Parveen Kumar Chhetri

Introduction

The natural upper boundary of a forest—or upper forest line—in mountain environments is a good indicator of past and current climate(Danzeglocke 2005;K?rner 2012).Recent increases in global average temperatures have resulted in an upward advance of forest lines,and higher recruitment(seedlings and saplings)near the forest line.Numerous studies have documented altitudinal expansion of the forest line in many mountain ranges of the world,including the Polar Urals,Russia(Devi et al.2008),the central Swiss Alps,Switzerland(Vittoz et al.2008),and the central Himalaya,Nepal(Gaire et al.2014).

Under conditions of upward movement of the forest line,it is important to delineate or map this boundary,enabling future monitoring of potential encroachment of the treeless alpine zone by the forest(Carlson et al.2013).Studies,such as those by Randin et al.(2009)and Forrest et al.(2012),have indicated that a shift of the forest line will shrink the alpine area and fragment the alpine ecosystem.Without knowledge about the area available for advance of the forest line,however,it is dif ficult to predict whether the alpine area will shrink in the future.Especially in areas such as the Himalayas,where the role of the topography in determining the extent of the alpine ecosystem is prominent,mapping the upper limit of sub-alpine forest and determining the area above the current forest limit available for encroachment is critical.

Satellite imagery,such as Landsat,in combination with digital elevation models(DEMs)can be utilized to identify and analyze the upper forest limit.Species-distribution models(SDMs)or habitat-suitability models(HSMs)can be used to understand species niche requirements and to predict the potential distribution of forest-line species(Hirzel et al.2006).These models are process based as well correlative,and use species presence and absence data,topo-climatic variables,and land-cover data to predict the potential habitat of forest-line species(Hirzel et al.2006;Carlson et al.2013).Nowadays,a hybrid approach,which combines correlative and process-based models into a single approach,is the most common(Boulangeat et al.2012).However,the combination of advanced remotesensing technology with physiology-based ecotone modeling represents a promising approach for species-distribution modeling in forest line areas(Coops et al.2013).

The forest line of the Nepalese Himalayas is one of the most diverse and least studied sub-alpine ecosystems of the world.Its habitat heterogeneity provides a home for many endangered species,but also makes it a center of agropastoralism.In Nepal,the upper forest line ranges from 3500 m in the West to 4100 m in the East,and Abies spectabilis,Betula utilis,Pinus wallichiana,and Juniperus indica are the main species forming the forest line(Gaire et al.2014;Chhetri and Cairns 2015;Shrestha et al.2015;Chhetri et al.2017).Several studies,such as the one by Xu et al.(2009),have indicated that the forest line will advance in response to increasing temperatures,which will result in a loss of habitat for endangered species such as Panthera uncia(snow leopard)(Forrest et al.2012)and endangered medicinal herbs such as Ophiocordyceps sinensis(caterpillar fungi).

Therefore,it is important to map the current forest line and predict how far and where this forest line will advance in the future.Consequently,my study aims to address the following research questions:(1)what is the current distribution pattern of sub-alpine forest in Makalu Barun National Park,eastern Nepal,and(2)how much area suitable for future encroachment is present above the current forest line.

To address these questions,the objectives of this study are to map the current distribution pattern of the forest line,develop a hybrid cartographic modelforforest-line advance,and identify areas where the upper forest line can advance.

Materials and methods

Study area

This study was carried out in the Makalu Barun National Park(MBNP)of eastern Nepal(Fig.1).The MBNP was established in 1992 and shares its western boundary with the Sagarmatha National Park,Nepal,and its northern boundary with the Qomolangama Nature Preserve,China.The park covers an area of 1500 km2,with elevations ranging from about 300 m in the Arun river area to 8025 m at the summit of the world’s fifth highest mountain Makalu(Byers 1996).This study is focused on U-shaped river valleys in the northern part of the park where the upper limit of the sub-alpine forest is present.The forest is undisturbed and the forest line is natural in most of the area(Chhetri and Cairns 2015,2016).

Major river valleys in the MBNP are the Saldima Khola(River),Kasuwa Khola,Barun Khola,Isuwa Khola,Apsuwa Khola,Sankhuwa Khola,Hongu Khola,and Inkhu Khola.These streams are glacier-fed and show evidence of Pleistocene glaciation at altitudes covered by subalpine forests today(Carpenter and Zomer 1996).The MBNP lies within the subtropical Asian monsoon zone,which is characterized by pronounced summer rainfall between June and September.

Hybrid cartographic model design

Many modeling approaches are available for predicting vegetation patterns at the alpine tree-line ecotone.However,due to its simplicity and wide popularity,I used a cartographic modeling approach.I developed a hybrid cartographic model(HCM)for predicting the future advance of the current upper forest limit to higher elevation based on expert knowledge and environmental variables such as elevation,slope,and aspect(slope direction).I used this modeling technique for identifying the area where the upper forest line can advance in the future,as well as the area that will be unsuitable for tree establishment.I have adapted a deductive approach derived from expert knowledge to create the suitability model,and have used five variables as well as expert knowledge(Table 1;Fig.2)to create the HCM.

Variables generation for the model

Landsat 8 imagery for the year 2014 was acquired from the United States Geological Survey(USGS)Earth Explorer website (http://earthexplorer.usgs.gov,accessed on 5 December 2015)to map the current distribution of forest and the upper forest line.The Landsat imagery was already precision-terrain corrected,orthorecti fied,and georeferenced when it was obtained from the USGS.Radiometric corrections(Chander et al.2009),an atmospheric correction using the dark-object subtraction method(Chaplin and Brabyn 2013),and image ratioing to reduce the topographic effect on the imagery(Colby 1991)were carried out.Maximum-likelihood classi fication,which had been used earlier by Danzeglocke(2005)and Panigrahy et al.(2010)for forest-line mapping,was used to generate a land-cover map of the Makalu Barun National Park(Fig.3).Land-cover polygons were generated and overlaid with digital elevation model(DEM)polygons to determine the upper limit of the forest.The upper forest line wasdelineated by connecting the uppermost patches of forest in the land-cover map.

Table 1 Environmental variables used to create the hybrid cartographic model

The Advanced Spaceborne Thermal Emission and Re flection Radiometer—Digital Elevation Model(ASTER–DEM)was obtained from the USGS(http://earthexplorer.usgs.gov,accessed on 5 December 2015).A 3×3 low-pass filter was used to smooth the DEM in order to reduce noise(Forkuor and Maathuis 2012)and a 3×3 Gaussian filter was used to eliminate edge interference in interpolated areas(Eckert et al.2005).Elevation values were extracted from the DEM(Fig.2a)to identify the upper limit of forest distribution in the MBNP.Chhetri and Cairns(2015)observed the forest line in the Barun valley in the MBNP at around 4000 m.The upper-treeline ecotonecoincideswith amean globaltemperatureof 6.4 ± 0.7 °C during the growing season(K?rner 2012).If we consider the growing-season temperature of the forest line at around 6.4 °C,a lapse rate of 0.52 °C 100 m-1in the central Himalaya(Kattel et al.2013),and a warming scenario of 2.5°C,forests can grow up to an elevation of 4500 m under future warming scenarios.Based on this concept,I have classi fied elevation into two zones:

1. Elevations below 4500 m(suitable);this is the elevation range where the upper forest line is currently observed and the temperature is suitable for tree growth in a warming scenario.

2. Elevations above 4500 m(unsuitable):temperatures in this zone would be too low to supporttree establishment.

Fig.2 Topographic variables used for creating the cartographic model.NDVI Normalized Difference Vegetation Index,NDWI Normalized Difference Water Index

Slope is one of the most important parameters controlling the spatial distribution pattern of the upper forest limit.A slope-angle map was generated from the DEM(Fig.2b)using the 4-cell method for slope modeling(Jenness 2013).Slope maps with slope angles above and below 34°were also created for implementation into the model.Areas with slopes above 34°are considered unsuitable for tree establishment because debris flows and snow avalanches endemic to such slopes will affect tree establishment(Walsh and Kelly 1990).In addition,tree establishment on slopes steeper than 34°is not possible due to a lack of soil formation(Brown 1994).The Normalized Difference Vegetation Index(NDVI)was calculated from atmospherically and radiometrically corrected Landsat imagery,using the standard equation(Chaplin and Brabyn 2013).Based on the NDVI values,a land-cover map of the park was created(Fig.2c).This NDVI-based land cover map was reclassified into two classes using a NDVI threshold value of 4.5.Values greater than 4.5 indicate dense forest,while values below 4.5 denote other land-cover types(Singh et al.2013).This new NDVI class map was incorporated into the model.

Fig.3 Land cover map of the Makalu Barun National Park(maximum likelihood classi fication).Forest line and alpine limit are indicated with dash lines

Fig.4 Map produced by the HCM to identify potential areas available for forest line advance

The Normalized Difference Water Index(NDWI)was used to delineate open-water features(McFeeters 1996).Areas with high positive NDWI values are bodies of water,while high negative values are typical for nonwater features such as terrestrial vegetation(Fig.2d).McFeeters(2013)used a threshold value of 0.3 to delineate swimming pools.Accordingly,I used a threshold value of 0.3 to differentiate water bodies from land and incorporated this information into the model,following the general principle that water bodies are not suitable for tree establishment.

Tree growth above the forest linedepends on the growing season temperature and duration of the growing season(K?rner 1998).Due to variations in local topography,some areas will receive more solar radiation than others,resulting indifferentgrowing-seasontemperaturesanddurations.The hypothetical illumination of a surface(hillshade)was calculatedtosimulateincomingsolarradiationusingthe DEM;this was done by determining illumination values for each cell in a raster,using ArcGIS 10.1.

Zenith angle,azimuth angle,slope,and altitude required for calculating hillshade were determined,considering the sampling date and time of each image(Pierce Jr.et al.2005).Areas receiving little sunlight are colored black,while those receiving large amounts of solar radiation are colored white(Fig.2e).Areas receiving little sunlight are considered unsuitable for tree establishment.Accordingly,the hillshade map was classi fied into two classes and included in the model.

Final map and accuracy assessment

Rasters(elevation,slope,NDVI,NDWI,and hillshade)obtained using the methods described above were reclassi fied to two final classes:suitable(suitable for forest encroachment)and unsuitable (unsuitable forforest encroachment).The five variables selected(Table 1)were added together without assigning weights.The final suitability map(Fig.4)was created by using the Spatial Analyst tools of ArcGIS 10.1.For accuracy assessment of the model,100 random location points were created in the final suitability map generated by the model and in highresolution Google Earth imagery.Three land-cover classes—existing forest,suitable habitat,and unsuitable habitat—were observed at these random location points.Observations were first carried out on the suitability map and then on the high-resolution Google Earth imagery to see whether the model predictions was correct.Expert and field knowledge was used to identify the correct land-cover class on the Google Earth imagery.

Results

The Makalu Barun National Park can be divided into seven major river valleys,which are,from east to west:Saldima,Barun,Isuwa,Apsuwa,Sankhuwa,Hongu,and Inkhu(Fig.1).Roughly 30%of the park’s area is above 5000 m,while 70%is below 5000 m.As shown in Fig.3,most of the park is covered by permanent snow and glaciers.The area below 5000 m is covered by vegetation,mostly dominated by sub-alpine forest.Abies spectabilis(Himalayan silver fir)is a dominant tree species in the subalpine forest and covers the south,north-,and east-facing slopes in most of the river valleys(Fig.3).

The upper forest limit is found to be below 3000 m in the lower portion of the river valleys,and up to about 4000 m along the upstream areas of the river valleys(Fig.3).This upper forest limit is higher on southern rather than on northern and eastern facing slopes.With regard to slope angle and forest distribution,most of the dense vegetation cover is present in areas where the slope angle is less than 35°and the elevation is below 5000 m.Lowvegetation density is observed above 5000 m,even where the slope is less than 35°due to unfavorable climatic conditions.

The land-cover map shows that the distribution of the upper forest line and alpine zone in different valleys of the MBNP is random,but it is highest in the Barun valley and lowest in the Inkhu valley(Table 2;Fig.3).The upper forest limit in most of the valleys is lower than the potential limit,which is around 4000 m,according to field observations by Chhetri and Cairns(2015).

Accuracy assessment and final map

The accuracy of the model in predicting different types of land cover is presented in Table 3.Overall accuracy of the model is 82%based on the comparison of 100 random points generated on the HCM final map and the actual conditions in the field using high-resolution Google Earth imagery and field-based observations.The map generated by the model(Fig.4)indicates that 65%of the park area is unsuitable for forest-line advance as most of this area is under permanent snow cover.The model also indicates that only 16%of the park area is suitable for future forest encroachment and that only 10%of the area is above the current upper-forest line limit.If the climate becomes favorable for tree establishment in the future,the forest line should advance in this 10%of the park area.Potential forest encroachment would occur along river valleys and adjacent south-and north-facing slopes.The model is sensitive to slope because an increase in slope causes the total area suitable for forest encroachment to increase.Removal of the NDWI does not affect the model much.However,a change in the threshold value of the NDVI and NDWI impacts the sensitivity of the model.An increase in the NDVI threshold decreases the total suitable area,while a decrease in the NDVI threshold increases the area suitable for tree encroachment.Similarly,an increase in the NDWIthreshold decreasesthe unsuitable areaand increases the suitable area.

Discussion

The accuracy of the model is comparable to other studies such as Lauver et al.(2002)and acceptable for land-cover mapping and cartographic-modeling studies.High accuracy in predicting unsuitable habitat might be due to the presence of permanent snow cover as the major land cover type of the area.Compared to complex land-cover types typical for lower elevations,it is very easy to determine that areas under permanent snow cover will be unsuitable for tree establishment.In areas with lower elevation,a numberoffactorssuch asslope angle,convexity,topographic breaks,and landform features must be considered before classifying an area as suitable or unsuitable.Therefore,chances of incorrect predictions are high in such complex topographic areas.

Table 2 Forest line and alpine limit in different river valleys of the MBNP based on the land cover map generated from Landsat 8 imagery using maximum likelihood classi fication

Table 3 Accuracy of the hybrid cartographic model(HCM)

Factors such as growing season,air temperature,wind,snow accumulation,light intensity,available tree species and genetic types,available seed and the dynamics of seed dispersal,disturbances,and natural and anthropogenic factors might have affected the character,extent,and spatial organization of the forest line in this alpine environment(Walsh and Kelly 1990).Fang et al.(2014)mentioned that the spatial distribution of the forest line is controlled by climate,topography,surface materials,and anthropogenic forces,whereas Malanson(2001)described that spatial patterns of the forest line are a result of the abiotic environment,such as lithology.Abiotic conditions are considered strong drivers of the spatial distribution of plant communities in the alpine environment(Carlson et al.2013).Snow cover,soil,and soil moisture are abiotic factors that in fluence the spatial pattern of the alpine treeline ecotone(Walsh et al.1994).Geomorphic control on vegetation was already demonstrated by studies such as Walsh et al.(2003)and Butler et al.(2007).Therefore,in this study I have considered mostly parameters related to topography for creating the hybrid cartographic model to predict forest-line expansion.

Analysis of climate records and climate model prediction results have indicated temperature increase trend throughout Nepal.This temperature rise can trigger the upslope expansion of sub-alpine forest.However,our study suggested topography related variables(aspect,slope,snow cover,and solar radiation)will prevent upslope forest expansion in most of the area of MBNP.

The slope is one of most important topographic variables controlling forest distribution in high elevation area.Slope angle regulates the forest distribution by limiting soil formation,and slope with angles greater than 34°often have too little soil to support forest growth.Rock outcrops and landslide paths will also not support forest establishment.Slope also controls solar radiation,stability,erosion,and moisture availability(Bader and Ruijten 2008).These parameters associated with the slope directly or indirectly control the spatial pattern of forests in the sub-alpine environment.The slope map of the MBNP(Fig.3)indicates that most of the park consists of slopes with angles greater than 34°and have low vegetation cover.The structure and composition of the forest line are also in fluenced by aspect because south-facing slopes receive more solar radiation than north-facing slopes in the northern hemisphere.In the Himalaya,the natural forest line is at higher elevations on south-facing slopes as compared to north-facing ones(Schickhoff 2005).Sometimes the difference in altitudinal position is up to several hundred meters(Schickhoff 2005).In this study,I recorded upper forest lines on the south-and north-facing slopes at 4221 m and 3985 m,respectively,which is a difference of almost 250 m.Other studies such as Paulsen and K?rner(2001)and Wang et al.(2013)did not observe any in fluence of slope exposure in their study areas.Use of remote sensing or satellite imagery for mapping areas,such as the Himalayas,is in fluenced by complex terrain.Topographic in fluence in remote-sensing images must be corrected before such images are used for forest-line mapping,which will reduce shadowing in the image.For this study,I tried to remove topographic effects through the use of a highquality DEM obtained from the USGS.However,due to the rugged topography of the area,topographic effects could not be removed completely.This might be the reason for the observed slope-exposure effect in our study.The use of terrain-corrected and high-resolution imagery can help minimize this issue.Chhetri et al.(2017)used highresolution(0.5 m spatial resolution)Digital Globe imagery and found no slope exposure in the Barun valley of Makalu Barun National Park,eastern Nepal.

Besides slope and aspect,forest line advances will depend on moisture availability,growing-season temperature,and microsite availability for seedling establishment.Also,the predicted suitable area must be absent of snow cover and snow avalanche-related disturbances because forest development and the spatial pattern of subalpine forest areas worldwide,would be affected,especially near their upper limit(Kajimoto et al.2004).

Traditionally,habitat-suitability models have been used to identify the current habitats of species or the environmental niches in which any particular species can occur(Elith and Leathwick 2009;Kearney and Porter 2009).Here,I have demonstrated the usability of a hybrid cartographic model for predicting how the present habitat would change in the future under a climate-change scenario.I have used five environmental variables(elevation,slope,NDVI,NDWI,and hillshade)that have direct physiological impact on plant communities to create the site suitability map.

For example,if climate became favorable for tree encroachment in the future,the forest line would only advance in 10%of the park area.However,the upward shift of tree-line species in these areas will depend on various factors in addition to the availability of a suitable area.These include:seed dispersal,a favorable climate,competition with other species,and the absence of human disturbances.Even though the model predicts that 10%of the area above the current forest line limit is available for future expansion,there is only a small chance that forest will occupy all of this available area under future climate-change scenarios.

The remaining 6%of the area available for forest-line expansion—as predicted by the model—is below the current forest line,where trees will probably expand into previously deforested areas.Harsch and HilleRisLambers(2016)found that a shift in forest distribution in response to climate change could occurupward ordownward.Accordingly,there is a chance that the A.spectabilis forest in the MBNP could move to previously unoccupied lower elevations.Conservationists and park managers can use this information to determine where the forest can advance and whether this area would overlap with the habitat of endangered species.This will help to take the necessary steps to conserve the habitat of species,such as the snow leopard and caterpillar fungi.The hybrid cartographic model can also be used to study the range shift of a species under environmental change conditions.The approach used in this study will help to address the issue of colonization of the alpine environment by the upward movement of thermophilic species by identifying the encroachment areas(Carlson et al.2013).

Conclusions

The hybrid cartographic model,which consists of the five topographic variables presented here,depicts areas above the current forest line that are available for forest encroachment.Most of this is unsuitable growing substrate.The inclusion of additional topographic variables and climatic variables—such as mean temperature,minimum temperature,growing season temperature,and precipitation in the model—will provide more reliable results.The use of a high-resolution digital elevation model and high-resolution satellite imagery could improve the accuracy of the model.This type of hybrid cartographic model can be utilized for habitat management of endangered species by park managers and conservationists.

Bader MY,Ruijten JJ(2008)A topography-based model of forest cover at the alpine tree line in the tropical Andes.J Biogeogr 35(4):711–723

Boulangeat I,Philippe P,Abdulhak S,Douzet R,Garraud L,Lavergne S,Lavorel S,Van Es J,Vittoz P,Thuiller W(2012)Improving plant functional groups for dynamic models of biodiversity:at the crossroads between functional and community ecology.Glob Change Biol 18(11):3464–3475

Brown DG(1994)Predicting vegetation types at treeline using topography and biophysical disturbance variables.J Veg Sci 5(5):641–656

Butler DR,Malanson GP,Walsh SJ,Fagre DB(2007)In fluences of geomorphology and geology on alpine treeline in the American West—more important than climatic in fluences?Phys Geogr 28(5):434–450

Byers AC(1996)Historical and contemporary human disturbance in the Upper Barun Valley,Makalu-Barun National Park and Conservation Area,East Nepal.Mt Res Dev 16(3):235–247

Carlson BZ,Randin CF,Boulangeat I,Lavergne S,Thuiller W,Choler P(2013)Working toward integrated models of alpine plant distribution.Alp Bot 123(2):41–53

Carpenter C,Zomer R(1996)Forest ecology of the Makalu Barun National Park and Conservation Area,Nepal.Mt Res Dev 16(2):135–148

Chander G,Markham BL,Helder DL(2009)Summary of current radiometric calibration coef ficients for Landsat MSS,TM,ETM+,and EO-1 ALI sensors.J Remote Sens Environ 113(5):893–903

Chaplin J,Brabyn L(2013)Using remote sensing and GIS to investigate the impacts of tourism on forest cover in the Annapurna Conservation Area,Nepal.Appl Geogr 43:159–168

Chhetri PK,Cairns DM(2015)Contemporary and historic population structure of Abies spectabilis at treeline in Barun valley,eastern Nepal Himalaya.J Mt Sci 12(3):558–570

Chhetri PK,Cairns DM(2016)Dendroclimatic response of Abies spectabilis at treeline ecotone of Barun Valley,eastern Nepal Himalaya.J For Res 27(5):1163–1170

Chhetri PK,Shrestha KB,Cairns DM(2017)Topography and human disturbances are major controlling factor in treeline pattern at Barun and Manang area in the Nepal Himalaya.J Mt Sci 14(1):119–127

Colby JD(1991)Topographic normalization in rugged terrain.Photogramm Eng Remote Sens 57(5):531–537

Coops NC,Morsdorf F,Schaepman ME,Zimmermann NE(2013)Characterization on an alpine treeline using airborne LIDAR data and physiological modeling. Glob Change Biol 19(12):3808–3821

Danzeglocke J(2005)Remote sensing of upper timberline elevation in the Alps on different scales.In:Oluic(ed.)New strategies for European remote sensing.MillPress,Rosterdam

Devi N,Hagedorn F,Moiseev P,Bugmann H,Shiyatov S,Mazepa V,Rigling A(2008)Expanding forests and changing growth forms of Siberian larch at the Polar Urals treeline during the 20th century.Glob Change Biol 14(7):1581–1591

Eckert S,Kellenberger T,Itten K(2005)Accuracy assessment of automatically derived digital elevation models from ASTER data in mountainous terrain.Int J Remote Sens 26(9):1943–1957

Elith J,Leathwick JR(2009)Species distribution models:ecological explanation and prediction across space and time.Annu Rev Ecol Evol Syst 40:677–697

Fang KY,Wilmking M,Davi N,Zhou FF,Liu CZ(2014)An ensemble weighting approach for dendroclimatology:drought Reconstructions for the Northeastern Tibetan Plateau.PLoS ONE 9(1):e86689

Forkuor G,Maathuis B(2012)Comparison of SRTM and ASTER Derived Digital Elevation Models over Two Regions in Ghana-Implications for Hydrological and Environmental Modeling.Studies on Environmental and Applied Geomorphology,InTech,pp 219–240

Forrest JL,Wikramanayake E,Shrestha R,Areendran G,Gyeltshen K,Maheshwari A,Thapa K(2012)Conservation and climate change:assessing the vulnerability of snow leopard habitat to treeline shift in the Himalaya.Biol Cons 150(1):129–135

Gaire NP,Koirala M,Bhuju DR,Borgaonkar HP(2014)Treeline dynamics with climate change at the central Nepal Himalaya.Clim Past 10(4):1277–1290

Harsch MA,HilleRisLambers J(2016)Climate warming and seasonal precipitation change interact to limit species distribution shifts across Western North America.PLoS ONE 11(7):e0159184

Hirzel AH,Lay GL,Helfer V,Randin C,Guisan A(2006)Evaluating the ability of habitat suitability models to predict species presences.Ecol Modeling 199(2):142–152

Jenness J(2013)DEM Surface Tools.Jenness Enterprises.Available at:http://www.jennessent.com/arcgis/surface_area.htm

Kajimoto T,Daimaru H,Okamoto T,Otani T,Onodera H(2004)Effects of snow avalanche disturbance on regeneration of subalpine Abies mariesii forest,northern Japan.Arct Antarct Alp Res 36(4):436–445

Kattel DB,Yao TD,Yang K,Tian L,Yang G,Joswiak D(2013)Temperature lapse rate in complex mountain terrain on the southern slope of the central Himalayas.Theoret Appl Climatol 113(3–4):671–682

Kearney M,Porter W(2009)Mechanistic niche modelling:combining physiological and spatial data to predict species ranges.Ecol Lett 12(4):334–350

K?rner C(1998)A re-assessment of high elevation treeline positions and their explanation.Oecologia 115(4):445–459

K?rner C(2012)Alpine treelines:functional ecology of the global high elevation tree limits.Springer,Basel

Lauver CL,Busby WH,Whistler JL(2002)Testing a GIS model of habitat suitability for a declining grassland bird.Environ Manage 30(1):88–97

Malanson GP(2001)Complex responses to global change at alpine treeline.Phys Geogr 22(4):333–342

McFeeters SK(1996)The use of Normalized Difference Water Index(NDWI)in the delineation of open water features.Int J Remote Sens 17(7):1425–1432

McFeeters SK(2013)Using the Normalized Difference Water Index(NDWI)within a geographic information system to detect swimming pools for mosquito abatement:a practical approach.Remote Sens.5(7):3544–3561

Panigrahy S,Anitha D,Kimothi MM,Singh SP(2010)Timberline change detection using topographic map and satellite imagery.Trop Ecol 51(1):87–91

Paulsen J,K?rner C(2001)GIS-analysis of tree-line elevation in the Swiss Alps suggests no exposure effect.J Veg Sci 12(6):817–824

Pierce KB Jr,Lookingbill T,Urban D(2005)A simple method for estimating potential relative radiation(PRR)for landscape-scale vegetation analysis.Landscape Ecol 20(2):137–147

Randin CF,Vuissoz G,Liston GE,Vittoz P,Guisan A(2009)Introduction of snow and geomorphic disturbance variables into predictive models of alpine plant distribution in the Western Swiss Alps.Arct Antarct Alp Res 41(3):347–361

Schickhoff U(2005)The upper timberline in the Himalayas,Hindu Kush and Karakorum:a review of geographical and ecological aspects.Mountain ecosystems.Springer,Berlin Heidelberg,pp 275–354

Shrestha KB,Hofgaard A,Vandvik V(2015)Recent treeline dynamics are similar between dry and mesic areas of Nepal,central Himalaya.J Plant Ecol 8(4):347–358

Singh PC,Panigrayh S,Parihar JS,Dharaiya N(2013)Modeling environmental niche of Himalayan birch and remote sensing based vicarious validation.Trop Ecol 54(3):321–329

Vittoz P,Rulence B,Largey T,Frelechoux F(2008)Effects of climate and land-use change on the establishment and growth of Cembran Pine(Pinus cembra L.)over the altitudinal treeline ecotone in the central Swiss Alps.Arct Antarct Alp Res 40(1):225–232

Walsh SJ,Kelly NM(1990)Treeline migration and terrain variability:integration of remote sensing digital enhancements and digital elevation models.Pap Proc Appl Geogr Conf 13:293–304

Walsh SJ,Butler DR,Brown DG,Bian L(1994)Form and pattern of alpine environments:an integrative approach to spatial analysis and modelling in Glacier National Park,U.S.A.In:Heywood DI,Price MF(eds)Mountain environments and GIS.Taylor and Francis,London,pp 189–216

WalshSJ,ButlerDR,MalansonGP,Crews-MeyerKA,MessinaJP,Xiao NC(2003)Mapping,modeling,andvisualizationofthein fluenceof geomorphic processes on the alpine treeline ecotone,Glacier National Park,MT,USA.Geomorphology 53(1–2):129–145

Wang WL,K?rner C,Zhang ZM,Wu RD,Geng YP,Shi W,Ou XK(2013)No slope exposure effect on alpine treeline position in the three parallel rivers region,SW China.Alp Bot 123(2):87–95

Xu JC,Grumbine RE,Shrestha A,Eriksson M,Yang XF,Wang Y,Wilkes A(2009)The melting Himalayas:cascading effects of climate change on water,biodiversity,and livelihoods.Conserv Biol 23(3):520–530