Impact of land use change on wetland water quality in the Keibul Lamjao Conservation Area (a proposed mixed UNESCO World Heritage site) from northeast India

Wetlands are among the most productive and threatened ecosystems in the world [1]. The ecosystems have been crucial to the development and survival of humanity. However, their use for short-term gain has driven its destruction and degradation [2]. Today, it is estimated that a quarter of all wetland species faces the uncertainty of extinction. Further, the lost in wetland ecosystem services is accounted to be US2ドル.7 trillion [3]. Between the span of 1970 and 2015, 35% of wetlands across the world were converted into other land use types [4]. Over the years, numerous studies that use remote sensing and GIS have identified the temporal land use / land cover (LULC) changes [5,6,7,8,9,10,11,12] and water quality degradation [13,14,15,16] of wetlands.

In India, 40% of wetlands in the country have seen a progressive decline in their quality to support and sustain aquatic life [17]. India has 99 Ramsar sites covering an area of 13,60,805 ha [18]. However, the Ramsar sites are burdened with encroachment for settlement, agriculture and pisciculture, pollution, invasive species infestation, climate change, unsustainable exploitation of resources and changes in water flow regime etc [19]. In Manipur, a state in the northeast India, the major share of wetlands is distributed in the central Imphal valley with a total of 153 wetlands covering an area of 528.59 km². For ages, the Loktak Wetland Complex (LWC), which includes Loktak (Ramsar site), Pumlen/Khoidum and Ikop/Kharung, have deep-rooted socio-cultural, economic and ecological role in the region [20–21]. The wetlands are covered with a unique ecosystem known as "phoomdi", which are floating mats of dense vegetation [22]. The phoomdi of Keibul Lamjao National Park (KLNP) serve as an ideal habitat for the endangered deer "Rucervus eldii eldii McClelland" with 260 + individuals in the world [23]. The LWC harbors waterfowl migrating on the Central Asia and East Asia / Australasia migratory flight pathway, with a total of 21 species reported [24,25,26]. Moreover, the locals have extracted wetland plants and other biotic resources for generations [24].

Therefore, the region which include Pumlen (43 km²) and its adjoining areas of Loktak (140 km²), including the only floating National Park in the world, KLNP (40 km²), has been proposed for inclusion as a "Mixed World Heritage Site" of UNESCO World Heritage Convention as the "Keibul Lamjao Conservation Area’ [21]. The inclusion aimed to provide international legal protection, increased awareness and attention, and enhanced management [27]. For example, the UNESCO World Heritage sites such as Mont-Saint-Michel and its Bay (France), Wood Buffalo National Park (Canada), Banc d’Arguin National Park (Mauritania), and Itsukushima Shinto Shrine (Japan), receive international support for protection and have access to needed resources for securing their ecosystem values [28].

However, in the LWC, construction of the Ithai barrage (1983) on the Imphal river for generation of hydropower (Loktak Multi Purpose project with an installed capacity of 105 MW) has converted the wetlands into an artificial reservoir. Today, the wetlands serve as collection units of silt, pollutants and biomass [25]. The wetland’s ecological fate is marked by rampant deforestation in the catchment area, resulting into excessive siltation, heavy encroachment for human settlement, agricultural and pisciculture expansion, infestation with invasive species such as water hyacinth and pollution of the 36 tributaries etc [29]. Similar to the LWC wetlands, several Ramsar sites in India, such as Deepor Beel (Assam), Ashtamudi lake (Kerala), Chilka lake (Odisha), Wular lake (Kashmir), and Rudrasagar lake (Tripura) have undergone rampant unsustainable growth leading to changes in land use pattern and water quality [30,31,32,33,34].

For the region, the major share of wetland research available is focussed on the Ramsar site in the Montreux records, i.e. Loktak [22,23,24,25, 29, 35,36,37,38,39,40,41,42,43,44,45,46]. The scientific investigations carried out in the Ramsar site focussed on LULC analysis [41–42]; watershed assessment [39]; water and sediment quality assessment [8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36, 43]; nutrients and metal bioaccumulation in phoomdi [35, 38]; ethnobotanical studies [44]; microplastics in fish [45]; limnological studies [44]; and livelihood assessment of the locals in and around the wetland [46] etc. On the other hand, there is a gap in scientific research for the remaining 152 wetlands of the Imphal valley. Today, the ecosystem services provided by other important Imphal valley wetlands such as Pumlen/Khoidum and Ikop/Kharung, and their current ecological status have been underreported and neglected. There only exist few reports on water quality assessment of the wetlands [47–48].

Hence, a detailed study of Pumlen/Khoidum and Ikop/Kharung wetlands, viz., encroachment status and impact of land use change on water quality were carried out. The present study was taken up with the aim (i) to use Remote Sensing and Geographic Information System (GIS) data for mapping the Land Use / Land Cover change (LULC) of the wetlands between 1988 and 2023 to understand the changing land use pattern; and (ii) to analyse water physico-chemical characteristics during 2022-23, influenced by land use change and calculate the Water Quality Index (WQI) for drinking and assess the irrigation water quality using different indices. The outcome of the study is expected to benefit the policymakers in managing the wetlands. Further, it will help in their cause to push for the inclusion of the region as a Mixed UNESCO World Heritage Site.

2.1 Study area

The Imphal valley (24.807° N and 93.938° E) is an intermontane valley located in the central portion of the Manipur state in the northeast India (Fig. 1). It covers an area of 2238 km² [49]. A number of small rivers, such as Imphal, Iril, Nambul, Kongba, Thoubal, Wangjing, Nambol, Khuga and Sekmai etc, originating from the surrounding hills traversed the valley, forming numerous wetlands in the low-lying regions. Among them, Pumlen/Khoidum (93o50’ to 94o0’ E and 24o20’ to 24o35’ N) in the Thoubal district is the second-largest freshwater wetland in the state. It has an area of 77.41 km². The wetland serves as a secondary reservoir of Loktak National Hydel Project maintained at 768.5 msl. It receives water from precipitation, surface run-off from the hill range on the southern edge, from surrounding agricultural fields and indirectly from Sekmai and Imphal rivers, respectively. The region receives an annual rainfall between 1500–1700 mm. The temperature range from 2o to 21o C minimum and 23o to 36o C maximum [50]. The climate of the region is sub-tropical monsoon type. Today, the wetland’s 2/3rd area are covered with phoomdi (Fig. 2a). For generations, the wetland has been a major source of income for the locals of 37 villages situated in the periphery. Ikop/Kharung (93o56’ to 93o94’ E and 24o35’ to 24o59’ N) is the third largest freshwater wetland in the valley. It has an area of 56.22 km². The wetland in the north-eastern boundary of Pumlen/Khoidum is fed by Arong and Heirok/Wangjing rivers. Similar to Pumlen/Khoidum, 2/3rd of the wetland is covered with phoomdi (Fig. 2b).

2.2 Land use / land cover change analysis

2.2.1 Data sources

In order to assess the encroachment status of Pumlen/Khoidum and Ikop/Kharung, 30 m spatial resolution, Tier-1, Level-2 Analysis Ready satellite imagery (Science Products Data) of Landsat 5TM dated 05-02-1988 (ID: LT05_L2SP_135043_19880205_20200917_02_T1) and Landsat 8 dated 05-02-2023 (ID: LC08_L2SP_135043_20230205_20230209_02_T1) were obtained from the United States Geological Survey (USGS) Earth Explorer website (http://earthexplorer.usgs.gov/) in Geo-TIFF data format. For the study, an image processing system called ERDAS Imagine 9 was used. Since the imagery were Level-2 analysis ready surface reflectance data, no further processing was required as detailed by the Landsat data document. 1988 Landsat 5TM data was used as base year data since there was no other data available before this date that was comparable in resolution to the later Landsat 8 data. 2023 Landsat 8 data was used as this was the latest/recent data available during the study.

2.2.2 Geospatial analysis

For the LULC change estimation, the Normalized Difference Vegetation Index (NDVI) was calculated. The step is followed by the Maximum Likelihood Supervised Classification using Near Infrared (NIR), Red, and Green bands with the help of classifier tools in ERDAS Imagine 9.

Initially, NDVI values were obtained using NIR (band-4 for Landsat5TM and band-5 for Landsat8) and Red (band-3 for Landsat5TM and band-4 for Landsat8) bands for both the satellite imagery to understand green and non-green cover. Accordingly, from the information collected during field visit and other ancillary data, the land use categories, viz., open water, thin vegetation, thick vegetation, agriculture/pisciculture farm and built-up could be identified. Later, the information was used as a guide for the selection of training pixels for the supervised classification (Fig. 3) [51]. The overall NDVI values of 1988 satellite imagery ranges from − 0.777778 to 0.591529, indicating the presence of diverse land cover features including water bodies and varying vegetation conditions. Similarly, the NDVI values for the study area range from − 0.154286 to 0.383606 for 2023 (Supplementary Table 1).

The Maximum Likelihood Supervised Classification was done using primary field data supported by high resolution Google Earth imagery and NDVI values [52]. For the classification, False Colour Composite (FCC) images of Band 4-3-2 and Band 5-4-3 of Landsat 5TM and Landsat 8 were used for training pixels (signature pixels) collection. Once the desired training pixels were collected, the supervised classification was performed. The output classes were then checked for further improvement and minor correction during the post classification rectification stage. When the results were satisfied, the "Recode" tool was used to combine the output classes belonging to the same classes. Ultimately, the entire LULC classes were reduced to 5 classes as mentioned previously.

2.2.3 Accuracy assessment

The output was assessed for accuracy using Accuracy Assessment tool, along with field data and high resolution Google Earth imagery as references. The step was performed using "random point method", generating 10 random points for each class to calculate the overall accuracy, kappa statistics and error matrix. The Overall Accuracy ranged between 93 and 96% for Pumlen/Khoidum and 94 to 96% for Ikop/Kharung, respectively. The accuracy range falls in the acceptable accuracy or strong agreement of LULC classification studies [53–54]. The accuracy analysis outputs (including error matrix) for both wetlands are provided as supplementary data (Supplementary Tables 2 and 3).

After the accuracy assessment, the LULC thematic layers obtained above were used to generate the temporal change thematic layer using raster union-overlay method. The change thematic layer is then used to quantify the changes and to prepare the temporal change maps showing the direction of changes within the LULC classes. These maps clearly represent the temporal changes of LULC classes from 1988 to 2023.

2.3 Surface water sampling and physico-chemical analysis

For the analysis of physico-chemical characteristics, samples were collected from 17 sampling locations (N = 51, with samples collected in triplicates from each site) of Pumlen/Khoidum (10 sampling locations) and Ikop/Kharung (7 sampling locations) (Table 1; Fig. 1), respectively. The PoM sampling was carried out during the October-November months of 2022 and the PM sampling during the March-April months of 2023, respectively.

The water samples were collected in 1 L polyethylene bottles which are already pre-cleaned with HNO₃ and rinsed with distilled water in the laboratory. During sampling, water was collected from a depth of 0.5 m below the surface while avoiding the formation of air bubbles. The sampling, preservation and analysis were carried out in accordance to the standard procedures given by the American Public Health Association [55].

Some of the physico-chemical parameters, viz., pH, electrical conductivity (EC), total dissolved solids (TDS) and dissolved oxygen (DO) were analyzed in situ using a water analyser kit (Systronics Water Analyser 371, India) for both seasons. The instrument for pH, EC and TDS analysis was calibrated to ensure the results’ reliability using the relevant standard solutions of HIMEDIA and EUTECH brands. The Nephelometric Turbidity Meter of Systronics Digital Nephelo Turbidity Meter 132, India was used for turbidity (T) analysis in laboratory. The instrument is calibrated using Formazin solutions. Alkalinity (Alk), chloride (Cl⁻), total hardness (TH) and calcium hardness (Ca²⁺) were estimated using titration methods [55]. Systronics Flame Photometer 128, India was used for the quantification of sodium (Na⁺) and potassium (K⁺). The instruments were calibrated using standards of the respective ions. The nutrients, viz., nitrate (NO₃⁻) and phosphate (PO₄^{3 −}) concentrations were estimated using a UV-Vis Spectrophotometer (Systronics UV-Vis Spectrophotometer 117, India) (Supplementary Table 4).

2.3.1 Evaluation of water quality index (WQI)

The WQI for drinking for both wetlands was estimated using the information available in the field [56,57,58]. The World Health Organisation [59] and the Bureau of Indian Standards [60] drinking water quality standards of the physico-chemical parameters were used for WQI calculations. In total, 14 physico-chemical parameters, viz., pH, EC, TDS, T, DO, Alkalinity, TH, Cl⁻, Ca²⁺, Mg²⁺, Na⁺, K⁺, NO₃⁻ and PO₄^{3 −}, were used. During the assessment, the 14 parameters were assigned a different weight (W_i) within the range of 1 to 5, based on the influence of the parameter on the overall water quality, with 1 as the least significant and 5 with the most vital influence, respectively [56] (Supplementary Table 5).

The relative weight (W_r) is given by

where, i = total number of parameters considered.

The quality rating scale (Q_i) for the parameters is calculated using

where, Ci = concentrations of the water parameters, S_i = drinking water standards as per WHO and BIS.

The sub-index (SI_i) of the parameters was calculated and then the aggregate was determined to give the WQI values.

Afterwards, the WQI values obtained were then classified using the categorization proposed by Yadav et al. [56].

2.3.2 Evaluation of nutrient pollution index (NPI)

The nitrogen and phosphorus enrichment status of the wetlands is presented by NPI values calculated using the equation [61]:

where, N and P are NO₃₋ and PO₄³⁻ concentrations for both wetlands; MAC_n and MAC_p are the limits of NO₃₋ and PO₄³⁻ according to WHO [59].

2.3.3 Evaluation of irrigation water quality

The irrigation water quality was assessed using the parameters such as pH, EC, TDS, Sodium Percentage (Na%), Sodium Adsorption Ratio (SAR), Magnesium Hazard (MH) and Kelly’s ratio (KR), respectively.

The Na% in water represents alkali hazard to crops in irrigation water [62].

The SAR in water represents sodium hazard to crops in irrigation water [63].

The MH index is expressed by the equilibrium state of calcium and magnesium in irrigation water [64].

The KR index represents alkali hazard due to excess alkali metal ion (Na⁺) against alkaline earth metal ions (Ca²⁺ and Mg²⁺) in irrigation water [65].

Further, Wilcox [62] and US Salinity Laboratory Staff [66] plots were used to assess the irrigation water quality. The plots between %Na and EC is represented in Wilcox diagram, while the USSLS diagram denotes the classification of the irrigation water quality into 16 classes based on plots between alkali hazards (SAR) and salinity hazards (EC).

2.4 Data analysis

The analysis of the experimental results represented as average of analyses ± standard deviation of three replicates. The calculations of the irrigation water quality indices such as Na%, SAR, MH and KR, and Wilcox and USSLS plots were done in Microsoft Excel 2010. SPSS 16.0 for windows was used for correlation analysis.

For Principal Component Analysis (PCA) analysis, prior to multivariate statistical modelling, the dataset was standardized using z-scores to eliminate the influence of differing units and variance ranges among parameters. The suitability of the water quality data for multivariate analysis was verified using the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity as done in Panjila et al. [67]. A KMO value > 0.5 and a significant Bartlett’s test result (p < 0.05) were considered prerequisites for proceeding. Principal Component Analysis was subsequently performed to identify the primary factors regulating the physico-chemical parameters. Principal components with Eigen values > 1 were retained (Kaiser’s criterion), and a Varimax rotation was applied to maximize the variance of the squared loadings, facilitating the interpretation of the latent factors. The statistical analysis was conducted in the R programming environment (v4.3.2) using the psych package [68] for factor extraction and the ggplot2 package [69] for biplot visualization. This multivariate approach is consistent with recent assessments of wetland and surface water quality [67, 70].

3.1 Land use / land cover change

For 2023, agriculture or pisciculture farms (31.79% of Pumlen/Khoidum and 44.84% of Ikop/Kharung), thick vegetation (33.28% of Pumlen/Khoidum and 26.27% of Ikop/Kharung), thin vegetation (26.40% of Pumlen/Khoidum and 16.63% of Ikop/Kharung), open water (3.54% of Pumlen/Khoidum and 11.42% of Ikop/Kharung), and built-up (4.99% of Pumlen/Khoidum and 0.84% of Ikop/Kharung) are the primary land use types. On the contrary, in 1988, Pumlen/Khoidum have majority of the wetland area covered with thin vegetation (43.99%), thick vegetation (42.82%) and open water (13.19%) land use classes. Similarly, thick vegetation (47.31%), thin vegetation (21.18%), agriculture or pisciculture farms (20.46%) and open water (11.05%) land use classes make up Ikop/Kharung in 1988 (Figs. 4 and 5; Table 2).

Between 1988 and 2023, Pumlen/Khoidum and Ikop/Kharung have lost 36.78% and 45.68% of the wetland areas to built-up and agriculture/pisciculture farm expansion. In total, the encroached areas were 28.47 km² for Pumlen/Khoidum and 25.68 km² for Ikop/Kharung, respectively.

For Pumlen/Khoidum, built-up inside the wetland increased to 3.86 km² in 2023. Brick or earthern or bamboo framed houses and various community buildings have started appearing. Moreover, hundreds of temporary farm shelters were seen within the wetland. During the time period, 3.6, 1.39, and 2.06 km² of open water, thick and thin phoomdi, respectively were converted into built-up areas (Supplementary Table 6). Similar to the study, Paonam and Chatterjee [42] in the LULC analysis of Loktak showed an overall increase in built-up area by 18.4% during 2009–2020. Kangabam et al. [71] in their study reported in increased of settlement land use class in Loktak to 4.42% in 2015 from 2.01% in 1977, respectively.

By 2023, a total of 24.61 km² wetland area was converted into agricultural fields or pisciculture farms. In total, 3.14, 8.68, and 12.92 km² of open water, thick and thin phoomdi, respectively were converted into agricultural fields or pisciculture farms. The deeper areas are reclaimed for pisciculture practices, whereas the drier portions are used for paddy farming. As a result, major portions of Lamjao pat, Kakyai pat, Waikhong pat and Laikom pat were converted into pisciculture ponds or agricultural fields. In total, the overall fish production of the wetland is 1652 kg per year [72]. Today, the majority of the harvest comes from artificial man-controlled pisciculture ponds in the wetlands. The common fish species include Ctenopharyngodon idella, Labeo rohita, Oreochromis mosambicus, Cyprinus carpio, Catla catla and Cyprinus carpio etc. For grass carps, the fodders such as Hydrilla verticillata, Ceratophyllum sp. and Echinochloa stagnina are available in the wetland. However, most of the ponds are fed with artificial feeds. In addition, poultry, piggery or cattle sheds are constructed in the wetland by locals. It releases waste directly into the surroundings. Overall, the conversion of shallow wetlands into agricultural fields or pisciculture farms is commonly reported by various researchers [6–7]. Similar conditions were also reported by Meitei et al. [11], where the pisciculture land use class increased by 74.92% during 1988 and 2023 (32.7 hectares to 57.2 hectares) in Heingang wetland in the Imphal urban region. Singh et al. [13] reported a similar pattern in Harike Ramsar site in Punjab, where the rate of increase in agricultural land was 18.87 ha/year during 2006–2014.

The thick vegetation cover or phoomdi areas reduced to 25.76 km² from 33.15 km², a decline of 22.29% and 7.39 km². Similarly, thin vegetation areas have been reduced by 39.97% (from 34.05 km² to 20.44 km²). From the overlay analysis, it is observed that 12.92 km² of thick vegetation and 8.68 km² of thin vegetation areas have been reclaimed for pisciculture or agricultural farms (Supplementary Tables 6 and Fig. 1). The majority of thick vegetation cover in the core area was converted into thin vegetation zones in 2023 (Fig. 4). Today, the maintenance of the water level at 768.5 m above msl throughout the year for the generation of power by Loktak Hydroelectric Project has caused dull and unhealthy phoomdi growth [29, 42]. The reduction of thick vegetation cover in the core area is a problem for the re-introduction of R. eldii eldii as proposed by the concerned authorities in Pumlen/Khoidum [36]. The thick floating vegetation was meant to serve as the ideal habitat of the deer, like the floating Keibul Lamjao National Park. The decreasing vegetation thickness will not support the weight of the deer. Similar observations were reported in Loktak by Kangabam et al. [71], where the thick phumdi coverage in the wetland reduced to 6.33% in 2023 from 9.82% in 1977. On the other hand, the thin phoomdi cover has seen a drastic change in area, with a total reduction of 49.38 km² during 1977 and 2023, respectively.

In 2023, the open water area was 2.74 km², which is a reduction of 73.16% from 1988. During the time period, only 0.77 km² of open water remains unchanged. 3.54% of the wetland area in 2023 is in the form of an open water zone. Most of the open water areas were taken over by uncontrolled thin (2.73 km²) and thick phoomdi (3.30 km²) proliferation in the wetland.

For Ikop/Kharung, the major land use pattern change is in the form of uncontrolled pisciculture or agricultural farms expansion (Fig. 5). It occupies 44.84% of the wetland (11.50 km² in 1988 to 25.21 km² in 2023) (Table 2). During the study period, 2.20 km² of open water area, 4.77 km² of thin vegetation and 11.23 km² of thick vegetation zone were converted into pisciculture or agricultural farms (Supplementary Tables 7 and Fig. 2). However, encroachment for settlement is comparatively less and occupies a mere 0.47 km² (0.84% in 2023). The land use change of the wetland has completely modified the freshwater ecosystem, with only 11.42% of the total area found to be open water zone in 2023.

Similar observations were made by Devi and Singh [73], where wetland and agricultural areas of the Thoubal district are converted into pisciculture farms. The pisciculture farms increased from 4.74% in 1991 to 13.04% in 2021. Moreover, the study concluded that the wetlands, viz., Phulou, Yaithibi and Waithou are almost converted into pisciculture or agricultural farms. Similarly to the study, Singh et al. [13] reported that the area under agriculture, forest and built-up increased in the Harike wetland, Punjab, whereas wasteland, water bodies and wetland have decreased during 2014–18. Over the years, the water spread area was reduced from 4073 ha and 4166 ha in 2002 and 2006 to 3918.98 ha and 3910.56 ha in 2014 and 2018, respectively. Likewise, the East Kolkata Wetlands, another Ramsar site in West Bengal, showed a 26% reduction in the wetland area due to urbanization and human activity during 2006 and 2012 [12].

3.2 Water characteristics

The sampling locations for water quality analysis were the peripheral areas near settlements or pisciculture or agricultural farms in the surroundings (Fig. 1). The wetland peripheral areas are almost entirely encroached (Figs. 4 and 5). Hence, the water quality will be heavily influenced by domestic waste water release, dumping of waste, agricultural fields and pisciculture ponds runoff, and discharge by the tributaries. Several wetland studies, viz., Bangweulu in Zambia [74], Lake Tana Basin in Ethiopia [75], Qionghai Lake in China [76], East Kolkata wetlands in India [12], and Lake Sapanca in Turkey [77] etc. reported the deteriorating water quality due to land use changes.

Tables 3 and 4 showed temporal and spatial variations in the water characteristics for PoM and PM seasons from both Pumlen/Khoidum and Ikop/Kharung, respectively.

Higher pH was observed for PM season. The values ranged from 6.63 ± 0.06 to 7.03 ± 0.036 (PM) in comparison to 6.38 ± 0.14 to 6.83 ± 0.086 (PoM) for Pumlen/Khoidum and 6.72 ± 0.15 to 7.51 ± 0.23 (PM) and 6.6 ± 0.025 to 7.09 ± 0.11 (PoM) for Ikop/Kharung, respectively (Supplementary Tables 8 and 9). pH levels for Ikop/Kharung and most of the locations of Pumlen/Khoidum for both seasons were within the permissible range. It thereby indicates a stable pH. The stable high pH can be attributed to greater photosynthetic performance of submerged aquatic plants, such as algae and phytoplanktons in the wetland bottom [78].

For Pumlen/Khoidum, P4 - Sarik Konjil showed a pH of 6.38 ± 0.14 during PoM season, indicating the slightly acidic nature of peripheral areas. The acidic pH is due to surface runoff from agricultural fields. The fields consume considerable amount of chemical fertilizers, insecticides and pesticides during the paddy cultivation period. The use of chemical fertilizers drastically changes pH, reducing the levels below the acceptable range of 6.5–8.5 [79]. One of the negative consequences of a sudden change in pH will be death of aquatic organisms previously acclimatized in the pisciculture farms [80]. Moreover, slow pH will trigger algal blooms, enhance leaching of toxic metals and trigger the loss of sensitive species [81]. Hence, a stable pH range is beneficial for the environment, wetland biota and livelihood security.

Higher EC was found in PM season. EC ranged from 212.6 ± 9.86 to 560.3 ± 255.6 μScm^{− 1} (PM) and 141.6 ± 3.78 to 393 ± 18.7 μScm^{− 1} (PoM) for Pumlen/Khoidum, respectively. Similarly, EC ranged from 211.33 ± 4.04 to 361.33 ± 6.027 μScm^{− 1} (PM) and 166 ± 21.51 to 238.33 ± 4.5 μScm^{− 1} (PoM) for Ikop/Kharung, respectively. The reason for high EC during PM is due to ions from agricultural runoff and drainage systems, aquaculture practices, runoff with fish feed, excreta and chemicals (CaCO₃) used in controlling ammonification of fish farms, shortage of rainfall and increased evaporation and residual deposits of salts and minerals etc. In aquatic ecosystems, high EC can lead to salinization issues, which is toxic to aquatic life. It inhibits the activities of soil microorganisms or cause nutrient toxicity to plants [82].

The TDS values were well within the permissible range for drinking, i.e. 500 mg L^{− 1} (Table 5). High TDS will lead to poor water taste and odour, thereby making the water less appealing for drinking [83]. Moreover, high TDS is associated with osmotic stress to plants, ionic imbalance, and reduction in DO, etc. It ultimately destroys aquatic life and hampers agricultural yield [84].

PoM season was more turbid. The elevated turbidity during PoM is due to increased runoff via tributaries and surrounding human settlements, agricultural and pisciculture farms. The region recieves an annual average rainfall of 1,467.5 mm, with the maximum rainfall during the monsoon months that extends from May to mid-October [85]. The massive deforestation in the catchment areas and its erosion leading to siltation in the wetlands is common [24–25]. However, a turbid wetland will reduce plant growth by limiting light penetration, disrupt the feeding behaviour of fish and harbour a wide range of harmful pathogens and bacteria [86].

For PM, DO range from 2.5 ± 0.71 to 5.6 ± 0.01 mg L^{− 1} for Pumlen/Khoidum, and 2.4 ± 0.02 to 4.8 ± 0.01 mg L^{− 1} for Ikop/Kharung, respectively. The low DO in PM season is associated with the rise in surface temperature of wetlands. The DO for PoM ranged from 3.8 ± 0.28 to 6.5 ± 1.55 mg L^{− 1} for Pumlen/Khoidum, and 3.72 ± 0.056 to 5.88 ± 0.62 mg L^{− 1} for Ikop/Kharung, respectively. The agricultural expansion has lead to nutrient enrichment and thereby triggering eutrophication, with uncontrolled algal bloom. The result is the accumulation of dead and decayed organic matter, which spikes the Biological Oxygen Demand (BOD). It eventually reduces DO levels [47]. The low DO is a sign of stress [87]. A moderate to high DO values is required to sustain healthy biodiverse aquatic life forms. The wetlands are wintering habitats for non-local migratory waterfowls, such as Anas clypeata, Tadorna tadorna and Aythya ferina etc [24]. Therefore, maintaining a healthy invertebrate and other small organism’s population is required to support the migration.

It was observed that the wetlands do not have major polluting sources of chloride ions (Tables 3 and 4). Similarly, alkalinity values of Pumlen/Khoidum for both seasons were lower than WHO permissible limit. For Ikop/Kharung, 14.29% of the PoM sample exceeded the WHO permissible limit [59]. As per the USGS hardness classification [88] (Table 6), 90% of Pumlen/Khoidum samples for PoM were moderately hard and 40% for PM were found to be hard water, respectively. Similarly, most of the water samples for Ikop/Kharung fall into moderately hard (85.71% samples for both PoM and PM) and hard (14.29% samples for both PoM and PM) water category. No soft water category was reported from Ikop/Kharung. Pumlen/Khoidum showed softer water in the PoM season due to dilution of dissolved Ca²⁺ and Mg²⁺ ions through rainfall and runoffs from major and minor tributaries. However, no significant temporal change in total hardness class was observed for Ikop/Kharung, with the water being moderately hard or hard throughout the year. The hardness is influenced by agricultural activities such as irrigation and fertilization, which introduce minerals, leading to the formation of hard water [47]. Moreover, it is observed that the hardness is because of both carbonate and non-carbonate types, which is indicated by the presence of alkali metal ions (Supplementary Tables 8 and 9).

The low levels of alkali metal ions (Na⁺ and K⁺) revealed the lack of polluting sources of the ions. For Pumlen/Khoidum, P4 - Sarikhonjin and P5 - Nongmaikhong during PM season had Mg²⁺ levels reaching 32.19 ± 14.7 and 32.16 ± 30.95 mg L^{− 1}. The high Mg²⁺ levels in the wetland will affect the solubility of ions such as Ca²⁺ and PO₄³⁻, reduce the availability of the nutrients to wetland plants or agricultural crops, thereby impeding growth [89].

PoM sampling showed elevated nitrate concentrations, which is attributed to the higher inflow of agricultural and pisciculture discharges containing nitrogen-rich residues. It is coupled with wastewater runoff from human habitations and large water volume addition through the tributaries carrying fertilizer residues. The valley reports an annual fertilizer consumption rate of 45 kg per hectare leaching into the surroundings [90]. From the LULC analysis, it was observed that 24.61 km² of Pumlen/Khoidum and 25.21 km² of Ikop/Kharung were converted into agricultural fields or pisciculture farms between 1988 and 2023.

For PoM, 40% samples for Pumlen/Khoidum and 100% samples for Ikop/Kharung showed PO₄^3– concentrations well above the permissible limit of 0.1 mg L^{− 1}. In wetlands, high phosphates are linked with increased plant growth, phytoplankton proliferation and species composition change and shading of higher plants in the bottom strata. The excess nutrient loading in wetlands is linked with alterations of hydrology and its associated changes in vegetation pattern and nutrient cycling [91].

The NPI values calculated for the wetlands showed no pollution (Supplementary Table 10). The enrichment of nutrients support uncontrolled algal bloom and deteriorate water quality. The result is a habitat where the algal growth eventually limits light penetration, reducing growth and caused die-off of plants in littoral zones. In addition, creation of a hypoxic or anoxic ‘dead zone’ which lack sufficient oxygen to support organisms is common [92]. For example, Arend et al. [93] reported dead zones during summer months in the Laurentian Great Lakes in North America. In the wetlands, the nutrient pollution check is attributed to the phytoextraction ability of phoomdi species [47]. The phoomdi of Pumlen showed a good phytoextraction capability for phosphorus with the species such as Eichhornia crassipes > Salvinia cucullata > Zizania latifolia > Phragmites karka > Saccharum munja > Echinochloa stagnina showing high bioaccumulation factor values of 1.1 ×ばつ 10⁴, 9.8 ×ばつ 10³, 9.4 ×ばつ 10³, 9.4 ×ばつ 10³, 8.4 ×ばつ 10³, and 7.6 ×ばつ 10³, respectively. Further, phoomdi mats create a shade on the surface. It thereby reducing the amount of sunlight penetrating the water column and prevent excessive algal growth.

However, the perception of wetlands as nutrient sinks has led to their use as wastewater disposal sites by local communities for a long period. It causes severe alteration in the structure and function of the ecosystems by eutrophication [24–25].

Similar study performed on the wetland by Devi et al. [48] reported deteriorating water quality of Ikop. The condition is influenced by rapid population explosion and the corresponding anthropogenic actions carried out in the surrounding periphery. Devi et al. [47] in their study reported low pH (5.86 ± 0.30), high EC (377.6 ± 5.55 μS m^{− 1}), high turbidity (33.2 ± 9.407 NTU) and high phosphate concentrations (0.79 ± 0.14 mg L^{− 1}), from Pumlen in 2019-21. Likewise, Kangabam et al. [36] reported deteriorating water quality of Loktak with low pH (6.43 ± 0.38), low DO (4.11 ± 0.76 mg L^{− 1}), and low BOD (1.74 ± 0.72 mg L^{− 1}), respectively. A similar study conducted by Laishram et al. [38] on Loktak reported low DO (0.42 mg L^{− 1}) and low BOD (0.42 mg L^{− 1}), which showed a polluted wetland ecosystem characteristics.

However, for Pumlen/Khoidum and Ikop/Kharung, there is a lack of elaborate studies in the past that assess the water characteristics. As such, because of the data deficiency for the particular wetlands, it is difficult to track the water quality trends.

3.3 Multivariate statistical analysis

3.3.1 Correlation among water characteristics

The degree of linear relationship between the physico-chemical parameters and their co-variability for both PoM and PM season has been measured using Pearson’s correlation coefficient (r) (Supplementary Tables 11 and 12). A significant correlation coefficient with the value of r in the range of ± 0.5 < r < ± 1 shows a single source of the physico-chemical parameters. For Pumlen/Khoidum, a strong correlation was found between EC and TDS, Na⁺, K⁺ and Cl⁻; TDS and alkalinity, Na⁺ and K⁺; hardness and Mg²⁺; Na⁺ and Cl⁻; K⁺ and Cl⁻, NO₃⁻, respectively. Throughout the seasons, EC and TDS have a strong correlation (r = 1 for PoM and r = 0.856 for PM). It showed that the high EC is because of dissolved inorganic ions inflows through domestic discharge, agricultural and pisciculture runoff. Moreover, EC showed strong relation with Na⁺ (r = 0.921 for PoM and r = 0.771 for PM) and NO₃⁻ (r = 0.824 for PoM and r = 0.786 for PM). It indicates that the presence of the ions results into higher EC. The non point sources of the ions are domestic non-treated sewages from the surrounding villages and agricultural areas. The result showed that an increase in the concentration of one parameter will eventually increase the concentration of the other. The ions share a similar origin source [94]. Similarly, TDS showed strong correlation with NO₃⁻ (r = 0.811 for PoM and r = 0.731 for PM). Mg²⁺ has a strong correlation with total hardness (r = 0.976 for PoM and r = 0.990 for PM). It showed that high levels of magnesium observed throughout the seasons have profound impact on the total hardness (bicarbonate hardness). A negative significant correlation was found between pH and DO. It showed that if pH increases, then DO tends to decrease. The condition arises in eutrophic ecosystems where high organic matter decomposition utilizes DO, which releases CO₂ that forms carbonic acids and lowers pH [95].

For Ikop/Kharung, strong correlation was found between EC and TDS, hardness, Ca²⁺, Mg²⁺, Na⁺ and Cl⁻; TDS and hardness, Ca²⁺, Mg²⁺, Na⁺ and Cl⁻; hardness and Ca²⁺, Mg²⁺ and Na⁺; Na⁺ and Cl⁻, respectively. Similar to Pumlen/Khoidum, EC and TDS showed a strong correlation (r = 0.993 for PoM and r = 0.998 for PM). It reveals that the high EC is because of dissolved inorganic ions such as Ca²⁺, Mg²⁺, Na⁺ and Cl⁻ inflows from surroundings. The statistically significant strong positive correlation among the physico-chemical parameters showed their common sources of origin in the wetlands.

3.3.2 Principal component analysis

Principle Component Analysis was performed to identify the probable sources of pollution associated with the land use changes. It is used to identify the primary factors regulating the physico-chemical parameters of water samples obtained from Pumlen/Kharung and Ikop/Kharung. Pumlen/Khoidum demonstrated a Kaiser-Meyer-Olkin (KMO) measure of 0.60, while Ikop/Kharung recorded 0.62, respectively. Both systems also produced significant results for Bartlett’s test of sphericity (Pumlen/Khoidum: χ² = 342.98, p < 0.001; Ikop/Kharung: χ² = 139.09, p < 0.001).

Supplementary Tables 13 and 14 record the factor loadings that quantify the strength of the association between parameters and principal components (PCs). Similarly, Fig. 6 graphically illustrates the clustering of parameters and the dispersion of sampling sites.

For Pumlen/Khoidum, the first three PCs explained 85.57% and 81.19% of the total variance, respectively. PC1 is identified as an anthropogenic-salinity gradient, characterized by strong positive loadings (> 0.75) for EC, TDS, Na⁺, K⁺, Cl⁻ and NO₃⁻, respectively. The tight clustering of these parameters along the PC1 axis (Fig. 6a and b) indicates simultaneous increases in nutrients and salinity. It is likely driven by combined sewage discharge and agricultural runoff. Conversely, PC2 reflects natural lithogenic controls, driven by pH and Ca2+ (PM), and Total Hardness (TH) and Mg2+ (PoM). The orthogonal orientation of these hardness vectors relative to nutrient vectors (Fig. 6b) confirms that natural mineral dissolution operates independently of anthropogenic nutrient inputs.

In Ikop/Kharung, the first three PCs explained a cumulative variance of 87.69% (PM) and 82.97% (PoM) of the physico-chemical variability. Unlike the findings for Pumlen/Khoidum, PC1 in this system is primarily governed by hardness and mineralization. Both seasons exhibited high loadings for TH, Mg2+, EC, and TDS. As illustrated in Fig. 6c and d, the vectors for TH and Mg2+ are the longest and most prominent along the PC1 axis. It signifies dominant water-rock interactions and mineral weathering processes in this system [96]. Furthermore, nutrient dynamics in Ikop/Kharung showed significant variations; NO3− exhibits inconsistent behaviour, loading significantly on PC2 or PC3 depending on the season. The biplots clearly illustrate this divergence, where the NO3− vector frequently decouples from the salinity cluster (EC/TDS). It suggests that nutrient enrichment in Ikop/Kharung is likely driven by localized point sources or specific land-use practices, rather than the general ionic enrichment processes seen elsewhere.

3.4 Water quality characterization

3.4.1 Drinking water quality

During the study period, temporal and spatial variations in the WQI for Pumlen/Khoidum and Ikop/Kharung drinking were observed (Fig. 7).

The WQI values ranged from 52.09 to 182.0 (PoM) and 38.86 to 108.9 (PM) for Pumlen/Khoidum, and 65.87 to 106.6 (PoM) and 30.12 to 62.06 (PM) for Ikop/Kharung, respectively (Table 7).

The PoM season showed higher WQI as a result of large volume of runoff entry through the tributaries that carries pollutants and nutrients. Moreover, surface runoff from agricultural, pisciculture farms and domestic habitats increases due to heavy rainfall in monsoon. For Pumlen/Khoidum, 60% samples were in the poor category, 10% in the very poor category and 30% in the unsuitable category for drinking during PoM. The water quality improved during PM period with 50% samples in the good category, 30% in the poor category, 10% each in the very poor and unsuitable category for drinking, respectively. However, P1 - Langmeidong, P4 - Sarikhonjin, P5 - Nongmaikhong, P8 - Phoubakchou and P9 - Sekmaijin Khunou sampling locations didn’t see much improvement in water quality with season change. The WQI category of the sampling locations remained in the poor, very poor and unsuitable category for drinking all the year around. Among the sampling locations, the P1 - Langmeidong region in both seasons showed high WQI values of 108.9 (PM) and 143.9 (PoM), respectively, which is un-suitable for drinking.

For Ikop/Kharung, 42.86% samples were in the poor category, 14.28% in the very poor category and 42.86% in the unsuitable category for drinking during PoM season. In comparison to Pumlen/Khoidum, water quality improves significantly during PM season with 85.72% samples in the good category and 14.28% samples in the poor category for drinking, respectively (Table 7). However, I6 - Tentha 2 remains in the un-suitable (106.6) and poor (62.06) category for both seasons.

Overall, the degraded wetland water quality is a result of wastewater discharge from the densely populated peripheral human villages, uncontrolled runoff from pisciculture and agricultural farms encroaching inside the wetlands, which uses extensive amount of toxic chemicals, insecticides and pesticides, garbage dumped in the wetlands, tributaries draining into the wetlands carrying nutrients and pollutants from the densely populated areas of the valley and catchment areas of Manipur hills etc.

Similar studies by several researchers showed the deteriorating water quality of the Loktak Wetland Complex [35–36, 47]. Kangabam et al. [36] reported WQI value that ranges from 64 to 77 for Loktak. It showed the poor water quality of the Ramsar site. Similarly, Mayanglambam and Neelam [37] in their study showed that 62.5% of the samples from Loktak during PM were in the poor category. From the northeast India, Sharma et al. [97] reported WQI values between 93.69 and 158.26 for Deepor Beel, a Ramsar site in Assam. The study showed that most of the sampling locations on the Ramsar site fall in the poor category of water quality. Biswas et al. [34] in the water quality assessment of Rudrasagar wetland, a Ramsar site from Tripura, reported 83.34% of water samples during PM falling in the un-suitable category for drinking.

However, the locals residing in the periphery still use the freshwater for drinking and various other domestic purposes, which is unsafe in today’s conditions without proper treatment.

3.4.2 Irrigation water quality

The Multi Purpose Loktak Hydro Electric National Project is meant to provide enough water to 24,000 hectares of farmlands in the peripheral areas of LWC [25]. Therefore, suitability of the wetlands for their use in irrigation was analyzed using the parameters such as pH, EC, TDS, Na%, SAR, MH and KR (Table 8).

According to Ayers and Wescot [98], 100% samples of Ikop/Kharung for both seasons falls in the desirable category for irrigation. For Pumlen/Khoidum, 10% of sampling location (P4 – Sarikhonjin) was found to be in the unsuitable category during PoM, with pH < 6.5. The low (acidic) or high (alkaline) pH conditions in the soil limit nutrient availability, causing nutrient deficiencies and poor plant growth [99]. Further, the solubility of fertilizers and pesticides are affected by high pH, which decreases with rising pH.

The wetlands have EC values below 750 μS cm^{− 1}, which is in the good and excellent irrigation water category [63]. High conductivity levels in soil will lead to increased concentrations of dissolved salts in the agricultural fields. It will affect the ability of crops to absorb water and various essential nutrients. Further, high salinity will disrupt the osmotic pressure balance in plant cells and hamper the uptake of nutrients like Ka⁺, Ca²⁺ and Mg²⁺ etc.

The TDS values for the sampling locations of Pumlen/Khoidum and Ikop/Kharung were < 1000 ppm, indicating a freshwater category [100]. A high TDS value in agricultural soil will create osmotic stress on crops, making it difficult for the plants to absorb water.

As per Wilcox [62], a high Na% > 60% in the agricultural fields is injurious. It will expose the crops to alkali hazard. The toxicity will stunt plant growth and cause a reduction in the permeability of soil by collapsing the soil structure and blocking soil pores. The majority of sampling locations showed good and excellent Na% for both seasons. Further, the Wilcox diagram plots between Na% and EC (Fig. 8) showed that the wetlands are in an excellent to good category.

As per Bouwer [101], the wetland water has safe SAR values for both seasons. In agricultural soil with high SAR, crops struggle to establish a healthy root system that is essential for water and nutrient uptake.

Based on Kelly’s Ratio [65], the wetlands are safe for irrigation. According to USSLS diagram plots between SAR and EC, it was observed that the alkali hazard falls in low category whereas the salinity hazard for some of the sampling locations were in medium hazard category (C2-S1) for both seasons, respectively (Fig. 9). The elevated levels of salinity in the agricultural fields will cause osmotic stress, imbalances in nutrient uptake or availability and stunted growth, which ultimately reduces yield [102]. Zheng et al. [103] reported that salinity stress reduced the grain yield of rice because of the reduction in the seed set rate, formation of effective number of panicles and grains per panicle. The stress also affects the quality of grains produced, which is detrimental for the local farmers.

90% PoM samples and 10% PM samples for Pumlen/Khoidum and 100% PoM samples and 14.29% PM samples for Ikop/Kharung showed MH values in the unsuitable category [64]. The elevated magnesium levels in agricultural soil will elevate the alkalinity of soil, affect crop health and affect the availability of calcium required by crops. Excess magnesium presence is reported to inhibit rice growth and yield because of the reduction in total chlorophyll content, net photosynthesis rate, and membrane stability in the plants [104].

Hence, the findings showed that the use of the wetlands for irrigation will put the agricultural crops to the risk of salinity and Mg stress. Ultimately, the long term soil impact from salinity hazard and high magnesium stress will be severe physical degradation of soil, nutrient imbalances, and reduction in microbial activities. At the end, it can lead to permanent degradation of the agricultural lands if not managed timely [89, 105].

3.5 Limitations of the study

The findings of the present study serve as a pioneer report that assesses the land use land cover change and its impact on the water quality status of Pumlen/Khoidum and Ikop/Kharung. The results obtained will provide a clear snapshot of the water characteristics of the wetlands, which must be updated regularly. Further, the LULC analysis will help understand the problem of uncontrolled encroachment carried out in the region. Thus, the findings will be useful in the preparation of a database on the changing water characteristics and land use pattern of the wetlands. It will benefit the environmental planners during the management of the ecosystems.

However, the study has its limitations and shortcomings. First, though the Landsat data has shown reliable accuracy in wetland studies, it still has some limitations too. As expected, the classes like "thin vegetation and thick vegetation" and "water bodies and agriculture/pisciculture farm" were difficult to separate and show certain overlapping pixel values (about 10%). In the present study, an effective approach utilized to address this issue was careful use of "Fill" tool (this tool allows to change the wrongly classified pixels to the desired class or category), along with high resolution Google Erath data during the post processing stage. Accordingly, the authors were able to achieve the desired accuracy. Second, more elaborate water sampling which covers all the areas of the wetlands is missing in the present study due to several constraints. Third, a detailed statistical or spatial comparison between LULC and WQI/NPI of the wetlands is missing. A more precise spatial overlay analysis showing that the wetland areas with the greatest LULC change have the worst water quality must be covered. Fourth, the present study does not include heavy metal analysis of water, plants and sediment samples. Fifth, the absence of microbiological analysis is a limitation for the study, which must be carried out in future studies. Therefore, upcoming research need to cover various shortcomings addressed in this particular section.

The present study showed that the two wetlands, Pumlen/Khoidum and Ikop/Kharung in the Loktak Wetland Complex, have been heavily encroached. The wetlands have lost 36.78% (Pumlen/Khoidum) and 45.68% (Ikop/Kharung) of its areas to human expansion between 1988 and 2023, respectively. Subsequently, a sharp decline in open water area was reported from both wetlands. In addition, the mismanagement of wastewater from households, agricultural and pisciculture runoff has polluted the wetlands and made the water unsafe for drinking, which is a serious health risk for locals. Unfortunately, the regular water quality monitoring of other wetlands apart from Loktak is still missing. Since Loktak is a Ramsar site, a major focus is poured onto the lone Wetland of International Importance. Further, increasing pollution of the wetlands has become a hazard to luxuriant growth of wetland vegetation and crops, on which the wildlife and human depend for generations because of salinity and magnesium hazards.

Today, the fate of the second and third-largest wetland of the state, i.e. Pumlen/Khoidum and Ikop/Kharung is uncertain. For example, uncontrolled and unmonitored expansion of agricultural and pisciculture farms has almost entirely taken over Ikop/Kharung.

Therefore, it is necessary that conservation laws are introduced and enforced strictly, appropriate environmental management plans are framed, awareness of the public is enhanced, the development of society is looked after, better coordination among various conservation efforts is made, and the communities are invited in the decision making process. At the end, the inclusion of the "Keibul Lamjao Conservation Area" into the mixed UNESCO World Heritage site will provide the help in the conservation of the degraded wetlands from various degradative forces in play today.