The regulatory frameworks for varietal approval and quality seed control concentrate on governing the seed supply. They do not intend to influence the direction of plant breeding. However, the indirect result of the variety release system is that breeders target high input levels, wide adaptation, and easily recognizable traits. In the extreme case, the main objective of the breeder is not to develop varieties for the farmers but to develop varieties that can be approved by the variety release committee. This often results in approved seeds that have little significance to farmers.

In most countries in the South, formal seed production has developed as part of the Green Revolution strategy of agricultural development. Plant breeding would increase yield potential under optimum growing conditions, seed production would be a necessary vehicle for technology transfer, and the seed and inputs would be subsidized. The centralized seed production infrastructure has been built to resemble the successful European and North American seed industries.
Similar to the development of formal seed production, seed laws in the South are strongly based on European or US seed laws. Such laws regulate variety release and official seed certification, and they establish the institutional framework of national seed councils and certification agencies. More recently, plant variety protection (PVP) has been introduced in the regulatory system. A major difference with the European systems is that the variety release and seed certification in Europe was originally based on a farmers' wish to regulate seed supply. Hence, farmers' organizations have a strong voice in the system. For example, farmers are represented in the Netherlands' seed certification foundation. Whereas in the South, such seed regulations were fitted in the existing bureaucratic structures. The seed quality control institutions became the driving force behind the development of seed legislation. They need legal backing when conflicts arise with seed producers.

Variety release systems
Variety release may become a goal in itself when regulatory systems are too rigid. Release is the yardstick on which the effectiveness of public plant breeding is measured. Their reward system is commonly based on the number of varieties released, not on the widespread use by farmers. Hence, the objectives of plant breeding are likely to be adapted to the variety release procedure.
A variety release system generally involves the following procedure: (a) application with a formal variety release committee and variety registration, including a variety description; (b) testing for the value for cultivation and use (VCU) of the variety involves different sites over a number of seasons; and (c) testing for distinctness, uniformity and stability (DUS). The test results are analyzed by the committee either to approve or reject for formal release. In all these stages there can be a bias towards favouring particular types of varieties. Since variety release is basically geared to supply varieties to the formal seed sector, strict uniformity standards of seed certification have to be acknowledged. Uniformity can be tested before a variety enters the trials, or parallel to this.
The management of variety testing makes it doubtful whether the trial results have any value to most farmers. Breeding is still expected to target favourable agro­ecological conditions for a monoculture of crops that are high yielding and have wide adaptations. Below are some of the management problems:

• Poor site selection is the result of variety trial stations which may not be representative of the agro­ecological systems where the variety is going to be marketed.
• Poor trial management increases residual variances. Residual variance reflects differences due to environment. Ideally, trials should be managed to reflect the input level of average farmers. Poor management trials distort the environment so that actual differences in varieties become less apparent. This results in the reduction in the number of varieties released or in the delay of release due to prolonged testing. Most trials contain small plots, often with considerable border effects. Sometimes high input levels are used as a matter of policy. In this view, variety trials represent the conditions of the better­off farmers. More often however, high input levels are used because they can improve the trial from a statistical point of view. High fertility levels, careful irrigation and pesticide can reduce field variation and thus reduce residual variances. Thereby, it increases the chance of approval of the higher yielding varieties. Finally, high input levels give 'beautiful crops' that make a trial credible to visitors. For example, trials are commonly weeded precisely when many visitors are expected. The result is that yield levels in official trials do not represent farmers' conditions at all. For example, in India the average yields in the official 1989/1990 sorghum variety trials were three times the farmers' average yields (approximately 2800 kg/ha against 900 kg/ha in farmers' fields).
• Poor identification of crop characteristics that are of relevance to farmers generally occurs. In the best cases diseases are screened but these often merely mark the presence or absence of a particular disease. Moreover, other characteristics that are extremely important for farmers are not taken into account. These includes aptitude to intercropping, shattering (e.g. soya bean), lodging when harvesting is delayed (e.g. maize), and cooking time of the produce (e.g. beans). Breeding thus tends to concentrate on yield as the most important characteristic without giving credit to the diverse needs of farmers.
• Analysis of the trials is often done using standard statistical analysis methods. The main characteristic of this analysis is that it leads towards breeding for wide adaptation. Since trials are pooled in one calculation, the variety having the highest average yield is again considered the best. However, this may not be the best variety in any of the testing sites. Standard variety release procedures rarely accept a variety that is specifically adapted to particular conditions, even though they may contain regional recommendations. Breeding for specific adaptation is thus not rewarded.

• Lack of participation and transparency in the closed system of the formal variety release leads to conservative trial designs and management. Parallel (demonstration) trials by the extension service, non­governmental organizations (NGOs) or private seed companies are rarely taken into account in the release decision. Farmers in the South are rarely well­represented in the variety release committee, or in the evaluation of varieties.

• Complete (vertical) resistance. Successes have been recorded of complete resistance to various pests and diseases, based on gene­for­gene relationships (see box on page 4) between the plant and the pathogen. Some of these vertical resistances proved durable, whereas others have broken because of random mutations in the pathogen population.
• Polygenic (often partial or horizontal) resistance. This is often but not always, based on several genes in one plant, and on the distribution of different resistance genes (different alleles of one gene) in a population which can result in significant resistance level.
• Different resistance genes in a multiline variety or in a variety mixture. The inclusion of different resistances in different plants of the same variety also reduces the epidemic. Most spores of a diseased plant will land on resistant types, which slows down the epidemic even when the disease consists of a population of spores with different pathogenity levels. The result of this breeding strategy is a multiline variety, i.e. a variety which is sufficiently uniform in its essential agronomic characteristics, but genetically heterogeneous for the resistance. A potential advantage of this strategy is that components in the variety that becomes heavily attacked may be replaced by other components with different resistance genes. Breeding multiline is expensive, because it involves numerous additional backcrosses. It further causes delay compared to pure line breeding, and it has additional cost in variety maintenance, seed production and seed quality control.

Releasing resistant varieties
The variety release systems tend to discriminate certain resistance strategies such as horizontal resistance and multiline varieties. On the other hand, the result of breeding for vertical resistance is a uniform variety that can be easily observed and is likely to be released. Breeding for horizontal resistance can also result in conventional uniform variety, which could pass a standard variety release procedure without many problems. Complications may arise however, when the testing facilities are poorly equipped or poorly staffed since varieties with partial resistance to a particular pathogen do carry symptoms of the disease. Their value lies in the reduced damage due to a slowing down of the epidemic when planted in large fields. Because they do carry symptoms they are likely to be recorded as susceptible and consequently rejected. Partially resistant varieties may not be identified in standard variety trials.
The problem with multiline varieties is that they show field resistance which is comparable to partial resistance. In addition, they are less uniform than conventional pure­line varieties. Especially where registration is based on the requirements for plant variety protection, multilines are often rejected. When multiline varieties have to be released individually, they may fail because they will not have better yields compared to all other varieties.
Finally, different registered varieties could be blended and sold to farmers. Such varietal mixtures could be constituted by seed producers or seed merchants, and if chosen carefully, could buffer disease epidemics. Most certification regulations, including those in some European countries, do not allow blended seed on the market. Control would be extremely difficult, but ruling out a potentially useful disease management strategy by law does not seem useful either.

Multiline wheat in the Netherlands
An illustrative case is a multiline wheat variety bred by the Netherlands' seed company Zelder in the late 1970s. This novelty put pressure on the Netherlands' variety release system. The variety, aptly named 'Tumult' (Dutch for 'commotion'), was insufficiently uniform, and thereby could not be released. The final option was to register the constituting lines as separate varieties. These are hardly distinctive because they are different only for the resistance gene. Moreover, testing of the lines individually for VCU would cause problems. The resistance to yellow rust of individual lines was below standard, but when tested as a multiline the variety was superior. Eventually, the registration authority was convinced by the breeder to test and approve the multiline variety as one entity. However, commercially it turned out to be a failure. The breeding and release procedure caused the multiline variety to be outdated before it even appeared on the market. Additionally, maintaining the lines and producing the seed separately was too expensive. The Zelder seed company has not produced multiline varieties since then.

Regulations as impediments
Even though breeding for vertical resistance is the easiest option, experience shows that this strategy bears some important risks. A resistance can be broken by the pathogen, causing massive losses. Such losses are more disastrous for risk­prone, small­scale farmers in low external input farming conditions, compared to their more commercial colleagues. The seed replacement rate in commercial farming is high. When a resistance is broken, farmers can purchase seeds of a new variety the next season if the breeders have done their job in being 'ahead of nature'. Their loss has to be absorbed for one or two years only, which can be done in the presence of effective rural credit. This was the case of the massive destruction of maize due to blight in the early 1970s in the USA.
Breeders for low external input farmers, who are mostly in the public sector, have different responsibilities. Their clients obtain seeds to a large extent through the local seed exchange mechanisms and variety replacement is slow. In the absence of effective rural credit schemes and commercial seed supply systems, a serious outbreak of a disease cannot be tackled by the market. Durability of resistances to diseases is therefore even more important in such systems. Breeding for polygenic resistance or multilines should be promoted, even though the methodologies are more complex. This is currently discouraged due to very ineffective variety release systems.
The importance of such regulatory frameworks in the breeding strategies has been acknowledged only very recently. Regulations are an impediment to novel approaches, such as breeding for specific adaptation and participatory plant breeding which are very valuable tools for targeting less endowed farmers.
Niels P. Louwaars

Centre for Plant Breeding and Reproduction Research (CPRO­DLO), P.O.Box 16, 6700 AA Wageningen, the Netherlands  Fax (+31) 317 418 094; E­mail n.p.louwaars@cpro.dlo.nl

Sources
S. Ceccarelli (1989), "Wide adaptation: How wide?" Euphytica 40, 197­205.

N.P. Louwaars and G.A.M. van Marrewijk (1996), Seed Supply Systems in Developing Countries. Wageningen, the Netherlands: CTA.

R. Tripp (1997), New Seeds and Old Laws. London, UK: Intermediate Technology Publications.

D.S. Virk, A.J. Packwood and J.R. Witcombe (1996), Varietal Testing and Popularisation and Research Linkages. Discussion papers series. Bangor, India: Centre for Arid Zone Studies.