1. Introduction
The movement of people from their habitual residence as a result of environmental change is not a new phenomenon, but it has been part of human history since pre-historical times (e.g. De Blij, Reference De Blij2010; Hinnawi & UNEP, Reference Hinnawi1985; Ionesco et al., Reference Ionesco, Mokhnacheva and Gemenne2014). The permanent or circular movement of people in response to climate change has been documented with climate anomalies associated with the North Atlantic Oscillation (Drake, Reference Drake2017), the late 4th-century drastic drop in temperature across Europe (Sedov, Reference Sedov1995, Reference Sedov and Mel'nikova1999), the ‘Medieval Warm Period’ (10th–12th centuries), the ‘Little Ice Age’ (13th–19th centuries) (Dobrydnev, Reference Dobrydnev2003), extreme climate events in the middle Qing Dynasty (1760–1820) (Li et al., Reference Li, Ge, Wang and Tao2017), the modern warm stage (20th century) (Klimenko, Reference Klimenko2016), the 1930s ‘Dust Bowl’ in the USA (McLeman, Reference McLeman2006), and many other cases. This process continues today, leading to border expansion, new land acquisitions, formation of new nations (Drake, Reference Drake2017), and is often motivated by people's desire to ensure better living conditions for generations to come.
However, it is in recent decades that global environmental changes in the form of gradual and sudden environmental events started receiving wider recognition as a driver of permanent or seasonal internal or cross-border migration (de Campos et al., Reference de Campos, Bell and Charles-Edwards2017; Groth et al., Reference Groth, Ide, Sakdapolrak, Kassa and Hermans2020; Laczko & Aghazarm, Reference Laczko, Aghazarm and Frank Laczko2009), a phenomenon defined and described with a variety of terms (Table 1). According to the International Organization for Migration (IOM), the environmental drivers of human migration are often divided into two broad groups as climate events associated with the sudden onset of extreme climate events (typhoons, storms, monsoon floods, glacial lake outbursts, hurricanes, and floods), and climate processes underlying slow and accumulated changes (desertification, and land-use processes, food insecurity, sea-level rise, and aridity trends) (Brown, Reference Brown2014). Non-climate factors such as societal vulnerability and low adaptive capacity transform natural hazards into natural disasters (Brown, Reference Brown2014; Hesse & Cotula, Reference Hesse and Cotula2006; IFRC, Reference Walter2004; McLeman, Reference McLeman2006).
Glossary of the terms used to define people's migration and displacement and the environmental dimension of these phenomena
The terms migration and displacement are interchangeably used in this paper.
The term environmental migration represents the movement of people because of extreme sudden events, slow-onset events, or accumulated events altogether or separately.
Several frameworks have been developed to address migration and displacement concerning climate change and environmental disasters. Some notable examples of frameworks, conventions, and treaties are shown in Table 2 (Abdiker et al., Reference Abdiker, Cessouma, Kidd, Maze, Konsimbo, Adepoju, Fumagalli and Nyabola2019; IOM, 2018; Kraler et al., Reference Kraler, Katsiaficas and Wagner2020; UN, 2018). The goal of these frameworks is often not only to contribute to the prevention and reduction of internal displacement risks globally but to serve as a template for regional or country-level initiatives, to promote regional or state engagement in environmental migration and research advancement. The African Union Convention for the Protection and Assistance of Internally Displaced Persons (or ‘Kampala Convention’) is the first legally binding regional convention in the world (Abdiker et al., Reference Abdiker, Cessouma, Kidd, Maze, Konsimbo, Adepoju, Fumagalli and Nyabola2019). This agreement, which was ratified by 28 African states and represents their commitment to prevent and address the conditions underlying internal displacement, still lacks systematic translation into action by individual states (ICRC, 2017).
Overview of initiatives/tools on environmental migration and displacement
Source: Abdiker et al. (Reference Abdiker, Cessouma, Kidd, Maze, Konsimbo, Adepoju, Fumagalli and Nyabola2019); IOM (2018); Kraler et al. (Reference Kraler, Katsiaficas and Wagner2020); UN (2018).
According to recent reports by the African Union Commission (AUC) and IOM, African migrants account for only 14% of global migration while Asia and Europe account for 41 and 24%, respectively (Achieng et al., Reference Achieng, Fadil, Righa, Adepoju, Fumagalli and Nyabola2020; McAuliffe et al., Reference McAuliffe, Bauloz, Nguyen, Qu, Kitimbo, Abel, Sawyer, Klatt, McAuliffe and Khadria2020). Despite the horrific stories of migration through the Mediterranean Sea that are brought to the attention of the public by the media, the majority of African migrants do not cross the oceans but rather cross land borders, a phenomenon that often remains understudied and poorly reported by the media. People from Africa typically move in search of better economic opportunities and a safe place for their families. It is still unclear to what extent the movement of people from and within Africa is influenced by environmental changes (Borderon et al., Reference McNamara, Farbotko, Thornton, Dun, Ransan-Cooper, Chevalier and Purevdulam2017). While various migration studies have tried to evaluate whether and to what extent migration is amplified by ongoing global environmental change, the debate still remains at large (Groth et al., Reference Groth, Ide, Sakdapolrak, Kassa and Hermans2020). There is a need for a better understanding of the role played in the migration and internal displacement process by other underlying non-environmental factors such as slow economic development, unemployment, corruption, lack of integrated natural resources management, demographic conditions, insufficient preparedness, lack of climate change policy and law, poor infrastructures, and other threats to livelihoods. Hydroclimatic changes may increase or decrease the plausibility of migration, but these outcomes highly depend on the underlying non-environmental contexts. How are these factors contributing to exposing communities in sub-Saharan Africa (SSA) to high vulnerability and associated susceptibility to migration in response to environmental change? Could migration as a result of environmental drivers be prevented with more efforts in planning for (and investing in) improved infrastructures and preparedness? It is still unclear to what extent migration occurs as an adaptive strategy for people dealing with adverse environmental conditions (Mcleman & Smit, Reference Mcleman and Smit2006; Müller-Mahn & Gebreyes, Reference Müller-Mahn and Gebreyes2019; Warner, Reference Warner2010). In the context of SSA, little is known to fully support one of these arguments (Ferris, Reference Ferris2020). Previous studies and meta-analyses have investigated different cases of environmental migration in SSA providing a wealth of information on socio-environmental conditions underlying relatively recent migrations in Africa (Borderon et al., Reference Borderon, Sakdapolrak, Muttarak, Kebede, Pagogna and Sporer2019; Hoffmann et al., Reference Hoffmann, Dimitrova, Muttarak, Crespo Cuaresma and Peisker2020; Kaczan & Orgill-Meyer, Reference Kaczan and Orgill-Meyer2019; Laczko & Aghazarm, Reference Laczko, Aghazarm and Frank Laczko2009; Lilleør & Van den Broeck, Reference Lilleør and Van den Broeck2011; Marchiori et al., Reference Marchiori, Maystadt and Schumacher2012; Naudé, Reference Naudé2010; Neumann & Hilderink, Reference Neumann and Hilderink2015; Rigaud et al., Reference Rigaud, de Sherbinin, Jones, Bergmann, Clement, Ober, Schewe, Adamo, McCusker, Heuser and Midgley2018). In most cases, however, conclusive empirical evidence of the causal relationship between environmental change and migration is still missing. Most of the empirical frameworks and statistical data published are based on the extrapolation of values from a small number of people residing in affected areas (Gemenne, Reference Gemenne2011; Marchiori et al., Reference Marchiori, Maystadt and Schumacher2012; Neumann & Hilderink, Reference Neumann and Hilderink2015; Trenberth et al., Reference Trenberth, Jones, Ambenje, Bojariu, Easterling, Klein, Parker, Rahimzadeh, Renwick, Rusticucci, Soden and Zhai2007). Despite the comprehensive reviews on environmental migration in 20 countries in Africa (Borderon et al., Reference Borderon, Sakdapolrak, Muttarak, Kebede, Pagogna and Sporer2019), robust evidence regarding the relationship between migration and climate anomalies is still missing. The recently published Intergovernmental Panel on Climate Change (IPCC) report also highlights the lack of extensive socioeconomic data, which limits our ability to evaluate whether African nations exhibit particularly strong exposure and vulnerability to environmental migrations or low adaptive capacity to respond to this phenomenon. The limited availability of socioeconomic data makes the understanding of the climate impacts and risks in the African continent particularly challenging (Trisos et al., Reference Trisos, Adelekan, Totin, Ayanlade, Efitre, Gemeda, Kalaba, Lennard, Masao, Mgaya, Ngaruiya, Olago, Simpson, Zakieldeen, Langsdorf, Löschke, Möller and Okem2022; Zhao et al., Reference Zhao, Guo, Ye, Gasparrini, Tong, Overcenco, Urban, Schneider, Entezari, Vicedo-Cabrera, Zanobetti, Analitis, Zeka, Tobias, Nunes, Alahmad, Armstrong, Forsberg, Pan and Li2021).
SSA strongly depends on hydroclimatic conditions for its food production (Collier et al., Reference Collier, Conway and Venables2008) and has low adaptive resilience (Downing et al., Reference Downing, Ringius, Hulme and Waughray1997) to the high levels of hydroclimatic variability, which expose the region to extreme conditions marked both by excess water (long wet seasons, flood, cyclone) and water-deficiency events (extended dry seasons, drought) (Brown et al., Reference Brown, Meeks, Hunu, Yu, Brown, Hunu, Meeks, Kennedy and Yu2010). This meta-analysis explores whether and how hydroclimatic variability has affected internal migration flows across SSA, a topic of growing interest that has remained understudied largely due to the lack of holistic spatiotemporal datasets. So far, the most comprehensive scientific studies published on environmental migration/climate displacement in Africa are the meta-analyses by Borderon et al., Reference Borderon, Sakdapolrak, Muttarak, Kebede, Pagogna and Sporer2019, Grote and Warner (Reference Grote and Warner2010), and the Groundswell report (Abdiker et al., Reference Abdiker, Cessouma, Kidd, Maze, Konsimbo, Adepoju, Fumagalli and Nyabola2019; Borderon et al., Reference Borderon, Sakdapolrak, Muttarak, Kebede, Pagogna and Sporer2019; Grote & Warner, Reference Grote and Warner2010; Rigaud et al., Reference Rigaud, de Sherbinin, Jones, Bergmann, Clement, Ober, Schewe, Adamo, McCusker, Heuser and Midgley2018). Grote and Warner (Reference Grote and Warner2010) compiled very good evidence of environmental change triggering migration, but the evidence was directly taken from various organizations and newsagent websites and can no longer be retrieved to further support their findings. Most scientific studies on the topic focus on one country, one sample community, or a group of countries from one region. In contrast, evidence from international migration organizations and humanitarian actors directly working with affected people provide a more integrated picture of human mobility in Africa (CRED, 2020; IOM, 2019). This meta-analysis overcomes the above shortcomings by consulting multidisciplinary data sources with detailed evidence showing the interaction between human movement and various environmental changes.
Here we draw on 87 case studies published in the literature (ranging from the environmental sciences to development economics, and migration research) or documented by research databases, reports, and international disaster datasets to develop a meta-analysis investigating the relationship between environmental changes and migration in 32 SSA countries. One of the main goals of this research is to identify the direct and indirect environmental changes causing migration while considering non-environmental underlying drivers. Direct environmental change drivers are defined as being a prompt reason/cause for a displacement of people while indirect drivers are defined as slowly emerging or accumulated environmental changes putting pressure on already vulnerable communities. Non-environmental underlying drivers are fundamentally political, economic, administrative, social, and development processes that lead to the depletion/degradation of natural resources. We develop a new conceptual framework for understanding the cascading flow of environmental change drivers on migration while reconstructing the movement pattern in SSA.
1.1. The debate
The term ‘environmental migration’ is not internationally agreed upon (Table 1). Therefore, researchers in this field used a variety of other terms to describe persons displaced as a result of extreme or slow-onset environmental conditions, including climate change refugees or migrants and environmental refugees or migrants (Ferris, Reference Ferris2020; Hinnawi & UNEP, Reference Hinnawi1985; IOM, 2007, 2014; McAdam, Reference McAdam2009; McNamara et al., Reference McNamara, Farbotko, Thornton, Dun, Ransan-Cooper, Chevalier and Purevdulam2017), while the phenomenon has been referred to as environmental migration, climate migration (Chazalnoël & Ionesco, Reference Chazalnoël and Ionesco2016), and disaster displacement (Table 1). The use of ‘climate refugees’ status to represent migration in the context of climate change fails to recognize human mobility from environmental degradation and more generally environmental change as well as voluntary migration from slow-onset processes (Chazalnoël & Ionesco, Reference Chazalnoël and Ionesco2016). In agreement with the definitions given by Brown (Reference Brown2014) and Brown et al. (Reference Brown, Hammill and McLeman2007) here we use the terms ‘environmental migration’ and ‘environmental displacement’ as the most comprehensive terminology for this research. It should be noted that the term includes people who migrate in response to environmental drivers that are not related to climate change (Table 1), such as volcanic eruptions, earthquakes, tsunamis, and also extreme hydrometeorological events that are not necessarily induced by climate change (Ferris, Reference Ferris2020).
The nexus between migration and environmental change is of great interest to many research fields addressing the human dimension of global change and related mitigation and adaptation strategies (Carr, Reference Carr2005). Scientists from these different fields have tried to document and explain this link. The heavily critiqued theory published in the early 1990s presents the maximalist point of view, whereby the environment is the primary cause of migration (Suhrke, Reference Suhrke1994). While the author failed to explain that not all migrations relate to environmental change, in the same manner, other authors (Adamo, Reference Adamo2014; Black, Reference Black1998; Castles, Reference Castles2002) highlighted that a community's vulnerability to environmental change exposure and the lack of adaptive capacity are the causes of migration rather than the changing environment itself. Black used his review to describe how migration caused by slow-onset events in the Sahel region is a coping cyclical mechanism rather than permanent displacement (Black, Reference Black1998). On the contrary, Myers and Kent argue that rapid population growth especially in developing countries is depleting resources and is a more critical push factor for people to migrate to a new place than slow-onset environmental changes (Myers & Kent, Reference Myers and Kent1995). Other scholars (Czaika & Godin, Reference Czaika and Godin2021; Lee, Reference Lee2001) tried to disentangle the complex drivers of environmental migration by exploring the migration–development nexus and its connection with environmental security. Their conclusions emphasized how the terms used to describe environmental migration have little to do with the understanding of the complexity of specific cases and how they are influenced by non-environmental factors (Czaika & Godin, Reference Czaika and Godin2021; Lee, Reference Lee2001). In response to the need for a better understanding of the environmental migration phenomenon, research in this field has been blooming, providing a number of empirical analyses and case studies linking the environment to migration (see Figure 1) (McLeman, Reference McLeman2012; Neumann & Hilderink, Reference Neumann and Hilderink2015). The notion that environmental change may contribute to migration has gained more acceptance and support from scholars working in this field (Black et al., Reference Black, Adger, Arnell, Dercon, Geddes and Thomas2011; McLeman, Reference McLeman2006; Piguet, Reference Piguet2012; Piguet et al., Reference Piguet, Pécoud and de Guchteneire2011). These scholars largely recognize that migration cannot be caused by environmental change alone except in the case of extreme events (Black et al., Reference Black, Adger, Arnell, Dercon, Geddes and Thomas2011; Castles & Miller, Reference Castles and Miller1998; Parnell & Walawege, Reference Parnell and Walawege2011). Some scholars are slowly moving from addressing the question of how environmental change influences migration to asking how migration might contribute to planned relocation as a climate-change adaptation measure (Ferris, Reference Ferris2020; Gemenne & Blocher, Reference Gemenne and Blocher2017; McLeman, Reference McLeman2006) to build resilience (Rockenbauch & Sakdapolrak, Reference Rockenbauch and Sakdapolrak2017; Sakdapolrak et al., Reference Sakdapolrak, Naruchaikusol, Ober, Peth, Porst, Rockenbauch and Tolo2016; Tebboth et al., Reference Tebboth, Conway and Adger2019). Climate change scientists often perceive migration and human displacement as a negative result of climate change that foregrounds the severity of the climate change threat. This view motivates disaster risk research with its focus on reducing the migration risk associated with environmental disasters by influencing policies and international funding (Ferris, Reference Ferris2020). On the contrary, some scholars look at migration as a coping mechanism or adaptation strategy to climate change (Ferris, Reference Ferris2020).
Flow diagram of research procedure and article selection.
Scholars have also been raising critical questions trying to broaden the understanding of the nexus between migration and environmental changes. It is still unclear what evidence exists in support of the hypothesis that environmental change may induce large movements of refugees and migrants (Castles, Reference Castles2002). There are no clear criteria to determine whether in complex situations environmental factors are primary drivers of migration (Myers & Kent, Reference Myers and Kent1995) or whether climatic conditions actually displace people (Beine & Jeusette, Reference Beine and Jeusette2019). Other studies also raised the question of whether migration is exacerbated as a result of rapid population growth in developing countries rather than occurring in response to climate change alone (Castles, Reference Castles2002). Indeed, it has been argued that ‘environmental conditionalities’ imposed by developed nations are condemning some developing countries to remain poor for life thereby triggering human displacement (Castles, Reference Castles2002). Despite the relatively rich body of literature on this subject, there is still some disagreement on whether environmental conditions may induce migrations. While some authors (e.g. Marchiori et al., Reference Marchiori, Maystadt and Schumacher2012) maintain that rainfall and temperature anomalies forced at least 5 million people to migrate from their place of residence in SSA from 1960 to 2000, others (e.g. Parsons & Beine, Reference Parsons and Beine2015) argue that there is no evidence that climate change is driving migration in the same period. In support of Marchiori's research, a few recent meta-analyses (Lilleør & Van den Broeck, Reference Lilleør and Van den Broeck2011; Reuveny & Moore, Reference Reuveny and Moore2009; Warner, Reference Warner2010) found that climate-related natural disasters and environmental depletion may cause migration in developing countries. A study focusing on SSA (Naudé, Reference Naudé2010) emphasized that environmental pressure may be an indirect driver of migration. Indeed, environmental hazards lower economic growth, fuel conflict, and increase emigration (le Blanc & Perez, Reference le Blanc and Perez2008; Myers & Kent, Reference Myers and Kent1995; Naudé, Reference Naudé2010). For many scholars, however, the line between climate change acting as a direct or indirect driver of migration in the global south remains blurred.
In recent years, new studies focusing primarily on environmental migration in Africa have addressed some of these unresolved questions about this phenomenon, though there is still a long way to go to collect all the data needed to address these knowledge gaps. Moreover, in recent years the first-ever African migration report by IOM and the AUC highlighted how climate change, disasters, and natural hazards are the leading drivers of migration across Africa (Abdiker et al., Reference Abdiker, Cessouma, Kidd, Maze, Konsimbo, Adepoju, Fumagalli and Nyabola2019). Despite the dominant narrative that migration from Africa results from violence and political conflict, the role of water scarcity and climate change has often been underestimated. Climate change research has shown how extreme weather events are becoming more frequent and intense (IPCC AR6 WGI, 2021; Trenberth et al., Reference Trenberth, Jones, Ambenje, Bojariu, Easterling, Klein, Parker, Rahimzadeh, Renwick, Rusticucci, Soden and Zhai2007). The people who are most affected by these events are always those living in vulnerable locations such as semi-arid and arid regions. For this reason, SSA is regarded as a global hotspot of conditions that favor vulnerability to the adverse effects of environmental and climatic stress, as a result of its dependence on rain-fed agriculture and lack of adaptive capacity (Groth et al., Reference Groth, Ide, Sakdapolrak, Kassa and Hermans2020; IOM and UN-OHRLLS, 2019; Niang et al., Reference Niang, Ruppel, Abdrabo, Essel, Lennard, Padgham, Urquhart, Barros, Field, Dokken, Mastrandrea, Mach, Bilir, Chatterjee, Ebi, Estrada, Genova, Girma, Kissel, Levy, MacCracken and Mastrandrea2014; Serdeczny et al., Reference Serdeczny, Adams, Baarsch, Coumou, Robinson, Hare, Schaeffer, Perrette and Reinhardt2017).
Reports published by the IPCC and Foresight project (Foresight, 2011; IPCC, 2014) showed that migration, which varies in scale and time, is becoming an emergency in the view of the projected climatic changes and the high number of affected people as a result of rapid and slow-onset weather events (Foresight, 2011; IPCC, 2014; Niang et al., Reference Niang, Ruppel, Abdrabo, Essel, Lennard, Padgham, Urquhart, Barros, Field, Dokken, Mastrandrea, Mach, Bilir, Chatterjee, Ebi, Estrada, Genova, Girma, Kissel, Levy, MacCracken and Mastrandrea2014; Trisos et al., Reference Trisos, Adelekan, Totin, Ayanlade, Efitre, Gemeda, Kalaba, Lennard, Masao, Mgaya, Ngaruiya, Olago, Simpson, Zakieldeen, Langsdorf, Löschke, Möller and Okem2022). There are increasing concerns about the way changes in the severity and frequency of extreme environmental events can enhance human migration in SSA and reshape the associated displacement patterns (Achieng et al., Reference Achieng, Fadil, Righa, Adepoju, Fumagalli and Nyabola2020; Black et al., Reference Black, Arnell, Adger, Thomas and Geddes2013; Clement et al., Reference Clement, Rigaud, Sherbinin, Jones, Adamo, Schewe, Sadiq and Shabahat2021; IDMC & NRC, 2020; Rigaud et al., Reference Rigaud, de Sherbinin, Jones, Bergmann, Clement, Ober, Schewe, Adamo, McCusker, Heuser and Midgley2018). Despite the surge of studies on environmental migrations, it is still unclear to what extent environmental drivers are directly or indirectly contributing to migrations in Africa. The paradox of environmental migration studies is that there is burgeoning literature on environmental refugees and a paucity of empirical research on migration–environment relations (Bates, Reference Bates2002). The reason why they have carried out little sound empirical work is the complexity of the matter, the multi-causality of the studies, and time lapses in causation (Bates, Reference Bates2002). To date, knowledge of the nexus between migration and extreme environmental events remains limited and fragmented. Therefore, there is a critical need for a synthesis capitalizing on the numerous existing case studies to achieve a holistic understanding of the environmental migration phenomenon. The drivers of environmental migration also vary from one country/region to another in terms of temporal and spatial distribution. Therefore, a mixed methodology is needed to analyze these dynamics. We use empirical analysis in combination with Qualitative Comparative Analysis (QCA) and statistical correlation methods thereby broadening our perspective on the topic. The main objective of this meta-analysis is to review and examine the various existing cases of migration caused by environmental change. We survey the literature on this topic to identify relevant case studies that can be used in a meta-analysis aiming at identifying primary and secondary environmental drivers of human displacement within SSA, mapping its geographic patterns, and evaluating regional differences in susceptibility to environmental migrations.
1.2. Societal resilience
Many SSA countries are undergoing rapid societal transformation and economic development fueled by agricultural intensification and energy transitions. These transformative changes, besides the increasing population and changing consumption patterns, create a rising demand for natural resources and ecosystem services (Hoff, Reference Hoff2011; Howells et al., Reference Howells, Hermann, Welsch, Bazilian, Segerström, Alfstad, Gielen, Rogner, Fischer, Van Velthuizen, Wiberg, Young, Alexander Roehrl, Mueller, Steduto and Ramma2013). Despite the rapid development, the region remains to be the most dynamic for the examination of human impacts on environmental changes and the pressure exerted on the limited natural resources (Bazilian et al., Reference Bazilian, Rogner, Howells, Hermann, Arent, Gielen, Steduto, Mueller, Komor, Tol and Yumkella2011; Zickgraf et al., Reference Zickgraf, Vigil, de Longueville, Ozer and Gemenne2016). The impact of environmental changes (sudden or slow) on human communities depends on their vulnerability and adaptive capacities directly influencing their societal resilience (Adger et al., Reference Adger, Lorenzoni and O'Brien2009; Niva et al., Reference Niva, Kallio, Muttarak, Taka, Varis and Kummu2021). Historically migration has been an adaptive response to various environmental changes, social deprivation, hardship, and conflict but it also offers an opportunity to increase income, expand knowledge, have access to better resources and living conditions, and increased societal resilience (Scheffran et al., Reference Scheffran, Marmer and Sow2011). Social resilience is ‘the ability of a community to withstand external shocks and stresses without significant upheaval’ (Adger et al., Reference Adger, Kelly, Winkels, Luong and Locke2002). Our research also investigates the extent of societal resilience of SSA countries through evidence gathering from past environmental migration studies and how communities have coped with the magnitude and occurrence of individual or cumulative climate stressors. The severe drought event between 2010 and 2011 affected over 13 million people in the Intergovernmental Authority on Development (IGAD) region (FAO, 2018). As a result, stakeholders in these countries developed IGAD Drought Disaster Resilience and Sustainability Initiative (IDDRSI), aimed at improving regional capacity to withstand future drought events and environmental shocks. As the unprecedented number of global refugees and internally displaced people (IDPs) prompted the UN to develop the New York Declaration for Refugees and Migrants (Table 2) (UN, 2018), the combination of an increased number of IDPs and recurrent droughts in the African region also led to the creation of the regional initiative ‘Building resilience in Africa's drylands’ to strengthen the resilience of vulnerable pastoralist and farming communities (FAO, 2018). Environmental migration is a critical resilience strategy for communities affected by climate change, when the study of it is coupled with resilience-building strategies, it will contribute to reducing vulnerability and promoting sustainable growth in the region.
2. Method and data analysis
2.1. Literature search and filtering
This meta-analysis (e.g. Ahn & Kang, Reference Ahn and Kang2018) collects all the available case studies on environmental migration in SSA we were able to find in the scholarly literature. It analyses each one of them to extract data on the magnitude of human displacement, its direct and underlying environmental drivers, timing, and geographic patterns (Table 3). Data harvested from this collection of published studies are then analyzed with a variety of methods to provide an integrated understanding of the phenomenon across regions and migration events. This approach connects the dots among in-depth case studies that look at the broader pattern characteristic of environmental migrations in SSA. Specifically, we present and analyze findings from the literature to investigate the relationship existing between extreme climate events, sudden and slow-onset environmental changes as drivers for migration in SSA (Figure 1).
Summary of 87 migration cases induced by environmental changes in SSA
Note: According to the latest United Nations Department of Economic and Social Affairs (UN/DESA) report on 42 countries of SSA, the median average household (HH) size is 4.8 persons (UN/DESA, 2019). For the sake of accuracy of the values in this research, we use the approximated 5 persons per HH.
The identification of literature started with Scopus (668 articles), Science Direct, Web of Science, and Google Scholar search engines, leading to the source journals (combined 207 articles) (Figure 1). More details on the way the literature was surveyed, and the relevant articles and case studies were selected are presented in Supplementary materials, Annex 1.
Identified literature abstracts were skimmed and screened as per the eligibility criteria. Out of the 188 studies identified through search engine selection followed by prescreening, 94 papers were then excluded after further reading. The eligible studies were then examined to extract the following primary data and metadata: region/country, date of the event, type of extreme event or environmental change drivers, and the number of people displaced/migrating as a result of the event. Additional secondary information was also coded, including additional effects (e.g. losses, consequences of environmental changes and exacerbating factors), type of relocation, type of migration, the profession of IDPs and migrants, land ownership status of IDPs, and migrants, hydrological data analysis or modeling techniques used, source of data and destination countries/regions were also extracted. Further reading of all these publications (94 + 75 = 169), led to the removal of duplicated studies and of other articles with exclusion criteria (articles that lack the above-mentioned primary data) leaving us with 79 papers, documenting 87 case studies that were included in this meta-analysis, see Figure 1 and Table 3.
All the information gathered through screening of different kinds of literature on environmental change-induced migration in SSA is compiled in Table 3. The table reports the country of the migrants when the environmental change/event started and ended, the type of events, additional exacerbating drivers, the number of people displaced because of these events, destination country/region (depending on the type of migration: internal/external), the author and date of the publication and some remarks as additional information.
2.2. Qualitative comparative analysis
Many of the sudden and slow-onset environmental changes are directly responsible for the internal displacement of people in SSA, while others exacerbate migration outcomes through a complex network of causal links (Foresight, 2011; Mastrorillo et al., Reference Mastrorillo, Licker, Bohra-Mishra, Fagiolo, Estes and Oppenheimer2016). The difficulty of grasping the complex interplays of the causal links arises from the use of quantitative and qualitative analysis methods separately by oversimplifying the causalities relation without acknowledging the complexity of multiple mutually interacting drivers with possible domino effects (Bilsborrow & Henry, Reference Bilsborrow and Henry2012; Groth et al., Reference Groth, Ide, Sakdapolrak, Kassa and Hermans2020). Several scholars attempted to overcome the shortcomings of understanding the complex migration linkages, such as the use of agent-based model simulations (e.g. Hassani-Mahmooei & Parris, Reference Hassani-Mahmooei and Parris2012; Kniveton et al., Reference Kniveton, Smith and Wood2011) or integration of survey/census data with Bayesian belief networks (e.g. Drees & Liehr, Reference Drees and Liehr2015). In recent years to overcome the lack of scaling options in ethnographic studies, participatory methods such as mobility mapping were employed (de Campos et al., Reference de Campos, Bell and Charles-Edwards2017) or mobile network data were used to capture short-term migration patterns for large areas (Lu et al., Reference Lu, Wrathall, Sundsøy, Nadiruzzaman, Wetter, Iqbal, Qureshi, Tatem, Canright, Engø-Monsen and Bengtsson2016). Another under-utilized but critical deciphering method that incorporates the benefits of qualitative and quantitative approaches and overcomes the existing methodological challenges is QCA (Groth et al., Reference Groth, Ide, Sakdapolrak, Kassa and Hermans2020; Haeffner et al., Reference Haeffner, Baggio and Galvin2018; Ide, Reference Ide2015; Kirchherr et al., Reference Kirchherr, Charles and Walton2016; Ragin, Reference Ragin1984).
Charles Ragin developed QCA in the 1970s as a research method (Simister & Scholz, Reference Simister and Scholz2017). QCA is used to analyze multiple cases in a complex setting, and it helps to explain why some factors cause changes but not others. QCA is a very powerful tool for observing the influence of combinations of various factors on a certain phenomenon (Schneider & Wagemann, Reference Schneider and Wagemann2012). The method allows complex causal links to be traced by using a systematic set-theoretic approach. QCA has the potential to improve our understanding of the interactions between direct, indirect, and underlying migration drivers, which still have a substantial knowledge gap in the field of environmental migration.
This research aims to understand the various levels of the complex phenomena between environmental drivers and migration in SSA. The author found QCA to be a more suitable method for this research because of its special design for small to intermediate size cases (5–50), its integration of qualitative and quantitative analysis, and its comparative and configurational abilities (Czaika & Godin, Reference Czaika and Godin2021; Kirchherr et al., Reference Kirchherr, Charles and Walton2016; Ragin, Reference Ragin1984; Ragin et al., Reference Ragin, Drass and Davey2017; Schneider & Wagemann, Reference Schneider and Wagemann2012). This research reviewed 87 works on migration induced by sudden, slow onset, and accumulated environmental changes from 32 countries in SSA. Fuzzy-set qualitative comparative analysis software, developed by Charles Ragin and Sean Davey, was used to conduct the combinations of several causal conditions. The number of possible combinations is 2k, where k is the number of causal conditions (Ragin et al., Reference Ragin, Drass and Davey2017).
2.2.1. QCA terminologies
Outcome: The existence of migrants because of one or more environmental drivers' combinations.
Variables: The migration drivers’ causal condition: the combination of variables explaining the outcome variable (number of migrants).
Logical remainder: Non-observed cases (migration drivers that have not displaced people).
QCA follows the following six stages to analyze a set of crisp data (Rihoux & De Meur, Reference Rihoux and De Meur2009):
(1) Building a dichotomous data table
• Constructing a ‘truth table’
(2) Resolving contradictory configurations
• Analysis and Boolean minimization
(3) Consideration of the ‘logical remainders' cases
(4) Interpretation
A detailed step-by-step explanation of the analysis method is provided in Supplementary Annex 3. QCA is based on Boolean data type, where cases are either included or not in a set based on presence indication as 1 and absence indication as 0 (Ragin, Reference Ragin1984; Ragin et al., Reference Ragin, Drass and Davey2017; Schneider & Wagemann, Reference Schneider and Wagemann2012). The list of environmental migration drivers gathered from the 32 SSA countries was coded to binary data, resulting in 98 cases. These cases were categorized into binary values 1 ‘migration’ and 0 ‘no migration’ because of the direct environmental driver. The crisp data passed through ‘specify analyses’ and ‘standard analyses’ to produce complex, parsimonious, and intermediate solutions.
In addition to the QCA, mosaic graphs were constructed using the R studio computer program to demonstrate the QCA binary truth table for selected five combinations between the various environmental change drivers. These five categorical combinations were derived from Table 3, Figures 3 and 4, and a careful literature review that highlighted recurrent pathways of human displacement driven by environmental conditions.
2.3. Statistical correlation
The correlation between floods, storms, cyclones, heavy rain, drought, landslides, heat shocks, and other mixed accumulated events helps us to relate the significance of the events as an inducing factor to migration/internal displacement in SSA. The author used an excel data matrix to feed to IBM SPSS software to conduct a bivariate correlation analysis between the above-mentioned environmental migration drivers. This method computes the Pearson correlation coefficient with significant levels based on a two-tailed test. The author related the various events by putting the hydrological extremes in one group and also by grouping high- and low-rainfall events together as a comparison of various relationships. Positive correlation leads to a ‘p’ value less than and equal to 0.01 or 0.05, indicating the significance of the relationship between the variables.
Our research analyzes cases reported in the literature. Thus, the metadata we have built may be inherently biased toward underreporting, a limitation of our approach which is based on secondary literature data. The empirical analysis of the indirect and underlying drivers of migration is seldom based on a rigorous assessment of causality relationships. The statistical correlation analysis here is used to complement and triangulate the QCA with other research methods in order to generate a better understanding of the interdependence existing among possible drivers (Czaika & Godin, Reference Czaika and Godin2019, Reference Czaika and Godin2021), despite possible concerns existing about the use of correlation analysis in meta-analyses (Rosenthal & DiMatteo, Reference Rosenthal and DiMatteo2001).
3. Results and discussion
3.1. Migration patterns in SSA
Table 3 includes 27 case studies in which the migration flow directions are reported. We used this metadata to map the migration flows (Figure 2). The rest of the 56 cases are about internal migration within the country's borders, while some cases from Chad, Mali, Mayotte, Republic of Congo, Rwanda, South Sudan, and Uganda had no information on the destination of migrants. The desirability of the destination country/location (pull factor) highly depends on distance/easy access (e.g. Burkina Faso, Ethiopia, Ghana), the opportunity to diversify income sources (e.g. Burkina Faso, Ghana, Kenya, Nigeria, Uganda), food security (e.g. Cote d'Ivoire, Ethiopia, Ghana, Kenya, Mali, Senegal, Uganda, Zambia), safety during sudden extreme events, the location is usually selected by first emergency responders, national and regional administration (Holloway et al., Reference Holloway, Chasi, de Waal, Drimie, Fortune, Mafuleka, Morojele, Nhambiu, Randrianalijaona, Vogel and Zweig2013; NRC, Reference NRC2015) (southern African region, Madagascar, Mozambique, Malawi, Zimbabwe), and reduced exposure to extreme climate conditions (e.g. Namibia, Niger, Madagascar, Mozambique, southern African region, Tanzania). On the contrary, climate migration is considered also a gendered migration from the example of northwestern Ghana and Tanzania. Female migrants in Ghana used to be denied from earning additional income when food insecurity persists during the dry season because of the bride price paid by the husband (Abdul-Korah, Reference Abdul-Korah2011; van der Geest, Reference van der Geest2011). The bride price gave the husband the right to control the woman from participating in circular migration and diversify income sources for the family. In the case of Tanzania, unmarried female migrants still lack access to land-based livelihood despite the vulnerable state they hold in the society, which led them to only engage in off-farm labor jobs and migrate (Atuoye et al., Reference Atuoye, Luginaah, Hambati and Campbell2020).
Map of the case studies included in this paper and the known environmental migration flows.
In Figure 2, the southern African region where all the arrows are converging in Botswana represents the centroid of the region. For the purpose of this study, Angola, Zambia, Zimbabwe, Mozambique, Namibia, Botswana, Swaziland, South Africa, and Lesotho are the countries taken into consideration to find the centroid location. In the various articles consulted for this research, people have migrated from Botswana, Madagascar, Mozambique, Namibia, South Africa, Zambia, and Zimbabwe within the South African region with no information on the specific destination, therefore this centroid point used in this figure represents the South African region at large (Bola et al., Reference Bola, Mabiza, Goldin, Kujinga, Nhapi, Makurira and Mashauri2013; Brouwer & Nhassengo, Reference Brouwer and Nhassengo2006; Holloway et al., Reference Holloway, Chasi, de Waal, Drimie, Fortune, Mafuleka, Morojele, Nhambiu, Randrianalijaona, Vogel and Zweig2013; IDMC, 2019; NRC, Reference NRC2015; Parkinson, Reference Parkinson2013).
The evidence analyzed in this study (Figure 2) demonstrates that environmental migration in Africa has been internal to SSA for the period between 1990 and 2021, particularly clustering on a regional level, in agreement with the first African migration report (Abdiker et al., Reference Abdiker, Cessouma, Kidd, Maze, Konsimbo, Adepoju, Fumagalli and Nyabola2019).
3.1.1. Environmental events triggering migrations
According to the metadata in Table 3, the two main environmental factors leading to human displacement in SSA as a result of environmental change (slow-onset, sudden, and accumulated) are associated either with high rainfall (i.e. heavy rainfall, high-intensity precipitation events) or water scarcity and drought. Figures 3 and 4 show two schematic diagrams with the interrelationship among environmental drivers in the case of high rainfall and water scarcity, respectively.
Schematic illustration of migration driven by high-rainfall events (including storms, flooding, cyclone, and heavy rain). In parentheses we indicate the corresponding case study in Table 3. The two arrow bars indicate directions of increasing intensity and severity. For instance, this figure is read by following the arrow starting from each driver. For example, cyclones caused direct displacement (first red arrow) in several case studies (3,10,29,30,31,32,33,34,39,40,42,43,44,46,70,87 (see Table 3)), cyclones (second red arrow) also led to flood which later caused displacement and following the third red arrow cyclone caused sea and river level rise which caused flood, riverbank bursting, and landslides, thereby displacing people.
Note: Severity is defined as the cumulative effect of harsh living conditions created by environmental change that pushes people to abandon their homes. The brown scale arrow on the right side of the figure points toward ascending level of severity. Intensity is defined as the amount of force exerted by and the persistence of environmental change-induced shocks. The orange arrow indicates the increase in intensity.
Schematic illustration of migration driven by low rainfall and water scarcity. In parentheses we indicate the corresponding case study in Table 3. The two arrow bars indicate directions of increasing intensity and severity. The number of displaced people for each event is shown in Table 3 and later used for QCA.
Excessive precipitation can be associated with cyclones, storms, and heavy rainfall. Cyclone and storm events are becoming a more common and frequent occurrence in islands and coastal countries of SSA and their immediate neighbors. These extreme sudden events caused displacement/migration as individual direct drivers or as a consequence of one another (‘domino effect’). Secondary to measures taken by local/regional governments to evacuate affected people, organizations like NRC, IDMC, IFRC, and UNOCHA are among the first respondents for assistance. The actual number of these migrants is usually published in reports provided by these first respondents and numbers could vary from all migrants who decided to permanently stay in the resettlement centers to those who returned back to what's left of their residence (Brouwer & Nhassengo, Reference Brouwer and Nhassengo2006; Holloway et al., Reference Holloway, Chasi, de Waal, Drimie, Fortune, Mafuleka, Morojele, Nhambiu, Randrianalijaona, Vogel and Zweig2013; IDMC, 2020; IDMC & NRC, 2020; IFRC, 2009, 2012a, 2016; NRC, Reference NRC2015; Parkinson, Reference Parkinson2013; Simatele & Simatele, Reference Simatele and Simatele2015; UNOCHA, 2012). Heavy rain caused displacement, partially in the Central, East, and South African regions. Heavy rain accompanied by a cyclone and storm-induced displacement in the Central African Republic, Democratic Republic of Congo, Republic of Congo, Burundi, Rwanda, Somalia, Zambia, Mozambique, and Namibia while the variation of heavy rain and lack of rain-triggered migration within Kenya from Samburu to Meru, Isiolo, Nanyuki, Nyeri, Karatina, Thika, and Nakuru counties. The above-listed countries faced large-scale property and infrastructure destruction, loss of life, agriculture, and livestock as the rapid events subsided (Caveirinha, Reference Caveirinha2013; IDMC, 2019, 2020; IFRC, 2016; NRC, Reference NRC2015; Wiles et al., Reference Wiles, Selvester and Fidalgo2005). Beside the loss of livelihood, Mozambique's health sector was highly affected by the year 2000 extreme event leaving communities stranded (Brouwer & Nhassengo, Reference Brouwer and Nhassengo2006). For small island countries like Comoros and Mayotte, the lack of less-exposed resettlement locations to shelter from future events was an additional challenge they had to face alone (IDMC, 2020).
Not only did these extreme low-pressure events displace people from their homes but also triggered a set of other environmental changes such as the rising of river/seawater levels and inundation (see case studies 46, 56, and 57, respectively). In Chad, Nigeria, the Republic of Congo, Sudan, and South Sudan the bursting and overflow of riverbanks and reservoirs (e.g. Kariba and Cabora Bassa dams) led to direct displacement (IDMC, 2020; IDMC & NRC, 2020). In other cases, human displacement was induced by landslides and floods. Mozambique and Namibia were the only two countries that had displacement unrelated to infrastructural damages. Landslides as a consequence of extreme rainfall events displaced many people in East Africa (Burundi, Kenya, Rwanda, Tanzania, Uganda, Zimbabwe) (DREF, 2012; IDMC, 2019; 2020; IDMC & NRC, 2020; UNOCHA, 2019). In Madagascar, the base of the Easter Mountain mudslide caused tremendous damage to 40,000 homes and killed 13,000 people triggering internal human displacement and migrations (Naeraa & Jury, Reference Naeraa and Jury1998; NRC, Reference NRC2015).
The combination of additional data extracted from the international disaster database in Supplementary Annex 1 with metadata from Table 3 shows that 38 out of 47 countries experienced immediate displacement because of floods. Floods did not just cause a one-time displacement; in countries such as Niger, Somalia (Gu rains), and Sudan the resettlement centers were flooded again, leading to 'secondary displacement’ to higher grounds (IDMC, 2020; IDMC & NRC, 2020). In some cases, flood and water-level rise did not cause immediate displacement but rather slow-onset adverse conditions, which are shown in gray in Figure 3 to stress the fact that they could occur either as extreme events or as slow-onset environmental drivers. The cascading connection of these migration drivers shows the severity and intensity scale of the events for people to stay in their habitual residence. The complex web of high precipitation-induced environmental impacts in Figure 3 resulted in internal displacement and migration in southern Africa, eastern Africa, and part of the Central African region (see Table 3) either as single drivers or in conjunction with others. The lack of well-documented, long-term hydro-climatological records of extreme events that have induced migration in the region points to an important gap in understanding the nexus between high precipitation and migration.
We then look at migrations induced by low rainfall and water scarcity (Figure 4). We found that dry spells and heatwave/extreme temperatures are associated with slow-emerging drivers of migrations within SSA. The crucial difference between Figures 3 and 4 is that each driver in Figure 3 causes migration on their own and in some countries cumulatively but in Figure 4 the cumulation of the drivers can only cause long-term migration. SSA is known for its high dependence on rain-fed agriculture to feed a large portion of its population. Farmers in many countries have developed an adaption method in the form of circular migration to make sure that they have enough food to feed their families. When facing an extended dry season, they migrate to the nearest city with more job opportunities and work as laborers. When the rainy season approaches, they return to their farm to prepare the land and work in the growing season, as reported by sample community surveys in Burkina Faso (Soum, Oudalan Nahouri, and Bougouriba provinces), northwestern Ghana (Nadowli district), and Kenya (Samburu county) (Abdul-Korah, Reference Abdul-Korah2011; Henry et al., Reference Henry, Schoumaker and Beauchemin2004; Ng'ang’a et al., Reference Ng'ang’a, Bulte, Giller, McIntire and Rufino2016; Rademacher-Schulz et al., Reference Rademacher-Schulz, Schraven and Mahama2013; van der Geest, Reference van der Geest2011). In these cases, only some family members migrate often temporarily without displacing the entire household in the form of rural–rural/rural–urban or north–south/south–north.
Nigeria, Tanzania, and Zambia are countries affected by heatwaves which were not direct drivers of displacement. Dry spells and extreme heat were associated with water scarcity in Mauritania, Rwanda, and Somalia, but they did not induce direct displacement. Instead, they contributed to the recurring drought event in Burkina Faso, Ethiopia, Ghana, Kenya, Mali, Mauritania, Nigeria, Rwanda, Somalia, Tanzania, Zambia, and Zimbabwe. These drought events displaced many people and trapped others in further food insecurity, leading to poverty and famine. Trapped communities are those people facing severe vulnerability to a dry climate, poverty, and food insecurity but do not have the resource to migrate to a better place (Black et al., Reference Black, Arnell, Adger, Thomas and Geddes2013; Nawrotzki & DeWaard, Reference Nawrotzki and DeWaard2018; Simatele & Simatele, Reference Simatele and Simatele2015). Even pastoralist communities that are well adapted to extreme weather events are facing water scarcity and grazing land degradation and seeking a more permanent settlement option in the case of West Pokot, Kenya, and Somalia (Ng'ang’a et al., Reference Ng'ang’a, Bulte, Giller, McIntire and Rufino2016; Tsui et al., Reference Tsui, Rugsdale and Shirwa1991). The combination of both rapid extreme events (flood, heatwave) and slow-onset events (drought) caused a large people displacement in Zimbabwe, Nigeria, Ethiopia, and Burkina Faso (Bola et al., Reference Bola, Mabiza, Goldin, Kujinga, Nhapi, Makurira and Mashauri2013; Dillon et al., Reference Dillon, Mueller and Salau2011; Henry et al., Reference Henry, Schoumaker and Beauchemin2004; IDMC, 2020; IDMC & NRC, 2020). The review of the literature (see Table 3) shows the recurrence of some extreme events for over 20 years.
Ethiopia had severe recurring droughts followed by accumulated environmental changes (see Figure 4) such as famine, ecological degradation, dryland expansion, famine, and locust invasion in 1960–1980, 1984–1994, 1985, 1994–2009 (Bilsborrow & DeLargy, Reference Bilsborrow and DeLargy1990; Ezra, Reference Ezra2000; Reference Ezra2001; Ezra & Kiros, Reference Ezra and Kiros2001; Gray & Mueller, Reference Gray and Mueller2012; IDMC, 2020; IDMC & NRC, 2020; Jacobsen & Wilkinson, Reference Jacobsen, Wilkinson and Weiner1993; Kane, Reference Kane and Brown1995a; Rahmato, Reference Rahmato1991). These events caused a displacement of almost 2 million people in total. In the last couple of years, the recurring drought was accompanied by floods and further displaced 367,000 people. The accumulated environmental changes over the 20 years that expanded the dryland and degraded the quality of the farmlands caused by high population growth followed by agricultural intensification could be the reason that exposed the communities to flood occurrence in 2019. Somalia also had arable land degradation and water scarcity in early 1970, which later developed into severe drought in 1973–1975 and led to food insecurity (Gebremedhin, Reference Gebremedhin1991; IDMC, 2020; IDMC & NRC, 2020; Molvær, Reference Molvær1991; Tsui et al., Reference Tsui, Rugsdale and Shirwa1991; Westing, Reference Westing1994). The unexpected tropical cyclone in 2018 caused heavy rainfall in the Arabian desert also created the perfect environment for locust breeding; the wind of yet another cyclone in 2019 facilitated the migration of locusts to East Africa exacerbating food insecurity (IDMC & NRC, 2020; Salih et al., Reference Salih, Baraibar, Kemucie Mwangi and Artan2020; WMO & FAO, 2016).
Besides other underlying non-environmental reasons in Somalia, the accumulated environmental changes led to slow-onset severe drought events developed over the years displacing 760,000 people. The neighboring Kenya and Tanzania also had accumulated and slow-onset environmental changes ranging from land scarcity, degradation, livestock diseases, and heatwave to drought resulting in the internal displacement of 90,420 people (Atuoye et al., Reference Atuoye, Luginaah, Hambati and Campbell2020; Charnley, Reference Charnley1997; Maurel & Kubik, Reference Maurel and Kubik2014; Mwakipesile, Reference Mwakipesile1976; Ng'ang’a et al., Reference Ng'ang’a, Bulte, Giller, McIntire and Rufino2016; Odgaard, Reference Odgaard1986). In the last decade, Kenya experienced 28 major droughts and Samburu county is one of the few places that was affected by both heavy rainfall and dry spell robbing the chance of farming entirely. Most families in the county now depend on remittances from family members who migrated to the neighboring counties (Ng'ang’a et al., Reference Ng'ang’a, Bulte, Giller, McIntire and Rufino2016).
Expanding the search from east to southern Africa showed us that the extended drought in Mozambique halved agricultural production and caused widespread food insecurity before the cyclone exacerbated the migration of people both permanently and temporarily (IDMC, 2015, 2020). In Zambia and Zimbabwe as well, the slow onset of severe drought in the late 1990s and early 2000 displaced more than 3 million people internally and externally to South Africa, respectively. From West SSA, the author found few temporal data on Mali, Mauritania, and Burkina Faso showing the development of accumulated environmental changes into slow-onset events (see Table 3: 1, 2, 26, 27, and 28) leading to displacement. These cases clearly demonstrate how soil erosion, deforestation, land degradation, desertification, water scarcity, and low rainfall/dry spell resulted in severe drought in these countries in 1960–1998, 1980–1990, and 2019.
3.2. QCA output
QCA is used to disentangle the complex causal conditions that lead to environmental migration from the collected datasets. It analyzed the combination of the direct environmental drivers whereas rapid/sudden events related to high-intensity downpours are grouped as one variable. These variables are (storm, cyclone, heavy rain), flood, landslides, heat shock, drought, and dry spells. From reviewing 87 case studies on environmental migration, we found that the direct drivers shown in Figure 5 have 98 cases with evidence of direct displacement immediately or through a certain length of time, therefore, the QCA output shown below focused on the following cases: flood (20), storm (8), cyclone (10), heavy rain (23), drought (13), landslides (9), heat shock (4), and dry spell (11). The indirect drivers (Figure 6, y-axis) have also contributed to the displacement of more people in SSA but not exclusively. One or more of these drivers accompanied the direct drivers creating widespread vulnerability and exacerbating the effect of direct drivers. In total, we had 51 cases from the secondary drivers, that is, famine (4), poverty (1), forest fire (1), locust invasion (2), dryland expansion (1), ecological degradation (6), river water-level rise (1), arable/grazing land degradation (7), food insecurity (7), water scarcity (5), riverbank bursting (4), land scarcity (3), agricultural intensification (1), farmland infertility (3), rainfall season shortening (2), soil erosion (1), desertification (1), and deforestation (1).
QCA parsimonious results.
Mosaic plot of direct (X-axis) vs indirect drivers (Y-axis) of environmental change.
Environmental migration decisions and drivers can only be explained by the interaction of several factors (conjunctural causation) (Foresight, 2011). The mosaic graph constructed from the QCA truth table demonstrates five distinct combinations between the various environmental change driver categories, we have encountered so far. The push from these environmental changes varies in the degree, scale, and intensity of the event. About 54% of the cases covered in this article focus on displaced people from rural communities relying on farming and pastoralism. Figure 6 shows the connection between 28 cases of arable/grazing land degradation and 21 cases of food insecurity as a result of direct migration drivers, respectively. Our finding further shows that 26 cases of ecological degradation and 16 cases of land scarcity were principal contributors to the displacement of communities dependent on agricultural production. The summary of direct and indirect drivers highlights that the agricultural community (farmers and pastoralists) dependent on seasonal rainfall is the most vulnerable and exposed to displacement as a result of environmental change extremes. Naturally, being vulnerable and exposed to recurrent environmental extreme events pushes people to change their place of residence, but the adaptive capacity in terms of economic and social status becomes the deciding factor.
Figure 7 displays the economic and social hardship faced by communities in terms of famine, poverty, and food insecurity as a push factor for out-migration. These hardship conditions were caused or worsened by climate change or extreme environmental shocks. Conversely, individual or household vulnerability to environmental migration can be enhanced by famine, poverty, and food insecurity. While here we are unable to evaluate the synergy between these direct and indirect drivers of environmental migration, we investigate their co-occurrence (Figure 7). Once these drivers have set roots, the occurrence of extreme hydrological events causes an immediate displacement from their residence. This mosaic plot shows the role of non-environmental underlying drivers on environmental migration. Our meta-analysis found that Tanzania, Nigeria, Rwanda, Burkina Faso, and Somalia were the most vulnerable to environmental migration as a result of economic and social hardship (see Supplementary Annex 4: Africa map with all environmental drivers) (Figures 8 and 9).
Mosaic plot of economic and social hardship (Y-axis) vs vulnerability to environmental change (X-axis).
Mosaic plot of the relationship between low- and high-rainfall events.
Mosaic plot of the relationship between low rainfall, degradation, and heat shock.
In several cases, migration occurred as a result of high-rainfall events (i.e. heavy rain and flood) happened in drylands or areas highly affected by drought and dry spells (39 cases, 67%). Heavy rainfall either destroyed the crop that has already suffered because of the extended period of no rain (Benin, Mali, Niger, and Nigeria) or ended up flooding the drylands (Ethiopia, Somalia, Tanzania, Zambia, Zimbabwe, and Botswana) leading to the displacement of people. The combination of low rainfall with extreme heat events and degradation of natural resources was highlighted in countries like Benin, Ghana, Niger, and Tanzania (Figure 3) with 33% of the total cases associated with arable and grazing land degradation.
The last mosaic plot (Figure 10) shows environmental migration cases related to poor infrastructure planning that led to torrential flooding from the overflowing of dams and reservoirs, bursting of riverbanks, and water-level rise in rivers. The combination of these drivers was highlighted in Chad, Nigeria, Central Africa Republic, Mozambique, Angola, and Zambia.
High rainfall vs poor infrastructure.
Figure 5 shows the optimal result (‘parsimonious solution’) obtained from the QCA on direct drivers. This paper only reports the parsimonious solution as it is considered the most robust and capable of reducing redundant variables (Baumgartner & Thiem, Reference Baumgartner and Thiem2017; Groth et al., Reference Groth, Ide, Sakdapolrak, Kassa and Hermans2020). See Supplementary Annex 3 for more details.
The parsimonious solution from QCA disentangles the migration decision pathways as a result of extreme hydrological events and drought. This method further analyzed the complex and intermediate pathways that could or could not lead to environmental migration (see Supplementary Annex 4). These pathways include – but are not limited to – the combined incidence of ‘flood’, ‘landslide’, ‘drought but not dry spell’. The results show the probable occurrence of displacement associated with floods in Central, East, and South Africa (20 cases, 62%) or landslides (9 cases, 28%) in East, West Africa, and Madagascar. The analysis further emphasizes that Mauritania and Eritrea are prone to drought-induced displacement (without dry spell) (5 cases, 15%) while Zambia is the only country showing the cumulative effect of flood, landslide, and drought on human displacement.
3.3. Statistical analysis
In addition to the above QCA and empirical analysis, we also measured the strength and direction of the linear relationship between variables or, in our case, the number of people displaced by each extreme event. The statistical correlation yields that there is no significant cross-correlation between multiple drivers. The detailed result of the correlation output from SPSS is provided in Supplementary Annex 5.
The correlation of multiple drivers as environmental migration inducers mainly depends on the selection of the destination. The destination of migrants is mainly governed by distance/access to better resources, climatic conditions, better income/job opportunities, and better living conditions/safety. The statistical analysis method does not consider this main governing factor. Figure 11 shows the environmental migrant population distribution from the case studies gathered. Each dot represents 4000 people, making Zimbabwe, Rwanda, Somalia, and Mozambique the top four countries with the highest environmental migrants in descending order (see Table 3). As previously mentioned, we also consulted the international disaster database (UCLouvain, Reference UCLouvain2020) from 2000 onward to complement the results generated from the review of 79 articles from the literature. Figure 12 shows the total environmental migration cases in SSA, see Supplementary Annex 2.
Environmental migrants' distribution per 32 countries.
Total environmental migration in SSA (flood is read as the combination of flood from the EMDAT dataset and flood from literature. Other drivers without the extension EMDAT are collected from the literature).
3.4. Conceptual framework analysis
We then propose a conceptual framework as a heuristic approach to structure the environmental drivers of migration with their underlying non-environmental factors into a systematic concept flow from evidence gathered in the literature. The literature review (Table 3) highlighted how environmental migrations are not the result of only environmental change drivers but of their complex inter-linkage with underlying non-environmental factors that enhance a population's vulnerability to environmental stressors. The following conceptual framework provides a clearer representation of the complex interrelationship of environmental migration drivers in SSA. The cumulative centripetal movement of drivers to the center (migration and displacement) is experienced when non-environmental factors existed for many years in each country. This centripetal movement of drivers is categorized into three main orbits based on their evolvement process.
The movement of people in search of a large quantity of untapped natural resources has existed for as long as history remembers. Nevertheless, what happens when these resources must be shared with more and more people? What about when other non-environmental factors prohibit the growing population from using these resources freely? Then communities would start to overexploit/deplete the resources available in their vicinity. The overuse of resources leads to degradation, exposing them to be severely impacted because of environmental changes over time. This theoretical framework of migration drivers has been translated for the 87 case studies of SSA as shown in Figure 13. Non-environmental underlying drivers such as poorly constructed infrastructures, limited economic development, high unemployment rate, lack of preparedness of local governments, unstable economic and political situation, some social and cultural norms (e.g. male role as a provider of the household, similar religion, and ethnicity; Davis et al., Reference Davis, D'Odorico, Laio and Ridolfi2013; Rademacher-Schulz et al., Reference Rademacher-Schulz, Schraven and Mahama2013; van der Geest, Reference van der Geest2011)), slow technological advancement, corruption, threats to livelihoods, lack of climate change policy, and law (Abdiker et al., Reference Abdiker, Cessouma, Kidd, Maze, Konsimbo, Adepoju, Fumagalli and Nyabola2019), lack of integrated natural resources management and demographic studies with no actionable impact left communities vulnerable and with little adaptive capacity to environmental changes in SSA.
Conceptual framework of environmental migration drivers in SSA.
Note: The figure depicts the three categories of environmental migration and internal displacement drivers translated from the 87 case studies in concentric elliptical layers. The figure is read from the out most layer of underlying non-environmental drivers that create suitable conditions for the emergence of indirect drivers exposing communities to slow-onset environmental changes. These indirect drivers in turn create a more suitable condition for the rising of direct drivers characterized by accumulated and extreme events. For instance: poorly constructed infrastructure due to the community's lack of adaptive capacity (outer layer) creates susceptibility of riverbanks for bursting (second layer), thus intensifying the effect of flood and landslides (third layer), thereby causing internal displacement and environmental migration (centroid).
From the early 1970s to 2000 population of SSA kept increasing along with the inequality in access to resources (Naudé, Reference Naudé2010). During this time SSA was strongly affected by natural disasters, particularly in Ethiopia (54 disaster occurrences), Tanzania (38), Kenya (28), Sudan (32), Madagascar (35), South Africa (56), Mozambique (46), and Nigeria (28) (Krishnamurthy, Reference Krishnamurthy2012; Naudé, Reference Naudé2010). These countries were found to be most susceptible to internal displacement. The outer ‘orbit’ of underlying non-environmental drivers (Figure 13) contributed to the emergence of accumulated environmental changes (indirect drivers) which over time became critical factors for direct migration drivers. Ecological degradation, arable/grazing land degradation, dry-land expansion, declining freshwater resources, desertification, soil infertility, land scarcity, loss of livestock, locust infestation, forest fire, malaria resurgence, and agricultural intensification fueled by high population growth are the indirect environmental/land-use drivers that enhanced the occurrence of more rapid/extreme events in the subsequent years. The direct drivers (Figure 13, innermost ‘orbit’) such as floods, storms, cyclones, heavy rain, heat shock, dry spell, landslides, and drought led to the human displacement and migration in SSA either alone or in combination with other environmental drivers. Thus, while direct and indirect environmental factors are important drivers of environmental migration in SSA, the underlying non-environmental factors are also expected to play a critical role in triggering internal displacement.
4. Conclusion
Environmental migrations have been a major focus of research on environmental change impacts. Indeed, empirical studies on environmental change drivers of migration have notably increased in the last few decades. To date, there is little agreement within the scientific community on the extent to which climate and environmental factors influence human mobility. Based on the secondary data collected from 32 countries, there are 149 cases in which various degrees of environmental changes contributed to migration in this paper.
This review quantified the effects of rapid, slow, and accumulated environmental changes on human mobility to better understand migration flows in SSA. The result from the theoretical analysis (Figure 13) and also the QCA reveals that the complex web of causal links are the main drivers of environmental migration in SSA while individual extreme or slow-onset events with high intensity and severity are very critical migration inducers. However, the existence of non-environmental underlying factors plays a significant role by exposing the population to cumulative environmental changes and hindering their resilience. The widely accepted notion that developing countries will endure the brunt of global environmental change in the not-too-distant future (Clement et al., Reference Clement, Rigaud, Sherbinin, Jones, Adamo, Schewe, Sadiq and Shabahat2021; Gahouma-Bekale, Reference Gahouma-Bekale2021) establishes the great importance of understanding these links (Carr, Reference Carr2005). The long-term hydro-climatological records of extreme events that have induced migration in the region point to an important gap in understanding the nexus between high precipitation and migration. This paper fills the gap in knowledge by presenting well-documented evidence of environmental migration cases from more than two decades of records. It investigates the complex web of environmental drivers and how they lead to human displacement. The quantitative and qualitative analysis further evaluates the pattern and weight of each migration driver in SSA.
The review of the literature on African environmental migration encountered some limitations such as the substantial lack of holistic evidence, the scarcity of comprehensive temporal and spatial data to fully understand the extent of human displacement in-depth, the lack of information about the future of IDPs permanent or temporary destination or secondary displacement, and the scale of data collection limited to the scope of (often scattered) past research and data collected by first responders. Limitations associated with the use of these metadata can be seen in the results of the statistical analysis. In fact, the statistical analysis shows no significant cross-correlation between multiple drivers of migration in SSA, in complete disagreement with the evidence, gathered in the literature as well as the results of theoretical, and QCA. Statistical analysis can only provide accurate and reliable outcomes when based on well-documented, systematic, and relatively complete datasets (Groth et al., Reference Groth, Ide, Sakdapolrak, Kassa and Hermans2020; Ide et al., Reference Ide, Brzoska, Donges and Schleussner2020). Therefore, with the currently available data compiled from the literature, the use of statistical correlation analysis may yield inaccurate results. ‘This requires transparency, information, and data-sharing between the African States on migration, and an openness to engage with the full range of stakeholders that play a significant role in migration knowledge production’ (Abdiker et al., Reference Abdiker, Cessouma, Kidd, Maze, Konsimbo, Adepoju, Fumagalli and Nyabola2019). Because about 85% of the African migration is an active cross-border movement within SSA there is a need for more data to document this phenomenon (Achieng et al., Reference Achieng, Fadil, Righa, Adepoju, Fumagalli and Nyabola2020). The adaptation and resilience of communities to cope with the scale of extreme and slow-onset environmental changes are quite limited. In fact, about 93% of the assessed countries had forced displacement as a coping mechanism to the environmental change drivers while the rest practiced short-term circular migration to increase their resilience. According to the IPCC estimates, the projected impacts and risks of frequent and intense extreme weather and climate events under 2°C and above climate warming (2021–2100) are expected to exacerbate and undermine food and water security by threatening water availability, food production, and access in SSA, likely leading to severe malnutrition and micro-nutrient deficiency (high confidence) (Pörtner et al., Reference Pörtner, Roberts, Poloczanska, Mintenbeck, Tignor, Alegría, Craig, Langsdorf, Löschke, Möller, Okem, Pörtner, Roberts, Tignor, Poloczanska, Mintenbeck, Alegría, Craig, Langsdorf, Löschke, Möller, Okem and Rama2022; Trisos et al., Reference Trisos, Adelekan, Totin, Ayanlade, Efitre, Gemeda, Kalaba, Lennard, Masao, Mgaya, Ngaruiya, Olago, Simpson, Zakieldeen, Langsdorf, Löschke, Möller and Okem2022). Food insecurity, in turn, is likely to contribute to humanitarian crises in more vulnerable and exposed communities, thereby leading to human displacement. African migration studies have often focused on fluxes from Africa to Europe while cross-border migration within SSA has remained understudied despite the greater number of people it affects (85% of the total migrants) (Achieng et al., Reference Achieng, Fadil, Righa, Adepoju, Fumagalli and Nyabola2020). The evidence-gathering process of this meta-analysis suggests that intra-African migrations are not well captured by past studies. Fortunately, some of the migratory fluxes are recorded by local governments, international organizations, and humanitarian groups. Thus, gathering and analyzing such records can shed light on environmental drivers of human displacement within SSA. To advance the understanding of this complex regional phenomenon there is a need for a more holistic approach with research focusing on the entire SSA region instead of single countries. This review has shown that understanding the link between migration and extreme environmental changes mainly depends on disentangling multi-causality relations, including not only environmental factors but also the vulnerability of SSA people to poverty, famine, and water scarcity.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/sus.2023.5.
Acknowledgment
Sinafekesh Girma Wolde is a research fellow under the NEWAVE – Next Water governance project at Politecnico Di Milano.
Author contributions
All three authors (Sinafekesh Girma Wolde, Paolo D'Odorico, and Maria Cristina Rulli) conceived the study and wrote the paper together. Sinafekesh Girma Wolde additionally performed data gathering and analysis.
Financial support
The NEWAVE project has received funding from the European Union's Horizon 2020 research and innovation program under grant agreement No. 861509.
Conflict of interest
The authors declare no conflict of interest and the study is original except where due acknowledgment is made in the text.
Research transparency and reproducibility
This research is reproducible by following the methods in which we transparently explain how the data were gathered and analyzed.
Open access
