A change in landscape: Lessons learned from abandonment of ancient Wari agricultural terraces in Southern Peru

Journal of Environmental Management xxx (2017) 1e11

Contents lists available at ScienceDirect

Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman

Research article

A change in landscape: Lessons learned from abandonment of ancient Wari agricultural terraces in Southern Peru ~ o a, *, Patrick Ryan Williams b, Megan L. Hart c Ana C. London a

Department of Earth Sciences, Lindenwood University, 209 S. Kingshighway, St. Charles, MO 63301, USA Department of Anthropology, Field Museum of Natural History, 1400 S. Lake Shore Drive, Chicago, IL 60605, USA c Department of Civil and Mechanical Engineering, University of Missouri Kansas City, 5110 Rockhill Road, Kansas City, MO 64110, USA b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 2 July 2016 Received in revised form 2 January 2017 Accepted 6 January 2017 Available online xxx

Ancient agricultural terrace practices have survived for millennia, sustaining populations through extreme climatic shifts and political regime changes. In arid regions with abrupt relief such as Southern Peru, agricultural terracing is undergoing a resurgence, as has seen revitalization of once abandoned terrace and hydraulic systems. Wari terraces at Cerro Baul provide clues to past cultural practices. They also document sustainable farming practices by using resilient land management techniques which can help combat desertification and degradation of arable lands. Three abandoned Wari terrace systems were mapped using microtopographic methods, the erosion patterns examined, the states of preservation compared, and then the design contrasted with modern terracing practices in the Moquegua Valley. In order to negate the harmful effects of desertification, rehabilitation and reconstruction of these terraces using ancient knowledge and techniques may be necessary. Rehabilitation must be conducted with consideration for preservation of cultural patrimony that may be encountered within the terrace treads or riser structures. With future climatic shifts impacting vulnerable dryland areas more than others, the ability to resiliently respond to these changes may be found in the lessons learned from ancient farming techniques such as the Wari. © 2017 Elsevier Ltd. All rights reserved.

Keywords: Peru Wari Terraces Erosion Resilience Sustainability

1. Introduction Abandoned agricultural terraces hold the key to understanding the cultural, environmental, and agricultural practices of ancient civilizations. In Southern Peru abandoned Wari agricultural terraces are found in the Moquegua Valley, mostly on the western slopes of Cerro Baul (2200 m a.s.l.) and on the south slopes of Cerro Mejia (2250 m a.s.l.) (Fig. 1A). In order to understand the causes of abandonment of Wari terraces, analysis and exploration of the relic features is necessary. The arid environment, limited vegetative cover, and known original configuration or morphology make these abandoned terraces ideal for determining the causes and the rates of degradation of these once productive lands. Understanding these aspects of land management provides hints at how the Wari resiliently negotiated local and global climatic shifts and environmental changes. This study examines the ancient use of abandoned Wari

agricultural terraces, the causes of abandonment, the current state of preservation, and examines how management of these terraces may impact current farming and land management techniques. Patterns of erosion are derived from detailed ground surveys and intensive three dimensional terrestrial laser scanning, basin analysis, isochronology, and soil characterization. This Wari agricultural system was abandoned around 1000 AD. Abandonment may possibly be due to changing climatic conditions including droughts, flash flooding and landslide events due to increased precipitation, and the political collapse of the Wari colony that destroyed the labor organization mechanisms needed to maintain the intervalley canal and terrace systems it supported (Williams, 2002). Since these terrace systems required extensive and labor intensive maintenance, and the slopes on which they were located were prone to be highly unstable, especially in times of intense precipitation, they were inherently vulnerable to climate shifts and human labor availability (Williams et al., 2005). 2. Study area

* Corresponding author. ~ o). E-mail address: [email protected] (A.C. London

The agricultural systems in this study are located in the slopes of

http://dx.doi.org/10.1016/j.jenvman.2017.01.012 0301-4797/© 2017 Elsevier Ltd. All rights reserved.

~ o, A.C., et al., A change in landscape: Lessons learned from abandonment of ancient Wari agricultural Please cite this article in press as: London terraces in Southern Peru, Journal of Environmental Management (2017), http://dx.doi.org/10.1016/j.jenvman.2017.01.012

2

~ o et al. / Journal of Environmental Management xxx (2017) 1e11 A.C. London

summer between December and March, yielding a total average of <25 mm annually. Ice core records of the Quelccaya ice cap (5670 m a.s.l.), located 350 km northeast of the study area, show fluctuations between drier and wetter conditions during the last thousand years (Thompson et al., 1985, 1986). The ice and lake Titicaca water level records shows wetter periods from 760 to 1040, and 1870e1984 A.D., and drier periods from 1040 to 1500 A.D. with particularly severe dry conditions during 1250e1310 A.D. (Thompson et al., 1985; Binford et al., 1997; Abbott and Anderson, 2009); wetter periods predominated from 1500 to 1720 A.D., followed by drier periods from 1720 to 1860 A.D. (Thompson et al., 1986; Binford et al., 1997; Abbott and Anderson, 2009). These patterns were confirmed in part by archaeological record that indicate a dry period and decreasing discharge in the Moquegua (Osmore) River €chtle and Eitel, 2012), and by historic records around 1000 AD (Ma from the Moquegua Valley which indicate a wet period between 1857 and 1870 (Ortlieb, 1995). Drought effects prevail at altitudes of ~ o events are present (Satterlee et al., 2000). 2000 m when El Nin 2.1. Climate and its influence on Wari agricultural and irrigation practices The Wari culture was an empire occupying the lowland areas of Southern Peru between 600 A.D.-1000 A.D. (Moseley et al., 2005). Irrigation systems were developed to divert water from nearby streams emanating in the highlands to agricultural fields on terraces constructed into the sides of the mountain and alluvial fans at ~ o et al., 2013: 508; Fig. 1B). Cerro Baul is the base Cerro Baul (London a towering 2000 foot high mesa, which formerly served as a regional political seat and ceremonial center (Nash and Williams, 2009), and based on radiocarbon dates on irrigation infrastructure as well as associations with dated archaeological settlements, we can date these terraces exclusively to the Wari period (Williams, 2003: 167). The principal irrigation canal that fed these terraces was the largest agricultural work constructed in the valley in premodern times and provided Wari a means of control over the surrounding landscape (Williams, 2006). Competition over the control of water was potentially critical during climate change events and likely proved to be a source of conflict between groups occupying the region (Williams, 2002). Wari people settled around Cerro Baul (Fig. 1A and B) around 600 A.D. (Moseley et al., 2005) and introduced terraced agriculture in the high sierra (2000e2500 m a.s.l.). Decreased precipitation between 660 and 680 A.D. and 700e720 A.D. caused the colonies to contract, while increased precipitation from 750 to 1000 A.D. led to expansion of agricultural activities (Williams, 2003; Moseley et al., 2005). 3. General Wari agricultural terrace construction Fig. 1. A) General location of sites investigated with Cerro Baul and Cerro Mejia as reference points. B) Known extents of Wari agricultural production and hydraulic works, 600 A.D.-1000 A.D. Adapted from Williams, 2006.

Cerro Baul. This terraced mountain consists of Paleogene-Neogene rocks of the Moquegua Formation, a sedimentary formation composed of conglomerates, coarse sandstones and lithic tuffs which form the cap rock of Cerro Baul (Bellido, 1979; Martinez and Zuloaga, 2000). Pleistocene alluvial deposits, and early Holocene sands and gravels form the lower slopes (Martinez and Zuloaga, 2000). In February 19, 1600 AD., a massive eruption of the Huaynaputina Volcano deposited a layer of ash over the study area (Thouret et al., 1999) forming isochronous marker horizons used to date natural and anthropogenic processes. Cerro Baul is located in an arid environment characterized by thin soil draped over rocky terrain. Precipitation in the study area is limited, but occurs as short, intense events during the Austral

Constructed farming terraces in this region consist of stone wall risers with a flat surface or terrace tread on which agricultural production occurs (Fig. 2A). Preserved stone risers show construction which used locally sourced stones assembled into a wall and spackled together with a mud based mortar. Vertical risers are continuous in lineal contour; they vary in height between 80 cm and 1 m with variations depending upon the slope of the natural terrain and (McEwan and Williams, 2012). Agricultural terraces in this area were installed along natural contour breaks such that gravity driven hydraulic distribution was possible. Cultivation of the slopes in a terraced manner allowed for a maximization of surface area for agricultural production while maintaining an equilibrium with water inflow for irrigation. Southern Peru's arid environment coupled with intensive rainfalls can cause extensive wash and erosion without careful management. The Wari people were focused on farming and environmental management by

~ o, A.C., et al., A change in landscape: Lessons learned from abandonment of ancient Wari agricultural Please cite this article in press as: London terraces in Southern Peru, Journal of Environmental Management (2017), http://dx.doi.org/10.1016/j.jenvman.2017.01.012

~ o et al. / Journal of Environmental Management xxx (2017) 1e11 A.C. London

3

Fig. 2. A) Cross-section of a Wari agricultural terrace near Cerro Petroglifo. B) Cross-section of El Paso Wari Canal along the contour of Cerro Mejia. C) Segment of El Paso Wari Canal along the slopes of Cerro Mejia (left). D) Remnant of Wari Distributary Irrigation Canal on the northern slope of Cerro Baul.

enabling cultivation while reducing erosion and maximizing irrigation water use, i.e. hydraulic equilibrium. In this manner, Cerro Baul terraces were sustainably built to compensate for the increased risk of erosion and climatic variance. The initial morphology of the terraces at the time of their abandonment can be reconstructed from an understanding of their original sustainable design and construction. The terraces were created by cutting slopes and filling of the natural slopes (Fig. 2A; Treacy and Denevan, 1994) using local soil and possibly enriched top soil from organic additions (Williams et al., 2005). Construction

of the terraces was phased to minimize labor output by first cutting the slope, placing the rock wall to add stability to the slope, and then infilling the cultivation surface area (McEwan and Williams, 2012). Rock walls were also carefully constructed so that at the base of the terrace wall or riser, a porous drainage layer of gravel was placed to alleviate the build-up of hydraulic pore pressure associated with rapid infiltration (McEwan and Williams, 2012). A layer of organically rich topsoil was spread on top of the gravel and the surface or terrace tread was leveled for cultivation and sloped to optimize water flow from the terrace tread to the next terrace down

~ o, A.C., et al., A change in landscape: Lessons learned from abandonment of ancient Wari agricultural Please cite this article in press as: London terraces in Southern Peru, Journal of Environmental Management (2017), http://dx.doi.org/10.1016/j.jenvman.2017.01.012

~ o et al. / Journal of Environmental Management xxx (2017) 1e11 A.C. London

4

slope (McEwan and Williams, 2012). Retaining walls were inclined into the slope to minimize the driving force and overtopping moment of lateral earth movement on severe slopes and to lend stability by allowing for maximize drainage of the fill material (Treacy and Denevan, 1994). In this manner the Wari construction utilized a common and modern canting technique of retaining walls that prevents the slope from straining during intense rainfalls and compensates for creep-strain over time (McEwan and Williams, 2012). Wari retaining wall construction was resilient over extended periods of time (~ma) and remain in place today, albeit at diminished proportions and height reductions. 3.1. Wari hydraulics systems If desertification processes are to be stopped on these terraces and the erosional processes reversed, then a crucial step to reclamation and cultural patrimony preservation is the rebuilding of a hydraulic system. The ancient hydraulic systems were open channel irrigation hydraulic structures utilizing basic basin drainage capture design with equanimity distribution. Being able to utilize the same or similar irrigation pathways for a rebuilt or modernized terraced agricultural area is crucial for establishing a resilient infrastructure and food source for this severely arid and resource poor nation. Agricultural terraces require intense water management. The Wari created an extensive water system to support and distribute water equitably (McEwan and Williams, 2012), for example, the El Paso canal which drew water from the Torata River to irrigate the slopes of Cerros Baul and Mejía (Fig. 1B). The principal canal was more than 14.2 km in length, this irrigation system brought water from the Tumilaca River to the slopes of Cerro Baul and into the Tumilaca drainage (Fig. 1B). A secondary canal flowed to the Quebrada Cocotea as a subsidiary of the El Paso canal and together with the El Paso canal created the longest canal system in Moquegua's pre-modern history. The principal canal along the main channel had a discharge capacity of 400 L of water per second at the point it reached the base of the urban settlements at Cerros Baúl and Cerro Mejía (Williams, 1997, Fig. 2B and C). This same canal irrigated up to 324 ha of agricultural land on the slopes of the colony's three hills (Fig. 2D; Williams, 2006). Upstream near its source, the canal likely had a discharge capacity in excess of 1000 L per second and irrigated hundreds of additional hectares of land (Williams, 2006). The original intake and the irrigated areas near the Torata River are under cultivation today, and only small fragments of the original Wari system remain intact in the upper reaches of the system. Canals conduct water, silts, and nutrients to the planting surfaces. The total irrigated area of the El Paso canal system would have approached 1000 ha of which less than half is preserved today (Williams, 2006). Construction costs for the lower half (324 ha) of the system near Cerro Baul is estimated at least 91,000 man-days of labor to construct the documented 228,000 m of terracing that is still preserved around the mountain (Williams, 2006). Given the lack of preservation of over half the system, it is reasonable to assume at least double the amount of labor required to build the preserved segments would have been needed to complete the entire project. The estimated original population of 2000 inhabitants would have required several years to construct the system, and a substantial effort to maintain it through time (Williams, 2006). 4. Current state In order to assess the current state of preservation of the Wari agricultural terraces and determine the pattern of degradation since abandonment, a systematic field survey of the individual

terraces and risers was completed using both hand surveying equipment and terrestrial laser scanning. Sites selected are part of the large group of stone-walled agricultural and cultural terraces constructed along the slopes of Cerro Baul and the adjacent alluvial fans (Fig. 1). The sites selected correspond to agricultural terraces constructed along hillsides with inclinations between 20 and 25 , and varying micromorphology from planar to concave slopes. 4.1. Microtopography hand surveying The microtopography of two sites within the study area (27  30 m and 20  30 m plots) was determined from a closelyspaced matrix of points surveyed with a Pro-shot™ rotating head laser level with 5 mm elevation accuracy. Measurements of elevation were made at each node of a grid with 1 m nodal spacing. A closer grid spacing (12e50 cm) was used where smaller-scale erosional features were encountered. A detailed geomorphic map of the terrace areas was made using the surveyed base map and combined with field mapping to delineate areas of channel and rill development and to assess the extent of wash and fan deposits. 4.2. Laser surveying analysis Microtopography of the three sites was collected using a Leica ScanStation 2 terrestrial laser scanner operating Cyclone version 7.2. ScanStation 2 specifications include a measurement range of 1e800 m with an accuracy of H10 mm (standard deviation) with a measurement rate capable of collecting 12,000 points per second in a field of view of 80  360 . Each terrace set is unique with different elevation changes and lateral extent which required terrace specific adjustments for data collection. In general, data was collected by measuring elevations on a 0.3 m nodal spacing for an entire terrace set. In order to minimize accumulation of registration errors inherent in laser scanning, wide baseline traverses were performed. Reflecting targets were first assembled and then placed in an appropriate portion of the terrace in order to maximize work efficiency and facilitate transition to each new vantage or station for data collection. Four scanner vantage points were necessary to cover the full extents of each terraced system, and to accurately sample all the terrace treads and risers. Dual axis compensation was not employed as settlement was minimal in these locations. Reflective targets’ coordinates were acquired using a differential GPS and subsequently were georeferenced so that exact location could be employed in registration of the point-cloud during post-processing. Each terrace was also photographically documented using the three dimensional built in digital camera of the ScanStation 2. Point cloud acquisition was then performed once all spatial references were confirmed and targets acquired. Pointclouds post-processing was performed using Cyclone 7.2. Each point-cloud was registered using the automatic transformation into a global coordinate system. Raw data was then “cleaned” to eliminate data noise and ghost data from individual scans. Point clouds were then delimited using spatial units and gathered into a modelspace view after being merged using the multiple vantage points. 5. Material characterization and interpretation In order to determine the current state of erosion on the intact terraces, basic soil characterization was performed for random grab samples of representative slope forming materials for the three terraced sites: Site 1, Site 2 and Spider Site (Fig. 1A). Site selection for sampling was determined based upon visual identification of the most representative material for the entire terraced system

~ o, A.C., et al., A change in landscape: Lessons learned from abandonment of ancient Wari agricultural Please cite this article in press as: London terraces in Southern Peru, Journal of Environmental Management (2017), http://dx.doi.org/10.1016/j.jenvman.2017.01.012

~ o et al. / Journal of Environmental Management xxx (2017) 1e11 A.C. London

using a terrace towards the base of the site. Grab samples were taken from the upper 30 cm of the terrace tread surface. Visual examination of the excavated portion of the terrace where sampling occurred shows armoring of the surface with loosely aligned gravel and cobble particles (Fig. 3A) with deflation noted in the top 10 cm of the sample. Below the loose gravel armoring, a more heterogeneous sample containing coarse and fine materials was noted. Bulk particle size distribution analysis was conducted on the grab samples containing both the armored and unarmored portions of the terraces and performed in accordance with ASTM D422. The predominant weighted fraction of the soils for all sites contained >50% gravel or coarse sand sized particles (See Supplementary Material for particle size distribution curves). However, over 13% of each sample passed the #200 sieve (0.075 mm diameter), yielding a substantial amount of fine grained material. Because this fine grained material is predominantly clays derived from igneous rocks, the fine grained materials were further wet sieved through the #200 sieve prior to dispersal for hydrometer particle size analysis. Material passing the #200 sieve was then dispersed in solution containing 40 mg/L sodium hexametaphosphate, slurried using a mechanical mixer, and placed in a graduated glass cylinder with a 152H hydrometer in order to determine fine particle size distribution. Particle size distribution curves yield a poorly graded coarse grained material with coefficients of uniformity and coefficients of curvature equal to 215 to 94 and from 0.17 to 0.99, respectively. These values are exceptionally large for Cu and very small for Cc, which are reflected in the very flat grain-size curves and poorly graded nature of the soils. Atterburg limits were determined using ASTM D4318 on the <#200 sieved material. Index properties

5

yielded values of the liquid limit (LL) around 30 for all three sites and a plasticity index (PI) ranging from 14 to 18. Organic matter content (ASTM D 2974) and specific gravity (ASTM D 854-00) were conducted for all three sites on the <#200 sieve sized particles (Table 1) and were approximately 3.5% for all three sites. Further mineralogical characterization of the fine grained portion of the samples was conducted using powdered x-ray diffraction (XRD) (STP479). XRD yielded strong kaolinite peaks at 7.1 Å and weak mixed montmorillonite peaks near 30 Å. In general, all three soils have similar particle size distribution shapes, curve coefficients and percentage of fine grained material, which indicate that their origins are comparable and the processes occurring since abandonment are similar. Mechanical composition and material properties of these three sites are reflective of an erosional surface with large amounts of coarse material, but also contain a fair amount of finer materials including a good portion of organic matter. Deflation materials are typically poorly graded soils Table 1 Soil classification and mechanical properties for all sites. USCS classification

Site 1

Site 2

Spider site

GC

GC-SC

GC-SC

Plasticity index (PI) Liquid limit (LL) Coefficient of uniformity (Cu) Coefficient of curvature (Cc) Specific gravity (Gs) g/cm3 Organic matter content (%) Critical state friction angle (Fcv’) Peak friction angle (Fmax’)

14 30 215 0.35 2.38 3.85 46 36

15 33 94 0.99 2.44 3.73 54 44

18 31 115 0.17 2.29 3.10 52 41

Fig. 3. A) Surface Deflation Processes Produced Surface Armoring of Terrace Treads. Sampling excavation sites show the abundant fine particles below the surface armoring of the terraced tread. The stones forming the terrace riser are observed on the far right of the sampling excavation. B) Processed Point Cloud Rendering of Erosional Features of Site 1. Note erosional features cutting through multiple terrace risers and treads. C) Geomorphic map of Wari terraces depicting erosional channels with parallel alignment and depositional features at the base of the risers. D) Reconstructed original surface from exposures of Huaynaputina Ash indicating the original morphology of the site; original tread inclination is preferentially to the SE.

~ o, A.C., et al., A change in landscape: Lessons learned from abandonment of ancient Wari agricultural Please cite this article in press as: London terraces in Southern Peru, Journal of Environmental Management (2017), http://dx.doi.org/10.1016/j.jenvman.2017.01.012

~ o et al. / Journal of Environmental Management xxx (2017) 1e11 A.C. London

6

with a clear amount of gravel/cobble sized materials protecting or coating a finer grained material below (Fig. 3A). Grain size distribution curves from these three sites match deflation material distribution curves (Novotny and Klimes, 2014). Direct shear at constant volume testing was conducted on remolded samples of the coarse material versus the fine grained material in order to obtain values of (Fcv), the critical state angle of shearing resistance of soil). Values for the three sites on the coarse portion of the material ranged from 46 to 54 degrees, while values for the finer grained material ranged from 36 to 44 . High values for the angle of shearing resistance corresponds well with the angle of repose of the sites visually obtained. Feldspathic mineralogy is commonly considered the determining feature in high values of shearing resistance, and is in relative abundance at these sites in angular to subangular grains.

mainly at the base of the risers, and occasionally as patches on the terrace treads. This ash is commonly present on Inca (1400e1532 AD) tread surfaces located in a different irrigation system to the north to northwest of Cerro Baul, but not much on the Wari terraces. Inca terraces are well maintained and in a good state of preservation, with higher risers (1e2 m high) and more consistent ~ o, 2008). In comparison, these continuity in tread surfaces (London Wari terraces show intense degradation since abandonment in 1000 AD. In general, the observed erosion on each tread surface indicates widely varying amounts of soil removal. Net soil accumulation occurs at the base of the terrace risers and progressively grades to net erosion with distance downslope on the terrace tread (Fig. 3C). Overall, there is more erosion and soil removal than accumulation of coarse debris, which is consistent with permanent abandonment.

6. Combined results from surveying

6.2. Site 2

Combined hand and laser surveys were made of the morphology, construction, management, and maintenance of the ancient terraces for insight into the Wari land management strategies. Each site is complex and differ in original slope configuration prior to terracing. Various states of preservation and erosion were encountered on these sites the surveys and erosional processes analysis of each individual site is broken down below.

Site 2 is situated in a concave, southwest facing slope on the western flank of Cerro Baul (Figs. 1B, 4A and 4B). The terrace system was constructed about 40 m down slope from the drainage divide in this small catchment. The microtopographic mapping of this site used two different techniques, a global 85  35 m fine-scan topographic laser survey, and a detail topographic and geomorphic mapping using a rotating head laser level of a subset of terrace treads expanded over 20  30 m area. This slope surface is covered by loose gravel and coarse sand rubble (Fig. 4C). Incised channels are the most common erosional feature (Fig. 4A, B, 4D), showing variable stages of development. At the top of the slope, channels originate in the middle of the terrace tread and stop at the edge of the terrace riser. Further downslope, erosional channels show bifurcation, extending through several tread levels, forming a dendritic network (Fig. 4A and D). Midway down slope, piracy becomes common where the erosion has been more intense, beginning what appears to be an anastomosing-like channel pattern (Fig. 4D). Erosional channels vary in width from 0.1 to 0.4 m and have an average depth of 0.15 m. Coarse grained material removed from the terrace treads are mainly deposited at the base of the terrace risers and distributed throughout the subsequent tread. Alluvial-like fans are less common than Site 1 and are very subtle in development. The hillslope on which terraces were constructed is currently covered by loose rock fragments and gravel indicative of deflation processes. Surface deflation topography and migration of fine material away from the terrace tread surface is present, and coarse sediment remains on the tread surfaces as a protective armor coating material. Gravel-size sediment partially infills most of the erosional channels (Fig. 4D). Materials accumulated at the base of the risers indicates that wash processes are occurring on the entire slope surface. Due to its location on a concave slope, the surface water overflow concentrates down slope, enhancing the formation of erosional features such as rills and gullies along the axis of the concavity (Fig. 4A and B). About two thirds of the way down slope, two incised channels have developed from intense gullying. These channels are long, of variable width (between 0.3 and 1 m) and are about 1 m in depth. The absence of the Huaynaputina ash is noticeable at this site indicating surface deflation and a more pronounced state of degradation than Site 1. Pronounced and intense rilling, gullying and erosional materials are present at Site 2 versus Site 1, which also speaks to the general state of preservation. Site 2 is significantly smaller in volume and surface area than Site 1, and possesses a concave shape making it more vulnerable to surface water erosional processes.

6.1. Site 1 Site 1 is located in a planar, northeast facing hillslope on the northwestern flank of Cerro Baul (Fig. 1A). Mapping of this site used two different techniques, a global 100  100 m fine laser scanning topographic survey, and a detailed topographic and geomorphic hand mapping using a rotating head laser level. Hand mapping utilized analysis of a subset of the overall system consisting of nine representative terrace treads expanded over a 27  30 m area. The subset of terrace treads shows an overall gently inclination towards the southeast. Erosional channel development is the predominant process on the slope since abandonment (Fig. 3B, C, 3D). Welldefined channels 20e40 cm in width and about 15 cm in depth are abundant across the surface area of the slope. Several generations of channels are present, some are only a single channel without tributaries (Fig. 3B, C, 3D). Other channels are interconnected and bifurcated, cutting different levels of terrace treads (Fig. 3C) and increasing in depth with down-slope distance (Fig. 3C). Other channels show bifurcation and drainage piracy with little apparent control on how or why erosion differs across similar space. Channel development is more abundant on the lower part of the terrace system, whereas in the upper part, failure of the terrace riser via spoon-shaped notches are more common but small channels are still present (Fig. 3C. Risers at this site show different states of preservation. On the east side of the system, some of the risers have been completely removed so that the terrace system resembles a planar slope (Fig. 3B and C). While some stones in the riser remain in the original configuration, the majority of the stones are missing. Base stones are still locatable indicating that these stones were transported away by natural processes rather than anthropogenic influence. On the west slopes, risers typically remain and range between 20 and 60 cm in height. Risers were originally configured such that they were laterally continuous from east to west. Risers in the west were much higher than those in the east which could be due to grading of the original site to create a level tread surface. This may also explain why rocks from the eastern risers are missing, while not from the western risers. Huaynaputina ash is rarely preserved at this site; it is found

~ o, A.C., et al., A change in landscape: Lessons learned from abandonment of ancient Wari agricultural Please cite this article in press as: London terraces in Southern Peru, Journal of Environmental Management (2017), http://dx.doi.org/10.1016/j.jenvman.2017.01.012

~ o et al. / Journal of Environmental Management xxx (2017) 1e11 A.C. London

7

Fig. 4. A) Incised, Bifurcated Erosional Channels Cutting Through Multiple Terrace Treads on Site 2, Concave Slope. Drainage pathways are forming a dendritic pattern near the toe of the slope. B) Processed Point Cloud Rendering of Site 2, Concave Slope Showing the Development of Erosional Rills and Gullies. C) Preserved Terrace Risers on Site 2, Concave Slope. Notice the gravelly nature of the tread materials D) Incised, anastomosing-like erosional channels cutting through multiple terrace treads on Concave Slope. Of special consideration is the accumulation of gravel within the erosional channel.

6.3. Spider site Spider site is approximately 120 m wide by 80 m long, located 10 m from the drainage divide, on a concave northwest facing slope on the western flank of Cerro Baul (Figs. 1A and 5A). The majority of the terraces of this site are located in the upper two thirds of this concave slope. Terraces were constructed along the contours of the slope and are laterally continuous in length. Terrace treads are the longest at the Spider Site versus Sites 1 and 2, making Spider Site the most susceptible to surface water erosional processes, especially those that are dispersed. Erosional processes on this site consist of diffuse slope wash and concentrated erosion in narrow and long channels. Sediment removed from the slope is carried downslope, accumulating at the base of the risers and further down slope onto the subsequent treads. The accumulated erosional materials form an apron that can mask the presence of terrace risers (Fig. 5B). Some of the risers have been breached, and concave notches have formed. While these features do not necessarily compromise the structural integrity of the riser, they do penetrate half of the riser height. Notched type failures, while uncommon, tend to form efficient conduit channels for concentrating surface water and debris. Water accumulation within these gullies can be conveyed at a higher energy gradient, more intensely eroding tread material and conveying larger diameter debris. Infilling of these channels is comprised of cobbles and gravels. The primary lowering of risers occurred by slope wash, with little to no evidence of wall structural failure. Comparatively, the current state of preservation of these terrace risers is moderate. In most cases terrace risers are about 30e50 cm in height, about half of the height used in most modern agricultural terraces in the area. Despite the lowered elevation of the risers, they are mostly laterally continuous, with some ‘broken” by surface water breaching (Fig. 5B).

Fig. 5. A) Processed Point Cloud Rendering of Spider Site. Notice the Preserved Terrace Risers and development of erosional gullies. B) Incised, Erosional Channels Cutting through Multiple Terrace Risers (walls) on Spider Site. Notice the apron of sediments at the base of the terrace risers.

The concave nature of the slope serves as a natural flow convergence mechanism where multiple drainages coalesce into one (Fig. 5A). This convergence allows overland flow to concentrate

~ o, A.C., et al., A change in landscape: Lessons learned from abandonment of ancient Wari agricultural Please cite this article in press as: London terraces in Southern Peru, Journal of Environmental Management (2017), http://dx.doi.org/10.1016/j.jenvman.2017.01.012

~ o et al. / Journal of Environmental Management xxx (2017) 1e11 A.C. London

8

at the center of the slope or concavity axis, increasing the water depth and its erosive power. Two thirds of the way down from the drainage divide at the lower portion of the terrace systems, a more pronounced and deep central channel developed (Fig. 5A). This channel is connected downslope to smaller tributaries which have developed a dendritic drainage pattern. The small tributaries have variable depths from a few centimeters up to half a meter. The main channel has also depth variability from half a meter to 1.5 m. 7. Discussion Marked differences in erosion style occur on the three surveyed sites. The sites on the concave slopes of Cerro Baul show development of erosional channels over the entire slope, with one or a few prominent and deep channels. On the planar slope, the channels are long and parallel, but no flow convergence has occurred. Deposition of sediment on the terrace treads and at the base of the terrace risers was common to all three sites. The contrasting pattern of erosion and deposition observed between the slopes probably resulted from the different elevation positions within the terraced systems on the hillslopes, and the original morphology of the natural slope (e.g. concave vs planar hillslope, Fig. 1A). Channels and rills develop where there is sufficient concentration of surface water flow to overcome the internal shear strength of the soil in the treads. This suggests a flow and water collecting area threshold for channel initiation, which may explain why the concave slopes develop more pronounced erosional features towards the top of the slope versus the planar slope developing channels about two-thirds down slope. While heavy channel incision was noted at all three sites, bifurcation and stream piracy are noted only on the concave slopes and not the planar, which also is evidence for surface water concentration initiating erosion (Figs. 3C, 4A and 4D, 5A). Because some of the terraces in Site 1 and Spider Site are so close to the drainage divide, the contributing water collection area above the observation point is small. The topography of the divide areas does not concentrate flow in the upper zone, resulting in diffuse sheet flow with insufficient power to heavily incise the terrace treads. Indeed, literature suggests that with increasing aridity, a larger surface area is needed for erosional channels to develop (Montgomery and Dietrich, 1988). The examples presented here indicate that there is a need for a larger surface water collection area and a greater distance from the drainage divide for incision to initiate. The three sites examined around Cerro Baul have extensive enough collection areas that wash processes are able to establish a distinct and pervasive network of rills and gullies towards the middle and lower portion of the slopes. Slope processes are similar throughout the three sites, but the emerging surficial erosional pattern is different. Materials from these slopes show that the organic content of the soil on all three sites is nearly identical. The granulometry of the terrace material of Sites 1 and 2 is quite similar, while at Spider Site 3 has less gravelsize material and more silt (Table 1). Erosion studies in northern Peru show materials to be more erodible when increases in silt and very fine sand content are also noted (Romero et al., 2007). Soil at Spider Site contains proportionally more silt than Sites 1 and 2, suggesting it should be more readily eroded than the other two sites based upon this analysis. In reality the opposite is found from the microtopography. This suggests that the differences in the pattern of erosion cannot be ascribed to differences in composition but rather to differences in basin morphology and proximity to the drainage divide. Loose rock fragments cover Wari slopes entirely. Where erosional channels are present, this accumulation of rock fragments has an important effect in erosion processes. Poesen et al. (1994) found the influence of the rock fragment on soil erosion and

sediment yield to be scale dependent. Fragments protect the surface against raindrop impact and flow detachment; they reduce the physical degradation of soil by increasing percolation and reducing runoff; and they retard the overland flow velocity. They further find that the size of the rock fragments affected the sediment yield: for small fragments (3 cm) sediment yield was reduced whereas for larger fragments this reduction was less pronounced or sediment yield may have actually increased. Rills and small channels on preColumbian terraces are commonly filled with gravel-size particles and lack finer material. The development of these erosional features and their infilling with rock fragments indicates that the rubble is not protecting the slope; they are enhancing the removal of fine-grained material by increasing the water percolation and subsurface water flow occurs. While surface water erosional processes such as slope wash dominate long term feature development, initial erosional features may have developed by the irrigation channels delivering uncontrolled, high turbidity water onto the terrace surfaces. Wari hydraulic systems began moving massive volumes of water (393e436 L/s) during a drought period from the Tumilaca River to the agricultural terraces using ancient engineering techniques and harnessing a 1e2 drop in hydraulic gradient over approximately 14.2 km (Williams, 2006). The sheer magnitude of water utilizing an earthen, gravity feed system, spanning such a long distance is remarkable for current engineering techniques, let alone ancient civilizations (McEwan and Williams, 2012). It is estimated that the total labor investment of the original construction was nearly 100,000 days, with multitude more for maintenance (McEwan and Williams, 2012). Studies on the stability of the canals post abandonment have not been done, and the amount of time they continued to deliver water into the irrigation surfaces is unknown. Modern analogs of earthen canals show that these irrigation channel structures are very sensitive to erosion. Water flow during irrigation exhibits a turbulent behavior, conveying a high content of suspended load (Fig. 6A and B). Modern analogs require constant maintenance to preserve efficient channel operation including silt ~ or Paulino removal and repair of the earthen and rock walls (Sen Quispe, personal communication; Dayton, 2008). Revetment of the walls of the channel is necessary to disperse the energy of the turbulent flow of water and without that revetment, erosional energy is maximized. Accumulation of silt more than likely began immediately after abandonment, consequently eroding sides and disconnecting them from the agricultural systems. 7.1. Lessons learned The evolution of society and agriculture is an ongoing process, but there is much that can be learned from measuring the degradation and current state of preservation of historical features in order to improve the future of arid agriculture. An understanding of the patterns and quantification of the long and short term rates of erosion in arid regions is essential for the establishment of sustainable and resilient agricultural practices and design of infrastructure in these generally fragile areas. In addition to presenting the current state of preservation and patterns of erosion on archaeological terraces in arid Southern Peru, this paper also seeks to compare modern agricultural practices versus ancient agricultural production and land management success and failures in order to more accurately assist in the redevelopment of a sustainable agriculture in arid environments. For the purposes of informing development of sustainable agricultural practices and the design of infrastructure such as transportation corridors or water conveyance structures in dry lands, however, the pattern and rates of erosion coupled with knowledge of ancient practices may be most appropriate for rehabilitation.

~ o, A.C., et al., A change in landscape: Lessons learned from abandonment of ancient Wari agricultural Please cite this article in press as: London terraces in Southern Peru, Journal of Environmental Management (2017), http://dx.doi.org/10.1016/j.jenvman.2017.01.012

~ o et al. / Journal of Environmental Management xxx (2017) 1e11 A.C. London

Fig. 6. A) Modern Distributary Irrigation Canal with High Velocity Turbulent Flow. Water is turbid, indicating erosion upslope. B) Modern Irrigated Agricultural Field on Steep Slope. Notice the abundant rills that have developed as a product of poor land management practices. C) Modern Concrete Lined Pasto Grande Irrigation Canal.

Terraced land productivity greatly diminished in the Moquegua Valley due directly to the loss of ancient mountainside terraces and infilling and disconnection of the irrigation system. While a modern concrete lined irrigation channel (Pasto Grande Irrigation Canal, Fig. 6C) and agricultural terraces have been constructed within the Valley, they farm mainly alfalfa and other water intensive annual row crops. Alfalfa is mainly used as feed for guinea pigs or other meat and milk producing farm animals. While this provides a livelihood for farmers in this area, these terraces do not necessarily produce a sustainable and nutritious supply of food for the population of this area or beyond. Abandonment of the ancient terraces and subsistence farming techniques has left approximately 1000 ha of previously arable, perennial based farming land area in the Moquegua Valley untilled and lost from agricultural production (Williams, 2006). Continuous poor land management has also contributed to a drop in soil fertility and loss of organic matter essential for agricultural

9

production (Fig. 6B). Ancestral knowledge and preservation of ancient agricultural practices which focused on crop rotation, addition of organic materials, and polyculture is lost in favor of quick yielding and short term crop production. Archaeological analysis of plant remains from Cerro Baul reveal a range of cultigens not in significant production today, including three taxa of likely agricultural importance: Chenopodium spp., S. molle, and Z. mays (Sayre et al., 2012; Williams et al., 2008). The two grains are important for subsistence farming and essential for nutritional support of the population. Approximately 40% of the Earth's current land surface is drylands, with 30% of that located in arid regions (Fig. 1A; World Meteorological Organization, 2005). Under the cyclic and intense variances in prevailing weather patterns induced by climate change, an increasing trend towards more arid regions is expected to continue (CEPES, 2011). Concerns about the vulnerability of populations in dryland environments led the United Nations to create the Convention to Combat Desertification (UNCCD) in 1996. With climatic change again impacting the reliability and supply of readily available nutritious food in southern Peru, sustainable terraced systems are being investigated for revitalization (Kendall and Rodriguez, 2009). The rehabilitation of ancient terraced agricultural systems and farming techniques is an alternative to mitigating desertification and combating climate change while providing a method for labor productivity in poor population bases (CEPES, 2011). A known pattern of long term and acute erosional processes is therefore essential for preserving the agricultural basis of agrarian societies and protecting the way of life for native populations. Archaeological earthworks, particularly agricultural terraces, are not in equilibrium with their environment. They have been constructed to modify rugged terrain for cultivation, decrease erosion, and increase the amount of land suitable for agriculture. Terracing also enables the planted areas to be more efficiently irrigated than flat land. Wari terraces were designed to evenly distribute irrigation water over the areal extent of the agricultural surface and such that each successive terrace used the excess run of the up slope terrace. Current practices involve blanket overland flow irrigation over a plot of land for the duration of water rights use (i.e. a farmer owns 8 h of water rights which results in one main channel continuously and without direction irrigating a land surface) with ~ or Paulino no subsequent runoff reuse in down slope terraces (Sen Quispe, personal communication). Plots that are irrigated in this manner are severely rilled with wash erosion that removes fertile soil and organic materials essential for sustainable agricultural practices. Lowland farmable agricultural surfaces below Cerro Baul are managed poorly by farmers lacking an understanding of ancestral farming practices and management techniques which were successful for Wari population growth and sustained food supply for millennia. Agricultural terraces cause strong modifications of the landscape and necessitate constant maintenance to avoid erosion of the retaining walls and tread surfaces. Abandonment and subsequent mismanagement of the terraces allowed progressive destruction of both the fertility of the soil but also the hydrology of the area reverted to the natural conditions existing prior to terrace construction (García-Ruiz, 1988). 7.2. Rebuilding and rethinking terraces in Southern Peru In order to combat arid land desertification and reclaim the ancient agricultural terraces, especially the Wari terraces associated with Cerro Baul, a systematic irrigation method or technique would need to be put in place that better distributes or directly applies the supply of water from precipitation or that captures or diverts water

~ o, A.C., et al., A change in landscape: Lessons learned from abandonment of ancient Wari agricultural Please cite this article in press as: London terraces in Southern Peru, Journal of Environmental Management (2017), http://dx.doi.org/10.1016/j.jenvman.2017.01.012

10

~ o et al. / Journal of Environmental Management xxx (2017) 1e11 A.C. London

from nearby rivers. Successful terrace rehabilitation may require direct flow of irrigation waters to the root of the plant, such as drip irrigation does in row crops or orchards. Terraces allow for constant, or at least more controlled humidity, which in turn allows for soil pore space to remain at saturation for longer periods of time. Drip irrigation may not be possible for these areas or may be economically too constraining for the populations farming this area. Understanding where the ancient irrigation channels are, reconnecting them to the river sources, and maintaining the channels is essential for terrace rehabilitation in these areas. In this manner, ancient land management combines with farming modern techniques to form a resilient and more sustainable nutrition culture (Rodríguez and Kendall, 2001). Reconstruction of the cut and fill terraces associated with the Wari on Cerro Baul would also require some reconstruction of most of stone wall risers, terraced surface area, and addition of fines and organic material lost during abandonment. Stone wall risers serve the purpose of supporting the fill materials, preventing lateral slope strain, as well as alleviating pore pressure increases. A base, permeable gravel or cobble layer underlying the larger stones of the wall is important to allow pore pressure dissipation. Multiple avenues for pore pressure release is important, and therefore some area near the base of the walls where water may pool should incorporate a second level of permeability. Resilient structures require multiple layers of redundancy and alleviation of rapid pore pressure increases is essential for critical state stability. Foundation stones should be entrenched within a 50 cm minimal depth key structure below the terrace level to lend stability and prevent rotation of the wall due to slope driving forces, without the need for canting the pitch of the wall upslope. Wall reconstruction should incorporate some form of cementing agent between stones, such as the native volcanic ash which may act in a pozzolanic manner when mixed with alumina silicates, calcium silica hydrates and calcium hydroxides already present in the fine material of the risers themselves. However, some void space may be appropriate for pore pressure dissipation during sudden and intense precipitation events, but especially if maintenance of the drainage basin is impractical or improbable. While the height of a retaining wall depends upon the width of the worked terrace, and the original slope of the terrain, the existing terraces were productive when the risers were between 80 and 100 cm in height. Empirically, it appears that the frictional resistance of these soils under long term, drained, shear loading associated with farming is maximized at or near these heights. Therefore maintaining the traditional height of the riser should be appropriate for polyculture crop production if the infrastructure is maintained. Management includes continuous maintenance by removing infilling sediments from drainage base, infilling of stone joints within risers where loss of cementing agent occurs due to physical or chemical removal, and prompt replacement of foundation or load bearing wall stones. Rapidity is essential in land management to prevent acute failures associated with sudden and intense rainfall events in arid lands. Terrace soils and other farming materials may need to be imported from off site, especially the fines required for plant root growth and establishment. Terraced surfaces should be relatively level but if overland flow irrigation is used, the slope of the surface should be laterally inclined to allow for even distribution of the irrigation waters. If extensive rehabilitation is performed the classic layered terrace system may be implemented in which a base layer of larger stones to allow drainage is placed below an intermediate layer made up of smaller stones, sand and clay, and finally an upper layer of 50e80 cm of agricultural soil with organics in excess 3% (FAO, 2001, 2002). Slopes should be slightly inclined towards the slope rim, allowing laminar water to flow gradually without

causing erosion, but also such that excess water and flow through can irrigate downslope terraces. Agro-forestry and polyculture fields are strongly recommended land management for sustainable production of food crops. Polyculture, or farming of a combination of annual, native crops such as cereals, and perennials such as fruit trees or avocados, is highly recommended. Production systems utilizing this type of structure generate both economic and ecological long-term benefits as farmers diversify produce production while enriching the land and reducing erosional processes associated with intensive monoculture production or row crop production for animal feed. Planting woody species near the wall risers may act as a wind barrier protecting crops upslope, and act as a pore pressure relief valve as they tend to rapidly remove excess water. Native species and production of polyculture terraces support the mitigation of desertification in three main ways 1) native species grow with minimal irrigation requirements and are often drought and intensive rain tolerant, 2) native species grow in soils derived from the area which limits the amount of fertilizing materials necessary for optimal production, and 3) because native species limit erosion and reduce runoff and therefore runoff conveyed pollutants receiving streams are cleaner which benefits native populations as well as downstream agricultural production. Increased food and water security, reduction of erosion and mitigation of excessive erosion all contribute positively in the battle of climate change. Terrace rehabilitation can be at odds with the preservation of cultural patrimony. Revitalization of once cultivated fields brings with it a commensurate population increase and traffic in areas of fragile or susceptible cultural sites. Excavations show a close relationship between ancient terraced areas and the archaeological monuments and settlements built by the people who farmed (Williams, 2003). The increased access to cultivation areas has negative impacts on adjacent sites and the cultural artifacts they hold. Likewise, terrace rehabilitation fundamentally changes the archaeological data encoded in abandoned terrace structures, including pollen data, soil chemistry, and even the terrace construction information itself. Agrarian rehabilitation should thus be accompanied by extensive funds for research on ancient agricultural infrastructure prior to revitalization and reserve zones that preserve segments of these archaeological treasures for future research. Likewise, mitigation to potential damage of related archaeological sites and monuments must be part of any rehabilitation project. 8. Conclusions Two main erosional patterns have developed over the last 1000 years upon the three ancient, abandoned, agricultural Wari terraces examined. Small erosional channels and spoon-shaped notches developed as a direct consequence of terrace abandonment. Drainage basin shape determines the development of erosional features; concave upward slopes in map view promote concentrated flow and development of dendritic channels while slopes planar in map view promote parallel channel growth at a rate that is a function of contributing area. Lateral inclination of terrace treads does not appear to affect the surface water flow direction. Excess flow generated from a large contributing area is minimally affected by terrace construction, generating erosional features paralleling the direction of maximum declivity of the natural slope. Small contributing drainage areas do not cause concentrated flow, producing only spoon-shaped notches. While the erosional patterns differed between the sites surveyed, the material properties of the soils do not differ drastically and therefore cannot explain these differences. States of preservation of the terraces are generally considered to

~ o, A.C., et al., A change in landscape: Lessons learned from abandonment of ancient Wari agricultural Please cite this article in press as: London terraces in Southern Peru, Journal of Environmental Management (2017), http://dx.doi.org/10.1016/j.jenvman.2017.01.012

~ o et al. / Journal of Environmental Management xxx (2017) 1e11 A.C. London

be poor but are directly influenced by the erosional patterns and rates established since abandonment. Lack of terrace riser stones, infilling of erosional scours and rills, and deflation surfaces are all examples of desertification that, if left unchecked, will continue to reduce the arability of this land. Drastic climatic shifts may intensify degradation, depleting the limited organic matter and soil phosphorus and nitrogen remaining under the deflation armoring where fine grained material can fix these nutrients essential for agricultural production. Agricultural practices have increased in the Moquegua Valley, but these practices may not be the best approach for preserving long term arable land in this arid environment. Reconstruction of the ancient terraces to full agricultural productivity may be necessary under the overriding changes in climate towards intense aridity in the region. If reconstruction is performed, or if modern agricultural terraces are to be successful, then ancient Wari farming techniques such as designed and maintained irrigation, polyculture, crop rotation and ancient cultigen propagation may be necessary. All this activities must be done with consideration of archaeological patrimony and artifact preservation. Acknowledgements This research was made possible in part by the generous support of National Geographic Grant #W259-12. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jenvman.2017.01.012. References Abbott, M.B., Anderson, L., 2009. Lake level fluctuations. In: Gornitz, V. (Ed.), Enc. of Paleo. and Anc. Env. Springer, Heidelberg, pp. 489e492. gica Nacional, Geología del Cuadra ngulo de Moquegua, Bellido, E., 1979. Carta Geolo Hoja; 35u. Boletín 15. INGEMMET. Binford, M., Kolata, A., Brenner, M., Janusek, J., Seddon, M., Abbott, M., Curtis, J., 1997. Climate variation and the rise and fall of an Andean civilization. Quat. Res. 47, 235e248. n de andenes: una CEPES (Centro Peruano de Estudios Sociales), 2011. Recuperacio tico. Rev. Agric. 129, 6e7. alternativa para mitigar el cambio clima Dayton, C., 2008. Late Prehistoric and Modern Irrigation Agriculture in Torata, Peru. PhD dissertation. Boston University, p. 328. FAO, 2001. Conservation Agriculture. Case Studies in Latin America and Africa. FAO Soils Bull. No. 78, Rome, p. 69. FAO, 2002. Farm management and economics. In: Training Modules on Conservation Agriculture, vol. 22. FAO Land Water Bull, Rome. n de la agricultura de montan ~ a y sus efectos sobre la García-Ruiz, J., 1988. La evolucio mica del paisaje. Rev. Estud. Agro-Soc. 146, 1e37. dina Kendall, A., Rodriguez, A., 2009. Desarrollo y perspectivas de los sistemas de tudes andines, andenería de los Andes centrales del Perú. Institut français d’e  de Las Casas, p. 312. Centro de Estudios Regionales Andinos Bartolome ~ o, A.C., 2008. Pattern and rate of erosion in arid environments derived from London the study of Inca agricultural terraces in Southern Peru. Geomorph 99 (1e4), 13e25. ~ o, A.C., Hart, Megan L., Williams, Patrick Ryan, Hente, Megan L., 2013. London Ground based lidar of ancient andean agricultural systems. In: Proceedings of the 2013 Digital Heritage International Congress. IEEE, New York, pp. 507e510. €chtle, B., Eitel, B., 2012. Fragile Landscapes, Fragile Civilizations d How Climate Ma

11

Determined Societies in the Pre-Columbian South Peruvian Andes. Catena, pp. 62e73.  gico del cuadra ngulo de Moquegua, Martinez, W., Zuloaga, A., 2000. Mapa geolo n digital. Bol. Ser. A Carta Geol. Nac. (no 15) INGEMMET. escala 1:50,000 versio McEwan, G., Williams, P.R., 2012. The Wari built environment: landscape and architecture of empire. In: Bergh, S. (Ed.), Wari: Lords of the Ancient Andes. Cleveland Museum of Art, Cleveland, pp. 65e81. Montgomery, D., Dietrich, W., 1988. Where do channels begin? Nature 336, 232e234. Moseley, M., Nash, D., Williams, P., DeFrance, S., Miranda, A., Ruales, M., 2005. Burning down the brewery: excavation and evacuation of an ancient imperial colony at Cerro Baúl, Perú. Proc. Natl. Acad. Sci. U.S.A. 102 (48), 17264e17271. http://dx.doi.org/10.1073/pnas.0508673102. Nash, D., Williams, P.R., 2009. In: Marcus, Moseley J., Williams, P.R. (Eds.), Wari Political Organization: the Southern Periphery. Andean Civilization: a Tribute to Michael E. CLoA, Los Angeles, pp. 257e276. Novotny, J., Klimes, J., 2014. Grain size distribution of soils within the Cordillera Blanca, Peru: an indicator of basic mechanical properties for slope stability evaluation. J. Mt. Sci. Engl. 11 (3) http://dx.doi.org/10.1007/s11629-013-2836-9. ~ o y episodios lluviosos en el Desierto de Atacama: EL Ortlieb, L., 1995. Eventos el Nin tudes Andin. En. línea 24, registro de los últimos dos siglos. Bull. Inst. Fr. e 519e537. Poesen, J., Torri, D., Bunte, K., 1994. Effects on rock fragments on soil erosion by water at different spatial scales: a review. Catena 23, 141e166.  n agrícola en los Andes: aspectos socioRodríguez, A., Kendall, A., 2001. Restauracio  micos de la rehabilitacio n de terrazas en regiones semi- econo aridas. II Encuentro sobre Hist. Medio Ambiente 24e26. Romero, C., Stroosnijder, L., Baigorria, G., 2007. Interrill and rill erodibility in the northern Andean Highlands. Catena 70, 105e113. ~ o disaster: Satterlee, D., Moseley, M., Keefer, D., Tapia, J., 2000. The Miraflores El Nin convergent catastrophes and prehistoric agrarian change in Southern Peru. And. Past 6, 95e116. Sayre, M., Goldstein, D., Whitehead, W., Williams, P.R., 2012. A Marked Preference: ~ Chicha de Molle and Wari State Consumption Practices. Nawpa Pacha 32, 231e258. Thompson, L., Mosley-Thompson, E., Bolzan, J., Koci, B., 1985. A 1500-year record of tropical precipitation in ice cores from the Quelccaya Ice Cap. Peru. Sci. 229, 971e973. Thompson, L., Mosley-Thompson, E., Dansgaard, W., Grootes, P., 1986. The little ice age as recorded in the stratigraphy of the tropical Quelccaya ice cap. Science 234, 361e364. Thouret, J., Davila, J., Eissen, J., 1999. Largest explosive eruption in historical times in the Andes at Huaynaputina volcano, A.D. 1600, Southern Peru. Geology 27, 435e438. Treacy, J., Denevan, W., 1994. The creation of cultivable land through terracing. In: Miller, N., Gleason, K. (Eds.), The Archaeology of Garden and Field. University of Pennsylvania Press, Philadelphia, pp. 91e110. Williams, P.R., 1997. Disaster in the Development of Agriculture and the Evolution of Social Complexity in the South-Central Andean Sierra. Ph.D. Dissertation. University of Florida. Williams, P.R., 2002. Rethinking disaster-induced collapse in the demise of the Andean highland states: Wari and Tiwanaku. World Arch. 33, 361e374. Williams, P.R., 2003. Hydraulic landscapes and social conflict in middle horizon Peru. In: Forte, M., Williams, P.R. (Eds.), The Reconstruction of Archaeological Landscapes through Digital Technologies. BAR Intl Series S1151, Oxford, pp. 163e172. Williams, P.R., 2006. Agricultural innovation, intensification, and sociopolitical development: the case of highland irrigation agriculture on the Pacific andean watersheds. In: Marcus, J., Stanish, C. (Eds.), Agricultural Strategies. Cotsen Inst. of Arch., UCLA, pp. 309e333. Williams, P.R., Keefer, D.K., Nash, D.J., Sims, K., Dayton, C., Moseley, M., 2005. Envisioning the Invisible: Reconstructing the Taphonomy of Ancient Andean Agricultural Landscapes through GIS Slope Stability Models. The Reconstruction of Archaeological Landscapes through Digital Technologies II, M. Forte, Ed. Oxford: BAR Intl Series 1379, pp. 53e59. Williams, P.R., Nash, D.J., Moseley, M., deFrance, S., Ruales, M., Miranda, A., Goldstein, D., 2008. Encuentros en el Reino Wari. Bol. de Arque. PUCP 9, 207e232. World Meteorological Organization, 2005. Climate and Land Degradation. WMO-No 989, p. 34.

~ o, A.C., et al., A change in landscape: Lessons learned from abandonment of ancient Wari agricultural Please cite this article in press as: London terraces in Southern Peru, Journal of Environmental Management (2017), http://dx.doi.org/10.1016/j.jenvman.2017.01.012