Early selection for growth vigor in rubber tree genotypes in northwestern São Paulo State (Brazil)

Early selection for growth vigor in rubber tree genotypes in northwestern São Paulo State (Brazil)

Paulo de Souza Gonçalves¹, Nelson Bortoletto², Fernando da Silva Fonseca³, Ondino Cleante Bataglia⁴ and Altino Aldo Ortolani⁵

¹Programa Seringueira, Centro de Café e Plantas Tropicais, Instituto Agronômico (IAC), Caixa Postal 28, 13001-970 Campinas, SP, Brasil. Send correspondence to P.S.G. Fax: [phone]. E-mail: [e-mail]

²Núcleo de Agronomia do Noroeste/IAC, Votuporanga, SP.

³Fellowship of the National Research Council - CNPq.

⁴Centro de Solos e Recursos Ambientais/IAC, Campinas, SP.

⁵Centro de Ecofisiologia e Biofísica/IAC, Campinas, SP.

ABSTRACT

Forty-five genotypes (clones) of rubber tree [Hevea brasiliensis (Willd. ex Adr. de Juss.) Müell. Arg.] of different origins were assessed for efficiency of early selection at the Votuporanga Experimental Station in northwestern São Paulo State, Brazil, using a randomized complete block design with three replications and six plants per plot. Girth at 120 cm above the highest point of the grafting union of each tree was taken at ages 12, 24, 36, 48, 60, 72, 84 and 96 months. Highly significant differences in girth were detected among the genotypes at all ages evaluated, except for 12 months. Estimates of genetic variance for two age sets showed a substantial increase with age, while the genotype variation decreased. Selection made at 24 months proved to be the most efficient, giving a superior gain per unit of time.

INTRODUCTION

Time is a critical factor in the genetic improvement of commercial traits for perennial species like rubber tree [Hevea brasiliensis (Willd. ex Adr. de Juss.) Müell. Arg.]. Breeding programs are usually designed to maximize genetic gain per unit of time, within the limits imposed by other constraints. For this reason, efficiency of early selection for growth before panel opening time is important for rubber tree breeders.

Most early selection studies estimate efficiencies and correlations based on comparisons between data from plants at various ages (Lambeth, 1980; Lambeth et al., 1983; Foster, 1986; Gill, 1987; Cotterill and Dean, 1988; King and Burdon, 1991;). These studies generally use environmental and site conditions commonly found in commercial plantations, and efficiency values are usually moderate to high for growth traits. Girth in rubber trees is an economically important character and is considered to be a measure of vigor. It determines the age at which a clone can be exploited.

The economic advantages of improved vigor are easily quantified and substantial. Justification of investment in an improvement program is more than met by the expected returns. The most expensive element of most tree breeding programs is the long process of evaluation, and rubber is no exception. Any reduction in evaluation test costs can greatly increase the financial viability of the breeding program.

Trees in the rubber tree plantation are tapped 120 cm above the point of the grafting union when 50% have attained at least 45 cm in girth, and six-seven years of age, depending on site quality (Bataglia et al., 1988). Selection cycles designed to allow comparisons of clone vigor would be prohibitively expensive in terms of plot size required, delay in realizing gain and length of generation cycle. To maximize the rate of genetic gain, the breeder must utilize evaluation results as soon as genotype performance can be reasonably predicted. The optimum time of assessment occurs when the additional gain achieved by more accurate later selections equals the additional costs of maintaining the evaluation tests. This means that anything that can be done to reduce the generation interval without severely reducing genetic gain per generation can result in a more efficient improvement program. The alternative is to carry out early selection, assessing genotypes or individuals at the youngest possible stage. Selection for growth vigor to achieve an earlier tapping age is effectively indirect selection of a juvenile girth with reliance on a correlated response in adult girth.

MATERIAL AND METHODS

Genetic materials used in the experiment were 45 hevea genotypes (clones) from different origins, which were introduced in a small scale trial (Table I) for evaluation. The clones were budded on to established rootstocks (Tjir 1 x Tjir 16) in a nursery. One-and-a-half-year-old rootstock seedlings raised in nurseries were used to budgraft the clonal materials. The successful budgrafts were uprooted and planted in plastic bags. The experiment was planted in the field after the first flush of leaves.

This work was done at the Votuporanga Experimental Station located in northwestern São Paulo State (Brazil) at 20º25'S latitude, 49º59'W longitude and 450 m elevation. Mean monthly temperatures varied from 20 to 25ºC, and annual rainfall totals for the duration of the experiment ranged from 1,087 mm to 1,537 mm. The winter drought varied from four to six dry months, with an average water deficiency of 180 mm.

The experimental design was a randomized complete block with three replications, using six trees per row with 7.0 m x 3.0 m spacing. Missing plants were replaced with spares during the first two years after planting to maintain plantation density, but were not scored. One row of the clone RRIM 600, acquired from a commercial nursery, was planted around the plot. Annual fertilizations consisted of 400 g of 10-10-10 formula NPK per plant, according to Balaglia and Gonçalves (1996).

The girth of each tree was measured once a year. The first measurement at 12 months was of diameter, because the plants were too small to measure girth. Plant diameter was measured 0.50 cm above ground level with a slide clipper. This measurement was converted to girth, assuming that the stem was cylindrical. Other measurements were taken at 24, 36, 48, 60, 72, 84 and 96 months at 120 cm above the highest point of the grafting union, using a measuring tape.

Biostatistical analysis

Analyses of variance

An analysis of variance was calculated for each age by considering only the general average as a fixed effect, and adopting the following statistical model:

where Y_ijk = observation of the ith genotype (clone) in the jth replication; m = general mean; g_i = effect of the ith genotype (i = 1, 2...g, g = 45); r_j = effect of the jth replication (j = 1, 2, ...r, r = 3); e_ij = experimental error associated within the ith genotype in the jth replication.

For the analysis of variance involving two different ages, the following statistical model was used, considering every effect as random except for the general mean and age:

where Y_ijk = observation of the jth genotype (clone) of the kth age in ith replication; m = general mean; r_i = effect of the ith repetition (i = 1, 2...r, r = 3); g_j = effect of the jth genotype (j = 1, 2, ...g, g = 45); b_ij = effect of the ith randomized block within the jth age; a_k = effect of kth age (k = 1, 2...a, a = 8); (ga)_jk = interaction between the jth genotype at kth age; (ra)_ik = interaction between the ith replication at the kth age; e_ijk = experimental error associated within the jth genotype at the kth age in the ith replication.

The estimates of genetic variances for each age and two age sets were calculated through "expected mean square" components (Table II).

Heritabilities

Estimates of genetic and phenotypic parameters for each age were obtained from the expected mean squares of the analysis using a procedure similar to the one presented by Vencovsky and Barriga (1992). The following expression was used for the estimation of the broad sense heritability ():

where is the total variance at ith age and is the environmental variance among the plots at ith age.

Genetic correlation

The genetic correlation (r_g) among the average performance of the genotypes at different evaluation months was estimated from algebric manipulation of the mean square (Table II) according to Vencovsky and Barriga (1992), to determine the degree of associations between given age pairs, using the following expression:

where Côv_g(xy) is the genetic covariance of ages x and y; . is the genetic variance among genotypes at ages x and y, respectively.

Genotype x age interaction breakdown

The breakdown of the component of genotype x age interaction () into simple and complex parts was made based on the expression proposed by Robertson (1959), adapted by Cruz and Castoldi (1991) and used by Marques et al. (1996) for situations in which genetic correlation is superior to 0.8, as was the case in this work, by the expressions:

According to Cruz and Castoldi (1991), the first part of the interaction component, + + , corresponds to the simple part. The remainder is denominated complex, due to a lack of correlation.

Correlated response to selection

To determine what occurs with genetic gain when selection is applied early, the correlated response to selection (CR_yx) was estimated following Falconer (1975):

where i = intensity of selection; = phenotypic variance among means.

Selection intensities of 11 and 20% were utilized for selection at 96 months. Fisher and Yates (1971) correction was used to calculate the intensity of selection, since the size of the genotype (clones) population was less than 50.

The formula for expected gain in percent of the mean can be written as:

where = represents the general mean of 96-month age girth.

RESULTS AND DISCUSSION

Highly significant differences in girth (P < 0.01) were detected among the genotypes at all ages evaluated, except for 12 months (Table III). Estimates of genetic variance () stood out, in which a substantial increase in the values along the ages was verified. At first, it could be inferred as an increase in genetic variability with age. However, when estimates of the genetic variation coefficient (CV_g) were used, it became apparent that variation tended to decrease with age compared to the average. So, it is evident that estimate values should be considered with reservation, since they are influenced by the magnitude of the data. This is corroborated by the heritability estimates (). The values at different ages are very similar and the small existing variation could be attributed to the error associated with these estimates. Both and values indicate that selection should be successful.

Significant differences (P < 0.01) were detected for every variation source in the analysis of variance involving the different ages (Table IV). Since the greatest interest is the interaction between genotypes evaluated at young ages and genotypes evaluated at tapping age, joint analyses of variance were done, considering the tapping age (96 months) and one of the other ages (Table V). It was found that tended to increase with ages closer to the tapping age, while the genotypes x age component () interaction had an inverse behavior. However, the / ratio always had small values and, when was broken down, the simple part of the interaction was always responsible for more than 60% of . This indicates that interaction was influenced by a difference in the genetic manifestation of the genotypes across different ages. This observation was reinforced by high, positive values of genetic correlation (r_g) of average genotype performance at juvenile age and girth at panel opening age.

Aiming to clarify what occurs with genetic gain when selection is applied early, the correlated response to selection was estimated (Table VI). Differences in estimated values of the correlated response to selection at different ages were small. They also differed very little from the estimate of gain at tapping age itself. If the focus is estimating the gain per unit of time the superiority of the correlated response to selection in relation to the selection done at tapping age was verified. This difference was more accentuated as the selection age was reduced. Under these circumstances, selection at 24 months would be recommended. It presented a superior gain to the selection done at 96 months.

ACKNOWLEDGMENTS

The authors are grateful to Mr. Ari de Camargo, agricultural technician from the Rubber Tree Program, and Mrs. Lígia Regina Lima Gouvêa for compiling the data. Research and publication supported by the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP). P.S.G., F.S.F., O.C.B. and A.A.O. are recipientes of CNPq fellowships.

RESUMO

Quarenta e cinco genótipos (clones) de seringueira [Hevea brasiliensis (Willd. ex Adr. de Juss.) Müell. Arg.] de diferentes origens foram avaliados para eficiência de seleção precoce na região noroeste do Estado de São Paulo. O ensaio foi conduzido na Estação Experimental de Votuporanga utilizando delineamento em blocos ao acaso com três repetições e seis plantas por parcela. O perímetro do caule de cada árvore/clone foi tomado a 120 cm de altura do calo de enxertia nas idades de 12, 24, 36, 48, 60, 72, 84 e 96 meses. Os resultados mostraram diferenças altamente significativas entre genótipos para todas as idades avaliadas exceto aos 12 meses. As estimativas da variância genética para idades combinadas mostraram um aumento substancial ao longo das idades, enquanto que as variâncias da interação genótipo x idade mostraram comportamento inverso. A seleção conduzida aos 24 meses mostrou melhor eficiência, propiciando um ganho superior por unidade de tempo. Os resultados da metodologia utilizada da seleção precoce mostraram que essa estratégia pode ser utilizada no melhoramento genético da seringueira.

(Received November 25, 1997)

• Bataglia, O.C. and Gonçalves, P. de S. (1996). Seringueira. In: Recomendaçăo de Adubaçăo e Calagem para o Estado de Săo Paulo (van Raij, B., ed). Instituto Agronômico, Fundaçăo IAC, Campinas, pp. 243 (Boletim Técnico No. 100).
• Bataglia, O.C., Cardoso, M. and Carreteiro, M.V. (1988). Situaçăo nutricional de seringais produtivos no Estado de Săo Paulo. Bragantia 47: 109-123.
• Cotterill, P.P. and Dean, C.A. (1988). Changes in the genetic control of growth of radiata pine to 16 years and efficiencies of early selection. Silvae Genet. 37: 138-146.
• Cruz, O.D. and Castoldi, F.L. (1991). Decomposiçăo da interaçăo genótipos x ambientes em partes simples e complexa. Rev. Ceres 38: 422-430.
• Falconer, D.S. (1975). Introduction to Quantitative Genetics. Oliver and Boyd, London, pp.365.
• Fisher, R.A and Yates, F. (1971). Tabelas Estatísticas para Pesquisa em Biologia, Medicina e Agricultura. Trad. de A.L. Haim, Editora Polígono, Săo Paulo, pp. 150.
• Foster, G.S. (1986). Trends in genetic parameters with stand development and their influence on early selection for volume growth in loblolly pine. For. Sci. 32: 944-959.
• Gill, J.G.S. (1987). Juvenile-mature correlations and trends in genetic variances in stika spruce in Britain. Silvae Genet. 36: 189-194.
• King, J.N. and Burdon, R.D. (1991). Time trends in inheritance and projected efficiencies of early selection in a large 17-year-old progeny test of Pinus radiata. Can. J. For. Res. 21: 1200-1207.
• Lambeth, C.C. (1980). Juvenile-mature correlations in Pinaceae and implication for early selection. For. Sci. 26: 571-580.
• Lambeth, C.C., van Buijtenen, J.P., Duke, S.D. and McCullough, R.B. (1983). Early selection is effective in a 20-year-old genetic test of loblolly pine. Silvae Genet. 32: 210-215.
• Marques, J.R., Andrade, H.B. and Ramalho, M.A.P. (1996). Assessment of the early selection efficiency in Eucalyptus cloeziana F. Müell in the Northwest of Minas Gerais State (Brazil). Silvae Genet. 45: 359-361.
• Robertson, A. (1959). Experimental Design on the Measurement of Heritabilities and Genetic Correlations. Biometrical Genetics Pergamon Press, New York, pp. 186.
• Vencosvsky, R. and Barriga, P. (1992). Genética Biométrica no Fitomelhoramento. Brazilian Genetics Society, Ribeirăo Preto, SP, pp. 469.

Publication Dates

• Publication in this collection
  01 Mar 1999
• Date of issue
  Dec 1998

History

• Received
  25 Nov 1997