by Desconhecido
Fig. 6. Graphs showing cereal-pollen percentages from each individual pollen site plotted on a time scale (site numbers refer to Fig. 5 and Appendix 1). Dark-yellow graphs show percentages of the pollen sum (the length of the vertical scales is 10%), whereas light-yellow graphs show an exaggeration by ×10. The year 1350 is indicated by a red line
Fig. 7. Bubble graph showing the change in cereal-pollen percentages around 1350, based on a comparison between the two periods 1250–1350 and 1350–1450. Each site is represented by a circle or a dot, plotted against altitude (vertical axis) and south-to-north position (horizontal axis). Blue circles represent sites that show an increase in cereal-pollen percentages from 1250–1350 to 1350–1450, and red circles represent sites that show a decrease. The sizes of the circles are proportional to the size of the increase or decrease, respectively. Black dots represent sites with no change
In some of the original diagrams, different types of cereal pollen have been distinguished. However, with the exception of rye, the distinction between different cereal-pollen types is difficult and sometimes arbitrary, and different analysts may use different identification criteria. Therefore, to enable comparison between the sites and calculations of mean values, they are all put together here into one common graph including all types of cereal pollen from each site.
The first impression from Figure 6 is that all the graphs look different. The differences reflect differences in the history of cereal growing between the sites, but that is not the whole truth. The pollen percentages are also influenced by other factors, like the size of the sampled basin (i.e. the peatland or lake) and the distance between the basin and the past cereal growing. Furthermore, the cereal-pollen percentages are affected by vegetation structure, which may have differed from site to site and also through time.
Hence, the graphs all look different, partly because of differences in their local agriculture and settlement histories and partly because of pollen-dispersal factors. Still it is possible to find patterns beyond the site-to-site variation. For instance, all sites show higher cereal-pollen percentages from the last millennium than from the previous one. Furthermore, most of the sites show a decrease during the last one or two centuries, reflecting the farm abandonment and forest plantation that characterised the uplands in the late nineteenth and twentieth centuries. Hence, the comparison of cereal graphs from different sites reveals both local variations and common regional trends. Also a closer look at the time of the Black Death and the late-medieval crisis, indicates great variation between the sites, but some important conclusions can be made. Of the 25 sites from southern Sweden, 11 show decreasing cereal-pollen percentages around 1350 (Grisavad, Östra Ringarp, Yttra Berg, Trälhultet, Rosts täppa, Flahult, Öggestorp, Store mosse, Skärpingegöl, Bråtamossen and Mattarp), others show little change, whereas only five sites show increasing values (Torup, Häggenäs, Skärsgölarna, Bocksten and Råshult). Thus, the number of sites showing decreasing cereal-pollen percentages around 1350 is higher than the number of sites showing increasing values.
Furthermore, from the same figure it can be concluded that the few sites that show increasing cereal-pollen percentages around 1350 are situated at relatively low elevation. This relationship is also evident from the bubble graph in Figure 7. The graph reveals clearly that two sites at low altitudes (Torup and Häggenäs) deviate strongly from the others by showing significantly increasing cereal-pollen percentages. These two sites are also the two southernmost sites, situated not far from the agricultural plains of Scania (sites 2 and 3 in Figs 5 and 6).
To summarise the evidence so far, pollen sites that show decreasing cereal-pollen percentages around 1350 are common in the South-Swedish Uplands, whereas two lowland sites in southernmost Sweden, on the contrary, show a distinct increase. The difference between lowland and upland sites is intriguing, but because of the low number of lowland sites any interpretations regarding the development in the lowlands have to be tentative and do not allow for any detailed analysis.
The same is true for much of middle and northern Sweden. Only three sites with detailed chronologies were available for the present study, and they represent rather different ecological and climatic zones. They are here used only as examples of agricultural development in different regions. Site Fjäturen, which is situated in a low-lying agricultural area north of Stockholm, has a cereal-pollen graph similar to many sites in the South-Swedish Uplands. It shows an increase in cereal-pollen percentages in the thirteenth century followed by a decrease in the fourteenth century. The latter indicates a late-medieval agricultural decline. It was followed by a second expansion in the seventeenth century. Lake Kalven is situated in an upland forest region and mining district of middle Sweden. It shows an increase in cereal-pollen percentages immediately before 1350, reflecting the first agricultural expansion in the area. The expansion was soon followed by stagnation and possibly decline after 1350. A renewed expansion started in the fifteenth century, later followed by a much stronger expansion in the seventeenth century. Lake Kassjön is the northernmost site, situated outside Umeå in northern Sweden. A continuous cereal-pollen graph begins in the fourteenth century. There is no clear indication of decline but rather a slow step-wise increase in cereal pollen, reflecting expansion periods in the sixteenth and eighteenth centuries.
The two sites from middle Sweden – Fjäturen and Kalven – show indications of late-medieval agricultural decline. In particular Lake Fjäturen is interesting, since it is the only lowland site in the data set that shows strong indication of agricultural decline in connection to the Black Death. Lake Kassjön in northern Sweden shows no such decline, but rather indicates an establishment of continuous crop growing at the same time. Written records are scarce and it is still an open question if the Black Death reached that far north.
Similar to lowland sites from southern Sweden, the sites from middle and northern Sweden provide interesting examples but are too few to allow for any representative analysis. In contrast, the South-Swedish Uplands are very well represented. As many as 21 pollen sites in the data set are from this relatively homogeneous upland region, which makes a solid basis for interpretation. Therefore the rest of this chapter will focus on the uplands. (Hopefully future research will provide more pollen sites from lowland areas as well as from regions further north.)23
We may now look at the average development in the South-Swedish Uplands by calculating mean cereal-pollen percentages for all the 21 upland sites together. This was done for 50-year time slices and the result is presented in Figure 8a. As evident from the diagram mean cereal-pollen percentages from upland sites are very low during much of the first millennium after Christ. Following an initial increase in the late eighth and early ninth centuries a significant increase starts in the late eleventh century and continues to the early fourteenth century. This increase reflects medieval colonisation and expansion, which is well known from other sources, like settlement remains, written documents, place names, etc. The expansion was part of a general agricultural expansion over much of Europe.24 The upward trend in the graph is broken in the mid-fourteenth century by an abrupt decrease in mean cereal-pollen percentages. There can be no doubt that this decrease reflects a significant decrease in cereal growing in the uplands after the Black Death. After the decline, values remain low for at least a century, until they start to increase slowly during the late fifteenth century onwards. In the sixteenth century they reach the same level as before the decline. Except for a possible stagnation in the early eighteenth century they continue to rise until the early nineteenth century. From the late nineteenth century onwards, cereal-pollen percentages decline, which reflects depopulation of the countryside and large-scale introduction of modern forest plantations in the uplands.
The fact that the mean cereal-pollen percentages for the upland sites drop sharply in connection to the Black Death is a strong and independent indication of agricultural decline. It shows that in spite of the site-to-site variation, and in spite of the
very low cereal-pollen percentages in general from upland sites, the regional trend is clear. The mean value decreases from 0.4% to 0.2% of the pollen sum, i.e. a halving.25
In order to estimate what this change in mean pollen-percentage values represents in the landscape, past vegetation cover for each 50-year time slice has been modelled using the so-called Landscape Reconstruction Algorithm.26 This modelling technique uses pollen counts, pollen-dispersal characteristics (pollen production and fall speed of pollen grains) and site characteristics (basin size) to translate the fossil pollen record to vegetation cover. Thus, the model enables us to describe the actual extent of cultivated fields, pastures and woodland. The model corrects for over- and under-representation of strong and weak pollen producers, respectively (for instance, most trees are over-represented in the pollen record and most herbs under-represented). Some of the problems connected with the interpretation of pollen percentages, for instance the over-representation of plants growing on the sampled peatland, also affects the model output, but the technique is the best available to reconstruct vegetation cover from pollen data. For this study pollen data and site characteristics for all the 21 upland sites have been used.27
Fig. 8. Bar charts showing (a) the mean cereal-pollen percentages for each 50-year time slice based on all the 21 upland sites, and (b) the corresponding cereal land-cover reconstruction using the Landscape Reconstruction Algorithm, submodel REVEALS. The year 1350 is indicated by a red line
The graph in Figure 8b shows the reconstructed land cover of cereal vegetation in the upland region for each 50-year time slice.28 The shape of the graph is similar to the one for cereal-pollen percentages, with only small differences. Most important in this context is that the reconstruction allows us to tentatively discuss not only pollen percentages but also the land cover of arable fields. According to the reconstruction, cultivated arable fields (excluding fallows) covered approximately 2% of the upland region in the late thirteenth and early fourteenth centuries. After a drop around 1350 they covered approximately 1% in the late fourteenth and early fifteenth centuries. After the late-medieval nadir they expanded to reach the maximum distribution of all time in the nineteenth century, when they covered 4–5% of the landscape. Thereafter they decreased and in the late twentieth century they covered approximately 2%.
Fig. 9. Cereal-pollen percentages from all the 21 upland sites. Each dot represents an original pollen sample with its corresponding cereal-pollen percentage (vertical axis) and original dating (horizontal axis). The samples were sorted chronologically and a running average was calculated (blue line). Four dots with very high values from the nineteenth century were excluded from the picture but included in the running-average. The year 1350 is indicated by a red line
Based on historical documents, the average land cover of arable fields for southern Sweden during the early eighteenth century has been estimated to approximately 5%, and for forested uplands to approximately 2%.29 The latter figure fits reasonably well with the land cover reconstruction based on pollen records presented here. According to the reconstruction, cereal cultivation covered approximately 3% of the uplands in the eighteenth century (Fig. 8b). The good agreement for the eighteenth century, from which there is plenty of documentary evidence, indicates that the pollen-based reconstruction is reliable enough also for earlier periods.
Regarding the drop around 1350, the reconstruction shows a halving of cereal cropland, from c. 2% to 1% of the land cover. Thus the pollen record – both pollen percentages and pollen-based land-cover reconstructions – gives support to earlier interpretations of historical records that suggest extensive farm abandonment in upland areas during the late-medieval crisis. Even though a halving of cereal production, as indicated by the pollen record, does not necessarily reflect a farm-desertion frequency of 50% (farms may have survived but reduced their acreage), it indicates that a desertion frequency of that size is not at all unlikely.
Some earlier pollen studies have similarly shown a decrease in cereal-pollen percentages after the Black Death. Evidence comes from several parts of northern and western Europe, for instance from Sweden, Norway, Denmark, Estonia, France, Netherlands, England and Ireland.30 Many of these studies are of high quality, with high-resolution analysis and detailed chronologies, but most of them are based on single pollen sites or occasionally a few sites. Dan Yeloff and Bas van Geel reviewed pollen-analytical evidence on late-medieval farm abandonment in Europe.31 Their review considers only positive evidence of abandonment and they emphasise that there are also many other pollen diagrams from Europe that instead show continuity. They arrived at the conclusion that the effect on the European rural landscape of the Black Death varied geographically. They also concluded that abandonment in some areas started already before the Black Death, for instance in France and England.
In spite of these earlier studies, it is difficult to draw any general conclusion on the late-medieval decline based on pollen data, in particular on a European scale but also on national and regional scales. However, the method used in this study – to combine a large number of local pollen studies from one region at such high time resolution – is a step forward. With this approach it has been possible to interpret the regional development and at the same time reveal local variation within the region.32 Also, the technique of landscape reconstruction modelling used here enables us to discuss the actual extent of arable and how it has varied through time.
Sudden impact of the Black Death
The late-medieval crisis was a complex and multifaceted process that lasted for more than a century and several different factors may have contributed to the circle of events. The Black Death of 1347–52 was certainly a major factor, but in some parts of Europe there was stagnation and even farm abandonment already before that period, in particular after the harvest failures leading to the Great Famine of 1315–22. There is no evidence of the latter from Sweden but that may very well be due to a paucity of documentary records.33 If the harvest failures struck Sweden, farm abandonment might be expected, not only connected to the Black Death, but also before.
Furthermore, abandonment may have continued long after the Black Death. Based on documentary evidence, Janken Myrdal suggested that widespread farm abandonment did not begin until the 1360s and 1370s. Before that time, vacant farms were probably taken over by the many landless and poor. Abandonment started when this reserve of the labour force was used up and it continued well into the fifteenth century. It was most extensive in the 1420s or somewhat later. According to Myrdal the prolonged process of abandonment was due to recurring outbreaks of plague in combination with social oppression and conflicts.34
The pollen data presented here may contribute to the discussion of timing. In the diagram for upland sites in Figure 8a, mean cereal-pollen percentages decrease exactly at 1350, but this is partly due to the construction of the diagram. To enable the calculation of mean values, the original samples from the different pollen sites were adjusted by interpolation to 50-year time slices.35 Therefore changes in the diagram are restricted to 50-year intervals. All that can be concluded from the diagram, regarding the time of the Black Death, is that the mean cereal-pollen percentages decrease significantly from the 1300–1350 time slice to the 1350–1400 time slice.
In an attempt to obtain a finer resolution than 50 years, a different type of diagram is presented in Figure 9. This diagram is also based on all the 21 upland sites, but instead of showing calculated mean values based on time slices, it shows all the original pollen samples from the different sites. In large measure the curve of the running average resembles the graph of mean values in Figure 8a. The important difference is that this one is not limited by the 50-year intervals of the time slices. Therefore it is interesting to note that it falls sharply exactly at 1350.
The chronologies of the different pollen diagrams included in the data set are all based on radiocarbon dating. Even though only pollen diagrams with chronologies that are regarded as sufficiently reliable have b
een included, radiocarbon dating is still a rather blunt technique in this context. Even if the radiocarbon dates are correct themselves they have a statistical error. Furthermore, the construction of the time/depth models used to calculate sample ages in the original studies include errors connected to interpolation.36 In spite of these possible chronological errors, which probably blur the picture, the mean cereal-pollen graph shows a distinct decline around 1350 (Fig. 8a) and, more important, the running-average curve for the original samples falls steeply at exactly 1350 (Fig. 9). Hypothetically, the fall in the latter would be even more concentrated to 1350 if the underlying chronologies were perfectly correct.
In conclusion, pollen data from the uplands indicate an immediate effect of the Black Death, which we know from other sources ravaged Sweden in the year 1350. They do not indicate a delayed effect, as suggested by Myrdal based on documentary evidence. However, the running-average curve seems to support his conclusion that abandonment continued into the early fifteenth century. Regarding the Great Famine in the early fourteenth century, the curve shows a slight decline from approximately 1320, but perhaps not strong enough to prove abandonment already at that time.
This conclusion of the timing of the decline based on pollen data may be compared with an entirely different type of source material, namely the dating of wood by dendrochronology. This dating method is very precise. If the outermost tree ring is preserved, the very year when the tree was cut down can be identified. If the outer tree rings are not preserved, the felling year may be estimated. The precision of this estimation varies, and depends on how much of the tree trunk that is preserved, but it is usually higher than for radiocarbon dating.