African Crop Science Journal African Crop Science Society ISSN: 1021-9730 EISSN: 2072-6589 Vol. 6, Num. 2, 1998, pp. 171-178

K.E. Giller* , F. Amijee , S.J. Brodrick and O.T. Edje¹

Department of Biological Sciences, Wye College, University of London, Wye, Ashford, Kent, TN25 5AH, U.K.
¹SADCC/CIAT Regional Programme for Beans in Southern Africa, Agricultural Research Institute, Selian, Arusha, Tanzania

(Received 31 December, 1996; accepted 2 February, 1998)

Addition of P fertilizer (26 kg P ha^-1) was shown to increase dramatically the number of root nodules and seed yields of Phaseolus vulgaris in experiments on farmers' fields in the West Usambara Mountains in northern Tanzania. Both nodulation and seed yields varied enormously between the different sites even where fertilizers were applied. Nitrogen fixation was measured in two of these experiments using the ¹⁵N isotope dilution method with a non-nodulating variety of P. vulgaris as a non-fixing reference plant. Without addition of P fertilizer the amount of N₂ fixed was only 2-8 kg N ha-1 (25-27 %N derived from N₂-fixation) and this was increased to 8-16 kg N ha-1 (48-51 %N derived from N₂-fixation) with addition of 26 kg P ha^-1. Inoculation with Rhizobium together with P addition increased nodulation and N₂-fixation marginally above that with P alone but did not increase seed yields. Seed yields were only marginally improved by addition of P or N fertilizers at two sites in the Kilimanjaro/Meru region of northern Tanzania. At one of these sites (Kilacha) inoculation with P more than trebled the number of nodules but only marginally increased total N accumulation. Across all sites, addition of N fertilizer (45 kg N ha^-1) increased yields slightly above those found with P alone. The need for further assessments of nodulation and N₂-fixation under soil fertility conditions prevailing on farmers' fields is emphasised.

Key Words: Inoculants, isotope dilution, N₂-fixation, 15N, phosphorus, seed yield

RÉSUMÉ

L'addition de la fertilisation de P (26 kg P ha^-1) a beaucoup augmenté le nombre de nodules racinaires ainsi que le rendement en graines de Phaseolus vulgaris en champs des fermiers dans la région montagneuse ouest d'Usambar au nord de la Tanzanie. La nodulation et le rendement en graines variaient amplement entre les différents sites traités à la fertilisation. La fixation de l'azote atmosphérique était mesurée dans deux de ces essais par la méthode de dilution de l'isotope ¹⁵N avec une variété non nodulante de P. vulgaris servant de témoin non fixateur. En l'absence de la fertilisation, seulement 2 à 8 kg N ha^-1 soient 25-27%, dérivaient de la fixation de N atmosphérique tandis qu'avec l'addition de 26 kg P ha^-1, la fixation azotée était élevée à 8-16 kg ha^-1 soient 48-51%. La combinaison de Rhizobium avec le fertilisant P a légèrement augmenté la nodulation ainsi que la fixation de N par rapport au traitement de fertlisant P mais sans toutefois augmenter le rendement en graines. Le rendement en graines était légèrement amélioré par l'addition de P et de N dans les deux sites au sein de la région de Kilimanjaro/Meru au nord de la Tanzanie. Dans l'une de ces deux sites (Kilacha), le traitement au P a plus que tripplé le nombre de nodules tandis que la quantité d'azote total s'est légèrement augmentée par rapport au seul traitement de P. Le besoin de poursuivre ce genre de travail en vue d'examiner la nodulation et la fixation de N dans les conditions de sol qui prévalent en plein champ des fermiers est souligné.

Mots Clés: Inoculants, dilution de l'isotope ¹⁵N, fixation de N₂, phosphore, rendement en graines

Bean (Phaseolus vulgaris L.) is a major grain legume crop in Tanzania and many other countries of east and central Africa, but yields on farmers' fields are generally poor and often in the range of 200-500 kg ha^-1. Poor soil fertility is thought to be the major constraint for bean production across most of the region (Mwandemele and Nchimbi, 1990). As Phaseolus vulgaris can nodulate and fix nitrogen abundantly given favourable conditions (Giller and Wilson, 1991), which could contribute to improved soil fertility, invesigations were conducted into the reasons for poor growth and nitrogen fixation of Phaseolus in Tanzania.

A survey of nodulation and growth of Phaseolus vulgaris on farmers' fields in Tanzania indicated that poor nodulation was associated with small concentrations of extractable phosphorus and variable populations of indigenous Phaseolus rhizobia (Amijee and Giller, 1998). To investigate whether concentrations of plant-available P were so small as to be limiting nodulation, N₂-fixation and growth of P. vulgaris, on-farm experiments were carried out at six locations in northern Tanzania to test the response to N and P fertilizer addition and Rhizobium inoculation. At two of these locations the amounts of N₂ fixed were measured using the ¹⁵N isotope dilution method using a non-nodulating mutant of P. vulgaris (Davis et al., 1988; Pedalino et al., 1992) as the non-fixing reference crop.

Field experiments were conducted in farmers' fields selected at the villages of Irente, Ubiri, Miegeo and Maghamba around Lushoto in the West Usambara Mountains, Maji-ya-Chai on the eastern slopes of Mount Meru, and Kilacha to the south of Mount Kilimanjaro, between March and June 1989. Details of these locations, including soil properties are described by Amijee and Giller (1998). The same experiment to study the responses of P. vulgaris to N and P fertilizer application and inoculation with Rhizobium was conducted at each of the locations. In addition at Irente and Ubiri the ¹⁵N isotope dilution method was used to determine the amount of N₂ fixed.

A small seeded local variety 'Kachina' was used in all experiments. This variety is widely grown in the Lushoto area and had previously been observed to nodulate well under favourable conditions. The four treatments were a control (no fertilizer or inoculant added); +P (26 kg P ha^-1 as triple super phosphate); +P+N (26 kg P ha^-1 and 45 kg N ha^-1 as ammonium sulphate); and +P+I (26 kg P ha^-1 as TSP and inoculated with Rhizobium). All fertilizers were applied to the seed furrows and roughly mixed into the soil immediately before planting. The inoculum was a peat-based mixed Rhizobium inoculant containing Rhizobium etli strain CIAT 632, Rhizobium sp. (Phaseolus) strains CIAT 151 and CIAT 274 and R. tropici strain CIAT 899. The inoculant contained at least 8 - 109 cells/g of Rhizobium. A complete randomised block design was used with six replicates. Plot sizes varied between different sites according to the area of land available. At Irente the individual plots were 4.5 x 4 m, at Ubiri 4.5 x 3 m and at all other sites 2.5 x 4 m. At all sites rows were spaced at 50 cm intervals and a single seed sown at 10 cm intervals within rows to give a planting density of 2 x 10⁵ plants ha^-1.

At Irente and Ubiri, each plot contained a microplot of 2.5 m x 1 m to which ¹⁵N-labelled fertilizer was added (10 kg N ha^-1 of ammonium sulphate; 11 atom % ¹⁵N excess), whilst the remainder of the plot received 10 kg N ha^-1 of unlabelled ammonium sulphate. The ¹⁵N-labelled fertilizer was applied in solution four days prior to sowing together with sucrose added at a C:N ratio of 10:1 to stimulate immobilisation of the ¹⁵N in the soil so that it would be released slowly (Giller and Witty, 1987). The non-fixing reference plant was a non-nodulating genotype of P. vulgaris NOD125 (Davis et al., 1988) and this was planted as a single row in each microplot due to shortage of seed.

All harvesting was done on an area basis and the number of plants was counted as plant stands were variable. This allows the data to be presented on a per plant or per m² basis. The number of nodules was counted, internal nodule colour assessed and an estimate of average nodule size (mm) made at 28-35 days after sowing on plants from a 0.4 m² area. At physiological maturity (~75 days) shoot material was harvested from a 0.5 m² area, dried, weighed and the nitrogen content determined by an automated nitrophenol method following semi-micro Kjeldahl digestion. Plants of the non-nodulating genotype were harvested from the ¹⁵N microplots at the same time, the N contents determined and N in the digests concentrated by a modified Conway microdiffusion method. ¹⁵N enrichments were determined using a Micromass 622 mass spectrometer (VG Isogas, Cheshire, UK). The amount of nitrogen fixed by symbiotic N₂-fixation was calculated by the isotope dilution method using the equations given by Witty (1983) using the non-nodulating Phaseolus variety as the reference plant.

The remainder of each plot was harvested at 90-100 days after sowing. The number of pods per plant, number of seeds per pod, one hundred seed weight, and total seed yield were determined. Data from all sites were analysed by a combined analysis of variance as described by Gomez and Gomez (1984) using the Genstat programme (Payne, 1987).

Nodulation. The number of nodules at flowering was increased by application of P fertilizers at all sites apart from Kilacha (Table 1a,b). When N fertilizer was applied together with P (+P+N), nodulation was similar to the control. The number of nodules with P fertilizer alone (+P) was two to three times that in the unamended control plots and plots where Rhizobium inoculant was also applied (+P+I) tended to have more nodules. At Kilacha there was an increase in nodulation only when P was added together with N fertilizer or Rhizobium inoculant. The number of nodules differed between sites with particularly poor nodulation at Maghamba and Irente. There were no consistent effects of any of the treatments on nodule size, although size of nodules in the control treatment was larger at Kilacha (c. 5 mm) than at the other locations (c. 3 mm) (Table 1c). All nodules were pink/red in colour irrespective of treatment and location (data not presented).

TABLE 1. Effect of P and N fertilizer application and inoculation with a mixture of Rhizobium strains on a) nodule number per plant. b) nodule number per m² and c) nodule size (mm) of Phaseolus vulgaris at flowering (28-35 d) in on-farm experiments in northem Tanzania

Dry matter and N accumulation. Dry matter production and N uptake were almost doubled by the application of P fertilizer at Ubiri, Irente and Miegeo (Table 2a,b). At all locations, dry matter production was greatest with the addition of N and P fertilizers. Inoculation with Rhizobium with P fertilizer added had no greater effect on dry matter or nitrogen yield than P fertilizer alone except at Kilacha where addition of N or Rhizobium together with P gave an increase in N uptake.

TABLE 2. Effect of P and N fertilizer application and inoculation with a mixture of Rhizobium strains on a) dry matter production and b) nitrogen uptake of Phaseolus vulgaris at physiological maturity (~75 d) in on-farm experiments in northern Tanzania

There were large differences in both dry matter production and N uptake between the sites with the crop at Irente showing very poor growth even where N and P fertilizers were added together. Dry matter production and N uptake were greatest across all treatments at Kilacha, but the treatments with N and P fertilizers at Ubiri and Miegeo gave similar production and N yield to that at Kilacha.

Measurements of N2-fixation. No nodules were found on the roots of the non-nodulating genotype of P. vulgaris and plants were yellow with a much lower N content (mean %N = 2.4) than the nodulated plants (mean %N =3.1). The proportion of N derived from N2-fixation was almost doubled by P application at Ubiri and was increased by 25% at Irente and the amount of N derived from N₂-fixation was more than doubled by P application at both sites (Table 3). Inoculation with Rhizobium together with P gave a slight increase in the %N fixed at both sites, but only increased the amount of N derived from N₂-fixation at Ubiri. The amount of N derived from N₂-fixation at Irente was much less than that at Ubiri reflecting the large differences in total N accumulation between the two sites (Tables 2b and 3).

TABLE 3. Effect of P fertilizer application and inoculation with a mixture of Rhizobium strains on N₂-fixation by Phaseolus vulgans at physiological maturity (~75 d) in on-farm experiments at two locations (Ubiri and Irente) in the Usambara Mountains, northern Tanzania. N₂-fixation was measured by ¹⁵N isotope dilution using a non-nodulating P. vulgaris variety as a reference plant

Seed yields. Differences in seed yields between treatments and locations (Table 4) reflected the differences in dry matter production and N uptake (Table 2). Overall yields were greatest where P and N fertilizers were added, intermediate with P fertilizer alone or P plus Rhizobium inoculation and poorest in the control treatment. There was no significant increase in yield due to P fertilizer plus inoculation with Rhizobium compared to P fertilizer alone. The yield response to P application compared with the control treatment was proportionally most pronounced at Ubiri, Irente and Miegeo although the absolute response in yield differed from 70 kg ha^-1 at Irente to 370 kg ha^-1 at Ubiri and 490 kg ha^-1 at Miegeo (Table 4b). There were no significant differences in yield between treatments at Kilacha. The fertilizer treatment by location interaction was significant (P < 0.05) and orthogonal contrasts between the sites in the Usambara Mountains versus the two sites close to Mount Kilimanjaro (Kilacha) and on the foothills of Mount Meru (Maji-ya-Chai) showed highly significant differences in response between these regions. There were also significant differences between the three sites in the Usambara Mountains. Mean yields varied from 205 kg ha^-1 at Irente to 1905 kg ha^-1 at Kilacha.

The 100 seed weight differed little between sites or treatments (data not presented). The number of seeds per pod did not differ significantly between fertilizer treatments but was least at Irente and most at Ubiri and Miegeo (Table 4c). Addition of P fertilizer significantly increased the number of pods per plant indicating that the response in yield to fertilizers was due to the overall increase in vigour increasing the number of flowering nodes (Table 4d). The mean number of pods per plant varied enormously between sites ranging from 2.7 at Irente to 8.2 at Maji-ya-Chai.

TABLE 4. Effect of P and N fertilizer application and inoculation with a mixture of Rhizobium strains on a) seed yield per hectare and b) number of pods per plant of Phaseolus vulgaris at final harvest (~ 100 d) in on-farm experiments in northern Tanzania

Effects of fertilizers on nodulation and N₂-fixation. Application of small amounts of P fertilizer dramatically increased nodulation and N accumulation of P. vulgaris growing in farmers' fields in northern Tanzania (Tables 1 and 2). Nitrogen fixation was only measured in the treatments where no fertilizer N was applied. This was because the ¹⁵N-labelled fertilizer had been added in a small amount (10 kg N ha^-1) together with sucrose to stimulate immobilisation of the ¹⁵N in the soil in order to reduce the rate of change of ¹⁵N enrichment of the available soil N (Witty and Ritz, 1984; Giller and Witty, 1987). As a result a treatment involving the addition of unlabelled N fertilizer cannot be included without drastically altering the rate of change of the ¹⁵N-enrichment of available soil N, which in turn will increase the likelihood of errors in the estimation of N₂-fixation (Witty, 1983; Witty and Giller, 1991).

The improvement in N yield with P fertilizer added alone was shown by the large improvements in nodulation (Table 1) and by the isotope dilution measurements at Ubiri and Irente (Table 3) to be due to large increases in the proportion of the plant N derived from N₂-fixation from less than 29% without P fertilizer to 40-50% with added P. Subsequent work has shown that the smaller response to P fertilizers at Irente than at Ubiri was due to acute deficiency of potassium in the soil at this site (Smithson et al., 1993) which was indicated by our earlier soil analyses (Amijee and Giller, 1998). Unfortunately, K treatments were not included in these experiments. In most cases the population of indigenous rhizobia which could nodulate P. vulgaris was obviously sufficient to form large numbers of nodules and inoculation with Rhizobium did not further improve nodulation. Responses to Rhizobium inoculation are not generally found if the population of indigenous, effective rhizobia for the legume of interest is larger than 50-100 cells g^-1 soil (Singleton and Tavares, 1986; Thies et al., 1991). Populations of Rhizobium nodulating P. vulgaris in these soils where no response to inoculation was found ranged from 4 x 10² cells g^-1 soil to 5.5 x 10³ cells g^-1 soil (Amijee and Giller, 1998). Inoculation responses would only be expected at such sites if the inoculant strain had been selected to be both highly competitive for nodulation sites on the plant and more effective than the indigenous rhizobia (Giller and Wilson, 1991; Sylvester-Bradley et al., 1991).

At Kilacha numbers of nodules were not increased by addition of P fertilizer alone but increased enormously from 580 m^-2 with P fertilizer alone to 2400 m^-2 when Rhizobium inoculant was also added (Table 1). No indigenous rhizobia had been detected at Kilacha (<1 cell g-1 soil; Amijee and Giller, 1998) and the increase in nodule numbers following inoculation also indicated that the indigenous Rhizobium population was small, although rhizobia effective with P. vulgaris were present and formed some nodules on uninoculated plants in the field. Kilacha was also the only site at which there was no increase in total N uptake when P fertilizer was added alone. There were, however, small increases in N uptake at Kilacha when N fertilizer or Rhizobium inoculant was also added indicating that P was not limiting growth or nodulation at this site.

Seed yields. Yields were increased significantly due to application of P fertilizer alone at the three sites in the Usambara Mountains (Ubiri, Irente and Miegeo) where similar responses in yield to fertilizer treatments were found. However, overall yields were much poorer at the locations in the Usambara Mountains than at the two sites in northern Tanzania, probably due to a combination of climatic effects (caused by differences in altitude, aspect and rainfall) and effects of soil fertility. There were no significant effects of fertilizers or Rhizobium inoculation on seed yields at Kilacha even though increases in nodulation and N uptake were found with inoculation at earlier harvests (Tables 2 and 4).

Addition of N and P fertilizers together gave little improvement in yield over the use of P fertilizers alone, as simply adding P gave increases in nodulation and N₂-fixation sufficient to produce yields equivalent to those where 45 kg N ha^-1 was added. It has recently been shown that applications of more than 80 kg P ha^-1 were required for maximum yields of P. vulgaris at locations in the Usambara Mountains, although such rates of fertilizer use are unlikely to be feasible for the farmer (Smithson et al., 1993). If P fertilizers are frequently applied at the recommended rate used in these experiments (26 kg P ha^-1) then it is likely that P will accumulate in the soil to give residual benefits in the subsequent years so that with time this rate of fertilizer use will be sufficient to sustain optimum yields.

Wider relevance of the results. The experiment described was the first in which the non-nodulating mutant of P. vulgaris isolated in our laboratory has been used as a non-fixing reference plant to measure N₂-fixation in P. vulgaris in the field. The non-nodulating genotype grew well given the poor fertility of the soils and we conclude that this genotype is suitable for use as a reference plant for measuring N₂-fixation in P. vulgaris in the field.

Although there is a large body of accumulated information on the amounts of N₂ fixed by grain legumes in the tropics, virtually all of the reports are of experiments conducted on research stations with addition of P fertilizers or where the soil P content is good due to a history of P fertilisation and the well-documented residual effects of P fertilizers. The amounts of N₂ fixed measured in the plots without added P fertilizers at these two experiments in the West Usambara mountains of northern Tanzania (2-8 kg N ha^-1) are probably representative of the amounts of N contributed from N₂-fixation by P. vulgaris in many parts of east and southern Africa where seed yields are commonly as poor as 200 to 500 kg ha^-1.

Recently, attempts have been made to estimate nutrient depletion in Africa by constructing nutrient balance models on both a national and local basis (e.g. Stoorvogel and Smaling, 1990). In the nutrient balances of Stoorvogel and Smaling (1990) it is assumed that 50% of legume nitrogen comes from N₂-fixation, but our results indicate that the rates of symbiotic N₂-fixation on farmers' fields are much less than this. For such models to be accurate, and in order to implement realistic and well-focused programmes for improvement of legume crops in the tropics, we urgently need a much better understanding of what proportion of the N harvested in grain legumes grown on farmers' fields is actually fixed from the atmosphere and our current efforts are directed at using the natural abundance of ¹⁵N in beans and companion crops for such estimates.

We are grateful to Mr. Jerome Mwaimu and his staff for management of the experiments in the field and Mr. Jon Fear for assistance with the chemical analyses. The UK Overseas Development Administration provided part of the funding for this work.

Amijee, F. and Giller, K.E. 1998. Environmental constraints to nodulation and nitrogen fixation of Phaseolus vulgaris L. in Tanzania. I. A survey of soil fertility and root nodulation and multi-locational inoculation responses. African Crop Science Journal 6:159-169.

Davis, J.H.C., Giller, K.E., Kipe-Nolt, J.A. and Awah, F.M. 1988. Nodulation mutants in common beans. Crop Science 28:859-860.

Giller, K.E. and Wilson, K.J. 1991. Nitrogen Fixation in Tropical Cropping Systems. CAB International, Wallingford.

Giller, K.E. and Witty, J.F. 1987. Immobilized 15N-fertilizer sources improve the accuracy of field estimates of N₂-fixation by isotope dilution. Soil Biology and Biochemistry 19: 459-463.

Gomez, K.A. and Gomez, A.A. 1984. Statistical Procedures for Agricultural Research. 2nd Edn., J. Wiley and Sons, New York. pp. 680.

Mwandemele, O.D. and Nchimbi, S.F. 1990. Country reports - Tanzania. In: Proceedings of Workshop on Bean Varietal Improvement in Africa. Smithson, J.B. (Ed.), pp. 163-178. CIAT African Workshop Series No. 4, Colombia.

Payne, R. W. (Ed.). 1987. GENSTAT 5, Reference Manual. Clarendon Press, Oxford. 740 pp.

Pedalino, M., Kipe-Nolt, J.A. and Giller, K.E. 1992. Genetic and physiological characteri-zation of the non-nodulating mutant of Phaseolus vulgaris L. - NOD125. Journal of Experimental Botany 43:843-849.

Singleton, P.W. and Tavares, J.W. 1986. Inoculation response of legumes in relation to the number and effectiveness of indigenous Rhizobium populations. Applied and Environmental Microbiology 51:1013-1018.

Stoovogel, J.J. and Smaling, E.M.A. 1990. Assessment of Soil Nutrient Depletion in Sub-Saharan Africa: 1983-2000. Volume 1: Main Report (second edition). Report 28, Winand Staring Centre, Wageningen.

Smithson, J.B., Edje, O.T. and Giller, K.E. 1993. Diagnosis and correction of soil nutrient problems of common bean (Phaseolus vulgaris) in the Usambara Mountains of Tanzania. Journal of Agricultural Science (Cambridge) 120:233-240.

Sylvester-Bradley, R., Mosquera, D., Asakawa, M. and Sanchez, G. 1991. Promiscuity and responses to rhizobial inoculation of tropical kudzu (Pueraria phaseoloides). Field Crops Research 27:267-279.

Thies, J.E., Singleton, P.W. and Bohlool, B.B. 1991. Influence of the size of indigenous rhizobial populations on establishment and symbiotic performance of introduced rhizobia on field grown legumes. Applied and Environmental Microbiology 57:19-28.

Witty, J.F. 1983. Estimating N₂-fixation in the field using ¹⁵N-labelled fertilizer: some problems and solutions. Soil Biology and Biochemistry 15:631-639.

Witty, J.F. and Giller, K.E. 1991. Evaluation of errors in the measurement of biological nitrogen fixation using ¹⁵N fertilizer. In: Stable Isotopes in Plant Nutrition, Soil Fertility and Environmental Studies. pp. 59-72. FAO/IAEA, Vienna.

Witty, J.F. and Ritz, K. 1984. Slow release ¹⁵N fertilizer formulations to measure N₂-fixation by isotope dilution. Soil Biology and Biochemistry 16:657-661.

Copyright 1998, African Crop Science Society