Data from: Fine-grain, large-domain climate models based on climate station and comprehensive topographic information improve microrefugia detection

Large-domain species distribution models (SDMs) fail to identify microrefugia, as they are based on climate estimates that are either too coarse or that ignore relevant topographic climate-forcing factors. Climate station data are considered inadequate to produce such estimates, a viewpoint we challenge here. Using climate stations and topographic data, we developed three sets of large-domain (450,000 km²), fine-grain (50 m) temperature grids accounting for different levels of topographic complexity. Using these fine-grain grids and the Worldclim data, we fitted SDMs for 78 alpine species over Sweden, and assessed over- versus underestimations of local extinction and area of microrefugia by comparing modelled distributions at species' rear edges. Accounting for well-known topographic climate-forcing factors improved our ability to model fine-scale climate, despite using only climate station data. This approach captured the effect of cool air pooling, distance to sea, and relative humidity on local-scale temperature, but the effect of solar radiation could not be accurately accounted for. Predicted extinction rate decreased with increasing spatial resolution of the climate models and with increasing number of topographic climate-forcing factors accounted for. About half of the microrefugia detected in the most topographically complete models were not detected in the coarser SDMs and in the models calibrated from climate variables extracted from elevation only. Although major limitations remain, climate station data can potentially be used to produce fine-grain topoclimate grids, opening up the opportunity to model local-scale ecological processes over large domains. Accounting for the topographic complexity encountered within landscapes permits the detection of microrefugia that would otherwise remain undetected. Topographic heterogeneity is likely to have a massive impact on species persistence, and should be included in studies on the effects of climate change.大区域物种分布模型（species distribution models, SDMs）无法识别微避难所（microrefugia），因为其依托的气候估算值要么分辨率过低，要么忽略了关键的地形气候驱动因子。学界普遍认为气候站数据不足以生成此类气候估算值，本文对此提出质疑。本研究利用气候站数据与地形数据，构建了三套覆盖45万平方千米大区域、50米细粒度的温度栅格，分别对应不同等级的地形复杂度。借助这些细粒度温度栅格与世界气候数据集（Worldclim），本研究为瑞典境内的78种高山物种拟合了物种分布模型，并通过对比物种分布后缘的模拟分布情况，评估了局域灭绝与微避难所面积的高估与低估情况。尽管仅使用气候站数据，但纳入已知的地形气候驱动因子后，我们对精细尺度气候的建模能力得到了显著提升。该方法能够反映冷空气堆积、距海距离与相对湿度对局域尺度温度的影响，但无法准确表征太阳辐射的作用。随着气候模型空间分辨率的提升以及纳入的地形气候驱动因子数量增加，预测的物种灭绝率随之下降。在地形参数最完整的模型中检测到的微避难所里，约有一半未在分辨率较低的物种分布模型，以及仅基于高程提取的气候变量校准的模型中被检出。尽管仍存在主要局限性，但气候站数据可用于生成细粒度的地形气候栅格，从而为大区域内局域尺度生态过程的建模提供了可行路径。对景观内的地形复杂度进行考量，能够发现原本难以被识别的微避难所。地形异质性可能对物种存续产生重大影响，因此在气候变化影响相关研究中应将其纳入考量范畴。