Journal of Animal Science

Appel, Anne K.;Voß, Barbara;Tönepöhl, Björn;von Borstel, Uta König;Gauly, Matthias

doi: 10.2527/jas.2013-6694pmid: 24158365

Abstract Mixing of unfamiliar pigs is a standard management procedure in commercial pig production and is often associated with a period of intense and physically damaging aggression. Aggression is considered a problem for animal welfare and production. The objective of the present paper was to investigate the genetic background of aggressive behavior traits at mixing of unfamiliar gilts under 2 different housing conditions. Therefore, a total of 543 purebred Pietrain gilts, from 2 nucleus farms (farm A: n = 302; farm B: n = 241) of 1 breeding company, were tested at an average age of 214 d (SD 12.2 d) for aggressive behavior by 1 observer. Observations included the frequencies of aggressive attack and reciprocal fighting during mixing with unfamiliar gilts. On farm A 41% of the gilts were purebred Pietrains, whereas 59% were purebred Landrace or Duroc gilts. On the farm B 42% of the gilts were purebred Pietrains, and 58% purebred Large White gilts. The average size of the newly mixed groups of gilts was 28 animals on farm A and 18 animals on farm B. The Pietrain gilts from the 2 herds were genetically closely linked. They were the offspring of 96 sires, with 64% of these sires having tested progeny in both farms. There were clear differences in the housing of the animals between the 2 farms. The test pen on farm A had a solid concrete floor littered with wooden shavings and was equipped with a dry feeder. On farm B there was a partly slatted floor, and the gilts were fed by an electronic sow feeder. Mean space allowance was 2.6 m2/gilt on farm A and 3.9 m2/gilt on farm B. Although large interindividual differences existed, gilts from farm B performed numerically more aggressive attack (mean 1.12, SD 1.42 vs. mean 0.71, SD 1.20) and reciprocal fighting (mean 0.78, SD 0.98 vs. mean 0.44, SD 0.82) when compared with gilts from farm A. The heritabilities and additive genetic variances for behavioral traits were estimated with a linear animal model and were on a low level in farm A (h2 = 0.11, SE = 0.07, and σ2a = 0.12 for aggressive attack and h2 = 0.04, SE = 0.07, and σ2a = 0.02 for reciprocal fighting) and on a moderate level in farm B (h2 = 0.29, SE = 0.13, and σ2a = 0.44 for aggressive attack and h2 = 0.33, SE = 0.12, and σ2a = 0.27 for reciprocal fighting). For both aggressive attack and reciprocal fighting, genetic correlation of the same trait between farm A and farm B was 1.0. Therefore, aggressive behavior does not seem to be influenced by genotype × environment interactions. Under these circumstances aggressions in group housing can be reduced by genetic selection against aggressive behavior. Therewith, the welfare and health of sows will ultimately increase. INTRODUCTION The mixing of pigs into new social groups is a routine procedure in pig farming. In general, aggression among pigs occurs directly after mixing of unacquainted animals (Mount and Seabrook, 1993; D'Eath and Turner, 2009), and it has been suggested that it serves the purpose of establishing a stable rank order (Ewbank, 1976). Nearly all unfamiliar individuals are involved in agonistic interactions during this period (Mount and Seabrook, 1993), although Arey and Franklin (1995) suggest that less than half of the possible pairs of unfamiliar pigs engage in fights when mixed. The formation of a stable social hierarchy takes place within 48 h postmixing (Meese and Ewbank, 1973; Arey and Edwards, 1998). The time to establish a stable rank order could be slowed down by the presence of highly aggressive animals (Erhard et al., 1997). After the establishment of a stable rank hierarchy, encounters between group members take place only at a minimum level of aggression (Løvendahl et al., 2005). Most aggressive encounters, such as nonreciprocal aggression or reciprocal fighting, result in skin lesions and injuries to the animals involved. Social stress and fear generated by mixing (reviewed by Kongsted, 2004) can additionally cause immunosuppression (Tuchscherer and Manteuffel, 2000). Consequently, aggression and thus social stress lead to decreased welfare and reduced longevity as well as compromised productivity of the animals (reviewed by von Borell et al., 2007; Spoolder et al., 2009), thus also affecting the profitability of commercial pig production. Because of these undesirable effects of aggressive behavior, breeding for calm and nonaggressive individuals seems to be a reasonable strategy for reducing aggressive interactions. Aggressiveness in pigs is known to be repeatable over time and across different situations (Jensen et al., 2002; Janczak et al., 2003; D'Eath, 2004). Previous genetic studies on this subject estimated moderate heritabilities of traits related to aggressive behavior after mixing in pigs (Løvendahl et al., 2005; Hellbrügge, 2007; D'Eath et al., 2009; Turner et al., 2009). However, constraints include the need to standardize the environment to allow all animals an equal opportunity to show their phenotype (Turner et al., 2010b). Otherwise, the risk of genotype × environment interactions may affect the success of selection (Turner et al., 2010a). Forkman et al. (2007) criticized the absence of such standardized behavior tests and missing information about their robustness, that is, the lack of information regarding what aspects can be changed without affecting the validity of the test. Genotype × environment interactions are well known for a number of traits in pigs (e.g., Merks, 1989; van Diepen and Kennedy, 1989; Guy et al., 2002; Wood et al., 2004; Wallenbeck et al., 2009). However, there is little information about the influence of environments on the expression of aggressive behavior traits and its corresponding variance components in gilts. Therefore, the objective of the present paper was to investigate aggressive behavior traits at mixing of unfamiliar gilts of a genetically highly connected purebred Pietrain population, tested by the same person under 2 different housing conditions. MATERIALS AND METHODS All data collection was done by 1 person from March 2010 to January 2012 on 2 nucleus farms (farms A and B) of the German breeding company BHZP GmbH. Animal care followed the general guidelines outlined in the European animal welfare regulations. Animals and Housing Conditions A total of 543 purebred Pietrain (of MHS-genotype NN) gilts, composed of 302 gilts on farm A and 241 gilts on farm B, were tested. The gilts of both farms were genetically closely related; they were the offspring of 280 dams and 96 sires, with 64% of these sires having tested progeny on both farms. Pedigree information contained sires and dams of 4 generations. In total 1,516 animals appeared in the pedigree file. The tested gilts were of similar age (214 d, SD 12.2 d). Farm A. At the average age of 212 d (SD 14.4 d), 23 to 34 gilts (mean 28.2), which were designated for internal replacements, were moved from the rearing building to a single quarantine pen in the breeding unit. From group housing in the rearing unit, on average, 6 gilts (1 to 15) were acquainted with each other. Each group of newly mixed gilts in the quarantine pen was composed of purebred German Landrace, purebred Pietrain, and purebred Duroc gilts. Pietrain gilts made up an average of 41% (29% to 50%), German Landrace 40% (29% to 50%), and Duroc gilts 19% (12% to 28%) of each test group. The total space allowance of the quarantine pen, where the observation took place, was 72 m2. This implies a mean space allowance per gilt of 2.6 m2. The quarantine pen was constructed with a solid concrete floor littered with wooden shavings. The installed feeding system included 3 nipple drinkers and 1 dry feeder (Big Dutchman International GmbH, Vechta, Germany). Gilts were fed a commercial pelleted diet ad libitum (12.2 MJ ME). Farm B. At the average age of 217 d (SD 17.7 d), 14 to 22 gilts (mean 18.3), which were designated for internal replacements, were moved from the rearing building into a pen in the gestation building. From group housing in the rearing unit, on average, 3 gilts (1 to 7) were acquainted with each other. Each group of newly mixed gilts was composed of purebred Large White and purebred Pietrain gilts. Pietrain gilts made up an average of 42% (32% to 53%) of each test group. In the gestation building there were 4 pens with a total space allowance of 71 m2 each. The mean space allowance per gilt was 3.9 m2. All pens were constructed in the same way. Panels divided each pen into a slatted activity area and a solid floor resting area. Each gestation pen was equipped with 4 nipple drinkers and 1 electronic sow feeder (Mannebeck GmbH, Schüttdorf, Germany). For enrichment a scratch brush was installed in every pen. Animals were fed the same commercial pelleted diet (12.2 MJ ME) as on farm A. However, the amount was restricted to 2.8 kg feed per animal per day. Behavior Leaving the rearing building, gilts on both farms were moved to a testing pen in the breeding unit. In each case the continuous observation period started when the last gilt had entered the pen and lasted, adapted from Hellbrügge (2007), for 30 min. All gilts were individually marked by a double-sided, numbered ear tag. During the testing time the observer stood on farm A quietly inside the pen. On farm B the observer stood in an alley outside the pen, from where she had a good view of the whole test area. The test design was set up in a way to minimize labor costs and the disturbance of the daily workflow to meet requirements of commercial farming and breeding. Aggression. During the observation period all aggressive encounters between the gilts were recorded by the observer. Aggressive encounters of the newly mixed gilts were subdivided into nonreciprocal aggression and reciprocal fighting (Hoy, 2009). The delivery of nonreciprocal aggression was defined as an aggressive interaction, such as biting or snapping, directed toward another gilt (D'Eath et al., 2009; Tönepöhl et al., 2013). Gilts delivering nonreciprocal aggression to other gilts were recorded when the attacked gilt showed a submissive reaction, that is, turned away from the attacking gilt, fled, or was displaced from an area, instead of retaliating (Tuchscherer et al., 1998; Langbein and Puppe, 2004). The identity of the attacked gilt was not recorded. The trait reciprocal fighting was defined as bilateral aggression between a minimum of 2 gilts. Reciprocal fights were displayed as bodily attacks such as “head-to-head knocks,” “head-to-body knocks,” “parallel–inverse parallel pressings,” “biting,” or “physical displacements” (Puppe, 1998). For each reciprocal fight the involved gilts were noted. Gilts delivering nonreciprocal aggression to other gilts and gilts involved in reciprocal fighting received a score for the aggressive attack trait. Statistical Analysis The data analysis was performed with the statistical software SAS version 9.2 (SAS Inst. Inc., Cary, NC). For the analysis of the aggressive attack and reciprocal fighting traits the following fixed effects were considered: farm (farm A or B), test batch (defined as a group of gilts, whose behavior was observed within a 9-wk period; 10 classes; considered as a fixed rather than a random effect because of computational considerations), the proportion of Pietrain gilts in a group of tested gilts (4 classes: less than or equal to 35%, greater than 35% up to and including 40%, greater than 40% up to and including 45%, and greater than 45% Pietrains per tested group), the number of animals that were acquainted with each other from the rearing unit, and the age of the tested gilts. An observer effect was not considered because all data were collected by the same person. The significance of the fixed effects and their interactions were tested with the procedure MIXED (SAS Inst. Inc., Cary, NC.) Effects that were not significant were not included in the final models. These nonsignificant effects included, for example, the number of gilts that were acquainted with each other from the rearing unit. For a better allocation numbers greater than 4 for the behavior trait aggressive attack were combined and counted as 4, and the numbers greater than 3 for reciprocal fighting were combined and counted as 3 for further statistical analysis. The variance components and their corresponding ratios (heritabilities) of the behavior traits aggressive attack and reciprocal fighting were estimated univariately within a linear animal model using the VCE-4 package (Neumaier and Groeneveld, 1998). On the basis of the significance of fixed effects from the mixed model analysis, the animal model for the observed behavior traits from the combined analysis of both farms included the fixed combined farm × test batch effect, with y being the respective dependent variable and e being the random error:  For observed behavior traits estimated separately for each farm the animal model included the parameter test batch as fixed effect, with y being the respective dependent variable and e being the random error:  The genetic correlations between the behavior traits were estimated bivariately using VCE-4 (Neumaier and Groeneveld, 1998). RESULTS The number of aggressive attacks performed by a Pietrain gilt during the 30-min observation period varied in the combined analysis of both farms (n = 543) from 0 to 11, with a mean of 0.90 (SD 1.3). Pietrain gilts of both farms were involved in 0 to 7 reciprocal fights. The mean frequency of reciprocal fights was 0.59 (SD 0.9). Pietrain gilts from farm B performed numerically more aggressive attacks (mean 1.12, SD 1.4 vs. mean 0.71, SD 1.2) and reciprocal fighting (mean 0.78, SD 0.9 vs. mean 0.44, SD 0.8) after mixing with unacquainted animals compared to gilts from farm A. The investigated behavior traits showed a large interindividual variation, indicating that individual differences in aggressive behavior existed. Of all tested Pietrain gilts on farm A 60.3% did not exhibit aggressive attacks during observation time. In comparison Pietrain gilts from farm B were involved in a higher frequency of aggressive attack. Only 40.7% of these gilts were not involved in any aggressive attack. On farm A 29.8% of the Pietrain gilts took part in reciprocal aggression, whereas 51.4% of all Pietrain gilts on farm B were seen at least once participating in reciprocal fights. Fixed Effects In the combined analysis of both farms the behavior traits aggressive attack (P ≤ 0.01) and reciprocal fighting (P ≤ 0.001) were significantly affected by the fixed combined farm × test batch effect. Variance Components and Heritabilities The estimated heritabilities of the observed behavior traits for the combined analysis of both farms were h2 = 0.20 (SE = 0.06) for aggressive attack and h2 = 0.16 (SE = 0.06) for reciprocal fighting. The additive genetic variance was σ2a = 0.25 for aggressive attack and σ2a = 0.12 for reciprocal fighting estimated for gilts of both farms together. The estimates of the residual variance for the combined analysis of both farms were σ2e = 1.02 for aggressive attack and σ2e = 0.53 for reciprocal fighting. On farm A the estimated heritabilities were at a low level compared to values of a medium magnitude estimated on farm B. The variance components estimated separately for the 2 farms are shown in Table 1. Table 1. Heritabilities (±SE), additive genetic variance (σ2a), and residual variance (σ2e) of the analyzed behavior traits for mixed Pietrain gilts (combined analysis of both farms A and B)   Aggressive attack  Reciprocal fighting  Item  h2 ± SE  σ2a  σ2e  h2 ± SE  σ2a  σ2e  Combined analysis (n = 543)  0.20 ± 0.06  0.25  1.02  0.16 ± 0.06  0.12  0.53  Farm A (n = 302)  0.11 ± 0.07  0.12  0.97  0.29 ± 0.13  0.44  1.06  Farm B (n = 241)  0.04 ± 0.07  0.02  0.50  0.33 ± 0.12  0.27  0.55    Aggressive attack  Reciprocal fighting  Item  h2 ± SE  σ2a  σ2e  h2 ± SE  σ2a  σ2e  Combined analysis (n = 543)  0.20 ± 0.06  0.25  1.02  0.16 ± 0.06  0.12  0.53  Farm A (n = 302)  0.11 ± 0.07  0.12  0.97  0.29 ± 0.13  0.44  1.06  Farm B (n = 241)  0.04 ± 0.07  0.02  0.50  0.33 ± 0.12  0.27  0.55  View Large Table 1. Heritabilities (±SE), additive genetic variance (σ2a), and residual variance (σ2e) of the analyzed behavior traits for mixed Pietrain gilts (combined analysis of both farms A and B)   Aggressive attack  Reciprocal fighting  Item  h2 ± SE  σ2a  σ2e  h2 ± SE  σ2a  σ2e  Combined analysis (n = 543)  0.20 ± 0.06  0.25  1.02  0.16 ± 0.06  0.12  0.53  Farm A (n = 302)  0.11 ± 0.07  0.12  0.97  0.29 ± 0.13  0.44  1.06  Farm B (n = 241)  0.04 ± 0.07  0.02  0.50  0.33 ± 0.12  0.27  0.55    Aggressive attack  Reciprocal fighting  Item  h2 ± SE  σ2a  σ2e  h2 ± SE  σ2a  σ2e  Combined analysis (n = 543)  0.20 ± 0.06  0.25  1.02  0.16 ± 0.06  0.12  0.53  Farm A (n = 302)  0.11 ± 0.07  0.12  0.97  0.29 ± 0.13  0.44  1.06  Farm B (n = 241)  0.04 ± 0.07  0.02  0.50  0.33 ± 0.12  0.27  0.55  View Large Genetic Correlation between the Behavior Traits A high, positive genetic correlation (rg = 0.95, SE = 0.04) was found between the behavior traits aggressive attacks and reciprocal fighting for the combined analysis of both farms. The correlations between the behavior traits were, for both farms, at a high positive level with low standard errors (farm A: rg = 1.00, SE = 0.02; farm B: rg = 0.96, SE = 0.03). However, with these correlations, a part-whole relationship between the behavior traits aggressive attack and reciprocal fighting has to be considered; that is, they are expected to be high because gilts participating in reciprocal fighting were, by definition, also recorded for the behavior trait aggressive attack. Between the behavior traits estimated for farm A and the behavior traits estimated for farm B positive genetic correlations were estimated. The genetic correlations between the same trait measured on farm A and on farm B were complete (rg = 1.00) for both traits (aggressive attack and reciprocal fighting). DISCUSSION Behavior Traits and Effects The test setup for aggressive behavior at mixing of unfamiliar gilts was based on a study by Hellbrügge (2007). Turner et al. (2010a) stated that for economic reasons the phenotyping of a trait such as postmixing aggressiveness in pigs for selection purposes must take less than 2 min. The present study was designed in a way that behavior traits were easy to record with little extra time and work required. For example, the observation period was set to a total of 30 min per mixing of the unfamiliar gilts. Therefore, data collection of the behavior traits fit under production conditions, which is a prerequisite for a trait to be considered as a possible selection criterion in pig breeding. The level of aggression may vary among environments; this has been shown inter alia in mice (Haemisch et al., 1994), humans (Miles and Carey, 1997), and pigs (e.g., van de Weerd and Day, 2009). Krauss (2011) suggested that the pen design, space allowance per animal, management, and animal-related factors influence the frequency of aggressive interactions. For the combined analysis of both farms in the present study the combined effect of the farm × test batch showed considerable influence on the behavior traits. A key reason for this effect could be the differences in the housing systems or physical environment. The housing system of farms A and B differed regarding the feeder design, feeding regimen, flooring design, and mean space allowance per gilt. Although with the present study design it was, unfortunately, impossible to tease apart the impact of the individual components, each of these differences in housing conditions or combinations thereof may have resulted in the observed differences in the expression of agonistic behavior. For example, feed was available on farm A from early in the beginning of the test, and therefore, it could be assumed that the gilts' motivation to feed masked some of the motivation to fight. It is also important to consider that the gilts on farm A were familiar with the type of feeder in the quarantine pen, whereas the pens on farm B were equipped with electronic sow feeders, which were unfamiliar to the animals. Furthermore, the flooring system of the 2 farms differed considerably. Results of various studies differ regarding the effect of flooring system and bedding on aggressive interactions (Matthews and Ladewig, 1994; Andersen et al., 1999; Salaün et al., 2002). Generally, it is suggested that provision of bedding aids in reducing agonistic interactions after mixing of unacquainted pigs (Spoolder et al., 2009). These findings are in line with results from the present study: gilts bedded with wooden shavings (farm A) showed a lower level of aggression compared to those kept without bedding (farm B). Numerous studies reported decreased aggression in connection with increasing space allowance as well as visual barriers (Weng et al., 1998; Barnett et al., 2001; Stukenborg, 2011). Nevertheless, compared to gilts on farm A, gilts on farm B were more aggressive during the observation period despite a slightly greater space allowance and the presence of visual barriers. However, in the present study gilts on either farm were offered ample space, exceeding the space allowance required by law. In addition to differences in the physical environment there are likewise differences in the social environment of the gilts. For pigs group size has an effect on the level of aggressive behavior (Rodenburg and Koene, 2007). Arey and Franklin (1995) reported that the amount of fighting in newly mixed growing pigs increased with the number of unfamiliar pigs in the pen. In a study by Turner et al. (2001) pigs from groups of 80 were less aggressive to unacquainted conspecifics than pigs from groups of 20 individuals, where they could still recognize familiar animals. However, more fights occurred after mixing in groups of 6 and 12 animals than in groups of 24 pigs, and the percentage of pigs participating in reciprocal fighting was lower in the largest group (Andersen et al., 2004). The quality of human-animal interactions can limit the productivity and welfare of these animals. Because of former positive human contact and gentle handling, gilts on farm A were not afraid of the presence of the quietly standing observer. Hemsworth et al. (1986) observed that pigs approached the experimenter significantly more often when he stood passively in the test area than when the pigs were actively approached. The design of the present study did not allow for a separate investigation of the effects of differences in feeding management, bedding, space allowance, and group composition. Therefore, no conclusions about the relative contribution of the different factors can be drawn. Some authors suggest that a standardized environment, which allows all animals equal opportunities to show their phenotype, is a basic prerequisite for successful phenotyping of traits (Turner et al., 2010b). It might be that not all gilts had the opportunity to show their “real” phenotype, or in contrast, the absence of a familiar feeder and bare food may have led to higher levels of agonistic interactions. Variance Components The present study showed differences in the heritability of the behavior traits aggressive attack and reciprocal fighting estimated for a genetically closely linked Pietrain population, tested under 2 different housing conditions. The residual variances for aggressive attack and reciprocal fighting estimated for farm A were in the range of the estimated values of residual variance from farm B. In contrast, the values of additive genetic variance tended to differ between the 2 farms. For this reason the reduced phenotypic variance and the marked differences in the estimated heritabilities between farms A and B seem to be caused by lower additive genetic variance in farm A. The moderate heritabilities of aggressive attack and reciprocal fighting estimated on farm B were in the range of the estimated heritabilities for offensive aggression of sows in groups (h2 = 0.17 to 0.24; Løvendahl et al., 2005). Turner et al. (2006) assessed postmixing aggressiveness of pigs by using an approach based on skin lesion scores. The authors calculated heritabilities (h2 = 0.11 to 0.22) that are within the range of the results of the present study. A moderate heritability for aggressive behavior of purebred Landrace sows in a provoked stress situation was calculated by Hellbrügge (2007). Moderate to high heritabilities were found for the delivery of nonreciprocal aggression (h2 = 0.31 and 0.34, respectively) and reciprocal aggression (h2 = 0.43 and 0.47, respectively) in growing pigs (D'Eath et al., 2009; Turner et al., 2009). In agreement with the results from farm A in the present study, Stukenborg (2011) estimated heritabilities of low magnitude for agonistic behavior traits of gilts, whereas Stukenborg (2011) found moderate heritabilities for participating in fights in growing pigs. Lush (1945) recommended that animals be kept for selection in environments in which they will be used in practice so that preferable genes have a chance to express their effects. In contrast, other authors propose that selection should be practiced in the most favorable environment to improve the accuracy of selection due to greater expression of genes of interest (Hammond, 1947). Superior environmental conditions are expected to result in improved production traits at the phenotypic level, allowing for better distinction between animals' production potential at the genetic level. However, in the present study there were lower additive genetic variances under the superior housing conditions, indicating that the assumption by Hammond (1947) may not hold for all types of traits. In particular, traits with undesirable effects on production parameters, such as detrimental behavior or susceptibility to diseases, can be expected to be better expressed under challenging housing conditions that are suboptimal rather than superior from a production point of view. Falconer (1952) introduced the concept of a genetic correlation between performances in different environments and used the ratio of indirect and direct responses to selection to determine the optimum environment for selection. The direction in which environment affects aggression, however, appears to differ between and within species (Haemisch et al., 1994; van Loo et al., 2002; van de Weerd and Day, 2009). The results of the present study emphasize that it is difficult to compare results of different studies, not only because of varying definition of the traits, experimental setups, and analysis methods but also because of the pronounced effect of management on the phenotypic expression of animals' genetic potential. Because of the differences in expression of genetic potential detected in the present study a challenging testing area appears to be required to obtain a good expression of the observed traits, as otherwise genetic differences may be hidden. Genetic Correlations For gilts from both farms high, positive genetic correlations could be found between the behavior traits aggressive attack and reciprocal fighting. Despite the part-whole relationship between the 2 behavior traits, it can be concluded that the strongly associated behavior traits share a common genetic basis. A high, positive genetic correlation, indicating that reciprocal aggression and delivery of nonreciprocal aggression in pigs share a similar genetic background, was found in several studies in pigs (Turner et al., 2008, 2009; D'Eath et al., 2009) as well as other species, such as mice (van Oortmerssen et al., 1984). A positive genetic correlation (rg = 0.28 to 0.50; Turner et al., 2009) between skin lesion scores 24 h and 3 wk after mixing indicates that selection to reduce aggression at mixing would also reduce aggression in the weeks after group formation, suggesting that aggressiveness may be reduced in several contexts rather than that specifically in response to regrouping (Rodenburg and Turner, 2012). Therefore, it could be deduced that selection for reduced aggression in gilts also has a positive effect on group behavior of older sows. The high genetic correlations between the same traits measured on the 2 different farms imply that although the behavior traits were observed under different housing conditions, they seem to be in part genetically controlled by the same genes. Because of the small sample size in the present study it can be assumed that the genetic correlations will decrease if a larger amount of behavioral data is available. Nevertheless, it can be expected that the genetic correlations will still remain at a high level. The concept of a genetic correlation between performances in different environments can be used as a measure of ranking difference due to genotype × environment interactions (Falconer, 1952). Robertson (1959) considered a genetic correlation of rg = 0.8 as a limit; values below this limit were associated with genotype-environment interactions. On the basis of these values there seem to be no appreciable genotype × environment interactions in the present study. From this it even be deduced that the selection of parents in 1 environment may also optimize progeny behavior in another environment. Low to moderate heritabilities could be estimated for the behavior traits aggressive attack and reciprocal fighting in newly mixed Pietrain gilts. A clear difference in the additive genetic variance could be found between the 2 farms, despite a genetically closely linked population and utilization of the same behavior traits and even the same observer. Selection for reduced aggression in group-housed animals seems to be feasible and desirable to improve the welfare of the animals. Nevertheless, before observed behavior traits can be included in the breeding criteria, it will be important to observe possible genetic associations with other traits with welfare or economic significance. LITERATURE CITED Andersen I. L. Bøe K. E. Kristiansen A. L. 1999. The influence of different feeding arrangements and food type on competition at feeding in pregnant sows. Appl. Anim. Behav. Sci.  65: 91– 104. Google Scholar CrossRef Search ADS   Andersen I. L. Nævdal E. Bakken M. Bøe K. E. 2004. Aggression and group size in domesticated pigs, Sus scrofa: When the winner takes it all and the loser is standing small. Anim. Behav.  68: 965– 975. Google Scholar CrossRef Search ADS   Arey D. S. Edwards S. A. 1998. Factors influencing aggression between sows after mixing and the consequences for welfare and production. Livest. Prod. Sci.  56: 61– 70. Google Scholar CrossRef Search ADS   Arey D. S. Franklin M. F. 1995. Effects of straw and unfamiliarity on fighting between newly mixed growing pigs. Appl. Anim. Behav. Sci.  45: 23– 30. Google Scholar CrossRef Search ADS   Barnett J. L. Hemsworth P. H. Cronin G. M. Jongman E. C. Hutson G. D. 2001. A review of the welfare issues for sows and piglets in relation to housing. Aust. J. Agric. Res.  52: 1– 28. Google Scholar CrossRef Search ADS   D'Eath R. B. 2004. Consistency of aggressive temperament in domestic pigs: The effects of social experience and social disruption. Aggress. Behav.  30: 435– 448. Google Scholar CrossRef Search ADS   D'Eath R. B. Roehe R. Turner S. P. Ison S. H. Farish M. Jack M. C. Lawrence A. B. 2009. Genetics of animal temperament: Aggressive behaviour at mixing is genetically associated with the response to handling in pigs. Animal  3: 1544– 1554. Google Scholar CrossRef Search ADS PubMed  D'Eath R. B. Turner S. P. 2009. The natural behaviour of the pig. In: Marchant-Forde J. N. editor, The welfare of pigs. Vol. 1.  Springer, Heidelberg, Germany. p. 13– 45. Google Scholar CrossRef Search ADS   Erhard H. W. Mendl M. Ashley D. D. 1997. Individual aggressiveness of pigs can be measured and used to reduce aggression after mixing. Appl. Anim. Behav. Sci.  54: 137– 151. Google Scholar CrossRef Search ADS   Ewbank R. 1976. Social hierarchy in suckling and fattening pigs: A review. Livest. Prod. Sci.  3: 363– 372. Google Scholar CrossRef Search ADS   Falconer D. S. 1952. The problem of environment and selection. Am. Nat.  86: 293– 298. Google Scholar CrossRef Search ADS   Forkman B. Boissy A. Meunier-Salaün M. C. Canali E Jones R. B. 2007. A critical review of fear tests used on cattle, pigs, sheep, poultry and horses. Physiol. Behav.  92: 340– 374. Google Scholar CrossRef Search ADS PubMed  Guy J. H. Rowlinson P. Chadwick J. P. Ellis M. 2002. Health conditions of two genotypes of growing-finishing pig in three different housing systems: Implications for welfare. Livest. Prod. Sci.  75: 233– 243. Google Scholar CrossRef Search ADS   Haemisch A. Voss T. Gärtner K. 1994. Effects of environmental enrichment on aggressive behavior, dominance hierarchies, and endocrine states in male DBA/2J mice. Physiol. Behav.  56: 1041– 1048. Google Scholar CrossRef Search ADS PubMed  Hammond J. 1947. Animal breeding in relation to nutrition and environmental conditions. Biol. Rev. Camb. Philos. Soc.  22: 195– 213. Google Scholar CrossRef Search ADS PubMed  Hellbrügge B. 2007. Genetic aspects of piglet losses and the maternal behaviour of sows. PhD Diss. Christian-Albrechts-Univ., Kiel, Germany. Hemsworth P. H. Barnett J. L. Hansen C. 1986. The influence of handling by humans on the behaviour, reproduction and corticosteroids of male and female pigs. Appl. Anim. Behav. Sci.  15: 303– 314. Google Scholar CrossRef Search ADS   Hoy S. 2009. Verhalten der Schweine. In: Hoy S. editor, Nutztierethologie. Verlag Eugen Ulmer, Stuttgart, Germany. p. 105– 139. Janczak A. M. Pedersen L. J. Bakken M. 2003. Aggression, fearfulness and coping styles in female pigs. Appl. Anim. Behav. Sci.  81: 13– 28. Google Scholar CrossRef Search ADS   Jensen K. H. Nielsen B. L. Olsen A. N. W. 2002. Results of experiments on development of test for identification of social traits in sows, with a view to predict risk of high aggression level in a group. DIAS Internal Rep. No. 155.  Danmarks Jordbrugsforskning, Foulum, Denmark. Kongsted A. G. 2004. Stress and fear as possible mediators of reproduction problems in group housed sows: A review. Acta Agric. Scand. A Anim. Sci.  54: 58– 66. Krauss A. V. 2011. Sozialverhalten in dynamischen Sauengruppen bei der Eingliederung neuer Sauen, PhD Diss. Justus-Liebig-Univ., Gießen, Germany. Langbein J. Puppe B. 2004. Analysing dominance relationships by sociometric methods—A plea for a more standardised and precise approach in farm animals. Appl. Anim. Behav. Sci.  87: 293– 315. Google Scholar CrossRef Search ADS   Løvendahl P. Damgaard L. H. Nielsen B. L. Thodberg K. Su G. Rydhmer L. 2005. Aggressive behaviour of sows at mixing and maternal behaviour are heritable and genetically correlated traits. Livest. Prod. Sci.  93: 73– 85. Google Scholar CrossRef Search ADS   Lush J. L. 1945. Animal breeding plans. Iowa State College Press, Ames. Matthews L. R. Ladewig J. 1994. Environmental requirements of pigs as measured by behavioral demand functions. Anim. Behav.  47: 713– 719. Google Scholar CrossRef Search ADS   Meese G. B. Ewbank R. 1973. The establishment and nature of the dominance hierarchy in the domesticated pig. Anim. Behav.  21: 326– 334. Google Scholar CrossRef Search ADS   Merks J. W. M. 1989. Genotype × environment interactions in pig breeding programmes. VI. Genetic relations between performances in central test, on-farm test and commercial fattening. Livest. Prod. Sci.  22: 325– 339. Google Scholar CrossRef Search ADS   Miles D. R. Carey G. 1997. Genetic and environmental architecture on human aggression. J. Pers. Soc. Psychol.  72: 207– 217. Google Scholar CrossRef Search ADS PubMed  Mount N. C. Seabrook M. F. 1993. A study of aggression when group housed sows are mixed. Appl. Anim. Behav. Sci.  36: 377– 383. Google Scholar CrossRef Search ADS   Neumaier A. Groeneveld E. 1998. Restricted maximum likelihood estimation of covariances in sparse linear models. Genet. Sel. Evol.  30: 3– 26. Google Scholar CrossRef Search ADS   Puppe B. 1998. Effects of familiarity and relatedness on agonistic pair relationships in newly mixed domestic pigs. Appl. Anim. Behav. Sci.  58: 233– 239. Google Scholar CrossRef Search ADS   Robertson A. 1959. The sampling variance of the genetic correlation coefficient. Biometrics  15: 469– 485. Google Scholar CrossRef Search ADS   Rodenburg T. B. Koene P. 2007. The impact of group size on damaging behaviours, aggression, fear and stress in farm animals. Appl. Anim. Behav. Sci.  103: 205– 214. Google Scholar CrossRef Search ADS   Rodenburg T. B. Turner S. P. 2012. The role of breeding and genetics in the welfare of farm animals. Anim. Front.  2( 3): 16– 21. Google Scholar CrossRef Search ADS   Salaün C. Callarec J. Toudic M. Dréan E. 2002. Effet du type de sol sur le bienêtre des truis gestantes en groupe alimentées au distributeur automatique de concentré (DAC). J. Rech. Porcine  34: 217– 223. Spoolder H. A. M. Geudeke M. J. van der Peet-Schwering C. M. C. Soede N. M. 2009. Group housing of sows in early pregnancy: A review of success and risk factors. Livest. Sci.  125: 1– 14. Google Scholar CrossRef Search ADS   Stukenborg A. 2011. Investigations on agonistic behaviour in pigs kept under commercial farm conditions. PhD Diss. Christian-Albrechts-Univ., Kiel, Germany. Tönepöhl B. Appel A. K. Voß B. König von Borstel U. Gauly M. 2013. Interaction between sows' aggressiveness post mixing and skin lesions recorded several weeks later. Appl. Anim. Behav. Sci.  144: 108– 115. Google Scholar CrossRef Search ADS   Tuchscherer M. Manteuffel G. 2000. The effect of psycho stress on the immune system. Another reason for pursuing animal welfare. Arch. Tierz.  43: 547– 560. Tuchscherer M. Puppe B. Tuchscherer A. Kanitz E. 1998. Effects of social status after mixing on immune, metabolic, and endocrine responses in pigs. Physiol. Behav.  64: 353– 360. Google Scholar CrossRef Search ADS PubMed  Turner S. P. D'Eath R. B. Roehe R. Lawrence A. B. 2010a. Selection against aggressiveness in pigs at re-grouping: Practical application and implications for long-term behavioural patterns. Anim. Welf.  19( S): 123– 132. Turner S. P. Horgan G. W. Edwards S. A. 2001. Effect of social group size on aggressive behaviour between unacquainted domestic pigs. Appl. Anim. Behav. Sci.  74: 203– 215. Google Scholar CrossRef Search ADS   Turner S. P. Roehe R. D'Eath R. B. Ison S. H. Farish M. Jack M. C. Lundeheim N. Rydhmer L. Lawrence A. B. 2009. Genetic validation of post-mixing skin injuries in pigs as an indicator of aggressiveness and the relationship with injuries under more stable social conditions. J. Anim. Sci.  87: 3076– 3082. Google Scholar CrossRef Search ADS PubMed  Turner S. P. Roehe R. Lawrence A. B. 2010b. Social behaviour in pigs (ID060). In: 9th World Congr. Genet. Appl. Livest. Prod., Leipzig, Germany. http://www.kongressband.de/wcgalp2010/assets/pdf/0060.pdf (verified 11/1/2013). Turner S. P. Roehe R. Mekkawy W. Farnworth M. J. Knap P. W. Lawrence A. B. 2008. Bayesian analysis of genetic associations of skin lesions and behavioural traits to identify genetic components of individual aggressiveness in pigs. Behav. Genet.  38: 67– 75. Google Scholar CrossRef Search ADS PubMed  Turner S. P. White I. M. S. Brotherstone S. Farnworth M. J. Knap P. W. Penny P. Mendl M. Lawrence A. B. 2006. Heritability of post-mixing aggressiveness in grower-stage pigs and its relationship with production traits. Anim. Sci.  82: 615– 620. Google Scholar CrossRef Search ADS   van de Weerd H. A. Day J. E. L. 2009. A review of environmental enrichment for pigs housed in intensive housing systems. Appl. Anim. Behav. Sci.  116: 1– 20. Google Scholar CrossRef Search ADS   van Diepen T. A. Kennedy B. W. 1989. Genetic correlations between test station and on-farm performance for growth rate and backfat in pigs. J. Anim. Sci.  67: 1425– 1431. Google Scholar CrossRef Search ADS PubMed  van Loo P. L. P. Kruitwagen C. L. J. J. Koolhaas J. M. van de Weerd H. A. van Zutphen L. F. M. Baumans V. 2002. Influence of cage enrichment on aggressive behaviour and physiological parameters in male mice. Appl. Anim. Behav. Sci.  76: 65– 81. Google Scholar CrossRef Search ADS   van Oortmerssen G. A. Benus I. Dijk D. J. 1984. Studies in wild house mice: Genotype-environment interactions for attack latency. Neth. J. Zool.  35: 155– 169. Google Scholar CrossRef Search ADS   von Borell E. Dobson H. Prunier A. 2007. Stress, behaviour and reproductive performance in female cattle and pigs. Horm. Behav.  52: 130– 138. Google Scholar CrossRef Search ADS PubMed  Wallenbeck A. Rydhmer L. Lundeheim N. 2009. GxE interactions for growth and carcass leanness: Re-ranking of boars in organic and conventional pig production. Livest. Sci.  123( 2): 154– 160. Google Scholar CrossRef Search ADS   Weng R. C. Edwards S. A. English P. R. 1998. Behaviour, social interactions and lesion scores of group-housed sows in relation to floor space allowance. Appl. Anim. Behav. Sci.  59: 307– 316. Google Scholar CrossRef Search ADS   Wood J. D. Nute G. R. Richardson R. I. Whittington F. M. Southwood O. Plastow G. Mansbridge R. da Costa N. Chang K. C. 2004. Effects of breed, diet and muscle on fat deposition and eating quality in pigs. Meat Sci.  67: 651– 667. Google Scholar CrossRef Search ADS PubMed  Footnotes 1 This study was performed within the project “Investigations of possibilities to integrate behavior traits in pig breeding programmes” funded by the Federal Ministry of Food, Agriculture and Consumer Protection (BMELV) via the Federal Office for Agriculture and Food (BLE) in the framework of the program “Innovation facilitation.” We thank the staff of BHZP GmbH for support during the recording of the data. American Society of Animal Science

Dufrasne, M.;Misztal, I.;Tsuruta, S.;Holl, J.;Gray, K. A.;Gengler, N.

doi: 10.2527/jas.2013-6684pmid: 24146157

Abstract Genetic parameters for birth weight (BWT), preweaning mortality (PWM), and HCW were estimated for a crossbred pig population to determine if BWT could be used as an early predictor for later performances. Sire genetic effects for those traits were estimated to determine if early selection of purebred sires used in crossbreeding could be improved. Data were recorded from 1 commercial farm between 2008 and 2010. Data were from 24,376 crossbred pigs from Duroc sires and crossbred Large White × Landrace dams and included 24,376 BWT and PWM records and 13,029 HCW records. For the analysis, PWM was considered as a binary trait (0 for live or 1 for dead piglet at weaning). A multitrait threshold-linear animal model was used, with animal effect divided into sire genetic and dam effects; the dam effects included both genetic and environmental variation due to the absence of pedigree information for crossbred dams. Fixed effects were sex and parity for all traits, contemporary groups for BWT and HCW, and age at slaughter as a linear covariable for HCW. Random effects were sire additive genetic, dam, litter, and residual effects for all traits and contemporary group for PWM. Heritability estimates were 0.04 for BWT, 0.02 for PWM, and 0.12 for HCW. The ratio between sire genetic and total estimated variances was 0.01 for BWT and PWM and 0.03 for HCW. Dam and litter variances explained, respectively, 14% and 15% of total variance for BWT, 2% and 10% for PWM, and 3% and 8% for HCW. Genetic correlations were −0.52 between BWT and PWM, 0.55 between BWT and HCW, and −0.13 between PWM and HCW. Selection of purebred sires for higher BWT of crossbreds may slightly improve survival until weaning and final market weight at the commercial level. INTRODUCTION Economic gain in commercial swine production results mainly from the sale of pigs reaching full market value. Sow prolificacy has been emphasized in many breeding programs with the intent of increasing the number of pigs reaching full market value while maintaining the same number of sows within the herd. This selection objective has resulted in a significant increase in the number of pigs born alive per litter. As a result of increased litter size, there has been a decrease in individual pig birth weight (BWT; Quiniou et al., 2002). Several studies have reported that piglet BWT is related to performances (e.g., piglet survival, growth rate, carcass composition, meat quality) and therefore is an important economic trait in pig production. Low BWT is related to a higher preweaning mortality (PWM), reduced weight gain from weaning to the finishing period, and a fatter carcass (Fix et al., 2010). Therefore, pigs with low BWT require more days on feed to reach market weight and potentially produce a lower-quality carcass (Gondret et al., 2005; Bérard et al., 2008; Rehfeldt et al., 2008; Fix et al., 2010). Economic loss associated with low BWT may be attributed to inefficient subsequent performance throughout the fattening period. However, selection for higher piglet BWT should be implemented carefully because of the negative relationship between BWT and litter size, which results in the necessity to select both traits simultaneously (Fix et al., 2010). The ability to make genetic improvement at the commercial level depends heavily on selection programs implemented on purebred lines at the nucleus level. Therefore, the genetic influence of both purebred parents on the commercial performance of crossbred progeny must be determined. Piglet BWT and PWM are strongly influenced by maternal effects (Arango et al., 2006); therefore, genetic parameters for these traits have historically been estimated from dam components, although a genetic effect of the sire is possible (Knol et al., 2002; Hamann et al., 2004). Knowledge of paternal genetic effects and heritabilities for BWT and survival could have a large economic impact if the inclusion of a paternal component is beneficial to implement within a commercial pig breeding program (Hamann et al., 2004). Moreover, a better understanding of the paternal genetic effect on piglet traits may lead to the ability to identify sires at an earlier age. The objective of this study was to estimate genetic parameters for BWT, PWM, and HCW for commercial crossbred pigs to assess the influence of paternal effects and to determine if BWT could be a good early predictor for subsequent performances. MATERIALS AND METHODS Data Animal Care and Use Committee approval was not obtained for this study because data were obtained from an existing database. Data were provided by Smithfield Premium Genetics (Rose Hill, NC). After discarding records with incomplete or inconsistent data, information recorded from 2008 through 2010 on 1 commercial farm was available for 24,376 crossbred pigs. Crossbred animals were produced from the mating of purebred Duroc boars with crossbred Large White × Landrace sows. Pedigree data were not available for crossbred dams. A description of the data is shown in Table 1. Piglet BWT and PWM status (dead or alive) were available for the 24,376 animals. The piglet BWT was recorded within 24 h of birth on the commercial farm. Of those pigs, 13,029 had subsequent HCW records. Mean age at slaughter was 192 ± 12 d. Pedigrees were traced back 2 generations, and a total of 26,136 animals and 2,016 litters were included. A total of 193 different sires and 1,671 dams had progeny with a recorded BWT and PWM status, and 191 sires and 1,639 dams had progeny with HCW. Distributions of records by sire and dam family are shown in Tables 2 and 3, respectively. On average, each sire was mated with 8.7 dams, and each dam had 1.2 litters. Sows had records for ≤10 parities, but records for parities 7 through 10 (<7% of records) were grouped together. The mean number of parity was 3.5 ± 1.9. Among the 1671 dams, 1413 had pigs recorded in parity >1. Contemporary groups (15) were defined on the basis of piglets born during the same year and month. Table 1. Description of data Item  Value  No. of records  24,376  No. of animals in pedigree  26,136  No. of litters  2,016  No. of dams  1,671  No. of sires  193  No. of contemporary groups  15  Parity        Mean  3.54      SD  1.90  Age at slaughter, d        Mean  192.37      SD  12.12  Item  Value  No. of records  24,376  No. of animals in pedigree  26,136  No. of litters  2,016  No. of dams  1,671  No. of sires  193  No. of contemporary groups  15  Parity        Mean  3.54      SD  1.90  Age at slaughter, d        Mean  192.37      SD  12.12  View Large Table 1. Description of data Item  Value  No. of records  24,376  No. of animals in pedigree  26,136  No. of litters  2,016  No. of dams  1,671  No. of sires  193  No. of contemporary groups  15  Parity        Mean  3.54      SD  1.90  Age at slaughter, d        Mean  192.37      SD  12.12  Item  Value  No. of records  24,376  No. of animals in pedigree  26,136  No. of litters  2,016  No. of dams  1,671  No. of sires  193  No. of contemporary groups  15  Parity        Mean  3.54      SD  1.90  Age at slaughter, d        Mean  192.37      SD  12.12  View Large Table 2. Distribution of records by sire family for birth weight (BWT), preweaning mortality (PWM), and HCW Trait  Mean  SD  Minimum  Maximum  BWT (n = 193)  126.30  118.72  6  741  PWM (n = 193)  126.30  118.72  6  741  HCW (n = 191)  68.21  67.49  2  398  Trait  Mean  SD  Minimum  Maximum  BWT (n = 193)  126.30  118.72  6  741  PWM (n = 193)  126.30  118.72  6  741  HCW (n = 191)  68.21  67.49  2  398  View Large Table 2. Distribution of records by sire family for birth weight (BWT), preweaning mortality (PWM), and HCW Trait  Mean  SD  Minimum  Maximum  BWT (n = 193)  126.30  118.72  6  741  PWM (n = 193)  126.30  118.72  6  741  HCW (n = 191)  68.21  67.49  2  398  Trait  Mean  SD  Minimum  Maximum  BWT (n = 193)  126.30  118.72  6  741  PWM (n = 193)  126.30  118.72  6  741  HCW (n = 191)  68.21  67.49  2  398  View Large Table 3. Distribution of records by dam family for birth weight (BWT), preweaning mortality (PWM), and HCW Trait  Mean  SD  Minimum  Maximum  BWT (n = 1671)  14.59  7.19  1  61  PWM (n = 1671)  14.59  7.19  1  61  HCW (n = 1639)  7.95  4.63  1  36  Trait  Mean  SD  Minimum  Maximum  BWT (n = 1671)  14.59  7.19  1  61  PWM (n = 1671)  14.59  7.19  1  61  HCW (n = 1639)  7.95  4.63  1  36  View Large Table 3. Distribution of records by dam family for birth weight (BWT), preweaning mortality (PWM), and HCW Trait  Mean  SD  Minimum  Maximum  BWT (n = 1671)  14.59  7.19  1  61  PWM (n = 1671)  14.59  7.19  1  61  HCW (n = 1639)  7.95  4.63  1  36  Trait  Mean  SD  Minimum  Maximum  BWT (n = 1671)  14.59  7.19  1  61  PWM (n = 1671)  14.59  7.19  1  61  HCW (n = 1639)  7.95  4.63  1  36  View Large Statistical Analysis An animal model was used to estimate genetic parameters. To separate the animal additive genetic effects into sire and dam components as in Zumbach et al. (2007), a model with sire additive genetic effects and dam effects was considered. This allowed the estimation of sire genetic covariance based on their crossbred progeny. Moreover, this model was better adapted as the dam effect had to include both genetic and environmental variations due to the lack of pedigree information for the crossbred dams. The equation for the general multiple-trait model was  where y is a vector of observations (BWT, PWM status, or HCW), β is a vector of fixed effects, s is a vector of additive genetic effects of the sire, d is a vector of dam effects composed of dam additive genetic effects and dam environmental effects, l is a vector of common litter effects assigned by litter of the dam and assumed to be uncorrelated, c is a vector of random contemporary group effect, X, Z, U, Q, and W are incidence matrices that relate observations to effects, and e is a vector of residual effects. For the observed traits, BWT and HCW were continuous, but PWM status was a binary trait (0 if the piglet was still alive at weaning or 1 if the piglet died before weaning). Fixed effects were sex and parity number for all traits. Contemporary groups were fitted as a fixed effect for BWT and HCW but as a random effect for PWM status to avoid the “extreme category problem” that would occur with contemporary groups with no dead piglets at weaning (Misztal et al., 1989). Age at slaughter was included as linear covariable for HCW only. For all traits, sire additive genetic, dam, common litter, and residual effects were included as random effects. In this model, the animal additive effect is partitioned into sire additive genetic effect, dam additive genetic effect included in the dam effect, and Mendelian sampling included in the residuals. The variance of the sire genetic effects describes 1/4 of the total additive genetic variance and represents the genetic component of the model. The residual variance for the binary trait was fixed to 1. The (co)variance matrices were assumed to be        where A is the additive relationship matrix and I is an identity matrix; traits 1, 2, and 3 refer to BWT, PWM, and HCW, respectively. Estimations of (co)variance components were obtained with a Gibbs sampling algorithm, using the THRGIBBSF90 program (Misztal et al., 2002; Montpellier, France) with flat priors for (co)variances. This program allows the estimation of (co)variance components and genetic parameters in threshold mixed models with combinations of categorical and continuous traits (Lee et al., 2002). The program POSTGIBBSF90 (Misztal et al., 2002; Montpellier, France) was used for post-Gibbs analysis. A single chain of 250,000 cycles with a burn-in of the first 50,000 iterations was run for the analysis. The stationary stage was confirmed by graphical inspection of plots of sampled values vs. iterations. Every 10th sample was retained to compute mean and SE, obtained as SD of the posterior distribution. Starting values for (co)variance components were obtained from preliminary analyses using linear models implemented with restricted maximum likelihood and bivariate threshold-linear analyses. RESULTS AND DISCUSSION The mean BWT of 1.40 kg with SD = 0.32 kg was similar to mean BWT reported in other studies (Grandinson et al., 2002; Knol et al., 2002; Arango et al., 2006; Fix et al., 2010). The mean HCW of 93.37 kg with a SD = 8.60 kg is in agreement with HCW reported by Fix et al. (2010) but is somewhat higher than the final weight reported by Zumbach et al. (2007) in similar crossbred populations. On average, litter had 12.1 ± 3.8 piglets born alive. The PWM rate for all piglets was 16.99%, rising from 15.26% in parity 1 to 19.58% in parity 7 or later. The PWM rate of 16.99% is higher than the rate of 11.8% reported by Arango et al. (2006) for piglets that were alive after birth but is similar to PWM rates in other studies (e.g., Knol et al., 2002; Quiniou et al., 2002; Cecchinato et al., 2010). In this study, piglets from parities up to 10 were included, with 20% of piglets from parities higher than 5, whereas Arango et al. (2006) reported only 4% of piglets were represented for parities of ≥5. Moreover, they found that PWM rate increased only for parities of ≥7. Therefore, the higher PWM rate in this study could be the result of the larger proportion of piglets from later parities. Phenotypic correlations were −0.25 between BWT and PWM, 0.20 between BWT and HCW, and −0.48 between PWM and HCW. As expected, PWM rate decreased as BWT increased, as shown in Fig. 1. Figure 1. View largeDownload slide Relationship of preweaning mortality (PWM) rate and birth weight (BWT). Figure 1. View largeDownload slide Relationship of preweaning mortality (PWM) rate and birth weight (BWT). Estimates of variance components for BWT, PWM, and HCW are in Table 4. Histograms of posterior distributions of estimated (co)variance components (figure not shown) were quasi-normal for all traits, and the Geweke test did not detect any lack of convergence. Estimated sire genetic variance was small for each trait (0.001 for BWT, 0.006 for PWM, and 2.028 for HCW). Estimated sire covariances were negative between BWT and PWM (−0.001) and between PWM and HCW (−0.014). Estimated sire covariance was positive between BWT and HCW (0.025). Estimated dam and litter variances were similar for BWT (0.015) and were higher than estimated sire variance. For PWM, estimated litter variance was higher than estimated sire variance; moreover, both were higher than estimated dam variance. For HCW, estimated dam variance was on the same order as sire variance but lower than estimated litter variance. Estimated dam covariances had the same sign than estimated sire covariances. Estimated residual variances were high compared to other variance components for each trait (0.070 for BWT, 1.000 for PWM, and 56.228 for HCW). The relatively high residual variances could be due to the variation of the Mendelian sampling, which is not included directly in the model (Zumbach et al., 2007). Therefore, the Mendelian sampling becomes part of the residual and contributes to the increasing of the estimated residual variance. Table 4. Estimates (SE) of (co)variances for sire genetic, contemporary group, dam, litter, and residual effects for birth weight (BWT), preweaning mortality (PWM), and HCW of crossbred pigs1 Effect  Trait  BWT  PWM  HCW  Sire genetic  BWT  0.001 (0.0004)  −0.001 (0.0009)  0.025 (0.0091)    PWM    0.064 (0.0039)  −0.014 (0.0275)    HCW      2.028 (0.4518)  Contemporary group  BWT        PWM    0.028 (0.0278)      HCW        Dam  BWT  0.015 (0.0017)  −0.015 (0.0056)  0.081 (0.0346)    PWM    0.039 (0.0143)  −0.077 (0.0502)    HCW      2.209 (0.7247)  Litter  BWT  0.015 (0.0015)        PWM    0.102 (0.0165)      HCW      4.914 (0.7987)  Residual  BWT  0.070 (0.0007)        PWM    1.000 (0.0080)      HCW      56.228 (0.7522)  Effect  Trait  BWT  PWM  HCW  Sire genetic  BWT  0.001 (0.0004)  −0.001 (0.0009)  0.025 (0.0091)    PWM    0.064 (0.0039)  −0.014 (0.0275)    HCW      2.028 (0.4518)  Contemporary group  BWT        PWM    0.028 (0.0278)      HCW        Dam  BWT  0.015 (0.0017)  −0.015 (0.0056)  0.081 (0.0346)    PWM    0.039 (0.0143)  −0.077 (0.0502)    HCW      2.209 (0.7247)  Litter  BWT  0.015 (0.0015)        PWM    0.102 (0.0165)      HCW      4.914 (0.7987)  Residual  BWT  0.070 (0.0007)        PWM    1.000 (0.0080)      HCW      56.228 (0.7522)  1Variances on diagonal; covariances above diagonal. View Large Table 4. Estimates (SE) of (co)variances for sire genetic, contemporary group, dam, litter, and residual effects for birth weight (BWT), preweaning mortality (PWM), and HCW of crossbred pigs1 Effect  Trait  BWT  PWM  HCW  Sire genetic  BWT  0.001 (0.0004)  −0.001 (0.0009)  0.025 (0.0091)    PWM    0.064 (0.0039)  −0.014 (0.0275)    HCW      2.028 (0.4518)  Contemporary group  BWT        PWM    0.028 (0.0278)      HCW        Dam  BWT  0.015 (0.0017)  −0.015 (0.0056)  0.081 (0.0346)    PWM    0.039 (0.0143)  −0.077 (0.0502)    HCW      2.209 (0.7247)  Litter  BWT  0.015 (0.0015)        PWM    0.102 (0.0165)      HCW      4.914 (0.7987)  Residual  BWT  0.070 (0.0007)        PWM    1.000 (0.0080)      HCW      56.228 (0.7522)  Effect  Trait  BWT  PWM  HCW  Sire genetic  BWT  0.001 (0.0004)  −0.001 (0.0009)  0.025 (0.0091)    PWM    0.064 (0.0039)  −0.014 (0.0275)    HCW      2.028 (0.4518)  Contemporary group  BWT        PWM    0.028 (0.0278)      HCW        Dam  BWT  0.015 (0.0017)  −0.015 (0.0056)  0.081 (0.0346)    PWM    0.039 (0.0143)  −0.077 (0.0502)    HCW      2.209 (0.7247)  Litter  BWT  0.015 (0.0015)        PWM    0.102 (0.0165)      HCW      4.914 (0.7987)  Residual  BWT  0.070 (0.0007)        PWM    1.000 (0.0080)      HCW      56.228 (0.7522)  1Variances on diagonal; covariances above diagonal. View Large Estimated heritability was 0.042 for BWT, 0.022 for PWM, and 0.124 for HCW (Table 5). For each trait, the estimated heritability was at the lower range of literature estimates, especially for HCW (e.g., Grandinson et al., 2002; Knol et al., 2002; Lund et al., 2002; Arango et al., 2006; Zumbach et al., 2007; Cecchinato et al., 2010). Comparison with literature estimates is difficult because of the different structures of data sets and different models. In many studies, mortality traits were modeled with linear models (van Arendonk et al., 1996; Knol et al., 2002; Mesa et al., 2006), which ignore the categorical nature of those traits. Also, animals in this study were crossbreds, and some traits in crossbred populations have lower heritabilities than in purebred populations (Lutaaya et al., 2001). Table 5. Estimates (SE) of heritability, sire genetic, dam, and litter effects for birth weight (BWT), preweaning mortality (PWM), and HCW of crossbred pigs Effect  BWT  PWM  HCW  Heritability  0.04 (0.015)  0.02 (0.013)  0.12 (0.024)  Sire genetic  0.01 (0.004)  0.01 (0.003)  0.03 (0.007)  Dam  0.15 (0.016)  0.03 (0.012)  0.03 (0.011)  Common litter  0.14 (0.015)  0.09 (0.014)  0.08 (0.012)  Effect  BWT  PWM  HCW  Heritability  0.04 (0.015)  0.02 (0.013)  0.12 (0.024)  Sire genetic  0.01 (0.004)  0.01 (0.003)  0.03 (0.007)  Dam  0.15 (0.016)  0.03 (0.012)  0.03 (0.011)  Common litter  0.14 (0.015)  0.09 (0.014)  0.08 (0.012)  View Large Table 5. Estimates (SE) of heritability, sire genetic, dam, and litter effects for birth weight (BWT), preweaning mortality (PWM), and HCW of crossbred pigs Effect  BWT  PWM  HCW  Heritability  0.04 (0.015)  0.02 (0.013)  0.12 (0.024)  Sire genetic  0.01 (0.004)  0.01 (0.003)  0.03 (0.007)  Dam  0.15 (0.016)  0.03 (0.012)  0.03 (0.011)  Common litter  0.14 (0.015)  0.09 (0.014)  0.08 (0.012)  Effect  BWT  PWM  HCW  Heritability  0.04 (0.015)  0.02 (0.013)  0.12 (0.024)  Sire genetic  0.01 (0.004)  0.01 (0.003)  0.03 (0.007)  Dam  0.15 (0.016)  0.03 (0.012)  0.03 (0.011)  Common litter  0.14 (0.015)  0.09 (0.014)  0.08 (0.012)  View Large To determine if the sire component of each trait was useful for sire selection in a breeding program, the sire genetic effect was calculated as the ratio of estimated sire variance to total variance. Because the estimated sire genetic variance was small compared with total variance for each trait, the sire genetic effect (Table 5) was small (0.011 for BWT, 0.005 for PWM, and 0.031 for HCW). The larger effect of sire on HCW compared with BWT and PWM could be the result of declining maternal effect over time. Because piglet traits such as BWT and PWM are strongly influenced by maternal effects, most studies have usually included only maternal effects in analyses. However, Hamann et al. (2004) estimated genetic parameters for litter size, which is a trait strongly affected by maternal effects, as both sow and boar traits. They found that the sire had a small but significant effect on that trait. The estimated dam effect was defined as the ratio between the estimated dam variance and the total variance (Table 5). Because of the lack of pedigree information for the crossbred dams, the dam effect is composed of genetic and environmental components. Estimated dam effects were higher than direct heritabilities for BWT (0.146) and for PWM (0.033) but lower for HCW (0.034). Also, estimated dam effects were higher than sire genetic effects for BWT and PWM but on the same order of values for HCW. As expected, the dam effect is more important than the sire genetic effect on early recorded traits, especially on BWT compared to PWM (van Arendonk et al., 1996; Grandinson et al., 2002; Knol et al., 2002; Lund et al., 2002; Arango et al., 2006). Moreover, when the dam effects for BWT and HCW are compared, it appears that the maternal influence is attenuated with age. For HCW, the dam effect is of the same magnitude as the sire genetic effect. However, the dam effect contains a genetic part and an environmental part. Therefore, either the sire genetic effect is higher than the dam genetic effect, or the dam effect is mainly genetic rather than environmental in later performances (Zumbach et al., 2007). The common litter effect was defined as the ratio of estimated litter variance to total variance. The litter effect (Table 5) explained a large portion of total variance for piglet traits (0.143 for BWT, 0.086 for PWM). However, the common litter effect is lower for HCW (0.075) compared to heritability. Cecchinato et al. (2010) found that the litter variance was larger than the sire variance for preweaning survival of piglets, which confirms that piglet survival is mainly affected by the litter effects. The smaller common litter effect for HCW compared with BWT indicates that effects common to littermates dissipate with age, like the dam effects. In the literature, the proportion of total variance explained by the litter effect at market age was 4% to 6% for backfat and 5% to 12% for weight per day of age and HCW (Lutaaya et al., 2001; Zumbach et al., 2007). Moreover, the birth litter explained a larger part of the total variance than the dam did for HCW, as found by Zumbach et al. (2007) with a similar model. A common issue with the dam effect is cross-fostering and possible confounding of maternal and permanent environmental effects. For cross-fostered piglets, the maternal genetic effect is different before and after cross-fostering, BWT vs. PWM, for example. One strategy to deal with that is to include in the model the effect of the adoptive dam. However, in this study, the sow that raised the piglet is assumed to be the real mother of the piglet because of the lack of information available about the adoptive dam. In this case, the effect of the nurse dam for PWM and, to a smaller extent, for HCW is assumed to be part of the common litter effect. Knol et al. (2002) studied piglet survival with the genetic effect of the adoptive dam. They had issues with the estimation of genetic parameters (i.e., negative heritabilities) and convergence. Piglet BWT was genetically correlated with PWM (−0.52 ± 0.33) and HCW (0.55 ± 0.15). The SE of these correlations were lower than their corresponding estimates and did not include zero, supporting the genetic association between traits. The genetic correlation between PWM and HCW was lower (−0.13 ± 0.24). The SE was greater than its corresponding correlation and did include zero. This indicates no genetic association between PWM and HCW. The greater SE may also be because fewer data were considered. Moreover, such a low correlation might partly be the result of the data structure because dead piglets at weaning had no HCW record. Phenotypic correlations had the same sign than genetic correlations but were lower between BWT and PWM (−0.25) and between BWT and HCW (0.20) and higher between PWM and HCW (−0.48). The dam correlations were also favorable between BWT and PWM (−0.62 ± 0.23) and between BWT and HCW (0.45 ± 0.19). These correlations indicate that piglets from a dam providing favorable genes and environment have a greater chance to survive until weaning and to reach a high final market weight. The dam correlation between PWM and HCW was low (−0.26 ± 0.17) but stronger than the sire genetic correlation. The negative genetic correlation between BWT and PWM is in agreement with other studies (Grandinson et al., 2002; Arango et al., 2006; Roehe et al., 2010) and indicates a favorable genetic link between BWT and piglet survival until weaning. However, selection for higher BWT should be carefully undertaken; very high BWT may increase farrowing mortality because of other problems such as dystocia or prolonged parturition (Grandinson et al., 2002). As reported by Herring et al. (2010), BWT is positively correlated with final weight. Thus, BWT may be a good indicator of final market weight, as BWT is heritable and genetically correlated with HCW. Moreover, BWT is expressed earlier in life and is recorded earlier. Therefore, selection on BWT as a way to improve HCW may provide an opportunity to accelerate genetic progress. The favorable genetic correlations of BWT with PWM and HCW indicate that selection for high BWT can improve survival until weaning and final market weight for crossbred pigs. Therefore, BWT could be used as an early predictor of subsequent performances. However, such selection should not be extreme because of the association between high BWT and higher farrowing mortality (Grandinson et al., 2002; Arango et al., 2006), which is not economically advantageous. Moreover, this situation is more common in sire lines with lower prolificacies, as was evident in this study. Indeed, fewer piglets per litter leads to heavier piglets and a higher frequency of dystocia, and that affects the survival for the whole litter (Ibáñez-Escriche et al., 2009). Therefore, a profitable selection on survival rate needs to balance survival and birth weight. A restricted selection index (Kempthorne and Nordskog, 1959) is often used in such situations where changes in 1 particular trait, such as BWT, are restricted to zero while selecting for correlated traits of direct interest, such as PWM and HCW. Litter size at weaning is an important economic trait. Many breeding programs focus on selection to increase the number of piglets born per litter as a way to improve litter size at weaning (Grandinson et al., 2002). Selection index often puts large economic values on litter size and PWM (De Vries, 1989). However, selection to increase litter size at birth does not guarantee survival until weaning and larger litter at weaning. Indeed, litter size has a negative impact on preweaning survival and is linked to an increasing number of light piglets per litter and higher variations of piglet BWT within litter (van Arendonk et al., 1996; Milligan et al., 2002). High BWT variations within litter lead to competitive exclusion of light piglets from access to productive teats. Therefore, differences in BWT between light and heavy piglets are often maintained or even increased until weaning, and smaller piglets at birth have lower survival rates, which has a negative economic impact for producers (Milligan et al., 2002). Therefore, litter size must not be forgotten in the selection goal because it has indirect influence on survival rate (Lund et al., 2002; Arango et al., 2005). However, selection for litter size should be coupled with maintaining a minimum threshold for BWT to avoid too light piglets with more risks of PWM. Approximate reliabilities of sire breeding values for the 3 traits were computed with the following formula:  where n is the total number of progeny of the sire and s2 is the ratio between the sire genetic variance and the total variance. The mean approximate reliabilities of sire breeding values were 0.21 (0.14) for BWT and PWM and 0.29 (0.18) for HCW. These mean approximate reliabilities are low because of low sire genetic variance. Also, reliability depends on the number of progeny of the sire. Therefore, sires with a large number of progeny will have breeding values with higher reliability. However, the number of progeny per sire is variable (Table 2), which leads to a low mean approximate reliability. Theoretically, reliabilities should be higher with an animal model because of the use of the genetic relationships among all animals instead of only relationships among sires. However, because this is a crossbred population, an animal model would be suboptimal. The genetic effect of the sire on BWT, PWM, and HCW was low. Thus, on the basis of these results, direct selection of purebred sires may not be very efficient to improve those traits in crossbred populations. Subsequently, it may be easier to improve BWT from the maternal side than from the paternal side. However, BWT was genetically correlated with PWM and HCW. Therefore, selection on BWT could be a way to improve survival of piglets until weaning and final market weight. It could also improve the genetic progress as BWT is recorded earlier in the life of the animal. However, selection should be focused not only on improvement of BWT but also on an optimum combination of BWT, survival, and litter size. LITERATURE CITED Arango J. Misztal I. Tsuruta S. Culbertson M. Herring W. 2005. Threshold-linear estimation of genetic parameters for farrowing mortality, litter size, and test performance of Large White sows. J. Anim. Sci.  83: 499– 506. Google Scholar CrossRef Search ADS PubMed  Arango J. Misztal I. Tsuruta S. Culbertson M. Holl J. W. Herring W. 2006. Genetic study of individual preweaning mortality and birth weight in Large White piglets using threshold-linear models. Livest. Sci.  101: 208– 218. Google Scholar CrossRef Search ADS   Bérard J. Kreuzer M. Bee G. 2008. Effect of litter size and birth weight on growth, carcass and pork quality, and their relationship to post-mortem proteolysis. J. Anim. Sci.  86: 2357– 2368. Google Scholar CrossRef Search ADS PubMed  Cecchinato A. González-Recio O. López de Maturana E. Gallo L. Carnier P. 2010. A comparison between different survival and threshold models with an application to piglet preweaning survival in a dry-cured ham-producing crossbred line. J. Anim. Sci.  88: 1990– 1998. Google Scholar CrossRef Search ADS PubMed  De Vries A. G. 1989. A model to estimate economic values of trait in pig breeding. Livest. Prod. Sci.  21: 49– 66. Google Scholar CrossRef Search ADS   Fix J. S. Cassady J. P. Herring W. O. Holl J. W. Culbertson M. S. See M. T. 2010. Effect of piglet birth weight on body weight, growth, backfat, and longissimus muscle area of commercial market swine. Livest. Sci.  127: 51– 59. Google Scholar CrossRef Search ADS   Gondret F. Lefaucheur L. Louveau I. Lebret B. Pichodo X. Le Cozler Y. 2005. Influence of piglet birth weight on postnatal growth performance, tissue lipogenic capacity, and muscle histological traits at market weight. Livest. Prod. Sci.  93: 137– 146. Google Scholar CrossRef Search ADS   Grandinson K. Lund M. S. Rydhmer L. Strandberg E. 2002. Genetic parameters for the piglet mortality traits crushing, stillbirth and total mortality, and their relation to birth weight. Acta Agric. Scand. A Anim. Sci.  52: 167– 173. Hamann H. Steinheuer R. Distl O. 2004. Estimation of genetic parameters for litter size as a sow and boar trait in German herdbook Landrace and Pietrain swine. Livest. Prod. Sci.  85: 201– 207. Google Scholar CrossRef Search ADS   Herring W. Holl J. Culbertson M. Misztal I. Chen C. Y. Fix J. See T. 2010. Use of terminal and nucleus birth weights to improve commercial pig performance. In: Proc. 9th World Congr. Genet. Appl. Livest. Prod., Leipzig, Germany. Communication No. 0284. Ibáñez-Escriche N. Varona L. Casellas J. Quintanilla R. Noguera J. L. 2009. Bayesian threshold analysis of direct and maternal genetic parameters for piglet mortality at farrowing in Large White, Landrace, and Pietrain populations. J. Anim. Sci.  87: 80– 87. Google Scholar CrossRef Search ADS PubMed  Kempthorne O. Nordskog A. W. 1959. Restricted selection indices. Biometrics  15: 10– 19. Google Scholar CrossRef Search ADS   Knol E. F. Ducro B. J. van Arendonk J. A. M. van der Lende T. 2002. Direct, maternal and nurse sow genetic effects on farrowing-, pre-weaning- and total piglet survival. Livest. Prod. Sci.  73: 153– 164. Google Scholar CrossRef Search ADS   Lee D. Misztal I. Bertrand J. K. Rekaya R. 2002. National evaluation for calving ease, gestation length and birth weight by linear and threshold model methodologies. J. Appl. Genet.  43: 209– 216. Google Scholar PubMed  Lutaaya E. Misztal I. Mabry J. W. Short T. Timm H. H. Holzbauer R. 2001. Genetic parameter estimates from joint evaluation of purebreds and crossbreds in swine using the crossbred model. J. Anim. Sci.  79: 3002– 3007. Google Scholar CrossRef Search ADS PubMed  Lund M. S. Puonti M. Rydhmer L. Jensen J. 2002. Relationship between litter size and perinatal and pre-weaning survival in pigs. Anim. Sci.  74: 217– 222. Google Scholar CrossRef Search ADS   Mesa H. Safranski T. J. Cammack K. M. Weaber R. L. Lamberson W. R. 2006. Genetic and phenotypic relationships of farrowing and weaning survival to birth and placental weights in pig. J. Anim. Sci.  84: 32– 40. Google Scholar CrossRef Search ADS PubMed  Milligan B. N. Dewey C. E. de Grau A. F. 2002. Neonatal-piglet weight variation and its relation to pre-weaning mortality and weight gain on commercial farms. Prev. Vet. Med.  56: 119– 127. Google Scholar CrossRef Search ADS PubMed  Misztal I. Gianola D. Foulley J. L. 1989. Computing aspects of a nonlinear method of sire evaluation for categorical data. J. Dairy Sci.  72: 1557– 1568. Google Scholar CrossRef Search ADS   Misztal I. Tsuruta S. Strabel T. Auvray B. Druet T. Lee D. H. 2002. BLUPF90 and related programs (BGF90). In: Proc. 7th World Congr. Genet. Appl. Livest. Prod., Montpellier, France. Communication No. 28-07. p. 743– 744. Quiniou N. Dagorn J. Gaudré D. 2002. Variation of piglets' birth weight and consequences on subsequent performance. Livest. Prod. Sci.  78: 63– 70. Google Scholar CrossRef Search ADS   Rehfeldt C. Tuchscherer A. Hartung M. Kuhn G. 2008. A second look at the influence of birth weight on carcass and meat quality in pigs. Meat Sci.  78: 170– 175. Google Scholar CrossRef Search ADS PubMed  Roehe R. Shrestha N. P. Mekkawy W. Baxter E. M. Knap P. W. Smurthwaite K. M. Jarvis S. Lawrence A. B. Edwards S. A. 2010. Genetic parameters of piglet survival and birth weight from a two-generation crossbreeding experiment under outdoor conditions designed to disentangle direct and maternal effects. J. Anim. Sci.  88: 1276– 1285. Google Scholar CrossRef Search ADS PubMed  van Arendonk J. A. M. van Rosmeulen C. Janss L. L. G. Knol E. F. 1996. Estimation of direct and maternal genetic (co)variances for survival within litters of piglets. Livest. Prod. Sci.  46: 163– 171. Google Scholar CrossRef Search ADS   Zumbach B. Misztal I. Tsuruta S. Holl J. Herring W. Long T. 2007. Genetic correlations between two strains of Durocs and crossbreds from differing production environments for slaughter traits. J. Anim. Sci.  85: 901– 908. Google Scholar CrossRef Search ADS PubMed  Footnotes 1 Marie Dufrasne acknowledges the support of FRIA through a grant scholarship and the support of the National Fund for Scientific Research (Brussels, Belgium). The authors would like to thank C. Y. Chen. Editorial help by Suzanne Hubbard is gratefully acknowledged. American Society of Animal Science

Sae-Lim, P.;Kause, A.;Mulder, H. A.;Martin, K. E.;Barfoot, A. J.;Parsons, J. E.;Davidson, J.;III, C. E. Rexroad,;van Arendonk, J. A. M.;Komen, H.

doi: 10.2527/jas.2012-5949pmid: 24085417

Vallée, A.;van Arendonk, J. A. M.;Bovenhuis, H.

doi: 10.2527/jas.2013-6490pmid: 24085407

Abstract Charolais sires can be mated to Montbéliard or Holstein dairy cows to produce crossbred calves sold for meat production. Heritabilities and correlations between traits can differ when they are calculated within Charolais × Montbéliard or within Charolais × Holstein population. Moreover, the genetic correlation between the same trait measured on Charolais × Montbéliard and on Charolais × Holstein crossbred calves is not necessarily unity. The first objective of this study was to estimate heritability and genetic correlation between traits within Charolais × Montbéliard and within Charolais × Holstein population. The second objective was to investigate if those traits are genetically identical between crossbred populations. Traits studied were calving difficulty, birth weight, height, bone thinness, and muscular development. Data included 22,852 Charolais × Montbéliard and 16,012 Charolais × Holstein crossbred calves from 391 Charolais sires. Heritabilities estimated separately within each crossbred population were similar. Stronger genetic correlations were observed in Charolais × Holstein population compared with Charolais × Montbéliard between calving difficulty and height (0.67 vs. 0.54), calving difficulty and bone thinness (0.42 vs. 0.27), birth weight and bone thinness (0.52 vs. 0.20), and birth weight and muscular development (0.41 vs. 0.18). Bivariate analysis considering observations on Charolais × Montbéliard and on Charolais × Holstein as different traits showed that genetic variances and heritabilities were similar for all traits except height. Birth weight and muscular development were genetically identical traits in each crossbred populations, with genetic correlations of 0.96 and 0.99. Genetic correlations were 0.91 for calving difficulty, 0.80 for height, and 0.70 for bone thinness and log-likelihood ratio tests indicated that they were significantly different from 1 (P ≤ 0.01). Results show evidence for reranking of Charolais sires for calving difficulty, height, and bone thinness depending on whether they are mated to Montbéliard or Holstein cows. INTRODUCTION To manage the replacement of their dairy cattle herds, farmers can choose to inseminate dairy cows having low milk production potential with semen of beef sires. Calves produced are sold at about 3 wk of age for beef production. Economic value of these crossbred calves is directly linked to their conformation and indirectly linked to calving difficulty (Luo et al., 2002; Hickey et al., 2007). In France, where the data were collected, the 2 main dairy cow breeds mated to Charolais sires are Holstein and Montbéliard. In various species, prenatal environment provided by the mother was shown to have consequences on progeny (Nicholas, 1996). Allen et al. (2004) used embryo transfer between larger Thoroughbred and smaller Pony mares and determined a difference of 15% for growth at birth. Studies in pigs or poultry have compared genetic parameters of the parental purebred lines to their terminal crossbred lines (Lutaaya et al., 2001). Zumbach et al. (2007) found genetic correlations lower than 1 for same production traits observed in purebred lines and in their reciprocal crosses, which was partly attributed to different environment conditions. In bovines, differences in performances and economic impact between crossbred and purebred calves have been reported by few studies (Wolfova et al., 2007; Dal Zotto et al., 2009). Interaction between sire and maternal breed was one explanation, among others, for low to medium correlations (from 0.01 to +0.46) between breeding values of beef sires for growth traits estimated on purebred and on crossbred progeny (Tilsch et al., 1989). However, to our knowledge, no genetic parameters have been estimated within different crossbred populations. In addition, information is lacking on genetic correlations for the same trait between different crossbred populations. The study will focus on traits measured on Charolais × Montbéliard and Charolais × Holstein crossbred calves including calving difficulty, birth weight, height, bone thinness, and muscular development. Traits observed in Charolais × Holstein and Charolais × Montbéliard populations might be genetically different. Therefore, the first objective is to estimate heritabilities and genetic correlations among traits in each crossbred population separately. Furthermore, the second objective is to estimate genetic correlations between the same trait measured in Charolais × Holstein and Charolais × Montbéliard populations. MATERIALS AND METHODS Animal Care and Use Committee approval was not obtained for this study because data used is routinely collected as part of the breeding program and collecting these phenotypes does not violate the integrity of the animals. Population Structure Data were from 38,864 crossbred calves originating from 391 purebred Charolais AI sires mated to dams from Montbéliard or Holstein breeds. Number of males was 20,168 (51.9%) and 18,696 (48.1%) for females. Number of Charolais × Montbéliard calves was 22,852 (58.8%) and 16,012 (41.2%) for Charolais × Holstein calves. Number of sires with offspring in both crossbred populations was 367. Number of sires with more than 30 calves in each crossbred population was 204. Sires had on average 99 offspring. Traits Traits included in this study were calving difficulty, birth weight, height, bone thinness, and muscular development. Data were collected through the national progeny testing program on calves born between 1986 and 2012. Herds were located in the mideastern part of France. Calving difficulty was recorded by farmers and was evaluated on a scale from 1 to 5, in which 1 corresponded to a calving process without difficulty or assistance and 5 corresponded to particular difficult circumstances where the calf died during calving. Birth weight was estimated by farmers immediately after calving and expressed in kilograms. Conformation traits of calves were recorded on average at 22 d of age and included height, bone thinness, and muscular development. These conformation traits were scored by 19 qualified classifiers who followed regular training sessions to score the traits in a consistent way. Classifiers scored both crossbred calves using the same trait definition. Height at withers was scored on a scale from 1 to 5, in which 1 corresponded to shortest calves; bone thinness was also scored on a scale from 1 to 5, in which 1 corresponded to thinnest bone structure. Muscular development was evaluated based on visual inspection of shoulders, back, and rump. Each location was scored on a scale from 1 to 9, in which 1 corresponded to light muscular development. The overall score for muscular development was obtained by averaging the scores for shoulders, back, and rump. Statistical Analysis Data were analyzed using the following animal model:  in which Yijklmn was the observation, µ was the overall mean, Si was the fixed effect of sex i (2 classes), Cj was the fixed effect of classifier j (19 classes), BYBSk was the fixed effect of the combination between the birth year (from 1986 to 2012) and the birth season defined as 4 classes where 3-mo periods were defined starting in December (104 classes), Animall was the random additive genetic effect of the lth calf ∼N(0, Aσa2), in which A corresponded to additive genetic relationship matrix and σ2a corresponded to the additive genetic variance, and eijkl was the random residual effect ∼N(0, Iσe2), in which I corresponded to the identity matrix and σ2e corresponded to residual variance. Only relations on the paternal side were used to construct the additive genetic relationship matrix. Pedigree information on the paternal side was traced back with a minimum of 3 generations. At first, univariate analyses were used to estimate heritabilities and bivariate analyses to estimate genetic correlations between different traits measured within the same crossbred population. Second, bivariate analyses were used to estimate heritabilities and genetic correlations between the same trait measured in the 2 different crossbred populations, as follow:  in which y1 represents traits measured on Montbéliard × Charolais crossbreds and y2 on Holstein × Charolais, X1 and X2 are the incidence matrices for fixed effects b1 and b2, Z1 and Z2 are the incidence matrices for random genetic effects u1 and u2, and e1 and e2 are the error terms. Covariances between e1 and e2 were 0 as traits were measured on different individuals. To test if genetic correlation was significantly different from 1, the log-likelihood ratio test was used. The likelihoods used were of the unconstrained model and of the model where genetic correlation was fixed at 0.998. Constraining genetic correlation at a value of exactly 1 is computationally not possible. Significance levels were obtained from a chi-square distribution with 1 degree of freedom. Breeding values of the 204 sires with a minimum of 30 calves in each crossbred population were estimated, using a univariate model in Charolais × Montbéliard and in Charolais × Holstein separately. Genetic parameters were estimated using ASREML (Gilmour et al., 2009). RESULTS Descriptive Statistics Number of observations, means, and standard deviations are given for each crossbred population in Table 1. Charolais × Montbéliard and Charolais × Holstein populations had similar means for calving difficulty and height. Calves from Montbéliard dams had 600 g heavier estimated weights compared with calves from Holstein dams, had 0.17 point higher bone thinness scores, and had 0.35 point higher muscular development scores. Standard deviations were similar in both crossbred populations for all traits. Table 1. Traits description with number of observations, means, and standard deviations for each crossbred population     Charolais × Montbéliard  Charolais × Holstein  Trait  Scale  n  Mean  SD  n  Mean  SD  Calving difficulty  1 (easy) to 5 (difficult)  20,806  1.67  0.69  15,580  1.74  0.70  Birth weight  kg  20,064  45.1  8.0  15,029  44.5  7.9  Height  1 (short) to 5 (tall)  18,759  3.15  0.85  12,641  3.11  0.92  Bone thinness  1 (thin) to 5 (thick)  18,772  2.65  0.85  12,647  2.48  0.85  Muscular development  1 (light) to 9 (heavy)  18,882  5.30  1.42  12,788  4.95  1.48      Charolais × Montbéliard  Charolais × Holstein  Trait  Scale  n  Mean  SD  n  Mean  SD  Calving difficulty  1 (easy) to 5 (difficult)  20,806  1.67  0.69  15,580  1.74  0.70  Birth weight  kg  20,064  45.1  8.0  15,029  44.5  7.9  Height  1 (short) to 5 (tall)  18,759  3.15  0.85  12,641  3.11  0.92  Bone thinness  1 (thin) to 5 (thick)  18,772  2.65  0.85  12,647  2.48  0.85  Muscular development  1 (light) to 9 (heavy)  18,882  5.30  1.42  12,788  4.95  1.48  View Large Table 1. Traits description with number of observations, means, and standard deviations for each crossbred population     Charolais × Montbéliard  Charolais × Holstein  Trait  Scale  n  Mean  SD  n  Mean  SD  Calving difficulty  1 (easy) to 5 (difficult)  20,806  1.67  0.69  15,580  1.74  0.70  Birth weight  kg  20,064  45.1  8.0  15,029  44.5  7.9  Height  1 (short) to 5 (tall)  18,759  3.15  0.85  12,641  3.11  0.92  Bone thinness  1 (thin) to 5 (thick)  18,772  2.65  0.85  12,647  2.48  0.85  Muscular development  1 (light) to 9 (heavy)  18,882  5.30  1.42  12,788  4.95  1.48      Charolais × Montbéliard  Charolais × Holstein  Trait  Scale  n  Mean  SD  n  Mean  SD  Calving difficulty  1 (easy) to 5 (difficult)  20,806  1.67  0.69  15,580  1.74  0.70  Birth weight  kg  20,064  45.1  8.0  15,029  44.5  7.9  Height  1 (short) to 5 (tall)  18,759  3.15  0.85  12,641  3.11  0.92  Bone thinness  1 (thin) to 5 (thick)  18,772  2.65  0.85  12,647  2.48  0.85  Muscular development  1 (light) to 9 (heavy)  18,882  5.30  1.42  12,788  4.95  1.48  View Large Heritability and Correlation within Crossbred Population Heritabilities and phenotypic and genetic correlations among birth and conformation traits for Charolais × Montbéliard population are presented in Table 2 and for Charolais × Holstein population in Table 3. Estimated heritabilities and phenotypic correlations were similar in both crossbred populations. Calving difficulty and birth weight had similar genetic correlation in Charolais × Montbéliard population (0.86) and in Charolais × Holstein population (0.87). Muscular development had in both populations genetic correlations of approximately zero with height and bone thinness (from –0.10 to 0.01). Stronger genetic correlations were observed in Charolais × Holstein population compared with Charolais × Montbéliard between calving difficulty and height (0.67 vs. 0.54), calving difficulty and bone thinness (0.42 vs. 0.27), birth weight and bone thinness (0.52 vs. 0.20), and birth weight and muscular development (0.41 vs. 0.18). Table 2. Heritability and phenotypic and genetic correlation for preweaning traits measured in Charolais × Montbéliard calves. Phenotypic correlations are presented above the diagonal (in italics) and genetic correlations are below. Heritabilities on diagonal and genetic variance (σ2a) were estimated on univariate analysis Charolais × Montbéliard  Calving difficulty  Birth weight  Height  Bone thinness  Muscular development  Calving difficulty  0.16 (0.02)1  0.41 (0.01)  0.21 (0.01)  0.18 (0.01)  0.17 (0.01)  Birth weight  0.86 (0.03)  0.26 (0.03)  0.48 (0.01)  0.36 (0.01)  0.27 (0.01)  Height  0.54 (0.06)  0.71 (0.04)  0.33 (0.03)  0.35 (0.01)  0.13 (0.01)  Bone thinness  0.27 (0.07)  0.20 (0.07)  0.44 (0.06)  0.32 (0.03)  0.11 (0.01)  Muscular development  0.47 (0.07)  0.18 (0.07)  –0.10 (0.07)  0.01 (0.07)  0.35 (0.03)  σ2a  0.07  15.7  0.22  0.18  0.60  Charolais × Montbéliard  Calving difficulty  Birth weight  Height  Bone thinness  Muscular development  Calving difficulty  0.16 (0.02)1  0.41 (0.01)  0.21 (0.01)  0.18 (0.01)  0.17 (0.01)  Birth weight  0.86 (0.03)  0.26 (0.03)  0.48 (0.01)  0.36 (0.01)  0.27 (0.01)  Height  0.54 (0.06)  0.71 (0.04)  0.33 (0.03)  0.35 (0.01)  0.13 (0.01)  Bone thinness  0.27 (0.07)  0.20 (0.07)  0.44 (0.06)  0.32 (0.03)  0.11 (0.01)  Muscular development  0.47 (0.07)  0.18 (0.07)  –0.10 (0.07)  0.01 (0.07)  0.35 (0.03)  σ2a  0.07  15.7  0.22  0.18  0.60  1Standard errors are between brackets. View Large Table 2. Heritability and phenotypic and genetic correlation for preweaning traits measured in Charolais × Montbéliard calves. Phenotypic correlations are presented above the diagonal (in italics) and genetic correlations are below. Heritabilities on diagonal and genetic variance (σ2a) were estimated on univariate analysis Charolais × Montbéliard  Calving difficulty  Birth weight  Height  Bone thinness  Muscular development  Calving difficulty  0.16 (0.02)1  0.41 (0.01)  0.21 (0.01)  0.18 (0.01)  0.17 (0.01)  Birth weight  0.86 (0.03)  0.26 (0.03)  0.48 (0.01)  0.36 (0.01)  0.27 (0.01)  Height  0.54 (0.06)  0.71 (0.04)  0.33 (0.03)  0.35 (0.01)  0.13 (0.01)  Bone thinness  0.27 (0.07)  0.20 (0.07)  0.44 (0.06)  0.32 (0.03)  0.11 (0.01)  Muscular development  0.47 (0.07)  0.18 (0.07)  –0.10 (0.07)  0.01 (0.07)  0.35 (0.03)  σ2a  0.07  15.7  0.22  0.18  0.60  Charolais × Montbéliard  Calving difficulty  Birth weight  Height  Bone thinness  Muscular development  Calving difficulty  0.16 (0.02)1  0.41 (0.01)  0.21 (0.01)  0.18 (0.01)  0.17 (0.01)  Birth weight  0.86 (0.03)  0.26 (0.03)  0.48 (0.01)  0.36 (0.01)  0.27 (0.01)  Height  0.54 (0.06)  0.71 (0.04)  0.33 (0.03)  0.35 (0.01)  0.13 (0.01)  Bone thinness  0.27 (0.07)  0.20 (0.07)  0.44 (0.06)  0.32 (0.03)  0.11 (0.01)  Muscular development  0.47 (0.07)  0.18 (0.07)  –0.10 (0.07)  0.01 (0.07)  0.35 (0.03)  σ2a  0.07  15.7  0.22  0.18  0.60  1Standard errors are between brackets. View Large Table 3. Heritability and phenotypic and genetic correlation for preweaning traits measured in Charolais × Holstein calves. Phenotypic correlations are presented above the diagonal (in italics) and genetic correlations are below. Heritabilities on diagonal and genetic variance (σ2a) were estimated on univariate analysis Charolais × Holstein  Calving difficulty  Birth weight  Height  Bone thinness  Muscular development  Calving difficulty  0.12 (0.02)1  0.40 (0.01)  0.23 (0.01)  0.17 (0.01)  0.15 (0.01)  Birth weight  0.87 (0.04)  0.20 (0.02)  0.51 (0.01)  0.39 (0.01)  0.28 (0.01)  Height  0.67 (0.06)  0.68 (0.05)  0.36 (0.04)  0.33 (0.01)  0.10 (0.01)  Bone thinness  0.42 (0.08)  0.52 (0.06)  0.45 (0.07)  0.30 (0.03)  0.13 (0.01)  Muscular development  0.49 (0.08)  0.41 (0.07)  0.01 (0.08)  –0.02 (0.08)  0.30 (0.03)  σ2a  0.06  11.3  0.26  0.17  0.49  Charolais × Holstein  Calving difficulty  Birth weight  Height  Bone thinness  Muscular development  Calving difficulty  0.12 (0.02)1  0.40 (0.01)  0.23 (0.01)  0.17 (0.01)  0.15 (0.01)  Birth weight  0.87 (0.04)  0.20 (0.02)  0.51 (0.01)  0.39 (0.01)  0.28 (0.01)  Height  0.67 (0.06)  0.68 (0.05)  0.36 (0.04)  0.33 (0.01)  0.10 (0.01)  Bone thinness  0.42 (0.08)  0.52 (0.06)  0.45 (0.07)  0.30 (0.03)  0.13 (0.01)  Muscular development  0.49 (0.08)  0.41 (0.07)  0.01 (0.08)  –0.02 (0.08)  0.30 (0.03)  σ2a  0.06  11.3  0.26  0.17  0.49  1Standard errors are between brackets. View Large Table 3. Heritability and phenotypic and genetic correlation for preweaning traits measured in Charolais × Holstein calves. Phenotypic correlations are presented above the diagonal (in italics) and genetic correlations are below. Heritabilities on diagonal and genetic variance (σ2a) were estimated on univariate analysis Charolais × Holstein  Calving difficulty  Birth weight  Height  Bone thinness  Muscular development  Calving difficulty  0.12 (0.02)1  0.40 (0.01)  0.23 (0.01)  0.17 (0.01)  0.15 (0.01)  Birth weight  0.87 (0.04)  0.20 (0.02)  0.51 (0.01)  0.39 (0.01)  0.28 (0.01)  Height  0.67 (0.06)  0.68 (0.05)  0.36 (0.04)  0.33 (0.01)  0.10 (0.01)  Bone thinness  0.42 (0.08)  0.52 (0.06)  0.45 (0.07)  0.30 (0.03)  0.13 (0.01)  Muscular development  0.49 (0.08)  0.41 (0.07)  0.01 (0.08)  –0.02 (0.08)  0.30 (0.03)  σ2a  0.06  11.3  0.26  0.17  0.49  Charolais × Holstein  Calving difficulty  Birth weight  Height  Bone thinness  Muscular development  Calving difficulty  0.12 (0.02)1  0.40 (0.01)  0.23 (0.01)  0.17 (0.01)  0.15 (0.01)  Birth weight  0.87 (0.04)  0.20 (0.02)  0.51 (0.01)  0.39 (0.01)  0.28 (0.01)  Height  0.67 (0.06)  0.68 (0.05)  0.36 (0.04)  0.33 (0.01)  0.10 (0.01)  Bone thinness  0.42 (0.08)  0.52 (0.06)  0.45 (0.07)  0.30 (0.03)  0.13 (0.01)  Muscular development  0.49 (0.08)  0.41 (0.07)  0.01 (0.08)  –0.02 (0.08)  0.30 (0.03)  σ2a  0.06  11.3  0.26  0.17  0.49  1Standard errors are between brackets. View Large Heritability and Genetic Correlation between Crossbred Populations Table 4 shows heritabilities and genetic correlations between the same trait measured in Charolais × Montbéliard and in Charolais × Holstein populations. Heritabilities estimates based on bivariate analysis were similar for Charolais × Montbéliard and Charolais × Holstein for all traits except for height where heritability was lower in Charolais × Montbéliard (0.34) than in Charolais × Holstein (0.55) populations. This difference is mainly due to a lower additive genetic variance in Charolais × Montbéliard (0.23) as compared to Charolais × Holstein (0.44) populations. This difference in heritability and in additive genetic variance was not as pronounced when estimating heritabilities on univariate analysis within population (Tables 2 and 3). Table 4. Heritabilities and genetic correlations between same traits measured in Charolais × Montbéliard and in Charolais × Holstein populations   Charolais × Montbéliard  Charolais × Holstein        σ2a1  σ2p2  h2 3  σ2a  σ2p  h2  rg4  P-value5  Calving difficulty  0.08  0.46  0.17 (0.02)6  0.07  0.47  0.14 (0.02)  0.91 (0.04)  0.01  Birth weight  17.8  60.8  0.29 (0.03)  13.1  58.1  0.23 (0.03)  0.96 (0.02)  0.05  Height  0.23  0.67  0.34 (0.03)  0.44  0.79  0.55 (0.05)  0.80 (0.04)  <0.001  Bone thinness  0.19  0.59  0.32 (0.03)  0.17  0.55  0.31 (0.03)  0.70 (0.05)  <0.001  Muscular development  0.61  1.69  0.36 (0.03)  0.58  1.65  0.35 (0.03)  0.99 (0.02)  0.75    Charolais × Montbéliard  Charolais × Holstein        σ2a1  σ2p2  h2 3  σ2a  σ2p  h2  rg4  P-value5  Calving difficulty  0.08  0.46  0.17 (0.02)6  0.07  0.47  0.14 (0.02)  0.91 (0.04)  0.01  Birth weight  17.8  60.8  0.29 (0.03)  13.1  58.1  0.23 (0.03)  0.96 (0.02)  0.05  Height  0.23  0.67  0.34 (0.03)  0.44  0.79  0.55 (0.05)  0.80 (0.04)  <0.001  Bone thinness  0.19  0.59  0.32 (0.03)  0.17  0.55  0.31 (0.03)  0.70 (0.05)  <0.001  Muscular development  0.61  1.69  0.36 (0.03)  0.58  1.65  0.35 (0.03)  0.99 (0.02)  0.75  1σ2a = genetic variance. 2σ2p = phenotypic variance. 3h2 = heritability. 4rg = genetic correlation. 5P-value is based on the log-likelihood ratio test and indicates if genetic correlations differ from unity. 6Standard errors are between brackets. View Large Table 4. Heritabilities and genetic correlations between same traits measured in Charolais × Montbéliard and in Charolais × Holstein populations   Charolais × Montbéliard  Charolais × Holstein        σ2a1  σ2p2  h2 3  σ2a  σ2p  h2  rg4  P-value5  Calving difficulty  0.08  0.46  0.17 (0.02)6  0.07  0.47  0.14 (0.02)  0.91 (0.04)  0.01  Birth weight  17.8  60.8  0.29 (0.03)  13.1  58.1  0.23 (0.03)  0.96 (0.02)  0.05  Height  0.23  0.67  0.34 (0.03)  0.44  0.79  0.55 (0.05)  0.80 (0.04)  <0.001  Bone thinness  0.19  0.59  0.32 (0.03)  0.17  0.55  0.31 (0.03)  0.70 (0.05)  <0.001  Muscular development  0.61  1.69  0.36 (0.03)  0.58  1.65  0.35 (0.03)  0.99 (0.02)  0.75    Charolais × Montbéliard  Charolais × Holstein        σ2a1  σ2p2  h2 3  σ2a  σ2p  h2  rg4  P-value5  Calving difficulty  0.08  0.46  0.17 (0.02)6  0.07  0.47  0.14 (0.02)  0.91 (0.04)  0.01  Birth weight  17.8  60.8  0.29 (0.03)  13.1  58.1  0.23 (0.03)  0.96 (0.02)  0.05  Height  0.23  0.67  0.34 (0.03)  0.44  0.79  0.55 (0.05)  0.80 (0.04)  <0.001  Bone thinness  0.19  0.59  0.32 (0.03)  0.17  0.55  0.31 (0.03)  0.70 (0.05)  <0.001  Muscular development  0.61  1.69  0.36 (0.03)  0.58  1.65  0.35 (0.03)  0.99 (0.02)  0.75  1σ2a = genetic variance. 2σ2p = phenotypic variance. 3h2 = heritability. 4rg = genetic correlation. 5P-value is based on the log-likelihood ratio test and indicates if genetic correlations differ from unity. 6Standard errors are between brackets. View Large Genetic correlations between crossbred populations for birth weight and for muscular development were not significantly different from 1 (P = 0.05 and P = 0.75). Genetic correlation between crossbred populations was 0.91 for calving difficulty, 0.80 for height, and 0.70 for bone thinness and all were significantly different from 1 (P ≤ 0.01). Comparison of Breeding Values Estimated within Crossbred Population Breeding values of sires estimated on either their Charolais × Montbéliard or their Charolais × Holstein crossbred offspring are shown in Fig. 1. Traits reported in Fig. 1 are bone thinness and muscular development, that is, a situation where the genetic correlation is significantly different from 1 (bone thinness) and a situation where the genetic correlation is not significantly different from 1 (muscular development). Figure 1. View largeDownload slide Breeding values of 204 sires originating from univariate analysis for bone thinness and muscular development estimated either on Charolais × Montbéliard or on Charolais × Holstein crossbred calves Figure 1. View largeDownload slide Breeding values of 204 sires originating from univariate analysis for bone thinness and muscular development estimated either on Charolais × Montbéliard or on Charolais × Holstein crossbred calves DISCUSSION Trait Means for Charolais × Montbéliard and Charolais × Holstein Populations Calves from crosses between Charolais sires and Montbéliard dams have on average higher birth weight, thicker bones, and higher grades for muscular development. This difference between Charolais × Montbéliard and Charolais × Holstein calves might be due to (maternal) genetic differences between Montbéliard and Holstein dams. However, as Holstein and Montbéliard cows are generally raised in different herds, we cannot exclude specific effects of Montbéliard and Holstein herds such as age at calving or criteria to select females used for terminal cross. In the present study, no information was available on specific farm conditions but Montbéliard and Holstein herds were located in the same region and we are not aware of any systematic differences in management between Montbéliard and Holstein herds. Therefore we expect that (maternal) genetic differences between Montbéliard and Holstein dams are the main reason for differences in mean values between both crossbred populations. Heritabilities Estimates of heritability obtained for calving difficulty are similar to that reported by Mujibi and Crews (2009) on purebred Charolais who analyzed scores transformed to a continuous scale. Heritability of birth weight is slightly lower than previous studies on purebred Charolais (Phocas and Laloe, 2003; Mujibi and Crews, 2009). Little information is available on genetic parameters for conformation traits of young calves because most studies considered postweaning traits. Heritability of bone thinness for Piemontese cows was 0.12 (Mantovani et al., 2010), which is lower than the present results. Heritability of muscularity at weaning for purebred Blonde d'Aquitaine and Limousin animals (Bouquet et al., 2010) was similar than results in the current study. Afolayan et al. (2007) analyzed height and muscularity from weaning to 600 d and estimated heritabilities from 0.42 to 0.60 for height and from 0.19 to 0.44 for muscularity. Heritability estimates for height tend to be higher than results of the current study, which might be due to the objective measurement of this trait in centimeters. Preweaning traits in beef cattle are affected by maternal effects (Manfredi et al., 1991; Brandt et al., 2010; McHugh et al., 2011). Therefore, most national cattle evaluation programs use statistical models accounting for direct genetic, maternal genetic, and maternal permanent environmental effects (Crews and Wang, 2007). Models used in the present study did not include a maternal (genetic) effect because interest is in genetic parameters for Charolais breed and maternal (genetic) effects would relate to the Holstein or Montbéliard breeds. Present heritabilities are comparable with other studies considering a maternal effect (Phocas and Laloe, 2003; Eriksson et al., 2004). In the current data set herd information was missing for two-thirds of the data and therefore the effect of herd was not included in the model. However, as farmers recorded calving traits, difference in trait values between herds might exist due to the subjective nature of recording. In addition, differences in management between herds might exist, which could affect the traits. Therefore, additional analyses were performed based on observations for which herd information was available. Herd variance explained between 1% (for bone thinness) and 23% (for birth weight) of the phenotypic variance. Herd variance was especially important for traits recorded by the farmer. Adjusting for herd effects led to similar estimates of genetic variance as those from the analysis when not including herd effects. Adjusting for herd effects does lead to a reduction of the residual variance for most traits and therefore might affect heritability estimates. This underlines the importance of herd identification and adjusting for herd effects. Genetic Correlation between Traits within Crossbred Population Calving difficulty and birth weight have high genetic correlations in both Charolais × Montbéliard and Charolais × Holstein populations (0.86 and 0.87). This is in agreement with estimates in purebred Charolais (Mujibi and Crews, 2009). These genetic antagonisms indicate the difficulty to improve simultaneously calving difficulty and birth weight, traits that are both of interest for the terminal cross. Calving difficulty is moderately correlated with height, bone thinness, and muscular development. Afolayan et al. (2007) reported similar genetic correlations between weight and height and between weight and muscle percentage at 400 d. Height and bone thinness are not correlated with muscular development, which offers opportunities to improve these traits independently. Results agree with correlations between height and muscular development in Normande dairy breed (Colleau et al., 1989). Genetic Correlations between Crossbred Populations Calving difficulty [genetic correlation (rg) of 0.91], bone thinness (rg of 0.70), and height (rg of 0.80) should be considered as genetically different traits depending on the breed of the dam. To our knowledge no similar study has been conducted in cattle. In pig breeding, different purebred populations are mated to produce terminal crossbred animals and this allows estimating genetic correlations between different crosses. Zumbach et al. (2007) presented a range of genetic correlations for production traits between purebreds and their reciprocal crosses with lowest value of 0.53 for growth. Although they concluded that low genetic correlations were due to difference in environment between populations, epistatic interaction could also have played a role. Genetic differences between traits measured in Charolais × Montbéliard and Charolais × Holstein, as quantified by genetic correlation, could be supported by differences in genetic variance in case of height. Furthermore, genetic correlations of calving difficulty, height, and bone thinness with other traits differ when evaluated in Charolais × Montbéliard or in Charolais × Holstein. As consequence of these genetic differences, reranking of sires evaluated in each crossbred population was observed. Tilsch et al. (1989) evaluated breeding values of beef sires through the performance of a minimum 6 of their purebred and crossbred progeny for various growth traits. The low correlations between breeding values (from 0.01 to +0.46) also suggest genetic differences between sires depending on the dam breeds they were mated to. In the context of pig breeding, even though purebred and crossbred populations are kept in different environments, several studies showed the influence of the breed of the population used for evaluation on the breeding value of purebred sires (Lo et al., 1993; Dekkers, 2007; Zumbach et al., 2007) and concluded on important reranking (Ibáñez-Escriche et al., 2011). Influence of Maternal Breed Genetic differences between traits measured in Charolais × Montbéliard and Charolais × Holstein might originate from several factors, one being differences in maternal environment, that is, environment before (and closely after) birth (e.g., Banos et al., 2007). For example, due to differences in morphology, Montbéliard and Holstein dams might provide different uterine conditions to their offspring, which might result in genotype × (uterine) environment interaction. Studies using embryo transfer and cross-fostering in mice resulted in significant uterine and nursing effects on tail length, body weight, and growth rate (Cowley et al., 1989; Rhees et al., 1999). Studies in human also showed effect of pre- and postnatal maternal environment on obesity or diabetes phenotypes (Barker, 1998). Alternatively, epistatic interactions might explain genetic differences between traits measured in Charolais × Montbéliard and Charolais × Holstein. Indeed, the effect of alleles from the Charolais sire might differ depending on the presence of alleles from either Montbéliard or from Holstein breed. In addition, interaction between the maternal genotype and the genotype of the offspring might play a role. This hypothesis is confirmed by observations in mice using embryo transfer, where various responses of the offspring genotypes on body weight were noticed depending on the kind of mother they developed in (Maestripieri and Mateo, 2009). Implications Calving difficulty, height, and bone thinness show significant genetic differences when measured in Charolais × Montbéliard or in Charolais × Holstein calves (P-value from 0.01 to < 0.001). Genetic difference for calving difficulty is smaller; however, this trait is of great interest for farmers. Consequently, selection of Charolais sires depends on the dam breed. Separated genetic evaluations for Charolais × Montbéliard and Charolais × Holstein crossbreds should be considered. This would offer breeding companies the possibility to label their commercial sires depending on their performance with a certain dam breed. Farmers would have the opportunity to make optimal choice when buying sires straws to inseminate their Holstein or Montbéliard females. LITERATURE CITED Afolayan R. A. Pitchford W. S. Delanda M. P. B. McKiernan W. A. 2007. Breed variation and genetic parameters for growth and body development in diverse beef cattle genotypes. Animal  1: 13– 20. Google Scholar CrossRef Search ADS PubMed  Allen W. R. Wilsher S. Tiplady C. Butterfield R. M. 2004. The influence of maternal size on pre- and postnatal growth in the horse: III Postnatal growth. Reproduction  127: 67– 77. Google Scholar CrossRef Search ADS PubMed  Banos G. Brotherstone S. Coffey M. P. 2007. Prenatal maternal effects on body condition score, female fertility, and milk yield of dairy cows. J. Dairy Sci.  90: 3490– 3499. Google Scholar CrossRef Search ADS PubMed  Barker D. J. P 1998. Mothers, babies, and health in later life. 2nd ed. Churchill Livingstone, Edinburgh, UK. Bouquet A. Fouilloux M.-N. Renand G. Phocas F. 2010. Genetic parameters for growth, muscularity, feed efficiency and carcass traits of young beef bulls. Livest. Prod.  129: 38– 48. Google Scholar CrossRef Search ADS   Brandt H. Müllenhoff A. Lambertz C. Erhardt G. Gauly M. 2010. Estimation of genetic and crossbreeding parameters for preweaning traits in German Angus and Simmental beef cattle and the reciprocal crosses. J. Anim. Sci.  88: 80– 86. Google Scholar CrossRef Search ADS PubMed  Colleau J. J. Beaumont C. Regaldo D. 1989. Restricted maximum likelihood (REML) estimation of genetic parameters for type traits in Normande cattle breed. Livest. Prod. Sci.  23: 47– 66. Google Scholar CrossRef Search ADS   Cowley D. E. Pomp D. Atchley W. R. Eisen E. J. Hawkins-Brown D. 1989. The impact of maternal uterine genotype on postnatal growth and adult body size in mice. Genetics  122: 193– 203. Google Scholar PubMed  Crews D. H.Jr Wang Z. 2007. Illustration of the maternal animal model used for genetic evaluation of beef cattle1. J. Anim. Sci.  85: 1842– 1848. Google Scholar CrossRef Search ADS PubMed  Dal Zotto R. Penasa M. De Marchi M. Cassandro M. Lòpez-Villalobos N. Bittante G. 2009. Use of crossbreeding with beef bulls in dairy herds: Effect on age, body weight, price, and market value of calves sold at livestock auctions. J. Anim. Sci.  87: 3053– 3059. Google Scholar CrossRef Search ADS PubMed  Dekkers J. C. M 2007. Marker-assisted selection for commercial crossbred performance. J. Anim. Sci.  85: 2104– 2114. Google Scholar CrossRef Search ADS PubMed  Eriksson S. Näsholm A. Johansson K. Philipsson J. 2004. Genetic parameters for calving difficulty, stillbirth, and birth weight for Hereford and Charolais at first and later parities. J. Anim. Sci.  82: 375– 383. Google Scholar CrossRef Search ADS PubMed  Gilmour A. R. Gogel B. J. Cullis B. R. Thomson R. 2009. ASREML ASReml User Guide Release 3.0. VSN International Ltd., Hemel Hempstead, UK. www.vsni.co.uk. Hickey J. M. Keane M. G. Kenny D. A. Cromie A. R. Amer P. R. Veerkamp R. F. 2007. Heterogeneity of genetic parameters for calving difficulty in Holstein heifers in Ireland. J. Dairy Sci.  90: 3900– 3908. Google Scholar CrossRef Search ADS PubMed  Ibáñez-Escriche N. Reixach J. Lleonart N. Noguera J. L. 2011. Genetic evaluation combining purebred and crossbred data in a pig breeding scheme1. J. Anim. Sci.  89: 3881– 3889. Google Scholar CrossRef Search ADS PubMed  Lo L. L. Fernando R. L. Grossman M. 1993. Covariance between relatives in multibreed populations: Additive model. Theor. Appl. Genet.  87: 423– 430. Google Scholar CrossRef Search ADS PubMed  Luo M. F. Boettcher P. J. Schaeffer L. R. Dekkers J. C. M. 2002. Estimation of genetic parameters of calving ease in first and second parities of Canadian Holsteins using Bayesian methods. Livest. Prod. Sci.  74: 175– 184. Google Scholar CrossRef Search ADS   Lutaaya E. Misztal I. Mabry J. W. Short T. Timm H. H. Holzbauer R. 2001. Genetic parameter estimates from joint evaluation of purebreds and crossbreds in swine using the crossbred model1. J. Anim. Sci.  79: 3002– 3007. Google Scholar CrossRef Search ADS PubMed  Maestripieri D. Mateo J. M. 2009. Maternal effects in mammals. University of Chicago Press, Chicago, IL. Google Scholar CrossRef Search ADS   Manfredi E. J. San Cristobal M. Foulley J. L. 1991. Some factors affecting the estimation of genetic parameters for cattle dystocia under a threshold model. Anim. Prod.  53: 151– 156. Google Scholar CrossRef Search ADS   Mantovani R. Cassandro M. Contiero B. Albera A. Bittante G. 2010. Genetic evaluation of type traits in hypertrophic Piemontese cows. J. Anim. Sci.  88: 3504– 3512. Google Scholar CrossRef Search ADS PubMed  McHugh N. Evans R. D. Amer P. R. Fahey A. G. Berry D. P. 2011. Genetic parameters for cattle price and body weight from routinely collected data at livestock auctions and commercial farms. J. Anim. Sci.  89: 29– 39. Google Scholar CrossRef Search ADS PubMed  Mujibi F. D. N. Crews D. H.Jr 2009. Genetic parameters for calving ease, gestation length, and birth weight in Charolais cattle. J. Anim. Sci.  87: 2759– 2766. Google Scholar CrossRef Search ADS PubMed  Nicholas F. W 1996. Introduction to veterinary genetics. Oxford Univ. Press, Oxford. UK. Phocas F. Laloe D. 2003. Evaluation models and genetic parameters for calving difficulty in beef cattle. J. Anim. Sci.  81: 933– 938. Google Scholar CrossRef Search ADS PubMed  Rhees B. K. Ernst C. A. Miao C. H. Atchley W. R. 1999. Uterine and postnatal maternal effects in mice selected for differential rate of early development. Genetics  153: 905– 917. Google Scholar PubMed  Tilsch K. Wollert J. Baumung A. 1989. Relationship between breeding values for growth of beef sires in purebreeding and crossbreeding. Livest. Prod. Sci.  21: 275– 285. Google Scholar CrossRef Search ADS   Wolfova M. Wolf J. Kvapilık J. Kica J. 2007. Selection for profit in cattle: II. Economic weights for dairy and beef sires in crossbreeding systems. J. Dairy Sci.  90: 2456– 2467. Google Scholar CrossRef Search ADS PubMed  Zumbach B. Misztal I. Tsuruta S. Holl J. Herring W. Long T. 2007. Genetic correlations between two strains of Durocs and crossbreds from differing production environments for slaughter traits. J. Anim. Sci.  85: 901– 908. Google Scholar CrossRef Search ADS PubMed  Footnotes 1 This study was funded by Gènes Diffusion. The authors thank Institut de l'Elevage for making data available, Weijia Gong for his helpful support, and all members of Gènes Diffusion's staff involved in the project. American Society of Animal Science

Paredes, S. P.;Kalbe, C.;Jansman, A. J. M.;Verstegen, M. W. A.;van Hees, H. M. J.;Lösel, D.;Gerrits, W. J. J.;Rehfeldt, C.

doi: 10.2527/jas.2013-6908pmid: 24126270

Abstract Postnatal (muscle) growth potential in pigs depends on the total number and hypertrophy of myofibers in skeletal muscle tissue. In a previous study an algorithm was developed to predict piglet BW at the end of the nursery period (10 wk of age) on the basis of BW at birth, at weaning, and at 6 wk of age. The objective of this study was to determine whether the differences in growth performance between poor (PP) and high (HP) performing piglets could be the result of different skeletal muscle properties. Therefore, from a total of 368 piglets (offspring from Hypor sows bred to TOPIGS sires) 2 groups with a divergent growth performance were selected at 6 wk of age: HP (n = 20, predicted BW at 10 wk of age 26.8–30.9 kg) and PP (n = 20, predicted BW at 10 wk of age 16.0–22.9 kg). Piglets were euthanized at 10 wk of age, and samples of the semitendinosus muscle (STN) were collected for histochemistry and gene expression analysis using quantitative PCR (qPCR). At 10 wk of age, realized BW did not differ from predicted BW in either group (P > 0.880). The HP piglets exhibited greater ADG and ADFI from 6 to 10 wk and greater BW at birth and 6 and 10 wk of age (P ≤ 0.002) compared with the PP piglets, whereas G:F ratio was similar (P = 0.417). Superior growth performance of HP piglets was associated with a 1.27-fold higher IGF1 plasma concentration at 10 wk compared with the PP piglets (P = 0.044). The greater weight and muscle cross-sectional area of STN in HP piglets was due to a 1.20-fold increase in total muscle fiber number (TFN; P = 0.009) and 1.34-fold increase in fiber cross-sectional area (FCSA; P = 0.004) compared with the PP piglets. The number of myonuclei per red and intermediate fiber was greater in HP piglets (P ≤ 0.097), but the nucleus-to-cytoplasm ratio was unaffected by the performance group (P = 0.861). The mRNA expression of proliferating cell nuclear antigen (PCNA), paired box 7 (PAX7), myogenic factor 5 (MYF5), and myogenic differentiation factor (MYOD) did not differ between groups (P ≥ 0.327). However, IGF2-specific mRNA expression was numerically higher in the HP piglets (P = 0.101). The greater myofiber number, the higher degree of myofiber hypertrophy, and the increased muscular mRNA expression of IGF2 indicate that HP piglets exhibit a greater capacity for lean accretion and may grow faster until market weight. In summary, pigs that were selected for predicted high BW at 10 wk of age using a complex selection model had a superior muscularity in terms of greater TFN and FCSA, which may be of advantage for lean mass accretion in later life and for meat quality. INTRODUCTION In recent years, within-litter variation in birth weight of piglets has increased because of a greater number of piglets born per sow per year (Foxcroft et al., 2007). However, farmers want to have a homogeneous litter at the end of the nursery period. Body weight at the end of the nursery period (10 wk) can be accurately predicted using an algorithm based on the data analysis of 3 data sets (n = 77,868 individual records) where we identified season of birth, and BW at birth, at weaning, and at 6 wk of age as main factors for the prediction (Paredes et al., 2012b). From the literature it can be derived that BW or muscle weight at birth mainly correlates with the number of myofibers formed during prenatal myogenesis (Gondret et al., 2006). In contrast, postnatal muscle growth is mostly associated with the hypertrophy of the existing fibers (reviewed by Rehfeldt et al., 2000), which in turn is associated with satellite cell activity and myonuclear accumulation (Allen et al., 1979). The emphasis on fiber number or fiber size alone could limit lean mass because of their trait antagonism (Rehfeldt et al., 2004). Therefore, it was of interest to determine how the predicted high or poor performance in piglets by the model of Paredes et al. (2012b) would be reflected in muscle microstructure. The concerted action of muscle development and growth is regulated by the IGF system (Oksbjerg et al., 2004; Kalbe et al., 2008) and the muscle regulatory factors (MRF; te Pas and Soumillion, 2001). Our own studies revealed that IGF2 could play a promoting role in both prenatal myogenesis and postnatal muscle growth (Kalbe et al., 2013). The aim of this study was to analyze muscle microstructure as well as mRNA expression of candidate genes of muscle growth and development in predicted high- or poor-performing piglets at 10 wk of age on the basis of BW at birth, at weaning, and at 6 wk of age and to determine whether muscle characteristics would play a role in the growth retardation. MATERIALS AND METHODS The experimental procedures with animals described in this study were conducted at the Nutreco Swine Research Centre (Sint Anthonis, the Netherlands) and were approved by the Animal Care and Use Committee of Utrecht University (Utrecht, the Netherlands). Animals The piglets used for this study originated from 35 litters born from October 26 to 30, 2011, from Hypor Libra sows (first to fourth parity) bred to TOPIGS P-line sires. One day before the expected farrowing date, all sows were treated with an intramuscular injection of 5 mg dinoprost (Dynolitic, Zoetis, Capelle aan den IJssel, the Netherlands) to synchronize the farrowing process. The average number of piglets born alive per litter was 13.5 ± 2.6. All piglets (n = 454) were weighed individually and identified immediately after parturition with an ear tag. At the same time piglets received a 0.3-mL intramuscular injection of ampicillin (Albipen, Intervet International, B.V., Boxmeer, the Netherlands) and 100 mg iron (Ursoferran, Serumwerk Bernburg AG, Bernburg, Germany). Birth date, individual time of birth of each piglet, and total duration of parturition were recorded. Cross-fostering was avoided unless a sow did not have a sufficient number of teats to accommodate all piglets born. This procedure was based on sow functioning mammary glands, and this was checked when the sows entered the farrowing room. In this case, the heaviest piglets of the litter were moved to a foster sow within 48 h after birth. Males were anaesthetized and castrated on d 5 of age according to farm procedures. Piglets were weaned at 21.6 ± 1.0 d of age and transferred to the nursery facilities, where they remained with their littermates until they reached 6 wk of age. Because of morbidity and mortality in the farrowing room, only 430 piglets remained at weaning. Selection of High- and Poor-Performing Piglets Piglets with a birth weight below 2 SD from the mean of the total population were excluded from the experiment. These piglets were considered as intrauterine growth-retarded animals (McMillen et al., 2001; Paredes et al., 2012b). To exclude animals with a clinically low health status as a factor for poor performance, 2 d before start of the trial (40 ± 1 d of age), all piglets were judged on alertness, appearance of a round belly, presence of nasal and eye secretions, ocular and oral mucosa color, and hair brightness. On d 41 and 69 of age ± 1 d, 10 mL of blood were collected from the jugular vein for determination of kidney (glucose, creatinine, sodium, potassium, calcium, phosphate, urea, total protein) and liver (albumin, direct and indirect bilirubin, alkaline phosphatase, glutamate dehydrogenase, total protein, glutamic pyruvic transaminase, γ-glutamyl transpeptidase, glutamic oxaloacetic transaminase) function. For general overview we analyzed creatine kinase, magnesium, chloride, and iron. To discard animals with signs of inflammation, acute phase proteins (haptoglobin and C-reactive protein) were determined. Animals in both groups were regarded as healthy by visual observation and veterinarian check throughout the study. Results for liver and kidney profiles were in agreement with the normal ranges as described by Kraft and Dürr (2005) for piglets of 6 and 10 wk of age. Acute phase protein results were within the normal range for 6- and 10-wk-old piglets (Biocheck GmbH, Leipzig, Germany). From weaning to 6 wk of age 62 piglets were excluded because of health problems. At 6 wk of age, 368 piglets were available for selection of the experimental piglets. On the basis of the algorithm developed by Paredes et al. (2012b) we included BW at birth, weaning, and 6 wk of age to predict the weight at the end of the nursery phase (10 wk of age) in 2 divergent populations of piglets (high performing, HP; poor performing, PP; each n = 30) balanced for sex and litter of origin. The 2 experimental populations were selected when the piglets were 6 wk (41.6 ± 1.0 d) of age. Piglets were regarded as PP if their predicted BW at 10 wk of age was between 16.0 and 22.9 kg and as HP piglets if their predicted BW at 10 wk of age was between 26.8 and 30.9 kg. Animal Housing and Feeding Piglets selected for the study (n = 60) were placed individually in 0.77 m2 pens and randomly distributed over 3 similar climate-controlled departments with 20 pens each. Piglets had ad libitum access to feed and water. During the first 24 h after arrival to the experimental facility lights were on continuously. Thereafter, a 16/8 h light/dark scheme was provided. Piglets were able to interact through the barred pen divisions and were offered toys to enhance welfare. All piglets had access to a commercial creep feed (Trouw Nutrition, Gent, Belgium) presented as gruel (2:1 water:feed), replaced twice a day to stimulate feed intake and supplied from d 14 of age until weaning. From weaning until 10 d postweaning all piglets were fed with a commercial weaner diet (Trouw Nutrition). From d 11 to 14 after weaning a gradual transition (75:25, 50:50, 25:75, 0:100) to a second diet (for composition see Table 1) took place. The experimental diet was formulated to meet or exceed requirements (NRC, 1998) using highly digestible protein sources. Table 1. Ingredient and calculated nutrient composition of the experimental diet (as fed) Item  Value  Ingredient, g/kg feed        Barley  250.0      Wheat  195.0      Corn  150.0      Oat groats  100.0      Whey protein concentrate  80.0      Soy protein concentrate  48.0      Potato protein  25.0      Sweet whey powder  24.6      Soya oil  23.0      Beet pulp  20.0      Sugar  20.0      Amino acid blend1  18.5      Vitamin and mineral blend2  12.0      Coconut oil  10.0      Sodium bicarbonate  6.5      Mono calcium phosphate  5.8      Limestone  4.0      Calcium formeate  3.0      Citric acid  3.0      Sodium chloride  1.0      Choline chloride blend  0.5      Phytase  0.1  Calculated composition3        CP, %  17.5      Ether extract, %  5.6      Crude fiber, %  3.1      Lactose, %  6.0      Net energy, MJ  10.7      AID lysine,4 %  1.4      AID methionine, %  0.5      AID methionine + cysteine, %  0.8      AID threonine, %  0.8      AID tryptophan, %  0.3      AID valine, %  0.9      Digestible phosphorus, %  0.4  Item  Value  Ingredient, g/kg feed        Barley  250.0      Wheat  195.0      Corn  150.0      Oat groats  100.0      Whey protein concentrate  80.0      Soy protein concentrate  48.0      Potato protein  25.0      Sweet whey powder  24.6      Soya oil  23.0      Beet pulp  20.0      Sugar  20.0      Amino acid blend1  18.5      Vitamin and mineral blend2  12.0      Coconut oil  10.0      Sodium bicarbonate  6.5      Mono calcium phosphate  5.8      Limestone  4.0      Calcium formeate  3.0      Citric acid  3.0      Sodium chloride  1.0      Choline chloride blend  0.5      Phytase  0.1  Calculated composition3        CP, %  17.5      Ether extract, %  5.6      Crude fiber, %  3.1      Lactose, %  6.0      Net energy, MJ  10.7      AID lysine,4 %  1.4      AID methionine, %  0.5      AID methionine + cysteine, %  0.8      AID threonine, %  0.8      AID tryptophan, %  0.3      AID valine, %  0.9      Digestible phosphorus, %  0.4  1Provided the following per kilogram of diet: methionine, 0.31%; lysine, 0.83%; threonine, 0.32%; tryptophan, 0.11%; and L-valine, 0.27%. 2Provided the following per kilogram of diet: vitamin A, 8,000 IU; vitamin D3, 2,000 IU; vitamin E, 30 mg; pantothenic acid, 12 mg; vitamin K3, 1.5 mg; vitamin B1, 1 mg; vitamin B2, 4 mg; vitamin B6, 1 mg; vitamin B12, 20 μg; nicotinic acid, 20 mg; folic acid, 0.3 mg; cobalt, 0.15 mg as basic cobaltous carbonate monohydrate; copper, 160 mg as cupric sulfate pentahydrate; iron, 100 mg as ferrous sulfate monohydrate; iodine, 1 mg as calcium iodate anhydrous; manganese, 30 mg as manganese oxide; zinc, 100 mg as zinc sulfate; selenium, 0.3 mg as sodium selenite. 3Calculated composition based on Product Board Animal Feed (2007) values. 4AID = apparent ileal digestibility (Product Board Animal Feed, 2007). View Large Table 1. Ingredient and calculated nutrient composition of the experimental diet (as fed) Item  Value  Ingredient, g/kg feed        Barley  250.0      Wheat  195.0      Corn  150.0      Oat groats  100.0      Whey protein concentrate  80.0      Soy protein concentrate  48.0      Potato protein  25.0      Sweet whey powder  24.6      Soya oil  23.0      Beet pulp  20.0      Sugar  20.0      Amino acid blend1  18.5      Vitamin and mineral blend2  12.0      Coconut oil  10.0      Sodium bicarbonate  6.5      Mono calcium phosphate  5.8      Limestone  4.0      Calcium formeate  3.0      Citric acid  3.0      Sodium chloride  1.0      Choline chloride blend  0.5      Phytase  0.1  Calculated composition3        CP, %  17.5      Ether extract, %  5.6      Crude fiber, %  3.1      Lactose, %  6.0      Net energy, MJ  10.7      AID lysine,4 %  1.4      AID methionine, %  0.5      AID methionine + cysteine, %  0.8      AID threonine, %  0.8      AID tryptophan, %  0.3      AID valine, %  0.9      Digestible phosphorus, %  0.4  Item  Value  Ingredient, g/kg feed        Barley  250.0      Wheat  195.0      Corn  150.0      Oat groats  100.0      Whey protein concentrate  80.0      Soy protein concentrate  48.0      Potato protein  25.0      Sweet whey powder  24.6      Soya oil  23.0      Beet pulp  20.0      Sugar  20.0      Amino acid blend1  18.5      Vitamin and mineral blend2  12.0      Coconut oil  10.0      Sodium bicarbonate  6.5      Mono calcium phosphate  5.8      Limestone  4.0      Calcium formeate  3.0      Citric acid  3.0      Sodium chloride  1.0      Choline chloride blend  0.5      Phytase  0.1  Calculated composition3        CP, %  17.5      Ether extract, %  5.6      Crude fiber, %  3.1      Lactose, %  6.0      Net energy, MJ  10.7      AID lysine,4 %  1.4      AID methionine, %  0.5      AID methionine + cysteine, %  0.8      AID threonine, %  0.8      AID tryptophan, %  0.3      AID valine, %  0.9      Digestible phosphorus, %  0.4  1Provided the following per kilogram of diet: methionine, 0.31%; lysine, 0.83%; threonine, 0.32%; tryptophan, 0.11%; and L-valine, 0.27%. 2Provided the following per kilogram of diet: vitamin A, 8,000 IU; vitamin D3, 2,000 IU; vitamin E, 30 mg; pantothenic acid, 12 mg; vitamin K3, 1.5 mg; vitamin B1, 1 mg; vitamin B2, 4 mg; vitamin B6, 1 mg; vitamin B12, 20 μg; nicotinic acid, 20 mg; folic acid, 0.3 mg; cobalt, 0.15 mg as basic cobaltous carbonate monohydrate; copper, 160 mg as cupric sulfate pentahydrate; iron, 100 mg as ferrous sulfate monohydrate; iodine, 1 mg as calcium iodate anhydrous; manganese, 30 mg as manganese oxide; zinc, 100 mg as zinc sulfate; selenium, 0.3 mg as sodium selenite. 3Calculated composition based on Product Board Animal Feed (2007) values. 4AID = apparent ileal digestibility (Product Board Animal Feed, 2007). View Large Piglet Performance and Sampling Piglet performance was analyzed weekly during the observational period (6 to 10 wk of age). Piglets were weighed at birth, at 1 d of age (21.7 ± 0.8 h), at weaning (21.6 ± 1.0 d), and weekly during the postweaning period until the end of the observational period (69.6 ± 1.0 d). Feed intake per animal was determined daily and calculated on a weekly basis during the observational period. At 10 wk of age, 20 piglets per group category were randomly selected to be sacrificed with an intracardiac injection of sodium pentobarbital (Euthasol 40%, ASTfarma B.V., Oudewater, the Netherlands) at a dose of 200 mg/kg BW. The right semitendinosus muscle (STN) was excised within 5 min postmortem, and its weight and circumference were recorded. Samples from the central portion of the muscle midbelly were snap-frozen in liquid nitrogen and thereafter stored at –80°C for histochemical and gene expression analyses. The STN muscle was used because it is a constituent of the ham as an economically important primal cut and because of its suitability for estimation of total fiber number (TFN). Muscle Histochemistry and Microscopy Muscle cross-sectional area (MCSA) of STN was calculated from the circumference of the muscle midbelly. Serial transverse sections of 10 µm were cut at –20°C in a cryostat (Leica, Nussloch, Germany). One section was stained for cytoplasm and nuclei with eosin and haematoxylin, respectively. Another section was exposed to the reaction for NADH tetrazolium reductase (Novikoff et al., 1961), which enables classification into red oxidative, intermediate oxidative, and white glycolytic fibers. Muscle structure was evaluated using a computerized image analysis system (AMBA, IBSB, Berlin, Germany) consisting of a camera, a microscope, and custom-made software specified for muscle structure analysis. Measurements were performed in the interactive mode. Using 100× magnification, 4 random fields of view were captured from each section. In total, 300 fibers were analyzed per sample. First, the eosin/haematoxylin-stained section was used to draw the fiber outlines and to detect myonuclei. Then, the resulting framework was superimposed on the photomicrograph of the NADH tetrazolium reductase–stained section, and each fiber was assigned to 1 of the 3 fiber types. The software displayed the proportion of each fiber type, fiber cross-sectional area (FCSA) for each fiber type and average, the number of nuclei per fiber for each type and average, and the FCSA per nucleus for each fiber type and average. For the estimation of TFN the fiber number per square centimeter was calculated (from average FCSA) and subsequently multiplied by the MCSA of the STN muscle. Microscopic analyses were all done by the same person. RNA Isolation, Reverse Transcription, and quantitative PCR of Myogenesis-Associated Genes Total RNA was isolated from STN muscle tissue with a RNeasy Fibrous Tissue Mini Kit (Qiagen, Hilden, Germany), as recommended by the supplier. This procedure includes the removal of genomic DNA with RNase-free DNase. RNA was quantified in a NanoDrop instrument (Peqlab, Erlangen, Germany). The quality of all RNA samples was assessed by the A260/A280 ratio and by denaturing agarose (1%) gel electrophoresis. In addition, from randomly selected samples (n = 27) the quality was monitored using the Experion Automated Electrophoresis System (Biorad, Munich, Germany) according to the manufacturer's protocol. All samples analyzed were classified by an RNA quality indicator (RQI; 10 = intact RNA, 1 = highly degraded RNA) in the best category (7 < RQI ≤ 10). Reverse transcription (RT) was performed with 2 µg of total RNA preparation, a mixture (2:1) of random primer p(dN)6 and anchored oligo (dT)18 primer (Roche, Mannheim, Germany), and Moloney mouse leukemia virus reverse transcriptase (M-MLV RT RNase H Minus Point Mutant, Promega, Mannheim, Germany) in 25 μL of the incubation buffer provided by the supplier, supplemented with deoxynucleoside triphosphates (deoxy-NTPs) (Roche) and RNasin (Promega), for 60 min at 42°C. The freshly synthesized cDNA samples were cleaned with the High Pure PCR Product Purification Kit (Roche) and eluted in 50 μL elution buffer. Transcript expression of myogenic factor 5 (MYF5), myogenic differentiation factor (MYOD), insulin growth factor 2 (IGF2), paired box 7 (PAX7), proliferating cell nuclear antigen (PCNA), and TATA box binding protein (TBP) genes was measured. For quantitative PCR (qPCR), 1.25 μL of each purified cDNA sample was amplified in duplicate with the LightCycler-FastStart DNA MasterPLUS SYBR Green I kit (Roche) in 10 μL total reaction volume. Primer information was described elsewhere [Kalbe et al. (2008): IGF2; Rehfeldt et al. (2012a): MYF5, MYOD; Erkens et al. (2006): TBP; Patruno et al. (2008a): PAX7]. Primers used to amplify 173 bp from PCNA (GenBank accession number DQ473295) mRNA were 5′- ACGCTAAGGGCAGAAGATAATGCAG (forward) and 5′- CGTGCAAATTCACCAGAAGGCATC (reverse). All primers were purchased from Sigma-Genosys (Steinheim, Germany) and, if possible, were derived from different exons to avoid amplification of residual genomic DNA. Amplification and quantification of generated products were performed in a LightCycler instrument 2.0 (Roche) under the following cycling conditions: preincubation at 95°C for 10 min, followed by 40 cycles denaturation at 95°C for 15 s, annealing for 10 s at 57°C (IGF2), 58°C (PAX7), 59°C (TBP), or 60°C (MYF5, MYOD, PCNA), extension at 72°C for 10 s, and single-point fluorescence acquisition for 6 s to avoid quantification of primer artifacts. The melting peaks of all samples were routinely determined by melting curve analysis to ascertain that only the expected products had been generated. Additionally, molecular sizes of PCR products were monitored by agarose gel electrophoresis analysis (not shown). To normalize for variations between individual LightCycler runs, an arbitrarily selected sample was coamplified as the calibrator. The relative quantification was performed with the LightCycler software version 4.5 using the quantification module: relative quantification − monocolor. Therefore, the relative expression ratio of the target gene is calculated on the basis of the PCR efficiencies and the crossing point deviation of an unknown sample vs. the calibrator and is expressed in comparison to an endogenous reference gene (TBP) as described by Pfaffl (2001). The TBP expression was not affected by group (P = 0.750) or by sex (P = 0.400). To calculate the PCR efficiency, routine dilutions of the gene-specific external standard (cloned PCR products) of known concentrations covering 5 orders of magnitude (5 × 10–16 to 5 × 10–12 g DNA) were coamplified during each run. Sequencing was performed with the automated sequencing system ABI PRISM 310 genetic analyzer using the ABI PRISMBig Dye kit (both from PE Applied Biosystems, Weiterstadt, Germany). Analysis of IGF1 in Plasma One day before piglets were euthanized, blood samples were obtained by puncture of the jugular vein after 1 h of fasting. Approximately 2.7 mL of blood were collected into EDTA tubes and centrifuged at 172 × g for 10 min at 4°C. Plasma samples were stored at –20°C before analyses. Plasma IGF1 concentrations were determined using a competitive electrochemiluminescence immunoassay as described by Rehfeldt et al. (2001). Statistical Analysis The UNIVARIATE procedure of SAS (version 9.1, 2002; SAS Inst. Inc., Cary, NC) was used to test residuals for normality with P-values > 0.05 for the Shapiro-Wilk test indicating the normal distribution of the data. Differences between groups in performance traits (ADG, ADFI, and G:F), muscle characteristics (microstructure and mRNA expression), and plasma IGF1 were evaluated using the MIXED procedure in SAS. The statistical model included group (HP or PP), sex, and their interaction (group × sex) as fixed factors. Differences between calculated and real BW at 10 wk of age were analyzed using the MIXED procedure of SAS including group (HP or PP) and method of BW determination (predicted or measured BW). The correlation between IGF1 and ADG was assessed by a Pearson correlation test using the CORR procedure in SAS. In all cases, pig was the experimental unit. Significant differences were identified at P < 0.05 and trends at P < 0.10. RESULTS Performance There was no difference in predicted and realized BW at 10 wk of age [HP = 29.2 kg (range: 26.8 to 30.9 kg) predicted vs. 29.8 ± 0.8 kg (range: 27.3 to 31.0 kg) measured, P = 0.880; PP = 19.1 kg (range: 16.0 to 22.9 kg) predicted vs. 19.0 ± 0.8 kg (range: 16.0 to 21.9 kg) measured; P = 0.912]. High-performing piglets displayed greater BW than piglets of the PP group at birth and 6 and 10 wk of age (P ≤ 0.002; Table 2). Table 2. Body weights, growth performance, and plasma IGF1 concentration of predicted high (HP, n = 20) and poor (PP, n = 20) performing piglets from 6 until 10 wk (least squares means ± SE)   Group  Sex  P-value  Item  HP  PP  SE  Castrate1  Female  SE  Group  Sex  BW, kg                      Birth  1.6  1.2  0.1  1.3  1.5  0.1  0.002  0.045      6 wk2  12.2  6.8  0.2  9.5  9.5  0.2  <0.001  0.870      10 wk3  29.8  19.0  0.8  25.0  23.8  0.8  <0.001  0.280  ADG 6 to 10 wk, g/d  629  436  25  554  510  26  <0.001  0.219  ADFI 6 to 10 wk, g/d  901  635  29  793  743  30  <0.001  0.229  G:F 6 to 10 wk  0.69  0.68  0.01  0.69  0.68  0.01  0.417  0.309  Plasma IGF13, ng/mL  431  339  31.2  393  378  32.5  0.044  0.727    Group  Sex  P-value  Item  HP  PP  SE  Castrate1  Female  SE  Group  Sex  BW, kg                      Birth  1.6  1.2  0.1  1.3  1.5  0.1  0.002  0.045      6 wk2  12.2  6.8  0.2  9.5  9.5  0.2  <0.001  0.870      10 wk3  29.8  19.0  0.8  25.0  23.8  0.8  <0.001  0.280  ADG 6 to 10 wk, g/d  629  436  25  554  510  26  <0.001  0.219  ADFI 6 to 10 wk, g/d  901  635  29  793  743  30  <0.001  0.229  G:F 6 to 10 wk  0.69  0.68  0.01  0.69  0.68  0.01  0.417  0.309  Plasma IGF13, ng/mL  431  339  31.2  393  378  32.5  0.044  0.727  1Castrate = castrated males. 26 wk = start of the observational period, 41.6 ± 1.0 d of age, when piglets were individually housed. 310 wk = end of the observational period, 69.6 ± 1.0 d of age View Large Table 2. Body weights, growth performance, and plasma IGF1 concentration of predicted high (HP, n = 20) and poor (PP, n = 20) performing piglets from 6 until 10 wk (least squares means ± SE)   Group  Sex  P-value  Item  HP  PP  SE  Castrate1  Female  SE  Group  Sex  BW, kg                      Birth  1.6  1.2  0.1  1.3  1.5  0.1  0.002  0.045      6 wk2  12.2  6.8  0.2  9.5  9.5  0.2  <0.001  0.870      10 wk3  29.8  19.0  0.8  25.0  23.8  0.8  <0.001  0.280  ADG 6 to 10 wk, g/d  629  436  25  554  510  26  <0.001  0.219  ADFI 6 to 10 wk, g/d  901  635  29  793  743  30  <0.001  0.229  G:F 6 to 10 wk  0.69  0.68  0.01  0.69  0.68  0.01  0.417  0.309  Plasma IGF13, ng/mL  431  339  31.2  393  378  32.5  0.044  0.727    Group  Sex  P-value  Item  HP  PP  SE  Castrate1  Female  SE  Group  Sex  BW, kg                      Birth  1.6  1.2  0.1  1.3  1.5  0.1  0.002  0.045      6 wk2  12.2  6.8  0.2  9.5  9.5  0.2  <0.001  0.870      10 wk3  29.8  19.0  0.8  25.0  23.8  0.8  <0.001  0.280  ADG 6 to 10 wk, g/d  629  436  25  554  510  26  <0.001  0.219  ADFI 6 to 10 wk, g/d  901  635  29  793  743  30  <0.001  0.229  G:F 6 to 10 wk  0.69  0.68  0.01  0.69  0.68  0.01  0.417  0.309  Plasma IGF13, ng/mL  431  339  31.2  393  378  32.5  0.044  0.727  1Castrate = castrated males. 26 wk = start of the observational period, 41.6 ± 1.0 d of age, when piglets were individually housed. 310 wk = end of the observational period, 69.6 ± 1.0 d of age View Large Average daily gain and ADFI from 6 to 10 wk of age were about 1.4-fold greater in the HP piglets compared with the PP piglets (P < 0.001). However, no differences in the computed G:F ratio between groups were noted (P = 0.417). At birth, female piglets were heavier than the males (P = 0.045), but no sex effect on the other growth performance traits was observed. No significant group × sex interactions were apparent for any of the performance traits (P ≥ 0.225). The concentrations of plasma IGF1 were 1.27-fold greater (P = 0.044) in the HP piglets when compared with the PP piglets (Table 2). No effects of sex (P = 0.727) and no interaction of sex × group (P = 0.751) were observed. In addition, there was a positive correlation between plasma IGF1 and ADG during the observational period (Table 3). This positive correlation was more pronounced in PP piglets (r = 0.585, P = 0.007) than in HP piglets (r = 0.473, P = 0.035). Table 3. Pearson's correlations between plasma IGF1 concentration at 10 wk of age and ADG during the experimental period (6 to 10 wk of age) within groups of predicted high (HP, n = 20) and poor (PP, n = 20) performing piglets at 10 wk of age and across the total number of observations1 Item  HP  PP  Total  r  0.473  0.585  0.576  P-value  0.035  0.007  <0.001  Item  HP  PP  Total  r  0.473  0.585  0.576  P-value  0.035  0.007  <0.001  110 wk = end of the observational period, 69.6 ± 1.0 d of age. View Large Table 3. Pearson's correlations between plasma IGF1 concentration at 10 wk of age and ADG during the experimental period (6 to 10 wk of age) within groups of predicted high (HP, n = 20) and poor (PP, n = 20) performing piglets at 10 wk of age and across the total number of observations1 Item  HP  PP  Total  r  0.473  0.585  0.576  P-value  0.035  0.007  <0.001  Item  HP  PP  Total  r  0.473  0.585  0.576  P-value  0.035  0.007  <0.001  110 wk = end of the observational period, 69.6 ± 1.0 d of age. View Large Structural Properties of Skeletal Muscle The weight of the STN muscle was 1.7-fold greater in HP piglets compared with PP piglets (116.9 ± 4.1 vs. 69.7 ± 4.1 g; P < 0.001). However, relative muscle weight expressed as a percentage of body weight was not greater in the HP piglets than in the PP piglets (P = 0.151). High-performing piglets had a 58% larger STN MCSA compared with PP piglets and differed in selected microstructural characteristics of STN muscle at 10 wk of age. Thus, HP piglets exhibited a 20% greater TFN (P = 0.009) and a 34% larger FCSA (P = 0.004) compared with the PP piglets (Table 4). The difference in average FCSA resulted from nearly equal relative differences in the FCSA of red, intermediate, and white fibers. There was a numerically higher number of nuclei per fiber (P = 0.102) in the HP piglets compared with the PP piglets, which was mainly due to differences in red (P = 0.087) and intermediate (P = 0.097) fibers. No differences in the FCSA per nucleus were found, which means there was an unchanged nuclei-to-cytoplasm ratio with increasing fiber size. The performance group of the piglets did not affect the proportions of the fiber types. The sex of the piglets had no influence on any microstructural muscle characteristics examined (P ≥ 0.142) with the exception of the proportion of red fibers (P ≥ 0.093), which tended to be greater in castrated males than in females. No interactions of sex × group were detectable (P ≥ 0.163). Table 4. Microstructural characteristics of the right semitendinosus muscle of predicted high (HP, n = 17) and poor (PP, n = 15) performing piglets at 10 wk of age (least squares means ± SE)1   Group  Sex  P-value  Item  HP  PP  SE  Castrate2  Female  SE  Group  Sex  Weight, g  116.9  69.7  4.1  93.0  93.5  4.2  <0.001  0.928  Circumference, cm  13.9  11.0  0.3  12.5  12.4  0.3  <0.001  0.827  MCSA,3 cm2  15.5  9.8  0.6  12.7  12.6  0.6  <0.001  0.912  Total fiber number, thousands  799  666  35  754  712  35  0.009  0.385  Proportion of fiber types, %                      Red  31.2  27.6  2.7  32.6  26.2  2.7  0.330  0.093      Intermediate  25.9  24.6  1.8  25.8  24.6  1.8  0.587  0.623      White  43.2  47.7  3.4  42.0  48.9  3.4  0.335  0.142  FCSA,4µm2                      Red  1,527  1,155  87  1,329  1,353  87  0.004  0.837      Intermediate  2,244  1,643  127  1,960  1,927  127  0.002  0.849      White  2,244  1,661  154  1,977  1,928  154  0.010  0.816      Average  2,009  1,499  122  1,738  1,770  122  0.005  0.847  No. of nuclei per                      Red fiber  0.9  0.7  0.1  0.8  0.8  0.1  0.087  0.803      Intermediate fiber  0.9  0.7  0.1  0.8  0.7  0.1  0.097  0.538      White fiber  0.8  0.6  0.1  0.7  0.7  0.1  0.134  0.620      Average  0.8  0.6  0.1  0.7  0.7  0.1  0.102  0.792  Average FCSA per nucleus, µm2  409  416  31  426  399  31  0.861  0.510    Group  Sex  P-value  Item  HP  PP  SE  Castrate2  Female  SE  Group  Sex  Weight, g  116.9  69.7  4.1  93.0  93.5  4.2  <0.001  0.928  Circumference, cm  13.9  11.0  0.3  12.5  12.4  0.3  <0.001  0.827  MCSA,3 cm2  15.5  9.8  0.6  12.7  12.6  0.6  <0.001  0.912  Total fiber number, thousands  799  666  35  754  712  35  0.009  0.385  Proportion of fiber types, %                      Red  31.2  27.6  2.7  32.6  26.2  2.7  0.330  0.093      Intermediate  25.9  24.6  1.8  25.8  24.6  1.8  0.587  0.623      White  43.2  47.7  3.4  42.0  48.9  3.4  0.335  0.142  FCSA,4µm2                      Red  1,527  1,155  87  1,329  1,353  87  0.004  0.837      Intermediate  2,244  1,643  127  1,960  1,927  127  0.002  0.849      White  2,244  1,661  154  1,977  1,928  154  0.010  0.816      Average  2,009  1,499  122  1,738  1,770  122  0.005  0.847  No. of nuclei per                      Red fiber  0.9  0.7  0.1  0.8  0.8  0.1  0.087  0.803      Intermediate fiber  0.9  0.7  0.1  0.8  0.7  0.1  0.097  0.538      White fiber  0.8  0.6  0.1  0.7  0.7  0.1  0.134  0.620      Average  0.8  0.6  0.1  0.7  0.7  0.1  0.102  0.792  Average FCSA per nucleus, µm2  409  416  31  426  399  31  0.861  0.510  110 wk = end of the observational period, 69.6 ± 1.0 d of age. 2Castrate = castrated males. 3MCSA = muscle cross-sectional area. 4FCSA = fiber cross-sectional area. View Large Table 4. Microstructural characteristics of the right semitendinosus muscle of predicted high (HP, n = 17) and poor (PP, n = 15) performing piglets at 10 wk of age (least squares means ± SE)1   Group  Sex  P-value  Item  HP  PP  SE  Castrate2  Female  SE  Group  Sex  Weight, g  116.9  69.7  4.1  93.0  93.5  4.2  <0.001  0.928  Circumference, cm  13.9  11.0  0.3  12.5  12.4  0.3  <0.001  0.827  MCSA,3 cm2  15.5  9.8  0.6  12.7  12.6  0.6  <0.001  0.912  Total fiber number, thousands  799  666  35  754  712  35  0.009  0.385  Proportion of fiber types, %                      Red  31.2  27.6  2.7  32.6  26.2  2.7  0.330  0.093      Intermediate  25.9  24.6  1.8  25.8  24.6  1.8  0.587  0.623      White  43.2  47.7  3.4  42.0  48.9  3.4  0.335  0.142  FCSA,4µm2                      Red  1,527  1,155  87  1,329  1,353  87  0.004  0.837      Intermediate  2,244  1,643  127  1,960  1,927  127  0.002  0.849      White  2,244  1,661  154  1,977  1,928  154  0.010  0.816      Average  2,009  1,499  122  1,738  1,770  122  0.005  0.847  No. of nuclei per                      Red fiber  0.9  0.7  0.1  0.8  0.8  0.1  0.087  0.803      Intermediate fiber  0.9  0.7  0.1  0.8  0.7  0.1  0.097  0.538      White fiber  0.8  0.6  0.1  0.7  0.7  0.1  0.134  0.620      Average  0.8  0.6  0.1  0.7  0.7  0.1  0.102  0.792  Average FCSA per nucleus, µm2  409  416  31  426  399  31  0.861  0.510    Group  Sex  P-value  Item  HP  PP  SE  Castrate2  Female  SE  Group  Sex  Weight, g  116.9  69.7  4.1  93.0  93.5  4.2  <0.001  0.928  Circumference, cm  13.9  11.0  0.3  12.5  12.4  0.3  <0.001  0.827  MCSA,3 cm2  15.5  9.8  0.6  12.7  12.6  0.6  <0.001  0.912  Total fiber number, thousands  799  666  35  754  712  35  0.009  0.385  Proportion of fiber types, %                      Red  31.2  27.6  2.7  32.6  26.2  2.7  0.330  0.093      Intermediate  25.9  24.6  1.8  25.8  24.6  1.8  0.587  0.623      White  43.2  47.7  3.4  42.0  48.9  3.4  0.335  0.142  FCSA,4µm2                      Red  1,527  1,155  87  1,329  1,353  87  0.004  0.837      Intermediate  2,244  1,643  127  1,960  1,927  127  0.002  0.849      White  2,244  1,661  154  1,977  1,928  154  0.010  0.816      Average  2,009  1,499  122  1,738  1,770  122  0.005  0.847  No. of nuclei per                      Red fiber  0.9  0.7  0.1  0.8  0.8  0.1  0.087  0.803      Intermediate fiber  0.9  0.7  0.1  0.8  0.7  0.1  0.097  0.538      White fiber  0.8  0.6  0.1  0.7  0.7  0.1  0.134  0.620      Average  0.8  0.6  0.1  0.7  0.7  0.1  0.102  0.792  Average FCSA per nucleus, µm2  409  416  31  426  399  31  0.861  0.510  110 wk = end of the observational period, 69.6 ± 1.0 d of age. 2Castrate = castrated males. 3MCSA = muscle cross-sectional area. 4FCSA = fiber cross-sectional area. View Large Gene Expression in Skeletal Muscle The mRNA expression of PCNA, PAX7, MYF5, and MYOD did not differ between groups (Table 5). However, IGF2-specific mRNA expression was numerically higher in the HP piglets than in the PP piglets (P = 0.101). No effect of sex and no group × sex interactions were observed for the gene-specific mRNA expressions. Table 5. The mRNA expression of selected genes in semitendinosus muscle of predicted high (HP, n = 20) and poor (PP, n = 20) performing piglets at 10 wk of age (least squares means ± SE)1   Group  Sex  P-value  Gene2  HP  PP  SE  Castrate3  Female  SE  Group  Sex  PCNA  1.12  1.09  0.07  1.15  1.06  0.08  0.814  0.410  PAX7  1.35  1.63  0.38  1.69  1.29  0.40  0.596  0.456  MYF5  1.05  0.88  0.12  1.06  0.88  0.13  0.327  0.295  MYOD  1.26  1.35  0.17  1.16  1.45  0.17  0.716  0.218  IGF2  0.77  0.62  0.06  0.76  0.64  0.06  0.101  0.179    Group  Sex  P-value  Gene2  HP  PP  SE  Castrate3  Female  SE  Group  Sex  PCNA  1.12  1.09  0.07  1.15  1.06  0.08  0.814  0.410  PAX7  1.35  1.63  0.38  1.69  1.29  0.40  0.596  0.456  MYF5  1.05  0.88  0.12  1.06  0.88  0.13  0.327  0.295  MYOD  1.26  1.35  0.17  1.16  1.45  0.17  0.716  0.218  IGF2  0.77  0.62  0.06  0.76  0.64  0.06  0.101  0.179  1Data are expressed as arbitrary units after normalization by the endogenous reference gene TATA box binding protein (TBP). 10 wk = end of the observational period, 69.6 ± 1.0 d of age. 2PCNA, proliferating cell nuclear antigen; PAX7, paired box 7; MYF5, myogenic factor 5; MYOD, myogenic differentiation factor; IGF2, insulin growth factor 2. 3Castrate = castrated males. View Large Table 5. The mRNA expression of selected genes in semitendinosus muscle of predicted high (HP, n = 20) and poor (PP, n = 20) performing piglets at 10 wk of age (least squares means ± SE)1   Group  Sex  P-value  Gene2  HP  PP  SE  Castrate3  Female  SE  Group  Sex  PCNA  1.12  1.09  0.07  1.15  1.06  0.08  0.814  0.410  PAX7  1.35  1.63  0.38  1.69  1.29  0.40  0.596  0.456  MYF5  1.05  0.88  0.12  1.06  0.88  0.13  0.327  0.295  MYOD  1.26  1.35  0.17  1.16  1.45  0.17  0.716  0.218  IGF2  0.77  0.62  0.06  0.76  0.64  0.06  0.101  0.179    Group  Sex  P-value  Gene2  HP  PP  SE  Castrate3  Female  SE  Group  Sex  PCNA  1.12  1.09  0.07  1.15  1.06  0.08  0.814  0.410  PAX7  1.35  1.63  0.38  1.69  1.29  0.40  0.596  0.456  MYF5  1.05  0.88  0.12  1.06  0.88  0.13  0.327  0.295  MYOD  1.26  1.35  0.17  1.16  1.45  0.17  0.716  0.218  IGF2  0.77  0.62  0.06  0.76  0.64  0.06  0.101  0.179  1Data are expressed as arbitrary units after normalization by the endogenous reference gene TATA box binding protein (TBP). 10 wk = end of the observational period, 69.6 ± 1.0 d of age. 2PCNA, proliferating cell nuclear antigen; PAX7, paired box 7; MYF5, myogenic factor 5; MYOD, myogenic differentiation factor; IGF2, insulin growth factor 2. 3Castrate = castrated males. View Large DISCUSSION The aim of the present study was to determine whether a large difference in piglet performance, high or poor, at 10 wk of age is reflected in skeletal muscle development and phenotype. We focus on the central role of the skeletal muscle as a tissue because it represents about 50% of body mass in piglets. The question arises whether muscle growth capacity might be the limiting factor for animal performance. In addition, it was of interest whether the HP and PP piglets chosen by the model of Paredes et al. (2012b) also differed in skeletal muscle properties. Previous studies demonstrated the importance of TFN for postnatal growth and carcass and meat quality in pigs (Rehfeldt et al., 2000; Bee, 2004). Because the STN is considered a standard muscle for the determination of TFN in pigs, muscle properties were analyzed in this muscle. Animal Performance High-performing piglets showed greater ADG and greater ADFI during the 4-wk observational period, resulting in similar computed G:F. The present study represents a subset of piglets (n = 40) from a parallel study (n = 60; Paredes et al., 2012a). Our observations regarding growth performance are in agreement with the aforementioned study. As shown by Paredes et al. (2012a), the differences in performance characteristics were not linked to nutrient digestibility. Superior growth performance of HP piglets was associated with a greater IGF1 plasma concentration, which is in line with the positive correlation of circulating IGF1 with growth rate in pigs (Buonomo et al., 1987). On the other hand, a greater plasma IGF1 concentration can also result from greater feed intake in pigs (Dauncey et al., 1990). Muscle Microstructure The algorithm used for the selection of the experimental piglets (Paredes et al., 2012b) provides a correct prediction of BW at 10 wk of age. This selection may also yield information on muscle structure because the groups selected for distinct growth performance also differed in TFN and FCSA. In addition, enhanced fiber hypertrophy was associated with a proportional increase in the number of myonuclei resulting in unchanged nucleus-to-cytoplasm ratio. In other cases such an enhanced muscle growth may reduce the nucleus-to-cytoplasm ratio of myofibers (Rehfeldt, 2007). Remarkably, the HP piglets exhibited both a greater TFN and a greater FCSA. In general, TFN and FCSA are inversely correlated with each other at the end of the intensive growth period, meaning that the growth of individual fibers is slower when a large number of fibers is present and vice versa (reviewed by Rehfeldt et al., 2000). However, both traits correlate positively with MCSA and thus contribute to the accretion of lean mass (Rehfeldt et al., 2004). Thus, those piglets not following this trait antagonism and simultaneously exhibiting high TFN and FCSA will grow faster, as realized by our HP piglets. Fiber number and size are influenced by genetic factors (species, gender, breed) and environmental factors such as age and nutrition (reviewed by Rehfeldt et al., 2004). Prenatal nutrition is of specific importance because inadequate nutrient supply in utero not only retards fetal growth but also impairs myogenesis. Piglets light at birth exhibit a smaller TFN than their heavier littermates (Handel and Stickland, 1987; Dwyer and Stickland, 1991; Gondret et al., 2006; Rehfeldt and Kuhn, 2006). Although a further increase in TFN occurs during the first postnatal weeks (Bérard et al., 2011), these light piglets exhibit a lower TFN throughout life (Rehfeldt and Kuhn, 2006) and display a lower postnatal growth rate (Wolter et al., 2002; Rehfeldt et al., 2008; Fix et al., 2010). Postnatal lean growth depends on the number of fibers and on the degree of fiber hypertrophy. The capacity limit for lean accretion is set by the number of fibers, as this is the prerequisite for postnatal muscle growth via fiber hypertrophy, which in turn does not continue beyond a certain limit. Once maximum fiber size is achieved, nutritional energy can no longer be used for protein accretion and is deposited as fat instead. This relationship may explain why piglets of low birth weight have deposited more fat and less lean in the carcasses until market weight and exhibit a poorer meat quality (Bee, 2004; Gondret et al., 2005, 2006; Rehfeldt et al., 2008). Therefore, one important criterion of genetic selection should be birth weight to favor piglets with a good capacity of muscle growth, even though some light piglets exhibit similar fiber numbers as heavier piglets (Dwyer et al., 1993; Gondret et al., 2005; Rehfeldt and Kuhn, 2006). In the case that low birth weight was associated with impaired prenatal muscle development, lean accretion until slaughter would be reduced (Rehfeldt et al., 2012a,b). In addition, Dwyer et al. (1993) stated that birth weight influences growth rate during the early stages of postnatal growth, whereas growth from d 70 of age until slaughter depended on TFN. When pigs were categorized within litter on the basis of their weight at slaughter (Nissen et al., 2004), the low- and medium-weight pigs had similar birth weights and TFN, but the medium-weight pigs had greater FCSA. The pigs with the highest slaughter weight and lean mass exhibited the highest TFN but had similar fiber hypertrophy as medium-weight pigs. That study showed that both TFN and a high degree of fiber hypertrophy are the prerequisites for fast postnatal growth. Our current divergent selection model included birth weight and BW at weaning and at 6 wk of age and generated 2 groups of piglets with differences in both TFN and FCSA at 10 wk of age. Our model therefore may predict differences in growth performance and carcass quality at market age. Taken together, a complex selection model seems to be superior over using only birth weight to predict further growth performance of the pigs. It still remains to be investigated whether a predicted BW at 10 wk of age is an indicator for good carcass quality at market age. Expression of Myogenesis-Associated Genes Postnatal muscle growth at the age of 10 wk is considered to result from hypertrophy by increasing diameter and length of the existing myofibers (Swatland and Cassens, 1972). Hypertrophic processes are attributed to satellite cells, which provide new nuclei to growing myofibers (Mauro, 1961). These stem cells are located between the plasmalemmal membrane and the basal lamina of the myofiber (Zammit and Beauchamp, 2001). Beside their anatomic location, quiescent satellite cells are characterized by the expression of PAX7 (Seale et al., 2000). Activated satellite cells coexpress the muscle regulatory factors MYOD (Yablonka-Reuveni and Rivera, 1994) or MYF5 (Kuang et al., 2007), which are markers for myogenic progenitor cells. Our study revealed no differences in the mRNA expression of PAX7, suggesting that the number of satellite cells is similar in HP and PP piglets at 10 wk of age. Other studies revealed conflicting results. Ropka-Molik et al. (2011) found greater PAX7 mRNA expression in gracilis muscle of the more muscular Pietrain pigs compared with Polish Landrace at 210 d of age, whereas Wang et al. (2012) showed opposite effects comparing satellite cell cultures derived from LM of Lantang pigs with those from Landrace pigs. In addition to muscularity, PAX7 expression may be related to muscular maturity, which is suggested from studies of Patruno et al. (2008b), who observed differences in PAX7 mRNA expression of myogenic cell cultures derived from STN of pigs at different ages. An equal proliferative activity of satellite cells in muscle of HP and PP piglets is suggested by their comparable PCNA mRNA expression. PCNA is considered to be a proliferation marker because the mRNA accumulates only in proliferating cells (Chang et al., 1990), such as satellite cells, adipocytes, and nerve cells in skeletal muscle tissue. In addition, equal numbers of satellite cells are supported by the similar nucleus-to-cytoplasm ratio between HP and PP piglets. This means that the number of myonuclei increased at the same rate as FCSA in HP piglets. Another indication of a comparable muscular maturity of the HP and PP piglets in our study is the lack of differences in the MYF5 and MYOD expression. The mRNA expression of both MRF is related to satellite cell activity at postnatal ages (Koishi et al., 1995; te Pas et al., 2000; Patruno et al., 2008a). Therefore, their expression is lower at postnatal than prenatal ages (Patruno et al., 2008a). It is difficult to compare our results with other studies because the expression values are age and muscle (and muscle region) specific. For instance, te Pas et al. (2000) analyzed MYF5 and MYOD mRNA in adult pigs selected for fast growth or for leanness and found no differences in the mRNA expression in the red region (near bone) of STN, whereas in the white region more MYF5 and MYOD mRNA was detected in fast growing pigs compared with lean-selected ones. Therefore, they concluded that the expression of MRF depends more on the selection for overall growth than for muscle deposition per se. In addition, Ropka-Molik et al. (2011) detected increased MYOD mRNA in different muscles of Polish Landrace pigs compared with Pietrain pigs at market weight. Nevertheless, from our mRNA expression analyses of genes encoding satellite-cell-associated transcription factors we could not conclude on a different muscle phenotype of HP and PP piglets at 10 wk of age. It is known that IGF2 stimulates both proliferation and differentiation of muscle cells (reviewed in Florini et al., 1996; Oksbjerg et al., 2004). In general, the mRNA expression of IGF2 in porcine skeletal muscle tissue is high in the embryo/fetus and declines in the perinatal/postnatal period (Lee et al., 1993; Gerrard et al., 1998). Nevertheless, the importance of IGF2 for postnatal skeletal muscle growth in pigs is evident from previous studies, although the underlying mechanisms are not completely understood. A paternally imprinted mutation in the IGF2 gene resulted in an increase in IGF2 mRNA expression in pigs (van Laere et al., 2003; Stinckens et al., 2007). This was associated with increased lean meat percentage related to postnatal muscle hypertrophy due to an increase in muscle fiber diameter and a higher proliferative capacity of satellite cells (van den Maagdenberg et al., 2008a,b). Furthermore, Rehfeldt et al. (2012b) found reduced IGF2 mRNA expression in LM of pigs at market age that originated from gilts fed low (6%) protein diets during gestation. Those pigs exhibited an opposite phenotype compared with that described by van den Maagdenberg et al. (2008a,b) because of impaired myogenesis with fewer myofibers and less lean meat and more fat in pigs of large litters. In our study, the HP piglets revealed a tendency for more IGF2 mRNA in skeletal muscle than PP piglets. This fits with their heavier STN weight and the greater MCSA, TFN, and number of myonuclei at 10 wk of age. Examining the skeletal muscle properties of HP piglets together with the knowledge of the IGF2-associated phenotypes, there is evidence that HP piglets will exhibit more lean mass than PP piglets at market age. In conclusion, pigs that were selected for predicted high or low BW at 10 wk of age using a complex selection model differed in TFN, FCSA, and myonuclear numbers. As indicated by the greater myofiber number, the higher degree of myofiber hypertrophy, and the increased muscular mRNA expression of IGF2, the HP piglets possess a greater capacity for lean accretion and are expected to exhibit faster growth until market weight and have a greater lean proportion in the carcass. This, however, remains to be investigated. The unchanged nucleus-to-cytoplasm ratio and mRNA expression of studied genes may suggest that HP and PP piglets grow in a similar manner but on different BW levels during the observational period between 6 and 10 wk of age. In addition, how the growth of PP pigs could be stimulated should be investigated. The most reasonable approaches to counteract the incidence of PP pigs could be optimizing the conditions in utero for the development of muscle and genetic selection for a simultaneous increase of TFN and FCSA. LITERATURE CITED Allen R. E. Merkel R. A. Young R. B. 1979. Cellular aspects of muscle growth: Myogenic cell proliferation. J. Anim. Sci.  49: 115– 127. Google Scholar CrossRef Search ADS PubMed  Bee G 2004. Effect of early gestation feeding, birth weight, and gender of progeny on muscle fiber characteristics of pigs at slaughter. J. Anim. Sci.  82: 826– 836. Google Scholar CrossRef Search ADS PubMed  Bérard J. Kalbe C. Lösel D. Tuchscherer A. Rehfeldt C. 2011. Potential sources of early-postnatal increase in myofibre number in pig skeletal muscle. Histochem. Cell Biol.  136: 217– 225. Google Scholar CrossRef Search ADS PubMed  Buonomo F. C. Lauterio T. J. Baile C. A. Campion D. R. 1987. Determination of insulin-like growth factor 1 (IGF1) and IGF binding protein levels in swine. Domest. Anim. Endocrinol..  4: 23– 31. Google Scholar CrossRef Search ADS PubMed  Chang C. D. Ottavio L. Travali S. Lipson K. E. Baserga R. 1990. Transcriptional and posttranscriptional regulation of the proliferating cell nuclear antigen gene. Mol. Cell. Biol.  10: 3289– 3296. Google Scholar CrossRef Search ADS PubMed  Dauncey M. J. Shakespear R. A. Rudd B. T. Ingram D. L. 1990. Variations in somatomedin-C/insulin-like growth factor-I associated with environmental temperature and nutrition. Horm. Metab. Res.  22: 261– 264. Google Scholar CrossRef Search ADS PubMed  Dwyer C. M. Fletcher J. M. Stickland N. C. 1993. Muscle cellularity and postnatal growth in the pig. J. Anim. Sci.  71: 3339– 3343. Google Scholar CrossRef Search ADS PubMed  Dwyer C. M. Stickland N. C. 1991. Sources of variation in myofiber number and within and between litters of pigs. Anim. Prod.  52: 527– 533. Google Scholar CrossRef Search ADS   Erkens T. van Poucke M. Goossens K. van Zeveren A. Peelman L. J. 2006. Development of a new set of reference genes for normalization of real-time RT-PCR data of porcine backfat and longissimus dorsi muscle, and evaluation with PPARGCIA. BMC Biotechnol.  6: 41. Google Scholar CrossRef Search ADS PubMed  Fix J. S. Cassady J. P. Herring W. O. Holl J. W. Culbertson M. S. See M. T. 2010. Effect of piglet birth weight on body weight, growth, backfat, and longissimus muscle area of commercial market swine. Livest. Sci.  127: 51– 59. Google Scholar CrossRef Search ADS   Florini J. R. Ewton D. Z. Coolican S. A. 1996. Growth hormone and the insulin-like growth factor system in myogenesis. Endocr. Rev.  17: 481– 517. Google Scholar PubMed  Foxcroft G. Bee G. Dixon W. Hahn M. Harding J. Patterson J. Putman T. Sarmento S. Smit M. Tse W. Town S. 2007. Consequences of selection for litter size on piglet development. In: Wiseman J. Varley M. A. McOrist S. Kemp B. editors, Paradigms of pig science.  Nottingham Univ. Press, Nottingham, UK. p. 207– 229. Gerrard D. E. Okamura C. S. Ranalletta M. A. Grant A. L. 1998. Developmental expression and location of IGF-I and IGF-II mRNA and protein in skeletal muscle. J. Anim. Sci.  76: 1004– 1011. Google Scholar CrossRef Search ADS PubMed  Gondret F. Lefaucheur L. Juin H. Louveau I. Lebret B. 2006. Low birth weight is associated with enlarged muscle fiber area and impaired meat tenderness of the longissimus muscle in pigs. J. Anim. Sci.  84: 93– 103. Google Scholar CrossRef Search ADS PubMed  Gondret F. Lefaucheur L. Louveau I. Lebret B. Pichodo X. Le Cozler Y. 2005. Influence of piglet birth weight on postnatal growth performance, tissue lipogenic capacity and muscle histological traits at market weight. Livest. Prod. Sci.  93: 137– 146. Google Scholar CrossRef Search ADS   Handel S. E. Stickland N. C. 1987. Muscle cellularity and birth weight. Anim. Prod.  44: 311– 317. Google Scholar CrossRef Search ADS   Kalbe C. Bérard J. Porm M. Rehfeldt C. Bee G. 2013. Maternal l-arginine supplementation during early gestation affects foetal skeletal myogenesis in pigs. Livest. Sci.  157: 322– 329. Google Scholar CrossRef Search ADS   Kalbe C. Mau M. Rehfeldt C. 2008. Developmental changes and the impact of isoflavones on mRNA expression of IGF-I receptor, EGF receptor and related growth factors in porcine skeletal muscle cell cultures. Growth Horm. IGF Res.  18: 424– 433. Google Scholar CrossRef Search ADS PubMed  Koishi K. Zhang M. McLennan I. S. Harris A. J. 1995. MyoD protein accumulates in satellite cells and is neurally regulated in regenerating myotubes and skeletal muscle fibers. Dev. Dyn.  202: 244– 254. Google Scholar CrossRef Search ADS PubMed  Kraft W. Dürr U. 2005. Klinische Labordiagnostik in der Tiermedizin. Schattauer GmbH., Stuttgart, Germany. p. 534. Kuang S. Kuroda K. Le Grand F. Rudnicki M. A. 2007. Asymmetric self-renewal and commitment of satellite stem cells in muscle. Cell.  129: 999– 1010. Google Scholar CrossRef Search ADS PubMed  Lee C. Y. Chung C. S. Simmen F. A. 1993. Ontogeny of the porcine insulin-like growth factor system. Mol. Cell. Endocrinol..  93: 71– 80. Google Scholar CrossRef Search ADS PubMed  Mauro A 1961. Satellite cell of skeletal muscle fibers. J. Biophys. Biochem. Cytol.  9: 493– 495. Google Scholar CrossRef Search ADS PubMed  McMillen I. C. Adams M. B. Ross J. T. Coulter C. L. Simonettta G. Owens J. A. Robinson J. S. Edwards L. J. 2001. Foetal growth restriction: Adaptations and consequences. Reproduction  122: 195– 204. Google Scholar CrossRef Search ADS PubMed  Nissen P. M. Jorgensen P. F. Oksbjerg N. 2004. Within-litter variation in muscle fiber characteristics, pig performance, and meat quality traits. J. Anim. Sci.  82: 414– 421. Google Scholar CrossRef Search ADS PubMed  Novikoff A. B. Shin W. Drucker J. 1961. Mitochondrial localization of oxidative enzymes: Staining results with two tetrazolium salts. J. Biophys. Biochem. Cytol.  9: 47– 61. Google Scholar CrossRef Search ADS PubMed  NRC 1998. Nutrient requirements of swine. 10th rev. ed. Natl. Acad. Press, Washington, DC. Oksbjerg N. Gondret F. Vestergaard M. 2004. Basic principles of muscle development and growth in meat-producing mammals as affected by the insulin-like growth factor (IGF) system. Domest. Anim. Endocrinol.  27: 219– 240. Google Scholar CrossRef Search ADS PubMed  Paredes S. P. Awati A. Jansman A. J. M. van Hees H. M. J. Verstegen M. W. A. Gerrits W. J. J. 2012a. Feed intake, rather than digestion is the growth-limiting factor in poor performing piglets. In: Proc. XII Int. Symp. Digestive Physiol. Pigs, Denver, CO. p. 121. Paredes S. P. Jansman A. J. M. Verstegen M. W. A. Awati A. Buist W. den Hartog L. A. van Hees H. M. J. Quiniou N. Hendriks W. H. Gerrits W. J. J. 2012b. Analysis of factors to predict piglet body weight at the end of the nursery phase. J. Anim. Sci.  90: 3243– 3251. Google Scholar CrossRef Search ADS   Patruno M. Caliaro F. Maccatrozzo L. Sacchetto R. Martinello T. Toniolo L. Regginani C. Mascarello F. 2008a. Myostatin shows a specific expression pattern in pig skeletal and extraocular muscles during pre-and post-natal growth. Differentiation  76: 168– 181. Google Scholar CrossRef Search ADS   Patruno M. Caliaro F. Martinello T. Mascarello F. 2008b. Expression of the paired box domain Pax7 protein in myogenic cells isolated from the porcine semitendinosus muscle after birth. Tissue Cell.  40: 1– 6. 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. Google Scholar CrossRef Search ADS PubMed  Product Board Animal Feed 2007. CVB chemical composition and nutritional values of feed ingredients. Prod. Board Anim. Feed, Zoetermeer, the Netherlands. p. 681. Rehfeldt C 2007. Satellite cell addition is/is not obligatory for skeletal muscle hypertrophy. J. Appl. Physiol..  103: 1104– 1106. Google Scholar CrossRef Search ADS PubMed  Rehfeldt C. Fiedler I. Dietl G. Ender K. 2000. Myogenesis and postnatal skeletal muscle cell growth as influenced by selection. Livest. Prod. Sci.  66: 177– 188. Google Scholar CrossRef Search ADS   Rehfeldt C. Fiedler I. Stickland N. C. 2004. Number and size of muscle fibres in relation to meat production. In: te Pas M. F. W. Everts M. E. Haagsman H. P. editors, Muscle development of livestock animals.  CAB Int., Wallingford, UK. p. 1– 38. Google Scholar CrossRef Search ADS   Rehfeldt C. Kuhn G. 2006. Consequences of birth weight for postnatal growth performance and carcass quality in pigs as related to myogenesis. J. Anim. Sci.  84: E113– E123. Google Scholar CrossRef Search ADS PubMed  Rehfeldt C. Kuhn G. Nürnberg G. Kanitz E. Schneider F. Beyer M. Nürnberg K. Ender K. 2001. Effects of exogenous somatotropin during early gestation on maternal performance, fetal growth and compositional traits in pigs. J. Anim. Sci.  79: 1789– 1799. Google Scholar CrossRef Search ADS PubMed  Rehfeldt C. Lefaucheur L. Block J. Stabenow B. Pfuhl R. Otten W. Metges C. C. Kalbe C. 2012a. Limited and excess protein intake of pregnant gilts differently affects body composition and cellularity of skeletal muscle and subcutaneous adipose tissue of newborn and weanling piglets. Eur. J. Nutr.  51: 151– 165. Google Scholar CrossRef Search ADS   Rehfeldt C. Stabenow B. Pfuhl R. Block J. Nürnberg G. Otten W. Metges C. C. Kalbe C. 2012b. Effects of limited and excess protein intakes of pregnant gilts on carcass quality and cellular properties of skeletal muscle and subcutaneous adipose tissue in fattening pigs. J. Anim. Sci.  90: 184– 196. Google Scholar CrossRef Search ADS   Rehfeldt C. Tuchscherer A. Hartung M. Kuhn G. 2008. A second look at the influence of birth weight on carcass and meat quality in pigs. Meat Sci.  78: 170– 175. Google Scholar CrossRef Search ADS PubMed  Ropka-Molik K. Eckert R. Piorkowska K. 2011. The expression pattern of myogenic regulatory factors MyoD, Myf6 and Pax7 in postnatal porcine skeletal muscles. Gene Expr. Patterns.  11: 79– 83. Google Scholar CrossRef Search ADS PubMed  Seale P. Sabourin L. A. Girgis-Gabardo A. Mansouri A. Gruss P. Rudnicki M. A. 2000. Pax7 is required for the specification of myogenic satellite cells. Cell  102: 777– 786. Google Scholar CrossRef Search ADS PubMed  Stinckens A. van den Maagdenberg K. Luyten T. Georges M. de Smet S. Buys N. 2007. The RYR1 g.1843C > T mutation is associated with the effect of the IGF2 intron3-g.3072G > A mutation on muscle hypertrophy. Anim. Genet.  38: 67– 71. Google Scholar CrossRef Search ADS PubMed  Swatland H. J. Cassens R. G. 1972. Muscle growth: The problem of muscle fibers with an intrafascicular termination. J. Anim. Sci.  35: 336– 344. Google Scholar CrossRef Search ADS PubMed  te Pas M. F. W. Soumillion A. 2001. Improvement of livestock breeding strategies using physiologic and functional genomic information of the muscle regulatory factors gene family for skeletal muscle development. Curr. Genomics.  2: 285– 304. Google Scholar CrossRef Search ADS   te Pas M. F. Verburg F. J. Gerritsen C. L. de Greef K. H. 2000. Messenger ribonucleic acid expression of the MyoD gene family in muscle tissue at slaughter in relation to selection for porcine growth rate. J. Anim. Sci.  78: 69– 77. Google Scholar CrossRef Search ADS PubMed  van den Maagdenberg K. Stinckens A. Claeys E. Buys N. de Smet S. 2008a. Effect of the insulin-like growth factor-II and RYR1 genotype in pigs on carcass and meat quality traits. Meat Sci.  80: 293– 303. Google Scholar CrossRef Search ADS   van den Maagdenberg K. Stinckens A. Lefaucheur L. Buys N. de Smet S. 2008b. The effect of mutations in the insulin-like growth factor-II and ryanodine receptor-1 genes on biochemical and histochemical muscle fibre characteristics in pigs. Meat Sci.  79: 757– 766. Google Scholar CrossRef Search ADS   van Laere A. Nguyen M. Braunschweig M. Nezer C. Collette C. Moreau L. Archibald A. L. Haley C. Buys N. Tally M. Andersson G. Georges M. Andersson L. 2003. A regulatory mutation in IGF2 causes a major QTL effect on muscle growth in the pig. Nature  425: 832– 836. Google Scholar CrossRef Search ADS PubMed  Wang X. Q. Yang W. J. Yang Z. Shu G. Wang S. B. Jiang Q. Y. Yuan L. Wu T. S. 2012. The differential proliferative ability of satellite cells in Lantang and Landrace pigs. PLoS ONE  7( 3): E32537. Google Scholar CrossRef Search ADS PubMed  Wolter B. F. Ellis M. Corrigan B. P. DeDecker J. M. 2002. The effect of birth weight and feeding of supplemental milk replacer to piglets during lactation on preweaning and postweaning growth performance and carcass characteristics. J. Anim. Sci.  80: 301– 308. Google Scholar CrossRef Search ADS PubMed  Yablonka-Reuveni Z. Rivera A. J. 1994. Temporal expression of regulatory and structural muscle proteins during myogenesis of satellite cells on isolated adult rat fibers. Dev. Biol.  164: 588– 603. Google Scholar CrossRef Search ADS PubMed  Zammit P. S. Beauchamp J. R. 2001. The skeletal muscle satellite cell: Stem cell or son of stem cell? Differentiation  68: 193– 204. Google Scholar CrossRef Search ADS PubMed  American Society of Animal Science

Onoda, T.;Yamamoto, R.;Sawamura, K.;Inoue, Y.;Murase, H.;Nambo, Y.;Tozaki, T.;Matsui, A.;Miyake, T.;Hirai, N.

doi: 10.2527/jas.2013-6523pmid: 24085406

Abstract Thoroughbred horses are seasonal mating animals, and their foals are born yearly in spring seasons. In northern regions or countries, the foals generally show a typical seasonal compensatory growth pattern, where their growth rate declines in winter and increases in the next spring. In this study, a new empirical approach is proposed to adjust for this compensatory growth when growth curve equations are estimated, by using BW of Japanese Thoroughbred colts and fillies raised in Hidaka, Hokkaido. Based on the traditional Richards growth curve equation, new growth curve equations were developed and fit to the weight–age data. The foals generally experience 2 major winter seasons before their debut in horseracing. The new equations had sigmoid subfunctions that can empirically adjust the first and second year compensatory growths, combined with the Richards biological parameter responsible for the maturity of animals. The unknown parameters included in the equations were estimated by SAS NLMIXED procedure. The goodness-of-fit was examined by using several indices of goodness-of-fit (i.e., Akaike's information criterion, Bayesian information criterion, –2 log likelihood, and residual sum of squares) for the multiple applications of the subfunctions. The indices indicated the best fit of the new equations including both subfunctions for the first and second compensatory growths to the weight–age data. The shapes of the growth curves were improved during the periods of compensatory growth. The proposed method is one of the useful approaches for adjusting multiple seasonal compensatory growths in growth curve estimations of Thoroughbreds and for the management of young horses during the compensatory periods. INTRODUCTION The Thoroughbred horse, a well-known horse breed for horseracing worldwide, is a seasonal breeder and engages in his or her mating activities during spring time so that their foals born 340 d later can take advantage of the milder temperatures and abundant forage. Due to the seasonal mating, young Thoroughbred foals experience their first winter season almost simultaneously. They exhibit seasonal compensatory growth (CG) pattern that is characterized by a reduced growth rate in the winter seasons and an accelerated growth rate in the next springs. The seasonal CG is clearly found in the change of BW or average daily gain of young Thoroughbreds especially in northern regions or countries (Brown-Douglas and Pagan, 2006). In Japan, the Equine Research Institute, Japan Racing Association (JRA), published standardized growth curves of BW for young Japanese Thoroughbreds based on the plots of longitudinal averages of the actual weight–age data up to 25 mo covering the first winter season (Equine Research Institute, 2004). The CG patterns were clearly observed in their growth curves. Recently, an empirical adjustment approach has been proposed to adjust a single (i.e., first year only) seasonal CG when growth curve equations are estimated in Japanese Thoroughbreds (Onoda et al., 2011). For the adjustment of the single CG, a new sigmoid subfunction was developed and combined with the traditional Richards growth curve equation. Thoroughbred horses generally experience 2 major winter seasons, first and second year seasons, before their debut in horseracing. Multiple applications of the proposed approach would be useful for considering multiple CG in Thoroughbred horses. During these multiple periods of CG, careful feeding management is required to ensure the optimum development of young Thoroughbred horses because the relationship between BW gain and feeding amount tends to be unbalanced in these periods. The objective of this study was to estimate the growth curve equations empirically adjusted for multiple (i.e., first and second year) seasonal CG for BW of young Japanese Thoroughbreds based on the approach of Onoda et al. (2011). MATERIALS AND METHODS All animal procedures were approved by the Animal Care and Use Committee at the JRA Hidaka Training and Research Center, and procedures for handling horses complied with those specified in the Basic Guidelines for Comprehensively Promoting Measures on the Welfare and Management of Animals (MEJ, 2006). Data Description For 39 Thoroughbred colts and 42 fillies, respectively, collected by Hidaka Training and Research Center, JRA, between 1999 and 2008, we analyzed a total of 3,961 and 4,341 BW (kilograms) and age (days) measurements. The maximum age in the data was about 1,100 d, which covers 2 major winter seasons before the foals' debut in horseracing. The frequency distributions of birth months of the foals are shown in Fig. 1. Figure 1. View largeDownload slide Frequency distribution of the analyzed young horses by birth months Figure 1. View largeDownload slide Frequency distribution of the analyzed young horses by birth months The JRA Hidaka Training and Research Center is located in Hidaka region in Hokkaido, the northern island in Japan. The Hidaka region is famous for the intensive production of racehorses where there are 815 racehorse farms corresponding to 82% of all Japanese racehorse farms (JBBA, 2012). The annual average temperature in Hidaka is 6.0°C (°F = 1.8 × °C + 32) and average monthly temperature for January and February are minus 7.9 and 7.2°C, respectively (JMA, 2013). Everywhere is covered by snow in Hidaka during the winter, and green pastures only begin to grow in the middle of May. During winter seasons, the foals are housed alone in stalls without room heating. Due to this coldness in winter, the typical CG phenomenon appears in the growth of foals. The weaning months of the foals is about 5 to 6 mo. Taming training for the foals begins generally in August to October of their yearling year. The rations in winter are hay and grains fed based on the foals' age in months after consulting the Japanese Feeding Standard for Horses (Equine Research Institute, 2004). Body weights of foals were measured weekly in the morning by using a 1 t load-cell type scale (Kubota Corporation, Osaka, Japan) and two 1.5 t electric balances (Mettler-Toledo International, Inc., Tokyo, Japan) for pre- and postweaning periods, respectively. Growth Models In our previous study (Onoda et al., 2011), 7 sigmoid growth curve equations (Logistic, Gompertz, von Bertalanffy, Brody, Richards, Bridges, and Janoscheck; see Köhn et al., 2007) were compared using Akaike's information criterion (AIC; Akaike, 1973). The chosen equation was the Richards equation (Richards, 1959) as follows:  In this traditional Richards equation, BW (kilograms) is described as a function of age (t, age in days). Staniar et al. (2004) discussed the biological interpretations of the 4 unknown parameters (A, B, k, and m). Briefly, A is the asymptotic limit for BW as t approaches infinity (i.e., mature BW), B is a scaling parameter that defines the degree of maturity, k is a rate constant that determines the spread of the curve along the time axis, and m is the point of inflection in the sigmoid curve in relation to age. Due to the lack of mature BW data of the analyzed foals, the mature BW (i.e., parameter A) of Japanese Thoroughbreds was fixed as 575.0 kg for both sexes based on The Japanese Feeding Standard for Horses (Equine Research Institute, 2004). The other parameters (B, k, and m) are estimated by the SAS NLMIXED (SAS Inst. Inc., Cary, NC) procedure separately in each sex. We used the Richards equation as the main growth curve equation in this study. Adjustments of Compensatory Growth To adjust the changes of growth rate during the first and second CG, the B parameter (i.e., degree of maturity) of the traditional Richards equation was modified by multiple applications of subfunctions [f′(t) and f″(t)] as follows:  in which  and  Based on Onoda et al. (2011), the sigmoid subfunctions f′(t) and f″(t) for the CG adjustment were developed as follows. A general sigmoid function f(t) is expressed as  in which e is the base of the natural logarithm, t is time, in which –∞ < t < ∞, α is a parameter determining the shape of the function, and 0 ≤ f(t) ≤ 1. A sigmoid function is a monotonous increase function, and the shapes of the function depend on the α values. The subtraction of 2 sigmoid functions with different α values leads to a function having only a single wave (down and up wave) such as Fig. 2. We chose arbitrary α values of 5 and 10. By using the subtraction between the 2 sigmoid functions with several modifications of the t parameter, the following sigmoid subfunction f′(t) in Eq. [2] above is obtained:  and the shape of the subfunction f′(t) is shown in Fig. 2. The f′(t) value crosses the horizontal axis at 432 d with the wave range of 268.49 × 2 d. These values of 432 and 268.49 are commonly estimated in both sexes. The 432 d is the averaging day of the first CG for all horses and determined by the crossing point between the actual data averages of all data with both sexes and traditional Richards equation, based on our previous study (Onoda et al., 2011). The half of the wave range, 268.49 d, is also determined by the preliminary parameter estimation by SAS NLMIXED procedure using the data of both sexes. This subfunction has zero value when t = –∞ or ∞. Only when t is around the averaging day of CG (i.e., around 432 d) do the typical nonzero f′(t) values appear as in Fig. 2. Then it was assumed that the f′(t) subfunction can empirically explain the changes of growth rate caused by CG (i.e., decreased growth rate in winter and increased growth rate in spring). Figure 2. View largeDownload slide Constructed sigmoid subfunction f′(t) for adjustment of a single seasonal compensatory growth with the averaging day of the compensatory growth of 432 d. Figure 2. View largeDownload slide Constructed sigmoid subfunction f′(t) for adjustment of a single seasonal compensatory growth with the averaging day of the compensatory growth of 432 d. The developed f′(t) subfunction was adapted for the B parameter, because B is a major maturity related parameter, as explained above. This subfunction is applied multiply for the B parameter to explain multiple CG with different winter seasons with minor modifications. For the construction of the subfunction f″(t) for the second year seasonal CG, the 432 d is simply modified as 797 (= 432 + 365) as shown in Eq. [2] above, assuming that the second year seasonal CG comes just 1 yr after from the first CG. The coefficient B′ and B″ in Eq. [2] are estimated in the parameter estimation. Parameter Estimation and Model Comparison Unknown parameters in the equation were estimated by the SAS NLMIXED procedure. The fundamental parameters (B, k, and m) were initially estimated using Eq. [1], and then B′ and B″ coefficients for subfunctions were estimated using Eq. [2] by fixing the estimated fundamental parameter values. This 2-step estimation is practical and computationally feasible. It is also useful for getting information about the main (i.e., baseline) growth curve without CG and additional modifications of the baseline due to CG. After the parameter estimation, indices of goodness-of-fit were calculated and used for the model comparisons among the models with or without subfunctions for CG. The chosen indices were AIC, Bayesian information criterion (BIC), –2 log likelihood, and the average of residual sum of squares (RSS). RESULTS AND DISCUSSION The scatter plot of the weight–age data is shown with gray dots in Fig. 3. In the figure, the tendencies of the seasonal CG are recognized around 432 d (and slightly 797 d) corresponding to the first and second CG periods, respectively. Figure 3. View largeDownload slide Scatter plot of the BW data of 3,961 colts (A) and 4,341 fillies (B) of Japanese Thoroughbreds (gray dots). Dashed lines indicate the averaging days of the periods of compensatory growth: 432 d (left) and 797 d (right) Figure 3. View largeDownload slide Scatter plot of the BW data of 3,961 colts (A) and 4,341 fillies (B) of Japanese Thoroughbreds (gray dots). Dashed lines indicate the averaging days of the periods of compensatory growth: 432 d (left) and 797 d (right) The newly developed f′(t) and f″(t) sigmoid subfunctions were simply added to the B parameter in Eq. [1] with the coefficients B′ and B″. Specifically, this leads to the replacement of B by B + B′f′(t) + B″f″(t) as shown in Eq. [2]. With these replacements and parameter estimation in each sex, the growth curve equations for male and female BW considering the 2 seasonal CG were obtained as  and  respectively. The subfunctions f′(t) and f″(t) were common in both sexes and the same as Eq. [2]. In this study, considering the consistency of the established growth curve equations of our previous studies, the parameter B and B′ were based on Onoda et al. (2011). Equations [5] and [6] look complicated but are a continuous single variable function of t (age in days). The insertions of f′(t) and f″(t) affect the BW only when t is within about 432 ± 268.49 d or 797 ± 268.49 d, respectively. The growth curves of the new Richards equation combined with these sigmoid subfunctions (i.e., Eq. [5] and [6]) are shown with black lines in Fig. 4, and they express the changes of growth rate in the first and second CG periods. Departures of the prematuring BW (i.e., BW around 1,000 d) in the analyzed data from the estimated growth curves are recognized due to the fixed usage of the maturing BW of 575.0 kg in this study. The body size of the recent colts and fillies are somewhat larger than the Japanese Thoroughbreds population standards on The Japanese Feeding Standard for Horses published earlier (Equine Research Institute, 2004). Figure 4. View largeDownload slide Scatter plot of the BW data of 3,961 colts (A) and 4,341 fillies (B) of Japanese Thoroughbreds (gray dots). Black lines indicate estimated growth curves (Eq. [5] and [6] for colts and fillies, respectively) Figure 4. View largeDownload slide Scatter plot of the BW data of 3,961 colts (A) and 4,341 fillies (B) of Japanese Thoroughbreds (gray dots). Black lines indicate estimated growth curves (Eq. [5] and [6] for colts and fillies, respectively) For male equations, the AIC value of traditional Richards equation (Eq. [5] without all subfunctions) was 33,666 (Table 1). The AIC value of the equation adjusting only the first seasonal CG {Eq. [5] without f″(t)} was 33,107 and showed a better fit than the traditional Richards equation. Based on the complete male equation of Eq. [5], the AIC value also decreased to 32,942, indicating the best fit to the actual weight–age data. For females, the complete equation of Eq. [6] showed the best fit, too. Concerning BIC, –2 log likelihood, and RSS, the same tendencies for the goodness-of-fit were recognized. In these model comparisons, the parameter values in each model were fixed as in Eq. [5] and [6]. Even if all the parameters were re-estimated by using each model, the same tendencies were also recognized. These results showed the strong evidence of the presence of both first and second CG in the analyzed Japanese Thoroughbred horse population. Table 1. Akaike's information criterion (AIC), Bayesian information criterion (BIC), –2 log likelihood (–2 Log L), and average of residual sum of squares (RSS) for Thoroughbreds growth curve models Models  Sex1  AIC  BIC  –2 Log L  RSS2  Traditional Richards  M  33,666  33,669  33,662  269.5 ± 6.8 a  Including f′(t)  M  33,107  33,110  33,103  233.4 ± 6.4 b  Including f′(t) and f″(t)  M  32,942  32,947  32,936  223.7 ± 6.3 c  Traditional Richards  F  37,407  37,411  37,403  305.2 ± 7.5 a  Including f′(t)  F  36,754  36,757  36,750  262.1 ± 7.2 b  Including f′(t) and f″(t)  F  36,594  36,599  36,588  252.4 ± 7.4 c  Models  Sex1  AIC  BIC  –2 Log L  RSS2  Traditional Richards  M  33,666  33,669  33,662  269.5 ± 6.8 a  Including f′(t)  M  33,107  33,110  33,103  233.4 ± 6.4 b  Including f′(t) and f″(t)  M  32,942  32,947  32,936  223.7 ± 6.3 c  Traditional Richards  F  37,407  37,411  37,403  305.2 ± 7.5 a  Including f′(t)  F  36,754  36,757  36,750  262.1 ± 7.2 b  Including f′(t) and f″(t)  F  36,594  36,599  36,588  252.4 ± 7.4 c  a–cAverages with different superscripts differ within sex (P < 0.05) 1M = Male; F = Female 2Average of RSS ± SE View Large Table 1. Akaike's information criterion (AIC), Bayesian information criterion (BIC), –2 log likelihood (–2 Log L), and average of residual sum of squares (RSS) for Thoroughbreds growth curve models Models  Sex1  AIC  BIC  –2 Log L  RSS2  Traditional Richards  M  33,666  33,669  33,662  269.5 ± 6.8 a  Including f′(t)  M  33,107  33,110  33,103  233.4 ± 6.4 b  Including f′(t) and f″(t)  M  32,942  32,947  32,936  223.7 ± 6.3 c  Traditional Richards  F  37,407  37,411  37,403  305.2 ± 7.5 a  Including f′(t)  F  36,754  36,757  36,750  262.1 ± 7.2 b  Including f′(t) and f″(t)  F  36,594  36,599  36,588  252.4 ± 7.4 c  Models  Sex1  AIC  BIC  –2 Log L  RSS2  Traditional Richards  M  33,666  33,669  33,662  269.5 ± 6.8 a  Including f′(t)  M  33,107  33,110  33,103  233.4 ± 6.4 b  Including f′(t) and f″(t)  M  32,942  32,947  32,936  223.7 ± 6.3 c  Traditional Richards  F  37,407  37,411  37,403  305.2 ± 7.5 a  Including f′(t)  F  36,754  36,757  36,750  262.1 ± 7.2 b  Including f′(t) and f″(t)  F  36,594  36,599  36,588  252.4 ± 7.4 c  a–cAverages with different superscripts differ within sex (P < 0.05) 1M = Male; F = Female 2Average of RSS ± SE View Large Some researchers may consider that the parameter B is biologically unimportant, as Richards (1959) noted. We tested an alternative to adjust parameter k by the subfunctions. If k was adjusted by the f′(t) and f″(t), however, the combined equation became more complex and we met difficulty in numerical computations. For the simplicity and computational feasibility of the combined equations, we adjusted the parameter B. There have been many studies investigating the seasonal changes of horses' growth rate (Hintz et al., 1979; Staniar et al., 2004, 2005; Brown-Douglas et al., 2005) or determining the appropriate growth curves for several horses (Santos et al., 1999; Morel et al., 2007). However, growth curve equations directly or empirically considering CG have not been estimated so far. Accounting for CG in Thoroughbreds, Staniar et al. (2004) assumed that the growth rate consists of baseline and other systematic deviation components. Even though they did not propose any unique growth curve equations handling CG, their idea was similar to our approach of using both traditional Richards function as the baseline and sigmoid subfunctions as the systematic deviations. France et al. (1996) showed a biphasic growth curve equation by combining 2 different mathematical equations with different phases or periods, and this is an alternative approach to handle CG instead of discarding the useful merits of a continuous equation. Yin et al. (2003) proposed a continuous bivariate equation handling 2 different growth phases by using an additional timing parameter tm, and this is another alternative to handle CG. Wan et al. (1998) and Porter et al. (2010) proposed flexible alternative equations for describing growth, but their equations are intended for the development of main (baseline) equations and seemed to be unsuitable for handling CG. One useful property of our approach is the introduction of continuous single variable growth curve equations. The developed subfunctions are applicable multiply for any biological parameters in any growth curve equations if necessary or computationally feasible. The combined growth curve equation with these subfunctions is always a unique mathematical equation of time. With this property, for example, empirical percentile growth curves with Z-values considering seasonal CG are easily generated based on the variations of the analyzed data around the estimated growth curves. The young Thoroughbreds generally undergo physical training for horseracing when they become about 2 yr of age even though they are still growing. As in our data (Fig. 3 and 4), the larger data variations of BW are recognizable in the second year season probably due to the growth and training. Better understanding of the development of young horse body compositions during their growing and training seasons are important based on some mathematical standard growth curves considering CG, as suggested by the recent studies of the body compositions of young Thoroughbreds (Tozaki et al., 2011; Fonseca et al., 2013). In conclusion, the proposed method in this study is one of the useful approaches for considering and understanding seasonal CG in growth curve estimations for young Thoroughbreds in northern regions or countries where the seasonal CG are distinctive. Knowledge of the estimated growth curves handling CG would be worthwhile for the horse raising activities that relate to secure development of Thoroughbred horses in winter and spring seasons. Based on this approach, the unique optimal growth curve equations considering multiple seasonal CG also could be empirically estimated for other seasonal breeding animals or other growth traits such as body heights or entire width of chest that also would be affected by seasonal CG. LITERRATURE CITED Akaike H 1973. Information theory and an extension of the maximum likelihood principle. In: 2nd Int. Symp. Inf. Theory, Budapest, Hungary, Tsahkadsor, Armenian S.S.R.  p. 267– 281. Brown-Douglas C. G. Pagan J. D. 2006. Body weights, wither height and growth rate in Thoroughbreds raised in America, England, Australia, New Zealand and India. In: Advances in equine nutrition, Vol. IV 2004–2008.  Kentucky Equine Research, Versailles, KY. p. 213– 220. Google Scholar CrossRef Search ADS   Brown-Douglas C. G. Parkinson T. J. Firth E. C. Fennessy P. F. 2005. Body weights and growth rate of spring- and autumn-born Thoroughbred horses raised on pasture. N. Z. Vet. J.  53: 326– 331. Google Scholar CrossRef Search ADS PubMed  Equine Research Institute 2004. Japanese feeding standard for horses (2004). Equine Research Institute, Japan Racing Association, Utsunomiya, Japan. Fonseca R. G. Kenny D. A. Hill E. W. Katz L. M. 2013. The relationship between body composition, training and race performance in a group of Thoroughbred flat racehorses. Equine Vet. J.  45: 552– 557. Google Scholar CrossRef Search ADS PubMed  France J. Dijkstra J. Thornley J. H. M. Dhanoa M. S. 1996. A simple but flexible growth function. Growth Dev. Aging  60: 71– 83. Google Scholar PubMed  Hintz H. F. Hintz R. L. Van Vleck L. D. 1979. Growth rate of Thoroughbreds. Effects of age of dam, year and month of birth, and sex of foal. J. Anim. Sci.  48: 480– 487. Google Scholar CrossRef Search ADS PubMed  JBBA 2012. Annual Statistics for Japanese Bloodhorse Production 2012. The Japan Bloodhorse Breeders' Association, Tokyo, Japan. JMA 2013. Climate statistics in Hidaka region during 1981-2010. Japan Meteorological Agency, Tokyo, Japan. www.data.jma.go.jp/gmd/risk/obsdl/index.php. (Accessed 24 September 2013.) Köhn F. Sharifi A. R. Simianer H. 2007. Modeling the growth of the Goettingen minipig. J. Anim. Sci.  85: 84– 92. Google Scholar CrossRef Search ADS PubMed  MEJ 2006. Basic guidelines for comprehensively promoting measures on the welfare and management of animals. Guideline No. 140, 31 October 2006, The Ministry of The Environment, Tokyo, Japan. Morel P. C. H. Bokor A. Rogers C. W. Firth E. C. 2007. Growth curves from birth to weaning for Thoroughbred foals raised on pasture. N. Z. Vet. J.  55: 319– 325. Google Scholar CrossRef Search ADS PubMed  Onoda T. Yamamoto R. Sawamura K. Inoue Y. Matsui A. Miyake T. Hirai N. 2011. Empirical growth curve estimation using sigmoid sub-functions that adjust seasonal compensatory growth for male body weight of Thoroughbred horses. J. Equine Sci.  22: 37– 42. Google Scholar CrossRef Search ADS PubMed  Porter T. Kebreab E. Kuhi H. D. Lopez S. Strathe A. B. France J. 2010. Flexible alternatives to the Gompertz equation for describing growth with age in turkey hens. Poult. Sci.  89: 371– 378. Google Scholar CrossRef Search ADS PubMed  Richards F. J 1959. A flexible growth function for empirical use. J. Exp. Bot.  10: 290– 300. Google Scholar CrossRef Search ADS   Santos S. A. Souza G. S. Oliveira M. R. Sereno J. R. 1999. Using nonlinear models to describe curves in Pantaneiro horses. Pesqi. Agropecu. Bras.  34( 7): 1133– 1138. Google Scholar CrossRef Search ADS   Staniar W. B. Kronfeld D. S. Treiber K. H. Splan R. K. Harris P. A. 2004. Growth rate consists of baseline and systematic deviation components in Thoroughbreds. J. Anim. Sci.  82: 1007– 1015. Google Scholar CrossRef Search ADS PubMed  Staniar W. B. Kronfeld D. S. Treiber K. H. Splan R. K. Harris P. A. 2005. Thoroughbred growth characterized by a baseline and systematic deviation. In: The growing horse: Nutrition and prevention of growth disorders. European Association for Animal Production publ. no. 114.  Wageningen Academic Publishers, Wageningen, Netherlands. p. 61– 63. Tozaki T. Sato F. Hill E. W. Miyake T. Endo Y. Kakoi H. Gawahara H. Hirota K. Nakano Y. Nambo Y. Kurosawa M. 2011. Sequence variants at the myostatin gene locus influence the body composition of Thoroughbred horses. J. Vet. Med. Sci.  73: 1617– 1624. Google Scholar CrossRef Search ADS PubMed  Wan X. Zhong W. Wang M. 1998. New flexible growth function and its application to the growth of small mammals. Growth Dev. Aging  62: 27– 35. Google Scholar PubMed  Yin X. Goudriaan J. Lantinga E. A. Vos J. Spiertz H. J. 2003. A flexible sigmoid function of determinate growth. Ann. Bot. (London)  91: 361– 371. Google Scholar CrossRef Search ADS   American Society of Animal Science

Amdi, C.;Krogh, U.;Flummer, C.;Oksbjerg, N.;Hansen, C. F.;Theil, P. K.

doi: 10.2527/jas.2013-6824pmid: 24085405

Abstract The increasing litter sizes of modern pig breeds have led to a significant number of piglets that are born undersized (“small” piglets) and some have been exposed to different degrees of intrauterine growth restriction (IUGR). The aim of this study was to investigate the physiology and capability to ingest colostrum of these small piglets, suffering from various degrees of IUGR, to see if their IUGR score could be a useful tool for easy identification of piglets in need of intervention in the colostrum period. Piglets were classified at birth based on head morphology. Piglets were classified either “normal,” “mildly IUGR” (m-IUGR), or “severe IUGR” (s-IUGR), based on head morphology. Blood samples were collected at birth and at 24 h, and colostrum intake during two 12-h periods and blood metabolites at 0 and 24 h were measured. At 24 h, piglets weighing <900 g at birth and the median piglet in birth order were sacrificed, and organ weights and hepatic glycogen were measured. Overall, there was an influence of the piglets' classification on most characteristics, with normal piglets having a greater colostrum intake between 0 and 12 h (P < 0.001) and between 12 and 24 h (P < 0.05), and higher birth weight, crown rump length, body mass index, and ponderal index (P < 0.001), and a tendency toward a higher vitality score (P < 0.069) than s-IUGR piglets. There was a time × IUGR interaction, with plasma glucose levels being lowered (P < 0.001) and lactate levels elevated (P < 0.001) in s-IUGR piglets at 24 h compared with normal and m-IUGR piglets. Some differences were found in electrolytes; sodium plasma concentrations were greatest for normal piglets (P < 0.05) and highest at 0 h (P < 0.05). At 24 h of age, s-IUGR piglets had a higher heart (P < 0.001) and brain percentage (P < 0.001), and a lower liver percentage (P < 0.001) relative to body weight, compared with normal piglets. In addition, s-IUGR piglets had less hepatic glycogen than m-IUGR piglets and normal piglets. The present study showed that the physiology of piglets in the colostrum period was affected by IUGR status at birth and their intermediary metabolism was altered due to different colostrum intakes. Furthermore, it was demonstrated that the head shape of newborn piglets is a good selection criteria for identifying piglets that need oral supplementation during the neonatal stage. INTRODUCTION The increasing litter sizes observed in modern pig breeds have led to a significant lower mean birth weight and increased percentage of small piglets born (Campos et al., 2012). In addition, various degrees of intrauterine growth restriction (IUGR) may occur in these small piglets (Foxcroft et al., 2006). To develop improved management interventions, more tools are needed to identify piglets with lower survival rates. It has recently been hypothesized that IUGR piglets can be characterized on 3 criteria, based on head morphology (Chevaux et al., 2010) and IUGR piglets have a “dolphin-like” head shape compared with “normal” piglets (Hales et al., 2013). Piglets exposed to IUGR have prioritized brain development due to the “brain sparing effect,” as part of a fetal adaptive reaction to placental insufficiency (Roza et al., 2008). However, it is important to know which organs or nutrient pools are being compromised when nutrients are being allocated preferentially to brain development. Glycogen pools in liver and muscles are important for neonatal piglet survival (Theil et al., 2011), but little is known about liver glycogen stores in IUGR piglets at this stage. Glycogen stores are the primary substrate for oxidation immediately after birth (Theil et al., 2012). Therefore, knowledge on the physiology of piglets exposed to different degrees of IUGR is important for increasing survival of susceptible piglets during the critical neonatal phase. The aim of this study was to evaluate traits related to piglet survival (colostrum intake, glycogen depletion) of normal, mildly IUGR, and severe IUGR piglets, and to evaluate performance characteristics, plasma metabolites, stomach content, and glycogen stores remaining in the liver at 24 h. We hypothesized that IUGR piglets would ingest less colostrum due to smaller physical stomach size, which in turn would affect plasma metabolites and depletion of glycogen stores during the colostrum period. MATERIAL AND METHODS All sampling, housing, and measurements were done in accordance with Danish laws and regulations for the humane care and use of animals in research [The Danish Ministry of Justice, Animal and Testing Act (consolidation Act no. 726 of 9 September 1993, as amended by Act No. 1081 of 20 December 1995)]. Furthermore, the Danish Animal Experimentation Inspectorate approved the study protocols and supervised the experiment. Animals and Design This study was conducted as part of a larger project investigating transition feeding on sow and piglet productivity (effects of diets will be published elsewhere). Thirty-six second parity sows (Danish Landrace × Danish Yorkshire) mated with Duroc semen were brought into farrowing crates at d 105 of gestation over 2 rounds (block, n = 18). All animals were housed at Aarhus University experimental research unit (Aarhus University-Foulum, Denmark). Sows were fed according to Danish recommendations (Jørgensen, 2005). Piglet Measurements at Birth Litters were video monitored continuously from 2 d prefarrowing until 24 h postfarrowing, such that farrowing time could be recorded. Based on video recordings, a vitality score at birth was given as described by Baxter et al. (2008). A vitality score of zero was given if there was no movement or breathing after 15 s. A vitality score of 1 was given if there was no movement after 15 s, but the piglet was breathing or attempting to breathe (coughing, spluttering, clearing its lungs). A vitality score of 2 was given if the piglet showed some movement within 15 s, breathing, or attempting to breathe. A vitality score of 3 was given if the piglet showed good movement, good breathing, and the piglet attempted to stand within 15 s. No score was given if the piglet was still and covered in the placental sac. Piglets were removed from the sow before they had suckled, ear tagged for identification purposes, dried, and the umbilical cord shortened to 15 cm and closed off with straps to prevent blood loss. Birth weight was recorded as an average of 3 weights on a precision scale to the nearest 1 g (Sartorius 12,100 g; Dansk Vægt Industri A/S, Skanderborg, Denmark). In addition, crown rump length (CRL) was measured from the crown of the head to the base of the tail (supine length of piglet) and gender was noted. The piglets were given a visual IUGR score from normal, mildly (m-IUGR), and severe (s-IUGR), according to Hales et al. (2013), recognizing the IUGR piglet as: 1) steep, dolphin-like forehead, 2) bulging eyes, and 3) wrinkles perpendicular to the mouth. A score of “s-IUGR” was given if 2 or 3 of the characteristics were present. A score of “m-IUGR” was given if 1 of the characteristics was present. Finally, if none of the characteristics were present, piglets were considered “normal” and represented control piglets. Immediately postpartum and before suckling, the first live born and then every second piglet (uneven eartag numbers) were held in dorsal recumbency and 6 mL of blood was drawn from vena jugularis externa/interna/communis, using 23-gauge needles into vacutainers containing heparin. Glucose and lactate values, and the following electrolytes, potassium (K+), sodium (Na+), calcium (Ca2+), and chloride (Cl–), were measured in whole blood (ABL 725 blood-gas analyzer; Radiometer A/S, Copenhagen, Denmark). Body mass index (BMI) and ponderal index (PI) of all piglets was calculated from the CRL and birth weight, using the following equations:    After measurements and procedures at birth, piglets were returned to their birth mother (within 20 min). At 12 and 24 h after birth of first born piglet, all piglets were weighed, as above, and weights recorded. Piglet Measurements at 24 h Twenty-four hours after the first piglet within a litter was born, blood samples were taken from all even-numbered piglets. The piglet born in the middle of the birth order within each litter was sacrificed, as were piglets born <900 g at 24 h after birth of first born piglet. In total, 80 piglets were sacrificed after a blood sample was collected and analyzed for blood metabolites, as described above, and weights of the heart, liver, stomach, and brain were recorded. A liver sample was collected from the quadrate lope, snap frozen in liquid nitrogen, and stored at –80°C for later pro-glycogen, macro-glycogen, and total glycogen analyses. The stomach was further given a score from 1 to 4, depending on content. A score of 1 was given if the stomach was full, 2 if it was more than half full, 3 if it was less than half full, and 4 if it was empty. Colostrum Intake Colostrum intake (CI) was estimated, according to Devillers et al. (2004), based on BW gain. According to Devillers et al. (2004), most of the variation in BW gain during the colostrum period is explained by colostrum intake (94.4%). Consequently, colostrum intake may be predicted by BW gain during the colostrum period. The following formula was adapted from Devillers et al. (2004):  Where 1.55 is the slope when regressing BW gain on CI (opposite relationship compared with what is needed here), 0.944 is the r2 value (used to derive the slope for the opposite regression coefficient), and 115 is a constant (amount of colostrum required to maintain BW from birth to 24 h of age). Based on the equation above, a similar equation was derived to predict the colostrum intake from 0 to 12 h and from 12 to 24 h:  Where the constants are identical to Eq. [3], except for the intercept (115 g/24 h), which was divided by 2 to account for the amount of colostrum required to maintain BW for a 12-h period only. Glycogen Analyses Twenty-five-milligram liver samples were prepared at –20°C. The muscle samples were precipitated in 0.5 mL 1.5 M perchloric acid for 20 min followed by centrifugation (2700 × g) in order to precipitate pro-glycogen leaving macro-glycogen in the supernatant as described previously (Adamo and Graham, 1998). From the supernatant, 0.25 mL were used for free glucose determination and the other 0.25 mL of the supernatant and the pellet (containing the pro-glycogen) were hydrolyzed for 2 h in 1 M HCl and the glucose units were analyzed by spectrophotometry as described previously (Passonneau and Lowry, 1993). The total glycogen pool in liver was calculated as:  Statistical Analyses Normal mixed models (MIXED procedure; SAS Inst. Inc., Cary, NC) were used to describe piglet traits, according to the following models:  where Yijk is blood metabolites and CI related to the piglet, μ is the overall mean, αi is the effect of piglet IUGR category (i = normal, m-IUGR, s-IUGR), βj is the effect of time (k = birth or 24 h), γk is the effect of block (j = 1, 2), (αβ)ij is the interaction between piglet IUGR category and time, and εijk is the residual error component, which is assumed independent and normal distributed. The interaction between piglet IUGR category and time was only included when significant. The total number of born piglets (p = 11, 12, 15,…, 27) was initially included in the model as a covariate but found not significant for all studied traits. Therefore, it was not included in the model. The other observed traits (CRL, BMI, organ weights, and glycogen stores) were tested with the described model without the time and time × IUGR interaction terms. For the data describing stomach content, a type 3 Poisson distribution in PROC GENMOD (SAS Inst. Inc.) was used. For all parameters, the random and residual error components were assumed to be independent and normally distributed, and their expectations were assumed to be zero. For the analysis of pro-glycogen, macro-glycogen, total glycogen, total liver glycogen, and liver glycogen g/100 g of wet weight, data were log transformed before analysis. Results presented for these data are the arithmetic data with confidence intervals. For the relationship between organ weights and IUGR category, a regression line was fitted with a simple correlation coefficient of determination. All other traits were separated, using the PDIFF option, and presented as LSMEANS ± SEM and considered significant when P < 0.05, and a trend when P < 0.10. RESULTS Piglet Characteristics The birth weight, weight at 12 h, and piglet characteristics are presented in Table 1. Of 619 piglets given an IUGR score at birth, 67.9% were normal (normal, n = 420 piglets), 24.7% had suffered mildly from IUGR (m-IUGR, n = 153 piglets), and 7.4% were deemed to have suffered severely from IUGR (s-IUGR, n = 46 piglets). Overall, there was a strong influence of the piglets' IUGR score on most characteristics, with normal piglets having a greater birth weight (P < 0.001), higher CRL (P < 0.001), higher BMI (P < 0.001), higher PI (P < 0.001), and tendency toward a higher vitality score (P < 0.07), compared with s-IUGR and m-IUGR piglets, whereas those of m-IUGR were intermediate. Table 1. Birth weight, colostrum intake, and piglet characteristics (crown rump length, body mass index, and vitality scores) in normal, mild (m-IUGR1), and severe (s-IUGR) piglets   IUGR score      Normal  m-IUGR  s-IUGR  P-value  n  420  153  46    Piglet characteristics          Birth weight, g  1,326 ± 12a  961 ± 16b  682 ± 23c  <0.001  Crown rump length, cm  25.3 ± 0.1a  23.3 ± 0.2b  21.1 ± 0.3c  <0.001  Body mass index, kg m–2  20.6 ± 0.2a  17.7 ± 0.3b  15.3 ± 0.5c  <0.001  Ponderal index, kg m–3  82.2 ± 0.8a  76.5 ± 1.3b  73.0 ± 2.5b  <0.001  Vitality scores  1.5 ± 0.1a  1.4 ± 0.1ab  1.2 ± 0.2b  0.069  Weight at 12 h, g  1,422 ± 14a  996 ± 19b  681 ± 27c  <0.001  Weight at 24 h, g  1,431 ± 14a  999 ± 20b  677 ± 26c  <0.001  BW change 0 to 24 h, %  7.4 ± 0.3a  2.6 ± 0.6b  –2.0 ± 1.0c  <0.001  Colostrum intake g          0 to 12 h, g  193 ± 4a  106 ± 7b  58 ± 13c  <0.001  12 to 24 h, g  73 ± 3a  59 ± 6b  45 ± 11b  0.002  0 to 24 h, g  268 ± 5a  163 ± 9b  97 ± 16c  <0.001    IUGR score      Normal  m-IUGR  s-IUGR  P-value  n  420  153  46    Piglet characteristics          Birth weight, g  1,326 ± 12a  961 ± 16b  682 ± 23c  <0.001  Crown rump length, cm  25.3 ± 0.1a  23.3 ± 0.2b  21.1 ± 0.3c  <0.001  Body mass index, kg m–2  20.6 ± 0.2a  17.7 ± 0.3b  15.3 ± 0.5c  <0.001  Ponderal index, kg m–3  82.2 ± 0.8a  76.5 ± 1.3b  73.0 ± 2.5b  <0.001  Vitality scores  1.5 ± 0.1a  1.4 ± 0.1ab  1.2 ± 0.2b  0.069  Weight at 12 h, g  1,422 ± 14a  996 ± 19b  681 ± 27c  <0.001  Weight at 24 h, g  1,431 ± 14a  999 ± 20b  677 ± 26c  <0.001  BW change 0 to 24 h, %  7.4 ± 0.3a  2.6 ± 0.6b  –2.0 ± 1.0c  <0.001  Colostrum intake g          0 to 12 h, g  193 ± 4a  106 ± 7b  58 ± 13c  <0.001  12 to 24 h, g  73 ± 3a  59 ± 6b  45 ± 11b  0.002  0 to 24 h, g  268 ± 5a  163 ± 9b  97 ± 16c  <0.001  a–cWithin a row, means without a common superscript differ (P < 0.05). 1IUGR = intrauterine growth restricted. View Large Table 1. Birth weight, colostrum intake, and piglet characteristics (crown rump length, body mass index, and vitality scores) in normal, mild (m-IUGR1), and severe (s-IUGR) piglets   IUGR score      Normal  m-IUGR  s-IUGR  P-value  n  420  153  46    Piglet characteristics          Birth weight, g  1,326 ± 12a  961 ± 16b  682 ± 23c  <0.001  Crown rump length, cm  25.3 ± 0.1a  23.3 ± 0.2b  21.1 ± 0.3c  <0.001  Body mass index, kg m–2  20.6 ± 0.2a  17.7 ± 0.3b  15.3 ± 0.5c  <0.001  Ponderal index, kg m–3  82.2 ± 0.8a  76.5 ± 1.3b  73.0 ± 2.5b  <0.001  Vitality scores  1.5 ± 0.1a  1.4 ± 0.1ab  1.2 ± 0.2b  0.069  Weight at 12 h, g  1,422 ± 14a  996 ± 19b  681 ± 27c  <0.001  Weight at 24 h, g  1,431 ± 14a  999 ± 20b  677 ± 26c  <0.001  BW change 0 to 24 h, %  7.4 ± 0.3a  2.6 ± 0.6b  –2.0 ± 1.0c  <0.001  Colostrum intake g          0 to 12 h, g  193 ± 4a  106 ± 7b  58 ± 13c  <0.001  12 to 24 h, g  73 ± 3a  59 ± 6b  45 ± 11b  0.002  0 to 24 h, g  268 ± 5a  163 ± 9b  97 ± 16c  <0.001    IUGR score      Normal  m-IUGR  s-IUGR  P-value  n  420  153  46    Piglet characteristics          Birth weight, g  1,326 ± 12a  961 ± 16b  682 ± 23c  <0.001  Crown rump length, cm  25.3 ± 0.1a  23.3 ± 0.2b  21.1 ± 0.3c  <0.001  Body mass index, kg m–2  20.6 ± 0.2a  17.7 ± 0.3b  15.3 ± 0.5c  <0.001  Ponderal index, kg m–3  82.2 ± 0.8a  76.5 ± 1.3b  73.0 ± 2.5b  <0.001  Vitality scores  1.5 ± 0.1a  1.4 ± 0.1ab  1.2 ± 0.2b  0.069  Weight at 12 h, g  1,422 ± 14a  996 ± 19b  681 ± 27c  <0.001  Weight at 24 h, g  1,431 ± 14a  999 ± 20b  677 ± 26c  <0.001  BW change 0 to 24 h, %  7.4 ± 0.3a  2.6 ± 0.6b  –2.0 ± 1.0c  <0.001  Colostrum intake g          0 to 12 h, g  193 ± 4a  106 ± 7b  58 ± 13c  <0.001  12 to 24 h, g  73 ± 3a  59 ± 6b  45 ± 11b  0.002  0 to 24 h, g  268 ± 5a  163 ± 9b  97 ± 16c  <0.001  a–cWithin a row, means without a common superscript differ (P < 0.05). 1IUGR = intrauterine growth restricted. View Large Colostrum Intake The colostrum intakes between 0 and 12 h and between 12 and 24 h are presented in Table 1. There was a strong influence of the piglets' IUGR score, in that normal piglets had a greater colostrum intake between 0 and 12 h (P < 0.001) and between 12 and 24 h (P = 0.022), than s-IUGR piglets, whereas that of m-IUGR was intermediate. Organ Weights and Liver Glycogen The average organ weights and liver glycogen content 24 h after birth are presented in Table 2. For the brain and liver, and brain to liver ratio, an interaction between IUGR score and sow diet was found and has been discussed elsewhere (Amdi et al., 2013), whereas no diet × IUGR score was observed for the other studied traits. The relationship between IUGR score and liver weight and organs at 24 h is shown (Fig. 1 A to C). In absolute terms, organ weights were all greater in normal piglets compared with m-IUGR and s- IUGR piglets. In relative terms, the piglets with an s-IUGR score had a higher relative weight of heart and brain percentage, and a lower relative weight of liver than normal piglets (Fig. 1). The average stomach weight (including content) decreased with increasing score (Fig. 2). Piglets suffering mildly and severely from IUGR (m-IUGR and s-IUGR) had a higher brain-to-heart ratio than normal piglets (P < 0.001). Overall, there was a strong influence of piglet IUGR score on glycogen results, as normal piglets had a greater amount of pro-glycogen (P < 0.001), macro-glycogen (P < 0.001), total glycogen (P < 0.001), total liver glycogen (P < 0.001), and relative liver glycogen (g glycogen/100 g of wet liver weight) (P < 0.001) than both s-IUGR and m-IUGR piglets. In addition, m-IUGR piglets had larger values of glycogen traits than s-IUGR piglets. Table 2. Actual organ weights, ratios, stomach content, and glycogen deposits at 24 h of age in normal, mild (m-IUGR1), and severe (s-IUGR) piglets   IUGR score      Normal  m-IUGR  s-IUGR  P-value  n  25  30  25    Glycogen deposits in liver          Pro-glycogen, µmol/g2  355a (227 to 371)  253b (167 to 265)  176c (111 to 184)  <0.001  Macro-glycogen, µmol/g2  239a (142 to 241)  140b (88 to 145)  95 (58 to 99)  <0.001  Total glycogen, µmol/g2  594a (388 to 609)  393b (268 to 409)  271c (184 to 292)  <0.001  Total liver glycogen, g2  4.3a (2.2 to 4.0)  1.3b (0.7 to 1.2)  0.6c (0.4 to 0.7)  <0.001  Liver glycogen, g/100 g of wet weight2  9.6a (6.3 to 9.9)  6.4b (4.3 to 6.6)  4.4c (3.0 to 4.7)  <0.001  Free glucose in liver  15.5 ± 1.1a  9.9 ± 1.0b  6.6 ± 1.1c  <0.001  Organ weights, g          Liver  38.8 ± 2.0a  18.4 ± 1.2b  13.7 ± 0.9c  <0.001  Heart  9.9 ± 0.4a  5.9 ± 0.2b  4.9 ± 0.2c  <0.001  Brain  34.0 ± 0.4a  30.4 ± 0.6b  29.7 ± 0.5b  <0.001  Stomach including content  35.7 ± 3.5a  12.8 ± 1.7b  8.5 ± 1.2b  <0.001  Stomach score  1.3 ± 0.1a  2.2 ± 0.2b  2.6 ± 0.2b  0.049  Ratios          Brain:heart ratio  3.6 ± 0.2a  5.2 ± 0.2b  6.2 ± 0.2c  <0.001  Relative organ weights (percentage)  Liver3  2.7 ± 0.1a  2.3 ± 0.1b  2.2 ± 0.1b  <0.001  Heart  0.7 ± 0.0a  0.8 ± 0.0b  0.8 ± 0.0c  <0.001  Brain3  2.5 ± 0.2a  4.1 ± 0.2b  5.2 ± 0.2c  <0.001    IUGR score      Normal  m-IUGR  s-IUGR  P-value  n  25  30  25    Glycogen deposits in liver          Pro-glycogen, µmol/g2  355a (227 to 371)  253b (167 to 265)  176c (111 to 184)  <0.001  Macro-glycogen, µmol/g2  239a (142 to 241)  140b (88 to 145)  95 (58 to 99)  <0.001  Total glycogen, µmol/g2  594a (388 to 609)  393b (268 to 409)  271c (184 to 292)  <0.001  Total liver glycogen, g2  4.3a (2.2 to 4.0)  1.3b (0.7 to 1.2)  0.6c (0.4 to 0.7)  <0.001  Liver glycogen, g/100 g of wet weight2  9.6a (6.3 to 9.9)  6.4b (4.3 to 6.6)  4.4c (3.0 to 4.7)  <0.001  Free glucose in liver  15.5 ± 1.1a  9.9 ± 1.0b  6.6 ± 1.1c  <0.001  Organ weights, g          Liver  38.8 ± 2.0a  18.4 ± 1.2b  13.7 ± 0.9c  <0.001  Heart  9.9 ± 0.4a  5.9 ± 0.2b  4.9 ± 0.2c  <0.001  Brain  34.0 ± 0.4a  30.4 ± 0.6b  29.7 ± 0.5b  <0.001  Stomach including content  35.7 ± 3.5a  12.8 ± 1.7b  8.5 ± 1.2b  <0.001  Stomach score  1.3 ± 0.1a  2.2 ± 0.2b  2.6 ± 0.2b  0.049  Ratios          Brain:heart ratio  3.6 ± 0.2a  5.2 ± 0.2b  6.2 ± 0.2c  <0.001  Relative organ weights (percentage)  Liver3  2.7 ± 0.1a  2.3 ± 0.1b  2.2 ± 0.1b  <0.001  Heart  0.7 ± 0.0a  0.8 ± 0.0b  0.8 ± 0.0c  <0.001  Brain3  2.5 ± 0.2a  4.1 ± 0.2b  5.2 ± 0.2c  <0.001  a–cWithin a row, means without a common superscript differ (P < 0.05). 1IUGR = intrauterine growth restriction. 2Data analysis was performed on log transformed data. Results presented here are on a normal scale (arithmetic) with confidence intervals. 3An interaction between sow diet and IUGR score was found (Amdi et al., 2013). View Large Table 2. Actual organ weights, ratios, stomach content, and glycogen deposits at 24 h of age in normal, mild (m-IUGR1), and severe (s-IUGR) piglets   IUGR score      Normal  m-IUGR  s-IUGR  P-value  n  25  30  25    Glycogen deposits in liver          Pro-glycogen, µmol/g2  355a (227 to 371)  253b (167 to 265)  176c (111 to 184)  <0.001  Macro-glycogen, µmol/g2  239a (142 to 241)  140b (88 to 145)  95 (58 to 99)  <0.001  Total glycogen, µmol/g2  594a (388 to 609)  393b (268 to 409)  271c (184 to 292)  <0.001  Total liver glycogen, g2  4.3a (2.2 to 4.0)  1.3b (0.7 to 1.2)  0.6c (0.4 to 0.7)  <0.001  Liver glycogen, g/100 g of wet weight2  9.6a (6.3 to 9.9)  6.4b (4.3 to 6.6)  4.4c (3.0 to 4.7)  <0.001  Free glucose in liver  15.5 ± 1.1a  9.9 ± 1.0b  6.6 ± 1.1c  <0.001  Organ weights, g          Liver  38.8 ± 2.0a  18.4 ± 1.2b  13.7 ± 0.9c  <0.001  Heart  9.9 ± 0.4a  5.9 ± 0.2b  4.9 ± 0.2c  <0.001  Brain  34.0 ± 0.4a  30.4 ± 0.6b  29.7 ± 0.5b  <0.001  Stomach including content  35.7 ± 3.5a  12.8 ± 1.7b  8.5 ± 1.2b  <0.001  Stomach score  1.3 ± 0.1a  2.2 ± 0.2b  2.6 ± 0.2b  0.049  Ratios          Brain:heart ratio  3.6 ± 0.2a  5.2 ± 0.2b  6.2 ± 0.2c  <0.001  Relative organ weights (percentage)  Liver3  2.7 ± 0.1a  2.3 ± 0.1b  2.2 ± 0.1b  <0.001  Heart  0.7 ± 0.0a  0.8 ± 0.0b  0.8 ± 0.0c  <0.001  Brain3  2.5 ± 0.2a  4.1 ± 0.2b  5.2 ± 0.2c  <0.001    IUGR score      Normal  m-IUGR  s-IUGR  P-value  n  25  30  25    Glycogen deposits in liver          Pro-glycogen, µmol/g2  355a (227 to 371)  253b (167 to 265)  176c (111 to 184)  <0.001  Macro-glycogen, µmol/g2  239a (142 to 241)  140b (88 to 145)  95 (58 to 99)  <0.001  Total glycogen, µmol/g2  594a (388 to 609)  393b (268 to 409)  271c (184 to 292)  <0.001  Total liver glycogen, g2  4.3a (2.2 to 4.0)  1.3b (0.7 to 1.2)  0.6c (0.4 to 0.7)  <0.001  Liver glycogen, g/100 g of wet weight2  9.6a (6.3 to 9.9)  6.4b (4.3 to 6.6)  4.4c (3.0 to 4.7)  <0.001  Free glucose in liver  15.5 ± 1.1a  9.9 ± 1.0b  6.6 ± 1.1c  <0.001  Organ weights, g          Liver  38.8 ± 2.0a  18.4 ± 1.2b  13.7 ± 0.9c  <0.001  Heart  9.9 ± 0.4a  5.9 ± 0.2b  4.9 ± 0.2c  <0.001  Brain  34.0 ± 0.4a  30.4 ± 0.6b  29.7 ± 0.5b  <0.001  Stomach including content  35.7 ± 3.5a  12.8 ± 1.7b  8.5 ± 1.2b  <0.001  Stomach score  1.3 ± 0.1a  2.2 ± 0.2b  2.6 ± 0.2b  0.049  Ratios          Brain:heart ratio  3.6 ± 0.2a  5.2 ± 0.2b  6.2 ± 0.2c  <0.001  Relative organ weights (percentage)  Liver3  2.7 ± 0.1a  2.3 ± 0.1b  2.2 ± 0.1b  <0.001  Heart  0.7 ± 0.0a  0.8 ± 0.0b  0.8 ± 0.0c  <0.001  Brain3  2.5 ± 0.2a  4.1 ± 0.2b  5.2 ± 0.2c  <0.001  a–cWithin a row, means without a common superscript differ (P < 0.05). 1IUGR = intrauterine growth restriction. 2Data analysis was performed on log transformed data. Results presented here are on a normal scale (arithmetic) with confidence intervals. 3An interaction between sow diet and IUGR score was found (Amdi et al., 2013). View Large Figure 1. View largeDownload slide The percentage of the organs compared with body weight (BW) at 24 h for normal piglets (□), piglets with mild intrauterine growth restriction (IUGR) (■), and piglets with a severe IUGR score (△) for (A), the liver; (B), the brain; and (C), the heart. Figure 1. View largeDownload slide The percentage of the organs compared with body weight (BW) at 24 h for normal piglets (□), piglets with mild intrauterine growth restriction (IUGR) (■), and piglets with a severe IUGR score (△) for (A), the liver; (B), the brain; and (C), the heart. Figure 2. View largeDownload slide Stomach content at 24 h presented in black for normal piglets, gray for piglets suffering mildly from intrauterine growth restriction (IUGR), and solid white for pigs suffering severe-IUGR. The score of 1 was given if the stomach was full, 2 if it was more than half full, 3 if it was less than half full, and 4 if it was empty. Figure 2. View largeDownload slide Stomach content at 24 h presented in black for normal piglets, gray for piglets suffering mildly from intrauterine growth restriction (IUGR), and solid white for pigs suffering severe-IUGR. The score of 1 was given if the stomach was full, 2 if it was more than half full, 3 if it was less than half full, and 4 if it was empty. Blood Metabolites The blood metabolite contents at 0 and 24 h of piglets with different IUGR scores are presented in Table 3. There was a time × IUGR interaction for glucose (P < 0.001), potassium (P < 0.049), calcium (P < 0.01), and chloride (P < 0.01). For glucose, no differences were found among IUGR scores at 0 h, but at 24 h the glucose concentrations were high, intermediate, and low in normal, m-IUGR, and s-IUGR piglets, respectively. Both m-IUGR and s-IUGR had similar glucose concentrations at 24 and 0 h, whereas that of normal piglets was highest 24 h after birth (P < 0.01; Table 3). For potassium, normal piglets had higher levels at 24 h than 0 h, and for calcium all piglet categories had lower values at 24 h compared with 0 h. For chloride, normal piglets had similar concentrations between 0 and 24 h, but m-IUGR and s-IUGR piglets had higher concentrations at 24 h. Plasma lactate was low, intermediate, and high (5.5, 5.9, and 7.4 mmol/L) in normal, m-IUGR, and s-IUGR piglets, respectively (P < 0.001). Plasma lactate concentrations were higher at birth compared with 24 h (7.3 ± 0.2 vs. 5.2 ± 0.2 mmol/L; P < 0.001). For sodium, the average concentrations were 133 ± 1, 130 ± 2, and 127 ± 3 mmol/L for normal, m-IUGR, and s-IUGR, respectively (P = 0.05). There was also an effect of time (P = 0.002), with sodium values being higher at birth (0 h) than 24 h (132 ± 1 vs. 128 ± 1 mmol/L). Table 3. Effects of time and intrauterine growth restriction (IUGR) on concentrations of plasma metabolites and plasma electrolytes at birth (0 h) and 24 h of age in normal, mild (m-IUGR), and severe (s-IUGR) piglets   0 h  24 h    IUGR score  IUGR score  P-value  normal  m-IUGR  s-IUGR  normal  m-IUGR  s-IUGR  IUGR  Time  IUGR × time  n  173  58  17  174  71  29        Glucose (mmol/L)  3.0 ± 0.1ad  2.9 ± 0.2ad  2.3 ± 0.3ac  4.9 ± 0.1b  3.3 ± 0.2d  1.9 ± 0.3c  <0.001  <0.001  <0.001  Lactate (mmol/L)  6.5 ± 0.2  7.1 ± 0.3  8.2 ± 0.6  4.5 ± 0.2  4.7 ± 0.3  6.5 ± 0.5  <0.001  <0.001  0.54  Potassium (mmol/L)  4.2 ± 0.1a  4.4 ± 0.1ac  4.2 ± 0.2ac  4.7 ± 0.1b  4.5 ± 0.1bc  4.5 ± 0.2ab  0.63  0.002  0.049  Sodium (mmol/L)  134 ± 1  135 ± 2  129 ± 4  132 ± 1  125 ± 2  126 ± 3  0.052  0.002  0.07  Calcium (mmol/L)  1.24 ± 0.03a  1.34 ± 0.05a  1.26 ± 0.09a  1.09 ± 0.03b  0.97 ± 0.04c  0.98 ± 0.07bc  0.74  <0.001  0.010  Chloride (mmol/L)  103 ± 2a  100 ± 3a  101 ± 6a  104 ± 2a  114 ± 3b  118 ± 4b  0.13  0.001  0.005    0 h  24 h    IUGR score  IUGR score  P-value  normal  m-IUGR  s-IUGR  normal  m-IUGR  s-IUGR  IUGR  Time  IUGR × time  n  173  58  17  174  71  29        Glucose (mmol/L)  3.0 ± 0.1ad  2.9 ± 0.2ad  2.3 ± 0.3ac  4.9 ± 0.1b  3.3 ± 0.2d  1.9 ± 0.3c  <0.001  <0.001  <0.001  Lactate (mmol/L)  6.5 ± 0.2  7.1 ± 0.3  8.2 ± 0.6  4.5 ± 0.2  4.7 ± 0.3  6.5 ± 0.5  <0.001  <0.001  0.54  Potassium (mmol/L)  4.2 ± 0.1a  4.4 ± 0.1ac  4.2 ± 0.2ac  4.7 ± 0.1b  4.5 ± 0.1bc  4.5 ± 0.2ab  0.63  0.002  0.049  Sodium (mmol/L)  134 ± 1  135 ± 2  129 ± 4  132 ± 1  125 ± 2  126 ± 3  0.052  0.002  0.07  Calcium (mmol/L)  1.24 ± 0.03a  1.34 ± 0.05a  1.26 ± 0.09a  1.09 ± 0.03b  0.97 ± 0.04c  0.98 ± 0.07bc  0.74  <0.001  0.010  Chloride (mmol/L)  103 ± 2a  100 ± 3a  101 ± 6a  104 ± 2a  114 ± 3b  118 ± 4b  0.13  0.001  0.005  a–dWithin a row, means without a common superscript differ (P < 0.05). Values are means ± SEM. View Large Table 3. Effects of time and intrauterine growth restriction (IUGR) on concentrations of plasma metabolites and plasma electrolytes at birth (0 h) and 24 h of age in normal, mild (m-IUGR), and severe (s-IUGR) piglets   0 h  24 h    IUGR score  IUGR score  P-value  normal  m-IUGR  s-IUGR  normal  m-IUGR  s-IUGR  IUGR  Time  IUGR × time  n  173  58  17  174  71  29        Glucose (mmol/L)  3.0 ± 0.1ad  2.9 ± 0.2ad  2.3 ± 0.3ac  4.9 ± 0.1b  3.3 ± 0.2d  1.9 ± 0.3c  <0.001  <0.001  <0.001  Lactate (mmol/L)  6.5 ± 0.2  7.1 ± 0.3  8.2 ± 0.6  4.5 ± 0.2  4.7 ± 0.3  6.5 ± 0.5  <0.001  <0.001  0.54  Potassium (mmol/L)  4.2 ± 0.1a  4.4 ± 0.1ac  4.2 ± 0.2ac  4.7 ± 0.1b  4.5 ± 0.1bc  4.5 ± 0.2ab  0.63  0.002  0.049  Sodium (mmol/L)  134 ± 1  135 ± 2  129 ± 4  132 ± 1  125 ± 2  126 ± 3  0.052  0.002  0.07  Calcium (mmol/L)  1.24 ± 0.03a  1.34 ± 0.05a  1.26 ± 0.09a  1.09 ± 0.03b  0.97 ± 0.04c  0.98 ± 0.07bc  0.74  <0.001  0.010  Chloride (mmol/L)  103 ± 2a  100 ± 3a  101 ± 6a  104 ± 2a  114 ± 3b  118 ± 4b  0.13  0.001  0.005    0 h  24 h    IUGR score  IUGR score  P-value  normal  m-IUGR  s-IUGR  normal  m-IUGR  s-IUGR  IUGR  Time  IUGR × time  n  173  58  17  174  71  29        Glucose (mmol/L)  3.0 ± 0.1ad  2.9 ± 0.2ad  2.3 ± 0.3ac  4.9 ± 0.1b  3.3 ± 0.2d  1.9 ± 0.3c  <0.001  <0.001  <0.001  Lactate (mmol/L)  6.5 ± 0.2  7.1 ± 0.3  8.2 ± 0.6  4.5 ± 0.2  4.7 ± 0.3  6.5 ± 0.5  <0.001  <0.001  0.54  Potassium (mmol/L)  4.2 ± 0.1a  4.4 ± 0.1ac  4.2 ± 0.2ac  4.7 ± 0.1b  4.5 ± 0.1bc  4.5 ± 0.2ab  0.63  0.002  0.049  Sodium (mmol/L)  134 ± 1  135 ± 2  129 ± 4  132 ± 1  125 ± 2  126 ± 3  0.052  0.002  0.07  Calcium (mmol/L)  1.24 ± 0.03a  1.34 ± 0.05a  1.26 ± 0.09a  1.09 ± 0.03b  0.97 ± 0.04c  0.98 ± 0.07bc  0.74  <0.001  0.010  Chloride (mmol/L)  103 ± 2a  100 ± 3a  101 ± 6a  104 ± 2a  114 ± 3b  118 ± 4b  0.13  0.001  0.005  a–dWithin a row, means without a common superscript differ (P < 0.05). Values are means ± SEM. View Large DISCUSSION Intrauterine growth restricted piglets are classically identified on their body weight distribution (Bauer et al., 1998b). However, birth weight alone does not indicate whether a piglet has been exposed to IUGR during fetal development. In the IUGR pigs, relatively more nutrients are redirected toward growth of the brain (brain sparing) and heart, compared with a normal piglet (Hammond, 1921), as part of a fetal adaptive reaction to placental insufficiency (Roza et al., 2008). This adaptive reaction is intended to ensure proper development of the brain (Baschat, 2004). It is important to know which organs or nutrient pools are being compromised when nutrients are being allocated preferentially to brain development. In the present study, the s-IUGR piglets did indeed have larger relative brain weights compared with m-IUGR piglets and larger relative weight of brains than normal piglets. Therefore, this identification method will allow a more accurate determination of susceptibility of piglets and their subsequent risk of death due to their IUGR status at birth. Some small piglets may still reach their genetic growth potential, displaying normal growth, whereas piglets suffering from IUGR have not (i.e., they have had asymmetrical growth) (Bauer et al., 1998a). Normal piglets have flat heads compared with the small dolphin-like piglets given an s-IUGR score (see Hales et al., 2013). Results from the present study indicate that piglets whose brains weigh <3% of birth weight at 24 h (Fig. 1) can be considered normal and will likely have higher survival chances than other littermates due to altered physiology and phenotype. Supply of energy, via glycogen mobilization and colostrum, is of major importance for the neonate piglet (Quesnel et al., 2012; Theil et al., 2011) until copious milk production begins ∼33 h after onset of parturition (Krogh et al., 2012). It has recently been shown that at least 200 g of colostrum per piglet is required to maintain life during the neonatal phase (Quesnel et al., 2012). However, the present study demonstrated that only normal, but not s-IUGR and m-IUGR piglets, ingested the required amount. Severe-IUGR piglets only receive approximately half the amount required (200 g) to ensure neonatal survival (Devillers et al., 2011; Quesnel et al., 2012). Nonetheless, some m-IUGR and s-IUGR piglets did indeed have full stomachs. However, based on falling plasma glucose levels and lower remaining glycogen depots in the liver at 24 h (discussed later), most IUGR piglets (m-IUGR and s-IUGR) in the present study did not ingest sufficient amounts of colostrum. Improving the quality and/or quantity of colostrum, thereby providing more essential nutrients, may be 1 way of compensating for the inherent difficulties faced by premature piglets, such as inability to suckle large quantities of colostrum. Based on the estimated colostrum intake between birth and 12 h, the s-IUGR piglets did indeed ingest a substantial amount of colostrum between 0 and 12 h and 12 and 24 h, which is in accordance with the stomach content found in m-IUGR and s-IUGR in the present study. Although some of the IUGR piglets had full stomachs, half of the IUGR piglets had either more than half-full or less than half-full stomachs, and more importantly they might not be fit for competing with normal piglets the first 24 h. As a consequence, many IUGR piglets in this study likely did not ingest sufficient amounts of colostrum to survive without intervention. One consequence of insufficient colostrum intake could be observed in plasma glucose values. The plasma glucose decreased from 0 to 24 h for s-IUGR piglets, whereas it stayed the same for m-IUGR piglets and increased for normal piglets. Our findings are in agreement with Baxter et al. (2008), who found higher plasma glucose levels in surviving piglets at 24 h (4.66 vs. 3.25 mmol/L) compared with dying piglets, with no differences in concentrations of plasma glucose (2.52 vs. 2.51 mmol/L) at birth (Baxter et al., 2008). In contrast, Tuchscherer et al. (2000) found glucose levels to be higher in piglets at risk (3.77 compared with 3.64 mmol/L) and concluded that reduced physiological maturity of piglets at birth is mainly responsible for a high risk of mortality during the first days of life (Tuchscherer et al., 2000). Generally, both their groups of piglets (survivors and piglets at risk) were heavier (1,368 and 1,063 g) than piglets in the present study. Differences in electrolytes were also observed between the 3 groups. In agreement with this, Wang et al. (2008) found that cellular signaling defects, redox imbalance, reduced protein synthesis, and enhanced proteolysis may be the major mechanisms responsible for abnormal absorption and metabolism of nutrients, as well as reduced growth and impaired development of the small intestine, liver, and muscle in IUGR piglets harvested at birth (Wang et al., 2008). This perhaps indicates that more electrolytes and/or more colostrum are needed for IUGR piglets. In addition, D'Inca et al. (2010) found IUGR in piglets to have a major impact on the length and weight of the small intestine, which was heavier in preterm IUGR neonates and longer and thinner in both preterm and term piglets at birth. Intestinal vulnerability in IUGR piglets may explain the more frequent and severe intestinal problems observed in the first few days of IUGR neonates (D'Inca et al., 2010), although it was not investigated in the present study. Severe-IUGR piglets had less glycogen left in their livers at 24 h compared with normal piglets, with that of m-IUGR piglets in between. The glycogen reserves of neonatal piglets are depleted quickly if piglets must metabolize glycogen in an attempt to maintain their body temperature (Herpin et al., 2002). Muscle glycogen concentration was halved in piglets over 24 h in a study by Pastorelli et al. (2009), emphasizing the importance to the neonatal piglet of an early external energy source (Pastorelli et al., 2009). In the current study, normal piglets had glycogen concentration in the liver comparable to that found in piglets at birth (Theil et al., 2011). However, not much is known on the ratio of pro- vs. macro-glycogen at birth or 24 h. Studies on pigs before slaughter have shown pro-glycogen, rather than macro-glycogen, to be the primary glycogen pool in the muscle from which energy was recruited for both exercise stress, as well as postmortem metabolism (Young et al., 2009). In the present study, s-IUGR piglets had less than half the total glycogen amount the normal piglets did and had depleted 10% more macro- than pro-glycogen, compared with the normal piglets. The liver weight and glycogen deposits increased with BW (Theil et al., 2011), indicating that a large birth weight automatically increases survival chances. Muscle glycogen was not assessed in the present study; however, it is likely that muscle glycogen is depleted, although at a slower rate, compared with that in the liver for the 3 groups of piglets. Based on the physiological data in this study, we suggest that it is possible to identify s-IUGR and m-IUGR piglets, which are in need of oral supplementation during the colostrum period. Additionally, the lowered capacity of these piglets to ingest colostrum seems to be of vital importance. Body mass index might be a good indicator of survival, where the higher the BMI value the greater the survival chance (Baxter et al., 2008), and BMI is a measure of piglet length and weight (weight per length). In addition, small piglets have a higher heat loss due to a high surface-to-volume ratio; i.e., they are less resistant to cooling (Theil et al., 2012) and therefore more likely to deplete energy stores. Piglets suffering from IUGR (s-IUGR) had lower BMI than piglets deemed normal. Hales et al. (2013) recently found that when 2 piglets have similar weight, the 1 with the highest BMI had the greater chance of surviving. This difference in BMI might be due to differences in muscle development of piglet categories. Small piglets have lower muscle mass (Rehfeldt and Kuhn, 2006) and a relative greater fat deposition at weaning (Amdi et al., unpublished), possibly due to this body type being better at withstanding restrictive environments (the mismatch theory; Gluckman et al., 2005). A higher BMI of 2 similar weight piglets might therefore indicate more muscle, more glycogen stores, and therefore an increased survival chance. Studies have shown that floor heating in the pen at birth significantly reduces mortality (Malmkvist et al., 2006) and that finding along with the data from the present study indicate that sufficient energy supply via colostrum is extremely important for increasing neonatal survival in large litters. Implications for Pig Production Identifying IUGR piglets on their head shape provides farmers with a fast and simple method of deciding which piglets need extra support to maximize neonatal survival. Piglets suffering from IUGR have less energy reserves left at 24 h compared with normal piglets and it can be speculated that many of these will be too weak to survive when second stage of lactogenesis is onset (Hartmann et al., 1984; Krogh et al., 2012) ∼33 h after parturition, whereas normal piglets would still be vigorous. Most differences were found between normal piglets and s-IUGR piglets, with less significant differences observed between m-IUGR and s-IUGR piglets, suggesting that if the piglet had some signs of IUGR, their energy reserves and body characteristics were altered. Although most piglets suffering from IUGR were still alive at the 24-h sampling time, their plasma glucose and lactate levels, as well as stomach content, all indicated that their energy supply and utilization of depots differed from that of normal piglets. Therefore, it is crucial to ensure that the milk supply for small piglets in the first few days of their life is as nutritious as possible. More research is needed on the fate and viability of piglets suffering from IUGR. In conclusion, piglet head shape (i.e., their IUGR score) in combination with weight is an easy way for the farmer to decide which piglets need an intervention. Piglets with a dolphin-like head shape ingest insufficient amounts of colostrum and hence have lower glucose levels at 24 h, and have been subjected to some degree of IUGR, as can be seen in their organ ratios. In addition, they have less glycogen left in the liver, both in total and relative liver weight. Based on observations in this study, if piglets are classified as m-IUGR or s-IUGR, glucose levels are altered, organ weights have suffered, and these factors combined suggest that if they do have a dolphin-like head shape, they are more susceptible to a low survival chance. LITERATURE CITED Adamo K. B. Graham T. E. 1998. Comparison of traditional measurements with macroglycogen and proglycogen analysis of muscle glycogen. J. Appl. Physiol.  84: 908– 913. Google Scholar CrossRef Search ADS PubMed  Amdi C. Hansen C. F. Krogh U. Oksbjerg N. Theil P. K. 2013. Less brain sparing in intrauterine growth restricted piglets born to sows fed Palm Fatty Acid Distillate. In:. Pluske J. Pluske J. editors, Manipulating pig production XIV.  Aust. Pig Sci. Assoc., Werribee, Australia. Baschat D. A. A. 2004. Fetal responses to placental insufficiency: An update. BJOG  111: 1031– 1041. Google Scholar CrossRef Search ADS PubMed  Bauer R. Walter B. Gaser E. Rösel T. Kluge H. Zwiener U. 1998a. Cardiovascular function and brain metabolites in normal weight and intrauterine growth restricted newborn piglets—Effect of mild hypoxia. Exp. Toxicol. Pathol.  50: 294– 300. Google Scholar CrossRef Search ADS   Bauer R. Walter B. Hoppe A. Gaser E. Lampe V. Kauf E. Zwiener U. 1998b. Body weight distribution and organ size in newborn swine (sus scrofa domestica)— A study describing an animal model for asymmetrical intrauterine growth retardation. Exp. Toxicol. Pathol.  50: 59– 65. Google Scholar CrossRef Search ADS   Baxter E. M. Jarvis S. D'Eath R. B. Ross D. W. Robson S. K. Farish M. Nevison I. M. Lawrence A. B. Edwards S. A. 2008. Investigating the behavioural and physiological indicators of neonatal survival in pigs. Theriogenology  69: 773– 783. Google Scholar CrossRef Search ADS PubMed  Campos P. H. R. F. Silva B. A. N. Donzele J. L. Oliveira R. F. M. Knol E. F. 2012. Effects of sow nutrition during gestation on within-litter birth weight variation: A review. Animal  6: 797– 806. Google Scholar CrossRef Search ADS PubMed  Chevaux E. Sacy A. Le Treut Y. Martineau G. 2010. Intrauterine Growth Retardation (IUGR): Morphological and behavioural description. In: Proc. 21st IPVS Congr., Vancouver Canada. p. 209. D'Inca R. Che L. Thymann T. Sangild P. T. Le Huërou-Luron I. 2010. Intrauterine growth restriction reduces intestinal structure and modifies the response to colostrum in preterm and term piglets. Livest. Sci.  133: 20– 22. Google Scholar CrossRef Search ADS   Devillers N. Le Dividich J. Prunier A. 2011. Influence of colostrum intake on piglet survival and immunity. Animal  5: 1605– 1612. Google Scholar CrossRef Search ADS PubMed  Devillers N. Milgen J. Prunier A. Le Dividich J. 2004. Estimation of colostrum intake in the neonatal pig. Anim. Sci.  78: 305– 313. Foxcroft G. R. Dixon W. T. Novak S. Putman C. T. Town S. C. Vinsky M. D. A. 2006. The biological basis for prenatal programming of postnatal performance in pigs. J. Anim. Sci.  84: E105– 108. Google Scholar CrossRef Search ADS PubMed  Gluckman P. D. Cutfield W. Hofman P. Hanson M. A. 2005. The fetal, neonatal, and infant environments–the long-term consequences for disease risk. Early Hum. Dev.  81: 51– 59. Google Scholar CrossRef Search ADS PubMed  Hales J. Moustsen V. A. Nielsen M. B. F. Hansen C. F. 2013. Individual physical characteristics of neonatal piglets affect preweaning survival of piglets born in a non-crated system. J. Anim. Sci.  91: 4991– 5003. Google Scholar CrossRef Search ADS PubMed  Hammond J. 1921. Further observations on the factors controlling fertility and foetal atrophy. J. Agric. Sci.  11: 337– 366. Google Scholar CrossRef Search ADS   Hartmann P. E. McCauley I. Gooneratne A. Whitely J. 1984. Inadequacies of sow lactation: Survival of the fittest. In: Peaker M. Vernon R. G. Knight C. H. editors, Physiological strategies in lactation.  Academic Press, London. p. 301– 26. Herpin P. Damon M. Le Dividich J. 2002. Development of thermoregulation and neonatal survival in pigs. Livest. Prod. Sci.  78: 25– 45. Google Scholar CrossRef Search ADS   Jørgensen L. 2005. Nutrient recommendations (Normer for næringsstoffer). Danish Pig Research Centre, Copenhagen, Denmark. Krogh U. Flummer C. Jensen S. K. Theil P. K. 2012. Colostrum and milk production of sows is affected by dietary conjugated linoleic acid. J. Anim. Sci.  90: 366– 368. Google Scholar CrossRef Search ADS PubMed  Malmkvist J. Pedersen L. J. Damgaard B. M. Thodberg K. Jørgensen E. Labouriau R. 2006. Does floor heating around parturition affect the vitality of piglets born to loose housed sows? Appl. Anim. Behav. Sci.  99: 88– 105. Google Scholar CrossRef Search ADS   Passonneau J. Lowry O. 1993. Enzymatic analysis: A practical guide. The Humana Press Inc., Clifton, NJ. Google Scholar CrossRef Search ADS   Pastorelli G. Neil M. Wigren I. 2009. Body composition and muscle glycogen contents of piglets of sows fed diets differing in fatty acids profile and contents. Livest. Sci.  123: 329– 334. Google Scholar CrossRef Search ADS   Quesnel H. Farmer C. Devillers N. 2012. Colostrum intake: Influence on piglet performance and factors of variation. Livest. Sci.  146: 105– 114. Google Scholar CrossRef Search ADS   Rehfeldt C. Kuhn G. 2006. Consequences of birth weight for postnatal growth performance and carcass quality in pigs as related to myogenesis. J. Anim. Sci.  84: E113– 115. Google Scholar CrossRef Search ADS PubMed  Roza S. J. Steegers E. A. Verburg B. O. Jaddoe V. W. Moll H. A. Hofman A. Verhulst F. C. Tiemeier H. 2008. What is spared by fetal brain-sparing? Fetal circulatory redistribution and behavioral problems in the general population. Am. J. Epidemiol.  168: 1145– 1152. Google Scholar CrossRef Search ADS PubMed  Theil P. K. Cordero G. Henckel P. Puggaard L. Oksbjerg N. Sørensen M. T. 2011. Effects of gestation and transition diets, piglet birth weight, and fasting time on depletion of glycogen pools in liver and 3 muscles of newborn piglets. J. Anim. Sci.  89: 1805– 1816. Google Scholar CrossRef Search ADS PubMed  Theil P. K. Nielsen M. O. Sørensen M. Lauridsen C. 2012. Lactation, milk and suckling. In: Bach K. E. Knudsen N. J. Kjeldsen H. D. Jensen B. B. editors, Nutritional physiology of pigs.  Pig Research Centre, Copenhagen, Denmark. p. 49. Tuchscherer M. Puppe B. Tuchscherer A. Tiemann U. 2000. Early identification of neonates at risk: Traits of newborn piglets with respect to survival. Theriogenology  54: 371– 388. Google Scholar CrossRef Search ADS PubMed  Wang J. Chen L. Li D. Yin Y. Wang X. Li P. Dangott L. J. Hu W. Wu G. 2008. Intrauterine growth restriction affects the proteomes of the small intestine, liver, and skeletal muscle in newborn pigs. J. Nutr.  138: 60– 66. Google Scholar CrossRef Search ADS PubMed  Young J. F. Bertram H. C. Oksbjerg N. 2009. Rest before slaughter ameliorates pre-slaughter stress-induced increased drip loss but not stress-induced increase in the toughness of pork. Meat. Sci.  83: 634– 641. Google Scholar CrossRef Search ADS PubMed  Footnotes 1 The research project was funded by the Ministry of Food, Agriculture and Fisheries of Denmark, grant no. 3405-11-0342. Charlotte Amdi's postdoc is funded by the Villum Foundation. American Society of Animal Science

Gao, Y.;Han, F.;Huang, X.;Rong, Y.;Yi, H.;Wang, Y.

doi: 10.2527/jas.2013-6528pmid: 24126267

Abstract This study hypothesized that the gut microbial populations, intestinal morphology, and cytokine production are differentially altered in 2 different pig breeds, namely, Chinese native Jinhua pigs and European Landrace pigs, after orally challenge with enterotoxigenic Escherichia coli (ETEC) K88. A total of 12 Jinhua pigs and 12 Landrace pigs were allocated to either the nonchallenged or the challenged groups (6 pigs per group). The challenged pigs were orally administered ETEC K88, and their nonchallenged counterparts were given sterile Luria-Bertani broth. Selected gut microbial populations, intestinal morphology, mRNA expression of tight junction proteins, and the levels of ileal cytokines and secretory immunoglobulin A (sIgA) production were measured in Jinhua and Landrace pigs. The results showed that the challenged Jinhua pigs exhibited a significantly (P < 0.05) lower incidence of diarrhea compared with their Landrace counterparts. The Escherichia coli (E.coli) population and the percentage of E. coli in the total bacteria population were increased in response to ETEC K88 challenge in both Jinhua and Landrace pigs. The challenged Landrace pigs shed more E. coli (P < 0.05) and had higher percentage of E. coli in the total bacteria population in the colon (P < 0.05) compared with their Jinhua counterparts. Both pig breeds tended to exhibit greater villous atrophy and crypt depth reduction in all of the intestinal segments with challenge. The expression of tight junction proteins decreased in response to ETEC K88 challenge in both pig breeds. The levels of the proinflammatory cytokines interferon (IFN)-γ, tumor necrosis factor-α, and IL-6 and the secretion of sIgA were positively altered whereas the levels of the anti-inflammatory cytokine IL-4 and transforming growth factor (TGF)-β were negatively altered by ETEC K88 challenge in both breeds. Jinhua pigs exhibited significantly higher levels of IFN-γ and TGF-β (P < 0.05) in the challenged group. Our findings provide valuable evidence to explain the differences in the intestinal physiology between Jinhua and Landrace pigs; that is, Jinhua pigs appeared to show better growth performance, a lower incidence of diarrhea, and a lower extent of immune activation in response to ETEC K88 challenge and a higher Lactobacillus population, a higher percentage of Lactobacillus in the total bacteria population, a higher ratio of Lactobacillus to E. coli, and higher levels of tight junction proteins with and without challenge. INTRODUCTION The small intestine is the major site of nutrient digestion and absorption. In addition, the intestinal epithelium, the gut associated lymphoid tissues, and the gut microbiota play crucial roles in the prevention of pathogenic microorganisms, toxins, and allergenic macromolecules from entering the interior space of the body and the maintenance of the gastrointestinal tract (GIT) homeostasis (Moreto and Perez-Bosque, 2009). As such, the maintenance of normal gastrointestinal functions is important for animal health. The intestinal functions of pigs may be altered by several physiological, pathological, psychological, and pharmacological factors (Lambert, 2009). Among those factors, piglets are susceptible to a number of pathogenic bacterial diseases, particularly during the weaning transition period. One of the most common diseases is postweaning colibacillosis (PWC), which is caused by serotypes of enterotoxigenic Escherichia coli (ETEC) in the first 2 wk after weaning (Kim et al., 2012). Enterotoxigenic E. coli uses extracellular fimbriae (e.g., K88, K99, and K18) that adhere to specific receptors on the surface of intestinal epithelia. After adhesion, ETEC can release at least 1 member of 2 defined groups of enterotoxins, namely, heat stable enterotoxin and heat-labile enterotoxin (LT) (Nataro and Kaper, 1998), which can induce fluid excretion and cause severe diarrheal disease in the weaning piglets. Infection with ETEC also results in a loss of gut barrier function (Ewaschuk et al., 2011), the induction of mucosal immune responses (Lessard et al., 2009), and the disruption of the microbial homeostasis in the GIT of pigs (Halas et al., 2010; Li et al., 2012). Pigs of different breeds have evolved diverse characteristics due to geographical separation and different domestication conditions. Chinese native pig breeds, such as Meishan, and European-originated breeds, such as Landrace, exhibit distinctive physiological and immunological alterations in response to stress (Sutherland et al., 2006). Nevertheless, the available data that compare the intestinal physiology between different pig breeds, particularly in response to ETEC infection, are scarce. Therefore, the present study aimed to compare the effects of ETEC K88 challenge on selected gut microbial populations, the intestinal morphology, the expression of tight junction proteins, and the production of cytokines and secretory immunoglobulin A (sIgA) between Chinese Jinhua and European Landrace pigs. The objective of this study was to determine the differential intestinal responses between the 2 pig breeds challenged with ETEC K88 and to gain valuable insight into the underlying mechanisms. MATERIALS AND METHODS Bacterial Strains and Culture Enterotoxigenic E. coli K88 C83907 was purchased from the China Institute of Veterinary Drugs Control (Beijing, China) and preserved by the National Engineering Laboratory of Bio-feed Safety and Pollution Prevention (Hangzhou, China). This strain was confirmed to be positive for K88 and the virulent factors, heat-stable enterotoxinii and LT, as determined by PCR genotyping. The bacteria was resuscitated in 3 mL of Luria-Bertani (LB) broth at 37°C for 24 h with shaking and plated onto LB agar. A single colony was inoculated into 50 mL of LB broth, cultured overnight at 37°C and 250 rpm, and then subcultured and serially diluted on LB agar for bacterial enumeration. A final concentration of 5.4 × 109 cfu/mL was used in this experiment. The following bacterial cultures were used as positive controls to generate standard curves that could be used to determine the select phyla and genera within the pig digesta through quantitative real-time PCR. Escherichia coli ATCC25922 and Bacillus subtilis were grown aerobically in LB broth at 37°C. Lactobacillus acidophilus ATCC4356 and Bifidobacterium suis ATCC27533 were grown anaerobically in peptone yeast glucose and de Man–Rogosa–Sharpe media, respectively, at 37°C. Animals, Challenge Procedures, and Sample Collection The animal protocols used in the present study followed the guidelines stated in the Guide for the Care and Use of Agricultural Animals in Research and Teaching and were approved by the animal care committee of Zhejiang University. A total of 24 weaned pigs, that is, 12 Jinhua pigs (with an initial body weight of 7.46 ± 0.22 kg) and 12 Landrace pigs (with an initial body weight of 8.10 ± 0.21 kg), were assigned to 4 experimental treatment groups (6 pigs per treatment): Jinhua nonchallenge treatment group, Landrace nonchallenge treatment group, Jinhua challenge treatment group, and Landrace challenge treatment group. All of the pigs were penned individually throughout the 10-d experiment and allowed ad libitum access to feed and water throughout the study. The pigs were inoculated orally using a syringe attached to a polyethylene tube on d 1, 3, 5, and 7 between 1400 and 1600 h. Each pig was administered 100 mL of LB broth containing 109 cfu/mL of ETEC K88 or 100 mL of sterile LB broth. The ADG, ADFI, and G:F of each pig were monitored throughout the experimental period. The number of pigs with diarrhea was recorded daily, and the diarrhea ratio was calculated using the following equation: diarrhea ratio = total number of pigs with diarrhea/(total number of experimental pigs × trial days) × 100. On d 11, all of the pigs were euthanized, and tissues from the duodenum, jejunum, and ileum as well as colonic and cecal digesta samples were removed and immediately frozen in liquid nitrogen. The samples were stored at –80°C until analysis. Gut Microbial Population Analysis The total genomic DNA from the different reference strains that were used to generate standard curves was extracted using the QIAamp UCP Pathogen Mini Kit (Qiagen, Valencia,, CA). The genomic DNA from the colonic and cecal digesta samples were extracted using the QIAamp DNA Stool Mini kit (Qiagen). The species-specific PCR primers used to quantify the total bacteria and the other 4 different bacterial species are listed in Table 1. The quantitative detections of total bacteria, E. coli, Lactobacillus, Bifidobacterium, and Bacillus were performed using a culture-independent quantitative real-time PCR method that was modified from previously studies (Layton et al., 2006; Furet et al., 2009). Briefly, standard plasmids that contained the 16s rRNA genes that were amplified using the primers listed in Table 1 were constructed, and the gene copy numbers were calculated using the following formula: (DNA concentration in μg/μL × 6.02 × 1023 copies/mol)/(DNA size (bp) × 660 ×106). Standard curves were generated from 10-fold dilutions of the plasmids with gene copy numbers ranging from 106 to 1013 by plotting the threshold cycles as a function of the bacterial gene copies. Table 1. Oligonucleotide primers used for the real-time PCR analysis of the selected microbial populations in colonic and cecal digesta Targeted bacterial group  Amplicon size (bp)  Primer sequence (5′-3′)1  Annealing temperature (°C)  Reference  Total bacteria  145  F: CGGTGAATACGTTCYCGG  50  Suzuki et al., 2000  R: GGWTACCTTGTTACGACTT  Escherichia coli  96  F: CATGCCGCGTGTATGAAGAA  57.5  Qi et al., 2011  R: CGGGTAACGTCAATGAGCAAA  Lactobacillus  186  F: CGATGAGTGCTAGGTGTTGGA  60  Petersson et al., 2010  R: CAAGATGTCAAGACCTGGTAAG  Bacillus  92  F: GCAACGAGCGCAACCCTTGA  60  Qi, H., 2011  R: TCATCCCCACCTTCCTCCGG  Bifidobacterium  121  F: CGCGTCCGGTGTGAAAG  55  Qi et al., 2011  R: CTTCCCGATATCTACACATTCC  Targeted bacterial group  Amplicon size (bp)  Primer sequence (5′-3′)1  Annealing temperature (°C)  Reference  Total bacteria  145  F: CGGTGAATACGTTCYCGG  50  Suzuki et al., 2000  R: GGWTACCTTGTTACGACTT  Escherichia coli  96  F: CATGCCGCGTGTATGAAGAA  57.5  Qi et al., 2011  R: CGGGTAACGTCAATGAGCAAA  Lactobacillus  186  F: CGATGAGTGCTAGGTGTTGGA  60  Petersson et al., 2010  R: CAAGATGTCAAGACCTGGTAAG  Bacillus  92  F: GCAACGAGCGCAACCCTTGA  60  Qi, H., 2011  R: TCATCCCCACCTTCCTCCGG  Bifidobacterium  121  F: CGCGTCCGGTGTGAAAG  55  Qi et al., 2011  R: CTTCCCGATATCTACACATTCC  1F = Forward primer; R = Reverse primer. View Large Table 1. Oligonucleotide primers used for the real-time PCR analysis of the selected microbial populations in colonic and cecal digesta Targeted bacterial group  Amplicon size (bp)  Primer sequence (5′-3′)1  Annealing temperature (°C)  Reference  Total bacteria  145  F: CGGTGAATACGTTCYCGG  50  Suzuki et al., 2000  R: GGWTACCTTGTTACGACTT  Escherichia coli  96  F: CATGCCGCGTGTATGAAGAA  57.5  Qi et al., 2011  R: CGGGTAACGTCAATGAGCAAA  Lactobacillus  186  F: CGATGAGTGCTAGGTGTTGGA  60  Petersson et al., 2010  R: CAAGATGTCAAGACCTGGTAAG  Bacillus  92  F: GCAACGAGCGCAACCCTTGA  60  Qi, H., 2011  R: TCATCCCCACCTTCCTCCGG  Bifidobacterium  121  F: CGCGTCCGGTGTGAAAG  55  Qi et al., 2011  R: CTTCCCGATATCTACACATTCC  Targeted bacterial group  Amplicon size (bp)  Primer sequence (5′-3′)1  Annealing temperature (°C)  Reference  Total bacteria  145  F: CGGTGAATACGTTCYCGG  50  Suzuki et al., 2000  R: GGWTACCTTGTTACGACTT  Escherichia coli  96  F: CATGCCGCGTGTATGAAGAA  57.5  Qi et al., 2011  R: CGGGTAACGTCAATGAGCAAA  Lactobacillus  186  F: CGATGAGTGCTAGGTGTTGGA  60  Petersson et al., 2010  R: CAAGATGTCAAGACCTGGTAAG  Bacillus  92  F: GCAACGAGCGCAACCCTTGA  60  Qi, H., 2011  R: TCATCCCCACCTTCCTCCGG  Bifidobacterium  121  F: CGCGTCCGGTGTGAAAG  55  Qi et al., 2011  R: CTTCCCGATATCTACACATTCC  1F = Forward primer; R = Reverse primer. View Large For the detection of bacteria in the digesta samples, real-time PCR assays were performed on a StepOnePlus Real-Time PCR System (Applied Biosystems, Life Technologies, Carlsbad, CA) using optical-grade 96-well plates. The reaction mixture (20 μL) consisted of 10 μL of SYBR Premix Ex Taq kit (Takara, Dalian, China), 1 μL of ROX Dye, 1 μL (10 nM) of each primer set, and 1 μL of the template DNA. The PCR conditions were 30 s at 95°C and 40 cycles for 5 s at 95°C, 30 s at the annealing temperature (Table 1), and 1 min at 72°C. Each of the PCR runs detected the standards and samples in triplicate. The gene copies were normalized for each digesta sample, and the data are presented as the log number of bacterial gene copies per gram of digesta sample. The lower detection limit for the bacterial enumeration was 106 copies/g of digesta samples. Intestinal Morphology The intestinal morphology was measured using the method described by Moeser et al. (2012). In general, middle duodenum and jejunum sections and distal ileum sections were fixed in 10% PBS buffered formalin and embedded in paraffin. Five-millimeter sections were stained with hematoxylin and eosin for histological analysis. The images were acquired using a Leica microscope (DM3000; Leica, Wetzlar, Germany) equipped with a CCD camera (DFC420; Leica). Before imaging, the system was calibrated at each magnification using a stage micrometer. The villi and crypts were measured using the “measure distances” tool of the Image-Pro Plus 6.0 software (MediaCybernetics, Rockville, MD) with the 20X objective. A minimum of 3 villi from each pig were measured. Analysis of the mRNA Expression of Intestinal Tight Junction Proteins Specific primers were designed based on the published sequences of the claudin-1, occludin, zonula occludens protein-1 (ZO-1), and zonula occludens protein-2 (ZO-2) genes in pigs using the National Center for Biotechnology Information online primer design tool (www.ncbi.nlm.nih.gov/tools/primer-blast; Table 2). The expression of 18S rRNA was used for normalization. The total RNA from tissues of different intestine segments (duodenum, jejunum, and ileum) was extracted using the TRIzol Reagent (Invitrogen). The yield and purity of the RNA extracts were determined using a Nanodrop 2000 spectrophotometer (Thermo-Fisher Scientific, Waltham, MA). The RNA integrity was inspected on a 1% (w/v) agarose gel after it was mixed with 2X RNA Loading Dye (Thermo-Fisher Scientific). Two micrograms of total RNA of each sample was used for reverse transcription in a 20-μL reaction system containing 1 μg of random hexamers, 1 mM deoxyribonucleotide triphosphate mixture, and 200 U of M-MuLV Reverse Transcriptase (Thermo-Fisher Scientific). The reactions were incubated at 25°C for 20 min and at 42°C for 60 min and terminated at 70°C for 10 min. The real-time PCR reactions were performed using the conditions described previously and the following temperature program: a precycling stage at 95°C for 30 s and 40 cycles of denaturization at 95°C for 5 s and annealing at 60°C for 34 s. The fluorescence was detected at the end of each annealing step, and the melting curves were monitored to confirm the specificity of the PCR products. The 2–ΔΔCT method (Livak and Schmittgen, 2001) was used to determine the mRNA expression levels. Table 2. Oligonucleotide primers used for the real-time PCR analysis of the mRNA expression levels of intestinal tight junction proteins Genes  Amplicon size (bp)  Primer sequence (5′-3′)  Annealing temperature (°C)  Reference  claudin-1  214  F: ATTTCAGGTCTGGCTATCTTAGTTGC  60  this study  R: AGGGCCTTGGTGTTGGGTAA  occludin  157  F: ATCAACAAAGGCAACTCT  60  this study  R: GCAGCAGCCATGTACTCT  ZO-1  147  F: AGCCCGAGGCGTGTTT  60  this study  R: GGTGGGAGGATGCTGTTG  ZO-2  89  F: ATTCGGACCCATAGCAGACATAG  60  this study  R: GCGTCTCTTGGTTCTGTTTTAGC  18S  122  F: CCCACGGAATCGAGAAAGAG  60  Shan et al., 2009  R: TTGACGGAAGGGCACCA  Genes  Amplicon size (bp)  Primer sequence (5′-3′)  Annealing temperature (°C)  Reference  claudin-1  214  F: ATTTCAGGTCTGGCTATCTTAGTTGC  60  this study  R: AGGGCCTTGGTGTTGGGTAA  occludin  157  F: ATCAACAAAGGCAACTCT  60  this study  R: GCAGCAGCCATGTACTCT  ZO-1  147  F: AGCCCGAGGCGTGTTT  60  this study  R: GGTGGGAGGATGCTGTTG  ZO-2  89  F: ATTCGGACCCATAGCAGACATAG  60  this study  R: GCGTCTCTTGGTTCTGTTTTAGC  18S  122  F: CCCACGGAATCGAGAAAGAG  60  Shan et al., 2009  R: TTGACGGAAGGGCACCA  View Large Table 2. Oligonucleotide primers used for the real-time PCR analysis of the mRNA expression levels of intestinal tight junction proteins Genes  Amplicon size (bp)  Primer sequence (5′-3′)  Annealing temperature (°C)  Reference  claudin-1  214  F: ATTTCAGGTCTGGCTATCTTAGTTGC  60  this study  R: AGGGCCTTGGTGTTGGGTAA  occludin  157  F: ATCAACAAAGGCAACTCT  60  this study  R: GCAGCAGCCATGTACTCT  ZO-1  147  F: AGCCCGAGGCGTGTTT  60  this study  R: GGTGGGAGGATGCTGTTG  ZO-2  89  F: ATTCGGACCCATAGCAGACATAG  60  this study  R: GCGTCTCTTGGTTCTGTTTTAGC  18S  122  F: CCCACGGAATCGAGAAAGAG  60  Shan et al., 2009  R: TTGACGGAAGGGCACCA  Genes  Amplicon size (bp)  Primer sequence (5′-3′)  Annealing temperature (°C)  Reference  claudin-1  214  F: ATTTCAGGTCTGGCTATCTTAGTTGC  60  this study  R: AGGGCCTTGGTGTTGGGTAA  occludin  157  F: ATCAACAAAGGCAACTCT  60  this study  R: GCAGCAGCCATGTACTCT  ZO-1  147  F: AGCCCGAGGCGTGTTT  60  this study  R: GGTGGGAGGATGCTGTTG  ZO-2  89  F: ATTCGGACCCATAGCAGACATAG  60  this study  R: GCGTCTCTTGGTTCTGTTTTAGC  18S  122  F: CCCACGGAATCGAGAAAGAG  60  Shan et al., 2009  R: TTGACGGAAGGGCACCA  View Large Enzyme-Linked Immunosorbent Assays of the Cytokine and Secretory Immunoglobulin A Concentration in the Ileum Intestinal segments were removed from distal ileum of each piglet, rinsed with 1x PBS to remove excess blood, homogenized in 1 mL of 1x PBS, and stored overnight at –20°C. After 2 freeze–thaw cycles to break up the cell membranes, the homogenate was centrifuged at 5,000 × g for 5 min at 4°C, and the supernatant was collected, aliquoted in a 1.5 mL tube, and stored at –20°C until use. The ileal cytokine interferon (IFN)-γ, tumor necrosis factor (TNF)-α, IL-6, IL-4, transforming growth factor (TGF)-β, and sIgA levels (pg/mL) were measured using a commercially available ELISA Kit (Cusabio Biotech Co., Ltd., Wuhan, China) according to the manufacturer's protocol. Statistical Analysis All of the statistical analyses were performed in SPSS 16.0 (SPSS Inc., Chicago, IL). The generalized linear model (GLM) procedure was used to analyze the breed, ETEC K88 challenge, and their interaction as a source of variation. The significance of the differences between all of the treatment groups was analyzed by 1-way ANOVA and means separation using Duncan's significant difference test. A significance level of 0.05 was used as default. RESULTS Growth Performance and Diarrhea Ratio Before the experiment, none of the animals showed clinical signs of postweaning diarrhea. The ETEC K88 challenge resulted in a decreased growth rate and an increased incidence of diarrhea in both Jinhua and Landrace pigs (Table 3). The ADG of Jinhua and Landrace pigs were decreased by 32.1 and 28.6%, respectively. The ADG of the challenged Jinhua pigs (0.19 ± 0.01 kg/d) was significantly higher (P < 0.05) compared with their Landrace counterparts (0.10 ± 0.02 kg/d). The breed (P < 0.001) and ETEC K88 challenge (P < 0.01) strongly affected the ADG throughout the experimental period. The breed exerted a significant effect on the ADFI (P < 0.05). The Jinhua pigs presented a significantly higher ADFI (P < 0.05) compared with the Landrace pigs in both the challenged and the nonchallenged groups. The ADFI was decreased by 14.0 and 12.5% in Jinhua and Landrace pigs, respectively, compared with the respective nonchallenged pigs. The interaction between breed and ETEC K88 challenge had no significant effect on the ADG (P = 0.186) or the ADFI (P = 0.660). No significant difference was observed in the G:F between the different experimental treatment groups. The challenged Jinhua pigs showed a significantly lower diarrhea ratio compared with their Landrace counterparts (7.1 vs. 26.2%, respectively). The breed (P = 0.001), ETEC K88 challenge (P < 0.001), and their interaction (P = 0.006) exerted significant effects on the diarrhea ratio. Table 3. Effects of enterotoxigenic Escherichia coli (ETEC) K88 challenge on weight gain, feed intake, feed efficiency, and incidence of diarrhea in Jinhua and Landrace pigs   Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  ADG, kg/d  0.28a  0.14bc  0.19b  0.10c  0.02  <0.001***  0.005**  0.186  ADFI, kg/d  0.57a  0.32b  0.49a  0.28b  0.03  <0.001***  0.224  0.660  G:F  0.50  0.44  0.39  0.37  0.02  0.413  0.084  0.741  Diarrhea ratio, %  0b  7.1b  2.4b  26.2a  0.03  0.001**  <0.001***  0.006**    Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  ADG, kg/d  0.28a  0.14bc  0.19b  0.10c  0.02  <0.001***  0.005**  0.186  ADFI, kg/d  0.57a  0.32b  0.49a  0.28b  0.03  <0.001***  0.224  0.660  G:F  0.50  0.44  0.39  0.37  0.02  0.413  0.084  0.741  Diarrhea ratio, %  0b  7.1b  2.4b  26.2a  0.03  0.001**  <0.001***  0.006**  a,b,cThe superscript letters indicate a significant difference within a row (P < 0.05). 1JN = nonchallenged Jinhua pigs; LN = nonchallenged Landrace pigs; JC = challenged Jinhua pigs; LC = challenged Landrace pigs. 2B = breed; C = ETEC K88 challenge; B × C, breed and ETEC K88 challenge interaction. *P < 0.05; **P < 0.01; ***P < 0.001. View Large Table 3. Effects of enterotoxigenic Escherichia coli (ETEC) K88 challenge on weight gain, feed intake, feed efficiency, and incidence of diarrhea in Jinhua and Landrace pigs   Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  ADG, kg/d  0.28a  0.14bc  0.19b  0.10c  0.02  <0.001***  0.005**  0.186  ADFI, kg/d  0.57a  0.32b  0.49a  0.28b  0.03  <0.001***  0.224  0.660  G:F  0.50  0.44  0.39  0.37  0.02  0.413  0.084  0.741  Diarrhea ratio, %  0b  7.1b  2.4b  26.2a  0.03  0.001**  <0.001***  0.006**    Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  ADG, kg/d  0.28a  0.14bc  0.19b  0.10c  0.02  <0.001***  0.005**  0.186  ADFI, kg/d  0.57a  0.32b  0.49a  0.28b  0.03  <0.001***  0.224  0.660  G:F  0.50  0.44  0.39  0.37  0.02  0.413  0.084  0.741  Diarrhea ratio, %  0b  7.1b  2.4b  26.2a  0.03  0.001**  <0.001***  0.006**  a,b,cThe superscript letters indicate a significant difference within a row (P < 0.05). 1JN = nonchallenged Jinhua pigs; LN = nonchallenged Landrace pigs; JC = challenged Jinhua pigs; LC = challenged Landrace pigs. 2B = breed; C = ETEC K88 challenge; B × C, breed and ETEC K88 challenge interaction. *P < 0.05; **P < 0.01; ***P < 0.001. View Large Gut Microbial Population Analysis The colonic and cecal microbial populations of total bacteria, E. coli, Bacillus, Bifidobacterium, and Lactobacillus (log 16S rRNA gene copies/g of digesta) as well as the percentage of E. coli in the total bacteria population, the percentage of Lactobacillus in the total bacteria population, and the ratio of Lactobacillus to E. coli in Jinhua and Landrace pigs are shown in Table 4. The effects of the breed and ETEC K88 challenge on the colonic and cecal microbial populations, percentages, and ratios were found to be site dependent. In the colon, the total bacterial population was not affected by the breed and ETEC K88 challenge but was significantly affected by their interaction (P = 0.029). The ETEC K88 challenge significantly increased the populations of E. coli (P < 0.05) in both pig breeds. The Jinhua pigs exhibited a significantly lower colonic (P < 0.05) population of E. coli compared with their Landrace counterparts, regardless of whether the pigs were challenged. Moreover, the colonic E. coli populations were strongly affected by the breed (P = 0.001) and ETEC K88 challenge (P < 0.001). The population of colonic Lactobacillus (P < 0.05) was significantly reduced in response to ETEC K88 challenge in Landrace pigs and was strongly affected by the interaction between breed and ETEC K88 challenge (P = 0.022). The population of Bacillus (P < 0.05) was increased only in the challenged Jinhua pigs and was significantly influenced by the interaction between breed and ETEC K88 challenge (P = 0.007). The colonic Bifidobacterium populations did not differ between the treatment groups. Table 4. Effects of enterotoxigenic Escherichia coli (ETEC) K88 challenge on colonic and cecal microbial populations of total bacteria, Escherichia coli, Lactobacillus, Bacillus, Bifidobacterium, and the ratio of Escherichia coli to total bacteria, Lactobacillus to total bacteria, and Lactobacillus to Escherichia coli (%) in Jinhua and Landrace pigs   Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  Bacterial population      Colon          Total bacteria  12.00  12.18  12.18  11.99  0.04  0.956  0.952  0.029*          Escherichia coli  8.60c  9.40b  9.72b  10.16a  0.15  0.001**  <0.001***  0.235          Lactobacillus  10.42ab  10.63a  10.56ab  10.33b  0.05  0.893  0.417  0.022*          Bacillus  9.24b  9.44b  9.58a  9.29ab  0.05  0.610  0.282  0.007**          Bifidobacterium  7.97  8.25  8.19  8.20  0.05  0.200  0.423  0.215      Cecum          Total bacteria  11.92  11.85  11.72  12.06  0.07  0.316  0.993  0.134          Escherichia coli  9.40b  9.50ab  9.62ab  10.00a  0.10  0.203  0.061  0.455          Lactobacillus  10.26  10.37  10.04  10.19  0.05  0.216  0.077  0.812          Bacillus  8.99  8.94  8.88  9.21  0.07  0.331  0.569  0.183          Bifidobacterium  7.82ab  7.75ab  8.16b  8.27a  0.08  0.011*  0.887  0.573  Bacterial ratio      Colon3          E. coli as % of total bact.  0.03b  0.32b  0.48b  1.33a  0.13  0.002**  <0.001***  0.093          Lact. as % of total bact.  3.06a  2.57ab  2.79ab  2.19b  0.13  0.031*  0.175  0.802          Lact:E. coli  70.66a  21.62b  5.81c  2.19c  7.91  <0.001*  <0.001*  <0.001*      Cecum          E. coli as % of total bact.  0.24c  0.38bc  0.80ab  0.98a  0.11  0.311  0.004**  0.876          Lact. as % of total bact.  2.25  2.31  2.13  2.17  0.18  0.889  0.734  0.972          Lact:E. coli  5.05  3.35  3.94  2.55  0.42  0.250  0.076  0.847    Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  Bacterial population      Colon          Total bacteria  12.00  12.18  12.18  11.99  0.04  0.956  0.952  0.029*          Escherichia coli  8.60c  9.40b  9.72b  10.16a  0.15  0.001**  <0.001***  0.235          Lactobacillus  10.42ab  10.63a  10.56ab  10.33b  0.05  0.893  0.417  0.022*          Bacillus  9.24b  9.44b  9.58a  9.29ab  0.05  0.610  0.282  0.007**          Bifidobacterium  7.97  8.25  8.19  8.20  0.05  0.200  0.423  0.215      Cecum          Total bacteria  11.92  11.85  11.72  12.06  0.07  0.316  0.993  0.134          Escherichia coli  9.40b  9.50ab  9.62ab  10.00a  0.10  0.203  0.061  0.455          Lactobacillus  10.26  10.37  10.04  10.19  0.05  0.216  0.077  0.812          Bacillus  8.99  8.94  8.88  9.21  0.07  0.331  0.569  0.183          Bifidobacterium  7.82ab  7.75ab  8.16b  8.27a  0.08  0.011*  0.887  0.573  Bacterial ratio      Colon3          E. coli as % of total bact.  0.03b  0.32b  0.48b  1.33a  0.13  0.002**  <0.001***  0.093          Lact. as % of total bact.  3.06a  2.57ab  2.79ab  2.19b  0.13  0.031*  0.175  0.802          Lact:E. coli  70.66a  21.62b  5.81c  2.19c  7.91  <0.001*  <0.001*  <0.001*      Cecum          E. coli as % of total bact.  0.24c  0.38bc  0.80ab  0.98a  0.11  0.311  0.004**  0.876          Lact. as % of total bact.  2.25  2.31  2.13  2.17  0.18  0.889  0.734  0.972          Lact:E. coli  5.05  3.35  3.94  2.55  0.42  0.250  0.076  0.847  a,b,cThe superscript letters indicate a significant difference within a row (P < 0.05). 1JN = nonchallenged Jinhua pigs; LN = nonchallenged Landrace pigs; JC = challenged Jinhua pigs; LC = challenged Landrace pigs. 2B, breed; C, ETEC K88 challenge; B × C, breed and ETEC K88 challenge interaction. 3E. coli as % of total bact. = Escherichia coli as percentage of total bacteria; Lact. as % of total bact. = Lactobacillus as percentage of total bacteria; Lact:E. coli, ratio of Lactobacillus to Escherichia coli. *P < 0.05; **P < 0.01; ***P < 0.001. View Large Table 4. Effects of enterotoxigenic Escherichia coli (ETEC) K88 challenge on colonic and cecal microbial populations of total bacteria, Escherichia coli, Lactobacillus, Bacillus, Bifidobacterium, and the ratio of Escherichia coli to total bacteria, Lactobacillus to total bacteria, and Lactobacillus to Escherichia coli (%) in Jinhua and Landrace pigs   Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  Bacterial population      Colon          Total bacteria  12.00  12.18  12.18  11.99  0.04  0.956  0.952  0.029*          Escherichia coli  8.60c  9.40b  9.72b  10.16a  0.15  0.001**  <0.001***  0.235          Lactobacillus  10.42ab  10.63a  10.56ab  10.33b  0.05  0.893  0.417  0.022*          Bacillus  9.24b  9.44b  9.58a  9.29ab  0.05  0.610  0.282  0.007**          Bifidobacterium  7.97  8.25  8.19  8.20  0.05  0.200  0.423  0.215      Cecum          Total bacteria  11.92  11.85  11.72  12.06  0.07  0.316  0.993  0.134          Escherichia coli  9.40b  9.50ab  9.62ab  10.00a  0.10  0.203  0.061  0.455          Lactobacillus  10.26  10.37  10.04  10.19  0.05  0.216  0.077  0.812          Bacillus  8.99  8.94  8.88  9.21  0.07  0.331  0.569  0.183          Bifidobacterium  7.82ab  7.75ab  8.16b  8.27a  0.08  0.011*  0.887  0.573  Bacterial ratio      Colon3          E. coli as % of total bact.  0.03b  0.32b  0.48b  1.33a  0.13  0.002**  <0.001***  0.093          Lact. as % of total bact.  3.06a  2.57ab  2.79ab  2.19b  0.13  0.031*  0.175  0.802          Lact:E. coli  70.66a  21.62b  5.81c  2.19c  7.91  <0.001*  <0.001*  <0.001*      Cecum          E. coli as % of total bact.  0.24c  0.38bc  0.80ab  0.98a  0.11  0.311  0.004**  0.876          Lact. as % of total bact.  2.25  2.31  2.13  2.17  0.18  0.889  0.734  0.972          Lact:E. coli  5.05  3.35  3.94  2.55  0.42  0.250  0.076  0.847    Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  Bacterial population      Colon          Total bacteria  12.00  12.18  12.18  11.99  0.04  0.956  0.952  0.029*          Escherichia coli  8.60c  9.40b  9.72b  10.16a  0.15  0.001**  <0.001***  0.235          Lactobacillus  10.42ab  10.63a  10.56ab  10.33b  0.05  0.893  0.417  0.022*          Bacillus  9.24b  9.44b  9.58a  9.29ab  0.05  0.610  0.282  0.007**          Bifidobacterium  7.97  8.25  8.19  8.20  0.05  0.200  0.423  0.215      Cecum          Total bacteria  11.92  11.85  11.72  12.06  0.07  0.316  0.993  0.134          Escherichia coli  9.40b  9.50ab  9.62ab  10.00a  0.10  0.203  0.061  0.455          Lactobacillus  10.26  10.37  10.04  10.19  0.05  0.216  0.077  0.812          Bacillus  8.99  8.94  8.88  9.21  0.07  0.331  0.569  0.183          Bifidobacterium  7.82ab  7.75ab  8.16b  8.27a  0.08  0.011*  0.887  0.573  Bacterial ratio      Colon3          E. coli as % of total bact.  0.03b  0.32b  0.48b  1.33a  0.13  0.002**  <0.001***  0.093          Lact. as % of total bact.  3.06a  2.57ab  2.79ab  2.19b  0.13  0.031*  0.175  0.802          Lact:E. coli  70.66a  21.62b  5.81c  2.19c  7.91  <0.001*  <0.001*  <0.001*      Cecum          E. coli as % of total bact.  0.24c  0.38bc  0.80ab  0.98a  0.11  0.311  0.004**  0.876          Lact. as % of total bact.  2.25  2.31  2.13  2.17  0.18  0.889  0.734  0.972          Lact:E. coli  5.05  3.35  3.94  2.55  0.42  0.250  0.076  0.847  a,b,cThe superscript letters indicate a significant difference within a row (P < 0.05). 1JN = nonchallenged Jinhua pigs; LN = nonchallenged Landrace pigs; JC = challenged Jinhua pigs; LC = challenged Landrace pigs. 2B, breed; C, ETEC K88 challenge; B × C, breed and ETEC K88 challenge interaction. 3E. coli as % of total bact. = Escherichia coli as percentage of total bacteria; Lact. as % of total bact. = Lactobacillus as percentage of total bacteria; Lact:E. coli, ratio of Lactobacillus to Escherichia coli. *P < 0.05; **P < 0.01; ***P < 0.001. View Large The microbial populations in the cecum were less altered compared with those of the colon. The breed, ETEC K88 challenge and their interaction had no effect on the cecal populations of total bacteria, E. coli, Lactobacillus, and Bacillus. Only the population of Bifidobacterium was markedly affected by the breed (P = 0.011), but it was not affected by ETEC K88 challenge and the interaction of breed and ETEC K88 challenge. The ETEC K88 challenge tended to affect the population of E. coli (P = 0.061) and Lactobacillus (P = 0.077). The percentage of E. coli in the total bacteria population in the colon was strongly affected by the breed (P < 0.001) and ETEC K88 challenge (P < 0.001). The ETEC K88 challenge significantly increased the colonic percentage of E. coli in the total bacteria population (P < 0.05) of the Landrace pigs. The challenged Landrace pigs exhibited a significantly higher colonic percentage of E. coli in the total bacteria population compared with their Jinhua counterparts. In contrast, the percentage of Lactobacillus in the total bacteria population was only significantly influenced by the breed (P = 0.031), and the percentages in both pig breeds tended to decrease after ETEC K88 challenge. The ratio of Lactobacillus to E. coli in the colon was strongly affected by the breed (P < 0.001), ETEC K88 challenge (P < 0.001), and their interaction (P < 0.001). The ETEC K88 challenge significantly decreased the ratio of Lactobacillus to E. coli in both pig breeds (P < 0.05), and the Jinhua pigs (challenged or not challenged) tended to present higher ratios of Lactobacillus to E. coli compared with Landrace pigs. In the cecum, however, the percentage of E. coli in the total bacteria population was only markedly affected by ETEC K88 challenge (P = 0.004). The ETEC K88 challenge strongly increased the percentage of E. coli in the total bacteria population in both pig breeds (P < 0.05). The percentage of Lactobacillus in the total bacteria population did not differ between all of the treatment groups. The ratio of Lactobacillus to E. coli tended to be affected by ETEC K88 challenge (P = 0.076), and both pig breeds exhibited a decrease in this ratio after ETEC K88 challenge. Intestinal Morphology The intestinal morphological data, including villus heights, crypt depth, and their ratio, in Jinhua and Landrace pigs are presented in Table 5. In the duodenum, the challenged Jinhua and Landrace pigs exhibited greater villous atrophy and crypt depth reduction compared with their nonchallenged counterparts. The villus heights, crypt depths and their ratio, in the challenged Jinhua pigs tended to be higher compared with the challenged Landrace pigs. The ETEC K88 challenge significantly affected the villus height (P = 0.046) and the ratio of villi to crypt (P = 0.016), and the effect of the breed on the villus heights tended to be significant (P = 0.083). In the jejunum, a similar intestinal morphological impairment was observed although the differences between the treatments were not significant. In the ileum, the ETEC K88 challenge caused significant reductions in the villus height in both Jinhua and Landrace pigs (P < 0.05). The crypt depth tended to be greater in the challenged animals, but the reduction in the villus heights in both breeds after the challenge resulted in a significant reduction in the ratio of villi to crypt. The challenged Landrace pigs exhibited significantly lower villus heights and a significantly lower ratio of villi to crypt when compared with their Jinhua counterparts. The breed and ETEC K88 challenge significantly influenced the villus heights (P < 0.001 for breed and P = 0.014 for ETEC challenge) and the ratio of villi to crypt (P = 0.010 for breed and P = 0.003 for ETEC K88 challenge), but their interactions did not affect the intestinal morphology throughout the experiment. Table 5. Effects of enterotoxigenic Escherichia coli (ETEC) K88 challenge on intestinal villus heights, crypt depth, and the ratio of villi to crypts in Jinhua and Landrace pigs   Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  Duodenum      Villus height, µm  242.34a  274.98ab  262.03ab  190.99b  21.99  0.083  0.046*  0.959      Crypt depth, µm  198.66  177.55  194.43  153.06  11.86  0.243  0.578  0.693      Villi:crypt  1.73a  1.55ab  1.37ab  1.25b  0.07  0.214  0.016*  0.784  Jejunum      Villus height, µm  289.05  240.25  253.93  234.45  11.51  0.167  0.389  0.532      Crypt depth, µm  163.17  130.71  163.12  150.52  10.81  0.367  0.686  0.684      Villi:crypt  1.78  1.85  1.63  1.58  0.06  0.956  0.114  0.633  Ileum      Villus height, µm  319.08a  248.17b  283.90bc  203.08c  14.07  <0.001***  0.014*  0.710      Crypt depth, µm  159.11  142.49  169.74  146.86  5.03  0.061  0.431  0.738      Villi:crypt  2.00a  1.75ab  1.68c  1.40c  0.07  0.010*  0.003**  0.868    Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  Duodenum      Villus height, µm  242.34a  274.98ab  262.03ab  190.99b  21.99  0.083  0.046*  0.959      Crypt depth, µm  198.66  177.55  194.43  153.06  11.86  0.243  0.578  0.693      Villi:crypt  1.73a  1.55ab  1.37ab  1.25b  0.07  0.214  0.016*  0.784  Jejunum      Villus height, µm  289.05  240.25  253.93  234.45  11.51  0.167  0.389  0.532      Crypt depth, µm  163.17  130.71  163.12  150.52  10.81  0.367  0.686  0.684      Villi:crypt  1.78  1.85  1.63  1.58  0.06  0.956  0.114  0.633  Ileum      Villus height, µm  319.08a  248.17b  283.90bc  203.08c  14.07  <0.001***  0.014*  0.710      Crypt depth, µm  159.11  142.49  169.74  146.86  5.03  0.061  0.431  0.738      Villi:crypt  2.00a  1.75ab  1.68c  1.40c  0.07  0.010*  0.003**  0.868  a,b,cThe superscript letters indicate a significant difference within a row (P < 0.05). 1JN = nonchallenged Jinhua pigs; LN = nonchallenged Landrace pigs; JC = challenged Jinhua pigs; LC = challenged Landrace pigs. 2B, breed; C, ETEC K88 challenge; B × C, breed and ETEC K88 challenge interaction. *P < 0.05; **P < 0.01; ***P < 0.001. View Large Table 5. Effects of enterotoxigenic Escherichia coli (ETEC) K88 challenge on intestinal villus heights, crypt depth, and the ratio of villi to crypts in Jinhua and Landrace pigs   Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  Duodenum      Villus height, µm  242.34a  274.98ab  262.03ab  190.99b  21.99  0.083  0.046*  0.959      Crypt depth, µm  198.66  177.55  194.43  153.06  11.86  0.243  0.578  0.693      Villi:crypt  1.73a  1.55ab  1.37ab  1.25b  0.07  0.214  0.016*  0.784  Jejunum      Villus height, µm  289.05  240.25  253.93  234.45  11.51  0.167  0.389  0.532      Crypt depth, µm  163.17  130.71  163.12  150.52  10.81  0.367  0.686  0.684      Villi:crypt  1.78  1.85  1.63  1.58  0.06  0.956  0.114  0.633  Ileum      Villus height, µm  319.08a  248.17b  283.90bc  203.08c  14.07  <0.001***  0.014*  0.710      Crypt depth, µm  159.11  142.49  169.74  146.86  5.03  0.061  0.431  0.738      Villi:crypt  2.00a  1.75ab  1.68c  1.40c  0.07  0.010*  0.003**  0.868    Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  Duodenum      Villus height, µm  242.34a  274.98ab  262.03ab  190.99b  21.99  0.083  0.046*  0.959      Crypt depth, µm  198.66  177.55  194.43  153.06  11.86  0.243  0.578  0.693      Villi:crypt  1.73a  1.55ab  1.37ab  1.25b  0.07  0.214  0.016*  0.784  Jejunum      Villus height, µm  289.05  240.25  253.93  234.45  11.51  0.167  0.389  0.532      Crypt depth, µm  163.17  130.71  163.12  150.52  10.81  0.367  0.686  0.684      Villi:crypt  1.78  1.85  1.63  1.58  0.06  0.956  0.114  0.633  Ileum      Villus height, µm  319.08a  248.17b  283.90bc  203.08c  14.07  <0.001***  0.014*  0.710      Crypt depth, µm  159.11  142.49  169.74  146.86  5.03  0.061  0.431  0.738      Villi:crypt  2.00a  1.75ab  1.68c  1.40c  0.07  0.010*  0.003**  0.868  a,b,cThe superscript letters indicate a significant difference within a row (P < 0.05). 1JN = nonchallenged Jinhua pigs; LN = nonchallenged Landrace pigs; JC = challenged Jinhua pigs; LC = challenged Landrace pigs. 2B, breed; C, ETEC K88 challenge; B × C, breed and ETEC K88 challenge interaction. *P < 0.05; **P < 0.01; ***P < 0.001. View Large Analysis of the mRNA Expression of Intestinal Tight Junction Proteins The expression levels of tight junction proteins (claudin-1, occludin, ZO-1, and ZO-2) in different intestinal segments of Jinhua and Landrace pigs are shown in Table 6. The mRNA expression levels of these tight junction proteins in the challenged animals were decreased in all of the intestinal segments. The Jinhua pigs (both challenged and not challenged) exhibited higher expression levels of these genes compared with the Landrace pigs. In the duodenum, the expression of occludin (P = 0.026) and ZO-2 (P = 0.026) was significantly influenced by the breed whereas only ZO-2 was strongly influenced by ETEC K88 challenge (P = 0.009). In the jejunum, the expression of occludin and ZO-2 was significantly affected by the breed (P < 0.001), ETEC K88 challenge (P = 0.002 for occludin and P = 0.003 for ZO-2), and their interaction (P = 0.006 for occludin and P = 0.027 for ZO-2). The expression of ZO-1 was only significantly affected by the breed (P = 0.02) and ETEC K88 challenge (P = 0.028). The decreased expression of the tight junction protein mRNAs was more pronounced in the challenged Jinhua pigs compared with their Landrace counterparts. However, the challenged Jinhua pigs still exhibited a significantly higher expression level of occludin compared with the challenged Landrace pigs. In the ileum, the breed had significant effects on claudin-1 (P < 0.001) and ZO-1 (P = 0.023) expression whereas the ETEC K88 challenge strongly affected the expression of claudin-1 (P = 0.005), ZO-1 (P = 0.006), and ZO-2 (P = 0.043). Only the ileal expression of claudin-1 was significantly affected by the interaction between breed and ETEC K88 challenge (P = 0.008). Table 6. Effects of enterotoxigenic Escherichia coli (ETEC) K88 challenge on the mRNA expression of intestinal tight junction proteins (claudin-1, occludin, ZO-1, and ZO-2) in Jinhua and Landrace pigs   Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  Duodenum      claudin-1  1.91  1.07  1.36  1.04  0.17  0.108  0.397  0.444      occludin  2.28a  1.00b  1.20ab  0.64b  0.23  0.026*  0.065  0.323      ZO-1  1.63  1.01  0.94  0.77  0.15  0.162  0.107  0.403      ZO-2  1.69a  1.01b  0.91b  0.76b  0.12  0.023*  0.009**  0.111  Jejunum      claudin-1  5.61  1.19  2.63  0.96  0.94  0.127  0.394  0.459      occludin  6.12a  1.07c  2.95b  0.73d  0.66  <0.001***  0.002**  0.006**      ZO-1  2.74a  1.10b  1.18b  0.62b  0.29  0.02*  0.028*  0.184      ZO-2  6.31a  1.58b  2.65b  0.83b  0.59  <0.001***  0.003**  0.027*  Ileum      claudin-1  2.63a  1.01b  0.56b  0.49b  0.27  <0.001***  0.005**  0.008**      occludin  1.64  1.05  0.59  0.54  0.21  0.415  0.065  0.485      ZO-1  2.18a  1.09b  0.89b  0.74b  0.18  0.023*  0.006**  0.062      ZO-2  1.59a  1.09ab  0.65b  0.76ab  0.15  0.477  0.043*  0.286    Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  Duodenum      claudin-1  1.91  1.07  1.36  1.04  0.17  0.108  0.397  0.444      occludin  2.28a  1.00b  1.20ab  0.64b  0.23  0.026*  0.065  0.323      ZO-1  1.63  1.01  0.94  0.77  0.15  0.162  0.107  0.403      ZO-2  1.69a  1.01b  0.91b  0.76b  0.12  0.023*  0.009**  0.111  Jejunum      claudin-1  5.61  1.19  2.63  0.96  0.94  0.127  0.394  0.459      occludin  6.12a  1.07c  2.95b  0.73d  0.66  <0.001***  0.002**  0.006**      ZO-1  2.74a  1.10b  1.18b  0.62b  0.29  0.02*  0.028*  0.184      ZO-2  6.31a  1.58b  2.65b  0.83b  0.59  <0.001***  0.003**  0.027*  Ileum      claudin-1  2.63a  1.01b  0.56b  0.49b  0.27  <0.001***  0.005**  0.008**      occludin  1.64  1.05  0.59  0.54  0.21  0.415  0.065  0.485      ZO-1  2.18a  1.09b  0.89b  0.74b  0.18  0.023*  0.006**  0.062      ZO-2  1.59a  1.09ab  0.65b  0.76ab  0.15  0.477  0.043*  0.286  a,b,cThe superscript letters indicate a significant difference within a row (P < 0.05). 1JN = nonchallenged Jinhua pigs; LN = nonchallenged Landrace pigs; JC = challenged Jinhua pigs; LC = challenged Landrace pigs. 2B, breed; C, ETEC K88 challenge; B × C, breed and ETEC K88 challenge interaction. *P < 0.05; **P < 0.01; ***P < 0.001. View Large Table 6. Effects of enterotoxigenic Escherichia coli (ETEC) K88 challenge on the mRNA expression of intestinal tight junction proteins (claudin-1, occludin, ZO-1, and ZO-2) in Jinhua and Landrace pigs   Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  Duodenum      claudin-1  1.91  1.07  1.36  1.04  0.17  0.108  0.397  0.444      occludin  2.28a  1.00b  1.20ab  0.64b  0.23  0.026*  0.065  0.323      ZO-1  1.63  1.01  0.94  0.77  0.15  0.162  0.107  0.403      ZO-2  1.69a  1.01b  0.91b  0.76b  0.12  0.023*  0.009**  0.111  Jejunum      claudin-1  5.61  1.19  2.63  0.96  0.94  0.127  0.394  0.459      occludin  6.12a  1.07c  2.95b  0.73d  0.66  <0.001***  0.002**  0.006**      ZO-1  2.74a  1.10b  1.18b  0.62b  0.29  0.02*  0.028*  0.184      ZO-2  6.31a  1.58b  2.65b  0.83b  0.59  <0.001***  0.003**  0.027*  Ileum      claudin-1  2.63a  1.01b  0.56b  0.49b  0.27  <0.001***  0.005**  0.008**      occludin  1.64  1.05  0.59  0.54  0.21  0.415  0.065  0.485      ZO-1  2.18a  1.09b  0.89b  0.74b  0.18  0.023*  0.006**  0.062      ZO-2  1.59a  1.09ab  0.65b  0.76ab  0.15  0.477  0.043*  0.286    Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items  JN  LN  JC  LC  SEM  B  C  B × C  Duodenum      claudin-1  1.91  1.07  1.36  1.04  0.17  0.108  0.397  0.444      occludin  2.28a  1.00b  1.20ab  0.64b  0.23  0.026*  0.065  0.323      ZO-1  1.63  1.01  0.94  0.77  0.15  0.162  0.107  0.403      ZO-2  1.69a  1.01b  0.91b  0.76b  0.12  0.023*  0.009**  0.111  Jejunum      claudin-1  5.61  1.19  2.63  0.96  0.94  0.127  0.394  0.459      occludin  6.12a  1.07c  2.95b  0.73d  0.66  <0.001***  0.002**  0.006**      ZO-1  2.74a  1.10b  1.18b  0.62b  0.29  0.02*  0.028*  0.184      ZO-2  6.31a  1.58b  2.65b  0.83b  0.59  <0.001***  0.003**  0.027*  Ileum      claudin-1  2.63a  1.01b  0.56b  0.49b  0.27  <0.001***  0.005**  0.008**      occludin  1.64  1.05  0.59  0.54  0.21  0.415  0.065  0.485      ZO-1  2.18a  1.09b  0.89b  0.74b  0.18  0.023*  0.006**  0.062      ZO-2  1.59a  1.09ab  0.65b  0.76ab  0.15  0.477  0.043*  0.286  a,b,cThe superscript letters indicate a significant difference within a row (P < 0.05). 1JN = nonchallenged Jinhua pigs; LN = nonchallenged Landrace pigs; JC = challenged Jinhua pigs; LC = challenged Landrace pigs. 2B, breed; C, ETEC K88 challenge; B × C, breed and ETEC K88 challenge interaction. *P < 0.05; **P < 0.01; ***P < 0.001. View Large Cytokine and Secretory Immunoglobulin A Concentration in the Ileum The effect of the ETEC K88 challenge on the concentrations of ileal cytokines and sIgA in Jinhua and Landrace pigs is presented in Table 7. The ETEC K88 challenge significantly increased the levels of proinflammatory cytokines IFN-γ (P < 0.05), TNF-α (P < 0.05), and IL-6 (P < 0.05) in Jinhua pigs and the levels of IFN-γ (P < 0.05) and TNF-α (P < 0.05) in Landrace pigs. In addition, a higher level of IFN-γ (P < 0.05) was observed in challenged Jinhua pigs compared with their Landrace counterparts. The levels of anti-inflammatory cytokines IL-4 and TGF-β were reduced (P < 0.05) in challenged pigs in both breeds, but only the level of TGF-β was significantly higher in the challenged Jinhua pigs compared with challenged Landrace pigs. The level of sIgA significantly increased (P < 0.05) in the challenged animals, but no significant difference was found between the 2 breeds with or without challenge. The breed significantly affected the proinflammatory cytokines IFN-γ (P < 0.001) and IL-6 (P = 0.015) and the anti-inflammatory cytokines IL-4 (P = 0.001) and TGF-β (P = 0.001) whereas the ETEC K88 challenge significantly affected all of the cytokines and the sIgA concentrations analyzed in this study. Only TNF-α (P = 0.027) and IL-4 (P = 0.020) were significantly influenced by the interaction between breed and ETEC K88 challenge. Table 7. Effects of enterotoxigenic Escherichia coli (ETEC) K88 challenge on the ileal concentrations of cytokines and secretory immunoglobulin A (sIgA) in Jinhua and Landrace pigs   Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items3  JN  LN  JC  LC  SEM  B  C  B × C  IFN-γ  68.77c  40.39d  120.44a  85.34b  7.59  <0.001***  <0.001***  0.259  TNF-α  96.75b  87.03c  107.46a  112.77a  2.89  0.473  <0.001***  0.027*  IL-6  98.28b  105.32a  107.24a  110.37a  1.40  0.015*  0.002**  0.300  IL-4  81.80b  92.06a  67.58c  69.66c  2.38  0.001**  <0.001***  0.020*  TGF-β  462.81a  373.448b  345.57b  257.72c  15.98  <0.001***  <0.001***  0.942  sIgA  197.84b  196.16b  220.45a  210.59a  3.37  0.153  0.001**  0.296    Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items3  JN  LN  JC  LC  SEM  B  C  B × C  IFN-γ  68.77c  40.39d  120.44a  85.34b  7.59  <0.001***  <0.001***  0.259  TNF-α  96.75b  87.03c  107.46a  112.77a  2.89  0.473  <0.001***  0.027*  IL-6  98.28b  105.32a  107.24a  110.37a  1.40  0.015*  0.002**  0.300  IL-4  81.80b  92.06a  67.58c  69.66c  2.38  0.001**  <0.001***  0.020*  TGF-β  462.81a  373.448b  345.57b  257.72c  15.98  <0.001***  <0.001***  0.942  sIgA  197.84b  196.16b  220.45a  210.59a  3.37  0.153  0.001**  0.296  a,b,cThe superscript letters indicate a significant difference within a row (P < 0.05). 1JN = nonchallenged Jinhua pigs; LN = nonchallenged Landrace pigs; JC = challenged Jinhua pigs; LC = challenged Landrace pigs. 2B, breed; C, ETEC K88 challenge; B × C, breed and ETEC K88 challenge interaction. 3IFN-γ = interferon gamma; TNF-α = tumor necrosis factor alpha; TGF-β = transforming growth factor beta. *P < 0.05; **P < 0.01; ***P < 0.001. View Large Table 7. Effects of enterotoxigenic Escherichia coli (ETEC) K88 challenge on the ileal concentrations of cytokines and secretory immunoglobulin A (sIgA) in Jinhua and Landrace pigs   Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items3  JN  LN  JC  LC  SEM  B  C  B × C  IFN-γ  68.77c  40.39d  120.44a  85.34b  7.59  <0.001***  <0.001***  0.259  TNF-α  96.75b  87.03c  107.46a  112.77a  2.89  0.473  <0.001***  0.027*  IL-6  98.28b  105.32a  107.24a  110.37a  1.40  0.015*  0.002**  0.300  IL-4  81.80b  92.06a  67.58c  69.66c  2.38  0.001**  <0.001***  0.020*  TGF-β  462.81a  373.448b  345.57b  257.72c  15.98  <0.001***  <0.001***  0.942  sIgA  197.84b  196.16b  220.45a  210.59a  3.37  0.153  0.001**  0.296    Treatment groups1      Nonchallenged  ETEC K88 challenged    P-value2  Items3  JN  LN  JC  LC  SEM  B  C  B × C  IFN-γ  68.77c  40.39d  120.44a  85.34b  7.59  <0.001***  <0.001***  0.259  TNF-α  96.75b  87.03c  107.46a  112.77a  2.89  0.473  <0.001***  0.027*  IL-6  98.28b  105.32a  107.24a  110.37a  1.40  0.015*  0.002**  0.300  IL-4  81.80b  92.06a  67.58c  69.66c  2.38  0.001**  <0.001***  0.020*  TGF-β  462.81a  373.448b  345.57b  257.72c  15.98  <0.001***  <0.001***  0.942  sIgA  197.84b  196.16b  220.45a  210.59a  3.37  0.153  0.001**  0.296  a,b,cThe superscript letters indicate a significant difference within a row (P < 0.05). 1JN = nonchallenged Jinhua pigs; LN = nonchallenged Landrace pigs; JC = challenged Jinhua pigs; LC = challenged Landrace pigs. 2B, breed; C, ETEC K88 challenge; B × C, breed and ETEC K88 challenge interaction. 3IFN-γ = interferon gamma; TNF-α = tumor necrosis factor alpha; TGF-β = transforming growth factor beta. *P < 0.05; **P < 0.01; ***P < 0.001. View Large DISCUSSION In the present study, we compared the gastrointestinal physiological responses to ETEC K88 challenge in Jinhua and Landrace piglets. In fact, the 2 pig breeds exhibited various differences in their baseline intestinal physiology. Under the current experimental conditions, the breed had marked effects on the incidence of diarrhea, the colonic microbial populations, the intestinal expressions of tight junction proteins, and the cytokines concentrations in response to ETEC K88 challenge but no effects on the intestinal morphology. Only a few studies have compared the gastrointestinal physiology or functions between different pig breeds. Meishan pigs were found to be less susceptible to ETEC strains bearing the colonization factor K88, 987P, F41, or F41 plus K99 compared with European Large White pigs (Duchetsuchaux et al., 1991). Moreover, the small intestinal nutrient transport and barrier functions in Yorkshire and Meishan gilts were found to respond differently to lipopolysaccharides (LPS) exposure (Albin et al., 2007). To the best of our knowledge, this study provides the first comparison of the intestinal physiology between Chinese Jinhua pigs and European Landrace Pigs using an ETEC K88 challenge model. These findings may help to explain how pigs of different breeds respond differently to pathogenic challenge by providing detailed information on the differences in the microbial populations, intestinal morphology, the expression levels of tight junction proteins, and mucosal immunity between Jinhua and Landrace pigs. Porcine E. coli challenge models have been widely used over the past decade to study various aspects of growth (Wellock et al., 2008b), nutrition (Bhandari et al., 2008; Yi et al., 2005; Lessard et al., 2009), and immunity (Bosi et al., 2004; Wellock et al., 2008a). Pigs susceptible to E. coli have shown retarded growth and greater incidence of postweaning diarrhea. In the present study, the breed was the main factor responsible for the differences in ADG and ADFI. The Jinhua pigs presented significantly higher ADG and ADFI compared with the nonchallenged and challenged Landrace pigs. However, the interaction between breed and ETEC K88 challenge had no effect on growth performance. One possible reason for this finding may be that the short experimental period of this study diminished the effect of the ETEC K88 challenge on growth performance. However, inoculation of ETEC K88 markedly affected the incidence of diarrhea in both pig breeds. We observed prompt diarrhea the day after the challenge, and this finding is in agreement with the results reported by Madec et al. (2000). The effect of the ETEC K88 challenge on the incidence of diarrhea was markedly greater in the Landrace pigs (the interaction between breed and ETEC K88 challenge; P = 0.006), and higher percentages of the Landrace pigs were diarrhea positive with challenge, which indicates that the Landrace pigs are more susceptible to ETEC K88 challenge. The commensal microbiota of the GIT contributes to the maintenance of mammalian host health and performance. In pigs, the composition and diversity of gut microbes are influenced by several factors, including weaning transition (Konstantinov et al., 2006), diet composition (Castillo et al., 2008; Hojberg et al., 2005), environment (Pluske et al., 2007), and pathogen infection (Konstantinov et al., 2008; Price et al., 2010). Pigs infected with Salmonella exhibited a shift in the composition of the fecal microbial community, but supplementation with a yeast fermentation product altered the composition by increasing the populations of Bacteroidetes and Lactobacillus (Price et al., 2010). Similar results were obtained in a study that used an F4 (K88)-positive E. coli challenge pig model: the administration of a probiotic at a safe threshold level increased the number of Lactobacillus and Bifidobacterium and reduced coliform shedding (Li et al., 2012). Pigs of different breeds, for example, Meishan and Landrace, also tend to present different predominant bacterial divisions in the distal gut under normal conditions (Guo et al., 2008). The oral inoculation of pathogenic strains of E. coli into a pig model of PWC resulted in the transient and immediate appearance of diarrhea and E. coli shedding in the feces (Madec et al., 2000). In the present study, we observed markedly increases in the E. coli populations and in the percentage of E. coli in the total bacterial populations in both pig breeds with ETEC K88 challenge. This result may help explain the higher incidence of diarrhea and the retarded growth performance observed in the challenged pigs. Although the current measurement did not discriminate the population of ETEC K88 from the entire E. coli strains, our data were consistent with the previous notion that ETEC are able to proliferate to large numbers in the gut before initiating diarrhea (Madec et al., 2000). Furthermore, the Landrace pigs exhibited a higher E. coli population and a higher percentage of E. coli in the total bacteria population compared with the Jinhua pigs with or without challenge, which suggests that the Landrace pig are most likely predisposed to shed more E. coli in the GIT. An increase in the ratio of Lactobacillus to Enterobacteria is considered beneficial for gut health (Castillo et al., 2008), and the inhibition of Enterobacteria may prevent or decrease the severity of diarrhea that appears after weaning (Melin et al., 2004). In our study, ETEC K88 challenge resulted in reduced percentage of Lactobacillus in the total bacteria population and the ratio of Lactobacillus to E. coli, which indicates that the challenged pigs of both breeds experienced a loss of normal gut physiology although the Landrace pigs presented a higher extent of intestinal damage compared with the Jinhua pigs. The effects of the breed, ETEC K88 challenge, and their interaction on the selected microbial populations were segment dependent. In the cecum, the selected microbial populations were less responsive to the breed and ETEC K88 challenge. We speculated that this finding can be attributed to the short experimental period used in the present study, which was less able to cause an apparent disturbance of the microbial community. The integrity of intestinal morphological structures is crucial for the maintenance of normal intestinal functions. The intestinal morphology can be affected by periweaning failure to thrive syndrome (Moeser et al., 2012), pathogenic infection (Price et al., 2010), and environmental stressors (Zhao et al., 2007). In the present study, the ETEC K88 challenge tended to decrease the villus height and crypt depth in all of the intestinal segments, and this finding is in agreement with a former study that used an ETEC K88 challenge pig model (Yi et al., 2005). However, the interaction between breed and ETEC K88 challenge had no effect on the intestinal morphology, which indicates that the 2 breeds responded similarly to the ETEC K88 challenge. We speculated that this result may be also attributed to the short experimental period and further morphological changes may occur over a more extended period. There were only marked breed differences in the villus heights and the ratio of villus to crypt in the ileum, which supports the hypothesis that Jinhua pigs may normally have a stronger intestinal structure. The effects of the ETEC K88 challenge on the intestinal morphologic structures were segment dependent, which could be attributed to the preference of ETEC bacteria to colonize the middle and distal portions of the small intestine (Nabuurs, 1998). It should be noted that the nonchallenged Landrace pigs exhibited significantly shorter villi compared with the Jinhua pigs. These shorter villi, plus the deeper crypts observed in Landrace pigs, reflect the presence of fewer absorptive and more secretory cells in the intestine, which can result in decreased absorption but increased secretion. Therefore, the higher amount of unabsorbed dietary material flowing to the hind gut would act as a substrate for ETEC and encourage bacterial proliferation (Pluske et al., 1997). Taken together, the observed changes in the intestinal morphology and gut microbial populations in response to ETEC K88 challenge may act concomitantly to increase the incidence of diarrhea and the susceptibility to ETEC K88 in Landrace pigs. The intestinal epithelia, particularly the internal tight junctions between enterocytes, also play important roles in the maintenance of intestinal permeability. The permeability of the intestinal epithelium can be increased by enteric pathogens and endotoxin translocation through the alteration of tight junctions (Moreto and Perez-Bosque, 2009). Infection with ETEC can negatively alter the tight junction protein expression in weanling piglets (Ewaschuk et al., 2011). Similarly, our findings indicated that challenge with ETEC K88 decreased the mRNA expression of tight junction proteins (claudin-1, occludin, ZO-1, and ZO-2) in all of the intestinal segments in both pig breeds. One speculated reason is that ETEC can significantly increase the paracellular permeability of the small intestine through either actomyosin ring alteration or occludin dephosphorylation and ZO-1 redistribution (Berkes et al., 2003). In addition, the increased permeability caused by impaired intestinal tight junction has been associated with increased incidence of secretory diarrhea (Moreto and Perez-Bosque, 2009); therefore, the reduced expression levels of tight junction proteins is also in agreement with the increased incidence of diarrhea in the challenged pigs observed in the present study. Moreover, the ETEC K88 challenge markedly affected the expression of occludin, ZO-1, and ZO-2 in the jejunum and claudin-1, ZO-1, and ZO-2 in the ileum, which indicates that the reduction in the expression of intestinal tight junction proteins in response to ETEC K88 challenge is also site dependent. This finding could be attributed to the gradual increase in gut-associated lymphatic tissues from the middle jejunum to the distal ileum, where the antigen uptake, immune stimulation, and the intestinal barrier disturbance can be further provoked by pathogen invasion. Different pig breeds may exhibit different intestinal barrier functions. For example, Meishan and Yorkshire pigs exhibit different barrier responses after LPS infusion (Albin et al., 2007). In the present study, the diminished mRNA expression of tight junction proteins was associated with ETEC K88 challenge in both breeds. The interaction between breed and ETEC K88 challenge significantly affected the levels of jejunal occludin and ZO-2 and ileal claudin-1, which indicates that the 2 pig breeds respond differently to ETEC K88 challenge. However, because the baseline mRNA levels of tight junction proteins were higher in Jinhua pigs and even higher with ETEC K88 challenge, Jinhua pigs may possess improved barrier functions that make them better able to maintain their intestinal physiology. Cytokines play crucial roles in the modulation of the inflammatory response in the GIT. Numerous proinflammatory cytokines (e.g., TNF-α, IFN-γ, and IL-6) are essential in the mediation of the inflammatory response caused by pathogen infection (Lippolis, 2008). Our results showed significantly increased levels of proinflammatory cytokines and sIgA with ETEC K88 challenge in both pig breeds, which suggests that the intestinal mucosal immune system becomes activated in the presence of ETEC and that inflammation occurred (Brandtzaeg, 2010). Anti-inflammatory cytokines (e.g., IL-4, IL-10, and TGF-β) are important for the attenuation and/or containment of the inflammatory process through the inhibition of the production of proinflammatory cytokines. However, a decreased production of IL-4 and TGF-β was observed in response to ETEC K88 challenge. This result is most likely due to the synergistic effects of cytokines on the inflammatory pathways and processes; that is, higher levels of proinflammatory cytokines may inhibit the production of anti-inflammatory cytokines. The interaction between breed and ETEC K88 challenge exerted an effect on cytokine production, which showed that the influence of ETEC K88 challenge on the production of TNF-α and IL-4 was greater in the challenged Landrace pigs compared with their Jinhua counterparts. These results suggests that the immune stress caused by ETEC K88 challenge is more pronounced in the Landrace pigs, which further supports the hypothesis that Landrace pigs were more susceptible to ETEC K88 challenge. Furthermore, the ETEC K88 challenge showed marked effects of on all of the cytokines and sIgA. This finding suggests that mucosal immunity was actively induced by ETEC K88 infection. The activation of the mucosal immune system results in the prevention of ETEC-induced diarrhea (Snoeck et al., 2003). Therefore, the reduction in the incidence of diarrhea may be attributed to the induction of cytokines. In addition, the activation of the mucosal immune system may exhibit a feedback modulation of the intestinal barrier functions and gut microbiota. Proinflammatory cytokines, such as TNF-α and IFN-γ, are known to induce the endocytosis of tight junction proteins and subsequently cause increased permeability (Capaldo and Nusrat, 2009). Secretory immunoglobulin A is produced by the plasma cells residing in the intestinal lamina propria, which fulfill the function of controlling the number of microbiota in the GIT (Sekirov et al., 2010). Therefore, the increased production of cytokines and sIgA with ETEC K88 challenge may in turn lead to the induction of intestinal permeation, which would allow the further entry of antigens, toxins, and pathogens into the circulatory system, and eventually result in triggering the consequent inflammatory cascades (Kim et al., 2012) and preventing the overgrowth of gut microbes. In summary, several physiological parameters, including growth performance, colonic microbial population, ileal morphology, expression of intestinal tight junction proteins, and cytokine production, were different between Jinhua and Landrace pigs without ETEC K88 challenge in the present study. However, the ETEC K88 challenge caused distinctive responses between the 2 breeds. In response to ETEC K88 challenge, the Jinhua pigs appeared to exhibit better growth performance, a lower incidence of diarrhea, and a lower extent of immune activation compared with the Landrace pigs. The influences of the challenge on the microbial populations and the mRNA expression of tight junction proteins were stronger in Jinhua pigs. However, the higher levels of Lactobacillus, the higher percentage of Lactobacillus in the total bacteria population, the higher ratio of Lactobacillus to E. coli, and the higher mRNA levels of tight junction proteins in the Jinhua pigs with and without challenge suggest that these pigs may possess a better ability to maintain the intestinal physiology homeostasis. However, further studies should address more underlying mechanisms because there are many other important aspects of the intestinal physiology and health that may also be affected by ETEC K88 challenge. Nevertheless, our findings may help the recognition of the excellent characteristics of Jinhua pigs. LITERATURE CITED Albin D. M. Wubben J. E. Rowlett J. M. Tappenden K. A. Nowak R. A. 2007. Changes in small intestinal nutrient transport and barrier function after lipopolysaccharide exposure in two pig breeds. J. Anim. Sci.  85: 2517– 2523. Google Scholar CrossRef Search ADS PubMed  Berkes J. Viswanathan V. K. Savkovic S. D. Hecht G. 2003. Intestinal epithelial responses to enteric pathogens: Effects on the tight junction barrier, ion transport, and inflammation. Gut  52: 439– 451. Google Scholar CrossRef Search ADS PubMed  Bhandari S. K. Xu B. Nyachoti C. M. Giesting D. W. Krause D. O. 2008. Evaluation of alternatives to antibiotics using an Escherichia coli K88(+) model of piglet diarrhea: Effects on gut microbial ecology. J. Anim. Sci.  86: 836– 847. Google Scholar CrossRef Search ADS PubMed  Bosi P. Casini L. Finamore A. Cremokolini C. Merialdi G. Trevisi P. Nobili F. Mengheri E. 2004. Spray-dried plasma improves growth performance and reduces inflammatory status of weaned pigs challenged with enterotoxigenic Escherichia coli K88. J. Anim. Sci.  82: 1764– 1772. Google Scholar CrossRef Search ADS PubMed  Brandtzaeg P 2010. Homeostatic impact of indigenous microbiota and secretory immunity. Benef. Microbes  1: 211– 227. Google Scholar CrossRef Search ADS PubMed  Capaldo C. T. Nusrat A. 2009. Cytokine regulation of tight junctions. Biochim. Biophys. Acta  1788: 864– 871. Google Scholar CrossRef Search ADS PubMed  Castillo M. Martin-Orue S. M. Taylor-Pickard J. A. Perez J. F. Gasa J. 2008. Use of mannan-oligosaccharides and zinc chelate as growth promoters and diarrhea preventative in weaning pigs: Effects on microbiota and gut function. J. Anim. Sci.  86: 94– 101. Google Scholar CrossRef Search ADS PubMed  Duchetsuchaux M. F. Bertin A. M. Menanteau P. S. 1991. Susceptibility of Chinese Meishan and European large white-pigs to enterotoxigenic Escherichia coli strains bearing colonization factor K88, 987P, K99, or F41. Am. J. Vet. Res.  52: 40– 44. Google Scholar PubMed  Ewaschuk J. B. Murdoch G. K. Johnson I. R. Madsen K. L. Field C. J. 2011. Glutamine supplementation improves intestinal barrier function in a weaned piglet model of Escherichia coli infection. Br. J. Nutr.  106: 870– 877. Google Scholar CrossRef Search ADS PubMed  Furet J. P. Firmesse O. Gourmelon M. Bridonneau C. Tap J. Mondot S. Dore J. Corthier G. 2009. Comparative assessment of human and farm animal faecal microbiota using real-time quantitative PCR. FEMS Microbiol. Ecol.  68: 351– 362. Google Scholar CrossRef Search ADS PubMed  Guo X. L. Xia X. J. Tang R. Y. Wang K. N. 2008. Real-time PCR quantification of the predominant bacterial divisions in the distal gut of Meishan and Landrace pigs. Anaerobe  14: 224– 228. Google Scholar CrossRef Search ADS PubMed  Halas D. Hansen C. F. Hampson D. J. Mullan B. P. Kim J. C. Wilson R. H. Pluske J. R. 2010. Dietary supplementation with benzoic acid improves apparent ileal digestibility of total nitrogen and increases villous height and caecal microbial diversity in weaner pigs. Anim. Feed Sci. Technol.  160: 137– 147. Google Scholar CrossRef Search ADS   Hojberg O. Canibe N. Poulsen H. D. Hedemann M. S. Jensen B. B. 2005. Influence of dietary zinc oxide and copper sulfate on the gastrointestinal ecosystem in newly weaned piglets. Appl. Environ. Microbiol.  71: 2267– 2277. Google Scholar CrossRef Search ADS PubMed  Kim J. C. Hansen C. F. Mullan B. P. Pluske J. R. 2012. Nutrition and pathology of weaner pigs: Nutritional strategies to support barrier function in the gastrointestinal tract. Anim. Feed Sci. Technol.  173: 3– 16. Google Scholar CrossRef Search ADS   Konstantinov S. R. Awati A. A. Williams B. A. Miller B. G. Jones P. Stokes C. R. Akkermans A. D. L. Smidt H. De Vos W. M. 2006. Post-natal development of the porcine microbiota composition and activities. Environ. Microbiol.  8: 1191– 1199. Google Scholar CrossRef Search ADS PubMed  Konstantinov S. R. Smidt H. Akkermans A. D. L. Casini L. Trevisi P. Mazzoni M. De Filippi S. Bosi P. de Vos W. M. 2008. Feeding of Lactobacillus sobrius reduces Escherichia coli F4 levels in the gut and promotes growth of infected piglets. FEMS Microbiol. Ecol.  66: 599– 607. Google Scholar CrossRef Search ADS PubMed  Lambert G. P 2009. Stress-induced gastrointestinal barrier dysfunction and its inflammatory effects. J. Anim. Sci.  87: E101– E108. Google Scholar CrossRef Search ADS PubMed  Layton A. McKay L. Williams D. Garrett V. Gentry R. Sayler G. 2006. Development of bacteroides 16S rRNA gene TaqMan-based real-time PCR assays for estimation of total, human, and bovine fecal pollution in water. Appl. Environ. Microbiol.  72: 4214– 4224. Google Scholar CrossRef Search ADS PubMed  Lessard M. Dupuis M. Gagnon N. Nadeau E. Matte J. J. Goulet J. Fairbrother J. M. 2009. Administration of Pediococcus acidilactici or Saccharomyces cerevisiae boulardii modulates development of porcine mucosal immunity and reduces intestinal bacterial translocation after Escherichia coli challenge. J. Anim. Sci.  87: 922– 934. Google Scholar CrossRef Search ADS PubMed  Li X. Q. Zhu Y. H. Zhang H. F. Yue Y. Cai Z. X. Lu Q. P. Zhang L. Weng X. G. Zhang F. Zhou D. Yang J. C. Wang J. F. 2012. Risks associated with high-dose Lactobacillus rhamnosus in an Escherichia coli model of piglet diarrhoea: Intestinal microbiota and immune imbalances. PLoS ONE  7( 7): e40666. Google Scholar CrossRef Search ADS PubMed  Lippolis J. D 2008. Immunological signaling networks: Integrating the body's immune response. J. Anim. Sci.  86: E53– E63. Google Scholar CrossRef Search ADS PubMed  Livak K. J. Schmittgen T. D. 2001. Analysis of relative gene expression data using real-time quantitative PCR and the 2–ΔΔCT method. Methods  25: 402– 408. Google Scholar CrossRef Search ADS PubMed  Madec F. Bridoux N. Bounaix S. Cariolet R. Duval-Iflah Y. Hampson D. J. Jestin A. 2000. Experimental models of porcine post-weaning colibacillosis and their relationship to post-weaning diarrhoea and digestive disorders as encountered in the field. Vet. Microbiol.  72: 295– 310. Google Scholar CrossRef Search ADS PubMed  Melin L. Mattson S. Katouli M. Wallgren P. 2004. Development of post-weaning diarrhoea in piglets. Relation to presence of Escherichia coli strains and rotavirus. J. Vet. Med. B Infect. Dis. Vet. Public Health  51: 12– 22. Google Scholar CrossRef Search ADS PubMed  Moeser A. J. Borst L. B. Overman B. L. Pittman J. S. 2012. Defects in small intestinal epithelial barrier function and morphology associated with peri-weaning failure to thrive syndrome (PFTS) in swine. Res. Vet. Sci.  93: 975– 982. Google Scholar CrossRef Search ADS PubMed  Moreto M. Perez-Bosque A. 2009. Dietary plasma proteins, the intestinal immune system, and the barrier functions of the intestinal mucosa. J. Anim. Sci.  87: E92– E100. Google Scholar CrossRef Search ADS PubMed  Nabuurs M. J. A 1998. Weaning piglets as a model for studying pathophysiology of diarrhea. Vet. Q.  20( Suppl. 3): 42– 45. Google Scholar CrossRef Search ADS PubMed  Nataro J. P. Kaper J. B. 1998. Diarrheagenic Escherichia coli. Clin. Microbiol. Rev.  11: 142] Google Scholar PubMed  Petersson A. Domig K. J. Schedle K. Windisch W. Kneifel W. 2010. Comparison of three methods to enumerate gut microbiota of weanling piglets fed insoluble dietary fibre differing in lignin content. J. Agric. Sci.  148: 225– 232. Google Scholar CrossRef Search ADS   Pluske J. R. Durmic Z. Payne H. G. Mansfield J. Mullan B. P. Hampson D. J. Vercoe P. E. 2007. Microbial diversity in the large intestine of pigs born and reared in different environments. Livest. Sci.  108: 113– 116. Google Scholar CrossRef Search ADS   Pluske J. R. Hampson D. J. Williams I. H. 1997. Factors influencing the structure and function of the small intestine in the weaned pig: A review. Livest. Prod. Sci.  51: 215– 236. Google Scholar CrossRef Search ADS   Price K. L. Totty H. R. Lee H. B. Utt M. D. Fitzner G. E. Yoon I. Ponder M. A. Escobar J. 2010. Use of Saccharomyces cerevisiae fermentation product on growth performance and microbiota of weaned pigs during Salmonella infection. J. Anim. Sci.  88: 3896– 3908. Google Scholar CrossRef Search ADS PubMed  Qi H 2011. Effects of different protein sources on intestinal micro flora and intestinal health of weaned piglets. PhD diss., Sichuan Agricultural University, Ya'an, China. Qi H. Xiang Z. Han G. Yu B. Huang Z. Chen D. 2011. Effects of different dietary protein sources on cecal microflora in rats. Afr. J. Biotechnol.  10: 3704– 3708. Sekirov I. Russell S. L. Antunes L. C. M. Finlay B. B. 2010. Gut microbiota in health and disease. Physiol. Rev.  90: 859– 904. Google Scholar CrossRef Search ADS PubMed  Shan T. Wu T. Reng Y. Wang Y. 2009. Breed difference and regulation of the porcine adipose triglyceride lipase and hormone sensitive lipase by TNF alpha. Anim. Genet.  40: 863– 870. Google Scholar CrossRef Search ADS PubMed  Snoeck V. Huyghebaert N. Cox E. Vermeire A. Vancaeneghem S. Remon J. P. Goddeeris B. M. 2003. Enteric-coated pellets of F4 fimbriae for oral vaccination of suckling piglets against enterotoxigenic Escherichia coli infections. Vet. Immunol. Immunopathol.  96: 219– 227. Google Scholar CrossRef Search ADS PubMed  Sutherland M. A. Niekamp S. R. Rodriguez-Zas S. L. Salak-Johnson J. L. 2006. Impacts of chronic stress and social status on various physiological and performance measures in pigs of different breeds. J. Anim. Sci.  84: 588– 596. Google Scholar CrossRef Search ADS PubMed  Suzuki M. T. Taylor L. T. DeLong E. F. 2000. Quantitative analysis of small-subunit rRNA genes in mixed microbial populations via 5′-nuclease assays. Appl. Environ. Microbiol.  66: 4605– 4614. Google Scholar CrossRef Search ADS PubMed  Wellock I. J. Fortomaris P. D. Houdijk J. G. Kyriazakis I. 2008a. Effects of dietary protein supply, weaning age and experimental enterotoxigenic Escherichia coli infection on newly weaned pigs: Health. Animal  2: 834– 842. Wellock I. J. Fortomaris P. D. Houdijk J. G. Kyriazakis I. 2008b. Effects of dietary protein supply, weaning age and experimental enterotoxigenic Escherichia coli infection on newly weaned pigs: Performance. Animal  2: 825– 833. Yi G. F. Carroll J. A. Allee G. L. Gaines A. M. Kendall D. C. Usry J. L. Toride Y. Izuru S. 2005. Effect of glutamine and spray-dried plasma on growth performance, small intestinal morphology, and immune responses of Escherichia coli K88(+)-challenged weaned pigs. J. Anim. Sci.  83: 634– 643. Google Scholar CrossRef Search ADS PubMed  Zhao J. Harper A. F. Estienne M. J. Webb K. E.Jr McElroy A. P. Denbow D. M. 2007. Growth performance and intestinal morphology responses in early weaned pigs to supplementation of antibiotic-free diets with an organic copper complex and spray-dried plasma protein in sanitary and nonsanitary environments. J. Anim. Sci.  85: 1302– 1310. Google Scholar CrossRef Search ADS PubMed  Footnotes 1 This work was supported by the National Natural Science Foundation for Distinguished Young Scholars of China (grant no. 31025027) and the China Postdoctoral Science Foundation funded project (no. 2012T50528). American Society of Animal Science

Welch, C. M.;Thornton, K. J.;Murdoch, G. K.;Chapalamadugu, K. C.;Schneider, C. S.;Ahola, J. K.;Hall, J. B.;Price, W. J.;Hill, R. A.

doi: 10.2527/jas.2013-6609pmid: 24085409

McNamara, K. A.;Knox, R. V.

doi: 10.2527/jas.2013-6867pmid: 24167000

Abstract Frozen-thawed boar sperm (FTS) has reduced motility and viability compared to cooled semen. Motility of FTS is related to in vitro and in vivo fertility, but this effect has not been determined in relation to the timing of ovulation. To test the effect of variable FTS motility in a multiple-AI system, ejaculates from 38 boars were collected and frozen in 0.5-mL straws. Upon thawing, samples were classified (mean ± SEM) by motility as poor (P, 20.2% ± 1.1%), moderate (M, 31.3% ± 0.9%), or good (G, 43.5% ± 0.8%). In replicates, mature gilts were synchronized and checked for estrus at 12-h intervals and assigned (n = 207) to receive 4.0 billion total sperm in each AI at 24 and 36 h after onset of estrus using the treatments: 1) P and M (P-M), 2) M and P (M-P), 3) G and M (G-M), and 4) M and G (M-G). For each treatment combination, a set of 3 boars was randomly selected within motility class for their allelic distinction with M sperm from a single boar represented across all treatments and sires used in both first and second inseminations. The insemination to ovulation interval (IOI) was determined using ultrasound every 12 h. Reproductive tracts were collected at approximately d 32 after AI. Treatment did not interact with IOI (P > 0.10) and did not affect (P > 0.10) pregnancy rate (57%, 67%, 71%, 76% ± 7.2%, pooled SEM) or total number of fetuses (9.2, 9.1, 9.5, 10.0 ± 0.8) for P-M, M-P, G-M, and M-G treatments, respectively. Treatment did affect (P < 0.05) the number of fetuses sired from the first AI (3.1, 7.2, 6.4, 6.3 ± 1.2) and second AI (5.7, 2.6, 3.0, 3.6 ± 0.9) for the P-M, M-P, G-M, and M-G treatments, respectively. The IOI also influenced (P < 0.05) the proportion of offspring sired by the second AI (30.0%, 57.7%, 51.3%, 18.3% ± 6.5%), as well as the number of fetuses sired by each AI. These results indicate FTS motility had no effect on pregnancy rate or litter size but did affect the number of fetuses sired from the first and second inseminations. The first AI appears to sire most of the litter except when P sperm was used. Number of fetuses sired was reduced when P sperm was used in either insemination compared to M, although no difference was evident between M and G. Fetal paternity appears to be a more sensitive marker for identifying the effects of sperm quality and IOI in a multiple-AI system with use of FTS. These results suggest that use of semen of various qualities can be used in combinations to aid in pregnancy establishment and contribute to litter size. INTRODUCTION Artificial insemination with cooled semen is used for breeding most U.S. sows (Weitze, 2000), although frozen-thawed sperm (FTS) is used in <1% of all matings (Johnson et al., 2000). With FTS, there is a decrease in fertility and number of live sperm (Roca et al., 2006) because of cryodamage (Colenbrander et al., 2000), which results in fewer AI doses and sperm with a shorter fertile life (Waberski et al., 1994; Watson, 2000). Further, with conventional AI, 4 to 6 billion FTS are required to produce optimal litter sizes similar to AI with cooled sperm (Watson and Behan, 2002; Reicks and Levis, 2008), and multiple inseminations are needed to compensate for variation in time of ovulation and to establish pregnancy (Soede and Kemp, 1997; Lamberson and Safranski, 2000). Despite these limits, commercial application (Knox, 2011; Didion et al., 2013) could provide opportunities for improved sample testing for disease, extended storage, and use of sperm from valued sires (Bailey et al., 2008). Motility is an important indicator for cooled semen use (Gadea, 2005), and >70% motility is the industry standard because fertility is reduced when used below this measure (Flowers, 1997; Gadea et al., 2004). However, most FTS is <60% motile (Purdy, 2008; Medrano et al., 2009) with variation among (Hernández et al., 2007) and within boars (Pelaez et al., 2006). Motility of FTS affects in vitro (Gil et al., 2008) and in vivo (Casas et al., 2010) measures of fertility, but there is no information on FTS motility in relation to interval from insemination to ovulation in a multiple-AI system. This information could be important for predicting fertility of FTS with different qualities, timing inseminations, and changing requirements for number of sperm. To develop insemination strategies with improved potential for fertility with FTS of variable quality in a double-AI system, we tested the effects of good-, moderate-, and poor-motility sperm when used in the first or second AI in relation to ovulation time on pregnancy establishment, litter size, and fetal paternity. MATERIALS AND METHODS: The use of animals for this experiment was approved by the institutional animal care and use committee of the University of Illinois. Animals and Synchronization of Estrus This experiment was performed in 7 replicates from January to October 2012 at the University of Illinois Swine Research Center. Terminal line gilts (n = 290; Genetiporc USA, Alexandria, MN) were moved from a finishing barn into pens in a gestation building between 147 and 180 d of age. Gilts were observed for estrous expression using the back-pressure test during fence line exposure to a mature boar. Gilts that had exhibited estrus (n = 268) were then moved into gestation stalls and synchronized by feeding 15 mg×gilt-1×d-1 of MATRIX (Altrenogest 2.2 mg/mL, Merck Animal Health, Summit, NJ) for 14 d as a top-dress on a standard sow gestation diet. In the first 2 replicates, gilts were treated with 5.0 mL of PG600 (400 IU eCG and 200 IU hCG, Merck Animal Health, Summit, NJ) 24 h following the last MATRIX feeding (LMF) to improve synchrony and expression of estrus. However, it was not used in the remaining replicates because of variation in follicle size and limited improvement in expression of estrus. Beginning on the third day following LMF, estrous detection was performed twice daily at 12-h intervals (0700 and 1900 h). Experimental Design To test the fertility effect of FTS with different motility classification for use in a multiple-AI system in relation to time of ovulation, ejaculates were collected from multiple boars and frozen in 0.5-mL straws. Upon thawing, samples were analyzed and classified on the basis of motility as good (G, ≥40%), moderate (M, 26% to 39%), or poor (P, 16% to 25%), similar to Pelaez et al. (2006) and Hernández et al. (2007). At onset of estrus, gilts (n = 207) were assigned to receive a total of 4.0 billion sperm (live + dead sperm) from a single motility class in the first and second AI performed 24 and 36 h after onset of estrus using the following treatment combinations: 1) P and M (P-M), 2) M and P (M-P), 3) G and M (G-M), and 4) M and G (M-G). For each set of the treatment combinations, 3 boars were randomly selected within each motility class for their allelic distinction with M sperm from a single boar represented across the treatment set and all sires used in both first and second inseminations (Table 1). This design allowed us to make direct treatment comparisons and indirect comparisons without having all treatment combinations represented, as described by Collins et al. (2008). The use of boars within each motility class with allelic distinction allowed us to identify fetal paternity, and using each sire in the first and second inseminations allowed determination for the effect of insemination order and sire effects. Gilts assigned to treatment averaged 229 ± 1 d of age (mean ± SEM) at first insemination. Fixed-time inseminations occurred at 24 and 36 h following onset of estrus. Transrectal real-time ultrasound (Aloka 500V, Tokyo, Japan) began 12 h following onset of estrus and continued at 12-h intervals to observe the number and size of ovulatory follicles (≥6.5 mm) and to determine when ovulation was complete (Knox and Althouse, 1999; Knox et al., 2002). These data were used to calculate the interval from onset of estrus to ovulation (EOI) and the insemination to ovulation interval (IOI) for first and second AI. The intervals calculated used the hours until ovulation was observed to have occurred. Table 1. Example allocation of gilts to treatment within a replicate by sire for first and second inseminations with frozen-thawed boar sperm classified as having poor (P), moderate (M), or good (G) motility1 Treatment  P-M  M-P  G-M  M-G  1A,B  2B,A  3C,B  4B,C  5D,E  6E,D  7F,E  8E,F  9A,E  10E,A  11C,E  12E,C  13D,B  14B,D  15F,B  16B,F  Treatment  P-M  M-P  G-M  M-G  1A,B  2B,A  3C,B  4B,C  5D,E  6E,D  7F,E  8E,F  9A,E  10E,A  11C,E  12E,C  13D,B  14B,D  15F,B  16B,F  A–FNumbers (1 to 16) represent the first 16/36 gilts assigned to treatment in a replicate based on order of estrus expression, with superscripts indicating some of the 38 sires that could be used (A–F) in order of insemination. 1Gilts were assigned in numerical order as they expressed estrus. For each treatment row, 3 boars were randomly selected within motility class and by their allelic distinction from the other boars used within the row. Also within a row, the same boar with moderate sperm was represented across all treatments. This treatment design allowed direct comparisons among treatments and indirect comparisons without performing all comparisons. Treatments were designed to obtain the best estimates without performing all comparisons between boars, similar to Collins et al. (2008). View Large Table 1. Example allocation of gilts to treatment within a replicate by sire for first and second inseminations with frozen-thawed boar sperm classified as having poor (P), moderate (M), or good (G) motility1 Treatment  P-M  M-P  G-M  M-G  1A,B  2B,A  3C,B  4B,C  5D,E  6E,D  7F,E  8E,F  9A,E  10E,A  11C,E  12E,C  13D,B  14B,D  15F,B  16B,F  Treatment  P-M  M-P  G-M  M-G  1A,B  2B,A  3C,B  4B,C  5D,E  6E,D  7F,E  8E,F  9A,E  10E,A  11C,E  12E,C  13D,B  14B,D  15F,B  16B,F  A–FNumbers (1 to 16) represent the first 16/36 gilts assigned to treatment in a replicate based on order of estrus expression, with superscripts indicating some of the 38 sires that could be used (A–F) in order of insemination. 1Gilts were assigned in numerical order as they expressed estrus. For each treatment row, 3 boars were randomly selected within motility class and by their allelic distinction from the other boars used within the row. Also within a row, the same boar with moderate sperm was represented across all treatments. This treatment design allowed direct comparisons among treatments and indirect comparisons without performing all comparisons. Treatments were designed to obtain the best estimates without performing all comparisons between boars, similar to Collins et al. (2008). View Large Semen Collection, Freezing, and Evaluation Boars (n = 38, Genetiporc USA, Alexandria, MN) from a commercial genetic supplier and used in a regular collection rotation were selected for this experiment. Ejaculates (n = 54) were frozen from July 2010 to August 2012. Multiple ejaculates were obtained for 14 of the boars with 13/14 having ejaculates in only 1 category although 1 boar had ejaculates in both M and P classes. At the boar stud, semen was collected and diluted with Modena extender (Swine Genetics International, Cambridge, IA) at a ratio of 1:1 and cooled to 17°C within 0.5 h after collection and shipped overnight to either the USDA-ARS National Center for Genetic Resources Preservation (Fort Collins, CO) for processing as described by Spencer et al. (2010) or to the University of Illinois (Urbana, IL) for processing using the procedures of Ringwelski et al. (2013). Briefly, on semen arrival at the freezing labs, the sample was evaluated for concentration and motility and centrifuged at 800 × g at 4°C for 12 min, and the supernatant was aspirated. The sperm pellet was resuspended with Androhep CryoGuard Cooling Extender (Minitube of America, Verona, WI) to a concentration of 2.8 × 109 sperm/mL and held at 5°C for 2.5 h before dilution with Androhep CryoGuard Freezing Extender (Minitube of America) to a final concentration of 1.4 × 109 sperm/mL. Straws (0.5 mL) were filled and then placed into the Ice Cube controlled-rate freezer (Minitube of America) using the freezing processes described by Spencer et al. (2010) and Ringwelski et al. (2013) and were stored in liquid nitrogen. All FTS samples were evaluated at the University of Illinois by the same technician. Post-thaw motility was determined by evaluating 3 straws independently for each ejaculate before final classification based on motility (Table 2). Straws were thawed at 50°C for 20 s, and the contents expelled into glass tubes at 37°C. Samples were diluted in Androhep CryoGuard Thawing Extender (Minitube of America) at 1:40 for motility and 1:400 for concentration. Motility evaluation was performed at 5, 30, and 60 min after thawing. Samples were examined under a phase-contrast microscope with a 37°C heated stage at 200× magnification. Ten fields were examined to evaluate 100 sperm, and motility was expressed as a percentage of the total number of sperm cells. After subjective analyses were completed and the motility of the 3 straws was within the same class, the initial measure was used for ejaculate classification. All samples were later reconfirmed for motility using CASA (Hamilton Thorne, Beverly, MA) with all sample values within ±5% of the subjective mean and with no change in motility class. Fluorescent staining was also performed on the same ejaculates to determine membrane integrity using propidium iodide (PI; Sigma Aldrich, St. Louis, MO) and acrosome integrity using fluorescein isothiocyanatelectin from Arachis hypogaea (FITC-PNA, Sigma Aldrich). Samples were thawed and diluted 1:50 in 26°C Beltsville Thawing Solution (Minitube of America) and coincubated for 10 to 15 min with the fluorescent stains and then fixed with 0.4% paraformaldehyde in PBS solution. A total of 300 sperm were analyzed for fluorescence using a Carl Zeiss AxioCamHRc (Carl Zeiss Microscopy, LLC., Thornwood, NY) at 400× magnification, and the percentage of PI negative and FITC-PNA positive sperm calculated from the total number of sperm evaluated (Table 2). Table 2. Means (±SEM) for frozen-thawed boar sperm measures of motility, viability, and live sperm with intact acrosomes from ejaculates used in treatment classification based on motility   Post-thaw sperm measures  Motility class1  n2  Motility, %  Viability, %  Live sperm with intact acrosomes, %  Good  16  43.5 ± 0.8x  51.9 ± 2.3x  80.3 ± 1.5  Moderate  23  31.3 ± 0.9y  40.5 ± 2.2y  77.9 ± 1.5  Poor  15  20.2 ± 1.1z  28.2 ± 2.0z  74.8 ± 1.8    Post-thaw sperm measures  Motility class1  n2  Motility, %  Viability, %  Live sperm with intact acrosomes, %  Good  16  43.5 ± 0.8x  51.9 ± 2.3x  80.3 ± 1.5  Moderate  23  31.3 ± 0.9y  40.5 ± 2.2y  77.9 ± 1.5  Poor  15  20.2 ± 1.1z  28.2 ± 2.0z  74.8 ± 1.8  x–zWithin a column, means without a common superscript are different (P < 0.05). 1Evaluation for motility was performed 5 min after thawing using microscopic assessment and then was confirmed using CASA (Hamilton Thorne, Beverly, MA) for all ejaculates, with motility classified as good (G, ≥40%), moderate (M, 26% to 39%), or poor (P, 16% to 25%). 2n = number of ejaculates collected from 38 individual boars. View Large Table 2. Means (±SEM) for frozen-thawed boar sperm measures of motility, viability, and live sperm with intact acrosomes from ejaculates used in treatment classification based on motility   Post-thaw sperm measures  Motility class1  n2  Motility, %  Viability, %  Live sperm with intact acrosomes, %  Good  16  43.5 ± 0.8x  51.9 ± 2.3x  80.3 ± 1.5  Moderate  23  31.3 ± 0.9y  40.5 ± 2.2y  77.9 ± 1.5  Poor  15  20.2 ± 1.1z  28.2 ± 2.0z  74.8 ± 1.8    Post-thaw sperm measures  Motility class1  n2  Motility, %  Viability, %  Live sperm with intact acrosomes, %  Good  16  43.5 ± 0.8x  51.9 ± 2.3x  80.3 ± 1.5  Moderate  23  31.3 ± 0.9y  40.5 ± 2.2y  77.9 ± 1.5  Poor  15  20.2 ± 1.1z  28.2 ± 2.0z  74.8 ± 1.8  x–zWithin a column, means without a common superscript are different (P < 0.05). 1Evaluation for motility was performed 5 min after thawing using microscopic assessment and then was confirmed using CASA (Hamilton Thorne, Beverly, MA) for all ejaculates, with motility classified as good (G, ≥40%), moderate (M, 26% to 39%), or poor (P, 16% to 25%). 2n = number of ejaculates collected from 38 individual boars. View Large Semen Thawing and Insemination Thawing of straws was performed in a thermally controlled water bath at 50°C for 20 s. The contents of the straws were expelled into 100-mL plastic AI bottles containing 80 mL of Androhep CryoGuard Thawing Extender held in a 26°C water bath. Within 15 min of thawing, intracervical insemination was performed using polygel-tipped AI catheters (Minitube of America). All inseminations were subjectively scored for fluid loss during and in the minute following AI catheter removal, with a score of 1 showing little or no fluid lost, 2 exhibiting moderate loss, and 3 indicating excessive fluid lost. The average scores for all first and second inseminations were 1.6 ± 0.1 and 1.7 ± 0.0, respectively. Reproductive Tract Processing for Pregnancy and Litter Responses Gilts were slaughtered at a local abattoir 31 to 35 d following AI (32 ± 1.0 d, mean ± SE). The reproductive tracts were collected and assessed for pregnancy status and the number of normal and abnormal fetuses. Fetuses were counted and weighed, and those with abnormal appearance in size or color and that were ≥1 SD below the average weight of the normal fetuses were classified as abnormal. A liver sample was removed from each individual fetus for DNA genotyping to determine paternity. Ovaries were examined for the number of corpora lutea and the presence of any abnormalities such as cystic (>12 mm) follicles or cystic corpora lutea. DNA Genotyping To determine the impact of FTS motility and IOI, parental identification of fetuses was performed using DNA obtained from the semen of all boars, the blood of all gilts bred, and the liver of all fetuses using a procedure described previously by Ringwelski et al. (2013). Briefly, samples were digested, and DNA was isolated using ZR-96 Quick-gDNA (Zymo Research, Irvine, CA). A panel of 14 microsatellite markers was chosen and primers were synthesized for use in PCR for fragment analysis based on size and fluorescent tag combination. Multiplex PCR products were combined and purified and sequenced as described by Meyers et al. (2010). Alleles were identified using GeneMarker software (SoftGenetics, LLC, State College, PA) and checked manually. Parentage of the fetus was determined manually using the genotypes from the dam and the 2 potential sires. Statistical Analysis Data were analyzed using ANOVA procedures in SAS (SAS Institute Inc., Cary, NC). Continuous response measures were analyzed using the PROC MIXED procedure for significance of the main effects using the F-test and differences between least squares means identified using the t test. Binary response measures were analyzed using PROC GENMOD, and significant main effects and differences between least squares means were identified using the x2 test. Binary analyses were performed using a binary distribution and a logit link. All models for the dependent variables included the main effects of treatment (4 levels), IOI for first and second AI and their interaction, and first AI and second AI sire and replicate. Of the other variables tested, only ovulation rate (P < 0.05) was significant for total and normal fetuses. The other variables, such as insemination score, interval from LMF to estrus, and duration of estrus, were included as class variables or covariates where appropriate and were not significant (P > 0.10) and were removed from final models. The assumptions of ANOVA for normal distribution of data were evaluated and tested using PROC UNIVARIATE, and those for homogeneity of variance were evaluated and tested using Levene's test. Significant differences were identified at P ≤ 0.05 and trends at P > 0.05 and ≤ 0.10. Gilts assigned to treatment were excluded from analyses for abnormalities that included an EOI > 60 h (n = 12), ovarian cysts at estrus or at slaughter (n = 21), or uterine infection at slaughter (n = 2). RESULTS Although a total of 4.0 × 109 sperm were used in each AI for all treatments, the actual number of motile sperm inseminated was 0.8 × 109 ± 0.2 for P, 1.2 × 109 ± 0.2 for M, and 1.7 × 109 ± 0.1 for G. The interval from LMF to estrus was 7.2 ± 0.8 d, and the duration of estrus averaged 44.1 ± 1.0 h. The EOI averaged 35.3 ± 0.8 h, and the ovulation rate averaged 14.8 ± 0.3 corpora lutea. There was no treatment × IOI interaction for first or second AI for any response measure assessed in this study (P > 0.10), and therefore, only main effects are presented. Sire used in the first and second AI was also not significant (P > 0.10) and was not included in the final models. Pregnancy Rate and Litter Responses There was no effect (P > 0.10) of treatment (Table 3) or IOI (Table 4) on pregnancy rate, number of normal fetuses, or average fetal weight. Table 3. Means (±SEM) for pregnancy rate, number of fetuses, embryo survival, and fetal paternity from each insemination as affected by first and second inseminations with frozen-thawed boar sperm (FTS) classified as having poor (P), moderate (M), or good (G) motility Treatment1  n2  Pregnant,3 %  Total fetuses/litter  Normal fetuses/litter  Embryo survival,4 %  Proportion of litter from first AI, %  Proportion of litter from second AI, %  Total number of fetuses from first AI  Total number of fetuses from second AI  P-M  41  57.1 ± 8.0  9.2 ± 1.1  8.7 ± 1.0  61.5 ± 6.5  38.1 ± 9.4  61.4 ± 9.2  3.1 ± 0.9x  5.7 ± 1.1x  M-P  43  67.4 ± 7.2  9.1 ± 0.7  8.8 ± 0.7  63.1 ± 4.4  67.6 ± 8.6  31.7 ± 8.5  7.2 ± 1.1y  2.6 ± 0.8y  G-M  45  71.1 ± 6.8  9.5 ± 0.7  9.3 ± 0.7  61.6 ± 4.6  65.5 ± 7.0  33.6 ± 7.0  6.4 ± 0.9y  3.0 ± 0.8y  M-G  42  75.6 ± 6.8  10.0 ± 0.8  9.6 ± 0.7  65.5 ± 4.4  65.0 ± 7.5  34.8 ± 7.6  6.3 ± 1.0y  3.6 ± 0.9y  Treatment1  n2  Pregnant,3 %  Total fetuses/litter  Normal fetuses/litter  Embryo survival,4 %  Proportion of litter from first AI, %  Proportion of litter from second AI, %  Total number of fetuses from first AI  Total number of fetuses from second AI  P-M  41  57.1 ± 8.0  9.2 ± 1.1  8.7 ± 1.0  61.5 ± 6.5  38.1 ± 9.4  61.4 ± 9.2  3.1 ± 0.9x  5.7 ± 1.1x  M-P  43  67.4 ± 7.2  9.1 ± 0.7  8.8 ± 0.7  63.1 ± 4.4  67.6 ± 8.6  31.7 ± 8.5  7.2 ± 1.1y  2.6 ± 0.8y  G-M  45  71.1 ± 6.8  9.5 ± 0.7  9.3 ± 0.7  61.6 ± 4.6  65.5 ± 7.0  33.6 ± 7.0  6.4 ± 0.9y  3.0 ± 0.8y  M-G  42  75.6 ± 6.8  10.0 ± 0.8  9.6 ± 0.7  65.5 ± 4.4  65.0 ± 7.5  34.8 ± 7.6  6.3 ± 1.0y  3.6 ± 0.9y  x,yWithin a column, means without a common superscript are different (P < 0.05). 1Gilts received a total number of 4.0 billion FTS assessed as P (20.2% ± 1.1%), M (31.3% ± 0.9%), or G (43.5% ± 0.8%) in 80 mL of extender at 24 and 36 h after detection of estrus. The number of motile FTS in each AI was 0.8 × 109 ± 0.2 for P, 1.2 × 109 ± 0.2 for M, and 1.7 × 109 ± 0.1 for G. 2Of all animals assigned to treatment, gilts were excluded from analyses because of an abnormally long estrus to ovulation interval (>60 h, n = 12) or if ovarian and reproductive tract abnormalities, such as ovarian cysts or uterine infection, were evident at estrus or slaughter (n = 23). 3Determined at slaughter at 31 to 35 d following AI. 4Embryo survival determined from number of normal fetuses/number of corpora lutea. View Large Table 3. Means (±SEM) for pregnancy rate, number of fetuses, embryo survival, and fetal paternity from each insemination as affected by first and second inseminations with frozen-thawed boar sperm (FTS) classified as having poor (P), moderate (M), or good (G) motility Treatment1  n2  Pregnant,3 %  Total fetuses/litter  Normal fetuses/litter  Embryo survival,4 %  Proportion of litter from first AI, %  Proportion of litter from second AI, %  Total number of fetuses from first AI  Total number of fetuses from second AI  P-M  41  57.1 ± 8.0  9.2 ± 1.1  8.7 ± 1.0  61.5 ± 6.5  38.1 ± 9.4  61.4 ± 9.2  3.1 ± 0.9x  5.7 ± 1.1x  M-P  43  67.4 ± 7.2  9.1 ± 0.7  8.8 ± 0.7  63.1 ± 4.4  67.6 ± 8.6  31.7 ± 8.5  7.2 ± 1.1y  2.6 ± 0.8y  G-M  45  71.1 ± 6.8  9.5 ± 0.7  9.3 ± 0.7  61.6 ± 4.6  65.5 ± 7.0  33.6 ± 7.0  6.4 ± 0.9y  3.0 ± 0.8y  M-G  42  75.6 ± 6.8  10.0 ± 0.8  9.6 ± 0.7  65.5 ± 4.4  65.0 ± 7.5  34.8 ± 7.6  6.3 ± 1.0y  3.6 ± 0.9y  Treatment1  n2  Pregnant,3 %  Total fetuses/litter  Normal fetuses/litter  Embryo survival,4 %  Proportion of litter from first AI, %  Proportion of litter from second AI, %  Total number of fetuses from first AI  Total number of fetuses from second AI  P-M  41  57.1 ± 8.0  9.2 ± 1.1  8.7 ± 1.0  61.5 ± 6.5  38.1 ± 9.4  61.4 ± 9.2  3.1 ± 0.9x  5.7 ± 1.1x  M-P  43  67.4 ± 7.2  9.1 ± 0.7  8.8 ± 0.7  63.1 ± 4.4  67.6 ± 8.6  31.7 ± 8.5  7.2 ± 1.1y  2.6 ± 0.8y  G-M  45  71.1 ± 6.8  9.5 ± 0.7  9.3 ± 0.7  61.6 ± 4.6  65.5 ± 7.0  33.6 ± 7.0  6.4 ± 0.9y  3.0 ± 0.8y  M-G  42  75.6 ± 6.8  10.0 ± 0.8  9.6 ± 0.7  65.5 ± 4.4  65.0 ± 7.5  34.8 ± 7.6  6.3 ± 1.0y  3.6 ± 0.9y  x,yWithin a column, means without a common superscript are different (P < 0.05). 1Gilts received a total number of 4.0 billion FTS assessed as P (20.2% ± 1.1%), M (31.3% ± 0.9%), or G (43.5% ± 0.8%) in 80 mL of extender at 24 and 36 h after detection of estrus. The number of motile FTS in each AI was 0.8 × 109 ± 0.2 for P, 1.2 × 109 ± 0.2 for M, and 1.7 × 109 ± 0.1 for G. 2Of all animals assigned to treatment, gilts were excluded from analyses because of an abnormally long estrus to ovulation interval (>60 h, n = 12) or if ovarian and reproductive tract abnormalities, such as ovarian cysts or uterine infection, were evident at estrus or slaughter (n = 23). 3Determined at slaughter at 31 to 35 d following AI. 4Embryo survival determined from number of normal fetuses/number of corpora lutea. View Large Table 4. Means (±SEM) for pregnancy rate, number of fetuses, embryo survival, and fetal paternity from each insemination as affected by interval from insemination ovulation when using frozen thawed boar sperm. Insemination to ovulation,1 h  n2  Pregnant,3 %  Total fetuses/litter  Normal fetuses/litter  Embryo survival, %  Proportion of litter from first AI, %  Proportion of litter from second AI, %  Total number of fetuses from first AI  Total number of fetuses from second AI  AI 1  AI 2  -36  -24  7  57.1 ± 20.2  6.3 ± 1.3  6.3 ± 1.3  44.7 ± 13.1  70.0 ± 23.8a,b  30.0 ± 23.8x,y  4.8 ± 2.1x,y  1.5 ± 1.2x  -24  -12  32  62.5 ± 8.7  8.4 ± 0.8  8.1 ± 0.9  57.8 ± 5.4  40.9 ± 9.8a  57.7 ± 9.7y  3.4 ± 0.9x  5.4 ± 1.1y  -12  0  66  70.0 ± 5.7  9.8 ± 0.7  9.6 ± 0.6  65.1 ± 4.3  49.4 ± 6.8a  51.3 ± 6.7y  4.9 ± 0.8x  5.0 ± 0.8y  0  12  55  68.5 ± 6.3  10.2 ± 0.1  9.8 ± 0.7  66.5 ± 4.1  80.1 ± 5.5b  18.3 ± 5.4x  8.4 ± 0.9y  1.6 ± 0.6x  Insemination to ovulation,1 h  n2  Pregnant,3 %  Total fetuses/litter  Normal fetuses/litter  Embryo survival, %  Proportion of litter from first AI, %  Proportion of litter from second AI, %  Total number of fetuses from first AI  Total number of fetuses from second AI  AI 1  AI 2  -36  -24  7  57.1 ± 20.2  6.3 ± 1.3  6.3 ± 1.3  44.7 ± 13.1  70.0 ± 23.8a,b  30.0 ± 23.8x,y  4.8 ± 2.1x,y  1.5 ± 1.2x  -24  -12  32  62.5 ± 8.7  8.4 ± 0.8  8.1 ± 0.9  57.8 ± 5.4  40.9 ± 9.8a  57.7 ± 9.7y  3.4 ± 0.9x  5.4 ± 1.1y  -12  0  66  70.0 ± 5.7  9.8 ± 0.7  9.6 ± 0.6  65.1 ± 4.3  49.4 ± 6.8a  51.3 ± 6.7y  4.9 ± 0.8x  5.0 ± 0.8y  0  12  55  68.5 ± 6.3  10.2 ± 0.1  9.8 ± 0.7  66.5 ± 4.1  80.1 ± 5.5b  18.3 ± 5.4x  8.4 ± 0.9y  1.6 ± 0.6x  a,bWithin a column, means without a common superscript are different (P = 0.1). x,yWithin a column, means without a common superscript are different (P < 0.05). 1Estrus detection and ultrasound were performed at 12-h intervals (0600 and 1800 h), and gilts received 4.0 billion total sperm assessed as having poor (P, 20.2 ± 3.4%), moderate (M, 31.3 ± 4.2%), or good (G, 43.5 ± 3.2%) motility in the first or second insemination at 24 and 36 h after detection of estrus. 2Gilts (n = 12) could not be included because of the inability to perform repeated transrectal ultrasound to confirm ovulation time. 3Determined at slaughter at d 31 to 35 following AI. View Large Table 4. Means (±SEM) for pregnancy rate, number of fetuses, embryo survival, and fetal paternity from each insemination as affected by interval from insemination ovulation when using frozen thawed boar sperm. Insemination to ovulation,1 h  n2  Pregnant,3 %  Total fetuses/litter  Normal fetuses/litter  Embryo survival, %  Proportion of litter from first AI, %  Proportion of litter from second AI, %  Total number of fetuses from first AI  Total number of fetuses from second AI  AI 1  AI 2  -36  -24  7  57.1 ± 20.2  6.3 ± 1.3  6.3 ± 1.3  44.7 ± 13.1  70.0 ± 23.8a,b  30.0 ± 23.8x,y  4.8 ± 2.1x,y  1.5 ± 1.2x  -24  -12  32  62.5 ± 8.7  8.4 ± 0.8  8.1 ± 0.9  57.8 ± 5.4  40.9 ± 9.8a  57.7 ± 9.7y  3.4 ± 0.9x  5.4 ± 1.1y  -12  0  66  70.0 ± 5.7  9.8 ± 0.7  9.6 ± 0.6  65.1 ± 4.3  49.4 ± 6.8a  51.3 ± 6.7y  4.9 ± 0.8x  5.0 ± 0.8y  0  12  55  68.5 ± 6.3  10.2 ± 0.1  9.8 ± 0.7  66.5 ± 4.1  80.1 ± 5.5b  18.3 ± 5.4x  8.4 ± 0.9y  1.6 ± 0.6x  Insemination to ovulation,1 h  n2  Pregnant,3 %  Total fetuses/litter  Normal fetuses/litter  Embryo survival, %  Proportion of litter from first AI, %  Proportion of litter from second AI, %  Total number of fetuses from first AI  Total number of fetuses from second AI  AI 1  AI 2  -36  -24  7  57.1 ± 20.2  6.3 ± 1.3  6.3 ± 1.3  44.7 ± 13.1  70.0 ± 23.8a,b  30.0 ± 23.8x,y  4.8 ± 2.1x,y  1.5 ± 1.2x  -24  -12  32  62.5 ± 8.7  8.4 ± 0.8  8.1 ± 0.9  57.8 ± 5.4  40.9 ± 9.8a  57.7 ± 9.7y  3.4 ± 0.9x  5.4 ± 1.1y  -12  0  66  70.0 ± 5.7  9.8 ± 0.7  9.6 ± 0.6  65.1 ± 4.3  49.4 ± 6.8a  51.3 ± 6.7y  4.9 ± 0.8x  5.0 ± 0.8y  0  12  55  68.5 ± 6.3  10.2 ± 0.1  9.8 ± 0.7  66.5 ± 4.1  80.1 ± 5.5b  18.3 ± 5.4x  8.4 ± 0.9y  1.6 ± 0.6x  a,bWithin a column, means without a common superscript are different (P = 0.1). x,yWithin a column, means without a common superscript are different (P < 0.05). 1Estrus detection and ultrasound were performed at 12-h intervals (0600 and 1800 h), and gilts received 4.0 billion total sperm assessed as having poor (P, 20.2 ± 3.4%), moderate (M, 31.3 ± 4.2%), or good (G, 43.5 ± 3.2%) motility in the first or second insemination at 24 and 36 h after detection of estrus. 2Gilts (n = 12) could not be included because of the inability to perform repeated transrectal ultrasound to confirm ovulation time. 3Determined at slaughter at d 31 to 35 following AI. View Large Proportion of Litter and Fetuses from First and Second Inseminations Treatment influenced the number of fetuses sired from the first and second AI (Table 3; P < 0.05). The number of fetuses sired from the first AI was greater than those sired from the second in all treatments except P-M. The frequency distribution of litters with number of fetuses sired within a litter from the first AI and second AI is shown in Fig. 1 and Fig. 2, respectively. The IOI also affected number of fetuses sired (Table 4; P < 0.05), with an increase in number sired when the first AI occurred at 0 h relative to ovulation compared to −12 and −24 h but not −36 h. There was no effect of treatment (Table 3) on the proportion of fetuses sired from the first AI, but there was a trend (P = 0.10) for the IOI from the first AI to impact the proportion (Table 4). The IOI for the second AI influenced the proportion of fetuses, resulting in smaller proportions sired when ovulation occurred at +12 or −24 h compared to 0 and −12 h (Table 4; P < 0.05). Figure 1. View largeDownload slide The effects of first and second inseminations using combinations of frozen-thawed sperm (FTS) classified by post-thaw motility as poor (P), moderate (M), or good (G) from unique sires on the frequency of litters with defined numbers of fetuses sired by the first AI within a litter. Figure 1. View largeDownload slide The effects of first and second inseminations using combinations of frozen-thawed sperm (FTS) classified by post-thaw motility as poor (P), moderate (M), or good (G) from unique sires on the frequency of litters with defined numbers of fetuses sired by the first AI within a litter. Figure 2. View largeDownload slide The effects of first and second inseminations using combinations of frozen-thawed sperm (FTS) classified by post-thaw motility as poor (P), moderate (M), or good (G) from unique sires on the frequency of litters with defined numbers of fetuses sired by the second AI within a litter. Figure 2. View largeDownload slide The effects of first and second inseminations using combinations of frozen-thawed sperm (FTS) classified by post-thaw motility as poor (P), moderate (M), or good (G) from unique sires on the frequency of litters with defined numbers of fetuses sired by the second AI within a litter. DISCUSSION This study was designed to evaluate insemination strategies when using FTS of variable motility in a double-AI system for improved fertility and to extend the use of valuable sperm. To accomplish this, our approach was to determine whether different classes of FTS motility used in the first or second AI would affect pregnancy establishment, litter size, and fetal paternity in relation to time of ovulation. Although we were not able to detect differences in pregnancy rate or litter size with FTS differing by ∼10% in motility, changes in the number of fetuses sired by each AI was sensitive enough to identify the effects of treatment and IOI. Regardless of the order, when P sperm was used there were, on average, 30% fewer fetuses sired when compared to use of M sperm. Interestingly, the first insemination sired ∼60% of all litters except when poor-motility sperm was used. In addition, when examining the number of fetuses sired, compared with G or M sperm, when P sperm was used in the first AI, fewer litters with more than 6 pigs were sired by that insemination. No differences in fertility were evident when G and M sperm were used in combination. This work has practical implications because the motility of FTS has been reported to affect in vitro (Gil et al., 2008) and in vivo (Casas et al., 2010) fertility. Of concern is that the motility of FTS is already reduced (Purdy, 2008; Medrano et al., 2009), with wide variation noted among boars (Woelders et al., 1995; Hofmo and Grevle, 2000; Hernández et al., 2007; Juarez et al., 2011) and between ejaculates within a boar (Pelaez et al., 2006). This can create problems as superior sires may never or only infrequently produce good-quality frozen sperm. If this is the case, they may not be used for cryopreservation or may only be used with high numbers of sperm in the AI dose (Cremades et al., 2005; Foxcroft et al., 2008; Spencer et al., 2010). Our results suggest that for valued sires with variable quality in the post-thaw motility from different ejaculates, the order for use in the double AI can affect fertility and may be used to determine how to best extend the use of limited numbers of fertile sperm. Classification for FTS quality based on motility is actually quite variable (Thurston et al., 2003; Hernández et al., 2007; Casas et al., 2010), with the majority of samples <60% motile. Detecting fertility effects among different levels of low-motility FTS may be challenging, but it would appear that paternity testing within the litter offers an opportunity to identify these subtle differences. Previous studies with cooled semen have shown that the proportion of offspring sired can be used as an indicator of fertility differences between boars when using pooled semen or first and second inseminations from different sires (Dziuk, 1970; Flowers, 1997). From these studies it was reported that sperm from 1 of the males will often fertilize a majority of the eggs (Dziuk, 1996) and that the seminal plasma components of high-fertility boars could be added to the sperm of low-fertility boars to improve their fertility (Flowers, 1997; Foxcroft et al., 2008). Interestingly, seminal plasma is removed during the processing procedures for FTS, and when added back to good-quality samples in homo- or heterospermic inseminations, no beneficial effect on fertility was observed (Pursel and Johnson, 1975). This might suggest that the fertility effect of seminal plasma might not be detected for FTS or only if poor-motility FTS were used. Ringwelski et al. (2013) also used heterospermic inseminations to detect an interaction of interval between inseminations and the EOI. Although use of pooled semen (heterospermic) is common in commercial pig production to reduce the risk of lowered pig production in groups of females inseminated with the same semen from a subfertile boar, this technique is not generally used at the genetic selection levels and may not be applied widely with use of FTS (Knox et al., 2008). Although it is not entirely clear how motility differences in FTS translate into different fertility outcomes, it has been reported that the addition of caffeine to FTS in vitro results in increased motility, which provides an improved estimate of viable sperm compared to samples tested without caffeine (Pelaez et al., 2006). This observation would tend to support the data from the present study and many other published studies in which the viability for FTS is often higher than the motility. It is also possible that fertility effects from motility may arise from sperm that are classified as G being more stable or resistant to changes in the female reproductive tract following AI when compared to lower-quality sperm (Hernández et al., 2007). Reduced motility of FTS might also be indicative of higher rates or degrees of cryoinjury and be associated with increased sperm backflow and phagocytosis in the uterus following AI (Roca et al., 2006). Supporting evidence for this effect may come from the work of Yamaguchi et al. (2009), who added caffeine and CaCl2 to FTS and observed an increase of sperm in the reservoirs and improved fertility following AI and hypothesized an effect through suppression of the uterine immune response. Our results also support the work of others where differences of ∼10% in FTS motility with use of moderate- to good-quality sperm did not alter fertility in vitro (Eriksson and Rodriguez-Martinez, 2000) or in vivo (Pursel and Johnson, 1975). However, Bwanga et al. (1991) did report changes in fertilization rates when ∼20% differences occurred between the M and G ranges. From a practical standpoint, since variation between and within boar ejaculates over time for pre- and postthaw quality has been reported, an obvious question becomes how or if to use these poor-quality samples to take advantage of limited genetic material. This study suggests that strategic use of FTS, including poor quality, could be used to help increase the use of genetically superior sires and improve numbers of pigs produced. Extrapolating our data for pig production with the use of a fertility index based on 100 sows mated and using the treatment means for pregnancy rate and number of healthy fetuses, the index results in a 10% to 13% stepwise reduction in potential pigs produced from the use of M-G (756), G-M (711), M-P (674), and P-M (561). These data indicate an importance to the order of insemination quality. Although the means for pregnancy and litter size were not statistically different, the numerical means for these and the production index suggest a fertility advantage for use of combinations of good- and moderate-motility sperm. Further, there is a clear advantage in fertility for the first AI except when poor-motility FTS is used. From a practical standpoint, if valued genetic FTS is available from multiple ejaculates from a single sire but with variable quality, it may be possible to use the sperm more efficiently using higher-quality sperm for the first AI and using P sperm in the second AI. Although our treatment design did not include all combinations with P-P, M-M, and G-G, our data would suggest reduced fertility with P-P, although increased fertility would not be expected with G-G compared with use of G-M, M-G, or M-M. However, it may be possible to improve fertility by mixing sperm of various qualities or by using the highest-quality sperm in the first AI. Further, for use of FTS, it would appear that there was no significant increase in fertility when the number of motile sperm inseminated was 1.7 billion compared to 1.2 billion, although use of 0.8 billion sperm in either the first or second AI resulted in a decrease in pigs sired. This information could be used to help extend the use of superior sires with FTS based on numbers of motile sperm. Single inseminations would be desirable for use of cooled or FTS and could each result in good fertility if they occurred at the optimal time before ovulation (Waberski et al., 1994; Soede et al., 1995). However, variation in the EOI has been reported in mature gilts (Spencer et al., 2010; Ringwelski et al., 2013) and in weaned sows (Weitze et al., 1994; Soede and Kemp, 1995; Nissen et al., 1997). This variation causes some inseminations to occur too early or too late, resulting in decreased pregnancy rates and litter sizes with cooled sperm (Soede et al., 1995; Nissen et al., 1997; Almeida et al., 2000) and with FTS (Spencer et al., 2010; Ringwelski et al., 2013). As a result of this variation and the inability to predict ovulation, the industry has adopted a protocol where 2 inseminations are spaced 18 to 24 h apart for cooled semen for optimal fertility compared to the use of single inseminations (Flowers and Esbenshade, 1993; Lamberson and Safranski, 2000; Koketsu, 2005). The timing and spacing of inseminations with FTS is even more critical because the duration of sperm fertility in-vivo is reduced compared to cooled semen (Waberski et al., 1994), and single inseminations with FTS often lead to reduced fertility (Johnson, 1985; Almlid et al., 1987). Studies that control ovulation time have been successful with a single AI occurring 6 h ahead of ovulation but not following ovulation (Bertani et al., 1997). Spencer et al. (2010) did not observe an effect of single compared to double insemination using FTS, but this may have been due to the 24 and 32 h fixed AI timing relative to detection of estrus in the evening, resulting in late-night inseminations that would not be practical for routine use of FTS. Surprisingly, we did not observe an effect for interval from first or second AI to ovulation on pregnancy rate or litter size but did detect differences in fetal paternity. It has been reported that ovulation in gilts usually occurs between 33 and 42 h after onset of estrus (Almeida et al., 2000; Bortolozzo et al., 2005), with induced prepubertal gilts ovulating at 33 h (Spencer et al., 2010) and mature gilts ovulating at 37 h (Ringwelski et al., 2013). According to previous studies, the life-span of FTS in-vivo is approximately 12 h (Waberski et al., 1994; Watson, 2000), but to match the fertility of cooled semen, an AI with FTS must occur within 4 h before ovulation. On the basis of these data, we fixed inseminations at 24 and 36 h after onset of estrus for practical AI timing and observed a mean EOI of 35 h. Approximately 34% of the gilts ovulated by 24 h, 41% by 36 h, 20% by 48 h, and 4% by 60 h. With our fixed-time inseminations, the majority of gilts ovulated within 12 h of receiving an insemination. It was interesting to note that 43% of all litters were sired in entirety by the first AI, 23% by the second AI, and the remaining 34% by the combination of both inseminations. Conclusions The results of this study indicate that when using a double insemination with FTS at 24 and 36 h after onset of estrus in mature gilts, the use of combinations of semen of different motility in the first and second AI did not affect pregnancy rate or litter size, but use of poor-quality sperm reduced piglets sired by that insemination. Further, there appears to be no fertility advantage for use of good- compared to moderate-motility FTS. Fetal paternity is a more sensitive tool for evaluating the effects of FTS quality and AI timing relative to ovulation. These data also suggest that insemination of 1.2 billion motile FTS sperm in a double AI could be used to help extend use of FTS compared to use of greater sperm numbers. The pig production index indicates that the order of insemination of sperm of different qualities will have an effect on overall fertility. LITERATURE CITED Almeida F. R. Novak S. Foxcroft G. R. 2000. The time of ovulation in relation to estrus duration in gilts. Theriogenology  53: 1389– 1396. Google Scholar CrossRef Search ADS PubMed  Almlid T. Stavne S. E. Johnson L. A. 1987. Fertility evaluation of the straw freezing technique for boar semen under practical artificial insemination conditions. Reprod. Domest. Anim.  22: 193– 202. Google Scholar CrossRef Search ADS   Bailey J. L. Lessard C. Jacques J. Brèque C. Dobrinski I. Zeng W. Galantino-Homer H. L. 2008. Cryopreservation of boar semen and its future importance to the industry. Theriogenology  70: 1251– 1259. Google Scholar CrossRef Search ADS PubMed  Bertani G. R. Scheid I. R. Fialho F. B. Rubin M. I. B. Wentz I. Gonçalves P. B. D. 1997. Effect of the time of artificial insemination with frozen-thawed or fresh semen on embryo viability and early pregnancy rate in gilts. Theriogenology  48: 933– 945. Google Scholar CrossRef Search ADS PubMed  Bortolozzo F. P. Uemoto D. A. Bennemann P. E. Pozzobon M. C. Castagna C. D. Peixoto C. H. Barioni W.Jr Wentz I. 2005. Influence of time of insemination relative to ovulation and frequency of insemination on gilt fertility. Theriogenology  64: 1956– 1962. Google Scholar CrossRef Search ADS PubMed  Bwanga C. O. Hofmo P. O. Grevle I. S. Einarsson S. Rodriguez-Martinez H. 1991. In-vivo fertilizing capacity of deep frozen boar semen packaged in plastic bags and maxi-straws. J. Vet. Med. A  38: 281– 286. Google Scholar CrossRef Search ADS   Casas I. Sancho S. Briz M. Pinart E. Bussalleu E. Yeste M. Bonet S. 2010. Fertility after post-cervical artificial insemination with cryopreserved sperm from boar ejaculates of good and poor freezability. Anim. Reprod. Sci.  118: 69– 76. Google Scholar CrossRef Search ADS PubMed  Colenbrander B. Gadella B. Feitsma H. Woelders H. 2000. Semen quality assessment, present and future. In: IVth Int. Conf. Boar Semen Preserv., Beltsville, MD. p. 35– 42. Collins E. D. Flowers W. L. Shanks R. D. Miller D. J. 2008. Porcine sperm zona binding ability as an indicator of fertility. Anim. Reprod. Sci.  104: 69– 82. Google Scholar CrossRef Search ADS PubMed  Cremades T. Roca J. Rodriguez-Martinez H. Abaigar T. Vazquez J. M. Martinez E. A. 2005. Kinematic changes during the cryopreservation of boar spermatozoa. J. Androl.  26: 610– 618. Google Scholar CrossRef Search ADS PubMed  Didion B. A. Braun G. D. Duggan M. V. 2013. Field fertility of frozen boar semen: A retrospective report comprising over 2600 AI services spanning a four year period. Anim. Reprod. Sci.  137: 189– 196. Google Scholar CrossRef Search ADS   Dziuk P. J 1970. Estimation of optimum time for insemination of gilts and ewes by double-mating at certain times relative to ovulation. J. Reprod. Fertil.  22: 277– 282. Google Scholar CrossRef Search ADS PubMed  Dziuk P. J 1996. Factors that influence the proportion of offspring sired by a male following heterospermic insemination. Anim. Reprod. Sci.  43: 65– 88. Google Scholar CrossRef Search ADS   Eriksson B. M. Rodriguez-Martinez H. 2000. Deep-freezing of boar semen in plastic film “cochettes.” J. Vet. Med. A  47: 89– 97. Google Scholar CrossRef Search ADS   Flowers W. L 1997. Management of boars for efficient semen production. J. Reprod. Fertil. Suppl.  52: 67– 78. Google Scholar PubMed  Flowers W. L. Esbenshade K. L. 1993. Optimizing management of natural and artificial matings in swine. J. Reprod. Fertil. Suppl.  48: 217– 228. Google Scholar PubMed  Foxcroft G. R. Dyck M. K. Ruiz-Sanchez A. Novak S. Dixon W. T. 2008. Identifying useable semen. Theriogenology  70: 1324– 1336. Google Scholar CrossRef Search ADS PubMed  Gadea J 2005. Sperm factors related to in-vitro and in-vivo porcine fertility. Theriogenology  63: 431– 444. Google Scholar CrossRef Search ADS PubMed  Gadea J. Selles E. Marco M. A. 2004. The predictive value of porcine seminal parameters on fertility outcome under commercial conditions. Reprod. Domest. Anim.  39: 303– 308. Google Scholar CrossRef Search ADS PubMed  Gil M. A. Almiñana C. Roca J. Vázquez J. M. Martínez E. A. 2008. Boar semen variability and its effects on IVF efficiency. Theriogenology  70: 1260– 1268. Google Scholar CrossRef Search ADS PubMed  Hernández M. Roca J. Gil M. A. Vázquez J. M. Martínez E. A. 2007. Adjustments on the cryopreservation conditions reduce the incidence of boar ejaculates with poor sperm freezability. Theriogenology  67: 1436– 1445. Google Scholar CrossRef Search ADS PubMed  Hofmo P. O. Grevle I. S. 2000. Development and commercial use of frozen use of frozen boar semen in Norway. In: Johnson L. A. Guthrie H. D. editors, Boar semen preservation IV. Allen, Lawrence, KS. p. 71– 86. Johnson L. A 1985. Fertility results when using frozen boar spermatozoa: 1970 to 1985. In: Deep freezing boar semen: Proc.  First Int. Conf. Deep Freezing Boar Semen, Uppsala, Sweden. p. 192– 222. Johnson L. A. Weitze K. F. Fiser P. Maxwell W. M. 2000. Storage of boar semen. Anim. Reprod. Sci.  62: 143– 172. Google Scholar CrossRef Search ADS PubMed  Juarez J. D. Parrilla I. Vazquez J. M. Martinez E. A. Roca J. 2011. Boar semen can tolerate rapid cooling rates prior to freezing. Reprod. Fertil. Dev.  23: 681– 690. Google Scholar CrossRef Search ADS PubMed  Knox R. Levis D. Safranski T. Singleton W. 2008. An update on North American boar stud practices. Theriogenology  70: 1202– 1208. Google Scholar CrossRef Search ADS PubMed  Knox R. V 2011. The current value of frozen–thawed boar semen for commercial companies. Reprod. Domest. Anim.  46: 4– 6. Google Scholar CrossRef Search ADS PubMed  Knox R. V. Althouse G. C. 1999. Visualizing the reproductive tract of the female pig using real-time ultrasonography. Swine Health Prod.  7: 207– 215. Knox R. V. Miller G. M. Willenburg K. L. Rodriguez-Zas S. L. 2002. Effect of frequency of boar exposure and adjusted mating times on measures of reproductive performance in weaned sows. J. Anim. Sci.  80: 892– 899. Google Scholar CrossRef Search ADS PubMed  Koketsu Y 2005. Six component intervals of nonproductive days by breeding-female pigs on commercial farms. J. Anim. Sci.  83: 1406– 1412. Google Scholar CrossRef Search ADS PubMed  Lamberson W. R. Safranski T. J. 2000. A model for economic comparison of swine insemination programs. Theriogenology  54: 799– 807. Google Scholar CrossRef Search ADS PubMed  Medrano A. Holt W. V. Watson P. F. 2009. Controlled freezing studies on boar sperm cryopreservation. Andrologia  41: 246– 250. Google Scholar CrossRef Search ADS PubMed  Meyers S. McDaneld T. Swist S. Marron B. Steffen D. O'Toole D. O'Connell J. Beever J. Sonstegard T. Smith T. 2010. A deletion mutation in bovine SLC4A2 is associated with osteopetrosis in Red Angus cattle. BMC Genomics  11: 337. Google Scholar CrossRef Search ADS PubMed  Nissen A. K. Soede N. M. Hyttel P. Schmidt M. D'Hoore L. 1997. The influence of time of insemination relative to time of ovulation on farrowing frequency and litter size in sows, as investigated by ultrasonography. Theriogenology  47: 1571– 1582. Google Scholar CrossRef Search ADS PubMed  Pelaez J. Breininger E. Alegre B. Pena F. J. Dominquez J. C. 2006. In-vitro evaluation in the quality of fertilizing capacity of boar spermatozoa in 0.25 mL straws. Reprod. Dom. Anim.  41: 153– 161. Google Scholar CrossRef Search ADS   Purdy P. H 2008. Swine gene banking: A quality control perspective on collection, and analysis of samples for a national repository. Theriogenology  70: 1304– 1309. Google Scholar CrossRef Search ADS PubMed  Pursel V. G. Johnson L. A. 1975. Freezing of boar spermatozoa: Fertilizing capacity with concentrated semen and a new thawing procedure. J. Anim. Sci.  40: 99– 102. Google Scholar CrossRef Search ADS PubMed  Reicks D. L. Levis D. G. 2008. Fertility of semen used in commercial production and the impact of sperm numbers and bacterial counts. Theriogenology  70: 1377– 1379. Google Scholar CrossRef Search ADS PubMed  Ringwelski J. M. Beever J. E. Knox R. V. 2013. Effect of interval between inseminations when using frozen-thawed boar sperm on fertility and fetal paternity in mature gilts. Anim. Reprod. Sci.  137: 197– 204. Google Scholar CrossRef Search ADS PubMed  Roca J. Hernández M. Carvajal G. Vázquez J. M. Martínez E. A. 2006. Factors influencing boar sperm cryosurvival. J. Anim. Sci.  84: 2692– 2699. Google Scholar CrossRef Search ADS PubMed  Soede N. M. Kemp B. 1995. Timing of AI and ovulation in sows. Reprod. Domest. Anim.  31: 201– 207. Google Scholar CrossRef Search ADS   Soede N. M. Kemp B. 1997. Expression of oestrus and timing of ovulation in pigs. J. Reprod. Fertil. Suppl.  52: 91– 103. Google Scholar PubMed  Soede N. M. Wetzels C. C. Zondag W. de Koning M. A. Kemp B. 1995. Effects of time of insemination relative to ovulation, as determined by ultrasonography, on fertilization rate and accessory sperm count in sows. J. Reprod. Fertil.  104: 99– 106. Google Scholar CrossRef Search ADS PubMed  Spencer K. W. Purdy P. H. Blackburn H. D. Spiller S. F. Stewart T. S. Knox R. V. 2010. Effect of number of motile, frozen-thawed boar sperm and number of fixed-time inseminations on fertility in estrous-synchronized gilts. Anim. Reprod. Sci.  121: 259– 266. Google Scholar CrossRef Search ADS PubMed  Thurston L. M. Holt W. V. Watson P. F. 2003. Post-thaw functional status of boar spermatozoa cryopreserved using three controlled rate freezers: A comparison. Theriogenology  60: 101– 113. Google Scholar CrossRef Search ADS PubMed  Waberski D. Weitze K. F. Gleumes T. Schwartz M. Willmen T. Petzoldt R. 1994. Effect of time of insemination relative to ovulation on fertility with liquid and frozen boar semen. Theriogenology  42: 831– 840. Google Scholar CrossRef Search ADS PubMed  Watson P. F 2000. The causes of reduced fertility with cryopreserved semen. Anim. Reprod. Sci.  60–61: 481– 492. Google Scholar CrossRef Search ADS PubMed  Watson P. F. Behan J. R. 2002. Intrauterine insemination of sows with reduced sperm numbers: Results of a commercially based field trial. Theriogenology  57: 1683– 1693. Google Scholar CrossRef Search ADS PubMed  Weitze K. F 2000. Update on the worldwide application of swine AI. In: Johnson L. A. Guthrie H. D. editors, IVth International Conference on Boar Semen Preservation. Allen Press, Lawrence, KS. p. 141– 146. Weitze K. F. Wagner-Rietschel H. Waberski D. Richter L. Krieter J. 1994. The onset of heat after weaning, heat duration, and ovulation as major factors in AI timing in sows. Reprod. Domest. Anim.  29: 433– 443. Google Scholar CrossRef Search ADS   Woelders H. Matthijs A. Besten M. D. 1995. Boar variation in “freezability” of the semen. Reprod. Domest. Anim.  31: 153– 159. Google Scholar CrossRef Search ADS   Yamaguchi S. Funahashi H. Muramaki T. 2009. Improved fertility in gilts and sows after artificial insemination of frozen-thawed boar semen by supplementation of semen extender with caffeine and CaCl2. J. Reprod. Dev.  55: 645– 649. Google Scholar CrossRef Search ADS PubMed  Footnotes 1 We extend our sincere thanks to Phil Purdy of the USDA and Genetiporc for their assistance in this experiment. We also thank J. Ringwelski, M. Bojko, S. Storms, B. Marron, and J. Beever for their technical assistance, as well as the University of Illinois Swine Research Center staff, G. Bressner and R. Alan, for the management and care of the animals used in this experiment. This project was supported by Agriculture and Food Research Initiative Competitive Grant no. 2010-85122-20620 from the USDA National Institute of Food and Agriculture. American Society of Animal Science