at SciVerse ScienceDirect

Journal of Arid Environments 86 (2012) 12e27

Contents lists available

Journal of Arid Environments

journal homepage: www.elsevier .com/locate/ jar idenv

Pre-farming environment and OSL chronology in the Negev Highlands, Israel

Y. Avni a,*, N. Porat a, G. Avni b

aGeological Survey of Israel, 30 Malkhe Israel St., Jerusalem 95501, Israelb Israel Antiquities Authority, P.O.B. 586, Jerusalem 91004, Israel

a r t i c l e i n f o

Article history:Received 10 January 2011Received in revised form5 December 2011Accepted 4 January 2012Available online 24 January 2012

Keywords:Desert agricultureErosionIsraelLoess sedimentsNegev HighlandsOSL dating

* Corresponding author. Tel.: þ972 25314345; fax:E-mail address: yavni@gsi.gov.il (Y. Avni).

0140-1963/$ e see front matter � 2012 Elsevier Ltd.doi:10.1016/j.jaridenv.2012.01.002

a b s t r a c t

The ancient agriculture in the southern Levant is very much dependant on the interaction between thegeological and geomorphological characteristics of the desert environment and the arid climaticconditions. Field observations and luminescence dating in the Negev Highlands, southern Israel, indicatethat deposition of fluvio-loess sediments occurred mainly during the late Pleistocene glacial period.These sediments, which were transformed into loessy soil, support a high natural biomass and have anagricultural potential. Major soil erosion started after 27 ka, exposing bedrock, increasing runoff anderosion. These feedback processes were intensified during the Holocene. Since the mid Holocene, theco-existence of soil and runoff created a unique setup which enabled the establishment of desert agri-culture in the southern Levant, based on runoff-harvesting techniques. Extensive construction of stoneterrace walls on top of diachronous middleelate Holocene alluvial units in the less degraded valleys ledto the deposition of the anthropogenic unit, consisting of fine grained re-deposited loess as a by-productof the flood irrigation. This process contributed to soil conservation and counteracted the continuousnatural soil erosion and desertification. The ability to practice desert agriculture is still preserved in thesouthern Levant, and historic climate changes are not required to explain the rise and fall of the greatfarming cultures.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Ancient agricultural fields dominate large parts of the desertlandscape in the southern Levant, covering vast areas in these aridregions. In the arid Negev Highlands of southern Israel (Fig. 1)ea hilly rocky region similar to other arid regions in the southernLevant e the ancient agriculture installations are comprised ofstone-built terrace walls, alongside channels built for collecting therunoff from themountain slopes and diverting floodwater from thewadies (Ackermann, 2007; Evenari et al., 1982). The total areaexhibiting these installations comprises ca. 30,000 acres (Avni et al.,2009; Kedar, 1967; Rubin, 1990).

The sharp contrast between the vast ancient cultivated fieldsand the present harsh desert climate (80e100 mm annual rainfall,average temperature 17e19 �C, evaporation 2000e2500 m/yr), ledthe researchers working in the Negev to consider two differentexplanations for this anomaly. The first one assumed thata considerable change in regional climate occurred in historictimes, leading to the demise of the ancient agriculture (e.g. Enzelet al., 2003; Issar et al., 1989, 1991; Issar and Zohar, 2009; Orland

þ972 25380688.

All rights reserved.

et al., 2009). According to this thesis, ancient desert agricultureflourished only during periods of significantly wetter climate,enabling the establishment of agriculture activities in the desertthat are basically similar to those practiced in the semi-arid zone tothe north of the Negev. The second explanation focused on theability of ancient societies to develop adaptation mechanisms tothe harsh desert conditions, including sophisticated agriculturetechnologies, which increased their resilience to environmentalhazards. According to this approach, the environment and climateof the Middle East deserts in historical times were very similar tothe present ones (Evenari et al., 1982; Horowitz, 1979; Liphschitzand Waisel, 1987; Liphschitz et al., 1987; Rosen, 2007; Rubin,1990). Basic research focusing on the technical aspects of theinstallations and on the reconstruction of the ways and reasons ofwater collection and use, presented two contradictory views: Kedar(1967) concluded that the main use of the sophisticated systemswas to create sustainable agricultural fields by accumulating loesssoils from the nearby slopes. Evenari et al. (1982) and Shanan(2000) claimed that these installations were constructed in orderto increase runoff from the slopes and provide more availablerainwater to the fields, a view supported by other scholars (Bruinsand Ore, 2008; Yair, 1983).

In order to address these debates a good understanding of thepresent desert environment and a reliable reconstruction of past

Fig. 1. Location maps: (A) Eastern Mediterranean, showing the southern Levant. (B) Location and rainfall distribution map of the Negev Highlands (after Evenari et al., 1982). (C)Topographic map of the Negev Highlands and the study sites (modified from Avni, 2005).

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e27 13

environment in the Levant are required. Both issues rely very muchon an understanding of the geologic and geomorphologicalcomponents of the region, including the evaluation of activegeomorphological changes that occurred in the natural environ-ment before, during and after the time in which desert agricultureflourished.

The agricultural installations are usually located on ancientgeomorphic features (such as valley bottoms, hills slopes and soils)that existed at the timewhen agriculturewas first introduced to theregion. In most cases, these geomorphic features were preservedunder the foundations of the agriculture installations. Therefore,a careful study of these physical features allows a clearer recon-struction of the paleo-environment of the region prior to humanintervention, and an evaluation of the environmental changes sincethen. Furthermore, the nature of the environmental conditions,

including the paleo-climate prevailing in the agricultural phase canbe deduced from reconstruction of the maintenance activitiespracticed by the ancient farmers, which reflect the nature of theerosive and sedimentation processes that prevailed in the Negevduring that time.

The chronology of the sediments underlying the agriculturalinstallations is not well known. Such sediments were described byBruins (1986) at Qadesh Barnea, at the western edge of the Negevhighlands, but were not dated. The only absolute ages are severalOSL ages from various sediment types at Nahal Zena (see below)that underlie the agricultural installations, which were in the rangeof 64 ka to 4 ka (Avni et al., 2006), suggesting a complex sedi-mentological history.

The aim of the present paper is to reconstruct the physicalenvironment that existed in the Negev Highlands of the southern

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e2714

Levant prior to the time in which the extensive ancient agricul-tural installations were erected. It will discuss the geomorpho-logical and environmental factors which facilitated agriculturalactivities in the deserts of the southern Levant, provide the age ofthe geological substrate below the agriculture installations usingOSL age determinations, and discuss the paleo-climate that pre-vailed during that time. The comparison between past and presentdesert conditions in the southern Levant underscores the possi-bilities of maintaining sustainable desert agriculture in thepresent.

1.1. The Negev Highlands

Most of the arid and semi-arid regions of the southern Levantare mountainous with hilly rocky terrains, consisting of marinecarbonate rocks (limestone, dolomite and chalk) of Mesozoic toTertiary age. These regions include most of Jordan, Syria, southernIsrael, northern Sinai and northwest Egypt. Among these, the NegevHighlands in southern Israel (Fig. 1) were selected for a detailedstudy. The lion’s share of the agriculture potential and the naturalrange of biomass in this region are concentrated in relativelynarrow valleys which are occasionally irrigated by flash floods. Thespecial significance of the valley bottom in desert environments isderived from their higher biological production, which depends onthe ability to increase water availability above that provided bydirect rainfall ewhich ranges between 80 and 150 mm/a. With theaddition of runoff from the rocky slopes, the flooded area in wadis(ephemeral streams) receive the equivalent of 300e500 mm peryear, the amount needed for most agricultural activities insemi-arid zones, including the cultivation of cereals and otherMediterranean crops (Evenari et al., 1982; Shanan, 2000; Yair,1983). In addition, the accumulated soil layer along the valleys isrelative deep, with good moisture storage during most of the year(Avni, 2005; Bruins, 1986; Shanan, 2000;Ward et al., 2001). Most ofthese soils originated from re-deposited fine loessy sediments,blown as dust from the nearby deserts, deposited on the slopes,later on washed into the valleys and accumulated within them toform relatively flat leveled bottoms (Avni, 2005; Avni et al., 2006).These naturally irrigated valleys also have a high value as preferredgrazing lands for Bedouin herds (Seligman et al., 1962).

Simultaneously, with the accumulation of fine grained sedi-ments within the valleys, the occasional powerful desert floodsmay erode these sediments by forming gullies and headcuts. Thesefeatures gradually retreat up the valleys, causing most of the soilerosion observed today within the valleys and endangering thenatural environment as well as modern infrastructures (Avni, 2005;Shanan, 2000; Valentin et al., 2005). Consequently, the naturalbiomass and the agricultural potential are severely reduced asa result of the concentration of the flood waters into the narrowgullies, causing a sharp reduction in irrigation efficiency (Avni,2005; Poesen et al., 2002).

The wide range and distribution of deposition and erosionphenomena within the cultivated desert valleys raises severalquestions, such as the time needed for the initiation of farming, thecurrent and the long term erosion rates, and how these relate toclimatic and population fluctuations during the Holocene. A majorquestion is the relationship between soil erosion processes and theestablishment of the ancient agriculture installations. Did theancient farmers establish their fields on bare rocks, exposed byaggressive soil erosion prior to historical times, or did theyconstruct their installations on fine grained alluvial sediment thatpredated their activities and could have been utilized directly forcultivation, with erosion taking over only afterwards? And moregenerally, what were the climatic conditions and how did thelandscape look at the time of the ancient agricultural activities?

The relations between sediment accumulation, soil formation,erosion and anthropogenic interventions in the Negev Highlands(Fig. 1) have been studied intensively (e.g. Avni, 2005; Avni et al.,2006, 2009; Shanan and Schick, 1980; Shanan, 2000; Yair, 1983).These studies benefit from the rich historical documentation andarcheological record, which began before the Early Holocene (e.g.Haiman, 1993; Lender, 1990). The ancient fields and agriculturalinstallations provide a variety of archaeologically dated featuresthat can be used as a basis for approximate estimations of soilformation and erosion rates, associated with geomorphic changessince the mid Holocene. Therefore, this region provides a perfectstudy area, enabling documentation and comparison of changes involumes and rates of erosion over several time scales, from annualto millennial, and throughout most of the Holocene (Avni, 2005;Avni et al., 2006, 2009).

1.1.1. GeologyThe Negev Highlands (Fig. 1) are a typical arid, hilly terrain at an

elevation of 600e1000 m, dissected by numerous wadies flowingnorthwestward to theMediterranean Sea and northeastward to theDead Sea basin. The valleys are incised to depths of up to severaltens of meters below the hills, and the valley floors are between 30and 200 m wide. In most cases the valley bottoms are filled withlate Pleistocene sediments, mainly fluvial, reworked loess andgravel, deposited in the region during the last glacial interval,approximately 71e18 ka (Avni et al., 2006; Bowman et al., 1986;Horowitz, 1979, 1992; Zilberman, 1992). Since the lastglacialeinterglacial transition (approximately 17 ka), erosion tookover, with the fine loess sediments eroded and transported downthe valleys (Avni et al., 2006; Horowitz, 1979, 1992), mostly to theirfinal drainage basins e the Mediterranean Sea or the Dead Sea,while the coarse gravels were deposited along the main drainagechannels.

1.1.2. ClimateThe Negev Highlands, located within the hot Sahara-Arabian

desert between N29�300-N31� latitude, has a mean annualtemperature range between 17 and 19 �C, depending on elevation.The average annual precipitation ranges from 120 mm in itsnorthern part to less than 80 mm in the south (Fig. 1); the annualevaporation range is 2000e2500 and the aridity/humidity indexP/PET (annual precipitation to potential evapotranspiration ratio) is0.16e00.6, (Beer Sheva and Sede Boqer stations, for the 1970e2002period, Kafle and Bruins, 2009), which, according to the classifica-tion of UNEP (1992; 1997), places the Negev Highlands within theArid Zone. The rainy season lasts from October to April, but most ofthe rainfalls during the winter (December to February) in low-intensity (less than 15 mm per hour) rainstorms originating fromthe Mediterranean Sea (Sharon, 1972; Sharon and Kutiel, 1986).Occasionally, small-scale (50e10 km2) short-lived (severalminutes) but intense (30e120 mm h�1) rainfalls occur (Sharon andKutiel, 1986), originating mainly from rare tropical air massespassing across the Red Sea. This synoptic system, known as the RedSea Trough, yields occasional floods. Some of these floods are lostalong the course of the drainage system due to transmission losses(Yair, 1983; Yair and Kossovsky, 2002). Yair (1983) reported that thethreshold amount of rainfall necessary to generate runoff on therocky surfaces is on the order of 1e3 mm and for the soil-coveredslopes it is higher, in the order of 3e5 mm. However, undernormal conditions, only rain events that exceed 10 mm generateconsiderable floods in the drainage basins. This limit is lower insmall rocky drainage basins (Evenari et al., 1982; Shanan, 2000;Shanan and Schick, 1980; Yair, 1983; Yair and Kossovsky, 2002).

The multi-annual distribution of the precipitation in the regionduring the last 60 years, as reflected in the Sede Boqer station

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e27 15

(Fig. 1), yields an average of 91 mm. Since 1970 the central Negevregion has become more arid and the P/PET index declined from ca.0.08 to 0.06, approaching the boundary with the hyper-arid zone(Kafle and Bruins, 2009).

1.2. The archaeological setting of the Negev Highlands

Human settlement in the Negev Highlands spans from theLower and Middle Paleolithic periods to present day. Throughhistory, the Negev Highlands had been a peripheral region, thepattern and intensity of its settlement closely linked to the fortunesof themore sedentary north. This is true of the earliest urban periodin the Early Bronze age (ca. 3200e2800 BCE), when the city of Aradprofoundly influenced the character of settlement in the NegevHighlands (Amiran, 1978; Cohen, 1999). The same pattern wasmaintained during the Iron Age (10the7th centuries BCE), whenmany settlements flourished under the political aegis of thenorthern Negev towns. The first evidence for extensive agriculturalsystems in the Negev Highlands is debated between archaeologists.Some scholars suggest that the earliest fields were connected to theEarly Bronze settlements, and continued to expand during the IronAge (for example Cohen, 1999; Haiman, 1994, 2003), while othersmaintain that the local population in the Negev Highlands in theseearly periods were pastoralists based on herdsmen husbandry(Finkelstein, 1984; Rosen and Finkelstein, 1992). The suggesteddating was usually based on the relationship of the fields to nearbysettlements and on pottery finds within the fields, but this provedto be an inaccurate methodology (and see the discussion below).

Large scale agriculturalfields (ca. 30,000 acres)were constructedin the Negev Highlands during the Byzantine and Early Islamicperiods (4th�10th centuries CE), incorporated with the extensivesettlement that reached a zenith in terms of the number of sites anddensity of population (Avni et al., 2009). Hundreds of farms andagricultural installations were spread throughout the region, andtogetherwith themainByzantine townsof the regioneElusa, Avdat,Mamshit, Shivta, Nessana, and Rehovot-in-the-Negev, formed anintensive settlement pattern (Mayerson,1960; Patrich, 1995; Rubin,1990; Shereshevski, 1991; Tsafrir, 1996). Extensive fieldwork con-ducted at the Negev proved that most of the farms and agriculturalsystems functioned during the Byzantine and Early Islamic times(Avni, 1996, 2008; Avni et al., 2009; Haiman, 1995; Lender, 1990).

The ancient agricultural fields cover large areas of the NegevHighlands, and appear in several forms: from minor terraced plotsand walls, built along riverbeds of secondary order, to largeagricultural farms which were constructed along the main valleysof the region (Evenari et al., 1982; Kedar, 1967; Mayerson, 1960;Rubin, 1990; Yair, 1983). In addition to major archaeologicalexcavations and surveys in the Negev, studies have been conductedon the ancient land and water exploitation of the region, focusingon the sophisticated agricultural rain collecting systemsconstructed along the valleys.

1.3. Present soil erosion

Since 1990 soil erosion is monitored in several sites in the NegevHighlands. The main features constantly monitored are headcutmigration and gully development and their influence on the desertecosystem. Along the upper segments of the valleys above theheadcuts the stream channels are shallow and flood water runsgently, irrigating large portions of the relatively leveled bottoms.This geomorphic configuration, sometimes supported by theconstruction of shallow terrace walls, favors agriculture cultivationand has high natural biomass that can be used for grazing andherding (Avni, 2005; Seligman et al., 1962). On the other hand, theformation of headcuts and gullies within the valleys forces flood

water to concentrate in narrow channels inwhich they run fast andthe irrigation effect is confined to the gully banks. The change inirrigation efficiency below the headcuts is reflected in a sharp dropin the floral biomass, estimated at 70e90%, causing a reduction inthe grazing range value of 83e99% (Avni, 2005; Stavi et al., 2010).As a result of this process the agriculture potential of the lowersegments decreases dramatically. Monitoring current soil erosion isimportant for evaluating the complex relations between thecurrent climate, including extreme weather events, the nature oflong term natural environmental change imposed on the desertenvironments and human interventions (Avni, 2005; Poesen et al.,2002, 2003; Shanan, 2000).

1.4. Recent agriculture cultivation

The ability of the present desert environment and climate tosupport agricultural activities was examined by the Evenari team(Evenari et al., 1982), who reconstructed several ancient agriculturesystems and utilized them to produce agriculture crops includingcereals and fruits. This research had emphasized the specialadaptation of Mediterranean crops to the desert environmentresulting from their ability to survive the long dry summers. Thedeep moisture storage of the loess soils after irrigation enablesthese crops to yield a reasonable harvest in the desert agricultureplots. Recent research has focused on the agriculture activity leadby the local Bedouins of the Negev Highlands (Ashkenazi et al.,2011). Earlier unpublished reports highlighted the lack of solidagricultural knowledge among the Bedouins which resulted in pooryields. Nonetheless, fruit trees (almonds, olives, figs, grapes) plan-ted in designated terraced plots throughout the Negev Highlandsby the Bedouins survive the present harsh conditions and areimportant for evaluating the capability of the present desertenvironment and climate to support agriculture.

2. Methods

2.1. Archaeological Surveys

Systematic archeological surveys have been conducted in theNegev Highlands between 1979 and 1989, covering large sections ofthe region (e.g. Avni, 1992; Baumgarten, 2004; Haiman, 1986, 1991,1993; Lender, 1990; Rosen, 1987, 1994). Among the thousands ofsurveyed sites, six were selected for the present study, particularlythose that exhibit clear relations between the foundations of theagricultural installations and the geological substrate (see below).

2.2. Field observations

Six sites were selected for a detailed study after evaluation ofancient agriculture systems. These present awide range of drainagebasin size, type of geological substrate and density of agriculturalinstallations.

In addition, modern soil erosion is being monitored in three ofthe sites, representing the local environmental changes in theNegev Highlands during the last 25 years. Observations in thesesites (Zipporim, Revivim and Boqer, Table 1) were taken annuallyand aftermajor flood events in order to calculate the annual and theoverall rates of headcut backward migration and soil erosion. Themethodology of the monitoring follows the protocol presented byAvni (2005).

2.3. OSL dating

Samples for OSL dating (Aitken, 1998) were taken from theselected study areas, from below the archaeological terraced plots

Table 1Comparison of modern gully head retreat and soil erosion in study sites, Negev Highlands, 1990e2010. A: Catchment area upstream of headcut (km2); B: Headcut composition;C: Total retreat 1990e2010 (m); D: Average annual retreat (m/y); E: Total eroded soil 1990e2010 (m3), tons (*); F: Average annual eroded soil (m3/y), tons/y (*); G: Averageannual eroded soil (tons/km2); H: Total areawhich lost agricultural and range potential 1990e2010 (m2); I: Average annual rate of lost area (hectare/y); (*) Calculation based on1500 kg/m3 (after Ninari and Berliner, 2002).

Headcut site A B C D E F G H I

Zipporim eastern 9 Chalk and loess 7.70 0.38Zipporim central 9 Loess 19.30 0.96Zipporim western 9 Loess and conglomerate 8.80 0.44Total for Zipporim 9 19.30 0.96 2160

3240(*)108162.0(*)

18.0 790 0.004

Revivim southern 15 Loess 29.70 1.48Revivim northern 15 Loess and conglomerate 8.70 0.43Total for Revivim 15 29.70 1.48 1200

1800(*)60.090.0(*)

6.0 860 0.004

Boqer southern 100 Loess 88.10 4.40Boqer northern 100 Loess w70 3.50Total for Boqer 100 Loess 88.10 4.40 3600

5400(*)180.0270.0(*)

2.7 1400 0.007

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e2716

(Fig. 1, Table 2), by drilling horizontally into naturally erodedsections. Very fine grained quartz (88e125 mm) was extracted fromthe samples in the laboratory under dim orange light using routineprocedures (Porat, 2007). Briefly, after selecting the grain size bywet sieving, carbonateswere dissolved byHCl. The rinsed and driedsamples then underwent magnetic separation to remove heavyminerals and most feldspars. Consequently HF etching was used todissolve any remaining feldspars and etch the outer part of thequartz grains.

The purified quartz grains were mounted on 10-mm aluminumdiscs using silicon spray as an adhesive. The equivalent dose (De)values were determined on up to 12 aliquots for each sample,measured using the optically stimulated luminescence (OSL) signaland the conventional single aliquot regenerative (SAR) doseprotocol (Murray and Wintle, 2000).

Dose rates were calculated from field and laboratory measure-ments. The gamma and cosmic dose rates were measured in thefield using a portable gamma scintillator. The alpha and beta doserates were calculated from the concentrations of the radioactive

Table 2Luminescence dating results from sediments underlying archaeological terraces. Quartz sanot include outliers. Water contents are estimated at 4 � 2%. Gammaþ cosmic dose ratesthe concentrations of the radioactive elements. *samples for which the gamma dose ratesdose from burial depth. No. of discs � number of aliquots used for De calculations out o

Sample Location Depth(m)

Fieldg (mGy/a)

K (%) U (ppm) Th (ppm

Mitzpe RamonYA-2 Old red unit 2.15 548* 0.29 1.9 2.6YA-1 0.8 719* 0.49 2.0 4.5YA-3 0.5 729* 0.53 1.9 4.3N. BoqerBOQ-1 Fluvial loess 2.6 895 0.95 2.0 5.8BOQ-2 Upper massive loess 0.6 892 0.95 1.8 5.4N. RevivimYRH-19 Fluvial loess w/pebbles 1.2 518 0.54 1.7 3.7YRH-18 Upper loess 0.5 888 0.81 1.9 4.4N. ZenaNZN-12 Terrace III e base 1.2 575 0.47 1.70 3.7NZN-1 Pre-terrace 1.3 481 0.17 0.99 1.18NZN-10 Pre-terrace 3.7 508 0.34 1.40 2.5N. ZipporimZIP-6 Old red unit 2.4 948 0.69 1.90 3.7ZIP-1 Fluvial with flint tools 2.5 434 0.15 2.00 1.15ZIP-5 Loessy terrace 1.8 736 0.51 2.00 3.2ZIP-2 Fluvial loess 2.85 774 0.48 2.25 4.0ZIP-3 Fluvial loess 0.7 736 0.51 1.90 3.4ZIP-4 Top fluvial loess 0.3 759 0.56 1.90 3.4N. LavanLVN-8 Basal conglomerate 1.5 442 0.06 2.1 0.6

elements U, Th and K, measured by ICP-MS/AES. In this arid regionaverage water contents were estimated at 4 � 2%.

3. Results

3.1. OSL ages

The OSL ages, along with the field and laboratory data are pre-sented in Table 2. The errors on the ages are between 8% and 30%;the large errors resulting from poor resetting of the OSL signal atthe time of deposition (see Discussion). The OSL signal of allsamples is bright, and is dominated by the fast component (Fig. 2a).Recycling ratios are within 1 �0.1 (Fig. 2b), suggesting that the SARprotocol corrects for any sensitivity changes. A plot of the equiva-lent dose (De) as a function of preheat temperatures shows that theDe values do not change as a function of temperature (Fig. 2c),indicating that a stable signal was used for dating. Dose distributionvaries among samples (Fig. 2d) and is related to their degree ofsignal resetting and sedimentological history.

mples, 88e125 mm. Preheats ranged from 200 to 260�C. The De is the mean and doesare mostly from field measurements. Alpha and beta dose rates were calculated fromwere calculated from the concentrations of the radioactive elements and the cosmicf those measured. The ages from N. Zena were published by Avni et al. (2006).

) Ext.a (mGy/a)

Ext.b (mGy/a)

Totaldose (mGy/a)

No.of discs

De (Gy) Age (ka)

7 489 1044 � 27 11/11 43.5 � 4.6 41.6 � 4.69 679 1406 � 31 11/11 18.8 � 4.3 13.4 � 3.08 688 1425 � 31 11/11 4.1 � 0.9 2.9 � 0.7

9 1015 1919 � 95 8/12 6.2 � 1.1 3.2 � 0.69 981 1881 � 95 8/12 2.2 � 0.7 1.2 � 0.4

7 647 1172 � 58 9/8 6.9 � 2.1 5.9 � 1.98 877 1773 � 94 10/12 1.5 � 0.3 0.8 � 0.2

7 624 1207 � 64 12/12 9.7 � 2.1 8.1 � 1.84 258 740 � 53 6/6 3.4 � 01.1 4.5 � 1.56 469 983 � 56 11/12 3.6 � 1.1 3.6 � 1.1

2 791 1740 � 100 10/10 77.0 � 3.4 44.3 � 3.21 357 791 � 50 8/10 25.1 � 0.6 31.7 � 2.11 660 1398 � 79 9/10 6.8 � 0.1 4.87 � 0.281 681 1457 � 83 9/10 3.5 � 0.1 2.4 � 0.151 653 1390 � 79 9/10 1.0 � 0.03 0.71 � 0.051 689 1450 � 81 9/10 1.2 � 0.01 0.83 � 0.05

6 315 763 � 49 12/12 14.5 � 1.9 19.1 � 2.8

a b

c d

Fig. 2. OSL results for sample LVN-8. a. An OSL signal for a natural sample. b. Dose response curve, De ¼ 16.1 � 0.7 Gy. Note repeated dose points (“recycling points” at ca. 4 Gy), withratios of 0.93 and 0.92. c. Preheat plot. Note that there is no change in De as a function of preheat temperature. d. Distribution of individually measured aliquots. Note the nearnormal distribution. The average De calculated from all measurements is 14.5 � 1.9 Gy.

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e27 17

3.2. The study sites

The six selected sites are presented from the smallest to thelargest drainage basin size:

3.2.1. Upper Nahal Zin, near Mizpe RamonAn area with shallow terraced plots was surveyed in the vicinity

of modern Mizpe Ramon, located north of the Ramon Crater at anelevation of ca. 900 m ASL (Fig. 1). The agricultural plots werelocated in small tributaries ranging in size from less than 1 km2 to

Fig. 3. Shallow terraces in the Mizpe Ramon region. The darker lines crossing the wadi bedbetween the walls are filled with lighter sediment, the re-deposited loess (anthropogenic u

several km2, all within the upper basin of Nahal Zin. The plotsextend across most of the width of the valley, where low-intensityflood water flows across the entire width and does not createcentral flow channels (Fig. 3). The terrace walls are low and built ofone or two courses of undressed fieldstones, behind which isa relatively thin loess accumulation of ca. 30 cm deep (Fig. 4).

A typical terraced plot was trenched using heavy machinery ina tributary of Nahal Zin, approximately 3 km north ofMizpe Ramon.The uppermost loessy layer trapped behind the terracewall yieldedOSL age of 2.9 � 0.7 ka, which corresponds to the Iron Age (Table 2

from left to right are low, 1e2 courses stone terrace walls. Note that the terraced plotsnit).

Fig. 4. A terrace wall near Mizpe Ramon composed of one course of fieldstone (indicate by arrow). Modern plowing marks, made by Bedouins, are visible in the forefront,emphasizing the long-term use of these ancient agricultural plots.

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e2718

and Fig. 5a). The soil was deposited directly on top of older fluvialsediment composed of a mixture of re-deposited loess and angulargravels, 8-5 cm in diameter, dated to 13.4 � 3 ka (Fig. 5a andTable 2). The mixture of loess and fine gravels continues to a depthof 1.75 m below surface and rests unconformably on top of a loessyunit, rich in calcite nodules, dated to 41.6 � 4.6 ka.

3.2.2. Nahal Zena basinThe Zena basin is located in the Matred Plateau, ca. 15 km north

of Mizpe Ramon (Fig. 1). This drainage basin occupies an area of5 km2 and was extensively exploited for agriculture, mainly duringthe Byzantine and Early Islamic times (Avni et al., 2006, 2009). Thealluvial section just below the foundations of the agriculturesystems is composed of a middleelate Holocene fluvial unit, con-sisting of conglomerate mixed with re-deposited loess lenses. Thisunit was named “the pre-anthropogenic unit” by Avni et al. (2006)and has provided different ages for different locations within thedrainage basin. Two OSL ages were obtained from this site:a sample taken below the base of the agricultural installations builtalong the main channel yielded an OSL age of 4.5 � 1.5 ka, and theunit at the base of a terrace wall built along one of the sidetributaries, at a distance of 600 m downstream, yielded an OSL ageof 3.6 � 1.1 ka (Avni et al., 2006).

3.2.3. Nahal ZipporimThe Nahal Zipporim drainage basin is located approximately

30 km northwest of Mizpe Ramon and drains the northwesternpart of the Matred Plateau toward the Mediterranean Sea viaNahal Lavan (Fig. 1). The drainage basin above the Zipporimheadcut comprises 9 km2. Most of the 80e100 m wide valley isdammed by shallow terrace walls, consisting of one to three stonecourses (Fig. 5b). The section below the terraces, exposed in theheadcut (Fig. 6), is 3.5 m thick and comprises two main units:a lower conglomerate of limestone and flint pebbles, 5e20 cm indiameter, lightly cemented by a mixture of loess and grit. The baseof this unit (unit B in Fig. 5b) is not exposed in the vicinity of thepresent headcut but in the eastern bank of the gully, approxi-mately 30 m downstream. The alluvial section is deposited on top

of the bedrock of the Maastrichtian Ghareb Formation (Shaw,1947) which comprises the bottom of the valley. This unit yiel-ded an OSL age of 31.7 � 2.1 ka. This conglomerate is buried by themain loessy unit (unit A in Fig. 5b) that is composed ofre-deposited fluvial loess with lenses of gravel, 2e3 cm in diam-eter. This unit yielded an OSL age of 2.4 � 0.15 ka at a depth of2.85 m below the surface. At depths of 0.7 m and 0.3 m below thesurface, just below the shallow agriculture terraces, the loessy unityielded similar OSL ages of 0.71 � 0.05 ka and 0.83 � 0.05 ka,respectively (Table 2).

300 m below the headcut, a residual hill composed of brownloess stands above the present flat valley. Its lower part, exposed bythe present Zipporim gully, yielded an OSL age of 44.3 � 3.2 ka.About 1 km upstream of the main Zipporim headcut, fluvial loessmixed with gravels exposed below the foundations of an agricul-ture terrace, yielded an OSL age of 4.9 � 0.3 ka (Table 2).

3.2.4. Nahal RevivimThe upper Revivim drainage basin is located in the northern

Negev Highlands, ca. 7 km south-west of modern Yeroham (Fig. 1).The drainage basin above the Revivim headcut comprises 15 km2,draining the valley between the Hatira and Rehme monoclinestoward the Mediterranean Sea. The valley in the vicinity of theheadcut is 100e120 m wide and sparsely dammed by shallowagriculture terrace walls, comprising 1e2 stone courses. Thesection exposed below the agricultural terraces by the2e2.5 m-high Revivim headcut (Fig. 7) is subdivided into two mainunits: a lower unit of conglomerate (unit B in Fig. 8), composed oflimestone and chert pebbles and cobbles, 5e50 cm in diameter,mixed with re-deposited loess lenses, and an upper unit, 1e1.5 mthick, of fluvial loess, containing conglomerate lenses (unit A inFig. 8). The loessy unit contains grit and large quantities of landsnails and plant remains, indicating that the fluvial loess wasdeposited in an environment that was rich in shrubs. The lowerconglomerate unit yielded a mid Holocene OSL age of 5.9 � 1.9 ka.The upper loessy units yielded an age of 0.8 � 0.2 ka at a depth of0.5 m below surface, which is 0.2 m below the base of the agri-culture installations.

a

b

Fig. 5. Schematic sections at the Mitzpe Ramon and Zipporim sampling sites, showing the different stratigraphic units.

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e27 19

3.2.5. Nahal LavanThe Nahal Lavan basin drains the northwestern part of the

Negev Highlands to the Mediterranean Sea. Large sections of thisbasin were utilized for agricultural activities during historicaltimes, mainly in the Byzantine and Early Islamic times (4the11thcenturies CE) (Avni et al., 2009; Baumgarten, 2004). Several largeagriculture farms, based on runoff-harvesting techniques, wereerected 3 km south of ancient Shivta, where the Nahal Lavandrainage basin is 53 km2. The fields in these farmswere constructeddirectly on top of a coarse conglomerate, dated by us to19.1 � 2.8 ka. The construction of the agricultural farm started

a sedimentation phase of an anthropogenic unit, entirely controlledby human intervention, yielding OSL ages of ca. 1.8 � 0.1 to1.0 � 0.1 ka (Avni et al., 2009).

3.2.6. Nahal BoqerThis drainage basin comprises approximately 100 km2, draining

a large area in the northern Negev Highlands toward the Mediter-ranean Sea. Ancient agricultural systems cover large sections of thebasin. The Boqer headcut (Fig. 1) exposed 3.5 m of the alluvialsection comprising the substrate below the agriculture installationsthat were placed within the main flow path, which is ca.

Fig. 6. The Zipporim headcut. Note the sharp drop in the vegetation cover below the headcut (people for scale). Remains of a Late Islamic agricultural terrace wall is visible on thebottom left. This terrace once crossed the now-eroded valley.

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e2720

100 m wide. This section is constructed of a thick unit ofre-deposited loess consisting of large fraction of fine sand depos-ited together with plant remains and land snails. In the northernbank of the Boqer gully which is 20e30 m wide, a basal conglom-erate is observed, composed of unconsolidated gravels, 5e10 cm indiameter, of limestone and flint clasts. In the southern sector of thegully, a loessy section with calcite nodules is exposed, indicatingthe development of a carbonate paleosol within the loess. A similarpaleosol was found in Nahal Zipporim 6 km south-west, where itwas dated to ca. 44 ka.

Fig. 7. Close view of the Revivim headcut, showing units A (fine, re-deposit

Two OSL samples were taken in the main headcut wall from thefluvial loess unit. The OSL ages are 3.2 � 0.6 ka and 1.2 � 0.4 ka atdepths of 2.6 m and 0.6 m below surface, respectively (Fig. 8).

No agricultural terraces were found near the headcut, butcultivation at this site was practiced by modern local Bedouins, asevidenced by underground storage devices used for storing wheatand barley grains that were found in the vicinity. It is possible thatterrace walls were not constructed in this locality as they were notnecessary due to the low gradient of the valley. However, shallowterraced plots were found a few hundred meters upstream. This

ed loess) and B (coarse conglomerate with lenses of loess) as in Fig. 8.

Fig. 8. Schematic sections of the Revivim and Boqer sampling sites, showing the different stratigraphic units. Legend as in Fig. 5.

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e27 21

segment contains at present abundant natural vegetation, indi-cating a high grazing value (Seligman et al., 1962).

3.3. The compiled stratigraphic framework

The units described above can be correlated between all thesites and have a common regional distribution (Fig. 9). From theoldest to the youngest these units are:

3.3.1. Brown loess unitBrown loess unit, almost free of gravel, exposed at the base of

most of the alluvial sections or comprising low hills above thepresent valleys. This unit was reported from Mizpe Ramon, NahalZipporim, Nahal Boqer, and previously from Nahal Zena, and NahalSa’ad (Avni et al., 2006). In all outcrops, the age of this unit rangesbetween 71 ka to 41 ka. These ages correspond to the last glacial

phase of dust deposition (Crouvi et al., 2008). The unit was trun-cated by erosion, indicating that its present distribution has a relictpattern.

3.3.2. Old conglomerate unitOld conglomerate unit, originally incised into the underlying

brown loess unit and exposed by the current gullies. This unit wasdeposited within a deep channel in Nahal Zipporim, demonstratingthe deep incision which occurred ca. 32 ka in the Negev Highlands.Younger conglomerates start appearing in the region during thelast glacial maximum (LGM, ca. 18e24 ka), as demonstrated by theconglomerate unit in Nahal Lavan (ca. 19 ka).

3.3.3. Fluvial loess and gravel unitFluvial loess and gravel unit, exposed in all sections, consisting

of a mixture of re-deposited loess and fluvial gravels and rock

Old conglomerateBrown loess

Fluvial loess and gravels

Re-deposited loess - the anthropogenic unit

Agriculture terrace

X

X

X

X

X

X

X

XX

30 ka

13-2 ka

2.9-0.7 ka

71-41 ka

Terrace stone wall

Fig. 9. Generalized composite section showing the stratigraphic relations of the units found in the study area (legend as in Fig. 5).

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e2722

fragments. The composition of this unit rangeswidely and is looselycemented. It is diachronic in age e ranging from a late Pleistocenepost-glacial age (13.4 � 3.0 ka in Mizpe Ramon) to a middleHolocene age (2.4 � 0.15 ka in Nahal Zipporim headcut). Itsdiachronic nature is demonstrated by the different ages insegments from the same drainage basin. For example, in the Zip-porim basin its age ranges between 4.9 � 0.3 ka and 2.4 � 0.15 ka,although the samples are only 1 km apart. Most of the agriculturalinstallations were constructed on top of this unit.

3.3.4. Re-deposited loess unit accumulated within agricultureinstallations e the anthropogenic unit

Re-deposited loess unit accumulated within agriculture instal-lations e the anthropogenic unit. (after Avni et al., 2006). It iscomposed of fine-grained loess re-deposited by floodwater withinand behind agricultural constructions. Its age varies between theIron Age (2.9 � 0.7 ka) in Mizpe Ramon, late Roman to Early Islamicin Nahal Zena and Nahal Lavan (Avni et al., 2006, 2009), up to EarlyIslamic in Nahal Boqer (1.2� 0.4 ka) and ca. 0.8 ka in Nahal Revivimand Nahal Zipporim. Detailed description of the farms is beyond thescope of this paper.

3.4. Recent monitoring and observations

3.4.1. Soil erosionBetween 1990 and 2010, the total linear gully retreat at the

monitored sites in the Negev Highlands (Table 1) ranged between19.3 and 88.1 m, at an average rate of 0.96e4.4 m/y�1 for each maingully head. The process was accompanied by erosion of soil whichhad high agricultural and range values. The total soil losses in thesesites range between 1200 and 3600m3, at an average annual rate of60e180 m3/y�1 for each gully head, which is equivalent to 90e270tons a year. Approximately 0.08e0.14 ha of land lost theiragricultural and grazing range values in each basin under study,at an average rate of 0.004e0.007 ha y�1 (Table 1). No recoveryeffects of the gully channels were found down the valley up toa distance of 15e20 km. The relatively rapid soil erosion generatesa multi-variant environmental change, which causes the degrada-tion of biomass, biodiversity, grazing value (Stavi et al., 2010), and

agriculture potential in the region (Avni, 2005). All these indicatethat the region is under an advancing natural desertificationprocesses which increases in proportion to the rate of the headcutmigration and gully expansion on a regional scale (Avni, 2005).

3.4.2. Recent sediment accumulationDuring the 2009e2010 floods, new alluvial sediments were

deposited on top of the older alluvial sections at several locations inthe Negev Highlands, such as in Nahal Boqer, Nahal Revivim and thetributaries of Nahal Zin. In most cases these sediments attaineda thickness of 15e20 cm, were composed of fine sand and silt, andaccumulated in the wider segments of the valleys on top of thenatural vegetation (Fig. 10). Simultaneously, in the narrowsegments the valleys and especially along the drainage channelsand gullies, stream power was strong enough to cause soil erosion,mainly by the activation of headcuts and gullies.

4. Discussion

4.1. Archaeology

Traditional archaeological dating of agricultural terraces in theNegev Highlands and elsewhere in the Near East was based on thelocation of fields in proximity to well-dated settlements. Thus,suggested dating ranged from Middle Bronze age (Evenari et al.,1982), Iron Age (Aharoni et al., 1958, 1960), the Nabatean e EarlyRoman period (Negev, 1986), and the Byzantine period (Mayerson,1960; Rubin, 1990). However, all these tentative datings were notderived from stratigraphically located finds within the terraces, butrather from ‘circumstantial evidence’. Bruins (2007) has recentlydated loess deposits within agricultural terraces in the NegevHighlands using 14C and found them to be as early as the MiddleBronze Age, and many terraces throughout the Negev highlandswere dated to the Iron Age and later (Bruins, 2007). Other studiesconcentrated on geomorphologic features of the fields and paidonly little attention to numerical dating (e.g. Ackermann, 2007;Shahack-Gross, 2007). Other than Avni et al. (2006) the unitsdirectly underlying the agricultural fields and installations were notdated.

Fig. 10. Recent sedimentation in the Boqer valley above the headcut after the 2009e2010 flood season.

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e27 23

The association of the ancient agricultural systems to theByzantine period is widely accepted by most scholars working inthe Negev (e.g. Baumgarten, 2004; Haiman, 1995; Mayerson, 1960;Rubin, 1990). However, this dating, based on pottery finds withinthe fields and in adjacent settlements, could not define preciselythe chronological framework of their existence e were the fieldsconstructed already in Roman times, as research in southern Jordanand north Africa suggested (Barker and Mattingly, 1996; Barker,2007), or, perhaps, was the ancient cultivation of agriculturalsystems in the Negev Highlands manifested only later in theByzantine period (Avni et al., 2009)?

4.2. The pre-agriculture landscape in the valleys

Several units construct the pre-agriculture landscape in theNegev Highlands. The oldest unit, dated to 71e41 ka (Avni et al.,2006) and exposed in most places, is the brown fluvially re-deposited loess which is typical of the last glacial period in thesouthern Levant (Avni et al., 2006; Bowman et al., 1986; Crouviet al., 2008; Horowitz, 1979, 1992; Zilberman, 1992). Its relictnature indicates that its original distribution in space and timecould have been much wider. It expresses the thick dust depositionduring that period when low temperature and a more favorablehumid regime prevailed in the region. However, Vaks et al. (2010)shows that in the Negev Highlands speleothems were not activeduring the last 86 ka due to lack of sufficient moisture. The climaticsignificance is that the amount of precipitation did not exceed200e275 mm during the glacial phase (after 73 ka), accompaniedby a temperature decrease of 3e10 C (Vaks et al., 2010). On theother hand, the development of calcic paleosols within the loessysediments shows that the mean annual precipitation was probablyin the range of 150e200 mm (Bowman et al., 1986).

Subsequently, prior to 32 ka, deep incision began, cutting wideerosive channels in the older loessy sediments. In these channelsa coarse conglomerate was deposited, indicating a high-energychannel flow during an erosive interval before the LGM. Thedetailed history of the region during this event is unclear due to thefragmentary documentation, and needs further research. Based on

the data published by Avni et al. (2006) from the nearby NahalSa’ad (Fig. 1), we can conclude that prior to the LGM these erosivechannels were already covered by younger fluvial loessy sediments(Avni et al., 2006). During the LGM, deep incision of the majorbasins is indicated from the deposition of the coarse conglomeratein Nahal Lavan, dated to ca. 19 ka. This major erosive event and itsfluvial units, which deposited conglomerates mixed withre-deposited loess, was dated in the Sa’ad basin to be between ca.27 � 3 ka and 17 � 3 ka (Avni et al., 2006).

4.3. The late Pleistocene to Holocene substrate below theagriculture installations

The layer immediately below the agriculture installations nearMizpe Ramon was dated to the termination of the Pleistocene(13.4 � 3.0 ka). It comprises a mixture of re-deposited fluvial loessand gravels representing high energy floods transporting coarsesediments contributed from the rocky slopes and from remnants ofold loess sections that still exist in the drainage basin. This layer andthe underlying fluvial units (units C and D in Fig. 5) are typical ofthe LGM and the post LGM units. The large hiatus between this unitand the sediments trapped by the agriculture installations e morethan 10 ka e points to the erosive regime that prevailed in theregion during most of the Holocene until the agriculture terracewalls were constructed in the valley.

In the Zipporim Valley the agriculture installations lie ona geological mixture of re-deposited loess and gravels with dia-chronous ages. About 1 km upstream of the main Zipporim head-cut, the geological substrate was dated to the mid Holocene(4.9 � 0.3 ka) while near the headcut the geological substrate it isyounger than 0.7 ka. In other drainage basins in the region thegeological units are of middle to late Holocene age. This clearlyindicates their diachronous nature, not only in adjacent basins butalso within individual segments of the same basin. This dia-chronous and even chaotic pattern is also demonstrated by the lackof any order or pattern in the time of deposition relative to the sizeof the drainage basin, or the order of the fluvial channels in whichsedimentation occurred.

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e2724

We can conclude that after the late Pleistoceneeearly Holocenemain erosive phase, some dynamic equilibrium was achieved. Atthis stage the wider channels provided space for the deposition ofthe fluvial loess and fine gravels. The lower conglomerate unit inNahal Revivim basin was deposited during the mid Holocene(5.9 � 1.9 ka), after the valley had been incised in a prior erosiveevent, probably during the main incision event following thePleistoceneeHolocene transition.

The large errors on some of the OSL ages (ca. 30%) indicate thatthese coarse sediments contained a large fraction of re-depositedmaterial carrying a residual OSL signal, possibly expressing highenergy and rapid transport which prevented full resetting of thesignal. The source of the Holocene loess is from the late Pleistocenebrown loess unit that was stored in the upper catchments andepisodically washed into the active channels during floods. Recy-cled loessy sediments within the drainage basins in the NegevHighlands are well described (e.g. Bowman et al., 1986; Horowitz,1979). In general, these relicts of alluvial e colluvial origin weredeposited during the last glacial phase, accumulated to a thicknessof several meters and serve since then as a major source of fineloessy sediments in the region (Crouvi et al., 2008, 2009).

Synchronous erosion or deposition events of regional scalecannot be identified in the Holocene (Avni et al., 2006, 2009).Instead, the available data hint at a sporadic non-continuousdeposition and erosion event acting on the basin scale. This indi-cates that almost constant erosion prevailed during most of theHolocene, locally interrupted by deposition intervals of “temporalstorage” within distinct drainage basins.

In the Northern Negev the late Pleistocene floodplains of themajor rivers that drain the Negev Highland show four cycles ofdeposition, soil development and erosion (Zilberman, 1992). Theirages were estimated on the base of 14C to 1) more than 50 ka, 2)35e30 ka, 3) 27e24 ka and 4) 14e12 ka BP. However, no depositionor erosion cycle is reported during the Holocene, downstream ofthe Negev Highlands.

Therefore, the rare events of deposition during the Holocenethat formed the geological substrate below the agricultureconstructions are considered the result of local and temporalaccumulation events facilitated by local geomorphic conditionssuch as high sediment availability in the upper catchments, highsediment supply during some of the flood events, low topographicgradients and dense shrub vegetation that stabilized the valleybottoms and locally favored accumulation. In this perspective thestudy by Bull (1997) of “discontinuous ephemeral streams” is veryuseful in understanding the alternations between depositionalenvironment and erosion in arid environments. A very similarmodel is presented by Shanan (2000) for the Negev Highlands. Bothstudies pointed to the non-climatic origin of these local scale anddiachronous erosionedeposition cycles.

This model is supported by present observations indicating thatsiltyesandy loess is still being accumulated during floods on top ofthe present vegetation in the wider segments of the valleys, whilein other segments erosion, channeling and gulling continues(Figs. 6 and 10).

4.4. Agriculture intervention e the deposition of the “anthropogenicunit”

As discussed above, the geological substrate below the agricul-ture constructions consists of re-deposited loess and gravel, namedby Avni et al. (2006) the “pre-anthropogenic unit”. This unit hasagricultural value and the ancient farmers could have cultivated thenatural plot. With the construction of agricultural systems, accu-mulation of fine grained sediments was the by-product of floodirrigation, initiated almost simultaneously with the first cultivation,

as the flood and runoff water carried the fine loess sediments fromthe upper drainage basin. Later accumulation of this unit dependedvery much on the structure of the agricultural installations. Inplaces where only one row of stones was constructed to formshallow terraces, accumulationwas minor and limited to the heightof the terrace wall. This type of terrace walls prevails in thesouthern region of Mizpe Ramon where sedimentation behindshallow constructions does not exceed 30 cm. Avni et al. (2006,2009) found that siltation of agricultural plots fed by floods is inthe range of 0.45e0.8 cm/y, indicating that a section of 30 cm of theanthropogenic unit could be deposited within 40e70 years.Therefore, we conclude that sedimentation occurred shortly afterthe construction of the agricultural installations and that the OSLages of their infill probably reflects the approximate time of theirconstruction e in this case the Iron Age (ca. 2.9 ka). After sedi-mentation occurred it was preserved since no later terraceconstruction (as well as the resulting soil accumulation) took place.This area is unique as, in contrast to other regions of the NegevHighlands further north, the intensive agriculture of the Byzantineperiod was not prevalent there (Avni et al., 2009).

On the other hand, when the terraced fields were designed totrap more water for irrigation, as in the case of plots with vines andorchards, the ancient farmers had to raise the terraces walls inorder to keep the water storage. Thus sedimentation intensified asa result of long term flood irrigation. In this case the anthropogenicunit reflects the time of cultivation, between the Late Roman andthe Early Islamic periods (ca. 1800e1100 y BP). This is the case inNahal Zena and Nahal Lavan farms, in which sediment was trappedbehind the terrace walls and accumulated up to 2e3.5 m high,forming the thick anthropogenic unit (Avni et al., 2006, 2009).

During this period, on-going maintenance of the fields isobserved in the Zena and Lavan farms as the occasional floodsdamaged the terraced plots, irrigation channels and terrace walls.The nature of the maintenance work serves as an indicator ofenvironmental conditions which prevailed during that time. Thepattern of gulling and soil erosion as reflected in the reconstructedinstallations hint at the continuation of erosive conditions fromancient times to present.

4.5. Present erosion patterns

Under the current conditions, very different geomorphicconditions are found in the same drainage basin and under thesame arid climate on either side of the present headcuts. While inthe upper segment above the headcuts sedimentation takes placeduring floods (Fig. 10), aggressive erosion prevails downstream ofthe headcuts and within the gullies and channels (Fig. 6). Theheadcut gradually retreats up the valley, at rates ranging from ca.1 m to more than 15 m annually (Avni, 2005; Shanan, 2000). Thisgeomorphic trend clearly indicates that most of the soils in the aridand semi-arid Negev Highlands, as well as other regions in thesouthern Levant, are undergoing a major phase of erosion anddesertification controlled mainly by natural processes (Avni, 2005).However, the fact that in some of the upper segments of the valleyssoils are still available and are irrigated by floods, indicates thatagricultural potential still exists in the southern Levant.

4.6. Present practice of desert agriculture

At present, the Bedouin population of the Negev Highlands usessome of the ancient fields for agricultural activities such as culti-vation of cereals and fruit trees (Fig. 11). During the winter of2006e2007 and 2009e2010, the annual precipitation at the NegevHighlands was above the average rainfall, enabling the harvest ofconsiderable amounts of wheat and barley by the local Bedouin

Fig. 11. Bedouin barley field in ancient terraced plots near Sede Boqer, May 2010. The green shrubs grow along the terrace wall. (For interpretation of the references to colour in thisfigure legend, the reader is referred to the web version of this article.)

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e27 25

population (Fig. 11). The old olive trees, spread over several sites inthe Negev Highlands (Ashkenazi et al., 2011), indicate the conti-nuity of the environmental conditions during the last half millen-nium and demonstrate the absence of catastrophic, extendeddroughts during this period. This hints at the ability of most of theMediterranean crops, when irrigated by floods, to survive thepresent harsh desert conditions. In addition, the experiment con-ducted by Evenari et al. (1982) to reconstruct several ancientagriculture farms, demonstrated the vitality of these installations tosupport agricultural crops, proves the functionality of thesesystems under the present-day arid climate.

Therefore, we conclude that the ability to maintain agriculturalactivities still exists at present with no indications for major envi-ronmental or climatic change since historical times, as previouslynoted in several studies (e.g. Horowitz, 1979, 1992; Liphschitz andWaisel, 1987; Liphschitz et al., 1987; Rubin, 1990). Some of theancient farms, originally erected ca. 1700 years ago are stillpreserved untouched, representing an excellent example of long-term soil conservation which opposes the general, natural trendof soil erosion and desertification.

4.7. Did a wetter climate enable desert agriculture, or can desertagriculture persist in today’s harsh desert climate?

Since the beginning of research on ancient agriculture systemsin the Negev Highlands, possible climate fluctuations duringhistorical times were considered as an essential component facili-tating and enabling these activities. Several studies (e.g. Bar-Matthews et al., 2000; Enzel et al., 2003; Issar et al., 1989, 1991;Issar and Zohar, 2009; Orland et al., 2009; Schilman et al., 2002)suggested that climate fluctuation enforced cultural shifts in thesouthern Levant for the time period of 1800 to 1300 years ago,including the rise and fall of the major phase of desert agriculture.

However, the fundamental issue regarding the design andconstruction of the desert agriculture installations is that theywerefed by surface runoff and by floods and not by direct rainfall. Asindicated by Sharon (1972), Sharon and Kutiel (1986) and Shanan

(2000), the generation of surface runoff and floods is by highintensity rains (although of short duration) which are a typicalfeature of the desert climate. This was also studied and described byShanan and Schick (1980) and Evenari et al. (1982), indicating thatthe desert agricultural installations were planned to fit the desertclimate conditions. In addition, Vaks et al. (2010) indicate thatspeleothems were not active during the Holocene in the regionsouth of the Beer Sheva valley, including the Negev Highlands, suchthat the amount of precipitation did not exceed 300e350 mm.Another signal excluding the possibility of significantly wetperiods during the Holocene is based on the lack of any substantialcalcic soils in the Holocene sequence in the Negev Highlands,although calcic nodules are well known feature in modern loessysoils in the 200 mm belt in the BeerShava Valley (Yaalon and Dan,1974). Therefore we can conclude that the Holocene climate in theNegev Highlands did not exceed the 200 mm limit, even not duringthe relatively humid phase in the Early Holocene (Horowitz, 1979).

The present data and its integrationwith that presented by Avniet al. (2006, 2009), indicates that a wetter climate was nota significant factor in the establishment of the historical agricul-tural farms, the deposition of the fine-grained sediments withinthem or their abandonment. This conclusion is strengthened bydata from Orland et al. (2009) which shows that climate conditionswere relatively drier during the Byzantine and the Early Islamicperiods, while the desert agriculture was at its peak during theseperiods.

Beyond that, erosion and gulling started long before peoplebecame important agents in the environment, suggesting thaterosion and the resulting desertification are controlled mainly bynatural mechanisms (Avni, 2005; Avni et al., 2006). This isconsidered as “natural desertification”, acting in the arid and semi-arid regions since post-glacial times. Any anthropogenic interven-tion in the environment such as overgrazing, land degradation orland conservation as discussed above is super-imposed on themajor natural trends.

Ancient desert agriculture utilized unique geomorphic condi-tions that evolved in the southern Levant during the Holocene,

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e2726

resulting from the convergence of several factors: Newly exposedbedrock generating high runoff; abundant fertile loess brought tothe region as desert dust; and floods resulting from the desertclimatic conditions in the southern Levant. These components,particularly the loess and bedrock, have an opposite dynamicfeedback effect. During most of the last glacial period the surface ofthe Negev highlands was covered with loess (Crouvi et al., 2008;Yair, 1983, 1987, Yair and Kossovsky, 2002). On gentle slopes havinglow gradient, accumulation of fine loess sediments prevailed,forming colluvial aprons attached to the slopes, as described byBowman et al. (1986). This requires almost no runoff during sedi-mentation. On the other hand, on much steeper rocky slopes, thehigh erodibility of the loess sediments and gentle runoff generatederosion and re-deposition of the eroded sediments within thenearby valleys, mixed with some gravels and rock fragmentsgenerated from the bear rock slopes. Furthermore, the erosiveevent circa 32e31 ka clearly indicated that for some time channelerosion did occur in the region during the glacial phase.

In the transition from the last glacial climate to the post-glacialand present interglacial climates, the gradual exposure of the rockyslopes by the erosive processes exposed the bedrock oncemore andtriggered an increase in the runoff coefficient (Yair, 1983, 1987, Yairand Kossovsky, 2002). More runoff reached the valleys from theslopes, and with it -abundant loess. The increased water flow irri-gated the uneroded segments of the valleys and created the envi-ronmental conditions which facilitated agriculture to be practicedunder the desert climate. These unique geomorphic conditionswere named the “Desert Agriculture Geomorphic Window” (Avniet al., 2006) that was utilized by the ancient farmers throughouthistorical times and is still functioning at present.

Since the pre-farming alluvial units are diachronous throughoutthe middleelate Holocene, as well as accumulations of theanthropogenic units within the agriculture installations, we canconclude that no special climatic conditions were responsible forthe establishment of agricultural activities. Furthermore, the sedi-ment composing the alluvial section deposited in Nahal Zipporim,Revivim and Boqer below the agriculture installations is similar tothe composition of the recent accumulation in present-day, widechannels after major floods. This indicates a continuity of envi-ronmental conditions since at least the Iron Age up to today. Thesurviving old olive trees (Ashkenazi et al., 2011), as well as thewidespread agricultural activities still practiced by the Bedouin,demonstrate that the present environmental and climate condi-tions still facilitate the same type of farming as in historical times.This indicates that climatic changes were not necessary for theinitiation or demise of the major agricultural phase. However,a dynamic non-climatic change is imposed on the region as itmoves towards increased soil erosion, gulling and natural deserti-fication (Avni, 2005; Avni et al., 2006).

5. Conclusions

1. During the last glacial phase, accumulation of fine-grained togravelly fluvio-loess beds took place within the drainage basinsin the Negev Highlands interrupted by some erosion events.Accumulation was accompanied by the development of calcicpaleosols, indicating a mildly arid to arid climate which pre-vailed during most of the last glacial period.

2. The major shift from aggradation to erosion happened at ca.17 ka. This overall trend accelerated since thePleistoceneeHolocene transition.

3. The continuous incision in the sediment-filled wadies of theNegev Highlands caused continuous degradation of waterirrigation efficiency, agricultural soils and natural biomass, allof them indicating accelerated desertification which continued

throughout the Holocene. Soil erosion and its resultantdesertification are controlled mainly by a natural mechanism.Therefore, the name “natural desertification” is attributed tothis long term process.

4. Since the mid Holocene, the co-existence of soil and runoffcreated unique geomorphic conditions termed the “DesertAgriculture Geomorphic Window” that enabled the establish-ment of desert agriculture and the construction of numerousterraced plots and farms. These activities were sustainable inthe arid regions of the southern Levant for several millennia.Parts of these agricultural constructions are still in use today.

5. During the Holocene, soil accumulation within the agriculturalinstallations, as by-products of irrigation, counteracted thenatural trend of soil erosion and can be considered as long-term land conservation efforts applied by the ancient farmersin the arid regions of the southern Levant. At the same time, thegeneral trend of sediment erosion continued uninterrupted inthe undisturbed basins, leading to the incision of channels andgullies and the continuation of natural desertificationprocesses.

6. We conclude that historic climate changes are not needed toexplain the rise and fall of the great farming cultures. Also theability of the present Bedouin population to practice desertagriculture, utilizing the same installations built by the ancientfarmers hundreds to thousands of years ago, indicates thecontinuation of the desert conditions throughout the lateHolocene with no significant climate change.

Acknowledgments

We would like to thank the anonymous reviewers whosecomments greatly improved on the previous version of themanuscript. Fieldwork was assisted by R. Madmon and Y. Rephael.Financial support for this long term research was provided by theRamon Science Center, Mizpe Ramon, and by the Ministry ofScience and Technology of Israel.

References

Ackermann, O., 2007. Reading the field: geoarchaeological codes in the Israelilandscape. Israel Journal of Earth Sciences 56, 87e106.

Aharoni, Y., et al., 1958. The ancient desert agriculture in the Negev III. Earlybeginnings. Israel Exploration Journal 8, 231e268.

Aharoni, Y., et al., 1960. The ancient desert agriculture of the Negev V. An Israeliteagricultural settlement at Ramat Matred. Israel Exploration Journal 10 23e36,97e111.

Aitken, M.J., 1998. An Introduction to Optical Dating. Oxford University Press,Oxford.

Amiran, R., 1978. Early Arad. Israel Exploration Society, Jerusalem.Ashkenazi, E., Chen, Y., Avni, Y., Lavee, S., 2011. Olive trees in past desert agriculture

in the Negev Highlands. Acta Horticulturae 888, 353e360.Avni, Y., Porat, N., Plakht, J., Avni, G., 2006. Geomorphologic changes leading to

natural desertification processes versus anthropogenic land conservation in anarid environment, the Negev Highlands, Israel. Geomorphology 82, 177e200.

Avni, G., Avni, Y., Porat, N., 2009. A new look at ancient agriculture in the Negev.Cathedra 133, 13e44 (Hebrew, English abstract).

Avni, G., 1992. Map of Har Saggi Northeast (225). Israel Antiquities Authority,Jerusalem.

Avni, G., 1996. Nomads, Farmers and Town-Dwellers e PastoralisteSedentistInteraction in the Negev Highlands, Sixth-Eight Centuries CE. Israel AntiquitiesAuthority, Jerusalem.

Avni, Y., 2005. Gully incision as a key factor in desertification in an arid environ-ment, the Negev Highlands, Israel. Catena 63, 185e220.

Avni, G., 2008. The ByzantineeIslamic transition in the Negev: an archaeologicalperspective. Jerusalem Studies of Arabic and Islam 35, 1e26.

Barker,, G.W., Mattingly,, D.J. (Eds.), 1996. Farming the Desert: The UNESCO LibyanArchaeological Survey. UNESCO, Paris.

Barker,, G.W. (Ed.), 2007. Archaeology and Desertification e The Wadi FaynanLandscape Survey, Southern Jordan. Oxbow, Oxford.

Bar-Matthews, M., Ayalon, A., Kaufman, A., 2000. Timing and hydrological condi-tions of sapropel events in the Eastern Mediterranean as evident from spe-leothems, Soreq cave Israel. Chemical Geology 169, 145e156.

Y. Avni et al. / Journal of Arid Environments 86 (2012) 12e27 27

Baumgarten, Y., 2004. Archaeological Survey of Israel, Map of Shivta (166). IsraelAntiquities Authority, Jerusalem.

Bowman, D., Karnieli, A., Issar, A., Bruins, H.J., 1986. Residual colluvio-aeolian apronsin the Negev Highlands (Israel) as a paleo-climatic indicator. Paleogeography,Paleoclimatology, Paleoecology 56, 89e101.

Bruins, H.J., Ore, G., 2008. Runoff from loess or bedrock? Hillslope geoarchaeologyof ancient runoff farming system at Horvat Haluqim and Har Eldad in thecentral NegevDesert. Israel Journal of Earth Sciences 57, 231e247.

Bruins, H.J., 1986. Desert Environment and Agriculture in the Central Negev andKadesh Barnea During Historical Times. Midbar Foundation, Nijkerk, theNetherlands.

Bruins, H.J., 2007. Radiocarbon dating the “Wilderness of Zin”. Radiocarbon 49 (2),481e497.

Bull, W.B., 1997. Discontinuous ephemeral streams. Geomorphology 19, 227e276.Cohen, R., 1999. Ancient Settlements of the Negev Highlands I: The Chalcolithic

Period, The Early Bronze Age and the Middle Bronze Age I. Israel AntiquitiesAuthority, Jerusalem.

Crouvi, O., Amit, R., Enzel, Y., Porat, N., Sandler, A., 2008. Sand dunes as a majorproximal dust source for late Pleistocene loess in the Negev desert, Israel.Quaternary Research 70, 275e282.

Crouvi, O., Amit, R., Porat, N., Gillespie, A.R., McDonald, E.V., Enzel, Y., 2009.Significance of primary hilltop loess in reconstructing dust chronology, accre-tion rates and sources: an example from the Negev desert, Israel. Journal ofGeophysical Research 114 (F02017), 1e16.

Enzel, Y., Bookman (Ken Tor),, R., Sharon, D., Gvirtzman, H., Dayan, U., Ziv, B., Stein, M.,2003. Late Holocene climates of the Near East deduced from Dead Sea level vari-ations and modern regional winter rainfall. Quaternary Research 60, 263e273.

Evenari, M., Shanan, L., Tadmor, N., 1982. The Negev e The Challenge of the Desert.Harvard University Press, Cambridge, Mass.

Finkelstein, I., 1984. The Iron age “Fortresses” of the Negev highlands: sedentari-zation of the Nomads. Tel Aviv 11, 189e209.

Haiman, M., 1986. Archaeological Survey of Israel, Map of Har Hamran South-West(198). Archaeological Survey of Israel, Jerusalem.

Haiman, M., 1991. Map of Mizpe Ramon Southwest (200). Israel AntiquitiesAuthority, Jerusalem.

Haiman, M., 1993. Map of Har Hamran Southeast (199). Israel Antiquities Authority,Jerusalem.

Haiman, M., 1994. The Iron Age sites of the western Negev Highlands. IsraelExploration Journal 44, 36e61.

Haiman, M., 1995. Agriculture and Nomad-State relations in the Byzantine and earlyIslamicperiods. Bulletinof theAmerican Schools ofOriental Research297, 29e53.

Haiman, M., 2003. The 10th Century B.C. settlement of the Negev highlands andIron age rural Palestine. In: Maeir, A., Dar, S., Safrai, Z. (Eds.), The Rural Land-scapes of Ancient Israel. BAR International Series 1121, Oxford, pp. 71e81.

Horowitz, A., 1979. The Quaternary of Israel. Academic Press, New York, 394 pp.Horowitz, A., 1992. Palynology of Arid Lands. Elsevier, Amsterdam, 546 pp.Issar, A., Zohar, M., 2009. Facing Global Environmental Change. Climate Change

Impacts on the Environment and Civilization in the Near East. Springer.Issar, A., Tsoar, H., Levin, D., 1989. Climate changes in Israel during historical times

and their impact on hydrological, pedological and socio-economic systems. In:Leinen, M., Sarthein, M. (Eds.), Paleoclimatology and Paleometeorology. KluwerAcademic Publ., Dordrecht, pp. 525e542.

Issar, A., Govrin, Y., Geyh, M.A., Wakshal, E., Wolf, M., 1991. Climate changes duringthe upper Holocene in Israel. Israel Journal of Earth Sciences 40, 219e223.

Kafle, H., Bruins, H.J., 2009. Climatic trends in Israel 1970e2002: warmer andincreasing aridity inland. Climatic Change 96, 63e77.

Kedar, Y., 1967. Ancient Agriculture in the Negev Highlands. Bialik Institute, Jer-usalem (Hebrew).

Lender, Y., 1990. Map of Har Nafha (196) 12-01. Israel Antiquities Authority, Jer-usalem, Israel.

Liphschitz, N., Waisel, Y., 1987. The climatic and vegetatial history of the last 5000years in the Sinai Peninsula: a dendroarcheological study. In: Dvirzman, G.,Gardus, Y., Bet Arieh, I., Harel, M. (Eds.), Sinai Part 1: Physical Geography. Eretzpublishing house. Ministry of Defense, Tel Aviv, pp. 510e524.

Liphschitz, N., Lev Yadun, S., Waisel, Y., 1987. A climatic history of the SinaiPeninsula in the light of dendrochronological studies. In: Dvirzman, G.,Gardus, Y., Bet Arieh, I., Harel, M. (Eds.), Sinai Part 1: Physical Geography. Eretzpublishing house. Ministery of Defense, Tel Aviv, pp. 525e531.

Mayerson, P., 1960. The Ancient Agricultural Regime of Nessana and the CentralNegev. British school of Archaeology in Jerusalem, London.

Murray, A., Wintle, A.G., 2000. Luminescence dating of quartz using an improvedsingle-aliquot regenerative-dose protocol. Radiation Measurements 32, 57e73.

Negev, A., 1986. Nabatean Archaeology Today. University Press, New York.Ninari, N., Berliner, P.R., 2002. The role of dew in the water and heat balance of bare

loess soil in the Negev desert: quantifying and actual dew deposition on the soilsurface. Atmospheric Research 64, 323e334.

Orland, I.J., Bar-Matthews, M., Kita, N.T., Ayalon, A., Matthews, A., Valley, J.W., 2009.Climate deterioration in the Eastern Mediterranean as revealed by ion micro-probe analysis of a speleothem that grew from 2.2 to 0.9 ka in Soreq Cave, Israel.Quaternary Research 71, 27e35.

Patrich, J., 1995. Church, State and the Transformation of Palestine e The Byzantineperiod. In: Levi, T.E. (Ed.), The Archaeology of Society in the Holy Land. LeicesterUniversity Press, London.

Poesen, J., Vandekerckhove, L., Nachtergaele, J., Oostwoud Wijdenes, D.,Verstraeten, G., Van Wesemael, B., 2002. Gully erosion in dryland environ-ments. In: Bull, L.J., Kirkby, L.J. (Eds.), Dryland Rivers: Hydrology and Geomor-phology of Semi-arid Channels. John Wiley & Sons. Ltd, pp. 229e262.

Poesen, J., Nachtergaele, J., Verstraeten, G., Valentin, C., 2003. Gully erosion andenvironmental change: importance and research needs. Catena 50, 91e133.

Porat, N., 2007. Analytical Procedures in the Luminescence Dating Laboratory.Geological Survey of Israel Technical Report TR-GSI/2/2007 (in Hebrew).

Rosen, B., Finkelstein, I., 1992. Subsistence patterns, carrying Capacity and settle-ment Oscillations in the Negev Highlands. Palestine Exploration Quarterly 124,42e58.

Rosen, S.A., 1987. Byzantine Nomadism in the Negev e Results from the EmergencySurvey. Journal of Field Archaeology 14, 29e42.

Rosen, S.A., 1994. Archaeological Survey of Israel: Map of Makhtesh Ramon (204).Israel Antiquities Authority, Jerusalem.

Rosen, A., 2007. Civilizating Climate e Social Response to Climatic Change in theAncient Near East. Altamira Press, Lanham.

Rubin, R., 1990. The Negev as Settled Land e Urbanisation and Settlement in theDesert in the Byzantine Period. Yad Ben Zvi (Hebrew), Jerusalem.

Schilman, B., Ayalon, A., Bar-Matthews, M., Kagan, E.J., Almogi-Labin, A., 2002. Sea-land paleoclimate correlation in the Eastern Mediterranean region during thelate Holocene. Israel Journal of Earth Sciences 51, 181e190.

Seligman, N., Tadmor, N., Raz, Z., 1962. Range Survey of the Central Negev. TheNational and University Institute of Agriculture, Bulletin no. 67. 44 pp. (inHebrew, English Abstract).

Shahack-Gross, R., 2007. Approaches to understanding formations of archaeologicalsites in Israel: materials and processes. Israel Journal of Earth Sciences 56 (2e4),73e86.

Shanan, L., Schick, A.P., 1980. A hydrological model for the Negev Desert Highlands:effects of infiltration, runoff and ancient agriculture. Hydrological Sciences-Bulletin-des Sciences Hydrologiques 25, 3e9.

Shanan, L., 2000. Runoff, Erosion, and the Sustainability of Ancient IrrigationSystems in the Central Negev Desert. In: The HydrologyeGeomorphologyInterface: Rainfall, Floods, Sedimentation, Land Use, Proceedings of the Jer-usalem Conference, May 1999. IAHS Publ. no. 261, pp. 75e106.

Sharon, D., Kutiel, H., 1986. The distribution of rainfall intensity in Israel, its regionaland seasonal variations and its climatological evaluation. Journal of Hydrology6, 277e291.

Sharon, D., 1972. The spottiness of rainfall in desert areas. Journal of Hydrology 17,161e175.

Shaw, S.H., 1947. Geological Map of Southern Palestine 1:250,000 with ExplanatoryNotes. Government printing office, Jerusalem, 42 pp.

Shereshevski, J., 1991. Byzantine Urban Settlements in the NegevDesert (Beer Sheva5) Beer Sheva.

Stavi, I., Perevolotsky, A., Avni, Y., 2010. Effect of gully formation and headcut retreaton primary production in an arid rangeland: natural desertification in action.Journal of Arid Environments 74, 221e228.

Tsafrir, Y., 1996. Some notes on the settlement and demography of Palestine in theByzantine Period: the archaeological evidence. In: Seger, J.D. (Ed.), Retrievingthe Past: Essays on Archaeological Research and Methodology in Honor of GusW. Van Beek, Winonna Lake, pp. 269e283.

UNEP, 1992. World Atlas of Desertification. In: Middleton, N. (Ed.), first ed. EdwardArnold, Nairobi.

UNEP, 1997. World Atlas of Desertification. In: Middleton, N., Thomas, D. (Eds.),second ed. Edward Arnold, London.

Vaks, A., Bar-Matthews, M., Matthews, A., Ayalon, A., Frumkin, A., 2010. Middle-LateQuaternary paleoclimate of northern margin of the Saharan-Arabian Desert:reconstruction from speleothems of the Negev desert, Israel. QuaternaryScience Reviews 29, 2647e2662.

Valentin, C., Poesen, J., Li, Y., 2005. Gully erosion: impacts, factors and control.Catena 63, 132e153.

Ward, D., Feldman, K., Avni, Y., 2001. The effect of loess erosion on soil nutrients,plant diversity and plant quality in the Negev desert wadis. Journal of AridEnvironments 48, 461e473.

Yaalon, D.H., Dan, J., 1974. Accumulation and distribution of loess-derived depositsin the semi-desert fringe areas of Israel. Zeitschrift fur Geomorphologie, Sup-plementband 20, 91e105.

Yair, A., Kossovsky, A., 2002. Climate and surface properties: hydrological responseof small arid and semi-arid watersheds. Geomorphology 42, 43e57.

Yair, A., 1983. Hillslope hydrology, water harvesting and areal distribution of someancient agricultural systems in the Northern Negev desert. Journal of AridEnvironments 6, 163e184.

Yair, A., 1987. Environmental effects of loess penetration into the northern Negevdesert. Journal of Arid Environments 13, 9e14.

Zilberman, E., 1992. The late Pleistocene sequence of the northwestern Negev floodplains e a key to reconstructing the paleoclimate of southern Israel in the lastglacial. Israel Journal of Earth Sciences 41, 155e167.