Clarification of key concepts

In the present review, we adopt the United Nations Population Fund’s definition of gender as “the economic, social and cultural attributes and opportunities associated with being male or female.” [17] Thus, we contend that gender-responsive CCA research and practice must move past mere comparisons of women’s and men’s perceptions, experiences and activities. The present review stands in alignment with the work of an increasing number of scholars who emphasise that gender-responsive CCA ought to account for “social relations of production, cultural norms and broader political-economic institutions [that mediate] the nature of exchanges, opportunities and the distribution of resources [and] contribute to the specific constructions and experiences of vulnerability, as well as capacities to respond and cope with climate stresses” [18,p.28].

Further, the present review employs the Intergovernmental Panel on Climate Change’s definition of CCA as “the process of adjustment to actual or expected climate change and its effects in order to moderate harm or exploit beneficial opportunities.” [19,p.43] We distinguish between autonomous and external CCA practices. Following Malik, Qin and Smith, we understand autonomous CCA as CCA practices adopted “by individuals and communities without deliberative government planning or intervention,” [20, p.14] which are nonetheless intertwined with “existing social, political, cultural and market institutions.” [20,p.15] On the other hand, we understand external CCA, also referred to as planned adaptation, as stemming from “a deliberative policy decision, based on an awareness that conditions have changed or are about to change and that action is required to return to, to maintain, or to achieve a desired state.” [20,p.4] We refer to such interventions as “external” rather than “planned” to underscore that autonomous CCA practices can also be planned.

Based on these conceptualisations of gender and CCA, we adopt Blythe et al.’s understanding of transformative CCA as an approach that addresses fundamental social, political and economic structures that together play a role in rendering populations more vulnerable and marginalised due to inequality. [21] Interventions employing GTAs embody this approach with specific focus on how gender ideology and norms embedded in these structures indirectly and directly shape women’s and men’s access to resources and to participation in decision-making fora. [22,23] GTAs are commonly conceptualised as spanning three primary dimensions: agency, relations, and institutional structures. [23] In this context, agency refers to “individual and collective capacities[…], attitudes, critical reflection, assets, actions, and access to services” [23,p.5]; relations entail “the expectations and cooperative or negotiation dynamics embedded within relationships between people in the home, market, community, and groups and organizations” [23,p.5]; and structures include “informal and formal institutional rules that govern collective, individual and institutional practices, such as environment, social norms, recognition and status” [23,p.5]. While GTAs are not explicitly addressed in the findings of this review, we return to the notions outlined here in the discussion.

The discussion also builds on the notion of livelihoods and livelihood transformations as conceptualised by Carr. [24] Carr notes that agrarian livelihoods are “project[s] of managing both social and natural processes to create and maintain particular socio-ecological states that further specific goals of those living in that system, particularly the goals of those whose authority provides them with privileges not enjoyed by others.” [24,p.71] According to Carr, gendered roles and identities tend to become more rigid when livelihoods are under environmental or social stress, and it is only when livelihood projects fail to ensure subsistence that spaces for re-negotiation and innovation tend to open up. [24] Such openings may present opportunities for re-arranging intra-household dynamics in a more equitable manner. However, the ensuing transitions are themselves characterised by power structures and pervasive norms and often pose new risks to different household members, especially those who are most vulnerable [24].

Search strategy

As is common for qualitative meta-syntheses, [28] our goal was to retrieve all studies relevant to our review objectives. The search strategy consisted of systematic database searches and supplementary hand searches that were developed through iterative trial searches. Supplementary hand searches were deemed necessary because it has been shown that database searches often do not suffice to identify all qualitative research on a given topic. [25] Recognizing that systematic searches for qualitative studies tend to include trade-offs between recall (i.e., identifying all relevant studies) and precision (i.e., identifying few non-relevant studies), [29] we chose to prioritise recall and thus adopted several complementary search techniques.

Our systematic database search spanned 8 databases covering a wide range of disciplines related to the gender-CCA nexus: Anthropology Plus, [30] Anthrosource, [31] International Bibliography of the Social Sciences, [32] Scopus, [33] SocINDEX, [34] Sociological Abstracts, [35] Web of Science, [36] and Women’s Studies International.[37] The systematic database search was conducted on 04.05.2022. A detailed search log, including number of hits and search strings used for each database is available in S1 Text.

The supplementary hand searches were conducted between June and August 2022 and consisted of a range of techniques that are commonly included in the umbrella term berrypicking [28,38]: We conducted forward and backward searches of all studies included during the screening process, we searched all publications of the first authors of included studies, and we conducted comprehensive hand searches of selected journals (Nature Climate Change, [39] Climate Policy, [40] Climate and Development, [41] and Gender and Development [42]) and one database (African Journals OnLine [43]) that we had identified as highly relevant to this review during the aforementioned scoping review. African Journals OnLine could not be searched systematically due to the limited advanced search functions available in this database. Further information regarding the search terms used for journal and database hand searches are available in S1 Text.

Lastly, a second systematic database search was conducted in Scopus [33] and Web of Science [36] on 17.06.2023 to enhance recall of the newest relevant studies. Scopus [33] and Web of Science [36] were selected for this search because these databases had rendered the most absolute and relevant hits during the first systematic database search. Further information is available in S1 Text.

Screening of records

An overview of the screening process is given in Fig 1. All hits were first saved in Zotero, [44] where duplicates were removed. All unique hits were then uploaded to Rayyan, [45] where titles and abstracts were screened independently by two researchers using predetermined inclusion criteria (except for hits resulting from the second database search, which were only screened by the first author due to time constraints). Incongruencies were resolved through discussion between the first and second authors. Next, full texts of all studies that had passed the title and abstract screening were screened using the same inclusion criteria. After this, Incongruencies were again resolved through discussion between the first and second authors.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flow chart of article inclusion.

https://doi.org/10.1371/journal.pclm.0000279.g001

The inclusion criteria were as follows: 1. general study characteristics (studies had to be peer-reviewed and published in English), 2. population (studies had to include participants from rural regions of East Africa, as defined by the United Nations Statistics Division [46]), 3. methodology (studies had to report findings from qualitative primary data collection methods), and 4. content (studies had to report findings that explicitly describe how men and women negotiate CCA practices within households).

Quality appraisal of included studies

There seems to be no scientific consensus concerning the approach to designing and conducting quality appraisals when synthesising qualitative research. [27] We decided to appraise the quality of all included studies because the full-text screening had indicated great discrepancies of quality between studies, and we expected that our analysis would benefit from a more systematic overview over the quality of included studies.

We based the quality appraisal on Saini and Shlonsky’s Qualitative Research Quality Checklist. [27] This 25-item checklist covers studies’ theoretical frameworks, settings, research designs, sampling procedures, data collection, ethical issues, researcher reflexivity, data analysis, and reporting of the findings. This tool was chosen because it allows for an assessment of quality across a wide range of qualitative research designs, it is more comprehensive than other comparable tools, and it was developed and piloted in a rigorous and transparent manner. [27] We used the first 22 items of the Checklist and left out the last three items relating to fairness and the promotion of justice, since our main aim was to assess the quality of the findings. Each included article was assessed by one of the authors. As is common practice for qualitative reviews, [27] we chose to consider the quality appraisal results during our analysis rather than excluding any studies from the analysis.

Data extraction and analysis

The analysis followed a two-step process. First, primary findings relevant to the review’s specific objectives were extracted from all included studies through an iterative, thematic coding process. Authors of the included studies were not contacted during this process due to time constraints. Findings of each article were coded by one reviewer. We chose an inductive approach (i.e., rather than using predetermined themes and codes, we relied on themes and codes emerging from the included studies) because this allows for the preservation of the original interpretations of primary studies, which is essential for qualitative meta-syntheses [25].

Specifically, we coded 9 of the included studies openly, i.e., assigning codes that were as close as possible to the original meaning of the respective text passages. Based on these codes, we then developed a coding framework which grouped related codes under themes. Serving as a basis for the extraction of findings from the remaining included articles, this framework was then developed and expanded in an iterative manner. The authors compiled their extractions in a shared Google Sheets [47] table. During the coding process, the authors also collected information pertaining to studies’ settings and methodologies.

Once the coding had been completed, the authors shared their reflections on each theme, code, and interconnections between different themes and codes in an interactive analysis session using the ConceptBoard digital collaboration software. [48] The results from this session and the shared data extraction sheet were then used as a starting point for further thematic analysis.

Study characteristics

The screening process outlined above resulted in 21 included articles. [18,49–68] Of these, 18 resulted from the first database search and 3 were identified through supplementary hand searches. General characteristics of the articles are summarised in Table 1, and detailed information for each article is available in S2 Table.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. General characteristics of the included articles.

https://doi.org/10.1371/journal.pclm.0000279.t001

As evident from Table 1, the final review sample included nearly equal amounts of qualitative and mixed-methods studies. Almost all of these studies employed focus group discussions, followed by key informant interviews and interviews with household members. Most studies appeared to have included both male and female participants, with some notable exceptions that only included women. [57,67,68] The included studies employed a wide range of theoretical frameworks and perspectives that commonly related to feminisms and intersectionality (see S2 Table for further details).

In terms of research settings, the included studies involved rural households pursuing a wide range of livelihood activities. While the most common livelihood profiles were dominated by agriculture, agropastoralism and pastoralism, several studies reported increasing diversification and non-farm activities, such as trade and wage labour. Such diversification was commonly described as a response to climatic stressors. Most settings experienced several stressors simultaneously, the most common one being increased frequency, duration or intensity of droughts. In many settings, climate-related livelihood diversification and other forms of intra-household CCA were clearly influenced by gender norms and relations. Moreover, CCA practises themselves also appeared to hold the potential to challenge and shift intra-household gender dynamics.

Study quality

This section provides a general overview of the results of the quality appraisal. Detailed assessments for each article are available in S1 Table. The quality appraisal suggested that the included articles varied substantially in quality (in the present review, we define quality in terms of the factors included in Saini and Shlonsky’s Qualitative Research Quality Checklist [27]). Several studies appeared to be of high overall quality, characterised by strong internal cohesion between research questions, study design, and reporting of findings; detailed descriptions of qualitative methods; and evident reflexivity of the authors. [50,54,55,63,66,67] At the other end of the spectrum, two mixed-methods studies exhibited significant inconsistencies between their objectives, study design and implementation of qualitative methodologies. [58,59] Specifically, in the article by Tambulasi et al., there appeared to be a disconnect between the mixed-methods study design and the research objectives, which were phrased in a quantitative manner.[58] In the article by Ndlovu and Mjimba, participants for qualitative methodologies were recruited using stratified random sampling, pointing to a lack of internal cohesion [59].

For most studies, the appraisal was difficult because the articles contained incomplete information regarding data collection and analysis, ethical considerations or study limitations. In some cases, this lack of information was severe enough to undermine a thorough assessment of the studies’ overall quality. [52,53,57,60–62,64] This issue was present in mixed-methods and purely qualitative studies alike. Remarkably, even studies that were perceived as of high overall quality tended to lack information regarding ethical considerations, [50,66,67] risk of bias and other limitations [50,54,63].

Intra-household CCA decision-making

In this section, we describe how male and female household members negotiated the ways in which their households adapted to perceived climatic stressors. While most of the evidence under this theme related to autonomous CCA, the section also includes some observations regarding intra-household decision-making about external CCA interventions.

Generally, men tended to be the primary decision-makers about autonomous and external CCA within households. [18,58] As one male participant from Mphampha village, Malawi, put it: “The man is the head of the family; therefore he must control everything at home. He is the leader, the driver. It’s by culture.” [58,p.195] However, some studies identified women as the primary implementers of CCA decisions because they played key roles in sustaining households’ livelihoods, [62,65,66] with one participant from Gwembe district, Zambia, noting: "The men decide […] leaving women to cope with even the unfavorable decisions." [65,p.537] This finding of ours is supported by quantitative evidence from a rural household survey in Tanzania’s Morogoro region, which showed that men held significantly higher decision-making power than their wives [69].

It should be noted that one study found that perceptions about CCA decision-making varied considerably between male and female participants: In focus group discussions about livestock-rearing among Borana pastoralists in Kenya, women claimed that they were rarely involved in decisions concerning animal breeding, while male participants claimed that no decisions were made without women. [63] The authors of said study offered no potential explanation for this apparent discrepancy between men’s and women’s perceptions of decision-making processes. [63] However, this observation appears to underscore the complexity of CCA decision-making processes and the various forms of participation and influence exercised by different household members. At any rate, even in this study, male and female participants agreed that men held the primary responsibility for intra-household CCA decision-making [63].

Regarding decision-making about external CCA interventions, the pattern of male dominance in CCA decision-making was compounded in some settings by the fact that women’s work was centred around the private sphere of the home [58,61]. In these settings, men were primarily responsible for engaging with actors in the public sphere, including governmental actors and NGOs. For instance, Rao et al. observed that in Kenya, “men dominated the state-provided aid and relief facilities during floods or droughts; [with] women relying on their male relatives to fulfil their needs” [49,p.967].

As indicated by this observation, in many settings, the division of labour within households and communities was strongly gendered, i.e., women and men were responsible for different tasks and domains. [50,55,63] In some cases, women appeared to be the primary CCA decision-makers within the domains that fell under their responsibility, such as planting seeds, preserving produce and preparing food. For instance, Anbacha and Kjosavik noted that, in the Ethiopian Borana households that they studied, women had limited decision-making power in the context of livestock rearing but were able to adapt to climatic stressors to some extent by choosing which crops to plant. [63] This notion of gendered domains for decision-making aligns with quantitative evidence from rural households in Tanzania’s Morogoro region, which suggests that decisions about cover crops and vegetable cultivation fell under the female domain while men were often responsible for decisions about cash-based CCA strategies [69].

However, effective household-level CCA would often have required comprehensive, integrated solutions because the impacts of climatic stressors were complex and extensive. Where women’s decision-making power was limited to traditionally female domains, this could lead to less adaptive or even maladaptive responses. For instance, one female participant in Kakamega County, Kenya, explained that “since it is the role of women to feed the family, most women reduce their food consumption during food scarcity and some skip meals to spare food for the children" [55,p.6].

Besides these cases where women made CCA decisions within traditionally female domains, one study referenced a female participant who stated that, in her marriage, decision-making power was primarily dependent on personal knowledge and skills: "After he retired, my husband came here and [now he] helps in the shop. He is not good at networking, nor does he have business ideas, but I can trust him to look after the shop when I am selling miraa. I know how to invest cash and get profit, so have the final say financially." [18,p.31] This dynamic appeared to be an exception, though.

Some external CCA initiatives attempted to enhance women’s adaptive capacity by providing them with physical resources required for adaptation, e.g., livestock and seeds. However, in most studies, it appeared that the provision of resources did not suffice to enhance women’s CCA agency because these interventions did not challenge the intra-household gender dynamics vesting decision-making power with men. For instance, Tambulasi et al. noted that, when external CCA stakeholders provided women with chicken in Malawi’s Chikwawa District, it was male household members who decided how these chickens should be used to meet household needs [58].

While most CCA decisions appeared to be taken either by men for the entire household or by men and women for their respective domains, three articles also reported some degree of joint decision-making. [59,63,65] For instance, Khoza et al. stated that, among Zambian small-scale farmers choosing to adopt Climate-Smart Agriculture, some male household members decided that they wished to involve their spouses in the decision-making process. [65] Two other articles reported that female participants were increasingly demanding to be more involved in CCA decision-making processes on the basis of their growing contributions to household economies. [18,63] Quantitative evidence from rural households in Tanzania’s Morogoro region suggests that, in some settings, earning money outside of the household may, indeed, be associated with greater CCA decision-making power among women [69].

Moreover, Ndlovu and Mjimba found that, among agropastoralists in Zimbabwe’s Umzingwane District, an increase in joint decision-making processes was necessitated by increased frequency and severity of droughts. However, the authors “found that such consultations between spouses often drag [on] and the associated time lag has a bearing on the effectiveness of the proposed interventions to reduce drought related risks”, [59,p.6] indicating that an increase in joint decision-making might bring about novel challenges.

Lastly, one study suggested that the dynamics of intra-household CCA decision-making were significantly dependent on the gender of household heads. While studying how households decided about the adoption of Climate Smart Agriculture in Malawi and Zambia, Khoza et al. found that "[w]omen could only make decision (sic) in cases of de jure female household heads […] with the outright absence of an adult man to lead decision making. Where an adult male relative was present within household (sic) (such as brother, son or grandson), the woman consulted him and would [be] likely to adopt his opinion on [CSA] adoption." [65,p.536] The authors found that these differences in decision-making dynamics were rooted in norms of ownership and control over key productive assets, stating that “[w]omen could only own major productive assets if they were de jure female [household heads] and had inherited assets from the late husband.” [65,p.538] This observation implies that control over household assets might mediate how cultural gender norms and dynamics influence intra-household CCA decision-making. The following section explores this influence in greater depth.

Factors influencing gendered CCA decision-making

This section explores underlying factors that influence the roles played by male and female household members in intra-household CCA decision-making processes. As noted above, intra-household division of labour, CCA decision-making and control over resources are strongly influenced by sociocultural gender norms. Moreover, the foregoing section explored how the gender of household heads might shape intra-household CCA negotiations. This section focuses on access, ownership and control over resources—a third set of factors that appear to mediate the influence of gender on CCA decision-making.

Generally, studies included in this review found that men owned most productive household assets and thus controlled a large share of household resources. [58,63] In most studies, women were typically able to access key productive resources through male household members or relatives. [59,61,65] However, women’s access to resources was often less secure than men’s, since women could lose access due to divorce, bereavement and estrangement from male household members and relatives. [18] This observation aligns with quantitative evidence from rural households in Tanzania’s Morogoro region, which showed that being married was crucial for women’s but not for men’s adaptive capacity because married women were more likely to have access to productive land than their unmarried, widowed or divorced counterparts [70].

Two articles included in our review highlighted that, even when women’s access to resources and decision-making power were codified in formal institutions, discriminatory cultural norms could hinder women’s ability to exercise their formal rights. [49,61] For instance, Wangui and Smucker found that, regarding women’s ability to access irrigation in Tanzania, "[o]ne of the main constraints […] is access to land. […] Though in principle the [Tanzania Village Land] Act supports gender equity, it leaves a lot of leeway to village government to define the process by which village land is distributed. The process is a greater obstacle for women, who are expected to gain access primarily through their husbands. Women’s adaptation decisions are therefore constrained" [61,p.373].

Furthermore, while access to resources certainly appeared to influence women’s and men’s adaptive capacity, ownership and control over resources seemed to be much more strongly associated with CCA decision-making power. As Mnimbo et al. observed in Tanzania: “[The] capacity to adapt is shaped by access to and control over resources, as well as power to make decisions. In the study area, males own and have dominant power over household resources. For example, they can decide on trading even households’ assets during drought.” [52,p.103] This observation is supported by quantitative evidence from the same study, which indicated a significant, gendered divergence between access to and control over resources. For instance, though all female participants could access land, 75% of land was under male control [52].

Moreover, it appears as though climatic stressors might further decrease women’s control over resources and thus further constrain their CCA decision-making power. For instance, Rao et al. found that, in Kenya, the "customary practice of allocating some [live-]stock for the use of […] wives and daughters” [18,p.34] was increasingly threatened by persistent droughts. This limited female household members’ ability to engage in CCA through livelihood diversification because women were reliant on this livestock to stem initial investments needed to set up small businesses. In a similar vein, Rao et al. also found that male herders increasingly migrated due to drought, and that “when men moved away with livestock, women lost control over milk for consumption and sale” [49,p.967]. Though the authors did not assess how these dynamics impacted women’s intra-household CCA decision-making power, it appears plausible that decreased control over resources within their traditional domains might decrease women’s adaptive capacity. Similar patterns have also been observed in quantitative research. For instance, Tavenner et al. found that, within rural East African households, women’s control over resources tended to decrease with increasing commercialization of farming, [71] which is a common autonomous CCA strategy in this region.

The observation that climatic stressors and related CCA strategies might contribute to decreased adaptive capacity among women shows how intra-household CCA is not only influenced by gender norms, but can itself also impact the dynamics between male and female household members. The following section describes this reciprocal influence in more detail.

CCA impacts on intra-household gender dynamics

As described in the previous section, gendered divisions of labour appeared to influence intra-household CCA decision-making. Simultaneously, numerous studies suggest that CCA practices were influencing intra-household division of labour, in turn. [50,52,55–57,63] It should be noted, however, that intra-household gender dynamics appeared to be influenced by a range of interacting factors, including but not limited to climatic stressors. Other such factors included changes in cultural norms, [18] economic strain, [56] and increased governmental regulation of rural livelihoods. [63] Acknowledging these complex interconnections, we focus on the influence of climatic stressors as it was identified by the authors of the studies included in the present review.

One frequent observation was that, due to persistent drought, men were increasingly struggling to provide for their households in the manner that traditional gender norms demand. [50,56] As a consequence, women in several settings were entering traditionally male-dominated domains, and vice versa in order to secure households’ livelihoods. [49,50,55,63,65] For instance, Rosen et al. observed this tendency in Zambia: “Participant narratives suggested that drought blurred a historically gendered division of employment roles. A shrinking labour market forced men and women to prioritise potential earnings over workforce preferences, pushing women into jobs with heightened manual labor demands and men into labor sectors traditionally dominated by women” [56,p.6].

Specifically, several studies found that women increasingly engaged in income-generating activities like petty trade and wage labour. [49,50,55–57,63] Often, this trend was rooted not only in the necessity to diversify livelihood portfolios, but also in the fact that men were spending more time away from home due to temporary or permanent migration. [49,56,57] This tendency was particularly pronounced in settings where pastoralism continued to play significant roles for sustaining livelihoods. This shift in gendered division of labour reportedly led to increased workloads for women [e.g. 56] and a growing proportion of de-facto female-headed households. [18,49,52,56] Shifts like the ones that emerged in our analysis have also been quantified. For instance, Anbacha and Kjosavik noted that, in Ethiopian Borana households experiencing increased droughts, gendered divisions of labor were becoming increasingly blurred in many domains, including hut making, herding and crop farming [50].

While none of the articles included in our review addressed whether and how this trend influenced women’s and men’s roles in intra-household CCA decision-making, there is some evidence that climate-induced changes in gendered division of labour might contribute to conflicts between male and female household members. For instance, Anbacha and Kjosavik noted that, in Ethiopia, “participation of women in petty trade is […] creating gender conflicts within their households […]. Women stated that men were not happy when their wives participated in petty trade. Some women were even beaten up and warned by their husbands to stop trading"[63,p.8]. In some settings, intra-household CCA negotiations were also associated with gender conflict due to the resource constraints that climatic stressors placed on households, e.g., Rosen et al. noted that, in Zambia, “[m]arital relationships were challenged in times of drought, particularly when disagreements around household purchases could not be reconciled. Tightened household incomes required more joint decision-making in even small household purchases” [56,p.7].

Besides experiencing an increased workload and negative repercussions from gender conflict, women were sometimes negatively impacted by CCA practices that placed them in precarious situations. For instance, two studies reported that female participants engaged in prostitution as a means of escaping utter destitution, [56,68] and Rosen et al. found that, in Zambia, “the financial insecurity propagated by drought forced girls into early marriages. While participants indicated child marriage was prevalent prior to drought, they explained that heightened poverty from drought perpetuated girls being married off by their parents or guardians because they are no longer able to care for them or are getting a financial return from the dowry" [56,p.7].

The impacts of external CCA interventions on intra-household gender dynamics have not been covered in detail. However, one study found that, if external CCA actors did not consider gendered division of labour when planning and implementing their interventions, these interventions could contribute to the increasing workload for women during climate change: “[T]he NGOs do not synchronize their initiatives, but instead increase the strain on the women and their roles and responsibilities by spreading meetings over weeks, which clash with other community processes and chores pertinent to women. Eventually, the women are left with less time to complete household chores and other productive duties” [68,p.276].

Despite these instances of CCA practices having negative consequences for female household members, there also seems to be some evidence that CCA might increase women’s agency under certain circumstances. For instance, Anbacha and Kjosavik note that “[a]lthough the participation of women in petty trade has obviously increased their workload, this is gradually challenging the existing gender roles and women are negotiating for change in gender relations" [63,p.8]. In this study, while women’s increased involvement in income-generating activities led to gender conflict in the short term, it also enabled women to access and control cash, thus enhancing their dependence on male household members. [63] In a similar vein, Rosen et al. found that, in Zambia, women’s increased involvement in financial decision-making during droughts may have enhanced their influence on intra-household CCA negotiations but was also associated with increased quarrels between husbands and wives. [56] In sum, the foregoing observations seem to indicate that CCA strategies certainly seem to have the potential to influence intra-household gender dynamics, but the pathways appear to be complex and the ultimate outcomes for male and female household members remain unclear.

Limitations

A number of limitations ought to be discussed. These relate to the quality and characteristics of the primary studies included in the analysis, the qualitative approach chosen for this review, and challenges inherent to studying intra-household CCA negotiations and decision-making. Firstly, two of the included mixed-methods studies did not distinguish clearly between findings resulting from quantitative and qualitative methodologies, [58,59] which made it hard to identify which findings to extract for our qualitative synthesis. Moreover, we identified some quality-related concerns in these two studies and other papers lacked clarity in reporting, rendering a quality assessment difficult. [53,57,60–62,64] As is common practice for qualitative reviews, [27] we chose not to exclude these studies from the analysis.

The findings of the present review were further limited by the relative scarcity of data in included primary studies regarding some of the issues we aimed to investigate. This limitation applies especially to intra-household negotiations about external CCA interventions and the impacts of such interventions on gender dynamics. Moreover, though many of the included studies explicitly or implicitly referenced intersectionality, the reporting of the findings in many cases lacked descriptions of multiple markers of difference, which made it difficult to consistently apply an intersectional approach in our analysis. Given that we excluded some otherwise relevant studies because they did not employ qualitative methods, [e.g. 72] it appears reasonable to assume that a mixed-methods review might have been able to draw on a richer base of primary data. We attempted to mitigate this limitation by contextualizing our findings with quantitative evidence from selected articles, which largely aligned with our qualitative results.

A mixed-methods or quantitative approach could also have mitigated some of the limitations inherent to all qualitative research, e.g., the limited generalizability of findings. [27] Since we synthesised qualitative primary data, the findings of the present review cannot necessarily be generalised to all rural households in East Africa. Rather, we hope that our findings may serve as a source of inspiration and a basis for reflection for researchers, policy-makers and practitioners engaging with CCA in East Africa and elsewhere.

Lastly, we encountered two challenges inherent in the field under study: First, as has been noted before, [8] the concept of CCA decision-making remains vague in many studies, and there appears to be no consensus about how decision-making should be assessed in qualitative and quantitative research. One study included in the present review found that perceptions of men’s and women’s involvement in intra-household CCA decisions differed greatly between male and female participants. [63] However, while many studies included in the present review did collect data from both male and female participants, [54–56,61–66] gendered differences or similarities in perceptions were rarely reported.

Second, as noted above (and applicable especially to our findings regarding the third specific objective), changes in gender dynamics in the settings under study appeared to be caused by a range of diverse, interconnected factors. Hence, it would be impudent to assume that all changes discussed above occurred purely in response to climatic stressors. In our analysis, we relied on the interpretations of the study authors, i.e., whenever a study identified climatic stressors as one reason for changes in gender dynamics, we assumed that to be true. Despite these limitations, we believe that the present review adds value to the discourses concerning the gendered nature of CCA in rural East Africa by virtue of its systematic and comprehensive approach to synthesising relevant qualitative evidence.

Intra-household CCA dynamics and gender transformation

From a theoretical perspective, our findings appear to reflect Carr’s framing of agrarian livelihoods as projects that structure household members’ roles and activities in pursuit of a specific, though ever-changing set of social and material goals (cf. Clarification of concepts, for a more detailed description of Carr’s livelihood model). [24] As described above, Carr posits that re-negotiation of gender roles and identities is most likely to occur when agrarian livelihoods are failing or threatening to fail, and that these re-negotiations are shaped by power structures and often pose new risks to different household members, especially those who are most vulnerable [24].

The present review has revealed that rural livelihoods in settings across East Africa are faced with a wide array of social and environmental stressors, including changes in climate. Households’ responses are varied and can be seen as falling on a spectrum from reinforcing rigid gender roles and identities to opening up spaces for re-negotiation. For instance, women’s tendency to skip meals in order to fulfil their social duty of ensuring that all other household members are fed, seems to reflect a rigidification of gendered livelihood roles at the expense of vulnerable household members. At the other end of the spectrum, women’s increased engagement in income-generating activities and the increase in de-facto female-headed households might point towards more transformative shifts in gender relations, which may have both positive and negative consequences for women and households.

At first glance, this wide range of adaptive responses appears to deviate from the findings of a recent review concerning “gendered dimensions of Climate-Smart Agriculture in Kenya” [7,p.1], which found that decision-making about Climate Smart Agriculture consistently reinforced inequitable gender norms and roles. [7] While our findings do indicate that CCA might reinforce inequities in many instances, we have also found evidence of autonomous CCA processes that appeared to open up spaces for transformative re-negotiation. However, our review found no instances of external CCA interventions leading to such positive transformations. Hence, given that Climate-Smart Agriculture tends to be promoted by stakeholders that are external to households and communities, [7] our findings could in fact be considered as being in alignment as those of Brisebois et al. concerning CSA adoption in Kenya [7].

In general, our findings seem to align with Carr’s observation that re-negotiations of gender dynamics are more likely to occur when livelihood projects fail or are threatening to fail to provide basic material security. [24] However, the threshold for such changes appears to vary between settings and households, with some households adhering to rigid gender roles even when these threaten the subsistence of individual household members. These differences appear to be partly determined by local adaptation contexts and households’ options for re-considering their livelihood projects, e.g. by engaging in new income-generating activities or employing novel agricultural strategies.

Moreover, our findings align with Carr’s observation that, when stressors lead to a re-defining of livelihood projects, the ensuing changes in roles and activities tend to be associated with distinct risks for different household members. [24] This was evident, for instance, in the experiences of women who had become de-facto household heads due to male out-migration: While their changed position within the household might have increased their decision-making power and autonomy, many had to contend with challenges of access to resources and social standing within their communities, [65] as well as increased workloads [56].

Male household members, too, appeared to be threatened by changes in livelihood projects and associated shifts in gender dynamics, though the risks they faced generally were of a more social, less existential nature. For instance, some studies included in the present review found that gender-based domestic violence had surged when women increased their engagement in income-generating activities because male household members perceived these activities as threatening for their identities as primary providers. [e.g. 63] In a similar vein, some male participants expressed a fear of social repercussions from other community members who might view them as incapable of providing for their households. [63] This finding aligns with Carr’s observation that livelihood projects serve as a means to obtain social as well as material goals, and that powerful household members, especially male household heads, tend to determine these goals and may use means of coercion to sanction other household members’ non-compliance [24].

However, female participants in the studies included in the present review tended to continue to engage in income-generating activities despite the attempted coercion because of the great perceived threat of failure of more traditional livelihood projects. Male household members appeared to eventually accept these activities once they recognized the associated increase in household income. According to Carr, male household members may be more likely to tolerate or even support increased productivity among women if they are themselves secure in their gendered identities as primary providers. [24] Unfortunately, we cannot conclude with certainty whether this tendency was present in the households analysed in this review because only few of the included studies consistently provided information about household wealth and material security.

Implications for gender-responsive adaptation interventions

The previous section suggested that our findings may hold valuable insights regarding the potential for gender transformation through CCA in East African rural households. This section explores resulting implications for gender-responsive adaptation policies and initiatives, i.e., external CCA interventions. According to Carr, most development and adaptation interventions that aim to foster more equitable gender dynamics in agrarian households are unsuccessful—either because they fail to challenge the underlying power relations that lead to inequitable outcomes, or because they disrupt current livelihood projects but fail to support the creation of viable, contextually appropriate alternatives. [24] Evidence regarding external adaptation interventions was scarce in the present review, with some notable exceptions. [49,58,65] The few interventions that were discussed predominantly fell under the first category, i.e., they did not challenge the root causes of inequality. For instance, studies included in this review found that interventions which attempted to “empower” women by providing access to credit, land, other productive resources or information were often ineffective because the provided resources were co-opted by male household members or because rigid gender norms prevented women from exercising control over the use of their knowledge, skills and assets [58,65].

Rather than ensuring that male household members were secure in their identities as primary providers and could thus concede more agency to female household members, [24] these interventions appeared to attempt to “empower” women by enhancing their standing vis-a-vis men. When applied in isolation, these approaches appear to disregard the complexity of intra-household power relations and the importance of livelihood projects as means for achieving social goals, and might thus add to gender conflict rather than relieving tensions and opening up spaces for effective re-negotiation.

There certainly is a case to be made for targeted interventions that provide immediate relief from the acute risks faced by many women and girls in East African rural households. We agree with Galiè and Kantor’s notion that “[b]oth gender accommodative and transformative approaches can add value, […] with the mix of approaches at different points in the change process determined by contextual conditions.” [22,p.195] However, it appears crucial that stakeholders involved in designing, implementing and monitoring gender accommodative interventions are aware that their initiatives are likely to cause some disruption to current livelihood strategies, which may lead to novel risks and unexpected consequences for the communities, households and individuals they attempt to support [24].

Findings from the present review suggest that all external CCA interventions in this context ought to recognize not only that rural livelihoods are under stress, but also that communities, households and individuals are actively responding to these stressors and, in some settings, have begun to re-negotiate livelihood projects and associated gender roles and identities. For instance, our findings suggest that there may be an increase in joint decision-making processes in some settings, and that joint decision-making might involve novel challenges, such as greater time demands and sources of potential conflict. An external intervention aiming to foster equitable and effective joint decision-making might be most promising when it is built on a deep and comprehensive understanding of the underlying sociocultural dynamics and actively encourages participants to co-create alternative livelihood projects that can benefit all members of a household or community.

This underscores the importance of working with local conceptualizations of “empowerment” in the context of GTAs, rather than attempting to impose externally generated notions of who should be empowered to do what and in which manner. [22] Moreover, this approach might deviate slightly from previous operationalizations of GTAs, which emphasise the importance of fostering reflections about gender norms among participants [23]: If one views gender roles and identities as resulting from livelihood projects that are developed and maintained in pursuit of a specific, though dynamic set of material and social goals, [24] then discussions about gender transformation and equity become inextricably linked to the livelihood project and goals in question. Interventions that adopt this premise might initiate discussions and re-negotiations of gender dynamics not by prompting reflections on gender dynamics directly, but rather through reflections on women’s and men’s shared and individual livelihood aspirations.

The present review also contributes to the gender-transformative design and implementation discourse in another manner. As explained above (cf. Background), GTAs are commonly conceptualised as spanning three primary dimensions: agency, relations and institutional structures. [23] The gender dynamics of intra-household CCA negotiations and decision-making are most closely aligned with the relational dimension of GTAs. Regarding the relational dimension of GTAs, our findings further showed that women’s approaches to managing their relationships with their husbands depended not only on the quality of the dyadic relationship between husband and wife, but were influenced by a complex web of reciprocal relations spanning women’s support networks and kin beyond the confines of the household. This observation indicates that gender-responsive adaptation interventions might do well to consider not only marital relationships, but the complex web of relations that different household members navigate within and beyond the household.

Furthermore, the findings of the present review underscore the importance of gender-transformative interventions that address all three primary domains, as suggested by Hillebrand et al. [23] For instance, our findings suggest that agency, e.g., in the form of control over cash and productive assets, appears to have a critical influence on household members’ decision-making power. [51,52,58,61,63,65] Further, formal and informal institutions, e.g., policies governing land ownership and sociocultural norms regarding gendered divisions of labour, also appeared to shape the roles and identities that women and men embodied in intra-household CCA negotiations. [e.g. 61] However, analyses of external adaptation interventions revealed that addressing any of these factors did not appear to generate more equitable outcomes if intra-household relations remained unchanged [e.g.58].

In sum, the present qualitative systematic review has shown that significant evidence gaps remain regarding the interplay of gender relations and CCA decision-making in rural East African households, especially concerning nuanced descriptions of intra-household CCA negotiations about external CCA interventions. Nonetheless, our analysis has revealed that the evidence base has grown substantially over the past five years. When synthesised, this knowledge significantly contributes to our understanding of the complex, context-dependent dynamics linking gender relations and intra-household CCA in rural East African households. We hope that the present review will provide guidance for policy-makers and practitioners who design, implement and evaluate gender-responsive CCA interventions in rural East Africa.