Smallholders in Mozambique call their fields machambas. Machambas constitute a key source of income and food to most smallholders in this region. Definitions of what constitutes a smallholder are manifold, but in the majority of cases, the term relates to the land area available for production. As such, “smallholder” is an umbrella term for diverse modes of production and cannot be easily defined, neither across continents, nor regionally. The characteristics of machambas are as diverse as the people cultivating them. This blog post provides a handful of impressions of the machambas in Northern Mozambique to illustrate this diversity. The material presented here was collected by a team of five MIDLAND researchers during a field trip across the three provinces of Niassa, Zambezia and Nampula in mid November 2021.

November is the very end of the dry season, a time when soils are extremely dry due to low rainfall over several months. In this period, smallholders commonly start to prepare land for the coming growing season. Overall the landscape was characterized by exposed dry soils, and road dust covered huts, cars, animals, and people likewise. Amidst the dry landscape, we saw patches of lush green alongside rivers and lakes.

Wetlands in drylands as seen by a drone.

These “wetlands in drylands” can be used for cultivating crops year-round, including water-intensive fruits and vegetables suitable for commercial activities. These lands were particularly fragmented, which indicates that these moist and fertile areas are shared among many community members.

Visiting the foothills of Mount Namuli, Mozambique´s second highest mountain, we observed that machambas covered the steep hill slopes, at elevations of 1,600m and beyond. These machambas are totally inaccessible to machinery or heavy equipment. Farmers climb the slopes to their fields carrying their tools, sometimes several hundred meters above their village.

Prepared field on the foothills of Mount Namuli.
Photo: Philippe Rufin

We visited a few sites bordering the Mecubúri Forest, Mozambique´s largest forest reserve. Here, cultivation requires manual clearing of trees and shrubs. Land is managed extensively, and fields are larger compared to the more consolidated regions nearby agglomerations. Farmers make use of fire for clearing land from remnants of vegetation and distribute seeds on the cleared areas. While remnants of the last harvest were still visible, we did not find signs of land preparation such as tillage. 

Traveling east towards the city of Nampula, the provincial capital, we noted dry and very sandy soils, appearing to be void of organic matter. Cassava is a key crop grown here, likely due to its´ drought tolerant nature. Moreover, cashew trees are scattered in the landscape, creating a characteristic landscape with plentiful of shaded spots. 

In sum, the machambas we encountered were highly diverse. They come in different sizes, with different quality, accessibility to markets and cities, and biophysical factors. These factors influence which crops are grown by smallholders and how land is managed, where and when. Some smallholders, especially in areas with greater connectivity, are able to generate production surplus, cultivate market-oriented crops, and sell them to local markets. Local cultures, customs and traditions also shape crop choices and the land management practices adapted to the cultivation of such crops. Who manages which crops and how tasks are distributed is also determined by gender, age, and roles of individual family members. Land is commonly shared amongst family members, and parcels are further subdivided and distributed to the youngest generation. Smallholders mostly use family labor and low-tech tools for land preparation. Some have access to improved seeds that can be bought in first or second generation.

Smallholder tools: hoe, axe, and knives.
Photo: Philippe Rufin

Small-scale producers in Northern Mozambique manage to achieve so much with so little, given the challenging conditions they’re exposed to, such as dry soils, uncertain rainfall, difficult access to water, absent or low use of fertilizers, and the very small and fragmented fields, often located in long distances from households and placed in inaccessible terrain. It was remarkable how heterogeneous machambas appear within small geographic space. When one is able to recognize the great diversity of farm systems (in terms of soil, crops, farm sizes, topography, climate), it is hard to conceive that untailored one-recipe advices aimed for rapid yield increases based on standard energy-intensive packages of fertilizers and pesticides could work equally well in such diverse contexts, especially when those may degrade the land in the long-term. Local farmers have a vast and deep knowledge about their territories, that has curated specific practices adapted to the local conditions over generations.

We left Mozambique wondering about the adequacy of decontextualized top-down development policies for such wide-ranging settings. The diversity of contexts and knowledge we encountered seemed instead an exceptional opportunity to co-create alternative pathways for decent livelihoods that could offer a diversity of crops for food security and resilience, and don’t compromise the sources of livelihoods for future generations. After discussing with several farmers, it became evident that a necessary element for real food system transformation would have to be horizontal learning, where knowledge is co-produced, reciprocal and collective. A dialogue of wisdom, where practical knowledge is combined with scientific and political knowledges, and where farmers are involved in both research and policy design of and for their own territories, seems not only suited to improve agricultural outcomes, but also to decentralize agricultural development and increase the political participation of local communities and networks to influence current agricultural policy paradigms.

Written by Philippe Rufin and Cristina Chiarella

Challenges in emerging frontiers

Our work on Northern Mozambique starts from the premise that this region might constitute an emerging frontier for land use investments and expansion. Before digging into the question of whether this premise is verified, let us discuss why we focus on “emerging frontiers”. 

A large share of current agricultural expansion and deforestation takes place in so-called “frontiers” – i.e., regions with rapid development of natural resource exploitation and land use changes, concentrating migration of smallholders as well as large-scale corporate investments. In 1893, Frederick Jackson Turner, American historian, wrote “The significance of the frontier in American history”, describing the expansion of the frontier and the rolling back of wilderness as an attempt to make livable space out of an uncooperative nature (see the Figure). Since then, intensive research efforts have been focused on describing and explaining the functioning of active deforestation frontiers in relation to agricultural expansion. A major conclusion of these works is that even though public and private governance interventions can indeed have some effectiveness to reduce deforestation and foster more sustainable land uses, these corrective measures are often hard to implement, face multiple obstacles and vested interests, and often remain fragile, contested and reversible. And even more importantly, a lot of environmental and social damage done is hard to reverse. 

“American Progress” (John Gast, 1872). Painting in the public domain. Obtained from

In contrast, research has been much less developed on unravelling how these frontiers emerged from territories considered as marginal in terms of agricultural productivity and connections to global markets. In the MIDLAND project, we aim to explain the processes that condition and shape the emergence of agricultural frontiers in territories considered as marginal in terms of agricultural productivity and global market connections. Much of this work is focused on the Miombo dry woodlands of Southern and Eastern Africa, with a more specific focus on Northern Mozambique. How and from what processes frontiers emerge in the first place, how and why is it that some places shift from being seen as “marginal”, remote, uninteresting, to becoming targets of interests and investments? Similar frontiers are emerging or re-emerging in other places, like the Paraguayan Chaco and Chiquitano (le Polain de Waroux 2019, Henderson et al. 2021), or boreal regions (Meyfroidt 2021), often overlooked by the international community of civil society, large NGOs, and international institutions. We build on a core assumption that steering the course of frontiers towards more sustainable outcomes might be easier when these frontiers emerge compared to when they are fully active, rampant frontiers operated by powerful actors. Of course these regions, like the North of Mozambique, already face critical sustainability challenges in terms of food security, livelihoods, environmental degradation and others, but prospects of large-scale commodity frontiers would add up a whole new dimension to these challenges.

Yet, an important caveat is that often, there’s good reasons why the abundant natural resources in these places have not triggered frontiers earlier. Remoteness, lack of infrastructures or other forms of agglomeration economies (market access, skilled labor, service providers…), unreliable investment contexts, lack of public authorities means, political instability or conflicts, critical barriers in agroenvironmental conditions, are all kinds of reasons that can typically explain why frontiers have not – yet – emerged in some areas. But these same factors can also hinder efforts towards governance towards sustainability. Active frontiers as in the Brazilian Amazon or Cerrado experience large-scale deforestation and high social impacts, but are also the places where innovative public-private governance mechanisms are developed, new monitoring tools and civil society mobilization trigger supply chain actors to take strong commitments to improve their sourcing, and detailed datasets allow for methodologically robust evaluation of policies’ impacts. 

With rampant poverty, increasing pressure on land, continuing deforestation, the status quo is not a desirable option in places like Mozambique. The challenge is thus about ramping up research efforts, datasets and monitoring tools development and mainstreaming, institutional strengthening, and policy responses, along with investments in agriculture and other economic activities. As team member Dilini Abeygunawardane often says, the challenge is about “attracting the right kind of investors, and investments”, to address sustainability challenges. 

References (see other blog posts and the Publications page for many more references on these issues)

Henderson, J., Godar, J., Frey, G. P., Börner, J., & Gardner, T. (2021). The Paraguayan Chaco at a crossroads: drivers of an emerging soybean frontier. Regional Environmental Change, 21(3), 1-14.

le Polain de Waroux, Y. (2019). Capital has no homeland: The formation of transnational producer cohorts in South America’s commodity frontiers. Geoforum, 105, 131-144.

Meyfroidt, P. (2021). Emerging agricultural expansion in northern regions: Insights from land-use research. One Earth, 4(12), 1661-1664.

Mapping large-scale tree plantation expansion and land use change trajectories in Northern Mozambique

This blog post is based on Bey et al. 2021 and this Twitter thread

In this study we present a remote sensing technique to map tree plantations with high accuracy in a challenging context (Miombo dry forests, little spectral & phenological differences between plantations and natural forests).

We show that in Northern Mozambique, ~70% of large-scale tree plantation expansion between 2001-2017 occurred on cropland, the remainder on natural forest and grasslands; suggesting stronger tradeoffs with livelihoods than with nature conservation.

Comparing our maps with government cadastral records, ~40% of plantation expansion occurred on lands not legally designated for this land use. Relying on cadastral records only for assessing drivers & impacts of land use change (or LSLA), and policy impacts, can be misleading.

The maps are available online in earth engine, they also cover other land use/covers (cropland, natural forests, grasslands) in the four northernmost provinces of Mozambique over four time periods: Link to interactive map in Google Earth Engine:

Theories on Long- and Short-Run Forest Dynamics: Forest Transition, Environmental Kuznets Curve, Ecologically Unequal Exchange

This blog post is based on Rodriguez García et al. 2021 and this Twitter thread

Many works assess Environmental Kuznets curve for forest; but crucial is to understand mechanisms linking economic development and de/reforestation (governance, changing economies, technologies, displacement..?). We assess this in this study using a co-integration approach on a long-term cross-country panel dataset.

Some insights: Over the long run, there are dynamic equilibrium relationships between forest cover area, economic development, agricultural area and rural population density. In other words, there are feedback mechanisms (to clarify) which, when forest cover tends to drop very low, tend to bring forest cover towards a dynamic equilibrium with these other variables. Does not mean automatic or guaranteed!

Agricultural intensification generally does not significantly influences country-level forest cover trajectories, on net (but links to agricultural area, and can hide crop and region-specific effects – see ).

For middle-income and pre-forest transition countries, improved governance positively affects forest area when the latter is low, i.e., good governance is key in situations of forest scarcity.

This approach, using cointegration to disentangle long and short run causal effects, and assessment of direct versus indirect effects, can be used for other similar questions (e.g. ).

Can strategic land-use planning contribute to the improvement of land governance in tropical, low-income regions?

This blog post is based on Oliveira and Meyfroidt 2021 and this Twitter thread

Land use plans and strategic spatial plans have been explored in specialized literature as the outcomes of separated planning processes. In fact, to date, few studies have devoted attention to the relationship between land use and strategic spatial planning processes. However, both can contribute to achieving sustainability objectives, including regulating the location, timing, and form of development.

In this study, we specifically focused on rural areas of the tropics. We explored how a closer interrelationship between land-use planning and strategic spatial planning supports the broader effort of understanding – strategic land-use planning – as a land governance instrument in tropical and subtropical regions, from humid forests to sub-humid or savannah and semi-arid landscapes.

We explored this relationship through a systematic literature review centered on identifying the instruments used within strategic approaches to land-use planning and how these instruments have been used. Our goal was to understand better how strategic land-use planning instruments can contribute to improving land governance in tropical landscapes.

We find “land-assessment instruments” such as “crop suitability surveys”; “land-zoning instruments” such as “agro-ecological zoning”; and “participatory instruments” such as “joint meetings between private interest groups and local communities”. The overall objective of this set of instruments is to support the decision-making of various land users, mainly farmers, regarding crop suitability or the identification of agronomic conditions as well as supporting land-use conflict resolution and better management of land-based resources.

Overall, in this review, we emphasize that to improve land governance in tropical landscapes, both planners and policy makers require reliable data regarding the existing land and its potential. We conclude with a summary of the mutually reinforcing perspective of the identified strategic land-use planning instruments and linkage to the main gaps, comprising the agenda for future research.

This exercise reinforces our initial claims that strategic-oriented, land-use planning processes have the capability of responding to current global land-use challenges in the tropics, thus supporting socioenvironmental governance.

How do successive waves of pioneers and investors build the conditions for land-use frontiers to emerge?

This blog post is based on Kronenburg García et al. 2022 and this Twitter thread

Thick ethnographic study in Niassa (Mozambique). Niassa is often described as the “forgotten” province of Mozambique – large, remote, sparsely populated, poor; but has also been “marketed” by the government as a land of opportunities with abundant natural resources and available land.

We trace the successive waves of actors that arrived in Niassa with different backgrounds, motives and business practices to establish farms or plantations – missionary farmers, white South African commercial farmers, large-scale forestry plantations. All repeatedly failed or have been struggling a lot, but leave sediments – legacies – that add up to gradually build the conditions for a frontier to emerge, and to give rise to a new wave by actors from within the region.

This new endogenous wave builds on legacies including social networks, financial capital, brownfields, legible land tenure, supporting institutions and policies, ways of dealing with land conflicts, and inclusive and diversifying business approaches.

What will happen next? The context in Northern Mozambique remains extremely difficult for both subsistence and commercial activities, navigating the trade-offs between deforestation and environmental concerns, livelihoods & development requires going beyond simplistic views and also investigating in a more nuanced way the perspective of these “investors”, often caricatured.

Theoretically, here’s what we argue in the final bits of the Conclusion..:
“… over time, through what may appear as stagnating regions, failures, or boom and bust cycles, legacies may accumulate to gradually change the conditions for a frontier to emerge (…) What previous frontier theories (…) would tend to consider as “noise,” or small irregularities largely irrelevant for the broader frontier pattern, may in fact constitute key dynamics to understand the mechanisms of frontier emergence (…) Gradual accumulation of legacies often remains unnoticed, and can (…) explain sudden transformations and seemingly surprising non-linear land-use transitions or regime shifts when these conditions overcome some tipping points…”

Resource frontiers and agglomeration economies: Understanding the diversity of investment logics in Southern and Eastern Africa

This blog post is based on Abeygunawardane et al. 2022 and this Twitter thread by @dilini_abey

We looked at transnational agriculture and forestry investments in understudied frontiers in Southern and Eastern Africa across Mozambique, Zambia, Tanzania, and Ethiopia. Our sample covers ~11% of the investments recorded in the four target countries.

We reconstructed the underlying logics these investments in a Bayesian network with firm and actor level interview and spatial data to assess where these investments are likely and how these vary across heterogeneous investors. We compiled a unique data set based on extended hours of interviews with managers including the top-level decision makers such as CEOs, CIOs, and managing directors often deemed unapproachable both at the individual and firm level.

We find that investment logics, characterized as a trade-off between resource frontiers and agglomeration economies, varied by investor’s related-skill sets and existing market reach. Experienced investors focusing on high-value crops invested preferentially in remote subsistence frontiers, seeking land with specific agroecological conditions. In contrast, newcomers – such as the typical speculators – focused more on proximity to infrastructures and markets. Investors with extensive skill sets and market reach expand the land use and deforestation frontier but are also likely to better survive the high transaction costs of operating in unfavorable markets.

Overall, we emphasize the importance of avoiding overly simplistic narratives, and we highlight the diversity of assets, skillsets, motivations and logics of investors in agriculture and forestry in Southern and Eastern Africa.

Complex relations between agricultural intensification and land use change: Assessing induced intensification, land sparing and rebound effect

This blog post builds on Rodriguez García et al. (2020)

Agricultural intensification can allow sparing land for nature, but it can also drive further expansion of cropland, i.e. a rebound effect. Conversely, constraints on cropland expansion may induce

We used an innovative cointegration approach to assess long and short-run mutual causation relationships between changes in cropland area and intensity, using a global cross-country panel dataset over 1961 to 2016.

We showed that in the short run, intensification resulted in a rebound effect for commodities with high
price-elasticity of demand, including rubber, flex crops (sugarcane, oil palm and soybean), and
tropical fruits, and for crops together in middle-income countries, which include many key agricultural producers strongly competitive in global agricultural commodity markets. These rebound effects remained over the long run for key commodities such as flex crops and rubber. Furthermore, intensification in low-income countries, driven by increases in total factor productivity, was associated with a stronger rebound effect than yields increases.

In contrast, intensification of staple cereals such as wheat and rice resulted in significant land sparing, and over the long run we found support for the induced intensification thesis for low-income countries.

Our study design, building on cointegration, error-correction models and tests for long-run causality direction could be used for addressing other complex long- and short-run causal dynamics in land and social-ecological systems.

Mapping smallholder and large-scale cropland dynamics in an emerging frontier of Mozambique

This website needs some more activity, and as we are step by step publishing and publicizing outputs from the MIDLAND project, it seems time to kick-off this blog section.
In a first series of blogs, we will compile info on key outputs of the project already published, and which were synthesized on Twitter.

So, let’s start with our work on mapping smallscale and large-scale cropland by remote sensing in a test study area in Mozambique, serving our aims to make progresses toward improved understanding on land use change and deforestation, food security, land acquisitions…
This blog post builds on Bey et al. (2020)

In that study, we focused on Gurué District (5606 km2) of Zambezia province of Mozambique, one of many countries in the region that has experienced a recent growth in foreign investments in agriculture through large-scale land acquisitions.

We built on open remote sensing data and tools including Collect Earth software and Google Earth Engine. We explored different compositing techniques, to show that median compositing performed best for disaggregating cropland by field size. Drawing upon Landsat data and using spectral and textural features, we characterized land use change over three time periods, 2006, 2012 and 2016, focusing on changes between small-scale cropland, large-scale mechanized cropland, and other land uses.

We showed that over 2007-2017, the largest source of change was small-scale cropland expansion. The gross area of small-scale cropland expansion (1214 km2) is approximately double the area of stable small-scale cropland (660 km2), and much larger than the large-scale cropland expansion that reached ~58 km2.

This method can be upscaled and applied in many parts of Africa with similar historic image availability challenges, and similar economic contexts with great disparities between small-scale unmechanized cropland and very large-scale mechanized cropland, to explore land consolidation dynamics and agent-specific pathways of land use change.

2021 Fieldwork in Mozambique

(Photo: Cristina Chiarella)

In November 2021, the MIDLAND team (Cristina Chiarella, Michelle Picoli, Philippe Rufin, Yara Ubisse, Patrick Meyfroidt) collected drone pictures of fields in Northern Mozambique, and exchanged with many farmers about their farming systems and livelihoods.

The team went from Lichina (Niassa province) to Gurué (Zambezia province) and Nampula and the Mecuburi area (Nampula province). The data will serve as reference data for mapping field sizes across the larger region.

The drone imagery shows the beauty and diversity of smallscale farming: