Popular media frequently depict the Republic of the Marshall Islands (RMI), a low-lying atoll nation, as “sinking” due to sea level rise, with the implication that the nation's inhabitants will have to relocate as climate change refugees.¹ The climate science on sea level rise and other climate change impacts may seem to justify these fears: Sea-level rise is projected to gradually cover much of the country's land area (Pachauri and Meyer, 2014; Owen et al., 2016), recent estimates of the economic toll of sea level rise in island nations reveal greater impacts than originally estimated (Martyr-Koller et al., 2021), and research warns that, without radical adaptation, inundations and contamination of freshwater lenses threaten atoll nations' habitability (Barnett and Adger, 2003; Storlazzi et al., 2018). However, how climate change-related impacts translate to human impacts is less clear, and mass forced migration cannot be assumed to be inevitable (Barnett, 2017; van der Ploeg et al., 2020).

Developments in climate mobilities research have critiqued narratives of climate change-induced mass migration (Boas et al., 2022). Citing millennia of Indigenous adaptation and resilience in the Pacific (Steiner, 2015; Teaiwa, 2018), recent research stresses that habitability is shaped by local knowledge and that the effects of climate change on atoll habitability are contestable (Farbotko and Campbell, 2022). In the face of adverse climate impacts, Indigenous people of the Pacific increasingly prefer to stay on their lands for cultural and spiritual reasons rather than relocate (Farbotko, 2018, 2022). In Tuvalu, grassroots anti-displacement mobilities and re-emplacement activities challenge dominant climate mobility narratives and reaffirm Indigenous rights (Farbotko, 2022), and, in the RMI, government and civil society strongly oppose the donor-supported relocation storyline (Bordner et al., 2020).

Research on environmental migration in the Pacific demonstrates variation in the extent to which climate change influences moves, ranging from infrequent references to environmental factors and criticism of climate change-driven migration narratives in Tuvalu (Mortreux and Barnett, 2009) to internal relocation within atolls as an adaptation strategy in French Polynesia (Duvat et al., 2022) to climate change concerns influencing future out-migration in Kiribati (Allgood and McNamara, 2017). Nevertheless, a recent systematic review refutes assumptions that migration due to climate change is already happening across atoll nations (Mortreux et al., 2023). In the RMI, research identifies education, healthcare, and economic and social factors as dominant migration drivers (van der Geest et al., 2020), with economic hardships that frustrate abilities to meet basic needs and new demands of changing lifestyles of particular concern (Rudiak-Gould, 2013, p. 148).

While climate change does not currently seem to be a major direct driver in the RMI, climate change impacts may influence migration via economic and social factors, such as shocks' effects on subsistence livelihoods and food security (Oakes, 2019). For example, outer island residents, especially, face high and increasing risks from dwindling sea life, declining rainfall, increasing erosion, and higher heat extremes that damage crops, which may influence adaptation and migration abilities and decisions (Rudiak-Gould, 2013, p. 57). The influence of non-environmental drivers reflects the interactive role of economic, political, demographic, and social drivers with environmental changes and highly contextual migration patterns, including existing migration corridors and non-climate anthropogenic drivers of risk, that make it difficult to reduce migration to a single influence (Black et al., 2011; Baldwin et al., 2019; Magnan et al., 2022). Indeed, a growing number of studies find that relationships between climate change and migration are more often indirect, small, and shaped by local contexts (de Sherbinin, 2020; Hoffmann et al., 2020; Boas et al., 2022).

A reductionist emphasis on climate change impacts driving migration raises questions regarding both the accessibility and necessity of migration as an adaptation strategy. While many mobility options were available to and practiced by Indigenous peoples in the Pacific, these options may be increasingly limited by degraded infrastructure, expensive travel costs, and political barriers (Boas et al., 2022). Climate change-related impacts on economic or social stability may reduce migration capacities (Warner et al., 2010; Zickgraf and Perrin, 2016), and damage to infrastructure necessary to support migration may block immediate moves, although infrastructure damage or disinvestment can operate as a migration push if poor infrastructure decreases accessibility and habitability (Cook and Butz, 2015; Hock et al., 2019; Olsen et al., 2021; Blondin, 2022). Socioeconomic and political contexts and colonial legacies also restrict adaptive capacities, increasingly compelling moves in some circumstances and undermining moving abilities in others (McLaughlin and Dietz, 2008; Bettini and Gioli, 2016; Bordner et al., 2020; Boas et al., 2022). For example, research in neighboring Tuvalu and Kiribati found that people were interested in migrating but unable to move, largely because they lacked the financial means (Milan et al., 2016; Oakes et al., 2016).

Recognizing that migration is often less preferred than in-place adaptation, this study investigates agency associated with migration in the RMI—how migration may be compelled or constrained—by examining common migration drivers and the roles of climate change and geographic isolation. Under what circumstances is migration a household or individual choice vs. forced or blocked? If few adaptation options are available, to what extent is any option a choice? Agency refers to the capacity of individuals and groups to act freely in culturally informed ways (Lazrus, 2012). Other definitions use different terminology to reflect similar concepts—the ability to act on one's values (Ibrahim and Alkire, 2007). Structure refers to sociopolitical, cultural, and economic systems that limit and inform choices and opportunities (Rossi, 1993).

Migration scholars have proposed several framings to understand the role and extent of choice in migration (McLeman and Hunter, 2010; Baldwin, 2014). Aspirations and abilities models, for example, aim to explain why some people move while others remain in place by assessing migration-related perceptions and desires and factors that limit abilities to act on aspirations (Carling, 2002; De Haas, 2021; De Sherbinin et al., 2022). Central to these models is reasoning that migration occurs along a voluntary-to-forced continuum, where more voluntary migration requires that origin conditions do not threaten severe conditions like health, starvation, or extreme poverty and that migrants are neither coerced to move nor without a bearable exit option (Hugo, 1996; Bakewell, 2010; Ottonelli and Torresi, 2013; Bartram, 2015). Despite this attention to agency in the broader migration literature, reviews of environmental migration research note limited scrutiny in environmental migrations (Obokata et al., 2014; Hunter et al., 2015). Environmental migration research considers agency in describing political and economic structures and inequalities that limit migration options and abilities (Alscher, 2011; Wrathall et al., 2014), including in the Pacific (Birk and Rasmussen, 2014), but agency is rarely a focus. However, growing attention in environmental migration research to the social, cultural, economic, and political factors that set contexts of environmental adaptation supports evaluation of migration agency related to these factors (Obokata et al., 2014; De Sherbinin et al., 2022). Agency also varies widely within communities (Sheller, 2018). Diversity in movement is shaped by individual and family capabilities and aspirations (De Haas, 2021), including resources necessary to relocate (Boas et al., 2022) and vulnerabilities that can increase risks of involuntary displacement (Adger et al., 2018).

Migration agency in the RMI is important to examine as the islands' geographic isolation and the distance between outer islands and the main island, Majuro, and between Majuro and the primary international migration destination, the United States, may generate exceptional vulnerability and agency-related conditions. The Pacific Islands have resilient, prosperous, and interconnected precolonial histories, with geographic isolation and vulnerability a product of how colonialism and capitalism refashion space (Hau'Ofa, 1995; Gaillard, 2007; Reenberg et al., 2008; Maldonado et al., 2013; Mitchell-Eaton, 2021). Related to the RMI's colonial history, limited economic opportunities, funds to invest in local adaptation, and access to goods and services may frustrate risk diversification and create circumstances where migration is the only feasible option (Kelman, 2007, 2020; Shultz et al., 2016; Doogan et al., 2018; Kim and Bui, 2019). These circumstances may restrict adaptive capacities and support the environmental scarcity thesis, where poor environmental conditions prompt out-migration in search of more secure livelihoods (see also Gray, 2009 for similar logic though different terminology; van der Geest, 2011, pp. 128–129; Hunter et al., 2015). Concurrently, however, high transportation costs may block moves. Difficult and costly moves and long migration distances constrain migration capacities (Henry et al., 2003; Gray, 2010; Massey et al., 2010; Nawrotzki et al., 2013; Birk and Rasmussen, 2014), and worsening conditions that affect livelihoods necessary to support challenging moves may lead to less long-distance migration, particularly for older and lower-income residents (Findley, 1994; Warner and Laczko, 2008; Alscher, 2011; Findlay, 2011; Roland and Curtis, 2020). These circumstances reflect the environmental capital thesis, where declining environmental conditions block out-migration by limiting access to resources needed to support migration (Gray, 2009; van der Geest, 2011, p. 128–129; Hunter et al., 2015). Noting the importance of scrutinizing contexts within which adaptation and migration decisions are made (Hoffmann et al., 2020), this study pays particular attention to how factors related to islands' geographic isolation may affect agency, including the possibility of opposing forces, where dynamics both compel and constrain migration.

Evaluating migration agency in a context that popular media frequently refers to as “ground zero” for climate change-related migration is important as debates around agency-related migration terminologies have wide-reaching policy and human consequences (Erdal and Oeppen, 2018). What conditions are considered sufficiently severe to classify migration as forced and what constitutes a bearable migration option are subject to normative beliefs and often defined by actors in the Global North. Migrants' perspectives on the circumstances of their own migrations seldom factor into classifications. Already, the political stakes surrounding classifications of voluntary vs. forced shape debates about definitions (Ottonelli and Torresi, 2013; Erdal and Oeppen, 2018). Environmental justice advocates may favor the term “climate change refugee” for suggesting a climate debt industrialized nations owe to those migrating, and, following similar reasoning, policymakers in the Global North may avoid the term (White, 2011; Felli, 2013). While Pacific Island leaders reject the refugee label for its disempowering connotations, some researchers argue that avoiding refugee classifications fails to capture the degree to which the environment affects adaptive capacities, particularly in cases where migration is one of few adaptation options (Farbotko and Lazrus, 2012; Lazrus, 2012). Acknowledging patterns of mixed migration, where people move along the same routes for different reasons, and the implications of migration classifications for siloed policy responses, scholars have called for a conceptual reconsideration of voluntary and forced distinctions (Landau and Segatti, 2009).

Recognizing a need to reexamine migration categorizing and classification (e.g., Renaud et al., 2011) and observing calls for research on factors that enable or hamper human mobility (Blondin, 2022) and scrutiny of policies and institutions that might support or obstruct local agency (Mortreux et al., 2023), this study uses interviews with a sample of internal migrants living in Majuro to examine perceptions of migration drivers and migration-related agency. Mobility is core to Marshallese identity (Jarillo and Barnett, 2021), and research on how migrants perceive the drivers that influenced their migration, their ongoing mobility, and their migration experiences, including challenges and barriers, may highlight disparities in adaptive capacities and inform policy to support the accessibility of multiple adaptive responses.

Data collection was conducted in Majuro from January to March and September to November 2019 with support from two organizations, the Marshall Islands Conservation Society (MICS) and the RMI office of the International Organization for Migration (IOM). These organizations were already involved in related projects and research on climate change adaptation, disaster response, and migration. Staff from these organizations and the RMI Office of the Chief Secretary assisted in developing research questions, identifying appropriate methods, and reviewing protocols.

Primary data collection involved 28 in-depth interviews with adult migrants from outer islands living on Majuro. Interviewing internal migrants of varying social and economic backgrounds and from different origin islands was necessary for comparisons between migration contexts, drivers, and experiences. Participants were sampled from a 2018 IOM internal migrant household survey. Interviews were intended to contextualize IOM survey results, which was a need highlighted by IOM staff and in the survey's final report (International Organization for Migration, 2018). The survey contacted 1,215 individuals across Majuro. Interview participants were recruited from survey respondents who had moved from an outer island, indicated a willingness to be recontacted, and provided an email address or phone number (n=220, 82 men and 138 women). Prospective participants were grouped by island of origin and randomly recruited from these groups to ensure a mix of migration origins in the sample. Interview participants varied widely by age (age 19–70), gender (6 men, 22 women), occupation, socioeconomic status, place of residence in Majuro, migration origin, and time of migration (arriving in Majuro between 1.5 and 41 years prior). Participants were mostly women as 63% of survey respondents were women. Participant variation was important to compare migration experiences.

The study's sample has several constraints. The overrepresentation of women may lead to overemphasis of certain drivers and migration experiences. This bias in the sample may be due to greater likelihoods of men migrating long distances and continuing to the U.S. and, thus, more women remaining in Majuro (Ravenstein, 1885; Silvey, 2006). Additionally, no non-migrants and ongoing migrants to the US were interviewed. Inclusion of these populations might offer a more complete picture of migration drivers, experiences, and agency in the RMI. Finally, the sample is more representative of non-circular internal migrants and longer-term internal migrants to Majuro as internal migrants to Majuro often continue to the U.S. within a few months of arriving on Majuro yet the time lag between survey and interviews prevented follow up data collection with these migrants. As the international move is expensive, migrants in the study sample may thus be less financially resourced than migrants who continue to the U.S. shortly after arriving on Majuro (Findley, 1994; Findlay, 2011).

Five interpreters supported participant recruitment and interview conduct. Four interpreters worked with MICS or IOM, which meant that they were familiar with the subject matter. This large number of interpreters was necessary for interview scheduling but also increased the number of key informants informing the research process and allowed for interpreter-participant matching by gender when schedules permitted (Edwards, 1998). Interpreters were trained together to promote consistency and encouraged to identify and ask follow-up questions to minimize translation-related disruptions (Adamson and Donovan, 2002; Williamson et al., 2011). Post-interview, interpreters translated and transcribed recordings.

Data collection also included 12 key informant interviews with elected officials, government staff, and members of civil society, conducted in English without interpreters. Participants included mayors, senators, and leaders of aid organizations and government departments. These interviews provided context on the conditions surrounding migration in the RMI and highlighted wider social implications and possible solutions. Senators and mayors from outlying islands discussed the specific circumstances on the atolls they represented.

Thematic analysis of both sets of interviews was conducted in NVivo. Codes were identified related to participants' migration influences and experiences and descriptions of vulnerability and resilience on outer islands and Majuro and organized thematically. To support accurate analysis, key points from interviews were discussed with interpreters following interviews and initial codes reviewed at the end of data collection (Nowell et al., 2017). Results include frequent use of direct quotations to reflect participant voice.

This study examines migration agency in a sample of internal migrants in the RMI, evaluating perceived migration drivers and roles of geographic isolation and climate change. Interviews with migrants living on Majuro and members of government and civil society reveal how social, economic, cultural, and environmental factors shape migration contexts and agency. Building on studies that consider effects of political and economic structures and inequalities on moving abilities (Alscher, 2011; Wrathall et al., 2014) and weigh the influence of geographic isolation-related factors on adaptive capacities (Kelman, 2007; Nawrotzki et al., 2013; Birk and Rasmussen, 2014; Kim and Bui, 2019; Roland and Curtis, 2020), this study reports on participants' perceived barriers to migration agency. While findings are specific to the RMI, findings contribute to limited agency-focused environmental migration research and may have relevance for other geographically isolated contexts (Obokata et al., 2014; Hunter et al., 2015).

Results suggest that geographic isolation-related factors may increase likelihoods of both more compelled and more constrained migration. Climate change impacts may exacerbate these opposing forces. For example, environment-related impacts on resource-dependent livelihoods may compel migration in search of new economic opportunities (the environmental scarcity thesis). However, despite improved mobility in the last two decades, largely due to rising incomes to support high transportation costs (Jarillo and Barnett, 2021), expensive travel continues to constrain migration for cash-poor residents, particularly return migration and visits. Worsening environmental conditions may exacerbate cost-related migration constraints by reducing the resources available to support migration, for example, if prospective migrants can no longer save for moves by producing copra (the environmental capital thesis). Inaccessible return migration raises the stakes of migration. Less economically successful migrants may want to return yet may be unable to afford travel costs. Cash-poor migrants less able to travel may also struggle to maintain social and cultural ties to origins. Regular return visits may allow more affluent migrants to maintain communal ties on outer islands, supporting an enduring sense of connection and encouraging further trips. However, cash-poor migrants may struggle to sustain similar ties.

Concurrently more compelled and more constrained migration and scenarios where both the environmental scarcity and capital theses apply may exacerbate vulnerability. Changes to the demographic structures of outer islands and Majuro resulting from out-migration may further complicate migration drivers and compelling and constraining dynamics. For example, as many working-age residents leave outer islands, those remaining in-place—largely older residents and children—may struggle to maintain resource-dependent livelihoods made increasingly strenuous from environmental changes. This demographic shift raises questions about the long-term viability of economic migration undertaken to support the in-place adaptation of family members remaining on outer islands (e.g., Farbotko et al., 2022).

This research, highlighting perceived barriers to migration agency and new risks from climate change, has policy implications for enhancing accessibility of adaptation options and reinforces policymakers' calls for funding necessary to implement the diverse strategies outlined in the RMI's National Adaptation Plan to strengthen adaptive capacities. Future research might examine the vulnerability implications of simultaneously increasing out-migration pressures and migration inaccessibility, effects of shifting demographic structures, heterogeneity related to population characteristics like age and gender, and projected impacts as environmental shocks and shifts grow increasingly frequent and extreme. Future research might also include control populations of non-migrants and migrants to the U.S. to expand the comparative analysis of migration contexts, drivers, and outcomes.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by the University of Wisconsin—Madison IRB: ID# 2018-1414. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

HR confirms sole responsibility for the following: study conception and design, data collection, analysis and interpretation of results, and manuscript preparation.