For further information about the research summarized here, please refer to Tehrani and Collard (2009) and Tehrani et al. (2010).
The evolution of rug weaving is a classic example of the Problem of Missing Links. The craft is practiced throughout the Middle East and Central Asia, but the poor preservation of textiles over time means that very little is known about its origins and development. It is likely that rug weaving was developed by nomadic pastoralists, who have easy access to raw materials such as wool from their sheep and goats and the ingredients for dyes extracted from wild plants and insects. Furthermore, a textile-based material culture is well adapted to the tribes’ mobile lifestyle since, unlike items made from wood or clay, textiles can be folded and rolled. Today, tribal women continue to manufacture a wide range of items, from the tents they live in to colourful rugs and bags for storing and transporting goods between winter and summer pastures. Each tribe has its own distinctive weaving style, although many designs and techniques are shared by different groups. It has often been claimed that these similarities can be traced back to common ancestral tribes. However, this hypothesis needs to be tested against the alternative possibility that contact and exchange among groups might have led to borrowing and blending among their weaving traditions.
Iranian textiles present an especially interesting context for investigating the phylogenesis/ethnogenesis problem because of the ways in which craft knowledge is transmitted among weavers. Between 2001 and 2003, I spent 6 months living among the tribes and interviewed over 60 weavers (all of whom were women, since weaving is an exclusively female activity) about how they learned to weave. The interviews revealed important differences in the ways that techniques and designs are transmitted. Techniques are almost always passed on from mother to daughter, usually at a young age (between 9 and 14 years old), over a period of several years. During their apprenticeship, young weavers also build up a repertoire of designs by collaborating with and imitating their mothers. However, whereas adult weavers rarely acquire new techniques once they begin to work independently, they frequently copy designs from their peers. We can therefore hypothesize that while the transmission of weaving techniques follow similar pathways to the transmission of genes (i.e. they are transmitted “vertically” between generations), the transmission of designs is likely to be much more complicated (since they can be transmitted “vertically” between generations and “horizontally” within generations).
To test this hypothesis, my colleagues and I used a phylogenetic technique known as cladistics. Cladistic analysis focuses on variation in the constituent parts, or “characters,” of a group of taxa. In biological species, characters may comprise DNA sequences or morphological traits. In languages, characters are usually based on lists of words, such as core vocabulary items. In the case of material culture, characters consist of stylistic and/or technological elements of assemblages, such as aspects of arrowhead design (O’Brien and Lyman 2003), forms of musical instruments (Temkin and Eldredge 2007), or, in this case, variations in textile ornaments and knotting techniques. It is worth emphasizing that the lack of naturally bounded units in material culture does not undermine the applicability of phylogenetic methods. As O’Brien et al. (2001) have pointed out, “whether a tooth represents one or multiple genes—replicators—is as yet unknown, but this does not hinder the efforts of palaeobiologists to determine and explain the evolutionary histories of the organisms whose phenotypic hard parts they study….Cultural traits conceived as ideas held in the mind of individuals are the replicators that are transmitted…If there is phenotypic change, and if over time enough variation is generated, cladistical analysis might indeed be able to detect the phylogenetic signal” (p.1134).
Cladistic analysis reconstructs relationships among taxa or classes by distinguishing characters that are evolutionarily novel (also termed apomorphic or derived), from those that were present in the last common ancestor of all the taxa under study, which are labelled ancestral or plesiomorphic. The presence of a derived trait in two or more taxa provides evidence that they are descended from a common ancestor of more recent origin than the ancestors they share with the other taxa under analysis. There are several methods to identify which traits are derived and which are ancestral, the most popular of which is outgroup analysis. An outgroup is defined as a taxon that shares a common ancestor with the taxa under analysis (the ingroup), but is of more distant origin than the ancestor the analyzed taxa share with each other. Since the outgroup does not share an exclusive common ancestor with any individual member of the ingroup, it follows that when a character occurs in two states among the study group, but only one of the states is found in the outgroup taxon, the former is considered the derived state and the latter the ancestral state.
Once the direction of change has been established for each character, the next step in a cladistic analysis is to construct a branching diagram that connects taxa according to their relative derived status. This diagram is known as a character cladogram. An example of a character cladogram is shown in Fig. 3 which concerns variations in a type of carpet ornament called a gul: the shape of the ornament is similar in all the taxa, but there are several differences in the interior design. In the outgroup taxon and taxon A, we can see what appear to be darts or birds protruding from the heart of the gul. In the three remaining taxa, these take a different form—that of clovers. Since the dart/bird form is found in the outgroup and the clover form is found only in the ingroup, we can infer that the clover evolved subsequent to the last common ancestor shared by the ingroup. In other words, the presence of the clover design provides evidence that taxa B, C, and D share a common ancestor that is not shared with taxon A. Studying the gul of these three taxa more closely, we can see that it is possible to make further distinctions. Thus, in the case of taxa C and D, the clover is divided into two stems; whereas in taxon B, the clovers have only one stem. Again, this suggests that the clover design has evolved in two forms. If we assume that C/D form is derived with respect to the B form, then this would imply that they share a common ancestor that is not shared with taxon B (although it should be noted that it is equally possible that the B form is derived, in which case we cannot be sure that C and D are more closely related to one another).
If descent and modification were the only cause or source of similarities among taxa, then all the character cladograms would be compatible with one another. Normally, however, a number of the character cladograms will suggest relationships that are incompatible because, as noted earlier, common descent is not the only source of similarity among taxa. How can we sort true family resemblances (known in phylogenetic terms as homologies) from similarities resulting from other processes such as independent evolution and borrowings (homoplasies)? The cladistic approach deals with this problem by generating an ensemble or consensus cladogram that is consistent with the largest number of characters and therefore requires the smallest number of evolutionary changes to account for the distribution of character states among the taxa. This approach is based on the principle of parsimony, the methodological injunction that explanations should never be made more complicated than necessary. Characters that are consistent with the ensemble cladogram can then be classified as homologous (i.e. similarities due to common descent) while those that are inconsistent with it can be classified as homoplastic (i.e. similarities that are due to other processes, such as borrowing and blending among lineages).
If it is true that horizontal transmission leads to complex and tangled patterns of cultural evolution, then, based on the description of craft learning above, we would predict that similarities among the designs used by Iranian tribes would be much more homoplastic than similarities among their techniques. This is because designs are transmitted both vertically and horizontally, whereas techniques are only usually transmitted vertically. We tested this prediction through a cladistic analysis of 122 decorative and technical characters from six tribal groups, the Yomut, Shahsevan, Qashqai, Boyer Ahmad, Papi and Bakhtiari. I employed a prehistoric Western Asian textile assemblage (from the Pazyrk Valley, Siberia, fourth to fifth century BCE) as an outgroup (Fig. 4). The first stage of the analysis inferred the most parsimonious tree for the textile assemblages. The second stage of the analysis examined how well decorative characters (n = 80) fitted the tree compared to the technical characters (n = 42). The goodness-of-fit for each set of characters was measured using the Retention Index (RI), which calculates the number of homoplastic changes a cladogram requires that are independent of the number of characters in the data (for a more detailed description, see Farris 1989).