**2. Materials and Methods**

#### *2.1. Study Area*

Madagascar is the fourth largest island on Earth. It is situated 300 km east of the African coast and has an area of 587,295 km2. The island has six large basins, divided into 32 macrobasins and 533 subbasins, all of which are distributed across 22 regions [7]. The island can be divided into four climatic ecoregions with four forest types: moist in the East, dry in the West, spiny in the South, and mangrove forests on the west coast [8]. Forest types are defined based on their inclusion in one of these four ecoregions. Madagascar's geographical position coupled with the island's irregular physical relief makes its climate extremely varied, which could be more accurately described as several climates differing by region. On the east coast, the climate is hot and humid, with the annual rainfall of 1100–3700 mm per year and the average temperature between 23 and 26 ◦C. The North and northwest regions have a tropical climate, with monsoon conditions driving rainfall in the summer. On the contrary, the southwest part is semiarid, with the annual rainfall of 500–700 mm per year. On the west coast, the climate is tropical, with a hot and dry summer. The annual average temperature varies between 24 and 27 ◦C. An interannual variation in temperature and precipitation is observed in the central highlands, with the annual rainfall of 900–1500 mm per year and the annual temperature range from 16 to 22 ◦C [12,13]. Previous studies pointed out that the annual rainfall decreases from 1500 to 400 mm per year from north to south across the west coast while the temperatures increased by 0.2 ◦C over northern Madagascar and by 0.1 ◦C over southern Madagascar [10,12].

This study focuses on the 12 major river basins that cover areas larger than 10,000 km2: Mangoky, Betsiboka, Tsiribihina, Mananara, Mangoro, Maningory, Mahajamba, Onilahy, Manambolo, Mahavavy, Sofia, and Mandrare (Figure 1).

**Figure 1.** Delineation of catchment boundaries and ecoregions.

#### *2.2. Methods*

Initially, we compared the river basin management in Madagascar and Japan by evaluating the existing framework documents in terms of the effects of climate change, and then estimated the impacts of climate change on the major river basins. The island country of Japan was chosen for comparison because of two reasons. The first is similarities between Japan and Madagascar. Both countries are islands with a diverse natural environment, prone to natural disasters, suffer from the impact of climate change, but have abundant water resources in contrast to many other island countries [14,15]. Japan and Madagascar rank first and fourth, respectively, among the ten countries most vulnerable to extreme weather events in 2018 [9]. The second reason is Japan's well-known experience in water management, as well as mitigation of climate change and water-related disasters. Japan is among the countries with higher levels of IRBM implementation. This view is supported by the findings in a comparative study on river basin management in Japan and other island countries [16,17]. Indeed, the traditional water wisdom of Japan enabled the achievement of its Millennium Development Goals in 2015; currently, Japan is a global leader in water technology developed by private companies [18]. In this paper, we focused on water availability and demand issues on a large river basin scale. The analysis of these issues is an essential component for water resources assessment, which is, therefore, a significant component of the evidence and analysis required for IWRM [19]. Unlike most other countries on the African continent, no river basin management plan has been completed by Madagascar. Therefore, the next step of the research was to investigate the future changes in water stress, water supply, and water demand for the major river basins. The data were obtained from Aqueduct projections using Coupled Model Intercomparison Project Phase 5 General Circulation Models provided by the World Resources Institute [20]. The Aqueduct Project is a data platform run by the World Resources Institute, an environmental research organization (Washington, DC, USA). It provides a global water risk atlas to help companies, governments, and civil society understand and respond to water risks—such

as water stress, variability from season to season, pollution, and water access. It intends to measure, map, and understand water risks around the globe. The Aqueduct Project is widely used by researchers across the globe because it uses open-source, peer-reviewed data to map water risks and collaborate with companies, governments, and research partners through the Aqueduct Alliance [21]. Projections of climate variables were driven primarily by the CMIP5 Project, and socioeconomic variables were based on the Shared Socioeconomic Pathways database from the International Institute for Applied Systems Analysis. The project computed water supply from the runoff values extracted from an ensemble of CMIP5 data. Herein, the total blue water or renewable surface water is used as an indicator of water supply. The projected change in the total blue water (the renewable surface water) is equal to the 21-year mean around the target year divided by the baseline period of 1950–2010. The data used by the World Resources Institute (WRI) to calculate the baseline include over 50 years of data across several indicators reported by country (FAO) to 2010. Water demand is measured as the sum of water withdrawals. The projected change in water withdrawal is equal to the total withdrawals in the target year divided by the baseline year of 2010.

Water withdrawals were modeled from the projected size, wealth, and other characteristics of the countries, for each of the three sectors as defined by the Food and Agriculture Organization of the United Nations (FAO): agricultural, industrial, and domestic. Water stress is an indicator of competition for water resources; it is informally defined as a societal demand for water, divided by available water. Water stress was computed as the ratio of water withdrawals to the available blue water on the average annual basis. We produced maps showing the projected changes in the water stress, water supply, and water demand from the baseline (1950–2010) to the future using geographic information systems and Aqueduct future-value data for the year 2040 under the Shared Socioeconomic Pathway 2 and Representative Concentration Pathway 8.5 scenarios. We also compared the future water stress scores of Madagascar and Japan using Aqueduct country and province ranking data (2020, 2030, and 2040). Higher scores on the scale from 0 to 5 correspond to greater competition among water users relative to the available surface water resources. All data are available at https://www.wri.org/aqueduct (accessed on 25 May 2020). Finally, an alternative solution for effective IWRM implementation and sustainable river basin management was considered based on the above comparisons and projections.
