Drought and households ’ adaptive capacity to water scarcity in Kasali, Uganda

The world is experiencing variability in precipitation, increased temperature, drought frequencies and intensities. Globally, approximately four billion individuals experience water scarcity due to drought. In Uganda about 10% of the population in the southern and northern parts of the country experience drought related water scarcity annually. This study aimed at assessing drought and households ’ adaptive capacity (AC) to water scarcity during drought in Kasali. This was done through determining drought trends from 1987 to 2017, assessing the impact of drought on water availability and the AC of households to manage water scarcity. Droughts were assessed based on the Reconnaissance Drought Index (RDI). The results show a decrease in the average annual rainfall, and the seasons of March-April-May (MAM), January-February (JF) while the seasons of September-October-November-December (SOND) and June-July-August (JJA) show an increase in rainfall trend. The average maximum and minimum annual and seasonal temperature increased signi ﬁ cantly by between 0.56 and 1.51 (cid:1) C. The minimum temperature increased more than the maximum temperature. Kasali experienced one extreme dry year and four moderate ones between 1987 and 2017. Above 70% of the households spend longer hours collecting water during dry years than wet years. The AC of households to water scarcity was low and drought negatively impacted water availability.


GRAPHICAL ABSTRACT INTRODUCTION
Increased temperature on the Earth's surface leads to changes in frequency and distribution of rainfall and droughts (Ghebrezgabher et al. ). Drought is an atmospheric event associated with insufficient availability of water for a long time in a given geographical area (Shamshirband et al. ). Drought is a complex climate change risk whose intensity and severity has risen globally in the past two decades causing adverse impacts on water resources, vegetation, people and their livelihoods (Gleick ; Masih et al. ). For instance, between 2002 and 2010, Australia experienced an extended period of droughts which affected ecosystems in its largest river system, Victoria and Murray-Darling Basin (MDB), and left thousands of people grappling with water scarcity (Leblanc et al. ). This drought was believed to have been the main cause of global apparent reversal in intensification of the water cycle that was observed in the subsequent years (Huntington ). Several studies have shown droughts in all the continents, with Africa having the highest number of drought events (Masih et al. ; EM-DAT ).
In Africa, The Greater Horn of Africa (GHA) and southern Africa regions were severely affected by droughts in the last one and half decades (Olufemi ). The GHA has experienced multiple droughts with varying onsets, duration and intensities. For example between 2009 and 2011 the region suffered the most severe drought due to its duration and intensity (FAO ). Climate change is undoubtedly becoming serious in the sub-Saharan African countries with studies showing increased annual temperatures and reduced mean annual precipitations in the GHA (Amjath et al. ). Drought has been severe in this region since the 1970s, and it has resulted in crises for millions of people together with their livestock and the situation has worsened due to climate variability and change (AfDB ).
The impact of drought on the hydrological system is less known. This is because of the human and physical feedback loops influencing the behaviour of the resources, being more at a local than a regional scale (Ghebrezgabher et al. ).
This remains a pressing issue which has to be studied because some adaptation strategies to water scarcity may result in mal-adaptation (Olufemi ). Drought has increased water scarcity in sub-Saharan Africa (Olufemi ), for example in early 2018 the city of Cape Town was hit by one of the worst drought periods in its history which led to water scarcity. During the last months of 2017 in Cape Town, the water reservoirs in the dams dwindled to so low a level that approximately 3.7 million metro area residents of Cape Town city risked experiencing a 'Day zero' situation where residents' taps would be switched off as a way of conserving water (Wolski ).
In Uganda, water scarcity due to drought poses the greatest threat to the rural population. Around 4.5 million people (or 10% of the population) are affected by water scarcity each year, mainly in the south-eastern and north-eastern regions of Uganda (World Bank ). Droughts in Uganda are a recurrent hazard and the country experienced notable drought events in the years 1967, 1979, 1987, 1999, 2002, 2005, 2008, 2010and 2017. The severe droughts affected many water resources in the country (NEMA ). Climate projections indicate that conditions will become even more severe in the 21st century (Hertel & Rosch ). Therefore, adverse impacts of drought will continue to pose threats to Uganda in future amidst the growing population and economy and increasing urbanization (Kilimani et al. ). The experiences of the local communities, their AC and risks need to be understood for proper measures to be put in place. This study seeks to address this gap in knowledge through determining the impact that drought has had on water quantity and assessing the AC of households to withstand such impacts.
In determining drought, various indices have been developed in the last 30 years to monitor the various droughtsagricultural, meteorological, hydrological, and socio-economics. Many studies have indicated that the Reconnaissance Drought Index (RDI) was useful for hydrological studies because it has been recognized as a more sensitive and conservative index (Shokoohi & Morovati ; Tigkas et al. ).
RDI also accounts for potential evapotranspiration (PET) and rainfall, unlike other indices, and therefore could potentially be useful in analysing droughts in tropical countries where PET is an important climatic factor. These data requirements are ideal for developing countries which monitor few climatic parameters. Combining drought information with a community household survey (HHS), Focus Group Discussions (FGD) and Key Informant Interviews (KII) provides robust tools to understand the community's experiences and assess the AC of households of Kasali.

STUDY AREA
The study area is in Kasali sub-county, southern Uganda  climate of the study area is governed by the passage of the Inter Tropical Convergence Zone (ITCZ) and influenced by convective rainfall influenced by the proximity to Lake Victoria (Mubiru et al. ), thereby experiencing a moderate rainfall distribution throughout the year with longer rains taking place between March and May and shorter rains between October and November, receiving a mean annual rainfall of 1180 mm. The dry months are January-February and June-August (GOU ). The mean annual maximum temperature of Kasali is around 25 C. The minimum temperature in the east (17.5 C) is higher than the west (15 C) (GOU ). The relative humidity ranges between 80 and 90% in the morning and 61-66% in the afternoon for the months of January-May. In June-August, the morning relative humidity decreases to around 77% and the same applies to the afternoon, which decreases to around 57% (GOU ).

METHODOLOGY Data collection
Both primary and secondary data were used in this study. The primary data was collected using a household questionnaire (HHQ), FGD and KIIs, while secondary data (monthly temperature and rainfall) for 1987-2017 was acquired from Uganda National Meteorological Authority (UNMA).
HHSs were undertaken to gather information on impacts of past droughts on water availability and assessing the AC of households. A total of 195 questionnaires were administered to the household heads or any other adult found in the household in the three parishes (Gayaza, Kigenya and Nkenge). The sample size was determined using Cochran's equation as shown in Equations (1) and (2): where n 0 ¼ sample size; z ¼ 1.96 for 95% confidence level; p ¼ estimated proportion of population (assumed to be 50% or 0.5); e ¼ margin of error (assumed to be 0.07).
Most meteorological stations in developing countries, where P ij and PET ij are the precipitation and potential evapotranspiration respectively in the jth month of the ith year and N is the total number of years. The normalized form (RDI n ) is computed using the following equation: in which α (i) k is the arithmetic mean of α (i) k . By assuming that the lognormal distribution is applied, the following equation can be used for the calculation of RDI st : in which y (i) is the ln (α (i) k ), y its arithmetic mean and σ y is its standard deviation. The drought classifications according to RDI values are shown in Table 1.

AC analysis
The main indicators influencing the AC of households to water scarcity during drought were selected based on  (6): with the interval of 0.5, the upper cut off point was determined as 3.00 þ 0.5 ¼ 3.50 and the lower limit as 3.00- The classification of the AC of the households is classified as low AC between 0.0 and 2.49, moderate AC between 2.50 and 3.49, and high AC between 3.50 and 5.00 (Table 2).
We used the χ 2 -test to assess if there were differences in AC across the six villages and also if there was a difference in travel time between the dry and wet season between the villages.

Rainfall
The annual monthly average rainfall for the period 1987-2017 for Kasali is shown in Figure 2. The results show that Kasali sub-county has a bimodal rainfall distribution with long rains being observed between March and May (MAM) and short rains between September and December (SOND). The study site has two dry seasons, the long season occurring in June-July-August (JJA) and the short dry season occurring in January-February (JF). May is the wettest month and receives the highest amount of rainfall averaging 178.43 mm (standard error, SE ± 12.33) and the driest month is July with a long-term average rainfall estimated at 27.15 mm (SE ± 3.6).

Temperature
The average Table 3 shows long-term trends in the rainfall, maximum and minimum temperature. The results show that annual and JF (-0.135, p ¼ 0.2919) rainfall declined but this was not statistically significant. The long dry seasons of JJA        and lowest in MAM (1.07 C).
In the 30-year study period there has been one extremely dry year, four moderately dry years and one extremely wet year, two very wet years, one moderately wet year, and 21 near normal years (Figure 4). The year 1991-1992 was extremely dry and 1988-1989, 1999-2000, 2008-2009

Impact of drought on water availability in households in
Kasali sub-county The impact of drought on water availability was determined through analysing the time taken by household members to collect water from the source during wet and dry years.
There was a statistical difference in the dry season (χ 2 2 ¼ 6.3 p ¼ 0.04) in terms of time spent collecting water in the six villages, while in the wet season there was no statistical difference in time used in the six villages (χ 2 2 ¼ 3.5 p ¼ 0.17). During wet years, 19% of the households in Kasali subcounty spend more than 1 hour collecting water, while 16%  spend 1 hour and the majority, 65% of the households, spend less than 1 hour collecting water. However, during drought years, the majority of the households (85%) spend more than 1 hour collecting water, while 12% spend up to 1 hour and only 3% spend less than 1 hour. The percentage of households spending more than 1 hour during dry years (85%) is considerably higher than that in wet years (19%), meaning that drought has a negative impact on the availability of water to the households (Figure 5(a) and 5(b)).

Time spent by households collecting water during dry and wet years
The time spent collecting water by each village during both wet and dry seasons differs due to their relative geography that determines water availability and water quality (refer to Supplementary material S2a for statistical significance).
Nkenge had the largest percentage of households which spent more than 1 hour followed by Kyango-Bigavu, 29.0, 16.1, 12.9, 9.7, 6.5 and 0.0% respectively. Kyango-Bigavu had the largest percentage of households that spent less than 1 hour (3.2%) and the other villages registered that none of its population spent less than 1 hour collecting water (Figure 6(a)).   1988-1989, 1999-2000, 2008-2009, 2016-2017 2016-2017, 2008-2009, 1999-2000  (Nkenge) has the largest percentage of households that spend more than 1 hour (32.3%), followed by Nkenge The three parishes selected for the study have a total of 26 water points which include boreholes (nine), shallow wells (10), water pumps (three) and protected well (one) ( Figure 1). Kigenya parish has the highest number of water points (15) and Gayaza has the lowest number of water points (three).

Indicators of AC to water scarcity in the villages of
Kasali sub-county The analysis indicated variation of AC across the villages and also among the five adaptive interventions, but it was not statistically significant (refer to Supplementary material 2B).  Table 7).

Drought trends
This study sought to understand the occurrence and fre- finding the area has not been ravaged by droughts because there are exceedingly high near-normal rainfall years (21 events) compared to either dry (five events) or wet years (four events). The moderate dry years occurred between the range of 7-10 years (1988-1989, 1999-2000, 2008-2009, and 2016-2017). In the earlier years, a drought occurred about once in 10 years, however, in recent times the frequency has reduced to every seven years. Additionally, one extreme drought was registered in 1991-1992.
Severe and extreme droughts are normally influenced by

Impact of drought on water access
Spending more than 1 hour collecting water is considered a vulnerability because water is collected more than once and At the village level, both access to water and the time spent to collect water vary. In the dry season the majority of villages (five out of the six) have more than 83% of their households spending more than 1 hour to collect water. One village in particular, Nkenge, has all the households spending more than 1 hour to collect water. The only village that has a moderate water access is Lwengwe because about 29% of its households could access water within 1 hour. Many households reported having unsustainable means of storing water for future use which leads to water scarcity. These findings are consistent with a study in Kaliro district in eastern Uganda where it was found that household members, especially women, struggle in accessing water during drought as a result of the drying up of nearby water sources like wells and river beds. This means that they have to walk miles in search of water (LWR ).
Some parishes such as Nkenge have six water points (three boreholes and three shallow wells) and Gayaza has three water points (two boreholes and one shallow well).
However, these sources are vulnerable to drought, for example the wells dry up during drought and boreholes become defunct due to over usage. The respondents indicated that the water from these water sources have reduced their flow due to a decrease in the water table during dry years. This also contributes to the length of time households spend while collecting water during drought. Households spend a shorter time collecting water during wet years because seasonal water sources such as ponds and wells are recharged by rains. Furthermore, households harvest water during wet years. Similar studies conducted on water scarcity have found that households access water much more easily during wet years (Dessalegn et al. ; Mubiru et al. ). As shown in this study the drought cycles are becoming shorter, therefore the impact on local communities will be large unless they come up with adaptation strategies that will help overcome these hurdles.

CONCLUSIONS AND RECOMMENDATIONS
In analysing the climate of Kasali between 1987 and 2017 we have found that the annual minimum and maximum temperatures have increased by 1.24 and 0.65 C respectively. The RDI has proved a useful tool to assess drought in Kasali. Our study reveals that the region has experienced five moderate to extreme drought years in the 30-year period . These droughts and other dry spells in the region result in water scarcity leading to more than 85% of the households requiring more than 1 hour to collect water contrary to only 19% in the wet seasons. The average time spent while collecting water in Kasali during dry years (1 hour) is higher than the UN goal of 30 minutes. The fact that residents need more than 1 hour to collect water during the wet seasons means that there is a shortage of water supply and there is a need for the government to develop more water access points that will help the residents to spend less time and increase their resilience to climate change impacts.
Additionally, we find that the overall AC of Kasali is low.

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