Journal of Threatened
Taxa | www.threatenedtaxa.org | 26 September 2025 | 17(9): 27433–27443
ISSN 0974-7907 (Online) | ISSN 0974-7893 (Print)
https://doi.org/10.11609/jott.9538.17.9.27433-27443
#9538 | Received 06 December 2024 | Final received 06 March 2025 |
Finally accepted 20 August 2025
Niche characterization
and distribution of Sikkim Himalayan
Begonia (Begoniaceae),
India: a niche modeling approach
Aditya Pradhan 1 , Dibyendu Adhikari 2 & Arun Chettri
3
1 Department of Botany, School of
Basic Sciences, SRM University Sikkim, Tadong, Gangtok, Sikkim 737102, India.
2 CSIR-National Botanical Research
Institute, Rana Pratap Marg, Lucknow, Uttar Pradesh 226001, India.
3 Department of Botany, School of
Life Sciences, Sikkim University, 6th Mile, Tadong,
Gangtok, Sikkim 737102, India.
1 apradhan512@gmail.com, 2 dibyenduadhikari@gmail.com,
3 achettri01@cus.ac.in (corresponding author)
Editor: K. Haridasan,
Palakkad, Thrissur, India. Date of publication: 26 September 2025 (online & print)
Citation:
Pradhan, A., D. Adhikari & A. Chettri (2025). Niche
characterization and distribution of Sikkim Himalayan Begonia (Begoniaceae), India: a niche modeling
approach. Journal of Threatened Taxa 17(9): 27433–27443. https://doi.org/10.11609/jott.9538.17.9.27433-27443
Copyright: © Pradhan et al. 2025. Creative Commons Attribution 4.0 International License.
JoTT allows unrestricted use, reproduction, and
distribution of this article in any medium by providing adequate credit to the
author(s) and the source of publication.
Funding: This study was entirely self-funded, with no financial support received from any external agency.
Competing interests: The authors declare no competing interests.
Author details: Dr. Aditya Pradhan is an assistant professor at SRM University Sikkim, with research interests in plant taxonomy and biodiversity conservation. Dr. Dibyendu Adhikari is a principal scientist at CSIR-NBRI, Lucknow, specializing in environmental science and forest ecology. Dr. Arun Chettri is an associate professor and head of the Department of Botany at Sikkim University, with a research focus on molecular taxonomy and ecology.
Author contributions: AP conducted the field survey, collected data, performed the modeling, and drafted the manuscript. DA refined the model and contributed to manuscript revision. AC was responsible for the research design, provided overall supervision, and approved the final version of the manuscript
Acknowledgments: We would like to express our sincere gratitude to the Department of Forest, Wildlife, and Environment, Government of Sikkim for granting the necessary permit for the field research.
Abstract: Understanding species’ ecological
niches and distribution patterns is crucial for biodiversity conservation and
management, particularly in ecologically sensitive regions. We used an
NDVI-based ecological niche modeling (ENM) approach for Begonia species
for this purpose, where we achieved high predictive accuracy (AUC: 0.82–0.97).
Niche breadth analysis revealed a positive correlation (r = 0.747, p = 0.003)
between broader niche breadth and larger predicted distribution areas, aligning
with the notion that better-performing models tend to capture either highly
specialized (narrow-breadth) or ecologically flexible (broad-breadth) niches.
Models for Begonia picta, B. panchtharensis,
B. sikkimensis, and B. xanthina
were classified as fair (0.8 < AUC < 0.9), and exhibited broader niche
breadth, with ranges extending from the western Himalaya to the eastern
Himalaya, encompassing Nepal, Bhutan, and China. In contrast, B. satrapis, B. gemmipara, and B.
nepalensis showed very good model performance
(AUC > 0.95) but had the narrowest niche breadth (0.102–0.195), suggesting
specialized habitat requirements and restricted distributions. Given their
limited ecological flexibility and smaller suitable areas, these species
warrant immediate conservation attention to mitigate extinction risks.
Keywords: Conservation, Darjeeling,
diversity, endemic, ENM, MaxEnt, NDVI, niche overlap,
niche breadth, northeastern India.
INTRODUCTION
For centuries, ecologists and
biologists have been fascinated on why species vary greatly in the extent of
its distribution. Some species have a narrow distribution range, whilst some
closely related species have a broader distribution, ranging from the continental
to the global scale (Willis 1922). It is widely believed that narrowly
distributed species have specialized environmental requirements while the
widely distributed species have broader environmental tolerance. Therefore, a
positive correlation between environmental niche breadth and range size is
widely accepted in macro ecological studies (Gaston 2000; Gaston & Spicer
2001; Slatyer et al. 2013). However, it is difficult
to conclude the above hypothesis because the environmental niche of a species
is usually defined by the set of occurrence records. Hence, a larger number of
presence locality data are likely to have a wider distribution range, unlike
species having a lesser number of occurrence records (Burgman
1989; Gaston & Blackburn 2000; Gregory & Gaston 2000; Gaston &
Spicer 2001). Therefore, the species-rich genus Begonia in Sikkim
Himalaya was chosen as the model plant to answer this question.
Begonia L. is the sixth largest genus of
flowering plants and provides several important ecosystem services. For
instance, they help stabilize soil in humid understory environments, support
local invertebrates, and contribute to microhabitat maintenance in forested
areas. Some Begonias also hold ornamental and economic value, being used in
horticulture for their diverse foliage, and showy flowers. In certain regions,
they have recognized medicinal uses, underscoring their cultural and economic
importance. Consequently, conserving Begonias will not only preserve the
essential ecological interactions but also safeguard potential benefits for
local communities.
Being one of the largest genera
of flowering plants, they provide an excellent opportunity to study the
processes underlying the theory of rapid radiation. A sufficient amount of
occurrence data is required to develop a robust distribution model and to test
the above mentioned theory (Moonlight 2017). However,
the unavailability of geo-referenced occurrence data in herbaria and other
online sources such as GBIF (Global Biodiversity Information Facility) limits
the use of such techniques. At present, there is a growing need to estimate the
species distribution range for theoretical as well as applied reasons, e.g.,
understanding species geography to its conservation. Limited species occurrence
data pose enormous challenge to the researchers. Moreover, quantifying the
environmental factors which contribute the most to the distribution of species
becomes even more complicated and challenging (Guisan
& Thuiller 2005; Colwell & Rangel 2009). The
factors that govern the distribution of species are biotic factors, abiotic
factors (soil and topography), species interaction, competition, predators, and
parasites (Gaston 2003). In practice, the species distribution model is
developed using only the occurrence data, and abiotic variables. Recently
several studies have indicated the importance of biotic interaction in shaping
the spatial distribution of species (Gotelli et al.
2010; Sunday et al. 2011). The factors such as biotic interaction and dispersal
are usually ignored, and their effect considered negligible at broader
geographical scale or spatial scales (Soberón 2007;
Colwell & Rangel 2009; Gotzenberger et al. 2012;
Araújo et al. 2014). Thus, abiotic factors, such as bioclimatic variables,
NDVI, slope, and aspect are often used in predicting, and identifying the
suitable habitat of species (Pradhan et al. 2020). The selection of predictor
variables is fundamental before modelling, yet the choice of input variables is
still debatable (Synes & Osborne 2011). The
ecologically relevant variables are capable of generating robust models and
vice versa. For example, the soil type variables might be good predictor
variables for plants whilst temperature, and forest fragmentation related
variables might be a good choice for animals. The use of NDVI contributes to
the modelling process by providing information about the phenological status,
canopy cover, and the water content variation (Amaral et al. 2007). In addition
to capturing phenological status and canopy cover, NDVI also provides insights
into spatial variation in plant health & productivity, reflecting factors
such as vegetation stress, and soil nutrient availability. Consequently, NDVI
data can be used as a proxy for detecting water deficits, drought stress, or nutrient
limitations, all of which are critical for understanding Begonia
establishment, and persistence. Thus, this study aimed to (1) predict the
suitable habitat of Begonia species in Sikkim Himalaya, and (2) define
the ecological niche of Begonia species and quantify the similarities
between them using ENM techniques. The ENMs constructed were compared to assess
the similarities of the ecological niche of the Begonia species, and to
know if they share the same ecological niche or not.
MATERIALS AND METHODS
Study area
The district of Darjeeling shares
a continuous geological and physiographic landscape with Sikkim, rendering the
two regions inseparable in these respects (Basu
2013). The region lies adjacent to Nepal in the east, China in the north, and Bhutan
in the west, making it a geopolitically, and biogeographically significant
segment of the Eastern Himalaya. The physical features of Darjeeling and Sikkim
are very similar, separated by rivers Teesta, and Rungit
which act as a natural boundary dividing the two geographically consonant
regions (Figure 1). Therefore, the state of Sikkim along with Darjeeling
together constitutes the Sikkim Himalaya. The two regions from herein will be
referred to as Sikkim Himalayas (Rai et al. 2000). The region lies amid the
eastern Himalayan regions, roofed by a snow clad-mountain in the north, and
planes in the south. It is bordered by countries such as Nepal in the east,
China in the north, and Bhutan in the west, and is tectonically one of the most
active areas of the Himalaya.
Collection of occurrence record
The primary
occurrence data or the presence data (i.e., geographic coordinate/Latitude and
Longitude) for 13 species of Begonia (viz., B. satrapis,
B. gemmipara, B. josephii,
B. picta, B. xanthina, B. cathcartii, B. flaviflora, B. megaptera, B. nepalensis, B. palmata, B. sikkimensis, B. panchtharensis, B. roxburghii)
were collected from the hills of Darjeeling and Sikkim Himalaya using Garmin
GPS (Global Positioning System). The occurrence data were collected with an
accuracy of 3–10 m.
The geographic coordinate was
collected in the form of Degree Minute Second (DMS) which was later converted
to decimal degrees (DD) using the formula:
DD = D + M/60 + S/3600
The converted presence data was
later rearranged in Microsoft Excel in the following order, i.e., species name,
longitude, latitude, and then saved in CSV (comma delimited) format,
and was later used for modelling.
Predictor Variables
The model was developed using
normalized difference vegetative index (NDVI) raster data for January to December
obtained from GLCF (Global Land Cover Facility) (University of Maryland, USA).
The NDVI is a numerical indicator that quantifies vegetation by measuring the
difference between near-infrared (which vegetation strongly reflects) and red
light (which vegetation absorbs), and is given by the formula:
NDVI = (NIR-RED) / (NIR+RED)
Where; NIR = near-infrared and
RED = Red light
The 12 NDVI variables were first
subjected to correlated tests (r>0.9) using ENM Tools 1.3 software (Warren
et al. 2010). Thus, out of 12 NDVI variables, 10 were used to model the
distribution of Begonia in Sikkim Himalaya along with altitude (Table
1). Although NDVI data for August and September were initially considered, both
months showed high correlation (r>0.9) with July NDVI, risking over fitting
if included simultaneously. Following best practices to reduce
multicollinearity, we retained July NDVI as representative of the monsoon peak
and excluded August and September. This approach helps ensure model parsimony
and avoids redundant variables.
Ecological Niche Modelling
MaxEnt v.3.3. 3k Software (Phillips
& Dudík 2008) was used to model the distribution
of Begonia species in Sikkim Himalaya. MaxEnt
modelling was used because it has a high accuracy rate and performs better with
small size (Elith et al. 2006). The 10 percentile training presence logistic threshold was used,
with 20 replicates run, and maximum of 5,000 iterations for each species. All
other settings were kept default as it has been calibrated with a wide range of
species (Phillips & Dudík 2008). From 20
replicated runs for each species, the average, and maximum, minimum, median,
and standard deviation was obtained. Each Begonia species were modelled
individually using a set of NDVI variables.
Using Niche Toolbox (http://shiny.conabio.gob.mx:3838/nichetoolb2/)
binary maps were obtained and suitable areas for each species of Begonia
were calculated using 10 percentile training presence logistic threshold cut
off values.
Niche Overlap
ENM Tools software was used to
examine the degree of niche overlap between Begonia species. Schoener’s D and Hellinger’s I metrics were used to
estimate the niche overlap between the species.
Schoener’s D is given by the formula:
,
Where pX,i and pY,I are the normalized
suitability scores for species X and Y in grid cell i,
similarly Hellinger’s I is given by the formula:
The niche similarity measures are
obtained after comparing the predicted suitable habitat calculated for each
grid cell from a model developed through MaxEnt. The
niche overlap values ranges vary from 0–1. The value 0 indicates no overlap and
the value of 1 indicates a complete overlap of niches. If only two ENM outputs
of two species are loaded in ENM Tools, single values of D and I
will be produced and if more than two populations of different species are
loaded pair wise D and I values will be produced in simple
Microsoft excel file (Warren et al. 2010).
Niche Breadth
Niche breadth was also assessed
using the same set of output predicted distribution models for each species
(Phillips et al. 2006; Warren et al. 2010).
Model evaluation and Performance
The model developed for each
species was classified and evaluated based on “area under the curve” or AUC
values. The model was further graded as: poor (AUC < 0.8), fair (0.8 <
AUC < 0.9), good (0.9 < AUC < 0.95), and very good (0.95 < AUC <
1.0) following Thuiller et al. (2005).
RESULTS
Occurrence record
A total of 108 occurrence records
or geographic coordinates (B. gemmipara = 8, B.
josephii = 12, B. satrapis
= 10, B. picta = 12, B. nepalensis = 4, B. palmata
= 14, B. panchtharensis = 5, B. sikkimensis = 8, B. cathcartii
= 7, B. megaptera = 5, B. xanthina = 5, B. flaviflora
= 5, B. roxburghii = 13) were collected
from Sikkim Himalayas. The individual occurrence data were then correlated with
the set of NDVI variables.
Predicted habitat distribution
ENM was computed individually for
each Begonia species. The model developed for 13 species of Begonia
are presented in Figure 2. The 10 percentile training
presence logistic threshold values for each species of Begonia are also
provided in table 4. Using the threshold values of individuals Begonia
species suitable habitat was calculated. Therefore based on NDVI dataset, B.
panchtharensis had the maximum area predicted to
be suitable (~4306.88 km2), followed by B. sikkimensis
(~3,804.62 km2), B. picta (~3,785.4
km2), B. cathcartii (~2,480.01 km2),
B. xanthina (~1,905.8 km2), B. josephii (~1,833.28 km2), B. flaviflora (~1,634.74 km2), B. megaptera (~1,412.58 km2), B. satrapis (~1,274.37 km2), B. gemmipara (~1,131.25 km2), B. palmata (~783.446 km2), B. roxburghii (~766.19 km2), B. nepalensis (~60.36 km2).
Model evaluation and validation
Our model performance showed high
accuracy and demonstrated high predictive ability based on AUC scores. The mean
AUC; ranged from 0.82 in B. panchtharensis and
B. sikkimensis to 0.97 in B. satrapis (Table 4).
Contributing variables and
Environmental constraints for Begonia species
The different NDVI variables used
to model the distribution of Begonia species in Sikkim Himalaya showed a
varying degree of contribution to each species model developed. The NDVI for
July contributed the most in B. cathcartii
(68.7 %) followed by B. sikkimensis (62.6 %), B.
josephii (50.5 %), B. flaviflora
(48.5 %), and B. gemmipara (47.6 %). The NDVI
for November contributed the most in the case of B. picta
(32.3 %), B. palmate (37.3 %), B. megaptera
(56.8 %), and B. roxburghii (29.6 %). The
NDVI for May, January, and March each contributed the most in B. satrapis (63.6 %), B. nepalensis
(24.5 %), and B. xanthina (53.1 %)
respectively to the final predictive model (Figure 3; Table 3). Considering the
permutation importance, the NDVI for July contributed the most in B. gemmipara (71.9 %), B. josephii
(39.5 %), B. catcarthii (60.6 %) and B. flaviflora (67.7 %). The NDVI for November contributed
the most in B. picta (72.7 %), B. palmata (40.3 %), and B. megaptera
(53.3 %), B. xanthina (33.6 %). The NDVI for
January, May, October, and December contributed the most in B. nepalensis (30.3 %), B. satrapis
(72.1 %), B. sikkimensis (43.0 %), and B. panchtharensis (53.3 %), and altitude contributed the
most in B. roxburghii (35.0 %) (Table 3).
Niche overlap
The niche overlap test resulted
in significantly different levels of overlaps in Begonia species. The
Hellinger’s I niche overlap values were highest between B. picta, B. sikkimensis,
and B. megaptera (overlap value = 0.96)
indicating the high level to niche overlap whereas the lowest level of niche
overlap was estimated between B. satrapis, and
B. flaviflora (overlap value = 0.35)
(Table 2).
Similarly, the Schoener’s D niche overlap values were highest between B.
picta, B. sikkimensis,
and B. megaptera (overlap value = 0.81)
indicating the high level to niche overlap whereas a low level of niche overlap
was estimated between B. satrapis, and
B. flaviflora (0.12) (Table 2).
Niche Breadth
The niche breadth analysis
resulted in narrower niches in some Begonia species. Begonia panchtharensis had the highest niche breadth value
(NBV) of 0.642, indicating broader niches compared to other related species of Begonia,
which also presented the broadest distribution of suitable habitat. Similarly,
the niche breadth for B. sikkimensis (NBV =
0.412) and B. picta (NBV = 0.384) were also
high with broader distribution of suitable habitat compared to other species of
Begonia. The lowest niche breadth value was estimated in B. satrapis (NBV = 0.102) indicating a very narrow niche.
Species like B. nepalensis (NBV = 0.110) and B.
palmata (NBV = 0.180) also showed low niche
breadth with a narrow distribution of suitable habitat (Table 4).
Relationship between predicted
suitable habitat and niche breadth
A strong positive correlation (r
= 0.747, p = 0.003) was observed between predicted suitable habitat and
niche breadth indicating that the species with higher predicted area retains
broader niche breadth and vice versa (Figure 4).
DISCUSSION
Niche characterization in Begonia
species
The distribution of Begonia species
is correlated with NDVI based on niche modeling. The importance of NDVI
variables contributing to the final predictive model varied across species. The
model developed for B. gemmipara, B. josephii, B. sikkimensis,
B. cathcartii, and B. flaviflora
showed the highest contribution by NDVI for July. Species like B. picta, B. palmata, B.
megaptera, and B. roxburghii,
usually flowers late after the monsoon, and thereby NDVI for November might
have been the most important predictor variables affecting the distribution of
species. Interestingly NDVI for November contributed the most in predicted the
distribution of B. nepalensis. The month of
November might have contributed the most in B. nepalensis
as the species flowers late during dry season i.e. December–January. Amongst
all the 13 species of Begonia, B. satrapis
is considered Critically Endangered and is endemic to Sikkim and Darjeeling
District of West Bengal (Adhikari et al. 2018). Due to narrow geographic range
having restricted distribution, such taxa are more sensitive to habitat
disturbance leading to extinction (Peterson & Watson 1998). The
distribution of B. satrapis is strictly
affected by NDVI for May, when the species begins to regenerate from the tuber.
Niche overlap and Niche Breadth
It is often assumed that closely
related species are morphologically and physiologically alike, and have similar
environmental requirements, i.e., niche retention (Futuyma
& Mitter 1996; Webb 2000; Violle
et al. 2011). The niche overlap test for
Begonia species resulted in great variability in niche overlap values
between morphologically similar species. The low niche overlaps values between
species of section Diploclinium, viz., B.
satrapis and B. josephii
might have resulted due to competitive interaction leading to niche
partitioning (Hardin 1960). Moreover, the highest niche overlap values between
species of section Diploclinium, viz., B. picta and Platycentrum,
viz., B. sikkimensis and B. megaptera support the ‘limiting similarity hypothesis’
(MacArthur & Levins 1967) which posits that
competitive exclusion among closely related species leads to the frequent
coexistence of more distantly related species within ecological communities.
The results of niche breadth
analysis support the idea that better-performing models are associated with more specialized and narrow niche breadth and vice-versa
(Fuchs et al. 2018). The model developed for Begonia species viz.
B. picta, B. panchtharensis,
B. sikkimensis, and B. xanthina
were considered fair (0.8 < AUC < 0.9), with higher niche breadth
indicating more ecological flexibility compared to other species of Begonia.
These species in addition to having broader niche breadth have larger
distribution areas, ranging from Western Himalaya to entire Eastern Himalaya,
covering countries like Nepal, Bhutan, and China (Rajbhandari
et al. 2010; Rana 2016; Camfield & Hughes 2018; Hughes et al. 2018; Pradhan
et al. 2019). Thus, these species have wider climatic tolerance with larger
variation within and amongst the population and sometimes even recognized at a
variety level (Camfield & Hughes 2018). In addition to having a wider niche
breadth, these species also have a wider predicted distribution area compared
to other species. A case apart in B. xanthina
with broader niche breadth and smaller area (~1905 km2) predicted to
be suitable. However, the model developed for B. satrapis,
B. gemmipara, and B. nepalensis
were considered a very good performing model with the lowest niche breadth
(ranging 0.102–0.195) indicating lesser ecological flexibility. Such species
with smaller niche breadth have lesser tolerance to
climatic variation preferring homogenous environmental conditions (Kassen 2002; Dennis et al. 2011). The study thus displays a
positive correlation between species’ niche breadth and suitable predicted
area, except in the case of B. xanthina the
results were otherwise. Niche breadth of most species was consistent with their
geographic distributions, as narrowly distributed species have smaller niche
breadth and broader distributed species have wider niche breadth (Gaston 1993; Kunin & Gatson 1997). The
narrow distribution range of B. satrapis, B.
gemmipara, and B. nepalensis
might be primarily due to narrow niche breadth. The study is in line with the
study on the Mexican genus of globular cacti and numerous other similar studies
(Zhu et al. 2016; Mosco 2017). Therefore, such rare species with narrow niche
breadth have a higher probability of extinction (Futuyma
& Moreno 1988; McKinney 1997) and thus require immediate conservation
initiatives to conserve the existing extant population.
CONCLUSION
The predictive distribution model
for Begonia species, like B. picta, B.
panchtharensis, B. sikkimensis,
and B. xanthina, showed wider niche breadths,
indicating greater ecological flexibility. In addition to their broader niche
breadths, these species have larger distribution areas that range from the
western to eastern Himalaya. In contrast, the models for B. satrapis, B. gemmipara,
and B. nepalensis demonstrated very strong performance
with narrower niche breadths, indicating less ecological flexibility. As a
result, these species require immediate attention, as their smaller suitable
habitats and narrow niche breadths make them more vulnerable to extinction.
Table 1. Correlation analysis for the 12 NDVI layers to check
multicollinearity using ENM Tools 1.3 (Warren et al. 2010).
|
Predictor variables |
eu1 (Jan) |
eu2 (Feb) |
eu3 (Mar) |
eu4 (Apr) |
eu5 (May) |
eu6 (Jun) |
eu7 (Jul) |
eu8 (Aug) |
eu9 (Sep) |
eu10 (Oct) |
eu11 (Nov) |
eu12 (Dec) |
|
Alt |
-0.56 |
-0.77 |
-0.24 |
-0.53 |
-0.57 |
-0.16 |
-0.76 |
-0.75 |
-0.61 |
-0.10 |
-0.53 |
-0.51 |
|
eu1 (Jan) |
|
0.57 |
0.15 |
0.45 |
0.35 |
0.07 |
0.58 |
0.61 |
0.60 |
0.13 |
0.46 |
0.45 |
|
eu2 (Feb) |
|
|
0.12 |
0.54 |
0.60 |
0.22 |
0.62 |
0.58 |
0.46 |
0.23 |
0.54 |
0.54 |
|
eu3 (Mar) |
|
|
|
0.03 |
-0.01 |
-0.16 |
0.18 |
0.22 |
0.22 |
-0.15 |
0.09 |
0.04 |
|
eu4 (Apr) |
|
|
|
|
0.67 |
0.56 |
0.69 |
0.67 |
0.60 |
0.59 |
0.69 |
0.75 |
|
eu5 (May) |
|
|
|
|
|
0.47 |
0.58 |
0.54 |
0.43 |
0.45 |
0.63 |
0.65 |
|
eu6 (Jun) |
|
|
|
|
|
|
0.39 |
0.35 |
0.29 |
0.73 |
0.51 |
0.59 |
|
eu7 (Jul) |
|
|
|
|
|
|
|
0.96 |
0.88 |
0.37 |
0.75 |
0.71 |
|
eu8 (Aug) |
|
|
|
|
|
|
|
|
0.92 |
0.35 |
0.74 |
0.69 |
|
eu9 (Sep) |
|
|
|
|
|
|
|
|
|
0.33 |
0.71 |
0.63 |
|
eu10 (Oct) |
|
|
|
|
|
|
|
|
|
|
0.56 |
0.72 |
|
eu11 (Nov) |
|
|
|
|
|
|
|
|
|
|
|
0.75 |
Table 2. Summary of niche overlap values
based on NDVI dataset [Schoener’s D (above diagonal)
and Hellinger’s I (below diagonal)].
|
Schoener’s D Hellinger’s I |
B. gemmipara |
B. josephii |
B. satrapis |
B. picta |
B. nepalensis |
B. palmata |
B. panchtharensis |
B. sikkimensis |
B. cathcartii |
B. megaptera |
B. xanthina |
B. flaviflora |
B. roxburghii |
|
B. gemmipara |
|
0.65 |
0.25 |
0.50 |
0.24 |
0.33 |
0.49 |
0.63 |
0.69 |
0.38 |
0.45 |
0.68 |
0.32 |
|
B. josephii |
0.89 |
|
0.22 |
0.55 |
0.27 |
0.46 |
0.49 |
0.68 |
0.76 |
0.46 |
0.43 |
0.68 |
0.37 |
|
B. satrapis |
0.53 |
0.49 |
|
0.43 |
0.55 |
0.38 |
0.32 |
0.37 |
0.20 |
0.43 |
0.34 |
0.12 |
0.28 |
|
B. picta |
0.81 |
0.81 |
0.73 |
|
0.44 |
0.67 |
0.57 |
0.81 |
0.49 |
0.81 |
0.70 |
0.48 |
0.66 |
|
B. nepalensis |
0.50 |
0.52 |
0.81 |
0.73 |
|
0.39 |
0.31 |
0.39 |
0.21 |
0.44 |
0.34 |
0.18 |
0.33 |
|
B. palmata |
0.61 |
0.71 |
0.68 |
0.91 |
0.68 |
|
0.37 |
0.56 |
0.35 |
0.75 |
0.52 |
0.32 |
0.67 |
|
B. panchtharensis |
0.79 |
0.75 |
0.63 |
0.85 |
0.59 |
0.67 |
|
0.64 |
0.49 |
0.48 |
0.65 |
0.53 |
0.40 |
|
B. sikkimensis |
0.89 |
0.89 |
0.68 |
0.96 |
0.68 |
0.82 |
0.88 |
|
0.64 |
0.67 |
0.63 |
0.60 |
0.54 |
|
B. cathcartii |
0.93 |
0.95 |
0.46 |
0.77 |
0.45 |
0.62 |
0.76 |
0.88 |
|
0.39 |
0.40 |
0.75 |
0.31 |
|
B. megaptera |
0.67 |
0.69 |
0.74 |
0.96 |
0.73 |
0.95 |
0.79 |
0.89 |
0.65 |
|
0.65 |
0.37 |
0.70 |
|
B. xanthina |
0.76 |
0.71 |
0.66 |
0.91 |
0.88 |
0.88 |
0.88 |
0.87 |
0.68 |
0.88 |
|
0.43 |
0.55 |
|
B. flaviflora |
0.90 |
0.90 |
0.35 |
0.77 |
0.41 |
0.60 |
0.82 |
0.87 |
0.93 |
0.65 |
0.75 |
|
0.34 |
|
B. roxburghii |
0.60 |
0.62 |
0.57 |
0.90 |
0.61 |
0.91 |
0.67 |
0.81 |
0.57 |
0.92 |
0.80 |
0.63 |
|
Table 3. Average contribution of input NDVI
variables to model output for each species of Begonia distributed in
Sikkim Himalaya.
|
Taxon Variables |
B. gemmipara |
B. josephii |
B. satrapis |
B. picta |
B. nepalensis |
B. palmata |
B. panchtharensis |
B. sikkimensis |
B. cathcartii |
B. megaptera |
B. xanthina |
B. flaviflora |
B. roxburghii |
|
Percentage contribution |
|||||||||||||
|
Eu1 (Jan) |
0 |
0.9 |
12.5 |
0.3 |
24.5 |
0 |
1.3 |
0.8 |
0 |
0.3 |
0.1 |
0 |
0.1 |
|
Eu2 (Feb) |
14.1 |
3.8 |
1.2 |
0 |
0.9 |
30.9 |
6.0 |
0 |
0.3 |
21.2 |
0 |
4.2 |
5.3 |
|
Eu3 (Mar) |
19.2 |
0 |
1.0 |
5.0 |
2.0 |
4.0 |
1.2 |
0 |
0 |
0 |
53.1 |
4.3 |
18.2 |
|
Eu4 (Apr) |
11.3 |
4.7 |
3.2 |
0 |
10.8 |
0.2 |
0.1 |
0.8 |
0.5 |
0.3 |
1.3 |
2.0 |
2.6 |
|
Eu5 (May) |
0 |
4.6 |
63.6 |
0 |
18.2 |
1.8 |
0 |
0 |
0.2 |
0 |
0.5 |
31.6 |
12.3 |
|
Eu6 (Jun) |
0.1 |
0 |
1.1 |
0 |
3.5 |
1.2 |
0.7 |
0 |
0.2 |
1.1 |
4.1 |
0.1 |
0.6 |
|
Eu7 (Jul) |
47.6 |
50.5 |
0 |
12.5 |
0 |
0.6 |
37.2 |
62.6 |
68.7 |
0.4 |
0.1 |
48.5 |
0 |
|
Eu10 (Oct) |
1.6 |
23.2 |
0.7 |
28.2 |
9.6 |
24.1 |
2.4 |
26.9 |
9.3 |
12.1 |
9.6 |
4.4 |
12.0 |
|
Eu11 (Nov) |
2.2 |
0 |
6.2 |
32.3 |
10.2 |
37.3 |
0 |
8.2 |
0 |
56.8 |
22.6 |
1.5 |
29.6 |
|
Eu12 (Dec) |
0 |
0 |
7.0 |
0 |
2.5 |
0 |
45.9 |
0 |
0 |
0.2 |
1.1 |
0 |
0.1 |
|
Altitude |
0 |
0 |
3.5 |
21.5 |
17.9 |
0.1 |
5.1 |
0.6 |
20.8 |
7.5 |
7.5 |
3.3 |
19.2 |
|
Permutation importance |
|||||||||||||
|
Eu1 (Jan) |
0 |
0.9 |
4.5 |
0.2 |
30.3 |
0 |
5.7 |
0 |
0 |
1.3 |
0 |
0.2 |
0 |
|
Eu2 (Feb) |
7.2 |
8.4 |
0.3 |
0.3 |
0.6 |
23.7 |
7.9 |
0 |
0 |
0 |
0 |
2.7 |
0 |
|
Eu3 (Mar) |
6.2 |
0 |
0.1 |
2.2 |
0.3 |
1.5 |
0 |
0 |
0 |
0 |
17.6 |
0.6 |
7.5 |
|
Eu4 (Apr) |
2.4 |
1.8 |
1.6 |
0 |
2.3 |
0 |
0.3 |
4.7 |
0.2 |
0.4 |
2.9 |
5.1 |
5.1 |
|
Eu5 (May) |
0 |
1.4 |
72.1 |
0 |
17.0 |
0.1 |
0 |
0 |
0 |
0 |
2 |
7.3 |
19.5 |
|
Eu6 (Jun) |
0.7 |
0 |
0.2 |
0 |
2.9 |
3.4 |
2.7 |
0.1 |
0.7 |
3.4 |
12.3 |
1 |
1.7 |
|
Eu7 (Jul) |
71.9 |
39.5 |
0 |
6.2 |
0 |
3.1 |
25.8 |
41.8 |
60.6 |
1.9 |
0 |
67.7 |
0 |
|
Eu10 (Oct) |
1.5 |
27.7 |
0.6 |
9.4 |
6.4 |
27.4 |
0.1 |
43.0 |
9.9 |
15.9 |
5.9 |
10.4 |
5.5 |
|
Eu11 (Nov) |
0.8 |
0.1 |
14.8 |
72.7 |
18.5 |
40.3 |
0 |
9.7 |
0 |
67.3 |
33.6 |
0 |
25.6 |
|
Eu12 (Dec) |
0 |
0 |
5.8 |
0 |
0.4 |
0 |
53.3 |
0 |
0 |
2.2 |
11.7 |
0 |
0.1 |
|
Altitude |
9.2 |
20.2 |
0 |
8.9 |
21.3 |
0.6 |
4.0 |
0.6 |
28.5 |
7.5 |
14 |
4.9 |
35.0 |
Table 4. Niche breadth values and predicted suitable area (10 percentile
training presence logistic threshold value).
|
|
Species |
AUC |
Niche breadth |
Threshold value/Area (km2) |
|
1 |
B. gemmipara |
0.93 |
0.195 |
0.304/1131.25 |
|
2 |
B. josephii |
0.94 |
0.225 |
0.384/1833.27 |
|
3 |
B. picta |
0.89 |
0.384 |
0.396/3785.40 |
|
4 |
B. satrapis |
0.97 |
0.102 |
0.133/1274.37 |
|
5 |
B. flaviflora |
0.91 |
0.231 |
0.416/1634.74 |
|
6 |
B. cathcartii |
0.91 |
0.206 |
0.283/2480.00 |
|
7 |
B. megaptera |
0.91 |
0.329 |
0.521/1412.57 |
|
8 |
B. nepalensis |
0.89 |
0.110 |
0.605/60.36 |
|
9 |
B. palmata |
0.95 |
0.180 |
0.469/783.44 |
|
10 |
B. panchtharensis |
0.82 |
0.642 |
0.333/4306.88 |
|
11 |
B. sikkimensis |
0.82 |
0.412 |
0.445/3804.61 |
|
12 |
B. xanthina |
0.84 |
0.565 |
0.449/1905.79 |
|
13 |
B. roxburghii |
0.94 |
0.191 |
0.448/766.19 |
Note: The highest and the lowest niche
breadth are highlighted in bold. Values range from 0–1: 0 is equal to one grid
cell being suitable (specialized niche); whereas 1 is where all grid cells are
suitable (broad niche).
For figures - - click here for
full PDF
REFERENCES
Adhikari, D.,
Z. Reshi, B.K. Datta, S.S. Samant, A. Chettri, K. Upadhaya,
M.A. Shah, P.P. Singh, T. Tiwary, K. Majumdar, A.
Pradhan, M.L. Thakur, N. Salam, Z. Zahoor, S.H. Mir, Z.A. Kaloo
& S.K. Barik (2018).
Inventory and characterization of new populations through ecological niche
modelling improve threat assessment. Current Science 114(3): 0468–0595. https://doi.org/10.185
20/c s%2Fv114%2Fi03%2F519-531
Amaral, S.,
C.B. Costa & C.D. Renno (2007). Normalized
Difference Vegetation Index (NDVI) improving species distribution models: an
example with the Neotropical genus Coccocypselum
(Rubiaceae). Anais XIII Simposio
Brasileiro de Sensoriamento
Remote, Florianopolis, Brazil, 21–26 April, INPE, pp 2275–2282.
Araújo, C.,
L. Marcondes-Machado & G. Costa (2014). The importance of
biotic interactions in species distribution models: A test of the Eltonian noise hypothesis using parrots. Journal of
Biogeography 41(3): 513–523. https://doi.org/ 10.1111/j bi.12234
Basu, S.K. (2013). Geology of Sikkim State and
Darjeeling District of West Bengal. Geological Society of India, Bangalore,
ix + 255 pp.
Burgman, M.A. (1989). The habitat
volumes of scarce and ubiquitous plants: A test of the model of environmental
control. The American Naturalist 133: 228–239. https://doi.org/10.1086/284912
Camfield, R.
& M. Hughes (2018).
A revision and one new species of Begonia L. (Begoniaceae,
Cucurbitales) in Northeast India. European Journal
of Taxonomy 396: 1–116. https://doi.org/10.5852/ejt.2018.396
Colwell, R.K.
& T.F. Rangel (2009). Hutchinson’s duality: The once and future niche. Proceedings
of the National Academy of Sciences 106: 19651–19658. https://doi.org/10.1073/pnas.0901650106
Dennis, R.L.,
L. Dapporto, S. Fattorini
& L.M. Cook (2011). The
generalism–specialism debate: the role of generalists
in the life and death of species. Biological Journal of the Linnean Society 104: 725–737. https://doi.org/10.111
1/j.1095-8312.2011.01789.x
Elith, J., C. Graham & the NCEAS
species distribution modelling group (2006). Novel methods
improve prediction of species’ distributions from occurrence data. Ecography 29: 129–151.
Fuchs, A., C.
Gilbert & J. Kamilar (2018). Ecological niche
modeling of the genus Papio. American Journal of
Biological Anthropology 166(4): 812–823.
Futuyma, D. & G. Moreno (1988). The evolution of
ecological specialization. Annual Review of Ecology Evolution and Systematics 19:
201–233.
Futuyma, D.J. & C. Mitter (1996). Insect-plant interactions: the evolution of
component communities. Philosophical Transactions of the Royal Society of
London. Series B: Biological Sciences 351: 1361–1366. https://doi.org/10.1098/rs
tb.1996.0119
Gaston, K.J. (1993). Rarity.
Chapman and Hall, London. 205 pp. https://doi.org/10.1007/978-94-011-0701-3
Gaston, K.J.
(2000). Global
patterns in biodiversity. Nature 405: 220–227. https://doi.org/10.1038/35012228
Gaston, K.J. (2003). The Structure
and Dynamics of Geographic Ranges. Oxford University Press, Oxford, UK, 278
pp.
Gaston, K.J.
& J.I. Spicer (2001). The relationship between range size and niche breadth: A test using
five species of Gammarus (Amphipoda).
Global Ecology and Biogeography 10: 179–188. https://doi.org/10.1046/j.1466-822x.200
1.00225.x
Gaston, K.J.
& T.M. Blackburn (2000). Pattern and Process in Macroecology. Wiley, U.S.,
377 pp.
Gotelli, N.J., G.R. Graves & C. Rahbek (2010). Macroecological signals of species interactions in the
Danish avifauna. Proceedings of the National Academy of Sciences of the
United States of America 107: 5030–5035. https://doi.org/10.1073/pnas.0914089107
Gotzenberger, L., F. de Bello, K.A. Brathen, J. Davison, A. Dubuis,
A. Guisan, J. Leps, R.
Lindborg, M. Moora, M. Partel,
L. Pellissier, J. Pottier,
P. Vittoz, K. Zobel & M. Zobel (2012). Ecological assembly rules in
plant communities approaches, patterns and prospects. Biological
Reviews 87: 111–127. https://doi.org/10.11 11/j.1469-185X.2011.00187.x
Gregory, R.D.
& K.J. Gaston (2000). Explanations of commonness and rarity in British
breeding birds: Separating resource use and resource availability. Oikos
88: 515–526. https://doi.org/10.1034/j.1600-0706.2000.880307.x
Guisan, A. & W. Thuiller (2005). Predicting
species distribution: offering more than simple habitat models. Ecology
Letters 8: 993–1009. https://doi.org/1 0.1111 /j.1461-0248.2005.00792.x
Hardin, G. (1960). The competitive
exclusion principle. Science 131: 1292–1297. https://doi.org/10.1126/science.131.3409.1292
Hughes, M.,
C.I. Peng, C.W. Lin, R.R. Rubite, P. Blanc & K.F.
Chung (2018). Chloroplast
and nuclear DNA exchanges among Begonia sect. Baryandra species (Begoniaceae)
from Palawan Island, Philippines, and descriptions of five new species. PLoS ONE 13(5) :
e0194877. https://doi.org/10.1371/journal.pone.0194877
Kassen, R. (2002). The experimental
evolution of specialists, generalists, and the maintenance of diversity. Journal
of Evolutionary Biology 15(2): 173–190. https://doi.org/10.1046/j.1420-9101.2002.00377.x
Kunin, W.E. & K.J. Gaston (1997). The Biology of
Rarity: Causes and Consequences of Rare Common Differences. Chapman and
Hall, London.
MacArthur, R.
& R. Levins (1967). The limiting
similarity, convergence, and divergence of coexisting species. The American
Naturalist 101(921): 377–385. https://doi.org/10.1086/282505
McKinney, M.L (1997). Extinction
vulnerability and selectivity: Combining ecological and paleontological views. Annual
Review of Ecology and Systematics 28: 495–516. http://doi.org/10.1146/annurev.ecolsys.28.1.495
Moonlight,
P.W. (2017). Linking
speciation and the niche: taxonomy, phylogeny and niche evolution in
neotropical Begonia. PhD thesis. University of Glasgow. http://theses.gla.ac.uk/id/eprint/8397
Mosco, A.
(2017). Niche
characteristics and potential distribution of Thelocactus
species, a Mexican genus of globular cacti. bioRxiv
124511. https://doi.org/10.1101/124511
Peterson,
A.T. & D.M. Watson (1998). Problems with areal definitions of endemism: the effects
of spatial scaling. Diversity and Distribution 4(4): 189–194. https://doi.org/10.1046/j.1472-4642.1998.00021.x
Phillips,
S.J., R.P. Anderson & R.E. Schapire (2006). Maximum entropy
modeling of species geographic distributions. Ecological Modelling 190
(3–4): 231–259. https://doi.org/10.1016/j.ecolmodel.2005.03.026
Phillips,
S.J. & M. Dudík (2008). Modeling of
species distributions with Maxent: new extensions and a comprehensive
evaluation. Ecography 31(2): 161–175. https://doi.org/101111/j.0906-7590.2008.5203.x
Pradhan, A.,
D. Rai, S.K. Barik & A. Chettri (2019). Begonia panchtharensis (Begoniaceae),
a new record to India from Sikkim, Eastern Himalaya. Journal of Japanese
Botany 94: 56–57.
Rai, L.K., P.
Prasad & E. Sharma (2000). Conservation threats to some important medicinal plants
of the Sikkim Himalaya. Biological Conservation 93: 27–33. https://doi.org/10.1016/S0006-3207(99)00116-0
Rajbhandary, S., M. Hughes & K. Shrestha
(2010). Three new
species of Begonia sect. platycentrum
from Nepal. Gardens’ Bulletin Singapore 62: 151–162.
Roy, P.S., P.
Meiyappan, P.K. Joshi, M.P. Kale, V.K. Srivastav, S.K. Srivasatava,
M.D. Behera, A. Roy, Y. Sharma, R.M. Ramachandran, P. Bhavani, A.K. Jain &
Y.V.N. Krishnamurthy (2016). Decadal Land Use and Land Cover Classifications across
India, 1985, 1995, 2005. ORNL DAAC, Oak Ridge, Tennessee, USA.
Slatyer, R.A., M. Hirst & J.P.
Sexton (2013).
Niche breadth predicts geographical range size: A general ecological pattern. Ecology
Letters 16: 1104–1114. https://doi.org/10.1111/ele.12140
Soberón, J. (2007). Grinnellian and Eltonian niches
and geographic distributions of species. Ecology Letters 10: 1115–1123. https://doi.org/10.1111/j.1461-0248.
2007.01107.x
Sunday, J.M.,
A.E. Bates & N.K. Dulvy (2011). Global analysis
of thermal tolerance and latitude in ectotherms. Proceeding of the Royal
Society B 278: 1823–1830. https://doi.org/10.1098/rspb.2010.1295
Synes, N.W. & P.E. Osborne (2011). Choice of
predictor variables as a source of uncertainty in continental-scale species
distribution modelling under climate change. Global Ecology and Biogeography
20(6): 904–914. https://doi.org/10.1 111/j.1466-8238.2010.00635.x
Thuiller, W., D.M. Richardson, P. Pyˇsek, G.F. Midgley, G.O. Hughes & M. Rouget (2005). Niche-based modelling as a tool for predicting the risk
of alien plant invasions at a global scale. Global Change Biology 11:
2234–2250. https://doi.org/10.1111/j.1365-2486.2005.001018.x
Violle, C., D.R. Nemergut,
Z. Pu & L. Jiang (2011). Phylogenetic limiting similarity and competitive
exclusion. Ecology Letters 14: 782–787.
Warren, D.,
G. Richard & M. Turelli (2010). ENM Tools: A
toolbox for comparative studies of environmental niche models. Ecography 33: 607–611.
Webb, C.O. (2000). Exploring
the phylogenetic structure of ecological communities: an example for rain
forest trees. The American Naturalist 156: 145–155. https://doi.org/10.1086/303378
Willis, J.C. (1922). Age and Area;
A Study in Geographical Distribution and Origin of Species. The University
Press, Cambridge, UK. https://doi.org/10.5962/bh l.title.30741
Zhu, G.P., H.Q. Li, L. Zhao, L.
Man & Q. Liu (2016).
Mapping the ecological dimensions and potential distributions of endangered
relic shrubs in Western Ordos Biodiversity Centre. Scientific Report 6: 26268.
https://doi.org/10.1038/s rep26268