Natural history collections (NHCs) are invaluable repositories of biological information documenting the biodiversity of our planet (Winker 2004). Increasingly, NHCs have proven useful in providing evidence of climate change, pollution, species decline, habitat loss, and threats to agriculture and public health (e.g.: Colla et al. 2012, Graham et al. 2004, Lavoie 2013, Lister 2011, Page et al. 2015, Pyke and Ehrlich 2010, Suarez and Tsutsui 2004, Winker 2004).

One force driving the acceleration of interest in NHCs is enhanced accessibility. In the past, using a biological collection has been limited to: physically visiting a museum’s holdings; requesting a loan of material; or less often, obtaining a spreadsheet or email detailing the label data associated with a specimen. These traditional means of access created an unintentional bottleneck where only a limited number of people were privy to the specimens and their associated data. Even if a museum had developed a catalog, frequently the included information was not very extensive and often limited to a species list that lacked specimen-level data. In contrast, digitization of a NHC serves specimen photographs, a database of associated label data (collection localities, coordinates, dates of collection, habitat notes, etc.), and georeferenced distribution maps online—where these data can be read, retrieved, and reused by anyone with an Internet connection. Complementary data such as field journal entries made by collectors or audio recordings of species’ songs can be added to the entry for a specimen. This approach places a wealth of previously inaccessible “dark” information into the hands of all interested parties (Page et al. 2015). Natural history collections and their data are on the forefront of biological research, and we are just now glimpsing the power and capabilities of this multidimensional source of biological data (Baird 2010, Scoble 2010, Suarez and Tsutsui 2004, Vollmar et al. 2010).

Researchers are using these digitized databases to answer a wide range of important questions addressing: ecological processes and the biodiversity of species (e.g., Nelson et al. 2012, Newbold 2010, Ward 2012), pollinator decline (Colla et al. 2012,) threats to agriculture (Sánchez-Cordero and Martínez-Meyer 2000), phenological changes due to climate change (Fenberg et al. 2016), and risk assessment of invasive species (Peterson and Vieglais 2001). Interagency working groups have prioritized resources for building support for these data sources, such as the Network Integrated Biocollections Alliance, Interagency Working Group on Scientific Collections, and National Science Foundation (Collections in Support of Biological Research, Advancing Digitization of Biodiversity Collections, and Postdoctoral Research Fellowships using Biological Collections). Moreover, to address the growing need for digitized NHC data directly relevant to human health, a renewed focus to establish a vector-specific database of specimen data, infectious disease agents, and patient samples has been proposed to combat emerging infectious diseases across the globe (DiEuliis et al. 2016).

A useful model of the digitization process has been described by Nelson et al. (2012). In brief, each specimen is photographed with its physical labels, the resulting image is processed and uploaded to a relational database, information associated with that specimen is entered into the database, and the locality information for the specimen is georeferenced as well. To avoid confusion among individuals, each specimen must be labeled with a unique catalog number that is used to access its records in the database. Unique identifiers are critical to managing and understanding biological variation because two individuals possess distinct properties, and can be genetically distinct, phenotypically unique, and potentially even separate morphologically cryptic species. Unique identifiers are essential to preserve identity, handle biological variation, and for repeatable science. A lack of unique identifiers is most noticeable, for example, when published species identifications are called into question and specific material is unable to be tracked unequivocally to material deposited in a NHC for re-identification (Turney et al. 2015). A unique identifier is often printed as an alphanumeric code on the specimen label, although the identifier may also be a two-dimensional barcode that can be scanned by an optical reader to ensure transcription fidelity (iDigBio. 2016). A data matrix barcode offers the benefit of storing large amounts of information in a relatively small graphic, an important feature when label size is constrained by storage space.

The Virginia Tech Insect Collection (VTEC) has begun digitizing its specimen data online to serve a wider audience. A member of the Symbiota Collections of Arthopods Network (SCAN, Gries et al. 2014), VTEC’s data are in the public domain (Creative Commons CCZero) to allow unimpeded reuse of the data (VPI-VTEC in SCAN). Dating to 1888, VTEC holds nearly a half-million specimens largely collected from the mid-Atlantic and with a strong focus on the biodiverse Appalachian region. The collection is the official repository for specimens in the state of Virginia and represents a rich insect diversity, including pollinators and many native species whose populations are in decline due to habitat loss. Appalachia is a topographically complex and ancient region with an exceptional repository of high species richness and relative rarity of taxa (Stein et al. 2000). Many species in the region are short-range endemics with a critically imperiled G1 conservation status as ranked by NatureServe (Master et al. 2012), representing irreplaceable biodiversity with strong conservation value to the state and nation. The specimen data found in VTEC encompasses much of this unique biodiversity and is an underutilized resource that could benefit research, extension, and teaching at the state, regional, and national stages.

However, like many NHCs, VTEC faces limited funding that impacts basic curation and management of the collection. In the interest of conserving funding as much as possible, and to share an open source resource with other collections, we wrote our own script to generate labels with unique specimen identifiers as a low-cost alternative to commercially available software or costly pre-printed labels. The VTEC uses 9-digit locally unique identifiers printed on labels and 36-digit universally unique identifiers (UUIDs) automatically assigned and stored in SCAN for specimen management. Here we describe a script we developed using open source libraries and a graphics package for creating specimen labels with a unique alphanumeric reference number that is also encoded in a data matrix barcode.

Use and customization

Broader applications

The concept of a natural history collection has expanded over time from whole, preserved specimens to include other biological collections, such as living stocks and cultures, and preserved tissues. Our labels can be used with these types of collections, and also as part of collection management in geology, paleontology, archeology, and related fields. For our purposes, the data matrix barcoded labels created by makelabels are an integral part of digitizing the VTEC collection (Fig. 2). However, these scannable labels can be used within a lab to quickly identify the contents of storage units, such as unit trays, and larger organizational units such as cabinets. We plan to label our Cornell insect drawers, which house individual specimens, using the same process but with larger data matrix barcodes positioned on the front of the drawer.

Figure 2.  

Photo of two-spotted bumble bee, Bombus bimaculatus, with data matrix barcode label, as prepared for the digitization project in the Virginia Tech Insect Collection (VTEC).

We thank an anonymous software engineer for assistance in developing the makelabels script and reviewing the manuscript, and Patricia Shorter for her assistance with proofing different barcode symbologies, the preliminary generation of data matrix barcodes for this project, and testing the final codes with barcode readers. We greatly appreciate the helpful suggestions of two reviewers that improved an earlier version of this paper.