A landmark CGIAR evaluation published in July 2025, covering 2,611 households across 33 villages in five districts, has produced the most granular picture yet of what smallholder farmers in northern Tanzania grow, how they grow it, and what the data reveals about why the region’s nutrition crisis is, at its root, a production crisis.

In October 2023, researchers from the International Food Policy Research Institute, the World Vegetable Center, the National Institute of Medical Research Mwanza, and Sokoine University of Agriculture fanned out across 33 villages in five districts in the Arusha and Kilimanjaro regions of Tanzania. Over the following three months, they surveyed 2,611 households with women of reproductive age, asking detailed questions about what each household grew, on how much land, with what inputs, and in what quantities.

The result, published in July 2025 as Vegetable, Fruit, and Staple Crop Production and Input Use: Baseline Findings from the FRESH End-to-End Evaluation (Tanzania Evaluation, Research Brief 3, July 2025), is the baseline document for the CGIAR FRESH end-to-end evaluation, the most comprehensive longitudinal study of fruit and vegetable farming and dietary change in northern Tanzania to date. It is Research Brief 3 in a series, following briefs on the overall evaluation design and on food and nutrition security outcomes.

What the brief documents is not a farming system in crisis in the dramatic sense of crop failure or acute food insecurity. It is something quieter and in some ways more intractable: a farming system that is largely organised around the wrong crops, applying inputs unevenly, achieving yields far below demonstrated potential for some of its most nutritious vegetables, and producing almost no fruit at all. Seven findings from the baseline data tell this story in full.

Finding One: Most Farmers Are Not Growing Vegetables, and Almost None Are Growing Fruit

Of the 1,968 farming households assessed, 79% cultivated staple crops during the long rainy season. Only 41% cultivated any vegetables. Only 14% cultivated any fruit. The three largest household categories in the sample were those farming staple crops only, at 52%, those farming staple crops and vegetables, at 28%, and those farming staple crops, vegetables, and fruit together, at just 6%.

Most farming households were smallholders operating less than two acres of land on a single plot, growing one or two crop types. Within these tight resource constraints, staple crops come first. Maize and kidney beans, the dominant staples, provide the food security foundation. Vegetables and fruit, while nutritionally important, compete poorly for the same land, labour, and inputs.

The structural dominance of staple crops is the foundational explanation for the dietary shortfall documented in the CGIAR nutrition briefs. When only 41% of farming households grow any vegetables and just 14% grow any fruit, the supply available to the broader population through local markets is constrained from the start. The dietary gap is, in large part, a production gap. And the production gap is not primarily a knowledge or technology problem. It is a resource allocation problem driven by rational decisions under constraint.

Finding Two: Tomatoes Are a Cash Crop, Not a Household Food

Only 11% of vegetable-farming households in the study grew tomatoes. But those who did invested more intensively in the crop than in any other vegetable: dedicating an average of 1.3 acres to it, giving it 47% of their cultivated area, applying fertiliser at a rate of 95%, using inorganic fertiliser at 82%, and applying pesticides at 87%. Among households producing more than 100 kilograms of tomatoes, 100% used both fertilisers and pesticides.

The commercial intent behind this investment is explicit. On average, 87% of tomato production was designated for sale. Among the larger producers, those harvesting more than 500 kilograms, an average of 95% of output went directly to market. Tomatoes in northern Tanzania’s smallholder system are not a dietary vegetable. They are a cash crop that happens to be nutritionally valuable.

This distinction matters for the nutrition agenda. Increasing tomato production will increase farm income and market supply. But it will not automatically improve household nutrition, because the households growing the most tomatoes are precisely the ones keeping the least for themselves. Any intervention that relies on tomato production as a pathway to improved diet quality needs to engage with this commercialisation dynamic, not assume it away.

Finding Three: Amaranth’s Yield Gap Is One of the Largest in the Dataset, and One of the Most Consequential

Amaranth, mchicha, was grown by 22% of vegetable-farming households in the study, making it the fourth most commonly cultivated vegetable across the sample. It is nutritionally dense, culturally embedded in the diet of both regions, and, on paper, one of the most promising crops for improving the nutritional quality of what households eat and sell.

The baseline records average amaranth yields of 4.5 tons per hectare across the farming households surveyed. Published field trial results from research conducted in the same study regions by the World Vegetable Center and colleagues, cited directly in the brief, demonstrate yields of 18.8 tons per hectare in the wet-cool season and 27.7 tons per hectare in the hot-dry season.

The gap between 4.5 and 27.7 tons per hectare is sixfold. It is not a marginal underperformance. It is evidence that the technology to produce dramatically more amaranth exists and has been demonstrated locally, but is not reaching farming households in any meaningful way. Only 24% of amaranth farmers used improved variety or hybrid seeds. Only 10% used inorganic fertiliser. Only 31% used any pesticide. Amaranth is being farmed as a low-investment subsistence crop when it is capable of being one of the most productive leafy vegetables in the system.

Finding Four: African Eggplant Is Widely Eaten and Almost Nobody Is Growing It

African eggplant, nyanya chungu or bitter tomato, is a frequently consumed vegetable across both Arusha and Kilimanjaro. It appears regularly in local cuisine, is nutritionally significant, and has been a priority crop for variety development by the World Vegetable Center. Nineteen improved vegetable varieties including African eggplant have been released for commercial production in Tanzania under the CGIAR programme.

The baseline found that just 3% of vegetable-farming households, 25 households in the entire sample of 815, were cultivating African eggplant. Average yield among those who did was 2.6 tons per hectare, against a national average of 4.5 tons per hectare for Mainland Tanzania. Use of improved variety seeds was 36%, meaning nearly two-thirds of African eggplant farmers are using local or recycled seed.

This is among the most striking disconnects in the brief: a crop that is widely consumed, for which improved varieties have been developed and released, is being grown by almost no one in the study area, and at yields well below national norms by those who do grow it. The FRESH programme has distributed African eggplant seed kits to farmers as part of its supply-side intervention, a recognition that seed access is the critical bottleneck. The baseline shows how much ground remains to cover.

Finding Five: Arusha and Kilimanjaro Are Two Different Farming Systems Wearing the Same Regional Label

The brief’s regional disaggregation reveals a divide between the two study regions that has significant implications for how interventions are designed and targeted.

In Arusha, 59% of farming households grew vegetables. In Kilimanjaro, only 28% did. The vegetables grown in each region also differ substantially. Arusha households more commonly cultivated African nightshade, collard greens, tomatoes, spinach, and green cabbage. Kilimanjaro households more commonly grew amaranth, pumpkin leaves, Ethiopian mustard greens, and cowpea leaves. The Arusha vegetable basket is more commercially oriented. Kilimanjaro’s is more diverse in traditional leafy greens.

Input use was consistently higher in Arusha across almost every vegetable examined. Fertiliser use for African nightshade was 88% in Arusha versus 80% in Kilimanjaro. Inorganic fertiliser use for the same crop was 30% in Arusha against 7% in Kilimanjaro. Pesticide use for amaranth was 45% in Arusha and 18% in Kilimanjaro.

Yet higher input use does not always produce higher yields. Kilimanjaro households achieved 7.6 tons per hectare for collard greens against Arusha’s 2.9, despite dedicating less land to the crop. For amaranth, Kilimanjaro recorded 3.2 tons per hectare against Arusha’s 2.4. These yield reversals suggest that Kilimanjaro farmers, working with lower input intensity, are better adapted to specific traditional leafy vegetables in ways that Arusha’s more commercially oriented system is not.

The practical implication is that the FRESH end-to-end intervention cannot be applied identically in both regions. Arusha needs work on yield improvement and variety upgrade for crops already in the system. Kilimanjaro needs work on expanding who is farming vegetables at all. These are different problems requiring different entry points.

Finding Six: Fruit Farming Is a Banana System, and That Is Not Enough

Fruit farming in the study area is defined by two facts: it involves only 14% of farming households, and among those households, 75% grow bananas. Avocados were grown by 24% of fruit-farming households, papayas by 15%, mangoes by 10%, oranges by 8%, and guavas by 6%. Every other fruit combined accounted for 17% of fruit-farming households.

Bananas in this system are a homestead crop, grown near the house rather than on dedicated commercial plots, harvested periodically, and consumed or sold informally in small quantities. Average banana production per farming household was 199 kilograms per year, at an average yield of 2.1 tons per hectare against a national average of 3.5. Input use was minimal: 3% used improved variety seeds, 2% used inorganic fertiliser, 3% used any pesticide.

This is not a fruit system capable of supplying the dietary levels that health requires. The companion nutrition briefs found that women in Arusha and Kilimanjaro consume an average of just 35 grams of fruit per day, against the WHO minimum that forms part of the 400-gram daily recommendation. The farming data explains why. When only 14% of farming households produce any fruit, and three quarters of those are producing bananas in home gardens, the aggregate supply of diverse fruit available to the population is structurally insufficient.

The most consequential single finding on fruit was for mangoes in Arusha. Average mango yield in the study area was 1.3 tons per hectare. The regional average for Arusha is 15.1 tons per hectare, nearly twelve times higher. The gap between what mango farming achieves in the study area and what it achieves in the broader region points to a set of production constraints, whether variety, management, or market incentive, that remain unaddressed.

Finding Seven: Tomatoes Are the Most Chemically Intensive Crop in the System, and That Raises a Food Safety Question

The final finding draws together the production data and the food safety research from the broader CGIAR programme. Tomatoes in the study area are not just commercially dominant. They are chemically dominant.

Chemical pesticide use among tomato-farming households was 43%, the highest of any vegetable in the study and roughly three to four times higher than for the next most intensively managed crops. Among commercial-scale producers, those harvesting more than 100 kilograms, 100% used both fertilisers and pesticides. Inorganic fertiliser use for tomatoes was 82%, and for green cabbage, the second most intensively managed crop, it was 73%. For every other vegetable in the study, inorganic fertiliser use was below 25%.

The companion CGIAR research brief on food safety, A Dual Threat: Postharvest Losses and Food Safety Issues in Tanzania’s Markets in Arusha and Kilimanjaro, found that chemical contamination of fresh produce is among the major food safety concerns in the region, alongside microbial contamination from unhygienic market conditions. Enforcement of existing regulations in informal markets is difficult. Knowledge of and investment in food safety practices is limited.

Together, these two briefs raise a question that the FRESH evaluation does not yet answer: what residue levels are reaching consumers on the tomatoes sold through informal markets in Arusha and Kilimanjaro? The safe vegetable production research, also from the same CGIAR programme, found that applying integrated pest management as part of good agricultural practices can reduce chemical pesticide application frequency on tomatoes by 30% while increasing yields by 144%. That is the pathway from the current high-input, safety-uncertain model to a more sustainable one. But reaching it at scale, among commercial tomato farmers who have built their practices around intensive input use, is one of the FRESH intervention’s most challenging tasks.

What the Baseline Means for the Evaluation Ahead

The FRESH end-to-end approach, under which this evaluation is being conducted, combines supply-side interventions, including climate-resilient cultivar distribution and training on safe and sustainable production, with demand-side and food environment interventions. The evaluation will assess impact on household vegetable production and fruit and vegetable intake among women of reproductive age across the study area.

The baseline has established what the evaluation is working against: a farming system where most households do not grow vegetables, where fruit production is thin and narrow, where some of the most nutritious crops are being produced at a small fraction of their demonstrated potential, where the most commercially important vegetable carries the heaviest chemical footprint, and where two neighbouring regions require meaningfully different intervention approaches.

None of this makes the FRESH programme’s task impossible. The evidence from the production trials and variety promotion briefs shows that yield improvements, input reduction, and farming system diversification are achievable under the right conditions. But the baseline makes clear that achieving them at the population level, across 33 villages in five districts, will require sustained effort, well-targeted support, and a realistic understanding of the structural constraints that have produced the numbers in this brief.

The endline evaluation will show how far the intervention has moved those numbers. The baseline has shown, with precision and at scale, exactly where they started.

This article draws on Vegetable, Fruit, and Staple Crop Production and Input Use: Baseline Findings from the FRESH End-to-End Evaluation (Tanzania Evaluation, Research Brief 3, July 2025), authored by Bliznashka L., Dione M., Zagré R. R., Boniface S., Dinssa F., Mwambi M., Mbwambo O., Mwombeki W., Jeremiah K., Malindisa E., Kinabo J., Cunningham K., Olney D., and Kumar N. Published by the International Food Policy Research Institute, Washington DC. Baseline data collected October 2023 to January 2024 across 2,611 households in 33 villages in five districts in the Arusha and Kilimanjaro regions of Tanzania. Research conducted in partnership with the World Vegetable Center, the National Institute of Medical Research Mwanza, and Sokoine University of Agriculture under the CGIAR Science Program on Better Diets and Nutrition (BDN). Full findings: cgiar.org/funders.

Based on: Vegetable, Fruit, and Staple Crop Production and Input Use: Baseline Findings from the FRESH End-to-End Evaluation (Tanzania Evaluation, Research Brief 3, July 2025)