The success of BNF in soybean in Brazil

Plant and Soil 252: 1–9, 2003. © 2003 FAO. Published by Kluwer Academic Publishers. Printed in the Netherlands. 1 The success of BNF in soybean in Brazil Bruno J. R. Alves1 , Robert M. Boddey & Segundo Urquiaga Embrapa Agrobiologia, Caixa Postal 74.505, Seropédica, 23851-970, Rio de Janeiro, Brazil. 1 Corresponding author∗ Received 7 January 2002. Accepted in revised form 23 July 2002 Key words: biological nitrogen fixation, Bradyrhizobium inoculants, Brazil, plant breeding, soybean Abstract Approximately forty years after commercial cropping of soybean in Brazil began, the total area under this crop has reached over 13 M ha with a mean productivity of 2400 kg ha−1 . Soybean varieties introduced from the USA and varieties rescued from early introductions in Brazilian territory were part of the Brazilian soybean-breeding programme which spread the crop from high to low latitudes. Disease-resistance, pest-resistance, tolerance to low fertility soils, as well as production of plants with pods sufficiently high above the ground for efficient mechanical harvesting, were all aims of the programme. Although BNF was not explicitly considered as a trait for selection in the breeding/selection programme, maximisation of biological nitrogen fixation (BNF) was favoured by conducting selection and breeding trials on soils low in N, in which the seeds were inoculated with efficient Bradyrhizobium inoculants but without N fertiliser application. Several efficient imported Bradyrhizobium strains were found to be unable to compete with native soil micro-flora and other previously-introduced Bradyrhizobium strains. Surprisingly, after being in the soil for many years one or two of these strains had become more competitive while maintaining their high BNF capacity. Today, these strains are included amongst the recommended Brazilian inoculants and have promoted significant improvements in grain yields. The breeding of soybeans in conditions that made grain yield highly dependent on BNF, and the continuous attention paid to the selection of Bradyrhizobium strains appropriate for the newly released varieties, have been the main contributors to today’s high yields and their great benefit to the Brazilian economy. There seems to be no reason why this ongoing research programme should not serve as an appropriate model to improve BNF inputs to grain legumes in other countries of the world. Introduction Information concerning the cultivation of soybean [Glycine max (L.) Merrill] dates from China in around 2500 BC. Nowadays this crop is spread throughout the world. The interest in soybeans can be explained by the elevated protein content of the seeds that can reach almost 40%. Soybean was introduced to Brazilian agriculture at the end of XIX century but was mainly grown on an experimental basis until the 1940s. The first report of significant production was registered for the State of Rio Grande do Sul in 1941 (total area ∗ FAX No: +55-21-2682-1230. E-mail: [email protected] planted 7651 ha with a production of 9146 t) (Medina, 1981). Growth in the area planted was slow until the 1960s when expansion began with a government campaign to increase wheat production. Soybean became the ideal summer crop when planted in rotation with wheat grown in the winter months. Soybean production has increased steadily (Figure 1), spreading from São Paulo and Rio Grande do Sul through Santa Catarina, Paraná and Minas Gerais in the 1960s and 1970s. More recently, production has expanded into the tropical savanna (Cerrado) regions of Mato Grosso do Sul, Goiás, western Bahia and towards Amazonia (Rondônia and Tocatins), especially in the State of Mato Grosso. In the past 5 years, the area planted to soybean (13.5 M ha in 2000) has overtaken that of maize (11.6 M ha) to occupy more area than any 2 Figure 1. Evolution of the area planted to soybeans (Mha) and total grain production (106 Mg) in Brazil (FAO, 2001). other crop. Today, soybean and its products constitute Brazil’s foremost agricultural export and the country produces 20% of the world’s soybeans (32.7 × 106 mg) second behind the largest producer (USA with 75.4 × 106 Mg). At 2400 kg ha−1 , yields are only slightly lower than those of the USA (2561 kg ha−1 ) and well above those of China (1820 kg ha−1 ) (FAO, 2001). Since in Brazilian soybean varieties almost all of the protein nitrogen produced is derived from BNF, the economic benefit in terms of a N-fertiliser saving is over US$ 2.5 billion per year. Adapting soybeans to regions, climate and soils through breeding As soybean is exotic to Brazil, much research was required to develop varieties adapted to the soils and climate and matching these varieties to efficient strains of Bradyrhizobium (see ‘Soja no Brasil’ Miyasaka and Medina, 1981). Selection/breeding of soybeans began in the early 1930s at the Instituto Agronômico de Campinas (IAC), in São Paulo State (latitude ≈23 ◦ S) and in the Secretariat of Agriculture as well as the Instituto de Pesquisas Agropecuárias do Sul, of the state of Rio Grande do Sul (temperate zone, 28 ◦ S). Due to favourable agronomic characteristics, as well as pest and disease tolerance, genetic material from the USA was generally best for planting in the South. The most planted varieties were Bragg, Davis, Hardee, Hill, and Hood (Vernetti et al., 1981). The lower (i.e. more northerly) latitude of the experimental areas of IAC in São Paulo favoured the improvement of soybean to overcome day-length limitations through crossing cultivars selected in the early 1940s (Abura, Aliança, Mogiana, etc.) with cultivars and lines from the USA belonging to the late maturity groups VII and VIII (Acadian, CNS, Pelican, etc.). This resulted in welladapted cultivars such as Santa Rosa that was widely planted during the 1960s and 1970s (Miranda et al., 1981). Since soybeans are reasonably well adapted to high latitudes and continental climates, selecting, and later breeding, for the southern region was not such a difficult problem. The real challenge came with the expansion of soybeans into the Cerrado (the central savanna of Brazil), which was encouraged by the low cost of land and by incentives in the form of subsidised credit, which attracted the farmers from the south of Brazil. This presented a difficult challenge for soybean breeders since yields would have to be maintained under an adverse climate, soil acidity/Al toxicity and physiological problems related to the short days. 3 Table 1. Grain yield (t/ha), plant and first pod heights (cm), flowering and maturity (days after emergence) of sixteen cultivars introduced in the cerrados of central Brazil (Spehar, 1995) Variety Yielda Santa Rosa 2.28 a IAC-2 2.21 ab IAC-4 2.18 abc UFV-1 2.06 abcd Florida 1.94 bcde Davis 1.92 bcdef Hardee 1.91 bcdef Paraná 1.89 cdef Viçoja 1.87 cdef Pampeira 1.79 defg Bienville 1.76 defg Planalto 1.67 efgh Bragg 1.60 fghi IAS-4 1.51 fghi Forest 1.41 hi Jupiter 1.36 i CV (%) 16.0 Table 2. Grain yield (t/ha), plant and first pod heights (cm), flowering and maturity (days after emergence) of thirteen cultivars selected in the cerrados of central Brazil (Spehar, 1995) Height Plant Pod Flowering Maturity Variety Yielda Height Flowering Maturity Plant Pod 56 90 57 57 49 47 48 55 39 36 32 35 42 31 48 85 42 47 43 46 44 36 39 37 34 34 28 37 29 29 31 66 109 117 114 118 98 101 104 95 102 96 93 97 92 95 99 134 Canarana BR 85 473-76 BR 85 487-88 BR 15 (Mato Gross) Cristalina BR 9 (Savana) Doko BR 40 (Itiquira) FT 11 (Alvorada) IAC-8 FT-Estrela FT-Eureka EMGOPA-304 CV (%) 3.79 a 3.50 a 3.41 bc 94 95 99 14 18 18 63 61 61 131 126 126 3.36 bcd 88 3.31 bcde 98 3.22 cdef 95 3.19 def 105 3.11 efg 79 3.03 fg 85 2.98 gh 104 2.80 gh 84 2.32 i 90 2.16 i 90 9.0 16 16 17 47 16 17 25 15 19 15 57 61 59 67 50 47 52 46 44 45 122 126 125 127 111 110 113 107 97 99 12 17 13 17 9 9 10 13 9 7 6 8 9 6 9 26 a Yields followed by the same letter are not statistically different at the 0.05 probability level, according to Duncan’s Multiple Range Test. Genetic variability in vegetative and reproductive cycles was demonstrated amongst soybeans growing in the states of São Paulo and Rio Grande do Sul and used to obtain late, and disease resistant, varieties as well as lines that grew well on poorly fertile soils. In this process genes for lateness such as PI 240664 (Kiihl and Miyasaka, 1970) were introduced into the breeding program. The late maturing cultivar Jupiter, obtained from an American breeding programme, along with varieties bred by the University of Viçosa, in Minas Gerais (latitude 21 ◦ C) such as ‘Viçoja’, lead to the development of varieties such as ‘Cristalina’ and ‘Doko’ which were introduced to the Cerrado region. Although both the improved and traditional varieties such as ‘Paraná’ and ‘Santa Rosa’ (Table 1), were well adapted to the day-length characteristics of the Cerrado they remained short in stature which posed a serious problem for mechanical harvesting (loss of pods low on stems) (Spehar et al., 1981). In recent years the breeding programme has largely been performed by the Embrapa soybean centre (Londrina, Parana, latitude 23.5 ◦ S) and the Embrapa Cerrado centre near Brasilia (latitude 16 ◦ S). As the soybean crop has expanded further into the Cerrado, a Yields followed by the same letter are not statistically different at the 0.05 probability level, according to Duncan’s Multiple Range Test. the importance of adaptation to lower latitudes has been overcome (Table 2). Present aims include higher yields, improved seed quality, production of cultivars with different maturities, tolerance to Al toxicity and low available Ca, as well as resistance to insects and diseases (Spehar et al., 1993; Spehar, 1995). Although BNF has not been specifically targeted as a selected trait, a recent report by Bohrer and Hungria (1998) showed that the products of these breeding program have produced soybean varieties which nodulate well with recommended strains for commercial inoculants of Bradyrhizobium. Breeding for biological nitrogen fixation In the 1960s the newly-created National Soybean Commission emphasised the need to prioritise BNF as an important part of the breeding programme. By this time it research results had shown that inoculated soybean was able to produce just as well as N-fertilised soybean (Weber, 1966a, b). Thus, the decision was made to omit nitrogenous fertilisers in the breeding programme and always inoculate the plants with effective Bradyrhizobium. Accordingly, surveys of imported and local ‘Rhizobium japonicum’ isolates were made. Local isolates were obtained from areas 4 Table 3. Dry matter accumulation and nodule dry weight of nine varieties of soybean inoculated with a mixture of 6 Bradyrhizobium sp. strains. Percentage of nodule occupancy by the strains for each variety is also showna Varieties Pérola Missões Bragg Paraná Hardee Planalto Santa Rosa Prata Pampeira Dry matter g/pot Dry nodules mg/pot Strains – % nodule occupancyb 513Re 527 532c 566 586 29W 8.51 9.43 9.42 9.72 13.52 10.21 13.92 14.80 12.55 932 806 1014 872 991 959 1121 1170 1281 0 0 0 0 0 0 2 0 5 98 93 92 86 100 100 98 100 88 0 0 2 3 0 0 0 0 2 2 5 4 11 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 2 2 0 0 0 0 0 2 a Study performed in sand filled Leonard jars with N-free nutrient solution. Data of Vidor, R.C.; Brose, E. and Pereira, J., 1978. IPAGRO – Secretary of Agriculture of Rio Grande do Sul (Jardim-Freire and Vidor, 1981). b Estimated by immuno-agglutination (Brockwell et al., 1977). Table 4. Effect of Bradyrhizobium strain on nodulation, nodule relative efficiency, ureide-N in xylem sap, seed yield and nitrogen harvest index of soybean. Pot experiment. (Neves et al., 1985) Bradyrhizobium strain Nodule dry weight mg plant−1 Relativea efficiency Ureide-N in xylem sap µg N ml−1 Nitrogen harvest indexb % Seed yield g plant−1 29W DF 395 SM1b 965 CB 1809 DF 383 230 a∗ 307 a 233 a 97 b 83 b 92 b 0.74 b 0.41 c 0.30 c 0.97 a 0.98 a 0.94 a 323 b 194 c 210 c 580 a 627 a 440 b 59 b 52 b 56 b 69 a 69 a 68 a 3.5 b 3.4 b 3.6 b 5.2 a 4.1 a 5.2 a a Relative nodule efficiency = (1 − H /C H ). 2 2 4 rmb Nitrogen harvest index = (Seed N/total N) × 100. ∗ Values followed by the same letter are not different at P = 0.05 level (Duncan’s multiple range test). where soybeans had been planted after inoculation with imported isolates. Strangely, in some cases these re-isolates were more effective than newly imported strains. A striking example is presented in the study by Vidor and his colleagues in Rio Grande do Sul (Jardim-Freire and Vidor, 1981). A ‘R. japonicum’ strain isolated form soybeans by J. R. Peres, a student of Johanna Döbereiner in Seropédica (Rio de Janeiro), was found to be far more successful in nodulating nine varieties of soybean than imported inoculant strains (Table 3). Based on the amount of N2 fixed per g of nodule biomass, Döbereiner et al. (1970) showed that soybean ‘R. japonicum’ strains could be divided into two groups. Subsequently, Neves et al. (1985) found that the more efficient strains (high ratio of N2 fixed to nodule dry weight) were hydrogen uptake positive (Hup+ ) and more efficient in transferring fixed N to developing grains than the Hup− strains (Table 4). These same strains were studied by Herridge and Peoples (1990) but they found no evidence of better efficiency in N2 fixing capacity for Hup+ strains. The latter study was carried out in the greenhouse while original study of Neves et al. (1985) was carried under natural, full-daylight, conditions. A follow-up study by Santos et al. (1996) comparing the performance of the wild-type Bradyrhizobium strain PJ17 and its Hup− mutant under full daylight and 30% shading, showed that relative efficiency (N2 fixed/g nodule) and the concentration of ureide in the stem sap were 5 Table 5. Effect of inoculation of soybean (cv BR-16) with different strains of Bradyrhizobium spp. on nodule occupancy, grain yield and total grain N at two sites (Londrina, Ponta Grossa) in Paraná (Hungria et al., 1997a, 1998) Site Inoculated strains Nodule occup- Grain yield Total grain N ancy (%)a kg ha−1 kg N ha−1 Londrina CB 1809 + CPAC 7 29 W + SEMIA 587 Non-inoculated Non inoc. + N∗ Coef. Variation % 45 65 – – 38 3789 a∗ 3110 b 2549 c 3455 ab 11.2 248 a 190 c 153 d 221 b 9.5 Ponta Grossa CB 1809 + CPAC 7 29 W + SEMIA 587 Non-inoculated Non inoc. + N∗ Coef. Variation % 44 54 – – 39 3215 a 2850 b 1834 c 2925 ab 10.8 209 a 174 b 110 d 189 ab 9.8 a % nodules occupied by inoculated strains. ∗ Values followed by the same letter are not statistically different (Tukey, P = 0.05). not significantly different under shaded conditions but both parameters were significantly higher under full daylight conditions. Now that two species of Bradyrhizobium that nodulate soybean have been re-classified as B. japonicum and B. elkanii (Kuykendall et al., 1992), it is clear that the Hup+ trait is mainly associated with B. japonicum, and that the only two strains used in commercial inoculants in Brazil until 1992 (29W and SEMIA 587), were B. elkanii and Hup− (Nishi et al., 1996; Rumjanek et al., 1993). Both strains are highly competitive. More efficient, Hup+ , B. japonicum strains such as CB1809 (SEMIA 586) have not been able to outcompete established populations of 29W and SEMIA 587. Nevertheless, a Bradyrhizobium strain, CPAC7, that was isolated from soybeans growing in a Cerrado field previously inoculated with CB1809, is a member of the CB1809 sero-group that has maintained the Hup+ trait as well as efficient translocation of N to grain. Remarkably, CPAC7 is able to compete for nodulation much more successfully against established populations of 29W and SEMIA 587 (Vargas et al., 1992, 1994). Increased competitiveness of CB1809 (RAPD techniques prove that CPAC7 is a variant of CB1809 – Nishi et al., 1996) maybe associated with polysaccharide composition (Scotti et al., 1993), which changed during adaptation of the strains to the soil (Coutinho et al., 1999). Inoculation with CPAC7 in the Cerrado can increase grain yields by up to 750 kg ha−1 . As a consequence, CPAC7 has been used as an inoculant in the region since 1992, and recently similar benefits have been demonstrated in two trials conducted in Paraná (Table 5; Hungria et al., 1998). It seems that Bradyrhizobium strains can, after several years of cycling between the soil and root nodules of soybean, change their capacity to compete for nodule occupancy. This has led research teams at the Embrapa centres to routinely search for efficient and competitive bradyrhizobia amongst strains isolated from soils used for many years for soybean production (Hungria et al., 1999). Almost all the arable land cropped to soybean has been inoculated at some time or other. As a consequence, elevated populations of Bradyrhizobium in the soil (1000–100 000 bacteria per g soil) are common. In 1992/1993 a network of experiments was established in the soybean production areas to examine the effect of re-inoculation. In the re-inoculation experiments performed in the Cerrado region, gains in grain production varied between 80 and 291 kg/ha (4– 12.5%) and in the Southern region the re-inoculation benefits were from 3.2 to 14.5% in grain production (Hungria et al., 1997a; Nishi and Hungria, 1996). Average effects of re-inoculation were calculated 4.5% increase in grain production. If the protein content was also considered the averaged increments reached 9% with a maximum of 25% at one site. This illustrates the results of Wani et al. (1995) who also found that re-inoculation often had little effect on soybean yields but N concentrations in seeds or other plant parts were increased over those of non-inoculated control plants. Effectiveness of BNF Soybeans demand large quantities of N that should be guaranteed by BNF. Otherwise depletion of the soil N reserves is likely to occur (Peoples et al., 1995). Occasionally, it has been observed that 10 or 12 days after emergence soybean leaves become pale green showing signs of N deficiency. This is the phase when seed reserves are being invested in nodule formation, which has led some workers to suggest that a starter dose of N fertiliser might help the plant to establish more quickly and eventually result in higher yields. However, this ‘starter dose’ of N has been tested many times in Brazil and if the inoculant is effective and efficient there is rarely any significant effect on grain 6 Figure 2. Effect of inoculation using a commercial Bradyrhizobium inoculant in the presence or absence of starter doses of 10, 20 or 30 kg N ha−1 of N fertiliser on the yield and total N in grain of the soybean variety IAC-2 cropped for the first year in the Cerrado region of Brazil (modified from Vargas et al., 1982). Same letters above the bars indicate no statistical differences between the means using Duncan’s multiple range test (P = 0.05). Figure 4. Effect of addition of 400 kg ha−1 of N fertiliser on the grain yield of the soybean variety ‘BR16’ inoculated with two different Bradyrhizobium strains in two consecutive growing seasons (92/93 and 93/94). In one of the treatments ∗ fertiliser was split into two doses of 200 kg, at planting and flowering, and in the other ∗∗ fertiliser was split into 10 equal doses during plant development (data from Nishi and Hungria, 1996). Figure 3. Effect of the application of gypsum and two doses of nitrogen fertiliser (30 and 60 kg N ha−1 ) on the soybean variety ‘Conquista’ in the Cerrado region of Brazil grown under ‘A’ conventional tillage, or ‘B’ no-tillage. N was applied at the beginning of reproductive stage (data from Reis et al., 2002). Error bars represent honest significant difference (P = 0.05, Tukey test). yields (Hungria et al., 1997b). Even in a soil of very low N availability, planted for the first time with soybean, Vargas et al. (1982) found that a starter dose of up to 30 kg N ha−1 had no significant effect on grain yield or grain N of soybean (Figure 2). After mid-pod fill, leaves become yellow and senesce exactly because their N is being re-translocated to the grain. Again, to the soybean producer this may appear that there is N deficiency at this time and a dose of N at the end of flowering might increase yields. Data of Hungria et al. (1997b) show that there is no significant benefit of adding N fertiliser at this time. More recently, Reis et al. (2002) showed that under conventional or no-tillage there was no benefit on grain yields of additions of 30 to 60 kg N ha−1 as ammonium sulphate at the start of pod filling stage (Figure 3). To show how symbiotic N2 fixation system can be very efficient in the soybean crop, Franco et al. (1978) added of up to 150 kg N ha−1 of N fertiliser (split in three doses) that increased total N yield of inoculated soybean, but produced no significant increase in grain yield. More recently, Nishi and Hungria (1996) added 400 kg ha−1 of N fertiliser to inoculated soybean (half at seeding and half at flowering, or split into 10 equal 7 doses during plant growth) and again found that grain yields were the same or lower than when the crop received no N fertiliser (Figure 4). Nevertheless, high amounts of N fertiliser usually lead to increases in leaf mass, as well as darker and larger leaves. Total dry weight may be higher than plants relying on BNF and almost always the total N accumulated by the whole plant is higher, but grain yields are usually unchanged. Role of BNF in soybean for the cropping system In the southern region of Brazil temperatures are high enough in winter, and there is sufficient rainfall to support year-round cropping. Virtually all soybean in this region is planted in rotation with winter crops such as wheat, oats or leguminous green-manures. In the Cerrado region, with its extended dry-season, soybean is generally followed by maize, sorghum or millet. The practice of no-tillage has spread very rapidly in recent years such that today almost 50% of these soybeanbased crop rotations are managed under no-till with no loss in productivity (Table 6). In well-managed fields 70–85% of the N required by soybeans is derived from BNF and present-day soybean yields range from 1.5 to 4.0 Mg ha−1 with BNF contributions in the order of 70 to 250 kg ha−1 of N. It is generally observed that when a winter cereal follows soybean, yields of these crops are higher than when preceded by maize or a fallow, and farmers attribute this to a net input of N derived from BNF by the soybean. However, the real benefit of BNF to the cropping system will depend on the proportion of plant N derived from BNF and the plant harvest index. Myers (1997) listed several N balance results (N exported in grain – BNF input to the crop) for soybean from different parts of the world that ranged from −132 to +104 kg N ha−1 . In the case of modern Brazilian soybean varieties, their high N harvest indices means that the quantity of N exported in the grain may be equal to, or higher than, the total N derived from the symbiotic process, leaving little in the soil (Alves et al., 1999; Zotarelli, 2000). In Brazil, the area under no-till is still increasing steadily, and under this system the BNF input to soybean and other legumes is favoured. The lower soil N availability under no-tillage compared to conventional (ploughed) systems, favours an earlier accumulation of nodule mass in soybeans (Figure 5), and the BNF contribution is consequently higher and the N balance for the systems more positive (Table 6). Figure 5. Nodule dry weight (mg plant−1 ) of soybeans growing under conventional (CT) and no-tillage (NT) systems conducted in the South (A - data from Zotarelli, 2000) and in the Cerrados (B – data from Reis et al., 2002) areas of Brazil. Differences between mean nodule mass for NT and CT are indicated by ∗ and ∗∗ for levels of significance of <0.05 and <0.01, respectively (Student ‘t’ test). One of the problems in computing N balances is that all root N is rarely taken into account since it is impossible to recover all fine roots as well as N exuded into the soil or derived from dying roots. Recently, a new technique has been developed that involves 15 N labelling, by feeding shoots with metabolites enriched with this isotope (for more details see Khan et al. 2002; Russel and Fillery, 1996). While no field data are yet available for soybean, assuming that roots of this crop behave in similar ways to those of faba bean (Vicia faba) and vetch (Vicia sativa), the amount of non-recoverable root N should account for approximately 30–35% of total plant N (Khan et al., 2002; McNeill et al., 1997). If extrapolations from this and other work conducted in Australia are valid (Peoples and Herridge, 2000), the true input of soybean BNF to the cropping system will be considerably more positive than previously thought. 8 Table 6. Grain yield, N derived from BNF (Ndfa) and N balancea in a soybean-based crop rotation planted under conventional and no-tillage in Paraná (after Zotarelli, 2000) Crop Crop conditions/history Grain yield Mg ha−1 Ndfa % N balance kg N ha−1 Maize Maize No-till. After oats (harvest 1997/1998) Conventional. After oats (harvest 1997/1998) 4.3 4.9 – – 16.2c 2.5c Soybean Soybean No-till. After oats (harvest 1997/1998) Conventional. After oats (harvest 1997/1998) 5.9 5.4 80.9 74.1 −6.9d∗ −24.3d Wheat Wheat No-till. After soybean (harvest 1998) Conventional. After soybean (harvest 1998) 2.5 2.3 – – −16.7c −15.4c Lupins Lupins No-till. After oats (harvest 1998) Conventional. After soybean (harvest 1998) 9.4b 11.2b 74.4∗ 68.8 202.5e 216.5e Maize Maize No-till. After lupins (harvest 1998/1999) Conventional. After lupins (harvest 1998/1999) – – −28.1c −46.3c 9.3∗∗ 7.8 ∗ , ∗∗ Difference between means significant at P < 0.05 and 0.01, respectively. a Partial total N balance = Total N exported in grain – BNF input – N fertiliser added. No estimates of gaseous or leaching N losses were made. b Total shoot dry matter. c Difference between N added as fertiliser and that exported in grain. d Difference between total grain N and N fixed by plant (roots included). e Total BNF contribution (15 N natural abundance technique). Conclusions Today, many competent farmers (especially those in the Cerrado region) are able to obtain soybean yields in excess of 4.0 Mg ha−1 . This quantity of grain contains between 250 and 280 kg N ha−1 , in most cases virtually all derived from BNF. The BNF system is so efficient that attempts to increase grain yields by addition of N fertiliser are hardly ever successful if the plants have been effectively inoculated with the recommended Bradyrhizobium strains. This great success of BNF technology in Brazil is a result of several component efforts: 1. The selection/adaptation to the shorter days, tolerance to pests and diseases and acid-soil conditions which allowed soybean to be grown in just about any region of Brazil. 2. The insistence that during breeding plants should be inoculated with the best available Bradyrhizobium strains and no N fertiliser should be applied. 3. The continuous interchange of information between plant breeders, agronomists and rhizobiologists, which led to the selection of Bradyrhizobium strains in parallel with soybean breeding. There seems to be no apparent reason why these principles cannot be applied to the introduction of other legumes in other parts of the world. References Alves B J R, Lara-Cabezas W A R, David E A and Urquiaga S 1999 Balanço de N em soja estabelecida em um Latossolo vermelho escuro do Triângulo Mineiro em condições de plantio direto e preparo convencional do solo. In Proceedings of Congresso Brasileiro de Ciência do Solo. CD-ROM. Brasília, July 1999. Sociedade Brasileira de Ciência do Solo, Viçosa-MG. Bohrer T R J and Hungria M 1998 Avaliação de cultivares de soja quanto à fixação biológica de nitrogênio. Pesq. Agropec. Bras. 33, 937–952. Brockwell J, Schwinghamer E A and Gault R R 1977 Ecological studies of root-nodule bacteria introduced into field environments. V. A critical examination of the stability of antigenic and streptomycin-resistance markers for identification of strains of Rhizobium trifolii. Soil Biol. Biochem. 9, 19–24. Coutinho H L da C, Kay H E, Manfio G P, Neves M C P, Ribeiro J R A, Rumjanek N G and Beringer J E 1999 Molecular evidence for shifts in polysaccharide composition associated with adaptation of soybean Bradyrhizobium strains to the Brazilian Cerrado soils. Environ. Microbiol. 1, 401–408. Dobereiner J, Franco A A and Guzman I 1970 Estirpes de Rhizobium japonicum de excepcional eficiência. Pesq. Agropec. Bras. 5, 155–161. FAO (2001) Internet site http://www.fao.org/. 9 Franco A A, Fonseca O O M and Marriel I E 1978 Efeito do nitrogênio mineral na atividade de nitrogenase e nitrato redutase, durante o ciclo da soja no campo. Rev. Bras. Ci. Solo 2, 110–114. Herridge D F and Peoples M B 1990 Ureide assay for measuring nitrogen fixation by nodulated soybean calibrated by 15 N methods. Plant Physiol. 93, 495–503. Hungria M, Vargas M A T and Campo R J 1997a A Inoculação de Soja. Embrapa CNPSo Circular Técnica 17, Embrapa Soja, Londrina-PR. 28 p. Hungria M, Vargas M A T, Campo R J and Galerani P R 1997b Adubação nitrogenada em soja? Embrapa CNPSo Com. Técnico 57, Embrapa Soja, Londrina-PR. 4 p. Hungria M, Boddey L H, Santos M A and Vargas M A T 1998 Nitrogen fixation capacity and nodule occupancy by Bradyrhizobium japonicum and B. elkanii strains. Biol. Fertil. Soils 27, 393–399. Hungria M, Vargas M A T, Andrade D de S, Campo R J, Chueire L M de O, Ferreira M C and Mendes I C 1999 Fixação biológica de nitrogênio em leguminosas de grãos. In Inter-relação Fertilidade, Biologia do Solo e Nutrição de Plantas. Eds. J O Siqueira et al. pp. 597–620. Sociedade Brasileira de Ciência do Solo, Universidade Federal de Lavras, Lavras-MG. Jardim-Freire J R and Vidor C 1981 Melhoramento de cultivares no Brasil: Cultura da soja nos Cerrados. In A Soja no Brasil. Eds. S Miyasaka and J C Medina. pp. 417–425. Instituto Agronômico de Campinas, Campinas-SP. Khan D F, Peoples M B, Chalk P M and Herridge D F 2002 Quantifying below-ground nitrogen of legumes. 2. A comparison of 15 N and non isotopic methods. Plant Soil 239, 277–289. Kiihl R A S and Miyasaka S 1970 Trabalhos de melhoramento de soja no Estado de São Paulo. In Proceedings of I Simpósio Brasileiro de Soja. Campinas, July 1970. Instituto Agronômico de Campinas, Campinas-SP. Kuykendall L D, Saxena B, Devine T E and Udell S E 1992 Genetic diversity in Bradyrhizobium japonicum Jordan 1982 and a proposal for Bradyrhizobium elkanii sp. nov. Can. J. Microbiol. 36, 501–505. McNeill A M, Zhu C and Fillery I R P 1997 Use of in situ 15 Nlabelling to estimate the total below-ground nitrogen of pastures legumes in intact soil-plant systems. Aust. J. Agric. Res. 48, 295– 304. Medina J C 1981 Introdução e evolução da soja no Brasil. In A Soja no Brasil. Eds. S Miyasaka and J C Medina. pp 17–20. Instituto Agronômico de Campinas, Campinas-SP. Miranda M A C de, Miyasaka S, Mascarenhas H A A and Rossetto C J 1981 Melhoramento de cultivares no Brasil no Estado de São Paulo. In A Soja no Brasil. Eds. S Miyasaka and J C Medina. pp 311–324. Instituto Agronômico de Campinas, Campinas-SP. Miyasaka S and Medina J C A 1981 A Soja no Brasil. Instituto Agronômico de Campinas, Campinas, SP, Brazil. 1062 pp. Myers R J K 1997 Soybean and sustainability of production systems. In Soybean Feeds the World. Ed. B Napompeth. pp 367–372. Proceedings of the Chiang Mai Soybean Conference, February 1994. Kasetsart University Press, Bangkok. Neves M C P, Didonet A D, Duque F F and Dobereiner J 1985 Rhizobium strain effects on nitrogen transport and distribution in soybeans. J. Exp. Bot. 36, 1179–1192. Nishi C Y M and Hungria M 1996 Efeito da reinoculação da soja [Glycine max (L.) Merril] em um solo com população estabelecida de Bradyrhizobium com as estirpes SEMIA 566, 586, 5019, 5079 e 5080. Pesq. Agropec. Bras. 31, 359–368. Nishi C Y M, Boddey L H, Vargas M A T and Hungria M 1996 Morphological, physiological and genetic characterisation of two new Bradyrhizobium strains recently recommended as Brazilian commercial inoculants for soybean. Symbiosis 20, 147–162. Peoples M B and Herridge D F 2000 Quantification of biological nitrogen fixation in agricultural systems. In Nitrogen Fixation: From Molecules to Crop Productivity. Eds. F Pedrosa et al. pp. 519–524. Kluwer Academic Publishers, Dordrecht. Peoples M B, Gault R R, Lean B, Sykes J D and Brockwell J 1995 Nitrogen fixation by soybean in commercial irrigated crops of Central and Southern New South Wales. Soil Biol. Biochem. 27, 553–561. Reis E H S, Lara Cabezas W A R, Alves B J R and Caballero S U 2002 Suplementação de nitrogênio mineral na cultura da soja (Glycine max) estabelecida em sistema plantio direto e convencional. In Proceedings of Congresso Brasileiro de Soja. pp. 215. Foz do Iguaçu, June 2002. Embrapa Soja, Londrina. Rumjanek N G, Dobert R C, Vanberkum P and Triplett E W 1993 Common soybean inoculant strains in Brazil are members of Bradyrhizobium elkanii. Appl. Env. Micro. 59, 4371–4373. Russell C A and Fillery I R P 1996 In situ 15 N labelling of lupin below-ground biomass. Aust. J. Agric. Res. 47, 1035–1046. Santos V A F, Neves M C P and Rumjanek N G 1996 Efficiency of soybean nodules related to rhizobia hydrogenase is influenced by light level. R. Bras. Fisiol. Veg. 8, 15–21. Scotti M R M M L, Neves M C P, Paiva E and Dobereiner J 1993 Effect of soybean roots on strain competitivity and protein profile of Bradyrhizobium japonicum adapted to cerrado soils. An. Acad. Bras. Ci. 65, 427–438. Spehar C R 1995 Impact of strategic genes in soybean on agricultural development in the Brazilian tropical savannahs. Field Crop. Res. 41, 141–146. Spehar C R, Vilela L, Mello de Souza, P I de and Vieira R D 1981 Melhoramento de cultivares no Brasil: Cultura da soja nos Cerrados. In A Soja no Brasil. Eds. S Miyasaka and J C Medina. pp. 341–345. Instituto Agronômico de Campinas, Campinas-SP. Spehar C R, Monteiro P M F O and Zuffo N L 1993 Soybean breeding in Central Brazil. In Procceedings of the Simpósio sobre a Cultura da Soja nos Cerrados. pp. 229–251. Uberaba, July 1992. Potash and Phosphate Institute, Piracicaba – SP. Vargas M A T, Peres J R R and Suhet A R 1982 Adubação nitrogenada, inoculação e épocas de calagem para a soja em um solo sob Cerrado. Pesq. Agropec. Bras. 17, 1127–1132. Vargas M A T, Mendes I C, Suhet A R and Peres J R R 1992 Duas novas estirpes de rizóbio para a inoculação da soja. Embrapa CPAC Com. Técnico 62, Embrapa Cerrados, Planaltina-DF. 3 p. Vargas M A T, Suhet A R, Mendes I C and Peres J R R 1994 Fixação biológica de nitrogênio em solos de Cerrados. Embrapa Cerrados, Planaltina, DF, Brazil. 83 pp. Vernetti F de J, Bonato E R, Cunha Gastal M F de, Dall’Agnol A, Raupp A A A and Velloso J A R de O 1981 Melhoramento de cultivares no Brasil pelo Instituto de Pesquisa Agropecuária do Sul. In A Soja no Brasil. Eds. S Miyasaka and J C Medina. pp. 283–292. Instituto Agronômico de Campinas, Campinas-SP. Wani S P, Rupela O P and Lee K K 1995 Sustainable agriculture in the semi-arid tropics through biological nitrogen fixation in grain legumes. Plant Soil 174, 29–49. Weber C R 1966a Nodulating and non-nodulating soybean isolines: Agronomic and chemical attributes. Agron. J. 58, 43–46. Weber C R 1966b Nodulating and non-nodulating soybean isolines: II Response to applied nitrogen and modified soil conditions. Agron. J. 58, 46–49. Zotarelli L 2000 Balanço de nitrogênio na rotação de culturas em sistemas de plantio direto e convencional na região de Londrina – PR. MSc Thesis, Universidade Federal Rural do Rio de Janeiro, Seropédica-RJ. 133 pp.