Implementing Causal Inference in Ecology Through the Structural Causal Model (SCM) Framework

Abstract

Ecologists are often interested in understanding causal relationships from ecological data. However, developed methods for causal inference, particularly for observational-based studies are often not taught or applied in ecology. This thesis overviews how the structural causal model (SCM) framework can be employed to increase both observational and experimental causal inference in ecology, increasing the validity of causal conclusions drawn from ecological data. First, this thesis presents a comprehensive review of the SCM framework geared towards a general ecology audience interested in observational causal inference. Next, the SCM framework is applied to a local observational coral reef dataset, to determine the causal drivers of coral reef regime shifts in Seychelles. This framework is also applied to a global observational coral reef dataset, to understand the global drivers of reef fish biomass. The cumulative results from the above two studies provide practical guidelines on how to apply the SCM to both local and global ecological data, each highlighting that novel, reliable causal conclusions can be drawn from this approach. Using theory and simulated data, this thesis further explains how the SCM framework can be used to ensure proper study design and analysis across quasi-experimental (e.g., matching methods, before control impact, regression discontinuity design, instrumental variables) approaches. Last, using key ecological examples, this thesis explores how the SCM framework can also be employed to help ensure valid causal conclusions are drawn from experimental data. Ultimately, the increased uptake of the SCM framework across ecology can increase the depth and pace at which we understand causal relationships in nature.