Markofsky, S. (2015) Perspectives on an Alluvial Margin: Settlement Patterns in the Northern Murghab Delta, Turkmenistan. IRAN (in press).

Perspectives on an Alluvial Margin: Settlement Patterns in the Northern Murghab Delta, Turkmenistan

This paper has been accepted for publication in the journal IRAN.

This is a pre-proof version: please do not redistribute without permission of the author.

Steven Markofsky, PhD

Complexity and Socio-ecological Dynamics

Institut Milà i Fontanals

Spanish National Research Council (CSIC)

[email protected]

+34 934 423 489 (x290)

Abstract

This paper presents the results of the Northern Murghab Delta Survey, an intensive survey de-signed to investigate Bronze Age settlement distributions in a transitional zone between delta anddesert in south-eastern Turkmenistan. The visually obstructed landscape complicates effective identification of past settlement patterns, a problem that can be significantly lessened (if not eliminated) via the integration of remote sensing data with intensive fieldwalking survey. The survey data is thus presented within the context of past and present landscapes, leading to a re-in-terpretation of Bronze Age occupation as a combination of varied processes, strongly influenced by but not fully constrained to major watercourses.

I. Introduction

Pivotal to a firm grasp on the archaeology of greater Iran and Central Asia is an understanding ofmarginality. The concept of ‘margins’, socially and environmentally nebulous regions that straddle the line between habitable, sustainable environments and adverse, often infertile zones, has been the focus of significant study and substantial theoretical shiftsi. Once viewed through deterministic lensesii, in which environmental deterioration and the consequent need to control water resources necessarily drove social development; such interpretations have since given way to more symbiotic explanations that consider factors such as the development of microenvironments, socio-ecological co-evolution and, niche constructioniii. These environmentally liminal regions, therefore, have become increasingly recognized as complex socio-ecological entities, dynamic boundary regions that can offer clues to the relationship between people and their environments that ultimately characterised human habitation.

Among the more enigmatic examples of such marginal environments are the inland alluvial fans, often referred to as 'oases', that occur throughout parts of greater Iran and Central Asia. The endorheic deltas of the Hirmand (Helmand) River in Iranian Sistan and the Tedjen and Murghab rivers in Turkmenistan are good examples of such environments, in which the extraordinarily

Markofsky Page 1

rich archaeological record both belies and illuminates the adversity inherent in inhabiting these regions. Even today, that adversity is manifested in the social and environmental tensions between the need to support burgeoning populations and the sustainability of the very resources needed for that support. The increasing recognition of the fragile balance of such regions—and how easy it is to tip that balance, has fostered unique developmental trajectories with specific implications for settlement dynamics, land use, irrigation and other socio-economic factors. Over the past few decades, these 'oasis civilizations' have come under increased scrutinyiv, and such investigations have resulted in new insights into socio-ecological development in marginal regions.

One of the better-known archaeological landscapes of this type is the Murghab alluvial fan in south-eastern Turkmenistan (Figure 1). Here, towards the end of the 3rd millennium BC and for reasons that are still not clear, a highly distinctive cultural tradition appeared, apparent in many elements of its material culture. Aspects of this unique identity included new ceramic forms, intricate stone and metalwork, zoomorphic and geometric iconography and, perhaps most conspicuously, a new form of highly stylised geometric architecturev. The distinctiveness of the archaeological finds as well as the near-concurrent appearance of a similar material culture as faraway as Bactria, a region comprising present-day southern Uzbekistan and northern Afghanistan,led its discoverer, V. Sarianidi, to propose a single complex for the entire region that he dubbed the Bactria Margiana Archaeological Complex (BMAC)vi. Widespread interest in the cultural richness of these sites, most notably Gonur Depevii, encouraged a number of research programmes designed to better understand the archaeological significance of this region in whichhundreds if not thousands of sites once dotted the late 3rd and early 2nd millennium landscape (Figure 2)viii. However, the keen academic interest in the BMAC phenomenon has often constrained research to larger, more significant sites while under-emphasizing local dynamics that could deepen our understanding of the complex trajectories of human occupation on the alluvial margin. Fortunately, recent projects have begun to rectify this situation, and researchers have undertaken a series of local-scale projects that seek to re-examine both sedentary-nomadic and rural-urban interaction in the context of broader socio-ecological, geomorphological and hydrological changes in the deltaix. However, much of this recent work has focused on the final phase of Bronze Age occupation in the region, when the nomadic presence became more prevalent. Far less is known about the role of small sedentary sites that dotted the alluvial fan throughout the Bronze Age.

Markofsky Page 2

Figure 1: Central Asia (NASA Blue Marble imagery)

To gain a deeper understanding of local settlement dynamics in this kind of unique alluvial margin, this paper considers as a case study a comparatively remote archaeological landscape in the northern fringe of the palaeodelta known as Egri Bogazx. Although several small sites have been documented in this areaxi and limited excavations have been conductedxii, very little is known about these seemingly isolated settlements or how they fit into the broader settlement dynamics of the Murghab delta. However, the presence of substantial archaeological material suggests that this area can reveal important information about local and regional settlement patterns and provide insight into the nature of occupation in this marginal and transitional region.This paper therefore seeks to address the following research questions: 1) to what extent can intensive survey in the northern Murghab delta be used to infer settlement dynamics at local and subregional levels; and 2) what may be learned of settlement processes in marginal and transitional alluvial zones?

Markofsky Page 3

Figure 2: Previous surveys in the Murghab, based on data from the AMMD project. Black dots represent sites

II. Study Area and Research Context

The inland delta of the Murghab River is situated in a broad geologic basin dominated by the Karakum Desert, a vast, arid region that comprises approximately 80% of the land mass of Turkmenistanxiii. The delta as it exists today represents only the most recent chapter in a dynamichydrological, geomorphological and anthropogenic historyxiv. Over the millennia, underlying geological trends have shifted the alluvial fan several dozen kilometres westward; and both environmental and human factors have contributed to a southward retraction: a process that began in earnest towards the beginning of the 2nd millennium BC and bears significantly on interpretations of occupational dynamics during the periodxv. The marginal and transitional aspects of the delta are evident both in the climatic regime as well as the geomorphology and

Markofsky Page 4

taphonomy of the region. The climate is arid, marked by very hot summers and cold winters, with annual rainfall averaging around 100mm. Irrigation is therefore essential for agriculture, and in addition to the modern Karakum canal, a network of irrigation canals has marked the landscape since the late 3rd millennium, comprising both built canals and modified watercoursesxvi. Along the main delta channels and the more substantial canals grows the riverine or tugai woodlandxvii, a distinctive micro-environment that in some cases likely existed since antiquityxviii and in other cases has been developed recently in conjunction with the present-day delta system to increase cotton and wheat cultivation. Beyond these fertile zones, the landscape is increasingly desolate. Windblown sand and loess that partially result from erosion caused by recent agricultural activity, road/pipeline building or other human endeavours, is ubiquitousxix. Although large sand dunes are not prevalent in the present-day delta, their frequency increases towards the outlying regions of the palaeodelta, lending a banded appearance to the landscape which adumbrates the long, north-south ridges that lay beyond the alluvial fan to the north-east.

Interspersed throughout the delta are hard, cracked clayey surfaces known regionally as takyrs (Figure 3). These features are analogous to the 'playas' of other desert basins and are sometimes associated with earlier alluvial surfaces, although they can form in sandy regions where drainage is poorxx. The smooth surface does not support vegetation, and plants only develop when sand is able to accumulate on the takyr surfacexxi. The encroachment of wind-blown sands from surrounding dunes may thus facilitate vegetation growth, which in turn stabilizes the sands and results in complex and nebulous boundaries between dune and takyr. It is in these boundary regions, the takyr perimeters, where archaeological material often tends to aggregate.

Markofsky Page 5

Figure 3: Murghab landscapes: a) Low dune hills b) Takyr. Note the development of vegetation in the jeep tracks on the takyr surface

This complex interplay of environmental, geomorphological and hydrological factors has hindered a clear understanding of occupational dynamics in the Murghab, although significant progress has been made. Until recently, the predominant interpretation of settlement in the region was as a series of distinct ‘micro-oases’xxii, discrete micro-environments within which settlement was possiblexxiii. This model derives largely from an interpretation of the late 3rd and

Markofsky Page 6

early 2nd millennium Karakum Desert as an adverse environment of wind-deposited sands extant throughout the Holocenexxiv, a model that until recently had fostered a strongly deterministic viewof settlement potential that restricts the Bronze Age communities to isolated, sand-bounded oaseswith sufficient access to water and fertile landxxv, beyond which cultivation would have been impossible (Figure 4).

Figure 4: 'Micro-oases' and continuous occupation in the Murghab delta

Although the concept of the oasis civilisation persists to a degree, many of its central tenets have been questioned by researchers associated with the Archaeological Map of the Murghab Delta (AMMD) project, a joint Italian-Turkmen programme of integrated investigations that has conducted research in the region for over two decades. Employing a combination of satellite imagery, aerial photography, excavation and survey, researchers have attempted to reconstruct the hydrology and geomorphology of the region and examine these in relation to diachronic settlement changexxvi. Their work has resulted in a different model of not only the peopling of thedelta, but also of the development of the alluvial fan itself. The new model challenges the static interpretation of the Karakum and posits a mid-2nd millennium date for the onset of the desertification process, an interpretation based in part on the widespread distribution of surface pottery along transects linking key sites throughout the delta. These findings suggest to the

Markofsky Page 7

researchers that material is not restricted to discrete, oasis-based settlement groups as per the earlier models, but reflects a vast alluvial floodplain where settlement and irrigation agriculture were common and pervasive. Furthermore, this interpretation downplays the restrictive effect that dunes may have had on occupation and posits an integrated settlement system in which the hydrology and geomorphology of the alluvial plain facilitated rather than impeded settlement. Because the proponents of this model see environmental change as a comparatively late development, they have tended to eschew desertification and environmental degradation as primary influences on Murghab settlement development, preferring to focus on social, political and economic factorsxxvii.

While much of this research has been regional in scope, a number of recent targeted studies provide a higher resolution view of Murghab settlement. Intensive survey conducted by Cleuziouxxviii highlights continuous and varied distributions of surface pottery at the intersite scale. More recently, intensive survey in the Auchin region has compared distributional patterns in regions of variable land-usexxix, and several recent studies have re-assessed local aspects of sedentary-nomadic relationships, exploring issues such as settlement location, socio-economic co-development and knowledge transferxxx. While much of the current small-site focus has concerned pastoral sites, a few sedentary occupations have come under renewed focus in the region south of Gonur Depe, and small-scale surveys around Gonur have sought a better understanding of the surrounding landscape and its socio-economic function with respect to the larger sites of Gonur north and southxxxi. Additionally, work in the Adji Kui region has contributed new perspectives to site dimensions and delineationsxxxii. Such investigations have been complemented by spatial-analysis approaches including Thiessen polygonsxxxiii, geostatisticsand anisotropic (directional) analysisxxxiv, all of which have contributed to a better understanding of regional, subregional and local settlement patterns in the Murghab alluvial fan.

To effectively build upon these recent theoretical and methodological developments in the context of local-scale settlement dynamics, it was necessary to identify a study area that offers avenues of continuity as well as points of departure. The Egri Bogaz region, a comparatively remote archaeological landscape in the northern portion of the delta, offers both. Situated near the margin of the alluvial fan where the desert/delta boundary is more immediate, this comparatively isolated region between the better-known settlement groups of Auchin and Kelleli contains few known sites. Four of these, initially identified by the Margiana Archaeological Expedition and designated Egri Bogaz 1-4, have been the target of limited excavation, but this work remains largely unpublished. More recently, surveys have identified additional Bronze Age, Parthian/Sasanian and Islamic sites among what researchers view as a continuous distribution of pottery across the delta landscapexxxv.

Ultimately, this research investigates an archaeological landscape; while the nominal focal pointis the site designated Egri Bogaz 4, this is not a site-specific study but an investigation into de-positional characteristics of a marginal alluvial landscape, for which a known site serves as auseful reference point. Furthermore, it should be stressed that while chronology of surface ma-terial is taken into account as is crucial in surface surveys, and was rigorously examined duringthe course of research, this is very much a study of marginal occupation vis-à-vis spatial relation-ships of surface material and depositional characteristics: a fine and reliable ceramic chronologyis not yet available in the northern part of the Murghab delta and remains secondary to the mainscope of this research. Because the region is poorly understood yet archaeologically rich, EgriBogaz provides an excellent research context in which to explore these spatial relationships en

Markofsky Page 8

route to a better understanding of variability in landscape and occupation, and ultimately toprovide insight into the peopling of an alluvial margin, with few a priori assumptions.

III. Methodology The Northern Murghab Delta Survey (NMDS), conducted from 2007 to 2009 with an additional data-collection season in 2013, employed an intensive survey strategy in which field walkers, spaced 20m apart, evenly surveyed an L-shaped region of 11 sq. km, reporting at 20m intervals the total number of sherds observed on the ground along their respective transects (Figure 5). Each 20m x 20m collection unit could then be assigned information including sherd total, land-cover type, or other geomorphological, hydrological or anthropogenic features. The resulting grid of 27,000 units provided a spatial continuum amenable to rigorous spatial and statistical analysis that could be applied over multiple analytical scales. Diagnostic material (e.g. rims, bases, handles, decorate fragments) was collected to provide broad chronological information (e.g. Bronze Age vs. Sasanian/Islamic) and, when possible, to identify chronological sub-categories of Bronze Age material (i.e. Middle Bronze, Late Bronze and steppe coarsewares). Because surface artefacts are subject to myriad post-depositional processes, a series of test pits was conducted in order to assess the degree to which the observed surface distributions could be deemed representative of actual subsurface archaeology.

Markofsky Page 9

Figure 5: Study area, showing a.) The survey area with analytical units and b.) the full surface pottery distri-bution

Markofsky Page 10

While the gridded system provided a structured methodology to the field survey, it was analytically restrictive due to its high degree of abstraction. Reducing a continuous distribution of surface material to a series of squares and grids can result in false linearities and patterning that bear little relation to archaeological reality. To mitigate this effect, the survey area was divided into a series of 9 analytical units that could be used as interpretative guides (Table 1). These analytical units, referred to hereafter as Areas 1-9 and 1E, were determined based on visual assessments of surface distributions and land cover, although it must be stressed that, as with the grids described above, such delineations are abstract and cannot be expected to represent actual boundaries in antiquity. However, they can be very useful for conceptual purposes. This goal was facilitated by the inclusion of raster data from high resolution Quickbirdand multispectral ASTER satellite imagery in order to assess the characteristics of both the surface scatters as well as the underlying landscape. In 2013, this imagery was supplemented with extremely high resolution visual and topographic data acquired via an unmanned aerial vehicle (UAV), graciously provided by the Ancient Merv Projectxxxvi and operated by Gaygysyz Joraev. These datasets were incorporated into a GIS system for spatial and statistical analysis.

IV.a. Results of the NMDS Survey

The following discussion will briefly outline some of the overarching characteristics of the NMDS survey area. The eastern portion of the region (Areas 1 and 1E) will then be examined, before discussion shifts westward via Areas 2 and 3 towards the western part of the survey area (Areas 4-8). In general terms, the study region contains locally varied geomorphology, a microcosm of the northern delta where sand dunes comprise about half the land cover, and takyrsanother third. Surface pottery was most prevalent on the perimeter of small takyrs and in over-sanded depressions. Indications of fluvial activity, usually marked by meandering trajectories visible both on the ground and in satellite imagery, occurred throughout the survey area. However, erosion and over-sanding often obscured watercourses (see Cattani 2008b: 127; Sarianidi 1990:55), so an accurate assessment of the local hydrological system was not always attainable.

Within this dynamic landscape, the site designated Egri Bogaz 4 is identified by a rather nebulous archaeological signature in the north-eastern portion of the survey area. According to a2008 ‘Site Passport’ document compiled by the Turkmen Ministry of Culture, the site comprises a 1-1.5m high mound measuring approximately 0.1ha in area. These dimensions are most likely based on Udeumuradov’s unpublished estimate of 1-2 ha; the discrepancy is probably due to his inclusion of surface scatters beyond this mound (Figure 6). In contrast, estimates provided by researchers associated with the AMMD project propose a much larger area of some 13 ha, described in their GIS database as a high mound surrounded by a central flat area, cut through bya pipeline.

Markofsky Page 11

Figure 6: Egri Bogaz 4 a) site 'passport' document and b) main mound immediately south of the road.

Topographic data from the 2013 UAV survey revealed significant variability over small areas, and clearly revealed where the pipeline and an associated road, both built in the 1970s, altered the topography of the site (Figure 7). Surface pottery was widely distributed, sometimes occurring on elevated areas adjacent to the road (including the mound specified above), althoughthe scatters were much denser in low lying areas, likely the result of both erosional (e.g. slopewash) and deflationary processes, the latter of which may in some cases leave ‘lag deposits’in which large sherds remain after the removal of small-grained deposits by the wind. A few of the mounds have been identified as tepesxxxvii, although these merge with dune ridges, making theidentification of specific anthropogenic mounds difficult. Furthermore, some of the pottery on these mounds was likely deposited recently as a result of construction activities, a possibility supported by the mixing of surface pottery with loose aeolian sands.

Markofsky Page 12

Figure 7: Digital Elevation Models of Egri Bogaz. Square 'features' are actually small dune hills, although upcast from the construction of the road and pipeline has contributed to recent anthropogenic modification.

Evidence of architecture was absent, although Udeumuradov has identified kiln remains here as with other Egri Bogaz sitesxxxviii; and concentrations of sometimes-vitrified brick debris found

Markofsky Page 13

during the NMDS survey support this assessment. Pottery scatters were almost exclusively of ‘Namazga’ type, a broad nomenclature derived from the Namazga site in southern Turkmenistan and used, sometimes too broadly, to characterize the pottery of the sedentary Bronze Age societies in the region. Both Middle and Late Bronze pottery was widespread (see below), as were occasional small pieces of copper and bronze. Two additional transects north of the survey area indicated that small aggregations of pottery and metal continued for at least 2 km. South of the central ‘mound’ of Egri Bogaz were dense scatters of brick, some of it vitrified, along with slag fragments of unidentified composition.

Other sites have been documented within Area 1, although confirmation of these as discrete habitation areas was not possible due to the complex geomorphology of the landscape and the often significant discrepancy between areas identified as 'tepes' (anthropogenic mounds) and actual concentrations of material (Figure 8). Chronological trends were not readily apparent in Area 1, although there were subtle indications that Middle Bronze material was slightly more prevalent toward the north of the survey area. For example, lug handles from hole-mouth jars and fragments of cups with straight, undifferentiated vertical rims similar to those found in Kelleli and the early phases of Gonur North tended to occur in the northern but not the southern part of the survey area. While circumstantial, this does support Salvatori's observations of ‘settlements and farms of the Middle Bronze Age’ north-east of the Egri Bogaz sitesxxxix.

Markofsky Page 14

Figure 8: Pottery distribution in Area 1 showing the relationship of surface material to previously identified sites

One striking aspect of the surface distribution in Area 1 was that pottery scatters to the east and to the west of the dense surface distribution were entirely different (Figure 9). To the west, a loose line of sherd scatters, some containing brick debris and copper or bronze fragments, extended broadly from south-west to north-east. Each of these pottery scatters comprised about 1 ha, and the general appearance is a partial arc of 'satellite' clusters that flank the western side ofthe primary area of occupation. No corresponding distributional pattern was present to the east,

Markofsky Page 15

however: here, surface material was sparse and consisted of clusters of small, heavily abraded sherds with few if any diagnostics. This sharp contrast with the dense scatters to the west suggests that occupation to the east may have been far more limited in scope. Topographic data from the UAV survey indicates a gradual downward slope westward towards a possible palaeochannel or canal, which could indicate preferential occupation of slightly higher ground in close proximity to watercourses. If accurate, this trend may have been spatially restricted: the geomorphological and hydrological character of the landscape to the east is similar in appearancealthough slightly higher in elevation, although largely devoid of surface material. While speculative, it is possible to suggest that the primary artery of occupation was a watercourse a few hundred meters to the west of the densest surface scatters, while some of the smaller channels to the east may have been exploited for small-scale irrigation agriculture rather than habitation.

Figure 9: Areas 1 and 1E. Note the tendency for moderate scatters to aggregate to the west of the main sur-face concentrations

Markofsky Page 16

Extending about 3km westward from these large and complex distributions in Area 1 was a region of comparatively sparse surface scatters. This region, comprising Areas 2 and 3, was situated to the north of a cultivated zone with cotton and wheat fields just barely extending into the survey area. Modern cultivation may be responsible for the texture of the exposed takyr surface which was soft and pliable; the dry, cracked unvegetated surfaces that occurred elsewhere in the survey area were not particularly common. The higher moisture content of the takyr surface here may be partially attributable to the raised water table resulting from these localagricultural activities, and additional evidence for a shallow water table may be found in a test pit conducted on the takyr surface just to the east (see below), where salt was detected about 50cm below the takyr surface.

Bronze Age material was sparse in the region and diagnostic material uncommon. A single scatter of incised coarseware, associated with the pastoral 'Andronovo' groups, was found in Area 2. Such pottery, while better attested further to the south where it has been found in close proximity to sedentary sites, has also been documented to the north-west in the Kelleli region, and recent excavations near Auchin in the north-eastern portion of the delta, as well as at sites 1211 and 1219 in the Takhirbai region, offer evidence that the interactions between sedentary and nomadic groups towards the end of the Bronze Age were particularly complex, although such interactions in the northern margin of the delta appear to have been less common than in other regions where Andronovo-type material is more prevalentxl.

Also occurring in Area 2, in sparse scatters around usually small takyr perimeters, were coarse, blackened sherds. Although it was not possible to accurately identify specific settlement locations based on these few scatters, it is reasonable to assume that, given the tendency of these materials to occur together at significant distances from the area of central occupation, there is anactual cultural element to the distribution of these surface scatters that is not sufficiently explained by post-depositional processes alone. One possibility may be that at least some of these scatters represent small-scale areas of domestic activity or, given the lack of architectural material, possibly seasonal or periodic occupations. Such an idea has been put forth to explain small scatters not associated with anthropogenic moundsxli. However, the direct association of a few sherds with a seasonal presence must be treated with suspicion given the complexity of the post-depositional processes in the northern Murghab. What does seem to be clear, however, is that the nature of occupation in these more remote regions has a significantly different character from the high-density regions not only in terms of the amount of surface material, but also in quality and type.

One feature worth noting in this northern portion of the NMDS survey area was a densely vegetated depression measuring approximately 80m in diameter, and situated immediately to the south of a line of three wells. Fleskins has described an innovative system of water collection called a chirle in which a series of wells is used to extract fresh groundwater trapped between thesandy subsurface and salinated water below the lensxlii. Usually, these reach a diameter of 10-12m, and it is perhaps more likely that the depression is, at least in part, natural. Although Bronze Age pottery was sparse here, the presence of Sasanian and Islamic sherds, the line of wells, and the present-day focus of the region as a remote yet developing agricultural zone suggest a history of significant water collection, and it is possible that the region once functionedas a primary source of water for caravans moving across the northern delta .

Markofsky Page 17

Areas 4-8, comprising the western portion of the survey area, was characterised by a rolling topography of stable or semi-stable dunes that gave way to a broad region of once-active palaeohydrography to the east (Figure 10a). While specific palaeochannels could not be confirmed, evidence from the remote sensing data indicates the presence of a watercourse east ofa large (0.75 km2) takyr. Two Bronze Age sites have been previously documented in the vicinity,as a result of transects conducted by researchers associated with the AMMD projectsxliii. This region of the survey area contained a series of moderate sherd clusters, extending from north to south and separated by about 300m. These contained both Middle and Late Bronze materials and specific scatters could not be individually dated: the mixed materials are palimpsests that likely developed as a result of site deflation, rendering their archaeological realities more complex than the ‘single period occupations’ that are sometimes attested for small sites in the regionxliv. The largest of these scatters sometimes contained brick debris as well, and notable surface finds included a violin-shaped figurine generally associated with the Middle Bronze Age and a torso that resembled Namazga III type figurines found in the Geoksyur oasis. This latter find, however, should not be taken to suggest earlier occupation in the region as it is a chance find without provenance; however it does support the circumstantial evidence of other pre-Middle Bronze materials attested in both the Kelleli and Adji Kui regionxlv. In addition to these Bronze Age scatters, Sasanian and early Islamic material was common throughout the NMDS survey area, noteworthy in light of the discovery of a ‘tomb’ containing Late Sasanian/Early Islamic material near Adji Kui 1xlvi. Such finds may indicate trajectories of movements northwardfrom Merv, and while no architectural evidence was found in the immediate survey area, a remote caravanserai exists at the edge of the palaeodelta 30km to the north-east (visited by the author and P. Wordsworth in 2013). Furthermore, several large bases with interior glaze were located during the survey that may be fragments of water jars. This suggests that at the very least, the Egri Bogaz region witnessed consistent trajectories of movement from at least the Sasanian/Islamic period, along routes that may have existed much earlier.

One further notable feature was found in Area 7, just north of a moderate scatter of Bronze Age pottery. Situated just east of a distinctly linear sand ridge that may possibly be the result of the infill of a canal, the site contained two low mounds, each less than 20m in diameter (Figure 10b, 10c). These were covered with dense scatters of brick, much of it vitrified, and slag. The absence of surface pottery or obvious wasters suggests that rather than a pottery kiln, the feature may reflect an industrial-grade furnace, perhaps a brick kiln. In this light, the proximity of the feature to a linear channel would be interesting, as the use of bricks has been documented in the construction of sluices for the automatic regulation of water (e.g. inscriptions at the Mesopotamian site of Diqdiqahxlvii), and texts at Lagash have documented the use of fired bricks in the construction of small dams to raise the water level in canals.xlviii The suggestion of a brick kiln rather than a metal production facility, such as a copper smelting furnace, is further supported by a preliminary in-field XRF analysis of four pieces of slag, which revealed substantial percentages of iron silicates but only trace percentages of copper or nickel. However,the site remains undated and cannot yet be definitively associated with Bronze Age productive capacities.

Markofsky Page 18

Figure 10: Western survey area. a.) Diagnostics in Areas 4-8, with prominent scatters identified; b.) the 'brickkiln' and c.) the location of the brick kiln next to the linear feature.

Markofsky Page 19

IV.b. Results: Surface Material

The diagnostic material collected during the survey comprised a total of 707 diagnostic sherds ofwhich 509 could be dated to the Bronze Age (Figure 11). Of this material, evidence pre-dating the Middle Bronze Age was entirely absent, with the possible exception of the figurine base described above. The absence of early material in the survey area suggests that any such occupation in the northern delta is likely to have been limited and perhaps constrained to central or southern portions of the delta now obscured by alluvium.xlix Deeply buried sherds from the Middle Bronze Age, recovered by M. Cremaschi (1998) in the heavily cultivated southern portion of the alluvial fan, support this possibility, as do sherds recovered in-situ from a buried pipe section south of Merv.l Additionally, Chalcolithic material has been attested at both Gonur and Adji Kui 1 and 9li.

Figure 11: Surface pottery types in Areas 1, 4 and 5

Markofsky Page 20

Subsequent Bronze Age periods, by contrast, are well represented in the diagnostic material—although differentiation was difficult because of the continuity of forms and the tendency for Middle and Late Bronze materials to occur together. Middle Bronze pottery included hole-mouth pots, bases with slightly inward-sloping walls similar to those found in stratigraphic unit 3on Adji Kui 9lii, and plain-rim vessels well attested in early phases of Gonur Northliii, as well as two violin-shaped terracotta figurines of a form well-established at sites such as Adji Kui 9 in theMurghab and Altyn Depeliv. Direct diagnostic evidence for Late Bronze occupation was more tenuous due to the continuity of forms, although the prevalence of dark red and grey fabrics, occasional trumpet-shaped bases and tall bottles with everted rims support the findings of previous surveyslv that the Late Bronze is also well-represented throughout the Egri Bogaz region.

Materials clearly datable to the final phase of the Bronze Age were difficult to identify, and Cattani has noted that production during this period “seems to be a continuation of the previous one and does not represent a real change in the cultural system”lvi. Nevertheless, this phase does not appear to be well-represented, as diagnostic markers such as concave walls, rounded, inflexed or everted rims and incised shoulder ornamentation were almost entirely absent in the survey arealvii. Steppe or ‘Andronovo’ type materials were extremely rare in the survey area, although occasional steppe pottery has been documented in the northern deltalviii. Furthermore, recent research at the site of Ojakly indicates a substantial degree of complexity both in the character of such settlements as well as their interaction with sedentary groupslix. While the occurrence of these coarsewares supports some degree of interaction between the steppe and the sedentary societies in the northern Murghab, it seems likely that the extent was quite limited.

IV.c. Results--Subsurface Analysis

The above analysis illustrates the highly variable character of the NMDS surface distribution andsuggests that surface pottery in the NMDS region, while partially reflective of settlement-deriveddeposition, has been strongly modified by post-depositional processes that may include natural processes such as fluvial transport, hill-slope erosion and deflation, as well as anthropogenic (or human-influenced) activities such as the movement of herds through takyrs or other low-lying areas. Because of these processes, the association between actual habitation processes and surface pottery may be highly tenuous, and calls into question the degree to which we can reliably extrapolate original settlement patterns from highly modified surface distributions. If the goal is to elicit marginal processes at the local level, this disconnect is particularly problematic. To address this issue, several lines of 1m x 1m test pits were dug in areas of high pottery density to establish subsurface ‘anchors’ by which to interpret the surface scatters. Thesewere dug to a depth of 1m, sufficient to determine the depth of cultural material. The pits were spaced 40m apart, and aligned east to west. This orientation is broadly perpendicular to both dune ridges and the general palaeohydrography of the delta, which trend N-S and NW-SSE, respectively.

The first line of test pits, in Area 4, traversed a large takyr flanked by a large dune on its eastern side. Beyond this dune, an interdune valley indicated a possible palaeochannel (Figure 12). Test pits dug into the takyr subsurface revealed natural alluvial stratigraphy, devoid of cultural material and characterised by a layer of friable clay, underlain by fine, laminar sands to a depth of approximately 60 cm, beneath which lay a second, more compact sandy-clay layer. This secondary clay layer may be compared with Lyapin’s findings in the Kelleli 1 area of a dense clay layer immediately below alluvial sands. He links this clay stratum with the

Markofsky Page 21

Eneolithic/Chalcolithic (Namazga II or III) alluvial surfaceslx, which suggests that Bronze Age alluvial deposits in the distal portion of the alluvial fan are not deeply buried.

Figure 12: Test Pits in Area 4

The takyr surface is generally flat although there was a slight downward slope to the east (2 cm decrease in elevation over 40m). While the downward slope of the surface sediments was minimal, there was a significant grade in the subsurface deposits. Here, the elevation of the sandy layer decreased 20cm over the same 40m distance between test pit AU4-0 and AU4-1W. This downward slope, which continued towards the edge of the dune, suggested that the takyr may have been part of a channel floodplain associated with a watercourse a short distance towardthe east. Similar topography has been observed in a natural channel near Takhirbai 3lxi.

In order to investigate this possibility, another pit was dug in an interdune 'valley’ on the east sideof the dune ridge (Figure 13). Beneath a thick layer of windblown surface sands, clayey sediments were exposed at a depth 40cm lower than in the pit 80m to the westlxii, which suggested a steepening of the downward grade of the alluvial surface.lxiii This suggests that the topographically low area reflected a pre-existing ditch or channel and was not formed entirely from more recent aeolian depositional phases. Furthermore, the clays within the takyr layer here

Markofsky Page 22

were poorly sorted and coarse, possibly indicating more active fluvial activity. In the western wall of the pit, at the transition between the surface deposit of Aeolian sand and the subsurface takyr-like soils, was a thick layer of ash that became more concentrated towards the bottom of the deposit. A few very small sherds and brick debris, less than 1cm in diameter, were found along with bits of plaster, although no charcoal was recovered. The buried and abraded sherd fragments east of the dune ridge stand in sharp contrast to the large and often diagnostic pottery on the takyr surface to the west, where the test pits revealed natural stratigraphy, and it is likely that the core of intensive occupation lies beneath the dune ridge. The increasing thickness of the ash deposit to the west, as well as the west-to-east downward grade mentioned above, suggests that actual settlement may have occurred on the alluvium or a natural rise quite close to the watercourse, now overlain by sands. While speculative, this may recall the type of occupational choices made further east in Area 1, in which the natural grade of the landscape in relation to occupation was similar in character.

Figure 13: Schematic diagram of Area 4 subsurface stratigraphy

Two additional lines of test pits were dug in Area 1, both north and south of the road (Figure 14).These were spaced at 20m rather than 40m to provide a higher-resolution understanding of archaeological processes in a region of apparently substantial occupation. The stratigraphy was

Markofsky Page 23

broadly comparable to that in Area 4, although a secondary layer of alluvial sands was reached inone pit north of the road at a depth of 60cm, beneath a compact clay layer. Additionally, the takyr subsurface in the northern line of test pits contained white particles that appeared to be salt or gypsum, in conjunction with small fragments (<1 cm) of brick and pottery. Soil in the region is highly salinated, a consequence of high evaporation rates in the desert heat as well as heavy and often poorly managed cultivation. Although high soil salinity has also been cited as a cause of agricultural decline in antiquitylxiv, more research is needed to evaluate the role that it might have had in influencing the Bronze Age agricultural regime.

Figure 14: Area 1 Test Pits

While substantial evidence of occupation was not found in the test pits, these small excavations offer both clues and restrictions on how to interpret the survey data in the northern delta. First, the estimation of specific site dimensions and locations based on material scatter is extremely unreliable, as deflationary processes, localised erosion and fluvial processeslxv have substantially transformed the surface deposition. The absence of large subsurface sherds or substantial cultural deposits, even in areas with dense surface scatters, suggests that while deflation has clearly played a role in creating the present-day surface distribution, redeposition of material,

Markofsky Page 24

primarily by water, is a substantial and ongoing process. Directional analysis of the survey data, not discussed in detail here, has revealed a prevailing NNW-SSE orientation to the material distributionlxvi. This phenomenon suggests that material tends to propagate in the direction of water flow, a process that would have been most pronounced in antiquity when the channels were still active; but that may still be influenced by heavy precipitation events in winter and spring.

V. Discussion

Perhaps the best way to situate the above data within the framework of marginality is to revisit the geomorphological, hydrological and environmental tensions between the idea of an oasis civilisation and that of a continuously settled alluvial plain. As noted above, one of the key reasons for the rejection of the oasis model by some recent researchers has been the apparent continuity of surface material, detected in multi-kilometre offsite transects across central and portions of the northern palaeodeltalxvii. The significance of this continuum is seen as both environmental and social: if offsite surface material is present throughout the alluvial fan, it suggests to the researchers that widespread occupation pre-dates desertification, and that settlement—or at least cultivation—was effectively uninterrupted throughout the entire landscape. Another element of this interpretation is that offsite material or ‘background scatter’ is broadly attributed to manuring, a concept derived from Wilkinson’s interpretations of such scatterslxviii. However, rigorous testing of such models that may be applicable in the Mediterranean or Near East have not been applied here and caution should be used when transferring survey-based interpretations from one archaeological environment to a completely different context lxix.

While the findings of the NMDS survey verify this picture of distributional continuity to an extent, they also illustrate that such continuity is by no means uniform and is, in fact, highly variable. To explore this variability, the following discussion considers first the material most likely to be associated with significant sedentary occupation, and then examines the sparse pottery scatters in the intermediate or ‘offsite’ regions. At the broadest level allowed by the spatial restrictions of the survey, markedly different settlement trajectories could be seen in the eastern and the western portion of the survey area. In Area 1, the surface material indicates a primary locus of occupation, flanked by a quasi-linear arc of moderately dense habitation or activity areas to the west. While both dunes and deflation prevent an accurate assessment of the size of this region of moderately dense material, sizeable scatters may be found over a region of some 60 ha, with small scatters of pottery, brick and copper/bronze fragments continuing at least 2 km further to the north, although this does not mean that the entire area was concurrently occupied, and we may very loosely estimate a region of around 5-8 ha of moderate to dense occupation based solely on surface scatters and the occasional subsurface material indicated by the test pits. Indeed, the conceptualization of this landscape of artefacts as a single archaeological entity--the proverbial ‘site’—overlooks the scope and complexity of occupation, factors that must be considered in order to understand settlement on the alluvial margin.

One way to consider the Area 1 surface material is as a series of aggregated and perhaps shifting settlement processes. Recently, Cattani has re-envisioned the archaeological landscape around sites such as Togolok 1 and Taip as large, perhaps multi-centric settlement complexes that could indicate large central placeslxx. In some cases, these could also be seen as contiguous chains of occupationlxxi. Recent work at the adjacent sites of Adji Kui 1 and 9, also a half-kilometre apart but with continuous surface material between them, offers another example of this kind of

Markofsky Page 25

aggregated occupation, although the chronological relationship of these two sites has not been firmly establishedlxxii.

Settlement aggregations such as these are not restricted to the Murghab. For example, at the site of Titrish Hoyuk in Anatolia, the main site is surrounded by an outer town as well as outlying ‘suburbs’. Algaze has suggested that sites with a shorter duration of occupation may spread out substantiallylxxiii, and this phenomenon may be in evidence in the northern Murghab where Bronze Age occupation most likely spanned no more than a few centuries. However, the fluvial hydrology is likely to have played a more deterministic role than on the dry-farming margin of Titrish Hoyuk, where rainfall could approach 300mm. For this reason, we might also turn to a regional parallel, that of Shahdad, where a pattern of horizontal rather than vertical stratigraphy has been attributed to occupational shifts as a means of coping with a changing and unstable fluvial systemlxxiv. The conjunction of opportunistic, short-duration settlement with a dynamic hydrological regime, in which local channel shifts and sand encroachment may have represented observable local manifestations of broader regional changes, may provide some clues to the organic and unsystematic nature of settlement in the northern Murghab; and such processes may also provide a rationale for the discrepancy between the low mounded sites throughout the Murghab, and the large depes only slightly further west in the wetter and more fertile Kopet Dag foothills.

In this frame of reference, occupation in Area 1 may be better represented as a settlement process—manifested as a complex, quasi-linear chain of occupation. But what might be the characteristics of this form of settlement? In this visually obstructed landscape, the actual scope of occupation is difficult to determine, but a general model may be proposed. The intensity and diversity of material in the centre of Area 1 indicates a primary settlement area, situated on a slight rise, comprising several hectares of uneven occupation. Secondary concentrations or activity areas are oriented approximately south to north, to the west of the main occupation area and possibly slightly lower in altitude, apparently loosely aligned along a watercourse, although whether natural, human-modified or constructed is not yet clear. This pattern is not restricted to the north, as can be seen in a comparison between statistically significant clusters of material in the NMDS survey area and the site of Togolok 1, where a quasi linear group of sites extends along a palaeochannel east of the site, while only two sites are found to the west (Figure 15).

Markofsky Page 26

Figure 15: Settlement aggregations. a.) Statistically derived clusters in Area 1 and b.) Togolok 1 region

A few kilometres to the west, the settlement pattern in Areas 4-8 bears some of the hallmarks of the denser Area 1 in terms of the apparent linearity and diagnostic material found, but is very different in both scale and character. Firstly, no apparent central area of occupation is evident: sherd scatters of similar dimension tend to align from north to south in aggregations of up to 2 hectares, a rough size estimate that has likely been inflated via post-depositional processes. Secondly, the heavy concentrations of vitrified brick, slag and occasional metal fragments recorded in Area 1 were largely absent here. Although some brick debris was found in Areas 4 and 5, clear indicators of production were absent with the possible exception of the ‘brick kiln’ mentioned above, although this was nearly a kilometre to the south and unlikely to be directly associated with the individual small settlements here. While speculative, it is possible that the feature may indicate larger-scale and more centralised production associated with the burgeoningarea of settlement in the central delta during the Late Bronze Age (see below), although the feature cannot yet be firmly dated to the period. Thirdly, the evidence from the test pits, namely the small fragments of cultural material and the topography of the partially buried takyr surface, suggests that unlike in Area 1, no large central area of occupation was present, and small settlements likely existed in very close proximity to a watercourse. While conclusions regarding water and settlement are tentative, the preceding analysis suggests that at least some of the moderate scatters along this corridor pertain to small sedentary occupations and cannot be written off solely as the result of post-depositional processes. If this is the case, a possible

Markofsky Page 27

relationship may be drawn between site size and proximity to channels. Small occupations both in Area 1 and in the western part of the survey area may have been located closer to watercourses, with larger settlements established slightly further away. Such a pattern may reflect the reduced capacity of smaller communities to undertake agricultural projects on the scale of their larger counterparts.

So far, the above discussion has focused on two clearly identifiable occupation areas, each of which appears to be associated with a distinct hydrological regime. We now move to the pro-cesses, occupational, post-depositional or otherwise, that have contributed to the ‘continuum’ of dispersed material throughout the survey area. As noted above, these have been attributed by some Murghab researchers to manuringlxxv, whereby the distribution of manure on agricultural soils resulted in the widespread dispersal of surface sherds contained in the refuse, which were then subject to further fragmentation and dispersal over the millennia. It is certainly possible thatmanuring contributed to some surface scatters in the northern Murghab. However, observed sherd totals were far lower than in comparable investigations in the Near East and the Aegeanlxxvi,and the variability of the background scatter suggests punctuated rather than continuous pro-cesses. Manuring, if practised, is likely to have been localised rather than widespread, perhaps contributing to the directional continuity of material along relict channel systems.

Statistical, geostatistical and directional analyses, not discussed here have been applied both to onsite and offsite sherd distributions in the NMDS survey arealxxvii. These analyses indicated statistically significant density peaks, sometimes 100-200m in diameter, that were not always observable in the field because of local dune obstruction. The sparse yet punctuated ceramic scatters, which did not exhibit the same quasi-linear orientation as the ‘satellite’ scatters in the western part of Area 1, or the north-south material distribution in Areas 5-7 (see above), likely reflect an amalgam of different processes. In certain cases, they may be the remains of transient occupation, or of activities likely to occur in the immediate vicinity of such settlements such as production, burial or activities associated with local herding. Livestock breeding was practised in the sedentary sites of the Murghab deltalxxviii, and unused agricultural land may therefore have seen a dual function as local pasturelxxix. If this is the case, small scatters may in some cases represent vestiges of encampments or transient movements by local herders rather than permanent or seasonal occupation. Even today, the remains of fires built by local herders can be seen in close proximity to agricultural fields, and it is possible that similar practices existed in antiquity. In still other cases, such scatters may indicate small-scale sedentary occupation where people were able to exploit locally productive agricultural land, a possibility that will be explored in more detail below.

The survey and test pit data suggest that while significant occupation and cultivation propagated along the channel systems, interfluvial zones may have been sufficiently watered to support localsubsistence of a very different character than the more densely occupied regions. The consistent but highly uneven presence of archaeological material may partially reflect processes of local mobility associated with the pastoral activities of a mixed economy as well as small-scale but opportunistic occupation facilitated by intermittent and shifting fertile zones, vestiges of small settlements whose surface record has long since been transformed via post-depositional processes. To understand these processes in the context of the geographical context of the northern Murghab, we can look to research conducted by Cazancli, who has modelled the behaviour of floodwaters in alluvial fans, finding that lateral erosion and channel shifts upstream may result in the constant reshaping of the distal regionslxxx. Related processes may be observed

Markofsky Page 28

in the present-day Murghab delta, where the boundary between active channels and dunes is often characterized by extensive pooling, and small ephemeral lakes. Beyond these regions of active channelization, the prevalence of these pools declines although one can still identify their former existence from the many small takyrs, some of which may still contain standing water. The existence of these pools, sometimes many kilometres away from significant channels, offers clues to the availability of widespread small pockets of fertile land, even in a potentially adverse environment of encroaching sands as well as complex and locally capricious channel systems. Inthis light we may consider Sherratt's suggestion that sheet flow associated with shallow canals may have been sufficient to facilitate non-canalised local irrigationlxxxi. It is likely that such micro-environments, often far removed from large channels, facilitated outlying subsistence during the Bronze Age.

The sporadic and opportunistic local subsistence practices outlined above may be seen in terms of niche-construction: in archaeological terms the process by which individuals or groups createdor modified their local environments, thereby influencing both environment and self-sustainabilitylxxxii. In Iran, for example, niche construction represents a developing framework for evaluating the social, political and economic implications of locally vs centrally managed water resourceslxxxiii. Modifications of the natural braided channels on the Tehran plain suggest tothe authors that niche construction mechanisms existed as early as the Neolithic, as distinct from later, imperial-sponsored large-scale irrigation projects (although large-scale need not imply a centrally administered project). Elsewhere in the region, similar indicators of niche construction—or at least serendipitous local exploitation—may occur at the site of Choga Mama in the KopetDag, where small transverse canals may have been constructed off of braided streamslxxxiv. A similar combination of practices may have existed in the Murghab, where creative self-sufficiency at the local level could have been supplemented with the acquisition of goods from more densely populated regions only a kilometre or two away. Even today, we can see the contrast between the widespread cotton and wheat fields of the heavily cultivated Merv oasis, and the scattered and often abandoned fields of wheat, melons and other crops towards the more remote fringes of the alluvial fan.

Because the research area is small, caution must be used when attempting to extrapolate these local-scale settlement dynamics over an entire alluvial fan. However, when viewed in a multi-scalar perspective, the results clearly suggest that prior conceptions of Egri Bogaz as a remote, isolated settlement group miss a larger and much more integrated picture, one that may be seen geographically as well as in the structure of occupation. From a geographical perspective, the 'isolation' of the Egri Bogaz region seems to be overstated in some earlier interpretations of the region, and we can examine the NMDS data in the context of the regional-scale AMMD work to support this assertion. Surveys conducted as part of the AMMD project indicated a line of Late Bronze occupation extending northward from the central delta (Figure 16), and it is plausible thatthe chain-like distribution of material, particularly in the western portion of the survey area (Areas 4-7) indicates a continuity of occupation from the central delta. Although beyond the scope of this paper, there is strong evidence to suggest that this region, comprising the Adji Kui and Adam Basan areas of occupation, underwent a burgeoning development during the Late Bronze Age, a possibility that is strengthened even further by the author’s identification via multispectral satellite imagery of a large and previously unknown depe just south of the Adji Kuiregion. The site was characterized by a 6 ha mound, approximately 3m high, covered with densescatters of primarily Late Bronze pottery, kiln remains, metal fragments and vitrified brick. There was also possible evidence of architecture based on UAV reconnaissance, although this

Markofsky Page 29

has not yet been confirmed on the ground (Figure 17). While purely speculative, the deepening focus of occupation in the Adji Kui region during the Late Bronze Age could reflect broad changes in response to an increasingly unreliable fluvial system in the alluvial margin. If this is the case, the ‘brick kiln’ discussed above could represent yet another manifestation of changing mechanisms of administration and hydrological management, although it remains undated and this possibility is purely speculative.

Figure 16: Survey area in regional context, showing AMMD site distributions in the central delta. Note the alignment of sites 965 through 1084 in relation to the western portion of the survey area

Markofsky Page 30

In terms of the specific settlement dynamics of the Egri Bogaz region, interpretation is somewhatlimited by the lack of available data on the development and decline of both the BMAC phenomenon and of Murghab occupation more generally. However, the high resolution of the NMDS survey data in the context of other projects in the region allows us to develop some informed speculation. The clustered, quasi-linear distributions of material, much of it dating from the Middle Bronze Age, suggest that the Egri Bogaz region, at least in the early phases of occupation, represents one locus in a much larger-scale, delta-wide process of settlement. New settlements likely appeared unevenly in fertile areas along river channels, at least initially, and the development of larger communities including the largest proto-urban centres was perhaps facilitated by greater access to cultivable land. The similarity of the distribution of the larger pottery aggregations in the NMDS study area to those found elsewhere in the delta (see the Togolok example above) suggest that such processes occurred on a fairly large scale towards the end of the Namazga V period, and the clear directionality in the orientation of these settlements, in accordance with the channels evident in the satellite imagery, demonstrates the influential role of the fluvial network in shaping the initial settlement patterns.

Figure 17: Site of 'AKF3', showing a.) regional context; b.) a DEM and c.) surface pottery

If the sherd scatters in the Egri Bogaz region reflect a delta-wide process of initial settlement, thedevelopmental trajectory may have been somewhat different, influenced in part by the transitional geography of the region, perhaps more so than at some of the largest and most

Markofsky Page 31

developed sites. Topographic data from SRTM imagery indicated that channels at the local and sub-regional level in the Egri Bogaz area are almost unidentifiable, as opposed to the easily-discernible palaeochannel networks throughout the central and southern delta and often associated with the larger site groups. The negligible slope of the landscape, and lack of evident deep cutting or levee-development of the palaeohydrology in the immediate region may have combined to create a more adverse and tenuous landscape for habitation or cultivation than may have been found in regions only a dozen or so kilometres away where the fluvial network may have been more reliable. The comparative lack of steppe material, as well, may also attest to the reduced potential for symbiotic development and transfer of goods or knowledge such as seen in Takhirbai or Auchin, perhaps limited by constrained development of the sedentary societies in this immediate region. The opportunistic exploitation of available, and sometimes non-channeled, areas of the landscape, may have been in part encouraged by limited central development, and while individual or community-level subsistence likely occurred throughout the delta, such phenomena were likely more pronounced in regions such as Egri Bogaz, in which the natural environment was perhaps less facilitative.

VI. Conclusion

This study, through a careful assessment of local and subregional distributions of surface pottery,has provided a new look at the variable processes that are likely to have been active on the desert-delta margin. Additionally, although the correlation between surface scatters and actual habitation is often tenuous, the multi-scalar interpretation of the characteristics of such scatters, in conjunction with subsurface analysis and an understanding of past and present geomorphological processes, is very useful in developing a picture of the fluidity and organic nature of occupation in a precarious alluvial margin.

As expected, the NMDS survey data confirm spatially the pivotal role of watercourses as influential factors in settlement establishment, as well as primary conduits for agricultural development and interaction between larger communities. However, a completely different storyhas emerged in regions more removed from these channels, where continuity is irregular and the anisotropic (linear) patterning much less evident. The story here appears to be one of serendipity: the variable landscape, while perhaps unproductive and even unusable in some areas, in others may have fostered opportunities for small-scale agriculture as well as local herd management and water acquisition, all practices of which are still evident in similar forms today.

In sum, the findings of this research suggest a framework for these kinds of delta-desert environments that falls somewhere between oasis and fertile alluvial plain. In this respect, transitionality and marginality effectively become active players in societal development, offering opportunities for economic exploitation but also demanding that such practices reflect a degree of creativity and adaptability in an already adverse, and perhaps deteriorating environment.

Markofsky Page 32

References Cited

Adams, R.M., 1965. Land Behind Baghdad: a History of Settlement on the Diyala Plains, Chicago: University of Chicago Press.

Algaze, G., Misir, A. & Wilkinson, T.J., 1992. Sanliurfa Museum/University of California excavations and surveys at Titris Höyük, 1991: A preliminary report. Anatolica, 18, pp.33–60.

Babaev, A.G., Fet, V. & Atamuradov, K.I., 1994. Landscapes of Turkmenistan. In Biogeography and Ecology of Turkmenistan. London: Kluwer Academic Publishers.

Balbo, A.L. et al., 2013. Contributions of geoarchaeology and remote sensing to the study of Holocene hunter–gatherer and agro-pastoral groups in arid margins: The case of North Gujarat (Northwest India). Quaternary International, 308-309, pp.53–65.

Barton, C.M., Ullah, I. & Bergin, S., 2010. Land use, water and Mediterranean landscapes: modelling long-term dynamics of complex socio-ecological systems. Philosophical transactions - Royal Society. Mathematical, Physical and engineering sciences, 368(1931), pp.5275–5297.

Bintliff, J.L. & Snodgrass, A.M., 1988. Off-site pottery distributions: A regional and interregional perspective. Current Anthropology, 29, pp.506–513.

Bondioli, L. & Tosi, M., 1998. Introduction. In A. Gubaev, G. Koshelenko, & M. Tosi, eds. The Archaeological Map of the Murghab Delta: Preliminary Reports 1990-1995. Rome: IsIAO, p. 97.

Brown, A.G., 1997. Alluvial Geoarchaeology: Floodplain Archaeology and Environmental Change D. Brothwell et al., eds., Cambridge: Cambridge University Press.

Casellato, U. et al., 2007. Physico-chemical and mineralogical study of ceramic findings from Mary City, Turkmenistan. Journal of Cultural Heritage, 8(4), pp.412–422.

Cattani, M. & Salvatori, S., 2008. Transects and other techniques for systematic sampling. In M. Cattani & S. Salvatori, eds. The Bronze Age and Early Iron Age in the Margiana Lowlands: Facts and Methodological Proposals for a Redefinition of the Research Strategies. Oxford: Archaeopress, pp. 1–29.

Cattani, M., Salvatori, S. & Tosi, M., 2008a. Excavations at sites no. 1211 and no. 1219 (Final Bronze Age). In S. Salvatori & M. Tosi, eds. The Bronze Age and Early Iron Age in the Margiana Lowlands: Facts and Methodological Proposals for a Redefinition of the Research Strategies. Oxford: Archaeopress, pp. 119–133.

Cattani, M., Salvatori, S. & Tosi, M., 2008b. The final phase of the Bronze Age and the “Andronovo Question” in Margiana. In S. Salvatori & M. Tosi, eds. The Bronze Age and Early Iron Age in the Margiana Lowlands: Facts and Methodological Proposals for a Redefinition of the Research Strategies. Oxford: Archaeopress, pp. 133–148.

Cazanacli, D., Paola, C. & Parker, G., 2002. Experimental steep, braided flow: application to flooding riskon fans. Journal of hydraulic engineering (New York, N. Y. Online), 128(3), pp.322–330.

Cerasetti, B., 2006. A 5000-Years history of settlement and irrigation in the Murghab Delta (Turkmenistan): an attempt of reconstruction of ancient deltaic system. BAR International Series, 1016, pp.21–28.

Cerasetti, B. & Mauri, M., 2002. The Murghab Delta palaeochannel reconstruction on the basis of remote sensing from space. Space Applications for Heritage Conservation.

Markofsky Page 33

Cerasetti, B., Codini, G.B., Rouse, L.M. (forthcoming) “Walking in the Murghab Alluvial Fan (Southern Turkmenistan): an Integrated Approach between Old and New Provides New interpretations about the Interaction between Settled and Nomadic People”, in B. Cerasetti, D. Frenez, C.C. Lamberg-Karlovsky, G. Marchesi (eds) My life is like the Summer Rose.Maurizio Tosi e l’archeologia come modo di vita. Festschrift in honor of Maurizio Tosi. British Archaeological Reports. International Series. Oxford.

Cerasetti, B., Salvatori, S. & Tosi, M., 2008. A GIS for the archaeology of the Murghab Delta. In S. Salvatori & M. Tosi, eds. The Bronze Age and Early Iron Age in the Margiana Lowlands: Facts and Methodological Proposals for a Redefinition of the Research Strategies. Oxford: Archaeopress, pp. 29–34.

Childe, V.G., 1934. New Light on the Most Ancient East. In London: Routledge.

Cleuziou, S., Gaibov, V. & Annaev, T., 1998. Off-site archaeological transects in N. Margiana. In A. Gubaev, G. Koshelenko, & M. Tosi, eds. The Archaeological Map of the Murghab Delta: Preliminary Reports 1990-1995. Rome: IsIAO, pp. 26–34.

Cremaschi, M., 1998. Palaeohydrography and Middle Holocene desertification in the northern fringe of the Murghab delta. In A. Gubaev, G. Koshelenko, & M. Tosi, eds. The Archaeological Map of the Murghab Delta: Preliminary Reports 1990-1995. Rome: IsIAO, pp. 15–25.

Cremaschi, M. & Zerboni, A., 2010. Human communities in a drying landscape. In I. P. Martini & W. Chesworth, eds. Landscapes and Societies. Dordrecht: Springer Netherlands, pp. 67–89.

Esearch, F.O.R.R. et al., 2013. Steven Markofsky 19340 Olney Mill Road. , pp.1–5.

Fleskins, L. et al., 2007. Desert water harvesting from takyr surfaces: assessing the potential of traditionaland experimental technologies in the Karakum. Land Degradation & Development, 18(1), pp.17–39.

Gerasimov, I.P. & Brice, W.C., 1978. Ancient rivers in the deserts of Soviet Central Asia. In The Environmental History of the Near and Middle East Since the Last Ice Age. New York: Academic Press, pp. 319–334.

Hakemi, A. & Sajjedi, M., 1988. Shahdad excavations in the context of the oasis civilisation. In G. Ligabue & S. Salvatori, eds. Bactria: An Ancient Oasis Civilisation From the Sands of Afghanistan. Rome: Centro di Studi Ligabue, pp. 141–152.

Harris, D., 2010. Origins of Agriculture in Western Central Asia, Philadelphia: University of Pennsylvania Press.

Hiebert, F.T., 1994. Origins of the Bronze Age Oasis Civilization in Central Asia, Cambridge, Massachusetts: Peabody Museum of Archaeology and Anthropology.

Jacobsen, T., 1960. The waters of Ur. Iraq, 22, pp.174–185.

Kendal, J., Tehrani, J.J. & Odling-Smee, J., 2011. Human niche construction in interdisciplinary focus. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, 366(1566),pp.785–92.

Kohl, P., 1984. Central Asia Palaeolithic Beginnings to the Iron Age: L’Asie Centrale Des Origines l'Age Du Fer,

Lamberg-Karlovsky, C.C., 2013. The Oxus Civilization. Cuadernos de Prehistoria y Arqueología Universidad Autónoma de Madrid, 39, pp.21–63.

Lees, S.H., Bates, D.G. & Bates, G., 2014. Society for American Archaeology The Origins of Specialized Nomadic Pastoralism : A Systemic Model. , 39(2), pp.187–193.

Markofsky Page 34

Luneau, E., 2013. Les mutations sociopolitiques de la civilisation de l’Oxus à la période finale (Asie centrale, âge du Bronze) : quels indicateurs de changements ? O. Brunet, C. Sauvin, & T. Al-Halabi, eds. Les marqueurs archéologiques du pouvoir, pp.221–241.

Lyapin, A.A., 1991. Paleogeographic features of the Murgab and Tedzhen deltas (eneolithic period, stone age). Problemy Osvoeniya Pustyn (Problems in Desert Development), 2, pp.63–70.

Mantellini, S., 2013. Settlement Dynamics, territory exploitation, and trade routes in teh ancient Samarkand Oasis (Uzbekistan). In Archeologia delle “Vie della Seta”: Percorsi, Immagini e CulturaMateriale II Ciclo di Conferenze.

Mantellini, S., Rondelli, B. & Stride, S., 2008. Analytical approach for representing the water landscape evolution in Samarkand Oasis (Uzbekistan). Computer Applications in Archaeology.

Marcolongo, B. & Mozzi, P., 1998. Outline of Recent Geological History of the Kopet-Dagh Mountains and the Southern Kara-Kum. In A. Gubaev, G. Koshelenko, & M. Tosi, eds. The Archaeological Map of the Murghab Delta: Preliminary Reports 1990-1995. Rome: IsIAO, pp. 1–14.

Markofsky, S., 2013. When Survey Goes East: Field Survey Methodologies and Analytical Frameworks in a Central Asian Context. Journal of Archaeological Method and Theory.

Markofsky, S. & Bevan, A., 2012. Directional analysis of surface artefact distributions: a case study from the Murghab Delta, Turkmenistan. Journal of Archaeological Science, 39(2), pp.428–439.

Masimov, I.S., 1979. Izuchenie pamatnikov epokhi bronzi nizovii Murgaba. Sovetskaya Arkheologiya, pp.111–131.

Masson, V.M., 1988. Altyn Depe.

Moore, K.M. et al., 1994. of Oxus. , 68, pp.418–427.

Orlovsky, L., Dourikov, M. & Babaev, A.G., 2004. Temporal dynamics and productivity of biogenic soil crusts in the central Karakum Desert, Turkmenistan. Journal of Arid Environments, 56(4), pp.579–601.

Pumpelly, R., 1908. Explorations in Turkestan, Expedition of 1904, Washington, DC: Carnegie Institutionof Washington.

Rosen, A.M., 2007. Civilizing Climate: Social Responses to Climate Change in the Ancient Near East, Lanham: AltaMira Press.

Rossi-Osmida, G., 2007. Citadella de la Statuette, Richere de la Etudi Ligabue.

Rossi-Osmida, G., 2011. Citadella de la Statuette Vol. 2, Richere de la Etudi Ligabue.

Rouse, L.M. & Cerasetti, B., 2014. Ojakly: A Late Bronze Age mobile pastoralist site in the Murghab Region, Turkmenistan. Journal of Field Archaeology, 39(1), pp.32–50.

Salvatori, S., 2002. A glimpse on copper and lead metalworking at Altyn-Depe (Turkmenistan) in the 3rd millennium BC. Ancient Civilizations from Scythia to Siberia, 8(1), p.69.

Salvatori, S., 2008a. A New Cylinder Seal from Ancient Margiana: Cultural Exchange and Syncretism in a World Wide Trade System" at the End of the 3rd Millennium BC. In S. Salvatori & M. Tosi, eds. The Bronze Age and Early Iron Age in the Margiana Lowlands: Facts and Methodological Proposals for a Redefinition of the Research Strategies. Oxford: Archaeopress.

Salvatori, S., 2007. About excavations at a bronze age site in Margiana (Turkmenistan). Rivista di Archaeologia, 31, pp.10–28.

Markofsky Page 35

Salvatori, S., 1998. Margiana Archaeological Map: The Bronze Age Settlement Pattern. In A. Gubaev, G. Koshelenko, & M. Tosi, eds. The Archaeological Map of the Murghab Delta: Preliminary Reports 1990-1995. Rome: IsIAO, pp. 57–66.

Salvatori, S., 2008b. The Margiana Settlement Pattern from the Middle Bronze Age to the Parthian-Sasanian: a Contribution to the Study of Complexity. In S. Salvatori & M. Tosi, eds. The Bronze Ageand Early Iron Age in the Margiana Lowlands: Facts and Methodological Proposals for a Redefinition of the Research Strategies. Oxford: Archaeopress, pp. 57–74.

Sarianidi, V., 1990. Drevnosti Strani Margush, Ashkhabad.

Sarianidi, V., 2002. Margus, Ashgabat: Turkmendowlethabarlary.

Sarianidi, V. & Dubova, N., 2011. Ongoing results of research at Gonur Depe 2008-2011 (in Russian). In Transactions of the Margiana Archaeological Expedition Vol 4. Gonur Depe Studies 2008-2011. Russian Academy of Sciences.

Sherratt, A., 1980. Water, soil and seasonality in early cereal cultivation. World Archaeology, 11(3), pp.313–330.

Spengler, J. 2014. Agriculturalists and pastoralists: Bronze Age economy of the Murghab alluvial fan, southern Central Asia. Vegetation History and Archaeobotany. DOI 10.1007/s00334-014-0448-0.

Staubwasser, M. et al., 2003. No Title. Geophysical Research Letters, 30(8), pp.7.1–7.4.

Stride, S., Rondelli, B. & Mantellini, S., 2009. Canals versus horses: political power in the oasis of Samarkand. World Archaeology, 41(1), pp.73–87.

Suslov, S.P. & Gershevsky, N.D., 1961. Physical Geograpy of Asiatic Russia, San Francisco and London: W.H. Freeman and Company.

Tamburrino, A., 2010. Water technology in ancient Mesopotamia. In W. Mays, ed. Ancient Water Technologies. New York: Springer, pp. 28–49.

Weiss, H., 1993. The genesis and collapse of third millennium north Mesopotamian civilization. Science, 261(20), pp.995–1004.

Wilkinson, T.J. et al., 2012. From human niche construction to imperial power: long-term trends in ancient Iranian water systems. Water History, 4, pp.155–176.

Wilkinson, T.J., 1993. Linear hollows in the Jazira: Upper Mesopotamia. Antiquity, 67, pp.548–562.

Wilkinson, T.J., 1982. The Definition of Ancient Manured Zones by Means of Extensive Sherd-Sampling Techniques. Journal of Field Archaeology, 9(3), pp.323–333.

Williams, T., 2012. Unmanned Aerial Vehicle Photography : Exploring the Medieval City of Merv , on theSilk Roads of Central Asia. , (15), pp.54–68.

Wittfogel, K.A., 1957. Oriental Despotism, New Haven: Yale University Press.

Markofsky Page 36

End Notes

Markofsky Page 37

i (Brown 1997; Barton et al. 2010; Cremaschi & Zerboni 2010, Spengler et al. 2014)ii (Pumpelly 1908; Childe 1934; Wittfogel 1957; Lees et al. 2014)iii (e.g. Wilkinson et al. 2012; Balbo et al. 2013)iv (Marcolongo & Mozzi 1998; Cattani & Salvatori 2008; Mantellini et al. 2008; Mantellini 2013; Stride et al. 2009; Harris 2010) v (Hiebert 1994; Sarianidi 1990; Sarianidi 2002; Luneau 2013)vi (Sarianidi 1990:74 ; Lamberg-Karlovsky 2013)vii A depe (or tepe) is an anthropogenic mound—the regional equivalent of the better-known near-eastern 'tell'.viii(Kohl 1984; Sarianidi 1990; Hiebert 1994; Cattani et al. 2008b; Rouse & Cerasetti 2014) ix(Cattani et al. 2008a; Sarianidi & Dubova 2011; Esearch et al. 2013; Markofsky 2013; Rouse and Cerasetti 2014) x The sites Egri Bogaz 1-4, discovered by Masimov in the 1970s and 1980s, are named after a well in the vicinity.xi (Sarianidi 1990)xii (Udeumuradov, pers. comm.)xiii (Orlovsky et al. 2004; Babaev et al. 1994)xiv (Cremaschi 1998; Cerasetti 2006; Cerasetti et al. 2008) xv (see discussions in Weiss 1993; Staubwasser et al. 2003; Rosen 2007; Harris 2010; Salvatori 2008b) xvi (Marcolongo & Mozzi 1998)xvii (Babaev et al. 1994)xviii (Moore et al. 1994)xix (Fleskins et al. 2007)xx (Fleskins et al. 2007)xxi (Suslov & Gershevsky 1961: 457-458)xxii (Hiebert 1994:39)xxiii (Kohl 1984; Sarianidi 1990)xxiv (Gerasimov & Brice 1978)xxv (Cremaschi 1998: 15)xxvi (Cremaschi 1998; Cerasetti & Mauri 2002; Cattani & Salvatori 2008)xxvii (Salvatori 2008b)xxviii (Cleuziou et al. 1998)xxix (Cerasetti et al. forthcoming)xxx (Cattani et al. 2008a; Cattani et al. 2008b; Rouse & Cerasetti 2014)xxxi (Sarianidi & Dubova 2011; Boroffka, pers. comm.)xxxii (Rossi-Osmida 2011; Rossi-Osmida 2007)xxxiii (Salvatori 1998)xxxiv (Markofsky & Bevan 2012; Markofsky 2013)xxxv (Bondioli & Tosi 1998; Cattani & Salvatori 2008)xxxvi (see Williams 2012 for applications in ancient Merv, Turkmenistan)xxxvii (see methodologies and classifications in Bondioli & Tosi 1998; Cattani & Salvatori 2008 also based on data in the GIS database for the AMMD project.)xxxviii (Udeumuradov, pers. comm.)xxxix (Salvatori 2008b: 62, footnote 6) xl (Cerasetti & Rouse n.d.; Rouse & Cerasetti 2014; Cattani et al. 2008a)xli (Cattani & Salvatori 2008)xlii (Fleskins et al. 2007)xliii Based on GIS data from the Archaeological Map of the Murghab Delta projects and Cattani and Salvatori 2008.xliv (Sarianidi 1990)xlv (Masimov 1979; Kohl 1984; Rossi-Osmida 2007; Rossi-Osmida 2011; Lamberg-Karlovsky 2013; but see criticism of proposed C14 dates in Salvatori 2007) xlvi (Casellato et al. 2007)xlvii (Jacobsen 1960)xlviii (Tamburrino 2010)xlix (cf Salvatori 2007)l (A. Ninfo, pers. comm.; Salvatori 2007: 18)li (Lamberg-Karlovsky 2013; Salvatori 2007)lii (cf Salvatori 2002)liii (Hiebert 1994)liv (Masson 1988: 27, 92; Rossi-Osmida 2007)lv (Cattani & Salvatori 2008; Salvatori 2008a: 65, Figure 5.5)

lvi (Cattani 2008a: 144)lvii (see forms in Cattani et al., 2008a: 144, Figure 10.10; Hiebert, 1994: 71, Figure 4-40) lviii (Cattani & Salvatori 2008)lix (Rouse & Cerasetti 2014)lx (Lyapin 1991)lxi (Cattani et al. 2008b: 140, Figure 10.7)lxii The test pit between these two, 40m west of the valley, was situated on a high dune ridge and not dug.lxiii Further research after the submission of this article suggest that the stratigraphic association between the clay layers eastand west of the dune ridge may not be so well-defined, and OSL dates, as-yet unpublished, suggest a substantial erosional event prior to the aeolian deposition. Further research is therefore needed to determine the character and continuity of the takyr-like deposits in the vicinity of the dune ridge.lxiv (e.g. Adams 1965)lxv (cf Brown 1997: 279)lxvi(Markofsky & Bevan 2012; Markofsky 2013)lxvii (Cattani & Salvatori 2008)lxviii (Wilkinson 1982)lxix (Markofsky 2013)lxx (Cattani & Salvatori 2008: 6)lxxi (Hiebert 1994: 20)lxxii (see Lamberg-Karlovsky, 2013; Gabriele Rossi-Osmida, 2011; Gabrielle Rossi-Osmida, 2007 for discussions on spa-tial relationships between the sites of Adji Kui 1 and 9)lxxiii (Algaze et al. 1992)lxxiv (Hakemi & Sajjedi 1988)lxxv (Cleuziou et al. 1998)lxxvi (cf Bintliff & Snodgrass 1988: 510)lxxvii (Markofsky & Bevan 2012; Markofsky 2013)lxxviii (Moore et al. 1994)lxxix (Wilkinson 1993: 558)lxxx (Cazanacli et al. 2002)lxxxi (Sherratt 1980)lxxxii (Kendal et al. 2011)lxxxiii (Wilkinson et al. 2012) lxxxiv (Sherratt 1980)

Acknowledgments

I wish to express my sincere gratitude to Tim Williams and Andrew Bevanfor their encouragement and support of this research. Thanks also go to Gai Joraev, MaurizioTosi, Barbara Cerasetti, and the University of Bologna researchers associated with the Ar-chaeological Map of the Murghab Delta without whose support this work would not have beenpossible. I also would like to thank Marco Madella, Andrea Balbo, Francesc Cecilia Conesa and the CaSEs team at IMF-CSIC for their support in developing this work in light of newer socio-ecological research and theoretical frameworks. I also wish to express my sincere thanks to Tony Wilkinson, who sadly passed away last year, for his encouragement of my re-search and his pioneering contributions to the field, which strongly shaped my own ways of thinking. Finally I would like to thank the anonymous reviewers for their helpful suggestions.