Plantations, outgrowers and commercial farming in Africa: agricultural commercialisation and implications for agrarian change

Whether or not investments in African agriculture can generate quality employment at scale, avoid dispossessing local people of their land, promote diversified and sustainable livelihoods, and catalyse more vibrant local economies depends on what farming model is pursued. In this Forum, we build on recent scholarship by discussing the key findings of our recent studies in Ghana, Kenya and Zambia. We examined cases of three models of agricultural commercialisation, characterised by different sets of institutional arrangements that link land, labour and capital. The three models are: plantations or estates with on-farm processing; contract farming and outgrower schemes; and medium-scale commercial farming areas. Building on core debates in the critical agrarian studies literature, we identify commercial farming areas and contract farming as producing the most local economic linkages, and plantations/estates as producing more jobs, although these are of low quality and mostly casual. We point to the gender and generational dynamics emerging in the three models, which reflect the changing demand for family and wage labour. Models of agricultural commercialisation do not always deliver what is expected of them in part because local conditions play a critical role in the unfolding outcomes for land relations, labour regimes, livelihoods and local economies.


Introduction
Debates continue about the relative merits of large and small farms, their implications for labour absorption, rural livelihoods and growth in Africa's farm sector (Lipton 2009;World Bank 2008;Collier and Dercon 2014;cf. Deininger and Byerlee 2011;Baglioni and Gibbon 2013). The recent onset of a 'global land grab' (Borras et al. 2011) and evidence of largescale land acquisitions in Africa has refocused attention on this longstanding debate, in a context of assumed land scarcity (Scoones et al. 2014). As part of this drive for investment, there has been a resurgence of plantations and large-scale estates that mimic (or in some cases revive) large colonial estates and state farms Anseeuw et al. 2012;White et al. 2012). Contract farming has also been promoted as an alternative to the 'land grab' that large-scale acquisitions often entail, and also as an 'inclusive business model' in which local smallholder farmers can participate and from which they can benefit (Cotula et al. 2009;von Braun and Meinzen-Dick 2009;cf. Little and Watts 1994;Oya 2013). Recent research has highlighted a third and increasingly significant category of commercial farming: the growth of 'middle farmers' as a key dynamic driving land concentration in Africa (Jayne, Chamberlin, and Headey 2014;Ariyo and Mortimore 2015).
Another set of relevant debates concerns the social relations of agrarian change (Bernstein 2010;O'Laughlin 2001;Razavi 2003;Tsikata 2009Tsikata , 2015. This discussion falls into two broad views about social relations. The first is represented by Sara Berry's study of the social dynamics of agrarian change in sub-Saharan Africa which focuses on Ghana, Nigeria, Kenya and Zambia (Berry 1993). Her analysis highlights the fluidity and dynamism of social and political relations in agrarian systems. She critiques the tendency to see the inexactitudes and open-endedness of customary arrangements as pathological, arguing that continuous negotiations and transactions constitute a strategy to retain influence and control and guarantee long-term benefits from reciprocal relations (Berry 1993). She also acknowledges, however, that the struggles for resources that are embedded in processes of agrarian change result in social differentiation in class, gender and kinship relations. Peters (2004) critiques this seemingly benign view of agrarian social relations for not taking account of the limits of negotiability and widespread processes of exclusion and deepening inequalities that occur as a result. Inequality, social differentiation and exclusion have been some of the visible outcomes of commoditisation, structural adjustment, market liberalisation and globalisation. She therefore advocates a theoretical shift -'away from privileging contingency, flexibility and negotiability that willy-nilly ends by suggesting an open field, to one able to identify those situations and processes … that limit or end negotiations and flexibility for certain social groups and categories' (Peters 2004, 269).
The recent literature on land grabbing raises similar questions about social relations, and processes of differentiation. These include the extent of dispossession and, within 'communities', questions about who has been most adversely affected, who has gained as a result of capitalist penetration into agrarian systems, the degree of social differentiation that results (Moyo, Jha, and Yeros 2013) and the extent to which commercial farming creates jobs or results in local labour redundancy (Li 2011). A particular focus has included the role of the African state and dominant social groups such as owners of capital, bureaucrats, chiefs and lineage heads (Amanor 2008;Lund 2008). Analysis has also focused on changes in customary law, and the erosion of communal property systems and kinship relations, as well as emerging class, gender and inter-generational inequalities (Daley and Pallas 2014;Doss, Summerfield, and Tsikata 2014;Tsikata and Yaro 2014;Verma 2014).
This JPS Forum joins these debates by examining processes of change, patterns of accumulation, investment dynamics, and expansion and contraction in sizes of farms, under three 'models' of agricultural commercialisation: plantation or estate farming; commercial farming areas; and outgrower and contract farming. The Forum also discusses how the trajectories of agricultural growth are implicated in the changing social relations in the countryside, drawing particular attention to gender and inter-generational relations. On the basis of nine case studies across three countries with different histories of agricultural commercialisation -Ghana, Kenya and Zambiathe Forum presents empirical answers to the question: What difference do the models make to the outcomes of commercialisation and to agrarian change? We analyse such differences in relation to four themes: land, labour, livelihoods and local economic linkages. For each, we ask how the outcomes of a particular form of land investment differ for different groups of people, notably by gender and age. While the papers in this Forum compare findings across the three models within each country, this introduction provides a broader perspective, presenting an overarching analysis of the dynamics both among the plantations/estates, outgrower schemes and commercial farming areas, and between the models, across the three countries.
There is a long history of attempts to encourage commercial agriculture across much of sub-Saharan Africa. But commercial farming has taken different formsvarying in scale as well as in institutional arrangements and labour regimeswith fluctuating political significance in different places and times. Past efforts in the colonial period have included the establishment of private estates and contract farming (Bates 1981;Heyer, Roberts, and Williams 1981), and in the period of immediate post-independence developmentalism also state-owned and -managed estates. Outgrower arrangements, too, are a model that has been encouraged across diverse cropscocoa, cotton, tobacco, sugarcane, coffee and teaas a means of integrating smallholder family farmers into commercial, and often transnational, value chains (Glover 1984;Oya 2012;Baglioni 2015). Liberalisation policies from the 1980s aimed to attract foreign direct investment into agriculture and other economic sectors. While in general, investment in farming remained low, there were some notable exceptions, typically in high-value export cropssuch as Kenya's expanding horticulture and floriculture sector (Dolan and Humphrey 2000) rather than in the production of staple crops.
Growing investor interest in farmland within the past decade has prompted the revival of large-scale commercial agriculture across many regions of Africa (Deininger and Byerlee 2011;Baglioni and Gibbon 2013), with resonances to past colonial and developmentalist schemes . In some cases, new farmland investments have been established on the same sites as defunct state farms and colonial estates (cf. Sulle and Smalley 2015;Paradza and Sulle 2015, among others), while in other cases large farms have been created in regions where they had not previously existed, such as in Northern Ghana (Tsikata and Yaro 2015). Associated with the growth of farmland deals and commercial farming in Africa is a policy and investor narrative concerning the 'win-win' possibilities arising from the convergence of investor capital with available land and abundant labour (FAO 2009). Whether there really is extensive available, unutilised land has been widely questioned, and many attempts to establish new estates have foundered when informal and customary rights of occupancy have been exercised (Edelman et al. 2015). Matching capital, land and labour through contract farming arrangements, with or without a core estate, has often been promoted, and a range of models have been proposed (Cotula and Leonard 2010). Yet the benefits of 'living under contract' have also been challenged (Little and Watts 1994).
In addition to the plantation or estate, and outgrower and contract farming models that have been widely discussed in the literature over decades, there is a third form of commercial agriculture worthy of attention. Small-and medium-scale commercial farming by independent farmers has been an important part of the African landscape since the colonial period. While these farmers have not been discussed much in the recent literature on commercial farming, there is evidence of their growing numbers and importance across Africa. Such farmers typically invest in agriculture with off-farm sources of income, and may acquire land through an endogenous process of agrarian differentiation, or through stateled schemes to establish commercial farming areas. Such 'middle farmers', as a new capitalist class of farmers, have considerable access to land in many countries in Africa, and some studies suggest that they may cumulatively be as or more significant in altering agrarian structure and spurring patterns of land concentration and accumulation as the expansion of estate farming (Jayne et al. 2015).
With some exceptions, little research and analysis is available on the implications and outcomes of different models of agricultural commercialisation (Cotula et al. 2015;Smalley 2013). Specifically, there is a gap in understanding the interactions between forms of commercial agriculture and land and labour relations, changing livelihood profiles and dynamics in the local economy surrounding these. With rapidly changing agrarian contexts across Africa, the large-versus-small debate (cf. World Bank 2009;Collier and Dercon 2014;Wiggins, Kirsten, and Llambi 2010) misses important dynamics, as differentiation happens simultaneously across different scales, with diverse factors impinging on outcomes. Changes are effected not just through corporate 'land grabs', but also significant demographic shifts, urbanisation and land acquisitions by local elites (Jayne, Chamberlin, and Headey 2014). All these processes are driving land concentration, facilitated often by inward investment from non-farm sectors. The outcomes for land access, labour relations, livelihoods and local economies are highly variable. As the papers in this Forum show, these outcomes are structured by the interactions among commercial farming models (of different kinds and scales), local farming sectors (already differentiated) and the rural non-farm economy. While several studies have addressed the relative efficiency of different scales and forms of commercial farming, in terms of land and labour productivity (Govereh and Jayne 2003;Lipton 2012;Dzanku 2015), our attention is on dynamics of agrarian change rather than factor productivity.
In this Forum, we ask who benefits and who loses out from agricultural commercialisation in Africa, how do agrarian transitions happen, what processes of accumulation and dispossession are generated, and where potentially do livelihoods get secured and for whom? How does each of these models articulate the relationships among land, labour and capital in commercial enterprises themselves and in the surrounding locality? Such questions need to be examined comparatively across the types of commercialisation, as the logic of capital operates in different ways depending on the form of the business model. This in turn poses questions about how people are able to compose their livelihoods as different processes of commercialisation unfold, and how such processes affect social differentiation in relation to class, gender and generation.
In what follows, we discuss the three models of commercial agriculture, and introduce our comparative research design, which involves three countries, three models and nine case studies, our qualitative methodologies and the design, sampling and analytical methods used in the household survey administered across our nine case study areas.

Three 'models' of commercial agriculture
Agricultural commercialisation involves diverse institutional arrangements, each with varied associated labour regimes and implications for land tenure, land concentration and for agrarian structures (Oya 2012). As already highlighted, following Smalley (2013), our study defines three broad models of commercial agriculture. In our work we were interested to find out how each performed against our four intersecting outcome themes, and whether the stereotypes often presented in wider policy commentary stood up to empirical scrutiny in our case study settings. Thus we examined, for example, whether estates/plantations result in mass dispossession and act as isolated enclaves with limited spillover benefits or, by contrast, whether they are the source of stable, relatively well-paid employment, including for women. We equally explored whether contract farming is indeed a 'win-win' solution, connecting capitalist production with smallholders and preventing land alienation or, by contrast, is the source of exploitative, risky contract relations that impede accumulation among smallholders. And, finally, we investigated the dynamic of medium-scale farming, and its impacts on land, labour, livelihoods and linkages, to assess whether this indeed demonstrates a new productive, inclusive and growth-generating pathway to sustainable commercialisation of agriculture. We now turn to characterising the three 'models', and their variants.
Plantations and estates have a long history in Africa, starting with colonial concession areas, and have been widely held as effective in providing substantial scales of wage employment, often combining permanent and casual labour. At the same time, because of the scale of land required for accumulation of capital in such farming systems, plantations/estates frequently displace local people. There are important variations, including whether or not there is on-farm processing, and production for domestic or export markets (Kydd and Christiansen 1982;Loewenson 1992). Plantations/estates may involve outright takeover of land and related resources, displacing other land users and uses, and there is some evidence that they are typically poorly integrated into their surrounding society and economy. One of the core reasons cited as to why plantations/ estates have relatively limited interaction with the local economy is that their value chains are often embedded in global markets. Ferguson (2006) has articulated a critique of plantations as constituting 'enclave economies' that source inputs (including migrant labour) from far afield and sell into foreign or national (rather than local) markets. This suggests that, while plantations/estates may achieve improved levels of productivity and output, weak forward and backward linkages into the local economy undermine their ability to contribute to dynamic rural economies and improved and diversified livelihoods in the surrounding area. We define plantations/estates as large, self-contained agribusiness farms that are vertically integrated into value chains. While they are often associated with one major crop, this is not always the case.
Outgrowing involves the development of processing facilities (usually but not always with a core commercially operated estate), through which small producers are incorporated into commercial value chains (Porter and Phillips Howard 1997). In this way, contract farming links capital to outgrowers via contracts, providing opportunities for accumulation by 'smallholders', but on terms of incorporation that may be more or less advantageous, depending on the institutional arrangements (Little and Watts 1994;Oya 2012Oya , 2013. Outgrowers are linked to a processing firm that sometimes also holds a nucleus estate, but augments its supply from these outgrowers who supply their produce. Such arrangements therefore create opportunities for wage employment not only in the processing mills, but also permanent and seasonal work on the nucleus estates. The terms of outgrowing contracts are highly diverse (Smalley 2013), generating different distributions of risk and benefit, and shaping the contours of new class relations in rural areas (Oya 2013). There are 'tight' contracts in which outgrowers are bound for future harvests, and 'loose' contracts in which outgrowers move in and out of supply agreements, and can choose to sell into local markets, depending on prevailing prices and demand from estates. Outgrowers are generally smallholders using their own land and labour for production, but with a commercial relationship for output marketing and often also input supply. In some cases, as in our case from Zambia (Matenga and Hichaambwa 2017, this volume), outgrowers may enter into a contract that cedes their land to an estate in return for receiving a dividend as land ownersa form of 'shareholder' outgrowing. In all this diversity, the common defining feature is not necessarily production by smallholders but the use of smallholders' land for contracted production.
Commercial farming areas in Africa feature a clustering of medium-scale commercial farmers who accumulate land via rental or sale, have contiguous or nearly contiguous landholdings in the same vicinity, and often specialise in the same crop or commodity. In some instances, these emerge through state planning, and the setting aside of blocks of land for commercial farming, as in our Zambia case (Sjaastad, Kalinda, and Maimbo 2010) and also elsewhere in Mozambique (Hammar 2010) and Nigeria (Ariyo and Mortimore 2015). In such instances, investments in medium-to large-scale farming require land consolidation and displacement, while creating various opportunities for on-farm employment and thus instigating new dynamics of social differentiation. In contrast, other commercial farming areas emerge through internal differentiation and marketisation, as well as purchase by outsiders, as we found in Kenya and Ghana. As depicted by Jayne et al. (2015), the new 'middle farmers' in these cases are mostly male, wealthy, middle-aged or retired, often from professional positions, including civil servants. We define commercial farming areas as localities dominated by medium-scale farms that are generally larger than those in the surrounding area, and are owned by individuals or small companies. They are often associated with mixed farming operations, but this is not always the case (Smalley 2013).
3. Three countries, three models, nine cases: a comparative research design and methodology The study was conducted in three countries: Ghana (West Africa), Kenya (East Africa) and Zambia (Southern Africa). As explained in more depth in the three other papers in this Forum (Matenga and Hichaambwa 2017 this volume;Yaro et al. 2017, this volume;Hakizimana et al. 2017, this volume), these countries have contrasting historical experiences with land tenure institutions (statutory and customary), large-scale commercial farming and agricultural liberalisation (Berry 1993;Dancer and Tsikata 2015). While Kenya is a former settler economy with a significant core of commercial agriculture, with large-scale farming accounting for 30 percent of marketed agricultural produce, including tea, coffee, maize, wheat and livestock (Dolan and Humphrey 2000;Kirsten et al. 2012), Zambia has repeatedly struggled to establish such a sector (Klepper 1979), and has liberalised through privatisation of state farms and provision of state infrastructure in 'farm blocks' (GRZ 2005). Ghana, by contrast, has an established system of peasant-based commercial agriculture undergirded by customary land tenure systems and informal land markets (Amanor-Wilks 2009;Tsikata and Yaro 2015). In all three countries, a policy environment favouring the liberalisation of land markets and foreign investment facilitates large-scale land acquisitions.
Another difference among the three countries is the nature of their institutional arrangements concerning land. In Ghana approximately 78 percent of land is held under customary tenure, 20 percent is state controlled and two percent is under shared ownership (Deininger 2003). Registration of individual landholdings acquired under customary law was first introduced under the Land Registry Act 1962 (Act 122) and extended under the Land Title Registration Act of 1986 (PNDC Law 152). However, many people acquiring these lands did not take up land title registration for a variety of reasons, among which were lack of access to land registries, cumbersome and expensive procedures and institutional inefficiencies. Zambia has administered a dual land tenure system of state land and customary land, with the possibility to convert customary land into state land, which is seen as attractive to investors, since 1995. There are no precise figures on the proportion of land held under customary and state tenure. Official figures from 1964 put state land at six percent; however, it is estimated to be as much as 10-20 percent today (Nolte 2012).
In Kenya and Zambia, the state instituted a regime of state expropriation of land through legislative reforms. The Zambian state declared Crown Lands in the colonial period and undertook extensive titling and registration of land in the 1960s and 1970s, respectively, following independence (Anthony and Uchendu 1970, 223;Adams 2003;Thurston 1987). In Kenya, post-independence land reforms, particularly in highland areas, resulted in the transfer of land to smallholders through settlement schemes (Leo 1981;Haugerud 1989). By contrast, in Ghana, there were no significant land law reforms until the World Bank-led Land Administration Project (LAP) from 2003, which sought to harmonise land policies and legislation with customary law (Tsikata and Yaro 2014).
Case study choices were informed not only by the type of institutional arrangement or 'model', but also different kinds of crop production, including production of staple grain crops, horticulture and feedstock for biofuels. Within each country, the findings compare outcomes across the three models (see Table 1), exploring which models of commercialisation are potentially more successful, across different indicators. Such an approach enabled us to understand the implications of the different commercial agriculture models across sectors, in countries with different agrarian structures and histories. Overall, the studies provide a series of cases, rather than a systematic comparative assessment; but they nevertheless offer some important insights, and challenge some of the simplistic narratives of agricultural commercialisation in Africa.
We provide below a brief summary of the nine case studies (see also Figure 1). Plantations and estates: In Ghana, our plantation/estate case is the 4500-ha Norpalm oil palm plantation, a former state-owned company, now partially privatised, which does onfarm processing for the production of palm oil for the domestic and export market, while the other plantation/estate cases are more diversified estates that produce for both domestic and export markets. In Kenya, we studied Kisima farm, a 6000-ha established estate owned by a fourth-generation Kenyan family of British origin, which processes flour and canola oil onsite, and has expanded from field crops into floriculture. In southern Zambia, the 10,000-ha Zambeef Chiawa estate produces a range of grainswheat, maize and soybeanprimarily for the domestic market, but does no on-farm processing. All are established plantations/ estates, but which have expanded operations, either geographically or into new crops, and vertically integrated their operations, within the past 10 years.
Contract farming: Our contract farming cases are, in Ghana, the fruit producers who supply passion fruit, papaya, pineapples and mangoes to Blue Skies company on contract, which in turn exports cut fruit to up-market European supermarkets; the French bean farmers who supply two Kenyan companies, Frigoken and Finlays, with their produce, for packaging and export to Europe; and the Magobbo outgrower scheme where farmers have ceded their land to form a single sugarcane estate, operated by a company contracted to Zambia Sugar. The latter model of consolidating outgrowers' landholdings is a growing trend in Southern Africa, and is effectively a lease-out model, where the block is farmed as a single unit by a management company, with outgrowers as shareholder-outgrowers who obtain variable dividends for their land which vary with output and prices, on the basis of original landholdings. Commercial farming areas: Two of our cases focus on areas where medium-scale commercial farmers have emerged, acquiring and consolidating landholdings and focusing on the same crop that dominates a given locality: mango farmers in the Somanya area of Ghana, and coffee farmers in Meru County in Kenya. The mango farmers emerged as a medium-scale sector during the 1990s. Together they operate on an estimated 5000 ha in this area, and have an association through which members acquire accreditation for export to the EU market. The medium-scale coffee farmers in Kenya have also emerged as a distinct sector since the 1990s, with diverse scales of coffee production, sometimes combined with agroforestry and other production, depending on the availability of offfarm incomes to pay for hired labour. In Zambia, by contrast, we focused on the state- designed commercial 'farm block' at Mkushi, together constituting a 176,000-ha area, where commercial farmers own and operate medium-to large-scale commercial farms.
In view of the inherent limitations of qualitative and quantitative approaches used alone (Tashakkori and Teddlie 2010; Teye 2012), we employed a sequential mixed-methods design for data collection and analysis. Our study was divided into three phases. In the first phase, we conducted in-depth qualitative studies on the origins, scale of operation, labour regimes, value chains and wider politics of the commercial farming cases, via indepth interviews, archival research and focus group discussions.
Based on insights gained from the qualitative results in the first phase, a structured questionnaire was designed and used to conduct a household survey of about 300 respondents selected from three case study areas in each of the three countries. We drew a random sample of the entire area, including the case study sites, within a 5-km radius from an identified centre point (N = approximately 100 households per case study). In each selected household, one representative completed the questionnaire. The quantitative data collected in the second phase was analysed using STATA. While our sampling was random, in analysing our data we distinguished between those households 'involved' as farmers, outgrowers or workers, and those households 'not involved' in our case studies of commercial agriculture, but living in the vicinity. Also during the quantitative data analysis stage, we constructed a wealth index based on responses provided on assets during the survey. Drawing on methods developed by McKenzie (2003) and Filmer and Prichett (2001), a composite wealth index was computed based on the value of land, livestock and other assets owned by the sampled households in the focal areas. This enabled us to cross-tabulate our findings against three wealth categories: rich, middle and poor households.
A third and final phase of field research involved in-depth life histories of rich, middle and poor households in each study site, including those 'involved' in various ways as well as those 'not involved', in order to locate our analysis in the context of longer historical shifts in livelihood strategies, and to explore intra-household relations and dynamics. The qualitative research also included in this final phase a process of mapping input and output markets, in order to identify the forms and extent of linkages between our case studies and the local economy. While we used a survey instrument, our analysis is not primarily driven by statistical analysis, but rather by our qualitative research, with illustrations from our survey data. Our research strategy resonates with sequential exploratory mixedmethods designs whereby emphasis is placed on qualitative data, but quantitative data is used to support qualitative analyses (see Creswell and Plano Clark 2011).
In the following section we look at the outcomes of agricultural commericalisation across our country case studies in relation to our four themesland, labour, livelihoods and local economic linkages. Within each theme we examine assumptions in mainstream narratives about commercial agricultural models, and assess these against the empirical evidence of our case studies. As we conclude, the actual outcomes across the cases are much more complex than the standard narratives suggest, and are affected by history, location and political economy, as well as the particular crop and company involved in the case. Nevertheless, as we show in the sections that follow, the study reveals some important patterns in the dynamics of commercialisation of agriculture in Africa.

Outcomes of agricultural commercialisation: land, labour, livelihoods and linkages
Agrarian change occurs through diverse processes, and its outcomes are contingent and require disaggregation. Here, we provide a synthesis of the key outcomes we observed from our case studies, distinguishing between outcomes for land and land relations including access, markets and tenure; labour including forms of waged and unpaid labour; livelihood composition, resilience and diversification; and the extent and nature of the linkages between commercial agriculture and the local farm and non-farm economy.

Land
In the literature on 'land grabbing', a pattern of land expropriation for large-scale plantation or estate agriculture is often described. A plantation or estate needs large areas of land, requiring conversion of customary or crown or state land to privately titled land or land under long-term lease, resulting in the removal of former residents. By contrast, contract farming does not require such land transfers, which is a reason for the widespread advocacy of contract farming as a 'win-win' means to securing investment while minimising land dispossession (cf. von Braun and Meinzen-Dick 2009;FAO 2009). The emergence of medium-scale commercial farming has received less attention, but this class of 'investorfarmers' cumulatively may result in significant land alienation through processes of land purchase, leasing and consolidation (Jayne et al. 2015).
But large-scale dispossession may be only one dimension of changing land relations as a result of agricultural commercialisation. For example, as seen in Tanzania's sugar industry, contract farming may involve the displacement of food crops (often controlled and farmed primarily by women) (Sulle and Smalley 2015). As medium-scale commercial farming expands, informal land markets may evolve (Chimhowu and Woodhouse 2006), with elites accumulating land and consolidating plots into larger farms. Such allocations may also happen through state action, including via land reform, as areas are allocated to medium-scale 'emergent' commercial farmers. The result is a complex process of both differentiation and dispossession, negotiated through the intersection of state and traditional authorities, which differ across contexts (Berry 1993;Peters and Kambewa 2007;Lund and Boone 2013). What, then, were the outcomes found across models in our three country cases?
In Ghana, in the area surrounding the commercial mango farming area, people experienced the most severely constrained access to land, due to the land consolidation involved in the establishment of the medium-scale commercial farms that dominate the area. In the less-populated area surrounding the long-established Norpalm plantation, access to land was less constrained, while the highest proportion of survey respondents who reported that they had access to land were in the outgrower area, as they retained their land and there was only a small core estate linked to the Blue Skies company.
In Zambia, by contrast, there is increasing land scarcity reported across all three cases, except those living around the Zambeef Chiawa estate. Land concentration around the commercial farming area at Mkushi was exacerbated by the emergence of further 'satellite' commercial farmers, farming on a medium scale in the areas adjacent to the larger commercial farms. Land conflicts were highest around the Magobbo sugar outgrower scheme. Since the scheme put the outgrowers' land under the management of the core estate, and many outsiders took advantage of the shareholder arrangement, this imposed land pressures on the surrounding areas. However, possibilities for people to expand their landholdings in the scheme are limited, as smallholders were each allowed to register only a predetermined plot size of 4-6 ha. Here, class differentiation is evident between those farmers who had ceded their land to the outgrower block, on the one hand, and those employed to work on it, with the former using their fairly substantial incomes generated from the scheme to invest in education and enterprises in the non-farm economy, while in parallel a landless class of (often young, male) labour has emerged.
In Kenya, the prior distribution of landholdings as part of post-independence land reform had a substantial impact on who was involved in commercial farming, as both farmers and workers. Immediately around the Kisima estate, land was settled as part of the post-independence Million Acre resettlement scheme (Leo 1981). This area has relatively equitable land access, meaning that smallholder farmers are able to sustain themselves from their farming, often together with other small enterprises. However, the younger generation and women without land seek work on the Kisima farm to supplement incomes. In the commercial farming and outgrower areas, land reforms involved land consolidation to facilitate cash crop production (specifically coffee and tea), with the aim of promoting an emerging 'yeoman' agrarian class (Leo 1981). Here, larger landholdings combine with higher levels of landlessness. Small plots of land are thus used, primarily by women, to engage in contracted production of French beans and other horticultural products, while men focus on staple crop production on other land parcels. Livelihoods in these areas combine agriculture with off-farm and remittance income. Alongside, more commercial producers have larger plots, and focus on coffee production, employing labour from nearby areas.
Land markets are evident in all cases. The growth of such marketsboth formal and informalcorrelates with changes in access, as some are excluded and others accumulate. Across all three countries, land purchases are infrequent in the smallholder areas where contract farming occurs. Across the countries, 95 percent of outgrower households have acquired land via inheritance, rather than through market sales. Short-term and informal rental markets are, however, vibrant, enabling outgrowers to move in and out of production, a pattern most notable in the French bean industry in Kenya. Such flexibility is not available, however, to the outgrowers in Zambia, whose land is locked into a block scheme. Increasing commodification of land occurs through other investments in addition to agriculture, for example around the Zambeef Chiawa estate in Zambia, where tourist operations and agricultural investors are driving an increased marketisation of land. Similarly, in Kenya and Ghana, land sales are on the rise among the commercial coffee and mango farmers, as the successful farmers seek to consolidate land holdings, and accumulate land for expanding production.
Reflecting this process of land commodification and growing land markets, land prices are perceived to be on the rise across our study sites, and this is not only due to farmland investments, but also the result of demographic pressures. In Ghana, around half of those surveyed across our three sites consider land prices to be increasing; this was highest in the plantation area, followed by the commercial farming area. High land prices are also prompting migration, as those displaced by expanding commercial farms move away from the area. In Zambia, in contrast, land prices were on the rise mostly in the outgrower and commercial farming areasbut this was less evident in the area surrounding the Zambeef estate, which is in an extensive dryland region under customary tenure. In Kenya, as in Ghana, land prices are rising dramatically in response to population growth and demand for land, including a process of consolidation of landholdings as retired professionals and others with capital to invest are buying up smaller parcels. There is also a growing conversion of farmland to non-agricultural purposes, particularly for the expansion of settlements, light industry and trading centres in these sites.
As we found across our cases, the standard narratives about different commercialisation models do not necessarily hold up. In our cases, the simple 'win-win' narrative around contract farming has to be qualified, given differential access and the intra-household dynamics observed. Different outcomes are evident across our cases. The plantations or estates in Ghana and Zambia have not brought about significant land dispossession in recent years, as they were established many years ago in areas with low population density, where agricultural potential was low or only enhanced through significant irrigation investment; in the Kenya case, the estate sat alongside settlement schemes where land had been redistributed earlier. In our cases, the emergence of medium-scale commercial farming within or near existing smallholder areas was where the greatest competition over land occurred. This dynamic of both local-level accumulation 'from below', but also significant investment from outside, is creating a new dynamic of land concentration and accumulation through agricultural commercialisation that, while not as dramatic as the extensive 'land grabs' that are also occurring, may have wider, deeper and more sustained impacts on land access and relations.

Labour
The substantial literature on African commercial agriculture points to the existence of diverse labour regimes, ranging from traditional, paternalistic 'domestic government' arrangements, typical of the settler economies, to much more marketised, industrial concerns, linked to larger agribusiness companies (Gibbon and Riisgaard 2014). While the latter may be linked into global markets and standards, and have 'corporate social responsibility' (CSR) requirements for improving labour conditions, older forms of large-scale family farming are highly reliant on an individual farmer for the provision of services and support, and conditions vary massively. Across all forms of commercial agriculture there has been a trend towards less permanent salaried work, except in management positions, and towards a greater reliance on casual, temporary work (Gibbon 2011). This has often been accompanied by a process of 'feminisation', as women are employed for particular tasks (Dolan and Humphrey 2000). This in turn is associated with a wider process of change in agrarian societies, as smallholder agriculture comes under pressure, generating a class of 'footloose', fragmented labour seeking their livelihoods through fragile piecework, combined with small-scale, subsistence agriculture (Bernstein 2010).
Employment outcomes across our cases were highly variable. The plantations or estates provided permanent, salaried employment in all cases, with wages sometimes being significantly higher than standard incomes in the areas. But such jobs were largely taken by highly qualified outsiders, and so did not improve opportunities locally. Temporary jobs were equally relatively better paid than local wage levels, and sometimes significantly so, when skilled labour was hired for agro-processing (such as at the Blue Skies plant in Ghana), though Kisima farm in Kenya paid slightly lower wages to casual workers than those paid by surrounding medium-scale commercial farmers. In these cases, employment opportunities opened up for women, and many women we interviewed commented on how a flexible employment arrangement allowed them to continue with farming at home, and engaging in reproductive labour (see for instance Hakizimana et al. 2017, this volume).
Youths also sought employment in the plantations and estates, as well as on mediumscale commercial farms, as an alternative or complement to smallholder farming. With land constraints increasing in some sites (e.g. near the Mkushi block in Zambia, and in the commercial farming areas of Ghana), and where extensive dryland farming is highly risky (as in the Zambeef Chiawa area of Zambia), wage employment, even if temporary and poorly paid, was attractive. In the Maggobo outgrower-shareholder scheme in Zambia, youths were largely excluded from the benefits, which were captured by local elites and influential people from nearby towns (Matenga and Hichaambwa 2017, this volume;Rocca 2014). Without access to land or dividends from a plot on the sugar block, they sought employment on the estate. Those in families with shareholder plots were given preferential access to employment, locking out some of the landless population, and making the institutional and political dynamics of labour access in this scheme especially complex.
With employment opportunities nearbyrather than involving long-distance migration to distant towns or minespeople may be able to combine wage employment with cashcrop production. This was the case among women in Kenya who involved in the production of French beans as outgrowers. Wage employment on nearby farms allowed for land leasing and expanding French bean production, and so allowed a flexibility and injection of cash into the farming enterprise. Straddling livelihood options by linking wage employment with farming was most possible in Kenya where commercial farms (both the large Kisima estate and commercial coffee farms) were in close proximity to smallholder areas. Thus, the geographic location of farms of different scales with different labour regimes in relation to each other has implications for employment, as for stronger local economic linkages (see below).
Across all the cases, there is an increasing monetisation of family labour. This is evident especially among the commercial mango farmers in Somanya, Ghana, to a lesser degree also among the contract farmers in all countries, among the commercial coffee farmers in Meru, Kenya, and the new 'satellite' commercial farmers around Mkushi block in Zambia. Paying family members (often adult sons) to work on a farm was due to this generation's higher education levels and the opportunity costs of working on a family farm rather than seeking employment elsewhere. The trend has clear gender and generational implications, with women often being excluded from farming operations (or moving into trading) as commercial farms become more established, and grown-up children, usually sons, being paid to remain or return from opportunities elsewhere to work, often to manage, the commercial mango and coffee farms (in Ghana and Kenya, respectively) of their parents or, more usually, fathers.
Thus, across the cases, there are highly diverse labour regimes, each with gender and generational consequences. The patterns do not conform to standard narratives of quality wage employment on estates providing a dynamic of proletarianisation, as wage labour becomes increasingly attractive compared to 'subsistence' agriculture (although elements of this trend are present). Instead, there is much more 'straddling' (Cowen 1981), in which people combine wage employment with small-scale farming, including on contract. The small-scale sector in our cases is far from reflecting simple 'subsistence', and is often highly commercialised itself, involving substantial wage employment, even among family members. These areas have diverse classes of labour with different relationships between petty commodity production and labour within and across households. Rather than seeing this only as a pattern of 'fragmented' classes of labour (Bernstein 2010) who struggle to make a living, for some this provides a more flexible option. Some were hiring in labour while selling their own, spreading risks, seeking opportunities in different markets, and diversifying livelihoods in ways that combine with reproductive labour. This can be advantageous, especially for women. However, there are on-going exclusions. Youths, both male and female, are increasingly finding it difficult to establish themselves, and must seek oftenprecarious employment opportunities, without the benefits of having a farm to fall back on. In some institutional arrangements for commercial agriculturenotably the shareholder outgrower set-up in Magobbo, Zambiathere are systematic exclusions in employment opportunity, creating a new agrarian class dynamic in the area.

Livelihoods
These different models of commercial agriculture provide a variety of livelihood options. This may allow diversification, which Dorward et al. (2009) characterise as 'stepping out' into wage employment or into the non-farm economy. It may also allow accumulation ('stepping up'), which may be from (re)investments from agriculture. We found a further trend, which was that of commercialisation driven by investments from outside of agriculture, such as retirement funds, remittances or on-going urban employmentwhat might be termed 'stepping in'. But there are also those who do not benefit from the opportunities of commercial agriculture, and are just 'hanging in', surviving off poorly paid wage labour, perhaps combined with small-scale agriculture. There are others, in turn, who 'drop out' in that they move out of farming entirely, and may also migrate elsewhere in search of other livelihoods. Young people in our case study areas often commented on the attraction of non-agricultural employment in towns and cities, for example. Each of these livelihood trajectories is evident across our case studies, but how are they linked to different commercial models?
As already mentioned, contract farming is often framed as an 'inclusive' business model that offers the benefits of a diversified livelihood, and an efficient connection of smallholders to commercial agricultural value chains, without land dispossession. In our cases, we observe a wide spectrum of contracts, ranging from tight contracts that bind landholders into long-term agreements with monopsony companies (as in Ghana), to loose arrangements that allow producers to move in and out of short-term agreements, without any sanctions for 'side-selling' (as in Kenya). There are also differences among outgrowing arrangements in which all the outgrowers' land is dedicated to a single cash crop grown on contract (as in Ghana and Zambia), and more diversified farming practices in which outgrowers retain diverse crops for consumption and sale, alongside contracted crops (as in Kenya). Outgrowing also does not always involve smallholders cultivating their own land and using their own labour. While in Ghana and Kenya, many households hire labour, with some simultaneously hiring in labour and engaging in wage work elsewhere, in Zambia households are not directly involved in production on their own land, except where individuals within the household are hired by the management company to work on the block farm. While producers enter into individual contracts in Ghana and Kenya, in the Zambian case, the 'contracts' are in fact shareholding agreements to facilitate large-scale cultivation rather than individual holdings.
For these reasons, the outcomes for livelihoods from contract farming are diverse. There are stark patterns of differentiation clearly evident in the Ghana and Zambia outgrower cases, while this is not evident in Kenya where relatively small holdings form the base for outgrowers' production and diversified livelihoods. Standard policy narratives present contract farming as a route towards accumulation for smallholder farmers by connecting them with potentially lucrative global markets (FAO 2009), but we find that this is not necessarily the case. The Kenyan outgrowers we studied are poorer than other farmers in the area, as measured by the value of key assets including land and livestock, among others (see Hakizimana et al. 2017, this volume). Those involved in outgrowing in Kenya are mostly women working small plots as part of a family farm. They are not able to accumulate capital to expand their operations, and those who are wealthier prefer not to engage in outgrowing arrangements. By contrast, in Zambia those with 'outgrower' shareholder contracts in the sugar block are considerably better off than their counterparts in the surrounding areas without such access, as the dividends they are paid far exceed returns from dryland farming. In Ghana too, outgrowers linked to Blue Skies gain considerable incomes relative to others in the same area.
Thus, across the cases, it depends who is doing the outgrowing in what agrarian context, and what the nature of the contract is. In Kenya, those women engaged in the 'loose' outgrowing arrangements can diversify their income-earning activities, and gain independent sources of income. In Ghana, access to outgrowing opportunities is important for accumulation, for men as well as women. Women additionally can diversify their livelihoods through access to wage employment at the Blue Skies processing plant. In Zambia, the shareholder dividend-based income is important for those who have access, but it results in both capture by male household heads and an increasing differentiation within the community around the outgrower block. In sum, there is such a variety of forms of contract farming that a simple narrative does not necessarily apply.
How do the other forms of commercialisation affect livelihoods of different groups? Outside the relatively well-paid managerial class, those employed full time on largescale plantations and estates are generally unable to diversify their livelihoodsespecially where there is land scarcity in the surrounding area. Links with homes elsewhere may be important via remittances, although wage rates tend to be low and remittances limited. As we discussed above, plantation or estate farming offers temporary, casual employment opportunities for some, and this may be relatively well paid for more skilled work. Such wage employment can be linked to farm-based production, and other livelihoods, as part of straddling strategies. For women in particular, this is especially key as income from wage labour can complement social reproductive obligations, and provide a source of independence and empowerment. Livelihood impacts are therefore shown to be influenced by locational factors, with the ability to move between wage employment on larger commercial farms and smallholder production proving to be crucial for some, as illustrated in particular in Kenya, and to a lesser degree also in Ghana and Zambia.
The creation of medium-scale commercial farms generates another dynamic. While such farms may generate employment, the quality of employment is relatively low. Such farms often re-create an old-style 'domestic government' style of commercial farming (Rutherford 2001), based on patriarchal control and now also with paid family labour, with poor conditions offered to other wage labourers including migrants living on the farm. However, medium-scale commercial agriculture can provide a pathway to accumulationsupported from outside, but generating a dynamic 'from below', as profits are made through specialised production linked to high value often export markets in, for example, coffee (as in Kenya), mangoes (as in Ghana) or grain crops (as in Zambia). Such accumulation, as noted earlier, may result in acquisition of land, employment of labour, and further investment in farm activity, including processing.
Such a pathway of commercial agriculture has often been envisaged in Africa, with repeated attempts to cultivate a 'yeoman' or 'emergent' farmer class, as in Kenya's Swynnerton Plan, and often with little success. In our cases we observe this occurring today, in the context of increasingly marketised production systems, where the advantages of lowcost and low-risk specialised commercial enterprises are seen. These by and large benefit male elites, either from the local area or those moving into farming from urban areas, where employment in industry or the civil service, for example, is increasingly fragile given repeated economic crises and the effects of structural reform. Except in some few cases, usually on the death of a husband, women were not benefitting from such new forms of medium-scale agriculture in our study areas. They may seek other options, linked to trading or agro-processing as an alternative source of income. Questions of inheritance and the willingness of a next generation to take over are also evident, often resulting in disputes over what happens to a farm, when parents become old or pass away.
How then does the articulation of agribusiness and other forms of capital investment into agriculture affect people's livelihood portfolios and trajectories? We find that these differ across cases quite dramatically, with different consequences for livelihood options, income levels and food security. Again, the standard narratives did not always fit. Contract farming may for some be a useful route to linking to external investment and gaining markets, but not always, and may be best when combined with other options to offset contract risks. Wage employment on plantations or estates may bring benefits for some, again especially if combined with other livelihood alternatives, and for those with few alternatives, such as the very poor, including the landless younger generation. Medium-scale commercial farming is a route to accumulation for those with start-up resources, and the ability to acquire land and invest, but the types of livelihoods it generates for workers may be highly precarious. In other words, an assessment of the impacts of commercial agriculture must take a rounded picture, across genders, generations and classes, and so take a wider livelihoods perspective, seeing how people combine accumulation, reproduction, investment and so on across different activities and spaces (Scoones 2015).

Linkages
How commercial agriculture affects local economies through linkages and multiplier effects has been much debated (Haggblade, Hazell, and Reardon 2007;Hazell et al. 2007). Classically, plantations or estates have been found to have few local linkages, operating as 'enclaves' (Ferguson 2006). They may employ migrant labour from outside the area who remit funds home rather than spend locally, as their consumption and service needs are supplied on the estate. And such enterprises may be linked into vertical value chains that generate little benefit nearby through transport, processing and other value addition. By contrast, smallholder contract farms and medium-scale commercial farms are seen to be more embedded in local economies, employing locally, and resulting in consumption and other linkages that benefit local businesses (Cotula and Leonard 2010).
How do these characterisations bear out in our cases? In Zambia, the Zambeef Chiawa estate employs few people, and the workforce includes a substantial number of migrants from outside the area. As an increasingly mechanised operation, the levels of employment and degree of integration with the local economy are low. There is no local procurement or sale of produce into the local economy. In other words, Zambeef operates as an enclave, and so fits the standard narrative. However, this is not true of the other cases. In Kenya, the Kisima farm employs many people, mostly as permanent workers whose wages and conditions of employment are well above the legal minimum and better than those offered by other, surrounding employers (see above). Other than via wages, Kisima contributes to livelihoods in the area through fairly extensive CSR initiatives such as a primary and a secondary school built and sponsored by the company, a health centre, and a scheme to sell potato seed to smallholderswhich also provides an additional income stream for the company.
Unlike the Zambia case, the extent of wage employment at Kisima reflects the extensive on-farm processing undertaken, as well as its recent expansion into labour-intensive floriculture that creates numerous jobs for women, especially off-farm permanent jobs. At the same time, other than via employment and its CSR initiatives, Kisima in Kenya has few links into the local economy, with the company not procuring inputs locally, and, other than its potato seed scheme, only selling into national and global markets, mostly via distributors based in Nairobi. In contrast, Ghana's Norpalm plantation is much more integrated into the local economy. The plantation employs many people although wages are low, but it also procures fresh fruit bunches of oil palm from local smallholders and sells some palm oil into local markets, which competes with local processing and production.
Overall, then, whether or not a plantation or estate is an 'enclave' in the local economy depends on a number of factors. First, it makes a difference if there is on-site processing that creates jobs in addition to those in primary production; second, it depends on the nature of the crop and the seasonality in labour demand, and whether or not the crop is one that smallscale farmers already grow locally; and, third, it depends on whether low-cost and smallscale processing of the crop is feasible and so sales from the plantation or estate to local small-scale processors are feasible.
Among the outgrower cases, in Zambia it is difficult to attribute dynamics in the local economy to the outgrower scheme, especially since the block farming model mimics the plantation or estate model. However, the increased incomes for landholders, through de facto rental, provide important purchasing power in the local economy. Linkages with the local economy are strongest among the French bean outgrowers in Kenya and the Blue Skies outgrowers in Ghana, both of whom acquire some inputs from the contracting company, but source others from small local agro dealers. In both cases, though, processing is centrally done by the company, and the produce exported. The increased incomes observed among outgrowers in all cases bring local benefits, and we see in all our sites those with contracts purchasing additional consumer goods, as well as food from local markets. As discussed earlier, outgrowing in the smallholder settings has employment effects, although this is qualified by the institutional arrangement in Zambia, which follows more the plantation or estate pattern.
The medium-scale commercial farmers are well embedded in local economies and, because they are accumulating and employing, have positive linkage effects. This is especially the case among the coffee growers in Kenya and the mango growers in Ghana. Our studies suggest that it makes a big difference whether the commercial farmer resides on the farm or is an absentee farmer with a manager on-site. In Kenya, for instance, the farming operations of the commercial coffee farmers living in the communities were found to be more embedded locally than those of the absentee farmers, in terms of technology transfer, forms of collective action and mentoring of other farmers. At Mkushi in Zambia, the linkages into the local economy are not only from the commercial farmers within the block, but also from the 'satellite farms' that have emerged around it, where smaller scale commercial farmers have taken advantage of the productive infrastructure of the block and are starting to accumulate both via farming and through non-farm enterprises. Here, diverse linkage effects are observed, generating growth in the local economy, alongside differentiation.

Conclusion
Commercial agriculture has a long history in Africa, from the colonial era to the present. In recent years, the growth of large-scale investments in commercial agriculture has raised the profile of agricultural commercialisation, linked to debates about 'land grabbing'. This has seen the establishment and rehabilitation of large-scale plantations and estates, focusing on some key crops, including oil palm and sugarcane, among others. These have had mixed success, and many have failed. But, clearly, for certain crops in certain areas where land is extensive and water is available, large-scale agriculture can have benefits, as we have seen in some of our cases. Benefits include wage employment, including flexible temporary employment that can be combined with other sources of livelihood. To varying extents, however, such investments have not had wider impacts on the local economy, and have remained mostly 'enclave' investments, separated geographically, socially and economically. In each case, the establishment of a plantation or estate had led to displacement and dispossession of existing farming communities.
Large-scale plantation and estate agriculture, however, is not the only pathway of commercialisation. In this study we looked at two others, involving various forms of contracting and outgrowing, and different types of medium-scale commercial agriculture. Across our cases we see much variety in both the form and impact of these commercialisation models. A simple conclusion is impossible. Opportunities for livelihood improvement and accumulation certainly existed among some of the outgrower cases, especially if linked with other livelihood options, as we found in Ghana. In contrast, in our Kenya case, outgrowing provided a crucial source of income and autonomy for poor women but without prospects for any significant accumulation, which is why better off smallholders, especially men, chose not to engage in the scheme. Incorporating smallholders as shareholders in an estate is another route employed in the sugar estate in Zambiaalthough the term 'outgrower' is rather a misnomer in this variant of outgrowing. The scheme resulted in significant dividends for landholders (mostly men), but also patterns of exclusion for others, including women and the younger generation (Rocca 2014). The employment effects of contract farming were varied, depending on the crop and level of investment, but an increasing marketisation of wage labour was observed, though most jobs generated were temporary and with poor conditions. Among the medium-scale commercial farms, we again saw much variety. This type of farm is an important and growing phenomenon in certain areas, associated with particular crops. Compared to large-scale land investments, such new medium-scale farms have a different effect on agrarian dynamics, employment relations and livelihood consequences. Such farming is much more embedded in the local economy, with strong forward and backward linkages, and is reliant on local rather than foreign investment. However, the effects on land access and impacts on agrarian class relations may be negative, as such farms are located in well-populated areas, and act to squeeze out opportunities for others, creating a new rural class of farming elites, with varying relations to labour, alongside growing landlessness.
Across our nine casesthree models in three countrieswe have gained some insights into emerging patterns and trends, but these are only hints. The sample was small, and the variations across cases huge. However, important new questions are posed, and gaps identified. For example, the spatial relationship between different commercial models appears to be especially important, and the ability to 'straddle' wage labour and farming livelihoods vital for some, making the large versus small farm debate often redundant; the question is rather how different farming models co-exist and the ways in which poor people can navigate opportunities among them. The longer term impacts of medium-scale farming investmentsespecially across generations and between genderson agrarian structures and class dynamics needs to be a major area for future work, as such enterprises expand and their effects become evident. Clearly, the 'win-win' narrative on outgrower schemes needs serious qualifying, although the opportunities for linking agribusiness capital and smallholder farming have potential, if flexibility is retained in the form of contract and, unlike in the sugarcane 'shareholder-outgrower' model, elite capture is prevented.
Beyond the specificities of particular cases and models of agriculture commercialisation, our findings speak to the long-standing debates about the social relations of agrarian change in African agriculture. We draw inspiration from Berry (1993), not least in selecting three of the four countries she studied, and in attending to struggles over resources, especially land, and the ways in which these both are shaped by and constitutive of class and gender relations. Yet, reflecting on our varied findings across the nine case studies, we find dispossession and exclusion to be significant features of some of the sites, reflecting Peters' (2004) argument that larger structural transformations limit negotiability in land and other social relations and, via social differentiation, deepen inequality.
Going beyond the simplistic narratives on land grabbing and commercial agriculture is a vital first step towards more nuanced and sophisticated policy frameworks. This must take into account local specificities of history, geography, agroecology and agrarian relations, and seek routes to more sustainable and inclusive pathways of commercialisation. Our cases offer some useful hints, important cautions and questions for further research.