1

Breeding for salinity tolerance – is it time for a new approach? P A Hollington1, 2; A Rashid3 and KS Gill4 1CAZS Natural Resources, University of Wales Bangor, UK; 3Soil Science Dept., NWFP Agricultural University, Peshawar, Pakistan; 4Dept. of Agronomy, Punjab Agricultural University, Ludhiana, India 4Corresponding author

Telephone: + 44 1248 382285

Email: p.a.hollington@bangor.ac.uk

Abstract Despite enormous effort and resources devoted to breeding salt tolerant crops over the years, the results in terms of new salt-tolerant varieties growing in farmers’ fields can only be described as disappointing. We put forward possible reasons for this: the difficulties involved in field screening for salinity are well known but other factors are also important, for example the lack of national testing for salt-tolerant varieties, the limited genetic background of the material used in many crossing programmes, and the concentration on salt tolerance to the exclusion of quality traits that are important to farmers. We suggest that the adoption of a client-oriented approach, which has already had enormous success in breeding several crops for drought-affected and other marginal lands in India, Nepal and Bangladesh, has the potential radically to improve the delivery and uptake by farmers of improved, salt-tolerant varieties that also satisfy the other requirements of farmers and increase biodiversity. Examples are given of successful client-oriented breeding for wheat, rice and maize, showing how the use of a low-cross number / high population strategy that takes account of stakeholder preferences avoids wasting resources in breeding material that will not be accepted by farmers, and details of how the process could be adapted for salt-affected environments are noted.

Keywords Participatory breeding, Drought, Salinity, Client orientation, Abiotic stress, South Asia

Introduction

Salinity, sodicity and waterlogging have long been major constraints on crop production, particularly in irrigated systems (Szabolcs 1994; Shannon 1997). Around 800 M ha are affected by salinity or sodicity worldwide (FAO 2000). Most is due to natural causes (primary salinity), but about 2% of dryland agriculture is affected by secondary salinity (resulting from human activities), as is 20% of irrigated land, which produces about one third of the world’s food (Munns 2005). In the developing world, salinity can reduce wheat yields by 65% (Quayyum & Malik 1988), leading to increased poverty and reliance on imports. The problem will worsen as population growth forces more land under irrigation, and climate change and increased competition for water (e.g. Hobbs & Gupta 2003) make it essential to exploit marginal lands and water.

Yield reductions due to these stresses lead to increased poverty, which is likely to be accentuated by global change (Yeo 1999; Hillel & Rozenweig 2002). Little extra land is available to increase the production area (Young 1999), increasing pressure to raise yields per unit area of soil and unit volume of water. Plant breeding for yield under drought may be the greatest challenge facing agricultural science (Reynolds et al. 2005), but so far traditional breeding has had limited effect on crop production in drought- and salt-affected areas (e.g. Araus et al. 2002; Flowers 2004). It is likely that increased farmer involvement in the whole breeding process would lead to better targeted varieties suited in particular to the needs of poor farmers in drought and salt-affected regions – the focus of this paper is on salinity, but the techniques are equally applicable to drought-affected and other marginal areas

Problems with conventional breeding

In breeding for low-potential stressed environments, one major problem is that selection efficiency decreases as the difference between the selection environment and the target environment increases (Ceccarelli & Grando 1999), due to high genotype x environment (G x E) interactions (Ceccarelli 1994, Ceccarelli et al. 1994). Thus genotypes selected on research stations under high input conditions do not usually do well in low-potential environments such as marginal farms in dry areas (Ceccarelli et al. 1998), or areas affected by salinity.

2

Salinity

Despite the many efforts to breed salt-tolerant crops, there has been little progress in producing varieties accepted and used by farmers (Flowers 2004; Gregorio & Cabuslay 2005), with fewer than 30 salt-tolerant cultivars released by 1995 (Flowers & Yeo 1995) and only three more were registered and one patented in the following nine years (Flowers 2004). Although transgenics with some salt tolerance have been produced, none have been field-tested and few of the claims made to success appear valid due to flawed testing methods, e.g. experiments under conditions of zero transpiration, or failure to compare the wild-type with transformants under appropriate salinity levels (Flowers 2004).

The multigenic nature of salt tolerance, and rapid spatial and temporal changes in field salinity (Richards 1983) make reliable screening difficult. In some crops this is compounded by additional stresses associated with salinity: mineral deficiencies and toxicities, submergence, deep water and drought may be important. These vary with time, so cultivar adaptability depends upon long term tolerance (Gregorio et al. 2002). The dynamism affects the level of salinity at which to screen, as large areas of a field may be of low or moderate salinity, with only small areas of high salinity. Richards (1992) concluded that selection at low salinity was preferable, as most yield was from the non-saline areas. However, he drew his conclusions from work on drying saline fields inappropriate to the irrigated conditions of many developing countries, and did not consider salinity/waterlogging interactions. Isla et al.(2003) agreed that on moderately saline soils the best strategy was to breed for high yield potential, but argued that under higher salinity breeding for both yield AND salinity tolerance was important.

In addition to the dynamic nature of salinity, difficulties are increased by plants preferentially extracting water from less saline areas of the rootzone, further decreasing selection efficiency. In rainfed environments with dryland salinity, drought lends an additional complication (Srivastava & Jana 1984). Breeding is also hampered by the need to select for productivity, and the difficulty of introducing tolerance traits without affecting flowering date and dry matter (DM) production. Tolerance also differs with growth stage (Shannon 1997) and environmental conditions (Maas 1990), and varies not only within species but within varieties (e.g. Abdus Salam et al. 1999 for wheat, Yeo et al. 1988 for rice), so screening needs to coincide with the likely stress period in the farm situation, to produce lines tolerant at particular growth stages which could be crossed to produce material for more complex situations.

Why are farmers not growing improved varieties?

Colleagues in Bangor surveyed the adoption of modern, high yielding varieties (HYVs) of rice in six states in India (Stirling & Witcombe 2004). In many districts, fewer than half the farmers were growing HYVs: these were also districts where yield was 1 t ha-1 or less, around half that of the districts with a higher adoption rate, and the more marginal the area, the fewer HYVs were grown. In many areas farmers continue to grow old varieties (for example average ages were 11 years in rice and 17 years in sorghum (Witcombe et al. 1996; Virk et al. 1996) that may be low yielding and susceptible to pests and diseases, or very old landraces (Witcombe et al. 1996). Farmers may not have been exposed to HYVs, and those that are released may be unsuitable for rainfed, marginal environments.

The situation is similar in salt-affected areas. In India, two wheat varieties have been released from the Central Soil Salinity Research Institute (CSSRI), KRL 1-4 and KRL19, and these are grown to some extent in salt-affected areas. However, when we interviewed farmers on saline land in the Indian Punjab they were unaware of these lines, and asserted that they had only been offered genotypes bred at the local agricultural university, which were unsuitable for their environments. In common with almost all the Indian genotypes developed from the old Rajasthan landrace Kharchia, KRL 1-4 has red grain unpopular with farmers (Dr KN Singh, CSSRI Karnal, Personal Communication), and is highly susceptible to rust, so leading to reduced uptake by farmers. However, a new variety, KRL 99, has just been registered for tolerance to salinity and to waterlogging, and also has excellent agronomic traits including amber grains (Singh et al. 2007). When we carried out testing of Indian genotypes in Pakistan, we found that KRL 1-4 did not do well, possibly as a result of denser soils and greater waterlogging, and conversely genotypes selected in Pakistan have not done well in India (Hollington et al. 2002).

In Pakistan varieties are not tested in a dedicated national programme for salt tolerance, and must therefore compete in terms of yield and disease resistance with lines from wheat breeders that incorporate the latest CIMMYT germplasm. Salinity is not a priority for the major crop breeding groups, and so it is left largely to non-specialists such as soil science and botany departments to produce this material. Such groups are clearly at a disadvantage when it comes to modern plant breeding methodologies and access to improved germplasm, and are also perceived to be subject to discrimination against their material in the official trials, as well as to competition from other departments within their institutions. As in India, salt-tolerant material is bred from old genotypes, for example much is based

3

upon a selection from the variety LU26, which dates back to the 1960s. This has serious consequences in terms of disease susceptibility and yield potential.

There is clearly an urgent need to identify and exploit new sources of salt-tolerance from a more diverse background. We also strongly believe that the lack of progress in developing successful varieties, in particular for Pakistan, has largely been due to the lack of simultaneous screening for salinity and waterlogging, which greatly exacerbates its effects (Barrett-Lennard et al. 1999; Barrett-Lennard 2003), although given current climatic conditions this is possibly not the constraint now that it once was. We do believe however that many of the examples of failure at the farm level could have been avoided if farmers had been consulted about their requirements at the start of breeding programmes.

Participatory approached

Our work in Bangor using participatory approaches to plant breeding has had enormous success in drought-affected and other marginal lands in India, Nepal and Bangladesh. For rice, examples include Joshi et al. (2007) in Bangladesh, Joshi et al. (1997) in Nepal, and Virk et al. (2003) in eastern India, while examples for maize include Witcombe et al. (2003) in western India, and Virk et al. (2005) in eastern India. We believe that such methods could radically improve the delivery and uptake by farmers of improved, salt-tolerant varieties that also satisfy the other requirements of farmers and which would increase biodiversity.

The strategy has been to use a low-cross number / high population size (Witcombe & Virk, 2001), taking account of stakeholder preferences to avoid wasting resources. We aim to provide a “Basket of choices” (Chambers 1989) of varied genetic material for the farmers to choose from. We obtain data from evaluation across environments, as well as measures of farmers’ preferences, and we include traits not usually considered by breeders, such as organoleptic evaluations for taste, smell and cooking characteristics (e.g. Virk et al. 2003). By giving seeds directly to farmers, even the poorest and most risk-averse can test new genotypes in their own fields.

Two complementary approaches can be used: participatory varietal selection (PVS), and participatory plant breeding (PPB). Recently (Witcombe et al. 2006) we have adopted the term client-oriented breeding (COB). This has been done as we feel that defining breeding programmes as “participatory” or as “formal”, “conventional” or “classical” develops an adversarial mindset between breeders who practice one method and those who practice the other, and overall it is better to describe the programme according to the degree of client-orientation. COB develops the purpose of high client-orientation, a systematic and explicit effort to involve the clients of the breeding programme, the farmers, while PPB describes the process.

Participatory variety selection

We have found PVS to be a rapid and cost-effective way to identify farmer-preferred cultivars, so long as a suitable choice of genotypes exists. If not, then it is necessary to produce novel variation by crossing, for which the more resource-demanding PPB may be used. PPB can use previously-identified genotypes from PVS as parents.

There are four main stages in PVS: first the identification of farmers’ requirements; secondly a search for suitable material for testing; thirdly experimentation with farmers; and finally the dissemination of farmer-preferred material. The identification phase allows farmers to be given appropriate material to test, and is carried out using a variety of techniques such as participatory rural appraisal (PRA), crop walks at harvest, or pre-selection by farmers of material in multi-entry trials. This phase allows identification of the best performing lines, comparison of the recommended and new lines with local material, and evaluation of diversity, and, importantly, ensures agreement between farmers’ names and phenotypes – very often farmers in different places may use different names for the same genetic material.

The search phase requires sourcing cultivars that meet farmers’ needs, in terms of maturity, height, grain quality and agro-environmental niche. These are selected from those released nationally and regionally, and may include old varieties and advanced pre-release material.

The third phase, of experiments on farmers’ fields, can range from the situation of very little participation, basically “on-station” trials moved to the farm, which are valuable to broaden the range of soils, pests and diseases encountered and encourage interactions with farmers, through to almost total participation with very little researcher input. We have found a simple method known as informal research and development (IRD) to be highly cost-effective. This is a good technique for organisations like NGOs, and allows them to provide farmers with acceptable improved genotypes even when they only have limited resources, Very often, IRD trials have only one new genotype to be tested alongside the farmer’s usual variety, and they record only simple data perceptions rather than

4

measured yields. An example is an IRD trial carried out on rice in Nepal (Joshi et al. 1997) in a high-potential environment where 90% of the rice grown was previously of one variety. The farmers were offered small amounts of seed of different varieties to grow without any researcher intervention, and within two years more than 35% of households had chosen to grow the new genotype again.

If new varieties are to be acceptable to farmers, it is essential that organoleptic tests are carried out as part of the post-harvest participatory assessment (e.g. Virk et al. 2003). Such tests assess traits like aroma, taste, grain consistency and cooking quality which, although rarely assessed in traditional plant breeding, are important for poor farmers. The method therefore has the advantage that expensive field evaluations of agronomically fit lines that will be rejected by consumers are avoided. It is important to do these evaluations in all the areas in which the variety is being tested, as consumers in one region may have different preferences to those in another (Stirling & Witcombe, 2004).

The final phase is the further dissemination of farmer-preferred cultivars, for which it is necessary to develop linkages between plant breeding and seed production organisations. A crucial concern is whether the cultivar identified by PVS is officially released in the area, as if not it cannot be recommended by the local extension services, and public-sector seed multiplication is difficult. For this reason, it is important to include released cultivars in a PVS programme, as otherwise a time-consuming release process needs to be gone through which often requires data unavailable from PVS. We therefore believe it is important that varietal release committees should consider data on farmer perceptions and demands and not just on yield and disease susceptibility as is usually the case.

In participatory plant breeding, farmers contribute to selection from segregating material. PPB is a logical extension of PVS, and creates new variability but it also consumes more resources. PPB should be used when PVS fails to identify suitable cultivars, and can use cultivars identified through PVS as parents for crosses. As with PVS, there can be varying degrees of participation, from growing all generations on-station to training expert farmers to make crosses and selections. In economic terms, the returns from PPB are higher than from conventional breeding: it is a cheaper process, and the benefits to farmers are realised earlier (Witcombe et al. 2003; Virk et al. 2003).

Mother and baby trials

One of the most common formats for client-oriented evaluation is the “Mother and Baby” trial (Snapp 1999). In general, the aim should be to site one mother trial per village, if possible located prominently near roads or paths. This gives the opportunity for farmers and others to see and discuss the varieties. Mother trials consist of several germplasm lines which meet the identified needs of farmers, and usually consist of one block of a randomised complete block design. They allow direct comparison of all entries, provide statistically analysable yield data, and are replicated across farmers. Mother trials are more effective than replicated on-station trials as they sample more environments (Johnson et al. 1992).

Baby trials consist of one, or sometimes two, lines from the Mother trial. They are grown by individual farmers to compare with local check and are far more numerous than Mother trials. Baby trials provide statistically analysable data on perceptions and on the acceptance of new germplasm – it is often unnecessary to record yield data, and simple comparisons with the standard variety are sufficient. They are carried out using farmer’s management, in the correct domain, and provide a large number of replications.

Client-oriented work on salinity

We have adopted client oriented methods in our work on salinity in India and Pakistan. In the SW Punjab in India we worked with farmers in 2 districts of the SW Punjab. Bhandi village in Bathinda district is largely affected by poor-quality (generally sodic) groundwater (average EC 7.35 dS m-1, RSC 33.3 meq l-1). In contrast, Muktsar district, where several villages were used, has large areas with shallow watertables, which are at 2 – 2.5 m in December, but rise in January/February and in July/August to about 1 –1.5 m. Crops are frequently subjected to waterlogging, and the area is affected by seepage from the Rajasthan Feeder Canal such that farmers had given up growing wheat altogether. The groundwater is brackish, especially further away from the canal.

Before the project, farmers had no knowledge of tolerant wheat genotypes, despite the CSSRI, which has the national remit for breeding salt-tolerant crops, being within the same region of India, and complained about low yields. They were interested in salt-tolerant varieties with acceptable quality in order to resume wheat production on their land. Farmer selection began in year 1, on the basis of the extent of the problem and also, in Bathinda,

5

availability, as sowing there was carried out early and it was not possible to visit large numbers of farms before the normal sowing dates.

We offered the farmers seed of salt-tolerant material, and gave them 5 kg seed of each of several improved genotypes to grow alongside their own lines (generally PBW 343 from the Punjab Agricultural University, Ludhiana), using their own management. At Bathinda the improved material was from CSSRI, and at Muktsar farmers were given this material plus genotypes from Pakistan to grow under conditions of saline waterlogging.

Farmers’ opinions were sought, and where possible comparisons made between new and old varieties using paired t-tests. Grain yields are reported, and varieties were evaluated visually at germination, tillering and reproductive stages.

Figure 1: CSSRI varieties vs. PBW 343 under sodic irrigation, Bhandi, Indian Punjab,1999-2002

In general farmers were happy about varieties given to them, although they did express reservations about grain colour. In the trials using poor quality water at Bhandi, yields of the CSSRI material (usually KRL 1-4) in the first year were significantly (P = 0.028) higher than the farmers' varieties. The average yield of the CSSRI genotypes was 3.96 t ha-1, and from the PAU genotypes 3.30 t ha-1, an improvement of 16.7%. In year 2, several farmers at Bhandi reported large increases in the area grown over the previous year. For example, Jagga Singh kept KRL23 as he was impressed with it (it gave 50% greater yields than his PBW343), and sowed 0.25 ha, as well as 5 kg of KRL 1-4. Other farmers also noted high improvements in individual trials. Taking the mean of the improved material (KRL23, KNS57 and KRL 1-4) compared with PBW 343 for those trials with common varieties in the three years, there were highly significant (p<0.001) increases in yield over the local (Figure 1) when grown under conditions of irrigation water sodicity.

6

Figure 2: KRL varieties vs. PBW 343 under saline waterlogging, Muktsar, Indian Punjab, 2000-2002.

Under waterlogged conditions at Muktsar, overall (Figure 2) KRL23 gave higher yields than PBW 343, with the mean difference significant at p = 0.005. However, the difference between KRL1-4 and PBW 343 was not significant (p=0.096). The Pakistan genotypes (not shown in the figure) were highly variable.

A number of problems became evident over the course of the work. Every year there were a number of discrepancies between the varieties noted as having been sown, and those actually recorded by several farmers. Several farmers also failed to provide results. Due to administrative confusion in Muktsar district, many of the farmers undertook participatory trials on non-stressed land in 1999/2000. Many also were not given a local check for comparison. Again, one farmer who had ploughed up a poorly establishing crop was later only given tolerant material. Waterlogging and salinity became particular problems at this site in late January 2000 due to excessive rain. In some areas it was almost impossible to obtain details of farmers’ opinions, while in some trials, no controls were used and farmers only grew the new material.

At the end of the first year, 90% of farmers preferred the new material and said they would grow it again, while in the second year, two of the farmers sowed their entire farms to salt tolerant wheat. However, even by the end of the project, there was no sign that the local extension service would recommend it over their own varieties.

Farmer participatory trials were also carried out in several villages in the North West Frontier Province, Pakistan (NWFP). In the first year, 2 villages were selected, Gundheri, near Nowshera, and Kass Kali near Mardan. Gundheri is in an area extensively affected by salinity and sodicity, the underground watertable is saline (4 dS/m) at a depth of 2 m, and the soil profile stratified of river alluvium, mainly sandy loam/silt loam, strongly calcareous and alkaline, with organic matter less than 1.0 %, sufficient phosphorus and low nitrogen. Kass Kali is in a rice growing area with dense saline sodic soils and a hard pan below the surface. The soils are of loess deposit, the watertable is shallow (less than 1 m). Good quality canal water is available for irrigation. In 1999/2000, 3 villages were selected, Gundheri, Kass Kali and Sheik Kali (Peshawar district), although the trials at Kass Kali were not continued due to local political problems. Sheik Kali is a site of low salinity, with a shallow water table, using canal water for irrigation. In 2000/01 trials were in Gundheri, Kot Kashmir (Bannu district, with saline soil), and at Regi, near the University campus. Regi is canal irrigated, and has a very shallow water table with

7

low salinity: KRL-1-4, KRL-13 and Bakthawar 92 were sown after rice. Kot Kashmir, in Bannu district in southern NWFP, has low to moderate salinity with 350 mm annual precipitation. The area has mostly fine soil, with supplementary canal irrigation as required. In the final season, trials were held in Gundheri and at Kot Kashmir. Table 1: Details of the participatory trials in NWFP, Pakistan Year Site No. of

farmers Quantity of seed

Check genotype Test genotypes

1998/99 Gundheri 16 1 kg Bakthawar 92 SARC 3, ICP 3 Kass Kali 14 1 kg Inqlab 91 SARC 3, ICP 3 1999/2000 Gundheri 14 1 kg Bakthawar 92, Inqlab 91 ICP 3, Ghaznavi 98 Kass Kali 11 1 kg SARC 3, ICP 3 Sheik Kali 8 1 kg Inqlab 91 ICP 3 2000/01 Gundheri 13 5 kg Bakthawar 92 ICP 3, SQ 92, SARC 3 Regi 10 5 kg Bakthawar 92 KRL 1-4, KRL 13 Kot Kashmir 2 5 kg Inqlab 91 SARC 3, ICP 3, SQ 92 2001/02 Gundheri 13 10 kg F. Sarhad ICP 3, SQ 92, SARC 3 Kot Kashmir 2 10 kg F. Sarhad ICP 3, SQ 92, SARC 3

Details of the trials are given in Table 1. Trials were farmer-managed and unreplicated, with replication across farmers in a village. Farmers were given 1 kg of salt-tolerant material at the start of the project, increasing to 5 kg in 2000/2001 and to 10 kg in 2001/02, and asked to grow it alongside their normal wheat, following their normal cultural practices. Varieties included both Indian and Pakistani genotypes, as well as material developed through laboratory and field testing in the UK and Spain. During growth and at maturity, project staff and a group of farmers assessed cultivar performance. Three 1 m2 samples were collected for yield data per cultivar from 1999/2000 onwards. Data was analysed using paired t-tests for each comparison except for 1998/99, when farmers opinions only were obtained.

Table 2. Summary of matrix rankings, participatory trials, NWFP, Pakistan 1998/99 Gundheri Kass Kali Genotype 1st 2nd 3rd 1st 2nd 3rd SARC 3 7 7 2 6 5 3 ICP 3 4 3 9 3 4 7 Bakthwar 92 5 6 5 - - -

Table 2 summarises the matrix ranking of the genotypes for 1998/99, after discussions with the farmers. SARC 3 was ranked first by seven farmers at Gundheri, and by six at Kass Kali. At Gundheri, ICP 3 was ranked third by nine farmers, while opinions on Bakthawar 92 were split. At Kass Kalli ICP 3 was placed third by most farmers, while the locally recommended Inqlab 91 was split.

Inqlab 91 - - - 5 5 4

Results of the PVS from 1999/2000 to 2001/02 are summarised in Table 3, but it should be borne in mind that there were substantial differences between sites. SARC 3 outyielded the locally-recommended genotypes with which it was compared by 4.3%, although comparisons with individual genotypes were not significant. ICP 3 outyielded the local genotypes by 16.1%, with significant yield increases over Ghaznavi 98 and F. Sarhad. There was only enough data to compare SQ 2 with local material in 2000/01 and 2001/02. In these years it was superior by 6.3%, with a significant improvement over F. Sarhad. Notable from the results is the fact that Bakthawar 92, although released for general cultivation rather than as a salt-tolerant variety, did not differ significantly from the salt-tolerant material in any trials. Of the salt-tolerant material, on average yields were highest in ICP 3.

Table 3: Summary of participatory trials in NWFP, Pakistan, 1999/2000 – 2001/02 (grain yield, kg ha-1). The shaded columns are yields for the improved variety in each comparison (SARC v local, ICP 3 v local, and SQ 92 v local). Significance levels are from paired t-tests. Improved salt-tolerant genotype Local genotypes

SARC 3 Local ICP 3 Local SQ 92 Local

Ghaznavi 98 2741±141ns 2537±145 2941±119* 2446±138 Insufficient data Inqlab 91 2790±141ns 2836±167 3039±107ns 2658±149ns Insufficient data Bakthawar 92 2348±168ns 2248±205 2620±163ns 2524±233 2447±237ns 2524±233 F. Sarhad 2950±225ns 2757±191 3349±192*** 2665±166 3070±192** 2665±166 Mean 2707 2595 2987 2573 2759 2595

8

The trials in Pakistan were less conclusive than those in India: in some there were no significant differences between new and old varieties, while in others, although there were no genotypic differences, as a whole the new material was better than the old. Many more areas and farmers were involved than in India, but it was equally difficult to get the “socio-economic” information and hard data on yields and salinity levels.

Conclusions

Our work with many crops and in different countries has clearly shown that COB or PVS are better than conventional varietal testing. In many breeding programmes fewer resources are allocated to testing more advanced lines: COB and PVS reverse this and allocate more resources to this important material. We have released four varieties from just three crosses in rice: two in India (Virk et al. 2003); one in Nepal (Gyawali et al. 2007) and one in Bangladesh (Joshi et al. 2007).

Using client oriented methods permits all sectors of agricultural society to input into and benefit from plant breeding programmes. As men and women may have different priorities for desired traits, both must be active in the selection of seeds to be saved for the next season’s crop, and other marginalised groups. PVS allows farmers to evaluate varieties for all traits and to trade-off traits, for example grain yield v fodder yield, maturity, or grain quality and salt tolerance, against each other. It can provide valuable additional information to breeding programmes, for example on traits for particular environments or cropping systems, and the specific traits wanted by farmers in particular areas. Overall, PVS, PPB and COB test varieties under realistic management, across more physical niches as trials are replicated in more locations, and across social niches where food preferences might vary. In addition, seed from participatory trials rapidly enters the informal seed system, leading to accelerated uptake of improved material.

While the benefits of farmer participation are not universal, we feel that client-orientation is an essential requirement for the success of any breeding programme, although intensive collaboration with farmers is not always needed. Variants of these methods have been very successful in many crops, and, despite the mixed results from or own work, there appear to be no valid reasons why they should not also work with salinity. If this were done, we feel that salinity breeding programmes would have much greater impact and provide a more cost-effective use of the large resources that are devoted to them.

Acknowledgements We wish to thank the many staff of the NWFP Agricultural University, Peshawar, Pakistan, and the Punjab Agricultural University, Ludhiana, India for their help and cooperation in this work. In particular, we thank the farmers who not only took part in the experimentation, but who also provided guidance, advice and hospitality during the project. We also wish to thank colleagues in Bangor for constructive inputs into this paper, and Dr Ed Barrett Lennard, Department of Agriculture and Food, Western Australia, for his initial inspiration. We are grateful to Dr SA Quarrie for supplying seed of SQ 92, to Prof RH Qureshi for seed of SARC 3, and to the Director, CSSRI for seed of KRL 1-4 and KRL 23. This work was supported by grants from DFID and the European Commission (INCO-DC Contract No ERBIC 18CT 980305, funded by DG XII).

References Abdus Salam, Hollington, PA, Gorham, J, Wyn Jones, RG & Gliddon, C. 1999. Physiological genetics of salt tolerance in wheat

(Triticum aestivum L.): Performance of wheat varieties, inbred lines and reciprocal F1 hybrids under saline conditions. J. Agron. Crop Sci. 183: 145 156.

Araus, JL, Slafer, GA, Reynolds, MP & Royo, C. 2002. Plant breeding and drought in C3 cereals : what to breed for. Ann. Bot. 89: 925 940.

Barrett-Lennard, EG. 2003. The interaction between waterlogging and salinity in higher plants: causes, consequences and implications. Plant Soil 253: 35 54

Barrett-Lennard, EG, van Ratingen, P & Mathie, MH. 1999. The developing pattern of damage in wheat (Triticum aestivum L.) due to the combined stresses of salinity and hypoxia: experiments under controlled conditions suggest a methodology for plant selection. Aust. J. Agric. Res. 50: 129 136.

Ceccarelli, S. 1994. Specific adaptation and breeding for marginal conditions. Euphytica 77: 205 219 Ceccarelli, S & Grando, S. 1999. Decentralised participatory plant breeding. ILEIA Newsletter 15: 3/4 , 36 37. Ceccarelli, S, Erskine, W, Hamblin, J & Grando, S. 1994. Genotype by environment interaction and international breeding

programmes. Exp. Agric. 30: 177 187 Ceccarelli, S, Grando, S & Impiglia, A. 1998. Choice of selection strategy in breeding barley for stress environments. Euphytica

103: 307 318. Chambers, R. 1989. Institutions and practical change. Reversals, institutions and change, in Farmer First, R Chambers, A Pacey

and LA Thrupp, eds., Overseas Development Institute, London, pp 181 195

9

FAO. 2000. Global Network on Integrated Soil Management for Sustainable Use of Salt-affected Soils. FAO Land and Plant Nutrition Service. http://www.fao.org/ag/agl/agll/spush/intro.htm. Accessed 13 August, 2007

Flowers, TJ. 2004. Improving crop salt tolerance. J. Exp. Bot. 55: 307 319 Flowers, TJ & Yeo, AR. 1995. Breeding for salinity resistance in crop plants: where next? Aust. J. Plant Physiol. 22: 875 884 Gregorio, GB & Cabuslay, GS. 2005 Breeding for abiotic stress tolerance in rice, in Abiotic Stresses: Plant Resistance through

Breeding and Molecular Approaches, M. Ashraf and P.J.C. Harris, eds., Food Products Press, The Haworth Press Inc, pp. 513 544

Gregorio, GB, Senadhira, D, Mendoza, RD, Manigbas, NL, Roxas, JP & Guerta, CQ. 2002. Progress in breeding for salinity tolerance and associated abiotic stresses in rice. Fld. Crops Res. 76: 91 101

Gyawali, S, Sunwar, S, Subedi, M, Tripathi, M, Joshi, KD & Witcombe, JR. 2007. Collaborative breeding with farmers can be effective. Fld. Crops Res. 101: 88-95

Hillel, D, & Rosenzweig, C. 2002. Desertification in relation to climate variability and change. Adv. Agron. 77: 1 38 Hobbs, PR, & Gupta, R. K., 2003. Rice-wheat cropping systems in the Indo-Gangetic plains: issues of water productivity in

relation to new resource-conserving technologies, in Water Productivity in Agriculture: Limits and Opportunities for Improvement, J. W. Kijne, R. Barker, and D. Molden, eds., CABI International, Wallingford, UK. ISBN 0 85199 669 8, pp. 239 - 253

Hollington, PA, Akhtar, J, Aragüés, R, Gill, KS, Hussain, Z, Quarrie, SA & Rashid, A. 2002. Third Annual Report, EU INCO-DC Project ERBIC 18CT 980305 “Assessment and development of salinity, sodicity and waterlogging tolerant wheat genotypes for India and Pakistan.” Centre for Arid Zone Studies, University of Wales, Bangor. 127 pp.

Isla, R, Aragüés, R & Royo, A. 2003. Spatial variability of salt-affected soils in the middle Ebro valley (Spain) and implications in plant breeding for increased productivity. Euphytica 134: 325-334

Johnson, JJ, Alldredge, JR, Ullrich, SE, & Dangi, O. 1992. Replacement of replications with additional locations for grain sorghum cultivar evaluation. Crop Sci..32: 43-46.

Joshi, KD, Musa, AM, Johansen, C, Gyawali, S, Harris, D &.Witcombe, JR. 2007. Highly client-oriented breeding, using local preferences and selection, produces widely adapted rice varieties. Fld. Crops Res. 100: 107-116

Joshi, KD, Subedi, M, Rana, RB, Kadayat, KB & Sthapit, BR. 1997. Enhancing on-farm varietal diversity through participatory varietal selection: a case study for Chaite rice in Nepal. Exp. Agric. 33: 335-344

Maas, EV. 1990. Crop salt tolerance. In: Tanji, KK (ed) Agricultural Salinity Assessment and Management. ACSE Manuals and reports on engineering practice No. 71. New York, ASCE. ISBN 0-87262-762-4, pp. 262-304

Munns, R. 2005. Genes and salt tolerance: bringing them together. New Phytol. 167: 645-663 Quayyum, MA & Malik, MD. 1988. Farm production losses in salt-affected soils. Proc. 1st Nat. Congr. Soil Sci. Lahore,

Pakistan, October 1985, pp 356-364 Reynolds, MP, Mujeeb-Kazi, A & Sawkins, M. 2005. Prospects for utilising plant-adaptive mechanisms to improve wheat and

other crops in drought- and salinity-prone environments. Ann. Appl. Biol. 146; 239-259. Richards, RA. 1983. Should selection for yield in saline regions be made on saline or non-saline soils? Euphytica 32: 431-438 Richards, RA. 1992. Increasing salinity tolerance of grain crops: Is it worthwhile? Plant Soil 146: 89-98. Shannon, MC. 1997. Adaptation of plants to salinity. Adv. Agron. 60: 75-120 Singh, KN, Kulshreshtha, N and Chatrath, R. 2007. KRL 99 – a new wheat germplasm developed for salinity, sodicity and

waterlogging stresses. CSSRI Karnal, India. Available at http://www.cssri.org/krl99.pdf, accessed August 21, 2007. Snapp, S. 1999. Mother and baby trials: a novel trial design being tried out in Malawi. In: TARGET. The Newsletter of the Soil

Fertility Research Network for Maize-Based Cropping Systems in Malawi and Zimbabwe. Jan. 1999 issue. CIMMYT, Zimbabwe

Srivastava, JP & Jana, S. 1984. Screening wheat and barley germplasm for salt tolerance, in Salinity Tolerance in Plants – Strategies for Crop Improvement, RC Staples and GH Toenniessen, eds., New York, Wiley., pp. 273-283

Stirling, CM & Witcombe, JR. 2004. Farmers and Plant Breeders In Partnership. 2nd edition. Dept. for International Development (DFID) Plant Sciences Research Programme (PSP) and Centre for Arid Zone Studies (CAZS). Bangor, UK: CAZS, University of Wales Bangor.

Szabolcs, I. 1994. Soils and salinization, in Handbook of Plant and Crop Stress, Pessarakli, M., ed., New York, USA: Marcel Dekker, pp. 3-11

Virk, DS, Chakraborty, M, Ghosh, J, Prasad, SC & Witcombe, JR. 2005. Collaborative and consultative participatory plant breeding of rice for the rainfed uplands of eastern India. Euphytica 132: 95-108.

Virk, DS, Singh, DN, Prasad, SC, Gangwal, JS & Witcombe, JR. 2003. Collaborative and consultative participatory plant breeding of rice for the rainfed uplands of eastern India. Euphytica 132: 95-108

Witcombe, JR & Virk, DS. 2001. Number of crosses and population size for participatory and classical plant breeding. Euphytica 122: 451-462

Witcombe, JR, Joshi, A & Goyal, SN. 2003 Participatory plant breeding in maize: a case study from Gujarat, India. Euphytica 130: 413-422

10

Witcombe, JR, Joshi, A, Joshi, KD & Sthapit, BR. 1996. Farmer participatory crop improvement. I. Varietal selection and breeding methods and their impact on biodiversity. Exp. Agric. 32: 445-460

Witcombe, JR., Joshi, KD, Gyawali, S, Musa, AM, Johansen, C, Virk, DS and Sthapit, BR. 2005. Participatory plant breeding is better described as highly client-oriented plant breeding. I. Four indicators of client-orientation in plant breeding. Exp. Agric. 41: 299 - 319

Yeo, AR. 1999. Predicting the interaction between the effects of salinity and climate change on crop plants. Sci. Hort. 78: 159-174 Yeo, AR, Yeo, ME & Flowers, TJ. 1988. Selection of lines with high and low sodium transport from within varieties of an inbreeding

species, rice (Oryza sativa L.). New Phytol. 110: 13-19 Young, A. 1999. Is there really spare land? A critique of estimates of available cultivable land in developing countries. Env, Dev and

Sust 1: 3-18.