Author's personal copy - BGUs personal copy ... because of climate change or anthropogenic arti cial...

Author's personal copy

Effects of prolonged drought on the vegetation cover of sanddunes in the NW Negev Desert: Field survey, remote sensingand conceptual modeling

Z. Siegal a,b, H. Tsoar a,⇑, A. Karnieli b

a Department of Geography and Environmental Development, Ben-Gurion University of the Negev, P.O. Box 653, Beer-Sheva 84105, Israelb The Remote Sensing Laboratory, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Sede-Boker Campus 84990, Israel

a r t i c l e i n f o

Article history:Available online 13 March 2013

Keywords:Sand dunesVegetated linear dunesMobilityStabilityBio-crustDrought

a b s t r a c t

Luminescence dating of stable sand dunes in the large deserts of the world has shown several episodes ofmobility during the last 30 k years. The logical explanation for the mobility of fixed dunes is severedrought. Though drought length can be estimated, the level of precipitation drop is unknown. The stabi-lized sand dunes of the northwestern Negev Desert, Israel have been under an unprecedented prolongeddrought since 1995. This has resulted in a vast decrease of shrubs cover on the fixed sand dunes, whichchanges along the rainfall gradient. In the north, an average of 27% of the shrubs had wilted by 2009, andin the drier southern area, 68% of the shrubs had withered. This loss of shrubbery is not expected toinduce dune remobilization because the existing bio-crust cover is not negatively affected by the drought.Eleven aerial photographs taken over the drier southern area from 1956 to 2005 show the change inshrub cover due to human impact and the recent severe drought.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

Huge, stabilized sand dune fields are a widely recognized geo-logical feature found throughout the world. Just two highly inves-tigated examples are the Kalahari Desert in southern Africa andmost parts of the Australian Deserts (Tsoar, 2013). It is clear thatevery stabilized dune field was once active and only became fixedbecause of climate change or anthropogenic artificial fixation (Ro-skin et al., 2011b). However, the processes and conditions thatcause stabilized dune fields to become reactivated are not so clear.There are documented cases of sand dune mobility engendered bystrong human impact on desert vegetation (Meir and Tsoar, 1996;Seifan, 2009). However, the cessation of human interference doesnot necessarily lead to the recuperation of the vegetation and re-fixation of the dunes, as continued dune activity has been observedin certain such cases (Arens and Geelen, 2006; Arens et al., 2004;Tsoar, 2005, 2008; Tsoar et al., 1995). Most researchers see thetransition from stabilized vegetated dunes into active ones asresulting from prolonged drought (Forman et al., 2001; Hugenholtzand Wolfe, 2005; Muhs and Holliday, 1995; Muhs et al., 1997a,b;Sridhar et al., 2006). Though droughts are considered to be a natu-ral ecosystem phenomenon, it was found that a decline in plantpopulation or even an increase in mortality occurs after a long

drought (Bowers, 2005a,b; Hereford et al., 2006; Mcauliffe andHamerlynck, 2010).

Since droughts come in various levels of severity, it is necessaryto define these extreme weather conditions. An extended period ofdeficient precipitation that lasts less than 5 years but is longer thana few months is referred to as drought. If this dry period lasts formore than a decade, it is called ‘‘increased aridity’’ (Hugenholtzand Wolfe, 2005), ‘‘megadrought’’ or ‘‘high magnitude drought’’(Hanson et al., 2009; Schmeisser et al., 2009; Woodhouse andOverpeck, 1998). These terms describe the longest and most severedroughts, such as the 16th century drought in the central UnitedStates (Woodhouse and Overpeck, 1998). Although the literatureis full of many conceptual models explaining the effects of vegeta-tion reduction on the activation of stabilized dunes, none of themdeal in a quantitative manner with how drought intensity andduration affect the levels of vegetation cover.

The introduction of luminescence dating methods providesresearchers with a pivotal tool to identify historic periods of sanddune activity. Since this technology reveals the time passed sincesand has been exposed to the sun, it enables, with some exceptions(Bateman et al., 2003), the determination of the end of mobilityepisodes and the initiation of stabilization (Forman et al., 2001;Mason et al., 2004; Miao et al., 2007; Sridhar et al., 2006). Locatingthese episodes of dune activity has enabled researchers to identifyperiods of prolonged drought or aridity (Lu et al., 2005; Muhs andZarate, 2001).

1875-9637/$ - see front matter � 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.aeolia.2013.02.002

⇑ Corresponding author. Fax: +972 8 6472821.E-mail address: [email protected] (H. Tsoar).

Aeolian Research 9 (2013) 161–173

Contents lists available at SciVerse ScienceDirect

Aeolian Research

journal homepage: www.elsevier .com/locate /aeol ia


The traumatic drought of the 1930s, known to be the driest per-iod of modern times in North America, did not activate the stabi-lized Nebraska Sand Hills. This resulted from the fact that despitethe low precipitation received by the Nebraska Sand Hills in the1930s, enough plants were able to persist during the droughtand long-term vegetation abundance was not altered (Manganet al., 2004). Modeling results indicate that even if the 1930sdrought would have continued for several decades, the dune vege-tation would still have persisted and protected the sand fromremobilization (Mangan et al., 2004). The 1930s drought is consid-ered to have been less severe and shorter than the thirteenth andsixteenth century megadroughts in this region (Stahle et al.,2007; Woodhouse and Overpeck, 1998). A tree-ring study in wes-tern Nebraska indicated that 22 droughts had occurred over a per-iod of 800 years, each having a longer duration than that of the1930s (Weakly, 1962).

The term megadrought, namely a prolonged, sharp decline inprecipitation, is commonly used to explain the historic activationof fixed dunes. However, over the last 100 years, such a drought–triggered remobilization of sand dunes has not been reportedand scientifically described. Hence, we have no quantitative datadescribing the intensity and length of drought conditions requiredto remobilize fixed dunes. All papers dealing with increased aeo-lian activities and the movement of previously stabilized dunesdeal with decreased precipitation in qualitative terms alone. Noneof them discuss quantitatively the minimal amount of rain thatwill convert fixed dunes into active ones (Forman et al., 2005; Han-son et al., 2009, 2010; Hesse and Simpson, 2006; Muhs and Holli-day, 1995; Sridhar et al., 2006).

It is crystal clear that there is a zero-growth limit for sand dunevegetation when the average amount of rainfall, in extremely aridzones, is insignificant (Thornthwaite, 1933). As rainfall increasesabove this limit, the density of perennial vegetation cover or theaboveground net primary productivity (NPP) starts to increase,either linearly or asymptotically (Eamus, 2003; Lauenroth, 1979;Noy-Meir, 1973; Sala et al., 1988; Wenjuan et al., 2005). The NPPfor different types of vegetation and soil tends to zero at differentcritical amounts of rainfall. According to the asymptotic curves ofthe NPP in the Lushi Basin of central China, grass tends to zero-yield at an average annual precipitation of 150 mm (Wenjuanet al., 2005), while for the North American grassland, it is around56–58 mm (Lauenroth, 1979; Sala et al., 1988). Satellite measure-ments taken along the rainfall gradient from the Sahara (rainfall0–100 mm) through the Saharan–Sahelian transition zone (100–200 mm) and to the Sahel region (200–400 mm) show that theprecipitation limit for the space-observed normalized differentvegetation index (NDVI) is below 150–160 mm (Tucker and Nich-olson, 1999). However, the regression line (Fig. 1c in Tucker andNicholson, 1999) reaches zero NDVI when the average rainfall(for 1980–1994) is 63 mm per year. With respect to vegetationon sand dunes, the amount of rainfall for zero NPP—known as‘‘ineffective precipitation’’—is not clearly known. In this paper,we describe the effect of prolonged drought on vegetationcover in the northern Negev Desert since 1996. The data acquiredin this study enable us to estimate the value of ineffectiveprecipitation for zero productivity of shrubs for sand dunes in thisregion.

The interdependence between rainfall and NPP also depends onother factors, including wind power, type of vegetation cover(shrubs vs. grass), soil quality and temperature. For example, thegrass cover on sand dunes is greater than that of shrubs. It wasfound that the threshold of dune activity in an area of grass-cov-ered sand dunes in south Texas was exceeded by two successivedrought years having a 30–50% reduction in precipitation (Formanet al., 2009). Drought and livestock grazing can activate stabilizedsand (Tsoar et al., 1995) or bring about the expansion of shrubs at

the expense of grass (Forman et al., 2001; Yanoff and Muldavin,2008).

1.1. Relative productivity of dune sand in deserts

There is a non-linear relationship between climate (wind andprecipitation) and sand dune vegetation cover (Werner et al.,2011; Yizhaq et al., 2009). Even in areas where annual average pre-cipitation should yield plant growth, the sand can remain bare ifhigh levels of wind power cause severe erosion, followed by sandtransport that prevents plant germination (Okin et al., 2001; Tsoar,2005; Yizhaq et al., 2007; Okin, 2008). But when wind power islow, dune sand is a relatively more favorable ecological habitatin the desert than adjacent non-sandy soils, such as loamy soils,despite their better water holding capacity (Fet et al., 1998; Noy-Meir, 1973; Sala et al., 1988; Tsoar, 1997, 2013).

Due to its low moisture tension, sand has the highest rate ofhydraulic conductivity, while clay has the lowest. For dune sandin humid areas with abundant precipitation, the main cause ofsoil-moisture loss in sand is the high rate of infiltration belowthe plant root layer. In arid and semi-arid lands, which have littleprecipitation and a high rate of evaporation, there is negligibleinfiltration. Unlike clayey soils, sandy soils offer the feature of deepinfiltration whereby moisture is protected from evaporation duringthe long dry periods (Tsoar, 1990). In the sand dunes of the NegevDesert, rainy year precipitation infiltrates to depths in excess of100 cm (Yair et al., 1997). The breakpoint between the advantageof coarse-textured sand (low evaporation) over fine-textured soil,composed of silt and clay, and its inherent disadvantages (highhydraulic conductivity and low field capacity) lies somewhere be-tween a mean annual precipitation of 300 and 500 mm in the sanddunes of the arid/semi-arid areas of Israel (Noy-Meir, 1973) or370 mm precipitation in the central grassland of the US (Salaet al., 1988) and is known as the ‘‘inverse texture effect’’.

Our conceptual model of the effect of drought on vegetationcover is shown in Fig. 1. The model is based on the relationshipbetween mean annual precipitation and the NPP discussed above.

Fig. 1. Conceptual model showing the response of vegetative cover (located in thedark gray area of the curve) to precipitation. The regressive character of the curveassumes that vegetation cover increases asymptotically as the precipitation rises. H,Z and M indicate locations along the rainfall gradient of the Negev Desert sanddunes (see Fig. 2). Following a long period of low precipitation (drought), the shrubcover is reduced due to plant adaptation to lower rainfall, and the curve moves tothe right. In this way, the decrease in the vegetation cover at each of the locationscan be discerned, as expressed by the wilting shrubs located in the light gray area.

162 Z. Siegal et al. / Aeolian Research 9 (2013) 161–173


Instead of NPP, we refer to the perennial vegetation cover, or themaximal population size of all plant species that can be sustainedunder a certain amount of rainfall (‘‘the carrying capacity’’). Thevegetation cover density is zero when the average precipitationis ineffective for plant growth in this soil. As the precipitation in-creases above this zero density level, vegetative production beginsincreasing as well. We assume that instead of a linear relationship,productivity on dune sand is characterized by an asymptotic curve,namely, that as precipitation continues to increase, the productiv-ity approaches a stage of no change or negligible increase. Thisstate is known as the maximal carrying capacity (Fernandezet al., 1991) (Fig. 1). This assumption is based on the above-dis-cussed inverse texture effect, which describes a situation in which,as the precipitation level passes a particular level (the precipitationbreakpoint), more additional rain water is lost by deep infiltration,beyond the root layer, and has no effect on increasing vegetationcover.

A prolonged drought will push the curve to the right, and in theinterval between the two curves, we get wilted shrubs that die be-cause of adaptation of the vegetation cover to lower rainfall (thelight gray area in Fig. 1). Here we aim to achieve a quantitative veg-etation cover/rainfall curve by measuring the amount of deadshrubs in the northwestern Negev dune field that was under a pro-longed drought, causing extensive wilting of shrubs along the rain-fall gradient.

1.2. Research area

The northwestern Negev Desert is a continuation of the north-ern Sinai Desert dune field, which covers an area of about13,000 km2, with the Negev Desert accounting for 10% of this ex-panse (Fig. 2). Northern Sinai and the northwestern Negev are geo-graphically inseparable regions; however, the differences in landuse make them very different. The Sinai sand dunes are under con-ditions of grazing, shrub gathering, trampling by sheep and goatflocks and other shrub destructive activities; the Negev dunes facevery little human impact (Tsoar, 2008; Tsoar and Moller, 1986).

Because of these differences in land use, there are strong albedodifferences between the Negev and the Sinai dunes (Karnieli andTsoar, 1995). Due to the sharp restrictions of human land use overthe last 50 years, vegetation recovery on the Negev side is rela-tively young and known, with some fluctuations (Meir and Tsoar,1996; Seifan, 2009).

The sand dunes of the Negev are mostly vegetated linear dunes,known as VLDs, that elongate rather than migrate (Roskin et al.,2011b; Tsoar, 2013; Tsoar et al., 2008). They are covered by peren-nial shrubs, annual plants and biogenic crusts, which are composedmostly of cyanobacteria (Karnieli et al., 1999; Verrecchia et al.,1995). VLDs with biogenic crusts are also known in other deserts,including those in Australia and the Kalahari (Hesse and Simpson,2006; Thomas and Dougill, 2007). Most of the perennial plants areshrubs, while the grasses are few (mostly on the dune crests). Theentire sand dune area of Sinai and the Negev had been under hu-man impact (grazing and shrubs gathering) for many years (Tsoar,1995). It is known from the Chihuahuan Desert in New Mexico thatstrong grazing leads to degradation and the invasion of shrubs atthe expense of grasses (Schlesinger et al., 1990).

The Negev sand dunes we have studied are located along therainfall gradient from the arid zone in the north (140 mm on an-nual average) to the extremely arid zone in the south (less than100 mm) (Fig. 2). Here, the rainy season is between Novemberand April, with no rain in summer. Winters are cool and summersvery hot. There is only one weather station in our research area.The average annual rainfall isohyet maps (Fig. 2) were compiledfrom the data of seven weather stations surrounding the field areafrom the north, east and south, following the method described inDunne and Leopold (1978). The wind power measured in Sde Hal-lamish was very low, which explains the dune-stabilizing effects ofthe shrubs and bio-crusts (Tsoar et al., 2008).

The episodes of sand dune encroachment into the NW Negevwere determined by optical stimulated luminescence (OSL) dating.Initial evidence for dune encroachment was dated to �24–18 ka,followed by a major post Last Glacial Maximum (LGM) encroach-ment at �18–11.5 ka. This latter encroachment resulted from a

Fig. 2. Map coordinates of the Negev Desert study area and its position in the overall southeastern Mediterranean region. The sand dunes are in gray. H, ZH, Z, MZ, and M arethe locations along the rainfall gradient (shown by the isohyets) where vegetation cover and dieback shrubs were surveyed. The isohyets are the averages for 1986–2006.

Z. Siegal et al. / Aeolian Research 9 (2013) 161–173 163


global post-LGM windy climate that is synchronous with the Hein-rich 1 and Younger Dryas cold-events. After 11.5 ka, the windpower was reduced, and the dunes reached their current spatialextent and became stabilized. The dunefield was generally stableduring the Holocene, though significant late Holocene dune mobi-lization is dated for �2–0.8 ka (Roskin et al., 2011a).

1.3. The recent prolonged Negev drought

The Negev Desert of Israel has been under continuous droughtsince the mid-1990s. Twenty-one years of rainfall data taken atthe Sde Hallamish weather station, located at the southern partof the dune field (Fig. 2), show that for this entire period, annualprecipitation averaged 83 mm. For the first seven relatively wetyears, the annual average was 114 mm, while for the remaining14 dry years, the average was 61 mm. This dry period started inwinter 1995–1996 and has exhibited a rainfall range of 67.6 mm,from 29.4 mm in winter 1998–1999 to 97 mm in winter 2006–2007, with a standard deviation of 22.5 (Fig. 3). The shrub cover

has been reduced since the late 1990s, a phenomenon that is asso-ciated with the continuous wilting of shrubs.

Drought strength can be described using three attributes:drought length (the number of consecutive years of negative raindeviations from the median), drought magnitude (an average ofthe negative deviations from the median of all successive dryyears) and drought severity (the product of drought length anddrought magnitude) (Petheram et al., 2008). As indicated above,the high hydraulic conductivity of dune sand brings rainwater todepths where it is stored and protected from evaporation. Accord-ingly, vegetation cover does not decrease instantly in a singledrought year, but reflects moisture availability over several years(Bullard et al., 1997). We, therefore, assume that a drought of 1or 2 years, after several rainy years, will not cause widespreadshrub wilting. Protracted droughts, however, are known to inducethe dieback of plants (Allen and Breshears, 1998; Breshears et al.,2005; Mueller et al., 2005; von Hardenberg et al., 2001). In theCanadian Prairies, for example, a drought of less than three dec-ades was observed to cause large-scale dune mobilization (Wolfe,

Fig. 3. Twenty-one years of rainfall distribution in Sde Hallamish, from winter 1988/1989–2008/2009.

Fig. 4. Rainfall maps of the NW Negev sand dunes. (A) Average annual rainfall isohyets for the wet years (1986–1995). (B) Average annual rainfall isohyets for the dry years(1996–2006).



1997). The prolonged drought in the Negev Desert started in thewinter of 1995/1996 with a severe drought, followed by two suc-cessive less-severe years (Fig. 3). Measurements of mortality andrecruitment of the different species in each community was carriedout in Sde Hallamish during the second and third year of thedrought (1996–1998). The study showed that, in general, mortalityincreased in the third year (Malkinson and Kadmon, 2007). Thelowest amount of rainfall on record in the Negev Desert was inthe fourth year of drought, followed by a very dry fifth year(Fig. 3). A high rate of shrub mortality was observed after the fifthyear of drought.

Fig. 4 shows two rainfall isohyets of the NW Negev, one for thewet years of 1986–1995 and the other for the dry years, 1996–2006. The drop in rainfall for 1996–2006 was relatively higher inthe drier sites of the dune field than in the less dry sites (Table 1).The drop in rainfall between the wet and dry years (Fig. 3; Table 1)led to a reduction in the shrub cover, as manifested by selectedshrub wilting. In Sde Hallamish (site H) in the south, the annualaverage rainfall dropped to values that could barely support peren-nial vegetation.

2. Materials and methods

A preliminary survey of wilting perennial plants conducted in2006 along the rainfall gradient at three sites showed a remarkabledieback, increasing from north to south. Consequently, a survey ofthe shrub cover was done during the spring (May–June) of 2007–2009 at five sites along the rainfall gradient from north to south,marked as M, MZ, Z, HZ, and H (Fig. 2). Four different samplingsof the shrub cover were carried out at each site, each of which com-prised an area of 5 � 20 m along the west-to-east alignment of thelinear dunes. The four samples covered the following natural hab-itats: 1. the south-facing lower slope, 2. the crest, 3. the north-fac-ing lower slope, and 4. the interdune (Fig. 5). Four repetitions weremade at each habitat of the five sites. A total of 80 surveys were

done each year, in 2007, in 2008 and in 2009. The lower north-fac-ing slope is the richest in shrub cover and is also covered by a crust,usually composed of cyanobacteria and sometimes of mosses (Yair,2001).

In the field, wilting was observed in all habitats at all sites alongthe rainfall gradient and in all perennial species. Visual assessmentof wilting was done by dividing the perennial plants into five cat-egories (Table 2). Only categories 4 and 5 were classified as deadplants.

The vegetation cover for each sampling area was calculatedaccording to the total area of the plants (in m2) divided by100 m2 (the sampling area). We measured the total (live and dead)shrub area, the total live shrub area, and the percentage of the deadshrub area in the total shrub area. Each habitat (Fig. 5) has a differ-ent average width. The weighted average of the shrub cover foreach habitat at each site was calculated by taking the average per-centage of the width of each habitat that was measured five timesat each site through aerial photography (Table 3).

Changes in the shrub cover of the Sde Hallamish dune field wereobtained by processing a series of 11 aerial photographs over a 50-year period (1956–2005; Table 4) applying GIS methodology. Using

Table 1The deficiency in rainfall for all the sites for 1986–1995 and 1996–2006. The locationof the sites is shown in Fig. 2.

Site H ZH Z MZ M

Average rainfall (in mm) for 1986–1995 112 118 134 145 155Average rainfall (in mm) for 1996–2006 69 75 90 98 115Deficiency in rainfall (%) 38 36 33 32 26

Fig. 5. Vegetated linear dunes at Sde Hallamish (site H) showing the four habitats sampled in the survey.

Table 2Categories used in the visual assessments of wilting.

Category Wiltingstage

Visual description

1 Normal No signs of wilting, proliferous plant growth2 Slightly

wiltedBeginnings of dieback

3 Wilted All leaves dead, but the stem green4 Dead Plant structure retained, but all above-ground

parts dead5 Dead Plant stems lying on the ground and starting to

disintegrate

Table 3The average width of each habitat and its relative size.

Habitat Average width (m) Relative size (%)

Crest 22 13South-facing slope 15 9North-facing slope 25 14Interdune 113 64Average width of the linear dunes 175 100



ERDAS IMAGINE software, all photographs were rectified to amaster orthophoto from 2003 using 15–20 ground control points(GCPs) in and around the dune field and a first or second polyno-

mial model. This process resulted in an average total root meansquare error (TRMSE) of 0.52, i.e., the size of about half a pixel (Ta-ble 4). Four sampling units, 400�400 m each, covering the fourdune habitats, were selected in the same location in all 11 photo-graphs. Shrub cover was obtained by the maximum likelihood clas-sifier algorithm. Note that with this method, there is no way todistinguish between the dead and live shrubs.

3. Results and discussion

3.1. Change in the perennial plant cover along the rain gradient

Fig. 6 presents the percentage of total, live, and dead shrubcover in the five experimental up-the-rainfall-gradient sites for2007–2009. The percentage of total, live and dead shrub cover

Table 4Data of aerial photographs used for monitoring temporal changes in vegetation cover.

Date Resolution (m) Scale TRMSEa (m)

24 July 1956 0.72 1:30,000 0.5226 July 1968 0.58 1:26,000 0.5112 June 1971 0.74 1:28,500 0.5324 April 1976 0.47 1:15,724 0.5017 April 1982 0.51 1:27,000 0.5720 July 1984 0.92 1:27,800 0.4704 February 1989 0.42 1:13,727 0.5030 April 1994 0.68 1:36,400 0.6004 December 1999 0.30 1:13,000 0.442003 1.00 – Base12 August 2005 0.63 1:23,100 0.57

a Total root mean square error.

68 80 91 106 123

2007 2008 2009

68 80 91 106 123

2007 2008 2009

68 80 91 106 123

2007 2008 2009

B

A

C

Fig. 6. (A) The total shrubs (live + dead) cover (in percent) surveyed in the sitesalong the rainfall gradient for 2007–2009. (B) The same for the live shrub cover (inpercent). (C) The same for the dead shrub cover as a percentage of the total shrubcover.

Fig. 7. The total shrub cover (in percent) in the four habitats surveyed along therainfall gradient.



for the four habitats of all sites along the rainfall gradient for thesame years are presented in Figs. 7–9. Table 5 details differencesin the compositions of the dominant perennial plants in the totaland live shrub cover of the several sites, habitats, and years.

Site M (the wettest northernmost site) is characterized byhomogeneity in the live shrub cover (Table 5). The dominant plantsof this site are species that are adjusted to stabilized sand, such asPanicum turgidum, Echiochilon fruticosum and Retama raetam,which are primary in the different habitats sampled. Psammophilicspecies that are adapted to sand movement, such as Artemisiamonosperma and Echinops philistaeus (Danin, 1996), sprout at theactive crests. There is not much change in shrub species cover aftermore than 30% of the shrubs have wilted.

Some species that are adapted to some sand movement, such asA. monosperma, and E. philistaeus (Danin, 1996), exist at site MZ(Table 5), together with shrubs that have adapted to stabilizedsand, such as R. raetam, P. turgidum and Gymnocarpos decander.Up to 60% of the shrubs at this site have wilted.

The total shrub cover at site Z is between 9% and 16% (Table 5).Here, the shrubs at the crests are adapted to shifting sand and in-clude A. monosperma, Moltkiopsis ciliate and Stipagrostis scoparia.On the south- and north-facing slopes and in the interdune, theshrub species are adapted to stabilized or partly stabilized sand(such as E. philistaeus, G. decander and R. raetam). Up to 63% ofthe shrubs at this site have wilted.

The total shrub cover at site HZ is between 8% and 16% (Ta-ble 5). The shrubs at the crest are adapted to shifting sand (A.monosperma, S. scoparia), and they did not change there afterthe long drought, except for the dieback of Noaea mucronata.More than 90% of the shrubs at the north-facing slope havewilted. After the long drought, there has been a great change inthe shrub species at this habitat in which the dominant shrubs(N. mucronata and A. monosperma) have wilted, and only shrubsadapted to shifting sand, such as Lycium shawii, have survived.G. decander is the dominant shrub at the interdune after the die-back of N. mucronata, A. monosperma and E. philistaeus during thedrought.

Fig. 8. The live shrub cover (in percent) in the four habitats surveyed along therainfall gradient.

Fig. 9. The percentages of dead shrub cover of the total shrub cover in the fourhabitats surveyed along the rainfall gradient.



The shrub cover at the driest area (site H) is the lowest, espe-cially with respect to live cover (Table 5). This site also has thehighest rate of wilted shrubs, approaching 95% at the north-facingslopes, while the percentage of the dieback shrubs is much lower atthe crest (13.5%). The shrubs that have survived at the activecrest after the long drought are psammophilic species, includingHeliotropium digynum, Cyperus macrorrhizus, A. monosperma andS. scoparia. Only species that are tolerant to drought, such asAnabasis articulate, have survived in the other three habitats(Table 5).

The shrub cover (total, live and dead) at the different sites alongthe rainfall gradient is shown in Fig. 6. There is a considerable de-crease in the live shrub cover along the rain gradient, from north tosouth (Fig. 6B). Similarly, the dead shrub cover increases signifi-cantly from north to south (Fig. 6C). These results fit our concep-tual model (Fig. 1) in which the decrease in live shrubs, alongwith the increase in dead plants, becomes greater when movingto the drier part of a dune field.

The shrub cover (total, live and dead) of the four habitats in thefive sites, under study for 3 years, is shown in Figs. 7–9, respec-tively. The effect of drought is discerned in the percentage oflive shrub cover (Fig. 8) and of dead shrub cover (as a percentof the total cover) (Fig. 9). The crests of the linear sand dunes re-acted differently to the long drought than did the other threehabitats.

3.2. Statistical analysis

The results of the three-year survey (2007–2009) include thetotal (live and dead) perennial shrub cover (in percent), the liveperennial shrub cover (in percent) and the percentage of the dry(dead) perennial shrubs in the total shrub cover. We carried outa significance test (2-way analysis of variance, ANOVA) for the dif-ferences between the various five sites and the four habitats for thesurvey years at a level of significance <0.05 (Table 6). A post hocLeast Significant Difference (LSD) test (Sokal and Rohlf, 1995)was conducted for all the cases where the ANOVA provided a sig-nificant result. Table 6 shows the results of the ANOVA and thepost hoc LSD test for the sites and habitats, with the p values givenin parentheses. The interaction between the sites and the habitatsindicates how changes in site location affect the differences be-tween the various habitats at each site along the rainfall gradient(Rohlf et al., 1990). The results indicate that the effect of site onhabitat behavior increases from north to south, down the rain gra-dient. In other words, the differences in shrub cover in the differenthabitats, with respect to the live and dead shrubs, increase as wemove southward (Figs 8 and 9), while there are no such differencesin the total shrub cover (Table 6). Therefore, as we move north, upthe rain gradient, the total shrub cover of the four habitats tends tobe identical (Fig. 7). The results, backed-up by the statisticalanalysis, show that there are small differences in the vegetation

Table 5Three year (2007–2009) average of total shrub (live + dead) cover, live shrub cover and percentage of dead shrub cover from total shrub cover. Dominant plant species in the fourhabitats of the five sites are also given.

Site Habitat The dominant species of the total shrubscover

The dominant species of the live shrubs Total shrubscover (%)

Live shrubscover (%)

Dead (dry) shrubs(percent of total)

M Crest Panicum turgidum; Retama raetam Panicum turgidum; Retama raetam 10.2 7.5 27.2South-facing slope

Panicum turgidum; Retama raetam Panicum turgidum; Retama raetam 13.1 8.8 31.3

North-facing slope

Echiochilon fruticosum; Convolvulus lanatus;Noaea mucronata

Echiochilon fruticosum; Convolvuluslanatus; Noaea mucronata

12.2 8.7 29.6

Interdune Panicum turgidum; Echiochilon fruticosum;Artemisia monosperma

Panicum turgidum; Echiochilonfruticosum

13.8 11.1 18.8

MZ Crest Artemisia monosperma; Moltkiopsis ciliata;Echinops philistaeus

Echinops philistaeus; Moltkiopsis ciliata 12.8 9.8 22.7

South-facing slope

Echinops philistaeus; Artemisia monosperma;Retama raetam

Echinops philistaeus; Retama raetam 11.7 5.2 60.7

North-facing slope

Echinops philistaeus; Panicum turgidum Echinops philistaeus; Panicum turgidum 14.1 7.1 51.7

Interdune Artemisia monosperma; Convolvulus lanatus;Gymnocarpos decander

Convolvulus lanatus; Retama raetam;Gymnocarpos decander

13.6 7.4 44.6

Z Crest Artemisia monosperma; Moltkiopsis ciliata Moltkiopsis ciliata; Artemisiamonosperma

12.6 10.2 15.4

South-facing slope

Gymnocarpos decander; Artemisiamonosperma; Echinops philistaeus

Echinops philistaeus; Retama raetam;Gymnocarpos decander

8.7 3.4 57.2

North-facing slope

Artemisia monosperma; Lycium shawii;Noaea mucronata

Lycium shawii; Anabasis articulata 15.6 6.4 63.4

Interdune Echiochilon fruticosum; Gymnocarposdecander

Echiochilon fruticosum; Gymnocarposdecander

14.6 10.1 41.1

HZ Crest Artemisia monosperma; Echinops philistaeus Artemisia monosperma; Echinopsphilistaeus

15.5 13.6 15.3

South-facing slope

Echinops philistaeus; Noaea mucronata;Gymnocarpos decander

Echinops philistaeus; Lycium shawii;Retama raetam

9.3 4.4 60.0

North-facing slope

Noaea mucronata; Artemisia monosperma Lycium shawii; Echinops philistaeus 11.3 1.2 91.2

Interdune Gymnocarpos decander; Noaea mucronata Gymnocarpos decander; Echiochilonfruticosum;

8.3 3.2 66.1

H Crest Heliotropium digynum; Cyperusmacrorrhizus

Heliotropium digynum; Cyperusmacrorrhizus

9.2 8.0 13.5

South-facing slope

Anabasis articulate; Artemisia monosperma;Noaea mucronata

Anabasis articulate; Helianthemumstipulatum

5.2 1.8 74.2

North-facing slope

Anabasis articulate; Artemisia monosperma;Noaea mucronata

Retama raetam; Anabasis articulate;Noaea mucronata

12.7 0.6 95.3

Interdune Anabasis articulate; Noaea mucronata;Artemisia monosperma

Anabasis articulate 7.6 0.9 89.7



cover along the rainfall gradient when the rainfall is above 100 mm(Fig. 7). Below 100 mm, there is a significant decrease in the shrubcover, excluding the vegetation at the crest (Figs 8 and 9).

3.3. Changes in the vegetation cover of Sde Hallamish between 1956and 2005 (GIS results)

Temporal fluctuations in the vegetation cover of Sde Hallamishduring the 50 years between 1956 and 2005 are given in Fig. 10.The slight increase between 1956 and 1968 and the decrease be-tween 1968 and 1982 are the result of changes in human impact(Meir and Tsoar, 1996; Tsoar, 2008). The human impact, as men-tioned above, was grazing and shrub gathering by the Bedouinshepherds. This human impact changed dramatically during theperiod from 1956 to 2005. The increase in vegetation cover be-tween 1982 and the mid-1990s was the result of vegetation recov-ery after human impact was removed (Meir and Tsoar, 1996;Seifan, 2009). The vegetation reached its maximal, balanced coverbetween 1989 and 1994 before the beginning of the drought(Fig. 10). During the latter half of the 1990s, the vegetation coverdecreased (Seifan, 2009), a result of shrub dieback, though Seifan(2009) did not refer to this phenomenon. From Fig. 10, one canclearly see the effect of the drought on vegetation cover between1994 and 1999. Based on the rain data (Figs. 3 and 4), one wouldexpect that the dieback of shrubs and their disintegration wouldstart to be conspicuous between 1996 and 1999 and continue atthe same rate through the beginning of the 21st century. This isconfirmed by the curve presented in Fig. 10.

The correlations between the moving average of annual precip-itation and vegetation cover were calculated in Sde Hallamish forthe years 1988/1989–2004/2005. The best correlation coefficients

Table 6Statistical analysis of the dune shrub measurements (2007–2009).

Total shrubs cover 2007 2008 2009

Differences between the sites ANOVA Significant Significant Not significantPost-hocLSD

H–HZ (0.04)H–Z (0.01)H–MZ (0.02)H–M (0.0009)

H–MZ (0.02)Z–MZ (0.01)

Differences between the habitats ANOVA Not significant Not significant Not significantInteraction between the sites and the habitats No No No

Live shrubs cover

Differences between the sites ANOVA Significant Significant SignificantPost–hocLSD

HZ–M (0.009)H–Z (0.01)H–MZ (0.02)H–M (0.0001)

H–MZ (0.0007)H–M (0.0006)

H–M (0.03)H–Z (0.01)

Differences between the habitats ANOVA Significant Significant SignificantPost-hocLSD

A–B (0.002)A–C (0.008)

A–B (0.0005)A–C (0.0008)

A–B (0.005)A–C (0.001)

Interaction between the sites and the habitats No Yes Yes

Dry shrubs cover

Differences between the sites ANOVA Significant Significant SignificantPost-hocLSD

MZ–M (0.03)Z–M (0.01)HZ–M (0.0001)H–M (0.0001)H–MZ (0.02)

MZ–M (0.009)HZ–M (0.0003)H–M (0.0001)MZ–H (0.03)H–Z (0.0006)

H–M (0.0002)H–Z (0.04)HZ–M (0.008)

Differences between the habitats ANOVA Significant Significant SignificantPost-hocLSD

A–B (0.0001)A–C (0.0001)A–D (0.0001)

A–B (0.0001)A–C (0.0001)A–D (0.0001)C–D (0.03)

A–B (0.0001)A–C (0.0001)A–D (0.0001)

Interaction between the sites and the habitats Yes Yes Yes

The five sites, going from north to south: M, MZ, Z, HZ, H. The four habitats are marked as A, crest; B, south-facing slope; C, north-facing slope; D, interdune. The parenthesesindicate the values of p.

Fig. 10. Changes in shrub cover at Sde Hallamish (site H) as measured from aerialphotographs from 1956 to 2005. Four sampling units (S.U.) were measured eachyear. The solid line reflects the average, and the dashed lines show ±1 standarddeviation (SD).

Table 7The correlation coefficients and probabilities for the movingaverages of rainfall and shrubs cover measured in Sde Hallamishbetween 1988/1989 and 2004/2005.

Moving average of the rainfall (years) R2 p-Value

5 0.33 0.36 0.62 0.17 0.84 0.028 0.83 0.039 0.88 0.01

10 0.79 0.0411 0.67 0.08



(R2) and significant probability (p-value <0.05) are for 7–10 yearmoving averages (Table 7).

4. Discussion and conclusions

4.1. Changes in shrub cover along the rainfall gradient (Fig. 6)

Site H in the south (Sde Hallamish) is the most significantly dif-ferent from the other sites, while its crest is the most significantlydifferent from all other habitats (Table 6). The total shrub covershows that site H in 2007 had less shrub cover than the other sites(Fig. 6A). The same goes for the live shrub cover in 2007, and tosome extent, in 2008 and 2009 (Fig. 6 and Table 6). The most sig-nificant differences were for the percentage of the dead shrubs be-tween the sites, where site H had the highest proportion of deadshrubs, which decreased up the rainfall gradient (Fig. 6C).

The results show that despite the apparently similar total shrubcover (especially in 2008 and 2009), which reflects the wetter per-iod before the beginning of the drought (Fig. 4A), the severe, longdrought reduced shrub cover in proportion to the precipitation(Fig. 6B and C). The distribution of the live shrub cover (Fig. 6B)corroborates the conceptual model shown in Fig. 1. The very lowcover of live shrubs at site H (<2%) indicates that 68 mm of averagerainfall is very close to the rainfall level representing zero NPP(ineffective precipitation).

4.2. Changes in the habitat shrub cover along the rainfall gradient(Figs. 7–9)

Statistically, there are no differences between the four habitatswith respect to total vegetation cover (Table 6). The live shrub cov-er and the percentage of wilted shrubs show significant differencesbetween the sites and the four habitats (Table 6). According to ourconceptual model (Fig. 1), the reaction to prolonged drought ismore conspicuous in the driest areas of the dune field. This is alsoevident for the habitats, excluding the crest (Figs. 8 and 9). Thecrest has a lower species diversity and lower shrub cover when

the crest is active (Fig. 5) (Tielboerger, 1997). The crest habitat re-acts differently to prolonged drought. It had more live vegetationthan the other habitats, mainly at sites H and HZ, and less deadshrubs at sites H, HZ, Z, and MZ (Figs. 8 and 9). The reason for thispeculiar behavior of the crest shrubs arises from the harsh condi-tions; high rates of sand erosion and deposition favor psammophil-ic shrubs in this habitat. The strong winds, the lack of bio-crustsand the high sand erosion at the crests give rise to shrubs thatare resistant to this severe ecology. These psammophilic shrubsare also more resistant to long droughts than the shrubs in theother habitats (Table 5). The same applies, but to a lesser extent,to the south-facing slope at sites H and HZ (Figs. 8 and 9).

As mentioned above, the north-facing slope and the interduneare the habitats with the maximal shrub cover (Fig. 5). This resultsfrom the relatively lower solar radiation on the north-facing slopeduring the long dry summer and the lower aeolian erosion at theinterdune. On the other hand, the shrubs at the crest are under se-vere conditions of very high aeolian erosion (up to 100 times therate of erosion in the interdune areas), which is known to be thelimiting factor for vegetation survival (Allgaier, 2008). For this rea-son, the crest is sometimes devoid of bio-crust and is covered bypsammophilic shrubs (Table 5). The south-facing slope receivesmuch higher solar radiation during the long dry summer, and forthat reason, the total shrub cover there is lower, especially at thedrier sites (Fig. 7). The lower bio-crust cover at this slope bringsabout higher aeolian erosion and reduced shrub cover.

In the drier sites (H, HZ), the habitats that were primarily strick-en by the drought are the north-facing slopes and the interdunes(Figs. 8 and 9). These two habitats were relatively indulged by soilsthat were relatively moist and by decreased solar exposure (for thenorth-facing slope) and lower wind erosion (for the interdune).During the prolonged drought, the shrubs of these two habitatssuffered from a severe reduction in soil moisture and were greatlyimpacted (Fig. 9).

The bio-crust cover is more evident at the north-facing slopeand the interdune. The crest is partly or fully devoid of crust, andonly the lower part of the south-facing slope is so covered(Fig. 5). Because the crust absorbs large amounts of rain, rainwater

0

1

2

3

4

5

6

7

0

10

20

30

40

50

60

70

80

90

100

110

Year

Fig. 11. The rain moving averages of 9 years calculated for Sde Hallamish showing the continuous drop in rainfall since the mid-1990s.



infiltration is reduced, along with water availability for the peren-nial shrubs (Ram and Aaron, 2007). This phenomenon is observablein the driest areas of sites H and HZ (Figs. 8 and 9). In addition, itshould be noted that seed germination on a bio-crust is relativelypoor as compared to exposed sand (Prasse and Bornkamm, 2000,2008).

4.3. The effects of drought length

The Negev, as an arid land, has been suffering from droughtsfrom biblically recorded times to our own days (Evenari et al.,1982). Perennial vegetation does not respond immediately todrought or excesses of rain as do annuals. Hence, 2–3 successiveyears of drought will not generate significant dieback of perennialsin the Negev. There is a time lag between the beginning of a longdrought and its effect on the shrub cover (Hugenholtz and Wolfe,2005). For this reason, it is preferable to evaluate shrub coveragainst a moving average of the rainfall. Table 7 indicates that after6 years of prolonged drought, a considerable drop in shrub cover isobserved. We found that the nine-year moving average providedthe best correlation between annual rainfall and the shrub coverfor Sde Hallamish (Table 7). Fig. 11 shows the nine-year movingaverages of precipitation and the levels of shrub cover between1989 and 2005. It also documents that the present drought inthe NW Negev Desert is the longest since rain was first recordedin the region of site H.

4.4. Shrub cover vs. average annual rainfall in the Negev Desert

Following the long drought, there is a major difference in the to-tal shrubs (Fig. 7) and live shrubs (Fig. 8). From Fig. 10 and previousstudies in the area (Meir and Tsoar, 1996; Schmidt and Karnieli,2002; Seifan, 2009), it is clear that the shrub cover increased con-siderably after 1982 thanks to the complete cessation of humanimpact (Tsoar, 2008), the lack of bio-crusts and the relatively highaverage rainfall (Fig. 4A). The shrub cover today is a result of thelower rainfall that prevailed in the area during the last 14 years.Taking the data of shrub cover vs. average rainfall during the wet-ter (1986–1995) and drought (1996–2009) periods, we observe alogarithmic change of shrub cover vs. precipitation (Fig. 12).Fig. 12 corroborates our conceptual model (Fig. 1). According toFig. 12, when the annual average rainfall is 52 mm, the shrub coverin the Negev Desert is zero. This value is close to the minimal pre-cipitation needed for vegetation growth on sand dunes that was gi-

ven by Danin (1996). In Sde Hallamish (site H), the average rainfallwas close to 60 mm, which should reflect a shift from dune stabil-ization to nearly full mobility. However, the bio-crust cover, whichis not affected by the continuous drought, prevents re-activation ofthe dunes. We can assume that if the average rainfall at H will con-tinue to be around 50–60 mm, and the bio-crust will be destroyedby trampling of animals and wind erosion, the area of Sde Halla-mish (Fig. 2) will become an active dune field. This would be thefirst example, known in the scientific literature, where such anoccurrence has been observed.

Acknowledgments

We gratefully acknowledge the help of Noam Lavin, Yossef Ash-kenazy, Itamar Giladi, Yaron Ziv, Avinoam Danin, and Avi Shmidafor their consistent advice and dedicated field activities.

References

Allen, C.D., Breshears, D.D., 1998. Drought-induced shift of a forest-woodlandecotone: rapid landscape response to climate variation. Proceedings of theNational Academy of Sciences of the United States of America 95, 14839–14842.

Allgaier, A., 2008. Aeolian sand transport and vegetation cover. In: Breckle, S.W.,Yair, A., Veste, M. (Eds.), Arid Dune Ecosystems. Springer, Berlin, pp. 211–224.

Arens, S.M., Geelen, L.H.W.T., 2006. Dune landscape rejuvenation by intendeddestabilisation in the Amsterdam water supply dunes. Journal of CoastalResearch 22, 1094–1107.

Arens, S.M., Slings, Q., Vries, C.N.d., 2004. Mobility of a remobilised parabolic dunein Kennemerland, the Netherlands. Geomorphology 59, 175–188.

Bateman, M.D., Thomas, D.S.G., Singhvi, A.K., 2003. Extending the aridity record ofthe Southwest Kalahari: current problems and future perspectives. QuaternaryInternational 111, 37–49.

Bowers, J.E., 2005a. Effects of drought on shrub survival and longevity in thenorthern Sonoran Desert. Journal of the Torrey Botanical Society 132, 421–431.

Bowers, J.E., 2005b. Influence of climatic variability on local population dynamics ofa Sonoran Desert platyopuntia. Journal of Arid Environments 61, 193–210.

Breshears, D.D., Cobb, N.S., Rich, P.M., Price, K.P., Allen, C.D., Balice, R.G., Romme,W.H., Kastens, J.H., Floyd, M.L., Belnap, J., Anderson, J.J., Myers, O.B., Meyer, C.W.,2005. Regional vegetation die-off in response to global-change-type drought.Proceedings of the National Academy of Sciences of the United States ofAmerica 102, 15144–15148.

Bullard, J.E., Thomas, D.S.G., Livingstone, I., Wiggs, G.F.S., 1997. Dunefield activityand interactions with climatic variability in the southwest Kalahari Desert.Earth Surface Processes and Landforms 22, 165–174.

Danin, A., 1996. Plants of Desert Dunes. Springer, Berlin.Dunne, T., Leopold, L.B., 1978. Water in Environmental Planning. W.H. Freeman &

Co., New York.Eamus, D., 2003. How does ecosystem water balance affect net primary productivity

of woody ecosystems? Functional Plant Biology 30, 187–205.Evenari, M., Shahan, L., Tadmor, N., 1982. The Negev: The Challenge of a Desert,

second ed. Harvard University Press, Cambridge, MA.

y = 10.751ln(x) - 42.556R² = 0.8511

0

2

4

6

8

10

12

14

0 20 40 60 80 100 120 140 160 180

Shru

bs c

over

(%)

Rainfall (mm)

Fig. 12. Change in shrub cover (in percent) with annual average rainfall as measured in the research area between 1986 and 2006.



Fernandez, R.J., Sala, O.E., Golluscio, R.A., 1991. Woody and herbaceousaboveground production of a Patagonian Steppe. Journal of RangeManagement 44, 434–437.

Fet, V., Polis, G.A., Sissom, W.D., 1998. Life in sandy deserts: the scorpion model.Journal of Arid Environments 39, 609–622.

Forman, S.L., Oglesby, R., Webb, R.S., 2001. Temporal and spatial patterns ofHolocene dune activity on the Great Plains of North America: megadroughtsand climate links. Global and Planetary Change 29, 1–29.

Forman, S.L., Marin, L., Pierson, J., Gomez, J., Miller, G.H., Webb, R.S., 2005. Aeoliansand depositional records from western Nebraska: landscape response todroughts in the past 1500 years. Holocene 15, 973–981.

Forman, S.L., Nordt, L., Gomez, J., Pierson, J., 2009. Late Holocene dune migration onthe south Texas sand sheet. Geomorphology 108, 159–170.

Hanson, P.R., Joeckel, R.M., Young, A.R., Horn, J., 2009. Late Holocene dune activity inthe Eastern Platte River Valley, Nebraska. Geomorphology 103, 555–561.

Hanson, P.R., Arbogast, A.F., Johnson, W.C., Joeckel, R.M., Young, A.R., 2010.Megadroughts and late Holocene dune activation at the eastern margin of theGreat Plains, north-central Kansas, USA. Aeolian Research 1, 101–110.

Hereford, R., Webb, R.H., Longpre, C.I., 2006. Precipitation history and ecosystemresponse to multidecadal precipitation variability in the Mojave Desert region,1893–2001. Journal of Arid Environments 67, 13–34.

Hesse, P.P., Simpson, R.L., 2006. Variable vegetation cover and episodicsand movement on longitudinal desert sand dunes. Geomorphology 81,276–291.

Hugenholtz, C.H., Wolfe, S.A., 2005. Biogeomorphic model of dunefieldactivation and stabilization on the northern Great Plains. Geomorphology70, 53–70.

Karnieli, A., Tsoar, H., 1995. Satellite spectral reflectance of biogenic crust developedon desert dune sand along the Israeli–Egypt border. International Journal ofRemote Sensing 16, 369–374.

Karnieli, A., Kidron, G.J., Glaesser, C., Ben-Dor, E., 1999. Spectral characteristics ofcyanobacteria soil crust in semiarid environment. Remote Sensing of theEnvironment 69, 67–75.

Lauenroth, W.K., 1979. Grassland primary production: North American grasslandsin perspective. In: French, N. (Ed.), Perspective in Grassland Ecology. Springer-Verlag, New York, pp. 3–24.

Lu, H.Y., Miao, X.D., Zhou, Y.L., Mason, J., Swinehart, J., Zhang, J.F., Zhou, L.P., Yi, S.W.,2005. Late Quaternary aeolian activity in the Mu Us and Otindag dune fields(north China) and lagged response to insolation forcing. Geophysical ResearchLetters 32. http://dx.doi.org/10.1029/2005GL024560.

Malkinson, D., Kadmon, R., 2007. Vegetation dynamics along a disturbancegradient: spatial and temporal perspectives. Journal of Arid Environments 69,127–143.

Mangan, J.M., Overpeck, J.T., Webb, R.S., Wessman, C., Goetz, A.F.H., 2004. Responseof Nebraska Sand Hills natural vegetation to drought, fire, grazing, and plantfunctional type shifts as simulated by the century model. Climatic Change 63,49–90.

Mason, J.A., Swinehart, J.B., Goble, R.J., Loope, D.B., 2004. Late-Holocene duneactivity linked to hydrological drought, Nebraska Sand Hills, USA. Holocene 14,209–217.

McAuliffe, J.R., Hamerlynck, E.P., 2010. Perennial plant mortality in the Sonoran andMojave deserts in response to severe, multi-year drought. Journal of AridEnvironments 74, 885–896.

Meir, A., Tsoar, H., 1996. International borders and range ecology: the case ofBedouin transborder grazing. Human Ecology 24, 39–64.

Miao, X.D., Mason, J.A., Swinehart, J.B., Loope, D.B., Hanson, P.R., Goble, R.J., Liu, X.D.,2007. A 10,000 year record of dune activity, dust storms, and severe drought inthe central Great Plains. Geology 35, 119–122.

Mueller, R.C., Scudder, C.M., Porter, M.E., Trotter, R.T., Gehring, C.A., Whitham, T.G.,2005. Differential tree mortality in response to severe drought: evidence forlong-term vegetation shifts. Journal of Ecology 93, 1085–1093.

Muhs, D.R., Holliday, V.T., 1995. Evidence of active dune sand on the Great-Plains inthe 19th-century from accounts of early explorers. Quaternary Research 43,198–208.

Muhs, D.R., Zarate, M., 2001. Late quaternary eolian records of the Americas andtheir paleoclimate significance. In: Markgraf, V. (Ed.), Interhemisphere ClimateLinkages. Academic Press, San Diego, pp. 183–216.

Muhs, D.R., Stafford, T.W., Been, J., Mahan, S.A., Burdett, J., Skipp, G., Rowland, Z.M.,1997a. Holocene eolian activity in the Minot dune field, North Dakota. CanadianJournal of Earth Sciences 34, 1442–1459.

Muhs, D.R., Stafford, T.W., Swinehart, J.B., Cowherd, S.D., Mahan, S.A., Bush,C.A., Madole, R.F., Maat, P.B., 1997b. Late Holocene eolian activity inthe mineralogically mature Nebraska sand hills. Quaternary Research 48,162–176.

Noy-Meir, I., 1973. Desert ecosystems: environment and producers. Annual Reviewof Ecology and Systematics 4, 25–51.

Okin, G.S., 2008. A new model of wind erosion in the presence of vegetation. Journalof Geophysical Research, Earth Surface 113, F02S10.

Okin, G.S., Murray, B., Schlesinger, W.H., 2001. Degradation of sandy arid shrublandenvironments: observations, process modeling, and management implications.Journal of Arid Environments 47, 123–144.

Petheram, C., McMahon, T.A., Peel, M.C., 2008. Flow characteristics of rivers in northernAustralia: implications for development. Journal of Hydrology 357, 93–111.

Prasse, R., Bornkamm, R., 2000. Effect of microbiotic soil surface crusts onemergence of vascular plants. Plant Ecology 150, 65–75.

Prasse, R., Bornkamm, R., 2008. Vascular plant response to microbiotic soil surfacecrusts. In: Breckle, S.W., Yair, A., Veste, M. (Eds.), Arid Dune Ecosystems.Springer, Berlin, pp. 337–351.

Ram, A., Aaron, Y., 2007. Negative and positive effects of topsoil biological crusts onwater availability along a rainfall gradient in a sandy and area. Catena 70, 437–442.

Rohlf, F.J., Chang, W.S., Sokal, R.R., Kim, J.Y., 1990. Accuracy of estimatedphylogenies – effects of tree topology and evolutionary model. Evolution 44,1671–1684.

Roskin, J., Porat, N., Tsoar, H., Blumberg, D.G., Zander, A.M., 2011a. Age, origin andclimatic controls on vegetated linear dunes in the northwestern Negev Desert(Israel). Quaternary Science Reviews 30, 1649–1674.

Roskin, J., Tsoar, H., Porat, N., Blumberg, D.G., 2011b. Palaeoclimate interpretationsof Late Pleistocene vegetated linear dune mobilization episodes: evidence fromthe northwestern Negev dunefield, Israel. Quaternary Science Reviews 30,3364–3380.

Sala, O.E., Parton, W.J., Joyce, L.A., Lauenroth, W.K., 1988. Primary production of thecentral grassland region of the United States. Ecology 69, 40–45.

Schlesinger, W.H., Reynolds, J.F., Cunningham, G.L., Huenneke, L.F., Jarrell, W.M.,Virginia, R.A., Whitford, W.G., 1990. Biological feedbacks in globaldesertification. Science 247, 1043–1048.

Schmeisser, R.L., Loope, D.B., Wedin, D.A., 2009. Clues to the medievaldestabilization of the Nebraska Sand Hills, USA, from ancient pocket gopherburrows. Palaios 24, 809–817.

Schmidt, H., Karnieli, A., 2002. Analysis of the temporal and spatial vegetationpatterns in a semi-arid environment observed by NOAA AVHRR imagery andspectral ground measurements. International Journal of Remote Sensing 23,3971–3990.

Seifan, M., 2009. Long-term effects of anthropogenic activities on semi-arid sanddunes. Journal of Arid Environments 73, 332–337.

Sokal, R.R., Rohlf, F.J., 1995. Biometry. W.H. Freeman and Company, New York.Sridhar, V., Loope, D.B., Swinehart, J.B., Mason, J.A., Oglesby, R.J., Rowe, C.M., 2006.

Large wind shift on the Great Plains during the Medieval Warm Period. Science313, 345–347.

Stahle, D.W., Fye, F.K., Cook, E.R., Griffin, R.D., 2007. Tree-ring reconstructedmegadroughts over North America since AD 1300. Climatic Change 83, 133–149.

Thomas, A.D., Dougill, A.J., 2007. Spatial and temporal distribution of cyanobacterialsoil crusts in the Kalahari: implications for soil surface properties.Geomorphology 85, 17–29.

Thornthwaite, C.W., 1933. The climates of the earth. Geographical Review 23, 433–440.

Tielboerger, K., 1997. The vegetation of linear desert dunes in the north-west Negev,Israel. Flora 192, 261–278.

Tsoar, H., 1990. The ecological background, deterioration and reclamation of desertdune sand. Agriculture, Ecosystems and Environment 33, 147–170.

Tsoar, H., 1995. Desertification in Northern Sinai in the eighteenth century. ClimaticChange 29, 429–438.

Tsoar, H., 1997. The physical properties of desert dune sand and its ecologicalimplications. Aarhus Geoscience 7, 79–90.

Tsoar, H., 2005. Sand dunes mobility and stability in relation to climate. Physica A357, 50–56.

Tsoar, H., 2008. Land use and its effect on the mobilization and stabilization of theNW Negev sand dunes. In: Breckle, S.W., Yair, A., Veste, M. (Eds.), Arid DuneEcosystems. Springer, Berlin, pp. 79–89.

Tsoar, H., 2013. Eolian environments: critical environments – sand dunes andclimate change. In: Shroder, J., Lancaster, N., Sherman, D.J., Baas, A.C.W. (Eds.),Treatise on Geomorphology, Aeolian Geomorphology, vol. 11. Academic Press,San Diego.

Tsoar, H., Moller, J.T., 1986. The role of vegetation in the formation of linear sanddunes. In: Nickling, W.G. (Ed.), Aeolian Geomorphology. Allen and Unwin,Boston, pp. 75–95.

Tsoar, H., Goldsmith, V., Schoenhaus, S., Clarke, K., Karnieli, A., 1995. Reverseddesertification on sand dunes along the Sinai/Negev border. In: Tchakerian, V.P.(Ed.), Desert Aeolian Processes. Chapman & Hall, London, pp. 251–267.

Tsoar, H., Blumberg, D.G., Wenkart, R., 2008. Formation and geomorphology of theNW Negev sand dunes. In: Breckle, S.W., Yair, A., Veste, M. (Eds.), Arid DuneEcosystems. Springer, Berlin, pp. 25–48.

Tucker, C.J., Nicholson, S.E., 1999. Variations in the size of the Sahara Desert from1980 to 1997. Ambio 28, 587–591.

Verrecchia, E., Yair, A., Kidron, G.J., Verrecchia, K., 1995. Physical properties of thepsammophile cryptogamic crust and their consequences to the water regime ofsandy soils, north-western Negev Desert, Israel. Journal of Arid Environments29, 427–437.

von Hardenberg, J., Meron, E., Shachak, M., Zarmi, Y., 2001. Diversity of vegetationpatterns and desertification. Physical Review Letters 87.

Weakly, H.E., 1962. History of drought in Nebraska. Journal of Soil and waterConservation 17, 271–274.

Wenjuan, D., Huimin, Q.Y.L., Dajie, Z., Duanhua, S., Liying, S., 2005. Modeling carbonand water budgets in the Lushi Basin with Biome-BGC. Chinese Journal ofPopulation, Resources and Environment 3, 27–34.

Werner, C.M., Mason, J.A., Hanson, P.R., 2011. Non-linear connections between duneactivity and climate in the High Plains, Kansas and Oklahoma, USA. QuaternaryResearch 75, 267–277.

Wolfe, S.A., 1997. Impact of increased aridity on sand dune activity in the CanadianPrairies. Journal of Arid Environments 36, 421–432.



Woodhouse, C.A., Overpeck, J.T., 1998. Two thousand years of drought variability inthe central United States. Bulletin of the American Meteorological Society 79,2693–2714.

Yair, A., 2001. Effects of biological soil crusts on water redistribution in the NegevDesert, Israel: a case study in longitudinal dunes. In: Belnap, J., Lange, O.L. (Eds.),Biological Soil Crusts: Structure, Function, and Management. Springer-Verlag,Berlin, pp. 303–314.

Yair, A., Lavee, H., Greitser, N., 1997. Spatial and temporal variability of waterpercolation and movement in a system of longitudinal dunes, western Negev,Israel. Hydrological Processes 11, 43–58.

Yanoff, S., Muldavin, E., 2008. Grassland–shrubland transformation and grazing: acentury-scale view of a northern Chihuahuan Desert grassland. Journal of AridEnvironments 72, 1594–1605.

Yizhaq, H., Ashkenazy, Y., Tsoar, H., 2007. Why do active and stabilized dunescoexist under the same climatic conditions? Physical Review Letters 98. Art. No.188001, May 4, 2007.

Yizhaq, H., Ashkenazy, Y., Tsoar, H., 2009. Sand dune dynamics and climate change:a modeling approach. Journal of Geophysical Research, Earth Surface 114,F01023. http://dx.doi.org/10.1029/2008JF001138.


Author's personal copy - BGUs personal copy ... because of climate change or anthropogenic arti cial...

Documents

Transcript of Author's personal copy - BGUs personal copy ... because of climate change or anthropogenic arti cial...