Materials and Methods

Aghanashini is one of the few rivers flowing towards the west without major anthropogenic interventions (free-flowing river). Earlier studies confirm diverse flora and fauna along the riverscape [4,15,16,18,19,41] compared to the adjacent river catchments. This west-flowing river originates at Manjaguni and Shankara Honda (Sirsi) [20,41,42] and traverses a distance of 128 km [43] and joins the Arabian Sea. The catchment area of Aghanashini is about 1449 km² [42]. It is spread across the coastal and hilly agro-climatic zones in Siddapura, Sirsi, Ankola, and Kumta taluks of Uttara Kannada district [44]. The population has increased by 9.2% from 221,562 (2001) to 241,884 (2011) in the catchment [45]. The population is projected to increase to 264,137 by 2021. Elevation in the river catchment ranges between 0 and 786 m ASL. The undulating terrain of the Sahyadrian (Ghats) has denser stream networks, and the coastal regions have a sparse stream network with the broader riverbeds. The soil in the catchment is mainly clayey skeletal, loamy skeletal, along with clayey, fine, sandy, and loamy soils [46]. Figure 1 depicts the location of the Aghanashini River basin in Uttara Kannada district, Karnataka State, India, with the population density, topography, lithology, and agro-climatic zones. These spatial layers were generated using open-source GIS (QGIS) with the data compiled from the secondary sources (topographic maps of the Survey of India, Census data) and field data (collected using pre-calibrated handheld global positioning system (GPS)).

Figure 1. Aghanashini River catchment—population density, topography, soil, and agro-climatic zones.

2.2. Method

The data used for the analyses are listed in Supplementary Table S1, which were collected from the field (primary source), and secondary sources such as spatial data (Remote sensing data—RS) acquired at regular intervals through space-borne sensors [47], rainfall data [48–50], extra-terrestrial solar radiation and temperature data [51,52], temporal population data [45], livestock data [53], meteorological data [49], agriculture and crop information [44,50], topographic maps [43], virtual online remote sensing data [54,55], and catchment conditions [56]. Remote sensing data were preprocessed to eliminate positional errors, and geometric corrections were made using ground control points obtained from field (using GPS), the Survey of India topographic maps of Scale 1:50,000, and virtual earth databases [54,55]. Radiometric corrections were made to enhance the scene radiometric properties (contrast enhancement) for better interpretation of the data [57,58]. The protocol adopted for assessing the eco-hydrologic and environmental regimes (physicochemical and biological integrity) with the landscape dynamics in the Aghanashini River catchment is given in Supplementary Figure S1.

Fieldwork was carried out for 38 months (during June 2016 to May 2019) to understand the seasonal variability of the water quality and flow characteristics at sampling locations (Figure 2) in various streams across various micro-watersheds of the Aghanashini River basin in the Central Western Ghats. The data collected from the field include training data for land use analysis, flow dynamics, physical, chemical, and biological quality of water in the selected streams, hydrological regime, and ecological footprint. Flow dynamics (discharge) in select streams of the micro watershed were gauged monthly at sampling locations (Figure 2) using a current meter (or float based on site conditions) with the area velocity relationships and extrapolated to ungauged streams [56]. Water quality changes in relation to land use, flow regime, and across seasons are assessed at the sub-catchment level (Chandikaholé) through continuous field monitoring for 28 months of 9 streams.

Figure 2. Locations of (i) gauging flow discharges, (ii) training data for remote sensing data analyses, and (iii) water sample collection (for assessing physical, chemical, and biological quality).

2.2.1. Land Use Dynamics

Land use analyses involved (i) generation of false-color composite (FCC) of remote sensing (RS) data (bands-green, red, and NIR). FCC helped in locating heterogeneous patches for choosing training polygons in the landscape; (ii) selection of training polygons covering 15% of the study area and polygons are uniformly distributed over the entire study area, covering all land use categories; (iii) loading these training polygons co-ordinates into pre-calibrated GPS; (vi) collection of the corresponding attribute data (land-use types) for these polygons from the field; (iv) supplementing this information with the data from the online data portal [54]; (v) 60% of the training data were used for classification; and (vi) the balance is used for validation or accuracy assessment. The land use analysis was performed using a supervised classification technique based on a Gaussian maximum likelihood algorithm with training data. Accuracy assessment (computation of Kappa statistics, overall accuracy, producer accuracy, and user accuracy) was carried out by comparing the classification output with the training data (field observations) collected using GPS [57–59].

2.2.2. Assessment of Hydrological Footprint, Ecological Footprint, and Eco-Hydrologic Footprint

Hydrological regime analyses involved the quantification of (i) run-off, (ii) infiltration, (iii) soil water availability, (iv) sub-surface (vadose) flow, (v) groundwater recharge, (vi) evapotranspiration (PET from vegetation), and (vii) assessment of the hydrologic regime as a function of various factors such as land use, precipitation, temperature, solar radiation, soil characteristics, geology, and topography [12,56,60,61]. Temporal rainfall data of all rain gauges in the catchment were compiled from India Meteorological Department (IMD) [49] and the Directorate of Economics and Statistics, GoK [50]. The average monthly and annual rainfall data [61, 62] were used to understand the spatial pattern of rainfall in the study area to derive the gross yield and net yield by considering interception [56].

Stream gauging also aided in calibrating the run-off model at sub-basin levels. Run-off is computed by considering land use and rainfall based on the rational formula [56]. The physical parameters for water supply include run-off (overland flow), infiltration (subsurface and groundwater recharge), and soil water availability. After precipitation, a portion of the rainfall that flows in the streams is (i) surface run-off or direct run-off and (ii) subsurface run-off. Surface run-off refers to the portion of water that directly enters into the streams during rainfall, which is estimated based on the empirical relationships [9–11,21,22] considering run-off coefficient, depending on land uses [56].

The portion of water that enters the subsurface (vadose and groundwater zones) during precipitation depends on land cover in the catchment. During field monitoring of streams in the forested catchment, overland flow is noticed in streams only after the saturation of subsurfaces. The water stored in sub-surfaces will flow laterally towards streams and contribute to streamflow during non-monsoon periods, referred to as pipe flow (during post-monsoon) and base flow (during summer).

Water demand assessment included the societal (water for domestic purposes, agriculture, horticulture, livestock, and industrial) and ecological (to maintain the terrestrial and aquatic integrity) requirements. The societal water demand for agriculture, domestic, and livestock sectors was compiled from field observations and supplemented with secondary data from government agencies. Agriculture and horticulture demand were quantified considering crop types, cropping patterns, growth phase, and water requirements per crop. Domestic water demand was estimated considering daily water demand (Table S1). Similarly, water demand for livestock was quantified by considering animal type, population, and water requirements per animal (Table S1) [60]. Ecological (aquatic) water demand in the river is assessed by considering the flow regimes and biodiversity in micro-watersheds. Terrestrial water demand was estimated considering vegetation type-wise and actual evapotranspiration (AET) using the modified Hargreaves Method in the diverse landscape (details are given in Table S1).

Month-wise water availability and demand were computed to understand the eco-hydrological footprint. The eco-hydrologic footprint helps in understanding water-scarce deficit (supply < demand) and surplus (supply > demand) situations. The hydrological flow regime in each catchment was assessed based on field observations, and streams were categorized into four groups [60]: A (perennial streams with 12 months of adequate water), B (8 months), C (6–8 months), and D (4 months, only during the monsoon).

The ecological footprint was assessed considering biotic elements (biodiversity), i.e., flora and faunal species. The spatial distribution and species richness of plants and animals in the river catchment were compiled from the field (transect based quadrat sampling) and the published literature—books [12,20,63–67], conference papers [18,41,68,69], journals [4,15,16,18,19,41,60,70–74], and web portals [75,76].

The eco-hydrological footprint of the Aghanashini River was evaluated considering the seasonal variability of water availability and water demand. The eco-hydrological footprint, forest cover, flow regime, and species distribution were compared across sub-basins to understand the linkages and inter-relationships among hydro-ecological aspects. Based on these assessments, streams in a sub-catchment were considered for water quality assessment in relation to land use, flow regime, and other characteristics.

2.2.3. Water Quality Assessment

Water quality assessment was carried out in select streams of the Chandikaholé sub-catchment of the Aghanashini River basin (catchment id—8), and locations were chosen based on the eco-hydrological footprint, distribution of flora and fauna, and flow duration. The Chandikaholé stream originates near Yaana and joins the main river—Aghanashini at Bagribailu. Flow regime, water quality in the stream, and land uses were assessed in the micro- and macro-watersheds. Yaana, Nanalli, Beilangi, Mastihalla, and Harita are the micro-watersheds connecting Aanegundi (AGT1), whereas Aanegundi (AGT1 and AGT2) and Bialgadde (BGT) are the macro-watersheds. Aanegundi AGT1 and AGT2 join near Yaana Cross along the Sirsi-Kumta Road.

The streams in this catchment were monitored (Figure 2) for 28 months to understand the seasonal dynamics of water quality (18 physical and chemical parameters) at 9 sampling locations—Beilangi (BE), Yaana (YK), Nanalli (YNK), Harita (HA), Bialgadde (BGT), Aanegundi (AG), Aanegundi tributary 1 (AGT1), Aanegundi tributary 2 (AGT2), and Mastihalla (MH). Water temperature (WT—laboratory thermometer), dissolved oxygen (DO—Winkler’s Method), discharge (current meter), electrical conductivity (EC), total dissolved solids (TDS), and pH (using Eutech: PCSTestr 35) were measured at the sampling location (on-site), while the other parameters such as total alkalinity (TA—titrimetric method); chemical oxygen demand (COD); biochemical oxygen demand (BOD); total hardness (TH) and calcium (Ca) using EDTA titrimetric method; magnesium (Mg); chloride (Cl—argentometric method); nitrate (phenol disulphonic acid method); orthophosphate (OP—stannous chloride method); sodium (Na) and potassium (K) using the flame emission photometric method were analyzed in the laboratory (off-site) according to the standard protocol [1,23,30,31,37,61]. Based on the temporal data, using a weighted arithmetic method, the water quality index (WQI) was computed [40,61,72] season-wise across sampling locations considering physicochemical parameters such as dissolved oxygen, electrical conductivity, total dissolved solids, pH, total alkalinity, total hardness, calcium, magnesium, chloride, and nitrate [40,61,72,73]. Water quality is graded as excellent (for WQI = 0 and 25); good (for WQI = 26–50); poor (for WQI = 51–75); very poor (for WQI = 76 and 100); and unfit for drinking (WQI > 100).

Multivariate analysis of season-wise and sampling location-wise water quality data through Pearson’s correlation coefficient (r), CA, and PCA was carried out using PAST software [35,37,42,61,62] to understand the contributing factors of pollution. PCA of water quality parameters [37] of nine streams was performed, and the scree plot shows principal components explaining variance. Components with an eigenvalue > 1 were considered significant, while < 1 were omitted from further analysis.