Abating eutrophication on urban lakes: a case study of Kabaka's Lake, Uganda

Urbanisation land use changes in cities around the world have reduced the natural pollutant filtration value of the wetlands in their catchments. This has led to the gradual eutrophication of available water bodies (Costa et al. 2018). Eutrophication, an indicator of high nutrient concentrations, and development of algal biomass, leads to water quality decline through discolouration, foul smells and tastes, diurnal pH changes, temperature changes, depletion of dissolved oxygen (DO) and degraded aesthetic values (Leng 2009). The effects of eutrophication are often realised while treating water for drinking, and through its negative impact on recreation and health. For instance, there is a link between eutrophication and the formation of trihalomethanes and other chlorinated organics in drinking water. Eutrophication has been associated with parasitic diseases of amphibians and carcinogens in humans (Paerl & Otten 2013). Cyanobacteria, the most important phytoplankton associated with harmful algae bloom, are known to be poisonous to animals and humans, and responsible for compounds like methylisoborneol and geosmin that cause off-flavours in municipal water systems, as well as in aquaculture-raised fish (Watson et al. 2016; Chorus & Welker 2021).

The challenges associated with eutrophication are mostly felt in developing countries, especially those in the tropics, where steady solar input and high temperatures sustain algal blooms throughout the year (Wells et al. 2015). There is an urgent need to use previously developed mitigation options, strategies, techniques and measures, to enable accurate assessment and effective long-term control of lake eutrophication. Extensive field research and process-based modelling of individual catchments, have improved understanding of the source and transport of nutrients from land to water, around the world. For example, the effects of factors that relate land use to water body quality have been quantified (Soranno et al. 2015). Statistical models have been used to: predict water quality conditions resulting from potential changes in lake operations and/or environmental conditions, conduct scenarios with respect to forecasting nutrient loads under different catchment management strategies, simulate thermal stratification and evaluate the trophic status in reservoirs (Dwarakish & Ganasri 2015) This has provided useful information for developing abatement management strategies for priority water quality management (McCutcheon 1990; Zyfi et al. 2014). The HELCOM Baltic Sea Action Plan (BSAP), used to combat eutrophication in a marine environment, relies on the integrated ecosystem approach to achieve good environmental status and sustainability, by setting eutrophication indicator targets (Jetoo 2018). Studies on eutrophication abatement external measures – advanced wastewater treatment, wastewater or runoff diversion, constructed wetlands, laws and regulations, and buffer strips – and internal measures such as chemical control, biomanipulation, dredging, plant harvesting, etc, are limited in developing countries. In this study, quantitative methods and multi-criteria assessment were applied (Kasim 2015) to evaluate strategies for sustainable management of eutrophication effects on water quality in Kabaka's Lake, Uganda, a typical urban lake catchment.

The removal of the lake's watershed has left the lake in need of the former wetland's primary ecosystem service, including; naturally cleansing received human waste, processing nutrients in the water, and releasing ‘filtered’ wastewater downstream – i.e., from which the bacteria and sediments have largely been removed. Loss of the wetlands has impacted recent disease outbreaks, including malaria and diarrhoea, in the area, degradation of the living environment, and damage to the lake's ecosystems. As a consequence, eutrophication abatement strategies are needed to address the lake's deteriorating water quality, due to the increasing change in land activities within its catchment/watershed.

Assessing the eutrophication abatement strategies for Kabaka's Lake necessitated the classification of the catchment's land use activities. The Multispectral Landsat satellite images for 1995, 2003 and 2019, produced by USGS in GeoTIFF format were downloaded from https://earthexplorer.usgs.gov on 1 May 2019. Using the Landsat natural and standard false colour [band 5 (SWIR 1), band 4 (NIR), band 3 (red), band 2 (green), and band 1 (blue)], composite images were generated to achieve a general description of LU/LC changes. GIS was applied to process the data, extract clipped Landsat scenes from the delineated catchment extent, and produce change detection maps. Areas showing the various defined land-covers were digitised from the aerial photographs and topographic maps.

To assess the lake's nutrient load, data from the Kabaka's Lake report (AWE 2017) were used while samples were collected at various depths, in shaded areas, from 7 points in the lake, in February and March 2019, using a Vendome 78-300 Fieldmaster. Measurements of pH and DO were taken immediately after sampling, using portable water quality metres (Hanna HI991003, USA and Milwaukee MW600, USA). The samples were then transported in a cool box to Makerere University Public Health and Environmental Engineering Laboratory for analysis. Total nitrogen and total phosphate were determined using the cadmium reduction and ascorbic acid method, after digestion with persulphate (APHA 2012), and final readings were made with a HACH DR/4000 spectrophotometer (USA). Chlorophyll-a was determined using the fluorometric method at the National Water and Sewerage Corporation laboratory.

Two abatement management strategies were assessed, constructed wetlands and dredging. Their selection was guided by an integrated ecosystem approach aimed at achieving good environmental status and sustainability (Jetoo 2018). First, a number of options proposed by (Kasim 2015) were evaluated (Table 1), and control technique longevity was noted as the most important criterion for eutrophication management. The increase in lake sedimentation, benchmarked on the 2017 bathymetric study, and the degraded wetland upstream of the lake's inlet and its attribute of reducing the lake inlet nutrient concentrations effectively, supported the ranking decision. On that basis, wetland construction (as an external measure) and dredging (internal control measure) were ranked best for this study.

Google Earth historical satellite imagery of the lake's receiving wetland footprint for the period 2002–2019 was used to delineate the receiving wetland footprint. Nutrient reduction efficiency ranges of 40–60% for TN and 30–40%) for TP were adopted, based on a number of studies (Mthembu et al. 2013; Ilyas & Masih 2017). The ranges were benchmarked from studies of constructed wetlands in Italy and Uganda (Okurut 2000; Foladori et al. 2013).

The lake's volume and depth were established to enable determination of the amount of dredging required. The lake's total area was sub-divided into 30 × 30 m grids using Arcmap. A handheld depth finder (H22FX Handheld Sonar System with LED Flashlight) was used to take water depth readings at various points along the grid lines, from which data were input to GIS software for bathymetric modelling.

The Trophic State Index (TSI) index – range: 0–100 – was used to assign a trophic state ‘grade’ to the lake (Table 2, Equations (3)–(8)). on the assumption that the current lake status is nitrogen limited, the relationship between total phosphorus and chlorophyll was established using Equations (9) and (10) for prediction modelling. A multiple linear regression (MLR) model (Equations (11) and (12)) was run to forecast the TSI of the lake for the next 30 years and determine the significance of the management strategies in eutrophication abatement, and validated through smoothing and de-seasonalising at 0.0001 tolerance (Carlson 1977). The model was used to predict the lake's TP and TN loads, and TSI for a 5-year range period (5, 10, 15, 20, 25, and 30 years), and an average end method (Equation (13)) was used to estimate the likely dredge volume.

Table 2

TSI equations

Secchi disk		(3)
		(4)
Chlorophyll-a		(5)
TP		(6)
TN		(7)
Nitrogen limited lakes (TN/TP) < 10)		(8)

Secchi disk		(3)
		(4)
Chlorophyll-a		(5)
TP		(6)
TN		(7)
Nitrogen limited lakes (TN/TP) < 10)		(8)

View Large

Green algae thrive in the lake's environmental conditions – pH > 8.5, low salinity (0.13 ± 0.00 mg/L), temperature of 25–35 °C – and are responsible for the lake water's green colour (874 ± 210 PtCo), hence exposing the lake to eutrophication enrichment. A Secchi disk transparency variation (0.5–0.1 m) was obtained along the lake's profile (Figure 5(b)), indicating a eutrophic factor. The combination of a large surface area and relatively shallow depth means that the lake does not react homogeneously, with mixing occurring at different times and to different degrees in different places. Biological accumulations in the lake have influenced fluctuations between TSI (SD = 18.66 and Chl-a = 18.65) and TSI (TN = 24.88 and TP = 20.146), confirming the dominance of algae in light attenuation. This is due to the similarity between the Secchi disk and Chlorophyll-a TSI values [TSI (Chl-a) = TSI (SD)], and the nitrogen factor limiting the algal biomass, confirmed by [TSI (TP) > TSI (Chl-a) = TSI (SD)], as argued by (Brown & Simpson 2001). Florida Department of Environmental Protection (FDEP), would classify a lake with a TSI of 21.57 as oligotrophic (TSI < 40) due to its low nutrient concentration. it has the potential, however, to support the highest level of biological productivity (e.g., an abundance of algae, aquatic plants, birds, fish, insects, and other wildlife).

Regression modelling was performed to predict the lake's TP and TN loads, and TSI for a 5-year range period (5, 10, 15, 20, 25, and 30 years). A 63% root mean square model fitting the data with fit goodness of minimal residual sum of squares (3.439) and acceptable P values (TN = 0.005492 and TP = 0.000139) was obtained. The model's normality standardised residual plots for TN and TP followed a normal distribution (spread homogenously along the line [y = 0]) validating its reliability. Plots show a reasonable correlation between the observed and predicted values. The model predicted an average annual increment of 1.14 (5.2%) in nutrient pollution load for the next 30 years.

The increase in wetland encroachment has translated into a gradual loss of nutrient reduction efficiency for the lake. Modelling of the lake's nutrient load over time shows that with the current 0.54 hectares of natural wetland, significant TP (p=0.0102) and TN reductions (p=0.000083) will continue to be noticed over the next 25 years. This implies that, if the current natural wetland is not destroyed, it will continue reducing the nutrient input load effectively for the next 25 years, after which the lake will be eutrophic, while the construction of a wetland in the catchment will increase the nutrient reduction efficiency over time.

The study's findings suggest that Kabaka's Lake's eutrophication is attributable largely to anthropogenic disturbance in the catchment, where increased human settlement has increased the lake's potential nutrient loading. The findings show that the lake is not yet eutrophic but is oligotrophic. With the current state of activities in the catchment, this will remain the case for about the next 25 years. Constructing a wetland with the current footprint will reduce the anticipated nutrient load significantly and sustain the lake's trophic status. Dredging the lake will not only increase its storage capacity but also enable thermal self-cleansing to remove the green colour. These findings are strong justification for watershed management as a sustainable solution to control nutrient flow into the lake.

The authors acknowledge AWE for the data used in this study. The authors acknowledge that this paper was partly based on an MSc study by Mutyaba (2019) under the supervision of A.N. and K.O.

All authors were involved in the study's conception. A.M. and A.N. were further involved in data acquisition, analysis, interpretation, and manuscript drafting and revisions. C.H. and K.O. reviewed and made comments on the draft manuscripts. All authors reviewed and approved the manuscript for submission.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.