Spatio-temporal variation of water quality and bio indicators of the Badulu Oya in Sri Lanka due to catchment disturbances

This article is published under the Creative Commons CC-BY-ND License (http://creativecommons.org/licenses/by-nd/4.0/). This license permits use, distribution and reproduction, commercial and non-commercial, provided that the original work is properly cited and is not changed anyway. Abstract: Safe and ample supply of freshwater is fundamental to humans and for the sustainability of ecosystem function. Therefore, impacts of catchment disturbances on surface water quality have been given special attention currently. The Badulu Oya sub-catchment of the Upper Mahaweli catchment area in Sri Lanka is one of the landscapes affected by intensive agricultural activities and urban development. This study was aimed at evaluating the spatial and temporal variation of water quality, stream physical habitat quality and macroinvertebrate bioindicators in the Badulu Oya catchment area in relation to disturbances due to agricultural and urban practices. Physicochemical water quality parameters and macroinvertebrate indices were evaluated at ten sampling sites in the Badulu Oya catchment area within a one year period. The results revealed that water quality variables such as electrical conductivity (EC), total solids (TS), total dissolved solids (TDS), dissolved oxygen (DO), alkalinity, nitrate-N(NO 3 -N) and physical variables such as stream channel quality index (CQI) and riparian quality index (RQI) catchment disturbances. Among the macroinvertebrate indices


INTRODUCTION
Aquatic ecosystems around the world are increasingly occurring in the respective catchment landscapes (Allan, 1995;Alam et al., 2006;Jayawardana et al., 2016).Alteration of land cover on the catchments of rivers and aquatic fauna in a number of ways.It has been reported that catchment alteration may impact stream ecological functions such as the dynamics of riverine carbon cycle and food web structure and functions (Webster et al., 1983;Gregory et al., 1991).Removal of natural vegetation in catchments may contribute to Subsequently, these changes may increase the sediment inputs to streams by erosion and alteration of substratum characteristics and channel morphology, thus reducing aquatic species diversity (Lake et al., 1985;Bunn et al., 1986).Altered catchment hydrology and land use can sources and enhance in-stream primary productivity resulting in changes of the aquatic trophic structure and benthic communities (Death & Winterbourn, 1995).the environment.The major rivers of Sri Lanka originate from the central hills and their catchment areas are subjected to a variety of land use changes as a result of development activities and intensive agricultural practices (Watawala et al., 2010).The Mahaweli River catchment is the largest catchment in Sri Lanka, which provides water to the Mahaweli River and the major reservoirs built along the river.This catchment is considered very important to the Sri Lankan economy as it also provides water needed to generate 55 % of the electricity requirement of the country, and to irrigate a Thiruchelvam, 2008).Badulu Oya originates from the Namunukula mountain range and is one of the main tributaries of the Mahaweli River.The catchment of this river has also been subjected to various land use changes during the last few decades.Previously forested catchment areas have been continuously cleared and converted into agricultural land and urban settlements.Evaluation of such land conversion impacts on stream ecological integrity is important for taking decisions in catchment management and for the adoption of better management practices.Effect of such land use changes on the ecological integrity of streams have to be assessed through holistic approaches taking into account physical, chemical and biological criteria.
In many countries in temperate regions of the world, macroinvertebrate bioindicators are commonly used to assess the stream health in conjunction with water quality data (Hilsenhoff, 1987;Muirhead-Thompson, 1987;Plafkin et al., 1989;Johnson et al., 1993;Lenat, 1993;Jayawardana et al., 2006a;2006b;2010;Jayawardana & Westbrooke, 2010).These bioindices are very effective compared to physico-chemical parameters.Such indices are also important for the accurate assessment of the status of river health and development of mechanisms for catchment management in Sri Lanka.The lack of baseline data on responses of aquatic bioindicators to land use impacts is one of the limitations of application of bioindices in river health monitoring programmes in Sri Lanka.Therefore, this study was conducted with the aim of evaluating the spatial and temporal variation of water quality and macroinvertebrate bioindicators in Badulu Oya catchment area with respect to agricultural and other land use disturbances.

Study area
Badulu Oya catchment is one of the sub-catchments of the Upper Mahaweli catchment area, which extends 318 km 2 .
Sri Lanka.The study region receives an annual average rainfall of 2000 mm.The Badulu Oya catchment receives rains during the wet season (October to March) and dry weather prevails in the catchment from April to September.The landscape of the catchment area varies from undulating, rolling topography to hilly, steeply dissected mountain terrain.The predominant soil type in the area is red yellow podzolic (De Alwis & Panabokke, 1972).The landscape of the catchment was covered with natural vegetation until the early 19 th century (Wickramagamage, 1988).Subsequently the forest cover of the catchment area has been greatly reduced due to the expansion of agricultural lands and urban development activities.Presently fragmented patches of natural forest are present in the catchment and the rest of the area is covered by tea, vegetable and paddy cultivated lands, and urban settlements.Vegetable crops are predominantly grown in the Badulu Oya catchment area during the dry season and paddy is cultivated during the rainy season.
Ten locations in 2 nd or 3 rd order tributaries of Badulu Oya were selected as sampling sites for the study (Figure 1).The catchment area land cover patterns were analysed using Landsat TM satellite images (Data or acquisition 10 th March 2011, with 30 m × 30 m resolution), 1:50000 topographic maps (Department of Survey, Sri Lanka), Google Earth satellite images and of the selected sampling tributaries ranged from pristine undisturbed natural forests to landscapes, which have been severely altered by deforestation, agriculture and urban settlements.The selected sampling sites were forested land cover in the respective micro-catchments of the sampling tributaries.The micro-catchments, disturbed sites and more than 35 % forest cover as less disturbed sites/ reference sites (Table 1).

Riparian and channel physical habitat quality estimation
Riparian quality index (RQI) and channel physical habitat quality index (CQI) were developed to assess the quality of riparian zone of streams and the channel habitat quality, respectively.RQI and CQI for the sampling reaches were developed using the criteria of rapid bio assessment protocols for use in streams and wadeable rivers (Barbour et al., 1999).For the development of RQI, attributes such as the extent of lateral extension of forest cover on either side of the river, riparian composition (woody trees/ grasses/ bare) and riparian continuity were assessed along a 300 m stream segment upstream from the sampling site.CQI was developed for each sampling reach using the primary channel characteristics (channel dominant substrate, water velocity, embeddedness) and secondary channel characteristics (velocity/depth regimes, degree of bank/channel alterations, sediment canopy cover).

Water quality and macroinvertebrate sampling and analysis
Water samples were collected at monthly intervals from 10 sampling locations covering 10 sub-catchments from August 2014 to July 2015.pH, electrical conductivity (EC), temperature and dissolved oxygen (DO) were measured in situ using a portable multi parameter (Hach-MM 156).In addition, physical and hydrological data of expandable Global Water Flow Probe) and substrate characters and embededness were visually estimated (Platts et al., 1983;Fitzpatrick et al., 1998).Duplicate water samples were collected from each sampling location and brought to the laboratory using 1.5 L sampling bottles under cold conditions for analysis.For the calculation of total solids (TS), 25 mL of samples Total suspended solids (TSS) was measured after weight was taken after drying at 105 between TS and TSS was calculated as total dissolved solids (TDS).NO 2 -N, NO 3 -N, PO 4 3-, SO 4 , and NH 3 -N were measured in the laboratory using a Hach DR 2700 spectrophotometer.Biochemical oxygen demand (BOD 5 ) was measured after incubating a water sample for 5 d following APHA (2005) procedures.Ephilithic biomass (g/m 2 ) was determined following the method adopted by Sponseller et al. (2001).Total coliform bacteria and faecal coliform bacteria were also measured in collected water samples adopting the US Environmental Protection Agency methodology (USEPA, 1978).
Macroinvertebrate samples were collected from study stream reaches (ten locations) using a surber sampler with the dry and rainy period of the catchment.Five replicate samples were collected from each sampling location.Macroinvertebrates were sorted and preserved in 70 % up to family level in the laboratory Macroinvertebrate total abundance, family richness, percentage EPT taxa (ephemeroptera: plecoptera: trichoptera), percentage chironomids and Shannon diversity index (SDI) were calculated using macroinvertebrate data.Taxa richness and percentage EPT were calculated because they are universally used macroinvertebrate metrics.These metrics are easily calculated and track water quality changes effectively (Wallace et al., 1996;Karr & Chu, 1999).Percentages of chironomidae were also estimated at the sites since they are indicators of the stress caused by pollution (Lencioni et al., 2012).Information from a number of sources was used to partition the invertebrate fauna into 5 major feeding categories, viz.shredders, collectors, predators, et al., 1984;Hauer & Lamberti, 1996;Gooderham & Tsyrlin, 2002).

Analysis of data
Principal component analysis (PCA) was conducted to explore the patterns of variation of the sites based on stream physical habitat quality, water quality and macroinvertebrate indices.Stream physical habitat quality, water quality data and macroinvertebrate indices were used as variable inputs for PCA.Since the input variables for PCA had different magnitudes and scales of measurements the data were standardised to produce a normal distribution of all variables.From the standardised correlation matrix of the data the initial factor solutions were extracted by multivariate principal components extraction.Components loading (correlation between the variables and the principal components components represented by high loadings were taken into consideration for evaluation of the components (Mazlum et al., 1999).Multivariate analysis was conducted using PRIMER-7 software (Plymouth Marine Laboratory, Plymouth, UK).Two-way analysis of variance (ANOVA) was conducted to evaluate the effect of land cover disturbance and seasonal impacts on stream water quality and macroinvertebrate indices.The degree of catchment disturbances (high: medium: low) and the season (dry and wet) were considered as main factors in the analysis.Data were checked for normality and homogeneity of variance before the analysis, and the data were log or square root transformed as necessary to meet the assumptions of linear models.Homogeneity of residuals was assessed from normal probability plots.

Multivariate analysis
PCA of the physical habitat, water quality and macroinvertebrate indices in response to catchment scale land cover is depicted in Figure 2. The results of the PCA could be used to interpret the underlying trends of variation of stream water quality, physical habitat quality and macroinvertebrate indices in the Badulu Oya with respect to catchment disturbance. of the data variability.Vector loading to the principal component axis indicated that SDI, NO 2 -N, family richness, RQI and CQI are highly associated with the PC1 axis.Conductivity, TDS, SO 4 2-, grazers and collectors were associated with PC2.There was a separation of sites along the PC1 axis based on the levels of disturbance of the catchments.Sites associated with a high percentage forest cover at the catchment were grouped together in the positive side of the PC1 and the sites with low percentage forest cover in the negative side of the axis.SDI, family richness, RQI and CQI were associated with high percentage forest cover in the catchment while NO 2 -N, NO 3 -N, TSS, PO 4 3-and percentage chironomidae were associated with highly disturbed sites.The results of the PCA indicate a pronounced impact to the stream ecological health exerted by the catchment disturbance.
The spatial and temporal variations of water quality in highly and less disturbed sites were tested using two-way ANOVA.The averages and ranges of the selected water quality variables reported at sampling sites during the sampling period are given in Table 2.Many of the water quality parameters tested were within limits of ambient water quality standards for inland waters Sri Lanka (CEA, 2001) except TSS, PO 4 3-and total and faecal coliform levels.
The results of two-way ANOVA indicated a parameters between seasons and the sites associated with varying degree of catchment disturbances (Table 3).Among the water quality variables tested, temperature is a vital factor for the growth and development of many aquatic species and for the function of all biological and biochemical reactions in the water.The temperature of the stream water can be affected by the level of solar radiation reaching the stream channel (Beschta &Taylor, 1988;Rutherford et al., 1997).According to the results of the present study, average stream temperature varied from 21 to 26 C among sampling sites.Results of the   variation of average stream temperature between the two associated with micro-catchments with varying levels of land disturbances (Table 3).Sri Lanka is a tropical island with no marked seasonal changes of temperature during the year as in many other temperate countries.
The Badulu Oya catchment receives comparatively high rainfall during the wet season than in the dry season.It is evident from the results of the present study that the locations could be related to dry and wet seasons of the catchment.
Among the other water quality variables tested in the present study, EC, TS, TDS, TSS, DO, alkalinity and NO 3 impacts and the degree of catchment disturbances.The EC of sampling sites ranged from 79.31 to 289.2 -1 during the dry period in comparison to the wet season and in sites associated with highly disturbed landscapes (Table 3 and Figure 3).TSS values varied between 62 to 151 mgL -1 among sampling sites, and in many sites these values exceeded the ambient water quality Lanka (CEA, 2001).It has been observed that in the given catchment most of the soil preparation activities for vegetable cultivation are conducted during the dry fertiliser, dissolved ions and soil particles coupled with the observed water quality variables in disturbed sites.Land degradation and water quality deterioration due to soil erosion by unplanned agricultural activities carried out in the Mahaweli catchment is also reported by many authors (Dharmarathne et al., 2008;Gunawardena et al., 2010;Hewawasam, 2010).NO 3 -N levels in sampling sites varied between 0.1 to 2.9 mgL -1 (p < 0.05) by land cover disturbances at the catchment (Table 3 and Figures 3 and 5).Similar results were reported by many authors that the land cover at the catchment scale is a good predictor of in-stream nutrient concentration, particularly nitrate (Omernik, 1977; Close & Davies-Colley, 1990; Johnson et al., 1997).It has been reported that the sediment input to streams resulting from soil erosion carries more nitrate and phosphate than that of in the dissolved fraction in water (Waters, 1995;Amarathunga et al., 2013) carried with runoff from vegetable cultivated lands may have increased the nutrient levels in the stream water in disturbed landscapes.There was also an increase of NO 2 -N, NH 3 -N and SO 4 2-levels in sites associated with highly disturbed catchments but these variations were authors that nitrites, ammonia containing compounds and organic loads contaminate waterways through wastewater discharge from urban areas (Allan, 1995).It can be expected that the waste from urban settlements in the catchment may have also contributed to such results.The microbiological quality of the water in study sites indicated elevated total and faecal coliform bacteria levels, which exceeded the proposed ambient water quality standards (total coliform < 20000 colonies /100 mL, faecal coliform; 250 600 colonies /100 mL) 2001) (Table 3).Two-way ANOVA results indicated a between the two seasons and among sites associated with micro-catchments with varying levels of land disturbances.The amount of faecal coliforms in the river Program, 2005).The elevated levels of faecal coliforms reported in highly disturbed sites in the Badulu Oya may have resulted from sewage input from urban areas and human settlements in the catchment.vegetation in terms of quality of vegetation structure (Tánago & Jalón, 2011).In the present study RQI scores ranged from 3 to 17 and high scores were recorded in less disturbed sites.PCA results also indicated that RQI is associated with sites with high percentage forest cover in the catchment.Human induced activities such as agriculture can have a Ferreira, 2005).In many sites of the study, encroachment of banks of the streams by cultivated lands was evident and the riparian vegetation is completely cleared or fragmented.Lack of riparian cover along the river in agricultural and urban areas may have contributed to the higher inputs of contaminants in stream water through runoff.et al., 1998).The results clearly demonstrated that stream with high level of catchment disturbances, and with the season (Table 3).The results suggest that stream channels in sites with disturbed landscapes are highly embedded 5).It is possible that the high input of soil particles from landscapes disturbed by vegetable cultivation and urban settlements in the given catchment may have contributed to the observed differences.

Macroinvertebrate bioindicators
During the study a total of 38 benthic macroinvertebrate families were recorded in the Badulu Oya catchment (Table 4).PCA and the results of two-way ANOVA indicated the factors contributing to the variation of macroinvertebrate indices in the sampling sites.Among the macroinvertebrate indices that were tested, total (Table 3; Figures 2 and 5).Macroinvertebrate total abundance and richness decreased with increasing in the PCA, improved water quality along with higher association with macroinvertebrate indices such as family richness and total abundance.It has been reported by many authors that stream features such as with thermal regime generate habitat templates upon which macroinvertebrate assemblages are structured (Southwood, 1977;Poff & Ward, 1990), which would be the case for the trends observed in the present study.However, other bioindices measured in the present study, i.e. percentage ETP taxa and percentage chironomidae, disturbance as depicted in the results of ANOVA test.However, in the PCA chironomidae percentage was associated with degraded catchments.This suggests that those indices need to be further tested for their application in the local context.The results of the present study also suggest that catchment scale land use activities features and water quality, ultimately determining the macroinvertebrate assemblage structure.
The present study shows that catchment disturbance through agriculture and urban activities also affect the functional feeding groups of macroinvertebrates.The functional feeding categories of biota in the streams are a better indicator of the energy dynamics within the streams as described in the river continuum concept (Vannote et al., 1980).The results of the present study indicated that the sites, which are impacted by land disturbance in the adjacent catchments are mostly dominated by collectors (Figure 6).The sites which are less affected by disturbance were dominated by grazers (Figure 6).It can with increased levels of organic sediments entering the streams in disturbed sites may have contributed to increased number of collector gatherers in the impacted sites.
indicate that catchment disturbance impacts by land use changes exert on stream ecological health of the Badulu Oya catchment.

CONCLUSION
that land use activities in the Badulu Oya catchment Chemical water quality variables such as EC, TS, TDS, DO, alkalinity and NO 3 -N; physical variables such as stream CQI and RQI indices; biological variables such as faecal coliform counts, macroinvertebrate indices like suggests that there is a possibility to incorporate such variables in river health monitoring programmes in the given catchment.
However, many of the water quality monitoring programmes in Sri Lanka mainly focus on onsite water quality analysis to estimate the level of pollution.The present approach is important for accurate assessment of stream ecological health since it incorporated diverse attributes to assess the stream health.Lack of baseline data on biological indices to measure stream health in Sri Lanka makes this approach even more important for further development of bioindices and their application in river health monitoring programmes in Sri Lanka.

Figure 1 :
Figure 1: Sampling locations in Badulu Oya were conducted using SPSS statistical software (IBM SPSS Statistics for Windows, Version 20.0 Armonk, NY: IBM Corp) and Microsoft Excel 2007.

Figure 2 :
Figure 2: PCA plot indicating site variation based on stream physical habitat, water quality and macroinvertebrate bioindices (two letters in the legend indicate level of disturbed land cover based on the percentage forest cover at catchment scale L = low disturbed sites; H = highly disturbed sites)

Figure 3 :
Figure 3: Monthly average water quality variation in sites associated with high (H_D) and low (L_D) catchment disturbances at entire catchment area

Figure 3 :Figure 3 :²
Figure 3: Monthly average water quality variation in site

Table 1 :
Sampling reaches of Badulu Oya catchment area

Table 2 :
Mean and the range of water quality at sampling sites during 12 month sampling period (1)the National Science Foundation of Sri Lanka 46(1)March 2018

Table 3 :
Results of main factors and their interaction for the tested water quality variables in the two-way ANOVA (p values) of the National Science Foundation of Sri Lanka 46(1) March 2018 Continued -March 2018 Journal of the National Science Foundation of Sri Lanka 46(1)-continued from page 59

Table 4 :
(Elosegi et al., 2010) recorded in sampling sites during wet and dry seasons (+ present; -absent) Among the physical attributes tested, channel physical within the channels.Improved habitat diversity within channels support diverse communities and increase the species richness(Barbour et al., 1999).In the present study CQI varied from 3.1 to 18 among the sites.High CQI values were recorded in sites associated with high forest cover and low CQI values resulted in sites associated with low forest cover.This suggests that local riparian vegetation has a strong impact on structuring stream channel features.Among the various functions of the riparian zone on structuring channel features, riparian canopy contributes to channel shade (Davies-Colley & Rutherford, 2005).Wood recruitment in riparian forests determines the amount of large woody debris in the river channel and plays an important role in determiningBrooks et al., 2003), velocity patterns, substrate composition and in-stream habitat heterogeneity(Elosegi et al., 2010).Embeddedness rubble, gravel) in the stream bed are surrounded or et al., 1983; Fitzpatrick Figure 4: Variation of macroinvertebrate indices in sites associated with high (H_D) and low (L_D) catchment disturbances at entire catchment area Percentage of EPT Percentage of chironomidae March 2018 Journal of the National Science Foundation of Sri Lanka 46(1) Figure 6: Variation of macroinvertebrate feeding groups in sites associated with high (H_D) and low (L_D) catchment disturbances at entire catchment area meso-habitat sequences (