HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Padi, F. K. Articles by Ehlers, J. D. Search for Related Content PubMed Articles by Padi, F. K. Articles by Ehlers, J. D. Agricola Articles by Padi, F. K. Articles by Ehlers, J. D. Related Collections Other Legumes Experiment Design

Effectiveness of Early Generation Selection in Cowpea for Grain Yield and Agronomic Characteristics in Semiarid West Africa

^a CSIR, Savanna Agricultural Research Institute, Box 52, Tamale, Ghana
^b Dep. of Botany and Plant Sciences, Univ. of California, Riverside, CA 92521-0124

^* Corresponding author (padifrancis{at}yahoo.co.uk).

	ABSTRACT

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The effectiveness of early generation selection for grain yield in a cowpea [Vigna unguiculata (L.) Walp.] population was examined in the Guinea and Sudan savanna agroecologies of Ghana where genotype x location interaction is known to be large. A set of 131 F_3:4 lines were developed from a cross between a local cultivar and an unadapted source of large grain size. Mild selection was practiced during line development at one location in the Guinea savanna zone to eliminate poorly adapted lines. Unreplicated F₃ plant data were collected on all the lines at the one location during the development of the lines. Multilocation trials were conducted with lines formed by bulk harvest of F₄ families to assess how effectively the early generation selection protocol was able to generate superior lines for the target agroecology. Genotypic correlation for grain yield between locations was high only between the two locations in the Guinea savanna zone. Narrow-sense heritability estimates were low and not different from zero for grain yield, but heritability estimates for days to flowering and seed size were large. F₄ lines derived from the highest 10% performing F₃ individuals were no higher yielding than F₄ lines derived from the remaining F₃ individuals, indicating that early generation selection for yield was ineffective. Single-seed descent (SSD) or bulk breeding methods will be more efficient than pedigree breeding for developing cowpea varieties with high yield potential for this agroecology.

Abbreviations: G x E, genotype x environment • G x L, genotype x location • G x Y, genotype x year • SSD, single-seed descent

	INTRODUCTION

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
DEVELOPING COWPEA [Vigna unguiculata (L.) Walp.] cultivars that are superior to existing cultivars in terms of yield potential and yield stability are important breeding objectives for cowpea improvement programs targeting the Guinea and Sudan savanna agroecologies of West Africa, where the bulk of the crop is produced (Ehlers and Hall, 1997; Hall et al., 2003). The significant genotype x environment (G x E) interaction for grain yield that characterizes these agroecologies are the result of both large and variable genotype x year (G x Y) and genotype x location (G x L) effects, because rainfall amount and distribution are highly variable over time and space, and because of large differences in soil water-holding capacity and fertility across these ecologies (Dingkuhn et al., 2006; Hall et al., 1997).

Breeders of autogamous species generally employ either pedigree, single-seed descent (SSD), or bulk breeding methods when the objectives are to combine complementary traits from two parents. Pedigree breeding, where selection is practiced among and within families in early generations, requires a substantial investment in field space and record-keeping activities during the line development phase but should result in a relatively small set of elite lines needing testing across the target production environment to identify varietal candidates. In contrast, SSD or bulk breeding methods incur only limited cost to generate a large number of lines, but cost of testing the large numbers of lines is high.

Genotype x year effects can hamper identification of promising lines with all these breeding approaches. With SSD or bulk breeding, where multiple-year evaluations of many lines might be needed to identify promising lines with stable performance, the effects occur at the end of the process of line development. With the pedigree approach, they occur during the process of line development and impact and restrict the types of lines selected as well as their performance characteristics.

Large G x L effects have been described for sets of fixed lines grown at locations in the Guinea and Sudan savanna (Padi, 2004). Selection under the target production environment has been advocated as a means of increasing gains from selection, particularly for regions where G x E interaction is prevalent (Ceccarelli et al., 1987; Annicchiarico and Pecetti, 1998). If these G x L effects are large, selection conducted at one environment will not produce valuable lines for other environments, and separate selection programs tailored to individual environments will have to be conducted. In the present study, we also assess the impact of G x L on early generation selection gains for grain yield and yield components in a breeding program.

In autogamous species such as cowpea, early generation selection typically involves initiating selection among F₂ individuals, especially for highly heritable traits such as grain quality characteristics (Hall et al., 1997). The usefulness of selecting in early generations for complex traits like grain yield that are affected by G x E interactions is debatable, but practical breeders will usually apply some selection pressure for productivity to eliminate very poor performing individuals that are unlikely to give rise to high-performing lines in later generations.

For many species, and in particular for large-seeded legumes such as cowpea, selection on an F₃ line basis is usually limited to performance assessments in a single environment conducted with small plots and few replications due to limitations of F₃ seed supply that is generated from single F₂ plants. If traditional pedigree breeding is employed to generate F₄ lines using seed from F₃ single-plant selections, similar seed-supply limitations will constrain performance-testing options in the F₄ generation.

Success in early generation selection in the breeding program depends on the heritability of the traits, that is, how accurately phenotypic assessments of traits reflect real genotypic differences between the lines and over the target production regions. Success also requires that these genetic differences will persist when testing is conducted on descendant lines that are approaching homozygosity. Theory (Cockerham, 1963) suggests that the expected genetic correlation (r_g) between line performance at an early generation and its descendant line at homozygosity is r_g = 0.667 to 0.707 for F₂-derived lines and r_g = 0.840 to 0.866 for F₃-derived lines, depending on the role of dominance (Bernardo, 2003). Thus, as noted by Bernardo (2003), the genetic correlations are high enough for early generation selection to be effective in self-pollinating crops as long as heritability values are reasonably high. Heritability of a trait is dependent not only on the genetic architecture of the trait and how the trait is influenced by G x E interactions but also by the testing protocol employed in the breeding program. In practice, early generation selection has been found to be effective in some situations. St. Martin and Geraldi (2002) observed that selection in the F₂ or F₃ for yield was effective in increasing genetic gain in soybean [Glycine max (L.) Merr.]. Similarly, Hegstad et al. (1999) discovered that selection for high-yielding lines in single-row plant row yield trials in the F₃ was effective in identifying superior lines in advanced yield trials in soybean. In cowpea, N'tare et al. (1984) found early generation testing was as effective as an SSD method in identifying lines that were not significantly different for yield. In other reports in self-fertilizing species, including wheat (Triticum aestivum L.) (Quail et al., 1989; Borghi et al., 1998) and chickpea (Cicer arietinum L.) (Rahman and Bahl, 1986), selection in the early generations (F₂ or F₃) was not effective for identifying lines with high yield potential in advanced testing.

The main objective of the present study is to assess the effectiveness of early generation selection for grain yield and it components in the Guinea and Sudan savanna zone of West Africa. Two complementary approaches were used to do this: (i) calculating trait heritabilities using variance component estimates and parent–offspring correlations, and (ii) empirically comparing performance of sets of F₄ lines developed by SSD that would have been selected vs. those not selected in early generations on the basis of single-plant performance in the F_3. If early generation selection were deemed successful, the pedigree approach would require that fewer numbers of advanced lines be tested in extensive multiyear and multilocation trials than bulk breeding to achieve a similar level of breeding progress.

	MATERIALS AND METHODS

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Genotypes and Experimental Conditions
One hundred and thirty-one F₃-derived F₄ lines were derived from a cross between Apagbaala, an adapted cultivar for northern Ghana, and UCR 01-11-52. Apagbaala is the most important commercial cultivar in northern Ghana because of its synchronous flowering but has small seed size (Padi et al., 2004). UCR 01-11-52 is a breeding line developed at the University of California, Riverside from a cross of ‘California Blackeye No. 46’ (CB46) and the large-seeded Brazilian cultivar Montiero. UCR 01-11-52 was selected for crossing with Apagbaala because of its very large attractive seed and earliness to flower. The F₂ population, consisting of 158 plants, was sown in the field at Nyankpala (9°25', 0°58' W) in the Guinea savanna agroecology of Ghana on 23 July 2002. Single seeds were sown at a spacing of 100 x 60 cm and grown under rain-fed conditions. At maturity, 127 individuals set reasonable amounts of seed, and each of these plants was harvested separately to develop F₂-derived F₃ families.

In July 2003, the 127 F_2:3 families were planted at Nyankpala in single stands at a spacing of 60 by 60 cm in four-row plots without replication. The adapted parent, Apagbaala, was planted in three replicates as a check. Plots were monitored regularly and the first plant to flower in each plot was tagged. At physiological pod maturity, the plant with the highest pod load in a plot based on visual assessment was also tagged. For each plot, the number of days to 50% flowering was noted. For each of the two tagged F_2:3 plants per plot, yield component data taken on each harvested F₃ plant included number of branches per plant, number of pods per plant, number of seeds per pod (average of 10 pods), and weight of a hundred seeds. Each tagged plant was harvested separately. In July 2004, F_3:4 families were advanced at Nyankpala without replication. Seeds were sown at 60 by 20 cm in five-row plots, 4 m long. In general, lines were advanced without selection, except that F_2:3 plants that produced too few seeds to permit planting a minimum of four rows in the F₄ were eliminated. This selection protocol reflects what most practical plant breeders would do during development of SSD lines, in other words, eliminate obviously poor-performing families and chose the best plant within a family to carry the family forward. In all, 131 F₄ families were harvested in bulk to obtain seeds, but only 90 families produced enough seed for trilocation yield evaluations. Thus assessment of early generation selection gain could be made using data from 131 lines in the F₃ generation, while estimation of G x L effects was conducted with a set of 90 F₄ bulk lines.

Replicated yield evaluation of the F₄ lines was conducted at three locations, including Manga (Sudan savanna, 11°01'), Nyankpala, and Yendi (Guinea savanna, 9°27' N), all located in northern Ghana. The adapted parent, Apagbaala, was included as a check at each location. For each location, seeds were sown in five-row plots measuring 4 m in length with three replications. Plants were spaced at 60 by 20 cm, with two plants per stand. Seeds were sown on 7 July 2005 at Manga, 26 July 2005 at Nyankpala, and 28 July 2005 at Yendi. The plants were sprayed three times, first at 31 to 33 d after sowing. The second and third sprays were usually done at 10-d intervals after the first spray to control flower thrips (Megalurothrips sjostedti) and a complex of pod-sucking insects. The insecticide used was lambda-cyhalothrin {[1(S*),3(Z)]-(±)-cyano(3-phenoxyphenyl)methyl-3-(2-chloro-3,3,3-trifluoro-1-propenyl)-2,2-dimethylcyclopropanecarboxylate} at the rate of 20 g a.i. per hectare. Data recorded on the three middle rows per plot at each location include days to 50% flowering, number of branches per plant, number of pods per plant, number of seeds per pod, weight of a hundred seeds, and grain yield.

Statistical Analyses
Parent–offspring correlations were computed between F₃ single plant performance and F₄ line performance at each location. In the F₄ line evaluations, simple correlation coefficients were used to assess the relationship between grain yield and other recorded traits. For evaluations conducted on the F₄ lines, analyses of variance (ANOVA) were first conducted separately for each dataset in each location. Genotypic (_g²) and error (_e²) variance components for grain yield were estimated from the mean squares and were used to estimate broad-sense heritability (H) on an entry mean basis as

[1]

where R is the number of replicates. The data was then pooled into a factorial structure of G x L for pairs of locations and submitted to ANOVA to obtain variance components to estimate the genotypic correlation for each trait between locations. The genotypic correlations (and their significance) between any two locations for a trait were estimated as suggested by Robertson (1959). The pooled genetic correlation among locations, which indicates consistency of genotype response across locations for a trait, was estimated (Annicchiarico, 2002) as

[2]

where S_g is the variance of the square root values of the genotypic variance component estimated for individual locations and _gl² is the variance due to G x L interaction.

Analysis of variance for each trait was then conducted for data over the three locations to estimate the extent of G x L interaction for each trait over the locations in which the study was conducted. Analyses of variance were conducted using the Statistix computer package (Analytical software version 7, Tallahassee, FL). Variance components were estimated from the mean squares for genotypes (_g²), G x L interaction (_gl²), and experimental error (_e²). Heritability of a trait in the broad-sense for the whole trial on an entry-mean basis was estimated as

[3]

where L is the number of locations. Two-sided confidence intervals (95%) were calculated to determine the precision of heritability estimates as described by Knapp et al. (1985). Heritability in the narrow sense was estimated by parent–offspring correlations between F₄ line entry means and F₃ data taken on a per plant basis, based on Pearson's product-moment correlation. Standard errors for parent–offspring correlation (r) were estimated following Ibrahim and Quick (2001) as

[4]

where n is the number of parent–offspring pairs.

The effectiveness of early generation selection was evaluated empirically by comparing the yields of F₄ lines derived from the top 20, 30, or 40% yielding F₃ individuals with F₄ lines that would have been rejected (during F₃ line evaluations) at the specified selection intensity at each location. A t test procedure was used to determine the statistical significance of the selected and rejected groups. We refer to this as retrospective selection.

	RESULTS

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Location and Year Differences
Rainfall amount and distribution were recorded at each location during the period of the experiment. In 2003 at Nyankpala, where single plant selections were made within and between F₃ families, total rainfall was similar to the long-term mean of 1000 mm for the Guinea savannah agroecology. Of this amount, 422.7 mm was received on 32 rainy days during the period of the experiment and provided very favorable conditions for growth. In 2005, during which replicated yield trials were conducted on F₄ lines, total rainfall in Nyankpala was 16% less than the long-term average, with a reduction in rainy days from 75 (long-term average) to 57 in 2005. Total rainfall during the experiment in Nyankpala was, therefore, only 316.2 mm, received on 22 rainy days (Table 1 ). In particular, a dry spell lasting 15 d across the Guinea savanna zone in August affected plant growth at this location. Plant growth was less severely affected by drought at Yendi, probably because of the presence of soils with greater water-holding capacity and higher soil organic-matter content. At Manga in the Sudan savannah agroecology, rainfall total and distribution was typical of the ecoregion, with 386.1 mm rain received on 32 rainy days. The rainfall distribution and amount, coupled with soil fertility characteristics, largely differentiated the locations for crop performance; Yendi was the most favorable and Manga was the least favorable, attributed mainly to a very poor sandy soil at Manga.

Retrospective selection in the F₃ was based on selection intensities of 20, 30, and 40% for grain yield. Effectiveness of this early generation selection was assessed on the expectation that the F₄ lines derived from higher-performing F₃ individuals will be significantly higher than F₄ lines derived from lower-performing F₃ individuals. Mean grain yield in the F₄ lines, however, was not significantly different between selected and rejected families at all selection intensities tested, though selection at 40% intensity only is presented (Table 4 ). Even at Nyankpala, where grain yields of the F₃ generation were assessed, mean yield of the selected lines was actually slightly lower than the rejected group. Early generation selection at 40% retained lines that were significantly later in flowering at all locations (albeit by a mean of 1 d), with a nonsignificant change in seed size.

	DISCUSSION

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The variation in traits studied and trait correlations observed in the population indicated high potential for selecting segregates that are earlier in flowering and have larger seed size than the adapted parent, without impacting negatively on yield potential. Earliness to flowering and large seed size are important traits for new cowpea cultivars in West Africa. Farmers can utilize early maturing varieties as a strategy to escape end-of-season drought or to permit double cropping within the single rainy season that characterizes the Guinea and Sudan savanna ecologies. Large-seeded varieties also command premium prices in cowpea trade in the West Africa (Langyintuo et al., 2003).

Agronomic characteristics that show significant correlations with grain yield could be important as surrogates for early generation selection for grain yield if these traits are easier to evaluate or less influenced by the environment. Promising traits in the present study include pods per plant and seeds per pod, each trait showing significant correlations with grain yield in the F₃. The F₃ correlation coefficients, when compared with the F₄ lines' correlation coefficients, suggest that these traits may be valuable as surrogates for grain yield only if the breeding lines are evaluated in single, widely spaced stands rather than in dense stands recommended for production. The increase in plant-to-plant competition in dense stands results in an increase in plant-to-plant coefficient of variation (Fasoula and Fasoula, 2002), and this may be responsible for the poor correlations in dense stands.

Results of the multilocation evaluation of F₄ lines were largely determined by different patterns of weather and soil conditions for plant growth at the different locations. The effects of F₄ line differences in response to these conditions were reflected in significant main effects for lines, locations, and their interaction for all traits included in the study. The large variation among the locations was reflected in the low broad-sense heritability estimates for grain yield, pods per plant and branches per plant. As indicated by the narrow-sense heritability estimates, days to flowering, seed size, and number of seeds per pod are much less influenced by environment and may be selected for early in the breeding program at one location for expression in other locations within the agroecology.

The locations in which the evaluations were conducted were selected by the National Agricultural Research System of Ghana as benchmark sites representing the major cowpea growing belts (similar to the definition of megaenvironments by Gauch and Zobel [1997]) existing within the Guinea and Sudan savanna ecoregions of Ghana. Based on the long-term yield data on which this choice of locations was made, location yield rank in the current evaluations was expected to follow a pattern of Yendi > Nyankpala > Manga. This trend was observed in this study in spite of the unusual 15-d dry spell experienced in August 2005 at Yendi and Nyankpala, compared with a more favorable rainfall distribution at Manga. This observation suggests that factors other than rainfall, such as soil fertility, probably play a major role in the relative yield rankings at these three locations. The pattern of the G x L interaction for grain yield was such that the genetic correlations for grain yield between locations decreased as the locations' mean yield decreased (or with increasing stress between locations), in concurrence with earlier observations by Bänziger et al. (1997) and Cooper et al. (1997). Selection for grain yield at any one of the two locations in the Guinea savannah zone should, therefore, identify genotypes with better adaptation to the other location within the Guinea savanna zone than at Manga in the Sudan savanna zone.

This established influence of the environment on grain yield expression indicated by the analyses of F₄ lines was reflected in the low F₃–F₄ line correlation coefficients and large standard error estimates for grain yield, underlining the strong influence of the semiarid environment on reliability of grain yield of single F₃ plants in predicting F₄ line performance. Early generation selection for grain yield was, therefore, ineffective based on the mean of selected lines not being different from mean of rejected lines. In addition to the influence of the environment, gene segregation from the F₃ to the F₄ lines may also cause low parent–offspring correlation coefficients. Compared with days to flowering and hundred seed weight that showed high narrow-sense heritability estimates, and with the F₃ and F₄ lines generations showing similar population mean and range for each of these traits, the more than sevenfold difference in the range for grain yield in the F₃ declined to a less than fourfold difference in the F₄ lines. This observation suggests that both environment and gene segregation may account for the apparent ineffectiveness of early generation selection in improving the population mean. The results of the present study are similar to that reported by Rahman and Bahl (1986), for which early generation selection was found effective for only highly heritable traits, including seed weight and seeds per pod in chickpea. The lack of significant difference per se between the mean of selected and rejected lines for grain yield observed in the present study should lead to increased comparative efficiency for early generation selection if early selection were associated with recovery of a significant proportion of elite families in the unselected population. In such a situation, fewer lines are handled at reduced transaction costs compared with an SSD approach, for example. Moreover, the probability of obtaining families with broad adaptation to the target region is higher when a high proportion of elite families are tested in multilocation evaluations. In the present study, however, the low recovery of elite families (2 to 4 of 9 per location, and 9 of 21 across locations) was associated with loss of all 4 families that showed potential for broad adaptation, suggesting that early generation selection in a pedigree breeding approach for the target agroecology is not effective for developing cultivars with high yield potential and wide adaptation.

The present study has emphasized the importance of G x E interaction in reducing the effectiveness of early generation selection for grain yield. F₃ evaluation under favorable conditions of weather, rainfall, and wide spacing at one location led to differences in the proportion of elite advanced breeding families recovered among locations that was related to location mean yield. Thus, of 9 elite families per location, 2 were recovered at the most stressed location (Manga), 3 at the location with moderate stress (Nyankpala), and 4 at the most favorable location (Yendi). It has been observed that in optimal environments, the genotypic variance increases relative to the error variance (Sinebo et al., 2002), leading to a better differentiation of genotype performance, which contributes to improved parent–offspring correlation because of the lower contribution of the environment to phenotypic expression (Ward, 1994). Though the observation of Simmonds (1996) that low-yielding lines in the early generation do not produce high-yielding segregates in a later generation appears valid, the key to progress in selection when following the pedigree breeding approach relies on effectively identifying the superior F₃ lines. Influence of the environment for selection and residual heterosis in the F₃ generation may be important in reducing the effectiveness of early generation selection in a pedigree breeding approach compared with an SSD or bulk breeding approach. The effectiveness of bulk or SSD approaches may be enhanced further by conducting early generation selection for traits having high narrow-sense heritability, such as grain size and flowering date.

	ACKNOWLEDGMENTS

 
This research was supported by funds from Project Number 6 of the Challenge Program for Water and Food, and USAID Bean/Cowpea Collaborative Research Support program grant no. DAN-G-SS-86-00008-00.

	NOTES

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher.

Received for publication May 10, 2007.

	REFERENCES

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 

• Annicchiarico, P. 2002. Genotype x environment interactions: Challenges and opportunities for plant breeding and cultivar recommendations. Available at http://www.fao.org/documents/pub_dett.asp?pub_id=103142&lang=en (verified 11 Nov. 2006). FAO Plant Protection and Production Papers 174. FAO, Rome.
• Annicchiarico, P., and L. Pecetti. 1998. Yield vs. morphological trait based criteria for selection of durum wheat in a semi-arid Mediterranean region (northern Syria). Field Crops Res. 59 :163–173.[CrossRef]
• Bänziger, M., F.J. Betran, and H.R. Lafitte. 1997. Breeding tropical maize for low N environments: I. Spillover effects from high N selection environments to low N target environments. Crop Sci. 37 :1103–1109.[Abstract/Free Full Text]
• Bernardo, R. 2003. On the effectiveness of early generation selection in self-pollinated crops. Crop Sci. 43 :1558–1560.[Abstract/Free Full Text]
• Borghi, B., M. Accerbi, and M. Corbellini. 1998. Response to early generation selection for grain yield and harvest index in bread wheat (T. aestivum L.). Plant Breed. 117 :13–18.[CrossRef]
• Ceccarelli, S., M.M. Nachit, G. Ortiz-Ferrara, M.S. Mekni, M. Tahir, J. van Leur, and J.P. Srivastava. 1987. Breeding strategies for improving cereal yield and stability under drought. p. 101–114. In J.P. Srivastava, E. Porceddu, E. Acevedo, and S. Varma (ed.) Drought tolerance in winter cereals. John Wiley & Sons, Chichester, UK.
• Cockerham, C.C. 1963. Estimation of genetic variances. p. 53–93. In W.D. Hanson and H.F. Robinson (ed.) Statistical genetics and plant breeding. Publication 982. National Academy of Science, National Research Council, Washington, DC.
• Cooper, M., R.E. Stucker, I.H. DeLacy, and B.D. Harch. 1997. Wheat breeding nurseries, target environments, and indirect selection for grain yield. Crop Sci. 37 :1168–1176.[Abstract/Free Full Text]
• Dingkuhn, M., B.B. Singh, B. Clerget, J. Chantereau, and B. Sultan. 2006. Past, present and future criteria to breed crops for water-limited environments in west Africa. Agric. Water Manage. 80 :241–261.[CrossRef]
• Ehlers, J.D., and A.E. Hall. 1997. Cowpea. Field Crops Res. 53 :187–204.[CrossRef]
• Fasoula, V.A., and D.A. Fasoula. 2002. Principles underlying genetic improvement for high and stable crop yield potential. Field Crops Res. 75 :191–209.[CrossRef]
• Gauch, H.G., and R.W. Zobel. 1997. Identifying mega-environments and targeting genotypes. Crop Sci. 37 :311–326.[Abstract/Free Full Text]
• Hall, A.E., B.B. Singh, and J.D. Ehlers. 1997. Cowpea breeding. Plant Breed. Rev. 15 :215–275.
• Hall, A.E., N. Cisse, S. Thiaw, H.O.A. Elawad, J.D. Ehlers, A.M. Ismail, R.L. Fery, P.A. Roberts, L.W. Kitch, L.L. Murdock, O. Boukar, R.D. Phillips, and K.H. McWatters. 2003. Development of cowpea cultivars and germplasm by the bean/cowpea CRSP. Field Crops Res. 82 :103–134.[CrossRef]
• Hegstad, J.M., G. Bollero, and C.D. Nickell. 1999. Potential of using plant row yields to predict soybean yield. Crop Sci. 39 :1671–1675.[Abstract/Free Full Text]
• Ibrahim, A.M.H., and J.S. Quick. 2001. Heritability of heat tolerance in winter and spring wheat. Crop Sci. 41 :1401–1405.[Abstract/Free Full Text]
• Knapp, S.J., W.W. Stroup, and W.M. Ross. 1985. Exact confidence intervals for heritability on a progeny mean basis. Crop Sci. 25 :192–194.[Abstract/Free Full Text]
• Langyintuo, A.S., J. Lowenberg-Deboer, D. Faye, D. Lambert, G. Ibro, B. Moussa, A. Kergna, S. Kushwaha, S. Musa, and G. Ntoukam. 2003. Cowpea supply and demand in west and central Africa. Field Crops Res. 82 :215–231.[CrossRef]
• N'tare, B.R., R.J. Redden, B.B. Singh, and M.E. Aken'ova. 1984. Evaluation of early generation selection procedures for yield in cowpea [Vigna unguiculata (L.) Walp.]. Crop Sci. 9 :91–100.
• Padi, F.K. 2004. Relationship between stress tolerance and grain yield stability in cowpea. J. Agric. Sci. 142 :431–443.[CrossRef]
• Padi, F.K., N.N. Denwar, F.Z. Kaleem, A.B. Salifu, V.A. Clottey, J. Kombiok, M. Haruna, A.E. Hall, and K.O. Marfo. 2004. Registration of ‘Apagbaala’ cowpea. Crop Sci. 44 :1486.[Free Full Text]
• Quail, K.J., R.A. Fischer, and J.T. Wood. 1989. Early generation selection in wheat: I. Yield potential. Aust. J. Agric. Res. 40 :1117–1130.[CrossRef][Web of Science]
• Rahman, M.A., and P.N. Bahl. 1986. Evaluation of early generation testing in chickpea. Plant Breed. 97 :82–85.[CrossRef]
• Robertson, A. 1959. The sampling variance of the genetic correlation coefficient. Biometrics 15 :469–485.[CrossRef][Web of Science]
• Simmonds, N.W. 1996. Family selection in plant breeding. Euphytica 90 :201–208.[CrossRef][Web of Science]
• Sinebo, W., R. Gretmacher, and A. Edelbauer. 2002. Environment of selection for grain yield in low fertiliser input barley. Field Crops Res. 74 :151–162.[CrossRef]
• St. Martin, S.K., and I.O. Geraldi. 2002. Comparison of three procedures for early generation testing of soybean. Crop Sci. 42 :705–709.[Abstract/Free Full Text]
• Ward, P.J. 1994. Parent-offspring regression and extreme environments. Heredity 72 :574–581.[CrossRef][Web of Science]

This article has been cited by other articles:

				R.-C. Yang When Is Early Generation Selection Effective in Self-Pollinated Crops? Crop Sci., October 22, 2009; 49(6): 2065 - 2070. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Padi, F. K. Articles by Ehlers, J. D. Search for Related Content PubMed Articles by Padi, F. K. Articles by Ehlers, J. D. Agricola Articles by Padi, F. K. Articles by Ehlers, J. D. Related Collections Other Legumes Experiment Design