Community livelihood vulnerability to drought in semi-arid areas of northern Cameroon

Drought is a devastating recurring climate hazard, with enormous impacts on livelihoods and attainment of sustainable development goals (SDGs) in semi-arid areas. The current study was conducted to evaluate households’ livelihood vulnerability to droughts in an African semi-arid region. A total of 384 households from a drought-prone region in Northern Cameroon were randomly sampled and interviewed using a structured questionnaire. Livelihood vulnerability was evaluated from the collected households’ data using the Livelihood Vulnerability Index (LVI) and Intergovernmental Panel on Climate Change Vulnerability Index (IPCC-VI) methods. The LVI results showed that food insecurity, livelihood strategy and social networking were the major drivers of livelihood vulnerability. This was because almost all of the interviewed households relied on rain-fed agriculture and there was limited early drought warning information, which limited the achievements of some SDGs such as zero hunger, water accessibility and poverty alleviation. Adoption of drought-resilient and early maturing crops together with the use of agriculture extension officers to disseminate drought warnings are advised to reduce the effects of the livelihood vulnerability drivers and ensure sustainable development. The IPCC-VI showed that drought exposure and adaptive capacity were low while drought sensitivity was high among the surveyed households. The high sensitivity was due to food and water insecurity, while low adaptive capacity was due to a lack of formal education and a high dependency rate. Heavy investment in education and reducing the number of dependents through family planning and campaigning against polygamy could enhance households’ adaptive capacity. It was concluded that households’ livelihood vulnerability in the study area was due to high sensitivity and low adaptive capacity to drought. Drought reversed positive steps that had been taken to achieve various SDGs; however, heavy investment in other SDGs such as quality education and health could enhance households’ adaptive capacity and system sustainability in African semi-arid regions.


Introduction
Drought is a recurring climate-related hazard, with enormous impacts on natural and production systems, especially among rural households that depend on agriculture for their livelihood [20]. It creates extensive costs with adverse effects on agriculture due to shortages of water caused by a decrease in precipitation [19,27]. The severity and 1 3

Study design and data collection
Five communities, namely Balaza Lawane, Dogba, Gazawa, Meri Ville and Yolde Naggue in the Diamare Division, were selected for the current study (Fig. 1). The communities were selected because of their willingness to participate in the study, easy accessibility and presence of social order. A household survey was used to collect qualitative data among the participating communities. A total of 384 households were randomly selected in the study area. The sample size was determined as per Bartlet et al. [10], whilst 50% of the population was estimated to be engaged in agricultural production and the confidence interval was 95%. The respondents were required to give verbal consent before the interview and only individuals who were > 20 years old were interviewed as they were expected to be involved in households' decisions and have a better understanding of drought impacts on their livelihoods. The respondents were interviewed by research assistants in their local languages using structured questionnaires with open and closed-ended questions. The questions administered included various LVI and IPCC-VI components such as socio-demographic profile, livelihood strategy, social network, food status, health condition, water status, and drought hazard. Survey respondents were encouraged to discuss their answers with other family members (partners and elderly relatives) to obtain more information regarding household practices. Secondary meteorological data were also collected from a local weather station (Maroua Salak Airport) and were used to determine the vulnerability of the study population. Collected survey data were coded and analysed using Statistical Package for Social Sciences version 20. The analysed results were presented in frequency, percentage or counts, which were either reported in the current paper or used to calculate different livelihood vulnerability components (see the section below). The dependency ratio in the Division was calculated as the sum of dependants (children < 14 years and elders > 65 years) divided by the number of those who were independent. Rainfall and temperature data were used to determine the exposure of the communities to drought hazards.

Vulnerability analytical framework
The LVI, as described by Hahn et al. [17] with sub-components modifications, was used to assess livelihood vulnerability. The LVI used seven major components and multiple indicators that were obtained through a household survey (Table 1). A balanced weighted average was employed in indicators' aggregation since indicators and sub-components were on diverse scales (i.e. percentage, count, inverses), which was necessary to ensure equal contribution to the overall index. The selected indicators were standardized (using Eq. 1) based on the Human Development Index as described by Hahn et al. [17].
where S c is the obtained value of the indicator, and S max and S min are the maximum and minimum values, respectively, for each of the sub-components.
The major component index was calculated by averaging the respective sub-components as presented in Eq. 2 below where n denotes the number of sub-components in each major component and M C is the value of respective major components. For household c, index Sci signifies the indicators, indexed by i, that comprise each major component. The major component values were averaged to obtain the LVI for each of the communities using Eq. 3: This can be written as where LVI is the livelihood vulnerability index for the community that is equal to the weighted average of the seven major components, and W Mi is the weight obtained from each major component. The LVI ranged from 0 (least vulnerable) to 1 (extremely vulnerable). Indicators used to determine the LVI (Eqs. [1][2][3][4] were also used to calculate the LVI-IPCC according to [17]. The indicators were aggregated into three vulnerability contributing factors (exposure, sensitivity and adaptive capacity, see Table 1) using Eq. 5.
where CF c is the IPCC contributing factor (sensitivity, adaptation or exposure) for community c, WM i is the weight of each major component, M ci is the major component for community c indexed, and n is the number of major components in each contributing factor.
The respective contributing factors were combined to establish the IPCC vulnerability index using Eq. 6 below; (1)  The IPCC-VI was scaled from − 1 (least vulnerable) to 1 (extremely vulnerable).

Livelihood vulnerability to drought impacts
Livelihood vulnerability indicators and indexes are shown in Tables 2 and 3, respectively; and there were variations among the semi-arid communities. Gazawa had the highest percentage of the households (12.3%) headed by females compared to other communities, while Yolde Naggue had the highest index of household heads with no formal education (0.67). The average household size index was the highest in Dogba (0.417) and lowest in Meri (0.280), while the dependency ratio was 1.28 in the Division. The aggregated socio-demographic index showed that Yolde Naggue was the most vulnerable (0.347) community, while Balaza Lawane was the least vulnerable (0.280) in the area. However, Balaza Lawane ranked lowest (0.721) in the aggregated livelihood strategy while Gazawa was the highest (0.628). Balaza Lawane also led in low household income, households whose income could not cover essential expenses and households with low livelihood diversification, which were 83%, 0.836 and 0.36, respectively. The results showed that over 90% of interviewed households in the division depended on agriculture (crop and livestock) as the main source of income. The aggregated social network vulnerability index was higher in Yolde Naggue (0.819) than in the rest of the communities. The proportion of the surveyed households that did not belong to any social group or organization and did not receive warning information was 77% and 97.5%, respectively, in Yolde Naggue, which was the highest in the Division. Balaza Lawane ranked the highest (0.712) in households that did not receive assistance from any social group/organization within their community and Gazawa was the lowest (0.506). The aggregated household health vulnerability score was low (0.093); however, surveyed households in Dogba had higher scores in households with sick or disabled household heads (0.247) compared to other communities. All of the surveyed households in the Division obtained their food primarily from their farms; however, Balaza Lawane (0.939) and Gazawa (0.938) had the highest overall food security vulnerability score in the area. Additionally, Gazawa ranked highest (0.535) in the aggregated water vulnerability index. Most households relied on natural sources such as rivers for water; but in Gazawa a large proportion of respondents (92.8%) claimed not to have a consistent water supply. Households in Gazawa (55.6%) and Meri (55.4%) reported water conflicts in their communities and respondents attributed this to water scarcity due to drought.
The aggregated drought disaster vulnerability was the highest in Gazawa (0.395) compared to other communities. The mean standard deviations of the monthly precipitation and temperature were 155.98 mm and 3.26 °C with the LVI of 0.15 and 0.13, respectively. The number of drought incidents experienced in the past 10 years (2010-2019) by households ranged from one to nine, with Gazawa having the highest score (0.3). Almost all respondents claimed to be affected by drought through crop failure, livestock death, malnutrition among children < 5 years, pasture and water shortage, reduced household income and high food prices. The overall LVI in the Division was 0.51 and that of Yolde Naggue was 0.534, which was above the average, making the community most vulnerable to drought impacts.

Major drivers of households' livelihood vulnerability to drought
The livelihood vulnerability drivers for the studied communities are shown in Fig. 2. Food insecurity, livelihood strategy and social network, in that order, were the main drivers of drought vulnerability in all communities except in Yolde Naggue where social network, food and livelihood strategy, in that order, were the main drivers. Overall, access to food, livelihood strategies, social networks and water availability were the key factors causing household livelihood vulnerability to drought.

Exposure, sensitivity and adaptive capacity of the communities (IPCC-VI)
The IPCC-VI aggregated score in the Division was − 0.090, while those in the study communities were − 0.068, − 0.074, − 0.081, − 0.095 and − 0.129 in Gazawa, Meri, Dogba, Balaza Lawane and Yolde Naggue, respectively (see Table 4). These IPCC-VI values indicated that Gazawa was the most vulnerable community and Yolde Naggue was the least vulnerable community. Gazawa had the highest exposure (0.395) and sensitivity (0.582) scores, while its drought   adaptive capacity was the lowest (0.512) among the study communities. The lowest level of exposure, sensitivity and high adaptive capacity made Yolde Naggue the least vulnerable community (Fig. 3). Figure 4 shows that high sensitivity and low adaptive capacity were the main drivers of vulnerability in the study area.

Drought sensitivity and adaptive capacity
High sensitivity and low adaptive capacity were the main contributing factors to drought vulnerability, while the exposure was low among the surveyed households in the semi-arid region (Fig. 4). The low exposure level indicated that to ensure household drought resilience and system sustainability in drylands, the focus should be on improving their adaptation capacity and sensitivity as was also noted by Thao et al. [42] and Mwatu et al. [31]. High sensitivity was because of high food and water insecurities due to reliance on rain for crop cultivation and replenishing of the natural water sources in the study area: which limited the achievements of SDGs i.e. zero hunger, water accessibility and poverty alleviation. The sensitivity included the health component as well; however, its contribution was minimal due to its low value compared to water and food components ( Table 3). The health component focused mostly on  household heads with long term illnesses, which is an important factor in the construction of vulnerability because they might use resources that are needed for food production while incapable of engaging in production activities [12,17]. However, aggregated health components did not account for other family members with long term illness or medical costs due to the high prevalence of diseases such as dysentery and cholera, which are common during drought [22]. There is a possibility that the health contribution to the sensitivity index could have been higher if more health indicators had been included as was done by Qaisrani, Umar, Siyal, and Salik [35] and Etwire, Al-Hassan, Kuwornu, and Osei-Owusu [16] in their vulnerability studies in rural semi-arid areas. Health service, being one of the SDGs, needs to be improved among rural households (despite its lower contribution to sensitivity in the current study) to enhance/maintain households' resilience against drought sensitivity in similar African drylands elsewhere. Nonetheless, the drought sensitivity (0.52) reported in the current study was higher than the 0.09-0.4 range noted by Etwire et al. [16], Mwatu et al. [31] and Qaisrani et al. [35] among farmers in semi-arid areas. The sensitivity differences are due to the low adaptive capacity (0.54) observed in the current study compared to the former studies.
Low adaptive capacity was due to the socio-demographic profile of the surveyed households whereby most household heads lacked formal education and the average household size was large (Table 2). Large household size could be due to early marriage, polygyny, lack of family planning or households caring for the children of relatives who died from violence or the HIV/AIDS pandemic, which is prevalent in most parts of Sub-Saharan Africa [3,9,34]. Large household size could provide the labour required for production but households with a high dependency ratio (high number of children < 14 years and elders > 65 years), such as that observed in the current study (1.28), offset production and livelihood sustainability [25]. Households usually opt to reduce the number of meals per day due to large family sizes. This practice is detrimental to their welfare, leading to health problems, especially among the most vulnerable members such as the young, the elderly, and pregnant women [31,38,46]. Family planning needs to be emphasized among rural households in African drylands and there is merit in campaigning against polygamy and early marriage so that the household adaptive capacity could be enhanced with improved household structure i.e. the number of dependants does not exceed the available labour force. Additionally, households that are more vulnerable and caring for the children of deceased relatives should receive government or social support to enhance their drought adaptive capacity. Lack of formal education was another contributing factor to low adaptive capacity among the surveyed households, which made access to written information difficult. Uneducated individuals lack the specialized skills needed to be employed in other industries [6,13], which was the case in the area. Adult education, as was done in Tanzania [13], should be introduced to achieve various SDGs i.e. gender equality, economic growth, inequality and poverty reduction, especially among women who did not receive any formal education during their childhoods. Education and training can improve the ability of households to acquire modern and appropriate drought adaptation and recovery information such as changes in daily weather and planting dates. Additionally, it can improve the ability of the agrarian population to diversify their income sources through the acquisition of salaried jobs that could not be possible due to lack of skills, and adaptation options, especially from written sources, that can play a great role in increasing productivity and fostering individual economic growth. Thao et al. [42] argues that education makes people more aware and able to adjust to climatic and environmental changes. The approach could be useful in other African drylands as well.

Household livelihood vulnerability
Households' livelihood vulnerability (LVI) was not the same among the communities in the studied area (see Tables 2  and 3). The observed overall LVI differences among communities are not new. Mekonen, Berlie, and Ferede [27], Mwatu et al. [31] and Segnon et al. [37] also reported variations in livelihood vulnerability even among the units within the same semi-arid area. The main livelihood vulnerability drivers in the Division were food insecurity, livelihood strategy, social network and water stress (Fig. 2), indicating the central role of agriculture in the rural economy and achieving the SDGs. Households depended on rain-fed agriculture and cultivated fewer crops, which increased their vulnerability to crop failures and food shortages during severe droughts. Vulnerability drivers reported in the current study were the same as those reported among the agrarian communities in African drylands by Adu et al. [4], Cassidy and Barnes [11], Lana et al. [21] and Maru et al. [23]. However, a weak social network, not food insecurity, was the main vulnerability driver in the Yolde Naggue community. The social network is an important vulnerability driver since households that have members in social groups can easily receive new technology and/or assistance to cope with the impacts of droughts and ensure system sustainability [6][7][8]. Yolde Naggue had a higher aggregated social networking index than other communities because there was an extremely high proportion of respondents who did not receive early drought warnings ( Table 2). Early drought warnings could be essential to households as they could save the available food to last longer and/or cultivate drought-resilient crops or varieties, which was the case in similar African drylands [30,32]. The limited access to drought warnings was probably due to the community's distance from the region's capital and poor media coverage (mobile phones, radio and television), which are the common channels used to share information [2]. African governments should improve their broadcasting infrastructure for timely dissemination of drought warnings in drylands; however, if this is expensive, agriculture extension officers could be used to ensure both timely dissemination of the warnings to improve social networking and sustainable development.
The reported overall LVI (0.51) and IPCC-VI (− 0.09) in the current study were not within the ranges of 0.4 to 0.5 for LVI [14,40] and − 0.40 to 0.34 for IPCC-VI [16,29] reported in the Asian highlands and African semi-arid areas, respectively. The variations could be attributed to the differences in production systems, selected indicators and assessment scales used. Furthermore, the LVI and IPCC-VI in the current study yielded inconsistent results: LVI showed Yolde Naggue (0.534) 1 3 to be the most vulnerable while IPCC-VI showed Gazawa (− 0.068) was the most vulnerable and Yolde Naggue (− 0.129) to be the least vulnerable. Dendir and Simane [14] and Rai, Bajgai, Rabgyal, Katwal, and Delmond [36] observed similar inconsistencies between LVI and IPCC-VI, while Adu et al. [4] and Sujakhu et al. [40] reported consistent findings between the two methods. The inconsistency was because Yolde Naggue had a higher adaptive capacity than Gazawa (Fig. 3) when the major components were aggregated into the three IPCC vulnerability contributing factors. The LVI results showed that although Yolde Naggue was the most vulnerable in almost all of the adaptive capacity components i.e. socio-demographic, livelihood strategy and social network (Table 3) it had the least IPCC-VI score, making it the overall least vulnerable community. This is so because the IPCC-VI function (Eq. 6) skewed the results and its outcomes were not realistic based on the components used and values obtained in the current study. The overall result of the LVI was realistic compared to the IPCC-VI when it was used to assess drought vulnerability impacts on the livelihood of the semi-arid population. Other studies [4,40] did not find the same inconsistency; and assessing the two methods is beyond the scope of the current study. We propose that both methods could be used in assessing livelihood vulnerability as they complemented each other in identifying livelihood vulnerability drivers and their contributing factors in semi-arid regions.

Conclusion
The current study examined the vulnerability of households to drought impacts in the semi-arid Diamare division of northern Cameroon by employing the Livelihood Vulnerability and IPCC-VI approaches. In general, the LVI and IPCC vulnerability indexes have been useful in determining the disparities in the vulnerability of selected communities in a semi-arid zone of northern Cameroon to drought. Both the LVI and IPCC-VI indicated variability in the vulnerability of households to the indicators of drought amongst the study communities. The study showed that high sensitivity and low adaptive capacity contributed to households' livelihood vulnerability, which in turn affected sustainability in the semi-arid zone of Cameroon. The main vulnerability drivers were food insecurity, livelihood strategy and social networking because the studied communities relied heavily on rain-fed agriculture, cultivated fewer crops and there was limited early warning information on impending droughts. Early drought warning could be essential to the households as they could prepare themselves for the upcoming drought. The preparation could be through the adoption of early maturing and drought-resilient crops, and livelihood diversification; hence, timely dissemination of weather information through media and/or agricultural extension officers could enhance drought resilience among semi-arid households and ensure sustainable development. Furthermore, households' drought adaptive capacity could be improved through increased access to formal education and improved household structure i.e. reduced number of dependants. The number of dependants could be reduced by campaigning against polygamy and early marriage, which will also contribute indirectly to various SDGs i.e. zero hunger, poverty alleviation, gender equality, health and wellbeing, especially among women. Generally, drought reduced system sustainability and has the potential to reverse positive steps taken to achieve various SDGs (zero hunger, poverty reduction and water access) in African semi-arid areas as revealed in this study. However, heavy investment in SDGs such as quality education and health could improve households' adaptive capacity and system sustainability. This study adds knowledge on natural hazards by examining the degree of drought vulnerability at the community level using primary data and suggests possible measures for vulnerability reduction in drought-prone areas. Additionally, since drought is a frequent phenomenon in semi-arid regions this study is relevant for policymakers, decision-makers, development planners and households to make informed decisions for proper drought interventions that are based on the underlying vulnerability conditions of a community.