (Ti) is considered to be immobile in peat, it can be used to reflect variations in allochthonous sediment input (Muller et al. 2008; Rothwell and Croudace 2015). To better illustrate effects arising from the sediment matrix, we used the ratio of Ti and the coherent (coh) scatter peaks of Mo (Ohlendorf et al., 2014). Iron (Fe) was normalized to Ti to better reflect the variations in autochthonous peatland dynamics.

OPEN IN VIEWER

Figure 3.

XRF-measured element ratios, total nitrogen/total carbon contents and stable isotope values for core Ro1A of Rouer peatland.

Further measured element variations were strongly influenced by allochthonous input of clastic sediment into the peat matrix, which is a typical feature for minerotrophic peatlands. Most notably in units A and B, before a homogenous peat body started to accumulate, the sediment was significantly characterized by mineral input.

Especially during the Late Pleistocene and Early to Middle Holocene sections of the core, the Ti/coh ratio was at its highest levels, which points to periods of high allochthonous sediment input. Here, low values, which were interpreted as periods of low allochthonous sediment input, were only observed at a depth of 340–330 cm (~11,800–11,500 cal. yr BP) and 285 cm (~8700–8500 cal. yr BP), with some shorter spells at 315–310 cm (~11,200–11,000 cal. yr BP), 265 cm (~7400 cal. yr BP) and 250 cm (~6400 cal. yr BP; Figure 3). After 6000 cal. yr BP, the Ti/coh ratio remained at consistently low levels until 2700 cal. yr BP. Another significant peak occurs only at 215 cm (~4200 cal. yr BP).

Above 170 cm (~2700 cal. yr BP), Ti/coh values showed a high variability at medium levels, consistently remaining higher than those during the previous time span (6000–2700 cal. yr BP). The highest values of the period after 2700 cal. yr BP were observed at around 170–110 cm (~2700–1800 cal. yr BP), 100 cm (~1200 cal. yr BP), 70 cm (~1000 cal. yr BP) and above 25 cm (~300–100 cal. yr BP), which indicates that allochthonous sediment input had increased again in the youngest layers of the peat deposit.

Within the Late Pleistocene and Early to Middle Holocene sections of the core, higher Fe/Ti values corresponded to periods of low allochthonous sediment input, reflected by low Ti/coh values (Figure 6). After 6000 cal. yr BP, Fe has been actively accumulated by the growing peat body. The periodic retention and release of Fe depended on the stability of the water table. Peaks in the Fe/Ti ratio indicate autochthonous enrichment of Fe in the capillary fringe under more oxic conditions.

The mid-Holocene period between 6000 and 2700 cal. yr BP showed the highest variability and the highest levels of Fe/Ti in the record. After 2700 cal. yr BP, high values of Ti/coh and low values of Fe/Ti reflected higher contents of introduced minerogenic matter. Subsequently, higher levels of Fe/Ti ratios were only observed at around 1400 cal. yr BP and at around 900–800 cal. yr BP.

Organic geochemistry

The C/N ratios determined in core Ro1A ranged from 14.93 to 67.1. The maximum range and the average were, however, influenced by measurement of a wood fragment, which was incorporated in the peat at a depth of 145–160 cm (~2700–3100 cal. yr BP; Figure 3). Excluding this measurement, the maximum C/N ratio of Ro1A was 21.17 at a depth of 180 cm (~4300 cal. yr BP) and a mean value of 17.67. In the upper 20 cm (last 200 years) of the Ro1A core the C/N ratios were notably lower than at greater depth. At a depth of 25–60 cm (~200–800 cal. yr BP) the values of the C/N ratio range from 18.39 and 19.66, and are therefore higher than in the upper part of the Ro1A core.

The δ¹³C and δ¹⁵N signatures showed a trend of increasing values with depth (Figure 3). The δ¹³C values of the Ro1A core varied in a range of −28.21‰ and −26.96‰. The maximum value was observed at a depth of 225 cm (~7800 cal. yr BP), the deepest sample measured. The minimum value was observed close to the surface at a depth of 15 cm. The δ¹⁵N signatures showed a similar trend as the δ¹³C signatures, increasing with depth, though less pronounced. The range of the δ¹⁵N values of Ro1A varied between −4‰ and +2.82‰. Noticeably, due to the presence of wood, the minimum value is at a depth of 150 cm (~6500 cal. yr BP) where the C/N ratio is the highest.

Pollen analysis and selected peat macro-remains

The pollen stratigraphy within the Rouer sequence stretches from the late Pleistocene to the middle of the 20th century. Figures 4 and 5 show the results of the pollen percentages (see also Supplemental Figures S1 and S2). All samples of the peat/organic sediment record yielded sufficient pollen for counting, although the preservation of pollen grains was partially poor, especially within the mid-Holocene section (alder forest peat). Overall, the pollen spectrum reflects a tree-dominated vegetation structure with high percentages of Alnus, Pinus, Betula, Corylus, and Quercus. Open-land grasses and herbs reached higher percentages only during the late Pleistocene and within the past 2000 years (Figure 4).

OPEN IN VIEWER

Figure 4.

Tree species pollen percentage diagram of core Ro1A of Rouer peatland.

OPEN IN VIEWER

Figure 5.

Pollen percentage diagram for the past 3000 years showing wetland types (green, excluded from the pollen sum), selected tree species (black) and selected open-land types (gray).

In total, 53 different pollen types were identified. Among the herbs, only Asteraceae and Brassicaceae reached percentages >20%, and only within the past 2000 years. None of the herbs appeared continuously throughout the record. Pollen of Cyperaceae, Lemnaceae, Alnus, Salix, Typha, and Caltha represent local wetland vegetation and were excluded from the pollen total. Wood particles, charred particles, Juncaceae diaspores, and Carex diaspores were counted in the sieved fractions (125–250 µm and 250 µm–2 mm) (Supplemental Figure S2).

Before the onset of the Holocene, the pollen spectrum is characterized by the dominance of Pinus pollen, which, as well as Betula, reached the highest percentages in the record. The high abundance of Salix suggests moderate climate conditions with pioneer trees during the Late Alleröd. Pollen of Poaceae, Asteraceae, Cichoriaceae, Brassicaceae, Apiaceae, and Caryophyllaceae point to a mainly open landscape.

The Preboreal, Boreal and the beginning of the Atlantic still showed high abundances of Pinus, but its dominance steadily decreased until about 7500 cal. yr BP. Betula also decreases in abundance toward the end of the Atlantic. Very low values of Carpinus and Tilia (lower than 1%) occurred during the Boreal. Decreasing open landscape-related pollen types such as Poaceae, Asteraceae, Rubiaceae, Brassicaceae (disappearing in the middle of the zone), Apiaceae and Artemisia suggest a decrease of open landscape vegetation in favor of a boreal forest. Pine trees were declining toward the end of the Boreal, while birch fluctuated. Other trees (Carpinus, Tilia, and Quercus) remained at low values. Wet-tolerant plants like Typha, Lemnaceae, and Caltha showed slightly increasing values. With the beginning of the Atlantic period, Alnus pollen significantly increased in abundance. Furthermore, Tilia showed a remarkable peak in this period, whereas Corylus, Pinus, Betula, and Salix tended to decrease. The first appearance of Fagus pollen (< ca. 2%) and an increase of Poaceae can be seen toward the end of the Atlantic period. Cyperaceae fluctuated at lower, but steady levels. Furthermore, wet-tolerant plants such as Lemnaceae, Caltha, and Typha remained abundant. Herbs generally remained at lower values; only Apiaceae showed increasing values toward the Atlantic/Subboreal transition.

The beginning of the Subboreal is marked by a high dominance of Alnus pollen. In this record, only low contents of Fagus pollen prevailed in this period. Light-loving trees such as Corylus benefited from this by showing relatively high abundances. Betula, Tilia, and Quercus showed relatively constant values; low and fluctuating values with single peaks occurred for Carpinus and Fraxinus. Pinus had decreased during the mid-Subboreal, but had increased again to medium abundances at the transition to the Subatlantic. Cyperaceae and Poaceae still fluctuated at lower values, but showed a peak around 4500 cal. yr BP. Open-land vegetation taxa (Asteraceae, Apiaceae) and those related to arable lands (Chenopodiaceae, Artemisia) and Plantaginaceae showed single high values within this zone.

The end of the Subboreal and the beginning of the Subatlantic stood out with higher abundances of Fagus and Quercus, and a steady decrease of Alnus pollen (Figure 5). Furthermore, Picea, Salix, Carpinus, Fraxinus, and Castanea increased in abundance. Pinus showed fluctuating, mainly low percentages, and have only risen to higher values toward the top of the record within the past 400 years. Juniperus and Larix showed only one occurrence with a percentage lower than 1%. The dominating types during the Subatlantic period were Cyperaceae, Poaceae, Asteraceae, Chenopodiaceae, Brassicaceae (high values up to nearly 10%), Linum and Plantaginaceae, which points to the increasing establishment of an open, deforested landscape during that period.

Geochemical indicators for peat formation and peat decomposition

The Ti/coh ratio clearly identified periods in which the peat was affected by the input of allochthonous minerogenic matter, originating from the catchment area. The input was likely triggered by rainfall events, especially under unstable climatic conditions, and depends on the density and characteristics of the vegetation cover. Hence, a stratigraphic analysis of the peat/organic sediment column allows inferences about the intensity of erosional processes, and therefore, can be linked to paleoclimatic changes. The overall effect depends on the productivity of the peatland ecosystem in relation to climatically and/or human-induced rates of weathering and erosion in the catchment area (Schittek et al., 2012).

Peaks in the Fe/Ti ratio are interpreted as indicators of changes in redox conditions due to water table fluctuations within the peatland (Schittek et al., 2016). High Fe/Ti ratios reveal an upward movement of Fe²⁺ from the anoxic peat to the upper aerated layers, followed by precipitation as Fe³⁺ oxide (Damman et al., 1992). Higher enrichment of Fe relative to Ti in the zone of water table fluctuations under oxic conditions is thus indicative of climatic conditions including episodic droughts, which affect the water saturation of the peatland (Shotyk, 1988).

The C/N ratio is a commonly used indicator for the decomposition of organic matter (Hornibrook et al., 2000; Krüger et al., 2015; Kuhry and Vitt, 1996). Because of higher loss of C relative to N during microbial decomposition, the C/N ratio becomes smaller in strongly decomposed peat.

The C/N ratios of the peat profile of the minerotrophic peatland “Rouer” vary in a range, which is considered typical for minerotrophic peatlands (Andersson et al., 2012), except for high ratios due to the wood fragment in Ro1A at 2700–3100 cal. yr BP (Figure 3). At this depth, exceptionally low δ¹⁵N values also occurred. Besides this, the C/N ratios showed a relatively uniform profile with depth, suggesting that there had been little variation in conditions for decomposition and litter inputs were of comparable quality (Jones et al., 2010). In the uppermost part of the core, the C/N values were notably smaller than at greater depth, indicating stronger decomposition due to past drainage or a higher input of minerogenic matter and nutrients, likely altering predominant vegetation (Gałka et al., 2019; Jones et al., 2010). Lower C/N ratios were also observed at the transition from the Middle Holocene to the Early Holocene, when minerogenic input had also increased. In the remaining part of the profile, C/N ratios did not consistently respond to changes in either Ti/coh or Fe/Ti ratios.

The δ¹³C signature varied in a narrow range around −27‰ and corresponded to δ¹³C values typically reported for peatland vegetation (Broder et al., 2012; O’Leary, 1981). More specifically, the measured δ¹³C signatures fell into the range of minerotrophic peatlands, which are generally somewhat lower than δ¹³C values of ombrotrophic peatlands (Hornibrook et al., 2000; Jones et al., 2010), an effect typically ascribed to a recycling of carbon from ¹³C-depleted methane (Jones et al., 2010; Nichols et al., 2009; Raghoebarsing et al., 2005), and/or differences in vegetation and decomposition (Biester et al., 2014; Gałka et al., 2019), but to a lesser extent to decomposition or diagenesis (Jones et al., 2010).

In general, an increasing trend with depth was observed for the δ¹³C profiles of the analyzed peat cores (Figure 3). The δ¹³C record may on the one hand be an indicator for changes in climate and vegetation (Gałka et al., 2019; Hornibrook et al., 2000; Jedrysek and Skrzypek, 2005). Drier and warmer conditions would lead to more negative δ¹³C values in bogs (Broder et al., 2012; Jedrysek and Skrzypek, 2005), where the C uptake of Sphagnum mosses depends on the thickness of water film on the mosses. The effect is therefore not easily applicable to wetlands without a dominance of Sphagnum mosses (Jones et al., 2010). On the other hand, the increase of δ¹³C values with depth here may be related to changes in decomposition. In most ombrotrophic peatlands, δ¹³C signatures are expected to show a uniform depth profile, or even slightly decreasing values, caused by limited decomposition and the accumulation of slowly decomposing substances depleted in ¹³C (e.g. lignin) (Alewell et al., 2011; Meyers, 1997). However, such residual enrichment of ¹³C-depleted, refractory compounds may not strictly apply to fens. In the presence of vascular plants with aerenchyma, oxygen is transported in the water-saturated peat and could enhance decomposition (Mainiero and Kazda, 2005). For example, this was described for a restored cutover peatland (Agethen et al., 2018), and also in an extreme case, of peatlands in Patagonia with a particularly high root density of cushion-forming plants with aerenchyma (Münchberger et al., 2019). Such enhanced decomposition, likely in the presence of oxygen, could then also lead to a breakdown of refractory compounds and residual enrichment of ¹³C, as experimentally validated for example, by Lehmann et al. (2002) for oxic versus anoxic decomposition, and as typically observed in aerated soils (Ågren et al., 1996; Nadelhoffer and Fry, 1988). Variations in δ¹³C may thus predominantly indicate differences in oxygen availability for decomposition. Indeed, we see congruent patterns of increasing values of δ¹³C around phases of high Fe/Ti ratios, indicating drier conditions (see e.g. 25 cm/330 cal. yr BP, 60 cm/840 cal. yr BP, 100–120 cm/1230–1400 cal. yr BP). It has to be noted, though, that patterns are in part shifted (see e.g. 130–150 cm/1530–1850 cal. yr BP) and the signals of anoxic versus oxic decomposition (Lehmann et al., 2002) are apparently obscured by other factors such as vegetation (Galka et al., 2019; Jones et al., 2010).

Signatures of δ¹⁵N in fens are not straightforward to interpret because N can enter fens through several sources in different organic and inorganic forms. Moreover, the processes in the N cycle leading to fractionation are highly complex (Jones et al., 2010). The δ¹⁵N values in bogs typically scatter around zero ‰ (Broder et al., 2012; Jones et al., 2010). This is also observed in the minerotrophic peat profile of the Rouer peatland. Yet, the N isotope signatures vary in a relatively wide range of −4‰ to 3.22‰ (Figure 2). Similar ranges were observed in bogs by Broder et al. (2012) and in forest soils of the temperate zone by Novák et al. (2003), while limnic sediments may be more enriched in ¹⁵N (Lehmann et al., 2002). The variability of the values should at least partly be related to the high variability of the δ¹⁵N values of different plant species (Krüger et al., 2015), but Liu et al. (2020) also reported that the initial ¹⁵N may be diagenetically altered, limiting its interpretation. There was a slight trend of increasing values with depth observable in the Rouer peat profile. Higher δ¹⁵N values may indicate stronger decomposed organic matter, as repeated cycling and mineralization of N leads to a preferential loss of ¹⁴N over ¹⁵N (Dijkstra et al., 2008; Nadelhoffer and Fry, 1988). Such enrichment of ¹⁵N would therefore be expected for decomposition under oxic conditions, while under anoxic conditions, a ¹⁵N depletion would be expected (Lehmann et al., 2002). Krüger et al. (2015) suggest that like δ¹³C values, δ¹⁵N values increase with depth in degraded peatlands due to higher microbial activity under aerobic conditions; conditions that might also apply to fens with intense decomposition in the presence of vascular vegetation with aerenchyma. This suggests that the organic matter is increasingly decomposed with depth, presumably with the availability of oxygen, confirming our interpretation of observed patterns in ¹³C (Krüger et al., 2015; Nadelhoffer and Fry, 1988). Smaller variations in the profile were indeed not related to other indicators of drought or wetness, thus the exact interpretation of distinct fluctuations may be limited due to diagenetic processes (Liu et al., 2020).

Late Pleistocene and Holocene environmental changes

Figure 6 summarizes the Late Pleistocene/Holocene environmental history of the Rouer fen in comparison with the paleoclimatic evidence from Greenland ice cores (Svensson et al., 2008; Vinther et al., 2009) and the sum of Holocene glacier advances (Wanner et al., 2011). The accumulation of organic-rich sediment, which is not solifluidally reworked, can be identified in the deposits of the Rouer fen at about 13,800 cal. yr BP during the Bølling interstadial. The Late Pleistocene was a period of dramatic climatic changes, which is also reflected in the oldest sediment layers of Rouer fen profile. During the Bølling period, solifluction and runoff within the Pall and Näerdenerbaach valley floors diminished, so that organically-enriched floodplain sediments started to accumulate at Rouer fen, interspersed by less organic-rich debris, as reflected by the oscillating Ti/coh values. During the Bølling and Alleröd interstadials (between 13,800 and 12,500 cal. yr BP), several short phases of significantly low Ti/coh values were identified, which points to a denser vegetation cover. According to Guiot and Couteaux (1992), the climate became increasingly warmer after 13,000 cal. yr BP and moisture was nearly as high as the present day. Higher abundances of grasses at this time indicate a landscape with an open character (Slotboom and van Mourik, 2015). At Rouer fen, pollen spectra show high scores of Pinus, Betula, and herbal/grass pollen. Apart from the cold-tolerant tree species (Pinus, Betula, Salix), deciduous arboreal pollen taxa are still very rare due to the long lag time for their arrival from their glacial refugia (Litt and Stebich, 1999). Pinus is the dominant pollen species, which shows that Pinus trees successfully colonized the open bare ground, where soils had yet to develop.

OPEN IN VIEWER

Figure 6.

The late Pleistocene/Holocene Ti/coh, Fi/Ti, TC/TN ratio, and grasses/herbs pollen sequences of Rouer peatland compared with NGRIP oxygen isotope data (Svensson et al., 2008), Northern Hemisphere temperature reconstruction data (Vinther et al., 2009) and the sum of glacier advances (Wanner et al., 2011) in comparison to the archaeological chronology (Paulke, 2019).

At around 12,700 cal. yr BP, this period of warmer climate ended due to a sudden southward shift of the oceanic polar front, returning to a position similar to that of the Last Glacial Maximum (Anderson et al., 2007). During this 1200 year-long cold period, the Younger Dryas, the climate was markedly drier due to a renewed strengthening of the continental high-pressure cells (Sirocko et al., 2016). The Younger Dryas cold event is clearly depicted in several paleoenvironmental records of the Eifel/Ardennes area (e.g. Brauer et al., 2008; Guiot and Couteaux, 1992; Sirocko, 2010; Sirocko et al., 2016). The mean July temperature depression of 7–8°C had a strong impact on the environment and led to a return of vegetation characteristic for subarctic conditions (Atkinson et al., 1987; Brauer et al., 2008; Lowe et al., 1994). Due to the harsher conditions, trees should have disappeared from the higher lying areas with harsher climate conditions and might have survived only in more protected valley floors. The pollen spectra of Rouer fen correspondingly show a decrease in Pinus, which was also evidenced by Guiot and Couteaux (1992) in the east of Luxembourg. The timing of the fluctuations in the Ti/coh values match the range and the oscillations of the Younger Dryas, as recorded by the Greenland ice δ¹⁸O isotope values. A drier and colder climate presumably led to an increased allochthonous sediment input into the fen area, which might have been a result of an overall reduced vegetation cover within the catchment area of Pall and Näerdenerbaach creeks. The end of the Younger Dryas is marked by a sudden temperature increase of about 8°C within only 5 years, at around 11,500 cal. yr BP, according to temperature reconstructions based on isotope data (Alley et al., 1993; Sirocko, 2010). The Rouer fen records this event apparently earlier at around 11,800 cal. yr BP, which might be due to insufficient age control in this section of the core. However, the variation pattern of the Ti/coh compares well to the Greenland isotope values.

The onset of the Holocene is marked by a rapid shift of temperature, right at the end of the Younger Dryas. According to Anderson et al. (2007), the mean annual temperature increased by 7°C within, at most, 50 years. This quick shift was attributed to changes in the North Atlantic drift system and an increase in sea surface temperatures, when the Gulf stream warmed abruptly (Alley et al., 1993; Knorr and Lohmann, 2007). The Rouer fen record documents this shift as a sudden decrease of Ti/coh values between 11,900 and 11,500 cal. yr BP. This period of low Ti/coh values correlates well with a contemporaneous increase in δ¹⁸O values of Greenland ice cores and can be assigned to the Friesland phase (Bos et al., 2005), which marks the beginning of the Holocene warming.

The Early Holocene experienced a series of important environmental changes. Steppe species declined and land became available for the expansion of an open woodland composed of Pinus, Betula and a little Corylus (Guiot and Couteaux, 1992). Following our Rouer fen pollen record, Alnus also became more abundant and might have colonized wet depressions and river valleys. Compared to the Late Pleistocene, especially during the warm Friesland phase, there was obviously a reduced fluvial discharge through rivers and brooks. River channels in the Rhine-Meuse area started to fill with organic sediment and peat, which significantly slowed down water run-off and resulted in a process of aggradation (Crombé, 2018).

After the warm beginning of the Holocene, a cold episode occurred from ca. 11,500–11,300 cal. yr BP, which led to a sudden increase of Ti/coh values again. This episode correlates with the Preboreal oscillation, which is characterized by a mean temperature decline of about 1–2°C for the subsequent 100–150 years (Björck et al., 1997; van der Plicht et al., 2004). Magny et al. (2007) presume wetter conditions in west-central Europe in response to the Preboreal oscillation, as inferred from higher lake water levels reconstructed for several European lakes between 11,300–11,150 cal. yr BP.

After this first prolonged Holocene cooling episode, reduced Ti/coh values indicate a return to a state with less input of allochthonous sediment at Rouer fen. This coincides with a warm episode, following the cold Preboreal oscillation (Fiłoc et al., 2016; Magny et al., 2007). However, this short warming was followed by the next abrupt climate cooling starting from ca. 11,100 cal. yr BP and lasting until 10,800 cal. yr BP (Bohncke and Hoek, 2007). At this point, a hiatus appears in the Rouer fen record. Crombé (2018) reported that a sedimentary hiatus in peat growth was observed in several, mostly smaller and medium-sized valleys of Rhine-Meuse area during that period. The main cause for this hiatus is likely climate oscillations during the Preboreal and a resulting instability of the environment with high amplitudes in fluvial discharge (Fiłoc et al., 2016). Paleoenvironmental information from the Rouer fen record is evident again at ca. 9600 cal. yr BP, when the pollen record shows noticeably high percentages of Pinus.

The abrupt isotope shifts in Greenland ice cores at around 9300 cal. yr BP and 8200 cal. yr BP represent the most sudden climatic shifts during the Holocene (Figure 6; Alley et al., 1993; Rasmussen et al., 2007). These are likely driven by catastrophic meltwater releases into the North Atlantic Ocean, which disrupted the thermohaline circulation and its attendant northward transport of warm water and air masses (Törnqvist et al., 2004; Yu et al., 2010). Sudden increases in Ti/coh values indicate a strong impact on allochthonous sedimentation in the Rouer record, associated with these sudden cooling events. The climate deteriorations at 9300 and 8200 cal. yr BP are part of a repeating pattern of longer-term anomalies during the Holocene, seemingly related to fluctuation in solar output. This pattern manifested as “Bond events” later on again at 5900, 4200, 2800, 1400 cal. yr BP, and most recently, during the Little Ice Age (Rohling and Pälike, 2005).

Lake sediment studies in eastern France also point to cooler temperatures and higher precipitation during the Bond events, due to a strengthened Atlantic westerly jet stream that resulted in more storm tracks across mid-latitude Europe (Anderson et al., 2007). The 8200 cal. yr BP cold anomaly affected all of central Europe (von Grafenstein et al., 1998) and lasted for about 120 years, according to the precisely-dated evidence from Lake Holzmaar in the nearby Eifel Mountains (Prasad and Baier, 2014). The pollen evidence from the Eifel maar lakes further shows that the Early Holocene forest structure in that area, mainly composed of Pinus and Corylus, did not recover after the 8200 cal. yr BP event. Instead, the development of a mixed oak forest started (Sirocko et al., 2016). The Rouer fen record shows a continuous increase of Alnus and Tilia after this event, as these species might have dominated the wetter valley grounds in the local catchment areas and have spread due to the warming after the 8200 cal. yr BP event.

However, environmental conditions remained relatively unstable until about 6000 cal. yr BP, as the Rouer fen record remained enriched with allochthonous titanium. Yet between 7000 and 6000 cal. yr BP, sediment input decreases steadily and the TOC content rises to levels above 30%. Dabkowski et al. (2015) suggest relative humidity stability and maximal forest development for the early Middle Holocene, inferred from mollusc fauna at Direndall tufa. The reconstruction of drier conditions between around 7100 and 6500 cal. yr BP by Granai et al. (2020), based on further malacological studies from the same site, is consistent with a general decrease of Ti/coh values, generally less negative values of δ¹³C, and lower values of C/N ratios. Guiot and Couteaux (1992), based on pollen data from the Echternach area, suggest warmer and wetter summers during the Atlantic period until 6000 cal. yr BP and a climatic change toward drier summers after 6000 cal. yr BP. At Rouer fen, by 6000 cal. yr BP, the alder forest had apparently fully developed and alder dominates the pollen spectrum for the next 3000 years. Pinus decreases, and deciduous trees, including Fagus, Carpinus, and Fraxinus, were fully established. From 6000 cal. yr BP onwards, temperatures in all seasons in central Europe had evened out to values within ± 1 K from modern values (Davis et al., 2003). The prevailing climate after 6000 cal. yr BP is characterized by short-term, periodic dry events, as reflected by the frequency of shifts in the Fe/Ti ratio, still low C/N ratios, but values of δ¹³C turning more negative. Sedimentation at Rouer fen had shifted from clastic-detrital deposition to an autochthonous sedimentation in the form of peat.

Climate remains relatively stable until about 4200 cal. yr BP. Afterwards, the frequency of Fe/Ti shifts reduces. Due to the evidence of tree line retreat in the Alps, a transition to a climate cooling with an increase in precipitation can be determined for the period after about 5000 cal. yr BP (Wick and Tinner, 1997). This is related to an intensification of the Gulf Stream, which at least from about 4000 cal. yr BP, also had an impact on increased cyclonic activity and a more meridional circulation pattern across Europe (Frenzel, 1998).

The Rouer fen record shows a marked peak of Ti/coh at 4200 cal. yr BP, which evidences a prominent climatic anomaly (e.g. Weiss, 2016). Dörfler et al. (1998) and Löhr and Neyses-Eiden (2001) mention an extreme weather event at the same time, which had led to the embedding of numerous tree trunks in abandoned meanders of Mosel and Kyll rivers in the nearby Eifel Mountains. A disastrous flood event was also evidenced in the sediments of Ulmener Maar Lake (Sirocko, 2010).

Between 4200 and 2800 cal. yr BP, during the Bronze Age, no further climate anomalies occurred. Conditions seem to have been rather stable, which is also reflected by more stable values of C/N and δ¹³C. Increasing percentages of grasses and herbs in the Rouer pollen spectrum indicate that Bronze Age people had extended the clearance of woodlands. The decrease of Tilia during that period might be related to the fact that the sites where Tilia grew often turned out to be suitable as farmland, and therefore, forest was preferentially cleared at these sites (Behre, 2007). In the Rouer pollen record, Tilia continuously decreases from 5500 cal. yr BP on, and especially between 3500 and 2900 cal. yr BP, abundance reduces to percentages below 1%. Alnus pollen levels, starting at 3800 cal. yr BP, rise to the highest in the record, reaching nearly 70% at around 3100 cal. yr BP.

Several paleoclimate records suggest that temperatures cooled further at the Subboreal-Subatlantic transition at around 2800 cal. yr BP (van Geel et al., 1996; Wanner et al., 2008; Zolitschka et al., 2003). Pare et al. (2010) remarked that the change toward a more cool-humid climate at 2800–2700 cal. yr BP led to a subsistence crisis and provoked migration in several regions during the early Hallstatt culture. The Rouer pollen record shows a contemporaneous decrease of grasses and herbs, and a strong increase of Fagus, Carpinus, Quercus, and Corylus, which indicates that open land was abandoned and forest recovered in the surroundings of Rouer fen. By the 2nd century BC, during the late Iron Age, settlement conditions in central Europe had stabilized again. City-like, fortified hilltop settlements of large dimensions were created all over the area of the Celtic culture.

During that period of cultural bloom, the transformation of the alder forest into an open sedge reed can be demonstrated. The first signs of clearing can be inferred from the appearance of larger numbers of charred particles at around 2200–2100 cal. yr BP. The appearance of Juncaceae seeds and Carex seeds indicates a more open vegetation. The presence of oospores from Characeae point to the existence of sun-exposed pools within the fen. Dabkowski et al. (2015) link the environmental change during the episode of Roman occupation to anthropogenic forest clearing.

The Roman period is known for a significant impact on vegetation through deforestation and intensification of agriculture. In the pollen diagram (Figure 5), clearing of the forest is shown by a reduction in tree pollen percentage of Alnus in particular, and an increase of Cyperaceae and Poaceae. Fagus also declines sharply, but, like Quercus, recovers by around 1800 cal. yr BP, when Cyperaceae and Poaceae again decrease in abundance.

The cold spell of the Migration Period, after the collapse of the Roman Empire, led to a decrease in population density, especially in rural areas (Kausch, 2006). Tree pollen, mainly Fagus and Alnus, show increasing values. Numerous types of herbs and shrubs such as Asteraceae, Cichoriaceae, and Brassicaceae show increased values, which indicates an intensive spread of these species on the former arable land.

The increased solar activity during the Middle Ages led to a positive North Atlantic Oscillation, which supplied mild and humid airmasses to Middle Europe (Sirocko, 2010). The recurred warming, known as the Medieval Climate Optimum, resulted in the highest population density in Middle Europe up to that point in time (Grupe, 1986). The population growth provoked a notable extension of the cultivated landscape (Bork et al., 1998). The Rouer peat archive witnessed these changes, containing the highest Poaceae pollen percentages during that period in the whole record, and the lowest overall tree species abundances. Alnus nearly disappears from the record, presumably due to the transformation of the remaining alder forest stands into wet grasslands. The warm period is further expressed in the occurrence of charred particles, which show by far the highest accumulation rates between 800 and 1400 AD. This could be a possible sign of slash-and-burn agriculture.

The spreading of Quercus between 1500 and 1700 cal. AD may be a result of its use for housebuilding, coppicing practice, and for tanneries. Furthermore, oak stands were widely used for forest pasturing practices during the Middle Ages and into the Modern Era, especially for feeding pigs acorns (Sudhaus and Friedmann, 2015). The increase in Brassicaceae is indicative of the expansion of arable land since the Middle Ages.

The Little Ice Age marks the most recent climatic shift to cold and humid climates: winter temperatures were lower and there was a sharp increase in precipitation, which led to severe soil erosion (Slotboom and van Mourik, 2015). Increased Ti/coh values is evidence of enhanced sediment input into the Rouer peat layers. In these upper peat layers, presumably stronger decomposition obscured concurrent responses in δ¹³C or C/N. After 1800 cal. AD, large areas of heathland and grassland in the Gutland were reforested, mainly with Pinus and Picea. Sylviculture, grassland farming, and agriculture characterize the youngest Modern Era.