Recovery of the critically endangered gharial (Gavialis gangeticus) amid escalating anthropogenic pressures in the Ganga River Basin

The gharial (Gavialis gangeticus), endemic to the Indian subcontinent, has undergone a severe population and range decline through the mid-20th century (1940s–1970s) driven by hunting, habitat loss, and incidental mortality in fishing gear, meaning the recovery of this Critically Endangered river-specialist crocodilian in the Ganga River Basin has depended heavily on decades of augmentation and protection, yet a comprehensive basin-wide assessment of its current status remains lacking. The gharial (Gavialis gangeticus), is endemic to the Indian subcontinent. In the recent years it has undergone a severe population decline due to habitat loss and degradation, incidental mortality in fishing nets and hunting in the past. In India, though the species has shown signs of recovery in the Ganga River Basin due to intense conservation measures adopted, yet a comprehensive basin-wide assessment of its current population status and habitat condition remains lacking. Conservation initiatives launched in the mid-1970s have contributed to its sustained recovery of the species. Nevertheless, information on the current population status, habitat condition and long-term recovery is lacking. Boat-based surveys were conducted across 22 rivers covering 7,680 km (60.3% of the total river length of 12,741 km) to determine the minimum population size (MPS) of gharial in the Ganga River Basin. We recorded 3037 gharials in 13 rivers, with the highest number from the Chambal (MPS = 2,097), followed by the Ghaghra (MPS = 463) and Girwa rivers (MPS = 158), while less than 10 individuals were observed in the Sharda, Kosi, Bagmati, and Babai rivers. Across 13 gharial occupied rivers encounter rates ranging from 0.02 ± 0.01 ind./km (Ganga) to 7.90 ± 4.29 ind./km (Girwa), highest in Chambal (5.18 ± 0.50 ind./km) and lowest in Ganga and Kosi (≤0.02 ind./km). The Chambal, Gandak, and Ghaghra rivers had individuals from all size classes, with a high proportion of larger individuals (>1.8 m) in the population. Habitat suitability analysis revealed that most of the river network (10,716 km; 84.1%) was very low suitable for gharials, whereas only 1,320 km (10.4%) was classified as highly suitable habitat. Population trend analysis based on long-term data from the National Chambal Sanctuary revealed an exponential rate of increase of 7% per annum between 1979 and 2024, indicating substantial population recovery in this key conservation stronghold. However, this trend is specific to the Chambal population and should not be extrapolated to the wider Ganga Basin, where our river-wise survey data reveal a markedly uneven population status. Growth was most rapid during the initial intensive grow and release phase from 1979 to 1993 (15.6% per annum). The identified river stretches potentially represent priority sites, therefore requiring immediate conservation measures to support gharial recovery in the Basin.

Globally, growing conservation optimism supported by effective recovery plans prevented the extinction of several endangered and threatened species in recent decades (Bolam et al., 2021; Grace et al., 2021). Long-term monitoring exercises yield reliable population trends resulting in successful yet cautious recovery programs (Crees et al., 2016; Di Silvestro, 2023; Jhala et al., 2025). Aquatic megafauna such as crocodilians are large-bodied, semi-aquatic top predators that serve as bioindicators of environmental change in the ecosystems they inhabit (Mazzotti et al., 2009; Somaweera et al., 2020). India is home to three crocodilian species: the gharial (Gavialis gangeticus), the mugger or marsh crocodile (Crocodylus palustris), and the saltwater crocodile (Crocodylus porosus) (Hussain, 1999). Of these, the gharial is a specialist, evolutionarily distinct member of the once-diverse family Gavialidae, with the most restricted distribution among the three species and a Critically Endangered status under the IUCN Red List, making it a flagship and iconic species for river conservation (Lang et al., 2019; Martin, 2019; Panda et al., 2023; Rio and Mannion, 2021; Sharma et al., 2021). Historically, the gharial inhabited major river basins of the Indian subcontinent, including the Indus, Ganga, Mahanadi, Brahmaputra, Kaladan, and Irrawaddy, spanning present-day Pakistan, India, Nepal, Bangladesh, Bhutan, and Myanmar (Choudhury and Chowdhury, 1986). By the early 1960s, gharial populations had severely declined due to widespread hunting, habitat destruction, trophy poaching, and mortality in passive fishing gear (Bustard, 1975; Katdare et al., 2011; Whitaker, 2007). Consequently, the species was pushed to the brink of extinction, with an over 80% decline in population within less than two generations and a substantial range contraction, triggering urgent conservation measures (Choudhury and Chowdhury, 1986). The alarming population decline prompted the inception of ‘Project Crocodile’ in the mid-1970s in India, which focused on habitat protection and a ‘grow and release’ program to stabilize the declining populations in the wild (Bustard, 1976; Singh, 1976). These efforts resulted in population recovery in some rivers, such as Chambal, Girwa in the Ganga Basin, and averted local extinction across several other rivers in India (Sharma et al., 2021; Vashistha et al., 2021).

Despite recovery through population augmentation efforts, gharial populations remains small, isolated, and restricted primarily to the Ganga, Mahanadi, and Brahmaputra basins in India and Nepal, with occasional records from Bangladesh (Lang et al., 2019; Panda et al., 2023; Rabbe Md. Fazle et al., 2023; Vashistha et al., 2021). The species is currently distributed across more than 14 spatially distinct population units, most characterised by small population sizes and limited connectivity (Lang et al., 2019). Conservation and long-term monitoring efforts have been largely concentrated in the Ganga and Mahanadi basins, particularly within the riverine Protected Areas (PAs) such as the National Chambal Sanctuary along the Chambal River and Katarniaghat Wildlife Sanctuary along the Girwa River in the Ganga Basin, and Satkosia Gorge Wildlife Sanctuary in the Mahanadi Basin (Choudhury and Chowdhury, 1986). Apart from these PAs, conservation efforts, including population monitoring, have been inconsistent, often conducted at irregular intervals using different methodologies (Lang et al., 2019; Nair et al., 2012; Nair and Katdare, 2013; Sharma et al., 2025a; Vashistha et al., 2023), leaving a significant gap in comprehensive, standardised population assessments across the Ganga Basin. (Nair and Katdare, 2013; Sharma et al., 2025a; Vashistha et al., 2023). Given the species’ highly fragmented distribution and persistent threats across its extent, the long-term survival of gharial populations is contingent upon timely and systematic population assessment. This underscores the need for basin-wide population assessment alongside identification of critical habitats suitable for recolonization, to guide targeted conservation action.

In the present study, we assessed the population status, including distribution, minimum population size (MPS), size-class composition, and habitat suitability, alongside long-term population trend to inform gharial conservation planning across the Ganga Basin and beyond. The study addressed the following research questions: (i) what is the overall population trend (ii) the current spatial distribution of gharials across the Ganga basin, (iii) how are population size and size-class composition structured across river systems, (iv) which environmental and anthropogenic factors most strongly influence gharial habitat suitability, and (v) what proportion of river stretch is suitable currently to support long-term conservation of gharial and potential recolonisation in the Basin. To address these questions, we surveyed 22 rivers covering 7,680 km of river stretches, representing 59.8% of the total river length, and combined field-based population surveys with spatial analysis of habitat suitability. The study identified three distinct components: (i) current minimum population size and distribution across 22 rivers, (ii) basin-wide habitat-suitability modelling, and (iii) a Chambal-specific analysis of long-term population change.

Study area

The population and habitat suitability assessments were conducted in the Ganga River Basin, one of the largest river systems in India, spanning 861,404 km 2 and covering approximately 26% of the country’s geographical area (Jain and Singh, 2020; Nandargi and Shelar, 2018). The basin is bounded by the Himalayas from the north, the Chota Nagpur Plateau in the south, the Aravalli Range in the west, and the Brahmaputra Ridge in the east (Hussain and Badola, 2001). The basin encompasses five distinct biogeographic zones, including the Himalayas, Semi-Arid, Deccan Peninsula, Gangetic Plain, and Coasts, with climatic conditions ranging from alpine to tropical (Rodgers and Panwar, 1988). Annual precipitation in the Basin varies between 500 and 2,500 mm, with 80% occurring during the monsoon season (Nandargi and Shelar, 2018). The Basin has an intricate network of rivers, of which 22 rivers with a length of 12,741 km were selected. Consolidate the repeated rationale into one sentence: The 22 rivers (totalling 12,741 km) selected for this study fall within the historic distribution range of the gharial and include stretches with documented past occurrences or sporadic sightings, making them ecologically relevant for evaluating current status and habitat suitability. About 7,680 km (59.8%) of these 22 rivers were surveyed to assess the current presence, distribution, and population structure of the gharial (Figure 1).

Population status

Data collection

Boat-based visual encounter river surveys were conducted to document the occurrence, minimum population size and size-class composition of gharial in select rivers in the Ganga River Basin (Supplementary Table 1). Surveys were undertaken during the post-monsoon (PoM) (November–March) periods between 2020 and 2023, coinciding with the winter basking period that maximizes the detection of gharials (Hussain, 1999). Each river was surveyed once, during daylight hours between 1000 and 1600 hrs, using a motorboat equipped with a 25-horsepower outboard engine, maintaining a constant speed of 8–10 km h -1 moving unidirectionally in the flow direction.This boat-based visual encounter approach, survey speed, and timing follow established, standard protocols for crocodilian population surveys (Bayliss, 1987; Hussain, 1999). Two observers equipped with 8 × 40 mm binoculars systematically scanned the riverbanks to detect basking gharial (Hussain, 1999). Gharials were identified based on standard morphological characteristics, following the identification guide (Brazaitis, 1973). For each sighting, the number of individuals, visually estimated total length, sex, and associated habitat variables were recorded. Sex was determined based on external morphological characteristics, with adult males identified by the presence of a nasal protuberance (ghara) (Brazaitis, 1973). Individuals were classified into five size classes based on total length: hatchlings (<0.6 m), yearlings (0.6–0.9 m), juveniles (0.9–1.8 m), sub-adults (1.8–3.0 m), and adults (>3.0 m), following established size–life stage relationships for gharials (Bustard and Maharana, 1982; Hussain, 1999). Partially submerged individuals were assigned to the nearest confident size class based on visible body proportions; individuals of uncertain size/sex were recorded as unclassified (excluded from size-class/sex-ratio analyses, retained in the total minimum count). Each surveyed river was divided into 5 km units, referred to as Biodiversity Evaluation Units (BEUs), using ArcGIS 10.6 (ESRI, Redlands, USA). For each 5-km BEU, we recorded active fishing gear or boats sighted (fishing intensity), people present on the riverbank or in the water (human presence), active or recent sand-extraction activity (sand mining), water off-take points or structures (water extraction), and ferry crossings or boats (ferry intensity). Raw counts were standardised by survey effort (counts per unit distance surveyed) and rescaled to a 0–1 index using min–max normalisation across all surveyed BEUs before use in the habitat-suitability model. Across all rivers, a total of approximately 2280 observer-hours were expended, corresponding to 190 days of active survey effort (Supplementary Table 1).

Detection assumptions, visibility, and size-class bias

Visual encounter surveys assume that all animals present and available for detection are observed (Bayliss, 1987; Hussain, 1999). Given the semi-aquatic and basking behavior of gharials, detection probability is highest during winter and post-monsoon periods when individuals spend prolonged periods basking on exposed sandbanks. To minimise visibility bias, surveys were conducted under favourable weather conditions (clear visibility, low wind, and no precipitation), and survey timing was standardised to mid-day hours when basking activity peaks. Survey days that did not meet these weather criteria were rescheduled to the next suitable day.

Despite these precautions, the problem of imperfect detection is unavoidable in a single-pass visual encounter survey due to factors such as partial submergence, channel width, and riverbank complexity. To account for these limitations, we adopted a conservative approach by reporting total gharial sighted as Minimum Population Size (MPS), which represents the minimum number of individuals observed and avoids assumptions of perfect detection (Balaguera-Reina et al., 2018; Barão-Nóbrega et al., 2022). As surveys were conducted only once per river during the study period, detection probability was not explicitly incorporated and modeled, and MPS values should therefore be interpreted as conservative estimates of true population size. Furthermore, visual estimation of total length is subject to observer-related error, particularly under varying visibility conditions and river widths. As a result, size-class data are presented with caution and are intended to reflect population structure rather than precise individual body lengths.

Data analysis

The total number of gharials observed during the surveys was treated as Minimum Population Size (MPS), defined as the minimum number of individuals present within the surveyed area (Barão-Nóbrega et al., 2022). The MPS provides a conservative estimate of population size (Balaguera-Reina et al., 2018). We further calculated encounter rate (individuals km -1) using the MPS. The encounter rate was calculated as the relative density index to facilitate comparisons among rivers with differing survey lengths (Bayliss, 1987). Encounter rate for each BEU was calculated as the number of gharials sighted divided by BEU length (5 km); river-specific encounter rate is the mean (± SEM) of BEU-level rates across all BEUs within the gharial-occupied length, including zero-count BEUs situated between the uppermost confirmed sighting and the downstream confluence. Zero-count BEUs were retained in this calculation because excluding them would inflate the river-specific encounter rate and make it inconsistent with the transparent total-count ÷ total-surveyed-length index reported as ERw in Table 1. We note that treating BEUs as independent replicates does not account for spatial autocorrelation along a river channel; the reported SEM should therefore be interpreted as a descriptive measure of spatial variability rather than a formal measure of statistical independence. The river-specific encounter rate was calculated only for gharial-occupied lengths, defined as the stretch between the uppermost BEU with confirmed gharial sightings and the downstream confluence with a larger river system. We note that defining the occupied length from the observed data itself may bias among-river comparisons, since a river with only a single downstream sighting would yield a short, high-density occupied stretch. We used one-way ANOVA to assess variations in channel depth and width across the surveyed river stretches. The pairwise post-hoc comparisons were conducted using Tukey’s Honestly Significant Difference (HSD). Results of pairwise comparisons are presented as compact letter displays (CLD), in which rivers sharing at least one letter do not differ significantly (α = 0.05) and rivers with no shared letters differ significantly. All statistical analyses were conducted in R version 4.4.1 (R Core Team, 2025) using RStudio version 2025.05.0 + 496 (Posit team, 2025).

Population recovery analysis

To quantify the long-term population trend of the gharial, we analyzed population data spanning 45 years (1979–2024). Due to inconsistent historical survey efforts in tributaries, data from the Chambal River, the species’ primary stronghold, was utilized as a robust surrogate for the overall basin-wide recovery trend. Population growth rate was estimated using an exponential growth model, based on three census data points spanning 1979-2024. We calculated the exponential rate of increase (r) using the formula:

where N t is the population size at time t, N 0 is the initial population size, and t is the number of years. The per annum growth rate (%) was derived from the exponential rate (Caughley, 1977). To evaluate shifts in population dynamics, growth rates were compared across two distinct conservation phases: the initial intensive grow and release phase (1979–1993) and the subsequent modern phase (1993–2024), alongside the overall recovery period (1979–2024).

Habitat suitability and priority stretch

Selection of variables

We considered a total of 19 explanatory variables, encompassing bioclimatic, hydrological, water quality, and anthropogenic stressors (Supplementary Table 2). The variables included channel width (CW), channel depth (CD), slope (SL), water discharge (WDIS), annual mean temperature (Bio1), isothermality (Bio3), temperature seasonality (Bio4), precipitation of wettest month (Bio13), precipitation seasonality (Bio15), precipitation of driest quarter (Bio17), precipitation of warmest quarter (Bio18), dissolved oxygen (DO), total dissolved solids (TDS), pH, sand mining (SM), water extraction (WE), human presence (HP), fishing intensity (FI), and ferry intensity (FER). Together, these predictors capture both natural environmental gradients and human-mediated disturbance processes shaping gharial occurrence at the basin scale. The selection of explanatory variables was guided by the species ecology and habitat requirements, which are strongly influenced by hydrological stability, thermal regimes, and human disturbance (Hussain, 2009; Katdare et al., 2011; Panda et al., 2023; Sánchez-Mercado and Ferrer-Paris, 2010; Sharma et al., 2025a; Vashistha et al., 2021). Bioclimatic variables were included to capture broad-scale climatic constraints affecting river discharge, basking behaviour, and nesting success, particularly temperature variability and precipitation patterns that influence sandbank formation and seasonal habitat availability. Hydrological variables were incorporated to reflect river width and depth, which are critical for foraging, movement, and refuge use by gharials (Das et al., 2024; Hussain, 2009; Sharma et al., 2025a). Anthropogenic variables, including proximity to human settlements, fishing pressure, sand mining, and infrastructure, were included to account for disturbance, habitat degradation, and direct mortality risks that have been identified as major drivers of gharial population decline across its range (Katdare et al., 2011; Nair and Katdare, 2013; Panda et al., 2023; Vashistha et al., 2021). Together, these variables capture both natural and human-mediated processes shaping gharial distribution at the basin scale. Multicollinearity among the 19 candidate predictors was tested using pairwise Pearson correlation analysis, and variables showing strong correlation (|r| ≥ 0.7) (Dormann et al., 2013) were excluded to avoid model overfitting. This screening retained 15 uncorrelated predictors: channel width (CW), channel depth (CD), slope (SL), annual mean temperature (Bio1), isothermality (Bio3), temperature seasonality (Bio4), precipitation of wettest month (Bio13), precipitation seasonality (Bio15), precipitation of driest quarter (Bio17), precipitation of warmest quarter (Bio18), dissolved oxygen (DO), total dissolved solids (TDS), pH, sand mining (SM), and water extraction (WE) (Supplementary Table 2).

We collected a total of 1049 gharial presence points across the basin, which were subsequently reduced to 556 presence points by retaining a single occurrence point within each 5 × 5 km grid cell using the spThin package in R. Spatial thinning was applied to minimize spatial autocorrelation and improve the reliability of habitat suitability predictions. In the absence of empirical estimates of gharial home range or movement distances, this limits the ability to define spatial independence based on species-specific movement ecology. Consequently, the choice of spatial resolution for explanatory variables and occurrence thinning was guided by methodological considerations. A 5 × 5 km grid size was selected as a compromise between overly coarse spatial resolutions (≥10 km²) that may obscure fine-scale habitat associations and finer resolutions (~1 km 2) that risk inflating model performance through spatial autocorrelation and overfitting in linear riverine systems.

Habitat suitability

We employed Maximum Entropy models (MaxEnt, version 3.4.4) to predict the potential distribution of the gharial using entire length of surveyed rivers in the Ganga Basin using presence-only data and selected environmental variables (Supplementary Table 2; Phillips et al., 2017). MaxEnt is a presence-only species distribution model (SDM) that applies the principle of maximum entropy to estimate the probability distribution of a species across geographic space, given a set of environmental predictors. It is widely recognized for its robust and reliable performance when compared to other approaches, such as Generalized Additive Models (GAMs), Genetic Algorithms for Rule-set Prediction (GARP), and BIOCLIM (Elith et al., 2006; Kaky et al., 2020). MaxEnt offers a distinct advantage over alternative models by achieving higher predictive accuracy when using presence-only data (Elith et al., 2011, 2006). It is suitable for both simple and complex models with diverse predictor variables. However, it is susceptible to biases arising from model overfitting and sample-selection bias in presence-only data (Phillips et al., 2009), which can compromise predictive performance, particularly when using extensive sets of predictor variables (Phillips and Dudık, 2008), and such inferences should be interpreted with caution (Yackulic et al., 2013).

To optimise model performance and reduce overfitting, we implemented a systematic model-tuning procedure by varying two key user-modifiable parameters: feature classes (L = Linear, Q = Quadratic, P = Product, T = Threshold, and H = Hinge) and regularization multipliers, ranging from 0.5 to 4.0, with an increment of 0.5 (Elith et al., 2011; Phillips and Dudík, 2008). The model complexity and overfitting can be controlled by careful selection of these parameters. A total of 21 feature classes single and combinations were evaluated (H, HT, L, LH, LHT, LP, LPT, LQ, LQH, LQHT, LQP, LQPT, LQT, LT, P, PT, Q, QHT, QP, QPT and QT) combined factorially with eight regularization multipliers (0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0), yielding 168 candidate models (Supplementary Table 3). We randomly sampled 10,000 background points within the spatial extent of the Ganga River Basin using the default background-sampling settings. Model evaluation was performed using a random k-fold cross-validation approach (block = “randomkfold” partitioning in ENMeval), in which occurrence records were randomly partitioned into 10 subsets, with each subset used once for testing and the remaining for nine for training. Because occurrence records were spatially thinned prior to model fitting, spatial autocorrelation between training and testing partitions is reduced but not eliminated; reported cross-validation statistics should therefore be interpreted as an upper-bound estimate of predictive performance. Fully spatially independent (river-block) cross-validation is recommended as a refinement for future basin-wide assessments. All other parameters were maintained at their default settings. Model performance and selection were based on Akaike’s Information Criterion corrected for small sample size (AICc), calculated using the ENMeval version 2.5.0.2 implemented in R (Muscarella et al., 2014). Two models met the ΔAICc ≤ 2 threshold (feature classes QT and PT, both with regularization multiplier = 1.5; ΔAICc = 0 and 0.70, respectively. Final habitat suitability was generated across the entire river length using an equal-weighted average of the predicted suitability from these two models. Variables importance was assessed using jackknife tests (Phillips et al., 2017). This test was used to evaluate the gain provided by each variable individually and, when omitted, aided in identifying the most influential predictor variable of gharial habitat suitability (Shcheglovitova and Anderson, 2013). The final habitat suitability map was classified into four categories based on predicted suitability score: high suitability (>0.70), moderate suitability (0.60–0.70), low suitability (0.50–0.60), and unsuitable/very low suitability (<0.50) (Das et al., 2022).

Additionally, surface air temperature and precipitation time-series data were obtained from CHELSA (Climatologies at high resolution for the earth’s land surface areas) for 1980 to 2024. These are downscaled model outputs having a resolution of 30 arc sec (~1 km) (Karger et al., 2017). The monthly temperature and precipitation datasets have been derived from the derivatives by ERA-Interim (developed at the European Centre for Medium-Range Weather Forecasts, ECMWF). Temperature and precipitation change for each river stretch were calculated as the difference between the mean of the most recent decade (2015–2024) and the mean of the earliest decade (1980–1989) of the CHELSA time series, expressed in °C and mm respectively. The suitable habitat areas of Gharial were overlaid on the changed temperature and precipitation to identify the impact of changes in temperature and precipitation on the Gharial habitats.

Population status

The study confirmed the presence of gharials in 13 of the 22 surveyed rivers (Figure 2A), with a basin-wide total of 3,037 gharials, representing the sum of river-specific minimum counts obtained during standardised, non-synchronous boat surveys conducted across 22 rivers between 2020 and 2023 (Table 1). Population size was highly unevenly distributed across rivers, with the Chambal River supporting the largest population (MPS = 2,097; 69.1%), followed by the Ghaghra River (MPS = 463; 15.2%) and the Girwa River (MPS = 158; 5.2%). In contrast, the lowest populations were recorded in the Kosi River (MPS = 1), the Bagmati River (MPS = 2), and the Babai River (MPS = 7). Encounter rate also varied substantially among the rivers (Figure 2B, and Table 1). Mean encounter rate (individual sighted per km) was highest in the Girwa River (7.90 ± 4.29) (mean ± SEM), followed by the Chambal River (5.18 ± 0.50) and Kauriyala River (1.80 ± 1.47). The lowest encounter rate was observed in the Kosi (0.02 ± 0.02) and Ganga (0.02 ± 0.007) rivers (Figure 2B, Table 1).

Size-class composition differed among rivers, with individuals representing all classes recorded in the Chambal, Gandak, and Ghaghra rivers (Figure 2C). The proportion of adult individuals (>3.0 m) was highest in the Chambal River (65.5%), followed by the Ramganga River (60.4%) and the Bagmati River (50.0%). Mid-sized individuals (subadults: 1.8–3.0 m) dominated in 4 of the 13 gharial-occupied rivers, with the highest proportions observed in the Kauriyala River (74.1%), followed by the Ganga (62.8%), Son (44.1%), and Ghaghra (40.8%) rivers. In contrast, smaller size classes (juvenile: 90–180 cm) constituted the majority of the population in the Babai (57.1%) and Sharda (44.4%) rivers. In the Yamuna and Gandak rivers, yearlings (0.6–0.9 m) accounted for the highest proportion. We recorded a total of 107 adult males (identified based on the presence of ghara) across the rivers surveyed. The highest number of males was recorded in the Chambal River (n=80), followed by the Ramganga (n=14), the Gandak (n=6), the Girwa (n=4), and the Ghaghra (n=3). No adult males were recorded in the other rivers.

River habitat condition

The average river depth across the surveyed rivers was 3.4 ± 2.9 m (mean ± standard deviation), with the highest mean depth observed in the Ganga River (5.7 ± 4.01 m) and the lowest in the Ajay River (0.92 ± 0.83 m) (Supplementary Figure 2). The average channel width was 339.9 ± 314.1 m, with the widest channel recorded in the Rupnarayan River (1149.9 ± 985.7 m) and the narrowest in the Sind River (77.7 ± 16.0 m). One-way ANOVA revealed a significant variation in depth (F 21, 1436 = 29.5, p < 0.001) and width (F 21, 1436 = 39.9, p < 0.001) across the surveyed rivers. Post-hoc Tukey HSD comparisons indicated that Ganga had significantly greater depth than the majority of other rivers, including Son, Ajay, Damodar, Kosi, and Babai. For width comparison, Rupnarayan width differed significantly from all other rivers except Sharda. The rivers with narrow width such as Sind and Gomti differed significantly from Rupnarayan, Sharda, Kauriyala, Ganga, Yamuna, and Girwa. Among all recorded pressures, fishing intensity and human presence were the most frequently observed threats (Supplementary Figure 3). Human presence was the only pressure detected in the Kauriyala River. Water extraction intensity was highest in the Betwa (93%) and Bagmati (91%) rivers and was negligible or absent in the Babai, Girwa, Kauriyala, and Sharda rivers. Sand mining was most prevalent in the Kosi (100%) and Betwa (87%) rivers, while no mining was recorded in the Girwa, Kauriyala, and Rapti rivers. Ferry activity was generally low, peaking in the Bagmati River (92%), and was absent in the Kauriyala River.

Population trend

The long-term population data from the Chambal River surrogate demonstrate a sustained recovery of the gharial since the initiation of targeted conservation programs.This analysis is specific to the Chambal River, used as a surrogate for the species’ primary stronghold, and is not extrapolated to other rivers in the Ganga Basin (see Section 3.1 for basin-wide minimum counts). Over the 45 years from 1979 to 2024, the population exhibited an overall exponential rate of increase (r) of 0.068, translating to a 7.0% per annum growth rate. The trend indicates a rapid population expansion during the initial 14 years (1979–1993), achieving an annual growth rate of 15.6%. In the subsequent three decades (1993–2024), the rate of recovery stabilized, dropping to 4.3% per annum (Figure 3).

Habitat suitability and priority stretch

A total of 168 candidate models were evaluated using multiple combinations of feature classes and regularization multipliers (Supplementary Table 3). The Area Under the Curve (AUC) for the top model was 0.95 ± 0.17 (QT-1.5; mean ± standard deviation), and for the second model, 0.97 ± 0.16 (PT-1.5; Supplementary Table 3). The jackknife test of variable importance revealed strong agreement between both models, with hydrological variables dominating model performance (Supplementary Figures 1A, B). For the QT model, the highest contribution was from water extraction (WE = 22%), followed by channel witdh (CW = 18%), channel depth (CD = 16%), and precipitation of the warmest quarter (Bio18 = 13%). Moderate influence was observed for human presence (HP = 9%) and dissolved oxygen (DO = 7%). In contrast, slope (SL = 4%), sand mining (SM = 3%), and precipitation of the driest quarter (Bio17 = 3%) showed minimal explanatory power. Similarly, in the PT model, water extraction (WE = 24%) remained the most influential predictor, followed by channel width (CW = 19%), channel depth (CD = 15%), and precipitation of the warmest quarter (Bio18 = 12%). Secondary contributions were from human presence (HP = 8%) and dissolved oxygen (DO = 6%). Variables such as slope (SL = 4%), sand mining (SM = 4%), and precipitation of driest quarter (Bio17 = 3%) contributed negligibly to predictive performance. Overall, hydrological and flow-related parameters explained nearly 65–70% of model gain, whereas geomorphological and disturbance variables accounted for only a minor proportion of habitat suitability prediction. These percentages represent MaxEnt’s internal ‘percent contribution’ heuristic, distinct from the jackknife test of gain shown in Supplementary Figures 1A, B and from permutation importance, which was not computed in this study.

Of the total 22 rivers, 10,716 km (84.1%) of the river stretches very low (Figure 2D, Supplementary Table 5). The largest extent of least suitable habitat occurred in the Ganga River (2325 km; 92.1% of its length), followed by the Yamuna (1220 km; 85.0%), Gomti (810 km; 87.1%), and Son (755 km; 96.2%). While the most suitable habitat was most extensive in the Chambal River (425 km; 44.0%), followed by the Ghaghra (205 km; 33.3%), Gandak (150 km; 50.8%), Yamuna (150 km; 10.5%), and Sharda (120 km; 26.4%). Among smaller rivers (<100 km in length), 100% of the Girwa River, 64.7% of the Babai River, and 50% of the Kauriyala River were classified as highly suitable for gharial. In contrast, rivers such as Ajay, Bagmati, Betwa, Damodar, Kosi, Mahananda, and Rupnarayan exhibited suitability only in the <0.50 category.

Overlay of climatic change on suitable habitats indicated that 1666 BEUs (8330 km) of river stretches encounter temperature change, corresponding to 24.3% of the total suitable habitat across the basin, whereas precipitation change was recorded in 1418 BEUs (7090 km), representing 28.6% of the suitable stretches (Supplementary Table 5). Temperature change was most prominent within suitable habitats of the Chambal, Yamuna, and Gandak rivers, where a large proportion of suitable BEUs coincided with warming trends, while comparatively limited temperature change occurred within suitable stretches of the Ghaghra and Girwa (Supplementary Table 5, Supplementary Figure 4). Similarly, precipitation change overlapped moderately with suitable habitats in the Chambal and Yamuna but remained minimal within several northern tributaries and smaller rivers (Supplementary Table 5, Supplementary Figure 5). Despite measurable climatic variation across BEUs, the spatial distribution of suitable habitats showed very little correspondence with temperature or precipitation change. Several rivers with large suitable stretches (e.g., Ghaghra and Girwa) occurred within relatively stable climatic zones, while others experiencing substantial climatic change (e.g., lower Ganga and semi-arid tributaries) remained predominantly low predicted relative suitability.

Across the 12,741 km river network, 1,666 BEUs (8,330 km; 65.4% of total river length) exhibited measurable temperature change, and 1,418 BEUs (7,090 km; 55.6% of total river length) exhibited measurable precipitation change (Supplementary Table 5). Of the 2,025 km classified as suitable habitat, 1070 km (52.8%) overlapped with temperature change and 860 km (47%) overlapped with precipitation change.

Gharial recovery in the Ganga River Basin has relied on population augmentation and counts obtained from sporadic surveys and geographically restricted surveys conducted at different time periods (Lang et al., 2019). Such sporadic monitoring has limited our ability to accurately assess population status, identify priority habitats, and implement effective, evidence-based conservation strategies for gharial across the Basin. In this study, we assessed the distribution, minimum population size (MPS), and size class structure of gharials across 22 rivers in the GB, through parallel surveys conducted between 2020 and 2023, alongside species’ recovery based on the long-term population trends over the last four decades. The presence of the largest gharial population in the Chambal River is not surprising, as over 600 km of the river segment is protected as the National Chambal Sanctuary and benefits from continuous conservation interventions, including supplementation under the grow and release program (Hussain, 2009, 1999; Rao, 1992; Sharma et al., 2021; Singh, 1976). Interestingly, we also recorded a high number of gharials in the Ghaghra River, despite no evidence of breeding (Vashistha et al., 2021). This suggests a possible sink population, potentially sustained by the dispersal of individuals from the Girwa population located upstream, which is continuously augmented through release of young gharials. This hypothesis, however, remains to be tested through mark-recapture or genetic studies (Vashistha et al., 2023, 2021). During the present survey, we also recorded a lower number of gharials in the Gandak River than previously reported (Panda et al., 2023), possibly due to the release of water from the upstream barrage observed during the survey. Such releases can increase flow levels, submerge basking sites, and displace individuals, thereby reducing sightings. So far, the Gandak River was considered to have the second-largest gharial population outside PAs (Panda et al., 2023). However, this study indicates that the Ghaghra River (MPS = 463) holds the second-largest gharial population outside PAs. Furthermore, we recorded over 1,500 adult gharials in the Ganga Basin, substantially exceeding the global estimate of 300–900 mature individuals reported by the IUCN Red List assessment (Lang et al., 2019). Because the IUCN estimate was also derived from comparable total counts rather than a statistical abundance model, we consider our estimate methodologically consistent and therefore comparable, highlighting a potentially substantial underestimation of the current gharial population.

The presence of all size classes in the Chambal, Gandak, and Ghaghra, with an overall higher contribution of large-sized individuals (>1.8 m; subadults and adults) in the population, is broadly consistent with patterns observed previously in gharial populations but does not confirm, a relatively stable population (Sharma et al., 2025b). This pattern is potentially due to the presence of a breeding population, which is well documented in the Chambal and Gandak rivers (Panda et al., 2023; Sharma et al., 2025b). The size-class structure observed in the Ghaghra may be associated with immigration from upstream (Girwa) breeding populations rather than local recruitment (Vashistha et al., 2023, 2021). In contrast, the dominance of smaller size classes (juvenile; 0.9–1.8 m) in the Babai and Girwa rivers, as well as the high proportion of yearlings (0.6–0.9 m) in the Yamuna and Gandak rivers, is consistent with, but not confirmatory of, recent recruitment or the success of grow and release programs in these rivers. Our single-survey, visually estimated size-class data cannot distinguish natural recruitment from continued releases, nor establish demographic stability, given known differences in detectability among size classes and the release histories summarised in Supplementary Table 4 (Choudhury and Chowdhury, 1986; Stevenson, 2015).

Adult males contribute less than 1% of the total gharial population in the basin and were overwhelmingly concentrated in the Chambal River, reinforcing its status as a self-sustaining population. This is consistent with previous studies that have identified the Chambal as the stronghold for the species (Sharma et al., 2025a). Although small numbers of adult males were also recorded in the Ramganga, Gandak, Girwa, and Ghaghra rivers, their scarcity is consistent with, though does not confirm, limited reproductive potential in thesesystems (Lang et al., 2019). While earlier studies reported no sightings of adult males in the upstream stretch of the Ghaghra River between Girijapuri and Challari Ghat (Vashistha et al., 2023), we recorded the presence of three adult males. Except for the Ghaghra, all rivers where adult males were recorded are known to support active breeding populations (Panda et al., 2023; Sharma et al., 2025b; Vashistha et al., 2022). This highlights the need for targeted nesting surveys to assess reproductive activity. Furthermore, the rivers lacking adult males may represent either remnant populations or habitats with low relative suitability for long-term persistence due to anthropogenic pressures (Nair and Katdare, 2013). This raises concern for gharial conservation and requires immediate conservation interventions, including habitat restoration, mitigation of anthropogenic stressors, and potential supplementation through translocation from more viable populations for sustaining these small populations in the future.

Despite decades of effort and the release of over 9,000 gharials in the rivers of the Ganga Basin (Supplementary Table 4), populations have failed to recolonize in much of their historical range, with most rivers in the basin supporting fewer than 100 individuals (Table 1). These findings highlight that release programs alone are insufficient without concurrent habitat protection and threat mitigation. The combined efforts of the grow and release program, along with the protection of suitable habitats, are associated with notable population increases in the Chambal and Girwa rivers: across the 11 rivers with both release records (Supplementary Table 4) and current survey data, historical release effort was strongly correlated with current minimum count (Pearson r = 0.85, p < 0.001; Spearman ρ = 0.86, p < 0.001). This correlation cannot, however, separate the relative contributions of survival of released individuals, natural recruitment, immigration, and detection. In contrast, where these interventions have been inconsistent or absent, population recovery has been limited. For instance, the Ken River, despite having a 16 km stretch within the Ken Gharial Sanctuary, no longer supports a gharial population, with the last recorded release occurring in 2007 (Nair and Katdare, 2013). While the Ghaghra River holds the second-largest gharial population and the presence of adult males, there is no evidence of breeding (Vashistha et al., 2023). Nevertheless, the presence of individuals across all size classes in the Ghaghra River is more plausibly explained by continued downstream dispersal of released and wild-bred individuals from the upstream Girwa population, together with ongoing releases within the Ghaghra itself, rather than by local in-river breeding, for which direct evidence (e.g., nests or hatchlings) is currently lacking (Basu and Singh, 2010; Vashistha et al., 2023).

Since gharial is a specialist species that requires suitable habitat and optimum prey to thrive on (Hussain, 2009, 1999), efforts are underway to reintroduce gharials in different potential sites, which are cautiously scrutinized for the availability of suitable habitats. Our habitat suitability analysis reveals that only 2025 km (15.9%) of the river stretches fall within the suitable category, and the remaining 10,716 (84.1%) are low suitable for a specialist like the gharial. Rivers with high encounter rates, such as the Chambal, Girwa, and Kauriyala, also contained the majority of highly suitable habitats, underscoring the strong linkage between habitat quality and population persistence. We note that the habitat-suitability model was itself trained using gharial occurrence records from the same rivers and surveys used to derive the minimum population counts; consequently, this correspondence does not constitute independent validation of the model, but rather reflects the expected internal consistency between the training data and model output. Other rivers, such as the Ghaghra, Gandak, Yamuna, and Sharda, are also potential sites for gharial recovery. Effective conservation outcomes in these systems will depend on habitat restoration and mitigation of anthropogenic pressures. The rivers, such as Ramganga and Son, have shown evidence of breeding activities within the past five years (Vashistha et al., 2022), but a majority of the river stretches are categorized as classified as having low predicted relative suitability, with small stretches as suitable. These rivers could support gharial populations if habitat is restored and anthropogenic pressures are mitigated (Nair and Katdare, 2013; Vashistha et al., 2022). Several rivers, such as the Ajay, Bagmati, Betwa, Damodar, Kosi, Mahananda, and Rupnarayan, exhibited low habitat suitability and currently lack gharial populations. Hence, these rivers require extensive restoration efforts to enhance their ecological potential before reintroduction can be considered. The habitat-suitability model is intended as a first-pass screening tool to prioritise sites for field verification, and its thresholds follow precedent from comparable riverine species-distribution studies; final reintroduction or supplementation decisions must additionally incorporate prey availability, nesting-bank condition, flow regulation, bycatch risk, social feasibility, genetics, and past release performance, rather than resting on the MaxEnt threshold alone. While examining further the relationship between suitability and climatic factors such as precipitation and temperature change, temperature change overlapped mainly with suitable habitats in the Chambal, Yamuna, and Gandak rivers. In contrast, comparatively smaller portions of suitable habitat in the Ghaghra and Girwa rivers coincided with temperature change (Supplementary Table 5). Similarly, precipitation change overlapped moderately with suitable habitats in the Chambal and Yamuna systems but showed limited overlap across several northern tributaries (Supplementary Figures S4, S5). However, suitable habitat stretches also occurred in river sections showing minimal climatic change, and several river segments experiencing substantial temperature or precipitation change remained very low suitable. Overall, no consistent spatial association was observed between the distribution of suitable habitats and areas experiencing temperature or precipitation change (Supplementary Figures S4, S5).

Gharials, like other crocodilians, exhibit temperature-dependent sex determination. The warming trends identified in our climatic overlay (Section 3.4), even though not spatially concentrated within currently suitable habitat, could plausibly skew hatchling sex ratios at nesting sites over time, with potential long-term demographic consequences for small, recovering populations. This study did not directly measure nest-site incubation temperatures or hatchling sex ratios and therefore cannot test this hypothesis directly. We recommend that future monitoring at active nesting sites (e.g., Chambal, Girwa) incorporate nest-temperature logging and periodic hatchling sex-ratio assessment, so that any climate-driven skew can be detected early and, where feasible, managed through nest-site shading, relocation, or artificial incubation.

The findings of this study align with previous studies emphasizing the importance of habitat quality and reduced anthropogenic pressures for gharial conservation (Hussain, 1999; Nair et al., 2012; Panda et al., 2023; Vashistha et al., 2023). The persistence of large unsuitable habitats across the basin underscores the need for integrated conservation strategies that address habitat restoration, pollution control, control of unregulated sand mining, and sustainable fishing practices. Although the full demographic consequences of ongoing riverscape modification may take decades to manifest, such impacts are likely inevitable given the scale and intensity of current pressures. Despite the widespread presence of multiple anthropogenic stressors across the surveyed rivers, variables such as sand mining and fishing intensity did not emerge as top contributing variables in the habitat suitability models, yet their impact cannot be overlooked (Katdare et al., 2011; Panda et al., 2023). This pattern likely reflects the dominance of hydrological predictors such as channel depth, channel width, and precipitation, which capture fundamental, basin-scale processes that determine gharial habitat availability (Hussain, 2009; Sharma et al., 2025a). Importantly, the relatively low contribution of anthropogenic variables should not be interpreted as evidence of limited conservation relevance (Katdare et al., 2011; Panda et al., 2023).

MaxEnt’s percent-contribution metric reflects each variable’s statistical contribution to explaining the spatial pattern of gharial occurrence relative to background points, which is strongly influenced by the degree of spatial variation (contrast) in a predictor across the study extent. Hydrological variables such as channel width and water extraction vary substantially among and within rivers and therefore provide strong discriminatory power for the model. Anthropogenic pressures such as fishing intensity and sand mining, by contrast, are widespread and co-occur across most surveyed BEUs (Section 3.4; Supplementary Figure 3), giving them comparatively low spatial contrast and therefore low statistical importance in a correlative model, even though their ecological and demographic impact (e.g., direct mortality, nest destruction, prey depletion) may be severe wherever they occur. MaxEnt variable importance therefore measures predictive/statistical contribution to the spatial occurrence pattern, not the magnitude of demographic or conservation impact; the two lines of evidence model-based habitat suitability and threat-based conservation concern are complementary rather than contradictory. Sand mining remains one of the most significant threats to gharial through the loss of basking and nesting sites, with cascading effects on recruitment and population demography (Katdare et al., 2011). Similarly, activities such as fishing, though they contributed minimally to the model, are known to increase mortality risks through entanglement and prey depletion (Katdare et al., 2011). Addressing these threats through a combination of habitat protection measures, which involve flow regulation and stricter enforcement against destructive activities like sand mining and fishing, is essential for ensuring the continued persistence of this critically endangered species. Consistent with earlier gharial habitat studies, physical habitat features such as channel depth and width show strong associations with occurrence probability (Dhami et al., 2023). Similar patterns have been observed in other aquatic species, where abiotic predictors often outweigh anthropogenic variables (Friedrichs-Manthey et al., 2020).

The population growth rate of the largest gharial population, in the Chambal River, increased at 7% per annum over the last four decades, underscoring the success of conservation interventions within this stronghold. This trajectory reflects a heavily protected, intensively supplemented population and should not be read as representative of the basin as a whole, which our river-wise MPS data (Table 1) show to be highly uneven, with most rivers supporting fewer than 100 individuals. However, the marked deceleration in the annual growth rate from 15.6% (pre-1993) to 4.3% (post-1993) suggests that the initial ‘quick recovery’ phase is potentially attributed to aggressive captive-rearing and release programs, and has transitioned into a stabilization phase. This stabilisation is one plausible explanation, among others (e.g., density-dependent effects, changes in release intensity over time, or escalating anthropogenic pressures outside protected limits), for the observed deceleration in growth rate, and would require independent demographic assessment to confirm (Caughley, 1977). These findings reiterate that timely interventions and legal protection under international and national laws can bring a species back from the verge of extinction. However, such efforts require continuous augmentation and monitoring, most importantly managing anthropogenic pressures that often affect specialist species.

While this study provides the most extensive assessment to date of gharial distribution, minimum population size, size-class structure, habitat suitability, and long-term population trend across the Ganga River Basin, certain limitations remain. This study has several limitations. First, detection probability was not explicitly modelled: population counts and body-length estimates were derived from single daylight boat-based surveys subject to imperfect detection from weather, time of day, water level, behaviour, and observer experience, so reported values represent minimum population sizes (MPS) rather than true abundance, and the absence of repeated surveys within the same season precluded formal detection-probability estimation. Second, habitat-suitability modelling using MaxEnt relies on presence-only data, which limits inference about true habitat absence and may inflate suitability in under-sampled rather than genuinely very low suitable areas; independent validation data were unavailable at the basin scale, so model evaluation relied on internal performance metrics that should be interpreted cautiously when informing management decisions. Third, established abundance-estimation approaches such as mark-recapture or unmarked models were not feasible given the wide-ranging, sparse distribution of gharials across 22 rivers and the long-term logistical and financial commitment repeated standardised surveys would require; we therefore prioritised spatial coverage and comparability across rivers over intensive estimation at a limited number of sites, and future monitoring would benefit from repeated surveys (Ahizi et al., 2021) and unmanned aerial vehicles to improve detection and validation (Jewitt et al., 2024; Thapa et al., 2018). Finally, the population-trend analysis is based on three discrete census counts (1979, 1993, 2024) from the Chambal stronghold; this endpoint-based approach cannot capture within-period fluctuations, and the Chambal-specific and basin-wide components of this study should be read as distinct.

This study provides one of the most comprehensive basin-wide assessments of gharial recovery, distribution, and minimum population size, population structure, and habitat suitability across the Ganga River Basin to date. The success of grow and release program and habitat protection is reflected in the initial 15.6% annual growth rate recorded in the Chambal stronghold. However, modern pressures require a shift toward aggressive habitat restoration to maintain this recovery momentum. By integrating synchronized field surveys across 22 rivers with habitat suitability modeling, we demonstrate that gharial persistence is highly uneven across the basin, and the Chambal River remains the stronghold, with the Ghaghra being the second largest gharial population. The recovery of the largest gharial population, in the Chambal River, reflects the combined success of the grow and release and habitat protection. The recovery trajectory should not be generalised to the Ganga Basin as a whole. However, many other rivers remain vulnerable, underscoring the need for urgent, integrated conservation strategies. Gharial conservation efforts should focus on safeguarding and strengthening existing strongholds, particularly the Chambal and Girwa rivers, through strict protection of nesting habitats, regulation of sand mining, and maintenance of ecologically appropriate flow regimes. Rivers identified as having high or moderate habitat suitability, such as the Ghaghra, Gandak, Yamuna, and Sharda, should be prioritized for targeted recovery interventions, including habitat restoration, reduction of anthropogenic pressures, and strategic supplementation under grow and release where appropriate. Rivers currently classified as low predicted relative suitability should be treated as lower near-term priorities for reintroduction pending field validation, rather than being categorically excluded from future consideration. Overall, this study emphasizes the importance of long-term data, and basin-scale monitoring frameworks that move beyond sporadic, site-specific assessments. Strategic prioritization of rivers and river stretches, informed by both population data and habitat suitability, will be critical for maximizing conservation outcomes. Sustained protection of high-quality habitats, coupled with targeted management interventions in rivers with recovery potential, offers the most realistic pathway for ensuring the long-term persistence of gharial and the ecological integrity of the Ganga River Basin.