1- College of Agriculture & Natural Resources, University of Tehran, Karaj, Iran
Abstract: (965 Views)
Extended Abstract
Background: Due to the continuous and intermittent droughts in the central desert and its outskirts, only camels are capable of survival and production, and other livestock are not able to continue living in most of these areas under these conditions. Given the declining trend of the camel population in Iran, extensive studies are necessary to select suitable animals, and the implementation of breeding programs is particularly necessary. The first step in breeding programs and animal selection is to identify important economic traits and to accurately measure or estimate these traits using correct and accurate methods. Due to the location and spatial conditions of camel life, it is not possible in most cases to access scales for weighing. Since the age of one year is the age of selection among camel breeders for remaining in the herd or sale, the importance of having a correct equation of the dependence of weight on biometric traits at this age becomes more apparent than ever. Therefore, the need to provide an appropriate method in this regard has led to the use of external body measurements to evaluate live camels in terms of weight characteristics. The mechanisms involved in the control of most biological traits in living organisms are highly complex for interpretation by univariate analysis. Because most traits are biologically correlated through the pleiotropic effects of genes and the linkage of gene loci, this correlation causes estimates to be biased; as a result, multivariate methods are used to analyze data with high correlation. The principal component analysis (PCA) method is a mathematical method for converting a number of correlated variables into a smaller number of uncorrelated variables. Using PCA makes it possible to predict the exact size of the desired trait that cannot be measured while measuring a series of traits related to the desired trait. The present study mainly aims to compare the PCA method with the multivariate linear regression and ordinary least squares methods, and ultimately to present a suitable, simple, accurate, and efficient equation for estimating live weight from biometric traits of one-year-old camels.
Methods: In this study, 250 one-year-old camels were randomly selected from different regions of Semnan Province, and their biometric traits and weights were recorded after recording their gender and age. The probability of correlation between two groups was determined using the coefficient of determination (R2) formula. The variance inflation factor (VIF) formula was used to detect the intensity of multicollinearity in ordinary least squares regression analysis. Data were edited with Excel software. Multiple regression analysis and PCA were performed with SAS software. The presence of multiple alignment in the data was examined by examining the correlation between independent variables. After estimating the relevant eigenvalues and vectors, a number of eigenvalues were used to justify the most changes to create a new regression equation. The Kaiser-Guttman rule was used to select the number of eigenvalues. In the final stage, the coefficients of the regression equation were obtained with standardized data.
Results: The body weight, body height, body length, abdominal circumference, and chest circumference were the traits measured and estimated with mean values of 170 kg, 145, 116, 174, and 134 cm, respectively, and with an R2of 0.92. The presence of a high R2could be due to variance inflation resulting from collocation and one of the earliest signs of the existence of multiple collocation, which was determined by examining the high correlation between most variables. The variance inflation values for body height, body length, abdominal circumference, and chest circumference as the independent variables were estimated at 45.66, 0.39, 0.92, and 2.05, respectively, with the body height trait having the highest variance compared to the critical value of 5 to 10, where multiple collocation would occur. According to the results of the ordinary least squares analysis, the body height, body length, abdominal circumference, and chest circumference were important traits at the levels of 0.50, 0.49, 0.48, and 0.51 in estimating body weight, respectively, with the chest circumference and abdominal circumference traits having the highest and lowest importance, respectively. According to the PCA results, the body height, body length, abdominal circumference, and chest circumference were important traits at the levels of 0.95, 0.62, 0.76, and 0.78 in estimating body weight, respectively, with the body height and body length traits having the highest and lowest importance, respectively.
Conclusion: The results of this study show that the multiple correlation problem in the data related to the relationship between the weight of Semnani camels and four independent variables related to this trait can be solved using the PCA method. In the case of multiple correlation in the data, the PCA method is recommended for estimating the regression equation compared to the ordinary least squares method. In this case, the standard errors of the estimates will be lower than those the ordinary least squares method. Finally, the obtained results suggest that PCA can be used in the presence of alignment in multivariate linear regression analyses and has more accurate estimates than ordinary least squares methods. Moreover, it can be used with an effective reduction in the number of biometric traits required for use in breeding programs and the selection of superior individuals.
Background: Due to the continuous and intermittent droughts in the central desert and its outskirts, only camels are capable of survival and production, and other livestock are not able to continue living in most of these areas under these conditions. Given the declining trend of the camel population in Iran, extensive studies are necessary to select suitable animals, and the implementation of breeding programs is particularly necessary. The first step in breeding programs and animal selection is to identify important economic traits and to accurately measure or estimate these traits using correct and accurate methods. Due to the location and spatial conditions of camel life, it is not possible in most cases to access scales for weighing. Since the age of one year is the age of selection among camel breeders for remaining in the herd or sale, the importance of having a correct equation of the dependence of weight on biometric traits at this age becomes more apparent than ever. Therefore, the need to provide an appropriate method in this regard has led to the use of external body measurements to evaluate live camels in terms of weight characteristics. The mechanisms involved in the control of most biological traits in living organisms are highly complex for interpretation by univariate analysis. Because most traits are biologically correlated through the pleiotropic effects of genes and the linkage of gene loci, this correlation causes estimates to be biased; as a result, multivariate methods are used to analyze data with high correlation. The principal component analysis (PCA) method is a mathematical method for converting a number of correlated variables into a smaller number of uncorrelated variables. Using PCA makes it possible to predict the exact size of the desired trait that cannot be measured while measuring a series of traits related to the desired trait. The present study mainly aims to compare the PCA method with the multivariate linear regression and ordinary least squares methods, and ultimately to present a suitable, simple, accurate, and efficient equation for estimating live weight from biometric traits of one-year-old camels.
Methods: In this study, 250 one-year-old camels were randomly selected from different regions of Semnan Province, and their biometric traits and weights were recorded after recording their gender and age. The probability of correlation between two groups was determined using the coefficient of determination (R2) formula. The variance inflation factor (VIF) formula was used to detect the intensity of multicollinearity in ordinary least squares regression analysis. Data were edited with Excel software. Multiple regression analysis and PCA were performed with SAS software. The presence of multiple alignment in the data was examined by examining the correlation between independent variables. After estimating the relevant eigenvalues and vectors, a number of eigenvalues were used to justify the most changes to create a new regression equation. The Kaiser-Guttman rule was used to select the number of eigenvalues. In the final stage, the coefficients of the regression equation were obtained with standardized data.
Results: The body weight, body height, body length, abdominal circumference, and chest circumference were the traits measured and estimated with mean values of 170 kg, 145, 116, 174, and 134 cm, respectively, and with an R2of 0.92. The presence of a high R2could be due to variance inflation resulting from collocation and one of the earliest signs of the existence of multiple collocation, which was determined by examining the high correlation between most variables. The variance inflation values for body height, body length, abdominal circumference, and chest circumference as the independent variables were estimated at 45.66, 0.39, 0.92, and 2.05, respectively, with the body height trait having the highest variance compared to the critical value of 5 to 10, where multiple collocation would occur. According to the results of the ordinary least squares analysis, the body height, body length, abdominal circumference, and chest circumference were important traits at the levels of 0.50, 0.49, 0.48, and 0.51 in estimating body weight, respectively, with the chest circumference and abdominal circumference traits having the highest and lowest importance, respectively. According to the PCA results, the body height, body length, abdominal circumference, and chest circumference were important traits at the levels of 0.95, 0.62, 0.76, and 0.78 in estimating body weight, respectively, with the body height and body length traits having the highest and lowest importance, respectively.
Conclusion: The results of this study show that the multiple correlation problem in the data related to the relationship between the weight of Semnani camels and four independent variables related to this trait can be solved using the PCA method. In the case of multiple correlation in the data, the PCA method is recommended for estimating the regression equation compared to the ordinary least squares method. In this case, the standard errors of the estimates will be lower than those the ordinary least squares method. Finally, the obtained results suggest that PCA can be used in the presence of alignment in multivariate linear regression analyses and has more accurate estimates than ordinary least squares methods. Moreover, it can be used with an effective reduction in the number of biometric traits required for use in breeding programs and the selection of superior individuals.
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |