Wari: Imperialism, Low Power, and Globalization in the Middle Horizon Central Andes

Abstract

Wari is sometimes described as the first empire of the Andes, conquering and controlling a broad region during the Middle Horizon (600–1000 CE). This article synthesizes archaeological research to offer a new perspective on Wari’s rise, expansion, and collapse. Wari emerged in a rapidly urbanizing environment as a set of ideas about the world and how it should work that blended foreign ideas with local traditions. Heartland cities were organized around elite kin groups who competed for followers by hosting small-scale gatherings. Wari-related ideas, objects, and people circulated far more widely, creating a dynamic cultural horizon of considerable heterogeneity. Efforts to centralize decision making in the ninth century CE may have led to the polity’s decline. Although this reconstruction of Wari politics differs from previous models, it is in keeping with contemporary interpretations of collective and low-power early expansive polities in other parts of the world.

Introduction

The Middle Horizon (600–1000 CE) was a significant turning point in central Andean history. After almost a millennium of more limited interactions, much of the region was entangled by a wide circulation of objects, ideas, and, to a lesser extent, people. Regional traditions that had been largely separate came together, creating connections that would grow stronger in the centuries that followed. From social organization (Jennings and Berquist 2023) to funerary practices (Isbell 1997a) and metallurgy (Lechtman 1980, 2003), the Middle Horizon spurred the development of more common ways of living across the central Andes, as well as more robust long-distance interaction networks. Critical aspects of the central Andean society that the Spaniards encountered in 1526 CE emerged during the Middle Horizon.

For those unfamiliar with the region, these networks penetrated coastal deserts, high mountains, and verdant jungle slopes (Kennan 2000; Thouret et al. 2009). There was no wheeled transport in the ancient Andes, and, even if there was, no domesticated animals that one could yoke or ride. Most rivers were not navigable. The sustained interregional relationships of the Middle Horizon were thus a monumental achievement of human ingenuity and perseverance. Huari was the predominant site in the Middle Horizon central Andes, a sprawling city in the highland Ayacucho Valley that housed about 60,000 people by the seventh century CE (Isbell 1997a). The art styles that coalesced in and around the city are also known as Wari—eponymously named but spelled differently here for clarity—as is the polity that grew around the city (Fig. 1). The geographic spread of Wari-style ceramics, textiles, and stonework would be used to define the Middle Horizon (Isbell 2009), with scholars later documenting at least 20 sites across the central Andes that feature Wari-style architectural elements like long corridors, interior patios, and, on occasion, D-shaped structures (Jennings and Craig 2001; McEwan and Williams 2012).

Fig. 1

Landsat image of the central Andes showing the approximate spatial extent of Wari influence and some of the sites mentioned in the text (base map from Google Earth)

Today, almost all scholars agree that Wari (and Huari) played a pivotal role in creating the Middle Horizon (e.g., Bergh 2012; Ochatoma Paravicino and Cabrera Romero 2021; Watanabe 2019). Yet, there remains significant disagreement on how best to define the era: Was Wari part of a greater Middle Horizon interaction sphere (MacNeish et al. 1975), the instigator of an era of early globalization (Jennings 2010a), or, as most would characterize it, an empire that conquered and controlled a large part of the central Andes (Schreiber 1992)? Offering a synthesis of research to date, I argue that Wari has elements of all three models. The limited extension of Wari sovereignty abroad had sweeping consequences on local populations. Some of it was direct—a colony inaugurating an ambitious agricultural project (Nash and Williams 2009)—while most of the time the impacts were more down-the-line and subtle.

Wari was an empire, in the sense of an expansionist polity that encompassed and rearranged a diversity of populations (after Burbank and Cooper 2010, p. 8; Pitts 2010, pp. 213-214). Wari’s leaders had ambitions of regional domination, and many people in foreign lands would come to identify themselves as part of a broader Wari community. Dreams of domination, however, did not come to pass. Central Andean topography and cultural diversity, in part, dissipated imperial ambitions. Yet, more importantly, Wari governance was built on repeated, small-scale, face-to-face contacts that shaped the kinds of interactions that could unfold. The result was a dynamic network of settlements on trails that snaked across the Andes, with each place associated in different ways with local groups and the Wari heartland. Space constraints limit what can be covered in this article; those seeking a history of Wari scholarship, for example, should look elsewhere (Bergh and Jennings 2012; Glowacki 1996; Isbell and McEwan 1991). My focus is on Wari origins, its governance, the globalizing impacts of colonization, and, woven throughout these discussions, on how Wari relates to more collective and low-power models of early expansionary polities in other regions of the world (e.g., Ando and Richardson 2017; Thurston and Fernández-Götz 2021).

Wari was the culmination of millennia of cultural development. The first mobile hunter-gatherers pioneered transportation routes through the central Andes some 14,000 years earlier (Dillehay et al. 2017). Monumental sites, like Caral, were first built in the fifth millennium BCE (Shady Solís 2015). Social stratification and craft specialization then emerged by at least the Formative Period (1200–200 BCE), with the last centuries of the era known for its shared religious iconography and wide circulation of goods (Burger et al. 2020). The Early Intermediate Period (200 BCE–CE 600) that followed witnessed the emergence of more regionally organized polities like Moche (Bourget 2016) and Nasca (Silverman and Proulx 2002) that were often predicated on violence, the display of prestige goods, and radical genealogical distance between commoners and elites. Wari is another chapter in this rich tradition of societal creation, initiating an era of unprecedented interconnectedness. Their intent was imperialistic—few doubt that warriors marched out of Huari to conquer enemies abroad—but much of the end result was accidental: the Middle Horizon was the unforeseen consequence of an early experiment in urbanism and political expansion.

Wari Origins

Systematic surveys since the 1970s have documented as many as 300 Early Intermediate Period sites in the Ayacucho Valley (Benavides 1976; MacNeish et al. 1981; Vivanco Pomacanchari and Valdez 1993). Often located close to rivers and springs, the sites tend to be accumulations of a few dozen houses—villages surrounded by irrigated agricultural terraces. Some of these settlements were established in the preceding Formative Period (Ochatoma Paravicino 1992, 1998; Pérez Calderón 2013, 2017), such that villagers maintained an agropastoral lifestyle over two millennia based on a range of domesticated plants (such as maize, beans, peppers, and potatoes) and animals (guinea pigs, llamas, and alpacas) (Leoni 2004, 2013; Lumbreras 1959, 1974b; MacNeish et al. 1981; Ochatoma Palomino 2011; Ochatoma Paravicino and Cabrera Romero 2010a; Pérez Calderón and Carrera Aquino 2015).

Dispersed across the valley, there are no significant differences in village size and, until at least the latter half of the Early Intermediate Period, homes tended to be spaced far apart with no evidence for settlement planning (Isbell 2001; Machaca Calle 1991). Houses were often single-room circular or rectangular structures, sometimes with an attached external patio where many household activities occurred. Archaeologists have found few potential status goods, only a handful of extended family compounds, and no plazas or public buildings (e.g., Vivanco Pomacanchari 2011; Vivanco Pomacanchari et al. 2003). Rather than collective burials, tombs tended to be single inhumations located in or around the home (Pérez Calderón 2019a).

The valley was also home to ceremonial centers like Wichqana (Flores Espinoza 1960; Lumbreras 1974b, 1981) and Chupas (Lumbreras 1974b) that drew visitors from many villages. These sites, like the villages that surrounded them, tended to be used for hundreds of years, with their platforms, plazas, and other monumental features always under construction. The result was a palimpsest of architectural choices that blended foreign influences with local traditions. Differences in ceremonial architecture across space and time in Ayacucho are striking, suggesting that sites competed for followers by offering a range of ideologies and practices (Cavero Palomino and Huamaní Díaz 2015; Leoni 2004, 2006; Pérez Calderón and Paredes 2015; Vivanco Pomacanchari and Mendoza 2015).

Ayacucho’s architecture and ceramics were significantly shaped by outside ideas in the Early Horizon (700–400 BCE) when Chavín and Paracas influences are clearly discernible (Lumbreras 1958, 1959). Engagement with the foreign by Ayacucho residents, however, waned after the Early Horizon through the first half of the Early Intermediate Period. The ongoing circulation of obsidian from the nearby Quispisisa source suggests that long-range interaction networks continued (Burger et al. 2000; Tripcevich and Contreras 2011)—the valley sits within a natural transportation corridor that both connects the Pacific to the Amazon and runs along the spine of the Andes—but archaeologists can detect little interest in infusing the wider world into the material culture of Ayacucho for almost a millennium.

There was a succession of overlapping pottery styles in the Formative and Early Intermediate Period, such as Aya Orqo, Wichqana, and Rancha (González Carré 1967; Lumbreras 1959). The best studied of these is Huarpa, a style featuring black on white geometric designs (Lumbreras 1974b; Menzel 1964). The style can be subdivided into an early (200–400 CE), middle (400–550 CE), and late phase (550–700 CE) (Knobloch 1983, 2013, 2023). Correlating these phases with ceramic survey data indicates that the valley’s population was increasing throughout the Early Intermediate Period (Isbell 2001; Valdez and Valdez 2021; Pérez Calderón 2017, 2019b). More people in the valley meant bigger and more numerous villages, with communities likely fissioning when they got too large (e.g., Bandy 2004). Over time, this process led to the valley’s transformation into a more anthropogenic landscape filled with settlements, fields, canals, and pathways.

By at least Late Huarpa, we also see more experimentation with new village lifeways. There is a compound at Tunasiypoq, for example, featuring four structures set around a central patio (Lumbreras 1974b), and a D-shaped structure at Huanca Qasa (Huamán Lira 2006). These changes may reflect the need to strengthen kin groups or other kinds of intermediate social organizations within expanding villages. At the same time, Ayacucho’s ceremonial centers were attracting larger, more permanent, populations, as well as incipient leaders who sought to separate themselves from the rest of the population through their roles in rituals, more ornate housing, and privileged access to foreign ideas (Isbell 2001; Pérez Cálderon 2017, 2019b; Valdez and Valdez 2021).

By far, the best understood of these burgeoning ceremonial centers is Ñawinpukyo, a mountain cult center with an emphasis on communal feasting (Fig. 2) (Leoni 2004, 2006, 2009). Located on a 500-m-long hill, Ñawinpukyo is capped by a large enclosure dubbed the East Plaza. The enclosure wall may have been as high as 2 m and had a single entrance in the southwest corner. Inside was a large plaza featuring a “circular temple” composed of three concentric walls (Leoni 2006, p. 286). The building’s lone entrance aligns with the snow-capped peak of Rasuwillka, while its interior may have been dedicated to food production. The handful of other buildings that fringed the plaza are also places where meals were prepared.

Fig. 2

Site plan of Ñawinpukyo, with most buildings probably dating toward the end of the Early Intermediate Period (courtesy of Juan Leoni)

Leoni (2006, p. 294) has identified an expansive Late Huarpa building 70 m west of the plaza as an “elite compound.” The function of the building remains unclear—it was heavily remodeled during the Middle Horizon—but its thick plaster and well-prepared floors set the structure apart from the more ephemeral housing elsewhere at the site that may have been erected during periodic visitations. The compound’s association with only Late Huarpa ceramics indicates it was added after Ñawinpukyo’s Middle phase rise to prominence. In the early Middle Horizon, the structure burned down, and the East Plaza was abandoned. Builders then remodeled the compound, creating the gallery and patio organization that was becoming associated with Wari governance (see next section).

Huari was another gathering place that was first used during the Formative Period. Research at the site has uncovered pockets of Huarpa occupations that may represent distinct sites (Bennett 1953; Cabrera Romero et al. 2021a; Cavero Palomino and Huamaní Díaz 2015; Knobloch 1976; Meddens 1978; Ochatoma Paravicino et al. 2015). Some of these sites had ceremonial structures that could host a couple dozen people, such as a rectangular columned, heavily plastered building excavated by Ochatoma Paravicino and colleagues (2015). This building, like others excavated in the same part of the site, was infilled with another ritually charged structure built directly above.

The Late Huarpa changes in Ayacucho were fueled by rising interregional engagement (Valdez 2017). Knobloch has linked changes in Middle and Late Huarpa iconography to influence from Cuzco, the Amazon, and, especially, Nasca (1976, 1983, 2023). Nasca’s influence intensified during Late Huarpa, with Huarpa motifs, pigments, and forms mimicking coastal examples (Doi et al. 2017; Knobloch 2005; Menzel 1964; Muro Ynoñán et al. 2023). Nasca potters and their families may have migrated to Ayacucho, and there are some indications of Ayacucho colonists in the Nasca Valley (Conlee 2021). The new motifs on Late Huarpa pottery indicate an increasing familiarity with such practices as Amazonian shamanism (Knobloch 2023) and Nasca trophy head taking (Menzel 1964). These foreign ideas entered into a turbulent political landscape of incipient elites and were likely used as both a wedge to heighten social distinctions and a mechanism of social cohesion.

At the same time, people were learning to live in larger groups and negotiate resource rights in a landscape crowded with anthropogenic features. These changes alone, however, do not explain why people began congregating into only a few settlements in the valley. Valdez and Valdez (2017, 2020, 2021) have argued that rising violence spurred aggregation into hilltop or walled settlements, and we see depictions of such settlements on a few ceramic models from the period (Pérez Calderón 2016). A Late Huarpa cache of eight fragmented human crania from Ñawinpukyo that were drilled, cut, and scraped has been interpreted as trophies from raiding (Finucane 2007, 2008).

By the end of Late Huarpa, Ñawinpukyo, Conchopata, and Huari were rapidly expanding as other settlements were abandoned (Pérez Calderón 2001; Valdez and Valdez 2020). The centers’ frenetic growth was unplanned with homes and temples quickly filling empty spaces (Isbell and Vranich 2004). There had been few commitments to highly public and visible celebrations in the fifth century CE—recall even Ñawinpukyo’s plaza was enclosed by a tall wall—and an even greater emphasis on small, enclosed ceremonial spaces emerged as these centers grew. In the hurly burly of early urbanization, experimentation in new ideologies, living arrangements, ceremonial structures, and funerary practices coalesced into a Wari identity (e.g., Cook 1994; Ochatoma Paravicino et al. 2015; Pérez Calderón 2019b; Valdez et al. 2002a, b, 2006).

Perhaps the only significant area of disagreement among archaeologists with the above characterization of life in Ayacucho during the Formative and Early Intermediate Period is the nature of Late Huarpa political organization. Some scholars have argued for a Late Huarpa state (Lumbreras 1974a, b; Pérez Calderón 2016, 2019a); others have posited competing chiefdoms (Isbell and Schreiber 1978; MacNeish 1981). I see little evidence for state organization at this time, at least in terms of thinking of states as regionally organized societies composed of a ruling class, commoner class, and a highly centralized and internally specialized government (e.g., Johnson and Earle 1987; Marcus and Feinman 1998). Chiefdoms are perhaps more plausible, but here I follow those scholars who conceive of sites like Ñawinpukyo more as ceremonial centers with little to no control over village life (Knobloch 2023; Leoni 2006; Valdez and Valdez 2020). As discussed in the next sections, the political centralization assumed for both states and chiefdoms was an aborted development centuries later in the late Middle Horizon.

I suggest that Late Huarpa is best characterized as a period of inchoate social stratification, when emerging elites used ceremonies conducted at Ñawinpukyo, Huari, and other centers to engender new kinds of relationships between themselves and the rest of Ayacucho’s inhabitants, both human and nonhuman. These early elites likely aspired to greater power, and, citing public benefits, directed community construction of agricultural infrastructure, temples, and even their own homes. Occurring amid substantial societal change, these political arrangements would have been fragile, highly contested, and likely intentionally ambiguous. More fieldwork is certainly desired on Late Huarpa society, along with many more radiocarbon dates from secure contexts, but the era’s political organization may have been unclear to the people who lived it: who they were and who they followed was a muddled work in progress.

Becoming Wari

Those who moved into Huari, Conchopata, and Ñawinpukyo from surrounding villages would have been hard-pressed to identify when Huarpa turned into Wari. The first styles that archaeologists define as Wari emerged as Huarpa styles were still in use (Cabrera Romero et al. 2021a; Isbell 2019), and much of what was once thought to be new to Wari’s religion, politics, and social structure—D-shaped structures, patio and gallery architecture, collective burials—first appeared by the end of Late Huarpa or, in the case of the Front-faced Deity motifs closely associated with the Tiwanaku polity in Bolivia, decades after Huari’s coalescence. Rather than a moment of change, becoming Wari was an ongoing negotiation within the context of rapid urbanizations, widening interregional interaction, and rising wealth disparities.

Although Ñawinpukyo’s growth stagnated, Huari and Conchopata expanded in a cacophony of construction projects that created a dense warren of dynamic architectural spaces (Fig. 3) (Isbell 2001a, b, 2009; Leoni 2007; Ochatoma Paravicino 2007; Ochatoma Paravicino et al. 2015; Pérez Calderón 1999). From above, the cities would soon look like places of “spatial confusion” marked by “enclosure and redundancy” (Isbell and Vranich 2004, p. 106), a jumble of “repetitive, apartment-like cells” and small open spaces (Isbell 2009, p. 209). A common means of meeting the challenges of rapid urbanization is to increase the size of decision-making units (Jennings 2016), such as by forming extended households or subordinating to more affluent families. Based on excavations at Conchopata and Huari, both strategies were taken in the Wari heartland as residences coalesced around dozens of expansive compounds with internal patios and narrow galleries (Isbell 2001; Ochatoma Paravicino and Cabrera Romero 2007; Young-Wolfe 2023).

Fig. 3

Site plan of the excavated portion of Conchopata (Map prepared by Patricia and Barbara Wolff, courtesy of Barbara Wolff). The site’s organization is similar to that of Huari during the first half of the Middle Horizon

Isbell (2006) refers to these compounds as “palaces,” and Blacker and Cook (2006) call them “lineage houses.” Both terms capture important aspects of these buildings. The plastered, painted, and well-drained homes stood in contrast to the more informal housing around them, creating elite spaces. The compounds were also “lineage houses” that hosted ritually charged feasts celebrating ancestors that, over time, were held in common by elite families and their followers as corporate groups expanded (Nash 2010, 2012; Rosenfeld 2011). Some mummified ancestors were buried in ornate tombs under compound floors. The tombs were occasionally reopened amid more intimate rituals to provide offerings and insert additional burials, with some capstones having a hole where liquids could be poured inside the tomb (Fig. 4) (Milliken 2006; Tung 2012).

Fig. 4

Collective Middle Horizon tomb types found in Wari elite compounds at Huari and Conchopata (courtesy of William Isbell)

The compounds at Huari and Conchopata are reminiscent of the kancha architecture of the Early Intermediate Period northern highlands. The northern compounds, featuring a ring of structures surrounding a central patio, were used for feasts and ancestor veneration among families who lived together when ceremonies were conducted (Berquist 2021; Lau 2002, J. Topic and T. Topic 2001; T. Topic and J. Topic 2010). The Late Huarpa/Wari innovation was to turn the kancha into a full-time elite residence that often housed extended families anchored by a cohort of closely related females (Tung 2012; Tung and Cook 2006). Lower-status families prepared most of the food and drink to be consumed at an event in their homes (Jennings et al. 2022a; Sayre and Whitehead 2017). Additional flourishes, such as serving llama meat in polychrome bowls and adding the psychedelic Anadentathera colubrina to molle beer, then occurred on site (Biwer et al. 2022; Cook and Glowacki 2003; Knobloch 2000). Although the frequency of these events remains unclear, they hinged on the elevation of everyday meals into a communal feast for the dead.

By the seventh century CE, repeated feasting and ancestor veneration appears to have resulted in the creation of a house society (e.g., Beck 2007; Joyce and Gillespie 2000; for Wari see Jennings and Berquist 2023). People thought of themselves as belonging to a corporate group, what I refer to as a “great house” that was anchored by a set of shared ancestors who were sometimes interred at the heart of elite residential compounds. Membership required regular participation in the events that took place within compounds, creating strong bonds via many hours of face-to-face interactions in intimate settings. The political power that emerged for great house leaders was considerable, yet tightly circumscribed—the size of a great house was limited by the number of personal connections that a leader could maintain (for a similar argument on kin-related groups in the Andes, see Isbell 1997a).

Coordination above the level of great house would have been necessary at Huari and Conchopata, and may have taken place within D-shaped structures (see Fig. 3). These buildings became more common in the Middle Horizon, taking on the standardized form of a niched curved wall, a straight wall with a single entrance, and, usually, a small courtyard around the entrance (Bragayrac 1991; Cook 2001; Reid 2023). The D-shaped structures also hosted collective rituals that may have focused more on external affairs (Tung and Knudson 2011; Sayre and Whitehead 2017). Smashed pottery in these contexts, for example, sometimes feature scenes of Wari warriors battling foreigners (Ochatoma Paravicino and Cabrera Romero 2010b) and a cache of trophy heads from one D-shaped structure came from people raised outside Ayacucho (Tung and Knudson 2008, 2010). Some scholars suggest the niches were used to display mummy bundles (Cook 2001, 2015; McEwan 1998), perhaps carried in procession along the corridors that often linked elite compounds to D-shaped structures (Young-Wolfe 2023). Since D-shaped structures tend to be smaller than compound patios, only a few elites from each great house could have entered them, creating a smaller group of deciders for larger-scale collective action.

The elite compounds, and Wari spaces more generally, were a constant work in progress, with the act of building used to create and sustain collectives (Groleau 2009, 2011; Isbell and Groleau 2010). When a hallway was decommissioned in Huari’s Vegachayuq Moqo sector during the Huarpa/Wari transition, for example, celebrants smashed 114 bowls and cups and dumped them into a trench being used for the foundation of a patio connected to a newly built D-shaped structure (Ochatoma Paravicino et al. 2021). The well-worn vessels likely came from the many households whose members contributed labor to the project. Similar actions occurred simultaneously in compounds and smaller homes—every addition of a room or replastering of a floor was an opportunity to define great house membership amid a dynamic social environment.

Like in other cities (Ortman et al. 2015), specialization and heterogeneity increased as the settlements grew. At Huari, some neighborhoods focused on shellwork, lapidary, and other industries (Lumbreras 2010). We know most about the organization of ceramic production at Conchopata (Ochatoma Paravicino and Cabrera Romero 2007; Pozzi-Escot 1985, 1991; Pozzi-Escot et al. 1994). In Late Huarpa, making pots was a cottage industry that took place in most households. Domestic production continued into the Middle Horizon, even as small workshops appeared inside or adjacent to elite compounds (Cook and Benco 2000; Grávalos et al. 2023; Wolff 2012). As with feasting, supporters appear to have assisted in workshop production, but it may have been the elites themselves who took a leading role in the production of large face-necked jars, urns, and other symbolically laden, difficult-to-make vessels that were then used as gifts or for display.

The care taken to kill, bury, and curate these vessels—they were sometimes “killed” with a single blow to the chest of a figure depicted on them—suggests that at least some of these vessels were seen as animate, powerful objects (Fig. 5) (Vazquez de Arthur 2020). The iconography of the earliest styles, such as Okros and Chakipampa, speak to predation, flight, and renewal that built off Huarpa themes (Knobloch 1976). The pots often feature fantastic creatures, such as the Ventral Animal, an undulating, insect-like organism with a triangular head and tail, four splayed legs, and a pointed proboscis (Menzel 1964). These early styles were joined a few generations later by styles like Robles Moqo and Conchopata (the style is named after the settlement) that drew inspiration from the Staffed Deity and Attendants of the Tiwanaku pantheon to the south (Cook 1994; Isbell and Knobloch 2006, 2009; Lumbreras 1960). The styles, at least in part, were contemporaneous, suggesting rival ideologies may have been in play as Huari and Conchopata grew (Isbell 2019).

Fig. 5

Wari face-necked jars from Conchopata. Large urns like these examples typically vary between 1.1 and 1.2 m in height (courtesy of José Ochatoma)

Urbanization also requires the creation of a countryside (Yoffee 2004). Since Huari and Conchopata grew by “virtually inhaling” Ayacucho’s other settlements (Schreiber 2005, p. 265), agropastoral systems developed to meet the needs of village economies needed to be retooled to meet the needs of an urban society. How this transition occurred remains unclear. Much of the valley was terraced and irrigated by Late Huarpa, and a maize-rich diet continued into the Middle Horizon (Finucane 2007, 2009; Lumbreras 1974b). Some infrastructural improvements may date to the Wari era, such as widening and covering of Huarpa-era canals to improve irrigation (Lumbreras 1974b). Yet, most innovations were likely logistical: labor organization, scheduling, fallowing, property rights, and other arrangements had to be renegotiated as people changed where and how they lived (see Bandy 2005 for the Tiwanaku region).

No centralized storage facilities have been found at Conchopata and Huari, and daily food production appears to have taken place in the home. At Conchopata, variation between domestic spaces in the ubiquity of different plant and animal species may speak to households’ idiosyncratic food production regimes (Sayre and Whitehead 2017), a possibility supported by the recent excavations of a compound in the Patipampa sector of Huari where beans dominated the assemblage (Taylor 2018). The presence of hoes and other agricultural implements indicates that at least some city residents went into the countryside to farm, with the Middle Horizon increase in the use of these implements in the valley’s remaining villages suggesting an intensification of agricultural production over other activities (Doi 2019; Ochatoma Paravicino and Cabrera 2001).

Different parts of the Ayacucho Valley are more closely associated with certain Wari ceramic motifs and styles (Anders 1986, 1998; Doi 2019). This raises the possibility that urban neighborhoods were more closely tied to different parts of the valley, reminiscent of how the residents of Teotihuacan’s apartment compounds were linked to disparate areas in central Mexico (Manzanilla 2017; for Tiahuanaco, see Janusek 2002). Middle Horizon terracing and irrigation canals in surrounding valleys suggest that extensification—bringing more distant land into food production (e.g., Styring et al. 2017)—was also required to support Conchopata and Huari residents, and the cities’ increased obsidian demand led to more intensive quarrying of the Quispisisa obsidian source located less than 100 km away (Tripcevich and Contreras 2011).

A surge in long-distance imports also followed Wari urbanization. As Rosenfeld and colleagues (2021) summarize, shells and other marine products came from the Pacific Ocean, hardwoods and coca arrived from the jungle, and minerals and metals were delivered from throughout the central highlands. Neutron activation analysis demonstrates that ceramics were also imported, at least at Huari, from more than 1000 km away (Williams et al. 2019a), with a more circumstantial argument for the mass import of Amazonian parrot feathers (Wilkinson 2018). Representatives of at least some of these import regions were featured on Wari pottery by at least the eighth century (Knobloch 2012), and social network analysis of these depictions demonstrate that these foreign representatives are associated more frequently with certain Wari-affiliated agents (Gibbon et al. 2022); factions within the Wari heartland may have had stronger ties to different regions.

These data can be used to argue for the development of an incipient prestige goods economy by the seventh century CE, wherein foreign goods were exchanged for ceramics, metalwork, textiles, and other objects made by Wari specialists (Earle and Jennings 2012). The foreign goods were then used as status makers by Wari’s burgeoning elite and employed as gifts to their followers who provided the labor and resources that sustained Huari and Conchopata. The intimacies of great house political action were thus primarily funded by an influx of foreign goods. For Wari to survive as a polity, leaders needed to keep these spigots open—the flow of Spondylus shell from Ecuador, for example, could not stop (Paulsen 1974). Maintaining ties to foreign elites was one way to keep the prestige economy going, as was colonization and conquest.

What was Wari? Wari began as an urban phenomenon, a suite of ideas about the world and how it works that developed as status differences widened and social groups expanded. Many of these ideas were foreign but became quickly domesticated as they were modified, reassembled, and embroiled in local interactions: a Ventral Animal slithered across a new vessel shape and a kancha became a residence surrounded by followers. Bundling Wari together culturally and politically within the Ayacucho Valley is understandable because Wari ideas engendered a shared identity. Everyone in growing cities like Huari and Conchopata would have been associated with a great house. They knew the same sets of gods and fed their ancestors in similar ways. A culture thus emerged that was infused with a set of ideas, one that extended into the countryside as many people routinely circulated through these cities over the course of a year.

One way to think of Wari is as a dynamic assemblage of people, objects, animals, and ideas (Jennings et al. 2021; for assemblage theory more generally, see DeLanda 2016 and Deleuze and Guattari 1987). Without too much hesitation we can speak of a heartland political assemblage arranged around great houses during the early Middle Horizon. Yet I hesitate to call this assemblage a state. A longstanding assumption is that urbanization was a product of state formation. We think that  a ruling class and centralized government were needed to run Wari’s affairs (e.g., Isbell 2009; Lumbreras 1974a; Schreiber 1992). This assertion is based on the idea of an urban revolution (Childe 1936) and the longstanding bundling of city, state, and civilization in Western thought (Jennings 2016). Cities are thought to require centralized administration. Archaeologists, however, have documented no royal palace, main temple complex, grand avenues, or great plazas at Huari. There is also no evidence for full-time bureaucrats, central storage facilities, or specialized craft complexes. No depictions in Wari art clearly show a ruler.

Wari scholars generally agree that competing elite matrilines and D-shaped structures were fundamental to Wari governance. Yet, the assumption that a city as large and wealthy as Huari needs a centralized state has led to attempts to find evidence for a shadow government that worked above Wari’s great houses. At Huari, a rectangular building on one side of Vegachayoq Moqo’s large D-shaped structure has been glossed a ruler’s palace (Isbell 2009), for instance, and two nearby heavily looted, monumental funerary complexes at Huari have been interpreted as royal burials (Cabrera Romero et al. 2021b; Ochatoma Paravicino and Cabrera Romero 2010b). Both locations likely date to the later half of the Middle Horizon, long after urbanization began, and, more importantly, appear to be linked to the more bottom-up efforts to bridge great house divides.

Recent scholarship has identified a wide range of more collective governance strategies in the earliest cities, including those, like Wari, that expanded through violence and colonization (e.g., Blanton and Fargher 2008; Carballo and Feinman 2024; Jennings 2016; Thurston and Fernández-Götz 2021). This scholarship suggests that competing corporate groups could create a city, transform a countryside, and organize long-distance trade. A wealth of data from the Wari heartland, from the analysis of agent depiction in art (Cook 1992; Gibbon et al. 2022) to the variation in architecture and diet between elite compounds (Taylor 2018; Young-Wolfe 2023), hint at competition and autonomy between great houses. The results of dozens of corporate groups pursuing their own agendas would have been often inefficient, messy, and conflict ridden, frustrating the efforts of those who sought to better organize the polity’s affairs. The churn of new and renovated D-shaped structures, small plazas, and other more public features are both the material signatures of ongoing attempts to bridge house divides and the failure of any coalition to endure. Based on current evidence, I suggest that no one was in charge when thousands of colonists left the Wari heartland in the seventh century CE.

Colonial Expansion and Cascading Influence

The flow of imports into Huari and Conchopata correlates with a colonial expansion that saw thousands of people leave their homeland and settle across the central Andes. Expansion followed pre-existing pathways, such that we can trace a spider web of Wari influence, objects, and, to a lesser degree, settlements emanating from the heartland (Jennings 2021; Williams 2017). Both depictions on Wari vessels (Knobloch 2012; Ochatoma Paravicino and Cabrera Romero 2002) and dioramas of Wari figurines (Tuni and Tesar 2011) suggest that expansion was, at least in part, a violent process, with warriors killing and capturing foreign combatants. Violence was endemic in Ayacucho by the Late Huarpa era (Valdez and Valdez 2020), and Ochatoma Paravicino and Cabrera Romero (2002) argue that a Wari warrior class developed with grades linked to one’s success in battle (also see Tung 2014).

External warfare is often used as a tool to consolidate political power (e.g., May 2018), and it is likely that leaders in Huari and Conchopota saw aggression as a means to both ensure resource access for growing settlements and cultivate a collective Wari identity (Arkush 2022; Tung 2008a, b, 2012, 2014). Fighting others is also seen as preferable to fighting each other, and expansion would have provided a shared goal to temporarily ameliorate conflict between great houses. Cranial trauma rates were high among elite males in Middle Horizon Ayacucho (Tung 2012), and depictions suggest occasional conflict between Wari leaders (Knobloch 2012). Waging war could have therefore served as a release valve, periodically propelling warriors outward from the Wari heartland to accrue prestige, loot, and captives for themselves and their house. The first of these warriors may have dreamed of annexing territory. If they did, they quickly realized that conquest and consolidation are two very different things.

Some of the earliest Wari sites outside Ayacucho suggest aspirations for an extractive economy that would assemble resources and funnel them back to the capital. These nodes, though often used by scholars to reconstruct Wari political economy (e.g., McEwan and Williams 2012), tended to be either quickly abandoned or left unfinished. Most colonists instead relied on the heartland’s more intimate forms of governance—the small-scale feasts and other rituals that took place in D-shaped structures and elite compounds—to maintain corporate groups and build relationships with local families (e.g., Jennings et al. 2021). Camelid caravans connected many of these colonies to Ayacucho and other settlements (Tomczyk and Gierz 2017; Tomczyk et al. 2019), exposing Wari objects and ideas to populations who used them in different ways to advance their own agendas. What was “Wari” quickly took on a life of its own as it left the hands and heads of colonists. A way of following the Warification of the central Andes is to travel along some of the pathways that emanated from Ayacucho.

Perhaps the heaviest pathways traveled were those to the Pacific Coast in the greater Nasca region (Edwards 2010, 2021; Edwards and Schreiber 2014). Two of the earliest sites in the region were Pacheco and Maymi. Located on the coastal plain, the settlements were ceramic production centers that likely housed heartland specialists intimately familiar with early Middle Horizon styles (Anders et al. 1994; Mejía Xespe 1926; Menzel 1964). They used local clays and imported pigments to create spectacular, often oversized, Wari-style vessels that sometimes incorporated local motifs and forms (Fig. 6) (Muro Ynoñán et al. 2023). The scale of production far exceeded local needs, and vessels that appear to have been manufactured at the two sites are found in surrounding coastal valleys (Druc et al. 2020), raising the possibility that Wari leaders may have sought to create a new production node in their prestige goods economy to obtain coastal goods (Earle and Jennings 2012, see Grávalos et al. 2023 for an assessment of this economy elsewhere). The Wari-style pottery and textiles in Ancón’s elite burials are in line with this strategy (Slovak 2007), and Lima polity elites also received Wari goods (Marcone 2010). The burial of a highland woman at Ancón raises the further possibility that marriage was part of these exchanges (Slovak et al. 2009; also see Meddens 2020; Nash 2015).

Fig. 6

An early Middle Horizon bottle in the shape of a feline. From a burial in the Nazca Valley, the bottle combines Wari elements with those of the south coast and northern highlands (height: 20.3 cm, 1996.290, courtesy of the Metropolitan Museum of Art).

Excavation results from a third site on the coastal plain, Huaca del Loro, provide a clearer understanding of settlement organization. The site has at least one D-shaped structure and a number of Wari-style compounds (Conlee et al. 2021; Strong 1957). Evidence for feasting with both Nasca and Wari-style pottery in these compounds hints at an incorporation strategy that built off of heartland practices, creating kin-like relations across cultural divides through repeated, small-scale ceremonies. If this were the case, it is interesting that many locals chose to settle in a neighborhood of Huaca del Loro set apart from where the Wari compounds were located. Perhaps, local families were hesitant to align with a specific great house in the first generations after the colonists’ arrival. Portions of patios and galleries at Pachecho and Maymi suggest compound-centered governance at these sites as well, but further work is needed to confirm their layouts (Anders et al. 1994; Mejía Xespe 1926).

Many people in the greater Nasca region welcomed the Wari incursion. Beyond the families who moved into Wari colonies, there is evidence of intermarriage at other sites (Buzon et al. 2012; Conlee et al. 2009) and widespread emulation of Wari motifs and funerary customs (Isla Cuadrado 2001; Menzel 1968). Other locals, however, fled the area around the colonies or pointedly turned their back on Wari ideas (Conlee 2010; Kerchusky 2018; Schreiber 1999; Spivak 2017). If Wari warriors sought to conquer the valley, Wari colonists subsequently extended an olive branch, seeking a foothold through gift-giving, marriage, and feasting. Both efforts were short-circuited by increasing aridity on the coast that stressed local populations (Eitel et al. 2005; Mächtle et al. 2009). The coastal plain of the greater Nasca region was largely abandoned by the eighth century CE, including Pacheco, Maymi, and Huaca del Loro (Schreiber 1999).

Settlers from Ayacucho nonetheless continued to travel to greater Nasca, congregating in the highest reaches of the valleys (2000–3500 masl) where agriculture was still possible. More than 40 Wari-affiliated sites have thus far been identified within these elevations, often placed near coastal-sierra trails (Edwards 2010, 2013; Edwards and Schreiber 2014; Isla Cuadrado and Reindel 2014, 2021; Sossna 2014). Many of these settlements appear to be Wari farmsteads, but there are also larger, more formal, sites featuring D-shaped structures and compound architecture. Archaeological, isotopic, and DNA evidence suggests that Wari/Nasca intermarriage continued, alongside the many small, face-to-face gatherings that took place within the region’s many compounds and D-shaped structures that were swiftly becoming the backbone of Wari governance (Conlee 2021; Conlee et al. 2009; Fehren-Schmitz et al. 2014; Jennings et al. 2021; Knobloch 1983, 2005).

By the end of the seventh century CE, at least one group from Ayacucho appears to have traveled southeast from Nasca along coastal trails to establish a colony at Cerro Baúl in the upper Moquegua Valley. Initially, this group passed through valleys with a few Wari imports and a wide range of local emulations (e.g., Jennings et al. 2015a, b). The inhabitants of some valleys, like Sihuas, had seemingly little interest in Wari, while Wari-inspired motifs can be found at almost every early Middle Horizon settlement in other valleys despite a lack of evidence for Wari colonists (e.g., Owen 2010; Jennings 2014). Cerro Baúl’s founders walked far past this mosaic of Wari-related interaction and influence, placing their settlement on an 8-ha mesa of religious importance to Moquegua’s Huaracane people. The settlement was a tightly packed mix of compounds, D-shaped structures, informal architecture, and platforms aligned to the region’s sacred peaks (Moseley et al. 2005; Williams 2001; Williams and Nash 2006, 2016). Huaracane families, perhaps intermingled with households originated from Nasca and other regions, lived in more ephemeral structures on the mesa’s flanks. An adjacent site, Cerro Mejía, held additional lower-status Wari colonists who also sought to bring coastal followers into their orbit (Nash 2021a, b).

The upper Moquegua Valley was uninhabited when the colonists arrived. The extensive system of agricultural terraces and irrigation canals that developed around Cerro Baúl required local laborers, and close to three decades of research in and around the settlement demonstrates how Wari colonists interacted with surrounding populations (Nash and Williams 2004; Williams and Isla 2002). There is no evidence for the application of force in Moquegua—Cerro Baúl and Cerro Mejía’s Wari inhabitants numbered in the low hundreds, and they had little contact with their homeland (Williams et al. 2019a, b; Reid et al. 2022). Instead, the settlement used the feasts and other ceremonies in its compounds, platforms, and D-shaped structures to create patron–client relationships tied to sacred knowledge, unique experiences, shared ancestry, and the display of irreplaceable Wari objects (Nash 2011, 2019; Williams and Nash 2002). Without regular access to Ayacucho, Cerro Baúl’s leaders sought to fund their activities by monopolizing access to some desired imports that were available in southern Peru, such as Alca source obsidian (Vining 2005). They also strove to create “a tightly controlled craft enterprise” that, in the case of ceramics, used an “exclusive recipe to fashion a diversity of Wari styles” (Sharratt et al. 2023, pp. 8–9). The Wari heartland’s great house governance, reshaped in Nasca, was thus tweaked again to organize affairs in Moquegua by incorporating local mountain worship and reconceptualizing the gift-giving of imports because of the settlement’s isolation from the Wari heartland.

Wari expansion into the southern highlands snaked along trails through the Andahuaylas region. Once again, the movement of people, ideas, and goods ignited a cascade of changes as many embraced the opportunities that Wari engendered (Grossman 1983; Kurin 2016). Closer to Ayacucho, there are traces of extensification, as farmers and herders shifted practices to meet growing demands for crops, meat, and camelid fiber (Pérez Calderón 2017; Pérez Calderón and Carrera Aquino 2015). Poorly understood sites like Patahuasi may have housed Wari colonists (Gómez Choque 2015), while other sites like Yako, with its rustic D-shaped structure that emulated some Wari building techniques, probably reflect efforts by local leaders to reinforce their own power in a rapidly changing political landscape (Meddens 1991; Meddens and Branch 2010). As the polity expanded, the distinction between who was “Wari” and who was not was likely itself an area of contestation.

Most colonists from Ayacucho passed through Anadahuaylas into Cusco. One of the most famous Wari sites in this later region is Pikillacta, a 47-ha site built before the end of the seventh century CE in the Lucre Basin (Fig. 7). The mega-complex features a 3-m-tall perimeter wall, plazas, audience halls, row after row of storerooms, and only a handful of entrances, all in a rigid, orthogonal architectural layout impressed upon an undulating landscape (McEwan 2005). The site looks like an administrative complex, as does the smaller contemporaneous site of Raqchi nearby, with 152 storehouses (Sillar et al. 2013). Wari elite compounds were typically built from the outside-in such that the perimeter was first defined and then the galleries were added (Spickard 1983). Pikillacta and Raqchi were built in a similar manner. First, a perimeter wall was constructed, the interior was then subdivided into sectors by walls, and then sectors were filled in one by one. The sites required millions of hours of labor to construct, and, in this case, it seems likely that the threat of force was a factor because of Cusco’s proximity to Ayacucho along well-traveled trails.

Fig. 7

Site plans of the rectangular enclosure sites of (A) Pikillacta and (B) Azangaro (plans courtesy of Gordon McEwan and the Martha Anders archives at the University of Calgary).

For McEwan (1987, p. 59), Pikillacta was designed to be the anchor of a “greater site,” an extensive complex of related smaller sites, fields systems, and canals. Sillar and colleagues (2013, p. 41) suggest similar aspirations for Raqchi. Although later Inca architecture obscures much of Raqchi, Pikillacta’s niched audience halls were an amalgamation of the heartland’s elite compounds and D-shaped structures where, according to McEwan (2005), hosts held feasts featuring both local and Wari ancestral mummy bundles (also see Topic and Topic 1992). For those who organized Pikillacta’s construction, the aspiration may have been to overcome the chaos of Ayacucho’s urban politics by nesting improved engines of societal integration into a more cohesive compound and broader campus. Great houses would stand together—all for one, one for all—instilling a more robust collective identity in both locals and incoming colonists. The rapid construction of Pikillacta and Raqchi was a “true measure” of Wari leaders’ ability to harness labor in the first years after expansion (McEwan and Williams 2012, p. 74). The great houses, however, had developed in the heartland to act as autonomous actors. Consolidation therefore proved difficult once again.

Most of Pikillacta was left unfinished, with only a quarter of the settlement ever occupied (McEwan 2005). Raqchi also appears to have never been finished, as does the adjacent site of Wanq'osiki (Sillar et al. 2013). Excavations show that the site’s storerooms were unused or temporarily employed as residences (McEwan 1991, 2005, see Sillar et al. 2013 for an argument that Raqchi was used as part of a state labor tax system). People from Cusco would come to Pikillacta and Raqchi for centuries, but in general the sites appear to have been ignored. Wari styles are rare even a few kilometers from the sites, and there were few changes in local settlement patterns, political organizations, and interregional exchange relationships (Covey et al. 2013). The down-the-line influence that often occurred on the central and southern coast was also significantly less in Cusco, suggesting a more limited interest in Wari ideas and objects (Bélisle 2015, 2019; Bélisle and Bauer 2020; Bélisle and Covey 2010; Bélisle et al. 2020).

Colonists from Ayacucho also showed little interest in Pikillacta and Raqchi. Most leapfrogged the sites to settle farther south at Huaro, a sprawling, loosely consolidated settlement that covers 9 km² (Glowacki 2002; Glowacki and McEwan 2021b, a; Zapata Rodríguez 1998, 2009). Huaro dates to about the same time as Pikillacta and is made up of residential clusters likely composed of “independent or semi-autonomous groups from Ayacucho” (Skidmore 2014, p. 313, but see Glowacki 2014 and Glowacki and McEwan 2021 for a different interpretation of the settlement’s relationship to the state). The settlers who came to Huaro lived in and around elite compounds. Interestingly, there are neither D-shaped structures nor Pikillacta-inspired audience halls at Huaro—nothing architecturally speaks to efforts to bridge great house divides. Instead, the site has the unruly, precocious feel of a frontier settlement, with residents maintaining ties to the Wari heartland but also interacting with Tiahuanaco and other Lake Titicaca Basin settlements farther south via long-distance interaction networks.

When Wari expanded into the northern highlands, one of the first sites built was Viracochapampa. A 33-ha site in Huamachuco, Viracochapampa has high walls and a regular orthogonal structure like Pikillacta. The site, however, was abandoned before it was finished (J. Topic 1991; Topic and Topic 2010), as may have been the nearby enclosure of Yamobama (Chirinos-Ogata 2017). At around the same time, Wari leaders were likely making offerings at nearby Cerro Amaru, a pilgrimage center associated with water that was tied to an elite lineage living at the site (Topic and Topic 1992, 2020). Pilgrims traveled hundreds of kilometers to make offerings at Cerro Amaru that were then placed in storage units. If the Wari-style offerings left at Cerro Amaru were gifts from Wari leaders, then perhaps they hoped to tap into the shrine’s long-distance relationships. These hopes, however, would be dashed as decreasing rainfall sapped the shrine’s power, leading to its decline soon after the Wari-style offerings were made. While colonists continued to migrate to Nasca and Cuzco, few would settle in the northern highlands during the first half of the Middle Horizon (T. Topic 1991).

As in other regions, Wari influence varied along the interregional pathways that connected the northern highlands to Ayacucho (e.g., Hastorf 1993; Lau 2002; Ravines 1977). Changes at another shrine in Huancayo, Wari-Wilka, hint at broader attempts in the northern highlands at co-opting longstanding ritual centers (Shea 1969), but early Middle Horizon Wari resource acquisition strategies appears to have focused more on the extension of their prestige goods network that funneled Wari-related ceramics, textiles, and other desired objects into the hands of the region’s elite (Isbell 2010; Lau 2012). One Wari goal may have been to strengthen ties to the Ecuadorian Spondylus shell trade (Knobloch 2021), with efforts also made to build relationships with the north coast Moche via trails descending from Cajamarca (Castillo 2000). Traders from Cajamarca had developed relationships with Moche by the end of the Early Intermediate Period, and the co-presence of Wari and Cajamarca wares in many contexts suggests that Wari leaders utilized Cajmarca’s pre-existing network to extend their reach (Cusicanqui 2022; Toohey and Chirinos-Ogata 2018).

We know the least about Wari’s eastern expansion into the river valleys that descend into the Amazon Basin (Villar 2023). A string of sites with Wari-related ceramics can be found along these routes, potentially providing access to coca, tropical feathers, and other jungle products (Raymond 1992, 2021; Raymond and Isbell 1969; Valdez et al. 2015; Wilkinson 2018). Without further research, however, it is unclear if these were local sites with Wari connections or the home of Wari colonists. More data exist for Espíritu Pampa, a settlement composed of elite compounds, agglutinated architecture, and D-shaped structures that had close ties to the heartland (Fonseca Santa Cruz and Bauer 2013, 2020). Another site, Inticancha in Chachapoyas, is reminiscent of Viracochapampa and Pikillacta and was also left unfinished (Church and Muscutt 2018; Ravines 1994). As in other regions, imperial ambitions of a more cohesive, tribute economy on the eastern flanks of the Andes appear to have quickly fizzled as colonists pivoted to variations on the more limited, more familiar, great house incorporation strategy centered on gift exchange and the intimate ceremonies taking place within elite compounds and D-shaped structures.

The story of Wari’s expansion is often told as a story of territorial conquest, with peripheral outposts serving as administrative sites in an extractive economy (e.g., McEwan and Williams 2012; Schreiber 1992). Violence undoubtedly played a role in the expansion, but Wari governance was predicated on creating and sustaining personal relationships. Foreigners had to become kin, and so the primary tools of Wari politics were feasting, gift-giving, and intermarriage. These, and other integrative mechanisms, were not the work of administrators following directives from central command—there was no state government to pull the strings—but rather the actions of great houses competing for wealth in people (e.g., Bledsoe 1980). The row of interconnected patios and galleries at sites like Pikillacta, Viracochcapampa, and Inticancha suggest a desire for greater coordination across great houses in the first years after expansion. Yet, these endeavors quickly collapsed, as did other attempts to co-opt shrines and create gift economies.

One way of looking at Wari’s expansion is in terms of dysfunction. Groups in different regions appear to have cycled through a variety of incorporation strategies. Classifying this as dysfunction, however, is based on the assumption that other early polities were able to effectively exert and sustain considerable power over others across great distances. Recently, a number of scholars have instead proposed “low-power” models wherein leaders often struggled to influence peripheral affairs (e.g., Ando and Richardson 2017; Khatchadourian 2016; Stein 1999). They picked, sometimes literally, their battles, and were quick to adjust their strategies to changing circumstances. In these low-power expansive polities, ambiguity was often a strength, and imperial rhetoric obscured a startlingly tenuous grasp over what happened on the ground (Grossman and Paulette 2020; Smith 2015). Seen in this light, Wari fumblings are just another example of an early polity in action, with leaders in different regions continually tweaking their relationships to find ones that worked. Ultimately those sites that endured like Huaro and Cerro Baúl were the ones that played to Huari’s strength: let great houses work to build kin-like relations between colonists and local groups.

The impact of Wari expansion can also be fruitfully explored through the lens of globalization theory (Jennings 2010a, b). Fernandini (2021), for instance, has attempted to separate the broader Wari cultural horizon from the far narrower footprint of the polity’s colonial expansion. As people, objects, and ideas left Ayacucho, they pulled away from the dynamic, often unruly, assemblage of heartland Wari society and entered into a wide variety of other entanglements. These entanglements uncoupled, reshaped, and extended what it meant to be “Wari” across space and time, even among people who had roots at Huari. Homogeneity, heterogeneity, standardization, and other characteristics of globalization can be found across the central Andes during the Middle Horizon, products of a dynamic “complex connectivity” that reshaped many lives amid the era’s heightened interregional flow of ideas, objects, and individuals (Tomlinson 1999).

A Wari-style pot initially employed in a gift exchange, for example, might be regifted again and again. The original ideas associated with its motifs might fade or become garbled as it became more akin to exotica (e.g., Helms 1988). The rationale for Wari elite privilege would also change in conversations between a Wari colonist and local leader, then change again as that leader attempted to assert a similar privilege amid their community’s traditions (e.g., Appadurai 1996). “Wari” as practiced in the cuisine prepared by a lone woman married into an Ancón elite family would be different than that practiced by a lower-status group newly settling into Huaro or a long-established higher-status one at Espíritu Pampa (e.g., Lightfoot 2004). Wari was transformed in each encounter as people navigated their lives in an era of widespread connections and often sweeping change.

Late Middle Horizon Reorganization, Dispersal, and the Wari Collapse

Menzel (1964, p. 69) suggests a “severe crisis” occurred midway through the Middle Horizon that led to increased secularization and diversity in Wari ceramic styles. Subsequent work has dated this shift to the beginning of the ninth century CE, correlating the stylistic transformation with a wide range of additional changes from settlement reorganization (Isbell 2009) to agricultural intensification (Schreiber 1992). These changes appear to be associated with increased political centralization in Ayacucho that also propelled Wari colonists, objects, and ideas even further afield (Schreiber 2001). The best evidence for a Wari state, as traditionally defined in archaeology, thus comes from the late Middle Horizon. Many of those who left Ayacucho may have been fleeing this inchoate state, with the idea of “Wari” morphing down-the-line once again as a result of their travels.

In the Vegachayoq Moqo sector of Huari, the largest D-shaped structure likely dates to this period (Bragayrac 1991; Finucane et al. 2007; González Carré et al. 1999). The temple’s courtyard is massive by Wari’s standards, allowing hundreds to gather. Nearby are the two excavated monumental funerary complexes that also likely date to the late Middle Horizon (Finucane et al. 2007). The tombs, like the larger D-shaped structure, are more-of-the-same, taking existing ideas of funerary architecture and making a deeper, more expansive, complex than previously known. Some scholars suggest that Vegachayoq Moqo was the precinct of a Huari ruler and that the two tomb complexes were royal mausoleums (Isbell 2006, 2009; Ochatoma Paravicino and Cabrera Romero 2010a, b). Although such assertions seem premature, it is clear that larger groups were being mobilized by the late Middle Horizon to build, feast, and bury the dead at Huari.

The increased size of corporate groups occurred at the expense of the cancha-inspired elite compounds. According to Isbell (1997b, 2001a, b, 2009), many of the compounds at Huari were abandoned, demolished, and covered by a layer of sand around the beginning of the ninth century CE. Immense trapezoidal enclosures were then built with large interior spaces, perhaps with the hopes of hosting the bigger collectives involved in their construction. The trapezoidal enclosures were never finished, however, and piles of building stone were left across the site. Many of Conchopata’s elite compounds were also abandoned in the ninth century CE, and the manufacture of elaborate and oversized vessels largely ceased (Isbell 2009; Wolff 2012). There would be no urban renewal at Conchopata, and those who lingered lived in single-room houses reminiscent of the Late Huarpa era (Isbell 2009). The elite compounds had been the anchors of the heartland’s great houses—the engines of societal integration—and without them the urban centers withered.

Huari’s aborted renewal occurred on the heels of efforts to streamline agricultural production. Azangaro (Anders 1986, 1991) and Jargampata (Isbell 1977) are two of several enclosures in the Ayacucho Valley with internal patios and apparent storage facilities that date to the late eighth century and later (see Fig. 7). Similar changes occurred in the nearby Sondondo Valley, where Wari-affiliated Jincamocco was radically expanded from 3.5 to 15.0 ha amid a valley-wide resettlement program designed to intensify maize production (Schreiber 1992). The success of these intensification efforts is unclear; Azangaro’s storerooms were not used to store agricultural products, and Jincamocco’s expansion may have been left unfinished—but their inauguration suggests a concern with food production in the buildup to the changes at Huari and Conchopata.

The ninth century also saw changes throughout the Wari interaction sphere. In the greater Nasca region, most of the sites in the upper valleys were abandoned (Isla Cuadrado and Reindel 2014; Jennings et al. 2022b; Schreiber 1999). Some groups from Ayacucho nonetheless appear to have traveled through Nasca and then turned southeast along the coastal plain trails. They left face-necked jars stuffed with feathered textiles in a longstanding ceremonial site at Corral Redondo (King 2013) and founded the small site of Pakaytambo at a crossroads in the upper Majes Valley (Reid 2023). Another group settled for a few years at Quilcapampa in Sihuas before vanishing after an ordered abandonment of the site (Jennings et al. 2021). Wari-affiliated ideas, objects, and people also moved farther down the coastal plain trails toward Moquegua (Cardona Rosas and Wise 2000; Nigra et al. 2017), where there was a “massive construction event” in Cerro Baúl at the beginning of the 10th century (Williams 2001, p. 79). The changes at Cerro Baúl created what excavators have called a “palace,” larger public spaces, and a greater focus on incorporating outsiders into settlement activities (Nash 2015; Nash and Williams 2009). Some of these outsiders were the Tiwanaku colonists lower in the drainage who had been largely ignored during the previous three centuries.

The ninth-century pulse of colonization along the southern Peruvian coast helped push ideas into the adjacent highlands, with Wari stylistic influence seen for the first time in the Chuquibamba (Goldstein 2010), Cotahuasi (Jennings and Yépez Álvarez 2015), and, to a lesser extent, Colca Valleys (de la Vera Cruz Chávez 1987, 1996). Long-distance interaction would decline by the end of the century, however (Jennings et al. 2015a, b). Rather than indexing relations to Ayacucho, Wari-related iconography, forms, and practices in the region may have been increasingly used to create a shared coast–highland regional identity (Jennings 2006b). A different experience was unfolding across much of the central coast down through Nasca (Fernandini 2021). Increasing aridity had led to the abandonment of most settlements on the coastal plain by the end of the eighth century. People nonetheless returned to these sites periodically to hold ceremonies and deposit the mummy bundles of elites and associated offering (Fig. 8) (e.g., Fernandini and Alexandrino 2016; Flores Espinoza et al. 2021; Pozzi-Escot et al. 2021; Segura Llanos and Shimada 2010). Many of the textiles and ceramics were Wari in style, but they appear alongside a wide variety of other prestigious styles from the period. Wari objects were thus employed as status markers, likely by elites seeking to maintain claims to ritually charged places to which they hoped to return. “Wari” on the central coast in the late Middle Horizon was thus different from “Wari” in the southern highlands, and both had little to do with what was happening in the Ayacucho heartland.

Fig. 8

Late Middle Horizon funerary context from the site of Cerro de Oro in the Cañete Valley that contains prestige goods from Wari and other cultures (courtesy of Francesca Fernandini, Programa Arqueología de Cañete-Cerro de Oro).

Pikillacta in Cusco was substantially expanded in the ninth century, perhaps to realize its earlier ambitions as a regional administrative center (Glowacki 2005). Ayacucho settlers, however, continued to largely avoid Pikillacta, and the surrounding local population took little interest in the site—the renovations were never completed (Covey et al. 2013; McEwan 2005). Huaro meanwhile grew in size and prosperity as Ayacucho colonists and local populations took advantage of rising interregional exchange in southern Peru and northern Bolivia in the absence of state oversight (Bélisle 2015; Skidmore 2014). “Wari” identity in late Middle Horizon Huaro may have been, at least in part, performed by families who had recently left the Wari heartland and were negotiating who they were amid a widening Tiwanaku diaspora after monumental construction ceased at the site in the mid-eighth century (Marsh et al. 2023).

Wari colonists returned to the northern highlands in the late Middle Horizon, settling largely at El Palacio located almost 100 km to the northwest of Viracochapampa (Watanabe 2014, 2016, 2019). Sprawling across 100 ha, El Palacio was organized around competing great houses and likely housed a mix of settlers from Ayacucho and locals from the Cajamarca region. The settlers seem to have had little involvement in the affairs of surrounding communities (Julien 1988; Watanabe 2014), focusing instead on trade relationships with communities connected to spondylus shell and green stone sources to the north in Ecuador and north coast groups to the west (Knobloch 2021; Mackey 1982; Paulsen 1974). Wari identity, once again, was performed against the backdrop of a different set of interregional relationships.

The north coast was reached via a well-traveled route that came down the Jequetepeque Valley (Castillo 2000; Castillo et al. 2012; Swenson and Berquist 2022; Watanabe 2014; Watanabe and Ugaz 2021). Wari-affiliated travelers taking these paths in the late Middle Horizon may have established an enclave with a D-shaped structure at Huaca Santa Rosa de Pucalá in Lambayeque (Bracamonte Lévano 2021a, b), as well as founded the site of Castillo de Huarmey on a portion of the central coast that had been previously free of Wari influence (Bernier and Chapdelaine 2018; Chapdelaine 2010; Nelson et al. 2010; Wilson 1988). The latter site has a spectacular tomb of what may have been a foreign elite and local wives (Knudson et al. 2017; Prądka-Giersz 2019). Intermarriage was therefore an important part of the colonists’ strategy of laying claim to a portion of the Huarmey Valley, as was the local manufacture of a wide range of Wari-style goods (Giersz et al. 2021, though see Kałaska et al. 2022 for the much wider circulation of metals evidenced at Huarmey).

In the valleys leading into the Amazon, the ninth-century transition is most clearly noted at Espíritu Pampa from a set of elite burials featuring depictions of an individual who survived capture and torture at the hands of other Wari actors (Isbell 2016; Knobloch 2016). The burials contain examples of south coast ceramics that have not yet been found at Huari or anywhere else in Ayacucho (Knobloch 2016), raising the possibility of extensive travels prior to the group’s burial at Espíritu Pampa or at least connections to other places that circumvented the Wari heartland. If this was the case, then it is interesting to think about the political ramifications of this agent’s followers being buried at Espíritu Pampa. The followers were, in one sense, clearly “Wari,” but they may have had a contentious relationship with those living in the Wari heartland after the dismantling of the region’s great houses.

The reasons behind the ninth-century pulse in colonization remains unclear. The razing of elite compounds at Huari and the abandonment of others at Conchopata, when combined with the possible evidence of internal conflict in Wari art, suggests a political shake up associated with the scaling up of social units and greater centralization. These changes appear to have pushed colonists outward, largely to Huaro and the newly established site of El Palacio that were at the fringes of the Wari interaction sphere. Still others traveled farther, establishing settlements in locations that had previously seen little to no Wari influence (e.g., Jennings et al. 2022b). At Castillo de Huarmey and Espíritu Pampa, some of the buried individuals could have been wealthy elites with close ties to Ayacucho. The late Middle Horizon Wari diaspora can be seen as an extension of empire (Giersz 2016; see Giersz and Pardo 2014 for Huarmey), but I suggest that many of those who left Ayacucho might be better characterized as refugees fleeing a Wari heartland where they were no longer welcome.

Huari and Conchopata were in steep decline by the end of the ninth century CE. The reorganization had failed, and at Huari we see evidence for abandoned temples, the looting of burials, and acts of iconoclasm (Cabrera Romero and Ochatoma Paravicino 2019; Ochatoma Paravicino et al. 2015). The energy expended can be impressive—a massive stone table, for example, was broken and deeply buried (Cabrera Romero and Ochatoma Paravicino 2019)—suggesting less a sacking of the city and more a collective loss of faith in the institutions that had structured city life over the last three centuries. Signs of conflict are also seen at Conchopata (Ochatoma Paravicino and Cabrera Romero 2000), and Azangaro “was abandoned abruptly amid what seem to be indications of conflict and confrontation” (Anders 1991, p. 190). More radiocarbon dating is required to elucidate the timing of the conflicts and the ultimate abandonment of these sites, but it is clear that there was little semblance of a Wari polity in Ayacucho by the 10th century even as people lingered at some sites until approximately 1000 CE (Finucane et al. 2007).

Wari as a cultural phenomenon very much continued elsewhere in Peru despite the heartland’s decline. Wari-inspired ideas spread farther and penetrated deeper into daily life in new parts of the central Andes during the 10th century CE, creating international identities that facilitated regional interaction and legitimized status differences (Isbell 2008). Colonies like Castillo de Huarmey and Cerro Baúl also prospered, with at least the latter site outlasting the last of Huari’s inhabitants (Williams 2001). When people finally left, there is little evidence for the violence and iconoclasm that plagued the heartland. Abandonment was steeped in ritual and orderly: participants infilled rooms, blocked doorways with stones, and made offerings along the way (e.g., Edwards 2017; Nash and Williams 2021). At Cerro Baúl, celebrants at the palace came together for one last feast, shattered their cups, and burned the building as they left the settlement (Moseley et al. 2005).

If Wari can be seen as a dynamic assemblage anchored by the heartland’s great houses, then what happened when these great houses fell? The severe crisis first identified by Menzel relates to a early ninth-century political reorganization that appears to have targeted these households, perhaps in the hopes of creating a much smaller group of decision makers within a more unified Wari polity. Eradicating the great houses, however, would have significant down-the-line implications because Wari sovereignty, as it existed, was kin-oriented and constituted in small, face-to-face rituals. Interestingly, late Middle Horizon colonies continued to build elite compounds and D-shaped buildings, suggesting that an alternate form of governance never took hold in the periphery. Some sites may have been founded by refugees, but even those living in longer-standing settlements had to face the same new reality: they would now need to cobble together their own Wari political assemblage from what they had at hand. At Espiritu Pampa, this might mean an assemblage that incorporated enemies of a swiftly dissolving Wari state, while at Cerro Baúl it meant ingratiating themselves with Tiwanaku neighbors. A different calculus was at play further down-the-line as other groups had to decide what “Wari” meant without the pull of a core region.

Defining Wari

Following an Inca-inspired model, Wari has long been seen as an empire that conquered and controlled the central Andes. Wari administrators are thought to have optimized yields and organized surplus extraction, funneling the wealth of conquered provinces back to the imperial heartland (Isbell and McEwan 1991; Lumbreras 1974a, 2000; McEwan 2005; Menzel 1977; Schreiber 1992; Williams and Isla 2002). There remained those who were unconvinced (e.g., Bawden and Conrad 1982; Moseley 1983), but the alternatives proposed for the Middle Horizon like Shady Solís’s (1982, 1988) peer polity model and Shea’s (1969) semiautonomous oracle model never gained traction. With the conquest empire so well ensconced in the late 20th century, Cerro Baúl was seen as a fortress that repulsed Tiwanaku enemies (Feldman 1998; Moseley et al. 1991), a kilometer-long irrigation canal that serviced Middle Horizon terraces around Pikillacta was an example of state agricultural intensification (Valencia Zegarra 2005), and the evidence for well-maintained trails near some Wari sites was an indication of a vast, state-made, interregional road network (Schreiber 1991).

Further research in the early 21st century, however, complicated the picture. For example, work at Conchopata emphasized the role played by competing “lineage houses” in daily life (Blacker and Cook 2006), and depictions of possible Wari leaders also hinted at cooperation and conflict (Knobloch 2010, 2012). In the periphery, it had become clear that Pikillacta and Viracochapampa had never been finished (McEwan 2005; Topic and Topic 2001), that some sites preliminarily identified as Wari administrative centers were made by other groups or dated to other periods (Jennings 2006a), and that relationships with local groups could be nonhierarchical (Topic and Topic 2010). This and other scholarship did not so much offer an alternative vision to the conquest empire hypothesis as muddy the status quo: we now have to acknowledge that there is much more to Wari than the prevailing model can explain (e.g., Castillo 2021; Kaulicke 2021; Makowski 2016).

The intellectual trajectory of Wari studies parallels that of other parts of the world (Feinman and Marcus 1998). Efforts to organize and classify past societies in the early 20th century gave way to the processual archaeology of the 1970s that sought universal mechanisms of state formation. More data revealed more complexities, leading to the critique of prevailing state models by the 1990s. Scholars became uneasy about long-enshrined stages of cultural evolution and embraced greater political, economic, and social diversity across space and time for regionally organized polities (e.g., Cusick 1998; Schwartz and Falconer 1994; Stein 1999, 2005). Much of the work over the past three decades has been about documenting this variability and offering alternatives to narrow definitions of “the early state” (e.g., Claessen and Skalník 1978). The questioning of the nature of Wari in the 2000s should be seen as part of this broader postprocessual moment in archaeology that continues to this day.

This article offers an alternative to the conquest empire hypothesis. Wari leaders began with imperial aspirations. Yet, like in other early expansive polities around the world (e.g., Worthington 2014), conquest proved far easier than consolidation. Over the past two decades, scholars have begun to think of many of these polities as “low-power states” (Richardson 2012, 2017). In these models, capacities are limited, and sovereignty must be “continually reassembled, reproduced, [and] brought into being” (Grossman and Paulette 2020, p. 3). In the confines of a D-shaped structure at Huari, one can imagine claims being made of unbridled Wari power. The reality on the ground, however, was that Wari sovereignty remained familial after expansion. Power over others was largely predicated on their continual participation in feasts and other small-scale ceremonies that created rival collectives organized around shared ancestors. Wari sovereignty, enacted through these ceremonies, was thus fractured between a settlement’s competing great houses and limited to the few kilometers that routine participants would have likely traveled. Gifts, of course, circulated more widely, but Wari leaders never developed political tools that enabled territorial governance.

Wari as a cultural phenomenon nonetheless spread across much of the central Andes. Some of this occurred via Wari actors, such as through the gift economy and co-option of sacred sites in northern Peru. These incorporation strategies could have formed the basis of a more robust imperial political economy (e.g., D’Altroy and Earle 1985), but they failed to gain significant traction. One reason why may have been that the heartland’s heterarchical organization of competing great houses inhibited the political centralization that could have better coordinated such efforts. A consequence of the sputtering attempts to extend sovereignty was nonetheless a surge in interregional interaction, creating the complex connectivity associated with an era of globalization (Jennings 2017). Many people were entangled with Wari during the Middle Horizon, but the Wari experience was different for each person. For most, Wari was down-the-line, filtered and transformed by the self-interest and perspectives of other actors. It was a llama caravan glimpsed from afar, a new decoration on a pot, or a bogeyman to scare children around the fire.

This re-envisioning suggests that Wari never conquered and controlled the central Andes. The Wari international identity that emerged there nonetheless proved pervasive and enduring (Isbell 2008). Many of the interregional connections spurred by Wari’s expansion would deepen in the centuries that followed, producing a robust circulation of shell, obsidian, feathers, and many other products despite the political fragmentation of the succeeding Late Intermediate Period. Ayacucho’s residents also helped shape ideas about household composition, elite status, mortuary customs, and other core aspects of society that continued long after the Spanish Conquest. Leaders at Huari had no say in many of these developments, but this should not diminish their impact. They helped create a cultural horizon that in many ways still endures in the central Andes more than a millennium later.

The model proposed in this article remains tentative. Admittingly, it is peppered with speculation, albeit speculation—the display of mummy bundles, shift in agricultural labor organization, the rationale for the acquisition of Wari-style goods, the work that Wari-style architecture did—that have often been shared with previous Wari scholarship. Although I argue that what is proposed here fits with the available evidence far better than the prevailing conquest empire model, much more work is required to better understand the Wari phenomenon. Few excavations have been conducted on rural sites in the Wari heartland, for example, and Huari still requires intensive pedestrian survey. There are also too few radiocarbon dates, insufficient studies on paleomobility, and a dearth of ceramic petrography. Perhaps most importantly, moving forward we need to continue to recognize that, to paraphrase Walt Whitman, Wari was large and contained multitudes, transforming across space and time during the Middle Horizon.