Gender norms and relations: implications for agency in coastal livelihoods

Improving livelihoods and livelihood opportunities is a popular thrust of development investments. Gender and other forms of social differentiation influence individual agency to access, participate in, and benefit from existing, new, or improved livelihood opportunities. Recent research illustrates that many initiatives intended to improve livelihoods still proceed as “gender blind,” failing to account for the norms and relations that will influence how women and men experience opportunities and outcomes. To examine gender in livelihoods, we employed empirical case studies in three coastal communities in Solomon Islands; a small island developing state where livelihoods are predominantly based on fisheries and agriculture. Using the GENNOVATE methodology (a series of focus groups) we investigated how gender norms and relations influence agency (i.e., the availability of choice and capacity to exercise choice). We find that men are able to pursue a broader range of livelihood activities than women who tend to be constrained by individual perceptions of risk and socially prescribed physical mobility restraints. We find the livelihood portfolios of women and men are more diverse than in the past. However, livelihood diversity may limit women’s more immediate freedoms to exercise agency because they are simultaneously experiencing intensified time and labor demands. Our findings challenge the broad proposition that livelihood diversification will lead to improvements for agency and overall wellbeing. In community-level decision-making, men’s capacity to exercise choice was perceived to be greater in relation to livelihoods, as well as strategic life decisions more broadly. By contrast, capacity to exercise choice within households involved spousal negotiation, and consensus was considered more important than male or female dominance in decision-making. The prevailing global insight is that livelihood initiatives are more likely to bring about sustained and equitable outcomes if they are designed based on understandings of the distinct ways women and men participate in and experience livelihoods. Our study provides insights to make these improvements in a Solomon Islands setting. We suggest that better accounting for these gendered differences not only improves livelihood outcomes but also presents opportunity to catalyze the re-negotiation of gender norms and relations; thereby promoting greater individual agency.


Introduction
In many developing countries and small island contexts, human wellbeing is tightly tied to primary productivity, often accessed via fisheries and agriculture. In these contexts, livelihoods are a common entry point to drive improvements to wellbeing (Ellis 2000;Vijaya et al. 2014). Here we broadly define a livelihood as means of generating income, securing food, or spending time (Jiao et al. 2017;Nielsen et al. 2013). A precondition for improving wellbeing through livelihoods is an understanding of how gender inequalities have implications for individual agency (Kabeer 1999a). Sen (1985, p. 203) defines agency as what a "person is free to do and achieve in pursuit of whatever goals or values he or she regards as important." Having choice and exercising choice, by this definition, are essential for individuals to access, participate in, and benefit from livelihoods opportunities to enhance their own wellbeing. While many studies propose a range of indicators of agency (e.g., Ibrahim and Alkire 2007;Kabeer 1999b;Sen 1985), having choice and exercising choice are frequently cited as central elements, and are strongly correlated with the manifestation of gender inequalities (Boudet et al. 2013;Ibrahim and Alkire 2007;Kabeer 1999b;Malhotra et al. 2002).
Agency can vary between individuals as a result of the differing sets of choices available to women and to men, and differences in their capacity to exercise these choices (Boudet et al. 2013). In short, the conditions shaping individual agency are gendered. In many development contexts, opportunity structures (i.e., education, information, extension services) tend to favor men, elevating them into positions where they are more able than women to access and control productive assets (i.e., land, income, equipment, technology) and natural resources (i.e., fish, land, and produce) (Meinzen-Dick et al. 2014). Women tend to be less able to make claims on natural resources and determine the direction of decisions related to assets and resource use within households and communities (Agarwal 1997;Okali 2006). In situations where income poverty and geographical remoteness are greater, the disparity in the choices available to women and men, and opportunities to exercise these choices, become more extreme (Boudet et al. 2013).
Gender inequalities in individual agency are underpinned by norms and relations that regulate the different roles, responsibilities, and expectations society ascribes to women and to men. Gender norms are the attitudes and informal "rules" that govern behaviors considered to be appropriate, acceptable, or desirable for women and for men within a particular society (Boudet et al. 2013). Gender relations refer to the relationships between women and men, and how these relationships are influenced by, and in turn influence, the social expectations of women and men in society (Agarwal 1997). These norms and relations are themselves expressions of, and produce, different manifestations of agency by shaping individuals abilities to act freely and have choice (Boudet et al. 2013). A study examining livelihoods across 20 primarily developing countries, found that women's agency is more closely bound by traditional gender norms and relations than men's (Boudet et al. 2013). Agency is not only influenced by gender but also intersects with other common "markers of disadvantage" including poverty, age, ethnicity, religion, and disability status that can accentuate the effects of gender norms and relations (World Bank 2013, p. 39). Development initiatives that alter, supplement, or diversify existing livelihoods will have gendered impacts; critically, even where they do not include an explicit focus on gender (Resurrección and Elmhirst 2009;Stacey et al. 2019). Yet, recent research highlights that many initiatives intended to improve livelihoods still proceed as "gender blind" (Kleiber et al. 2019;Stacey et al. 2019). These initiatives tend to focus on either women or men and not account for the influence of gender norms and relations on opportunities and outcomes. The way livelihood initiatives are designed and delivered can reinforce, maintain, or shift gendered patterns in the divisions in labor, participation in decision-making, and access to/ control over assets and resources in households and communities (Okali 2006;Resurrección and Elmhirst 2009). If gender is not accounted for, inequalities in women's and men's agency to negotiate their socio-economic conditions and maintain their wellbeing may be perpetuated or even exaggerated (Kawarazuka et al. 2016;Nightingale 2006;Resurrección and Elmhirst 2009). In order to improve opportunities and outcomes for both women and men, livelihood initiatives (i.e., those that introduce new and/or altered farming and fishing methods/management, or marketing strategies) need to be designed to consider how gender inequalities affect individual agency.
Yet, in many contexts understandings of the influence of gender upon agency and upon livelihoods is lacking. Specific knowledge gaps relate to how gender norms and relations shape the different choices individuals have to access and participate in livelihoods and their ability to exercise that choice. In this paper, the overarching research question we seek to answer is: how do gender norms and relations influence the expressions of agency of women, men, and youth in their livelihoods? We address this question by capturing the gender-differentiated experiences of women, men, and youth using an established methodology that combines qualitative and quantitative techniques. We develop this understanding in three coastal, relatively remote communities of Solomon Islands, a small island developing state, where the majority of the population are highly reliant on coastal resources. Our empirical data relate to current livelihoods and include some inisghts from externally delivered livelihood initiatives these communities have previously engaged with. We structure our results according to the description of agency that Boudet et al. (2013) offer which distinguishes between (a) choices of individuals to access and participate in livelihoods and (b) individual capacity to exercise choice in livelihoods (including the new or altered livelihood initiatives being facilitated). We make this distinction to avoid the common assumption that access to livelihoods enables or equates to an individual's ability to exercise choice (Boudet et al. 2013;Kabeer 1999b). As Kabeer (1999b) expresses, it is import to differentiate between access to choice as measure of potential ability and the actualization of choice.

Methods and study sites
Context Solomon Islands is a small island developing state situated in the south-western Pacific Ocean. The nation is the third most populous of the Pacific Islands Countries and Territories with approximately 600,000 inhabitants; 80% of whom reside in rural areas (Solomon Islands National Statistics Office 2009). Solomon Islands is ranked low in human development and is placed at 152 of 189 countries in the UNDP human development index (UNDP 2018). Land (and to some extent coastal marine areas) is governed and allocated through customary tenure systems, and 87% of land is customarily owned (AusAid 2008). The majority of the population are dependent on subsistence and/or small-scale agriculture (89%) and fisheries (60%) for household food and income, with under 20% of the population participating in salaried employment (Solomon Islands National Statistics Office 2009).
Customary practices and beliefs, Christianization, colonization, and more recently, independence have profoundly affected the contemporary culture, social structure, economy, and the use and status of natural resources in Solomon Islands (Allen et al. 2013;Bennett 1987;Foale and Macintyre 2000). These historical periods have shaped deep-seated gender norms and relations that are known to influence decision-making related to land and coastal resources (Akin 2003;Foale and Macintyre 2000), divisions in labor, and broader expectations of moral behavior (Cohen et al. 2016;Pollard 2000). In terms of livelihood activities, women tend to be primarily responsible for crop farming, with an estimated 71% of women engaged in subsistence farming in comparison to 51% of men (JICA 2010). Men's participation is higher in reef and offshore fishing, while women generally participate in inshore coastal environments, such as lagoons and mangrove areas (Kronen and Vunisea 2009). While most people in Solomon Islands' sustain their daily livelihood needs through primary production, people living in more geographically remote areas simultaneously experience what is described as "poverty of opportunity" (Lightfoot et al. 2001). This means there are few opportunities for people to change from subsistence livelihoods or to bring about improvements to their living situations. In an attempt to redress this, many development initiatives have used different approaches to improve or diversify opportunities available for women and men in coastal rural areas-reflecting a broader Pacific and global trend (Cinner and Bodin 2010;Gillett et al. 2008).

Study area
The research was undertaken between September 2014 and September 2015 in three coastal communities; one community in Western Province and two communities in Malaita Province (Fig. 1). Each community comprised of a cluster of between four and 10 villages. These villages are geographical proximate and have historical social alliances, and the purposes of this paper, we refer to these village clusters as one community.
Western and Malaita Provinces reflect the national trend of high rates of participation in subsistence and/or small-scale farming (Western 93% and Malaita 95% of households) and fishing (Western 83%, Malaita 49% of households). The differences between engagements in fisheries in the two provinces are partly explained by the relatively greater proportion of the population residing in coastal areas in Western Province compared to Malaita. Western Province has a higher rate of salaried employment (20%) compared to Malaita (9%), mostly attributable to the greater rates of commercial logging and tourism in the West (Solomon Islands National Statistics Office 2009).

Study sample
Our selection of Solomon Islands as a case study was opportunistic; as it was a focus of the CGIAR Research Program on Aquatic Agricultural Systems and projects funded by the Australian Centre for International Agricultural Research that were focused on community engagement to realize improvements to livelihoods. Communities involved in the program were selected because they had the following: (i) a high dependence on aquatic (i.e., mangrove, reefs, and coastal) resources and/or terrestrial (i.e., forest and agricultural plots); (ii) experienced resource decline associated with fisheries and/or agriculture; and, (iii) expressed an interest in receiving support to improve livelihood opportunities and the condition of their productive resources. At the time of data collection, the communities studied had been involved in the program between 1 and 2 years. Activities undertaken prior to this study included community consultations for preliminary scoping and agreement to research and the participatory development of broad-scope community action plans. Preliminary and collaborative activities that followed included the development of fisheries management plans, inter and intra-community sharing of farming techniques, and training in organic farming methods. In Community 1, a women's savings club 1 was also initiated by an external organization prior to engagement with program.

Data collection
Data were collected using four different focus-group discussion (FGD) formats (Table 1). Questions were designed to explore how gender norms and relations influenced the wellbeing of community members. The FGD formats used were contextualized versions of the tools developed for GENNOVATE 2 , a comparative global research initiative examining gender norms and agency in natural resource management (see Badstue et al. 2018;Petesch et al. 2018). The GENNOVATE methods were selected because they had been developed and peer reviewed by a group of gender and development experts and were designed explicitly to examine capacities (including agency) to innovate in livelihood domains and offered an opportunity for subsequent global comparison.
FGDs were designed and written in English, and then translated, tested and modified by local researchers fluent in English and Pijin to clarify any ambiguities. FGDs were conducted in Solomon Islands Pijin, which was the common language across all communities (and between researchers and respondents). Prior to commencing research, we held, with community leaders, a community meeting where the research objectives were discussed and a schedule of FGDs was drafted. After the meeting, respondents volunteered their participation for the FGDs and provided verbal informed consent. Due to community confidentiality agreements, community and respondent identities remain anonymous.
A total of 24 FGDs were conducted with 232 respondents across three communities. Eight FGDs were held in each community over a period of 5 days with separate groups of (a) adult men (n = 79), (b) adult women (n = 92), (c) male youth (n = 45), and (d) female youth (n = 16). Respondent demographic data were collected prior to each FGD. Respondents were aged between 16 and 70 years of age, resided locally and were actively engaged in fishing and/or agriculture (Table 2). Youth were unmarried and between the ages of 16-24. FGDs were held with between five and 20 people and took between 40 min and 5 h.

Data analysis
FGDs were recorded digitally and in writing. Transcripts were translated from Pijin into English using digital and written recordings. Data were coded in NVivo10. Preliminary coding was undertaken using a coding structure developed from the GENNOVATE study, which consisted of theory-driven codes (overarching themes included gender norms, agency, and agricultural innovation) and data-driven codes based on sampling a sub-set of transcripts. We then analyzed data for emergent themes through an iterative process that involved the comparison of data between respondent groups and between Table 1 Objectives and scope of focus-group discussions Format Objective FGD 1 To explore women's and men's experiences with and perceptions of norms and relations shaping gender roles, household, and community decision-making related to livelihoods, women's physical mobility, and access to livelihood opportunities FGD 2 To explore women's and men's experiences with and perceptions of factors enabling and constraining innovation, opportunities for agriculture entrepreneurship, social cohesion, and networks FGD 3 To explore women's and men's experiences with and perceptions of inequality and social differentiation, including factors shaping socioeconomic status and mobility and their gender dimensions FGD 4 To explore female and male youths' experiences with and perceptions of norms and relations shaping gender roles, household, and community decision-making related to livelihoods, women's physical mobility, access to livelihood opportunities, social cohesion, and networks communities. Data from FGDs were predominantly qualitative, however, FGD format one and two involved collection of Likert scale data. These quantitative data were analyzed using Pearson's chi-square test for independence.

Livelihood choices
Our study sought to explore the manifestations of agency through the livelihood choices available to women and men. Demographic data illustrated that primary or secondary livelihood activities of respondents primarily focused on the production of food and income through fisheries and agriculture in close proximity to their communities. Across all three communities, 91% of women reported farming as a primary livelihood activity, and 81% of men reported they either farmed or fished as their primary livelihood activity. It was common across all communities for households to have small agricultural plots (referred to as gardens) located on family-owned land. Community 1 was geographically closer to a large regional center than Communities 2 and 3. Discussions of both women and men in Community 1 reflected greater access to salaried employment than the respondents of Communities 2 and 3, where pursing salaried employment would require migration to an urban center.
In this section, we present the results and discuss how gender norms and relations (and the shifts in these expectations) have shaped divisions in labor ("Gender norms and divisions in labor" section), physical mobility ("Gender norms and physical mobility" section) and influenced the livelihood activities individuals were able to pursue. We then examine the gendered impacts of livelihood diversification in response to the introduction of new or altered livelihood initiatives ("Gendered impacts of livelihood diversification" section).

Gender norms and divisions in labor
To supplement our demographic data on livelihood roles FGDs explored underlying norms. Discussions of norms shaping gender roles in FGD formats one and four indicated distinct perceptions of divisions in labor and livelihood activities that were socially appropriate for women and men to participate in. These perceptions in turn influenced the livelihood choices available to individuals. Adult women reported that "gardening is our work," whereas men reported a diversity of livelihood practices in addition to gardening, such as building and selling hand-carved dugout canoes, cutting and selling firewood, building houses for informal salaries, and fishing for both food and income. Men in Community 1 also reported they were able to work for nearby logging companies. Despite women's greater access to salaried employment in Community 1, across all communities, adult women suggested they had access to limited opportunities, "some of us women only have our garden for our livelihoods." Men had a greater set of choices to access and participate in livelihoods. This is consistent with the findings of Scheyvens (2003) who suggests that the program of the early missionaries in Solomon Islands aimed at restricting women to the domestic sphere while encouraging men to be part of the growth of the modern economy.
Respondents reported that customary beliefs influenced contemporary livelihoods. Both female and male respondents reported that, in the past, domestic labor was primarily a woman's responsibility, and according to local custom, men were forbidden from cooking and washing women's clothes. As one adult woman reported, "Men were the boss and were served by the women, like a chief." There was evidence that these expressions of masculine status were maintained in present-day livelihood practices. Both female and male respondents reported women's ability to leave the household or community (e.g., to attend markets) was hindered because in her absence her husband would have to undertake "women's work." Both female and male respondents reported that in this situation the husband and wife risk criticism from other community members and " … they will say she must be the boss of her husband." While we do not have data to suggest this was the case here, it has been found elsewhere that where people are exhibiting increased agency in a way that challenges existing power relations, there is a risk of increased tension in relationships, and even violence, towards individuals or groups that exhibit greater agency (Boudet et al. 2013).

Gender norms and physical mobility
Across all three communities, female and male respondents reported women faced restrictions in traveling to sell their produce at markets in regional centers. An adult male respondent expressed "there is no reason for a woman to go out marketing, she is supposed to be staying at home with the kids." Another male respondent suggested there was also some level of distrust associated with women traveling away from the community; "[a] woman makes the husband work hard in doing everything at home from looking after the kids, cooking, washing and going to the garden whilst the woman is out somewhere doing marketing. Who knows what she is doing? She can do anything she wishes in the absence of her husband …" However, different views were shared in Community 1 where a women's savings club initiative had appeared to increase the social support and acceptance women received to attend markets. One woman reported "before we had a savings club our husbands didn't let us do anything. If we came back late they would be cross. But now if we come back late they hug us!" Individual physical mobility (and migration more broadly) is viewed as a fundamental capability shaping individual freedom (de Haas 2009). In the instances where women migrate or become more physically mobile, the potential for gender roles to shift becomes greater, allowing more flexibility in divisions in labor and livelihood pursuits (de Haas 2009). Yet, research in other developing contexts has established that even in the instances where there are no restrictions on women's physical mobility, the jobs that women undertake outside of their communities are merely extensions of their domestic roles (i.e., teaching, nursing, or cooking) (Boudet et al. 2013;Start and Johnson 2004). This suggests that physical mobility freedoms do not necessarily correlate with greater agency to pursue a range of livelihoods, as there is still a need to reconcile livelihood choices with the norms that determine the appropriate roles for women and men to undertake.

Gendered impacts of livelihood diversification
We found evidence that historically strict gendered divisions in labor had become more relaxed and the livelihoods women and men were participating in had diversified. An adult woman reported "before, because of custom, men did not do the work of women, like washing clothes. Now it's changed. If the men husk coconuts, the women husk coconuts too." Female respondents reported they had more recently become actively engaged in net fishing, an activity once only undertaken by men. In the instances where livelihood initiatives had led to increased income and food for the household (i.e., from the women's savings club and organic farming), our results indicated a correlation with the destabilization of norms related to divisions in household labor. In these situations, men became more prepared to undertake "women's work" to allow their wives to continue production and sale of products. One adult man reported "today women can instruct their husbands to clean the house when they are away. This is not something that was practised before." One adult male respondent reported, "since [external organization] have come into our village I've realized the term 'gender' and today women and men in the village share responsibilities. Men can wash dishes and women can cut the firewood too." Some respondents suggested these shifts in norms were facilitated by exposure to external organizations, population growth, and increased pressure for primary production. Other studies suggest that traditional divisions in labor have destabilized overtime due to an increase in women accessing formal education (Pollard 2000), and specific to coastal livelihoods, shifts towards cash-based economies (Barclay et al. 2018) leading women and men to question traditional norms.
While we found evidence of shifts in divisions in labor, these were not always perceived as positive. One adult woman reported "Life now is hard … [in the past] men had their own work, and women had their own work. Nowadays women's work is heavy … Before, carrying water and hoeing the garden only the men did. But now, the women are doing this work." Female respondents in Community 1 reported that because of the savings club, they now not only needed to maintain their domestic roles but also needed to find time to participate in new livelihood activities. Our results are consistent with other studies from Solomon Islands that suggest women's labor demands continue to escalate as livelihood activities diversify (Cohen et al. 2016;McDougall 2014;Pollard 2000). Through the frame of adaptive capacity, higher livelihood diversity is considered to be a positive attribute correlated with a greater ability to adapt in the event of social or ecological change (Cinner and Bodin 2010;Cohen et al. 2016). Yet, women's own reflections indicated that this had led to a greater labor burden. These findings indicate that a diversity of livelihood choices may limit women's more immediate freedoms to exercise agency as a result of increasing responsibilities and time pressures. As argued by Start and Johnson (2004), having many livelihood choices does not necessarily equate to women's freedom to depart from entrenched gender roles.
We explored factors shaping individual choice to trial and adopt new agricultural activities. An example raised in discussion was organic farming practices that varied from the popular practices of renewing a garden plot referred to as "slash and burn." Across all three communities, 91% of women reported they were primarily responsible for land-based food provisioning. Female respondents reported their adoption of new practices was limited by perceived risks associated with experimenting with new methods; "those who practice organic farming go hungry for some time until they start to reap the yields." These results suggested that women held greater concerns about limited or delayed rewards in trialing new agricultural practices (such as those introduced externally). We found among women and men, the willingness to adopt new practices would be higher with prior evidence of success, "people in the village want to see results first before they try new things." This finding emphasizes the importance of addressing these risks for women, and the perceptions of risks among both men and women (as opposed to initiatives focusing solely on technical or knowledge gaps), before any progress might be made along an agricultural-livelihood improvement pathway.
Risk perception and exposure affects the choices available to an individual (Gustafson 1998). In many contexts, women are found to be more risk averse than men as perceptions of risk are reinforced by norms that promote the reproductive responsibilities of women, leaving women with less time and physical space to experiment and innovate (Fothergill 1996;Gustafson 1998). Willingness to bear risk in trialing new or altered livelihoods is found to be influenced by the presence, absence, or quality of relationships with external organizations affiliated with livelihood initiatives (Cohen et al. 2016). Initiatives that can help carry the cost of innovation, with particular recognition to the constraints of women, are more likely to enhance opportunities to access, participate in, and lead to improved outcomes (Cohen et al. 2016).
Scholars examine historical patterns of labor as a starting point to understand the contemporary cultural distinction and views of women's and men's roles in livelihoods (Pearson 2000). The destabilization of these gender norms that shaped historical divisions in labor can open up spaces for women to innovate and experiment with new or altered livelihood activities (Boudet et al. 2013). In developing country contexts, high livelihood diversity can be viewed as a safety net to maintain basic needs where risk is spread (Ellis 2000). On the surface, women's increased participation in net fishing might be interpreted as an indication of greater livelihood choice, however, our findings signaled that diversification in this case was associated with intensification of women's labor. In case studies that employed the same methodologies in Cambodia and Philippines, Locke et al. (2017) found that diversified livelihoods of women could also represent family hardship, where a woman needs to add value to her husband's enterprise-the consequence being a greater labor burden and further constraints upon agency. Interestingly, perception of risk (found to be greater for women) was in fact a strong reason given for not trialing new livelihoods. In our case, diversified livelihoods did not represent increased choice in the way in which individuals generate income, secure food, or spend their time. Just as important to understand, although not examined here, are the structural factors such as the political institutions, constraints of geography and rurality, market and economic opportunities, and the state (Agarwal 1997;Start and Johnson 2004).

Capacity to exercise choice in livelihoods
Understanding agency in the context of livelihoods requires identifying both gender differences in the availability of choice, but also individual capacity to exercise that choice. The conditions in which individuals exercise choice in livelihoods is affected by their ability to make strategic life decisions, not only in relation to livelihoods but within households and communities broadly (Boudet et al. 2013). Community and household decision-making domains are commonly analyzed separately in gender literature in order to distinguish between the different gender-based constraints at these scales (Agarwal 1997;Kabeer 1999b;Malhotra et al. 2002). Using a visual representation of a "power" ladder, adult females and males scored their power and freedom to make "all" to "none" of their own life decisions. Quantitative results indicated only a marginal difference in the experiences of women and men, where slightly more women (57%) than men (50%) indicated that they had the power and freedom to make "most" to "all" of their own life decisions (n = 64, df = 4, p = > 0.01). Although the quantitative difference was slight, our examination of qualitative responses illustrated that women and men were in fact referring to decision-making within different domains. Men's discussions on power in decision-making were dominated by references to decisions made at the community level (i.e., in relation to schools, the church, and management of coastal resources, specifically land, fish, and reefs). By contrast, women most commonly discussed decision-making at the household level (i.e., in relation to children, crop farming, and household consumption). These results illustrate two quite different points. First, that a relative measure of agency is dependent on particular settings, social hierarchies and individual values. These findings expose the different realms of decision-making women and men are exposed to, and, consistent with Sen's definition of agency, their ability to act on behalf of what an individual values and has reason to value. Second, the distinctions in our qualitative data highlight some limitations of quantitative methodologies, described as "simple windows on complex realities" for social and gender analysis (Kabeer 1999b, p. 447). The pairing of the both quantitative and qualitative data here provided a view of women's and men's relative agency in different decision-making domains; it was the qualitative data only that enabled us to see distinctions in the type of decisions being made. In the following sections, we explore women's and men's power and freedom to make decisions in the community (including through external support structures) ("Exercising choice in communal domains" section) and within the household ("Exercising choice in the household domain" section) in more detail.

Exercising choice in communal domains
Decisions made in communal domains can affect the sets of livelihood choices individuals have and capacities to exercise those choices (Agarwal 1997;Meinzen-Dick et al. 1997). In Solomon Islands, community leadership structures and relationships of power are underpinned by customary tenure systems as well as gender norms and relations that influence the different opportunities individuals have to benefit from livelihoods and to participate in their governance (Hviding 1998;JICA 2010).
Land and marine tenure operated through a system of matrilineal descent in Community 1 and patrilineal descent in Communities 2 and 3. While we did not conduct an in-depth exploration of the influence of customary tenure rights on agency, our results reinforced the findings of others who challenge the assumption that matrilineal descent systems transmit greater decision-making power to women (see Macintyre 2008;Meinzen-Dick et al. 1997). Discussions among men in all communities inferred that in practice, men ultimately act as spokespeople and negotiators in extra-household decisions regarding the use of land and marine resources, irrespective of inheritance systems. Discussions in FGD format one suggested women overall, and men who did not hold primary land rights, were less able to determine the direction of decisions about the use and management of coastal and terrestrial resources-regardless of inheritance systems. Respondents from Community 2 reported that as the number of people involved in farming had increased, the land available for gardening had become scarce and disputes over land and tenure had increased. Literature examining tenure rights pay close attention to gender disparities (Meinzen-Dick et al. 1997;Okali 2006). Yet studies that use measures of women's access to or control over coastal or terrestrial resources often fail to demonstrate how such rights translate into agency, specifically freedom to exercise choice (Kabeer 1999a).
FGD formats one and four explored norms and relations that shaped gender roles and decision-making power. Responses suggested that men had more explicit involvement in local community and traditional governance structures than women. The importance of intersectionality became clear in Communities 2 and 3 where it was reported that migrants and/or men who do not have tenure rights experienced substantially less agency in community decisions. Formal leadership positions (e.g., clan chiefs, village chiefs, village chairmen, church leaders) in all three communities were predominantly held by men. These roles prescribed responsibilities for negotiating, enforcing, and sanctioning (e.g., through compensation payments) community activities and rules. The only exceptions were the leadership positions held by women in women's, youth, church, or community groups. Women's exclusion from communal decisions that enforce or modify rules governing a community (e.g., decisions that affect customary tenure rights, control over coastal or terrestrial resources, and access to external support structures and opportunities) will have adverse implications for their livelihoods (Agarwal 1997;Boudet et al. 2013). For example, a study exploring community-based coastal marine management in Solomon Islands found that women rarely participated in management decisions and associated their exclusion with the closure of an area commonly fished by women (Rohe et al. 2018).
There is evidence to suggest that women may have greater agency when they have access to supportive community groups (Boudet et al. 2013). We found the explicit efforts of non-government organizations to support the contributions of women in the activities they led (i.e., through the women's savings club in Community 1) appeared to increase women's self-efficacy, brought social and economic security to households and communities, and led to improvements in overall wellbeing. One woman explained, "when organizations come into our community we see the light. Like when you [external organization] come, you educate us and open our minds. That's why we know we have the right to make decisions and we feel free to speak out. Before our mouths were zipped. We had good ideas, but we never voiced them. If we voiced our ideas, no one would follow them." From an etic perspective, we observed that the very act of engaging with external organizations increased the self-confidence of both women and men in dealing with "outsiders." Nonetheless, this confidence is not necessarily stable as flow on effects for the empowerment of women through access to community groups are certainly not guaranteed. Malhotra et al. (2002, p. 8) contend that women's access to external support and resources should be perceived as "enabling factors" and not be interpreted as "proxies" for empowerment. This was true in our case where respondents reported that external livelihood initiatives had sought to increase women's representation in decision-making positions through encouraging women into community governance structures and leadership (i.e., in marine management committees). While this had transformed the formal governance arrangements, female respondents suggested that this did not necessarily translate into women's greater voice or influence (see also Cornwall 2003). Further research might examine whether increased confidence and greater representation translates to broader agency and overall wellbeing of both women and men.

Exercising choice in the household domain
When exploring individual agency to make decisions within households, our quantitative results indicated that women perceived they had a moderate degree of power. During discussions, some women suggested they were "the boss of the house," simultaneous views reported that men function as the "household head." These discussions suggested women had power to make small decisions relating to their family (i.e., how many crops to sell versus consume), they were less able to contribute to strategic life decisions (e.g., about large household expenditures). In FGD format one, which explored perceptions of gendered decision-making power, a hypothetical situation was presented (twice with roles reversed) about the autonomy a wife/husband would have to purchase an item without requiring the wife's/husband's approval ( Fig. 2; Table 3). Women's and men's power in household decisions varied with a weak positive correlation indicating women faced slightly more difficulty than men. We found the views of youth were more polarized; the majority of both female and male youth reported it would be "hard"-"very hard" (52%) for a wife, but "very easy" (56%) for a husband to make decisions regarding the use of money in the absence of his wife's support. The more rigid views held by youth were expressed through references to men as the "head of the house" and "the boss" more frequently than adults. While our findings do not provide evidence of the persistence of such beliefs, it is possible they may relate to the geographic isolation of the communities in which we worked. Consistent with Whitehead et al. 2007, it is also feasible that youth tended to represent more strongly in group discussions the views of how they felt things should be. Whitehead et al. suggest that the formation of gendered identities among youth are shaped by both implied and overt expectations held by family and wider social networks, which are influenced by gender and cultural norms. This would be an area worthy of further exploration in this specific context.
Results indicate the statistical significance (p < 0.05) of responses and respondent category. The relation between these variables was not significant in all cases.  Fig. 2 Responses across all three communities indicating the ease or difficulty with which a a wife or b a husband could proceed with a purchase (i.e., a sewing machine or an outboard engine, respectively) without approval from their spouse There was a strong consensus in discussions among adults that a husband and wife should share decision-making within the household. This emphasis on negotiation and the need to maintain intra-household gender relations was far more prevalent than expressions that there was, or should be, overt gendered dominance in household decision-making. Even so, men, women, and youth all conveyed that in practice men had the final say in decisions. This finding is consistent with Montgomery et al. (1996), who suggest joint household decision-making can mask male dominance. Other studies find that men tended to support the idea of spousal cooperation and shared opinions in household decision-making, however, only in the instances where they did not disturb existing household power dynamics (Boudet et al. 2013;Locke et al. 2017).
While these results present rigid differences in household decision-making, both women's and men's capacity to contribute to decision-making at the household level is context dependent. This fluid nature of decision-making will often influence the extent to which women will have the power to contribute to the final decision (Agarwal 1997). For example, Pollard (2000) highlights the social and cultural complexity of decision-making in Solomon Islands by suggesting that the dominant ideology that women are subordinate to men, is paralleled by women's own conceptions of their centrality within their households and society more broadly. Creating spaces for women to exercise choice through participatory approaches have become common practice for initiatives. However, external ideals of equality can be inappropriate because, in some instances, women themselves may have a stake in patriarchal arrangements and overtly challenging these arrangements could risk women's means of negotiation (Cornwall 2003). The overt compliance, and the importance some women place on maintaining gender relations and roles, may give women "room to maneuver" through maintaining harmony (Cornwall 2003(Cornwall , p. 1331. Although, Cornwall also highlights the tension that not challenging inequitable relations runs the risk of "reinforcing stakes that maintain a status quo that the marginal have tactics to grapple with, but no possibility of realizing strategies for change because they lack the power and agency to do so (cf. de Certeau, 1984)" (Cornwall 2003(Cornwall , p. 1331).

Conclusion
As development initiatives increasingly turn their attention to livelihoods as an entry point for improving human wellbeing, it becomes important that these efforts consider the way gender influences how individuals experience opportunities and benefits differently (Boudet et al. 2013;Meinzen-Dick et al. 1997;Okali 2006). Despite years of research and best practice guidance, many livelihood initiatives are gender blind, and often persist with a narrow focus on bringing new livelihoods to women (Stacey et al. 2019). Yet, unless norms, roles, and aspirations of both women and men are understood and carefully navigated, there is a risk that initiatives may amplify women's existing workloads under the banner of "participation" or "empowerment," risk backlash from family and community members, and have contradictory consequences for individual agency and overall wellbeing (Cornwall 2003;Resurrección and Elmhirst 2009). Our empirical case studies offer some points of guidance for livelihood investments in coastal communities in Solomon Islands. The methodology we used provides readily accessible tools that could be utilized prior to commencing livelihood initiatives.
By applying the description of agency that Boudet et al. (2013) offer, we emphasize the distinction between the sets of livelihoods available to women and men, and the differences in their capacity to exercise choice between and among these livelihood pursuits. The research tools we employed did not explore livelihood aspirations of women, men, and households, but this is a critical foundation for initiatives seeking to change, add, or improve livelihoods directly. Our findings challenge the broad proposition that diverse livelihoods serve as a safety net to maintain basic needs and spread risk (Ellis 2000). This proposition underplays the risk that diversification may simultaneously increase labor burdens. We found that the livelihoods women and men choose to pursue were restricted by social and gendered expectations and gender-influenced perceptions of risk. The cross-case analysis that employed the same methodology illustrated that these social constructions are not rigid, in that if initiatives work to change beliefs (for example, by legitimizing women's mobility beyond the village) this can lead to a renegotiation of gender relations that expand women's agency to experiment and innovate (see Locke et al. 2017). Further, if compelling opportunities become available, then both women and men may have agency to innovate in ways that defy existing norms (Locke et al. 2017). We found that capacity to exercise choice in livelihoods is correlated with the dynamic interactions that govern individual choice between the community and household settings. Most men (not all) had greater capacity to exercise choice and determine the direction of decisions in community settings. In the household setting, the gendered difference was less distinct, with both women and men emphasizing spousal cooperation as the priority. Individual perceptions of their capability to exercise control in situations that affect their lives is an important mechanism shaping agency (Bandura 1990). Our results highlighted the different domains of decision-making women and men were exposed to and may value. Consistent with Sen (1985), it is important that livelihood initiatives seeking to drive improvements to wellbeing also recognize individual abilities to act on behalf of what an individual values and has reason to value.
Research focused on gender differentiation of roles, expectations, and aspirations can offer critical guidance to ensure that livelihood initiatives, and the outcomes they seek to promote are equitable and contribute towards both sustainable and locally perceived improvements to wellbeing. At the frontier of gender research are gender transformative approaches which suggest that certain interventions can serve as a catalyst for the re-negotiation of gender norms and relations (e.g., Cole et al. 2018). We found some evidence of shifts in norms and relations from engagement with the women's savings club. This engagement had, in part, increased women's and men's openness to new roles and responsibilities within the household and in community governance. Nonetheless, and as cautioned by others (e.g., Nightingale 2006; Resurrección and Elmhirst 2009), whether livelihood initiatives intentionally acknowledge and engage with gender or not, they will interact with gender-in ways that may reinforce, or alternatively, shift existing gender norms and relations thereby having implications for the agency of different individuals. Our findings add weight to others (e.g., Buvinić 1986;Okali 2006) who have established that livelihood initiatives are more likely to bring about sustained and equitable outcomes if they are designed and delivered based on understandings of how women and men participate in and experience livelihood opportunities differently. However, considerable scope remains for research to investigate the manner in which livelihood initiatives can apply this knowledge in a way that challenges and shifts the underlying norms and relationships that perpetuate gender inequality.