On-farm conservation

Part of a series on

Conservation

Native potato varieties, each with a card bearing its name, on a display table at Santa Teresa in Cusco, Peru, in 2026.

Cannabis conservation·Grow-out priorities·Populations

Threat Categories

Extinction

Genetic Threats

Genetic erosion·Genetic contamination·Introgression·Gene flow·Hybridisation·Population bottleneck·Genetic drift·Inbreeding depression

Habitat & Environmental Threats

Habitat loss

Eradication Tracking & Analysis

Eradication·Herbicide spraying of cannabis·Herbicide spraying in Mexico·Herbicide spraying in the United States·Herbicide spraying in Colombia·Herbicide spraying in Belize·Herbicide spraying in South Africa·US eradication abroad·Yukthiya·Cannabis eradication in Kirivong·Cannabis eradication in Fiji·Campaigns by country

International Enforcement

Cannabis prohibition·US eradication abroad·INCB

Legal & Policy

Farmers' rights·Access and benefit-sharing·Intellectual property·Single Convention on Narcotic Drugs·1988 Convention·Cannabis legalisation movement

Population Assessment

Minimum viable population·Population viability analysis·Effective population size

Conservation Monitoring

Population monitoring·Genetic monitoring

Conservation Methods - Ex Situ

Ex situ conservation·Gene bank·Cryopreservation·In vitro conservation·Core collection·Seed storage

Conservation Methods - In Situ

In situ conservation·On-farm conservation·Genetic reserve·Community seed bank·Feral and ruderal populations

Genetic Conservation

Germplasm·Genetic resources·Crop wild relative·Conservation genetics

Sampling & Collection

Germplasm collecting·Germplasm sampling

Preservation Strategies

Conservation prioritisation·Complementary conservation

Recovery & Restoration

Reintroduction·Genetic rescue

Conservation Science

Landrace conservation

Portal·Category

On-farm conservation is the conservation of crop diversity by the farmers who grow it, within their own farming systems. Nigel Maxted and three co-authors defined it in 1997 as "the sustainable management of genetic diversity of locally developed crop varieties (landraces), with associated wild and weedy species or forms, by farmers within traditional agricultural, horticultural or agri-silvicultural systems".^: 18 With the genetic reserve for wild species, it is one of the two forms of in situ conservation of plant genetic resources. Its aim is to keep going the processes by which crops change under farmers' management: their selection, the exchange of seed and gene flow between crop populations. Particular genes or genotypes come second.^: 261: 138

For Cannabis, hemp farmers in the Tai'an district of Shandong in China ran their own seed system in the mid-1990s. The fibre crop was cut before it flowered. Seed for the next year came from fields sown for seed and from escaped plants by roads and ditches.^: 63–64

Farmers' practice and projects

Stephen Brush distinguished two senses of in situ conservation of crops in 2000. In the first, diversity persists through farmers' everyday practice. It is most visible where farmers keep local varieties although bred varieties are available to them. In the second, governments, international programmes and private organisations run projects to support that practice. Such projects may use techniques that are not local.^: 4 Brush took landraces to be the main object of such work. In his description they are named farmer varieties, usually of narrow range, often diverse and adapted to local conditions. In his view the science of conserving them in the field was less developed than that for wild relatives of crops.^: 5

When the international effort to store crop germplasm took shape in the 1970s, on-farm conservation was considered and set aside. Brush gave three reasons. Development was assumed to require the replacement of local crop populations. Farmers were expected to need a subsidy to go on growing old varieties once bred ones were on offer. Diversity left in farmers' fields was of little direct use to breeders who worked on experiment stations.^: 7 By 2000 he held that in situ and ex situ conservation complement each other and that neither is sufficient alone. Farming systems go on producing new variation after a sample has been collected. Populations on farms are a backup for gene banks, which are themselves exposed to drift in their collections and to failures of equipment and funding. The Convention on Biological Diversity calls for in situ conservation.^: 7–11 Brush held that on-farm conservation should be aimed at a few places and groups of farmers, was not an alternative to agricultural modernisation and "will not succeed through administrative coercion".^{: 9, 12}

What farmers decide

Devra Jarvis and Toby Hodgkin listed five kinds of decision by which farmers may change the genetic diversity of a crop population: which traits to select, how to manage the crop, where to plant it, how many plants to grow and where the seed comes from.^: 270 Selection may take place in the field during the season or on the harvested crop. The person selecting may be a man or a woman, young or old. Planting on a particular site exposes a population to that site's conditions, as with barley sown in Ethiopia on soils prone to waterlogging and rice grown at up to 3,000 m in Nepal.^: 270–271 The smaller the population planted, the more it is open to genetic drift, inbreeding and the loss of alleles.^: 272

Farmers keep several varieties of one crop for different fields, risks, markets and uses. Brush's example is a farmer in Turkey who may grow different wheats for hillside and valley fields and for irrigated and rain-fed plots. The wheats serve different uses: bread made at home, grain for the towns, straw and fodder.^: 3–4 In the Peruvian Andes, where soils and growing conditions do not explain the number of potato varieties kept, Brush took part of the explanation to be the value that Andean culture places on diversity itself.^: 15 He divided the value of a local variety to its growers into three kinds. Direct value is its yield and uses. Indirect value is what it does for the farming system, such as the growing of maize with beans and squash in Mexico. Option value is the wish to pass the variety on or to keep it for future needs. Farmers may not value the genetic resource itself, only the practical and sometimes aesthetic qualities of the crops that carry it.^: 15–17

Seed lots and seed flow

Dominique Louette followed the maize of 39 farmers, a fifth of those in the indigenous community of Cuzalapa, over six seasons from 1989 to 1991. The valley lies in the Sierra de Manantlán Biosphere Reserve in Jalisco, Mexico, within the region where maize is thought to have originated.^: 110–113 Louette separated a seed lot, the kernels that one farmer selects and sows in one season to reproduce a type of maize, from a variety, the set of seed lots that share a name.^: 112 The farmers grew 26 varieties. Six were local and covered more than 80 per cent of the maize area. The other 20 had been brought in and changed from season to season.^: 113–114 Of 484 seed lots sown in the six seasons, 53 per cent came from the farmer's own harvest, 36 per cent from another farmer in the valley and 11 per cent from outside it.^: 115 Farmers had several reasons to look for seed from others. Insects often damaged ears kept in store for more than six months. Some farmers had no land of their own or worked as sharecroppers. Farmers generally held that a variety's seed should be renewed to keep it productive. An introduced seed lot that looked like a local variety took its name.^: 118–121

More than 30 per cent of the seed lots were made up from fewer than 40 ears, which by theoretical estimates is too few to keep rare alleles. Yet the purple-grained variety Negro, 70 per cent of whose seed lots came from fewer than 40 ears, was as diverse at isozyme loci as varieties sown from larger lots. Louette put this down to cross-pollination between neighbouring fields of different varieties, which farmers did nothing to prevent.^: 125–128 Farmers chose their seed ears after harvest for size, soundness and type. In a trial over two seasons this choice reduced the effect of crossing on a variety's ear traits and kept its grain colour steady. It did not stop the change at isozyme loci.^: 129–133 For Louette the maize of the valley formed a metapopulation, in which seed lots die out when not replanted and new ones are founded through exchange. He concluded that the unit to conserve was the whole group of varieties grown in the area, with the processes that create its diversity.^: 136–138

Oliver Coomes and 16 co-authors set out in 2015 to rebut four views of farmer seed networks: that they spread seed inefficiently, that they are closed and conservative, that they give every farmer ready and equal access to seed and that they are bound to disappear.^: 42 They found reasonable the global estimates that 80 to 90 per cent of seed sown passes through such networks, farmers' own saved seed included. Formal seed systems in developing countries often supply well under 10 per cent.^: 43 The networks carry wild plants and bred varieties onto farms. In Vanuatu farmers give away taro so that a variety lives on beyond their own gardens, a practice the authors count as storing seed in the network.^: 44 Access is unequal: marriage rules among the Fang of Gabon and boundaries between language groups limit who receives seed. Widows, orphans and tenant farmers can be left out.^: 45 The authors expected the networks to last alongside commercial supply, since farmers pass purchased seed through them and rules on exchanging seed are costly to enforce.^: 46

Supporting farmers

Brush sorted the ways of raising the value of local varieties to farmers into market and non-market methods.^: 17 One market method is to sell landrace produce to buyers who will pay more for it, as urban consumers in Peru did for selected Andean potatoes. Another is to certify a product's origin through an appellation. Brush doubted that the sale of crop genetic resources themselves, under contract or as intellectual property, could pay for conservation, since breeders have little demand for material from farmers' fields.^: 17–20 Among non-market methods, organisations in the Peruvian Andes held "diversity fairs" at which farmers from different villages showed the crops they grew. In participatory plant breeding, farmers and breeders select together among variable, segregating material. In participatory varietal selection, farmers choose among finished lines from a breeding programme. Brush judged the first more favourable to conservation, because it works from material like that already grown. He expected the second to replace local populations with new and less variable ones.^: 20–21

Stacks of harvested sorghum stalks in fields by Lake Hayq in South Wollo, Ethiopia, in 2014. The Seeds of Survival programme kept sorghum landraces on farms at sites in Wollo, among them Loga Haik.

The Seeds of Survival programme in Ethiopia, begun in 1988 by the national gene bank with support from the Unitarian Service Committee of Canada, kept landraces on farms under the farmers' own practices. The farmers set the plot sizes and seed rates.^{: 144, 149–150} Farmers and scientists selected sorghum and maize landraces together. In 2000 Melaku Worede and colleagues reported 3,102 farmers growing the sorghum selections and 2,999 the maize. About 500 farmers were paid on contract to keep material likely to be abandoned and to multiply selected landraces for distribution. Samples of the original and the selected seed went to the national gene bank.^: 151–152 Composite landraces of durum wheat, selected for yield under stress, were grown on nearly 4,000 farms in Central and Eastern Shewa.^: 153 The programme planned community seed banks, owned and run by local cooperatives, each with a seed store and a store of germplasm for local crop improvement.^{: 157, 159}

In 1995 the International Plant Genetic Resources Institute began a project with national programmes in nine countries: Burkina Faso, Ethiopia, Nepal, Vietnam, Peru, Mexico, Morocco, Turkey and Hungary. It set out to learn where and why farmers keep local cultivars, how their decisions change genetic variation over time and how continued selection of local cultivars could be supported. Its planners specified that it should not dissuade farmers from adopting new varieties that raise food supply or income.^: 263

Measuring what is kept

Brush suggested measures of success besides counts of alleles and genotypes. Two of them were the number of farmers in an area who keep local crop populations and manage them by local criteria and the flow of farmer varieties within and between communities.^: 9 Up to 2000, most studies had related farmers' choices to diversity at a single point in time.^{: 262, 274} Valeria Negri and colleagues judged, on anecdotal evidence, that the people keeping landraces are almost always elderly. In their judgement only a handful of on-farm projects were actively maintaining landrace diversity.^: 8–9

George Dyer and colleagues set 38 case studies of maize on Mexican farms, from 1995 to 2009, against a nationally representative survey of rural households in 2002 and 2007. The case studies suggested that the number of varieties per farm was rising. The survey found it falling, from 1.43 to 1.22. It also found less exchange of seed between 2002 and 2007 than in the five years before. The authors attributed the difference to the preference of case studies for sites of high diversity. They called the decline genetic erosion. Brush, Mauricio Bellon and four co-authors replied that differences between the case studies explained more of the variation than time did. They held that the survey's effort to capture minor varieties was unclear. Counts by household, they added, cannot show whether a variety dropped by one household survives elsewhere in the community.

Cannabis seed on farms

The hemp landrace of Tai'an that Robert Connell Clarke studied, known locally as Lai Wu or Fei Cheng hemp, was almost entirely dioecious. Clarke saw little if any intentional selection of it by farmers. Two seed types, one large and one small, grew in the same fields, the large-seeded making up about 65 per cent of plants and the small-seeded about 20 per cent. They flowered at very different times, so few hybrids formed between them.^: 61 Farmers said they preferred the large-seeded type, which gave longer stalks, but the small-seeded type kept its share of the population.^: 62 The fibre crop was harvested before it flowered and set no seed. Most seed came from fields sown in late May or early June for the purpose. The rest came from escaped plants on ditch banks and roadsides, which the farmers called "wild".^: 63–64 The seed fields were sown in clusters in rows about one metre apart, so that the plants could branch. Farmers often removed the tops of the female plants as they began to flower, to make them branch further. They sometimes thinned up to 90 per cent of the male plants at early flowering. Most of the seed was bought by other hemp farmers of the district.^: 64–65 Seed was also exchanged between counties.^: 62

In Himachal Pradesh in India, the state cabinet approved a pilot study of cannabis cultivation in January 2025. A coordinator of the Himalaya Niti Abhiyan told Down to Earth that locally grown seed, called bijo, had been used for fibre and food until the mid-1990s. A former village head of Mandi District urged that local seed be preserved for its cultural and economic value.

Hemp landraces recorded in villages of Romania between 2000 and 2008 are described under landrace conservation. Morocco's rules on seed for licensed growers are covered in the article on in situ conservation.

See also

• Community seed bank
• Complementary conservation
• Ex situ conservation
• Farmers' rights
• Gene bank
• Genetic erosion
• Genetic reserve
• In situ conservation
• Landrace Cannabis
• Landrace conservation
• Seed saving