Buckley Hannah L, Day Nicola J, Case Bradley S, Lear Gavin
School of Science, Auckland University of Technology, Auckland, New Zealand.
School of Biological Sciences, Victoria University of Wellington, Wellington, New Zealand.
PeerJ. 2021 Apr 8;9:e11096. doi: 10.7717/peerj.11096. eCollection 2021.
Effective and robust ways to describe, quantify, analyse, and test for change in the structure of biological communities over time are essential if ecological research is to contribute substantively towards understanding and managing responses to ongoing environmental changes. Structural changes reflect population dynamics, changes in biomass and relative abundances of taxa, and colonisation and extinction events observed in samples collected through time. Most previous studies of temporal changes in the multivariate datasets that characterise biological communities are based on short time series that are not amenable to data-hungry methods such as multivariate generalised linear models. Here, we present a roadmap for the analysis of temporal change in short-time-series, multivariate, ecological datasets. We discuss appropriate methods and important considerations for using them such as sample size, assumptions, and statistical power. We illustrate these methods with four case-studies analysed using the R data analysis environment.
如果生态研究要为理解和应对当前环境变化做出实质性贡献，那么描述、量化、分析和检验生物群落结构随时间的变化，就需要有效且可靠的方法。结构变化反映了种群动态、生物量变化、分类单元相对丰度的变化，以及在不同时间收集的样本中观察到的定殖和灭绝事件。以前大多数关于表征生物群落的多变量数据集中时间变化的研究，都是基于短时间序列，这些短时间序列并不适用于诸如多变量广义线性模型等对数据需求较大的方法。在此，我们提出了一个用于分析短时间序列、多变量生态数据集时间变化的路线图。我们讨论了合适的方法以及使用这些方法时的重要注意事项，如样本量、假设和统计功效。我们用四个使用R数据分析环境进行分析的案例研究来说明这些方法。
