Bambara groundnut V. subterranea is an indigenous leguminous crop in
Africa. It is an important source of protein in the diet, particularly to poorer
communities who can hardly afford expensive animal protein. There is currently,
a growing concern that meat is a source of various non communicable ailments,
such as high blood pressure, cancer and lowered immunity to infectious diseases
(Sesay et al., 1996). Studies show that Bambara groundnut is a rich energy
and protein source with 16-25.2% protein, 42.8-65% carbohydrate on a dry weight
basis and that the seed protein content compares well with that of valued grain
legumes (Linnemann and Azam-Ali, 1993). The crop has good agronomic values in
that it is endowed with the advantages of being resistant to insect pests and
diseases, has a wide range of genetic variability, drought resistant, high adaptation
to marginal conditions and a good nitrogen fixer (Azam-Ali, 1992; Sesay et
al., 1996). It is therefore a crop suited to low input farming systems that
makes it popular amongst farmers with limited resources. However, the problem
of securing consistent, satisfactory yields in Bambara groundnut is of practical
concern in the region. Yields of Bambara groundnut are not only generally low,
varying between 650 and 850 kg ha-1 (Stanton et al., 1966)
and perhaps as low as 60-110 kg ha-1 (Linnemann and Azam-Ali, 1993)
but also, notoriously erratic. Other workers have reported yield levels of Bambara
groundnut, for instance 300 kg ha-1 for current landraces and 399-1310
kg ha-1 for promising entries (Mulila-Mitti and Kanenga, 1996). Occasionally,
yields of as high as 2000 and 3900 kg ha-1 have been observed in
experimental plots (Mulila-Mitti and Kanenga, 1996).
Linnemann and Azam-Ali (1993) had associated low yields to poor germination and late establishment while Squire et al. (1996) indicated that variability in growth and development of individual plants is the main cause of low yields in this crop. Yield, however, is a complex character determined by a number of components that follow a developmental sequence (Karikari and Tabone, 2003). The organs developed earlier can have a profound influence on those produced later (Grafius, 1978; Hamid and Grafius, 1978). Thus selection and improvement for yield requires manipulation of quantitative characters that may correlate and influence among themselves and yield. However, limited work is available that correlates paths of influence among developmental variables in Bambara groundnut for instance those of Karikari (1972), Linnemann (1994) and Karikari and Tabone (2003). The latter noted that yield is a complex terminal outcome of growth to which there are diverse and interrelated developmental tracks. Thus no single character that is absolutely important for yield, necessitating development of selection criteria comprising both vegetative (first order components) and reproductive (second order components) variables and hence aid breeders in improvement of yield through a component selection approach. A detailed examination of the nature of association among the components will assist in developing a more reliable criterion for selection and minimize risks of component compensation in yield improvement.
The present investigation analyses more critically the interrelationships and influential patterns among yield and components of yield in Bambara groundnut using the path coefficient analysis tool after Wright (1921) and as revised by Dewey and Lu (1959). Results of this investigation will shed light on variables to consider in improvement of bambara groundnut for yield improvement in this location and other similar ecologies.
MATERIALS AND METHODS
An experiment was conducted in the screen house of the Department of Crop Science and Production at the Sokoine University of Agriculture (SUA), Morogoro, Tanzania during the January-May 2000 growing season.
The materials used in this study were obtained from the collection of Bambara groundnut germplasm maintained by the Department of Crop Science and Production. The collected materials consisted of a mixture of landraces from different areas of Tanzania, viz. Bukoba (West Lake), Songea (South) and Mbinga (South West).
The experiment was laid out in a Completely Randomized Design (CRD) in which nine landraces (Table 1) were replicated four times in the screen house. Planting was done in 5 L size plastic pots, each containing 2 plants. The pots, each regarded as a plot, were laid on a bench in four rows at spacing of 50 cm both within and between rows. Soil was collected from the Crop Museum plots of SUA and 4 kg of soil were placed in each 5 L plastic pot. Three seeds were sown per pot and the seedlings were thinned at 3 weeks after sowing, leaving two plants per pot. Gap filling was done during the first week after emergence. Appropriate husbandry practices were done to ensure adequate moisture and protection against pests.
Data collected from each plot included number of days to first flower, average
values for plant height and number of full open trifoliolate leaves, leaf length
of median leaf blades and petiole length. Other variables included average weight
of 100 seeds, number of pods per plant and seed yield per plant. The mean values
of the variables for each plot were subjected to ANOVA and phenotypic correlations
using SAS (1997) computer software. Phenotypic correlations were further partitioned
into components of direct and indirect effects using path coefficient analysis
after Wright (1921) and Dewey and Lu (1959). The coefficients were obtained
by solving sets of simultaneous equations arranged in matrix notation which
show the relationships between correlations and path coefficients as shown below:
||Effects of vegetative and reproductive variables on seed yield per plant
and effects of first order components on weight of 100 seeds:
P36+r14 P46+r15 P56
P36+r24 P46+r25 P56
||List of local landraces of Bambara groundnut that were characterized
and used in the analysis
||r13 P16+r23 P26+P36+r34
||r14 P16+r24 P26+r34
||r15 P16+r25 P26+r35
r12 P26+2P16 r13 P36+
2P16 r14 P46+2P16 r15
P56+2P26 r23 P36+2P26
r24 P46+ 2P26 r25 P56+2P36
r34 P46+2P36 r35 P56+2P46
||Effects of first order components on number of pods per plant:
P37+r14 P47+r15 P57
P37+r24 P47+r25 P57
||r13 P17+r23 P27+P37+r34
||r14 P17+r24 P27+r34
||r15 P17+r25 P27+r35
r12 P27+2P17 r13 P37+
2P17 r14 P47+2P17 r15
P57+2P27 r23 P37+2P27
r24 P47+2 P27 r25 P57+2P37
r34 P47+2P37 r35 2P57+2P47
||Effects of second order components on seed yield per plant:
In the above equations, rs are the phenotypic correlations between variables, Ps are the direct effects (coefficients) of one variable upon another and rij pijs are the indirect effects. The residual effect, Px8, is composed of effects other than those included in the model.
The influence of vegetative variables on seed yield per plant: The paths of influence among vegetative variables on seed yield are shown in Fig. 1 and Table 2. Only leaf length was positively and significantly correlated with seed yield per plant and this correlation was predominantly attributed to the direct contribution of leaf length on yield.
The influence of reproductive variables on seed yield per plant: The interrelationships among reproductive variables on seed yield are indicated in Fig. 2 and Table 3. Seed size (100 seed weight) had a low correlation with seed yield; however, its direct effect was relatively high. The high direct effect was reduced to a low correlation by the negative indirect effect of 100 seed weight through number of pods. The latter was attributed to the negative correlation existing between weight of 100 seeds and number of pods while the direct effect of number of pods on yield was high and positive.
||Relations between vegetative (growth) variables with seed
|| Relations between reproductive variables with seed yield
|* Significant; ** Highly significant
|| 1st order components on weight of 100 seeds
The high and significant correlation between number of pods with seed yield was largely due to its direct effect on seed yield. Number of days to flowering was significantly and negatively correlated with seed yield and this was largely due to its negative direct effect on yield. In each correlation, weight of 100 seeds and number of pods interacted negatively on their effects on seed yield.
Two stage relations
First order components (vegetative variables) on seed size: The paths of
influence of vegetative variables on seed size are shown in Fig.
3 and Table 4. Petiole length had a significant negative
relationship with 100 seed weight and this was predominantly due to the negative
direct effect of this variable on seed size.
|| 1st order components on number of pods/plant
|** Highly significant
|| 2nd Order components on seed yield/plant
|** Highly significant
First order components (vegetative variables) on number of pods per plant: Results of influence of first order components on number of pods are shown in Table 5 and Fig. 3. Plant height was significantly and positively correlated with number of pods and this was mainly due to its positive direct contribution on number of pods. Similarly, the significant positive correlation between leaf length with number of pods was largely due to its direct effect on number of pods.
||Path diagram and coefficients of factors on the influence
of vegetative variables on seed yield of Bambara groundnut. Pijs are
the direct effects, rijs are the correlation coefficients (* Significant;
** Highly significant)
||Path diagram and coefficients of factors on the influence
of reproductive variables on seed yield of Bambara groundnut. Pijs
are the direct effects, rijs are the correlation coefficients (**
Second order components on seed yield per plant: Effects of second order
components (weight of 100 seeds and number of pods) on seed yield are shown
in Fig. 3 and Table 6. The high positive
independent effect of seed size on yield was reduced to a low correlation by
the negative indirect effect of seed size on yield through number of pods per
plant. The latter was in turn attributed to the negative correlation between
seed size and number of pods per plant while the direct effect of number of
pods per plant was high and positive.
||Path diagram and coefficients of factors on the influence
of first order on second order components and the latter on yield of Bambara
groundnut. Pijs are the direct effects, rijs are the correlation
Number of pods produced was significantly and positively correlated with seed
yield and this was mainly due to the high direct effect of pods per plant on
The positive relationship and contribution of leaf length on pod formation and yield suggests that plant architecture for Bambara groundnut crop that favors longer leaves results to more pods and higher seed yield. Karikari and Tabone (2003) advocated that canopy spread was among variables that could be used for indirect selection for drought tolerance. Leaf and petiole lengths determine spread of the canopy above the ground and hence provide the necessary ground cover against moisture loss from the soil. Increased photosynthetic area from the increased leaf size as a result of increased leaf area index may have contributed to increased agronomic performance. Pod formation in Bambara groundnut is an important yield attribute as the current investigation indicates strong contribution of this variable on seed yield. Other investigations (Karikari, 1972; Wigglesworth, 1996; Karikari and Tabone, 2003) have also indicated the importance of higher poding in yields of Bambara groundnut although the variable indicated low heritability suggesting that selection for high pod number may be difficult.
Component compensation was noted in the relationships among seed size, pod number and yield of Bambara groundnut. Thus the non-significant relationship between seed size on yield was due to the sacrificial and unfavorable influence through pod number. Results suggest that if pod variation was to be held constant, increased seed size would increase yield. However, the consistent negative relationship between pod number and seed size compensated and sacrificed the relationship between seed size and yield. While Karikari and Tabone (2003) also reported high independent contribution of seed size on yield of Bambara groundnut, they found positive and significant correlation between seed size and yield. Differential environments and genotypes used might have contributed to the different responses. Various factors have been advocated in contributing to negative relationships among plant components including competition for ambient resources such as nutrients, moisture, light; genetic factors such as linkage and pleiotropy. Wigglesworth (1996) noted the importances of reducing inter and intraplant competition by raising the crop under stress free conditions so as to minimize yield sacrifice through component compensation.
Genetic improvement for more pods should be accompanied by optimum husbandry
practices that minimize adverse relationships among components of yield in order
to realize the benefits of genetic manipulation. Crossing, selfing and selection
among segregates may break unfavorable linkages resulting to progenies combining
genes of higher number of pods and larger seeds. The significant positive relationships
among the seed size variables viz. weight of 100 seeds, seed width and seed
length suggest that these variables are influenced by similar physiological
and genetic patterns and that simultaneous selection for them can be done in
a breeding programme.
The present investigation suggests that late flowering has a detrimental effect on seed yield of Bambara groundnut in the environment of investigation. In areas with marginal rains, earlier flowering confers an advantage of forming more pods and seeds and consequently higher yields. Early pod induction may cause plants to use assimilates for reproductive growth at the expense of continued vegetative growth. Flowering earliness in Bambara groundnut confers drought tolerance through drought escape according to reports of Karikari (1972) and Karikari and Tabone (2003).
Plant stature determines the number of nodes and sites where pod-bearing branches
are formed. This study suggests that taller plants produce more pods emanating
from more branching that in turn results to increased yield (Table
5 and 6). Breeders in these areas therefore, should develop
earlier genotypes with architecture of Bambara groundnut varieties that are
tall with more pods and longer foliage. However, the programmes should aim at
developing genotypes producing amounts of pods that do not largely sacrifice
seed size for optimum production of yield. Research is needed to investigate
husbandry practices that minimize component compensation effects for realizing
higher yields of Bambara groundnut. Studies should also be done under field
conditions so as to investigate the consistency of such findings.
The authors wish to thank the Tanzanian Government for providing the necessary financial assistance for the study. We appreciate the comments and suggestions of anonymous reviewers, which greatly improved the quality of our paper.