Implementing Reproducible Fisheries Research: A Decade of Experience With the Kahawai Reporting System

TL;DR

This paper discusses a decade-long approach to making fisheries research fully reproducible through code and consistent computing environments.

Contribution

The paper presents a reproducible research framework that has been successfully applied in fisheries and other disciplines.

Findings

A reproducible research framework has been successfully used for a decade in fisheries data analysis.

The framework is transferable to other disciplines and improves trust in scientific publications.

Despite initial overhead, reproducibility increases transparency, collaboration, and efficiency.

Abstract

Scientific publishing is widely perceived to be in a state of crisis. A contributing factor is reproducibility: the extent to which the results can be replicated is key to assessing the reliability of a study. Reproducibility rates have often been found to be low, a situation complicated by the fact that many studies provide insufficient details of their methods.

For the last decade, we have been using a framework that allows us to fully reproduce the results of analyses of fisheries data. Key features are that the analyses are fully defined in code and run within a consistent computing environment. The approach has proven to be transferable, with the framework being adopted in a range of other disciplines.

Incentivising the broader adoption of fully reproducible analyses would assist in re‐establishing trust in scientific publications, addressing the “reproducibility crisis” while also providing a basis for strengthened peer review processes and increasing the likelihood that questionable research practices are identified.

Implementing fully reproducible analyses has some overhead, especially where the software tools that support the approach are unfamiliar. We have found that the benefits of openness, transparency, and efficiency, together with increased collaboration, make this overhead worthwhile.

Figures