Animal fibre use in the Keriya valley (Xinjiang, China) during the Bronze and Iron Ages: A proteomic approach

Highlights

• A large corpus of archaeological hides and textiles is identified by proteomics.

• New peptide markers for keratin differentiation are proposed.

• The study reveals a choice of the raw material used in the fabrication of textiles.

Abstract

Textile technology strongly advanced with sedentism and pastoralism. During prehistory, many populations settled in central Eurasia, a place of extensive exchange and cultural contact. In the Taklamakan desert, the dry climate enabled good preservation of ancient textiles. The study presented here aimed to identify animal fibres from Bronze Age and Iron Age sites in the Keriya valley (Xinjiang, China) using proteomics. A large corpus of 109 keratin extracts obtained from raw fibres or textiles was analysed, enabling us to establish a corrected and improved list of peptide markers for the identification of species, not only among the family Bovidae, but also for camels and humans. In total, we were able to identify 97% of the sampled objects to the taxonomic level of tribe and 85% of caprines to the level of genus. The assemblage was dominated by hair of sheep (57.8%) followed by goat (16.5%), cattle (8.3%), camel (0.9%), human (0.9%) and non-differentiated Caprinae (sheep or goat) (12.8%). The study showed a continuity between the two sites in this respect. It revealed a choice of raw material linked to the function of the textile, with most woven textiles being made from sheep's wool and most pelts being obtained from goat. Comparison with the bone assemblage of one of the sites provided insight into the herd management strategies. The results confirm the heuristic potential of the proteomic approach for the determination of archaeological fibres and for textile studies in general. Data are available via ProteomeXchange with identifier PXD012189.

Introduction

Textile technology played a central role in the development of past societies (Good, 2001; Hardy, 2007). The first archaeological evidence for pelts dates back to 90,000–80,000 years ago, whereas the first evidence for twisted fibres appears as clay impressions 27,000 years ago (Hardy, 2007). The invention of looping and weaving opened up many possibilities for the creation of textile products, from nets to clothing. Starting in the Holocene, domestication and agriculture enabled the increase of textile production by supplying animal or vegetal fibres (Bender Jørgensen et al., 2018; Rast-Eicher, 2005). Therefore, the study of textiles can provide much information about resources, manufacturing techniques, trade, and culture through time. In contrast to other materials, such as ceramics, bone, or metal, textiles are very sensitive to degradation and are rarely preserved in archaeological contexts. Preservation occurs in specific environments, such as deserts, bogs, or permafrost (Gleba, 2011; Good, 2001; Hardy, 2007; Strand et al., 2010).

The preserved fibres are generally studied from the point of view of technology (of weaving, dyes, etc.), but their specific origin (vegetal or animal) is rarely studied in depth. A taxonomic determination of the animal or plant species at the origin of the textile is often made, but the rarity of the remains does not allow questions related to either the production methods of these fibres (organisation of livestock or crops) or the cultural choices underlying the activities to be addressed. Such questions are addressed using information from animal bones and teeth (often more numerous) or plant remains (generally rarer) found in archaeological sites, but it is difficult to use these proxies to define in detail the types of textiles produced and their use. When there are many textiles on a site, we can increase the number of taxonomic identifications and directly acquire information on the animals, plants, herds, and fields that are at their origin.

Generally, fibre identification is undertaken by microscopy. The morphological observation of animal fibres by optical and scanning electron microscopy normally permits specialists to clearly separate animal families. To achieve this, they focus on different parameters, such as the colour and diameter of the fibre, the morphology of the scales, and the structure of the medulla (Houck, 2010; Thomas et al., 2012; Rast-Eicher, 2016). However, alteration of the fibre due to diagenesis can make these techniques inapplicable to archaeological remains. In addition, these techniques cannot always differentiate between certain related species, such as sheep and goat. More importantly, wool characteristics have changed through time, so the general morphology of ancient fibre may not correspond to that of modern wool from the same species (Rast-Eicher, 2016). For any one of these reasons, the differentiation of archaeological fibres can be very challenging.

Palaeoproteomics has recently emerged as a powerful method to characterise the fraction of proteins preserved over time, providing a molecular signature related to the nature and taxonomy of archaeological samples (Cappellini et al., 2014; Hendy et al., 2018). This approach involves the hydrolysis of protein extracts, followed by analysis by mass spectrometry. It can generate diagnostic peptides that help discriminate among animal species, through their m/z value and fragmentation pattern related to their amino acid sequence (Cleland and Schroeter, 2018). The main proteins present in hair and fur are keratins. They constitute a large family of proteins forming heteropolymeric filaments (Plowman, 2007; Schweizer et al., 2006) incrusted into a matrix of keratin-associated proteins (KAPs) that stabilise the filament structure through extensive disulphide bonding (Gong et al., 2012; Marshall et al., 1991). Keratins are classified into two families: type I keratins (40–55 kDa), which are acidic, and type II keratins (55–70 kDa), which are basic or neutral. The revised nomenclature of mammalian keratins names type I keratins from K31 to K40 and type II keratins from K81 to K87 (Schweizer et al., 2006). Keratin K33 generally exists as two isoforms, named K33a and K33b. However, the annotation of keratins in databases is often heterogeneous or related to previous classifications. Thus, wool microfibrillar keratins are often named based on the old nomenclature, which relies on electrophoresis separation: components 8C-1, 8C-2, 8A and 8B for type I and components 5, 7A, 7B, and 7C for type II (Plowman, 2003; Miao et al., 2018). A correspondence between the nomenclatures is provided in the Supplementary Material Table S1. Proteomic analysis of hair, fur, and textiles was first conducted on modern samples to assess the quality of textile products (Hollemeyer et al., 2002; Hollemeyer and Heinzle, 2007; Clerens et al., 2010; Plowman et al., 2012, 2018; Paolella et al., 2013; Vineis et al., 2014; Li et al., 2016). Analysis of ancient samples in order to characterise remains of Neolithic human hair and clothes (Hollemeyer et al., 2008, 2012; Fresnais et al., 2017) and for other, archaeozoological applications (Solazzo, 2017; Solazzo et al., 2017, 2014, 2013, 2011) has permitted researchers to refine taxonomic attributions, in some cases increasing knowledge of the fabrication of ancient textiles (Fresnais et al., 2017; Solazzo et al., 2011). The high sequence similarity between keratins belonging to the same type and between proteins of closely related species makes keratin identification difficult (Plowman, 2007). Nevertheless, diagnostic peptides have been proposed to differentiate between related animal species, such as a fragment of K33, which differs by only one amino acid between goat and sheep (Solazzo et al., 2014, 2013). These markers are not always detected in proteomic studies of archaeological fibres, which leaves many unidentified samples. New markers and identification keys are essential for the improvement of archaeological fibre identification.

Here we present the results of a proteomic study aimed at identifying ancient fibres from the Xinjiang region at the species level. Xinjiang is an arid region of northwestern China particularly rich in well-preserved organic remains, including textiles and furs (Wang, B., 1999; Good, 1998; Keller et al., 2001; Liu et al., 2011; Mallory and Mair, 2000; Zhao, 2004, 2002, 2001, 1999).

Located in the eastern part of Central Asia, Xinjiang was a nodal point within the large Eurasian sphere of contact and interaction known as the Silk Roads, and within their forerunners. Many sites from the Bronze Age and Iron Age have been discovered so far (Chen and Hiebert, 1995; Jia et al., 2009; 2010), the oldest dating to around 2500-2000 BCE. Isotopic analyses on vegetal remains made it possible to enlighten us on human diet, subsistence strategies, and cross-cultural contacts from the Late Bronze Age to early historic times, both in northern and in southern Xinjiang (Wang et al., 2017a . Wang et al., 2017b). However, the textile remains found to date come mainly from eastern Xinjiang (the Wupu cemetery, near Hami); from the Lopnor region (including the Xiaohe and Gumugou cemeteries) (Li, 2007; Liang et al., 2012; Qiu et al., 2014; Qu et al., 2017; Wang, 2014; Yang et al., 2014a) and from the Tarim Basin, in both the Turfan region (including the Yanghai and Subeixi cemeteries) and the southern part of the Taklamakan desert (notably the Keriya Valley and the Chärchan area) (Wang, B., 1999; Good, 1998; Wang et al., 2016).

We focused on two sites in the Tarim basin (southern Xinjiang), where large corpuses of textiles and furs have been unearthed: Djoumboulak Koum (also called Yuansha gucheng), which is an Iron Age fortified settlement, and the Northern Cemetery, which dates from the Bronze Age (Fig. 1) (Debaine-Francfort and Idriss, 2001; Debaine-Francfort, 2013). Discovered by the Sino-French Archaeological Mission in Xinjiang, both are located in the now-dry protohistoric delta of the Keriya River, which crosses the Taklamakan desert from south to north. Both have yielded large corpuses of textiles and furs that are being studied in a multidisciplinary way. Our research not only aims to discriminate between native and exogenous fauna, it also focusses on the methods used to process the fibres and initiates wider research on the evolution of fleeces and textiles from the Bronze Age to the Iron Age and a comparison with contemporaneous fleeces and textiles found in Europe. This information is cross-referenced with that provided by archaeozoology to supplement our knowledge on the history of breeding, transhumance practices, and hunting. This work, which has already revealed the complexity and diversity of weaving techniques and dying practices, suggests that the textile industry had a major importance in the ancient societies of the Keriya valley (and beyond) (Cardon et al., 2013; Debaine-Francfort and Idriss, 2001). The often poor state of preservation of the fibres, which is sometimes insufficient to enable their taxonomic determination using microscopy, has led us to adopt non-zooarchaeological analytical methods to overcome this taphonomic handicap and to thus address unanswered questions about the raw materials used in the manufacture of these textiles.

In this article, we report on how a proteomics approach was used on raw fibres and textiles in order to identify the origin of animal fibres from these two archaeological sites. The taxonomic specificity of keratins was evaluated on a large corpus of samples, with the aim to define new peptide markers for species differentiation. The fibre identification was used to figure out the relation between domestic animal species and type of textile produced and to document the diachronic development of textile use in Xinjiang from the Bronze Age to the Iron Age.