Groundnut (Arachis hypogaea L.) is an important food crop popularly cultivated for its seed consumption. Nigeria is globally ranked as the third largest groundnut producing country and the largest producer in Africa  . The crop has helped fight malnutrition in Africa as the seed is nutritionally rich  . Breeding effort is ongoing to meet different purposes    . The crop has been largely characterized morphologically for improvement purposes, but previous studies only captured few landraces or cultivars using limited traits   . Morphological traits which are the phenotypic expression of genetic information     are grouped into two: quantitative and qualitative traits. The latter is controlled by one or few genes while the former is controlled by interactions of many genes called polygenes   . Characters interrelate to contribute to the overall function of plant and they may relate positively or negatively    . The degree of relationship may be explored as morphological markers to monitor inheritance or likely appearance of other characters. When two characters are closely related, improving one of the characters may affect the other character  . Character association is a basic tool in plant breeding and systematics      .
Nigeria is committed to high productivity of groundnut as the highest producer in Africa. Despite this, comprehensive information on the interrelationship amongst traits of the crop is scarce and often misapplied. The aim of the present study was to evaluate the degree of relationship and association among selected characters of groundnut cultivars. The outcome may help users of the information make decisions in cultivar selection. This is a technical approach that may save time and resources in the evaluations of traits that are not instantly measurable. It may also provide useful information that will unravel the complex nature of factors affecting the overall survival and yield of the crop.
2. Materials and Methods
118 cultivars of groundnut (Arachis hypogaea L.) were planted at the University of Agriculture Makurdi Research Farm (Latitude 07˚45.53'N; Longitude 008˚37.41'E 99 metres above sea level). Farm sized used was 891.75 m2. Seeds were sowed on 41 long ridges with an inter ridge distance of 0.75 m. The entire field was partitioned into two: Plot 1 and 2. Planting was done in a perpendicular orientation to the ridges. One (1) seed of each cultivar was sowed and replicated 5 times within 1m space on the ridges. This was done for all 118 cultivars in each plot. The design structure was a completely randomized design in a 118 × 5 × 2 arrangement. Therefore, a total of 1180 plant stands were assessed. Characterization was based on qualitative and quantitative evaluations of accessions  . A total of 62 characters were evaluated as listed in Table 1.
Computation was done in Excel work book. Data was analysed on the SPSS software (20.0 versions). Correlation analysis was applied on Qualitative traits
Table 1. List of characters evaluated in Arachis hypogaea.
using the Spearman’s Rank and the Kendall Tau’s methods of correlation. Pearson’s Product Moment Correlation method was applied on quantitative characters. Significant correlations were flagged at 0.05 and 0.01 levels. Strength of relationship was classified as follows: Weak (0.00 - 0.39), Moderate (0.40 - 0.69), High (0.70 - 0.89) and Very high (0.90 - 1.00) correlations   . Dependence among traits was tested using Chi square at 0.05 level of significance. Multiple regression analysis was carried out using a predictive model for seed yield. The equation connecting seed yield and the determinants followed a multiple regression patterns of Y = a + B1X1 + B2X2 + B3X3 + B4X4 + B5X5 + B6X6.   .
3. Results and Discussion
Table 2 gives the Pearson’s correlation matrix of 26 quantitative traits. Most of the traits are very weakly correlated either positively or negatively. Day to emergence weakly correlates with Day to 50% flowering (R = 0.237, p < 0.05) and Number of flower (R = 0.221, p < 0.05). Emergence of first shoot does not affect the number of flower produced. Day to first flowering correlates with Day to maturity (R = 0.39, p < 0.01), Pod length (R = 0.237, p < 0.05), pod weight (R = 0.292, p < 0.01), number of seed (R = 0.324, p < 0.01) and seed weight (R = 0.345, p < 0.01). These are all weak relationships. It appears that timing of flowering significantly affects the maturity time of groundnut but no effect on the yield. Number of flower correlates with number of pods (R = 0.535, p < 0.01), pod length (R = 0.285, p < 0.05), pod weight (R = 0.312, p < 0.01), number of seeds (R = 0.409, p < 0.01) and seed weight (R = 0.295, p < 0.05). Thus flower production capacity tends to moderately affect both the pod and seed production potential of the plant (yield) but no effect on the sizes of pod and seed.
Length of floral whorl largely affects the petiole length of the flower with a significant correlation coefficient of 0.865 (p < 0.01). These characters have no effect on any vegetative and reproductive trait. Plant height moderately correlates with leaflet length only (R = 0.354, p < 0.01) indicating that tall cultivars are likely to develop long leaflets. The extent to which these traits affect each other (R2) is 13%. Other factors that may determine the sizes of leaflets account for 87%. Thus, height of the crop is not a good indicator of leaf sizes. When groundnut is used as a folder crop in animal feeds, cultivars with excellent combination of the two characters may be selected. Plant biomass moderately correlates with number of pod (R = 0.494, p < 0.01), pod length (R = 0.395, p < 0.01), pod width (R = 0.473, p < 0.05), number of seed (R = 0.468, p < 0.01) and seed weight (R = 0.548, p < 0.01). It appears that overall weight of the plant in the field has significant effect on yield as well as sizes of pods and seeds. From theory, plant biomass is a function of gravity, proplasmic content and water relationship. This suggests that the crop needs maximum care to maintain good biomass. From the matrix, plant biomass has no relationship with other vegetative traits such as plant height and leaf sizes.
Most of the high positive correlations recorded were observed amongst pod and seed parameters. Number of pod correlates with pod weight (R = 0.674,
p < 0.01) and number of seeds (R = 0.698, p < 0.01). To achieve maximum seed yield, high pod yield is necessary. Pod length and width significantly affects pod weight (R = 0.738, p < 0.01) and the number of seeds present (R = 0.551, p < 0.01). The coefficient of determination between pod sizes and number of seed (R2) is 30%. Large pod may therefore determine the number of seeds it contains to an extent. 70% of this relationship is accounted for by other unknown factors. As observed in the field, some large pods may contain fewer numbers of seeds than small pods most especially when the seeds are of big sizes. Seed weight and number of seeds produced per plant are highly correlated (R = 0.826, p < 0.01).
Table 3 presents the Kendall’s correlation matrix among qualitative traits with varying features. Qualitative traits have no effect on each other as all coefficients are very low. Those with significant values are very weakly related. Growth habit correlates with stem branching pattern (R = 0.214, p = 0.01) and standard petal colour (R = 0.221, p = 0.01). These are very weak relationships. Growth could assume any habit (procumbent, decumbent or erect) irrespective of all stem morphological features such as branching pattern, pigmentation and lateral branching habit. Growth habit has no relationship with all other qualitative characters as all coefficients of relationship are very weak and insignificant.
Table 3. Kendall’s correlation matrix of qualitative characters with varying features.
Legend: Code: GH, Character: growth habit, Variables: Procumbent, Decumbent, Erect; Code: SBP, Character: stem branching pattern, Variables: Alternate, Sequential, Irregular with flowers on main stem, Irregular without flowers on main stem; Code: SP, Character: stem pigmentation, Variables: Green, Brown; Code: SH, Character: stem hairiness, Variables: Scarce, Abundant, Very abundant; Code: LBH, Character: lateral branch habit, Variables: Distichous, Non-distichous; Code: LC, Character: leaf colour, Variables: Dark green, Light green; Code: HYL, Character: hairiness of young leaflet, Variables: Almost glabrous, Sparse and short, Sparse and long, Profuse and short, Profuse and long; Code: HML, Character: hairiness of mature leaflet, Variables: Almost glabrous, Sparse and short, Sparse and long, Profuse and short, Profuse and long; Code: SPC, Character: standard petal colour, Variables: Yellow only, Yellow with white patches; Code: LTX, Character: leaf texture, Variables: Hard, Soft; Code: TD, Character: traces of disease, Variables: Susceptible, Resistant; Code: PB, Character: pod beak, Variables: Absent, Slight, Moderate, Prominent, Very prominent; Code: PC, Character: pod constriction, Variables: None, Slight, Moderate, Deep, Very deep; Code: SC, Character: seed colour, Variables: Rosy pink, Rosy brown, Chimeric. *Correlation is significant at the 0.05 level (2-tailed); **Correlation is significant at the 0.01 level (2-tailed).
Stem hairiness also recorded very weak relationships with all qualitative characters. The only significant correlation between this trait and standard petal colour (R = 0.221, p = 0.01) was a very weak value. Pod beak and constriction pattern are significantly correlated (R = 0.284, p = 0.01) but very weak. The absence or prominence of beak does not affect the constriction pattern. Pod could have slight or moderate or deep or very deep constriction regardless of other qualitative characters. Tolerance to diseases is not affected by any of the qualitative traits. Also, no trait determines seed colour. The only high positive correlation observed among all qualitative traits is between hairiness of young and mature leaflets (R = 0.837, p = 0.01). Hairiness is classified as: Almost glabrous, Sparse/Short, Sparse/Long, Profuse/Short and Profuse and long. Pubescence level in young leaflets determines level and nature of pubescence in the mature leaflets of a particular plant. However, leaf and stem pubescence have no relationship.
Table 4 shows the relationship between Pod constriction and Pod size as revealed by Spearman and Kendall’s method. Pod constrictions had very low negative correlation with Pod sizes in the two non-parametric correlations (Kendall R = 0.029; Spearman R = 0.038). Pods could be slightly or moderately or deeply constricted irrespective of their sizes.
Chi square tests (Table 5) revealed significant associations among the variables tested (p < 0.05): Cultivar type and Disease susceptibility (χ2 = 17.93, p = 0.000); Cultivar type and Plant survivorship (χ2 = 39.97, p = 0.000); Cultivar type and Pod production (χ2 = 69.695, p = 0.008).
The equation connecting seed production capacity and six other regressors is given in Table 6. Among the regressors, Plant biomass PB (p = 0.035) and Number of pods NP (p = 0.00) were significant on seed production. Seed yield could be predicted by substituting the values of the predictors in the linear equation:
where: Y = number of seed, NF = number of flower, FS = length of the flower, PH = plant height, LS = leaflet length, PB = plant biomass, NP = number of pod. This level of relationship is confirmed in the scattergram of partial regression plots shown in Figures 1-4. Predicting the number of seed produced is necessary
Table 4. Spearman rank and Kendall’s tau correlations between pod constriction and pod size.
Figure 1. Regression plot of number of seed and number of pod per plant.
Figure 2. Regression plot of number of seed and plant biomass.
Table 5. Chi square tests between cultivar type and other characters.
Figure 3. Regression plot of number of seed and plant height.
Figure 4. Regression plot of number of seed and flower size.
Table 6. Regression coefficients and equation for groundnut seed production.
Seed Yield (Y) = 7.463 + 0.181(NF) − 0.733(FS) + 0.011(PH) − 0.886(LS) + 0.019(PB) + 0.889(NP). a. Dependent Variable: No of seeds/plant.
after pod harvesting when pods are to be stored for a long time without break. The coefficient of relationship of the overall model is 0.716. This model explains 51% of variation in seed production (R2 = 51.3%) hence a fit model. 48% of variation in seed production is not accounted for the predictors (Table 7).
In the present report, Arachis hypogaea can be described as a crop having moderate level of character association in quantitative traits. This is contrary to the findings in other crops where quantitative characters are very highly related    . High level of character divergence in vegetative features has many implications. It implies high level of diversity which may form the basis of varietal separation   . The major botanical varieties of Arachis hypogaea L. were previously classified based on foliar and stem morphological differences   . It also implies that genes coding for the vegetative traits are not linked together in groundnut and that no trait could be used as a morphological marker of other vegetative characters. For instance, Leaflet sizes could be small or large irrespective of the plant height among the cultivars.
Most vegetative and reproductive traits are not related. This may hinder breeding practice of indirect selection on the crop because cultivars cannot be easily selected for pod yield on the basis of plant height or leaf sizes or the growth pattern of standing plants in the field before they are harvested. Over the years, most improvement program on groundnut targeted yield related parameters    . Therefore, any trait or attribute that may help predict or improve yield would attract the attention of groundnut breeders. The observed moderate correlation between plant biomass and seed weight could be attributed to normal physiological processes such as water absorption potentials and photosynthetic activities   . Larger canopies may manufacture more food to be deposited in the seed as food reserves  . Adequate water intake is also needed for seed production in groundnut   .
The present report totally agrees with other studies where high correlations in reproductive characters have been reported  . It is expected that sizes of pods should determine the sizes and number of seeds present. The number of seed should also correlate well with its weight. Breeders aiming at improving seed yield may then focus only on improving the crop for pod yield and pod sizes regardless of the constriction pattern of the pod. Similar findings were previously reported on the crop   . It is also expected that the length of pedicel should determine the overall size of the flower, though this is unrelated with pod and seed yield. The present report has buttressed basic genetic theories. Traits such as disease susceptibility, plant survivorship and pod production depend on the type of cultivar. This may be due to the inherent genetic differences among the
Table 7. Model summary for groundnut seed production.
aPredictors: (Constant) : NF, FS, PH, LS, PB, NP.
cultivars    . Hence, selection for or against these traits in groundnut should be cultivar dependent only. It appears that the two traits that can help predict seed production in groundnut are pod yield and plant biomass, although a large proportion (48%) of seed production determinant is unaccounted for.
Pod yield is a quantitative trait which may involve the interactions of polygenes. This trait thus requires QTL (Quantitative Trait Loci) analysis for a more detailed explanation. The plant biomass factor is a weight related factor and cannot be separated from the moisture content of the plant. This implies that the crop requires high amount of water for maximum pod and seed production. Groundnut pods, as the name implies, are buried underground and they are attached to the plant root system where water absorption takes place. This information accounts for why the crop is popularly cultivated in the raining season in Nigeria  .
From the present report, qualitative traits do not affect each other. Hence they cannot be used as morphological markers of other traits. Any qualitative feature cannot be predicted by physical observation of any other trait. They all show a wide separation of relationship but are excellent features that provide systematic resolution. Traits such as growth habit, branching pattern, stem pigmentation, pod characteristics and seed colour can help differentiate cultivars. Among the qualitative traits, only disease resistant cultivars could be of interest to breeders.
The present study has successfully determined the level and nature of relationships among all qualitative and quantitative characters of Arachis hypogaea. The most significant high levels of relationships were found among quantitative reproductive traits especially in yield related properties. Vegetative characters are weakly related. No relationship was established among all qualitative features with the exception of pubescence of young and mature leaflets. The implications of these results have been extensively discussed. The information provided in this report is very crucial to groundnut growers and breeders as well those interested in systematic studies of the crop.
 Mustapha, S., Muhammed, U.M., Adeosun, N.O., Mohammed T.J., Mohammed, S.S. and Ibn-Aliyu, A. (2015) Nutritional and Functional Characterization of Undecoticated Groundnut (A.hypogaea L.) Seeds from Bosso Market, Minna, Nigeria. American Journal of Food Science and Technology, 3, 126-131.
 Upadhaya, H.D., Ortiz, R., Bramel, P.J. and Singh, S. (2003) Development of a Groundnut Core Collection Using Taxonomical, Geographical and Morphological Descriptors. Genetic Resources and Crop Evolution, 50, 139-148.
 Janila, P., Nigam, S.N., Manish, K.P, Pandey, Nagesh, P. and Ranjeev, K.V. (2013) Groundnut Improvement: Use of Genetic and Genomic Tools. Frontier in Plant Science, 4, 1-33.
 Katzamindis, S., Stavropoulus, N. and Ipsilandus, C.G. (2006) Classification and Evaluation of Greek Groundnut (Arachis hypogea L.) Using 17 Main Agronomic and Quality Traits. Pakistan Journal of Biological Science, 9, 1021-1027.
 Aguoru, C.U., Idakwo, F. and Olasan, J.O. (2015a) Partitioning of Eggplant Cultivars in North Central Nigeria Using Gross Vegetative and Reproductive Characters. Advanced Studies in Biology, 7, 413-422.
 Bayat, M., Amirnia, R. and Rahimi, M. (2017) Phenotypic and Genotypic Relationships between Traits in Crocus sativus L. As Revealed by Path Analysis. Elsevier Journal of Applied Research on Medicinal and Aromatic Plants, 5, 33-40.
 Olasan, J.O., Aguoru, C.U., Omoigui, L.O., Danmaigona, C. and Ugbaa, M.S. (2017a) Assessment of Genetic Diversity in Groundnut (Arachis hypogaea L.) Using Principal Component Analysis and Cluster Segmentation. Proceedings of the Genetic Society of Nigeria 41st Annual Conference, University of Agriculture Makurdi, Makurdi, 15-19 October 2017, 427-436.
 Andargie, M., Rem, S.P., Geoffrey, M.M. and Michael, P.T. (2013) Quantitative Trait Loci of Flowering Time Related Traits Identified in Recombinant Inbred Lines of Cowpea (Vigna unguiculata). Genome, 56, 289-294.
 Desawi, H.T., Sentayehu, A.K. and Daniel, E.G. (2014) Assessment of Genetic Variability, Genetic Advance, Correlation and Path Analysis for Morphological Traits in Sesame Genotypes. Asian Journal of Agricultural Research, 8, 181-194.
 Ogunniyan, D.L. and Olukojo, S.L. (2015) Genetic Variation, Heritability, Genetic Advance and Agronomic Character Association of Yellow Elite Inbred Lines of Maize (Zea mays L). Nigerian Journal of Genetics, 28, 24-28.
 Singh, S., Prakash, A., Chakraborty, N.R., Wheeler, C., Agarwal, P.K. and Ghosh, A. (2016) Genetic Variability, Character Association and Divergence Studies in Jatropha curcas for Improvement in Oil Yield. Trees, 30, 1163-1180.
 Aguoru, C.U., Okoli, B.E. and Olasan, J.O. (2016) Comparative Gross Morphology of Some Species of Sesamum L. International Journal of Current Research in BioSciences and Plant Biology, 3, 21-32.
 Aguoru, C.U., Offia, K.L. and Olasan, J.O. (2016) Insecticidal Efficacy of Plumeria spp Leaf Extracton Two Economically Important Insects Populations: Mosquitoes (Anopheles) and Bean Weevil (Callosubrachus maculatus). Journal of Herbal Medicinal Research, 1, 0001-0009.
 Olasan, J.O., Aguoru, C.U., Ugbaa, M.S. and Ekefan, E.J. (2017) Physiological Characterization of Groundnut (Arachis hypogaea L.) Seedlings Using Vigour and Performance Indices. Proceedings of the Genetic Society of Nigeria 41st Annual Conference, Makurdi, 15-19 October 2017, 84-91.
 Udom, G.N., Ndaeyo, N.U. and Yunus, Y.S. (2009) Effect of Sowing Dates and Varieties on the Yield Performance of Groundnut. Proceedings of the 43rd Annual Conference of the Agricultural Society of Nigeria, Abuja, 20-23 October 2009, 189-191.