Report: Yemen’s agriculture in distress

Summary

Agriculture is the lifeblood of the Yemeni economy and its culture. However, our assessment of environmental change during Yemen’s armed conflict reveals severe agricultural distress that is widespread throughout the country. Our analysis of the potential drivers of these changes indicates that, for the most part, they have occurred because of factors linked to its conflict. In many cases, the impact of these factors has been amplified by historical policies, especially around water access, which have increased vulnerability.

Yemen is food insecure, and many Yemenis face famine conditions. Because of this it is important that urgent steps are taken to limit further losses to its agricultural sector. The COVID-19 pandemic has reduced humanitarian assistance to Yemen, impeding access and reducing donor funding. Protecting and restoring Yemen’s agricultural sector in an economically, environmentally and culturally sustainable way will be critical for increasing the resilience of Yemen’s people.

In this report, Eoghan Darbyshire uses open-source datasets to show the agricultural areas in distress and investigate to what extent the conflict has been responsible for the deterioration.

Key findings

• 257,000 hectares of cropland is exhibiting signs of distress, approximately equivalent to the total cropland in Jordan or Lebanon. Particularly hard hit is the southern Tihamah (near Hodeidah), Jawf, the southern deltas (near Aden), the upper Hadhramaut valley, the northern highland plains (near Sana’a), and Marib.
• The distress has led to a reduction in food security because of a loss of livelihoods and, in remote areas, food for subsistence.
• Key conflict related drivers include: direct attacks to farms and agricultural infrastructure, the economic war and war economy reducing access to water, agricultural inputs and markets, and the collapse of governance.
• The immediate outlook is bleak in the face of locusts, floods, cumulative damage, and the COVID-19 related loss of humanitarian assistance.
• When the security situation allows, agricultural recovery programmes must ensure that they are environmentally sustainable, and provide increased productivity, food security and secure livelihoods.

Contents

Key agricultural areas in distress

Relationship to food security

Drivers of agricultural stress

    1. Armed violence

    2. Population dynamics

    3. Access to water

    4. Extreme weather

   5. Pesticides and fertilisers

   6. Pests

    7. Changing agricultural practises

    8. Governance failures and institutional collapse

Closing remarks

Appendix

Key agricultural areas in distress

In order to assess the location and areas of cultivated land under distress, we created a simple distress index. The index captures where there have been significant biomass losses between the pre-conflict period (2009-2013), and the period during the conflict (2014-2019).¹ The index incorporates changes in both total-biomass and relative-biomass, which accounts for inter-annual variability in the weather. Where the disparity between the two is particularly large – this may indicate water resource stresses.

Our approach in comparing two long time periods is, of course, a simplification and there are changes within these periods that are not captured. However, we have not pursued a more granular approach for the country as a whole as there are very different and complex narratives in each location – as is evident in our accompanying report into agricultural decline in the Tihamah. Nonetheless, this approach is successful in capturing the largest and most significant agricultural areas in distress, and allows us to investigate country-wide drivers of degradation.

This report shows the agricultural areas that are in particular distress, with the index for each area illustrated alongside the data used to calculate it. Where relevant, the way in which relative-biomass has declined – either as a year-on-year trend, or a step-change in a particular year is shown.² In general, our results from satellite analyses agree well with official statistics from the Yemeni Ministry of Agriculture and Irrigation.

A Yemen-wide analysis of vegetation dynamics is available in the appendix, in addition to an environmental baseline assessment, and explains how the agricultural biomass losses this report explores are occurring in-spite of CO₂ and climate driven greening trends elsewhere in Yemen.

The southern Tihamah

The Tihamah is a flat plain about 30km wide, which runs parallel to the Red Sea. It is a nationally important area for agriculture and includes numerous wadis that support spate agricultural systems. This involves the capture and storage of seasonal rains, rivers and floodwater in ponds, lakes, canals and dams for later release. In total, 32% of the cropland area in the southern Tihamah is in distress, equivalent to 1,072km². In particular, wadis Zabid, Rima, Siham, Surdud, Mawr are showing signs of deterioration.

Because of the severity of the agricultural pressure in the Tihamah, we have studied the environmental impact of the conflict in wadis Zabid and Rima in an article to accompany this report, allowing a more granular analysis than the country-scale one presented here.

The southern deltas

Along the south-western coast of Yemen lie two agriculturally productive zones – the Tuban delta, which flows through Aden, and the Abyan delta, which flows through Zibjibar. As in the Tihamah, these regions were traditionally farmed with spate agriculture, with different rules and priorities for different sections of the wadis, although in recent times groundwater abstraction has become increasingly important. Farmers here grow the subsistence crops sorghum, millet, wheat, and maize, with cash crops such as fruits, cotton or sesame grown only after these have been harvested.

In total, 42% of the cultivated area in these southern deltas is in distress, equivalent to 167km². In the Tuban delta the agricultural distress has mainly been upstream and along the periphery of the delta, with no clear year or trend dominating the relative-biomass loss. The Abayan delta is distressed throughout its upper and mid-streams, with relative-biomass losses occurring mainly as step changes between 2015-17, although the downstream plantations south of the town of Musaymir have suffered large total-biomass losses.

Jawf

North-east of Sana’a lies Jawf, one of the most remote agricultural areas in the country. Our research shows that 57% of the cultivated area is in distress, equivalent to 110km². The relative-biomass loss has been particularly severe, and a mixture of long-term trends and step losses in 2015-16.

Because of Jawf’s remoteness there is very little information on agriculture in the area, both in the grey or academic literature, at least in English. We do know that it has been an agricultural area since ancient times, using spate irrigation of the wadis Madhab and Kharid – although this water is relatively scarce. There was a significant change in 1985 when a large-scale World Bank project introduced groundwater irrigation. This caused immediate violent conflict and a proliferation of uncontrolled new wells. Agriculture in this area is mainly subsistence and so these biomass drops may have significantly reduced the food security of local people. The primary crop is sorghum, whilst wheat, barley, sesame and some fruit and vegetables are grown close to groundwater wells

Upper Hadhramaut

Wadi Hadhramaut is a striking 100km long valley sunk 300m into the Jol Plateau, which sits at around 1km above sea level. Many tributary wadis feed into the main valley, and again the agriculture relies on spate water and, following World Bank and Soviet interventions, groundwater irrigation. The main crops include date palms, wheat sorghum, sesame and alfalfa.

In total, 39% of the cultivated area in the upper valley is in distress, equivalent to 166km². This is primarily from relative-biomass losses, which has occurred mainly as step-changes in 2018 and 2019. Like Jawf, much of the Hadhramaut is remote and there are few humanitarian actors in the area.

Marib

Up until 2019, the city and region was viewed as a beacon of stability and safety, successfully incorporating a large increase in its population. Atop Marib sits a dam constructed in the 1980s by the UAE to better manage spate and ground waters, and which limits the possible area of cultivation.

We found that 43% of this cropland is under distress, equivalent to 71km². This is primarily because of relative-biomass losses in 2019. Satellite radar indicates that the fields furthest from the dam are under stress. These changes may be connected with the area becoming an active conflict front. The area behind the dam wall has had some relative-biomass loss following the filling of the dam by above normal rainfall.

Northern highland plains

The highland plains are home to the majority of Yemen’s population and, owing to the cooler temperatures and damper environment than elsewhere, also home to much of its agriculture. This is predominantly groundwater fed to grow cash crops. The northern highland plains show signs of agricultural distress, particularly to the north and west of Sana’a, where 37% of the cropland is affected, equivalent to 209km². This is primarily driven by a total-biomass loss – there are actually gains in relative-biomass.

This discrepancy suggests either the more efficient use of the water available, or an increase in reliance on groundwater, another potential sign of agriculture under stress. The areas of greatest difference are where the dominant crop is qat, the narcotic leaf chewed by approximately 80% of the population. Qat uses up to 15% of the total cultivated area in Yemen. And, in order to be cultivated up to five times a year, it uses a disproportionate amount of water, pesticides and fertilisers, exacerbating food insecurity by tying up land, livelihoods and labour.

Relationship to food security

Given the extent of agricultural land under distress during the conflict – 2,565km² – the clear question arising is how has this impacted food security, given that Yemen has been suffering a near-famine during the conflict?

The food security situation results from a complex web of interactions, many of which are underlain by pre-existing economic vulnerabilities. Often these are linked to the agricultural sector and changes that have taken place since the 1970s, when international capital financed large-scale water diversion projects, and a widescale move to groundwater irrigation. This resulted in unsustainable, unfair and ineffective farming practices. Prior to the current conflict, Yemen imported approximately 90% of its food for consumption, despite nearly 30% of the population being employed in agriculture and more than 70% relying on income from the sector.

These underlying fractures predicated the scale of destruction from the ‘economic war’ during the current conflict. Summarised very simply, currency depreciation and a widespread loss of incomes mean that people cannot afford food – the availability of which has been reduced, and cost of which has increased, due to port blockades and the difficulty of getting domestic produce to markets. A recent World Peace Foundation policy brief reports in detail how the economic war and famine are linked across all-sectors.

For most Yemenis the biggest food security impact of agricultural stress has been via loss of livelihoods, rather than the loss of crops themselves – although this is important for subsistence farmers in more remote areas. Unfortunately, owing to significant UN funding shortfalls in 2020, the World Food Programme has had to halve the rations it supplies and which keep much of the population only just above famine. Furthermore, COVID-19 is stretching the resources of other humanitarian aid, a situation compounded by the Houthis’ obstruction of aid shipments.

In this new context, restoring agricultural livelihoods and produce may be seen as a pathway to ease the food security situation. However, through an investigation of the macro-level drivers responsible for the agricultural distress we have observed, we show that economic and environmental reasons make such a restoration unfeasible in the short-term, underscoring the continuing importance of international aid.

Drivers of agricultural stress

1. Armed violence

Based on our analysis of the Armed Conflict Location and Event Data Project (ACLED) database, fighting – intentionally or unintentionally – has directly impacted the agricultural sector throughout the conflict. In total we have identified more than 1,600 events;³ the year and approximate location of these are plotted in Fig 10.

It is clear that many elements of agricultural infrastructure indispensable to the survival of communities have been damaged during the conflict. The distribution of the events changes over time. In general, targets in the early years of the conflict were in the central upland areas, whereas in recent years there has been an increase in the Tihamah and the north western uplands. This correlates with the shifting conflict lines. In total there have been 489 incidents of attacks on agricultural sites or infrastructure in the southern Tihamah, 220 in the northern highland plains, 67 in Jawf, 45 in the southern deltas, 19 in Marib and 5 in the upper Hadhramaut.

Whilst our filtering may not capture all events in the database, nor the database capture all events on the ground, the agricultural damage is widespread and countrywide. Without more detailed information, and a case-by-case analysis, it is difficult to ascertain incontrovertibly if the damage is intentional or collateral. Yet the sheer number of events is compelling. An earlier and more forensic analyses of conflict incidents, using a different dataset, did conclude that the attacks on agriculture were deliberate, and a strategy of the Saudi-led coalition.

Many incidents that are not considered ‘agricultural’ by our filtering algorithm will still be important to the sustainability of the agricultural sector. In particular, the damage to water facilities – be they pipes, dams, wells or other – has also been widespread. Also important has been the damage to roads, bridges and transportation infrastructure, such that moving produce or livestock to or from market has become nearly impossible in some locations.

Conflict parties have regularly been reminded of their obligations under international humanitarian law, and the need to protect civilians and civilian objects indispensable for survival. For example, UN Security Council Resolution 2417 called on parties to the conflict to take “constant care to spare civilian objects, including objects necessary for food production and distribution such as farms, markets, water systems, mills, food processing and storage sites, and hubs and means for food transportation, and refraining from attacking, destroying, removing or rendering useless objects that are indispensable to the survival of the civilian population, such as foodstuffs, crops, livestock, agricultural assets, drinking water installations and supplies, and irrigation work”.

2. Population dynamics

Population dynamics will help explain some of the agricultural changes that we have documented, predominantly through people abandoning farms owing to economic hardship and insecurity. In some situations, the biomass losses will cause population movements, but anecdotally it seems in most cases that migration is a cause of biomass loss, rather than an effect. However, such movements have been concentrated in certain places. According to an International Labour Organisation report, in the first two years of the war the agricultural sector lost almost 50% of its workers. These were predominantly in Hodeidah governorate (73,000 jobs). Comparatively, in Aden and Marib there has actually been a significant increase in the population – however agriculture in these areas is still in distress.

Figure 11 summarises the spatial changes in population, based on movements using the displacement tracking global network, which details the movements of the 3.65 million Yemenis who have been displaced, and on direct conflict deaths from the ACLED database.

3. Access to water

It is estimated that approximately 40% of Yemen’s cultivated land now relies on groundwater. In our forthcoming report on groundwater trends, our analysis of satellite observations indicates significant spatial variability across the country with some pronounced changes during the conflict. We hypothesise that some of these observed changes are related to the economic war, especially the access to energy for running pumps – be that diesel or solar power.

For example, at the end of 2015, there was a reversal to the trend of declining groundwater levels in two areas in agricultural distress, the Tihamah and the southern deltas, with levels subsequently rising to above normal levels. This coincided with cuts to diesel subsidies and economic blockades, and so groundwater pumping stopped in these locations. This led to the abandonment of farms, especially those growing water-hungry fruits like banana and mango. Water levels rose in response to the higher than normal rainfall and permeable geology of the basin have allowed groundwater levels to rise.

Counter-intuitively, in some stressed locations there was an acceleration in the decline of groundwater, in particular the northern highland plains. This was despite the above normal rainfall and diesel access issues. However, we speculate that groundwater abstraction has continued apace due to the increased use of solar panels – readily visible from space and often subsidised by international aid organisations – in particular to ensure that qat production remained at the same level.

However, in areas not identified as being under stress, there are similar positive and negative groundwater trends, indicating groundwater levels alone are unlikely to explain the agricultural distress across the country. In another of our forthcoming reports on water infrastructure, we highlight highlights other, more localised, controls on water accessibility and quality for agriculture. These include damage to the water resource infrastructure, such as wastewater recycling, either following direct attacks or in the absence of good management, the damming of wadi watercourses upstream to horde water, and increased intrusion of seawater into the groundwater resources in coastal locations.

4. Extreme weather

The role of extreme weather events in causing damage to crops is difficult to quantify but likely to be important in some locations. According to the emergency events database (EM-DAT),¹ during the conflict there have been seven major flooding events, five cyclones with associated flooding, and one landslide.

Whilst floods are fairly typical in Yemen, high magnitude events are often devastating, leading to the loss of crops, trees, livestock, infrastructure and human life. In the locations hit by the brunt of these events, crop and soil damage has almost certainly occurred. Without a more comprehensive analysis it is difficult to ascertain the full effect, but imagery and videos show disruption in Sana’a , Aden, Marib, Taizz, Hadhramaut, and Dhamar. Two of the severe flooding events – in Aden and Hadhramaut – have occurred in recent months and so are not captured by our maps, but as water structures have been damaged, they are likely to have severe local implications for this year’s harvest.

Whilst there are not enough data points to make a statistically robust conclusion, that the majority of cyclones to make landfall have occurred in the past five years suggests a new and emergent threat. This is consistent with recent research that suggests climate change will increase the severity and frequency of tropical storms. Other facets of climate change may also make agriculture throughout Yemen more precarious, or create new opportunities. These include increasing temperatures in its desert regions, or decreased sunlight in the uplands. In the appendix we show heterogenous trends in rainfall indicators across the country (amount, intensity, number of days, maximum). However, although Yemen is deemed one of the countries most vulnerable to climate change, since the onset of the conflict no country-scale climate assessment has been conducted, nor funds made available to do so. Such activities will be essential for any restoration of the agricultural sector.

5. Pesticides and fertilisers

The economic war has led to a fall in the use of agricultural inputs, in particular pesticides and fertiliser. A private sector survey of farmers in Hodeidah and Lahj for the UNDP indicated that this lack of inputs has played a leading role in crop losses.

Although the total use of pesticides and fertilisers may have declined, there is widespread evidence to suggest that the use of illegal and harmful alternatives has risen, as unscrupulous entrepreneurs and smugglers have taken advantage of the loss of sanitary controls at ports. Although a small fraction has been seized by the government, they are hamstrung by poor coordination between agencies and limited testing capacity, with only one laboratory. Some dealers are reported to have manufactured the pesticides locally. Yemeni investigative journalists found pesticide shops in Sana’a to be stocked with banned pesticides and fertilisers – the majority of these shops were unlicensed and of those that were, 90% still violated a law stipulating the technical expertise required. These products are especially popular with qat growers, who are reported to use as many as 80 different pesticides, as they can increase annual harvests from two to five, and preserve freshness.

The use of these pesticides brings a host of health issues. They are often carcinogenic, responsible for 15% of childhood poisonings, linked to type-2 diabetes, and cause a host of other acute and chronic issues. The wider environmental effects on soil, groundwater and biodiversity in Yemen are likely to be significant but are largely unknown and understudied. In an intersection with the water crisis, the pesticides are more likely to be ingested as fresh food is rarely washed due to water shortages. Prior to the conflict, there were renewed efforts to tackle the problem – for example in discussions at the National Dialogue Conference. However, in 2019 it was reported in Yemen’s sixth national report to the Convention on Biological Diversity (CBD) that ‘weak law enforcement related to pesticides and fertilizers use’ remains one of the key obstacles preventing sustainable agriculture.

6. Pests

The agricultural losses may also be related to an increasing prevalence of pests, either owing to natural variation, changes in management, or a collapse of control measures.

The primary threat to crops in Yemen has been desert locusts. They are monitored by the FAO, which funds in-country surveys and collates data on its Locust Hub. These form the basis of monthly reports, which have consistently highlighted how the conflict has prevented surveys and control measures. Even before the conflict, control infrastructure was limited and, whilst the Sana’a-based Desert Locust Monitoring and Control Center did receive FAO funding in 2015 to repair its offices damaged during the 2011 political upheaval, there have been subsequent reports of underfunding. Because of the active conflict lines, this programme has only been able to support relatively small areas. Whilst a secondary control centre has been set up in Aden, their resources appear even more limited – even cars have to be hired – and so unable to serve more distant areas such as Hadhramaut. Here, farmers have had to resort to basic and ineffective methods, such as waving them away with cardboard cartons, or burying them in trenches.

Nonetheless, the data that was collected – see figure 13 – illustrates the scale of the desert locust problem throughout the country. Swarms of adult and juvenile locusts (hopper bands) have been reported in the highland plains and terraces, Hadhramaut, wadi Tuban, Jawf, Marib and the northern Tihamah. Numbers have been high in the past 18 months, but especially so in the past six months (Fig. 14), and the ongoing effect on agriculture may be significant for livelihoods and food security. Fortunately, there has been an increase in control measures too, but these miss key areas in Hadhramaut, west of Aden, Jawf, and the southern highlands. Furthermore, as can be seen by the number of locust events in proximity to control measures, they are only successful to a degree. The increase in control measures is partly associated with a scaling-up of the FAO’s desert locust appeal; how sustainable this will be in the new funding landscape is unclear. Although a menace to crops, locusts have actually helped alleviate hunger in Yemen to some extent – they are harvested from crops at night and sold as a low-cost snack.

The FAO has scaled up its desert locust appeal in response to the crisis in East Africa, where 4.9 million people faced starvation this summer. These locust swarms originated in Yemen in August 2019, before migrating to Ethiopia, where their progeny moved to southeast Ethiopia (October), to Kenya and Somalia (December), then within Kenya, Ethiopia, northern Tanzania and northern Uganda (January), and South Sudan (March). Subsequent generations have hatched in many of these locations.

The next generation is expected to transit through South Sudan and potentially into Chad and West Africa. Given the inadequate controls described above, it is likely that the conflict conditions in Yemen prevented proper control measures, and have directly led to an environmental and humanitarian crisis elsewhere.

It is a remarkable example of the environmental dimensions of conflict reverberating across a continent, potentially destabilising areas thousands of kilometres away. An expert from the FAO’s desert locust team told the Guardian that Yemen had “become a reservoir” that will stoke the East Africa crisis because of the inability to control breeding. Areas of East Africa are already insecure and the combination of Yemeni derived locust threats and recent climate shocks – in particular in South Sudan – may have significant consequences.

Increased rainfall, and particularly from cyclones, has created highly favourable conditions for locusts. Given the potential for an increased frequency of cyclones because of climate change – and the seemingly intractable nature of the conflict – locusts in, and originating from Yemen may remain problematic for some time. Several tools are now available for open-source data researchers to track developments.

Yemen has to contend with other problematic pests too. In Hadhramaut, 2013 saw the introduction of the red palm weevil. This beetle’s larvae feed on date palm trunks, killing the trees. Whilst an emergency programme was launched in 2014 by the FAO, it is unclear how successful this has been – the conflict may have stymied further control measures. In mid-2018, the fall armyworm arrived from the African continent, where it had already destroyed $4.6 billion worth of crops, and spread across Yemen, hitting maize and cereal crops. Control of the fall armyworm, red palm weevil, tomato leaf minder, wheat rust, and dubas bug are likely to be important for ensuring crop yields and livelihoods in the coming years.

7. Changing agricultural practises

It is possible that the biomass losses are related to changes in agricultural practises or crops, potentially in a response to conflict pressures. A move towards increased production of qat, as anecdotally reported, may have negatively affected agriculture elsewhere by extending its domination over scarce inputs and resources. In the conflict economy such a move would make sense and maximise income security – the market value is guaranteed as the crop is so essential, with approximately 80% of Yemenis chewing daily. Qat is also essential to the Houthi leadership, who use the proceeds of associated taxes to fund the war effort, use free qat as a recruiting tool, and ensure that growers have sufficient access to diesel. Crowded qat markets remain open despite the COVID-19 transmission risk, which chewing and spitting the drug may exacerbate.

Unfortunately, we have been unable to assess if the area or intensity of qat cultivation has increased – to do so would require a more sophisticated machine learning approach to overcome the challenges of different geographies (terraces versus plains), and distinguish from similar evergreens such as coffee. However, official statistics suggest that qat area and yield has stayed at the same level throughout the conflict – in stark contrast to other crops. Furthermore, as already discussed, the continued drop in groundwater levels in the qat growing areas may be evidence of either an intensification of cultivation, or a need to abstract more groundwater to counter unfavourable conditions. Satellite imagery shows the proliferation of solar panels in qat growing areas during the conflict, indicating the means to continue pumping groundwater despite the fuel crisis.

Likewise, it has not been possible to verify anecdotal reports of the restoration of long-abandoned agricultural terraces following urban to rural migration. Following qualitative inspections of satellite imagery, we can find no evidence for this trend. Through viewing current and historical satellite imagery, we have noticed an increase in unknown agricultural buildings in the highland plains. These may be greenhouses, which benefit from efficient water use, although it is our understanding that greenhouse technology is still in its infancy in Yemen. Various agencies including Food4humanity, Oxfam, US-Aid, Saudi-Aid and the FAO are currently running pilot programmes. Further investigation is required, with a more sophisticated algorithm needed to map the buildings’ prevalence over time.

One factor which may be overlooked is the collapse of beekeeping, which is critical to both agriculture and biodiversity. According to the Yemeni Organization for Honey and Agriculture Development, this has occurred as the cost of honey is too great for most Yemenis, economic blockades have reduced exports by around 50%, the seasonal movement of breeders to find fresh flowers has been inhibited, and farms have been directly attacked.

8. Governance failures and institutional collapse

The conflict has negatively impacted governmental agencies that were already weak. The economic war and the war economy, and in particular the non-payment of wages, limited budgets and corruption, has led to a loss of operations in many instances. Further to this, the essential infrastructure to run agriculture-relevant agencies and departments has been lost, be that labour, buildings, expertise, paperwork, or computers. Much of the support offered to farmers has been totally suspended, but likely the most important was the loss of diesel subsidies. Moreover, projects to improve agriculture, or the infrastructure that supports it, have been suspended in the face of the conflict – the maxim that conflict is sustainable development in reverse is particularly true for Yemen.

Closing remarks

We have documented a number of agricultural areas exhibiting different levels of stress, and a list of mostly conflict-related drivers that we know to be contributing in some way to this deterioration. It is difficult to attribute the losses between these drivers at the country scale as they often exert a collective influence. These factors can also act in different directions at different locations and times, yet cumulatively cause stress that is visible from space. Nonetheless, it is clear that – as before the conflict – water is a key factor determining vulnerability, be that access to groundwater or damage from floods.

The immediate outlook for Yemen on all fronts is bleak; this is particularly true for the agricultural sector. Not only are locust numbers on the rise, but the recent flood events will likely lead to reduced yields particularly of wheat and sorghum. This is on top of the accumulated losses already sustained as a result of the conflict. In a seemingly intractable conflict, where humanitarian funding is being cut and COVID-19 mitigations are inadequate, there is little prospect of agricultural recovery any time soon; as such the food security situation will continue to deteriorate.

Longer term, in the case of peace, sustainable recovery is possible and could lead to a coherent agricultural system that is more environmentally sustainable and provides increased productivity, food security and secure livelihoods. Some changes may seem simple – such as conversion from qat to less water dependent crops such as coffee or honey, which have a high export value, although these solutions need scrutinising to ensure no unintended consequences. For example, the proliferation of solar panels to pump groundwater may seem promising in democratising energy access, but they may promote the unsustainable abstraction of groundwater and only be accessible to those farmers with the most capital, i.e. those growing the water intensive cash crops.

Some of the changes may seem backward looking – such as restoring the terrace systems – but are actually an example of a nature-based solution to climatic change. Better enforcement of existing legislation would help reduce water conflicts. Many of the recommendations in a World Bank report from before the conflict will remain relevant, in particular the localisation of new policy development, implementation and management – in stark contrast to the international approach in the past.

The sector will of course face challenges from the reverberating effects of the conflict, be that governance, unexploded ordnance or a smaller workforce. A response to increased locust activity will be required if it is associated with climatic changes, although the response is complex and the current activity may be transient.

Dr Eoghan Darbyshire is a researcher at CEOBS specialising in open source data analysis. Our thanks to Walid Saleh, Lindsay Stringer, Lina Eklund, Paul Scholte, and those interviewed in anonymity for their feedback on this investigation and data analysis.

Appendix

A. Data and methods used in this article

All data used in this CEOBS analysis is from open-source datasets. In most cases this data has then been used to derive further data products using code developed in the Matlab, Google Earth Engine and QGIS programming environments. Where feasible, the data and code used here can be shared upon request. We are also happy to share our figures in higher resolution formats.

A1. Agricultural Distress Index

The data products used to calculate the distress index were:

• Total biomass produced annually, acquired from the FAO WAPOR platform on a 100m grid and referred to in this work as total-biomass. The product is calculated by aggregating the dekadal (every 10 days) net primary productivity. This is the carbon uptake owing to photosynthesis minus the losses by plant metabolism, calculated using solar radiation, weather, and Normalised Difference Vegetation Index (NDVI) derived products (land use type and therefore light use efficiency, fraction of photosynthetic radiation used and soil moisture stress).
• Gross Biomass Water Productivity (GBWP), also acquired from the FAO WAPOR platform on a 100m grid and referred to in this work as relative-biomass. GBWP is the ratio between annual evapotranspiration and total biomass produced. Evapotranspiration is the amount of water consumed by the plants through evaporation, transpiration and interception (i.e. rainfall). These terms themselves are a function of precipitation, temperature, humidity, soil moisture stress and land cover, all of which are accounted for based on data from earth observations, assumptions and model outputs. Large changes in relative-biomass over short time frames indicate a likely human driver, although effects arising from natural disasters will also be captured. Because GBWP is a highly derived product it is imperfect, and may not capture and assimilate all elements sufficiently – for instance it does not account for fog, which is a primary source of moisture for natural vegetation in Yemen’s coastal facing mountainous regions.
• The crop fraction for a given pixel, from Proba-V 2015 100m Land Cover maps. Where this was over 15%, a pixel was considered cropland. Qualitative validation of the product against visual imagery indicates a high success rate, although there are locations were classification is not perfect – particularly in the mountainous regions.

Any given pixel was deemed as distressed if any of the following conditions were met:

• The difference between the mean pre-conflict (2009-2013) total-biomass and mean post-conflict (2014-2019) total-biomass was greater than 2 standard deviations of the pre-conflict period (2009-2013), and the change was negative (i.e. loss of biomass).
• The difference between the mean pre-conflict (2009-2013) total-biomass and mean post-conflict (2014-2019) total-biomass when expressed as a percentage was greater than 50%, and the change was negative (i.e. loss of biomass).
• The difference between the mean pre-conflict (2009-2013) relative-biomass and mean post-conflict (2014-2019) relative-biomass was greater than 2 standard deviations of the pre-conflict period (2009-2013), and the change was negative (i.e. loss of biomass).
• The difference between the mean pre-conflict (2009-2013) relative-biomass and mean post-conflict (2014-2019) relative-biomass when expressed as a percentage was greater than 50%, and the change was negative (i.e. loss of biomass).
• The change in total-biomass was positive and the change in relative-biomass was negative, where the difference between the percentage changes was greater than 50%.
• The change in total-biomass was negative and the change in relative-biomass was positive, where the difference between the percentage changes was greater than 50%.

The statistics for each region were based on the following definitions:

• Southern Tihamah: Bounding Box: Lon 42.5->43.5, Lat 13->15, Elevation < 300m
• Jawf: Bounding Box: Lon 44.3->45, Lat 15.9->16.5
• Marib: Bounding Box: Lon 45.1->45.7, Lat 15.3->15.7
• Aden & Abayan Deltas: Bounding Box: Lon 44.6-> 45.7, Lat 12.7->13.4, Elevation < 300m
• Highland Plains: Bounding Box Lon 43.5 -> 45, Lat = 13.5->15.9, Elevation > 2000m, Slope < 2.5e6
• Upper Hadhramaut: Bounding Box Lon 47.8->49.2, Lat 15->16.4, Elevation  < 750m, Slope < 2.5e6

Where the elevation and slope data was acquired from the Shuttle Radar Topography Mission (STRM) digital elevation model at 30m spatial resolution. Data processing was undertaken in Matlab and QGIS.

A2. Biomass step-change algorithm

Where there were negative changes in relative-biomass, we used a simple algorithm to determine if the change was the result of a step change or year-on-year trend, and if the former the year of the change-point.

To identify the locations experiencing a step drop in relative-biomass four numbers were calculated:

1. relative-biomass mean for the period encompassing the year in question and the two proceeding years (one in the case of 2018, and none for 2019)
2. relative-biomass mean for the period encompassing 2009 to the preceding year
3. relative-biomass standard deviation for the period encompassing 2009 to the preceding year
4. (b) minus two multiplied by (c), representing the size of change required for a step change

Where (a) was less than (d), and (d) was greater than a minimal change threshold of 0.15kg m^-3, a negative step change was identified. The magnitude of this drop was then converted to a percentage by: ((a-b)/b)*100

For pixels where there was no step change, a trend was identified via least-squares if the Mann-Kendal test was passed with a 1% significance level.

A3. Filtering the ACLED database

The notes for each entry into the Armed Conflict Location & Event Data Project (ACLED) database for Yemen between Jan 2015 and Jun 2020 was filtered using keywords to find events associated with agriculture and water security. To avoid false positives, we also searched for exemptions, e.g. we identify entries with ‘field’ but then blacklist these if the field is ‘oil field’, ‘field commander’, ‘field marshal’. The filtered entries have undergone some, but not exhaustive manual verification.

The keyword list for agriculture was:

‘ soil’; ‘ crop’; ‘agriculture’;’ farm’;’ fish ‘; ‘fisherman’; ‘fisheries’;’ bee ‘;’ beekeeper’;’ plants’; ‘ salt ‘;’ field’;’ agri’; ‘agro’; ‘vegetable’;  ‘flower’; ‘plantation’; ‘onion’; ‘tomato’; ‘okra’; ‘legume’; ‘zucchini’; ‘mulukhiyah’; ‘ tree ‘;’ qat ‘; ‘ fruit ‘; ‘ coffee ‘; ‘palm’; ‘banana’; ‘mango’; ‘pomegranate’; ‘melon’; ‘cotton’; ‘orchard’; ‘garden’; ‘ arbor’; ‘ tractor ‘;’polytunnel ‘; ‘ nursery’; ‘greenhouse’; ‘dairy’;’ milk’; ‘livestock’;’ graze ‘; ‘grazing’;’ sheep ‘;’ goat ‘;’chicken’; ‘cattle’; ‘ cow ‘;’ cows ‘; ‘donkey’; ‘donkeys’; poultry’; ‘camel’; ‘horse’; ‘ruminant’; ‘forage ‘; ‘foraging ‘;’pasture’; ‘ herd’; ‘sorghum’; ‘millet’; ‘corn’; ‘cereal’; ‘barley’; ‘maize’; ‘sesame’.

And the exemption list:

‘seller’; ‘market’; ‘dealer’; ‘harbour’; ‘oil field’; ‘field leader’; ‘field commander’; ‘Raydan field’; ‘Yamani Dairy Factory’.

A4. Official data from the Ministry of Agriculture and Irrigation

To help validate the earth observation data, we have sought to present data from the ground. The Ministry of Agriculture and Irrigation (MAI) is the only organisation collating agricultural data for the whole country, and we have compiled this into time-series presented in figure A1. The MAI is based in Sana’a and under Houthi control, and therefore liable to political manipulation; this may not be the case here, but such a risk is one of the problems of using data from conflict parties. Through the conflict there are drops in the yield and area, in particular for cereals, fodder, vegetables and fruits, consistent with out satellite analyses. Notably, the only major crop which has remained constant is Qat, with yield reportedly increasing by 30% in 2019. Note, there are improvements in most crops yield in 2019 – this is not entirely inconsistent with the satellite data, with biomass showing an increase in 2019, however the scale of the increase is not consistent and as such we treat these numbers with some skepticism.

Figure A1. Changes to crop yield and area since 2007, from official data from the Sana’a based Ministry of Agriculture and Irrigation (MAI) year books 2011-2019.

 B. Selected environmental baseline data for Yemen

B1. Landscape

There is a great diversity of landscapes in Yemen, with distinct climates, land uses and challenges owing to the ongoing conflict and climatic changes. The first step to investigating these changes is to understand this diversity. Fig. B1 displays the distributions of climate, topography, land-use and population density alongside simplified geo-climate zones that we have derived.

The spectacular Sarawat mountains rise above 3500m as they run the length of western Yemen, parallel to the Red Sea but separated by a 30km coastal plain, the Tihamah. Plains also run along the southern coast but are more discontinuous, punctured by the escarpments of the Hadramaut plateau – this stretches inland to near the Saudi border and is interspersed with sunken wadis (dry river beds). The Hadramaut is enclosed by desert, the Ramlat al-Sab’atayn to the west and depressed empty-quarter to the north-east. The island archipelago of Socotra, ‘the Galapagos of the Indian Ocean’, lies 350km off the coast of Yemen and consists of coastal plains, plateaus and tall mountains.

The majority of Yemenis live in the Sarawat, including the capital Sana’a, as the wetter and cooler climate is more hospitable – for both humans and plant life alike. Owing to its steep escarpments and a predominantly westerly wind generating orographic precipitation, more rain falls in the Sarawat then anywhere else in Yemen. The western coastal plains are more inhospitable, with the high temperatures, high salinity and aridity resulting in desert-like conditions where only limited vegetation can prosper. Vegetation is most prosperous in wadis.

Whilst traditionally home to natural mangrove scrub and Doum palms, over the twentieth century the wadis were turned over to agriculture, including sorghum, dates and figs, with livestock grazing on by-products, residues and weeds. The heat, humidity and isolation of Socotra has led to spectacular and unique vegetation, much of it endemic, such as the dragon’s blood tree, the cucumber tree and many aloes. The hyper-arid dunescape of the empty quarter means very little vegetation grows, apart from in some places, scattered xeric shrubs. In the far east of Yemen are the coastal forests of the Al Hawf fog-oasis, a potential UNESCO site, and which abut the border with Oman.

B2. Climate

Figure B1 summaries the key climate changes across Yemen in recent decades. The dominant climate trends are rising temperatures, in-line with global heating, and a fall in relative humidity i.e. a drying, especially in the Sarawat. During the conflict there has been above normal rainfall across the mainland but in particular in the Sarawat and Tihamah. Likewise, the west of the country has seen higher temperatures and drier conditions than pre-conflict. This is a simple analysis intended to show the basic climatic conditions during the conflict, and how these compare to the climatology. More detailed analysis, such as that conducted for precipitation in the next section, are important to elucidate second order climate changes.

Figure B1. Landscape and climate parameters of Yemen. A) Simplified land use classification from the WaPOR platform at 250m spatial resolution. B) Shuttle Radar Topography Mission (STRM) digital elevation model at 30m spatial resolution. C) 6 geo-climate zones defined for this work based on landscape, land use and climatic differences. D) Estimated population density per 100m x 100m grid square for the year 2010 provided by the WorldPop project. E-I) Climate parameters displayed as (i) the climatology (i.e. mean field) over 1981-2018, (ii) the difference between the mean fields during the conflict (2014-18) and pre-conflict (2009-13) periods, and (iii) the annual anomaly (i.e. distance from mean field) for each geo-climate zone against a 1981-2018 baseline. Precipitation data provided by the CHIRPS 0.05° product. Temperature, relative humidity and wind speed data provided by the ECMWF ERA5 reanalysis project.

B3. Precipitation

There is a complex picture of precipitation trends varying between east-west, north-south and a dry-wet season. In general, the wet areas are getting wetter and the dry areas drier, but there are more rainy days across the country and the intensity of rainfall is falling. Briefly these findings can be summarised as:

• Total precipitation increasing in the west and decreasing in the east, especially in the empty quarter.
• Maximum daily and maximum 5-daily precipitation totals are increasing in the north-west, in particular near Sana’a during the wet season. Comparatively they are declining in the south-west near Ibb and Taiz during the dry season, and the east year-round.
• The number of rainy days are increasing, more so in the west and in particular at altitude and during the wet season.
• Precipitation intensity is falling across the country, in particular during the dry season, with the exception of the north west during the wet season.
• The number of extreme rainy days is increasing in the north west of the country, but remains constant elsewhere.

More work is needed to unpick these trends in greater detail, and understand the drivers. We present this simple analysis to show the complex picture and help shed light on the ongoing trends that are likely important to consider from an environmental peacebuilding perspective. (n.b. Whilst the heavy rains associated with cyclones are important events for flooding, this type of analysis looks at the more ordinary precipitation, especially important for spate agriculture)

Figure B2. Statistically significant annual trends in rainfall properties over Yemen during the period 1981-2019.

Our precipitation analysis uses precipitation estimates from the Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) is dataset. This product combines satellite precipitation and cloud data with in-situ station data to create gridded rainfall time series, also accounting for biases introduced by complex terrain. Here, we use the daily 0.05° version of the data.

We followed the approach in Harrison et al., (2019), splitting the analysis into wet-season, dry-season and annual components for each pixel, and used  definitions from the Expert Team on Climate Change Detection and Indices. The Harrison et al., (2019) paper found CHIRPS v2.0 data ranked highest out of multiple datasets when compared to sparse gauge data in sub-Saharan Africa.

The blending algorithm used to produce the CHIRPS data does not include uncertainty information; this represents a substantial weakness of the product and means we cannot present uncertainties associated with the precipitation anomalies calculated. Apparently, work is ongoing to provide standard error fields, which may be available in future CHIRPS releases. In nearby Ethiopia, which has a somewhat similar climate, Dinku., et al (2018) found CHIRPS had high skill when compared to independent rainfall assessments.

It must be noted that the ground gauge data going into CHIRPS from Yemen ceased after the start of the conflict, although the coverage from surrounding countries remains good ensuring some level of ground constraint on the satellite measurements.

B4. Landscape wide vegetation dynamics

The change in the average total-biomass produced for all vegetation between the pre-conflict (2009-2013) and during-conflict (2014-2019) periods is presented in Figure B3. The dominant colour on the map is blue, indicating an increase in biomass, especially in the upland areas in the west. These large-scale positive biomass changes are perhaps surprising given the agricultural distress outlined above.

Figure B3. Change in total-biomass between the pre-conflict and conflict period

However, this biomass gain is over a large area and is predominantly natural vegetation, indicating that larger scale primary drivers may be at play. Inter-annual variability in meteorology, illustrated in Fig. B1, may have been more favourable for vegetation growth during the conflict years. To account for the changes in biomass accounting for the influence of this inter-annual variability in meteorology, in Fig. B4 we present relative-biomass.

Although the scale of the losses in some agricultural areas is either greater or reverses, the use of relative-biomass appears to make little difference to the large-scale picture – the predominant feature is still biomass gain in the upland. A more granular look at these gains, rather than just comparing the average of two periods, indicates they are similar year on year, rather than discrete step changes. This linear trend stretches back till at least 2002 based on an analysis of NDVI (not shown here) and is thus consistent with a large-scale driver – but what might this be?

Curiously, up to half of global vegetation is greening and three main drivers have been identified: (1) land-use changes, (2) favourable climate changes, and (3) CO₂ fertilisation – i.e. a turbocharging of photosynthesis as CO₂ concentrations increase. Across 1982-2015, data presented in the recent IPCC Special Report on Climate Change and Land suggests that for western Yemen these drivers are similar in scale. However, during the conflict, there has been no large-scale land use or cover change in the areas of biomass gain, which have remained mainly natural vegetation. Likewise, a large climate driven shift in vegetation seems unlikely as despite the clear long-term climate trends in Fig. B1, average temperature is relatively stable across the pre-conflict and conflict periods, whilst the decrease in humidity likely puts plants under stress rather than promoting growth. However, very recent research indicates that the role of CO2 fertilisation has been significantly underestimated in past studies, and is the overwhelming driver of greening at the latitudes of Yemen.

There are biomass gains in every geo-zone, and these greening drivers are likely in operation across the country. That there is widespread biomass loss in spite of these greening drivers highlights the magnitude of the driver(s) causing the loss.

Figure B4. Change in relative-biomass between the pre-conflict and conflict period