Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/taylor-francis/habitat-suitability-modelling-for-lagotis-cashmeriana-royle-rupr-a-Css9ZDhsTG
References
References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.
- Publisher
- Taylor & Francis
- Copyright
- © 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group on behalf of the International Water, Air & Soil Conservation Society(INWASCON).
- ISSN
- 2474-9508
- DOI
- 10.1080/24749508.2020.1816871
- Publisher site
- See Article on Publisher Site
Abstract
GEOLOGY, ECOLOGY, AND LANDSCAPES 2022, VOL. 6, NO. 4, 241–251 INWASCON https://doi.org/10.1080/24749508.2020.1816871 RESEARCH ARTICLE Habitat suitability modelling for Lagotis cashmeriana (ROYLE) RUPR., a threatened species endemic to Kashmir Himalayan alpines Nadeem Salam , Zafar A. Reshi and Manzoor A. Shah Department of Botany, University of Kashmir, Srinagar, India ABSTRACT ARTICLE HISTORY Received 30 March 2020 Rare and endemic species comprise globally a priority conservation concern in view of being at Accepted 6 August 2020 a higher risk of extinction. Recording the occurrence data for such species, especially in hardly accessible alpine habitats, is a rather challenging task. Modelling serves as effective tool for KEYWORDS predicting habitat suitability and practising artificial introductions for such species with Conservation; Maxent; encouraging conservation implications. We used Maxent modelling to predict the habitats Lagotis cashmiriana; ground suitable for Lagotis cashmiriana (ROYLE) RUPR., a threatened species endemic to Kashmir validation; reintroduction Himalaya. Our modelling approach consisted of two rounds of modelling and one round of ground validations. The first round of modelling was based on the published literature and herbarium records and the second round included the new records that were obtained from ground validations based on first model predictions. Through this approach, we were able to identify several new populations of L. cashmiriana and reported a significant increase in its range size. We also identified the suitable areas for reintroduction of L. cashmiriana in the western Himalayan region after identifying a broad range of environmental factors that determine the distribution of this species. Introduction species requires a detailed knowledge about the suita- ble habitats for the species occurrence. However, From the last few decades several factors like climate assessment of suitable habitats has been done for change, habitat fragmentation, alien species invasion, a small number of species due to time constraints. over-exploitation and pollution have emerged as Ecological Niche Modelling tools serve as one of the major threats to global biodiversity. Of particular con- most appropriate solutions in solving the problem of cern are threatened and endemic species. conservation in many ways (Botts et al., 2012; Characterised by restricted geographic ranges, high Huisman & Millar, 2013). They aid in estimating the degree of habitat specialization, small population distribution of species in conservation assessments size, low reproductive capacity and limited geographic and have been widely applied in conservation biology distribution, these species are at greater risk as com- to predict the potential distributions of species pared to other widely distributed species (Markham, (Chefaoui et al., 2005; Peterson & Vieglais, 2001; 2014; McKinney, 1999; Myers et al., 2000; Silva et al., Rushton et al., 2004). They are now being increasingly 2017). At the same time these species have important used to determine various factors which govern the role in ecosystem functioning as they help in main- distribution of species (Elith et al., 2006; Guisan & taining the ecosystem diversity, make ecosystems Zimmermann, 2000; Kozak et al., 2008). Recently, resistant to invasion and act as indicators of general numerous modelling techniques have been success- patterns of species diversity (Lyons et al., 2005; Lyons fully applied in case of artificial introductions or & Schwartz, 2001), thus, substantiating their conserva- selecting appropriate sites for their conservation and tion on priority basis. management (Elith & Leathwick, 2009; Gaston, 1996; Effective conservation strategies of threatened spe- Pecl et al., 2017; Watson et al., 2014). Recent develop- cies demand a proper knowledge of their geographic ments in Ecological Niche Modelling (ENM) have distributions and determining suitable areas for their explored applications to diverse conservation issues, reintroduction (Nazeri et al., 2010; Polak & Saltz, including suitable habitat and species range estimates 2011; Rodríguez-Salinas et al., 2010; Seddon et al., (Bellard et al., 2012; Chefaoui et al., 2005; Gaubert 2014). However, data about geographic distribution et al., 2006; Gritti et al., 2013), protected area prior- of plants remains often biased for being collected itization and network design (Huisman & Millar, from easily accessible areas and during specific periods 2013; Ishihama et al., 2019; Margules & Austin, 1994; of the year (Funk & Richardson, 2002; Hijmans et al., Rondinini et al., 2005; Sanchez-Cordero et al., 2005; 2000). Reintroduction of threatened and endemic Solano & Feria, 2007; Thorn et al., 2009), effects of CONTACT nadeem salam nadeemsalam22@gmail.com Department of Botany, University of Kashmir, Srinagar, Jammu and Kashmir 190 006, India © 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group on behalf of the International Water, Air & Soil Conservation Society(INWASCON). This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 242 N. SALAM ET AL. habitat disturbance on species distributions (Araújo & species. Keeping in view the above-mentioned sce- Peterson, 2012; Banks et al., 2005; Bing et al., 2019), to nario we used Ecological Niche Modelling approach aid in IUCN categorization of species (Pena et al., to model Lagotis cashmiriana with following objec- 2014) and projecting future distributions under cli- tives: (i) to develop reliable, statistically accurate, pre- mate change (Bellard et al., 2012; Franco et al., 2018; diction maps depicting the potential distribution of Moraitis et al., 2019; Wang et al., 2018). ENM the species, (ii) to find the suitable combinations of approach combines species occurrence data with eco- environmental variables driving distribution of spe- logical/environmental variables (temperature, precipi- cies, (iii) to locate new populations and assess the tation, elevation, geology, and vegetation) to create population status in the predicted habitats through a model representing species distributions compatible field validation and relate it with model thresholds, with the environment (Elith & Leathwick, 2009). and (iv) to identify suitable habitats for species Availability of high-resolution satellite imageries, reintroductions. downscaling tools for environmental variables and interpolated spatial datasets on climate and vegetation Methodology has enhanced the accuracy of prediction of the models manifold. Presently there is a wide range of environ- Study area mental niche models for studying species distributions We focussed on the Kashmir Himalaya a part of such as Bioclim (Busby, 1991), Domain (Carpenter Indian Himalayan region situated in the northwest et al., 1993), linear, multivariate and logistic regres- Himalayan biogeographic zone between 33°20–34° sions (Mladenoff et al., 1995), generalized linear mod- 54 N latitudes and 73°55–75°35E longitudes, covering elling and generalized additive modelling (Frescino an area of 16, 000 km2. Two of the 12 important et al., 2001), discriminant analysis (Manel et al., biogeographic zones of India namely Northwest 1999), genetic algorithms (Stockwell & Peters, 1999), Himalaya (including the Kashmir region together artificial neural networks (Manel et al., 1999; Moisen with Tilel, Gurais, Keran and Karnah) and Trans- & Frescino, 2002), and support vector machines (Guo Himalaya (including the Ladakh region) occur in this et al., 2005). region (Rogers & Panwar, 1988). The Kashmir Kashmir Himalaya which is a part of Indian Himalayan region is comprised of lofty mountains of Himalayan region occupies a pivotal position in repre- the Pir Panjal in the South and Southwest and by the senting a unique biospheric unit in the region. The Great Himalayan range in the North and East with mountainous region lies between 32°20ʹ–34°50ʹ North a deep elliptical bowl-shaped Kashmir valley in the latitude and 73°55ʹ–75°35ʹ East longitude . Two thou- middle. The altitude of the region ranges from 1500 sand plant species have been recorded from the region (Valley) to 5, 420 m Kolahoi (the highest peak). (Dar et al., 2002), grouped under 710 genera and 132 families out of which 8% species are exclusively ende- mic to Kashmir. Despite the region comprises of only Study species 0.48% landmass of India (Dar et al., 2008) its contri- Lagotis cashmiriana Royle Rupr. (Kashmir Hare’s Ear) bution to the country’s angiosperm flora is 12% of is a perennial herb of family Scrophulariaceae found in which 3% are endemic (Dar et al., 2012). Most of alpine slopes of Kashmir Himalayas. It is found at an these endemic species are restricted to the alpine and altitude ranging from 3,000 to 4,000 m. It is usually sub-alpine habitats. Three hundred and fifty-five spe- found in shady and wet areas and among rock cre- cies of plants have been rendered threatened due to vices. The species is not only over-exploited in view of various anthropogenic activities, such as habitat loss myriad medicinal uses but its individuals are also or modification, over-exploitation of economically damaged by herbivores in various populations. important plants, alien species invasion, unchecked Further, landslides, excessive tourist flow and con- grazing, agricultural expansion, unplanned develop- struction of roads are the other factors which in con- ment and influx of tourists (Dar et al., 2008; Dar & junction with hostile habitat conditions contribute to Naqshi, 2002). Lagotis cashmiriana an endemic and the present threat status of this endemic species. threatened plant species of Kashmir Himalaya is facing an imminent threat due to over-grazing, fragile habitat, landslides, excessive tourist flow, construction The modelling framework of roads and over-exploitation for local use (Dar et al., Species distribution modelling and ground 2006; Tali et al., 2014). Earlier Dar et al. (2006) carried validations studies on reproductive ecology and Exsitu conserva- tion strategies of Lagotis cashmiriana as a means for its We followed stepwise modelling approach to get recovery and restoration. Identification of suitable insights into the ecology and distribution of habitats for reintroduction of Lagotis is the next L. cashmiriana. Our modelling approach consisted of important step in recovery and restoration of the two rounds of Species Distribution Modelling (SDM) GEOLOGY, ECOLOGY, AND LANDSCAPES 243 and one round for ground-validation processes. We Google Earth to ascertain the actual habitat condition based our initial model on fourteen secondary occur- prevailing in the areas of occurrence. rence records obtained through herbarium specimens from Kashmir University Herbarium (KASH), pub- Environmental data lished sources and field researchers. All these records were confirmed through repeated field surveys and We used bioclimatic variables to model the distribu- accordingly were fitted in the model. Prior to fitting tions of L. cashmiriana. These GIS data sets character- the occurrence localities in our models we considered ize global climates from 1950 to 2000 using average issue of spatial autocorrelation among the presence monthly weather station data and are available at localities. Although our surveys focussed on collecting different spatial resolutions. (Hijmans et al., 2005) coordinates from the entire region of our study; how- and are known to influence species distributions ever, the records which were used to build the first (Root et al., 2003). To fit the models at the local model were spatially autocorrelated. We used “SDM scale, besides climate we explored the possible effects tool box” (Brown, 2014) to remove spatially autocor- of several other types of physiographic factors such as related localities. elevation and slope (Table 1). Climatic predictors were We carried out extensive surveys during 2014, 2015 obtained from WorldClim (Hijmans et al., 2005) with and 2016 to explore the robustness and pertinence of a spatial resolution of 30 arc-seconds (http://world the model in predicting the population status of the clim.org/current) and resampled to a 500 m cell size species in each locality of occurrence as predicted for usage in the regional scale models. All spatial under various model thresholds. We surveyed twenty procedures were implemented in ArcGIS 10.1. We new sites across Kashmir Himalayan region as pre- tested all the predictor variable for pair-wise correla- dicted by the model to be highly suitable for the tions using the Spearman’s rank correlation test, and species. The newly located sites (new populations) only those with a correlation coefficient lower than were simultaneously recorded. These new records 0.85 were taken (Elith et al., 2006;). Based on the were added to previous records and the final model correlation and taking into consideration the species was created to predict the suitable habitats for ecology and the importance of extreme environmental L. cashmeriana. To select localities for sampling, we conditions, a set of eleven variables was finally selected converted the model predictions to binary maps (sui- to fit our models. table/unsuitable) using the lowest presence threshold (LPT; also known as minimum training presence), i.e., Habitat suitability modelling the lowest prediction value returned by Maxent for a location with observed presence of the species. SDM software Maxent 3.3.2 (Phillips et al., 2006) Assessment of the actual habitat type of the species was used to estimate the potential range of the in the localities of occurrence was done through species. Maxent predicts the distribution of species repeated field surveys. We also superimposed the pre- using the principle of maximum entropy, which dicted potential areas on Google Earth Ver.6 (www. finds the probability that is the closest to uniform google.com/earth) imageries for habitat quality assess- combining environmental data with presence local- ment. The predicted suitability maps were exported in ities and background records sampled from the KMZ format using Diva GIS ver. 7.3 (www.diva-gis. overall study area (Phillips et al., 2006). Since the org). KMZs are zipped Keyhole Mark up Language sample size for our species was low, we used only (KML) files which specify a set of features such as the linear and quadratic features (Phillips et al., place marks, images, polygons, 3D models or textual 2004). All other parameters were maintained at descriptions for display in Google Earth. The exported default settings. As recommended by Phillips et al. KMZ files were overlaid on satellite imageries in (2006) we used default settings. The Maxent default Table 1. Environmental variables used in modelling L. cashmiriana. Predictors Source Annual Mean Temperature World clim; Hijmans et al. (2005) Mean Diurnal Range (Mean of monthly (max temp – min temp) World clim; Hijmans et al. (2005) Isothermality (BIO2/BIO7) (* 100) World clim; Hijmans et al. (2005) Max Temperature of Warmest Month World clim; Hijmans et al. (2005) Mean Temperature of Coldest Quarter World clim; Hijmans et al. (2005) Precipitation of Driest Month World clim; Hijmans et al., (2005) Precipitation Seasonality (Coefficient of Variation) World clim; Hijmans et al. (2005) Precipitation of Wettest Quarter World clim; Hijmans et al. (2005) Precipitation of Warmest Quarter World clim; Hijmans et al. (2005) Elevation USGS Hydro-1 K dataset) Slope USGS Hydro-1 K dataset 244 N. SALAM ET AL. setting removes duplicate presence record. This Results programme has been in vogue for quite a long Habitat suitability modelling time now to model species distributions since its inception. The model works on presence only data Our first model showed most of the alpine habitats of of species records. Like other SDMs, it estimates Kashmir Himalaya to be highly suitable for the relationship between species records at sites L. cashmiriana which is rather coherent with the and the environmental and/or spatial characteristics described geographic range of the species. Besides, of those sites (Franklin, 2009). In comparison with northern parts of Pakistan (Pakistan occupied other presence only Species Distribution Models Kashmir) were also predicted to be the suitable habi- (SDMs), Maxent has been found more valid in tats for the target species (Figure 1). The extensive field delimiting the species fundamental niche (Elith surveys carried out during 2014, 2015 and 2016 based et al., 2006). While running the Maxent model on our model thresholds covered almost the entire one can increase the number of replicates which predicted area. At five sites (Sinthan Top, Pehjan, facilitates the cross validation and subsequently the Peer Ki Gali, Kousarnag and Margan) new popula- model calibration. Maxent uses machine learning tions of L. cashmiriana were successfully located. Our technique which estimates the distribution of final model (2) which included the newly identified a species while conforming the empirical averages sites showed rather widespread distribution adding of the climate information associated with the some parts of Himachal, Uttrakhand and additional occurrence data (Phillips et al., 2004). It is one parts of Pakistan occupied Kashmir Himalaya to be amongst the “presence-only” group of species dis- also suitable. tribution modelling methods which has been widely used and has the capacity to handle low sample Model calibration and factors determining species sizes. distribution To validate the model robustness, we executed 20 replicated model runs for the species with Our both the models attained an AUC value of>0.90 a threshold rule of 10 percentile training presence. (0.99 ± 0.0009, 0.99 ±.0004) and thus can be consid- In the replicated runs, we employed cross- ered as excellent . Both can successfully discriminate validation technique where samples were divided between suitable and unsuitable habitats. Jackknife into replicate folds and each fold was used for tests of model validations also confirmed that test data. Other parameters were set to default as Maxent predicted the species’ occurrence significantly the program is already calibrated on a wide range better than random expectations (p < 0.05). Based on of species datasets (Phillips & Dudik, 2008). Model the analysis of variable contribution as given by quality was evaluated based on Area Under Curve Maxent, Precipitation Seasonality had the highest con- (AUC) value and the model was graded following tribution in both the models followed by Mean Thuiller et al. (2005) as: poor (AUC < 0.8), fair (0.8 Diurnal Range and Precipitation of Coldest Quarter < AUC < 0.9), good (0.9 < AUC < 0.95) and very as shown in Table 2 and Figure 2. The response curves good (0.95 < AUC < 1.0). Further, potential area of for the environmental predictors most determinant distribution and/or reintroduction were categorized for the species distribution of L. cashmiriana are pre- into five classes based on logistic threshold of sented in Figure 3. Overall, the response curves reveal 10 percentile training presence, i.e., very-high (0.- that the species is mainly distributed in areas with 762–1), high (0.572–0.761), medium (0.381–0.571), lower values of Precipitation of Coldest Quarter, low (0.325–0.570) and very low (0–0.324). The lack Precipitation Seasonality and low to medium tempera- of absence data especially for those species which tures, which is coherent with the known distribution have not been well documented and that are rare of the species along the north-western Himalaya poses a major limitation of many studies of species (Kashmir Himalaya). distributions (Chefaoui and Lobo, 2008; Wisz & Guisan, 2009). However, several authors have Habitats for reintroduction (Chefaoui and Lobo, 2008; Elith et al., 2006; Graham et al., 2004) proposed random creation of Our field surveys and habitat analysis in the species pseudo-absences as an alternative way to overcome occupied habitats showed that L. cashmiriana occu- these limitations of presence only datasets, thus pies the alpine slopes with moist and rocky habitats. It making the predictions more reliable and accurate sometimes grows in rock crevices and prefers pebbled (Elith et al., 2006). Pseudo-absences were generated and sandy soils at an altitudinal range of 3000–4000 m by selected randomly assigning unoccupied grid (Table 3). Superimposing the predicted potential habi- cells within a polygon containing the collectively tat map of the species on Google Earth satellite ima- known distribution of each species within the geries revealed a mosaic of habitats to be suitable for study region. the species persistence (Figure 1(c,d)). High to very GEOLOGY, ECOLOGY, AND LANDSCAPES 245 Figure 1. Habitat suitability map for L. cashmiriana: (a) map based on initial records, (b) map based on final species occurrence data points, (c) highly suitable areas for reintroduction of L. cashmiriana and (d) habitat suitability using Google Earth imageries. 2 2 Table 2. Estimates of relative contributions of the predictor medium, 132 km highly suitable and 117 km very environmental variables to the MaxEnt model. highly suitable areas for L. cashmiriana. Percent contribution Model Model Discussion Variable 1 2 Mean Diurnal Range (Mean of monthly max temp – 28.2 35.4 Rare and endemic species have acquired top priority for min temp) Precipitation Seasonality (Coefficient of Variation) 38.2 33.3 conservation worldwide because these species are at Mean Temperature of Coldest Quarter 6.4 11 higher risk of extinction. Mapping potential habitats Temperature Annual Mean 7.2 10.3 Isothermality (BIO2/BIO7) (* 100) 15.2 6 for rare and endemic species can aid in conservation Precipitation of Driest Month 1.6 1 planning and management. We used Ecological Niche Elevation 0.5 0.2 Modelling approach as an important tool for conserva- Max Temperature of Warmest Month 0.7 0.1 Precipitation of Warmest Quarter 0.2 0.2 tion of Lagotis cashmiriana an endemic and threatened Precipitation of Wettest Quarter 0.2 0.3 plant species of Kashmir Himalaya. Our models fitted Slope 0.3 2.0 with both climatic and non-climatic predictors depict, from a robust modelling approach, the potential range high habitat suitable areas for the species were con- of the species besides identifying the most suitable areas tinuous alpine patches of north-western Himalayan for its occurrence. Moreover, our models were success- region. Medium to low habitat suitability areas were ful in predicting the previous distribution range of the subalpine slopes among evergreen forests. The species target species and identifying highly suitable areas was found to be closely associated with Juniper spp., which are coincident with grid cells where the species Rhododendron spp. and Bergenia spp. which form have not been recorded yet. Based on our model pre- thick mats and help to maintain moist conditions dictions followed by extensive field surveys we were able and also serve as safe refuge for the species. Besides to locate five new populations of L. cashmiriana thus in certain instances the species occurred between rock validating our spatial projections. Our spatial projec- crevices assisted by moist habitats. Analysis of habitat tions can support targeted surveys to collect additional suitability under current climatic conditions reveals records for the species, help identifying source and sink that overall suitable area for species reintroduction is populations, and support the selection of populations to 2 2 2 951 km of which 472 km is less suitable, 230 km target urgent conservation measures. 246 N. SALAM ET AL. Figure 2. Results of jackknife evaluation procedure on the relative importance of predictor variables for L. cashmiriana for model 1 (a) and model 2(b). Our study also explains the role of habitat suitabil- However, while taking species reintroduction plans ity modeling in identifying the habitats for reintroduc- into consideration, appropriate habitats should be tion of threatened and endemic plant species. Analysis carefully selected based on field observations. of habitat suitability under current climatic conditions Reintroduction of the target species in the identified revealed that overall suitable area for species reintro- habitats would help a great deal in rehabilitating the duction is 951 km for L. cashmiriana of which species population and in improving conservation sta- 2 2 2 472 km is less suitable, 230 km medium, 132 km tus hence in conserving the overall biodiversity of the highly suitable and 117 km very highly suitable. The region. predicted suitable areas compose a mosaic of habitats Our study can provide a road map for applying including alpine rocky slopes, grasslands, pastures and distribution modelling as a conservation tool for also forest areas in upper reaches. Rocky alpine slopes, other species which are threatened and need immedi- moist areas near alpine streams, and habitats among ate conservation efforts. Our Model predictions and Juniperous patches and Betula patches are among high field assessments reveal that L. cashmiriana has probability areas for the species; hence, these areas a limited potential distribution. Majority of the pre- could be used for in situ conservation of the species. dicted habitats are also highly affected by human GEOLOGY, ECOLOGY, AND LANDSCAPES 247 Figure 3. Maxent response curves (logistic output: probability of presence) for predictors with highest contribution for L. cashmiriana (a, b and c) for final model. Table 3. Habitat characteristics of the species occupied sites. Locality Coordinates Habitat Major associated Species Sinthan N33 ° 34.859 Steep sloppy with longer patches of soil and Rhododendron spp., Bergenia spp., Picrorhiza spp. Rheum Top E75 ° 36.518 intermingled by big sedimentary rocks wibianum, Corydalis cashmeriana, Juniper spp. Pehjan N33°57 .012 Very steep sloppy, thin pebelled soil as well as big Rhododendron, Aquilegia pubiflora , E74°22 836’ sedimentary rocks throughout Podophyllum spp., Lagotis cashmeriana, Juniper spp. Peer Ki N33°37.797 Extremely steep slope with thin pebelled soil layer Rhododendron spp., Corydalis cashmeriana, Bergenia spp. Gali E74° 31.217 on rocky slope Kousar N 33°30 .320 Very steep slope, thin pebelled soil as well as big Picrorhiza spp. Rheum spp., Corydalis cashmeriana. nag E 74° 46 .780’ sedimentary rocks Margan 33°45 26.8452 ’N75° Steep slope, pebelled soil, shady moist areas Rhododendron spp., Bergenia spp.,Picrorhiza spp.Rheum spp. top 29 15.5796 ’E activities like road construction, over-exploitation, modelling technique. Similarly, De Siqueira et al. habitat fragmentation etc. Thus there is an immediate (2009) modelled a rare plant with only seven presence need to go for conservation measures both Insitu and records by using GARP software. In our study we also Exsitu. Earlier Dar et al. (2006) worked on vegetative used Maximum Entropy Modelling approach as it has propagation methods for L. cashmiriana however been one of the most widely used among the best these studies need to be further supplemented by performing methods. Maxent has a high predictive micropropagation techiniques for mass multiplication performance for both small and large sample sizes of the target species. The plants can then be reintro- (Elith et al., 2006; Hernandez et al., 2006; Pearson duced in their natural suitable habitats using et al., 2007; Wisz et al., 2008). Our models provide an Ecological Niche Modelling tool. excellent discriminatory ability following Thuiller Recently, several studies have used numerous mod- (2003) showing AUC values above 0.90 for elling approaches to successfully map distributions of L. cashmiriana. Our final models which were calibrated species with fewer occurrences. For instance, Pearson with rather large number of occurrences predicted et al. (2007) used only five species occurrence records comparatively larger areas as suitable in comparison and modelled Geckos in Madagascar by using Maxent with known distribution, thus increasing the chances to 248 N. SALAM ET AL. add new locations of occurrence for the species. In our temperatures, which would support more accurate fore- study many areas predicted by our models as suitable casts if climate change scenarios are applied. were without the target species. Possible reasons for We agree that the variables that we selected for our this could be either due to commission error by the study did not account for species dispersal and biotic model or because of the inability of the species to interactions which are important factors for determin- disperse to these locations. Discovery of additional ing species distributions (Kearney & Porter, 2009); populations of the target species is important since however, we do not possess enough biological knowl- the species current habitats are being rapidly fragmen- edge about the target species to account for such ted by humans. Extensive field surveys are required to interactions. further identify unrecovered populations. For this pur- pose those areas should be focused first which are Conclusion predicted as highly suitable so that new populations can be discovered in the short term. L. cashmeriana is an endemic and threatened plant The choice of suitable number and combination of species of Kashmir Himalayan region. It is facing an environmental predictors is crucial when modelling rare imminent threat due to over-grazing, fragile habitat, and endemic species. It has been observed in ecological landslides, excessive tourist flow, construction of science that a few variables account for about 95% of the roads and over-exploitation for local use, thus variation in distribution. Hence while modelling species demanding its immediate conservation. We used distributions suitable number of relevant environmental Ecological Niche Modelling approach as a first- variables should be selected. Too many variables can hand conservation tool to predict suitable habitats lead to over-fitting and over-prediction of distributions for L. cashmiriana. Our results complement the and too fewer variables can lead to under prediction. growing body of literature that indicates the signifi - The inclusion of climatic and topographic information, cance of Ecological Models to predict potential spe- to a certain extent, minimizes the potential source of cies distributions, identify new populations of error in prediction. However, with the improvement in threatened and endemic species and to locate suita- technology both spatially and temporally, there could be ble habitats for species reintroductions. On the basis better availability of data on vegetation, bioclimatic and of model predictions, we successfully reconfirmed topography for reliable prediction on the species distri- previous populations and located five new popula- bution pattern. In our study we included both climatic tions of L. cashmiriana. We were able to identify and topographic variables which we believe have direct most suitable habitats were the target species can be relevance to the species. The temperature variables reintroduced. We believe that well-designed exten- describe the thermal tolerance of the species, the annual sive field surveys in the predicted regions will further precipitation describes water availability and the physio- improve the estimates of range size, which may likely graphic factors describe the macro-habitat characteris- reduce the current threat status for L. cashmiriana. tics of the species. Out of nineteen bioclimatic variables Although we have successfully predicted distribution we cautiously selected only those variables which were under current climate conditions, the distribution of not highly correlated and which gave maximum con- L. cashmiriana needs to be studied under future tribution to the maxent model. Our models show that climate by applying various climate change scenar- a mixture of climatic and non-climatic variables is ios. Our modelling method can be applied for other needed to explain endemic species distributions in endemic and threatened species of the region. Western Himalaya. The response curves reveal that the species is mainly distributed in areas with lower values of Precipitation of Coldest Quarter, Precipitation Acknowledgements Seasonality and low to medium temperatures, which is We are thankful to Deobandu Adhikari, NEHU, Shillong, coherent with the known distribution of the species India for his help and very useful discussions regarding the along the north-western Himalaya (Kashmir ecological niche modelling. Authors are also highly thankful Himalaya). Our model can also inform in advance the to the editor and reviewers for their constructive comments range dynamics of the target species under climate and secessions in improving the manuscript. change scenarios as reflected clear involvement of cli- matic variables in delimiting the distribution of L. cashmiriana. Climate change has been found to Disclosure statement cause distribution shifts in many plant species (Hoegh- No potential conflict of interest was reported by the authors. Guldberg et al., 2008) and narrowly endemic species are believed to face more extinction risks as compared to others species (Thomas et al., 2004). Specifically for our ORCID test species the model highlighted a larger dependence on features of the precipitation regime and low Nadeem Salam http://orcid.org/0000-0001-9865-6910 GEOLOGY, ECOLOGY, AND LANDSCAPES 249 Proceedings of International Seminar on Multidisciplinary References Approaches in Angiosperm Systematics. Kolkata Araújo, M. B., & Peterson, T. (2012). Uses and misuses of Dar, G. H., & Naqshi, A. R. (2002). Threatened flowering bioclimatic envelope modeling. Ecology, 93(7), plants of Kashmir Himalaya– A checklist. Oriental 1527–1539. https://doi.org/10.1890/11-1930.1 Journal of Science, 6(1), 23–53. Banks, S. C., Finlayson, G. R., Lawson, S. J., Elith, J., Graham, C. H., Anderson, R. P., Dudik, M., Lindenmayer, D. B., Paetkau, D., Ward, S. J., & Ferrier, S., Guisan, A., Hijmans, R. J., Huettmann, F., Taylor, A. C. (2005). The effects of habitat fragmentation Leathwick, J. R., Lehmann, A., Li, J., Lohmann, L. G., due to forestry plantation establishment on the demogra- Loiselle, B. A., Manion, G., Moritz, C., Nakamura, M., phy and genetic variation of a marsupial carnivore, Nakazawa, Y., Overton, J. M. M., Peterson, A. T., Antechinus Agilis. Biological Conservation, 122(4), Phillips, S. J., . . . Zimmermann, N. E. (2006). Novel 581–597. https://doi.org/10.1016/j.biocon.2004.09.013 methods improve prediction of species’ distributions Bellard, C., Bertelsmeierm, C., Leadley, P., Thuiller, W., & from occurrence data. Ecography, 29(2), 129–151. Courchamp, F. (2012). Impacts of climate on the future of https://doi.org/10.1111/j.2006.0906-7590.04596.x biodiversity. Ecology Letters, 15(4), 365–377. https://doi. Elith, J., & Leathwick, J. R. (2009). Species distribution org/10.1111/j.1461-0248.2011.01736.x models: Ecological explanation and prediction across Bing, X. W., Svenning, J. C., Chen, G. K., Zhang, M. G., space and time. The Annual Review of Ecology, Huang, J. H., Chen, B., Ordonez, A., & Ma, K. (2019). Evolution, and Systematics, 40(1), 677–697. https://doi. Human activities have opposing effects on distributions org/10.1146/annurev.ecolsys.110308.120159 of narrow-ranged and widespread plant species in China. Franco, J. N., Tuya, F., Bertocci, I., Rodríguez, L., Proceedings of the National Academy of Sciences of the Martínez, B., Sousa-Pinto, I., & Arenas, F. (2018). The United States of America, 116(52), 26674–26681. https:// ‘golden kelp’ Laminaria ochroleuca under global change: doi.org/10.1073/pnas.1911851116 Integrating multiple eco-physiological responses with Botts, E. A., Erasmus, B. F., & Alexander, G. (2012). species distribution models. Journal of Ecology, 106(1), Methods to detect species range size change from biolo- 47–58. https://doi.org/10.1111/1365-2745.12810 gical atlas data: A comparison using the South African Franklin, J. (2009). Mapping species distributions: Spatial frog atlas project. Biological Conservation, 146(1), 72–80. inference and prediction. Cambridge University Press. https://doi.org/10.1016/j.biocon.2011.10.035 Frescino, T. S., Edwards, T. C., & Moisen, G. G. (2001). Brown, J. L. (2014). SDM toolbox: A python-based GIS Modeling spatially explicit forest structural attributes toolkit for landscape genetic, biogeographic and species using generalized additive models. Journal of Vegetation distribution model analyses. Methods in Ecology and Science, 12(1), 15–26. https://doi.org/10.1111/j.1654- Evolution, 5(7), 694–700. https://doi.org/10.1111/2041- 1103.2001.tb02613.x 210X.12200 Funk, V. A., & Richardson, K. S. (2002). Systematic data in Busby, J. R. (1991). BIOCLIM – A bioclimate analysis and biodiversity studies: Use it or lose it. Systematic Biology, prediction system. In C. R. Margules & M. P. Austin 51(2), 301–313. https://doi.org/10.1080/10635150252 (Eds.), Nature conservation: Cost effective biological sur- veys and data analysis (pp. 64–68). CSIRO. Gaston, K. J. (1996). What is biodiversity? In K. J. Gaston Carpenter, G., Gillison, A. N., & Winter, J. (1993). (Ed.), Biodiversity: A biology of numbers and difference DOMAIN: A flexible modeling procedure for mapping (pp. 1–9). Blackwell Science. potential distributions of plants and animals. Biodiversity Gaubert, P., Papes, M., & Peterson, A. T. (2006). Natural and Conservation, 2(6), 667–680. https://doi.org/10.1007/ history collections and the conservation of poorly known BF00051966 taxa: Ecological niche modeling in central African rain- Chefaoui, R., Lobo, J., & Hortal, J. (2008). Assessing the forest genets (Genetta spp.). Biological Conservation, 130 effects of pseudo-absences on predictive distribution (1), 106–117. https://doi.org/10.1016/j.biocon.2005.12.006 model performance. Ecological Modelling, 210(4), 478- Graham, C. H., Ron, S. R., Santos, J. C., Schneider, C. J., & 486.https://doi.org/10.1016/j.ecolmodel.2007.08.010 Moritz, C. (2004). Integrating phylogenetics and environ- Chefaoui, R. M., Hortal, J., & Lobo, J. M. (2005). Potential mental niche models to explore speciation mechanisms in distribution modelling, niche characterization and con- dendrobatid frogs. Evolution, 58(8), 1781–1793. https:// servation status assessment using GIS tools: A case study doi.org/10.1111/j.0014-3820.2004.tb00461.x of Iberian Copris species. Biological Conservation, 122(2), Gritti, E. S., Gaucherel, C., Crespo-Perez, M. V., & Chuine, I. 327–338. https://doi.org/10.1016/j.biocon.2004.08.005 (2013). How can model comparison help improving spe- Dar, A. R., Dar, G. H., & Reshi, Z. (2006). Recovery and cies distribution models? PloS One, PLoS One, 8, e68823. restoration of some critically endangered endemic https://doi.org/10.1371/journal.pone.0068823 angiosperms of the Kashmir Himalaya. Journal of Guisan, A., & Zimmermann, N. E. (2000). Predictive habitat Biological Sciences, 6(6), 985–991. https://doi.org/10. distribution models in ecology. Ecological Modelling, 135 3923/jbs.2006.985.991 (2–3), 147–186. https://doi.org/10.1016/S0304-3800(00) Dar, A. R., Dar, G. H., & Reshi, Z. (2008). Narrow endemic 00354-9 angiosperms of the Kashmir Himalaya: Threat assess- Guo, Q., Kelly, M., & Graham, C. (2005). Support vector ment and conservation. In M. Z. Chisti & A. Fayaz machines for predicting distribution of Sudden Oak (Eds.), Science for better tomorrow (pp. 31–39). Death in California. Ecological Modeling, 128(1), 75–90. Universal Printers. https://doi.org/10.1016/j.ecolmodel.2004.07.012 Dar, G. H., Bhagat, R. C., & Khan, M. A. (2002). Biodiversity Hernandez, P. A., Grahamk, C. H., Master, L. L., & of the Kashmir Himalaya. Valley Book House. Albert, D. L. (2006). The effect of sample size and species Dar, G. H., Khuroo, A. A., & Nasreen, A. (2012). Endemism characteristics on performance of different species distri- in the angiosperm flora of Kashmir Valley, India: bution modeling methods. Ecography, 29(5), 773–785. Stocktaking. In S. K. Mukherjee & G. G. Maiti (Eds.), https://doi.org/10.1111/j.0906-7590.2006.04700.x 250 N. SALAM ET AL. Hijmans, R. J., Cameron, S. E., Parra, J. L., Jones, P. G., & Ecological Modeling, 157(2–3), 209–225. https://doi.org/10. Jarvis, A. (2005). Very high resolution interpolated climate 1016/S0304-3800(02)00197-7 surfaces for global land areas. International Journal of Moraitis, M. L., Valavanis, V. D., & Karakassis, I. (2019). Climatology, 25(15), 1965–1978. https://doi.org/10.1002/ Modelling the effects of climate change on the distribution joc.1276 of benthic indicator species in the Eastern Mediterranean Hijmans, R. J., Garrett, K. A., Huama´n, Z., Zhang, D. P., Sea. Science of the Total Environment, 667, 16–24. https:// Schreuder, M., & Bonierbale, M. (2000). Assessing the doi.org/10.1016/j.scitotenv.2019.02.338 geographic representativeness of genebank collections: Myers, N., Mittermeier, R. A., Mittermeier, C. G., The case of Bolivian wild potatoes. Conservation DaFonseca, G. A. B., & Kent, J. (2000). Biodiversity hot- Biology, 14(6), 1755–1765. https://doi.org/10.1111/j. spots for conservation priorities. Nature, 403(6772), 1523-1739.2000.98543.x 853–858. https://doi.org/10.1038/35002501 Hoegh-Guldberg, O., Mumby, P., Hooten, A. J., Nazeri, M., Jusoff, K., Bahaman, A., & Madani, N. (2010). Steneck, R. S., Greenfield, P., Gomez, E., Harvell, C., Modeling the potential distribution of wildlife species in Sale, P., Edwards, A., Caldeira, K., Knowlton, N., the Tropics. World Journal of Zoolgy, 5(3), 225–231. Eakin, C. M., Iglesias-Prieto, R., Muthiga, N., Pearson, R. G., Raxworthy, C. J., Nakamura, M., & Bradbury, R., Dubi, A., & Hatziolos, M. (2008). Coral Peterson, A. T. (2007). Predicting species distribution reefs under rapid climate change and ocean from small occurrence records: A test case using cryptic acidification. Science, 318(5857), 1737–1742. https://doi. geckos in Madagascar. Journal of Biogeography, 34(1), org/10.1126/science.1152509 102–117. https://doi.org/10.1111/j.1365-2699.2006.01594.x Huisman, J. M., & Millar, A. J. K. (2013). Australian seaweed Pecl, G. T., Araújo, M. B., Bell, J. D., Blanchard, J., collections: Use and misuse. Phycologia, 52(1), 2–5. Bonebrake, T. C., & Chen, I.-C. (2017). Biodiversity redis- https://doi.org/10.2216/12-089.1 tribution under climate change: Impacts on ecosystems Ishihama, F., Takenaka, A., Yokomizo, H., & Kadoya, T. and human well-being. Science, 355(6332), 1–9. https:// (2019). Evaluation of the ecological niche model doi.org/10.1126/science.aai9214 approach in spatial conservation prioritization. PLoS Pena, J. C. C., Kamino, L. H. Y., Mariano-Neto, M. R. E., & ONE, 14(12), e0226971. https://doi.org/10.1371/journal. Siqueira, M. F. (2014). Assessing the conservation status pone.0226971 of species with limited available data and disjunct Kearney, M., & Porter, W. P. (2009). Mechanistic niche distribution. Biological Conservation, 170, 130–136. modelling: Combining physiological and spatial data to https://doi.org/10.1016/j.biocon.2013.12.015 predict species’ ranges. Ecology Letters, 12(4), 334–350. Peterson, A. T., & Vieglais, D. A. (2001). Predicting species https://doi.org/10.1111/j.1461-0248.2008.01277.x invasions using ecological niche modeling: New Kozak, K., Graham, C., & Wiens, J. J. (2008). Integrating approaches from bioinformatics attack a pressing GIS-based environmental data into evolutionary biology. problem. Bioscience, 51(5), 363–371. https://doi.org/10. Trends in Ecology & Evolution, 23(3), 141–148. https:// 1641/0006-3568 (2001)051[0363:PSIUEN]2.0.CO;2 doi.org/10.1016/j.tree.2008.02.001 Phillips, S. J., Anderson, R. P., & Schapire, R. E. (2006). Lyons, K. G., Brigham, C. A., Traut, B. H., & Maximum entropy modeling of species geographic Schwartz, M. W. (2005). Rare species and ecosystem distributions. Ecological Modelling, 190(3–4), 231–259. functioning. Conservation Biology, 19(4), 1019–1024. https://doi.org/10.1016/j.ecolmodel.2005.03.026 https://doi.org/10.1111/j.1523-1739.2005.00106.x Phillips, S. J., & Dudik, M. (2008). Modelling of species Lyons, K. G., & Schwartz, M. W. (2001). Rare species loss distribution with Maxent: New extensions and alters ecosystem function - invasion resistance. Ecology a comprehensive evaluation. Ecography, 31(2),161 Letters, 4(4), 358–365. https://doi.org/10.1046/j.1461- −175. https://doi.org/10.1111/j.0906-7590.2008.5203.x 0248.2001.00235.x Phillips, S. J., Dudik, M., & Schapire, R. E. (2004). A maximum Manel, S., Dias, J. M., Buckton, S. T., & Ormerod, S. J. entropy approach to species distribution modelling. (1999). Alternative methods for predicting species distri- Appearing in Proceedings of The 21st International bution: An illustration with Himalayan river birds. Conference on Machine Learning. Banff, Canada. Journal of Applied Ecology, 36(5), 734–747. https://doi. Polak, T., & Saltz, D. (2011). Reintroduction as an ecosystem org/10.1046/j.1365-2664.1999.00440.x restoration technique. Conservation Biology: The Journal Margules, C. R., & Austin, M. P. (1994). Biological models of the Society for Conservation Biology, 25(3), 424–425. for monitoring species decline: The construction and use https://doi.org/10.1111/j.1523-1739.2011.01669.x of data bases. Philosophical Transactions of the Royal Rodríguez-Salinas, P., Riosmena-Rodriguez, R., Arango, H., Society B: Biological Sciences, 344(1307), 69–75.https:// Gustavo, M. S., & Raquel. (2010). Restoration experiment doi.org/10.1098/rstb.1994.0053 of Zostera marina L. in a subtropical coastal lagoon. Markham, J. (2014). Rare species occupy uncommon niches. Ecological Engineering, 36(1), 12–18. https://doi.org/10. Scientific Repeports, 4, 63–65. https://doi.org/10.1038/ 1016/j.ecoleng.2009.09.004 srep06012 Rogers, W. A., & Panwar, P. S. (1988). Planning a wildlife McKinney. (1999). High rates of extinction and threat in poorly protected area network in India (Vols. 1-2). Wildlife studied taxa. Conservation Biology, 13(6), 1273–1281. https:// Institute of India. doi.org/10.1046/j.1523-1739.1999.97393.x Rondinini, C., Stuart, S., & Boitani, L. (2005). Habitat suitability Mladenoff, D. J., Sickley, T. A., Haight, R. G., & Wydeven, A. P. models reveal shortfall in conservation planning for African (1995). A regional landscape analysis and prediction of vertebrates. Conservation Biology, 19(5), 1488–1497. https:// favorable grey wolf habitat in the northern Great Lakes doi.org/10.1111/j.1523-1739.2005.00204.x region. Conservation Biology, 9(2), 279–294. https://doi.org/ Root, T. L., Price, J. T., Hall, K. R., Schneider, S. H., 10.1046/j.1523-1739.1995.9020279.x Rosenzweig, C., & Pounds, J. A. (2003). Fingerprints of Moisen, G. G., & Frescino, T. S. (2002). Comparing five global warming on wild animals and plants. Nature, 421 modeling techniques for predicting forest characteristics. (6918), 57–60. https://doi.org/10.1038/nature01333 GEOLOGY, ECOLOGY, AND LANDSCAPES 251 Rushton, S. P., Ormerod, S. J., & Kerby, G. (2004). New Huntley, B., van Jaarsveld, A. S., Midgley, G. F., paradigms for modelling species distributions? Journal of Miles, L., Ortega–Huerta, M. A., Peterson, A. T., Applied Ecology, 41(2), 193–200. https://doi.org/10.1111/ Phillips, O. L., & Williams, S. E. (2004). Extinction risk j.0021-8901.2004.00903.x from climate change. Nature, 427(6970), 145–148. Sanchez-Cordero, V., Cirelli, V., Munguía, M., & Sarkar, S. https://doi.org/10.1038/nature02121 (2005). Place prioritization for biodiversity representation Thorn, J. S., Nijman, V., Smith, D., & Nekaris, K. A. I. using ecological niche modeling. Biodiversity Informatics, (2009). Ecological niche modelling as a technique for 2, 211–223. https://doi.org/10.17161/bi.v2i0.9 assessing threats and setting conservation priorities for Seddon, P. J., Griffiths, C. J., Soorae, P. S., & Asian slow lorises (Primates: Nycticebus). Diversity and Armstrong, D. P. (2014). Reversing defaunation: Distributions, 15(2), 289–298. https://doi.org/10.1111/j. Restoring species in a changing world. Science, 345 1472-4642.2008.00535.x (6195), 406. https://doi.org/10.1126/science.1251818 Thuiller, W. (2003). BIOMOD - optimizing predictions of Silva, T. R., Medeiros, M. B., Noronha, S. E., & Pinto, J. R. R. species distributions and projecting potential future shifts (2017). Species distribution models of rare tree species as under global change. Global Change Biology, 9(10), an evaluation tool for synergistic human impacts in the 1353–1362. https://doi.org/10.1046/j.1365-2486.2003. Amazon rainforest. Brazilian Journal of Botany, 40(4), 00666.x 963–971. https://doi.org/10.1007/s40415-017-0413-0 Thuiller, W., Richardson, D. M., Pyˇsek, P., Midgley, G. F., Siqueira, M., Durigan, G., De Marco Júnior, P., & Hughes, G. O., & Rouget, M. (2005). Niche based model- Peterson, A. (2009). Something from nothing: Using ling as a tool for predicting the risk of alien plant inva- landscape similarity and ecological niche modeling to sions at a global scale. Global Change Biology, 11(12), find rare plant species. Journal for Nature Conservation, 2234–2250. https://doi.org/10.1111/j.1365-2486.2005. 17(1), 25–32. https://doi.org/10.1016/j.jnc.2008.11.001 001018.x Solano, E., & Feria, T. P. (2007). Ecological niche modeling Wang, R., Li, Q., He, S., Liu, Y., Wang, M., & Jiang, G. and geographic distribution of the genus Polianthes (2018). Modeling and mapping the current and future L. (Agavaceae) in Mexico: Using niche modeling to distribution of Pseudomonas syringae pv. actinidiae improve assessments of risk status. Biodiversity and under climate change in China. PloS One, 13(2), Conservation, 16(6), 1885–1900. https://doi.org/10.1007/ e0192153. https://doi.org/10.1371/journal.pone.0192153 s10531-006-9091-0 Watson, J. E. M., Dudley, N., Segan, D. B., & Hockings., M. Stockwell, D., & Peters, D. (1999). The GARP modeling (2014). The performance and potential of protected areas. system: Problems and solutions to automated spatial Nature, 515(7525), 67–73. https://doi.org/10.1038/ prediction. International Journal of Geographical nature13947 Information Science, 13(2), 143–158. https://doi.org/10. Wisz, M., & Guisan, A. (2009). Do pseudo-absence selection 1080/136588199241391 strategies influence Species Distribution Models and their Tali, B., Ganie, A., Nawchoo, I. A., Wani, A., & Reshi, Z. A. predictions? An information-theoretic approach based (2014). Assessment of Threat status of selected endemic on simulated data. BioMedCentral Ecology, 9(1), 8. medicinal plants using IUCN regional guidelines: A case https://doi.org/10.1186/1472-6785-9-8 study from Kashmir Himalaya. Journal for Nature Wisz, M. S., Hijman, R. J., Peterson, A. T., Graham, C. H., Conservation, 23,80–89. http://dx.doi.org/10.1016/j.jnc. Wisz, M. S., Hijman, R. J., Peterson, A. T., & 2014.06.004 Graham, C. H. (2008). Effects of sample size on the Thomas, C. D., Cameron, A., Green, R. E., Bakkenes, M., performance of species distribution models. Diversity Beaumont, L. J., Collingham, Y. C., Erasmus, B. F. N., de and Distributions, 14(5), 763–773. https://doi.org/10. Siqueira, M. F., Grainger, A., Hannah, L., Hughes, L., 1111/j.1472-4642.2008.00482.x
Journal
Geology Ecology and Landscapes – Taylor & Francis
Published: Oct 2, 2022
Keywords: Conservation; Maxent; Lagotis cashmiriana; ground validation; reintroduction