Ecological Engineering Predicting distribution of ...

2 downloads 0 Views 8MB Size Report
et al., 2012; Manish et al., 2016; Ranjitkar et al., 2014; Telwala et al.,. 2013). However, more ...... Rawat, P.K., Tiwari, P.C., Pant, C.C., 2012. Climate change ...
Ecological Engineering 97 (2016) 593–609

Contents lists available at ScienceDirect

Ecological Engineering journal homepage: www.elsevier.com/locate/ecoleng

Predicting distribution of major forest tree species to potential impacts of climate change in the central Himalayan region Anusheema Chakraborty a,∗ , P.K. Joshi b , Kamna Sachdeva a a b

Department of Natural Resources, TERI University, New Delhi 110070, India School of Environmental Sciences, Jawaharlal Nehru University, New Delhi 110067, India

a r t i c l e

i n f o

Article history: Received 28 December 2015 Received in revised form 11 July 2016 Accepted 5 October 2016 Keywords: Niche modelling Habitat suitability MaxEnt Climate change Himalaya

a b s t r a c t Predicting climatic niche of species and projecting their potential range shifts in geographic distribution under future climate scenarios is essential for assessing impacts of climate change. Ecological nichebased models are widely used to map habitat suitability of current and future potential distribution of species, using precise coordinates of species occurrences, along with climatic and various environmental variables. Despite the importance of high dependence on forest resources in the Himalayan region, the direct impacts of climate change on major forest tree species is not well-documented. In the present study, we used MaxEnt (or maximum entropy) modelling to predict current distribution and changes in the future distributions of four ecologically and economically dominant forest tree species (Quercus leucotrichophora, Q. semecarpifolia, Q. floribunda, and Pinus roxburghii) in the central Himalayan region. Future predictions were based on representative concentration pathways (RCPs) for two time periods (2050s and 2070s). We demonstrated the use of MaxEnt by combining different climatic, geomorphologic, and pedologic variables as predictor variables to model the potential climatic niches. We evaluated the model performance with an average AUC value varying as 0.809 (±0.020), 0.982 (±0.008), 0.966 (±0.006), and 0.803 (±0.025) for Q. leucotrichophora, Q. semecarpifolia, Q. floribunda and P. roxburghii, respectively. Depending upon the RCPs, the results show both increase and decrease in suitable habitat range of these species across all future climate scenarios. The shifts in geographic distributions of climatic niches show unusual patterns, implying the need for urgent adaptive forest management strategies. Our approach can be used as a baseline database for broad-scale applicability in forest tree species restoration and conservation planning. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Despite great uncertainty surrounding climate change-induced impacts, global estimates show marked influences on species extinction rates and distribution patterns, vegetation phenology, and ecosystem structure and composition (Chen et al., 2011; Jewitt et al., 2015; Mantyka-Pringle et al., 2015; Trumbore et al., 2015; Urban, 2015). Associated with climate change, several other concurrent stressors, such as invasive species, degradation, over-exploitation, pollution and plant diseases, which either act independently or in combination, further aggravate the impacts due to climate change (Lewis et al., 2015; Mantyka-Pringle et al., 2015). Over the next century, with expected changes in projected global climate as a result of increasing atmospheric CO2 levels and

∗ Corresponding author. E-mail address: [email protected] (A. Chakraborty). http://dx.doi.org/10.1016/j.ecoleng.2016.10.006 0925-8574/© 2016 Elsevier B.V. All rights reserved.

other greenhouse gases (GHGs) (Friend et al., 2014), it is anticipated that it will lead to loss of individual species with substantial changes in their optimal conditions for growth and survival (Dirnböck et al., 2011; Gallagher et al., 2013), especially as climate change progresses towards the extremes. In plant ecology, one of the oldest observations is the relationship between geographic patterns of species and climate (Chakraborty et al., 2013; Reu et al., 2011), with climate acting as a primary factor which regulates spatial distribution patterns of many tree species (Woodward and Williams, 1987). Most forest tree species are particularly sensitive to climate change (Hansen and Phillips, 2015), and are adapted to a range of climatic conditions, which is referred to as their climatic niche (Pearson and Dawson, 2003; Peterson, 2011). Due to the long life-span of tree species as well as their slow migration rates (Lindner et al., 2010; Pearson, 2006; Zhu et al., 2012), unprecedented rapid climate change will not allow immediate adaptation to newer climatic conditions in their current locations (Aitken et al., 2008). In such cases,

594

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

species exposed to climate conditions outside their climatic niches will likely have three possible fates: migrating to other locations by tracking their ecological niches spatially, adapting to different conditions in current locations, or eventually leading to local extinction. This will invariably result in profound shifts in the distribution and abundance of species making them highly vulnerable, with range of possible impacts so far that have been underestimated (Lindner et al., 2014; Moritz and Agudo, 2013). Therefore, understanding the spatial distribution of climatic niches of forest tree species and projecting their potential range shifts for future climate change scenarios is important for assessing their vulnerability so as to develop appropriate adaptive forest management strategies under a rapidly changing climate, including assisted migration (Koralewski et al., 2015). Hutchinson’s (1957) fundamental niche defines ‘climatic niche’ as the climatic component within which species survive and grow, constituting a bioclimatic envelope which has its foundations in ecological niche theory (Pearson and Dawson, 2003). Most bioclimatic models are based on empirical relationships between observed species distribution and environmental variables (Araújo and Williams, 2000; Elith et al., 2006; Guisan and Zimmermann, 2000; Miller et al., 2007; Peterson, 2003). Such correlative models represent the realized niche since observed species’ distributions are constrained by non-climatic factors, including biotic interactions (Pearson and Dawson, 2003). On the contrary, other bioclimatic models look for physiological representation based on mechanistic relationship between climate parameters and species responses (González-Moreno et al., 2015; Kearney and Porter, 2009), thereby, identifying the fundamental niche by modelling physiological limiting mechanisms in a given environment or geographic space (Monahan, 2009). The crucial issue in ecological niche-based modelling is the selection of explanatory variables to create species-environment profile that supposedly predicts distribution and abundance of species (Dormann, 2007). Nonetheless, since the causal mechanism of geographic distribution of species is not readily quantifiable, we often resort to substitute and proxies (Minor and Urban, 2007). With poorly studied taxa, correlative models are more advantageous since they require little knowledge of mechanistic links between species and its surrounding environment (Kearney and Porter, 2009). This proves to beneficial given the paucity in the amount of data available in some regions (BarbetMassin et al., 2012). Scientific literature using niche models has had an overwhelming success in the recent past, while debate about usefulness of the approach has also followed (Araújo and Peterson, 2012; Wiens et al., 2009). In general, a ‘climatic niche’ model rather predicts a suitable habitat of the species, than its actual distribution, which may involve series of evolutionary and ecological processes (Aitken et al., 2008). The Himalayan forests are undergoing changes in their distribution patterns, and are potentially expected to show shifts in the forest-cover boundaries (Alekhya et al., 2015; Chakraborty et al., 2013; Chaturvedi et al., 2011; Gopalakrishnan et al., 2011; Joshi et al., 2012; Manish et al., 2016; Ranjitkar et al., 2014; Telwala et al., 2013). However, more comprehensive detailed assessments of climatic niches of major forest tree species and their potential impacts in the face of climate change are required (Gairola et al., 2013). The need for this information in the Himalayan region is very important since it will allow forest-dependent communities, forest managers and policy makers to assess vulnerability and climate change impact for species adaptation and conservation (Keenan, 2015). Since regional climate models show that temperature and precipitation in the Himalayan region is likely to continue to increase in future (Kulkarni et al., 2013), the knowledge on potential impacts of climate change on forest tree species should be properly established and interpreted. In the light of these considerations, we provide a scientific basis with the application of correlative MaxEnt

(or maximum entropy) model to map the current potential distribution of four forest tree species (Quercus leucotrichophora, Q. semecarpifolia, Q. floribunda, and Pinus roxburghii) and to predict changes in the potential distribution of these species under future climate change scenarios based on representative concentration pathways (RCPs) in the central Himalayan region, for two time periods (2050s and 2070s). We assume that both the extent of suitable climatic habitat and range of the tree species would possibly alter with changes in future climate of the study area.

2. Study area The study area is located in the Kumaon division in the state of Uttarakhand, falling in the central Himalayan region. It has a total area of approximately 20,397 km2 and is mostly mountainous consisting of a forest-dominated landscape. The region varies between east longitudes of 80◦ 10 to 79◦ 27 and north latitudes of 30◦ 48 to 28◦ 52 , with an altitude ranging from 157 m at the foot hills to 6980 m at the plateau. This region is bounded on the north by China, on the east by Nepal, on the south by the state of Uttar Pradesh, and on the west by the Garhwal division. Due to the steep altitudinal gradient from south to north, there is significant diversity in the natural vegetation of this region (Singh and Mal, 2014). Depending upon elevation, temperature regime in this area may resemble that of temperate region (above 2000 m elevation) or of tundra region (alpine belt) (Singh and Singh, 1992). Literature indicates forests in upper areas of western and central Himalayan region being vulnerable to projected impacts of climate change, while forests in eastern Himalayan region are comparatively more resilient (Chaturvedi et al., 2011; Gopalakrishnan et al., 2011; Joshi et al., 2012; Shrestha et al., 2012). This highlights the sensitivity of different vegetation types in eastern Himalayan region as relatively stable to both temperature and precipitation variables, which is not the case for western and central Himalayan region. The geographically dominant forest types include Upper or Himalayan chir pine forest (9/C1b), Ban oak forest (12/C1a), and moist Siwalik sal forest (3C/C2a). Quercus spp. is a large genus with many species, wherein the Himalayan region itself is represented by 35 species. The spatial distribution of Quercus spp. is mostly between 1000 m and 3600 m, above mean sea level. In the central Himalayan region, which includes the Kumaon division in the state of Uttarakhand, only five species of oak can be found (Singh et al., 2012). However, given the geographical dominance along with ecological and economic importance, we chose four major tree species, Q. leucotrichophora, Q. semecarpifolia, Q. floribunda, and P. roxburghii for our study. Locally known as banj or ban oak, Q. leucotrichophora, is a valuable keystone species with great societal relevance in the central Himalayan region. It is among the main forest-forming species in the densely populated mid-altitudinal zones and provides a variety of ecosystem services (Singh et al., 2014). Q. floribunda (locally known as moru oak) is another dominant oak species found between 1700 m and 2600 m elevation range forming the evergreen climax forests (Negi et al., 1996). The other important oak species, Q. semecarpifolia (locally known as kharsu oak), occupies area less than approximately 350 km2 , and has an elevation range higher than all the other evergreen oak forests in Himalaya. Nonetheless, its canopy is often severely disturbed by cutting and lopping (Singh et al., 1997; Vetaas, 2000). On the other hand, P. roxburghii occurs extensively in the low-tomid montane belt of central and western Himalayan region. Locally known as chir pine (or simply, pine), it is a gregarious, fire-resistant, indigenous tree species, often forms pure forest stands. Its capacity to rapidly colonize degraded habitats and high volume returns make this species a precious resource in the region (Chaturvedi and Singh, 1987).

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

595

Table 1 Climatic, geomorphologic, and pedologic variables used for modelling climatic niches. Type

Variable Name

Code

Data Source

Resolution

Climatic

Annual Mean Temperature Mean Diurnal Range Isothermality (BIO 2/BIO 7) (×100) Temperature Seasonality (Standard Deviation×100) Max Temperature of Warmest Month Min Temperature of Coldest Month Temperature Annual Range (BIO 5-BIO 6) Mean Temperature of Wettest Quarter Mean Temperature of Driest Quarter Mean Temperature of Warmest Quarter Mean Temperature of Coldest Quarter Annual Precipitation Precipitation of Wettest Month Precipitation of Driest Month Precipitation Seasonality (Coefficient of Variation) Precipitation of Wettest Quarter Precipitation of Driest Quarter Precipitation of Warmest Quarter Precipitation of Coldest Quarter Elevation Slope Aspect Solar radiation data (annual) Soil type

BIO 1 BIO 2 BIO 3 BIO 4 BIO 5 BIO 6 BIO 7 BIO 8 BIO 9 BIO 10 BIO 11 BIO 12 BIO 13 BIO 14 BIO 15 BIO 16 BIO 17 BIO 18 BIO 19 Elev Slope Aspect DNI Soil

WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim WorldClim ASTER-GDEM ASTER-GDEM ASTER-GDEM MNRE, GoI MAFW, GoI

30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 arc-seconds 30 m 30 m 30 m 10 km 1:50000 scale

Geomorphologic

Pedologic

3. Materials and methods 3.1. Species occurrences data The species occurrence data used for this study were collected from our own field surveys conducted in six districts (Almora, Bageshwar, Champawat, Nainital, Pithoragarh, and Udham Singh Nagar) of the Kumaon division in the state of Uttarakhand between 2014 and 2015. Owing to the vast geographical area, it wasn’t possible to cover the entire region with extensive field visits to account for many field occurrence points, which was essentially due to inaccessibility in some areas. Therefore, we relied on the forest type map provided by the Uttarakhand Forest Department, which maps different forest types in the Uttarakhand state based on Champion and Seth (1968) classification scheme. Altogether, 500 species occurrence points distributed in the study region were used to model the potential distribution for Q. leucotrichophora (ban oak) species. Similarly, 100 species occurrence points for Q. semecarpifolia (kharsu oak), 150 species occurrence points for Q. floribunda (moru oak) and 500 species occurrence points for P. roxburghii (pine) were used in the study. To minimize any error due to these additional reference points, species occurrences points were further validated through visual interpretation of available Google Earth® images. 3.2. Climate and other environmental data To model potential distribution of major tree species across our study area, we first collected nineteen (19) bioclimatic layers, three topographic layers, annual solar radiation data, and soil type database (Table 1). We obtained grid-based bioclimatic layers from the WorldClim database (Hijmans et al., 2005) at 30 arc-seconds spatial resolution to represent current climatic condition. Bioclimatic variables are derived from monthly temperature and rainfall values in order to generate biologically meaningful variables which define annual trends, seasonality and extremes of temperature and rainfall (Heikkinen et al., 2006). We used Digital Elevation Model (DEM) from the Global DEM (GDEM) product generated from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) dataset (Version 2) available at 30 m spatial resolution (Yamaguchi et al., 1998). We extracted three features from DEM: (i)

elevation, (ii) slope, and (iii) aspect, as the predictor variables. The estimate of annual average direct normal irradiance (DNI) at 10 km spatial resolution is based on hourly estimates of radiation over 10 years (2002–2011). This was obtained from the Solar Energy Centre, Ministry of New and Renewable Energy (MNRE), Government of India (GoI). As a function of pedologic data, the soil type map available at 1:50000 scale in vector format was acquired from the Ministry of Agriculture & Farmers Welfare (MAFW), Government of India (GoI). These variables, or similar to these variables, have already been used successfully in many species modelling studies conducted elsewhere (Chitale et al., 2014). To determine the potential future distribution of the species under different climate scenarios, we used datasets of future climate using the RCPs of four GHGs concentration trajectories adopted by the Intergovernmental Panel on Climate Change (IPCC) for its fifth Assessment Report (AR5). RCPs, namely, RCP 2.6, RCP 4.5, RCP 6.0, and RCP 8.5 were used, both available for near-term (2050s) and long-term (2070s) future climate projections. Predictions from the Hadley Centre Global Environment Model version 2-Earth System (HadGEM2-ES) climate model was used for all four RCPs. HadGEM2-ES is an Earth system model, and it is preferred over other climate models since it simulates enhanced climateinduced ecosystem and hydrological processes (Ahlström et al., 2012; Betts et al., 2015; Jones et al., 2011). This model was developed by the Met Office, Hadley Centre for the Coupled Model Intercomparison Project (CMIP5) (Collins et al., 2011) centennial experiments including ensembles of simulations of the RCPs. RCP 2.6 is the lowest GHG concentration pathway in which radioactive forcing (global energy imbalances) levels reach 3.1 W/m2 by mid-century and drop upto 2.6 W/m2 by 2100, RCP 4.5 is the concentration pathway in which total radiative forcing reaches to 4.5 W/m2 by 2100 and stabilizes due to employment of a range of technologies and strategies for reducing GHG emissions. Likewise, RCP 6.0 also represents stabilization by 2100, this time at 6.0 W/m2 by 2100, and RCP 8.5 shows rising radiative forcing pathway leading to 8.5 W/m2 by 2100 (Moss et al., 2010; Vuuren et al., 2011). 3.3. Methodology We used freely available MaxEnt software to model the potential current and future distributions of major forest tree species in

596

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

Table 2 General statistics showing major bioclimatic profiles of forest tree species based on the occurrences data used in the study. Variable Name

Mean Diurnal Range (◦ C) Isothermality (Dimensionless) Temperature Seasonality (◦ C) Temperature Annual Range (◦ C) Annual Precipitation (mm) Precipitation of Driest Quarter (mm) Precipitation of Warmest Quarter (mm) Precipitation of Coldest Quarter (mm)

Ban

Kharsu

Moru

Pine

Mean

SD

Min.

Max.

Mean

SD

Min.

Max.

Mean

SD

Min.

Max.

Mean

SD

Min.

Max.

10.3 0.41 47.3 24.6 1675 97 726 144

0.9 0.02 1.6 1.3 186 9 91 19

8.1 0.37 44.6 21.4 1207 69 524 96

11.9 0.45 51.8 27.7 2202 118 996 222

9.4 0.40 48.8 23.3 1388 106 696 202

0.3 0.01 1.0 0.4 122 12 79 26

8.3 0.37 44.8 21.9 1156 88 523 150

9.9 0.41 51.4 24.1 1805 129 797 256

9.6 0.41 47.6 23.3 1475 101 634 166

0.5 0.01 1.1 0.6 116 7 79 21

8.5 0.37 44.6 21.9 1209 87 520 135

10.9 0.43 49.4 25.0 1909 125 849 250

10.5 0.41 47.4 25.1 1654 97 714 140

0.7 0.01 1.6 1.2 168 7 86 14

8.6 0.38 44.7 22.5 1305 70 533 102

11.9 0.44 53.1 28.2 2277 116 1036 222

the central Himalayan region. MaxEnt is a broader-scale machine learning approach which estimates the probability distribution of species occurrences based on environmental conditions of a location in which presence of the species is found by calculating distribution of maximum entropy, i.e. the most spread out distribution in space for a given set of constraints (climate and other environmental variables) (Elith et al., 2011). MaxEnt outperforms other methods by showing higher predictive accuracy and has an easy to use, user-friendly interface (Bowler, 2014; Elith et al., 2006; Warren and Seifert, 2011). MaxEnt has been particularly used to estimate potential range shifts of species due to climate change as well (Heikkinen et al., 2006; Hijmans and Graham, 2006; Wiens et al., 2009). With limited data availability, it is of particular advantage since MaxEnt uses presence-only records with continuous and categorical variables (Warren and Seifert, 2011), which makes its usage a much easier functionality, despite few assumptions that it considers in predicting climatic niches. 3.3.1. Selecting predictor variables We checked all predictor variables for high cross-correlations using Pearson’s correlation coefficient (r ≥ 0.80 or r ≤ −0.80). We included only one variable from a set of highly correlated variables which was biologically relevant to the species, to reduce problems due to multi-collinearity (Dormann et al., 2013). In the end, we used thirteen (13) layers from the set of twenty four (24) predictor variables, namely, BIO 2, BIO 3, BIO 4, BIO 7, BIO 12, BIO 17, BIO 18, BIO 19, elevation, slope, aspect, solar radiation and soil type. We used the eight bioclimatic variables for modelling current distribution of the species (BIO 2, BIO 3, BIO 4, BIO 7, BIO 12, BIO 17, BIO 18, BIO 19) along with other environmental variables. Similarly, for consistency, we used future projections of the eight bioclimatic variables (RCPs for two time periods), while keeping the other geomorphologic and pedologic layers constant (elevation, slope, aspect, solar radiation and soil type). Table 2 lists the major bioclimatic profiles of the species occurrence data used in the study. 3.3.2. Ecological niche-based model Several notable attempts have been made by using various statistical models to predict the spatial distribution of species. For example, it includes classical regression methods such as logistic regression (Pearce and Ferrier, 2000), resource selection functions (Johnson et al., 2004), Bayesian approach (Convertino et al., 2011a), generalized additive models (GAMs) (Drexler and Ainsworth, 2013), generalized linear models (GLMs) and generalized linear mixed models (GLMMs) (Swanson et al., 2013), among the few. On the other hand, species distribution modelling based on machine learning algorithms have become increasingly popular in the recent years, some of them include models based on artificial neural networks (Pearson et al., 2002), ecological niche factor analysis (ENFA) (Hirzel et al., 2002), maximum entropy (MaxEnt) (Phillips et al., 2006), classification and regression trees (CART)

(Vayssières et al., 2000), boosted regression trees (BRT) and random forests (RF) (Mainali et al., 2015b). Of these wide arrays of methods available, MaxEnt is often described as the most popular method as it easily handles complex interactions between predictor and response variables (Elith et al., 2011), and is often less sensitive to small sample sizes (Wisz et al., 2008). MaxEnt’s user-interface model is known for its simplicity and easier implementation, whilst typically outperforming other methods based on predictive accuracy (Merow et al., 2013); thereby, making it the most used ecological niche-based model. Based on presence records, MaxEnt estimates the probability of presence of any species by randomly generating background points by finding the maximum entropy of the species distribution (Phillips et al., 2006). By using regularization parameters, it controls over-fitting and handles both categorical and continuous variables. MaxEnt uses six different features classes (linear, quadratic, product, threshold, and hinge and category indicator) that constrain the geographical distribution of a species (Elith et al., 2011), describing the species’ responses to environmental conditions. This helps in accounting for different types of input variables available that can be used in modelling the MaxEnt predictions (Phillips and Dudík, 2008). Depending on amount of adequate species’ data available, MaxEnt’s default approach allows all feature types in the model runs, which can be modified subjected to user’s preferences (Elith et al., 2011). Further details are provided in Phillips and Dudík (2008) and Elith et al. (2011). The generated output from MaxEnt estimates the habitat suitability (which is representative of a climatic niche) for species that generally varies from 0 (lowest) to 1 (highest) (Kumar et al., 2014).

3.3.3. Model implementation Ideally, one should test different settings in MaxEnt by varying regularization parameters, number of iterations and feature types. In our case, however, default settings in MaxEnt yielded the best results given the actual presence of the particular tree species being modelled. By default, the MaxEnt model uses original presence data (input) against 10,000 random background points (pseudoabsences) for modelling the habitat distribution. We imported the MaxEnt output data and three arbitrary categories were reclassified into classes of habitat suitability: low (25–50% probability of occurrence), medium (50–75% probability of occurrence) and high (75% probability of occurrence) by omitting the values below 25% as non-suitable habitat based on the logistic threshold. Area under the receiver-operator curve (AUC) is used as a threshold, as an independent measure of predictive accuracy in the MaxEnt literature (Merow et al., 2013). It is based only on the ranking of locations and is interpreted as the probability that a randomly chosen presence location is ranked higher than a randomly chosen background point (Anderson and Gonzalez, 2011). The model performance and robustness is commonly evaluated by AUC values that range from 0 to 1, where AUC values between 0.5–0.7

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

597

Fig. 1. The receiver operating characteristic (ROC) curve for: (a) Q. leucotrichophora (ban oak), (b) Q. semecarpifolia (kharsu oak), (c) Q. floribunda (moru oak), and (d) P. roxburghii (pine). Black curves show the mean response and grey margins are +/− one standard deviation calculated over 10 replicates.

are considered low, 0.7–0.9 are moderate and above 0.9 are high values (Shrestha and Bawa, 2014). We used AUC values to determine our model’s performance accuracy. Although, it is important to use more than one metric to assess model performances (Elith and Graham, 2009), but since we are using only MaxEnt model for predicting different species, we used the frequently applied AUC metric (Syfert et al., 2013) for evaluating distribution discrimination ability. In addition to this, we used the Jackknife method to assess the importance of variables in the final model (Phillips et al., 2006). In case of small number of occurrence records, Jackknife tests are often considered useful for their high predictive ability to guide field surveys (Pearson et al., 2007). It is mostly known for methodical assessment of variable’s importance based on dropping one variable at a time to determine the respective variable’s importance in the model (Rengstorf et al., 2013; Shcheglovitova and Anderson, 2013). This one-factor-at-a-time (OAT) method is usually reliable if only the input factors used are not highly interactive (Convertino et al., 2014). We employed the MaxEnt software version 3.3.3k, which runs on the Java runtime environment (JRE) (available from https:// www.cs.princeton.edu/∼schapire/maxent/) for current potential habitat distribution of species. We tested correlation coefficients among predictor variables used in the study with ENMTools v.1.3 (Warren et al., 2010). For future potential distributions of tree species, we fitted MaxEnt model for each species with default settings for the MaxEnt function from the ‘dismo’ package (Hijmans et al., 2015) within R (version 3.1.2, 32-bit) open-source statistical software (R Development Core Team, 2011). ESRI’s ArcGIS 10 software was used for spatial analysis of outputs produced from the aforementioned software for further analysis.

4. Results Some of the ecological niche-based models predict species distribution patterns based on presence and absence occurrence points, while other models perform with presence-only occurrence points. Often comparisons between two different scenarios (climate change or otherwise) are conducted using the same species occurrence database along with environmental predictors, and different climatic variables. However, it should be noted that often these models perform best only under specific circumstances (Harte and Newman, 2014). Their applicability should be considered with more extensive field-based methodologies, so as to use these predictions and their derived database through a realistic approach while considering different mechanisms in ecology.

4.1. Model performance The probability of occurrence for the tree species is not randomly distributed, with regions of higher occurrence values specific to certain locations in the study area. Nonetheless, Q. leucotrichophora (ban oak) and P. roxburghii (pine) have a more dominated presence in this region, which also leads to low AUC values in comparison to Q, semecarpifolia (kharsu oak) and Q. floribunda (moru oak) tree species (Fig. 1). For current potential distribution, with an AUC of 0.809 (average of 10 replicate models) and a standard deviation of 0.020, predictive qualities of the MaxEnt model for Q. leucotrichophora (ban oak) can be considered only moderate. Similarly, with AUC values of 0.803 (±0.025) for P. roxburghii (pine), the predictive model again can be considered moderate, given the large presence of this species throughout the study region. However, Q. semecarpifolia (kharsu

598

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

Fig. 2. Relative importance of different predictor variables based on results of Jackknife tests in MaxEnt. Graphs show variable contributions to AUC (area under the ROC curve) for: (a) Q. leucotrichophora (ban oak), (b) Q. semecarpifolia (kharsu oak), (c) Q. floribunda (moru oak), and (d) P. roxburghii (pine) shown averaged over 10 replicate runs.

oak) and Q. floribunda (moru oak) show rather good model performance with AUC values, 0.982 (±0.008) and 0.966 (±0.006), respectively. In any model, closer AUC value is to its potential maximum can be only be assessed if the species studied is region specific, i.e. only if the species is dominant in its environmental niche. In case of widespread environmental niche, it automatically corresponds lower AUC value (Elith et al., 2011; Phillips et al., 2006). This can be seen in our model performance for four tree species. In fact, higher values of AUC for Q. semecarpifolia (kharsu oak) and Q. floribunda (moru oak) are related, since they represent the forest type that occupies more defined and restricted ecological niche in the study area. While for Q. leucotrichophora (ban oak) and P. roxburghii (pine), AUC values are lower due to partial overlap of niche of dominant species that characterizes corresponding forest type.

Table 3 Relative variable contributions (%) to the MaxEnt model for four forest tree species.

4.2. Influencing predictor variables

The aforementioned predictor variables present higher gain (contain most information) compared to other variables, which also contribute in the model to some extent (refer Table 3 for details). After analysing the variables contribution to the species distribution model, elevation, as one might also expect is generally the major predictor variable, which distinguishes the presence of these forest tree species. Nonetheless, we can also further deduce that other predictor variables were crucial to characterizing and further differentiating the distribution of the different suitable habitats of the tree species. Interestingly, predictor variable such as aspect only contributes in the detection of Q. leucotrichophora (ban oak), and not for other oak species. Similarly, solar radiation is more important in the MaxEnt model for Q. leucotrichophora (ban oak) and P. roxburghii (pine), and not for Q. semecarpifolia (kharsu oak) and Q. floribunda (moru oak) species, despite the fact that they have a wider spread climatic niche. Other bioclimatic variables vary differently for each

The MaxEnt model’s internal test of variable importance showed that elevation represents the most important and useful layer in explaining tree species distribution. The most influential predictor variables (from their relative contribution to 10 model runs) for current potential distribution for all forest tree species are detailed in Table 3. In particular, temperature seasonality (BIO 4) and elevation contribute the most for Q. leucotrichophora (ban oak). For Q. semecarpifolia (kharsu oak), elevation, temperature annual range (BIO 7) and precipitation of coldest quarter (BIO 19) contribute the most in the MaxEnt model in this decreasing order. Elevation, temperature annual range (BIO 7), and temperature seasonality (BIO 4) contribute the most for Q. floribunda (moru oak), and temperature seasonality (BIO 4) along with elevation contribute most for P. roxburghii (pine) species.

Variable

Ban oak Kharsu oak Moru oak Pine

Elevation (m) Slope (m) Aspect (m) Mean Diurnal Range (◦ C) Isothermality Temperature Seasonality (◦ C) Temperature Annual Range (◦ C) Annual Precipitation (mm) Precipitation of Driest Quarter (mm) Precipitation of Warmest Quarter (mm) Precipitation of Coldest Quarter (mm) Solar radiation (KWh/m2 /day) Soil type

14.1 1.1 2.1 1.9 0 69.7 0.5 3.4 0.4 1 0.4 4.8 0.6

37.8 0.4 0 0.2 0.4 0.5 30.7 0 0.7 0.2 25.5 0.2 3.5

49.3 0.6 0.1 0 0 19.8 22.9 1.6 3.6 0.5 0.4 0.2 1

33.4 1.3 0.4 0 0.8 50.9 0.6 0.7 1.9 0.3 0.4 7.2 2

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

tree species, showing the importance for the need for considering each predictor variable in the species distribution model. Individual Jackknife tests for each tree species also show that elevation have the highest predictive power (highest AUC) for all tree species (Fig. 2). Temperature seasonality had high predictive power for Q. leucotrichophora (ban oak) and P. roxburghii (pine), while temperature annual range had high predictive power for Q. semecarpifolia (kharsu oak) and Q. floribunda (moru oak) tree species. Relevant individual response curves for current potential distribution for show elevation influencing the presence of the species in certain altitudinal ranges only (Fig. 3). For instance, Q. leucotrichophora (ban oak) species is found mainly between 1000 m to 3000 m approximately, Q. semecarpifolia (kharsu oak) is found between 2500 m to ∼3200 m, Q. floribunda (moru oak) between 2000 m to 3000 m and P. roxburghii (pine) is found mainly between 1000 m to 3000 m, approximately. Temperature seasonality (BIO 4) negatively influences Q. leucotrichophora (ban oak), Q. floribunda (moru oak) species and P. roxburghii (pine) species, and temperature annual range (BIO 7) negatively influences Q. semecarpifolia (kharsu oak) and Q. floribunda (moru oak) species. Temperature annual range (BIO 7) more or less negatively influences all the Quercus spp. On the other hand, precipitation of the driest quarter (BIO 17) positively influences P. roxburghii (pine) till values of 100 mm and above and thereafter, influences negatively. Interestingly, precipitation of the coldest quarter (BIO 19) influences only positively Q. semecarpifolia (kharsu oak) species significantly. The probability of Q. leucotrichophora (ban oak) and P. roxburghii (pine) is positively influenced by the direct normal irradiance, but then sharply decreases after a value of 6 KWh/m2 /day.

4.3. Potential current species distribution Our study demonstrates for the first time the potential current distribution of four forest tree species: Q. leucotrichophora (ban oak), Q. semecarpifolia (kharsu oak), Q. floribunda (moru oak), and P. roxburghii (pine) in the Kumaon division of the state of Uttarakhand (Fig. 4). Potential habitats (climatic niches) with high suitability thresholds were distributed more in the higher elevation zones for two Quercus spp., with approximate forest area coverage up to 246.74 km2 and 332.28 km2 for Q. semecarpifolia (kharsu oak) and Q. floribunda (moru oak), respectively. However, Q. leucotrichophora (ban oak) and P. roxburghii (pine) have highly suitable potential habitat areas of values ranging up to 788.56 km2 and 2375.04 km2 , respectively. Medium suitable areas for Q. leucotrichophora (ban oak), Q. semecarpifolia (kharsu oak), Q. floribunda (moru oak), and P. roxburghii (pine) is up to 1072.93 km2 , 387.93 km2 , 326.30 km2 , and 1771.97 km2 , respectively. For potentially low suitable areas, area coverage approximately equals to 2787.17 km2 , 131.80 km2 , 464.88 km2 , and 3196.82 km2 for Q. leucotrichophora (ban oak), Q. semecarpifolia (kharsu oak), Q. floribunda (moru oak), and P. roxburghii (pine) species, respectively. Our field surveys in the study region also reveal that the predicted potential habitats were mostly located in the similar altitudinal zones, with high prevalence of Pinus spp. in areas throughout the study region; even extending to altitudes beyond 3000 m. Sparse tree cover, grasslands, cultivated lands (agriculture and fallow land), and forest scrub mostly fall within areas of medium to low habitat suitability climatic niches. However, these potential habitat areas are representatives of ‘fundamental niche’ of tree species. In reality, it may be over-estimating or under-estimating the actually present ‘realized niche’ of the species. Nevertheless, given the paucity of available data, modelling climatic niches through this approach may be the only congruent option available in this highly complex heterogeneous terrain.

599

Table 4 Overall predicted climate change impacts for four tree species. Tree species

Increase/Decrease

Range shift

Significant change

Ban oak Kharsu oak Moru oak Pine

+/− – ∼∼ ∼∼∼/+ +

↑↑ ↓ ∼ ↑↑↑

Yes No No Yes

* In this case, the signs refer to: (i) + for increase in potential area, (ii) − for decrease in potential area, (iii) ∼ for unusual patterns, (iv) ↑ for upward shift, and, (v) ↓ for downward shift.

4.4. Predicted climate change impacts To estimate the potential impact of climate change on the distribution patterns of the trees species, the change in estimated spatial distribution in current and corresponding future climate scenarios was estimated. Fig. 5 shows zoomed-in views of the study region for potential climate change impacts for Q. leucotrichophora (ban oak) species. On the other hand, Q. semecarpifolia (kharsu oak) shown in Fig. 6 and Q. floribunda (moru oak) shown in Fig. 7, both have relatively less significant change in spatial distribution, in terms of total area cover and changes over an altitudinal gradient. Fig. 8 shows potential climate change impacts in case of P. roxburghii (pine) across all scenarios in two different time periods. The total predicted area of the tree species more or less appeared to be constant; however, it showed changes spatial distribution of low, medium and high suitable areas. A brief summary of the changes due to potential future climate scenario for four tree species observed over two time periods is given in Table 4. The maximum expansion was seen in case of P. roxburghii (pine) from 2375.04 km2 (current scenario) to 2671.32 km2 (under RCP 8.5 by 2070) for high suitable areas, and slight increase for medium suitable areas. On the contrary, low suitable areas for P. roxburghii (pine) reduced with future climate scenarios. This indicates prevalence of highly suitable areas for pine species as we progress toward extremes of climate change. Quercus spp. on the other hand showed decrease in suitable areas with future climate scenarios. For instance, Q. leucotrichophora (ban oak) shows decrease in area from 788.56 km2 to approx. up to 775 km2 (near-term) and 765 km2 (long-term scenario) for high suitable area, and up to 990.37 km2 (long-term scenario) for medium suitable areas. Q. semecarpifolia (kharsu oak) found in higher altitudes show similar trends with decrease in suitable area up to 30 km2 and increase in low suitable area up to 25 km2 . Q. floribunda (moru oak) showed similar decreasing trends in suitable areas to roughly 30 km2 , while increase in low suitable areas up to 70 km2 . With changes in bioclimatic layers from current climate to four RCP trajectories, the potential distribution was for tree species in Himalayan region show different trends, which depict potential impacts of climate change. These results show that changes in future climate scenarios can potentially decrease the climatic niches of tree species with small to large extent depending upon the GHG emission scenario. Although, change in area may be of lesser value, the potential impact can be expected to be at a much larger scale, given the fact that disturbances due to human factors in this region hasn’t been accounted in this modelling exercise. 5. Discussion Our work represents the first attempt to model impacts of future climate change on the spatial distribution of major forest tree species in the Himalayan region. Despite previous forestrelated research in this region, there is lack of association with climate change uncertainties (Negi et al., 2012). Q. leucotrichophora (ban oak), Q. semecarpifolia (kharsu oak), Q. floribunda (moru oak) and P. roxburghii (pine) represent the flagship tree species of

600

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

Fig. 3. Relationships between top predictor variables and probability of presence of tree species: (a) Q. leucotrichophora (ban oak), (b) Q. semecarpifolia (kharsu oak), (c) Q. floribunda (moru oak), and (d) P. roxburghii (pine). Black curves show the mean response and grey margins are +/− one standard deviation calculated over 10 replicates.

the Himalaya. With the sole criteria of appropriately predicting the potential presence or absence of these tree species, MaxEnt modelling procedure performed well with the climatic and other environmental data, representing moderate to high AUC scores (>0.8) in the study region. Since climate change has already impacted species’ habitats including those of several forest ecosystems worldwide (Koralewski et al., 2015; Lindner et al., 2014), it will likely affect the distribution of forests in the Himalayan region as well (Mainali et al., 2015a). In response to climate change, many mountain plant species are expected to migrate to much higher elevations from their current locations (Braunisch et al., 2014; Du et al., 2014). While evidence supports this prediction (Alexander et al., 2015), attempts at modelling responses of mountain plant species to warming have shown contrasting results, with some models predict range expansions (Chen et al., 2011) and others predict contractions (Zhu et al., 2012). Such inconsistencies can be due to the complex interactions between regional macro-climatic conditions and local factors that influence structure and function of mountain tree species (Beniston, 2003; Kerr and Dobrowski, 2013; Körner, 2007).

5.1. Availability of presence-absence data records We used correlative MaxEnt model which can be used to predict how the distribution of species might be affected by changes in future climate depending upon the different RCPs. It is trained on presence-only records, which may result in inadequate and/or inaccurate classifications of the species (Chakraborty and Joshi, 2016).

Since we collected only presence information of the tree species studied, we preferred using this correlative model. However, we believe addition of presence or abundance, along with absence or may be both can significantly improve model predictions. This would also involve the exploration of other ecological niche-based models that could play important role in predicting the probability distribution of the Himalayan forest tree species. Nonetheless, despite the understanding of having more all-inclusive species data points in the Himalayan region, historical data on species distribution are significantly lacking (Shrestha and Bawa, 2014).

5.2. Assumptions with ecological niche-based models In most cases, the species’ presence and absence information, along with environmental as well as climatic data is limited to either space or time. Since the species-environment relationship that is generated through ecological niche-based models is based on this limited information, it can only provide us an overview of the expected potential species distributions. In these modelling approaches, the safest convenient way of working around such predictive modelling is to assume that the species modelled would not attain complete but partial pseudo-equilibrium with its environment (Guisan and Thuiller, 2005). Literature suggests that at times the information regarding environment and climatic parameters at the predictive species’ sites show disagreement with the correlative niche modelling techniques found at presence records datasets (Anderson, 2013). For realistic expectations about the outcomes from the models, the theoretical paradigm of the selection crite-

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

601

Fig. 4. Potential geographic distributions of current climatic niches for a) Quercus leucotrichophora (ban oak), (b) Quercus semecarpifolia (kharsu oak), (c) Quercus floribunda (moru oak), and (d) P. roxburghii (pine). In this case, the probability of occurrence of the tree species is categorized as low, medium and high suitable areas (however, this may vary depending on varying conditions on field sites).

ria of the models should be carefully considered and highlighted given the methodological constraints of any region (Chakraborty and Joshi, 2016). To add to this, there are several statistical obstacles when it comes to interpreting these modelling results (Elith et al., 2011). Given these limitations, the first and foremost factor that needs to be checked is the availability of data for the region in concern. The collection of data at different spatial resolutions, all through different time periods, along with collection under diverse taxonomic and ecological hierarchies creates greater difficulty, depending on the ultimate purpose of niche modelling exercise. Therefore, to integrate all the data available in one plat-

form itself, is an immense challenge and requires specific dedicated analysis, unlike done otherwise. 5.3. Accuracy of ecological niche-based models The type of input data used (occurrence records and environmental variables) through any niche modelling attempt significantly affects the quality of species’ distribution outputs produced in the end (Rocchini et al., 2011). For instance, elevation as a relevant indicator should be included as a predictor variable in niche models for mountain plant species (Oke and Thompson,

602

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

Fig. 5. Climate change impacts for ban oak over four RCPs for 2050 and 2070.

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

603

Fig. 6. Climate change impacts for kharsu oak over four RCPs for 2050 and 2070.

2015). With the precise choice of choosing any model and its careful estimation of parameters, including adequate amount of sample size of species, geographically uniform occurrence data and regularization parameters such as in MaxEnt, it can produce reliable predictions of habitat suitability in species distribution maps (Convertino et al., 2011b). However, geographically biased distributional data will automatically distort the species distribution representation, along with their environmental responses (Graham et al., 2008; Hortal et al., 2008). If locational errors are consistent across species (Graham et al., 2008), such biases in geographical space will translate into biases in environmental space as well (Araújo and Guisan, 2006), thus, ultimately limiting the model’s performance in space. Since predictive accuracy of niche models is sensitive to both sample size and biases in the distribution of data (Araújo and New, 2007), very small sample size of species occurrence data could seriously hinder the usefulness different modelling approaches (Araújo and Peterson, 2012; Wiens et al., 2009; Wisz et al., 2008). This drawback will also affect ecological niche-based model such as MaxEnt, even though it is regarded to perform relatively well with small sample sizes (Hernandez et al., 2006).

The use of environmental variables as predictors for niche models can also be a source of uncertainty (Rocchini et al., 2011). Correspondingly, environmental variables derived from remote sensing data may also be prone to classification errors (Sohl, 2014). Using biased environmental data, especially those created from climatic models and geographical interpolations (Chen et al., 2015), may lead to errors in distribution maps depending on the scale of the study (Convertino et al., 2011b) and accuracy of these models (Hijmans and Graham, 2006; Thuiller, 2004). Therefore, for more robust assessments of potential effects of climate change on different regions with poorly known taxa, projections of bioclimatic envelope models can be implemented in congruence with climate change metrics which are applied independently of species occurrence data (Garcia et al., 2016).

5.4. Sensitivity and uncertainty analysis Many studies explore the comparison between different ecological niche-based models (Barbet-Massin et al., 2012; Hernandez et al., 2006; Mainali et al., 2015b; Navarro-Cerrillo et al., 2011), while others focus on capturing the effects of uncertainty of different environmental variables, either considering one at a time

604

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

Fig. 7. Climate change impacts for moru oak over four RCPs for 2050 and 2070.

(Convertino et al., 2013, 2012b), or considering variability of all input factors at the same time, such as global sensitivity and uncertainty analysis (GSUA) (Convertino et al., 2014). To the best of our knowledge, the uncertainty of MaxEnt models related to internal input variables, or interaction among all possible combinations of environmental covariates and internal input variables is often not evaluated systematically, except in few cases where models and model parametizations have been explored (Briscoe et al., 2016; Chu-Agor et al., 2012; Convertino et al., 2014; Ochoa-Ochoa et al., 2016). Although, usually adopted in species distribution modelling, variable importance is often calculated through information of the influence of one input factor while keeping all other factors constant one at a time, such as Jacknife tests in MaxEnt models (Elith et al., 2011; Pearson et al., 2007). Through such sensitivity analysis, niche models can help in the local management regimes by providing quantitative species specific decisions by allowing different management scenarios based on the variable’s importance for land resource managers (Rengstorf et al., 2013). For instance, GSUA can help in indicating ecosystem management options in terms of their ramifications for ecological resilience (Perz et al., 2013). However, since we explored only the climatic, geomorphologic, pedologic variables in this study, we have restricted the use of

GSUA, as input variables of such scale are beyond a certain scope of local land resource management. Nonetheless, use of environmental variables such as land use land cover information (Convertino et al., 2014, 2012a), proximity to road and settlements (Mainali et al., 2015b), conservation interventions (Velásquez-Tibatá et al., 2013), or socio-economic variavles (Ray et al., 2016), are dynamic enough to be influenced land management decisions. Therefore, while considering such dynamic variables, the modulations of their impacts on ecosystems and subsequent uncertainty for species distribution models, often require GSUA. Our study was conducted at a much larger scale to understand the potential impacts of climate change on different forest tree species, using environmental variables beyond the control of local land management regimes. Hence, our study can be further extended explicitly to examine the dependencies among different manageable input variables and their importance in decision making for conservation and resource management.

5.5. Potential future predictions of tree species The greatest difficulty in the assessment of impacts of climate change on forest tree species is the uncertainty in future climate

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

Fig. 8. Climate change impacts for pine over four RCPs for 2050 and 2070.

605

606

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

predictions (Rosenzweig et al., 2014), which makes developing adaptive strategies for forest management slightly challenging (Wang et al., 2015). According to available datasets, there are four GHG emission scenarios which includes the RCPs and over 20 general circulation models (GCMs) in the latest IPCC reports (IPCC, 2013). These models show different projections of future climate conditions. The choice of one of these models, or maybe more to be used in predicting a future potential climatic niche is problematic. It is resource demanding and time consuming, and sometimes, the usage of different models, for instance, climate simulations from the CMIPs (CMIP3 and CMIP5) do not necessarily change predictions of future habitat suitability of different species (Wright et al., 2015). Despite the advances in simulating the future climate from third phase of the CMIP (CMIP3) contributed to the IPCC fourth assessment report (AR4) to the fifth phase (CMIP5) contributed to the AR5, inter-model uncertainty over projected changes still persists (Trenberth, 2010). Therefore, from a modelling perspective, the large range of future climate scenarios makes it almost impossible to accommodate a range of potential future results from such niche modelling experiments. Hence, there is an increasing need for additional forest management decisions throughout the 21st century with more field-based experiments, accommodating impacts of human disturbances as well (Turner et al., 2015). For more precise interpretation, the realization of these caveats is essential, as one may take up the endeavour to solve such ecological riddles.

plant physiology, but it is definitely associated with changes in meteorological factors such as atmospheric pressure, solar radiation, cloud cover, precipitation and solar radiation, which directly and indirectly influences presence or absence of many tree species in mountainous regions (Körner, 2007). Although, elevation alone cannot fully explain species distributions, our results suggest that the integration of elevation with other geomorphologic, pedologic and climatic variables can yield more accurate results for defining climatic niches. In summary, this study provides a strong baseline database for the assessment of climate change impact on tree species and for developing broad-scale adaptive strategies for forest tree species restoration and conservation planning, in consideration of the uncertainty of future climate. Acknowledgements AC would like to acknowledge HSBC Climate Scholarship of TERI University for funding her doctoral research. PKJ and KS would like to acknowledge the Ministry of Environment, Forests and Climate Change (MoEF&CC), Government of India (GoI) for their support (Project Serial Number: R&D/NNRMS/2/2013-14). The authors are grateful to the anonymous reviewers and the editorial board for providing valuable comments to enhance the quality of the previous version of the manuscript. References

5.6. Comparison with previous studies Scientific literature indicates reduced distribution pattern of Q. semecarpifolia (kharsu oak) from 40% and 76% with every 1 ◦ C and 2 ◦ C rise in temperature, respectively (Saran et al., 2010). This corresponds to our modelling results as well, which show similar response, however, to a much lesser extent with scenarios of future RCPs. According to our results, there is decrease Quercus spp. and increase in the distribution of P. roxburghii (pine) over the progressive RCPs for 2050 and 2070. Comparable results have been reported where anthropogenic disturbances along with climate change replaces oak species with other tree species, especially pine along the Lesser Himalayan belt (between 750 and 2500 m elevations) (Rawat et al., 2012). Despite the fact that the western Himalaya have always been associated with the climate change sensitive species such as Q. leucotrichophora (ban oak) and P. roxburghii (pine) (Negi et al., 2012), empirical evidence to support the same theory is still lacking. Our study reports for the first time, response of dominant forest trees in the Himalayan region to predicted impacts of climate change; although, our modelling attempt does not account for human disturbances such as encroachment to forest-covered areas, fire, cutting, over-grazing, intensive agriculture, or shifting-cultivation predominant in this region. 6. Conclusion In this study, we modelled the impacts of climate change on potential climatic niches of four important forest tree species (Q. leucotrichophora, Q. semecarpifolia, Q. floribunda, and P. roxburghii) in the central Himalayan region, with moderate to high level of model accuracies. This was achieved despite limited species data points (presence and absence information), highlighting the usefulness of the approach employed for this study to assess the impacts of climate change on forests. Our results demonstrate effective modelling of species with sufficient consideration given to the ecological characteristics of tree species in the modelling procedures. Certain predictor variables such as elevation proved to be highly relevant since it significantly improved accuracies of the distribution model, which proves that elevation may not directly influence

Ahlström, A., Schurgers, G., Arneth, A., Smith, B., 2012. Robustness and uncertainty in terrestrial ecosystem carbon response to CMIP5 climate change projections. Environ. Res. Lett. 7, 44008, http://dx.doi.org/10.1088/1748-9326/7/4/044008. Aitken, S.N., Yeaman, S., Holliday, J.A., Wang, T., Curtis-McLane, S., 2008. Adaptation, migration or extirpation: climate change outcomes for tree populations. Evol. Appl. 1, 95–111, http://dx.doi.org/10.1111/j.1752-4571. 2007.00013.x. Alekhya, V.P., Pujar, G., Jha, C., Dadhwal, V., 2015. Simulation of vegetation dynamics in Himalaya using dynamic global vegetation model. Trop. Ecol. 56, 219–231. Alexander, J.M., Diez, J.M., Levine, J.M., 2015. Novel competitors shape species/’ responses to climate change. Nature 525, 515–518, http://dx.doi.org/10.1038/ nature14952. Anderson, R.P., Gonzalez Jr., I., 2011. Species-specific tuning increases robustness to sampling bias in models of species distributions: an implementation with Maxent. Ecol. Modell. 222, 2796–2811, http://dx.doi.org/10.1016/j.ecolmodel. 2011.04.011. Anderson, R.P., 2013. A framework for using niche models to estimate impacts of climate change on species distributions. Ann. N.Y. Acad. Sci. 1297, 8–28, http:// dx.doi.org/10.1111/nyas.12264. Araújo, M.B., Guisan, A., 2006. Five (or so) challenges for species distribution modelling. J. Biogeogr. 33, 1677–1688. Araújo, M.B., New, M., 2007. Ensemble forecasting of species distributions. Trends Ecol. Evol. 22, 42–47. Araújo, M.B., Peterson, A.T., 2012. Uses and misuses of bioclimatic envelope modeling. Ecology 93, 1527–1539, http://dx.doi.org/10.1890/11-1930.1. Araújo, M.B., Williams, P.H., 2000. Selecting areas for species persistence using occurrence data. Biol. Conserv. 96, 331–345. Barbet-Massin, M., Jiguet, F., Albert, C.H., Thuiller, W., 2012. Selecting pseudo-absences for species distribution models: how, where and how many? Methods Ecol. Evol. 3, 327–338. Beniston, M., 2003. Climate change in mountain regions: a review of possible impacts. Clim. Change 59, 5–31. Betts, R., Golding, N., Gonzalez, P., Gornall, J., Kahana, R., Kay, G., Mitchell, L., Wiltshire, A., 2015. Climate and land use change impacts on global terrestrial ecosystems and river flows in the HadGEM2-ES Earth system model using the representative concentration pathways. Biogeosciences 12, 1317–1338. Bowler, M.G., 2014. Species abundance distributions, statistical mechanics and the priors of MaxEnt. Theor. Popul. Biol. 92, 69–77, http://dx.doi.org/10.1016/j.tpb. 2013.12.002. Braunisch, V., Coppes, J., Arlettaz, R., Suchant, R., Zellweger, F., Bollmann, K., 2014. Temperate mountain forest biodiversity under climate change: compensating negative effects by increasing structural complexity. PLoS One 9, e97718. Briscoe, N.J., Kearney, M.R., Taylor, C.A., Wintle, B.A., 2016. Unpacking the mechanisms captured by a correlative species distribution model to improve predictions of climate refugia. Global Change Biol. 22, 2425–2439, http://dx. doi.org/10.1111/gcb.13280. Chakraborty, A., Joshi, P.K., 2016. Simulation-based approaches for ecological niche modelling: a geospatial reference. In: Samui, Pijush (Ed.), Handbook of Research on Advanced Computational Techniques for Simulation-Based Engineering. IGI Global, Hershey, PA USA, pp. 148–170.

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609 Chakraborty, A., Joshi, P.K., Ghosh, A., Areendran, G., 2013. Assessing biome boundary shifts under climate change scenarios in India. Ecol. Indic. 34, 536–547. Chaturvedi, O.P., Singh, J.S., 1987. The structure and function of pine forest in Central Himalaya. I. Dry matter dynamics. Ann. Bot. 60, 237–252. Chaturvedi, R.K., Gopalakrishnan, R., Jayaraman, M., Bala, G., Joshi, N.V., Sukumar, R., Ravindranath, N.H., 2011. Impact of climate change on Indian forests: a dynamic vegetation modeling approach. Mitig. Adapt. Strat. Global Change 16, 119–142. Chen, I.-C., Hill, J.K., Ohlemüller, R., Roy, D.B., Thomas, C.D., 2011. Rapid range shifts of species associated with high levels of climate warming. Science 333, 1024–1026, http://dx.doi.org/10.1126/science.1206432. Chen, J., Brissette, F.P., Lucas-Picher, P., 2015. Assessing the limits of bias-correcting climate model outputs for climate change impact studies. J. Geophys. Res.: Atmos. 120, 1123–1136. Chitale, V.S., Behera, M.D., Roy, P.S., 2014. Future of endemic flora of biodiversity hotspots in India. PLoS One 9, e115264, http://dx.doi.org/10.1371/journal. pone.0115264. ˜ R., Kiker, G.A., Aiello-Lammens, M.E., Akc¸akaya, Chu-Agor, M.L., Munoz-Carpena, H.R., Convertino, M., Linkov, I., 2012. Simulating the fate of Florida Snowy Plovers with sea-level rise: exploring research and management priorities with a global uncertainty and sensitivity analysis perspective. Ecol. Modell. 224, 33–47, http://dx.doi.org/10.1016/j.ecolmodel.2011.10.021. Collins, W.J., Bellouin, N., Doutriaux-Boucher, M., Gedney, N., Halloran, P., Hinton, T., Hughes, J., Jones, C.D., Joshi, M., Liddicoat, S., Martin, G., O’Connor, F., Rae, J., Senior, C., Sitch, S., Totterdell, I., Wiltshire, A., Woodward, S., 2011. Development and evaluation of an Earth-system model–HadGEM2. Geosci. Model Dev. 4, 1051–1075, http://dx.doi.org/10.5194/gmd-4-1051-2011. ˜ R., Convertino, M., Donoghue, J.F., Chu-Agor, M.L., Kiker, G.A., Munoz-Carpena, Fischer, R.A., Linkov, I., 2011a. Anthropogenic renourishment feedback on shorebirds: a multispecies Bayesian perspective. Ecol. Eng. 37, 1184–1194, http://dx.doi.org/10.1016/j.ecoleng.2011.02.019. ˜ R., Chu-Agor, M., Fischer, R., Linkov, I., Convertino, M., Kiker, G., Munoz-Carpena, 2011b. Scale-and resolution-invariance of suitable geographic range for shorebird metapopulations. Ecol. Complexity 8, 364–376. ˜ Convertino, M., Bockelie, A., Kiker, G.A., Munoz-Carpena, R., Linkov, I., 2012a. Shorebird patches as fingerprints of fractal coastline fluctuations due to climate change. Ecol. Process. 1, 1–17, http://dx.doi.org/10.1186/2192-1709-1-9. ˜ R., Kiker, G.A., Chu-Agor, M.L., Fischer, Convertino, M., Welle, P., Munoz-Carpena, R.A., Linkov, I., 2012b. Epistemic uncertainty in predicting shorebird biogeography affected by sea-level rise. Ecol. Modell. 240, 1–15, http://dx.doi. org/10.1016/j.ecolmodel.2012.04.012. Convertino, M., Troccoli, A., Catani, F., 2013. Detecting fingerprints of landslide drivers: a MaxEnt model. J. Geophys. Res. Earth Surf. 118, 1367–1386, http:// dx.doi.org/10.1002/jgrf.20099. ˜ R., Chu-Agor, M.L., Kiker, G.A., Linkov, I., 2014. Convertino, M., Munoz-Carpena, Untangling drivers of species distributions: global sensitivity and uncertainty analyses of MaxEnt. Environ. Modell. Softw. 51, 296–309, http://dx.doi.org/10. 1016/j.envsoft.2013.10.001. R Development Core Team, (2011). R: A Language and Environment for Statistical Computing. Dirnböck, T., Essl, F., Rabitsch, W., 2011. Disproportional risk for habitat loss of high-altitude endemic species under climate change. Global Change Biol. 17, 990–996, http://dx.doi.org/10.1111/j.1365-2486.2010.02266.x. Dormann, C.F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré, G., Marquéz, J.R.G., Gruber, B., Lafourcade, B., Leitão, P.J., Münkemüller, T., McClean, C., Osborne, P.E., Reineking, B., Schröder, B., Skidmore, A.K., Zurell, D., Lautenbach, S., 2013. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography 36, 27–46, http://dx.doi.org/10.1111/ j.1600-0587.2012.07348.x. Dormann, C.F., 2007. Promising the future? Global change projections of species distributions. Basic Appl. Ecol. 8, 387–397. Drexler, M., Ainsworth, C.H., 2013. Generalized additive models used to predict species abundance in the Gulf of Mexico: an ecosystem modeling tool. PLoS One 8, e64458. Du, J., He, Z., Yang, J., Chen, L., Zhu, X., 2014. Detecting the effects of climate change on canopy phenology in coniferous forests in semi-arid mountain regions of China. Int. J. Remote Sens. 35, 6490–6507, http://dx.doi.org/10.1080/ 01431161.2014.955146. Elith, J., Graham, C.H., 2009. Do they? How do they? WHY do they differ? On finding reasons for differing performances of species distribution models. Ecography 32, 66–77, http://dx.doi.org/10.1111/j.1600-0587.2008.05505.x. Elith, J., Graham, C., Anderson, R., Dudik, M., Ferrier, S., Guisan, A., Hijmans, R., Huettmann, F., Leathwick, J., Lehmann, A., Li, J., Lohmann, L., Loiselle, B., Manion, G., Moritz, C., Nakamura, M., Nakazawa, Y., Overton, J., Peterson, A., Phillips, S., Richardson, K., Scachetti-Pereira, R., Schapire, R., Soberon, J., Williams, S., Wisz, M., Zimmermann, N., 2006. Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29, 129–151, http://dx.doi.org/10.1111/j.2006.0906-7590.04596.x. Elith, J., Phillips, S.J., Hastie, T., Dudı´ık, M., Chee, Y.E., Yates, C.J., 2011. A statistical explanation of MaxEnt for ecologists. Divers. Distrib. 17, 43–57. Friend, A.D., Lucht, W., Rademacher, T.T., Keribin, R., Betts, R., Cadule, P., Ciais, P., Clark, D.B., Dankers, R., Falloon, P.D., Ito, A., Kahana, R., Kleidon, A., Lomas, M.R., Nishina, K., Ostberg, S., Pavlick, R., Peylin, P., Schaphoff, S., Vuichard, N., Warszawski, L., Wiltshire, A., Woodward, F.I., 2014. Carbon residence time dominates uncertainty in terrestrial vegetation responses to future climate

607

and atmospheric CO2. PNAS 111, 3280–3285, http://dx.doi.org/10.1073/pnas. 1222477110. Gairola, S., Proches¸, S¸., Rocchini, D., 2013. High-resolution satellite remote sensing: a new frontier for biodiversity exploration in Indian Himalayan forests. Int. J. Remote Sens. 34, 2006–2022, http://dx.doi.org/10.1080/01431161.2012. 730161. Gallagher, R.V., Hughes, L., Leishman, M.R., 2013. Species loss and gain in communities under future climate change: consequences for functional diversity. Ecography 36, 531–540, http://dx.doi.org/10.1111/j.1600-0587.2012. 07514.x. Garcia, R.A., Cabeza, M., Altwegg, R., Araújo, M.B., 2016. Do projections from bioclimatic envelope models and climate change metrics match? Global Ecol. Biogeogr. 25, 65–74, http://dx.doi.org/10.1111/geb.12386. González-Moreno, P., Diez, J.M., Richardson, D.M., Vilà, M., 2015. Beyond climate: disturbance niche shifts in invasive species. Global Ecol. Biogeo., http://dx.doi. org/10.1111/geb.12271. Gopalakrishnan, R., Jayaraman, M., Bala, G., Ravindranath, N.H., 2011. Climate change and Indian forests. Curr. Sci. 101, 348–355. Graham, C.H., Elith, J., Hijmans, R.J., Guisan, A., Townsend Peterson, A., Loiselle, B.A., 2008. The influence of spatial errors in species occurrence data used in distribution models. J. Appl. Ecol. 45, 239–247. Guisan, A., Thuiller, W., 2005. Predicting species distribution: offering more than simple habitat models. Ecol. Lett. 8, 993–1009. Guisan, A., Zimmermann, N.E., 2000. Predictive habitat distribution models in ecology. Ecol. Modell. 135, 147–186, http://dx.doi.org/10.1016/S03043800(00)00354-9. Hansen, A.J., Phillips, L.B., 2015. Which tree species and biome types are most vulnerable to climate change in the US Northern Rocky Mountains? For. Ecol. Manage. 338, 68–83, http://dx.doi.org/10.1016/j.foreco.2014.11.008. Harte, J., Newman, E.A., 2014. Maximum information entropy: a foundation for ecological theory. Trends Ecol. Evol. 29, 384–389, http://dx.doi.org/10.1016/j. tree.2014.04.009. Heikkinen, R.K., Luoto, M., Araújo, M.B., Virkkala, R., Thuiller, W., Sykes, M.T., 2006. Methods and uncertainties in bioclimatic envelope modelling under climate change. Prog. Phys. Geogr. 30, 751–777. Hernandez, P.A., Graham, C.H., Master, L.L., Albert, D.L., 2006. The effect of sample size and species characteristics on performance of different species distribution modeling methods. Ecography 29, 773–785, http://dx.doi.org/10. 1111/j.0906-7590.2006.04700.x. Hijmans, R.J., Graham, C.H., 2006. The ability of climate envelope models to predict the effect of climate change on species distributions. Global Change Biol. 12, 2272–2281, http://dx.doi.org/10.1111/j.1365-2486.2006.01256.x. Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A., 2005. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965–1978. Hijmans, R.J., Phillips, S., Leathwick, J., Elith, J., Hijmans, M.R.J., 2015. Package dismo—species distribution modeling (v. 1.0-12). Circles 9, 1. Hirzel, A.H., Hausser, J., Chessel, D., Perrin, N., 2002. Ecological-niche factor analysis: how to compute habitat-suitability maps without absence data? Ecology 83, 2027–2036. Hortal, J., Jiménez-Valverde, A., Gómez, J.F., Lobo, J.M., Baselga, A., 2008. Historical bias in biodiversity inventories affects the observed environmental niche of the species. Oikos 117, 847–858. IPCC, 2013. Climate change 2013: the physical science basis. In: Stocker, T.F., Qin, D., Plattner, G.-K., Tignor, M., Allen, S.K., Boschung, J., Nauels, A., Xia, Y., Bex, V., Midgley, P.M. (Eds.), Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press Cambridge, United Kingdom and New York, NY, USA. Jewitt, D., Erasmus, B.F.N., Goodman, P.S., O’Connor, T.G., Hargrove, W.W., Maddalena, D.M., Witkowski, E.T.F., 2015. Climate-induced change of environmentally defined floristic domains: a conservation based vulnerability framework. Appl. Geogr. 63, 33–42, http://dx.doi.org/10.1016/j.apgeog.2015. 06.004. Johnson, C.J., Seip, D.R., Boyce, M.S., 2004. A quantitative approach to conservation planning: using resource selection functions to map the distribution of mountain caribou at multiple spatial scales. J. Appl. Ecol. 41, 238–251, http:// dx.doi.org/10.1111/j.0021-8901.2004.00899.x. Jones, C.D., Hughes, J.K., Bellouin, N., Hardiman, S.C., Jones, G.S., Knight, J., Liddicoat, S., O’Connor, F.M., Andres, R.J., Bell, C., Boo, K.-O., Bozzo, A., Butchart, N., Cadule, P., Corbin, K.D., Doutriaux-Boucher, M., Friedlingstein, P., Gornall, J., Gray, L., Halloran, P.R., Hurtt, G., Ingram, W.J., Lamarque, J.-F., Law, R.M., Meinshausen, M., Osprey, S., Palin, E.J., Parsons Chini, L., Raddatz, T., Sanderson, M.G., Sellar, A.A., Schurer, A., Valdes, P., Wood, N., Woodward, S., Yoshioka, M., Zerroukat, M., 2011. The HadGEM2-ES implementation of CMIP5 centennial simulations. Geosci. Model Dev. 4, 543–570, http://dx.doi.org/10.5194/gmd-4-543-2011. Joshi, P.K., Rawat, A., Narula, S., Sinha, V., 2012. Assessing impact of climate change on forest cover type shifts in Western Himalayan Eco-region. J. For. Res. 23, 75–80. Körner, C., 2007. The use of altitude in ecological research. Trends Ecol. Evol. 22, 569–574, http://dx.doi.org/10.1016/j.tree.2007.09.006. Kearney, M., Porter, W., 2009. Mechanistic niche modelling: combining physiological and spatial data to predict species’ ranges. Ecol. Lett. 12, 334–350. Keenan, R.J., 2015. Climate change impacts and adaptation in forest management: a review. Annals of Forest Science1 1–223, http://dx.doi.org/10.1007/s13595014-0446-5.

608

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609

Kerr, J.T., Dobrowski, S.Z., 2013. Predicting the impacts of global change on species, communities and ecosystems: it takes time. Global Ecol. Biogeogr. 22, 261–263. Koralewski, T.E., Wang, H.-H., Grant, W.E., Byram, T.D., 2015. Plants on the move: assisted migration of forest trees in the face of climate change. For. Ecol. Manage. 344, 30–37, http://dx.doi.org/10.1016/j.foreco.2015.02.014. Kulkarni, A., Patwardhan, S., Kumar, K.K., Ashok, K., Krishnan, R., 2013. Projected climate change in the hindu Kush–Himalayan region by using the high-resolution regional climate model PRECIS. Mt. Res. Dev. 1, 142–151, http://dx.doi.org/10.1659/MRD-JOURNAL-D-11-00131.1. Kumar, S., Graham, J., West, A.M., Evangelista, P.H., 2014. Using district-level occurrences in MaxEnt for predicting the invasion potential of an exotic insect pest in India. Comput. Electron. Agric. 103, 55–62, http://dx.doi.org/10.1016/j. compag.2014.02.007. Lewis, S.L., Edwards, D.P., Galbraith, D., 2015. Increasing human dominance of tropical forests. Science 349, 827–832. Lindner, M., Maroschek, M., Netherer, S., Kremer, A., Barbati, A., Garcia-Gonzalo, J., Seidl, R., Delzon, S., Corona, P., Kolström, M., Lexer, M.J., Marchetti, M., 2010. Climate change impacts, adaptive capacity, and vulnerability of European forest ecosystems. For. Ecol. Manage. 259, 698–709. Lindner, M., Fitzgerald, J.B., Zimmermann, N.E., Reyer, C., Delzon, S., van der Maaten, E., Schelhaas, M.-J., Lasch, P., Eggers, J., van der Maaten-Theunissen, M., Suckow, F., Psomas, A., Poulter, B., Hanewinkel, M., 2014. Climate change and European forests: what do we know, what are the uncertainties, and what are the implications for forest management? J. Environ. Manage. 146, 69–83, http://dx.doi.org/10.1016/j.jenvman.2014.07.030. Mainali, J., All, J., Jha, P.K., Bhuju, D.R., 2015a. Responses of montane forest to climate variability in the central himalayas of Nepal. Mt. Res. Dev. 35, 66–77, http://dx.doi.org/10.1659/MRD-JOURNAL-D-13-00101.1. Mainali, K.P., Warren, D.L., Dhileepan, K., McConnachie, A., Strathie, L., Hassan, G., Karki, D., Shrestha, B.B., Parmesan, C., 2015b. Projecting future expansion of invasive species: comparing and improving methodologies for species distribution modeling. Global Change Biol. 21, 4464–4480, http://dx.doi.org/ 10.1111/gcb.13038. Manish, K., Telwala, Y., Nautiyal, D.C., Pandit, M.K., 2016. Modelling the impacts of future climate change on plant communities in the Himalaya: a case study from Eastern Himalaya, India. Model. Earth Syst. Environ. 2, 1–12, http://dx. doi.org/10.1007/s40808-016-0163-1. Mantyka-Pringle, C.S., Visconti, P., Di Marco, M., Martin, T.G., Rondinini, C., Rhodes, J.R., 2015. Climate change modifies risk of global biodiversity loss due to land-cover change. Biol. Conserv. 187, 103–111, http://dx.doi.org/10.1016/j. biocon.2015.04.016. Merow, C., Smith, M.J., Silander, J.A., 2013. A practical guide to MaxEnt for modeling species’ distributions: what it does, and why inputs and settings matter. Ecography 36, 1058–1069, http://dx.doi.org/10.1111/j.1600-0587. 2013.07872.x. Miller, J., Franklin, J., Aspinall, R., 2007. Incorporating spatial dependence in predictive vegetation models. Ecol. Modell. 202, 225–242. Minor, E.S., Urban, D.L., 2007. Graph theory as a proxy for spatially explicit population models in conservation planning. Ecol. Appl. 17, 1771–1782. Monahan, W.B., 2009. A mechanistic niche model for measuring species’ distributional responses to seasonal temperature gradients. PLoS One 4, e7921, http://dx.doi.org/10.1371/journal.pone.0007921. Moritz, C., Agudo, R., 2013. The future of species under climate change: resilience or decline? Science 341, 504–508, http://dx.doi.org/10.1126/science.1237190. Moss, R.H., Edmonds, J.A., Hibbard, K.A., Manning, M.R., Rose, S.K., van Vuuren, D.P., Carter, T.R., Emori, S., Kainuma, M., Kram, T., 2010. The next generation of scenarios for climate change research and assessment. Nature 463, 747–756. Navarro-Cerrillo, R.M., Hernández-Bermejo, J.E., Hernández-Clemente, R., 2011. Evaluating models to assess the distribution of Buxus balearica in southern Spain. Appl. Veg. Sci. 14, 256–267, http://dx.doi.org/10.1111/j.1654-109X. 2010.01112.x. Negi, A.S., Negi, G.C.S., Singh, S.P., 1996. Establishment and growth of Quercus floribunda seedlings after a mast year. J. Veg. Sci. 7, 559–564, http://dx.doi.org/ 10.2307/3236305. Negi, G.C.S., Samal, P.K., Kuniyal, J.C., Kothyari, B.P., Sharma, R.K., Dhyani, P.P., 2012. Impact of climate change on the western Himalayan mountain ecosystems: an overview. Trop. Ecol. 53, 345–356. Ochoa-Ochoa, L.M., Flores-Villela, O.A., Bezaury-Creel, J.E., 2016. Using one vs. many, sensitivity and uncertainty analyses of species distribution models with focus on conservation area networks. Ecol. Modell. 320, 372–382, http://dx.doi. org/10.1016/j.ecolmodel.2015.10.031. Oke, O.A., Thompson, K.A., 2015. Distribution models for mountain plant species: the value of elevation. Ecol. Modell. 301, 72–77, http://dx.doi.org/10.1016/j. ecolmodel.2015.01.019. Pearce, J., Ferrier, S., 2000. An evaluation of alternative algorithms for fitting species distribution models using logistic regression. Ecol. Modell. 128, 127–147, http://dx.doi.org/10.1016/S0304-3800(99)00227-6. Pearson, R.G., Dawson, T.P., 2003. Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful? Global Ecol. Biogeogr. 12, 361–371, http://dx.doi.org/10.1046/j.1466-822X.2003.00042.x. Pearson, R., Dawson, T., Berry, P., Harrison, P., 2002. SPECIES: a spatial evaluation of climate impact on the envelope of species. Ecol. Modell. 154, 289–300. Pearson, R.G., Raxworthy, C.J., Nakamura, M., Townsend Peterson, A., 2007. ORIGINAL ARTICLE: Predicting species distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. J. Biogeogr. 34, 102–117, http://dx.doi.org/10.1111/j.1365-2699.2006.01594.x.

Pearson, R.G., 2006. Climate change and the migration capacity of species. Trends Ecol. Evol. 21, 111–113, http://dx.doi.org/10.1016/j.tree.2005.11.022. ˜ R., Kiker, G., Holt, R.D., 2013. Evaluating ecological Perz, S.G., Munoz-Carpena, resilience with global sensitivity and uncertainty analysis. Ecol. Modell. 263, 174–186, http://dx.doi.org/10.1016/j.ecolmodel.2013.04.024. Peterson, A.T., 2003. Predicting the geography of species’ invasions via ecological niche modeling. Q. Rev. Biol. 78, 419–433. Peterson, A.T., 2011. Ecological niche conservatism: a time-structured review of evidence. J. Biogeogr. 38, 817–827. Phillips, S.J., Dudík, M., 2008. Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecography 31, 161–175, http://dx. doi.org/10.1111/j.0906-7590.2008.5203.x. Phillips, S.J., Anderson, R.P., Schapire, R.E., 2006. Maximum entropy modeling of species geographic distributions. Ecol. Modell. 190, 231–259. Ranjitkar, S., Kindt, R., Sujakhu, N.M., Hart, R., Guo, W., Yang, X., Shrestha, K.K., Xu, J., Luedeling, E., 2014. Separation of the bioclimatic spaces of Himalayan tree rhododendron species predicted by ensemble suitability models. Global Ecol. Conserv. 1, 2–12, http://dx.doi.org/10.1016/j.gecco.2014.07.001. Rawat, P.K., Tiwari, P.C., Pant, C.C., 2012. Climate change accelerating land use dynamic and its environmental and socio-economic risks in the Himalayas: mitigation through sustainable land use. Int. J. Clim. Change Strategies Manage. 4, 452–471. Ray, D., Behera, M.D., Jacob, J., 2016. Predicting the distribution of rubber trees (Hevea brasiliensis) through ecological niche modelling with climate, soil, topography and socioeconomic factors. Ecol. Res. 31, 75–91, http://dx.doi.org/ 10.1007/s11284-015-1318-7. Rengstorf, A.M., Yesson, C., Brown, C., Grehan, A.J., 2013. High-resolution habitat suitability modelling can improve conservation of vulnerable marine ecosystems in the deep sea. J. Biogeogr. 40, 1702–1714, http://dx.doi.org/10. 1111/jbi.12123. Reu, B., Proulx, R., Bohn, K., Dyke, J.G., Kleidon, A., Pavlick, R., Schmidtlein, S., 2011. The role of climate and plant functional trade-offs in shaping global biome and biodiversity patterns. Global Ecol. Biogeogr. 20, 570–581, http://dx.doi.org/10. 1111/j.1466-8238.2010.00621.x. Rocchini, D., Hortal, J., Lengyel, S., Lobo, J.M., Jiménez-Valverde, A., Ricotta, C., Bacaro, G., Chiarucci, A., 2011. Accounting for uncertainty when mapping species distributions: the need for maps of ignorance. Prog. Phys. Geogr. 35, 211–226. Rosenzweig, C., Elliott, J., Deryng, D., Ruane, A.C., Müller, C., Arneth, A., Boote, K.J., Folberth, C., Glotter, M., Khabarov, N., Neumann, K., Piontek, F., Pugh, T.A.M., Schmid, E., Stehfest, E., Yang, H., Jones, J.W., 2014. Assessing agricultural risks of climate change in the 21 st century in a global gridded crop model intercomparison. PNAS 111, 3268–3273, http://dx.doi.org/10.1073/pnas. 1222463110. Saran, S., Joshi, R., Sharma, S., Padalia, H., Dadhwal, V.K., 2010. Geospatial modeling of brown oak (Quercus semecarpifolia) habitats in the kumaun himalaya under climate change scenario. J. Indian Soc. Remote Sens. 38, 535–547. Shcheglovitova, M., Anderson, R.P., 2013. Estimating optimal complexity for ecological niche models: a jackknife approach for species with small sample sizes. Ecol. Modell. 269, 9–17, http://dx.doi.org/10.1016/j.ecolmodel.2013.08. 011. Shrestha, U.B., Bawa, K.S., 2014. Impact of climate change on potential distribution of chinese caterpillar fungus (Ophiocordyceps sinensis) in Nepal himalaya. PLoS One 9, e106405, http://dx.doi.org/10.1371/journal.pone.0106405. Shrestha, U.B., Gautam, S., Bawa, K.S., 2012. Widespread climate change in the himalayas and associated changes in local ecosystems. PLoS One 7, e36741. Singh, R.B., Mal, S., 2014. Trends and variability of monsoon and other rainfall seasons in Western Himalaya. India. Atmos. Sci. Lett. 15, 218–226, http://dx. doi.org/10.1002/asl2.494. Singh, S.P., Singh, J.S., 1992. Forests of the Himalayas: Structure, Function and Impact of Man. Gyanodaya Prakashan, Nainital, India. Singh, S., Rawat, Y., Garkoti, S., 1997. Failure of brown oak(Quercus semecarpifolia) to regenerate in central Himalaya: a case of environmental semisurprise. Curr. Sci. 73, 371–374. Singh, G., Rai, I.D., Rawat, G., 2012. The mortality of banj oak (Quercus leucotrichophora A. Camus) trees in Mussoorie, Uttarakhand: is it an alarming call for rapid degradation? Curr. Sci. 102, 1622–1623. Singh, V., Thadani, R., Tewari, A., Ram, J., 2014. Human influence on banj oak (Quercus leucotrichophora, A. camus) forests of central himalaya. J. Sustainable For. 33, 373–386, http://dx.doi.org/10.1080/10549811.2014.899500. Sohl, T.L., 2014. The relative impacts of climate and land-use change on conterminous United States bird species from 2001 to 2075. PLoS One 9, e112251. Swanson, A.K., Dobrowski, S.Z., Finley, A.O., Thorne, J.H., Schwartz, M.K., 2013. Spatial regression methods capture prediction uncertainty in species distribution model projections through time. Global Ecol. Biogeogr. 22, 242–251, http://dx.doi.org/10.1111/j.1466-8238.2012.00794.x. Syfert, M.M., Smith, M.J., Coomes, D.A., 2013. The effects of sampling bias and model complexity on the predictive performance of MaxEnt species distribution models. PLoS One 8, e55158, http://dx.doi.org/10.1371/journal. pone.0055158. Telwala, Y., Brook, B.W., Manish, K., Pandit, M.K., 2013. Climate-induced elevational range shifts and increase in plant species richness in a Himalayan biodiversity epicentre. PLoS One 8, e57103.

A. Chakraborty et al. / Ecological Engineering 97 (2016) 593–609 Thuiller, W., 2004. Patterns and uncertainties of species’ range shifts under climate change. Global Change Biol. 10, 2020–2027, http://dx.doi.org/10.1111/j.13652486.2004.00859.x. Trenberth, K., 2010. More knowledge, less certainty. Nat. Rep. Clim. Change, 20–21, http://dx.doi.org/10.1038/climate.2010.06. Trumbore, S., Brando, P., Hartmann, H., 2015. Forest health and global change. Science 349, 814–818. Turner, D.P., Conklin, D.R., Bolte, J.P., 2015. Projected climate change impacts on forest land cover and land use over the Willamette River Basin, Oregon, USA. Clim. Change 1–14, 4, http://dx.doi.org/10.1007/s10584-015-1465-4. Urban, M.C., 2015. Accelerating extinction risk from climate change. Science 348, 571–573, http://dx.doi.org/10.1126/science.aaa4984. Vayssières, M.P., Plant, R.E., Allen-Diaz, B.H., 2000. Classification trees: an alternative non-parametric approach for predicting species distributions. J. Veg. Sci. 11, 679–694, http://dx.doi.org/10.2307/3236575. Velásquez-Tibatá, J., Salaman, P., Graham, C.H., 2013. Effects of climate change on species distribution, community structure, and conservation of birds in protected areas in Colombia. Reg. Environ. Change 13, 235–248. Vetaas, O.R., 2000. The effect of environmental factors on the regeneration of Quercus semecarpifolia Sm. in Central Himalaya, Nepal. Plant Ecol. 146, 137–144, http://dx.doi.org/10.1023/A:1009860227886. Vuuren, D.P., van, Edmonds, J., Kainuma, M., Riahi, K., Thomson, A., Hibbard, K., Hurtt, G.C., Kram, T., Krey, V., Lamarque, J.-F., Masui, T., Meinshausen, M., Nakicenovic, N., Smith, S.J., Rose, S.K., 2011. The representative concentration pathways: an overview. Clim. Change 109, 5–31, http://dx.doi.org/10.1007/ s10584-011-0148-z. Wang, T., Wang, G., Innes, J., Nitschke, C., Kang, H., 2015. Climatic niche models and their consensus projections for future climates for four major forest tree species in the Asia-Pacific region. For. Ecol. Manage., http://dx.doi.org/10.1016/ j.foreco.2015.08.004.

609

Warren, D.L., Seifert, S.N., 2011. Ecological niche modeling in Maxent: the importance of model complexity and the performance of model selection criteria. Ecol. Appl. 21, 335–342. Warren, D.L., Glor, R.E., Turelli, M., 2010. ENMTools: a toolbox for comparative studies of environmental niche models. Ecography 33, 607–611, http://dx.doi. org/10.1111/j.1600-0587.2009.06142.x. Wiens, J.A., Stralberg, D., Jongsomjit, D., Howell, C.A., Snyder, M.A., 2009. Niches, models, and climate change: assessing the assumptions and uncertainties. Proc. Natl. Acad. Sci. 106, 19729–19736. Wisz, M.S., Hijmans, R., Li, J., Peterson, A.T., Graham, C., Guisan, A., 2008. Effects of sample size on the performance of species distribution models. Divers. Distrib. 14, 763–773. Woodward, F.I., Williams, B.G., 1987. Climate and plant distribution at global and local scales. Vegetatio. Adv. Veg. Sci. 69, 189–197. Wright, A.N., Schwartz, M.W., Hijmans, R.J., Shaffer, H.B., 2015. Advances in climate models from CMIP3 to CMIP5 do not change predictions of future habitat suitability for California reptiles and amphibians. Climatic Change 1–13, 3, http://dx.doi.org/10.1007/s10584-015-1552-6. Yamaguchi, Y., Kahle, A.B., Tsu, H., Kawakami, T., Pniel, M., 1998. Overview of advanced spaceborne thermal emission and reflection radiometer (ASTER). IEEE Trans. Geosci. Remote Sens. 36, 1062–1071, http://dx.doi.org/10.1109/36. 700991. Zhu, K., Woodall, C.W., Clark, J.S., 2012. Failure to migrate: lack of tree range expansion in response to climate change. Global Change Biol. 18, 1042–1052, http://dx.doi.org/10.1111/j.1365-2486.2011.02571.x.