De-agrarianisation and re-agrarianisation in patches: understanding microlevel land use change processes in Nepalese smallholder landscapes

ABSTRACT This paper investigates the ongoing reproduction crisis in Nepal. We utilize farmer context-specific actions of ‘hanging in’, ‘stepping out’ and ‘stepping up’ to unpack the pathways of de-activation, de-agrarianisation and re-agrarianisation in four spatially and socially differentiated landscapes. We detail a continuum of land use and labour use intensity, the microlevel variations and repertoires of actions in relation to landscape, shrinking farm sizes, labour shortages, forest expansion and increasing wildlife encroachment. The analysis focuses on specific landscape and social contexts and shows how smallholders are fine-tuning agricultural practices to meet subsistence needs. But account also must be taken of ecological variability and socially differentiated access to land to understand how households allocate labour between different land uses and between farm and off-farm activities. Household survival depends as much on the allocation of scarce labour resources as on that of scarce land. It suggests that household rather than just land has become a key unit of production.


Introduction
Reproduction crises (Bernstein 2010) seen as the continuous struggle to maintain households' social and economic viability have been a recurrent theme in many locations in the Global South.A classic example of this is Murray Li's account (Li 2014) of how 'land ran out' for a group of indigenous highlanders in Sulawesi, Indonesia.Land that had previously been managed in common became privatized driven by the emergence of capitalist relations fueled by the cultivation of cacao.For some, cacao growing was a route to relative prosperity.But many became landless with few prospects either in their homeland, for off-farm labour or migration to towns and cities.
The long trajectory of agrarian relations in the Nepalese hills also reveals several crises for household reproduction from a farm-based economy, although agriculture remains an essential livelihood component to almost 66% of Nepal's population (MoALD 2019).But, in contrast to Sulawesi, 'lands end' has not meant in Nepal the complete loss of access to land by smallholders as a result of the development of capitalist agriculture.Rather farm size has shrunk driven by population increase and inheritance practices, diminishing the subsistence component of living that can be drawn from land as households.The historical roots of this long-term crisis lie in the legacies of prior feudal like land relations and the role of caste structures determining inequitable access to land, the burdens of rent and taxes and low agricultural productivity (Wily et al. 2009;Sugden et al. 2016) and these are explored in detail in our study sites.Many households have lived on the margins between subsistence and destitution (Regmi 1978) and faced recurrent food crises (see Blaikie et al. 1980).This led in the 19 th century to a significant out-migration in search of productive farming land and better work opportunities in northern India and Bhutan (Kansakar 1984;Adhikari 2001;Hutt 2003).With the opening of the lowlands in the Tarai in the early 1950s as malaria was controlled, land shortages drove a significant migration from the hills to the plains.But by the 1980s migration to the Tarai had decreased, the Nepalese population had doubled over the 30-year period and yet again the Mid-Hills was seen to be in a state of crisis and confronting an agrarian and environmental dead end (Blaikie et al. 1980).
However, it was evident from the turn of the 21 st century that this had in part been relieved by the growth of transnational migration of rural labour to the Middle East and East Asia creating a remittance economy in the Mid-Hills (Blaikie et al. 2002).But a major food security crisis in 2008 at a time of the sharp global rise in food prices (Hobbs 2009) pointed to the limits of migrant income.Price rise effects overlaid a domestic food production crisis driven by a decade of conflict from the mid-90s (see Lawati and Pahari 2010), a neglect of agriculture, and adverse climate conditions (Bardsley and Hugo 2010; Berchin et al. 2017;Rijal et al. 2022).The need for income in rural Nepal, fueled by land scarcity, declining production and climate change especially in the Mid-Hills, has continued to drive labour migration.By 2014, some 56% of households received remittances which accounted for an estimated 30% of GDP (WorldBank 2016).But migration in turn has contributed to an acute shortage of farm labour, exacerbated by a decline in household fertility rates, reducing household sizes and changing household demographic structures (Uematsu et al. 2016).The insufficiency of remittances to maintain households (Sugden et al. 2018) has ensured that rural households have hung on to what land they have even when production falls far short of meeting simple reproduction needs (Bernstein et al. 2018).
A body of literature has emerged about the drivers of this temporary labour migration (Adhikari 2001;Thieme and Wyss 2005;Maharjan et al. 2020;Sunam et al. 2021) and the consequences of this for the agricultural economy (Fox 2018;Poudel 2019;Maharjan et al. 2020).There is debate as to the extent that remittance flows have (Pant 2006;Wagle 2012) or have not contributed to alleviating poverty rates and food security and supported Nepal's agrarian economy (Gartaula et al. 2012;Sunam and McCarthy 2016;KC and Race 2019;Sugden et al. 2022) but it is evident that these effects are differentiated according to the context in the country's mountain geography.There is general agreement that most remittances are used for basic consumption purposes (Thieme and Wyss 2005;Gartaula et al. 2012;Sunam and McCarthy 2016) or invested in house construction and children's education (Dhakal 2011).However, the effects of remittances and changing labour dynamics must be seen in terms of two long-term processes of change in Nepal's rural landscape -declining farm size and expanding forest area -and how these materialize in diverse land use systems.
There has been an inexorable shrinkage in farm size, overlying pre-existing deep inequalities in land ownership patterns, although this has marked regional dimensions (Wily et al. 2009).Already by 2009(Wily et al. 2009) ) an estimated 58% of farmers or 2.7 million rural households were functionally landless with less than 0.5 ha.By 2021 it has been reported that average farm size had shrunk in Nepal from 1 ha in 1970 to 0.7 ha (Rigg 2020) leaving little margin for surplus production.As we shall see, many households have much less than this although we found few completely landless households.There is no evidence of a classic agrarian transition (Timmer 2014) taking root in Nepal through commercialized agriculture and land consolidation despite the Nepalese's government efforts to promote it (GoN 2015) through high value crops (see also Khatri et al. 2023).While there may be evidence of a growing engagement in cultivating cash crops in areas well connected to urban markets (Chhetri et al. 2023), for many, maintaining the subsistence component of farm production has been a priority (Holmelin 2021;Khatri et al. 2023).
It is evident that in common with many other South Asian countries (Kelly 2011;Rigg et al. 2018, McCarthy 2020;Rigg 2020) smallholders are hanging on and not deserting the agrarian economy as the agricultural transition orthodoxy would have them do (Timmer 2014).This is not because they see much of a future in agriculture but rather because they cannot get out.As Rigg (2020) argues this persistence reflects an ongoing household reproduction crisis and as McCarthy (2020) puts it with respect to Indonesia, households are moving sideways.Households have needed to keep a foot in agriculture to support a distributional or sharing economy (Ferguson 2015) to secure household reproduction under circumstances where the wage rates of migrant labour are insufficient in amount and reliability to meet the costs of household reproduction (Sugden et al. 2022) and leaving agriculture is not possible.As both Li (2014) and McCarthy (2020) emphasize, the dynamics of agrarian change are highly contingent and empirically complex, depending on context.This paper argues with respect to Nepal's mountain landscapes that an understanding of micro-environmental variability is central to making sense of the social and often parallel land use outcomes of agrarian change processes as different kinds of households struggle to secure a subsistence contribution to the household from land.
As we shall see, most households are determined to hang on to their land where they can, even if they don't actively farm it all.Rural household economies are made up of a varying and shifting mix of on-farm, off-farm and non-farm activities that stretch from rural to urban and from Nepal to abroad, tilting the axis of the unit of production from land to household labour and its deployment (Harriss-White 2023).At the same time, a forest transition has been achieved (see Mather 1992) and the recovery of Nepal's forests, particularly in areas of community forestry, has been widely documented (Chhetri et al. 2023).The significant point is that the forestry transition has not taken place in response to an agrarian transition whereby increasing agricultural productivity has driven a retreat from more marginal land (Pain et al. 2021) allowing forest expansion.Rather forest policy has had a more consequential effect.
The effect of rules around forest management and forest expansion has been to drive a decline in forest use (Marquardt et al. 2016;Poudel 2019;Poudyal et al. 2023) including a reduction in livestock numbers which are today kept installed and mainly fed with on-farm tree fodder supplies.The dual outcomes of forest use regulations and the loss of farm labour through migration and shrinking household sizes have encouraged a retreat of the cultivation boundary from more marginal land, as well as contributing to an increase of trees also on farmland.But the recovering forest areas has in turn created new threats, not least from crop losses due to the depredations of increasing wildlife populations (Baral et al. 2021;Bista and Song 2022).Thus, on top of some of the traditional drivers of a reproduction crisis for rural households -incorporation into commodity markets, shrinking farm size and low productivity often associated with water scarcity -more recent causes have included significant crop losses due to wildlife damage compounding the effects of a changing climate.
The combined effects of shrinking farm sizes, forest expansion and remittance flows appear to be driving three diverse processes and outcomes.One effect has been to withdraw resources from agriculture or de-activate agricultural production (van der Ploeg 2008) and a withdrawal of labour linked to labour migration has been a significant cause of this.As noted, this has led to a retreat in the cultivation boundary but this should not necessarily be seen as land abandonment (Marquardt et al. 2020) and may reflect a shift to less intensive land use such as fodder production.A second possible effect has been that of deagrarianisation whereby rural households move entirely out of agriculture but this seems not to be taking place at a large scale in Nepal although there are cases of abandoned land (Khanal and Watanabe 2006;Jaquet et al. 2015;Maharjan et al. 2020).A third effect that has been noted (Sunam et al. 2021) is incipient re-agrarianisation whereby remittance income is invested in various forms of commercial agriculture and this appears to be the source of commercialization in agriculture found by Chettri et al. (2021).
These stylized pathways of de-activation, de-agrarianisation or re-agrarianisation need to be seen in the context of specific actions by farming household in concrete locations and what they offer in terms of adaptation possibilities.For the purposes of this paper, therefore, we have adopted Dorward et al. (2009) corresponding characterization for describing specific farmer's actions as, 'hanging in', 'stepping out' or 'stepping up' to explore the microlevel variation and repertoire of actions that farmers pursue for farming to contribute to household simple reproduction.As various authors note (Rigg 2020;Sunam et al. 2021) land use systems are context specific and this is the core focus of this paper.In its exploration of socially and ecologically differentiated actual land use practices and their historical roots in four specifically contrasted landscapes in Nepal, one in the Inner Tarai and three in the Mid-Hills, the paper seeks to understand what land use changes and processes of agricultural intensification are taking place and their consequences for different households.It is interested in both the contrasts in agricultural land use between sites as well as those between households and social groups and their location within sites.

Methodology
We selected four case study villages, to capture contrasts in agro-ecology, farming practices and smallholders' livelihood possibilities.Three of these are in the Nepalese Mid-Hills and one in the plains of the Inner Tarai 1 (hereafter referred to as the Tarai, see Figure 1 and Table 1).The interaction of altitude, landscape position, rainfall, water resources, soil quality and vegetation influences what crops can be grown in each location.
Chyasku in the Mid-Hills and Jhunga in the Tarai exemplify the range of land use intensity in Nepal's landscapes.Jhunga (JH) is sub-tropical, rich in water resources and has significant areas of irrigated farmland (khet) used annually for double or triple cropping.Villagers exploit the broadleaf community forest which contains valuable timber resources.In contrast, Chyasku (CH) suffers from prolonged periods of drought, water constraints and low crop productivity on the rain fed land (bari).Moreover, the area of land that has been deactivated (idled) is expanding, wildlife populations are increasing, and there is a sharp decrease in use of the 30-year-old pine community forests.
There are also important land use differences within the case villages as seen in Khimti and Kalang with contrasts between upland and lowland use.In Khimti (KH), farming is prospering in the lowland, whereas the surrounding upland is depopulating and deagrarianizing rapidly.In Kalang (KA), however, the higher part of the village has become an area of commercial interest for kiwi production due to road access, and the irrigated valley bottom, previously important, is now seen as less attractive.This has given more marginal social groups (e.g.Jirel and Dalits 2 ) access to this productive irrigated farmland.
Preliminary fieldwork was undertaken in November 2019 with two further rounds of data collection between July 2020 and September 2021.The villages contain several settlements/hamlets with households from different castes, ethnic groups and economic status.In each village, we selected respondents to cover this diversity.In the first round of fieldwork, we held in-depth household interviews, focusing on land size and land management, investments in land, livestock rearing and livelihood risks and responses.Additionally, we undertook transect walks, group discussions and participatory resource mapping to understand the local landscape in relation to the different land uses and social groups in the village.In the second round, we purposively focused on a subset of households  and life-story timelines, mapping life trajectories.This added information on changes in land holdings, land use practices, livestock keeping, forest use and income portfolios with smallholders, and these provide the key narrative evidence of this paper.We further arranged group discussions on forest use in local forest user groups.Table 2 summarises the methods used and the data collected during fieldwork.Table 3 locates the social identity of the informants for the life story interviews by site and position in the caste hierarchy.

Findings
In the case villages, we identified three types of land use, the first two types corresponding to the notion of 'hanging-in' and the third to 'stepping up'.'Stepping out' might be expected to be reflected in land sales, but there was no evidence of this and the few households that left the village either rented or sharecropped out their land.The first type, (I), was use for subsistence which clearly remains a key farming objective although few households gain more than 50% of their household reproduction needs from land.The second type was idling land (II), a form of land deactivation which is often referred to as abandoned land in the Nepalese literature (Khanal and Watanabe 2006;Jaquet et al. 2015;Ojha et al. 2017;Maharjan et al. 2020).These areas are mostly not abandoned but remain extensively managed for fodder and timber trees.These replace the contribution from community forests for many smallholder households.The third type (III) is intensified land use, in which investments and farming practices are oriented toward production for the market.
A fourth type of 'Stepping out' and abandoning land was actually very rare in all study sites and we therefore focus our discussion on the first three land use types described in Table 4. Importantly, all these three land use types can often be found on different parts of a single farm and landscape in variable proportions.Figure 2 gives an overview of the spectrum of land use intensities in each of the study sites, the arrows indicating the direction of changes.Our findings suggest that farmers are strategic in their practices matching household resources to landscape position and the categories should be seen as a continuum of land uses (see Figure 2).However, variations in agroecology and altitude between and within the study villages lead to different combinations and

I) Maintenance of subsistence farming
Farming to meet household food needs.Fields are cropped with traditional food crops, using customary practices and a minimum of inputs.Found on both rainfed bari and irrigated khet lands.Smaller numbers of livestock are kept.Corresponds to the notion of 'hanging in'.

II) Idling use of land
An extensive use of land with low production potential (mostly bari land), exposure to crop loss through wildlife, or limited water resources.Fields are managed with a minimum labour inputs for fodder, fruit and timber trees.Corresponds to the notion of 'hanging in' when some product/income is still harvested.III) Intensified use of land A land use implemented on the most productive land (mostly khet land) and on land strategically located.The farmer intensifies these cropping areas or/and the livestock keeping.The production is on semi-commercial scale, sometimes with use of micromechanisation such as use of hand power tillers.The intensification is based on an increased use of external inputs in the farming system.Corresponding to the notion of 'stepping up'.locations of these land use types.At a village level, these are in turn structured by caste identities and legacies of land control before the implementation of the Birta 3 Abolition Act in 1959 (see Adhikari 2011).In the following sections, we will describe each of the study sites in descending altitude from Kalang.For each case village, we give a short historical background of social dynamics and local land uses, we then present land use variations and ongoing changes.

Kalang: a higher altitude village with a selective move into perennial cash crops
Kalang stretches vertically up a mountain landscape; with cold winters but with reliable rainfall, it supports a rich mix of vegetation.There are irrigated khet fields in the valley at around 1000 masl, rainfed bari plots on the rising slope and at the top, a forested area rises to more than 3000 masl.The 800 households (HH) are spread between three sub-villages at lower, middle and higher altitudes but clustered by social identity.The largest social groups (more than 85% of the population) are the Jirel (310 HH), Sherpa (260 HH) and Kshetri (120 HH) families while Tamang, Dalit and Bhujel households comprise the remaining part of the population.The original settlers were the Sherpa and Jirel people, the Sherpa living in the upper part close to their grazing areas and the Jirel in the lower areas.Some six generations ago, a Kshetri man who was a revenue collector was granted extensive areas of land under birta in the village leading over time to an expanding Kshetri community that rapidly acquired from the Jirel large parts of the irrigated lowland and bari land.The average land holdings in Kalang are larger than in the other case villages, but these differ between the social groups within the village.On average, Kshetri households now own around 4 ha, a Jirel family 2-2.5 ha, and other social groups have 1 ha or less.
Kalang used to be a remote village, where most households' subsistence economy was supplemented with off-season off-farm activities (e.g.wage labour, selling firewood, bamboo items and tourism) and many young men out-migrated (KA3, KA5).However, in the 1980s, a road to Jiri was constructed through the upper slopes (at 2400 masl) of the village.This created opportunities for land investment and the social groups with more resources (i.e.rich Kshetri and Jirel) gradually started to move upwards selling their irrigated khet land to other groups living in the lower area (poorer Jirel and Dalit).The Dalits have retained their traditional occupation as village blacksmiths but working now on a cash basis selling knives to the villagers and passing tourists.The Sherpa people actively use the forest in the higher altitude areas and have stayed there.
Kalang is a fertile area; 'this place is the best for crop production in the region' (KA6) a Jirel man said.The mid area is dominated by bari land producing maize, millet, wheat, potatoes and 'all vegetables grow here if you can cultivate them' (KA1, KA4).In the lower khet fields, rice, potatoes and mixed vegetables are grown.The livestock kept are mostly goats but also ox, cow and buffalos, and their manure is an important input in the cropping (KA1).There are abundant fodder resources in the surrounding landscape; grass and tree fodder are collected mainly from private idling land, while families living close to the community forest gather grass and tree fodder (mostly Khashru or Quercus cassura/obtusifolia) from the forest (KA3, KA4, KA7).Some families keep a small number of chicken and pigs for their own needs.Sherpa families also sell non-timber forest products (NTFPs) collected from the community forest and public land (KA4) (see Supplemental Photo 1).
The villagers are food sufficient for up to 6 months per year but this reflects a lack of farm labour and not of land ('how much can a single person collect/carry?' asked KA4 rhetorically).With labour constraints, most families have concentrated their cropping on land around their houses (KA7).More distant bari plots have gradually been transferred into idling land (KA1).This change of land use has happened during the last 10 years and today about 50% of the mid area of Kalang is idled.Several families reported that they now harvest and sell Utis (Alnus nepalensis) and Pinus (Pinus wallichiana) timber from these plots (KA3, KA5, KA7).Many also source most of the winter fodder for livestock from here, i.e. khasru (Quercus cassura/obtusifolia) trees.But as the perennial vegetation takes over the mid area, the wildlife moves in, which in turn makes cropping on the remaining bari fields increasingly risky (KA5).A farmer described how '[when] I graze my goats [-] the deer graze with them' (KA6).Others mentioned how porcupines feed on the growing crops and farmers rushed off to chase monkeys in the middle of the interviews (KA1).Some khet land has also been transferred into idling land.Although farming is hard, households continue to farm and don't want to sell their land (KA1, KA5, KA8) feeling that farming is still important and that their children and grandchildren will need it when they are too old to work in cities ('they would curse me [if I sold the land]!',KA1).
The traditional cash crop, potato, which everyone grows has not been that profitable to all HHs and they have started to explore new potential cash crops such as kiwi.Close to the road, rich Kshetri and Jirel families have started kiwi production, and two households are large scale (1.5 ha) producers.Several families have planted smaller areas of kiwi (KA1, KA3, KA5), hoping to expand as the early kiwi growers have done.A smaller number of households are dedicated to tea-(KA4), apple-(KA2) and shitake-(KA2) production.
The forest on the top of Kalang was made a community forest in 2006.This affected the Sherpa community, as their traditional collection of medical plants/herbs and bamboo and grazing livestock became limited and they lost a part of their key income source.Villagers living in lower Kalang seldom visited the community forest and collected their fodder on the private land and were not affected by the new community forest restrictions in the same way.
In sum, the road construction changed the land use dynamics of Kalang and today's land use changes in Kalang point in different directions, although the historical legacies of socially differentiated land ownership remain.Some households are 'stepping up' production through commercialisation and diversifying, for example, into large-scale kiwi production, even on bari land in the upper part of the village.This interest in kiwi production has opened up opportunities for marginal groups, e.g.Dalits and poorer Jirel, giving them access to the more productive irrigated farmland in the lower areas.Holding khet land increases the cropping options which might lead to commercialisation.The 'stepping up' route is, however, not open for all as it requires capital and land close to the road.The Sherpa community in Kalang has been affected by the community forestry program as traditional grazing and NTFP practices have become more restricted.Most Sherpa households, besides potato cultivation, therefore minimally collect NTFP.
The largest land area, however, is the wide belt of land between the upper kiwi farming areas (i.e.Upper Kalang) and the khet fields (i.e.Kalang Valley).This contains an expanding area of idling land and a decreasing area of cropped bari fields.Here, households are abandoning the idea of active farming and their 'giving up' has been accelerated by the increasing wildlife population.As in the other two Mid-Hill cases, farmers shift bari land into idling land, where timber and fodder are generating some income while the households continue producing for subsistence on land close to the homestead.These perennial tree crops make it possible for labour poor rural households to use their land and meet some food and income needs.The strongest agricultural opportunities seen presently in Kalang are from perennial crops such as kiwi, tea as well as timber and fodder.The official view is, however, that timber and fodder are not crops and should not be grown on farmland.

Chyasku -a mountain village with limited access to water
Chyasku sits on a plateau along a mountain ridge, surrounded by forest and bari fields on the terraced slopes.About half of the 240 Chyasku households are Tamang who settled several generations ago moving from Dolakha.These were followed by Dalits (now about 30 households) who came for the forest resources allowing the production of the charcoal required for blacksmithing and later by Newar households (now about 90 households).
Chyasku lies in a rain shadow and is drier than many other neighbouring villages.Drought is central to the village history with long memories of agricultural hardships.Many villagers recalled periods of hunger (CH3, CH5, CH6, CH9, CH10), especially a famine in 1994-95.Even now, households can only self-provision for 3 to 7 months of the year and off-farm incomes are necessary for survival.When the road network developed in the area in the 2010s, traditional portering incomes disappeared and the Chyasku farmers had to find other off-farm incomes.All interviewed Chyasku households listed waged labour in agriculture, house and road construction and industry as supplementary income sources to farming.The large earthquake in 2015 destroyed many of their houses.But house reconstruction with government assistance provided local work opportunities from which women also benefited (CH10).
Land holdings in Chyasku are small, ranging from 0.25 to rare cases of 2 ha of bari land.However, Tamang owns most of the more productive land with average land holdings of more than 1 ha, while the Newar also holds productive land but on average has holdings of only 0.3 ha.Dalits on the other hand have average holdings of about 0.5 ha, but these are on the least fertile soils and they remain the poorest group in the village.There is no irrigated khet land.Households, however, said that the red soil is productive for the traditional crops (CH5) such as maize, mustard, millet, buckwheat, different legumes and the recently introduced sosta bean.Farmers emphasised how the subsistence crops were reliable and always produced something even during years of drought and windstorms (see Supplemental Photo 2).
Livestock is one of the main sources of income and all interviewed households keep goats.Some households have small areas of cropland but also own pakho plots, which are areas of grass and tree fodder, and rely more on livestock rearing.One Dalit household owned 0.65 ha of pakho land in steep areas which provided fodder for their 14-15 goats and they sold 3-4 goats every year (CH5).Others have converted less productive bari for fodder tree cultivation, effectively idling the land and today 15-20% of the bari land is idle land.Accordingly, fodder is abundant in the landscape, but the lack of water restricts livestock numbers.The water supply became even worse after the 2015 earthquake, when several of the water sources dried up (CH7).
While subsistence crops are a 'safe card', more commercial crops are risky in Chyasku.They need a share of the scarce water for irrigation and crop inputs are expensive.A few Chyasku farmers have created pockets of intensified vegetable production (with tomatoes, chili pepper, potatoes, garlic, onion) using stored rain water and household wastewater.Recently, increased access to pumped drinking water has helped intensifying vegetable farming, but this water is expensive.They have specialised in off-season production as they cannot compete on price with the main season vegetables from the Tarai.An emerging practice by those who can afford to wait is to plant orange trees (CH1, CH2, CH4, CH7).Further, the water constraints have made the villagers focus on smaller livestock such as goats, pigs and chickens.Goat and pig farming have a long tradition in Chyasku, and the scale has increased in recent years (CH1, CH2, CH4, CH9, CH10).
The Chyasku villagers are deeply frustrated with the increasing wildlife populations as the 'deer graze as goats in the fields' (CH3), 'come 24 hours a day' (CH7) and the monkeys have become 'uncontrollable' (CH8, CH7).The wild animals thrive in the four surrounded community forests and in the growing idling area and invade the crops causing significant harvest losses.One farmer said, 'I stopped cultivating legumes and soybeans due to the attacks of deer and monkeys' (CH3).Another decided 2 years ago to switch from food crops to growing sweet-oranges (junar) due to wildlife eating his bean crop (CH1).Some of the Chyasku farmers estimate they only achieve around 25% of their possible maize harvest (CH12, CH15).The forest areas are rarely visited by many households but as a Tamang farmer put it, 'the person who feeds on daily wages will need the forest' (CH1).Most households continue to use wood for cooking together with recent adoption of liquefied petroleum gas.
In sum, farming in Chyasku offers poor returns given the constraints of rainfall, low crop productivity and high crop losses due to wild animals that affect all households, especially those having land in forest fringes, although there are underlying differences in land resources between the different social groups.While farmers are 'hanging in', they are at the same time de-activating land.However, half of the interviewed households have also bought more bari (CH6, CH 10, CH4) or pakho land (CH2, CH4, CH7, CH5) in the village during their lifetime funded through off-farm employment.Although these land purchases are very small, this emphasizes the subsistence imperative and the need to secure food supply even when the rains and work opportunities fail.The small but growing local market in neighbouring urban centres of Ramechhap and Manthali also provides opportunities for sale of small surpluses of crop and livestock products.Tree fodder is an abundant resource in the Chyasku landscape and may offer potential for commercial goat production if water constraints are solved.

Khimti: a village with dynamics of upland and lowland farming
Khimti is located in a narrow V-shaped river valley, with steep mountain slopes on both sides.The village centre is in the valley but on the surrounding forested hills, there are more than 10 satellite villages framed by terraced bari fields which form Upper Khimti.About 240 households live in Khimti village centre and 500 households in the surrounding upland villages.About half of all households are Kshetri, the second largest groups of villagers are the Brahmin (ca 100 HH) and Newar (ca 100 HH), Sunuwar (ca 60 HH), Tamang (ca 60 HH), and there are also smaller groups of Magar, Majhi and Dalit families living in Khimti.
The first people living in the lower Khimti valley were the Majhis, who were skilled fishermen living on the riverside (KH2).The Kshetri people were mostly settled in Upper Khimti but gradually bought khet land in the valley particularly after 1959 when birta was abolished from a Brahmin landlord who had held the land.When the Kshetri farmers first arrived during the 1950s, they settled in the uplands given the prevalence of malaria in the lowland.They established villages, rainfed bari fields and fodder areas there while also constructing irrigated khet fields during short stays (during sowing and harvesting seasons) in the valley (KH1, KH2, KH3, KH11).It was a time when most households experienced food scarcity and villagers would work in neighboring areas for payment in food (KH1, KH2, KH3, KH4, KH5).Once the malaria outbreaks were controlled in the 1980s, upland villagers started to move permanently into the valley by their khet fields.But the migration was rapid after the construction of the road in Khimti in 2000.More recently, increasing crop losses to wild animals and landslides have enforced a move out of the Upper Khimti villages.The arrival of the two Dalit families in the village is more recent as the first generation of migrants is still alive.Brahmin's remain the largest landowners in the valley owning on average about 0.3 ha of khet land and about 1 ha of bari land per household, but most of the bari lands are idled.The Kshetri owns on average about 0.2 ha of khet and around 1 ha of bari land, while the holdings of khet land by Majhi and Sunuwar people are much less on average at 0.05 ha, and they have almost no bari land (see Supplemental Photo 3).
Large parts of the uphill bari land are now not cropped, and mainly old people remain in the semi-abandoned satellite villages (KH7).In contrast, the village centre in the valley has a growing business centre and increasing land prices.The former upland farmers focus their farming on their small khet fields.They visit their upland fields and fodder areas when possible.Some try to find additional off-farm incomes through military service, teaching, and periods of international migration.Khimti's location on a flood plain is risky and a constant concern as villagers have experienced several floods.The latest was in 2017 when a flash flood washed away a significant area of khet land.One affected household was one of the Dalit families who lived on the riverbank.They lost all their land and belongings (KH5), and as a consequence they had to send one of the sons who was studying in grade 11 to Malaysia to earn cash income.
But there is still a commitment to keeping the upland.'Whatever happens you should not leave your land' said a Kshetri woman living in Upper Khimti (KH6).She argued that as long as one had land, one has a place to live and grow food for subsistence as a complement to remittance income.The upland production is focused on maize and millet and smaller amounts of potatoes and vegetables.Though the upland soil is not bad (KH6), water is scarce and harvests are poor due to wildlife damage.In certain areas, the level of crop losses due to monkeys and porcupines (KH1, KH3, KH4, KH10, KH11) is so high that the villagers have stopped farming completely as they 'cannot guard the fields [against wild animals] all the time' (KH10).The consequence of this agricultural withdrawal is self-reinforcing as bari fields become idling land and wildlife thrives in an increasingly forested landscape.The benefit is that there is no shortage of fodder or fuel wood for the upland households, limiting the need to extract resources from the community forest.
On the other hand, in lower Khimti, the irrigated khet land can give three harvests in a year and more intensive commercial farming is rapidly expanding in the valley.The number of hand tillers is increasing and an agro-vet shop has opened in the village.There are now commercial farms growing paddy, vegetables and rearing poultry and goats for the market.There is also more specialized production of dairy products, pigs, mushrooms, fish and poultry.One Brahmin farmer with 21 milking cows has built up a dairy shop (KH11).He grows grass on his khet land, but when he thinks about the possibility of expanding production, he has started to look towards the fodder resources in the uphill areas.Other semi-commercial livestock farmers are also considering moving into livestock production in the uphill areas because of fodder and grazing land availability (KH3, KH7).Some Khimti respondents were also experimenting with growing tree crops in the uplands, e.g.planting lemon, mango and jackfruit trees (KH3, KH6, KH7, KH11) and trying out monkey safe crops such as turmeric, garlic and ginger.
The khet holdings in Khimti are small (between 0.05 and 0.3 ha) and many want more land to farm but cannot afford to buy it.Instead, they enter into sharecropping (KH3, KH5, KH10) or bandaki (KH5, KH9, KH10) agreements.Bandaki is a practice of providing a loan with no interest to a landowner in exchange for using his/her land until the loan is repaid.
A Sunuwar lady described it as 'eating the interest from the production' (KH9).
Though the community forests in Khimti are expanding and encroaching on the uphill agriculture land, several villagers felt that the quality of the regrowth forest was poor.They said that 'there is nothing in the forest' any longer (KH3), and pointed to a decrease in species diversity related to pine tree dominance, seen by many as a low quality timber tree.With more fodder and firewood on their private land, they also said that they 'don't use the community forest [any longer]' (KH6).
So Khimti seems to have two divergent agricultural futures in its upland and lowland landscapes.Some 'stepping up' is happening in a range of commercial farming activities at a relatively small scale on the khet fields in the valley.At the same time, there is deactivation of land but households are 'hanging in' on the upland areas, where the upland bari fields are increasingly covered with forest and a few new fruit trees.But there is limited area in the valley to scale up commercial agricultural production.For farmers with livestock, the upland forest vegetation is the next fodder supply.Different agricultural futures seem possible for the two agroecological zones.While the valley land has possibilities to intensify production, the upland areas have the potential to produce fodder for growing livestock numbers (while supplying manure to the lowland).A ruminant focused production might develop in strategically located corners of the upland areas connected with basic infrastructure.However, traditional food cropping in less accessible parts of Upper Khimti will decrease, although less labour intensive fodder and fruit crops might be grown.But the constraint of farm size, particularly for Majhi and Sunuwar people, and increasing construction activities in the existing khet land will limit their agrarian futures in the valley.

Jhunga -a village with opportunities of intensification
Jhunga stretches as a thin strip between the southern side of the Kamala River and hills with sal (Shorea robusta) forest.It is hot and humid.Highly productive irrigated khet fields cover the area closest to the river, whereas border of rainfed bari fields and houses nestle on the forest edge.Of the 300 Jhunga households, two-thirds belong to the Danuwar community, the second largest group is the Majhi, and then there are about 20 households each of the Magars, Brahmins/Kshetris and Dalits.
Originally Jhunga was part of the Danuwar people's farming, fishing and hunting territory and they remain the largest landholders.The forest resources were rich, but endemic malaria made it a hard place to live.After malaria eradication in the 60s, Pahade ('hill persons') started to move in.The Pahade was Dalit and Magar people who were attracted by the forest resources, the possibility to buy or sharecrop khet land (JH4, JH8), to do wage labour on Danuwars' khet land (JH3, JH6, JH7) or work as blacksmiths in the case of the Dalits.The khet land along the river was highly productive but vulnerable to recurrent floods.In the monsoon of 1993, the Kamala River flooded and large areas of irrigated khet land were washed away.In one blow, most Jhunga villagers lost a substantial part of their best cropland (JH1, JH2, JH9).The time after is remembered as a period of hunger and extreme hardship (JH1, JH2, JH3, JH9) when the forest became key to survival.After the flood, the village relocated to the forest edge.
Recently, several households discovered that their land was registered in someone else's name and they have lost or risk of losing part of their farm land (JH1, JH2, JH4, JH6, JH3).This uncertainty over land ownership seems to be a result of unclear registration processes before and after the relocation of the village, an ongoing informal land trade and two muddled land survey processes.In 2020, the blacktopped Chure Highway 4 which runs through Jhunga was completed.With the new road the village started to develop as a local market hub and land is now 'more expensive than gold' (JH1) in Jhunga.
Both khet and bari land are used for subsistence farming in Jhunga, but not all households have both land types and not everyone has much farmland.The Danuwars are considered landlords who have on average larger holdings (0.7-2.7 ha), the Majhi have slightly smaller farms (average 0.7-1.35ha) and the other villagers have much less land (average 0.07-0.5 ha).The khet fields produce 2-3 crops per year of rice, maize, wheat, mustard, potatoes and other vegetables.The bari land gives one to two crops of maize, mustard, millet and lentils.The sand banks are also used for growing maize or grazing.
However, given the high land productivity, a very small area can provide a significant contribution to a household's food budget.Even the land poor see subsistence farming as an essential foundation to meet migration ambitions.A Dalit women explained that 'if the persons have 2-4 kattha (0.07-0.14 ha) of land, then they can send their family member abroad, as they could [then] save some money' (JH8).She reasoned that with no subsistence production, all remittance money would be spent on food.This would not allow any saving for buying land or other assets that could improve the household's longer-term prospects and make the migration effort remunerative.The need for cash to meet health and education costs has led many families to send someone abroad to work.One Majhi woman said (JH2), 'I want my husband to go for income because we need cash for health, education and festivals.I and my sisters-in-law can do agriculture'.Another Danuwar family also wanted his sons to work overseas to provide future health security and pay back the loan that was taken for past health problems (JH9).
Raising cows is the activity of choice and most households typically have 4-6 cows and some even sharecrop the rearing of them (JH2, JH7).In second place come buffalos and thirdly goats.Whereas the Brahmin/Kshetri farmers collect fodder in the forest and carry it back to their stall-fed animals at the homestead, Danuwar and Majhi households entrust their cattle to specialised herders.In contrast to the other study villages, Jhunga has no idling land and there is insufficient fodder for the number of livestock in the village.With the livestock herding and collection of fodder in the forest, wild animals normally keep away and Jhunga does not experience the wildlife damage found in the other villages.But at times elephants come to eat bananas and trample crops (see Supplemental Photo 4).
Intensive paddy production in khet areas has long been an established cash crop.Increasingly, power tillers are used for land preparation and threshing machines at harvest.Some households grow vegetables for sale, and many more say that they would like to do so if they had irrigation on their land.There are many stories in Jhunga of intentions to secure income from livestock.A Kshetri lady reported how for a long period she had tried to do this from selling milk and curd from her buffalos.However, the buffaloes always died 5 and after 17 dead buffalos she finally gave up (JH4).Today, cattle are seldom profitable and households are mainly focused on poultry, although pigs and ducks are described as future livestock possibilities.
To conclude, people moved to Jhunga in search for land, but many lost a large part of their land from a major flood in 1993.More recently, several households have faced severe anxiety over the loss of their land due to claims of invalid ownership papers.As with the other three sites, there are differences in the size of land holdings by social group and highly productive and commercial production of rice and vegetables compete with raising livestock.For those with mid-and larger sized farms, Jhunga seems to have a quite bright commercial agrarian future ahead.Its location close to a larger town (Bhiman and Sindhuli bazar) provides a broad range of opportunities to sell agricultural products and the new highway facilitates such trading.The highway has rapidly changed village-life, opening up opportunities for new non-agrarian incomes within Jhunga, e.g.tea shops, hotels and land speculation.Such more capital-intensive land uses are starting to out-compete the smallscale farming on the fertile land.However, once again, this route is not open to all.The poorest villagers have not been able to acquire or access sufficient land nor start commercial cropping and will not benefit from the road either.They are dependent on periods of agricultural wage labour and forest resources for livestock rearing, and they observe this development from a distance, like respondent JH7 who has never in her life been paid in cash but only in food.

Discussion
The findings from the four sites and their diverse location specific land use changes underscore the importance of taking account of landscape position in a mountain economy in understanding agrarian change processes.But also evident from the village cases in the mapping of social identities to particular locations and land use practices is the enduring relevance of the history of land settlement and caste structures to spatial patterns of change.In Kalang and Khimti where a high caste individual had held land under 'birta', farm size remains sharply differentiated by social identity.But even in Chyasku without that history of 'birta', Dalits at the bottom of the caste hierarchy are to be found on the most marginal land.Yet these are also socially dynamic landscapes and in three of the sites -Kalang, Khimti and Jhunga -, infrastructural development such as roads or the control of malaria have led to movement of social groups within a village landscape, accompanied at times by in-migration of others.
As Jodha (1992) emphasized, mountains have core features of inaccessibility, fragility (or risk), marginality and heterogeneity.These in turn give rise to niches of opportunity and human adaptation mechanisms to these features.The contrasts between the four sites we have examined speak usefully to Jodha's core features.
Jhunga in the lowland (inner Tarai) provides the point of comparison with the other three sites.It is more homogenous agro-ecologically and is highly productive although historically inaccessible because of malaria.That has changed and with a new road, market access has dramatically expanded, reducing its physical marginality, and fueling a commercial economy and intensive use of all agricultural land.Yet it does not entirely evade mountain effects given the risks of flooding from the major river that flows next to the village.On the other hand, the mountain site that is comparable to Jhunga in terms of landscape features (relative flatness) and uniformity, is Chyasku.Located at 1600 masl without access to irrigation, lying in a rain shadow and drought prone, it is an agriculturally marginal area.It has few niches of opportunity and there are severe limits to adaptation possibilities.
The other two sites of Khimti and Kalang bring out the significance of altitudinal landscape transects from low land to highland (625-1500 masl) and the consequences in terms of a mosaic of land use practices.While the current locations of more marketoriented agriculture in both sites were inaccessible in the past, the control of malaria in Khimti and the building of a road in Kalang have shifted the priority location for agriculture.In the case of Khimti, the upland has been de-emphasized for the more productive lowlands leading to resettlement in the river plain.In contrast, a new road upslope in Kalang has favored commercial cropping leading to the deemphasizing of the more productive irrigated land at the bottom of the hill and into which previously marginal Dalits and Jirels have bought into.Yet this shift has only been possible because of the combination of the road and sufficient rainfall to support cropping without irrigation.In sum, as Sugden et al. (2018) noted agrarian change processes are strongly mediated by local agro-ecological contexts.
Patterned across these four contrasting sites are a set of common land use practices, although the significance of each depends on location.There are high potential locations with differences in extent and underlying causes of their potential in three of the sites -Jhunga where a commercial crop production is widespread, and Khimti and Kalang where it is a more niche specific and restricted in scale.These are places where 'stepping up' is possible (see Dorward et al. 2009) although it is restricted in scale and commitment by farm size and the need for households to meet simple reproduction needs.A move towards a more market-oriented farming can be seen in both small-scale livestock production, commercial vegetable production and kiwi growing.All efforts to expand production focus on productivity, use low input levels, e.g.urea, hybrid seeds, and pesticides but there are increasing constraints of water, wildlife damage and labour shortage.In Chyasku the rains failed for six successive years 2012-2017.In other places, intense storms lead to flash floods and loss of cropland as seen in both Jhunga and Khimti.
Common to Khimti, Kalang and Chyasku is the coexistence of two distinct land use practices on a single farm.Some land is allocated for simple reproduction needs but only a small number of households meet their full subsistence requirements from this (e.g. in Jhunga).This is combined with land deactivation whereby previously cultivated land, almost exclusively bari land, is no longer cultivated.This is not necessarily an abandonment of land but rather a de-activation or idling of land, particularly by social groups with more land (e.g. the Khsetri in Khimti and Kalang), reducing the intensity of use and diverting land use to less labour-demanding production of fodder and trees for timber.There is a paradoxical outcome where meeting simple reproduction needs is a core objective of farming that few households achieve but at the same time less productive land is being deactivated or idled.
The explanation is to be found in the lack of labour to meet the labour intensity of mountain farming given the limits to mechanization.But the continuing allocation of a portion of the farm area to meet subsistence requirements also points to the insufficiency of returns from off-farm incomes (labour and remittance) to ensure even simple household reproduction (see Rigg 2020).This de-activation, in turn, has created new risks.With the expansion of forest area and the re-treeing of farm landscapes there has been an increase in wildlife populations leading to serious crop losses (Bista and Song 2022), contributing to further land deactivation.Hunting wild animals is prohibited by the Nepalese Forest Act (MoFE 2019), which means that farmers have no legal means to control them.This all means that there is an ongoing squeeze on simple reproduction (Bernstein et al. 2018) driven by a combination of unequal access to land, declining farm sizes, labour shortage, wildlife encroachment and insufficient returns from precarious off-farm work both overseas and in Nepal's towns (see also Rigg 2020).It is happening across all the sites, particularly for the land poor accounting for why there is no significant process of deagrarianisation.Households are not 'stepping out' and nor can one find the emergence of a subset of larger farms.Households will not sell land as the evidence makes clear although they will buy scraps of land when they can, but few are completely landless.These are households caught in a poverty trap, unable to leave agriculture and move into a more secure economy and they are constantly searching for marginal opportunities that will allow them to 'tread water' (Sunam and McCarthy 2016).As Rigg (2020) observes, many rural households are squeezed by having to stay but need to get out in order to survive, with few opportunities to prosper.While these are clearly rural households by virtue of residence, most can scarcely be called farming households given the limited contribution that farm production provides for household survival and the necessity of off-farm income to live.
In common with much of South Asia (Rigg 2020) there is little evidence to support the emergence of a classic form of agrarian transition in Nepal.The specifics of a mountain economy offer little opportunity for widespread land consolidation and the emergence of an intensive market oriented agricultural economy.There are claims that the occupational multiplicity found in Nepal's rural economy offers a route to a stable future (Chhetri et al. 2023).But much will depend on the ability of a remittance economy based on the overseas migration of semi-skilled labour to support it.The agrarian question as Bernstein (2006) has suggested is now much more a question of labour and of finding sufficient decent work to support the simple reproduction of the household.In this sense, it is primarily the household rather than land that needs to become the central unit of analysis since it is the deployment of labour that is central to a household's persistence and viability.The micro-level changing contours of land use in Nepal's marginal mountain economies are best understood with respect to how household labour is allocated, both on-farm between land of different qualities and between on-and off-farm and the extent to which time bound labour for farm work may constrain the search for off-farm work.

Conclusion
The findings from this exploration of land use change in four contrasted landscapes in Nepal evidence a continuum of different land uses and labour use intensity.These range from intensive 'stepping up' to extensive practice of 'hanging in', related to the possible outputs and yields, position in landscape and exposure to wildlife.But all interviewees practice semi-subsistence farming to ensure household reproduction even if remittance incomes are a major part of the rural economy.Classic agricultural development is not even a remote possibility as a route out of poverty for many in the Mid-Hills of Nepal.There are valley areas with potential for intensive commercial agriculture (as in Jhunga and Khimti valley) with some agricultural future but this is less clear in more marginal mountain landscapes such as Chyasku and Upper Khimti although these households will not abandon ('stepping out') their land.
For many poorer households, overseas migration is not possible and farming will need to be combined with off-farm work in a poor rural economy.A sufficient market-oriented economy to support households is likely to be highly site-specific.It is possible that more productive, labour extensive uses of land that is currently idled or deactivated can be found as in Kalang.This might include payments for the re-treeing of farmland as a contribution to the provision of environmental services or a livestock economy.There may be localized opportunities from road construction and irrigation development in the hills.This might lead to reinvesting in currently idled land.But the greater likelihood is that Nepal's rural households will continue to balance their labour allocation between securing a subsistence contribution from land and working in insecure off-farm activities.
But this fine-tuning of agricultural practices to balance land and labour resources with household reproduction needs may simply be treading water to stay in the same place.In some respects, the historical outmigration of Nepalese before the 1950s points to the fact that 'land ended' a long time ago.There has been an enduring household reproduction crisis (see Bernstein 2010) that has stretched over decades.In some respects, through labour migration and petty commodity production, households have been drawn into a modern capitalist economy but under deeply disadvantageous terms that reinforce precarity.In turn, this requires efforts to squeeze the best out of what resources of land and labour households have.For a few, by virtue of location in the complex geography in the Nepalese mid-hills, there may be the possibility to do a little better.But for most, there can be no agrarian transition as they are trapped in liminality, both in and out of modernity (McCarthy 2020).

Notes
1. Inner Tarai is the elongated valley between Chure and Mahabharat hills of Nepal.The climate and temperature are similar to Tarai (i.e.sub-tropical), which is the southern flat land of Nepal.2. Jirel is one of many indigenous groups in Nepal which are collectively denoted as Janajati, Dalit refers to the lowest caste communities which were categorized as 'untouchables' in the past.
3. Birta was the land granted to key officials by the rulers in lieu of salary on an inheritable and tax-exempt basis (Regmi 1978), these grants were terminated by the Birta Abolition Act in 1959.The lands were converted into private property or raikar after the implementation of this Act.4. The Madan Bhandari Highway. 5. Unclear from what disease but probably from foot and mouth disease.

Figure 1 .
Figure 1.Map of Nepal and case villages' locations.

Figure 2 .
Figure 2. Illustration of land use intensity in relation to altitude in case villages.The filled areas indicate the main land uses, the shaded areas show emerging land uses and the arrows appoint the directions of change.

Table 3 .
Life story informants (coded) mapped by site, social group and position in the caste hierarchy.

Table 2 .
Methods and data collection.

Table 4 .
Categories of agricultural land use.