Palaeontologia Electronica palaeo-electronica.org Anthropologically introduced biases in natural history collections, with a case study on the invertebrate paleontology collections from the middle Cambrian Spence Shale Lagerstätte Anna F. Whitaker and Julien Kimmig ABSTRACT Natural history collections are critical for modern scientific investigations, which are greatly expanding on the potential data applications of historic specimens. How- ever, using these specimens outside their original intent introduces biases and poten- tial misinterpretations. Anthropogenic biases can be introduced at any point during the life of museum specimens, from collection, preparation, and accession, to digitization. These biases can cause significant effects when the user is unaware of the collection context, as specific collection biases are often known anecdotally, but not ubiquitously. In this case study, the University of Kansas collection of Spence Shale Lagerstätte material was examined for anthropogenic biases using a collections inventory, inter- views with stakeholders, and a literature review. Biases were found related to collector interest, locality preference, and researcher interest and specialization. These biases create a distorted view on the diversity and ecology of the Spence Shale, and need to be considered in future research. Anna F. Whitaker. Department of Chemical and Physical Sciences, University of Toronto Mississauga, 3359 Mississauga Road, Mississauga, ON L5L 1C6, Canada. [email protected] Julien Kimmig. Earth and Mineral Sciences Museum & Art Gallery, Pennsylvania State University, University Park, Pennsylvania, 16802, USA. [email protected] Keywords: Natural history collections; invertebrate paleontology; collections bias; soft-bodied preservation; Cambrian Submission: 25 June 2020. Acceptance: 20 November 2020. Whitaker, Anna F. and Kimmig, Julien. 2020. Anthropologically introduced biases in natural history collections, with a case study on the invertebrate paleontology collections from the middle Cambrian Spence Shale Lagerstätte. Palaeontologia Electronica, 23(3):a58. https://doi.org/10.26879/1106 palaeo-electronica.org/content/2020/3238-collections-biases Copyright: December 2020 Paleontological Society. This is an open access article distributed under the terms of Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0), which permits users to copy and redistribute the material in any medium or format, provided it is not used for commercial purposes and the original author and source are credited, with indications if any changes are made. creativecommons.org/licenses/by-nc-sa/4.0/ WHITAKER & KIMMIG: COLLECTIONS BIASES INTRODUCTION history collections are nonrandom and represent a ‘presence only’ dataset, and must be treated as Natural history collections (NHCs) are under- such (Wehi et al., 2012; Lipps, 2018; Hedrick et al., going a transformative role within science during 2020). Examining sources of anthropogenic bias the twenty-first century. More so than ever, the can help determine whether absences are true specimens they contain are being applied to “big absences, or false signals due to introduced condi- data” interdisciplinary problems such as disease tions. control, the global biodiversity and ecological cri- This paper provides an overview on anthropo- ses, and mitigation and measurement of the effects logically introduced biases in natural history sci- of global warming (Suarez and Tsutsui, 2004; ences that create discrepancies between the Winker, 2004; Baird, 2010; Pyke and Ehrlich, 2010; published literature, museum collections, and the Bakker et al., 2020; Hedrick et al., 2020). New reality of the natural world. As a case study, anthro- technological applications and methodologies, pogenic biases in the University of Kansas collec- such as genomics, computational statistics, CT tion of the Spence Shale Lagerstätte material are scanning, and machine learning are bringing identified through a collections inventory, types of renewed interest and application to historic collec- material in the collection, comparison with the pub- tions, decades or even centuries after they have lished literature, and identification of the collection been established (Wandeler et al., 2007; Nelson context by discussion with the collectors, collection and Ellis, 2018). Natural history collections span managers, and researchers who have used it from the seventeenth century on, making them through time. By identifying these biases, this work invaluable representations of a record of anthropo- will serve as a guide for future researchers using genic change on a global scale (Nelson and Ellis, the Spence Shale material, as well as an example 2018). The value of these historic specimens lies in for others concerned about collection bias in their their ability to be used in time-series data, repre- studies. senting multiple collecting events of a locality or generations of a species over time (Lister and Cli- INTRODUCTION OF BIASES IN NATURAL mate Change Research Group, 2011; Habel et al., HISTORY COLLECTIONS 2014; Holmes et al., 2016). Despite the nearly unlimited potential of natural history collections, The challenge of accounting for collections they are under threat from lack of funding, cuts to bias is due to the extreme variability in types of stewardship positions (curators, collections man- bias. Natural history collections can be biased in agers, researchers), and lack of space. However, infinite combinations relating to specimen taxon, there are several reasons natural history collec- preservation type, when the collection was made, tions must be maintained: e.g., historic collections the field collector, the collecting institution, etc. To are impossible to replicate on the same scale, they discuss these issues, this paper is organized represent taxa and localities that no longer exist, according to when a source of bias enters a collec- and they are cost-effective to maintain in compari- tion (Figure 1), with examples from across biologi- son to trying to accumulate new data sets (Allmon cal/paleontological collections. and Poulton, 2000; Suarez and Tsutsui, 2004; Field Bias Lister and Climate Change Research Group, 2011). Field collection is the moment a living organ- While the application of collections-based ism or natural object is removed from its environ- data grows, the potential distance between the ment and becomes a specimen of study. There are contextual knowledge of the data and end users is numerous biases introduced at this stage, often also increasing. Database aggregators and digiti- compounding as the distance between collection zation efforts are mobilizing data on exponential event and use in study occurs. To begin, the speci- scales, allowing researchers access to millions of men’s physical characteristics influence selection. specimen records across thousands of institutions Overly large or bulky specimens may be left in the (Nelson and Ellis, 2018; Hedrick et al., 2020). field due to the difficulty of transporting them While this increases the potential scope of the (Nekola et al., 2019). Paleontological material that analyses, it means the researcher is often removed is too large, or too remote to remove without spe- from personal familiarity of each data point. cialized mechanical equipment (e.g., cranes or Museum collections data have inherent vices that helicopters) must be left in situ, or be accounted for researchers must be made aware of, lest they mis- in the budget of the collecting trip. Amateur collec- interpret the resulting signals and patterns. Natural tors, or collectors from smaller institutions, are less 2 PALAEO-ELECTRONICA.ORG FIGURE 1. Examples of anthropological actions that create bias in natural history collections and when they occur. likely to be able to fund such coordinated efforts specimens, but should be cataloged and marked with external contractors. In modern and fossil col- as such. Thus, even for taxonomically-oriented lections, smaller specimens requiring specialized research, poorly-preserved or unidentifiable mate- equipment (microscopes, sieves) to locate may be rial should be considered for collection. ignored due to practicality. In contrast, coring or Specimens collected live (and to a certain drilling samples may exclude large specimens and extent fossil) in the field can be biased based on favor smaller specimens. In paleontological taxo- the stage of their lifecycle (Panchen et al., 2019), nomic collection, complete or articulated speci- as well as towards the most common times of field- mens may be favored over incomplete/ work: during the off-time of the academic calendar, disarticulated specimens, leading to a complete- holidays, etc. (Nekola et al., 2019). Additionally, the ness bias. While completeness of specimens is geopolitical landscape affects collection opportuni- crucial, or at least beneficial, to most types of ties; this is especially the case in times of high geo- research (e.g., systematics, biostratigraphy), there political unrest, leading towards variable collection are other research questions that can still be inves- over time (Panchen et al., 2019). tigated despite incomplete preservation, e.g., mac- The purpose of the collecting event often roevolution, paleoecology, stratigraphic/geographic defines what information is associated with the distributions (Lieberman and Kaesler, 2000; specimen. A major bias results from the fact that Allmon, 2005), and sometimes an incomplete fieldwork is often taxonomically-oriented,