Rapid Assessment of Natural Resource Degradation in Refugee Impacted Areas in Northern Uganda Technical Report June 2019 Updated based on April 2019 refugee statistics . © 2019 International Bank for Reconstruction and Development / The World Bank and The Food and Agriculture Organization of the United Nations The designations employed and the presentation of material in this information product do not imply the expression of any opinion whatsoever on the part of The World Bank or The Food and Agriculture Organization of the United Nations (FAO) concerning the legal or development status of any country, territory, city or area, or of its authorities. 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 or FAO concerning the endorsement or acceptance of such boundaries. The World Bank and FAO do not guarantee the accuracy of the data included in this work. The mention of specific companies or products of manufacturers, whether or not these have been patented, does not imply that these have been endorsed or recommended by The World Bank and FAO in preference to others of a similar nature that are not mentioned. The findings, interpretations and conclusions expressed in this information product are those of the author(s) and do not necessarily reflect the views or policies of The World Bank, World Bank Board of Executive directors, the FAO, or the governments and members represented by these institutions. Rights and Permissions The World Bank and FAO encourage the use, reproduction and dissemination of material in this information product. Except where otherwise indicated, material may be copied, downloaded and printed for private study, research and teaching purposes, or for use in non-commercial products or services, provided that appropriate acknowledgement of The World Bank and FAO as source and copyright holders is given. All queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Cover photo: © FAO / Rebecca Tavani CONTENTS ACKNOWLEDGMENTS ............................................................................................................................................. III ACRONYMS ............................................................................................................................................................. IV .............................................................................................................................................. V EXECUTIVE SUMMARY 1. INTRODUCTION ................................................................................................................................................ 1 1.1 BACKGROUND 1 1.2 OBJECTIVES OF THE ASSESSMENT 1 1.3 AREA OF INTEREST 2 2. SOCIOECONOMIC FINDINGS ............................................................................................................................. 4 2.1 REFUGEE AND HOST COMMUNITY POLITICAL FRAMEWORK 4 2.2 POPULATION AND HOUSEHOLD CHARACTERISTICS 5 2.3 WOODFUEL CONSUMPTION 6 2.4 ACCESS TO WOODFUEL 8 2.5 COOKING STOVES AND PRACTICES 9 3. WOODY BIOMASS RESOURCES FINDINGS ....................................................................................................... 13 3.1 BIOPHYSICAL FIELD MEASUREMENTS 13 3.2 LULC MAPPING AND CHANGE DETECTION 14 3.3 LINKING WOODFUEL DEMAND AND SUPPLY 24 4. RECOMMENDED TECHNICAL INTERVENTIONS ................................................................................................ 26 4.1 DEVELOPMENT OF AGROFORESTRY SYSTEMS 26 4.2 ESTABLISHMENT OF WOODLOTS FOR ENERGY AND OTHER PURPOSES 28 4.3 REHABILITATION OF DEGRADED FORESTS 31 4.4 ENHANCEMENT OF ENERGY EFFICIENCY 34 4.5 ADDITIONAL RECOMMENDED MEASURES 36 5. CONCLUSIONS ................................................................................................................................................ 39 REFERENCES ........................................................................................................................................................... 41 ANNEX 1: METHODOLOGIES ................................................................................................................................... 43 ANNEX 2: RAPID WOODFUEL DEMAND QUESTIONNAIRE ....................................................................................... 53 i FIGURES FIGURE 1. AOI: 5 AND 15 KM BUFFER ZONES AROUND REFUGEE SETTLEMENTS, NORTHERN UGANDA ......................................................... 3 FIGURE 2. A MAKESHIFT MARKET IN BIDIBIDI SETTLEMENT .................................................................................................................. 6 FIGURE 3. TRADITIONAL CHARCOAL KILN NEAR BIDIBIDI SETTLEMENT .................................................................................................... 7 FIGURE 4. FUELWOOD SOURCES FOR HOUSEHOLDS ........................................................................................................................... 8 FIGURE 5. TYPES OF HOUSEHOLD COOKSTOVES ................................................................................................................................. 9 FIGURE 6. TYPICAL OUTDOOR KITCHEN SETTING: A LORENA STOVE AND PILE OF FIREWOOD (REFUGEE HOUSEHOLD - BIDIBIDI) ........................ 10 2 FIGURE 7. LULC PER SETTLEMENT (AREA IN KM ) WITHIN THE 15 KM BUFFER ...................................................................................... 15 FIGURE 8. TREE COVER LOSS (IN HECTARES) USING 10 PERCENT TREE COVER THRESHOLD WITHIN 5 AND 15 KM BUFFERS (2001–2016) ......... 16 FIGURE 9. BIOMASS STOCK WITHIN THE SETTLEMENTS AND 15 KM OF SETTLEMENT BOUNDARIES, EARLY 2018 ........................................... 17 FIGURE 10. BIOMASS STOCK CHANGES BETWEEN 2013 AND 2018 WITHIN SETTLEMENTS AND 15 KM BUFFERS .......................................... 18 FIGURE 11. AREA 1: PLOT ALLOCATIONS ON 2015 LULC MAP ......................................................................................................... 44 FIGURE 12. AREA 2: PLOT ALLOCATIONS ON 2015 LULC MAP ......................................................................................................... 45 FIGURE 13. PLOT DESIGN FOR THE BIOPHYSICAL INVENTORY .............................................................................................................. 46 FIGURE 14. NUMBER OF SATELLITE IMAGES (LANDSAT 7 ETM+) OF THE TIME SERIES FOR BOTH PERIODS .................................................. 50 TABLES TABLE 1. SUMMARY OF THE INDICATIVE COSTS OF THE RECOMMENDED INTERVENTIONS .......................................................................... VII TABLE 2: REFUGEE SETTLEMENTS INCLUDED IN STUDY ........................................................................................................................ 2 TABLE 3. GENDER OF HOUSEHOLD HEAD ......................................................................................................................................... 5 TABLE 4. HOUSEHOLD INCOME ..................................................................................................................................................... 5 TABLE 5. REFUGEE AND HOST WOODFUEL CONSUMPTION, KG PPPD ..................................................................................................... 6 TABLE 6. ESTIMATED TOTAL WOODFUEL CONSUMPTION IN THE TARGET REFUGEE SETTLEMENTS ................................................................. 7 TABLE 7. BIOMASS STOCK BY LULC CATEGORY ............................................................................................................................... 13 TABLE 8. BIOMASS GROWTH (DRY MATTER) FOR SELECTED LULC CLASSES, AS NATIONAL AVERAGES AND FOR SEMI-MOIST LOWLANDS ............. 14 TABLE 9. LOSS AND DEGRADATION (HA) AND BIOMASS (AGB) CHANGES IN SELECTED LAND COVER CLASSES WITHIN 5 AND 15 KM OF THE REFUGEE SETTLEMENT BOUNDARIES ................................................................................................................................. 19 TABLE 10. DEGRADATION AND LOSS WITHIN THE REFUGEE SETTLEMENTS AND 5 KM BUFFER ZONE, WITH NET BIOMASS ESTIMATES (DM) FOR 2010–2013 AND 2014–2018 ..................................................................................................................................... 21 TABLE 11. SUMMARY OF DEGRADATION AND LOSS (IN HA) PER SETTLEMENT WITHIN 5 KM ..................................................................... 23 TABLE 12. ESTIMATED WOODFUEL DEMAND AND SUPPLY IN THE TARGET REFUGEE SETTLEMENTS AND WITHIN 5 KM BUFFER ZONE ................. 24 TABLE 13. SCENARIOS WITH REFUGEE POPULATION REDUCTIONS ....................................................................................................... 25 TABLE 14. INDICATIVE COSTS OF AGROFORESTRY INTERVENTION, PER HECTARE BASIS ............................................................................. 27 TABLE 15. INDICATIVE COSTS OF AGROFORESTRY INTERVENTION WITHIN REFUGEE SETTLEMENTS AND 5 KM BUFFERS ................................... 28 TABLE 16. INDICATIVE INVESTMENT AND OPERATIONAL COSTS OF ESTABLISHING WOODLOTS FOR ENERGY, PER HECTARE BASIS ....................... 29 TABLE 17. INDICATIVE COSTS TO SET UP A NURSERY ......................................................................................................................... 30 TABLE 18. WOODLOT REQUIREMENTS FOR ENERGY AND INDICATIVE ESTABLISHMENT AND MAINTENANCE COSTS OVER FIVE YEARS .................. 31 TABLE 19. INDICATIVE COSTS FOR NATURAL REHABILITATION OF DEGRADED AREAS, PER HECTARE BASIS OVER FIVE YEARS .............................. 32 TABLE 20. INDICATIVE COSTS FOR ASSISTED NATURAL REGENERATION, PER HECTARE BASIS OVER FIVE YEARS ............................................... 33 TABLE 21. INDICATIVE COSTS TO SET UP A NURSERY FOR ASSISTED NATURAL REHABILITATION, PER HECTARE BASIS ........................................ 33 TABLE 22. INDICATIVE COSTS OF REHABILITATION OF DEGRADED AND LOST WOODLAND AND BUSHLAND IN THE TARGET REFUGEE SETTLEMENTS ................................................................................................................................................................................ 34 TABLE 23. INDICATIVE COSTS FOR ENERGY EFFICIENCY ENHANCEMENTS ............................................................................................... 36 TABLE 24. INDICATIVE COST FOR THE PROVISION OF IMPROVED CHARCOAL KILNS AND TRAINING PACKAGES BY REFUGEE SETTLEMENTS ............. 36 TABLE 25. MATRIX OF RESULTED CLASSES OBTAINED BY COMBINING THE LULC MAP WITH THE MASK OF LOSS AND DEGRADATION ................. 49 ii Acknowledgments Between February and July 2018, the World Bank (WB) commissioned the Food and Agriculture Organization of the United Nations (FAO)1 to undertake a ‘Rapid Diagnostic Assessment of Land and Natural Resources Degradation in Areas Impacted by the South Sudan Refugee Influx in Uganda’. This work was undertaken in close collaboration with the Ugandan Ministry of Water and Environment (MWE). The aim of the assessment was to determine the environmental impacts of the refugee influx, with a focus on forest resources, and propose appropriate intervention options to mitigate pressure on the environment and support energy access to the refugee and host communities. A Technical Report summarizing the findings and recommendations of the assessment was published in October 2018. This updated version was produced in June 2019 to reflect a downwards revision in the refugee population that took place after a countrywide validation exercise. It supersedes the previous report. The original report and 2019 update were prepared jointly by the World Bank task team comprised of Ross Hughes (Task Team Leader), Matthew Owen (Senior Consultant), Lesya Verheijen (Senior Operations Officer), Christine Kasedde (Environmental Specialist), and Herbert Oule (Senior Environmental Specialist) and following authors at FAO (listed alphabetically): John Begumana, Laura D’Aietti, Arturo Gianvenuti, Inge Jonckheere, Eva Kintu, Erik Lindquist, Rebecca Tavani, and Zuzhang Xia. The authors would like to thank the staff of the WB, FAO and the United Nations High Commissioner for Refugees (UNHCR) in Kampala, the FAO Resilience Team for East Africa, and the National Forestry Authority of Uganda for their support and inputs, as well as the United Nations Institute for Training and Research (UNITAR) headquarters in Geneva for providing very high-resolution satellite imagery. The authors also wish to acknowledge and thank those in the West Nile region for their invaluable assistance with the field work: the FAO Office in Yumbe, the UNHCR Sub-Offices in Arua and Yumbe, the Office of the Prime Minister, and the District Local Governments of Adjumani, Arua, Moyo, and Yumbe. The authors would also like to extend special thanks to the following colleagues who contributed to this report with comments, views, and information: Shukri Ahmed, Robert Basil, Nelly Bedijo, Cecilia Bonacchi, Alexa Caesar, Benjamin Caldwell, Cyril Ferrand, Priya Gujadhur, Leonidas Hitimana, Orjan Jonsson, Koen Joosten, Lauri Vesa, Jacqueline Were, and Sheila Wertz (all FAO), Ranya Sherif (UNHCR), Charles Ariani, Julius Ariho, John Diisi, Levi Etwodu, Stephen Galima, Maxwell Kabi, Xavier Mugumya, Robert Otuko, and Edward Ssenyonjo (NFA). Preparation of the original report was financed by the TerrAfrica Leveraging Fund. The update was financed by the The State and Peacebuilding Fund (SPF). SPF is a global fund to finance critical development operations and analysis in situations of fragility, conflict, and violence. The SPF is kindly supported by: Australia, Denmark, Germany, The Netherlands, Norway, Sweden, Switzerland, The United Kingdom, as well as International Bank for Reconstruction and Development. 1 WB Contractual Agreement no. 7185743; FAO Project Symbol: OSRO/GLO/801/WBK. iii Acronyms AGB Above-ground Biomass AOI Area of Interest BFAST Breaks for Additive Season and Trend CRRF Comprehensive Refugee Response Framework DBH Diameter at Breast Height DEM Digital Elevation Model DM Dry Matter DRDIP Development Response to Displacement Impacts Project FAO Food and Agriculture Organization of the United Nations GFC Global Forest Change GoU Government of Uganda HH household IBEK Improved Basic Earth Kiln IDA International Development Association (World Bank Group) IPCC Intergovernmental Panel on Climate Change LULC Land Use and Land Cover MWE Ministry of Water and Environment NBS National Biomass Study NDMI Normalized Difference Moisture Index NFA National Forestry Authority NGO Non-governmental Organization NDP II Second National Development Plan OPM Office of the Prime Minister pppd per person per day ReHoPE Refugee and Host Population Empowerment RRP Refugee Response Plan SEPAL System for Earth Observations, Data Access, Processing and Analysis for Land Monitoring SPGS III Sawlog Production Grant Scheme, Phase III SRTM Shuttle Radar Topography Mission STA Settlement Transformation Agenda THF Tropical High Forest UNHCR United Nations High Commissioner for Refugees UBOS Uganda Bureau of Statistics UN United Nations UNDP United Nations Development Program UNITAR United Nations Institute for Training and Research WB World Bank WG Working Group iv Executive summary The ongoing refugee crisis in South Sudan has led to the establishment of some of the world’s largest refugee settlements over the border in northern Uganda. By April 2019, over 815,000 South Sudanese refugees and asylum seekers had migrated to Uganda. Uganda is also hosting refugees from Burundi, the Democratic Republic of the Congo, and Somalia, making it the largest refugee host country in Africa (and second in the world), with a total of 1.2 million refugees and asylum-seekers. The influx of refugees was reported to have exacerbated a range of ongoing environmental impacts and associated challenges, including land degradation and woodland loss, resulting in inadequate access to energy for cooking and competition with local people for water and other natural resources. Supporting more sustainable use of those resources, especially forests and other woodlands, could help address environmental degradation and improve energy access. The World Bank (WB) commissioned the Food and Agriculture Organization of the United Nations (FAO) to undertake a rapid assessment of natural resource degradation around the refugee settlements in northern Uganda, with a focus on forest resources, and to identify possible interventions to mitigate pressure on the environment and support energy access for both the refugee and host communities. This report summarizes the main findings and recommendations of the assessment, updated to reflect the most recent (April 2019) refugee population figures. These are expected to guide WB support to the Government of Uganda (GoU)—including the Development Response to Displacement Impacts Project (DRDIP) and an IDA disbursement window for refugee-affected countries—as well as provide information of wider strategic value to other agencies concerned with the impacts of refugees on natural resources in Uganda. A similar analysis is being undertaken for the refugee settlements in west and south-west Uganda and will result in a second assessment report that will add to the evidence base for the WB/GoU interventions. Main findings The assessment revealed the following key findings: • The refugee influx from South Sudan has led to an increase in the rate of degradation and tree loss, both inside the West Nile refugee settlements and around their boundaries, with accelerated land cover changes in bushland and woodland. Deforestation and forest degradation are not new phenomena in Uganda, and the refugee presence has added to existing pressures on the environment, due to increased demand for wood as cooking fuel. Competition for available resources could become a source of tension between the refugees and host communities. • Land cover change analysis shows an increase in tree cover loss and degradation both within and around the refugee settlements after the start of the refugee influx from South Sudan. Within a 5 km buffer zone from the settlement boundaries, total tree cover loss between 2010 and 2013 was 1,919 ha, while degradation covered 5,664 ha (in woodland and bushland, including the areas of the settlements themselves). Meanwhile from 2014 to 2018, there was 34,112 ha of loss and 29,604 ha of degradation. Between the two periods, there was an average increase of around 14 percent of degradation and loss in woodland, bushland and cropland within 5 km of the settlement boundaries, and additional loss and degradation in an extended 15 km buffer - though the latter more likely reflects ongoing degradation by host communities rather than refugee-related impacts. • Refugee and host households are highly dependent on forests and other woodlands as sources of woodfuel for cooking and for income generation, contributing to their livelihood resilience. Average daily consumption of firewood by the refugees is 1.6 kg per person and among host communities is 2.1 kg, about 30 percent higher. Taking into account the additional use of charcoal, average daily fuel consumption rises to 1.8 kg per person in firewood equivalent among refugees and 2.2 kg among host community households. • Total cooking fuel demand in the 14 targeted refugee settlements is about 345,000 metric tons of wood per year (on dry weight basis), based on the April 2019 refugee population. This is about four v times the quantity of tree growth within the settlements and the 5 km buffer zone, which could result in an annual biomass loss of about 8 percent. Only around the Maaji (I and II) settlements in Adjumani District is there an apparent surplus of woodfuel in excess of demand within the 5 km buffer zone. Based on the woodfuel demand and supply assessment, the refugee settlements with greatest pressure on the surrounding forests and other woodlands are Pagirinya, Nyumanzi, Imvepi, Palorinya, Bidibidi, and Ayilo (I). • Refugee woodfuel consumption at Bidibidi settlement has significantly reduced, to about half the amount recorded in a March 2017 survey, probably due to greater wood shortage, a more diverse diet with fresher food, drier firewood, and more efficient stoves and cooking practices. • Both refugees and locals have a tradition of building improved mud-stoves from locally available materials. A higher proportion of refugee households use such improved stoves than host communities, and in Bidibidi there has been a marked increase in their adoption since the March 2017 survey. Modern prefabricated cookstoves are also available in regional markets but are too expensive for most refugees and locals. Improved mud-stoves are likely to remain a practical cooking solution and are well-known and culturally acceptable. There would be value in confirming thermal efficiency, pollutant emissions, and safety of the adopted mud-stoves to identify areas of possible improvement. • Households need additional wood to build and maintain living structures. A majority of households have constructed semi-permanent structures and have improved their homes with latrines and kitchen shelters. A few have bathing shelters, animal sheds, and poultry/bird pens. The quantity of wood used was not measured under this study. • Although natural resource depletion is a concern for GoU and partners, there are few organizations working on environment and energy-related activities in in refugee-affected areas. Those organizations that do so generally operate at a small scale on 12-month budget cycles. To ensure a more effective and harmonized approach with appropriate technical expertise and adequate resourcing, there is a need for a coordinated package of interventions implemented on a multi-year basis through a multi-agency program. This would effectively address environmental degradation associated with the presence of the refugees and ongoing local drivers. The assessment recommends a range of costed interventions and additional measures to improve environmental management, ensure access to woodfuel resources for both refugee and host communities, and contribute to building livelihood resilience: 1) Development of agroforestry systems on household plots and farmland, where trees and woody perennials are interplanted along boundaries and with crops for energy, food, and fodder. This intervention should target the residential plots assigned to refugees and the cultivated fields of both host and refugee communities surrounding refugee settlements. 2) Establishment of woodlots for energy and other purposes such as building poles, fruits, and fodder. This intervention should target areas owned by host communities and individuals, protected areas managed by the National Forestry Authority (NFA), and areas assigned for refugee settlements. 3) Rehabilitation of degraded forests using both natural and assisted regeneration. This intervention should target areas owned by host communities and individuals, protected areas managed by the NFA, and areas assigned to the refugees. 4) Enhancement of energy efficiency, to reduce demand for woodfuel through more efficient cooking practices and charcoal production techniques. This intervention should target both host and refugee populations. Table 1 gives estimated costs for the implementation of the proposed interventions for 14 refugee settlements in northern Uganda. vi Table 1. Summary of the indicative costs of the recommended interventions % of Recommended intervention Cost (US$) total Development of agroforestry systems 61,513,000 60.1 Establishment of woodlots for energy and other purposes 20,171,000 19.7 Rehabilitation of degraded forests 15,007,000 14.7 Enhancement of energy efficiency 5,656,000 5.5 Total 102,347,000 100 The recommended interventions should be coordinated under an integrated energy and environment program with sufficient institutional capacity and resources to undertake more in-depth analysis at the site level; carry out monitoring and evaluation; support systematic efforts to promote the interventions across the associated host communities; and ensure sound learning, sharing, and interaction with other programs of a similar nature in Uganda and elsewhere. This will ensure that the measures do not take place in isolation or in a scattered or short-term manner. Such an integrated energy and environment program could complement the community-driven approaches adopted under the DRDIP. vii 1. Introduction 1.1 Background The refugee crisis in South Sudan has led to the establishment of some of the world’s largest refugee settlements in northern Uganda. By April 2019, over 815,000 South Sudanese refugees and asylum-seekers had migrated to Uganda.2 Uganda is also hosting refugees from Burundi, the Democratic Republic of Congo, and Somalia, making it the largest refugee host country in Africa (and second in the world), with a total of 1.4 million refugees and asylum seekers. Forest resources play a key role in supporting livelihoods in Uganda, providing the country’s main source of domestic energy for cooking through firewood and charcoal, and supporting significant biodiversity for increasing resilience, adaptation and strengthening the provision of essential ecosystem services. Woodfuels3 are the primary source of energy for more than 90 percent of households in Uganda (UBOS, 2018) and an even higher proportion of refugees. The energy needs of a large number of refugees are increasingly difficult to meet in a situation of declining tree cover and agricultural expansion, and extraction of wood for fuel may contribute to degradation of soils, forests, and woodlands. The refugee influx has reportedly had a range of environmental impacts and associated challenges, including land degradation and woodland loss, resulting in inadequate access to energy for cooking and competition for natural resources. Insufficient arable land continues to impair the ability of refugees to grow their own food (UNHCR 2018), despite the allocation of plots for agricultural and residential purposes ranging in size from 30 x 30 m to 100 x 100 m per household. Given the large number of people who have crossed into Uganda, there is a pressing need to develop strategies for sustainable energy access and forest resource management targeting both refugees and hosts. A joint assessment conducted by FAO and the United Nations High Commissioner for Refugees (UNHCR) in one settlement (Bidibidi) in March 20174 concluded that the aboveground biomass (AGB) stock within the settlement area could meet the needs of the population for only three years, in the absence of any intervention. Measures were proposed to reduce demand for wood (for example, fuel-efficient stoves) and increase supply (for example, woodlots and multipurpose tree planting), to build resilience and to create opportunities for sustainable development. Uganda is benefiting from a new IDA18 sub-window for refugees and host communities.5 The country’s progressive refugee policies enhance its prospects for support under this window. Uganda is also benefitting from ongoing support to refugee-hosting areas under an ongoing IDA investment project—the Development Response to Displacement Impacts Project (DRDIP, P152822). The WB commissioned FAO to undertake a ‘Rapid Diagnostic Assessment of Land and Natural Resources Degradation in Areas Impacted by South Sudan Refugee Influx in Kenya and Uganda’.6 The assessment was expected to provide a clear profile of the scope of the environmental impacts of the refugee influx, with a focus on forest resources, management challenges, assessment of possible intervention strategies, and practical proposals for interventions for potential inclusion in financing packages submitted to the IDA18 sub-window for refugees, and to inform ongoing WB support under the DRDIP. 1.2 Objectives of the assessment The purpose of the assessment was to conduct a rapid diagnostic assessment of land and forest resources degradation around the 14 refugee settlements in northern Uganda to identify potential intervention options to mitigate pressure on the environment, ensure access to energy for cooking, and contribute to building the resilience of displaced and host communities. 2 Source - UNHCR, Government of Uganda, Office of the Prime Minister. https://ugandarefugees.org/en/country/uga 3 In FAO’s terminology, ‘woodfuels’ are a category of biofuels where the original composition of the wood is preserved. For this study, only firewood and charcoal are considered. ‘Firewood’ is synonymous with ‘fuelwood’. 4 FAO and UNHCR. 2017. http://www.fao.org/3/a-i7849e.pdf. 5 The refugee sub-window was created under the 18th replenishment of IDA. 6 WB Contractual Agreement no. 7185743; FAO Project Symbol: OSRO/GLO/801/WBK. 1 The study involved a combination of a desk review, field survey, and remote sensing analysis. The field survey comprised a socioeconomic assessment of woodfuel consumption and associated challenges in two refugee settlements and in selected villages in the local area, as well as a study of biophysical parameters of woodlands and bushlands in preselected hotspots in Adjumani, Arua, Moyo, and Yumbe Districts. The assessment builds on the methodology developed in the joint FAO-UNHCR technical handbook, Assessing Woodfuel Supply and Demand in Displacement Settings (FAO & UNHCR, 2016).7 The methodology comprised three components: (1) assessment of woodfuel demand and associated challenges; (2) assessment of woodfuel supply, including AGB stock, land cover classification, and changes; and (3) identification of interventions to address issues related to energy access, natural resource degradation, and livelihoods. The methodology for the socioeconomic analysis, biophysical field inventory, and remote sensing analysis is described in detail in the Annex 1. 1.3 Area of interest Table 2 lists the 14 refugee settlements in northern Uganda, with districts and establishment dates. Table 2: Refugee settlements included in study No. Settlement name District Establishment date 1 Bidibidi Yumbe August 2016 2 Imvepi Arua February 2017 3 Rhino extension - Omugo Arua January 2017 4 Agojo Adjumani January 2016 5 Ayilo I Adjumani January 2015 6 Ayilo II Adjumani July 2014 7 Boroli I/II Adjumani January 2014 a 8 Maaji I Adjumani January 1997 a 9 Maaji II Adjumani January 1997 a 10 Maaji III Adjumani January 1997 11 Nyumanzi Adjumani January 2014 12 Pagirinya Adjumani January 2016 13 Palorinya Moyo December 2016 14 Palabek Lamwo April 2017 Note: a. Settlements established in 1997 and reopened in 2015. The area of interest (AOI) for this assessment was the ‘buffer zone’8 up to 5 km of the boundaries of the 14 refugee settlements, this being the assumed limit for routine firewood collection. A wider AOI up to 15 km away was also assessed to understand trends and dynamics within host communities (Figure 1). 7 http://www.fao.org/3/a-i5762e.pdf. 8 https://www.supermap.com/en/online/deskprodotnet/Features/Analyst/Vector/bufferanalyst/HowBufferWork.htm. 2 Figure 1. AOI: 5 and 15 km buffer zones around refugee settlements, northern Uganda Source: UNHCR - Settlements extents, administrative data. Note: The boundary of Maaji I settlement was not available, so the settlement center point was used to create the buffer. The boundaries, names and designations on this map do not imply official endorsement or acceptance by the United Nations. 3 2. Socioeconomic findings 2.1 Refugee and host community political framework Coordination of the refugee protection and response system in Uganda is led by the Office of the Prime Minister (OPM), while operational response is co-led by the OPM and UNHCR, supported by UN agencies and partners. Uganda’s policy toward refugees is unique in Africa. In accordance with the Refugee Act (2006) and Refugee Regulations (2010), the Government of Uganda (GoU) has developed national frameworks with an inclusive approach, granting refugees freedom of movement and the right to work, establish business, and access public services such as education, on a par with nationals. The Second National Development Plan (NDP II) (2015/16– 2019/20) provides for refugee management and protection as a priority in development planning and implementation by the OPM of the Settlement Transformation Agenda (STA) to promote socioeconomic development in refugee-hosting areas. The allocation of plots of land where refugees can live and farm is a practice that has significant implications for the planning of community-based environmental interventions and for intervening to address environmental degradation. Host districts are required to develop Integrated District Development Plans that incorporate the development needs of host communities and refugees. Refugee and Host Population Empowerment (ReHoPE) is a policy framework launched in 2017 by the GoU in collaboration with UN agencies and the WB. ReHoPE seeks to foster a multiyear, multisectoral program to bridge humanitarian and development approaches. It provides guidance for a comprehensive response to address refugees’ and host communities’ needs and to build the capacity of hosting districts in planning and providing services to refugee and host communities. ReHoPE supports the GoU to address environmental degradation in refugee-hosting areas through improved natural resource management and energy access. Uganda’s Comprehensive Refugee Response Framework (CRRF) was launched by the OPM and UNHCR in March 2017. It has five pillars: 1. admission and rights, 2. emergency response and ongoing needs, 3. resilience and self-reliance of refugees, 4. expansion of solutions through resettlement and complementary pathways, and 5. voluntary repatriation. In October 2017, a high-level, government-led Steering Group, facilitated by UNHCR, was established to bring together humanitarian and development actors, local government, and the private sector, to engage and provide guidance on refugee matters. The CRRF Steering Group also documents lessons from the Uganda refugee experience to inform relevant global, regional, and national initiatives, as well as the development of the Global Compact on Refugees. The Steering Group has established a secretariat to support the application of the CRRF. The secretariat serves as a knowledge hub and platform for strategic discussions, building on initiatives already in place to manage and find solutions for refugees. The Working Group on Energy and Environment (‘WorkGrEEn’) operates under the umbrella of the CRRF to coordinate the country-wide energy and environment response for ReHoPE, in line with NDP II, the STA, and the Uganda Refugee Response Plans (RRPs).9 The Working Group is co-chaired by the OPM, UNHCR and the United Nations Development Program (UNDP), and its mandate is anchored in existing strategies including ReHoPE (Objective 4), the STA (Pillars 1, 2, 4, and 5), and the RRP (Strategic Priority 6, Objective 3). The Working Group has been constituted specifically to support refugee-affected districts and has a representation of relevant actors in the environment sector (government, non-governmental organizations [NGOs], UN agencies). The Working Group is leading on Objective 4 of ReHoPE and will validate and enhance the results and indicators in the next revision of the ReHoPE strategy. 9 https://ugandarefugees.org/wp-content/uploads/Uganda-I-RRP-2018pdf.pdf. 4 2.2 Population and household characteristics Household size and gender The socio-economic survey covered 174 households (HHs) in the refugee settlements of Bidibidi (Yumbe District) and Maaji (Adjumani District), as well as 168 host community households in Ciforo (Adjumani) and Okangali (Yumbe) sub-counties. The majority of respondents were female in both the refugee and host communities, where they constituted 91 percent and 83 percent of respondents, respectively. The average refugee household was found to be larger than the average host household (7.9 versus 6.4 persons). Table 3 shows the gender of household heads. Among the refugee respondents, 75.9 percent and 62.1 percent were female-headed in Maaji (Adjumani District) and Bidibidi (Yumbe District), respectively, while in the host communities in Ciforo and Okangali, only 32.5 percent and 24.7 percent, respectively, were female-headed. These figures are aligned with UNHCR’s socioeconomic assessment report 2017, which indicates that 63.8 percent of refugee households in Uganda are female headed, but only 30.5 percent of host community households are female headed (UNHCR 2017a). Table 3. Gender of household head Refugee communities Host communities Female (%) Male (%) Female (%) Male (%) Adjumani District 75.9 24.1 32.5 67.5 Yumbe District 62.1 37.9 24.7 75.3 Livelihoods The majority of refugee and host households engage in agriculture-based livelihoods, usually subsistence farming. A small proportion of refugee households have other income (for example, cash transfers, brewing, selling woodfuel, tailoring, teaching, transporting items, selling cooking oil, blacksmithing, selling dried fish or casual work in local food outlets). Host community households are also engaged in other income-earning activities such as casual labor, selling woodfuel and non-wood forest products, exchanging food, and cooking and selling food. Table 4 shows the proportions of refugee and host community households with and without a source of income. Table 4. Household income Refugee communities Host communities HH with no HH with HH with no HH with income (%) income (%) income (%) income (%) Adjumani District 26.4 73.6 14.1 85.9 Yumbe District 18.4 81.6 4.2 95.8 As expected, there are more households in refugee communities without an income. It is also important to note that among refugee settlements, 30 percent of those households with income-earners had more than one person earning an income. Since the assessment conducted by FAO and UNHCR in Bidibidi in March 2017, the proportion of refugee households with members earning an income has risen from 26 percent to 81.6 percent. This could be a sign that the population is transitioning from an emergency situation to a more stable way of life. It was observed that the land allocated to refugees in the Maaji settlements (Adjumani District) has greater arable potential than that in Bidibidi settlement (Yumbe District). Refugees in Maaji were more likely to be engaged in commercial scale farming on their plots and renting additional land to grow more crops. In addition to the good soils, this finding also reflects the establishment of the Maaji settlements dating back to 1997. 5 Figure 2. A makeshift market in Bidibidi settlement ©FAO/Eva Kintu 2.3 Woodfuel consumption An average of 97 percent of households across the refugee and host communities use firewood for cooking (Table 5). Refugee households are more likely to use charcoal than host communities (16.7 percent versus 6 percent) and a few of them use both charcoal and firewood. A number of households also burn crop residues such as cassava stalks and maize cobs and stalks (when available). Per capita woodfuel consumption The household survey reveals that the average firewood consumption of a refugee household in both districts is a little lower than that of a host household (Table 5). The figures provided in Table 5 are average of woodfuel consumption expressed as kilogram per person per day (kg pppd) by the users. Table 5. Refugee and host woodfuel consumption, kg pppd Charcoal Firewood Population using Population using consumption consumption firewood (%) charcoal (%) (kg pppd wood (kg pppd) a equivalent) Refugees - Adjumani 94.3 1.73 25.3 1.25 Refugees - Yumbe 98.9 1.57 8.0 1.40 Refugees - total 96.6 1.65 16.7 1.30 Hosts - Adjumani 98.8 2.14 7.2 1.35 Hosts - Yumbe 96.5 2.13 4.7 1.25 Hosts - total 97.6 2.13 6.0 1.30 Note: a. Expressed in firewood equivalent, assuming 20 percent conversion of firewood to charcoal by weight. Kilogram of firewood pppd is expressed on an air-dry basis. Notably, the daily firewood consumption of refugee households in Bidibidi settlement has declined significantly from 3.5 to 1.6 kg pppd since March 2017. A possible reason is a move from a diet dominated by dry beans to a more diverse diet with more fresh food that cooks faster. It was also observed that refugees were using drier wood in 2018 than in 2017, when green wood was often collected and burned. A slight increase in charcoal consumption and a greater use of improved stoves were also observed in 2018. 6 A few refugee households use firewood for commercial purposes (Maaji: 2.4 percent; Bidibidi: 1.4 percent). This is more common in host communities, for commercial activities such as charcoal production, brewing, alcohol distillation, tobacco curing, and brick making. Figure 3. Traditional charcoal kiln near Bidibidi settlement ©FAO/Rebecca Tavani Total refugee woodfuel consumption Table 6 indicates the total woodfuel consumption for all refugee settlements in northern Uganda. The figures for each settlement are based on weighted averages extrapolated from the proportions of woodfuel users (Table 5) drawn from the household surveys conducted in Bidibidi and Maaji. Total woodfuel consumption takes into account both firewood (expressed on an air-dry basis) and charcoal (expressed in firewood equivalent, assuming a conversion efficiency of 20 percent). The April 2019 population data suggest total woodfuel consumption of 421,019 metric tons per year (t/yr) in firewood equivalent. Table 6. Estimated total woodfuel consumption in the target refugee settlements Population Total woodfuel consumption Settlement (April 2019) (t/yr firewood equivalent) Bidibidi 225,808 149,262 Imvepi 57,758 38,178 Rhino extension - Omugo 24,533 16,217 Agojo 6,661 4,403 Ayilo I 23,837 15,757 Ayilo II 13,722 9,070 Boroli I/II 14,841 9,810 Maaji I 518 342 Maaji II 16,174 10,691 Maaji III 14,947 9,880 Nyumanzi 39,505 26,113 Pagirinya 35,803 23,666 Palorinya 119,587 79,049 Palabek 43,238 28,581 Total 636,932 421,019 7 Woodfuel source The dominant source of firewood for both refugee and host community households is bushland, followed by woodland. Host communities also source wood from cropland (Figure 4). Figure 4. Fuelwood sources for households 2.4 Access to woodfuel Around 60 percent of both refugee and host households (refugee households - Adjumani: 59 percent; Yumbe: 73 percent; host households - Adjumani: 59 percent; Yumbe: 65 percent) collect more than three headloads of firewood per week. In refugee households, 84 percent of respondents spend two or more hours per trip collecting firewood (Adjumani: 78 percent; Yumbe: 89 percent), while in host communities, about 69 percent spend two or more hours, with a higher proportion in Adjumani (82 percent) than Yumbe (56.7 percent). The most commonly mentioned challenge for refugees in firewood collection is its scarcity, which results in women walking long distances (exposing themselves to more risks and challenges). Some refugee respondents reported a fear of being attacked/beaten by host communities or of encountering wild animals during firewood collection. Other challenges mentioned include fear of arrest by forest guards, assault/rape, inadequate or insufficient tools for collecting firewood, and flooded streams leading to inaccessibility in the rainy season. Some refugees also reported issues indirectly related to firewood collection, such as children missing school, lack of food or cash to exchange for firewood, lack of transport, and a language barrier impeding communication with the host community. Host communities mentioned similar challenges in firewood collection, with the most common being its scarcity and fear of encountering hazards such as snakes and scorpions. Other challenges mentioned include lack of tools, rain interference, conflicts and tensions with landlords, and lack of transport. 8 2.5 Cooking stoves and practices The majority of refugee and local households have constructed improved cookstoves, with hosts in Adjumani having the highest proportion and those in Yumbe having the lowest. The most common improved cookstove used by refugee households is the mud-stove for firewood, sometimes represented by the Lorena stove (with two pot holes, one fireplace, and a chimney or smoke vent). It is common practice for households that have the mud-stove for firewood to construct another mud-stove for charcoal, as well as to use a 3-stone fire—a trend for multiple hearths previously noted in the 2017 FAO-UNHCR assessment in Bidibidi. The survey found that among refugee communities, 62.1 percent use an improved mud-stove with firewood and 23 percent with charcoal (Figure 5). A large proportion of refugee households (45 percent) use the 3-stone fire (38 percent in Maaji and 53 percent in Bidibidi), sometimes in combination with an improved stove (16 percent in Maaji and 33 percent in Bidibidi). In the host communities, 52 percent use a mud-stove with firewood and 6 percent with charcoal. The 3-stone fire is used by an average of 52 percent of host community respondents in Adjumani and Yumbe (respectively 19 percent and 85 percent), of which 11 percent (Adjumani) and 32 percent (Yumbe) use it in combination with an improved stove. Figure 5. Types of household cookstoves 9 Figure 6. Typical outdoor kitchen setting: a Lorena stove and pile of firewood (refugee household - Bidibidi) ©UNHCR/Ranya Sherif Cookstoves are set up both in dedicated indoor kitchens and in outdoor settings. Indoor cooking is more common during the rainy season, while outdoor cooking is cooler during the hot season. Indoor kitchens are often poorly ventilated so cooking outdoors also reduces smoke inhalation. Typical 2-pot mud-stove (host kitchen - Adjumani) Typical 1-pot mud-stove (refugee kitchen - Bidibidi) 10 Type of mud-stove constructed in the Mud-stove used alongside 3-stone veranda (host household - Yumbe) fire (refugee household - Maaji) Young woman lighting wire-mesh-like Woman heating water in indoor kitchen charcoal stove (Bidibidi) ©FAO/Eva Kintu (all 6 photos) Source of cookstoves Over 91 percent of refugee households with improved cookstoves constructed or sourced the stoves themselves (Maaji: 97 percent; Bidibidi: 86 percent). A smaller proportion said they received them from an NGO (18 percent—Maaji: 7.8 percent; Bidibidi: 27 percent) while some were supported by relatives (1.5 percent—Maaji: 3.1 percent: Bidibidi: 0 percent) and a small number purchased from the market (0.8 percent— Maaji: 0 percent; Bidibidi: 1.5 percent). Diet and food preparation Although the refugee diet in Bidibidi is more varied than in 2017, maize and beans are still the dominant food, with beans especially requiring a long cooking time. The household survey showed an average of 69 percent (Maaji: 64 percent; Bidibidi: 74 percent) of households in the refugee communities cook beans on five or more days per week, compared to 42 percent (Adjumani: 27 percent; Yumbe: 57 percent) in the host communities that cook beans five or more days in a week. This is understandable considering that households in host communities are able to grow a range of crops, whereas refugee households have land constraints. 11 It was observed that both the refugee and host households often place the beans in water, add ash solution to soften them, and reduce time for cooking by pre-boiling them for about 15 minutes, skinning them through a grinding action, and putting them back on the fire to cook. There is therefore good evidence of energy-saving cooking practices being applied. Another food prepared by most refugee households (84 percent—Maaji: 82 percent; Bidibidi: 86 percent) and host households (79 percent—Adjumani: 78 percent; Yumbe: 80 percent) on a daily basis is ugali, a dough made from maize, sorghum, or cassava flour that is boiled in water for 15–20 minutes and mixed (’mingled’) until firm. The survey also sought to establish the most common foods prepared by the interviewed households, and a more varied diet for refugee communities in Bidibidi was observed compared to the 2017 survey. The refugees now prepare vegetables, fish, and other food such as fresh roots (cassava, sweet potatoes, yams), groundnut or sesame paste, varieties of peas, and other foods cooked in different proportions: chapati, chicken, cassava leaves, rice, eggs, milk, and soya. 12 3. Woody biomass resources findings 3.1 Biophysical field measurements Table 7 illustrates the main results of the biophysical field assessment and gives an indication of the potential woody biomass available in each land use and land cover (LULC) category. Only 67 sample plots were surveyed out of the planned 95, due to problems in accessing the intact areas. Table 7. Biomass stock by LULC category Deadwood LULC (main land use) Source No. of plots No. of trees per ha AGB (t per ha) (t per ha) Intact woodland NBS 15 567 ± 103 38.0 ± 7.0 No data Intact bushland NBS 10 708 ± 257 27.8 ± 5.0 No data Cropland This survey 21 391 ± 270 9.14 ± 5.23 4.61 ± 7.2 Degraded woodland This survey 7 880 ± 923 25.3 ± 18.5 0.30 ± 0.25 Degraded bushland This survey 14 120 ± 76 3.94 ± 3.95 0.25 ± 0.24 Note: NBS = National Biomass Study, Forest Department, Uganda. A total of 70 tree species were recorded in the field, of which Acacia hockii, Combretum collinum, Combretum fragrans, and Lannea fruticosa were dominant. As the table shows, AGB in degraded bushlands is approximately 4 t per ha and was derived by analyzing data in the grassland category, as there were no field observations for the strata. Woodland plots with the greatest indicators of degradation were analyzed and yielded total AGB of 25.3 t per ha. This estimate, however, has the greatest uncertainty, with a confidence interval of ±18.5 t due to the wide variability found in this class. Woody biomass in cropland is estimated at 9.1 t per ha, which is approximately the national average. The category includes plots that were measured in areas described as ‘young fallow’, which are commonly found where crop cultivation has recently advanced into woodlands. Analysis of intact woodland sites using the NBS dataset for 15 plots resulted in an average AGB of about 38 t per ha. Analysis of intact bushland yielded 10 plots containing an average AGB of approximately 28 t per ha. The NBS provides estimated growth rates as national averages, and for agroecological zones (Forest Department 2002), the target settlements are all in the semi-moist lowland zone (Table 8). 13 Table 8. Biomass growth (dry matter) for selected LULC classes, as national averages and for semi-moist lowlands National biomass Semi-moist lowland biomass growth LULC class growth (t per ha) (t per ha) Woodland 4.265 3.583 Bushland 0.853 0.256 Grassland 0.853 1.024 Subsistence farmland 0.853 1.450 Source: Forest Department, Uganda, 2002. Note: A conversion factor of 0.853 was applied to convert woody biomass from air-dry matter to dry matter which corresponds to moisture content of 14.7% (NBS). 3.2 LULC mapping and change detection The aim of the remote sensing analysis was to map degradation and loss before and after the establishment of the refugee settlements, as a means of estimating land cover and biomass changes over time for the AOI, and to validate the consumption data generated by the surveys within the settlements and host communities. The LULC map is part of Uganda’s national mapping system and was used in this study to gain a better understanding of the dominant LULC classes. As shown in Figure 7, the AOI is characterized by relatively homogeneous distribution of the main land cover types (bushland, grassland, and subsistence farmland). The Maaji settlements in Adjumani District seem to be the richest in vegetation, particularly tree cover. Details on the methodology and datasets used in the remote sensing analysis are provided in the Annex 1. 14 Figure 7. LULC per settlement (area in km2) within the 15 km buffer Source: NFA maps 2015. Note: NFA = National Forestry Authority. 15 Slopes within the AOI were computed using a digital elevation model (DEM) (RCMRD 201510) and show a range from 0.5 percent to 18 percent. In general, the area is flat or gently undulating, and steepness is not likely to be a factor constraining access by either refugees or local people. Steeper slopes characterize areas near Moyo along the wester side of the River Nile and onthe southwest of Ayilo II settlement. The Global Forest Change (GFC) dataset (Hansen et al. 2013) was used to compute tree cover loss from 2001 to 2016. Figure 8 shows the loss detected within both the 5 and 15 km buffer zones. Figure 8. Tree cover loss (in hectares) using 10 percent tree cover threshold within 5 and 15 km buffers (2001– 2016) Source: GFC data. Tree cover loss shows one peak (in 2014) for the 5 km buffer and two peaks (in 2011 and 2014) for the 15 km buffer. The 2011 peak was three times higher for the 15 km buffer than the 5 km buffer, although possible causes were not investigated. The 2014 peak could be linked to the establishment of some refugee settlements and the GFC dataset may have detected clear-cuts covering extensive areas. Considering only 2014–2016, the GFC dataset does not show a significant increase in tree cover loss that might be associated with the refugees’ arrival. This could be partly explained by the challenges in detecting changes in complex landscapes (Mitchard et al. 2015, Hansen et al. 2013, Tyukavina et al. 2015). However, the map presenting biomass changes between 2013 and 2018 (Figure 10) shows a reduction in biomass stocks across the whole area, especially northern Bidibidi and around Ayilo and Palabek. Details are provided in Table 9 and Table 10. Degradation and loss within the settlements and the 5 and 15 km buffer zones was mapped by combining existing LULC maps (2010 and 2015) and clipping to the AOI with a ‘degradation/loss mask’ obtained from the Breaks for Additive Seasonal and Trend (BFAST, 2010)11 algorithm to detect vegetation cover changes for the two periods. The results were used to create two biomass maps for 2010–2013 and for 2014–2018 (Figure 9) by applying the biomass stocking factors from the biophysical survey. The 2010 and 2015 LULC maps were reclassified into just four classes based on their prominence in the landscape, accessibility, and biomass stocking: 1. woodland, 2. bushland, 3. cropland, and 4. other. The classes of the land cover maps were combined with the two classes of the change maps (loss and degradation). In more detail, ‘intact woodland’ and ‘intact bushland’ are vegetated areas that remain ‘stable’, without degradation and loss. Degraded classes refer to partial vegetation removal while loss occurs when there is complete vegetation removal. 10 The data represent the 30 m DEM from Shuttle Radar Topography Mission. (http://geoportal.rcmrd.org/layers/servir%3Auganda_srtm30meters). 11 For more information on BFAST: http://bfast.r-forge.r-project.org/. 16 Figure 9. Biomass stock within the settlements and 15 km of settlement boundaries, early 2018 17 Figure 10. Biomass stock changes between 2013 and 2018 within settlements and 15 km buffers 18 According to the results for the 5 km buffer zone (Table 9), the total tree cover loss between 2010 and 2013 was about 1,919 hectares (ha), while degradation covered about 5,664 ha (in woodland and bushland, including the areas of the settlements themselves). Meanwhile, from 2014 to 2018, there was 34,112 ha of loss and 29,604 ha of degradation. Between the two periods, there was about a 12 percent increase in areas affected by degradation and loss on the total areas within the 5 km buffer zone. Total biomass loss accounts for the total loss, including the loss from degraded land.12 The overall picture indicates a significant increase in loss and degradation, not only within the 5 km buffer near the refugee settlements but also in the extended 15 km buffer from their boundaries. The latter is especially interesting as it is unlikely to have any direct link to the presence of the refugees but suggests extensive ongoing degradation cause by host communities. Table 9. Loss and degradation (ha) and biomass (AGB) changes in selected land cover classes within 5 and 15 km of the refugee settlement boundaries 5 km buffer 15 km buffer Loss and 2010–2013 2014–2018 2010–2013 2014–2018 degradation Total area AGB Total AGB Total AGB Total AGB (ha) stock (t) area (ha) stock (t) area (ha) stock (t) area (ha) stock (t) Loss in woodland 157 5,961 3,288 124,950 536 20,358 9,253 351,614 Loss in bushland 703 19,532 6,998 194,543 1,428 39,696 14,015 389,624 Loss in cropland 1,060 10,521 23,826 236,591 2,141 21,255 54,311 539,306 Total loss 1,919 36,015 34,112 556,084 4,104 81,309 77,579 1,280,544 Degraded woodland 1,425 36,088 10,558 267,427 4,073 103,164 25,872 655,341 Degraded bushland 4,240 16,704 19,047 75,044 8,797 34,660 38,787 152,822 Total degradation 5,664 29,604 12,870 64,660 Biomass loss in — 27,169 — 201,336 — 77,668 — 493,381 degraded woodland Biomass loss in — 44,728 — 200,942 — 92,809 — 409,207 degraded bushland Total biomass loss — 71,897 — 402,277 — 170,477 — 902,588 from degraded land Total biomass loss 107,912 958,361 251,786 2,183,132 Table 10 shows estimates of loss and degradation in the settlements and within the 5 km buffer. The remaining AGB or net woody biomass is the sum of the biomass from degraded classes with the biomass from the intact classes. Only changes derived from the time series analysis (BFAST) are considered, rather than changes between the intact woodlands and bushlands derived from the two mapped periods. In other words, changes within ‘intact’ classes (that is, intact woodland, intact bushland, and cropland) between the two periods (2010–2013 and 2014– 2018) should not be compared with the change estimates (loss and degradation) resulting from BFAST results, since they refer to two different datasets and approaches. Details are provided in the annexed methodology. 12 The biomass factor used to compute biomass loss in degraded land is taken as the difference between the biomass factors for intact woodland (38 t per ha) and degraded woodland (25.3 t per ha), which is 12.7 t per ha. Similarly, for the bushland class, it is the difference between the biomass factors for intact bushland (27.8 t per ha) and degraded bushland (3.9 t per ha), which is 23.9 t per ha. 19 The LULC maps were used to classify changes and provide insight on possible drivers. For example, it is evident that the loss observed in the Rhino Camp extension and Palabek is mainly related to loss in woodland, probably due to agricultural expansion. Meanwhile in Agojo and Ayilo II, major losses are found in cropland and bushland, while in Nyumanzi it is bushland that is most affected by human impact. Overall, the results presented in Table 10 show an increase in degradation and loss in both woodland and bushland after the refugees’ arrival. For instance, in Bidibidi, AGB decreased from 1.6 million t in 2013 to about 1 million t in 2018 and the area of degraded woodland increased from 470 to 4,409 ha in the 5 km buffer zone. Ayilo settlements (I and II) are the most affected in terms of degradation, especially in woodland. Agojo, Nyumanzi, and Rhino extension also show degradation, though at a more restricted scale. Other settlements showing an increase in loss and degradation are Imvepi and Maaji I, Nyumanzi, and Palabek. More details are provided in Table 10. However, while there is an increase in observed degradation, the spatial distribution of biomass loss (as mapped in Figure 10) does not provide strong evidence that this results primarily (or even majorly) from refugee woodfuel harvesting. The highest losses are seen in host community areas set back from the settlement boundaries. 20 Table 10. Degradation and loss within the refugee settlements and 5 km buffer zone, with net biomass estimates (DM) for 2010–2013 and 2014– 2018 Bidibidi Imvepi Rhino extension - Omugo Agojo 2010–2013 2014–2018 2010–2013 2014–2018 2010–2013 2014–2018 2010–2013 2014–2018 Land cover class ha t ha t ha t ha t ha t ha t ha t ha t Intact woodland 11,592 440,513 5,713 217,095 2,620 99,549 1,482 56,327 1,438 54,648 233 8,854 94 3,570 28 1,081 Intact bushland 22,102 614,449 6,888 191,481 2,597 72,205 1,269 35,273 541 15,050 407 11,315 2,189 60,846 289 8,024 Cropland 62,222 568,705 59,168 540,798 18,583 169,850 15,637 142,920 5,938 54,271 6,465 59,090 8,511 77,789 8,710 79,614 Other 62,498 — 66,801 — 17,469 — 16,410 — 14,792 — 12,094 — 3,506 — 1,987 — Degradation Degraded woodland 470 11,916 4,409 111,692 137 3,458 1,635 41,411 61 1,543 1,181 29,912 3 82 68 1,710 Degraded bushland 1,446 5,697 8,145 32,092 170 670 1,588 6,258 29 116 695 2,738 169 667 570 2,246 Total AGB remaining (t) 1,641,279 1,093,157 45,733 282,189 125,628 111,908 142,955 92,674 Loss Loss in woodland 52 1,969,92 1,287 48,903 12 469 393 14,949 3 123 233 8,871 0 10 19 728 Loss in bushland 170 4,726 3,324 92,396 28 776 460 12,788 2 48 125 3,463 15 413 188 5,217 Loss in cropland 424 3,876 5,284 48,300 153 1,398 2,829 25,858 31 279 878 8,029 77 703 2,714 24,805 Total loss (ha) 646 9,895 193 3,682 35 1,237 92 2,921 Ayilo I Ayilo II Boroli I/II Maaji I 2010–2013 2014–2018 2010–2013 2014–2018 2010–2013 2014–2018 2010–2013 2014–2018 Land cover class ha t ha t ha t ha t ha t ha t ha t ha t Intact woodland 2,855 108,479 16 602 3,289 124,974 14 551 254 9,641 36 1,378 542 20,602 259 9,860 Intact bushland 1,666 46,307 1,967 54,689 1,947 54,121 1,760 48,924 1,709 47,508 1,605 44,618 2,852 79,281 417 11,589 Cropland 4,022 36,762 6,988 63,874 2,009 18,366 4,553 41,619 6,739 61,593 6,946 63,485 3,997 36,535 3,782 34,568 Other 4,764 — 1,410 — 3,862 — 1,356 — 2,113 — 902 — 329 — 2,210 — Degradation Degraded woodland 158 4,008 5 121 152 3,857 18 456 15 392 27 684 16 397 116 2,939 Degraded bushland 244 960 1,496 5,895 229 901 1,692 6,667 104 408 1,017 4,006 100 395 334 1,314 Total AGB remaining (t) 196,516 125,180 202,219 98,216 119,542 114,172 137,210 60,270 Loss Loss in woodland 34 1,303 — — 37 1,416 1 34 1 31 1 31 3 116 44 1,662 Loss in bushland 51 1,419 662 18,39 66 1,821 986 27,407 7 208 134 3,733 14 378 236 6,563 Loss in cropland 155 1,412 1,412 12,904 85 778 1,265 11,560 126 1,152 397 3,632 33 300 458 4,188 Total loss (ha) 240 2,073 188 2,252 134 532 50 738 21 Maaji II Maaji III Nyumanzi Pagirinya 2010–2013 2014–2018 2010–2013 2014–2018 2010–2013 2014–2018 2010–2013 2014–2018 Land cover class ha t ha t ha t ha t ha t ha t ha t ha t Intact woodland 3,562 135,340 3,032 115,210 28 1,081 1,036 39,374 217 8,249 89 3,379 651 24,751 13 489 Intact bushland 6,263 174,117 1,237 34,387 289 8,024 711 19,771 3,348 93,062 480 13,331 3,201 88,986 261 7,263 Cropland 2,327 21,268 4,343 39,699 8,710 79,614 4,938 45,136 6,561 59,968 5,850 53,471 2,093 19,132 4,248 38,825 Other 135 — 3,070 — 1,987 — 3,110 — 2,727 — 4,052 — 8,121 — 7,974 — Degradation Degraded woodland 42 1,074 277 7,021 68 1,710 490 12,418 4 89 164 4,158 85 2,154 121 3,073 Degraded bushland 185 730 194 765 570 2,246 370 1,458 97 384 1,077 4,245 460 1,811 494 1,947 Total AGB remaining (t) 332,529 197,082 92,674 118,157 161,751 78,584 136,834 51,597 Loss Loss in woodland 6 239 36 1,358 19 728 73 2,784 — — 24 930 8 287 141 5,342 Loss in bushland 48 1,331 146 4,046 188 5,217 227 6,318 5 128 259 7,188 77 2,142 168 4,671 Loss in cropland 24 223 254 2,318 2,714 24,805 731 6,684 17 158 971 8,877 51 465 1,293 11,817 Total loss (ha) 79 436 2,921 1,032 22 1,254 135 1,602 Palorinya Palabek 2010–2013 2014–2018 2010–2013 2014–2018 Land cover class ha t ha t ha t ha t Intact woodland 6,534 248,278 3,595 136,626 17,939 681,668 1,347 51,184 Intact bushland 9,699 269,633 4,665 129,681 1,721 47,833 425 11,827 Cropland 11,510 105,204 7,976 72,903 8,998 82,240 33,070 302,257 Other 19,662 — 21,164 — 38,889 — 23,706 — Degradation Degraded woodland 126 3,187 2,494 63,177 457 11,579 1,476 37,376 Degraded bushland 1,602 6,312 5,277 20,791 64 253 403 1,586 Total AGB remaining (t) 632,614 423,178 823,572 404,230 Loss Loss in woodland 12 462 375 14,251 37 1,406 1,118 42,473 Loss in bushland 321 8,915 1,631 45,344 4 123 121 3,353 Loss in cropland 150 1,373 2,420 22,115 49 445 6,488 59,302 Total loss (ha) 483 4,426 90 7,727 22 Table 11 highlights the total degradation and loss (including the partial loss in degraded bushland and woodland). The settlements most affected by major changes in woodland, bushland and cropland can be noted by comparing the total loss and degradation within the 5 km buffer zone from the boundaries of each settlement (plus the areas of the settlements themselves) over the two periods. Table 11. Summary of degradation and loss (in ha) per settlement within 5 km 2010–2013 2014–2018 Total Degradation Loss % loss and Degradation Loss % loss and Settlement District area (ha) (ha) (ha) degradation (ha) (ha) degradation Bidibidi Yumbe 161,131 1,916 646 −1.6 12,555 9,895 −13.9 Imvepi Arua 41,765 307 193 −1.2 3,223 3,682 −16.5 Rhino ext. Arua 22,884 90 35 −0.5 1,876 1,237 −13.6 - Omugo Agojo Adjumani 14,568 173 92 −1.8 638 2,921 −24.4 Ayilo I Adjumani 13,949 402 240 −4.6 1,501 2,073 −25.6 Ayilo II Adjumani 11,640 381 188 −4.9 1,710 2,252 −34.0 Boroli I/II Adjumani 11,061 119 134 −2.3 1,044 532 −14.2 Maaji I Adjumani 7,854 116 50 −2.1 450 738 −15.1 Maaji II Adjumani 12,589 228 79 −2.4 471 435 −7.2 Maaji III Adjumani 11,714 638 2,921 −30.4 860 1,032 −16.2 Nyumanzi Adjumani 12,962 101 22 −0.9 1,242 1,254 −19.3 Pagirinya Adjumani 14,709 545 135 −4.6 615 1,602 −15.1 Palorinya Moyo 49,633 1,728 483 −4.5 7,771 4,426 −24.6 a Palabek Lamwo 68,131 521 90 −0.9 1,878 7,727 −14.1 Average (%) -4.5 Average (%) -18.1 Note: a. Changes in Palabek consider only the most recent years, 2017–2018. 23 3.3 Linking woodfuel demand and supply Table 12 shows estimated woodfuel supply and demand for each refugee settlement, including both firewood and charcoal (the latter converted to firewood equivalent). Potential supply takes into account annual AGB growth from woodland and bushland within 5 km of the settlement boundaries. Woodfuel demand estimates are based on official refugee population data from April 2019. Table 12. Estimated woodfuel demand and supply in the target refugee settlements and within 5 km buffer zone Woodfuel Annual Refugee demand AGB stock Annual AGB Annual AGB net Settlement population refugees (t) growth (t/yr) loss/gain (t/ yr) loss/gain (Apr 2019) (t/yr DM)* (%) Bidibidi 225,808 122,395 1,093,157 29,214 -93,181 -8.5% Imvepi 57,758 31,306 282,189 7,913 -23,393 -8.3% Rhino extension - Omugo 24,533 13,298 111,908 2,486 -10,812 -9.7% Agojo 6,661 3,610 92,674 379 -3,232 -3.5% Ayilo I 23,837 12,921 125,180 793 -12,128 -9.7% Ayilo II 13,722 7,437 98,216 890 -6,548 -6.7% Boroli I/II 14,841 8,044 114,172 793 -7,251 -6.4% Maaji I 518 280 60,270 1,244 964 1.6% Maaji II 16,174 8,767 197,082 11,549 2,782 1.4% Maaji III 14,947 8,102 118,157 4,553 -3,548 -3.0% Nyumanzi 39,505 21,413 78,584 870 -20,543 -26.1% Pagirinya 35,803 19,406 51,597 364 -19,042 -36.9% Palorinya 119,587 64,820 423,178 18,170 -46,651 -11.0% Palabek 43,238 23,436 404,230 6,767 -16,669 -4.1% Total 636,932 345,236 3,250,598 85,984 -259,251 -8.0% *Note: DM = dry matter. Woodfuel demand converted to dry basis assuming 18 percent moisture content. AGB growth rates taken from the NBS as averages for the agro-ecological zone of the AOI, which is classified as semi-moist lowland (see Table 8). Growth rates of degraded woodland and bushland estimated by using correction factors of 0.33 and 0.85, respectively, derived from the ratio of AGB stock of the degraded to the intact classes. Estimate of annual AGB loss takes into account HH woodfuel demand based on April 2019 refugee population, though field observations highlighted other demand for woody biomass for construction, energy for commercial and economic activities, agricultural activities, and losses to fire. The impacts of a reduced refugee population were also considered. In a scenario with a refugee population that is reduced in the target area by 15 percent, 30 percent, and 45 percent, assuming woodfuel demand per person remains stable, the annual biomass loss would decrease from the current estimated 8 percent to 6.4 percent, 4.8 percent, and 3.2 percent, respectively (Table 13). 24 Table 13. Scenarios with refugee population reductions Refugee population (−15%) Refugee population (−30%) Refugee population (-45%) Annual Woodfuel Annual Annual Woodfuel Annual Annual Woodfuel Annual Annual AGB stock AGB demand AGB net demand AGB net demand AGB net Settlements (t) growth refugees loss/gain loss/gain refugees loss/gain loss/gain refugees loss/gain loss/gain (t/yr) (t/yr - DM) (t/yr) (%) (t/yr - DM) (t/yr) (%) (t/yt- DM) (t/yr) (%) Bidibidi 1,093,158 29,214 104,036 -74,822 -6.8% 85,677 -56,463 -5.2% 67,317 -38,103 -3.5% Imvepi 282,189 7,913 26,611 -18,698 -6.6% 21,915 -14,002 -5.0% 17,219 -9,306 -3.3% Rhino ext. - 111,909 2,486 11,303 -8,817 -7.9% 9,308 -6,822 -6.1% 7,314 -4,828 -4.3% Omugo Agojo 92,675 3,069 -2,690 -2.9% 2,527 -2,148 -2.3% 1,986 -1,607 -1.7% 379 Ayilo I 125,181 793 10,982 -10,189 -8.1% 9,044 -8,251 -6.6% 7,106 -6,313 -5.0% Ayilo II 98,217 890 6,322 -5,432 -5.5% 5,206 -4,316 -4.4% 4,091 -3,201 -3.3% Boroli I/II 114,171 793 6,838 -6,045 -5.3% 5,631 -4,838 -4.2% 4,424 -3,631 -3.2% a Maaji I 60,270 1,244 239 1,005 1.7% 197 1,047 1.7% 154 1,090 1.8% a Maaji II 197,082 11,549 7,452 4,097 2.1% 6,137 5,412 2.7% 4,822 6,727 3.4% a Maaji III 118,157 4,553 6,886 -2,333 -2.0% 5,671 -1,118 -0.9% 4,456 97 0.1% Nyumanzi 78,584 870 18,201 -17,331 -22.1% 14,989 -14,119 -18.0% 11,777 -10,907 -13.9% Pagirinya 51,597 364 16,495 -16,131 -31.3% 13,584 -13,220 -25.6% 10,674 -10,310 -20.0% Palorinya 423,178 18,170 55,097 -36,927 -8.7% 45,374 -27,204 -6.4% 35,651 -17,481 -4.1% Palabek 404,230 6,767 19,921 -13,154 -3.3% 16,405 -9,638 -2.4% 12,890 -6,123 -1.5% Total 3,250,598 85,984 293,452 -207,468 -6.4% 241,665 -155,681 -4.8% 189,881 -103,897 -3.2% 25 4. Recommended technical interventions Wood is the main source of energy for both refugee and host communities in northern Uganda. Demand for woodfuel is expected to increase with rising population, as other energy options for cooking are unaffordable or inferior. This could widen the gap between demand and sustainable supply, placing growing strains on the well- being of both hosts and refugees and causing degradation of woody resources in and around the refugee settlements. On the other hand, the leading driver of conversion from woodland to other LULC categories is agriculture, which is dominated by host populations. The following intervention options can help support sustainable environmental management, ensure energy access for cooking, and contribute to building livelihood resilience in both refugee and host communities. 1) Development of agroforestry systems: Consisting in interplanting of trees and crops for different purposes such as energy, food, and fodder. 2) Establishment of woodlots for energy and other purposes: Establishment of woodlots with trees planted at a high density to maximize biomass production and with short rotation length for a sustainable source of fuelwood as well as for poles to construct shelter and other products such as fruits, leaves, and fodder. 3) Rehabilitation of degraded forests: A combination of natural and assisted regeneration to restore areas of degraded native forest and boost productivity over the longer term. 4) Enhancement of energy efficiency: To reduce demand for woodfuel through improvement in household cooking efficiency and charcoal production efficiency. Each option is described in more detail below. 4.1 Development of agroforestry systems Agroforestry is an intervention designed to address land degradation while also providing woodfuel, food (for example, fruits, nuts, edible leaves), timber, fodder for livestock, and other non-wood products. The integration of trees into farming systems can enhance livelihood opportunities and increase the resilience of both host and refugee communities, contributing to food and nutrition security and generating income. In addition, agroforestry represents a suitable activity for the restoration of degraded lands, bringing people involved to identify and implement specific practices in which woody perennials (trees, shrubs, palms, and bamboos) are combined with agricultural crops and/or animals on the same land management unit. Trees planted in agroforestry systems can provide a number of other benefits, for example, fixing nitrogen, stabilizing the soil, providing shade, defining boundaries, and supporting pollination services. The establishment of trees and shrubs in strategic places, especially the residential plots assigned to the refugees, can diversify and increase agricultural production while also providing an opportunity to bridge the humanitarian response and sustainable development. These systems can take advantage of small patches of land to produce woodfuel, food, and fodder and to make living fences for delineation of refugees’ household plots. Other areas suggested for this type of intervention are the cultivated fields of both host and refugee communities in the surroundings of the refugee settlements. For example, Palabek, Ayilo I and II, Agoyo, Bidibidi, Maaji II, and Pagirinya, where large areas are under cultivation. Local landlords, cooperatives, and other group or individuals of refugee and host communities can also be supported through incentive schemes (for example, microfinance) to motivate investments in agroforestry and cover the start-up costs. Possible species for agroforestry interventions in this context are calliandra (Calliandra calothyrsus) and other multipurpose trees such as sesbania (Sesbania spp.), tephrosia (Tephrosia spp.), pigeon pea (Cajanus cajan), gliricidia (Gliricidia sepium), and moringa (Moringa oleifera), which fix nitrogen and provide woodfuel, mulch, and 26 fodder. The most important crops in northern Uganda for possible intercropping are cassava, beans, groundnuts, sesame, millet, sorghum, maize, and okra. The use of bamboo species in agroforestry could fulfil an interesting multipurpose role such as in providing building materials, erosion control, stream bank stabilization, livestock fodder and demarcation. Before introducing bamboo species as part of agroforestry systems, it is important to evaluate effective management strategies to avoid risks of invasiveness and other possible negative impacts on the environment. As part of this intervention, it is important to introduce training to raise awareness on the benefits of agroforestry, provide technical support and extension services, and encourage both host and refugee communities to adopt agroforestry systems. The involvement of the District Forestry and Agricultural Offices could start with the support of relevant partners to establish demonstration plots, tree nurseries, and training centers in the refugee settlements and surrounding villages. The World Agroforestry Centre (also known as ICRAF) has recently implemented an agroforestry pilot project at Rhino Camp and Imvepi refugee settlements, showing that agroforestry systems can be rolled out in a refugee situation targeting both refugee and host communities. A major challenge to implementing agroforestry interventions occurs when there is land (and hence tree) tenure insecurity, which is why the refugees’ own household plots are considered particularly important for this intervention. The time required before harvesting depends on the species selected, and this might create a disincentive to invest in trees, particularly in view of uncertainty over the refugees’ duration of stay. Multipurpose and fast-growing woody species (for example, pigeon pea, moringa, caliandra, leucaena) should be considered to increase the motivation of people to manage trees effectively, by providing several benefits such as materials for fencing, fruits, fodder, and ecosystem services such as soil conservation and soil fertility. Table 14 summarizes the estimated costs for agroforestry on a hectare basis. In this scenario, labor for land preparation, harvesting, and other field operations is deemed to be provided by the households, so it is not costed. Table 14. Indicative costs of agroforestry intervention, per hectare basis Years 1 2 Cost (US$) Community tree/garden center US$ per ha (one per 30 ha of agroforestry) Establishment 312 1 312 Management 26.5 1 1 53 Agricultural inputs US$ per ha Seeds 20 1 1 40 Fertilizers 60 1 1 120 Training package US$ per ha Agroforestry experts and communication 20 1 1 40 Total US$ per ha 565 Table 15 costs a scenario in which both refugee and hosting populations are involved in agroforestry within the settlements and the 5 km buffer zone. The potential cropland for an agroforestry intervention is estimated to estimate the cost. On average, the hosting population in northern Uganda cultivates 1.7 ha per household (Mwaura 2016), while the land allocated to the refugee households for production differs by settlement and may be 30 x 30 m, 50 x 50 m, or 100 x 100 m. For costing purposes, an average of 50 x 50 m is assumed. The number of households is derived from the total refugee and hosting populations and the average number of members (7.9 and 6.3, respectively) and by considering that 25 percent of the refugee population is engaged in farming activities (UNHCR 2017b), while 48 percent of host households in northern Uganda depend on subsistence farming (UBOS 2017). 27 Table 15. Indicative costs of agroforestry intervention within refugee settlements and 5 km buffers No. of Local Estimated No. of local Agroforestry Refugee refugee population land for Settlement agricultural investment population agricultural within 5 km agroforestry a households (US$) households buffer (ha) Bidibidi 225,808 7,146 436,782 33,279 58,360 32,973,470 Imvepi 57,758 1,828 45,901 3,497 6,402 3,617,253 Rhino extension 24,533 776 11,116 847 1,634 923,140 Agojo 6,661 211 74,193 5,653 9,662 5,459,289 Ayilo I 23,837 754 21,326 1,625 2,951 1,667,207 Ayilo II 13,722 434 15,017 1,144 2,054 1,160,295 Boroli I/II 14,841 470 21,473 1,636 2,899 1,637,753 Maaji I 518 16 12,101 922 1,571 887,878 Maaji II 16,174 512 7,454 568 1,093 617,788 Maaji III 14,947 473 10,728 817 1,508 851,897 Nyumanzi 39,505 1,250 33,340 2,540 4,631 2,616,438 Pagirinya 35,803 1,133 5,611 428 1,010 570,655 Palorinya 119,587 3,784 73,918 5,632 10,520 5,943,936 Palabek 43,238 1,368 32,701 2,492 4,578 2,586,361 Total 636,932 20,156 801,661 61,079 108,873 61,513,361 Note: a. Population estimates (Source: CIESIN 2016 as local population (‘pop2015’) according to the local population density calculated in 5 km buffers around the settlements. 4.2 Establishment of woodlots for energy and other purposes Firewood and charcoal are the main sources of energy for refugee and host communities in northern Uganda and the rapid increase of population due to the arrival of refugees has inevitably increased pressure on natural resources and resulted in an imbalance between demand and available supply within accessible walking distance. Interventions for the establishment of woodlots are recommended over a period of at least three to five years, to ensure sufficient time to establish adequate production capacity and proper transfer of knowledge to ensure sustainability. The objective should be to maximize biomass production in a short time and increase tree density to reach the optimum growth per unit of area. Fast-growing tree species and short-rotation coppice management should be adopted to enable early harvesting for fuelwood. In addition, the use of multipurpose species can increase people’s motivation to manage trees effectively because of the provision of other benefits (for example, building poles, fence posts, non-wood forest products such as fruits and fodder, and ecosystem services such as soil conservation and soil fertility). It is important to highlight that labor needed for planting and tending for trees is particularly intense for at least the initial three years before they produce an appreciable quantity of biomass. This intervention should target • Areas owned by host communities and individuals; • Protected areas managed by the NFA; and • Areas assigned to the refugees. Most species can be used for fuel, but quality varies greatly. Some species burn very fast while others produce a lot of smoke and are more difficult to dry. In Uganda, eucalyptus is mainly grown for domestic and industrial fuelwood, but other species have also been promoted for energy purposes such as Gmelina arborea, Grevelia robusta, Markhamia lutea, Acacia mangium, and Acacia auriculiformis. 28 It is important that refugee and host communities are involved and are given the responsibility for tree planting and management and for other aspects of this intervention (including dialogue and decision making). Beneficiaries should be organized into groups to encourage and promote tree planting. A participatory approach through consultation at all levels is required to allocate land for plantation and to agree on implementation modalities. Rules and rights need to be communicated and enforced. For refugee communities, site-specific formal agreements supporting tree planting are required to provide clarity on the land ownership of new plantations, including the land, trees, and other assets, and who will benefit from the eventual harvest of wood (FAO and UNHCR 2018). In addition to the land already assigned to the refugees for agricultural activities, refugee groups could acquire other communal land with the support of the OPM using the same process by which land is secured for refugee households. The local committees including the OPM, District Forestry Office, landlords, host and refugee community leaders, and relevant partners should be established in each settlement to identify available land and discuss in detail the management and ownership of proposed woodlots. FAO and the UNHCR are already investigating options for expanding the Sawlog Production Grant Scheme, Phase III (SPGS III) model into the refugee-hosting areas of northern Uganda, where land ownership is mainly communal, and to move to results-based financing for medium and large-scale tree planting. A verification process should be carried out to ensure physical establishment of plantations and adherence to quality standards (for example, use of appropriate species, seeds quality, survival rate of seedlings, conservation of ecosystem practices, and social issues such as labor management and community relations). Rather than paying 100 percent of funds up front for employed labor, only a portion of the financial support could be paid at the time of woodlot establishment followed by retrospective disbursements after verification of outputs and tree survival rates at agreed milestones. This would be an incentive for planning carefully the expected returns from the investment and at the same time guarantee that both refugee and host community groups have sufficient funds to establish and manage the plantations. Institutional woodlots should also be supported. There is a need to explore the possibility of obliging all institutions to have a minimum number or acreage under trees, with clear objectives and management plans. Institutions offer defined land ownership and can provide opportune locations to increase tree planting in the region, for example, faith-based communities, educational establishments, health facilities, and government offices at parish, subcounty, and district levels. Authorities are also willing to put in place bylaws that oblige initiatives that remove trees to plant others in return; for instance, when space is opened to settle refugees, trees should be planted along the new roads opened. Table 16 provides indicative costs of investment and operations for the energy woodlot working cycle. Establishment costs can vary significantly from district to district and are dependent on land type, vegetation, and other site-specific biophysical and socioeconomic factors. Table 16. Indicative investment and operational costs of establishing woodlots for energy, per hectare basis Years 0 1 2 3 4 5 Cost (US$) Site preparation US$ per ha Land clearing 79.9 1 79.9 Land preparation 68.1 1 68.1 Marking and pitting 38.5 1 38.5 Other preplant operations 118.4 1 118.4 Planting US$ per ha Outplanting 192.4 1 192.4 Survival count 1.8 1 1.8 Blanking 38.5 1 38.5 Postplanting and protecting US$ per ha Ring-hoeing 20.7 3 62.1 29 Years 0 1 2 3 4 5 Cost (US$) Slashing 26.6 1 1 53.2 Weeding 37.0 3 2 185.0 Termite control 118.4 1 118.4 Tending 50.3 1 50.3 Fire protection 14.2 1 1 1 1 1 71.0 Road works US$ per ha Road construction 17.8 1 17.8 Road maintenance 3.0 1 1 1 9.0 Harvesting US$ per ha Coppicing and pollarding 108.6 1 108.6 Overhead cost US$ per ha Land lease 7.4 1 1 1 1 1 1 44.4 Surveying 26.6 1 1 53.2 Technical management 3.0 1 1 1 1 1 15.0 Administration 1.5 1 1 1 1 1 7.5 Total US$ per ha 1,333.0 Table 17 provides indicative costs to set up a nursery for raising seedlings with an annual production capacity of 250,000 seedlings (able to support up to 50 ha of woodlot establishment). Table 17. Indicative costs to set up a nursery Description Years 1 2 3 4 5 Cost (US$) Nursery construction US$ Water supply (tank, pump, irrigation system) Protection (fence, shed net) 8,333 1 8,333 Structure (poles, bricks, polythene sheet) Others Maintenance (10%) 833 1 1 1 1 1 4,165 Labor US$ Bed construction Seed sowing 280 1 1 1 1 1 1,400 Watering Weeding and so on Tools US$ Assorted (wheelbarrows, rakes, hoes, knives, sprayers) 500 1 500 Consumables US$ Chemicals, poles, nails, food, and so on 250 1 1 1 1 1 1,250 Total US$ 15,648 Average cost per hectare US$ per ha 313 Under this intervention, productive woodlots in Uganda commonly achieve mean annual increments of 20–26 m3 per ha. Assuming average wood density of 600 kg per m3 and a biomass expansion factor of 1.5 (to include bark and branches in the mean annual increment), the total AGB increment achievable with tree plantations would be 18.0–23.4 t per ha per year. To compensate fully for the estimated annual loss of biomass (Table 12) and guarantee a fuel security for cooking, the minimum area of woodlots needed to meet the total woodfuel demand of the current refugee population in each settlement has been calculated (Table 18). 30 Table 18. Woodlot requirements for energy and indicative establishment and maintenance costs over five years Minimum woodlot AGB loss/gain Woodlot area Cost of woodlots and Settlement area per household (t/yr) (ha) nurseries (US$) (ha) Bidibidi -93,181 3,982 6,554,895 0.14 Imvepi -23,393 1,000 1,645,636 0.14 Rhino extension -10,812 462 760,556 0.15 Agojo -3,232 138 227,346 0.16 Ayilo I -12,128 518 853,152 0.17 Ayilo II -6,548 280 460,595 0.16 Boroli I/II -7,251 310 510,106 0.16 Maaji I 964 - - - Maaji II 2,782 - - - Maaji III -3,548 152 249,609 0.08 Nyumanzi -20,543 1,994 3,281,689 0.40 Pagirinya -19,042 712 1,172,616 0.16 Palorinya -46,651 1,994 3,281,689 0.13 Palabek -16,669 712 1,172,616 0.13 Total -259,251 12,254 20,170,505 0.15 Note: AGB loss/gain refers to the settlement areas plus a 5 km buffer. 4.3 Rehabilitation of degraded forests The rehabilitation of degraded forests surrounding the refugee settlements is a relatively cost-effective means of sustainably managing native resources, in which wood harvesting can be controlled and regulated by a continual series of felling cycles through dedicated harvesting plans, in accordance with practical needs and the socioeconomic and ecological characteristics of specific sites. The objective is to restore forest productivity with a view to producing a sustainable supply of woodfuel and ecosystem services. Extraction of woodfuel seems to be one of the drivers of degradation and loss, after the expansion of agricultural activities (although agricultural activities do play an important part in refugee integration and development). This intervention should target • Areas owned by host communities and individuals; • Protected areas managed by the NFA; and • Areas assigned to the refugees. Rehabilitation of woodland can be achieved through a combination of scattered tree planting and measures to assist natural regeneration as a mechanism of recovery. The field survey determined that wildlings and saplings, especially coppice shoots, are common in degraded woodland and bushland. This intervention could involve enrichment planting using nursery-grown seedlings of native species to accelerate the natural rehabilitation process. Capacity building at the refugee and host community levels should focus on strengthening of existing tree nurseries to produce appropriate species. The species selection for rehabilitation and protection should take into account the suitable measures for rehabilitation at the site level, although the preference is for species that are fast-growing, are adapted to the local climate and topography, and have strong root systems. A total of 70 tree species were recorded in the field work of this assessment, of which Acacia spp., Combretum spp. and Lannea spp. were dominant. 31 Maintenance is needed in the early years after out-planting to reduce the impact of weeds. Grasses need to be slashed to enhance the growth of wildlings and planted seedlings in the first two to three years; fire protection must also be undertaken to protect areas under rehabilitation. An important element of sustainable forest management is community participation management. In fact, it is vital for the success of this approach that the right to access the land and to harvest wood and non-wood forest products should be understood and agreed with local communities (including with the refugees). Experiences of participatory forest management in Uganda include • Community-based forest management, whereby forest resource management is exclusively based on efforts of the communities, and user rights over the forest resources belong to the community; • Collaborative forest management, where communities and other key stakeholders work in partnership on the management of forests; and • Private forests, where local community members manage their own trees on private land. (Turyahabwe et al. 2012). These approaches emphasize decentralization or devolution of forest management rights and responsibilities to communities. Sustainable use of the forest resources in and around the settlements can contribute significantly to the resilience of refugee and host communities by providing access to additional income, food, and other household resources. It is therefore important that both refugees and hosts are engaged in the rehabilitation of degraded forests through a participatory approach to ensure the wise use of natural resources and provide both groups with ongoing benefits. This passive rehabilitation strategy should be carefully planned, as the nature and extent of recovery depend on the ecology and disturbance of the areas and the condition of the landscape. Detailed land use assessment is required for each settlement to define areas for regeneration and restoration of forest productivity. The biophysical and socioeconomic barriers to rehabilitation require in-depth site assessment to determine the suitability of different rehabilitation measures. The intervention should include natural rehabilitation of degraded areas as well as assisted natural regeneration of areas with total woodland and bushland loss. Table 19 first provides indicative costs for the natural rehabilitation of degraded woodland and bushland, by protecting remnant trees from firewood harvesting, livestock grazing, and other destructive agents (for example, fire). Table 19. Indicative costs for natural rehabilitation of degraded areas, per hectare basis over five years Years 1 2 3 4 5 Cost (US$) Protection US$ per ha Fire protection 14.2 1 1 1 1 1 71.0 Watching 24.0 1 1 1 1 1 120.0 Tree marking 3.5 1 1 7.0 Overhead cost US$ per ha Surveying 26.6 1 1 53.2 Technical management 3.0 1 1 1 1 1 15.0 Administration 1.5 1 1 1 1 1 7.5 Total US$ per ha 274.0 The proposed rehabilitation intervention in the areas of woodland and bushland with total loss, detected through the remote sensing analysis, also includes enrichment planting of additional trees and further maintenance operations in the form of fire protection and weed control (year 1–2). Table 20 provides indicative costs for rehabilitation in areas of major degradation and loss through assisted natural regeneration. 32 Table 20. Indicative costs for assisted natural regeneration, per hectare basis over five years Years 1 2 3 4 5 Cost (US$) Site preparation US$ per ha Land preparation 11.3 1 11.3 Marking and pitting 6.4 1 6.4 Other preplant operations 19.7 1 19.7 Planting US$ per ha Planting 32.0 1 32.0 Survival count 1.8 1 1 3.6 Blanking 6.4 1 6.4 Post-planting slashing 4.4 1 1 8.8 Post-planting weeding 6.2 3 2 31.0 Protection US$ per ha Fire protection 14.2 1 1 1 1 1 71.0 Watching 24.0 1 1 1 1 1 120.0 Overhead cost US$ per ha Surveying 26.6 1 1 53.2 Technical management 3.0 1 1 1 1 1 15.0 Administration 1.5 1 1 1 1 1 7.5 Total US$ per ha US$ per ha 386.0 Table 21 summarizes the indicative costs per hectare (if divided over the nursery served) to set up a nursery with an annual production capacity of 250,000 seedlings. Assuming that enrichment planting for rehabilitation would require 400–800 seedlings per ha, the nursery costed would typically cover 470 ha of rehabilitation. Table 21. Indicative costs to set up a nursery for assisted natural rehabilitation, per hectare basis Description Years 1 2 3 4 5 Cost (US$) Nursery construction US$ Water supply (tank, pump, irrigation system) Protection (fence, shed net) 8,333 1 8,333 Structure (poles, bricks, polythene sheet) Others Maintenance (10%) 833 1 1 1 1 1 4,165 Labor US$ Bed construction Seed sowing 280 1 1 1 1 1 1,400 Watering Weeding and so on Tools US$ Assorted (wheelbarrows, rakes, hoes, knives, sprayers, etc) 500 1 500 Consumables US$ Chemicals, poles, nails, food, and so on 250 1 1 1 1 1 1,250 Total US$ 15,648 Average cost per hectare US$ per ha 33 Combining the costings from the three previous tables, Table 22 summarizes the total cost of rehabilitation for each refugee settlement over five years, according to the measured extent of degradation and loss of woodland and bushland within 5 km. Costs of rehabilitation can vary significantly from district to district and are dependent 33 on land type, vegetation, and other site-specific biophysical and socioeconomic factors. Further investigations are required to analyze site-specific conditions and assess feasibility. Table 22. Indicative costs of rehabilitation of degraded and lost woodland and bushland in the target refugee settlements Degraded Cost of natural Area of loss in Cost of assisted woodland and rehabilitation of woodland and natural Total cost Settlement bushland degraded areas bushland regeneration (US$) (ha) (US$) (ha) (US$) Bidibidi 12,555 3,436,304 4,611 1,931,548 5,367,851 Imvepi 3,223 882,135 853 357,322 1,239,457 Rhino ext. - Omugo 1,876 513,461 358 149,966 663,427 Agojo 638 174,621 207 86,712 261,333 Ayilo I 1,501 410,824 662 277,312 688,136 Ayilo II 1,710 468,027 987 413,454 881,481 Boroli I/II 1,044 285,743 135 56,552 342,294 Maaji I 450 123,165 280 117,292 240,457 Maaji II 471 128,913 182 76,240 205,153 Maaji III 860 235,382 300 125,670 361,052 Nyumanzi 1,242 339,935 283 118,549 458,484 Pagirinya 615 168,326 309 129,440 297,766 Palorinya 7,771 2,126,923 2,006 840,313 2,967,236 Palabek 1,878 514,009 1,239 519,017 1,033,026 Total 35,834 9,807,766 12,412 5,199,387 15,007,153 Note: Covers area of lost and degraded woodland and bushland within settlements and 5 km buffers. 4.4 Enhancement of energy efficiency Although the assessment shows that to a certain extent the refugee communities have embraced and adopted improved fuel-saving cookstoves, much can be done to increase coverage. Improved mud-stoves remain the most appropriate cooking solution and are already well-known and culturally acceptable to the refugee and local population. Although the results show a reasonable adoption of the improved mud-stoves for firewood among refugees (62.1 percent) and host communities (51.8 percent), there are still significant proportions using the 3- stone fire, particularly in the host communities. Therefore, extending the use of improved cookstoves to ensure that all households will shift at least to an improved mud-stove is also an intervention option to consider, to reduce the pressure on natural resources. Work is needed mainly to continue sensitization campaigns and demonstrations, especially in host communities where coverage is still low. From the perceptions of both refugee and host respondents, there is an indication that many people still need to be sensitized on how to improve the construction and use of improved cookstoves to enhance further energy efficiency. Modern pre-fabricated cookstoves are available in regional markets, but neither refugees nor locals have the funds to buy them in significant numbers, and free distribution should carefully consider a combination of local 34 specific factors to minimize uptake failure. Modern pre-fabricated stoves with very high efficiency should be installed at institutional levels (for example, schools, clinics, reception, and administrative centers) as well as at commercial level such as restaurants and bakeries. Other fuel-saving technologies and practices should be explored in relation to other common economic and commercial activities practiced in northern Uganda such as charcoal production, brick making, and tobacco curing. Reducing demand for fuelwood while providing access to alternative, locally sourced fuels can reduce the exposure of women and children to associated risks and reduce the time needed for collecting fuel and could thus have a significant impact on the quality of life, releasing some of their time for productive activities, education, or leisure. Reducing the amount of wood needed for cooking and providing alternative and sustainable livelihood opportunities can also help reduce environmental degradation, reduce expenditure on fuel, and improve food and nutrition security. Along with the use of more fuel-efficient cookstoves, the following energy-saving measures should be promoted to reduce energy consumption for cooking (FAO and UNHCR 2017): • The soaking of beans and grinding or cutting of food into smaller pieces. For example, beans should be soaked overnight for 8–14 hours and cooked the next day, so they will cook in a shorter time. • Drying fuel before use and processing into smaller pieces. Using dry wood would increase cooking energy efficiency, reducing the quantity required for cooking and with the side benefit of reducing harmful smoke emissions. • The use and production of heat retention boxes and bags using locally available materials, which can reduce fuel consumption by more than 40 percent. • The use of suitable lids for all cooking tasks to help contain heat so that food cooks faster. • Sharing cooking facilities among families. • Using traditional clay cooking pots—although more delicate, these absorb and retain heat longer than metal pots and, when hot, they require less fuel than metal pots to continue the cooking process. Support to the development of more sustainable charcoal value chains should also be considered under this intervention, including the provision and training on use of improved charcoal kilns. Through this intervention, technical and business skills and entrepreneurship training should be provided to groups of refugees and host communities. Links with existing microfinance services should be established for these groups. A shift from traditional charcoal kilns to a more efficient alternative could increase the wood conversion efficiency from 15– 20 percent to 25–30 percent, with better preparation and stacking of the wood and more careful management of the pyrolysis process. The use of more efficient kilns means the more efficient use of wood, thereby increasing output and reducing inputs in terms of wood and labor. Improved charcoal kilns can be produced in Uganda in various sizes, and key advantages should be considered—such as mobility. A portable steel kiln was considered in the costing analysis (Table 23) with a production capacity of 150 t of charcoal operating 300 days per year. This type of portable kiln might have a cost up to US$2,200 per unit plus US$500 per unit for other costs for the start-up and US$1,000 per unit for a training package to improve technical and business skills. In addition to the improved portable kiln, this intervention proposes the improvement of management of traditional kilns such as the improved basic earth kiln (IBEK) through training, exchange, and dialogue between charcoal producers to enhance energy efficiency by making small adjustments to the technology already widely in use. A training package at the household level is also included in this intervention to enhance energy-saving practices for cooking. 35 Table 23. Indicative costs for energy efficiency enhancements Years 1 2 Cost (US$) Household training package US$ per HH Demonstrations for energy-saving measures at the 5 1 1 10 household level Equipment and materials 15 1 15 Total per household (HH) US$ 25 Improved charcoal production US$ per unit Improved kiln (portable or IBEK) 2,200 1 2,200 Start-up cost 500 1 500 Kiln demonstration and training 1,000 1 1 2,000 Total per charcoal unit US$ 4,700 Note: HH = household. Table 24 shows the indicative costs of provision of improved kilns and a training package, taking into account the total households and the current charcoal consumption in the refugee and surrounding host communities living within the 5 km buffer zone of each settlement. Table 24. Indicative cost for the provision of improved charcoal kilns and training packages by refugee settlements Charcoal Household Number Improved consumption Refugee Host Total training of charcoal Settlement in settlement Total (US$) households households households packages improved production and 5 km (US$) kilns (US$) buffer (t/yr) Bidibidi 28,583 69,330 97,914 2,447,844 6,066 40 190,059 2,637,903 Imvepi 7,311 7,286 14,597 364,925 1,177 8 36,871 401,796 Rhino ext. 3,105 1,764 4,870 121,747 452 3 14,166 135,913 Agojo 843 11,777 12,620 315,496 528 4 16,545 332,040 Ayilo I 3,017 3,385 6,402 160,061 499 3 15,642 175,702 Ayilo II 1,737 2,384 4,121 103,015 303 2 9,493 112,509 Boroli I/II 1,879 3,408 5,287 132,176 357 2 11,201 143,376 Maaji I 66 1,921 1,986 49,659 77 1 2,416 52,075 Maaji II 2,047 1,183 3,231 80,763 299 2 9,362 90,124 Maaji III 1,892 1,703 3,595 89,872 298 2 9,336 99,208 Nyumanzi 5,001 5,292 10,293 257,317 816 5 25,566 282,883 Pagirinya 4,532 891 5,423 135,567 599 4 18,780 154,347 Palorinya 15,138 11,733 26,871 671,765 2,316 15 72,572 744,338 Palabek 5,473 5,191 10,664 266,595 871 6 27,305 293,900 Total 80,624 127,248 207,872 5,196,802 14,659 98 459,314 5,656,116 4.5 Additional recommended measures The recommended technical interventions should be coordinated under an integrated energy and environment program that has sufficient institutional capacity and resources to undertake more in-depth analysis, implementation, and management at the site level; carry out monitoring and evaluation; support systematic efforts to promote these interventions across the associated host communities; and ensure sound learning, sharing, and interaction with other programs of a similar nature both in Uganda and elsewhere. This will ensure 36 that the measures do not take place in isolation or in a scattered, ineffectual, and short-term manner. Such integrated energy and environment program could complement the community-driven approaches adopted under the DRDIP which is likely to focus on shorter-term development needs of host communities. The following additional measures are recommended to ensure that the proposed interventions are grounded in a holistic and effective institutional structure and are well informed by suitable contextual information and deep understanding of the issue: • Development of forest management plans. Forest management plans would support the energy needs of the refugee and host communities and reduce the environmental and social impacts caused by the overexploitation of natural resources and by deforestation and forest degradation. When designing forestry interventions, the following aspects should be considered: mobilization of relevant stakeholders for a coordinated response at local and national levels, identification and demarcation of potential sites for the rehabilitation of degraded forests and the establishment of woodlots, clarification of the tree and land tenure regimes, assessment of site suitability for intervention, identification of stewards who will maintain the woodlot and appropriate agroforestry systems, review of existing land use plans, and identification of land use arrangements among local stakeholders. After site demarcation, site suitability assessments should be conducted by forestry experts and local authorities to assess physical and socioeconomic attributes of selected sites (for example, road accessibility, natural regeneration, terrain, edaphic conditions, distance, water availability, hydrology, risks, and other local conditions). • Trials for species suitability. Trials should be set up at the institutional level to test and demonstrate the suitability of a range of species (and species mixes) for different purposes such as high planting densities to maximize woodfuel yields on specific sites and agroforestry systems to grow trees, crops, and/or livestock on the same plot, providing a range of goods and ecosystem services. • Field testing of cookstoves performance. Improved cookstoves and traditional methods for cooking in use at household and institutional levels should be tested through internationally agreed protocols that measure efficiency, pollutant emissions, and safety, to design site-specific interventions to enhance energy efficiency including possible improvements in the cooking practices. • Promotion of integrated approaches. To improve the management and use of natural resources as well as to enhance the resilience of refugees and host communities, participatory forest management approaches should be adopted. An integrated approach to the management of natural resources, including forests and other woodlands, is a prerequisite, given the links between the biophysical, social, economic, and political dimensions of the proposed interventions and recognizing the importance of stakeholder participation in their management and development at local and national levels. • Establishment of local associations/cooperatives. The establishment of local associations or cooperatives should be explored as a way of boosting the economic benefits of specific environmental and energy interventions, with arrangements that provide equal opportunities for participation by both the refugee and host communities. • Promotion of entrepreneurship. An incentive mechanism (for example, microcredit scheme) should be created to integrate and support refugees and host communities to become entrepreneurs capable of contributing to Uganda’s socioeconomic development by enhancing business skills and capacities to provide forest-related services and thereby assist in the implementation of the interventions proposed in this study. • Local capacity building. Efforts should be made to build capacity among local authorities and partners to increase the technical and managerial skills needed for the rehabilitation of forests and other woodlands and the management of plantations and agroforestry systems. There is a need to identify specific areas of 37 need and relevant targeted people/groups for skills enhancement and to develop local technical capacities for the sustainable collection, production, and use of woodfuel. Capacity development of local governments should include monitoring and managing of woodfuel supply and demand; developing forestry management plans that support both host and refugee populations; and linking the importance of collective action on environmental conservation measures (for example, sustainable forest management, energy-saving measures at institutional and household levels) to improved livelihoods, which in turn will contribute to ensuring food security and nutrition. Relevant stakeholders should use or revive/strengthen existing structures such as local environment committees and resource/water user committees. These structures can then be linked to local government structures to ensure service provision at local levels. • Secure land and tree tenure. Issues regarding secure land and tree tenure need to be cleared and include incentive mechanisms for the adoption of sustainable land management by refugees and host communities. The allocation of additional land for specific purposes to ensure a sustainable supply of fuelwood such as the establishment of dedicated energy woodlots needs to be agreed upon between the parties regarding land ownership, period of use, right to harvest, and security. The OPM should adopt the approach it uses to negotiate with landlords to acquire land for settling refugees, but with the specific objective of tree planting and suitable terms. The area of land potentially available for interventions needs to be identified in situ through the participation of host communities, refugees, the OPM, and district authorities. Secure tree and land tenure is a prerequisite for the broader engagement of refugee and host communities to undertake tree planting. • Awareness raising on sustainable forest management. Awareness should be raised about the importance of sustainable forest management and the business potential of wood energy plantations, agroforestry systems, and enhancement of energy efficiency to ensure full understanding and support among the refugee and host communities and other stakeholders. • Monitoring. Degradation of land and other natural resources in areas affected by refugee influx should be monitored continuously with the support of the NFA. This will also include monitoring the progress made by implementation of activities. 38 5. Conclusions The population in northern Uganda has increased dramatically following the settlement of over 700 thousand refugees since 2014, and this presents a risk of competition with host communities for natural resources such as land, water, and wood, which will ultimately cause deforestation and/or environmental degradation. Woodland and bushland in areas surrounding the refugee settlements and nearby villages are the main sources of the wood needed as fuel for cooking, while cropland represents an additional source of firewood for the host communities. Impacts on the surrounding environment of refugee settlements resulting from the collection and production of fuelwood and charcoal can be lasting and damaging. This assessment indicates a steady increase in degradation and vegetation loss over the hosting area, and map comparisons reveal increased land cover changes in the woodland and bushland. The areas within the settlements and the buffer zone of 5 km around their boundaries have been subjected to changes after the refugee arrival, and in some of the target settlements, competition for the available resources could be a source of tension between the refugee communities and hosts living in the immediate surroundings. The livelihoods of refugee and host households are highly dependent on forests and other woodlands as primary sources of woodfuel for cooking. Average daily consumption of firewood by refugees is 1.6 kg per person and about 30 percent higher among host communities (2.1 kg). Taking into account the use of charcoal, the average daily fuel consumption rises to 1.8 kg per person in firewood equivalent among refugees and 2.2 kg among households of host communities. Refugee woodfuel consumption at Bidibidi has significantly reduced, about half the amount as identified in a survey conducted in March 2017, probably due to greater wood shortages, a more diverse diet with fresher food, the use of drier wood, and improved stoves. Total cooking fuel demand in the target refugee settlements is about 345,000 t per year—about four times the quantity of tree growth within 5 km of the settlements—which means that harvesting exceeds sustainable limits (pending verification of population data, which may have a significant bearing on this conclusion). Due to the straight connection between the estimated cooking fuel demand and the refugee numbers, and since the refugee verification exercise in Uganda is ongoing at the time of writing, the estimated woodfuel demand may vary. Communities are constructing improved cookstoves from locally available materials. Generally, there are more refugee households using such devices than host communities, and in Bidibidi settlement there is a noticeable increase in their adoption and use compared to the previous year. However, the challenges associated with firewood access and use are still preeminent—a problem for both refugee and host communities. Modern prefabricated cookstoves are available in regional markets, but neither refugees nor locals have the funds to buy them, and free distribution should carefully consider a combination of local-specific factors to minimize uptake failure. Improved mud-stoves are likely to remain a practical cooking solution and are a ‘technology’ already well- known and culturally acceptable to the refugee population. The livelihoods of refugee and host households are highly dependent on natural resources such as land and water for subsistence farming as well as woodland and bushland as a source of fuelwood for cooking. In addition, the majority of households have constructed semipermanent structures and have improved their homes with latrines and dedicated kitchens, and a few have bathing shelters, animal sheds, and poultry or bird pens. In this regard, households also need wood to build and maintain these structures. Although natural resources depletion is a major concern for the government and partners, very few organizations working in refugee-affected areas are focusing on the issue of environment and energy. The few organizations that do work in the sector are mainly operating at a small scale with 12-month budget cycles. To ensure an effective and harmonized approach with suitable technical expertise and adequate resourcing, there is a need for 39 a joint action to implement multicomponent interventions through a multiyear and multiagency arrangement. This will effectively address the environmental degradation factors. Planning for the sustainable supply of energy plays a crucial role in minimizing environmental impacts and conflicts with host communities over the use of natural resources. Dedicated woodlots provide for a sustainable supply of woodfuel and rehabilitation interventions on degraded land enhance availability and productivity of forest products (wood and non-wood forest products) and ecosystem services. Agroforestry interventions along with a more efficient use of energy for cooking and charcoal production can reduce these environmental impacts. It is expected that refugee and host communities will continue using fuelwood and charcoal for the foreseeable future as their primary sources of energy. Therefore, responsible planning for sustainable harvesting, production, and use of fuelwood is crucial for enabling sustainable development by ensuring energy access and, in turn, building resilience in the refugee-affected areas and contributing to food and nutrition security. 40 References BFAST (Breaks for Additive Seasonal and Trend). 2010. Spatial R package. www.loicdutrieux.net/bfastSpatial/. Chave, J., Rejou-Mechain, M., Burquez, A., Chidumayo, E., Colgan, M.S., Delitti, W.B.C., Duque, A., Eid, T., Fearnside, P.M., Goodman, R.C., Henry, M., Martinez-Yrizar, A., Mugasha, W.A., Muller-Landau, H.C., Mencuccini, M., Nelson, B.W., Ngomanda, A., Nogueira, E.M., Ortiz-Malavassi, E., Pelissier, R., Ploton, P., Ryan, C.M., Saldarriaga, J.G. and Vieilledent, G. 2014. Improved allometric models to estimate the aboveground biomass of tropical trees. Global Change Biology, 20(10): 3177–3190. CIESIN (Center for International Earth Science Information Network). 2016. High Resolution Settlement Layer. Source Imagery 2016 DigitalGlobe, Inc. Columbia University, New York. www.ciesin.columbia.edu/data/hrsl/. Accessed February 2018 DeVries, B., A. K. Pratihast, J. Verbesselt, L. Kooistra, and M. Herold. 2016. “Characterizing Forest Change Using Community-Based Monitoring Data and Landsat Time Series.” PLoS ONE 11 (3): e0147121. www.ncbi.nlm.nih.gov/pmc/articles/PMC4809496/pdf/pone.0147121.pdf. Dutrieux, L. P., J. Verbesselt, L. Kooistra, and M. Herold. 2015. “Monitoring Forest Cover Loss Using Multiple Data Streams, a Case Study of a Tropical Dry Forest in Bolivia.” ISPRS Journal of Photogrammetry and Remote Sensing, 107(2015): 112-125. FAO (Food and Agriculture Organization of the United Nations). 2015. Global Forest Resources Assessment 2015. Rome. www.fao.org/3/a-i4782e.pdf. FAO, and UNHCR (United Nations High Commissioner for Refugees). 2016. Assessing Woodfuel Supply and Demand in Displacement Settings. Rome. www.fao.org/3/a-i5762e.pdf. ———. 2017. Rapid Woodfuel Assessment: 2017 Baseline for the Bidibidi Settlement, Uganda. Rome, FAO, and Geneva, Switzerland, UNHCR. www.fao.org/3/a-i7849e.pdf. ———. 2018. Managing Forests in Displacement Settings: Guidance on the Use of Planted and Natural Forests to Supply Forest Products and Build Resilience in Displaced and Host Communities. Rome, FAO and UNHCR. www.fao.org/3/I8309EN/i8309en.pdf. Forest Department. 2002. National Biomass Study. Technical Report. Kampala, Ministry of Water Lands and Environments. Hansen, M. C., Potapov, P.V., Moore, R., Hancher, M., Turubanova, S. A., Tyukavina, A., Thau, D., Stehman, S.V., Goetz, S. J., Loveland, T. R., Kommareddy, A., Egorov, A., Chini, L., Justice, C.O., Townshend, J. R. G. 2013. “High- Resolution Global Maps of 21st-Century Forest Cover Change.” Science 342(6160): 850–853. Lambin, E. F. 1999. “Monitoring Forest Degradation in Tropical Regions by Remote Sensing: Some Methodological Issues.” Global Ecology and Biogeography 8 (3–4): 191–198. http://doi.wiley.com/10.1046/j.1365-2699.1999.00123.x. Mitchard, E., Viergever, K., Morel, V., and Tipper, R. 2015. Assessment of the Accuracy of University of Maryland (Hansen et al.) Forest Loss Data in 2 ICF Project Areas – Component of a Project That Tested an ICF Indicator Methodology. https://ecometrica.com/wp- content/uploads/2015/08/UMD_accuracy_assessment_website_report_Final.pdf. Mwaura, F. M., and A. Adong. 2016. “Determinants of Households’ Land Allocation for Crop Production in Uganda.” Journal of Sustainable Development 9 (5): 229-246. 41 RCMRD (Regional Centre for Mapping of Resources for Development). 2015. Digital Elevation Model (DEM) from Shuttle Radar Topography Mission (SRTM). Available online: http://geoportal.rcmrd.org/layers/servir%3Auganda_srtm30meters Silveira, E. M. O., I. T. Bueno, F. W. Acerbi-Junior, J. M. Mello, J. R. S. Scolforo, and M. A. Wulder. 2018. Using Spatial Features to Reduce the Impact of Seasonality for Detecting Tropical Forest Changes from Landsat Time Series. Remote Sensing 2018 (10): 808: doi:10.3390/rs10060808. Tyukavina, A., A. Baccini, M. C. Hansen, P. V. Potapov, S. V. Stehman, R. A. Houghton, A. M. Krylov, S. A. Turubanova, and S. J. Goetz. 2015. “Aboveground Carbon Loss in Natural and Managed Tropical Forests from 2000 to 2012.” Environmental Research Letters 10 (7): 074002. http://iopscience.iop.org/article/10.1088/1748- 9326/10/7/074002/meta;jsessionid=B9B73099EC8D83B90613AB8104E7CD66.c1. Turyahabwe, N., J. G. Agea, M. Tweheyo, and S. B. Tumwebaze. 2012. Collaborative Forest Management in Uganda: Benefits, Implementation Challenges. College of Agricultural and Environmental Sciences, Makerere University, Kampala. UBOS (Uganda Bureau of Statistics). 2017. Uganda National Household Survey 2016/17. Kampala, Uganda Bureau of Statistics (UBOS). UBOS. (2018). Uganda National Household Survey 2016/2017. Kampala, Uganda; UBOS. Available at: https://www.ubos.org/wp-content/uploads/publications/03_20182016_UNHS_FINAL_REPORT.pdf UNEP-WCMC. 2016. World Database on Protected Areas User Manual 1.4. UNEP-WCMC: Cambridge, UK. Available at: http://wcmc.io/WDPA_Manual. UNHCR (United Nations High Commissioner for Refugees). 2017a. Livelihoods Socio-economic Assessment in the Refugee Hosting Districts. Reev Consult International, Kampala. ———. 2017b. Livelihoods Strategy. UNHCR Uganda Operation 2017-2020, Kampala. ———. 2018. Uganda Country Refugee Response Plan. The Integrated Response Plan for Refugees from South Sudan, Burundi and the Democratic Republic of the Congo. UNHCR, Regional Refugee Coordination Office (RRC), Nairobi, Kenya. Verbesselt, J., R. Hyndman, G. Newnham, and D. Culvenor. 2010. “Detecting Trends and Seasonal Changes in Satellite Image Time Series.” Remote Sensing of Environment 2010 (114): 106–115. 42 Annex 1: Methodologies Woodfuel data collection and analysis A woodfuel demand assessment was conducted in March 2018 by an FAO team supported by OPM representatives and four enumerators. A quantitative household questionnaire (see annex 2) and qualitative interviews in the refugee and host communities generated information on energy consumption for cooking, average time spent by households to collect fuelwood, types of cooking system used, associated challenges, and related livelihood issues. The survey was implemented in 174 refugee households in Bidibidi settlement (Yumbe) and Maaji settlement (Adjumani), as well as in 168 households in host communities in Ciforo (Adjumani) and Okangali (Yumbe) subcounties. Data from these locations were extrapolated across the other target refugee settlements in northern Uganda. The selection of the sample sites took into account the establishment date of the settlements. Maaji is part of the group of settlements dating back to 1997, while Bidibidi is one of the settlements established as a result of the new refugee influx after 2014. In addition, the team agreed to return to Bidibidi for monitoring the trends of data from the earlier woodfuel assessment conducted in the same settlement by FAO and the UNHCR in 2017. The analysis considered that the other settlements present similar characteristics in terms of woodfuel consumption with no significant invalidation of data collected in the selected sample sites. The enumerators were pretrained on questionnaire implementation and the use of a weighing scale to measure firewood and charcoal consumption. Data collection was guided and supervised by FAO and OPM staff. Systematic sampling was employed for the selection of households, selecting every tenth households in each location. Key informant interviews were also carried out by the field supervisor on specific areas of interest. Additional observations were made daily and shared with the field supervisor. They included differences between refugee and host communities in income sources, foods, and cookstove use. Data collection also included photographs, some of which are shared in this report. The team was able to perform a quality check of completed questionnaires before leaving the sampled locations. Overall checking of the questionnaires was carried out by the FAO field supervisor. The quantitative analysis was integrated with data from qualitative data and document review. Biophysical field inventory Biophysical field data were collected to estimate biomass stocks for the different strata described in the scoping report and adjusted (as explained below) into five classes: woodland, bushland, cropland, woodland depleted, and bushland depleted. The latter two classes were created by overlaying loss/depletion layers on existing LULC maps developed by the NFA. For this study, ‘intact’ refers to those areas within the bushland and woodland classes where BFAST did not indicate change. Since the focus of the assessment was on LULC classes with potential woodfuel resources, grasslands were not considered as they contain very low AGB. However, field results show that an overwhelming majority of the plots expected to fall within the degraded bushland class were classified by enumerators as ‘grassland’. Given that grassland is considered to be the resulting class of bushland that is slowly degraded over time, its biomass expansion factor was therefore used to calculate the AGB of the degraded bushland class. Originally planned to be included in the biophysical survey, tropical high forests (THFs) were ultimately excluded as their location was found to be too far for refugees to access, situated 10 km south of the Maaji settlements (Adjumani District). Furthermore, this stand represents the only stand of THF in the AOI and is intact because it falls within the Zoka Central Forest Reserve, considered off-limits for fuelwood collection. Biophysical data were gathered from plots located in two preidentified sampling zones (as described in the intermediate report) and used to estimate the AGB stock for the selected LULC classes. Hotspot 1 covered a heavily 43 affected area located between the three refugee settlements of Bidibidi, Rhino, and Imvepi, which together host more than 380,000 refugees.13 The dominant land use is subsistence cropland and grassland, with remnants of previously intact woodland. Hotspot 2 was on the opposite bank of the Nile amidst dense woodland vegetation some 10 km south of the Maaji settlements, albeit fragmented due to agricultural expansion. The distribution of woody biomass was mapped using remote sensing, and stock changes within the AOI were measured for 2010–2013 (‘before South Sudan crisis’) and 2014–2018 (‘after South Sudan crisis’). Plot sampling approach A statistical stratified sampling design was adopted, with 95 plots spread between Hotspot 1 and Hotspot 2 (Figure 11 and 12). Plot allocation targeted an equal distribution across classes (15 plots per class regardless of the area proportion14) and ensured that rare classes (in particular degraded woodland and degraded bushland) were well represented. A preassessment of the plots was carried out using Collect Earth to validate their land cover type and the actual presence of degradation (for the depleted land cover classes) and to reach the target sample number for each stratum. A total of 67 out of 95 plots were measured in the field. The majority of the plots south of the Maaji settlements turned out to be inaccessible and could not therefore be measured. Figure 11. Area 1: Plot allocations on 2015 LULC map 13 Population data from OPM-UNHCR, April, 2019: Bidibidi 225,808; Imvepi 57,758; Rhino 104,912. Rhino extension - Omugo (24,533) is not directly within the hotspot. 14 This is also because the team would end up having more plots in cropland and due to limitations in time and human resources. 44 Figure 12. Area 2: Plot allocations on 2015 LULC map Note: The boundaries, names and designations on the above maps (Fig. 11-12) do not imply official endorsement or acceptance by the United Nations. Sources: UNHCR Settlements extents, Border crossing, Villages/towns locations. Protected areas: UNEP-WCMC (2016). World Database on Protected Areas User Manual 1.4. UNEP-WCMC: Cambridge, UK. Available at: http://wcmc.io/WDPA_Manual Plot design and field data collection At each sampling location, a circular plot of 0.05 ha (12.6 m radius) was established. The plot size in cropland was increased by 0.1 ha (18 m radius) due to high variability of tree biomass in cropland attributable to sparse distribution of trees. Within each plot (Figure 13), sub-plots of 4 m radius were measured to capture the biomass of small trees and shrubs, which are popular sizes used for firewood. 45 Figure 13. Plot design for the biophysical inventory Source: FAO & UNHCR, 2017 Within the first quadrant of the plot (between points 2 and 3), shrubs were measured (including basal diameter, crown diameter and average height, and number of stems [in the case of clustered shrubs]). All standing trees (live and dead) of at least 3 cm diameter at breast height (DBH) were also measured in the first quadrant. In the rest of the plot, the minimum measured DBH was 5 cm. Other tree parameters recorded were species and total height. Within a smaller radius of 4 m (giving a circle of 0.01 ha), all saplings and deadwood were measured. Four photographs were taken in the cardinal directions and the following variables were also recorded: • Land use • Major LULC type(s) of the surrounding area • Degradation indicators such as o Fire evidence; o Grazing intensity; o Vegetation cover; and o Number of stumps. The biophysical data were collected by members of the NFA inventory team between March 13 and April 5, 2018, and were recorded on tablets using Open Foris Collect Mobile, an Android App for fast, intuitive, and flexible data collection for field-based surveys.15 Enumerators classified the vegetation into categories associated with those within the LULC maps: woodland (closed/open), planted forests, bushland/shrubland, grassland, cultivated land, bare/open land, built-up, water 15 http://www.openforis.org/tools/collect-mobile.html. 46 body, THF depleted, and so on. For the rapid assessment, the most common and most likely to be accessed LULCs in the AOI were grouped into five strata and analyzed: woodlands (intact and depleted), bushlands (intact and depleted), and croplands. Overlaying the LULC map with Hansen and time series analysis of overstory loss allowed for the creation of the depleted woodland and depleted bushland strata. The remaining ‘THFs depleted’ in the area (Hotspot 2), located in the southern part of Adjumani District, 10 km south of the Maaji settlements, were considered too far for refugees to walk for fuelwood collection. Furthermore, they are located in the Zoka Central Forest Reserve, considered off-limits for fuelwood collection. For this reason, THFs were not included in the biomass analysis. Estimating biomass stocks Only AGB (in trees and shrubs) and deadwood were targeted in this study, as they represent the primary source of woodfuel for refugees and local people. In each plot, AGB was calculated using the allometric equations of Chave et al. (2014), which were also used in Uganda’s NBS. R scripts developed for REDD+16 in the NBS were used to estimate stocks. Plot-level results were aggregated into LULC classes, as assigned to plots during the field inventory. Shrub biomass was estimated using the NBS equation for small trees. To categorize field sites as degraded or intact, indicators such as number of stumps, presence of fire, erosion, or grazing woodlands were captured. Those plots with any number of these indicators were considered to be degraded. To estimate the biomass of the intact areas, NBS data from the West Nile region were ultimately used, as the field crew experienced problems accessing many of the intact rapid assessment sites. Data for closed woodlands measured by the NBS between 1998 and 1999 were reviewed to indicate the average standing stock of woodlands in the region before the impact of degradation. Average annual biomass increments were also obtained from the NBS, which provides these for the various LULC classes in each of Uganda’s agroecological zones (Forest Department 2002). The target area in northern Uganda is in the semi-moist lowland zone. Remote sensing analysis Datasets used • A DEM (RCMRD 2015)17 was used to compute slope in the AOI. • LULC maps for 2010 and 2015 were used to describe the hotspot areas and to derive changes and degradation in woody vegetation classes (in this case, woodland, bushland and cropland). The maps were vector-based but were converted to pixel-based products to facilitate interpolation with other datasets. • Landsat time series imagery was analyzed using BFAST (2010), to detect where and with what intensity changes have occurred. BFAST enables per-pixel detection of the date and magnitude of change over time. Overlaying BFAST results with LULC maps indicated where changes had occurred since the 2015 LULC.18 The use of Landsat’s dense time series imagery has been demonstrated for mapping land cover changes, such as deforestation, forest degradation, and impact of fire (Silveira et al. 2018). 16 REDD+ = Reducing Emissions from Deforestation and Forest Degradation. 17 The data represent the 30 m DEM from the SRTM http://geoportal.rcmrd.org/layers/servir%3Auganda_srtm30meters. 18 2015 LULC map is used as ‘proxy’ of the 2014 situation in the area. 47 • The GFC dataset (Hansen et al. 2013) was used to compute statistics on tree cover loss as a first analysis of the trends in tree cover using existing products, which currently cover 2001–2016. Tree cover loss is defined as complete over-story removal occurring on land with at least 10 percent initial tree cover in 2000 (the only available GFC dataset that includes tree cover percentage). Therefore, statistics were computed using tree canopy cover for 2000 (defined as “canopy closure for all vegetation taller than 5 m in height, as percentage per output grid cell”19) subtracting the loss (defined as a “stand-replacement disturbance, or a change from a forest to non-forest state”) observed in the selected periods. Nevertheless, this dataset could partially depict the real situation since only tree cover changes are considered and sparse tree cover (lower than 10 percent)20 was excluded. This could therefore confirm the accuracy of the results in detecting changes when tree cover is higher than 20 percent (Hansen et al. 2013) and that human-affected areas could be characterized by other vegetation forms (that is, shrubs). • Population data at 1 x 1 km resolution from Columbia University’s Connectivity Lab and Center for International Earth Science Information (CIESIN 2016)21 was used to estimate the host population in the 5 km buffers around each settlement, as an indication of total population to compute estimates for each settlement. Classification and change detection To provide spatially and temporally explicit information on biomass and changes in biomass over time, a time series approach was used incorporating available Landsat satellite imagery from the U.S. Geological Survey. The Landsat sensor records the electromagnetic reflectance of Earth’s surface in multiple wavelengths at a spatial resolution of 30 x 30 m. This reflectance information can be used to determine land surface biophysical characteristics, such as vegetation type. Changes in vegetation result in correlate changes in the detected reflectance. Time series analysis, in which every available satellite image acquired over the study area was analyzed, enabled tracking the reflectance over a long period to detect both subtle and unambiguous changes on the land surface. In the case of this study, subtle negative changes correspond to land surface degradation, and strong negative changes correspond to complete overstory removal. All processing was carried out in the FAO System for Earth Observation Data Access, Processing & Analysis for Land Monitoring (SEPAL)22 platform. The results of the BFAST23 algorithm (DeVries et al. 2016 Dutrieux et al. 2015; Verbesselt et al. 2010) (reclassified into loss and degradation maps for the two periods of interest, 2010–2013 and 2014–2018) were overlaid to the LULC maps (NFA data24) (2010 and 2015 map for the two periods, respectively) to know in which land cover types (that is, woodland, bushland, and cropland) changes occurred. LULC maps for 2010 and 2015 were reclassified and the 13 land cover classes identified during the initial scoping work were reduced to just four based on their prominence in the landscape, accessibility, and biomass content: 1. woodland, 2. bushland, 3. cropland, and 4. other. The classes of the land cover maps were combined with the two classes of the change maps (loss and degradation) as per the matrix in Table 25. In more detail, ‘intact woodland’ and ‘intact bushland’ are vegetated areas that remain ‘stable’, without degradation and loss. Degraded classes refer to a partial removal of vegetation while loss 19 https://earthenginepartners.appspot.com/science-2013-global-forest/download_v1.4.html. 20 Tree cover in 2000, defined as canopy closure for all vegetation taller than 5 m. Encoded as a percentage per output grid cell, in the range of 0 to 100. 21 Center for International Earth Science Information Network. Columbia University. High Resolution Settlement Layer. Source Imagery 2016 DigitalGlobe, Inc. www.ciesin.columbia.edu/data/hrsl/. 22 https://sepal.io/. 23 For more information on BFAST: http://bfast.r-forge.r-project.org/. 24 http://redd.unfccc.int/files/annex_8_mapaa_methodologyresults_ug_frl_1_.pdf. 48 occurs when there is complete vegetation removal. For these last classes, woody biomass is assumed to be zero. The maps for the two periods (‘’before arrival’) and (‘after arrival’) are shown in Figure 9 and Figure 10. Table 25. Matrix of resulted classes obtained by combining the LULC map with the mask of loss and degradation MAP Combination Degradation and loss classes Original map Reclassified LULC classes Loss Degradation No change code code Plantations and woodlots—deciduous Loss in Degraded Intact 1 1 trees/broadleaves (‘hardwood’) woodland woodland woodland Loss in Degraded Intact 2 1 Plantations and woodlots—coniferous trees woodland woodland woodland 3 6 THF—normally stocked Loss in other Other Other 4 6 THF—depleted/encroached Loss in other Other Other Woodland—trees and shrubs Loss in Degraded Intact 5 1 (average height > 4m) woodland woodland woodland Bushland—bush, thickets, scrub Loss in Degraded Intact 6 2 (average height < 4m) bushland bushland bushland Grassland—rangelands, pastureland, open 7 6 savannah; may include scattered trees shrubs, Loss in other Other Other scrubs, and thickets. Wetlands – wetland vegetation; swamp areas, 8 6 Loss in other Other Other papyrus and other sedges Subsistence farmland – mixed farmland, 9 3 smallholdings in use or recently used, with or Loss in other Cropland Cropland without trees Uniform commercial farmland—mono-cropped, 10 6 non-seasonal farmland usually without any trees Loss in other Other Other for example tea and sugar estates 11 6 Built up area—urban or rural built-up areas Loss in other Other Other 12 6 Open water—lakes, rivers, and ponds. Loss in other Other Other 13 6 Impediments (bare rocks and soils) Loss in other Other Other Note: The map codes are as follows: loss in woodland (10), degraded woodland (11), intact woodland (1), intact bushland (2), loss in bushland (20), degraded bushland (22), cropland (3), loss in other (9), and other (6). Code of the BFAST loss/degradation map are as follows: loss (1), degradation (2), and no change (0). On the left are the ‘original code’ and ‘reclassified code’ of the LULC maps. From this, it is possible to know how the original 13 map classes were reclassified. The BFAST methodology tracks a single vegetation index, the Normalized Difference Moisture Index (NDMI), through time to detect both unambiguous and subtle changes in vegetation cover. It requires several parameters to be set to define the scope of the analysis, including the time over which the analysis will be carried out, the historical period defining an expected behavior for each pixel, and a monitoring period indicating ‘from’ and ‘to’ dates for detecting any deviations (breaks) from ‘normal’ pixel behavior. Therefore, breaks can be considered the variations from the seasonal patterns, as a result of either abrupt changes (for example, deforestation, fires) or more gradual changes (for example, encroachment, gradual land degradation). The advantages of using indexes rather than original band observations include minimizing the soil and other background effects, providing a degree of standardization for comparison, and enhancing the vegetation signal (Silveira et al. 2018). BFAST time series analysis was performed for two time periods, 2010–2013 and 2014–2018. The parameters used for this analysis were as follows: 49 • For the changes between 2010 and 2013 o Beginning of historical period: January 1, 2005 o Beginning of monitoring period: January 1, 2010 o End of monitoring period: December 31, 2013 • For the changes between 2014 and 2018 o Beginning of historical period: January 1, 2010 o Beginning of monitoring period: January 1, 2014 o End of monitoring period: April 16, 2018 The output of the time series analysis is ‘magnitude’ of change. Magnitude can vary from strongly negative (for example, deforestation) to strongly positive (for example, reforestation or revegetation). Classification of magnitude values requires creating thresholds to distinguish change classes and create classes capable of being summarized and mapped. To relate ‘magnitude’ values obtained in the analysis with on-the-ground change, the results need to be calibrated based on reliable data. Results in this study were calibrated with field-based observations and very high spatial resolution imagery from Google Earth and Worldview2, 3 and GeoEye1 imagery provided by the United Nations Institute for Training and Research (UNITAR) and using the socioeconomic information on consumption. The processing generated a three-band raster dataset covering the AOI, where the date of break and the magnitude of detected change are recorded for each pixel (band 1 and band 2 of the resulting output). To identify the changes within the AOI, the layer of change magnitude was used. This is computed as the median residual (‘difference or distance’) between the predicted and observed values within the monitoring period. According to the different intensities of change, (very) large negative changes were used as proxy for complete tree cover loss and medium negative changes used as potential areas for degradation. The final results were further calibrated based on the socioeconomic results. The time series Landsat data were created automatically in the SEPAL platform. SEPAL was also used for the processing of the algorithm itself. The computer-intensive process analyzed about 980 Landsat images relating to the AOI (Figure 14). The validation of the maps was carried out using field data and the very high spatial resolution imagery Digitalglobe25 satellite images provided by UNITAR. Figure 14. Number of satellite images (Landsat 7 ETM+) of the time series for both periods 25 https://discover.digitalglobe.com/ 50 Technical considerations This section explains some of the technical complications involved in this study and helps explain the discrepancies between the biomass consumption estimates derived from the household study with those obtained from the remote sensing-based analysis in which area deforested and degraded was multiplied by a biomass expansion factor. Differences between the estimates derived from the remote sensing-based analysis and the household study can likely be ascribed to the following reasons: • Inaccuracy of the LULC maps (2010 and 2015 maps). Even though the LULC national maps are the result of intensive work carried out by the NFA, the application of vector-based products over pixel-based change maps may compound errors, given the probable map errors. It is generally discouraged to combine datasets of different types (vector versus raster) and different spatial accuracy. Furthermore, land cover maps utilized for the study are national scale maps and not intended to be used on a subnational basis. However, due to limited time, using existing and endorsed national products was considered the best approach. • Definitions and Land Cover Classification System used. The classification system and associated definitions of woodland, bushland, and cropland are those adopted by GoU in its national mapping activities. However, this classification is rather complex: for example, land cover classes with a tree cover component in the LULC maps for the AOI include THF depleted, woodland, bushland, grassland, subsistence farmland, savannah, and wetland. Furthermore, distinction between bushland and woodland is rather difficult to assess in remote sensing because the height of the objects in the imagery is unknown. • Assumption of absolute loss. For pixels classified as loss, biomass was set to zero. In other words, the assumption for the sake of the study was that there is no remaining biomass after overstory removal, when in reality there is partial loss. For example, inside the settlements (and where loss was mapped out) there is still scattered vegetation. • Inaccuracy of the biomass factors applied for each LULC class selected. Due to the limited time and resources for a more in-depth assessment, only 67 sample plots were surveyed in the field to derive the biomass expansion factor. This meant some rather high margins of error. For example, AGB on degraded woodlands was found to be 25.3 t/ha ± 18.5, meaning that the AGB for this LULC could be anywhere from 6.8 to 43.8 t per ha. Variability would be decreased if there were a larger number of plots surveyed in this class. These estimates were then expanded over the vector data for the respective LULCs. • Changes in grassland were not considered because of low biomass for wood fuel collection. Grassland is one of the major classes in Bidibidi, but grasslands were not considered in the study because they have very low AGB and therefore are unlikely to meet the fuelwood needs of both the host and refugee communities. However, misclassifications on the LULC map are possible given that it was produced as a national product. Therefore, it is possible that some areas where fuelwood collection is indeed occurring were omitted. • Validation of remote sensing findings with field data. Discrepancies found between data collected on the ground and those used in the remote sensing analysis could partly be related to the different spatial resolution of the two sources (spatial resolution of the field plots versus the spatial resolution of the images) and GPS measurements errors (that is, how precise the instrument was able to collect the coordinates for that plot). In addition, the interpretation of the land cover features during the field data collection should be in line with the interpretation of the very high spatial resolution imagery and the data collection phase. 51 • The data presented in the socioeconomic findings might present some deviations resulting from using the indicator fuelwood consumption per person per day assessed in the households sampled to then extrapolate the total woodfuel demand of the whole refugee settlements in the AOI. Overall, loss changes were mapped with higher certainty with respect to the changes classified as ‘degradation’ (especially in bushland), which were spatially diffused around the AOI. Difficulties were found in discriminating real changes from soil moisture changes, especially in croplands. BFAST is a relatively new approach to assessing forest degradation and is continually being improved. Making distinctions between vegetation cover changes and degradation processes is problematic when dealing with complex landscapes and change processes. Characterizing a disturbance event is complicated by the fact that deforestation is preceded by several years of forest degradation when driven by subsistence agriculture (DeVries et al. 2016). Due to the complexity of the area, as suggested by Lambin (1999), a “more practical definition of degradation would be a continuous measurable value (for example, in terms of canopy cover).” DeVries et al. (2016) well explain implications of using definitions based on area, height, and canopy cover thresholds.26 Therefore, the use of classes such as tree cover and shrub cover percentage could be a preferred option to obtain estimates of biomass stock and changes without using the LULC classification system and its associated classification errors. Indeed, extending the analysis to grassland would require more time and a separate assessment most likely using satellite images with better spatial resolution, covering the periods of the analysis and further field data collection. It is therefore important to underline that estimates presented from the remote sensing analysis may provide an overview of the lands prone to degradation and further natural resources exploitation, and the reasons provided above may not completely reflect the socioeconomic findings. 26 For example, the Inter-governmental Panel on Climate Change (IPCC) defines degradation as changes negatively affecting carbon stocks in forests which remain forests, where a forest is defined based on area, height, and canopy cover thresholds. Degradation can occur when a forest is completely cleared, but the total area cleared is less than the area threshold (that is, 0.5 ha). Degradation can alternatively occur when a larger area of forest experiences negative changes in forest canopy cover, but the canopy fraction still remains above a defined forest threshold (for example, 20 percent). Finally, using the area-based definition implies the evaluation of the total area affected from the disturbance (DeVries et al. 2016). 52 Annex 2: Rapid Woodfuel Demand Questionnaire Country: Settlement or village: District: Block number: Name of Enumerator: Plot number: Date: Before starting the interview: • Begin the session by explaining the format and objectives of the interview. • Ensure the interviewee can choose in advance not to participate if they are uncomfortable in any way. • Specify that confidentiality will be maintained at all times. Thus, no record will be kept of participants’ names. 1. HOUSEHOLD INFORMATION How many structures on the plot? Date of arrival to the camp Walling material Roofing material (e.g. Type of structure on the plot: (e.g. wood poles, straw, bamboo, iron sheet) mud brick) House -o How many rooms? Kitchen hut -o Latrine -o Animal shed - o Other -o Fencing -o Interviewee age: Relationship with the household (head, spouse, son, daughter, other): Head of household gender: Interviewee gender: 1-o Male 2-o Female 1-o Male 2-o Female Total number of household members: Number of income/wage earners in the household: Number of adults > 18 years Number of children (2-18 years) Number of infants < 2 years Male Female Male Female Male Female 53 Current livelihood category Current status of household Agriculture=A, Agropastoral=AP, Pastoral=P, Fishing =F A AP P F Other (specify) IDP Refugee Returnee Host What are the current sources of household income? Exchange or Selling No income Cash transfers Remittances Other sale of food firewood If other, specify income source and earner: What kind of income generation activity would you like to undertake? 2. CURRENT SOURCES OF FUEL FOR COOKING, HEATING WATER What fuels do you use for cooking and heating in the household? Crop residues Firewood Charcoal Grass/straw Animal dung Other fuel (specify) If ‘other fuel’, give details: What is the usual quantity of fuel you consume per day in the household? Main uses: C=cooking; H=heating; AG=agricultural uses (e.g. curing Measured quantity tobacco, drying food, etc.); CM= commercial uses (e.g. baking bread, Fuel type (kg/day) brewing alcohol, making food for selling). Type of firewood harvested: DW= deadwood; GW= green wood Firewood Charcoal Grass/Straw Crop residues Animal dung Other If ‘Other’, specify daily household consumption: 54 Where do you source your fuel? (multiple responses allowed) Collect from Collect UN or NGO Collect from Collect from Buy from the natural from shrub Others distribution woodlots farmland market forest land If ‘Others’, specify the source: If you collect firewood, how many headloads per week are gathered by people in this household? 1 2 3 4 5 6 or more Who within your household is primarily responsible for collecting fuel? (insert number for each box) Male adult Female adult Female child Male child Other If ‘Other’, please specify: How many hours does the total collection trip take in average? (Going from your house to the main collecting area, cut and collect firewood and back) What challenges are you facing during collection of firewood? 3. TECHNOLOGIES FOR COOKING AND HEATING What method/ stoves are currently used for cooking and heating? (if more than one type of stove is observed, tick multiple boxes) Three Mudstove Mudstove Ceramic Ceramic Metal stove Metal stove Others stone fire (firewood) (charcoal) (firewood) (charcoal) (firewood) (charcoal) If ‘Others’, specify which stoves: Where is the stove(s) located? (multiple responses allowed) In a dedicated In a room used also for In the In a separate Outdoors kitchen sleeping living area building/structure Why do you like to cook with this cooking system? 55 Does it have any disadvantages? (Note for the enumerator: response choices should not be read, tick all that apply) Food is Expensive to use Too much smoke It requires a lot of fuel Other undercooked because of fuel costs If ‘Other’, specify these disadvantages: If you currently have a stove, where did you get it from? Market NGO/UN org. Self-produced Relatives Other If ‘Other’, specify the source: What is the main cooking technology you would prefer to use if you had a choice? Three Mudstove Mudstove Ceramic Ceramic Metal Metal stone Others (firewood) (charcoal) (firewood) (charcoal) (firewood) (charcoal) fire If ‘Others’, specify preference: 4. TYPE OF FOOD AND PREPARATION What are the main types of food usually cooked? Indicate the typical method of cooking for each food and how many times per week it is prepared. Type of food Method of cooking (e.g. Boiling, Stewing, Roasting, Times of preparation in a Frying, Baking, Dried food) week ………………………………………… ………………………………………………………………………… ……………………………………… ………………………………………… ………………………………………………………………………… ……………………………………… ………………………………………… ………………………………………………………………………… ……………………………………… 56