Department of Ecology and Evolutionary Biology, University of Toronto, Toronto, ON, M5S3B2, Canada.
Trends Ecol Evol. 2012 Nov;27(11):637-47. doi: 10.1016/j.tree.2012.07.002. Epub 2012 Aug 14.
Many central questions in ecology and evolutionary biology require characterizing phenotypes that change with time and environmental conditions. Such traits are inherently functions, and new 'function-valued' methods use the order, spacing, and functional nature of the data typically ignored by traditional univariate and multivariate analyses. These rapidly developing methods account for the continuous change in traits of interest in response to other variables, and are superior to traditional summary-based analyses for growth trajectories, morphological shapes, and environmentally sensitive phenotypes. Here, we explain how function-valued methods make flexible use of data and lead to new biological insights. These approaches frequently offer enhanced statistical power, a natural basis of interpretation, and are applicable to many existing data sets. We also illustrate applications of function-valued methods to address ecological, evolutionary, and behavioral hypotheses, and highlight future directions.
许多生态学和进化生物学中的核心问题都需要描述随时间和环境条件变化的表型。这些特征本质上是函数，而新的“函数值”方法利用了传统的单变量和多变量分析通常忽略的数据的顺序、间隔和功能性质。这些快速发展的方法可以解释感兴趣的性状如何响应其他变量而连续变化，并且优于传统的基于汇总的分析方法，例如生长轨迹、形态形状和对环境敏感的表型。在这里，我们解释了函数值方法如何灵活地使用数据并带来新的生物学见解。这些方法通常提供增强的统计能力、自然的解释基础，并且适用于许多现有的数据集。我们还举例说明了函数值方法在解决生态、进化和行为假设方面的应用，并强调了未来的方向。
