Hybrid datasets: integrating observations with experiments in the era of macroecology and big data
Hybrid datasets: integrating observations with experiments in the era of macroecology and big data
复制标题
DOI：
10.1002/ecy.2504
复制
发表时间：
2018-12-01
期刊：
影响因子：
4.8
通讯作者：
Rindi, Luca
中科院分区：
文献类型：
--
作者：
Benedetti-Cecchi, Lisandro;Bulleri, Fabio;Rindi, Luca
插文理解人类对生物圈日益增长的控制如何影响地球上的生命是一个关键的研究挑战。开源数据库的日益可用性促进了这一任务，使生态学家能够在前所未有的空间和时间尺度上解决科学问题。大型数据集大多是观察性的，因此它们发现变量之间因果关系的能力可能有限。实验更适合于归因于因果关系，但它们通常在范围上受到限制。我们提出了混合数据集，从观察与实验数据的整合，作为一种方法来利用的范围和能力，在生态研究中的因果关系。我们展示了如何用时间序列分析（收敛交叉映射）和宏观生态学（联合物种分布模型）中的新兴技术分析混合数据集，可以对非生物和生物过程的因果效应产生新的见解，否则很难实现。我们用海洋生态系统中的两个案例研究来说明这些原则，并讨论了在环境、物种和生态过程中推广的潜力。如果明智地使用，混合数据集的分析可能会成为标准的研究目标，寻求大规模生态现象的因果解释的方法。
Box Understanding how increasing human domination of the biosphere affects life on earth is a critical research challenge. This task is facilitated by the increasing availability of open-source data repositories, which allow ecologists to address scientific questions at unprecedented spatial and temporal scales. Large datasets are mostly observational, so they may have limited ability to uncover causal relations among variables. Experiments are better suited at attributing causation, but they are often limited in scope. We propose hybrid datasets, resulting from the integration of observational with experimental data, as an approach to leverage the scope and ability to attribute causality in ecological studies. We show how the analysis of hybrid datasets with emerging techniques in time series analysis (Convergent Cross-mapping) and macroecology (Joint Species Distribution Models) can generate novel insights into causal effects of abiotic and biotic processes that would be difficult to achieve otherwise. We illustrate these principles with two case studies in marine ecosystems and discuss the potential to generalize across environments, species and ecological processes. If used wisely, the analysis of hybrid datasets may become the standard approach for research goals that seek causal explanations for large-scale ecological phenomena.