GENOTYPE X ENVIRONMENT INTERACTIONS IN SUGARCANE YIELD TRIALS IN THE CENTRAL-WESTERN REGION OF VENEZUELA

Ramón Rea and Orlando De Sousa Vieira

Ramón Rea. Engineer and M.Sc. in Agronomy, Central University of Venezuela. Ph.D. in Agronomy, Mississippi State University, USA. Researcher, Instituto Nacional de Investigaciones Agricolas (INIA). Address: INIA-Yaracuy, Apartado 110, San Felipe- Yaracuy 3201, Venezuela. e-mail: ramonrea@hotmail.com

Orlando De Sousa-Vieira. B.S. in Agronomy, Universidad Centro-Occidental Lisandro, Venezuela. M. Sc. in Agronomy, Mississippi State University, USA. Ph.D., Louisiana State University USA. Researcher, INIA, Venezuela.

Summary

The presence of genotype by environment (G x E) interaction is a major concern to plant breeders, since large interactions can reduce gains from selection and complicate identification of superior cultivars. The objectives of this study were to determine the relative magnitude of G x E interaction effects and to evaluate phenotypic stability in sugarcane (Saccharum spp. Hybrid) in terms of regression coefficient, mean square deviation from regression, ecovalence, coefficient of determination, and coefficient of variation. Fourteen genotypes and three sugarcane cultivars were evaluated for 2yr at six locations in Venezuela. The genotype x location interaction for cane yield and apparent sucrose content (pol % cane) indicated that genotypes ranked different or changes in the magnitudes of differences between genotypes from one environment to another. The second order interaction was not significant for both traits. The clones B80-549, B80-408, and B81-503 were significantly superior to the rest of genotype for cane yield. B81-509, V84-25, B81-494, and B80-408 had the best performance for pol % cane. This study suggests that the stability analysis can contribute with supplementary information on the performance of new sugarcane selections prior to release for commercial cultivation and can increase the efficiency of cultivar development programs.

Resumen

La presencia de interacción genotipo x ambiente (G x A) es de suma importancia para los mejoradores, debido a que interacciones muy grandes pueden reducir la ganancia de la selección, complicando la identificación de cultivares superiores Los objetivos de este trabajo fueron determinar la magnitud de la interacción G x A y evaluar la estabilidad fenotípica en caña de azúcar (Saccharum spp. Hybrid) mediante el uso del coeficiente de regresión, la desviación de la regresión, la ecovalencia, el coeficiente de determinación y el coeficiente de variabilidad. Catorce genotipos y tres cultivares de caña de azúcar fueron evaluados en seis localidades durante dos años en Venezuela. La interacción G x A para rendimiento en caña y contenido de azúcar aparente (pol % caña) indica que los genotipos se comportan diferentes en cada localidad. La interacción de segundo orden no fue significativa para ambos caracteres. Los genotipos B80-549, B80-408 y B81-503 fueron significativamente superiores al resto de los materiales para rendimiento en caña, y B81-509, V84-25, B81-494 y B80-408 se destacaron en pol % caña. Este estudio sugiere que el análisis de estabilidad fenotípica puede contribuir con información suplementaria sobre el comportamiento de nuevas selecciones de caña de azúcar antes de la liberación como cultivares promisorios y puede incrementar la eficiencia en los programas de obtención de cultivares.

Resumo

A detecção da interação genotipo x ambiente (G x A) é do principal interesse para os melhoristas, sendo que as interações muito grandes podem reduzir o progresso esperado e complicar a identificação de cultivares superiores. Este trabalho teve por objetivo determinar a magnitude da interação G x A e avaliar a estabilidade fenotípica em cana-de-açúcar (Saccharum spp. Hybrid) mediante o uso do coeficiente de regressão, desvios da regressão, ecovalença, coeficiente de determinação e coeficiente de variação. Catorze genótipos experimentales e três cultivares testemunhas de cana-de-açúcar foram avaliados em seis localidades durante dos anos em Venezuela. A interação G x A para o rendimento de cana e o conteúdo de açúcar aparente (Pol % cana) indicou que os genótipos tem um comportamento diferente em cada ambiente. A interação de segundo ordem foi não-significativa para ambos caracteres. Os genótipos B80-549, B80-408 e B81-503 foram significativamente superiores ao resto dos materiais em rendimento de cana. B81-509, V84-25, B81-494 e B80-408 foram as mais destacadas em Pol % cana. Este estudo também sugere que a análise da estabilidade fenotípica pode contribuir com informação suplementar sobre o comportamento de novas seleções de cana-de-açúcar antes da liberação como cultivares promissórios e pode aumentar a eficiencia dos programas de obtenção de cultivares.

KEYWORDS / Adaptability / G x E Interaction / Phenotypic Stability /

Recibido: 10/05/2002. Modificado: 26/08/2002. Aceptado: 13/09/2002

Introduction

Genotype by environment (G x E) interaction complicates selection and testing of plant genotypes. Measuring G x E is important in order to determine an optimum strategy for selecting genotypes with adaptation to target environments (Romagosa et al., 1993; De Lacy et al., 1994; Annicchiriarico, 1997). In plant breeding programs, many potential genotypes are usually evaluated in different environments (locations and years) before selecting desirable genotypes. For quantitative traits such as yield, the relative performance of different genotypes often varies from one environment to another. Such statistical interaction results from changes in the relative ranking of the genotypes or changes in the magnitudes of differences between genotypes from one environment to another. Changes in ranking make it difficult for the plant breeder to decide which genotype should be selected (Nguyen et al., 1980).

The importance of G x E interactions in sugarcane selection is widely recognized (Milligan et al., 1990). A strong regional basis to G x E interactions has been identified in Queensland (Australia). Its magnitude varies from negligible in central Queensland to highly significant in North Queesland, where it is difficult to identify broadly adapted varieties. (Jackson et al, 1991; Bull et al., 1992; Jackson and Hogarth, 1992). In addition, Hogarth and Bull (1990) found that family x environment interactions in the Southern region were of sufficient magnitude to affect response to selection. Mirzawan et al. (1993) reported clone x location interactions as the more important source of clone x environment interactions and were much larger than location x crop-year and clone x location x crop-year interaction in Australia. They suggested that more emphasis should be placed on sampling a greater number of locations than on the testing of clonal ratooning ability within locations in early stages of selection in sugarcane. Given the diversity of sugarcane adaptation in Venezuela, there is an urgent need to test genotypes over a wider range of environments.

Several statistical methods have been developed for the analysis of G x E interaction (Hill, 1975; Lin et al., 1986; Wescott, 1986; Crossa, 1990; Flores et al., 1998). Plaisted and Peterson (1959) using a combined analysis of variance with pairs of cultivars suggested that lines with the smallest cultivar x location interaction were the most stable cultivars. Campbell and Kern (1982) used this analysis to study the stability of 10 sugarbeet (Beta vulgaris L) cultivars. The linear regression of genotype values on site mean yield (Finlay and Wilkinson, 1963; Eberhart and Russell, 1966; Pedersen et al., 1978), commonly termed joint regression analysis, is certainly the most popular method for stability analysis (Becker and León, 1988) due to its simplicity and the fact that its information on adaptive response is easily applicable to locations (Annicchiarico, 1997). Tai et al. (1982) evaluated phenotypic stability of sugarcane cultivars by measuring regression coefficient (b) and mean square deviations from regression (s2d) for some important traits. Ecovalence stability index (w) or the contribution of a genotype to the G x E interaction sum of squares proposed by Weber and Wricke (1990) have been used in sorghum (Sorghum bicolor L. Moench, tall fescue (Festuca arundinacea Schreb.), orchardgrass (Dactylis glomerata L.) and, sugarcane (Gray, 1982; Galvez, 1980; Nguyen et al., 1980; Kang and Miller, 1984). Shukla (1972) modified the ecovalence in order to give an unbiased estimate of the G x E variance for every genotype which he termed ‘stability variance’ (s2). Francis and Kannenberg (1978) utilized the coefficient of variation (cvi) of each genotype as a stability measure. Pinthus (1973) used the coefficient of determination (r2) and Lin and Binns (1988) developed a superiority index (Pi), defined as the mean squared distance between the genotype’s response and the maximum response averaged over all environments.

Although the regional variety trials have been a part of the sugarcane breeding programme at the Central-Western region of Venezuela for many years, the relative magnitudes of G x E interaction have not been documented. This paper reports and discusses the importance of G x E interaction and the utilization of stability analysis in the selection of sugarcane genotypes in Venezuela. The experiments were conducted to i) evaluate cane yield and sucrose content potential, ii) determine the nature of G x E interactions, and iii) study the adaptation of different sugarcane genotypes using stability parameters.

Materials and Methods

Fourteen experimental genotypes and three commercial cultivars of sugarcane were grown in replicated trials in the Central-Western region of Venezuela. The experimental genotypes were: B81-503, B80-549, B81-328, B81-66, B80-529, B81-42, B81-494, B80-621, B81-570, B80-408, B81-509, B81-59, V84-24, and V84-25. The cultivars evaluated were PR980, PR61-632, and V64-10. All materials were evaluated at six locations (Río Turbio and San Nicolás in Lara State, Yaritagua, Marín, Agua Negra and Matilde in Yaracuy State), each with 2 crop-years (plant crop and first ratoon) during 1996 and 1997. The traits studied were cane yield (tons cane·ha-1; TCH) and apparent sucrose content (pol % cane). The trial was laid out in a randomized block design (RCB) with three replications at each location. Plots were 3 rows wide, with 1.5m between rows, and 10m long. All three rows were harvested for measuring cane yield. The cane was burned and then cut by hand. A 10-stalk sample was randomly taken from each plot and weighed. The samples were milled and the crusher juice was analyzed for sucrose content.

Analysis of Variance. Data were combined over locations and analyzed as combined series of RCB’s with repeated measures (crop-year) using the General Linear Models (GLM) procedure (SAS, 1988a). All effects were considered fixed in the statistical model (Nguyen et al., 1980; Tai et al., 1982). Means were compared by Fisher‘s protected least significant difference (LSD; Steel and Torrie, 1980)

Stability analysis. Stability of the 17 genotypes for the two traits was estimated by using the coefficient of regression (b), mean squared deviations from regression (s2d), ecovalence stability index (w), coefficient of variation (cvi) and coefficient of determination (r2) using the GLM procedure (SAS, 1988a). A genotype with a regression coefficient >1.0 is responsive to increasingly favorable conditions with respect to site mean yield; a genotype with a regression coefficient <1.0 is considered not responsive. Small values of w and cvi are considered to be more stable. A genotype with a coefficient of determination of 1.0 would be more stable.

Correlation analysis. Phenotypic correlation of the mean cane yield of the 17 clones at one location with the yields at each of the other locations was used to determine location similarities (Campbell and Kern, 1982). Correlation coefficients among stability parameters for cane yield and pol % cane were also calculated using the SAS CORR procedure (SAS, 1988b).

Results and Discussion

The analyses of variance for TCH and Pol % Cane are presented in Table I. Genotype, genotype x location, crop-year, crop-year x genotype and crop-year x location interaction were significant for both traits. Such statistical interaction resulted from changes in the relative ranking of the genotypes or changes in the magnitudes of differences between genotypes from one environment to another. The variance due to crop-year x genotype and crop-year x location were significant, which revealed that ratooning ability and more than one site should be considered in the final stages of selection. Differently, Jackson et al. (1991) and Jackson and Hogarth (1992) found that clone x location interactions were more important than clone x crop-year interactions in Australia. Further, Jackson and Hogarth (1992) has suggested that, for early stages of selection where many clones are evaluated, testing only a plant crop may be satisfactory. The 3-way interaction was not significant. The relative large genotype mean square indicated that cultivars differed in their genetic potential for TCH and Pol % cane.

Correlation of cane yield (2yr means) of the 17 genotypes at pairs of locations did not reveal any unique locations (Table II). Correlation coefficients for more widely separated locations tended to be lower than those for locations nearer to each other, denoting gradual environmental change.

Yields for individual environments ranged from 84.3 TCH at Agua Negra to 146.5 TCH at Marin (Table III). It was assumed that these environments provided a representative sample of growing conditions in the Central-Western region of Venezuela. It is possible that selection of stable genotypes of sugarcane would be different if tested in a wider range of environments. However, this investigation makes apparent the magnitude of G x E interactions that must be confronted in sugarcane breeding and demonstrates genotype differences in response to several environmental conditions.

Two-year mean cane yields ranged from 84.41 to 119.66 TCH while pol % cane ranged from 12.44 to 14.65% (Table IV). B80-549, B80-408, and B81-503 were significantly (P £0.05) superior to the rest of genotypes for cane yield. The clones B81-509, V84-25, B81-494, and B80-408 had the best performance for pol % cane.

The regression coefficients of the genotypes ranged from 0.48 to 1.45 and from 0.01 to 1.76 for cane yield and pol % cane, respectively. The large variation in the regression coefficients indicated that genotypes had different environmental responses. Pfahler and Linsken (1979) pointed out that variability among environments determined, to a large extent, the usefulness of this regression response parameter. B80-408 and B81-503 appeared to be more responsive to favorable environments than the other genotypes as indicated by the relatively high regression coefficients and high yields (TCH) in higher yielding environments. B81-570 and B81-66 were less responsive to environmental change, as indicated by the lower regression coefficient for TCH and Pol % cane. In higher yielding environments, these genotypes lacked the ability to respond to the favorable conditions.

Mean yield was plotted against cvi following the methodology of Francis and Kannenberg (1978) (Figures 1 and 2). Mean cvi and grand mean (TCH, Pol % cane) divide the graph into four groups:

Group A: High yield, small variation

Group B: High yield, large variation

Group C: Low yield, small variation

Group D: Low yield, large variation

Figure 1. Mean TCH vs CVi for 17 genotypes over 12 environments

Figure 2. Mean Pol % cane vs CV for 17 genotypes over 12 environments

A stable genotype is one that provides high yield and consistent performance across environments. According to this definition, only group A can be considered as stable. Group C is consistent but is low in cane yield in most environments. This method could be used to identify genotypes on a group basis rather individually; however, the method can also be used in a plant-breeding context (Francis and Kannenberg, 1978).

The ecovalence stability index (w), and the coefficient of determination stability index (r2) complement the stability analysis.

Correlation between the mean and stability parameters ranged from very low to moderate (Table V). The r2, s2d, cvi, b, and w were significantly correlated with one another for TCH while the coefficient of regression (b), cvi, and r2, were significantly associated with each other for pol % cane. The correlations were significant, but not particularly high. According to Nguyen et al. (1980), the most desirable statistic among r2, s2d, and w would be the coefficient of determination (r2) because it is a standardized form and the results are comparable between experiments directly regardless of the measurement scale used.

Langer et al (1979) reported high correlations among r2, s2d, and w for three groups of oat (Avena sativa L.) cultivars. They concluded that any of these would be a satisfactory parameter for measuring stability.

Evaluations based on several crop-years and locations provide useful information to determine adaptation and stability of cultivars and provide knowledge of the magnitude and cause of the environmental effects in sugarcane breeding programs. Based on the different stability analyses, genotypes B80-549, B80-408, B81-503, and B80-529 were the most stable in TCH across environments tested showing broader adaptability. These data also suggest that the coefficient of variation or other stability statistics could be used in addition to mean yield by the sugarcane breeder in the selection process when genotype x environment interactions are present. We also recommend plotting the cvi against yield of each genotype in order to select genotypes with high yield and low cvi. Additionally, similar emphasis should be placed on sampling locations and ratooning ability within locations in sugarcane regional tests in the Central-Western area of Venezuela.

ACKNOWLEDGMENTS

The authors acknowledge the field assistance of Herman Nass, Milagros Niño, Anfer Ortíz and José George; the helpful feedback on the manuscript by Clarence W. Watson Jr, Mississippi State University; and statistical analysis by SAS. The research was supported by the Instituto Nacional de Investigaciones Agrícolas (INIA, Yaracuy).

REFERENCES

1. Annicchiarico P ( 1997) Joint regression vs. AMMI analysis of genotype-environment interactions for cereals in Italy. Euphytica 94: 53-62.        [ Links ]

2. Becker HC, León J ( 1988) Stability analysis in plant breeding. Plant Breeding 101: 1-23.        [ Links ]

3. Bull JK, Hogarth DM, Basford KE ( 1992) Impact of genotype x environment interaction on response to selection in sugarcane. Aust. J. Agric. Res. 32: 731-737.        [ Links ]

4. Campbell LG, Kern JJ ( 1982) Cultivar x environment interactions in sugarbeet yield trials. Crop Sci. 22: 932-935.        [ Links ]

5. Crossa J ( 1990) Statistical analyses of multi-location trials. Adv. Agron. 44: 55-85.        [ Links ]

6. De Lacy IH, Fox PN, Corbett JD, Crossa J, Raaram S, Fischer RA, Van Ginkel M ( 1994) Long-term associations of locations for testing spring wheat. Euphytica 72: 95-106.        [ Links ]

7. Eberhart SA, Russell WA ( 1966) Stability parameters for comparing varieties. Crop Sci. 6: 36-40.        [ Links ]

8. Finlay KW, Wilkinson GN ( 1963) The analysis of adaptation in a plant-breeding programme. Aust. J. Agric. Res. 14: 742-752.        [ Links ]

9. Flores FM, Moreno T, Cubero JI ( 1998) A comparison of univariate and multivariate methods to analyze G x E interaction. Field Crops Res. 47: 117-127.        [ Links ]

10. Francis TR, Kannenberg LW ( 1978) Yield stability studies in short-season maize. I. A descriptive method for grouping genotypes. Can. J. Plant Sci. 58: 1029-1034.        [ Links ]

11. Galvez G (1980) The genotype-environment interaction in experiments of sugarcane variety trials (Saccharum spp): Comparison of three stability methods. In Lopez MB, Madrazo CM (Eds.) Proc. XVII Congr. Int. Soc. Sugar Cane Technol. Manila, Philippines. pp. 1152-1160.        [ Links ]

12. Gray E ( 1982) Genotype x environment interactions and stability analysis for forage yield of orchardgrass clones. Crop Sci. 22: 19-23.        [ Links ]

13. Hill J ( 1975) Genotype-environment interactions-a challenge for plant breeding. J. Agric. Sci. 85: 477-493.        [ Links ]

14. Hogarth DM, Bull JK (1990) The implication of genotype by environment interactions for evaluation of sugarcane families. I. Effect on selection. In Kang MS (Ed.) Genotype-by-environment interaction and plant breeding. Department of Agronomy, Louisiana State University. pp. 335-346.        [ Links ]

15. Jackson PA, Hogarth DM ( 1992) Genotype x environment interactions in sugarcane. I. Patterns on response across sites and crop-years in North Queensland. Aust. J. Agric. Res. 43: 1447-1460.        [ Links ]

16. Jackson PA, Horsley D, Foreman J, Hogarth DM, Wood AW ( 1991) Genotype x environment (GE) interactions in sugarcane variety trials in the Herbert. Proc. Aust. Soc. Sugarcane Technologist 13: 349-355.        [ Links ]

17. Kang MS, Miller JD ( 1984) Genotype x environment interactions for cane and sugar yield and their implications in sugarcane breeding. Crop Sci. 24: 435-440.        [ Links ]

18. Langer SK, Fey J, Balley T ( 1979) Associations among productivity, production response, and stability indexes in oat varieties. Euphytica 22: 17-24.        [ Links ]

19. Lin CS, Binns MR ( 1988) A superiority measure of cultivar performance for cultivar x location data. Can. J. Plant Sci. 68: 193-198.        [ Links ]

20. Lin CS, Binns MR, Lefkovitch LP ( 1986) Stability analysis: where do we stand?. Crop Sci. 26: 894-900.        [ Links ]

21. Milligan SB, Gravois KA, Birchoff KP, Martin FA ( 1990) Crop effects on broad-sense heritabilities and genetic variances of sugarcane yield components. Crop Sci. 30: 344-349.        [ Links ]

22. Mirzawan PD, Cooper M, Hogarth DM (1993) The impact of genotype x environment interactions for sugar yield on the use of indirect selection in Southern Queensland. Aust. J. Agric. Res. 33: 629-638        [ Links ]

23. Nguyen HT, Sleper DA, Hunt K ( 1980) Genotype x environment interactions and stability analysis for herbage yield of tall fescue synthetics. Crop Sci. 20: 221-224.        [ Links ]

24. Pedersen AR, Everson EH, Grafius JE ( 1978) The gene pool concept as a basis for cultivar selection and recommendation. Crop Sci. 18: 883-886.        [ Links ]

25. Pfahler PL, Linskens HF ( 1979) Yield stability and population diversity in oats (Avena sp.). Theor. Appl. Genet. 54: 1-5.        [ Links ]

26. Pinthus MJ ( 1973) Estimate of genotypic value: A proposed method. Euphytica 22: 121-123.        [ Links ]

27. Plaisted RL, Peterson LC ( 1959) A technique for evaluating the ability of selections to yield consistently in different locations or seasons. Am. Potato J. 36: 381-384.        [ Links ]

28. Romagosa I, Fox PN, García del Moral LF, Ramos JM, García del Moral B, Roca de Togores F, Molina-cano JL ( 1993) Integration of statistical and physiological analyses of adaptation of near-isogenic barley lines. Theor. Appl. Genet. 86: 822-826        [ Links ]

29. SAS (1988a) SAS/STAT User‘s Guide. Version 6.3, 4th Ed., Vol. 2. SAS Institute, Inc., Cary, NC. 1028 pp.        [ Links ]

30. SAS (1988b) SAS procedures Guide. Version 6.0, 3rd Ed. SAS Institute, Inc., Cary, NC. 441 pp.        [ Links ]

31. Shukla, GK ( 1972) Some statistical aspects of partitioning genotype-environmental components of variability. Heredity 29: 237-245.        [ Links ]

32. Steel RG, Torrie JH ( 1980) Principles and procedures of statistics. 2nd. Ed. McGraw-Hill. New York.        [ Links ]

33. Tai PYP, Rice ER, Chew V, Miller JD ( 1982) Phenotypic stability analyses of sugarcane cultivar performance tests. Crop Sci. 22: 1179-1184.        [ Links ]

34. Weber WE, Wricke G (1990) Genotype x environment interaction and its implication in plant breeding. In Kang MS (Ed.) Genotype-by-environment interaction and plant breeding. Department of Agronomy, Louisiana State University. pp. 1-19.        [ Links ]

35. Wescott B ( 1986) Some methods of analyzing genotype-environment interaction. Heredity 56: 243-253.        [ Links ]