Using performance data and reproductive measurements to predict fertility in replacement beef heifers

Using performance data and reproductive measurements to predict fertility in replacement beef heifers Arthur L. Jones,* Roy D. Berghaus, Matthew W. Studstill, Jacob S. Segers, Jason D. Duggin, # Patsie T. Cannon, and R. Lawton Stewart, Jr,1 *Department of Population Health, The University of Georgia, Tifton, GA; Department of Population Health, The University of Georgia, Athens, GA; Department of Animal and Dairy Science, The University of Georgia, Athens, GA, 30602; Department of Animal and Dairy Science, The University of Georgia, Tifton, GA, 31793; # Department of Animal and Dairy Science, The University of Georgia, Calhoun, GA, 30701 ABSTRACT: This study investigated the predictive value of reproductive tract evaluation and growth characteristics measured 30 70 d prior to the breeding season on 1) pregnancy outcome and 2) time to conception in replacement beef heifers. A total of 1,992 heifers (BW 329 ± 42 kg; age 347 ± 27 d) were delivered for enrollment in the Georgia Heifer Evaluation and Reproductive Development (HERD) program between 2006 and 2011 at two locations. Physical traits were selected to assess management of heifers prior to entering the program in addition to developmental traits traditionally measured in the HERD program and included: reproductive tract maturity score (RTS), weight 70 d prior to breeding as a percentage of target weight, hip height (HH) 40 50 d prior to breeding, and average daily gain 40 50 d prior to breeding. Cattle entered in the program were of similar age and subjected to comparable nutritional and management programs. Chi-square test of homogeneity (pregnancy status) and the Kaplan Meier product limit method (number of days from initial breeding to conception) were used to analyze univariate associations with predictor variables. Multivariate analyses of pregnancy status and time to conception were performed using logistic regression and Cox regression, respectively. The odds of pregnancy increased by 15% for every 2.5 cm increase in HH (P = 0.001), and by 20% for every 30-d increase in heifer age at the start of the breeding period (P = 0.019). Although RTS was associated (P = 0.015) with pregnancy status in the univariate analysis, after adjusting for the other variables included in the final multivariable model there was no significant association (P > 0.05). RTS and heifer age were not associated (P > 0.05) with time to conception in the multivariable analysis and were not included in the final model. However, HH was significantly (P = 0.005) associated with the time to conception after adjusting for location and year of enrollment. After 35 d, the hazard rate for conception increased 15% for every 2.5 cm increase in HH [hazard ratio (95% confidence interval) = 1.15 (1.04, 1.26); P = 0.005]. Variables intended to indicate prior management (average daily gain and weight 70 d prior to breeding as a percentage of target weight) were not found to be associated with pregnancy or time to conception. The results suggest that factors relating to maturity can be used to select heifers that are more likely to achieve pregnancy and have reduced times to conception. Key words: cattle, fertility, growth traits, heifer development Published by Oxford University Press on behalf of American Society of Animal Science 2018. This work is written by (a) US Government employees(s) and is in the public domain in the US.This Open Access article contains public sector information licensed under the Open Government Licence v2.0 (http://www.nationalarchives.gov.uk/doc/open-government-licence/version/2/). Transl. Anim. Sci. 2018.2:74 80 doi: 10.1093/tas/txy002 1 Corresponding author: lawtons@uga.edu Received October 11, 2017. Accepted February 27, 2018. 74

Predicting fertility in beef heifers 75 INTRODUCTION Effective selection and management of replacement beef heifers has a positive influence of the reproduction efficiency of beef herds (Patterson et al, 1992). Purchasing or developing replacement heifers is a significant capital investment for producers; therefore, it is important for cattle producers to consider the economics of replacement heifers (Hughes, 2013). Heifers are expected to calve at 2 yr of age and maintain a 12-mo calving interval. In order to reach puberty and achieve pregnancy in this time frame, various physiological events must take place in the heifer (Gasser, 2013). Puberty is attained when the heifer exhibits her first ovulatory estrus followed by a luteal phase of normal duration (Atkins et al., 2013). Age of puberty in heifers is associated with age of conception. Since heifers in a development program are normally bred during a defined period of 60 70 d, heifers that reach puberty prior to, or early in the breeding season, are more likely to breed (Perry and Cushman, 2013). Heritability of reproductive traits, such as first-service conception rate, is considerably low (0.03 ± 0.03) thus, management practices can greatly affect reproductive efficiency (Bormann et al., 2006). Selection of profitable heifers can be arduous in that many variables affect the ability of the animal to be productive over her lifetime. Selecting replacement females based on growth characteristics, such as hip height (HH), reproductive tract scoring, and pelvis area, prior breeding period have been shown to increase lifetime reproductive efficiency (Endecott et al., 2013); therefore, these growth characteristics need to be evaluated as true predictors of fertility. The objective of this study was to identify growth and reproductive measurements that related to pregnancy and time to conception in a group of beef heifers and assess the value of these measurements as a selection tool to increase reproductive efficiency in a heifer development program. MATERIALS AND METHODS All procedures involving animals were verified and approved by the University of Georgia s Office of Animal Care and Use. Animal Management The University of Georgia Heifer Evaluation and Reproductive Development Program (HERD) allows producers in Georgia and surrounding states to consign yearling heifers to a development program for educational purposes. For this study, a total of 946 and 949 heifers were assessed at HERD program facilities in Tifton and Calhoun, Georgia between the years of 2006 and 2011. Females in Tifton were born in September and November of each year and were delivered to HERD facilities during early October the following year. Heifers in the Calhoun program were born in December and February of each year and delivered to HERD facilities in early December the following year. Heifers had a mean age of 347 ± 27 d upon arrival at HERD facilities. Additionally, similar management practices were utilized at each location, including nutrition, estrous synchronization, and stocking rate. Methods of synchronization and/or heat detection differed across years. For years 2006 and 2007, the estrus synchronization protocol used in the heifers was the 7-d CIDR-PGF (an intravaginal device containing progesterone, controlled internal drug release CIDR, was inserted on d 0, removed 7 d later, and a dose of prostaglandin F2 α [PGF] was administered at the time of CIDR removal), and heifers were artificially inseminated upon detection of estrus. For years, 2008 2011, the estrus synchronization program was the 14-d CIDR-PGF with timed artificial insemination at 66 ± 2 h post PGF (a CIDR was inserted on d 0 and removed 14 d later; a single dose of PGF was administered 16 d after the CIDR was removed and heifers were artificially inseminated at 66 ± 2 h). However, all heifers were subjected to artificial insemination and placed with clean-up bulls for 58 d for a total breeding season ranging from 60 to 70 d. Pregnancy status was assessed with ultrasound 35 d after bulls were removed. Data Collection Heifers were weighed upon arrival at HERD facilities and again at intervals to assess 22-d, 37-d, 75-d, and 112-d average daily gain. HHs were measured at 60 d, while reproductive tract maturity score (RTS) and pelvic area (PA) measurements were assessed 30 d prior to the breeding season. HHs were determined as heifers passed by a measuring stick located on the back panel of the working chute. Pelvic measurements were assessed using a Rice Pelvimeter (Lane Manufacturing, Inc. Denver, CO). PA was attained by multiplying height and width of the pelvic opening. Heifers that did not achieve a PA of 140 cm 2 30 d prior to breeding were considered noneligible to sale due to potential for dystocia (Johnson et al., 1988). PA was used solely as a selection tool and was not analyzed as a predictor of fertility in this study. RTS were recorded by a veterinarian in each year of the study. Scores were given based on procedures described by

76 Jones et al. Anderson et al. (1991), with scores 1 (infantile) to 5 (mature and cycling). Similar to PA, heifers were considered noneligible for sale prior to the breeding period if a minimum RTS of 2 was not achieved to decrease the likelihood of heifers not breeding due to infantile reproductive tracts, absent tracts, or if the heifer was pregnant. Although females that did not meet the minimum requirements from PA and RTS were removed from the program, they could still proceed with the breeding protocol at the discretion of the owner. If a heifer was removed from the program, no data on that individual was used for analysis. In addition to these growth traits, weight gain during the time of development at HERD facilities, and prior to arrival at HERD facilities was measured. Management of heifers prior to arrival was unknown, thus, it was presumed that this may factor in to pregnancy status at the end of the breeding season. Therefore, average daily gain over the first 22 d was used to assess management prior to arrival based on potential compensatory gains. Statistical Analysis All statistical tests assumed a two-sided alternative hypothesis, and (P < 0.05) was considered significant. All analyses were performed using commercially available statistical software (Stata version 12.1, StataCorp LP, College Station, TX.) Prebreeding reproductive data were examined for correlations with reproduction outcomes. To identify variables that influence pregnancy, univariate associations between categorical predictor variables and the pregnancy status at the end of the breeding period was evaluated using a chi-square test of homogeneity. A multivariable logistical regression model was then constructed. Multivariable model selection proceeded from a maximum model containing all variables that were associated (P < 0.20) with pregnancy status in the univariate Table 1. Descriptive statistics for characteristics of 1,914 heifers enrolled in a Georgia beef heifer development program between 2006 and 2011 Variable Min, Max Mean (SD) Age at delivery, d 279, 404 347 (27) Age at the beginning of the 365, 476 426 (26) breeding period, d Weight at delivery, kg 214, 503 329 (42) Weight at delivery as a 63.6, 129.9 91.5 (10.1) percentage of target breeding weight, % Hip height 3 4 wk after delivery, cm 110, 137 122 (4) Average daily gain during the first 3 4 wk after delivery, kg/d 1.20, 3.67 1.04 (0.66) analysis. With the exception of enrollment year and location, variables were removed from the maximum model in a manual stepwise process until only those with (P < 0.05) remained. Enrollment year and location were included in all models regardless of their significance because of their theoretical importance as confounding variables. After reaching a preliminary main-effects model, all two-way interactions with heifer characteristic variables were evaluated. Both categorical and continuous forms of continuous predictor variables were assessed. The form of predictors that led to the lowest value of Akaike s Information Criterion (AIC) was preferred. The goodness of fit of the final model was evaluated using the Hosmer Lemeshow goodness of fit test (Hosmer and Lemeshow, 2000). Table 2. Characteristics and pregnancy outcomes of 1,914 heifers enrolled in a Georgia beef heifer development program between 2006 and 2011 Variable No. heifers (% pregnant) a P value b RTS 0.015 1 9 (55.6) 2 197 (81.7) 3 762 (84.7) 4 700 (86.1) 5 246 (89.4) Weight at delivery as a 0.934 percentage of target weight 63.6 79.9% 249 (84.3) 80.0 89.9% 622 (85.1) 90.0 99.9% 657 (85.8) 100.0 129.9% 386 (85.8) Hip height (cm) 3 4 wk <0.001 after delivery 111 118 283 (78.5) 119 122 740 (84.1) 123 126 548 (87.6) 127 137 343 (90.4) Age (d) at the beginning 0.047 of the breeding period 365 394 276 (81.5) 395 424 626 (83.9) 425 454 752 (87.0) 455 476 260 (88.5) Average daily gain (kg/d) during 0.642 the first 3 4 wk after delivery 1.20 to 0.58 481 (83.6) 0.59 1.03 482 (85.9) 1.04 1.43 460 (85.9) 1.44 3.67 491 (86.2) Total 1,914 (85.4) a Number of heifers in each category and the percentage that were identified as pregnant following a 60 70 d breeding period. b Chi-square test of homogeneity.

Predicting fertility in beef heifers 77 Table 3. Multivariable logistic regression model for the prediction of pregnancy status at the end of a 60 70 d breeding period in 1,914 heifers enrolled in a Georgia beef heifer development program between 2006 and 2011 Variable Coefficient (SE) OR (95% CI) a P value b Year Referent Referent Referent 2006 0.590 2007 0.241 (0.227) 0.79 (0.50, 1.23) 0.290 2008 0.076 (0.243) 0.93 (0.58, 1.49) 0.756 2009 0.289 (0.222) 0.75 (0.47, 1.19) 0.218 2010 0.289 (0.222) 0.75 (0.48, 1.16) 0.193 2011 0.357 (0.220) 0.70 (0.45, 1.08) 0.105 Location Calhoun Referent Referent Referent Tifton 0.078 (0.131) 1.08 (0.84, 1.40) 0.553 Hip height (2.5 cm increments) 3 4 wk after delivery 0.141 (0.044) 1.15 (1.06, 1.25) 0.001 Age (months) at the beginning of the breeding period 0.183 (0.078) 1.20 (1.03, 1.40) 0.019 Constant 7.42 (2.14) NA 0.001 CI = confidence interval; NA = not applicable. a Odds ratio and 95% CI. b Based on Wald statistics. Data were screened for outliers by examining plots of the predicted probabilities versus delta-deviance, delta-beta, and delta-χ 2. Univariate associations between categorical predictors and the time from the beginning of the breeding period to conception were evaluated using the Kaplan Meier product limit method in conjunction with the log rank test. Restricted mean conception times were calculated as the area under Kaplan Meier survival curves. Multivariable analysis of the time to conception was performed using Cox regression. Model selection followed the same approach as previously described for logistic regression models. The proportional hazards assumption was graphically evaluated for predictors by plotting ln ( ln) survival curves, and statistically evaluated by using a score test based on the scaled Schoenfeld residuals (Hosmer et al., 2008). Stratification was used to adjust for variables that failed to meet the proportional hazard assumption if they were primarily considered to be important as confounders. When it was desirable to estimate the effect of a variable that failed to meet the proportional hazards assumption, it was allowed to interact with a function of time in an extended Cox model (Kleinbaum and Klein, 2005). The most appropriate form of the interaction with time was determined by comparison of AIC values for competing models. RESULTS AND DISCUSSION Of the 1,992 heifers that were delivered for enrollment in the Georgia beef heifer development program between 2006 and 2011, 78 heifers left the program either before the start of the breeding period or before their pregnancy status could be determined. The reasons that heifers left the program included: small PA (n = 26), low RTS (n = 10), pregnant at arrival (n = 5), free martin (n = 4), death (n = 4), injury (n = 2), genetic disorder (n = 2), and unspecified (n = 25). Characteristics for the 1,914 heifers that completed the program are summarized in Table 1, and the distribution of heifers by categorical variables and the percentages in each category that were identified as pregnant at the end of a 60- to 70-d breeding period are summarized in Table 2. In the univariate analysis, RTS, HH 3 4 weeks after delivery, and age at the beginning of the breeding period were associated (P < 0.05) with the percentage of pregnant heifers (Table 2). Results of the multivariable analysis for the prediction of pregnancy status are summarized in Table 3. HH (P = 0.001) and heifer age (P = 0.019) were associated with pregnancy status after adjusting for the year and location of enrollment, respectively. The odds of pregnancy increased by 15% for every 2.5 cm increase in HH, and by 20% for every 1-mo increase in heifer age at the start of the breeding period. Although RTS was significantly associated with pregnancy status in the univariate analysis, it was not significantly associated with pregnancy after adjusting for the other variables that were included in the final multivariable model. The Hosmer Lemeshow goodness of fit test indicated that the final logistic regression model provided a good fit to the data (P = 0.660).

78 Jones et al. Table 4. Mean and median number of days to conception for 1,914 heifers enrolled in a Georgia beef heifer development program between 2006 and 2011 Variable No. heifers Mean (95% CI) Median (95% CI) P value a RTS 0.021 1 9 42.6 (23.8, 61.4) 57 (3, ) 2 197 24.9 (21.3, 28.5) 15 (4, 22) 3 762 21.1 (19.4, 22.9) 4 (3, 10) 4 700 20.1 (18.3, 21.9) 4 (3, 4) 5 246 19.0 (16.2, 21.8) 4 (3, 10) Weight at delivery as a 0.871 percentage of target weight 63.6 79.9% 249 21.8 (18.8, 24.8) 10 (4, 15) 80.0 89.9% 622 20.7 (18.8, 22.6) 4 (3, 10) 90.0 99.9% 657 20.4 (18.5, 22.2) 4 (3, 9) 100.0 129.9% 386 21.9 (19.4, 24.4) 4 (3, 14) Hip height (cm) 3 4 weeks after delivery 0.002 111 118 283 24.3 (21.2, 27.5) 9 (4, 15) 119 122 740 21.8 (20.0, 23.6) 4 (4, 12) 123 126 548 19.1 (17.1, 21.1) 3 (3, 4) 127 137 343 19.4 (17.0, 21.8) 4 (3, 10) Age (days) at the beginning of the 0.047 breeding period 365 394 276 23.5 (20.4, 26.6) 4 (3, 15) 395 424 626 22.2 (20.2, 24.1) 4 (4, 14) 425 454 752 19.6 (17.9, 21.2) 4 (3, 5) 455 476 260 19.5 (16.7, 22.3) 4 (3, 10) Average daily gain (kg/d) during the 0.840 first 3 4 weeks after delivery 1.20 to 0.58 481 21.2 (19.0, 23.4) 4 (3, 15) 0.59 1.03 482 20.8 (18.6, 23.0) 4 (4, 14) 1.04 1.43 460 20.1 (17.9, 22.3) 4 (3, 5) 1.44 3.67 491 21.7 (19.6, 23.8) 4 (3, 10) Total 1,914 21.0 (19.9, 22.1) 4 (4, 5) CI = confidence interval. a Based on a log rank test for equality of survivor functions. Literature notes the ability of HH to predict mature weights by calculating frame scores (Dhuyvetter, 1995). However, there is little research reporting the effect of HH on pregnancy outcome. Eler et al. (2014) reported that low genetic correlations between HH and heifer pregnancy (0.00) suggest different genes are responsible for influencing these traits. This is consistent with work by Silva Ii et al. (2003) who reported a genetic correlation between HH and heifer pregnancy status of (0.10) in Nelore cattle. These reports differ to findings of the current study that HH has a significant correlation to pregnancy outcome. However, genetic correlations were not assessed in the present study. In this study, it is thought that by selecting heifers with greater HH, indirect selection of heifers that were more mature most likely occurred. Holm et al. (2009) and Pence (2007) both reported significant (P < 0.01) univariate correlations between RTS category and pregnancy status after a 60- to 70-d breeding period, which agrees with findings of the current study. However, in multivariate logistical regression, Holm et al. (2009) reported that RTS was significantly associated (P < 0.01) with pregnancy outcome. These results differ from findings of the current study in that age and HH explained more variation in pregnancy outcome than RTS. Different veterinarians assigned RTS over the years of this study. Thus, it is possible that variation could have resulted in making this variable less reliable. This agrees with work by Rosenkrans and Hardin (2003) who found, on a scale of 0.0 1.0 (low to high), only moderate agreement (.46) between veterinarians assigning RTS. As previously mentioned, age was valuable as a predictor of fertility in this population of heifers in univariate and multivariate analysis. This observation disagrees with findings of Pence (2007) who reported an insignificant (P = 0.139)

Predicting fertility in beef heifers 79 Figure 1. Kaplan Meier survival curves illustrating the time to conception for 1,914 beef heifers by location (A), RTS (B), hip height 3 4 weeks after delivery (C), and age at the beginning of the breeding period (D). univariate correlation between age at the beginning of the breeding period and pregnancy status after a 60- to 70-d breeding period. Furthermore, using a multivariate logistical regression model, Holm et al. (2009) reported that age was not associated (P = 0.76) to pregnancy status after the first breeding season. Survival analysis was used to determine whether heifer characteristics were associated with the number of days from the beginning of the breeding period to conception. The time to conception is summarized by categories of predictor variables in Table 4. As in the analysis of pregnancy status as a dichotomous outcome, RTS (P = 0.021), HH (P = 0.002), and age at the beginning of the breeding period (P = 0.047) were all significantly associated with time to conception in the univariate survival analyses. Kaplan Meier survival curves for all variables that had an association (P < 0.20) with time to conception in the univariate analysis are shown in Figure 1. In the multivariable Cox regression analysis, location, year of enrollment, and HH all failed to meet the proportional hazards assumption. Consequently, location and year of enrollment were included in the model by stratification, and HH was included by allowing it to interact with a function of time that divided the follow-up period into segments before and after 35 d. Only HH was significantly associated with the time to conception after adjusting for location and year of enrollment. The effect of HH differed during the breeding period. HH was not associated with conception during the first 35 days (P = 0.204). However, after 35 d, the hazard rate for conception increased by 15% for every 2.5 cm increase in HH [hazard ratio (95% confidence interval) = 1.15 (1.04, 1.26); P = 0.005]. RTS and heifer age were not significantly associated with time to conception after adjusting for HH, enrollment year, and location, and consequently were not included in the final multivariable Cox regression model. Hazard ratio estimates were not available for enrollment year and location because these variables were incorporated in the final model by stratification. Results of survival analysis in the current study disagree with work by Pence (2007), who reported that time to conception for heifers with a RTS of 4 or 5 was significantly (P < 0.001) higher than heifers with RTS of 1 or 2. Little research exists that reports the relationship between HH and timing of conception. However, Eler et al. (2014) reported a negative genetic correlation between HH and age at first calving ( 0.01), which disagrees with the current study and suggests little relationship exists between HH and time that conception occurs. However, results of this study suggest that increased heifer HH was indicative of maturity, leading to more successful pregnancy outcomes earlier in the breeding period.

80 Jones et al. Results of this work suggest that maturity of a heifer at time of breeding plays a significant role in determining ability to achieve pregnancy and timing of conception. Univariate analysis showed that older heifers with higher RTS and HH were more likely to achieve pregnancy early in the breeding season. However, multivariate analysis showed in many of these variables are confounding. Reproductive tract scoring is based primarily on maturity of heifers which could be better accounted for by assessing age and HH prior to entering the breeding period in this population of heifers. Heifers were subjected to two methods of heat detection/synchronization across years in this study. The 14-d CIDR-PG protocol used in the final 4 years of this study is designed to induce estrus in peripubertal heifers. The estrogen combined with progestin affects the endocrine system of heifers in a way that mimics blood hormones around the time of puberty (Patterson et al., 2013). As a result, heifers in the final years of this study may have been more receptive to the synchronization protocol which could explain the discrepancies between the current study and results of Pence et al. (2007). Although heifers were subjected to various management practices prior to entering HERD facilities, body weight prior to arrival at HERD facilities, or weight gain during the developmental period did not affect final pregnancy outcome. Thus, variables selected in this study to investigate plane of nutrition prior to the breeding period was not associated with pregnancy outcome or timing of conception considering the variables tested in this study. Furthermore, it is believed that nutritional management from the time that heifers arrived at HERD facilities to first breeding period (approximately 70 d) was such that heifers were on similar planes of nutrition by first AI date, regardless of previous management. IMPLICATIONS Based on findings of this research, it is evident that growth characteristics that related to maturity were most important when selecting heifers that would achieve pregnancy early within a defined breeding season. Future research in this area should focus on the relationship of HH to heifer pregnancy outcomes and time to conception. It was apparent in this study that proper nutritional management during the last 70 d of the developmental period was adequate time achieve the same plane of nutrition across heifers coming from different management backgrounds. LITERATURE CITED Anderson, K.J., D.G. LeFever, J. S. Brinks, and K.G. Odde. 1991. The use of reproductive tract scoring in beef heifers. Agri-Practice. 12:106 111. Atkins, J. A., K. G. Pohler, and M. F. Smith. 2013. Physiology and endocrinology of puberty in heifers. Vet. Clin. North Am. Food Anim. Pract. 29:479 492. doi:10.1016/j. cvfa.2013.07.008 Bormann, J. M., L. R. Totir, S. D. Kachman, R. L. Fernando, and D. E. Wilson. 2006. Pregnancy rate and first-service conception rate in angus heifers. J. Anim. Sci. 84:2022 2025. doi:10.2527/jas.2005-615 Dhuyvetter, J. 1995. Beef Cattle Frame Scores. Extension Bulletin AS-1091. North Dakota State Univ., Fargo. Eler, J. P., A. B. Bignardi, J. B. Ferraz, and M. L. Santana, Jr. 2014. Genetic relationships among traits related to reproduction and growth of Nelore females. Theriogenology. 82:708 714. doi:10.1016/j.theriogenology.2014.06.001 Endecott, R. L., R. N. Funston, J. T. Mulliniks, and A. J. Roberts. 2013. Joint Alpharma-beef species symposium: implications of beef heifer development systems and lifetime productivity. J. Anim. Sci. 91:1329 1335. doi:10.2527/jas.2012-5704 Gasser, C. L. 2013. Joint Alpharma-beef species symposium: considerations on puberty in replacement beef heifers. J. Anim. Sci. 91:1336 1340. doi:10.2527/jas.2012-6008 Holm, D. E., P. N. Thompson, and P. C. Irons. 2009. The value of reproductive tract scoring as a predictor of fertility and production outcomes in beef heifers. J. Anim. Sci. 87:1934 1940. doi:10.2527/jas.2008-1579 Hosmer, D., and S. Lemeshow. 2000. Applied Logistic Regression. 2nd ed. New York: John Wiley & Sons, Inc. Hosmer, D., S. Lemeshow, and S. May. 2008. Applied Survival Analysis. 2nd ed. New York: John Wiley & Sons, Inc. Hughes, H. 2013. Raised replacement heifers: some economic considerations. Vet. Clin. North Am. Food Anim. Pract. 29:643 652. doi:10.1016/j.cvfa.2013.07.013 Johnson, S. K., G. H. Deutscher, and A. Parkhurst. 1988. Relationships of pelvic structure, body measurements, pelvic area and calving difficulty. J. Anim. Sci. 66:1081 1088. Kleinbaum, D., and M. Klein. 2005. Survival analysis: a self-learning text. New York: Springer. Patterson, D. J., R. C. Perry, G. H. Kiracofe, R. A. Bellows, R. B. Staigmiller, and L. R. Corah. 1992. Management considerations in heifer development and puberty. J. Anim. Sci. 70:4018 4035. doi:10.2527/1992.70124018x Patterson, D. J., J. M. Thomas, N. T. Martin, J. M. Nash, and M. F. Smith. 2013. Control of estrus and ovulation in beef heifers. Vet. Clin. North Am. Food Anim. Pract. 29:591 617. doi:10.1016/j.cvfa.2013.07.009 Pence, M., Ensley, D., Berghaus, R., Rossi, J., Wilson, T.,Cannon, P. T. 2007. Improving reproductive efficiency through the use of reproductive tract scoring in a group of beef replacement heifers. Bov. Pract. 41:35 40. Perry, G. A., and R. Cushman. 2013. Effect of age at puberty/conception date on cow longevity. Vet. Clin. North Am. Food Anim. Pract. 29:579 590. doi:10.1016/j.cvfa.2013.07.011 Rosenkrans, K. S., and D. K. Hardin. 2003. Repeatability and accuracy of reproductive tract scoring to determine pubertal status in beef heifers. Theriogenology. 59:1087 1092. Silva Ii, J. A. I. V., J. P. Eler, J. B. S. Ferraz, B. L. Golden, and H. N. Oliveira. 2003. Heritability estimate for stayability in Nelore cows. Livest. Prod. Sci.79: 97 101.