Regional setting of the study areas
Geographical and historical setting
The two study areas, the Reinhardswald and the Kellerwald-Edersee National Park, are about 60 km apart from one another. They are located in the northern part of Hesse, a German federal state characterized by low mountain landscapes and a high proportion (42 %) of woodland cover (Fig. 1). The climate can be characterized as subatlantic with low subcontinental influence. In both study areas, natural woodlands would be dominated by European beech (Fagus sylvatica), with some exceptions at moist, wet or extremely dry sites and steep, rocky slopes, which are characterized by black alder (Alnus glutinosa) and birches (Betula spp.) or oaks (Quercus robur and Q. petraea), respectively (BfN 2010).
The Reinhardswald is bordered by the rivers Fulda, Weser and Diemel and represents the northernmost connected forest area of Hesse. The study area is 20,600 ha in size. The underlying bedrock is predominantly sandstone from the Buntsandstein, a lithostratigraphic unit of the Lower Triassic series (Rapp 2002).
The Kellerwald-Edersee National Park is located south of the river Eder, whose deeply incised valley has been flooded by a 27-km long artificial lake (Edersee) since 1914, when the Eder was dammed. The national park is 5700 ha in size and was established in 2004 with the aim of protecting large, unfragmented areas of semi-natural acidophilous beech forests and other deciduous forest types. The underlying bedrock is mainly composed of Carboniferous clay shale and greywacke. Since 2011, about 1500 ha of the Kellerwald-Edersee National Park are part of the UNESCO World Heritage Site “Primeval Beech Forests of the Carpathians and the Ancient Beech Forests of Germany” (Succow et al. 2012; Vološčuk 2014; Menzler and Sawitzky 2015).
According to forest site survey data, the site conditions are dominated by acidic soils with mesotrophic nutrient status in both study areas. The western parts of the Reinhardswald are characterized by flat or gently sloping plateaus, where seasonally moist soils predominate. While fresh (moderately fresh, fresh or markedly fresh) soils are largely missing on these flatter areas, they are much more abundant on the slopes of the eastern parts of the Reinhardswald (Bonnemann 1984; Rapp 2002). In the Kellerwald-Edersee National Park, moderately dry, moderately fresh and fresh sites prevail and almost no seasonally moist or moist soils occur (Menzler and Sawitzky 2015).
In high medieval times (ca. AD 1100 to 1300), large parts of both study areas were not covered by forest, but used as agricultural land. Relicts of ancient fields and lynchets (i.e., banks of earth that build up on the downslope of a field ploughed over a long period of time), which are visible in the field as well as in DTMs (Fig. 2), provide strong evidence for the former agricultural land use. From the early 14th century onwards, the associated settlements and their fields were abandoned due to a complex of causes, particularly climate deterioration and epidemic plagues. Forest spread out quite quickly, and already in the mid-15th century dense woodland had re-covered the landscape (Höhle 1929; Jäger 1951; Jäger 1958; Born 1961; Bonnemann 1984; Stephan 2010). As relatively accurate topographic maps, the scale is at least 1:100,000, from the 18th century (Reinhardswald: Leopold 1719; Rüstmeister 1724) and early 19th century (Kellerwald-Edersee National Park: von Le Coq 1805) show, both study areas were almost completely covered with deciduous forest stands, interspersed with small meadows and heathlands. At this time, the portion of coniferous forests was less than 1 % (Mackeldey 1971; Zarges 1999).
Today, the forest vegetation is dominated by beech, with shares of about 40 % in the Reinhardswald and 66 % in the Kellerwald-Edersee National Park. The portion of oaks is 17 % in the Reinhardswald and 7 % in the Kellerwald-Edersee National Park. Coniferous tree species, mainly spruce (Picea abies), reach total covers of 40 % in the Reinhardswald and 20 % in the Kellerwald-Edersee National Park.
Both forest areas are nowadays owned by the federal state of Hesse. The whole Reinhardswald area belonged for centuries to the Landgraviate of Hesse-Kassel. Ownership transferred in 1803 to the Electorate of Hesse, which was merged into the Prussian province of Hesse-Nassau in 1866 (Bonnemann 1984). In contrast, the area of the current Kellerwald-Edersee National Park was divided between three different territories for a long time (Fig. 3). The larger eastern part was ruled by the County of Waldeck (since 1712 Principality of Waldeck) and was not merged into the Prussian province of Hesse-Nassau until 1929. The smaller western part (Itter estate) belonged to the Landgraviate of Hesse-Marburg from 1589, and was later assigned to the Landgraviate of Hesse-Darmstadt (since 1806 Grand Duchy of Hesse). This part was also merged into the Prussian province of Hesse-Nassau in 1866. A small area at the southern edge of the Kellerwald-Edersee National Park belonged, like the Reinhardswald, to the Landgraviate of Hesse-Kassel (Curtze 1850; Demandt 1972; Waldeyer 2014).
Pre- and early industrial charcoal burning and customers in the study areas
From the middle of the 16th century onwards, smelting works and hammer mills were established in the vicinity of the studied forest areas. Prior to this period, iron and iron products were made in small, decentralized smelting furnaces and smithies (Wick 1910; Sippel 2005). From the mid-17th century, the charcoal makers used temporary charcoal kilns on-site in the forest coupes. These charcoal kilns consisted of a carefully stacked wood pile which was covered by a gas-tight layer of soil, turf and moss (Klein 1836; von Berg 1860). Previously, the production of charcoal took place in charcoal pits (Schäfer 1977; Wick 1910). For these reasons, we conclude that the majority of the charcoal kiln sites in both study areas were established between the mid-17th century and the end of the 19th century. In the following section, we summarize the main developments with regard to charcoal and iron production in the study areas.
Reinhardswald
The first written evidence of charcoal making dates from 1302 (Bonnemann 1984). A number of later archival documents indicate that secondary woodland, which had developed after the abandonment of farmland in medieval times, was cleared for agriculture in the 16th century. The harvested wood was used for glass and charcoal production (Jäger 1951). Forest decrees issued by the Landgraviate of Hesse-Kassel in 1593, 1629 and 1683 highlight that there was a huge amount of lying wood (originating from windthrow, for example) which should be used for charcoal production. But also regularly harvested wood was otherwise carbonized. Trees that were suitable for timber production or technical purposes must not be allocated to the charcoal makers (Landgraf zu Hessen 1593; Landgraf zu Hessen 1629; Landgraf zu Hessen 1683).
The main customers of charcoal were the initially privately operated hammer mills and ironworks at Lippoldsberg (1555–1873, since 1583 state-owned) and Heisebeck (1555–ca. 1564), and the state-owned ironworks at Vaake (1581–1583), Knickhagen (1591–1666), and Veckerhagen (1666–1903). From about 1700 to 1730, a copperworks existed at the Olbetal near Veckerhagen (Fig. 4). Following a forest description from 1774, an annual wood volume of ca. 1200 m3 could be sustainably harvested in the ca. 120-ha forest compartment “Stickelhalbe” to provide charcoal for the hammer mill at Lippoldsberg (Henne 1997). By 1767, the Veckerhagen ironworks needed ca. 4800 m3 and by 1802 ca. 5500 m3 of wood per year (Cancrinus 1767; Laurop 1802). The ironworks purchased the wood from neighboring state forests and coordinated the charcoal makers. The iron ore was mostly obtained from the Hohenkirchen mining district, which is situated 18 km southwest of Veckerhagen (Fig. 1). From the early 19th century onwards, the importance of charcoal for iron smelting decreased, since coal-derived coke was increasingly available (Cancrinus 1767; Laurop 1802; Wick 1910; Lotze 1997).
Kellerwald-Edersee National Park
The importance of charcoal making in the area of the current Kellerwald-Edersee National Park is highlighted by a forest decree issued by the Prince of Waldeck in 1741 (Fürst von Waldeck-Pyrmont 1741). This decree emphasizes the increasing demand from mines, smelting works and hammer mills for charcoal, fuelwood, and construction timber, and provides detailed instructions for protecting the forest resources from depletion. In particular, the forest administration was instructed to supervise the cutting of wood for charcoal making very strictly. Trees that were suitable for timber production or technical purposes had to be excluded from charcoal production. Besides charcoal, also wood ash was made in the area, the relics of ash production sites can be found in the forest (Sippel 2009).
In Waldeck, the early industrial iron production had its heyday between the 16th and 18th century. During this time, a multitude of mostly small ironworks and hammer mills (e.g. Vornhagen 1540–1710, Kleinern 1657–1870, Gellershausen 1658–1664) were in operation in the vicinity of the current Kellerwald-Edersee National Park (Fig. 3) (Mannel 1908; Schäfer 1977). The most important complex of ironworks and associated hammer mills was located in the valleys of the rivers Eder and Werbe, ca. 2 km north of the Kellerwald-Edersee National Park. The first written record of this proto-industrial area is from 1623. The central ironworks was the Bericher Faktorei (smelting works and associated hammer mills), which had its peak period in the early 18th century. The iron ore was purchased from the Adorf mining district, which is situated 35 km northwest of the ironworks (Fig. 1). Charcoal and fuelwood were mostly obtained from the adjacent state-owned forests. Between the 17th and 19th century the charcoal and fuel wood supply of the Bericher Faktorei had been strictly regulated by the governmental forest administration. Various measures were aimed at preventing both forest over-exploitation and the wasting of wood. The price of the wood was regulated, and wood delivery to the ironworks and their charcoal makers was made dependent on the available felling volume. There were certain situations in which the ironwork in Berich had to cease its operations due to fuel shortage, or was compelled to purchase charcoal from other regions (Curtze 1850; Mannel 1908; Schäfer 1977). Between 1750 and 1806, the annual wood consumption of the Bericher Faktorei amounted to a mean of 7500 m3. This factory had faded in significance by 1833 and subsequently closed down in 1875 (Mannel 1908).
In the territory of the former Landgraviate of Hesse-Kassel next to the village of Frankenau, 2 km south of the Kellerwald-Edersee National Park, an ironworks operated from 1576 to 1663. Charcoal and fuelwood were, however, presumably obtained from the Itter estate (Boucsein 2009). The ironworks and hammer mills of the Haina Hospital, ca. 12 km southeast of the Kellerwald-Edersee National Park, had been operating between the early 16th and the 19th century and were able to gain the charcoal and fuelwood supply from the surrounding woods (Wick 1910; Friedrich 1990; Zarges 1999; Boucsein 2009). In the village of Thalitter in the Itter estate, 8 km northwest of the Kellerwald-Edersee National Park, a copperworks was in operation from 1712 to 1868. This factory also consumed large quantities of charcoal and fuelwood (Cancrinus 1767; Tasche 1849; Paul 1939). In those parts of the Kellerwald-Edersee National Park which belonged to the Itter estate, charcoal making had its peak in the 19th century (Zarges 1999).
Charcoal kiln site distribution
Digital terrain model (DTM) and landscape attributes
For the systematic mapping of charcoal kiln sites, we used a digital terrain model (DTM), which was derived from state-wide airborne laser scanning (ALS) data by the Hessian State Office of Land Management and Geoinformation (HVBG). The ALS data were recorded between 2009 and 2012 and the applied DTM provided a spatial resolution of 1 m2 (Doneus et al. 2008; Gertloff 2011; Risbøl et al. 2013; HVBG 2015). Two different hillshade images were produced from the DTM, which varied with regard to the lighting conditions. This enabled us to visually identify charcoal kiln sites with great accuracy (Fig. 2).
For each kiln site data point, we determined the parameter values of four landscape attributes from digital forest inventory maps (“tree species composition”), forest site survey maps (“water supply status”, “nutrient supply status”), and soil maps (“soil complex classes”). Soil complex classes have been mapped at a scale of 1:50,000 (HNUG 2002), the scale of forest inventory maps was 1:50,000 and that of the site survey maps was 1:25,000 (HMULF 2002). The three landscape attributes “altitude”, “exposition” and “inclination” were derived from the DTM. Since the foundations of charcoal kilns are always small plateaus, the parameter values of the attributes “exposition” and “inclination” were calculated as mean values of 15-m-wide rings around 15-m-wide buffer circles at the kiln site data points. The landscape attributes mentioned above were also determined for the total areas of each studied forest landscape. All geospatial analyses were performed using GRASS GIS 7 (GRASS Development Team 2015) and QGIS 2.11 software (QGIS Development Team 2015).
Comparison of DTM-based surveys and field inventories of charcoal kiln sites
Georeferenced data from field inventories of charcoal kiln sites in the northern Reinhardswald (Koch 1990; Stephan 2010) enabled us to compare DTM-derived information on charcoal kiln sites with data collected in the traditional way and without the use of GPS (Ludemann 2012). Within an area of 847 ha, we identified 414 charcoal kiln sites by interpreting the DTM, while merely 328 of these (79 %) were located during the field surveys. Apart from the greater effort of field inventories, the positional accuracy of the DTM-derived point data was much higher. Divergences of more than 50 m were not unusual between the field survey points and the actual kiln site locations.
Statistical analysis
All statistical analyses were performed by using the R software version 3.2.2 (R Development Core Team 2015) with the “vegan” package (Oksanen et al. 2012) and the “cluster” package (Maechler et al. 2015). Significance of statistical tests was noted as follows: *** = p ≤ 0.001; ** = p ≤ 0.01; * = p ≤ 0.05; n.s. = p > 0.05.
Univariate analysis
Using a chi-square test, we checked whether the frequency distributions of charcoal kiln sites with respect to the levels of the various landscape attributes were proportional to the distribution of these landscape attributes in the total study areas. We applied the likelihood ratio test (cf. Gould et al. 2006):
$$ {G}^2=2{\displaystyle {\sum}_1^c}{n}_i \log\ \left({n}_i/\widehat{\mu_l}\right)\kern2em \mathrm{with}\kern0.5em i = 1(1)\mathrm{c} $$
(1)
G
2 is χ
2-distributed with c-1 degrees of of freedom, where n is the total number of charcoal kiln sites, n
i
the observed charcoal kiln sites in the i-th factor level, and \( \widehat{\upmu_l}={p}_i\times n \) the theoretical values of the total area in this level.
Multivariate and cluster analysis
In order to better specify the multivariate structure in the data of the charcoal kiln sites, we conducted a principal component analysis (PCA; cf. Venables and Ripley 2002). For this purpose, the nominal variables “water supply status”, “nutrient supply status”, “soil complex classes”, “exposition”, and “main tree species” were transformed into dichotomous variables with regard to the respective factor levels. Only dichotomous variables were used for further analysis. In order to fit the dichotomous variables of both studied forest areas onto the PCA plot, we used the function “envfit” provided by the “vegan” package in R (Oksanen et al. 2012). A biplot was created, which allowed for the analysis of the correlation between the different variables and both forest areas.