NaviDAM: A collaborative decision-support tool for navigating between detection and attribution methods in ecology

Disentangling drivers behind biodiversity change implies confidently detecting such shifts and being able to properly isolate and measure the influence of a given pressure, ruling out rival explanations. Although experiments are the gold standard for achieving such attribution, they are not always feasible, depending on the study's scale and the species populations involved. The need to leverage observational data for detection and attribution has led to the uptake of practices, hypotheses and methods inherited from causal inference. However, navigating the many modelling options and identifying suitable approaches for a given project is not straightforward. This is why we have introduced NaviDAM: an online decision-support tool designed to guide method selection by assessing users' case studies against a set of criteria. Depending on the main objective — effect estimation, change detection or scenario projection, for instance — options regarding data structure, affordable assumptions, desired model properties and programming languages must be selected. The tool covers a wide range of approaches, from predictive models to causal estimators. Methods are furthermore briefly described on dedicated pages. Redirection links lead to existing online materials that detail them, as well as to the reference method paper and application articles in ecology (with and without remote sensing data). Finally, their assessment table provides transparency on how the method is filtered. Method descriptions and assessments can be discussed and iteratively refined: NaviDAM is an open-source project that welcomes contributions from the community. The website is completed by good practices and gallery pages that respectfully cover cross-cutting subjects (e.g. review articles, causal graphs and collinear inputs) and provide examples of NaviDAM usage. Through collaborative efforts, the aim is to develop a tool which is: i) user-oriented, with a set of criteria to filter down candidate methods suited to the input case study; ii) method-inclusive, to match the variety of users' needs; and iii) resource-oriented, linking method materials and outstanding applications to help users get started on their own projects.