Sensitivity of floodplain geoecology to human impact in the Dijle catchment

Many fluvial systems in West and Central European have undergone important changes in their sediment dynamics and sediment storage during the Holocene, and it is important to understand the driving forces of these changes in order to understand the sensitivity of fluvial systems to environmental changes. The overall objective of this PhD thesis is to provide a better understanding of the sensitivity of floodplain geomorphology and ecology ( geoecology ) to human impact in the landscape and to attain a detailed insight in the timing, nature and extent of human impact on floodplain geoecology. The Dijle catchment (758 km²), located within the Belgian loess belt, can serve as a model for most lowland rivers in the European loess area and is selected as the main study area in this thesis. Previously gathered data on sediment dynamics and floodplain geomorphology were combined with new detailed field-based sediment data, palynological data and radiocarbon ages to provide a more comprehensive overview of the changing floodplain geoecology in the Dijle catchment for the Holocene period. These data showed that during the Middle Holocene, the floodplain was mainly a relatively stable environment and consisted of a strongly vegetated marsh (dominated by an alder carr forest), which resulted in peat accumulation. Throughout the Late Holocene, floodplain geoecology changed with clearing of the Alder carr forest, the development of a single channel river and the dominance of minerogenic overbank sedimentation.A detailed chronology of the floodplain changes indicated that the end of the peat growth and the transition towards clastic overbank deposition was diachronous at the scale of the entire catchment, ranging between 6500 and 500 cal a BP. Moreover, a well-dated cross-section at the downstream end of the study area suggested that were the floodplain is wide, floodplain changes were asynchronous at cross-section scale, caused by a gradual increasing sediment input in the floodplain that first affected the near-channel parts. These results indicate that conclusions with respect to past changes in sedimentation rate and floodplain morphology should be based on a multi-transect and multi-core dating approach.In order to understand the role human impact has played in changing the floodplain geoecology, a reconstruction of vegetation changes and a semi-quantification of anthropogenic land cover changes were made, based on palynological and statistical analysis (cluster analysis and non-metric multidimensional scaling (NMDS)). The results show that the pollen records and the NMDS analysis could not detect Mesolithic or Neolithic human activities in the Dijle catchment. In these periods, human impact in the catchment was limited to local disturbances and small-scale forest clearances. The Dijle catchment was covered by deciduous forest, dominated by Quercus, Corylus and Tilia, the natural vegetation for the catchment. Only from the Bronze Age onwards (ca. 3900 cal a BP) human impact clearly increased and vegetation gradually changed. Human impact further increased from the Iron Age onwards, except for a temporary halt between ca. 1900 and 1600 cal a BP, possibly coupled with the Migration Period in Europe. The intensity and onset of the increasing human impact differs between the different study sites. The relationship between pollen deposited in the floodplain and the vegetation abundance in the landscape was provided based on simulations in the HUMPOL software suite. Such model simulations are needed to translate the modern pollen signal from alluvial sites into vegetation abundance in the surroundings of the sampling site. Our results indicate that HUMPOL simulations provide a better similarity to the observed pollen signal in the modern alluvial deposits, compared to the observed pollen and the vegetation in the catchment of the sampling site. However, more information on the importance of fluvial pollen transport is needed to correctly interpret the outcome of the multiple scenario approach (MSA), applied to fossil pollen assemblages from alluvial sites and to select the most likely reconstruction of past vegetation.Finally, the observed floodplain changes were evaluated against the available data on human impact, for the headwaters and for the entire Dijle catchment. This integrated approach demonstrated that during the Neolithic Period, human impact in the catchment was nearly absent and floodplains consisted of a strongly vegetated marshy environment where organic material accumulated, which is considered as the natural state of the floodplain. The increase in human impact from Bronze Age onwards caused an increase in soil erosion and in hillslope-floodplain connectivity. Consequently, sediment input in the floodplain system increased and floodplain geoecology changed towards a more open floodplain dominated by clastic overbank deposits, mainly as the indirect result of an intensification of agricultural activities. At the scale of the entire Dijle catchment, the gradual changes in floodplain morphology coincided with the gradually increasing human impact in the catchment, which suggests a linearity between the external forcing (human impact)and geomorphic response (floodplain change). However, at the narrow floodplains in the headwaters, the gradual increase in human impact contrasts with the abrupt change in floodplain geoecology, only triggered when human impact reached a threshold. Observed differences in the process-response model at catchment scale are attributed to differences in hillslope-floodplain connectivity, the location within the catchment and mainly to differences in the timing and intensity of human activities between different subcatchments.status: publishe