Raster and original working data for the paper Holocene matters: landscape history accounts for current species richness of vascular plants in forests and grasslands of eastern Central Europe

Aim: Current species-richness patterns are sometimes interpreted as a legacy of landscape history, but historical processes shaping the distribution of species during the Holocene are frequently omitted in biodiversity models. Here, we test their importance in modelling current species richness of vascular plants in forest and grassland vegetation.

Location: Western Carpathians and adjacent regions.

Taxon: Vascular plants.

Methods: Numbers of all species and of habitat specialists were extracted from plot records of forest and grassland vegetation. For each plot, environmental and historical data were derived from thematic maps. Historical data related to the persistence of (i) temperate taxa during the Late Glacial and Early Holocene, (ii) open-landscape taxa during the Middle Holocene, and (iii) taiga species during the Late Holocene were based on 112 fossil pollen profiles. Boosted regression trees were used to model spatial patterns in species richness.

Results: Historical variables always appeared among the best predictors of current species richness. In light forests, species richness highly mirrored both the Late Glacial (12.5% contribution) and Middle-Holocene (8.6%) landscape history. The latter factor became an important predictor also for species richness of steppe grasslands (8.3%) along with temperature seasonality (11.9%). Species richness of dark coniferous forests was best predicted by the Late-Holocene occurrence of taiga forests (14.8%), which had an even stronger effect on the richness of habitat specialists (20.5%). 

Main conclusions: Landscape changes since the Last Glacial Maximum are important predictors of current plant species richness. The historical effects were found to be habitat-specific and, because they may interact with recent environmental conditions and anthropogenic pressures, they often show a non-linear relationship with species richness. We provide one possible direction of incorporating past landscape changes into the models of species richness.

Methods
Numbers of all species and of habitat specialists were extracted from plot records of forest and grassland vegetation (the EVA database). Environmental data were derived from GIS-based thematic maps. As historical data we used the pollen site scores on the first axis of principal coordinate analysis (PCoA) performed by Jamrichová et al. (2017 J Biog) and interpolated them using ordinary kriging with an exponential semi-variogram model. We also weighted each spatial interpolation by the regional relationship between PCoA scores and elevation. In the regions where this relationship was relatively stronger (as measured by the R2 of linear regression model), the interpolated PCoA scores depend more on elevation, whereas in regions where this relationship was weak, the interpolated PCoA scores depend more on spatial similarity among sites with pollen profiles.

## 研究目标
当前物种丰富度格局常被解读为景观历史的遗留效应，但全新世期间塑造物种分布的历史过程却常被排除在生物多样性模型之外。本研究旨在检验此类历史过程在模拟森林与草原植被现存维管植物（vascular plants）物种丰富度中的重要性。

## 研究区域
西喀尔巴阡山脉及其毗邻区域。

## 研究类群
维管植物（vascular plants）。

## 研究方法
从森林与草原植被的样地记录中提取所有物种及生境专性物种的数量（数据来源于EVA数据库）。针对每个样地，其环境与历史数据均源自专题地图。其中，历史数据基于112个化石花粉剖面构建，涵盖三类类群的存续情况：(i) 晚冰期与早全新世的温带类群；(ii) 中全新世的开阔生境类群；(iii) 晚全新世的泰加林物种。

本研究采用Jamrichová等（2017，《Journal of Biogeography》）通过主坐标分析（principal coordinate analysis, PCoA）得到的花粉样地第一轴得分作为历史数据，并使用搭载指数半方差模型的普通克里金插值法完成空间插值。同时，根据主坐标分析得分与海拔的区域相关性对每次空间插值进行加权：在该相关性较强（以线性回归模型的决定系数R²衡量）的区域，插值得到的主坐标得分更多依赖于海拔；而在相关性较弱的区域，插值得分则更多依赖于存在花粉剖面的样地间的空间相似性。最终采用提升回归树（boosted regression trees）模拟物种丰富度的空间格局。

## 研究结果
历史变量始终是现存物种丰富度的最优预测因子之一。在疏林（light forests）中，物种丰富度高度响应晚冰期（贡献占比12.5%）与中全新世（8.6%）的景观历史。中全新世的景观历史同时也是草原物种丰富度的重要预测因子（贡献占比8.3%），与温度季节性（11.9%）共同发挥调控作用。暗针叶林的物种丰富度则以晚全新世泰加林分布为最优预测因子（贡献占比14.8%），该因子对生境专性物种丰富度的影响更为显著（20.5%）。

## 主要结论
末次冰盛期以来的景观变化是现存植物物种丰富度的重要预测因子。历史效应具有生境特异性，且可与现代环境条件及人为压力产生交互作用，因此常与物种丰富度呈现非线性相关关系。本研究为将古景观变化纳入物种丰富度模型提供了一种可行方向。