2023 ESA Annual Meeting

Reproducibility in science has been an important driver in adopting standards for sharing data and ultimately participating in global biodiversity knowledge. Many standards, such as Darwin Core, have been proposed to share data among scientists and data beneficiaries, such as governmental agencies and conservation organizations. Organizations with the objective of managing a large amount of data may be tempted to implement complex database systems, such as relational databases, which have been considered best practices to address all data needs. However, recent trends in the software development community have favored decoupled architectures, where each data layer answers a specific need.

As a national biodiversity observation network, we are required to work with varied data structures, such as abundance, time-series, occurrence records, and more. Data acquired through surveys will also vary in quality and precision and will have to meet different requirements throughout its life cycle, such as flexibility in attribute definition to accommodate survey teams and taxonomic revisions. Our data management systems also have to address the different data needs of our organization, such as data integration, storage, analysis, and visualization. Our initial efforts to accommodate those needs using complex data models such as Darwin Core for data storage required intensive work to develop and were challenging to maintain. A decoupled data layer architecture has proven efficient in designing and maintaining data stores to accommodate our varied data needs, simplify the implementation of new data structures and use cases, and maintain those systems. Each data layer is managed independently to fit specific needs. Specific data layers and interfaces are developed to share and integrate data to and from standard formats for collaboration, such as Darwin Core.

COS 40-2 - Managing biodiversity data : Why we don't use Darwin Core to store data

Presenting Author(s)

Coauthor(s)