The International Tree-Ring Data Bank at Fifty: Status of Stewardship for Future Scientific Discovery

Among the many remarkable features of tree rings is that networks of tree-ring chronologies accumulate information (Fritts 1976), empowering discoveries across multiple earth systems (Babst et al. 2018). The dendrochronology community recognized the synergistic capacity of combining individual site chronologies across large areas relatively early in the history of the science, leading to the founding of the International Tree-Ring Data Bank (ITRDB) in 1974 (Babcock 1974). The ITRDB was created to encourage global cooperation among the various branches of dendrochronology, especially dendroclimatology, and provide a permanent location for the storage of dendrochronological data from around the world. As the name states, the ITRDB originally operated as a sort of bank from which researchers could deposit and request data. Harold Fritts at the University of Arizona’s Laboratory of Tree-Ring Research led the effort to create the ITRDB and oversaw it until 1989 when it was transferred to the National Geophysical Data Center at the United States National Oceanic and Atmospheric Administration (NOAA). At that point the ITRDB became fully publicly available, and currently resides with the World Data Service for Paleoclimatology (WDS-Paleo) at NOAA’s National Centers for Environmental Information (NCEI), housed and managed along with more than a dozen other types of global paleoenvironmental proxy data. The ITRDB makes up roughly half of the total data holdings at WDS-Paleo, echoing the commitment of the dendrochronology community to develop and share tree-ring chronologies for public use.

In understanding global change and its effects, the ITRDB has proven invaluable. By capitalizing on the ways in which the network records variation in climate (St. George and Ault 2014), it has been used to develop large-scale, regional to continental reconstructions of drought (Cook and Krusic 2004; Cook et al. 2010a, 2020; Boucher et al. 2011; Stahle et al. 2016; Herrera and Ault 2017), temperature (Mann et al. 1999; PAGES 2k Consortium 2013; Wilson et al. 2016), soil moisture (Williams et al. 2020; Zhang et al. 2020), and seasonal precipitation (Stahle et al. 2020), along with hundreds of more spatially specific reconstructions of runoff and other climatic features and phenomena. The strengths of dendroclimatic collections have also enabled assessments of climate and land-surface models (Cook et al. 2010b; Woodhouse et al. 2010; Ault et al. 2014; Jeong et al. 2021) and provided accessible data to resource managers (Rice et al. 2009). Although some studies have demonstrated limitations for ecological forecasting because of the targeted nature of tree-ring collections in the ITRDB (Nehrbass-Ahles et al. 2014; Klesse et al. 2018), the suite of tree-ring growth sensitivity to climate variation embedded in the network has nonetheless underscored vulnerabilities of global forests to anthropogenic climate change (Salzer et al. 2009; Liu et al. 2013; Williams et al. 2013; Charney et al. 2016; Babst et al. 2019). As the dendrochronology community continues to grow into “frontier” regions (e.g.Solomina et al. 2022; Zuidema et al. 2022) and develops new data collection and analysis techniques (Pearl et al. 2020; Griffin et al. 2021), the future of the science is promising, highlighting the continued need to expand and maintain the ITRDB as a central repository for tree-ring chronologies (Babst et al. 2017).

Our team operates the World Data Service for Paleoclimatology at NOAA’s NCEI. The WDS-Paleo is certified by the CoreTrustSeal as a Trustworthy Data Repository that complies with established standards in data stewardship. We receive federal funding to support stewarding the ITRDB and associated paleo archives with the mission of making these valuable environmental data publicly accessible, updated, and secure. It is our job to work with the tree-ring community to build the ITRDB as a resource by curating new submissions in a timely manner, performing quality checks, and collaborating on building new tools and adding linked resources to the data bank. These efforts are guided by the dendrochronological community through the ITRDB advisory committee, which is currently composed of active dendrochronologists from around the world. Our work has benefited immensely from multiple independent efforts of dendrochronologists within the community, including an exhaustive review of datasets in the ITRDB by Zhao et al. (2019), who error-checked and reformatted numerous measurement files. We can now report that all the issues identified by Zhao et al. have been resolved on the ITRDB. Other efforts to improve the ITRDB have been helpful and inspiring, and we encourage such efforts to continue in collaboration with our team.

At the time of writing, the ITRDB contains more than 9000 tree-ring measurement files, the raw data of dendrochronology. This is still a growing network, with hundreds of new or updated sites submitted each year. Each site has associated metadata that describe the location, species, investigators, related publications, and funding sources. Measurement data, including total or partial ring widths and other derived measurements such as density and blue intensity, are provided in separate text files in “Tucson decadal” format and the tab-delimited, spreadsheet-like NOAA/WDS-Paleo archival template format. An additional text file provides chronology statistics from the COFECHA program (Holmes 1983), which is used as a rough check on measurement data quality as described below. All data files associated with an ITRDB site are stored in an HTTPS web-accessible folder (Table 1). The tree-ring measurement data in both Tucson and NOAA formats are stored in the same directory, while the crossdating statistics from COFECHA are stored in a separate directory for tree series statistics. All of the files are linked from the site's landing page for easy access. Users may obtain site metadata in several ways (Table 1), including searching for landing pages via the NCEI WDS-Paleo search engine, through an interactive map, and programmatically through our web service, an application programing interface (API; Gross et al. 2022). Various researchers have used these services to build tools that access and display the data (e.g. Dendrobox, Zang 2015). Since the release of dplR (Bunn 2008), a growing suite of Open Source tools for analyzing tree-ring measurements is available in the R programming language (R Core Team 2020). Our team is currently developing tools in R to aid in searching and accessing NCEI WDS-Paleo resources, including the ITRDB and climate reconstructions. As a prelude to those future packages, we provide an R script that imports the entirety of ITRDB metadata and raw measurement files into R (see Guiterman 2023).

All of the measurement data housed on the ITRDB meet the FAIR guiding principles, making them “findable, accessible, interoperable, and reusable” (Wilkinson et al. 2016). We recognize limitations in the Tucson decadal format but emphasize that because of ongoing efforts to clean existing data sets and quality-check new ones, every one of the raw data (.rwl) files is machine readable. Specifically, the files can be read into R by using the dplR function read.rwl() with the added parameter, format=”tucson”. In addition, we provide the same data in a tab-delimited format in NOAA template files. Following community needs (e.g.Zhao et al. 2019), each site is assigned a unique and permanent dataset digital object identifier (DOI). These identifiers are important for documenting the completion of data management plans to many funding agencies (including the US National Science Foundation) and in meeting Open Data standards at a growing number of peer-reviewed journals. The data DOIs, from one of the earliest contributions (Fritts 1997) to one of the most recent (Khan 2022), can also be cited directly in research papers and reports. The citability of data DOIs is particularly useful in instances of datasets without an associated peer-reviewed publication, which should also be cited along with the data DOI. To further aid in identifying and searching for ITRDB records, we have implemented the use of standardized variable names in the NCEI WDS-Paleo dataset search (Table 1) (PaST; Morrill et al. 2021).

We welcome and accept dataset contributions at any time. Researchers are encouraged to submit their data in advance of journal submission and funding agency deadlines; we will work with you to plan the timing of dataset release. For instance, some journals require an embargo period on data until a paper is published, or contributors may independently want to hold the release of data for publication. In such situations, we can delay the data release until a specific date, ensuring that datasets are finalized and data DOIs are minted well ahead of the deadlines. Before submitting data, please consult our contributions page for instructions and guidelines (Table 1). Note that all raw measurement files must be readable by read.rwl() in the dplR library, and additionally pass basic quality standards. Although we rely on contributors to guarantee the accuracy of the tree-ring measurements and their associated crossdating, we provide COFECHA output for users to assess the data quality according to their own standards. We have, however, only accepted submissions that have fewer than 40% “problem segments” and greater than 0.35 mean series intercorrelation, based on standards provided to us decades ago. Contributors are recommended to check that they achieve these standards in their own COFECHA output, and to also ensure the readability of their Tucson decadal files in R prior to submission. Once the data are assimilated into the ITRDB and publicly released, it will be available through the NCEI WDS-Paleo search page and site landing page (Table 1).

As the dendrochronology community expands and builds greater diversity of its people, along with the regions, species, and methods it represents, we are excited to continue to support the community and its collections. We recognize the challenges this growth poses to the ITRDB, which may have rigid statistical standards of crossdating for trees from some areas and is inherently limited in capturing the rich data and metadata of many subfields of dendrochronology, including archaeology and ecology. From our position of support to the paleosciences communities, we look forward to discussions about how to encourage and accommodate this essential growth, whether that be within the ITRDB or in a new data archive.

The foresight of dendrochronologists to create the ITRDB a half-century ago generated innumerable insights into earth systems and societal responses to global change, and with continued commitment to making tree-ring data publicly accessible and usable, the future of dendrochronology is equally bright.

We are grateful to the founders of the ITRDB and to its stewards over the last 50 years, including in particular Hal Fritts and Bruce Bauer, along with Wendy Gross, Eugene Wahl, and Imke Durre. We also thank the members of the tree-ring community who have graciously shared their data on the ITRDB to both build and enrich the global tree-ring network. We thank Peter Brewer and Andy Bunn for their assistance and support, and Malcolm Hughes for generously providing information about the establishment of the ITRDB and its conveyance to NOAA. More information about this history is available at https://www.ncei.noaa.gov/pub/data/paleo/about/itrdb/. We are also grateful for constructive comments on this manuscript by Brooke Adams and two anonymous reviewers. This manuscript and our stewardship are supported by NOAA Cooperative Agreements NA17OAR4320101 and NA22OAR4320151 to CIRES, and the NCEI Science and Data Stewardship Support 1332KP19FNEEN0003 to Riverside Technology, Inc.