Acta Ecologica Sinica xxx (xxxx) xxx

Contents lists available at ScienceDirect

Acta Ecologica Sinica

journal homepage: www.elsevier.com/locate/chnaes

Distribution modelling of Tor putitora (Hamilton, 1822), an endangered

cyprinid in the Himalayan river system using MaxEnt
Ranjit Mahato a, Santoshkumar Abujam b, Dhoni Bushi a, Oyi Dai Nimasow c, Gibji Nimasow a, *,
Debangshu Narayan Das b
a
Department of Geography, Rajiv Gandhi University, Arunachal Pradesh, India
b
Department of Zoology, Rajiv Gandhi University, Arunachal Pradesh, India
c
Department of Botany, Rajiv Gandhi University, Arunachal Pradesh, India

A R T I C L E I N F O A B S T R A C T

Keywords: Freshwater fishes are threatened by habitat modification, overfishing, pollution, invasive species, and climate
Freshwater fish change. Hence, habitat assessment using geospatial technology is essential in predicting the distribution of
Environmental variables threatened candidates. Tor putitora, an endangered game fish is declining due to indiscriminate fishing and
Habitat predilection
habitat destruction. Meager attempts have been made so far to understand habitat predilection of the species. In
Conservation
this study, we used Maximum Entropy (MaxEnt) model with occurrence records and environmental variables.
Eco-restoration
The occurrences and environmental variables were collected from survey and secondary sources and used after
removing biases, multicollinearity test and contribution analysis. The model performed well with area under
curve (AUC) 0.867, partial AUC 0.895 and true skill statistic (TSS) 0.756. The main contributing variables were
upstream elevation (35.3%) and flow length (21.4%). The model predicted 10% suitable, 33% slight to moderate
suitable and 57% unsuitable. So, modelling is useful in identifying river networks for conservation and eco-
restoration of freshwater fishes.

1. Introduction fins show tint of reddish golden colour as well as at maturity the body
above its lateral line shows usually golden colour, which might be absent
Tor putitora [1] is known popularly as Golden mahseer or Himalayan in the juvenile stage [8]. According to IUCN, [9], T. putitora is enlisted as
mahseer and it belongs to the family Cyprinidae under order Cypri­ an endangered (EN) species.
niformes. It is one of the largest freshwater fish of the Indian subconti­ Across the global range, a total of 17 species of Tor are considered
nent inhabiting mainly the rivers of the Himalayan foothill region [2]. valid [10]. Out of which, eight (8) species have been reported from India
The species has a restricted area of occupancy in general within India, namely Tor barakae [11], Tor khudree [12], Tor kulkarnii [13], Tor
Afghanistan, Pakistan, Nepal, Bhutan, Myanmar, and Bangladesh [3–5]. malabaricus [14], Tor mosal [1], T. putitora [1], Tor remadeviae [15] and
Behaviourally, being potamodromous (migrating within freshwaters) Tor tor [1]. In the North-eastern states of India, only 3 species of genus
one, the species mostly uses smaller hill streams as their spawning Tor namely Tor barakae, T. putitora and T. tor have been reported to date.
grounds (generally rapid streams, riverine pools, and lakes character­ Interestingly, T. putitora is the state fish in many states of India viz.,
ized by large boulders, pebbles and gravel) with water temperature Arunachal Pradesh, Himachal Pradesh, Uttarakhand and the union ter­
range of 11 to 30.5 ◦ C, alkaline pH and dissolved oxygen concentration ritory of Jammu and Kashmir. The Indus, Ganges and Brahmaputra river
in the range of 6.4–11 mg/l [6,7]. Compared to the other congeners, basins are the native range of T. putitora [16]. Among the Indian species
T. putitora is elongated but laterally compressed having longer head and of the genus, T. putitora is the important food and a popular game fish
head length is significantly greater than body depth at dorsal fin origin. after T. tor. It constitutes an amazing fishery in different foothill water
Further, the head is slightly blunt with a large mouth and prominent bodies of the country and nowadays, many reservoirs are being stocked
lateral line stripes whereas the body is dim greenish dorsally, yellowish with the species both for fishery and ecotourism development. However,
and silvery white ventrally. The distal margin of pelvic, anal and caudal the creation of dams across certain rivers has alarmingly degraded

* Corresponding author at: Department of Geography, Rajiv Gandhi University, Rono Hills, Doimukh, 791112, Arunachal Pradesh, India.
E-mail address: [email protected] (G. Nimasow).

https://doi.org/10.1016/j.chnaes.2022.01.004
Received 9 January 2022; Received in revised form 23 January 2022; Accepted 24 January 2022
1872-2032/© 2022 Ecological Society of China. Published by Elsevier B.V. All rights reserved.

Please cite this article as: Ranjit Mahato, Acta Ecologica Sinica, https://doi.org/10.1016/j.chnaes.2022.01.004
R. Mahato et al. Acta Ecologica Sinica xxx (xxxx) xxx

habitat and shrink their natural breeding grounds. Furthermore, high records and environmental factors very precisely and successfully. It is
mortality of brood and juvenile fishes often have been witnessed due to expected that such outcome would be able to provide important infor­
the act of indiscriminate overfishing in the rivers and the species has mation for rational aquatic resource management and ex-situ or in-situ
been estimated to decline by more than 50% already [16]. T. putitora is conservation of the species like golden mahseer in its natural habitat.
also popular and important high-value indigenous fish species in Nepal
[17]. The species is also reportedly declining from its natural habitat in 2. Materials and methods
Nepal [18]. However, in recent years, the success in artificial breeding at
some research stations has provided enthusiasm on developing it for 2.1. Species occurrence data
commercial cultivation in natural waters [19]. In Bhutan, it is
commercially important game fish, and has a very high table value. It is As per the available literature, the targeted species is mainly
locally called as Serngya (Yellowfish), and represents the symbol of good distributed in the rivers of the Himalayan region. Hence, the study area
luck in the Bhutanese belief system [20]. Unfortunately, the natural encompasses the union territory of Jammu and Kashmir in the west to
stock in the country is also declining due to deterioration of the natural the North-East Indian states in the east including the countries of Nepal
habitat to an irreparable extent. and Bhutan (Fig. 1). The occurrences of T. putitora in the study area were
Species Distribution Modelling (SDM) predicts the distribution of a collected from both primary and secondary sources. The primary
species across geographic space and time using environmental data and occurrence data were recorded during the field survey using handheld
occurrence records. Hence, the significance of ecological factors like Garmin GPSMAP 64csx. The dismo package of R (Version 4.0.3) has
topographic, climatic, land cover, physico-chemical properties of soil been used to access the secondary occurrence data from the Global
and water, etc. cannot be ignored in determining the distribution pattern Biodiversity Information Facility [27] following Hijmans et al. [28]. The
of freshwater fishes [21]. Further, the alteration in thermal regimes, search string ‘T. putitora’ in GBIF returned 138 occurrence data and
land use and hydrology has been reported to cause fish isolation and another 42 species occurrences were consulted from different published
fragmentation within the terrestrial landscapes leading to changes in the journal articles. Thus, a total of 180 occurrence records were collected.
distributional patterns [22–25]. Therefore, looking at the declining Out of this, the species records lacking geographic coordinates and
population of natural stock, estimating the potential distribution of this falling outside the study area were removed. Further, a snapping toler­
sought after freshwater fish through ecological modelling is highly ance of 1 km was used to move the occurrence points to the closest
indispensable for policy making and planning particularly towards freshwater pixel and the points falling outside these pixels were
habitat restoration and implementing conservation strategies. So far, removed. The new geographical coordinates (104 Nos.) were used as
hardly any attempts have been made to model the distribution of presence records for testing the geographical biases (Supplementary - 1).
freshwater fishes, particularly T. putitora in the river systems of the The occurrence records are often reported to be biased due to easy
Himalayas. This research article hitherto revealed that Maximum En­ accessibility of an area or near road or areas of high population density
tropy (MaxEnt) approach [26] can predict the distribution range of an that can distort the species’ niche resulting into error in estimates of the
endangered Cyprinid fish (T. putitora) on the backdrop of the occurrence model performances [29,30]. Hence, the biases in the occurrence

Fig. 1. Map of the study area. Generated by the authors using ArcGIS 10.3, http://esri.com.

2
R. Mahato et al. Acta Ecologica Sinica xxx (xxxx) xxx

records were removed using the spThin package in R, which uses an Table 1
algorithm to randomize the removal of occurrence locations within a Post collinearity variables assessed for inclusion in the model.
specified distance threshold [31]. The resulting dataset retained the Variable code Variable name Unit Source
records of species present within 10 km spatial grid and returned a
elv_01 Upstream elevation (min) Meter HydroSHEDS
dataset of 58 occurrence points for the model (Supplementary - 2). [m]
slope_01 Upstream slope (min) Degree * HydroSHEDS
2.2. Environmental variables 100
flow_acc_01 Flow length Count HydroSHEDS
hydroclim_03 Isothermality (BIO2/BIO7) * 100 WorldClim
For generating the distribution modelling of Tor putitora, a set of (×100)
near-global environmental variables with 1 km resolution were down­ hydroclim_15 Precipitation Seasonality * 100 WorldClim
loaded from the website of EarthEnv project (www.earthenv.org) using (Coefficient of Variation)
R. EarthEnv is a collaborative project of biodiversity scientists and lc_avg_01 Evergreen/Deciduous [%] Consensus land cover
Needle leaf Trees (CLC)
remote sensing experts for monitoring and modelling biodiversity,
lc_avg_03 Deciduous Broadleaf Trees [%] Consensus land cover
ecosystems, and climate. The sources of these variables are WorldClim, (CLC)
Consensus Land Cover, HydroSHEDS and World soil information (ISRIC) lc_avg_04 Mixed/Other Trees [%] Consensus land cover
as explained in detail by Domisch et al. [32]. The environmental vari­ (CLC)
ables include 19 layers of hydroclimatic variables, 12 layers of land lc_avg_05 Shrubs [%] Consensus land cover
(CLC)
cover variables, 10 layers of soil variables, 4 layers each of elevation and lc_avg_07 Cultivated and Managed [%] Consensus land cover
slope, and 1 layer each of stream length and flow accumulation. Addi­ Vegetation (CLC)
tionally, 4 variables namely dissolved oxygen, total alkalinity, town and lc_avg_09 Urban/Built-up [%] Consensus land cover
road proximity was generated through the Inverse Distance Weighting (CLC)
lc_avg_10 Snow/Ice [%] Consensus Land
(IDW) interpolation method from the water samples (field survey) and
Cover
published literatures using ArcMap 10.3 [33]. The coordinate points lc_avg_11 Barren [%] Consensus land cover
were snapped to the closest freshwater pixel with a snapping tolerance (CLC)
of 1 km before the interpolation. All the predictor variables were clipped lc_avg_12 Open Water [%] Consensus land cover
to the area of interest (AOI) for further processing. (CLC)
soil_avg_04 Silt content mass fraction [%] World soil
information (ISRIC)
2.3. Selection of best environmental variables soil_avg_07 Cation exchange capacity [cmol/ World soil
kg] information (ISRIC)
To achieve parsimonious model, all the predictors with insignificant soil_avg_08 Bulk density of the fine [kg / m3] World soil
earth fraction information (ISRIC)
contribution to the model were removed through multicollinearity test
soil_avg_09 Depth to bedrock (R [cm] World soil
and analysis of percent contributions. Most of the statistical methods horizon) up to max. 240 cm information (ISRIC)
face the problem of multicollinearity due to high correlation among the do Dissolved oxygen [mg/l] IDW interpolation in
predictor variables [34] that can lead to model uncertainties [35–37]. ArcGIS 10.3
Hence, Merow et al. [38] recommended minimizing the highly corre­
lated environmental variables because the complex features created by
MaxEnt are often already highly correlated. In this study, we used Table 2
Variance Inflation Factor (VIF) correlation approach to reduce multi­ Variables retained for the final model (shown in bold) with percent contribution
collinearity problems using usdm package in R [39]. In general, a cor­ and permutation importance (averaged over 5 replicate runs).
relation coefficient threshold of >0.7 is considered to be an appropriate Variable code Variable name Contribution Permutation
indicator of collinearity [40–44] and a VIF greater than 10 is a signal (%) importance (%)
that the model has a collinearity problem [45]. Accordingly, in this elv_01 Upstream elevation 35.3 41.7
study, out of 55 variables, 19 predictor variables with correlation co­ (min)
efficient threshold values of <0.7 [46,47] and VIF < 7 were retained for flow_acc_01 Flow length (m) 21.4 4.4
additional analysis (Table 1). lc_avg_03 Deciduous broadleaf 7.8 2.1
trees
Then, initially we run MaxEnt model with all 19 predictor variables lc_avg_01 Evergreen/Deciduous 5.4 8.7
to examine the variable contributions. Based on the results, 12 variables needle leaf trees
(elv_01, hydroclim_15, flow_acc_01, lc_avg_01, lc_avg_03, lc_avg_05, lc_avg_05 Shrubs 3.9 0.6
lc_avg_07, lc_avg_09, lc_avg_11, soil_avg_09, slope_01, and do) with more hydroclim_15 Precipitation 3.5 9.1
Seasonality (coefficient
than 1% percent contribution were retained. Out of which, 6 best
of variation)
environmental variables (elv_01, hydroclim_15, flow_acc_01, lc_avg_03, lc_avg_07 Cultivated and managed 3.5 6.6
soil_avg_09, and do) were selected after consulting available literature vegetation
on the habitat information of the species and removing the persistent do Dissolved oxygen 2.3 2.2
variables of land cover. Finally, we performed 5 replicate runs with 58 Slope_01 Upstream slope (min) 2.3 0.7
lc_avg_11 Barren 2.2 5.9
presence records and the 6 environmental variables to predict the dis­ lc_avg_09 Urban/Built-up 2.2 0.4
tribution of T. putitora in the Himalayan river system (Table 2). soil_avg_09 Depth to bedrock (R 1.4 1.5
horizon) up to max.
2.4. Model setup and evaluation 240 cm

MaxEnt (version 3.4.1) was used to model the potential distribution [38]. The specific settings used in the MaxEnt model were logistic output
of T. putitora. Out of a range of traditional and machine-learning format, cross-validate replicate run type, response curves, jackknife
modelling approaches, MaxEnt is reported to perform well [26,48–51] measures of variable importance, 10 percentile training presence, and
with high predictive accuracy [52,53]. MaxEnt is a machine-learning rest kept as default settings. The MaxEnt output in .ASC format was
algorithm that builds relationships between presence-only records and imported to ArcGIS 10.3 and the relative habitat suitability [38] ranging
a set of environmental variables to estimate a target probability distri­ from 0 (unsuitable) to 1 (highly suitable) was reclassified into five
bution of maximum entropy based on the most uniform distribution

3
R. Mahato et al. Acta Ecologica Sinica xxx (xxxx) xxx

classes viz. < 0.1 = unsuitable, 0.1 to 0.3 = slightly suitable, 0.3 to Table 3
0.5 = moderately suitable, 0.5 to 0.7 = suitable, and > 0.7 = highly Results of model evaluation using full and partial ROC-AUC metrics.
suitable. The flowchart of the methodology followed in this study is Models Full AUC metric Partial AUC metric
given in Fig. 2. (mean ± SD)
AUC at 0.5 AUC at 0.05 level AUC
To test the model predictive accuracy, we used 5-fold cross- level (mean ± SD) ratio
validation, where the occurrence data was randomly split into several (mean) (mean)
equal-size groups called ‘folds’, and the models are created leaving out Replicate_1 0.871 ± 0.030 0.5 0.890 ± 0.040 1.780
each fold in turn. The overall performance of the model was evaluated Replicate_2 0.818 ± 0.046 0.5 0.892 ± 0.040 1.784
using the threshold independent, area under curve (AUC) of the receiver Replicate_3 0.891 ± 0.039 0.5 0.894 ± 0.050 1.789
operating characteristic (ROC) curve and True Skill Statistic (TSS). The Replicate_4 0.879 ± 0.037 0.5 0.892 ± 0.040 1.785
Replicate_5 0.875 ± 0.026 0.5 0.891 ± 0.040 1.782
AUC value ranges from 0.5 to 1.0 where a value of 1.0 indicates perfect
Averaged 0.867 ± 0.036 0.5 0.895 ± 0.040 1.790
fit model while a value near 0.5 indicates a fit not better than a random
[54,55]. We also used the partial AUC metric to evaluate the perfor­
mance of models [56–58] using Niche Toolbox; available online at upstream elevation (elv_01) showed the highest relative importance of
http://shiny.conabio.gob.mx:3838/nichetoolb2/. The toolbox calcu­ 35.3%, followed by flow length (flow_acc_01) with 21.4%. The impor­
lates the ratio of random AUC (at 0.5 levels) and actual AUC (at a tance of rest of the variables ranged between 1.4 and 7.8% only, indi­
defined level of omission, e.g. 0.05) using the occurrence data and model cating intermediate to lower contribution in predicting the suitable
prediction raster map. Further, we calculated the True Skill Statistic habitat of T. putitora. The logistic output (response curves) shows a
(TSS) that takes into account both omission and commission errors, and negative relationship of T. putitora with high upstream elevation (peak
success as a result of random guessing. The TSS value ranges from − 1 to towards lower values 0 to 1000 m) and deciduous broadleaf trees
+1, where +1 indicates perfect agreement and values of zero or less (peaked towards lower values of 5 to 10%). On the other hand, there was
indicate a performance no better than random [59]. a positive association with flow length (peak towards higher values of
8000 to 30,000 m), precipitation seasonality (peak towards higher
3. Results values), dissolved oxygen (peak towards higher values of 5 to 12 mg/l),
and depth to bedrock (peaked towards higher values of 220 to 235 cm).
3.1. Model evaluation The response curves of the environmental variables are shown in Fig. 4.

The final model used six best environmental variables (minimum

3.2. Distribution modelling of Tor putitora
upstream elevation, precipitation seasonality, deciduous riparian
broadleaf trees, flow length, dissolved oxygen, and depth to bedrock)
The final suitability map was calculated in terms of the total river
representing topographic, flow accumulation, land cover, hydro­
length. Out of the total river length of 60,290.56 km, the model pre­
climatic, soil and water quality. These variables are known to play a
dicted 34,265.75 km (56.83%) as unsuitable, followed by 12,979.36 km
significant role in the spawning, feeding habits, and overall distribution
(21.53%) as slightly suitable and 6807.67 km (11.29%) as moderately
of Tor putitora. The averaged model (5 replicate runs) show high training
suitable. The suitable and highly suitable categories constitute
AUC values (> 0.894) across the models and also higher than the test
5003.62 km (8.30%) and 1234.16 km (2.05%) respectively (Table 4 &
AUC values. The test AUC values, which show the actual model pre­
Fig. 5). Countrywise, India represents a total river length of 3652 km,
dictive power, were higher than 0.818. The partial AUC values of
followed by Nepal with 2248 km under suitable and highly suitable
replicated models ranged in between 0.890 and 0.894 (Table 3). The
categories. Bhutan showed the lowest river length of 338 km under
distribution of AUC ratio (AUCpartial/ AUCrandom) was significantly
suitable category. Among the Indian states and union territories, the
higher than the random expectations (Fig. 3). The model performance
largest river length under suitable and highly suitable categories was
tests showed satisfactory results (mean AUCfull 0.867, mean AUCpartial
found in Arunachal Pradesh with 1626 km, followed by Uttarakhand
0.895 and TSS 0.756) indicating that the final model is reasonably good
with 960 km and Himachal Pradesh with 772 km. The union territory of
and consistent. Out of the six environmental variables, minimum
Jammu & Kashmir, Darjeeling district of West Bengal and Sikkim

Fig. 2. Flowchart of methodology followed in the study.

4
R. Mahato et al. Acta Ecologica Sinica xxx (xxxx) xxx

Fig. 3. Partial AUC distribution for Tor putitora generated after 500 iterations with 5% omission in the receiver operating characteristic (ROC) space. Blue curve and
shaded bars show the frequency distribution of the ratios between AUC from model prediction and AUCrandom. Red curve show the distribution of AUC ratios for
random models. The higher distributional range of AUC ratios indicates the better predictive ability of the MaxEnt model.

Fig. 4. Species response curve (derived from MaxEnt). Predictor variables are shown in x-axis and species response in y-axis represented by Minimum Upstream
Elevation (elv_01), Flow Length (flow_acc_01), Precipitation Seasonality (hydroclim_15), Deciduous Broadleaf Trees (lc_avg_03), Depth to Bedrock (soil_avg_09), and
Dissolved Oxygen (do).

recorded suitable river lengths of 165 km, 69 km and 61 km respectively

Table 4
(Supplementary - 3A & B).
Suitable categories of Tor putitora in the Himalayan river system.
Generated by the authors using ArcGIS 10.3, http://esri.com
Categories Suitability index River length (km) Percentage A visual examination of the final model shows majority of the upper
Unsuitable < 0.1 34,265.75 56.83 reaches of the Himalayan rivers as unsuitable for T. putitora due to
Slightly suitable 0.1–0.3 12,979.36 21.53 higher elevation. On contrary, the relatively low altitude reaches,
Moderately suitable 0.3–0.5 6807.67 11.29 mainly comprising of the Indian states of Himachal Pradesh, Uttarak­
Suitable 0.5–0.7 5003.62 8.30
Highly suitable > 0.7 1234.16 2.05
hand, West Bengal, Sikkim, and Arunachal Pradesh including the
Total 60,290.56 100.00 countries of Nepal and Bhutan represents the suitable and highly suit­
able habitat of T. putitora (Fig. 5). Besides, the model also predicted
majority of the suitable areas in the river flow length of 8000 to

5
R. Mahato et al. Acta Ecologica Sinica xxx (xxxx) xxx

Fig. 5. Predictive habitat suitability of Tor putitora in the Himalayan river system.

30,000 m. The suitable habitats were mainly predicted in the stems of Indian subcontinent due to suitable topographic and hydroclimatic
the major rivers like the Sutlej (Himachal Pradesh), Ganga and Ram­ characteristics.
ganga (Uttarakhand), Trishuli and Kosi (Nepal), Teesta (West Bengal The model predicted only 10.35% of the total river length as suitable
and Sikkim), Puna Tsang Chhu (Bhutan), and Kameng, Subansiri, Siang, habitat of T. putitora in the entire Himalayan river system that point
Dibang, Lohit, and Noa-Dihing (Arunachal Pradesh). These localities towards its limited area of habitation. The model also predicted about
represent the main spawning and feeding grounds of T. putitora in the 11% of the total river length under moderately suitable that forms the

6
R. Mahato et al. Acta Ecologica Sinica xxx (xxxx) xxx

future potential habitats of the species and needs proper exploration and covered with sand, silt and small boulders [6].
confirmation. Considering the endangered status and reported declining The findings of wide distribution and restricted area of occupancy of
population in wild stocks, the model results would help to identify the T. putitora conforms to the report of IUCN, [9]. The results also predicted
priority areas for conservation and restoration of T. putitora in the Hi­ suitable areas in the rivers of Northeast Himalayas particularly Kameng,
malayan river system. Subansiri, Siang, Dibang, Lohit and Noa-Dihing in Arunachal Pradesh
that have been reported as unexplored potential habitats of T. putitora
4. Discussion [2]. Hence, these predictions are important in the context of a 50%
decline in the past and 80% decline in future wild stocks of the species
The results show the occurrences of T. putitora in larger streams and due to regulations in its river habitats [9,75]. As per available literature,
rivers in low minimum upstream elevation and elongated flow lengths. habitat alteration by dams, barrages, illegal sand/boulder mining,
Besides, low percentage of riparian deciduous broadleaf trees, high poaching and indiscriminate fishing of brooders are largely responsible
precipitation seasonality, high dissolved oxygen, and high depth to for the endangerment of T. putitora [75–81]. Besides, the life history and
bedrock also play low to moderate role in predicting the distribution of ecological traits like migratory habit, low fecundity and delayed sexual
T. putitora. Based on the results, the species did not occur in the absolute maturity have been also suggested to play a vital role in evaluating the
smallest streams in conformity with its occurrence reports from the vulnerability of the species [63,82–84]. Therefore, conservation and
major rivers and tributaries of Himalayan river system [2]. The model restoration strategies should be focused on the unexplored areas and the
showed a negative relationship of T. putitora with high upstream critical suitable habitats. According to experts, ex-situ conservation of
elevation by restricting the predictions to the elevation range of the species through development of hatcheries would be a sustainable
<1000 m (with highest possible range up to 2100 m). The results are in conservation action but less genetic variability has been observed in
agreement with the reported occurrence of the species up to an altitude hatchery stocks compared to its wild populations [85]. So, priority must
of 2000 m [60], up to 850 m [5], 600–1200 m [61], 70 to 1891 m [62] in be given to the potential unexplored habitats by restricting human ac­
India, and 2100 m in Nepal [63]. The hydroclimatic variables like tivities like fishing, extraction of sand, gravels and boulders, and by
temperature and precipitation are highly correlated with the elevation. withdrawing the proposed dams or proper environmental impact
Hence, the high predictive power of the model in the lower altitude assessment before construction of dams for hydropower. As of now,
areas can also be attributed to the temperature and precipitation re­ numbers of dams have been proposed in the major rivers of the Hima­
quirements of the species. As per previous studies, the spawning grounds layan region which may possibly further reduce the already fragmented
of T. putitora are characterized by water temperature varying from 11 to habitat of T. putitora.
30.5 ◦ C [6,7] and the feeding grounds in the range of 14 to 22 ◦ C [6].
The river habitat is an important determinant of the condition of fish 5. Conclusion
assemblages in rivers [64,65] and there is a reciprocal relationship be­
tween river habitat and the flow length. In simple words, an elongated Our study shows the predictive distribution of T. putitora in the Hi­
flow of river offers the possibilities of more diverse river habitats. Flow malayan river system using MaxEnt modelling technique. We used
length is a linear aspect of drainage network and refers to the total topographic, hydroclimatic, land cover, soil and water quality variables
length of stream channels in the drainage basin [66]. The flow length to evaluate the performance of model. The performance of model was
had a positive relationship, with higher values of flow length predicting high and reasonably good. The model predicted only 10.35% of the total
a high likelihood of T. putitora occurrences. Generally, Tor species prefer river length as suitable to highly suitable indicating its limited area of
fast-flowing rivers and streams with rocky/stony substrate [62]. Hence, habitation. About 11% of the river length has been predicted as
the presence of higher pools and ripples in the lower gradients along moderately suitable which calls for further explorations and confirma­
elongated stretches of the stream offers suitable habitats for the species. tion of the species occurrence. Based on the reported declining wild
Riparian forests are an important habitat for freshwater fish world­ stocks of the species, we reccommed recommend focussed conservation
wide, because forests provide fish with both shelter and food including activities towards unexplored and critical habitats following the model
fruits, seeds, and canopy insects [67–69]. Hence, the moderate contri­ predictions. We also recommend lessening anthropogenic activities like
butions of deciduous broadleaf trees in the model predictions could be indiscriminate fishing, sand/boulder mining, and construction of dams
attributed to the feeding requirements of the species. The rivers along over the suitable river habitats of T. putitora. Further, our model appears
the Himalayan foothills, considered to be the actual geographic range of to be constrained by the vastness of the study area. So, we recommend
T. putitora is mostly characterized by deciduous broadleaf forest. researchers to carry out distribution modelling at watershed level by
Therefore, the decayed and semi-decayed plant matter of the riparian integrating the local level environmental factors for more reliable pre­
forests forms essential food for the species. The feeding habit of diction of habitat suitability.
T. putitora on plant matters has been reported in India [70,71],
Bangladesh [72] and Nepal [73]. Funding
We also found a positive association of T. putitora with high precip­
itation seasonality, dissolved oxygen and depth to bedrock. Seasonality The study does not receive any funding.
plays a vital role in influencing the perseverance of living organisms.
Seasonal shifts in climatic conditions influence the availability of re­
Ethics approval
sources, which determines the presence or absence of species in an
environment at temporal and spatial scale [74]. Precipitation season­
The protocol was approved by the Institutional Ethical Committee of
ality, a measure of the range of annual precipitation plays a key role in
the Rajiv Gandhi University, Itanagar, India (Ref. No. IEC/RGU/01).
breeding and feeding habits of freshwater fishes. Dissolved oxygen is an
important parameter in assessing water quality because of its influence
on the organisms living within a body of water. According to Bhatt et al., Consent to participate
[6], the dissolved oxygen varies from 5.2 to 12.9 mg/l in feeding
grounds while it ranges between 6.4 and 11 mg/l in the spawning The authors give consent to participate.
grounds of T. putitora [6,7]. The preference of depth to bedrock varies
among different freshwater fishes. The model prediction towards the Consent for publication
higher values of depth to bedrock may be attributed to the habitat
preferences of T. putitora in large water volumes and river beds mostly The authors give consent for the publication of the article.

7
R. Mahato et al. Acta Ecologica Sinica xxx (xxxx) xxx

Availability of data and material [15] B.M. Kurup, K.V. Radhakrishnan, Tor remadevii, a new species of Tor (Gray) from
Chinnar wildlife sanctuary, Pambar River, Kerala, southern India, J. Bombay Nat.
Hist. Soc. 107 (2011) 227–230.
The authors ensure the availability of data and material as and when [16] B.R. Jha, A. Rayamajhi, N. Dahanukar, A. Harrison, A. Pinder, Tor putitora, in: The
asked for. IUCN Red List of Threatened Species, 2018, p. 2018, https://doi.org/10.2305/
IUCN.UK.2018-2.RLTS.T126319882A126322226.en, e.T126319882A126322226.
[17] A.K. Rai, A.K. Rana, K.B. Bastola, Comparative growth study of indigenous fish
Code availability species Mahseer, Katle and Asla fed with local feed ingredients, J. Nep. Agric. Res.
1 (1997) 57–61.
[18] J.D. Bista, S.K. Wagle, S. Prasad, T.B. Gurung, A.K. Rai, Growth and Maturity of
Not applicable.
Sahar Tor putitora in Controlled Environment in Mid Hill Pokhara Valley, Nepal,
Asian Fisheries Society, Indian branch, Kochi, India, 2007, p. 143.
Author contributions [19] A.K. Rai, S.K. Wagle, N. Pradhan, S.R. Basnet, A.P. Baidya, T.B. Gurung, J.D. Bista,
M.K. Shrestha, A.B. Thapa, Status of sahar Tor putitora and its prospects in Nepal,
in: First International Symposium of the Mahseer, Kualampur, Malasiya, 2006,
Ranjit Mahato and Santoshkumar Abujam prepared the initial draft pp. 1–23.
of the manuscript. Ranjit Mahato prepared the thematic layers, and [20] Ministry of Agriculture and Forests, Bhutan, Bhutan: A Pioneer in Mahseer
conservation. http://www.moaf.gov.bt/bhutan-a-pioneer-in-mahseer-conservation
Dhoni Bushi collected the fish sample. Oyi Dai Nimasow and Gibji , 2020 (accessed 11 June 2021).
Nimasow commented, edited and revised the manuscript by consulting [21] S. Roy, S. Ray, S.K. Saikia, Indicator environmental variables in regulating the
the relevant literature. Debanshu Narayan Das further revised the distribution patterns of small freshwater fish Amblypharyngodon mola in India and
Bangladesh, Ecol. Indic. 120 (2021), 106906, https://doi.org/10.1016/j.
manuscript after proofreading. Finally, the manuscript was read by all
ecolind.2020.106906.
the authors and approved it to submit for possible publication. [22] V. Greshishchev, N. Onikura, H. Iyooka, Environmental factors influencing fish
species distribution in irrigation channels around Ariake Sea, Kyushu, Japan,
Irrigat. Drainage Syst. Eng. 4 (2015) 139, https://doi.org/10.4172/2168-
Declaration of Competing Interest 9768.1000139.
[23] C.M. Hamilton, B.L. Batemana, J.M. Gorzoa, B. Reid, W.E. Thogmartinc, M.
Z. Peeryd, P.J. Heglunde, V.C. Radeloffa, A.M. Pidgeona, Slow and steady wins the
The author(s) declare no competing interests. race? Future climate and land use change leaves the imperiled Blanding’s turtle
(Emydoidea blandingii) behind, Biol. Conserv. 222 (2018) 75–85, https://doi.org/
10.1016/j.biocon.2018.03.026.
Acknowledgments [24] J.A. Johnson, R. Parmar, K. Ramesh, S. Sen, R.S. Murthy, Fish diversity and
assemblage structure in Ken River of Panna landscape, central India, J. Threat.
The authors are thankful to the Soil and Limnological Laboratory, Taxa 4 (2012) 3161–3172, https://doi.org/10.11609/JoTT.o3024.3161-72.
[25] A. Ruiz-Navarro, P.K. Gillingham, J.R. Britton, Predicting shifts in the climate
Department of Geography, Rajiv Gandhi University, Rono Hills, Doi­
space of freshwater fishes in Great Britain due to climate change, Biol. Conserv.
mukh, Arunachal Pradesh (India) for facilitating necessary infra­ 203 (2016) 33–42, https://doi.org/10.1016/j.biocon.2016.08.021.
structural and laboratory facilities to carry out the work. The authors [26] S.J. Phillips, R.P. Anderson, R.E. Schapire, Maximum entropy modeling of species
also acknowledge GBIF, EarthEnv, and the authors of other published geographic distributions, Ecol. Model. 190 (2006) 231–259, https://doi.org/
10.1016/j.ecolmodel.2005.03.026.
articles for making available the data sources for this work. [27] GBIF, The Global Biodiversity Information Facility, What Is GBIF?. https://www.gb
if.org/what-is-gbif, 2019 (accessed 5 May 2021).
[28] R.J. Hijmans, S. Phillips, J. Leathwick, J. Elith, Dismo Package for R (Version
Appendix A. Supplementary data
4.0.3). https://cran.r-project.org/package=dismo, 2011.
[29] R. Kadmon, O. Farber, A. Danin, Effect of roadside bias on the accuracy of
Supplementary data to this article can be found online at https://doi. predictive maps produced by bioclimatic models, Ecol. Appl. 14 (2004) 401–413,
https://doi.org/10.1890/02-5364.
org/10.1016/j.chnaes.2022.01.004.
[30] S.D. Veloz, Spatially autocorrelated sampling falsely inflates measures of accuracy
for presence-only niche models, J. Biogeogr. 36 (2009) 2290–2299, https://doi.
References org/10.1111/j.1365-2699.2009.02174.x.
[31] M.E. Aiello-Lammens, R.A. Boria, A. Radosavljevic, B. Vilela, R.P. Anderson,
spThin: an R package for spatial thinning of species occurrence records for use in
[1] F. Hamilton, An Account of the Fishes Found in the River Ganges and Its Branches,
ecological niche models, Ecography 38 (2015) 541–545, https://doi.org/10.1111/
George Ramsay and Co., Edinburgh, 1822.
ecog.01132.
[2] J.P. Bhatt, M.K. Pandit, Endangered Golden mahseer Tor putitora Hamilton: a
[32] S. Domisch, G. Amatulli, W. Jetz, Near-global freshwater-specific environmental
review of natural history, Rev. Fish Biol. Fish. 26 (2016) 25–38, https://doi.org/
variables for biodiversity analyses in 1 km resolution, Sci. Data 2 (2015), 1–13
10.1007/s11160-015-9409-7.
Data available online at, http://www.earthenv.org, https://doi.org/10.1038/sdat
[3] W. Oo, Inland fisheries of the Union of Myanmar, in: T. Petr, S.B. Swar (Eds.), Cold
a.2015.73.
Water Fisheries in the Trans-Himalayan Countries vol. 431, FAO Fish, 2002,
[33] Environmental Systems Research Institute, ArcMap 10.3 (Version 10.3) [Computer
pp. 71–77.
Software], Environmental Systems Research Institute, Redlands, California, USA,
[4] K.G. Rajbanshi, I. Csavas, Aquaculture development in Bhutan - a report prepared
2016.
for the establishment of fish seed production Centre project, FAO FI:DP/BHU80/
[34] J. Miller, Species distribution modeling, Geogr, Compass 4 (2010) 490–509,
007, Field Doc 1 (1982) 46.
https://doi.org/10.1111/j.1749-8198.2010.00351.x.
[5] P.K. Talwar, A.G. Jhingran, Inland Fishes of India and Adjacent Countries vol. 1,
[35] V. Braunisch, J. Coppes, R. Arlettaz, R. Suchant, H. Schmid, K. Bollmann, Selecting
Rotterdam, A.A. Balkema, 1991, p. 541.
from correlated climate variables: a major source of uncertainty for predicting
[6] J.P. Bhatt, P. Nautiyal, H.R. Singh, Status (1993–1994) of the endangered fish
species distributions under climate change, Ecography 36 (2013) 971–983, https://
Himalayan mahseer Tor putitora (Hamilton) (Cyprinidae) in the mountain reaches
doi.org/10.1111/j.1600-0587.2013.00138.x.
of the river ganga, Asian Fish. Sci. 17 (2004) 341–355.
[36] M.H. Graham, Confronting multicollinearity in ecological multiple regression,
[7] C.B. Joshi, Alikhad (Sutlej river system): Juvenile habitat and breeding grounds, in:
Ecology 84 (2003) 2809–2815, https://doi.org/10.1890/02-3114.
P. Nautiyal (Ed.), Mahseer the Game Fish, Jagdamba Publication, Dehradun, 1994,
[37] K.P. Vatcheva, M. Lee, J.B. McCormick, M.H. Rahbar, Multicollinearity in
pp. B102–B111.
regression analyses conducted in epidemiologic studies, Epidemiology 6 (2016)
[8] A. Darshan, S.K. Abujam, D.N. Das, Biodiversity of Fishes in Arunachal Himalaya,
227, https://doi.org/10.4172/2161-1165.1000227.
Academic Press, Elsevier, Massachusetts, 2018.
[38] C. Merow, M.J. Smith, J.A. Silander, A practical guide to MaxEnt for modeling
[9] IUCN, Red List of Threatened Species, 2021, https://doi.org/10.2305/IUCN.
species’ distributions: what it does, and why inputs and settings matter, Ecography
UK.2018-2.RLTS.T126319882A126322226.en (accessed 1 June 2021).
36 (2013) 1058–1069, https://doi.org/10.1111/j.1600-0587.2013.07872.x.
[10] R. Fricke, W.N. Eschmeyer, R. Van der Laan, Eschmeyer’s Catalog of Fishes:
[39] B. Naimi, M.B. Araújo, Sdm: a reproducible and extensible R platform for species
Genera, Species, Online Version, http://researcharchive.calacademy.org/research
distribution modeling, Ecography 39 (2016) 368–375, https://doi.org/10.1111/
/ichthyology/catalog/fishcatmain.asp, 2021 (accessed 1 March 2021).
ecog.01881.
[11] L. Arunkumar, C. Basudha, Tor barakae, a new species of mahseer fish (Cyprinidae:
[40] C.F. Dormann, J. Elith, S. Bacher, C. Buchmann, G. Carl, G. Carré, R.G. Marquéz,
Cyprininae) from Manipur, India, Aquaculture 4 (2003) 271–276.
B. Gruber, B. Lafourcade, P.J. Leitão, T. Münkemüller, C. McClean, P.E. Osborne,
[12] W.H. Sykes, On the fishes of the Deccan, Proc. Zool. Soc. London 6 (1839)
B. Reineking, B. Schröder, A.K. Skidmore, D. Zurell, S. Lautenbach, Collinearity: a
157–165.
review of methods to deal with it and a simulation study evaluating their
[13] A.G.K. Menon, Taxonomy of mahseer fishes of the genus Tor Gray with description
performance, Ecography 36 (2013) 27–46, https://doi.org/10.1111/j.1600-
of a new species from the Deccan, J. Bombay Nat. Hist. Soc. 89 (1992) 210–228.
0587.2012.07348.x.
[14] T.C. Jerdon, On the fresh-water fishes of southern India, Madras J. Lit. Sci. 15
(1849) 302–346.

8
R. Mahato et al. Acta Ecologica Sinica xxx (xxxx) xxx

[41] A. Farrell, G. Wang, S.A. Rush, J.A. Martin, J.L. Belant, A.B. Butler, D. Godwin, Himalaya, in: J.S. Datta Munshi (Ed.), Modern Trends in Fish Biology Research,
Machine learning of large-scale spatial distributions of wild turkeys with high- Narendra Publishing House, New Delhi, 2000, pp. 1–18.
dimensional environmental data, Ecol. Evol. 9 (2019) 5938–5949, https://doi.org/ [62] P. Nautiyal, Review of the art and sciences of Indian Mahseer (game fish) from
10.1002/ece3.5177. nineteenth century: road to extinction or conservation? Proc. Nat. Acad. Sci. India
[42] X. Feng, D.S. Park, Y. Liang, R. Pandey, M. Papeş, Collinearity in ecological niche Sect. B 84 (2014) 215–236, https://doi.org/10.1007/s40011-013-0233-3.
modeling: confusions and challenges, Ecol. Evol. 9 (2019) 10365–10376, https:// [63] T.K. Shrestha, The Mahseer in the Rivers of Nepal Disrupted by Dams and Ranching
doi.org/10.1002/ece3.5555. Strategies, R.K. Printers, Teku, Kathmandu, 1997.
[43] S.A. Manzoor, G. Griffiths, M. Lukac, Species distribution model transferability and [64] L. Wang, P.W. Seelbach, R.M. Hughes, Introduction to landscape influences on
model grain size–finer may not always be better, Sci. Rep. 8 (2018) 7168, https:// stream habitats and biological assemblages, in: R.M. Hughes, L. Wang, P.
doi.org/10.1038/s41598-018-25437-1. W. Seelbach (Eds.), Landscape Influences on Stream Habitats and Biological
[44] R.K. Sony, S. Sen, S. Kumar, M. Sen, K.M. Jayahari, Niche models inform the effects Assemblages, American Fish Society Symposium 48, Bethedsda, 2006, pp. 1–23.
of climate change on the endangered Nilgiri Tahr (Nilgiritragus hylocrius) [65] A.P. Fialho, L.G. Oliveira, F.L. Tejerina-Garro, B.de Mérona, fish-habitat
populations in the southern Western Ghats, India, Ecol. Eng. 120 (2018) 355–363, relationship in a tropical river under anthropogenic influences, Hydrobiologia 598
https://doi.org/10.1016/j.ecoleng.2018.06.017. (2008) 315–324, https://doi.org/10.1007/s10750-007-9165-3.
[45] S. Chatterjee, A.S. Hadi, Analysis of collinear data, in: W.A. Shewhart, S.S. Wilks [66] R.E. Horton, Erosional development of streams and their drainage basins: hydro-
(Eds.), Regression Analysis by Example, John Wiley & Sons Inc., New Jersey, 2006, physical approach to qualitative morphology, Bull. Geol. Soc. Am. 56 (1945)
pp. 221–258. 275–370, https://doi.org/10.1177/030913339501900406.
[46] R. Engler, L.T. Waser, N.E. Zimmermann, M. Schaub, S. Berdos, C. Ginzler, [67] M. Goulding, The Fishes and the Forest, University of California Press, Los Angeles,
A. Psomas, Combining ensemble modeling and remote sensing for mapping 1980.
individual tree species at high spatial resolution, For. Ecol. Manag. 310 (2013) [68] S.D. Gerking, Feeding Ecology of Fish, Academic Press, San Diego, 1994.
64–73, https://doi.org/10.1016/j.foreco.2013.07.059. [69] R.H. Lowe-McConnell, Ecological Studies in Tropical Fish Communities,
[47] N.E. Zimmermann, T.C. Edwards, G.G. Moisen, T.S. Frescino, J.A. Blackard, Cambridge University Press, Cambridge, 1987.
Remote sensing-based predictors improve distribution models of rare, early [70] B. Kishor, J.P. Bhatt, V.S. Rawat, P. Nautiyal, Variations in food habit of the
successional and broadleaf tree species in Utah, J. Appl. Ecol. 44 (2007) Himalayan mahseer – Tor putitora (ham.) inhabiting the ganga river system in
1057–1067, https://doi.org/10.1111/j.1365-2664.2007.01348.x. Garhwal region, Indian J. Fish. 45 (1998) 113–118.
[48] G. Nimasow, O.D. Nimasow, J.S. Rawat, G. Tsering, T. Litin, Remote sensing and [71] P. Nautiyal, M.S. Lal, Food and feeding habits of fingerlings and juveniles of
GIS-based suitability modeling of medicinal plant (Taxus baccata Linn.) in Tawang mahseer (Tor putitora ham) in Nayar river, J. Bombay Nat. Hist. Soc. 81 (1984)
district, Arunachal Pradesh, India, Curr. Sci. 110 (2016) 219–227, https://doi.org/ 642–646.
10.18520/cs/v110/i2/219-227. [72] M.A. Alam, S. Dewan, M.I. Hossain, M.G. Hussain, Food and feeding habit of T.
[49] A.T. Peterson, M. Papeş, M. Eaton, Transferability and model evaluation in putitora (Hamilton), Univ. J. Zool. Rajshahi Univer. 21 (2002) 21–22.
ecological niche modeling: a comparison of GARP and Maxent, Ecography 30 [73] V.K. Mahaseth, Food and feeding habit of T. putitora of Mahakali River, Nepal,
(2007) 550–560, https://doi.org/10.1111/j.2007.0906-7590.05102.x. Bibechana 13 (2015) 121–131.
[50] S. Phillips, M. Dudík, Modeling of species distributions with Maxent: new [74] G.A. Dayton, Seasonality, in: S.E. Jørgensen, B.D. Fath (Eds.), Encyclopedia of
extensions and a comprehensive evaluation, Ecography 31 (2008) 161–175, Ecology, Elsevier B.V, Amsterdam, 2008, pp. 3168–3171.
https://doi.org/10.1111/j.0906-7590.2008.5203.x. [75] K.K. Sharma, V.C. Mohan, U. Kouser, Comparative accounts of merestic count and
[51] A. Radosavljevic, R.P. Anderson, Making better Maxent models of species morphometric measurements of Golden mahseer (Tor putitora) among Chenani
distributions: complexity, overfitting and evaluation, J. Biogeogr. 41 (2014) hydroelectric dam, Jhajjar stream and Dansar stream, (J&K) India, Indian J. Appl.
629–643, https://doi.org/10.1111/jbi.12227. Res. 5 (2015) 772–774.
[52] R. Moreno, R. Zamora, J.R. Molina, A. Vasquez, M.Á. Herrera, Predictive modeling [76] N. Gupta, K. Sivakumar, V.B. Mathur, M.A. Chadwick, The ‘tiger of Indian rivers’:
of microhabitats for endemic birds in south Chilean temperate forests using stakeholders’ perspectives on the golden mahseer as a flagship fish species, Area 46
maximum entropy (Maxent), Ecol. Informat. 6 (2011) 364–370, https://doi.org/ (2014) 389–397, https://doi.org/10.1111/area.12124.
10.1016/j.ecoinf.2011.07.003. [77] B. Khajuria, S. Langer, N.K. Tripathi, Status of Golden mahseer (Tor putitora) in
[53] X.Q. Yang, S. Kushwaha, S. Saran, J. Xu, P. Roy, Maxent modeling for predicting Jammu region (J&K), Int. J. Recent Sci. Res. 4 (2013) 1154–1156.
the potential distribution of medicinal plant, Justicia adhatoda L. in lesser [78] W.S. Lakra, M. Goswami, U.K. Sarkar, Conservation biology of Indian mahseer,
Himalayan foothills, Ecol. Eng. 51 (2013) 83–87, https://doi.org/10.1016/j. Indian, J. Anim. Sci. 80 (2010) 98–108.
ecoleng.2012.12.004. [79] P. Nautiyal, The Golden Mahseer (a Threatened Fish of Himalaya), Lambert
[54] R. Baldwin, Use of maximum entropy modeling in wildlife research, Entropy 11 Academic Publishing, Leipzig, 2011.
(2009) 854–866, https://doi.org/10.3390/e11040854. [80] M.K. Pandit, R.E. Grumbine, Potential effects of ongoing and proposed,
[55] A.H. Fielding, J.F. Bell, A review of methods for the assessment of prediction errors hydropower development on terrestrial biological diversity in the Indian
in conservation presence/ absence models, Environ. Conserv. 24 (1997) 38–49, Himalaya, Conserv. Biol. 26 (2012) 1061–1071, https://doi.org/10.1111/j.1523-
https://doi.org/10.1017/S0376892997000088. 1739.2012.01918.x.
[56] J.M. Lobo, A. Jiménez-Valverde, R. Real, AUC: a misleading measure of the [81] S.S. Pathani, Biology of tor and putitor mahseer, in: P. Nautiyal (Ed.), Mahseer the
performance of predictive distribution models, Glob. Ecol. Biogeogr. 17 (2008) Game Fish, Jagdamba Publications, Dehradun, 1994, pp. B86–B90.
145–151, https://doi.org/10.1111/j.1466-8238.2007.00358.x. [82] P. Nautiyal, H.R. Singh, River projects and endangered hill stream mahseer Tor
[57] A.T. Peterson, M. Papeş, J. Soberón, Rethinking receiver operating characteristic putitora in Indian uplands of Garhwal Himalaya, a critical evaluation, Himalayan J.
analysis applications in ecological niche modeling, Ecol. Model. 213 (2008) 63–72, Environ. Zool. 3 (1989) 129–134.
https://doi.org/10.1016/j.ecolmodel.2007.11.008. [83] S.N. Ogle, Mahseer breeding and conservation and possibilities of commercial
[58] D. Adhikari, Z. Reshi, B.K. Datta, S.S. Samant, A. Chettri, K. Upadhaya, M.A. Shah, culture: The Indian experience, in: T. Petr, S.B. Swar (Eds.), Coldwater Fisheries in
P.P. Singh, R. Tiwary, K. Majumdar, A. Pradhan, M.L. Thakur, N. Salam, Z. Zahoor, the Trans-Himalayan Countries, FAO Fisheries Technical Paper 431, 2002,
S.H. Mir, Z.A. Kaloo, S.K. Barik, Inventory and characterization of new populations pp. 193–212.
through ecological niche modeling improve threat assessment, Curr. Sci. 114 [84] W.L.C. William, J.P. Tony, P. Daniel, A fuzzy logic expert system to estimate
(2018) 519–531. intrinsic extinction vulnerabilities of marine fishes to fishing, Biol. Conserv. 124
[59] O. Allouche, A. Tsoar, R. Kadmon, Assessing the accuracy of species distribution (2005) 97–111, https://doi.org/10.1016/j.biocon.2005.01.017.
models: prevalence, kappa and the true skill statistic (TSS), J. Appl. Ecol. 43 (2006) [85] N. Pradhan, S.K. Wagle, T.B. Gurung, Genetic variation of mahseer (Tor putitora)
1223–1232, https://doi.org/10.1111/j.1365-2664.2006.01214.x. populations from hatchery, lake and rivers of Nepal, in: T.B. Gurung, P.K. Pokharel,
[60] K.D.B. Cordington, Notes on Indian Mahseer, J. Bombay Nat. Hist. Soc. 46 (1946) C.R. Upreti, B.R. Joshi, S.M. Pradhan, S.K. Wagle (Eds.), Proceedings of the 8th
334–336. National Workshop on Livestock and Fisheries Research, Nepal Agricultural
[61] H.R. Singh, N. Kumar, Some aspects of ecology of hillstreams; stream morphology, Research Council (NARC), Nepal, 2011, pp. 35–43.
zonation, characteristics, and adaptive features of ichthyofauna in Garhwal