layeranalyzer: Inferring correlative and causal connections from time series data in r

Abstract

Distinguishing correlative and causal connections among time series is an important challenge in evolutionary biology, ecology, macroevolution and palaeobiology. Here, we present layeranalyzer, an r package that uses linear stochastic differential equations as a tool for parametrically describing evolutionary and ecological processes and for modelling temporal correlation and Granger causality between two or more time series. We describe the basic functions in layeranalyzer and briefly discuss modelling strategies by demonstrating our tool with three disparate case studies. First, we model a single time series of phenotypic evolution in a bird species; second, we extract cyclical connections in the well-known hare-lynx dataset; third, we infer the correlative and causal connections among the genus origination and extinction rates of brachiopods and bivalves. We summarize the advantages and limitations of using linear stochastic differential equations and layeranalyzer for studying correlative and causal connections.