Journal of Animal Science

Duncan, Natalie, B;Meyer, Allison, M

doi: 10.1093/jas/sky448pmid: 30476110

Abstract Changes in locomotor activity of beef females during the 72 h prepartum were determined in 3 experiments: (i) a 2-yr study with spring-calving multiparous cows (Exp. 1; n = 34 and 27 for years 1 and 2, respectively); (ii) spring-calving primiparous (first pregnancy; n = 13) and multiparous (n = 21) dams (Exp. 2); and (iii) fall-calving multiparous cows (Exp. 3; n = 33). For all experiments, IceQube activity monitors (iceRobotics, Edinburgh, UK) were placed above the left hind fetlock of pregnant females ≥3 d prepartum. During the calving season, females were housed in 18 × 61 m drylots with ad libitum access to hay or haylage. Parturition was closely monitored, and time of birth was noted. Motion index, standing and lying time, step count, and the number of lying bouts for each dam (summed per hour) were determined for the 72 h preceding calving. Within experiment, data were analyzed by day (days −3 to −1 prepartum), by 6-h period during the final 24 h prepartum, and by hour during the final 6 h prepartum using a mixed model with time as a repeated effect. Year was also included as a fixed effect in Exp. 1. Fixed effects of parity and time prepartum × parity were included for Exp. 2. In all 3 experiments, motion index, standing time, step count, and the number of lying bouts were greater (P < 0.001) on day −1 compared with days −2 and −3 prepartum. In the 24 h prepartum, dams had greater (P < 0.01) motion index, standing time, step count, and the number of lying bouts during 6 h preceding parturition compared with −11 to −6 h in all experiments. Motion index, step count, and lying bouts changed (P ≤ 0.02) during the last 6 h in all experiments. Primiparous dams had more (P ≤ 0.01) lying bouts than multiparous dams during the last day and −11 to −6 prepartum. In all experiments, the number of lying bouts more than doubled (P < 0.001) from −2 to −1 prepartum, with no effect of year (P = 0.57) in Exp. 1 or parity (P ≥ 0.29) in Exp. 2. This suggests that lying bout changes may be the most reliable of parameters measured in detection of calving. Moreover, fall-calving cow behavioral patterns were similar to changes observed in spring-calving females, suggesting that calving season may have minimal effects on pre-calving behavior. Overall, electronic locomotor activity monitors can detect behavioral changes peripartum in beef heifers and cows. More research is necessary to determine if these can be used to remotely sense early signs of parturition in beef cattle. INTRODUCTION In 2010, 20.9% of reported nonpredator beef calf preweaning death loss was due to calving problems (USDA-APHIS, 2010). Individual monitoring of dams is required to identify calving difficulties as early as possible, allowing for human intervention to minimize the negative impacts of dystocia and prolonged calving on calf survival (Dargatz et al., 2004; Wehrend et al., 2006). Remote detection of calving in a herd setting through prediction technology could reduce calf mortality caused by dystocia or other calving-related health problems such as hypothermia (Saint-Dizier and Chastant-Maillard, 2015). Beef producers would benefit from this technology if it was economical, required minimal additional animal handling, was useful in large herd settings, and allowed for a diverse array of animal housing dependent upon operation type. Behavioral monitoring using accelerometers and pedometers has been used in the dairy industry for health and estrus detection, allowing for behavioral observation of multiple cows without the need to physically observe each animal individually (Fogsgaard et al., 2015; Tsai, 2017; Finney et al., 2018). These devices can be left on for long periods of time, making timing of application and removal flexible, as well as decreasing the amount of handling required near calving. Commercially available accelerometer devices such as IceTag and IceQube products (iceRobotics, Ltd., Edinburgh, UK) have been well validated against visual behavioral monitoring in dairy cattle (Nielsen et al., 2010), where use of this technology for early recognition of parturition is being explored (Huzzey et al., 2005; Miedema et al., 2011b; Borchers et al., 2017). Use of accelerometers in beef cattle has been limited to research in the early detection of feedlot morbidity (Pillen et al., 2016; Richeson et al., 2018), but dairy data suggest that these may have utility in detection of calving in beef females as well. Monitoring of calving in beef cattle is often concentrated in heifers due to increased dystocia associated with first parity females (Berger et al., 1992; Zaborski et al., 2009). Behavioral activity is likely influenced by many factors including housing, season, and environment (Brzozowska et al., 2014). Locomotor behavior of dairy dams peripartum has been reported to differ by parity (Miedema et al., 2011a; Titler et al., 2015). Because of this, behavioral differences associated with parity may also impact the use of electronic activity monitoring, but to our knowledge, peripartum locomotor activity and factors affecting it have not been studied in beef cattle. We hypothesized that both multiparous and primiparous beef dams would have increased locomotor activity prior to calving, with primiparous dams being more active near parturition than multiparous dams. Our specific objectives were to determine peripartum changes in locomotion behavior in spring- and fall-calving cows, as well as differences between behavior of primiparous and multiparous dams in the final 3 d, 24 h, and 6 h prepartum. MATERIALS AND METHODS The University of Missouri Animal Care and Use Committee approved animal care and use in this study (Protocols 7936, 8952, and 9045). Gestating Animal Management All experiments. In 3 experiments, cows or heifers were housed in well-drained 18 × 61 m drylots with limestone-based surface (Figure 1, for all experiments) at the University of Missouri Beef Research and Teaching Farm and observed during the peripartum period. More females were observed in each experiment than were used in final data analysis to provide adequate numbers. Cows were fed ad libitum hay or haylage in round bale feeders located near the center of the pens on 9.1 × 9.1 m concrete pads to prevent mud accumulation around the feeders. Automatic waterers and feed bunks (9.8 m bunk length per pen) were located at the front of pens. Sheds were kept closed to animals unless individuals were moved under cover due to inclement weather during spring-calving (Exp. 1 and 2). For Exp. 1 and 2, ~20% of the pen area at the opposite end from the waterers and feed bunks were bedded with fescue straw to help mitigate calf cold stress. No bedding was used in Exp. 3, as it occurred during the fall season. Experiment 1. A 2-yr study was conducted in which multiparous, spring-calving, and Sim-Angus crossbred beef cows (year 1 n = 48, year 2 n = 56) were monitored during late gestation. The calving season was 79 d (beginning January 31, 2015) in year 1 and 44 d (beginning on February 2, 2016) in year 2. As part of a forage system study (Niederecker et al., 2018), cows were allocated to either strip-graze endophyte-infected stockpiled tall fescue or receive endophyte-infected tall fescue hay in drylots during late gestation. Drylots used for the hay forage system were the same as those used in calving (Figure 1). Cows were kept within the same forage system treatment groups, but those grazing stockpiled tall fescue were moved to additional drylots adjacent to cows receiving hay for observation 10.2 ± 8.6 (SD throughout) d (year 1) and 9.2 ± 4.3 d (year 2) pre-calving and fed harvested haylage. Dams were penned in groups of 8 to 10 animals based on previous treatments. Figure 1. View largeDownload slide Calving pen layout (18 × 61 m) for Exps 1, 2, and 3; hay feeders were on 9.1 × 9.1 m concrete pads; feed bunks were 9.8 m/pen. Figure 1. View largeDownload slide Calving pen layout (18 × 61 m) for Exps 1, 2, and 3; hay feeders were on 9.1 × 9.1 m concrete pads; feed bunks were 9.8 m/pen. Experiment 2. Primiparous (n = 23; dams pregnant with first calf) and multiparous (n = 65) spring-calving Sim-Angus crossbred beef females were moved to 6 drylots for observation 6.7 ± 3.0 d prior to calving. The period of calving monitoring was 15 d beginning on January 28, 2017. Dams were penned in groups of 12 to 15 animals by prior management group (parities 1 and 2 vs. parity ≥3). Animals were allowed ad libitum access to endophyte-infected tall fescue hay and supplemented with dried distillers grains with solubles at approximately 1700 hours daily. Experiment 3. Fall-calving Sim-Angus crossbred beef females (n = 45) were moved to 6 drylots for observation 17.5 ± 1.2 d prior to calving following individual feeding in smaller pens (4 cows per pen) for research using a Calan gate system. Dams were then assigned to calving pens in groups of 10 to 12 animals. The calving season was 36 d beginning on September 7, 2017. Animals were allowed ad libitum access to endophyte-infected tall fescue hay and supplemented with a soyhulls and dried distillers grains with solubles-based supplement at approximately 1700 hours daily. Calving Management and Data Collection One IceQube activity monitor (iceRobotics, Edinburgh, UK) was placed above the left hind fetlock of each pregnant female per manufacturer instructions. IceQube activity monitors were placed on dams 44.3 ± 21.7 d (year 1) and 41.0 ± 1.7 d (year 2) prior to calving in Exp. 1, 6.7 ± 0.3 d prior to calving in Exp. 2, and 34.4 ± 0.9 d prior to calving in Exp. 3. These accelerometers allow for tri-axial movement detection, continuous data logging, and data summary into specific increments of time (Richeson et al., 2018). These locomotion devices have been validated for accuracy against video monitoring in late gestation dairy cows by other groups (Mattachini et al., 2013; Borchers et al., 2016). Calving pens were well-lit, and lights were kept on during all or most of the night (depending on night calving density and experiment) to allow for observation of calving. Personnel monitored cows and heifers for physical signs of labor by walking through pens at least once per hour from 0600 to 2400 hours in each study, with additional monitoring between 0000 and 0600 hours during heavy calving periods. Females were monitored continuously by on-site personnel from the time of visible evidence of stage II parturition (presence of amniotic membranes or calf feet), and actual time of birth was recorded for each calf (expulsion of entire calf, including all 4 legs). Minimal human interference occurred during parturition except to assess progress or assist as needed if there were concerns of dystocia. No animals with dystocia requiring assistance were included in the dataset. Data were excluded from cows that calved prior to IceQube placement, with unknown calving times, that calved outside of the observation period, or with IceQube malfunction, resulting in the numbers and animals described in Table 1. Data from an individual day were removed for females moved outside of their normal patterns during the final 72 h prepartum. Movement outside normal patterns constituted cows that left pens due to movement to a working facility for other data collection, rearranging animals among pens, or relocation to a covered area if calving during inclement weather (Exp. 1 or 2). Descriptive data are provided for animals included in analysis for Exp. 1, 2, and 3 in Table 1. Table 1. Description of beef females included in analysis for locomotor activity during the 72 h prior to calving in 3 experiments1 Exp. 1 Exp. 2 Exp. 3 Variable Year 1 Year 2 Primiparous Multiparous n 34 27 13 21 33 Parity 4.5 ± 2.5 4.8 ± 2.3 1 ± 0 4.7 ± 3.1 3.4 ± 1.2 Prepartum body weight, kg 682 ± 74 671 ± 75 552 ± 44 670 ± 63 708 ± 89 Prepartum body condition score2 5.3 ± 0.5 5.3 ± 0.5 4.9 ± 0.5 5.4 ± 0.5 5.5 ± 0.7 Gestation length3, d 284 ± 2.7 279 ± 2.7 277 ± 2.0 276 ± 2.5 284 ± 3.1 Average calving date February 21, 2015 February 16, 2016 February 5, 2017 February 5, 2017 September 19, 2017 Exp. 1 Exp. 2 Exp. 3 Variable Year 1 Year 2 Primiparous Multiparous n 34 27 13 21 33 Parity 4.5 ± 2.5 4.8 ± 2.3 1 ± 0 4.7 ± 3.1 3.4 ± 1.2 Prepartum body weight, kg 682 ± 74 671 ± 75 552 ± 44 670 ± 63 708 ± 89 Prepartum body condition score2 5.3 ± 0.5 5.3 ± 0.5 4.9 ± 0.5 5.4 ± 0.5 5.5 ± 0.7 Gestation length3, d 284 ± 2.7 279 ± 2.7 277 ± 2.0 276 ± 2.5 284 ± 3.1 Average calving date February 21, 2015 February 16, 2016 February 5, 2017 February 5, 2017 September 19, 2017 1Least square means ± SD are presented for each variable. 2BCS evaluated on 1 to 9 scale (1 = emaciated, 9 = obese). 3Calculated only for cows that conceived from artificial insemination. View Large Table 1. Description of beef females included in analysis for locomotor activity during the 72 h prior to calving in 3 experiments1 Exp. 1 Exp. 2 Exp. 3 Variable Year 1 Year 2 Primiparous Multiparous n 34 27 13 21 33 Parity 4.5 ± 2.5 4.8 ± 2.3 1 ± 0 4.7 ± 3.1 3.4 ± 1.2 Prepartum body weight, kg 682 ± 74 671 ± 75 552 ± 44 670 ± 63 708 ± 89 Prepartum body condition score2 5.3 ± 0.5 5.3 ± 0.5 4.9 ± 0.5 5.4 ± 0.5 5.5 ± 0.7 Gestation length3, d 284 ± 2.7 279 ± 2.7 277 ± 2.0 276 ± 2.5 284 ± 3.1 Average calving date February 21, 2015 February 16, 2016 February 5, 2017 February 5, 2017 September 19, 2017 Exp. 1 Exp. 2 Exp. 3 Variable Year 1 Year 2 Primiparous Multiparous n 34 27 13 21 33 Parity 4.5 ± 2.5 4.8 ± 2.3 1 ± 0 4.7 ± 3.1 3.4 ± 1.2 Prepartum body weight, kg 682 ± 74 671 ± 75 552 ± 44 670 ± 63 708 ± 89 Prepartum body condition score2 5.3 ± 0.5 5.3 ± 0.5 4.9 ± 0.5 5.4 ± 0.5 5.5 ± 0.7 Gestation length3, d 284 ± 2.7 279 ± 2.7 277 ± 2.0 276 ± 2.5 284 ± 3.1 Average calving date February 21, 2015 February 16, 2016 February 5, 2017 February 5, 2017 September 19, 2017 1Least square means ± SD are presented for each variable. 2BCS evaluated on 1 to 9 scale (1 = emaciated, 9 = obese). 3Calculated only for cows that conceived from artificial insemination. View Large IceQube activity monitors were removed ≥2 d postpartum. IceManager 2012 software (iceRobotics, Edinburgh, UK) was used to obtain and sum each cow and heifer’s motion index, standing time, lying time, step count, and number of lying bouts per hour, then exported to an Excel spreadsheet (Microsoft Corp., Redmond, WA). Step count indicates the number of steps taken by the leg on which the IceQube is attached. Standing and lying time within an hour sum up to 60 min and therefore have an inverse relationship. Lying bouts are any periods in which an animal laid down and stood back up, and vary in duration length. Motion index was provided by the IceManager software using a proprietary algorithm. Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. For example, if a cow calved at 1250 hours, 1200 to 1300 hours was set as 0 h; whereas if a cow calved at 1210 hours, 1100 to 1200 hours was set as 0 h. This was done to prevent postpartum behavior from being most of the time occurring during 0 h (calving), which would likely influence results. All other time periods prepartum were based on this point. Statistical Analysis Data were analyzed using repeated measures ANOVA [MIXED procedure of SAS 9.4 (SAS Inst. Inc., Cary, NC)] in 3 separate analyses for each experiment: by day during the final 72 h prepartum (days −3, −2, and −1), by 6-h period during the final 24 h prepartum (hours −23 to −18, −17 to −12, −11 to −6, and −5 to 0), and by hour during the final 6 h prepartum (hours −5, −4, −3, −2, −1, and 0). For Exp. 1, the fixed effects of time prepartum and year were included in the model. The fixed effects of time prepartum, parity, and their interaction were included in the model for Exp. 2. For Exp. 3, only the fixed effect of time prepartum was included in the model. Animal was the experimental unit for all experiments. Time prepartum was considered a repeated effect for all models, and the best-fit covariance structure (chosen from compound symmetry, heterogeneous compound symmetry, autoregressive, heterogeneous autoregressive, and unstructured) was used for all analyses. Least squares means were separated using least significant difference and considered significant when P ≤ 0.05. Main effects and interactions were reported when P ≤ 0.05 with tendencies considered when P ≤ 0.10 and P > 0.05. In the absence of interactions, main effects of time and parity are reported for Exp. 2. RESULTS Final 72 h Prepartum Day affected (P < 0.001) motion index, standing time, lying time, step count, and number of lying bouts for spring-calving multiparous dams in Exp. 1 (Table 2). Motion index was greater (P < 0.001) on day −1 than days −2 and −3. Cows spent greater (P ≤ 0.001) time standing on day −1 than day −2 or −3. Because standing and lying are inversely related, dams spent less (P ≤ 0.001) time lying on day −1 than day −2 or −3. Step count was also greater (P ≤ 0.001) on day −1 when compared with day −2 or −3. Additionally, cows had a greater (P < 0.001) number of lying bouts on day −1 than on days −2 and −3. There were no differences (P ≥ 0.1) between days −3 and −2 for any measures of activity. There was an effect of year (P = 0.03) for step count, but year did not affect any other parameters measured (P ≥ 0.39). Table 2. Locomotor activity during the 72 h prior to calving in multiparous spring-calving (Exp. 1) and fall-calving (Exp. 3) beef cows Day1 P-value Item −3 −2 −1 SEM Day Year2 Spring-calving (Exp. 1)3  Motion index4 4,040b 3,871b 7,049a 417 <0.001 0.39  Standing time, min 736.5b 752.5b 910.6a 16.8 <0.001 0.83  Lying time, min 703.5a 687.5a 529.4b 16.8 <0.001 0.83  Step count 980b 949b 1,725a 71.1 <0.001 0.03  Lying bouts5 10.4b 10.3b 22.0a 0.70 <0.001 0.82 Fall-calving (Exp. 3)6  Motion index 6,478b 6,521b 9,740a 536 <0.001 —  Standing time, min 707.1b 721.9b 907.3a 31.3 <0.001 —  Lying time, min 732.9a 718.1a 532.7b 31.3 <0.001 —  Step count 1,496b 1,499b 2,408a 145 <0.001 —  Lying bouts 10.9b 10.7b 17.4a 0.89 <0.001 — Day1 P-value Item −3 −2 −1 SEM Day Year2 Spring-calving (Exp. 1)3  Motion index4 4,040b 3,871b 7,049a 417 <0.001 0.39  Standing time, min 736.5b 752.5b 910.6a 16.8 <0.001 0.83  Lying time, min 703.5a 687.5a 529.4b 16.8 <0.001 0.83  Step count 980b 949b 1,725a 71.1 <0.001 0.03  Lying bouts5 10.4b 10.3b 22.0a 0.70 <0.001 0.82 Fall-calving (Exp. 3)6  Motion index 6,478b 6,521b 9,740a 536 <0.001 —  Standing time, min 707.1b 721.9b 907.3a 31.3 <0.001 —  Lying time, min 732.9a 718.1a 532.7b 31.3 <0.001 —  Step count 1,496b 1,499b 2,408a 145 <0.001 —  Lying bouts 10.9b 10.7b 17.4a 0.89 <0.001 — a,bWithin an item, main effect means differ (P ≤ 0.05). 1Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. 2Because data were collected over the course of 2 calving seasons (2015 and 2016), year was included in the statistical model for that experiment only. 3n = 34 and 27 in years 1 and 2, respectively. 4Proprietary index in which the value is arbitrary. 5Any length of time in which dam laid down and then returned to standing. 6n = 33. View Large Table 2. Locomotor activity during the 72 h prior to calving in multiparous spring-calving (Exp. 1) and fall-calving (Exp. 3) beef cows Day1 P-value Item −3 −2 −1 SEM Day Year2 Spring-calving (Exp. 1)3  Motion index4 4,040b 3,871b 7,049a 417 <0.001 0.39  Standing time, min 736.5b 752.5b 910.6a 16.8 <0.001 0.83  Lying time, min 703.5a 687.5a 529.4b 16.8 <0.001 0.83  Step count 980b 949b 1,725a 71.1 <0.001 0.03  Lying bouts5 10.4b 10.3b 22.0a 0.70 <0.001 0.82 Fall-calving (Exp. 3)6  Motion index 6,478b 6,521b 9,740a 536 <0.001 —  Standing time, min 707.1b 721.9b 907.3a 31.3 <0.001 —  Lying time, min 732.9a 718.1a 532.7b 31.3 <0.001 —  Step count 1,496b 1,499b 2,408a 145 <0.001 —  Lying bouts 10.9b 10.7b 17.4a 0.89 <0.001 — Day1 P-value Item −3 −2 −1 SEM Day Year2 Spring-calving (Exp. 1)3  Motion index4 4,040b 3,871b 7,049a 417 <0.001 0.39  Standing time, min 736.5b 752.5b 910.6a 16.8 <0.001 0.83  Lying time, min 703.5a 687.5a 529.4b 16.8 <0.001 0.83  Step count 980b 949b 1,725a 71.1 <0.001 0.03  Lying bouts5 10.4b 10.3b 22.0a 0.70 <0.001 0.82 Fall-calving (Exp. 3)6  Motion index 6,478b 6,521b 9,740a 536 <0.001 —  Standing time, min 707.1b 721.9b 907.3a 31.3 <0.001 —  Lying time, min 732.9a 718.1a 532.7b 31.3 <0.001 —  Step count 1,496b 1,499b 2,408a 145 <0.001 —  Lying bouts 10.9b 10.7b 17.4a 0.89 <0.001 — a,bWithin an item, main effect means differ (P ≤ 0.05). 1Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. 2Because data were collected over the course of 2 calving seasons (2015 and 2016), year was included in the statistical model for that experiment only. 3n = 34 and 27 in years 1 and 2, respectively. 4Proprietary index in which the value is arbitrary. 5Any length of time in which dam laid down and then returned to standing. 6n = 33. View Large Motion index, standing time, lying time, step count, and number of lying bouts were also affected (P < 0.001) by day in fall-calving multiparous dams in Exp. 3 (Table 2). These cows had the same changes in activity as Exp. 1 for all parameters (P < 0.001), with the day prior to calving being different than days −2 and −3. In Exp. 2, the interaction of day × parity affected (P = 0.002) number of lying bouts in spring-calving females (Table 3). Both primiparous and multiparous dams had a greater (P ≤ 0.001) number of lying bouts on day −1 compared with days −2 and −-3, but primiparous dams had more (P = 0.02) lying bouts than multiparous dams on day −1. Despite this, parity and the parity × day interaction did not affect (P > 0.17) all other parameters. There was an effect of day (P <0.001) for motion index, standing and lying time, and step count. Motion index, standing time, and step count were greater (P < 0.001), and lying time less (P < 0.001), on day −1 than days −2 and −3, as observed in Exp. 1 and 3. Table 3. Locomotor activity during the 72 h prior to calving in primiparous and multiparous spring-calving beef cows (Exp. 2) Day1 Parity P-value Item −3 −2 −1 SEM 12 ≥23 SEM Day Parity Day × parity Motion index4 5,842b 5,727b 8,344a 664 6,949 6,326 742 <0.001 0.52 0.23 Standing time, min 784.1b 780.7b 908.5a 18.4 843.1 805.8 20.9 <0.001 0.17 0.81 Lying time, min 655.9a 659.3a 531.5b 18.4 596.9 634.2 20.9 <0.001 0.17 0.81 Step count 1,458b 1,434b 2,093a 151 1,720 1,603 171 <0.001 0.60 0.22 Lying bouts5 — — — — — — — <0.001 0.02 0.002  Primiparous 10.94y 9.54yz 24.62w 1.76 — — — — — —  Multiparous 8.49z 9.33yz 19.10x 1.38 — — — — — — Day1 Parity P-value Item −3 −2 −1 SEM 12 ≥23 SEM Day Parity Day × parity Motion index4 5,842b 5,727b 8,344a 664 6,949 6,326 742 <0.001 0.52 0.23 Standing time, min 784.1b 780.7b 908.5a 18.4 843.1 805.8 20.9 <0.001 0.17 0.81 Lying time, min 655.9a 659.3a 531.5b 18.4 596.9 634.2 20.9 <0.001 0.17 0.81 Step count 1,458b 1,434b 2,093a 151 1,720 1,603 171 <0.001 0.60 0.22 Lying bouts5 — — — — — — — <0.001 0.02 0.002  Primiparous 10.94y 9.54yz 24.62w 1.76 — — — — — —  Multiparous 8.49z 9.33yz 19.10x 1.38 — — — — — — a,bWithin an item, main effect means differ (P ≤ 0.05). x–zWithin an item, interactive means differ (P ≤ 0.05). 1Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. 2Primiparous n = 13. 3Multiparous n = 21. 4Proprietary index in which the value is arbitrary. 5Any length of time in which dam laid down and then returned to standing. View Large Table 3. Locomotor activity during the 72 h prior to calving in primiparous and multiparous spring-calving beef cows (Exp. 2) Day1 Parity P-value Item −3 −2 −1 SEM 12 ≥23 SEM Day Parity Day × parity Motion index4 5,842b 5,727b 8,344a 664 6,949 6,326 742 <0.001 0.52 0.23 Standing time, min 784.1b 780.7b 908.5a 18.4 843.1 805.8 20.9 <0.001 0.17 0.81 Lying time, min 655.9a 659.3a 531.5b 18.4 596.9 634.2 20.9 <0.001 0.17 0.81 Step count 1,458b 1,434b 2,093a 151 1,720 1,603 171 <0.001 0.60 0.22 Lying bouts5 — — — — — — — <0.001 0.02 0.002  Primiparous 10.94y 9.54yz 24.62w 1.76 — — — — — —  Multiparous 8.49z 9.33yz 19.10x 1.38 — — — — — — Day1 Parity P-value Item −3 −2 −1 SEM 12 ≥23 SEM Day Parity Day × parity Motion index4 5,842b 5,727b 8,344a 664 6,949 6,326 742 <0.001 0.52 0.23 Standing time, min 784.1b 780.7b 908.5a 18.4 843.1 805.8 20.9 <0.001 0.17 0.81 Lying time, min 655.9a 659.3a 531.5b 18.4 596.9 634.2 20.9 <0.001 0.17 0.81 Step count 1,458b 1,434b 2,093a 151 1,720 1,603 171 <0.001 0.60 0.22 Lying bouts5 — — — — — — — <0.001 0.02 0.002  Primiparous 10.94y 9.54yz 24.62w 1.76 — — — — — —  Multiparous 8.49z 9.33yz 19.10x 1.38 — — — — — — a,bWithin an item, main effect means differ (P ≤ 0.05). x–zWithin an item, interactive means differ (P ≤ 0.05). 1Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. 2Primiparous n = 13. 3Multiparous n = 21. 4Proprietary index in which the value is arbitrary. 5Any length of time in which dam laid down and then returned to standing. View Large Final 24 h Prepartum In spring-calving multiparous dams (Exp. 1), 6-h time period within the last 24 h prepartum affected all parameters (P ≤ 0.01; Table 4). Motion index and step count were greater (P < 0.001) during the last 6 h prepartum than all other periods (Table 4). Standing time was greater (P ≤ 0.01) during the −5 to 0 h and −23 to −18 h periods than during the −11 to −6 h period, with lying time being the inverse. Lying bout number during the final 24 h increased (P < 0.001) during the −17 to −12 h period and again during the final 6-h period. There was an effect of year (P ≤ 0.02) for all parameters except step count (P = 0.22) in this experiment. Table 4. Locomotor activity by 6-h period during the 24 h prior to calving in multiparous spring-calving (Exp. 1) and fall-calving (Exp. 3) beef cows Time period1, h P-value Item −23 to −18 −17 to −12 −11 to −6 −5 to 0 SEM Period Year2 Spring-calving (Exp. 1)3  Motion index4 1,359b 1,292b 1,460b 3,053a 179 <0.001 0.02  Standing time, min 239.1a 224.5ab 211.5b 240.7a 9.7 0.01 0.01  Lying time, min 120.9b 135.5ab 148.5a 119.4b 9.7 0.01 0.01  Step count 326.5b 312.5b 357.2b 725.3a 69.5 <0.001 0.22  Lying bouts5 2.06c 3.39b 3.88b 12.42a 0.69 <0.001 0.002 Fall-calving (Exp. 3)6  Motion index 2,083b 2,122b 1,946b 3,477a 268 <0.001 —  Standing time, min 222.3ab 231.6a 198.6b 250.9a 14.7 <0.001 —  Lying time, min 137.7ab 128.4b 161.4a 109.1b 14.7 <0.001 —  Step count 493.8b 512.3b 488.0b 890.2a 72.4 <0.001 —  Lying bouts 2.39c 2.56bc 3.17b 9.39a 0.65 <0.001 — Time period1, h P-value Item −23 to −18 −17 to −12 −11 to −6 −5 to 0 SEM Period Year2 Spring-calving (Exp. 1)3  Motion index4 1,359b 1,292b 1,460b 3,053a 179 <0.001 0.02  Standing time, min 239.1a 224.5ab 211.5b 240.7a 9.7 0.01 0.01  Lying time, min 120.9b 135.5ab 148.5a 119.4b 9.7 0.01 0.01  Step count 326.5b 312.5b 357.2b 725.3a 69.5 <0.001 0.22  Lying bouts5 2.06c 3.39b 3.88b 12.42a 0.69 <0.001 0.002 Fall-calving (Exp. 3)6  Motion index 2,083b 2,122b 1,946b 3,477a 268 <0.001 —  Standing time, min 222.3ab 231.6a 198.6b 250.9a 14.7 <0.001 —  Lying time, min 137.7ab 128.4b 161.4a 109.1b 14.7 <0.001 —  Step count 493.8b 512.3b 488.0b 890.2a 72.4 <0.001 —  Lying bouts 2.39c 2.56bc 3.17b 9.39a 0.65 <0.001 — a–cWithin an item, main effect means differ (P ≤ 0.05). 1Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. 2Because data were collected over the course of 2 calving seasons (2015 and 2016), year was included in the statistical model for that experiment only. 3n = 34 and 27 in years 1 and 2, respectively. 4Proprietary index in which the value is arbitrary. 5Any length of time in which dam laid down and then returned to standing. 6n = 33. View Large Table 4. Locomotor activity by 6-h period during the 24 h prior to calving in multiparous spring-calving (Exp. 1) and fall-calving (Exp. 3) beef cows Time period1, h P-value Item −23 to −18 −17 to −12 −11 to −6 −5 to 0 SEM Period Year2 Spring-calving (Exp. 1)3  Motion index4 1,359b 1,292b 1,460b 3,053a 179 <0.001 0.02  Standing time, min 239.1a 224.5ab 211.5b 240.7a 9.7 0.01 0.01  Lying time, min 120.9b 135.5ab 148.5a 119.4b 9.7 0.01 0.01  Step count 326.5b 312.5b 357.2b 725.3a 69.5 <0.001 0.22  Lying bouts5 2.06c 3.39b 3.88b 12.42a 0.69 <0.001 0.002 Fall-calving (Exp. 3)6  Motion index 2,083b 2,122b 1,946b 3,477a 268 <0.001 —  Standing time, min 222.3ab 231.6a 198.6b 250.9a 14.7 <0.001 —  Lying time, min 137.7ab 128.4b 161.4a 109.1b 14.7 <0.001 —  Step count 493.8b 512.3b 488.0b 890.2a 72.4 <0.001 —  Lying bouts 2.39c 2.56bc 3.17b 9.39a 0.65 <0.001 — Time period1, h P-value Item −23 to −18 −17 to −12 −11 to −6 −5 to 0 SEM Period Year2 Spring-calving (Exp. 1)3  Motion index4 1,359b 1,292b 1,460b 3,053a 179 <0.001 0.02  Standing time, min 239.1a 224.5ab 211.5b 240.7a 9.7 0.01 0.01  Lying time, min 120.9b 135.5ab 148.5a 119.4b 9.7 0.01 0.01  Step count 326.5b 312.5b 357.2b 725.3a 69.5 <0.001 0.22  Lying bouts5 2.06c 3.39b 3.88b 12.42a 0.69 <0.001 0.002 Fall-calving (Exp. 3)6  Motion index 2,083b 2,122b 1,946b 3,477a 268 <0.001 —  Standing time, min 222.3ab 231.6a 198.6b 250.9a 14.7 <0.001 —  Lying time, min 137.7ab 128.4b 161.4a 109.1b 14.7 <0.001 —  Step count 493.8b 512.3b 488.0b 890.2a 72.4 <0.001 —  Lying bouts 2.39c 2.56bc 3.17b 9.39a 0.65 <0.001 — a–cWithin an item, main effect means differ (P ≤ 0.05). 1Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. 2Because data were collected over the course of 2 calving seasons (2015 and 2016), year was included in the statistical model for that experiment only. 3n = 34 and 27 in years 1 and 2, respectively. 4Proprietary index in which the value is arbitrary. 5Any length of time in which dam laid down and then returned to standing. 6n = 33. View Large In fall-calving multiparous dams (Exp. 3), period within the final 24 h prepartum also affected (P ≤ 0.001) all parameters (Table 4). Again both motion index and step count were greater (P < 0.001) during the last 6 h prepartum compared with all other periods. Standing time was greater (P < 0.001) from −5 to 0 h than −11 to −6 h time period, but −5 to 0 h was not different (P ≥ 0.09) from −23 to −18 and −17 to −12 h periods. The same was true for lying time, with cows lying less (P < 0.001) from −5 to 0 h than −11 to −6 h. Lying bout number was greatest (P < 0.001) during the −5 to 0 h time period. Additionally, lying bout number was greater (P = 0.02) during −11 to −6 h compared with −23 to −18 h; however, −17 to −12 h was not different (P ≤ 0.06) from −11 to −6 h or −23 to −18 h periods. Parity and the interaction of period x parity did not affect (P ≥ 0.19) motion index, standing time, lying time, or step count when analyzed by 6-h periods during the final 24 h of Exp. 2 (Table 5). There was an interaction of period × parity for lying bouts (P = 0.03), where primiparous females had a greater (P < 0.001) number of lying bouts than multiparous females during the −11 to −6 h period. Lying bouts increased (P < 0.001) during the last 6-h period when compared with the previous 18 h for both primiparous and multiparous dams. Additionally, multiparous dams had more (P = 0.03) lying bouts from −17 to −12 h than −23 to −18 h and primiparous dams had more (P = 0.002) lying bouts from −11 to −6 h than −17 to −12 h. There was an effect of period (P ≤ 0.01) on all other parameters. Motion index and step count were greater (P ≤ 0.001) during the final 6 h prepartum than all previous 6-h periods. Females spent more (P ≤ 0.02) time standing from −5 to 0 h prepartum than the 2 preceding 6-h periods. Time spent standing was also decreased (P = 0.04) during the −11 to −6 h period when compared with −23 to −18 h. Table 5. Locomotor activity by 6-h period during the 24 h prior to calving in primiparous and multiparous spring-calving beef cows (Exp. 2) Time period1, h Parity P-value Item −23 to −18 −17 to −12 −11 to −6 −5 to 0 SEM 12 ≥23 SEM Period Parity Period × parity Motion index4 1,699b 1,493b 1,368b 3,783a 368 2,187 1,985 194 <0.001 0.42 0.37 Standing time, min 239.4ab 211.8bc 202.7c 254.6a 11.8 231.5 222.8 7.3 0.01 0.35 0.19 Lying time, min 120.6bc 148.2ab 157.3a 105.4c 11.8 128.5 137.2 7.3 0.01 0.35 0.19 Step count 432.2b 375.8b 354.3b 930.4a 90.7 540.4 505.9 45.5 <0.001 0.56 0.26 Lying bouts5 — — — — — — — — <0.001 0.02 0.03  Primiparous 2.31yz 2.77y 4.46x 15.08w 1.51 — — — — — —  Multiparous 1.86z 2.57y 2.33yz 12.33w 1.18 — — — — — — Time period1, h Parity P-value Item −23 to −18 −17 to −12 −11 to −6 −5 to 0 SEM 12 ≥23 SEM Period Parity Period × parity Motion index4 1,699b 1,493b 1,368b 3,783a 368 2,187 1,985 194 <0.001 0.42 0.37 Standing time, min 239.4ab 211.8bc 202.7c 254.6a 11.8 231.5 222.8 7.3 0.01 0.35 0.19 Lying time, min 120.6bc 148.2ab 157.3a 105.4c 11.8 128.5 137.2 7.3 0.01 0.35 0.19 Step count 432.2b 375.8b 354.3b 930.4a 90.7 540.4 505.9 45.5 <0.001 0.56 0.26 Lying bouts5 — — — — — — — — <0.001 0.02 0.03  Primiparous 2.31yz 2.77y 4.46x 15.08w 1.51 — — — — — —  Multiparous 1.86z 2.57y 2.33yz 12.33w 1.18 — — — — — — a–cWithin an item, main effect means differ (P ≤ 0.05). w–zWithin an item, interactive means differ (P ≤ 0.05). 1Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. 2Primiparous, n = 13. 3Multiparous, n = 21. 4Proprietary index in which the value is arbitrary. 5Any length of time in which dam laid down and then returned to standing. View Large Table 5. Locomotor activity by 6-h period during the 24 h prior to calving in primiparous and multiparous spring-calving beef cows (Exp. 2) Time period1, h Parity P-value Item −23 to −18 −17 to −12 −11 to −6 −5 to 0 SEM 12 ≥23 SEM Period Parity Period × parity Motion index4 1,699b 1,493b 1,368b 3,783a 368 2,187 1,985 194 <0.001 0.42 0.37 Standing time, min 239.4ab 211.8bc 202.7c 254.6a 11.8 231.5 222.8 7.3 0.01 0.35 0.19 Lying time, min 120.6bc 148.2ab 157.3a 105.4c 11.8 128.5 137.2 7.3 0.01 0.35 0.19 Step count 432.2b 375.8b 354.3b 930.4a 90.7 540.4 505.9 45.5 <0.001 0.56 0.26 Lying bouts5 — — — — — — — — <0.001 0.02 0.03  Primiparous 2.31yz 2.77y 4.46x 15.08w 1.51 — — — — — —  Multiparous 1.86z 2.57y 2.33yz 12.33w 1.18 — — — — — — Time period1, h Parity P-value Item −23 to −18 −17 to −12 −11 to −6 −5 to 0 SEM 12 ≥23 SEM Period Parity Period × parity Motion index4 1,699b 1,493b 1,368b 3,783a 368 2,187 1,985 194 <0.001 0.42 0.37 Standing time, min 239.4ab 211.8bc 202.7c 254.6a 11.8 231.5 222.8 7.3 0.01 0.35 0.19 Lying time, min 120.6bc 148.2ab 157.3a 105.4c 11.8 128.5 137.2 7.3 0.01 0.35 0.19 Step count 432.2b 375.8b 354.3b 930.4a 90.7 540.4 505.9 45.5 <0.001 0.56 0.26 Lying bouts5 — — — — — — — — <0.001 0.02 0.03  Primiparous 2.31yz 2.77y 4.46x 15.08w 1.51 — — — — — —  Multiparous 1.86z 2.57y 2.33yz 12.33w 1.18 — — — — — — a–cWithin an item, main effect means differ (P ≤ 0.05). w–zWithin an item, interactive means differ (P ≤ 0.05). 1Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. 2Primiparous, n = 13. 3Multiparous, n = 21. 4Proprietary index in which the value is arbitrary. 5Any length of time in which dam laid down and then returned to standing. View Large Final 6 h Prepartum In spring-calving multiparous dams (Exp. 1), there was an effect of hour (P ≤ 0.001) for motion index, step count, and lying bouts during the final 6 h prepartum (Table 6). Motion index increased (P = 0.002) at −3 h and again at −1 h prior to calving. Step count was greater (P = 0.001) during the last 4 h, with −1 h having a greater (P ≤ 0.05) number of steps taken than −3 and 0 h relative to calving. Lying bouts increased (P ≤ 0.001) from −5 h to −3 h relative to calving. The number of lying bouts more than doubled (P < 0.001) from −2 to −1 h and increased (P = 0.001) during the hour in which calving occurred. There tended to be an effect of hour (P = 0.09) for standing and lying time, where cows spent greater (P = 0.02) time standing at −2 h compared with −5, −4, −3, and 0 h relative to calving. There was an effect of year (P ≤ 0.009) for all parameters except lying bouts (P = 0.57). In fall-calving multiparous dams (Exp. 3), hour affected (P ≤ 0.02) motion index, step count, and lying bouts (Table 6). Motion index increased (P = 0.001) from −5 to −4 h, and cows had greater (P ≤ 0.03) motion index during 0 and −1 h than −4 and −5 h. Step count was greater (P = 0.02) from −3 to 0 h than at −5 h relative to calving, with no difference (P ≥ 0.06) among the last 4 h pre-calving. Lying bout number for fall-calving dams was greater (P < 0.001) at −1 and 0 h relative to calving than other hours, and more than doubled (P < 0.001) from −2 to −1 h prior to calving. Lying bout number also increased (P < 0.001) from −3 to −2 h. Unlike the spring-calving cows in Exp. 1, there was no effect of hour (P = 0.92) for standing and lying time in the fall-calving herd. Table 6. Locomotor activity by hour during the 6 h prior to calving in multiparous spring-calving (Exp. 1) and fall-calving (Exp. 3) beef cows Hour1 P-value Item −5 −4 −3 −2 −1 0 SEM Hour Year2 Spring-calving (Exp. 1)3  Motion index4 304.8c 356.1c 485.3b 561.3b 730.8a 637.6ab 96.5 <0.001 0.02  Standing time, min 36.38b 37.73b 40.78b 44.72a 42.67ab 39.06b 2.76 0.09 0.009  Lying time, min 23.62a 22.27a 19.22a 15.28b 17.33ab 20.94a 2.76 0.09 0.009  Step count 77.2c 90.0c 121.1b 139.6ab 173.5a 138.3b 23.3 0.001 0.004  Lying bouts5 0.77d 1.09cd 1.29c 1.42c 3.06b 4.81a 0.50 <0.001 0.57 Fall-calving (Exp. 3)6  Motion index 356.4c 500.9b 567.4ab 649.9ab 686.8a 715.8a 66.2 0.006 —  Standing time, min 39.53 43.39 41.46 42.99 42.11 41.41 3.45 0.92 —  Lying time, min 20.47 16.61 18.54 17.01 17.89 18.59 3.45 0.92 —  Step count 93.0c 127.6bc 145.2ab 170.0ab 182.0a 172.5ab 17.8 0.02 —  Lying bouts 0.48c 0.64c 0.76c 1.18b 2.73a 3.61a 0.49 <0.001 — Hour1 P-value Item −5 −4 −3 −2 −1 0 SEM Hour Year2 Spring-calving (Exp. 1)3  Motion index4 304.8c 356.1c 485.3b 561.3b 730.8a 637.6ab 96.5 <0.001 0.02  Standing time, min 36.38b 37.73b 40.78b 44.72a 42.67ab 39.06b 2.76 0.09 0.009  Lying time, min 23.62a 22.27a 19.22a 15.28b 17.33ab 20.94a 2.76 0.09 0.009  Step count 77.2c 90.0c 121.1b 139.6ab 173.5a 138.3b 23.3 0.001 0.004  Lying bouts5 0.77d 1.09cd 1.29c 1.42c 3.06b 4.81a 0.50 <0.001 0.57 Fall-calving (Exp. 3)6  Motion index 356.4c 500.9b 567.4ab 649.9ab 686.8a 715.8a 66.2 0.006 —  Standing time, min 39.53 43.39 41.46 42.99 42.11 41.41 3.45 0.92 —  Lying time, min 20.47 16.61 18.54 17.01 17.89 18.59 3.45 0.92 —  Step count 93.0c 127.6bc 145.2ab 170.0ab 182.0a 172.5ab 17.8 0.02 —  Lying bouts 0.48c 0.64c 0.76c 1.18b 2.73a 3.61a 0.49 <0.001 — a–dWithin an item, main effect means differ (P ≤ 0.05). 1Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. 2Because data were collected over the course of 2 calving seasons (2015 and 2016), year was included in the statistical model for that experiment only. 3n = 34 and 27 in years 1 and 2, respectively. 4Proprietary index in which the value is arbitrary. 5Any length of time in which dam laid down and then returned to standing. 6n = 33. View Large Table 6. Locomotor activity by hour during the 6 h prior to calving in multiparous spring-calving (Exp. 1) and fall-calving (Exp. 3) beef cows Hour1 P-value Item −5 −4 −3 −2 −1 0 SEM Hour Year2 Spring-calving (Exp. 1)3  Motion index4 304.8c 356.1c 485.3b 561.3b 730.8a 637.6ab 96.5 <0.001 0.02  Standing time, min 36.38b 37.73b 40.78b 44.72a 42.67ab 39.06b 2.76 0.09 0.009  Lying time, min 23.62a 22.27a 19.22a 15.28b 17.33ab 20.94a 2.76 0.09 0.009  Step count 77.2c 90.0c 121.1b 139.6ab 173.5a 138.3b 23.3 0.001 0.004  Lying bouts5 0.77d 1.09cd 1.29c 1.42c 3.06b 4.81a 0.50 <0.001 0.57 Fall-calving (Exp. 3)6  Motion index 356.4c 500.9b 567.4ab 649.9ab 686.8a 715.8a 66.2 0.006 —  Standing time, min 39.53 43.39 41.46 42.99 42.11 41.41 3.45 0.92 —  Lying time, min 20.47 16.61 18.54 17.01 17.89 18.59 3.45 0.92 —  Step count 93.0c 127.6bc 145.2ab 170.0ab 182.0a 172.5ab 17.8 0.02 —  Lying bouts 0.48c 0.64c 0.76c 1.18b 2.73a 3.61a 0.49 <0.001 — Hour1 P-value Item −5 −4 −3 −2 −1 0 SEM Hour Year2 Spring-calving (Exp. 1)3  Motion index4 304.8c 356.1c 485.3b 561.3b 730.8a 637.6ab 96.5 <0.001 0.02  Standing time, min 36.38b 37.73b 40.78b 44.72a 42.67ab 39.06b 2.76 0.09 0.009  Lying time, min 23.62a 22.27a 19.22a 15.28b 17.33ab 20.94a 2.76 0.09 0.009  Step count 77.2c 90.0c 121.1b 139.6ab 173.5a 138.3b 23.3 0.001 0.004  Lying bouts5 0.77d 1.09cd 1.29c 1.42c 3.06b 4.81a 0.50 <0.001 0.57 Fall-calving (Exp. 3)6  Motion index 356.4c 500.9b 567.4ab 649.9ab 686.8a 715.8a 66.2 0.006 —  Standing time, min 39.53 43.39 41.46 42.99 42.11 41.41 3.45 0.92 —  Lying time, min 20.47 16.61 18.54 17.01 17.89 18.59 3.45 0.92 —  Step count 93.0c 127.6bc 145.2ab 170.0ab 182.0a 172.5ab 17.8 0.02 —  Lying bouts 0.48c 0.64c 0.76c 1.18b 2.73a 3.61a 0.49 <0.001 — a–dWithin an item, main effect means differ (P ≤ 0.05). 1Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. 2Because data were collected over the course of 2 calving seasons (2015 and 2016), year was included in the statistical model for that experiment only. 3n = 34 and 27 in years 1 and 2, respectively. 4Proprietary index in which the value is arbitrary. 5Any length of time in which dam laid down and then returned to standing. 6n = 33. View Large There was no interaction of day × parity (P ≥ 0.61) or main effect of parity (P ≥ 0.29) for any variable during the final 6 h prior to parturition in Exp. 2 (Table 7). There was an effect of hour (P < 0.001) on motion index, step count, and lying bouts. Motion index was greater (P ≤ 0.001) during −2, −1, and 0 h prepartum compared with −5, −4, and −3 h, with no difference (P ≥ 0.33) among the last 3 h prior to calving. Step count was greater (P ≤ 0.001) during −1 and −2 h than −3, −4, and −5 h, with no difference (P ≥ 0.06) among the last 3 h prior to calving. Lying bouts increased (P ≤ 0.001) from −4 to −3 h, −3 to −2 h, and more than doubled from −2 to −1 h. Lying bout number during −1 h was not different (P = 0.50) from 0 h relative to calving. There tended to be an effect of hour (P = 0.08) for standing and lying time, where greater (P = 0.02) time was spent lying at 0 h prior to calving when compared with −2 and −4 h. There were no differences (P ≥ 0.06) among −5, −4, −3, and −2 h or between −1 and 0 h relative to calving for standing or lying time. Table 7. Locomotor activity by hour during the 6 h prior to calving in primiparous and multiparous spring-calving beef cows (Exp. 2) Hour1 Parity P-value Item −5 −4 −3 −2 −1 0 SEM 12 ≥23 SEM Hour Parity Hour x Parity Motion index4 408.4b 400.4b 521.6b 799.6a 864.8a 788.4a 116.1 623.8 637.3 96.4 <0.001 0.91 0.67 Standing time, min 44.38abc 45.78ab 42.60abc 46.06a 38.56bc 37.19c 2.77 41.99 42.87 2.42 0.08 0.78 0.61 Lying time, min 15.62abc 14.22bc 17.40abc 13.94c 21.44ab 22.81a 2.77 18.01 17.13 2.42 0.08 0.78 0.61 Step count 108.5c 103.8c 135.7bc 205.6a 206.1a 170.8ab 21.1 150.3 159.9 22.8 <0.001 0.74 0.94 Lying bouts5 0.49d 0.72d 1.24c 2.17b 4.75a 4.34a 0.59 2.51 2.06 0.33 <0.001 0.29 0.67 Hour1 Parity P-value Item −5 −4 −3 −2 −1 0 SEM 12 ≥23 SEM Hour Parity Hour x Parity Motion index4 408.4b 400.4b 521.6b 799.6a 864.8a 788.4a 116.1 623.8 637.3 96.4 <0.001 0.91 0.67 Standing time, min 44.38abc 45.78ab 42.60abc 46.06a 38.56bc 37.19c 2.77 41.99 42.87 2.42 0.08 0.78 0.61 Lying time, min 15.62abc 14.22bc 17.40abc 13.94c 21.44ab 22.81a 2.77 18.01 17.13 2.42 0.08 0.78 0.61 Step count 108.5c 103.8c 135.7bc 205.6a 206.1a 170.8ab 21.1 150.3 159.9 22.8 <0.001 0.74 0.94 Lying bouts5 0.49d 0.72d 1.24c 2.17b 4.75a 4.34a 0.59 2.51 2.06 0.33 <0.001 0.29 0.67 a–dWithin an item, main effect means differ (P ≤ 0.05). 1Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. 2Primiparous, n = 13. 3Multiparous, n = 21. 4Proprietary index in which the value is arbitrary. 5Any length of time in which dam laid down and then returned to standing. View Large Table 7. Locomotor activity by hour during the 6 h prior to calving in primiparous and multiparous spring-calving beef cows (Exp. 2) Hour1 Parity P-value Item −5 −4 −3 −2 −1 0 SEM 12 ≥23 SEM Hour Parity Hour x Parity Motion index4 408.4b 400.4b 521.6b 799.6a 864.8a 788.4a 116.1 623.8 637.3 96.4 <0.001 0.91 0.67 Standing time, min 44.38abc 45.78ab 42.60abc 46.06a 38.56bc 37.19c 2.77 41.99 42.87 2.42 0.08 0.78 0.61 Lying time, min 15.62abc 14.22bc 17.40abc 13.94c 21.44ab 22.81a 2.77 18.01 17.13 2.42 0.08 0.78 0.61 Step count 108.5c 103.8c 135.7bc 205.6a 206.1a 170.8ab 21.1 150.3 159.9 22.8 <0.001 0.74 0.94 Lying bouts5 0.49d 0.72d 1.24c 2.17b 4.75a 4.34a 0.59 2.51 2.06 0.33 <0.001 0.29 0.67 Hour1 Parity P-value Item −5 −4 −3 −2 −1 0 SEM 12 ≥23 SEM Hour Parity Hour x Parity Motion index4 408.4b 400.4b 521.6b 799.6a 864.8a 788.4a 116.1 623.8 637.3 96.4 <0.001 0.91 0.67 Standing time, min 44.38abc 45.78ab 42.60abc 46.06a 38.56bc 37.19c 2.77 41.99 42.87 2.42 0.08 0.78 0.61 Lying time, min 15.62abc 14.22bc 17.40abc 13.94c 21.44ab 22.81a 2.77 18.01 17.13 2.42 0.08 0.78 0.61 Step count 108.5c 103.8c 135.7bc 205.6a 206.1a 170.8ab 21.1 150.3 159.9 22.8 <0.001 0.74 0.94 Lying bouts5 0.49d 0.72d 1.24c 2.17b 4.75a 4.34a 0.59 2.51 2.06 0.33 <0.001 0.29 0.67 a–dWithin an item, main effect means differ (P ≤ 0.05). 1Hour 0 was defined as the hour increment nearest to parturition in which the majority of time represented prepartum behavior. 2Primiparous, n = 13. 3Multiparous, n = 21. 4Proprietary index in which the value is arbitrary. 5Any length of time in which dam laid down and then returned to standing. View Large DISCUSSION An understanding of maternal locomotor activity changes as beef females near parturition may enable the use of quantitative behavioral data to predict time of calving, rather than qualitative behavioral monitoring alone. Remote detection of calving in a herd setting through prediction technology could reduce calf mortality caused by dystocia or other calving-related problems (Saint-Dizier and Chastant-Maillard, 2015). While previous research has indicated that locomotion behavioral changes are useful for calving prediction in dairy cattle, this is the first study to examine this in beef cattle to our knowledge. Overall, the results of this study suggest that accelerometers could be used to recognize early signs of parturition in beef cattle because changes in activity can be observed during the last 6 h prepartum in both multiparous and primiparous beef dams, as well as both spring-calving and fall-calving herds. The results of these studies indicate that future research using locomotion behavior in beef dams could be explored for the creation of calving prediction technologies, although other parameters that may influence maternal behavior detection such as enclosure size and calving environments should also be investigated. Locomotion Changes at Calving Standing and lying time. In all 3 experiments, females had greater standing time and less lying time on the day of calving compared with 2 and 3 d prior to calving. In dairy cattle, an 80% increase in standing activity during the 24 h prior to calving occurred during the peripartum period (defined as 24 h pre- and post-calving) compared with the preceding days, indicating that animals spent less time recumbent (Huzzey et al., 2005). This increase in standing time was also reported on the last day, during the 24 h prior and 12 h prior to calving (Huzzey et al., 2005; Titler et al., 2015; Borchers et al., 2017). Across our 3 experiments, more time was spent recumbent at −11 to −6 h period compared with the last 6-h period prepartum, regardless of parity or calving season. This agrees with previous data from dairy cows where less time was spent lying during the last 6- and 2-h periods (Miedema et al., 2011b; Borchers et al., 2017). When comparing the last 6 h pre-calving, there was no effect of hour for standing and lying time during the last 6 h in Exp. 3, but there tended to be a difference in Exp. 1 and 2 with an effect of year in Exp. 1. Both Exp. 1 and 2 utilized spring-calving females, suggesting that standing and lying time during the last 6 h may be sensitive to differences in ambient temperature, individual herd dynamics (as spring- and fall-calving herds are maintained separately on this operation), or dam temperament specific to each herd. These parameters may still be helpful in determining when cows are near calving, but additional research is necessary to investigate factors that affect them. Step count. Recent research with dairy cattle has shown that electronic data loggers did not detect a change in step count during the 2 wk prior to calving (Borchers et al., 2017); however, an earlier dairy study reported an increase in number of steps within the 12 h immediately prior to calving (Titler et al., 2015). Our data demonstrate an increase in number of steps taken during the final day when compared with days −3 and −2 prior to calving, with an increase in step number most specifically occurring during the last 3 h prepartum. This increase in steps could be due to females walking in search for a safe place to calve, seeking isolation from the herd, or pacing due to discomfort (Wehrend et al., 2006; Proudfoot et al., 2014). This increase may occur in small enough increments that previous research comparing individual days did not detect a difference; however, it may be useful in determining behavioral differences within a few hours of calving. Step count differences may also be detected in beef cattle more readily than in dairy cattle because of differences in temperament or housing. In the current study, beef dams were housed outdoors and in larger pens as opposed to indoor barns often used in the dairy industry. Lying bout number. In agreement with the current study, an increase in number of lying bouts within the final 24 h prepartum has been shown in dairy cattle (Miedema et al., 2011b; Titler et al., 2015; Ouellet et al., 2016; Borchers et al., 2017). Our study examined smaller units of time pre-calving than previous studies, allowing for the observation of a significant increase in lying bouts during the last 6 h. In all 3 experiments, average lying bout number more than doubled from −2 to −1 h prepartum. In spring-calving cows (Exp. 1), lying bouts continued to increase, and females had the greatest number of lying bouts during the hour of calving (0 h). Despite this, an increase in lying bouts between −1 and 0 h did not occur in Exp. 2 and 3. This difference between calving seasons may again be due to differences in dam temperament between herds or because fall-calving dams were mildly heat stressed and therefore less active after finding a comfortable place to calve. Although data were not compared statistically among current experiments, cows in Exp. 3 had the numerically fewest lying bouts during the hour of calving. In Exp. 1, the number of lying bouts did not have an effect of year, unlike other parameters measured during the last 6 h prior to calving. Because the lying bout increase −1 h prepartum in all 3 experiments was consistently around twice that of −2 h, this parameter appears to be the most reliable indicator of behavioral changes relative to calving in the current study. This increase in frequency of standing and lying back down is another example of restlessness associated with earlier signs of labor (Owens et al., 1985; Wehrend et al., 2006). Furthermore, this time period precalving would be a useful time for notification of impending calving to monitor for dystocia. With our method of analysis, time of beginning stage II parturition and time of fetal membrane rupture are included some time prior to the time of calf delivery, which may be within 0 h or earlier, depending on the duration of labor. Initial parturition-related recumbent behavior in cattle often begins as the calf enters the birth canal (Schuenemann et al., 2011). From an observational standpoint, many dams continue to be restless and frequently get up and down after stage II parturition begins, including after fetal membrane rupture. This discomfort may be expressed as the behavioral changes detected immediately prior to calving. Future studies where labor duration or specific events of parturition (stages, membrane rupture, presence of calf feet, head, etc.) are observed may be useful in further determining the utility of lying bouts in calving detection. Motion index. The motion index algorithm was developed by iceRobotics to indicate the total activity using a combination of all locomotor variables measured, where a greater numerical motion index indicates that cattle are more active. In the current study, the predominant increase in motion index occurred during the last few hours prior to calving, with this increase likely attributed to the activity of changing standing and lying status frequently as observed with the increase in number of lying bouts in addition to the number of steps taken. Motion index was summed by day in a dairy study with no difference during the last 4 d of gestation; however, overall activity was increased on the day prior to calving when compared with a week prior (Borchers et al., 2017). This indicates that motion indexing may be capable of predicting changes in behavior by combining multiple motion variables that are necessary in the establishment of calving prediction and detection technologies as used with research in dairy cattle (Titler et al., 2015; Borchers et al., 2017). Parity Effects In dairy cattle, primiparous females spent less time standing during the final 24 h pre-calving compared with multiparous cows (Titler et al., 2015). Additionally, primiparous dairy heifers spent more of the final 2 h prepartum recumbent compared with multiparous cows in another study (Miedema et al., 2011a). This was attributed to primiparous females spending more time in labor and having more contractions during parturition (Miedema et al., 2011a; Schuenemann et al., 2011). Other studies in dairy cattle reported differences between parities with increased restlessness a week prior to calving in primiparous dams, demonstrated by increased neck movement activity, pawing, and tail swishing (Wehrend et al., 2006; Borchers et al., 2017). While an effect of parity has been observed for multiple measurements of restlessness in dairy dams, few differences were noted for primiparous versus multiparous dams in Exp. 2. The restlessness attributed to primiparous dams may be evidenced in other variables for beef dams such as number of lying bouts, as parity affected this parameter in a time-dependent manner in the current study. On the day prior to calving and during the final 6-h period prior to calving, primiparous dams had an increased number of lying bouts when compared with multiparous dams. This suggests that primiparous dams are more restless and change their activity from lying to standing more frequently resulting in shorter, more frequent lying bouts. Parity did not affect lying bout number within the final 6 h precalving, indicating that this increase in restlessness of heifers occurs earlier during or prior to stage I of parturition. When analyzing the last 6 h prior to calving, lying bouts increased for both primiparous and multiparous dams, as the mean number of lying bouts more than doubled from hours −2 to −1 relative to calving. This suggests that lying bouts may be a good tool for prediction of calving in beef females regardless of parity. Results of the current study are similar to others in dairy cattle where the number of lying bouts increased on the day prior to calving with no effect of parity (Jensen, 2012; Borchers et al., 2017). Because these studies did not compare time periods within the 24 h prior to calving, it is important to note that the algorithm created for calving detection in dairy dams may not be applicable to beef cattle where parity had an effect on lying bouts when comparing by day. There was no difference in motion index between multiparous and primiparous dams, suggesting that the increase in activity occurs independent of parity. Previously an effect of parity was detected 4 h prior to calving in dairy cattle which was attributed to discomfort at calving (Borchers et al., 2017). Drivers of the difference in motion index between parities during these last 4 h cannot be interpreted due to the proprietary nature of the motion index algorithm. Alternate index development may be useful as further research is conducted and accounts for factors such as difference in beef and dairy cattle, environment, and parity. Other Factors That May Impact Calving Behavior While season effects were not statistically tested in the current study, similar trends were observed in behavior of both spring- and fall-calving multiparous dams (Exp. 1 and 2 vs. Exp. 3). There were numerical differences for most parameters between the 2 seasons which were likely due to differences in animal temperament, herd dynamics, weather, and feeding behavior. Despite this, data interpretations are largely similar between calving seasons. Because increases in locomotion behavior were observed at similar time points, we believe that prepartum behavioral changes in beef females can be detected by accelerometers regardless of calving season. The same spring cow herd was used in both years of Exp. 1, with some animal additions (3-yr olds entering the mature herd) and culling. Thus, year effects (means not shown) could be due to acclimation to increased human interaction, differences in neonatal sampling, weather differences between years, or other unknown factors. In all 3 experiments, dams were housed in the same calving pens. Future research is necessary to determine if behavior varies among calving environments, as a change in pen size, footing, and forage delivery may lead to a difference in herd dynamics and overall activity. There may also be behavioral differences based on location of shelters and with animals on pasture. This is supported by research in dairy cattle where conflicting behavioral changes were reported between studies where females were moved to maternity pens at the appearance of calf’s feet outside the vulva in one study and moved to maternity pens at least a day prior to calving in another study (Huzzey et al., 2005; Titler et al., 2015). In dairy cattle, the movement of animals at various stages of parturition has shown to have an effect on behavior of dams (Proudfoot et al., 2013). Because of this, animals that moved outside of normal patterns for the current study were not included to prevent any influence on overall calving behavioral data interpretation. CONCLUSION In conclusion, the current study demonstrates that locomotion behavior of beef cows and heifers increases during the 24 h prior to calving. This increased activity was most notable during the final 6 h prepartum, with many additional differences observed within the last 2 to 4 h before calving. In this study, the number of lying bouts had the most consistent change during this time period and was unaffected by parity. Lying bout changes occurred in both spring- and fall-calving seasons. This behavior may be a key tool in predicting nearness of calving in beef dams. More research in this area may allow for development of methods for detecting calving via machine learning combined with accelerometers with remote sensing capability. This technology likely presents an opportunity to optimize precision calving management, decreasing calf mortality or neonatal morbidity due to dystocia and other challenges associated with parturition. Additionally, this technology could allow for improved calving detection without human interference in research settings where the physiological or behavioral data or samples are collected in the peripartum period. Conflict of interest statement. None declared. ACKNOWLEDGMENTS The authors thank Brian Vander Ley (Great Plains Veterinary Education Center, Clay Center, NE); graduate students Katlyn Niederecker, Jill Larson, and Emma Stephenson; undergraduate assistants from the Meyer lab; and the University of Missouri Beef Research and Teaching Farm for assistance with this project. LITERATURE CITED Berger , P. J. , A. C. Cubas , K. J. Koehler , and M. H. Healey . 1992 . Factors affecting dystocia and early calf mortality in Angus cows and heifers . J. Anim. Sci . 70 : 1775 – 1786 . doi:10.2527/1992.7061775x Google Scholar Crossref Search ADS PubMed Borchers , M. R. , Y. M. Chang , K. L. Proudfoot , B. A. Wadsworth , A. E. Stone , and J. M. Bewley . 2017 . Machine-learning-based calving prediction from activity, lying, and ruminating behaviors in dairy cattle . J. Dairy Sci . 100 : 5664 – 5674 . doi: https://doi.org/10.3168/jds.2016-11526 Google Scholar Crossref Search ADS PubMed Borchers , M. R. , Y. M. Chang , I. C. Tsai , B. A. Wadsworth , and J. M. Bewley . 2016 . A validation of technologies monitoring dairy cow feeding, ruminating, and lying behaviors . J. Dairy Sci . 99 : 7458 – 7466 . doi: https://doi.org/10.3168/jds.2015-10843 Google Scholar Crossref Search ADS PubMed Brzozowska , A. , M. Łukaszewicz , G. Sender , D. Kolasińska , and J. Oprządek . 2014 . Locomotor activity of dairy cows in relation to season and lactation . Appl. Anim. Beh. Sci . 156 : 6 – 11 . doi: https://doi.org/10.1016/j.applanim.2014.04.009 Google Scholar Crossref Search ADS Dargatz , D. A. , G. A. Dewell , and R. G. Mortimer . 2004 . Calving and calving management of beef cows and heifers on cow-calf operations in the United States . Theriogenology 61 : 997 – 1007 . doi: https://doi.org/10.1016/S0093-691X(03)00145-6 Google Scholar Crossref Search ADS PubMed Finney , G. , A. Gordon , G. Scoley , and S. J. Morrison . 2018 . Validating the IceRobotics IceQube tri-axial accelerometer for measuring daily lying duration in dairy calves . Livest. Sci . 214 : 83 – 87 . doi: https://doi.org/10.1016/j.livsci.2018.05.014 Google Scholar Crossref Search ADS Fogsgaard , K. K. , T. W. Bennedsgaard , and M. S. Herskin . 2015 . Behavioral changes in freestall-housed dairy cows with naturally occurring clinical mastitis . J. Dairy Sci . 98 : 1730 – 1738 . doi: https://doi.org/10.3168/jds.2014–8347 Google Scholar Crossref Search ADS PubMed Huzzey , J. M. , M. A. von Keyserlingk , and D. M. Weary . 2005 . Changes in feeding, drinking, and standing behavior of dairy cows during the transition period . J. Dairy Sci . 88 : 2454 – 2461 . doi: https://doi.org/10.3168/jds.S0022-0302(05)72923-4 Google Scholar Crossref Search ADS PubMed Jensen , M. B . 2012 . Behaviour around the time of calving in dairy cows . Appl. Anim. Beh. Sci . 139 : 195 – 202 . doi: https://doi.org/10.1016/j.applanim.2012.04.002 Google Scholar Crossref Search ADS Mattachini , G. , E. Riva , C. Bisaglia , J. C. Pompe , and G. Provolo . 2013 . Methodology for quantifying the behavioral activity of dairy cows in freestall barns . J. Anim. Sci . 91 : 4899 – 4907 . doi: https://doi.org/10.2527/jas.2012-5554 Google Scholar Crossref Search ADS PubMed Miedema , H. M. , M. S. Cockram , C. M. Dwyer , and A. I. Macrae . 2011a . Behavioural predictors of the start of normal and dystocic calving in dairy cows and heifers . Appl. Anim. Behav. Sci . 132 : 14 – 19 . doi: https://doi.org/10.1016/j.applanim.2011.03.003 Google Scholar Crossref Search ADS Miedema , H. M. , M. S. Cockram , C. M. Dwyer , and A. I. Macrae . 2011b . Changes in the behaviour of dairy cows during the 24 h before normal calving compared with behaviour during late pregnancy . Appl. Anim. Beh. Sci . 131 : 8 – 14 . doi: https://doi.org/10.1016/j.applanim.2011.01.012 Google Scholar Crossref Search ADS Niederecker , K. N. , J. M. Larson , R. L. Kallenbach , and A. M. Meyer . 2018 . Effects of feeding stockpiled tall fescue versus tall fescue hay to late gestation beef cows: I. Cow performance, maternal metabolic status, and fetal growth . J. Anim. Sci . doi: https://doi.org/10.1093/jas/sky341 Nielsen , L. R. , A. R. Pedersen , M. S. Herskin , and L. Munksgaard . 2010 . Quantifying walking and standing behaviour of dairy cows using a moving average based on output from an accelerometer . Appl. Anim. Beh. Sci . 127 : 12 – 19 . doi: https://doi.org/10.1016/j.applanim.2010.08.004 Google Scholar Crossref Search ADS Ouellet , V. , E. Vasseur , W. Heuwieser , O. Burfeind , X. Maldague , and É. Charbonneau . 2016 . Evaluation of calving indicators measured by automated monitoring devices to predict the onset of calving in Holstein dairy cows . J. Dairy Sci . 99 : 1539 – 1548 . doi: https://doi.org/10.3168/jds.2015-10057 Google Scholar Crossref Search ADS PubMed Owens , J. L. , T. N. Edey , B. M. Bindon , and L. R. Piper . 1985 . Parturient behaviour and calf survival in a herd selected for twinning . Appl. Anim. Beh. Sci . 13 : 321 – 333 . doi: https://doi.org/10.1016/0168-1591(85)90012–7 Google Scholar Crossref Search ADS Pillen , J. L. , P. J. Pinedo , S. E. Ives , T. L. Covey , H. K. Naikare , and J. T. Richeson . 2016 . Alteration of activity variables relative to clinical diagnosis of bovine respiratory disease in newly received feedlot cattle . Bov. Pract . 50 : 1 – 8 . Proudfoot , K. L. , M. B. Jensen , P. M. Heegaard , and M. A. von Keyserlingk . 2013 . Effect of moving dairy cows at different stages of labor on behavior during parturition . J. Dairy Sci . 96 : 1638 – 1646 . doi: https://doi.org/10.3168/jds.2012-6000 Google Scholar Crossref Search ADS PubMed Proudfoot , K. L. , M. B. Jensen , D. M. Weary , and M. A. von Keyserlingk . 2014 . Dairy cows seek isolation at calving and when ill . J. Dairy Sci . 97 : 2731 – 2739 . doi: https://doi.org/10.3168/jds.2013-7274 Google Scholar Crossref Search ADS PubMed Richeson , J. , T. Lawrence , and B. White . 2018 . Using advanced technologies to quantify beef cattle behavior . Transl. Anim. Sci . 2:223–229. doi: https://doi.org/10.1093/tas/txy004 Saint-Dizier , M. , and S. Chastant-Maillard . 2015 . Methods and on-farm devices to predict calving time in cattle . Vet. J . 205 : 349 – 356 . doi: https://doi.org/10.1016/j.tvjl.2015.05.006 Google Scholar Crossref Search ADS PubMed Schuenemann , G. M. , I. Nieto , S. Bas , K. N. Galvão , and J. Workman . 2011 . Assessment of calving progress and reference times for obstetric intervention during dystocia in Holstein dairy cows . J. Dairy Sci . 94 : 5494 – 5501 . doi: https://doi.org/10.3168/jds.2011-4436 Google Scholar Crossref Search ADS PubMed Titler , M. , M. G. Maquivar , S. Bas , P. J. Rajala-Schultz , E. Gordon , K. McCullough , P. Federico , and G. M. Schuenemann . 2015 . Prediction of parturition in Holstein dairy cattle using electronic data loggers . J. Dairy Sci . 98 : 5304 – 5312 . doi: https://doi.org/10.3168/jds.2014-9223 Google Scholar Crossref Search ADS PubMed Tsai , I. C . 2017 . Differences in behavioral and physiological variables measured with precision dairy monitoring technologies associated with postpartum diseases . MS Thesis. University of Kentucky , Lexington. Wehrend , A. , E. Hofmann , K. Failing , and H. Bostedt . 2006 . Behaviour during the first stage of labour in cattle: influence of parity and dystocia . Appl. Anim. Beh. Sci . 100 : 164 – 170 . doi: https://doi.org/10.1016/j.applanim.2005.11.008 Google Scholar Crossref Search ADS Zaborski , D. , W. Grzesiak , I. Szatkowska , A. Dybus , M. Muszynska , and M. Jedrzejczak . 2009 . Factors affecting dystocia in cattle . Reprod. Domest. Anim . 44 : 540 – 551 . doi: https://doi.org/10.1111/j.1439-0531.2008.01123.x Google Scholar Crossref Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the American Society of Animal Science. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Massender,, Erin;Brito, Luiz, F;Cánovas,, Angela;Baes, Christine, F;Kennedy,, Delma;Schenkel, Flavio, S

doi: 10.1093/jas/sky455pmid: 30500934

Abstract Genetic parameters were estimated for growth, ultrasound, and carcass traits in a Canadian crossbred heavy lamb population. Traits analyzed included birth, weaning, post-weaning, and ultrasound scanning weights; pre- and post-weaning average daily gain; ultrasonically measured eye muscle and fat depths; hot carcass weight; fat depth at the GR site (110 mm from the midline on the 12th rib); carcass conformation scores; saleable meat yield; price grid value; and total carcass value. The impact of three alternative slaughter endpoints (slaughter age, carcass weight, and carcass fatness) on genetic parameter estimates was also evaluated. In general, carcass traits were found to be moderately heritable, with heritability estimates ranging from 0.17 ± 0.02 for hot carcass weight at a constant slaughter age to 0.34 ± 0.02 for average carcass conformation score at a constant carcass weight. Heritability estimates were similar when observations were adjusted to alternative slaughter endpoints, but for some traits, phenotypic variance and genetic correlation estimates differed. Genetic correlations between carcass traits and growth and ultrasound traits were typically favorable. Ultrasonically measured eye muscle depth and fat depth were found to be moderately to strongly positively correlated with hot carcass weight (0.33 ± 0.15 to 0.71 ± 0.19) and fat depth at the GR site (0.38 ± 0.14 to 0.74 ± 0.12), respectively, reaffirming the usefulness of selection on ultrasound traits to improve carcass yield and quality. Genetic correlations among carcass traits were generally favorable, with the exception of moderate unfavorable positive genetic correlations between fat depth at the GR site and primal cut carcass conformation scores (0.31 ± 0.05 to 0.60 ± 0.05). Overall, the results of this research suggest that there is potential to improve carcass yield and quality through genetic selection and provides the population-specific genetic parameter estimates needed for the genetic evaluation of carcass traits in the Canadian sheep population. Nevertheless, the optimal endpoint for carcass trait genetic evaluations will need to be further investigated, considering both the current findings and additional information on production practices in the industry. INTRODUCTION The Canadian sheep industry faces many challenges in providing a consistent supply of high-quality lamb. Improved production efficiency and lamb quality are critical to meet the domestic demand for Canadian lamb products (Gooch et al., 2006). The Canadian Sheep Genetic Evaluation System (CSGES) provides genetic evaluations for reproduction, growth, and ultrasound traits. Animals are also ranked for six selection indexes designed to meet various Terminal and Maternal economic breeding objectives, as proposed by Quinton et al. (2014). However, genetic evaluations for carcass traits, which have a major impact on the profitability of meat lamb production, are currently unavailable. Lambs marketed at the ideal weight, age, and fatness level are of increased value to processors and yield greater producer profit under a price grid classification system. Carcass traits, such as carcass weight, fat depth, and conformation, have been considered economically important traits for many years in the CSGES (Tosh and Wilton, 2002; Quinton et al., 2014), but infrequent phenotyping has prevented their genetic evaluation. Since 2007, all heavy lambs in the province of Quebec (lambs under 1 yr of age with a carcass weight of at least 16.4 kg) have been marketed through the Heavy Lamb Sales Agency (HLSA). Producer payment through the HLSA utilizes a price grid classification system, thus rewarding producers that meet target weight, muscularity, and fatness levels and providing the phenotypes needed for the genetic evaluation of these traits (Les Éleveurs d’ovins du Québec, 2017). Although genetic parameters have not been previously estimated for these traits in a Canadian sheep population, published genetic parameter estimates suggest that carcass traits are moderately heritable (Safari et al., 2005) and carcass trait genetic evaluations have been successfully implemented in other national sheep breeding programs (Beef + Lamb New Zealand Genetics, 2017; Swan et al., 2017). Nonetheless, population-specific genetic parameter estimates are needed for the implementation of carcass trait genetic evaluations in the Canadian sheep industry. A slaughter endpoint is a criterion used to decide when to market animals for slaughter. The choice of slaughter endpoint is often regionally dependent, with slaughter age, carcass weight, and carcass fatness level being common decision support criteria. The Canadian sheep industry is highly decentralized and production practices vary greatly by flock size and geographical region (Quinton et al., 2014), thus it is unknown which slaughter endpoint(s) would be the most suitable for the genetic evaluation of carcass traits in the CSGES. The slaughter endpoint used in genetic evaluations may have implications for the genetic correlations between traits (Pollott et al., 1994), and consequently, influence the selection response achieved through the use of multiple-trait selection indexes. Thus, it is important to evaluate the impact of alternative slaughter endpoints to ensure that proposed carcass trait genetic evaluations are representative of diverse breeding objectives among commercial sheep producers. Breeding objectives utilized to derive existing CSGES selection indexes (Quinton et al., 2014) assumed that rail-graded lambs were marketed at a constant slaughter age. Quinton et al. (2014) noted that producers typically aim to market lambs at a constant weight but commercial data demonstrated a wide range of weight and age endpoints. The effect of alternative slaughter endpoints on carcass trait genetic parameter estimates has been studied more extensively in beef cattle, as reviewed by Ríos-Utrera and Van Vleck (2004), but, to the best of our knowledge, evaluation of all three slaughter endpoints in sheep is limited to studies by Pollott et al. (1994) and Conington et al. (1998). These studies utilized a sample of animals from designed experiments in British sheep populations and differed from the present study with respect to the breeds, production systems, and traits analyzed. Furthermore, the genetic parameter estimates presented by Pollott et al. (1994) and Conington et al. (1998) typically had large standard errors, which limited the interpretation of genetic correlation results. Thus, it is unclear if the results presented by Pollott et al. (1994) and Conington et al. (1998) would be applicable to the breeds and production practices in the Canadian sheep industry. Consequently, the objectives of this research were 1) to estimate the first genetic parameters for carcass traits in a Canadian crossbred heavy lamb population; 2) to estimate the correlations between carcass, growth, and ultrasound traits; and 3) to evaluate the impact of alternative slaughter endpoints on carcass trait genetic parameter estimates. MATERIALS AND METHODS Animals Data used in this research were obtained from commercial producer and abattoir records, thus, animal care approval was not required. As part of the routine grading procedures of the HLSA, carcass yield and quality measurements on over 80,000 heavy lambs raised under commercial conditions were recorded between January 2011 and August 2013. Animal identification (tattoo number) was used to link carcass measurements to pedigree and management records of 16,565 lambs enrolled in the CSGES. Consistent with the Canadian sheep population, approximately 29% of the lambs with carcass records were purebred, while in the total dataset approximately 60% of animals were purebred. The breeds represented included Polled Dorset (DP; 26%), Rideau Arcott (RI; 22%), Romanov (RV; 21%), Suffolk (SU; 13%), Polypay (PO; 8%), Canadian Arcott (CD; 4%), Hampshire (HA; 3%), and rare breeds and unknown crosses (3%). The major breeds are typical of the Canadian sheep population and included highly prolific Maternal breeds (RV, RI, PO), Maternal or Maternal sire breeds (DP), and Terminal sire breeds (SU, HA, CD). Data Management information (sex, date of birth, breed composition, age at weighing, flock identification, and producer-defined management group) and growth and ultrasound trait measurements were retrieved from the CSGES for all animals with carcass records and their relatives. Contemporary groups were defined as management group within year and flock. Abattoir identification was unavailable, but slaughter groups were formed as unique month-year combinations of slaughter date to account for seasonal or market differences that may have influenced carcass characteristics. The difference between birth and slaughter dates was used to calculate slaughter age (SAGE, days). Lambs used in this research were an average of 172.0 ± 40.9 d of age at slaughter and ranged from 54 to 353 d of age. Only carcasses that met the heavy lamb classification criteria (HCW ≥ 16.4 kg and SAGE ≤ 365 d) and animals with growth trait records that were within the CSGES trait limits, as per Schaeffer and Szkotnicki (2015), were retained. Traits analyzed in this research included growth, ultrasound, and carcass traits. Birthweight (BWT, kg) was measured within 24 h of birth, while records for weaning weight (WWT, kg), post-weaning weight (PWWT, kg), and ultrasound scanning weight (WTUS, kg) were measured at an average age of 54.5 ± 9.8, 97.5 ± 11.5, and 97.8 ± 11.8 d, respectively. Adjusted 50-d and 100-d weights were used to calculate pre-weaning average daily gain (ADG50, kg) and post-weaning average daily gain (ADG100, kg), assuming linear growth during each period (Schaeffer and Szkotnicki, 2015). Eye muscle depth (EMDUS, mm), a measure of the longissimus dorsi muscle and average fat depth (FATUS, mm) were measured by accredited ultrasound technicians at a site halfway between the last rib and the hip bone between the third and fourth lumbar vertebrae. Hot carcass weight (HCW, kg) was recorded after the carcass was dressed, following the specifications for Canadian lamb processing (Government of Canada, 1992). Carcass fat depth (FATGR, mm) was measured as the total tissue depth at the GR site (110 mm from the midline on the 12th rib) (Kirton and Johnson, 1979). Carcass conformation scores, ranging from 1 (poor muscling) to 5 (excellent muscling), were used to assess the muscularity of carcasses in three primal cuts: shoulder (SHOUL), loin (LOIN), and leg (LEG). Primal cut conformation scores were then averaged and rounded to the nearest whole number to calculate an average carcass conformation score (AVGCONF). Saleable meat yield (SMY, % of HCW) was predicted from FATGR and AVGCONF using equations derived by Jones et al. (1996). Price grid value (CINDEX) was derived from price grid class (20.0 to 24.0 kg and <20.0 or >24.0 kg), FATGR, and AVGCONF measurements, as per the Heavy Lamb Sales Agency Producer’s Guide (Les Éleveurs d’ovins du Québec, 2017). Total carcass value (PRICE, $CAD) was estimated based on CINDEX and HCW, assuming a base price per kilogram of HCW of $7.85 (Les Éleveurs d’ovins du Québec, 2017). Both R (R Core Team, 2017) and SAS (SAS Institute Inc., 2013) statistical software were used for preliminary data editing. Animals with an unknown dam, contemporary group with fewer than three animals, or that were cross-fostered or bottle-fed were excluded from the final dataset. Fixed-effects models, as described in Table 1, were used to adjust the observations for each trait and any animal with a residual more than 3 SDs from the mean was further excluded to remove potential outliers. There were 29,923 animals with records in the final dataset, including 14,441 animals with carcass records. Table 1. Fixed and random effects fitted in the genetic parameter estimation models for various trait groups1 Growth traits Ultrasound traits Carcass traits Effect BWT WWT PWWT ADG50; ADG100 WTUS Age constant Weight constant Age constant Weight constant Fat constant Fixed categorical effects Sex + + + + + + + + + + Dam age at parity + + + + + + + + + + Birth type + − − − − − − − − − Birth-rearing type − + + + + + + + + + Slaughter group − − − − − − − + + + Fixed covariate effects Breed composition + + + + + + + + + + SAGE − − − − − − − + − − HCW − − − − − − − − + − FATGR − − − − − − − − − + Age at weaning − + − − − − − − − Age at post-weaning − − + − − − − − − − Age at ultrasound − − − − + + − − − − Weight at ultrasound − − − − − − + − − − Random effects Additive genetic + + + + + + + + + + Contemporary group + + + + + + + + + + Maternal genetic + + + + + − − − − − Maternal permanent environmental + + + + + − − − − − Growth traits Ultrasound traits Carcass traits Effect BWT WWT PWWT ADG50; ADG100 WTUS Age constant Weight constant Age constant Weight constant Fat constant Fixed categorical effects Sex + + + + + + + + + + Dam age at parity + + + + + + + + + + Birth type + − − − − − − − − − Birth-rearing type − + + + + + + + + + Slaughter group − − − − − − − + + + Fixed covariate effects Breed composition + + + + + + + + + + SAGE − − − − − − − + − − HCW − − − − − − − − + − FATGR − − − − − − − − − + Age at weaning − + − − − − − − − Age at post-weaning − − + − − − − − − − Age at ultrasound − − − − + + − − − − Weight at ultrasound − − − − − − + − − − Random effects Additive genetic + + + + + + + + + + Contemporary group + + + + + + + + + + Maternal genetic + + + + + − − − − − Maternal permanent environmental + + + + + − − − − − 1ADG50 = pre-weaning average daily gain; ADG100 = post-weaning average daily gain; BWT = birthweight; FATGR = fat depth at the GR site; HCW = hot carcass weight; PWWT = post-weaning weight; SAGE = age at slaughter; WTUS = scanning weight; WWT = weaning weight. View Large Table 1. Fixed and random effects fitted in the genetic parameter estimation models for various trait groups1 Growth traits Ultrasound traits Carcass traits Effect BWT WWT PWWT ADG50; ADG100 WTUS Age constant Weight constant Age constant Weight constant Fat constant Fixed categorical effects Sex + + + + + + + + + + Dam age at parity + + + + + + + + + + Birth type + − − − − − − − − − Birth-rearing type − + + + + + + + + + Slaughter group − − − − − − − + + + Fixed covariate effects Breed composition + + + + + + + + + + SAGE − − − − − − − + − − HCW − − − − − − − − + − FATGR − − − − − − − − − + Age at weaning − + − − − − − − − Age at post-weaning − − + − − − − − − − Age at ultrasound − − − − + + − − − − Weight at ultrasound − − − − − − + − − − Random effects Additive genetic + + + + + + + + + + Contemporary group + + + + + + + + + + Maternal genetic + + + + + − − − − − Maternal permanent environmental + + + + + − − − − − Growth traits Ultrasound traits Carcass traits Effect BWT WWT PWWT ADG50; ADG100 WTUS Age constant Weight constant Age constant Weight constant Fat constant Fixed categorical effects Sex + + + + + + + + + + Dam age at parity + + + + + + + + + + Birth type + − − − − − − − − − Birth-rearing type − + + + + + + + + + Slaughter group − − − − − − − + + + Fixed covariate effects Breed composition + + + + + + + + + + SAGE − − − − − − − + − − HCW − − − − − − − − + − FATGR − − − − − − − − − + Age at weaning − + − − − − − − − Age at post-weaning − − + − − − − − − − Age at ultrasound − − − − + + − − − − Weight at ultrasound − − − − − − + − − − Random effects Additive genetic + + + + + + + + + + Contemporary group + + + + + + + + + + Maternal genetic + + + + + − − − − − Maternal permanent environmental + + + + + − − − − − 1ADG50 = pre-weaning average daily gain; ADG100 = post-weaning average daily gain; BWT = birthweight; FATGR = fat depth at the GR site; HCW = hot carcass weight; PWWT = post-weaning weight; SAGE = age at slaughter; WTUS = scanning weight; WWT = weaning weight. View Large Pedigree The pedigree package (Coster, 2013) was used to trim branches of the full CSGES pedigree without carcass trait data. The final pedigree contained 37,885 animals over 21 generations. Due to the editing procedure, dam information was known for all animals with records, but sire identification information was missing for about 19% of animals with records, consistent with the common sheep production practice of group mating. Animals with records were the progeny of 2,760 known sires and 17,451 dams, and had an average pedigree depth of 10.4 generations. Known sires and dams had an average of 8.8 and 1.7 progeny with records, respectively. Statistical Analyses The MEANS and GLM procedures of SAS (SAS Institute Inc., 2013) were used to estimate descriptive statistics (Table 2) and test the significance of fixed effects, respectively. (Co)variance components were estimated using mixed linear animal models in the ASReml statistical software (Gilmour et al., 2015). The fixed and random effects included in the models for the various trait groups are summarized in Table 1 and described below. Univariate analyses were used to estimate the heritability of each trait, while genetic and phenotypic correlations were estimated from bivariate analyses. For the purposes of comparison, the heritability estimates were categorized as low (<0.15), moderate (0.15 to 0.30), or high (>0.30), while genetic correlations were categorized as being weak (<0.30), moderate (0.30 to 0.70), or strong (>0.70). Phenotypic correlations between growth and carcass traits are presented in Supplementary Tables S1 to S3. Preliminary descriptive analysis and genetic parameter estimates for a subset of the traits studied were initially presented in Massender et al. (2018). Table 2. Trait abbreviations and descriptive statistics1 Trait Abbreviation n Mean ± SD Range CV (%) Growth traits  Birthweight, kg BWT 22,494 4.3 ± 1.1 1.2–8.0 25.6  50-d weaning weight, kg WWT 29,082 19.8 ± 6.0 3.5–40.0 30.5  100-d post-weaning weight, kg PWWT 26,763 33.0 ± 8.1 8.0–65.0 24.5  Pre-weaning average daily gain, kg ADG50 29,082 0.3 ± 0.1 0.0–1.1 31.4  Post-weaning average daily gain, kg ADG100 26,693 0.3 ± 0.1 0.0–1.5 31.4  Scanning weight, kg WTUS 1,299 38.1 ± 8.1 15.4–64.0 21.1 Ultrasound traits  Ultrasonic eye muscle depth, mm EMDUS 1,299 26.8 ± 3.8 10.6–36.6 14.2  Ultrasonic fat depth, mm FATUS 1,299 3.8 ± 1.4 0.9–10.8 36.9 Carcass traits  Hot carcass weight, kg HCW 14,441 23.2 ± 2.3 16.4–31.7 9.9  Carcass fat depth at the GR site, mm FATGR 14,441 11.0 ± 3.3 1.0–19.0 30.0  Slaughter age, days SAGE 14,441 172.0 ± 40.9 54.0–353.0 23.8  Predicted saleable meat yield, % SMY 14,441 77.2 ± 1.5 72.0–82.0 2.0  Leg conformation, score LEG 14,441 2.9 ± 0.5 1.0–4.0 18.2  Loin conformation, score LOIN 14,441 3.5 ± 0.6 2.0–5.0 16.8  Shoulder conformation, score SHOUL 14,441 3.0 ± 0.6 1.0–5.0 21.5  Average carcass conformation, score AVGCONF 14,441 3.1 ± 0.6 2.0–4.0 17.8  Carcass price grid value, score CINDEX 14,441 101.8 ± 3.5 85.0–106.0 3.4  Total carcass value, $CAD PRICE 14,441 184.4 ± 16.5 124.1–218.6 9.0 Trait Abbreviation n Mean ± SD Range CV (%) Growth traits  Birthweight, kg BWT 22,494 4.3 ± 1.1 1.2–8.0 25.6  50-d weaning weight, kg WWT 29,082 19.8 ± 6.0 3.5–40.0 30.5  100-d post-weaning weight, kg PWWT 26,763 33.0 ± 8.1 8.0–65.0 24.5  Pre-weaning average daily gain, kg ADG50 29,082 0.3 ± 0.1 0.0–1.1 31.4  Post-weaning average daily gain, kg ADG100 26,693 0.3 ± 0.1 0.0–1.5 31.4  Scanning weight, kg WTUS 1,299 38.1 ± 8.1 15.4–64.0 21.1 Ultrasound traits  Ultrasonic eye muscle depth, mm EMDUS 1,299 26.8 ± 3.8 10.6–36.6 14.2  Ultrasonic fat depth, mm FATUS 1,299 3.8 ± 1.4 0.9–10.8 36.9 Carcass traits  Hot carcass weight, kg HCW 14,441 23.2 ± 2.3 16.4–31.7 9.9  Carcass fat depth at the GR site, mm FATGR 14,441 11.0 ± 3.3 1.0–19.0 30.0  Slaughter age, days SAGE 14,441 172.0 ± 40.9 54.0–353.0 23.8  Predicted saleable meat yield, % SMY 14,441 77.2 ± 1.5 72.0–82.0 2.0  Leg conformation, score LEG 14,441 2.9 ± 0.5 1.0–4.0 18.2  Loin conformation, score LOIN 14,441 3.5 ± 0.6 2.0–5.0 16.8  Shoulder conformation, score SHOUL 14,441 3.0 ± 0.6 1.0–5.0 21.5  Average carcass conformation, score AVGCONF 14,441 3.1 ± 0.6 2.0–4.0 17.8  Carcass price grid value, score CINDEX 14,441 101.8 ± 3.5 85.0–106.0 3.4  Total carcass value, $CAD PRICE 14,441 184.4 ± 16.5 124.1–218.6 9.0 1CV (%) = coefficient of variation; n = number of records; SD = standard deviation. View Large Table 2. Trait abbreviations and descriptive statistics1 Trait Abbreviation n Mean ± SD Range CV (%) Growth traits  Birthweight, kg BWT 22,494 4.3 ± 1.1 1.2–8.0 25.6  50-d weaning weight, kg WWT 29,082 19.8 ± 6.0 3.5–40.0 30.5  100-d post-weaning weight, kg PWWT 26,763 33.0 ± 8.1 8.0–65.0 24.5  Pre-weaning average daily gain, kg ADG50 29,082 0.3 ± 0.1 0.0–1.1 31.4  Post-weaning average daily gain, kg ADG100 26,693 0.3 ± 0.1 0.0–1.5 31.4  Scanning weight, kg WTUS 1,299 38.1 ± 8.1 15.4–64.0 21.1 Ultrasound traits  Ultrasonic eye muscle depth, mm EMDUS 1,299 26.8 ± 3.8 10.6–36.6 14.2  Ultrasonic fat depth, mm FATUS 1,299 3.8 ± 1.4 0.9–10.8 36.9 Carcass traits  Hot carcass weight, kg HCW 14,441 23.2 ± 2.3 16.4–31.7 9.9  Carcass fat depth at the GR site, mm FATGR 14,441 11.0 ± 3.3 1.0–19.0 30.0  Slaughter age, days SAGE 14,441 172.0 ± 40.9 54.0–353.0 23.8  Predicted saleable meat yield, % SMY 14,441 77.2 ± 1.5 72.0–82.0 2.0  Leg conformation, score LEG 14,441 2.9 ± 0.5 1.0–4.0 18.2  Loin conformation, score LOIN 14,441 3.5 ± 0.6 2.0–5.0 16.8  Shoulder conformation, score SHOUL 14,441 3.0 ± 0.6 1.0–5.0 21.5  Average carcass conformation, score AVGCONF 14,441 3.1 ± 0.6 2.0–4.0 17.8  Carcass price grid value, score CINDEX 14,441 101.8 ± 3.5 85.0–106.0 3.4  Total carcass value, $CAD PRICE 14,441 184.4 ± 16.5 124.1–218.6 9.0 Trait Abbreviation n Mean ± SD Range CV (%) Growth traits  Birthweight, kg BWT 22,494 4.3 ± 1.1 1.2–8.0 25.6  50-d weaning weight, kg WWT 29,082 19.8 ± 6.0 3.5–40.0 30.5  100-d post-weaning weight, kg PWWT 26,763 33.0 ± 8.1 8.0–65.0 24.5  Pre-weaning average daily gain, kg ADG50 29,082 0.3 ± 0.1 0.0–1.1 31.4  Post-weaning average daily gain, kg ADG100 26,693 0.3 ± 0.1 0.0–1.5 31.4  Scanning weight, kg WTUS 1,299 38.1 ± 8.1 15.4–64.0 21.1 Ultrasound traits  Ultrasonic eye muscle depth, mm EMDUS 1,299 26.8 ± 3.8 10.6–36.6 14.2  Ultrasonic fat depth, mm FATUS 1,299 3.8 ± 1.4 0.9–10.8 36.9 Carcass traits  Hot carcass weight, kg HCW 14,441 23.2 ± 2.3 16.4–31.7 9.9  Carcass fat depth at the GR site, mm FATGR 14,441 11.0 ± 3.3 1.0–19.0 30.0  Slaughter age, days SAGE 14,441 172.0 ± 40.9 54.0–353.0 23.8  Predicted saleable meat yield, % SMY 14,441 77.2 ± 1.5 72.0–82.0 2.0  Leg conformation, score LEG 14,441 2.9 ± 0.5 1.0–4.0 18.2  Loin conformation, score LOIN 14,441 3.5 ± 0.6 2.0–5.0 16.8  Shoulder conformation, score SHOUL 14,441 3.0 ± 0.6 1.0–5.0 21.5  Average carcass conformation, score AVGCONF 14,441 3.1 ± 0.6 2.0–4.0 17.8  Carcass price grid value, score CINDEX 14,441 101.8 ± 3.5 85.0–106.0 3.4  Total carcass value, $CAD PRICE 14,441 184.4 ± 16.5 124.1–218.6 9.0 1CV (%) = coefficient of variation; n = number of records; SD = standard deviation. View Large Fixed effects. The final model for each trait (Table 1) included fixed effects of sex (male or female), dam age at parity (1, 2, …, 7), birth (single, twin, or triplet or more) or birth-rearing type (born as single raised as single, born as multiple raised as single, born as twin raised as twin, born as triplet or more and raised as twin, and born as triplet or more and raised as triplet or more), and linear covariates of fractional breed composition for the seven major breeds (DP, RV, RI, SU, PO, CD, and HA). Analyses of WWT, PWWT, and WTUS included age of the animal at measurement as a linear covariate. Ultrasound measurements were adjusted to either a fixed age or WTUS. Slaughter group was included in all carcass trait models and the effect of slaughter endpoint on genetic parameter estimates was evaluated by modeling carcass traits at each of three alternative slaughter endpoints (age at slaughter, carcass weight, and carcass fat depth). Random effects. Random effects in each model included contemporary group and animal additive genetic effect (Table 1). Contemporary groups were assumed to be uncorrelated, but the covariance between traits due to the contemporary group effect was fitted in the bivariate analyses. As maternal additive genetic and maternal permanent environmental effects were significant (P < 0.01), both effects were retained in the final growth trait models, with no covariance assumed between the direct and maternal genetic effects. Some ultrasound and carcass traits have been found to be weakly to moderately influenced by maternal effects (e.g., Mortimer et al., 2010; Einarsson et al., 2015); however, the structure of data used in this research was inadequate to estimate these effects. RESULTS AND DISCUSSION Descriptive Statistics The number of observations varied from 1,299 for ultrasound measurements and WTUS to 29,082 for WWT and ADG50 (Table 2). Ultrasound measurements are recorded infrequently in Canada due to their cost and lack of access to ultrasound technicians. This suggests that the genetic evaluation of carcass trait records generated through rail-graded marketing systems may be a promising method to improve carcass quality. However, ultrasound traits remain valuable indicators of carcass quality because rail-graded marketing systems are uncommon outside of Quebec and carcass traits cannot be measured directly on breeding candidates. According to the HLSA, the ideal carcass is between 20.0 and 24.0 kg with high muscularity scores (4 to 5) and a target FATGR measurement between 7 and 12 mm (Les Éleveurs d’ovins du Québec, 2017). In this dataset, average HCW (23.2 ± 2.3) and FATGR (11.0 ± 3.3) were on the high end of the ideal range, while the average AVGCONF (3.1 ± 0.6) was low (Table 2). Approximately 56%, 61%, and 23% of observations were within the ideal ranges for HCW, FATGR, and AVGCONF, respectively. Phenotypic Variation and Heritability Growth traits. Direct heritability estimates for growth traits were generally moderate, with maternal heritability and maternal permanent environmental variance ratio estimates decreasing and direct heritability estimates increasing as age at measurement increased (Table 3), consistent with results presented by Tosh and Kemp (1994) and Boareki (2017) in purebred Canadian sheep populations. Maternal heritability and maternal permanent environmental variance ratio were the largest for BWT (0.15 ± 0.01 and 0.10 ± 0.01, respectively) and the smallest for ADG100 (0.02 ± 0.01 and 0.03 ± 0.01, respectively) (Table 3). Although significant maternal effects have been reported for carcass traits in other populations (e.g., Mortimer et al., 2010; Einarsson et al., 2015), the low maternal effect estimates observed for PWWT and ADG100 suggest that maternal effects are likely to be small for ultrasound and carcass traits. Table 3. Phenotypic variance (σp2) ⁠, ratio of contemporary group effect variance to phenotypic variance (g2), maternal permanent environmental effect variance ratio (c2), direct heritability (hd2) ⁠, maternal heritability (hm2) ⁠, and total heritability (ht2) estimates for growth and ultrasound traits1 Trait2 σp2 g2 c2 hd2 hm2 ht2 Growth traits  BWT 0.62 0.19 ± 0.01 0.10 ± 0.01 0.18 ± 0.01 0.15 ± 0.01 0.26 ± 0.01  WWT 19.23 0.30 ± 0.01 0.05 ± 0.01 0.23 ± 0.01 0.05 ± 0.01 0.26 ± 0.01  PWWT 37.08 0.34 ± 0.01 0.03 ± 0.01 0.25 ± 0.01 0.05 ± 0.01 0.27 ± 0.01  WTUS 39.68 0.29 ± 0.04 – 0.30 ± 0.06 – 0.30 ± 0.06  ADG50 0.01 0.33 ± 0.01 0.05 ± 0.01 0.21 ± 0.01 0.06 ± 0.01 0.24 ± 0.01  ADG100 0.01 0.45 ± 0.01 0.03 ± 0.01 0.15 ± 0.01 0.02 ± 0.01 0.15 ± 0.01 Ultrasound traits  EMDUSa 10.92 0.28 ± 0.04 – 0.16 ± 0.06 – 0.16 ± 0.06  EMDUSw 6.26 0.32 ± 0.04 – 0.35 ± 0.06 – 0.35 ± 0.06  FATUSa 2.16 0.49 ± 0.04 – 0.22 ± 0.05 – 0.22 ± 0.05  FATUSw 1.60 0.56 ± 0.03 – 0.22 ± 0.05 – 0.22 ± 0.05 Trait2 σp2 g2 c2 hd2 hm2 ht2 Growth traits  BWT 0.62 0.19 ± 0.01 0.10 ± 0.01 0.18 ± 0.01 0.15 ± 0.01 0.26 ± 0.01  WWT 19.23 0.30 ± 0.01 0.05 ± 0.01 0.23 ± 0.01 0.05 ± 0.01 0.26 ± 0.01  PWWT 37.08 0.34 ± 0.01 0.03 ± 0.01 0.25 ± 0.01 0.05 ± 0.01 0.27 ± 0.01  WTUS 39.68 0.29 ± 0.04 – 0.30 ± 0.06 – 0.30 ± 0.06  ADG50 0.01 0.33 ± 0.01 0.05 ± 0.01 0.21 ± 0.01 0.06 ± 0.01 0.24 ± 0.01  ADG100 0.01 0.45 ± 0.01 0.03 ± 0.01 0.15 ± 0.01 0.02 ± 0.01 0.15 ± 0.01 Ultrasound traits  EMDUSa 10.92 0.28 ± 0.04 – 0.16 ± 0.06 – 0.16 ± 0.06  EMDUSw 6.26 0.32 ± 0.04 – 0.35 ± 0.06 – 0.35 ± 0.06  FATUSa 2.16 0.49 ± 0.04 – 0.22 ± 0.05 – 0.22 ± 0.05  FATUSw 1.60 0.56 ± 0.03 – 0.22 ± 0.05 – 0.22 ± 0.05 1Parameter estimates are followed by their approximate standard error. 2ADG50 = pre-weaning average daily gain; ADG100 = post-weaning average daily gain; BWT = birthweight; EMDUS = ultrasonically measured eye muscle depth; FATUS = ultrasonically measured fat depth; PWWT = post-weaning weight; WTUS = scanning weight; WWT = weaning weight. Ultrasound eye muscle and fat depth measurements adjusted to a constant age (EMDUSa, FATUSa) or scanning weight (EMDUSw, FATUSw). View Large Table 3. Phenotypic variance (σp2) ⁠, ratio of contemporary group effect variance to phenotypic variance (g2), maternal permanent environmental effect variance ratio (c2), direct heritability (hd2) ⁠, maternal heritability (hm2) ⁠, and total heritability (ht2) estimates for growth and ultrasound traits1 Trait2 σp2 g2 c2 hd2 hm2 ht2 Growth traits  BWT 0.62 0.19 ± 0.01 0.10 ± 0.01 0.18 ± 0.01 0.15 ± 0.01 0.26 ± 0.01  WWT 19.23 0.30 ± 0.01 0.05 ± 0.01 0.23 ± 0.01 0.05 ± 0.01 0.26 ± 0.01  PWWT 37.08 0.34 ± 0.01 0.03 ± 0.01 0.25 ± 0.01 0.05 ± 0.01 0.27 ± 0.01  WTUS 39.68 0.29 ± 0.04 – 0.30 ± 0.06 – 0.30 ± 0.06  ADG50 0.01 0.33 ± 0.01 0.05 ± 0.01 0.21 ± 0.01 0.06 ± 0.01 0.24 ± 0.01  ADG100 0.01 0.45 ± 0.01 0.03 ± 0.01 0.15 ± 0.01 0.02 ± 0.01 0.15 ± 0.01 Ultrasound traits  EMDUSa 10.92 0.28 ± 0.04 – 0.16 ± 0.06 – 0.16 ± 0.06  EMDUSw 6.26 0.32 ± 0.04 – 0.35 ± 0.06 – 0.35 ± 0.06  FATUSa 2.16 0.49 ± 0.04 – 0.22 ± 0.05 – 0.22 ± 0.05  FATUSw 1.60 0.56 ± 0.03 – 0.22 ± 0.05 – 0.22 ± 0.05 Trait2 σp2 g2 c2 hd2 hm2 ht2 Growth traits  BWT 0.62 0.19 ± 0.01 0.10 ± 0.01 0.18 ± 0.01 0.15 ± 0.01 0.26 ± 0.01  WWT 19.23 0.30 ± 0.01 0.05 ± 0.01 0.23 ± 0.01 0.05 ± 0.01 0.26 ± 0.01  PWWT 37.08 0.34 ± 0.01 0.03 ± 0.01 0.25 ± 0.01 0.05 ± 0.01 0.27 ± 0.01  WTUS 39.68 0.29 ± 0.04 – 0.30 ± 0.06 – 0.30 ± 0.06  ADG50 0.01 0.33 ± 0.01 0.05 ± 0.01 0.21 ± 0.01 0.06 ± 0.01 0.24 ± 0.01  ADG100 0.01 0.45 ± 0.01 0.03 ± 0.01 0.15 ± 0.01 0.02 ± 0.01 0.15 ± 0.01 Ultrasound traits  EMDUSa 10.92 0.28 ± 0.04 – 0.16 ± 0.06 – 0.16 ± 0.06  EMDUSw 6.26 0.32 ± 0.04 – 0.35 ± 0.06 – 0.35 ± 0.06  FATUSa 2.16 0.49 ± 0.04 – 0.22 ± 0.05 – 0.22 ± 0.05  FATUSw 1.60 0.56 ± 0.03 – 0.22 ± 0.05 – 0.22 ± 0.05 1Parameter estimates are followed by their approximate standard error. 2ADG50 = pre-weaning average daily gain; ADG100 = post-weaning average daily gain; BWT = birthweight; EMDUS = ultrasonically measured eye muscle depth; FATUS = ultrasonically measured fat depth; PWWT = post-weaning weight; WTUS = scanning weight; WWT = weaning weight. Ultrasound eye muscle and fat depth measurements adjusted to a constant age (EMDUSa, FATUSa) or scanning weight (EMDUSw, FATUSw). View Large Direct heritability of BWT was estimated to be 0.18 ± 0.01 and was similar to published estimates (Table 3), which range from 0.07 ± 0.03 in an Australian Maternal crossbred population (Ingham et al., 2007) to 0.22 ± 0.04 in Australian Merino sheep (Mortimer et al., 2017a). Birthweight has generally been found to be less heritable in Canadian sheep populations (Tosh and Kemp, 1994; Boareki, 2017), with published heritability estimates ranging from 0.07 for RV sheep to 0.39 for HA sheep (Tosh and Kemp, 1994). The direct heritability estimate for WWT (0.23 ± 0.01; Table 3) was consistent with literature estimates, which varied from 0.09 ± 0.05 in a multi-breed New Zealand Terminal and Dual-Purpose crossbred population (Payne et al., 2009) to 0.40 ± 0.03 in Australian Merino sheep (Huisman and Brown, 2008). However, WWT is often measured at a later age in other countries and lower heritability estimates (0.05 to 0.21) have been reported for WWT and adjusted WWT in Canadian sheep populations (Tosh and Kemp, 1994; Boareki, 2017). Direct heritability for PWWT was estimated to be 0.25 ± 0.01 (Table 3). Post-weaning weight is measured at a wide range of ages (100 to 180 d), but, Maxa et al. (2007a) reported a lower direct heritability for PWWT (0.17 ± 0.04) for a population of SU sheep and Terminal crosses from the Czech Republic at a similar age. The direct heritability estimates reported by Tosh and Kemp (1994) and Boareki (2017) for PWWT and adjusted PWWT in Canadian populations ranged from 0.14 to 0.39. The heritability of WTUS (0.30 ± 0.06; Table 3) was found to be similar to recent literature results (Brito et al., 2017; Mortimer et al., 2017a). Direct heritability estimates for ADG50 and ADG100 were 0.21 ± 0.01 and 0.15 ± 0.01, respectively (Table 3). The direct heritability estimate for ADG100 was similar to the estimate of 0.16 ± 0.03 by Maximini et al. (2012) in a multi-breed Austrian sheep population. Ultrasound traits. Ultrasound traits were found to be moderately to highly heritable (0.16 ± 0.06 to 0.35 ± 0.06; Table 3). Direct heritability estimates for EMDUS in the literature have ranged from 0.32 ± 0.02 in Lleyn sheep (Ceyhan et al., 2015) to 0.40 ± 0.05 in Scottish Blackface sheep (Karamichou et al., 2007) and 0.20 ± 0.06 (Mortimer et al., 2017a) to 0.42 in an Icelandic sheep population (Einarsson et al., 2015) for age- and weight-constant EMDUS, respectively. Direct heritability estimates ranging from 0.23 ± 0.01 in a crossbred Australian population (Walkom and Brown, 2016) to 0.37 ± 0.02 (Ceyhan et al., 2015) and 0.12 ± 0.07 in Danish Shropshire sheep (Maxa et al., 2007b) to 0.42 (Einarsson et al., 2015) have been reported for age- and weight-constant FATUS, respectively. Greater phenotypic variation was observed for age-constant EMDUS (10.92 mm2) compared to weight-constant EMDUS (6.26 mm2) (Table 3), while additive genetic variance estimates remained similar, leading to higher direct heritability estimates for weight-constant EMDUS (0.35 ± 0.06 vs. 0.16 ± 0.06). However, the direct heritability estimates for FATUS were the same at alternative endpoints (0.22 ± 0.05), due to similar phenotypic and additive genetic variation in the weight- and age-constant analyses. Higher phenotypic variation for age-constant EMDUS was also reported by Fernandes et al. (2004) in a Canadian sheep population, although they reported higher direct heritability estimates from age-constant analyses. Fernandes et al. (2004) concluded that both weight- and age-adjusted ultrasound measurements were useful because they provide information on proportionality of the carcass and growth rate, respectively. Mortimer et al. (2014) found that adjusting ultrasound trait measurements to a constant weight and age reduced phenotypic variation, resulting in higher direct heritability estimates from the weight- and age-adjusted analyses when compared to age-adjusted analyses. Although the genetic parameter estimates must be interpreted cautiously due to the small sample size in the present research, these results suggest that the covariate used in ultrasound genetic evaluations could have an impact on their usefulness as indicators of carcass yield and quality. Carcass traits at alternative slaughter endpoints. There were considerable differences in the phenotypic variance of some traits when observations were adjusted to alternative slaughter endpoints (Table 4). Reduced phenotypic variation could make it more difficult to identify superior individuals, thus, the phenotypic variance of traits may be of interest in determining appropriate slaughter endpoint(s) for carcass trait genetic evaluations. Pollott et al. (1994) also noted that traits highly dependent on a particular endpoint were less variable than when adjusted to the other endpoints, for example, conformation traits that were adjusted to a constant fatness. Phenotypic variance for most traits was the highest when observations were adjusted to a constant SAGE, in agreement with the results reported by Pollott et al. (1994). This trend makes sense given that animals often differ in growth rate and maturation, likely leading to greater variability in carcass traits at a constant age. Table 4. Phenotypic variance (σp2) ⁠, ratio of contemporary group effect variance to phenotypic variance (g2), and direct heritability (hd2) estimates for carcass traits at alternative slaughter endpoints1 Slaughter endpoint Slaughter age Carcass weight Carcass fatness Trait2 σp2 g2 hd2 σp2 g2 hd2 σp2 g2 hd2 HCW 4.87 0.33 ± 0.02 0.17 ± 0.02 − − − 4.54 0.35 ± 0.02 0.18 ± 0.02 FATGR 9.54 0.21 ± 0.01 0.30 ± 0.02 8.44 0.19 ± 0.01 0.33 ± 0.03 − − − SMY 2.13 0.17 ± 0.01 0.26 ± 0.02 1.96 0.15 ± 0.01 0.28 ± 0.02 0.25 0.11 ± 0.01 0.26 ± 0.02 LEG 0.25 0.12 ± 0.01 0.32 ± 0.02 0.24 0.12 ± 0.01 0.33 ± 0.02 0.23 0.10 ± 0.01 0.30 ± 0.02 LOIN 0.28 0.17 ± 0.01 0.27 ± 0.02 0.27 0.16 ± 0.01 0.29 ± 0.02 0.24 0.12 ± 0.01 0.24 ± 0.02 SHOUL 0.35 0.14 ± 0.01 0.31 ± 0.02 0.34 0.14 ± 0.01 0.32 ± 0.02 0.30 0.11 ± 0.01 0.28 ± 0.02 AVGCONF 0.26 0.14 ± 0.01 0.32 ± 0.02 0.24 0.13 ± 0.01 0.34 ± 0.02 0.23 0.10 ± 0.01 0.30 ± 0.02 CINDEX 11.72 0.08 ± 0.01 0.23 ± 0.02 11.31 0.07 ± 0.01 0.22 ± 0.02 8.27 0.09 ± 0.01 0.22 ± 0.02 PRICE 236.91 0.28 ± 0.01 0.19 ± 0.02 53.36 0.10 ± 0.01 0.20 ± 0.02 243.45 0.29 ± 0.01 0.18 ± 0.02 Slaughter endpoint Slaughter age Carcass weight Carcass fatness Trait2 σp2 g2 hd2 σp2 g2 hd2 σp2 g2 hd2 HCW 4.87 0.33 ± 0.02 0.17 ± 0.02 − − − 4.54 0.35 ± 0.02 0.18 ± 0.02 FATGR 9.54 0.21 ± 0.01 0.30 ± 0.02 8.44 0.19 ± 0.01 0.33 ± 0.03 − − − SMY 2.13 0.17 ± 0.01 0.26 ± 0.02 1.96 0.15 ± 0.01 0.28 ± 0.02 0.25 0.11 ± 0.01 0.26 ± 0.02 LEG 0.25 0.12 ± 0.01 0.32 ± 0.02 0.24 0.12 ± 0.01 0.33 ± 0.02 0.23 0.10 ± 0.01 0.30 ± 0.02 LOIN 0.28 0.17 ± 0.01 0.27 ± 0.02 0.27 0.16 ± 0.01 0.29 ± 0.02 0.24 0.12 ± 0.01 0.24 ± 0.02 SHOUL 0.35 0.14 ± 0.01 0.31 ± 0.02 0.34 0.14 ± 0.01 0.32 ± 0.02 0.30 0.11 ± 0.01 0.28 ± 0.02 AVGCONF 0.26 0.14 ± 0.01 0.32 ± 0.02 0.24 0.13 ± 0.01 0.34 ± 0.02 0.23 0.10 ± 0.01 0.30 ± 0.02 CINDEX 11.72 0.08 ± 0.01 0.23 ± 0.02 11.31 0.07 ± 0.01 0.22 ± 0.02 8.27 0.09 ± 0.01 0.22 ± 0.02 PRICE 236.91 0.28 ± 0.01 0.19 ± 0.02 53.36 0.10 ± 0.01 0.20 ± 0.02 243.45 0.29 ± 0.01 0.18 ± 0.02 1Parameter estimates are followed by their approximate standard error. 2AVGCONF = average carcass conformation score; CINDEX = carcass price grid value; FATGR = fat depth at the GR site; HCW = hot carcass weight; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield. View Large Table 4. Phenotypic variance (σp2) ⁠, ratio of contemporary group effect variance to phenotypic variance (g2), and direct heritability (hd2) estimates for carcass traits at alternative slaughter endpoints1 Slaughter endpoint Slaughter age Carcass weight Carcass fatness Trait2 σp2 g2 hd2 σp2 g2 hd2 σp2 g2 hd2 HCW 4.87 0.33 ± 0.02 0.17 ± 0.02 − − − 4.54 0.35 ± 0.02 0.18 ± 0.02 FATGR 9.54 0.21 ± 0.01 0.30 ± 0.02 8.44 0.19 ± 0.01 0.33 ± 0.03 − − − SMY 2.13 0.17 ± 0.01 0.26 ± 0.02 1.96 0.15 ± 0.01 0.28 ± 0.02 0.25 0.11 ± 0.01 0.26 ± 0.02 LEG 0.25 0.12 ± 0.01 0.32 ± 0.02 0.24 0.12 ± 0.01 0.33 ± 0.02 0.23 0.10 ± 0.01 0.30 ± 0.02 LOIN 0.28 0.17 ± 0.01 0.27 ± 0.02 0.27 0.16 ± 0.01 0.29 ± 0.02 0.24 0.12 ± 0.01 0.24 ± 0.02 SHOUL 0.35 0.14 ± 0.01 0.31 ± 0.02 0.34 0.14 ± 0.01 0.32 ± 0.02 0.30 0.11 ± 0.01 0.28 ± 0.02 AVGCONF 0.26 0.14 ± 0.01 0.32 ± 0.02 0.24 0.13 ± 0.01 0.34 ± 0.02 0.23 0.10 ± 0.01 0.30 ± 0.02 CINDEX 11.72 0.08 ± 0.01 0.23 ± 0.02 11.31 0.07 ± 0.01 0.22 ± 0.02 8.27 0.09 ± 0.01 0.22 ± 0.02 PRICE 236.91 0.28 ± 0.01 0.19 ± 0.02 53.36 0.10 ± 0.01 0.20 ± 0.02 243.45 0.29 ± 0.01 0.18 ± 0.02 Slaughter endpoint Slaughter age Carcass weight Carcass fatness Trait2 σp2 g2 hd2 σp2 g2 hd2 σp2 g2 hd2 HCW 4.87 0.33 ± 0.02 0.17 ± 0.02 − − − 4.54 0.35 ± 0.02 0.18 ± 0.02 FATGR 9.54 0.21 ± 0.01 0.30 ± 0.02 8.44 0.19 ± 0.01 0.33 ± 0.03 − − − SMY 2.13 0.17 ± 0.01 0.26 ± 0.02 1.96 0.15 ± 0.01 0.28 ± 0.02 0.25 0.11 ± 0.01 0.26 ± 0.02 LEG 0.25 0.12 ± 0.01 0.32 ± 0.02 0.24 0.12 ± 0.01 0.33 ± 0.02 0.23 0.10 ± 0.01 0.30 ± 0.02 LOIN 0.28 0.17 ± 0.01 0.27 ± 0.02 0.27 0.16 ± 0.01 0.29 ± 0.02 0.24 0.12 ± 0.01 0.24 ± 0.02 SHOUL 0.35 0.14 ± 0.01 0.31 ± 0.02 0.34 0.14 ± 0.01 0.32 ± 0.02 0.30 0.11 ± 0.01 0.28 ± 0.02 AVGCONF 0.26 0.14 ± 0.01 0.32 ± 0.02 0.24 0.13 ± 0.01 0.34 ± 0.02 0.23 0.10 ± 0.01 0.30 ± 0.02 CINDEX 11.72 0.08 ± 0.01 0.23 ± 0.02 11.31 0.07 ± 0.01 0.22 ± 0.02 8.27 0.09 ± 0.01 0.22 ± 0.02 PRICE 236.91 0.28 ± 0.01 0.19 ± 0.02 53.36 0.10 ± 0.01 0.20 ± 0.02 243.45 0.29 ± 0.01 0.18 ± 0.02 1Parameter estimates are followed by their approximate standard error. 2AVGCONF = average carcass conformation score; CINDEX = carcass price grid value; FATGR = fat depth at the GR site; HCW = hot carcass weight; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield. View Large Direct heritability estimates for carcass traits ranged from 0.17 ± 0.02 for HCW at a constant SAGE to 0.34 ± 0.02 for AVGCONF at a constant carcass weight (Table 4). Heritability estimates generally differed by less than the standard errors of the estimates between alternative slaughter endpoints with the exception of FATGR and the carcass conformation traits where differences between 0.03 to 0.05 were observed between analyses. There was a trend for direct heritability estimates at a constant fatness and constant carcass weight to be the lowest and the highest, respectively. The lower direct heritability estimates for carcass traits at a constant fatness could indicate that adjusting for fatness removes more of the additive genetic variation in the traits compared to the other two endpoints. These trends were in agreement with results reported by Pollott et al. (1994) and Conington et al. (1998); however, Pollott et al. (1994) generally reported larger differences in direct heritability estimates between alternative slaughter endpoints. Overall, the small differences in the direct heritability estimates of the carcass traits at alternative slaughter endpoints are likely to be of little practical significance. The direct heritability of HCW was estimated to be 0.17 ± 0.02 and 0.18 ± 0.02 when observations were adjusted to a constant age and fatness, respectively (Table 4). Published direct heritability estimates for HCW adjusted to a constant SAGE have ranged from 0.20 ± 0.05 (Payne et al., 2009) to 0.59 in a multi-breed Australian Merino and Terminal crossbred population (Daetwyler et al., 2012). Carcass fat depth was found to be highly heritable (0.30 ± 0.02 to 0.33 ± 0.03; Table 4), in agreement with most literature estimates (Mortimer et al., 2010; Brito et al., 2017). Direct heritability estimates for FATGR adjusted to a constant carcass weight have ranged from 0.20 ± 0.02 (Brito et al., 2017) in a New Zealand Terminal crossbred population to 0.47 ± 0.08 (Ingham et al., 2007) in the literature. Direct heritabilities estimated with age and weight (Mortimer et al., 2010) or age (Mortimer et al., 2017b) slaughter endpoints have also been reported in Australian multi-breed and Merino populations (0.50 ± 0.05 and 0.23 ± 0.11, respectively). Direct heritability estimates for carcass conformation traits ranged from 0.24 ± 0.02 to 0.34 ± 0.02 (Table 4) and were consistent with published estimates, which ranged from 0.14 ± 0.05 (Karamichou et al., 2007) to 0.45 ± 0.05 (Maxa et al., 2007b), assessed on a range of European sheep breeds graded using the EUROP classification system (Meat and Livestock Commercial Services Ltd., n.d.). The composite traits (SMY, CINDEX, and PRICE) were all found to be moderately heritable (0.18 ± 0.02 to 0.28 ± 0.02; Table 4). Published direct heritability estimates have ranged from 0.19 ± 0.06 (Karamichou et al., 2007) to 0.23 ± 0.10 in Scottish Blackface sheep (Conington et al. 1998) for PRICE with observations adjusted to a constant age, and 0.32 ± 0.12 for observations adjusted to a constant fatness (Conington et al. 1998). Published direct heritability estimates for SMY ranged from 0.29 ± 0.11 (Mortimer et al., 2017b) to 0.32 (Daetwyler et al., 2012) and are similar to the estimates presented here (0.26 ± 0.02 to 0.28 ± 0.02). Genetic Correlations Growth traits. Consistent with literature estimates, body weight traits were all moderately to strongly correlated (0.46 ± 0.04 to 0.89 ± 0.01; Table 5), with weights measured at closer ages having stronger phenotypic and genetic correlations than those measured at more distant ages (Fischer et al., 2006; Ingham et al., 2007; Huisman and Brown, 2008; Boareki, 2017). Positive genetic correlations between WWT, PWWT, and WTUS (0.82 ± 0.06 to 0.89 ± 0.01; Table 5) were favorable, as selection for these traits would be expected to increase production efficiency. However, it is well reported that the positive genetic correlations between BWT and other weight traits are concerning due to the potential for reduced lambing ease or number of lambs born as BWT increases (Brown, 2007; Li and Brown, 2016; Boareki, 2017). Quinton et al. (2014) and Boareki (2017) concluded that utilizing selection indexes is important to balance the effects of selection for reproductive and growth traits. Where estimable, moderate to strong positive genetic correlations were generally observed between body weight traits and pre- and post-weaning average daily gain (0.27 ± 0.05 to 0.89 ± 0.01; Table 5), although the correlations were weaker than those reported by Maximini et al. (2012) between WTUS and average daily gain (0.96 ± 0.01). Table 5. Genetic and phenotypic correlations among growth and ultrasound traits1,2 BWT WWT PWWT WTUS ADG50 ADG100 EMDUSa EMDUSw FATUSa FATUSw BWT 0.36 ± 0.01 0.31 ± 0.01 0.28 ± 0.03 0.35 ± 0.01 0.10 ± 0.01 0.15 ± 0.04 −0.07 ± 0.04 0.13 ± 0.04 −0.02 ± 0.04 WWT 0.50 ± 0.04 0.77 ± 0.01 0.71 ± 0.02 NC3 0.11 ± 0.01 0.42 ± 0.03 −0.11 ± 0.04 0.39 ± 0.03 0.04 ± 0.04 PWWT 0.46 ± 0.04 0.89 ± 0.01 NC 0.75 ± 0.01 0.69 ± 0.01 0.64 ± 0.02 −0.01 ± 0.04 0.52 ± 0.03 0.03 ± 0.05 WTUS 0.57 ± 0.12 0.82 ± 0.06 NC 0.69 ± 0.02 0.62 ± 0.02 0.66 ± 0.02 −0.02 ± 0.04 0.52 ± 0.03 −0.05 ± 0.05 ADG50 0.49 ± 0.04 NC 0.89 ± 0.01 0.77 ± 0.06 0.11 ± 0.01 0.41 ± 0.03 −0.11 ± 0.04 0.38 ± 0.03 0.03 ± 0.04 ADG100 0.27 ± 0.05 0.49 ± 0.04 0.82 ± 0.02 0.56 ± 0.09 0.51 ± 0.04 0.44 ± 0.03 0.00 ± 0.04 0.32 ± 0.04 −0.01 ± 0.05 EMDUSa 0.38 ± 0.18 0.36 ± 0.12 0.54 ± 0.08 0.40 ± 0.14 0.34 ± 0.13 0.22 ± 0.14 NC 0.41 ± 0.03 0.10 ± 0.04 EMDUSw −0.17 ± 0.13 −0.41 ± 0.11 −0.28 ± 0.11 −0.47 ± 0.15 −0.37 ± 0.11 −0.25 ± 0.11 NC 0.06 ± 0.04 0.11 ± 0.04 FATUSa 0.21 ± 0.15 0.47 ± 0.10 0.63 ± 0.08 0.58 ± 0.10 0.40 ± 0.10 0.25 ± 0.12 0.42 ± 0.16 −0.13 ± 0.16 NC FATUSw −0.16 ± 0.14 0.02 ± 0.12 −0.02 ± 0.11 −0.07 ± 0.15 −0.03 ± 0.12 −0.03 ± 0.12 0.29 ± 0.19 0.32 ± 0.13 NC BWT WWT PWWT WTUS ADG50 ADG100 EMDUSa EMDUSw FATUSa FATUSw BWT 0.36 ± 0.01 0.31 ± 0.01 0.28 ± 0.03 0.35 ± 0.01 0.10 ± 0.01 0.15 ± 0.04 −0.07 ± 0.04 0.13 ± 0.04 −0.02 ± 0.04 WWT 0.50 ± 0.04 0.77 ± 0.01 0.71 ± 0.02 NC3 0.11 ± 0.01 0.42 ± 0.03 −0.11 ± 0.04 0.39 ± 0.03 0.04 ± 0.04 PWWT 0.46 ± 0.04 0.89 ± 0.01 NC 0.75 ± 0.01 0.69 ± 0.01 0.64 ± 0.02 −0.01 ± 0.04 0.52 ± 0.03 0.03 ± 0.05 WTUS 0.57 ± 0.12 0.82 ± 0.06 NC 0.69 ± 0.02 0.62 ± 0.02 0.66 ± 0.02 −0.02 ± 0.04 0.52 ± 0.03 −0.05 ± 0.05 ADG50 0.49 ± 0.04 NC 0.89 ± 0.01 0.77 ± 0.06 0.11 ± 0.01 0.41 ± 0.03 −0.11 ± 0.04 0.38 ± 0.03 0.03 ± 0.04 ADG100 0.27 ± 0.05 0.49 ± 0.04 0.82 ± 0.02 0.56 ± 0.09 0.51 ± 0.04 0.44 ± 0.03 0.00 ± 0.04 0.32 ± 0.04 −0.01 ± 0.05 EMDUSa 0.38 ± 0.18 0.36 ± 0.12 0.54 ± 0.08 0.40 ± 0.14 0.34 ± 0.13 0.22 ± 0.14 NC 0.41 ± 0.03 0.10 ± 0.04 EMDUSw −0.17 ± 0.13 −0.41 ± 0.11 −0.28 ± 0.11 −0.47 ± 0.15 −0.37 ± 0.11 −0.25 ± 0.11 NC 0.06 ± 0.04 0.11 ± 0.04 FATUSa 0.21 ± 0.15 0.47 ± 0.10 0.63 ± 0.08 0.58 ± 0.10 0.40 ± 0.10 0.25 ± 0.12 0.42 ± 0.16 −0.13 ± 0.16 NC FATUSw −0.16 ± 0.14 0.02 ± 0.12 −0.02 ± 0.11 −0.07 ± 0.15 −0.03 ± 0.12 −0.03 ± 0.12 0.29 ± 0.19 0.32 ± 0.13 NC 1Phenotypic and genetic correlation estimates are presented above and below the diagonal, respectively, and are followed by their approximate standard error. 2ADG50 = pre-weaning average daily gain; ADG100 = post-weaning average daily gain; BWT = birthweight; EMDUS = ultrasonically measured eye muscle depth; FATUS = ultrasonically measured fat depth; PWWT = post-weaning weight; WTUS = scanning weight; WWT = weaning weight. Ultrasound eye muscle and fat depth measurements adjusted to a constant age (EMDUSa, FATUSa) or scanning weight (EMDUSw, FATUSw). 3NC: Model did not converge. View Large Table 5. Genetic and phenotypic correlations among growth and ultrasound traits1,2 BWT WWT PWWT WTUS ADG50 ADG100 EMDUSa EMDUSw FATUSa FATUSw BWT 0.36 ± 0.01 0.31 ± 0.01 0.28 ± 0.03 0.35 ± 0.01 0.10 ± 0.01 0.15 ± 0.04 −0.07 ± 0.04 0.13 ± 0.04 −0.02 ± 0.04 WWT 0.50 ± 0.04 0.77 ± 0.01 0.71 ± 0.02 NC3 0.11 ± 0.01 0.42 ± 0.03 −0.11 ± 0.04 0.39 ± 0.03 0.04 ± 0.04 PWWT 0.46 ± 0.04 0.89 ± 0.01 NC 0.75 ± 0.01 0.69 ± 0.01 0.64 ± 0.02 −0.01 ± 0.04 0.52 ± 0.03 0.03 ± 0.05 WTUS 0.57 ± 0.12 0.82 ± 0.06 NC 0.69 ± 0.02 0.62 ± 0.02 0.66 ± 0.02 −0.02 ± 0.04 0.52 ± 0.03 −0.05 ± 0.05 ADG50 0.49 ± 0.04 NC 0.89 ± 0.01 0.77 ± 0.06 0.11 ± 0.01 0.41 ± 0.03 −0.11 ± 0.04 0.38 ± 0.03 0.03 ± 0.04 ADG100 0.27 ± 0.05 0.49 ± 0.04 0.82 ± 0.02 0.56 ± 0.09 0.51 ± 0.04 0.44 ± 0.03 0.00 ± 0.04 0.32 ± 0.04 −0.01 ± 0.05 EMDUSa 0.38 ± 0.18 0.36 ± 0.12 0.54 ± 0.08 0.40 ± 0.14 0.34 ± 0.13 0.22 ± 0.14 NC 0.41 ± 0.03 0.10 ± 0.04 EMDUSw −0.17 ± 0.13 −0.41 ± 0.11 −0.28 ± 0.11 −0.47 ± 0.15 −0.37 ± 0.11 −0.25 ± 0.11 NC 0.06 ± 0.04 0.11 ± 0.04 FATUSa 0.21 ± 0.15 0.47 ± 0.10 0.63 ± 0.08 0.58 ± 0.10 0.40 ± 0.10 0.25 ± 0.12 0.42 ± 0.16 −0.13 ± 0.16 NC FATUSw −0.16 ± 0.14 0.02 ± 0.12 −0.02 ± 0.11 −0.07 ± 0.15 −0.03 ± 0.12 −0.03 ± 0.12 0.29 ± 0.19 0.32 ± 0.13 NC BWT WWT PWWT WTUS ADG50 ADG100 EMDUSa EMDUSw FATUSa FATUSw BWT 0.36 ± 0.01 0.31 ± 0.01 0.28 ± 0.03 0.35 ± 0.01 0.10 ± 0.01 0.15 ± 0.04 −0.07 ± 0.04 0.13 ± 0.04 −0.02 ± 0.04 WWT 0.50 ± 0.04 0.77 ± 0.01 0.71 ± 0.02 NC3 0.11 ± 0.01 0.42 ± 0.03 −0.11 ± 0.04 0.39 ± 0.03 0.04 ± 0.04 PWWT 0.46 ± 0.04 0.89 ± 0.01 NC 0.75 ± 0.01 0.69 ± 0.01 0.64 ± 0.02 −0.01 ± 0.04 0.52 ± 0.03 0.03 ± 0.05 WTUS 0.57 ± 0.12 0.82 ± 0.06 NC 0.69 ± 0.02 0.62 ± 0.02 0.66 ± 0.02 −0.02 ± 0.04 0.52 ± 0.03 −0.05 ± 0.05 ADG50 0.49 ± 0.04 NC 0.89 ± 0.01 0.77 ± 0.06 0.11 ± 0.01 0.41 ± 0.03 −0.11 ± 0.04 0.38 ± 0.03 0.03 ± 0.04 ADG100 0.27 ± 0.05 0.49 ± 0.04 0.82 ± 0.02 0.56 ± 0.09 0.51 ± 0.04 0.44 ± 0.03 0.00 ± 0.04 0.32 ± 0.04 −0.01 ± 0.05 EMDUSa 0.38 ± 0.18 0.36 ± 0.12 0.54 ± 0.08 0.40 ± 0.14 0.34 ± 0.13 0.22 ± 0.14 NC 0.41 ± 0.03 0.10 ± 0.04 EMDUSw −0.17 ± 0.13 −0.41 ± 0.11 −0.28 ± 0.11 −0.47 ± 0.15 −0.37 ± 0.11 −0.25 ± 0.11 NC 0.06 ± 0.04 0.11 ± 0.04 FATUSa 0.21 ± 0.15 0.47 ± 0.10 0.63 ± 0.08 0.58 ± 0.10 0.40 ± 0.10 0.25 ± 0.12 0.42 ± 0.16 −0.13 ± 0.16 NC FATUSw −0.16 ± 0.14 0.02 ± 0.12 −0.02 ± 0.11 −0.07 ± 0.15 −0.03 ± 0.12 −0.03 ± 0.12 0.29 ± 0.19 0.32 ± 0.13 NC 1Phenotypic and genetic correlation estimates are presented above and below the diagonal, respectively, and are followed by their approximate standard error. 2ADG50 = pre-weaning average daily gain; ADG100 = post-weaning average daily gain; BWT = birthweight; EMDUS = ultrasonically measured eye muscle depth; FATUS = ultrasonically measured fat depth; PWWT = post-weaning weight; WTUS = scanning weight; WWT = weaning weight. Ultrasound eye muscle and fat depth measurements adjusted to a constant age (EMDUSa, FATUSa) or scanning weight (EMDUSw, FATUSw). 3NC: Model did not converge. View Large Ultrasound traits. Genetic correlation estimates among EMDUS and FATUS adjusted to either age- or weight-constant endpoints (−0.13 ± 0.16 to 0.42 ± 0.16; Table 5) were generally positive and weak to moderate in magnitude with large standard errors, suggesting that selection on EMDUS to improve carcass conformation may result in unfavorable increases in FATUS. These results contrast with those described by Fernandes et al. (2004), who found a weak negative correlation between the traits when adjusted to a constant weight (−0.17). Carcass traits. An unfavorable weak positive genetic correlation between HCW and FATGR at a constant age was identified (0.15 ± 0.07; Table 6), although this correlation was weaker than some literature estimates (Ingham et al., 2007; Brito et al., 2017). Hot carcass weight had a strong positive genetic correlation with PRICE at both age and fatness endpoints (0.87 ± 0.02 to 0.90 ± 0.02; Tables 6 and 8), whereas FATGR and PRICE had favorable weak to moderate (−0.53 ± 0.05 to −0.17 ± 0.07; Tables 6 and 8) negative correlations at both age and fatness endpoints. These results were in agreement with those reported by Karamichou et al. (2007) and indicate that total price was more strongly associated with carcass yield than carcass quality. Thus, genetic selection to increase HCW would be expected to favorably increase carcass value independent of the endpoint used in the analysis. Among the carcass conformation traits, moderate to strong positive genetic correlations were observed at all endpoints (0.54 ± 0.05 to 0.89 ± 0.02; Tables 6 to 8), suggesting that selection for AVGCONF could be beneficial to simultaneously improve all primal cut conformation traits. Unfavorable moderate positive genetic correlations between FATGR and carcass conformation scores at age and weight endpoints (0.31 ± 0.05 to 0.60 ± 0.05; Tables 6 and 7) indicate that selection to reduce FATGR will reduce muscularity. Similarly, Einarsson et al. (2015) identified a moderate positive genetic correlation (0.38) between FATGR and EUROP carcass conformation score. Adjusting AVGCONF to a constant weight or age resulted in weak unfavorable negative correlations (−0.18 ± 0.06 and −0.19 ± 0.06, respectively) with SMY, while at a constant fatness, the traits were strongly positively correlated (0.96 ± 0.01), due to the fact that SMY is predicted based on AVGCONF and FATGR (Tables 6 to 8). The endpoint used in the analysis was found to have an influence on both the magnitude and direction of some genetic correlations. Furthermore, the unfavorable correlations observed between some traits underscore the importance of using selection indexes for balanced carcass trait genetic improvement. Table 6. Phenotypic and genetic correlation estimates for carcass traits adjusted to a fixed slaughter age endpoint1,2 HCW FATGR SMY LEG LOIN SHOUL AVGCONF CINDEX PRICE HCW 0.35 ± 0.01 −0.30 ± 0.01 0.23 ± 0.01 0.23 ± 0.01 0.18 ± 0.01 0.22 ± 0.01 −0.19 ± 0.01 0.88 ± 0.01 FATGR 0.15 ± 0.07 −0.94 ± 0.01 0.24 ± 0.01 0.39 ± 0.01 0.37 ± 0.01 0.35 ± 0.01 −0.54 ± 0.01 0.13 ± 0.01 SMY −0.15 ± 0.07 −0.94 ± 0.01 −0.02 ± 0.01 −0.21 ± 0.01 −0.13 ± 0.01 −0.09 ± 0.01 0.62 ± 0.01 −0.04 ± 0.01 LEG 0.03 ± 0.07 0.31 ± 0.05 −0.03 ± 0.06 0.36 ± 0.01 0.38 ± 0.01 0.62 ± 0.01 0.10 ± 0.01 0.27 ± 0.01 LOIN 0.00 ± 0.07 0.58 ± 0.05 −0.38 ± 0.06 0.57 ± 0.05 0.45 ± 0.01 0.57 ± 0.01 −0.01 ± 0.01 0.22 ± 0.01 SHOUL 0.08 ± 0.07 0.53 ± 0.05 −0.26 ± 0.06 0.63 ± 0.04 0.65 ± 0.04 0.77 ± 0.01 0.02 ± 0.01 0.19 ± 0.01 AVGCONF 0.04 ± 0.07 0.48 ± 0.05 −0.19 ± 0.06 0.83 ± 0.02 0.75 ± 0.03 0.89 ± 0.02 0.10 ± 0.01 0.26 ± 0.01 CINDEX −0.24 ± 0.08 −0.59 ± 0.05 0.69 ± 0.04 0.08 ± 0.07 −0.07 ± 0.07 −0.04 ± 0.07 0.01 ± 0.07 0.23 ± 0.01 PRICE 0.87 ± 0.02 −0.17 ± 0.07 0.18 ± 0.07 0.04 ± 0.07 −0.06 ± 0.07 0.01 ± 0.07 0.00 ± 0.07 0.24 ± 0.07 HCW FATGR SMY LEG LOIN SHOUL AVGCONF CINDEX PRICE HCW 0.35 ± 0.01 −0.30 ± 0.01 0.23 ± 0.01 0.23 ± 0.01 0.18 ± 0.01 0.22 ± 0.01 −0.19 ± 0.01 0.88 ± 0.01 FATGR 0.15 ± 0.07 −0.94 ± 0.01 0.24 ± 0.01 0.39 ± 0.01 0.37 ± 0.01 0.35 ± 0.01 −0.54 ± 0.01 0.13 ± 0.01 SMY −0.15 ± 0.07 −0.94 ± 0.01 −0.02 ± 0.01 −0.21 ± 0.01 −0.13 ± 0.01 −0.09 ± 0.01 0.62 ± 0.01 −0.04 ± 0.01 LEG 0.03 ± 0.07 0.31 ± 0.05 −0.03 ± 0.06 0.36 ± 0.01 0.38 ± 0.01 0.62 ± 0.01 0.10 ± 0.01 0.27 ± 0.01 LOIN 0.00 ± 0.07 0.58 ± 0.05 −0.38 ± 0.06 0.57 ± 0.05 0.45 ± 0.01 0.57 ± 0.01 −0.01 ± 0.01 0.22 ± 0.01 SHOUL 0.08 ± 0.07 0.53 ± 0.05 −0.26 ± 0.06 0.63 ± 0.04 0.65 ± 0.04 0.77 ± 0.01 0.02 ± 0.01 0.19 ± 0.01 AVGCONF 0.04 ± 0.07 0.48 ± 0.05 −0.19 ± 0.06 0.83 ± 0.02 0.75 ± 0.03 0.89 ± 0.02 0.10 ± 0.01 0.26 ± 0.01 CINDEX −0.24 ± 0.08 −0.59 ± 0.05 0.69 ± 0.04 0.08 ± 0.07 −0.07 ± 0.07 −0.04 ± 0.07 0.01 ± 0.07 0.23 ± 0.01 PRICE 0.87 ± 0.02 −0.17 ± 0.07 0.18 ± 0.07 0.04 ± 0.07 −0.06 ± 0.07 0.01 ± 0.07 0.00 ± 0.07 0.24 ± 0.07 1Phenotypic and genetic correlation estimates are presented above and below the diagonal, respectively, and are followed by their approximate standard error. 2AVGCONF = average carcass conformation score; CINDEX = carcass price grid value; FATGR = fat depth at the GR site; HCW = hot carcass weight; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield. View Large Table 6. Phenotypic and genetic correlation estimates for carcass traits adjusted to a fixed slaughter age endpoint1,2 HCW FATGR SMY LEG LOIN SHOUL AVGCONF CINDEX PRICE HCW 0.35 ± 0.01 −0.30 ± 0.01 0.23 ± 0.01 0.23 ± 0.01 0.18 ± 0.01 0.22 ± 0.01 −0.19 ± 0.01 0.88 ± 0.01 FATGR 0.15 ± 0.07 −0.94 ± 0.01 0.24 ± 0.01 0.39 ± 0.01 0.37 ± 0.01 0.35 ± 0.01 −0.54 ± 0.01 0.13 ± 0.01 SMY −0.15 ± 0.07 −0.94 ± 0.01 −0.02 ± 0.01 −0.21 ± 0.01 −0.13 ± 0.01 −0.09 ± 0.01 0.62 ± 0.01 −0.04 ± 0.01 LEG 0.03 ± 0.07 0.31 ± 0.05 −0.03 ± 0.06 0.36 ± 0.01 0.38 ± 0.01 0.62 ± 0.01 0.10 ± 0.01 0.27 ± 0.01 LOIN 0.00 ± 0.07 0.58 ± 0.05 −0.38 ± 0.06 0.57 ± 0.05 0.45 ± 0.01 0.57 ± 0.01 −0.01 ± 0.01 0.22 ± 0.01 SHOUL 0.08 ± 0.07 0.53 ± 0.05 −0.26 ± 0.06 0.63 ± 0.04 0.65 ± 0.04 0.77 ± 0.01 0.02 ± 0.01 0.19 ± 0.01 AVGCONF 0.04 ± 0.07 0.48 ± 0.05 −0.19 ± 0.06 0.83 ± 0.02 0.75 ± 0.03 0.89 ± 0.02 0.10 ± 0.01 0.26 ± 0.01 CINDEX −0.24 ± 0.08 −0.59 ± 0.05 0.69 ± 0.04 0.08 ± 0.07 −0.07 ± 0.07 −0.04 ± 0.07 0.01 ± 0.07 0.23 ± 0.01 PRICE 0.87 ± 0.02 −0.17 ± 0.07 0.18 ± 0.07 0.04 ± 0.07 −0.06 ± 0.07 0.01 ± 0.07 0.00 ± 0.07 0.24 ± 0.07 HCW FATGR SMY LEG LOIN SHOUL AVGCONF CINDEX PRICE HCW 0.35 ± 0.01 −0.30 ± 0.01 0.23 ± 0.01 0.23 ± 0.01 0.18 ± 0.01 0.22 ± 0.01 −0.19 ± 0.01 0.88 ± 0.01 FATGR 0.15 ± 0.07 −0.94 ± 0.01 0.24 ± 0.01 0.39 ± 0.01 0.37 ± 0.01 0.35 ± 0.01 −0.54 ± 0.01 0.13 ± 0.01 SMY −0.15 ± 0.07 −0.94 ± 0.01 −0.02 ± 0.01 −0.21 ± 0.01 −0.13 ± 0.01 −0.09 ± 0.01 0.62 ± 0.01 −0.04 ± 0.01 LEG 0.03 ± 0.07 0.31 ± 0.05 −0.03 ± 0.06 0.36 ± 0.01 0.38 ± 0.01 0.62 ± 0.01 0.10 ± 0.01 0.27 ± 0.01 LOIN 0.00 ± 0.07 0.58 ± 0.05 −0.38 ± 0.06 0.57 ± 0.05 0.45 ± 0.01 0.57 ± 0.01 −0.01 ± 0.01 0.22 ± 0.01 SHOUL 0.08 ± 0.07 0.53 ± 0.05 −0.26 ± 0.06 0.63 ± 0.04 0.65 ± 0.04 0.77 ± 0.01 0.02 ± 0.01 0.19 ± 0.01 AVGCONF 0.04 ± 0.07 0.48 ± 0.05 −0.19 ± 0.06 0.83 ± 0.02 0.75 ± 0.03 0.89 ± 0.02 0.10 ± 0.01 0.26 ± 0.01 CINDEX −0.24 ± 0.08 −0.59 ± 0.05 0.69 ± 0.04 0.08 ± 0.07 −0.07 ± 0.07 −0.04 ± 0.07 0.01 ± 0.07 0.23 ± 0.01 PRICE 0.87 ± 0.02 −0.17 ± 0.07 0.18 ± 0.07 0.04 ± 0.07 −0.06 ± 0.07 0.01 ± 0.07 0.00 ± 0.07 0.24 ± 0.07 1Phenotypic and genetic correlation estimates are presented above and below the diagonal, respectively, and are followed by their approximate standard error. 2AVGCONF = average carcass conformation score; CINDEX = carcass price grid value; FATGR = fat depth at the GR site; HCW = hot carcass weight; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield. View Large Table 7. Phenotypic and genetic correlation estimates for carcass traits adjusted to a fixed carcass weight endpoint1,2 FATGR SMY LEG LOIN SHOUL AVGCONF CINDEX PRICE FATGR −0.93 ± 0.01 0.18 ± 0.01 0.35 ± 0.01 0.32 ± 0.01 0.30 ± 0.01 −0.52 ± 0.01 −0.42 ± 0.01 SMY −0.94 ± 0.01 0.05 ± 0.01 −0.16 ± 0.01 −0.08 ± 0.01 −0.03 ± 0.01 0.60 ± 0.01 0.48 ± 0.01 LEG 0.31 ± 0.05 −0.04 ± 0.06 0.33 ± 0.01 0.34 ± 0.01 0.60 ± 0.01 0.15 ± 0.01 0.13 ± 0.01 LOIN 0.60 ± 0.05 −0.40 ± 0.06 0.58 ± 0.05 0.42 ± 0.01 0.55 ± 0.01 0.03 ± 0.01 0.04 ± 0.01 SHOUL 0.51 ± 0.05 −0.24 ± 0.06 0.64 ± 0.04 0.64 ± 0.04 0.76 ± 0.01 0.06 ± 0.01 0.05 ± 0.01 AVGCONF 0.48 ± 0.05 −0.18 ± 0.06 0.84 ± 0.02 0.75 ± 0.03 0.89 ± 0.02 0.15 ± 0.01 0.13 ± 0.01 CINDEX −0.56 ± 0.05 0.67 ± 0.04 0.09 ± 0.06 −0.06 ± 0.07 −0.02 ± 0.07 0.02 ± 0.06 0.85 ± 0.01 PRICE −0.53 ± 0.05 0.58 ± 0.05 0.03 ± 0.07 −0.08 ± 0.07 −0.09 ± 0.07 −0.06 ± 0.07 0.93 ± 0.02 FATGR SMY LEG LOIN SHOUL AVGCONF CINDEX PRICE FATGR −0.93 ± 0.01 0.18 ± 0.01 0.35 ± 0.01 0.32 ± 0.01 0.30 ± 0.01 −0.52 ± 0.01 −0.42 ± 0.01 SMY −0.94 ± 0.01 0.05 ± 0.01 −0.16 ± 0.01 −0.08 ± 0.01 −0.03 ± 0.01 0.60 ± 0.01 0.48 ± 0.01 LEG 0.31 ± 0.05 −0.04 ± 0.06 0.33 ± 0.01 0.34 ± 0.01 0.60 ± 0.01 0.15 ± 0.01 0.13 ± 0.01 LOIN 0.60 ± 0.05 −0.40 ± 0.06 0.58 ± 0.05 0.42 ± 0.01 0.55 ± 0.01 0.03 ± 0.01 0.04 ± 0.01 SHOUL 0.51 ± 0.05 −0.24 ± 0.06 0.64 ± 0.04 0.64 ± 0.04 0.76 ± 0.01 0.06 ± 0.01 0.05 ± 0.01 AVGCONF 0.48 ± 0.05 −0.18 ± 0.06 0.84 ± 0.02 0.75 ± 0.03 0.89 ± 0.02 0.15 ± 0.01 0.13 ± 0.01 CINDEX −0.56 ± 0.05 0.67 ± 0.04 0.09 ± 0.06 −0.06 ± 0.07 −0.02 ± 0.07 0.02 ± 0.06 0.85 ± 0.01 PRICE −0.53 ± 0.05 0.58 ± 0.05 0.03 ± 0.07 −0.08 ± 0.07 −0.09 ± 0.07 −0.06 ± 0.07 0.93 ± 0.02 1Phenotypic and genetic correlation estimates are presented above and below the diagonal, respectively, and are followed by their approximate standard error. 2AVGCONF = average carcass conformation score; CINDEX = carcass price grid value; FATGR = fat depth at the GR site; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield. View Large Table 7. Phenotypic and genetic correlation estimates for carcass traits adjusted to a fixed carcass weight endpoint1,2 FATGR SMY LEG LOIN SHOUL AVGCONF CINDEX PRICE FATGR −0.93 ± 0.01 0.18 ± 0.01 0.35 ± 0.01 0.32 ± 0.01 0.30 ± 0.01 −0.52 ± 0.01 −0.42 ± 0.01 SMY −0.94 ± 0.01 0.05 ± 0.01 −0.16 ± 0.01 −0.08 ± 0.01 −0.03 ± 0.01 0.60 ± 0.01 0.48 ± 0.01 LEG 0.31 ± 0.05 −0.04 ± 0.06 0.33 ± 0.01 0.34 ± 0.01 0.60 ± 0.01 0.15 ± 0.01 0.13 ± 0.01 LOIN 0.60 ± 0.05 −0.40 ± 0.06 0.58 ± 0.05 0.42 ± 0.01 0.55 ± 0.01 0.03 ± 0.01 0.04 ± 0.01 SHOUL 0.51 ± 0.05 −0.24 ± 0.06 0.64 ± 0.04 0.64 ± 0.04 0.76 ± 0.01 0.06 ± 0.01 0.05 ± 0.01 AVGCONF 0.48 ± 0.05 −0.18 ± 0.06 0.84 ± 0.02 0.75 ± 0.03 0.89 ± 0.02 0.15 ± 0.01 0.13 ± 0.01 CINDEX −0.56 ± 0.05 0.67 ± 0.04 0.09 ± 0.06 −0.06 ± 0.07 −0.02 ± 0.07 0.02 ± 0.06 0.85 ± 0.01 PRICE −0.53 ± 0.05 0.58 ± 0.05 0.03 ± 0.07 −0.08 ± 0.07 −0.09 ± 0.07 −0.06 ± 0.07 0.93 ± 0.02 FATGR SMY LEG LOIN SHOUL AVGCONF CINDEX PRICE FATGR −0.93 ± 0.01 0.18 ± 0.01 0.35 ± 0.01 0.32 ± 0.01 0.30 ± 0.01 −0.52 ± 0.01 −0.42 ± 0.01 SMY −0.94 ± 0.01 0.05 ± 0.01 −0.16 ± 0.01 −0.08 ± 0.01 −0.03 ± 0.01 0.60 ± 0.01 0.48 ± 0.01 LEG 0.31 ± 0.05 −0.04 ± 0.06 0.33 ± 0.01 0.34 ± 0.01 0.60 ± 0.01 0.15 ± 0.01 0.13 ± 0.01 LOIN 0.60 ± 0.05 −0.40 ± 0.06 0.58 ± 0.05 0.42 ± 0.01 0.55 ± 0.01 0.03 ± 0.01 0.04 ± 0.01 SHOUL 0.51 ± 0.05 −0.24 ± 0.06 0.64 ± 0.04 0.64 ± 0.04 0.76 ± 0.01 0.06 ± 0.01 0.05 ± 0.01 AVGCONF 0.48 ± 0.05 −0.18 ± 0.06 0.84 ± 0.02 0.75 ± 0.03 0.89 ± 0.02 0.15 ± 0.01 0.13 ± 0.01 CINDEX −0.56 ± 0.05 0.67 ± 0.04 0.09 ± 0.06 −0.06 ± 0.07 −0.02 ± 0.07 0.02 ± 0.06 0.85 ± 0.01 PRICE −0.53 ± 0.05 0.58 ± 0.05 0.03 ± 0.07 −0.08 ± 0.07 −0.09 ± 0.07 −0.06 ± 0.07 0.93 ± 0.02 1Phenotypic and genetic correlation estimates are presented above and below the diagonal, respectively, and are followed by their approximate standard error. 2AVGCONF = average carcass conformation score; CINDEX = carcass price grid value; FATGR = fat depth at the GR site; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield. View Large Table 8. Phenotypic and genetic correlation estimates for carcass traits adjusted to a fixed carcass fatness endpoint1,2 HCW SMY LEG LOIN SHOUL AVGCONF CINDEX PRICE HCW 0.11 ± 0.01 0.14 ± 0.01 0.10 ± 0.01 0.08 ± 0.01 0.12 ± 0.01 −0.01 ± 0.01 0.91 ± 0.01 SMY −0.09 ± 0.07 0.62 ± 0.01 0.52 ± 0.01 0.67 ± 0.01 0.75 ± 0.01 0.36 ± 0.01 0.22 ± 0.01 LEG −0.03 ± 0.07 0.86 ± 0.02 0.30 ± 0.01 0.32 ± 0.01 0.59 ± 0.01 0.28 ± 0.01 0.23 ± 0.01 LOIN −0.18 ± 0.08 0.71 ± 0.04 0.54 ± 0.05 0.36 ± 0.01 0.50 ± 0.01 0.27 ± 0.01 0.18 ± 0.01 SHOUL −0.07 ± 0.07 0.86 ± 0.02 0.61 ± 0.05 0.54 ± 0.05 0.74 ± 0.01 0.29 ± 0.01 0.17 ± 0.01 AVGCONF −0.09 ± 0.07 0.96 ± 0.01 0.83 ± 0.03 0.68 ± 0.04 0.87 ± 0.02 0.38 ± 0.01 0.24 ± 0.01 CINDEX −0.20 ± 0.08 0.56 ± 0.06 0.42 ± 0.06 0.55 ± 0.06 0.47 ± 0.06 0.49 ± 0.06 0.34 ± 0.01 PRICE 0.90 ± 0.02 0.03 ± 0.07 0.07 ± 0.07 −0.03 ± 0.08 0.03 ± 0.07 0.02 ± 0.07 0.21 ± 0.08 HCW SMY LEG LOIN SHOUL AVGCONF CINDEX PRICE HCW 0.11 ± 0.01 0.14 ± 0.01 0.10 ± 0.01 0.08 ± 0.01 0.12 ± 0.01 −0.01 ± 0.01 0.91 ± 0.01 SMY −0.09 ± 0.07 0.62 ± 0.01 0.52 ± 0.01 0.67 ± 0.01 0.75 ± 0.01 0.36 ± 0.01 0.22 ± 0.01 LEG −0.03 ± 0.07 0.86 ± 0.02 0.30 ± 0.01 0.32 ± 0.01 0.59 ± 0.01 0.28 ± 0.01 0.23 ± 0.01 LOIN −0.18 ± 0.08 0.71 ± 0.04 0.54 ± 0.05 0.36 ± 0.01 0.50 ± 0.01 0.27 ± 0.01 0.18 ± 0.01 SHOUL −0.07 ± 0.07 0.86 ± 0.02 0.61 ± 0.05 0.54 ± 0.05 0.74 ± 0.01 0.29 ± 0.01 0.17 ± 0.01 AVGCONF −0.09 ± 0.07 0.96 ± 0.01 0.83 ± 0.03 0.68 ± 0.04 0.87 ± 0.02 0.38 ± 0.01 0.24 ± 0.01 CINDEX −0.20 ± 0.08 0.56 ± 0.06 0.42 ± 0.06 0.55 ± 0.06 0.47 ± 0.06 0.49 ± 0.06 0.34 ± 0.01 PRICE 0.90 ± 0.02 0.03 ± 0.07 0.07 ± 0.07 −0.03 ± 0.08 0.03 ± 0.07 0.02 ± 0.07 0.21 ± 0.08 1Phenotypic and genetic correlation estimates are presented above and below the diagonal, respectively, and are followed by their approximate standard error. 2AVGCONF = average carcass conformation score; CINDEX = carcass price grid value; HCW = hot carcass weight; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield. View Large Table 8. Phenotypic and genetic correlation estimates for carcass traits adjusted to a fixed carcass fatness endpoint1,2 HCW SMY LEG LOIN SHOUL AVGCONF CINDEX PRICE HCW 0.11 ± 0.01 0.14 ± 0.01 0.10 ± 0.01 0.08 ± 0.01 0.12 ± 0.01 −0.01 ± 0.01 0.91 ± 0.01 SMY −0.09 ± 0.07 0.62 ± 0.01 0.52 ± 0.01 0.67 ± 0.01 0.75 ± 0.01 0.36 ± 0.01 0.22 ± 0.01 LEG −0.03 ± 0.07 0.86 ± 0.02 0.30 ± 0.01 0.32 ± 0.01 0.59 ± 0.01 0.28 ± 0.01 0.23 ± 0.01 LOIN −0.18 ± 0.08 0.71 ± 0.04 0.54 ± 0.05 0.36 ± 0.01 0.50 ± 0.01 0.27 ± 0.01 0.18 ± 0.01 SHOUL −0.07 ± 0.07 0.86 ± 0.02 0.61 ± 0.05 0.54 ± 0.05 0.74 ± 0.01 0.29 ± 0.01 0.17 ± 0.01 AVGCONF −0.09 ± 0.07 0.96 ± 0.01 0.83 ± 0.03 0.68 ± 0.04 0.87 ± 0.02 0.38 ± 0.01 0.24 ± 0.01 CINDEX −0.20 ± 0.08 0.56 ± 0.06 0.42 ± 0.06 0.55 ± 0.06 0.47 ± 0.06 0.49 ± 0.06 0.34 ± 0.01 PRICE 0.90 ± 0.02 0.03 ± 0.07 0.07 ± 0.07 −0.03 ± 0.08 0.03 ± 0.07 0.02 ± 0.07 0.21 ± 0.08 HCW SMY LEG LOIN SHOUL AVGCONF CINDEX PRICE HCW 0.11 ± 0.01 0.14 ± 0.01 0.10 ± 0.01 0.08 ± 0.01 0.12 ± 0.01 −0.01 ± 0.01 0.91 ± 0.01 SMY −0.09 ± 0.07 0.62 ± 0.01 0.52 ± 0.01 0.67 ± 0.01 0.75 ± 0.01 0.36 ± 0.01 0.22 ± 0.01 LEG −0.03 ± 0.07 0.86 ± 0.02 0.30 ± 0.01 0.32 ± 0.01 0.59 ± 0.01 0.28 ± 0.01 0.23 ± 0.01 LOIN −0.18 ± 0.08 0.71 ± 0.04 0.54 ± 0.05 0.36 ± 0.01 0.50 ± 0.01 0.27 ± 0.01 0.18 ± 0.01 SHOUL −0.07 ± 0.07 0.86 ± 0.02 0.61 ± 0.05 0.54 ± 0.05 0.74 ± 0.01 0.29 ± 0.01 0.17 ± 0.01 AVGCONF −0.09 ± 0.07 0.96 ± 0.01 0.83 ± 0.03 0.68 ± 0.04 0.87 ± 0.02 0.38 ± 0.01 0.24 ± 0.01 CINDEX −0.20 ± 0.08 0.56 ± 0.06 0.42 ± 0.06 0.55 ± 0.06 0.47 ± 0.06 0.49 ± 0.06 0.34 ± 0.01 PRICE 0.90 ± 0.02 0.03 ± 0.07 0.07 ± 0.07 −0.03 ± 0.08 0.03 ± 0.07 0.02 ± 0.07 0.21 ± 0.08 1Phenotypic and genetic correlation estimates are presented above and below the diagonal, respectively, and are followed by their approximate standard error. 2AVGCONF = average carcass conformation score; CINDEX = carcass price grid value; HCW = hot carcass weight; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield. View Large Growth and ultrasound traits. Weight and gain traits were generally found to be positively correlated with age-constant EMDUS and FATUS, with correlation estimates ranging from 0.22 ± 0.14 to 0.54 ± 0.08 and 0.21 ± 0.15 to 0.63 ± 0.08, respectively (Table 5). In contrast, correlations between weight traits and weight-constant EMDUS and FATUS were generally negative and ranged from −0.47 ± 0.15 to −0.17 ± 0.13 and −0.16 ± 0.14 to 0.02 ± 0.12 (Table 5). Most estimates of EMDUS and FATUS found in the literature were adjusted to a constant weight, and negative correlations between weight traits and EMDUS and FATUS have been previously identified (Maximini et al., 2012; Mortimer et al., 2014). Mortimer et al. (2014) found that genetic correlations between weight traits and age-constant EMDUS and FATUS were positive while negative correlations were found when adjusting EMDUS and FATUS observations to a constant age and weight. Growth and carcass traits. Most genetic correlations between growth and carcass traits were weak to moderate in strength (Tables 9 to 11). Hot carcass weight was generally positively correlated with all growth traits at both age and fatness slaughter endpoints (0.07 ± 0.07 to 0.38 ± 0.05; Tables 9 and 11), indicating that current selection for weight and gain traits is expected to indirectly improve HCW. Thus, indirect genetic improvement for carcass yield remains achievable for commercial producers that market on a live-weight basis without the use of ultrasound measures. Interestingly, correlations between HCW and WWT and HCW and PWWT were not considerably different at either age (0.38 ± 0.05 versus 0.38 ± 0.05; Table 9) or fatness (0.21 ± 0.06 versus 0.18 ± 0.05; Table 11) endpoints, indicating that selection on WWT is likely adequate for HCW genetic improvement. However, Brito et al. (2017) reported a much stronger genetic correlation of 0.92 ± 0.02 between HCW and liveweight at 180 d in a New Zealand sheep population, and Australian research has indicated that WWT and PWWT are both moderately to strongly correlated with HCW, with genetic correlation estimates above 0.60 in many studies (Ingham et al., 2007; Greeff et al., 2008; Mortimer et al., 2010). Thus, it may be beneficial to examine the genetic correlation between a weight trait measured more closely to SAGE to allow for better indirect selection to improve HCW in the Canadian sheep population. With the exception of moderate negative genetic correlations between FATGR and BWT (−0.36 ± 0.06 to −0.35 ± 0.06), weak negative genetic correlations were observed between growth traits and FATGR (−0.16 ± 0.14 to −0.09 ± 0.05; Tables 9 and 10). Published genetic correlation estimates between FATGR and liveweights have varied widely. Most genetic correlation estimates between these traits have been positive when FATGR was unadjusted (Brito et al., 2017) or adjusted to a constant weight (Ingham et al., 2007; Greeff et al., 2008), although Mortimer et al. (2010) reported a moderate negative correlation when FATGR was adjusted to a constant weight. In general, conformation traits had weak negative correlations with the growth traits (−0.27 ± 0.06 to −0.08 ± 0.06; (Tables 9 to 11), indicating that selection on growth traits alone is not expected to substantially increase carcass muscularity. Overall, genetic correlations between growth and carcass traits were generally favorable, with the exception of unfavorable negative genetic correlations between growth traits and carcass conformation traits. Table 10. Genetic correlations between growth traits, ultrasound traits, and carcass traits adjusted to a fixed carcass weight endpoint1,2 BWT WWT PWWT ADG50 ADG100 FATGR −0.35 ± 0.06 −0.14 ± 0.05 −0.16 ± 0.05 −0.14 ± 0.05 −0.12 ± 0.05 SMY 0.33 ± 0.06 0.08 ± 0.05 0.09 ± 0.05 0.08 ± 0.05 0.06 ± 0.06 LEG −0.13 ± 0.06 −0.14 ± 0.05 −0.20 ± 0.05 −0.13 ± 0.05 −0.18 ± 0.05 LOIN −0.25 ± 0.06 −0.22 ± 0.05 −0.22 ± 0.05 −0.21 ± 0.05 −0.16 ± 0.05 SHOUL −0.27 ± 0.06 −0.25 ± 0.05 −0.27 ± 0.05 −0.25 ± 0.05 −0.17 ± 0.05 AVGCONF −0.24 ± 0.06 −0.21 ± 0.05 −0.27 ± 0.05 −0.21 ± 0.05 −0.22 ± 0.05 CINDEX 0.24 ± 0.07 −0.05 ± 0.06 −0.07 ± 0.06 −0.06 ± 0.06 −0.06 ± 0.06 PRICE 0.13 ± 0.07 −0.13 ± 0.06 −0.13 ± 0.06 −0.12 ± 0.06 −0.07 ± 0.06 WTUS EMDUSa EMDUSw FATUSa FATUSw FATGR −0.16 ± 0.14 0.05 ± 0.19 0.20 ± 0.14 0.38 ± 0.14 0.64 ± 0.12 SMY 0.06 ± 0.16 0.03 ± 0.21 −0.02 ± 0.15 −0.38 ± 0.15 −0.55 ± 0.14 LEG −0.26 ± 0.13 0.10 ± 0.17 0.32 ± 0.12 0.14 ± 0.14 0.39 ± 0.13 LOIN −0.24 ± 0.14 −0.08 ± 0.19 0.13 ± 0.14 0.18 ± 0.15 0.48 ± 0.15 SHOUL −0.21 ± 0.14 0.24 ± 0.19 0.40 ± 0.13 0.25 ± 0.15 0.51 ± 0.14 AVGCONF −0.20 ± 0.13 0.10 ± 0.17 0.28 ± 0.12 0.25 ± 0.14 0.50 ± 0.13 CINDEX 0.04 ± 0.18 0.29 ± 0.22 0.27 ± 0.16 −0.01 ± 0.18 −0.11 ± 0.17 PRICE −0.03 ± 0.17 0.31 ± 0.22 0.32 ± 0.16 0.00 ± 0.18 0.00 ± 0.17 BWT WWT PWWT ADG50 ADG100 FATGR −0.35 ± 0.06 −0.14 ± 0.05 −0.16 ± 0.05 −0.14 ± 0.05 −0.12 ± 0.05 SMY 0.33 ± 0.06 0.08 ± 0.05 0.09 ± 0.05 0.08 ± 0.05 0.06 ± 0.06 LEG −0.13 ± 0.06 −0.14 ± 0.05 −0.20 ± 0.05 −0.13 ± 0.05 −0.18 ± 0.05 LOIN −0.25 ± 0.06 −0.22 ± 0.05 −0.22 ± 0.05 −0.21 ± 0.05 −0.16 ± 0.05 SHOUL −0.27 ± 0.06 −0.25 ± 0.05 −0.27 ± 0.05 −0.25 ± 0.05 −0.17 ± 0.05 AVGCONF −0.24 ± 0.06 −0.21 ± 0.05 −0.27 ± 0.05 −0.21 ± 0.05 −0.22 ± 0.05 CINDEX 0.24 ± 0.07 −0.05 ± 0.06 −0.07 ± 0.06 −0.06 ± 0.06 −0.06 ± 0.06 PRICE 0.13 ± 0.07 −0.13 ± 0.06 −0.13 ± 0.06 −0.12 ± 0.06 −0.07 ± 0.06 WTUS EMDUSa EMDUSw FATUSa FATUSw FATGR −0.16 ± 0.14 0.05 ± 0.19 0.20 ± 0.14 0.38 ± 0.14 0.64 ± 0.12 SMY 0.06 ± 0.16 0.03 ± 0.21 −0.02 ± 0.15 −0.38 ± 0.15 −0.55 ± 0.14 LEG −0.26 ± 0.13 0.10 ± 0.17 0.32 ± 0.12 0.14 ± 0.14 0.39 ± 0.13 LOIN −0.24 ± 0.14 −0.08 ± 0.19 0.13 ± 0.14 0.18 ± 0.15 0.48 ± 0.15 SHOUL −0.21 ± 0.14 0.24 ± 0.19 0.40 ± 0.13 0.25 ± 0.15 0.51 ± 0.14 AVGCONF −0.20 ± 0.13 0.10 ± 0.17 0.28 ± 0.12 0.25 ± 0.14 0.50 ± 0.13 CINDEX 0.04 ± 0.18 0.29 ± 0.22 0.27 ± 0.16 −0.01 ± 0.18 −0.11 ± 0.17 PRICE −0.03 ± 0.17 0.31 ± 0.22 0.32 ± 0.16 0.00 ± 0.18 0.00 ± 0.17 1Correlation coefficients are followed by their approximate standard error. 2ADG50 = pre-weaning average daily gain; ADG100 = post-weaning average daily gain; AVGCONF = average carcass conformation score; BWT = birthweight; CINDEX = carcass price grid value; EMDUS = ultrasonically measured eye muscle depth; FATGR = fat depth at the GR site; FATUS = ultrasonically measured fat depth; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; PWWT = post-weaning weight; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield; WTUS = scanning weight; WWT = weaning weight. Ultrasound eye muscle and fat depth measurements adjusted to a constant age (EMDUSa, FATUSa) or scanning weight (EMDUSw, FATUSw). View Large Table 10. Genetic correlations between growth traits, ultrasound traits, and carcass traits adjusted to a fixed carcass weight endpoint1,2 BWT WWT PWWT ADG50 ADG100 FATGR −0.35 ± 0.06 −0.14 ± 0.05 −0.16 ± 0.05 −0.14 ± 0.05 −0.12 ± 0.05 SMY 0.33 ± 0.06 0.08 ± 0.05 0.09 ± 0.05 0.08 ± 0.05 0.06 ± 0.06 LEG −0.13 ± 0.06 −0.14 ± 0.05 −0.20 ± 0.05 −0.13 ± 0.05 −0.18 ± 0.05 LOIN −0.25 ± 0.06 −0.22 ± 0.05 −0.22 ± 0.05 −0.21 ± 0.05 −0.16 ± 0.05 SHOUL −0.27 ± 0.06 −0.25 ± 0.05 −0.27 ± 0.05 −0.25 ± 0.05 −0.17 ± 0.05 AVGCONF −0.24 ± 0.06 −0.21 ± 0.05 −0.27 ± 0.05 −0.21 ± 0.05 −0.22 ± 0.05 CINDEX 0.24 ± 0.07 −0.05 ± 0.06 −0.07 ± 0.06 −0.06 ± 0.06 −0.06 ± 0.06 PRICE 0.13 ± 0.07 −0.13 ± 0.06 −0.13 ± 0.06 −0.12 ± 0.06 −0.07 ± 0.06 WTUS EMDUSa EMDUSw FATUSa FATUSw FATGR −0.16 ± 0.14 0.05 ± 0.19 0.20 ± 0.14 0.38 ± 0.14 0.64 ± 0.12 SMY 0.06 ± 0.16 0.03 ± 0.21 −0.02 ± 0.15 −0.38 ± 0.15 −0.55 ± 0.14 LEG −0.26 ± 0.13 0.10 ± 0.17 0.32 ± 0.12 0.14 ± 0.14 0.39 ± 0.13 LOIN −0.24 ± 0.14 −0.08 ± 0.19 0.13 ± 0.14 0.18 ± 0.15 0.48 ± 0.15 SHOUL −0.21 ± 0.14 0.24 ± 0.19 0.40 ± 0.13 0.25 ± 0.15 0.51 ± 0.14 AVGCONF −0.20 ± 0.13 0.10 ± 0.17 0.28 ± 0.12 0.25 ± 0.14 0.50 ± 0.13 CINDEX 0.04 ± 0.18 0.29 ± 0.22 0.27 ± 0.16 −0.01 ± 0.18 −0.11 ± 0.17 PRICE −0.03 ± 0.17 0.31 ± 0.22 0.32 ± 0.16 0.00 ± 0.18 0.00 ± 0.17 BWT WWT PWWT ADG50 ADG100 FATGR −0.35 ± 0.06 −0.14 ± 0.05 −0.16 ± 0.05 −0.14 ± 0.05 −0.12 ± 0.05 SMY 0.33 ± 0.06 0.08 ± 0.05 0.09 ± 0.05 0.08 ± 0.05 0.06 ± 0.06 LEG −0.13 ± 0.06 −0.14 ± 0.05 −0.20 ± 0.05 −0.13 ± 0.05 −0.18 ± 0.05 LOIN −0.25 ± 0.06 −0.22 ± 0.05 −0.22 ± 0.05 −0.21 ± 0.05 −0.16 ± 0.05 SHOUL −0.27 ± 0.06 −0.25 ± 0.05 −0.27 ± 0.05 −0.25 ± 0.05 −0.17 ± 0.05 AVGCONF −0.24 ± 0.06 −0.21 ± 0.05 −0.27 ± 0.05 −0.21 ± 0.05 −0.22 ± 0.05 CINDEX 0.24 ± 0.07 −0.05 ± 0.06 −0.07 ± 0.06 −0.06 ± 0.06 −0.06 ± 0.06 PRICE 0.13 ± 0.07 −0.13 ± 0.06 −0.13 ± 0.06 −0.12 ± 0.06 −0.07 ± 0.06 WTUS EMDUSa EMDUSw FATUSa FATUSw FATGR −0.16 ± 0.14 0.05 ± 0.19 0.20 ± 0.14 0.38 ± 0.14 0.64 ± 0.12 SMY 0.06 ± 0.16 0.03 ± 0.21 −0.02 ± 0.15 −0.38 ± 0.15 −0.55 ± 0.14 LEG −0.26 ± 0.13 0.10 ± 0.17 0.32 ± 0.12 0.14 ± 0.14 0.39 ± 0.13 LOIN −0.24 ± 0.14 −0.08 ± 0.19 0.13 ± 0.14 0.18 ± 0.15 0.48 ± 0.15 SHOUL −0.21 ± 0.14 0.24 ± 0.19 0.40 ± 0.13 0.25 ± 0.15 0.51 ± 0.14 AVGCONF −0.20 ± 0.13 0.10 ± 0.17 0.28 ± 0.12 0.25 ± 0.14 0.50 ± 0.13 CINDEX 0.04 ± 0.18 0.29 ± 0.22 0.27 ± 0.16 −0.01 ± 0.18 −0.11 ± 0.17 PRICE −0.03 ± 0.17 0.31 ± 0.22 0.32 ± 0.16 0.00 ± 0.18 0.00 ± 0.17 1Correlation coefficients are followed by their approximate standard error. 2ADG50 = pre-weaning average daily gain; ADG100 = post-weaning average daily gain; AVGCONF = average carcass conformation score; BWT = birthweight; CINDEX = carcass price grid value; EMDUS = ultrasonically measured eye muscle depth; FATGR = fat depth at the GR site; FATUS = ultrasonically measured fat depth; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; PWWT = post-weaning weight; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield; WTUS = scanning weight; WWT = weaning weight. Ultrasound eye muscle and fat depth measurements adjusted to a constant age (EMDUSa, FATUSa) or scanning weight (EMDUSw, FATUSw). View Large Table 11. Genetic correlations between growth traits, ultrasound traits, and carcass traits adjusted to a fixed carcass fatness endpoint1,2 BWT WWT PWWT ADG50 ADG100 HCW 0.11 ± 0.07 0.21 ± 0.06 0.18 ± 0.05 0.20 ± 0.06 0.11 ± 0.06 SMY −0.14 ± 0.07 −0.19 ± 0.05 −0.22 ± 0.05 −0.18 ± 0.05 −0.15 ± 0.06 LEG −0.08 ± 0.06 −0.10 ± 0.05 −0.17 ± 0.05 −0.09 ± 0.05 −0.16 ± 0.05 LOIN −0.19 ± 0.07 −0.19 ± 0.05 −0.19 ± 0.05 −0.18 ± 0.06 −0.14 ± 0.06 SHOUL −0.22 ± 0.06 −0.23 ± 0.05 −0.24 ± 0.05 −0.22 ± 0.05 −0.15 ± 0.06 AVGCONF −0.19 ± 0.06 −0.18 ± 0.05 −0.24 ± 0.05 −0.18 ± 0.05 −0.21 ± 0.05 CINDEX −0.08 ± 0.07 −0.20 ± 0.06 −0.23 ± 0.06 −0.21 ± 0.06 −0.17 ± 0.06 PRICE 0.06 ± 0.07 0.09 ± 0.06 0.06 ± 0.06 0.09 ± 0.06 0.03 ± 0.06 WTUS EMDUSa EMDUSw FATUSa FATUSw HCW 0.13 ± 0.16 0.50 ± 0.21 0.33 ± 0.15 0.01 ± 0.17 −0.11 ± 0.16 SMY −0.22 ± 0.14 0.20 ± 0.18 0.41 ± 0.13 0.16 ± 0.15 0.38 ± 0.15 LEG −0.25 ± 0.14 0.17 ± 0.18 0.38 ± 0.13 0.13 ± 0.14 0.35 ± 0.14 LOIN −0.24 ± 0.16 −0.04 ± 0.20 0.16 ± 0.15 0.15 ± 0.17 0.41 ± 0.16 SHOUL −0.21 ± 0.15 0.35 ± 0.21 0.48 ± 0.14 0.21 ± 0.17 0.43 ± 0.16 AVGCONF −0.19 ± 0.14 0.17 ± 0.19 0.34 ± 0.13 0.25 ± 0.15 0.46 ± 0.14 CINDEX −0.09 ± 0.15 0.20 ± 0.19 0.30 ± 0.14 0.18 ± 0.16 0.28 ± 0.16 PRICE 0.12 ± 0.17 0.69 ± 0.19 0.56 ± 0.14 0.12 ± 0.18 0.03 ± 0.17 BWT WWT PWWT ADG50 ADG100 HCW 0.11 ± 0.07 0.21 ± 0.06 0.18 ± 0.05 0.20 ± 0.06 0.11 ± 0.06 SMY −0.14 ± 0.07 −0.19 ± 0.05 −0.22 ± 0.05 −0.18 ± 0.05 −0.15 ± 0.06 LEG −0.08 ± 0.06 −0.10 ± 0.05 −0.17 ± 0.05 −0.09 ± 0.05 −0.16 ± 0.05 LOIN −0.19 ± 0.07 −0.19 ± 0.05 −0.19 ± 0.05 −0.18 ± 0.06 −0.14 ± 0.06 SHOUL −0.22 ± 0.06 −0.23 ± 0.05 −0.24 ± 0.05 −0.22 ± 0.05 −0.15 ± 0.06 AVGCONF −0.19 ± 0.06 −0.18 ± 0.05 −0.24 ± 0.05 −0.18 ± 0.05 −0.21 ± 0.05 CINDEX −0.08 ± 0.07 −0.20 ± 0.06 −0.23 ± 0.06 −0.21 ± 0.06 −0.17 ± 0.06 PRICE 0.06 ± 0.07 0.09 ± 0.06 0.06 ± 0.06 0.09 ± 0.06 0.03 ± 0.06 WTUS EMDUSa EMDUSw FATUSa FATUSw HCW 0.13 ± 0.16 0.50 ± 0.21 0.33 ± 0.15 0.01 ± 0.17 −0.11 ± 0.16 SMY −0.22 ± 0.14 0.20 ± 0.18 0.41 ± 0.13 0.16 ± 0.15 0.38 ± 0.15 LEG −0.25 ± 0.14 0.17 ± 0.18 0.38 ± 0.13 0.13 ± 0.14 0.35 ± 0.14 LOIN −0.24 ± 0.16 −0.04 ± 0.20 0.16 ± 0.15 0.15 ± 0.17 0.41 ± 0.16 SHOUL −0.21 ± 0.15 0.35 ± 0.21 0.48 ± 0.14 0.21 ± 0.17 0.43 ± 0.16 AVGCONF −0.19 ± 0.14 0.17 ± 0.19 0.34 ± 0.13 0.25 ± 0.15 0.46 ± 0.14 CINDEX −0.09 ± 0.15 0.20 ± 0.19 0.30 ± 0.14 0.18 ± 0.16 0.28 ± 0.16 PRICE 0.12 ± 0.17 0.69 ± 0.19 0.56 ± 0.14 0.12 ± 0.18 0.03 ± 0.17 1Correlation coefficients are followed by their approximate standard error. 2ADG50 = pre-weaning average daily gain; ADG100 = post-weaning average daily gain; AVGCONF = average carcass conformation score; BWT = birthweight; CINDEX = carcass price grid value; EMDUS = ultrasonically measured eye muscle depth; FATUS = ultrasonically measured fat depth; HCW = hot carcass weight; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; PWWT = post-weaning weight; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield; WTUS = scanning weight; WWT = weaning weight. Ultrasound eye muscle and fat depth measurements adjusted to a constant age (EMDUSa, FATUSa) or scanning weight (EMDUSw, FATUSw). View Large Table 11. Genetic correlations between growth traits, ultrasound traits, and carcass traits adjusted to a fixed carcass fatness endpoint1,2 BWT WWT PWWT ADG50 ADG100 HCW 0.11 ± 0.07 0.21 ± 0.06 0.18 ± 0.05 0.20 ± 0.06 0.11 ± 0.06 SMY −0.14 ± 0.07 −0.19 ± 0.05 −0.22 ± 0.05 −0.18 ± 0.05 −0.15 ± 0.06 LEG −0.08 ± 0.06 −0.10 ± 0.05 −0.17 ± 0.05 −0.09 ± 0.05 −0.16 ± 0.05 LOIN −0.19 ± 0.07 −0.19 ± 0.05 −0.19 ± 0.05 −0.18 ± 0.06 −0.14 ± 0.06 SHOUL −0.22 ± 0.06 −0.23 ± 0.05 −0.24 ± 0.05 −0.22 ± 0.05 −0.15 ± 0.06 AVGCONF −0.19 ± 0.06 −0.18 ± 0.05 −0.24 ± 0.05 −0.18 ± 0.05 −0.21 ± 0.05 CINDEX −0.08 ± 0.07 −0.20 ± 0.06 −0.23 ± 0.06 −0.21 ± 0.06 −0.17 ± 0.06 PRICE 0.06 ± 0.07 0.09 ± 0.06 0.06 ± 0.06 0.09 ± 0.06 0.03 ± 0.06 WTUS EMDUSa EMDUSw FATUSa FATUSw HCW 0.13 ± 0.16 0.50 ± 0.21 0.33 ± 0.15 0.01 ± 0.17 −0.11 ± 0.16 SMY −0.22 ± 0.14 0.20 ± 0.18 0.41 ± 0.13 0.16 ± 0.15 0.38 ± 0.15 LEG −0.25 ± 0.14 0.17 ± 0.18 0.38 ± 0.13 0.13 ± 0.14 0.35 ± 0.14 LOIN −0.24 ± 0.16 −0.04 ± 0.20 0.16 ± 0.15 0.15 ± 0.17 0.41 ± 0.16 SHOUL −0.21 ± 0.15 0.35 ± 0.21 0.48 ± 0.14 0.21 ± 0.17 0.43 ± 0.16 AVGCONF −0.19 ± 0.14 0.17 ± 0.19 0.34 ± 0.13 0.25 ± 0.15 0.46 ± 0.14 CINDEX −0.09 ± 0.15 0.20 ± 0.19 0.30 ± 0.14 0.18 ± 0.16 0.28 ± 0.16 PRICE 0.12 ± 0.17 0.69 ± 0.19 0.56 ± 0.14 0.12 ± 0.18 0.03 ± 0.17 BWT WWT PWWT ADG50 ADG100 HCW 0.11 ± 0.07 0.21 ± 0.06 0.18 ± 0.05 0.20 ± 0.06 0.11 ± 0.06 SMY −0.14 ± 0.07 −0.19 ± 0.05 −0.22 ± 0.05 −0.18 ± 0.05 −0.15 ± 0.06 LEG −0.08 ± 0.06 −0.10 ± 0.05 −0.17 ± 0.05 −0.09 ± 0.05 −0.16 ± 0.05 LOIN −0.19 ± 0.07 −0.19 ± 0.05 −0.19 ± 0.05 −0.18 ± 0.06 −0.14 ± 0.06 SHOUL −0.22 ± 0.06 −0.23 ± 0.05 −0.24 ± 0.05 −0.22 ± 0.05 −0.15 ± 0.06 AVGCONF −0.19 ± 0.06 −0.18 ± 0.05 −0.24 ± 0.05 −0.18 ± 0.05 −0.21 ± 0.05 CINDEX −0.08 ± 0.07 −0.20 ± 0.06 −0.23 ± 0.06 −0.21 ± 0.06 −0.17 ± 0.06 PRICE 0.06 ± 0.07 0.09 ± 0.06 0.06 ± 0.06 0.09 ± 0.06 0.03 ± 0.06 WTUS EMDUSa EMDUSw FATUSa FATUSw HCW 0.13 ± 0.16 0.50 ± 0.21 0.33 ± 0.15 0.01 ± 0.17 −0.11 ± 0.16 SMY −0.22 ± 0.14 0.20 ± 0.18 0.41 ± 0.13 0.16 ± 0.15 0.38 ± 0.15 LEG −0.25 ± 0.14 0.17 ± 0.18 0.38 ± 0.13 0.13 ± 0.14 0.35 ± 0.14 LOIN −0.24 ± 0.16 −0.04 ± 0.20 0.16 ± 0.15 0.15 ± 0.17 0.41 ± 0.16 SHOUL −0.21 ± 0.15 0.35 ± 0.21 0.48 ± 0.14 0.21 ± 0.17 0.43 ± 0.16 AVGCONF −0.19 ± 0.14 0.17 ± 0.19 0.34 ± 0.13 0.25 ± 0.15 0.46 ± 0.14 CINDEX −0.09 ± 0.15 0.20 ± 0.19 0.30 ± 0.14 0.18 ± 0.16 0.28 ± 0.16 PRICE 0.12 ± 0.17 0.69 ± 0.19 0.56 ± 0.14 0.12 ± 0.18 0.03 ± 0.17 1Correlation coefficients are followed by their approximate standard error. 2ADG50 = pre-weaning average daily gain; ADG100 = post-weaning average daily gain; AVGCONF = average carcass conformation score; BWT = birthweight; CINDEX = carcass price grid value; EMDUS = ultrasonically measured eye muscle depth; FATUS = ultrasonically measured fat depth; HCW = hot carcass weight; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; PWWT = post-weaning weight; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield; WTUS = scanning weight; WWT = weaning weight. Ultrasound eye muscle and fat depth measurements adjusted to a constant age (EMDUSa, FATUSa) or scanning weight (EMDUSw, FATUSw). View Large Ultrasound and carcass traits. Hot carcass weight and PRICE were generally found to be positively correlated with EMDUS (0.31 ± 0.22 to 0.73 ± 0.18; Tables 9 to 11), with the strongest correlations being observed when observations were adjusted to a constant age. The moderately strong positive genetic correlations between HCW and EMDUS were in agreement with published estimates (Brito et al., 2017; Mortimer et al., 2017b). Genetic correlations between FATGR and FATUS were moderate to strong at all endpoints (0.38 ± 0.14 to 0.74 ± 0.12; Tables 9 and 11) and were within the range of published estimates (Brito et al., 2017; Mortimer et al., 2017b), suggesting that FATUS is a good indicator of GR site carcass fatness. Most measures of carcass muscularity and quality (LEG, LOIN, SHOUL, AVGCONF, SMY, CINDEX) had weak to moderate genetic correlations with the ultrasound traits (Tables 9 to 11). Carcass conformation traits were generally found to be positively correlated with both EMDUS (−0.08 ± 0.19 to 0.48 ± 0.14) and FATUS (0.13 ± 0.14 to 0.55 ± 0.15) at all slaughter endpoints (Tables 9 to 11). Similarly, Einarsson et al. (2015) reported a moderate genetic correlation between EUROP carcass conformation score and EMDUS (0.53). The genetic correlations between EMDUS and the carcass conformation traits, and FATGR and FATUS, were found to be the strongest in the weight-constant ultrasound analyses (Tables 9 to 11), suggesting that ultrasound traits should continue to be evaluated with observations adjusted to a constant weight in the CSGES to provide the best indication of their corresponding carcass trait measures. The results of this study reaffirm the usefulness of ultrasound measurements as indicators of carcass yield and quality in Canadian sheep breeding programs. In the future, genetic correlations should be re-estimated with a larger ultrasound trait dataset to clarify the optimal ultrasound trait endpoint to maximize indirect response in carcass quality traits. Table 9. Genetic correlations between growth traits, ultrasound traits, and carcass traits adjusted to a fixed slaughter age endpoint1,2 BWT WWT PWWT ADG50 ADG100 HCW 0.07 ± 0.07 0.38 ± 0.05 0.38 ± 0.05 0.37 ± 0.05 0.21 ± 0.06 FATGR −0.36 ± 0.06 −0.09 ± 0.05 −0.11 ± 0.05 −0.09 ± 0.05 −0.09 ± 0.06 SMY 0.34 ± 0.07 0.03 ± 0.05 0.04 ± 0.05 0.05 ± 0.06 0.04 ± 0.06 LEG −0.14 ± 0.06 −0.12 ± 0.05 −0.18 ± 0.05 −0.12 ± 0.05 −0.17 ± 0.05 LOIN −0.26 ± 0.06 −0.19 ± 0.05 −0.19 ± 0.05 −0.18 ± 0.05 −0.15 ± 0.06 SHOUL −0.25 ± 0.06 −0.18 ± 0.05 −0.20 ± 0.05 −0.18 ± 0.05 −0.12 ± 0.05 AVGCONF −0.24 ± 0.06 −0.17 ± 0.05 −0.22 ± 0.05 −0.16 ± 0.05 −0.19 ± 0.05 CINDEX 0.22 ± 0.07 −0.11 ± 0.06 −0.13 ± 0.06 −0.12 ± 0.06 −0.10 ± 0.06 PRICE 0.12 ± 0.07 0.28 ± 0.05 0.27 ± 0.05 0.26 ± 0.05 0.16 ± 0.06 WTUS EMDUSa EMDUSw FATUSa FATUSw HCW 0.23 ± 0.16 0.71 ± 0.19 0.44 ± 0.15 0.24 ± 0.17 0.10 ± 0.17 FATGR −0.13 ± 0.15 0.25 ± 0.19 0.36 ± 0.14 0.49 ± 0.14 0.74 ± 0.12 SMY 0.01 ± 0.17 −0.18 ± 0.21 −0.18 ± 0.16 −0.47 ± 0.15 −0.62 ± 0.14 LEG −0.26 ± 0.13 0.17 ± 0.17 0.39 ± 0.12 0.16 ± 0.14 0.41 ± 0.13 LOIN −0.22 ± 0.15 0.03 ± 0.19 0.22 ± 0.14 0.25 ± 0.16 0.54 ± 0.15 SHOUL −0.16 ± 0.15 0.36 ± 0.19 0.46 ± 0.13 0.33 ± 0.16 0.55 ± 0.15 AVGCONF −0.17 ± 0.14 0.19 ± 0.18 0.34 ± 0.12 0.31 ± 0.14 0.53 ± 0.13 CINDEX 0.00 ± 0.17 0.16 ± 0.22 0.18 ± 0.16 −0.06 ± 0.18 −0.11 ± 0.17 PRICE 0.23 ± 0.17 0.73 ± 0.18 0.54 ± 0.14 0.20 ± 0.17 0.07 ± 0.17 BWT WWT PWWT ADG50 ADG100 HCW 0.07 ± 0.07 0.38 ± 0.05 0.38 ± 0.05 0.37 ± 0.05 0.21 ± 0.06 FATGR −0.36 ± 0.06 −0.09 ± 0.05 −0.11 ± 0.05 −0.09 ± 0.05 −0.09 ± 0.06 SMY 0.34 ± 0.07 0.03 ± 0.05 0.04 ± 0.05 0.05 ± 0.06 0.04 ± 0.06 LEG −0.14 ± 0.06 −0.12 ± 0.05 −0.18 ± 0.05 −0.12 ± 0.05 −0.17 ± 0.05 LOIN −0.26 ± 0.06 −0.19 ± 0.05 −0.19 ± 0.05 −0.18 ± 0.05 −0.15 ± 0.06 SHOUL −0.25 ± 0.06 −0.18 ± 0.05 −0.20 ± 0.05 −0.18 ± 0.05 −0.12 ± 0.05 AVGCONF −0.24 ± 0.06 −0.17 ± 0.05 −0.22 ± 0.05 −0.16 ± 0.05 −0.19 ± 0.05 CINDEX 0.22 ± 0.07 −0.11 ± 0.06 −0.13 ± 0.06 −0.12 ± 0.06 −0.10 ± 0.06 PRICE 0.12 ± 0.07 0.28 ± 0.05 0.27 ± 0.05 0.26 ± 0.05 0.16 ± 0.06 WTUS EMDUSa EMDUSw FATUSa FATUSw HCW 0.23 ± 0.16 0.71 ± 0.19 0.44 ± 0.15 0.24 ± 0.17 0.10 ± 0.17 FATGR −0.13 ± 0.15 0.25 ± 0.19 0.36 ± 0.14 0.49 ± 0.14 0.74 ± 0.12 SMY 0.01 ± 0.17 −0.18 ± 0.21 −0.18 ± 0.16 −0.47 ± 0.15 −0.62 ± 0.14 LEG −0.26 ± 0.13 0.17 ± 0.17 0.39 ± 0.12 0.16 ± 0.14 0.41 ± 0.13 LOIN −0.22 ± 0.15 0.03 ± 0.19 0.22 ± 0.14 0.25 ± 0.16 0.54 ± 0.15 SHOUL −0.16 ± 0.15 0.36 ± 0.19 0.46 ± 0.13 0.33 ± 0.16 0.55 ± 0.15 AVGCONF −0.17 ± 0.14 0.19 ± 0.18 0.34 ± 0.12 0.31 ± 0.14 0.53 ± 0.13 CINDEX 0.00 ± 0.17 0.16 ± 0.22 0.18 ± 0.16 −0.06 ± 0.18 −0.11 ± 0.17 PRICE 0.23 ± 0.17 0.73 ± 0.18 0.54 ± 0.14 0.20 ± 0.17 0.07 ± 0.17 1Correlation coefficients are followed by their approximate standard error. 2ADG50 = pre-weaning average daily gain; ADG100 = post-weaning average daily gain; AVGCONF = average carcass conformation score; BWT = birthweight; CINDEX = carcass price grid value; EMDUS = ultrasonically measured eye muscle depth; FATGR = fat depth at the GR site; FATUS = ultrasonically measured fat depth; HCW = hot carcass weight; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; PWWT = post-weaning weight; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield; WTUS = scanning weight; WWT = weaning weight. Ultrasound eye muscle and fat depth measurements adjusted to a constant age (EMDUSa, FATUSa) or scanning weight (EMDUSw, FATUSw). View Large Table 9. Genetic correlations between growth traits, ultrasound traits, and carcass traits adjusted to a fixed slaughter age endpoint1,2 BWT WWT PWWT ADG50 ADG100 HCW 0.07 ± 0.07 0.38 ± 0.05 0.38 ± 0.05 0.37 ± 0.05 0.21 ± 0.06 FATGR −0.36 ± 0.06 −0.09 ± 0.05 −0.11 ± 0.05 −0.09 ± 0.05 −0.09 ± 0.06 SMY 0.34 ± 0.07 0.03 ± 0.05 0.04 ± 0.05 0.05 ± 0.06 0.04 ± 0.06 LEG −0.14 ± 0.06 −0.12 ± 0.05 −0.18 ± 0.05 −0.12 ± 0.05 −0.17 ± 0.05 LOIN −0.26 ± 0.06 −0.19 ± 0.05 −0.19 ± 0.05 −0.18 ± 0.05 −0.15 ± 0.06 SHOUL −0.25 ± 0.06 −0.18 ± 0.05 −0.20 ± 0.05 −0.18 ± 0.05 −0.12 ± 0.05 AVGCONF −0.24 ± 0.06 −0.17 ± 0.05 −0.22 ± 0.05 −0.16 ± 0.05 −0.19 ± 0.05 CINDEX 0.22 ± 0.07 −0.11 ± 0.06 −0.13 ± 0.06 −0.12 ± 0.06 −0.10 ± 0.06 PRICE 0.12 ± 0.07 0.28 ± 0.05 0.27 ± 0.05 0.26 ± 0.05 0.16 ± 0.06 WTUS EMDUSa EMDUSw FATUSa FATUSw HCW 0.23 ± 0.16 0.71 ± 0.19 0.44 ± 0.15 0.24 ± 0.17 0.10 ± 0.17 FATGR −0.13 ± 0.15 0.25 ± 0.19 0.36 ± 0.14 0.49 ± 0.14 0.74 ± 0.12 SMY 0.01 ± 0.17 −0.18 ± 0.21 −0.18 ± 0.16 −0.47 ± 0.15 −0.62 ± 0.14 LEG −0.26 ± 0.13 0.17 ± 0.17 0.39 ± 0.12 0.16 ± 0.14 0.41 ± 0.13 LOIN −0.22 ± 0.15 0.03 ± 0.19 0.22 ± 0.14 0.25 ± 0.16 0.54 ± 0.15 SHOUL −0.16 ± 0.15 0.36 ± 0.19 0.46 ± 0.13 0.33 ± 0.16 0.55 ± 0.15 AVGCONF −0.17 ± 0.14 0.19 ± 0.18 0.34 ± 0.12 0.31 ± 0.14 0.53 ± 0.13 CINDEX 0.00 ± 0.17 0.16 ± 0.22 0.18 ± 0.16 −0.06 ± 0.18 −0.11 ± 0.17 PRICE 0.23 ± 0.17 0.73 ± 0.18 0.54 ± 0.14 0.20 ± 0.17 0.07 ± 0.17 BWT WWT PWWT ADG50 ADG100 HCW 0.07 ± 0.07 0.38 ± 0.05 0.38 ± 0.05 0.37 ± 0.05 0.21 ± 0.06 FATGR −0.36 ± 0.06 −0.09 ± 0.05 −0.11 ± 0.05 −0.09 ± 0.05 −0.09 ± 0.06 SMY 0.34 ± 0.07 0.03 ± 0.05 0.04 ± 0.05 0.05 ± 0.06 0.04 ± 0.06 LEG −0.14 ± 0.06 −0.12 ± 0.05 −0.18 ± 0.05 −0.12 ± 0.05 −0.17 ± 0.05 LOIN −0.26 ± 0.06 −0.19 ± 0.05 −0.19 ± 0.05 −0.18 ± 0.05 −0.15 ± 0.06 SHOUL −0.25 ± 0.06 −0.18 ± 0.05 −0.20 ± 0.05 −0.18 ± 0.05 −0.12 ± 0.05 AVGCONF −0.24 ± 0.06 −0.17 ± 0.05 −0.22 ± 0.05 −0.16 ± 0.05 −0.19 ± 0.05 CINDEX 0.22 ± 0.07 −0.11 ± 0.06 −0.13 ± 0.06 −0.12 ± 0.06 −0.10 ± 0.06 PRICE 0.12 ± 0.07 0.28 ± 0.05 0.27 ± 0.05 0.26 ± 0.05 0.16 ± 0.06 WTUS EMDUSa EMDUSw FATUSa FATUSw HCW 0.23 ± 0.16 0.71 ± 0.19 0.44 ± 0.15 0.24 ± 0.17 0.10 ± 0.17 FATGR −0.13 ± 0.15 0.25 ± 0.19 0.36 ± 0.14 0.49 ± 0.14 0.74 ± 0.12 SMY 0.01 ± 0.17 −0.18 ± 0.21 −0.18 ± 0.16 −0.47 ± 0.15 −0.62 ± 0.14 LEG −0.26 ± 0.13 0.17 ± 0.17 0.39 ± 0.12 0.16 ± 0.14 0.41 ± 0.13 LOIN −0.22 ± 0.15 0.03 ± 0.19 0.22 ± 0.14 0.25 ± 0.16 0.54 ± 0.15 SHOUL −0.16 ± 0.15 0.36 ± 0.19 0.46 ± 0.13 0.33 ± 0.16 0.55 ± 0.15 AVGCONF −0.17 ± 0.14 0.19 ± 0.18 0.34 ± 0.12 0.31 ± 0.14 0.53 ± 0.13 CINDEX 0.00 ± 0.17 0.16 ± 0.22 0.18 ± 0.16 −0.06 ± 0.18 −0.11 ± 0.17 PRICE 0.23 ± 0.17 0.73 ± 0.18 0.54 ± 0.14 0.20 ± 0.17 0.07 ± 0.17 1Correlation coefficients are followed by their approximate standard error. 2ADG50 = pre-weaning average daily gain; ADG100 = post-weaning average daily gain; AVGCONF = average carcass conformation score; BWT = birthweight; CINDEX = carcass price grid value; EMDUS = ultrasonically measured eye muscle depth; FATGR = fat depth at the GR site; FATUS = ultrasonically measured fat depth; HCW = hot carcass weight; LEG = leg carcass conformation score; LOIN = loin carcass conformation score; PRICE = total carcass value; PWWT = post-weaning weight; SHOUL = shoulder conformation score; SMY = predicted saleable meat yield; WTUS = scanning weight; WWT = weaning weight. Ultrasound eye muscle and fat depth measurements adjusted to a constant age (EMDUSa, FATUSa) or scanning weight (EMDUSw, FATUSw). View Large Alternative Slaughter Endpoints The ideal slaughter endpoint to use in genetic evaluations will depend on a producer’s current breeding objective, which is in turn contingent on local production and marketing systems. Age, weight, and fatness slaughter endpoints each have their own advantages and disadvantages from a practical standpoint and further industry consultation is necessary to determine the optimal endpoint(s) for genetic evaluations. The aim of age-constant carcass trait genetic evaluations is to improve growth rate and production efficiency, but it may result in greater carcass variability, as indicated by the larger phenotypic variance observed when observations were adjusted to a constant SAGE. Nevertheless, an age endpoint may be useful for annual lambing systems where overwintering is expensive. Using weight as the endpoint in genetic evaluations would allow improvements to carcass uniformity and quality without increasing weight. However, the strong positive correlations between HCW and PRICE suggest that increasing HCW is economically advantageous under the current HLSA price grids. Marketing lambs at heavier weights was found to be common in the dataset with over 35% of carcasses having a HCW greater than the HLSA’s ideal weight of 24.0 kg. Marketing animals at a constant fatness threshold aims to improve uniformity, carcass quality, and saleable meat production. However, fatness is expensive to objectively measure via ultrasound, the genetic correlation between producer-measured subjective fat scores and FATGR are unknown in the Canadian sheep population, and heritability and phenotypic variance estimates tended to be the lowest in the fat-constant analyses. Given the diverse range of production practices, it is unlikely that a single endpoint would be ideal for all Canadian sheep producers. Consequently, it may be optimal to design two alternative Terminal sire selection indexes that are optimized for SAGE and carcass weight endpoints to target extensive and intensive producers, respectively. This would increase flexibility and allow Canadian sheep producers to evaluate the current performance of their flocks in order to determine whether selection for greater production efficiency or carcass quality would be optimal. CONCLUSIONS The consistency and quality of Canadian lamb carcasses has been previously identified as a barrier to the expansion of the Canadian lamb industry. This study presents the first genetic parameter estimates for carcass traits in a Canadian heavy lamb population and provides the parameters necessary for their genetic evaluation. Carcass traits were found to be moderately to strongly heritable, indicating that there is considerable potential to improve carcass yield and quality through genetic selection. Alternative slaughter endpoints had little impact on carcass trait heritability estimates, but genetic correlations were sensitive to the endpoint used to derive the estimates and further industry consultation is necessary to determine the ideal slaughter endpoint(s) for genetic evaluations. Growth and ultrasound traits were generally found to be favorably correlated with carcass traits, which suggests that current genetic selection practices should have a beneficial impact on carcass yield and quality. Future research will evaluate the relative efficiency of direct and indirect carcass trait selection as well as the value of including carcass traits into new Terminal sire selection indexes for the Canadian sheep industry. Footnotes 1 Funding for this research (project # 485324-2015) was provided by the National Science and Engineering Research Council, Ontario Sheep Farmers, the Canadian Sheep Breeders’ Association, and the Centre d’expertise en production ovine du Québec. Les Éleveurs d’ovins du Québec provided the carcass data used in this research. The authors would like to thank Bill Szkotnicki and Dr. Larry Schaeffer for their assistance with the analyses. LITERATURE CITED Beef + Lamb New Zealand Genetics . 2017 . Better sheep breeding: ram buying decisions . https://www.blnzgenetics.com/files/1507488803_Ram%20buyers%20guide%202017.pdf (accessed 12 March 2018). Boareki , M . 2017 . Genetic improvement of ewe reproductive traits in Rideau-Arcott. [thesis] . University of Guelph , Canada . http://hdl.handle.net/10214/11558 (accessed 12 March 2018). Brito , L. F. , J. C. McEwan , S. Miller , W. Bain , M. Lee , K. Dodds , S. A. Newman , N. Pickering , F. S. Schenkel , and S. Clarke . 2017 . Genetic parameters for various growth, carcass and meat quality traits in a New Zealand sheep population . Small Rumin. Res . 154 : 81 – 91 . doi: https://doi.org/10.1016/j.smallrumres.2017.07.011 Google Scholar Crossref Search ADS Brown , D. J . 2007 . Variance components for lambing ease and gestation length in sheep . Proc. Assoc. Advmt. Anim. Breed. Genet . 17 : 268 – 271 . http://livestocklibrary.com.au/handle/1234/5842. Ceyhan , A. , K. Moore , and R. Mrode . 2015 . The estimation of (co)variance components growth, reproduction, carcass, FECS and FECN traits in Lleyn sheep . Small Rumin. Res . 131 : 29 – 34 . doi: https://doi.org/10.1016/j.smallrumres.2015.07.024 Google Scholar Crossref Search ADS Conington , J. , S. Bishop , A. Waterhouse , and G. Simm . 1998 . A comparison of growth and carcass traits in Scottish Blackface lambs sired by genetically lean or fat rams . Anim. Sci . 67 : 299 – 309 . doi: https://doi.org/10.1017/S1357729800010067 Google Scholar Crossref Search ADS Coster , A . 2013 . pedigree: pedigree functions . https://cran.r-project.org/package=pedigree. (accessed 4 January 2018). Daetwyler , H. D., A. A. Swan , J. H. van der Werf , and B. J. Hayes . 2012 . Accuracy of pedigree and genomic predictions of carcass and novel meat quality traits in multi-breed sheep data assessed by cross-validation . Genet. Sel. Evol . 44 : 33 . doi: https://doi.org/10.1186/1297-9686-44-33 Google Scholar Crossref Search ADS PubMed Einarsson , E. , E. Eythórsdóttir , C. R. Smith , and J. V. Jónmundsson . 2015 . Genetic parameters for lamb carcass traits assessed by video image analysis, EUROP classification and in vivo measurements . Icelandic Agric. Sci . 28 : 3 – 14 . doi: https://doi.org/10.16886/IAS.2015.01 Google Scholar Crossref Search ADS Fernandes , T. L. , J. W. Wilton , and J. J. Tosh . 2004 . Estimates of genetic parameters for ultrasound-measured carcass traits in sheep . Can. J. Anim. Sci . 84 : 361 – 365 . doi: https://doi.org/10.4141/A03-080 Google Scholar Crossref Search ADS Fischer , T. M. , J. H. J. van der Werf , R. G. Banks , A. J. Ball , and A. R. Gilmour . 2006 . Genetic analysis of weight, fat and muscle depth in growing lambs using random regression models . Anim. Sci . 82 : 13 – 22 . doi: https://doi.org/10.1079/ASC200511 Google Scholar Crossref Search ADS Gilmour , A. R. , B. J. Gogel , B. R. Cullis , S. J. Welham , and R. Thompson . 2015 . ASReml user guide release 4.1 structural specification . VSN International Ltd, Hemel Hempstead, HP1 1ES, UK. p. 1–375. https://www.vsni.co.uk/downloads/asreml/release4/UserGuideStructural.pdf (accessed 12 January 2018). Gooch , M. , T. Moore , A. Mussel , K. Grier , D. Webb , and S. Dessureault . 2006 . Market opportunity analysis for the Canadian lamb industry . George Morris Centre , Guelph, Canada . http://www.georgemorris.org/publications/Canadian_Lamb_-_Market_Opportunities_and_Challenges.pdf (accessed 10 December 2018). Government of Canada . 1992 . Livestock and poultry carcass grading regulations . https://laws-lois.justice.gc.ca/PDF/SOR-92-541.pdf (accessed 10 December 2018). Greeff , J. C., E. Safari , N. M. Fogarty , D. L. Hopkins , F. D. Brien , K. D. Atkins , S. I. Mortimer , and J. H. van der Werf . 2008 . Genetic parameters for carcass and meat quality traits and their relationships to liveweight and wool production in hogget Merino rams . J. Anim. Breed. Genet . 125 : 205 – 215 . doi: https://doi.org/10.1111/j.1439-0388.2007.00711.x Google Scholar Crossref Search ADS PubMed Huisman , A. E. , and D. J. Brown . 2008 . Genetic parameters for bodyweight, wool, and disease resistance and reproduction traits in Merino sheep. 2. Genetic relationships between bodyweight traits and other traits . Aust. J. Exp. Agric . 48 : 1186 – 1193 . doi: https://doi.org/10.1071/EA08120 Google Scholar Crossref Search ADS Ingham , V. M. , N. M. Fogarty , A. R. Gilmour , R. A. Afolayan , L. J. Cummins , G. M. Gaunt , J. Stafford , and J. E. Hocking Edwards . 2007 . Genetic evaluation of crossbred lamb production. 4. Genetic parameters for first-cross animal performance . Aust. J. Agric. Res . 58 : 839 – 846 . doi: https://doi.org/10.1071/AR06368 Google Scholar Crossref Search ADS Jones , S. D. M. , W. M. Robertson , M. A. Price , and T. Coupland . 1996 . The prediction of saleable meat yield in lamb carcasses . Can. J. Anim. Sci . 76 : 49 – 53 . doi: https://doi.org/10.4141/cjas96-007 Google Scholar Crossref Search ADS Karamichou , E. , B. G. Merrell , W. A. Murray , G. Simm , and S. C. Bishop . 2007 . Selection for carcass quality in hill sheep measured by X-ray computer tomography . Animal 1 : 1 – 13 . doi: https://doi.org/10.1017/S1751731107413684 Google Scholar Crossref Search ADS PubMed Kirton , A. H. , and D. L. Johnson . 1979 . Interrelationships between GR and other lamb carcass fatness measurements . Proc. New Zeal. Soc. Anim. Prod . 39 : 194 – 201 . http://www.nzsap.org/system/files/proceedings/1979/ab79026.pdf (accessed 24 February 2018). Les Éleveurs d’ovins du Québec . 2017 . Heavy Lamb Sales Agency producer’s guide . http://ovinquebec.com/upload/pdf/agence_de_vente/guide_producteur/PG_2017_ENG_LEOQ.pdf (accessed 7 August 2018). Li , L. , and D. J. Brown . 2016 . Estimation of genetic parameters for lambing ease, birthweight and gestation length in Australian sheep . Anim. Prod. Sci . 56 : 934 – 940 . doi: https://doi.org/10.1071/AN14129 Google Scholar Crossref Search ADS Massender , E. , L. F. Brito , D. Kennedy , A. Canovas , and F. S. Schenkel . 2018 . Genetic parameter estimates for post-weaning growth, ultrasound, and carcass traits in Canadian heavy lambs . In: Proc. 11th World Congr. Genet. Appl. Livest. Prod. Electronic, Auckland, New Zealand. p. 1–5 . http://www.wcgalp.org/proceedings/2018/genetic-parameter-estimates-post-weaning-growth-ultrasound-and-carcass-traits (accessed 7 August 2018). Maxa , J. , E. Norberg , P. Berg , and M. Milerski . 2007a . Genetic parameters for body weight, longissimus muscle depth and fat depth for Suffolk sheep in the Czech Republic . Small Rumin. Res . 72 : 87 – 91 . doi: https://doi.org/10.1016/j.smallrumres.2006.04.018 Google Scholar Crossref Search ADS Maxa , J. , E. Norberg , P. Berg , and J. Pedersen . 2007b . Genetic parameters for carcass traits and in vivo measured muscle and fat depth in Danish Texel and Shropshire . Acta Agric. Scand. Sect. A . 57 : 49 – 54 . doi: https://doi.org/10.1080/09064700701440439 Maximini , L. , D. J. Brown , R. Baumung , and B. Fuerst-Waltl . 2012 . Genetic parameters of ultrasound and computer tomography scan traits in Austrian meat sheep . Livest. Sci . 146 : 168 – 174 . doi: https://doi.org/10.1016/j.livsci.2012.03.007 Google Scholar Crossref Search ADS Meat and Livestock Commercial Services Ltd . n.d. The value of independence: sheep carcase authentication and verification services . MLC Services Ltd ., Kenilworth, United Kingdom . http://www.mlcsl.co.uk/publications/Sheep-carcass-classification.pdf (accessed 24 January 2018). Mortimer , S. I. , S. Hatcher , N. M. Fogarty , J. H. J. van der Werf , D. J. Brown , A. A. Swan , J. C. Greeff , G. Refshauge , J. E. Hocking Edwards , and G. M. Gaunt . 2017a . Genetic parameters for wool traits, live weight, and ultrasound carcass traits in Merino sheep . J. Anim. Sci . 95 : 1879 – 1891 . doi: https://doi.org/10.2527/jas2016.1234 Mortimer , S. I. , S. Hatcher , N. M. Fogarty , J. H. J. van der Werf , D. J. Brown , A. A. Swan , R. H. Jacob , G. H. Geesink , D. L. Hopkins , J. E. Hocking Edwards , et al. 2017b . Genetic correlations between wool traits and carcass traits in Merino sheep . J. Anim. Sci . 95 : 2385 – 2398 . doi: https://doi.org/10.2527/jas2017.1385 Mortimer , S. I. , A. A. Swan , D. J. Brown , and J. H. J. van der Werf . 2014 . Genetic parameters revisited for ultrasound scanning traits in Australian sheep . In: Proc. 10th World Congr. Genet. Appl. Livest. Prod . Vancouver, Canada. p. 1 – 3 . http://www.wcgalp.org/proceedings/2014/genetic-parameters-revisited-ultrasound-scanning-traits-australian-sheep (accessed 7 August 2018). Mortimer , S. I. , J. H. J. van der Werf , R. H. Jacob , D. W. Pethick , K. L. Pearce , R. D. Warner , G. H. Geesink , J. E. Hocking Edwards , G. E. Gardner , E. N. Ponnampalam , et al. 2010 . Preliminary estimates of genetic parameters for carcass and meat quality traits in Australian sheep . Anim. Prod. Sci . 50 : 1135 – 1144 . doi: https://doi.org/10.1071/AN10126 Google Scholar Crossref Search ADS Payne , G. M. , A. W. Campbell , N. B. Jopson , J. C. McEwan , C. M. Logan , and P. D. Muir . 2009 . Genetic and phenotypic parameter estimates for growth, yield and meat quality traits in lamb . Proc. New Zeal. Soc. Anim. Prod . 69 : 210 – 214 . doi: https://doi.org/10.13140/2.1.1736.6400 Pollott , G. , D. Guy , and D. Croston . 1994 . Genetic parameters for lamb carcass characteristics at three end-points: fat level, age and weight . Anim. Sci . 58 : 65 – 75 . doi: https://doi.org/10.1017/S0003356100007091 Google Scholar Crossref Search ADS Quinton , C. D. , D. Kennedy , K. Stachowicz , and S. P. Miller . 2014 . Economic breeding objectives for Canadian lamb . In: Proc. 10th World Congr. Genet. Appl. Livest. Prod. Genetic Imp. Vancouver, Canada. p. 1–3 . http://www.wcgalp.org/proceedings/2014/economic-breeding-objectives-canadian-lamb. R Core Team . 2017 . R: a language and environment for statistical computing . R Foundation for Statistical Computing, Vienna, Austria. https://www.r-project.org/ (accessed 4 April 2018). Ríos-Utrera , A. , and L. Van Vleck . 2004 . Heritability estimates for carcass traits of cattle: a review . Genet. Mol. Res . 3 : 380 – 394 . http://www.funpecrp.com.br/gmr/year2004/vol3-3/gmr0102_abstract.htm Google Scholar PubMed Safari , E. , N. M. Fogarty , and A. R. Gilmour . 2005 . A review of genetic parameter estimates for wool, growth, meat and reproduction traits in sheep . Livest. Prod. Sci . 92 : 271 – 289 . doi: https://doi.org/10.1016/j.livprodsci.2004.09.003 Google Scholar Crossref Search ADS SAS Institute Inc . 2013 . Base SAS® 9.4 procedures guide . SAS Institute Inc ., Cary, NC . https://support.sas.com/documentation/cdl/en/procstat/66703/PDF/default/procstat.pdf (accessed 4 April 2018). Schaeffer , L. R. , and W. J. Szkotnicki . 2015 . Genetic evaluations of sheep in Canada . Centre for Genetic Improvement of Livestock, Department of Animal & Poultry Science, University of Guelph , Guelph, Canada . http://www.aps.uoguelph.ca/~lrs/ELARES/GE2015.pdf (accessed 7 August 2018). Swan , A. A. , R. G. Banks , D. J. Brown , and H. R. Chandler . 2017 . An update on genetic progress in the Australian sheep industry . Proc. Assoc. Advmt. Anim. Breed. Genet . 22 : 365 – 368 . http://www.aaabg.org/aaabghome/AAABG22papers/83Swan22365.pdf (accessed 1 November 2018). Tosh , J. J. , and R. A. Kemp . 1994 . Estimation of variance components for lamb weights in three sheep populations . J. Anim. Sci . 72 : 1184 – 1190 . doi: https://doi.org/10.2527/1994.7251184x Google Scholar Crossref Search ADS PubMed Tosh , J. J. , and J. W. Wilton . 2002 . A terminal-sire index for selecting rams . Can. J. Anim. Sci . 82 : 591 – 593 . doi: https://doi.org/10.4141/A02-024 Google Scholar Crossref Search ADS Walkom , S. F. , and D. J. Brown . 2016 . Genetic evaluation of adult ewe bodyweight and condition: relationship with lamb growth, reproduction, carcass and wool production . Anim. Prod. Sci . 57 : 20 – 32 . doi: https://doi.org/10.1071/AN15091 Google Scholar Crossref Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the American Society of Animal Science. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Hess, Andrew, S;Lunney, Joan, K;Abrams,, Samuel;Choi,, Igseo;Trible, Ben, R;Hess, Melanie, K;Rowland, Raymond R, R;Plastow, Graham, S;Dekkers, Jack C, M

doi: 10.1093/jas/sky446pmid: 30496411

, van Grevenhof, Elizabeth M;Vandenplas,, Jérémie;Calus, Mario P, L

doi: 10.1093/jas/sky433pmid: 30423111

Abstract Future genomic evaluation models to be used routinely in breeding programs for pigs and poultry need to be able to optimally use information of crossbred (CB) animals to predict breeding values for CB performance of purebred (PB) selection candidates. Important challenges in the commonly used single-step genomic best linear unbiased prediction (ssGBLUP) model are the definition of relationships between the different line compositions and the definition of the base generation per line. The use of metafounders (MFs) in ssGBLUP has been proposed to overcome these issues. When relationships between lines are known to be different from 0, the use of MFs generalizes the concept of genetic groups relying on the genotype data. Our objective was to investigate the effect of using MFs in genomic prediction for CB performance on estimated variance components, and accuracy and bias of GEBV. This was studied using stochastic simulation to generate data representing a three-way crossbreeding scheme in pigs, with the parental lines being either closely related or unrelated. Results show that using MFs, the variance components should be scaled appropriately, especially when basing them on estimates obtained with, for example a pedigree-based model. The accuracies of GEBV that were obtained using MFs were similar to accuracies without using MFs, regardless whether the lines involved in the CB were closely related or unrelated. The use of MFs resulted in a model that had similar or somewhat better convergence properties compared to other models. We recommend the use of MFs in ssGBLUP for genomic evaluations in crossbreeding schemes. INTRODUCTION In pig and poultry breeding, crossbreeding programs are generally used. The breeding objective is therefore to improve crossbred (CB) performance. Traits expressed in purebred (PB) and CB individuals are genetically not the same (Wei and Van der Werf, 1995; Wientjes and Calus, 2017). Therefore, it seems reasonable to use performance and genotypic data on CB individuals for genomic prediction of CB performance. However, collecting CB information might be difficult and expensive. In breeding programs using genomic selection, single-step genomic best linear unbiased prediction (ssGBLUP) is the model of choice, as it enables to use phenotypes of both animals with and without genotypes (Aguilar et al., 2010; Christensen and Lund, 2010). In the implementation of ssGBLUP, ensuring compatibility between the pedigree-based relationship matrix and the genomic relationship matrix is one of the main issues (Christensen, 2012; Legarra et al., 2014, 2015). In crossbreeding, genomic prediction enables to accurately link CB phenotypes to PB animals, and considers multiple breed compositions simultaneously. Important challenges are the definition of relationships between different line compositions and to appropriately define the different base generations. A proposed solution to both make the pedigree based and the genomic relationship matrix compatible and to appropriately deal with multiple base generations is the use of metafounders (MFs), which are pseudo-individuals that are included in the pedigree as founders without known parents (Legarra et al., 2015). These MFs are arbitrarily grouped based on, for example line, sex, and age, similar to genetic groups. Genetic groups are considered unrelated, while MFs are considered to be related, and their relationships are computed by genotypes of their descendants. Xiang et al. (2017) showed that single-step genomic evaluation with MFs performs at least as good as the breed-of-origin-based ssGBLUP in genomic prediction for crossbreeding breeding programs. Our objective was to investigate the effect of using MFs in genomic prediction for CB performance, depending on the relatedness of lines involved in the cross, on the accuracy and bias of GEBVs. In addition, the impact of the use of MFs on estimated variances was evaluated. To address these questions, we used simulated data for a three-way cross reflecting a pig breeding scheme. MATERIAL AND METHODS Data Simulation To investigate the effect of using MFs in genomic prediction on the accuracy and bias of GEBV, data for the historical, PB, and CB lines were simulated using the software QMSim (Sargolzaei and Schenkel, 2009). Phenotypes and genotypes of the individuals were simulated using a crossbreeding scheme. We simulated five correlated traits; one trait for each line composition, respectively, the three PB lines 1, 2, and 3, and the CB animals 23 and 123. Phenotypes and true breeding values (TBVs) for the line composition to which they belonged were simulated under additive gene action using a custom Fortran program. The traits were correlated, by assuming the same correlations among QTL effects as the genetic correlations between traits. Genetic correlations between traits were randomly sampled in the range of 0.2 to 0.8 from a uniform distribution (Table 1), and heritabilities were randomly sampled in the range of 0.2 to 0.4 from a uniform distribution. Within a line composition, 4,500 QTLs that explained 95% of the total additive genetic variance, and a residual polygenic effect that explained 5% of the total additive genetic variance, were underlying the associated simulated trait. TBVs were computed as the sum of the products of the simulated allele substitution effects with the genotypes of the 4,500 QTLs coded as 0, 1, and 2, and a polygenic effect. Allele substitution effects of QTLs were sampled from a multinormal distribution with means of 0 and variances of 1. Within each line composition to which a trait belongs, the variance explained by all QTLs was computed as the sum of the variances across all QTLs, assuming no correlation between the QTLs. The variance of each jth QTL was calculated as σj2=2pj(1−pj)αj2 ⁠, where pj is the allele frequency and αj is the allele substitution effect of jth QTL. Within each line composition, the allele substitution effects of the associated trait were rescaled to obtain a variance explained by all the QTLs equal to 1. Finally, the phenotypes for each animal for the trait associated with its line composition were generated by summing the TBVs and a residual error sampled from a normal distribution with a mean 0 and a variance computed such that the heritability within a line composition was equal to the simulated heritability. Marker and QTL mutation rates of 2.5 × 10–5 were assumed. In total, 52,908 markers were available with a minor allele frequency (MAF) > 0.05, spread across 18 chromosomes representing the pig genome. Table 1. Genetic correlations used for simulated true breeding values for related and unrelated lines Line PB-1 PB-2 PB-3 CB-23 PB-2 0.46 PB-3 0.27 0.80 CB-23 0.33 0.58 0.30 CB-1(23) 0.55 0.31 0.26 0.69 Line PB-1 PB-2 PB-3 CB-23 PB-2 0.46 PB-3 0.27 0.80 CB-23 0.33 0.58 0.30 CB-1(23) 0.55 0.31 0.26 0.69 View Large Table 1. Genetic correlations used for simulated true breeding values for related and unrelated lines Line PB-1 PB-2 PB-3 CB-23 PB-2 0.46 PB-3 0.27 0.80 CB-23 0.33 0.58 0.30 CB-1(23) 0.55 0.31 0.26 0.69 Line PB-1 PB-2 PB-3 CB-23 PB-2 0.46 PB-3 0.27 0.80 CB-23 0.33 0.58 0.30 CB-1(23) 0.55 0.31 0.26 0.69 View Large The simulation process was started with the simulation of a historical population with 100 generations. The size of the historical generations was set to 18,840, with equal numbers of males and females, for the first 70 generations. In the next 10 generations, the population gradually decreased to 390 individuals to mimic a bottleneck. During the last 20 generations (81 to 100) the population size increased up to 18,840 again. The number of males in the last generation was 90. After formation of the historical generations, breeding of lines 1, 2, and 3 started. Each line used 30 founder males and 1,000 founder females. A litter size of 2 was assumed with one male and one female progeny, such that each generation consisted of 2,000 individuals. All animals were replaced each generation. Matings were done at random between 30 males (randomly selected) and the 1,000 females. This scheme of line breeding was continued for 10 generations to represent a scenario with closely related lines and 100 generations to represent a scenario with unrelated lines, before starting the three-way crossbreeding program. Hereafter, these will be referred to as the related and unrelated scenarios, respectively. Starting from the last of those 10 or 100 generations, a three-way crossbreeding program with nine generations of random selection was simulated (see Figure 1 for a schematic overview). Random selection was used for simplicity, as selection would especially complicate the interpretation of estimated variances. In generations 1 to 3 of line breeding only pedigree was recorded, no genotypes or phenotypes. From generation 6 onwards crossbreeding started by crossing lines 2 and 3, after which this two-way cross was crossed with line 1, creating a three-way cross representing a pig breeding scheme. This crossbreeding was performed in generations 6 to 8. To mimic a practical situation where not all animals are phenotyped, and to limit the total number of phenotypes to enable computations within reasonable time, about 15,000 PB phenotypes were randomly recorded for generations 4 to 8, and about 3,500 CB phenotypes were randomly recorded for generations 6 to 8. About 5,250 PB genotypes were randomly recorded for generations 6 to 8, and about 925 CB genotypes were randomly recorded for generations 7 and 8. Random mating was applied in generations 1 to 8. Figure 1. View largeDownload slide Schematic overview of the simulation for the unrelated scenario, indicating which animals were genotyped or phenotyped, and the average numbers across replicates. Figure 1. View largeDownload slide Schematic overview of the simulation for the unrelated scenario, indicating which animals were genotyped or phenotyped, and the average numbers across replicates. In total, about 2,125 individuals had both a phenotype and a genotype. Finally, the 9th generation consisted of selection candidates for which only genotypes were available. The 9th generation contained 6,000 individuals, i.e., 2,000 for each of the PB lines 1, 2, and 3. Additionally, the same simulations were run with 500 individuals genotyped for each line composition within each genotyped generation, with the aim to test the influence of the number of genotypes on the estimation of MF relationships, variance components, and GEBV. Results were, however, very similar and therefore only the results for the initial scenarios are presented in this paper. The complete simulation was replicated 10 times. Statistical Analysis A five-trait ssGBLUP model (Aguilar et al., 2010; Christensen and Lund, 2010; Legarra et al., 2014) was used where the five traits modeled the PB performance of lines 1, 2, and 3 and the CB performance of crosses 23 and 1(23). The ssGBLUP model uses the inverse of a matrix with combined pedigree and genomic relationships. Inverses of the different combined pedigree genomic relationship matrices were computed using calc_grm (Calus and Vandenplas, 2016), considering MFs or not. The different inverses are described below. The variance components were estimated using Gibbs2f90 (Misztal et al., 2002) for which 50,000 samples were used, a burn-in of 3,500 and each 10th sample being stored. To limit the computational burden for the variance components estimation, all the genotyped animals of generation 9 were discarded from the datasets. The GEBV were computed using MiXBLUP (ten Napel et al., 2017). When the MFs were not considered, a genomic relationship matrix G required for the computation of the inverse of the combined pedigree-genomic relationship matrix H−1 was computed without line-specific adjustments. The matrix G was equal to: G=0.95Ga+0.05A22 where A22 stores the pedigree relationships among genotyped animals, and the adjusted genomic relationship matrix Ga is computed as follows: Ga=(1−fp−)G*+2fp−J where G* is a raw genomic relationship matrix computed following the first method of VanRaden (2018) using current allele frequencies computed from all genotyped animals, J is a matrix of ones, and fp− is the average pedigree inbreeding coefficient across genotyped animals, according to the FST method (Powell et al., 2010; Vitezica et al., 2011). When the MFs were considered in the ssGBLUP model (ssGBLUP_MF), one MF was assigned to each PB line, making a total of three MFs. Self-relationships and relationships between MFs were estimated based on genotypes of their descendants, and pedigree information, following the generalized least squares (GLS) method for multiple populations as shown by Garcia-Baccino et al. (2017), and implemented in the software createHmf (Legarra, 2016b). Briefly, the MF (self-)relationships are computed as twice the (co)variances of the estimated allele frequencies for the base generation of the pedigree. These base population allele frequencies were computed using the GLS method and all PB and CB genotypes (Garcia-Baccino et al., 2017). The computation of the inverse of the combined pedigree-genomic relationship matrix including MFs, H(γ)−1 ⁠, was performed using the software calc_grm (Calus and Vandenplas, 2016), following Legarra et al. (2015) and assuming a residual polygenic effect of 5%, by giving a weight of 0.05 to A22 as explained above, while in this case Ga=MM'/(1/2)n ⁠, where n is the number of SNPs and M stores the genotypes coded as {−1,0,1} ⁠. Note that this Ga can be obtained using the first method of VanRaden (2018) assuming that all allele frequencies are equal to 0.5. Finally, for reasons of comparison, the same model was also applied using the ordinary inverse of the pedigree based relationship matrix A−1 ⁠. This model is hereafter referred to as PBLUP. Evaluation of Model Performance Several aspects of the results were evaluated, between analyses with and without MFs. Estimated genetic variances were compared against true variances. True variances were empirically calculated as the variances of TBV of all the PB 2,000 animals in generation 1, and of all the 2,000 CB animals in generation 4. Similarly, true residual variances were empirically calculated as the variances of errors of all the 2,000 PB animals in generation 1, and of all the 2,000 CB animals in generation 4. Genetic variances estimated with the ssGBLUP_MF model were rescaled to get them on the same scale as the estimates of the other models where the genetic parameters relate to a base generation of supposedly unrelated animals (Legarra et al., 2015; Xiang et al., 2017). This scaling involved multiplying the genetic variances for the PB traits with (1−(γPB/2)) ⁠, where γPB is the self-relationship in the corresponding PB line. For the CB traits, this transformation should be done for each breed-of-origin-specific genetic variance component, and then summing across breed of origins. We did not consider breed of origin in the model, however, computed a weighted average of the scaling factor (1−(γPB/2)) across the PB lines involved in the CB animals, where weights were based on the breed composition of the cross. This approach is valid under the assumption that the genetic variance for CB performance is the same for each PB line. Finally, estimated genetic correlations were compared to simulated values. For ssGBLUP_MF, the estimates were computed from the unscaled estimated (co-)variances following Xiang et al. (2017). The accuracy of GEBV for both PB and CB performance was computed as the correlation between the TBV and the GEBV for the PB selection candidates in generation 9. The bias of the level of the GEBV and the bias of the scale of the GEBV were evaluated, respectively, as the intercept and slope of the regression of the TBV on the GEBV. Accuracies and bias were computed for each PB line separately. Finally, the convergence of ssGBLUP was compared in both situations with and without MFs. RESULTS Genetic Differentiation Between Lines For the two scenarios, i.e. related and unrelated scenarios, the level of genetic differentiation between the three PB lines was measured using the global Wright’s FST statistic, as implemented in the software Genepop (4.2) (Raymond and Rousset, 1995; Rousset, 2008). Using the genotypes of all PB animals in generation 6, the estimated global Wright’s FST statistics were on average equal to 0.06 for the related scenario, and to 0.36 for the unrelated scenario, across the five replicates. Relationships Among MFs and Estimated Variance Components The estimated self-relationships of the MFs were around 0.17 for the related and around 0.74 for the unrelated scenario (Table 2). The relationships among MFs showed to be very similar in the scenarios with related or unrelated lines, ranging from 0.045 to 0.049. Table 2. Relationships among metafounders for related and unrelated scenarios (average of 10 replicates; SE within brackets) Related scenario Unrelated scenario Line 1 Line 2 Line 3 Line 1 Line 2 Line 3 Line 1 0.171 (0.005) 0.049 (0.002) 0.047 (0.002) 0.746 (0.020) 0.046 (0.005) 0.045 (0.006) Line 2 0.049 (0.002) 0.171 (0.007) 0.046 (0.002) 0.046 (0.005) 0.741 (0.016) 0.046 (0.005) Line 3 0.047 (0.002) 0.046 (0.002) 0.171 (0.006) 0.045 (0.006) 0.046 (0.005) 0.743 (0.020) Related scenario Unrelated scenario Line 1 Line 2 Line 3 Line 1 Line 2 Line 3 Line 1 0.171 (0.005) 0.049 (0.002) 0.047 (0.002) 0.746 (0.020) 0.046 (0.005) 0.045 (0.006) Line 2 0.049 (0.002) 0.171 (0.007) 0.046 (0.002) 0.046 (0.005) 0.741 (0.016) 0.046 (0.005) Line 3 0.047 (0.002) 0.046 (0.002) 0.171 (0.006) 0.045 (0.006) 0.046 (0.005) 0.743 (0.020) View Large Table 2. Relationships among metafounders for related and unrelated scenarios (average of 10 replicates; SE within brackets) Related scenario Unrelated scenario Line 1 Line 2 Line 3 Line 1 Line 2 Line 3 Line 1 0.171 (0.005) 0.049 (0.002) 0.047 (0.002) 0.746 (0.020) 0.046 (0.005) 0.045 (0.006) Line 2 0.049 (0.002) 0.171 (0.007) 0.046 (0.002) 0.046 (0.005) 0.741 (0.016) 0.046 (0.005) Line 3 0.047 (0.002) 0.046 (0.002) 0.171 (0.006) 0.045 (0.006) 0.046 (0.005) 0.743 (0.020) Related scenario Unrelated scenario Line 1 Line 2 Line 3 Line 1 Line 2 Line 3 Line 1 0.171 (0.005) 0.049 (0.002) 0.047 (0.002) 0.746 (0.020) 0.046 (0.005) 0.045 (0.006) Line 2 0.049 (0.002) 0.171 (0.007) 0.046 (0.002) 0.046 (0.005) 0.741 (0.016) 0.046 (0.005) Line 3 0.047 (0.002) 0.046 (0.002) 0.171 (0.006) 0.045 (0.006) 0.046 (0.005) 0.743 (0.020) View Large The average variance component estimates (and SD) for the related and unrelated scenarios are presented for PBLUP, ssGBLUP and ssGBLUP_MF (Tables 3 and 4). For comparison, presented genetic variances estimated with the ssGBLUP_MF model were rescaled as described in a previous section. The estimated variances were compared against the empirically calculated true values outlined in Table 5. For both the related and unrelated scenarios, estimated residual variances were close to the empirically calculated true values for all three models, with deviations from the simulated values ranging from −5.3% to 2.6%. Estimated genetic variances differed for the related and unrelated scenarios. In the related scenario, the genetic variances were in all cases overestimated, with deviations from the simulated values ranging from 1.7% to 32.4%. Genetic variances were on average overestimated by 12.9, 14.9, and 11.5%, respectively, with the models PBLUP, ssGBLUP, and ssGBLUP_MF. In the unrelated scenario, the most extreme estimates across the models underestimated the genetic variance by 4.1% or overestimated it by 27.3%. The genetic variance was on average underestimated by 3.8% and 0.3% by PBLUP and ssGBLUP_MF, respectively, while it was overestimated by 16.8% for ssGBLUP. For both scenarios, not performing the scaling of the estimates for ssGBLUP_MF yielded genetic variances that were overestimated by 22.0% and 58.7% for the related and unrelated scenarios, respectively (Supplementary Table 3). Table 3. Estimated variance components for the related scenario for three different models: PBLUP, ssGBLUP, and ssGBLUP using metafounders. Residual and genetic variances estimates are presented for purebred traits 1, 2, and 3 and crossbred traits 23 and 1(23) Genetic variances and correlations1 Model Trait Residual 1 2 3 23 1(23) PBLUP 1 2.672 1.141 0.614 0.236 0.534 0.505 2 1.659 0.614 1.063 0.291 0.466 0.212 3 3.699 0.236 0.291 1.182 0.308 0.195 23 1.907 0.534 0.466 0.308 1.092 0.512 1(23) 3.456 0.505 0.212 0.195 0.512 1.286 ssGBLUP 1 2.656 1.173 0.653 0.096 0.576 0.506 2 1.651 0.653 1.079 0.186 0.528 0.249 3 3.690 0.096 0.186 1.202 0.288 0.183 23 1.862 0.576 0.528 0.288 1.155 0.517 1(23) 3.497 0.506 0.249 0.183 0.517 1.256 ssGBLUP-MF2 1 2.689 1.110 0.754 0.035 0.550 0.474 2 1.652 0.754 1.071 0.095 0.521 0.376 3 3.687 0.035 0.095 1.196 0.253 0.215 23 1.846 0.550 0.521 0.253 1.159 0.530 1(23) 3.539 0.474 0.376 0.215 0.530 1.160 Genetic variances and correlations1 Model Trait Residual 1 2 3 23 1(23) PBLUP 1 2.672 1.141 0.614 0.236 0.534 0.505 2 1.659 0.614 1.063 0.291 0.466 0.212 3 3.699 0.236 0.291 1.182 0.308 0.195 23 1.907 0.534 0.466 0.308 1.092 0.512 1(23) 3.456 0.505 0.212 0.195 0.512 1.286 ssGBLUP 1 2.656 1.173 0.653 0.096 0.576 0.506 2 1.651 0.653 1.079 0.186 0.528 0.249 3 3.690 0.096 0.186 1.202 0.288 0.183 23 1.862 0.576 0.528 0.288 1.155 0.517 1(23) 3.497 0.506 0.249 0.183 0.517 1.256 ssGBLUP-MF2 1 2.689 1.110 0.754 0.035 0.550 0.474 2 1.652 0.754 1.071 0.095 0.521 0.376 3 3.687 0.035 0.095 1.196 0.253 0.215 23 1.846 0.550 0.521 0.253 1.159 0.530 1(23) 3.539 0.474 0.376 0.215 0.530 1.160 1Variances are on the diagonal, correlations are on the off-diagonal. Standard errors of the variances ranged 0.031 to 0.062, and from 0.023 to 0.130 for the genetic correlations. Standard errors for each estimate are presented in Supplementary Table 1. 2Genetic variances after scaling are presented. View Large Table 3. Estimated variance components for the related scenario for three different models: PBLUP, ssGBLUP, and ssGBLUP using metafounders. Residual and genetic variances estimates are presented for purebred traits 1, 2, and 3 and crossbred traits 23 and 1(23) Genetic variances and correlations1 Model Trait Residual 1 2 3 23 1(23) PBLUP 1 2.672 1.141 0.614 0.236 0.534 0.505 2 1.659 0.614 1.063 0.291 0.466 0.212 3 3.699 0.236 0.291 1.182 0.308 0.195 23 1.907 0.534 0.466 0.308 1.092 0.512 1(23) 3.456 0.505 0.212 0.195 0.512 1.286 ssGBLUP 1 2.656 1.173 0.653 0.096 0.576 0.506 2 1.651 0.653 1.079 0.186 0.528 0.249 3 3.690 0.096 0.186 1.202 0.288 0.183 23 1.862 0.576 0.528 0.288 1.155 0.517 1(23) 3.497 0.506 0.249 0.183 0.517 1.256 ssGBLUP-MF2 1 2.689 1.110 0.754 0.035 0.550 0.474 2 1.652 0.754 1.071 0.095 0.521 0.376 3 3.687 0.035 0.095 1.196 0.253 0.215 23 1.846 0.550 0.521 0.253 1.159 0.530 1(23) 3.539 0.474 0.376 0.215 0.530 1.160 Genetic variances and correlations1 Model Trait Residual 1 2 3 23 1(23) PBLUP 1 2.672 1.141 0.614 0.236 0.534 0.505 2 1.659 0.614 1.063 0.291 0.466 0.212 3 3.699 0.236 0.291 1.182 0.308 0.195 23 1.907 0.534 0.466 0.308 1.092 0.512 1(23) 3.456 0.505 0.212 0.195 0.512 1.286 ssGBLUP 1 2.656 1.173 0.653 0.096 0.576 0.506 2 1.651 0.653 1.079 0.186 0.528 0.249 3 3.690 0.096 0.186 1.202 0.288 0.183 23 1.862 0.576 0.528 0.288 1.155 0.517 1(23) 3.497 0.506 0.249 0.183 0.517 1.256 ssGBLUP-MF2 1 2.689 1.110 0.754 0.035 0.550 0.474 2 1.652 0.754 1.071 0.095 0.521 0.376 3 3.687 0.035 0.095 1.196 0.253 0.215 23 1.846 0.550 0.521 0.253 1.159 0.530 1(23) 3.539 0.474 0.376 0.215 0.530 1.160 1Variances are on the diagonal, correlations are on the off-diagonal. Standard errors of the variances ranged 0.031 to 0.062, and from 0.023 to 0.130 for the genetic correlations. Standard errors for each estimate are presented in Supplementary Table 1. 2Genetic variances after scaling are presented. View Large Table 4. Estimated variance components for the unrelated scenario for three different models: PBLUP, ssGBLUP, and ssGBLUP using metafounders in the model Genetic variances and correlations1 Model Trait Residual 1 2 3 23 1(23) PBLUP 1 2.673 1.112 0.604 0.115 0.352 0.410 2 1.670 0.604 1.039 0.187 0.516 0.300 3 3.720 0.115 0.187 1.099 0.269 0.106 23 1.856 0.352 0.516 0.269 0.904 0.510 1(23) 3.581 0.410 0.300 0.106 0.510 0.900 ssGBLUP 1 2.615 1.244 0.597 0.069 0.341 0.517 2 1.640 0.597 1.126 0.151 0.490 0.332 3 3.651 0.069 0.151 1.245 0.245 0.158 23 1.827 0.341 0.490 0.245 1.041 0.533 1(23) 3.544 0.517 0.332 0.158 0.533 1.018 ssGBLUP-MF2 1 2.688 1.080 0.729 0.183 0.468 0.506 2 1.680 0.729 1.026 0.292 0.475 0.469 3 3.714 0.183 0.292 1.100 0.297 0.094 23 1.903 0.468 0.475 0.297 0.826 0.554 1(23) 3.518 0.506 0.469 0.094 0.554 0.835 Genetic variances and correlations1 Model Trait Residual 1 2 3 23 1(23) PBLUP 1 2.673 1.112 0.604 0.115 0.352 0.410 2 1.670 0.604 1.039 0.187 0.516 0.300 3 3.720 0.115 0.187 1.099 0.269 0.106 23 1.856 0.352 0.516 0.269 0.904 0.510 1(23) 3.581 0.410 0.300 0.106 0.510 0.900 ssGBLUP 1 2.615 1.244 0.597 0.069 0.341 0.517 2 1.640 0.597 1.126 0.151 0.490 0.332 3 3.651 0.069 0.151 1.245 0.245 0.158 23 1.827 0.341 0.490 0.245 1.041 0.533 1(23) 3.544 0.517 0.332 0.158 0.533 1.018 ssGBLUP-MF2 1 2.688 1.080 0.729 0.183 0.468 0.506 2 1.680 0.729 1.026 0.292 0.475 0.469 3 3.714 0.183 0.292 1.100 0.297 0.094 23 1.903 0.468 0.475 0.297 0.826 0.554 1(23) 3.518 0.506 0.469 0.094 0.554 0.835 Residual and genetic variance estimates are presented for purebred traits 1, 2, and 3 and crossbred traits 23 and 123. 1Variances are on the diagonal, correlations are on the off-diagonal. Standard errors of the variances ranged from 0.035 to 0.074, and from 0.040 to 0.130 for the genetic correlations. Standard errors for each estimate are presented in Supplementary Table 2. 2Genetic variances after scaling are presented. View Large Table 4. Estimated variance components for the unrelated scenario for three different models: PBLUP, ssGBLUP, and ssGBLUP using metafounders in the model Genetic variances and correlations1 Model Trait Residual 1 2 3 23 1(23) PBLUP 1 2.673 1.112 0.604 0.115 0.352 0.410 2 1.670 0.604 1.039 0.187 0.516 0.300 3 3.720 0.115 0.187 1.099 0.269 0.106 23 1.856 0.352 0.516 0.269 0.904 0.510 1(23) 3.581 0.410 0.300 0.106 0.510 0.900 ssGBLUP 1 2.615 1.244 0.597 0.069 0.341 0.517 2 1.640 0.597 1.126 0.151 0.490 0.332 3 3.651 0.069 0.151 1.245 0.245 0.158 23 1.827 0.341 0.490 0.245 1.041 0.533 1(23) 3.544 0.517 0.332 0.158 0.533 1.018 ssGBLUP-MF2 1 2.688 1.080 0.729 0.183 0.468 0.506 2 1.680 0.729 1.026 0.292 0.475 0.469 3 3.714 0.183 0.292 1.100 0.297 0.094 23 1.903 0.468 0.475 0.297 0.826 0.554 1(23) 3.518 0.506 0.469 0.094 0.554 0.835 Genetic variances and correlations1 Model Trait Residual 1 2 3 23 1(23) PBLUP 1 2.673 1.112 0.604 0.115 0.352 0.410 2 1.670 0.604 1.039 0.187 0.516 0.300 3 3.720 0.115 0.187 1.099 0.269 0.106 23 1.856 0.352 0.516 0.269 0.904 0.510 1(23) 3.581 0.410 0.300 0.106 0.510 0.900 ssGBLUP 1 2.615 1.244 0.597 0.069 0.341 0.517 2 1.640 0.597 1.126 0.151 0.490 0.332 3 3.651 0.069 0.151 1.245 0.245 0.158 23 1.827 0.341 0.490 0.245 1.041 0.533 1(23) 3.544 0.517 0.332 0.158 0.533 1.018 ssGBLUP-MF2 1 2.688 1.080 0.729 0.183 0.468 0.506 2 1.680 0.729 1.026 0.292 0.475 0.469 3 3.714 0.183 0.292 1.100 0.297 0.094 23 1.903 0.468 0.475 0.297 0.826 0.554 1(23) 3.518 0.506 0.469 0.094 0.554 0.835 Residual and genetic variance estimates are presented for purebred traits 1, 2, and 3 and crossbred traits 23 and 123. 1Variances are on the diagonal, correlations are on the off-diagonal. Standard errors of the variances ranged from 0.035 to 0.074, and from 0.040 to 0.130 for the genetic correlations. Standard errors for each estimate are presented in Supplementary Table 2. 2Genetic variances after scaling are presented. View Large Table 5. Empirically calculated true variance components for the related and unrelated scenarios Trait Related scenario Unrelated scenario Residual variancea Additive genetic variancea Heritabilityb Residual variancea Additive genetic variancea Heritabilityb 1 2.720 1.040 0.28 2.670 1.060 0.28 2 1.635 1.045 0.39 1.645 1.061 0.39 3 3.697 1.074 0.22 3.857 1.070 0.22 23 1.859 0.985 0.35 1.873 0.817 0.30 1(23) 3.580 0.971 0.21 3.554 0.871 0.20 Trait Related scenario Unrelated scenario Residual variancea Additive genetic variancea Heritabilityb Residual variancea Additive genetic variancea Heritabilityb 1 2.720 1.040 0.28 2.670 1.060 0.28 2 1.635 1.045 0.39 1.645 1.061 0.39 3 3.697 1.074 0.22 3.857 1.070 0.22 23 1.859 0.985 0.35 1.873 0.817 0.30 1(23) 3.580 0.971 0.21 3.554 0.871 0.20 aAll standard errors were <0.045. bAll standard errors were <0.01. View Large Table 5. Empirically calculated true variance components for the related and unrelated scenarios Trait Related scenario Unrelated scenario Residual variancea Additive genetic variancea Heritabilityb Residual variancea Additive genetic variancea Heritabilityb 1 2.720 1.040 0.28 2.670 1.060 0.28 2 1.635 1.045 0.39 1.645 1.061 0.39 3 3.697 1.074 0.22 3.857 1.070 0.22 23 1.859 0.985 0.35 1.873 0.817 0.30 1(23) 3.580 0.971 0.21 3.554 0.871 0.20 Trait Related scenario Unrelated scenario Residual variancea Additive genetic variancea Heritabilityb Residual variancea Additive genetic variancea Heritabilityb 1 2.720 1.040 0.28 2.670 1.060 0.28 2 1.635 1.045 0.39 1.645 1.061 0.39 3 3.697 1.074 0.22 3.857 1.070 0.22 23 1.859 0.985 0.35 1.873 0.817 0.30 1(23) 3.580 0.971 0.21 3.554 0.871 0.20 aAll standard errors were <0.045. bAll standard errors were <0.01. View Large Estimates of the genetic correlations among PB lines showed large deviations from the simulated values, and were on average underestimated, both for the unrelated and related scenarios. Estimated genetic correlations between the PB lines 1, 2, and 3 and the CB 23 and 1(23) animals were generally close to the simulated values. Across models, estimated genetic correlations were similar, both within the related and the unrelated scenario. The largest differences were observed for the related scenario, where the estimated genetic correlations of the PBLUP and ssGBLUP model were on average 0.06 to 0.07 lower than those of ssGBLUP_MF, whose estimates were closer to the simulated values. Accuracy and Bias A total of 2,000 genotyped selection candidates per line were used for computing accuracy and bias. Across the related and unrelated scenarios, for PB performance the accuracies ranged from 0.37 to 0.47 with PBLUP (Supplementary Table 4), and from 0.47 to 0.59 for ssGBLUP (Figure 2; Supplementary Table 4). For CB performance the accuracies ranged from 0.13 to 0.27 with PBLUP, and from 0.27 to 0.40 with ssGBLUP. Accuracies of ssGBLUP and ssGBLUP_MF within the same scenario were very similar, with any differences being smaller than the standard errors (Figure 2; Supplementary Table 4). Accuracies of PBLUP were very similar between the related and unrelated scenario, because effectively no information was used across lines due to lack of pedigree links between the lines, leading to very similar amounts of information being available in both scenarios. Accuracies of ssGBLUP were comparable across the related and unrelated scenarios, except for PB performance of lines 1 and 3, where the accuracies were higher for the related scenario (Figure 2; Supplementary Table 4). Figure 2. View largeDownload slide Accuracies of GEBV for purebred selection candidates in generation 9, either for purebred or crossbred performance, using ssGBLUP models with or without metafounders, for lines with related and unrelated pedigrees in purebred and crossbred performances. Red (blue) bars represent models with (without) metafounders. Figure 2. View largeDownload slide Accuracies of GEBV for purebred selection candidates in generation 9, either for purebred or crossbred performance, using ssGBLUP models with or without metafounders, for lines with related and unrelated pedigrees in purebred and crossbred performances. Red (blue) bars represent models with (without) metafounders. The mean values of all sets of (G)EBV were unbiased, as the intercepts of the regression of TBV on EBV were in most cases not significantly different from 0 (Supplementary Table 5). The coefficients of the regression of TBV on EBV were in all cases close to 1 for PB performance (Figure 3; Supplementary Table 6). The regression coefficients for CB performance were in most cases smaller than 1, indicating that the variance of the GEBV tended to be somewhat inflated. Intercepts and regression coefficients for ssGBLUP and ssGBLUP_MF were very similar within the same scenario. Figure 3. View largeDownload slide Bias, defined as the regression slope of the true on the GEBV for purebred selection candidates in generation 9, either for purebred or crossbred performance, using obtained for ssGBLUP models with and without using metafounders for lines with related and unrelated scenarios in purebred and crossbred performances. Red (blue) bars represent models with (without) metafounders. Figure 3. View largeDownload slide Bias, defined as the regression slope of the true on the GEBV for purebred selection candidates in generation 9, either for purebred or crossbred performance, using obtained for ssGBLUP models with and without using metafounders for lines with related and unrelated scenarios in purebred and crossbred performances. Red (blue) bars represent models with (without) metafounders. Convergence of ssGBLUP In the closely related scenario, ssGBLUP and ssGBLUP_MF required a similar number of iterations to reach convergence. In the unrelated scenario ssGBLUP needed substantially more iterations compared to ssGBLUP_MF, resulting in ~30% additional computation time (Figure 4). Figure 4. View largeDownload slide Convergences of ssGBLUP models with and without using metafounders (MFs) for related and unrelated scenarios. Red bars represent models with MF. Figure 4. View largeDownload slide Convergences of ssGBLUP models with and without using metafounders (MFs) for related and unrelated scenarios. Red bars represent models with MF. DISCUSSION The models ssGBLUP and ssGBLUP_MF have been compared in terms of estimated variance components, accuracy, bias, and computational efficiency in order to evaluate the possible benefit of MFs in genomic evaluations for a crossbreeding program. Our results showed that using MF in genomic prediction for CB performance does not affect the prediction accuracies, while it may speed up convergence in specific cases. At the same time, estimated variances for ssGBLUP_MF, after appropriate scaling, were in closer agreement with the empirical true values than ssGBLUP. Relationships Among MFs Models used in breeding value estimation commonly assume that parents with unknown ancestors are sampled from an infinite base population with common genetic variance, and that these base animals are unrelated. In practice, due to pedigree incompleteness, in addition to animals from the oldest generation in the pedigree, in later generations there usually are also animals with unknown ancestors. In this case, animals from the same generation may in fact be more closely related to each other. This is commonly solved by allocating genetic groups to animals with unknown parents that can be grouped based on line, generation, birth date, sex or a combination of these or other factors (Westell et al., 1988). All base animals within the same genetic group are assumed to come from ancestors with similar breeding values, while the animals between genetic groups all have a considered relationship of zero. By using MFs instead of genetic groups, relationships between the pseudo individuals representing genetic groups are computed based on the genotypes of the descendants (Legarra et al., 2015), and used in the model. Because MFs are considered to represent a finite-size pool of gametes, the MFs also have a self-relationship (Legarra et al., 2015). We obtained a self-relationship of the MFs of ~0.17 for the related and ~0.74 for the unrelated scenario. This suggests that the base generation of the related scenario is much more diverse than the base generation of the unrelated scenario. In fact, the base generation of the unrelated scenario had its base generation after 90 generations more of line breeding than the related scenario, and was therefore subject to considerably more accumulated inbreeding. This was reflected in the higher self-relationship of the MF for the unrelated compared to the related scenario. The self-relationship of the MFs in the unrelated scenario is very similar to the values found for pigs (Xiang et al., 2017), and close to the expected value of 2/3 when assuming that base generation allele frequencies are uniformly distributed (see Supplementary Appendix 1). Other reported values in literature varied from values of 0.55 for Holstein and 0.77 for Jersey cattle (Legarra et al., 2015), and 0.30 to 0.47 for dairy goat and sheep (Legarra et al., 2015; Colleau et al., 2017). The latter values are closer to the level of the self-relationship of the MFs in our related scenario, suggesting higher diversity in the base generations of those populations. It should be noted that in all those cases, including our study, a 50k type of chip was used, where the SNPs were selected based on MAF, which is expected to have some impact on the estimated MF relationships. If the relationships among MF would be computed using whole-genome sequence instead, considering that this would have a U-shaped rather than a uniform distribution of allele frequencies, it is expected that higher values would have been obtained in all those cases. Estimated Variance Components The estimated residual variances were similar across the different models and not significantly different from the empirical true values. However, this was not the case for all the estimated genetic variances of the three models. Estimates of the models PBLUP and ssGBLUP should be expressed in an unrelated base population. While the estimated genetic variances for the PBLUP models were similar to the empirical true values, genetic variance estimates for the ssGBLUP model overestimated the empirical true variances. This could be explained by the fact that across-breed allele frequencies and across-breed adjustments of the genomic relationship matrix were used to make it compatible with the pedigree relationship matrix. While such across-breed adjustments may not affect the accuracy (Makgahlela et al., 2014; Lourenco et al., 2016), they may affect the compatibility between the two types of relationships and the estimates of genetic (co)variances (Legarra, 2016a; Wientjes et al., 2017). For the ssGBLUP_MF model, estimated genetic variances were similar to the empirical true genetic variances, after rescaling. Rescaling for the ssGBLUP_MF model was needed because the estimated genetic variance components from the ssGBLUP_MF model are expressed in a hypothetical related base population with allele frequency of 0.5 for all SNPs (Legarra et al., 2015; Garcia-Baccino et al., 2017). Estimated genetic correlations were similar across the three models, even if some deviations were observed from the simulated values. For example, the estimated genetic correlation among the PB lines 2 and 3 especially deviated from the simulated value, most likely because of the weak link between the lines, and because of the limited amount of information available for this particular genetic correlation. On the other hand, estimated genetic correlations between PB and CB performances, for which more information was available, were generally close to the simulated values. For the unrelated scenario, overall ssGBLUP_MF in fact yielded estimated genetic correlations that were closest to the simulated values. This superiority for the unrelated scenario compared to PBLUP may be due to the higher importance of having genomic information to provide stronger links between the different categories of animals, while ssGBLUP_MF additionally profits from making pedigree and genomic relationships better compatible, and therefore may have more correct estimated variance components compared to ssGBLUP. This could be explained by the fact that across-breed allele frequencies and across-breed adjustments of the genomic relationship matrix were used to make it compatible with the pedigree relationship matrix. Based on these results, further studies are required to develop and validate an approach to estimate easily (co)variance components for ssGBLUP_MF in the context of crossbreeding and multivariate evaluations, when switching from PBLUP (or ssGBLUP) routine evaluations to ssGBLUP_MF evaluations. A straightforward approach would be to re-estimate variance components, however, such an approach may be time-consuming. Legarra et al. (2015) proposed an approach to compute variance components for ssGBLUP_MF by scaling the ones from PBLUP (or ssGBLUP) with the following factor: k=1+diag(Γ)¯/2−Γ− ⁠, where the matrix Γ describes the relationships among MFs. According to Legarra et al. (2015), the scaling factor k should be <1, meaning that the genetic variances assuming related founders are larger in comparison to the ones assuming unrelated founders. This is also what we observed for our estimated genetic variances, especially for the unrelated scenario. However, scaling the estimated genetic variances for PBLUP or ssGBLUP as proposed by Legarra et al. (2015) for the related and unrelated scenarios did not result in estimated genetic variances of ssGBLUP_MF that were in agreement with the empirical true values. The scaling factor for the related scenarios was close to 1 (i.e., 0.996), and the one for the unrelated scenario was larger than 1 (i.e., 1.09), meaning that the estimates for the unrelated scenario only deviated more from the empirical true values (results not shown). Based on our results, a third approach could be to compute variance components expressed in a related base population from variance components obtained with PBLUP (or ssGBLUP) and MFs’ relationships. Covariance components could be computed from genetic correlations estimated with PBLUP (or ssGBLUP) and variance components expressed on a related base population. Effect of MFs on Performance of Genomic Evaluations Adding the MF in ssGBLUP did not affect the prediction accuracy. It did reduce the number of iterations until convergence by ~27% for the unrelated scenario. For the unrelated lines, the G and A matrix may be less compatible, because the considered base generation falls after 100th generations of line breeding, compared to only 10 for the related lines. Poor compatibility of G and A may have affected the convergence of ssGBLUP. The use of MFs likely results in a more consistent relationship matrix in ssGBLUP_MF, as it adjusts the base of the pedigree relationships to have the same base as the genomic relationships (Garcia-Baccino et al., 2017). This is the likely explanation for the observation that the use of MFs for the unrelated scenario resulted in improved convergence and estimated genetic variances and genetic correlations that were closer to the simulated values compared with ssGBLUP. Results from Xiang et al., (2017) show that in terms of model-based reliabilities and predictive abilities, ssGBLUP_MF performs at least as well as ssGBLUP using the breed of origin of alleles in the CB animals which requires a step of phasing genotypes and of assigning breed of origin of alleles in CB animals. These additional steps are computationally time-consuming. Use of MFs only requires to compute the relationships among MFs, which can be done using the general least squares estimator of base generation allele frequencies (McPeek et al., 2004; Garcia-Baccino et al., 2017), whose computing time using sparse matrices (Strandén et al., 2017) is trivial relative to all computations needed for ssGBLUP (Aldridge et al., 2018). The ssGBLUP_MF model is therefore more convenient while achieving similar accuracies and biases. Also, while this issue was not considered in this study, fitting genetic groups in ssGBLUP is not as straightforward as for PBLUP, and requires additional computations for the contributions of genotyped animals to genetic groups (Misztal et al., 2013). Using MFs instead only influences the computation of the inverse of pedigree-based relationship matrix. Finally, in genomic evaluations with multiple lines or breeds it is not easy to scale G and A properly (Legarra et al., 2015), unless relationships are dissected by breed of origin (Christensen et al., 2014, 2015), but this is straightforward with the use of MF. Therefore, there are several advantages and no clear obstructions to use MFs in genomic evaluations, and especially in crossbreeding schemes. CONCLUSIONS Based on the results in our study, the ssGBLUP model using MFs is the preferred model for implementation of genomic prediction for CB performance in practical breeding programs. The MFs can easily accommodate for differences in base populations for different lines involved, as the genomic and pedigree relationships are compatible by construction. In comparison to ssGBLUP, this leads, potentially, to improved convergence behavior of the iterative solver, without affecting the prediction accuracies. Our results also suggest that rescaled variance components estimated with ssGBLUP_MF may be more accurate than those of ssGBLUP. Further studies are needed for developing and validating approaches to easily compute or approximate variance component estimates for ssGBLUP_MF. Conflict of interest statement. None declared. ACKNOWLEDGMENTS This article is part of the project Feed-a-Gene. Mario Calus and Jérémie Vandenplas are financially supported by the Dutch Ministry of Economic Affairs (TKI Agri & Food project 16022) and the Breed4Food partners Cobb Europe, CRV, Hendrix Genetics, and Topigs Norsvin. The use of the HPC cluster has been made possible by CAT-AgroFood (Shared Research Facilities Wageningen UR). REFERENCES Aguilar , I. , I. Misztal , D. L. Johnson , A. Legarra , S. Tsuruta , and T. J. Lawlor . 2010 . Hot topic: a unified approach to utilize phenotypic, full pedigree, and genomic information for genetic evaluation of Holstein final score . J. Dairy Sci . 93 : 743 – 752 . doi: https://doi.org/10.3168/jds.2009–2730 Google Scholar Crossref Search ADS PubMed Aldridge , M. N. , J. Vandenplas , and M. P. L. Calus . 2018 . Efficient and accurate computation of base generation allele frequencies . J. Dairy Sci . doi:10.3168/jds.2018-15264. Calus , M. P. L. , and J. Vandenplas . 2016 . Calc_grm – a program to compute pedigree, genomic, and combined relationship matrices . ABGC, Wageningen UR Livestock Research . Christensen , O. F . 2012 . Compatibility of pedigree-based and marker-based relationship matrices for single-step genetic evaluation . Genet. Sel. Evol . 44 : 37 . doi: https://doi.org/10.1186/1297-9686-44-37 Google Scholar Crossref Search ADS PubMed Christensen , O. F., A. Legarra , M. S. Lund , and G. Su . 2015 . Genetic evaluation for three-way crossbreeding . Genet. Sel. Evol . 47 : 98 . doi: https://doi.org/10.1186/s12711-015-0177-6 Google Scholar Crossref Search ADS PubMed Christensen , O. F. , and M. S. Lund . 2010 . Genomic prediction when some animals are not genotyped . Genet. Sel. Evol . 42 : 2 . doi: https://doi.org/10.1186/1297-9686-42-2 Google Scholar Crossref Search ADS PubMed Christensen , O. F., P. Madsen , B. Nielsen , and G. Su . 2014 . Genomic evaluation of both purebred and crossbred performances . Genet. Sel. Evol . 46 : 23 . doi: https://doi.org/10.1186/1297-9686-46-23 Google Scholar Crossref Search ADS PubMed Colleau , J. J., I. Palhière , S. T. Rodríguez-Ramilo , and A. Legarra . 2017 . A fast indirect method to compute functions of genomic relationships concerning genotyped and ungenotyped individuals, for diversity management . Genet. Sel. Evol . 49 : 87 . doi: https://doi.org/10.1186/s12711-017-0363-9 Google Scholar Crossref Search ADS PubMed Garcia-Baccino , C. A., A. Legarra , O. F. Christensen , I. Misztal , I. Pocrnic , Z. G. Vitezica , and R. J. Cantet . 2017 . Metafounders are related to F st fixation indices and reduce bias in single-step genomic evaluations . Genet. Sel. Evol . 49 : 34 . doi: https://doi.org/10.1186/s12711-017-0309-2 Google Scholar Crossref Search ADS PubMed Legarra , A . 2016a . Comparing estimates of genetic variance across different relationship models . Theor. Popul. Biol . 107 : 26 – 30 . doi: https://doi.org/10.1016/j.tpb.2015.08.005 Google Scholar Crossref Search ADS Legarra , A . 2016b . createHmf . https://github.com/alegarra/metafounders – ( accessed April 3 2017 ). Legarra , A. , O. F. Christensen , I. Aguilar , and I. Misztal . 2014 . Single step, a general approach for genomic selection . Livest. Sci . 166 : 54 – 65 . doi: https://doi.org/10.1016/j.livsci.2014.04.029 Google Scholar Crossref Search ADS Legarra , A., O. F. Christensen , Z. G. Vitezica , I. Aguilar , and I. Misztal . 2015 . Ancestral relationships using metafounders: finite ancestral populations and across population relationships . Genetics . 200 : 455 – 468 . doi: https://doi.org/10.1534/genetics.115.177014 Google Scholar Crossref Search ADS PubMed Lourenco , D. A., S. Tsuruta , B. O. Fragomeni , C. Y. Chen , W. O. Herring , and I. Misztal . 2016 . Crossbreed evaluations in single-step genomic best linear unbiased predictor using adjusted realized relationship matrices . J. Anim. Sci . 94 : 909 – 919 . doi: https://doi.org/10.2527/jas.2015-9748 Google Scholar Crossref Search ADS PubMed Makgahlela , M. L., I. Strandén , U. S. Nielsen , M. J. Sillanpää , and E. A. Mäntysaari . 2014 . Using the unified relationship matrix adjusted by breed-wise allele frequencies in genomic evaluation of a multibreed population . J. Dairy Sci . 97 : 1117 – 1127 . doi: https://doi.org/10.3168/jds.2013-7167 Google Scholar Crossref Search ADS PubMed McPeek , M. S., X. Wu , and C. Ober . 2004 . Best linear unbiased allele-frequency estimation in complex pedigrees . Biometrics . 60 : 359 – 367 . doi: https://doi.org/10.1111/j.0006-341X.2004.00180.x Google Scholar Crossref Search ADS PubMed Misztal , I. , S. Tsuruta , T. Strabel , B. Auvray , T. Druet , and D. H. Lee . 2002 . BLUPF90 and related programs (BGF90) . In: Proceedings of the 7th World Congress on Genetics Applied to Livestock Production , Montpellier, France ; p. 743 – 744 . Misztal , I., Z. G. Vitezica , A. Legarra , I. Aguilar , and A. A. Swan . 2013 . Unknown-parent groups in single-step genomic evaluation . J. Anim. Breed. Genet . 130 : 252 – 258 . doi: https://doi.org/10.1111/jbg.12025 Google Scholar Crossref Search ADS PubMed ten Napel , J. , J. Vandenplas , M. Lidauer , I. Stranden , M. Taskinen , E. Mäntysaari , M. P. L. Calus , and R. F. Veerkamp . 2017 . MiXBLUP, user-friendly software for large genetic evaluation systems – manual V2.1-2017-08 , Wageningen, the Netherlands . Powell , J. E., P. M. Visscher , and M. E. Goddard . 2010 . Reconciling the analysis of IBD and IBS in complex trait studies . Nat. Rev. Genet . 11 : 800 – 805 . doi: https://doi.org/10.1038/nrg2865 Google Scholar Crossref Search ADS PubMed Raymond , M. , and F. Rousset . 1995 . An exact test for population differentiation . Evolution . 49 : 1280 – 1283 . doi: https://doi.org/10.1111/j.1558-5646.1995.tb04456.x Google Scholar Crossref Search ADS PubMed Rousset , F . 2008 . Genepop’007: a complete re-implementation of the genepop software for windows and linux . Mol. Ecol. Resour . 8 : 103 – 106 . doi: https://doi.org/10.1111/j.1471-8286.2007.01931.x Google Scholar Crossref Search ADS PubMed Sargolzaei , M. , and F. S. Schenkel . 2009 . Qmsim: a large-scale genome simulator for livestock . Bioinformatics . 25 : 680 – 681 . doi: https://doi.org/10.1093/bioinformatics/btp045 Google Scholar Crossref Search ADS PubMed Strandén , I., K. Matilainen , G. P. Aamand , and E. A. Mäntysaari . 2017 . Solving efficiently large single-step genomic best linear unbiased prediction models . J. Anim. Breed. Genet . 134 : 264 – 274 . doi: https://doi.org/10.1111/jbg.12257 Google Scholar Crossref Search ADS PubMed VanRaden , P. M . 2018 . Efficient methods to compute genomic predictions . J. Dairy Sci . 91 : 4414 – 4423 . doi: https://doi.org/10.3168/jds.2007-0980 Google Scholar Crossref Search ADS Vitezica , Z. G., I. Aguilar , I. Misztal , and A. Legarra . 2011 . Bias in genomic predictions for populations under selection . Genet. Res. (Camb.) 93 : 357 – 366 . doi: https://doi.org/10.1017/S001667231100022X Google Scholar Crossref Search ADS PubMed Wei , M. , and J. H. van der Werf . 1995 . Genetic correlation and heritabilities for purebred and crossbred performance in poultry egg production traits . J. Anim. Sci . 73 : 2220 – 2226 . doi: https://doi.org/10.2527/1995.7382220x Google Scholar Crossref Search ADS PubMed Westell , R. A. , R. L. Quaas , and L. D. Vanvleck . 1988 . Genetic groups in an animal model . J. Dairy Sci . 71 : 1310 – 1318 . doi: https://doi.org/10.3168/jds.S0022-0302(88)79688-5 Google Scholar Crossref Search ADS Wientjes , Y. C. J., P. Bijma , J. Vandenplas , and M. P. L. Calus . 2017 . Multi-population genomic relationships for estimating current genetic variances within and genetic correlations between populations . Genetics . 207 : 503 – 515 . doi: https://doi.org/10.1534/genetics.117.300152 Google Scholar PubMed Wientjes , Y. C. J. , and M. P. L. Calus . 2017 . Board invited review: the purebred-crossbred correlation in pigs: a review of theory, estimates, and implications . J. Anim. Sci . 95 : 3467 – 3478 . doi: https://doi.org/10.2527/jas.2017.1669 Google Scholar PubMed Xiang , T., O. F. Christensen , and A. Legarra . 2017 . Technical note: genomic evaluation for crossbred performance in a single-step approach with metafounders . J. Anim. Sci . 95 : 1472 – 1480 . doi: https://doi.org/10.2527/jas.2016.1155 Google Scholar PubMed Footnotes The project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement no. 633531. © The Author(s) 2018. Published by Oxford University Press on behalf of the American Society of Animal Science. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Ring, Siobhán, C;Graham, David, A;Kelleher, Margaret, M;Doherty, Michael, L;Berry, Donagh, P

doi: 10.1093/jas/sky430pmid: 30412254

Abstract Genetic selection is an inexpensive and complementary strategy to traditional methods of improving animal health and welfare. Nonetheless, endeavors to incorporate animal health and welfare traits in international breeding programs have been hampered by the availability of informative phenotypes. The recent eradication program for bovine viral diarrhea (BVD) in the Republic of Ireland has provided an opportunity to quantify the potential benefits that genetic selection could offer BVD eradication programs elsewhere, as well as inform possible eradication programs for other diseases in the Republic of Ireland. Using a dataset of 188,085 Irish calves, the estimated direct and maternal heritability estimates for the birth of persistently infected calves following likely in utero exposure to BVD virus ranged from not different from zero (linear model) to 0.29 (SE = 0.075; threshold model) and from essentially zero (linear model) to 0.04 (SE = 0.033; threshold model), respectively. The corresponding genetic SD for the direct and maternal effect of the binary trait (0, 1) ranged from 0.005 (linear model) to 0.56 (threshold model) units and ranged from 0.00008 (linear model) to 0.20 (threshold model) units, respectively. The coefficient of direct genetic variation based on the linear model was 2.56% indicating considerable genetic variation could be exploited. Based on results from the linear model in the present study, there is the potential to reduce the incidence of persistent infection in cattle by on average 0.11 percentage units per year which is cumulative and permanent. Therefore, genetic selection can contribute to reducing the incidence of persistent infection in cattle. Moreover, where populations are free from persistent infection, inclusion of the estimated genetic merit for BVD in national breeding indexes could contribute to a preservation of a BVD-free status. Results from the present study can be used to inform breeding programs of the potential genetic gains achievable. Moreover, the approaches used in the present study can be applied to other diseases when data become available. INTRODUCTION Bovine viral diarrhea (BVD) which is caused by BVD virus in the genus pestivirus, family Flaviviridae, is a contagious viral disease of cattle with global significance (Houe, 1999; Gunn et al., 2005). Production losses associated with BVD infection cost Irish dairy and beef producers an estimated €102 million per annum (Stott el al., 2012). Therefore, it is not surprising that BVD is ranked as one of the most important endemic infectious diseases affecting the productivity and international competitiveness of the Irish livestock industry (More et al., 2010). Infection with BVD virus in cattle manifests as either persistent or transient infection; the type of infection that develops, and the clinical signs that ensue, are dependent on the timing of infection relative to conception (Grooms, 2004; Hansen et al., 2010), the immune-competence status of the developing fetus (Smirnova et al., 2012), the virus biotype (Bachofen et al., 2010), and the virulence of the virus (Grooms, 2004). With the exception of the in utero development period, an animal infected with BVD virus at any stage of its lifetime (i.e., from birth to death) becomes transiently infected; such animals shed BVD virus for a short period before a specific antibody response develops and these animals typically develop life-long immunity to BVD virus infection (Brownlie et al., 1987; Lindberg and Houe, 2005). Many transient infections are subclinical (Brownlie et al., 1987; Barrett et al., 2011), but there is evidence that immunosuppression, arising from transient infection, can lead to secondary bacterial and/or viral illnesses such as diarrhea, respiratory disease, and reproductive disorders (Houe, 2003; Grooms, 2004). Persistently infected animals are only created when the unborn fetus is infected with BVD virus between ~30 and 120 d of gestation (Grooms, 2004; Hansen et al., 2010; Smirnova et al., 2012). Persistently infected animals shed BVD virus throughout their life and are the main source of inter-animal transmission of BVD virus (Grooms, 2004; Moennig et al., 2005). Persistently infected animals may appear without clinical signs or they may succumb to mucosal disease, characterized by ill thrift, oral lesions, lameness, and death. It is also well recognized that persistently infected animals are immunosuppressed, have greater susceptibility to other diseases, and often die before reaching maturity (Houe, 2003; Moennig et al., 2005). BVD virus has been either successfully eliminated or the prevalence of persistently infected cattle has substantially reduced in several countries (e.g., Norway, Sweden, Denmark, Austria, Switzerland, and more recently the Republic of Ireland) and specific regions of France, Germany, The Netherlands, Italy, and United Kingdom through the implementation of BVD virus eradication schemes (Stahl and Alenius, 2012; Reardon et al., 2018); such schemes typically target the identification and culling of persistently infected animals, followed by subsequent surveillance testing and biosecurity measures to prevent the re-introduction of BVD virus. New strains of the virus, as well as the live-trade of pregnant animals, continue to pose a significant threat for re-introduction of the disease (Houe, 1999). Such a threat materialized in the Shetland Islands which had achieved a BVD-free status, but lost this status after the purchase of a persistently infected animal (Barrett et al., 2011). Genetic variation is known to exist in many health traits in cattle (Berry et al., 2011). However, there is no information on the extent of genetic variability in the susceptibility of cattle to BVD virus. The objective of the present study was to quantify the extent of genetic variability among Irish dairy and beef calves with regards them being born persistently infected following likely exposure to BVD virus in utero. The results in this instance will be useful in determining the feasibility of genetic selection for cattle to reduce the incidence of persistently infected progeny (both in Ireland and elsewhere) through direct selection of animals (i.e., dams and/or sires) as parents of the next generation. Moreover, breeding programs that exploit the known inherent genetic variability could help to maintain a BVD-free status where persistent infection has been eradicated. In addition, the methods applied in the present study can be used when considering forthcoming eradication programs for other diseases internationally (e.g., Infectious Bovine Rhinotracheitis or Johne’s disease). MATERIALS AND METHODS Data Since 2013, an industry-led compulsory national BVD eradication program has been operating in the Republic of Ireland; this program commenced after a voluntary program in 2012. As part of the eradication program, producers are required by law to obtain an ear biopsy sample from each calf (i.e., alive, aborted, or stillborn calf) ≤20 d after birth (Anon, 2018a,b). Using either an ELISA or a reverse transcription polymerase chain reaction, ear biopsy samples are tested for the presence of BVD virus in one of 14 designated laboratories. Test results are classified as positive, inconclusive, or negative based on the manufacturers’ guidelines for the respective tests. Further details on the eradication program are published elsewhere (Clegg et al., 2016). For the purpose of the present study, 11,643,625 BVD virus test results were obtained from the Irish Cattle Breeding Federation’s database on 11,553,516 cattle from 89,686 Irish bovine herds from 1 January 2012 to 17 November 2017, inclusive. Tests undertaken using the ELISA method constituted over half (i.e., 56%) of the tests. Prior to trait definition, of the 3,387 animals that yielded two positive test results within 21 d of each other, the latter test result for each animal was discarded to allow for differentiation between transient and persistent infection. In addition, 1,296 inconclusive test results were not considered further. Moreover, data from one laboratory which tested <0.01% of samples (i.e., 83 samples) were also removed. Phenotype Definition Animals with BVD virus test result(s) were defined as either persistently infected (PI; PI = 1) or not persistently infected (PI = 0). Where each test result of an animal yielded a positive result, irrespective of the number of tests available for the animal, that animal was deemed persistently infected (PI = 1). Where each test result of an animal yielded a negative result, that animal was deemed not persistently infected (PI = 0). Where an animal yielded both positive and negative test results, animals were deemed not persistently infected (PI = 0), provided only 1 positive test result existed; otherwise, animals were not deemed as either persistently infected or not persistently infected and these animals were later removed from the dataset. The phenotype of each animal’s dam was also defined, where possible, using her own BVD virus test result(s) by the aforementioned criteria; otherwise, where a dam did not have a BVD virus test result, the phenotype of her progeny were used. A cow that produced ≥1 not persistently infected progeny was herself deemed not persistently infected (PI = 0). The phenotype of a cow that only produced persistently infected progeny was not defined and that cow’s progeny were later removed from the dataset. Exposure Definition As nulliparous females are generally managed separately to multiparous cows on Irish farms, exposure to BVD virus in utero was defined within herd according to the management group of the animal’s dam at conception. Potential exposure to BVD virus was defined for all animals except those born to a purchased dam. Animals were deemed potentially exposed to BVD virus in utero if they were born ±90 d of a persistently infected (PI = 1) contemporary in the same herd-management group. Data Pruning Animals born from multiple births (435,297 animals) were removed from the dataset. Animals for which test results were reported for the first time >30 d of age were discarded (1,018,021 animals) as were a further 744 calves not tested in their birth herd. Moreover, to maximize the likelihood of equal lifetime exposure to BVD virus among herd mates, animals born to purchased dams were removed (i.e., both the animal and its dam were not born in the same herd; 3,092,837 calves). A total of 186 remaining calves without a defined BVD phenotype were also discarded. In addition, 2,242 calves born to dams of unknown status or dams deemed to be persistently infected (PI = 1) were omitted from the dataset as it was assumed that a PI dam always produces a PI calf (Brownlie et al., 1987). Calves from primiparous dams that calved <545 d of age (i.e., 15 months; 6,892 calves) were also removed as were calves from multiparous dams that calved >666 d (i.e., 22 months; 174,240 calves) from the parity median. Furthermore, 2,025,221 calves with an unknown sire were removed as were a further 828,653 calves with an unknown maternal grand-sire. Heterosis and recombination loss coefficients for each animal and its dam were derived using methods described by VanRaden and Sanders (2003). Pedigree information for each animal was traced back to founder animals (where possible). In addition, only animals deemed potentially exposed to BVD virus (previously defined) were considered while exposure groups with <5 calves were omitted from the analyses. Following edits, 188,085 calves with a BVD status phenotype from 5,399 herds remained. Statistical Analyses Variance components for the binary trait of persistently infected or not persistently infected were estimated in ASReml version 4.1 using both a logit threshold model and a linear mixed model for all 188,085 calves in the one analysis (Gilmour et al., 2009). The fitted model can be written in matrix form as y=Xβ+Zsire+Zmgs+Zpe+e where y is the vector of the trait observed (i.e., persistently infected or not persistently infected); β is the vector of fixed effects; sire is the vector of direct additive genetic effects of the animal’s sire; mgs is the vector of direct additive genetic effects of the animal’s maternal grand-sire; pe is a vector of permanent environmental effects of the animal’s dam; e is the vector of residual effects; X is the incidence matrix for fixed effects; and Z is the incidence matrix relating to the random additive genetic effects. The fixed effects included in the model were exposure group, heterosis coefficient of the animal (categorized as 0.00, 0.01 to 0.09, 0.10 to 0.19, …0.90 to 0.99 and 1.00), heterosis coefficient of the animal’s dam (categorized as 0.00, 0.01 to 0.09, 0.10 to 0.19, …0.90 to 0.99, and 1.00), recombination loss coefficient of the animal (categorized as 0.00 to 0.09, 0.10 to 0.29, 0.30 to 0.49, and ≥0.50), recombination loss coefficient of the animal’s dam (categorized as 0.00 to 0.09, 0.10 to 0.29, 0.30 to 0.49, and ≥0.50), animal gender, and the interaction between parity of the animal’s dam when the animal was born (i.e., 1, 2, 3, 4, and ≥5) and the age (in months) of the animal’s dam at calving relative to the parity median. Heritability was calculated as (4*additive genetic variance of the animal’s sire)/(phenotypic variance) for the direct estimate and (4*additive genetic variance of the animal’s maternal grand sire)/(phenotypic variance) for the maternal estimate. RESULTS Descriptive Statistics and Fixed Effects Of the 188,085 calves used to estimate variance components, 7,780 (i.e., 4.14%) calves were deemed persistently infected; 11.82% (of the 188,085 calves) were born to a dam that had >1 progeny in the dataset. Even after edits were applied, the majority (i.e., 74%) of exposure groups included only one calf deemed persistently infected. The apparent prevalence of persistent infection was greatest in beef-breed calves compared to dairy-breed calves (Figure 1). In addition, the apparent prevalence of persistent infection was greatest in progeny born to first parity dams (i.e., 5.60%) and lowest for progeny born to fifth or greater parity dams (i.e., 2.36%; Figure 1). Relative to a first parity cow, the odds of a third, fourth, and fifth parity cow producing a persistently infected calf was 0.74 (95% CI: 0.68 to 0.81), 0.63 (95% CI: 0.57 to 0.71), and 0.39 (95% CI: 0.35 to 0.44), respectively (P < 0.001); there was no difference in the odds of a second parity cow producing a persistently infected calf relative to a first parity cow. No other fixed effect considered in the mixed model was associated with persistent infection (P > 0.05). Figure 1. View largeDownload slide Prevalence (%) and number of beef (white) and dairy (black) breed animals (in brackets) deemed persistently infected when born to first, second, third, fourth, and ≥5th parity dams. Figure 1. View largeDownload slide Prevalence (%) and number of beef (white) and dairy (black) breed animals (in brackets) deemed persistently infected when born to first, second, third, fourth, and ≥5th parity dams. Variance Components An indication of underlying genetic variability for persistent infection is captured in Figure 2 which illustrates the distribution of the mean progeny prevalence of persistent infection from sires that had at least 50 progeny deemed exposed to BVD virus in utero; those bulls sired progeny in at least five distinct herds. The average prevalence of persistent infection in the progeny of those sires when born to first, second, third, fourth, and fifth parity or greater dams was 6%, 4%, 3%, 2%, and 1%, respectively. Nonetheless, one bull sired a total of 146 calves, of which 63 were born to primiparous dams while the remaining 83 were born to multiparous dams. Of the 63 calves born to primiparous dams, 40% were deemed persistently infected. The average prevalence of persistent infection among his 83 progeny that were born to multiparous dams was still 3.5 to six times higher than the average prevalence of persistent infection of progeny born to dams of the same parity. Interestingly, based on estimated breeding values, derived by the ICBF (http://www.icbf.com; May 2018), for both calf mortality and cow survival, that same AI sire was ranked in the lowest (i.e., worst) 12th percentile (EBV reliability 99%) and 10th percentile (EBV reliability 88%) of all beef breeds, respectively; although not investigated in the present study, and results should thus be interpreted with caution, these results suggest that bulls which are genetically more likely to produce calves born persistently infected are also those that are most likely to produce calves that tend to die at birth or result in cows with reduced longevity. Figure 2. View largeDownload slide Distribution of the mean prevalence of persistently infected calves, born to first (black), second (white), third (grey), fourth (striped), and ≥5th (speckled) parity dams, in the progeny of sires that produced at least 50 progeny (per dam parity) in at least five herds (per dam parity) where progeny were deemed exposed to BVD virus in utero. Figure 2. View largeDownload slide Distribution of the mean prevalence of persistently infected calves, born to first (black), second (white), third (grey), fourth (striped), and ≥5th (speckled) parity dams, in the progeny of sires that produced at least 50 progeny (per dam parity) in at least five herds (per dam parity) where progeny were deemed exposed to BVD virus in utero. The direct and maternal heritability estimates for the birth of persistently infected cattle following likely in utero exposure to BVD virus based on the threshold model were 0.29 (SE = 0.075) and 0.04 (SE = 0.033), respectively. The corresponding genetic SD for the direct and maternal effect was 0.56 and 0.20 units, respectively. The estimated direct and maternal heritability estimates using the linear model for the birth of persistently infected cattle following likely in utero exposure to BVD virus were 0.0007 (SE = 0.0012) and zero, respectively. The corresponding genetic SD for the direct and maternal effect was 0.005 and 0.00008 units, respectively. The coefficient of direct genetic variation based on the linear model was 2.56%, indicating that considerable exploitable genetic variation does exist, despite the low heritability. Irrespective of the whether the linear or threshold model was used, the absence of a maternal permanent environmental variance (P > 0.05) suggests no repeatability of persistent infection within dam (i.e., a dam that produces a persistently infected calf is unlikely to produce persistently infected progeny in subsequent calvings). DISCUSSION Breeding to Improve Animal Health and Welfare Genetic selection for (re)production in cattle (Berry et al., 2014; Berry, 2017), and other species (Havenstein et al., 2003; Berry, 2017), is responsible for a considerable proportion of observed on-farm changes. Although there are exceptions (e.g., Nordic countries), the same genetic gain for health and welfare traits in cattle have generally not been realized. The reason for the lag in genetic gain is not, however, due to a lack of genetic differences among individuals (i.e., genetic variability) per se, but more likely a lack of routine access to informative data (Berry et al., 2011) to accurately distinguish the genetically elite from the genetically inferior animals. To establish the potential rate of genetic gain achievable, knowledge of the extent of genetic variability is required; variance components are also necessary to inform the mixed model equations for the estimation of breeding values. The heritability statistic is also useful to determine the number of records required to achieve accurate estimates of genetic merit. Put simply, the presence of large genetic variability suggests greater genetic differences among individuals, and thus (all else being equal) faster genetic gain can be achieved. The higher the heritability, the fewer the records required to achieve that potential gain with a high degree of precision; nonetheless, low heritability traits can achieve the same degree of precision as high heritability traits, provided ample data are available (Berry et al., 2011). In any case, measures of genetic variability, genetic merit or heritability can only be generated from informative phenotypes. That said, with the increasing international pressure to improve overall animal health and welfare standards, the abundance of (disease) phenotypes that may be generated from research studies can be used to assess the potential gains achievable through breeding; results will determine the desirability and feasibility of considering such traits in national breeding programs as part of national disease eradication programs as well as maintenance of a disease-free status. Based on results from the linear model in the present study, there is the potential to reduce the incidence of persistent infection in cattle by on average 0.11 percentage units per year (based on an annual genetic gain of 0.215 SDs; Schaeffer, 2006); with just 100 half-sib progeny records, a reduction in the incidence of persistent infection of 0.017 percentage units per year could be achieved with an accuracy of selection of 13%. Variance Components Although to our knowledge the present study is the first to quantify variance components for persistent infection, genetic variance and heritability estimates for bovine respiratory disease (BRD) have been documented when linear models were used (Muggli-Cockett et al., 1992; Snowder et al., 2005, 2006, 2010), although no estimates exist from threshold models. When results from the linear model were compared to those reported elsewhere (Muggli-Cockett et al., 1992; Snowder et al., 2005, 2006, 2010), both the non-significant heritability estimate and the genetic variance were lower in the present study. For example, using a mixed linear sire model, Schneider et al. (2010) reported a heritability estimate on the observed scale for BRD, classified as a binary trait (i.e., not treated or treated for respiratory reasons), ranging from 0.07 (SE = 0.04) to 0.11 (SE = 0.06) for preweaned and feedlot U.S. beef cattle; the direct genetic SD in that study ranged from 0.036 to 0.094 while neither a maternal genetic or a maternal permanent effect were considered in those genetic analyses. Based on an animal linear model, Snowder et al. (2005) reported direct heritability estimates (P < 0.05) ranging from 0.09 to 0.22 and maternal heritability estimates ranging from 0.00 (SE = 0.02) to 0.13 (SE = 0.07) for BRD, also classified as a binary trait (i.e., healthy or affected by BRD) in pre-weaned beef calves at the U.S. Meat Animal Research Centre; the direct genetic SD in that study ranged from 0.035 to 0.135. Snowder et al. (2005) also noted no dam permanent environmental variance, consistent with the results of the present study, while the dam genetic effect was only significant (P < 0.05) in a few of the beef breeds investigated. The estimates resulting from linear models of categorical traits are incidence dependent; therefore, our results based on the linear model are likely not directly comparable with other genetic studies on BRD (Muggli-Cockett et al., 1992; Snowder et al., 2005, 2006, 2010) which reported a higher disease incidence (ranging from 8.3% to 23.9%) than the present study (4.14%). Threshold models endeavor to account for differences in trait incidence (Gianola and Foulley, 1983), thus they are expected to generate more unbiased estimates than linear models. Using both the linear model heritability estimates as well as the mean incidence of BRD reported in other studies, heritability estimates were transformed to the underlying liability scale to mimic threshold models as described by Robertson and Lerner (1949). When the respective heritability estimates of studies (that were different from zero) were transformed to the underlying scale, the heritability estimates were 0.18 for Snowder et al. (2006) and ranged from 0.20 to 0.65 for Snowder et al. (2005), making them more in agreement with our estimate derived from the threshold model (i.e., 0.29). When the heritability estimate from the threshold model in the present study was converted to the observed scale the heritability estimate was 0.06. These results suggest that the estimates from the threshold model in the present study better reflect the variance components of viral diseases in cattle than the linear model, especially when the incidence of disease is low. Evidence of Possible Genetic Variability in Establishment of Persistent Infection in the Fetus The mechanism of generating persistent infections in the fetus involves a complexity of interactions between the pregnant cow, the growing fetus, and the connecting placenta (Hansen et al., 2015). Cattle have a synepitheliochorial placenta, which inhibits contact between maternal and fetal circulatory systems (Chucri et al., 2010); as a result, the transfer of antibodies to the fetus is prevented (Tizard, 2013) yet maternal viral infections, including BVD virus, can be transferred across the placenta to the fetus (Hansen et al., 2015). For many decades, it has been hypothesized that the generation of persistent infection requires firstly for BVD virus to evade the innate immune system of the pregnant cow, secondly for BVD virus to evade the adaptive system of the pregnant cow, and finally, for BVD virus to successfully establish within the fetus without actually killing the fetus (Smirnova et al., 2012). Due to immature stage of development of the fetal immune system, it has been assumed (Grooms, 2004; Lanyon et al., 2014) that the fetus cannot recognize BVD virus as foreign; thus, the fetus may not produce an effective humoral or cellular adaptive immune response (Peterhans and Schweizer, 2010). The lack of a functioning immune system at this stage is believed to result in immune-tolerance enabling the virus to establish in the fetus resulting in persistent infection (Hansen et al., 2010). Nevertheless, recent studies (Smirnova et al., 2012, 2014; Hansen et al., 2015) have detected both fetal innate and adaptive immune responses following inoculation of their pregnant dams with BVD virus during the first trimester of gestation. In each of these studies (Smirnova et al., 2012, 2014; Hansen et al., 2015), fetal viremia was reduced following a fetal immune response and this suggests that although the fetus was unable to clear the virus completely, the developing fetus was competent and functioning in the first trimester of gestation, which is contrary to what was believed for many years. Clearance of BVD virus in a developing fetus may have occurred in a study by McClurkin et al. (1984) who demonstrated that following injection of two fetuses with BVD virus at 125 d of gestation, one calf seroconverted in utero (i.e., developed antibodies to BVD virus) thus becoming transiently infected while the other calf did not seroconvert (i.e., no antibodies to BVD virus were developed) but became persistently infected. Similarly, Stokstad and Løken (2002) noted that four calves born following intranasal infection of pregnant dams on days 74 to 82 of pregnancy were not persistently infected while 18 calves born to their contemporaries infected during the same period of gestation were persistently infected. Moreover, Schoder et al. (2004) reported the birth of two dizygotic twins, one of which was born persistently infected while the other was not persistently infected; their dam and the twin that was not persistently infected both tested antibody positive, indicating that they had been exposed to the virus. Each of these studies indicates possible variability in immune system development between fetuses and a possible explanation for a large genetic variance but much smaller maternal genetic variance. In addition, the chain of events that are required for the establishment and birth of a persistently infected calf (e.g., for the dam to become infected, for the virus to reach and cross the placenta, for the fetus to become infected and the fetus not to mount an immune response) may each be subject to variability among individuals, some of which (as quantified by the present study) is likely due to genetic variation under the control of both the developing fetus in utero and its dam. Following on from the present study, it could be interesting to in utero estimate breeding values for BVD of fetuses at conception and subsequently challenge genetically divergent fetuses (and their dams) with BVD virus. Should differences in the observed prevalence of persistent infection actually exist in the resulting calves born, then in depth investigation on the rational for such differences could aid a better understanding of the mechanisms of the virus and the factors contributing to differences in susceptibility. Limitations of the Present Study Although the size of dataset used in the present study was large (relative to other genetic studies on viral diseases), the study is not without its limitations. For example, the mixed model used in the present study, which is used internationally for genetic analyses, assumes that after accounting for herd and other systemic environmental effects (e.g., dam parity), all individuals within a herd-group (e.g., herd parity) are managed uniformly (e.g., all animals of the same parity within a herd are subject to the same vaccination protocol). Although this assumption is generally true in Irish production systems, there are always exceptions to the norm. Such inconsistencies do not perform well in mixed model analysis as their effects are not accounted for; the resulting estimates from such analyses have large residual variation. Large residual variation was observed in the present study when the linear model was used. Much of these inaccuracies in uniformity can be overcome when the prevalence of disease is high or the number of records available is large; however, in the present study, the animal-level apparent prevalence was low (4.14% after edits). In addition, only 26% of exposure groups included in the present study had two or more persistently infected calves. When the analyses were restricted to include only animals deemed exposed by ≥2 animals that were defined as persistently infected, only 49% (91,877 records) of the initial dataset were available with an animal-level apparent prevalence of 4.55%; in this scenario using the linear model none of the observed variation for persistent infection could be attributed to genetic effects. Moreover, incorrect parentage deflates heritability by p2, where p is the proportion of animals with correctly identified sires (Van Vleck, 1970). Therefore, the true heritability for persistent infection in the present study, after accounting for pedigree registration errors (Purfield et al., 2016), ranges from 0.0009 (linear model) to 0.39 (threshold model). In addition, the use of imperfect tests contributes to an underestimation of genetic variation and subsequent heritability (Bishop and Woolliams, 2010). According to Presi et al. (2011), sensitivity and specificity values for BVD virus ELISA tests are 98% and 99.8%, respectively; sensitivity and specificity values for BVD virus reverse transcription polymerase chain reaction tests are 97.1% and 100%, respectively (Presi et al., 2011). Coupled with the issue of imperfect tests is that just 52% of animals deemed persistently infected in the present study had a confirmatory positive test result; the remaining animals were likely culled immediately after the initial positive test result or stillborn. Moreover, the phenotype considered in the present study considered only calves carried to full-term, in the context of the national eradication program in the Republic of Ireland. The present study did not have access to phenotypes from aborted fetuses which may also have been persistently infected. In addition, calving dates, and subsequent conception dates which were used as a proxy to determine likely exposure to BVD virus in utero could contain error. Moreover, incomplete exposure, a topic explored in detail by Bishop and Woolliams (2010, 2014), may also have biased results from the present study. In a scenario of incomplete exposure, uninfected animals may represent both animals that are resistant to a pathogen and animals that are not resistant to that same pathogen but have yet to be exposed to an infectious dose (Bishop and Woolliams, 2014). Such a scenario will bias heritability estimates downwards by ε, where ε is the proportion of the population exposed to the pathogen (Bishop and Woolliams, 2014). Twomey et al. (2016) applied the formula reported by Bishop and Woolliams (2010) to determine the impact of imperfect exposure on heritability estimates using a phenotype of liver damage caused by F. hepatica in Irish cows; if the exposure probability were 0.6, 0.7. 0.8, and 0.9, the true heritability of liver damage on the observed scale would be 0.10 (SE = 0.20), 0.06 (SE = 0.10), 0.04 (SE = 0.06), and 0.03 (SE = 0.05), respectively (Twomey et al., 2016). Therefore, the multiple inaccuracies in the recording of phenotypes and pedigree relationships as well as imperfect tests and exposure are likely to have contributed to a deflated heritability estimate. While the present study did not have access to herd-level vaccination usage, variability in the birth of persistently infected calves may differ among animals in vaccinated and non-vaccinated herds. For the estimation of variance components of another immune response trait that results in respiratory disease (immune response to bovine herpesvirus) in Irish cattle, Ring et al. (2018) analyzed vaccinated herds separately from non-vaccinated herds; a negligible difference in the variance components was reported. Provided data on vaccination usage were available, further research could analyze vaccinated herds separately from non-vaccinated herds to quantify variance components for the birth of persistently infected calves. CONCLUSIONS The present study provides quantitative evidence that considerable genetic variability exists for the birth of persistently infected cattle following likely in utero exposure to BVD virus although no maternal genetic variance exists; genetic selection can contribute to a reduction in the incidence of persistent infection. Results from the present study can be used to inform international breeding programs of the gains achievable in reducing the incidence of persistent infection in cattle. In addition, the methods used in the present study can be applied to other diseases when data from future disease mitigation or eradication programs become available. For example, the data pruning procedures used in the present study attempted to maximize the possibility of equal lifetime exposure to the pathogen among herd-mates. Biologically plausible phenotype and exposure definitions were created to reflect both management practices and the lifecycle of the pathogen investigated; the same methods applied in the present study can be applied to other disease traits. LITERATURE CITED Anon . 2018a . Terms & Conditions of BVD Scheme for Dairy Breed Animals born in 2018 . – [accessed 28 Feb 2018 ]. https://www.agriculture.gov.ie/animalhealthwelfare/diseasecontrol/bovineviraldiarrhoeabvd/. Anon . 2018b . Terms and Conditions of BVD Scheme for Beef Breed Animals born in 2018 . – [accessed 28 Feb 2018 ]. https://www.agriculture.gov.ie/animalhealthwelfare/diseasecontrol/bovineviraldiarrhoeabvd/. Bachofen , C. , U. Braun , M. Hilbe , F. Ehrensperger , H. Stalder , and E. Peterhans . 2010 . Clinical appearance and pathology of cattle persistently infected with bovine viral diarrhea virus of different genetic subgroups . Vet. Microbiol . 141 : 258 – 267 . doi: https://doi.org/10.1016/j.vetmic.2009.09.022 Google Scholar Crossref Search ADS PubMed Barrett , D. J. , S. J. More , D. A. Graham , J. O’Flaherty , M. L. Doherty , and H. M. Gunn . 2011 . Considerations on BVD eradication for the Irish livestock industry . Ir. Vet. J . 64 : 12 . doi: https://doi.org/10.1186/2046-0481-64-12 Google Scholar Crossref Search ADS PubMed Berry , D. P . 2017 . Breeding a better cow—will she be adaptable ?. J. Dairy Sci . 101 : 3665 – 3685 . doi: https://doi.org/10.3168/jds.2017-13309 Google Scholar Crossref Search ADS PubMed Berry , D. P. , M. L. Bermingham , M. Good , and S. J. More . 2011 . Genetics of animal health and disease in cattle . Ir. Vet. J . 64 : 5 . doi: https://doi.org/10.1186/2046-0481-64-5 Google Scholar Crossref Search ADS PubMed Berry , D. P. , E. Wall , and J. E. Pryce . 2014 . Genetics and genomics of reproductive performance in dairy and beef cattle . Animal . 8 ( s1 ): 105 – 121 . doi: https://doi.org/10.1017/S1751731114000743 Google Scholar Crossref Search ADS PubMed Bishop , S. C. , and J. A. Woolliams . 2010 . On the genetic interpretation of disease data . PLoS One . 5 : e8940 . doi: https://doi.org/10.1371/journal.pone.0008940 Google Scholar Crossref Search ADS PubMed Bishop , S. C. , and J. A. Woolliams . 2014 . Genomics and disease resistance studies in livestock . Livest Sci . 166 : 190 – 198 . doi: https://doi.org/10.1016/j.livsci.2014.04.034 Google Scholar Crossref Search ADS PubMed Brownlie , J., M. C. Clarke , C. J. Howard , and D. H. Pocock . 1987 . Pathogenesis and epidemiology of bovine virus diarrhoea virus infection of cattle . Ann. Rech. Vet . 18 : 157 – 166 . Google Scholar PubMed Chucri , T. M. , J. Monteiro , A. Lima , M. Salvadori , J. K. Junior , and M. A. Miglino . 2010 . A review of immune transfer by the placenta . J. Reprod. Immunol . 87 : 14 – 20 . doi: https://doi.org/10.1016/j.jri.2010.08.062 Google Scholar Crossref Search ADS PubMed Clegg , T. A. , D. A. Graham , P. O’Sullivan , G. McGrath , and S. J. More . 2016 . Temporal trends in the retention of BVD+ calves and associated animal and herd-level risk factors during the compulsory eradication programme in Ireland . Prev. Vet. Med . 134 : 128 – 138 . doi:10.1016/j.prevetmed.2016.10.010 Google Scholar Crossref Search ADS PubMed Gianola , D. , and J. Foulley . 1983 . Sire evaluation for ordered categorical data with a threshold model . Genet. Sel. Evol . 15 : 201 – 224 . doi:10.1186/1297-9686-15-2-201 Google Scholar Crossref Search ADS PubMed Gilmour , A. R. , B. Gogel , B. Cullis , R. Thompson , and D. Butler . 2009 . ASReml user guide release 3.0 . Hemel Hempstead, UK: VSN Int. Ltd . Grooms , D. L . 2004 . Reproductive consequences of infection with bovine viral diarrhea virus . Vet. Clin. North Am. Food Anim. Pract . 20 : 5 – 19 . doi:10.1016/j.cvfa.2003.11.006 Google Scholar Crossref Search ADS PubMed Gunn , G. J. , H. W. Saatkamp , R. W. Humphry , and A. W. Stott . 2005 . Assessing economic and social pressure for the control of bovine viral diarrhea virus . Prev. Vet. Med . 72 : 149 – 162 . doi:10.1016/j.prevetmed.2005.08.012 Google Scholar Crossref Search ADS PubMed Hansen , T. R. , N. P. Smirnova , H. Van Campen , M. L. Shoemaker , A. A. Ptitsyn , and H. Bielefeldt‐Ohmann . 2010 . Maternal and fetal response to fetal persistent infection with bovine viral diarrhea virus . Am. J. Reprod. Immunol . 64 : 295 – 306 . doi:10.1111/j.1600-0897.2010.00904.x Google Scholar Crossref Search ADS PubMed Hansen , T. R. , N. P. Smirnova , B. T. Webb , H. Bielefeldt-Ohmann , R. E. Sacco , and H. Van Campe . 2015 . Innate and adaptive immune responses to in utero infection with bovine viral diarrhea virus . Anim. Health Res. Rev . 16 : 15 – 26 . doi:10.1017/S1466252315000122 Google Scholar Crossref Search ADS PubMed Havenstein , G. , P. Ferket , and M. Qureshi . 2003 . Carcass composition and yield of 1957 versus 2001 broilers when fed representative 1957 and 2001 broiler diets . Poult. Sci . 82 : 1509 – 1518 . doi:10.1093/ps/82.10.1509 Google Scholar Crossref Search ADS PubMed Houe , H . 1999 . Epidemiological features and economical importance of bovine virus diarrhea virus (BVDV) infections . Vet. Microbiol . 64 : 89 – 107 . doi:10.1016/S0378-1135(98)00262-4 Google Scholar Crossref Search ADS PubMed Houe , H . 2003 . Economic impact of BVDV infection in dairies . Biologicals . 31 : 137 – 143 . doi:10.1016/S1045-1056(03)00030-7 Google Scholar Crossref Search ADS PubMed Lanyon , S. R. , F. I. Hill , M. P. Reichel , and J. Brownlie . 2014 . Bovine viral diarrhea: pathogenesis and diagnosis . Vet. J . 199 : 201 – 209 . doi:10.1016/j.tvjl.2013.07.024 Google Scholar Crossref Search ADS PubMed Lindberg , A. , and H. Houe . 2005 . Characteristics in the epidemiology of bovine viral diarrhea virus (BVDV) of relevance to control . Prev. Vet. Med . 72 : 55 – 73 . doi:10.1016/j.prevetmed.2005.07.018 Google Scholar Crossref Search ADS PubMed McClurkin , A. W., E. T. Littledike , R. C. Cutlip , G. H. Frank , M. F. Coria , and S. R. Bolin . 1984 . Production of cattle immunotolerant to bovine viral diarrhea virus . Can. J. Comp. Med . 48 : 156 – 161 . Google Scholar PubMed Moennig , V. , H. Houe , and A. Lindberg . 2005 . BVD control in Europe: current status and perspectives . Anim. Health Res. Rev . 6 : 63 – 74 . doi:10.1079/AHR2005102 Google Scholar Crossref Search ADS PubMed More , S. J. , K. McKenzie , J. O’Flaherty , M. L. Doherty , A. R. Cromie , and M. J. Magan . 2010 . Setting priorities for non-regulatory animal health in Ireland: results from an expert Policy Delphi study and a farmer priority identification survey . Prev. Vet. Med . 95 : 198 – 207 . doi:10.1016/j.prevetmed.2010.04.011 Google Scholar Crossref Search ADS PubMed Muggli-Cockett , N. , L. Cundiff , and K. Gregory . 1992 . Genetic analysis of bovine respiratory disease in beef calves during the first year of life . J. Anim. Sci . 70 : 2013 – 2019 . doi:10.2527/1992.7072013x Google Scholar Crossref Search ADS PubMed Peterhans , E. , and M. Schweizer . 2010 . Pestiviruses: how to outmaneuver your hosts . Vet. Microbiol. 2010 . 142 : 18 – 25 . doi:10.1016/j.vetmic.2009.09.038 Purfield , D. C. , M. McClure , and D. P. Berry . 2016 . Justification for setting the individual animal genotype call rate threshold at eighty-five percent . J. Anim. Sci . 94 : 4558 – 4569 . doi:10.2527/jas.2016-0802 Google Scholar Crossref Search ADS PubMed Presi , P. , R. Struchen , T. Knight-Jones , S. Scholl , and D. Heim . 2011 . Bovine viral diarrhea (BVD) eradication in Switzerland—experiences of the first two years . Prev. Vet. Med . 99 : 112 – 121 . doi:10.1016/j.prevetmed.2011.01.012 Google Scholar Crossref Search ADS PubMed Reardon , F. , D. A. Graham , T. A. Clegg , J. A. Tratalos , P. O’Sullivan , and S. J. More . 2018 . Quantifying the role of Trojan dams in the between-herd spread of bovine viral diarrhea virus (BVDv) in Ireland . Prev. Vet. Med . 152 : 65 – 73 . doi:10.1016/j.prevetmed.2018.02.002 Google Scholar Crossref Search ADS PubMed Ring , S. C. , D. A. Graham , R. G. Sayers , N. Byrne , M. M. Kelleher , M. L. Doherty , and D. P. Berry . 2018 . Genetic variability in the humoral immune response to bovine herpesvirus-1 infection in dairy cattle and genetic correlations with performance traits . J. Dairy Sci . 101 : 6190 – 6204 . doi:10.3168/jds.2018–14481 Google Scholar Crossref Search ADS PubMed Robertson , A. , and I. M. Lerner . 1949 . The heritability of all-or-none traits: viability of poultry . Genetics 34 : 395 – 411 . Google Scholar PubMed Schaeffer , L . 2006 . Strategy for applying genome‐wide selection in dairy cattle . J. Anim. Breed. Genet . 123 : 218 – 223 . doi:10.1111/j.1439-0388.2006.00595.x Google Scholar Crossref Search ADS PubMed Schneider , M. J. , R G. Tait , Jr , M. V. Ruble , W. D. Busby , and J. M. Reecy . 2010 . Evaluation of fixed sources of variation and estimation of genetic parameters for incidence of bovine respiratory disease in preweaned calves and feedlot cattle . J. Anim. Sci . 88 : 1220 – 1228 . doi:10.2527/jas.2008-1755 Google Scholar Crossref Search ADS PubMed Schoder , G. , K. Möstl , V. Benetka , and W. Baungartner . 2004 . Different outcome of intrauterine infection with bovine viral diarrhea (BVD) virus in twin calves . Vet. Record . 154 : 52 – 53 . doi:10.1136/vr.154.2.52 Google Scholar Crossref Search ADS Smirnova , N. P. , B. T. Webb , H. Bielefeldt-Ohmann , H. Van Campen , A. Q. Antoniazzi , S. E. Morarie , and T. R. Hansen . 2012 . Development of fetal and placental innate immune responses during establishment of persistent infection with bovine viral diarrhea virus . Virus Res . 167 : 329 – 336 . doi:10.1016/j.virusres.2012.05.018 Google Scholar Crossref Search ADS PubMed Smirnova , N. P. , B. T. Webb , J. L. McGill , R. G. Schaut , H. Bielefeldt-Ohmann , H. Van Campen , R. E. Sacco , and T. R. Hansen . 2014 . Induction of interferon-gamma and downstream pathways during establishment of fetal persistent infection with bovine viral diarrhea virus . Virus Res . 183 : 95 – 106 . doi:10.1016/j.virusres.2014.02.002 Google Scholar Crossref Search ADS PubMed Snowder , G. D., L. D. Van Vleck , L. V. Cundiff , and G. L. Bennett . 2005 . Influence of breed, heterozygosity, and disease incidence on estimates of variance components of respiratory disease in preweaned beef calves . J. Anim. Sci . 83 : 1247 – 1261 . doi:10.2527/2005.8361247x Google Scholar Crossref Search ADS PubMed Snowder , G. , L. D. Van Vleck , L. V. Cundiff , and G. L. Bennett . 2006 . Bovine respiratory disease in feedlot cattle: environmental, genetic, and economic factors . J. Anim. Sci . 84 : 1999 – 2008 . doi:10.2527/jas.2006-046 Google Scholar Crossref Search ADS PubMed Ståhl , K. , and S. Alenius . 2012 . BVDV control and eradication in Europe—an update . Jpn. J. Vet. Res . 60 ( Suppl .): S31 – S39 . Google Scholar PubMed Stokstad , M. , and T. Løken . 2002 . Pestivirus in cattle: experimentally induced persistent infection in calves . J. Vet. Med., Series B . 49 : 494 – 501 . doi:10.1046/j.1439-0450.2002.00600.x Google Scholar Crossref Search ADS Stott , A. W. , R. W. Humphry , G. J. Gunn , I. Higgins , T. Hennessy , J. O’Flaherty , and D. A. Graham . 2012 . Predicted costs and benefits of eradicating BVDV from Ireland . Ir. Vet. J . 65 : 12 . doi:10.1186/2046-0481-65-12 Google Scholar Crossref Search ADS PubMed Tizard , I. R . 2013 . Veterinary immunology—E-book . China : Elsevier Health Sciences . Twomey , A. J., R. G. Sayers , R. I. Carroll , N. Byrne , E. O. Brien , M. L. Doherty , J. C. McClure , D. A. Graham , and D. P. Berry . 2016 . Genetic parameters for both a liver damage phenotype caused by and antibody response to phenotype in dairy and beef cattle . J. Anim. Sci . 94 : 4109 – 4119 . doi:10.2527/jas.2016-0621 Google Scholar Crossref Search ADS PubMed VanRaden , P. M. , and A. H. Sanders . 2003 . Economic merit of crossbred and purebred US dairy cattle . J. Dairy Sci . 86 : 1036 – 1044 . doi:10.3168/jds. S0022-0302(03)73687-X Google Scholar Crossref Search ADS PubMed Van Vleck , L. D . 1970 . Misidentification in estimating the paternal sib correlation . J. Dairy Sci . 53 : 1469 – 1474 . doi:10.3168/jds.S0022-0302(70)86416-5 Google Scholar Crossref Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the American Society of Animal Science. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Bennett, Gary L; Tait, Richard G; Shackelford, Steven D; Wheeler, Tommy L; King, David A; Casas, Eduardo; Smith, Timothy P L

doi: 10.1093/jas/sky451pmid: 30476168

Abstract The objective of this study was to enhance estimates of additive, dominance, and epistatic effects of marker polymorphisms on beef carcass and quality traits. Myostatin (MSTN) F94L SNP and the µ-calpain (CAPN1) 316 and 4751 SNP haplotype have previously been associated with fat and muscle traits in beef cattle. Multiyear selection in a composite population segregating these polymorphisms increased minor allele (F94L L) and chosen haplotype (CAPN1 CC and GT) frequencies to intermediate levels resulting in more precise estimates of additive and nonadditive genetic effects. During the 3 yr after selection, 176 steers were evaluated for growth, carcass, meat quality, tenderness (n = 103), and meat color traits. The statistical model included year, age of dam, age of the steer, and genotype in a random animal model. The 9 genotypes (3 CAPN1 diplotypes × 3 F94L genotypes) affected marbling score, ribeye area, adjusted fat thickness, vision yield grade (all P < 0.001), slice shear force (P = 0.03), and CIE L* reflectance (P = 0.01). Linear contrasts of the 9 genotypes estimated additive, recessive, and epistatic genetic effects. Significant additive effects of the F94L L allele decreased marbling score, adjusted fat thickness, vision yield grade, and slice shear force; and increased ribeye area and CIE L* reflectance. The homozygous F94L FF and LL genotypes differed by 1.3 to 1.9 phenotypic SD for most carcass traits and by 0.8 to 0.9 SD for slice shear force and CIE L* reflectance but carcass weight differed by only 3 kg (0.1 SD). The L allele was partially recessive to F for ribeye area (P = 0.02) and the heterozygous FL means tended to be closer to the FF genotype than the LL genotype for other carcass traits but differences from additive were not significant. The CAPN1 additive × F94L additive effect on slice shear force was the only significant epistatic estimate. The F94L L allele is prevalent in Limousin but nearly absent in other U.S. purebreds. This allele had about half of the effects on birth weight, muscle, and fat traits reported for severe MSTN mutations in Belgian Blue and Piedmontese breeds. The interaction between MSTN and CAPN1 genotypes may reflect the strong additive effects of MSTN F94L L allele on fat and muscle traits interfering with the phenotypic effect of CAPN1 genotype on meat tenderness. INTRODUCTION Some polymorphisms in the myostatin (MSTN) gene in cattle reduce functionally effective myostatin resulting in muscle hypertrophy. Before breeders knew what caused muscle hypertrophy and before DNA tests were developed, some breeders selected their cattle for increased muscling (Arthur, 1995). Selection greatly increased frequencies of some MSTN alleles in breeds such as Piedmontese and Belgian Blue (Grobet et al., 1997; Kambadur et al., 1997). Surveys of some European cattle breeds for DNA variants in MSTN revealed many polymorphisms (Grobet et al., 1998; Dunner et al., 2003). Among those was F94L, a phenylalanine to leucine substitution at amino acid position 94 in myostatin (Grobet et al., 1998). The leucine allele frequency was high in Limousin and low or absent in many other breeds. Subsequently, this allele was associated with increased muscling and reduced fatness in Limousin crossbreds (Esmailizadeh et al., 2008; Alexander et al., 2009). However, mating designs limited statistical power for estimating dominance effects (Esmailizadeh et al., 2008). Polymorphisms in µ-calpain (CAPN1) are associated with postmortem proteolysis of muscle leading to increased meat tenderness (Page et al., 2004; White et al., 2005; Casas et al., 2006; Robinson et al., 2012). Also, some MSTN polymorphisms suppressing production of functional myostatin increase meat tenderness (Wheeler et al., 2001). Casas et al. (2001) identified an interaction of a QTL affecting meat tenderness on BTA 4 with heterozygous MSTN and homozygous normal progeny of a Belgian Blue crossbred bull. Epistasis between MSTN and CAPN1 is possible because both genes are thought to affect protein turnover of muscle. The objective of this study was to estimate additive and nonadditive effects associated with the F94L polymorphism in MSTN and SNP 316 and 4751 haplotypes in CAPN1. Estimates were enhanced by increasing minor allele frequency of F94L and increasing frequencies of chosen CAPN1 haplotypes through selection and by using bulls heterozygous for F94L and for CAPN1. MATERIALS AND METHODS The U.S. Meat Animal Research Center (USMARC) Institutional Animal Care and Use Committee approved the experiment following recommendations by FASS (1999). Composite Population A composite cattle population known as MARC I was formed beginning in 1978 and consisted of 0.125 Angus, 0.125 Hereford, 0.25 Braunvieh, 0.25 Charolais, and 0.25 Limousin (Gregory et al., 1991). From 1992 through 1999, the composite was divided into 2 lines: a calving ease selection line and a control line (Bennett, 2008). After completing the selection experiment, cows from both lines were bred to the same bulls and their progeny treated as a single population. During this period from 2000 through 2006, the MARC I population was continued with about 205 calves from 18 sires born each year. Approximately half of sires were replaced each year resulting in the use of 68 bulls selected from within the herd. Genetic Markers Selected markers were from 2 genes thought to affect muscles, especially protein turnover and proteolysis. They were the large subunit of micromolar activated calpain (CAPN1) and myostatin (MSTN). The SNP marker chosen for MSTN on BTA 2 was a phenylalanine to leucine substitution at amino acid position 94 in myostatin (F94L; rs110065568, Grobet et al., 1998). Two SNP markers, CAPN1_316 (BTA29; rs17872000) and CAPN1_4751 (BTA29; rs17872050), were used for the CAPN1 gene located on BTA 29. CAPN1_316 segregates C and G alleles, while CAPN1_4751 segregates C and T alleles. Initial findings associated the C allele of CAPN1_316 with tender beef (Page et al., 2002). Common haplotypes determined from CAPN1_316 and CAPN1_4751 SNP are CC (CAPN1hCC), GT (CAPN1hGT), and GC (CAPN1hGC). A fourth haplotype, CT (CAPN1hCT), is rare. White et al. (2005) found the largest difference for 14 d Warner-Bratzler shear force between CAPN1hCC and CAPN1hGT. The most frequent haplotype in that study (CAPN1hGC) was intermediate in tenderness to CAPN1hCC and CAPN1hGT. Samples of DNA were extracted from blood or semen. Extraction of DNA was done using a Qiagen QIAmp DNA mini blood kit (Qiagen, Valencia, CA). Blood samples were collected in 10 mL syringes with 4% EDTA. Blood was frozen until DNA was extracted. Genotyping was performed using a primer extension method with mass spectrometry-based analysis of the extension products on a MassArray system as suggested by the manufacturer (Sequenom, Inc., San Diego, CA) and described by Stone et al. (2002). When necessary, genotype assays were repeated to reduce missing genotypes. Base, Selection, and Evaluation Phases The experiment was conducted in 3 phases: base, selection, and evaluation. The base phase surveyed live animals and some frozen semen from 4 populations (Angus and 3 composites; MARC I, MARC II, and MARC III) completing the calving ease selection experiment (Bennett, 2008), for CAPN1 allele and haplotype frequencies. Frequencies of CAPN1hCC, CAPN1hGT, and CAPN1hGC were 0.47, 0.28, and 0.25 for Angus; 0.20, 0.21, and 0.58 for MARC I; 0.20, 0.39, and 0.40 for MARC II; and 0.22, 0.46, and 0.32 for MARC III. Allele L (F94LaL) is more frequent than allele F (F94LaF) in the Limousin breed but absent or near zero in most other breeds in the United States (Dunner et al., 2003). Limousin makes up 0.25 of MARC I and it was assumed to be the only population segregating F94L. The estimated frequency of F94LaL in MARC I was 20.7% for 363 animals born before 2004. The Limousin specificity was verified later by genotyping Angus (n = 564), MARC II (Angus, Hereford, Gelbvieh, and Simmental; n = 538), and MARC III (Angus, Hereford, Red Poll, and Pinzgauer; n = 747) populations born in 2009, 2010, and 2011 for F94L. All were homozygous for the phenylalanine allele (F94LaF) except for 1 Angus, 1 MARC II, and 2 unrelated MARC III heterozygotes, a frequency less than assumed genotyping error rate. MARC I was chosen for this experiment, because F94LaL was present in the population. Selection was applied from 2004 through 2006 with the goal of increasing frequencies of F94LaL, CAPN1hCC, and CAPN1hGT to 0.5 and eliminating CAPN1hGC. Calves were bled before weaning, the DNA was extracted, and then genotyped. Marker genotypes were used to select replacement bulls and heifers soon after weaning. Selection of replacement animals in this phase was based on the presence of F94LaL, CAPN1hCC, and CAPN1hGT and absence of CAPN1hGC. The evaluation phase (birth years 2007, 2008, and 2009) increased the number of animals evaluated for carcass traits. This phase also used bulls heterozygous for both F94L and CAPN1 if available. These bulls increase the number of within sire comparisons across any combination of progeny genotypes. A heterozygous bull can sire progeny with heterozygous or either homozygous genotype. Thirty bulls sired calves in the evaluation phase. Bulls heterozygous for CAPN1hCC/CAPN1hGT (n = 15) sired 70% of calves. Bulls heterozygous for F94L (n = 23) sired 87% of calves. Bulls heterozygous for both CAPN1 and F94L (n = 10) sired 59% of calves. Blood samples from spring-born progeny were collected before weaning and genotyped. Calves with incomplete genotypes were removed from the experiment. Replacement bulls were randomly sampled within sire from among males heterozygous for both F94L and the chosen CAPN1 haplotypes. The remaining males were castrated by banding soon after weaning and genotyping. Steers consumed corn and corn silage-based diets until harvest. Weights were taken at birth (mean date = April 13), at weaning (mean age = 159 d, SD = 18 d), and as yearlings (mean age = 344 d, SD = 22 d). All experimental steers were harvested on a single day each year at a commercial abattoir at an average age of 487 d. In 2010 (born 2009), 29 steers were removed from the experiment based on either having unselected haplotypes (CAPN1hGC or CAPN1hCT) or having genotypes in common with many steers (reduced at random). Carcasses were weighed hot, electrically-stimulated, and chilled using the commercial facility’s proprietary system. At 36 h postmortem, carcasses were ribbed between the 12th and 13th ribs and an image analysis based (VBG2000) grading system (Shackelford et al., 2003) assessed adjusted fat thickness, ribeye area, USDA marbling score, CIE L* of the longissimus muscle, and calculated vision yield grade. A longissimus steak from the 13th rib region was returned to USMARC to evaluate slice shear force at 14 d postmortem (Shackelford et al., 1999). Statistical Analysis Either trait measurements or logarithms of measurements (marbling score; slice shear force) were analyzed with a mixed model using MTDFREML (Boldman et al., 1995). The model was: Yi,j,k,l=μ + Yeari + Aod5j +Genotypek + b × Agei,j,k,l+ ai,j,k,l+ ei,j,k,l where Yi,j,k,l is the observation or its logarithm for the i, j, k, l-th animal, µ is the mean, Yeari is birth year 2007, 2008, or 2009, Aod5j is age of dam (2, 3, 4, or ≥ 5 yr), b is a linear regression coefficient on the i, j, k, l-th animal’s age (Agei,j,k,l) in days, Genotypek is 1 of the 9 combinations of 3 CAPN1 diplotypes and 3 F94L genotypes, ai,j,k,l is the additive polygenic animal effect, and ei,j,k,l is of the residual effect of the i, j, k, l-th observation. Covariances of polygenic effects were assumed proportional to the pedigree relationship matrix. Residual effects were assumed independent with constant variance. The pedigree used to calculate relationships included more than 6,600 animals. Heritabilities were constrained between 0.20 and 0.70 because of few observations and imprecise genetic variance estimates. Similar ranges in heritabilities for these traits were estimated from larger, related populations (Gregory et al., 1994; Bennett and Gregory, 1996). Skewed distributions of marbling scores and meat tenderness values were transformed using base 10 logarithms to determine P values but reported means and contrasts are from analyses of untransformed data. A P < 0.10 for the genotype effect (9 genotypes) was used as a guideline for then calculating linear contrasts for additive, dominance, and epistasis effects associated with CAPN1 haplotypes and F94L alleles similar to Tait et al. (2016). Linear contrast coefficients used to estimate genetic effects (Table 1) which were divided by their SE to determine significance based on a t-test. Only testing genetic effect contrasts after meeting an overall genotype probability test (e.g., P < 0.10 guideline) protects against probability inflation due to multiple testing. A significance level of P < 0.05 was used for individual contrasts. Table 1. Linear contrast coefficients used to estimate additive, dominance, and epistasis effects for µ-calpain (CAPN1) haplotype and F94L SNP Genotype mean1 . F94L2 . CAPN12 . F94L × CAPN13 . F94L . CAPN1 . A . D . A . D . AA . AD . DA . DD . FF NN-NN −1 −1 −1 −1 1 1 1 1 FF CC-NN −1 −1 0 2 0 −2 0 −2 FF CC-CC −1 −1 1 −1 −1 1 −1 1 FL NN-NN 0 2 −1 −1 0 0 −2 −2 FL CC-NN 0 2 0 2 0 0 0 4 FL CC-CC 0 2 1 −1 0 0 2 −2 LL NN-NN 1 −1 −1 −1 −1 −1 1 1 LL CC-NN 1 −1 0 2 0 2 0 −2 LL CC-CC 1 −1 1 −1 1 −1 −1 1 Divisor4 6 6 6 6 1 2 2 4 Genotype mean1 . F94L2 . CAPN12 . F94L × CAPN13 . F94L . CAPN1 . A . D . A . D . AA . AD . DA . DD . FF NN-NN −1 −1 −1 −1 1 1 1 1 FF CC-NN −1 −1 0 2 0 −2 0 −2 FF CC-CC −1 −1 1 −1 −1 1 −1 1 FL NN-NN 0 2 −1 −1 0 0 −2 −2 FL CC-NN 0 2 0 2 0 0 0 4 FL CC-CC 0 2 1 −1 0 0 2 −2 LL NN-NN 1 −1 −1 −1 −1 −1 1 1 LL CC-NN 1 −1 0 2 0 2 0 −2 LL CC-CC 1 −1 1 −1 1 −1 −1 1 Divisor4 6 6 6 6 1 2 2 4 1Estimated genotype means are identified by the 9 combinations of 3 MSTN F94L genotypes and 3 µ-calpain (CAPN1) diplotypes. F94L genotypes are designated by FF (homozygous F94LaF), FL (heterozygotes), and LL (homozygous F94LaL). CAPN1 diplotypes are designated NN-NN (homozygous CAPN1hNN), CC-NN (heterozygotes) and CC-CC (homozygous CAPN1hCC). 2Linear contrast coefficients multiplied by genotype means to estimate F94LaL and CAPNhCC haplotype additive (A) and dominance (D) effects. 3Linear contrast coefficients for 2-factor epistatic effects identified with 2 letters. The first letter is the F94LaL effect and the second letter is the CAPN1hCC haplotype effect, e.g., AD is additive F94L × CAPN1 dominance epistatic effect. 4Actual coefficients used were the whole numbers in table divided by this number. Open in new tab Table 1. Linear contrast coefficients used to estimate additive, dominance, and epistasis effects for µ-calpain (CAPN1) haplotype and F94L SNP Genotype mean1 . F94L2 . CAPN12 . F94L × CAPN13 . F94L . CAPN1 . A . D . A . D . AA . AD . DA . DD . FF NN-NN −1 −1 −1 −1 1 1 1 1 FF CC-NN −1 −1 0 2 0 −2 0 −2 FF CC-CC −1 −1 1 −1 −1 1 −1 1 FL NN-NN 0 2 −1 −1 0 0 −2 −2 FL CC-NN 0 2 0 2 0 0 0 4 FL CC-CC 0 2 1 −1 0 0 2 −2 LL NN-NN 1 −1 −1 −1 −1 −1 1 1 LL CC-NN 1 −1 0 2 0 2 0 −2 LL CC-CC 1 −1 1 −1 1 −1 −1 1 Divisor4 6 6 6 6 1 2 2 4 Genotype mean1 . F94L2 . CAPN12 . F94L × CAPN13 . F94L . CAPN1 . A . D . A . D . AA . AD . DA . DD . FF NN-NN −1 −1 −1 −1 1 1 1 1 FF CC-NN −1 −1 0 2 0 −2 0 −2 FF CC-CC −1 −1 1 −1 −1 1 −1 1 FL NN-NN 0 2 −1 −1 0 0 −2 −2 FL CC-NN 0 2 0 2 0 0 0 4 FL CC-CC 0 2 1 −1 0 0 2 −2 LL NN-NN 1 −1 −1 −1 −1 −1 1 1 LL CC-NN 1 −1 0 2 0 2 0 −2 LL CC-CC 1 −1 1 −1 1 −1 −1 1 Divisor4 6 6 6 6 1 2 2 4 1Estimated genotype means are identified by the 9 combinations of 3 MSTN F94L genotypes and 3 µ-calpain (CAPN1) diplotypes. F94L genotypes are designated by FF (homozygous F94LaF), FL (heterozygotes), and LL (homozygous F94LaL). CAPN1 diplotypes are designated NN-NN (homozygous CAPN1hNN), CC-NN (heterozygotes) and CC-CC (homozygous CAPN1hCC). 2Linear contrast coefficients multiplied by genotype means to estimate F94LaL and CAPNhCC haplotype additive (A) and dominance (D) effects. 3Linear contrast coefficients for 2-factor epistatic effects identified with 2 letters. The first letter is the F94LaL effect and the second letter is the CAPN1hCC haplotype effect, e.g., AD is additive F94L × CAPN1 dominance epistatic effect. 4Actual coefficients used were the whole numbers in table divided by this number. Open in new tab RESULTS AND DISCUSSION Frequencies of chosen CAPN1 haplotypes and F94LaL changed during the 3 phases of the experiment (Fig. 1). The frequency of F94LaL and CAPN1hCC approached 0.5 during the evaluation phase. The combined frequencies of the 2 selected CAPN1 haplotypes were about 0.4 during the base phase and increased to 0.75 during the evaluation phase (Fig. 1). About 40% of animals still had at least one CAPN1hGC haplotype during the evaluation phase. Because CAPN1hCC was more frequent than CAPN1hGT, analyses of all traits except slice shear force used CAPN1hCC and all other haplotypes (CAPN1hNN) instead of CAPN1hGT specifically. Only 103 animals with unambiguous diplotypes consisting of only CAPN1hGT and (or) CAPN1hCC haplotypes were used to analyze slice shear force. This method of addressing the CAPN1hGC haplotypes was chosen for slice shear force because a direct effect of CAPN1 on this trait was expected. This method should have no bias and a straight forward interpretation of CAPN1hCC and CAPN1hGT associations but does result in fewer animals and less power. The CAPN1hNN method was used for all other traits because CAPN1 effects, if any, are expected to be less direct. The CAPN1hNN method maximized the number of steers that could be used to estimate F94L effects for traits other than slice shear force while still accounting for the possibility of CAPN1 influence. Figure 1. Open in new tabDownload slide Frequencies for MSTN F94L allele L (▲) and µ-calpain 316–4751 haplotypes CC (■) and GT (♦) by birth year. Base, selection, and evaluation phases of the experiment are identified by vertical dashed lines. Figure 1. Open in new tabDownload slide Frequencies for MSTN F94L allele L (▲) and µ-calpain 316–4751 haplotypes CC (■) and GT (♦) by birth year. Base, selection, and evaluation phases of the experiment are identified by vertical dashed lines. Table 2 shows the combinations of CAPN1 diplotypes and F94L genotypes for the 176 harvested steers and the 103 used in slice shear force analyses. Table 3 characterizes the opportunity for increasing power through within sire comparisons across genotypes. Median values for 28 sires used were 4.5 progeny distributed among 3 of the 9 possible genotypes (Table 3). Four sires had progeny in 7, 8, or 9 genotype classes making especially strong contributions to increasing power. For example, the 2 steers in the smallest CAPN1 × MSTN combination for slice shear force had 18 half-sibs among 7 or the other 8 CAPN1 × MSTN combinations. Table 2. Number of harvested MARC I steers by genotype CAPN1 diplotype1 . F94L genotype2 . Total . . FF . FL . LL . . CAPN1hNN, CAPN1hNN 26 26 9 61 CAPN1hGT, CAPN1hGT 9 12 2 23 CAPN1hCC, CAPN1hNN 24 40 16 80 CAPN1hCC, CAPN1hGT 17 18 10 45 CAPN1hCC, CAPN1hCC 8 17 10 35 Total 58 83 35 176 CAPN1 diplotype1 . F94L genotype2 . Total . . FF . FL . LL . . CAPN1hNN, CAPN1hNN 26 26 9 61 CAPN1hGT, CAPN1hGT 9 12 2 23 CAPN1hCC, CAPN1hNN 24 40 16 80 CAPN1hCC, CAPN1hGT 17 18 10 45 CAPN1hCC, CAPN1hCC 8 17 10 35 Total 58 83 35 176 1Diplotypes composed of µ-Calpain (CAPN1) haplotypes. Any haplotype other than CAPN1hCC is represented by CAPN1hNN and these animals were used for analyses of most traits. Only animals with diplotypes consisting of CAPN1hGT and CAPN1hCC were used to analyze slice shear force (n = 103) and are a subset of the 176 steers and indicated by italics. 2MSTN F94L homozygous F94LaF (FF), heterozygous (FL), and homozygous F94LaL (LL) genotypes. Open in new tab Table 2. Number of harvested MARC I steers by genotype CAPN1 diplotype1 . F94L genotype2 . Total . . FF . FL . LL . . CAPN1hNN, CAPN1hNN 26 26 9 61 CAPN1hGT, CAPN1hGT 9 12 2 23 CAPN1hCC, CAPN1hNN 24 40 16 80 CAPN1hCC, CAPN1hGT 17 18 10 45 CAPN1hCC, CAPN1hCC 8 17 10 35 Total 58 83 35 176 CAPN1 diplotype1 . F94L genotype2 . Total . . FF . FL . LL . . CAPN1hNN, CAPN1hNN 26 26 9 61 CAPN1hGT, CAPN1hGT 9 12 2 23 CAPN1hCC, CAPN1hNN 24 40 16 80 CAPN1hCC, CAPN1hGT 17 18 10 45 CAPN1hCC, CAPN1hCC 8 17 10 35 Total 58 83 35 176 1Diplotypes composed of µ-Calpain (CAPN1) haplotypes. Any haplotype other than CAPN1hCC is represented by CAPN1hNN and these animals were used for analyses of most traits. Only animals with diplotypes consisting of CAPN1hGT and CAPN1hCC were used to analyze slice shear force (n = 103) and are a subset of the 176 steers and indicated by italics. 2MSTN F94L homozygous F94LaF (FF), heterozygous (FL), and homozygous F94LaL (LL) genotypes. Open in new tab Table 3. Characterization of potential for within sire comparisons among genotypes Progeny distribution measures for 28 sires . Value . Median number of progeny per sire 4.5 Sires with 3 progeny F94L genotypes1 12 Average F94L progeny genotypes per sire 2.04 Sires with 3 progeny CAPN1 diplotypes2 7 Average CAPN1 progeny diplotypes per sire 2.11 Sires with 7 to 9 CAPN1 × F94L progeny genotypes 4 Sires with 1 to 3 CAPN1 × F94L progeny genotypes 16 Median CAPN1 × MSTN progeny genotypes per sire 3 Progeny distribution measures for 28 sires . Value . Median number of progeny per sire 4.5 Sires with 3 progeny F94L genotypes1 12 Average F94L progeny genotypes per sire 2.04 Sires with 3 progeny CAPN1 diplotypes2 7 Average CAPN1 progeny diplotypes per sire 2.11 Sires with 7 to 9 CAPN1 × F94L progeny genotypes 4 Sires with 1 to 3 CAPN1 × F94L progeny genotypes 16 Median CAPN1 × MSTN progeny genotypes per sire 3 1Myostatin (MSTN) F94L genotypes were homozygous F94LaF, heterozygous F94L, and homozygous F94LaL. 2µ-Calpain (CAPN1) 316–4751 diplotypes were homozygous CAPN1hCC, the heterozygotes, and homozygous CAPN1hNN. Open in new tab Table 3. Characterization of potential for within sire comparisons among genotypes Progeny distribution measures for 28 sires . Value . Median number of progeny per sire 4.5 Sires with 3 progeny F94L genotypes1 12 Average F94L progeny genotypes per sire 2.04 Sires with 3 progeny CAPN1 diplotypes2 7 Average CAPN1 progeny diplotypes per sire 2.11 Sires with 7 to 9 CAPN1 × F94L progeny genotypes 4 Sires with 1 to 3 CAPN1 × F94L progeny genotypes 16 Median CAPN1 × MSTN progeny genotypes per sire 3 Progeny distribution measures for 28 sires . Value . Median number of progeny per sire 4.5 Sires with 3 progeny F94L genotypes1 12 Average F94L progeny genotypes per sire 2.04 Sires with 3 progeny CAPN1 diplotypes2 7 Average CAPN1 progeny diplotypes per sire 2.11 Sires with 7 to 9 CAPN1 × F94L progeny genotypes 4 Sires with 1 to 3 CAPN1 × F94L progeny genotypes 16 Median CAPN1 × MSTN progeny genotypes per sire 3 1Myostatin (MSTN) F94L genotypes were homozygous F94LaF, heterozygous F94L, and homozygous F94LaL. 2µ-Calpain (CAPN1) 316–4751 diplotypes were homozygous CAPN1hCC, the heterozygotes, and homozygous CAPN1hNN. Open in new tab Averages for steer traits, their estimated heritabilities, SD, and phenotypic SD are shown in Tables 4 and 5. The P-values for genotypes (8 df) were not significant for any weights from birth through harvest (Table 5). Genotype effects were significant (P < 0.05) for all carcass and meat traits except hot carcass weight. Estimated genotype and diplotype means for all traits are shown in Table 6. Table 4. Averages and SD of unadjusted measurements on 176 harvested steers Trait . Average . SD . Birth weight, kg 41.0 5.5 Weaning weight, kg 202 25 Yearling weight, kg 409 44 Final weight, kg 604 55 Hot carcass weight, kg 376 37 Marbling score1 329 38 Ribeye area, cm2 94.0 9.2 Adjusted fat thickness, mm 9.3 3.6 Vision yield grade2 2.26 0.73 Slice shear force3, kg 14.8 4.0 CIE L*4 35.6 2.0 Trait . Average . SD . Birth weight, kg 41.0 5.5 Weaning weight, kg 202 25 Yearling weight, kg 409 44 Final weight, kg 604 55 Hot carcass weight, kg 376 37 Marbling score1 329 38 Ribeye area, cm2 94.0 9.2 Adjusted fat thickness, mm 9.3 3.6 Vision yield grade2 2.26 0.73 Slice shear force3, kg 14.8 4.0 CIE L*4 35.6 2.0 1300 = Slight00; 400 = Small00 (USDA, 1997). 2Prediction of USDA Yield Grade. Smaller numbers indicate greater yield of boneless, closely trimmed retail cuts. 3Values for 103 steers having only diplotypes consisting of CAPN1hCC and (or) CAPN1hGT and analyzed for slice shear force. 4CIE L* measure of lightness. Greater values indicate lighter lean color. Open in new tab Table 4. Averages and SD of unadjusted measurements on 176 harvested steers Trait . Average . SD . Birth weight, kg 41.0 5.5 Weaning weight, kg 202 25 Yearling weight, kg 409 44 Final weight, kg 604 55 Hot carcass weight, kg 376 37 Marbling score1 329 38 Ribeye area, cm2 94.0 9.2 Adjusted fat thickness, mm 9.3 3.6 Vision yield grade2 2.26 0.73 Slice shear force3, kg 14.8 4.0 CIE L*4 35.6 2.0 Trait . Average . SD . Birth weight, kg 41.0 5.5 Weaning weight, kg 202 25 Yearling weight, kg 409 44 Final weight, kg 604 55 Hot carcass weight, kg 376 37 Marbling score1 329 38 Ribeye area, cm2 94.0 9.2 Adjusted fat thickness, mm 9.3 3.6 Vision yield grade2 2.26 0.73 Slice shear force3, kg 14.8 4.0 CIE L*4 35.6 2.0 1300 = Slight00; 400 = Small00 (USDA, 1997). 2Prediction of USDA Yield Grade. Smaller numbers indicate greater yield of boneless, closely trimmed retail cuts. 3Values for 103 steers having only diplotypes consisting of CAPN1hCC and (or) CAPN1hGT and analyzed for slice shear force. 4CIE L* measure of lightness. Greater values indicate lighter lean color. Open in new tab Table 5. Heritability and phenotypic SD estimates and P-values for sources of variation Trait . Year . Dam age . Calf age1 . Genotype . h2 . σP . Birth weight, kg 0.10 0.001 0.09 0.38 0.68 5.3 Weaning weight, kg 0.91 <0.001 <0.001 0.73 0.30 19.0 Yearling weight, kg <0.001 <0.001 <0.001 0.99 0.31 24.4 Final weight, kg 0.005 0.001 <0.001 0.85 0.50 48.5 Hot carcass weight, kg <0.001 0.001 <0.001 0.92 0.42 31.4 Marbling score2,3 0.13 0.60 0.53 <0.001 0.47 34 Ribeye area, cm2 0.004 0.04 0.16 <0.001 0.39 7.1 Adjusted fat thickness, mm 0.15 0.61 0.23 <0.001 0.51 3.2 Vision yield grade4 0.22 0.35 0.10 <0.001 0.52 0.62 Slice shear force3, kg 0.08 0.94 0.34 0.03 0.205 4.0 CIE L* reflectance6 0.03 0.86 0.05 0.01 0.44 1.9 Trait . Year . Dam age . Calf age1 . Genotype . h2 . σP . Birth weight, kg 0.10 0.001 0.09 0.38 0.68 5.3 Weaning weight, kg 0.91 <0.001 <0.001 0.73 0.30 19.0 Yearling weight, kg <0.001 <0.001 <0.001 0.99 0.31 24.4 Final weight, kg 0.005 0.001 <0.001 0.85 0.50 48.5 Hot carcass weight, kg <0.001 0.001 <0.001 0.92 0.42 31.4 Marbling score2,3 0.13 0.60 0.53 <0.001 0.47 34 Ribeye area, cm2 0.004 0.04 0.16 <0.001 0.39 7.1 Adjusted fat thickness, mm 0.15 0.61 0.23 <0.001 0.51 3.2 Vision yield grade4 0.22 0.35 0.10 <0.001 0.52 0.62 Slice shear force3, kg 0.08 0.94 0.34 0.03 0.205 4.0 CIE L* reflectance6 0.03 0.86 0.05 0.01 0.44 1.9 1Julian birthday linear covariate was used for all traits and is a proxy for age for all traits (except birth weight) because a single harvest date was used each year. 2300 = Slight00; 400 = Small00 (USDA, 1997). 3Logarithm (base 10) values of traits were analyzed. 4Prediction of USDA Yield Grade. Smaller numbers indicate greater yield of boneless, closely trimmed retail cuts. 5Constrained to 0.20 ≤ h2 ≤ 0.70. 6CIE L* measure of lightness. Greater values mean lighter lean color. Open in new tab Table 5. Heritability and phenotypic SD estimates and P-values for sources of variation Trait . Year . Dam age . Calf age1 . Genotype . h2 . σP . Birth weight, kg 0.10 0.001 0.09 0.38 0.68 5.3 Weaning weight, kg 0.91 <0.001 <0.001 0.73 0.30 19.0 Yearling weight, kg <0.001 <0.001 <0.001 0.99 0.31 24.4 Final weight, kg 0.005 0.001 <0.001 0.85 0.50 48.5 Hot carcass weight, kg <0.001 0.001 <0.001 0.92 0.42 31.4 Marbling score2,3 0.13 0.60 0.53 <0.001 0.47 34 Ribeye area, cm2 0.004 0.04 0.16 <0.001 0.39 7.1 Adjusted fat thickness, mm 0.15 0.61 0.23 <0.001 0.51 3.2 Vision yield grade4 0.22 0.35 0.10 <0.001 0.52 0.62 Slice shear force3, kg 0.08 0.94 0.34 0.03 0.205 4.0 CIE L* reflectance6 0.03 0.86 0.05 0.01 0.44 1.9 Trait . Year . Dam age . Calf age1 . Genotype . h2 . σP . Birth weight, kg 0.10 0.001 0.09 0.38 0.68 5.3 Weaning weight, kg 0.91 <0.001 <0.001 0.73 0.30 19.0 Yearling weight, kg <0.001 <0.001 <0.001 0.99 0.31 24.4 Final weight, kg 0.005 0.001 <0.001 0.85 0.50 48.5 Hot carcass weight, kg <0.001 0.001 <0.001 0.92 0.42 31.4 Marbling score2,3 0.13 0.60 0.53 <0.001 0.47 34 Ribeye area, cm2 0.004 0.04 0.16 <0.001 0.39 7.1 Adjusted fat thickness, mm 0.15 0.61 0.23 <0.001 0.51 3.2 Vision yield grade4 0.22 0.35 0.10 <0.001 0.52 0.62 Slice shear force3, kg 0.08 0.94 0.34 0.03 0.205 4.0 CIE L* reflectance6 0.03 0.86 0.05 0.01 0.44 1.9 1Julian birthday linear covariate was used for all traits and is a proxy for age for all traits (except birth weight) because a single harvest date was used each year. 2300 = Slight00; 400 = Small00 (USDA, 1997). 3Logarithm (base 10) values of traits were analyzed. 4Prediction of USDA Yield Grade. Smaller numbers indicate greater yield of boneless, closely trimmed retail cuts. 5Constrained to 0.20 ≤ h2 ≤ 0.70. 6CIE L* measure of lightness. Greater values mean lighter lean color. Open in new tab Table 6. Means of traits by myostatin F94L genotypes and µ-calpain (CAPN1) diplotypes Trait . F94L1 . CAPN12 . SED3 . . FF . FL . LL . NN-NN . NN-CC . CC-CC . SEDhom . SEDhet . Birth weight, kg 40.7 41.4 43.3 41.7 41.3 42.4 1.2 1.0 Weaning weight, kg 202 199 206 204 202 201 4.6 3.9 Yearling weight, kg 409 410 412 408 409 413 8.5 7.1 Final weight, kg 608 603 598 599 604 607 11.6 9.7 Hot carcass weight, kg 375 375 378 374 377 377 7.6 6.3 Marbling score4 345 329 294 325 327 316 8.2 6.8 Ribeye area, cm2 89 93 103 95 95 95 1.7 1.4 Adjusted fat thickness, mm 10.5 9.1 6.4 8.9 8.9 8.3 0.8 0.6 Vision Yield Grade5 2.59 2.26 1.57 2.18 2.18 2.05 0.15 0.12 Slice shear force, kg 16.2 14.5 12.8 15.1 13.7 14.8 0.96 0.81 CIE L* reflectance6 34.65 35.58 36.20 35.40 35.58 35.44 0.45 0.38 Trait . F94L1 . CAPN12 . SED3 . . FF . FL . LL . NN-NN . NN-CC . CC-CC . SEDhom . SEDhet . Birth weight, kg 40.7 41.4 43.3 41.7 41.3 42.4 1.2 1.0 Weaning weight, kg 202 199 206 204 202 201 4.6 3.9 Yearling weight, kg 409 410 412 408 409 413 8.5 7.1 Final weight, kg 608 603 598 599 604 607 11.6 9.7 Hot carcass weight, kg 375 375 378 374 377 377 7.6 6.3 Marbling score4 345 329 294 325 327 316 8.2 6.8 Ribeye area, cm2 89 93 103 95 95 95 1.7 1.4 Adjusted fat thickness, mm 10.5 9.1 6.4 8.9 8.9 8.3 0.8 0.6 Vision Yield Grade5 2.59 2.26 1.57 2.18 2.18 2.05 0.15 0.12 Slice shear force, kg 16.2 14.5 12.8 15.1 13.7 14.8 0.96 0.81 CIE L* reflectance6 34.65 35.58 36.20 35.40 35.58 35.44 0.45 0.38 1MSTN F94L homozygous F94LaF (FF), heterozygous (FL), and homozygous F94LaL (LL) genotypes. 2CAPN1 diplotypes are designated NN-NN (homozygous CAPN1hNN), CC-NN (heterozygotes) and CC-CC (homozygous CAPN1hCC). 3Approximate within gene SED for the difference between homozygotes (SEDhom) and between the heterozygote and either of the homozygotes (SEDhet). 4300 = Slight00; 400 = Small00 (USDA, 1997). 5USDA vision yield grade. Smaller numbers indicate greater yield of boneless, closely trimmed retail cuts. 6CIE L* measure of lightness. Lighter lean color results in greater values. Open in new tab Table 6. Means of traits by myostatin F94L genotypes and µ-calpain (CAPN1) diplotypes Trait . F94L1 . CAPN12 . SED3 . . FF . FL . LL . NN-NN . NN-CC . CC-CC . SEDhom . SEDhet . Birth weight, kg 40.7 41.4 43.3 41.7 41.3 42.4 1.2 1.0 Weaning weight, kg 202 199 206 204 202 201 4.6 3.9 Yearling weight, kg 409 410 412 408 409 413 8.5 7.1 Final weight, kg 608 603 598 599 604 607 11.6 9.7 Hot carcass weight, kg 375 375 378 374 377 377 7.6 6.3 Marbling score4 345 329 294 325 327 316 8.2 6.8 Ribeye area, cm2 89 93 103 95 95 95 1.7 1.4 Adjusted fat thickness, mm 10.5 9.1 6.4 8.9 8.9 8.3 0.8 0.6 Vision Yield Grade5 2.59 2.26 1.57 2.18 2.18 2.05 0.15 0.12 Slice shear force, kg 16.2 14.5 12.8 15.1 13.7 14.8 0.96 0.81 CIE L* reflectance6 34.65 35.58 36.20 35.40 35.58 35.44 0.45 0.38 Trait . F94L1 . CAPN12 . SED3 . . FF . FL . LL . NN-NN . NN-CC . CC-CC . SEDhom . SEDhet . Birth weight, kg 40.7 41.4 43.3 41.7 41.3 42.4 1.2 1.0 Weaning weight, kg 202 199 206 204 202 201 4.6 3.9 Yearling weight, kg 409 410 412 408 409 413 8.5 7.1 Final weight, kg 608 603 598 599 604 607 11.6 9.7 Hot carcass weight, kg 375 375 378 374 377 377 7.6 6.3 Marbling score4 345 329 294 325 327 316 8.2 6.8 Ribeye area, cm2 89 93 103 95 95 95 1.7 1.4 Adjusted fat thickness, mm 10.5 9.1 6.4 8.9 8.9 8.3 0.8 0.6 Vision Yield Grade5 2.59 2.26 1.57 2.18 2.18 2.05 0.15 0.12 Slice shear force, kg 16.2 14.5 12.8 15.1 13.7 14.8 0.96 0.81 CIE L* reflectance6 34.65 35.58 36.20 35.40 35.58 35.44 0.45 0.38 1MSTN F94L homozygous F94LaF (FF), heterozygous (FL), and homozygous F94LaL (LL) genotypes. 2CAPN1 diplotypes are designated NN-NN (homozygous CAPN1hNN), CC-NN (heterozygotes) and CC-CC (homozygous CAPN1hCC). 3Approximate within gene SED for the difference between homozygotes (SEDhom) and between the heterozygote and either of the homozygotes (SEDhet). 4300 = Slight00; 400 = Small00 (USDA, 1997). 5USDA vision yield grade. Smaller numbers indicate greater yield of boneless, closely trimmed retail cuts. 6CIE L* measure of lightness. Lighter lean color results in greater values. Open in new tab Linear contrasts for genetic effects were estimated for significant carcass and meat traits (Table 7). Additive effects of MSTN F94L were significant (P < 0.01) for all these traits. Reduced fat thickness, larger ribeye area, and better yield grade were associated with F94LaL. It was also associated with lower marbling scores, more tender meat, and lighter meat color. The F94LaL was partially recessive to F94LaF for ribeye area resulting in the heterozygote being closer to the F94LaF homozygote (Tables 6 and 7). The estimated differences of homozygous F94LaF, heterozygous, and homozygous F94LaL on traits with significant additive and dominance effects are shown in Fig. 2. Differences are standardized by subtracting the average of the homozygotes and dividing by phenotypic SD. Although genotype effect (8 df) was not significant in this study (P = 0.38; Table 5), birth weight is also shown because it could affect use of F94LaL in mating systems, approached significance (P = 0.06) in heifers that were sibs to these steers (Cushman et al., 2015), and has shown significant increases in other homozygous MSTN mutations (e.g., Casas et al., 2004). Table 7. Estimated marker associated additive and nonadditive effects for traits with overall P < 0.10 for genotype Marker effect1 . Marbling score2 . Ribeye area, cm2 . Adjusted fat thickness, mm . Vision Yield Grade . Slice shear force2, kg . L* reflectance . . Value ± SE . Value ± SE . Value ± SE . Value ± SE . Value ± SE . Value ± SE . F94L A −25.4*** ± 3.9 6.6*** ± 0.8 −2.06*** ± 0.36 −0.57*** ± 0.07 −1.7** ± 0.7 0.78*** ± 0.21 F94L D 9.4 ± 5.6 −2.7* ± 1.2 0.61 ± 0.52 0.18 ± 0.10 0.0 ± 0.9 0.16 ± 0.30 CAPN1 A −4.7 ± 4.3 0.3 ± 0.9 −0.32 ± 0.40 0.07 ± 0.08 −0.1 ± 0.7 0.02 ± 0.24 CAPN1 D 5.8 ± 5.5 0.0 ± 1.1 0.30 ± 0.51 0.07 ± 0.10 −1.3 ± 0.9 0.16 ± 0.30 A × A 13.4 ± 20.4 −2.3 ± 4.3 0.61 ± 1.90 0.06 ± 0.37 7.3* ± 3.7 0.49 ± 1.12 A × D 9.2 ± 15.0 5.2 ± 3.1 −1.03 ± 1.40 −0.23 ± 0.27 −0.8 ± 2.5 0.80 ± 0.82 D × A 13.2 ± 14.4 −2.6 ± 3.0 −0.01 ± 1.37 0.21 ± 0.26 −0.3 ± 2.4 0.96 ± 0.79 D × D 4.1 ± 10.3 2.3 ± 2.2 1.06 ± 1.02 0.05 ± 0.19 −0.1 ± 1.7 −0.51 ± 0.56 Marker effect1 . Marbling score2 . Ribeye area, cm2 . Adjusted fat thickness, mm . Vision Yield Grade . Slice shear force2, kg . L* reflectance . . Value ± SE . Value ± SE . Value ± SE . Value ± SE . Value ± SE . Value ± SE . F94L A −25.4*** ± 3.9 6.6*** ± 0.8 −2.06*** ± 0.36 −0.57*** ± 0.07 −1.7** ± 0.7 0.78*** ± 0.21 F94L D 9.4 ± 5.6 −2.7* ± 1.2 0.61 ± 0.52 0.18 ± 0.10 0.0 ± 0.9 0.16 ± 0.30 CAPN1 A −4.7 ± 4.3 0.3 ± 0.9 −0.32 ± 0.40 0.07 ± 0.08 −0.1 ± 0.7 0.02 ± 0.24 CAPN1 D 5.8 ± 5.5 0.0 ± 1.1 0.30 ± 0.51 0.07 ± 0.10 −1.3 ± 0.9 0.16 ± 0.30 A × A 13.4 ± 20.4 −2.3 ± 4.3 0.61 ± 1.90 0.06 ± 0.37 7.3* ± 3.7 0.49 ± 1.12 A × D 9.2 ± 15.0 5.2 ± 3.1 −1.03 ± 1.40 −0.23 ± 0.27 −0.8 ± 2.5 0.80 ± 0.82 D × A 13.2 ± 14.4 −2.6 ± 3.0 −0.01 ± 1.37 0.21 ± 0.26 −0.3 ± 2.4 0.96 ± 0.79 D × D 4.1 ± 10.3 2.3 ± 2.2 1.06 ± 1.02 0.05 ± 0.19 −0.1 ± 1.7 −0.51 ± 0.56 1Epistatic effects are listed as myostatin F94L effect × µ-Calpain (CAPN1) haplotype effect. Linear contrast coefficient used to estimate additive (A), dominance (D), and epistatic (A × A, D × A, A × D, D × D) effects are shown in Table 1. 2Means and SE estimates from actual values. Significance determined from analyses of logarithms of data. *P < 0.05; **P < 0.01; ***P < 0.001. Open in new tab Table 7. Estimated marker associated additive and nonadditive effects for traits with overall P < 0.10 for genotype Marker effect1 . Marbling score2 . Ribeye area, cm2 . Adjusted fat thickness, mm . Vision Yield Grade . Slice shear force2, kg . L* reflectance . . Value ± SE . Value ± SE . Value ± SE . Value ± SE . Value ± SE . Value ± SE . F94L A −25.4*** ± 3.9 6.6*** ± 0.8 −2.06*** ± 0.36 −0.57*** ± 0.07 −1.7** ± 0.7 0.78*** ± 0.21 F94L D 9.4 ± 5.6 −2.7* ± 1.2 0.61 ± 0.52 0.18 ± 0.10 0.0 ± 0.9 0.16 ± 0.30 CAPN1 A −4.7 ± 4.3 0.3 ± 0.9 −0.32 ± 0.40 0.07 ± 0.08 −0.1 ± 0.7 0.02 ± 0.24 CAPN1 D 5.8 ± 5.5 0.0 ± 1.1 0.30 ± 0.51 0.07 ± 0.10 −1.3 ± 0.9 0.16 ± 0.30 A × A 13.4 ± 20.4 −2.3 ± 4.3 0.61 ± 1.90 0.06 ± 0.37 7.3* ± 3.7 0.49 ± 1.12 A × D 9.2 ± 15.0 5.2 ± 3.1 −1.03 ± 1.40 −0.23 ± 0.27 −0.8 ± 2.5 0.80 ± 0.82 D × A 13.2 ± 14.4 −2.6 ± 3.0 −0.01 ± 1.37 0.21 ± 0.26 −0.3 ± 2.4 0.96 ± 0.79 D × D 4.1 ± 10.3 2.3 ± 2.2 1.06 ± 1.02 0.05 ± 0.19 −0.1 ± 1.7 −0.51 ± 0.56 Marker effect1 . Marbling score2 . Ribeye area, cm2 . Adjusted fat thickness, mm . Vision Yield Grade . Slice shear force2, kg . L* reflectance . . Value ± SE . Value ± SE . Value ± SE . Value ± SE . Value ± SE . Value ± SE . F94L A −25.4*** ± 3.9 6.6*** ± 0.8 −2.06*** ± 0.36 −0.57*** ± 0.07 −1.7** ± 0.7 0.78*** ± 0.21 F94L D 9.4 ± 5.6 −2.7* ± 1.2 0.61 ± 0.52 0.18 ± 0.10 0.0 ± 0.9 0.16 ± 0.30 CAPN1 A −4.7 ± 4.3 0.3 ± 0.9 −0.32 ± 0.40 0.07 ± 0.08 −0.1 ± 0.7 0.02 ± 0.24 CAPN1 D 5.8 ± 5.5 0.0 ± 1.1 0.30 ± 0.51 0.07 ± 0.10 −1.3 ± 0.9 0.16 ± 0.30 A × A 13.4 ± 20.4 −2.3 ± 4.3 0.61 ± 1.90 0.06 ± 0.37 7.3* ± 3.7 0.49 ± 1.12 A × D 9.2 ± 15.0 5.2 ± 3.1 −1.03 ± 1.40 −0.23 ± 0.27 −0.8 ± 2.5 0.80 ± 0.82 D × A 13.2 ± 14.4 −2.6 ± 3.0 −0.01 ± 1.37 0.21 ± 0.26 −0.3 ± 2.4 0.96 ± 0.79 D × D 4.1 ± 10.3 2.3 ± 2.2 1.06 ± 1.02 0.05 ± 0.19 −0.1 ± 1.7 −0.51 ± 0.56 1Epistatic effects are listed as myostatin F94L effect × µ-Calpain (CAPN1) haplotype effect. Linear contrast coefficient used to estimate additive (A), dominance (D), and epistatic (A × A, D × A, A × D, D × D) effects are shown in Table 1. 2Means and SE estimates from actual values. Significance determined from analyses of logarithms of data. *P < 0.05; **P < 0.01; ***P < 0.001. Open in new tab Figure 2. Open in new tabDownload slide Means for MSTN F94L genotypes divided by their phenotypic SD and deviated from the average of F94LaF (FF) and F94LaL (LL) homozygotes. All differences between divergent homozygotes are significant except birth weight. Heterozygotes were different from the average of homozygotes for ribeye muscle area (P < 0.05). Birth weight is included for comparison with previous literature reports of significant differences. Figure 2. Open in new tabDownload slide Means for MSTN F94L genotypes divided by their phenotypic SD and deviated from the average of F94LaF (FF) and F94LaL (LL) homozygotes. All differences between divergent homozygotes are significant except birth weight. Heterozygotes were different from the average of homozygotes for ribeye muscle area (P < 0.05). Birth weight is included for comparison with previous literature reports of significant differences. Multiple variants in MSTN that decrease functional activity and cause muscular hypertrophy in cattle have been found (Dunner et al., 2003). An 11 base pair deletion in the Belgian Blue breed causes a frame shift in the translation frame that prevents translation of the active signaling domain of the protein. In Piedmontese, there is a single base change that eliminates a proteolytic self-cleavage site immediately proximal to the signaling domain that prevents process to the active form. These 2 highly disruptive variants contrast with the relatively minor substitution of an aliphatic side chain (leucine) for an aromatic (phenylalanine) in the propeptide domain resulting from the F94L variant, because both are nonpolar residues. However, it is possible this substitution affects protein folding, stability, or trafficking of the myostatin protein with the primary evidence for this being the effect on muscle growth in cattle (Alexander et al., 2007). Differences between F94L homozygote means in this study ranged from 1.3 to 1.9 phenotypic SD for fat and muscle traits. Most carcass traits show a nonsignificant tendency for the heterozygote to be partially recessive to the F94LaF allele (the mean being closer to the homozygous F94LaF mean than the homozygous F94LaL mean), although the effect did reach significance for ribeye area. Esmailizadeh et al. (2008) reported on Limousin-Jersey back-cross families in Australia and New Zealand. They found no significant effects of the F94LaL SNP on birth and live weights. Additive and dominance effects on Longissimus muscle area were significant but not as large as the additive effects in this study. Several measures of fatness were also decreased, and meat weights increased. Alexander et al. (2009) found increased muscle area and reduced marbling in a Wagyu-Limousin F2 family. Differences in birth weight, fat, and muscle traits between homozygotes for some of the severe MSTN variants estimated in 2 other experiments exceed the values for F94L estimated in this experiment. Casas et al. (2004) compared homozygous active and inactive F2 progeny from Belgian Blue F1 × F1 matings in the same location and under similar management as this experiment. Short et al. (2002) compared F2 Piedmontese under similar management but in a different location. Both experiments found significant differences between divergent homozygotes for birth weight, ribeye area, marbling score, fat thickness, and yield grade. The average percentage differences from the active MSTN homozygotes were 15%, 32%, −32%, −62%, and −78%, respectively, compared with 7%, 15%, −14%, −39%, and −38% for F94L in this experiment. The effects of the F94L mutation were about half the percentage differences of the Belgian Blue and Piedmontese mutations. The CAPN1 haplotypes had no significant additive or dominance associations with any measured trait including meat tenderness. A significant CAPN1 additive by F94L additive epistatic effect is illustrated in Fig. 3. One way of characterizing this epistasis is that CAPN1hCC reduced slice shear force (increased tenderness) in animals homozygous for the common F94LaF allele, had no effect in F94L heterozygotes, and decreased tenderness in F94LaL homozygotes. Increased tenderness is often observed in meat from animals with heterozygous and homozygous MSTN mutations resulting in nonfunctional myostatin (e.g., Wheeler et al., 2001). Because few animals are homozygous CAPN1hCC in most common populations surveyed (White et al., 2005), F94LaL would be associated with increased tenderness in most populations using the epistatic estimates. Frequency of CAPN1hCC was increased in 2 similar experiments. In an Angus population (Tait et al., 2014a), the estimated additive effect of CAPN1aCC (compared to CAPN1aGT) on slice shear force was −1.05 ± 0.25 kg and in the composite MARC III population (Tait et al., 2014b) was −1.15 ± 0.48 kg. Using only estimated means for homozygous F94LaF, the equivalent additive estimate was −1.93 ± 0.97 kg in the current study. The usual relationship between CAPN1 haplotypes and meat tenderness appears to be disrupted by the F94LaL allele. Previous QTL discovery in progeny of an F1 Piedmontese sire and an F1 Belgian Blue sire heterozygous for active and inactive myostatin found interactions of the MSTN variants with other QTL located on BTA 4 (Casas et al., 2001) and BTA 5 (Casas et al., 2000) for meat tenderness. Figure 3. Open in new tabDownload slide MSTN F94L × µ-calpain genotypic means for slice shear force. F94L homozygous F94LaF, heterozygous, and homozygous F94LaL genotypes are designated by FF, FL, and LL, respectively. CAPN1 diplotypes are designated GT-GT (homozygous CAPN1hGT), CC-GT (heterozygotes) and CC-CC (homozygous CAPN1hCC). The additive effect is significant for F94L (P < 0.01) and the additive F94L × additive µ-calpain effect is significant (P < 0.05). Variation is shown as ±1 SEM. Figure 3. Open in new tabDownload slide MSTN F94L × µ-calpain genotypic means for slice shear force. F94L homozygous F94LaF, heterozygous, and homozygous F94LaL genotypes are designated by FF, FL, and LL, respectively. CAPN1 diplotypes are designated GT-GT (homozygous CAPN1hGT), CC-GT (heterozygotes) and CC-CC (homozygous CAPN1hCC). The additive effect is significant for F94L (P < 0.01) and the additive F94L × additive µ-calpain effect is significant (P < 0.05). Variation is shown as ±1 SEM. Research using heifer half-sibs from this population showed delayed age of puberty due to F94LaL alleles (Cushman et al., 2015). However, conception was not reduced or delayed in this population and management system. This moderate form of MSTN mutation with high frequency in the Limousin breed shows the potential for variation among the effects of mutations in some genes. In this case, the resulting increase in lean meat yield, moderate birth weight increase, and limited effect on heifer conception is a useful resource for improving efficiency of lean beef production in conventional production systems. Because F94L effects are mostly additive, 0, 1, or 2 copies of F94LaL can be used to create 3 product types with increasing lean meat yield and decreasing marbling. Terminal cross bulls with 1 or 2 copies of F94LaL mated to cows with 0 copies would produce 0 and 1 copy progeny (1 copy bulls) or 1 copy progeny (2 copy bulls). Based on the epistatic estimates for slice shear force, CAPN1 selection would be less effective for progeny with 1 copy but effective for progeny with 0 copies. Footnotes 1 Mention of trade name, proprietary product, or specified equipment does not constitute a guarantee or warranty by the USDA and does not imply approval to the exclusion of other products that may be suitable. 2 The USDA is an equal opportunity provider and employer. 3 The authors acknowledge the important contributions of technicians and cattle operations staff to the conduct and completion of this study. LITERATURE CITED Alexander , L. J. , T. W. Geary, W. M. Snelling, and M. D. Macneil. 2007 . Quantitative trait loci with additive effects on growth and carcass traits in a wagyu-limousin F2 population . Anim. Genet . 38 : 413 – 416 . doi: 10.1111/j.1365-2052.2007.01616.x Google Scholar Crossref Search ADS PubMed WorldCat Alexander , L. J. , L. A. Kuehn, T. P. Smith, L. K. Matukumalli, B. Mote, J. E. Koltes, J. Reecy, T. W. Geary, D. C. Rule, and M. D. Macneil. 2009 . A limousin specific myostatin allele affects longissimus muscle area and fatty acid profiles in a wagyu-limousin F2 population . J. Anim. Sci . 87 : 1576 – 1581 . doi: 10.2527/jas.2008-1531 Google Scholar Crossref Search ADS PubMed WorldCat Arthur , P. F . 1995 . Double muscling in cattle: a review . Aust. J. Agric. Res . 46 : 1493 – 1515 . doi: 10.1071/AR9951493 Google Scholar Crossref Search ADS WorldCat Bennett , G. L . 2008 . Experimental selection for calving ease and postnatal growth in seven cattle populations. I. Changes in estimated breeding values . J. Anim. Sci . 86 : 2093 – 2102 . doi: 10.2527/jas.2007-0767 Google Scholar Crossref Search ADS PubMed WorldCat Bennett , G. L. , and K. E. Gregory. 1996 . Genetic (co)variances among birth weight, 200-day weight, and postweaning gain in composites and parental breeds of beef cattle . J. Anim. Sci . 74 : 2598 – 2611 . doi: 10.2527/1996.74112598x Google Scholar Crossref Search ADS PubMed WorldCat Boldman , K. G. , L. A. Kriese, L. D. Van Vleck, C. P. Van Tassell, and S. D. Kachman. 1995 . A manual for use of MTDFREML. A set of programs to obtain estimates of variance and covariances . USDA, Agricultural Research Service , Washington, DC . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Casas , E. , G. L. Bennett, T. P. Smith, and L. V. Cundiff. 2004 . Association of myostatin on early calf mortality, growth, and carcass composition traits in crossbred cattle . J. Anim. Sci . 82 : 2913 – 2918 . doi: 10.2527/2004.82102913x Google Scholar Crossref Search ADS PubMed WorldCat Casas , E. , S. D. Shackelford, J. W. Keele, R. T. Stone, S. M. Kappes, and M. Koohmaraie. 2000 . Quantitative trait loci affecting growth and carcass composition of cattle segregating alternate forms of myostatin . J. Anim. Sci . 78 : 560 – 569 . doi: 10.2527/2000.783560x Google Scholar Crossref Search ADS PubMed WorldCat Casas , E. , R. T. Stone, J. W. Keele, S. D. Shackelford, S. M. Kappes, and M. Koohmaraie. 2001 . A comprehensive search for quantitative trait loci affecting growth and carcass composition of cattle segregating alternative forms of the myostatin gene . J. Anim. Sci . 79 : 854 – 860 . doi: 10.2527/2001.794854x Google Scholar Crossref Search ADS PubMed WorldCat Casas , E. , S. N. White, T. L. Wheeler, S. D. Shackelford, M. Koohmaraie, D. G. Riley, C. C. Chase , Jr, D. D. Johnson, and T. P. Smith. 2006 . Effects of calpastatin and micro-calpain markers in beef cattle on tenderness traits . J. Anim. Sci . 84 : 520 – 525 . doi: 10.2527/2006.843520x Google Scholar Crossref Search ADS PubMed WorldCat Cushman , R. A. , R. G. Tait , Jr, A. K. McNeel, E. D. Forbes, O. L. Amundson, C. A. Lents, A. K. Lindholm-Perry, G. A. Perry, J. R. Wood, A. S. Cupp,et al. 2015 . A polymorphism in myostatin influences puberty but not fertility in beef heifers, whereas µ-calpain affects first calf birth weight . J. Anim. Sci . 93 : 117 – 126 . doi: 10.2527/jas.2014-8505 Google Scholar Crossref Search ADS PubMed WorldCat Dunner , S. , M. E. Miranda, Y. Amigues, J. Cañón, M. Georges, R. Hanset, J. Williams, and F. Ménissier. 2003 . Haplotype diversity of the myostatin gene among beef cattle breeds . Genet. Sel. Evol . 35 : 103 – 118 . doi: 10.1051/gse:2002038 Google Scholar Crossref Search ADS PubMed WorldCat Esmailizadeh , A. K. , C. D. Bottema, G. S. Sellick, A. P. Verbyla, C. A. Morris, N. G. Cullen, and W. S. Pitchford. 2008 . Effects of the myostatin F94L substitution on beef traits . J. Anim. Sci . 86 : 1038 – 1046 . doi: 10.2527/jas.2007-0589 Google Scholar Crossref Search ADS PubMed WorldCat FASS . 1999 . Guide for the care and use of agricultural animals in agricultural research and teaching . 1st rev. ed. Fed. Anim. Sci. Soc ., Savoy, IL . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gregory , K. E. , L. V. Cundiff, and R. M. Koch. 1991 . Breed effects and heterosis in advanced generations of composite populations for preweaning traits of beef cattle . J. Anim. Sci . 69 : 947 – 960 . doi: 10.2527/1991.693947x Google Scholar Crossref Search ADS PubMed WorldCat Gregory , K. E. , L. V. Cundiff, R. M. Koch, M. E. Dikeman, and M. Koohmaraie. 1994 . Breed effects, retained heterosis, and estimates of genetic and phenotypic parameters for carcass and meat traits of beef cattle . J. Anim. Sci . 72 : 1174 – 1183 . doi: 10.2527/1994.7251174x Google Scholar Crossref Search ADS PubMed WorldCat Grobet , L. , L. J. Martin, D. Poncelet, D. Pirottin, B. Brouwers, J. Riquet, A. Schoeberlein, S. Dunner, F. Ménissier, J. Massabanda,et al. 1997 . A deletion in the bovine myostatin gene causes the double-muscled phenotype in cattle . Nat. Genet . 17 : 71 – 74 . doi: 10.1038/ng0997-71 Google Scholar Crossref Search ADS PubMed WorldCat Grobet , L. , D. Poncelet, L. J. Royo, B. Brouwers, D. Pirottin, C. Michaux, F. Ménissier, M. Zanotti, S. Dunner, and M. Georges. 1998 . Molecular definition of an allelic series of mutations disrupting the myostatin function and causing double-muscling in cattle . Mamm. Genome 9 : 210 – 213 . doi: 10.1007/s003359900727 Google Scholar Crossref Search ADS PubMed WorldCat Kambadur , R. , M. Sharma, T. P. Smith, and J. J. Bass. 1997 . Mutations in myostatin (GDF8) in double-muscled Belgian Blue and Piedmontese cattle . Genome Res . 7 : 910 – 916 . doi: 10.1101/gr.7.9.910 Google Scholar Crossref Search ADS PubMed WorldCat Page , B. T. , E. Casas, M. P. Heaton, N. G. Cullen, D. L. Hyndman, C. A. Morris, A. M. Crawford, T. L. Wheeler, M. Koohmaraie, J. W. Keele,et al. 2002 . Evaluation of single-nucleotide polymorphisms in CAPN1 for association with meat tenderness in cattle . J. Anim. Sci . 80 : 3077 – 3085 . doi: 10.2527/2002.80123077x Google Scholar Crossref Search ADS PubMed WorldCat Page , B. T. , E. Casas, R. L. Quaas, R. M. Thallman, T. L. Wheeler, S. D. Shackelford, M. Koohmaraie, S. N. White, G. L. Bennett, J. W. Keele,et al. 2004 . Association of markers in the bovine CAPN1 gene with meat tenderness in large crossbred populations that sample influential industry sires . J. Anim. Sci . 82 : 3474 – 3481 . doi: 10.2527/2004.82123474x Google Scholar Crossref Search ADS PubMed WorldCat Robinson , D. L. , L. M. Cafe, B. L. McIntyre, G. H. Geesink, W. Barendse, D. W. Pethick, J. M. Thompson, R. Polkinghorne, and P. L. Greenwood. 2012 . Production and processing studies on calpain-system gene markers for beef tenderness: consumer assessments of eating quality . J. Anim. Sci . 90 : 2850 – 2860 . doi: 10.2527/jas.2011-4928 Google Scholar Crossref Search ADS PubMed WorldCat Shackelford , S. D. , T. L. Wheeler, and M. Koohmaraie. 1999 . Evaluation of slice shear force as an objective method of assessing beef longissimus tenderness . J. Anim. Sci . 77 : 2693 – 2699 . doi: 10.2527/1999.77102693x Google Scholar Crossref Search ADS PubMed WorldCat Shackelford , S. D. , T. L. Wheeler, and M. Koohmaraie. 2003 . On-line prediction of yield grade, longissimus muscle area, preliminary yield grade, adjusted preliminary yield grade, and marbling score using the MARC beef carcass image analysis system . J. Anim. Sci . 81 : 150 – 155 . doi: 10.2527/2003.811150x Google Scholar Crossref Search ADS PubMed WorldCat Short , R. E. , M. D. MacNeil, M. D. Grosz, D. E. Gerrard, and E. E. Grings. 2002 . Pleiotropic effects in hereford, limousin, and piedmontese F2 crossbred calves of genes controlling muscularity including the piedmontese myostatin allele . J. Anim. Sci . 80 : 1 – 11 . doi:10.2527/2002.8011 Google Scholar Crossref Search ADS PubMed WorldCat Stone , R. T. , W. M. Grosse, E. Casas, T. P. Smith, J. W. Keele, and G. L. Bennett. 2002 . Use of bovine EST data and human genomic sequences to map 100 gene-specific bovine markers . Mamm. Genome 13 : 211 – 215 . doi: 10.1007/s00335-001-2124-9 Google Scholar Crossref Search ADS PubMed WorldCat Tait , R. G. , Jr, R. A. Cushman, A. K. McNeel, E. Casas, T. P. Smith, H. C. Freetly, and G. L. Bennett. 2016 . Estimates of epistatic and pleiotropic effects of casein alpha s1 (CSN1S1) and thyroglobulin (TG) genetic markers on beef heifer performance traits enhanced by selection . J. Anim. Sci . 94 : 920 – 926 . doi: 10.2527/jas.2015-9860 Google Scholar Crossref Search ADS PubMed WorldCat Tait , R. G. , Jr, S. D. Shackelford, T. L. Wheeler, D. A. King, E. Casas, R. M. Thallman, T. P. L. Smith, and G. L. Bennett. 2014a . µ-Calpain, calpastatin, and growth hormone receptor genetic effects on preweaning performance, carcass quality traits, and residual variance of tenderness in Angus cattle selected to increase minor haplotype and allele frequencies . J. Anim. Sci . 92 : 456 – 466 . doi: 10.2527/jas.2013-7075 Google Scholar Crossref Search ADS WorldCat Tait , R. G. , Jr, S. D. Shackelford, T. L. Wheeler, D. A. King, J. W. Keele, E. Casas, T. P. Smith, and G. L. Bennett. 2014b . CAPN1, CAST, and DGAT1 genetic effects on preweaning performance, carcass quality traits, and residual variance of tenderness in a beef cattle population selected for haplotype and allele equalization . J. Anim. Sci . 92 : 5382 – 5393 . doi: 10.2527/jas.2014-8211 Google Scholar Crossref Search ADS WorldCat USDA . 1997 . Official United States standards for grades of carcass beef . Agric. Market. Serv., USDA , Washington, DC . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Wheeler , T. L. , S. D. Shackelford, E. Casas, L. V. Cundiff, and M. Koohmaraie. 2001 . The effects of piedmontese inheritance and myostatin genotype on the palatability of longissimus thoracis, gluteus medius, semimembranosus, and biceps femoris . J. Anim. Sci . 79 : 3069 – 3074 . doi: 10.2527/2001.79123069x Google Scholar Crossref Search ADS PubMed WorldCat White , S. N. , E. Casas, T. L. Wheeler, S. D. Shackelford, M. Koohmaraie, D. G. Riley, C. C. Chase , Jr, D. D. Johnson, J. W. Keele, and T. P. Smith. 2005 . A new single nucleotide polymorphism in CAPN1 extends the current tenderness marker test to include cattle of Bos indicus, Bos taurus, and crossbred descent . J. Anim. Sci . 83 : 2001 – 2008 . doi: 10.2527/2005.8392001x Google Scholar Crossref Search ADS PubMed WorldCat Published by Oxford University Press on behalf of the American Society of Animal Science 2018. This work is written by (a) US Government employee(s) and is in the public domain in the US. Published by Oxford University Press on behalf of the American Society of Animal Science 2018.

Freking, Brad A; Bennett, Gary L

doi: 10.1093/jas/sky474pmid: 30561683

Abstract Objectives were to estimate reciprocal effects of Romanov and Rambouillet breeds on survival, growth, and reproductive traits of F1 progeny and direct breed effects (Suffolk and Composite – ½ Columbia, ¼ Hampshire, and ¼ Suffolk) on survival and growth traits of the subsequent terminally sired lambs. Mature Rambouillet ewes (n = 243) were exposed to 20 Romanov rams over two seasons producing 621 lambs for evaluation of growth and survival traits with 274 F1 ewes being evaluated for reproduction traits through 4 yr of age. Similarly, mature Romanov ewes (n = 116) were exposed to 20 Rambouillet rams producing 601 lambs for evaluation of growth and survival traits with 176 F1 ewes being evaluated for reproduction traits through 4 yr of age. A total of 433 of those F1 ewes produced 3,431 lambs (1,552 litters) from 1,634 exposures to terminal sires over 4 yr. Terminal sires consisted of 38 Suffolk and 44 Composite rams. Reciprocal crossbred ewe lambs were produced from dramatically different uterine and neonatal environments, with litter size at birth from Romanov dams exceeding those from Rambouillet dams by 1.52 lambs (P < 0.001) and birth weight of lambs from Romanov dams averaged 3.41 kg compared with 4.26 kg from Rambouillet dams. Differences in BW were still evident at 140 d (P < 0.001) for dam-reared lambs. However, reciprocal ewe first breeding BW of both types were similar (P = 0.38). Minimal differences were observed in performance of reciprocal cross ewes through 4 yr for productivity, longevity, or progeny growth and survival. One exception was BW at 140 d where an interaction of dam breed with terminal sire breed reached significance for both dam-reared (P = 0.05) and nursery-reared (P = 0.02) lambs. This interaction was due to the lower weight of Composite-sired lambs out of reciprocal cross ewes born from Rambouillet dams. Composite rams increased number born (P < 0.01) and number weaned (P < 0.05) of the reciprocal cross ewes. Suffolk rams increased (P < 0.001) BW and growth rates from birth to 140 d of terminal progeny. Thus, there were little cumulative differences accrued over the 4 yr and no differences were detected for cumulative kilogram of lamb generated at 140 d per ewe exposed. The practical outcome of this evaluation was that performance levels of both types of Romanov crossbred ewes was similar allowing the industry to produce the desired crossbred ewes without needing large purebred ewe flocks of the less numerous Romanov breed. INTRODUCTION Improved ewe productivity has been identified as a critical priority to allow sustained competitiveness in the global market for U.S. lamb production (ASI, 2016). Within the common annual production system of fall breeding and spring lambing, biological efficiency was shown to be most influenced by survival, fertility, and prolificacy (Wang and Dickerson, 1991a,b,c). Romanov sheep have documented superior levels of performance for survival, fertility, prolificacy, and length of seasonal fertility (Casas et al., 2004, 2005; Freking et al., 2004). Efficiency of commercial sheep production could be improved markedly by greater industry use of specialized superior dam lines as maternal contributors in terminal crossbreeding systems. An important issue for commercial producers is the relative performance of F1 replacement ewes sired by Romanov rams compared to F1 ewes produced by Romanov dams. Potential genetic causes of reciprocal breed effects include maternal and paternal breed effects, maternal and paternal imprinting effects, and mitochondrial effects. While little is known about imprinting and mitochondrial effects in sheep, maternal breed effects are well documented (Bradford, 1972). Significant maternal effects on prenatal survival, birth weight, postnatal survival, and postnatal growth are often detected and associated with differences in litter size. Maternal effects on reproductive traits have been less studied. Rambouillet and Romanov represent an extreme contrast in average litter size of roughly 2.0 lambs and thus are excellent resources to investigate reciprocal breed effects. Our objectives in this experiment were first to estimate Rambouillet and Romanov reciprocal breed effects on survival and growth traits of F1 lambs and on reproductive traits of F1 ewes. Secondly, our objective was to estimate direct breed effects on survival and growth traits of crossbred lambs sired by Suffolk and Composite rams. MATERIALS AND METHODS General Experimental Design and Traits Recorded The U.S. Meat Animal Research Center (USMARC) Institutional Animal Care and Use Committee approved the experiment following recommendations by FASS (1999). The primary experimental objective is to estimate reciprocal breed (Romanov vs. Rambouillet) effects on reproductive traits of F1 ewes (Figure 1). The intent was to produce at least 200 breeding ewes of each reciprocal cross over 2 yr (2005 and 2006) for subsequent evaluation of reproductive traits under annual production systems at 1, 2, 3, and 4 yr of age (2006 to 2010). Given the expected reproductive rate of the breeds, twice as many Rambouillet ewes were needed to generate roughly equivalent numbers of crossbred females. Actual numbers of ewes exposed to rams for each year and mating type are represented in Table 1. The experiment consisted of two phases that represented distinct generations: production of reciprocal crosses of F1 lambs; and production of terminally-sired lambs by reciprocal crosses of F1 ewes. Romanov ewes and rams were sampled from the USMARC flock which represented most of the available diversity of the Romanov breed in the U.S. Purebred Rambouillet ewes and rams were purchased from seven different producers in Texas. Breeding flocks of Romanov and Rambouillet ewes were similar in age distribution and all ewes at least 3 yr of age at breeding. Twenty rams (10 per year) from each breed contributed progeny for the production and evaluation of F1 crosses in this experiment. Table 1. Number of Romanov, Rambouillet, and reciprocal F1 cross ewes joined by mating types and year of lambing subclasses . Year of lambing . Breed of Ewe . Breed of Ram . 2005 . 2006 . 2007 . 2008 . 2009 . 2010 . Romanov Rambouillet 116 94 Rambouillet Romanov 243 178 F1 Paternal Romanov Suffolk 74 132 124 120 52 F1 Paternal Romanov Composite 75 133 121 117 54 F1 Paternal Rambouillet Suffolk 50 83 75 70 33 F1 Paternal Rambouillet Composite 50 85 81 72 33 . Year of lambing . Breed of Ewe . Breed of Ram . 2005 . 2006 . 2007 . 2008 . 2009 . 2010 . Romanov Rambouillet 116 94 Rambouillet Romanov 243 178 F1 Paternal Romanov Suffolk 74 132 124 120 52 F1 Paternal Romanov Composite 75 133 121 117 54 F1 Paternal Rambouillet Suffolk 50 83 75 70 33 F1 Paternal Rambouillet Composite 50 85 81 72 33 Open in new tab Table 1. Number of Romanov, Rambouillet, and reciprocal F1 cross ewes joined by mating types and year of lambing subclasses . Year of lambing . Breed of Ewe . Breed of Ram . 2005 . 2006 . 2007 . 2008 . 2009 . 2010 . Romanov Rambouillet 116 94 Rambouillet Romanov 243 178 F1 Paternal Romanov Suffolk 74 132 124 120 52 F1 Paternal Romanov Composite 75 133 121 117 54 F1 Paternal Rambouillet Suffolk 50 83 75 70 33 F1 Paternal Rambouillet Composite 50 85 81 72 33 . Year of lambing . Breed of Ewe . Breed of Ram . 2005 . 2006 . 2007 . 2008 . 2009 . 2010 . Romanov Rambouillet 116 94 Rambouillet Romanov 243 178 F1 Paternal Romanov Suffolk 74 132 124 120 52 F1 Paternal Romanov Composite 75 133 121 117 54 F1 Paternal Rambouillet Suffolk 50 83 75 70 33 F1 Paternal Rambouillet Composite 50 85 81 72 33 Open in new tab Figure 1. Open in new tabDownload slide Picture depicting the design contrast of this reciprocal cross experiment. Purebred populations of these two breeds of ewes have an extreme difference in reproductive rate. The top two panels represent Rambouillet rams mated to Romanov ewes where the average litter size of mature ewes is about 3.7 lambs born per ewe lambing. The bottom two panels represent Romanov rams mated to Rambouillet ewes where the average litter size of mature ewes is about 1.7 lambs born per ewe lambing. Figure 1. Open in new tabDownload slide Picture depicting the design contrast of this reciprocal cross experiment. Purebred populations of these two breeds of ewes have an extreme difference in reproductive rate. The top two panels represent Rambouillet rams mated to Romanov ewes where the average litter size of mature ewes is about 3.7 lambs born per ewe lambing. The bottom two panels represent Romanov rams mated to Rambouillet ewes where the average litter size of mature ewes is about 1.7 lambs born per ewe lambing. Traits evaluated were divided into those measured from the start of the breeding season to weaning of the lambs, those measured postweaning to 140 d of age of the lambs, and those measured cumulatively after ewes were 4 yr of age. Traits measured from the start of the breeding season to weaning were weight of the ewe at the beginning of the first breeding season, conception rate (ewe exposed but did not lamb = 0; ewe lambed = 1), number born, litter birth weight, number of dam and nursery-reared weaned lambs, and dam and nursery-reared litter weaning weight. Traits measured at 140 d of age were the number of dam and nursery-reared lambs, dam and nursery-reared litter weight per ewe lambing, and dam and nursery-reared litter weight per ewe exposed. Total productivity through 4 yr of age for each ewe entering the breeding flock was calculated as the sum of 140-d weights for dam or nursery-reared lambs. Traits measured at weaning and 140 d of age were defined separately for dam-reared and nursery-reared lambs to evaluate both aspects of reproduction of crossbred ewes. Phase I: Production of Reciprocal Crosses of F1 Lambs Breeding flocks of Romanov and Rambouillet ewes were similar in age distribution and ewes were at least 3 yr of age at breeding. A total of 116 (28 3 yr old; 50 4 yr old; 27 5 yr old; and 11 6 yr old) mature Romanov ewes were single-sire mated to 10 Rambouillet rams (ranged 4 to 7 yr age) during a 35-d breeding season beginning September of 2004. A total of 94 of those remaining Romanov ewes were again exposed to 10 new Rambouillet rams (ranged 1 to 4 yr of age) the following September of 2005. Likewise, 243 (51 3 yr old; 41 4 yr old; 61 5 yr old; 54 6 yr old; 22 7 yr old; and 14 8 yr old) mature Rambouillet ewes were single-sire mated to 10 Romanov rams (ranged 1 to 5 yr of age) in September of 2004, with 178 of the remaining ewes exposed to the second sample of 10 Romanov rams (ranged 1 to 3 yr of age) in September of 2005. Purebred ewes were shorn about 30 d before the start of lambing and separated by breed in drop pens. Ewes were limited to rearing two lambs, with additional lambs reared in the nursery without regard to sex of the lamb, although smaller lambs were typically selected. This approach was intended to help standardize early growth and development of the two types of crossbred ewe lambs available for subsequent evaluation. Ram lambs were castrated and tails docked on all lambs. Lambs were weighed at 0 (birth), 56 (weaning), 70, and 140 d of age. Nursery-reared lambs were weaned at ~35 d from milk replacer as they transitioned to creep feed rations and later re-joined contemporaries at the time of weaning for dam-reared lambs (56 d of age). Lambs were provided ad libitum access to a total mixed diet (18% CP) from creep to about 27 kg BW and then fed ad libitum a total mixed diet (2.96 Mcal of ME/kg of DM with 14.5% CP) during the finishing period. To account for variation in range of ages for a common weight date, individual lamb BW was adjusted to the intended target ages using the individuals own ADG for the period involved. After 140 d weights were recorded, replacement ewe lambs were identified and moved to pasture. Selection of F1 ewe lambs was based on routine culling for structural abnormalities and health, with additional culling of the bottom 10% to 15% based on adjusted 140 d BW, ignoring rearing status. Culling based on this BW was conducted separately within each line to reduce bias against potential growth differences between lines. Using this criterion, 132 dam-reared and 44 nursery-reared F1 ewes out of Romanov dams were selected for evaluation. Likewise, 249 dam-reared and 25 nursery-reared F1 ewes out of Rambouillet dams were initially selected. BW of all F1 ewes evaluated for reproduction was also recorded at the time of first breeding. Phase II: Production of Terminally Sired Lambs by Reciprocal Crosses of F1 Ewes F1 ewes were maintained as a single contemporary group except during breeding and lambing. Ewes were primarily managed on grass pastures with supplementation only if forage was limiting. Multi-sire breeding groups on pasture exposed half of the ewes to Suffolk rams the other half to Composite rams (Leymaster, 1991) during 35 d breeding seasons beginning each September. This composite was developed from ½ Columbia, ¼ Suffolk, and ¼ Hampshire germplasm. Each ewe was randomly assigned to either Suffolk or Composite rams for the duration of the experiment to facilitate investigation of potential effects of ram mating breed (Suffolk and Composite) on ewe longevity. To sample the ram breeds adequately, roughly one ram was used for each 15 ewes. These rams consisted of roughly equal numbers of 1 and 2 yr old rams at the beginning of the experiment. Once a ram was used in multi-sire mating, he continued to be used in subsequent years unless health or fertility problems occur. A total of 38 Suffolk and 44 Composite rams were utilized over the 4 yr of the experiment. The F1 ewes were evaluated for reproductive traits over 4 yr of production. Ewes born in 2005 were exposed to lamb in 2006 to 2009. Ewes born in 2006 were exposed to lamb in 2007 to 2010. Ewes were shorn about 30 d before the start of lambing and separated by reciprocal type in drop pens to avoid incorrect line assignment data due to mis-mothering. Ewes were limited to rearing two lambs, with additional lambs reared in the nursery without regard to sex of the lamb. Management and data recorded on Phase II lambs was the same as previously described for Phase I. Ewe longevity was recorded as presence or absence in the herd at about 50 mo age. Statistical Analysis Data were analyzed with the mixed-model analysis of variance procedure of SAS (SAS Inst., Inc., Cary, NC). In Phase I of the experiment where the reciprocal cross progeny were generated, the models included fixed effects of year (2005, 2006), and ewe line (Rambouillet, Romanov) for traits recorded on ewes (Table 3). Additional fixed effects for sex of lamb (male, female), type of birth (1, 2, 3, or 4+) for birth weight and survival traits to weaning, or rearing (1, 2) for subsequent BW traits, were added for analysis of traits recorded on lambs (Tables 2 and 4). Nursery-reared lambs were analyzed separately for traits after birth. Interactions among fixed effects that included line of ewe were also fitted. The random effect of individual rams nested within ewe line was included. Levels of significance associated with the effects of ewe line were tested with this individual ram within ewe line mean square and are considered approximations due to unbalanced data. Remaining fixed effects and interactions were tested against the residual mean square. Table 2. Levels of significance, least squares means, standard errors, and number of lambs for the effect of ewe breed producing F1 reciprocal cross lambs for growth and survival traits recorded on F1 lambs Item . Least squares means (N for F1 lambs) for breed of ewe . Romanov . Rambouillet . Level of significance . Birth weight, kg 3.41 ± 0.08 (601) 4.26 ± 0.10 (621) <0.0001 56d weight on dam, kg 14.10 ± 0.24 (237) 16.36 ± 0.17 (488) <0.0001 56d weight nursery, kg 13.71 ± 0.21 (184) 14.57 ± 0.38 (54) 0.05 70d weight on dam, kg 16.78 ± 0.30 (235) 19.22 ± 0.21 (485) <0.0001 70d weight nursery, kg 16.41 ± 0.27 (182) 17.24 ± 0.47 (51) 0.13 140d weight on dam, kg 47.53 ± 0.60 (234) 50.04 ± 0.49 (477) 0.002 140d weight nursery, kg 40.35 ± 0.74 (171) 39.65 ± 1.10 (48) 0.61 Survival to wean on dam, % 63.2 ± 2.8 (411) 82.3 ± 2.5 (568) <0.0001 Survival to wean nursery, % 78.0 ± 6.1 (190) 88.0 ± 7.2 (53) 0.17 Survival to 140 d on dam, % 60.5 ± 2.8 (411) 80.6 ± 2.6 (568) <0.0001 Survival to 140 d nursery, % 76.0 ± 6.5 (190) 85.3 ± 7.6 (53) 0.22 Ewe lamb breeding weight, kg 41.0 ± 0.46 (176) 41.6 ± 0.45 (274) 0.38 Item . Least squares means (N for F1 lambs) for breed of ewe . Romanov . Rambouillet . Level of significance . Birth weight, kg 3.41 ± 0.08 (601) 4.26 ± 0.10 (621) <0.0001 56d weight on dam, kg 14.10 ± 0.24 (237) 16.36 ± 0.17 (488) <0.0001 56d weight nursery, kg 13.71 ± 0.21 (184) 14.57 ± 0.38 (54) 0.05 70d weight on dam, kg 16.78 ± 0.30 (235) 19.22 ± 0.21 (485) <0.0001 70d weight nursery, kg 16.41 ± 0.27 (182) 17.24 ± 0.47 (51) 0.13 140d weight on dam, kg 47.53 ± 0.60 (234) 50.04 ± 0.49 (477) 0.002 140d weight nursery, kg 40.35 ± 0.74 (171) 39.65 ± 1.10 (48) 0.61 Survival to wean on dam, % 63.2 ± 2.8 (411) 82.3 ± 2.5 (568) <0.0001 Survival to wean nursery, % 78.0 ± 6.1 (190) 88.0 ± 7.2 (53) 0.17 Survival to 140 d on dam, % 60.5 ± 2.8 (411) 80.6 ± 2.6 (568) <0.0001 Survival to 140 d nursery, % 76.0 ± 6.5 (190) 85.3 ± 7.6 (53) 0.22 Ewe lamb breeding weight, kg 41.0 ± 0.46 (176) 41.6 ± 0.45 (274) 0.38 Open in new tab Table 2. Levels of significance, least squares means, standard errors, and number of lambs for the effect of ewe breed producing F1 reciprocal cross lambs for growth and survival traits recorded on F1 lambs Item . Least squares means (N for F1 lambs) for breed of ewe . Romanov . Rambouillet . Level of significance . Birth weight, kg 3.41 ± 0.08 (601) 4.26 ± 0.10 (621) <0.0001 56d weight on dam, kg 14.10 ± 0.24 (237) 16.36 ± 0.17 (488) <0.0001 56d weight nursery, kg 13.71 ± 0.21 (184) 14.57 ± 0.38 (54) 0.05 70d weight on dam, kg 16.78 ± 0.30 (235) 19.22 ± 0.21 (485) <0.0001 70d weight nursery, kg 16.41 ± 0.27 (182) 17.24 ± 0.47 (51) 0.13 140d weight on dam, kg 47.53 ± 0.60 (234) 50.04 ± 0.49 (477) 0.002 140d weight nursery, kg 40.35 ± 0.74 (171) 39.65 ± 1.10 (48) 0.61 Survival to wean on dam, % 63.2 ± 2.8 (411) 82.3 ± 2.5 (568) <0.0001 Survival to wean nursery, % 78.0 ± 6.1 (190) 88.0 ± 7.2 (53) 0.17 Survival to 140 d on dam, % 60.5 ± 2.8 (411) 80.6 ± 2.6 (568) <0.0001 Survival to 140 d nursery, % 76.0 ± 6.5 (190) 85.3 ± 7.6 (53) 0.22 Ewe lamb breeding weight, kg 41.0 ± 0.46 (176) 41.6 ± 0.45 (274) 0.38 Item . Least squares means (N for F1 lambs) for breed of ewe . Romanov . Rambouillet . Level of significance . Birth weight, kg 3.41 ± 0.08 (601) 4.26 ± 0.10 (621) <0.0001 56d weight on dam, kg 14.10 ± 0.24 (237) 16.36 ± 0.17 (488) <0.0001 56d weight nursery, kg 13.71 ± 0.21 (184) 14.57 ± 0.38 (54) 0.05 70d weight on dam, kg 16.78 ± 0.30 (235) 19.22 ± 0.21 (485) <0.0001 70d weight nursery, kg 16.41 ± 0.27 (182) 17.24 ± 0.47 (51) 0.13 140d weight on dam, kg 47.53 ± 0.60 (234) 50.04 ± 0.49 (477) 0.002 140d weight nursery, kg 40.35 ± 0.74 (171) 39.65 ± 1.10 (48) 0.61 Survival to wean on dam, % 63.2 ± 2.8 (411) 82.3 ± 2.5 (568) <0.0001 Survival to wean nursery, % 78.0 ± 6.1 (190) 88.0 ± 7.2 (53) 0.17 Survival to 140 d on dam, % 60.5 ± 2.8 (411) 80.6 ± 2.6 (568) <0.0001 Survival to 140 d nursery, % 76.0 ± 6.5 (190) 85.3 ± 7.6 (53) 0.22 Ewe lamb breeding weight, kg 41.0 ± 0.46 (176) 41.6 ± 0.45 (274) 0.38 Open in new tab Table 3. Levels of significance, least squares means, and standard errors for the effect of ewe breed producing F1 reciprocal cross lambs for traits recorded on purebred ewes . Ewe breed . Item1 . Romanov . Rambouillet . Level of significance . Conception rate, % 85.4 ± 2.7 81.6 ± 2.1 0.28 Number born 3.36 ± 0.06 1.84 ± 0.04 <0.0001 Litter birth weight, kg 10.27 ± 0.23 8.62 ± 0.18 <0.0001 Number weaned2 2.37 ± 0.07 1.60 ± 0.05 <0.0001 Litter 56 d wt2, kg 32.54 ± 0.93 24.76 ± 0.68 <0.0001 Litter 70 d wt2, kg 38.82 ± 1.11 29.07 ± 0.81 <0.0001 Litter 140 d wt2, kg 101.57 ± 3.27 75.08 ± 2.51 <0.0001 . Ewe breed . Item1 . Romanov . Rambouillet . Level of significance . Conception rate, % 85.4 ± 2.7 81.6 ± 2.1 0.28 Number born 3.36 ± 0.06 1.84 ± 0.04 <0.0001 Litter birth weight, kg 10.27 ± 0.23 8.62 ± 0.18 <0.0001 Number weaned2 2.37 ± 0.07 1.60 ± 0.05 <0.0001 Litter 56 d wt2, kg 32.54 ± 0.93 24.76 ± 0.68 <0.0001 Litter 70 d wt2, kg 38.82 ± 1.11 29.07 ± 0.81 <0.0001 Litter 140 d wt2, kg 101.57 ± 3.27 75.08 ± 2.51 <0.0001 1All traits are on a per ewe lambing basis except for conception rate. 2Calculated for genetic birth dam, combining both dam- and nursery-reared lambs. Open in new tab Table 3. Levels of significance, least squares means, and standard errors for the effect of ewe breed producing F1 reciprocal cross lambs for traits recorded on purebred ewes . Ewe breed . Item1 . Romanov . Rambouillet . Level of significance . Conception rate, % 85.4 ± 2.7 81.6 ± 2.1 0.28 Number born 3.36 ± 0.06 1.84 ± 0.04 <0.0001 Litter birth weight, kg 10.27 ± 0.23 8.62 ± 0.18 <0.0001 Number weaned2 2.37 ± 0.07 1.60 ± 0.05 <0.0001 Litter 56 d wt2, kg 32.54 ± 0.93 24.76 ± 0.68 <0.0001 Litter 70 d wt2, kg 38.82 ± 1.11 29.07 ± 0.81 <0.0001 Litter 140 d wt2, kg 101.57 ± 3.27 75.08 ± 2.51 <0.0001 . Ewe breed . Item1 . Romanov . Rambouillet . Level of significance . Conception rate, % 85.4 ± 2.7 81.6 ± 2.1 0.28 Number born 3.36 ± 0.06 1.84 ± 0.04 <0.0001 Litter birth weight, kg 10.27 ± 0.23 8.62 ± 0.18 <0.0001 Number weaned2 2.37 ± 0.07 1.60 ± 0.05 <0.0001 Litter 56 d wt2, kg 32.54 ± 0.93 24.76 ± 0.68 <0.0001 Litter 70 d wt2, kg 38.82 ± 1.11 29.07 ± 0.81 <0.0001 Litter 140 d wt2, kg 101.57 ± 3.27 75.08 ± 2.51 <0.0001 1All traits are on a per ewe lambing basis except for conception rate. 2Calculated for genetic birth dam, combining both dam- and nursery-reared lambs. Open in new tab Table 4. Levels of significance, number of lambs, least squares means and average standard errors of growth traits for the interaction effect of breed of ewe by type of birth on birth weight or rearing type on dam-reared only wean weight of phase I lambs . Least squares means for breed of ewe . Level of significance . Trait Romanov (N) Rambouillet (N) Birth weight, kg <0.0001  Single 4.26 ± 0.27 7 5.38 ± 0.08 87  Twin 3.36 ± 0.09 68 4.70 ± 0.05 446  Triplet 3.08 ± 0.07 156 4.04 ± 0.08 84  Quad and above 2.95 ± 0.05 370 2.90 ± 0.37 4 Adjusted 56 d wt on dam, kg 0.006  Single 14.91 ± 0.40 62 16.74 ± 0.24 134  Twin 13.28 ± 0.26 175 15.98 ± 0.26 354 . Least squares means for breed of ewe . Level of significance . Trait Romanov (N) Rambouillet (N) Birth weight, kg <0.0001  Single 4.26 ± 0.27 7 5.38 ± 0.08 87  Twin 3.36 ± 0.09 68 4.70 ± 0.05 446  Triplet 3.08 ± 0.07 156 4.04 ± 0.08 84  Quad and above 2.95 ± 0.05 370 2.90 ± 0.37 4 Adjusted 56 d wt on dam, kg 0.006  Single 14.91 ± 0.40 62 16.74 ± 0.24 134  Twin 13.28 ± 0.26 175 15.98 ± 0.26 354 Open in new tab Table 4. Levels of significance, number of lambs, least squares means and average standard errors of growth traits for the interaction effect of breed of ewe by type of birth on birth weight or rearing type on dam-reared only wean weight of phase I lambs . Least squares means for breed of ewe . Level of significance . Trait Romanov (N) Rambouillet (N) Birth weight, kg <0.0001  Single 4.26 ± 0.27 7 5.38 ± 0.08 87  Twin 3.36 ± 0.09 68 4.70 ± 0.05 446  Triplet 3.08 ± 0.07 156 4.04 ± 0.08 84  Quad and above 2.95 ± 0.05 370 2.90 ± 0.37 4 Adjusted 56 d wt on dam, kg 0.006  Single 14.91 ± 0.40 62 16.74 ± 0.24 134  Twin 13.28 ± 0.26 175 15.98 ± 0.26 354 . Least squares means for breed of ewe . Level of significance . Trait Romanov (N) Rambouillet (N) Birth weight, kg <0.0001  Single 4.26 ± 0.27 7 5.38 ± 0.08 87  Twin 3.36 ± 0.09 68 4.70 ± 0.05 446  Triplet 3.08 ± 0.07 156 4.04 ± 0.08 84  Quad and above 2.95 ± 0.05 370 2.90 ± 0.37 4 Adjusted 56 d wt on dam, kg 0.006  Single 14.91 ± 0.40 62 16.74 ± 0.24 134  Twin 13.28 ± 0.26 175 15.98 ± 0.26 354 Open in new tab Phase II of the experiment evaluated the reciprocal types of ewes over 4 yr of production when mated to either Suffolk or Composite rams under an annual production system. Models for traits recorded on the terminal-sired lambs included fixed effects of year (2006, 2007, 2008, 2009, and 2010), age of ewe (1, 2, 3, and 4), reciprocal ewe line (Romanov or Rambouillet as paternal parents), terminal sire line (Suffolk or Composite), sex of lamb (male, female), and type of birth (1, 2, or 3+) or rearing (1, 2). Two-way interactions among fixed effects that included reciprocal ewe line were also fitted (Table 5). The random effect of individual sire of each ewe within reciprocal ewe line was fitted to test effects associated with reciprocally produced ewe line. Models for traits recorded as traits of the ewe (Table 6) included fixed effects of contemporary mating groups (five groups over 4 yr), ewe age (1, 2, 3, and 4), reciprocal ewe line (Romanov or Rambouillet as paternal parents), terminal sire line (Suffolk or Composite), and the two-way interactions among fixed effects that included reciprocal ewe line. Models for cumulative traits calculated per ewe were similar but did not include the effects of contemporary groups or ewe age. Table 5. Levels of significance, least squares means, and standard errors for the interaction effect of ewe reciprocal line and terminal sire ram breed for growth and survival traits recorded on Phase II lambs. Trait . Dam breed of F1 terminal breed . Least squares means for parental breeds interaction1 . Avg. SEM . Level of significance . Rom × Suff . Rom × Comp . Ramb × Suff . Ramb × Comp . Dam line . Sire line . Interaction . Birth weight, kg 4.60 4.47 4.61 4.44 0.06 0.84 <0.0001 0.53 56 d weight on dam, kg 17.29 16.20 17.24 15.92 0.21 0.53 <0.0001 0.38 56 d weight nursery, kg 14.97 15.14 14.86 14.51 0.31 0.28 0.73 0.85 70 d weight on dam, kg 20.41 19.07 20.33 18.73 0.26 0.54 <0.0001 0.39 70 d weight nursery, kg 17.71 17.98 17.60 17.21 0.38 0.29 0.85 0.86 140 d weight on dam, kg 42.87 40.49 42.20 38.83 0.38 0.02 <0.0001 0.05 140 d weight nursery, kg 34.98 34.21 34.17 30.83 0.64 0.01 <0.0001 0.02 Survival to wean on dam, % 84.3 85.2 86.2 86.3 1.40 0.28 0.74 0.78 Survival to 140 d on dam, %2 97.8 96.7 96.7 97.0 0.80 0.68 0.61 0.34 Trait . Dam breed of F1 terminal breed . Least squares means for parental breeds interaction1 . Avg. SEM . Level of significance . Rom × Suff . Rom × Comp . Ramb × Suff . Ramb × Comp . Dam line . Sire line . Interaction . Birth weight, kg 4.60 4.47 4.61 4.44 0.06 0.84 <0.0001 0.53 56 d weight on dam, kg 17.29 16.20 17.24 15.92 0.21 0.53 <0.0001 0.38 56 d weight nursery, kg 14.97 15.14 14.86 14.51 0.31 0.28 0.73 0.85 70 d weight on dam, kg 20.41 19.07 20.33 18.73 0.26 0.54 <0.0001 0.39 70 d weight nursery, kg 17.71 17.98 17.60 17.21 0.38 0.29 0.85 0.86 140 d weight on dam, kg 42.87 40.49 42.20 38.83 0.38 0.02 <0.0001 0.05 140 d weight nursery, kg 34.98 34.21 34.17 30.83 0.64 0.01 <0.0001 0.02 Survival to wean on dam, % 84.3 85.2 86.2 86.3 1.40 0.28 0.74 0.78 Survival to 140 d on dam, %2 97.8 96.7 96.7 97.0 0.80 0.68 0.61 0.34 1Two reciprocal F1 ewe types crossed with two terminal sire breeds. Rom = Romanov dams producing F1 ewes. Ramb = Rambouillet dams producing F1 ewes. Suff = Suffolk ram breed. Comp = Composite ram breed. 2Postweaning survival of those that survived to weaning. Open in new tab Table 5. Levels of significance, least squares means, and standard errors for the interaction effect of ewe reciprocal line and terminal sire ram breed for growth and survival traits recorded on Phase II lambs. Trait . Dam breed of F1 terminal breed . Least squares means for parental breeds interaction1 . Avg. SEM . Level of significance . Rom × Suff . Rom × Comp . Ramb × Suff . Ramb × Comp . Dam line . Sire line . Interaction . Birth weight, kg 4.60 4.47 4.61 4.44 0.06 0.84 <0.0001 0.53 56 d weight on dam, kg 17.29 16.20 17.24 15.92 0.21 0.53 <0.0001 0.38 56 d weight nursery, kg 14.97 15.14 14.86 14.51 0.31 0.28 0.73 0.85 70 d weight on dam, kg 20.41 19.07 20.33 18.73 0.26 0.54 <0.0001 0.39 70 d weight nursery, kg 17.71 17.98 17.60 17.21 0.38 0.29 0.85 0.86 140 d weight on dam, kg 42.87 40.49 42.20 38.83 0.38 0.02 <0.0001 0.05 140 d weight nursery, kg 34.98 34.21 34.17 30.83 0.64 0.01 <0.0001 0.02 Survival to wean on dam, % 84.3 85.2 86.2 86.3 1.40 0.28 0.74 0.78 Survival to 140 d on dam, %2 97.8 96.7 96.7 97.0 0.80 0.68 0.61 0.34 Trait . Dam breed of F1 terminal breed . Least squares means for parental breeds interaction1 . Avg. SEM . Level of significance . Rom × Suff . Rom × Comp . Ramb × Suff . Ramb × Comp . Dam line . Sire line . Interaction . Birth weight, kg 4.60 4.47 4.61 4.44 0.06 0.84 <0.0001 0.53 56 d weight on dam, kg 17.29 16.20 17.24 15.92 0.21 0.53 <0.0001 0.38 56 d weight nursery, kg 14.97 15.14 14.86 14.51 0.31 0.28 0.73 0.85 70 d weight on dam, kg 20.41 19.07 20.33 18.73 0.26 0.54 <0.0001 0.39 70 d weight nursery, kg 17.71 17.98 17.60 17.21 0.38 0.29 0.85 0.86 140 d weight on dam, kg 42.87 40.49 42.20 38.83 0.38 0.02 <0.0001 0.05 140 d weight nursery, kg 34.98 34.21 34.17 30.83 0.64 0.01 <0.0001 0.02 Survival to wean on dam, % 84.3 85.2 86.2 86.3 1.40 0.28 0.74 0.78 Survival to 140 d on dam, %2 97.8 96.7 96.7 97.0 0.80 0.68 0.61 0.34 1Two reciprocal F1 ewe types crossed with two terminal sire breeds. Rom = Romanov dams producing F1 ewes. Ramb = Rambouillet dams producing F1 ewes. Suff = Suffolk ram breed. Comp = Composite ram breed. 2Postweaning survival of those that survived to weaning. Open in new tab Table 6. Levels of significance, least squares means, and standard errors for the interaction effect of ewe reciprocal line and terminal sire ram breed for reproductive traits recorded on reciprocal F1 ewes Trait . Dam breed of F1 terminal breed . Least squares means for parental breeds interaction 1 . Avg. SEM . Level of significance . Rom × Suff . Rom × Comp . Ramb × Suff . Ramb × Comp . Dam line . Sire line . Interaction . Conception rate3, % 95.3 94.9 93.9 95.8 1.40 0.89 0.49 0.30 Number born2 2.19 2.27 2.15 2.24 0.04 0.40 0.01 0.98 Litter birth wt2, kg 10.09 10.05 9.88 9.78 0.16 0.21 0.60 0.80 Number weaned2 1.90 1.99 1.91 1.99 0.04 0.98 0.02 0.82 Litter 56 d wt2, kg 31.62 30.88 31.64 29.96 0.65 0.54 0.03 0.38 Litter 70 d wt2, kg 35.90 35.60 36.08 34.64 0.84 0.69 0.20 0.41 Litter 140 d wt2, kg 74.73 73.95 72.71 69.71 1.70 0.12 0.19 0.44 Ewe longevity to 50 mo3, % 72.8 78.4 80.2 82.8 4.40 0.24 0.29 0.69 Cumulative number born3 7.32 7.83 7.35 7.83 0.33 0.97 0.07 0.95 Cumulative number weaned3 6.35 6.89 6.51 6.95 0.30 0.76 0.06 0.85 Cumulative litter birth wt3, kg 33.61 34.39 33.62 34.18 1.44 0.90 0.52 0.86 Cumulative litter 140 d wt3, kg 247.2 251.9 246.0 241.2 11.5 0.65 0.99 0.63 Trait . Dam breed of F1 terminal breed . Least squares means for parental breeds interaction 1 . Avg. SEM . Level of significance . Rom × Suff . Rom × Comp . Ramb × Suff . Ramb × Comp . Dam line . Sire line . Interaction . Conception rate3, % 95.3 94.9 93.9 95.8 1.40 0.89 0.49 0.30 Number born2 2.19 2.27 2.15 2.24 0.04 0.40 0.01 0.98 Litter birth wt2, kg 10.09 10.05 9.88 9.78 0.16 0.21 0.60 0.80 Number weaned2 1.90 1.99 1.91 1.99 0.04 0.98 0.02 0.82 Litter 56 d wt2, kg 31.62 30.88 31.64 29.96 0.65 0.54 0.03 0.38 Litter 70 d wt2, kg 35.90 35.60 36.08 34.64 0.84 0.69 0.20 0.41 Litter 140 d wt2, kg 74.73 73.95 72.71 69.71 1.70 0.12 0.19 0.44 Ewe longevity to 50 mo3, % 72.8 78.4 80.2 82.8 4.40 0.24 0.29 0.69 Cumulative number born3 7.32 7.83 7.35 7.83 0.33 0.97 0.07 0.95 Cumulative number weaned3 6.35 6.89 6.51 6.95 0.30 0.76 0.06 0.85 Cumulative litter birth wt3, kg 33.61 34.39 33.62 34.18 1.44 0.90 0.52 0.86 Cumulative litter 140 d wt3, kg 247.2 251.9 246.0 241.2 11.5 0.65 0.99 0.63 1Two reciprocal F1 ewe types crossed with two terminal sire breeds. Rom = Romanov dams producing F1 ewes. Ramb = Rambouillet dams producing F1 ewes. Suff = Suffolk ram breed. Comp = Composite ram breed. 2Trait is on a per ewe lambing basis. 3Trait is on a per ewe exposed basis Open in new tab Table 6. Levels of significance, least squares means, and standard errors for the interaction effect of ewe reciprocal line and terminal sire ram breed for reproductive traits recorded on reciprocal F1 ewes Trait . Dam breed of F1 terminal breed . Least squares means for parental breeds interaction 1 . Avg. SEM . Level of significance . Rom × Suff . Rom × Comp . Ramb × Suff . Ramb × Comp . Dam line . Sire line . Interaction . Conception rate3, % 95.3 94.9 93.9 95.8 1.40 0.89 0.49 0.30 Number born2 2.19 2.27 2.15 2.24 0.04 0.40 0.01 0.98 Litter birth wt2, kg 10.09 10.05 9.88 9.78 0.16 0.21 0.60 0.80 Number weaned2 1.90 1.99 1.91 1.99 0.04 0.98 0.02 0.82 Litter 56 d wt2, kg 31.62 30.88 31.64 29.96 0.65 0.54 0.03 0.38 Litter 70 d wt2, kg 35.90 35.60 36.08 34.64 0.84 0.69 0.20 0.41 Litter 140 d wt2, kg 74.73 73.95 72.71 69.71 1.70 0.12 0.19 0.44 Ewe longevity to 50 mo3, % 72.8 78.4 80.2 82.8 4.40 0.24 0.29 0.69 Cumulative number born3 7.32 7.83 7.35 7.83 0.33 0.97 0.07 0.95 Cumulative number weaned3 6.35 6.89 6.51 6.95 0.30 0.76 0.06 0.85 Cumulative litter birth wt3, kg 33.61 34.39 33.62 34.18 1.44 0.90 0.52 0.86 Cumulative litter 140 d wt3, kg 247.2 251.9 246.0 241.2 11.5 0.65 0.99 0.63 Trait . Dam breed of F1 terminal breed . Least squares means for parental breeds interaction 1 . Avg. SEM . Level of significance . Rom × Suff . Rom × Comp . Ramb × Suff . Ramb × Comp . Dam line . Sire line . Interaction . Conception rate3, % 95.3 94.9 93.9 95.8 1.40 0.89 0.49 0.30 Number born2 2.19 2.27 2.15 2.24 0.04 0.40 0.01 0.98 Litter birth wt2, kg 10.09 10.05 9.88 9.78 0.16 0.21 0.60 0.80 Number weaned2 1.90 1.99 1.91 1.99 0.04 0.98 0.02 0.82 Litter 56 d wt2, kg 31.62 30.88 31.64 29.96 0.65 0.54 0.03 0.38 Litter 70 d wt2, kg 35.90 35.60 36.08 34.64 0.84 0.69 0.20 0.41 Litter 140 d wt2, kg 74.73 73.95 72.71 69.71 1.70 0.12 0.19 0.44 Ewe longevity to 50 mo3, % 72.8 78.4 80.2 82.8 4.40 0.24 0.29 0.69 Cumulative number born3 7.32 7.83 7.35 7.83 0.33 0.97 0.07 0.95 Cumulative number weaned3 6.35 6.89 6.51 6.95 0.30 0.76 0.06 0.85 Cumulative litter birth wt3, kg 33.61 34.39 33.62 34.18 1.44 0.90 0.52 0.86 Cumulative litter 140 d wt3, kg 247.2 251.9 246.0 241.2 11.5 0.65 0.99 0.63 1Two reciprocal F1 ewe types crossed with two terminal sire breeds. Rom = Romanov dams producing F1 ewes. Ramb = Rambouillet dams producing F1 ewes. Suff = Suffolk ram breed. Comp = Composite ram breed. 2Trait is on a per ewe lambing basis. 3Trait is on a per ewe exposed basis Open in new tab RESULTS AND DISCUSSION General Results Levels of significance and least squares means are reported for effects of ewe breed producing reciprocal cross lambs, and both ewe and sire breed for the production of terminal-sired progeny from the evaluation of reproduction in the reciprocal cross ewes. Estimated differences between these ewe breeds can be the result of direct additive genetic effects as well as any differences due to maternal effects, imprinted and transgenerational effects, and differences in mitochondrial origin. In particular, it was expected that we could observe maternal genetic effects associated with differences in litter size. Effects of year and its interaction with ewe breed and sire breed were fit in the model but are not discussed because conditions contributing to these effects could not be identified, the effects cannot be predicted to recur in the future, and it is likely that decisions on genetic choices by breeders will do so with average year effects in mind. Genotype (breeds) and environmental interactions (year) did not exhibit consistent trends and were outside of the primary objectives of the study. Age of ewe effects was consistently detected as highly significant for most ewe traits analyzed (the exception being conception rate) and results were consistent with the well-known influences on ewe productivity traits where ewe lamb performance is typically less than in the subsequent two to three parities and did not interact with the main effects of ewe breed or sire breed. Therefore, results associated with ewe age were not tabulated or further discussed; however, it is notable that differences between ewe breeds were not detected even at the first parity where one might anticipate the impact of this contrast to be most pronounced. Phase I: Production of Reciprocal Crosses of F1 Lambs Effects of ewe breed producing F1 reciprocal cross lambs are presented for traits recorded on a per lamb basis (Table 2) and for traits recorded on a per ewe basis (Table 3). The Romanov breed originated in northwestern Russia and excels in adaptability, length of breeding season, age at puberty, prenatal and postnatal survival, maternal behavior, and ewe productivity (Fahmy, 1996; Freking et al., 2000). Although the reproductive rate advantages seem to be associated with many genes each with small effects rather than genetic differences at few loci with larger effects, recent evidence of a mutation in the Romanov breed (Heaton et al., 2017) for the same gene (BMPR-1B) containing the Booroola allele is worthy of future investigation. The Rambouillet breed is one of the most predominant wool breeds used throughout the extensively managed areas such as Texas and western range of the United States with a lower reproductive rate, although they can responded to selection applied for reproduction traits (Burfening et al., 1993). Over the 2 yr period of matings, mature Romanov ewes (n = 116) produced 601 lambs and mature Rambouillet ewes (n = 243) produced 621 lambs for evaluation of growth and survival traits. This result highlights the contrast of these breeds that would be informative for producers trying to make breeding system decisions. Despite no detectable difference in conception rate (P = 0.28; Table 3), twice as many Rambouillet ewes compared to Romanov ewes were required at the start of the experiment to produce similar numbers of contemporary reciprocal crossbred progeny. These higher litter sizes were associated with differences in survival rate of reciprocal crossbred lambs. Differences in ewe breeds influenced (P < 0.001) survival of dam-reared progeny by showing an increase of 20% from Rambouillet dams at weaning and 140 d compared to progeny from Romanov dams. Survival rates of nursery-reared progeny were similar between the two ewe breeds (P > 0.17). Subsequent BW measures on dam-reared lambs were significantly higher (P < 0.001) at 56, 70, and 140 d for Rambouillet. Litter weight measured out to d 140 indicated an advantage of >25 kg for Romanov ewes on a per ewe lambing basis. However, on an individual lamb basis no differences were detected for ewe breed between reciprocal F1 ewe types from the recorded first breeding weight at ~7 mo. The impact of ewe breed was much less for nursery-reared lambs and was not significant (P = 0.61) by the time lambs reached 140 d. The dramatically different maternal environments resulted in detectable interactions of breed of ewe with type of birth for birth weight and with type of rearing for weaning weight (Table 4). The higher litter sizes in the Romanov litters restricted individual fetal size compared to similar genetic fetuses gestated in smaller litter sizes in Rambouillet uterine environments. As indicated from the larger standard errors, very few singles were born to Romanov ewes and very few triplets, and a single set of quads were born to Rambouillet ewes. Single, twin, and triplet F1 lambs were lighter when gestated in Romanov ewes compared to Rambouillet. There was a 1.6 kg difference between singles and twins reared by Romanov dams as opposed to only a 0.76 kg difference between singles and twins reared by Rambouillet ewes. Phase II: Production of Terminally Sired Lambs by Reciprocal Crosses of F1 Ewes A total of 433 reciprocal cross F1 ewes produced 3,431 lambs (1,552 L) from 1,634 exposures to terminal sires over 4 yr. Terminal sires consisted of 38 Suffolk and 44 Composite rams. Effects of reciprocal cross ewe breed and terminal sire breed are presented for traits recorded on a per lamb basis (Table 5) and for traits recorded on a per ewe basis (Table 6). Considering only the main effect of dam line, birth weight of terminal sired lambs was similar (P = 0.84) from the two reciprocal cross ewe breeds and BW after birth revealed few detectable differences from weaning through 140 d for either dam-reared or nursery-reared progeny. The presence of an interaction of dam line with sire line for 140 d weight is discussed later in the manuscript. Reciprocal cross ewe breed also did not influence (P = 0.28) survival rate of terminal-sired progeny. Terminal sire breed effects were evident (P < 0.0001) at birth with lambs born from Suffolk rams exceeding those from Composite rams by 0.15 kg. Suffolk-sired lambs continued to grow at a faster rate than Composite-sired lambs and were heavier by 1.2, 1.5, and 2.8 kg at 56-, 70-, and 140-d, respectively. The two terminal sire breeds were similar for survival rates to weaning (P = 0.74) and 140 d (P = 0.61). Survival rates were high in this phase of the experiment highlighting the importance of maximizing both maternal and individual lamb heterosis. These higher survival rates in this phase of the experiment can also be attributed in part to less variable litter sizes in the F1 ewes, especially compared to the variability in the purebred Romanov. Leeds et al. (2012) reported results from a terminal sire study that included the USMARC Composite and Suffolk as sire breeds mated to Rambouillet ewes under extensive rangeland systems and did not detect sire breed differences in ewe fertility or productivity but approached significance for crossbred lamb survival to weaning. In that study, an interaction was detected between sire breeds and birth weight, indicating that lightweight Suffolk-sired lambs had a greater risk of death than lightweight Composite-sired lambs. Similar to the current study, Suffolk-sired progeny in the rangeland study had greater pre- and postweaning growth than Composite-sired progeny (Notter et al., 2012). Minimal differences were observed in performance of reciprocal cross ewes through their age 4 yr seasons for productivity, longevity, or progeny growth and survival. One exception was BW at 140 d where an interaction of dam breed with terminal sire breed reached significance for both dam-reared (P = 0.05) and nursery-reared (P = 0.02) lambs. In both cases, the interaction was due to the lower than average weight of Composite-sired lambs out of reciprocal cross ewes born from Romanov sires and Rambouillet dams. Composite terminals increased number born (P = 0.01) and number weaned (P = 0.02) leading to greater litter weight at 56 d (P < 0.05) from the reciprocal cross ewes. Although the interaction of dam breed with terminal sire breed did not reach significance for litter birth weight (P = 0.80), one could speculate that the apparent improved embryonic survival rates could be due to less extreme embryonic growth from the Composite sires. Suffolk rams increased (P < 0.001) BW and growth rates from birth to 140 d of terminal progeny. Comparing the relative differences between the two terminal-sire breeds, Composite sires increased numbers of live terminal progeny while Suffolk sires increased the growth rates. Thus, there were very little cumulative differences accrued over the 4 yr and no differences were detected for cumulative kilogram of lamb generated at 140 d per ewe exposed. Production systems utilizing terminal sire breeds can affect ewe productivity of maternal lines by increasing fitness traits of crossbred progeny as well as growth and carcass traits. These results were consistent with previous research comparing these two breeds as terminal sires, where similar growth rates but improved fitness traits were observed for Composite-sired lambs (Leymaster, 1991). We did not detect any large differences in reproductive traits between reciprocal crosses of these two maternal breeds, despite the dramatic differences in litter sizes at birth they were derived from. Limited amounts of evidence in the literature from other experiments conducted in sheep would also suggest that these differences are limited. Shrestha et al. (1983) were unable to detect any reciprocal effects on reproductive traits from five breeds. Contributions to reciprocal differences theoretically could be due to impacts of genes with specific maternally imprinted expression patterns. However, experimental evidence to document the effect of maternally imprinted genes is difficult to distinguish from maternal environment effects (Goddard and Whitelaw, 2014). The practical outcome of this specific reciprocal evaluation is that performance levels of both types of Romanov crossbred ewes will be similar allowing the industry to produce the desired crossbred ewes without needing large purebred ewe flocks of the less numerous Romanov breed. ACKNOWLEDGMENTS The authors acknowledge Kreg A. Leymaster (retired) who provided the primary leadership for conceiving, designing, and conducting this experiment. The authors also acknowledge Stephanie Schmidt for assistance with manuscript preparation and the USMARC sheep operations for animal husbandry. Conflict of interest statement. None declared. Footnotes 1 Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture. The U.S. Department of Agriculture (USDA) prohibits discrimination in all its programs and activities on the basis of race, color, national origin, age, disability, and where applicable, sex, marital status, familial status, parental status, religion, sexual orientation, genetic information, political beliefs, reprisal, or because all or part of an individual’s income is derived from any public assistance program. (Not all prohibited bases apply to all programs.) Persons with disabilities who require alternative means for communication of program information (Braille, large print, audiotape, etc.) should contact USDA’s TARGET Center at (202) 720-2600 (voice and TDD). To file a complaint of discrimination, write to USDA, Director, Office of Civil Rights, 1400 Independence Avenue, S.W., Washington, DC 20250–9410, or call (800) 795-3272 (voice) or (202) 720-6382 (TDD). USDA is an equal opportunity provider and employer. LITERATURE CITED American Sheep Industry Association, Inc . 2016 . U.S. Sheep industry research, development, and education priorities. ASI website . http://www.sheepusa.org/Resources_Publications_ResearchEducationPriorities2016 (Accessed 28 September 2018). Bradford , G. E . 1972 . The role of maternal effects in animal breeding. VII. Maternal effects in sheep . J. Anim. Sci . 35 : 1324 – 1334 . doi:10.2527/jas1972.3561324x Google Scholar Crossref Search ADS PubMed WorldCat Burfening , P. J. , S. D. Kachman, K. J. Hanford, and D. Rossi. 1993 . Selection for reproductive rate in Rambouillet sheep: estimated genetic change in reproductive rate . Small Ruminant Research . 10 : 317 – 330 . doi: 10.1016/0921-4488(93)90136-6 Google Scholar Crossref Search ADS WorldCat Casas , E. , B. A. Freking, and K. A. Leymaster. 2004 . Evaluation of dorset, finnsheep, romanov, texel, and montadale breeds of sheep: II. Reproduction of F1 ewes in fall mating seasons . J. Anim. Sci . 82 : 1280 – 1289 . doi: 10.2527/2004.8251280x Google Scholar Crossref Search ADS PubMed WorldCat Casas , E. , B. A. Freking, and K. A. Leymaster. 2005 . Evaluation of dorset, finnsheep, romanov, texel, and montadale breeds of sheep: V. Reproduction of F1 ewes in spring mating seasons . J. Anim. Sci . 83 : 2743 – 2751 . doi: 10.2527/2005.83122743x Google Scholar Crossref Search ADS PubMed WorldCat FASS . 1999 . Guide for the care and use of agricultural animals in agricultural research and teaching . 1st rev. ed. Savoy, IL : Federation Animal Science Societies . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Freking , B. A. , and K. A. Leymaster. 2004 . Evaluation of dorset, finnsheep, romanov, texel, and montadale breeds of sheep: IV. Survival, growth, and carcass traits of F1 lambs . J. Anim. Sci . 82 : 3144 – 3153 . doi: 10.2527/2004.82113144x Google Scholar Crossref Search ADS PubMed WorldCat Freking , B. A. , K. A. Leymaster, and L. D. Young. 2000 . Evaluation of dorset, finnsheep, romanov, texel, and montadale breeds of sheep: I. Effects of ram breed on productivity of ewes of two crossbred populations . J. Anim. Sci . 78 : 1422 – 1429 . doi:10.2527/2000.7861422x Google Scholar Crossref Search ADS PubMed WorldCat Goddard , M. E. , and E. Whitelaw. 2014 . The use of epigenetic phenomena for the improvement of sheep and cattle . Front. Genet . 5 : 247 . doi: 10.3389/fgene.2014.00247 Google Scholar Crossref Search ADS PubMed WorldCat Heaton , M. P. , T. P. L. Smith, B. A. Freking, A. M. Workman, G. L. Bennett, J. K. Carnahan, and T. S. Kalbfleisch. 2017 . Using sheep genomes from diverse U.S. Breeds to identify missense variants in genes affecting fecundity . F1000Res . 6 : 1303 . doi: 10.12688/f1000research.12216.1 Google Scholar Crossref Search ADS PubMed WorldCat Leeds , T. D. , D. R. Notter, K. A. Leymaster, M. R. Mousel, and G. S. Lewis. 2012 . Evaluation of columbia, USMARC-composite, Suffolk, and texel rams as terminal sires in an extensive rangeland production system: I. Ewe productivity and crossbred lamb survival and preweaning growth . J. Anim. Sci . 90 : 2931 – 2940 . doi: 10.2527/jas.2011-4640 Google Scholar Crossref Search ADS PubMed WorldCat Leymaster , K. A . 1991 . Straightbred comparison of a composite population and the Suffolk breed for performance traits of sheep . J. Anim. Sci . 69 : 993 – 999 . doi:10.2527/1991.693993x Google Scholar Crossref Search ADS PubMed WorldCat Notter , D. R. , T. D. Leeds, M. R. Mousel, J. B. Taylor, D. P. Kirschten, and G. S. Lewis. 2012 . Evaluation of columbia, USMARC-Composite, Suffolk, and texel rams as terminal sires in an extensive rangeland production system: II. Postweaning growth and ultrasonic measures of composition for lambs fed a high-energy feedlot diet . J. Anim. Sci . 90 : 2941 – 2952 . doi: 10.2527/jas.2011-4641 Google Scholar Crossref Search ADS PubMed WorldCat Prolific Sheep . 1996 . The Romanov. In: M. H. Fahmy, ed. Prolific Sheep. pp. 47−72. Wallingford, U.K: CAB International. Shrestha , J. N. B. , W. E. Rempel, W. J. Boylan, and K. P. Miller. 1983 . General, specific, maternal and reciprocal effects for ewe productivity in crossing five breeds of sheep . Can. J. Anim. Sci . 63 : 497 – 509 . doi:10.4141/cjas83-059 Google Scholar Crossref Search ADS WorldCat Wang , C. T. and G. E. Dickerson. 1991a . A deterministic computer simulation model of life -cycle lamb and wool production . J. Anim. Sci . 69 : 4312 – 4323 . doi:10.2527/1991.69114312x Google Scholar Crossref Search ADS WorldCat Wang , C. T. , and G. E. Dickerson. 1991b . Simulated effects of reproductive performance on life-cycle efficiency of lamb and wool production at three lambing intervals . J. Anim. Sci . 69 : 4338 – 4347 . doi:10.2527/1991.69114338x Google Scholar Crossref Search ADS WorldCat Wang , C. T. and G. E. Dickerson. 1991c . Simulation of life-cycle efficiency of lamb and wool production for genetic levels of component traits and alternative management options . J. Anim. Sci . 69 : 4324 – 4337 . doi:10.2527/1991.69114324x Google Scholar Crossref Search ADS WorldCat Published by Oxford University Press on behalf of the American Society of Animal Science 2018. This work is written by (a) US Government employee(s) and is in the public domain in the US. Published by Oxford University Press on behalf of the American Society of Animal Science 2018.

Gong,, Hua;Zhou,, Huitong;Bai,, Lingrong;Li,, Wenhao;Li,, Shaobin;Wang,, Jiqing;Luo,, Yuzhu;Hickford, Jon G, H

doi: 10.1093/jas/sky465pmid: 30535023

Abstract The keratin-associated proteins (KAPs) are important constituents of wool fibers. Of the many mammalian KAP genes (KRTAPs) identified, KRTAP20-1 has been described in humans, but it has not been described in any other species. A search of the sheep genome using the human KRTAP20-1 sequence revealed a homologous open reading frame on chromosome 1, which would encode a high glycine-tyrosine KAP. PCR-single-stranded conformational polymorphism (PCR-SSCP) analysis identified 8 different banding patterns representing 8 unique DNA sequences (named A to H). The sequences had highest similarity to the human KRTAP20-1 sequence, and this suggests that they are variants of ovine KRTAP20-1. Among these variants, a 12-bp insertion/deletion and 6 single nucleotide poly- morphisms (SNPs), including one 5′ untranslated region (UTR) SNP, one 3′ UTR SNP, and 2 nonsynonymous SNPs, were detected. Variant A was found to be associated with a decrease in mean fiber diameter, fiber diameter standard deviation, and prickle factor, whereas variant C was associated with increased greasy fleece weight and decreased wool yield. These associations persisted after adjusting for the effect of 2 nearby KRTAPs (KRTAP6-3 and KRTAP22-1) that have also been reported to associate with these wool traits. This suggests that variation in KRTAP20-1 affects wool yield and mean fiber diameter–associated traits, and that this effect is unlikely to be the result of the clustering of these KRTAPs on chromosome 1. INTRODUCTION Sheep are raised principally for their meat, milk, and wool. Wool production is particularly important in fine wool breeds like the Merino, and the value of the wool produced by a sheep depends on wool characteristics, with fiber diameter and fleece weight being the most economically important characteristics (Holman and Malau-Aduli, 2012). Wool is a natural fiber and principally made of keratins and keratin- associated proteins (KAPs). Keratins are assembled into keratin intermediate filaments (KIF), whereas KAPs form a matrix that cross-links these KIFs. The KAPs are therefore believed to play an important role in determining the characteristics of the wool fiber. The KAPs are either rich in cysteine or have a high content of glycine and tyrosine. They can be categorized into 3 broad groups: high sulphur KAPs (HS; ≤30 mol% cysteine), ultra-high sulphur KAPs (UHS; >30 mol% cysteine), and high glycine-tyrosine KAPs (HGT; 35 to 60 mol% glycine and tyrosine). The KAPs can further be placed into families based on their sequence similarities, and in humans, 25 families (KPA1−KAP13, KAP15−KAP17, and KAP19−KAP27) have been described (Rogers and Schweizer, 2005; Rogers et al., 2007; Rogers et al., 2008). The KAPs are encoded by small intronless genes called KRTAPs. There are 88 functional KRTAPs described in humans (Rogers and Schweizer, 2005; Rogers et al., 2007; Rogers et al., 2008), but to date, only 33 KRTAPs have been reported in sheep (Gong et al., 2016; Li et al., 2017a, 2017b; Wang et al., 2017b; Bai et al., 2018), a species for which research in KAPs was pre- viously focused, due to the economic importance of wool. Despite the near accomplishment of a complete sheep genome sequence, many of the human KRTAP homologs remain to be identified and their effect on wool characteristics remains under investigation. KAP20 is a member of the HGT-KAP fami- ly, and in humans this KAP family contains 2 proteins, KAP20-1 and KAP20-2 encoded by KRTAP20-1 and KRTAP20-2, respectively (Rogers and Schweizer, 2005). KRTAP20-2 has been recently identified in sheep and goats (Wang et al., 2017a; Bai et al., 2018), and a nonsense mutation in the ovine KRTAP20-2 has been shown to affect wool fiber curvature (Bai et al., 2018), whereas variation in caprine KRTAP20-2 affects cashmere fiber weight and length (Wang et al., 2017a). KRTAP20-1 has not been reported in any other species. In this study, we describe the identification of ovine KRTAP20-1, reveal variation in this gene, and investigate associations between this genetic variation and variation in wool traits. MATERIALS AND METHODS All research involving animals were carried out in accordance with the Animal Welfare Act 1999 (New Zealand Government), and the collection of sheep blood drops by the nicking of their ears is covered by Section 7.5 Animal Identification, in Code of Welfare: Sheep and Beef Cattle (2016), a code of welfare issued under the Animal Welfare Act 1999 (New Zealand Government). Sheep Blood and Wool Samples A total of 579 sheep were used to search for variation in KRTAP20-1. These included 89 New Zealand (NZ) Romney sheep (sourced from 5 farms), 92 Merino sheep (sourced from 5 farms), and 398 Southdown × Merino-cross sheep (sourced from the same farm, but from 6 sire-lines). Among these, only the 398 Southdown × Merino-cross sheep had wool data collected, and hence association study was only carried out on the Southdown × Merino-cross sheep. All the lambs were ear-tagged with unique identification number within 12 h of birth, and their birth dates, birth weights, birth ranks (i.e., whether they were a single, twin, or triplet), gender, and dam identity were recorded. All sheep were managed as a single mob on the same farm and were shorn at 12 mo of age. At shearing, greasy fleece weight (GFW) was measured, and a wool sample was collected from the mid-side region of each sheep for wool trait measurement at the New Zealand Wool Testing Authority Ltd. (NZWTA, Napier, New Zealand) using International Wool Testing Organization (IWTO) standardized methods. This included measurement of wool yield (Yield), mean staple length (MSL), mean staple strength (MSS), mean fiber diameter (MFD), fiber diameter standard deviation (FDSD), coefficient of variation of fiber diameter (CVFD), mean fiber curvature (MFC), and prickle factor (PF; the percentage of fibers of diameter greater than 30 microns). Clean fleece weight (CFW) was calculated from the GFW and Yield. A blood sample from each sheep was collected onto TFN paper (Munktell Filter AB, Sweden) and genomic DNA was purified using a two-step washing procedure as described in Zhou et al. (2006). Search for the Human KRTAP20-1 Homolog in the Sheep Genome The coding sequence of a human KRTAP20-1 sequence (GenBank accession NM_181615) was used to BLAST search the Ovine Genome Assembly v4.0. A genome sequence that showed the highest homo- logy with the human KRTAP20-1 coding sequence was presumed to be the notional ovine KRTAP20-1, and the sequences flanking this segment were used to design PCR primers for amplifying the notional gene. PCR Amplification of Ovine KRTAP20-1 Two PCR primers (5′-TCATATTCTGCAA GCAAAGGC-3′and 5′-GCTGATGGGTCTCAG TCAC-3′) were desi gned to amplify a fragment of notional length of 290 bp from the putative KRTAP20-1, based on the sequence information obtained above. These primers were synthesized by Integrated DNA Technologies (Coralville, IA). PCR amplification was performed in a 15-μL reaction containing the genomic DNA on one 1.2-mm punch of TFN paper, 0.25 μM of each primer, 150 μM of each dNTP (Eppendorf, Hamburg, Germany), 2.5 mM of Mg2+, 0.5 U of Taq DNA polymerase (Qiagen, Hilden, Germany), and 1× reaction buffer supplied with the enzyme. The thermal profile consisted of an initial denaturation for 2 min at 94 °C, followed by 35 cycles of 30 s at 94 °C, 30 s at 60 °C, and 30 s at 72 °C, and with a final extension of 5 min at 72 °C. Amplification was carried out in S1000 thermal cyclers (Bio-Rad, Hercules, CA). Screening for Sequence Variation, Variant Sequencing, and Sequence Analyses PCR amplicons of ovine KRTAP20-1 were subject to single-stranded conformational polymorphism (SSCP) analysis to screen for potential variation in the gene. A 0.7-μL aliquot of each amplicon was mixed with 7 μL of loading dye (98% formamide, 10-mM EDTA, 0.025% bromophenol blue, and 0.025% xylene-cyanol). After denatu- ration at 95 °C for 5 min, samples were rapidly cooled on wet ice and then loaded on 16 × 18 cm, 14% acrylamide:bisacrylamide (37.5:1; Bio-Rad) gels with addition of 1% glycerol. Electrophoresis was performed using Protean II xi cells (Bio-Rad), at 390 V for 18 h at 8 °C in 0.5 × TBE buffer. The gels were silver-stained according to the method of Byun et al. (2009). PCR amplicons representative of different SSCP patterns from sheep that appeared to be homozygous for KRTAP20-1 were sequenced at the Lincoln University DNA Sequencing Facility. For those variants that were only found in heterozygous sheep, they were sequenced using a rapid approach described previously (Gong et al., 2011a). In this approach, a band corresponding to the allele was excised as a gel slice from the polyacrylamide gel, macerated, and then used as a template for reamplification with the original primers. This second amplicon was then sequenced. Sequence alignments, translations, and comparisons were carried out using DNAMAN (version 5.2.10, Lynnon BioSoft, Vaudreuil, Canada). A phylogenetic tree was constructed based on the predicted amino acid sequence for the coding regions using MEGA version 7.0. The BLAST algorithm was used to search the NCBI GenBank (www.ncbi.nlm.nih.gov/) databases for homologous sequences. Genotyping of KRTAP6-3 and KRTAP22-1 The 319 Southdown × Merino-cross sheep used for association study were also genotyped for KRTAP6-3 and KRTAP22-1 using a PCR-SSCP technique described previously for these 2 genes (Li et al., 2017b, 2017c). Briefly, KRTAP6-3 was amplified using the PCR pri- mers 5′-CCGAGAACAACCTCAACTAC-3′ and 5′-GTAGAGGATGAGAGTCTTTCT-3′ that would notionally produce a fragment of 236 or 281 bp in length [KRTAP6-3 has been described to have length variation (Li et al., 2017b)], whereas KRTAP22-1 was amplified using the primers 5′-CCGAGAACAACCTCAACTAC-3′ and 5′-GTAGAGGATGAGAGTCTTTCT-3′, which would notionally produce a fragment of 305 bp in length. After denaturation, the KRTAP6-3 amplicons were electrophoresed using 12% acrylamide: bisacrylamide (37.5:1; Bio-Rad) gels containing 3.5% vol/vol glycerol, at 17 °C, 350 V for 18 h, whereas amplicons of KRTAP22-1 were electrophoresed using 14% acrylamide: bisacrylamide (37.5:1; Bio-Rad) gels at 18 °C, 300 V for 16 h. For each gene, PCR amplicons of the previously described variants (Li et al., 2017b, 2017c) were included as references to determine sample genotypes in each gel. Statistical Analyses of Associations Statistical analyses were undertaken using Minitab version 16 (Minitab Inc., State College, PA). Unless indicated, all P values were considered statistically significant when P < 0.05. Trends were noted when 0.05 ≤ P < 0.1. General Linear Mixed Models (GLMMs) were used to evaluate the effect of the presence/absence (encoded as 1 or 0, respectively) of 3 of 8 KRTAP20-1 variants (those present at a frequency of ≥5%) detected on the wool traits that had been measured (or calculated). In these models, gender and sire were found to affect (P < 0.05) all the wool traits, and so they were included in the models as fixed and random factors, respectively. Birth rank was not found to affect (P > 0.1) wool traits, and therefore, it was not included in the models. For those wool traits where associations were found between the presence/absence of the KRTAP20-1 variants and variation in the trait in the above models, a second series of “multi-gene GLMMs” that included the genotype of 2 nearby KAP genes (KRTAP6-3 and KRTAP22-1), which have previously been described to affect wool traits (Li et al., 2017b, 2017c), were undertaken. Only those KRTAP6-3 and KRTAP22-1 genotypes with a frequency of over 5% and that affected the wool traits (P < 0.05) were fitted as factors in these models, and gender and sire were again included in the models as fixed and random factors, respectively. RESULTS Identification of KRTAP20-1 in the Sheep Genome A BLAST search of the Ovine Genome Assembly (Oar_v4.0) using a human KRTAP20-1 coding sequence (NM_181615) identified a region with 73% identity on chromosome 1. Analysis of the sequence in this region led to the identification of a 192-bp open reading frame (ORF; NC_019458.2:123265164_123265355). This ORF was clustered with 15 previously described KRTAPs, with KRTAP11-1, KRTAP7-1, KRTAP8-1, KRTAP8-2, KRTAP20-2, KRTAP6-5, KRTAP6-2, KRTAP6-4, KRTAP6-1, KRTAP22-1, and KRTAP6-3 being located upstream and KRTAP15-1, KRTAP13-3, KRTAP26-1, and KRTAP24-1 being positioned downstream (Figure 1). Figure 1. View largeDownload slide KRTAPs identified on sheep chromosome 1 including the newly identified KRTAP20-1 (boxed). Vertical bars represent the approximate location of the different KRTAPs and the arrowheads indicate the direction of transcription. The numbers below the bars indicate the name of the respective KAP genes (i.e., 11.1 is KRTAP11-1). The nucleotide distances are approximate and refer to NC_019458.2. Figure 1. View largeDownload slide KRTAPs identified on sheep chromosome 1 including the newly identified KRTAP20-1 (boxed). Vertical bars represent the approximate location of the different KRTAPs and the arrowheads indicate the direction of transcription. The numbers below the bars indicate the name of the respective KAP genes (i.e., 11.1 is KRTAP11-1). The nucleotide distances are approximate and refer to NC_019458.2. The ORF identified above would encode a 63 amino acid protein, which would possess a high content of glycine and tyrosine (31.8 mol% and 28.6 mol%, respectively). Phylogenetic analysis of the predicted amino acid sequences of this ORF and all of the sheep HGT-KAP genes identified to date, together with the KAP20-n amino acid sequences from goats and humans, revealed that this ORF was different from all known ovine HGT-KAP genes, but was more related to KAP20-1 gene from human than any other known HGT-KAP from sheep (Figure 2). This suggests that the ORF represents the ovine ortholog of the human KAP20-1 gene. Figure 2. View largeDownload slide Phylogenetic tree of the HGT-KAPs identified in sheep, together with human and goat KAP20-n sequences. The tree was constructed using the predicted amino acid sequences for the genes. The numbers at the forks indicate the bootstrap confidence values and only those equal to, or higher than 50%, are shown. The sheep KAPs are indicated with a prefix “s,” whereas the sequences from human and goat are indicated with “h” and “g,” respectively. The newly identified sheep KAP20-1 sequence is indicated with an arrow. The GenBank accession numbers for the other HGT-KAP gene sequences are NM_001193399 (sKAP6-1), KT725832 (sKAP6-2), KT725837 (sKAP6-3), KT725840 (sKAP6-4), KT725845 (sKAP6-5), X05638 (sKAP7-1), X05639 (sKAP8-1), KF220646 (sKAP8-2), MH071391 (sKAP20-2), KX377616 (sKAP22-1), NM_181615 (hKAP20-1), NM_181616 (hKAP20-2), and MF973462 (gKAP20-2). Figure 2. View largeDownload slide Phylogenetic tree of the HGT-KAPs identified in sheep, together with human and goat KAP20-n sequences. The tree was constructed using the predicted amino acid sequences for the genes. The numbers at the forks indicate the bootstrap confidence values and only those equal to, or higher than 50%, are shown. The sheep KAPs are indicated with a prefix “s,” whereas the sequences from human and goat are indicated with “h” and “g,” respectively. The newly identified sheep KAP20-1 sequence is indicated with an arrow. The GenBank accession numbers for the other HGT-KAP gene sequences are NM_001193399 (sKAP6-1), KT725832 (sKAP6-2), KT725837 (sKAP6-3), KT725840 (sKAP6-4), KT725845 (sKAP6-5), X05638 (sKAP7-1), X05639 (sKAP8-1), KF220646 (sKAP8-2), MH071391 (sKAP20-2), KX377616 (sKAP22-1), NM_181615 (hKAP20-1), NM_181616 (hKAP20-2), and MF973462 (gKAP20-2). Sequence Variation Identified in Ovine KRTAP20-1 By optimizing SSCP conditions, 8 different SSCP banding patterns representing 8 variants were detected for the KRTAP20-1 amplicons (Figure 3). One of the variant sequences (SHEEP-KRTAP20-1*A) was identical to the sheep genome assembly sequence, whereas the others were unique, but shared high sequence identity to the genome assembly sequence. These variants were named SHEEP-KRTAP20-1*A to SHEEP-KRTAP20-1*H in accordance with the latest KAP/KRTAP nomenclature (Gong et al., 2012), and the sequences were deposited in GenBank with accession numbers MH243552−MH243559, for variants A to H, respectively. Figure 3. View largeDownload slide PCR-SSCP patterns for ovine KRTAP20-1. Eight different banding patterns representing 8 variants (A to H) are shown in either homozygous or heterozygous forms. Figure 3. View largeDownload slide PCR-SSCP patterns for ovine KRTAP20-1. Eight different banding patterns representing 8 variants (A to H) are shown in either homozygous or heterozygous forms. Sequence analysis revealed a 12-bp insertion/deletion (indel) and 6 SNPs in the gene, including 4 coding SNPs, one 5′ untranslated region (UTR) SNP, and one 3′ UTR SNP (Figure 4). Two of 4 coding SNPs were nonsynonymous and would result in amino acid changes of p.Gly10Ser and p.Gly20Asp. The 5′ UTR SNP was located 17 bp upstream of the ATG start codon, whereas the 3′ UTR SNP was located 24 bp downstream of the stop codon. The indel occurred in the middle of the coding region and would lead to the gain/loss of 4 amino acids “Asn-Tyr-Gly-Cys,” in the middle region of the protein. Figure 4. View largeDownload slide Alignment of the ovine KRTAP20-1 sequence variants. Nucleotides in the coding region are shown in upper case and those in the untranslated regions are shown in lower case. Dashes represent nucleotides identical to the top sequence, and dots represent the deletion of nucleotides. The single nucleotide polymorphisms (SNPs) found in this gene are shown above the sequences, and the positions of those SNPs are shaded. The nonsynonymous SNPs are indicated with their amino acid changes and the primer binding regions are indicated with horizontal lines. Figure 4. View largeDownload slide Alignment of the ovine KRTAP20-1 sequence variants. Nucleotides in the coding region are shown in upper case and those in the untranslated regions are shown in lower case. Dashes represent nucleotides identical to the top sequence, and dots represent the deletion of nucleotides. The single nucleotide polymorphisms (SNPs) found in this gene are shown above the sequences, and the positions of those SNPs are shaded. The nonsynonymous SNPs are indicated with their amino acid changes and the primer binding regions are indicated with horizontal lines. Association of Ovine KRTAP20-1 Variants with Variation in Wool Traits In the 398 Merino-cross sheep used for the association study, while all of the 8 variants were detected, only 3 variants occurred at a frequency of over 5%, these being A (70.4%), B (6.7%), and C (12.9%). All of the other variants (D to H) were observed at a frequency lower than 5%, and the sheep containing these rare variants were removed from the association study. This left 319 sheep for the statistical analyses. In the single gene models, where only KRTAP20-1 itself was considered, the presence of variant A was found to be associated with a decrease in MFD, FDSD, and PF, whereas the presence of variant C was associated with increased GFW and decreased Yield (Table 1). No associations were detected for other wool traits. Table 1. Associations between the absence/presence of KRTAP20-1 variants and various wool traits Trait1 Variant2 Mean ± SE3 P Absent Present GFW (kg) A 2.47 ± 0.07 2.38 ± 0.05 0.140 B 2.40 ± 0.05 2.33 ± 0.08 0.310 C 2.35 ± 0.05 2.48 ± 0.06 0.028 Yield (%) A 71.8 ± 1.00 73.0 ± 0.71 0.222 B 72.6 ± 0.68 74.1 ± 1.23 0.167 C 73.6 ± 0.74 71.2 ± 0.88 0.007 MFD (µm) A 20.3 ± 0.28 19.2 ± 0.20 <0.001 B 19.5 ± 0.20 19.7 ± 0.35 0.561 C 19.4 ± 0.22 19.6 ± 0.25 0.436 FDSD (µm) A 4.32 ± 0.10 4.01 ± 0.07 0.002 B 4.08 ± 0.07 4.14 ± 0.13 0.616 C 4.09 ± 0.08 4.08 ± 0.09 0.956 PF (%) A 4.05 ± 0.52 2.09 ± 0.37 <0.001 B 2.51 ± 0.36 3.23 ± 0.65 0.213 C 2.61 ± 0.40 2.45 ± 0.47 0.736 Trait1 Variant2 Mean ± SE3 P Absent Present GFW (kg) A 2.47 ± 0.07 2.38 ± 0.05 0.140 B 2.40 ± 0.05 2.33 ± 0.08 0.310 C 2.35 ± 0.05 2.48 ± 0.06 0.028 Yield (%) A 71.8 ± 1.00 73.0 ± 0.71 0.222 B 72.6 ± 0.68 74.1 ± 1.23 0.167 C 73.6 ± 0.74 71.2 ± 0.88 0.007 MFD (µm) A 20.3 ± 0.28 19.2 ± 0.20 <0.001 B 19.5 ± 0.20 19.7 ± 0.35 0.561 C 19.4 ± 0.22 19.6 ± 0.25 0.436 FDSD (µm) A 4.32 ± 0.10 4.01 ± 0.07 0.002 B 4.08 ± 0.07 4.14 ± 0.13 0.616 C 4.09 ± 0.08 4.08 ± 0.09 0.956 PF (%) A 4.05 ± 0.52 2.09 ± 0.37 <0.001 B 2.51 ± 0.36 3.23 ± 0.65 0.213 C 2.61 ± 0.40 2.45 ± 0.47 0.736 1GFW = greasy fleece weight; Yield = wool yield; MFD = mean fiber diameter; FDSD = fiber diameter standard deviation; CVFD = coefficient of variation of fiber diameter; PF = prickle factor (percentage of fibers over 30 microns). 2A total of 319 sheep are included in the model when variants that occurred at a frequency under 5% are excluded. Variant A is present in 266 sheep and absent in 53 sheep, variant B is present in 33 sheep and absent in 286 sheep, and variant C is present in 74 sheep and absent in 245 sheep. 3Predicted means and standard error of those means are derived from GLMMs with gender and sire included in the models as fixed and random factors, respectively. P < 0.05 are in bold. View Large Table 1. Associations between the absence/presence of KRTAP20-1 variants and various wool traits Trait1 Variant2 Mean ± SE3 P Absent Present GFW (kg) A 2.47 ± 0.07 2.38 ± 0.05 0.140 B 2.40 ± 0.05 2.33 ± 0.08 0.310 C 2.35 ± 0.05 2.48 ± 0.06 0.028 Yield (%) A 71.8 ± 1.00 73.0 ± 0.71 0.222 B 72.6 ± 0.68 74.1 ± 1.23 0.167 C 73.6 ± 0.74 71.2 ± 0.88 0.007 MFD (µm) A 20.3 ± 0.28 19.2 ± 0.20 <0.001 B 19.5 ± 0.20 19.7 ± 0.35 0.561 C 19.4 ± 0.22 19.6 ± 0.25 0.436 FDSD (µm) A 4.32 ± 0.10 4.01 ± 0.07 0.002 B 4.08 ± 0.07 4.14 ± 0.13 0.616 C 4.09 ± 0.08 4.08 ± 0.09 0.956 PF (%) A 4.05 ± 0.52 2.09 ± 0.37 <0.001 B 2.51 ± 0.36 3.23 ± 0.65 0.213 C 2.61 ± 0.40 2.45 ± 0.47 0.736 Trait1 Variant2 Mean ± SE3 P Absent Present GFW (kg) A 2.47 ± 0.07 2.38 ± 0.05 0.140 B 2.40 ± 0.05 2.33 ± 0.08 0.310 C 2.35 ± 0.05 2.48 ± 0.06 0.028 Yield (%) A 71.8 ± 1.00 73.0 ± 0.71 0.222 B 72.6 ± 0.68 74.1 ± 1.23 0.167 C 73.6 ± 0.74 71.2 ± 0.88 0.007 MFD (µm) A 20.3 ± 0.28 19.2 ± 0.20 <0.001 B 19.5 ± 0.20 19.7 ± 0.35 0.561 C 19.4 ± 0.22 19.6 ± 0.25 0.436 FDSD (µm) A 4.32 ± 0.10 4.01 ± 0.07 0.002 B 4.08 ± 0.07 4.14 ± 0.13 0.616 C 4.09 ± 0.08 4.08 ± 0.09 0.956 PF (%) A 4.05 ± 0.52 2.09 ± 0.37 <0.001 B 2.51 ± 0.36 3.23 ± 0.65 0.213 C 2.61 ± 0.40 2.45 ± 0.47 0.736 1GFW = greasy fleece weight; Yield = wool yield; MFD = mean fiber diameter; FDSD = fiber diameter standard deviation; CVFD = coefficient of variation of fiber diameter; PF = prickle factor (percentage of fibers over 30 microns). 2A total of 319 sheep are included in the model when variants that occurred at a frequency under 5% are excluded. Variant A is present in 266 sheep and absent in 53 sheep, variant B is present in 33 sheep and absent in 286 sheep, and variant C is present in 74 sheep and absent in 245 sheep. 3Predicted means and standard error of those means are derived from GLMMs with gender and sire included in the models as fixed and random factors, respectively. P < 0.05 are in bold. View Large Given that 2 nearby KAP genes, KRTAP22-1 and KRTAP6-3, have been reported to affect wool yield and MFD-associated traits, respectively (Li et al., 2017b, 2017c), we tested whether the associations detected for KRTAP20-1 are a result of linkage with these genes. Accordingly, the genotype of KRTAP22-1 or KRTAP6-3 was fitted into multigene GLMMs. All of the associations detected for KRTAP20-1 in the single-gene models persisted in the multigene models, and the previously reported effects for KRTAP6-3 and KRTAP22-1 were also confirmed (Table 2). Table 2. Effect of KRTAP20-1 on selected wool traits adjusted for the effect of KRTAP6-3 or KRTAP22-1 Trait1 Variant2 Other KRTAP effect KRTAP20-1 effect3 Gene P Absent Present P GFW, (kg) A KRTAP22-1 0.141 2.51 ± 0.07 2.40 ± 0.05 0.147 B KRTAP22-1 0.053 2.43 ± 0.05 2.31 ± 0.09 0.120 C KRTAP22-1 0.105 2.38 ± 0.05 2.52 ± 0.07 0.027 Yield, (%) A KRTAP22-1 0.074 71.7 ± 1.11 72.5 ± 0.77 0.442 B KRTAP22-1 0.031 72.2 ± 0.74 74.1 ± 1.28 0.106 C KRTAP22-1 0.051 72.2 ± 0.79 70.6 ± 0.98 0.006 MFD, (µm) A KRTAP6-3 0.002 20.5 ± 0.31 19.5 ± 0.23 0.001 B KRTAP6-3 <0.001 19.7 ± 0.22 20.0 ± 0.38 0.482 C KRTAP6-3 <0.001 19.6 ± 0.24 19.9 ± 0.28 0.232 FDSD, (µm) A KRTAP6-3 0.071 4.39 ± 0.11 4.08 ± 0.08 0.003 B KRTAP6-3 0.014 4.15 ± 0.08 4.23 ± 0.14 0.510 C KRTAP6-3 0.014 4.14 ± 0.09 4.18 ± 0.10 0.659 PF, (%) A KRTAP6-3 0.007 4.56 ± 0.56 2.61 ± 0.42 <0.001 B KRTAP6-3 <0.001 3.05 ± 0.41 3.88 ± 0.69 0.149 C KRTAP6-3 0.001 3.04 ± 0.44 3.14 ± 0.52 0.823 Trait1 Variant2 Other KRTAP effect KRTAP20-1 effect3 Gene P Absent Present P GFW, (kg) A KRTAP22-1 0.141 2.51 ± 0.07 2.40 ± 0.05 0.147 B KRTAP22-1 0.053 2.43 ± 0.05 2.31 ± 0.09 0.120 C KRTAP22-1 0.105 2.38 ± 0.05 2.52 ± 0.07 0.027 Yield, (%) A KRTAP22-1 0.074 71.7 ± 1.11 72.5 ± 0.77 0.442 B KRTAP22-1 0.031 72.2 ± 0.74 74.1 ± 1.28 0.106 C KRTAP22-1 0.051 72.2 ± 0.79 70.6 ± 0.98 0.006 MFD, (µm) A KRTAP6-3 0.002 20.5 ± 0.31 19.5 ± 0.23 0.001 B KRTAP6-3 <0.001 19.7 ± 0.22 20.0 ± 0.38 0.482 C KRTAP6-3 <0.001 19.6 ± 0.24 19.9 ± 0.28 0.232 FDSD, (µm) A KRTAP6-3 0.071 4.39 ± 0.11 4.08 ± 0.08 0.003 B KRTAP6-3 0.014 4.15 ± 0.08 4.23 ± 0.14 0.510 C KRTAP6-3 0.014 4.14 ± 0.09 4.18 ± 0.10 0.659 PF, (%) A KRTAP6-3 0.007 4.56 ± 0.56 2.61 ± 0.42 <0.001 B KRTAP6-3 <0.001 3.05 ± 0.41 3.88 ± 0.69 0.149 C KRTAP6-3 0.001 3.04 ± 0.44 3.14 ± 0.52 0.823 1GFW = greasy fleece weight; Yield = wool yield; MFD = mean fiber diameter; FDSD = fiber diameter standard deviation; PF = prickle factor (percentage of fibers over 30 microns). 2After removal of the sheep that contained the rare variants of ovine KRTAP20-1, there are 319 left for association analyses. When corrected for KRTAP22-1, ten sheep that had rare KRTAP22-1 genotypes are removed, leaving 309 for these association analyses. Of these, variant A is present in 262 sheep and absent in 47 sheep, variant B is present in 33 sheep and absent in 276 sheep, and variant C is present in 68 sheep and absent in 241 sheep. When corrected for KRTAP6-3, 8 sheep that contained rare KRTAP6-3 genotypes are removed, leaving 311 for these association analyses. Of these, variant A is present in 264 sheep and absent in 47 sheep, variant B is present in 33 sheep and absent in 278 sheep, and variant C is present in 68 sheep and absent in 243 sheep. 3Predicted means and standard error of those means are derived from GLMMs with gender and sire included in the models as fixed and random factors, respectively. P < 0.05 are in bold. View Large Table 2. Effect of KRTAP20-1 on selected wool traits adjusted for the effect of KRTAP6-3 or KRTAP22-1 Trait1 Variant2 Other KRTAP effect KRTAP20-1 effect3 Gene P Absent Present P GFW, (kg) A KRTAP22-1 0.141 2.51 ± 0.07 2.40 ± 0.05 0.147 B KRTAP22-1 0.053 2.43 ± 0.05 2.31 ± 0.09 0.120 C KRTAP22-1 0.105 2.38 ± 0.05 2.52 ± 0.07 0.027 Yield, (%) A KRTAP22-1 0.074 71.7 ± 1.11 72.5 ± 0.77 0.442 B KRTAP22-1 0.031 72.2 ± 0.74 74.1 ± 1.28 0.106 C KRTAP22-1 0.051 72.2 ± 0.79 70.6 ± 0.98 0.006 MFD, (µm) A KRTAP6-3 0.002 20.5 ± 0.31 19.5 ± 0.23 0.001 B KRTAP6-3 <0.001 19.7 ± 0.22 20.0 ± 0.38 0.482 C KRTAP6-3 <0.001 19.6 ± 0.24 19.9 ± 0.28 0.232 FDSD, (µm) A KRTAP6-3 0.071 4.39 ± 0.11 4.08 ± 0.08 0.003 B KRTAP6-3 0.014 4.15 ± 0.08 4.23 ± 0.14 0.510 C KRTAP6-3 0.014 4.14 ± 0.09 4.18 ± 0.10 0.659 PF, (%) A KRTAP6-3 0.007 4.56 ± 0.56 2.61 ± 0.42 <0.001 B KRTAP6-3 <0.001 3.05 ± 0.41 3.88 ± 0.69 0.149 C KRTAP6-3 0.001 3.04 ± 0.44 3.14 ± 0.52 0.823 Trait1 Variant2 Other KRTAP effect KRTAP20-1 effect3 Gene P Absent Present P GFW, (kg) A KRTAP22-1 0.141 2.51 ± 0.07 2.40 ± 0.05 0.147 B KRTAP22-1 0.053 2.43 ± 0.05 2.31 ± 0.09 0.120 C KRTAP22-1 0.105 2.38 ± 0.05 2.52 ± 0.07 0.027 Yield, (%) A KRTAP22-1 0.074 71.7 ± 1.11 72.5 ± 0.77 0.442 B KRTAP22-1 0.031 72.2 ± 0.74 74.1 ± 1.28 0.106 C KRTAP22-1 0.051 72.2 ± 0.79 70.6 ± 0.98 0.006 MFD, (µm) A KRTAP6-3 0.002 20.5 ± 0.31 19.5 ± 0.23 0.001 B KRTAP6-3 <0.001 19.7 ± 0.22 20.0 ± 0.38 0.482 C KRTAP6-3 <0.001 19.6 ± 0.24 19.9 ± 0.28 0.232 FDSD, (µm) A KRTAP6-3 0.071 4.39 ± 0.11 4.08 ± 0.08 0.003 B KRTAP6-3 0.014 4.15 ± 0.08 4.23 ± 0.14 0.510 C KRTAP6-3 0.014 4.14 ± 0.09 4.18 ± 0.10 0.659 PF, (%) A KRTAP6-3 0.007 4.56 ± 0.56 2.61 ± 0.42 <0.001 B KRTAP6-3 <0.001 3.05 ± 0.41 3.88 ± 0.69 0.149 C KRTAP6-3 0.001 3.04 ± 0.44 3.14 ± 0.52 0.823 1GFW = greasy fleece weight; Yield = wool yield; MFD = mean fiber diameter; FDSD = fiber diameter standard deviation; PF = prickle factor (percentage of fibers over 30 microns). 2After removal of the sheep that contained the rare variants of ovine KRTAP20-1, there are 319 left for association analyses. When corrected for KRTAP22-1, ten sheep that had rare KRTAP22-1 genotypes are removed, leaving 309 for these association analyses. Of these, variant A is present in 262 sheep and absent in 47 sheep, variant B is present in 33 sheep and absent in 276 sheep, and variant C is present in 68 sheep and absent in 241 sheep. When corrected for KRTAP6-3, 8 sheep that contained rare KRTAP6-3 genotypes are removed, leaving 311 for these association analyses. Of these, variant A is present in 264 sheep and absent in 47 sheep, variant B is present in 33 sheep and absent in 278 sheep, and variant C is present in 68 sheep and absent in 243 sheep. 3Predicted means and standard error of those means are derived from GLMMs with gender and sire included in the models as fixed and random factors, respectively. P < 0.05 are in bold. View Large DISCUSSION The ortholog of human KRTAP20-1 was identified on sheep chromosome 1, in a region where it is clustered with 15 other known KRTAPs. The detection of 8 sequence variants in a small population of sheep suggests that ovine KRTAP20-1 is highly polymorphic. The level of polymorphism appears to be higher than has been reported for the other gene from the same family (i.e., KRTAP20-2), and for which only 2 sequence variants resulting from a nonsense SNP have been described (Bai et al., 2018). The sequence variation detected in ovine KRTAP20-1 includes SNPs and a small, nonframeshift indel. The SNPs include synonymous, nonsynonymous, 5′ UTR, and 3 UTR SNPs. These may affect the structure and/or expression of KAP20-1 protein and hence wool fiber traits. SNPs and small nonframeshift indels are observed for other KRTAPs, including KRTAP1-1 (Rogers et al., 1994), KRTAP5-4 (Gong et al., 2010), KRTAP6-1 (Gong et al., 2011a), and KRTAP6-3 (Zhou et al., 2016). This variation appears to be a characteristic feature of the KRTAPs. Typically the KRTAPs appear to be clustered in chromosome regions with structures that appear to be conserved between sheep and humans (Gong et al., 2016; Rogers et al., 2006), but the location of KRTAP20-1 is quite different in these 2 species. In humans, KRTAP20-1 is paired with KRTAP20-2 (within a distance of approximately 19 kb) and these genes are located between KRTAP8-1 and KRTAP6-n (Rogers et al., 2002). In sheep, KRTAP20-1 appears to be located between KRTAP6-n and KRTAP15-1, and it is separated from KRTAP20-2 by approximately 131 kb (Figure 1). In addition, the direction of transcription of KRTAP20-1 relative to the other KRTAPs is also different between these 2 species. This suggests that KRTAP20-1 may have evolved via different pathways in sheep and humans. What drives this evolution is currently unknown, but investigations into the biological functions of the genes may shed some light on this. The association results suggest that variation in KRTAP20-1 affects some wool traits including GFW, Yield, MFD, FDSD, and PF. However, care is needed in interpreting these results as KRTAP20-1 is clustered with many other KRTAPs, and association with these wool traits has also been reported for the nearby genes KRTAP22-1 and KRTAP6-3 (Li et al., 2017b, 2017c). KRTAP22-1 is located approximately 51 kb from KRTAP20-1 (Figure 1), and it has been reported to affect wool Yield (Li et al., 2017b). Yield is the proportion of GFW that is CFW, so the associations observed between variation in KRTAP20-1 and variation in GFW and Yield were therefore adjusted for KRTAP22-1. Equally, KRTAP6-3 is approximately 44 kb from KRTAP20-1 (Figure 1), and it has been reported to affect MFD-associated traits (Li et al., 2017c). The association of variation in KRTAP20-1 with variation in MFD, FDSD, and PF was accordingly adjusted for KRTAP6-3. The persistence of KRTAP20-1 associations in the multigene GLMMs (i.e. when adjusted for either KRTAP22-1 or KRTAP6-3) suggests that these effects are unlikely to be due to the linkage of other KRTAPs, but instead reflect the independent effect of KRTAP20-1. The results from multigene GLMMs also confirmed the previous findings that KRTAP22-1 affects Yield (Li et al., 2017b) and that KRTAP6-3 affects MFD-associated traits (Li et al., 2017c). Variant C of KRTAP20-1 was found to be associated with increased GFW and decreased Yield, but not with CFW (Tables 1 and 2). This suggests that C does not affect the quantity of wool fibers produced (and hence does not affect CFW), but possibly increases the amount of other materials present in the raw wool, such as water, wax, and suint (which leads to a higher GFW and lower Yield). Variant C contains a nucleotide substitution c.59G>A which would lead to an amino acid change p.Gly20Asp. Aspartic acid is 1 of 2 acidic amino acids, and its hydrophilic nature may enable the KAP to bind more water. Acidic amino acids are not common in other HGT-KAPs, with the exception of sheep KAP8-2, a protein that is absent in humans (Gong et al., 2014). Whether this substitution directly affects the Yield by allowing wool fibers to absorb more water is certainly worthy of further investigation. Variant A was found to be associated with MFD-associated traits, but no associations were detected for variants B and C (Tables 1 and 2). Variant A differed from variant B by a SNP in the 3′ UTR, and from variant C by 1 nonsynonymous SNP (Figure 4). As discussed above, this nonsy- nonymous SNP would cause a rare amino acid substitution in the context of the known HGT-KAPs, and it may possibly also have an impact on the structure of the protein and/or the cross-linking of the KAP with KIFs. This might consequently affect the fiber diameter–associated traits. SNPs in the 3′ UTR have previously been reported to affect gene expression, either by affecting mRNA stability and translation or by changing a mirRNA regulation process (Amini and Ismail, 2013). The association results appear to be in agreement with the variant frequencies found in the different breeds studied. Although being common in both the Merino and Romney breeds, variant A, which was associated with favorable MFD-associated traits (Tables 1 and 2) was found at a higher frequency in Merino sheep (76%) than in Romney sheep (59%). This is consistent with the observation that Merino wool is in generally finer and has lower FSDS and PF than Romney wool, but it would not in any way explain all of the very large difference in MFD between Romney and Merino sheep wool. Despite KRTAP20-1 being clustered with other HGT-KRTAPs, the effects detected for KRTAP20-1 appear to be different from those reported for other HGT-KRTAPs, including KRTAP6-1, KRTAP6-3, KRTAP20-2, and KRTAP22-1. KRTAP6-1 and KRTAP6-3 have been shown to affect MFD-associated traits (Li et al., 2017c; Zhou et al., 2015), KRTAP22-1 has been reported to affect Yield (Li et al., 2017b), and KRTAP20-2 reportedly affects fiber curvature (Bai et al., 2018). This suggests that the individual HGT-KRTAPs may play distinc- tive roles in the fiber assembly process and hence affect wool fiber traits in different ways. This highlights the importance of characterizing all of the KRTAPs and better understanding their indivi- dual effects, especially if they are to be of use in the development of gene markers for improving wool production. ACKNOWLEDGMENTS We acknowledge the support of the AGMARDT Postdoctoral Fellowship to H.G. and the New Zealand Guardian Trust for the Vernon Willey Trust Fellowship to H.Z. LITERATURE CITED Amini , F. , and E. Ismail . 2013 . 3’-UTR variations and G6PD deficiency . J. Hum. Genet . 58 : 189 – 194 . doi: https://doi.org/10.1038/jhg.2012.155 Google Scholar Crossref Search ADS PubMed Bai , L. , H. Gong , H. Zhou , J. Tao , and J. G. H. Hickford . 2018 . A nucleotide substitution in the ovine KAP20-2 gene leads to a premature stop codon that affects wool fiber curvature . Anim. Genet . 49 : 357 – 358 . doi: https://doi.org/10.1111/age.12668 Google Scholar Crossref Search ADS PubMed Byun , S. O. , Q. Fang , H. Zhou , and J. G. Hickford . 2009 . An effective method for silver-staining DNA in large numbers of polyacrylamide gels . Anal. Biochem . 385 : 174 – 175 . doi: https://doi.org/10.1016/j.ab.2008.10.024 Google Scholar Crossref Search ADS PubMed Gong , H. , H. Zhou , J. M. Dyer , and J. G. Hickford . 2014 . The sheep KAP8-2 gene, a new KAP8 family member that is absent in humans . Springerplus 3 : 528 . doi: https://doi.org/10.1186/2193-1801-3-528 Google Scholar Crossref Search ADS PubMed Gong , H. , H. Zhou , R. H. Forrest , S. Li , J. Wang , J. M. Dyer , Y. Luo , and J. G. H. Hickford . 2016 . Wool keratin-associated protein genes in sheep – a review . Genes 7 : 24 . doi: https://doi.org/10.3390/genes7060024 Google Scholar Crossref Search ADS Gong , H. , H. Zhou , and J. G. H. Hickford . 2011a . Diversity of the glycine/tyrosine-rich keratin-associated protein 6 gene (KAP6) family in sheep . Mol. Biol. Rep . 38 : 31 – 35 . doi: https://doi.org/10.1007/s11033-010-0074-6 Google Scholar Crossref Search ADS Gong , H. , H. Zhou , G. W. McKenzie , Z. Yu , S. Clerens , J. M. Dyer , J. E. Plowman , M. W. Wright , R. Arora , C. S. Bawden , et al. 2012 . An updated nomenclature for keratin-associated proteins (KAPs) . Int. J. Biol. Sci . 8 : 258 – 264 . doi: https://doi.org/10.7150/ijbs.3278 Google Scholar Crossref Search ADS PubMed Gong , H. , H. Zhou , J. E. Plowman , J. M. Dyer , and J. G. Hickford . 2010 . Analysis of variation in the ovine ultra-high sulphur keratin-associated protein KAP5-4 gene using PCR-SSCP technique . Electrophoresis 31 : 3545 – 3547 . doi: https://doi.org/10.1002/elps.201000301 Google Scholar Crossref Search ADS PubMed Holman , B. , and A. Malau-Aduli . 2012 . A review of sheep wool quality traits . Annu. Rev. Res. Biol . 2 : 1 – 14 . Li , S. , H. Zhou , H. Gong , F. Zhao , J. Hu , Y. Luo , and J. G. H. Hickford . 2017a . Identification of the ovine keratin-associated protein 26-1 gene and its association with variation in wool traits . Genes 8 : 225 . doi: https://doi.org/10.3390/genes8090225 Google Scholar Crossref Search ADS Li , S. , H. Zhou , H. Gong , F. Zhao , J. Wang , X. Liu , Y. Luo , and J. G. H. Hickford . 2017b . Identification of the ovine keratin-associated protein 22-1 (KAP22-1) gene and its effect on wool traits . Genes 8 : 27 . doi: https://doi.org/10.3390/genes8010027 Google Scholar Crossref Search ADS Li , S. , H. Zhou , H. Gong , F. Zhao , J. Wang , Y. Luo , amd J. G. H. Hickford . 2017c . Variation in the ovine KAP6-3 gene (KRTAP6-3) is associated with variation in mean fiber diameter-associated wool traits . Genes 8 : 204 . doi: https://doi.org/10.3390/genes8080204 Google Scholar Crossref Search ADS Rogers , G. R. , J. G. Hickford , and R. Bickerstaffe . 1994 . Polymorphism in two genes for B2 high sulfur proteins of wool . Anim. Genet . 25 : 407 – 415 . Google Scholar Crossref Search ADS PubMed Rogers , M. A. , L. Langbein , S. Praetzel-Wunder , and K. Giehl . 2008 . Characterization and expression analysis of the hair keratin associated protein KAP26.1 . Br. J. Dermatol . 159 : 725 – 729 . doi: https://doi.org/10.1111/j.1365-2133.2008.08743.x Google Scholar Crossref Search ADS PubMed Rogers , M. A. , L. Langbein , S. Praetzel-Wunder , H. Winter , and J. Schweizer . 2006 . Human hair keratin-associated proteins (KAPs) . Int. Rev. Cytol . 251 : 209 – 263 . doi: https://doi.org/10.1016/S0074-7696(06)51006-X Google Scholar Crossref Search ADS PubMed Rogers , M. A. , L. Langbein , H. Winter , C. Ehmann , S. Praetzel , and J. Schweizer . 2002 . Characterization of a first domain of human high glycine-tyrosine and high sulfur keratin-associated protein (KAP) genes on chromosome 21q22.1 . J. Biol. Chem . 277 : 48993 – 49002 . doi: https://doi.org/10.1074/jbc.M206422200 Google Scholar Crossref Search ADS PubMed Rogers , M. A. , and J. Schweizer . 2005 . Human KAP genes, only the half of it? Extensive size polymorphisms in hair keratin-associated protein genes . J. Invest. Dermatol . 124 : vii – vix . doi: https://doi.org/10.1111/j.0022-202X.2005.23728.x Google Scholar Crossref Search ADS PubMed Rogers , M. A. , H. Winter , L. Langbein , A. Wollschläger , S. Praetzel-Wunder , L. F. Jave-Suarez , and J. Schweizer . 2007 . Characterization of human KAP24.1, a cuticular hair keratin-associated protein with unusual amino-acid composition and repeat structure . J. Invest. Dermatol . 127 : 1197 – 1204 . doi: https://doi.org/10.1038/sj.jid.5700702 Google Scholar Crossref Search ADS PubMed Wang , J. , L. Che , J. G. Hickford , H. Zhou , Z. Hao , Y. Luo , J. Hu , X. Liu , S. Li . 2017a . Identification of the caprine keratin-associated protein 20–2 (KAP20-2) gene and its effect on cashmere traits . Genes 8 : 328 . doi: https://doi.org/10.3390/genes8110328 Google Scholar Crossref Search ADS Wang , J. , H. Zhou , J. Zhu , J. Hu , X. Liu , S. Li , Y. Luo , and J. G. H. Hickford . 2017b . Identification of the ovine keratin-associated protein 15-1 gene (KRTAP15-1) and genetic variation in its coding sequence . Small Rumin. Res . 153 : 131 – 136 . doi: https://doi.org/10.1016/j.smallrumres.2017.06.007 Google Scholar Crossref Search ADS Zhou , H. , H. Gong , S. Li , Y. Luo , and J. G. Hickford . 2015 . A 57-bp deletion in the ovine KAP6-1 gene affects wool fiber diameter . J. Anim. Breed. Genet . 132 : 301 – 307 . doi: https://doi.org/10.1111/jbg.12138 Google Scholar Crossref Search ADS PubMed Zhou , H. , H. Gong , J. Wang , J. M. Dyer , Y. Luo , and J. G. Hickford . 2016 . Identification of four new gene members of the KAP6 gene family in sheep . Sci. Rep . 6 : 24074 . doi: https://doi.org/10.1038/srep24074 Google Scholar Crossref Search ADS PubMed Zhou , H. , J. G. Hickford , and Q. Fang . 2006 . A two-step procedure for extracting genomic DNA from dried blood spots on filter paper for polymerase chain reaction amplification . Anal. Biochem . 354 : 159 – 161 . doi: https://doi.org/10.1016/j.ab.2006.03.042 Google Scholar Crossref Search ADS PubMed Footnotes This study was financially supported by the Lincoln University Gene-Marker Laboratory and the program of Basic Research Creative Groups of Gansu Province (18JR3RA190). © The Author(s) 2018. Published by Oxford University Press on behalf of the American Society of Animal Science. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Kayser, William, Christian;Carstens, Gordon, E;Parsons, Ira, Loyd;Welsh, Thomas, H;Washburn, Kevin, E;Lawhon, Sara, D;Pinchak, William, E;Richeson, John, T;Chevaux,, Eric;Skidmore, Andrew, L

doi: 10.1093/jas/sky447

Ying,, Zhixiong;Ge,, Xiaoke;Zhang,, Hao;Su,, Weipeng;Li,, Yue;Zhou,, Le;Zhang,, Lili;Wang,, Tian

doi: 10.1093/jas/sky457pmid: 30508105

Abstract This study was conducted to investigate the effects of methionine restriction (MR) on growth performance, insulin sensitivity, and hepatic and muscle glucose metabolism in intrauterine growth retardation (IUGR) pigs at 49 and 105 d of age. At weaning (day 21), 30 female normal birth weight (NBW) piglets were fed control diets with adequate methionine (NBW-CON), whereas 60 female IUGR piglets were fed either the control diets (IUGR-CON) or MR diets which were 30% reduced in methionine (IUGR-MR) (n = 6 replicates (pens) with five piglets per replicate). At 49 and 105 d of age, one pig with a BW near to the mean of each replication was selected for biochemical analysis. Compared with NBW-CON pigs, IUGR-CON pigs exhibited lower relative daily gain (RDG) and homeostasis model assessment of insulin resistance (HOMA-IR) index at day 49 (P < 0.05), but higher RDG and HOMA-IR index at day 105 (P < 0.05). Hepatic phosphoenolpyruvate carboxykinase and glucose-6-phosphatase (G6Pase) activities were higher in IUGR-CON than NBW-CON pigs at both days 49 and 105 (P < 0.05), while hepatic glycogen synthase and glycogen phosphorylase activities were lower in IUGR-CON pigs at both two ages (P < 0.05). In addition, compared with NBW-CON pigs, IUGR-CON pigs (105-d old) had lower protein kinase B phosphorylation (PKB/Akt) in liver (P < 0.05), but not in muscle (P > 0.05). Compared with IUGR-CON pigs, IUGR-MR pigs had lower RDG at day 49, less blood glucose at day 105, and lower HOMA-IR index at both days 49 and 105 (P < 0.05). Additionally, compared with IUGR-CON pigs, MR decreased IUGR-MR pigs’ hepatic G6Pase activities and increased their hepatic glycogen contents at day 105 (P < 0.05), as well as increased their hepatic and muscle PKB/Akt phosphorylation (P < 0.05). In conclusion, the ability of dietary MR to restrict IUGR pigs’ growth and to reduce blood glucose appeared, respectively, in earlier and later period, but MR improved IUGR pigs’ insulin sensitivity at both days 49 and 105. INTRODUCTION Intrauterine growth retardation (IUGR) is usually defined as a failure of the fetus to achieve the intrinsic growth potential (Rosenberg, 2008), affecting approximately 7% to 15% of all newborns in human (Saleem et al., 2011) and 15% to 20% in pig production (Quiniou et al., 2002; Su et al., 2007). Individuals born with IUGR are at high risks of chronic metabolic diseases because of changes in the structures and functions of important metabolic organs during fetal period (Barker, 2012). A great number of human epidemiological investigations and animals researches have demonstrated that IUGR individuals are prone to the development of type 2 diabetes mellitus (T2DM), which was associated with IUGR-induced insulin resistance, increased hepatic glucose production, and reduced hepatic and/or muscle glycogen synthesis (Simmons et al., 2001; Selak et al., 2003; Martin-Gronert and Ozanne, 2007; Whincup et al., 2008; Li et al., 2010). Additionally, it has been found that the catch-up growth occurring in postnatal life was in a close relation with the progress of insulin resistance and glucose metabolic disorder in IUGR individuals (Simmons et al., 2001; Morrison et al., 2010). On the contrary, inhibition of postnatal catch-up could help to attenuate IUGR-induced adverse effects on insulin sensitivity and metabolic activities (Singhal et al., 2003; Lim et al., 2011). Dietary methionine restriction (MR) is a nutritional intervention technique, which decreases methionine content in diets. Besides slowing aging and improving metabolic health in normal birth weight (NBW) animals (Ables et al., 2012; Lees et al., 2014; Stone et al., 2014), our previous study has also showed that MR decreased BWs, improved insulin sensitivity, and reversed hyperglycemia in 180-d-old IUGR pigs, which suggested that dietary MR may be beneficial to lower IUGR-induced high risk of T2DM (Ying et al., 2017). Early postnatal stage is a key window where the IUGR-induced programming can be reversed (Vickers et al., 2005; Hochberg et al., 2011), so it is worth further research about the effects of MR on young IUGR pigs, especially at different ages. Therefore, this study was performed to investigate the effects of MR on growth, blood glucose concentration, insulin sensitivity, and hepatic and muscle glucose metabolism in IUGR pigs at 49 and 105 d of age, which can also investigate potential different effects at these two ages. Because the high similarity of anatomy between pigs and human and the suitability of using pigs as an animal model for human T2DM (Bellinger et al., 2006), this study may provide some information for relevant research in human. METHODS AND MATERIALS Animals and Treatments During the preparation, healthy pregnant sows with similar expected date of confinement (<3 d) and parity (second and third) were preselected and fed a commercial diet according to professional production standard. At farrowing, approximately 90 sows that had similar litter sizes (11 to 13 piglets) and IUGR offspring were further selected. The birth weight and sex of each piglet (Landrace × Yorkshire) were recorded carefully. In each litter, one female NBW piglet (~1.52 kg) and two same-sex naturally occurring IUGR littermates (~0.87 kg) were preselected according to their birth weights (D’inca et al., 2010); specifically, a newborn piglet with a birth weight near the mean (within 0.5 SD) was defined as normal NBW piglet, whereas a 2 SD lower birth weight was identified as IUGR piglet. At weaning (21 d of age), 30 female NBW piglets were randomly selected and fed control diets, whereas their IUGR littermates were randomly assigned to the control diets or MR diets. Thereafter, 30 female NBW piglets and 60 same-sex IUGR piglets were allocated to three groups: NBW-CON group (~6.55 kg), IUGR-CON group (~4.85 kg), and IUGR-MR (~4.84 kg); each group consisted of six replicates (pens) with five piglets per replicate. The control diets were formulated with adequate methionine according to the NRC (NRC, 2012), and the MR diets were 30% reduced in methionine. The two kinds of diets were isonitrogenous by adjusting with l-alanine. The composition and nutrient levels of the diets are given in Supplementary Table 1. Feed and water were provided ad libitum. The BW and feed consumption of pigs were recorded approximately every 2 wk on the basis of pen (including those at 49 and 105 d of age). Then the relative daily gain (RDG; BW gain (g) · day (d)−1· starting BW (kg)−1), relative daily feed intake (RDFI; feed consumption (g) · day (d)−1· mean BW (kg)−1), and G:F (BW gain (g) · feed consumption (g)−1) were calculated. All experiments were approved by the Institutional Animal Care and Use Committee of Nanjing Agricultural University (NJAU-CAST-2015-098). Sample Collection At 49 and 105 d of age, after overnight fasting, one pig with a BW near to the mean of each replication was selected. Then approximately 10 mL heparinized blood samples were taken by jugular venepuncture, and the pigs were sacrificed by intramuscular injection of sodium pentobarbital (50 mg/kg BW). Liver tissues (left lobe) and skeletal muscle samples (left semitendinosus) were immediately collected and stored in liquid nitrogen for future biochemical assay. Plasma was obtained by centrifuging at 2,000 × g for 10 min at 4 °C and stored at −80 °C until biochemical assay. Determinations of Plasma Hormones and Glucose Concentrations Plasma glucose concentration was determined by using a commercial kit (361500; Rongsheng Biotechnology Company, Shanghai, China). Plasma insulin and glucagon concentrations were measured by using Insulin RIA Kit and Glucagon RIA Kit (Beijing North Institute of Biological Technology, Beijing, China). Plasma IGF-1 was determined by using an enzyme-linked immunosorbent kit (CSBE06829p; Cusabio Biotech Company, Wuhan, China). Insulin sensitivity was evaluated by using the homeostasis model assessment of insulin resistance (HOMA-IR = [fasting glucose (mmol/L) × fasting insulin (μU/mL)]/22.5) (Matthews et al., 1985). Assessment of Hepatic and Muscle Glucose Metabolism The activities of phosphoenolpyruvate carboxykinase (PEPCK), glycogen synthase (GYS), and glycogen phosphorylase (GYP) in liver, and muscle GYS activity were measured by using commercial kits based on colorimetric method. Kits for PEPCK (PEPCK-1-Y), GYS (GCS-1-Y), and GYP (GPA-1-Y) were purchased from Suzhou Comin Biotechnology Company, Suzhou, China). Hepatic and muscle glycogen contents were measured by using commercial kits according to the anthracenone method (A043; Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Hepatic glucose-6-phosphatase (G6Pase) activity was assayed according to a previous method (Jia et al., 2012) with slight modifications. Liver tissues were homogenized (1:9, w/v) in 0.25 mol/L sucrose solution, and supernatant were obtained by centrifuged at 20,000 × g for 10 min at 4 °C. For determination of enzyme activity, the supernatant was added into an assay mixture (26.5 mmol/L glucose-6-phosphate and 1.8 mmol/L EDTA) and incubated at 37 °C for 10 min, and then the reaction was quenched by a final concentration of 4% perchloric acid. After that, the inorganic phosphate content in the supernatant was measured using a Pi detection kit (C006-3; Nanjing Jiancheng Bioengineering Institute, Nanjing, China). The G6Pase activity was calculated as release rate of inorganic phosphate (nmol·min−1·mg prot−1). Additionally, total protein concentrations used in the calculation of enzymes’ activities were determined by commercial kits based on BCA method (A045-3; Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Assay of Hepatic and Muscle mRNA Expressions Total RNA was isolated from approximately 50 mg snap-frozen samples (9108; Takara Biotechnology, Japan) and reverse-transcribed into complementary DNA (RR037A; Takara Biotechnology, Dalian, China) by using commercial kits. Real-time PCR was carried out on a QuantStudio 5 real-time PCR system (Applied Biosystems, Foster City, USA). The SYBR Green PCR assay system was 20 µL in total, consisted of 10 µL SYBR Premix Ex Taq, 0.4 µL of the forward and reverse primers, 0.4 µL of ROX reference dye, 6.8 µL of double-distilled H2O, and 2 µL of cDNA template. The reaction conditions were as follows: 30 s at 95 °C, 40 cycles of 5 s at 95 °C, and 30 s at 60 °C. The relative mRNA expression was calculated using the 2−ΔΔCt method after normalization to GAPDH (Pfaffl, 2001). The values of NBW-CON group were used as a calibrator. The primer sequences are shown in Supplementary Table 2. Assay of Hepatic and Muscle Protein Kinase B (PKB/Akt) Phosphorylation Levels The procedure of western blot was conducted according to our previous study (Ying et al., 2017). Proteins were extracted from approximately 50 mg tissues by grinding in RIPA lysis containing protease and phosphatase inhibitors. Equal amounts of protein (70 μg/lane) were separated by SDS–PAGE and then transferred to activated polyvinylidene difluoride membranes. The membranes were blocked using 5% bovine serum albumin dissolved in TBST (TBST; 0.1 % Tween-20, 100 mM Tris–HCl, and 150 mM NaCl, pH 8.0). Primary antibodies for p-PKB/Akt (ser473) (#9271; Cell Signaling Technology, Boston) and total PKB/Akt (#9272; Cell Signaling Technology), and goat-anti-rabbit secondary antibody were purchased from Cell Signaling Technology. The usage of antibodies was according to productions’ instruction. The blots were developed using a chemiluminescence kit (Millipore Corporation, Billerica), and the visualized bands were obtained by a Luminescent Image Analyzer LAS-4000 system (Fujifilm Company, Tokyo, Japan). After that, the intensity of the bands was quantified by Gel-Pro Analyzer 4.0 software (Media Cybernetics, Rockville). Statistical Analysis Data were analyzed by SPSS 20.0 statistical software (SPSS, Chicago) and presented as means ± SEs. After analysis of homogeneity test, statistical differences between groups were determined via one-way ANOVA analysis and Tukey’s post hoc test for multiple comparisons. Differences were considered statistically significant at P < 0.05. RESULTS Growth Performance From 21 to 49 d of age, IUGR decreased (P < 0.05) the RDG and G:F of IUGR-CON group when compared with those of NBW-CON group (Table 1), but did not cause significant changes in RDFI (P > 0.05). Compared with IUGR-CON group, dietary MR decreased IUGR-MR group’ RDG and G:F (P < 0.05), but also did not affect RDFI (P > 0.05). Table 1. Effects of dietary methionine restriction on the growth performance of intrauterine growth retardation pigs at 49 and 105 d of age Item1 NBW-CON IUGR-CON IUGR-MR Days 21 to 492 RDG, g·d−1·kg−1 58.17 ± 1.38a 51.31 ± 1.30b 42.65 ± 1.43c RDFI, g·d−1·kg−1 47.83 ± 1.18b 50.25 ± 1.21ab 53.07 ± 1.61a G:F 0.67 ± 0.02a 0.60 ± 0.02b 0.50 ± 0.01c Days 50 to 1053 RDG, g·d−1·kg−1 32.99 ± 1.53b 44.48 ± 1.59a 45.64 ± 1.68a RDFI, g·d−1·kg−1 38.04 ± 0.48b 44.65 ± 1.22a 45.59 ± 0.67a G:F 0.45 ± 0.01 0.44 ± 0.01 0.44 ± 0.01 Item1 NBW-CON IUGR-CON IUGR-MR Days 21 to 492 RDG, g·d−1·kg−1 58.17 ± 1.38a 51.31 ± 1.30b 42.65 ± 1.43c RDFI, g·d−1·kg−1 47.83 ± 1.18b 50.25 ± 1.21ab 53.07 ± 1.61a G:F 0.67 ± 0.02a 0.60 ± 0.02b 0.50 ± 0.01c Days 50 to 1053 RDG, g·d−1·kg−1 32.99 ± 1.53b 44.48 ± 1.59a 45.64 ± 1.68a RDFI, g·d−1·kg−1 38.04 ± 0.48b 44.65 ± 1.22a 45.59 ± 0.67a G:F 0.45 ± 0.01 0.44 ± 0.01 0.44 ± 0.01 a–cWithin a row, means with different superscript letters differ, P < 0.05. 1NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets; RDG = relative daily gain (BW gain (g)· day (d)−1 · starting BW (kg)−1); RDFI = relative daily feed intake (feed consumption (g) · day (d)−1 · mean BW (kg)−1); G:F = feed efficiency (BW gain (g) · feed consumption (g)−1). 2Results are presented as mean ± SE (n = 6 replicates (pens) with five piglets per replicate). 3Results are presented as mean ± SE (n = 6 replicates (pens) with four piglets per replicate). View Large Table 1. Effects of dietary methionine restriction on the growth performance of intrauterine growth retardation pigs at 49 and 105 d of age Item1 NBW-CON IUGR-CON IUGR-MR Days 21 to 492 RDG, g·d−1·kg−1 58.17 ± 1.38a 51.31 ± 1.30b 42.65 ± 1.43c RDFI, g·d−1·kg−1 47.83 ± 1.18b 50.25 ± 1.21ab 53.07 ± 1.61a G:F 0.67 ± 0.02a 0.60 ± 0.02b 0.50 ± 0.01c Days 50 to 1053 RDG, g·d−1·kg−1 32.99 ± 1.53b 44.48 ± 1.59a 45.64 ± 1.68a RDFI, g·d−1·kg−1 38.04 ± 0.48b 44.65 ± 1.22a 45.59 ± 0.67a G:F 0.45 ± 0.01 0.44 ± 0.01 0.44 ± 0.01 Item1 NBW-CON IUGR-CON IUGR-MR Days 21 to 492 RDG, g·d−1·kg−1 58.17 ± 1.38a 51.31 ± 1.30b 42.65 ± 1.43c RDFI, g·d−1·kg−1 47.83 ± 1.18b 50.25 ± 1.21ab 53.07 ± 1.61a G:F 0.67 ± 0.02a 0.60 ± 0.02b 0.50 ± 0.01c Days 50 to 1053 RDG, g·d−1·kg−1 32.99 ± 1.53b 44.48 ± 1.59a 45.64 ± 1.68a RDFI, g·d−1·kg−1 38.04 ± 0.48b 44.65 ± 1.22a 45.59 ± 0.67a G:F 0.45 ± 0.01 0.44 ± 0.01 0.44 ± 0.01 a–cWithin a row, means with different superscript letters differ, P < 0.05. 1NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets; RDG = relative daily gain (BW gain (g)· day (d)−1 · starting BW (kg)−1); RDFI = relative daily feed intake (feed consumption (g) · day (d)−1 · mean BW (kg)−1); G:F = feed efficiency (BW gain (g) · feed consumption (g)−1). 2Results are presented as mean ± SE (n = 6 replicates (pens) with five piglets per replicate). 3Results are presented as mean ± SE (n = 6 replicates (pens) with four piglets per replicate). View Large From 50 to 105 d of age, IUGR increased (P < 0.05) RDG and RDFI in IUGR-CON group when compared with NBW-CON group, but had no significant effects on G:F (P > 0.05). Compared with IUGR-CON group, MR did not significantly affect these three parameters in IUGR-MR group (P > 0.05). Plasma Glucose and Hormone Concentrations At 49 d of age, the levels of plasma insulin, glucagon, IGF-1, and HOMA-IR index were lower in IUGR-CON group than NBW-CON group (P < 0.05), whereas there were no significant differences in plasma glucose concentrations between the two groups (P > 0.05) (Table 2). Compared with IUGR-CON group, IUGR-MR group had a decreased plasma IGF-1 concentration and lower HOMA-IR index (P < 0.05), but similar plasma glucose, insulin, and glucagon concentrations (P > 0.05). Table 2. Effects of dietary methionine restriction on the plasma glucose and hormone levels of intrauterine growth retardation pigs at 49 and 105 d of age1 Item2 NBW-CON IUGR-CON IUGR-MR Day 49 Glucose, mmol/L 7.68 ± 0.70 6.72 ± 0.64 6.07 ± 0.48 Insulin, pmol/L 169.47 ± 13.09a 119.19 ± 9.49b 83.63 ± 13.22b Glucagon, pg/mL 236.27 ± 12.97a 178.96 ± 16.02b 149.88 ± 13.87b IGF-1, ng/mL 68.72 ± 2.92a 53.62 ± 2.90b 39.34 ± 3.39c HOMA-IR 8.06 ± 0.36a 5.06 ± 0.48b 3.21 ± 0.60c Day 105 Glucose, mmol/L 5.09 ± 0.52ab 6.47 ± 0.45a 4.43 ± 0.32b Insulin, pmol/L 116.69 ± 9.56ab 152.65 ± 13.52a 107.20 ± 10.52b Glucagon, pg/mL 135.61 ± 16.05 150.11 ± 18.88 124.83 ± 12.85 IGF-1, ng/mL 30.98 ± 3.14b 44.96 ± 3.08a 32.49 ± 4.52b HOMA-IR 3.71 ± 0.36b 6.27 ± 0.68a 3.10 ± 0.52b Item2 NBW-CON IUGR-CON IUGR-MR Day 49 Glucose, mmol/L 7.68 ± 0.70 6.72 ± 0.64 6.07 ± 0.48 Insulin, pmol/L 169.47 ± 13.09a 119.19 ± 9.49b 83.63 ± 13.22b Glucagon, pg/mL 236.27 ± 12.97a 178.96 ± 16.02b 149.88 ± 13.87b IGF-1, ng/mL 68.72 ± 2.92a 53.62 ± 2.90b 39.34 ± 3.39c HOMA-IR 8.06 ± 0.36a 5.06 ± 0.48b 3.21 ± 0.60c Day 105 Glucose, mmol/L 5.09 ± 0.52ab 6.47 ± 0.45a 4.43 ± 0.32b Insulin, pmol/L 116.69 ± 9.56ab 152.65 ± 13.52a 107.20 ± 10.52b Glucagon, pg/mL 135.61 ± 16.05 150.11 ± 18.88 124.83 ± 12.85 IGF-1, ng/mL 30.98 ± 3.14b 44.96 ± 3.08a 32.49 ± 4.52b HOMA-IR 3.71 ± 0.36b 6.27 ± 0.68a 3.10 ± 0.52b a–cWithin a row, means with different superscript letters differ, P < 0.05. 1Results are presented as mean ± SE (n = 6). 2NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets; IGF-1 = insulin-like growth factor 1; HOMA-IR = homeostasis model assessment of insulin resistance. View Large Table 2. Effects of dietary methionine restriction on the plasma glucose and hormone levels of intrauterine growth retardation pigs at 49 and 105 d of age1 Item2 NBW-CON IUGR-CON IUGR-MR Day 49 Glucose, mmol/L 7.68 ± 0.70 6.72 ± 0.64 6.07 ± 0.48 Insulin, pmol/L 169.47 ± 13.09a 119.19 ± 9.49b 83.63 ± 13.22b Glucagon, pg/mL 236.27 ± 12.97a 178.96 ± 16.02b 149.88 ± 13.87b IGF-1, ng/mL 68.72 ± 2.92a 53.62 ± 2.90b 39.34 ± 3.39c HOMA-IR 8.06 ± 0.36a 5.06 ± 0.48b 3.21 ± 0.60c Day 105 Glucose, mmol/L 5.09 ± 0.52ab 6.47 ± 0.45a 4.43 ± 0.32b Insulin, pmol/L 116.69 ± 9.56ab 152.65 ± 13.52a 107.20 ± 10.52b Glucagon, pg/mL 135.61 ± 16.05 150.11 ± 18.88 124.83 ± 12.85 IGF-1, ng/mL 30.98 ± 3.14b 44.96 ± 3.08a 32.49 ± 4.52b HOMA-IR 3.71 ± 0.36b 6.27 ± 0.68a 3.10 ± 0.52b Item2 NBW-CON IUGR-CON IUGR-MR Day 49 Glucose, mmol/L 7.68 ± 0.70 6.72 ± 0.64 6.07 ± 0.48 Insulin, pmol/L 169.47 ± 13.09a 119.19 ± 9.49b 83.63 ± 13.22b Glucagon, pg/mL 236.27 ± 12.97a 178.96 ± 16.02b 149.88 ± 13.87b IGF-1, ng/mL 68.72 ± 2.92a 53.62 ± 2.90b 39.34 ± 3.39c HOMA-IR 8.06 ± 0.36a 5.06 ± 0.48b 3.21 ± 0.60c Day 105 Glucose, mmol/L 5.09 ± 0.52ab 6.47 ± 0.45a 4.43 ± 0.32b Insulin, pmol/L 116.69 ± 9.56ab 152.65 ± 13.52a 107.20 ± 10.52b Glucagon, pg/mL 135.61 ± 16.05 150.11 ± 18.88 124.83 ± 12.85 IGF-1, ng/mL 30.98 ± 3.14b 44.96 ± 3.08a 32.49 ± 4.52b HOMA-IR 3.71 ± 0.36b 6.27 ± 0.68a 3.10 ± 0.52b a–cWithin a row, means with different superscript letters differ, P < 0.05. 1Results are presented as mean ± SE (n = 6). 2NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets; IGF-1 = insulin-like growth factor 1; HOMA-IR = homeostasis model assessment of insulin resistance. View Large At 105 d of age, IUGR dramatically elevated IUGR-CON group’s plasma IGF-1 concentration and the HOMA-IR index when compared with NBW-CON group (P < 0.05), but did not cause significant changes in plasma glucose, insulin, and glucagon concentrations (P > 0.05). Compared with IUGR-CON group, MR decreased IUGR-MR group’s plasma glucose, insulin, IGF-1, and HOMA-IR levels (P < 0.05), but did not influence plasma glucagon concentrations (P > 0.05). Hepatic Gluconeogenic Enzyme Activities At both 49 and 105 d of age, the activities of hepatic PEPCK and G6Pase were greater in IUGR-CON group than NBW-CON group (P < 0.05) (Table 3). MR alleviated IUGR-induced increases in the G6Paes activity at 105 d of age (P < 0.05), but did not affect PEPCK activity at 105 d of age and G6Pase activity at 49 and 105 d of age (P > 0.05). Table 3. Effects of dietary methionine restriction on the hepatic gluconeogenic enzyme activities of intrauterine growth retardation pigs at 49 and 105 d of age1 Item2 NBW-CON IUGR-CON IUGR-MR Day 49 PEPCK, nmol·min−1·mg prot−1 28.03 ± 1.62b 48.02 ± 1.92a 43.87 ± 1.34a G6Pase, nmol·min−1·mg prot−1 154.63 ± 47.47b 257.43 ± 33.39a 201.58 ± 44.04ab Day 105 PEPCK, nmol·min−1·mg prot−1 20.16 ± 0.79b 25.15 ± 1.22a 22.45 ± 1.25ab G6Pase, nmol·min−1·mg prot−1 118.95 ± 15.26b 236.85 ± 25.77a 158.69 ± 18.17b Item2 NBW-CON IUGR-CON IUGR-MR Day 49 PEPCK, nmol·min−1·mg prot−1 28.03 ± 1.62b 48.02 ± 1.92a 43.87 ± 1.34a G6Pase, nmol·min−1·mg prot−1 154.63 ± 47.47b 257.43 ± 33.39a 201.58 ± 44.04ab Day 105 PEPCK, nmol·min−1·mg prot−1 20.16 ± 0.79b 25.15 ± 1.22a 22.45 ± 1.25ab G6Pase, nmol·min−1·mg prot−1 118.95 ± 15.26b 236.85 ± 25.77a 158.69 ± 18.17b a,bWithin a row, means with different superscript letters differ, P < 0.05. 1Results are presented as mean ± SE (n = 6). 2NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets; PEPCK = phosphoenolpyruvate carboxykinase; G6Pase = glucose-6-phosphatase. View Large Table 3. Effects of dietary methionine restriction on the hepatic gluconeogenic enzyme activities of intrauterine growth retardation pigs at 49 and 105 d of age1 Item2 NBW-CON IUGR-CON IUGR-MR Day 49 PEPCK, nmol·min−1·mg prot−1 28.03 ± 1.62b 48.02 ± 1.92a 43.87 ± 1.34a G6Pase, nmol·min−1·mg prot−1 154.63 ± 47.47b 257.43 ± 33.39a 201.58 ± 44.04ab Day 105 PEPCK, nmol·min−1·mg prot−1 20.16 ± 0.79b 25.15 ± 1.22a 22.45 ± 1.25ab G6Pase, nmol·min−1·mg prot−1 118.95 ± 15.26b 236.85 ± 25.77a 158.69 ± 18.17b Item2 NBW-CON IUGR-CON IUGR-MR Day 49 PEPCK, nmol·min−1·mg prot−1 28.03 ± 1.62b 48.02 ± 1.92a 43.87 ± 1.34a G6Pase, nmol·min−1·mg prot−1 154.63 ± 47.47b 257.43 ± 33.39a 201.58 ± 44.04ab Day 105 PEPCK, nmol·min−1·mg prot−1 20.16 ± 0.79b 25.15 ± 1.22a 22.45 ± 1.25ab G6Pase, nmol·min−1·mg prot−1 118.95 ± 15.26b 236.85 ± 25.77a 158.69 ± 18.17b a,bWithin a row, means with different superscript letters differ, P < 0.05. 1Results are presented as mean ± SE (n = 6). 2NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets; PEPCK = phosphoenolpyruvate carboxykinase; G6Pase = glucose-6-phosphatase. View Large Hepatic Glycogen Metabolism At 49 d of age, IUGR-CON pigs exhibited decreased hepatic GYS and GYP activities (P < 0.05), but similar hepatic glycogen contents (P > 0.05) when compared with NBW-CON pigs (Table 4). Compared with IUGR-CON pigs, MR increased IUGR-MR pigs’ hepatic GYS activities (P < 0.05), but did not affected their hepatic glycogen contents and GYP activities (P > 0.05). Table 4. Effects of dietary methionine restriction on the hepatic glycogen metabolism of intrauterine growth retardation pigs at 49 and 105 d of age1 Item2 NBW-CON IUGR-CON IUGR-MR Day 49 Glycogen, mg/g tissue 15.84 ± 3.53 14.49 ± 3.57 16.93 ± 3.39 GYS, nmol·min−1·mg prot−1 23.31 ± 3.94a 17.43 ± 3.99b 25.93 ± 2.82a GYP, nmol·min−1·mg prot−1 9.27 ± 0.83a 6.39 ± 0.58b 6.23 ± 0.79b Day 105 Glycogen, mg/g tissue 8.23 ± 0.82ab 7.12 ± 0.75b 10.53 ± 0.87a GYS, nmol·min−1·mg prot−1 16.61 ± 1.18b 14.36 ± 0.97b 22.63 ± 1.46a GYP, nmol·min−1·mg prot−1 16.73 ± 0.90a 11.75 ± 0.76b 14.14 ± 0.47b Item2 NBW-CON IUGR-CON IUGR-MR Day 49 Glycogen, mg/g tissue 15.84 ± 3.53 14.49 ± 3.57 16.93 ± 3.39 GYS, nmol·min−1·mg prot−1 23.31 ± 3.94a 17.43 ± 3.99b 25.93 ± 2.82a GYP, nmol·min−1·mg prot−1 9.27 ± 0.83a 6.39 ± 0.58b 6.23 ± 0.79b Day 105 Glycogen, mg/g tissue 8.23 ± 0.82ab 7.12 ± 0.75b 10.53 ± 0.87a GYS, nmol·min−1·mg prot−1 16.61 ± 1.18b 14.36 ± 0.97b 22.63 ± 1.46a GYP, nmol·min−1·mg prot−1 16.73 ± 0.90a 11.75 ± 0.76b 14.14 ± 0.47b a,bWithin a row, means with different superscript letters differ, P < 0.05. 1Results are presented as mean ± SE (n = 6). 2NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets; GYS = glycogen synthase; GYP = glycogen phosphorylase. View Large Table 4. Effects of dietary methionine restriction on the hepatic glycogen metabolism of intrauterine growth retardation pigs at 49 and 105 d of age1 Item2 NBW-CON IUGR-CON IUGR-MR Day 49 Glycogen, mg/g tissue 15.84 ± 3.53 14.49 ± 3.57 16.93 ± 3.39 GYS, nmol·min−1·mg prot−1 23.31 ± 3.94a 17.43 ± 3.99b 25.93 ± 2.82a GYP, nmol·min−1·mg prot−1 9.27 ± 0.83a 6.39 ± 0.58b 6.23 ± 0.79b Day 105 Glycogen, mg/g tissue 8.23 ± 0.82ab 7.12 ± 0.75b 10.53 ± 0.87a GYS, nmol·min−1·mg prot−1 16.61 ± 1.18b 14.36 ± 0.97b 22.63 ± 1.46a GYP, nmol·min−1·mg prot−1 16.73 ± 0.90a 11.75 ± 0.76b 14.14 ± 0.47b Item2 NBW-CON IUGR-CON IUGR-MR Day 49 Glycogen, mg/g tissue 15.84 ± 3.53 14.49 ± 3.57 16.93 ± 3.39 GYS, nmol·min−1·mg prot−1 23.31 ± 3.94a 17.43 ± 3.99b 25.93 ± 2.82a GYP, nmol·min−1·mg prot−1 9.27 ± 0.83a 6.39 ± 0.58b 6.23 ± 0.79b Day 105 Glycogen, mg/g tissue 8.23 ± 0.82ab 7.12 ± 0.75b 10.53 ± 0.87a GYS, nmol·min−1·mg prot−1 16.61 ± 1.18b 14.36 ± 0.97b 22.63 ± 1.46a GYP, nmol·min−1·mg prot−1 16.73 ± 0.90a 11.75 ± 0.76b 14.14 ± 0.47b a,bWithin a row, means with different superscript letters differ, P < 0.05. 1Results are presented as mean ± SE (n = 6). 2NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets; GYS = glycogen synthase; GYP = glycogen phosphorylase. View Large At 105 d of age, IUGR-CON pigs had decreased hepatic GYP activities (P < 0.05), but similar hepatic glycogen contents and GYS activities (P > 0.05) when compared with NBW-CON pigs. Compared with IUGR-CON pigs, MR elevated IUGR-MR pigs’ hepatic glycogen contents and GYS activities (P < 0.05) without affecting hepatic GYP activities (P > 0.05). Muscle Glycogen Synthesis At 49 d of age, IUGR-CON pigs had lower muscle glycogen contents and GYS activities (P < 0.05) when compared with NBW-CON pigs (Table 5). MR enhanced the muscle GYS activities (P < 0.05) but did not affect the muscle glycogen contents (P > 0.05) of IUGR-MR pigs when compared with those of IUGR-CON pigs. Table 5. Effects of dietary methionine restriction on the muscle glycogen synthesis of intrauterine growth retardation pigs at 49 and 105 d of age1 Item2 NBW-CON IUGR-CON IUGR-MR Day 49 Glycogen, mg/g tissue 4.50 ± 0.35a 3.04 ± 0.33b 3.54 ± 0.44ab GYS, nmol·min−1·mg prot−1 92.48 ± 7.66a 46.45 ± 5.66c 71.31 ± 4.51b Day 105 Glycogen, mg/g tissue 1.75 ± 0.18 1.94 ± 0.21 1.62 ± 0.20 GYS, nmol·min−1·mg prot−1 67.56 ± 3.58 70.56 ± 4.42 55.76 ± 5.43 Item2 NBW-CON IUGR-CON IUGR-MR Day 49 Glycogen, mg/g tissue 4.50 ± 0.35a 3.04 ± 0.33b 3.54 ± 0.44ab GYS, nmol·min−1·mg prot−1 92.48 ± 7.66a 46.45 ± 5.66c 71.31 ± 4.51b Day 105 Glycogen, mg/g tissue 1.75 ± 0.18 1.94 ± 0.21 1.62 ± 0.20 GYS, nmol·min−1·mg prot−1 67.56 ± 3.58 70.56 ± 4.42 55.76 ± 5.43 a–cWithin a row, means with different superscript letters differ, P < 0.05. 1Results are presented as mean ± SE (n = 6). 2NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets; GYS = glycogen synthase. View Large Table 5. Effects of dietary methionine restriction on the muscle glycogen synthesis of intrauterine growth retardation pigs at 49 and 105 d of age1 Item2 NBW-CON IUGR-CON IUGR-MR Day 49 Glycogen, mg/g tissue 4.50 ± 0.35a 3.04 ± 0.33b 3.54 ± 0.44ab GYS, nmol·min−1·mg prot−1 92.48 ± 7.66a 46.45 ± 5.66c 71.31 ± 4.51b Day 105 Glycogen, mg/g tissue 1.75 ± 0.18 1.94 ± 0.21 1.62 ± 0.20 GYS, nmol·min−1·mg prot−1 67.56 ± 3.58 70.56 ± 4.42 55.76 ± 5.43 Item2 NBW-CON IUGR-CON IUGR-MR Day 49 Glycogen, mg/g tissue 4.50 ± 0.35a 3.04 ± 0.33b 3.54 ± 0.44ab GYS, nmol·min−1·mg prot−1 92.48 ± 7.66a 46.45 ± 5.66c 71.31 ± 4.51b Day 105 Glycogen, mg/g tissue 1.75 ± 0.18 1.94 ± 0.21 1.62 ± 0.20 GYS, nmol·min−1·mg prot−1 67.56 ± 3.58 70.56 ± 4.42 55.76 ± 5.43 a–cWithin a row, means with different superscript letters differ, P < 0.05. 1Results are presented as mean ± SE (n = 6). 2NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets; GYS = glycogen synthase. View Large At 105 d of age, both muscle glycogen contents and muscle GYS activities were similar between the three groups (P > 0.05). Hepatic and Muscle mRNA Expressions At both 49 and 105 d of age, IUGR-CON pigs had greater mRNA expressions (P < 0.05) of hepatic phosphoenolpyruvate carboxykinase 1 (PCK1), phosphoenolpyruvate carboxykinase 2 (PCK2), and glucose-6-phosphatase catalytic subunit (G6PC), but lower mRNA expressions (P < 0.05) of hepatic GYS 2 and GYP, as well as a similar muscle GYS1 mRNA expression (P > 0.05) when compared with NBW-CON pigs (Figure 1). MR had no effects on these mRNA expressions when IUGR-CON pigs and IUGR-MR pigs were compared (P > 0.05). Figure 1. View largeDownload slide Effects of dietary methionine restriction on the hepatic and muscle mRNA expressions of intrauterine growth retardation pigs at 49 (A) and 105 (B) d of age. The column and its bar represented mean and SE (n = 6), respectively. Means without a common letter differ, P < 0.05. NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets; PCK1 = phosphoenolpyruvate carboxykinase 1; PCK2 = phosphoenolpyruvate carboxykinase 2; G6PC = glucose-6-phosphatase catalytic subunit; GYS2 = glycogen synthase 2 (liver); PYGL = liver glycogen phosphorylase; GYS2 = glycogen synthase 1 (muscle). Figure 1. View largeDownload slide Effects of dietary methionine restriction on the hepatic and muscle mRNA expressions of intrauterine growth retardation pigs at 49 (A) and 105 (B) d of age. The column and its bar represented mean and SE (n = 6), respectively. Means without a common letter differ, P < 0.05. NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets; PCK1 = phosphoenolpyruvate carboxykinase 1; PCK2 = phosphoenolpyruvate carboxykinase 2; G6PC = glucose-6-phosphatase catalytic subunit; GYS2 = glycogen synthase 2 (liver); PYGL = liver glycogen phosphorylase; GYS2 = glycogen synthase 1 (muscle). Hepatic and Muscle PKB/Akt Phosphorylation Levels As shown in Figure 2, IUGR-CON pigs had lower PKB/Akt (ser473) phosphorylation levels in liver (P < 0.05), but not in muscle (P > 0.05) when compared with NBW-CON pigs. MR elevated hepatic and muscle PKB/Akt (ser473) phosphorylation levels when IUGR-CON pigs and IUGR-MR pigs were compared (P < 0.05). Figure 2. View largeDownload slide Effects of dietary methionine restriction on the phosphorylation levels of hepatic (A) and muscle (B) protein kinase B (PKB/Akt) of intrauterine growth retardation pigs at 105 d of age. The column and its bar represented mean and SE (n = 6), respectively. Means without a common letter differ, P < 0.05. NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets. Figure 2. View largeDownload slide Effects of dietary methionine restriction on the phosphorylation levels of hepatic (A) and muscle (B) protein kinase B (PKB/Akt) of intrauterine growth retardation pigs at 105 d of age. The column and its bar represented mean and SE (n = 6), respectively. Means without a common letter differ, P < 0.05. NBW-CON = normal birth weight pigs fed control diets; IUGR-CON = intrauterine growth restriction pigs fed control diets; IUGR-MR = intrauterine growth restriction pigs fed methionine restriction diets. DISCUSSION In the present study, IUGR decreased IUGR-CON pigs’ growth rates during the period from 21 to 49 d of age, but caused growth rates’ increases from 50 to 105 d of age, reflected by the data of RDG. G:F is indicator reflecting the capacity of utilizing nutrients for body growth. Su et al.’s study showed that IUGR decreased 49-d-old pigs’ G:F by impairing their intestinal structure and function, which consequently reduced these pigs’ growth rates (Su et al., 2017). Therefore, in our study, the decreased G:F of IUGR-CON pigs from days 21 to 49 may be responsible for the reduced growth rates during this period and the recovered G:F together with the increased feed intake (reflected by increased RDFI) of IUGR-CON pigs from days 50 to 105 were important reasons for the increased growth rates of these elder IUGR-CON pigs. In parallel with the contradictory changes in IUGR-CON pigs’ growth rates at different ages, IUGR-CON pigs’ also exhibited a decreased plasma IGF-1 level at 49 d of age, but an increased one at 105 d of age. IGF-1 is an important hormone, which is closely related with pre- and postnatal growth (Cianfarani et al., 2006). Previous studies showed that IUGR lowered fetal arterial IGF-1 levels (Thorn et al., 2009) and that postnatal catch-up growth was associated with increased blood IGF-1 (Özkan et al., 1999; Cianfarani et al., 2002). Therefore, it was likely that the decreased IGF-1 of 49-d-old IUGR-CON pigs derived from fetal defects and that the increased IGF-1 of 105-d-old IUGR-CON pigs was associated with accelerated growth of these pigs. However, further researches are needed to investigate the mechanism for the relation between increased IGF-1 and accelerated growth in IUGR individuals. Similarly, the inconsistent changes in the plasma insulin and glucagon concentrations between 49- and 105-d-old IUGR-CON pigs may be also related with their changed growth rates. Blood glucose homeostasis is a result of an intricate balance, which is mainly controlled by exogenous glucose uptake, endogenous glucose production, and glucose removal from the blood. Hepatic glucose production (gluconeogenesis and glycogen hydrolysis) and hepatic and muscle glycogen storage are the important pathways for endogenous glucose production and blood glucose removal, respectively (Sharabi et al., 2015). Our previous study found that the combination of increased hepatic gluconeogenesis (reflected by increased gluconeogenic enzymes’ activities) and decreased muscle glycogen contents resulted in hyperglycemia in 180-d-old IUGR pigs (Ying et al., 2017). However, in the present study, the same combination found in 49-d-old IUGR-CON pigs did not affect blood glucose concentrations. These contradictory findings made us further notice the decreased G:F of the 49-d-old IUGR-CON pigs, which may suggest reduced absorption of exogenous nutrients (i.e., glucose). Therefore, the reason why the 49-d-old IUGR-CON pigs could maintain normal blood glucose concentrations despite of increased hepatic gluconeogenesis and decreased muscle glycogen may be related to the reduced absorption of exogenous glucose, and such reduced absorption of exogenous glucose could also partly explain why muscle glycogen was only reduced in the 49-d-old IUGR-CON pigs, but not in 105-d-old IUGR-CON pigs. In addition, blood glucose concentrations of IUGR-CON pigs were also unchanged at 105 d of age. Such age-related changes in blood glucose concentrations were also found in Simmons et al. (2001) study and Poore and Fowden (2002) study where IUGR animals only developed hyperglycemia at old ages but not at young ages. Despite the effects of IUGR on IUGR-CON pigs’ the growth rates and blood glucose concentrations were different between days 49 and 105, IUGR consistently increased these pigs’ hepatic PEPCK and G6Pase activities at the two ages. Such increased hepatic PEPCK and G6Pase activities were also found in newborn IUGR pigs (Jia et al., 2012) and adult IUGR pigs (Ying et al., 2017) and rats (Nyirenda et al., 1998). Therefore, all these results suggested that IUGR-induced effects on hepatic gluconeogenic enzymes’ activities were consistent irrespective of different ages. Real-time PCR data showed the mRNA expressions of PEPCK-encoding gene (i.e., PCK1 and PCK2) and G6Pase-encoding gene (i.e., G6PC) were increased in the IUGR-CON pigs of both ages, which may be, respectively, responsible for the increased PEPCK and G6Pase activities of these pigs. Insulin-induced down-regulation of gene expression is a major pathway to reduce hepatic gluconeogenic enzymes’ amount and activities, in which PKB/Akt phosphorylation is a key step (Barthel and Schmoll, 2003). Although this study only measured the PKB/Akt phosphorylation at day 105, the found decreases in the hepatic PKB/Akt phosphorylation of IUGR-CON pigs could partly explain the up-regulated mRNA expressions of the hepatic PCK1, PCK2, and G6PC in these pigs, which were also previously proved in 180-d-old IUGR pigs (Ying et al., 2017). Interestingly, although IUGR decreased IUGR-CON pigs’ hepatic PKB/Akt phosphorylation, it did not change these pigs’ muscle PKB/Akt phosphorylation levels, which may be because that liver was more susceptible to insulin resistance than skeletal muscle (Kraegen et al., 1991). Besides the age-related changes in the growth rates and blood glucose concentrations of IUGR-CON pigs, IUGR even caused completely opposite effects on insulin sensitivity between 49- and 105-d-old IUGR-CON pigs, reflected by data of HOMA-IR index (Matthews et al., 1985); namely, the insulin sensitivity of IUGR-CON pigs was improved at 49 d of age but decreased at 105 d of age. During intrauterine period, IUGR fetal had less access to nutrients and therefore they developed selective adaptations, including enhanced insulin sensitivity to achieve a greater nutrient storage (Thorn et al., 2009). Therefore, we presumed that this selective adaptation could persist until early postnatal stage and resulted in the improved insulin sensitivity of IUGR-CON pigs at 49 d of age; however, with age, catch-up growth and excessive feed intake, IUGR-CON pigs eventually turned from insulin hypersensitivity to insulin resistance in late life. Similar with the effects of MR on NBW animals (Ables et al., 2012; Stone et al., 2014), MR decreased the RDG and G:F of IUGR-CON pigs during the period from 21 to 49 d of age, which may be associated with increased energy expenditure caused by MR (Malloy et al., 2006). However, such an ability of MR to inhibit IUGR pigs’ growth disappeared during the period from 50 to 105 d and the period from 106 to 180 d (RDG of IUGR-CON pigs: 17.27 ± 0.79 vs. RDG of IUGR-MR pigs: 17.91 ± 0.57; P > 0.05; unpublished) of age. Because the period from 50 to 105 d and the period from 106 to 180 d (RDG of NBW-CON pigs: 13.72 ± 0.38 vs. RDG of IUGR-CON pigs: 17.27 ± 0.79; P < 0.05; unpublished) of age were the times when IUGR-CON pigs developed postnatal catch-up growth, it was likely that the inhibition of MR in the postnatal growth of IUGR pigs was eliminated by the effect of catch-up growth. In the present study, MR decreased IUGR-MR pigs’ blood glucose concentrations at 105 d of age, but did not affect this parameter at day 49. The decreased blood glucose of 105-d-old IUGR-MR pigs may be associated with the increased hepatic glycogen contents of these pigs, which may be caused by the enhanced hepatic GYS activity, because GYS is a rate-limiting enzyme of glycogen synthesis. Insulin signal is an important regulator of GYS activity, in which PKB/Akt phosphorylation is a pivotal trigger (Barthel and Schmoll, 2003). Therefore, the increased PKB/Akt phosphorylation of the 105-d-old IUGR-MR pigs may be the main reason for the increased hepatic GYS activities of these pigs when the hepatic GYS-encoding gene (i.e., GYS 2) was not affected by MR. In addition to the increased hepatic glycogen contents of the 105-d-old IUGR-MR pigs, the decreased hepatic G6Pase activity may also play a role in decreasing the blood glucose of these 105-d-old IUGR-MR pigs by reducing hepatic gluconeogenesis, because G6Pase is the rate-limiting enzyme catalyzing the final step of gluconeogenesis (Barthel and Schmoll, 2003). However, although 49-d-old IUGR-MR pigs also had increased hepatic GYS activities as the 105-d-old IUGR-MR pigs, such increased GYS activities did not result in increased hepatic glycogen contents in the 49-d-old IUGR-MR pigs. This unchanged hepatic glycogen content of the 49-d-old IUGR-MR pigs may be associated with the unchanged hepatic G6Pase activities of these pigs, because as the last enzyme of hepatic gluconeogenesis (Barthel and Schmoll, 2003), a decrease in G6Pase activity may stimulate hepatic glycogen synthesis by increasing hepatic glucose-6-phosphate which is the substrate of glycogen synthesis (Ferrer et al., 2003). Meanwhile, increased consumption of glucose-6-phosphorate via glycogen synthesis could in return reduce G6Pase activity by decreasing G6Pase’s access to this substrate, which could be the reason for the different changes in the G6Pase activities between the 49- and 105-d-old IUGR-MR pigs especially when MR did not affect the mRNA expression of G6Pase encoding gene. In addition to decreasing blood glucose concentrations of 105-d-old IUGR-MR pigs, MR also decreased blood glucose, increased hepatic glycogen, and inhibited hepatic G6Pase activities in 180-d-old IUGR pigs (Ying et al., 2017). All these results together may imply time-dependent effects of MR on IUGR pigs’ glucose metabolism and that the occurrence of these effects may somehow have a relationship with catch-up growth of IUGR pigs. Although MR caused different effects on IUGR pigs’ growth and glucose metabolism at different ages, MR consistently improved IUGR pigs’ capacity to control blood glucose concentration, reflected by the decreased HOMA-IR index of IUGR-MR pigs (Matthews et al., 1985). Additionally, MR produced no effect on the IUGR pigs’ mRNA expressions of different enzymes involved in gluconeogenesis and glycogen metabolism no matter the ages of these, which suggested that the effects of MR glucose metabolic enzymes did not involve in gene regulation. In conclusion, dietary MR caused different effects on the growth rates and the glucose homeostasis between 49- and 105-d-old IUGR pigs; namely, MR did not affect IUGR pigs’ blood glucose concentrations at 49 d of age, but it lowered this parameter in 105-d-old IUGR pigs, which was associated with increased hepatic glycogen synthesis and decreased hepatic gluconeogenesis. Additionally, MR improved IUGR pigs’ insulin sensitivity at both days 49 and 105. It was also worthwhile to mention IUGR-induced different changes that IUGR decreased growth rates and improved insulin sensitivity in 49-d-old IUGR-CON pigs, but resulted in opposite results in 105-d-old IUGR-CON pigs. Conflict of interest statement. None declared. ACKNOWLEDGMENT The authors thank their laboratory colleagues for their assistance. LITERATURE CITED Ables , G. P. , C. E. Perrone , D. Orentreich , and N. Orentreich . 2012 . Methionine-restricted C57BL/6J mice are resistant to diet-induced obesity and insulin resistance but have low bone density . PLoS One 7 : e51357 . doi: https://doi.org/10.1371/journal.pone.0051357 Google Scholar Crossref Search ADS PubMed Barker , D. J. P . 2012 . Developmental origins of chronic disease . Public Health 126 : 185 – 189 . doi: https://doi.org/10.1016/j.puhe.2011.11.014 Google Scholar Crossref Search ADS PubMed Barthel , A. , and D. Schmoll . 2003 . Novel concepts in insulin regulation of hepatic gluconeogenesis . Am. J. Physiol. Endocrinol. Metab . 285 : E685 – E692 . doi: https://doi.org/10.1152/ajpendo.00253.2003 Google Scholar Crossref Search ADS PubMed Bellinger , D. A. , E. P. Merricks , and T. C. Nichols . 2006 . Swine models of type 2 diabetes mellitus: insulin resistance, glucose tolerance, and cardiovascular complications . ILAR J . 47 : 243 – 258 . doi: https://doi.org/10.1093/ilar.47.3.243 Google Scholar Crossref Search ADS PubMed Cianfarani , S. , C. Geremia , C. D. Scott , and D. Germani . 2002 . Growth, IGF system, and cortisol in children with intrauterine growth retardation: is catch-up growth affected by reprogramming of the hypothalamic-pituitary-adrenal axis ? Pediatr. Res . 51 : 94 – 99 . doi: https://doi.org/10.1203/00006450-200201000-00017 Google Scholar Crossref Search ADS PubMed Cianfarani , S. , C. Ladaki , and C. Geremia . 2006 . Hormonal regulation of postnatal growth in children born small for gestational age . Horm. Res . 65 ( Suppl. 3 ): 70 – 74 . doi: https://doi.org/10.1159/000091509 Google Scholar PubMed D’Inca , R. , M. Kloareg , C. Gras-Le Guen , and I. Le Huërou-Luron . 2010 . Intrauterine growth restriction modifies the developmental pattern of intestinal structure, transcriptomic profile, and bacterial colonization in neonatal pigs . J. Nutr . 140 : 925 – 931 . doi: https://doi.org/10.3945/jn.109.116822 Google Scholar Crossref Search ADS PubMed Ferrer , J. C. , C. Favre , R. R. Gomis , J. M. Fernández-Novell , M. García-Rocha , N. de la Iglesia , E. Cid , and J. J. Guinovart . 2003 . Control of glycogen deposition . FEBS Lett . 546 : 127 – 132 . doi: https://doi.org/10.1016/S0014-5793(03)00565-9 Google Scholar Crossref Search ADS PubMed Hochberg , Z. , R. Feil , M. Constancia , M. Fraga , C. Junien , J. C. Carel , P. Boileau , Y. Le Bouc , C. L. Deal , K. Lillycrop , et al. 2011 . Child health, developmental plasticity, and epigenetic programming . Endocr. Rev . 32 : 159 – 224 . doi: https://doi.org/10.1210/er.2009-0039 Google Scholar Crossref Search ADS PubMed Jia , Y. , R. Cong , R. Li , X. Yang , Q. Sun , N. Parvizi , and R. Zhao . 2012 . Maternal low-protein diet induces gender-dependent changes in epigenetic regulation of the glucose-6-phosphatase gene in newborn piglet liver . J. Nutr . 142 : 1659 – 1665 . doi: https://doi.org/10.3945/jn.112.160341 Google Scholar Crossref Search ADS PubMed Kraegen , E. W. , P. W. Clark , A. B. Jenkins , E. A. Daley , D. J. Chisholm , and L. H. Storlien . 1991 . Development of muscle insulin resistance after liver insulin resistance in high-fat-fed rats . Diabetes . 40 : 1397 – 1403 . doi: https://doi.org/10.2337/diab.40.11.1397 Google Scholar Crossref Search ADS PubMed Lees , E. K. , E. Król , L. Grant , K. Shearer , C. Wyse , E. Moncur , A. S. Bykowska , N. Mody , T. W. Gettys , and M. Delibegovic . 2014 . Methionine restriction restores a younger metabolic phenotype in adult mice with alterations in fibroblast growth factor 21 . Aging Cell 13 : 817 – 827 . doi: https://doi.org/10.1111/acel.12238 Google Scholar Crossref Search ADS PubMed Li , Y. , Y. He , L. Qi , V. W. Jaddoe , E. J. Feskens , X. Yang , G. Ma , and F. B. Hu . 2010 . Exposure to the Chinese famine in early life and the risk of hyperglycemia and type 2 diabetes in adulthood . Diabetes 59 : 2400 – 2406 . doi: https://doi.org/10.2337/db10-0385 Google Scholar Crossref Search ADS PubMed Lim , K. , J. A. Armitage , A. Stefanidis , B. J. Oldfield , and M. J. Black . 2011 . IUGR in the absence of postnatal “catch-up” growth leads to improved whole body insulin sensitivity in rat offspring . Pediatr. Res . 70 : 339 – 344 . doi: https://doi.org/10.1203/PDR.0b013e31822a65a3 Google Scholar Crossref Search ADS PubMed Malloy , V. L. , R. A. Krajcik , S. J. Bailey , G. Hristopoulos , J. D. Plummer , and N. Orentreich . 2006 . Methionine restriction decreases visceral fat mass and preserves insulin action in aging male Fischer 344 rats independent of energy restriction . Aging Cell 5 : 305 – 314 . doi: https://doi.org/10.1111/j.1474-9726.2006.00220.x Google Scholar Crossref Search ADS PubMed Martin-Gronert , M. S. , and S. E. Ozanne . 2007 . Experimental IUGR and later diabetes . J. Intern. Med . 261 : 437 – 452 . doi: https://doi.org/10.1111/j.1365-2796.2007.01800.x Google Scholar Crossref Search ADS PubMed Matthews , D. R. , J. P. Hosker , A. S. Rudenski , B. A. Naylor , D. F. Treacher , and R. C. Turner . 1985 . Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man . Diabetologia . 28 : 412 – 419 . doi: https://doi.org/10.1007/BF00280883 Google Scholar Crossref Search ADS PubMed Morrison , J. L. , J. A. Duffield , B. S. Muhlhausler , S. Gentili , and I. C. McMillen . 2010 . Fetal growth restriction, catch-up growth and the early origins of insulin resistance and visceral obesity . Pediatr. Nephrol . 25 : 669 – 677 . doi: https://doi.org/10.1007/s00467-009-1407-3 Google Scholar Crossref Search ADS PubMed NRC . 2012 . Nutrient requirement of swine . 11 th rev. ed. Washington, DC: National Academic Science . Nyirenda , M. J. , R. S. Lindsay , C. J. Kenyon , A. Burchell , and J. R. Seckl . 1998 . Glucocorticoid exposure in late gestation permanently programs rat hepatic phosphoenolpyruvate carboxykinase and glucocorticoid receptor expression and causes glucose intolerance in adult offspring . J. Clin. Invest . 101 : 2174 – 2181 . doi: https://doi.org/10.1172/JCI1567 Google Scholar Crossref Search ADS PubMed Özkan , H. , A. Aydın , N. Demir , T. Erci , and A. Büyükgebiz . 1999 . Associations of IGF-I, IGFBP-1 and IGFBP-3 on intrauterine growth and early catch-up growth . Neonatology . 76 : 274 – 282 . doi: https://doi.org/10.1159/000014169 Google Scholar Crossref Search ADS Pfaffl , M. W . 2001 . A new mathematical model for relative quantification in real-time RT-PCR . Nucleic Acids Res . 29 : e45 . doi: https://doi.org/10.1093/nar/29.9.e45 Google Scholar Crossref Search ADS PubMed Poore , K. R. , and A. L. Fowden . 2002 . The effect of birth weight on glucose tolerance in pigs at 3 and 12 months of age . Diabetologia 45 : 1247 – 1254 . doi: https://doi.org/10.1007/s00125-002-0849-y Google Scholar Crossref Search ADS PubMed Quiniou , N. , J. Dagorn , and D. Gaudre . 2002 . Variation of piglets’ birth weight and consequences on subsequent performance . Livest. Prod. Sci . 78 : 63 – 70 . doi: https://doi.org/10.1016/S0301-6226(02)00181-1 Google Scholar Crossref Search ADS Rosenberg , A . 2008 . The IUGR newborn . Semin. Perinatol . 32 : 219 – 224 . doi: https://doi.org/10.1053/j.semperi.2007.11.003 Google Scholar Crossref Search ADS PubMed Saleem , T. , N. Sajjad , S. Fatima , N. Habib , S. R. Ali , and M. Qadir . 2011 . Intrauterine growth retardation–small events, big consequences . Ital. J. Pediatr . 37 : 41 . doi: https://doi.org/10.1186/1824-7288-37-41 Google Scholar Crossref Search ADS PubMed Selak , M. A. , B. T. Storey , I. Peterside , and R. A. Simmons . 2003 . Impaired oxidative phosphorylation in skeletal muscle of intrauterine growth-retarded rats . Am. J. Physiol. Endocrinol. Metab . 285 : E130 – E137 . doi: https://doi.org/10.1152/ajpendo.00322.2002 Google Scholar Crossref Search ADS PubMed Sharabi , K. , C. D. Tavares , A. K. Rines , and P. Puigserver . 2015 . Molecular pathophysiology of hepatic glucose production . Mol. Aspects Med . 46 : 21 – 33 . doi: https://doi.org/10.1016/j.mam.2015.09.003 Google Scholar Crossref Search ADS PubMed Simmons , R. A. , L. J. Templeton , and S. J. Gertz . 2001 . Intrauterine growth retardation leads to the development of type 2 diabetes in the rat . Diabetes 50 : 2279 – 2286 . doi: https://doi.org/10.2337/diabetes.50.10.2279 Google Scholar Crossref Search ADS PubMed Singhal , A. , M. Fewtrell , T. J. Cole , and A. Lucas . 2003 . Low nutrient intake and early growth for later insulin resistance in adolescents born preterm . Lancet 361 : 1089 – 1097 . doi: https://doi.org/10.1016/S0140-6736(03)12895-4 Google Scholar Crossref Search ADS PubMed Stone , K. P. , D. Wanders , M. Orgeron , C. C. Cortez , and T. W. Gettys . 2014 . Mechanisms of increased in vivo insulin sensitivity by dietary methionine restriction in mice . Diabetes 63 : 3721 – 3733 . doi: https://doi.org/10.2337/db14-0464 Google Scholar Crossref Search ADS PubMed Su , G. , M. S. Lund , and D. Sorensen . 2007 . Selection for litter size at day five to improve litter size at weaning and piglet survival rate . J. Anim. Sci . 85 : 1385 – 1392 . doi: https://doi.org/10.2527/jas.2006-631 Google Scholar Crossref Search ADS PubMed Su , W. , H. Zhang , Z. Ying , Y. Li , L. Zhou , F. Wang , L. Zhang , and T. Wang . 2017 . Effects of dietary L-methionine supplementation on intestinal integrity and oxidative status in intrauterine growth-retarded weanling piglets . Eur. J. Nutr . 57:2735–2745. doi: https://doi.org/10.1007/s00394-017-1539-3 Thorn , S. R. , T. R. Regnault , L. D. Brown , P. J. Rozance , J. Keng , M. Roper , R. B. Wilkening , W. W. Hay , Jr , and J. E. Friedman . 2009 . Intrauterine growth restriction increases fetal hepatic gluconeogenic capacity and reduces messenger ribonucleic acid translation initiation and nutrient sensing in fetal liver and skeletal muscle . Endocrinology 150 : 3021 – 3030 . doi: https://doi.org/10.1210/en.2008-1789 Google Scholar Crossref Search ADS PubMed Vickers , M. H. , P. D. Gluckman , A. H. Coveny , P. L. Hofman , W. S. Cutfield , A. Gertler , B. H. Breier , and M. Harris . 2005 . Neonatal leptin treatment reverses developmental programming . Endocrinology 146 : 4211 – 4216 . doi: https://doi.org/10.1210/en.2005-0581 Google Scholar Crossref Search ADS PubMed Whincup , P. H. , S. J. Kaye , C. G. Owen , R. Huxley , D. G. Cook , S. Anazawa , E. Barrett-Connor , S. K. Bhargava , B. E. Birgisdottir , S. Carlsson , et al. 2008 . Birth weight and risk of type 2 diabetes: a systematic review . Jama 300 : 2886 – 2897 . doi: https://doi.org/10.1001/jama.2008.886 Google Scholar Crossref Search ADS PubMed Ying , Z. , H. Zhang , W. Su , L. Zhou , F. Wang , Y. Li , L. Zhang , and T. Wang . 2017 . Dietary methionine restriction alleviates hyperglycemia in pigs with intrauterine growth restriction by enhancing hepatic protein kinase B signaling and glycogen synthesis . J. Nutr . 147 : 1892 – 1899 . doi: https://doi.org/10.3945/jn.117.253427 Google Scholar Crossref Search ADS PubMed Footnotes This work was supported by the National Natural Science Foundation of China (grant number 31572418) and National Key Research and Development Program of China (018YFD0501100). © The Author(s) 2018. Published by Oxford University Press on behalf of the American Society of Animal Science. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)