Ecological studies require quality data to describe the nature of ecological processes, relationships, and increasingly to understand ecosystem changes. Increasing access to big data collected via different technologies has magnified both the burden and the complexity of ensuring quality data. The costs of errors in ecological studies include low utilization of data, increased time spent cleaning data, and poor reproducibility that can result in a misunderstanding of ecosystem processes and dynamics. These costs erode the efficiency and trust of ecological research. While the imperative for data quality is broadly recognized, the specific principles of ensuring data quality are rarely discussed. "QA/QC" may be referenced with little consensus as to what specific quality steps entail as well as the activities which apply across data types and sub-disciplines of ecological research and application. Consequently, a common framework is necessary to describe quality assurance (QA) and quality control (QC) and how these processes, separately and together, improve data quality. We present a holistic view of QA and QC with examples from broad-scale ecological monitoring programs (e.g., NEON, BLM AIM), community led science (e.g. LandPKS), and small-scale research labs.
Results/Conclusions
In a new treatment of QA and QC, we highlight opportunities for preventing and detecting errors beyond data collection to every aspect of the data flow. We propose that preventing errors at each stage should receive greater emphasis than common reactionary approaches to data quality control. Careful planning to identify methods, data structures, roles and responsibilities, and timelines can drastically reduce downstream errors. Calibrating field observers as well as instruments is critical for detecting methodological errors before they propagate throughout a dataset. During data archiving, storage, analysis, and publication steps, reproducible data manipulation is necessary to avoid inadvertently introducing data structure and value errors. Improving and maintaining data quality should be adaptive: upon learning from past mistakes, we place emphasis on iterating QA and QC processes to improve over time. Our vision is to empower every ecologist producing or encountering a dataset to ensure, preserve and, where possible, improve the integrity of data, thereby increasing the value of ecosystem scientific discovery and decision making.