Wildlife Biology, Big Data, and ReproducibleResearch

摘要

Changes in technology have made it possible to gather vast amounts of data, often of highquality, that in turn can improve the quality of wildlife biology. However, with this growth in data, practicessuch as data management, exploratory data analysis, data-sharing, and reproducibility of an analysis havebecome increasingly complex. These practices often depend heavily on computer scripting languages, and areoften hidden from the peer-review process despite their influence on the final results. Although these issueshave been discussed in the literature, they are generally dealt with in a piecemeal fashion, preventingsynthesis, and thereby slowing progress. We offer a conceptual framework to illustrate relationships amongthese practices, and show where wildlife biology as a field has embraced these changes, where awareness isgrowing, and where it lags behind other fields. We then present several case studies to emphasize theimportance of adopting these practices. Any of these case studies could have been conducted with littleattention to these practices or employing scripting languages, but there are many disadvantages to thisapproach including increased chance of errors, inefficiency, and lack of reproducibility. We suggest that achange in the culture of how wildlife biology is conducted is required and that this change will be fostered byintegrating these practices into wildlife biology education, implementation, and embracing the idea of opendata and open computer code.