About Bioline  All Journals  Testimonials  Support Bioline  News

African Crop Science Journal
African Crop Science Society
ISSN: 1021-9730 EISSN: 2072-6589
Vol. 6, Num. 2, 1998, pp. 129-135
African Crop Science Journal, Vol

African Crop Science Journal, Vol. 6. No. 2, pp. 129-135, 1998

Heritability of traits associated with Striga [Striga hermonthica (Del.) Benth.] resistance in an open-pollinated maize population

L. Akanvou and E. V. Doku 1

IDESSA B.P. 121 Ferkéssédougou, Côte d'Ivoire
1University of Ghana, Crop Science Department, P.O.Box 44, Legon, Accra, Ghana

(Received 10 June, 1997; accepted 2 February, 1998)

Code Number:CS98014
Sizes of Files:
      Text: 43K
      Graphics: No associated graphics files


Heritability and genetic correlations were estimated for traits associated with resistance to Striga hermonthica (Del.) Benth., a weed parasite of the Scrophulariaceae family. This study was conducted on 320 full-sib families generated from Design 1 crosses in the open-pollinated maize population, TZL Composite-1 (C1). Estimates of narrow-sense heritability were highest for traits such as ear Striga rating (h2n = 0.43±0.16), Striga syndrome rating (h2n = 0.33±0.15) and for the yield of infested plants (h2n = 0.31±0.13). These values indicated that a sufficient proportion of additive genetic variance might be available in the population under study to make progress through selection. For those traits, recurrent selection procedures such as mass selection, S1 progeny testing, or half-sib family selection is suggested for improving that population for resistance to S. hermonthica. Estimates of genetic correlations showed that grain yield was negatively correlated with Striga syndrome rating 1, taken at 8 weeks after planting (rg = -0.92±0.93), ear Striga rating at harvest (rg = -0.88±1.28) and, to a lesser extent, to Striga count 1 taken at 8 weeks after planting (rg = -0.22±0.46). Yield was positively correlated to the number of ears harvested on the infested plants (rg = 0.99±1.12).

Key Words: Genetic improvement, parasitic weed, recurrent selection


L'héritabilité et les corrélations génétiques ont été évaluées en considérant les caractères liés à la résistance de Striga hermonthica (Del.) Benth., une herbe parasite de la famille des Scrophulariaceae. Cette étude a été conduite avec 320 familles full-sib générées à partir de croisements du type Design 1 dans une population de maòs à pollinisation libre, TZL Composite-1 (C1). Les estimations de l'héritabilité au sens restreint étaient plus élevées pour les caractères tels que le syndrome du Striga sur les épis (hn = 0.43±0.16), le syndrome du Striga visible sur la plante (hn= 0.33±0.15) et pour le rendement des plantes infesteés (hn = 0.31±0.13). Ces valeurs indiquent qu'une proportion suffisante de variance génétique additive pourrait étre présenteé dans la population étudiée, ce qui permettrait de faire des progrès en sélection. Pour ces caractères, la sélection récurrente telle que la sélection massale, le testage de progénies S1, ou la sélection des familles half-sib est indiquée pour améliorer la résistance de cette population au S. hermonthica. Les corrélations génétiques évalueés ont montré que le rendement était négativement corrélé au syndrome du Striga sur les plantes infesteés 8 semaines après le semis (rg = -0.92±0.93), au syndrome du Striga sur l'épis à la récolte (rg = -0.88±1.28), et à un degré moindre au nombre de plantes de Striga 8 semaines après le semis (rg = -0.22±0.46). Le rendement était positivement corrélé au nombre d'épis récoltés sur les plantes infesteés (rg = 0.99±1.12).

Mots Clés: Amélioration génétique, parasites de mauvaises herbes, sélection récurrente


Striga hermonthica (Del.) Benth. is a root parasitic plant of the Scrophulariaceae family, and is one of the major constraints to maize (Zea mays L.) production in the sub-Saharan Africa. In maize, crop losses due to Striga vary between 10% yield reduction to total crop failure (Smaling et al., 1991). Control methods using herbicides and nitrogen fertilizers are costly for African farmers. Also, hoeing or hand-pulling of Striga is time consuming, very tedious and not effective.

Breeding for resistant varieties combined with legume/cereal crop rotation are among the most effective and economical Striga control methods. However, better knowledge of the mode of gene action controlling the inheritance of resistance will enhance the identification and development of resistant varieties. Estimates of heritabilities of the most important characters associated with Striga resistance express the degree of their reliability in improving resistance to Striga when incorporated into breeding lines.

The objectives of this study were to: (i) provide estimates of heritability of traits associated with Striga resistance in the open-pollinated maize population, TZLComposite-1 (C1), and (ii) estimate genetic correlation among the traits studied.


Experimental design. A bulk of the maize population TZLComposite-1(C1) was planted at the International Institute of Tropical Agriculture (IITA) Ibadan, Nigeria and North Carolina Design 1 crosses were made: 80 random plants designated as male were each crossed with 4 other random plants designated as female. Seeds for plants used as males were planted 4 days after planting the female parents. Every plant used as male parent was selfed. At harvest, a set of 4 female ears representing a cross to a particular male was selected and constituted a single male group (half-sib family). A total of 80 such male groups were obtained, resulting in 320 full-sib families which were grown under S. hermonthica artificial infestation in a randomised complete block design with two replications at two locations, Abuja and Mokwa (Northern parts of Nigeria). The 320 full-sib families were divided into 20 sets which contained 16 families and also 2 hybrid checks (9022-13, the resistant check and 8338-1, the susceptible check) each. Thus, a total of 18 entries were assigned to each block. Within blocks, each plot consisted of two rows of 22 plants. One row of each plot was artificially infested with S. hermonthica while the other row remained non- infested. The rows were 5m long and 0.75m apart. Plants were spaced 50 cm within rows. Three kernels per hill were planted and the stand thinned to 2 plants per hill. Infested and non-infested plots were separated by a 2 m alley to avoid contamination.

Striga infestation and field management. Screening for Striga resistance in the 320 full-sib families was done on artificially infested fields at Abuja and Mokwa. In both locations, maize was planted on ridges which provided enough moisture and prevented waterlogging. Fertilizer N.P.K (15:15:15) was applied to each row after planting at a rate of 60 kg ha-1. The nitrogen rate which was half the recommended rate for maize in Northern Nigeria soils was used to ensure maximum development of S. hermonthica. Half of the nitrogen fertilizer (30 kg ha-1) was applied at planting and the other half was top-dressed 4 to 5 weeks later. The experiment was kept free of weeds other than S. hermonthica.

Striga seeds were mixed with sieved sand which was used as a carrier material to provide adequate volume for rapid and consistent infestation. The amount of sand and Striga seeds needed for artificial infestation was calculated following the method outlined by Berner et al. (1993). Prior to Striga infestation and maize planting, Ethylene gas was used to stimulate suicidal germination of existing Striga seeds in the soil. A scoop that held approximately 5ml of water was used for infestation. Holes of 10 cm in diameter and 8 cm in depth were dug out on the ridges. Infestation was then done by spreading the contents of a scoop filled with Striga seed mixed with sieved sand into each of the holes of the infested rows. About 3,000 germinable S. hermonthica seeds were sowed in each hole. Striga seeds were then lightly covered with surrounding soil. Maize seeds were planted the same day, both in the non-infested and infested hills above the Striga seeds.

Data collection. Data were collected on infested and non-infested rows for the following parameters: Date of flowering, plant height, ear height, number of Striga plants emerged per row at 8 weeks after planting (Striga count 1), Striga syndrome on host plants at 8 weeks after planting (Striga rating 1), number of ears per row, and Striga effect on ears (ear Striga rating). Yield was then computed. Date of flowering was recorded when approximately 50% of the plants were shedding pollen. Plant and ear heights were measured in cm from ground level to the tip of the tassel and uppermost ear-bearing node, respectively. Data on the number of emerged Striga plants and Striga syndrome on host plants were taken on the entire infested rows. Plant aspect was taken on host plants in non-infested rows only. Number of ears was counted from each infested and non-infested row. Moisture was measured on infested and non-infested rows by using a moisture meter (Dicky-John Corp., U.S.A). Estimates of moisture were determined on samples of grains shelled from the middle areas of at least 5 ears for each row. The Striga syndrome rating (1 = no symptoms, 9 = severe symptoms) used to measure tolerance of maize to Striga was based on several indicators: Foliar damage which increases with the severity of Striga infestation (small 1-3mm, round, white chlorotic blotches, gradually extending longitudinally into streaks), loss of turgor, and in severely affected plants 'scorching' or firing of the leaf tissue, particularly around the margins. Affected plants exhibited spindly and weak stems, small and poorly filled ears, shortening of internodes which gave a dwarfing appearance. This scale was designed at IITA by Kim et al. (1991).

Statistical analysis. An analysis of variance (ANOVA) was carried out for the seven traits which may be important in selecting for Striga resistance. The mean values for each plot were used in the analysis of variance using Proc GLM from SAS (SAS, 1987) for each character for the two locations combined. The form (Table 1) used for the pooled analysis over sets for both locations was given by Hallauer and Miranda (1988).

TABLE 1. General form of the combined analysis of variance of Design 1 experiment pooled across sets for two locations (Abuja and Mokwa)

Source of Variation
















Rep (Set x Loc)




Male (Set)

s (m-1)



Fem.(Set x Male)

sm (f-1)



Male x Loc (Set)

s (m-1) (1-1)



Fem. x Loc (Set x Male)

sm (f-1) (l-1)




ls (r-1) (mr-1)







a Loc = locations = 2; s = number of sets = 20; r = replications per set = 2; m = male group per set = 4; f = females mated to same male =4; s2 = random plot to plot variation; s2m= variance due to genetic differences among males; s2f = variance due to genetic differences among females mated to the same male; s2ml = variance due to interaction of male genotypes with locations; s2fl= variance due to interaction of female genotypes with locations

Estimates of the components of variance and their standard errors. By using the appropriate mean squares obtained from the analysis of variance (Table 1), estimates of the variance among males (s2m), and variance among females mated to the same male (s2f) were computed according to the formula given by Lindsey et al. (1962).

Under assumptions outlined by Comstock and Robinson et al. (1949), additive variance (s2a) and dominance variance (s2d) were estimated as follows:

s2a= 4*s2m and s2d= 4*(s2f-s2m)

Standard errors for s2a and s2d were computed by using the general formula given by Anderson et al. (1952).

Estimates of heritability and their standard errors. The estimates of narrow-sense heritability were based on the selection of full-sib families (Hanson, 1963). Estimates of narrow sense heritability (h2n) based on the combined data analysis of both locations were:

h2n=[2 * s2m]/[( s2m + s2f) + (s2ml + s2fl)/l + s2 (rl)]

Where s2m and s2f are the male and female variances, respectively,

s2 = estimate of pooled error variability, l = number of locations, Standard errors of h2n were computed by an approximate procedure outlined by Hanson (1989).

Estimates of genotypic and phenotypic correlations. Genotypic correlations (rg) and phenotypic correlations (rp) were calculated by using the mean products and estimates of genetic and phenotypic variances as outlined by Falconer (1981). The standard error (S.E) of the genetic correlations were computed using the formula given by Scheinberg (1966).


Estimates of the components of variance and their standard errors. Estimates of additive genetic variance (s2a) and dominance variance (s2d) are presented in Table 2. Comparison of the estimates of variances showed that the estimates of s2a were lower than those of the s2d for Striga count 1, yield uninfested and delayed silking. Estimates of additive variance were, however, greater than those of the dominance variance for Striga rating 1, ear Striga rating 1, yield of infested plants, and for anthesis silk-interval of infested plants. The additive genetic variance and the dominance variances were almost of the same magnitude for the yield of non-infested plants. Negative estimates of s2d or s2a may be due to either sampling error or lack of random mating in making the half-sib family groups.

TABLE 2. Across locations estimates of additive genetic variance (s2a), dominante variance (s2d), narrow sense heritability (h2n) and their standard errors





Striga count 1




Striga rating 1




Ear Striga rating




Yield infested plants




Yield non infested plants




Anthesis silk-interval




Delay silk




**, h2n not estimable because s2a was negative

Estimates of heritability. Narrow sense heritability estimates (h2n) of the seven characters studied (Table 2) showed that ear Striga rating had the highest heritability : h2n = 0.437. Variables Striga rating 1 and yield infested plants had lower heritabilities compared to variable ear Striga rating: h2n = 0.330 and 0.317, respectively. Variables such as Striga count 1, yield non-infested plants and ASI showed a low heritability estimate: h2n = 0.139, 0.155 and 0.173, respectively. The heritability of variable delay silking was not estimable because the additive variance was zero.

Estimates of phenotypic and genotypic correlations. Genotypic and phenotypic correlation coefficients (rg and rp, respectively) computed for the seven characters studied are summarised in Table 3. Positive and low genotypic correlations were obtained between Striga count 1 and Striga rating 1 (rg = 0.20±0.41), Striga count 1 and ear Striga rating (rg = 0.03±0.50). Positive and high genotypic correlations were obtained between Striga rating 1 and ear Striga rating (rg = 1.08±0.71). Negative and low genotypic correlations were obtained between Striga count 1 and yield of infested plants (rg = -0.22±0.46) and Striga count 1 and anthesis silking interval (rg = -0.29±0.45). Negative and high genotypic correlations were obtained between Striga rating 1 and yield of infested plants (rg = -0.92±0.93) and Striga rating 1 and anthesis silking interval (rg = -0.84±1.39). Plant height was negatively correlated with Striga rating 1 (rg = -0.68±0.45) and ear Striga rating (rg = -0.33±0.54). Yield showed a high genotypic correlation with plant height (rg = 0.61±0.54) and with ears harvested for infested plants (rg = 0.99±1.12), which is a component of yield.

TABLE 3. Genotypic correlations, rg (in bold) with their standard errors and phenotypic correlations rp between characters studied

count 1

rating 1

Ear Striga

Ear harvested
infested plants

Yield infested

silking interval

Infested plants







count 1







rating 1




-1.1 2±0.70



Ear Striga







Ear harvested
infested plants







infested plants







Anthesis silking interval








Estimates of genetic variances and heritability estimates. Results of the analysis of the data across locations (Table 2) showed that the proportion of s2a was greater than that of the s2d (close to zero) for traits like Striga rating 1, ear Striga rating and anthesis silk-interval of infested plants. This suggested that the s2a was a high proportion of the total genetic variance for these characters, whereas non-additive gene action plays a major role in Striga emergence as previously reported by Akanvou et al. (1997).

Heritability estimates indicate to what extent selection is likely to be effective (Robinson et al., 1949, 1951; Dudley and Moll, 1969). In our study, estimates of narrow-sense heritability were generally low (Table 2). This indicates that at the early cycle of recombination (cycle 1) the proportion of the additive genetic variance, although present in the total genetic variance, was low for most of the characters under study. Estimates of narrow-sense heritability were higher for ear Striga rating (h2n = 0.43±0.16), Striga rating (h2n = 0.33±0.15) and yield of infested plants (h2n = 0.31±0.13). These values indicated that a sufficient proportion of additive genetic variance might be available in the population under study. This is confirmed by the large additive variance estimates found for these characters (Table 2).

Estimates of genetic correlations. Genetic correlations (rg) for characters studied showed that yield was negatively correlated with Striga count 1 (rg = -0.22±0.46), Striga rating 1 (rg = -0.92±0.93) and with ear Striga rating (rg = -0.88±1.28). Such negative associations were expected since Striga reduces yield through its adverse effects on the physiology of the infested plants. A low correlation was found between Striga count 8 and Striga rating 8 (rg = 0.20±0.41). This indicates that the number of Striga plants is not linearly related to the ability of the host plants to tolerate Striga effects. For instance, a host plant can support few Striga plants or no Striga plants at all above ground, but show heavy Striga symptoms. There was a strong relationship between grain yield and ASI of infested plants (rg = 1.05±1.30). This indicates that selecting for longer ASI would increase the yield which depends on the capacity of the genotypes to synchronise male and female flowering. Thus, selecting for ASI and for yield under Striga infestation may be an efficient method to breed for Striga resistance in the population under study. Genetic correlation values above 1 may be due to sampling variation.


This study showed that for traits such as Striga rating and ear Striga rating, additive gene action appears to be the primary source of variation in the open-pollinated maize population under study. Narrow-sense heritability estimates also indicated that the additive variance was a major proportion of the genotypic variance for characters like Striga rating, ear Striga rating and yield of infested plants. Recurrent selection such as mass selection, S1 progeny testing, or half-sib family selection is suggested for improving that population for resistance to S. hermonthica. The choice of the selection scheme will depend on the length of time and the amount of effort required to complete a selection cycle. The trait Striga count 1 is, however, controlled by non-additive gene action which could be interpreted as dominance but, as no test for epistatic variation was made, such a distinction can not be drawn.


This paper is part of a Ph.D thesis submitted by the first author to the University of Ghana, Legon-Accra, Ghana. Academic and Research fundings were provided by WINROCK/IITA fellowship. The first author expresses her gratitude to Dr Kling, maize breeder at IITA, Nigeria for her suggestions.


Akanvou, L., Doku, E. V. and Kling, J. 1997. Estimates of genetic variances and interrelationships of traits associated with Striga [Striga hermonthica (Del.) Benth.] resistance in an open-pollinated maize population. African Crop Science Journal 5: 1-8.

Anderson, R. L. and Bancroft, T. A. 1952. Statistical Theory in Research. Mc Graw-Hill Book co., New York. pp. 321.

Berner, D. K., Winslow, M. D., Awad, A. E., Cardwell, K. F. and Raj, D. R. 1993. Striga Research Methods. A manual. Draft edition prepared by the IITA Striga research group, Ibadan, Nigeria: IITA. 71pp.

Dudley, J. W. and Moll, R. H. 1969. Interpretation and use of estimates of heritability and genetic variances in plant breeding. Crop Science 9: 257-262.

Falconer, D. S. 1981. Introduction to Quantitative Genetics. Second edition. Longman, New York. 432pp.

Hallauer, A. R. and Miranda, J. B. 1988. Hereditary variance. In: Quantitative Genetics in Maize Breeding. Hallauer, A. R. and Miranda, J. B. (Eds.), pp. 44-111. Iowa State University Press, Ames, Iowa.

Hanson, W. D. 1963. Heritability. In: Statistical Genetics and Plant Breeding. Hanson, W. D. and Robinson, H. F. (Eds.), pp. 125-140. Washington D.C. NAtional Academic of Science, National Research Council Pub-lication 982.

Hanson, W. D. 1989. Standard errors for heritability and expected selection response. Crop Science 29:1561-1562.

Kim, S. K. 1991. Breeding maize for Striga tolerance and the development of field infestation technique. In: CombatingStriga in Africa.Proceedings of the international workshop organized by IITA, ICRISAT and IDRC. Kim, S.K. (Ed.), pp. 96-108. IITA Ibadan, Nigeria.

Lindsey, M. F., Lonnquist, J. H. and Gardner, C. O. 1962. Estimates of genetic variance in open-pollinated varieties of cornbelt. Crop Science 5:125-129.

Robinson, H.F., Comstock, R.E. and Harv, H. 1949. Estimates of heritability and the degree of dominance in corn. Agronomy Journal 41: 353-359.

Robinson, H.F., Comstock, R.E. and Harvey, P.H. 1951. Genotypic and phenotypic correlations in corn and their implications in selection. Agronomy Journal 43:282-287.

SAS Institute. 1987. SAS/STAT Guide for Personal computers. Version 6.04 ed. SAS Institute, Inc., Cary, NC.

Smaling, E.M.A., Stein, A. and Sloot, A. 1991. A Statistical analysis of the influence of Striga hermonthica on maize yields in Kenya. Plant and Soil 138:1-8.

Scheinberg, E. 1966. The sampling variance of the correlation coefficients estimated in genetic experiments. Biometrics 22:187-191.

Copyright 1998, African Crop Science Society

Home Faq Resources Email Bioline
© Bioline International, 1989 - 2014, Site last up-dated on 14-Apr-2014.
Site created and maintained by the Reference Center on Environmental Information, CRIA, Brazil
System hosted by the Internet Data Center of Rede Nacional de Ensino e Pesquisa, RNP, Brazil