Present-Day Vegetation Helps Quantifying Past Land Cover in Selected Regions of the Czech Republic

The REVEALS model is a tool for recalculating pollen data into vegetation abundances on a regional scale. We explored the general effect of selected parameters by performing simulations and ascertained the best model setting for the Czech Republic using the shallowest samples from 120 fossil sites and data on actual regional vegetation (60 km radius). Vegetation proportions of 17 taxa were obtained by combining the CORINE Land Cover map with forest inventories, agricultural statistics and habitat mapping data. Our simulation shows that changing the site radius for all taxa substantially affects REVEALS estimates of taxa with heavy or light pollen grains. Decreasing the site radius has a similar effect as increasing the wind speed parameter. However, adjusting the site radius to 1 m for local taxa only (even taxa with light pollen) yields lower, more correct estimates despite their high pollen signal. Increasing the background radius does not affect the estimates significantly. Our comparison of estimates with actual vegetation in seven regions shows that the most accurate relative pollen productivity estimates (PPEs) come from Central Europe and Southern Sweden. The initial simulation and pollen data yielded unrealistic estimates for Abies under the default setting of the wind speed parameter (3 m/s). We therefore propose the setting of 4 m/s, which corresponds to the spring average in most regions of the Czech Republic studied. Ad hoc adjustment of PPEs with this setting improves the match 3–4-fold. We consider these values (apart from four exceptions) to be appropriate, because they are within the ranges of standard errors, so they are related to original PPEs. Setting a 1 m radius for local taxa (Alnus, Salix, Poaceae) significantly improves the match between estimates and actual vegetation. However, further adjustments to PPEs exceed the ranges of original values, so their relevance is uncertain.REVEALS模型（REVEALS model）是一款可将区域尺度花粉数据重新换算为植被丰度的工具。本研究通过模拟实验探究了所选参数的通用影响，并基于120个化石样点的浅层样本与半径60 km范围内的实际区域植被数据，确定了适配捷克共和国的最优模型设置。研究结合CORINE土地覆盖（CORINE Land Cover）地图、森林清查数据、农业统计资料与生境制图数据，获取了17个植物类群的植被占比。模拟结果表明，调整所有类群的样点半径，会显著影响重花粉粒或轻花粉粒类群的REVEALS模型估算结果。减小样点半径与增大风速参数的调控效果相近。但仅针对本地类群设置1 m的样点半径（即便该类群花粉粒较轻），可获得更低且更准确的估算结果，尽管其花粉信号较强。增大背景半径不会对估算结果产生显著影响。通过将模型估算结果与7个区域的实际植被数据对比，本研究发现，精度最高的相对花粉生产率估算值（Relative Pollen Productivity Estimates，PPEs）来自中欧与瑞典南部。在风速参数默认设置（3 m/s）下，初始模拟与花粉数据得到的冷杉属（Abies）估算结果不符合实际情形。据此，本研究建议将风速参数调整为4 m/s，该值对应捷克共和国多数研究区域的春季平均风速。采用该设置对PPEs进行临时调整后，估算结果与实际植被的匹配度提升了3至4倍。研究认为，除4个例外情况外，该调整值均较为合理，因其处于标准误范围内，且与原始PPEs具有相关性。为本地类群桤木属（Alnus）、柳属（Salix）与禾本科（Poaceae）设置1 m的样点半径，可显著提升估算结果与实际植被的匹配度。但进一步对PPEs进行调整会超出原始值的取值范围，因此其相关性尚不明确。