lable at ScienceDirect

Journal of Arid Environments 102 (2014) 27e33

Contents lists avai

Journal of Arid Environments

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

Spatial patterns provide support for the stress-gradient hypothesisover a range-wide aridity gradient

Jotham Ziffer-Berger a,*, Peter J. Weisberg b, Mary E. Cablk c, Yagil Osemd

aHerbarium of the Hebrew University, Giv’at Ram, Jerusalem, IsraelbDept. of Natural Resources and Environmental Science, University of Nevada, Reno, NV, USAcDivision of Earth and Ecosystem Sciences, Desert Research Institute, Reno, NV, USAd Institute of Plant Sciences, Agricultural Research Organization, Volcani Center, Bet Dagan, Israel

a r t i c l e i n f o

Article history:Received 16 May 2013Received in revised form13 November 2013Accepted 19 November 2013Available online 11 December 2013

Keywords:Artemisia tridentataDistribution rangeFacilitationGreat BasinMojave DesertPinus monophyllaStress gradient

* Corresponding author. Tel.: þ972 2 6584456; fax:E-mail addresses: jotham.z@gmail.com (J. Ziffer-B

edu (P.J. Weisberg), Mary.Cablk@dri.edu (M.E. Cab(Y. Osem).

0140-1963/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.jaridenv.2013.11.006

a b s t r a c t

We examined variations in the relative importance of facilitation vs. competition, in light of the Stress-Gradient Hypothesis (SGH) by assessing plant interactions along an aridity gradient over biogeographicscales. We surveyed the relationship between a shrub species (Artemisia tridentata) and pine seedlings(Pinus monophylla) across the Great Basin and the Mojave Desert, USA, encompassing the entire range ofP. monophylla. Using 69 sites we evaluated the spatial association between P. monophylla seedlings andA. tridentata shrubs, quantified with an electivity index, and implemented multiple regression analysis onthe effects of macro- and micro-environmental factors: precipitation, temperature, monsoonality index,topography, substrate and litter cover. We identified annual precipitation as a main factor, which wasnegatively related to shrub-seedling association. Additionally, shrub-seedling association was stronger inthe hot- than in the cold-desert, and was negatively related to litter cover. Effects of monsoonality,summer temperature, and bedrock type were not significant. We also considered nonlinear functionalforms of a precipitationeelectivity relationship, but the negative linear model proved most predictive.Our observations match SGH predictions. Studying the role of interspecific interactions in shapingspecies range shifts may lead to improved predictions of distribution ranges and changes in drylandvegetation under global change scenarios.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Facilitation in natural plant communities has been characterizedas an important but under-developed topic of study (Brooker andCallaway, 2009; Brooker et al., 2008; Bruno et al., 2003). Researchover the past decade has shown that the ecological role of facili-tation can vary along gradients of resource availability and abioticstress (Holmgren and Scheffer, 2010; Lortie and Callaway, 2006;Maestre et al., 2006, 2009; Michalet, 2006). The stress-gradienthypothesis (SGH) is a widely supported unifying paradigm (Heet al., 2013) which suggests that the importance of interplantpositive interactions (i.e., facilitation) increases along gradients ofincreasing abiotic stress and, correspondingly, the importance ofnegative interactions (i.e., competition) decreases (Bertness andCallaway, 1994). Accordingly, one would expect to find evidence

þ972 2 6584741.erger), pweisberg@cabnr.unr.lk), yagil@volcani.agri.gov.il

All rights reserved.

of the SGH by observing the net balance between competitive andfacilitative interactions over a species’ range. This is because thebiogeographic ranges of species usually lie along environmentalgradients, and their boundaries are often influenced by abioticstress (Normand et al., 2009) and competitive interactions (Sextonet al., 2009). Given that stress gradients at the scale of a species’range are often strongly linked to climate, the use of the SGH maybe extended from discussion of patterns of interspecific in-teractions to predictive models of species distribution in thecontext of climate change. Thus, the SGH is inherently a question ofbiogeographic scope. However to date only few studies haveaddressed the SGH over extensive portions of a focal species’ range(Armas et al., 2011; Cavieres et al., 2006; Dohn et al., 2012; Schöbet al., 2013; Soliveres et al., 2011). For this reason there is a needto critically examine the SGH over broader spatial scales, and thepresent study attempted to fill this gap. In contrast to previousstudies, which are mainly based on an experimental approach ofexamining plant interactions through controlled manipulations(Armas and Pugnaire, 2005; Callaway et al., 2002; Chambers, 2001;Gómez-Aparicio et al., 2005; Maestre and Cortina, 2004), or alter-natively observational approaches comparing different species in

J. Ziffer-Berger et al. / Journal of Arid Environments 102 (2014) 27e3328

different systems (Holzapfel et al., 2006; Kikvidze et al., 2005,2011), we aimed to detect plant interactions between two focalspecies, in situ, with the goal of describing the functional form ofthe variation of inter-specific plant interactions along abiotic stressgradients, over a biogeographic scale.

Whereas the SGH generally predicts a positive linear relation-ship between abiotic stress level and the relative importance offacilitation, Maestre and Cortina (2004) as well as Michalet (2006)proposed that under certain circumstances nonlinear relationships,such as unimodal or asymptotic ones might be expected. Michalet(2006) and Maestre et al. (2009) further hypothesized that thefunctional form of the relative importance of facilitation might notonly depend on the stress level, but rather might also depend onmultiple additional factors, including the nature of the stress fac-tors, the life history of the target species and ontogenetic stage.

We conducted our study in a water-limited system by assessingthe pair-wise relationship between mature shrubs and tree seed-lings across the entire geographic range of the tree species. Wecarried out an extensive field study of the single-leaf pinyon pine(Pinus monophylla Torr. and Frém.) as a model species, focusing onthe importance of facilitation during its early establishment phase.P. monophylla has already been characterized as relying on facili-tation for early establishment (Chambers, 2001; Chambers et al.,1999), and it co-occurs throughout its range with its facilitatingshrub, the polymorphic species Artemisia tridentata Nutt. (sage-brush) (Durant McArthur and Welch, 1982).

Our approach used the level of spatial association betweenyoung P. monophylla seedlings and mature A. tridentata as a sur-rogate for the relative importance of facilitation (Callaway et al.,1996; Greenlee and Callaway, 1996). We expected that the mostfundamental abiotic stress gradient within the geographic range ofP. monophyllawould be that of water availability, which constitutesthe main limiting factor for seedling establishment (Chambers,2001). Accordingly, we projected that water-related environ-mental factors such as precipitation and temperature regimeswould influence the nature of the inter-specific interaction. Wetested two alternative hypotheses: (1) that in accordance with theSGH the spatial association between P. monophylla and matureA. tridentata would increase consistently with increasing waterstress, which would indicate the relative importance of facilitationover the entire species range; and (2) that the increasing relativeimportance of facilitation with increasing water stress would belimited to only a fraction of the water-stress gradient within thespecies range. In the latter case, asymptotic or unimodal relation-ship patterns might emerge, corresponding to patterns proposedby Maestre and Cortina (2004).

2. Methods

2.1. Target species

P. monophylla Torr. and Frém. is a long-lived tree speciesdistributed in the U.S. Intermountain West, from southern Idaho toBaja California (Farjon and Styles, 1997; Little, 1971; Fig. 1). Its seedsare dispersed by seed-caching rodents and birds, which cache seedsunder and near shrubs, as well as in intershrub open spaces with noclear microsite preference (Vander Wall, 1997). Previous work re-ported that P. monophylla seedlings are more likely to establishunder shrub canopies than in intershrub open spaces (Chambers,2001).

A. tridentata is a widely distributed shrub that occursthroughout most of western North America, from British Columbiato Baja California. It is highly drought resistant, growing in regionswith annual rainfall as little as 160 mm (Kolb and Sperry, 1999). Itconstitutes a major understory component in P. monophylla

woodlands (West et al., 1978) and has been documented as theprimary P. monophylla nurse shrub species (Callaway et al., 1996;Chambers, 2001).

2.2. Study area

Our study encompassed the entire distribution range ofP. monophylla within the semi-arid Intermountain region of theUnited States (Fig. 1). P. monophylla woodland landscapes aretypically patchy mosaics of trees, shrubs and bare interspaces, avegetation formation that is typical of many semi-arid ecosystems(Webster and Maestre, 2004). These woodlands are usually poor inannual species and geophytes, with a shrub dominated understory.

The climatic conditions within the geographic range ofP. monophylla are highly varied. Annual precipitation ranges from200 to 700 mm (PRISM, 2010) and varies with both latitude andelevation. The temporal distribution and form of precipitation arealso highly variable: most precipitation occurs as snow in thewinter season, especially in the western part of the range, withincreasing frequency of monsoonal rains towards the eastern part(Adams and Comrie, 1997). Bedrock types are diverse and includeplutonic, metamorphic and also various sedimentary rockformations.

2.3. Field survey structure

We designed the field survey to quantify the spatial associationbetween P. monophylla seedlings and mature A. tridentata shrubsover a broad range of environmental conditions. We designatedsurvey sites by using stratified random sampling throughout thegeographic range of the P. monophylla. We divided its distributionarea (Little, 1971) into three yearly precipitation levels: 250e350 mm, 351e500 mm, and 501e650 mm, and three regions:eastern Central Great Basin, western Central Great Basin (east-westdivision through the geographical center of Nevada), and MojaveDesert. This allowed us to obtain an equal representation of all partsof the P. monophylla range. A total of 69 spatially random surveysites were visited (Fig. 1). In cases where a randomly computer-generated site was inaccessible or lacked any pine recruitmentwe used the closest suitable location. At each site we randomlyselected a reference point and marked the 25 young seedlingsclosest to the reference point. We laid out a square plot around theselected group of seedlings and therefore the plot size varied ac-cording to seedling density. The 69 plots ranged in area from 80 to155 m2, with a mean plot area of 105 m2. We included onlyestablished seedlings with distinct juvenile characteristics (glau-cous color, short needles) in the survey, and excluded seedlings thatlacked these characteristics and those that had recently germinatedand lacked bark and branches.

2.4. Site data

We characterized each study site by using the following spatiallyextrapolated climate variables, obtained from PRISM (PRISMGroup,2010), and used them to form climatic gradients: (1) annual pre-cipitation, to form a gradient of aridity; (2) minimum Januarytemperature, to form a gradient of extreme low temperature; and,(3) maximum July temperature, to form a gradient of extreme hightemperature. Additional information relevant to each plotincluded: (1) bedrock type from Ludington et al. (2005), classifiedto bedrock types: calcareous and non-calcareous, a property whichnoticeably affects nutrient availability (Larcher, 2003), (2) “mon-soonality index”, calculated as the proportion of annual precipita-tion occurring from July through September (Romme et al., 2009);(3) “topographic position index” (TPI), calculated as the relative

Fig. 1. The intermountain region of the Western United States, including the Basin and Range and the Mojave ecoregions. The gray polygons represent the distributional range ofPinus monophylla (after Little 1971), and the dots represent the survey sites.

J. Ziffer-Berger et al. / Journal of Arid Environments 102 (2014) 27e33 29

difference in elevation between a 0.008-km2 grid cell and the 400-km2 area surrounding it where lower values represent lower hill-slope positions, i.e. proximity to basin habitats; (4) “ecoregion”based on Omernik (1987), classified to the Central Basin and Range,characterized as cold desert with snow as the main winter

precipitation, and the Mojave Basin and Range including its west-ern margins (Fig. 1), characterized as warm desert with rain as themain winter precipitation form (Allen and Geiser, 2011); (5) slope,measured with an inclinometer; and (6) soil texture, classified tocrumbled or not crumbled. Soil texture, used as a measure of

J. Ziffer-Berger et al. / Journal of Arid Environments 102 (2014) 27e3330

organic matter and clay content in the soil, was determined withthe ball test (Soil Survey Division Staff, 1993), by formingmoistenedsoil samples and observing whether the soil collapsed into acrumbled structure or broke into pieces, when dropped onto a hardsurface.

We surveyed only north-facing slopes to reduce the potentiallyconfounding influence of varied slope aspects.

2.5. Microsite characteristics and availability

In order to characterize the abiotic features of microsites(Table 1), we centered a 0.25-m2 quadrat around each seedling andwithin each quadrat we estimated the cover of rocks and litter,quantified in 10% increments. In order to characterize the rela-tionship between P. monophylla seedlings and A. tridentata shrubsin each microsite, we recorded the relationship as one of two cat-egories: (i) shrub microsite and (ii) intershrub space. A shrubmicrosite was within the range of influence of the shrub, defined asbeing shaded by it, either growing directly under it or within 15 cmof the canopy’s downhill margin, during at least part of the day, inwhich case both aboveground and belowground interactions mightbe expected. An intershrub space was outside the shrub’s shadeinfluence. We used this spatial relationship according to theobservation of Chambers (2001), who found shading to be criticalto pinyon seedling survival. To quantify the overall availability ofshrub microsites at each site, we sampled points along two lineartransects within each plot. We oriented the transects parallel to theslope, and fixed their length and distance from each other ac-cording to the plot size, so that they divided the plot area into threeequal rectangles. We recorded microsite characteristics at 25equally spaced points along the transects, by using 0.25-m2 quad-rants as described above for the seedling microsites.

2.6. Spatial association between pine seedlings and shrubs

We used the electivity index (E) (Jacobs, 1974) to quantify theshrub-seedling spatial association. This index relates the propor-tion of shrub microsites used by seedlings to the proportion ofavailable shrub microsites. In the present context, the electivity

Table 1List of variables recorded in the survey, measure units and recorded value range;variables in italics were found uncorrelated and were included in the multipleregression analysis.

Variable Unit Range surveyed

Soil texture Crumbled/non crumbledLitter cover 10% increments 10e100Rock cover 10% increments 0e100Annual precipitation mm/year 210e877Monsoonality (proportion

of summer rain withintotal precipitation)

% 2e28

Max. summer temperature(multiannual maximumJuly mean temperature)

�C 18e32

Min. winter temperature(multiannual minimumJanuary mean temperature)

�C �2 to �13

Topographic position index Meters �428 to 789Elevation Meters 1365e2615Ecoregion Categorical

binary1: Within the Basinand Range (cold desert)0: Mojave Desert(warm desert).

Bedrock type Categoricalbinary

1: Calcareous 0:Non calcareous

Slope inclination Degrees 0e46

index can be interpreted as the level of seedling affinity to shrubmicrosites. The electivity index is calculated as:

E ¼ ln�rð1� pÞpð1� rÞ

�

inwhich r is the proportion of seedlings located in shrub micrositesthus under the influence of shrubs, and p is the proportion ofavailable shrubmicrositeswithin a plot. Positive values of E indicatea positive association between seedlings and shrubs, whereasnegative values indicate a negative association and values near zeroindicate a lack of preference to or against shrub microsites. Weregarded the electivity index as a proxy for the nature and level ofthe interspecific interaction, i.e., facilitation or competition, be-tween tree seedlings and mature shrubs, and used it as a contin-uous response variable in multiple regression analysis while usingvarious site and microsite environmental factors as independentvariables. In sites where shrub cover is very close to 100% or 0%, Ebecomesmeaningless andwe excluded such sites from the analysis.

2.7. Data analysis

We calculated Pearson’s correlation coefficients for all pairwisecombinations of independent numerical variables, in order to avoidmulticollinearity in multiple regression (Graham, 2003). Afterchoosing a subset of independent variables, we implemented for-ward and backward stepwise linear regression, combined withAkaike Information Criterion (AIC) scoring (Akaike, 1974), using theJMP (SAS 2007) software, version 7, with P-to-enter ¼ 0.25 and P-to-remove ¼ 0.10. After identifying the most relevant environ-mental variables, we examined various candidate models,including possible interactions. We examined residuals from theselected/final multiple regression model for spatial autocorrelationby using Moran’s I global autocorrelation statistic and by plottingomnidirectional correlograms (Legendre and Fortin, 1989), imple-mented in R software, version 2.11.

Since we found precipitation to be a key environmental variablewith the highest increment to the coefficient of determination(Table 2), we conducted a multiple linear regression with alterna-tive nonlinear relationships concerning precipitation, i.e., a log xmodel for an asymptotic relationship and a second-degree poly-nomial model for a unimodal relationship.

3. Results

We omitted eight of 69 plots from the electivity calculationsbecause shrub cover was either too low (nearly zero) or too high(nearly 100%) for meaningful estimation of the electivity index. Weobtained Pearson’s correlation coefficients equal to or greater than0.5 for several pairs of environmental variables: litter cover andgravel cover; TPI and maximum July temperature; elevation andmaximum July temperature; and ecoregion and minimum Januarytemperature. We therefore conducted several alternative stepwiseregressions, each using a different subset of independent variables.Additionally, we found soil texture to be similar among most of the

Table 2Parameter estimates of a linear multiple regression model following a stepwiseparameter selection of seedling electivity to shrub microsites (R2 ¼ 0.42, N ¼ 61,DF ¼ 3, p < 0.0001).

Parameter b SE R2 increment p-value

Precipitation �0.006 0.001 0.31 <0.0001Ecoregion �0.38 0.135 0.06 0.0069Litter cover �0.11 0.049 0.05 0.0254

J. Ziffer-Berger et al. / Journal of Arid Environments 102 (2014) 27e33 31

sites and therefore omitted it. The final set of variables that weconsidered formultiple regression analysis comprised bedrock type(calcareous or not), precipitation, slope, maximum July tempera-ture, monsoonality, ecoregion, and litter cover. Stepwise regressionanalysis identified precipitation, ecoregion, and litter cover assignificantly related to electivity, and we included them as com-ponents of the most parsimonious model (R2 ¼ 0.42, P < 0.0001,n ¼ 61, AIC ¼ �10.92; Table 2). We examined two models based onthese environmental variables that incorporated the potential in-teractions between ecoregion and precipitation, and betweenecoregion and litter cover. Neither interaction was statisticallysignificant (P ¼ 0.59 and P ¼ 0.37, respectively).

Precipitation, the most influential environmental variable inaffecting the level of spatial association between shrubs andseedlings, was negatively related to electivity (b ¼ �0.006). Elec-tivity was highest (E z 2.5) at the xeric end of the gradient whereannual precipitation approximated 200 mm and decreased toslightly negative values (E z �0.5) at the mesic end where annualprecipitation approximated 650. Electivity was negatively relatedto litter cover at the microsite scale (b ¼ �0.11; SE ¼ 0.50) and wasconsiderably lower (b ¼ �0.38; SE ¼ 0.14) in the cold desert (Basinand Range) than in the warm desert (Mojave Basin).

We tested two alternative nonlinear models, i.e., a 2nd-degreepolynomial model (R2 ¼ 0.45, P < 0.0001, n¼ 61, AIC¼ �11.34) anda log x model (R2 ¼ 0.44, P < 0.0001, n ¼ 61, AIC ¼ �13.05) forelectivity. The models included precipitation as the transformedterm, and ecoregion and litter cover as covariates. We comparedthese to the equivalent linear regression model. These showedslightly higher R2 values though they did not reveal any consider-able difference in pattern from the negative linear relationship thatwas suggested by a linear model (R2 ¼ 0.31, P < 0.0001, n ¼ 61,AIC ¼ �4.37; Fig. 2). Results from Moran’s I on model residualsshowed a lack of spatial autocorrelation over any lag distance(I ¼ 0.068, P ¼ 0.513).

4. Discussion

We identified precipitation amount as a main environmentalfactor influencing the level of spatial association betweenP. monophylla seedlings and A. tridentata, as indicated by the elec-tivity index. Electivity shifted from positive values at the more xericend of the precipitation gradient to negative ones at the mesic endof the gradient. This pattern suggests that the relative importanceof positive interspecific interactions increases with increasingdrought stress throughout the entire range of P. monophylla, and isconsistent with the stress-gradient hypothesis as originally pro-posed by Bertness and Callaway (1994).

Mean annual precipitation is commonly considered as a quan-titative variable to identify or characterize spatial gradients ofwater availability (Holzapfel et al., 2006; Maestre and Cortina,2004; Tielbörger and Kadmon, 2000). Therefore the increasinglevel of shrub-seedling association with decreasing precipitationlevel probably can be interpreted to indicate increasing relative

Fig. 2. Comparison of alternative models (N ¼ 61) for the relationship between electivity (aspolynomial (graph C) models do not show any substantial difference to linear model (grap

importance of facilitation by shrubs with decreasing water avail-ability. In such cases facilitation by shrubs is likely realized throughamelioration of the water stress experienced by tree seedlings.Previous work on the interaction between A. tridentata andP. monophylla seedlings had already demonstrated the existence offacilitation by shrubs, mainly through shading and improvement ofnutrient availability (Chambers, 2001). An additional possiblemechanism of facilitation relates to the phenomenon of hydrauliclift by Richards and Caldwell (1987) and demonstrated inA. tridentata shrublands (Caldwell et al., 1998).

Similar spatial patterns have been recorded on a global scale inalpine plants (Kikvidze et al., 2005, 2011). Their finding that plantaggregation increased with colder climate is analogous to ourfinding of increased spatial associationwith increased aridity in ourwater-limited study area. Though each case demonstrates adifferent, but overriding stress gradient, both confirm the SGHprediction. The nature of planteplant interactions along the stressgradient is an important determinant of spatial patternwithin plantcommunities.

In addition to precipitation, ecoregion and litter cover weresignificant in explaining the variation in shrub-seedling associa-tion. The shrub-seedling association was stronger in the warmdesert than in the cold desert and also increased with decreasinglitter cover. The ecoregion factor represents average temperatureand seasonal temperature extremes in addition to growing seasonlength and the proportion of snow versus rain precipitation events(Rundel and Gibson, 1996). These differences result in overall lowerpotential evapotranspiration and consequently a more favorablewater balance for a given precipitation level in the cold than in thewarm desert. Dohn et al. (2012) report a similar effect of ecoregion.Litter cover expresses a microsite effect on water availability; itreduces evaporative loss and improves the seedling water balance(Ibáñez and Schupp, 2002). Litter cover is also related to otherfactors that might enhance P. monophylla seedling establishment,such as improved nutrient availability and inhibition of competitors(Cavieres et al., 2007).

Although the three factors that were significantly related toshrub-seedling association are all assumed to be strongly related towater availability, two other factors that may also influence thewater balance did not show any clear relationship with electivitylevel; they were summer maximum temperature and monsoon-ality. It might have been expected that in sites with higher summertemperatures water loss through evapotranspiration would in-crease and the shrub-seedling association would, therefore, bestronger, but we suggest that summer temperature maxima are lessindicative of water availability for the pine seedlings since in thisseason growth of P. monophylla is nearly nil. It might also have beenexpected that better water conditions for seedlings and conse-quently lower levels of shrub-seedling association would occur insites with higher summer rainfall (higher monsoonality), but wesuggest that the lack of influence of monsoonality on electivity canbe attributed to the seedlings’ inability to exploit the short pulses ofrain that occur during the hot dry season, as previously reported by

a proxy for facilitation) and precipitation: the logarithmic (graph B) and the 2nd-degreeh A) in their trend.

J. Ziffer-Berger et al. / Journal of Arid Environments 102 (2014) 27e3332

Huxman et al. (2004) for other Great Basin and Mojave Desertperennials. Bimodal patterns of primary productivity over thecourse of the growing season, caused by discrete periods of wateravailability from winter snowmelt and monsoonal rains (Notaroet al., 2010), are commonly observed in the southwestern U.S. butless so in the Great Basin. Furthermore, Lyr and Garbe (1995)documented extremely low growth rates of Pinus sylvestris sap-lings irrigated during summer when root zone temperatures werehigh. Thus, summer rains do not necessarily improve water avail-ability for P. monophylla seedling establishment and consequentlyexert no significant effect on pineeshrub association.

Michalet et al. (2006) and Maestre et al. (2009) have suggestedthat under conditions of extreme environmental stress, such ascould be expected at the xeric end of the P. monophylla range, therelative importance of competition might increase. This wouldlead to a nonlinear, unimodal relationship pattern in which thenet balance between competition and facilitation would tilt to-wards competition at both the xeric and the mesic ends of thegradient. Our findings could not provide any support for thishypothesis.

Application of the SGH to modeling interspecific interactionsover range-wide scales, as implemented here, holds promise foraddressing one of the major challenges in using niche-based spe-cies distribution models to predict climate change responses, andspecifically the incorporation of interspecific interactions(Butterfield et al., 2013; Guisan and Thuiller, 2005; He et al., 2013;Thuiller et al., 2008). The SGH represents a unifying ecologicaltheory resting upon environmental gradients that can be readilymapped. Thus, predictions based on the SGH can be tied directly tomodels of biotic response to climate change, perhaps leading toimproved predictions of shifts in species geographical ranges andchanges in plant community composition.

Our results suggest that disturbance which involves removal ofshrub cover might accelerate shifts at both the trailing and leadingedges of tree species ranges, in response to climate changeinvolving decreased annual precipitation (Aráoz and Grau, 2010;Murphy et al., 2010). Vegetation removal should result in lessfacilitation for seedling establishment at the drier end (trailing edgeof the range), whereas at the humid or leading edge, it should leadto diminished competition, favoring species range shifts (i.e.,“march forward”) rather than changes merely in the central ten-dency of the distribution of abundance (i.e. “lean forward” sensuBreshears et al. 2009). There is a need for more studies that testpredictions of the SGH over biogeographic scales, and for devel-opment of quantitative models that incorporate response functionsdescribing how interspecific interactions vary in type and magni-tude along extensive ecological gradients. Research on how speciesinteractions vary over the extent of species’ geographical rangesshould progress most efficiently through the linking of controlledfield and laboratory experiments (Armas and Pugnaire, 2005;Chambers, 2001; Gómez-Aparicio et al., 2005; Maestre andCortina, 2004) to observational studies across biogeographicallyrelevant scales.

Acknowledgments

We are very grateful to the International Arid Land Consortiumfor funding this project. We would like to thank Todd Granberry forhis assistance in field work, Tom Dilts for processing the GIS data,and Jian Yang for his useful comments on the manuscript.

References

Adams, D.K., Comrie, A.C., 1997. The North American monsoon. Bull. Am. Meteorol.Soc. 78, 2198e2213.

Akaike, H., 1974. A new look at the statistical model identification. IEEE Trans.Automat. Control 19, 716e723.

Allen, E.B., Geiser, L.H., 2011. The North American Desert. In: Pardo, L.H., Robin-Abbott, M.J., Driscoll, C.T. (Eds.), Assessment of Nitrogen Deposition Effect andEmpirical Critical Loads of Nitrogen for Ecoregions in the United States. USDA,Forest Service, Northern Research Station, Newton Square, PA.

Aráoz, E., Grau, H.R., 2010. Fire-mediated forest encroachment in response to cli-matic and land-use change in subtropical Andean treelines. Ecosystems 13,992e1005.

Armas, C., Pugnaire, F.I., 2005. Plant interactions govern population dynamics in asemiarid plant community. J. Ecol. 93, 978e989.

Armas, C., Rodríguez-Echeverría, S., Pugnaire, F.I., 2011. A field test of the stress-gradient hypothesis along an aridity gradient. J. Veg. Sci. 22, 818e827.

Bertness, M.D., Callaway, R.M., 1994. Positive interactions in communities. TrendsEcol. Evol. 9, 191e193.

Breshears, D.D., Myers, O.B., Meyer, C.W., Barnes, F.J., Zou, C.B., Allen, C.D.,McDowell, N.G., Pockman, W.T., 2009. Tree die-off in response to global change-type drought: mortality insights from a decade of plant water potential mea-surements. Front. Ecol. Environ. 7, 185e189.

Brooker, R.W., Maestre, F.T., Callaway, R.M., Lortie, C.L., Cavieres, L.A., Kunstler, G.,Liancourt, P., Tielbörger, K., Travis, J.M.J., Anthekme, F., Armas, C., Coll, L.,Crocket, E., 2008. Facilitation in plant communities: the past, the present, andthe future. J. Ecol. 96, 18e43.

Brooker, R.W., Callaway, R.M., 2009. Facilitation in the conceptual melting pot.J. Ecol. 97, 1117e1120.

Bruno, J.F., Stachowicz, J.J., Bertness, D.M., 2003. Inclusion of facilitation intoecological theory. Trends Ecol. Evol. 18, 119e125.

Butterfield, B.J., Cavieres, L.A., Callaway, R.M., Cook, B.J., Kikvidze, Z., Lortie, C.J.,Michalet, R., Pugnaire, F.I., Schöb, C., Xiao, S., Zaitchek, B., Anthelme, F.,Björk, R.G., Dickinson, K., Gavilan, R., Kanka, R., Maalouf, J.P., Noroozi, J.,Parajuli, R., Phoenix, K., Reid, A., Ridenour, W., Rixen, C., Wipf, S., Zhao, L.,Brooker, R.W., 2013. Alpine cushion plants inhibit the loss of phylogenetic di-versity in severe environments. Ecol. Lett. 16, 478e486.

Caldwell, M.M., Dawson, T.E., Richards, J.H., 1998. Hydraulic lift: consequences ofwater efflux from the roots of plants. Oecologia 113, 151e161.

Callaway, R.M., DeLucia, E.H., Moore, D., Nowak, R., Schlesinger, W.H., 1996.Competition and facilitation: contrasting effects of Artemisia tridentata ondesert vs. montane pines. Ecology 77, 2130e2141.

Callaway, R.M., Brooker, R.W., Choler, P., Kikvidze, Z., Lortie, C.J., Michalet, R.,Paolini, L., Pugnaire, F.I., Newingham, B., Aschehoug, E.T., Armas, C., Kikodze, D.,Cook, B.J., 2002. Positive interactions among alpine plants increase with stress.Nature 417, 844e848.

Cavieres, L.A., Badano, E.I., Sierra-Almeida, A., Gómez-González, S., Molina-Montenegro, M.A., 2006. Positive interactions between alpine plant species andthe nurse cushion plant Laretia acaulis do not increase with elevation in theAndes of central Chile. New Phytol. 169, 59e69.

Cavieres, L.A., Chacón, P., Peñaloza, A., Molina-Montenegro, M., Arroyo, M.T.K., 2007.Leaf litter of Kageneckia angustifolia D. Don (Rosaceae) inhibits seed germina-tion in sclerophyllous montane woodlands of central Chile. Plant Ecol. 190, 13e22.

Chambers, J.C., 2001. Pinus monophylla establishment in an expanding Pinus-Juni-perus woodland: environmental conditions, facilitation and interacting factors.J. Veg. Sci. 12, 27e40.

Chambers, J.C., Vander Wall, S.P., Schupp, E.W., 1999. Seed and seedling ecology ofpiñon and juniper species in the pygmy woodlands of western North America.Bot. Rev. 65, 1e38.

Dohn, J., Dembele, F., Karembe, M., Moustakas, A., Amevor, K., Hanan, N.P., 2012.Tree effects on grass growth in savannas: competition, facilitation and thestress-gradient hypothesis. J. Ecol. 101, 202e209.

Durant McArthur, E., Welch, B.L., 1982. Growth ate differences among Big Sagebrush[Artemisia tridentata] accessions and subspecies. J. Range Manag. 35, 396e401.

Farjon, A., Styles, B.T., 1997. Flora Neotropica e Pinus (Pinaceae). New York BotanicalGarden Press, New York.

Gómez-Aparicio, L., Gómez, J.M., Zamora, R., Boettinger, J.L., 2005. Canopy vs. soileffects of shrubs facilitating tree seedlings in Mediterranean montane ecosys-tems. J. Veg. Sci. 16, 191e198.

Graham, M.H., 2003. Confronting multicollinearity in ecological multiple regres-sion. Ecology 84, 2809e2815.

Greenlee, J.T., Callaway, R.M., 1996. Abiotic stress and the relative importance ofinterference and facilitation in montane bunchgrass communities in WesternMontana. Am. Nat. 148, 386e396.

Guisan, A., Thuiller, W., 2005. Predicting species distribution: offering more thansimple habitat models. Ecol. Lett. 8, 993e1009.

He, Q., Bertness, M.D., Altieri, A.H., 2013. Global shifts towards positive speciesinteractions with increasing environmental stress. Ecol. Lett. 16, 695e706.

Holmgren, M., Scheffer, M., 2010. Strong facilitation in mild environments: thestress gradient hypothesis revisited. J. Ecol. 98, 1269e1275.

Holzapfel, C., Tielbörger, K., Parag, H.A., Kigel, J., Sternberg, M., 2006. Annual planteshrub interactions along an aridity gradient. Basic Appl. Ecol. 7, 268e279.

Huxman, T.E., Snyder, K.A., Tissue, D., Leffler, A.J., Olge, K., Pockman, W.T.,Sandquist, D.R., Potts, D.L., Schwinning, S., 2004. Precipitation pulses and car-bon fluxes in semiarid and arid ecosystems. Oecologia 141, 254e268.

Ibáñez, I., Schupp, E.W., 2002. Effects of litter, soil surface conditions, and micro-habitat on Cercocarpus ledifolius Nutt. seedling emergence and establishment.J. Arid Environ. 52, 209e221.

J. Ziffer-Berger et al. / Journal of Arid Environments 102 (2014) 27e33 33

Jacobs, J., 1974. Quantitative measurement of food selection: a modification of theforage ratio and Ivlev’s electivity indices. Oecologia 14, 413e417.

Kikvidze, Z., Pugnaire, F.I., Brooker, R.W., Choler, P., Lortie, C.J., Michalet, R.,Callaway, R.M., 2005. Linking patterns and processes in alpine plant commu-nities: a global study. Ecology 86, 1395e1400.

Kikvidze, Z., Michalet, R., Brooker, R.W., Cavieres, L.A., Lortie, C.J., Pugnaire, F.I.,Callaway, R.M., 2011. Climatic drivers of planteplant interactions and diversityin alpine communities. Alpine Bot. 121, 63e70.

Kolb, K.J., Sperry, J.S., 1999. Differences in drought adaptation between subspecies ofsagebrush (Artemisia tridentata). Ecology 80, 2373e2384.

Larcher, W., 2003. Physiological Plant Ecology, fourth ed. Springer, Berlin.Legendre, P., Fortin, M.J., 1989. Spatial pattern and ecological analysis. Vegetatio 80

(2), 107e138.Little Jr., E.L., 1971. Atlas of United States trees, Volume 1: Conifers and Important

Hardwoods. U.S. Department of Agriculture.Lortie, C.J., Callaway, R.M., 2006. Re-analysis of meta-analysis: support for the

stress-gradient hypothesis. J. Ecol. 94, 7e16.Ludington, S., Moring, B.C., Miller, R.J., Flynn, K., Hopkins, M.J., Stone, P., Bedford, D.R.,

Haxel, G.A., 2005. Preliminary Integrated Geologic Map Databases for the UnitedStates e Western States: California, Nevada, Arizona, and Washington [USGSopen-file report]. URL: http://mrdata.usgs.gov/sgmc/ca.html.

Lyr, H., Garbe, V., 1995. Influence of root temperature on growth of Pinus sylvestris,Fagus sylvatica, Tilia cordata and Quercus robur. Trees 9, 220e223.

Maestre, F.T., Cortina, J., 2004. Do positive interactions increase with abiotic stress?A test from a semi-arid steppe. Proc. R. Soc. B 271 (Suppl.), 331e333.

Maestre, F.T., Valladares, F., Reynolds, J.F., 2006. The stress-gradient hypothesis doesnot fit all relationships between plant-plant interactions and abiotic stress:further insights from arid environments. J. Ecol. 94, 17e22.

Maestre, F.T., Callaway, R.M., Valladares, F., Lortie, C.J., 2009. Refining the stress-gradient hypothesis for competition and facilitation in plant communities.J. Ecol. 97, 199e205.

Michalet, R., 2006. Is facilitation in arid environments the result of direct or com-plex interactions. New Phytol. 169, 3e6.

Michalet, R., Brooker, R.W., Cavieres, L.A., Kikvidze, Z., Lortie, C.J., Pugnaire, F.I.,Valiente-Banuet, A., Callaway, R.M., 2006. Do biotic interactions shape bothsides of the humpedeback model of species richness in plant communities.Ecol. Lett. 9, 767e773.

Murphy, H.T., VanDerWal, J., Lovett Doust, J., 2010. Signatures of range expansionand erosion in eastern North American trees. Ecol. Lett. 13 (10), 1233e1244.

Normand, S., Treier, U.A., Randin, C., Vittoz, P., Guisan, A., Svenning, J.C., 2009.Importance of abiotic stress as a range limit determinant for European plants:insights from species responses to climatic gradients. Glob. Ecol. Biogeogr. 18(4), 437e449.

Notaro, M., Liu, Z., Gallimore, R.G., Williams, J.W., Gutzler, D.S., Collins, S., 2010.Complex seasonal cycle of ecohydrology in the Southwest United States.J. Geophys. Res. 115, 1e20.

Omernik, J.M., 1987. Ecoregions of the conterminous United States. Ann. Assoc. Am.Geogr. 77, 118e125.

PRISM Climate Group. URL: http://www.prism.oregonstate.edu/ (accessed11.01.12.).

Richards, J.H., Caldwell, M.M., 1987. Hydraulic lift: substantial nocturnal watertransport between soil layers by Artemisia tridentata roots. Oecologia 73, 486e489.

Romme, W.H., Allen, C.D., Bailey, J.D., Baker, W.L., Bestelmeyer, B.T., Brown, P.M.,Eisenhart, K., Floyd-Hanna, L., Huffman, D.W., Jacobs, B.F., Miller, R.F.,Muldavin, E.H., Swetnam, T.W., Tausch, R.J., Weisberg, P.J., 2009. Historicaland modern disturbance regimes, stand structures, and landscape dynamicsin pinyon-juniper vegetation of the Western U.S. Range Ecol. Manag. 62,203e222.

Rundel, P.W., Gibson, A.C., 1996. Ecological Communities and Processes in a MojaveDesert Ecosystem: Rock Valley, Nevada. Cambridge University Press, Cambridge.

Schöb, C., Armas, C., Guler, M., Prieto, I., Pugnaire, F.I., 2013. Variability in functionaltraits mediates plant interactions along stress gradients. J. Ecol. 101, 753e762.

Sexton, J.P., McIntyre, P.J., Angert, A.L., Rice, K.J., 2009. Evolution and ecology ofspecies range limits. Annu. Rev. Ecol. Evol. Syst. 40, 415e436.

Soil Survey Division Staff, 1993. Soil Survey Manual. Soil Conservation Service[Handbook 18]. U.S. Department of Agriculture.

Soliveres, S., Eldridge, D.J., Maestre, F.T., Bowker, M.A., Tighe, M., Escudero, A., 2011.Microhabitat amelioration and reduced competition among understorey plantsas drivers of facilitation across environmental gradients: towards a unifyingframework. Perspect. Plant Ecol. Evol. Syst. 13, 247e330.

Thuiller, W., Albert, C., Araujo, M.B., Berry, P.M., Cabeza, M., Guisan, A., Hickler, T.,Midgley, G.F., Paterson, J., Schurr, F.M., Sykes, M.T., Zimmermann, N.E., 2008.Predicting global change impacts on plant species’ distributions: future chal-lenges. Perspect. Plant Ecol. Evol. Syst. 9, 137e152.

Tielbörger, K., Kadmon, R., 2000. Temporal environmental variation tips thebalance between facilitation and interference in desert plants. Ecology 81,1544e1553.

Vander Wall, S.B., 1997. Dispersal of singleleaf piñon pine (Pinus monophylla) byseed-caching rodents. J. Mammal. 78, 181e191.

Webster, R., Maestre, F.T., 2004. Spatial analysis of semi-arid patchy vegetation bythe cumulative distribution of patch boundary spacings and transition proba-bilities. Environ. Ecol. Stat. 11, 257e281.

West, N.E., Tausch, R.J., Rea, K.H., Tueller, P.T., 1978. Taxonomic determination,distribution, and ecological indicator values of sagebrush within the pinyon-juniper woodlands of the Great Basin. J. Range Manag. 3, 87e92.