| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ANIMAL GENETICS |
Department of Animal Science, Iowa State University, Ames 50011
Abstract
The objective of the study was to estimate variance components, heritability, and repeatability of ultrasound longissimus muscle area (ULMA) measures. Data included 4,653 serial ULMA measures from 882 purebred Angus bulls and heifers. Animals were born over a 4-yr period from 1998 to 2001. Each year, bulls and heifers were ultrasonically scanned four to eight times, with a 4- to 6-wk interval between scans. Initially, data were subdivided by scan session across years and were analyzed in a multitrait model (MTM). Data pooled across years and scan session were then analyzed using random regression models (RRM) to estimate trends in genetic parameter estimates. Additive direct genetic variance increased with advancing scan session ranging from 8.67 cm4 at the first scan (mean age = 35 wk) to a maximum of 19.48 cm4 at the sixth scan (mean age = 56 wk). Heritability of ULMA increased from 0.35 at first scan to a maximum of 0.48 at the fourth scan (mean age = 50 wk). Additive direct genetic variance and heritability values at about 1 yr of age (fifth scan) were 18.24 cm4 and 0.45, respectively. Estimates from RRM also showed an increase in 
and h2with age. Trends in 
estimates, although tending to fluctuate, also increased with age. Additive direct genetic variance at 1 yr of age ranged from 15.8 cm4 to 17.0 cm4 for the different models. Heritability of yearling ULMA measures ranged from 0.40 to 0.42 and repeatabilities ranged from 0.80 to 0.84. For the range of ages used in the current study, both MTM and RRM showed close to maximum heritability values at around 1 yr of age. Therefore, phenotypic differences in yearling ULMA between Angus cattle are better indicators of genetic differences than earlier measurements. Angus breeders could, therefore, use ULMA measures made at around 1 yr of age to select next generation parents.
Key Words: Beef Cattle • Composition • Heritability • Ultrasound
Introduction
Several studies have been conducted to evaluate the use of ultrasound longissimus muscle area (ULMA) for genetic evaluation of beef cattle. Initial studies were limited to evaluating prediction accuracy by comparing slaughter and ultrasound measures (Perkins et al., 1992a
,b; Duello, 1993; Hassen et al. 1998). Subsequent studies based on field and experimental data have shown medium to high heritabilities of ULMA (Hough and Herring, 1999; Wilson et al., 1999; Reverter et al., 2000). Furthermore, genetic correlations of 0.48 to 0.71 between bull ULMA and half-sib steer carcass LMA measures suggested that these two traits are controlled by common genes (Kriese and McElhenny, 1995; Wilson et al. 1999).
Another issue in the use of ULMA and other ultrasound traits for genetic evaluation of body composition is the need to understand changes in genetic parameter estimates with ages at measurement. National cattle evaluation programs for the American Angus Association run routine genetic evaluations for ULMA and other ultrasound traits for bulls and developing heifers measured at about 365 and 395 d of age, respectively. These measurement ages were implemented based on practical herd management and also to allow breeding cattle to better differentiate themselves genetically compared with earlier measurements. However, to maximize genetic response to selection, ultrasound data should be collected at the earliest possible time, when individual animal phenotypic differences are the best indicators of genetic ranking. However, literature that relates changes in genetic parameter estimates for ultrasound traits to a wide range of ages at measurement does not exist.
The objective of this study was to estimate variance components, heritability, and repeatability of serial ULMA data in purebred Angus bulls and heifers measured from around weaning to 15 mo of age.
Materials and Methods
Source of Data
Animals in the present study came from the Iowa State University beef cattle breeding project. A detailed description of the project, including selection of base parents and management procedures, can be found in Hassen et al. (2003a). The project was designed to develop two lines of beef cattle for use as a research base to answer questions that influence genetic improvement of beef cattle. The two selection lines include a Quality line (Q-line) and a Retail line (R-line). Bulls in the Q-line are primarily selected for ultrasound-predicted percentage of intramuscular fat EPD and R-line bulls are selected primarily for ultrasound-measured longissimus muscle area (ULMA) and percentage of retail product (PRP) EPD.
Breeding and Management Procedures
The project was started in 1997 with the purchase of 285 spring-1996-born, purebred registered Angus heifers. The heifers were purchased from two herds in Nebraska and three herds in South Dakota. The heifers were randomly assigned to the two selection lines and were managed under similar condition at the Rhodes Research and Demonstration Farm located in central Iowa. Each year, breeding took place in June and July, with calving the following spring. In the first year of breeding, the heifers were bred artificially with semen from industry sires. In addition, mature cows and virgin heifers from the previous breeding project were implanted with a combination of fresh and frozen embryos obtained from three industry herds. In subsequent breeding, foundation cows and replacement heifers were bred artificially using semen from industry sires as well as from bulls selected in the project. Females within each line were bred as yearlings and were removed from the herd for reproductive failure. Criteria used in the selection of industry sires during the first and subsequent breeding are found in Hassen et al. (2003a).
After weaning, bull calves were fed a 1.3 Mcal NEg/kg (DM basis) corn-corn silage based diet to allow a mean weight gain of 1.5 kg/d. Replacement heifers were fed a 1.1 Mcal NEg/kg (DM basis) corn silage-based diet to allow a mean daily weight gain of 0.70 to 1.1 kg/d.
Animals and Scanning Procedure
Serial measures were collected on spring-born progeny from 1998 to 2001. For the first 3 yr, weaned bull and heifer calves were scanned four to six times for ULMA and other ultrasound traits starting at minimum ages of 28 wk, with an average interval of 4 to 6 wk between scans. The minimum age for 2001-born calves was 27 wk, and progeny were scanned up to eight times. Bulls and heifers were scanned using an Aloka 500V real-time ultrasound machine, equipped with a 3.5-MHz, 17.2-cm linear array transducer (Coromertics Medical Systems Inc., Wallingford, CT) or a Classic Scanner-200, equipped with a 3.5-MHz, 18-cm transducer (Classic Ultrasound Equipment, Tequesta, FL).
Data Analyses
Edits and Preliminary Evaluations.
Individual animal plots with unusual growth trends were often associated with poor-quality images in one or more scan session. Bulls and heifers with such trends including those with single observation per animal were excluded from the analysis. After editing, 4,653 observations from 396 bulls and 486 heifers were used for further analysis. The calves were from 47 sires and 336 dams. Ages at measurement were expressed in weeks and included 36 age points ranging from 27 to 62 wk.
Data were first analyzed based on mixed, linear models (SAS Inst. Inc., Cary, NC) to determine the level of fixed Legendre polynomial of age at measurement to model mean ULMA trend.
Data were then subjected to multiple-trait model (MTM) analysis to generate initial information on the general trend in variance component estimates, heritabilities, and genetic correlations. Serial ULMA data were divided into six subsets based on scan sessions across years. Few individuals were represented in Scans 7 and 8 and were therefore excluded from this part of the analysis. Data were analyzed using a six-trait animal model that included fixed effects of contemporary group (CG) (birth year, sex, and pen), linear effect of age at scan, random effects of animal, and an error term.
Random Regression Models.
Variance components were estimated by an average information REML algorithm using DXMRR (Meyer, 1998a). Meyer and Hill (1997) and Meyer (1998b) provided comprehensive information on estimation of (co)variance functions and likelihood values.
Data were subjected to genetic models that included fixed effects of CG (birth year, sex, pen, and scan session), fixed Legendre polynomial of age at measurement, and random regression coefficients on Legendre polynomial of age at measurement for direct genetic and direct permanent environmental effects. All genetic models assumed heterogeneous error variances based on preliminary information from phenotypic models. Different error variances were used for age ranges of 27 to 41, 42 to 47, and 48 to 62 wk. In comparing genetic models, the degree of Legendre polynomials of age at measurement (k) was sequentially increased by 1, to a maximum of k = 5, starting from a model that fits an intercept (k = 1). In all cases, the interest was to use simpler models with the same order of polynomial to characterize direct additive genetic and direct permanent environmental effects. The order of fixed polynomial for all models was kept constant at k = 4 (cubic) based on results from the preliminary evaluation.
The current study included relatively limited numbers of observations. Therefore, best-fitting models involving fewer numbers of parameter estimates were of interest. However, maximum log likelihood values often favor models with the largest possible dimensionality (Mills and Prasad, 1992) and Akaike’s information criteria (AIC) values may sometimes lead to model overfitting (Hurvich and Tsai, 1989; Mills and Prasad, 1992). Therefore, Schwartz’s Bayesian information criteria (BIC) were used as the primary criteria to evaluate models.
The general genetic model used is as follows.
 |
 |
 |
where
, 
Results and Discussion
Frequency distribution and means of ULMA measures for Angus bulls and heifers are shown in Figure 1. Except for extreme age groups, the distribution of ULMA measures across ages was nearly uniform. Number of observations per age group ranged from 24 at ages of 27 and 28 wk, to a maximum of 200 at 40 wk of age.
|
). The relative variation in bull and heifers data as determined by the coefficient of variation (not shown) were similar, ranging from 7 to 16% and from 7 to 20%, respectively.
Results from the MTM analysis are shown in Table 1. There was a general increase in with advancing scan session. Literature reports on the relationship between variance component estimates and measurement ages are not only limited in number but also lack consensus as to the direction of such a relationship. For instance, Johnson et al. (1993) reported an increase in for measurements made on Brangus cattle from 14.67 cm4 at weaning (29 wk) to 21.60 cm4 at 1 yr of age. After excluding data from progeny of embryo transfer cows, the authors reported of 15.04 and 20.58 cm4 for weaning and yearling measures, respectively. In contrast, Crews and Kemp (2001) reported a decline in estimates for composite bulls and heifers measured at 12 and 14 mo of age. Direct additive genetic variance of bull measures declined from 32.5 to 28.9 cm4. Heifer values for the corresponding ages were 18.45 and 16.30 cm4, respectively. Generally, differences in breed composition of cattle, improvement in ultrasound technology over the years, as well as data analysis procedures used in these studies may partly account for lack of unanimity of research results.
|
reported h2 of 0.11 for Angus cattle averaging 8 to 20 mo of age. However, based on Angus field records, Wilson et al. (1999) reported h2 of 0.37. Although measurement ages are out of the range as compared with the current study, Reverter et al. (2000) reported a pooled h2 estimate of 0.42 for Angus cattle averaging 66 to 70 wk of age. Correlations between ULMA measures at different scans increased as the number of weeks between scans decreased. Genetic correlations between yearling measures (Scan 5) and those of Scans 1, 2, 3, 4, and 6 were 0.91, 0.95, 0.96, 0.99, and 0.97, respectively. Phenotypic correlations (not shown) between yearling ULMA and those of Scans 1, 2, 3, 4, and 6 were 0.64, 0.68, 0.75, 0.85, and 0.83, respectively.
The maximum of log likelihood (log L) values and information criteria for genetic models are shown in Table 2. Maximum log likelihood values increased as the order of orthogonal polynomials of age increased. Similarly, AIC values favored a model with the highest degree of polynomial. The largest change in log L and AIC values occurred between Models I and II. However, according to BIC values, Model III remained the best model to describe the current data with the fewest possible numbers of parameter estimates. Although Model IV showed the lowest BIC value, this marginal improvement as compared to Model III is associated with eight additional parameter estimates.
|
) showed that factors associated with the first eigenvalue accounted for 94 to 98% of the 
. The corresponding value for permanent environmental (PE) covariance function ranged from 78 to 80%. On the other hand, for both covariance functions, the third and fourth eigenvalues were closer to zero.
|
. However, log L and AIC values for the model were –8,376 and 16,776, respectively. Although Model-k23 showed the best fit to the current data, estimates based on Models III and IV will also be included in further discussion.
Covariances and correlations between random regression coefficients based on the different models are shown in Table 3. Regardless of the model used, individual animal intercepts often showed larger variances than other coefficients for both direct additive genetic and PE effects. Coefficients for direct PE effect often showed larger variation as compared with corresponding values for direct genetic coefficients. However, coefficients for direct genetic effects were more strongly correlated with each other than those of PE coefficients.
The trends in based on the estimated covariance function for Models III, IV, and k23 are shown in Figure 2. Direct additive genetic variance increased as measurement age increased. Models III and IV showed similar values for ages 30 to 53 wk. However, estimates from Model IV were exaggerated for the oldest bulls and heifers that were represented by a relatively fewer number of observations. For most of the ages, values from Model-k23 were larger than those of Models III and IV. However, as compared to the rest of the models, estimates based on Model-k23 were often closer to values from MTM. At 1 yr of age, estimates from the different models ranged from 15.8 to 17.0 cm4.
|
, though tending to fluctuate at different intervals, showed a general increase with age (Figure 3). At 1 yr of age, ranged from 15.8 to 18.5 cm4. However, the difference in estimates between models is more apparent at extreme ages that were again represented by relatively few numbers of observations.
|
. Parameter estimates were averages of a class interval of 4 wk. Heritability values ranged from 0.29 to 0.44 and generally increased with age. Differences in estimates between models for extreme age groups were mainly caused by the relatively poor variance estimates. Mean yearling heritability estimates ranged from 0.39 to 0.42, and these values agree closely with previous estimate of 0.37 from Angus field records (Wilson et al., 1999). Currently, the National Cattle Evaluation program for the American Angus Association uses h2 of 0.38 for ULMA measures from bulls and heifers.
|
reported a pooled repeatability of 0.42 for crossbred feedlot cattle averaging 10 to 12 mo at the start of feeding. However, when data were analyzed by ultrasound technician, repeatability of ULMA ranged from 0.79 to 0.92. The current results are also in agreement with Perkins et al. (1992b), who reported a pooled repeatability of 0.81 based on measures from 36 steers representing four breed types. In agreement with results from the MTM, the genetic correlation between ULMA measures increased as the number of days between measurements decreased (not shown). Mean genetic correlations between yearling ULMA and measurements made at 35, 40, 46, 50, and 56 wk of age were 0.98, 0.99, 1.00, 1.00, and 1.00, respectively. The corresponding phenotypic correlations were 0.68, 0.75, 0.80, 0.80, and 0.80.
Generally, both MTM and RRM have demonstrated that bulls and heifers were more able to differentiate themselves genetically with advancing scan sessions or ages at measurement. Therefore, the use of a repeatability model to evaluate current data may not be appropriate. When data were analyzed based on such a model, , , and 
were estimated at 14.1, 13.6, and 11.9 cm4, respectively. Heritability and repeatability were estimated at 0.30 and 0.70, respectively. When heterogeneous error variances were assumed, and were estimated at 13.0 and 13.8 cm4, respectively. Heritability and repeatability ranged from 0.32 to 0.39 and from 0.66 to 0.80, respectively.
Considering the results from RRM, the trend in variance component estimates at extreme ages was generally unstable. Although these age groups were represented by fewer numbers of bulls and heifers, the problem could also be partly attributed to the general feature associated with fitting polynomials. As the use of Model IV involved estimation of more parameters than the rest of the models, this may exacerbate the problem for predicted variance values for relatively young and old ages. Furthermore, the results for the oldest age group could also be influenced by selection practices. Bulls that were available for scans at these ages were, for the most part, top-ranking bulls from their respective lines that were selected for breeding. Hence, results for these ages should be viewed with caution.
For the range of ages used in the current study, results suggest a medium level of genetic control for ULMA measures and that h2 values are close to maximum at around 1 yr of age. Therefore, differences in yearling ULMA measures in Angus cattle are good indicators of genetic potential for the trait than earlier measurements. In addition, previous reports of optimal h2 values for other ultrasound traits including percentage of intramuscular fat, 12- to 13th-rib fat thickness, and rump fat thickness (Hassen et al. 2003a,b) supports the current practice of scanning Angus cattle for all these traits at around 1 yr of age. The strong genetic association between yearling data and measurements made at other ages suggests that ULMA measures at different ages are controlled by the same set of genes. Therefore, selecting for ULMA at any of these mean ages would also increase yearling measures.
Implications
Ultrasound measures remain an important source of information to speed up genetic progress for compositional traits. The association between phenotypic measures and breeding values for ultrasound longissimus muscle area are close to maximum at around 1 yr of age. Any additional measurement made beyond this age may not bring a substantial improvement in the accuracy of selecting bulls and heifers. Data from good-quality images taken at 12 mo or at 13 to 14 mo of age for heifers can be used to accurately rank Angus cattle for genetic merit for ultrasound longissimus muscle area.
Footnotes
1 Journal paper of the Iowa Agric. and Home Econ. Exp. Stn., Ames; Project No. 3608, supported by the Hatch Act and State of Iowa funds. 
2 Correspondence: 239 Kildee Hall (e-mail: hassen{at}iastate.edu).
Received for publication July 17, 2003. Accepted for publication January 27, 2004.
Literature Cited
American Angus Association. 2003. Angus body composition genetic evaluation using ultrasound measures. Available: http://www.angus.org/sireeval/ultrasound.html. Accessed July 10, 2003.
Crews, D. H., Jr., and R. A. Kemp. 2001. Genetic parameters for ultrasound and carcass measures of yield and quality among replacement and slaughter beef cattle. J. Anim. Sci. 79:3008–3020.
Duello, D. A. 1993. The use of real-time ultrasound measurements to predict composition and estimates of genetic parameters of carcass traits in live beef cattle. Ph.D. Diss., Iowa State Univ., Ames.
Hassen, A., D. E. Wilson, and G. H. Rouse. 2003a. Estimation of genetic parameters for ultrasound-predicted percentage of intramuscular fat in Angus cattle using random regression models. J. Anim. Sci. 81:35–45.
Hassen, A., D. E. Wilson, and G. H. Rouse. 2003b. Heritability and repeatability of back fat and rump fat thickness in Angus cattle. J. Anim. Sci. 81(Suppl.1):348. (Abstr.)
Hassen, A., D. E. Wilson, R. L. Willham, G. H. Rouse, and A. H. Trenkle. 1998. Evaluation of ultrasound measurements of fat thickness and longissimus muscle area in feedlot cattle: Assessment of accuracy and repeatability. Can. J. Anim. Sci. 78:277–285.
Hough, J., and W. Herring. 1999. Genetic evaluation of ultrasound measurements for Hereford cattle. Pages 199–203 in Proc. Beef Improv. Fed. 31st Res. Symp. Annu. Mtg., Roanoke, VA.
Hurvich, C. M., and C. L. Tsai. 1989. Regression and time series model selection in small samples. Biometrika 76:297–307.
Johnson, M. Z., R. R. Schalles, M. E. Dikeman, and B. L. Golden. 1993. Genetic parameter estimates of ultrasound-measured longissimus muscle area and 12th rib fat thickness in Brangus cattle. J. Anim. Sci. 71:2623–2630.[Abstract]
Kriese, L. A., and W. H. McElhenney. 1995. Carcass and live animal evaluation, live animal measures. Pages 35–43 in Proc. 5th Genetic Prediction Workshop, Kansas City, MO.
Meyer, K. 1998a. "DXMRR"—a program to estimate covariance functions for longitudinal data by restricted maximum likelihood. Proc. 6th World Congr. Genet. Appl. Livest. Prod., Armidale, NSW Australia 27:465–466.
Meyer, K. 1998b. Estimating variance functions for longitudinal data using random a regression model. Genet. Sel. Evol. 30:221–240.
Meyer, K., and W. G. Hill. 1997. Estimation of genetic and phenotypic covariance functions for longitudinal or ‘repeated’ records by restricted maximum likelihood. Livest. Prod. Sci. 47:185–200.
Mills, J. A., and K. Prasad. 1992. A comparison of model selection criteria. Econometric Reviews 11:201–233.
Perkins, T. L., R. D. Green, and K. E. Hamlin. 1992a. Evaluation of ultrasonic estimates of carcass fat thickness and longissimus muscle area in beef cattle. J. Anim. Sci. 70:1002–1010.[Abstract]
Perkins, T. L., R. D. Green, K. E. Hamlin, H. H. Shepard, and M. F. Miller. 1992b. Ultrasonic prediction of carcass merit in beef cattle: Evaluation of technician effects on ultrasonic estimates of carcass fat thickness and longissimus muscle area. J. Anim. Sci. 70:2758–2765.[Abstract]
Reverter, A., D. J. Johnston, H.-U. Graser, M. L. Wolcott, and W. H. Upton. 2000. Genetic analysis of live-animal ultrasound and abattoir carcass traits in Australian Angus and Hereford cattle. J. Anim. Sci. 78:1786–1795.
Shepard, H. H., R. D. Green, B. L. Golden, K. E. Hamlin, T. L. Perkins, and J. B. Diles. 1996. Genetic parameter estimates of live animal ultrasonic measures of retail yield indictors in yearling breeding cattle. J. Anim. Sci. 74:761–768.[Abstract]
Wilson, D., G. Rouse, C. Hays, and A. Hassen. 1999. Genetic evaluation of ultrasound measures: Angus. Pages 187–198 in Proc. Beef Improv. Fed. 31st Res. Symp. Annu. Mtg., Roanoke, VA.
This article has been cited by other articles:
|
|
 |
|
  D. G. Riley, S. W. Coleman, C. C. Chase Jr., T. A. Olson, and A. C. Hammond Genetic parameters for body weight, hip height, and the ratio of weight to hip height from random regression analyses of Brahman feedlot cattle J Anim Sci, January 1, 2007; 85(1): 42 - 52. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
