Experimental tobacco plot size and design to breeding lines evaluation

Experimental tobacco plot size and design to breeding lines evaluation CORESTA CONGRESS September 2012 Sapporo, Japan

Experimental tobacco plot size and design to breeding lines evaluation TOLEDO, F.H.R.B.(1); RAMALHO, M.A.P.(2); PULCINELLI, C.E.(3); BRUZI, A.T.(4) (1) Department of Genetics, University of São Paulo, Piracicaba, Brazil; (2) Biology Department, Federal University of Lavras, Minas Gerais, Brazil; (3) Souza Cruz S.A., Tobacco Breeding Centre, Rio Negro, Brazil; (4) Agriculture Department, Federal University of Lavras, Minas Gerais, Brazil

Introduction Value for Cultivation and Use: VCU The experimental Value for Cultivation and Use (Piepho and Mohring, 2006) is a requirement of the Ministry of Agriculture, Livestock and Supply (MAPA) (Law Number 9456 of 04.25.1997), for the new cultivars release (BRAZIL, 1997); The rules for the VCU carry out are different according to the species and are set by the MAPA; Several aspects should be considered in the experimental strategy of VCU's, including the size, plot design and number of replications;

Introduction Value for Cultivation and Use:VCU To the tobacco crop information regarding to the plot size and design are scarce (FEHR, 1987). In the case of tropical and subtropical conditions were not found references. The most usual plot size for tobacco lines evaluation is formed by six rows, with seven plants per line (CREW and JOHNES, 1972; GUPTON and ARCHER, 1968); This relatively large plot limits, the number of the new lines to be tested (close to 10 lines).

Introduction Value for Cultivation and Use:VCU For the tobacco crop the rules for the VCU carry out were not defined; And tow major questions should be answered;

Introduction Value for Cultivation and Use:VCU Is it necessary the use of plots with borders? How is the number of plants per plot and their design to obtain reliable estimates of the cultivars mean under evaluation?

Objective The aim with this work was to obtain information about the strategies for carrying out Tobacco VCU trials to be evaluated at southern Brazil, in relation to the plot size and design.

Material and Methods Local The data was recorded from two different trials (for the set of 10 Burley and Virginia breeding lines), carried out by the Souza Cruz Company, in Rio Negro, Paraná State, Southern Brazil by 2009/2010 season.

Material and Methods Local The data was recorded from two different trials (for the set of 10 Burley and Virginia breeding lines), carried out by the Souza Cruz Company, in Rio Negro, Paraná State, Southern Brazil by 2009/2010 season.

Material and Methods Experimental Design and Evaluated Traits The experimental design was randomized complete blocks design with four replications, with ten lines in the final stage of evaluation. The plot was performed by six rows of seven plants each, total of 42 plants per plot. From each plant, referencing the position of its plant in the plot (row and column), was recorded the following traits: cured leaf weight (g/plant), total sugar content (%) and total alkaloids content (%);

Material and Methods Plot Arrangement: Row and Column 7 Utile Area Columns 6 5 4 3 2 1 1 2 3 4 5 6 Rows Data was recorded from individual plants Border Area

Material and Methods Experimental Plan: Trial Map 37,0 meters Front Line Sense Lenght Back 1 Plot CMF 641296 Line 1 Treat 6 lines of 7 plants 7.2 m x 3.5 m 36,00 meters Left Front of the Trial Right Corridor 1 m Plot 1 2 3 4 5 Line CMF 641296 CMF 641423 CMF 641427 CMF 641380 CMF 502109 Treat 1 4 5 2 9 8 7 6 3 10 Corridor 1 m Replication I CMF 502085 CMF 641474 CMF 641447 CMF 641400 CSC 444 10 9 8 7 6 1 8 9 10 6 Corridor 1 m CMF 641296 CMF 502085 CMF 502109 CSC 444 CMF 641447 11 12 13 14 15 4 3 7 2 5 Corridor 1 m CMF 641423 CMF 641400 CMF 641474 CMF 641380 CMF 641427 Replication II 20 19 18 17 16 9 8 1 3 6 Corridor 1 m CMF 502109 CMF 502085 CMF 641296 CMF 641400 CMF 641447 21 22 23 24 25 10 4 5 2 7 Corridor 1 m CSC 444 CMF 641423 CMF 641427 CMF 641380 CMF 641474 Replication III 30 29 28 27 26 10 1 9 7 8 Corridor 1 m CSC 444 CMF 641296 CMF 502109 CMF 641474 CMF 502085 31 32 33 34 35 5 4 6 2 3 Corridor 1 m CMF 641427 CMF 641423 CMF 641447 CMF 641380 CMF 641400 Replication IV 40 39 38 37 36 Corridor 1 m Back of the Trial

Material and Methods Data Analysis and Estatistical Procedure It was performed the variance analysis following the split- plot model according to Pimentel-Gomes (2009) in order to answer the question: Is it necessary plots with border?

Material and Methods Data Analysis and Estatistical Procedure The variance analysis was performed using three different schemes: Between location in the plot (Utile or Border plants); Between rows; Between columns; Columns 7 Utile Area 6 5 4 3 Border Area 2 1 1 2 3 4 5 6 Rows

Material and Methods Data Analysis and Estatistical Procedure Estatistical Model in which: Y ijk = μ+r i +t j +(tr) ij +s k +(rs) ik +(ts) jk +e ijk y ijk is the observation in the replication i, treatment j, position k; μ is a constant, the average of observation; r i is the replication i, with i = 1, 2,..., 4; t j is treatment j, with j = 1, 2,..., 10; (tr) ij is the error (a), between plots; s k is the position effect; (rs) ik is the error (b); (ts) ik is the interaction effect treatment i position k; and ε ijk experimental error, error (c), with ε N(0, σ 2 ).

Material and Methods Data Analysis and Estatistical Procedure It was also performed the resampling method (simulation) and Maximum Curvature Method of the Variation Coefficient (CV%) in order to answer the question: How is the number of plants per plot and their design to obtain reliable estimates of the cultivars mean under evaluation?

Material and Methods Data Analysis and Estatistical Procedure The resampling method was utilized as follows: an algorithm (bootstrap) on the data sheet draw lots of n plants for each plot, and the variance analysis was carried out for each one of these lots; The number of sampled plants (n) varied from 2 to 41 randomly selected without replacement and the process was repeated 2000 times for each n; All analyze were performed and/or implemented using the software R (R Development Core Team, 2011). Columns 7 Utile Area 6 5 4 3 2 1 1 2 3 4 5 6 Rows Border Area

Material and Methods Data Analysis and Estatistical Procedure For the maximum curvature of the coefficient of variation estimation the variance analysis errors was used to estimate the autocorrelation coefficient of first-order (ρ); According to Parnaiba, Francis & Mitchell (2009) the point of maximum curvature of the coefficient of variation can be obtained by;

Material and Methods Data Analysis and Estatistical Procedure Maximum curvature of the coefficient of variation in which: X 0 is the maximum curvature point of coefficient variation (CV%); ρ is the correlation coefficient; σ 2 is the experimental error variance; and Z is the plot average.

Results and Discussion Table 1. Summary of variance analysis for Burley varietal tobacco group, for cured leaf weight (g / plant), sugar content (%) and total alkaloids content (%), considering the border and the utile area plot. Rio Negro, Parana, 2009/2010. Source of Variation D.F. Curead Leaves (g/plant) Total Sugar (%) Tota Alkalloids (%) M.S. P (value) M.S. P (value) M.S. P (value) Replication 3 17421.52 0.09 3.39 < 0.01 0.89 0.05 Lines (L) 9 14993.3 0.07 0.35 0.58 1.16 < 0.01 Error (a) 27 7119.69 0.42 0.3 Border vs Utile (P) 1 37831.89 <0.01 0.06 0.12 0.58 < 0.01 Error (b) 3 561.2 0.02 0.07 Interaction L x P 9 651.01 0.27 0.03 0.33 0.05 0.56 Error 27 487.54 0.02 0.06 Average Accuracy Border Utile 282.29 259.52 302.96 0.72 1.68 1.65 1.71 0.86 3.97 4.06 3.89 -

Results and Discussion Table 2. Summary of variance analysis for Virginia varietal tobacco group, for cured leaf weight (g / plant), sugar content (%) and total alkaloids content (%), considering the border and the utile area plot. Rio Negro, Parana, 2009/2010. Source of Variation D.F. Curead Leaves (g/plant) Total Sugar (%) Tota Alkalloids (%) M.S. P (value) M.S. P (value) M.S. P (value) Replication 3 2547.6 < 0.01 2.48 0.55 0.23 < 0.01 Lines (L) 9 2594.44 < 0.01 1.86 0.84 0.22 < 0.01 Error (a) 27 202.5 3.49 0.04 Border vs Utile (P) 1 29220.68 < 0.01 0.25 0.42 0.19 < 0.01 Error (b) 3 17.8 0.58 0.01 Interaction L x P 9 176.82 0.54 0.27 0.68 0.01 0.41 Error 27 195.86 0.37 0.01 Average 249.36 5.68 2.26 Border 276.32 5.62 2.3 Utile 229.38 5.75 2.2 Accuracy 0.96 0.92 -

Results and Discussion The results for the two tobacco varieties groups for the three traits were very similar. The interaction lines position was not significant for the three traits. Buy another way, the behavior of the lines was similar regardless they were in the border or useful area.

Results and Discussion Table 5. Summary of variance analysis for Burley varietal tobacco group, for cured leaf weight (g/plant), sugar content (%) and total alkaloids content (%), considering the different plot rows. Rio Negro, Parana, 2009/2010. Source of Variation D.F. Curead Leaves (g/plant) Total Sugar (%) Tota Alkalloids (%) M.S. P (value) M.S. P (value) M.S. P (value) Replication 3 52962,55 0,08 10,11 < 0,01 2,67 0,05 Lines (L) 9 45716,19 0,06 1,08 0,57 3,49 < 0,01 Error (a) 27 21469,26 1,26 0,92 Betweenn Rows (P) 5 1442,44 0,42 0,18 0,03 0,29 0,06 Error (b) 15 2395,20 0,11 0,17 Interaction L x P 45 1079,46 0,86 0,05 0,86 0,20 0,04 Error 135 1431,44 0,07 0,13

Results and Discussion Table 6. Summary of variance analysis for Virginia varietal tobacco group, for cured leaf weight (g/plant), sugar content (%) and total alkaloids content (%), considering the different plot rows. Rio Negro, Parana, 2009/2010. Source of Variation D.F. Curead Leaves (g/plant) Total Sugar (%) Tota Alkalloids (%) M.S. P (value) M.S. P (value) M.S. P (value) Replication 3 7639,01 < 0,01 7,38 0,55 0,68 < 0,01 Lines (L) 9 7754,71 < 0,01 5,55 0,83 0,68 < 0,01 Error (a) 27 622,96 10,25 0,11 Betweenn Rows (P) 5 570,02 0,22 3,65 0,01 0,05 0,07 Error (b) 15 418,77 1,17 0,02 Interaction L x P 45 431,34 0,36 1,02 0,59 0,03 0,26 Error 135 398,61 1,09 0,03

Results and Discussion Table 7. Summary of variance analysis, for Burley varietal group, for cured leaf weight (g/plant), sugar content (%) and total alkaloids content (%), considering the different columns of the plot. Rio Negro, Parana, 2009/2010. Source of Variation D.F. Curead Leaves (g/plant) Total Sugar (%) Tota Alkalloids (%) M.S. P (value) M.S. P (value) M.S. P (value) Replication 3 61788,27 0,08 11,80 < 0,01 3,12 0,05 Lines (L) 9 53335,87 0,06 1,26 0,57 4,08 < 0,01 Error (a) 27 25047,23 1,47 1,08 Betweenn Columns (P) 6 68108,13 < 0,01 0,03 0,35 0,61 < 0,01 Error (b) 18 2717,62 0,02 0,11 Interaction L x P 54 2101,43 0,10 0,03 0,26 0,12 0,03 Error 162 1592,72 0,02 0,08

Results and Discussion Table 8. Summary of variance analysis, for Virginia varietal group, for cured leaf weight (g/plant), sugar content (%) and total alkaloids content (%), considering the different columns of the plot. Rio Negro, Parana, 2009/2010. Source of Variation D.F. Curead Leaves (g/plant) Total Sugar (%) Tota Alkalloids (%) M.S. P (value) M.S. P (value) M.S. P (value) Replication 3 8916,42 < 0,01 8,57 0,55 0,80 < 0,01 Lines (L) 9 9084,63 < 0,01 6,53 0,83 0,78 < 0,01 Error (a) 27 731,20 12,02 0,12 Betweenn Columns (P) 6 45891,77 < 0,01 1,48 0,07 0,20 < 0,01 Error (b) 18 789,63 2,28 0,04 Interaction L x P 54 528,71 0,91 0,98 0,09 0,02 0,23 Error 162 726,04 0,74 0,02

Results and Discussion It was found that there were no differences in average performance of lines in the different rows in the plot; The average performance of the lines varied between columns; The lines position interaction was not significant. That is, regardless of the row or column used was coincidence in the lines performance;

Results and Discussion FC Virginia Burley Figure 1. Curvature of the coefficient of variation (CV%) depending on the size of the plot (a) and point of maximum curvature (b). For cured leaves weight (g/plant), sugar content (%) and alkaloid content (%), Burley variety groups (top) and Virginia (below). Rio Negro, Parana, 2009/2010.

Conclusions The use of border is not justified because the interactions between lines position in the plots was not significant, showing that the behavior of the lines was similar in the different positions; The performance of the plants varied with the position column in the plot; The experimental precision evaluated by CV% increased with increasing the plot size, however, the method of maximum curvature of the coefficient of variation showed no significant increases in accuracy if the number of plants is greater than seven; The ideal number of plants for a tobacco plot would be larger than seven. The ideal plots would be formed by two rows of plants.

Siyabonga Muito Obrigado Many Thanks Alah hamdulila alah akbar Grazie mille Arigatougozaimasu बह त बह त धन यव Asante sana 非常感謝 Tack så mycket Terima kasih banyak Xie Xie ne Cảm ơn bạn rất nhiều Je vous remercie beaucoup Vielen Dank Dankie Çok teşekkür ederim