Groundnut (Arachis hypogaea L.), is cultivated annually on about 24.63 million hectares worldwide with annual production of 41.27 million tons in shell with a productivity of about 1.85 t ha^–1 (FAO, 2012). It is highly adapted to tropical and subtropical climates of the world and cultivated in nearly 100 countries. It is a key crop for small scale farmers especially in Africa and Asia where the crop serve as a valuable source of dietary protein, oil, and fodder for livestock. It contains 48-50% oil and 26-28% protein, and is a rich source of dietary fibre, minerals, and vitamins (Janila et al., 2013). In addition, groundnut has the ability to fix atmospheric nitrogen to the soil to help in the maintenance of soil fertility. In West Africa, Nigeria is the largest producer of groundnuts with a production of 3.07 million tons on about2.4 million hectares (FAO, 2012). Despite the economic, social and cultural importance of groundnuts, its productivity is severely constrained by several biotic and abiotic factors. Among them groundnut chloroticrosette disease (GCRD) causes severe crop losses The most serious yield losses were reported during the year 1975 when an epidemic in northern Nigeria destroyed approximately 0.7 million hectares of groundnut, with an estimated loss of US$250 million (Yayock et al., 1976).The GCRD is characterised by small, chlorotic, twisted and distorted leaflets with shortened internodes and thickened stems. Affected plants especially those infected at a young stage are severely stunted (Bock et al., 1990). The disease also affects both quality of the haulm and the pod.

Population development and phenotype evaluation

The study involved the use of nine experimental lines comprising of three aphid resistant (ICGX – SM 00020/5/9, ICGX – SM 00017/5/P₁₀/P₁ and ICGX – SM 00020/5/P₄/P₁ and three GCRD resistant (ICGV IS 07890, ICGV IS 07899 and ICIAR-19BT) genotypes (these genotypes were previously evaluated for three years (2008 – 2010) at the Institute for Agricultural Research (IAR), Ahmadu Bello University (ABU),Zaria, Nigeria and were confirmed to have field resistance to aphids and GCRD)obtained from the International Crop Research Institute for Tropical Agriculture (ICRISAT) inMali. Three widely cultivated varieties (SAMNUT14, KWANKWASO, and MANIPENTA) were also included as parents in the population development. The pedigree descriptions of the nine genotypes are presented in Table 1. The genotypes were manually cross-pollinated in a half diallel mating scheme at the screen house of IAR, SamaruIAR) Samaru, (11°10.00?N and 7°38.00? E, 693 m), and ABU in 2011. Additional manual cross-pollinations were made at IAR during the 2011 rainy season. Seed limitations for multi-location evaluation were overcome by advancing F₁ seeds to next generation (F₂) as suggested by (Hallauer et al., 2008).

The nine parental lines along with the 36 F₂ progenies were evaluated for GCRD resistance using a 9 x 5lattice design with two replications at two locations (Samaru, Kaduna state, and Lafia, Nasarawa state (8^°32"N, 7°42"E ) during the 2011/2012 growing seasons using an infector – row techniques (susceptible SAMNUT 14 genotype was planted in alternate rows with test materials) as described by Olorunju et al. (2001) at the two locations. The infector rows were planted 2 weeks prior to the test materials to allow the build-up of inoculum. Two row plots of 4.0 m in length with inter and intra-row spacing of 0.75 m x 0.25 m, respectively, were used.

Aphid and GCRD resistance evaluation

A. craccivora colonies were collected from infested cowpea Vigna unguiculata L., and groundnut A hypogaea plants at different locations in groundnut producing area in Nigeria to cover the different isolates that may be present in the country. The colonies (presumed to be viruliferous) were each maintained on susceptible groundnut genotype SAMNUT 14 in a screen house.

Two wingless (apterae) aphids were transferred onto 7 to 14-day-old seedlings of nine parental lines and their 36 F₂ progenies grown at IAR. Each genotype was observed for the presence or absence of aphid colonies (adults as well as nymphs) 7 days after infestation. Plants with no aphid colonies were re-infested with aphids seven days after the first infestation. It is rare to find plants without aphids in choice tests because the aphids are free to roam to find suitable plant hosts. Aphids that appeared to be transient, possibly probing for feeding sites, are often observed on resistant plants in choice tests, along with dead aphids. Sometimes several viviparous aptera, surrounded by a few nymphs, were observed on resistant plants without the development of established colonies. Rate of aphid infestation was evaluated using 0-4 scale, combined with GCRD incidencedeveloped by Mensah et al. (2005, 2008). Aphid infestation three time (at forthright interval) by a 0 = No aphid, 0.5 = fewer than 10 aphids per plant, no colony formed, 1.0 =11–100 aphids per plant, plants appear healthy, 1.5 = 101–150 aphids per plant, plants appear healthy, 2.0 = 151–300 aphids per plant, mostly on the young leaves or tender stems, plants appear healthy, 2.5 = 301–500 aphids per plant, plants appear healthy, 3.5 = 501–800 aphids per plant, young leaves and tender stems are covered with aphids, leaves appear slightly curly and shiny, 3.5 = more than 800 aphids per plant, plants appear stunted, leaves appear curled and slightly yellow, no sooty mould and few cast skins and 4.0 = more than 800 aphids per plant, plants appear stunted, leaves appear severely curled and yellow and are covered with sooty mould and cast skins. An aphid damage index (DI) for each line was calculated by the following formula: DI = ∑ (scale value x no. of plants in the category)/ (4 x total no. of plants) x 100. The DI ranges between 0 for no infestation and 100 for the most severe damage (Mensah et al., 2005). The DI was used as an indicator of aphid resistance and was applied in the analysis. The disease severity was recorded as the amount of plant tissue that is diseased, chlorotic rosette.

Reaction to GCRD was recorded on a scale of 1 to 9 as described by GGP (2000) as follows: 1 = No apparent rosette symptoms, 3 = 10 – 20% rosette symptoms, 5 = 20 to 60% rosette symptoms, 7 = 60 – 80% rosette symptoms and 9 = 100% rosette symptoms. The results of these observations were transformed to compute infection responses as measured by area under disease progressive curve (AUDPC) based on Moldovan et al. (2005).

Agronomic data such as pod weight per plant (g), and sound kernel weight per plant (g) were measured on five randomly selected plants per plot.

Data analysis

Genetic analysis of resistance to GCRD

Analysis of the diallel for general combining ability (GCA)andspecific combiningability (SCA) for all traits were based on the Model I, method 2 proposed by Griffing (1956). Parents and one set of F₂’s, excluding reciprocal F₂’s, were included in the analysis combining abilities.

Trait values were predicted based on traits mean value to produce a balanced data set. Diallel data were analysed using the Diallel SAS-05 program (Zhang et al., 2005). GCA and SCA effects were determined for parents and the 36F₂’s, respectively. The following linear mixed model was fitted to data to estimate variance components for single and multi-location diallel tests. The model for the analysis of variance for single location was:

Model for the analysis of variance for multi-location was:

General and specific combining ability for the traits

For the traits studied, the mean square (MS) values for both GCA and SCA were significant (P < 0.05) (Table 2). The SCA mean squares were higher than the GCA mean squares for DI and AUDPC (Table 2). The magnitude of GCA x location interaction for the majority of traits were relatively small compare to the GCA mean squares. Significant SCA × location interactions (p < 0.05) were observed for all traits. The GCA and SCA variance components were significantly different from zero for all the traits (Table 3). To understand the relative importance of general and specific combining abilities for DI and AUDPC, estimates of components of GCA and SCA that approximates variances estimated according to Bakers ratio (Baker, 1978) indicates a ratio of closer to unity (0.73) for aphid damage index and low value 

Both GCA and SCA made significant and important contribution to progeny variation for DI and AUDPC. All the parents of the most GCRD-resistant crosses; ICGX – SM 00017/5/P₁₀/P₁, ICIAR-19BT, ICGX – SM 00020/5/9 and ICGX – SM 00020/5/P₄/P₁ had appreciable resistance and also favourable GCA resistance values. This suggests that, although resistance to GCRD tends to be at least partly dominant (Clements et al., 2004), optimal resistance in progenies will require crossing parental genotypes that are both GCRD resistant, supporting the findings of Loffler et al. (2011) and Hung and Holland (2012). Therefore, selection for resistance should not be confined to a single group but should be performed in parallel in all groups. The study showed that crosses between two susceptible genotypes resulted in progenies with susceptibility to GCRD. In contrast, Loffler et al. (2011) observed that hybrids often had more disease than their parental inbreds, perhaps because they used highly susceptible tester lines and higher inoculum pressure for hybrids than parents. SCA as well as GCA were important for evaluating resistance of the progeny. Significant SCA detected in 13 of 36 possible combinations indicated the presence of non-additive gene effect. Significant SCA effects were observed for the combinations ICGX – SM 00017/5/P₁₀/P₁ X ICIAR-19BT (aphid resistant//rosette resistant), ICGX – SM 00020/5/9 X ICGX – SM 00020/5/P₄/P₁ (aphid resistant//aphid resistant), ICGX – SM 00020/5/P_4/P₁ X SAMNUT 14 (aphid resistant//rosette susceptible) and ICGV IS 07899 X KWANKWASO ((aphid resistant//rosette susceptible). These results indicate resistance of these progenies was higher than would be expected from average of their expected parents based on AUDPC symptom rating. The largest positive SCA effects correspond to ICGX – SM 00020/5/P4/P₁ X MANIPENTA. This combination was more susceptible than predicted average parent performance indicating the importance of non-additive gene effect in this particular cross. Kenga et al. (2004) suggest that the difficulty in predicting the resistance level of the hybrid, on the basis of GCA alone should necessitate testing of specific male-female combinations. The SCA values provide important information about the performance of the hybrid relative to its parents. Arunga et al. (2010) found that the SCA effect alone has limited value for parental choice in breeding programs. The authors suggested that per se performance of the lines, SCA and GCA effects should be considered in selecting desirable parents in a breeding programme which is desired by any breeder. Furthermore, it was observed that crosses involving one good combiner and one average or poor combiner showed negative SCA effects. For example, MANIPENTA and KWANKWASO had poor GCA values for GRD resistance, while their crosses with ICGV IS 07899 and ICGX – SM 00020/5/9, respectively, had significant and desirable SCA effects. This is in agreement with Habarurema et al. (2012) which made similar conclusion in a study on bacterial blight ((Xanthomonas oryzae pv.oryzae) in rice.

The combining ability ratio, also known as Baker’s ratio, for resistance to GCRD observed in this study was less than unity. According to Baker (1978), when combining ability ratio approaches unity, GCA alone cannot predict the performance of the parents. Thus, the GCA scores could not be used to predict the performance of the parents in the present study, because the value of Baker’s ratio is much lower than the theoretical maximum of unity. Low Baker’s ratio observed for AUDPC in this study highlighted the importance of SCA variance, and hence the importance of dominance and/or epistatic gene effects for increasing resistance to GRD. This implies that selection should be done on latter generations, based on better hybrid combinations, rather than the performance of the parents involved in crossing programs. Partitioning G x L into variance to GCA x L and SCA x L interaction effects indicated significant variances of both GCA x L and SCA x L effects. The significant of GCA x L variance implied that GCRD symptom rating (AUDPC) was sensitive to environmental conditions and data from additional environments or seasons will lead to precise GCA .