DESCRIPTION AND COMPARISON OF GROWTH PARAMETERS IN CHIANINA AND NELORE CATTLE BREEDS *

Rapid growth until slaughter weight is an important goal for increased meat production. Curves that relate weight or size with age have been used to describe growth. Growth parameters such as mature weight and general maturity rate, estimated from weights taken periodically during the life of the animal, can be used to evaluate development of animals (Brody, 1945). Growth equations are important in estimating parameters that are biologically uninterpretable, such as age at point of inflection of the growth curve, mature weight and maturity rate (Richards, 1959). These estimates should be used under the hypothesis that they simulate the true biological model (Brown et al., 1972). The growth functions of Richards, Logística, von Bertalanffy, Gompertz and Brody have been used to describe the Nelore growth (Duarte, 1975). Results suggested the function of von Bertalanffy was the most reasonable in describing the rapid growth and regular development of Nelore cattle. These growth functions have also been used by others (Wada et al., 1983) to describe bovine growth and, according to the authors, of all the growth models compared, the von Bertalanffy model was the easiest from a computational point of view. The curves derived from polynomial and von B rtalanffy models have been used to describe bovine growth (Vaccaro and Rivero, 1985). The authors verified that the von Bertalanffy model resulted in a curve that agreed better with conventional standards of biological growth. The von Bertalanffy function was selected as the most appropriate to evaluate development of beef cattle (López De Torre et al., 1992). The criteria used by the authors to compare the functions of von Bertalanffy, Brody and Richards were computational difficulty, goodness of fit and absence of weight bias at maturity. From the point of view of bovine development measured in terms of body weight at different ages, growth is a combination of hereditary and environmental effects. High heritability values for mature weight have been found by Northcutt and Wilson (1993) and Bullock et al. (1993). According to the authors, selection based on this characteristic should be effective, and the parameter mature weight can be genetically altered through selection. Adjustment of non-linear models algebraic functions to bovine weights could help detect genetic variability, a fundamentally important factor for genetic selection in breeding programs. From a genetic aspect, relationships that may eventually exist between the different non-linear growth parameters could provide alternatives for genetic selection programs. Genetic antagonism has been found between mature weight and maturity rate (Taylor and Fitzhugh Jr., 1971; Brown et al., 1972; Nelsen et al., 1982; Perotto, 1992; Bullock et al., 1993), indicating that selection for DESCRIPTION AND COMPARISON OF GROWTH PARAMETERS IN CHIANINA AND NELORE CATTLE BREEDS*


INTRODUCTION
Rapid growth until slaughter weight is an important goal for increased meat production.Curves that relate weight or size with age have been used to describe growth.Growth parameters such as mature weight and general maturity rate, estimated from weights taken periodically during the life of the animal, can be used to evaluate development of animals (Brody, 1945).
Growth equations are important in estimating parameters that are biologically uninterpretable, such as age at point of inflection of the growth curve, mature weight and maturity rate (Richards, 1959).These estimates should be used under the hypothesis that they simulate the true biological model (Brown et al., 1972).
The growth functions of Richards, Logística, von Bertalanffy, Gompertz and Brody have been used to describe the Nelore growth (Duarte, 1975).Results suggested the function of von Bertalanffy was the most reasonable in describing the rapid growth and regular development of Nelore cattle.These growth functions have also been used by others (Wada et al., 1983) to describe bovine growth and, according to the authors, of all the growth models compared, the von Bertalanffy model was the easiest from a computational point of view.
The curves derived from polynomial and von Bertalanffy models have been used to describe bovine growth (Vaccaro and Rivero, 1985).The authors verified that the von Bertalanffy model resulted in a curve that agreed better with conventional standards of biological growth.The von Bertalanffy function was selected as the most appropriate to evaluate development of beef cattle (López De Torre et al., 1992).The criteria used by the authors to compare the functions of von Bertalanffy, Brody and Richards were computational difficulty, goodness of fit and absence of weight bias at maturity.
From the point of view of bovine development measured in terms of body weight at different ages, growth is a combination of hereditary and environmental effects.High heritability values for mature weight have been found by Northcutt and Wilson (1993) and Bullock et al. (1993).According to the authors, selection based on this characteristic should be effective, and the parameter mature weight can be genetically altered through selection.Adjustment of non-linear models algebraic functions to bovine weights could help detect genetic variability, a fundamentally important factor for genetic selection in breeding programs.
From a genetic aspect, relationships that may eventually exist between the different non-linear growth parameters could provide alternatives for genetic selection programs.Genetic antagonism has been found between mature weight and maturity rate (Taylor and Fitzhugh Jr., 1971;Brown et al., 1972;Nelsen et al., 1982;Perotto, 1992;Bullock et al., 1993), indicating that selection for higher maturity rates could lead to lighter weights at maturity.On the other hand, positive and high genetic correlations were found between mature weight and birth weight, as well as between mature weight and yearling weight (Bullock et al., 1993) and, according to Northcutt and Wilson (1993), heifer weight taken before maturity could be used in genetic evaluations of mature weight.
The purpose of the present study was to analyze growth of Nelore (Bos taurus indicus) and Chianina (Bos taurus taurus) animals.The non-linear growth model proposed by von Bertalanffy (1938) was used and the growth parameters were estimated to describe and to compare the development of these two meat-producing bovine breeds.

Nelore breed animals
Nelore animals came from a herd of the Bonsucesso farm, Guararapes, State of São Paulo.They were kept in colonião grass pasture (Panicum maximum), with mineral salt and bone meal supplements available ad libitum.Mating took place in the field, year round; consequently, calves were born during all months of the year.Animals were weighed successively and monthly from birth to 18 months.Only healthy and pasture-fed pure animals (males and females) with a sufficient number of weights were included in the analyses.The sample was composed of 1358 animals, including monthly weight data from 1972from to 1978from and 1980from to 1982. Data were standardized for birth weight, 90, 150, 205, 365, 455 . Data were standardized for birth weight, 90, 150, 205, 365, 455 and 550 days.

Chianina breed animals
Data referring to the Chianina breed came from six herds from the States of São Paulo and Rio de Janeiro, and one herd from the State of Goiás.Animals were raised under a pasture regime.The reproduction system was by natural service; therefore, calves were born throughout the year.Weights were measured trimesterly from birth until 18 months with a tolerance of 15 days, limiting variation (in days) for age groups.Only healthy and pure animals (males and females) with all consecutive weights until 18 months of age were used.The 257-animal sample included data from 1973from to 1978from and 1980from to 1982. Trimester weights were standardized for birth weight, 90, 150, 205, 365, 455 and 550 . Trimester weights were standardized for birth weight, 90, 150, 205, 365, 455 and 550 days.

Obtaining growth parameters
The form of the model used in this study was a function of von Bertalanffy, given by the equation: Yt = A (1-Be -kt ) 3 , where Yt is weight at time t; t is the time (age of animal since birth); A is the asymptotic weight, final weight, or mature weight; B is the integration constant, and k is the maturity rate or measurement of the exponential function variation.
Model parameters can be interpreted biologically: 1. Asymptotic weight or mature weight is represented by the letter A in the equation.It is the weight achieved by organisms in growth, which is not surpassed by the elapsing exponential decline in the maturity rate.2. Inflection point occurs when the estimated maturity rate passes from a growing function to a declining function.In other words, the rate of change in weight is at its maximum and an exponential decline begins with an increasing reduction in the maturity rate.3. Slope is represented by the letter k.The estimate of the slope of a non-linear equation is the measurement of the rate of approximation to its asymptotic value.It is referred to as growth rate or maturity rate in relation to mature weight.The value k -1 is the interval of time spent to attain maturity and serves to measure changes in the degree of maturity (Taylor, 1965).4. Y-intercept is related to the initial weight of the animal.It represents weight t, when time equals zero and estimates birth weight.In the model, it is assigned the letter B and has no biological interpretation.It can be called the integration constant (Richards, 1959;Fitzhugh Jr., 1976).
The von Bertalanffy function parameters were adjusted according to the methodology presented by Draper and Smith (1966).

Standardizing weight
Monthly weights were standardized for birth weight, 90,150,205,365,455 and 550 days, using the following methodology (Warwick and Legates, 1979) where ADG is the average daily gain of the time interval in days relative to PA i (weight before standard age i) and PP i (weight after standard age i) and ∆days the interval in days among the weights considered.

Correcting data for environmental effects
Corrections for environmental effects were made using the program "Mixed Model Least Squares and Maximum Likelihood Computer Program" (Harvey, 1987).
∆days Growth curves in Chianina and Nelore cattle For Nelore breed animals, Y ijk = µ + a i + F j + e ijk where Y ijk is the dependent variable; µ is the general average; a i is a group of random effects (in this case the effect of the bull); F j is the group of fixed effects (month and year of birth, age of mother and sex), and e ijk is the experimental error.
For Chianina animals, where Y ijkl is the dependent variable; µ is the general average, a i is the group of random effects (representing herds); b ij is a hierarchical random effect (bull within the herd); F k is other group of fixed effects (year and month of birth, age of mother and sex), and e ijkl is the experimental error.

Estimates of heritability and genetic, phenotypic and environmental correlations
Estimates of heritability and genetic, phenotypic and environmental correlations were obtained through the methodology described in specialized literature.Heritability estimates were based on interclass correlations between paternal half brothers.Genetic, phenotypic and en-vironmental correlations among growth parameters were based on variance and covariance components between and among sires and error, respectively.

Environmental variation factors of growth parameters
For weights observed in Nelore animals, analysis of variance (Table I) showed that effects of the bull, year and month of birth, age of mother and sex were significant for all parameters.These results agreed with those obtained by other authors.Growth parameters have been influenced by effect of the age of the mother (Quaas, 1983;Wada and Nishida, 1987;Northcutt et al., 1994).The effect of the season was significant for the maturity rate (Nuru et al., 1981), as well as for mature weight and maturity rate (McLaren et al., 1982).The year of birth proved to be significant for the growth parameters and the effect of the bull's lineage for mature weight (DeNise, 1982;Quaas, 1983).The effects of the bull, month and year of birth and sex have been significant for the parameters mature weight and integration constant, while the maturity rate had been influenced by the effect of the bull (Bianchini Sobrinho and Duarte, 1991;Souza and Bianchini Sobrinho, 1994).Statistical analysis applied to growth parameter values obtained from Chianina animal weights (Table II) showed that the effect of the bull within the herd and sex were significant for all parameters, and the effect of age of the mother proved to be significant for the parameters integration constant and age at inflection point.

Heritability estimates for growth parameters
The heritability estimates found for Nelore animals were low (Table III).For the Chianina, heritability estimates were also low (Table IV), mainly for mature weight, age at inflection point, weight at inflection point and maturity interval.Such results suggest that selection for growth, based solely on curve parameters, may not lead to a satisfactory response.These results agreed with those obtained in the literature (Bianchini Sobrinho and Duarte, 1991;Souza and Bianchini Sobrinho, 1994).However, high heritabilities for the growth parameters were found (Wada and Nishida, 1987).High heritabilities for the parameter mature weight were observed by others (Northcutt and Wilson, 1993;Bullock et al., 1993), suggesting that this parameter could be genetically altered by selection.

Growth parameter values obtained with the von Bertalanffy model
In addition to the basic growth parameters, such as mature weight (A), maturity rate (k) and integration constant (B), the model can generate diverse growth measurements that permit the study of variations among the developmental standards that may not be reflected in the basic parameters.Among these are age at inflection point of the growth curve (T (I) ), weight at inflection point of the growth curve (P (I) ) and average maturity interval (1/k).This additional information allows more precise comparisons among the developmental standards.These statistics proved to be useful information regarding bovine growth (Duarte, 1975).
Average values were obtained for growth parameters of Nelore (Table V) and Chianina (Table VI) breed animals.Note that average mature weight (A) estimated for the Chianina breed (751.38 kg) is more than twice the value estimated for the Nelore breed (312.87 kg).Although presenting heavier average mature weights, Chianina exhibited maturity rates (k) inferior to those of Nelore by approximately 25%, which is deemed relevant considering the magnitude of this parameter.
Parameters A and k showed a pronounced inverse relationship in both breeds.Animals with heavier average A values presented the slowest average k values.On the other hand, those that grew with fast maturity rates were lighter at maturity.An inverse relationship between mature weight and maturity rate was also observed in the literature (Taylor and Fitzhugh Jr., 1971;Brown et al., 1972;Duarte, 1975;Nelsen et al., 1982;Perotto, 1992;Bullock et al., 1993;Perotto et al., 1994).
An inverse relationship between parameters k and 1/k was observed in both breeds.These observations agree with other authors' results (Brown et al., 1976;López De Torre and Rankin, 1978;Perotto et al., 1994).Note that the Chianina breed had slower maturity rates, and consequently, their maturity intervals were longer than Nelore.Average values for parameters T (I) , P (I) and 1/k were respectively: 3.297 months, 92.702 kg and 8.048 months for Nelore animals and 6.640 months, 222.63 kg and 10.989 months for Chianina animals.Observe that Nelore breed animals were younger and lighter at inflection point and at maturity than Chianina breed.

Correlations among growth parameters
Correlations among growth parameters of Nelore animals were estimated (Table VII).Observe that parameter k exhibited a negative genetic correlation coefficient with A, indicating that selection for faster maturity rates could result in a reduction mature weights.Negative genetic correlation coefficients between maturity rate and mature weight were also found in the literature (Taylor and Fitzhugh Jr., 1971;Brown et al., 1972;Calo et al., 1973;Duarte, 1975;López De Torre and Rankin, 1978;Nelsen et al., 1982;Perotto, 1992).
The parameter k was negatively correlated genetically with 1/k, agreeing with results found in the literature (Brown et al., 1976;López De Torre and Rankin, 1978;Perotto et al., 1994).This result indicates that selection for faster maturity rates could result in individuals that reach maturity much more quickly.The parameter k was negatively correlated genetically with B, P (I) and T (I) , meaning that selection for faster maturity rates could result in lower birth weights and weight at inflection point, and therefore, could reduce the age at inflection point.
Parameter A showed positive genetic correlation with 1/k, agreeing with Duarte (1975) and Bullock et al. (1993).The genetic correlation among A and T (I) was also positive, and there was an equal genetic correlation of +1.0 between parameters A and P (I) , suggesting that selection for heavier weights at the inflection could increase mature weight; however, animals would also be older at maturity.
The parameter 1/k had positive genetic correlations with B, T (I) and P (I) .Genetic correlations were positive between B and T (I) , B and P (I) , as well as between T (I) and P (I) , meaning that selection for heavier birth weights could result in older and heavier animals at inflection point, and selection for younger animals at inflection point could reduce weight at inflection point.

Correlations among different parameters and weights at diverse ages
The correlations among different parameters and weights at diverse ages of Nelore animals were obtained (Table VIII).Examine that mature weight had positive genetic correlation with birth weight.This result was also verified by Northcutt et al. (1994).This value tended to increase as age increased, agreeing with Bullock et al. (1993).For maturity rate, agreeing with results in the literature which indicate negative genetic correlation between parameters A and k, one would expect that genetic correlation between k and weight at different ages would tend to decrease as age increased.However, in this study such correlations tended to grow until 150 days of age, after which values tended to decrease.
For the period studied, from birth to 18 months, the results indicate that selection could be made at 150 days, choosing animals with faster maturity rates without reducing mature weight.Selection after or before 150 days for heavier weights could result in choosing animals with slower maturity rates that reach mature weight older.
Genetic correlation for the parameter 1/k tended to decrease until 150 days of age, after which it tended to increase with age even though values were negative.Considering the time interval studied, selection performed at 150 days, aiming to decrease the maturity interval, could result in heavier and younger animals at maturity.Selection for shorter maturity intervals before or after 150 days could result in animals that reach mature weight younger, and are therefore, lighter at maturity.Among P (I) and weight at different ages, the genetic correlations were positive and tended to increase with age, suggesting that selection for heavier weights at in-  (1975).Parameter B had low genetic correlation coefficients with weight at different ages, which although negative until the weight at 205 days, tended to increase with age.Genetic correlations between T (I) and weight at different ages tended to increase with age, even though values were negative.
For the period analyzed, the practice of selection for heavier weights at 150 days could lead to obtainment of animals that grow with faster maturity rates, are heavier and younger at inflection point and maturity.

Correlations among growth parameters
Correlation coefficients among growth parameters of Chianina animals were estimated (Table IX).The genetic correlation between A and k was negative, agreeing with results found in literature (Talyor and Fitzhugh Jr., 1971;Brown et al., 1972;Calo et al., 1973;Duarte, 1975;López De Torre and Rankin, 1978;Nelsen et al., 1982;Perotto, 1992).The genetic correlation between k and P (I) was also negative.Such results indicate that selection to faster maturity rates could lead to lighter weights at inflection point and maturity.The parameters A and P (I) had equal genetic correlation +1.0.These results indicate that selection for heavier weights at inflection point could lead to heavier mature weights.Parameter A presented negative genetic correlation with T (I) .Genetic correlation between P (I) and T (I) was also negative, meaning that selection for younger animals at inflection point could increase weight at inflection point and mature weight.These results did not agree with those found in literature.
Between 1/k and k the genetic correlation was negative, indicating that with faster maturity rates, maturity is reached at a shorter interval of time.These results were also observed by Brown et al. (1976), López De Torre andRankin (1978) and Perotto et al. (1994).The parameter 1/k was negatively correlated genetically with parameters A, B and P (I) , suggesting that selection for shorter maturity intervals could increase birth weight and weight at inflection point.These results do not agree with those found in literature.However, high standard errors were found suggesting that interpretations should be made cautiously.One of the factors that could explain this finding could be related to the small number of observations in the sample.
The parameter 1/k was positively correlated genetically with T (I) , meaning that selection for shorter maturity intervals could reduce age at inflection point.Parameters B and P (I) had positive genetic correlations with A, indicating that selection for heavier birth weights could result in heavier weights at inflection point and heavier mature weights.

Correlations among different parameters and weight at different ages
Correlation coefficients between different parameters and weight at different ages of Chianina animals were obtained (Table X).Note that standard error values varied between ±0.25 and ±0.50, which can be considered high.The small sample size of Chianina could have resulted in these values.Genetic correlations between A and weight at different ages tended to increase with age.This result agreed with those of Bullock et al. (1993).Therefore, selection for heavier weights at any age could lead to heavier mature weights.
The parameter k and weight at different ages presented genetic correlations which increased until 205 days of age; afterwards, it decreased.For this group of animals and the interval of ages analyzed, results suggested that selection for heavier weights at 205 days could lead to faster maturity rates without reducing mature weight.Selection for heavier weights, before or after this age, could result in animals with slower maturity rates and lighter mature weights.
The parameters A and P (I) showed genetic correlations with weight at different ages which were equal and tended to increase as age increased, inferring that evaluation of genetic potential for weight at maturity could be made early, reducing the interval between generations and, consequently, leading to better genetic gains per unit of time.
For parameter 1/k and weight at different ages, genetic correlations tended to decrease as age increased, until 205 days of age, after which it increased as age increased, although with negative values.Considering the developmental period analyzed, it is suggested that selection at 205 days for shorter maturity intervals could lead to faster maturity rates and heavier mature weights.From this age on, or before it, selection for shorter maturity intervals could result in lighter animals at maturity.The integration constant (B) had low genetic correlations that grew with age, although they were negative until 205 days of age.

CONCLUSIONS
Low heritabilities found for both breeds, mainly for mature weight, age at inflection point of growth curve, weight at inflection point of growth curve and maturity interval, suggest that selection for growth based solely on phenotypic values of curve parameters may not lead to a satisfactory response.
Nelore animals grow faster with shorter maturity intervals, and reach mature weight younger than Chianina.Despite its slow maturity rates and longer maturity intervals, Chianina exhibited mature weights superior to those of Nelore.
Selection for heavier weights at inflection point of growth curve leads to increased mature weight, in both breeds, allowing for early estimates of the genetic potential of the animals, contributing to increased genetic gains per unit of time.
Considering the developmental period studied, from birth to 18 months, the results suggested that the practice of genetic selection for heavier weights at 150 and 205 days of age for Nelore and Chianina animals, respectively, leads to animals that grow with faster maturity rates and reach the maturity younger and heavier.
Nelore animals reach the age considered ideal for selection for growth characters younger than Chianina animals, implicating on reduction of the interval between generations on Nelore breed animals, allowing for greater annual genetic gains on that breed.

ACKNOWLEDGMENTS
The authors greatly thank Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) for their financial support that made this work possible.Publication supported by FAPESP.

Table I
Analysis of variance for estimated growth parameters by von Bertalanffly's model, obtained from Nelore weight data.

Table III
Estimates of heritability of growth parameters estimated by the model of von Bertalanffy in Nelore animals.

Table V
Average values of growth parameters estimated by the model of von Bertalanffy obtained from growth data adjusted for significant environmental effects in Nelore animals.

Table VI
Average values of growth parameters estimated by the model of von Bertalanffy obtained from growth data adjusted for significant environmental effects in Chianina animals.

Table VII
Estimates of genetic, phenotypic and environmental correlations between growth parameters estimated by the model of von Bertalanffy for Nelore animals.

Table VIII
Estimates of genetic, phenotypic and environmental correlations between growth parameters estimated by the model of von Bertalanffy and weight at different ages for Nelore animals.

Table X
Estimates of genetic, phenotypic and environmental correlations between growth parameters estimated by the model of von Bertalanffy and weight at different ages forChianina animals.