the Impact of Watershed Land Use on River Water Quality: Implication for Planning in the Case of Kebena River Watershed, Addis Ababa, Ethiopia

Background: The impact of watershed land-use on surface water quality, especially in urbanized areas, has been investigated in numerous studies in developed countries, however, the issue is one of the under researched areas in developing cities of Africa. To study the impact of watershed land-use on surface water quality, we used the main land use types of the Kebena watershed from aerial photograph and collected river water samples during the dry and wet seasons for two consecutive years at different points from the river course. We calculated the share of each land use using ArcGIS and tested the water quality during each season. The variations in water quality parameters relating to the different land use types of the sub-watersheds were analyzed using ANOVA. Results: Kebena watershed is mainly covered by 39.14% forest, 32.51% built-up area and 27.25% cultivated land. At sub-catchment level, Denkaka, (agriculture dominated) sub-catchment with 44.90 % cultivated land, Little Kebena, (forest dominated) sub-catchment, with 60.87% forest cover and Gine (urban dominated) sub-catchment with 90.44% built-up area were identied. The variations in water quality parameters relating to the different land use types of Kebena watershed revealed the signicantly high seasonal relationship between the concentration of the water quality indicators during the dry season at (P<0.001, P<0.05). Furthermore, there is a strong positive relationship between the urban and forest dominated sub-catchments and water quality indicators during both the wet and dry seasons than agriculture dominated sub-catchment.

nature and location of development which mostly is decided by urban planners signi cantly in uences both the generation and resolution of environmental problems (Patricia et al., 2013;Barbec, 2009;Carter, 2007). Though urban plans play a vital role to balance the various development needs and the provisional capacity of available land, it is not common to integrate anticipated environmental pressures and further study the impacts of upcoming developments on natural resources (Carter, 2007). This is especially true with river water quality related issues that are addressed by a different entity apart from major urban planning factors, which have greatly affected rivers as most projects originate from urban planning decisions (Cettner, et al., 2013;Patricia et al., 2013;Carter, 2007).
To maintain the balance between nature conservation and development demand in an urban setting, watershed planning and management is becoming a key approach to minimize the impact of watershed land use on surface waters. Historically, watershed planning and management were used to solve ood-related problems (Ward et al., 2012). Currently, however, attempts are being done to study the relationship between land use, the ow and storage of water, and water quality (Doerfer and Urbonas, 2004). Moreover, habitat conditions, ecology, and community values are taken into consideration as the vital aspect of watershed-based planning and management approach (EPA, 2013;EPA, 1995). Watershed planning and management also enhance program e ciency, improve coordination among institutions, and balance the allocation of resources and funding (Browner, 1996;Doerfer and Urbonas, 2004).
In Addis Ababa, rivers and riversides are among the natural resources that are negatively affected by the rapid watershed urbanization (Worku and Giweta, 2018;AACSPPO, 2017;Yohannes, and Elias, 2017;Mazhindu et al., 2012;Gebre and Rooijen, 2009;Alemayehu, 2001).Land-use change and deforestation in Addis Ababa have resulted in high level of river water pollution, which has become a threat to human health and native biodiversity causing habitat loss and fragmentation (AACSPPO, 2017;Weldegebriele et al., 2012;Beyene et al. 2009). Kebena River is one of the main tributaries of the Awash River basin, originating from the northern tip of the Entoto Mountain in Addis Ababa. Despite the fact that many people use Kebena river water for various domestic purposes, the river is continually exposed to contamination by solid and liquid wastes from different sources (Worku and Giweta, 2018;Seyoum et al., 2017).
Thus, the aim of this study is to examine the impact of watershed land use on Kebena river water within its spatial and seasonal variation. Furthermore, the study assessed the river's water pollution linkage with the dominant land-use types within the three catchments of Kebena watershed. The ndings of the study are, therefore, utilized to forward pertinent recommendations for major watershed-level planning and management interventions to improve water quality of the Kebena River.

Description of the study Area
This study was conducted in the Kebena river watershed draining the upper most central part of the city of Addis Ababa (Fig. 1).
Generally, the Addis Ababa river catchments are drained by two major river systems such as Tinishu Akaki (Little Akaki) and Tiliku Akaki (Great Akaki) rivers. The two rivers originate from the Entoto Mountain chains and ow down to the man-made Lake, the Aba Samuel water reservoir, at the lower end of the City (Gebre & Rooijen, 2009;Beyene et al., 2009). Kebena River is characterized by mountainous landscape that dissects the central part of the City and its watershed is situated in the northern-central part of the city, within a geographical location between of 8° 9' 50'' N -9° 6' 60''N latitude, and 38° 45' 30'' -38° 49' 30''E longitude (Fig. 1). The total estimated area is about 5,150 ha with an elevation range of 2000 to 3200 m.a.s.l.
The ow of water in the rivers varies during the dry and wet seasons, thus varying the pollution character. In Addis Ababa, the dry season extends from December to May while the rainy season extends from June to September. During the dry season, the city experiences an average temperature of 6°C to 29°C. There is also a short rainy season that falls between February to March, during this period the city experiences mild rains. Whereas, the long rainy season is from June to September and the largest amount of rain falls during this period, which averages to about 900 mm as calculated from 30 years of meteorological data (Fig. 2).

Land use and Watershed Analysis
Major land uses in the Kebena watershed were mapped using manual digitization of recti ed aerial photograph of the year 2016 to categorize the land use types using ArcGIS 10.1 software. With the assumption that different land use types have varying level of potential effects on water quality, the land use types were categorized into four groups based on the dominant land use : (1) forested land use including natural forest, Eucalyptus globulus plantation, and riverine vegetation, (2) built-up area (impervious surfaces) accommodating residential, light industries, hospitals, schools other land uses (3) Cultivated land (pervious surfaces including grassland, and agriculture land), and (4) bare land (degraded bare soil/rock surface ) (Figure 3).
The study also used a 30m Digital Elevation Model (DEM) to determine the ow-path directions of the three sub-watersheds using the hydrology tool of spatial analyst tool set in ArcGIS version 10.1. The outlet points of the sub-watersheds were used as sampling sites whereby, the ow-path-direction grid cells were used to produce upstream catchment for each site. Based on the result of the hydrology tools, three sub-watersheds were identi ed as Urban, Agriculture, and Forest dominated sub-watersheds within Kebena watershed. The impact of the lower part of the main watershed is included as part of Kebena watershed. Databases have been transformed into a common digital format, projected onto a common coordinate system (UTM) and analyzed as vectors in ArcGIS (Sliva, and Williams, 2001).

Water Quality Analysis
Water quality tests were conducted at the main watershed and sub-watershed levels. Eight sampling points were selected to study the impact of watershed land use on river water quality. Seasonal sample collection was conducted in two consecutive years and a total of 128 water samples were collected during the two major seasons of the years 2016 and 2017. Wet season samples were taken between June and August while dry season samples were taken between January and March.
Water samples were collected at a depth of 20 cm below the water surface and at 1.5 m distance from the edge of the land surface into the mid of the river using a 2L polyethylene plastic bottle. The bottles were rinsed very well with sample water and samples were taken by dipping the opening of the sampler upslope. The water samples were stored in iceboxes until the analyses were conducted.
Samples were taken to Addis Ababa Environmental Protection Authority laboratory within 4 hours after collection and analyzed for selected water quality parameters.
Acid-cleaned polyethylene gloves were used while handling all materials during sample collection and analysis. A total of 10 parameters were selected to assess the pollution level of the river water. These were pH, EC, TDS, TSS, BOD, COD, Total Nitrogen, Total Phosphorous, Ammonia, and Chloride. All the parameters were analyzed in the laboratory involving standard experimental protocols as outlined in APHA (2005). The pretreatment and the determination of the parameters in the laboratory followed national standard methods of examining water and wastewater, which were provided by the Environment, Forest and Climate Change Commission, Ethiopia. The values of the parameters that are presented in this paper are the means of composite analyses. The mean values of the water sample test results were calculated and compared to international and local water quality standards for various purposes.
Furthermore, the pollution level of Kebena river water was examined.

Statistical Analysis
Statistical analyses were performed to study the relationships between land-use variables and water quality parameters. A one-way ANOVA test was used to compare mean values of water quality parametersamong the sub-watersheds for different water quality indicators for the dry and wet seasons. Independent Test was conducted to compare the mean values of the different water quality indicators for the wet and dry seasons. Multiple regression analysis was applied to determine whether watershed land use factors have positive or negative influence on separate water quality indicators and how strong this interaction is using three predictor land use variables (agricultural, forested, and urban land use variables).

Watershed Land Use types and distribution
Kebena river watershed accounts for a total area of 5,247.60 ha of land of which 39.14% is forest composed of Eucalyptus globulus plantation and riverine vegetation. The second dominant land use in the watershed is the settlement or built-up area covering 32.51%. Thirdly, the Kebena watershed contains cultivated land use accounting for 27.25% of the total area, while the remaining 1.1% of Kebena watershed is considered as bare land consisting of bare soil and rocks.
Along with the other sub-watersheds, the agriculture dominated sub-watershed (Denkaka sub-watershed) covers a total area of 2474.20 ha of land, out of which 44.90% is eld crop, 34.14% forest, 19.17% built-up area, and 1.78% bare land. Likewise, the forest dominated sub-watershed in the middle (Little-Kebena sub-watershed), is predominantly covered with 60.87% of natural forest mixed with Eucalyptus globulus plantation covering a total area of 1010.10ha (Table 1). The second-largest land use within the subwatershed is the built-up area which covers 21.50% of the total land followed by cultivated land and bare land accounting for 17.00% and 0.63%, respectively. In addition to the dominant land use type, which is forested area, this sub-watershed is covered by a highly congested settlement, largely occupying the lower part of the sub-watershed covering areas of the riverbanks. The urban dominated sub-watershed with a total area of 387 ha of land accommodates 90.44% of built-up area, 7.49% of eld crop including urban farms, 1.5% of forest including parks and riverine vegetation, and 0.52% of bare land (Table 1).

Seasonal Variations in Water Quality of the Kebena watershed
The result of the analysis showed that there are signi cantly high seasonal variations in the concentration of the water quality indicators. As shown in Table 2 below, the river water carries signi cantly higher loads of pollutants during the dry season than in the wet season (P<0.001, P<0.05). Accordingly, mean values of all parameters except TSS, COD, and BOD have variation at P-Value of at 0.000.
The mean values of TSS, EC, Ammonia, and TN are signi cantly higher during both the wet and dry seasons compared to surface water quality standards set by WHO, which is 25 mg/l, 0.60 -0.18µs/cm, 1.5 mg/l, 10 mg/l, respectively. On the contrary, Chloride and COD showed signi cantly lower values in the wet season compared to surface water quality standards set by UNEP, and WHO which is 200-300mg/l and 250mg/l, respectively (Table 2).

Seasonal and spatial variations of water quality in the sub-watersheds
In aquatic life, pH is an essential chemical limiting factor if excessively acidic or alkaline streams can interfere with aquatic organisms' biochemical reactions, which either harm or destroy stream organisms (Dumago et al., 2018). The results from this study show that in the wet season, the mean pH level at the source of the river within the forest watershed is 6.73. The lowest mean value of pH was recorded in the agriculture dominated Denkaka sub-watershed, which is 5.64, followed by forest dominated little-Kebena sub-watershed with a value of 5.92 and urban dominated Gin e sub-watershed with a value of 6.64. Besides, in the dry season, the mean value of the river water at the source is 7.04, whereas lower pH value was reported to be 7.65 in the urban dominated sub-watershed of Gin le, 8.00, in forest dominated sub-watershed of (Little-Kebena) and 8.08 in agricultural dominated sub-watershed of Denkaka.
Electrical Conductivity (EC) is a capacity of how easily electrical current can pass through water, hence depending on the presence of charged ions, whereas Total Dissolved Solids (TDS) contain mineral and salt impurities dissolved in water (Moore, Richards, and Story, 2008;Dumago et al., 2018). These impurities, when present in high quantities, affect the quality of surface waters to provide various ecosystem services. Total Suspended Solids also affect water color due to higher water TSS concentration, higher turbidity worsens water clarity (Dumago et al., 2018).
Accordingly, the result showed that EC and TDS values in the three catchments and Kebena watershed are higher with no noticeable variations during the dry season, and the ANOVA result also reveals (F=0.79 and 0.486 for EC and TDS, respectively, while p-value is > 0.05).
However, the TSS value displays substantial variability with F=25,56 and p-value < 0.01 during the dry seasons. On the other hand, during the wet season, the ndings of ANOVA showed that the mean values of both EC, TDS and TSS display major differences across the three catchments and Kebena with F = 122,51, 35,87 and 22,77 respectively, and p-value < 0,01).
COD is the amount of total oxygen needed to oxidize into carbon dioxide and water both biologically accessible and inert organic matter (Lee and Nikraz, 2015), mainly used to determine the organic pollutant present in the river water. BOD on the other hand is the calculation of the amount of oxygen biologically required to break down organic matter under aerobic conditions (Lee and Nikraz, 2015), which is a widely used for measuring organic pollution of surface waters. The results, therefore, indicated that at the source of the Little-Kebena catchment, the value of COD accounts for 14 mg/l and 12mg/l while the mean value of BOD is registered to be 2.4 mg/l and 10.8 mg/l during the wet and dry seasons, respectively.
The value of COD increase to 55 mg/l and 34mg/l, during the wet and dry seasons, respectively, at the lower end of the catchment.
Similarly, the value of COD at the lower part of Denkaka catchment increases in both the rainy (39mg/l) and dry seasons (46mg/l). Moreover, at the Gin e catchment, the value of COD is registered to be 114.00 mg/l during the rainy and 157.00 mg/l during the dry seasons. The BOD value also increases for the two catchments during the dry season, 40.2mg/l for Little-Kebena and 23mg/l for Denkaka catchments, however, the lowest value is registered for Gin e catchment which is 12.1mg/l.
The ANOVA result also shows the signi cant variation between the three catchments, Kebena watershed and COD results during both the wet and dry seasons with F = 165.74 and 287.12, respectively and p-value < 0.01). Furthermore, the ANOVA result revealed signi cant differences between the three catchments, Kebena watershed and BOD values during the wet and dry seasons with F = 287.12, and 754.83, respectively and p-value <0.001.
During the wet and dry seasons, the mean value of Ammonia at the source of Little-Kebena catchment was found to be 1.25, and 0.001 mg/l, respectively. During the wet season, higher value was registered at the Gin e catchment with 18.25 mg/l, whereas during the dry season, higher level of Ammonia 138.1mg/l was recorded at Gin e catchment. The ANOVA result also shows the signi cant disparity between the three catchments, Kebena watershed and Ammonia results during both the wet (F = 648.15, p-value <.01) and dry (F = 32.94, p-value <0.001) seasons.
According to the results, the mean values of Total Phosphorous (TP) at the source of Little-Kebena are found to be 0.5mg/l and 8.6mg/l during the wet and dry seasons, respectively. While the results of Total Nitrogen (TN) are registered to be 14 mg/l and 9 mg/l during the wet and dry seasons, respectively. The results of TP showed an increase towards the lower part of Little-Kebena and Gin e catchment while decreasing in Dekaka catchment. With regards to TN the value was found to be lower during the rainy season in all the three sub-watersheds with a mean value of 4.1, 3, and 10mg/l in Little-Kebena, Denkaka and Gin e sub-watersheds, respectively.
During the dry season, these values increased intensely in all three catchments and Kebena watershed whereby the highest mean value was recorded at Gin e catchment with a value of 129mg/l.
The ANOVA result also shows the signi cant relationship between the three catchments, Kebena watershed and TP results during wet seasons with (F = 46.65, p-value < 0.001), where the result of Gin e catchment was found to be very high with 1.55mg/l. During the dry season The ANOVA result also con rmed to this signi cant variation with (F = 11.58, p-value < 0.001), where a high value was recorded from Gin e catchment with 16.09 mg/l. Furthermore, the ANOVA results con rmed the signi cant differences between the three catchments, Kebena watershed and of TN during wet seasons (F=47.05, p-value < 0.001), and dry seasons with (F = 25.77, p-value < 0.001).
The effect of Chloride concentration or conductivity in surface waters is considered as general urban impact indicator (Paul and Meyer, 2001). The result of the Chloride analysis at the source of Little Kebena was found to be 22.2 and 10.4 during the wet and dry seasons respectively. The high level of Chloride 102.8mg/l during the dry season is registered in the Denkaka catchment while the lowest value was recorded to be 79.2mg/l at Gin e catchment. During the rainy season, a high amount of Chloride, i.e., 78.8mg/l was recorded in Gi e catchment while the lowest was found to be 36.61mg/l at Denkaka watershed. The ANOVA results also revealed the strong differences between the three catchmenets, Kebena watershed and Chloride at (F=110.34 and p-value <0.001) during the wet and at F= 35.66 and p-value <0.001) during the dry seasons

Watershed land use and water quality indicators
Compared to water quality indicators at the source of the forest dominated catchment, the Urban dominated (Gin e) and the forest dominated (Little-Kebena) sub-watersheds appeared to have the greatest effect on water quality during both the wet and dry seasons than the agriculture dominated (Denkaka) sub-watershed (Table 4). During the wet season, the Gin e sub-watershed showed a strong positive relationship with all the water quality indicators except pH. While, forest dominated sub-watershed showed a positive relationship with all the water quality indicators except pH, TN, and TP. In contrast, TN showed a negative relationship with all the watersheds at P < 0.01 level, with R 2 value of 0.86. Whereas TP showed a strong negative relationship with all the sub-watersheds except Gin e with R 2 of 0.76, and pH showed a strong negative relationship with (Little-Kebena) and Denkaka sub-watersheds at s P < 0.1 level.
The urban dominated sub-watershed appeared to have the greatest effect on water quality (Table 4). The TDS, TSS, EC, Ammonia, TN, TP, Chloride and COD showed strong positive relationship at P < 0.01. The pH showed a positive relationship at P < 0.05, while BOD showed no relationship. Forest dominated sub-watershed showed higher value of water quality indicators during the dry season in which the pH, TDS, EC, Ammonia, T-N, Chloride, COD, and BOD were strongly and positively related ( Table 4). The agriculture dominated sub-watershed though revealing lower water quality indicators during the dry season, still exhibits a strong positive relationship with pH, TDS, EC, Ammonia, TN, Chloride, COD, and BOD at P < 0.01 level.
In general, the Kebena water quality result at the end of the watershed when compared with the source revealed higher more serious effect during both the wet and dry seasons since all the water quality indicators showed strong positive relationship except BOD and pH. The latter two parameters were found to be negatively related with the watershed's land use.

Seasonal variations of water pollution in the Kebena River
The water quality in the Kebena River was found to be deteriorated during the wet and dry seasons due to pollutants originating from various sources in the watershed. The results in this study are in agreement with earlier studies in the watershed (AACSPPO, 2017; Worku and Giweta, 2018). The clear deterioration in the dry season water quality suggest that perhaps the pollution sources are mainly from point sources located at different parts of the river course (Worku and Giweta, 2018). This is possibly true because in most of the places along the river course, liquid wastes from nearby residential settlements and other land uses are directly connected to the river through open ditches and road side drainage lines, aggravating the impact during the dry season as there is no rain to dilute the river water.
During the wet season, the mean value of pH (6.23) of Kebeba river was below the range of the permissible limits which is 6.5-8.5/9, (DENR, 2016;WHO, 2004;EPA, 2003) showing mainly of acidic character assumed to be resulting from the heavy rain during the wet season. This limits the potential of the river water to deliver various ecosystem services during the rainy seasons. However, during the dry season, the average pH value of the river water is 7.87 which makes it suitable to provide various ecosystem services (WHO, 2004).
The fact that mean values of TSS, EC, Ammonia, TP, and TN are signi cantly higher during both the wet and dry seasons compared to surface water quality standards is assumed to be associated with high amount of direct sewage discharges (U. S. E.P. A., 2002;Gibb, 2000). Furthermore, the value of EC is highly affected by temperature which validates the higher value of EC in Kebena river during the dry season (Moore, Richards, and Story, 2008). Large nutrient loads of nitrogen (TN) and phosphorus (P) can have somewhat speci c routes, partially due to land use, building drainage and impermeable paved surfaces, and leaf litter dropping from cultivated trees, (Teurlincx et al., 2019;). The potential sources of phosphorus specially during the dry seasons in urban catchments might be related to wastewater disposals, and in-stream bathing and laundry activities (Dumago et al., 2018;Pieterse, Bleuten and Jørgensen, 2003;Paul and Meyer, 2001).
On the contrary, Chloride and COD showed signi cantly lower values in the wet season compared to surface water quality standards, this might be related to the diluting effect of the rain water during this season which affects the concentration of salts and other minerals (Giglioli, and King, 1966).

Relation between watershed land use and water quality
The study also compared water quality indicators from source of the forest-dominated catchment of the Little Kebena River with samples collected from the lower point of Little-Kebena watershed itself, the urban-dominated Gin e catchment, and the agriculturaldominated Denkaka catchment in order to identify the catchment with a higher effect on the water quality of the river. Accordingly, the result revealed the strong positive relationship between the urban dominated catchment land use and various water quality indicators such as, Total Nitrogen, Total Phosphorous, Chloride, COD and BOD, are observed during both the dry and wet seasons. The high amount of TP observed in Little-Kebena and Gin e catchments during the wet seasons, therefore, mainly is the result of the existence of large impervious surfaces coverage within the catchments. Therefore, the water quality of Gin e river revealed high water pollution during both the wet and dry seasons, which is highly supported by many studies that concluded that water quality of rivers found in highly urbanized watersheds are strongly positively related to various water quality parameters (Brabec, 2009;Sliva and Williams, 2001). According to earlier studies, high runoff in urban watersheds is directly related to high watershed imperviousness; Schueler (1995) and Klien (1979) conversed that the degradation of the river water quality is extreme to the point where it is almost unavoidable at 30% of the imperviousness of watersheds.
Contrary to most researches such as Ding et al., (2015), Sliva and Williams (2001) the forest dominated sub-watershed with 60.87% of forest cover indicates a positive relationship with a number of water quality indicators such as Chloride, COD and BOD, and TSS during the wet season while pH, TDS, EC, Ammonia, TN, Chloride, COD and BOD show strong positive relationship during the dry season. The strong positive relationship of the river water with the above mentioned indicators could be attributed to the discharge of liquid waste from closely developed settlements in the middle and lower catchments of Little Kebena River, as discussed by (Brabec, 2009); whereby the location of development within a given watershed plays an important role in water quality variability in addition to the percentage of watershed imperviousness. The structural plan also con rmed that some parts of the upper watersheds are occupied by informal settlements and industries (AACSPPO, 2017). Land degradation is also another major challenge aggravating soil erosion, specially within the Eucalyptus plantation (AACSPPO, 2017) which explains the large amount of TSS and TDS in Kebena River water.
The agriculture dominated sub-watershed with 19.17% and forest dominated sub-watershed with (21.50%) of their land being built-up, lie within the imperviousness cover ratio of 12-30% where the impact of the watershed is considered as minimal. Thus, controlling further impervious surface increment of the watersheds, while delineating proper river buffer zones, guiding upcoming developments, and implementing proper water pollution instruments is signi cantly crucial for the improvement of Kebena river water quality.
Planning documents along with Building Permit Directives can be used as tools to guide and monitor development and imperviousness ratio of institutions and individual compounds.

Planning Implication to water pollution mitigation in the urban settings
The City Structural Plan con rmed that high pollution of river waters emanates from domestic and industrial wastewater discharges as well as storm runoff, thus, suggested various ways to prevent pollution of rives (AACSPPO, 2017). In line with this, the structure plan outlined that one of the effective ways to lower point source pollution in ux to Kebena river catchments is by introducing new and upgrading of existing Waste Water Treatment Plants (WWTPs) (Paul and Meyer, 2001;Björklund, 2015). Furthermore, the structure plan delineated river buffer zone around the rivers of Addis with the width of 10-30m to be covered with indigenous trees and shrubs, the structure plan also forwarded a way of considering the slope of the buffer zone and various ways of developing the sloped riversides (AACSPPO, 2017).
However, the suggested river water pollution prevention methods lack a watershed level perspective that will contribute to the fragmented and site-speci c river management gaps of the study area. More importantly, the proposed environmental solutions of the Structural Plan are not synergized with the current and proposed land use, road networks and other planning features. This highly demeans the contribution to the planning decisions to prevent the river water pollution which is one of the main challenges of the city.
Watershed planning, therefore, becomes as one of the vital tools to addresses water quality problems in a holistic manner by fully assessing the potential contributing causes and sources of pollution, then prioritizing restoration and protection strategies to address these problems (Ding et al., 2015;Arnold and Gibbons, 1996).
In the case of the Kebena river watershed, different watershed level planning approaches are required for the three sub-catchments so as to address the varying issues. For instance, in Gin e catchment where 90% of the watershed is built up, there is signi cant need for a comprehensive watershed planning that emphasizes on preventing further river degradation while focusing on site level design that reduces runoff and imperviousness, (Arnold, and Gibbons,1996). Designing riversides in a manner that removes and cleanses much of the stormwater pollution that comes from the city before it joins the river is vital (Prominski et al., 2015). Furthermore, planning solutions should emphasize on emission reduction and enhancement of ecosystem conservation, as one of the Dublin Principles of 1992 (ICWE, 1992) which concerns river basin management at the lowest appropriate level (Kemper et al., 2007). In line with this, it is necessary to remove polluting land uses, incorporate open ditches, small streams, and roads with environmentally suitable storm water management network, and establish adequate river buffer while reconnecting green patches within the catchment. Furthermore, it is crucial to incorporate water pollution mitigation schemes into urban planning decisions through provision of compact development approach, and brown eld development approaches to reduce natural landscape composition fragmentation (Xu et al., 2019), and riparian buffers development to control further increment of impervious surfaces (Xu et al., 2019;Chouli, Aftias, and Deutsch., 2007;Mander et al, 1997).
With regards to Little-Kebena catchment, where there is lower imperviousness ratio, river pollution mitigation measures should focus on preventive measures to retain the existing natural green spaces through open space planning, development and management schemes (Arnold, and Gibbons,1996). Moreover, location of development within the catchment plays a signi cant role in affecting water quality in an urbanized watershed whereby upland land use are critical in determining overall stream function, degradation, and rehabilitation potential Booth and Jackson (1997) and Barbec (2009). Thus, in this catchment, there is a possibility of extending the forest in the upper catchment to the lower catchment by limiting the various land uses that are polluting the river.
In Denkaka catchment, on the other hand, there is a crucial need for the provision of large river buffer zones to control runoff pollution coming mainly from crop elds, and settlements developed close to the river.
While developing watershed planning is crucial to preventing urban river pollution, its contribution will be limited if not adequately integrated with urban plans to facilitate its implementation and ensure its sustainability (Bernhardt and Palmer, 2007). Therefore, the ndings of this study suggest that land use planning need to be combined with scienti c analysis of land uses and possible sources of pollutants in order to achieve effective restoration and maintenance of river water quality.

Conclusions
The ndings have indicated that Kebena river watershed, has a highly polluted river water, which limits its potential to deliver various ecosystem services. The overall nding revealed that pollutants originating from various sources ( The ndings in this study can be considered as an addition of new knowledge to the river restoration research in order to understand how watershed landuse impacts river water quality in urbanizing cities. It may further indicate that small projects in the future, if not well studied and integrated at watershed level may not be effective in protecting river water pollution in the City and in the country at large.
The ndings in this study can also be considered as an addition to policymakers in their decision-making process on how to integrate the crucial role of watershed planning into overall city land use planning activities in general and surface water quality issues in particular. Finally, future research should focus on developing tools, which help to transfer the main principles of watershed planning into practice. Further research can be conducted to identify and integrate the impact of landscape characters of sub watersheds such as slope, vegetation, and soil characters in addition to land use. Such studies will signi cantly advance our understanding on the integration of watershed planning principles into current river restoration activities in the country in particular and more widely in similar urban settings elsewhere.

List Of Abbreviations
If abbreviations are used in the text are de ned in the text at rst use.

Declarations
Ethics approval and consent to participate "Not applicable" Consent for publication "Not applicable" Availability of data and materials All data and materials applied for the research are presented in the main paper.
Competing interests "The authors declare that they have no competing interests".   The reference water quality indicator is Forest source point ** and * are signi cant at 1% and 5% levels, respectively.