Genetic associations between behavioral traits and direct-social effects of growth rate in pigs 1

Transcription

1 Published January 0, 015 Genetic associations between behavioral traits and direct-social effects of growth rate in pigs 1 L. Canario,* S. P. Turner, R. Roehe, N. Lundeheim,* R. B. D Eath, A. B. Lawrence, E. Knol, R. Bergsma, and L. Rydhmer* *Department of Animal Breeding and Genetics, Swedish University of Agricultural Sciences, Box 703, Uppsala, Sweden; Department of Animal Genetics, UMR 0444, LGC, INRA, 3136 Castanet-Tolosan, France; Scottish Agricultural College, West Mains Road, Edinburgh EH9 3JG, United Kingdom; and Institute for Pig Genetics, P.O. Box 43, 6640 AA Beuningen, the Netherlands ABSTRACT: This study examined the behavioral consequences of selecting pigs using a social genetic model for growth. Calculations enable each member of a group of pigs to be given a direct breeding value (DBV) and a social breeding value (SBV), which can be summarized into a total breeding value (TBV) for growth. Selection for growth TBV could affect animal behavior because social effects account for within-group interactions. Data were recorded from 96 groups of Yorkshire and Yorkshire Landrace pigs in a nucleus herd. Each group contained 15 pigs fed ad libitum from feeders; the space allowance was 0.85 m²/pig. Average daily gain was quantified from 35 to 100 kg of BW. Fighting and bullying activity at mixing (period 1), lying frequency 3 wk after mixing (period ), and counts of skin lesions in periods 1 and were recorded. The DBV for these traits were estimated with a classic animal model. We simulated different correlations between the direct genetic effect and the social genetic effect on growth rate (r DS ), components that respectively determine a pig s genetic capacity to grow and its genetic influence on growth of group mates: r DS was successively assumed to be 0 and ±0.1, ±0.0, ±0.9, and ±0.58. Finally, the correlations between DBV, SBV, and TBV for ADG, as well as the Key words: activity, aggressiveness, pig, skin lesions, social interactions, social model 01 American Society of Animal Science. All rights reserved. J. Anim. Sci : doi:10.57/jas INTRODUCTION Social genetic models play an important role in the genetic improvement of pig growth because a sizeable 1 The authors thank the pig breeder; the technicians, U. Schmidt and B. Olsson; and S. Ison, M. Jack, and M. Farish, who analyzed the video tapes. Corresponding author: laurianne.canario@toulouse.inra.fr Received April 19, 01. Accepted August 8, DBV for behavior and skin lesions, were calculated and tested for a level of significance at P < The gradient from negative to positive values of r DS refers to a progressive path running from genetic antagonism to genetic mutualism for growth. If rds in the population truly ranged between 0.58 and 0.0, correlations for TBV for ADG with DBV for fighting and bullying progressively increased with rds. Consequently, if rds was low (between 0.1 and +0.1) or positive (>+0.1), pigs with high TBV for ADG had higher DBV for bullying other pigs in the group and for fighting than pigs with lower TBV for ADG. Pigs with high TBV for ADG did not differ from other pigs in their DBV for lesions to the anterior part of the body, but they had a lower DBV for posterior lesions, whereas in period, they had higher DBV for posterior lesions and lower DBV for lying. Under genetic mutualism for growth and in housing conditions similar to those in the present study, selection for growth TBV would promote the rapid establishment of the dominance relationships, with more aggressive contests among group mates at mixing. Pigs would subsequently be more active but, judging by skin lesions, less willing to fight in a more stable social situation. proportion of the genetic variation in growth rate is modulated by social interactions with group mates that determine feed intake and global activity (Bijma et al., 007b; Bijma, 011). In growing-finishing pigs, social interactions explain two-thirds of the genetic variation in growth rate (Bergsma et al., 008). These models incorporate both a direct genetic effect (i.e., the influence of a pig s genotype on its own growth rate) and a social genetic effect (i.e., the influence of a pig s geno-

2 Behavior and direct social effects of growth 4707 type on growth rates in group mates). Because the social genetic effect accounts for within-group interactions, we hypothesized that this type of selection affects social behavior. Aggressiveness in pigs is known to be repeatable over time and across different situations (Jensen et al., 00; Janczak et al., 003; D Eath, 004); it also is known to be influenced in part by genetics (Løvendahl et al., 005; Turner et al., 008). Decisions to engage in aggressive behavior might be made by pigs according to the relative costs and benefits of behavior in different aggressive strategies, which will vary depending on, among other factors, resource scarcity and the strategies played by others (Enquist and Leimar, 1983). Similarly, the correlation between the direct genetic effect and the social genetic effect on growth (r DS ) could vary, depending on the social dynamics in play (Bijma et al., 007b), from negative (i.e., genetic antagonism for growth) to neutral (independence of a pig s growth from that of its group mates) to positive values (genetic mutualism in respect of growth). The aim of the current study was to estimate genetic associations between growth and traits with an effect on welfare under different values for r DS. Given that expressions of aggressiveness observed in unstable and stable situations might be inconsistent, we examined records of behavior and skin lesions obtained shortly after mixing (when entering the growing-finishing facility) and records obtained 3 wk later. MATERIALS AND METHODS Ethical Consideration The procedures employed in this study were approved by the Animal Experiments Committee at the Scottish Agricultural College, and permission to conduct the study was granted by Sweden s animal protection authority (Djurskyddsmyndigheten). Governmental licensing was not required. Animals, Housing, and Growth Performance The pigs studied were part of a Swedish dam line nucleus herd. The piglets (750 females, 479 intact males, and 10 castrates) were born to sows raised in individual farrowing pens without crates and were the progeny of 39 dams and 8 boars. Two genetic lines were studied: pure Yorkshire (n = 809) and crossbred Yorkshire Landrace pigs (n = 630). All castrates were crossbred, most intact males were purebred, and the females were evenly distributed over both lines. The piglets were weaned at approximately 5 wk but remained with their littermates until approximately 10 wk. At 10 wk, the pigs were mixed into novel social groups and moved into the growing-finishing facilities. Each mixed group included 15 animals (3 pigs from each of 5 litters) of the same sex and same breed. A total of 96 groups were formed from 31 litters deriving from 14 farrowing batches. Pigs were grouped so that animals of similar BW were kept together. The groups were placed in partly slatted pens (9% slats, 71% lightly bedded solid flooring) with 0.85 m²/pig total floor area. They were fed ad libitum from single-space feeders in each pen, and they had permanent access to 1 nipple drinker. The average ambient temperature was 19.4 C (SD =.9). Among the 1,439 pigs studied,.6% died during the finishing period, and data from these pigs were not included in the analyses. Pigs were individually weighed before mixing at 10 wk (SD = 4.3 d) and again 85 d later (SD = 5.0 d). The average BW were 8 (SD = 5.5) and 106 kg (SD = 11.9). The data were collected between October 005 and January 007. Further details of the experimental design and protocol are provided by Turner et al. (009). Behavior Records In 69 of the 96 groups, pig behavior was video recorded during the first 4 h after mixing (period 1) and 3 wk later for another 4 h (period ). Three observers were involved in the video analyses; the degree of interobserver agreement for traits studied was high (r = 0.83 for the evaluation from three 1-h samples of data, P < 0.001). The total duration and frequency of reciprocal fighting and the infliction and receipt of non-reciprocal aggression, referred to as bullying, were recorded by continuous observations during period 1 (Turner et al., 006a). Reciprocal fighting was defined as a fight lasting 1 s in which two pigs were pushing, head knocking, or biting each other. Receipt of bullying was recorded when a pig was attacked but did not retaliate. An animal was identified as the winner when its opponent turned away or ran away. If the animals stopped fighting without a clear winner or the fight was interrupted by a third pig, the outcome was recorded as unclear. During period, scan samples were performed at 1-h intervals for a 4-h period to record the number of times each pig was observed lying (either laterally or ventrally). On average, animals involved in aggressive encounters initiated 4 (SD = 4) fights and inflicted 4 (SD = 6) bullying acts in period 1. During period, pigs were observed lying on the majority of occasions [19 (SD = )/4 scans]. Count of Skin Lesions The number of fresh skin lesions was recorded by a single observer immediately before and 4 h after mixing (period 1). Fresh skin lesions were characterized by a bright red color or apparently recent scabs. Skin lesions

3 4708 Canario et al. on anterior (head, neck, shoulders, and anterior legs) and posterior (rump, hind legs, and tail) parts of the body were recorded separately as anterior lesions and posterior lesions, respectively. The number of skin lesions resulting from mixing was calculated by subtracting the premixing count from the postmixing count. At 18. d (SD =.9) after mixing (period ), skin lesions that were judged to be less than 4 h old were again counted by the same observer. On average, pigs showed 19 (SD = 18) anterior lesions and 5 (SD = 6) posterior lesions in period 1 and 10 (SD = 5) anterior lesions and 5 (SD = 4) posterior lesions in period. Skin lesions located on the anterior part of the body are associated with a high level posterior part of the body are the result mainly of being bullied while being chased by other pigs or attempting to Genetic Analysis of Behavior and Skin Lesions Descriptive data on behavior and skin lesions are given in Table 1. Preliminary analyses using SAS software (SAS Inst. Inc., Cary, NC) showed that their distribution was substantially positively skewed. To decrease formation was applied. Traits were analyzed with the linear model y =Xb + Z a + Wc + Vg + e, [1] D D Table 1. Number of groups, means (SD) of lesion scores, and aggressive and lying behavioral traits dur- mixing (period ) Trait No. of groups No. of pigs involved per group Observations per pig involved Period 1 Anterior lesions 1, (1) 1 (7) Posterior lesions (3) 7 (3) 69 1 () 5 () () 4 (1) () 3 () Bullying acts delivered () 5 () Bullying acts received (1) 4 () Period Anterior lesions (1) 10 () Posterior lesions () 5 (1) Lying frequency 7 15 (0) 19 (1) 1 The number of lesions in period 1 is the difference of lesions before and after mixing, whereas in period, this number is based on a single measurement of fresh lesions. In the 96 studied groups, on average, 14 (SD = 1) of the 15 pigs had anterior lesions. These 14 pigs had, on average, 1 lesions (SD = 7). where y is the observation for each animal and X, Z, W, V are incidence matrices and b fects, including the breed of the pig (Yorkshire or Yorkshire Landrace), a batch effect according to date at vector c is a vector of random pen effects (i.e., physical environment in which several groups can be raised) withc~ N(0, Iσ c ); g is a vector of random group effects with g ~ N(0, Iσ g ), and e is a vector of residuals, with e~ N(0, Iσ e ). The matrix Z is an incidence matrix D for the direct genetic effects. The vector a D refers to random direct genetic effects related to the direct estimated breeding value (DBV), which are assumed to follow a normal distributiona~ N(0, Aσ A ), with A being the relationship matrix. Estimating Breeding Values for Growth Rate Growth rate was measured as ADG between mixing at 8 kg to ultrasonic back fat measurement at 106 kg. group mates on individual growth, the model adopted for the analysis of ADG was y D D S S =Xb+Z a +Za +Wc+Vg+e, [] fects and Za D D a D indicating the direct genetic effects related to the DBV. The incidence matrix Z S corresponds to social genetic effects; a S corresponds to the summed social genetic effects applied from 1 animal from a group of n individuals on its group mates and is related to the social breeding value (SBV). The DBV and the SBV represent the heritable components of direct and social genetic genetic effects originating from multiple individuals, the breeding values of individuals can be summarized in a total breeding value (TBV) expressing an individual s value for genetic improvement (Bijma et al., 007b). The DBV is expressed in the individual s own phenotype, and the SBV is expressed in the phenotypes of its group mates. So the TBVi accounting for direct and social effects of individual i was calculated as TBVi = titative genetic equation TBVi = ad, i + 14aS, i with the covariance structure of the additive genetic effects: ad σa σ D ad, as var A a = σ S ad, a σ S as where σ is the direct genetic variance, σ is the social ad as genetic variance, and σ is the term for covariance ad, as,

4 Behavior and direct social effects of growth 4709 between direct and social genetic effects. It is linked to r DS, the correlation between the direct genetic effect and the social genetic effect on ADG; A is the relationship matrix between pigs. In the Swedish breeding program, pigs are weighed netic evaluation for growth. In the present study, ADG was measured from mixing (approximately 35 d) to the end of the study period. To estimate the direct and social genetic effects accurately, a larger number of animals than were available in the present population is required. Hence, variance components for the direct and the social genetic effects on ADG estimated from a Dutch pig population (n = 14,03 pigs) were used: σ σ ad = and as (Bergsma et al., 008). Although the variance for social genetic effects is low, it is important to realize that the contribution of social effects depends on the number of group mates. The total genetic variance available for response to selection, σ TBV is calculated as follows: σtbv = σ A + ( n 1) σ, ( 1) D AD A + n σ S A (Bijma et S cial group dynamics, in which the pigs displayed negative (genetic antagonism), neutral, or positive (genetic mutualism) interactions. To obtain a range from a high antagonism between the direct genetic effect and the social genetic effect for growth (r DS < 0) to a high positive correlation between the genetic components, r DS was successively assumed to be 0 and ±0.1, ±0.0, ±0.9, and ±0.58. A negative genetic correlation (r DS < 0) implies genetic antagonism between the growth of 1 individual and the growth of its group mates that might be the result of non-cooperative behavior. In contrast, a positive genetic correlation implies that the growth of 1 individual and the growth of its group mates are mu- ences. The underlying behaviors relate to the absence of competition for food or contagiously limited and limiting growth. Correlations between BV for Growth Rate and BV for Behavior and Skin Lesions Components of genetic variance in behavior and skin lesions were estimated using the ASReml software (Gilmour et al., 004). The pedigree used for the analyses included information about grandparents and greatgrandparents. Correlations between the DBV, SBV, and TBV for ADG and the DBV for behavior and skin lesions were calculated. These correlations provide an approximation of the genetic correlations that could have been estimated with a larger data set. RESULTS Descriptive data on the behavioral characteristics and skin lesions at group level are presented in Table 1. havior at mixing, with, on average, 1 pigs/group initi- exhibited more anterior lesions than posterior lesions, even in period. Social Characteristics of Pigs with High Growth DBV Estimates of correlations between DBV, SBV, and TBV for ADG and DBV for behavior and skin lesions obtained with different values of r DS are presented in more, they received fewer posterior lesions than pigs with lower DBV for ADG, but the number of anterior lesions did not differ with DBV for ADG (P ure 3). In period, the DBV for ADG was not associated with skin lesions, but pigs with a high DBV for ADG Social Characteristics of Pigs with High Growth SBV or High Growth TBV Variation of DBV for behavior and skin lesions with SBV and TBV for ADG are described below according to 3 distinct patterns of group dynamics (i.e., genetic antagonism, neutral, or genetic mutualism for growth with group mates). The correlations between the DBV for behavioral characteristics and skin lesions with TBV for ADG were ADG ranged from null to moderate values. The strongest correlations and greatest variation with r DS were obtained between the DBV for bullying traits and growth TBV. Case of Genetic Antagonism for Growth (r DS <. In cases of genetic antagonism between the DBV and SBV for ADG, pigs with a high SBV for ADG ing. They also had a lower DBV for anterior lesions and a higher DBV for posterior lesions than pigs with a low were correlated with the SBV for ADG only where the genetic antagonism was very high (r DS ever, under these circumstances pigs with a high SBV pigs with high SBV for ADG (except when r DS

5 4710 Canario et al. Figure 1. Correlations between direct breeding values for fighting activity in period 1 and direct (diamonds), social (triangles), and total (circles) breeding values for growth rate depending on various assumed correlations between the direct and the social genetic effect on growth rate (horizontal axis: r DS < 0 refers to genetic antagonism and r DS > 0 refers to genetic mutualism between group mates). Solid symbols indicate that the correlations with direct, social, and total breeding values for growth rate differed from zero at P < Nonsignificant results are indicated by open symbols. The correlations between TBV for ADG and DBV for fight initiation and fights won were null when r DS was highly negative (r DS = 0.58) but positive for higher values of r DS. Moreover, the TBV for ADG and DBV for fights lost (period 1) and for lying (period ) were independent when r DS = 0.58 (Figure 4). The correlation between TBV for ADG with DBV for posterior lesions in period was positive, except for r DS = 0.58, where it was null. The correlation between TBV for ADG and DBV for receipt of bullying differed with r DS ; it was positive when r DS = 0.58 but negative for more moderate values of r DS. Case of Neutral Genetic Interactions for Growth (r DS 0.1 and r DS 0.1). The assumed range of true r DS encompasses that estimated in the Swedish Yorkshire population (r DS = 0.07; Canario et al., 010). Pigs with a high SBV for ADG also had a high DBV for losing fights and for both bullying and being bullied in period 1; they also had a positive DBV for lying in period, and they displayed high DBV for posterior lesions in periods 1 and but lower DBV for anterior lesions than pigs with low SBV for ADG. The correlations between SBV for ADG and DBV for fight initiation and success were not significant. Correlations of TBV for ADG with DBV for fighting and bullying others in period 1 and with DBV for posterior lesions in period were both positive; however, with DBV for the receipt of bullying and posterior lesions in period 1 and lying in period, they were negative. Case of Genetic Mutualism for Growth (r DS > 0.1). The scenario in which it is assumed that the true correlation is r DS = 0.0 is equivalent to that estimated in the Dutch population (Bergsma et al., 008). In period 1, pigs with a high SBV for ADG had higher DBV for initiating, winning, and losing fights and for inflicting bullying than pigs with a low SBV for ADG; however, except when r DS = 0.58, they did not differ from other pigs in their DBV for lesions in period 1 or in their DBV for receiving bullying. In period, pigs with a high SBV for ADG had a high DBV for posterior lesions. In addition, they had a high DBV for anterior lesions in period when r DS > 0.0. Pigs with high SBV for ADG did not differ from other pigs in their DBV for lying frequency unless r DS = 0.58; however, when r DS = 0.58, the correlations between DBV for behavior and skin lesions and DBV, SBV, and TBV for ADG were similar; pigs with high breeding value (BV) for ADG (compared with those with lower BV) were more active, initiated and won more fights, inflicted more bullying, and received less bullying. Correlations with DBV for ADG vs. TBV for ADG The correlations obtained with DBV for ADG and (for comparison) TBV for ADG were similar across the different correlations between the direct and social genetic effects on ADG (r DS ), other than in a situation of extreme genetic antagonism for ADG (r DS = 0.58). The DBV for fighting and posterior lesions in period 1 and the DBV for lying in period were then associated with DBV for ADG but not TBV for ADG. Similarly, the DBV for anterior lesions in period 1 was associated with TBV for ADG but not DBV for ADG. Last, when r DS = 0.58, pigs receiving more bullying had a low DBV for

6 Behavior and direct social effects of growth 4711 Figure. Correlations between direct breeding values for bullying activity in period 1 and direct (diamonds), social (triangles), and total (circles) breeding values for growth rate depending on various assumed correlations between the direct and the social genetic effect on growth rate (horizontal axis: r DS < 0 refers to genetic antagonism, and r DS > 0 refers to genetic mutualism between group mates). Solid symbols indicate that the correlations with direct, social, and total breeding values for growth rate differed from zero at P < Nonsignificant results are indicated by open symbols. Figure 3. Correlations between direct breeding values for skin lesions and direct (diamonds), social (triangles), and total (circles) breeding values for growth rate depending on various assumed correlations between the direct and the social genetic effect on growth rate (horizontal axis: r DS < 0 refers to genetic antagonism, and r DS > 0 refers to genetic mutualism between group mates). Solid symbols indicate that the correlations with direct, social, and total breeding values for growth rate differed from zero at P < Nonsignificant results are indicated by open symbols.

7 471 Canario et al. Figure 4. Correlations between direct breeding values for lying frequency observed in period and direct (diamonds), social (triangles), and total (circles) breeding values for growth rate depending on various assumed correlations between the direct and the social genetic effect on growth rate (horizontal axis: r DS < 0 refers to genetic antagonism, and r DS > 0 refers to genetic mutualism between group mates). Solid symbols indicate that the correlations with direct, social, and total breeding values for growth rate differed from zero at P < Nonsignificant results are indicated by open symbols. H0: correlation not significantly different from zero. ADG but a high TBV for ADG. DISCUSSION Genetics of Pig Aggressiveness among Group Mates Establishing a dominance hierarchy among new group mates at mixing minimizes aggressiveness and energy expenditure over the long term (Edwards, 1991), especially when pigs are raised in a competitive environment in which resources are limited (Andersen et al., 000). Pigs closely matched in age and BW fight to establish or reinforce their social position (Enquist and Leimar, 1983; Rushen, 1987; Andersen et al., 000) and, as a result, allocate less energy to growth than they could (Stookey and Gonyou, 1994). In addition to reciprocated fighting, pigs use bullying to maintain social relationships (Beilharz and Cox, 1967). Genetic parameters for the number of skin lesions and for fighting and bullying activity indicate that the categories of traits (behavior and lesions) are moderately heritable and positively correlated (Turner et al., 009). After mixing the intensity of social contests decreases over time until, eventually, social relationships stabilize (Fraser et al., 1995). Nonetheless, an ongoing, lower level of aggression persists even in stable social groups, and it results in the continued accumulation of a small number of fresh lesions many weeks after mixing (Meese and Ewbank, 1973; Turner et al., 000). In our study population, the aggressiveness of pigs at mixing was genetically correlated with lesions 3 wk after mixing (Turner et al., 009) but not with global activity (D Eath et al., 009). Methodology Direct selection for growth has been shown to increase aggressiveness and decrease activity in poultry (Bokkers and Koene, 003; Kjaer and Mench, 003). In fish, the outcome of selection for growth is a function of genotype-environment interactions (GxE) that depend on the availability of food (Ruzzante and Doyle, 1991). In pigs, the influence of selection for growth on aggressiveness and activity is poorly documented; however, Barnett et al. (1988) inferred from a comparison of genotypes that selection for high growth rate results in increased general activity and a decreased motivation to participate in social interactions. Using classic animal models, Jonsson and Jorgensen (1989) reported a positive genetic correlation between aggressiveness and growth, whereas Turner et al. (006b) found the traits to be genetically independent. Several decades ago, Griffing (1967, 1977) showed that, given the substantial effect of social interactions on individual performance, quantitative traits should be analyzed with an extended model including within-group interactions. Drawing on the recommendations of Griffing (1967, 1977), Muir (005) developed a social model that was further developed by Bijma et al. (007b) to estimate social genetic parameters. Several authors have been concerned that if the social genetic effect is not taken into account in genetic evaluations of growth rate, biases of additive genetic and residual variances will result (e.g., Cappa and Cantet, 006; Bijma et al., 007b). Chen et al. (009) estimated an empirical gain in response to selection averaging 1 g of ADG per unit of SBV for the top 10% of boars (SD =.4 g; n = 4 lines). Thus, the inclusion of a social genetic effect in the model might provide a more optimal response to selection for growth (Griffing, 1967, 1977; Muir, 005; Bijma et al., 007a). Here we estimated correlations between direct and social breeding values for growth and direct breeding values for behavior and skin lesions. It should be noted that these correlations only approximate to the genetic correlation between growth over the fattening period and welfare traits measured in the first 3 wk following mixing. Because breeding values are regressed to the mean, depending on their accuracy, we observed low values for correlations. The significance values of the correlations also could be biased because the breeding values were estimated without error. The estimates for the social genetic effect are sensitive to environmental conditions (Bijma et al., 007a,b; Bijma, 010; Canario et al., 010; Bijma, 011). It can be assumed that the behavioral response to selection for growth would be quite different if pen size, group size, and thus density, as well as food availability, were varied. For instance, GxE can arise when different feeding strategies are used between herds (several pigs at a trough vs. 1 pig at the electronic feeder) even when all pigs are fed ad libitum; however, the population studied was reared in farming conditions similar to those from many European pig selection companies. Selective breeding and commercial environments differ according to these factors. The existence of GxE might influence the effect of selection on the perfor-

8 Behavior and direct social effects of growth 4713 mance of commercial pigs (Bijma, 011). The SBV values and the behaviors that underlie them could differ between the systems of production. Characteristics of Pigs with High Direct Genetic Merit for Growth Although previous studies failed to find a substantial long-term effect of postweaning mixing on growth (Greer, 1987; Spoolder et al., 000), animals that express aggressiveness toward pen mates at mixing are important for the group. If they are winners, they facilitate rapid settlement of dominance relationships (Tindsley and Lean, 1984). In the present study, the majority of pigs in a pen fought in the 4-h period after mixing. We found that pigs with high direct breeding value for ADG were socially successful. Specifically, they have higher genetic merit for both initiating and winning fights and bullying other pigs, behaviors in part genetically controlled by the same genes (r g = 0.64 in sows, Løvendahl et al., 005; r g = 0.84 in the population used in the present study, Turner et al., 009). Furthermore, these pigs have a genetic predisposition to be bullied less frequently and to display fewer lesions on the posterior part of the body. Our results on DBV for ADG agree with those obtained by bivariate classic analyses of ADG with behavior and injury traits at mixing (L. Canario, unpublished data). Taken together, the findings suggest that ongoing selection for growth might increase aggressiveness in pigs, as previously shown in broilers (Kjaer and Mench, 003). Selection for direct effects on ADG also might produce more active pigs, as suggested by Barnett et al. (1988). Characteristics of Pigs with a High Social Genetic Merit for Growth There is commercial interest in selecting pigs with both high DBV and high SBV for growth. The current study investigates for the first time the effect of such selection on aggressiveness and activity, and it shows how sensitive the resulting genetic traits are to variations in the direction and size of the correlation between direct and social genetic components for growth that might be encountered in different populations. The gradient from negative to positive values of r DS refers to a progressive path running from genetic antagonism to genetic mutualism. To date, few estimates of r DS in real populations have been published. Genetic mutualism for growth was found in a Dutch population (Bergsma et al., 008; r DS = 0.0, SE = 0.10), whereas neutral social interactions for growth were estimated in the Swedish Yorkshire population (Canario et al., 010; r DS = 0.07, SE = 0.13). Chen et al. (009) found positive to negative social interactions for growth in populations in the United States (r DS = 0.37 to 0.74). When the social environment induces a genetic antagonism between the growth of a pig and the growth of group mates (r DS < 0.1), pigs with high genetic merit for the growth of their group mates (high SBV) displayed a genetic predisposition to inflict and receive bullying, to lose fights more frequently, and to accumulate fewer anterior lesions and more posterior lesions in the 4-h period after mixing. It was interesting that these pigs, which received more aggression, also bullied others. Such bullying could benefit the group as a whole if it serves to stabilize dominance relationships. When the genetic antagonism was very strong (r DS = 0.58), pigs with a positive influence on the growth of group mates also initiated and won fewer fights. These results support the observation that dominant pigs intensify their resource defense in competitive situations to secure privileged access to food at the expense of subordinates (Mendl and Held, 001). It has been observed in large groups, where the cost of establishing dominance is greater than it is in smaller groups (Turner and Edwards, 004), that pigs minimize the duration of fights. Our results also suggest that aggressive strategies change under highly competitive conditions. Selection for growth TBV would encourage the delivery of bullying and would limit the involvement of pigs in mixing contests only when r DS is highly negative, as reflected by pigs with a lower genetic predisposition for anterior lesions and losing fights. Under these competitive conditions, individual growth is promoted by the avoidance of costly fights (Mendl and Deag, 1995; Estevez et al., 007). When the correlation between direct and social effects for growth is weak (r DS 0.1 and r DS +0.1), pigs with genes that benefit their own growth also have genes with a low effect on the growth of group mates. Under these circumstances, pigs with a high social breeding value for growth did not differ from other pigs in their fight initiation and success; however, they lost more fights than other pigs. They also displayed more posterior lesions in both unstable and stable situations. On the other hand, they displayed fewer anterior lesions. Losers of fights might not be severely penalized in their access to resources (Estevez et al., 007) compared with a situation of genetic antagonism for growth. Pigs with a beneficial effect on the growth of group mates might be more inclined to accept defeat in aggressive contests. The level of anterior lesions, which we expected to be associated with aggressive encounters between dominant pigs (Turner et al., 006a), was not a discriminating factor between pigs with more or less high growth TBV. This implies that genetic selection for growth TBV would not seriously impair welfare. Where there is genetic mutualism for growth (r DS > +0.1), pigs with genes that benefit their own growth also have genes with a beneficial effect on the growth of

9 4714 Canario et al. group mates. The opposite expression of this relationship also might be apparent, whereby genetic effects simultaneously depress an individual s growth and that of its group mates. In contrast to situations in which there was either no relationship or a negative relationship between the direct and social effects on growth, when r DS was positive, we observed that pigs with a high social breeding value for growth initiated more fights than other pigs. This finding suggests that pigs that benefited the rest of the group might have actively participated in the establishment of dominance relationships by being more challenging relative to a situation of genetic antagonism or neutral interactions for growth. These pigs also had a higher genetic merit to win fights and to inflict bullying on others when genetic mutualism increased, but they did not differ in their receipt of bullying, nor did they differ in their skin lesions within 4 h of mixing. When counted 3 wk after mixing, pigs with a high SBV had more posterior lesions than other pigs, probably indicating their involvement in both reciprocal fighting and the receipt of bullying. When interdependency between the growth of group mates is too high (high r DS ), a more aggressive strategy at mixing might be adopted. Such a strategy might be successful, but it also might result in a high level of aggressiveness at mixing and the subsequent receipt of bullying in the following weeks, which could have implications for pig welfare. This situation could reflect excessive investment in establishing social relationships at the expense of later growth. Social breeding values for growth were unrelated to general activity, although pigs with a high SBV were more active when the correlation between direct and social effects was large and positive. Characteristics of Pigs with a High Total Genetic Merit for Growth The pigs in the present study were fed ad libitum with no space restriction, which probably corresponds to a situation of no competition. In situations of both neutral and positive genetic interactions between DBV and SBV for growth, pigs with a high TBV for growth fought more frequently regardless of the outcome and inflicted more bullying but were bullied less than other pigs. They received fewer posterior lesions in period 1 but more posterior lesions in period, when they also displayed less lying behavior. Under these circumstances, selection for growth TBV would mainly make pigs more prone to bully their group mates at mixing and presumably less willing to fight and more active in more stable situations. The importance of subordinate pigs avoiding dominant ones could not be assessed. Genetic selection for TBV seems to be insufficient to counterbalance the increase in general activity resulting from direct selection for growth. This could be partly explained by the need for higher feed intake in faster-growing pigs (Labroue, 1995). The social genetic model might offer an effective way to improve performance and decrease aggressiveness among conspecifics (Rodenburg et al., 010; Bijma 011). The phenotypic expression of a pig s aggressiveness relates not only to the physical environment and previous social experience (Mendl, 1993; D Eath, 004) but also to the genetics of the focal pig and that of its conspecifics. It also is a key feature in stable social hierarchies (Hessing et al., 1994; Erhard et al., 1997). Consequently, a direct selection against aggressiveness, if considered, should be addressed with caution. It also can be questioned whether a genetic change in social behavior achieved through selection for growth would persist over several generations for pigs that can, to some extent, modulate their social strategy according to that of their group mates (Mendl and Deag, 1995; Erhard et al., 1997). Conclusions This study shows that the effect on behavior and skin lesions within 3 wk after mixing of selection for growth over the fattening period using a TBV that incorporates both direct and social effects would not lead to radical changes in aggressiveness compared to a classic approach based on direct effects alone. Where there is negligible competition for food, the selected pigs would initiate more bullying behavior immediately after mixing as they establish dominance relationships and would be at greater risk of being attacked in a stable situation than currently observed. On the basis of the effects on skin lesions and general activity, however, we conclude that animal welfare should not be excessively affected by such a selection. LITERATURE CITED Andersen, I. L., H. Andenaes, K. E. Boe, P. Jensen, and M. Bakken The effect of weight asymmetry and resource distribution on aggression in groups of unacquainted pigs. Appl. Anim. Behav. Sci. 68: Barnett, J. L., P. H. Hemsworth, G. M. Cronin, C. G. Winfield, T. H. McCallum, and E. A. Newman The effects of genotype on physiological and behavioral responses related to the welfare of pregnant pigs. Appl. Anim. Behav. Sci. 0: Beilharz, R. G., and D. F. Cox Social dominance in swine. Anim. Behav. 15: Bergsma, R., E. Kani, E. F. Knol, and P. Bijma The contribution of social effects to heritable variation in finishing traits of domestic pigs (Sus scrofa). Genetics 178: Bijma, P Multilevel selection 4: Modeling the relationship of indirect genetic effects and group size. Genetics 186: Bijma, P., 011. Breeding for social interaction, for animal welfare. In Encyclopedia of Sustainability Science and Technology. R. A. Meyers, ed. Springer, New York. doi: / Bijma, P., W. M. Muir, E. D. Ellen, J. B. Wolf, and J. A. M. van Arendonk.

10 Behavior and direct social effects of growth b. Multilevel selection : Estimating the genetic parameters determining inheritance and response to selection. Genetics 175: Bijma, P., W. M. Muir, and J. A. M. van Arendonk. 007a. Multilevel selection 1: Quantitative genetics of inheritance and response to selection. Genetics 175: Bokkers, E. A. M., and P. Koene Behavior of fast- and slow growing broilers to 1 wks of age and the physical consequences. Appl. Anim. Behav. Sci. 81:59 7. Canario, L., N. Lundeheim, and P. Bijma Pig growth is affected by social genetic effects and social litter effects that depend on group size. In Proc. 9th World Congr. Genet. Appl. Livest. Prod., Leipzig, Germany. CD-ROM Communication ID170. Cappa, E. P., and R. J. C. Cantet Bayesian estimation of direct and competition additive (co)variance in individuals mixed models. In Proc. 8 th World Congr. Genet. Appl. Livest. Prod., Belo Horizonte, MG, Brazil. CD-ROM Communication No Chen, C. Y., R. K. Johnson, S. Newman, S. D. Kachman, and L. D. van Vleck Effects of social interactions on empirical responses to selection for average daily gain of boars. J. Anim. Sci. 87: D Eath, R. B Consistency of aggressive temperament in domestic pigs: The effects of social experience and social disruption. Aggress. Behav. 30: D Eath, R. B., R. Roehe, S. P. Turner, S. H. Ison, M. Faris, C. Jack, and A. B. Lawrence Genetics of animal temperament: aggressive behavior at mixing is genetically associated with the response to handling in pigs. Animal 3: Edwards, S. A Scientific perspectives on loose housing systems for dry sows. Pig Vet. J. 8: Enquist, M., and O. Leimar Evolution of fighting behavior: Decision rules and assessment of relative strength. J. Theor. Biol. 10: Erhard, H. W., M. Mendl, and D. D. Ashley Individual aggressiveness of pigs can be measured and used to reduce aggression after mixing. Appl. Anim. Behav. Sci. 54: Estevez, I., I. L. Andersen, and E. Naevdal Group size, density and social dynamics in farm animals. Appl. Anim. Behav. Sci. 103: Fraser, D., D. L. Kramer, E. A. Pajor, and D. M. Weary Conflict and cooperation: Sociobiological principles and the behavior of pigs. Appl. Anim. Behav. Sci. 44: Gilmour, A. R., B. J. Gogel, B. R. Cullis, S. J. Welham, and R. Thompson ASReml user guide release.0. VSN Int. Ltd., Hemel Hempstead, UK. Greer, E. B Lack of effect of regular movement and of mixing groups on the performance of growing pigs. Aust. J. Exp. Agric. 7:17. Griffing, B Selection in reference to biological groups. I. Individual and group selection applied in population of unordered groups. Aust. J. Biol. Sci. 10: Griffing, B Selection for populations of interacting phenotypes. Pages in Proc. nd Int. Conf. Quant. Genet. E. Pollak, O. Kempthorne, and T. B. Bailey, ed. Iowa State Univ. Press, Ames. Hessing, M. J. C., W. G. P. Schouten, P. R. Wiepkema, and M. J. M. Tielen Implications of individual behavioural characteristics on performance in pigs. Livest. Prod. Sci. 40: Janczak, A. M., L. J. Pedersen, and M. Bakken Aggression, fearfulness and coping styles in female pigs. Appl. Anim. Behav. Sci. 81:13 8. Jensen, K. H., B. L. Nielsen, and A. N. W. Olsen. 00. 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 Foulum, Denmark. Jonsson, P., and J. N. Jorgensen Selektion in der Schweinezucht unter Berqcksichtigung des Dominanzverhaltens. Arch. Tierz. 3: Kjaer, J. B., and J. A. Mench Behavior problems associated with selection for increased production. Pages 67 8 in Poultry Genetics, Breeding and Technology. W. M. Muir and S. E. Aggrey, ed. CAB Int., Wallingford, UK. Labroue F., Facteurs de variation génétiques de la prise alimentaire chez le porc en croissance: Le point des connaissances. INRA Prod. Anim. 8: Løvendahl, P., L. H. Damgaard, B. L. Nielsen, K. Thodberg, G. Su, and L. Rydhmer Aggressive behavior of sows at mixing and maternal behavior are heritable and genetically correlated traits. Livest. Prod. Sci. 93: Meese, G. B., and R. Ewbank The establishment and nature of the dominance hierarchy in the domesticated pig. Anim. Behav. 1: Mendl, M Are aggressiveness and social status of group housed sows predictable from observations of earlier behavior. Page 87 in Proc. Br. Soc. Anim. Sci., UK. Mendl, M., and J. M. Deag How useful are the concepts of alternative strategy and coping strategy in applied studies of social behavior? Appl. Anim. Behav. Sci. 44: Mendl, M., and S. Held Living in groups: An evolutionary perspective. Pages 7 35 in Social Behavior in Farm Animals. J. J. Keeling, and H. W. Gonyou, ed. Univ. of Bristol, Bristol, UK. Muir, W. M Incorporation of competitive effects in forest tree or animal breeding programs. Genetics 170: Rodenburg, T. B., P. Bijma, E. D. Ellen, R. Bergsma, S. de Vries, J. E. Bolhuis, B. Kemp, and J. A. M. Van Arendonk Breeding amiable animals? Improving farm animal welfare by including social effects in breeding programmes. Anim. Welf. 19(S):77 8. Rushen, J A difference in weight reduces fighting when unacquainted newly weaned pigs first meet. Can. J. Anim. Sci. 67: Ruzzante, D. E., and R. W. Doyle Rapid behavioral changes in medaka, Oryzias latipes, during selection for competitive and noncompetitive growth. Evolution 45: Spoolder, H. A. M., S. A. Edwards, and S. Corning Aggression among finishing pigs following mixing in kennelled and unkennelled accommodation. Livest. Prod. Sci. 63: Stookey, J. M., and H. W. Gonyou The effects of regrouping on behavioral and production parameters in finishing swine. J. Anim. Behav. Sci. 7: Tindsley, W. E. C., and I. J. Lean Effects of weight range at allocation on production and behavior in fattening pig groups. Appl. Anim. Behav. Sci. 1:79 9. Turner, S. P., and S. A. Edwards Housing immature domestic pigs in large social groups: Implications for social organization in a hierarchical society. Appl. Anim. Behav. Sci. 87: Turner, S. P., M. Ewen, J. A. Rooke, and S. A. Edwards The effect of space allowance on performance, aggression and immune competence of growing pigs housed on straw deep-litter at different group sizes. Livest. Prod. Sci. 66: Turner, S. P., M. J. Farnworth, I. M. S. White, S. Brotherstone, M. Mendl, P. Knap, P. Penny, and A. B. Lawrence. 006a. The accumulation of skin lesions and their use as a predictor of individual aggressiveness in pigs. Appl. Anim. Behav. Sci. 96: Turner, S. P., R. Roehe, R. B. D Eath, S. H. Ison, M. Farish, M. C. Jack, N. Lundeheim, L. Rydhmer, and A. B. Lawrence Genetic validation of post-mixing injuries in pigs as an indicator of aggressiveness and the relationship with injuries under more stable social conditions. J. Anim. Sci. 87: Turner, S. P., R. Roehe, W. Mekkawy, M. J. Farnworth, P. W. Knap, and A. B. Lawrence Bayesian analysis of genetic associations of lesions and behavioral traits to identify genetic components of individual aggressiveness in pigs. Behav. Genet. 38: Turner, S. P., I. M. S. White, S. Brotherstone, M. J. Farnworth, P. W. Knap, P. Penny, M. Mendl, and A. B. Lawrence. 006b. Heritability of postmixing aggressiveness in grower-stage pigs and its relationship with production traits. Anim. Sci. 8: Weary, D. M., and D. Fraser Partial tooth-clipping of suckling pigs: Effects on neonatal competition and facial injuries. Appl. Anim. Behav. Sci. 65:1 7.