African Crop Science Journal African Crop Science Society ISSN: 1021-9730 EISSN: 2072-6589 Vol. 6, Num. 3, 1998, pp. 273-282

C. S. Wortmann, M. Silver-Rwakaikara¹ and J. Lynch²

CIAT, P.O. Box 6247, Kampala, Uganda
¹Department of Soil Science, Makerere University, P.O. Box 7062, Kampala, Uganda
²Department of Horticulture, Pennsylvania State University, University Park, PA 16802, USA

(Received 16 June, 1998; accepted 2 September, 1998)

Inadequate nitrogen nutrition is a major constraint to the performance of common bean (Phaseolus vulgaris L.) in sub-Saharan Africa. Twenty six lines selected for good performance under N-limiting conditions were further evaluated at two locations in Uganda to determine the mechanisms allowing good performance under soil N-limiting conditions. Two non-nodulating lines were included whose mean yield was 44% more with N application. The test lines showed little response to applied N implying that all fixed significant amounts of N₂; N derived from the atmosphere was estimated to average 15 and 43% of plant N for two locations using the natural abundance ¹⁵N technique. Mean yield of test lines ranged from 810 to 1450 kg ha^-1 across two locations with N-limiting conditions. Characteristics contributing to good performance under N-limiting conditions varied with genotypes which were grouped into two clusters, with four out-lying genotypes, based on N acquisition and utilisation characteristics. Nitrogen utilisation efficiency in vegetative growth was not important to yield under N-limiting conditions. Increased N acquisition during podfill was important to the yield of some genotypes, and presumably compensated for less biomass and plant N at the beginning of podfill. Yield was positively related to biomass and plant N at physiological maturity (R9), but not to N utilisation efficiency in biomass production at R9. Efficient partitioning of N from the vegetative parts to the grain was important to growth under N-limiting conditions, as was N utilisation efficiency in grain formation.

Key Words: BILFA, low N tolerance, N-limiting conditions, Phaseolus vulgaris

La nutrition inadéquate d'azote est une contrainte majeure pour la performance du haricot commun (Phaseolus vulgaris L.) en Afrique sub-saharienne. Vingt six lignées sélectionnées pour leur bonne performance sous les conditions limitantes d'azote ont évaluées dans deux localités en Uganda afin de déterminer les mechanismes liés à cette performance sous les conditions limitantes d'azote. Deux lignées non nodulantes ayant un rendement moyen de plus de 44 % avec l'application d'azote ont été inclues. Les lignées testées ont montré peu de réponse à l'application d'azote impliquant que toutes les lignées ont fixé des uantités importantes d'azote. L'azote obtenu de l'atmosphère a été estimé à une moyenne de 15 à 43% d'azote du plant dans deux localités en utilisant la technique de l'abondance naturelle de ¹⁵N. Le rendement moyen des lignées testées se rangeait de 810 à 1450 kg/ha à travers deux localités charactérisées par les conditions limitant d'azote. Les charactéristiques contribuant à la bonne performance sous les conditions limitantes d'azote variait pour les génotypes qui étaient regroupés en trois groupes avec 4 génotypes basés sur l'acquisition d'azote et sur les characteristiques d'utilisation. L'efficacité d'utilisation d'azote pendant la croissance végétative n'a pas été importante pour le rendement sous les conditions limitantes d'azote. L'acquisition accrue d'azote pendant le remplissage des gousses a été importante par rapport au rendement de quelques génotypes, et a compensé le déficit de la biomasse et de l'azote de la plante au commencement du remplissage des gousses. Le rendement a été positivement lié à la biomasse et l'azote de la plante à la maturité physiologique (R9), mais pas à l'efficience d'utilisation d'azote dans la production de la biomasse au stade de la maturité physiologique. La repartition effective d'azote des parties végétatives à la graine était importante pour la croissance sous les conditions limitantes d'azote comme a été l'efficience d'utilisation dans la formation des graines.

Mots Clés: BILFA, faible tolérance à l'azote, conditions limitantes d'azote, Phaseolus vulgaris

Inadequate soil N availability has been identified as a major constraint to bean production in sub-Saharan Africa with an estimated annual loss of production of 679,000 tonnes per year (Wortmann et al., in press). Unlike some legumes, bean typically derives little of its N from the atmosphere under low input agriculture although N2 fixation can be substantial if soil P is adequate (Giller et al., 1998). Bean is genetically variable for ability to obtain N from the soil, for N2 fixation and for partitioning of N (Graham, 1981; Rennie and Kemp, 1983). Thus, a regional effort called Bean Improvement for Low Fertility in Africa (BILFA) was initiated in 1990 to identify genotypes for different soil fertility-related constraints, including inadequate soil N with moderate soil acidity (Wortmann et al., 1994).

A better understanding of plant mechanisms of good performance under soil N-limiting conditions will enable more efficient germplasm improvement. A conceptual framework for analyzing N acquisition and utilisation in bean consists of several basic components (Lynch and Rodriguez, 1994): "rate and duration of N acquisition (through N fixation as well as soil N uptake), efficiency of N utilization in vegetative growth, timing of the transition to reproductive growth (seed filling), rate and duration of N accumulation in seed, and efficiency of N utilization in seed formation." The objective of this research was to determine the importance of the first, second and last of the above mentioned components on bean performance under N-limiting conditions.

Plant materials and locations. Twenty six genotypes previously selected from the BILFA for good performance under soil N-limiting conditions, two non-nodulating lines (isolines of Ex Rico 23 and INIAP 404), and two well-adapted and high yielding check varieties were grown on N-limited sites at Kawanda Agricultural Research Institute (KARI) and Namulonge Agriculture and Animal Research Institute (NAARI) in Uganda. The KARI (0° 24'N, 32° 31'E, 1190 m asl) soil was a Kandiudalfic Eutrudox clay loam and the NAARI (0° 31'N, 32° 36'E, 1140 m asl) soil was a Rhodic Kandiudalf sandy clay loam. Organic carbon was 1.4 and 1.9%, total N was 0.13 and 0.15%, and pH in water was 5.0 and 5.3, for the 0-20 cm soil depth at the KARI and NAARI sites, respectively.

Trial design and management. Two trials, with contrasting rates of N application, were conducted at each site during the first season of 1996 in a randomized complete block design of four replications. Plot size was 2 m x 5 m.

Fertilizer was applied to all plots at the rates of 22.4 kg P and 41.5 kg K ha^-1. Urea at 40 kg N ha^-1 was applied to the applied-N trial and no N was applied to the N-limited trial. Bean was sown in 50 cm rows to achieve a density of 20 plants m^-2 on 26 March at KARI and 3 April at NAARI. Fungicide (Banrot as seed dressing) and insecticide (carbosulfan) were applied for root rot and bean stem maggot (Ophiomyia spp.) control. Carbosulfon was applied prior to flowering for control of aphids (Aphis fabae) and flower thrips (Megalurothrips sjostedi Trybom).

Data collection. Ten plants per plot were randomly selected and sampled at the beginning of podfill (R8) which occurred between 49-59 days after sowing at KARI and 51-58 days at NAARI. Total above-ground dry weight (TDWR8) was determined. Ten plant samples were again taken at physiological maturity (R9); total above-ground dry weight (TDWR9), stover weight (ST) and grain dry weight (GDW) were determined (Table 1). The remaining plants were harvested when dry to determine plot grain yield. Severity and incidence of plant diseases were rated using a 1-9 method of scoring where 1 indicates absence of the disease and 9 indicates death of the plant due to the disease (CIAT, 1987). Nodulation was qualitatively rated as 1 (absent to very little nodulation) to 3 (well nodulated) in consideration of the number and size of nodules.

TABLE 1. Denotation of codes used in this paper

Nitrogen concentration (Kjeldahl method, Bremner, 1965) was determined for the samples collected at R8 of all genotypes, and for the stover and grain samples of 12 genotypes which contrasted in grain yield under soil N-limiting conditions. Total N acquisition at R8 (PNR8) and R9 (PNR9), N utilisation efficiency (biomass produced per unit of N or the inverse of N concentration; NUER8 and NUER9), N acquisition during podfill (NPF) and stover N (STN) were calculated (Table 1). Grain N (GN and GN%), grain produced per unit of PNR9 (GDW/PNR9), N harvest index (NHI), and remobilisation of N from vegetative parts to grain (NRMB = (PNR8 - STNR9)/PNR8 * 100) were also calculated.

Dinitrogen fixation was determined in samples taken at R8 using the 15N natural abundance method (Giller and Wilson, 1991), whereby percent of plant N derived from the atmosphere was estimated from the following equation:

                                                            d 15N(soil) -  d 15N (legume N)
%N from N2 fixation = ----------------------------- X 100
                        d 15N(soil) - B 

where B = -1.97 for bean (Mariotti, 1983). Analysis for ¹⁵N was done using a Europa Scientific Roboprep automated C/N analyser couple to a 20:20 isotope ratio mass spectrometer. The non-nodulating bean varieties, INIAP 404 and Ex Rico 23, each occurred once in each replicate as the reference crops.

Analysis of variance was calculated across locations (Analytical Software, 1996); nodule and disease scores were transformed to the square root, but means are based on the actual scores. Pearson's correlation coefficients were determined using variety means by N-limited trial. Values for the non-nodulating lines were excluded from these analyses. Cluster analysis, using the means from N-limited trials for N acquisition and utilization characteristics, was used to group similar genotypes.

Effects of location and N application. Mean biomass and grain yields in N-applied trials were slightly larger than in N-limited trials (Table 2). Performance of the non-nodulating lines, however, was increased with N application resulting in more biomass at R8 (+29%) and R9 (+52%), and in more grain yield (+44%) confirming that soil N was limiting. Presumably, nodulating lines obtained an amount of N through N2 fixation (Ndfa; N derived from the atmosphere) on the N-limited soils which was comparable to the amount of N recovered from the applied urea. Nitrogen level by genotype interaction effect was significant for biomass at R9, but not for biomass at R8 and grain yield.

TABLE 2. Trial means of 30 genotypes grown under N-limited and N-applied conditions at two locations

Plant biomass at R8 was less at KARI than at NAARI, but the differences were small at physiological maturity (Table 2). Although plant N at R8 was more at NAARI, N2 fixation was more at KARI (Table 3). The means at R8 were 14.7 and 43.3% Ndfa, 0.67 and 1.14 g per 10 plants Ndfa, and 8.2 and 16.0 kg ha-1 N2 fixed, at NAARI and KARI, respectively. Nodulation was less with N application (Table 2). The NAARI soil had more organic carbon and higher pH which maybe resulted in more N release from mineralisation of organic matter during the season, leading to suppression of N2 fixation.

TABLE 3. Trial means for performance of 12 contrasting genotypes grown at two locations in trials of N-limited and N-applied conditions. Sample size was 10 plants

Nitrogen acquisition during podfill (NPF) was much more when N was applied, and more at KARI (Table 3). Plant N at R9 and concentration and amount of N in grain were more where N was applied. The reduced rate of growth at NAARI following the R8 sampling may have been due to diseases. Common bacterial blight (Xanthomonas campestris pv. phaseoli), angular leaf spot (Phaeoisariopsis griseola) and floury leaf spot (Mycovellosiella phaseoli) were moderately severe at both locations. Angular leaf spot may have affected growth at both sites during the podfill stage; the correlation coefficients of angular leaf spot scores with yield were -0.56 and -0.71 for KARI and NAARI, respectively. Severity of the other diseases was not related to grain or biomass production.

Biomass produced per unit of N was more at KARI than at NAARI at R8, but NUER9 and NHI were similar across sites. Vegetative NUE was not much affected by N application, but NHI was less with N applied (Table 3). Remobilization of N from vegetative parts of the plant to grain was much increased with N application, and more at NAARI than at KARI where stover contained more N.

Nitrogen acquisition. Estimated amounts of N2 fixed at the beginning of podfill ranged from 0.59 to 1.29 g per 10 plants (7.8 to 17.0 kg ha^-1) for MMS250 and UBR(92)09, respectively (Table 4). Percent of Ndfa under N-limiting conditions varied from 20.7% for RWK 5 to 39.3% for UBR(92)09 when averaged across the two locations. At KARI, Ndfa reached 58.4% for UBR(92) and 24.1 kg ha^-1 for MORE 90040. Other genotypes with high Ndfa were MMS 253, RWR109, UBR(92)11, UBR(92)17, UBR(92)25, XAN 76, and MCM5001. Genotypes which obtained relatively little N from the atmosphere were UBR(92)12, MCM 1016, MMS 250, RWK 5 and CAL 96. This range of values for Ndfa are similar to results obtained with beans in northern Tanzania (Giller et al., 1998). Dinitrogen fixation and nodule scores were largely independent of biomass and plant N accumulated at R8 (Table 6). Some genotypes which were efficient in capturing scarce soil N derived proportionally less N from the atmosphere. UBR(96)17, UBR(96)25, XAN 76 and MCM 5001 combined good N2 fixation with good acquisition of soil N (Table 4). MMS 253, RWR 382 and UBR(92)09 had high proportions of Ndfa, but apparently were less efficient in acquisition of scarce soil N. BAT 1297, MMS 243 and MUS 97 captured soil N efficiently, but derived relatively little N from the atmosphere.

TABLE 4. Means across locations for 30 genotypes grown under soil N-limited conditions

TABLE 5. Nitrogen utilisation characteristics of 12 genotypes grown under soil N-limited conditions

The estimates of Ndfa might have been more accurate if non-nodulating lines had been planted more frequently in the field, as variations in available N and 15N in the soil can be considerable.

Nitrogen utilisation efficiency. Plant biomass at R8 was not predictive of plant biomass at R9 (r = 0.29), possibly due to the effects of angular leaf spot. MORE 90040, XAN 76, RWK 5, MCM 5001, UBR(92)25 and UBR(92)17 had relatively high biomass and plant N at R8, but of these only MORE 90040 had relatively high biomass at R9.

Nitrogen uptake during the podfill stage (NPF) was less with genotypes which had more plant N, and more with higher N utilisation efficiency (lower N concentration), at R8 (Table 5) suggesting that N stress in the vegetative biomass favoured NPF. NPF contributed more to PNR9 than to grain N. NPF had little effect on yield under N-limiting conditions for some genotypes, as much of the N stayed in the vegetative parts of the plant; NPF was negatively related to NHI but positively related to the ratio of stover N to PNR8. NPF was, however, apparently important for the good performance of CNF 5513 and some other genotypes. We cannot determine if the N acquired during podfill came primarily from the soil or from N2 fixation. Considering, however, the relatively high NPF at KARI where soil N was apparently least and N2 fixation was high, a considerable amount presumably was derived from the atmosphere.

Nitrogen concentration in the grain was negatively related to yield on low N soils confirming the importance of N utilisation efficiency in grain yield under N-limiting conditions (Table 6). Allocation of N to grain formation as indicated by NHI tended to be higher with the better performing genotypes. Remobilisation of N from vegetative plant parts to the grain varied widely with genotypes and was not consistently related to performance under N-limiting conditions.

Mechanisms of tolerance to N-limiting conditions. Heavier yield under N-limiting conditions, or low N tolerance, can be due to one mechanism or a combination of several mechanisms. Susceptibility to angular leaf spot was a significant determinant of yield in these trials. The four selected genotypes with the highest yields in the N-limited trials had a mean score of 3.9 for angular leaf spot while the six selected genotypes with the least yield had a mean of 5.9.

Ability to acquire N and achieve greater plant biomass by R8 may have contributed to higher yield for MORE 90040, XAN 76 and RWK 5. Dinitrogen fixation apparently was a major mechanism of early N acquisition for MORE 90040 and XAN 76, but RWK 5 acquired much of its N from the scarce soil N supply. Nitrogen utilisation efficiency in vegetative growth apparently did not contribute much to performance under N-limiting conditions.

TDWR9 was related to yield under N-limiting conditions and appeared to be important to MORE 90040 and CNF 5513. Nitrogen acquisition during podfill contributed to performance of MORE 90040 and CNF 5513 under N-limiting conditions. NPF was very high for UBR(92)20 with much plant growth during R8 but it had poor partitioning of N to the grain. NUE at R9 favoured performance of CNF 5513.

Nitrogen utilisation efficiency in grain formation was important in MORE 90040, CNF 5513 and RWK 5. Remobilisation of N to grain was important for genotypes which had less N acquisition during podfill, including XAN 76, RWK 5, MCM 5001 and UBR(92)25. Efficient partitioning of N to grain was important; NHI was relatively high with the more tolerant varieties.

TABLE 6. Pearson correlation coefficients for selected characteristics under soil N-limited conditions

1 NPF is related to TDWR9 on a site basis; r = 0.72 and 0.89 for KARl and NAARI,, respectively; NHI is negatively related to STNPNR8 (KARl-0.86, NAARI -0.70); Correlation coefficients were significant (P £ 0.05) when greater than the absolute value of 0.4.1 and 0.56 when for both and single sites, respectively
2 Denotations of the codes are given in Table 1

TABLE 7. Two clusters of genotypes, and four independent genotypes, based on similarities in N acquisition and utilisation characteristics

Eight genotypes fell into two clusters, while four genotypes were independent of clusters, when grouped according to similarities for N acquisition and utilisation characteristics (Table 7). NPF, NRMB, NUER9 and NUESD were significant (P = 0.05) determining variables of the clusters; Ndfa, NHI, PNR9, PNR8 and NUER8 were not significant determinants.

XAN 76, MCM 5001, UBR(92)25 and IBR(92) 43 were members of the first group which had intermediate yield and was distinctive for its high remobilisation of N to the grain. This cluster was also characterised by moderately high early N acquisition, in which N2 fixation was important, but also by less N acquisition during podfill resulting in relatively less plant N at R9.

Cluster II consisted of MORE 90040, RWK 5, MLB(92)89A and CAL 96, and was distinguished by little N acquisition during podfill. It had a moderately high mean yield but the members differed widely in yield. This cluster was not outstanding for any N efficiency characteristics, and inefficient in N2 fixation.

CNF 5513 was efficient for N utilisation at physiological maturity. It was also efficient in allocation of N to grain. UBR(92)17 was inefficient in N utilisation, but fixed much N2 and acquired N early with little N acquisition during podfill.

UBR(92)12 had high NUE in grain formation, but relatively less N2 fixation and moderately low N remobilisation.

UBR(92)20 acquired much N during podfill resulting in much total plant N at physiological maturity. It was inefficient in N remobilisation and allocation to the grain.

The results indicate potential for improving bean performance under N-limiting conditions through improving N acquisition and utilisation characteristics. These characteristics interacted little with growing conditions indicating high repeatability of performance. Relative yield was less stable across trials. Simple selection for yield under N-limiting conditions would identify better adapted genotypes, generally with improved NUE in grain formation; this implies, however, a decrease in the protein content of the harvested product which is of dietary importance for people throughout the region.

Improved N acquisition efficiency might be achieved through breeding using UBR(92)20 as one parent with CNF 5513 or a parent from cluster I. Nitrogen fixation might be improved through crosses of members of cluster I with UBR(92)17. Crosses of CNF 5513 with members of Cluster I should yield progeny of high N utilisation efficiency, combining efficiencies in N remobilisation and N allocation. Plant breeders may wish to focus on individual genotypes rather than clusters in selecting parents, however, due to variation in yield within clusters I and II.

A better understanding of mechanisms of performance under N-limiting conditions has been gained and several conclusions can be made.

Dinitrogen fixation was less, and a higher proportion of plant N was obtained from the soil, at Namulonge, apparently due to higher soil N availability.

Nitrogen utilisation efficiency in vegetative growth was generally not important to yield under N-limiting conditions.

Increased N acquisition during podfill stage was important to the yield performance of some genotypes, and presumably compensated for low biomass and plant N at R8.

Yield was weakly related to biomass and plant N at R9, but unrelated to the amount of biomass produced per unit of N at R9.

Efficient partitioning of N from the vegetative parts to the grain was important to N-limited yield performance.

Nitrogen utilisation efficiency in grain formation was very important to yield on N-limiting soils.

Angular leaf spot affected grain yield and may have distorted the effects of N-efficiency mechanisms as the plants approached physiological maturity.

Genotypes interacted less with growing conditions for N acquisition and utilisation characteristics than for yield. Improvements in N efficiency characteristics should be effective over a range of growing conditions.

Genotypes, and groups of genotypes, differed for efficiencies in N acquisition and utilisation suggesting potential for further improving performance on N-limiting soils through breeding.

We are grateful to: Ken Giller and Noreen Matheson for their advice on the use of the natural abundance technique and for the 15N analysis, and for Ken's helpful comments on the manuscript; Douglas Beck for providing the non-nodulating lines; the Directors of Kawanda ARI and Namulonge AARI who provided land and facilities needed for this research; and the Canadian International Development Agency and the United States Agency for International Development for the financial support given for this and other bean research in Africa.

Bremner, J.H., 1965. Total nitrogen. In: C.A. Black, D.D. Evans, J.L. White, L.E. Ensminger, F.E. Clark and R.C. Dinauer (eds.). Methods of Soil Analysis. Part 2. ASA, Madison, Wisc., USA. pp.1149-1178.

CIAT (Centro Internacional de Agriculture Tropical), 1987. Standard system for the evaluation of bean germplasm. van Schoonhoven, A. and Pastor-Corrales, M.A. (compilers). Cali, Colombia. 54p.

Giller, K.E., Amijee, F., Brodrick, S.J. and Edje, O.T. 1998. Environmental constraints to nodulation and nitrogen fixation of Phaseolus vulgaris L. in Tanzania. II. Response to N and P fertilizers and inoculation with Rhizobium. African Crop Science Journal 6(2):171-178.

Giller, K.E. and Wilson, K.J. 1991. Nitrogen fixation in tropical cropping systems. CAB International, Wallingford, 313 pp.

Graham, P.H., 1981. Some problems of nodulation and symbiotic nitrogen fixation in Phaseolus vulgaris L.: a review. Field Crops Research 4:93-112.

Lynch, J. and Rodriguez H., N.S. 1994. Photosynthetic nitrogen-use efficiency in relation to leaf longevity in common bean. Crop Science 34:1284-1290.

Mariotti, A. 1983. Atmospheric nitrogen is a reliable standard for natural 15N abundance measurements. Nature 303:685-687.

Rennie, R.J. and Kemp, G.A. 1983. N2 fixation in field beans quantified by 15N isotope dilution: II. Effect of cultivars of beans. Agronomy Journal 75:645-649.

Statistix for Windows, 1996. User's manual, Analytical Software, Tallahassee, FL.

Wortmann, C.S., Kirkby, R.A., Eledu, C.A. and Allen, D.J. 1998. An Atlas of Common Bean (Phaseolus vulgaris L.) Production in Africa. CIAT. In press.

Wortmann, C.S., Lunze, L., Ochwoh, V.A. and Lynch, J. 1994. Bean improvement for low fertility soils in Africa. African Crop Science Journal 3(4):469-477.

Copyright 1998, African Crop Science Society