Historical land-use and landscape change in southern Sweden and implications for present and future biodiversity

The two major aims of this study are (1) To test the performance of the Landscape Reconstruction Algorithm (LRA) to quantify past landscape changes using historical maps and related written sources, and (2) to use the LRA and map reconstructions for a better understanding of the origin of landscape diversity and the recent loss of species diversity. Southern Sweden, hemiboreal vegetation zone. The LRA was applied on pollen records from three small bogs for four time windows between AD 1700 and 2010. The LRA estimates of % cover for woodland/forest, grassland, wetland, and cultivated land were compared with those extracted from historical maps within 3-km radius around each bog. Map-extracted land-use categories and pollen-based LRA estimates (in % cover) of the same land-use categories show a reasonable agreement in several cases; when they do not agree, the assumptions used in the data (maps)-model (LRA) comparison are a better explanation of the discrepancies between the two than possible biases of the LRA modeling approach. Both the LRA reconstructions and the historical maps reveal between-site differences in landscape characteristics through time, but they demonstrate comparable, profound transformations of the regional and local landscapes over time and space due to the agrarian reforms in southern Sweden during the 18th and 19th centuries. The LRA was found to be the most reasonable approach so far to reconstruct quantitatively past landscape changes from fossil pollen data. The existing landscape diversity in the region at the beginning of the 18th century had its origin in the long-term regional and local vegetation and land-use history over millennia. Agrarian reforms since the 18th century resulted in a dramatic loss of landscape diversity and evenness in both time and space over the last two centuries leading to a similarly dramatic loss of species (e.g., beetles).

SWEREF 99. The reform maps include detailed information on land-use/vegetation units of generally high technical and geometric quality (Örback, 1998), which allows rectification with a relatively low degree of error. be separated from hay meadows and other unspecified open land areas. Therefore, all classes were grouped into a single class "Grassland" (LuV2), however carefully taking into account the written information related to the maps. Some of the meadow classes might have been half-wooded land as wooded hay meadows were a common feature of the traditional cultural landscape of southern Sweden (Emanuelsson, 2007;Berglund, 1991;, but the meadow class was more often specified as "meadow without tree". Therefore, "meadows" were classified as "Grassland". We combined bogs and fens with other wetland classes such as mowed meadow on wet soils and unspecified wetlands into a single "Wetland" class (LuV3). As deciduous woodland was not always separated from conifers, all classes with tree cover were grouped into a "Woodland/Forest" class (LuV4). It includes managed forests/tree plantations as well as non-managed woodland as these are not separated in the maps. It also includes the class "Outland" documented in some maps, representing woodland areas that were generally part of the commons of villages, outside the "infields" with cultivated fields and hay meadows (Berglund 1991). In these maps, "Outland" that was not wooded is sometimes illustrated by some generalized tree symbols scattered over the area and specified in related notes and tables as "Outland without trees". Those "Outland" areas where classified as "Grassland".

Assignment of pollen types to the maps' LuV classes
The Landscape Reconstruction Algorithm (LRA, see end of Appendix under "LRA reconstructions") reconstruction requires a number of known parameters of which pollen productivity estimates (PPEs) are particularly difficult to obtain. In this paper we use 25 pollen taxa for which PPEs are available (Mazier et al., 2012;Cui et al., 2013) (Table 3). As the LRA estimates of Land-Use/Vegetation classes (LuVs) (LRA-LuVs) will be compared with LuV data extracted from historical maps (map-LuVs), and because the LuVs are characterised by groups of plant taxa, the 25 pollen taxa used for the LRA reconstructions have to be assigned to the LuVs.
In this study, Calluna (heather) is the most problematic case because it exhibits high pollen counts and percentages (20-30%) at both Stavsåkra (TW3-4) and Storasjö (TW1-4) (see Fig. 3), implying high LRA estimates of its cover in the vegetation ( that between-site variation in abundance is very high in both grassland and woodland. We assume that Calluna was also growing in the three major vegetation types (grassland/heath, wetland, and woodland) in the past and exhibited a similar between-site variation in abundance.
Therefore, we chose to examine three different scenarios of assignment of pollen taxa to the   Sugita, 2005;Mazier et al., in press;in Fredh, 2012). In this study we tested three methods (test described below) and chose to use the map-DWLuVs calculated with the LuV-fall speeds obtained with the method 2, as described below.

Fall speed of pollen for LuVs (LuV-fall speed)
Values of taxon-specific fall speed of pollen are available for the 25 taxa included in the LRA reconstruction (Table 3). Fall speed can be either measured or calculated with the Stoke's law (Gregory, 1973) using size measurements of the pollen grains from the actual plant taxa (e.g. Mazier et al., 2012). As we need fall speed values for LuVs, we have to calculate fall speed for groups of pollen taxa (here assigned to LuVs To do so, Nielsen and Sugita (2005) used the taxon-specific fall speeds and weighted them with the proportion of each pollen taxon in the assemblage of all pollen taxa used in the study before calculating the mean fall speed for the group of pollen taxa assigned to the LuVs in the maps. This method can also be used without weighting the taxon-based fall speeds (our first alternative Alt 1). Mazier et al. (in press, in Fredh, 2012) summed the mean diameter of all pollen types included in a LuV to get the "taxa-group pollen diameter"). The taxa-group fall speed of pollen was then calculated using the Stoke's law and the "taxa-group pollen diameter" (our second alternative Alt 2). A fourth possible approach is to calculate the "taxa-group pollen diameter" by weighting the diameter of each pollen taxon in the group with its proportion in the pollen assemblage used in the study (comparable to the first approach above, our third alternative Alt 3).
Here we use the three alternative methods (Alt 1-3 above) to assess their effect on the four TWs and two sites (results shown only for one site and one TW in Fig. 1). LuV-fall speed is always slightly higher when method 2 is used. The differences between scenarios are due to the alternative Calluna assignments to LuVs. The variation in LuV-fall speed is much larger when Method Alt3 is used, both between scenarios and TWs. Moreover, the values are very different from those obtained with methods 1 and 2. This is of course due to the weighting of pollen-type sizes.

Distance weighting plant abundance for LuVs
We extracted absolute LuV data (in m 2 ) from the four harmonized maps (TW1-4, seeTable 1) in 10-m increments from the edge of the coring site out to 3 km using ArcGIS. The nonpollen-producing areas (NPP in Fig. 2    The map-DWLuVs (in proportion) for the time windows TW1-4 and scenarios I-V were calculated using the program DWPA calculator v6.0 (Sugita, 02 May 2012, unpublished) with the map-extracted LuV data at 10m-increments (in m 2 ) and the three sets of LuV-fall speed obtained with method Alt 1-3 as input data (Fig.2). We used the Prentice's pollen dispersal model for bogs (Prentice, 1985) and the mean radius of the 3 bog sites as the "common radius".
The differences between the map-DWLuVs using fall speeds Alt 1 or Alt 2 are very small, while the differences are larger between using fall speeds Alt 2 or Alt 3. The maximum differences are found for LuV2 (-12% at Stavsåkra, -7% at Storasjö) and LuV4 (+8% at both Stavsåkra and Storasjö) in scenario I (TW4) (Fig. 2). However, the effect on map-DWLuVs of the method used to calculate the fall speed for groups of pollen taxa is generally not very large. Therefore, the differences in map-DWLuVs wouldn't affect the test of the LRA's performance significantly. However, weighting taxon-specific fall speed of pollen or pollen diameters with the pollen proportions in the pollen assemblages used for the LRA reconstruction appears to be a circular approach. The LRA-estimated and map-extracted DWLuVs should be independent of each other if the LRA reconstructions are to be tested.
Therefore, we propose that the method used by Mazier et al. (in press;in Fredh, 2012), here Alt 2, is the soundest one in this case, and we recommend using this method in similar tests.
There are almost no between-scenario differences in map-DWLuVs when FSP2 is used, which is explained by small between-scenario differences in fall speed Alt 2 (FSPs 2) (Fig. 1).
There are large between-site differences in map-DWPLuVs when fall speed Alt 2 (FSP2) is used (Fig. 2)

Pollen records
The chronologies for the pollen counts used in this study are from Olsson et al. (2010) for Stavsåkra and Storasjö (LRA runs), Cui (2013) for Notteryd (LRA run), and Cui et al. (2013) for Kansjön and Trummen (REVEALS runs). Detailed information on the pollen data used and discussed in the main article is presented in Table 1 of the main article and Figure 3 of this Appendix. Figure 3 presents the pollen data in percentages and related pollen-based LRA estimates over the last 3000 years for the three small sites Stavsåkra, Notteryd, and Storasjö. The complete pollen data in percentages over the last 11 000 years are published in Greisman and Lemdahl (2009)  respectively. Therefore, the LRA-estimates are best to compare with TW2 than with TW1.  Figure 3 Taxon-based REVEALS estimates of regional plant abundance using the pollen records Trummen and Kansjön (two large lakes) and taxon-based LRA estimates of local