Promoting Green Urban Development in Africa: Enhancing the relationship between urbanization, environmental assets and ecosystem services A PRELIMINARY INVESTIGATION OF THE POTENTIAL COSTS AND BENEFITS OF REHABILITATION OF THE NAKIVUBO WETLAND, KAMPALA Promoting Green Urban Development in Africa: Enhancing the relationship between urbanization, environmental assets and ecosystem services A PRELIMINARY INVESTIGATION OF THE POTENTIAL COSTS AND BENEFITS OF REHABILITATION OF THE NAKIVUBO WETLAND, KAMPALA Authors Jane Turpie, Liz Day, Dambala Gelo Kutela, Gwyneth Letley, Chris Roed and Kat Forsythe Prepared for AECOM on behalf of The World Bank and the KCCA Prepared by Anchor Environmental Consultants 8 Steenberg House, Silverwood Close, Tokai 7945 www.anchorenvironmental.co.za in association with Freshwater Consulting Group 2016 COPYRIGHT © 2016 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. September 2016 RIGHTS AND PERMISSIONS The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to the Publishing and Knowledge Division, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org.                                                  Page iv Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala PREFACE AND ACKNOWLEDGEMENTS This study forms one of the case studies of a larger study on Green Urban Development commissioned by the World Bank and led by AECOM. Anchor Environmental Consultants (Anchor) was subcontracted by AECOM to undertake case studies in three cities: Kampala (Uganda), Dar es Salaam (Tanzania) and Durban (South Africa). Each city was consulted as to the focus of the case study. In the case of Kampala, the city requested a study to evaluate the potential costs and benefits of rehabilitating the Nakivubo wetland to enable the development of a recreational park. The study was led by Dr Jane Turpie of Anchor Environmental Consultants (Anchor). Dr Liz Day of Freshwater Consulting Group undertook the ecological and restoration aspects of the study, assisted by Kat Forsythe of Anchor and by Chris Roed of West Coast Engineering on the engineering and costing aspects. Gwyn Letley of Anchor and Dr. Dambala Gelo Kutela of the University of Cape Town handled the econometric analyses of the water treatment costs and recreational valuation studies, respectively, while Gwyn Letley also undertook the cost-benefit analysis. Dr. David Kyamboto of Makere University and his graduate students assisted with the household survey. We are grateful to Roland White and Chyi-Yun Huang of the World Bank, Diane Dale, Brian Goldberg and John Bachmann of AECOM, and Timm Kroeger of The Nature Conservancy for inputs and discussions during the project planning phase, as well as to the inputs received from reviewers Jeff Wielgus, Mike Toman, Urvashi Narain and Glen-Marie Lange of the World Bank. We are also grateful to the Kampala Capital City Authority (KCCA) for their interest and support of this project - in particular to Najib Lukooya Bateganya for assistance with meetings and data collation. We are also grateful to Susan Namaalwa and Stephen Tumwebaze of the National Water and Sewerage Corporation (NWSC) for provision of water treatment works data and for a tour of the Ggaba water treatment works. Thanks also to the citizens of Kampala who willingly participated in the household survey.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page v This page intentionally blank.                                                  Page vi Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala EXECUTIVE SUMMARY Rapid urbanisation threatens existing natural areas within An estimate was made of the design requirements and cities and the ecosystem services that they provide. This costs of the interventions outlined above in order to case study forms part of a broader study that investigates inform a high level cost-benefit analysis. These took the benefits of investing in Green Urban Development existing commitments to sanitation infrastructure in African cities. The Kampala case study focuses on the improvements into account. The overall estimated Nakivubo wetland, one of several large wetland systems capital costs amounted to $53 million, with ongoing that occur within and around the city. This wetland has maintenance and operating costs of $3.6 million per become severely degraded by polluted water from the annum. In addition, the cost of forgone agricultural city that passes through the wetland before entering production in the wetland area was estimated to be Inner Murchison Bay. In the late 1990s, it was ascertained $141,500 per year. that the water treatment service performed by the wetland yielded a significant cost saving for the nearby The benefits associated with wetland restoration that Ggaba Water Treatment Works. However, as the city were included in the cost-benefit analysis were water has continued to grow, pollution flows into the wetland treatment cost savings and recreational benefits. have increased significantly, the size and assimilative It was estimated that a reduction in phosphorous capacity of the wetland has decreased, and the costs concentrations in the lower wetland as a result of of water treatment have increased. These concerns, as the restoration interventions would have significant well as the increasing shortage of public open space impacts on the chemical costs of water treatment, areas in the city that are available for recreation, have amounting to a saving of some $1.143 million per year. led to the city’s consideration of the rehabilitation of the The overall recreational benefit was estimated using Nakivubo wetland, both to restore its functioning and survey-based revealed- and stated-preference methods, to create the opportunity for a recreational area with which suggested that the welfare gains to households associated possibilities for economic development. This in Kampala would potentially amount to about $22.05 study provides a preliminary evaluation of the state of million per annum. the Nakivubo wetland, the potential costs and benefits of its rehabilitation and the implications for the city’s The estimates derived in this study suggest that the expansion plans. interventions would have an overall net benefit. Using a discount rate of 6%, the net present value of the project Based on an analysis of the current functioning and over 15 years was estimated to be $80 million (Table 1). capacity of the Nakivubo wetland, a set of interventions This was further tested under varying assumptions of was identified that would be needed to restore the costs, benefits and discount rate. Under the worst case wetland to a level where economic benefits could scenario (upper bound estimates of costs, lower bound be realised. The primary objectives were defined as estimates of benefits), the Net Present Value (NPV) was (1) effecting a measurable improvement of water negative at discount rates of 6% and 9%. The internal rate quality passing out of the Nakivubo wetland into Inner of return (IRR) was estimated to be 20%, but sensitivity Murchison Bay, (2) ensuring sustainable management of analysis yielded a range of 4 to 34%. the Nakivubo wetland, (3) reducing water quality impacts on human health and (4) opening up opportunities Table 1 Results of the cost-benefit analysis for safe recreational use of the lower wetland. A sequential set of interventions (treatment train) was Net present value Discount Rate recommended to achieve these objectives, and included (US $ millions) both infrastructure upgrades and wetland rehabilitation and conservation measures, as well as investment in 3% 6% 9% recreational facilities: Best 220 158 83 Step 1. Prevent pollution at source through improved Base estimate 121 80 51 sanitation infrastructure and measures; Worst 4 -13 -24 Step 2. Prevent residual pollution entering the wetland; Step 3. Improve Waste Water Treatment Works (WWTW) effluent quality; Step 4. Rehabilitate the upper wetland for secondary waste water treatment / polishing; Step 5. Restore and protect the lower wetland; and Step 6. Establish recreational space and facilities.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 1 The following key messages emerge from the study: 4: This will require fundraising and co-ordination This study estimates that a total initial expenditure of at 1: Wetlands have limited capacity to solve urban least $53 million is required in order to achieve the goals pollution problems outlined, depending on the actual shortfall in sanitation Nakivubo wetland no longer has any positive impact on infrastructure. Given recent and planned expenditure the ecological condition of Inner Murchison Bay, with the in the city, the Nakivubo rehabilitation may only be lower wetland and bay having reached a hypertrophic feasible with a significant contribution of donor funding. state that is characterised by frequent, often toxic, algal Separating the issues and addressing sanitation first blooms, as well as being severely contaminated with will lower the risks and make funding for direct wetland pathogens that carry a risk to human health. Wetlands rehabilitation work more accessible. cannot substitute for wastewater treatment works and can only improve the quality of low volumes of A detailed plan will need to be drawn up. The actions moderately polluted water. described in this report pertain to multiple government institutions. In order for effective action to take place, 2: Rehabilitation will be costly but worthwhile coordination between these bodies is necessary. It will also As is often the case in the rehabilitation of natural be necessary for one of them, probably the National Water systems, the financial costs of the required interventions and Sewerage Corporation, to take the lead in this co- can be expected to be high. Many of these interventions ordination, with National Water and Sewerage Corporation could be regarded as obligations that should be met focusing on the bulk of the investments required including irrespective of the restoration of wetlands, for the treatment wetlands, Kampala Capital City Authority focusing mere purpose of providing essential services required on the investment in recreational facilities, and the Ministry for maintaining human health and dignity. While these of Natural Resources focusing on the management of the investments would not be particularly cost-effective in remaining natural wetlands. Similarly, there will need to be terms of the impacts on water treatment costs alone clarity regarding the responsibility and capacity of different (given the investment that has already been made in institutions to undertake the ongoing oversight and a new plant further afield), taken together with their management actions. potential recreational benefits, they are likely to be worthwhile. This is not surprising, given the dire lack 5: Avoid a repeat experience of managed green open space available to Kampala There are important lessons to be learned from this residents for outdoor recreation. As the economy of study. Considerable environmental and economic costs Uganda continues to develop and incomes improve, have been incurred by delaying investments that are the demand for these kinds of amenities is only set inevitably required. This potentially holds true for the to increase. Such benefits would also potentially be many wetland areas that are about to become engulfed reflected more tangibly in property values in the area. In by the growing city of Kampala. Kampala is the second addition, the restoration of wetlands and water quality in fastest growing city in Eastern Africa with an annual Inner Murchison Bay would be likely to have benefits for population growth rate of 3.9%. The population of the biodiversity and fisheries. Greater Kampala Metropolitan Area (GKMA) is projected to grow from just over 3 million in 2012 to 13 million 3: There are no shortcuts in 2040. Most wetlands within the existing urban area For the rehabilitation effort to be successful, have already been effectively lost. Without proactive implementation of all the interventions outlined in interventions, the wetlands outside of the present urban this study is required. The order of implementation core will also be destroyed and the cumulative impacts is not particularly important as long as they are all on Murchison Bay and any economic activities around implemented. For example, Step 2 could be implemented the bay, including the viability of a future waterfront as a stopgap while Step 1 is finalised. In addition, the development, could be significant. sooner that work begins on the wetland-edge filter strips and on wetland rehabilitation, the lower the risk of further human encroachment into the wetland when it comes to the implementation of this step, because people will not be able to settle or carry out agriculture in the re-established permanently wet areas.                                                  Page 2 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Thus the following recommendations should be considered: 1. Sanitation measures need to keep ahead of the growing urban population, and until this is the case, runoff from areas that are not well serviced needs to be diverted to a WWTW; 2. Water quality standards for waste water treatment need to be increased and enforced, and investments need to be made in the technological innovation required to deal with this; 3. There needs to be stricter regulation and control of industrial discharges. This should form part of a review into industrial effluent guidelines and current practices; 4. Legal protection of wetlands needs to be strengthened, and tough and/or innovative measures need to be taken to prevent the reclamation, farming and settlement of wetland areas; 5. Construction of roads and railways through wetlands should be avoided, or at least be done in such a way as to minimise impacts on hydro- and sediment dynamics and aesthetics; 6. Monitoring programmes need to be established, including continuous monitoring of flows into the wetland, and monthly measures of water quality of inflows and within all wetland areas in and around Kampala. Water quality monitoring data should be made readily available on an online platform for potential use by academic institutions, consultants, civil society etc. in order to encourage transparency and accountability as well as facilitate ongoing analysis and adaptive management. One of the main challenges in achieving the above would be institutional. Greater Kampala extends well beyond the boundaries of the KCCA, which originally encompassed the entire city, and unless the KCCA area is adjusted accordingly (as has been done in other countries), the problems that will arise in a growing city will be in areas under multiple other jurisdictions. Meanwhile, it may be more pragmatic for these issues to to be managed by the relevant national institutions such as NWSC.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 3 ACRONYMS AND ABBREVIATIONS B M BIC Bayesian Information Criterion ML Mega Litre BOD Biological Oxygen Demand MWTP Marginal Willingness to Pay BWWTW Bugolobi Waste Water Treatment Works N C NPV Net Present Value NTU Nephelometric Turbidity Units CBA Cost Benefit Analysis NWSC National Water and Sewerage Corporation CFU Colony-forming Units CL Conditional Logit COD Chemical Oxygen Demand CV Compensation Variation P CVM Contingent Valuation Method PAC Poly-aluminium Chloride PES Present Ecological State D DO Dissolved Oxygen R DWAF Department of Water Affairs and Forestry RPL Random Parameter Logit E S EAI Environmental Assessment Institute SBDC Single Bounded Dichotomous Choice EC Electrical Conductivity SD Standard Deviation SUDS Sustainable Urban Drainage System F FSTP Faecal Sludge Treatment Plant T TEEB The Economics of Ecosystems and Biodiversity TP Total Phosphorous G TSS Total Suspended Solids GIS Geographic Information System GUD Green Urban Development U USh Ugandan Shillings I IMB Inner Murchison Bay IRR Internal Rate of Return W WHO World Health Organisation K WTP Willingness to Pay WTW Water Treatment Works KCCA Kampala Capital City Authority WWTW Waste Water Treatment Works KSP Kampala Sanitation Programme L LCM Latent Class Model LV WatSan Lake Victoria Water and Sanitation                                                  Page 4 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala CONTENTS I. Introduction..................................................................................................................................................................... 9 II. Papyrus wetlands and their functioning......................................................................................................................... 11 III. The deterioration of the Nakivubo wetland.................................................................................................................... 13 IV. Current capacity for water treatment............................................................................................................................. 15 V. Interventions required to meet water quality and recreational objectives...................................................................... 16 A. Objectives.......................................................................................................................................................................16 B. Proposed interventions..................................................................................................................................................17 C. ...................................................................................................................................................................22 Overall costs. D. Cumulative effect on water quality................................................................................................................................23 VI. Potential water treatment cost savings.......................................................................................................................... 24 VII. Potential recreational benefits....................................................................................................................................... 26 VIII. Cost-benefit analysis...................................................................................................................................................... 27 IX. Conclusions and policy implications .............................................................................................................................. 30 X. References..................................................................................................................................................................... 32 Appendix 1. Selected water quality measures and their interpretation..................................................................................39 Appendix 2. Water quality amelioration by wetlands..............................................................................................................41 Appendix 3. Recent changes to Nakivubo wetland..................................................................................................................42 Appendix 4. Pollution inputs ...................................................................................................................................................46 Appendix 5. Current condition of aquatic ecosystems in terms of water quality....................................................................47 Appendix 6. Assessment of interventions required to meet water quality and recreational objectives.................................58 Appendix 7: Impacts of water quality changes on water treatment costs...............................................................................87 93 Appendix 8. Recreational demand for open space areas in Kampala and potential benefits of restoring Nakivubo wetland .                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 5 FIGURES Figure 1 Map showing the location and extent (1996) of the Nakivubo wetland in Kampala.....................................................9 Figure 2 A typical longitudinal section of the lower Nakivubo swamps showing the dominant vegetation types and related morphological features, not to horizontal scale. ............................................................................................11 Figure 3 Estimated historical extent (green) and current (red) extent of mostly natural vegetation within the Nakivubo wetland. ......................................................................................................................................................14 Figure 4 Estimated past and current capacity of the wetland for sedimentation /uptake of phosphorus, and estimated hypothetical removal if wetland habitats were still intact, with the volume of total phosphorus passing out of the system (i.e. “untreated” load)........................................................................................................15 Figure 5 Treatment train required to achieve water quality objectives in the lower Nakivubo wetland and Inner Murchison Bay, with brief summary of actions required at each step. (Note: the sixth step, the establishment of recreational facilities, is not depicted here).....................................................................................17 Figure 6 Areas for rehabilitation as filter strips (bright green areas) and treatment wetlands (light green areas) in the Nakivubo wetlands. . .........................................................................................................................................20 Figure 7 ..............................................................................21 Concept vision for the Nakivubo Wetland Park used in this study. Figure 8 Schematic diagram of the cumulative effect of the five intervention steps on the trophic state of the lower wetland area......................................................................................................................................................23                                                  Page 6 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala TABLES Table 1 Results of the cost-benefit analysis................................................................................................................................1 Table 2 Estimated capital costs and ongoing maintenance and operating costs of the recommended interventions, broken into component costs. The minimum and maximum range for cost estimates are included in brackets adjacent to the estimated cost value for each intervention. ......................................................................................22 Table 3 Results of the cost-benefit analysis..............................................................................................................................26 Table 4 Present value of costs and benefits under expected best case, base estimate and worst case scenarios (2015 US Dollars, 6% discount rate, 15 years)..............................................................................................................28 Table 5 NPV Sensitivity Analysis using discount rates of 3%, 6%, 9%.......................................................................................28                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 7 This page intentionally blank.                                                  Page 8 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala I. INTRODUCTION Urbanisation is taking place at an unprecedented rate In Kampala, much of the remaining natural habitat in Africa, with the rate of growth often outpacing comprises wetlands that have survived because of the plans and the capacity of city managers to provide high risk of flooding. However, these wetland areas have the necessary services. As a result, existing natural become increasingly degraded as a result of pollution, areas within cities that provide a range of ecological, reclamation for agriculture, drying out and encroachment amenity and engineering benefits are becoming by human settlements. Wetlands are well known for their smaller and more degraded, and problems such as provision of valuable ecosystem services. In particular, flooding, air pollution and water pollution are becoming the papyrus-dominated Nakivubo wetland in Kampala worse. These problems are likely to escalate with (Figure 1) has been shown in the past to provide an continued movement of the poor into cities and will be important function in the amelioration of water quality in exacerbated by climate change. Inner Murchison Bay. Polluted water from Kampala’s city central district is channelled via the Nakivubo wetland into Inner Murchison Bay, which is also the main water supply area for the city, with water being treated for domestic consumption at the Ggaba Water Treatment Plant just 3km south of the wetland (Emerton et al. 1998). Figure 1 Map showing the location and extent (1996) of the Nakivubo wetland in Kampala Redrawn from: Kansiime & Nalubega 1999                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 9 This polluted water comprises waste water runoff These concerns, as well as the increasing shortage of generated in residential and industrial areas, treated public open space areas in the city that are suitable effluent from the main sewage treatment works, and for recreation, have led to the city’s consideration of raw sewage from slums within the catchment area a possibility of restoring some of the functions and (Emerton et al. 1998, Kayima et al. 2008). In the late economic value of its remaining natural assets, with a 1990s, it was ascertained that the service performed view to creating green urban recreational space and by the wetland yielded a significant cost saving for the associated possibilities for economic development. Ggaba Water Treatment Works (Emerton et al. 1998). One of the areas under consideration is the Nakivubo However, as the city has continued to grow, waste water wetland, a papyrus dominated wetland system. flows into the wetland have increased significantly, overwhelming the wetland’s capacity to assimilate the This study provides a preliminary ecological and economic excess pollutants before reaching Murchison Bay. This, evaluation of the potential for rehabilitation of the as well as encroachment of informal settlements and Nakivubo wetland system. Technical detail is provided in agricultural activities in the wetland (Kansiime & Nalubega a series of appendices. The study provides an overview of 1999) has led to increasing health risks (Stalder 2014), the the current status and functioning of the Nakivubo system deterioration of both the wetland and Inner Murchison and investigates the measures that would be required (a) Bay, and increasing costs of water treatment (Ooyo 2009). to maximise its ecological capacity for removal of wastes, and (b) to restore the health of the system to the extent that would create the opportunity for the development of a recreational area in the lower wetland and adjacent lakeshore area. The indicative costs and benefits of these options are compared, and the policy implications of the findings are discussed.                                                  Page 10 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala II. PAPYRUS WETLANDS AND THEIR FUNCTIONING Papyrus wetlands are associated with shallow, Papyrus grows in dense stands, either rooted to the permanent and slow moving freshwater systems and substrate, or in the form of floating mats. The floating occur extensively around the fringes of Lake Victoria mats initially form as an extension of the rooted stands, (Kansiime et al. 2007, Van Dam et al. 2011). Their natural but sections may break free to form floating papyrus location and extent varies over time with changing rafts (Kansiime et al. 2007). Once these small sections lake levels (Morrison & Harper 2009). However, the are floating, they provide a nucleus for further growth papyrus wetlands around Lake Victoria are threatened by and can lead to expansive vegetative mats, such as conversion to agriculture, grazing and human settlement, those found around Lake Victoria. Even within healthy over harvesting and altered quantity and quality of papyrus swamps, sections of the mat will break off and freshwater inputs. In Kenya, iup to 50% of the papyrus float away. However, increased disturbance (for example wetlands along Lake Victoria have been cleared (Owino through over-harvesting) can destabilise existing mats, & Ryan 2007). These pressures are particularly intense in leading to their more rapid fragmentation. The lower, and around urban areas. floating portions of the wetlands have a high degree of exchange with the lake through seiches (regular, small rises in water level due to oscillations in broader lake water level) which occur on average every 135 minutes and have been estimated to exchange over 100 000 m3 of water per day in the Nakivubo wetland (Kansiime & Nalubega 1999). Figure 2 A typical longitudinal section of the lower Nakivubo swamps showing the dominant vegetation types and related morphological features, not to horizontal scale. Source: Kansiime & Nalubega 1999                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 11 While papyrus wetlands are dominated by the large Water entering wetlands from urbanised catchments in sedge species Cyperus papyrus, other species are many African and other countries generally has elevated often also present in particular zones of the wetland amounts of sediments, nutrients and pollutants from (Figure 2). Within Lake Victoria, the wetland grass industrial effluents, as well as treated and, in many areas, Miscanthidium violaceum, is the next most abundant untreated sewage and other wastes. Excess nutrients wetland plant species. While normally occurring on the can stimulate algal and other plant growth in aquatic fringes of wetlands, Miscanthidium can also form floating ecosystems, potentially leading to a deterioration in mats in deeper water. aquatic ecosystems (e.g. reduced biodiversity, possible toxicity, low concentrations of dissolved oxygen as a The position of papyrus wetlands in the landscape means result of high levels of decomposition of plant material, that they are important both in terms of their influence poor habitat quality and the knock-on effects of these on water quality entering the lake, and as sheltered issues), while heavy metals and pathogens pose a risk and productive refugia for aquatic biodiversity. Where to both natural wetland-associated fauna and flora and they are adequately protected, papyrus wetlands are to human health. Polluted water flowing into wetlands important for biodiversity and provide habitat for a is slowed down as it spreads out into the wetlands and suite of highly specialised and in many cases threatened passes between plants, allowing sediments to settle out, species that are adapted to live within these unique nutrients to be assimilated and pathogens to be destroyed environments. Such species include the sitatunga by UV light (see Appendix 1 for more information on water antelope Tragelaphus spekei, crested crane (Uganda’s quality measures and Appendix 2 for a detailed description national bird), papyrus yellow warbler Chlrorpeta of how wetlands improve water quality). gracilirostris and papyrus gonolek Laniarius mufumbiri (Owino & Ryan 2007). The papyrus wetlands within the The ability of wetlands to perform water quality Lake Victoria system also provide important refugia for amelioration services depends on many factors, juvenile fish of species such as tilapia which constitute including the flow of water through them (hydraulic an important fishery within Lake Victoria (Balirwa 1995, efficiency), their vegetation and management. It is very Kiwango & Wolanski 2008). important to understand that there is an upper limit to the amount of pollution that a wetland can remove, as While healthy wetlands are used to harvest papyrus and well as to the amount of pollution that can be added for fishing and hunting, the capacity of the wetlands to to a wetland without having a significant impact on its supply these resources and to provide a nursery service functioning and biodiversity. for lake fisheries is linked to their ecological condition. The most well-known ecosystem service provided by papyrus wetlands around Lake Victoria is “water quality amelioration”, or “water treatment”. This means that polluted water entering the wetlands is cleaned up to some extent before it passes into Lake Victoria.                                                  Page 12 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala III. THE DETERIORATION OF THE NAKIVUBO WETLAND In its natural state, the Nakivubo wetland would have At the same time, increasing urbanisation has increased been dominated by perennial papyrus swamp, with the the volume of flows passing down the Nakivubo channel swamp area extending much further up the wetland as a result of increased rate of runoff and generation than now. However, the wetland has undergone massive of large volumes of waste water. Pollution enters the changes with the expansion and growth of Kampala. A wetland mainly via the Nakivubo channel, which receives railway line was constructed across the wetland from contaminated runoff and litter from poorly-serviced Port Bell to the city centre in 1923, resulting in the urban and industrial areas abutting the channel, and both gradual infilling and drying out of the upper parts of the treated and untreated sewage effluent from the Bugolobi wetland. This process would have been accelerated by Waste Water Treatment Works (WWTW). In addition, the the construction of channels in the 1950s to carry waste wetland receives treated sewage effluent from the Luzira water from the city, by the manipulation of the wetland Prison and contaminated runoff from the settlements drainage for small-scale farming, and the concentration immediately around the wetland. During the rainy season, of surface flows through a single culvert under the the channel blocks with litter and other waste and tends railway line. By the 1990s, the swamp vegetation was to spread widely across the surrounding area, comprising largely confined to the lower wetland (i.e. below the extant and former, now largely cultivated, portions of the railway line)1. The drying of the upper wetland has Nakivubo wetland (Stalder 2014). More information on probably also contributed to further degradation as pollution inputs is given in Appendix 4. a result of the establishment of unserviced informal settlements along its margins, and increasing cultivation Current levels of pollution in the wetland are hazardous (mainly cocoyams). for human health. The drainage channels have high levels of human excreta, and expose downstream The original extent of the swamp area was estimated users to large loads of parasitic nematodes and faecal to be over 500 ha, but this had been reduced to about bacteria. In addition they impact on wetland and lake 400 ha by 1955. Taylor (1991) estimated the area of ecological function. High loads of organic material, as natural vegetation (the swamp) to be about 280 ha in well as contaminated sediments and nutrients, result in 1990, while Kansiime & Nalubega (1999) estimated deteriorating ecological conditions (particularly reduced 190 ha less than a decade later. Based on Google Earth dissolved oxygen) in the lower Nakivubo wetland. This imagery, we estimated that the current (2015) functional further impacts on lake biota, such as fish populations, area now only covers about 90 -91 ha (Figure 3; see while high concentrations of orthophosphates Appendix 3 for time series). While there was a central entering the Murchison Bay area further increase plant section of wetland vegetation until the mid-1990s, the productivity, contributing to phytoplankton blooms and upper Nakivubo wetland was almost entirely under their associated aesthetic and human health effects and cultivation by 2000. A recently-published study by Isunju costs of treatment for water supply. & Kemp (2016) which is based on more detailed satellite data analysis suggests that 62% of wetland vegetation was lost between 2002 and 2014, mostly attributable to crop cultivation. In recent years, the lower wetland has begun to break up, likely as a result of the intensive use of this area for agriculture. By mid-2015, a large section of Miscanthidium had broken away as a result of agricultural activities. The only detailed studies of the vegetation of the wetland were 1 carried out in the mid-1990s (Kansiime & Nalubega 1999). By then, only small areas of papyrus swamp were recorded in the upper wetland and the remaining lower wetland was dominated by papyrus (floating as well as rooted) and Miscanthidium grass.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 13 Figure 3 Estimated historical extent (green) and current (red) extent of mostly natural vegetation within the Nakivubo wetland.                                                  Page 14 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala IV. CURRENT CAPACITY FOR WATER TREATMENT Emerton et al. (1998) suggested that the Nakivubo wetland treated almost all effluents entering the system and significantly improved the quality of water entering Inner Murchison Bay. However, the overall assimilative capacity of the wetland is assumed to have decreased substantially since then. Kansiime & Nalubega (1999) showed that uptake rates of total nitrogen by the functional papyrus areas of the wetland were in the order of 475 kg N/ha/year, and total phosphorus reduction (through sedimentation and uptake) was in the order of 77 kg P/ha/year. Miscanthidium violaceum mats had lower rates of Figure 4 Estimated past and current capacity of the wetland for sedi- reduction of both phosphorus and nitrogen nutrients mentation /uptake of phosphorus, and estimated hypothetical than C. papyrus. removal if wetland habitats were still intact, with the volume of total phosphorus passing out of the system (i.e. “untreated” load) About 70 ha of papyrus and 20 ha of Miscanthidium currently remains in the Nakivubo wetland, compared with about 91 ha and 23 ha respectively in 1996. Kansiime & Nalubega (1999) noted that more of the polluted inflows came into contact with the Miscanthidium than with papyrus, but papyrus is more efficient at nutrient removal and had higher overall removal rates. Based on their findings, it was estimated that the wetland removed about 15% of nitrogen and 20% of phosphorus inputs in 1996, but that this has now decreased to 8% and 10%, respectively, as a result of both increases in nutrient inputs and absolute decrease in the assimilative capacity of the wetland. Based on these findings, if the full extent of the wetland habitat was restored, it could hypothetically remove about 15% and 19% of current N and P inputs. If the restored areas were managed as treatment wetlands, this amount would be further increased (see Chapter 5 and Appendix 6).                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 15 V. INTERVENTIONS REQUIRED TO MEET WATER QUALITY AND RECREATIONAL OBJECTIVES A. Objectives On the basis of the present condition and extent One of the main objectives of this study was to identify of the wetland and its current pollution load (see the interventions that would be required to improve the Appendices 2, 4 and 5), it must be stressed that quality of water passing from the Nakivubo wetland into the above objectives cannot be met simply by Inner Murchison Bay, in order to address concerns about expanding the natural wetland treatment capacity potable water supply for the city, to enable the potential of the lower Nakivubo wetland (see Figure  4). The development of recreational facilities in the Nakivubo volumes of waste water being produced in the wetland and Inner Murchison Bay, and to address other upstream catchment and its quality are well beyond environmental problems such as the impact of water the capacity of the wetland in its current state or quality on biodiversity and fisheries. potential rehabilitated state (e.g. artificial wetlands) to effect adequate treatment on a long term basis. The objectives were defined as Indeed, while many wetlands are recognised as playing a valuable role in effecting the polishing of 1. effecting a measurable improvement of water quality contaminated water passing through them, they do passing out of the Nakivubo wetland into Inner not provide effective mechanisms for the treatment Murchison Bay; of large volumes of highly contaminated waste with high loads of sediments and nutrients. Recognition 2. to ensure sustainable management of the Nakivubo of this point is a critical factor underpinning wetland; the approach recommended in this section to address water quality issues, and thereby drive 3. reduce water quality impacts on human health; and other activities and/or opportunities such as the rehabilitation of parts of the Nakivubo wetlands and 4. open up opportunities for safe recreational use of the the development of recreational opportunities within lower wetland. the wetlands.                                                  Page 16 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala B. Proposed interventions A sequential series of interventions, also known as a treatment train, is required to achieve the broad and ambitious objectives outlined in Figure 5. The interventions include both infrastructure upgrades and wetland rehabilitation and conservation measures. The six interventions are described briefly below, after which the high-level cost estimates are summarised and the cumulative impact of the interventions on water quality is discussed. A fuller description of the background, design and assumptions around each set of interventions is provided in Appendix 6. Step 1. Prevent pollution at source Fewer than 10% of households in Kampala have piped sewage (NWSC Corporate Plan 2012), with the remaining population depending on various forms of on-site sanitation such as pit latrines, septic tanks, public toilets and open defecation. The collection and disposal of faecal sludge generated in the pit latrines is inadequate and sanitation in informal settlements is very poor. In order to prevent raw sewage from entering the wetland and Inner Murchison Bay, both sanitation coverage and waste water treatment works (WWTW) capacity will need to be further increased by improving sanitation in areas not covered by water-borne sewage, expanding Figure 5 Treatment train required to achieve water quality objectives in the lower Nakivubo wetland and Inner Murchison Bay, with brief sewage systems, along with the provision of public summary of actions required at each step. (Note: the sixth step, the health education. Sanitation measures ought to be in establishment of recreational facilities, is not depicted here). place as minimum standards in any high-density urban environment. These efforts should be complemented by improving sweeping and litter collection to reduce the amount of sediment and solid waste entering the wetland. Some of these required improvements are already underway. New WWTW are under construction at Lubigi and Nakivubo, plus new sewers in Nakivubo and Kinawataka areas, which will increase the sanitation coverage to 30% of households in Kampala (NWSC 2014). The new 45 ML Nakivubo WWTW will replace the Bugolobi WWTW with its current design capacity of 12 ML and effective capacity of 6 ML. This will now deal with waste water from the Kinawataka catchment as well as the Nakivubo catchment, so there is some uncertainty as to how much of the Nakivubo and Kinawataka catchments respectively will be serviced. The total wastewater load from the Nakivubo catchment, if fully serviced, would be about 65 ML. Full coverage of waterborne sewage may not be feasible or desirable, however. It is probably more reasonable to aim for coverage of 60-70%, while implementing other measures such as VIP latrines in the remaining areas and having waste water treatment works that can handle waste water flows channelled from those areas as well as efficient sludge collection services and treatment facilities.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 17 The Lake Victoria WatSan Project will address the Step 2. Prevent residual pollution from entering the water supply and sanitation situation for some of wetland Kampala’s informal settlements, including faecal sludge It is recognised that the implementation of water-borne management, sewerage and building public toilets. sewage is unlikely to be feasible for all households in About €3.41 million of the budget will be spent on the catchment, especially if there is further population upgrading sanitation within the Nakivubo catchment. growth in the catchment. Therefore alternative measures (NWSC 2015). The project does not include the need to be put in place to address outstanding issues installation (or subsidisation) of pit latrines nor will it surrounding polluted runoff, such as the installation and achieve full servicing of currently under-serviced areas management of litter and sediment traps, installation in terms of sludge removal. This project will also cover of diversion drains to convey waste water from informal public education, which is not considered further here. settlements to the WWTW, and the creation of wetland filtration strips at the original wetland edges to filter While the ongoing projects are expected to greatly overflows before they reach the main wetland. The improve the situation, there is clearly still a shortfall combination of these measures with the above (Step in terms of sanitation. Given that the new Nakivubo 1) is considered the most effective, long-term approach WWTW will also have to handle waste from Kinawataka, to addressing the dire pollution levels passing into the it is likely that some further expansion of capacity will be Nakivubo channel. required to adequately service the Nakivubo catchment. There is also still a need to install pit latrines, increase To control and limit the amount of litter and sediment sludge treatment facilities, and to further increase flowing into the Nakivubo channel an expanded channel WWTW capacity to deal with inputs from formal and litter trap, instream litter and sediment traps, about 12 informal areas. Based on the original plan to develop hydraulic drops constructed from gabions to control for an 8ML WWTW in the Kinawataka catchment, it was flow and particle transport, and continuous cleaning of assumed that the capacity of the Nakivubo WWTW the channel will be required, at a cost of about $1.75 needs to be increased by at least this amount in order to million plus $0.95 million per year. Diversion drains meet sewage treatment requirements in the catchment, are required to intercept dry season surface flows (i.e. at an estimated cost of $8.8 million. Based on available flows not driven by rainfall events) and shallow seepage information (Ministry of Health 2010) we estimated that from unlined areas of informal settlements and would some 12.6% of the population in the catchment would be routed to the nearest existing sewer and on to the still require pit latrines, and 40% of the existing latrines WWTW for treatment. This would cost approximately need access to proper servicing. Using data from Isunju $1 million. In addition, any excess runoff, most likely et al. (2013), the costs of these were estimated to be during wet conditions, should be diverted into sediment about $1.15 million plus $0.67 million per year. Currently traps and through reed bed wetland filtration systems, the collection and treatment of faecal sludge in Kampala established between diversion drain outlets and the main is 43% or 390 m3 per day. Based on this information Nakivubo channel. The reed beds, which are necessary and the total number of households in the catchment for filtering sediments and reducing the volume of area using pit latrines, it was also estimated that a 200 organic waste should be designed as broad wetland m3 faecal sludge treatment plant (FSTP) is needed to filtration strips that run parallel with the channel. It is treat faecal sludge from pit latrines in the catchment recommended that a linear series of papyrus wetlands adequately. The total construction cost of a new sludge should be installed along the wetland edge in the vicinity treatment plant was estimated to be $1.44 million, with of the prison, occupying the eastern edge of the lower annual maintenance costs of around $0.07 million. wetland, and along the wetland edge of the informal settlements occupying the western edge of the lower In addition to sanitation measures, it was also assumed wetland (Figure 6). These wetlands would cost about that current levels of street sweeping and litter collection $1.65 million to establish, and then would require need to be improved at a cost of about $0.6 million per maintenance on a cyclical basis, which would include the annum, to further prevent sediments and litter from cutting and disposal of plant material, and the dredging entering the storm water system. and disposal to waste of contaminated sediment as necessary to restore wetland treatment capacity. While implementation of these measures would require considerable capital expenditure of over $11 million, given the significant impact it would have on instream and downstream ecological function, as well as its implications for human health, and the opening up of opportunities for safe recreation downstream, it should be regarded as an essential medium-term rather than long-term objective.                                                  Page 18 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Step 3. Improve WWTW effluent quality Step 4. Rehabilitate upper wetland for waste water The new Nakivubo WWTW will include mechanical, treatment biological and chemical treatment of municipal sewage Rehabilitating the wetland for waste water polishing and industrial effluent, sludge digestion and cogeneration (i.e. supplementary treatment of final effluent) requires using biogas. It is recommended that additional installing and maintaining an extensive series of tertiary treatment measures such as filters to decrease treatment wetlands throughout the upper Nakivubo nutrient concentration and maturation ponds to reduce wetland (Figure 6). pathogens are also added. This would cost about $22 million (Table 1). The area needs to be far more extensive than has been previously considered, because of the large volume Improving the aeration of treated effluent from the of wastewater that has to be treated, and should existing and planned WWTW, by installing aeration ideally replace all of the upper wetland area that has sprays within the final effluent maturation ponds, would been converted to agriculture. This would entail the increase dissolved oxygen concentrations to at least construction of multiple shallow wetland cells separated 4mg/L. Upgrading technology at the WWTW to allow by berms with controlled pipe outlets or multiple for a final effluent concentration of 1.5 mg PO4-P /l overflows linking each cell to downstream cells. Based on (phosphorus in orthophosphate) instead of the current information from other areas and taking economies of standard of 10 mg P/L (i.e. meeting effluent limits) scale into account, the cost of constructing these wetlands would substantially reduce phosphorous loads entering would be in the order of $2.3 million. As in Step 2, these the wetland, thereby improving conditions for aquatic wetlands would require maintenance on a cyclical basis, ecosystems in Murchison Bay. which would include the cutting and disposal of plant material, and the dredging of contaminated sediment to restore wetland treatment capacity. Note that achieving required levels of tertiary polishing of effluent will be dependent on the size of the constructed wetland area and the concentration of the final effluent produced by the WWTW. Effluent concentrations need to be reduced significantly below those that are legally permissible in Uganda if specified volumes are to be treated effectively by rehabilitated wetland areas. Rehabilitated wetlands will therefore help to achieve downstream benefits in terms of resource quality, by helping to make the lower wetland safe for recreational purposes, improving fisheries, and potentially controlling the expansion of informal settlements into wetland areas.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 19 Figure 6 Areas for rehabilitation as filter strips (bright green areas) and treatment wetlands (light green areas) in the Nakivubo wetlands. Source: Google Earth imagery Step 5. Restore and protect lower wetland To avoid further breaking up and loss of these wetlands, Improving the functioning of the lower wetland requires agricultural activities need to be managed to prevent improving the spread of flow into this area from upstream, cultivation and livestock grazing on the floating islands. reducing agricultural damage and encroachment, Recommended measures include fencing where encouraging conservation and the regular harvesting of necessary, and signage and patrols to enforce these papyrus plant material to maintain wetland function. regulations. In order to promote nutrient storage within the wetland and to stimulate growth, papyrus needs to It is recommended that at least three new channels be harvested at certain times of the year and at different are created via a series of pipes and culverts under growth stages. A specific harvesting plan for papyrus the railway line in order to spread the flow effectively plant material in the lower wetland should be initiated from the treatment wetlands along the width of the and this should focus on educating local people about lower wetland. This would cost about $1.38 million. how best to harvest the plant material. The ongoing costs Flows from the Nakivubo channel should pass into the would therefore include the managing and monitoring lower wetland as at present, with possible allowance of the harvesting programme as part of the larger for further spreading of flows into degraded Papyrus conservation effort in the lower wetland. wetland.                                                  Page 20 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Figure 7 Concept vision for the Nakivubo Wetland Park used in this study Step 6. Establish a recreational space with facilities If well managed, this in itself could provide tourism The development of a Nakivubo Wetland Park is opportunities. The park would also contain a wetland envisaged to be integrally and physically linked to the visitor information centre located in between the planned waterfront area at Port Bell with the provision landscaped areas and bird sanctuary to enhance visitor of a reasonably large park area adjacent to (not in) experience. The area in the vicinity of the prison site and the wetland, with access to both the lake-front and Port Bell would be developed for retail and restaurants vegetated wetlands, with picnic facilities and walking with some of these being on the waterfront. In addition paths in and around the landscaped area and wetland to the central wetland recreational area at Port Bell it (Figure 7). Therefore the main landscaped recreational is also envisaged that landscaping of smaller areas on area would be close to Port Bell, but smaller sites could the opposite side of the lower wetland and the possible also be set up at other locations around the wetland (See construction of wide paved walking and cycle routes along Appendix 6 for more detail). one or both banks of the wetland extending from the lake to 5th bridge. These walkways and cycle paths would It is envisioned that a central part of the wetland provide a safe and pleasant passage into the city which could be developed as a bird sanctuary which would would also encourage non-motorised commuting such as retain its natural characteristics as far as possible. The the use of bicycle taxis. sanctuary would contain rustic boardwalks and hides.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 21 The costs of providing public amenities such as pathways, C. Overall costs boardwalks, foot bridges, landscaped park space for picnic and leisure, bird hides, public toilet facilities, The overall estimated costs of the interventions are parking areas and a visitor centre are estimated to summarised in Table 2. These estimates are over and be about $11.5 million. This does not include private above the expenditure that has already been committed development costs associated with the development of to in terms of sanitation upgrades. restaurants and retail outlets. Table 2 Estimated capital costs and ongoing maintenance and operating costs of the recommended interventions, broken into component costs. The minimum and maximum range for cost estimates are included in brackets adjacent to the estimated cost value for each intervention. Capital Costs Maintenance and (US $ Millions) operating costs (US $ Millions per year) 1. Prevent pollution at source 11.39 1.78 Expand pit latrine access and servicing in Nakivubo catchment 1.15 (0.23 – 2.4) 0.67 (0.49 – 2.45) Increase capacity of the Nakivubo WWTW by 8 ML 8.8 (6.6 – 11.0) 0.44 (0.33 – 0.55) Construct a 200 m3/day Faecal Sludge Treatment Plant 1.44 (1.08 – 1.80) 0.07 (0.05 – 0.09) Increased street sweeping and litter collection - 0.60 (0.4 – 0.8) 2. Removal of pollution upstream of natural water resources 4.44 1.01 Installation and management of litter and sediment traps 1.75 (1.5 – 2.0) 0.95 (0.48 – 1.4) Install diversion drains from informal settlements to WWTW 1.04 (0.3 – 1.77) 0.02 (0.006 – 0.04) Installation and management of wetland filtration strips 1.65 (1.24 – 2.06) 0.03 (0.02 – 0.04) 3. Improvement in WWTW effluent quality 21.96 0.44 Filters to decrease nutrient concentrations 16.50 (12.4 – 20.6) 0.33 (0.25 -0.41) Maturation ponds to reduce pathogens 5.46 (4.10 – 6.83) 0.11 (0.08 -0.14) 4. Rehabilitate parts of wetland for treatment 2.30 0.05 Installation and management of treatment wetlands 2.30 (1.73 – 2.88) 0.05 (0.03 – 0.06) 5. Improvement in function of the lower wetland 1.41 0.07 Pipes/culverts to improve hydraulic efficiency in lower wetlands 1.38 (1.04 – 1.72) 0.04 (0.03 – 0.05) Measures to control agricultural activities 0.03 (0.02 – 0.04) 0.01 (0.005 – 0.01) Harvesting of Papyrus plant material to stimulate growth 0.00 0.015 (0.01 – 0.02) 6. Recreational facilities 11.48 0.23 Boardwalks, signage etc. to control recreational impacts 2.52 (1.9 – 3.1) 0.05 (0.04 – 0.06) Visitor centre, landscaping, picnic facilities, ablution facilities 8.96 (6.7 – 11.2) 0.18 (0.13 – 0.22) TOTAL COSTS 52.98 3.57                                                  Page 22 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala D. Cumulative effect on water quality While Steps 1 and 2 are likely to reduce nutrient levels Note that the accuracy of predicted implementation significantly, these steps alone would not take the effects, and in particular their knock-on downstream wetland out of its hypertrophic state. Adding Step 3 is implications, is severely limited by poor data relating likely to achieve eutrophic conditions. Implementation to water quality and hydrology in the catchment, and of Steps 1, 2 and 3 are considered an essential pre- would be significantly improved by a structured flow requirement for achieving the outcomes of Steps 4 and water quality monitoring programme, designed to and 5, i.e. providing the opportunity for alternative improve decision-making around the short and long beneficial uses of the lower wetland (e.g. recreation, term benefits and costs likely to be associated with the tourism, fishing industry). Steps 4 and 5 are expected to recommended measures. bring the system close to its natural mesotrophic state. The cumulative modelled effect of the five intervention steps on the trophic state of the Nakivubo wetland is shown in Figure 8. Figure 8 Schematic diagram of the cumulative effect of the five intervention steps on the trophic state of the lower wetland area.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 23 VI. POTENTIAL WATER TREATMENT COST SAVINGS The city of Kampala is supplied with drinking water Water treatment data from Ggaba WTW for the period from Lake Victoria, with the treatment and distribution February 2007 – May 2015 supplied by the NWSC were being operated by the National Water and Sewerage analysed. However, there were inconsistencies in some Corporation (NWSC) and Kampala Water (KW) (NWSC of the cost data and unexpected trends and results 2014). The Ggaba Water Treatment Works (WTW) is from the treatment analysis. As a result, we were not located just 3 km south of the Nakivubo wetland, making able to model costs using the local data provided. For it vulnerable to the impacts of urban activities on water the detailed analyses and more information about the quality in Inner Murchison Bay. empirical data see Appendix 7. Discussions with the NWSC confirmed that contrary to findings from the Both high sediment and nutrient loads typically lead to analysis, water quality deterioration still has a significant increased water treatment operation and maintenance impact on costs and will continue to do so at the new costs. Increases in sedimentation and associated plant. In the absence of a useable model for the Ggaba increases in turbidity in river and wetland systems plant, it was decided to value the potential cost savings have become a common challenge facing many cities using a benefits transfer method. This is essentially worldwide (McDonald & Shemie 2014). In the context of the use of a model developed elsewhere, but with treating water, the biggest effects of increased sediment parameters adjusted as far as possible to correct for and nutrient loads on the cost of treating water include differences between the source and receiving valuation increased use of coagulants, increased sludge output, site. Therefore the potential cost savings were estimated increased occurrence of algal blooms, and clogging of using a model from a similar WTW in South Africa (see reticulation systems (Graham 2004, McDonald & Shemie Appendix 7). 2014, Rangeti 2014). It has previously been suggested that the Nakivubo wetland buffered these effects by Based on existing data, and the assumption that the significantly improving water quality before it reached restoration interventions would achieve a mesotrophic Murchison Bay, leading to water treatment cost savings. state, the pre- and post-restoration phosphorous However, more recently it has been suggested that water concentrations were estimated to be 2.5 mg/L and 0.047 treatment costs have in fact been rising as a result of mg/L, respectively. This would suggest a reduction in the subsequent deterioration of water quality in Inner loads from over 8000 kg to about 150 kg. The model Murchison Bay. In fact, Ooyo (2009) found intensive algal was updated using average values for water quality blooms to be particularly high, resulting from elevated parameters found in the raw water in Inner Murchison phosphorous levels in Inner Murchison Bay. Between Bay and it was estimated that the saving per ML of 1993 and 2007 the aluminium sulphate dosage, used treated water would be almost half of the current day during the treatment process to remove algae, increased cost, amounting to some $845 000 per year. significantly from 20 mg/l to almost 70 mg/l (Ooyo 2009). This effect has been sufficient to lead to various The water currently being treated is much less than the upgrades of the water treatment works, and finally to quantity demanded, and there are several “dry zones” the planning of a new facility at some distance from the in Kampala. The city has planned for growth in capacity. urban environment. Works are currently underway at the existing Ggaba site, which will increase the capacity of these works to Ggaba Water Treatment Works comprises three some extent, and then a new treatment works will be treatment plants with a total water supply capacity of constructed outside of Inner Murchison Bay, at some 170,000 m3 per day (NWSC pers. comm. 2015). Upgrades distance from the city, to meet future demands2. It is have occurred in 1992 and in 2007 to make the plant assumed that once these upgrades are completed, the fully automated (Ooyo 2009). In 2010, the intake pipeline Ggaba plants will be able to increase its outputs from at Ggaba III was extended further into Lake Victoria in an 170 ML to the overall design capacity of about 230 ML. effort to extract water of a better quality for treatment. Thus the cost saving was estimated under full capacity, as From October 2010, the new pipeline was in operation $1.143 million per annum. (NWSC pers. comm.). It should be noted that the extra cost of building the latter plant 2 further away due to water quality issues is already a sunk cost. Although it is far away, the pollution inputs into IMB via the Nakivubo wetland will eventually have an effect on costs of water treatment at the new plant, but this is unlikely to be significant within the time frame of this analysis.                                                  Page 24 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Not only is it expected that treatment costs will decrease, but associated costs such as sludge removal and backwashing are also expected to decrease with improved water quality. One assumes that the significant reduction in phosphorus and other nutrients as a result of restoration will decrease the frequency and intensity of algal blooms in Inner Murchison Bay and will decrease the need for continuous backwashing of filters at the Ggaba WTW. Sludge dewatering and sludge removal can be very expensive and it is assumed that the improved water quality will decrease the amount of sludge produced during the water treatment process and will therefore reduce the time and costs involved in having it removed. In the Rio Camboriu watershed in Brazil, preliminary results have shown that a significant part of the cost saving associated with improved water quality was in the savings associated with sludge disposal (Timm Kroeger, TNC, in litt.)                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 25 VII. POTENTIAL RECREATIONAL BENEFITS A study was undertaken to estimate the possible welfare Given a total of over 400 000 households in Kampala City, gains from the creation of a Nakivubo Wetland Park, the results of the revealed preference analysis suggest a using both revealed preference methods (analysis of the potential overall net benefit to Kampalans of some $17.0 use of existing sites to infer the willingness to pay for the - $29.4 million per annum. Using the mean household proposed site) and stated preference methods (asking size, mean adult:child ratio, and mean expected level of people to state their willingness to pay to use a proposed use per household (2.3 times), the results of the stated site). Data were collected by means of a survey of over preference analysis yield a slightly lower overall estimate 600 households in a representative set of parishes of $14.7 - $15.9 million per annum. Nevertheless the in all five divisions of Kampala. Respondents were estimates are relatively close. These minimum and asked to answer questions about their socio-economic maximum estimations were used to determine an circumstances, about household use of multiple outdoor average estimate of $22.05 million in overall recreational recreational sites during the last 12 months and about benefit (Table 3). These results highlight the demand for how they felt about the restoration of the Nakivubo outdoor recreational opportunities in Kampala and the wetland for its development as a recreational park. These associated welfare gains with improving current outdoor included questions about their current expenditure open green spaces. and how much they would be willing to pay to use a park at Nakivubo. Data were analysed using a range of econometric models (for the full study, see Appendix 8). Based on different models, the revealed preference analysis suggested that the compensating variation (CV) or total willingness to pay (WTP) for the Nakivubo Wetland Park would be from $18 (on average) to $75 (for one sector of the population) per household per year, suggesting a substantial welfare gain of designing a new site that involves such changes. The stated preference analysis suggested that households were willing to pay between $5.64 and $6.11 per adult, and anticipated visiting the park on average 2.3 times per year. Table 3 Results of the cost-benefit analysis Revealed Preference Stated Preference Willingness to Pay ($/hh/year) 40 -70 35 – 38 Aggregate benefit (US $ millions) 17 – 29.4 14.7 – 15.9 Average estimate used in CBA (US $ millions) 22.05                                                  Page 26 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala VIII. COST-BENEFIT ANALYSIS Cost-benefit analysis is a conceptual framework and tool Note that the time for recovery of the lake system is used to evaluate the viability and desirability of projects unknown. This is difficult to estimate since nutrient and policies based on costs and benefits accumulating cycling is a complex process. Phosphorous has been over time (Hanley & Spash 1993, Pearce et al. 2006). accumulating in Inner Murchison Bay for decades. Cost-benefit analysis involves the adjustment of future The nutrient fluxes and transportation are not well values to their present value equivalent through the understood in this system. To some extent there is process of discounting at a rate which reflects the exchange with the rest of Lake Victoria, but the nature potential rate of return on alternative investments or the of Inner Murchison Bay is that water is retained within rate of time preference. For a project to be considered the bay for long periods, and sediments and nutrients viable, the net present value (NPV) must be positive. accumulate in this area. The phosphorous that has This places greater weight on values occurring closer accumulated in the system will take some years to to the present, which means that the future benefits of be depleted and/or flushed from the system, and will restoration projects will be down-weighted compared continue to become available for plant uptake during with the upfront investment costs, and have to be periods of disturbance (e.g. by wind). Therefore, even substantial in order for a project to be viewed positively. if nutrient inputs into Inner Murchison Bay were to be Projects can also be evaluated by estimating the internal reduced to very low levels by the interventions described rate of return (IRR), which is the discount rate at which in this report, it may take years before this has an impact the total net present value of the project falls to zero. on the trophic status of the lower wetland and Inner Murchison Bay. Thus we have incorporated a time delay The implicit assumption of the above is that the costs of 5 years before the benefits of the project are felt. and benefits of a project can be determined with certainty. In reality however, accurately estimating all The estimates derived in this study suggest that the variables in a cost-benefit analysis becomes a challenge interventions would have a net benefit. Using a discount as a result of the way in which estimates are assessed rate of 6%, the net present value of the project over and forecast (EAI 2006). Studies are limited by availability 15 years was estimated to be $80 million (Table 4). of data and resources, as well as uncertainty in the This was further tested under varying assumptions of consideration of changes in factors such as land use, costs, benefits and discount rate. Under the worst case climate, household incomes and urbanisation (EAI 2006). scenario (upper bound estimates of costs, lower bound It is therefore important to incorporate some form estimates of benefits), the NPV was negative at discount of sensitivity analysis so as to adequately assess the rates of 6% and 9%, which suggests there is some risk of reliability of the estimates. the project costs outweighing the benefits (Table 5). The internal rate of return (IRR) was estimated to be 20%, but In addition to the costs of the interventions and the sensitivity analysis yielded a range of 4 to 34%. Monte water treatment and recreational benefits described Carlo Simulations were conducted using 10 000 runs on above, the opportunity costs associated with the loss the base, best and worst NPV estimates using a triangular of agricultural land were also taken into account. It is distribution for 3%, 6% and 9% discount rates. The results estimated that there are currently 135 ha of agricultural suggest that the probability of a negative outcome is 0%, land within the wetland that need to be rehabilitated. 2% and 7% for these three discount rates, respectively. The value added by the past conversion of wetland to crop farming was estimated to be between $91 000 to $192 000 (average $141 500) per year, based on Emerton et al. (1998) and Kakuru et al. (2013), respectively.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 27 Table 4 Present value of costs and benefits under expected best case, base estimate and worst case scenarios (2015 US Dollars, 6% discount rate, 15 years) Present Value (US $ millions) Best Base Estimate Worst Costs Restoration 35.6 48.6 62.0 Maintenance 21.1 32.1 57.6 Loss of agricultural land 0.83 1.30 1.76 Total present value of costs 57.6 82.0 121.3 Benefits Recreation 206.9 155.2 103.4 Water treatment Savings 8.7 7.0 5.2 Total present value of benefits 215.6 162.1 108.7 Net Present Value 158.0 80.1 -12.6 Internal Rate of Return (IRR, %) 34% 20% 4% Table 5 NPV Sensitivity Analysis using discount rates of 3%, 6%, 9% Net present value (US $ millions) Discount Rate 3% 6% 9% Best 220 158 8 3 Base estimate 121 80 51 Worst 4 -13 -24                                                  Page 28 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala It should be noted that the estimate of benefits are The restoration of water quality in Nakivubo wetland highly conservative. Firstly, the recreational benefits may also be beneficial to fisheries, but this is dependent associated with the restoration today are likely to on a number of factors. Habitat degradation and poor understate future values of these benefits, which water quality has led to the movement of fishermen out are expected to increase as the city becomes larger of Inner Murchison Bay to more productive fishing areas and incomes rise. Secondly, the cost-benefit analysis in Lake Victoria. Fish communities have been impacted does not take into account a range of other potential by low levels of oxygen as a result of the eutrophication benefits of the restoration interventions. These include of wetland areas and the increased nutrients in the lake improvements in human health, property value and resulting in excessive algal and plankton growth (Muyodi fisheries, as well as intangible benefits that may arise et al. 2011). It is expected that with improved effluent from restoring the biodiversity of the area. control, siltation control and restored riparian habitat within the Nakivubo wetland and along the lake shore, Human health is strongly related to access to water fish stocks will start to increase in Inner Murchison and sanitation (Alexander & McInnes 2012, Wetlands Bay and this would have a positive influence on local International 2010). Many people living within the fisheries. Studies elsewhere have shown that wetland Nakivubo catchment are exposed to poor sanitation restoration can curtail the decline and loss of local conditions. In addition, the people living along the fisheries (Alexander & McInnes 2012). However, this is edge of the Nakivubo wetland and those using the dependent on how the other catchments and wetlands wetland for farming have direct contact with the water around Inner Murchison Bay are managed, the time (and parasites) and are exposed to considerable risk. taken for lake water quality to change, and the way in An improvement in sanitation and the disposal and which the fisheries are managed. treatment of waste water will have a significant positive influence on human health and welfare in Kampala, with Finally, the restoration of the wetland would lead to an further benefits in terms of reduced pressure on health improvement in its flora and fauna, which is something services and avoided loss of productivity. that many members of society, even beyond Kampala or Uganda, would value. These kinds of values, referred An improved local environment is also likely to be to in the literature as non-use or existence values, are reflected in property prices in the area. It is well known intangible and difficult to quantify, even with best- that well-managed green open space areas lead to higher practice stated preference methods (Dubgaard 2003, property prices, whereas degraded environments lower Olsen & Shannon 2010). While the study area is not property prices (Behrer 2010, Letley & Turpie 2015). particularly important from a conservation perspective, In brief discussions with property agents in Kampala, e.g. as a habitat for rare and endangered species, its it was suggested that property values are positively conservation is likely to play a role in the overall health influenced by altitude, with hilltop homes having better and viability of aquatic ecosystems in and around views, whereas the properties in the valley areas close Murchison Bay, and thus the overall levels and diversity to the wetlands had lower value, and this tended to of flora and fauna that can be supported in the area. be very strongly negatively influenced by the fact that While this study has not attempted to estimate existence the wetland areas are the areas available for informal value, this benefit should be acknowledged. settlement and the development of slums. Where slums do occur, it is likely to be this factor, rather than the state of the wetland, that drives property prices. However, away from slums, the health of wetlands is more likely to have an impact. In the Nakivubo wetland area, while some of the wetland is fringed with slums, there are areas where property values may potentially benefit from wetland restoration. If residents do value open space and associated amenities then it would be expected that the willingness to pay for these amenities should be revealed in property prices3. This could be captured through land value capture initiatives in 3 the areas where there is planned development. Instruments such as tax increment financing (TIF), joint development agreements and betterment tax could be used to capitalise on these benefits. Tax increment financing, for example, uses property tax increases to repay public infrastructure investment required by a project. It encourages investment in deteriorating areas by allowing local governments to use future property tax revenues to finance the current infrastructure costs needed to attract development. Local government could also levy a betterment tax against real estate in the area, which is a tax that is dedicated to a specific project that helps to improve an area.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 29 IX. CONCLUSIONS AND POLICY IMPLICATIONS Key message 1: Wetlands have limited capacity to solve Key message 3: There are no shortcuts urban pollution problems For restoration to be successful, implementation of The Nakivubo wetland has deteriorated significantly since all the interventions outlined in this study is required. the 1990s, when it was believed to perform a valuable Ad hoc investments not driven by detailed plans will service in ameliorating the quality of water passing fail to deliver the outcomes required, resulting in both though it into Inner Murchison Bay, even though at that wasted money and time. The order of implementation stage the system was already impacted by structures is not particularly important as long as they are all and human activities. The pressures on the wetland, implemented. For example, Step 2 could be implemented including pollution inflows, agriculture and settlement, as a stopgap while Step 1 is finalised. In addition, the have increased markedly, and the functional part of sooner that work begins on the wetland-edge filter the wetland is now less than a quarter of its original strips and on wetland rehabilitation, the lower the risk extent. The wetland no longer has any positive impact of further human encroachment into the wetland when on the trophic status of Inner Murchison Bay, with the it comes to the implementation of this step, because lower wetland and bay having reached a hypertrophic people will not be able to settle or carry out agriculture state that is characterised by frequent, often toxic, algal in the re-established permanently wet areas. blooms, as well as being severely contaminated with pathogens that carry a risk to human health. Key message 4: This will need fundraising and co- ordination Timely interventions to keep pace with the city’s growing In order for the restoration interventions to be sanitation needs would have prevented the deterioration successfully realised, the city will need to carry out a of water quality in Nakivubo and Inner Murchison Bay. suite of actions relating to sourcing and management of Waste water treatment standards are also a contributing funds, sorting out responsibilities and co-ordinating the factor. The lack of urgency in dealing with these issues efforts required. may have been partly due to the assumption that the environment would be able to deal with the remainder This study estimates that a total initial expenditure of at of the pollution load without any significant impact on least $53 million is required in order to achieve the goals human welfare. This is a myth that needs to be quickly outlined, depending on the actual shortfall in sanitation dispelled. Wetlands cannot substitute wastewater infrastructure. If, for example, the total Nakivubo treatment works and can only improve the quality of restoration took place over four years, this would entail limited volumes of moderately polluted water. an average annual capital expenditure of 13.25 per year, which is equal to 39-74% of the city’s capital budget in Key message 2: Rehabilitation will be costly but the last three financial years up to 2014/15 ($18, $25 worthwhile and $34 million, respectively) and 11-23% of planned As is often the case in the restoration of natural systems, annual capital expenditures over the next two years ($57 the estimated costs of the required interventions can and $123 million, respectively, which is largely donor- be expected to be high. Many of these interventions funded). This suggests that the Nakivubo restoration could be regarded as obligations that should be met is only feasible with a significant contribution of donor irrespective of the restoration of wetlands, for the funding. The challenge this represents may look different mere purpose of providing essential services required depending on the potential funding sources and whether for maintaining human health and dignity. While these these are fungible (i.e. interchangeable). For example investments would not be particularly cost effective in there may be an opportunity to use a donor source that terms of the impacts on water treatment costs alone has a pre-determined preference to spend on these (given the investment that has already been made in issues. Raising funds for further sanitation investments a new plant further afield), taken together with their does not necessarily have to be linked to wetland potential recreational benefits, they are likely to be restoration. Separating the issues and addressing worthwhile. This is not surprising given the dire lack sanitation first will lower the risks and make funding for of managed green open space available to Kampala the direct wetland rehabilitation work more accessible. residents for outdoor recreation. As the economy of Uganda continues to develop and incomes improve, the demand for these kinds of amenities is likely to increase. Such benefits would also potentially be reflected more tangibly in property values in the area. In addition, the restoration of the wetlands and water quality in Inner Murchison Bay would be likely to have benefits for biodiversity and fisheries.                                                  Page 30 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala A detailed plan will need to be drawn up. The actions Thus the following recommendations should be described in this report pertain to multiple government considered: institutions, particularly the KCCA, the NWSC, the Ministry of Lands, Housing and Urban Development, and 1. Sanitation measures need to keep ahead of the the Ministry of Natural Resources. In order for effective growing urban population, and until this is the case, action to take place, coordination between these bodies runoff from areas that are not well serviced needs to is necessary. It will also be necessary for one of these be diverted to a WWTW; bodies, probably the NWSC to take the lead in this co-ordination, with NWSC focusing on the bulk of the 2. Water quality standards for waste water treatment investments required including treatment wetlands, need to be made more stringent, particularly with KCCA focusing on the investment in recreational facilities, regards to phosporus, and enforced, and investments and the Ministry of Natural Resources focusing on need to be made in the technological innovation the management of the remaining natural wetlands. needed to deal with this; Similarly, there will need to be clarity regarding the responsibility and capacity of different institutions to 3. There needs to be stricter regulation and control undertake the ongoing oversight management actions. of industrial discharges. This should form part of a review into industrial effluent guidelines and current Key message 5: Avoid a repeat experience practices; There are important lessons to be learned from this study. Wetland function has been detrimentally impacted 4. Legal protection of wetlands needs to be by direct and indirect interventions that have in many strengthened, and tough and/or innovative measures cases wasted opportunities to reap benefits that might need to be taken to prevent the reclamation, farming be associated with a healthy wetland. For example, and settlement of wetland areas; increasing the spread of flows under the railway line would have retained a much larger area of functional 5. Construction of roads and railways through wetlands wetland downstream, and thus increased wetland should be avoided, or at least be done in such a way services. Given the pollution loading into and through as to minimise impacts on hydro- and sediment the wetland, even such measures would not however dynamics and aesthetics; have protected the wetland adequately, or allowed it to address its present pollution load to any significant 6. Monitoring programmes need to be established, level. Considerable environmental and economic costs including continuous monitoring of flows into the have thus been incurred by delaying investments that wetland, and monthly measures of water quality of are inevitably required, including the additional costs inflows and within all wetland areas in and around that have been sunk into water treatment facility Kampala. Water quality monitoring data should modifications and the location of new treatment works be made readily available in an online platform for at distant sites. This potentially holds true for the many potential use by academic institutions, consultants, wetland areas that are about to become engulfed by the civil society etc. in order to encourage transparency growing city of Kampala. Kampala is the second fastest and accountability as well as facilitate ongoing growing city in Eastern Africa with an annual population analysis and adaptive management. growth rate of 3.9% (KCCA 2012). In 2012 the population of the Greater Kampala Metropolitan Area (GKMA) was One of the main challenges in achieving the above will be just over 3 million and was projected to reach 5 million institutional. Greater Kampala extends well beyond the in 2020 and 13 million in 2040 (KCCA 2012). The area of boundaries of the KCCA, which originally encompassed developed land within the GKMA increased from 27% the entire city, and unless the KCCA area is adjusted in 1989 to 78% in 2010 (KCCA 2012). Land conversion accordingly (as has been done in other countries), the has already resulted in the substantial loss of wetland problems that will arise in a growing city will be in areas area in Kampala with almost 50% of the original wetland under multiple other jurisdictions. Meanwhile, it may be areas within Kampala’s core city area (KCCA) having more pragmatic for these issues t to be managed by the been converted by 1999, and only about 8% remaining relevant national institutions such as NWSC. highly functioning (KCCA 2014b). Without proactive interventions, the wetlands outside of the present urban core will also be destroyed and the cumulative impacts on Murchison Bay and any economic activities around the bay, including the viability of a future waterfront development, could be significant.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 31 X. REFERENCES African Development Fund 2008. Uganda: Kampala Sanitation Programme Environmental and Social Impact Assessment Summary (ESIA). Prepared by Fichtner, July 2008. Alexander, S. & McInnes, R. 2012. The benefits of wetland restoration. Ramsar Scientific and Technical Briefing Note no. 4. Gland, Switzerland: Ramsar Convention Secretariat. Anderson, S. T., & West, S. E. (2006). Open space, residential property values, and spatial context. Regional Science and Urban Economics, 36:773–789. Anderson, S. T., West, S. E., Geoghegan, J., Jim, C. Y., Chen, W. Y., Yamagata, Y., Monson, M. (2013). Measuring the impact of culture using hedonic analysis. Land Use Policy, 1: 1–50. Armitage, N., Rooseboom, A., Nel, C. & Townshend, P. 1998. The removal of urban litter from stormwater conduits and streams. Water Research Commission (South Africa) Report No. TT 95/98, Pretoria. Armitage, N. 2003. The removal of urban solid waste from stormwater drains. Department of Civil Engineering, University of Cape Town, Private Bag Rondebosch, 7701, South Africa. Armitage, N., Vice, M., Fisher-Jeffes, L., Winter, K., Spiegel, A. & Dunstan, J. 2013. Alternative Technology for Stormwater Management: the South African Guidelines for Sustainable Drainage Systems. Report to the Water Research Commission. WRC Report No: TT 558/13. Balirwa, J.S. 1995. The Lake Victoria environment: Its fisheries and its wetlands – a review. Wetlands Ecology and Management 3: 209-224. Behrer, P. 2010. Building in the mountains: a hedonic analysis of the value of degraded mountain views using GIS modelling. Discussion Paper 2009-15, Cambridge, Mass.: Harvard Environmental Economics Program, May 2010. Billore, S. K., Singh, N., Sharma, J. K., Dass, P. & Nelson, R. M. 1999. Horizontal subsurface flow gravel bed constructed wetland with Phragmites karka in central India. Water Science and Technology 40(3): 163–171. Boxall, P.C. & Adamowicz, W. 2002. Understanding heterogeneous preference: A latent class approach. Environmental and Resource Economics 23: 421-446 Breffle, W. & Morey, E. 2000. Investigating Preference Heterogeneity in a Repeated Discrete- Choice Recreation Demand Model of Atlantic salmon Fishing. Marine Resource Economics 15: 1–20. Brinson, M.M. 2000. Changes in the functioning of wetlands along environmental gradients. Wetlands 13: 65-74. Carlson, R.E. 1977. A Trophic State Index for Lakes. Limnol. Oceanogr. 22: 361-368. Carlsson F., P. Frykblom & C. Liljenstolpe. 2003. Valuing wetland attributes: an application of choice experiments. Ecological Economics 47: 95– 103 Centre for Watershed Protection (CWP) 1998. Costs and Benefits of Storm Water BMPs: Final Report 9/14/98. Centre for Watershed Protection, Ellicott City, Maryland. Cited in Taylor (2005). Cerfonteyn, M. & Day, E. 2010. Diep River Water Quality Study. Draft report to the City of Cape Town. Chiesura, A. 2004. The role of urban parks for the sustainable city, Landscape and Urban Planning. 68: 129-138. City of Cape Town. 2009. Policy on minimising the impact of stormwater from urban development on receiving waters. Roads and Stormwater Department: Catchment, River and Stormwater Management Branch Cooke, G.D., Welch, E.B., Peterson, S.A. & Nichols, S.A. 2005. Restoration and Management of Lakes and Reservoirs. 3rd Edition. Taylor and Francis Group LLC. pp 591. COWI 1998. Kampala Water Quality Monitoring Programme: Murchison Bay Water Quality Project. Report prepared for Ministry of Natural Resources National Water and Sewerage Corporation by COWI in association with VKI, Kampala.                                                  Page 32 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Dallas, S., Scheffe, B., & Ho, G. 2004. Reed beds for greywater treatment – case study in Santa Elena-Monteverde, Costa Rica, Central America. Ecological Engineering 23: 55–61. Day, E. & Clark, B. 2012. Development of a Water Quality Index and Related Reporting Tools and Information for Inland and Coastal Waters of the City of Cape Town. Application of the City of Cape Town Water Quality Index to historical water quality data: 1978 to 2011. Report to the City of Cape Town. Debo, T.N. & Reese, A.J. 2003. Municipal Stormwater Management, 2nd Edition, Lewis Publishers. De Luca, G. 2008. SNP and SML estimation of univariate and bivariate binary choice models. Stata Journal 8: 190–220. De Morais, L. J., Vymazal, J., Bartolomeu, F. & Dias, V. N. 2003. Operation, maintenance and costs of constructed wetlands. In: Proc Conf. The Use of Aquatic Macrophytes for Wastewater Treatment in Constructed Wetlands. Eds. Dias, V. & Vymazal, J. ICN and INAG, Lisbon, Portugal, pp. 315–323. Department of Water Affairs and Forestry (DWAF), South Africa. 2002. South African National Water Quality Monitoring Programmes Series. National Eutrophication Monitoring Programme. Implementation Manual. Department of Water Affairs and Forestry. Pretoria. Department of Water Affairs and Forestry (DWAF), South Africa. 2008. Methods for determining the Water Quality component of the Ecological Reserve. Second Draft. Prepared by P-A Scherman of Scherman Consulting. Department of Water Affairs and Forestry (DWAF). 1996a. South African Water Quality Guidelines (second edition), Volume 7: Aquatic Ecosystems. Department of Water Affairs and Forestry (DWAF). 1996b. South African Water Quality Guidelines (second edition), Volume 2: Recreational Use. Dzikiewicz, M. 1996. The use of macrophytes for wastewater treatment in rural areas of Poland – social aspects. Environ. Res. Forum 5–6, 249–254. Dubgaard, A. 2003. Cost-Benefit Analysis of Wetland Restoration. International Conference “Towards natural flood reduction strategies”, Warsaw. September 2003. Emerton, L., Iyango, L., Luwum, P. & Malinga, A. 1998. The present economic value of the Nakivubo Urban Wetland, Uganda. IUCN – The World Conservation Union, Eastern Africa Regional Office, Nairobi and National Wetlands Programme, Wetlands Inspectorate Division, Ministry of Water, Land and Environment, Kampala. Environmental Assessment Institute (EAI). 2006. Risk and Uncertainty in Cost Benefit Analysis: Toolbox Paper. Version 1.0. ISBN no.: 87-7992-041-1. Ferrini, S. & Scarpa, R. 2007. Designs with a priori information for nonmarket valuation with choice experiments: a Monte Carlo study. Journal of Environmental Economics and Management 53:342-363. Fletcher, T., Duncan, H., Lloyd, S, & Poelsma, P. 2003. Stormwater Flow and Quality and the Effectiveness of Non- proprietary Stormwater Treatment Measures. Draft report for the NSW EPA. April 2003. Cooperative Research Centre for Catchment Hydrology, Melbourne, Victoria. Forbes, G., Dickson, K.R., Golden, T.D., Hudak P. & Doyle, R.D. 2004. Dissolved phosphorus retention of light- weight expanded shale and masonry sand used in subsurface flow treatment wetlands. Environmental Science Technolology 38: 892-898. Fuhrimann, S., Stalder, M., Winkler, M.S., Niwagaba, C.B., Babu, M., Masaba, G., Kabatereine, N.B., Halage, A.A., Schneeberger, P.H.H., Utzinger, J. & Cissé, G. 2015. Microbial and chemical contamination of water, sediment and soil in the Nakivubo wetland area in Kampala, Uganda. Environmental Monitoring and Assessment 187: 475. DOI 10.1007/s10661-015-4689-x. Gelo, D. & Koch, S. 2015. Contingent valuation of community forestry programs in Ethiopia: Controlling for preference anomalies in double-bounded CVM. Ecological Economics, 114:79–89. Graham, P. 2004. Modelling the water quality in dams within the Umgeni Water operational area with emphasis on algal relations. PhD Thesis, North West University.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 33 Greene, W.H. & Hensher, D.A. 2003. A latent class model for discrete choice analysis: contrasts with mixed logit. Transportation Research Part B 37: 681-698 Gunes, K., Tuncsiper, B., Masi, F., Ayaz, S., Leszczynska, D., Findik Hecan, N. & Ahmad, H. 2011. Conctruction and maintenance cost analyzing of constructed wetland systems. Water Practice & Technology 6(3): doi: 10.2166/ wpt.2011.043. Hanley, N. & Spash, C. L. 1993. Cost Benefit Analysis and the Environment. Edward Elgar Publishing Limited, Cheltenham, United Kingdom. Hannemann, W.M. & Kanninen, B. 1996. The statistical analysis of discrete-response CV data. In: Valuing Environmental Preferences: Theory and Practice of the Contingent Valuation Method in the US, EC, and Developing Countries. I.J. Bateman and K.G. Willis (eds), Oxford University Press, Oxford. Hunt, L. M. 2006. Recreational Fishing Site Choice Models: Insights and Future Opportunities. Human Dimensions of Wildlife, 10:153–172. Hunter, G. 2003. Engineering Coordinator Stormwater, Penrith City Council, New South Wales, personal communication. Data provided as part of the CRC s BMP costing survey of Australian stormwater managers. Cited in Taylor (2005). Hynes, S., Hanley, N. & Scarpa, R. 2008. Effects on welfare measures of alternative means of accounting for preference heterogeneity in recreational demand models. American Journal of Agricultural Economics 90: 1011–1027 IRIDRA. 2002. Database of Italian constructed wetlands. Electronic version. Isunju, J. B., Etajak, S., Mwalwega, B., Kimwaga, R., Atekyereza, P., Bazeyo, W. & Ssempebwa, J. C. 2013. Financing of sanitation services in the slums of Kampala and Dar es Salaam. Health 5(4): 783-791. http://dx.doi.org/10.4236/ health.2013.54104. Johnston, C. A., Detenbeck, N. E. & Niemi, G. J. 1990. The cumulative effect of wetlands on stream water quality and quantity. A landscape approach. Biogeochemistry 10: 105 -141. Jordan, T. E., Whigham, D. F., Hofmockel, K. H. & Pittek, M. A. 2003. Wetlands and Aquatic Processes: Nutrient and Sediment Removal by a Restored Wetland Receiving Agricultural Runoff. Journal of Environmental Quality 32: 1534–1547. Kadlec, R.H. & Wallace, S. D. 2009. Treatment wetlands, 2nd edition. CRC Press, Taylor and Francis Group LLC, Boca Raton, Florida. Kakuru, W., Turyahabwe, N. & Mugisha, J. 2013. Total Economic Value of Wetlands Products and Services in Uganda. The Scientific World Journal, Volume 2013, Article ID 192656. Kansiime, F. & Nalubega, M. 1999. Wastewater Treatment by a Natural Wetland: The Nakivubo Swamp, Uganda: Processes and Implications. PhD Dissertation, A.A. Balkema Publishers, Rotterdam, Netherlands. Kansiime, F., Saunders, M. J. & Loiselle, S. A. 2007. Functioning and dynamics of wetland vegetation of Lake Victoria: an overview. Wetland Ecology Management 15:443-451. Kaplan, R. 1983. The analysis of perception via preference: a strategy for studying how the environment is experienced. Landscape and Urban Planning 12: 161–176. Kayima, J., Kyakula, M., Komakech, W. & Echimu, S. P. 2008. A study of the degree of pollution in Nakivubo Channel, Kampala, Uganda. Journal of Applied Sciences and Environmental Management 12(2): 93-98. KCCA 2012. Kampala Physical Development Plan: Updating Kampala Structure Plan and Upgrading the Kampala GIS Unit, Draft Final Report. KCCA 2014a. City Partnerships for Urban Sanitation Service Delivery in Africa and South Asia. Analytical Report Improving Faecal Sludge Management (FSM) for On-site Sanitation in Kampala City, Uganda. Supported by Bill and Melinda Gates Foundation and DFID. August 15, 2014.                                                  Page 34 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala KCCA 2014b. KCCA Strategic Plan 2014/15-2018-2019. Laying the Foundation for Kampala City Transportation. Kiwango, Y.A. & Wolanski, E. 2008 Papyrus wetlands, nutrients balance, fisheries collapse, food security, and Lake Victoria level decline in 2000-2006. Wetland Ecology Management 16: 89-96 Klein, R.W. & Spady R.H. 1993. An efficient semiparametric estimator for binary response models. Econometrica 61: 387–421. Krinsky, I. & Robb, A. 1986. On approximating the statistical properties of elasticities. Review of Economics and Statistics 68: 715– 719. Kriström, B. 1997. Spike models in contingent valuation. American Journal of Agricultural Economics 79: 1013-1023. Kroeger, T. 2008. Open space property value premium analysis. PhD, Conservation Economic Program. National Council for Science and the Environment 2006 Wildlife Habitat Policy Research Program. Kyambadde, J., Kansiime, F., Gumaelius, L. & Dalhammar, G. 2004. A comparative study of Cyperus papyrus and Miscanthidium violaceum-based constructed wetlands for wastewater treatment in a tropical country. Water Research 38: 475-485. Kyambadde, J., Kansiime, F. & Dalhammar, G. 2005. Nitrogen and phosphorus removal in substrate-free pilot constructed wetlands with horizontal surface flow in Uganda. Water Air and Soil Pollution 165: 37–59. La Notte, A., Maes, J., Thieu, V., Bouraoui, F. & Fabio, M. 2012. Biophysical Assessment and Monetary Valuation of Ecosystem Services. Scenario analysis for the case of water purification in Europe. European Commission, Joint Research Centre, Scientific and Policy Reports: EUR 25638 EN. Lane, H. 2004. Water Sensitive Urban Design in the Snowy Monaro Region. Proceedings of the SIA 2004 Regional Conference: Use It or Lose It - It Makes Sense. Shoalhaven, New South Wales. Layard, R. & Glaister, S. 1994. Cost-benefit analysis. Cambridge, UK: Cambridge University Press, pp. 1-56. ISBN 9780521466745 Leinster, S. 2004. Shaun Leinster, Senior Engineer, Ecological Engineering Pty Ltd, Brisbane, Queensland, personal communication. Cited in Taylor (2005). Letley, G. & Turpie, J.K. 2015. The contributions of natural and man-made green open space to property value in Durban, South Africa: a hedonic pricing study. Paper presented at EfD Annual Meeting, Shanghai, November 2015. Lloyd, S.D., Wong, T.H.F. & Chesterfield, C.J. 2002. Water Sensitive Urban Design: A Stormwater Management Perspective. Industry Report No. 02/10. Cooperative Research Centre for Freshwater Ecology, Melbourne, Victoria. Marais, M.J. & Armitage N.P. 2004. The measurement and reduction of urban litter entering stormwater drainage systems: Paper 2 – Strategies for reducing the litter in the stormwater drainage systems. Water SA 30(4): 483-492. ISSN 0378-4738, Water Research Commission. Masi, F., Giovannelli, L., Cortés, V., Innocenti, A. & Pucci, B. 2006. Reed bed treatment systems as an affordable and reliable opportunity for solving sanitary emergencies in developing countries with tropical climate. In Proc. 10th Internat. Conf. Wetland Systems for Water Pollution Control, MAOTDR 2006, Lisbon, Portugal, pp. 1124. McDonald, R.I. & Shemie, D. 2014. Urban Water Blueprint: Mapping conservation solutions to the global water challenge. The Nature Conservancy: Washington, D.C. McFadden, D. 1974. Conditional logit analysis of qualitative choice behaviour. In: Zarembka, P. (Ed.), Frontiers in Econometrics, Academic Press, New York. McKee, L.J., Eyre, B.D. & Hossain, S. 2000. Transport and retention of nitrogen and phosphorus in the sub-tropical Richmond River estuary, Australia — a budget approach. Biogeochemistry 50:241-278. Metcalf & Eddy, Inc. 2003. Wastewater engineering: treatment and reuse. McGraw-Hill.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 35 Ministry of Health, 2010. Annual Health Sector Performance Report 2009/10. Republic of Uganda Mitsch, W.J. & Gosselink, J.G. 1993. Wetlands. 2nd Edition. John Wiley, New York. Mitsch, W. J. & Gosselink, J. G. 2000. Wetlands. 3rd Edition. John Wiley and Sons, Inc. Morrison, E.H.J. & Harper, D.M. 2009 Ecohydrological principals to underpin the restoration of Cyperus papyrus at Lake Naivasha, Kenya. Ecohydrology and hydrobiology 9(1):83-97 Muyodi, F. J., Mwanuzi, F. L. & Kapiyo, R. 2011. Environmental Quality and Fish Communities in Selected Catchments of Lake Victoria. The Open Environmental Engineering Journal 4: 54-65. Nansubuga, I., Banadda, N., Babu, M., Verstraete, W. & Van de Wiele. T. 2013 Effect of polyaluminium chloride water treatment sludge on effluent quality of domestic wastewater treatment. African Journal of Environmental Science and Technology 7: 145-152. National Water and Sewerage Corporation (NWSC) 2013. Annual Activity Report for the Financial Year 2012/13. Performance Review Report for the period July 2012 - June 2013. Document Ref. No: NWSC-HQ/F&A-CP/GR/003. National Water and Sewerage Corporation (NWSC) 2014. Lake Victoria WATSAN project package 5a: concept development and feasibility study for the improvement of the water supply and sanitation situation in informal. Settlement sin Kampala, feasibility study report. September 2014. National Environment Management Agency (NEMA) 1999. National Environment Regulations: Standards for Discharge of Effluent into Water or on Land. Report, National Environment Management Authority (NEMA). National Water and Sewage Corporation (NWSC) 2012. Corporate Plan July 2012 – June 2015. NWSC, Kampala, Uganda. Olsen, N. & Shannon, D. 2010. Valuing the net benefits of ecosystem restoration: The Ripon City Quarry in Yorkshire. Ecosystem Valuation Initiative, Case Study No. 1. IUCN/Aggregate Industries UK. Ooyo, R. 2009. Deteriorating Water Quality in the Lake Victoria Inner Murchison Bay and its impact on the drinking water supply for Kampala, Uganda. National Water and Sewerage Corporation (NWSC), Kampala. Owino, A.O. & Ryan, P.G. 2007. Recent papyrus habitat loss and conservation implications in western Kenya. Wetlands Ecology and Management 15: 1-12 Pearce, D., Atkinson, G. & Mourato, S. 2006. Cost-Benefit Analysis and the Environment. Recent Developments, Organisation for Economic Co-operation and Development (OECD). Platzer, M., Cáceresy, V., Fong, N. & Haberl, R. 2002. Investigations and experiences with subsurface flow constructed wetlands in Nicaragua, Central America. In Proc. 8th Internat. Conf. on Wetland Systems for Water Pollution Control, Arusha, Tanzania, University of Dar es Salaam and IWA, pp. 350–365. Ponnamperuma, F.N. 1972. The Chemistry of Submerged Soils. Advances in Agronomy 24: 29-96. Provencher, B. & Bishop, R.C. 1997. An Estimable dynamic model of recreation behavior with an application to Great Lakes angling. Journal of Environmental Economics and Management 33: 107-127. Rangeti, I. 2014. Determinants of Key Drivers for Potable Water Treatment Cost in uMngeni Basin. Master of Technology Thesis. Environmental Health in the Faculty of Health Sciences, Durban University of Technology. Revelt, D. & Train, K. 1998. Mixed logit with repeated choices: households’ choices of appliance efficiency level. Review of Economics and Statistics 80: 647-657 Rousseau, D. P. L., Vanrolleghem, P. A. & De Pauw, N. 2004. Constructed wetlands in Flanders: a performance analysis. Ecological Engineering 23: 151–163. Saunders, D. L. & Kalff, J. 2001. Nitrogen retention in wetlands, lakes and rivers. Hydrobiologia 443: 205–212. Scarpa, R. & Thiene, M. 2005. Destination Choice Models for Rock-Climbing in the North-Eastern Alps: A Latent-Class Approach Based on Intensity of Participation. Land Economics 81: 426–44.                                                  Page 36 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Scarpa, R., Thiene, M. & Tempesta, T. 2007. Latent Class Count Models of Total Visitation Demand: Days out Hiking in the Eastern Alps. Environmental and Resource Economics 38: 447–60. Shrestha, R. R., Haberl, R. & Laber, J. 2001. Constructed wetland technology transfer to Nepal. Water Science and Technology 43(11): 345–350. Stewart, E. 2005. Evaluation of Septic Tank and Subsurface Flow Wetland for Jamaican Public School Wastewater Treatment. MSc. Thesis, Michigan Technological University. Silva, N.M.L.P. & Braga, J.P.M. 2006. Construction and maintenace costs of constructed wetlands in the centre of Portugal. In Proc. 10th International Conference Wetland Systems for Water Pollution Control, MAOTDR 2006, Lisbon, Portugal. Stalder, M. 2014. Microbial and industrial contamination along the major wastewater chain in Kampala, Uganda. MSc Thesis, Environmental Sciences. Swiss Tropical and Public Health Institute, Swiss Federal Institute of Technology, Zurich. Steiner, G. R. & Combs, D. W. 1993. Small constructed wetlands systems for domestic wastewater treatment and their performance. In: Constructed Wetlands for Water Quality Improvement. Ed. Moshiri, G. A. CRC Press/Lewis Publishers, Boca Raton, Florida, pp. 491–498. Swait, J. 1994. A structural equation model of latent segmentation and product choice for cross-sectional, revealed preference choice data. Journal of Retailing and Consumer Services 1:77–89. Taylor, A. R. D. 1991. Status of Nakivubo and Luzira swamps in Kampala. Report prepared for the Ministry of Environment Protection, Natural Wetlands conservation and management programme, Kampala- Uganda. Taylor, A. 2005. Structural Stormwater quality BMP Cost/Size Relationship Information from the Literature. Urban Stormwater Quality Program, Cooperative Research Centre for Catchment Hydrology. www.toolkit.net.au/Tools/ DownloadPublication.aspx?id=1000020, accessed 22 September 2015. TEEB, 2010. The Economics of Ecosystems and Biodiversity Ecological and Economic Foundations. Edited by Pushpam Kumar. Earthscan, London and Washington. Tebandeke, P. 2013. Determination of pollutant loads from selected urban streams and Nakivubo Channel, Kampala (Unpublished master’s dissertation). Makerere University, Kampala, Uganda. Tilton, D.L. & Kadlec, R.H. 1979. The utilization of a fresh-water wetland for nutrient removal from secondarily treated waste water effluent. Journal of Environmental Quality 8:328- 334. Train, K. 1998. Recreation demand models with taste differences over people. Land Economics 74(2): 230– 239. Ulrich, R.S. & Addoms, D.L. 1981. Psychological and recreational benefits of a residential park. Journal of Leisure Research 13: 43-65. United States Environmental Protection Agency (US EPA). 2000. Constructed Wetlands Treatment of Municipal Wastewater. EPA 625/R-99/010, U.S. Environmental Protection Agency, Cincinnati, Ohio. United States Environmental Protection Agency (US EPA). 2001. National Menu of Best Management Practices for Storm Water Phase II. United States Environmental Protection Agency online guideline. Cited in Taylor (2005). Van Dam, A. Kipkemboi, J., Zaal, F. & Okeyo-Owuor, J. B. 2011. The ecology of livelihoods in East African papyrus wetlands (ECOLIVE). Rev Environ Sce Biotechnol DOI 10.1007/s11157-011-9255-6 Walsh, G. 2001. Gary Walsh, Senior Engineer, Melbourne Water, Victoria, personal communication. Cited in Taylor (2005). Weber, T. 2001, 2002. Tony Weber, Senior Waterways Program Officer (Water Quality), Brisbane City Council, Queensland, personal communication. Cited in Taylor (2005). Wetlands International. 2010. Wetlands & Water, Sanitation and Hygiene (WASH) – understanding the linkages. Wetlands International, Ede, Netherlands.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 37 Wise, C. C. & Armitage, N. P. 2002. The Use of GIS in Litter Generation Modelling and Strategy Determination. Proceedings of the International Waste Management Biennial Congress and Exhibition of the IWMSA, Wastecon 2002, Vol. 1, pp 112 – 120, Durban, South Africa. World Health Organization (WHO). 2006. Guidelines for safe recreational water environments. Volume 1, Coastal and Fresh Waters. Geneva, Switzerland. 219 pp. WSS Services (2015) Development of a pollution management strategy to improve long term water quality status in the Inner Murchinson Bay, Lake Victoria & initiate implementation of its initiatives: Situation Analysis Report Presentation. WSS Services (U) LTD, Kampala, Uganda.                                                  Page 38 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala APPENDIX 1. SELECTED WATER QUALITY MEASURES AND THEIR INTERPRETATION A. General Measures pH Specific Conductivity • Definition: This is the measurement of the • Definition: This is the measurement of the hydrogen-ion concentration in the water. A pH ability of water to conduct an electric current below 7 is acidic (the lower the number, the more - the greater the content of ions in the water, acidic the water, with a decrease of one full unit the more current the water can carry. Ions are representing an increase in acidity of ten times) dissolved metals and other dissolved materials. and a pH above 7 (to a maximum of 14) is basic (the Conductivity is reported in terms of microsiemens higher the number, the more basic the water). per centimetre (µS/cm). Natural waters tend • Importance: High pH values tend to facilitate to vary between 50 and 1500 µS/cm. the solubilisation of ammonia, heavy metals • Importance: Specific Conductivity may be and salts. The precipitation of carbonate salts used to estimate the total ion concentration of (marl) is encouraged when pH levels are high. the water, and is often used as an alternative Low pH levels tend to increase carbon dioxide measure of dissolved solids. It is often possible and carbonic acid concentrations. Lethal effects to establish a correlation between conductivity of pH on aquatic life may occur below pH 4.5 and dissolved solids for a specific body of water and above pH 9.5, although it is noted that some aquatic ecosystems are naturally within these Turbidity ranges and their fauna and flora have adaptations that allow them to withstand these conditions. • Definition: This is a measurement of the suspended particulate matter in a water body which interferes Dissolved Oxygen (DO) with the passage of a beam of light through the water. Materials that contribute to turbidity are silt, • Definition: This is a measure of the amount clay, organic material, or micro-organisms. Turbidity of oxygen dissolved in water. Typically the values are generally reported in Nephelometric concentration of dissolved oxygen in surface water Turbidity Units (NTU). Pure distilled water would is less than 10 mg/L. The DO concentration is have non-detectable turbidity (0 NTU). The subject to diurnal and seasonal fluctuations that extinction depth (for lakes), measured with a Secchi are due, in part, to variations in temperature, disc, is an alternative means of expressing turbidity. photosynthetic activity and river discharge. The • Importance: High levels of turbidity increase the maximum solubility of oxygen (fully saturated) total available surface area of solids in suspension ranges from approximately 15 mg/L at 0°C to 8 upon which bacteria can grow. High turbidity mg/L at 25°C (at sea level). Natural sources of reduces light penetration; therefore, it impairs dissolved oxygen are derived from the atmosphere photosynthesis of submerged vegetation and algae. or through photosynthetic production by aquatic In turn, the reduced plant growth may suppress plants. Natural re-aeration of streams can fish productivity. Turbidity interferes with the take place in areas of waterfalls and rapids. disinfection of drinking water and is aesthetically • Importance: Dissolved oxygen is essential to unpleasant. Drinking water should be < 1 NTU. the respiratory metabolism of most aquatic organisms. It affects the solubility and availability Suspended Solids (Residue, Non-filterable) of nutrients, and therefore the productivity of aquatic ecosystems. Low levels of dissolved • Definition: This is a measure of the particulate oxygen facilitate the release of nutrients from the matter that is suspended within the water column. sediments. Oligotrophic (low nutrient) lakes tend Non-filterable residue values are reported in mg/L. to have increased concentrations of dissolved • Importance: High concentrations of non-filterable oxygen in the hypolimnion (deeper waters) relative residue increases turbidity, thereby restricting to the epilimnion (defined as orthograde oxygen light penetration (hindering photosynthetic profiles). Eutrophic (high nutrient) lakes tend to activity). Suspended material can result in have decreased concentrations of dissolved oxygen damage to fish gills. Settling suspended solids in the hypolimnion relative to the epilimnion can cause impairment to spawning habitat (defined as clinograde oxygen profiles). by smothering fish eggs. Suspended solids interfere with water treatment processes.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 39 Total Dissolved Solids - TDS (Residue, Filterable) Nitrite (NO2-) • Definition: This is a measure of the amount • Definition: This is a measure of a form of nitrogen of dissolved material in the water column. It that occurs as an intermediate in the nitrogen is reported in mg/L with values in fresh water cycle. It is an unstable form that is either rapidly naturally ranging from 0-1000 mg/L. Dissolved salts oxidized to nitrate (nitrification) or reduced such as sodium, chloride, magnesium and sulphate to nitrogen gas (de-nitrification). This form of contribute to elevated filterable residue values. nitrogen can also be used as a source of nutrients • Importance: High concentrations of TDS limit for plants. Nitrite is generally reported in either the suitability of water as a drinking source and µg/L or mg/L. It is normally present in only minute irrigation supply. High TDS waters may interfere quantities in surface waters (<0.001 mg/L). with the clarity, colour and taste of manufactured • Importance: Since nitrite is also a source of products. Drinking water should have <500 mg/L. nutrients for plants its presence encourages plant proliferation. Nitrite is toxic to aquatic Colour, true life at relatively low concentrations. Drinking water limits tend to be about 1 mg/L. • Definition: This is a measure of the dissolved colouring compounds in water. The colour of Nitrate (NO3-) water is attributed to the presence of organic and inorganic materials; different materials absorb • Definition: This is the measurement of the different light frequencies. Colour is expressed as most oxidized and stable form of nitrogen in Pt-Co units according to the platinum-cobalt scale. a water body. Nitrate is the principle form of Water colour can naturally range from 0-300 Pt-Co. combined nitrogen found in natural waters. It Higher values are associated with swamps and bogs. results from the complete oxidation of nitrogen • Importance: Colour is regarded as a pollution compounds. It is generally reported in µg/L or problem in terms aesthetics, but is not generally mg/L. Without anthropogenic inputs, most surface considered a detriment to aquatic life. Increased waters have less than 0.3 mg/L of nitrate. colour may interfere with the passage of • Importance: Nitrate is the primary form of light, thereby impeding photosynthesis. nitrogen used by plants as a nutrient to stimulate growth. Excessive amounts of nitrogen may result B. Nutrients (Nitrogen and Phosphorus) in phytoplankton or macrophyte proliferations. The sources of the following nutrients include At high levels it is toxic to infants. Drinking sewage treatment plant effluents, agriculture, urban water limits tend to be about 10 mg/L. developments, recreation, industrial effluents and mining Total Organic Nitrogen Total Ammonia (NH3 & NH +) 4 • Definition: This is a measure of that portion of nitrogen that is organically bound. Organic • Definition: This is a measure of the most reduced nitrogen includes all organic compounds such inorganic form of nitrogen in water and includes as proteins, polypeptides, amino acids, and dissolved ammonia (NH3) and the ammonium ion urea. It is reported as mg/L. Dissolved organic (NH4+). Nitrogen is an essential plant nutrient and nitrogen can often constitute over 50% of although ammonia is only a small component of the the total soluble nitrogen in fresh water. nitrogen cycle, it contributes to the trophic status of a body of water. Ammonia is generally reported • Importance: Organic nitrogen is not in either µg/L or mg/L. Natural waters typically immediately available for biological activity. have ammonia concentrations less than 0.1 mg/L. Therefore, it does not contribute to furthering plant proliferation until decomposition to • Importance: Excess ammonia contributes to the inorganic forms of nitrogen occurs. eutrophication of water bodies. This results in prolific algal growths that have deleterious impacts on other aquatic life, drinking water supplies, and recreation. Ammonia at high concentrations is toxic to aquatic life.                                                  Page 40 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Total Nitrogen APPENDIX 2. WATER QUALITY AMELIORATION BY WETLANDS • Definition: This is a measure of all forms of Water entering wetlands from developed catchments nitrogen (organic and inorganic). Nitrogen generally has elevated amounts of sediments, nutrients is an essential plant element and is often and pollutants from catchment activities, industrial the limiting nutrient in marine waters. effluents, treated and untreated sewage and other • Importance: The importance of nitrogen in the wastes. Excess nutrients can stimulate algal growth, aquatic environment varies according to the leading to deterioration in aquatic ecosystems, with a relative amounts of the forms of nitrogen present, whole suite of knock-on effects, while other heavy metals be it ammonia, nitrite, nitrate, or organic nitrogen and pathogens pose a risk to human health. (each of which are discussed in detail above). There are a number of different process through which Total Phosphorus wetlands remove sediments, nutrients and pollutants from the inflowing water (Figure A2.1). Nutrients that • Definition: This is a measure of both inorganic are introduced in dissolved form can be taken up directly and organic forms of phosphorus. Phosphorus can by plants and incorporated into plant tissue as they grow. be present as dissolved or particulate matter. It is Most of the phosphorous that is introduced to wetlands an essential plant nutrient and is often the most is attached to sediment and settles to the bottom, limiting nutrient to plant growth in fresh water. where it can remain inactive (Brinson 2000). However, if It is rarely found in significant concentrations in sediments are stirred up then some of this phosphorous surface waters. It is generally reported in µg/L can go back into solution and become available for use or mg/L. The total phosphorus concentrations in by plants. The uptake of dissolved phosphorous will most lakes not affected by anthropogenic inputs continue as long as there is room for further plant growth is generally less than 0.01 mg/L (10 µg/L). (in terms of space, oxygen or plant size limits), after which the system will reach some kind of equilibrium • Importance: Since phosphorus is generally the most in which the uptake is balanced by the senescence, limiting nutrient, its input to fresh water systems death and rotting of plant material which reintroduces can cause extreme proliferations of algal growth. nutrients into the water column (remineralisation). Inputs of phosphorus are the prime contributing At this point there would be no further net uptake of factors to eutrophication in most fresh water nutrients by the wetland unless nutrients are being systems. A general guideline regarding phosphorus exported out of the system (e.g. by harvesting plants or and lake productivity is: <10 µg/L phosphorus dredging and removal of sediments), or unless there is a yields is considered oligotrophic, 10-25 µg/L P natural process of peat formation, which does occur in will be found in lakes considered mesotrophic, Nakivubo wetland (Kansiime & Nalubega 1999). and >25 µg/L P will be found in lakes considered eutrophic. Drinking water limits tend to be 10 µg/L. Nitrogen is removed in wetlands mainly by the nitrification–denitrification process (Saunders & Kalff Orthophosphate (PO4-3) 2001). Nitrification is the microbially-mediated oxidation of ammonium (NH4) to nitrite (NO2) and then nitrate • Definition: This is a measure of the inorganic (NO3). This process consumes oxygen and thus occurs oxidized form of soluble phosphorus. It is in aerobic areas of the wetland. Nitrate then diffuses generally reported in µg/L or mg/L. to anaerobic areas of the wetland where it may be • Importance: This form of phosphorus is the most denitrified. This is the rate-limiting step in the removal readily available for uptake during photosynthesis. of nitrogen from flooded systems. In the denitrification High concentrations of orthophosphate generally process nitrate (NO3) is reduced to gaseous nitrous oxide occur in conjunction with algal blooms. (N2O) and nitrogen gas (N2), which are then released to Source: https://www.for.gov.bc.ca/hts/risc/pubs/aquatic/interp/index.htm the atmosphere (Mitsch & Gosselink 1993). This occurs mainly in sediments with abundant organic matter that provides a carbon source for denitrifying bacteria. Bacteria concentrations are reduced in wetlands by exposure to UV-light. The degree to which this occurs is linked to the duration of water retention within the system. In the Nakivubo wetland, the concentrations of pollutants are also decreased in the lower part of the wetland by dilution with lake waters.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 41 The ability of wetlands to perform water quality APPENDIX 3. RECENT CHANGES TO amelioration services depends on their area and type NAKIVUBO WETLAND of vegetation as well as to their overall health and From the satellite images taken in 2000 and 2015 it can management. Hydraulic efficiency, which is the degree be seen that there has been extensive encroachment to which a wetland disperses inflow over its area, is also around the sides of the Nakivubo wetland (Figure A3.1). important (Jordan et al. 2003). This maximizes contact These mostly consist of informal dwellings as well as area and it can be assumed that it serves to increase expansion of industrial areas. There is also evidence for detention time as well. There is an upper limit to the sediment being dumped in the upper sections alongside amount of pollution that a wetland can remove, as the canal (possible dredge spoils from the canal itself). well as to the amount of pollution that can be added to a wetland without having a significant impact on its The upper Nakivubo wetland was almost entirely under functioning and biodiversity. At high phosphorus loading cultivation already by the year 2000 (Figure A3.2). rates wetlands may eventually become a phosphorus This is in contrast to the maps in Kansiime & Nalubega source rather than a sink (Tilton & Kadlec 1979, Forbes (1999), which indicated a remaining central section of et al. 2004). This also varies seasonally. Wetlands are wetland vegetation in the mid-1990s. The channel is thought to be better at removing total suspended solids, also much less defined in 2000 and is further north than phosphorus and ammonia during high flow periods that in 2015. (when sediment loads entering the wetland increase), but better at removing nitrates during low flow periods The lower sections of the wetland in the most recent (Johnston et al. 1990, McKee et al. 2000). picture have changed quite dramatically from that in 2000 (Figure A3.3). It is clear that some of the papyrus floating islands are breaking apart and seem to have decreased in size. The exact cause of these changes is unknown, however, degradation of the papyrus through cutting, burning or replacing with agriculture could be leading to these patterns observed. On close inspection of the island in the southern end in the 2015 image, extensive agriculture can be seen. This area was identified as a thick Miscanthiduim violaceum mat in Kansiime & Nalubega (1999) and appears mainly intact in the 2000 image. Figure A2.1 Summary of water quality amelioration services by natural systems Source: Turpie, 2015                                                  Page 42 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Figure A3.1 Changes in Nakivubo wetland between 2000 (top picture) and 2015 (bottom picture) from satellite pictures (Google Earth). Red outline indicates soil dumping, blue indicates informal encroachment and yellow indicates industrial expansion.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 43 Figure A3.2 Changes in upper Nakivubo wetland between 2000 (top picture) and 2015 (bottom picture) from satellite pictures (Google Earth).                                                  Page 44 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Figure A3.3 Changes in lower Nakivubo wetland between 2000 (top picture) and 2015 (bottom picture) from satellite pictures (Google Earth).                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 45 APPENDIX 4. POLLUTION INPUTS Sources of pollution affecting the wetland are routed The nutrient loading into the wetland has been described primarily along the Nakivubo channel, which receives in a number of studies. Of these, three estimate the total contaminated runoff and litter from poorly-serviced urban daily loading into the wetland. Kansiime & Nalubega areas abutting the channel, treated sewage effluent from (1999) estimated that 365 000 kg of total N and 28 288 the Bugolobi Waste Water Treatment Works (WWTW), kg of total P were deposited into the wetland annually and treated sewage from the Luzira Prison. during the mid-1990s. Estimates from COWI (1998) which was conducted during a similar period were not Flows have been recorded in various parts of the vastly different (Table A4.2). More recent estimates catchment in the 1990s (Kansiime & Nalubega 1999, from Tebandeke (2013) give an average and a range of COWI 1998) as well as in a recent study (Tebandeke 2013; estimates; however these are from smaller streams as Table A4.1). These data suggest that flows entering the well as the Nakivubo Channel. We therefore assume wetland may have increased from about 73 000 m3 per that the Nakivubo is the highest estimates as we know day in the 1990s (COWI 1998) to about 106 000 m3 per it contributes the most into the wetland. These upper day in 2013 (based on Tebandeke 2013). If this is the limits were about a 50% increase in terms of N and P case, the increase (4% per year) is in line with population loads from 1998 to 2013. This seems feasible in terms of increase in the catchment over the same period. the increased population of Kampala over this period. Households within the catchment receive water from outside the catchment (piped in to houses and water Water quality and sediment data collected by Stalder points), and runoff is increased as a result of hardened (2014) & Fuhrimann et al. (2015) showed that: surfaces. About 92% of the flow into the wetland reaches the railway culvert and flows into the lower part of the • Bacterial contamination decreased along the wetland (COWI 1998). Nakivubo channel with distance from Kampala City; Table A4.1 Flow data for different sites upstream of the Nakivubo wetland from available literature sources. Flow is expressed in m3/day and for some data sources was separated into average flows for dry and wet seasons. Study Sampling Location Dry Season Flow Wet Season Flow Average Flow (m3/day) (m3/day) (m3/day) Kansiime & Railway culvert 50 000-60 000 ~500 000 101 575 Nalubega 1999 COWI 1998 Fire Station 22 464 Kayuanga drainage area 6 048 Kitante drainage area 9 504 Bugolobi WWTW 5 184 Nakivubo Channel (5th street) 72 576 Railway Culvert 66 528 Tebandeke 2013 Kayuanga drainage area 2 851 12 787 Kitante drainage area 3 024 18 317 City Abattoir 778 864 Nakivubo Channel 42 509 170 208 Table A4.2 Nutrient loadings (total nitrogen and total phosphorous) for the Nakivubo wetland (kg/y) from available literature sources. Time of sampling Annual loading N (kg/y) Annual loading P (kg/y) Reference Mid 1990s 365 000 28 288 Kansiime & Nalubega 1999 1997 310 980 38 325 COWI 1998 Nov 2010-May 2012 457 544* 60 366* Tebandeke 2013 * Individual values not given for sampling sites, however assuming highest values correspond to the Nakivubo Channel entering the wetland.                                                  Page 46 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala • With distance further downstream, high APPENDIX 5. CURRENT CONDITION OF levels of point-source contamination of faecal AQUATIC ECOSYSTEMS IN TERMS OF coliform bacteria were however associated WATER QUALITY with inflows from the BWWTW (final effluent This section describes the quality of various sources of E. coli and Salmonella spp. counts ranged from water considered in this study, in terms of its human 3x106 to 9.7x107 (mean 2.5x107) and 1 x 102 health and/or ecological effects, using the guidelines/ – 3.6 x 104 (mean 1.5 x 103) respectively); criteria presented below. • Point source inflows of water contaminated with Salmonella bacteria strains appeared in samples A5.1 Water quality limits and guidelines used from sites further downstream from the BWWTW Interpretation of water quality in the Nakivubo Channel, outlet, and were assumed to be sourced from wetlands and Inner Murchison Bay area was informed informal settlements to the west of the wetland by existing guidelines relating to both human and (Kasanvu, Namuwongo A and B and Yoka zones); ecological health. From an ecological perspective, these • Bacterial contamination decreased with distance hinged mainly on water quality guidelines relating to into Lake Victoria, and with depth from the surface. trophic (nutrient) condition, since this is the key area Kansiime & Nalubega (1999) showed that waste water of ecosystem response to nutrients, considered in the is not evenly distributed through the Nakivubo swamp, present study to comprise one of the main impacts with the bulk of water from the Nakivubo channel affecting the Nakivubo wetland and its potential for funnelling through the central portion of the swamp, future development/beneficial use. The trophic status while the edges of the swamp receive low volumes of of freshwater ecosystems allows them to be broadly waste water, or none at all – this is attributed to the fact classified into one of four trophic categories associated that the edges of the swamp are emergent and attached, with different levels of nutrients (mainly phosphorus whereas the lake-side (downstream) portions of the and nitrogen; Table A5.1). The ranges of key nutrients swamp are floating, and result in channelization of water (phosphorus and nitrogen nutrients) as well as of and (effective) short-circuiting of effluent flows. chlorophyll-a (a measure of phytoplankton (green algae) in the water) that are (broadly) associated with different trophic states in open water systems are provided in Table A5.2. An explanation of water quality variables is given in Appendix 1. Table A5.1 Broad classification of trophic condition (based on DWAF 1996) Trophic status Level of nutrients Generic description Typical situation for lakes Oligotrophic Low Systems with moderate species Clear waters and a rocky or sandy diversity; usually low productivity shoreline. Both planktonic and rooted systems with rapid nutrient cycling plant growth are sparse and no nuisance growth of algae or aquatic plants Mesotrophic Moderate Systems that usually exhibit high Shallow with a soft, silty bottom. Rooted species diversity; productive systems, plant growth is abundant along the prone to nuisance growths of aquatic shores and out into the lake, and algal plants and blooms of (seldom toxic) blooms are not unusual. Water clarity is blue-green algae usually poor. If deep enough to thermally stratify, the bottom waters are often devoid of or low in oxygen. Eutrophic High Systems associated with low species Characteristics between meso- and diversity, highly productive systems, hypertrophic prone to nuisance growths of aquatic plants and blooms of (sometimes toxic) blue-green algae Hypertrophic Extremely high Systems that usually exhibit very Bottom level anoxia is common and the low species diversity; very highly systems are prone to blooms of blue- productive; prone to nuisance growths green as well as green algae of aquatic plants and blooms of (sometimes toxic) blue-green algae                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 47 The ecological condition of a wetland can also be assessed From a human health perspective, WHO (2006) in terms of its similarity to the natural condition. A set guidelines ought ideally to be used with regard to the of guidelines has been devised to determine the health identification of risk thresholds associated with different of southern African systems, expressed in terms of concentrations of bacterial contamination. However, classes A (natural) to F (very highly degraded), based on WHO (2006) guidelines are based on concentrations of ranges of concentrations for key water quality variables intestinal Enterococci bacteria in water. These data were (Table A5.3). Present Ecological State (PES) categories not available for any of the aquatic ecosystems addressed reflect assumed deviation in water quality from natural in this study, and as a result, human health guidelines / reference conditions. The thresholds in Table A5.3 are that covered available data (e.g. South African Water based on data for open water systems in southern Africa Quality Guidelines for Recreational Use (DWAF 1996b) (DWAF 2008). While these have not been tested in East were used in this study. These guidelines, developed for African lakes such as Lake Victoria, it is probable that the application in freshwater environments are based on ranges at least at the degraded end of the scale (that is, concentrations of faecal coliform bacteria. Such data are in the range D to F) apply to most open water bodies. also available for the freshwater ecosystems included in For the purposes of broadly defining target management the present study area. The DWAF (1996b) water quality objectives for treating urban runoff, the values in Table guidelines are thus considered realistically applicable to A5.3 should be adequate. the present study, and are summarised in Table A5.4, with categories of recreational exposure to water as defined/described in Table A5.5. Table A5.2 Summary of ranges of phosphorus and/or nitrogen based nutrients associated with different trophic conditions in in-lake aquatic ecosystems Trophic Average summer inorganic Average summer inorganic Chlorophyll-a (ug/l) state nitrogen concentrations (mg/l) phosphorus concentrations (mg/l) (DWAF 1996a) (DWAF 2002) Oligotrophic <0.5 < 0.015 < 3.5 Mesotrophic 0.5-2.5 > 0.015-0.047 3.5- 9 Eutrophic 2.5-10 >0.047-0.130 9-25 Hypertrophic >10 > 0.130 >25 Table A5.3 Threshold values for key water quality variables for determining Present Ecological State (PES) after Day & Clark (2012) – For all variables except Dissolved oxygen (DO), listed values represent the upper threshold for each PES Category. For DO, values shown represent the lower threshold for each PES Category. TIN is Total Inorganic Nitrogen PES Category PES description (based on deviation Chl-a TOTAL P NH3-N DO TIN from assumed reference condition) A No change 5 0.005 0.015 >8 0.25 B Small change 10 0.015 0.04375 >7 0.7 C Moderate change 20 0.047 0.0725 >6 1 D Large change 30 0.13 0.1 >4 4 E Serious change 40 1 0.12875 >2 10 F Extreme change >40 >1 >0.5 >0 >20                                                  Page 48 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Interpretation of effluent quality data from the BWWTW It should be noted upfront, however, that while adequate has been carried out with reference to effluent limits as guidelines exist for the interpretation of data, the water cited in the (Ugandan) National Environment (Standards quality data for the effluent, channelled, wetland and for Discharge of Effluent into Water or on Land) in-lake environments considered in this study are limited Regulations (S.I. No 5/1999) (Under section 26 and 107 to the results of a few discrete studies, the data for which of the National Environment Act, Cap 153). Data relevant have seldom been presented in a manner that permits to this study are summarised in Table A5.6. easy dissemination or application to other questions/ research areas, and flow data from which loading could be calculated were limited and unreliable. Table A5.4 South African DWAF (2005b) water quality limits for the use of water for recreational purposes. Faecal coliform bacteria (#cfu) Description of effect Full contact Intermediate contact range range Target water quality range 0-130 0-1000 Some risk of gastrointestinal effects Moderate risk of gastrointestinal effects 130-600 1000-4000 Noticeable health effects expected in swimmer and bather population, especially 600-2000 if such events occur frequently Increasing levels of risk of gastrointestinal illness, with only small volumes of > 2000 >4000 water needed to be ingested to be associated with human health impacts # The concentrations of bacteria in water is usually expressed as numbers of colony-forming units (CFU) per 100 ml, which is a measure of the number of viable cells in a particular sample, where a colony represents an aggregate of cells derived from a single progenitor cell Table A5.5 Categories of recreational use of a water body (DWAF 1996b) Category Description Full contact Full-body water contact, and includes full immersion activities such as swimming and diving. recreation Intermediate All forms of contact recreation excluding those listed under full contact recreation. Includes some activities contact recreation which involve a high degree of water contact (e.g. water-skiing, wading and wind-surfing) as well as those which involve relatively little water contact (e.g. canoeing and angling). Compared with the above, full immersion is likely to occur only occasionally and among novices of a water sport in respect of the latter category, the age of users (water sports such as water-skiing and windsurfing are usually practised by adults rather than by young children), and health status of users (strenuous water sports are generally practised by water users in a fairly good state of health). Non-contact All forms of recreation which do not involve direct contact with water such as picnicking and hiking recreation alongside water bodies, and scenic appreciation of water by those residing or holidaying on the shores of a water body. These activities are primarily concerned with the scenic and aesthetic appreciation of water. Table A5.6 Effluent Limits as cited in the Ugandan National Environment (Standards for Discharge of Effluent into Water or on Land) Regulations, S.I. No 5/1999 (Under section 26 and 107 of the National Environment Act, Cap 153). Only variables referred to in this study are included Variable Concentration Variable Concentration Biological Oxygen Demand (BOD) 50 (mg/L) Total phosphate (mg/L) 10 Chemical Oxygen Demand (COD) 100 (mg/L) Orthophosphate (PO4-P) (mg/L) 5.0 Total nitrogen 10 (mg/L) Total suspended solids (TSS) (mg/L) 100 Coliform organisms 10 000 counts/100 ml                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 49 A5.2 Water from the BWWTW A5.3 The Nakivubo channel Water quality data for the first six months of 2015 Useful data for the channel are limited, and the most only were available for BWWTW final effluent, and are recent, useable data that could be sourced are presented presented graphically in Figure A5.1. No reliable water in Table A5.7 and Table A5.8, with assumed site flow data were available, although it is understood that locations as shown in Figure A5.3. the works when upgraded would treat up to 45 ML/day. The effluent quality data, interpreted with reference to These data suggested that: Table A5.6, suggest that: • Concentrations of both orthophosphate • Out of the 24 samples analysed, total (PO4-P) (the dissolved, biologically available phosphorus concentrations fell within legal part of total phosphorus) and total nitrogen fell effluent discharge limits only seven times; well within the range considered associated with hypertrophic conditions in aquatic • COD concentrations were also problematic, ecosystems, as outlined in Table A5.2. only complying in six samples, although BOD showed better compliance, with nine • Concentrations of orthophosphate were variable, samples only exceeding legal limits; but increased substantially overall with distance downstream, as did loading of total phosphorus • Total suspended solids complied with (TP), suggesting that additional sources of this legal limits in just eight samples; nutrient entered the channel along its length • Effluent data generally showed a reduction at least as far as the railway line culvert; in concentrations of TSS, faecal coliforms • Total ammonium nitrogen concentrations were and orthophosphate in April / May elevated well above target concentrations for samples, assumed to reflect a dilution this variable (e.g. DWAF 1996) but, assuming pH effect of rainfall on effluent streams. <7 and temperatures < 20⁰C, were all still below In the event that effluent data comprised the full stream the range at which concentrations of un-ionised of water in the channel, it would be associated with the ammonia (NH3) would reach assumed “chronic following aquatic ecosystem attributes (interpreted from toxicity” concentrations for aquatic ecosystems. Table A5.2 and with reference to DWAF 1996a): However, given that high rates of photosynthesis may increase pH, at least during the day, high • Orthophosphate concentrations were concentrations of total ammonia in association always at least an order of magnitude with orthophosphate (which increases algal greater than the threshold for hypertrophic productivity) should still be viewed with concern; conditions and if this water comprised in- lake water, it would fall within the range of • Biological Oxygen Demand was relatively high a Category F, as described in Table A5.3; in the channel (nearly double the maximum legal limit for waste water effluent at the two • Orthophosphate usually comprised more upstream sites; see Table A5.6) and reflected than 50% of total phosphorus, indicating that high levels of organic material in the channel; removal efforts would need to allow for large levels of uptake by plants, rather than focus • Suspended sediment concentrations were highest only on sedimentation of particulate material; in the upstream channel sites, but dropped substantially (by an order of magnitude) with • Faecal coliform data were always in excess of safe distance downstream. Sediment loading increased limits for even intermediate contact recreation; at the 5th Street site, reflecting large volumes Dissolved oxygen data, although limited to two samples, of diluted inflows from the BWWTW upstream showed levels that were critically low (as per DWAF of this site. The fact that there was a substantial 1996a), and likely to pose acute threats to non air- reduction in TSS load by the railway line site breathing aquatic life. They fell within the PES range for a is indicative of high levels of sedimentation in Category F for this variable (Table A5.3). the channel upstream. Such sedimentation, along with some degree of instream biological processes, is assumed to account also for the reduction in BOD loading by the railway culvert;                                                  Page 50 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Figure A5.1 Water quality data for final effluent from the BWWTW                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 51 Table A5.7 Undated water quality data for the Nakivubo channel after COWI (1998) TN TP PO4 BOD TSS Flow TN TP NO3 NH4 PO4 BOD TSS Station pH EC (kg/ (kg/ (kg/ (kg/ (kg/ (m3 (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) day) day) day) day) day) day) Fire 6.7 456 11.54 2.03 0.13 3.09 0.4 98.3 260 239 42 13 2.038 5 840 22464 Station 5th 6.89 497 11.99 1.69 0.11 2.58 0.13 98.6 172 852 105 67 10.195 12 483 72576 street Railway 6.79 429 8.88 1.67 0.05 3.45 0.68 44.1 26 755 112 39 3.274 1 729 66528 Culvert Figure A5.2 Nakivubo channel at 5th Street Bridge • The substantial increase in loading of total The data described above do not reflect any seasonal phosphorus with distance downstream reflects trends. Descriptive data presented in Kansiime & both the fact that additional sources of phosphorus Nalubega (1999) suggest, however, that water quality enter the channel from informal settlements may deteriorate in the channel as a result of runoff – between the 5th Street site and the railway, and this scenario is typical of many urban catchments with that the proportion of total phosphorus that poor servicing and high levels of informal settlement was dissolved rather than in particulate form (e.g. Cerfonteyn & Day 2012), because instead of increased with distance downstream as well, diluting pollution streams, rainwater may simply wash thus reducing the efficiency of sedimentation contaminants from the broader catchment into the as a process for phosphorus removal. watercourse. Even if there is some dilution impact, net pollutant loading increases. No data indicating this process were available for the study area.                                                  Page 52 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Table A5.8 Effluent Limits as cited in the Ugandan National Environment (Standards for Discharge of Effluent into Water or on Land) Regulations, S.I. No 5/1999 (Under section 26 and 107 of the National Environment Act, Cap 153). Only variables referred to in this study are included Concentrations Value Range Variance Range Value Range Variance Range (all streams and channels) pH 6.3-7.6 6.6-8.4 DO (mg/l) 0.0-3.0 0.1 0.0-2.2 1.2 EC (uS/cm) 215-1597 100-523 252-2028 60-897 TSS (mg/l) 61-877 12-350 25-514 10-44 Temp (°C) 23-25.5 1-0.9 24.9-28.9 0.3-0.9 TN (mg/l) 2.28-164.07 0.31-22.85 3.49-173 0.84-3.1 TP (mg/l) 0.77-23.33 0.24-11.55 0.94-38.18 0.51-9.71 COD (mg/l) 100.7-1480.5 5.58-582.4 96.19-1352.48 31.44-880 BOD (mg/l) 82.28-839.4 18.36-372.4 69.31-1108.05 34.01-333.16 FC (cFu/100ml) 3.8*10 -2.1*10 6 7 3.5*10 -1.3*10 5 7 2.6*10 -5.4*10 6 7 1.5*106-4.8*107 Figure A5.3 2015 Google Earth Image showing main features referred to in this section. Yellow labels indicate assumed locations of water quality sites referred to in the text and Table A5.7 and Table A5.8.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 53 A5.4 The Nakivubo wetlands Data collected by Stalder (2014) showed that both the Water passing along the Nakivubo channel at least as upper and lower portions of the wetland are affected by: far as the railway line generally carries high loads of phosphorus and nitrogen nutrients, as well as sediments. • Elevated faecal coliform bacteria, including The high chemical and biological oxygen demands (COD Escherichia coli, as well as Salmonella and BOD) of this waste stream are likely to contribute as a result of receipt of contaminated to downstream effects in the lower wetland, including water from the Nakivubo channel; reduced levels of dissolved oxygen and high levels of plant productivity (Kansiime & Nalubega 1999). • Parasitic nematode eggs and larvae, assumed to derive from raw and treated waste water The dataset presented in Table A5.7 does not include - hookworm and several other nematode bacterial data. Data in Table A5.8 are of limited value, species were detected in the channel, the in that they do not allow differentiation between sites, lower wetland and its Murchison Bay outlet but they do indicate extremely high counts of faecal at concentrations above the safe limits bacteria in the samples, with even minimum values being recommended by WHO (2006), while Ascaris well above high risk thresholds for intermediate contact lumbricoides (giant roundworm) were present recreation, suggesting that at all times during the period in the channel, but not in the lower wetland; sampled by Tebandeke (2013), contact with water in the • Heavy metal contamination of both water and channel would have been a potentially high risk activity sediments as a result of industrial pollution, from a human health perspective. This interpretation is with copper, iron and cadmium all occurring at supported by (again, non-site-specific) data from Stalder concentrations that exceeded cited NEMA (1999) (2014) for both bacterial and human parasite data. maximum acceptable concentrations in the channel, the lower wetland / swamp and the lake shores, The condition of the lower Nakivubo wetland is considered to be deteriorating, with shrinkage • Elevated suspended solids, orthophosphate and of areas dominated by indigenous wetland plant total ammonia concentrations, as well as high levels communities (e.g. Cyperus papyrus community), as well of both Biological and Chemical Oxygen Demand as the detachment of substantial portions of floating (BOD and COD), associated with organic waste – the Miscanthidium violaceum islands and their passage into study did not however indicate the degree of spatial the greater Murchison Bay area, as described above. variability in concentrations of these variables in the channel, lower wetland and lake environments. Figure A5.4 View of the Nakivubo wetlands Figure A5.5 View across the lower wetland towards Murchison Bay.                                                  Page 54 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala A5.5 Murchison Bay Between the edge of the Nakivubo wetland and the No useful recent water quality data for Murchison Bay Ggaba intake, any pollution undergoes dilution through at the outlet of the Nakivubo channel were sourced in mixing with the lake water. The levels of BOD, TP, EC this study. Although water quality data were available for and Ammonium decreased by 51%, 96%, 67% and 98% the Ggaba Water Treatment Plant in the bay, even these respectively (Ooyo 2009). However, there are other did not include nutrient or oxygen data. Visual evidence potential sources of pollution within Inner Murchinson from Google Earth imagery (e.g. Figure A5. 2) suggests Bay that will also contribute to the levels of pollution however that water in the bay is affected by periodic encountered at the Ggaba water intake plant. In 2001, algal blooms, consistent with hypertrophic in-lake levels of NH4+ detected in the Inner Murchinson Bay conditions, as described in Table A5.2, and it is assumed were isolated to the area at the edge of the Nakivubo on this basis that the bay is subject to substantially wetland with values measuring 2-3 mg/l (Table A5.6). elevated concentrations of at least phosphorus nutrients. By 2010, most of the northern section of the Inner Murchison Bay measured at least 5 mg/l, and in 2014, Kansiime & Nalubega (1999) did measure water quality in around the edge of Nakivubo wetland and around Port the bay, with distance from the lower Nakivubo wetland. Bell these measurements were about 10 mg/l. Levels of Although this study did not provide raw data, they noted total Phosphorous were quite evenly distributed across the following: the Inner Murchison Bay during 2001, however by 2011 high values (1mg/l) were concentrated in the northern • Temperature, pH and Dissolved Oxygen half of the bay (Table A5.6). increased with passage from the wetland interface to the open waters of the bay; The only apparent source of Total Suspended Solids (TSS) • Water was still hypoxic at 750 m offshore in 2001 were coming from a source near to the Ggaba from the wetland interface; intake, which measured less that 100mg/l (Figure A5. 8). There were no detectable TSS coming from the edge • Conductivity, ammonium-nitrogen and total of Nakivubo wetland. By 2010 this pattern had however phosphorus concentrations sharply decreased in completely shifted and the highest levels in the Inner the open waters of the bay, just 1 km offshore, Murchison Bay came from the edge of the Nakivubo presumably as a result of dilution – the graph wetland measuring up to 400mg/l. provided by these authors appears to indicate a range of approximately 1.0 to 1.5 mg P/l for Measurements of Biological Oxygen Demand were samples taken 750m from the shore, reducing to low during 2001 (less than 5 mg/l) (Figure A5. 9). a range of approximately 0.1-0.6 mg/l for samples Measurements in 2014 indicated that comparatively 2.5 km from the shore. Interpreting these data much higher levels of BOD were encountered across in terms of the ranges presented in Table A5.2, the Bay with measurements highest (over 20 mg/l) even the latter are however still clearly in the around the Nakivubo wetland edge, Kasanga wetland hypertrophic zone for orthophosphate, noting immediately south of Nakivubo and Wankolokolo however that the above authors presented data for wetland to the east of Port Bell. total phosphorus only and not orthophosphate; • On the basis of Table A5.3, the above data would all place Murchison Bay in a Category F in terms of phosphorus, and a Category B in terms of un- ionised ammonia. Data for the other variables considered in Table A5.3 were not available.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 55 Figure A5.6 Ammonium measurements in 2001, 2010 and 2014 across Inner Murchison Bay (WSS Services 2015). Figure A5.7 Total Phosphorous measurements in 2001 and 2011 across Inner Murchison Bay (WSS Services 2015).                                                  Page 56 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Figure A5.8 Total Suspended Solids measurements in 2001 and 2011 across Inner Murchison Bay (WSS Services 2015). Figure A5.9 Biological Oxygen Demand measurements in 2001 and 2014 across Inner Murchison Bay (WSS Services 2015).                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 57 APPENDIX 6. ASSESSMENT OF Step 1. Prevent pollution from catchment INTERVENTIONS REQUIRED TO MEET area WATER QUALITY AND RECREATIONAL Existing sanitation systems and projects OBJECTIVES Currently the piped sewerage coverage is only between The objectives of the rehabilitation measures were 6.5 and 10% of households in Kampala (NWSC Corporate assumed to be as follows: Plan 2012). The remaining population depends on various forms of on-site sanitation: pit latrines (55- 65%), improved (VIP) pit latrines (27.5%), septic tanks 1. Effect a measurable improvement in the quality (20%), public toilets (1%), and open defecation (African of water passing out of the Nakivubo wetland into Development Fund 2008, KCCA 2012, 2014a). There is Murchison Bay inadequate collection and disposal of faecal sludge that is generated in pit latrines, with only 43% (390m3 per day) a. Improvement of water quality to a Category D or of the faecal sludge generated being collected for proper better (Table A5.3) would be recommended for all disposal. Statistics for the population living within the ecological variables, catchment area of the Nakivubo wetland are unknown, however. Based on 2014 census data, an estimated 324 b. The “low risk” category for at least intermediate 000 people live in about 77 100 households within the contact recreational use of water (see Table A5. catchment, which if serviced would generate about 65 4) is assumed to be a necessary target in terms of ML4 of sewage per day (assuming 200 litres/person/day). faecal bacteria counts; This does not include the effluent generated by industry. 2. Ensure sustainable management of the existing The National Water and Sewerage Corporation (NWSC) in Nakivubo wetland; Kampala currently operates two waste water treatment works in the form of the conventional sewage treatment 3. Reduce impacts on human health as a result of works at Bugolobi with a capacity to treat up to 12 ML exposure to per day and waste water stabilisation ponds at Lubigi (outside the Nakivubo catchment area) that have a a. faecal bacteria, parasitic nematodes and other combined capacity of 5.4 ML per day. The Bugolobi pests associated with exposure to human faecal WWTW currently receives piped sewage predominantly waste; from the central business district in Kampala and extends to areas of Old Kampala, Mengo, Katwe, Nsambya, Kibuli, b. blooms of algae including blue-green algae Mbuya, Nakawa, Naguru, Bukoto and Kamwokya (NWSC (Murchison Bay); website, accessed 2015), as well as from medium and large industrial facilities (COWI 1998). 4. Open up opportunities for the safe recreational use of the lower wetland. 1 ML = 1000 m3 = 1 million litres 4 Each intervention in the treatment train is presented in detail below (Step 1 – 6). Information about the measures required, design specifications and costs, and the expected outcomes for each intervention are described. The approaches are presented here in the order in which they need to be implemented if there is to be any chance of achieving these objectives.                                                  Page 58 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala During the period 2009 - 2012 funding was secured for Another major project is underway to address the water the Kampala Sanitation Programme (KSP) which forms supply and sanitation situation for Kampala’s informal part of the larger Lake Victoria Protection Project (LVP; settlements. This project, the Lake Victoria WatSan NWSC Annual Activity Report 2013). The LVP was to Project (NWSC 2015) has identified six different aspects be implemented in two separate phases. LVP Phase that should be addressed: 1) water supply, 2) sanitation 1 reportedly included the completion of the Lubigi and hygiene marketing, 3) faecal sludge management, Sewage and Faecal Sludge Plant and the upgrading 4) sewerage, 5) schools and 6) public toilets. The project of the Bugolobi WWTW, at a total cost of about €15 does not go as far as to install pit latrines for households, million (NWSC Annual Activity Report 2013), although but does market them. The faecal sludge management the latter had not occurred by the time of this study, aspect includes a collection service, transfer stations and possibly due to a change in plan (see below). The LVP treatment plants. While the project is for Kampala as a Phase 2 incorporates major works to construct and whole, based on the percentage of the area of each of operate a new Nakivubo Waste Water Treatment Plant the parishes inside the Nakivubo catchment, we estimate (WWTP), construct a Kinawataka WWTW, rehabilitate that approximately €1.16 million will be spent on and extend the Nakivubo Sewer Network and construct upgrading water supply and €3.41 million on upgrading the Kinawataka Sewer Network (NWSC Annual Activity sanitation within the informal areas within the Nakivubo Report 2013). Kinawataka is to the north of the Nakivubo catchment. Estimates for improving sewerage were only catchment area, and drains into the Murchison Bay given for two parishes, only one of which, Kibuli, was to the east of Nakivubo and Port Bell. The cost of this within our catchment. phase of the project was reported to be €84 million with the funding coming from KfW (€10 million), Further measures required African Development Bank (AfDB; €38 million) and the In order to prevent raw sewage from entering the Government of Uganda/NWSC (€36 million; NWSC wetland and Murchison Bay, both sanitation coverage Annual Activity Report 2013). and WWTW capacity will need to be further increased. Due to soil conditions, the plans for construction of Sanitation measures ought to be in place as minimum the new Nakivubo WWTW had to be moved from the standards in any high density urban environment. For a original site near the lower reaches of the Nakivubo situation such as currently found in Kampala, there is an wetland to Bugolobi, at the site of the existing WWTW. urgent requirement for improving sanitation through: The construction of the Kinawataka WWTW was put on a. Retrofitting unserviced areas with appropriate hold and the scope scaled down as it extended beyond infrastructure to allow for the conveyance and the available budget (NWSC Annual Activity Report treatment of waste water including sewage 2013). The LV WatSan Sanitation Plan states that the generated in developed areas of the upstream original Kinawataka WWTW would not be constructed catchment; due to limited investment costs but would be replaced by a pre-treatment facility and pumping station which b. Expansion of existing sewage works capacity as will transport sewage to the new Nakivubo WWTW required to accommodate the additional sewage at the Bugolobi site (NWSC 2014). The new Nakivubo volume generated by; and and Kinawataka Sewers are reported to be under construction with one third of the pipes having been laid. c. Effective long-term maintenance of new and Phase 2 of the LVP was due to be completed towards the existing sewage collection, conveyance and end of 2015. However, construction had just begun at treatment facilities, including long-term policing the Bugolobi WWTW at the time of a site visit (October and pollution tracking. 2015), and will probably take another two years. It is however recognised that the financial costs of With the construction of new WWTW at Lubigi and retrofitting unserviced areas will be high, though Nakivubo, plus new sewers in Nakivubo and Kinawataka noting that such cost analyses often ignore hidden cost areas, the sanitation coverage is expected to rise to offsets such as human health benefits. The planned 30% of households in Kampala (NWSC 2014). The new interventions described above will substantially increase Nakivubo WWTW is planned to increase overall capacity sanitation coverage and WWTW capacity, but will still to 45 ML per day. If this serviced the Nakivubo catchment only meet a fraction of what is needed from a human only, this could meet two-thirds of demand. However, health, let alone environmental, perspective for Kampala this needs to deal with waste water from the Kinawataka as a whole. While full coverage is not likely to be feasible, catchment as well. it is more reasonable to aim for waterborne sanitation coverage of 60-70%, while implementing cheaper measures such as VIP latrines in the remaining areas and having waste water treatment works that can handle waste water flows channelled from those areas.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 59 While the WatSan Project aims to improve sanitation Using a similar approach, we derived an estimate which in these areas, it does not include the installation (or included the whole catchment area rather than only the subsidisation) of pit latrines nor will it achieve full directly neighbouring parishes, based on GIS data of the servicing of currently under-serviced areas in terms catchment and parishes and 2014 census data. It was of sludge removal. These measures generally need to assumed that 87.4% of the population had pit latrine go hand in hand with a comprehensive public health access (Ministry of Health 2010) and only the remaining education campaign. Such sanitation measures are being 12.6% required new pit latrines. Updated prices for addressed by the WatSan Project and are not considered construction were obtained from Isunju et al. (2013), and further here. the cost of having the sludge removed from the pit latrines twice a year (Isunju et al. 2013) was included for the new The planned capacity of the Nakivubo WWTW could latrines as well as for the estimated 40% of the existing provide an acceptable level of service for the Nakivubo latrines that are not regularly serviced (KCCA 2012). This catchment. However, because the Nakivubo WWTW yielded an estimated capital cost of $1.15 million and will have to deal with waste water from Kinawataka, the additional annual costs of $0.67 million (Table A6.1). combined treatment capacity for these two catchments will need to be increased in order to meet this level. The study by Isunju et al. (2013) suggested that although the willingness to pay for operations and In addition, attention needs to be paid to catchment maintenance of public sanitation facilities was high management functions such as street sweeping and litter (61.2% of respondents in a Kampalan slum), this was collection, to prevent sediments and litter from entering also very dependent on level of awareness, costs of visits the storm water system. It is therefore estimated that the and the management of the facility. Due to the high following additional measures would be required in order costs involved in setting up new facilities, the transient to significantly reduce polluted inflows into the Nakivubo residence of people in slums and the socioeconomic wetland from the catchment: status of most people in these areas, the overall willingness to pay fell short of the funds required to maintain the infrastructure. This is evident in the low a. Improve sanitation in areas that are not serviced level of maintenance of existing latrines. There is thus by water borne sewage a need for public funding to extend to the financing of these. This is considered crucial to prevent the high loads b. Further expand WWTW capacity of pollution that are washed into the Nakivubo system during the rainy season. c. Street sweeping and litter collection In Kampala, there is inadequate collection and disposal Design and costs of faecal sludge generated by pit latrines, and existing Improve sanitation faecal sludge treatment plants (FSTP) across the city are The additional investment required for installation and overwhelmed. With the increase in pit latrine access servicing of pit latrines and the costs associated with the and additional servicing in the Nakivubo catchment, it is construction of a new faecal sludge treatment facility in deemed essential that a new FSTP be constructed to deal the catchment were estimated. Emerton et al. (1998) with increased outputs of sludge from informal areas. had estimated that the cost of improving sewerage and Approximately only 43% (390 m3 per day) of faecal sludge sanitation facilities in low-cost areas adjacent to Nakivubo generated in Kampala is being collected and disposed of would be in the order of 97.59 USh million/year. That properly (KCCA 2014a). This suggests that if sludge from study assumed that about one third of the population of pit latrines was collected with a 100% coverage, the total each of the directly surrounding parishes were considered amount generated per day and needing disposal would low-income; that population and number of households be 907 m3 . Based on this information and the total in each parish had increased since the 1991 census data number of households in the catchment area using pit at the average population growth rate; that elevated pit latrines, it was estimated that a 200 m3 FSTP would be latrines would need to be constructed at a 1999 price of needed to adequately address the treatment of sludge USh 625 000 each; that each pit latrine is shared between from pit latrines in the catchment. Cost estimates from five households; and that the lifespan of each pit latrine is the LV WatSan Project (NWSC 2014, Annex 11) for the approximately 10 years. construction of a 400m3 per day FSTP were used to determine the costs associated with a FSTP of half that size (200 m3 per day). The total construction cost was estimated to be $1.44 million, with annual maintenance costs of around $0.07 million.                                                  Page 60 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Table A6.1 Estimated costs of improving pit toilet access and servicing within the Nakivubo catchment area based on method from Emerton et al. (1998). Total # households 77 102 % without sanitation 12.6% Additional % unserviced 40% Installation ratio (hh: pit latrines) 5 Average cost of installation per latrine (range) $598 ($125-1250) Average cost of biannual sludge removal per latrine (range) $23.58 ($17-85.8) Initial installation costs (based on average prices) $1 152 000 Annual maintenance cost (based on average prices) $675 000 Waste water treatment capacity Street sweeping and litter collection Given that the new Nakivubo 45 ML WWTW will be Sediments and litter are ubiquitous problems in urban receiving sewage from both the Nakivubo and Kinawataka environments. They accumulate until they are either catchments, it is assumed that the capacity of the new manually removed or are transported by the wind and/ plant may not be sufficient. It is difficult to estimate or stormwater runoff into the drainage system. Once in exactly how much sewage will be pumped to the WWTW the drainage system, they can contribute to blockages from these areas as there is little information about the and increased flood risk, as well as providing health risks. exact number of sewer connections in either of these These problems should ideally be managed as part of an catchments or the extent and location of the sewer integrated catchment management strategy which includes network. Based on census data it is estimated that the planning controls (e.g. restrict use of certain areas), source Nakivubo catchment alone generates around 65ML per controls (e.g. education programmes, litter bins), and day. If we assume a long term target of 70% coverage structural controls (stormwater treatment and litter traps). in the Nakivubo catchment whilst implementing other sanitation measures such as increased pit latrine access In South Africa, studies indicate that sweeping once and servicing, and improved sludge treatment facilities, a day removes about 83% of litter, whereas sweeping then the design capacity of 45 ML is expected to be three times a day can remove as much as 99% (Armitage sufficient. However, due to changes in the planned et al. 1998, Marais & Armitage 2004). The efficiency is Sanitation Program, and the now increased flow of sewage strongly linked to the frequency of sweeping relative to from adjacent catchments to the Nakivubo WWTW via the frequency of stormwater-producing rainfall events pumping stations, it is assumed that the overall capacity (greater than about 5mm rainfall). This means that effort of the plant may be exceeded. The original Sanitation Plan could vary seasonally, but must prepare for the “first included a new WWTW in the Kinawataka catchment with flush” after the dry season. The method is also important. a design capacity of 8 ML. If one assumes that this WWTW Use of street flushing, for example, would exacerbate the was designed based on the catchment population size and problem. Numerous studies have been undertaken on sanitation targets then it is expected that an extra 8 ML of litter generated in urban environments. Based on these, sewage will be required. Wise & Armitage (2002) estimated litter loads generated by a variety of residential, industrial, retail and other urban Based on a database of capital costs of WWTW typologies. These estimates, together with generalised constructed throughout Africa maintained by GIBB costs in Armitage (2004) suggest that as much as 3800 Consulting Engineers in Cape Town, the cost of tonnes of litter may be generated annually in the Nakivubo constructing a new WWTW is expected to be in the order catchment (Table A6. 2). of $1.1 million per ML of works constructed. Ongoing maintenance of the plants are estimated at 2% of the In order to achieve a street sweeping efficiency of 60%, total capital portion of the WWTW for civil and building this would incur an annual cost of about $2.4-4.8 million works, and 3% for the mechanical and electrical portions per year. Assuming that no litter traps are in place in the of the WWTW. Therefore increasing the capacity of the catchment, and that street sweeping currently achieves an Nakivubo WWTW is estimated to cost approximately $8.8 efficiency of 50%, the residual requirement for sweeping million, with maintenance required for this increase in would be approximately $0.4 – 0.8 million per annum. plant capacity estimated at $0.44 per annum.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 61 Expected outcome The outcome of these measures along with the already It should be emphasised, however, that all of these planned improvements in sanitation would be to data provide broad comparisons only, and moreover significantly improve the quality of water entering the do not allow for the change in discharge that would be wetlands via the Nakivubo Channel. In addition, dry associated with diversion of effluent to the sewers, which season runoff would be reduced, and reflect genuine would mainly affect low flow data. The results of these stormwater flows, and not dry weather pollution streams. comparisons are provided in Table A6. 3 in terms of both Further polishing of stormwater would be possible in the loading and stormwater concentrations, and presented lower Nakivubo wetland as outlined for Step 4. graphically in Figure A6. 1. The latter shows the data as concentrations, in relation to the trophic state guidelines In order to quantify changes in loading of major water presented in Table A5.2. quality variables into the lower wetland and Murchison Bay as a result of implementation of this activity, a TSS Data in Table A6. 3 indicate that an assumed 80% TSS removal efficiency of 80% is considered a reasonable removal rate upstream would have a significant effect target objective. This target is in line with typical on instream water quality (concentration) – this positive Sustainable Urban Drainage System (SUDS) approaches outcome would be passed on to the downstream to urban stormwater management (e.g. Debo & Reese environment of the lower Nakivubo wetland and 2003), with the latter demonstrating treatment train Murchison Bay (loading). The most useful representation efficiencies of 61-95% for TSS removal. An important of the outcome of this measure is indicated in Figure A6. component of TSS removal from stormwater in the 1, which shows the current highly eutrophic status of Kampala area is the assumed corresponding removal water in the channel. This would conceivably be reduced of particulate phosphorus- and nitrogen-enriched to mesotrophic conditions in the case of nitrogen organic sediments, thus addressing in part nutrient nutrients and, in the case of the more problematic enrichment and chemical and biological oxygen demand phosphorus nutrients associated with phytoplankton issues. Debo & Reese (2003) also demonstrated total blooms in the downstream environment, from strongly phosphorus removal rates of up to 89% (group median hypertrophic to near-eutrophic levels. Such changes 34 ± 33%) for different treatment options and removal would be considered of significant magnitude to have of 45% total phosphorus is a standard requirement in a good likelihood of resulting in measurable change the stormwater management policies of some cities in terms of downstream plant production, although (e.g. City of Cape Town 2009: Management of Urban phosphorus concentrations would remain high. Dilution Stormwater Impacts Policy). effects from lake water and lake water seiches would moreover also be likely in the bay and lower wetland The (albeit very limited) water quality and flow data respectively, and would further improve water quality if available for the Nakivubo Channel in the present project not nutrient loading. were used to estimate possible outcomes in terms of stormwater quality, assuming implementation of this Reduced organic loading, reflected in assumed significant treatment option. Estimated wet- and dry-season 5 reductions in BOD and COD in downstream water, discharge data were based on summary data presented would be assumed to be of benefit to general aquatic by Tebandeke (2013) for the Nakivubo Channel (position ecosystem function downstream. For example, it undefined), and loading rates calculated for these would be likely to improve the availability of habitat of discharges were calculated for Total Nitrogen (TN), Total suitable quality for fish, thus potentially improving the Phosphorus (TP), BOD and TSS, using water quality data downstream Tilapia fishery. Such knock-on effects have provided in Table A5. 7 for the Railway Culvert. The not been quantified in this study. proportions of TP and TN comprising orthophosphate, and 6 ammonium and nitrate nitrogen, respectively, An unquantified reduction in bacterial contamination were also calculated and used to derive broad estimates of downstream water bodies would also be assumed (by difference) of the proportion of phosphorus and as a result of implementation of these measures. nitrogen nutrients that could be expected to be removed However, given the high rates of Salmonella and parasitic as particulate matter in sediments. 40% of both TP nematodes associated at present with even treated and TN was thus calculated to be the proportion likely sewage effluent, alternative approaches to the treatment to be available in particulate form, and these values of these pests would need to be found in addition to were included in calculations of removal rates if 80% those outlined above. TSS was removed. The data, though of low confidence, do not seem unreasonable - 45% removal rates of Implementation of this measure would require total phosphorus are the standard requirement for considerable capital expenditure and political will. Given management of stormwater runoff in the previously the significant impact it would have on instream and cited City of Cape Town stormwater management policy. downstream ecological function, as well as its implications for human health and dignity, and the opening up of 5 Tebaneke (2013) presented mean wet and dry season discharge data opportunities for safe recreation downstream, it should be of 1.97m3/s and 0.492 m3/s respectively regarded as an essential medium-term rather than long- 6 Note that nitrite nitrogen data were missing from datasets – this component of total nitrogen was assumed to be very low term objective in urban centres.                                                  Page 62 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Table A6.2 Estimate of annual litter loads generated in the Nakivubo wetland catchment, based on preliminary estimates of land cover and litter generation values given in Armitage (2004) Sub catchment Kayunga Kitante 5th Str Lower TOTAL Area (ha) 170 420 1370 830 Land-use - approx. % Low Density Res 40% 10% 20% Med Density Res 25% 5% 5% 5% High Density Res 40% 10% 10% Informal Res 40% 5% 10% 20% Industrial 15% 5% Retail 5% 15% 15% Offices 10% Halls, Stadiums and Entertainment 5% Facilities Taxi Ranks 5% 5% 5% Schools 5% 10% 10% 5% Hospitals 5% Golf course 20% Wetland/ fields 35% Litter Load (kg/year) 491 938 414 267 1 828 402 1 016 792 3 751 398 Vegetation load (kg/year) 4 760 10 290 27 058 20 543 62 650 Total load (kg/year) 496 698 424 557 1 855 460 1 037 334 3 814 048 Table A6.3 Preliminary estimates of changes in loading of key variables before and after treatment step 1, based on assumptions of 80% TSS removal. Total nitrogen Total Phosphorus Total suspended Biological Oxygen (kg / month) loading sediment (TSS) Demand (BOD) (kg / LOADING (kg / month) (kg / month) month) Dry Wet Dry Wet Dry Wet Dry Wet season season season season season season season season Pre implementation 11 482 45 973 2 159 8 646 33 617 134 606 57 020 228 313 After implementation 1 378 5 517 259 1 038 6 724 26 921 No No conversion conversion data data CONCENTRATION Total nitrogen Total Phosphorus Total suspended Biological Oxygen (mg/l) (mg/l) sediment (TSS) Demand (BOD) (mg/l) (mg/l) Pre implementation 8.88 1.67 26.0 44.1 After implementation 1.07 0.20 5.2 No conversion Wet and dry flow data derived from estimates in Tebandeke (2013) for the Nakivubo Channel, and reflected as loading per month, noting that wet and dry seasons span approximately six months of the year each. Note however that water quality data do not reflect seasonal variation, so wet and dry season data should be considered as ranges in loading. Water quality data used to calculate loading and to depict concentrations for the pre- and post-implementation scenarios based on Table A1 data for the Railway Culvert (after COWI 1998). Data for the Railway culvert used because this point includes most of the significant current sources of contamination into the Nakivubo channel that have been targeted by Steps 1 and 2. Concentrations calculated as a percentage of existing concentrations, assuming 80% TSS removal. 80% TSS reduction assumed, and 40% TP and TN reduction allowed for, on the 80% removed through TSS. These proportions calculated from proportion of TP and TN respectively comprising (NH4 +NO3)/TN and PO4-P/ TP – the limited data indicate 40% for both variables. Data shown in table reflect nutrients and sediments left in the system – hence 60% of 20% remaining TSS, if 80% TSS removed and 40% Nitrogen and phosphorus nutrients removed.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 63 Figure A6.1 Changes in concentrations of total nitrogen and total phosphorus compared to assumed current conditions if sediment removal rates of 80% are achieved. Coloured horizontal bars indicate trophic state at different nutrient concentrations, using data from Table A5.2 after DWAF (1996a and 2002). Step 2. Prevent residual pollution entering Design and costs Nakivubo wetland Sediment and litter traps Measures required This would require: While it might be reasonable to aim for waterborne sanitation coverage of 60%, we recognise that implementation of water-borne sewage is unlikely to i) Installation of litter and sediment traps on inflows be feasible for all households for a number of reasons. into the Nakivubo channel – this would apply to catch Therefore alternative measures will need to be put pits, drains and minor channels leading into the main in place to address the residual problems of polluted channel runoff, which include installation and servicing of litter and sediment traps, diversion of waste water flows to ii) Design and installation of an instream litter and the waste water treatment works, and filtration strips to sediment trap in the Nakivubo channel in the vicinity help deal with wet season overflows from these systems. of the 5th Street channel crossing – this would aim The latter measures will also provide some level of to remove sediment and solid waste in a reach pollution control during the interim period during which affected by high TSS loading, and would reduce the the sewage systems are expanded. This combination of requirement for channel dredging downstream; measures is considered the most effective, long-term approach to addressing the dire pollution levels passing iii) A second in-channel litter and sediment trap should into the Nakivubo channel. be installed in the channel immediately upstream of the railway culvert, and downstream of the informal The main objective of this set of measures is to minimise settlements abutting the channel in this area; the volume of polluted waste, including sediment, heavy metals, solid waste and (particularly) organic waste in iv) Allowance would need to be made for the ongoing the form of untreated sewage, domestic grey water, and maintenance (dredging) of the above facility, and for urban detritus, entering the Nakivubo wetland, and thus the disposal at an appropriate waste disposal site of becoming part of the catchment drainage system. The spoil thus generated; following measures must be included in this activity: To control and limit the amount of sediment flowing into a. Installation and management of litter and the Nakivubo wetland, the flow of any rivers and channels sediment traps should be controlled to prevent particle sizes larger than the selected size from flowing into the wetland (see Box b. Install diversion drains to convey waste water 1). Normal practice would be to prevent particles of say from informal settlements to WWTW 0.5mm or 1mm from being transported. Typically the flow speed should be limited. To control the gradient it is normal c. Create wetland filtration strips to filter drain practice to install hydraulic drops along the river or channel overflows before reaching main wetland To install an expanded channel litter trap and to control the flow of the Nakivubo channel from the Olde Timey Railway crossing to 5th Street to the Nakivubo channel, a distance of approximately 5km to the Nakivubo railway crossing, is estimated to cost between $1.5m to $2.0m. This includes 12 hydraulic drops constructed from gabions, a litter trap, and cleaning of the channel.                                                  Page 64 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Box 1. The sediment trapping process Deposition of transported sediments happens when a river loses energy, e.g. when a river enters a wetland or lake. To control and limit the amount of sediment flowing into the Nakivubo wetland, the flow of any rivers and channels should be controlled so that they are deposited before entering the wetland. Normal practice would be to prevent particles of say 0.5mm or 1mm from being transported. Using the Manning equation, Q = [A.R2/3.i1/2] / n, for channels and assuming a channel width of 4m, a Manning coefficient of 0.048, and a depth of between 25mm to 100mm, the flow velocities can be controlled as follows: For a gradient of 1:50 (assumed existing gradient) Water depth Flow velocity Flow Maximum Particle Size that [mm] [cm/s] [m3/hr] can be transported [mm] 8 11.9 14 2 16 19.0 45 6 25 24.7 87 8 33 29.8 140 11 For a gradient of 1:2000 (proposed new gradient) Water depth Flow velocity Flow Maximum Particle Size that [mm] [cm/s] [m3/hr] can be transported [mm] 25 4.0 14 0.6 50 6.2 45 1.0 75 8.1 87 1.1 100 9.7 140 1.1 By controlling the gradient of the Nakivubo channel, the particle sizes being transported would be greatly reduced, reducing the sediments and nutrients entering the Nakivubo wetland. To control the gradient it is normal practice to install hydraulic drops along the river or channel as shown below. The installation is considered relatively inexpensive and will achieve immediate results.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 65 Example of a hydraulic drop installed along a river to control the river’s gradient and the carrying velocity of particles. Even with source measures in place, there will be transfer of sediment and litter into the drainage system. In areas that have combined sewer systems (e.g. Europe, North America) removal can be achieved at the sewage works, as well as at overflows in very wet weather. Where stormwater drainage is separate from the sewage system, the sediments and litter must be trapped and removed along the watercourse. A range of designs exist that are applicable to different circumstances, from concrete structures within canalised reaches to in-channel excavated depressions. Sediment traps in particular will retard velocities to facilitate settling out of sediments. Litter traps work best when they incorporate some form of screen (Armitage et al. 1998). Of the different designs of litter traps, two patented designs were highlighted by Armitage et al. (1998) as being the most effective – the Stormwater Cleaning Systems (SCS) structure (similar to the Baramy Gross Pollutant Trap in Australia), and the Urban Water Environmental Management (UWEM) concept. The advantage of the latter is that it can trap silt and sewage as well as litter, and can be deigned to handle very large flows. While this was all technology that was developed around the 1990s there has not been much more done to advance this. The success of these measures depends on the ongoing maintenance of the traps. Below are examples of sediment and grit traps developed by Armitage et al. (1998). Enviroscreen installed at the Vygekraal Canal (Armitage et al. 1998)                                                  Page 66 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala The Enviroscreen model The improved Uyse model View of the expanded channel model Cross section through the improved Uys structure. Plan and long section through the expanded channel litter channel.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 67 Diversion drains Wetland filtration strips Installation of diversion drains would be required to Excess runoff (e.g. in wet conditions) from diversion intercept surface flows and shallow seepage from unlined drains should be diverted into sediment traps and areas of informal settlements, where there is a high through reed bed filtration systems, established likelihood that such water comprises untreated sewage between diversion drain outlets and the main Nakivubo and grey water runoff, with high bacterial, parasite, channel. Such reed beds, which should be designed as nutrient and general organic waste content. Diversion broad wetland filtration strips running parallel with the drains would need to be routed into the nearest existing channel, would be required to filter sediment and reduce sewer with capacity, and routed to the WWTW for the volume of organic waste in particular that passes into treatment. Alternatively, new sewers would need to be the channel. laid to cater for this waste stream. The diversion of such waste is usually effective only in dry flow scenarios, as A linear series of Cyperus papyrus wetlands should be it is likely to be practically limited to a low volume only. designed and installed along the wetland edge of the This could require further expansion of existing sewage prison, occupying the eastern edge of the lower wetland, works capacity to accommodate the additional sewage and along the wetland edge of the informal settlements volumes created. Assuming that the planned capacity of occupying the western edge of the lower wetland (Figure the sewage works will accommodate this, the costs of A6. 2). These would help to delineate the wetland the drains themselves are estimated to be in the region edge, preventing further encroachment, although it is of $1.04 million. likely that the wetlands would be crossed to access the agricultural areas within the wetland. These wetlands and/or traps should be maintained on a cyclical basis, with the following measures being required (exact frequency would need to be determined in detailed design and operational phases): • Cutting of plant material and disposal as a means of nutrient removal: at least annually to stimulate growth; and • Dredging of contaminated sediment to restore wetland treatment capacity – this activity would be required more frequently in upslope wetlands and those associated with the treatment of seepage from informal settlements and the Luzira Prison (see iii and iv above). Dredged sediment would need to be disposed of appropriately – disinfection and beneficial use in fertiliser or manure pellets could be considered. Based on the costs of developing and maintaining treatment wetlands (described under Step 4), it was estimated that the construction of the filter strips would cost in the order of $1.65 million, and maintenance costs thereafter would be roughly $300 000 per year. Figure  A6.2 Proposed location of linear series of Cyperus papyrus wetland strip filters                                                  Page 68 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Expected outcome Step 3. Improve WWTW effluent quality This step serves to complement the activities in Step 1, providing a second sweep to remove any remaining Existing technology and capacity pollutants as far as possible just before they enter the wetland. Nevertheless, thorough implementation of this The Bugolobi WWTW uses conventional sewage approach in all identified areas of concern is likely to treatment technology involving trickling filters (Box result in significant downstream improvement from both 2). Trickling filters are a well proven and documented a human health and ecosystem function perspective. method of treating sewage with the earliest trickling filters constructed in the early twentieth century. They While this step could be implemented independently are a low energy and effective means of treating sewage, of Step 1 as an alternative set of measures, it would require relatively little maintenance and are easy to not achieve the same degree of pollution abatement. maintain with parts being readily available. Trickling However, these measures can make a short-term filters are therefore ideally suited for Kampala. difference, since they are faster to implement than those in Step 1. Box 2. Current treatment of sewage through the Bugolobi WWTW (Source: NWSC) Preliminary Treatment: The inlet filter removes debris, large suspended solids and organic matter using metallic bar screens, sand traps and grit removal. The waste water then flows into grit chambers where heavy solids sink to the bottom and are removed. These chambers remove solids >0.3mm in diameter. Primary Treatment: Organic solids are separated from the liquid in the first sludge sedimentation tank (primary clarifier). Offensive solids removed through settling and floating materials (scum, oils, grease) are removed through skimming in circular sedimentation basins. Removes on average 45-50% of suspended solids and 25-30% of the BOD in the incoming waste water. Settled sludge is collected and transported to a sludge thickener. Secondary Treatment: This purification stage involves treatment by means of biological trickling filter where waste water flowing over the surface of a well aerated stone bed where bacteria remove any organic matter, commonly known as bio-filters or aeration tanks. After this any excess sediments, suspended matter and break-off bacterial film is removed as sludge through a secondary clarifier. The final treated sewage effluent is tested and must comply with the national environmental standards before being discharged into the receiving environment (Nakivubo) through an artificial wetland. Sludge is moved to anaerobic sludge digester beds and then onto sand drying beds. Dried sludge is sold to farmers to use as fertiliser. Additional Treatment: Vegetation in the Nakivubo wetland provides final “tertiary treatment”, i.e. the removal of excess inorganic nutrients. Illustration of the conventional sewage treatment process                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 69 Figure A6.3 Bugolobi WWTW from Google Earth on 18 July 2015. The various unit processes indicated are as follows: (1) Primary Settling Tanks, (2) Digesters, (3) Large Trickling Filters, (4) Small Trickling Filters, (5) Secondary Settling Tanks, (6) Sludge Drying Beds, (7) Demolished struc- tures, most probably old trickling filters. The capacity of the BWWTW was estimated based on the Indeed, the Bugolobi WWTW is not currently used to size of the trickling filters in operation. From Figure A6. 3 its full capacity and does not operate efficiently. It relies taken in 2015, it appears that about half of the trickling on a system of siphons and pumping stations to deliver filters have been demolished. Using the measurement more than half of its sewage. Frequent operational tool in Google Earth, the diameter of each of the existing problems results in untreated sewage being discharged large trickling filters are 32m, and each of the small into the environment. When treatment occurs, the trickling filters are 12m. facility does not comply with nutrient and coliform removal standards (African Development Fund 2008). According to Metcalfe & Eddie (2004), stone media will have a surface area of about 70 m2 per cubic metre of For trickling filters to work effectively, they require filter media. Using a total surface area for all trickling continuous wetting of the substrate media otherwise filters of 7100 m2 and a filter media depth of 4.0 m deep, the organic matter attached to the media will die. Re- the flow capacity of the trickling filters would be 12 463 commissioning a trickling filter to its design capacity m3/day. With a shallower filter media depth of 3.5m generally takes two to three months to develop effective the flow capacity of the trickling filters would be 10 905 organic matter on the substrate media. It is therefore m3/day. These estimates are more or less in line with important to ensure that there is always electrical power the influent flow as stated by Nansubuga et al. (2013) to supply pumping equipment and also that there is that the WWTW is receiving an average inflow of 12 standby equipment should a pump fail or be in the 000 m3 per day. However, the inflows are likely to have process of being serviced. increased, which means that the capacity of the plant can be considered to be exceeded at this stage, even if all the trickling filters were working and if the WWTW was running at full capacity.                                                  Page 70 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala The most recent Google Earth image (July 2015) shows Pathogens can be reduced using chemical treatment that only four of the eight large trickling filters are or exposure to sunlight in maturation ponds. The working. This can be seen from their darker colour, latter is recommended here. Assuming that there indicating that the filters are wet. Thus the Bugolobi is no disinfection happening at the WWTW such as WWTW is running at about 50% of its capacity which is chlorination of the final effluent, the ponds should be the same as having at least 50% of the raw sewage enter designed with a surface area large enough to reduce the Nakivubo wetland untreated (or more than 50% if the faecal coliform loading. The maturation ponds act input has increased). as solar ponds where the sun’s ultra violet rays destroy pathogens. Maturation ponds also provide a buffer The new Nakivubo WWTW will replace the Bugolobi between the WWTW and the environment should any WWTW and is planned to increase overall capacity major spills at the plant occur. Maturation ponds are to 45 ML per day. The treatment process will include generally designed as flat shallow ponds with a depth mechanical, biological and chemical treatment of of one metre (to prevent reeds and other aquatic plants municipal sewage and industrial effluent, sludge taking root. There need to be a minimum of three ponds digestion and cogeneration using biogas (NWSC). The in series. For a flow of 12 ML per day, and allowing for a biogas will be captured during the anaerobic processes 99.8% reduction in faecal coliforms, three maturations and will be converted into electricity using combined ponds, the surface area required for each pond needs biogas heat and power systems. to be 4800m2 (about 0.5 ha) To cater for a flow of 45 ML/day, 13 ponds would be required (7.5 ha). Note Further measures required that these ponds are not the same as the waste water The following measures are included in this step: treatment wetlands described further below. To construct maturation ponds below the existing a. Improve aeration of treated effluent from the BWWTW, a cost of $420 000 per pond is estimated. Thus existing and planned WWTW, by the installation three ponds is estimated at $1.26 million and a further of aeration sprays within final effluent maturation seven ponds are estimated at $2.94 million. Maintenance ponds – this would have the aim of increasing DO is estimated at 2% per annum of the construction costs concentrations to at least 4mg/l; (Table A6. 4). b. Improve effluent standards with regard to phosphorus, by upgrading technology at Expected outcome the WWTW to allow for final effluent with Consideration of treated effluent volumes and quality concentrations of 1.5 mg orthophosphate/litre provides an interesting picture in relation to proposed instead of the current standard of 10mg/l; and efforts to address water quality issues in the Nakivubo wetlands and associated Murchison Bay areas. The c. Include measures to reduce parasitic nematode limited 2015 dataset for the Bugolobi WWTW suggest infection of final effluent. that current effluent treatment does not meet legal effluent limits outlined in Table A5. 6. Figure A6. 4 Design and costs illustrates effluent loading on receiving water bodies for the 2015 dataset for Total phosphorus and TSS data only, Given that work has already commenced on the new assuming actual water quality at the time, but projecting 45 ML plant, the only additional interventions required the planned future design effluent volume onto these are the tertiary treatment facilities (sand filters and data. The calculated total phosphorus and TSS loads maturation ponds). are compared in the figure against loads that would be The tertiary treatment for phosphate removal is achieved achieved if legal effluent limits were met, and the figures using sand filters. The cost of this is estimated at about show that substantially reduced loads would occur, if a third of the capital cost of a plant up to the secondary effluent limits were met. settling tanks. Therefore the construction of sand filters was estimated to be $16.5 million (Table A6. 4). Table A6.4 Estimated capital and operating costs of upgrading the WWTW to improve water quality standards in Nakivubo wetland Initial cost Operating and maintenance Intervention $ million costs $ million per annum Further improvements to decrease nutrient concentrations (sand filters) 16.5 0.33 Further improvements to reduce pathogens (maturation ponds for 45 ML output) 5.46 0.11                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 71 Figure A6.4 Total phosphorus and TSS loading estimates in BWWTW final effluent assuming treatment volumes of 33 000m3/d. Loading based on 2015 data, with red line showing loading if legal effluent limits shown in Table A5. 6 are achieved.                                                  Page 72 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala In fact, an upgrading of the BWWTW is already planned. An important assumption of this study is that the new plant will be required to be designed so as to meet at least the legal limits for effluent discharge, as required by Uganda’s own legislation. Aeration of final effluent is recommended, with a view to increasing Dissolved Oxygen Concentrations in final effluent to > 4mg/l. This is because 4mg/l is believed to be a threshold below which non air breathing aquatic fauna may be critically affected (DWAF 2008). The reason for the low current dissolved oxygen concentrations in water in the lower wetland and zone of discharge into Murchison Bay is primarily due to the high biological and chemical oxygen demands (BOD and COD) of the combined effluent and stormwater. Oxygen is required for process such nitrification (that is, the oxidation Figure A6.5 Bioballs are light and easy to handle. of ammonia or ammonium to nitrite, followed by the Source Broadreach (Pty) Ltd oxidation of nitrite to nitrate) and is generally required in the decomposition of organic material. Aeration of effluent water is one manner in which additional dissolved oxygen can be made available for such processes, with the intended outcome that more of the BOD and COD can thus be met, resulting in higher concentrations in dissolved oxygen passing into the lower wetland and Murchison Bay, thus improving conditions for aquatic ecosystems including the Murchison Bay fishery. Potential rapid intervention to improve capacity of Bugolobi WWTW With the BWWTW running at 50% capacity and being overloaded and with the time for the new proposed works coming on line, a contingency plan is needed urgently to alleviate the pollution into the Nakivubo wetlands and into Lake Victoria. The capacity of the existing trickling filters can be almost Bioballs are easily transported, unloaded and packed due to their instantly increased replacing the stone media with high lightness and robustness. surface area packing or bioballs (Figure A6. 5). Whereas the existing stone media has a surface area of between 65 to 75 m2 per cubic metre of stone, proprietary packing and bioballs typically have a surface area ranging from 175 m2 to 300 m2 per cubic metre of media. This equates to an increase in area of up to 3.5 to 4 times. This alone would increase the capacity of the treatment works considerably. It is recommended that this is done in stages over the period of a year, to accommodate the time for the bioballs to become colonised and fully functional. It is assumed that the already-planned expansion of the WWTW is designed to meet legal effluent standards. Therefore, additional measures would have to be installed in order to reduce the nutrient loads and pathogens beyond that required by national policy, in Organically loaded (dirty) water being sprinkled over the bioballs in a order to achieve the goals of wetland rehabilitation and trickling filter. downstream water quality fit for recreational use.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 73 Tertiary filtration aided by chemical addition can reduce Step 4. Rehabilitation of upper wetland for total phosphorus concentrations in the final effluent waste water treatment to very low levels. To achieve very low phosphorus Note that unless the measures outlined in Steps concentrations, chemicals must be added to waste water 1 - 3 above can be achieved, it is unlikely that the to associate phosphorus with solids that can then be measures outlined in the following section will have any successfully removed through filtration. Aluminum- or measurable effect on wetland function or fitness of water iron-based coagulants and polymer are the chemicals quality from an aesthetic or human heath perspective. most commonly used for this purpose. Measures required Traveling sand bed filters, mixed- media gravity filters, The following measure is required to improve the and variations of these filtration technologies as well as supplementary treatment for final treated effluent. The membrane filters are commonly used. Filtration has been extent depends on the extent to which final effluent is employed for many years to treat drinking water and treated (Step 3). more recently applied to treat waste water. Selection of a filtration technology includes the usual considerations Design and costs such as: desired effluent quality; reliability of treatment This step involves the creation of multiple shallow equipment; capital, operating and maintenance costs; wetland cells (standing water depth 300-500mm or less), equipment footprint, and future expandability. separated by berms and with controlled pipe outlets or multiple overflows linking each cell to downstream cells A two-stage filtration process generally produces the (Figure A6. 6). The excavated spoil could be used in part lowest phosphorus levels. Two-stage treatment may be for the creation of berms, and should ideally be treated achieved through use of a first and second stage filter initially to remove parasite loads. Excess fill would need or by providing tertiary clarification prior to filtration. to be disposed of outside of the wetland areas. Excellent treatment results have been obtained by using a two-stage treatment process consisting of chemical Allowance needs to be made for maintenance of the addition with tertiary settling in advance of their sand treatment wetlands on a cyclical basis, with the following bed filters. measures being required (exact frequency would need to be determined in detailed design and operational The treatment provided also removes other pollutants phases): which commonly affect water quality to very low levels. COD and TSS are routinely less than 2 mg/l and fecal coliform bacteria less than 10fcu/100 ml. Turbidity of 1. Cutting of plant material and disposal as a means the final effluent is very low which allows for effective of nutrient removal: at least annually to stimulate disinfection using ultraviolet light (by means of growth; maturation ponds), rather than chlorination. 2. Dredging of contaminated sediment to restore The amount of bioballs or packing media required to wetland treatment capacity; and replace the stone media in, say, four trickling filters would be 12 900 m3. Assuming a material cost of $270 3. Disposing of dredged sediment appropriately – per cubic metre and allowing for transportation, removal disinfection and beneficial use in fertiliser or manure of the existing media, and placement costs is estimated pellets could be considered. at $7m.                                                  Page 74 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Figure A6.6 Indicative areas for rehabilitation as treatment wetlands in the Nakivubo wetlands. The proposed location of filters are also shown (darker green) for reference. Constructed wetlands require extensive landscaping and Maintenance of the constructed wetland systems excavation work and should be designed with careful include inspections and monitoring of mosquitos, consideration for factors such as access, prevention of algae and sediment build up (Armitage et al. 2013) and litter and debris entering the inlet zone, and regulation of the frequency of such is dependent on the individual the water level within the constructed wetlands (Armitage system. In the UK, low frequency monitoring is every 12 et al. 2013). The following cost estimates are based on months and high frequency is every month (Armitage information collated from the literature (Table A6. 5. et al. 2013). The routine maintenance of constructed wetlands include the removal of litter, the cutting of grass 4. (Table A6. 6) that focused on best management banks and general management of the vegetation and practices for sustainable drainage systems. Costs should include the inspection and cleaning of inlet and vary significantly from country to country. The main outlet pipes (Armitage et al. 2013). The following table expenses involved in construction wetlands include represents typical maintenance rates (Table A6. 8, based acquiring land, excavation, planting soil, pipelines, on 2010 Rands, Armitage et al. 2013). vegetation and on-site work (Kadlec & Wallace 2009, Gunes et al. 2011). Table A6. 7 includes an estimate for a hypothetical 1 ha constructed wetland system as described by Kadlec & Wallace (2009).                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 75 Taylor’s (2005) review of Australian best management Given that the study area is likely to require maximal practices in this field reported that annual maintenance rehabilitation of the upper wetland, of about 75 ha, there costs for constructed wetlands are 2% of the construction are likely to be economies of scale when it comes to costs (cited CWP 1998, Weber 2001, U.S. EPA 2001). labour and construction of the treatment wetlands. Thus In the same review it is reported that the macrophyte we applied the equations of Lloyd et al. (2002), which zone should be replaced every 20-50 years at a cost of takes into consideration economies of scale, to estimate 50% of the initial construction cost (cited Fletcher et al. the construction costs. These were adjusted to 2015 US 2003). In Taylor (2005), it was reported that Lloyd et al. $ values. Based on this the design and construction costs (2002) estimated the following equation that could be could be expected to be in the order of $2.287 million, used for estimating the landscaping maintenance costs: with an annual maintenance cost (estimated to be 2% of Maintenance Costs (Aus $) = 9842.20*(surface treatment this), of about $45 700 per year. area in ha)^0.4303. Table A6.5 Estimated construction and design costs for constructed treatment wetlands typically treating municipal waste water (extracted from reviews by Taylor 2005, Silva & Bragga 2006, Gunes et al. 2011, La Notte et al. 2012). Reference Wetland type/size Cost per unit Currency Leinster 2004 Small-scale wetland $90 - $100 per m2 Australian $ Leinster 2004 Large-scale wetland (reticulated $65 per m 2 Australian $ lake) Hunter 2003 Large wetland $500,000 per ha Australian $ Weber 2002 Standard constructed wetland $3400 – $17,900 per ha of area treated or Australian $ $730,000 per ha of total wetland area Walsh 2001 Greenfields wetland $120,000 per ha of area treated Australian $ Lloyd et al. 2002 (equation) Greenfields wetland Construction cost ($) = 343,913 x Ln * (surface Australian $ treatment area in ha) + 738,607 Lane 2004 Standard constructed wetland $16 000 for outlet and CDS unit plus the costs of Australian $ $75 per m2 Stewart 2005 Horizontal Flow (HF) wetland $86 per m2 ($74 -97) US $ Dzikiewicz 1996 Horizontal Flow (HF) wetland €31 per m (€10 – 83) 2 Euro Rousseau et al. Horizontal Flow (HF) wetland €257 per m (€237 – 277) 2 Euro 2004 IRIDRA 2002 Horizontal Flow (HF) wetland €125 per m2 (€38 – 247) Euro Masi et al. 2006 Horizontal Flow (HF) wetland €115 per m2 (€101 – 129) Euro Steiner & Combs Horizontal Flow (HF) wetland €74 per m (€27 - 144) 2 Euro 1993 Billore et al. 1999 Horizontal Flow (HF) wetland €29 per m2 Euro Platzer et al. 2002 Horizontal Flow (HF) wetland $61 per m ($22 - 229) 2 US $ U.S. EPA 2000 Horizontal Flow (HF) wetland $67 per m ($32 - 125) 2 US $ Dallas et al. 2004 Horizontal Flow (HF) wetland $33 per m2 US $ De Morais et al. Horizontal Flow (HF) wetland €96 per m 2 Euro 2003 Shrestha et al. Horizontal Flow (HF) wetland $31 – 72 per m2 US $ 2001                                                  Page 76 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Table A6.6 Percentage breakdown of capital costs for the construction of the treatment wetland beds Plumbing Control Country example Excavation Gravel Liner Vegetation Other (Pipes) Structures Spain 15 27 33 2 6 5 12 Czech Republic 7 53 13 7 12 - 8 Portugal 12.5 37.5 25 5 - 11 9 Source: La Notte et al. 2012 Table A6.7 Estimated capital costs for a hypothetical 1 ha wetland system, in 2009 US $ Component Unit Quantity Unit Cost ($) Total Cost ($) Land acquisition ha 1 10,000 10 000 Site evaluation Lump sum 1 2,000 2,000 Clear & grub ha 1 8,000 8,000 Earthworks m 3 10,000 7 70,000 Liner m 2 12,000 8 96,000 Planting soil m2 3,000 10 30,000 Plants & planting plant 20,000 3 60,000 Structures Lump sum 5 2,000 10,000 Conveyance m 400 35 14,000 Site Work Lump sum 1 20,000 20,000 Total direct cost 320,000 Engineering 15% 48,000 Construction & observation 5% 16,000 Start-up service 5% 16,000 Non-construction costs 5% 16,000 Contingency 20% 64,000 Total indirect cost 160,000 Total Cost 480,000 Source: Kadlec & Wallace 2009 Table A6.8 Typical routine inspection and monitoring rates for constructed wetlands based on data provided in Armitage et al. (2013) Description Unit Rate (Rands) Inspections Visit 210 Vegetation management (large) Visit.m 2 0.60 Vegetation management (pocket wetlands) Visit.m2 2.00-2.40 Sediment removal (standard wetland) m3 Site dependent, > 160 Source: 2010 Rands                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 77 Expected outcome -- Miscanthidium violaceum mats provided The following factors and assumptions have a bearing on a barrier between underlying swamp the calculations and approaches informing this section: waters, which were poorly treated as a result of passage through the • Groundwater influence – this is assumed to be mats; however, the efficacy of use of negligible, with much of the swamp underlain Miscanthidium as a treatment for irrigated by clays (Kansiime & Nalubega 1999); waste water was not investigated in • Channel length and inflows: any of the above studies, and may have yielded different results to testing of its -- Kansiime & Nalubega (1999) estimate the total role in an in situ flow-through system; length of the Nakivubo channel as 12.3km, with the upper swamp zone (upstream of • Hydraulic retention time: 5 days has been the railway line) comprising 1.1km and the assumed (based on Kyambadde et al. 2004), lower swamp zone being a further 1.2km noting again that no allowance for seiche effects has been made. This time is however likely to • Influence of seiches: these affect in-swamp have been reduced, as a result of more recent retention time, and flush sediments into Lake detachment of large portions of the floating Victoria (Kansiime & Nalubega 1999) – however, island of the wetland (see Appendix 3); they affect the middle portions of the swamp (along the main flow path) and the lakeward • Seasonality (after Stalder 2014): side, and do not result in flushing of the whole -- Wet seasons: March to May and October swamp, with thick peat and areas with high to November (152 mm maximum) levels of suspended solids having a “dampening” -- Dry Seasons: July (driest month), effect on flushing (Kansiime & Nalubega 1999); and December to February • Swamp-lake exchange: this increases Even if the water treatment regulations are met for all longitudinally with distance from the railway the WWTW discharging into the Nakivubo wetland, these towards the lake shore (also see above regarding standards will not be enough (in conjunction with Steps seiche effects) – swamp lake exchange has 1 – 2) to bring about the targeted trophic condition of been discounted in this assessment; the lower wetland and Inner Murchison Bay, or fulfil the • Waste water influences: objective of improving wetland quality and recreational -- Waste water flow is not well distributed and opportunities in the lower wetland. Objectives for throughout the swamp – Kansiime & Nalubega achieving such measures should be based on the passage (1999) show that the area of the swamp of water that is not enriched beyond eutrophic levels, exposed at the time of their (1999) study to as outlined in Table A5.2. In terms of the Condition waste water effluent was some 785 000 m2 out categories outlined in Table A5.3, this would equate to of a total area of 1 150 000 m2 – by the time of conditions that were not worse than Category D. the present project, this area had decreased considerably (see Chapter 4 and Appendix 3); Given that reduction in phosphorus concentrations is considered the most challenging of the variables -- Water depth, vegetation type and density to address, and that achieving these requirements and flow depth are the main reasons for phosphorus means that there is a high likelihood affecting preferential flow in the wetland that these would also be achieved for other variables, (Kansiime & Nalubega 1999), with calculations around the effect of wetlands and other data from the above authors as well as measures on WWTW effluent quality have focused Kyambadde et al. (2004) suggesting: on total phosphorus. Achieving a PES Category D for -- Cyperus papyrus wetland allows for higher phosphorus means that a target concentration of no rates of both nitrogen and phosphorus higher than 0.15 mg/L total phosphorus would be required removal than Miscanthidium violaceum, at the railway culvert, noting that even this value lies just with plant uptake and storage being the above the eutrophic /hypertrophic threshold. main factors accounting for nitrogen and phosphorus removal in the former,                                                  Page 78 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala With treatment wetlands being suggested as a • If effluent is produced to the legal limits for total potentially useful tertiary polishing mechanism to phosphorus of 10mg/l, there is not sufficient improve effluent quality, data from Kyambadde et al. area in the wetland to be able to further treat (2005) for total phosphorus removal rates were used it to its required standard of 0.15 mg P/l. In to calculate approximate areas required to achieve the event that effluent remains in the high the target water quality concentrations with regard to hypertrophic zone, even if its concentrations have total phosphorus. These authors assessed performance reduced considerably, it is unlikely that there efficiencies in constructed Cyperus papyrus wetlands will be a measurable reduction in symptoms of in Uganda, and their data are therefore considered hypertrophic conditions, such as phytoplankton appropriate to the current circumstances. They estimated blooms, low oxygenation of bottom waters, total phosphorus removal rates by treatment wetlands of affecting aquatic habitat quality (e.g. for fish); 191 ± 9 kg/ha/year. These data were used to inform the • In order to achieve the required levels of tertiary graphs shown in Figure A6. 4. polishing of effluent, the final effluent itself needs to be produced with substantially lower Figure A6. 7 shows the areas of treatment wetland that concentrations of total phosphorus than those would be required to treat effluent to the target value that are legally permissible in Uganda. In fact, a of 0.15 mg/L total phosphorus, assuming different final effluent of 1.5 mg/L would be required if a concentrations of total phosphorus in final effluent, treatment wetland of 85 ha was utilised, while starting from the legal limit of 10 mg/l, and reducing an area of 117 ha would be required for final to much lower concentrations. Actual areas of nominal effluent quality of 2 mg/l for total phosphorus; parts of the Nakivubo wetland are shown in Figure A6. 6 to illustrate the implications of the areas shown in Figure • If lower volumes of effluent were produced (e.g. A6. 7. A total of 134 ha is illustrated in the figure as the if the improved quality of effluent generated in maximum area for rehabilitation as treatment wetlands. the new works allowed for its beneficial reuse in industry or agriculture elsewhere), then smaller The following conclusions can be drawn regarding the wetland areas would be required, as also indicated use of treatment wetlands in achieving tertiary polishing in Figure A6. 7 (e.g. a reduction in wetland area of effluent: from 85ha to 65 ha to achieve the same final water quality, if the volume was reduced to 25 000 m3/d); • Such low concentrations are achievable in WWTWs – but would need to be included in the design of the planned new facility. Figure A6.7 Estimated wetland area required to treat effluent to a fixed target of 0.15 mg/L, for two different final effluent volumes, when different final effluent concentrations are achieved.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 79 In addition to achieving downstream benefits in terms Step 5. Improve the functioning of the lower of resource quality, opening the lower wetland for wetland safe recreational purposes, likely improvements in Measures required fisheries and other benefits discussed in terms of Step 4, The following measures are included in this step: rehabilitated wetlands would also be likely to control the expansion of informal settlement into the wetlands. a. Improve the spread of flow into the lower wetlands; There are caveats to the above discussion. Hydrological data for the Nakivubo channel are very limited, and b. Prevent further damage from agriculture; calculations of loading in the channel (presented for Step 1) use single values for assumed wet and dry c. Encourage regular harvesting of Papyrus plant channel flows. The calculation of required wetland area material to stimulate growth; and to address effluent of a known quantity and assumed quality ignores the fact that during high flows, if Step d. In the event that recreational use of the island 1 only is pursued, then there would be a substantial occurs, control the effects of increased human improvement in the quality of outflows from the channel, passage by: which in wet conditions might exceed treated effluent volumes by more than fourfold (data from Tebandeke i) Creating boardwalks to avoid where possible (2013) suggest that wet season flows are in the order of sensitive floating wetland areas; 170 000 m3/d). If these flows took place in the context of full implementation of Step 1, then considerable ii) Ensure that adequate provision is made for wet season dilution of effluent flows would occur. In litter collection; and the event that Step 1 was not implemented and only Step 2 was followed, this argument would not apply, iii) Ensure that sewage and grey waste water as the latter is based primarily on dry season diversion are disposed of into the sewerage and do of contaminated flows, and these would be largely not pass into the water body. ineffectual in the wet season. The last measures are necessary for both the protection of the wetland and as facilities for visitors, and are thus incorporated into Step 6. Figure A6.8 Relationship between surface phosphorus and surface chlorophyll in lakes. Data sourced from Cooke et al. (2005) after Carlson (1977). Equation for relationship (3rd order polynomial) calculated in present study.                                                  Page 80 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Design and costs Improve the spread of flow Harvest wetland vegetation The treatment wetlands and filtration strip wetlands Papyrus needs to be harvested at certain times of the (Step 3) should be connected to the lower wetland via a year and at different growth stages to stimulate growth series of pipe or other culverts created at intervals along and promote nutrient storage within the wetland. A the railway line. At least three new drainage lines are specific harvesting plan for papyrus plant material in the envisaged. It was assumed that three new drainage lines lower wetland should be initiated and this should focus would be needed to spread the flow effectively from on educating local people about how best to harvest the treatment wetlands through to the lower wetland the plant material. The ongoing costs would therefore via a series of pipes and/or culverts. Typical capital and include the managing and monitoring of the harvesting maintenance costs for conventional drainage design were programme as part of the larger conservation effort in taken from Armitage et al. (2013) and inflated to 2015 US the lower wetland. This was based on estimates taken $. These costs were based on estimates for constructing from Kakuru et al. (2013) for wetland management and unlined channels, using pipe culverts 600mm in diameter conservation. and dewatering subsoil to fit the drainage lines. The estimated capital and maintenance costs involved Flows from the Nakivubo channel should pass into the in protecting the lower section of the wetland are lower wetland as at present, with possible allowance summarised in Table A6. 9. for further spreading of flows into degraded Papyrus wetland, if there is sufficient water flow. Expected outcome Achievement of this objective is contingent on Prevent further damage from agriculture implementation of Steps 1 to 3. The resultant improved Agricultural activities need to be managed to prevent water quality flowing into the lower wetland and more cultivation and livestock grazing in the floating Cyperus particularly into the Murchison Bay area would be papyrus or M. violaceum islands, to prevent further associated with the following impacts: breaking up and loss of these wetlands. • Improved oxygenation of wetland waters, improving Measures to control agricultural activities include fencing wetland / lake habitat for aquatic organisms where necessary around the lower wetland to prevent including fish, reported in some studies (e.g. Stalder agriculture within the designated wetland area, and 2014) to be limited by the build-up of anoxic sludge signage and patrols to enforce these regulations. It was on the lake bottoms and poor water quality; assumed that approximately 2 km of fencing would be • Improved growth of Cyperus papyrus wetland needed and that patrolling the wetland would require vegetation throughout the Nakivubo wetlands full time staff at a cost of $6000 per year (updated from (as a result of establishment of extensive Kakuru et al. 2013 who estimated wetland management treatment wetlands (Step 4)) but particularly costs including salaries of staff). in the lower wetlands, where reduced sediment would allow improved uptake of orthophosphate by this species in water channelled into degraded areas from upstream; Table A6.9 Estimated capital and ongoing maintenance costs associated with interventions to conserve the lower wetlands (2015 US $ millions) Initial cost Operating and maintenance costs Intervention $ million $ million per annum Pipes/culverts to improve the spread of flow into the lower wetlands. 1.38 0.04 Measures to control agricultural activities 0.03 0.007 Encourage regular harvesting of Papyrus plant material to stimulate 0.00 0.02 growth Total costs: conservation of lower wetland 1.41 0.07                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 81 • Improved aesthetic conditions, as a result Step 6. Establish a recreational space with of reduced litter and sedimentation (Steps facilities 1 and 2), as well as substantially reduced Vision for the Nakivubo Wetland Park human health risks for intermediate contact The KCCA envisages a future development path in which recreation, allowing opportunities for the use wetlands areas are enhanced as recreational green open of the lower wetland for safe recreational / space areas to be enjoyed by the city’s inhabitants. This tourism opportunities to be explored; and has been recently reinforced by a directive from the • Decreased plant production in Ministry of Environment to restore the city’s wetlands. Murchison Bay, potentially: In addition, the city envisages a lakefront development -- improving its aesthetic qualities if decreased in the vicinity of Port Bell in which there will be various production rates result in measurable attractions such as waterside restaurants (Figure A6. improvement in water clarity and 9). Such waterside developments within or alongside decreased phytoplankton blooms; and working harbours have been very successful in other cities, such as Cape Town. -- potentially reducing purification costs. It should, however, be noted that these impacts are only The extent and details of the envisaged Nakivubo likely to be measurable / visually apparent if Steps 1 to Wetland Park have not been developed. For this study 3 have been rigorously implemented. Figure A6. 8 plots it was necessary to articulate what this might mean. We values from Cooke et al. (2005) that are applicable to all envisage that this will be integrally and physically linked lake conditions, and were derived from Carlson (1977). to the planned waterfront area at Port Bell. The graph shown in Figure A6. 8 shows that, once in-lake What we have envisaged for the Nakivubo-Wetland water is in a hypertrophic state, substantial reduction in recreational area is the provision of a reasonably large total phosphorus concentration needs to be achieved to parkscape area adjacent to the wetland, with access to show a reduction in surface chlorophyll concentrations. both lakefront and the vegetated wetlands, with picnic Other data from the same sources (not shown in this facilities and walking paths in and around the landscaped figure) highlight the related fact that improved water area and wetland (Figure A6. 10). The main landscaped clarity as a result of nutrient reduction also requires recreational area would thus be close to Port Bell, but the water body to be shifted towards mesotrophic smaller sites could also be set up at other locations conditions, and achieving major reductions in surface around the wetland. phosphorus concentrations may have little visible effect unless actual trophic state thresholds can be crossed. A core part of the wetland could be set up as a bird sanctuary which retains its natural characteristics as No quantification of the effects of reduced phosphorus far as possible and into which access would require a concentrations on in-lake chlorophyll concentrations, bit more effort in order to keep human disturbance particularly in the vicinity of the Ggaba water inlet have to a low level. The bird sanctuary would contain rustic been made in this study. Water quality data for the boardwalks and hides (of a design that would not latter include TSS concentrations, but no nutrient data. encourage unintended uses. This in itself could provide Although it is assumed that the measured TSS comprised tourism opportunities if well managed. The park would wholly of phytoplankton, the data themselves are not also contain a wetlands information centre located in transposable to any of the variables included in the between the landscaped areas and bird sanctuary to Carlson (1977) relationships. enhance the visitor experience.                                                  Page 82 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Figure A6.9 Concept plan for the study area taken from the Kampala Physical Development Plan Various other facilities (e.g. sporting facilities) could In addition to a core recreational area in the vicinity be included depending on the space available. It is of the Prison site to Port Bell area we also envisage envisaged that the immediate surrounding area in the landscaping of smaller areas on the opposite side of vicinity of the current prison site and Port Bell would be the lower wetland, and the possible creation of a wide developed for retail, restaurants or kiosks with some of paved walking/cycle way along one or both banks of this being on the waterfront. We do not envisage any the wetland extending over parts or all of the way from further transport routes across the wetland however, as the lake to 5th bridge. Such a pathway would be on this would likely be counter to the objectives of wetland the inside of the wetland buffer strips described in the restoration, conservation and aesthetic improvement of previous chapter and could potentially create a safe the area. and pleasant passage into town which encourages non- motorised commuting and the use of cycle taxis rather than the highly polluting boda-boda’s. Extending this idea to the whole length of the wetland would, however, require the relocation of houses from the informal slum areas. Should this be feasible, its proper construction would discourage the return of these informal dwellings. In this study, we have not considered the addition of this feature due to the difficulty of estimating the relocation costs. However, it is recommended that this be considered in the future.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 83 Table A6.10 Estimated capital and maintenance costs for the proposed Nakivubo Wetland Park based on the concept vision used for this study Initial cost Operating and maintenance costs Work/Amenity $ million $ million per annum Vegetation clearing 0.12 0.002 Boardwalk & foot bridges 0.65 0.013 Fencing 0.08 0.002 Pathways (walkways, cycle paths etc.) including amenities such as 1.54 0.031 benches, litter bins and lighting Landscaping of park area for picnic and leisure activities 2.59 0.052 Public toilets 0.43 0.009 Parking areas 1.62 0.032 Visitor centre 4.32 0.086 Hides/resting pavilions 0.12 0.002 TOTAL 11.48 0.230 Estimated costs of recreational facilities A high level estimate was made of the costs of Costs associated with constructing the wetland park developing the recreational park. This does not include and associated amenities, and measures to control the the private development costs that would be anticipated, effects of increased human recreational use within the such as the development of restaurant, entertainment wetland park were estimated using costs available in and retail outlets, but only the cost of providing the literature on similar projects based elsewhere in the public amenities. These were estimated to include the world combined with costings provided by an industry following: professional in South Africa. All values were converted to 2015 US $. South African cost estimates were multiplied • 13 km main paths around the wetland area of at by a factor of 1.5 as it was assumed that the associated least 2 m width, designed for walking and cycling, costs in Kampala would be higher due to influences such with amenities such as benches, bins and lighting; as fewer suppliers and imported materials. Maintaining • 5 km of board walks within the and servicing these amenities was assumed to be 2% wetland, and 5 foot bridges; of capital costs. Some of the ongoing servicing and maintenance costs for the public restrooms could • 15 ha of landscaped park space for possibly be covered by a pay-per-use initiative. picnic and leisure activities; • 8 resting pavilions/bird hides; It is estimated that the total capital costs involved in constructing the wetland park will be approximately • 10 public toilet blocks; $11.5 million, with annual maintenance and operating • 5 parking/access areas of 0.5 ha; and costs of $0.23 million per annum (Table A6. 10). The • 1 single-storey visitor centre of about 0.5 ha in size. largest cost being the construction of parking areas, a visitor centre and the landscaping of 15ha of park area.                                                  Page 84 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Figure A6.10 Concept vision for the Nakivubo Wetland Park used in this study                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 85 Appendix 7: Impacts of water quality changes on water treatment costs A7.1 Introduction Ggaba Water Treatment Works is made up of three In the context of treating water the biggest effects of treatment plants. Ggaba I has been operational since increased sediment and nutrient loads on the cost of 1930, Ggaba II since 1992 and Ggaba III since 2007, and treating water are as follows: (Source: Graham 2004, they have a total water supply capacity of approximately McDonald & Shemie 2014, Rangeti 2014): 170,000 m3 per day (NWSC pers. comm. 2015). In 1992 the plant was upgraded to include clarification stages • Increased usage of coagulants; of the treatment process and in 2007 the system was upgraded again to become fully automated (Ooyo • Increased amount of time water 2009). In doing so the process became more accurate in spends in settling ponds; terms of chemical dosages during the various treatment • Increased waste water sludge (costly stages. From Ggaba the water is pumped to reservoirs to treat and transport); at Muyenga, Gunhill and Naguru Hills which have a total • Increased sediment loads preventing storage capacity of 65 220 m3 (NWSC 2014). In 2010 the adequate filtration and disinfection intake pipeline at Ggaba III was extended further into of other pathogens and algae; Lake Victoria in an effort to extract higher quality water for treatment. From October 2010 the new pipeline was • Increased occurrence of algal in operation (NWSC pers. comm.). blooms and toxic algae; • Increased dominance by blue-green algae; • Clogging of reticulation systems by filamentous algae; • Increased occurrence of taste and odour issues in potable water and the need for activated carbon usage to eliminate these; • Wasted water on more frequent backwashing to clean clogged filters; and • Increased nutrients require the use of more complex and costly treatment technologies. Figure A7.1 Rising aluminium sulphate dosage at Ggaba WTW 1993 – 2007 Source: Ooyo 2009                                                  Page 86 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala A7.2 Water treatment pre-2007 A7.3 Water treatment post-2007 (empirical data) Previous studies have shown that one of the major Water treatment data from Ggaba WTW was analysed for economic benefits of the Nakivubo wetland has been the period February 2007 – May 2015. These data were its ability and function in treating pollution so as to supplied by the NWSC and included monthly volumes of maintain a certain standard of water in Inner Murchison raw and treated water, monthly chemical usage and costs Bay which influences water treatment at Ggaba only 3 and water quality data for the raw and treated water. km away (Emerton et al. 1998). However, over the years From the chemical cost data spreadsheets provided by the ability of the wetland to maintain this function has the NWSC it was noted that the overall usage of each decreased and the water quality entering Murchison chemical (kilograms) had been multiplied by current Bay has deteriorated significantly thereby increasing costs for each chemical (shillings per kg) (i.e. they water treatment costs at the Ggaba WTW. In 1998, COWI multiplied the amount of chemical used in 2007 by the reported that the Inner Murchison Bay had medium to current day cost of the chemical). Therefore all costs are high levels of eutrophication with Total Phosphorous assumed to be in present-day Shillings (2015). loads of 183 kg/day and Total Nitrogen loads of 1125 kg/day entering the lake. Of this it was determined During 2010 there was a period of inconsistent cost data that 85% of the phosphorous and 75% of the nitrogen that did not match the volume data or the water quality entering the inner bay was from the Nakivubo wetland data and as a result was removed from the analysis. This (COWI 1998). Ooyo (2009) found that the water quality inconsistent data occurred during the same period that in the Inner Murchison Bay was poor but improved as the new pipeline was being constructed and it is assumed one moved further out from the Nakivubo channel to that the changeover and building of the pipeline resulted the outer bay. Phosphorous levels were particularly in unusual values that were significantly different to all high, being above the recommended maximum level for other monthly observations. It was decided that these surface water of 0.1 mg.l-1, ranging from 1 – 4 mg.l-1 values should be removed from the analysis. across the monitoring stations (Ooyo 2009). Ooyo (2009) determined that the high levels of phosphorous caused Data were provided on monthly measurements of TSS intensive algal blooms in the inner bay and as a result and turbidity as well as other parameters such as colour influenced the amount of aluminium sulphate needed and E coli. Contrary to expectation, water treatment during the treatment process to adequately remove costs and TSS were negatively correlated and were not algae and associated odour and taste problems. The significant (Figure A7. 2 and Figure A7. 3). Previous aluminium sulphate dosage increased significantly from studies and the current state of Inner Murchison approximately 20 mg/l in 1993 to almost 70 mg/l in 2007 Bay would suggest that TSS, a measure of algae and (Figure A7. 1, Ooyo 2009). The level of aluminium residue suspended sediments, would have a positive and remaining in treated water can pose a threat to human significant impact on the water treatment costs at Ggaba. health and during the period 1993 - 2007 the residue Other water quality variables, such as iron and colour increased in Kampala’s drinking water from 0.01 to 0.16 were also negatively correlated with costs. Turbidity and mg.l-1, only slightly less than the maximum permissible E coli showed a positive relationship with costs but these level of 0.20 mg.l-1 (Ooyo 2009). relationships were weak and not significant. Multiple regression yielded a significant model but the signs were not as would be expected. Figure A7.2 Total treatment costs (Ugandan Shillings per m3 treated water) and Total Suspended Solids (TSS, mg/l) for the period 2007 – 2015.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 87 Figure A7.3 The negative relationship between treatment costs (Ugandan Shillings per m3 treated water) and TSS from 2007 – 2015. Figure A7.4 (a) Increasing water treatment costs (Ugandan Shillings per m3) over the period 2007 – 2010 and (b) increasing residual aluminium levels (mg/l) in treated water for the same period. Figure A7.5 (a) Increasing water treatment costs (Ugandan Shillings per m3) over the period 2007 – 2010 and (b) increasing residual aluminium levels (mg/l) in treated water for the same period.                                                  Page 88 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala In 2007 Ggaba III became fully automated, making The use of PAC during the treatment process resulted chemical dosage more accurate and the water treatment in a decrease in the use of Alum, thereby reducing the process more efficient. However, water treatment costs residual aluminium in the drinking water. The use of PAC continued to rise from 2007 until mid-2010, as did the as a flocculent also reduced the amount of backwashing residual aluminium sulphate present in the treated needed to remove algae clogged in filters. This can be drinking water (Figure A7. 4). seen in Figure A7. 6 below; the difference in the volume of raw water entering the plant and the amount of As a result of the rising residual aluminium sulphate treated water decreases significantly from mid-2010 levels in the treated drinking water, the NWSC switched when the polymer PAC was used for the first time. to using a combination of aluminium sulphate and polyaluminium chloride (PAC) in the treatment process It was expected that water quality would have continued in 2010. PAC is a synthetic polymer that dissolves in to deteriorate over time, but with an abrupt change to water and is preferred to Alum because it has a lower overall better quality in late 2010 when the extended dosage requirement, often does not require any form pipe came into operation, followed by a new path of of neutralising agent, has a shorter flocculation time, deterioration. However, no such trends were observed produces less sludge and requires less backwashing for either turbidity or TSS (Figure A7. 7). E coli counts (Graham 2004). The use of PAC at Ggaba III commenced did increase over time, but without any improvement in at approximately the same time as the extension of 2010 (Figure A7. 7). It should be noted that water quality the abstraction pipe further into Murchison Bay. It is measurements are taken where the piped water enters assumed that these two changes had a significant impact the plant. on both the water treatment costs and the quality of the drinking water being produced as both costs and residual alum levels started to decrease towards the end of 2010 (Figure A7. 5). Figure A7.6 The difference in the volume of raw water and treated water at Ggaba 2007 - 2015                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 89 Figure A7.7 TSS, E coli and Turbidity trends over time. After meeting with the NWSC in Kampala and discussing A model was developed for the Durban Heights Water with them the results of the data analysis it was agreed Treatment Works in South Africa (as part of the Durban that the main drivers of use of chemicals in water case study), using monthly treatment cost data over a treatment are E.coli and nutrient concentrations within period of five years (July 2010 – June 2015), data on raw the raw water. Other relevant water quality data, such water quality and on nutrient loads entering the Nagle as chlorophyll levels were not made available in time Dam from which the raw water is drawn. The model had for this study and in any case it is expected that these to be designed in such a way that it could be applied in a data are unlikely to be able to explain the decrease in situation of more limited data from Inner Murchison Bay. treatment costs since 2010. It is expected that these Thus the model was developed with treatment costs as a decreases in costs are a result of improved human function of phosphorous loads, coliform counts, colour, efficiency in operating the treatment process at the WTW temperature and conductivity. Using these values and and of the change to PAC as discussed above. Variables a standard water treatment cost model developed for such as human efficiency are unobservable within the water treatment in Durban we were able to calculate analysis and it was decided that the use of a benefit the water treatment cost savings as a result of improved transfer approach from another water treatment plant water quality entering Murchison Bay. The technology that is not complicated by such factors would be a more and chemicals used to treat water at Durban Heights beneficial approach to use here. and Ggaba WTW are comparable as are the dimensions of the water supply reservoir in Durban and Inner A7.4 Estimation using benefit transfer Murchison Bay. Discussions with the NWSC confirmed that water quality deterioration still has an impact on costs and The water treatment cost savings at Ggaba were will continue to do so under the new plant. Therefore estimated based on the assumption that total in the absence of a useable model for the Ggaba plant, phosphorous concentrations in the wetland would it was decided to value the potential cost savings using improve substantially through the restoration a benefits transfer method. This is essentially the use interventions and, as a result, total phosphorous of a model developed elsewhere, but with parameters concentrations in Inner Murchison Bay would ultimately adjusted as far as possible to correct for differences decrease to a mesotrophic condition, resulting in between the source and receiving valuation site. treatment cost savings due to fewer algal blooms. The cost saving estimate is that obtained when this level of lake condition is reached.                                                  Page 90 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Table A7.1 Estimated capital and maintenance costs for the proposed Nakivubo Wetland Park based on the concept vision used for this study Coefficient Std. Error t.value Pr(>|t|) Intercept 47.780 23.007 2.077 0.044 * Phosphorous Load 0.015 0.005 3.381 0.002 ** Temperature 0.685 0.674 1.016 0.316 Coliforms 0.006 0.003 2.261 0.029 * Colour 0.501 0.403 1.242 0.222 Conductivity 0.980 1.239 -0.791 0.434 R-squared 0.481 Adjusted R-squared 0.416 Notes: (1) ***p<0.001, **p<0.01, *p<0.05, .p<0.10. Table A7.2 Estimated capital and maintenance costs for the proposed Nakivubo Wetland Park based on the concept vision used for this study Pre-Restoration Post- Restoration Phosphorous concentration (mg/L) 2.5 0.047 Phosphorous load (kg/yr.) 8088 152 Water treatment cost ($/ML) 22.14 8.52 Saving per ML ($/ML) 13.60 Annual cost saving (US $ millions) 0.845 Data on phosphorous concentrations in the wetland and The treatment cost model estimates are shown in average flow data for the Nakivubo wetland were used Table A7.1. The result from the model was computed to calculate phosphorous loads entering Inner Murchison as a Rands per ML (R/ML) value. This value was then Bay (IMB) under a pre-restoration scenario. Currently converted into US $ using current 2015 conversion rate. the phosphorous concentrations in the lower wetland The results of the analysis show that the water treatment are extremely high at around 2.5 mg/L. The restoration costs would decrease from $22.14 to $8.52 per ML of interventions are aimed at reducing these concentrations treated water (Table A7.2). This is a saving of $13.60 per to ensure the wetland is in a mesotrophic state, which ML which equates to $845,000 per year based on the at the upper end of this state, has a phosphorous current supply of 170 ML per day at the Ggaba WTW. concentration of 0.047 mg/L. This value was used to Currently the average cost of treating one ML of water at determine the post-restoration phosphorous load using Ggaba is approximately $30/ML (Figure A7. 2). Therefore flow rates from the wetland. Based on these values, the the water treatment savings that would occur as a result pre-restoration annual phosphorous load entering the of restoration are significant with post-restoration costs Lake was estimated to be 8088 kg and post- restoration being 38% of pre-restoration costs. 152 kg. The model was updated using average values for water quality parameters found in the raw water The difference between the model estimate of $22.14/ in Inner Murchison Bay. Average temperature, average ML pre-restoration and the actual current cost at Ggaba colour and average conductivity were used based on of $30/ML is possibly a result of having to use average data analysed from NWSC for the raw water at the coliform data from Durban as well as external effects Ggaba WTW. Because Coliform data was not available such as plant efficiency which are unobservable and for Inner Murchison Bay or for the raw water at Ggaba, expected to differ between the Durban Heights and the average coliform count from Durban was applied. Ggaba WTW. It was assumed that not only will phosphorous loads decrease as a result of wetland restoration but so too will the coliform counts. The decrease in coliforms and in phosphorous was then assumed to improve the colour of the water being treated as a result of fewer nutrients, bacteria and algae in the raw water. A 66% reduction in coliforms and in colour was applied to the post- restoration scenario.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 91 A7.5 Conclusions Appendix 8. Recreational demand for The empirical data that was provided by the NWSC open space areas in Kampala and potential produced unexpected results that differed from benefits of restoring Nakivubo wetland personal comments and understanding by the staff of A8.1 Introduction the current water treatment situation at Ggaba. The Urbanisation in Africa is occurring at an unprecedented general understanding is that water treatment costs are rate as the rural poor move into cities in search of still increasing as a result of poor water quality in Inner employment and other opportunities. This has had a Murchison Bay. Current pollution levels entering Inner significant impact on natural and semi-natural green Murchison Bay from the Nakivubo wetland are high and open space areas in and around urban centres and the have been increasing consistently over the years as the benefits that they provide to city inhabitants. Urban wetland condition has deteriorated. It was expected green open space not only provides opportunities therefore that the water treatment data would show for recreation and tourism, it provides a refuge from deteriorating water quality in Inner Murchison Bay and urban dis-amenities such as noise, traffic congestion related increases in water treatment costs. This, however, and pollution (Anderson & West 2006, Kroeger 2008), was not the case and the data did not conform in any helps to reduce stress (Ulrich & Addoms 1981, Kaplan way to what was expected. Although the technology at 1983), and promotes emotional, psychological and Ggaba WTW has been upgraded and the water intake physical health (Chiesura 2004, Anderson et al. 2013). A pipeline has been extended further into Inner Murchison decrease in availability of suitable green open space can Bay, one would still expect to see increasing treatment lead to increased travel costs and congestion, as well as costs as a result of the poor water quality. As a result of decreased health, productivity and economic output. It not being able to use the empirical data to determine a can also contribute to urban sprawl, as city inhabitants water treatment cost model the benefit transfer method move to the urban fringe in order to have access to green was used. This proved to be relatively successful and was open space. However, investment in the maintenance of based on the assumption that the effect of phosphorous green open space areas is often given low priority due to loads on algae and on treatment costs were similar at a lack of appreciation of their value (TEEB 2010). Durban Heights and Ggaba. The analysis showed that a reduction in phosphorous concentrations in the lower In Uganda, the city of Kampala has undergone a period wetland as a result of the restoration interventions of rapid urbanization that has significantly contributed would have significant impacts on chemical costs and to the degradation of the quality of the city’s natural the saving per ML of treated water would be almost half environment. The city is located on the edge of Lake of the current day cost. An annual saving of $845 000 is Victoria, and has developed on a hilly landscape that is estimated if restoration interventions for the Nakivubo drained via a number of wetlands that are connected wetland are implemented. to the lake. Since terrestrial green open space areas within the city have either been lost to development or Other associated treatment costs expected to decrease are under private ownership, the wetlands are the last as a result of improved water quality entering Murchison publicly-accessible green open space within the urban Bay include sludge removal and backwashing. The area. However, they have been used as conduits for reduction in phosphorous and other nutrients as a result waste water, have been altered and encroached upon by of restoration initiatives are expected to decrease the agriculture and informal settlements, and are generally frequency and intensity of algal blooms in the inner bay highly polluted and degraded. With few opportunities in and will therefore decrease the need for continuous the city, Kampala residents tend to travel to the outskirts backwashing of the filters. Sludge dewatering and sludge of the city or beyond to visit outdoor recreational removal can be very expensive and it assumed that the areas, contending with and contributing to heavy traffic improved water quality will decrease the amount of congestion along the way. sludge produced during the water treatment process and will therefore reduce the time and costs involved in One of the main wetland systems in Kampala is the having it removed. Nakivubo wetland which has its source in the city centre and flows into Lake Victoria in Murchison Bay, close to the harbour area at Port Bell. While highly degraded at present, the lower part of the wetland has been recognised as a potential location for the development of a recreational park that encompasses wetland and lake shore waterfront areas. Bringing this about may require considerable investment in restoration and infrastructure which the city might be reluctant to make unless there is evidence that the benefits would outweigh the costs. This requires estimation of the potential demand for a new/improved recreational site, as well as the other co-benefits that will result from wetland restoration.                                                  Page 92 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Importantly, the addition of a new recreational site close A8.2 Conceptual Framework to the city centre has the possibility of generating welfare A8.2.1 Multi-site choice model of revealed preference for citizens that were not able to access the currently- available sites due to the costs involved. Our econometric framework of modelling multi-site choice of outdoor recreation is informed by a random The demand for outdoor recreational areas is typically utility framework. In the latter, individual choices aren’t analysed using a random utility maximisation model completely deterministic, instead, choices are affected framework that links the frequency of site visitation to by alternatives that are not included in the survey and individual attributes, the characteristics of alternative other unobservable individual’s and site characteristics sites available, and the travel costs required to reach (Hannemann & Kanninen 1996). McFadden (1974) made each site (Babatunde et al. 2012). The models are then use of this intuition in developing the random utility used to infer the value that households place on access model, which was accomplished through the inclusion of to sites and/or changes to site characteristics. Such an error term in (1). information is very useful to managers of these public open space areas. Recreational demand studies that involve multiple sites In this specification, we assume a linear-in-parameters have been dominated by studies of angler site choice specification for indirect utility, where we denote the (Hunt 2006). Although one has to take care to minimise individual with subscript q, the choice with i, and choice issues of endogenous stratification and truncation, alternative with t. Note also that α_iq is the individual studies of user groups such as anglers are probably alternative-specific intercept that captures the intrinsic best carried out using on-site survey methods, because preference for the alternative, γ_i captures systematic the proportion of anglers in any population is usually preference heterogeneity related to socioeconomic fairly low. Some of these studies have explored the characteristics, where s_q is a vector of socio-economic implications of a reduction in the number of quality characteristics; β_q captures systematic preference of angling sites (Provencher & Bishop 1997). Few, if heterogeneity related to program attributes, x_iqt is any studies have been carried out on site choice for the vector of attributes (including costs) for alternative general outdoor recreation by urban residents, or have and ε_iqt is stochastic and accounts for observational investigated the potential gains in welfare associated deficiencies due to unobservable components in the with the restoration or addition of recreational sites. model that are assumed to be uncorrelated with the Unlike angling, general recreation (e.g. swimming, observed components. picnicking, and walking) is likely to have a very high level of participation among urban households, and The above functional specification leads to a conditional whereas a loss of a site may simply reduce the welfare logit model, which is often used to model discrete in an existing user group, in a case where opportunities choice behaviour under the random utility framework may be increased, this may induce the participation (Greene & Hensher 2003, Train 1998). However CL suffers of new households. Therefore it was considered more from restrictive assumption of IIA, and, therefore, fails appropriate to gather our data from a general household to account for unobserved heterogeneity (parameter survey. variation) and potential correlation between available choices. In light of this limitation recourse to more Recreational studies typically employ revealed or stated flexible approaches are desirable. One such o is the preference methods or some combination thereof. While RPL. In this study, in addition to the base CL model, we stated preference methods are usually regarded as less applied RPL model preferences for outdoor recreation reliable than revealed preference methods, which model sites. The RPL generalizes the CL by allowing coefficients actual behaviour, they may provide important validation to vary randomly over individuals, rather than being where the situation involves introduction of a new choice fixed (Train 1998), and, therefore, it relaxes IIA and can that may be outside the current set of experiences represent any substitution pattern. Furthermore, the RPL and thus simulation of the new situation may involve explicitly accounts for unobserved heterogeneity (Train extrapolation beyond the range of parameters used 1998, Carlsson et al. 2003). to estimate the model. In this study, we used revealed preference data on frequency of travel to a number of sites in and around Kampala, and augmented this with data from the same respondents on their interest in utilising and their willingness to pay for access to the proposed Nakivubo Wetland Park.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 93 Conditional Logit In general, (1) is a logit model with both alternative- varying and alternative-invariant regressors. As such, it constitutes a mixed logit model. However, following However, the integral in (8) does not have an exact the literature, we restrict γ_i=γ and α_iq=α, since the solution. Therefore, we estimate via simulated maximum real interest is in attribute preferences, rather than likelihood (Revelt & Train 1998, Train 1999). Furthermore, alternative-specific effects. Assuming β_q=β and that it is assumed that there is correlation between the each of the errors is identically and independently randomly distributed parameters, and, therefore, we distributed (IID) type 1 extreme value, a CL can be estimate the full variance-covariance matrix of the estimated based on the following probability model. parameter vector, Σ_β, assuming a normal distribution, i.e., β_q~N(β ̅,Σ_β ). Given the Cholesky decomposition of the variance-covariance matrix, β_q=β ̅+Cη_q, where η_q is a vector of standard normals. In other words, we estimate both β ̅ and C, such that CC^’=Σ_β. Estimation can proceed with data that is pooled over all Despite the desirable properties of RPL, allowing of the choice experiments. for individual heterogeneity and correlation across alternatives, it is subject to restrictive assumptions. In Finally, using parameter estimates from either model, this case, those assumptions are based on the assumed we calculate both marginal WTP for a change in each of distribution of the coefficient vector. The two most selected attributes of recreations sites and total WTP common are: (a) the log-normal distribution, which for a joint change in these attributes. By restricting restricts all individuals to have positive coefficients for total WTP to a joint change of attributes that matches the variable, and (b) the normal distribution, which with improvement in Nakivubo wetland as specified in allows for both positive and negative values. However, CVM, we estimate the welfare impact of developing the there is no rule of thumb to select the distribution. As wetland park via revealed preference approach. with any sort of model misspecification, the estimated results could be biased if the distribution is misspecified Random Parameters Logit and Latent Class Logit Models (Carlsson et al. 2003). The RPL, on the other hand, extends the CL, by allowing the coefficient vectorβ_q, to vary across the population An alternative to resolve the problem is a recourse according to the densityf(β|θ), where θ is a vector of the to non-parametric or semi-parametric models that parameters of the distribution, while the CL assumes that requires no or few assumptions about the parameters the preceding density is degenerate. Assuming the error distribution. Latent Class model (LCM) provides the terms are IID type 1 extreme value, an RPL or mixed logit remedy to distributional problem as it is semi-parametric model (Train 1998) can be specified. Following Carlsson model. Latent class model largely resemble RPL by way et al. (2003), the conditional probability of alternative i of accounting for preferences heterogeneity, albeit for individual q in choice situation t can be specified. the heterogeneity is modelled as discrete parameter variation (Greene & Hensher 2003). In this study, we will also estimate LCM in addition to RPL estimation to compare welfare estimates of the program change. In LCM, individual are sorted into set of class or segments One of the maintained assumptions in the RPL inherent of the population which is not observed by the analyst. in (6) is that individual utilities vary, but are stable across Assume that there exist segments in the study the different choice experiments (Train 1999). population that individual belongs to segment . The Given (6), the conditional probability of observing utility function in (1) can thus, be re-specified as a sequence of choices is simply the product of the individual choice probabilities for each choice set. Denoting j(q,t) as the sequence of choices from all of the Utility parameters are now segment specific (Boxall & experiments, the conditional probability can be written, Adamowicz 2002) and equation (2) becomes as in (7). Where and are segment specific utility and scale Due to the variation in the β_q parameter vector, the parameters respectively. preceding conditional probability needs to be integrated over the assumed density, in order to arrive at the unconditional probability in (8).                                                  Page 94 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Following Swait (1994), we specify that the probability that an individual be in segment is given as; Where a vector of socio-economic and psychometric variables, is a vector of parameters and is a scale factor (Boxall & Adamowicz 2002). The joint probability of a randomly chosen individual chooses alternative and be in segment is given as where T is the bid offered to the respondent, belongs to and is a continuous and increasing function such that Substituting this for choice equation (6) and membership there is a spike (discontinuity) at zero. The spike model equation (7) probabilities yields uses two basic valuation questions (a) whether or not the respondent has non-zero WTP and (b) whether or not the respondent would want to pay the price, T suggested A8.2.2 Contingent valuation in the dichotomous choice question. Define an indicator In the second part of the study, we estimate WTP for use that indicates if the respondent has nonzero WTP or not: of the proposed Nakivubo Wetland Park in a referendum- . Similarly, let indicate if the respondent is willing to pay style contingent valuation study. We estimate a simple the suggested bid price, T: .The log-likelihood for the sub- spike model to analyse closed-ended responses and sample is then given by: allow respondents to be indifferent about the project (Kriström 1997). Zero responses are common in open- Assuming that WTP has a logistic distribution and the ended valuation studies and the spike model specification WTP function is linear in the parameters, the response allows us to take zero-bids into account. For public goods, function is given by: closed-ended styled contingent valuation studies often assume that all respondents are “in-the-market”. Popular distributional assumptions (e.g. log-logistic, lognormal, or Weibull) together with other popular models using continuous distributions (e.g. logit, probit models) imply that all respondents have positive WTP (Kriström 1997). These assumptions exclude respondents who have zero where the vector of socio-economic characteristics of the WTP for the good in question. respondents is, is the corresponding parameter vector and is the marginal utility of income. The mean WTP is Respondents may select not to reveal their true WTP then given by and the median WTP is given by (Kriström, for using Nakivubo Wetland Park. They may find that 1997). the Wetland Park doesn’t contribute positively to their welfare. One obvious reason being availability of alternative substitute outdoor recreation sites, which A8.3 Data collection may be better and cheaper in terms travel cost of visit. A8.3.1 Recreational sites and attributes Another reason has to do with sentiment, as has been voiced during the survey, that displacing poor people Prior to the household survey, we used the help of who are currently living around the proposed wetland local experts to compile a list of outdoor recreational Park without compensation is not morally supported. sites, and to create a map of these sites. Data were The simple spike model allows us to split the sample into then collected on 10 purposely selected major outdoor two categories: respondents with zero WTP and positive recreation sites in Kampala. The site attributes chosen WTP for the establishment of Nakivubo Wetland Park. were presence of natural amenities; forest, wildlife, To determine if the respondent is indifferent or not (i.e. water view and facilities; paved walkway, parking lot, if the respondent has a zero WTP) we use the answer to restroom, playground and picnic tables and boats. These the follow-up open-ended question. attributes were expected to be salient attractions of the sites to drive outdoor recreation demand. The closed-ended question asks the respondent whether he/she will accept or reject the Nakivubo Wetland Park for a given bid T. The project is represented as the change of environmental and physical qualities (infrastructure, environmental amenity etc.). We define WTP for this change as:                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 95 A8.3.2 Survey design development in the future. For each respondent who A survey was conducted in Kampala, Uganda over replied affirmatively, we randomly assigned an entry the period of 20 - 28 July 2015. A two stage sampling fee from the bid vector asked if he/she is willing to pay design method was used, in which a set of parishes it as an entry to visit the new park in the future. This (an administrative unit equivalent to a suburb) was closed-ended question was followed by an open-ended selected, and then households were selected within question; “what would be the daily per-person entrance each of these parishes. In the first stage, parishes were fee for which you would decide not to visit the site randomly selected from all five divisions of Kampala (assuming children pay half-price)”? (Central, Kawempe, Lubaga, Makindye and Nakawa). In the second stage, drawing on the 2014 Ugandan census A8.3.3 Description of data data, we randomly selected households within each Table A8.1 presents descriptive statistics of the data used Parish. A total of 644 households were interviewed, with in the empirical analysis. Although the primary purpose interviews being conducted with an adult member of the of our analysis is to examine preferences for various types household (aged 20 or more). The survey was conducted of outdoor recreation sites in Kampala, it is expected that by 11 trained and experienced enumerators, and was individual, household and site level characteristics are closely supervised by one of the authors. likely to affect the demand for various recreations sites and the expressed willingness to pay for new wetland park The survey instrument comprised three parts: socio- establishment. The study included variables that vary at economic questions, revealed preference questions the individual (household) level: age, gender education about household use of multiple outdoor recreational and income of a respondents and household head. We sites during the last 12 months and a contingent also included household income range. valuation section on the restoration of the Nakivubo wetland for its development as a recreational park. Key In terms of socio-economic covariates, we found that 52% questions regarding multi-site choice were whether an of the respondents were male, the average age was 30, individual had visited any outdoor recreation site during and average education was 15 years. Corresponding to the last 12 months, how many trips they made to visit this, we observed that 87% of households were headed by the sites during the same period and the travel cost males, the average household head was 43 years old and incurred during the last visit. has 17 years of education. About 2.3% of the households fell in the Ugandan lowest income class whereas 23.4% In the contingent valuation part of our questionnaire, falls in top income class of the same. These estimates we described proposed change in quality of Nakivubo compare with Uganda National Household Survey 2012/13 wetland from its current degraded (deteriorated) state estimates of 2.1% and 23.5% respectively. to one of being Recreation Park for Kampala public. The changed to a new state was described as improved water On average, households visited outdoor recreation quality in the area, and provision of facilities such as sites about 5 times within the last 12 months and lakefront picnic areas, and walking and cycling paths in paid an average of 18 744 USh as travel cost per visit. and around the wetlands. This was followed by questions Almost all households (92%) anticipated visiting the to elicit the households’ willingness to pay to use the Nakivubo Wetland Park after it is developed, and 72% of proposed area. households were willing to pay at least 2000 USh as entry fee to visit the park. We chose single-bounded dichotomous (SBDC) CVM value elicitation format because of its incentive We expect that household income is positively related compatibility advantage compared to other formats. to outdoor recreation demand as well as increase the As SBDC is a closed-ended value elicitation format, this likelihood of willingness to pay for development of new required the design and generation of a price vector of park ceteris paribus. From economic theory, travel cost starting bids. The bid vector was obtained from a pilot is expected to be inversely related to the frequency of study of 40 randomly-selected households, in which visits to outdoor recreation sites, but is likely to raise an open-ended CVM question format was used. Using propensity of willingness to pay for a new recreation park the response to open-ended CVM question of the pilot as a substitute alternative. As natural beauty is critically study, we generated a vector of five starting bids; {2000, scarce for households residing in CBD areas, respondents 6000, 12000, 14,000 and 20,000}. We used entry fees from such areas are expected to demand more outdoor as the payment vehicle. Respondents were asked if recreation than others. they anticipated visiting the new Nakivubo Park after its                                                  Page 96 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Table A8.1 Descriptive statistics used in the empirical analysis Variable Description Mean Std.Dev. Min Max age Respondent’s age 29.55 9.719 19 75 qn10a der Frequency of site visit in a year 4.689 3.119 1 9 qn2b Respondent’s gender 0.528 0.499 0 1 qn3 Respondent’s education level in years 15.21 5.230 3 21 qn4a Household head’s age 43.23 12.10 23 87 qn4b Household head’s gender 0.867 0.340 0 1 qn4c Household head’s education in years 16.98 4.810 4 22 qn5 Member household earning income 1.987 1.010 1 5 qn19a1 Travel cost to a site 18744 15079 500 65000 qn24 Willing to visit Nakivubo park regardless 0.918 0.274 0 1 qn26 Willing to pay to visit Park establishment 0.718 0.450 0 1 qn27 Open-ended willingness to pay for Nakivubo park 16935 13650 1000 50000 establishment income1 Income range: 50,000 USh 0.0234 0.152 0 1 income2 Income range: 50,000-100,000 USh 0.0714 0.258 0 1 income3 Income range: 100,000-200,000 USh 0.127 0.333 0 1 income4 Income range 2000,000-300,000 USh 0.0714 0.258 0 1 income5 Income range 300,000-500,000 USh 0.206 0.405 0 1 income6 Income range: 500,000- 1,000,000 USh 0.266 0.442 0 1 income7 Income range >1,000,000 USh 0.234 0.424 0 1 A8.4 Results In comparing CL and RPL models, we observe that the A8.4.1 Revealed Preference Analysis latter outperformed the former for the following major Utility Parameter Estimates preseasons: (a) the standard deviations of coefficients all site attributes attribute are statistically significant We estimated the utility function parameters ( using implying heterogeneity in preferences among the conditional logit (CL), random parameter logit (RPL) respondents (b) following Greene & Hensher (2003) model and latent class model (LCM). Table A8.2 presents comparing this owing to log-likelihood value shows that the estimated utility function parameters from the RPL is an improvement over CL model. standard conditional, random parameter logit model and latent class model. Furthermore, the result of latent class model as evidenced by statistically significant class probabilities In all models, we included one common alternative- attested the presence of discrete preference specific intercept for the ten alternatives (sites). In RPL, heterogeneity. Based on Bayesian information criterion we assumed coefficient of travel cost to be fixed and not (BIC) statistics, we selected three classes of study sample. randomly distributed for two reasons: (a) the distribution The statistically significant class probabilities suggest the of the marginal willingness to pay for site attribute is the presence of discrete preference heterogeneity across same as the distribution of that attribute’s coefficient this class, which makes LCM different from CL. As RPL is (b) we wish to restrict the travel cost coefficient variable not nested within the LCM, there is no comparison test to be non-positive for all individuals. All of the non- available for these two models within classical statistical/ price attributes are randomly distributed with normal econometric tradition. However, this limitation can distribution. We also included some socio-economic be overcome by Bayesian model selection method to variables age, income class and household head’s choose the model that best fits the data at hand. Overall, education as fixed coefficients variables and these the heterogeneity in preferences for attributes among interact with the alternative-specific coefficient. individuals in RPL and across population classes in latent class model (LCM), make us prefer these models to the conditional logit model.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 97 Table A8.2 Estimates of utility parameters Random parameter logit Latent class logit Conditional Variables logit Coefficient Standard Class 1 Class 2 Class 3 deviation Paved 0. 8029 1.4136 3.3428 3.6689 0.31107 -22.6268 (0.1479)*** (0.6831)*** (1.9020)*** (13.256) (0.1724)** (154.152) Size in ha -0. 0179 -0.0574 0.0523 -4.5326 0.01858 -0.00875 (0.0046)** (0.0273)*** (0.0200)*** (30.808) (0.00679)*** (00860) forest -0.9927 -1.5680 1.6301 2.7513 -2.3348 -0.9854 (0.2364)*** (0.46960)*** (0.7440)*** (28.599) (0.4333)*** (0.2803)*** Water-view 0.03310 0.2919 1.241 4.5326 -0.54090 23.7677 (0.2664) (0.4069) (0.5743)*** (30.809) (0.4879) (154.152) Kiosk 0.9590 1.8312 1.4636 13.2013 3.4583 1.10884 (0.1952)*** (0.7498)*** (1.2997) (52.259) (0.58710)*** (0.2562)*** Travel Cost -0.12260D-04 -0.1662D-04 -0.3713D-04 -0.20556D-04 -0.7063D-04 (0.7047D-05)* (0.1029D-04)* (0.146D-04)** (0.8119D-05)** (0.10287D-04) Intercept 1.068 1.536 -16.3511 4.5816 24.4192 (0.371)*** (0.623)*** (30.506) (0.8852) (154.152) Age -0.03350* 0. 03174 -0.1244 -0.10627 -0.01929 (0. 0173) (0. 01689) (0.0672)** (0.0489)*** (0144) income 0.1342 0. 1291 0. 3374 -0.4405 0.1126 (0.0914) (0.0896) (.01593)*** (0.35424) (0868) education -0.0828** -0.07945 -0.33419 27.7228 -0.00406 (0.0298) (0.0296) (0.1555)*** (71.8976) (0.00545) Respondent 548 548 size Observation 5,480 5,480 size Class 0.3584 0.38003 0.26149 probability (0.0408)*** (0.0388)*** (0. 04737)*** Log- likelihood -839.7574 -831.9344 -792.8639 -792.8639 -792.8639 Standard error in parentheses, *** p<0.01, ** p<0.05, * p<0.1 In all models, the estimate of travel cost parameter RPL offers more information. It shows that preferences is negative and statistically significant, corroborating for paved walkway are heterogeneous among visitors with economic theory; as travel costs rises, demand for as suggested by standard deviation (SD) parameter of outdoor recreations falls. Moreover, household income the preferences for this attribute. In fact, its SD is the class coefficient has positive sign in all but class 2 of LCM. highest indicating that the preferences are the most For the most part, its effect is statistically insignificant, heterogeneous one for this attribute. However, SD but in class 1 of LCM, the latter of which supports our parameter for kiosk attribute is not statistically significant priori expectation from economic theory. suggesting that there is no individual level preference heterogeneity for this attribute. For both attributes With regards to attributes, only two of five selected though, we see that preferences vary a great deal across attributes; paved walk way and presence of kiosk and classes of LCM model. canteens, have positive and statistically significant effect on the demand for outdoor recreation. The results show Of the remaining attributes, the results show that water that these attributes are the most attractive feature of view attribute has no effect on visitor’s preference in all outdoor recreation’s sites in Kampala. the models. To our surprise though, results from CL and RPL show that Kampala visitors strictly prefer smaller size of recreation sites and absence of forest although these preferences are heterogeneous across individuals (see SD of RPL parameters). Results from LCM analysis also largely support these finding although we still observe class level preference heterogeneity for both of the attributes. Overall, our evidences of heterogeneous across visitors and classes of visitors are in line with general findings in related literature (see Hynes et al. 2008, Scarpa et al. 2007, Scarpa & Thiene 2005, Breffle & Morey 2000).                                                  Page 98 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Table A8.3 Marginal Willingness to Pay (MWTP) and Compensating Variation for selected attributes Latent class logit Conditional Random Variables logit parameter logit Class 1 Class 2 Class 3 Size -1465.6073 -3457.9932 -18692.716 904.204250 -1240.136 (945.54723) (2732.683) (183763D+12) (511.0674) (2189.484) Paved 65465.017 85056.478 98793.0645 15132.654 -0.32037D+07 (40000) (61048.84) (356959D+12) (10035.522) (21856D+10) Forest -80950.87 -94353.592 74086.2002 -113584.173 -139530.645 (49992.026) (60806.33) (770095D+12) (51187.837) (209435.16) Water view -2699.656 -17563.4380 122051.138 -26313.307 -336521D+07 (21722.469) (25672.117) (829579D+12) (26500.780) (218557D+10) Kiosk 78198.6344 110185.125 122051.138 168236.237 156998.201 (45704.22) (71234.9460) (829579D+12) (74445.716) (231437.258) CV(WTP) 145688 66034.6039 116639.845 270874.695 -0.310017D+07 (73438.015) (78718.58) 356959D+12 111274.775 0.218557D+10 Welfare Measures Returning to LCM results, we see a slightly different set The preceding coefficient estimates, although interesting, of results compared to the results of the two preceding do not provide for straightforward interpretation, due models. Particularly, marginal willingness to pay of each to the differences in the estimation models and the attribute varies a great deal across population classes scale factor associated with these values (Greene & (segments). Paved walk way appear to have positive Hensher 2003). In order to improve interpretability, welfare effect in class 2, but statistically insignificant marginal rates of substitution between the attributes effect in rest of the classes. Likewise, we find that were computed, using the negative of the travel cost average individuals in class 2 and 3 derives positive coefficient as the denominator, and the distribution of welfare gain from kiosk attribute, but this effects appears these ratios is obtained via the Krinsky-Robb (Krinsky & to be negligible for average individual in class 1 of our Robb 1986) method. These ratios can be interpreted as study sample. In terms of remaining attributes, variation the marginal WTP for a change in each attribute (Greene of MWTP estimates were observed across LCM classes & Hensher 2003, Train 1998, Carlsson et al. 2003, although, for most part, they appear to be statistically Hannemann & Kanninen 1996). The calculated marginal insignificant as vindicated by their corresponding high rates of substitution or marginal willingness (MWTP) to standard deviation SD. For site size though, we observe pay between the attributes are presented in Table A8.3. a positive and statistically significant MWTP in class 2, Because the parameters are normally distributed, and but zero MWTP in remaining classes. Presence of forest the travel cost coefficient is fixed, the calculated marginal in sites is welfare reducing in class 2, but has no welfare WTP is also normally distributed. effects in rest of the classes. Water view attribute has no significant welfare effect in any of the classes as is the Marginal willingness to pay (MWTP) estimates of both case in CL and RPL analysis. CL and RPL models, consistent with utility parameter results, shows that only paved walk way and kiosk In addition to MWTP analysis, we also estimated attributes offer a positive welfare gain. In either models, compensation variation (CV) or total willingness to pay we see that average visitor enjoys the highest welfare (WTP) for a joint change in all selected attributes as gain from presence of kiosk and canteens followed by opposed MWTP, which estimate the welfare effect of a that of presence of paved walkway in a recreation site. change in a single attribute. As can be seen from last row The remaining attributes have negative but statistically of Table A8.3, CV estimates are positive and significant insignificant welfare effects in both models. in all of our models, but it turns out to be insignificant in class 1 and 3 of LCM. The CV estimates ranged from 660 346 USh ($18.34) in RPL to 2 708 755 USh ($75.24) in class 2 of LCM. The results suggest that a simultaneous change in all of these attributes, as for example for establishment of new outdoor recreation site, is welfare improving for average visitors in Kampala.                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 99 A8.4.2 Stated preference analysis Analysis of bid function WTP Estimates Table A8.3 presents the estimated results from the The bid-response data conform to a priori expectations, simple spike model and Klein & Spady (1993) semi- as informed by economic theory; where quantity parametric (KSS) specification of the binary choice demanded falls with increasing price (Figure A8.1). model. A total of 620 observations were included in the analysis. Our discussion of the bid function Mean WTP was estimated from a simple spike model as analysis is however, based on the results from the 18 077 USh ($5.64) with the ratio of the 95% confidence latter specification for the following reasons. Normally, interval to the mean being 0.072. This result rises to 19 parameters of discrete-choice models are estimated 552 USh ($6.11) with 0.12 ratio of the 95% confidence by maximum likelihood which imposes assumptions interval to the mean when estimated from standard on the distribution of the underlying error terms. probit model. Although the WTP estimates are different, Under correctly-specified distributional assumptions, there is significant overlap in their confidence intervals, parametric maximum likelihood estimators such as implying that equality of WTP cannot be rejected. Note spike and probit models are known to be consistent and also that both of the mean WTP estimates are larger than asymptotically efficient. Nevertheless, departures from that obtained from an open-ended question, which was the distributional assumptions may lead to inconsistent 16 935 USh ($5.29). This supports findings from previous estimation (De Luca 2008). In our case, the likelihood studies in which SBDC format is followed by open-ended ratio test of the probit model against the KSS model elicitation question (Gelo & Koch 2015). rejects the Gaussianity assumption of normal distribution suggesting that ordinary probit function or simple spike specification with covariates may not be appropriate for our data. This result is also supported by estimated marginal densities of the error terms (Figure A8.2). Figure A8.1 Survival function showing proportion of respondents willing to pay the suggested fee in relation to the fee posed in the questionnaire Figure A8.2 Marginal density of error terms from KSS model                                                  Page 100 Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala In the analysis, the parameters, which capture the link A8.5 Estimated aggregate recreational between socio-economic covariates and WTP, for the benefit most part, are consistent with our a priori expectations. Based on a total of over 400 000 households in Kampala The results suggested that female respondents were City, the results of the revealed preference analysis more likely to visit Nakivubo in the future and be suggest an overall net benefit to Kampalans of some willing to pay entry fees. However, education reduced $17.0 - $29.4 million per annum. Using the mean willingness to pay an entry fee for visiting the park. Age household size, mean adult:child ratio, and mean of the respondent, on the other hand, did not seem to be expected level of use per household (2.3 times), the of significance in influencing WTP for visiting the park. results of the stated preference analysis yield a slightly lower overall estimate of $14.7 - $15.9 million per Mean WTP increased with household income, but annum. Nevertheless the estimates are relatively close. decreased with the entry fee (price) of visiting the Wetland Park. In other words, outdoor recreation services of the new park are seen as normal goods as its demand increases with income. Furthermore, the demand for this service is consistent with the law of demand: increase in entry fee (price) reduces the WTP (demand for the services). Additional significant demand shifters, other than income and respondents’ characteristics, were the education and age of the household head. The results suggested that respondents from households with older and more educated household heads were more likely to visit Nakivubo Wetland Park and be willing to pay entry fees. Table  A8.4 Determinants of bid function Variable Parametric model KSS model Intercept 3.928*** -1.452*** (0.212) (0.378) Respondent’s age 0.00961 -0.00197 (0.00904) (0.00802) Respondent’s education 0.0129 -0.0954*** (0.0154) (0.0123) Respondent’s gender 0.102 0.293* (0.162) (0.155) Household head’s age -0.00473 0.00953* (0.00641) (0.00524) Household head’s education 0.0308* 0.0413*** (0.0166) (0.0132) Income class 0.0923* 0.0911** (0.0495) (0.0387) Bid -0.000128*** -0.000213*** (1.36e-05) (1.18e-05) Share -zero WTP 0.282 0.282 (13650) (13650) Log-likelihood -642.961 -533.648 LR test of Probit model against SNP model 97.47*** Number of observations 620 620                                                  Cost-benefit analysis of rehabilitation of the Nakivubo wetland, Kampala Page 101