This tool is based on two main principles: a causal path diagram and a Structural Equation Model (SEM) framework. The development of causal path analysis allows for the visual representation of the different cause-and-effect relationships among the environmental components from which ecosystem function and services emerge. They simplify the system by mapping how variables influence one another (e.g., seabed type -> oxygen regime -> carbon recycling & storage -> climate regulation) and help identify direct, indirect and feedback effects that are difficult to detect. These diagrams rely on available information and first principles and represent testable hypotheses that can then be verified with available data using the SEM framework. SEM is a quantitative framework able to quantify and partition the relative direct and indirect pathways between multiple processes within a system. Some forms of the SEM approach consist of multiple sub-models within a global model, which allows for testing of direct and indirect effects within a causal path diagram to tease out the mechanisms driving ecosystem functioning and service provisioning. Once the causal path diagram is defined and the appropriate data formatted, the SEM models are fitted after checking for collinearity. The are refined and simplified iteratively, where necessary, until the model is determined to be an adequate representation of the data (determined by a statistical test). Pathways can be modified (added or removed) to fit the data better whilst retaining ecological realism. Equally statistically valid alternative models are compared, and the recurring parameters and consistent pathways are highlighted. Typical output represents a network of interactions going from the left (descriptors = environmental features) to the right (ecosystem function and services), positive (black) and negative (red) relationships are represented alongside the magnitude of effect (arrow thickness) and significance (stars). The tool is data greedy, and all components of the diagram must be supported by data to be tested appropriately, such that often a compromise between complexity and data availability is necessary. SEM works as a system of multivariate correlations so continuous data for each descriptor, and the responses is essential. Although categorical data can technically be accommodated, they are generally dealt with through individual SEM versions for each data category separately.