Introduction

Migration is a complex phenomenon, intertwined with a multiplicity of economic, social, and security outcomes, which have themselves become increasingly complex with globalization (IOM 2020). As a concept, migration encompasses a wide variety of movements and engages people with multiple experiences and backgrounds. Migration can be voluntary — when people decide to migrate, or forced by existing circumstances, such as extreme weather events (Mallick and Schanze 2020; Wiegel et al. 2021). The decision to migrate (or not) is influenced by social (e.g., place-based attachment: the bond between people and places), political (e.g., democratic institutions), economic (current or discounted income), demographic (e.g., age structure), and environmental factors (e.g., hazard frequency) (Foresight 2011a; Adams 2016; IOM 2020; Biswas and Mallick 2021; Khalil and Jacobs 2021). By offering life-improving opportunities as a livelihood strategy, migration can provide an exit for populations exposed to crises, conflicts, and environmental hazards (Sly and Wrigley 1986; Naudé 2008; Sheffran et al. 2019; IOM 2020).

In addition to the rich migration-environment literature (Hunter 2005; Abu et al. 2014; Cattaneo and Peri 2016), there has been increasing recognition that some populations exposed to environmentally hazardous conditions do not move (Bhusal et al. 2021; Wiegel et al. 2021). Emerging scholarship on non-migration reveals that staying can be voluntary, as opposed to only involuntary as a result of the incapacity to move (Mallick and Shanze 2020; Bhusal et al. 2021). This article contributes to environmental non-migration scholarship by reviewing literature specifically on the drivers of environmental non-migration decisions in Africa.

Background and study motivation

Environmental migration has accompanied the global surge in environmental hazards in the last three decades, often pushing affected population to move to new places (Idemudia and Boehnke 2020). Research suggests that the rising frequency of extreme climate-induced events is contributing to increasing migration (Afifi 2011; Bhattacharyya and Wer 2012; UNDESA 2016, 2017). For instance, it was predicted that climate change alone accounts for up to 25 million environmental migrants (Ionesco et al. 2017). The United Nations Environment Programme equally predicted that 50 million persons were displaced in 2010 due to environmental hazards. Estimates further hold that by 2050, over 700 million environmental migrants will be registered, that is, out of every 11 people, 1 person will be displaced due to environmental hazards (Christian Aid 2007). The latter statistic provided the motivation for the geographic focus of this study, especially since Africa is also projected to supply the highest percentage of environmental migrants in the future (Black et al. 2011; Borderon et al. 2019).

Migration research has typically focused on movement as a key livelihood strategy (Idemudia and Boehnke 2020). However, intensifying climate-induced disasters — particularly in Africa — has extended migration research into its role in adaptation to environmental challenges and its related livelihood outcomes (Hunter 2005; Mallick and Vogt 2012; Balgah and Kimengsi 2019; Mpandeli et al. 2020). Recent scholarship has also explored decisions to not migrate (Biswas and Mallick 2021; Bhusal et al. 2021). While migration research is not new to Africa,Footnote 1 this shift in contemporary research towards non-migration intentions, drivers, decisions of environmental migrantsFootnote 2 is very contemporary (Doevenspeck 2011; Graham 2020). Such a topical emphasis is justified for several reasons. First, surging extreme events in Africa can yield substantial increases in actual and potential environmental migrants (Boko et al. 2007; Marchiori et al. 2012; IOM 2020). Between 75 and 250 million residents are estimated to be exposed to water stressors (e.g., droughts and floods) in Africa alone in the twenty-first century, an increase from 130,000 between 1960 and 2000 (Boko et al. 2007; Marchiori et al. 2012). Second, the potential for non-migration is intensified as environmental challenges further impoverish poor households — creating poverty traps by pushimg them deeper into poverty — while also pushing non-poor into poverty (Carter and Barrett 2006). Third, weak formal institutions challenge disaster risk reduction efforts (Balgah 2011; Borderon et al. 2019; Balgah and Kimengsi 2022) .

Environmental non-migration

To avoid a conceptual muddle, it is necessary to differentiate environmental migration from non-migration. Environmental non-migration defines situations where residents do not move from their areas of habitation despite the occurrence of environmentally stressing events, such as floods, droughts, and earthquakes (Renaud et al. 2007; Foresight 2011b). An environmental migrant is a person who, mainly due to sudden or slow developing environmental changes adversely affecting living conditions, leaves their habitual residence either voluntarily or because no other option remains (IOM 2014). Environmental migration is, therefore, an effort to adapt to environmental challenges and/or escape from environmental threats (Biswas and Mallick 2021). Yet, not every affected person migrates when extreme weather-related events occur (Buchenrieder et al. 2017; Khan et al. 2018), and, as a result, it is necessary to also examine environmental non-migration. Understanding how and why decisions to migrate or not are made can inform the policy processes to ensure successful outcomes linked to non-migration.

Initially, the concept of environmental non-migration was used to describe the involuntary immobility of those without the means to move, and especially in contexts characterized by slow — or rapid-onset environmental hazards and disasters. In this way, environmental non-migration was linked to the concept of trapped populations (Ayeb-Karlsson et al. 2018; Wiegel et al. 2021). It soon became clear, however, that staying is not always involuntary and that some populations may even stay when negative effects of environmental hazards are imminent (Adams 2016; Mallick and Schanze 2020). This called for an expansion of the non-migration concept to include persons voluntarily (or involuntarily) living in places that are vulnerable to environmental change (Mallick et al. 2020). The concept of being trapped describes the inability to move away from risk-prone environmental areas, as a consequences of inadequate capacity and resources (Foresight 2011a). As such, the notion of trapped populations recognizes the existence of complex human–environment interactions that may render impoverished people “trapped” and increasingly vulnerable to environmental threats and yet less able to move away from them (Foresight 2011a; Ayeb-Karlsson et al. 2018). However, non-migration decisions not only are the outcomes of being trapped but also can be influenced by social factors such as place-based attachment (Mallick and Schanze 2020), demographic factors such as the size of the affected population (Foresight 2011a), political decisions such as government barriers to migration (Adams 2016), and economic factors such as income at the time when an environmentally damaging event takes place (Black et al. 2011), or discounted benefits of migration (Mellander et al. 2011) . As such, non-migration cannot be simply equated to immobility. To emphasize this, Mallick and Schanze 2020) suggest dimensions of particular relevance to environmental non-migration research: (1) voluntary non-migration: when people do not migrate due to high levels of aspirations and capabilities to stay in their places of origin; (2) involuntary non-migration: when people exhibit low aspirations to stay — they are trapped and cannot migrate mainly due to their incapability to move. Considering such perspectives in environmental non-migration research can provide novel insights into what drives non-migration decisions across space and time and why not everyone moves under conditions of high environmental risks.

Interest in environmental non-migration research has stimulated an evolving literature on the topic (Schewel 2019; Bhusal et al. 2021). Expanding empirical studies contends that environmental non-migration decisions are driven by a variety of factors. These include perceived risk (Buchenrieder et al. 2021; Wiegel et al. 2021), the scale of impact on livelihoods (Faye 2021), and household coping capacity (McLeman 2018). Also important are place-based attachmentsFootnote 3 (Adams 2016; Mallick et al. 2020), family obligation (Jónsson 2010), and personal preferences (Carling 2002). Local interpretations of environmental hazards (Wiegel et al. 2021) also influence non-migration decisions as do elements of the socio-political, economic, and institutional environments (Bhusal et al. 2021). Yet, emerging African scholarship on environmental non-migration (e.g., Sly and Wrigley 1986; Henry et al. 2003; Jónsson 2010; Mpandeli et al. 2020) is still to gain significant science and policy relevance (Borderon et al. 2019). To aid in moving this body of work forward, we review literature to identify and summarize the state-of-the-art on the drivers of environmental non-migration. We do so as a contribution to framing forward-looking migration and non-migration research and policy agendas for Africa.

The next section presents and succintly discusses neoclassical migration theory as we lean heavily on this work within our interpretation. We then present the Foresight conceptual framework adopted in the paper. Methodological issues are elaborated next before presenting and discussing the results. Our concluding section explores how to advance future non-migration research in Africa, highly vulnerable to environmental change effects. The relevance of a holistic approach to researching on environmental non-migration drivers is highlighted; and the usefulness of the Foresight framework for analyzing non-migration decisions even beyond the African continent is contemplated.

Neoclassical migration theory revisited

Neoclassical thinking has occupied a central position in migration theorizing due to its logical plausibility that humans are rational and self-interested. This is quite an attractive proposition in explaining migration decisions. According to the neoclassical perspective, in order to maximize utility or expected benefits, individuals are likely to relocate especially if discounted migration benefits outweigh costs (Tiebout 1956; Sell and Dejong 1978). As examples, regions with higher real wage differentials tend to have positive net migration, while lower-waged ones are often characterized by out-migration (Sly and Wrigley 1986; Mellander et al. 2011). The dichotomy between individual and bounded rationality — the achievement of goals as constrained by existing individual and collective resources respectively (Dandy et al. 2019) — stimulated an expansion within the perspective. On the one hand, micro-economic theorists have often opined that migration is an individual choice, with decisions motivated by the possibility of maximizing personal gains, however defined (Tiebout 1956; Massey et al. 1993). On the other hand, new household economists — while accepting the rationality paradigm, argue that migration (as well as non-migration) decisions go beyond individual motivations to include the intention to create (or conserve) benefits for broader economic units, such as the family or household (Stark and Bloom 1985; Bakewell and Jonsson 2013).Footnote 4 Whatever the unit of analysis, rationality provides a strong theoretical ground for migration decisions. This supports Ravenstein’s initial (1889) opinion that migration decisions are influenced by economic motives. Yet we argue below that, in spite of its attractiveness, neoclassical theories have failed to provide explanations for all decisions to migrate or not. For instance, despite imminent economic risks and suffering, many people voluntarily stay in areas affected by environmental hazards.

Why will people deliberately stay in areas affected by, or prone to, environmental hazards? Some scholars see such populations as “trapped,” given their vulnerability and limited capacity to move away from the threat (Ayeb-Karlsson et al. 2018; Nawrotzki and DeWaard 2018). Skeptics go beyond such generalizations, to opine that the plethora of non-migration drivers transcend economic gain and inherent capacities, to include subjective, socio-cognitive, and endogenous factors such as risk perception, place-based attachment, information assymetry, and informal, community-based safety nets (Whisler et al. 2008; Black et al. 2011; Mellander et al. 2011; Buchenrieder et al. 2021; Wiegel et al. 2021). Also, being left behind can be a deliberate decision to facilitate others’ migration, while ensuring that stayers engage in certain activities (e.g., caring for other family members or carrying out investments) on behalf of the migrated (Jónsson 2010). Such conceptualization legitimizes research that holistically considers the multiplicity of factors that may influence non-migration decisions (Ayeb-Karlsson et al. 2018). We draw on the Foresight (2011a) multi-dimensional framework to systematically analyze the multiplicity of factors that shape environmental non-migration decisions on the African continent.

Conceptual framework: Leaning on Foresight (2011a)

The decision of individuals to migrate or not is the outcome of a complex interaction of a series of economic and non-economic factors (Sell 1983; Boswell 2007; Ayeb-Karlsson et al. 2018). An effective conceptual framework for analyzing environmental non-migrants’ decisions should be able to provide explanations for such decisions (Boswell 2007). According to the Foresight (2011a) report, an effective policy-oriented framework should meet at least the following two criteria:

  1. i

    Identifying the underlying causes of migration

    The resilience and usefulness of migration frameworks lies in their ability to comprehensively capture the drivers of migration and non-migration, not just the symptoms. This is valid also for the decision to migrate or not, due to environmental disasters.

  2. ii

    Broad perspective

    In practice, migrating or not is an outcome of the interaction of economic, political, social, demographic, and environmental factors. As such, concepts such as place, trapped, movers, stayers, and even environmental disasters are just “building blocks” on which migration decisions are made. Therefore, decisions to migrate or not are based on the cumulative and intersecting influence of multifaceted factors (Beauchemin et al. 2020), even if it remains a logical assumption that rising environmental degradation increases the pressure to migrate (Meze-Hausken 2000; Vag 2009). The conceptual strength of the Foresight (2011a) lies in the premise that migration decisions (to move or not) are outcomes of interaction between multiple drivers, whose understanding may improve knowledge on how decisions to migrate or not are made. In addition, the Foresight (2011a) framework has been lauded for its capacity to model population movements influenced by global environmental changes, and to inform policy decisions needed to build resilience towards environmental hazards (Ayeb-Karlsson et al. 2018).

We adopt this conceptual framework (Fig. 1) which elaborates a variety of migration drivers needed to better understand (non)-migration decisions in Africa, especially given the continent’s high vulnerability to climate change and other environmental hazards (Balgah and Kimengsi 2019) . As emphasized by Black et al. (2011), the migration intentions of those affected by environmental hazards are mediated by political factors (e.g., government policy incentives, state intervention), demographic factors (e.g., population size and disease prevalence), economic factors (e.g., employment opportunities and income), social factors (e.g., family/social obligations and informal safety nets), and environmental factors (e.g., hazard exposure, hazard experience, and ecosystem services). Based on the robustness of this conceptualization, we use these factor categories to categorize and rate the importance of drivers of environmental non-migration decisions across the selected literature in Africa. The sub-categories are explained in more detail below.

Fig. 1
figure 1

Source: adapted from Foresight (2011a) and Black et al (2011). Note: HH, household

Non-migration decision making in the context of environmental change.

Full size image

Methodology

Identification and selection of review papers

We review the growing literature on environmental non-migration in Africa, with a thrust on non-migration drivers. Reviews like this can provide summative evidence on non-migration studies in the continent, with conclusions that can inform furture research and policy agendas (Borderon et al. 2019).Footnote 5 In the first step, keywords (e.g., Environmental Change, OR Non-migration, Stayers, OR Voluntary Non-migration, OR Resilient Communities, Trapped Populations, OR Immobile, OR Non-migration Determinants, OR Non-migration Drivers, OR Migration Issues, AND environmental non-migration, AND Sub-Saharan Africa (SSA)/Africa) were used to identify and select key publications from Science Direct, Web of Knowledge, and Google Scholar, and the Climig-database.Footnote 6 The search process ensured that only articles published in peer-reviewed scientific journals were identified. Each term was used individually, and in combination with SSA or Africa. Over 500 hits were obtained. A process to identify only migration and non-migration journal articles written on Africa, African regions, and/or countries reduced the number of items to 222. This sample constituted only articles that explicitly mentioned SSA, Africa, or an African country in the title, abstract, introduction, or methodology section. We then checked for duplicates, which reduced the articles to 106. At this stage, 35 articles that explicitly mentioned non-migration either in the title or abstract or introduction were immediately retained, while 17 articles focusing purely on migration studies were removed. The remaining 54 articles were read by both authors, and 50 articles which explicitly mentioned non-migration at least once were additionally retained. In a final step, we scanned through the 85 retained papers to identify only empirically oriented non-migration research (that is, articles which contained quantitative and qualitative empirical results of primary or secondary data on environmental non-migration), to increase the chances of identifying non-migration drivers. This left us with the 77 articles (43 country-specific case studies and 26 multiple country cases that used primary data, and 8 regional articles based on secondary data) which were used in the review. Overall, 35 (45.45%) of the retained articles applied quantitative methods, 31 (40.26%) qualitative methods, and 11 (14.29%) mixed methods (specific details are provided in Annex I). The literature search was completed in February 2020.

Data coding and analysis

We employed the deductive approach (Chandra and Shang 2019) to code the data; the selection of themes was informed by the Foresight (2011a) framework (Table 1) and previous migration-based reviews in Africa (see Epule et al. 2015; Borderon et al. 2019). Content analyses were performed for each driver category in the famework. All retained articles were reviewed twice by each author to reduce coding and misinterpretation errors (Janson and Olsson 2001). Predesigned sheets were used in the review process and articles were numbered and studied independently by both authors in alphabetical order (Annex 1). The data initially recorded in Microsoft Word was exported to Microsoft Excel 10 where descriptive statistical analysis was performed. Key results are presented and discussed subsequently in this article.

Table 1 Key drivers of non-migration decisions and their subcategories
Full size table

Inter-rater reliability (IRR) analysis was performed using the predesigned variable categories and sub-categories, again based on the Foresight (2011a) framework. This approach was used to identify the strongest determinants for environmental non-migration decisions across the retained sources. IRR demonstrates the degree of agreement or correlation of specific variables among two or more raters (Gwet 2014) and represents the degree of homogeneity or consensus with respect to each driver across all the retained case studies (Quarfoot and Levine 2016). The choice of IRR for the review is threefold. First, IRR scores from at least two raters are more reliable than those of a single rater (Johnson et al. 2001). Second, its application in the climate change/non-migration literature is limited in application, although commonly applied in the medical and social sciences.Footnote 7 Third, it can be easily estimated as the frequency/percentage of the categorical (nominal and ordinal) drivers in the selected case studies that are consistently agreed upon (or not) with respect to their influence on non-migration decisions.Footnote 8

Using three or more raters can lead to more robust results compared to only two raters, especially when measured variables are interval and ratio-scaled (Gwet 2014; Quarfoot and Levine 2016). However, there is no advantage when nominal and ordinal data with no missing scores are evaluated by only two raters (Janson and Olsson 2001). In addition, emerging discrepancies are better managed by the two raters as compared to multiple raters (Johnson et al. 2001).We opted for the two-rater approach given that the study made use of only nominal and ordinal variables. The decision was further motivated by the facts that (1) the scoring sheets were developed by both raters (thereby minimizing interpretation discrepancies) and (2) their different scholarly backgrounds (Development Studies and Geography respectively) already offered two relevant perspectives for the review.

In practice, each major driver category (e.g., political, economic) had a binary connotation with the value 1 as long as any of the subcategories (clear disaster management policy, disaster relief, etc.) was identified as a key driver in the empirical case study, and 0 otherwise.Footnote 9 A driver category was considered important as long as at least one of the subcategories was reported in the rating process as significant in influencing non-migration intentions and/or decisions in any retained article. A template designed in Microsoft Word was used to facilitate the process of data acquisition and eventual analysis.

Drivers were scored (1 if present, 0 if absent) across all 77 reviewed papers by each of the two raters, at the end of which the mean scores per rater (computed by dividing the total scores for each variable category by 77 — the total number of articles) were combined and averaged again. Following Hallgren (2012), Cohen’s kappa was computed for each driver sub-category (e.g., placed-based attachement and social obligations) and averaged to generate to single IRR index for each driver category (in this example, social drivers). The choice of applying the kappa statistic was based on its proven relaiblity when (1) dealing with categorical data and (2) only two raters are involved, as was the case in this study (Johnson et al. 2001; Hallgren 2012). Other statistical approaches (e.g., product moment correlation, and interclass correlation coefficients) take precedence over the kappa statistic only when 3 or more raters are involved (Norman and Streiner 1966) . The newly computed indices were then used for the IRR analysis. For convenience, each mean index was rounded up to two decimal places, as presented in Table 2. The higher the index, the more frequently it was found as a driver of non-migration decisions in the sample (Quarfoot and Levine 2016) . According to Soberg et al. (2008), whenever the IRR approach is used, mean scores less than 0.2. are considered poor: 0.21–0.40 fair, 0.41–0.60 moderate, 0.61–0.80 good, and 0.81–1.00 excellent, in terms of explanatory power across the sample. The data recorded was analyzed using Microsoft Excel, version 10. A logical summary of the key research steps and processes is presented in Fig. 2. Results of descriptive statistics and the IRR analyses are presented and discussed in the next section.

Table 2 IRR analysis of retained literature of non-migration drivers
Full size table
Fig. 2
figure 2

Flow diagram with key research steps. Note: SSA = Sub-Saharan Africa. Note: I, identification of papers; II, article selection procedures; III, adopted analytical approaches; IV, variable categories; V, data coding and analysis steps; and VI, outputs

Full size image

Results

Descriptive statistics

African non-migration literature according to year of publication

Examining the temporal pattern of non-migration coverage offers an indication of the salience of the topic to the research community. To this end, Fig. 3 presents the African non-migration literature across time. Non-migration literature in Africa has been evolving over the last 4 decades, first peaking between 2000 and 2002 and then 2010 to 2012, with the the highest peak between 2013 and 2017 (Fig. 3). The first and second peaks coincide with major recorded environmental hazards in the continent including the Mozambique floods of 2000 and drought-related famine in Somalia between 2010 and 2012, respectively. The third peak coincides with the 2015 World Conference on disaster risk reduction (DRR) in Sendai Japan, whose outcome, the Sendai Framework for DRR (2015–2030), emphasized a research agenda targeting inclusiveness, preparedness, and (environmental) disaster response strategies (Balgah and Kimengsi 2022). Overall, 60% of the identified publications were realized between 2011 and 2020, indicating a significant upward trend in coverage of non-migration in relation to environmental pressure.

Fig. 3
figure 3

African migration literature between 1975 and 2020

Full size image

Distribution of studies by country or region

An analysis of the distribution of empirical case studies by country or region of study from the retained publications indicate that the highest concentration (26 papers: above one-third) of the case studies analyzed different environmental hazards across many countries (hereinafter referred to as “multiple” case studies). As far as individual (country) case studies are concerned, Ghana and Ethiopia both occupy the first position with 8 cases each, followed by Tanzania and Burkina Faso with 6 and 4 cases respectively. Of all these countries, only Ethiopia appears among the top three countries in Africa with sufficiently reported natural hazards. None of the other top countries is considered to be especially disaster-prone (that is, Kenya and South Africa).Footnote 10 This incongruence further justifies the need for a broader non-migration empirical research agenda in Africa that increasingly focuses on countries hardest hit by climate-induced disasters.

Description of major types of stressors in the African-based literature

On the stressors linked to environmental non-migration, the analysis reveals the dominance of the diverse stressor discourse in the literature focusing on Africa with almost 73% of retained articles reporting different stressors, such as droughts and economic deprivation in Burkina Faso (Hampshire 2002) and Ethiopia (Gray and Mueller 2012), floods and gender inequality in Ghana (Afriyie et al. 2018), land degradation, political failure, and cultural diversity in Benin (Doevenspeck 2011), rainfall variability and household demography in Tanzania (Afifi et al. 2014), floods, place-based attachment, and risk perceptions in Cameroon (Buchenrieder et al. 2021), and climate-related forest degradation and socio-demographic factors in Ghana (Abu et al. 2014). The recurrence of multiple stressors close to 75% of the articles validates the multifaceted nature and complexity of decisions to migrate or not in Africa even under conditions characterized by environmental disasters.

Relative importance of non-migration

The review sought to determine if “non-migration” was explicitly mentioned in the title of the paper, the abstract, or the introductory sections, in an attempt to estimate the importance of the non-migration discourse in the African-focused literature. We assume here that mentioning non-migration already in the title or abstract is an indicator that an article was prepared with a strong focus on non-migration. This was found in 46% of the reviewed papers, suggesting a weak representation of non-migration literature. It therefore supports the fact that non-migration research is still in the early stages in Africa, with a cautious but rather optimistic movement towards including non-migration within a much older migration scholarship. The bulk of the literature mentioning “non-migration” (over 70%) was from the period 2011–2020.

Inter-rater reliability analysis of non-migration drivers

The IRR results reveal that social factors (particularly place-based attachment social/family obligations and cultural values) were the most important drivers of environmental non-migration in Africa (IRR score = 0.67). This was particularly evident in Ethiopia and Cameroon. Buchenrieder et al. (2021), for instance, report that the decision of floods victims in northwest Cameroon to stay in their communities was strongly explained by their unwillingness to be separated from their dead relatives. This was in spite of psychosocial trauma emanating from the floods. In West Africa (Ghana), the 1970–1980s drought which depicted environmental deterioration did not significantly trigger migration (Foresight 2011b) due to satisfaction with local politics, and optimism towards economic development (Carr 2005). Non-migration decisions in Africa are therefore increasingly shaped by social factors such as place-based attachment, which are themselves influenced by existing and anticipated political and economic environments.

Environmental factors, such as disaster frequency and coping capacity, were the second major drivers of non-migration decision across the case studies (IRR score = 0.56), as exemplified by the study of Findley (1994) on multiple droughts in Mali. This is closely followed by demographic characteristics which appear as the third most important driver of environmental non-migration decisions (0.48). These include age, household size as reported by Ezra (2001), and Abu et al. (2014) for northern Afifi et al. (2014) for Tanzania; household heads’ intentions and perceptions reported by Buchenrieder et al. (2021); employment of household head (Ezra 2001); and gender and literacy rates as observed by Boyle (1994) in Burkina Faso. Additionally, economic motives such as income at the time of disaster, financial impact of the environmental event, crop and livestock value, and market and credit opportunities were rated as 0.27 (Table 2). Economic interest is reported by Dreie and Sow (2015) as a key driver to migrate or not from northern Benin to Nigeria. The least common driver was the political environment (0.26) which was reflected by the (non)availability of state support, political will/stability, ethnic tensions/civil conflicts, clear hazard management policies, disaster relief, and democracy. Dovenspeck (2011) reports that the political environment was an important determinant for migration decisions in rural Benin.

Discussion

Environmental non-migration literature trends in Africa

The literature on environmental non-migration in Africa is growing, with the greatest bulk registered in the decade 2011 to 2020. This is not surprising, considering that the effects of environmental change in the continent were highest during this decade, which coincided with the elaboration of the current global framework for disaster risk reduction (Balgah and Kimengsi 2022). Rising publication interest on this subject could also be explained by lower than expected outmigration, which was logically expected to accompany the growing climatic and environmental disasters across the continent (Borderon et al. 2019). The rising number of studies is probably also a cumulative outcome of growing numbers of climate and non-migration datasets, and increasing analytical capacity across Africa (e.g., Borderon et al. 2019; Idemudia and Boehnke 2020).

For instance, in North-Western Burkina Faso, Vinke et al. (2021) established that not all people migrate in the face of climate-induced disasters; most women formed part of the non-migrant population under such climatic caprices.

The decision to not migrate is triggered by the perception of community members that migration has negative consequences for their socio-economic situation and their health (Epule et al. 2015; Vinke et al. 2021). This evidence further shows that non-migration may become a feasible and increasingly contemplated option in many parts of Africa, especially if the trend of environmental disasters increase (Vinke et al. 2021). Additionally, climate-induced non-migration is shaped by the climate literacy levels of community members (Helbling et al. 2021).

Regional variations in the studies on environmental non-migration

On the spatial distribution of the literature, it is observed that more than one-third of the literature emanates from multiple case studies. This result suggests the increasing recognition of multi-scale and multi-country dimensions in the appreciation of environmental disasters. Put succinctly, scholars are increasingly recognizing the fact that environmental disasters need to be better understood at multiple scales — this will further inform non-migration decisions. Our results corroborate the findings of a recent study carried out by Helbling et al. (2021) using a dataset of 37,000 individuals across 30 African countries, which concluded with the need for multi-scale studies on environmental non-migration. Such studies can improve environmental monitoring systems and broaden understanding of the complex effects of environmental change on non-migration decisions across Africa (Afolayan and Adelekan 1999). In addition, the results that environmental non-migration studies have been prioritized in East and West African countries. For instance, Ghana, Ethiopia, Tanzania, and Burkina Faso account for 26 of the 77 cases. This corroborates the findings of Ihinegbu (2021) who reported that close to 60% of the growing literature on climate and environmental disasters emanate from these sub-regions, 30% in East Africa, and 27% in West Africa. Surprisingly, the countries classified as disaster-prone (Kenya and South Africa) did not feature on the top list of published literature. While this evidence shows a stark contrast with the results of Ihinegbu (2021) who reported that close to 40% of the literature on climate and environmental disasters emanate from South Africa, it further suggests that despite rising environmental disasters, the Southern African region is found to be wanting in terms of the number of studies on environmental non-migration. Clear reasons for this discrepancy requires further research.

Description of major types of stressors in the African-based literature

The growing literature on environmental migration suggests that multiple environmental stressors tend to shape people or household’s decision to migrate or not (Khan et al. 2018; Mallick et al. 2020; Buchenrieder et al. 2021). These multiple interlinked forces include climate change effects such as droughts, floods, and land degradation with social, economic, and demographic variables. However, single case studies focusing on hydro-meteorological factors such as droughts, floods, and rainfall unreliability were also identified. These results reiterate the context-specificity of weather-related stressors, which in addition to their frequency and impacts can affect non-migration decisions differently across space and time. In fact, droughts and floods remain the most frequent environmental disasters across Africa. These results therefore reflect contextual reality. That almost three-quarters of all the empirical studies mentioned multiple stressors is an indication of the complexity that abounds climate change and non-migration decisions in Africa (Boko et al. 2007). In this light, Perch-Nielsen et al. (2008) conclude that while there is a connection between environmental disasters and the decision to (not) migrate, this connection is by no means deterministic; it rather depends on other vulnerability-linked factors of the population in question, as well as the characteristics of the region in question. The findings suggest the need for non-migration research to consider holistic analytical approaches including both climate change–related and other forces in play that might influence the decision to migrate or not, even in areas suffering from negative environmental change effects, as recently attoned by Khalil and Jacobs (2021).

Relative importance of non-migration

The analysis showed that less than 50% of the literature made explicit use of the term non-migration either in the title, abstract, or introduction. This further corroborates the fact that non-migration in the environmental literature is quite nascent in the African setting. Going by recent studies, it could be deduced that this scenario does not only apply to Africa. For instance, in the Himalayan Region, Bhusal et al. (2021) contend that questions on environmental non-migration aspirations in this disaster-prone region account only for a negligible number of studies. This view is further shared by Ahsan et al. (2022) who report on the dearth of literature on environmental non-migration, even in a highly disaster-prone setting such as Bangladesh. Thus, while Africa is not completely left out of the discourse on environmental non-migration, research acceptance and implementation are largely contemporary. Comparatively, Africa could be doing well in terms of embracing non-migration research. Additionally, the bulk of the literature (two-thirds) which mentions non-migration came from multiple cases. On the whole, the growing non-migration research in Africa is expected, given that (1) it is highly vulnerable to environmental hazards (Epule et al. 2015; Borderon 2019; Mpandeli et al. 2020), and (2) the decision to stay may increasingly become important in reducing the surging rate of out-migration being observed across the continent (De Haas 2008; Bauchemin et al. 2020).

Inter-rater reliability analysis of non-migration drivers

Based on the IRR results, we conclude that place-based attachment, social/family obligations, and cultural values, as social drivers, significantly drive environmental non-migration in Africa (IRR score = 0.67). This showed strong evidence in East Africa, as reported for example by Tegegne (2016) for Ethiopia and Central Africa as revealed by Buchenrieder et al. (2021) in Cameroon. This result aligns with previous contentions that environmental non-migration is strongly influenced by place-based attachment (Adams 2016; Mallick and Schanze 2020; Khalil and Jacobs 2021).

Environmental forces represent the second major drivers of the decision not to migrate across the African case studies (IRR score = 0.48). This can be explained by the fact that most people affected by environmental hazards live in vulnerable and high-risk areas, where extreme events are frequent and repetitive (Askman et al. 2018). Constant exposure induces coping capacity, which renders voluntary relocation less attractive, especially when collective action and community solidarity are central components of hazard management (Balgah et al. 2015; Bhusal et al. 2021), or when innovative, adaptive, and coping strategies (e.g., constructing high foundations in flood areas) are developed by affected populations (Buchenrieder et al. 2021). Thus, while those highly affected by environmental hazards live in areas characterized by resource scarcity and rampant poverty (Boko et al. 2007; Carter and Barrett 2006; Borderon 2019), their perceptions and endogenous coping capacities developed through multiple hazard experiences can positively influence their decision to stay. This partly explains why demographic characteristics only surface as the third most important driver of non-migration decisions (0.48).

Economic motivation was reported to have minimal effects on non-migration decisions in the reviewed papers (IRR = 0.27), questioning the relevance of the neoclassical migration theory in environmental non-migration research beyond conserving previously accrued economic benefits (Bakewell and Jonsson 2013), which is very unlikely when extreme environmental events occur. Environmental victims are likely to stay put, only if their discounted economic benefits of staying after extreme events outweigh those from migrating. This result, while supporting the neoclassical theories of migration (Tiebout 1956; Bakewell and Jonsson 2013), reflects the low tendency for the poor to migrate, especially under conditions of uncertainty. For such vulnerable households, it might be a rational decision to stay put in their habitual residential communities for the sake of sustaining their fragile livelihoods, based on experiential knowledge and community survival strategies (such as solidarity and reciprocity), than to engage in migration processes with uncertain outcomes (Buchenrieder et al. 2017).

The least relevant driver for non-migration from the retained studies was the political environment (IRR = 0.26). On the political environment, it can be understood that there is a general lack of confidence in political (hazard management) institutions, due to the perpetuation of bad governance practices in many African countries, such as corruption, elite capture, and the politicization of disaster relief (Makara 2018; Balgah and Kimengsi 2022). In such political contexts, affected populations are likely to sideline state support from environmental hazard management package, albeit ready to grasp any sporadic support from state-led disaster management institutions (Buchenrieder et al. 2017; 2021). Given the intimate relationship between political stability and sustained socio-economic growth (Amin 2000), it is only logical that the absence of one affects the other on non-migration decisions. Scientific evidence suggests that governments with strong political institutions, active democracies, and vibrant economies are more benevolent in supporting climate coping, adaptation, and resilience activities, and therefore the decision not to migrate (Rogerson 2016; Grasham et al. 2019; Graham 2020). Jacobs and Munis (2020) report in a recent study based on conjoint experiments that federal politics strongly influenced attachment to place in some federal states in the USA. The results of this study suggest similar trends, as non-migration decisions in moderate democracies like Ghana were not affected by the political environment (Abu et al. 2014), compared for example with Benin (Doevenspeck 2011). Improving the political (and consequently the economic) environment is likely to positively influence non-migration decisions in Africa (Scheffran et al. 2019), at least through functioning state-led climate change/disaster management institutions, and climate insurance markets.

Conclusion

Contemporary research interest on environmental non-migration in Africa is rising. To provide a comprehensive knowledge base to advance a future research agenda, this study reviewed the current status of environmental non-migration studies in the continent. This has led to a number of conclusions. First, in spite of growing interest on non-migration research, few studies explicitly address environmental non-migration. This can been understood as part of the transition from a dominant migration research to an emerging non-migration agenda. As a consequence, we urge scholars to increasingly engage these questions since increasing weather-related extreme events continue to perpetuate poverty in Africa, bringing non-migration issues to the frontline. Research is particularly needed in hardest hit countries in order to generate findings of immediate contextual relevance. Regular reviews, such as provided here, are also necessary to inform environmental non-migration policy beyond single countries.

Secondly, social drivers such as place-based attachment, social/family obligations, and cultural values are critical determinants of environmental non-migration decisions in Africa. The social environment therefore lends itself as fertile ground for context-specific, in-depth non-migration research. Environmental drivers were of second importance. Constant exposure to environmental hazards is likely to induce livelihood diversification (Biswas and Mallick 2021), adaptation, and resilience capacities (Bhusal et al. 2021), thereby rendering voluntary relocation less attractive, even for resource-poor households.

Third, economic and political drivers had the least influence on non-migration decisions, defying the contentions held in the neoclassical migration theory. This, we attribute to the generally low economic development and weak policy environment in many African countries, which render state and market institutions for environmentally related disaster management ineffective (Balgah and Kimengsi 2022). However, given the intimate relationship between politics, the economy, and social development (Huggins and Thompson 2015), it seems incumbent for political institutions to be put right, to enhance and sustain economic growth, reduce the impacts of climate-induced events, and promote social development in climate-vulnerable environments.

Finally, this review reiterates the role of place-based attachment in influencing non-migration decisions (Adams 2016; Khalil and Jacobs 2021), moderated by existing (political and economic) coping capacity (van der Geest 2011a, b; McLeman 2018; Bhusal et al. 2021). Similar findings have been reported from empirical case studies out of the continent (see for instance the works of Khan et al. (2018) in Bangladesh; and Bhusal et al. (2021) in Nepal). Therefore, and in spite of the differences in the strength of different drivers in explaining non-migration decisions in Africa, the results signal complex linkages between social, environmental political, demographic, and economic drivers of non-migration. The decision to move or stay due to environmental change is neither simple nor linear, as it is mediated by complex interaction of many factors (Dandy et al. 2019). This complexity needs to be explored in future non-migration research in and beyond Africa. The study results suggest a high relevance of the Foresight (2011a) framework for analyzing complex environmental non-migration decisions. Further application of this framework is encouraged to enhance its analytical capacity across space and time.

We acknowledge that it is impossible to completely analyze the entire African environmental cum climate change non-migration literature with a single review. In addition, the article selection process was limited to articles published and indexed in the selected databases, on which the literature search was done. An analysis of dominant methodological approaches applied was not done. Further research should (i) expand the literature base to include papers which are published in other languages (apart from the English language) on the subject, (ii) comprehensively review applied methodological approaches with an explicit objective of identifying those relevant for the study of non-migration issues focusing on Africa, and (iii) apply more qualitative approaches (e.g., key informant interviews and focused group discussions) with political and economic actors, in a bid to better capture perspectives on non-migration decisions from different stakeholders. A broader focus on climate change can provide insights on the state-of-the art in Africa beyond environmental issues. Meanwhile, climate adaptation and resilience building efforts should continuously be promoted across Africa, given the ever growing capacity of environmental disasters to stimulate forced migration, or to trap populations, across the continent. Disaggregating and analyzing case studies according to methods applied (that is, either qualitative, quantitative, or mixed) can provide novel insights on the role of methods on knowledge generation, precision, and accuracy (Perch-Nielson et al. 2008) , with potential to better guide methodological choices in the growing research on non-migration in Africa. The science and policy implications of informed knowledge generation on environmental disasters and non-migration decisions for sustained growth and stable livelihoods across Africa that is highly exposed to environmental risks therefore cannot be overemphasized.