Whale bones are regularly found during archaeological excavations. Identification of these specimens to taxonomic levels is problematic due to their fragmented state. This lack of taxonomic resolution limits understanding of the past spatiotemporal distributions of whale populations and reconstructions of early whaling activities. To overcome this challenge, we performed Zooarchaeology by Mass-Spectrometry on an unprecedented selection of 719 archaeological and palaeontological specimens of probable whale bone from Atlantic European contexts, from the Middle to Late Neolithic (c.3500–2500 BCE) to the eighteenth century CE.

The results show high numbers of Balaenidae (most likely North Atlantic right whale (Eubalaena glacialis)) and grey whale (Eschrichtius robustus) specimens, two species no longer present in the eastern North Atlantic. Many of these specimens derive from contexts associated with the known medieval whaling cultures of the Basques, northern Spaniards, Normans, Flemish, Frisians, Anglo-Saxons, and Scandinavians. This association raises the likelihood that pre-industrial whaling impacted these taxa, contributing to their extinction and extirpation respectively. Much lower numbers of other large whale taxa were identified, suggesting that it was once abundant and accessible whales that suffered the greatest long-term impact. The pattern of natural abundance leading to over-exploitation, well-documented for other taxa, is thus applicable to early whaling.

Whale bone samples were taken using a ©Dremel rotary tool removing a small piece of bone weighing up to c.500 mg. For 474 specimens, collagen was extracted using a modified Longin (1971) method as detailed in Seiler et al. (2019), with the addition of a lipid extraction step and the use of a higher acid concentration, at the National Laboratory for Age Determination, Norwegian University of Science and Technology, Norway.

Initially, the samples were crushed into small pieces and cleaned in an ultrasonic bath with 18.2 MΩ-cm ultrapure water (type 1) (three times five minutes). The samples were then ultrasonicated for 15 minutes in dichloromethane and methanol (2:1). This step was repeated three or more times until the solution was clear. Following this, the material was demineralized overnight using 2.44 M HCl (50 ml of solution per 100 mg of bone) in glass tubes within a vacuum desiccator kept at room temperature. The samples were then washed with ultrapure water until a pH of 3 to 4 and 4 ml of 0.5 % NaOH was added for 2–4 hours at room temperature to dissolve humic acids. After washing with ultrapure water till pH<10, 5 ml of 1.22 M HCl was added for 1–2 hours to remove atmospheric CO₂. The sample was then washed with ultrapure water till pH=3±0.2 and hydrolysed to gelatine at 70 °C overnight. Finally, the gelatine was filtered through a prebaked (900 °C) quartz filter (Merck Millipore, AQFA04700, 99.998 % capture for 0.3 μm particles) and the filtrate freeze-dried. The final collagen was taken to the Henry Wellcome Laboratory for Biomolecular Archaeology, University of Cambridge, UK, for ZooMS analysis. 

Collagen extraction failed for 25 specimens and when sufficient original bone material remained a subsample was taken directly to the Henry Wellcome Laboratory. Additionally, another 237 specimens were analysed by ZooMS without prior collagen extraction. For these 262 (25+237) specimens, approximately 30 mg subsamples were taken and processed at the Henry Wellcome Laboratory for Biomolecular Archaeology. The bones were demineralised in 0.6 M hydrochloric acid at 4°C for two weeks. Each sample was then centrifuged, the hydrochloric acid discarded, and the retained material rinsed three times with 200 μl of 50 mMol ammonium bicarbonate (AmBic) pH 8.0 solution, before being gelatinised in 100 μl of AmBic solution at 65˚C for one hour. For the 449 collagen samples prepared at NTNU (474 samples, minus the 25 which failed), 0.1 mg of collagen was also placed in AmBic solution at 65˚C for one hour.

For both the 262 and 449 sample groups, the gelatinised collagen, still in AmBic, was incubated with 0.4 μg of trypsin at 37˚C overnight, and subsequently acidified with 0.1% trifluoroacetic acid (TFA). The collagen was then purified using a 100 μl C18 resin ZipTip® pipette tip (EMD Millipore) with conditioning and eluting solutions composed of 50% acetonitrile and 0.1% TFA and washing solution composed of 0.1% TFA; the samples were eluted in a volume of 50 μl. Equal amounts of the collagen extract and α-cyano-hydroxycinnamic acid matrix solution (1% in conditioning solution) were mixed (1 μl each) and spotted onto a 384 spot MALDI target plate. Each sample was externally calibrated against an adjacent spot containing a mixture of six peptides (des-Arg1-bradykinin m/z = 904.681, angiotensin I m/z = 1295.685, Glu1-fibrinopeptide B m/z = 1750.677, ACTH (1–17 clip) m/z = 2093.086, ACTH (18–39 clip) m/z = 2465.198 and ACTH (7–38 clip) m/z = 3657.929). Samples were spotted in triplicate, and run on a Bruker ultraflex III MALDI TOF/TOF mass spectrometer with a Nd:YAG smart beam laser. A SNAP averaging algorithm was used to obtain monoisotopic masses (C 4.9384, N 1.3577, O 1.4773, S 0.0417, H 7.7583). Averaged spectra were created from the replicates for each specimen using mMass software (Strohalm et al., 2008), and then visually compared to published m/z markers for mammals, as presented in Buckley et al. (2009), Kirby et al. (2013), Buckley et al. (2014) and Hufthammer et al. (2018). An additional eight specimens were previously analysed at BioArch at the University of York following the protocol outlined in Rodrigues et al. (2018) and the results incorporated into this study.

For additional information see manuscript.

MALDI-TOF MS data was analysed using mMass software v5.5.0 (Niedermeyer & Strohalm, 2012); details can be found in the main manuscript.