“Taken-left” dynamics? Rethink the livelihood changes of affected villagers in the era of the global land rush

When large-scale common land is taken from villagers by investors with little compensation, their labour unneeded, villagers’ livelihoods tend to be largely destroyed. This implies a tendency to focus on what has been taken from villagers during the land-based change, which has valid and has far-reaching social relevance. But as the rise of the industrial tree plantation (ITP) sector in Guangxi shows, some villagers are capable of having their livelihoods maintained and even expanded when big investors come and acquire massive amounts of land. This seems to be an anomaly at a first glance, considering what has been taken from villagers, but these unexpected and positive livelihood changes can be explained when one closely examines the dynamic of what is taken and what has been left to villagers. During this ITP boom, although large-scale collectively owned forestland is taken by investors and few work opportunities are created to incorporate those affected, villagers’ control over farmland plots and their access to off-farm work opportunities remain (including the land system and labour dynamics). This paper highlights the importance of analysing “taken-left” dynamics to more fully capture diverse livelihood changes.


Introduction
Currently, massive land grabs occur globally under the convergence of the contemporary food, fuel, and environmental crises (Borras Jr et al. 2011) and this trend has intensified given the multiple crises caused by the recent pandemic. In this work, land grabs are understood as "control grabbing", and are essentially an attempt to control land via varying control mechanisms and for desired profits (Borras et al. 2012;Ribot and Peluso 2003). These land grabs can sometimes be simply speculation (Tsing 2000), but on other occasions are related to the rapid expansion of the cultivation of cash crops (i.e. crop booms) and thus lead to massive changes in land use (Hall 2011a, b). Such changes in land use and land control tend to significantly affect local villagers' livelihoods.
Large-scale land deals, of crop, tend to lead to significant negative impacts. Because in the course of land acquisitions, those who previously accessed land, particularly the marginalized, who depend on land, tend to lose their vital, possibly their only, livelihood source. This has occurred in many regions of the world (Kenney-Lazar 2012; Hall et al. 2011;Hall 2011a, b), although a few scholars have acknowledged the differentiated distribution of the impacts of land (Xu 2019a;Yenneti and Day 2016). In some cases, the impacts of land dispossession on affected villagers might be partially alleviated if villagers are able to get sufficient compensation to explore alternative livelihoods (Zhan 2019) or are incorporated via either employment or contract farming (Barney 2004), although such incorporation may be less beneficial and even adverse for those capital-poor villagers as the lion's share of the profits is already reaped by those who controlled the capital and the upstream and downstream business, as seen in the "soy complex" in Bolivia (McKay and Colque 2016) and oil palm boom in Indonesia (McCarthy 2010). In many other cases, the situation is even worse, when those dispossessed do not have the opportunity to be included, which was vividly described as, "their land is needed, but their labour is not" . In other words, when these villagers are dispossessed of their key livelihood sources, they are not offered any other, albeit usually less favourable, livelihood opportunities. As a response, those dispossessed and possibly displaced villagers are observed to resist in varying ways (Borras Jr and Franco 2013;Moreda 2015;Mamonova 2015).
Within land grabbing, a large part of the villagers' livelihoods tends to be partially or completely destroyed, particularly when villagers' capacity to maintain their land control and land use is low. In research, there has been a tendency to focus on what has been taken from villagers during the landbased change. This is very valid and has far-reaching social relevance. However, it cannot fully capture the dynamics.
For example, in Guangxi, the Industrial Tree Plantation (ITP) sector has been expanding rapidly and this has led to extensive changes in land use and land control since the beginning of the 2000s. During this process, around 0.2 million ha of land were acquired by two foreign investors (namely APP from Indonesia and Stora Enso from Finland) to plant eucalyptus trees to secure a stable supply of raw materials for their production (Liu 2010;Stora Enso 2016). Meanwhile, the ITP sector creates almost no job opportunities for the local population (as will be elaborated on in Sect. 3). Thus, this case seems to fall into the worst situation outlined above. However, in reality, the rise of the ITP sector took place in Guangxi without full displacement. In the course of the land acquisitions, the villagers' livelihood changes were highly diversified: a few villagers did not engage with the sector, while others were able to be incorporated, and even to become grabbers themselves. Thus, some villagers, rather than completely losing their livelihoods, were able to maintain and even expand their livelihoods. 1 Villagers in rural Guangxi were not pushed into the vulnerable situation experienced by villagers in other regions of the world. To see how these relatively positive livelihood changes occur has implications for the analysis of a land grab.
To understand these dynamics, this paper explores the "taken and left" dynamics (i.e. what is taken from and what is left to the villagers based on the associated contextual settings in rural Guangxi), largely based on an extensive set of primary data collected from the author's two fieldwork trips in the Guangxi Province in the spring of 2015 and 2016. In 2015, the author conducted interviews with various actors, including 47 in-depth interviews with villagers in 7 counties across 5 cities in the southern part of Guangxi where there are distinct dynamics within the development of the ITP sectors. In 2016, the author conducted a household survey of 106 respondents in four villages using snowball sampling. These four sites involve distinct forms of land grabbing at varying scales, including foreign company-dominated, state-owned farm-dominated, individual entrepreneur-dominated and local villager-dominated land grabs. To deepen the understanding of villagers' answers, this survey was conducted in combination with interviews, and the questionnaires were used as a checklist during the process. During this fieldwork visit, the major challenge for the author was to build trust within a relatively short period (around two months) as an outsider (who cannot speak the dialect). Although the author has tried to partly address this challenge by living with a local household and hiring a local villager in each site to help introduce and translate when necessary, it is still difficult to gain sufficient trust to access some sensitive information (e.g. household income). In addition, data from secondary sources, including official reports, statistical yearbook and news reports and observation from the preliminary fieldwork in 2014 are also deployed during the analysis.
This piece focuses on two key contextual settings that closely affect "taken and left" dynamics, namely, the rural land system and internal migration. It demonstrates that, on the one hand, the rural land system in Guangxi enables investors to acquire massive amounts of cheap land for the development of the ITP sector without completely displacing villagers. Thus, villagers' access to resources and livelihoods are not undermined. On the other hand, these villagers are in less vulnerable positions, because of their diverse livelihood sources based on their (albeit uneven) access to off-farm work opportunities. This paper does not intend to generalize this livelihood change trajectory to the broader geographic scope or to champion these positive livelihood changes in a land grabbing case, as all these dynamics have emerged under a certain context and some villagers indeed do become more vulnerable. Instead, it hopes to argue that over-focusing on what and how much has been taken away by grabbers might hinder a more comprehensive understanding of the diverse livelihood changes and that it is equally important to have a look at what and how much is left to villagers in certain contexts.
The remainder of this paper is structured as follows. The next (second) section offers an alternative framing, namely, "taken-left" dynamics, to analyse livelihood changes underlying land grabbing. After this, the third section offers the background of the development of the ITP sector in Guangxi. In the fourth section, this paper uses three typical illustrative cases to show the importance of using the "taken-left" dynamics to better understand the livelihood changes that occurred in Guangxi. The fifth section analyses the "taken-left" dynamics (i.e. what is taken from and what is left to villagers) and maps the specific rural land-labour relations that are embedded in the area. These contextual settings are closely associated with not only what has been taken by the investors but also what has been left to villagers when a land grab takes place. On one hand, the rural land system (re)shapes the schemes of land grabbing, enabling massive changes in land use and land control without full displacement. In other words, after investors' land acquisitions, most farmland plots are still left to villagers. On the other hand, the prevalence of rural-urban migration indicates that although many villagers are not employed by the ITP investors, they can use the labour resources left to them to conduct more profitable off-farm work and thus obtain additional income.

"Taken-left" dynamics
Among the current media reports and academic studies on land grabbing, "what is taken" from the villagers by grabbers tends to receive much attention. This is both a theoretically grounded and practically validated angle. On the one hand, dispossessing villagers of their means of production is a key process of capital accumulation (Marx 1992;Harvey 2003). In other words, the development of capitalism essentially requires "taking" something (e.g., land) from a certain group of people, in order to maintain the supply of cheap labour and natural resources. On the other hand, once villagers' access to land and other livelihood resources is eroded, it has significant impacts on both the livelihoods of those affected individuals and the stability of the society. Thus, to make such cases visible is imperative to mitigating their negative impacts, avoiding the naturalisation of the exploitation process and contributing to more accountable politics.
However, this does not mean that this is the single element that should be taken into account when analysing land politics. It is also important to see the flip side of the coin, namely "what is left" to villagers, considering its profound theoretical and empirical relevance.
Large-scale land acquisitions by investors do not always lead to the full displacement of the villagers in question. Investors are keen not to take all the resources from villagers per se, namely, by displacing them, but instead aim to capture surplus value and maximise profit, considering the nature of capital (Marx 1992). Thus, they are motivated to acquire resources when those resources can bring them benefits. Put it in another way, they might intentionally leave certain resources behind, if the context makes this economic.
On the other hand, villagers' livelihoods might not necessarily be damaged even when resources (i.e., lands) are taken, because villagers' agencies and reactions are primarily linked to what is left to them, as described in James Scott's (Scott 1977) analysis of peasants' moral economy in Southeast Asia. According to Scott, "[t]he test for the peasant is more likely to be 'What is left?' than 'How much is taken?' This subsistence test offers a very different perspective on exploitation than theories which rely only on the criterion of surplus value expropriated." (Scott 1977, p. 7). Thus, James Scott argues that "[i]t was the smallness of what was left rather than the amount taken (the two are obviously related, but by no means are they identical) that moved peasants to rebel." (Scott 1977, p. 7). This idea of moral economy originated from Edward Palmer Thompson's finding from the empirical analysis of the English crowd (Edelman 2012), where Thompson pointed out that popular riots erupted when existential rights were violated (Thompson 1971). Similarly, Michael Watts also pointed out the "strength and resilience" of peasants in response to the shortages of food in Hausaland, although he concluded that such dynamics have a class character and were "not always especially moral" (Watts 2013, p. 146). In this vein, the affected villagers are still able to make a living even under squeezed circumstances, through self-exploitation (i.e., consuming less and working harder) and strategic uses of the remaining resources (van der Ploeg 2013). In other words, villagers tend to remain silent if what is left is still sufficient for them to maintain subsistence through drudgery and adaption, as happened in Ukraine (Mamonova 2015).
Therefore, analysing what is left to villagers, is equally important to exploring what has been taken, and opens a new angle to a more comprehensive understanding of the diverse and unexpected livelihood changes in the course of land grabbing. Without it, we cannot understand why, while some villagers are pushed into more vulnerable situations, others are able to make ends meet and even prosper during a land grab. However, what can be taken and what is left are neither spontaneous nor random, but are indispensable to certain contextual components, particularly those factors that are linked to villagers' capability and autonomy in maintaining and gaining access to resources (particularly land and financial resources) and livelihoods (both on-farm and off-farm work).
Thus, what can be taken by investors and what is left to villagers is not fixed, but changing with power dynamics. As in the case of Guangxi, faced with various forms of resistance, some investors did give back villagers small plots of land after large-scale acquisition.
What is left to villagers after a land grab also tends to be differentiated among villagers, who are not a homogeneous group, but highly differentiated in terms of access to land and labour resources (including access to work opportunities). Based on distinct resources left to them, villagers then will have diverse livelihood changes. In this sense, what is left to villagers is inherently a relationship between individuals, groups, and institutions that are embedded in a contested dynamic.
Based on the agrarian political economy analytical framework, this work advances a "taken-left" dynamics framing as shown in Fig. 1. This framing takes both what is taken from and what is left to villagers into account based on contextual settings that dynamically shape the land, labour and financial conditions of rural villagers. These settings are closely linked to what can be taken by investors and the (both visible and invisible) economic and social costs associated with the taking process. Meanwhile, because of their critical roles in governing people's access to land, labour and financial resources (i.e. what is left to villagers), these settings significantly affect villagers' capacity to maintain (or even to expand) their livelihoods during a land grab.
In the context of rural China, two related contexts include the rural land system (Household Responsibility System reform or HRS reform) and massive internal migration (rural-urban migration). In addition, other factors that directly or indirectly contribute to change of the income of rural households (e.g., increasing public monetary transfer under the rural revitalization strategy) are also relevant to the villagers' livelihood changes.

Background: the rise of the ITP sector
The ITP sector emerged and has been expanding massively in Guangxi since the early 2000s. Among the industrial tree crops, eucalyptus trees, particularly the fast-growing tree species (e.g. E.urophylla) and are the most common. 2 According to Yang (2019), the area covered by eucalyptus trees in Guangxi expanded over ten times from 0.15 million hm 2 in 2000 to 1.78 million hm 2 in 2015. Most of the ITPs in Guangxi are established in the southern part of the province given the suitable natural conditions for the cultivation of eucalyptus trees. The products of the ITPs are used as industrial raw materials, including for the produce of the paper-pulp and wood-based panels.
The ITP sector is land intensive. To build ITPs, a large tract of land is acquired from rural villagers by foreign and domestic companies (including state-owned farms) or entrepreneurs. These land acquisition practices are largely shaped by the rural land system. Mainly, they result in taking from villagers collectively-owned forestland rather than family farmland (as will be elaborated in "Land system and "taken-left" dynamics around land" section). While many large-scale land-based projects initiated by Chinese investors got public attention but did not actually gain ground or failed/ were abandoned during the process (Borras et al. 2022;Mora 2022;Bräutigam and Zhang 2013), land grabs do happen inside China, not only by the state actors for infrastructure development and urban construction but also by the private sector for agro-production (Andreas et al. 2020).
Meanwhile, the ITP sector is labour-saving and only requires seasonal labour in the first two years for planting, weeding and fertilizing. 3 This indicates that its massive changes in land-use and land control creates limited Fig. 1 Schema on "taken-left" dynamics in a land grab job opportunities. What is worse, such job opportunities are not always offered to local populations. According to the author's fieldwork in Guangxi in 2014, 2015 and 2016; instead of local villagers, many investors tend to employ relatively cheaper labour from other provinces (e.g. Yunnan and Guizhou) via brokers and even cross-border migrant workers from Vietnam. 4 According to a focus group discussion in one village group in Hepu County where around 300 mu of forestland has been leased out to a foreign company, only eight to ten local villagers are employed by the investor for two or three days each year to fertilize trees (fieldnotes, 18 March 2015). 5 In a few cases, some of the investors employ few local people as guards to protect the plantations. But such opportunities are very limited, as "one person can take care of 1000-2000 mu plantations" and the people hired are normally local elites (fieldnotes 18 March 2015).
Thus, this seems to be a typical scenario of what is vividly summarized by Tania Li as "land is needed, but labour is not" . Moreover, as observed in other regions, when the land taken from villagers is mainly common land, villages tend to be put in a more vulnerable position (Dell'Angelo et al. 2017;Gerber and Haller 2021). However, a closer examination of villagers' livelihood change reveals a distinct scenario.

Three cases of livelihood changes
After large-scale forestland is taken by investors for the rise of the ITP, villagers' livelihoods changed. Such changes were not unified and did not always lead to extremely negative outcomes. As shown in the three illustrative cases below, 6 there are varying directions of livelihood changes during the land-based change: some villagers chose to engage with the ITP sector, while others do not. Although these villagers' land had been taken, their livelihoods were largely maintained and at times even expanded. They call for a careful examination of the "left-take" dynamics during these land-based changes.
Case 1: A villager in W County used to do migrant work in Guangdong Province. When he saw a foreign company entered his village and leased forestland owned by his village collective to plant eucalyptus trees, 7 he felt that it is not worthwhile to lease out land. So, he decided to follow and built his own eucalyptus tree plantations. In 2005, he rented 30 mu of forestland from his own village collective at the annual rent of 8 yuan per mu and started to cultivate eucalyptus trees with his family labour (his wife and himself). With the profits he gained from these trees, he gradually expanded his ITPs, via leasing, to more than 100 mu in his village. With the expansion of his ITPs, he transferred the production mode and now is completely dependent on employed labour for planting and harvesting trees. Moreover, in addition to directly acquiring land, he also controls large-scale ITPs owned by other villagers via indirect channels (e.g., using loans). He purchases eucalyptus trees from other nearby planters (mainly small-scale ones), then employs specialized workers to log and transport trees, and sells trees to whomever offers the highest prices. He also lends money to planters who are in short of money to purchase fertilizers. In return, he gets the promise of these planters that their trees will be sold to him at a certain lower price after they are mature. In this way, he becomes better off and is known as a "big boss" (Da laoban) by his fellow villagers. However, he also feels that the ITP business carries a huge risk. His 70-mu plantation was destroyed by the typhoon last year, and he needs to spend extra money to hire workers to replant trees, (Field notes, 18 March 2015).
This case shows how some villagers are able to have their livelihood expanded when a land grab emerged. When these villagers lost part of their livelihood resources, they retook and gained control over forestland from other villagers to expand. During the process, based on the financial resources they accessed from the previous off-farm work, they have shifted to commercial farming.
Case 2 A villager in X County, similar to many of his fellow villagers, chose to become an independent planter of eucalyptus trees in 2010. He maintained his control over the farmland plots allocated to his household during the Household Responsibility System reform (HRS reform), when forestland was leased to investors. He converted the use of some farmland plots (3mu) from sugarcane production to eucalyptus trees. Meanwhile, he retains the cultivation of food and other cash crops (e.g., sugarcane) on other land plots. Because, as explained by this villager, his household also needs crops that could bring income annually to cover their daily expenses. He explains that his strategy of crop choice is to cultivate one type of crop with a short growth cycle that can generate income yearly while planting two kinds of crops with a longer growth cycle, which he vividly described as "one short and two long". In this way, he told me confidently "(the 3 mu eucalyptus trees planted) are my pension when I grow old. Then I can enjoy my twilight years." (Field notes, 19 February 2016). 8 For his household, planting trees is actually offering an additional livelihood choice. 9 In contemporary China, a rural household usually has more than one source of income, be it farm, nonfarm, or, in most cases, both. Villagers like him are not in a situation where there is little choice but to plant a certain crop in a certain way and for a monopolized market, as happened in the sugarcane boom in Uganda (Martiniello 2020) or in Bolivia (McKay and Colque 2016). Instead, these villagers are able to produce and harvest more flexibly and there are distinct purchasers, including middlemen, timber processing mills of different sizes and paper-pulp companies, partly because of the multiple uses of eucalyptus trees.
In this case, these villagers chose to engage in the expansion of the ITP sector as land investors, unlike the livelihood choices of villagers in Yunnan province (Hua et al. 2021). They are able to maintain their livelihoods and to some extents have had their resilience enhanced.
Case 3 A villager in W county remained in off-farm work in urban areas after 2000 mu of the collectively owned forestland has been leased to a foreign investor in his village. Although he complained that there is no place to pick firewood after the operation of ITPs and the paddy cultivation became more difficult as the water is extracted by those trees, he pointed out that "before the (eucalyptus) trees were planted, (I) have gone out for migration work. Now when the trees are planted here (by investors, (I) still went out for (wage work)", because "the income from wage work for one month is better than the income (from farming) for one year". (Fieldnotes, 3 March 2016). Meanwhile, his household remained reliant on their farmland plots for food production.
This case shows that although some of the livelihood resources are dispossessed by investors, their livelihoods are still largely maintained as long as their access to off-farm work opportunities is undamaged. In addition, the continuing farmland access also supplemented their livelihoods.
These cases indicate some common scenarios in the course of the land-based changes in Guangxi. Although part of these villagers' livelihoods is taken, the negative impacts of land acquisition on livelihoods are less significant than those in other regions, particularly when land grabs take place on common land (Dell'Angelo et al. 2017;Gerber and Haller 2021). To understand these livelihood changes, following the discussions of the framing in ""Taken-left" dynamics" section above, it is important to investigate what is left to villagers while examining what has been taken by investors, particularly about the key contextual settings that shape these "taken-left" dynamics.

Contextual settings and "taken-left" dynamics in Guangxi
As shown in Fig. 2, with the rise of the eucalyptus trees in Guangxi, two interlinked contextual settings are closely tied to what is left to the villagers, and alter the trajectories of these villagers' livelihood changes, namely, (i) the rural land system in China and (ii) the off-farm work opportunities. This particular land system enables investors to acquire massive quantities of cheap land without displacing villagers. Villagers' access to farmland is largely retained even when large tracts of land are taken for the ITP boom. Meanwhile, labour transfers to non-farm sectors continue, partly driven by the land system in rural China, which diversifies villagers' livelihood sources and offers villagers channels to earn additional income. Because of income gained from non-farm wage jobs, villagers are able to make more flexible livelihood choices to maintain and even expand their livelihoods during changes.

Land system and "taken-left" dynamics around land
The landscape in Guangxi is rather hilly, with 15.27 million ha of forestland accounting for around 64% of the total land area of the province and over 3 times as much as the farmland there. 10 Regarding the land property of rural land in Guangxi, aligned with the other regions in China, HRS reform was launched at the beginning of the 1980s after the initiative of intra-community land redistribution in Xiaogang village in Anhui province in 1978. 11 Under this land reform, the user rights to the land were contracted to rural households and the property rights remained collectively owned (Unger 2002;Ye 2015). Out of the principle of fairness, the land, mainly farmland, was firstly divided into different grades according to quality (i.e. fertility) and then distributed according to household size. As a result, each rural household was allocated to several tiny and scattered farmland plots of different grades (qualities) considering the unique local natural geographic conditions. After this initial round of land distribution in the 1980s, although land reallocations in response to demographic change still occur among a few villagers, "small adjustment along with a stability as a whole" is the main principle promoted by the state (Kung 2000;Chen and Gan 2019). This means that the land reallocation on the ground is infrequent and partial and land entitlements remain essentially unchanged since the HRS reform in many villages based on my observation during the fieldwork in Guangxi. Thus, the fragmentation of farmland remains in  rural Guangxi. According to a survey conducted by the author in Guangxi province, the farmland (including both paddy land and dry land) controlled by a rural household is 10.1 mu in size divided into more than 11 plots on average. To be specific, as shown in Figs. 3 and 4, the majority (around 77%) of the households in question reported having fewer than 5 mu of paddy land, while 67% of them have 3-10 paddy land plots. Meanwhile, 68% of these households have less than 10 mu dry land and 63% of them have more than 3 dry land plots.
In the early 1980s, villagers were not allowed to transfer the land plots allocated to them due to the restriction by law. Later, the state loosened control over land transfers, but these were still limited to a few trial sites. In the late 1990s, rural rental land markets emerged and, since then, land commodification has developed rapidly with the introduction of a series of state policies that promote land transfers and land consolidation (Zhang et al. 2004;Zhang and Wu 2017;Ye 2015). This change in the land market in Guangxi de facto paved the way for the advent of land acquisitions for the construction of ITPs.
Although this reform leads to the fragmentation of farmland in rural China, it ensures the villagers' de facto control over farmland and enhances villagers' autonomy and capacity in three ways. First, with access to farmland plots, villagers are able to control the process of agro-production and exchange/circulation on these land plots. 12 Second, villagers are also able to mortgage or lease out the user rights of their land-either partly or fully-in exchange for the financial resources that may be used to support their alternative livelihood choices (e.g. a small business). Third, villagers' control over land can be treated as a final "safety net"-securing a minimal level of subsistence if their ventures in other nonagricultural sectors fail. Thus, villagers' control over farmland under this land system enables them to flexibly adapt their livelihood strategy when encountering changes.
The forestland in Guangxi has a distinct trajectory in terms of land control change. In Guangxi, more than 90% of the forestland is owned by the collective, and the rest is owned by state-owned forest farms. 13 In line with the HRS reform, the forestland was also intended to be contracted to rural households together with farmland at the beginning of the 1980s. In 1981, a central policy (the Sanding policy) was issued, aiming to "determine forestland ownership, the reserved forestland and the responsibility for forest/ plantation production. 14 Redistribution of the collectively owned forestland was taking place in several pilot rural sites across the country. However, this reform was found to lead to excessive deforestation in these regions, particularly in Southern China. Thus, in 1987 a new policy was launched by the Chinese central state to curb this reform, stating that "collectively-owned concentrated and contiguous forestland that has not yet been distributed must not be distributed" in Southern China. 15 This means that except for a small part of forestland that had already been allocated to rural households in a few villages, most of the collectivelyowned forestland in Guangxi remained undistributed before the rise of the ITP sector. In 2008, a new round of forestland reform was introduced. 16 Under this reform, the user rights to forestland were formally cleared and distributed, although a large part of forestland has already been de facto allocated to either investors via leasing or to villagers via both customary occupation and leasing due to the ITP boom. Thus, compared with farmland, forestland in Guangxi is relatively concentrated and contiguous.
Linking this with the dynamics of the ITP boom in Guangxi, this land system has a prominent role in shaping what can be taken from and what is left to the villagers in question. This system features fragmented farmland on the one hand and contiguous forestland on the other. 17 This means that investors will need to negotiate and have contracts with numerous smallholders to acquire farmland plots, indicating that, in rural China, acquisitions of farmland have much higher transaction costs and risks than those of forestland (Borras Jr et al. 2018). This largely hinders investors' attempts of building large-scale eucalyptus plantations on farmland. The relatively concentrated and contiguous forestland then becomes the target of big companies' land acquisitions. This is encapsulated by the cases of Stora Enso's and APP's land acquisitions in Guangxi (Xu 2018b 17 According to the law of China, some forestland and some farmland plots (excluding the permanent basic farmland) are permitted for aboriculture (see Article 36 of the Land Administration Law of the People's Republic of China (2020 Amendment), but villagers are also observed to plant trees on the permanent basic farmland as there are no explicit rules on related punishment. 12 The decision making is not completely free, but is continuously (re)shaped by interventions from "above", from "below" and from the intermingling of the two (Xu 2019b). 13 According to the Guangxi Province's Twelfth Five-year Plan of the Development of Eucalyptus (2010Eucalyptus ( -2015. that they acquired in Guangxi are forestland, either from state-owned farms, rural collectives, or to a much more limited extent, from individual villagers (Liu 2010;StoraEnso 2016). Before these land acquisitions, the collectively owned forestland was largely neither claimed nor used by villagers. There were no obvious profitable crops suitable to plant on hilly forestland before the ITP boom. This means that, to use those forestland plots, extensive labour and/or financial resources would be required to prepare the land. Thus, many villagers did not strongly resist the land acquisitions, as explained by a villager that "if we had not given the land to him (referring to the foreign investor), we also did not have money to invest at that time" (fieldnote, 3 March 2016). 18 However, this does not mean that villagers' livelihoods were unaffected during the process. Some villagers did get income from the forestland before it was taken by investors. According to my fieldwork in Guangxi in March 2016, a group of villagers informed me that they used to cut and sell firewood from the collectively owned forestland and this was a vital cash income source for them (focus group discussion note, 3 March 2016). This implies that when these land plots were contracted to foreign companies, these villagers lost part of their livelihood sources, although such income from forestland was usually minor compared with other income sources (e.g., wage income). Villagers also lost the opportunity to benefit from the sharp increase of the forestland value as the low land rent has been locked in by the land acquisitions. During these land acquisitions, the term of land leases was usually 30 years (or 15 years for contracts signed later) and land rent usually remained unchanged as it was when it was largely leased to investors and middlemen at the beginning of the ITP boom. 19 But, land rent had increased more than fivefold, from 8 yuan per mu per year in 2000 to 42 yuan per mu per year in 2015 in one village I visited (fieldnotes, 18 March 2015). This implies that villagers lost the opportunity to gain sufficient economic profit from the forestland when it was taken at a relatively low price and for a long period. In this sense, for many villagers, this case includes a typical dispossession process. Their land is taken but they get little compensation. Thus, this is different from what is framed as accumulation without dispossession (AWD) when villagers can get abundant compensation in form of "valuable flats, commercial venues and/or secure jobs" Zhan (2019, p. 447).
In short, collectively owned forestland was taken from the villagers, and this led to negative impacts on many villagers' livelihoods. But when companies are aimed at taking forestland rather than farmland, villagers were capable of at least maintaining their access to farmland during the massive land control change, as shown in the above cases (1, 2 and 3). Thus, villagers' access to farmland is largely left intact, because this land system enables investors to acquire massive amounts of cheap land without completely dispossessing villagers. To put it in another way, the villagers in question then were not deprived of their corresponding livelihoods from these farmland plots they could access. 20

Internal migration and "taken-left" dynamics around labour and financial access
Even before TPI, the labour conditions in rural Guangxi have been changing significantly in the past two decades, with the prevalence of villagers' non-farm work. On one hand, Chinese villagers are seeking additional income, as the population pressure makes the land plots distributed too small for the villagers to maintain their subsistence purely based on farming, called the "involution" of agriculture by Philp Huang (Huang et al. 2012). On the other, a huge demand for cheap labour in urban areas exists, because of the rapid urbanization and industrialization (Andreas et al. 2020). In response, a majority of villagers work off-farm in the towns nearby or in cities further away, turning into peasant workers (nong ming gong), while leaving other household members, usually the elderly, the weak, children and sometimes women at home (Murphy 2021;Ye 2018). These peasant workers are difficult to permanently settle in the urban areas due to a series of contextual and economic factors. 21 Thus, their migration is typically cyclical and their wages are mainly remitted back home, while other family members are conducting small-scale farming at home under a household division of labour (Zhang 2015;Murphy 2002).
According to the Peasant Worker Survey Report issued by the Chinese National Bureau of Statistics, the number of peasant workers has increased from 225.42 million in 2008 to 288.36 million in 2018, as shown in Fig. 5. They accounted for 31.8% of the total rural population in 2008, and increased to over 50% of the rural population in 2018. This means that more than half of the villagers in China are currently doing off-farm work. The labour condition in Guangxi also follows this trend. As shown in Fig. 6, the majority (89%) of the rural households surveyed by the author are doing off-farm work either in urban areas, in the nearby counties/townships, or both.
These peasant workers are of low cost, because their reproduction is partially complemented by the agriculture production with the land plots allocated to them in the countryside, and their bargaining power is normally weakened since they are usually locally rootless and lack support from local communities (Huang et al. 2012;Chuang 2015;Ye et al. 2013). Among these peasant workers, according to the Survey Report 2018, 22 around half are working for industrial sectors (mainly in the manufacturing and construction industries). This reflects the swift development of industrial sectors in urban areas, which is largely dependent on cheap labour from rural areas.
This shift has significant relevance to the development of the ITP sector in Guangxi. It explicates the change in rural labour conditions, which fits neatly with the cultivation of eucalyptus trees. During the author's fieldwork in Guangxi, many villagers, particularly those young and strong, expressed their preference to migrate out to do wage jobs, because the incomes from wage work are normally much higher than that of farming the tiny and fragmented farmland plots distributed to them. As a villager explained: "without doing wage work, a household (might) gain no income after annual consumption. If (a villager) only conducts farming, it is impossible to support a child's education" (Fieldnotes, 16 February 2016). When the young and the physically strong labourers have migrated out for more profitable wage jobs, the "left-behind" are usually children, women and the elderly (Ye et al. 2013;Murphy 2021;Ye 2018). As a result, there is a shortage of labour in rural areas, although the degree might vary from one household to another. This can be seen in the heavy reliance on cross-boundary labour from Vietnam during the production of sugarcane in Guangxi province, particularly during the harvesting and crushing season (Borras Jr et al. 2018). 23 Thus, this labour dynamic, at least partly, explained some households' land-use decisions towards a more labour-saving crop, namely, eucalyptus trees as shown in case 2 above. At the same time, this migrant rush also means a shift of the composition in rural households' income. As shown in Fig. 7, the average rural household's share of wages in total disposable income in Guangxi has astonishingly increased from 3% in 1991 to 30% in 2018. 24 This implies that the non-agriculture endeavour has become an ever-increasingly vital source of income for villages. Such income, generated from wage work, offers Guangxi villagers an opportunity to access financial resources to actively make further livelihood choices. They can either intensify/expand production or diversify livelihood sources when encountering massive changes in land use and land control, as happened to the villager in case 1. According to Yan Hairong and Chen Yiyuan (2015), financial access via off-farm work greatly conditioned the "capitalization from below" in rural China.
However, access to financial resources from wages is unequal among households. This is related to both the rural households' distinct labour endowments and their uneven access to social capital (Zhang 2015). Some households are able to prosper with more profitable job opportunities in urban areas, while others are left with only unskilled and temporary wage work. As a result, the differentiated capacity to access non-farm incomes, as Forrest Zhang (Zhang 2012) argued, leads to rural inequality. Based on their advantageous access to financial resources, some rural households are able to accumulate at the expense of their fellow villagers (Xu 2018a), deepening social differentiation.
In addition to wages, the financial resources of rural villagers are also shaped by the state policy related to public monetary transfer. According to the Guangxi Statistical Yearbook, the share of transfer income (mainly public transfer through subsidies and pensions) in total disposable income in Guangxi has increased sharply from 5.6% (292 yuan) in 2010 to 25% (3109 yuan) in 2018. This change is related to a shift of national policies towards rural revitalization and poverty alleviation and does have an impact on the income of rural households. Such increases in of income tend to offer some support for villagers to secure their subsistence during land-based changes.
Therefore, as shown in Fig. 2, the ITP boom in Guangxi takes large tracts of land from rural communities and does not bring sufficient work opportunities to incorporate those affected villagers. But, situated in the contextual settings in rural China, villagers are not fully displaced and are left with, at least, those farmland plots that have been allocated to them during the HRS reform, and although many villagers are unable to get employed by investors of ITPs, they do use the labour resources left to them to access off-farm work, and thus obtain additional income. Thus, with the land and labour that are left, the affected villagers are able to access land and financial resources, which then facilitate their distinct active livelihood changes during the rise of the ITP sector.
Underlying these "taken-left" dynamics, the villagers in question still have (albeit differentiated) autonomy and capacity to make active livelihood choices and have relatively positive livelihood change. When villagers are left with off-farm wage work opportunities undamaged, they can either use the money earned from wage work to acquire extra land to large-scale ITP entrepreneur themselves (as case1) or actively step out of the sector, largely dependent on wage income as the main livelihood sources (as case 3). In both cases, farmland plots left to these villagers are also 23 According to Borras et al.'s research, the shortage could be as much as 50,000 labourers in Guangxi. And there are similar dynamics in China-Yunnan border (Borras et al. 2021;Hua et al. 2019). 24 As already mentioned in the introduction, the author was unable to collect relevant data on household income during the survey. Thus, the general data from official yearbooks are used here to show the shift of the composition in rural households' income. an integral, but not a main, part of their livelihoods. Meanwhile, when villagers are left with farmland plots, they are sometimes able to make more active land-use choices to either become independent planters (as case 2) or step out of this sector (plant other crops).

Conclusion
When large-scale common land is taken from villagers by investors with little compensation and their labour is not needed, villagers' livelihoods tend to be largely destroyed, as shown in many critical studies on land grabbing. This implies a tendency to focus on what has been taken from villagers during land-based change. This is very valid and has far-reaching social relevance. But the case of ITP shows that some villagers are capable of maintaining their livelihood and even expanding in this circumstance. This seems to be an anomaly at first glance, when only considering what is taken from villagers. But these relatively positive changes can be explained when one closely examines both what is taken and what has been left to villagers underlying certain institutions ("taken-left" dynamics).
With the rise of the ITP sector in Guangxi, although large-scale forestland is taken by investors and few work opportunities are created to incorporate affected villagers, villagers' control over farmland plots and their access to alternative livelihood opportunities (e.g. off farm work) are remain under the context of the Chinese agrarian transformation and certain settings around land and labour. To be specific, when investors acquire land for large-scale ITPs, they prefer relatively contiguous and undistributed forestland plots rather than fragmented and distributed farmland plots, considering the high transaction costs of the latter. In this sense, the Chinese rural land system does empower the villagers and prevent their full displacement. To put it in another way, villagers' access to the farmland plots is mostly retained during this ITP expansion. Thus, these villagers are at least able to change the use of their farmland plots, and either engage with the booming sector or step out of it, towards a more profitable direction based on their understanding. Meanwhile, Chinese villagers have recently tended to highly diversify their livelihoods. Many villagers have additional, in some cases the majority of, their income from off-farm sectors. In addition, rural villagers, particularly the poor, do receive financial support from the state under the rural revitalization and poverty alleviation strategies. Thus, although part of villagers' livelihoods might be affected due to the loss of previously collectively used forest land, their access to off-farm work secured their subsistence and, in a few cases, provided the financial capital required to extend their livelihoods.
These dynamics are embedded in certain political economic contexts in China, but the impacts and implications of analysing these dynamics go beyond the Chinese context. This study demonstrates that analysing what is left to villagers is as important as exploring what has been taken from them, when analysing land grabbing. Neither negative nor positive livelihood changes are intrinsic to capital accumulation. Without examining what is left, we cannot understand that while many villagers are pushed into more vulnerable situations, others are able to make ends meet and even become better off during massive changes in land use and land control. Thus, this paper highlights the importance of examining the "taken-left" dynamics and the associated contextual settings that (re)shape the land, labour and financial conditions of rural villagers. This framework of "taken-left" dynamics enables us to better understand livelihood changes not only in China, but also in other regions more globally, including why villagers in Ukraine were able to survive and co-exist with large farm enterprises (Mamonova 2015) and why villagers in Thailand were able to get involved in the eucalyptus and oil palm boom when investors did acquire large-scale land (Barney 2004). In this way, this paper hopes to offer a framework to understand the diverse and unexpected livelihood changes more comprehensively in the era of the global land rush.
Acknowledgements The final write-up work of this manuscript is funded by the European Research Council (ERC) grant to the . I would like to thank the organizers and participants of the Writeshop-Workshop in Critical Agrarian Studies and Scholar-Activism, for their comments on an earlier version of this paper. I would like to thank the investigated villagers for their patience and support. I am sincerely grateful to three anonymous reviewers for their critical and constructive comments that helped us sharpen our argument and improve the overall quality of the paper. I would like to thank Linda McPhee for the excellent copyediting of the text.

Conflict of interest
The authors have no competing interests to declare that are relevant to the content of this article.
Ethical approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent Informed consent was obtained from all individual participants included in the study.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.