The discovery of lapis lazuli pigment preserved in the dental calculus of a religious woman in Germany radiocarbon-dated to the 11th or early 12th century, a rare pigment used in illuminated manuscripts.
During the European Middle Ages, the opening of long-distance Asian trade routes introduced exotic goods, including ultramarine, a brilliant blue pigment produced from lapis lazuli stone mined only in Afghanistan. Rare and as expensive as gold, this pigment transformed the European color palette, but little is known about its early trade or use. Here, we report the discovery of lapis lazuli pigment preserved in the dental calculus of a religious woman in Germany radiocarbon-dated to the 11th or early 12th century. The early use of this pigment by a religious woman challenges widespread assumptions about its limited availability in medieval Europe and the gendered production of illuminated texts.
In nature, blue pigments are relatively rare, occurring in mineral seams that must be mined. Throughout the European medieval period (5th to 15th centuries AD), only a small number of natural and synthetic blue pigments were known, including ultramarine, azurite, Egyptian blue, smalt, and vivianite (table S1). Among these blues, ultramarine, made by grinding and purifying lazurite crystals from the ornamental stone lapis lazuli (1-4), was, by far, the most expensive, reserved along with gold and silver for the most luxurious manuscripts (2, 5). Unlike other blues, such as azurite and vivianite, ultramarine is both brilliant and highly stable, even at high temperatures, and when made from high-quality lapis lazuli and well purified using oil flotation, a deep blue hue can be achieved (4, 6). Mined from a single region in Afghanistan (7), lapis lazuli was a quintessential luxury trade good in the Eurasian pre–Modern period, and its waxing and waning availability in artistic centers throughout Eurasia reflects both its enormous expense and the circuitous supply lines along which it was traded over thousands of miles (8, 9).
Within the context of medieval art, the application of highly pure ultramarine in illuminated works was restricted to luxury books of high value and importance, and only scribes and painters of exceptional skill would have been entrusted with its use (5). Before the 15th century, however, scribes seldom signed their works, raising questions as to the identity of early scribes and illuminators (5, 10). Even among books in women’s monastery libraries, fewer than 15% bear female names or titles, and before the 12th century, fewer than 1% of books can be attributed to women (11). Consequently, it has long been assumed that monks, rather than nuns, were the primary producers of books throughout the Middle Ages (5). Recent historical research, however, has challenged this view, revealing that religious women were not only literate but also prolific producers and consumers of books (10-12). In Germany and Austria, religious women played a particularly active role in book production, and their work as scribes and illuminators can be traced to as early as the late eighth century (10, 12). Although surviving examples of these early works are rare and relatively modest, there is a growing body of evidence that women’s monasteries were actively producing books of the highest quality by the 12th century (10). The dual-sex monastery of Admont in Salzburg, for example, supported a community of nuns who copied many of the more than 200 surviving books from the monastery’s 12th-century book collections, and Diemut, a 12th-century female scribe at the monastery of Wessobrunn in Bavaria, was recorded to have produced more than 40 books, including a richly illuminated gospel (10). From the 13th to the 16th centuries, during which documentary evidence and record keeping in Germany is more complete, more than 4000 books attributed to over 400 women scribes have been identified, and active scriptoria have been identified at 48 women’s monasteries (11). However, identifying the early contributions of religious women to medieval book production is challenging due to the limited number of surviving books, the precarious documentation of women’s monasteries, and the tendency of scribes to leave their work unsigned (10). As a result, individual female scribes remain poorly visible in the historical record, and it is likely that most of their scribal work has gone unrecognized.
Recently, microscopic analyses have revealed that dental calculus (calcified tooth tartar) can entrap and preserve a wide range of microdebris related to craft activities (13, 14). Here, we report the identification of lazurite and phlogopite crystals, in the form of powder consistent in size and composition with lapis lazuli–derived ultramarine pigment, that were found embedded within the dental calculus of a middle-aged woman buried in association with a 9th- to 14th-century church-monastery complex at Dalheim, Germany (Fig. 1). Radiocarbon-dated to AD 997–1162, this woman represents the earliest direct evidence of ultramarine pigment usage by a religious woman in Germany. Moreover, because the monastery and the entirety of its contents were destroyed during a 14th-century fire, this finding of lapis lazuli potentially represents the sole surviving evidence of female scribal activity at the site. Our results suggest that dental calculus can be used to help identify scribes and artists in the archaeological record and to aid in the historical reconstruction of women’s monasteries and their role in book production. In addition, although the importation of this expensive foreign pigment into medieval Europe is first materially attested in the 10th century (15), its presence in an otherwise unremarkable women’s community in northern Germany powerfully testifies to the expansion of long-distance trading circuits during the 11th-century European commercial revolution.
Blue Particle Identification in Dental Calculus
In 2014, during a separate study on the identification of plant microremains in dental calculus (16), numerous particles of blue color (Fig. 2) were observed embedded within the dental calculus of a 45- to 60-year-old woman buried in association with a medieval church-monastery complex at the site of Dalheim near Lichtenau, Germany (Fig. 1). Radiocarbon-dated to cal. AD 997–1162 (95% probability; fig. S1), this individual, B78, was otherwise unexceptional, presenting no notable skeletal pathologies or evidence of trauma or infection (16, 17). Further osteological investigation did not detect indications of hard labor, while dental analysis revealed heavy calculus deposits on the anterior teeth (fig. S2) and only mild to moderate periodontal disease accompanied by the antemortem loss of two molars, likely due to caries (16). Female biological sex was confirmed using both osteological and genetic methods (16), and the skeletal remains are now curated at the Institute of Evolutionary Medicine at the University of Zürich. Few historical records survive for the church-monastery complex, which now stands in ruins. A stone church dedicated to St. Peter was likely first constructed at the site during the ninth century and later expanded. Although the founding date of the Dalheim women’s monastery is unknown, four nearby Benedictine and Cistercian women’s monasteries were founded in AD 1127, 1140, 1142, and 1149 (18). The earliest surviving texts documenting the women’s community at Dalheim date to AD 1244, 1264, and 1278 and describe it as a house of Augustinian canonesses attached to a church dedicated to St. Peter (18). Excavated by the Westphalian Museum of Archaeology from 1988 to 1991, the monastery is believed to have housed approximately 14 religious women until its destruction by fire following a series of 14th-century battles (18, 19). An unmarked cemetery, from which B78 was excavated, is located immediately adjacent to the church.
To isolate the blue particles for further study, we first sought to demineralize the surrounding dental calculus using a dilute HCl solution (0.05 M), as is typically performed during microbotanical analysis. However, we found that this procedure led to color instability and loss (fig. S3); by comparing colors of the acid-demineralized calculus to reference pigments, we confirmed that using an acid as a decalcifying agent is detrimental to color stability and particle size in lapis lazuli, azurite, malachite, and vivianite (fig. S4; see the Supplementary Materials). We then tested an alternate approach on a second dental calculus sample from the same individual, decontaminating the calculus surface and then disrupting the calculus structure by sonication in ultrapure water. Calculus fragments and mineral particles released by this procedure were transferred to a microscope slide without mounting media or coverslip and allowed to dry under controlled conditions. Inspection under light microscopy revealed more than 100 particles of deep blue color (Fig. 2), many of which were observed in situ still encased within fragments of dental calculus (Fig. 2B). All subsequent analyses used sonication for pigment isolation.
Distribution of Blue Particles in Dental Calculus
In most cases, the blue particles appeared singly—not as clumps—having the appearance of a blue powder dispersed across many dental calculus fragments (fig. S5, A to E). Blue particles were observed across calculus pieces originating from different teeth, suggesting that the particles entered the calculus in different episodes rather than as a single localized event, and over a period of time as the dental calculus matrix calcified. Average particle size was 10.9 ± 9.5 μm (SD), which is consistent with published data for natural lapis lazuli pigment (20) and our own measurements of 10.8 ± 8.0 μm (SD) for reference Afghan lapis lazuli pigment (table S2 and data file S1).
Elemental Composition of Blue Particles Consistent with Lazurite
We next compared the blue particles optically to a selected reference panel of blue mineral pigments, followed by elemental analysis using scanning electron microscopy with energy-dispersive x-ray spectroscopy (SEM-EDS) (Fig. 3). With the exception of lazurite (the dominant blue mineral in lapis lazuli), all blue pigments that were available and used during the medieval period contain metal (copper, cobalt, or iron) as a major element in their composition (table S1). SEM-EDS analysis of the archaeological particles and reference blue pigments allows a clear distinction between major element compositions (Fig. 3). The archaeological blue particles lack copper, cobalt, and iron, thereby excluding pigments containing these metals as major elements, but they closely resemble the elemental composition of lazurite, the sulfur-containing tectosilicate that gives lapis lazuli its dark blue color.
Identification of Lapis Lazuli Minerals Lazurite and Phlogopite
To confirm the identification of lapis lazuli, we analyzed the archaeological particles using micro-Raman spectroscopy. The micro-Raman spectra generated from the archaeological blue particles yielded a positive match to modern reference lapis lazuli pigment (Fig. 4) and to a previously published medieval lapis lazuli pigment identified from a medieval fresco painting (21), as well as to other lazurite spectra available in the RRUFF database (22). The spectra taken from the B78 sample show the characteristic modes for lazurite, Na3CaAl3Si3O12S, at 258, 548, 803, and 1096 cm−1, with the strongest modes being the symmetric S3− ν1-stretching mode at 548 cm−1 and its overtone at 1096 cm−1, as well as the S3− ν1-bending mode at 258 cm−1 (20). These spectral characteristics allow for an unambiguous identification of the mineral particles as lazurite.
In addition, using micro-Raman spectroscopy, we identified a colorless, translucent mineral accompanying the blue lazurite crystals within the archaeological sample as phlogopite (Fig. 5). The phyllosilicate phlogopite, K(Fe, Mg)3(Si3Al)O10(F,OH)2, is an accessory mineral found to accompany the tectosilicate lazurite in lapis lazuli stone (7). Iron-rich phlogopite has some characteristic vibrational modes that are activated if Fe substitutes Mg (23). In the presence of Fe, the Si-Ob-Si mode at 681 cm−1 forms a triplet, and an additional mode appears near 550 cm−1 as observed in our sample. In addition, a systematic downshift of the Si-Ob-Si peaks has been reported with increasing iron content (23). The latter could provide an opportunity to correlate the lazurite with specific mining locations where the characteristic Fe/Mg ratios are known. Overall, our analyses show that the blue pigment found in the B78 sample is lazurite and that the colorless mineral is iron-rich phlogopite. Together, lazurite and phlogopite allow for a positive identification of the archaeological blue particles as originating from lapis lazuli.
How a middle-aged woman living a life of apparently low physical labor and buried in a cemetery associated with a woman’s religious community came to have such a rare and expensive mineral pigment in her dental calculus is not entirely certain, but we propose four possible scenarios: (i) B78 was a scribe or book painter engaged in the production of illuminated manuscripts, (ii) B78 was employed in the preparation of artist materials for herself or other scribes, (iii) B78 consumed lapis lazuli in the context of lapidary medicine, or (iv) B78 performed emotive devotional osculation of illuminated books produced by others.
Scenario 1: Book Production
The most parsimonious scenario is that individual B78 was a woman engaged in the production of high-quality manuscripts. The commissioning of a talented female scribe to produce deluxe liturgical books using expensive materials has precedent in Germany at this time. For example, a pair of letters dated to between AD 1140 and 1168—nearly contemporaneous with the burial of B78—detail an exchange between Sindold, the keeper and corrector of books (armarius) of the men’s monastery at Reinhardsbrunn, and the women’s monastery of Lippoldsberg where his sisters lived, located only 70 km east of Dalheim (see the Supplementary Materials). In his letter, Sindold commissions the “skillful” production of a deluxe, illuminated matutinal (liturgical book) to be produced by sister “N” using parchment, leather, pigment, and silk that he provided for that purpose (24). That the Reinhardsbrunn armarius would outsource the production of such an important and valuable book to a women’s monastery speaks to the reputation of women as makers of books by the 12th century. While Sindold does not elaborate on the specifics of the pigments being sent, judging by the amount of parchment (the equivalent of 384 pages) and the inclusion of silk, it can be assumed that the pigments were at least commensurate in quality and expense. Among surviving books from Germany that have been tested and are known to contain lapis lazuli pigment, the earliest putatively attributed to a woman scribe is a copy of the Liber Scivias (Heidelberg University Library, Codex Salemitani X,16) by Hildegard of Bingen of the women’s monastery at Rupertsberg and produced circa AD 1200 (15); however, the unsigned paintings were colored by at least two anonymous individuals (15).
In Germany, women’s monastic communities, especially during earlier periods, were largely made up of noble or aristocratic women. Many were highly educated, and devotional reading was encouraged as an expression of piety. These women would have led lives largely free of hard labor, consistent with the absence of occupational skeletal stress observed for B78. Work was encouraged within the monastery, however, and activities related to book production were considered worthy pursuits. In adding detail to their illuminations, it is plausible to assume that artists would have occasionally licked their brushes to make a fine point, a practice that later artist manuals refer to explicitly (4). In doing so, pigments, such as lapis lazuli, may have been introduced into the oral cavity, where they could have become entrapped within dental calculus. The repeated activity of inserting the tip of the brush into the mouth could explain the distribution pattern of in situ blue particles observed across multiple calculus fragments.
Scenario 2: Pigment Presentation
It is possible, although less likely, that the lapis lazuli pigment was introduced into the oral cavity of B78 through pigment production rather than painting. Ultramarine pigment production from lapis lazuli stone is described in numerous late medieval instruction manuals (4), of which the Italian 15th-century text Il Libro dell’Arte by Cennino Cennini (AD 1437) is perhaps the most detailed and best known (25). In it, Cennini describes a laborious process of grinding, progressively washing, and levigating lapis lazuli stone powder followed by oil flotation to remove impurities and concentrate the blue-bearing lazurite crystals, and he warns that the mortar in which the lapis lazuli stone is ground should be covered so that “it may not go off in dust” (see the Supplementary Materials). This airborne dust could potentially come into contact with dental calculus through accidental inhalation, either during the pigment preparation itself or afterward, such as during workshop cleaning activities. Experimental work confirms that it is possible for particles to enter the oral cavity by pigment preparation, even when only a limited amount of airborne dust is produced (see the Supplementary Materials; fig. S6). In addition, the handling of the dry pigment powder can itself create airborne dust that can settle on the face and lips, as well as enter the oral cavity.
Cennini also notes that the work of pigment production is typically performed by women, but this gendered division of labor may be a late medieval development associated with the professionalization of trade and crafts. During earlier periods, it is not specified how or who produced finished pigments from raw materials, and few recipe books are known from the 12th century AD and earlier (2, 4). Although high-quality lapis lazuli pigment first appears in European manuscripts as early as the 10th century (4, 15), the Arabic method of oil flotation described by Cennini, which is necessary to produce a brilliant blue rather than a dull bluish-gray pigment, is not attested in European artist manuals before the 15th century (4). As such, this raises questions as to whether scribes of the 11th and 12th centuries produced their own lapis lazuli pigments or received them as imported finished products from trading centers such as Alexandria, via Italian merchants. It is possible that the scribes themselves prepared their own—possibly lower quality—pigments or that scribes may have been provisioned with finished pigments produced by others, either locally or abroad. If a religious woman at Dalheim was preparing lapis lazuli pigment, then it is likely that it was for her own use or for another female scribe within her religious community.
Scenario 3: Lapidary Medicine
Alternatively, B78 may have consumed powdered lapis lazuli as a form of lapidary medicine. Since antiquity, lapis lazuli stone has been ascribed magical and healing powers by many Old World cultures, who used it primarily as an amulet stone and as a component of eye ointments (3, 26). The first-century Greek medical text De Materia Medica by Dioscorides describes the medicinal libation of lapis lazuli to treat scorpion bites, ulcers, eye growths, pustules, and herniated membranes (27), and an inventory of a Jewish apothecary in Cairo dating to the 13th and 14th centuries refers to the use of lapis lazuli as both an antivenom and an eye treatment (28). Medical lapis lazuli was particularly important in the medieval Islamic world, where it is amply attested in numerous medical recipe books (3, 29). By contrast, lapis lazuli first appears in European medical texts only in the 11th century (29) in such early works as the late 11th-century Liber de lapidibus by Marbod of Rennes (30), the 12th-century Physica by Hildegard of Bingen (31), and the 12th-century Circa Instans (32). Although these works describe the medical use of lapis lazuli (see the Supplementary Materials), there is little evidence that the Mediterranean and Islamic method of ingesting lapis lazuli pigment was widespread or even practiced in 11th- and 12th-century Germany. Consequently, although the ingestion of medical lapis lazuli by B78 cannot be ruled out, it appears unlikely given the paucity of evidence for this practice.
Scenario 4: Devotional Osculation
Last, women’s devotional reading has a long history, with documentation dating back to at least the sixth century in Merovingian Gaul (10). Beginning in the 14th and 15th centuries, devotional reading became more emotive under the influence of the Netherlands reformer Geert Grote of Deventer, and ritual osculation (kissing) of painted figures in illuminated prayer books became more common. In response, illuminators began adding decorative “osculation tablets” to these books to deflect kissing away from the painted figures, which suffered damage—including paint removal—through repeated kissing (33). Although it is possible that the lapis lazuli in the calculus of B78 entered the oral cavity through the kissing of painted images, the radiocarbon date of the skeleton makes it highly unlikely because such forms of emotive devotion are not attested until nearly three centuries later. In addition, osculation would have likely lifted clumps of pigment together with associated mounting media, which we did not observe in the B78 dental calculus.
Together, the presence of lapis lazuli in the dental calculus of B78 is best explained by its accidental incorporation during painting and/or pigment preparation. The discovery of lapis lazuli in the dental calculus of an 11th-century religious woman is without precedent in the European medieval archaeological record and marks the earliest direct evidence for the use of this rare and expensive pigment by a religious woman in Germany. Within medieval Europe, women’s monasteries in Germany are noteworthy for their active role in book production, but limited historical information is available for female scribes and the books they produced between the Anglo-Saxon missionary period beginning in the 8th century and the great monastic expansion of the 12th and 13th centuries (10-12). In Germany, only five women’s scriptoria are known to have been active in the 8th through the 11th centuries (11), and it is likely that most women’s book production during this period was produced more informally by individual female scribes who left few records of their work.
As was the case for many early women’s religious communities, Dalheim has left very few traces in the historical record. No books survive from the monastery, either from its libraries or in any other surviving works. Nearly invisible in the historical record, the women of Dalheim are known to us today nearly exclusively through the archaeological record and a handful of brief textual references (18, 19). The case of Dalheim raises questions as to how many other early women’s communities in Germany, including communities engaged in book production, have been similarly erased from history. Archaeological research, in combination with emerging techniques for the recovery of microremains from dental calculus and analytical methods such as SEM-EDS and micro-Raman, offers great promise for illuminating the lives of the modest and pious women who quietly produced the books of medieval Europe.
- M. M. P. Merrifield, Medieval and Renaissance Treatises on the Arts of Painting: Original Texts with English Translations (Courier Corporation, 1999).
- M. Clarke, The Art of All Colours: Mediaeval Recipe Books for Painters and Illuminators(Archetype Publications Ltd., 2011).
- G. Frison, G. Brun, Lapis lazuli, lazurite, ultramarine ‘blue’, and the colour term ‘azure’up to the 13th century. J. Int. Col. Assoc. 16, 41–55 (2016).
- R. Fuchs, D. Oltrogge, in Europäische Technik im Mittelalter 800 bis 1200—Tradition und Innovation: Ein Handbuch, U. Lindgren, Ed. (Gebr. Mann Verlag, 2001), pp. 435–450.
- J. J. G. Alexander, Medieval Illuminators and Their Methods of Work (Yale Univ. Press, 1992).
- A. Coccato, L. Moens, P. Vandenabeele, On the stability of mediaeval inorganic pigments: A literature review of the effect of climate, material selection, biological activity, analysis and conservation treatments. Heritage Sci. 5, 12 (2017).
- J. Wyart, P. Bariand, J. Filippi, Lapis lazuli from Sar-e-Sang, Badakhshan, Afghanistan. Gems Gemmol. 17, 184–190 (1981).
- J. Kirby, S. Nash, J. Cannon, Trade in Artists’ Materials (Archetype Books, 2010)
- D. V. Thompson, The Materials and Techniques of Medieval Painting (Courier Corporation, 1956), vol. 327.
- A. I. Beach, Women as Scribes: Book Production and Monastic Reform in Twelfth-Century Bavaria (Cambridge Univ. Press, 2004).
- C. J. Cyrus, The Scribes for Women’s Convents in Late Medieval Germany (University of Toronto Press, 2009).
- F. Lifshitz, Religious Women in Early Carolingian Francia: A Study of Manuscript Transmission and Monastic Culture (Oxford Univ. Press, 2014).
- A. Radini, “Particles of Everyday Life: Past Diet and Living Conditions as Evidenced by Micro-debris Entrapped in Human Dental Calculus, A Case Study from Medieval Leicester and Surrounding,” PhD thesis, University of York (2016).
- C. Warinner, C. Speller, M. J. Collins, A new era in palaeomicrobiology: Prospects for ancient dental calculus as a long-term record of the human oral microbiome. Phil. Trans. R. Soc. B 370,20130376 (2015).
- R. Fuchs, D. Oltrogge, in Blau: Farbe der Ferne, H. Gercke, Ed. (Verlag Das Wunderhorn, 1990).
- C. Warinner, J. F. M. Rodrigues, R. Vyas, C. Trachsel, N. Shved, J. Grossmann, A. Radini, Y.Hancock, R. Y. Tito, S. Fiddyment, C. Speller, J. Hendy, S. Charlton, H. U. Luder, D. C. Salazar-García, E. Eppler, R. Seiler, L. H. Hansen, J. A. S. Castruita, S. Barkow-Oesterreicher, K. Y. Teoh,C. D. Kelstrup, J. V. Olsen, P. Nanni, T. Kawai, E. Willerslev, C. von Mering, C. M. Lewis Jr.., M. J.Collins, M. T. P. Gilbert, F. Rühli, E. Cappellini, Pathogens and host immunity in the ancient human oral cavity. Nat. Genet. 46, 336–344 (2014).
- M. I. Hofmann, T. Böni, K. W. Alt, U. Woitek, F. J. Rühli, Paleopathologies of the vertebral column in medieval skeletons. Anthropologischer Anzeiger 66, 1–17 (2008).
- W. Niemeyer, Gründung und Erinnerung in der Geschichte des Klosters Dalheim. Mitteilungen der Deutschen Gesellschaft für Archäologie des Mittelalters und der Neuzeit 27, 215–226(2014).
- R. Pieper, Kloster Dalheim: Eine kurze Geschichte (Landschaftsverband Westfalen-Lippe, 2000).
- I. Osticioli, N. F. C. Mendes, A. Nevin, F. P. S. C. Gil, M. Becucci, E. Castellucci, Analysis of natural and artificial ultramarine blue pigments using laser induced breakdown and pulsed Raman spectroscopy, statistical analysis and light microscopy. Spectr. Acta Part A Mol. Biomol. Spectr. 73, 525–531 (2009).
- P. Ballirano, A. Maras, Mineralogical characterization of the blue pigment of Michelangelo’s fresco “The Last Judgment”. American Mineralogist 91, 997–1005 (2006).
- B. Lafuente, R. T. Downs, H. Yang, N. Stone, in Highlights in Mineralogical Crystallography, T. Armbruster, R. M. Danisi, Eds. (W. De Gruyter, 2015), pp. 1–30.
- A. Wang, J. J. Freeman, B. L. Jolliff, Understanding the Raman spectral features of phyllosilicates. J. Raman Spectr. 46, 829–845 (2015).
- F. Peeck, Die Reinhardsbrunner Briefsammlung (MGH Epistolae Selectae, 1952).
- C. D. A. Cennini, The Craftsman’s Handbook: The Italian “Il Libro dell’ Arte”, translated by D. V. Thompson Jr. (Dover,1954).
- C. J. Duffin, The pharmaceutical use of Lapis Lazuli in the Ancient East. Pharmaceutical Historian 44, 84–87 (2014).
- T. A. Osbaldeston, R. Wood, The Herbal of Dioscorides the Greek (Ibidis Press, 2000).
- E. Lev, Drugs held and sold by pharmacists of the Jewish community of medieval (11–14th centuries) Cairo according to lists of materia medica found at the Taylor–Schechter Genizah collection, Cambridge. J. Ethnopharmacol. 110, 275–293 (2007).
- F. E. Glaze, Salerno’s Lombard Prince: Johannes ‘Abbas de Curte’ as Medical Practitioner.Early Science and Medicine 23, 177–216 (2018).
- J. M. Riddle, Marbode of Rennes’ (1035-1123) De lapidibus considered as a medical treatise with text, commentary and CW King’s translation, together with text and translation of Marbode’s minor works on stones. Sudhoffs Archiv 20, 1–144 (1977).
Originally published by ScienceAdvances 5:1 (01.09.2019), DOI:10.1126/sciadv.aau7126, under the terms of a Creative Commons Attribution-NonCommercial 4.0 International license.