Transparency and reproducibility in invasion science

Fabio Mologni , Jason Pither

Policymakers and practitioners overseeing invasive species management depend on reliable research for guidance. Transparency and reproducibility are core features of reliable research, and prerequisites for outcomes to be independently replicated within the same or different systems.  These features are evidently lacking in many science disciplines, including Ecology.  In this Discussion paper, we first report the findings of an assessment of 49 primary research studies that were part of a systematic mapping effort, showing that invasion science research exhibits the same shortfalls as ecology research more broadly. For instance, only one study explicitly considered statistical power in the methods describing study design, and only 2 studies provided access to both data and code, which is the minimum requirement for computational reproducibility.  We then discuss the implications that low statistical power has for published invasion science research, for designing studies, and for policymakers and practitioners relying on primary research to inform their decisions. We then make specific recommendations, targeting the same stakeholders as well as publishers, on how to maximize the reliability of invasion science research moving forward.  This includes explicitly considering and ideally estimating statistical power, undertaking a study pre-registration, making all relevant code and non-sensitive raw data accessible and useable, and devising and upholding clear and consistent policies on transparent reporting and open materials.