Modelling of riverine ecosystems by integrating models: conceptual approach, a case study and research agenda
Modelling of riverine ecosystems by integrating models: conceptual approach, a case study and research agenda
复制标题
DOI：
10.1111/jbi.12009
复制
发表时间：
2012-12-01
影响因子：
3.9
通讯作者：
Fohrer, Nicola
中科院分区：
文献类型：
--
作者：
Jaehnig, Sonja C.;Kuemmerlen, Mathias;Fohrer, Nicola
目的水圈和生物圈之间高度复杂的相互作用，以及多因素的关系，表征溪流和河流之间的相互联系作用的不同元素的景观。在这些生态系统中应用物种分布模型（SDMs）需要特别注意，因为河流是线性系统，其非生物和生物条件以线性方式结构化，上游/下游或邻近地区的横向影响显着。我们的目标是建立一个河流生态系统中底栖无脊椎动物的建模框架，并在数据丰富的研究流域中测试我们的方法。位置我们提出了一个案例研究的低地基尔斯陶河位于北方德国的9公里的一段。方法结合水文模型、水力模型和物种分布模型，预测双壳类角质球壳在河流系统中的生境适宜性。水文模型产生的结果作为水力模型的输入，水力模型用于模拟河道内的水位、流速和泥沙排放。结果集成模型获得了良好的评价分数（受试者工作特征曲线下面积0.96; kappa 0.86;真实技能统计量0.95;敏感性86.14;特异性85.75）。采样点变量的平均值与预测分布的值无显著差异（MannWhitney U检验P > 0.05）。预计在基尔斯陶河9公里段的下游一半发生概率很高。模型中最重要的变量是输沙量（占40%），其次是水深（30%），流速（19%）和水流功率（11%）。主要结论的水文和水力模型是能够产生预测，在不同的空间尺度，这是已知的影响河流生物，这反过来，使用的空间数据模型作为输入。我们的案例研究取得了良好的结果，这符合我们的研究生物的生态知识。虽然这种方法是可行的，在当地规模的栖息地适宜性的预测（在这里：达到一个小流域），我们讨论未来的建模方法和大规模的应用程序仍然存在的挑战。
Aim Highly complex interactions between the hydrosphere and biosphere, as well as multifactorial relationships, characterize the interconnecting role of streams and rivers between different elements of a landscape. Applying species distribution models (SDMs) in these ecosystems requires special attention because rivers are linear systems and their abiotic and biotic conditions are structured in a linear fashion with significant influences from upstream/downstream or lateral influences from adjacent areas. Our aim was to develop a modelling framework for benthic invertebrates in riverine ecosystems and to test our approach in a data-rich study catchment. Location We present a case study of a 9-km section of the lowland Kielstau River located in northern Germany. Methods We linked hydrological, hydraulic and species distribution models to predict the habitat suitability for the bivalve Sphaerium corneum in a riverine system. The results generated by the hydrological model served as inputs into the hydraulic model, which was used to simulate the resulting water levels, velocities and sediment discharge within the stream channel. Results The ensemble model obtained good evaluation scores (area under the receiver operating characteristic curve 0.96; kappa 0.86; true skill statistic 0.95; sensitivity 86.14; specificity 85.75). Mean values for variables at the sampling sites were not significantly different from the values at the predicted distribution (MannWhitney U-test P > 0.05). High occurrence probabilities were predicted in the downstream half of the 9-km section of the Kielstau. The most important variable for the model was sediment discharge (contributing 40%), followed by water depth (30%), flow velocity (19%) and stream power (11%). Main conclusions The hydrological and hydraulic models are able to produce predictors, acting at different spatial scales, which are known to influence riverine organisms; which, in turn, are used by the SDMs as input. Our case study yielded good results, which corresponded well with ecological knowledge about our study organism. Although this method is feasible for making projections of habitat suitability on a local scale (here: a reach in a small catchment), we discuss remaining challenges for future modelling approaches and large-scale applications.