The Landscape and Settlement Patterns of the Sidekan Subdistrict

The mountainous Sidekan subdistrict, with its small valley floors suitable for agriculture surrounded by steep hillsides and mountains, lacks many of the natural characteristics that would pull precipitate large settlements. Without rare minerals and with only limited acreage of arable land, occupation of this region required catalyzing factors to lead to a population size sufficient to form a recognized political entity. Why occupants chose to live in the Sidekan subdistrict underlies many of the research questions. The objectives of this dissertation include determining the factors motivating settlement of the Sidekan subdistrict, understanding the chronological extent of occupation in the Sidekan subdistrict, and exploring why Muṣaṣir-era sites dominate the archaeological assemblage. One approach for understanding settlement motivations and determinants is settlement ecology, a theory that considers the relationships and interplay between human and environmental factors (Brannan and Birch 2017, 55). The Rowanduz Archaeological Program’s (RAP) site survey and multiple excavations yielded a multi-scalar regional data source that, with the assistance of Neo-Assyrian and Urartian historical records concerning Muṣaṣir, provides an integrative data source for these analyses.

Settlement Ecology

Settlement ecology is a theoretical approach to studying settlement created in response to the research question of “why do people settle in a given place during a specific time and in a particular arrangement” (Kellet and Jones 2017a, 1). Glenn Davis Stone, in his foundational manuscript Settlement Ecology: The Social and Spatial Organization of Kofyar Agriculture (1996), developed a diachronic integrative approach borrowing anthropology, economics, geography, and ecology to answer those questions. Building off previous studies focusing on the spatial relationship of villages and farms (Stone 1992) and the ethnic dynamics of settlement abandonment (Stone 1993), his comprehensive book charts the decision factors underlying Kofyar settlements in Nigeria (Stone 1996). His research observed the expansion and migration of the Namu valley by the Kofyar agriculturalists, which led to modeling the interrelated factors that pushed and pulled settlements towards nucleation and dispersal. In promoting settlement ecology as the theoretical approach for this analysis, Stone built on a long lineage of scholars settling settlement. Later archaeological settlement ecology studies brought in tools like GIS to determine the spatial relationship between the landscape, cultural features, and social dynamics. As an integrative methodology utilizing the vast interdisciplinary research on settlement, agriculture, and spatial dynamics, the historical foundations of settlement ecology are central to operationalizing the theory.

Foundations of Settlement Ecology

##### The Geographers

Before Stone’s contribution to settlement ecology, archaeologists and geographers had advanced a litany of varying reasons or theoretical frameworks to understand why people settled in given places. Among the theoretical forebearers were geographers like Johann Heinrich Von Thunen ([1826] 1966), Walter Christaller ([1933] 1966), and Ester Boserup (1965), who advanced the ideas of proximity-access, Central Place Theory, and population’s effect on agricultural intensification, respectively.

In the 19th century, the emergence of the social sciences and the movement towards the scientific study of the quantifiable world led the geographer von Thunen to model the interplay of settlement, land use, and spatial distance ([1826] 1966). In the process of trying to better understand the value of land and rents for landowners in the early Industrial Revolution, he generated a mathematical model paralleling the spatial arrangement of land. He based his model on a theoretical single market town located in an idealized homogenous agricultural plain, where the marginal productivity of each plot of land is determined by the capability of the land minus transportation costs to transport goods to the central market. In von Thunen’s model of the “isolated state,” concentric rings representing progressively decreasing profitability emanate around the central market town. With the model’s assumption of uniform fertility and a single market, the transportation costs to the center to the center dictate optimal land uses for each circle – the nearest engaging in most intensive cultivation, middle rings with forestry or rotating fallow, and ranching at the furthest extent (von Thunen 1966). Von Thunen’s use of profit maximization and economic rent was imperfect but served as a simple proxy to represent the many variables of agricultural productivity and demonstrate the principle that proximity to markets directly dictates land use in an observable and repeatable process.

While often dismissed as overly broad or nonrepresentative of the complicated dynamics of field use, von Thunen’s findings highlighted the outsized importance of agricultural proximity that served as the foundation for Christaller’s more dynamic principle of settlement geography, Central Place Theory. Like von Thunen, Christaller ([1933] 1966) created a hypothetical geographical model with nearly all variables held constant to evaluate the impact of changing relevant inputs. His model assumed terrain, transportation facilities, and population density are constants, with consumers accepting perfect competition and producers abandoning the motivation for excess profit. In contrast to von Thunen’s isolated state, Christaller’s model for Central Place Theory added additional markets on the idealized landscape as settlements of tiered sizes between villages and cities. The size of each settlement dictates the types of goods or services offered. Further, he represented the primary measure of cost with the maximization of time to reach markets. Markets and resources on the landscape serve as attractions for the settlers, using a simple model with a rudimentary weighting of features ([1933] 1966, 84-133).

Christaller’s model resulted in his Central Place Theory, that central places serve as points of attraction that support outlying settlements through goods and services. Central functions “are produced and offered at a few necessarily central points in order to be consumed at many scatter points,” e.g., towns and villages, where transportation can provide those goods and services over longer ranges (Christaller [1933] 1966, 19–21). In an environment with constant terrain, the resulting pattern is an evenly distributed hierarchical system of towns and cities, with a hexagonal-shaped hinterland surrounding a settlement. Cities, with the largest markets and greatest capacity for goods and services, form hexagonal hinterlands between themselves and other cities. Towns and villages follow the same pattern, with smaller hexagonal hinterlands around each tier of settlement. This geometry relies on the market principle, where central places maximize the range of goods produced in order to optimally minimize transportation (Christaller [1933] 1966, 66–72). At the border of a city, the furthest distance from the central market, the cost of traveling to the city outweighs the value of the goods and services there. As the threshold required for producing a product rise, the border around the city increases, because the rise in transportation costs is outweighed by the higher costs of production at a closer settlement locus. Christaller’s model can be mistaken as generalizing to the point of inaccuracy, but like von Thunen proximity access rings, the hexagonal and hierarchical structure serves to illustrate a single principle at play – the importance of markets and their positioning.

While Von Thunen and Chisholm’s models differ in their emphasis on agrarian land use versus settlement placement, both operate on the same premise that effort is a constant in the pursuit of maximal productivity. In its simplest form, when taking effort as a fixed value (x), the combined effort of transportation (y) and effort of production (z) must equal total effort (x = y + z). An increase in effort required for transportation must be accompanied by a proportionate decrease in effort for production (x = (y+1)+(z-1)). The geometric shapes of concentric circles and overlapping hexagons are the results of overlaying this mathematical theory on a hypothetical version of the real world. Transportation and distance form the foundation of settlement geography, but the early geographers’ models did not account for dynamic effects of cultivation and population size.

Boserup (1965) sought to model the relationship between population growth and food production, manipulating the two variables of demographics and agricultural productivity. Precipitated in part by a rejection of the Malthusian view of fundamentally inelastic food production, implicitly reflected in the settlement models of von Thunen and Christaller, Boserup’s model used the growth in population as the independent variable that affects the methods and intensity of agricultural activity (1965, 1). Like the previous geographers, the model held most factors constant, such as settlement pattern, to focus on the independent and dependent variables of interest, rather than trying to capture the myriad possible inputs related to agriculture. Another assumption of Boserup’s model is the Law of Least Effort, that farmers will expend effort the minimal effort necessary to satisfy their needs (1965, 28-32). Use of the Law of Least effort is often critiqued as a Eurocentric view of a false dichotomy between work and leisure that ignores the cultural differences in the perception of effort and productivity or requires an external force to force surplus production (Morrison 1994, 130-31; Erickson 2006, 336). While valid criticisms of the underlying motivations, the model’s necessary assumption of only producing required outputs enables a narrow focus on the relevant interplay between population and land use.

With population as the independent variable, Boserup’s model predicts that a growth in population may lead to technological advances and increased intensification of the available land, moving from extensive to intensive cultivation (1965, 41-42). She used a simplified view of fallowing and intensification frequency, creating a single spectrum that spanned low effort extensive or high effort intensive cultivation (Boserup 1965, 15-18). When the rural population is low, farmers shift seasonal production between fields, fallowing the unused fields for multiple growing seasons. Those long periods of fallow result in highly productive land, with zero or minimal effort expended during the fallow seasons (Boserup 1965, 12). With increased population and a requirement for greater output, farmers decrease the fallow time of fields, with additional labor required to match or surpass the productivity of highly fallowed land. As population rises, the intensification of the land increases, resulting in a higher marginal labor cost to produce the same output. The resulting interaction can be displayed as a graphical representation of efficiency and population concentration (Boserup 1965, 23–55). Her model indicates productivity narrowly associated with agricultural output is negatively correlated with population growth but notes that associated population concentration and social organization may lead to second-order effects on the population more broadly (1965, 116-120).

##### The Settlement Archaeologists

Amidst the backdrop of geographers’ growing and increasingly complex models, settlement archaeologists began utilizing some of that discipline’s insights in order to better understand archaeological landscapes and settlement patterns. Willey’s (1953) analysis of the Viru Valley in Peru used aerial photography to document over 300 site locations and separate them into typologies, describing the distribution of types in different ecological zones. While lacking the quantitative rigor of some earlier geographers’ economic models or studies by future settlement archaeologists, Willey’s regional model was one of the first archaeological studies that emphasized the importance of spatial relationships between human features on the landscape. Not long after, Binford’s (1964) propagation of statistical analysis for intra-site and regional archaeological analysis initiated a new wave of archaeologists adopting ecological tools and an accompanying quantitative rigor for the study of agrarian or hunter-gatherer settlement locations (Binford 1964; 1980; 1990; Deetz 1968; Whallon 1968; Ashmore 1981; Kelly 1983).

Following the greater utilization of statistical and quantitative tools for the descriptive analyses of archaeological settlement, scores of archaeologists used tools adopted from the fields of geography and ecology to understand the placement of archaeological sites. In parallel to the theories of proximity access propagated by von Thunen and Christaller, Vita-Finzi and Higgs (1970) introduced to archaeology the methodology of catchment analysis, studying the spatial relationship between sites, the “natural resources lying within the economic range of individual sites,” and mobility (1970, 5). Combining the tenets of site catchment and central place theory, Flannery (1976) began by applying the methodology to the Etla Valley in Oaxaca, Mexico. His study described a series of steps for settlement growth, beginning near river fords, spreading symmetrically to daughter settlements before eventually filling in the interstitial space between existing settlements. Although using the basic principles of site catchment, he pointed out the methodology’s difficulty in the binary identification of agricultural vs. non-agricultural land and the reasons farmers may distribute fields in unorthodox patterns as disaster mitigation (Flannery 1976: 92). In a further effort to explain settlement patterning, Trigger (1981) analyzed subsistence availability, political institutions, technology, population, among other factors, as causal determinants of settlement location. Another study by Sanders included a far greater list of ecological determinants, like rainfall, zonal soil patching, temperature, among others, but all of these models failed to provide a system for prioritization or optimization of these variables (Hamond 1981; 1981).

These models do not fall prey but circle the dangerous trap of the axiom “correlation does not equal causation,” implying correlations between the observed order and position of settlements to ecological or geographic factors. Flannery (1976, 162) acknowledged that these approaches to settlement patterns use a set of probabilistic “rules” in search of the original reasons for occupation. However, probabilistic rules alone cannot account for the multitude of factors that contribute to human behaviors and the difficulty of creating mathematical representations of those components. Stone points out causal issues of many of these models, problems of equifinality – arriving at the same end point or conclusion by many potential means – a common issue of archaeological models (Hodder and Orton 1976; Crumley 1979; Stone 1996, 7). He addressed the problem of equifinality and the concept of settlement rules, with many intersecting overriding factors leading to a shared final point (the observed pattern), by thinking of the priorities by “their varying strength” (Stone 1996, 8). This concept that many rules with priorities of varying strengths contribute to the decision-making framework behind settlement locations formed the foundation of settlement ecology.

Theory of Settlement Ecology

Building on the intellectual scaffolding of the geographers and settlement archaeologists, Stone (1996) identified that each of the geographic and archaeological models influenced an aspect of settlement patterning and agrarian land use, with details of each interacting in dynamic and unexpected ways. His objectives in the study of the Kofyar people were to understand the mechanisms governing agrarian settlement patterns and attempt to predict and explain why people settled at a given time and place on the landscape. The pursuit of these questions led him to create the theory of settlement ecology. Among the many ideas operating in Stone’s conception of settlement ecology was the idea of priorities of varying strengths, in that the many rules that govern settlement and agrarian land use push and pull with one another, resulting in often unexpected outcomes from a set of models (Stone 1996, 8).

The rules determining settlement patterning can include a near-infinite list of factors, but Stone utilized some of the foundational principles established by geographers like von Thunen, Christaller, and Boserup. Stone termed the transportation principle demonstrated in von Thunen and Christaller’s models as “proximity access theory” – the key implication that closeness to an important feature on the landscape is preferable to distance to that feature (Stone 1996, 14). Despite that preference, other motivations or rules can override the benefits of proximity. For example, Boserup’s theory of agricultural intensification is at odds with proximity access. As populations are drawn closely to desirable fields because of proximity access, the effort required for progressively intense cultivation makes the land less desirable. Simultaneously, occupants are pulled elsewhere to the central places of Christaller’s model, under the same laws of Central Place Theory – goods and services like labor markets, religious facilities, or defense exist at the concentrated settlement locus. The dynamic push and pull between settlement rules, reacting to past and present choices of populations, is the core ramification of the study.

Stone’s focus on Kofyar agriculture led him to first lay out an agroecological overview of agricultural intensification, building on Boserup’s idealized models with models of ecological adaptation, forming a rough model for what agrarian settlement systems should look like without the effect of historical and cultural factors (Stone 1996, 32–56). In observing the Kofyar, he observed how weak social factors originally drew residences together, while stronger factors, like improved soil elsewhere, led to dispersed settlements nearer to the advantageous agrarian landscape. Intensification further altered the importance of water for settlements, drawing the populace towards agriculturally superior soil (Stone 1996, 132–61). In illustrating the successive steps of the push and pull of settlement factors, he emphasizes that intensification is not a given. Rather, the populace makes choices between intensification and abandonment, where other “rules” like the social organization of labor contribute to the outcome (Stone 1996, 182).

In a rebuttal to some of the geographic principles at play, Stone showed that farmers’ conception of proximity in traveling from their homestead to fields operates on a “threshold” model, where distances shorter than 700 m did not affect their willingness to travel (1996, 132). Existing geographic proximity models calculated proximity as a continual spectrum where 100 m is closer and thus more preferable than 150 m. The addition of the real perception of distance allowed Stone’s study to better understand the motivations behind farmers clustering into small, nucleated homesteads. In comparing the distribution of Kofyar farms and settlements with the idealized arrangement suggested by Boserup’s agricultural model, he observed the farmers do not act as her intensification model suggested. Rather the Kofyar settlement was “not an optimal solution to the agroecology of the Namu Plains,” adapting to the pressure and rules of settlements rather than strictly conforming to them (Stone 1996, 186).

Critiques of Stone’s study of the Kofyar and propagation of a methodological toolset of settlement ecology found few skeptics, with criticisms limited to an absence of utilizing cross-cultural information concerning frontier expansion (Picchi 1998, 174). Stone’s later work departed from the broad studies of settlement patterning, focusing narrowly on agricultural decision-making. Using improved GIS technology and ethnographic data from new study areas, he and co-authors modeled how Indian farmers chose seed types based on social pressure rather than maximizing crop yields (Flachs et al. 2017), how farmers make decisions about crops and land as part of a social display (Stone 2018), and why farmers choose not to plant vitamin-rich Gold Rice (Glover et al. 2020). While archaeologists cannot fully replicate Stone’s contemporary observation of the Kofyar, simultaneously tracking settlement expansion and evolution with ample information on social and cultural factors, his use of disparate data types and the dynamic interplay of complex settlement rules established a new methodology for understanding archaeological settlement.

Archaeological adoption of the methodological approach of settlement ecology followed Stone’s 1996 publication, acknowledging the theory explicitly and implicitly. Landscape archaeologists using settlement ecology also used the theoretical framework of historical ecology. Historical ecology emphasizes the dynamic nature of human-landscape interaction, with the landscape as an active participant in the creation of human cultural activities, not a static background that acts like a constraint or limitation on adaptation (Crumley 1994; Balée and Erickson 2006; Balée 2006). Like historical ecology, settlement ecology “acknowledges that landscapes are the products of people’s interactions with their environments” (Anschuetz et al. 2001, 168). Kellet and Jones (2017b), in the introduction to their comprehensive edited volume The Settlement Ecology of the ancient Americas, define and outline the five principles of settlement ecology in archaeology, building on Stone’s foundational anthropological work with direct application to archaeology.

Settlement ecology is applicable to “societies of all types,” with any “specific characteristics (e.g., degree of social complexity, mobility/sedentism),” and any era of human occupation, although Stone (1996, 5) proposed a model narrowly applicable to agrarian societies.

In contrast to processual settlement archaeology or geographical models, settlement ecology is a “time and space contingent” methodological approach requiring detailed knowledge of the “specific and local environmental, social, political, economic, ideological, and historical conditions” that limits universalizing cross-cultural comparisons.

Ecological interactions between entities lead to the “push and pull” of settlement prioritization, where reactions and adaptations to “ecological conditions, needs, pressures, and relationships” lead to subsequent reactions and adaptations.

The human agency of “conscious decisions made by people” in response to environmental, cultural, and social factors creates settlement patterns. Preexisting environmental traits or social characteristics do not determine settlement patterning without the intentional choices of people.

Spatial relationships between physical landscape features, settlements, cultural boundaries, social traits, and other factors are the primary analytical toolset of settlement ecology. Kellet and Jones argue a dichotomy between sites and non-site landscapes is integral for the use of GIS and other spatial technologies, the “best methodological approach in which to unravel the complex nature of prehistoric settlement patterns” (Kellett and Jones 2017b, 11–13).

While not all studies utilizing settlement ecology wholly follow all five principles, they form a helpful framework in which to structure research projects. Unlike Stone’s ethnographic and anthropological analysis of the Kofyar, archaeological projects cannot simultaneously observe the changing trajectory of settlement patterning and collect detailed information about ethnic divisions alongside environmental documentation. Rather, archaeologists must reconstruct the rules of settlement and associated prioritizations by isolating each factor, analyzing its impact, and qualitatively reconstructing the strengths and interplay of factors. Through the use of GIS, environmental reconstructions, and archaeological material culture, scholars used settlement ecology to expand on concepts of proximity and movement, population reconstruction, agricultural intensification, intergroup violence, and settlers’ decision-making framework.

Applications of Settlement Ecology

As Kellett and Jones (2017b, 13) argue in their listing of principles of settlement ecology, GIS and other spatial technologies greatly enable the analysis of spatial relationships at the core of settlement ecology. Stone’s (1996) volume on the Kofyar settlement ecology utilized early versions of GIS, devoting an entire chapter to quantitatively analyzing the spatial positioning of settlements, ethnicity, agricultural fields, and environmental data, supplementing the qualitative and descriptive explanations of settlement decisions. Advances in computer technology enabled GIS to better comprehensively manage environmental, cultural, historical data alongside multi-scalar archaeological data, enabling more advanced quantitative studies integrating and weighing the competing variables that contribute to settlement decisions (Maschner 1996; Wheatley and Gilling 2005).

GIS and other spatial tools granted the ability to better manage spatial data as well as create new types of datasets and more rigorously analyze the significance of correlations. Using freely available datasets like Digital Elevation Models (DEM) collected from satellites, scholars utilizing GIS can derive environmental and topographic data related to settlement decision factors. Least Cost Paths (LCP) calculate a route between two points by determining the least amount of energy to move from one pixel to another, using various types of cost surfaces though most often elevation and its derived slope, creating a line that represents the most natural path of movement between two points (Conolly and Lane 2006; White and Surface-Evans 2012). Another GIS operation calculates the viewshed from a point, often an archaeological site, outputting the area visible from that point (Jones 2006; Wheatley and Gillings 2000). A related GIS algorithm calculates the topographic prominence, determining the point of highest elevation in a given area (Llobera 2001; Christopherson 2003). Topographic prominence can help determine which locations have the most defensibility while viewsheds reveal which sites are most visible, a useful characteristic for religious or ritual places. Further, the addition of metrics and evaluation of statistical significance, such as Ordinary Least Squares (OLS), alongside GIS facilitated validation of proposed settlement decision factors deviating from expectations (Hasenstab 1996; Kvamme 1999).

An important feature of settlement ecology’s determination of the motivations and factors behind settlement decisions is the environmental characteristic of the study landscape, specifically factors related to sustenance. As populations require food and water as a core necessity, obtaining sustenance is not only one of the strongest priorities of settlers but the first principle for understanding the expected settlement patterning. Studies of contemporary populations can collect information concerning agricultural suitability or utilize land use data from global and regional geospatial databases, but archaeological studies face additional difficulties.

Studies reconstructing the environment of the last thousand years can utilize modern environmental datasets with only moderate changes. Working backward with land use data like LANDSAT multi-spectral satellite imagery or governmental agricultural surveys with recorded documentation of known changes in the global and regional climates, like sea-level rise or river damming, yields largely accurate information regarding agricultural soil quality and water accessibility (Hasenstab 1996; Maschner 1996; Jones 2010; Jones and Ellis 2016; Kellet and Jones 2017a). Even extrapolating past conditions from extremely detailed soil class data, using the current topography to model and extrapolate changes over time, often results in useful, albeit imperfect, data (Posluschny et al. 2012). However, studies further in the past must use more complex models to reconstruct the environmental and possible agricultural capability. Models using paleo-environmental data are the most comprehensive method for reconstructing the entire landscape and habitat of a given study area. Paleo-floral data from archaeological sites, including pollen (Bottema 1999) and charcoal (Guibal 1999; Vernet 1999), enable narrow reconstructions of an area’s past environment by examining the types and health of vegetation during a site’s occupation.

Combining site-level proxy environmental data across multiple sites leads to a more detailed understanding of the broader landscape in a region. By integrating floral and faunal paleo-environmental data into a GIS and simulating conditions, Brouwer Burg (2013) created an accurate facsimile of the landscape of Post-Glacial central Netherlands. In instances where the archaeological data lacks high-resolution paleo-environmental data or high-quality modern land use information, the archaeological record can assist in environmental reconstructions. In a recent article, Hughes et al. (2018) created a cross-cultural model for reconstructing land use by inputting dozens of variables, including the known caloric intake of populations in the area, soil conditions, the dietary archaeobotanical evidence, and settlement size. Using a concentric circle model of land use, parallel to von Thunen and Christaller’s proximity-based models, enables overlaying a proposed division over the observed environment. Unfortunately, all the detailed environmental reconstructions rely on inputting extensive high-quality local data or the availability of comprehensive geospatial datasets. While not all study areas have access to that material, settlement ecology studies utilize accessible information in parsing the factors of settlement decisions.

##### Case studies

Mobility and transportation are significant factors in influencing settlement decisions and the increased accessibility of GIS-assisted tools like LCP, based on geographical and culturally based cost surfaces. Originally based solely on DEMs and the physical restrictions of traveling the topography, cost surfaces define the cost of traveling from one point to another (Gietl et al. 2007). While DEMs and their derived slope are historically the most commonly used cost surface by archaeologists, given the accessibility of base data and experimental movement evidence, archaeologists increasingly use other constraining variables like vegetation, soil type, route visibility, or socio-cultural factors (Llobera 2000; Verhagen et al. 2019, 226-30). With the physical surface of a DEM as the background, recent studies added factors such as indigenous travel knowledge (Supernant 2017), pilgrimage sites (Kristensen and Friese 2017), and visibility of cultural waypoints (Bell and Lock 2000) as additional costs in the creation of cost surfaces. Merging physical factors like slope or soil type with the important but incomparable cultural parameters requires using statistical tools like multi-criteria analysis and the weighting of inputs to evaluate significance for the creation of LCPs (Howey 2007; Howey 2011). Parsing out the impact of the many factors contributing to a cost surface parallels settlement ecology’s codification of rules and priorities of varying strengths.

Minimizing travel cost and distance by increased proximity to points of interest is one of the most powerful deciders of settlement position and studying its role emphasizes the other motivations pushing against reducing costs. Carballo and Pluckhahn (2007) generated transportation corridors, a function related to LCP that outputs the best corridor to move through a region, to evaluate the growth of urbanization and political expansion in Tlaxcala, Mexico. The corridors’ path and relative ranked travel time parallel the growth of ceremonial centers and territorial expansion, suggesting that accessibility was a primary motive behind cultural and political changes. Loughlin’s (2017) settlement ecology based study of the small El Melón basin builds off previous work on generating settlement corridors, modeling how the collapse of the nearby La Venta created a power vacuum that precipitated a new concentration of power and economic exchange in El Melón. The beneficial characteristics of the physical landscape led to increased trade while political organization pushed towards further growth and consolidation. Herrera’s (2017) work in the same volume explored how the use of topographic markers like glyphs served as mnemonics for navigation in the highlands of Columbia. The markers served as central places, anchors for attracting settlement into the mountainous micro-environments, leading to diverse settlement types adapted to the ecological niches with “flexible social networks” (Herrera 2017, 216). Initiated as a tool to assist in navigating the difficult topography of the mountains, the waypoints pulled settlement towards arable portions of the landscape, creating a feedback loop between agricultural availability and proximity.

Stone’s analysis of the Kofyar agrarian population, using the theories of agricultural intensification, demonstrated the impact of agricultural variability and field use on the macro-trends in settlement decisions. Settlement ecology studies of archaeological populations first require estimates of the size and makeup of the populace to investigate the dynamics of fields, proximity, and intensity. Brannan and Birch (2017) compare the roofed area at the Mississippian site of Singer-Moye with comparable sites to estimate population by period and conclude its population was directly affected by the utilization of the surrounding environs. To evaluate the effect of drought and adaptions to wet or dry periods in the American Southwest, Ingram (2017) used the sum of rooms in each watershed by period as a proxy for the watershed’s ability to support high or low levels of population. Comparing the counterfactual situation where drought directly causes drops in population and subsequent rises during wet periods indicated that people in high-density areas were more likely to move as a response to drought. Lemonier (2017), lacking visible agricultural structures for the study of agrarian adaptations of the Maya Lowlands, used the position of households and neighborhoods to extrapolate likely field positioning. In comparing Bio Bec’s hypothesized agrarian spatial layout to the documented fields of La Joyanca, the household positioning method yielded accurate results. Using locations of the fields, households, and elite structures, Lemonier determined that La Joyanca residents were pulled towards greater proximity to elite residences.

Few settlement ecology studies, even in the only edited volume dedicated to operationalizing the theory, fully explicate the long list of settlement factors and the weight that occupants assigned each in their decision-making framework. Jones (2017) attempts such a task by creating a simple model to evaluate which settlement factors deviate from expectation. The model is based on the hierarchy of risk, the inverse of Maslow’s hierarchy of needs, where factors are ranked from highest to lowest priority, with sustenance scarcity the highest settlement risk factor and site vulnerability one of the lowest. A multi-layered risk map, created from the environmental and cultural factors around Piedmont Village Tradition settlements in the American Southeast, represents the amount of risk at every point in the study area (Jones 2017, 39-42). The average value of each contributing factor in a 2 km catchment around each known site was calculated and compared against the expected ranking of risk mitigation. Deviation of average risk factors from the expectation in the hierarchy of risk indicated when settlement decisions were influenced by other influences, such as warfare leading to increased defensibility in lieu of water accessibility. Based on the same underlying Piedmont Village Tradition archaeological and environmental data, an earlier article by Jones and Ellis (2016) compared risk factors at each site versus a random sample of background points. Running a discriminant function analysis outputted quantifiable metrics of the most and least important factors but, unlike Jones’s 2017 book chapter, did not explore the settlers’ decisions.

Comparison of the observed factors against the counterfactual of an idealized or random situation is a useful framework in the explication of multiple settlement decisions, as deviation from an expected situation warrants explanation. Jazwa and Jazwa’s (2017, 157-8) article studying settlement patterns of Bronze Age Messenia bases its counterfactual idealized settlement model on “ideal free distribution” (IFD) of habitat suitability. IFD is a Human Behavioral Ecology model that measures habitat suitability and how settlement spreads into new habitats as population density increases in existing habitats. The authors compared the size, hierarchy, and distribution of archaeological sites in Messenia against the predicted IFD, observing a high degree of conformity to the ideal model, indicating the Bronze Age settlers based their decisions primarily on the environmental conditions of the landscape. However, the primary deviations from the IFD occurred related to the relationship of sites to the main elite center at the Palace of Nestor, suggesting that the cultural pull of the palace affected nearby settlements more than those further afield (Jazwa and Jazwa 2017, 164-67). The use of multi-factor risk and suitability models compared to idealized or hypothetical distributions enables settlement ecology studies to evaluate many of the decision-making factors contributing to settlement but requires robust datasets of environmental and archaeological data. In studies without high quality or large quantities of data, isolating variables over time is an additional method for understanding some of the factors that contribute to settlement outcomes.

Settlement Ecology of Sidekan

The question underlying the research objectives of this dissertation is what were the factors that motivated settlement and abandonment in the Sidekan subdistrict. Specifically, why are the sites where they are? Why did the residents choose to settle in this area? Is the prevalence of Muṣaṣir era sites an accident of discovery, or does it represent the actual disproportionate types of settlement in the archaeological record? Landscape archaeology cannot wholly answer the final question but analysis of the characteristics of the known sites, rather than a focus on the unknown, reveals qualities of the settlement pattern that suggests Muṣaṣir’s existence brought attention and prosperity to this small network of valleys. As the previous literature review section demonstrates, scholars approached these questions from various directions, from purely quantitative with the use of GIS to extremely qualitative analyses of written and ethnographic records. Studies of movement and accessibility provide insight into the significance of those factors affecting the chronology of Sidekan and Muṣaṣir while the land use in the Topzawa Valley around Gund-i Topzawa reveals aspects of the growth and contraction of the region’s settlement.

Given the biased nature of the Sidekan survey data – biased through discovery methods alongside road cuts and following the knowledge of pre-existing sites – many of the techniques that rely on full area coverage and a much larger set of sites are unsuitable for this project. Rather, I use two techniques to focus on three factors: movement corridors to explain the origins of the earliest material in Sidekan and the micro-analysis of fields around excavated sites to understand land use and intensification.

Chronology and Settlement Change in the Sidekan Region

Evidence from the region’s historical overview (Chapter 3), archaeological excavations in the Soran district (Chapter 2), and excavation and site surveys in the Sidekan subdistrict (Chapters 4, 5) indicate notable occupation began in Sidekan during the Late Bronze Age (LBA). Radiocarbon results from Gund-i Topzawa East provide the earliest archaeological date in the Sidekan subdistrict, the 13th-12th centuries BCE. The earliest historical reference, of Aššur-uballiṭ I, the subduer of Muṣru, in the 14th century, suggests Muṣaṣir existed in some form by at least that century. In contrast, archaeological artifacts from Soran date as far back as the Paleolithic Period and include Neolithic Period, Early Bronze Age (EBA), and later occupation.

While non-existent in Sidekan, evidence of pre-LBA occupation is plentiful a few kilometers away, on the Diana Plain to the west and in the Urmia Basin to the east. Solecki’s (1998) cave survey of the Baradost and Safar’s (1950) excavation of the cave sites of Bastoon and Hawdian contained Paleolithic and Neolithic artifacts. Safar’s cave soundings, up to 10 feet deep, included distinct Neolithic and Early Bronze Age type wares (e.g. Hasuna, Ubaid, Early Dynastic, Uruk) typical in Mesopotamia and Iran. Gird-i Banahilk’s extensive excavation of Halaf material culture material demonstrates substantial Neolithic occupation on the core of the Diana Plain (Braidwood and Howe 1960).

RAP’s excavation of Gird-i Dasht, the only major archaeological mound in the Soran district, recorded multiple examples of the Khabur Ware ceramic type, a clear indicator of Early and Middle Bronze Age (MBA) occupation (Oguchi 1997). This unique painted ware spread from Mesopotamia across the Near East, into the intermontane valley systems of the Trans-Tigridian corridor and onto the Iranian Plateau, specifically at the site of Hasanlu, located ca. 50 km east of the Kelishin Pass (Danti, Voigt, and Dyson 2004, 586–92). The absence of Khabur Ware in the excavated and surveyed material from the Sidekan subdistrict, with its presence to its east and west, provides circumstantial evidence that occupation by pottery-making populations did not begin in a significant way until at least the LBA.

While the absence of pottery or historical records is not proof that the area was unoccupied, the environmental characteristics would suggest the populace were likely transhumanist pastoralists of some type, without evidence easily detectable through archaeological survey. However, unlike the cave-rich limestone Baradost Mountains, the geologic character of the Sidekan subdistrict is ill-suited for cave formation (Jassim and Goff 2006; Sissakian 2013). As a result, Sidekan’s settlement desirability is far less the valleys to the west, surrounded by caves, and would likely not have attracted large transhuman populations. Thus, the archaeological evidence is consistent with sedentary occupation beginning in the mid-second millennium and presents the research question of why sedentary occupation emerged at that comparatively late date.

While one could propose hundreds of possible reasons why settlement in the Sidekan subdistrict did not begin until the comparatively late LBA, the area’s isolation is characteristic underlying historical and contemporary discussions of the region. Movement into and out of Sidekan is the foundational principle of access and isolation and forms the theoretical and methodological approach for explaining the impact and change of the region’s isolation. While the emergence of Muṣaṣir as a political entity in the late second and early first millennia could have served as a cultural catalyst for sedentism, that inverts the cause and effect – Muṣaṣir required a pre-existing population. However, a new form of movement entered the Near East during the second millennium, instigating political and cultural changes elsewhere: the horse and its associated riding technology.

While the domestication of horses occurred as early as the fifth of fourth millennia on the Eurasian steppe, horses only became commonplace in Mesopotamia and Iran by the early-to-mid-second millennium (Anthony 2007, 397-403). Zooarchaeological evidence of domesticated horses from sites in Central Asia occurs by at least in the fourth millennium, although the extent of domestication as pack animals or for riding remains a question (Kohl et al. 2006, 138–40). As the genetic markers of equid domestication are insufficient for identifying horse domestication, given that domesticated males can breed with wild mares, the wear on teeth from biting bits and pictorial depictions serve as the primary indicators of the spread of the animal (Anthony 2007, 193-220). The only skeletal evidence of equids in Mesopotamia and its immediate environs until approximately 2500 BC was of onagers (Downs 1961, 1196). However, Mesopotamians were aware of horses before that time, with Ur III texts referencing them as the “ass of the mountains” (Anthony 2007, 416).

Art historical depictions in the third millennium show rudimentary carts and chariots towed by donkeys, onagers, or other pack animals like oxen (Moorey 1970). By the early second millennium, terracotta plaques begin showing people riding horses, and in ca. 1900 BCE, a cylinder seal from Karum Kanesh in Anatolia depicts a horse-drawn chariot (Littauer and Crouwel 1987, 41; Anthony 2007, 403). Skeletal evidence of horse bones and teeth with wear patterns associated with bits and riding occurs between 2100-2000 BCE at the sites of Malyan and Godin Tepe in Iran, the first zooarchaeological signs of domesticated horses entering the Mesopotamian cultural sphere (Anthony 2007, 416). Textual documentation parallels the spread and adoption of horse riding in Mesopotamia and Iran. In the eighteenth century, texts from Syria describe packs of horses harnessed together with grooms and trainers at Mari (Moorey 1986, 198). However, horseback riding had not reached ubiquity, as a contemporary text condemns Mari’s king, Zimri-Lim, for riding a horse (Anthony 2007, 418). Full economic and cultural adoption of horses for transportation and warfare did not occur until migrating groups underscored the animal’s benefit.

Domesticated horses and their associated riding technology spread from the Central Asian steppe outwards, east and west, alongside trade and the migration of Proto-Indo-European riders (Anthony 2007). Horse bones at sites in eastern Anatolia from the Early Bronze Age support that migration from the steppes was the origin of horses (Collins 1996, 24). While domesticated horses spread peacefully through trade, the full-scale adoption of horses followed eastern ethnic groups' utilization of the animals for warfare. In the MBA, Kassites and Mitanni conquered populations of Babylonia and Syria, respectively, due to their expertise in horse rearing and militarization. The name of the Mitanni, maryanni, becomes associated with horse warriors because of their equestrian proficiency (Boyce 1987, 508). Mitanni, an Indo-European elite class ruling over an ethnically Hurrian population in Syria, as well as the likely Indo-European Kassite rulers of Babylonia, were early adopters of the horse-drawn chariot for warfare (Moorey 1986, 197). Kassite texts extensively discussed horses, horse breeding, and aspects of charioteering, emphasizing a core characteristic of their power (Malko 2014). With the display of military prowess, the bulk of the Mesopotamian populace adopted horses for warfare, commercial activities, and improved conveyance by the latter half of the second millennium (Kohl et al. 2006, 141).

Along with military benefits, horses brought extensive economic and transportation advantages. Herding, for example, became more efficient with horseback riding. A pedestrian pastoralist can herd 200 sheep while one on horseback can drive 500 (Anthony 2007, 222). Transportation assisted by horses shows similarly significant increases. Animals like the ox, donkeys, and onagers were harnessed to sleds or wagons for transportation but could not move goods as quickly and as far as horses (Wilkinson 2014, 48-49; Kohl et al. 2006, 145). A two-wheeled cart, more well suited for horses than donkeys, has 40% less draft than a four-wheeled version, resulting in 60% more efficient transportation of the same amount of goods (Anthony 2007, 65-69). Compared to an ox, a horse can walk twice as long with a full load, four hours versus two, and travel 60 km in a daily workload compared to the ox’s 25 km distance (Bökönyi, 1991, 553). This drastically improved transportation ability led to cultural and political changes in the Middle and Late Bronze Ages, increasing interconnectivity between urban centers and rural settlements. As the horse’s presence in the Near East occurred in the centuries preceding the earliest archaeological evidence and textual references to Muṣaṣir, could this phenomenon have enabled the beginning of sedentary occupation and later developments in the area of Sidekan?

The emergence of domesticated horses, riding technology, and carts assisting in transportation occurring nearly simultaneously to the LBA archaeological material in the Sidekan subdistrict warrants an analysis of the significance of the nascent transportation method. Horses' impact on transportation and connectivity between sites in the Sidekan subdistrict and surrounding regions must be at a level necessary to spur the beginning of growth in sedentarism. A method in determining the impact is calculating the travel time and distance between the Diana Plain and the Sidekan subdistrict highlands to compare pedestrian versus horse transportation and movement. LCP between the Early Bronze Age (EBA) site of Gird-i Dasht, on the Diana Plain, and Mudjesir, the proposed core of Muṣaṣir, yield different routes that are combined with data on travel time.

A major variable used to calculate the cost of crossing terrain that generates the LCP is the velocity of travel, denoted in GIS as the vertical factor table (Becker et al. 2017). This variable conveys how different slopes, going upwards and downwards, change the speed or provide additional friction for movement. For most archaeological LCP analyses, the path modeling is based on the hiking equation by Tobler (1993), created by experimentally observing how humans on foot traverse the terrain at different slopes. That method has proven effective for many studies, even if a bit simplistic (Conolly and Lane 2006). While far less utilized in the literature, some archaeological studies attempted using LCP with non-pedestrian locomotion, including horses (Sunseri 2015; Verhagen, Nuninger, and Groenhuijzen 2019). The critical difference for generating horse-based LCP is the change in velocity, indicated with the vertical factor table. The most common calculation uses Tobler’s hiking function as a base and multiplies exponential function by 1.25, described by Tobler (1993) and based on earlier research of horse cost movement by Imhof (1950). Other publications of cost formulas used racehorse velocity on slopes and equestrian treadmills for the dataset (Eaton et al. 1995; Self, Spence, and Wilson 2012). Archaeological LCP studies most often use Tobler’s modified hiking function, although the other formulas reveal that the horse variation of Tobler’s formula overemphasizes the benefit of horses on steep slopes (Lugo and Alatriste-Contreras 2020, 4–6).

To evaluate the possible benefits of horse assisted versus pedestrian transportation over the mountains surrounding Sidekan, I generated two LCP between Gird-i Dasht and Mudjesir, using Tobler’s pedestrian hiking and modified horse hiking functions. The origin was set at the site of Gird-i Dasht to model a hypothesized travel or trading journey to the population center at Mudjesir. Using ArcMap 10.8, I first generated a slope raster from a DEM as the initial cost surface, which I then combined with a cubically weighted ranked waterways raster to account for the difficulty of crossing large rivers. The Path Distance used the cost surface, with a vertical factor table based on Tobler’s hiking equation as one version with another using the modified hiking equation to represent horse travel. The Path Distance function generated a raster representing the relative costs of traveling from pixel to pixel starting at the site of Gird-i Dasht. The Cost Path function created rasterized routes between the two sites representing the least amount of travel expended from those two surfaces, with one version outputting a metric for travel hours and another representing the accumulated cost (Figure 6.1).

The dual LCPs reveal minimal differences in routes between the two modes of transport, but the associated travel times and efficiencies indicate substantial benefits for horse-assisted transit. Both routes’ rough corresponding paths nearby the modern road from the Diana Plain to the town of Sidekan support the accuracy of the LCP. However, the pedestrian LCP shows a 4% longer route (18.3 vs. 17.6 km), avoiding the steeper slope of the descent into the Hawilan Basin by descending on the basin’s edges. While the difference in distance is minimal, the one-way route time crosses the threshold for significance. The pedestrian route is estimated to take 4.8 hours of constant travel while the horse’s last 3.89 hours. As noted, horses can travel four hours with a full load before requiring rest and ride 65 km total in a day (Bökönyi, 1991, 553). At the generated time (3.8 hours) and distance (17.6 km), a horse could travel round-trip between Gird-i Dasht and Mudjesir in one day, while a pedestrian or an ox would likely require rest before the return. In addition, the overall difference in costs, 29.5 for horse and 45.95 for pedestrian, equal a 43.6% overall improvement in horse transportation. With the tall grass on the mountain slopes supporting traveling horses, the EBA occupants could far more easily access and travel to the valleys of the Sidekan subdistrict.

The horse’s impact on transportation to Sidekan, while demonstrable and substantial, is not, on its own, sufficient to prove horses led to the founding and development of polity that became Iron Age Muṣaṣir. Further research, specifically of the excavated faunal bones, may provide additional evidence for the appearance and importance of horses. The prevalence or absence of horse bones and bit-worn teeth in the lowest levels of the site of Gird-i Dasht and the existence of bones around the site of Mudjesir could show the capability of horses to improve communication, trade, and movement between the sites as evidence of their use at this time. However, the proposed propagation of domesticated horses and their use for transportation as one of the factors that affect the rules of settlement ecology that Stone discusses provides a means of exploring the apparent establishment of sedentism in the Sidekan area.

While horse-based transportation may have been a factor precipitating the start of archaeologically visible sedentary occupation in the Sidekan subdistrict in the LBA, other factors contributed to the contraction of occupation after the 8th-7th centuries BCE. The biased and limited survey sample size constrains the direct evidence of contraction or abandonment in Sidekan, a few data points support at least a moderate reduction in settlement: the burning of the final occupation levels at Gund-i Topzawa in Iron III, parallel burning at sites along the Topzawa Valley, multiple Achaemenid burial sites in the valley, nearly non-existent Achaemenid settlement evidence across the subdistrict, and a near absence of post-Achaemenid artifacts until the Islamic period.

The major destruction event at Gund-i Topzawa Building 1-W Phase B provided a bounty of artifacts and information about room use but showed little sign of violence towards the inhabitants. Given the building's type and quantity of objects, it is unlikely the residents abandoned the building, and a fire destroyed the structure after their departure. However, the abandonment of the upper levels and possible squatter occupation for some time, without an apparent later rebuild nearby, suggests the fire may have precipitated abandonment. The extreme destruction of Qalat Mudjesir on top of an Assyrian-style doorway raises the intriguing possibility of Scythian destruction of the area in parallel with their attacks against Urartu in the 7th century but requires further research of Qalat Mudjesir to establish that connection. The surveyed sites to the west, specifically Gund-i Manga with its comparable ceramics, displayed similar burning in the road cut section.

At Gund-i Topzawa, an Achaemenid burial formed the final archaeological phase after the building’s primary use. Further east down the valley, Ghaberstan-i Topzawa’s nearly contemporaneous burial suggests a transformation of the valley from settlement and occupation into inhumation on the outskirts of Muṣaṣir. Despite multiple Achaemenid or post-Achaemenid burials, archaeological evidence for Achaemenid occupation is almost non-existence. The Achaemenid style column bases at Mudjesir suggest their presence, at least at that site. However, survey and excavation of the Mudjesir fields and Qalat Mudjesir recovered no clear Achaemenid style pottery. In addition, the excavated and surveyed pottery of the whole area provides evidence of Sasanian-era occupation, ephemeral on the small plain of Sidekan, but no additional types until the Islamic period. The totality of these factors suggests settlement in Sidekan post-Iron III at the very least contracted from its peak contemporary to Urartu. Given the transformation in RAP’s data of the Topzawa Valley from a population center to a location for burials in the mid-first millennium, the study of the Topzawa Valley’s settlement organization can reveal not valuable data about land use in the Sidekan area but shed insights into chronological questions.

Population and Land Use in the Topzawa Valley

Despite the incomplete survey dataset for a study of Sidekan’s landscape, the extensively excavated and intensively surveyed site of Gund-i Topzawa provides a unique perspective to analyze the characteristics of settlement from the micro-level and build upwards towards regional conclusions. The settlement insights Gund-i Topzawa provides come from the use of rooms in the excavated Building 1-W Phase B, the types of archaeobotanical remains, and the detailed breakdown of other buildings at the site. Using those data with ethnographic studies enables reconstruction and modeling of population sizes and broad insights about land use around the site. While the other similar sites along the Topzawa Valley (RAP23, 21, 22) lack Gund-i Topzawa’s specificity concerning excavated material or the number and types of buildings, the distribution of site locations enables the expansion of conclusions about Gund-i Topzawa’s immediate environs to the whole Topzawa Valley. Further, understanding the land use in the Topzawa Valley provides a major pillar in the explication of Sidekan and Muṣaṣir’s settlement patterns. The room type and usage of the excavated parts of Gund-i Topzawa, largely from Building 1-W Phase B, provide relevant data connections to ethnographic and archaeological studies of population sizing. Archaeologists utilize various techniques for estimating the population of archaeological sites, from estimates of total site area, natural resources in the area, and extrapolation from features of individual dwellings, among others (Zorn 1994, 32–35).

In Kramer’s (1982) ethnography of the pseudonymous village of Aliabad, she lays out the features of each dwelling in the village, including the number of bins and ovens and the total dwelling and compound area. With that information, she included the number of families in each house. In total, she lists the characteristics of 30 houses of two stories, providing a dataset to find the average number of bins, storerooms, and square footage per family. The size of a family requires some discussion and overview of other ethnographies. Kramer lists the household size as ranging from 5.1-6.3 person while Watson’s ethnography of a nearby village lists four to five people, with a mean of 4.6 (Watson 1979, 47; Kramer 1982, 123–24). Others enumerate the family size between 3.5 and 8 individuals (Zorn 1994, 33). Given the wide range of Watson and Kramer’s numbers, ranging from 4 - 6, I use five people per family as a simplified value to encompass the findings of the many ethnographies examining pre-modern Middle Eastern cultures. Textual accounts from Mesopotamia could provide a different perspective, but the demonstrable difference in material cultures indicates that data may not be transferable to the Zagros Mountains highland people.

Kramer’s dataset of dwelling features and number of families yields the average numbers of families per feature used to extrapolate the population of Gund-i Topzawa. On average, the houses of Aliabad had 2.25 bins/family, 1.75 storerooms/family, and a total area of 46.6 sq/m per family (Kramer 1982, 114–15). The fully excavated Gund-i Topzawa Building 1-W Phase B contained bins and storerooms, enabling the calculation of families in the building with those metrics. 1-W Phase B contained two medium-sized bins and 2-3 storerooms. Rooms 2 and 3 were surely storerooms of some kind, while Room 1 seemingly served a dual purpose. According to Kramer’s observations, that yields 4.5 families using the bins and 3.5-5.2 families based on storerooms. Note that the bin dimensions described at Aliabad were significantly larger than those in Room 2. The Aliabad residents reserved the second-floor rooms for living space, which corresponds to the interpretation of Gund-i Topzawa.

Calculating the total square footage of Gund-i Topzawa Building 1-W Phase B is complicated by the second story and unknown extent of the southern portion of the building. However, the collapsed remains indicated that the second story extended only over Rooms 2 and 3. Measuring the building as only the visible extent, with two same-sized rooms over Rooms 2 and 3, the total square footage of Gund-i Topzawa Building 1-W Phase B equaled 73.5 sq/m. Using the dwelling space per family value yields 1.6 families in Gund-i Topzawa Building 1W Phase B. As the square footage represents the minimum possible size, 1.6 families should be considered the minimum size, not representative of the projected size. In addition, courtyards were an important component of household compounds, but the excavated material at Gund-i Topzawa provides little insight into the size or existence of courtyards. Therefore, the three metrics, bins, storerooms, and square footage, result in 4.5, 3.5-5.2, or 1.6 families at Gund-i Topzawa. Kramer notes that of the features of Aliabad, bins are most likely to correspond to population size. However, a conservative estimate, given the size of the bins and incomplete information on square footage, is 3 families living in this building at Gund-i Topzawa, with a total of 15 people.

Extrapolating the estimated population of one building at Gund-i Topzawa to the entire site requires assumptions based on the number of rooms in each building. Building 1-W Phase B had five rooms, including the two upper stories. Buildings 2W, 3W, and 4W each had two rooms visible but lacked the triangular wall that defined Building 1-W Phase B’s Room 3. Given the similar elevation in the section and types of structures, they likely had second stories, totaling six rooms. At roughly two-thirds the size of Building 1-W Phase B, their population can be estimated as two families, or ten people per building, equaling an additional 30 individuals at Gund-i Topzawa West. The population estimate at Gund-i Topzawa East is slightly more complicated. From the excavated material, that area of the site is not easily identifiable as contemporary to Building 1-W Phase B. However, the upper phase of Building 2-E, cleaned but unexcavated, parallels the rebuilding and reuse between Building 1-W Phase B and 1-W Phase A, suggesting at least some contemporaneous occupation in the east. With six rooms in the east, the estimated population is five families with 25 people. Thus the total population of Gund-i Topzawa in the 8th century is estimated at 70 people.

The total estimated population of Iron Age Gund-i Topzawa, while interesting, does not provide much insight into the land use surrounding the site and distribution of settlement in the valley. Another ethnographic study from the nearby village of Rust in 1956 provides detailed information on fields surrounding a settlement (Galloway 1958). The brief publication lists the number of houses (130), total population (700 people), and a detailed map of every field and its type around the village. While the distribution of crops cannot be directly compared to the Iron Age, as some cash crops were introduced from the New World, the fallow patterns and amount of cropland per person can serve as valuable proxies for the similar environment of Gund-i Topzawa. To accurately capture the area of the fields and the accompanying characteristic of the land, like slope, I vectorized the map and georeferenced the vectorized map in ArcGIS, converting it to editable shapefiles (Figure 6.2). The shapefiles provided the total area of each field type as well as the total field area (Table 7). Notably, as referenced by Galloway in the article, the amount of fallowed land was only 11.7%. While minimal fallowing is often a sign of intensification, as discussed by Boserup, Galloway’s description of the fields suggests the fertility of the soil requires less fallowing than more arid environments.

The georeferenced Rust field data indicates a total of 2.078 sq/km of fields, including the .244 sq/km under fallow. Assuming Galloway’s enumeration of the total population of 700 is correct, that equals 2968 sq/m of fields per person. In addition, I joined the slope derived from the DEM to the Rust field shapefiles, taking the average value for each field’s covered area. Although the exact field locations are imperfect, as the combination of georeferencing and a hand-drawn map from the 1950s does not yield perfectly located polygons, the slope for each feature type broadly aligns with the assumed slope. For example, the average slope of fruit trees is 33 degrees, the rivers are 16 degrees, and the remaining fields equaled about 20 degrees. Thus, the average slope of all the fields is 20.8 degrees, with a standard deviation of 7.24. While the village of Rust was larger than Gund-i Topzawa’s estimated population and Rust’s geography slightly differs from Topzawa’s, the nearly identical climates, similar topography, and cultural continuity enable a comparison to the land use around Gund-i Topzawa.

Table 8: Cropland Area in Vectorized Rust Fields

As a way to estimate if the accuracy of the fields per person and population of Gund-i Topzawa were roughly accurate, I created a raster that represents the possible arable land in Sidekan, including the Topzawa Valley. While many archaeological studies use complicated methods to derive the arable land, ranging from multi-spectral imagery of contemporary soil as proxies for agriculture in antiquity to derived geospatial analysis of time constant geographic features, I opt for a simpler method based on the observed Rust fields (French, Duffy, and Bhatt 2012; Codding and Jones 2013; Jones and Ellis 2016; Howey and Brouwer Burg 2017). This method took the range of slopes observed in the fields, i.e., any slope below 28 degrees, and a cost distance raster of 800 m distance from waterways. The result was a raster of a single value indicating possible agricultural land. As Figure 6.3 indicates, this is not a wholly accurate facsimile of potential agricultural land. It serves, rather, as a maximal view of agricultural use. Around the point of Gund-i Topzawa, I generated a polygon buffer that included 207,804 sq/m of arable land, as indicated in Figure 6.3.

Figure 6.3: Catchment of Agriculturally Capable Land around Gund-i Topzawa

The outline of the required arable land around Gund-i Topzawa for the given population estimates of the site and estimated field acreage per person derived from Rust presents two takeaways. First, the area consumes the immediate environs of the valley, including the equally fertile land to the south of the river. Thus, the combination of two estimated variables yields an area that, from the available data, passes the so-called eye test. Second, the southwestern border of the polygon forms a border with the immediate catchment of Gund-i Manga (RAP23). Without information on the number of walls, I cannot duplicate this procedure for that site, but the general size of the catchment can be assumed as similar in size, reaching further south. Unfortunately, we did not survey the road to the west of Gund-i Topzawa, so we cannot evaluate if that catchment area would overlap a neighboring site.

While only a single polygon based around two estimated measures, the land use around Gund-i Topzawa gives the insight that the populace likely extensively utilized the valley floor at the height of the settlement’s size. Assuming Gund-i Manga followed similar patterns, it is likely that during the Iron III period, the height of Muṣaṣir and Urartu’s focus on the area, residents of the Sidekan region used much of the Topzawa Valley’s high-quality agricultural land. With the Topzawa Stele’s location marking Urartian kings’ travel down the valley from Kelishin Pass, historical information further reinforces the importance and intensification of the valley during the period.

Using the same method of estimating the amount of arable land enables calculating the relative intensity of Muṣaṣir’s agricultural and occupation of the Sidekan region. While the archaeological estimate of Muṣaṣir’s population is impossible from the available data, Sargon II’s description of capturing the city provides a metric on which to base estimates. On line 349, the text describes taking 6,110 people of Muṣaṣir away to his camp. While Neo-Assyrian campaign accounts likely overestimate victories and number of captured enemies, using 6,110 people to generate a polygon of possible arable land around Muṣaṣir, based on the Rust person per field calculus, yielded an area roughly covering Sidekan, indicated in Figure 6.4. Much like the Gund-i Topzawa polygon, the results are rudimentary and not intended as representative of the true area covered. However, assuming Sargon II exaggerated the number of captives, but the overall population of Muṣaṣir was similar to his stated metrics, Sidekan and the kingdom seemingly followed the proposed intensification of land in the Topzawa Valley. The possible intensification of settlement to cover nearly the full extent of Sidekan by the Iron Age kingdom combined with the catalyzing force of horse transportation in the LBA simultaneous to the rise of Muṣaṣir in the textual record helps answer the original queries of settlement ecology in Sidekan.

Settlement Decision Factors for Sidekan

With multiple factors of settlement decision captured in the preceding sections, we can return to the central questions that initially spurred this specific study: Why are the site where they are? Why did residents choose to settle in this area? Is the prevalence of Muṣaṣir-era sites an accident of discovery, or does it represent the actual disproportionate types of settlement in the archaeological record? The available information cannot fully answer these prompts, but the insights about transportation access and land use intensification permit a narrative explanation of the spread and contraction of occupation in the area that indirectly addresses the research questions.

The available archaeological data suggest that occupation in Sidekan pre-Bronze Age was either minimal or confined to nomadic populations, with an uptick of settlement in the Late Bronze Age. Increased transportation mobility precipitated by the prevalence and accessibility of equine pack animals is postulated as one of the factors behind this increase. In addition, the reduced transit time from the Diana Plain and greater use of the Kelishin Pass connected the Sidekan valley system to neighboring cultures focused attention along the route between the Iranian Plateau and the Mesopotamian plains. Fundamentally, Sidekan and the outlying valleys are areas of marginal settlement that require additional effort for substantial agriculture. With interconnectedness enhanced through more expedient routes, settlement began to grow and expand.

As archaeological evidence of the extent of Late Bronze Age occupation in Sidekan is severely curtailed, with only scant physical data indicating activity at all, the current research project cannot determine the amount or intensity of that period’s settlement. However, the historical evidence of campaigns against Muṣru and its linguistic descendants indicate, at minimum, a loose political entity of some size in the region by that time. One aspect not directly concerning settlement patterns and answered with the discussed data is the early reference of Muṣru as a sacred or holy city. Assuming the correctness of that interpretation and accuracy of Musri as Muṣaṣir, the site's religious significance undoubtedly served as another catalyst for increased settlement. Further, if the holy city or cult center held meaning for the populace in nearby areas, the increased ease of access would serve doubly serve to facilitate settlement growth. The question of early Muṣaṣir/Muṣru’s religious position and importance for outlying groups directly connects to the origin of Urartu seen through the lens of the Ḫaldi cult. Chapter 7’s conclusions discuss the possibilities and implications of such association.

By the Iron Age, around the time of the emergence of Urartu around Lake Van, Sidekan/Muṣaṣir had almost assuredly grown to a sizeable entity with a concentrated core around the Ḫaldi temple. While Gund-i Topzawa Building 2-E may date to the 10th or 9th centuries BC, the construction of the drain at Mudjesir provides a clue to the state of Muṣaṣir before the Urartian conquest. The radiocarbon date of the drain is between 895 and 833 BC, which, given the nature of a drain, would indicate the drain ended use around that time. The thick fill above the drain was seemingly intentional, postulated as part of a leveling operation that would correspond to the elevation of Ḫaldi as Urartu’s supreme deity concurrent with the royal journey commemorated in the Kelishin Stele. Regardless of the nature of the rebuilt structure, the monumental drain in the 9th century demonstrates a well-built monumental center. In the outskirts of the political entity, at least in the Topzawa Valley, the settlement decision locations seemingly relied on the slope of the hillsides, placing settlements at the upper bounds of optimal agricultural land. By the 8th century, the population of these hillside settlements was intensely cultivating the valley. Given the relatively limited agriculturally capable land, the location of settlement enabled this maximal agriculture. The archaeological evidence cannot currently provide insights regarding the character and extent of Iron Age settlements in the western valleys and basins. However, the occupation surrounding the modern-day field of Mudjesir was far more extensive than that in the Topzawa Valley.

Determining the extent and date of abandonment at the end of Iron III related to Urartu, Neo-Assyria, and the major historical events largely relies on the burn layer at Gund-i Topzawa Building 1-W Phase B with other circumstantial evidence from the limited excavations of Qalat Mudjesir. The exact date of the destruction of Gund-i Topzawa is complicated by the radiocarbon date’s ambiguity owing to the Halstatt Plateau. However, the probability it was the result of Sargon II’s invasion is minuscule. Two questions influence that interpretation: whether the burning was before Sargon II’s campaign or after the Urartian reconquest and whether the burning resulted from military activity or a natural event that corresponded to a larger abandonment of the valley system. In either case, the extremely hot burning at Qalat Mudjesir, over a Neo-Assyrian style threshold, would suggest violent destruction of the temple after Sargon II’s occupation of Muṣaṣir. However, given the Urartian importance on Ḫaldi, a demolition of that structure without rebuilding is unlikely. Rather, a probable explanation is the Scythian migration and attacks against Urartu and eastern Neo-Assyrian settlements in 5th and 6th centuries BC also reached Muṣaṣir and the Ḫaldi temple. If so, the destruction and simultaneous chaos in Urartu decreased the attention and support of Muṣaṣir, causing, if not abandonment, contraction to the core around Mudjesir.

The nature of Achaemenid material partly explains the post-Urartian and Neo-Assyrian settlement of Sidekan in the area. Archaeological evidence of Achaemenid occupation is sparse and unevenly distributed. Excavated sites of the period include only burials at the uppermost layer of Gund-i Topzawa Building 1 and Ghaberstan-i Topzawa. We recovered no other Achaemenid material in the Topzawa Valley. At Mudjesir, despite the vast quantities of Iron III pottery, none of the collected pottery clearly dates to the Achaemenid Period. However, the Achaemenid style column-base provides at least circumstantial evidence that worship of Ḫaldi continued at that site. Combined with the latest known mention of Ḫaldi (technically “Son of Ḫaldita”) from the Behistun Inscription in 521 BC, a case exists that the Ḫaldi cult persisted through at least that time. With the archaeological evidence restricted to burials with elite goods, the cult possibly consisted exclusively of the religious facilities and their direct support, with worshipers of Ḫaldi from across the Achaemenid Empire visiting on pilgrimages of some support. The lack of domestic Achaemenid material could be explained by a change in settlement decision towards lower-lying areas utilized in the Iron III for agriculture. Still, regardless, that change would signify a changed prioritization away from maximal agriculture production.

By the Sasanian period, seen through the excavation of Sidekan Bank (ca. 5th-6th centuries CE), the population of Sidekan may have transitioned or reverted towards mobile or transient behaviors. That site had characteristics of a temporary location for storage of goods, with wood or other burnable materials covering the short-lived shelter. The subsequent periods are ill-represented in the archaeological and historical record, but ceramics of the Islamic Period suggests a possible increase in population by that point. The later archaeological evidence of settlements in Sidekan remains sparse, and survey has only begun the process of understanding the nature of Islamic material in the area. However, the accounts of the rise of the Sorani Emirate, Muhammad Kor’s forced conquest of the Pireseni tribe, and the travels of Jewish traders into Sidekan to meet with the tribal leaders indicate the population of Sidekan was largely transient and separated politically and culturally from the residents of the Diana Plain through at least the 19th century. The combined historical and archaeological evidence supports the hypothesis that the more significant amount of Iron Age ceramics and sites in the archaeological record reflects the reality of a uniquely large and settled population at that time, providing a temporary answer to the prevalence of Muṣaṣir-era sites. With the proposed settlement peak with Muṣaṣir, the question of why people settled in this area is in part answered by the nature of Muṣaṣir – a religious center with vital importance to surrounding empires. With the Ḫaldi cult and associated activities at Muṣaṣir, wealth and population flowed into Sidekan. If Sargon II’s account of sacking the Muṣaṣir treasury has any accuracy, the types of fine goods and metals, like silver and gold, in Muṣaṣir originated elsewhere, as the area has no know sources of those metals. The nature of the religious center, its early founding, and its connection to Urartu remain a major question in understanding Sidekan’s archaeological history.