texto en 
Inglés (pdf)
Artículo en XML
Referencias del artículo
Traducción automática
Enviar artículo por email
Citado por SciELO 
Similares en
SciELO 
Permalink
Introduction
Because of their high forage production, species such as elephant grass (Pennisetum purpureum) are of great importance in livestock production (Pegoraro et al., 2009). The elephant grass is a highly variable, tropical perennial plant able to adapt to the oscillating climatic conditions prevailing in Brazil (Valle et al., 2009). The productive potential of elephant grass fostered breeding programs for this species (Souza Sobrinho et al., 2005).
The genetic diversity of elephant grass is high in biometric and molecular traits, and it can be used by breeding programs (Cavalcante and Lira, 2010). The evaluation and selection of superior materials for specific areas depend mainly on the genotype × environment interaction (Silva et al., 2010). Productivity, as well as most quantitative traits, is polygenic in nature and highly influenced by the environment. Therefore, the genotype × environment interaction exerts a great influence on the expression of these traits (Schmildt et al., 2011).
In experiments with successive harvests and periodic evaluations over time, important parameters such as stability can be estimated. The stability or performance consistency of genotypes through a range of environments, which can also be expressed as their lowest average variation, which is dependent on the predictability of genotype response (Cruz et al., 2012). The method of Eberhart and Russell (1966) can be used to estimate the stability of each genotype in each environment, using linear regression analysis. This is computed as a single linear regression of the response variable for each genotype in each environment, weighted by the mean of each environment and by the overall mean (Ramos et al., 2011).
The aim of this study was to estimate stability parameters by the Eberhart and Russel method (1966) and select elephant grass genotypes with forage production stability and high yield in Campos dos Goytacazes, RJ, Brazil.
Materials and methods
The experiment was conducted at Centro Estadual de Pesquisa em Agroenergia e Aproveitamento de Resíduos, Pesagro RJ, located in Campos dos Goytacazes, RJ (21° 19’ 23” S and 41° 19’ 40” W; 13 m altitude). According to the Köppen (1948) classification, the climate is a hot and humid tropical Aw type, with annual precipitation at around 1152 mm. The soil was classified as a Distrophic Argisol (Embrapa, 2006), having: 18 mg dm-3 P2O5, 83 mg∙dm-3 K2O; 4.6 cmolC dm-3 Ca; 3.0 cmolC dm-3 Mg; 0.1 cmolC dm-3 Al; 4.5 cmolC dm-3 H + Al and 1.6 % C.
The experimental design was a randomized complete block with 80 treatments (genotypes) and two replicates. Plot was a 5.5 m-long row with 2 m between rows, totaling 11 m2. The dry matter yield (DMY) was calculated from the percentage of DM and tiller weight in 1.5 m of each plot. Results were converted to Mg ha-1.
The elephant grass was planted in February 2011, and two complete harvests were made to standardize plant growth, in December 2011 and in March 2012. After the standardization, another five evaluation harvests were performed: two in the dry season (June and August, 2012) and three in the rainy season (October 2012 and February and May, 2013). Over this period, the following 80 genotypes were evaluated: (1) Elefante da Colômbia, (2) Mercker, (3) Três Rios, (4) Napier Volta Grande, (5) Mercker Santa Rita, (6) Pusa Napier N 2, (7) Gigante de Pinda, (8) Napier N 2, (9) Mercker S. E. A, (10) Taiwan A-148, (11) Porto Rico 534-B, (12) Taiwan A-25, (13) Albano, (14) Hib. Gigante Colômbia, (15) Pusa Gigante Napier, (16) Elefante Híbrido 534-A, (17) Costa Rica, (18) Cubano Pinda, (19) Mercker Pinda, (20) Mercker Pinda México, (21) Mercker 86 México, (22) Taiwan A-144, (23) Napier S. E. A., (24) Pusa Napier N 1, (25) Elefante de Pinda, (26) Mole de Volta Grande, (27) Napier, (28) Mercker Comum, (29) Teresópolis, (30) Taiwan A-46, (31) Duro de Volta Grande, (32) Turrialba, (33) Taiwan A-146, (34) Cameroon - Piracicaba, (35) Taiwan A-121, (36) Vruckwona, (37) P241 Piracicaba, (38) IACCampinas, (39) Elefante C. Itap., (40) Capim Cana D’África, (41) Gramafante, (42) Roxo, (43) Guaçu/I.Z.2, (44) Cuba-115, (45) Cuba-116, (46) Cuba-169, (47) King Grass, (48) Roxo Botucatu, (49) Mineirão IPEACO, (50) Vruckwona Africano, (51) Cameroon, (52) CPAC, (53) Guaçu, (54) Napierzinho, (55) EMPASC 308, (56) EMPASC 310, (57) EMPASC 309, (58) IJ 7136 cv. EMPASC 307, (59) IJ 7139, (60) EMPASC 306, (61) Goiano, (62) CAC-262, (63) Ibitinema, (64) Australiano, (65) 13 AD, (66) 10 AD IRI, (67) 07 AD IRI, (68) Pasto Panamá, (69) BAG 92, (70) 09 AD IRI, (71) 11 AD IRI, (72) 06 AD IRI, (73) 01 AD IRI, (74) 04 AD IRI, (75) 13 AD IRI, (76) 03 AD IRI, (77) 02 AD IRI, (78) 08 AD IRI, (79) BAG UENF 79, and (80) BAG UENF 80.
An ANOVA was run for each variable in each harvest (environment). After checking the homogeneity of residual variances, a split-plot combined analysis was performed, considering genotypes as factor A and harvests as factor B. The following model was utilized:
where Y ijk = value observed in subplot i, j, k; μ = a constant inherent to every observation; ai = i-th level of factor a (i = 1, 2, ..., I); b k = effect of block k (k = 1, 2, ..., K); ab ik = experimental error at the plot level; b j = effect of the j-th level of factor B (j = 1, 2, ..., J); ab ij = effect of the interaction between factors A and B; e ijk = experimental error at the subplot level. The mean value in each harvest as well as the overall means of the genotypes in the five harvests were grouped using the Scott-Knott test (p≤0.05).
The method of Eberhart and Russell (1966) was used to obtain stable estimates, considering the genotypes and successive harvests as evaluation environments. The model used by Eberhart and Russell (1966) is described below:
where Y ij = performance of genotype i in environment j; mi = overall mean; bi = regression coefficient, which describes the response of the variation of genotype i at harvest j; I j = coded environmental index; d ij = deviation from the regression of genotype i in environment j; e ij = mean experimental error. Student’s t was performed to test the hypotheses H 0: b i = 1 and H 0: b i = 0. The hypothesis H0: s2 di = 0 was evaluated by the F test. Analyses were performed with Genes computer software (Cruz, 2013).
Results and discussion
The source of variation genotype was significant (F; p≤0.05), and the sources of variation harvest and genotype × harvest were also significant (F; p≤0.01) (Table 1). Thus, the significance of the interaction supports the study of adaptability and stability to identify genotypes with predictable performance and high yield.
Table 1 Summary of the analyses of variance for dry matter yield (DMY), in Mg ha-1, in the evaluation of 80 elephant grass genotypes over five harvests (different harvests were considered environments) (Campos dos Goytacazes, RJ, 2012/2013).
| Source of variation | D.F | S.Q. | Mean Squared | F |
|---|---|---|---|---|
| Blocks | 1 | 0.836 | 0.836 | |
| Genotype | 79 | 727.033 | 9.203 | 1.176† |
| Error a | 79 | 422.039 | 5.342 | |
| Cut | 4 | 3409.256 | 852.315 | 17.065† |
| Error b | 4 | 178.825 | 44.705 | |
| G x C | 316 | 1720.258 | 5.444 | 1.561† |
| Error c | 316 | 1101.929 | 3.487 | |
| Residue | 320 | 1280.755 | 4.002 | |
| Total | 799 | 7560.179 |
†Significant p≤0.01 and p≤0.05 % probability level, respectively, according to the F test
The significant interaction between genotype and harvest showed that the response of genotypes is not consistent throughout successive harvests; in other words, there are differences between their means in the evaluation of their performance over the five harvests. As it is a perennial crop, the elephant grass should be productive throughout its cultivation, so although there was a significant genotype × harvest interaction, what matters for the producer is that the genotypes have high performance over the harvests (Souza Sobrinho et al., 2005).
The results show genotypes with the highest DM yield (above 5.5 Mg ha-1) in the overall mean of the 80 genotypes from the five harvests. These genotypes were: Elefante da Colômbia (1), Gigante de Pinda (7), Hib. Gigante Colômbia (14), Elefante de Pinda (25), P241 Piracicaba (37), Gramafante (41), Guaçu/ I.Z.2 (43), Vruckwona Africano (50), CPAC (52), EMPASC 309 (57), IJ 7136 cv. EMPASC 307 (58), CAC-262 (62), Australiano (64), Pasto Panamá (68), 02 AD IRI (77), and 08 AD IRI (78). The genotype with the highest overall mean was 68 (Pasto Panamá), with 8.4 Mg ha-1 (Table 2; Figure 1).
Table 2 Arrangement of the most productive elephant grass genotypes ordered by overall means for dry matter yield (Yi), in Mg ha-1, and estimate of stability parameters (Yi and s2 di) proposed by Eberhart and Russell (1966), in five evaluation harvests (environments) (Campos dos Goytacazes, RJ, 2012/2013).
| Genotype | Y i | b i | s 2 di | Genotype | Y i | b i | s 2 di |
|---|---|---|---|---|---|---|---|
| 68 | 8.4 | 1.6349¶ | 0.0360 § | 41 | 5.8 | 1.8456† | 3.8708¶ |
| 58 | 7.1 | 1.5571¶ | -1.3059 § | 77 | 5.8 | 1.6890¶ | 1.4152 § |
| 43 | 6.7 | 1.9669 † | 8.0959 † | 14 | 5.7 | 1.1512 § | -0.0160 § |
| 1 | 6.3 | 1.1783 § | 4.4353¶ | 50 | 5.5 | 1.0062 § | 4.9344¶ |
| 57 | 6.3 | 1.2486 § | -1.1091 § | 52 | 5.5 | 1.0287 § | 1.0779 § |
| 25 | 6.2 | 1.6415¶ | 2.9670¶ | 62 | 5.5 | 1.3661 § | -1.4491 § |
| 37 | 6.1 | 1.1540 § | -1.3753 § | 78 | 5.5 | 2.0484† | 2.1703 § |
| 64 | 5.9 | 1.3643 § | -1.1791 § | 7 | 5.5 | 1.6625¶ | 2.5256 § |
† Significant at p≤0.05 %
¶ Significant at p≤0.01 %
§ Not significant (t test), respectively
The method of Eberhart and Russell (1966), evaluated the individual performance of elephant grass genotypes in response to temporal variations, by analyzing the harvests. This information is important in breeding programs because it allows the selection of genotypes with predictable responses (Souza Júnior et al., 2002). According to Cunha et al. (2013), the different methods to estimate forage production stability in Pennisetum spp. allow a better characterization of productive performance and, therefore, greater safety during selection.
Genotypes 68 (Pasto Panamá), 58 (IJ 7136 cv. EMPASC 307), 62 (CAC-262), 77 (02 AD IRI), 78 (08 AD IRI), and 7 (Gigante de Pinda) formed the group with the highest overall mean for DM yield, desirable regression coefficients, and s2 di not significant by the F test (Table 2). These results show that these six genotypes are able to respond to a favorable environment and have high yield capacity in adverse environmental conditions (Peluzio et al. 2010). Genotypes 43 (Guaçu/I.Z.2), 25 (Elefante de Pinda), and 41 (Gramafante) showed high DM yield, desirable regression coefficients, but significant bi and s2 di (p≤0.05 and p≤0.01).
Genotype Gramafante (41) showed a DMY of 14 Mg ha-1, one of the highest (Table 3; Figure 1). Genotypes 43 (Guaçu/I.Z.2) and 78 (08 AD IRI) had a DMY of 1.4 Mg ha-1 in the second harvest (Table 3; Figure 1). Leão et al. (2012) evaluated the forage production of hybrids between elephant grass and millet and observed the highest DMY, of 9.8 Mg ha-1, in elephant grass genotype Pioneiro, which is below the highest value found in our study (14.0 Mg ha-1). Meinerz et al. (2011) observed the same result when evaluating the elephant grass genotype Mercker Pinda for forage production in agro-ecological and conventional conditions, with a highest DMY (10.1 Mg ha-1) in the conventional tillage system.
Table 3 Dry matter yield, in Mg ha-1, of the most productive elephant grass genotypes in five harvests (environments) (Campos dos Goytacazes, RJ, 2012/2013).
| Genotype | Cut (environments) | Genotype | Cut (environments) | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | 1 | 2 | 3 | 4 | 5 | ||
| 1 | 8.6a | 3.2a | 3.3a | 11.1a | 5.2b | 52 | 8.8a | 3.7a | 3.2a | 8.2a | 3.5b |
| 7 | 11.8a | 3.1a | 1.8a | 7.8a | 2.8b | 57 | 10.8a | 4.2a | 3.4a | 7.5a | 5.6b |
| 14 | 7.9b | 3.2a | 1.9a | 8.1a | 7.1a | 58 | 12.2a | 4.4a | 2.8a | 7.7a | 8.2a |
| 25 | 11.9a | 3.1a | 1.8a | 5.0a | 9.2a | 62 | 10.2a | 3.6a | 1.7a | 6.3a | 5.5b |
| 37 | 9.4a | 3.4a | 3.2a | 7.7a | 7.0a | 64 | 10.7a | 4.2a | 2.1a | 7.0a | 5.4b |
| 41 | 14.0a | 2.9a | 2.2a | 4.9a | 5.1b | 68 | 12.9a | 7.1a | 2.4a | 10.4a | 9.3a |
| 43 | 13.2a | 1.4a | 2.3a | 5.5a | 11.2a | 77 | 12.8a | 2.8a | 2.5a | 6.1a | 4.7b |
| 50 | 6.9b | 2.9a | 1.7a | 6.3a | 9.8a | 78 | 13.6a | 1.4a | 1.6a | 6.8a | 3.9b |
Means with different letter in a column are statistically different (Scott-Knott test; p≤0.05).
Conclusions
There are differences between the mean values of the genotypes or in the classification of their performance over the five harvests. Genotypes Pasto Panamá, IJ 7136 cv. EMPASC 307, CAC-262, 02 AD IRI, 08 AD IRI, and Gigante de Pinda showed high forage production and phenotypic stability over the five harvests.
