A review of the heterogeneous landscape of biodiversity databases: opportunities and challenges for a synthesized biodiversity knowledge base

Abstract

Aim: Addressing global environmental challenges requires access to biodiversity data across wide spatial, temporal and biological scales. Recent decades have witnessed an exponential increase of biodiversity information aggregated by biodiversity databases (hereafter ‘databases’). However, heterogeneous coverage, protocols, and standards of databases hampered the data integration among databases. To stimulate the next stage of data integration, here we present a synthesis of major databases, and investigate i) how the coverages of databases vary across taxonomy, space, and record type; ii) the degree of integration among databases; iii) how integration of databases can increase biodiversity knowledge; iv) the barriers to databases integration.Location: GlobalTime period: ContemporaryMajor taxa studied: Plants and VertebratesMethods: We reviewed the scope of twelve well-established databases and assessed the status of their integration. We synthesized information from these databases to assess major knowledge gaps and barriers to fully integration. We estimated how improved integration can increase the coverage and depth of biodiversity knowledge. Results: Each reviewed database had unique focus of data coverages. Data flows were common among databases, though not always clearly documented. Functional trait databases were more isolated than those pertaining to species distributions. Poor compatibility between taxonomic systems used by different databases posed a major challenge to integration. We demonstrated that integration of distribution databases can lead to greater taxonomic coverage that corresponds to 23 years’ advancement in knowledge accumulation, and improvement in taxonomic coverage could be as high as 22.4% for trait databases. Main conclusions: Rapid increase of biodiversity knowledge can be achieved through the integration of databases, providing the data necessary to address critical environmental challenges. Our synthesis provides an overview of the integration status of databases. Full integration across databases will require tackling the major impediments to data integration – taxonomic incompatibility, lags in data exchange, barriers to effective data synchronization, and isolation of individual initiatives.

Ten (mostly) simple rules to future-proof trait data in ecological and evolutionary sciences

“…Train the world: There is an urgent need for more comprehensive trait data across the globe and the tree of life (Feng et al, 2022) and thus increasing access to training globally is critically important. Open access publications, tools, data, and educational resources provide ways to help lower the barriers to participation (Evans & Reimer, 2009).…”

“…examples in Kattge et al, 2020). Thus, these studies also often involve linking to different types of data, which requires interoperability between datasets (Feng et al, 2022;Gallagher et al, 2020).…”

Representation of global change drivers across biodiversity datasets

“…In this study, we present a framework to test data representation across different global change drivers over space and time that can be applied to other datasets. We recommend that future syntheses explicitly include tests of the representation of their data for the global change drivers being tested in addition to highlighting other data gap 26,33,38,65,[71][72][73][74] .…”

“…A lot of the focus in the literature is on filling geographic 24,26,32,34 , temporal 24,26,36,37 , and taxonomic 33,37,38 biodiversity data gaps, but this focus should be shifted towards prioritizing regions that undersample global change. For example, we are currently lacking biodiversity data from places with high magnitudes of climate change including Arctic and boreal forest regions, as well as tropical regions that are currently entering non-analog climate spac 90 .…”

“…To upscale the findings of syntheses of local-scale data to estimates of global or mean biodiversity change, the underlying data should be representative across multiple dimensions: space, time, taxonomic variation, as well as variation in the drivers of biodiversity change [32][33][34][35] . There is already recognition of the biases associated with space and geography 24,26,32,34 , time and historical baselines 24,26,36,37 and taxonomy 33,37,38 . In contrast, sampling biases with respect to drivers of change are rarely emphasized in the existing literature (but see 39 for spatio-taxonomic biases in North American butterfly occurrence records).…”