Using random regression models to estimate genetic variation in growth pattern and its association with sexual maturity of Thai native chickens.

1. Genetic (co)variances and parameters between body weights (BW) across the growth trajectory were estimated using a univariate random regression (RR) animal model. The effect of growth rates (GH) on age at first egg (AFE) and egg weight at first egg (EWFE) were explored using a series of univariate and bivariate analyses. 2. Body weights were taken from Thai native chickens at hatch day to 168 days of age. The model included interactions between age with hatch nested within year and sex as fixed effects, and random effects of direct additive genetic, direct permanent environmental, maternal genetic and maternal permanent environmental effects. All random effects were fitted as regressions to animals' age via quadratic Legendre polynomials and fitting six classes of residual variances was identified as an optimal variance structure to estimate parameters. 3. Genetic and phenotypic variances for BW increased with increasing age. Estimated heritabilities for direct additive (h2 a) and maternal genetic (h2 m) effects on BW traits ranged from 0.34 to 0.54, and 0.04 to 0.06, respectively. Estimated variance ratios for direct (c2 ape) and maternal permanent environmental (c2 mpe) effects ranged from 0.19 to 0.48 and 0.10 to 0.12, respectively. Estimated correlations between weights at different ages were high for all random effects. 4. Estimated h2 a for six GH traits ranged from 0.06 to 0.28, while for AFE and EWFE these were 0.24 and 0.16, respectively. Estimated h2 m and c2 mpe were low for GH. Estimated genetic correlations between GH and AFE ranged from -0.22 to 0.02 and, between GH and EWFE, ranged from -0.05 to 0.40. These estimates suggested that selecting high GH chickens at 28 days of age can be expected to reduce AFE and to increase EWFE.

Genetic parameters for haemoglobin levels in sows and piglets as well as sow reproductive performance and piglet survival.

Genetic parameters were estimated for haemoglobin (Hb) levels in sows and piglets as well as sow reproductive performance and piglet survival. Reproductive traits were available between 2005 and 2014 for 7857 litters from 1029 Large White and 858 Landrace sows. In 2012 and 2013, Hb levels, sow BW and sow back fat depth were measured on 348 sows with 529 litters 5 days prior to farrowing. In addition, Hb levels were available for 1127 one-day-old piglets from 383 litters (a maximum of three piglets per litter) of 277 sows with Hb levels. The average Hb levels in sows (sow Hb), their litters (litter Hb, based on average Hb of three piglets) and individual piglets (piglet Hb) were 112 ± 12.6 g/l, 103 ± 15.3 g/l and 105 ± 21.7 g/l, respectively. Heritabilities for Hb levels were 0.09 ± 0.07 for sow Hb, 0.19 ± 0.11 for litter Hb and 0.08 ± 0.05 for piglet Hb. Estimates for the permanent environment effect of sows were 0.09 ± 0.09 for sow Hb, 0.11 ± 0.12 for litter Hb and 0.12 ± 0.03 for piglet Hb. In comparison, heritabilities for both number of stillborn piglets and pre-weaning survival were lower (0.05 ± 0.01 and 0.04 ± 0.01). Sow BW had no significant heritability, while sow back fat depth was lowly heritable (0.10 ± 0.08). Positive genetic correlations were found between sow Hb and litter Hb (0.64 ± 0.47) and between litter Hb and sow back fat depth (0.71 ± 0.53). Higher litter Hb was genetically associated with lower number of stillborn piglets (-0.78 ± 0.35) and higher pre-weaning survival (0.28 ± 0.33). Negative genetic correlations between sow Hb and average piglet birth weight of the litter (-0.60 ± 0.34) and between piglet Hb and birth weight of individual piglets (-0.37 ± 0.32) indicate that selection for heavier piglets may reduce Hb levels in sows and piglets. Similarly, selection for larger litter size will reduce average piglet birth weight (rg: -0.40 ± 0.12) and pre-weaning survival (-0.57 ± 0.13) and may lead to lower litter Hb (-0.48 ± 0.27). This study shows promising first results for the use of Hb levels as a selection criterion in pig breeding programs, and selection for higher Hb levels may improve piglet survival and limit further reduction in Hb levels in sows and piglets due to selection for larger and heavier litters.

Contemporary group estimates adjusted for climatic effects provide a finer definition of the unknown environmental challenges experienced by growing pigs.

Environmental descriptors derived from mean performances of contemporary groups (CGs) are assumed to capture any known and unknown environmental challenges. The objective of this paper was to obtain a finer definition of the unknown challenges, by adjusting CG estimates for the known climatic effects of monthly maximum air temperature (MaxT), minimum air temperature (MinT) and monthly rainfall (Rain). As the unknown component could include infection challenges, these refined descriptors may help to better model varying responses of sire progeny to environmental infection challenges for the definition of disease resilience. Data were recorded from 1999 to 2013 at a piggery in south-east Queensland, Australia (n = 31,230). Firstly, CG estimates of average daily gain (ADG) and backfat (BF) were adjusted for MaxT, MinT and Rain, which were fitted as splines. In the models used to derive CG estimates for ADG, MaxT and MinT were significant variables. The models that contained these significant climatic variables had CG estimates with a lower variance compared to models without significant climatic variables. Variance component estimates were similar across all models, suggesting that these significant climatic variables accounted for some known environmental variation captured in CG estimates. No climatic variables were significant in the models used to derive the CG estimates for BF. These CG estimates were used to categorize environments. There was no observable sire by environment interaction (Sire×E) for ADG when using the environmental descriptors based on CG estimates on BF. For the environmental descriptors based on CG estimates of ADG, there was significant Sire×E only when MinT was included in the model (p = .01). Therefore, this new definition of the environment, preadjusted by MinT, increased the ability to detect Sire×E. While the unknown challenges captured in refined CG estimates need verification for infection challenges, this may provide a practical approach for the genetic improvement of disease resilience.

Evaluation of sire by environment interactions for growth rate and backfat depth using reaction norm models in pigs.

Reaction norm models for sires were used to evaluate genotype by environment interactions for lifetime average daily gain (ADG) and backfat depth (BF) in pigs recorded at 143 days. Data for ADG and BF were available for 265 103 pigs recorded in Australia between 2000 and 2010 in nine herds. The full data set and two subsets based on minimum numbers of progeny per sire of 50 and 100 were analysed. The environmental descriptor was quantified as least square means of herd-by-birth month (HBM) contemporary groups which varied from 540 to 738 g/day for ADG and from 8.2 to 13.8 mm for BF. Several models were evaluated for estimation of variance components in terms of predictive ability for sire intercepts and slopes. The accuracy of genetic parameter estimates was improved by increasing family size, fitting models with a fixed regression coefficient on the environmental descriptor instead of fixed HBM effects and heterogeneous residual variances. Significant sire by environment interactions were found for ADG but not for BF. Heritability estimates for ADG ranged from approximately 0.21 in average environments to approximately 0.30 in the most unfavourable environments. Estimates of sire intercepts and slopes varied by 98.5 g/day and 0.253 g/day per g/day for ADG, respectively. Lowly negative correlations between sire intercept and slope do not support the hypothesis that high growth is associated with larger environmental sensitivity.

Economic weights for maternal traits of sows, including sow longevity.

The objective of this study was to develop a transparent, comprehensive, and flexible model for each trait for the formulation of breeding objectives for sow traits in swine breeding programs. Economic values were derived from submodels considering a typical Australian pig production system. Differences in timing and expressions of traits were accounted for to derive economic weights that were compared on the basis of their relative size after multiplication by their corresponding genetic standard deviation to account for differences in scale and genetic variability present for each trait. The number of piglets born alive had the greatest contribution (27.1%) to a subindex containing only maternal traits, followed by daily gain (maternal; 22.0%) and sow mature weight (15.0%). Other traits considered in the maternal breeding objective were preweaning survival (11.8%), sow longevity (12.5%), gilt age at puberty (8.7%), and piglet survival at birth (3.1%). The economic weights for number of piglets born alive and preweaning piglet survival were found to be highly dependent on the definition of scale of enterprise, with each economic value increasing by approximately 100% when it was assumed that the value of extra output per sow could be captured, rather than assuming a consequent reduction in the number of sows to maintain a constant level of output from a farm enterprise. In the context of a full maternal line index that must account also for the expression of direct genetic traits by the growing piglet progeny of sows, the maternal traits contributed approximately half of the variation in the overall breeding objective. Deployment of more comprehensive maternal line indexes incorporating the new maternal traits described would lead to more balanced selection outcomes and improved survival of pigs. Future work could facilitate evaluation of the economic impacts of desired-gains indexes, which could further improve animal welfare through improved sow and piglet survival. The results justify further development of selection criteria and breeding value prediction systems for a wider range of maternal traits relevant to pig production systems.

Economic weights for performance and survival traits of growing pigs.

The objective of this paper was to derive economic weights for performance and survival traits of growing pigs including feed conversion ratio (FCR), daily feed intake (DFI), ADG, postweaning survival of the growing pig (SG), and carcass fat depth at the P2 site (CFD). An independent model was developed for each trait to derive economic values directly based on a typical Australian production system. This flexible approach may be used to customize economic values for different production systems and alternative trait combinations in breeding objectives. Discounted genetic expressions were used as a means of taking into account differences in frequency and timing of expression of traits to obtain economic weights. Economic values for SG were derived based on a cost-saving and a lost-revenue approach. The correct formulation of the economic value of ADG depends on how feed cost is included in the breeding objective. If FCR is defined as a breeding objective trait, then savings in feed costs through earlier slaughter should not be counted in the economic value of ADG. In contrast, if DFI is included in the breeding objective instead of FCR, then feed-cost savings through earlier slaughter need to be attributed to the economic value for ADG, as a benefit from faster ADG. The paper also demonstrates that economic weightings in indexes for FCR can potentially be overestimated by 70% when it is assumed that DFI or FCR records taken from a limited duration test period reflect the corresponding trait over the full lifetime of the growing pig destined for slaughter. Postweaning survival of the growing pig was the most important breeding objective trait of growing pigs. The relative importance of each breeding objective trait in a sire-line index based on the genetic SD of each trait was 44.5, 27.0, 17.4, and 11.1% for SG, FCR, ADG, and CFD, respectively. Further studies to better clarify the extent of genetic variation that exists in SG under nucleus-farm and commercial-farm conditions are warranted, given the high economic importance of this survival trait of growing pigs.

Genetic parameters for haemoglobin levels in pigs and iron content in pork.

Genetic parameters were obtained for iron content in m. longissimus dorsi (2255 records) and haemoglobin levels recorded at 5 (4974 records) and 21 (2405 records) weeks of age in two sire lines from September 2009 until January 2011. The measure of iron in pork was the mean of two replicates. Genetic associations of haematological traits with meat quality traits (2255 records), as well as growth rate and backfat (close to 60 000 records), were estimated. Analyses were based on an animal model using residual maximum likelihood procedures. Iron content in pork was moderately heritable (0.34 ± 0.07) and genetic correlations with haemoglobin measures ranged from 0.39 ± 0.24 to 0.58 ± 0.13, indicating their potential use as selection criteria for increasing iron levels in pork. However, heritabilities for haemoglobin levels were low, ranging from 0.04 ± 0.2 to 0.18 ± 0.04. Procedures to measure haemoglobin on farm may require refinement. Redness of pork, quantified by a* value, had high genetic correlations with iron content (0.90 ± 0.04 to 0.94 ± 0.03) and moderate genetic correlations with haemoglobin levels (0.31 ± 0.22 to 0.55 ± 0.15). Iron content had significant genetic associations with L* measures (-0.61 ± 0.14 to -0.54 ± 0.23), b* value (0.60 ± 0.14 for dorsal b* measure, 0.50 ± 0.15 for average of dorsal and ventral b* measures) and pH at 45 min post mortem (-0.42 ± 0.14). These high genetic correlations between colour measurements and iron content in pork provide further avenues for selection strategies to improve iron content in pork. Current selection practices are not expected to affect iron content in pork, as no significant genetic correlations between performance and haematological traits were found.

Correlated responses in sow appetite, residual feed intake, body composition, and reproduction after divergent selection for residual feed intake in the growing pig.

Residual feed intake (RFI) has been explored as an alternative selection criterion to feed conversion ratio to capture the fraction of feed intake not explained by expected production and maintenance requirements. Selection experiments have found that low RFI in the growing pig is genetically correlated with reduced fatness and feed intake. Selection for feed conversion ratio also reduces sow appetite and fatness, which, together with increased prolificacy, has been seen as a hindrance for sow lifetime performance. The aims of our study were to derive equations for sow RFI during lactation (SRFI) and to evaluate the effect of selection for RFI during growth on sow traits during lactation. Data were obtained on 2 divergent lines selected for 7 generations for low and high RFI during growth in purebred Large Whites. The RFI was measured on candidates for selection (1,065 pigs), and sow performance data were available for 480 sows having from 1 to 3 parities (1,071 parities). Traits measured were sow daily feed intake (SDFI); sow BW and body composition before farrowing and at weaning (28.4 ± 1.7d); number of piglets born total, born alive, and surviving at weaning; and litter weight, average piglet BW, and within-litter SD of piglet BW at birth, 21 d of age (when creep feeding was available), and weaning. Sow RFI was defined as the difference between observed SDFI and SDFI predicted for sow maintenance and production. Daily production requirements were quantified by litter size and daily litter BW gain as well as daily changes in sow body reserves. The SRFI represented 24% of the phenotypic variability of SDFI. Heritability estimates for RFI and SRFI were both 0.14. The genetic correlation between RFI and SRFI was 0.29 ± 0.23. Genetic correlations of RFI with sow traits were low to moderate, consistent with responses to selection; selection for low RFI during growth reduced SDFI and increased number of piglets and litter growth, but also increased mobilization of body reserves. No effect on rebreeding performance was found. Metabolic changes previously observed during growth in response to selection might explain part of the better efficiency of the low-RFI sows, decreasing basal metabolism and favoring rapid allocation of resources to lactation. We propose to consider SRFI as an alternative to SDFI to select for efficient sows with reduced input demands during lactation.

Exploring breeding opportunities for reduced thermal sensitivity of feed intake in the lactating sow.

The aims of this study were, first, to evaluate the effects of climatic variables on daily feed intake of lactating sows and, second, to establish whether the response of sows to variation in temperature on feed intake during lactation was heritable. A total of 82,614 records for daily feed intake during lactation were available for 848 sows with 3,369 litters farrowing from January 2000 to December 2007. Climatic parameters available from the nearest weather station were maximum 24 h outside temperature, day length changes, and humidity. Although ambient room temperature was modified at the animal level in the farrowing shed, these climatic variables still had a significant effect on feed intake during lactation. Regression coefficients temperature and humidity were 0.01385 ± 0.00300 (temperature) - 0.00031 ± 0.00009 (temperature(2)) and 0.01443 ± 0.00620 (humidity) - 0.00009 ± 0.00004 (humidity(2)). There was an interaction between temperature and humidity, partly due to the climate control in the farrowing shed. At low temperature, feed intake increased considerably with greater humidity, in contrast to a small reduction in feed intake with greater humidity at high temperature. Day length change was modeled with a cosine function. At the start of autumn (September 21), sows ate 0.36 ± 0.056 kg/d less feed than at the start of spring (March 21). Daily feed intake during lactation was described as a function of days in lactation and as a function of both days in lactation and temperature using random regression models. The average heritability and repeatability summarized over the day in lactation at the mean temperature were 0.21 and 0.69, respectively. Genetic variance of temperature response on feed intake was less than 20% of the day effect. The permanent environmental variance was 2-fold (day) and 4-fold (temperature) greater than the corresponding additive genetic variance. Heritabilities of daily feed intake were greater during the first week of lactation compared with the rest of lactation. The genetic correlation between days decreased as time increased down to about 0.2 between the first and last day in lactation. The genetic correlation between feed intake records at the extreme temperatures decreased to about -0.35. It was concluded that random regression models are useful for research and results may be used to develop simpler models that can be implemented in practical breeding programs. An effect of temperature on lactation feed intake was found even in this climate-controlled environment located in a temperate climate zone. Larger effects are expected in more extreme climatic conditions with less temperature-controlled farrowing sheds.

Genetic relationships between composition of pork bellies and performance, carcase and meat quality traits.

Belly traits including predicted fat percentage of the belly (FATPC), combined area of the rib bone and muscle (RBMA), intermuscular fat area (IMFA) and subcutaneous fat area (SFA) were recorded on 2403 pigs along with carcase fat depth at the P2 site (P2). Belly traits were derived from image analysis of the anterior side of pork bellies. Further data available for pigs with belly data and their contemporaries included lifetime growth rate, ultrasound backfat and loin muscle depth (35 406 records), along with meat quality traits (3935 records). There were 4586 feed intake records and 18 398 juvenile insulin-like growth factor-I (IGF-I) records available, which included the majority of pigs with belly data. Genetic parameters were estimated based on an animal model using Residual Maximum Likelihood procedures. Heritability estimates for belly traits ranged from 0.23 to 0.34 (±0.05 to 0.06) while the common litter effect varied from 0.04 to 0.07 (±0.03). Genetic correlations between FATPC, individual belly fat measurements and carcase P2 fat depth differed significantly from unity, ranging from 0.71 to 0.85 (±0.05 to 0.08). Genetic correlations between IMFA and subcutaneous fat measurements varied from 0.47 to 0.63 (±0.08 to 0.13). Genetic correlations between belly and performance traits show that selection for reduced juvenile-IGF-I, reduced feed intake and reduced backfat along with increased loin muscle depth will reduce overall fat levels in the belly. Only loin muscle depth had a significant genetic correlation with RBMA (0.32 ± 0.10), thereby assisting selection for improved lean meat content of the belly. Ultimately, genetic improvement of belly muscles requires specific measurements of lean meat content of the belly. For this to be effective, measurements are required that can be routinely recorded on the slaughter line, or preferably on the live animal.

Economic weights for feed intake in the growing pig derived from a growth model and an economic model.

Economic weights are obtained for feed intake using a growth model and an economic model. The underlying concept of the growth model is the linear plateau model. Parameters of this model are the marginal ratio (MR) of extra fat and extra protein deposition with increasing feed intake (FI) and the maximum protein deposition (Pd(max)). The optimum feed intake (FI0) is defined as the minimum feed intake that meets energy requirements for Pd(max). The effect of varying FI and MR on performance traits was determined. An increase in FI results in a larger increase in growth rate with lower MR. For a given MR, feed conversion ratio is lowest when FI equals FI0. Lean meat percentage (LMP) is largest for a low MR in combination with a low FI. The decrease in LMP with higher FI islargest when FI exceeds FI0. Economic weights for FI, MR and Pd(max) depend on FI in relation to FI0. Economic weights for FI are positive when FI is less than FI0 and negative when FI is larger than FI0. The MR has only then a negative economic weight, when FI is below FI0. Economic weights of FI and MR have a larger magnitude with lower MR and lower Fl. In contrast, economic weights for growth rate and FI derived from the economic model only change in magnitude and not in sign with different levels of these traits. The economic model always puts a negative economic weight on FI since it expresses profit due to a decrease in FI with constant growth rate and LMP. This holds the risk of continuous decrease in FI in pig breeding programs. In contrast, the use of growth models for genetic improvement allows direct selection for an optimum feed intake which maximizes feed efficiency in combination with maximum lean meat growth. It is concluded that recording procedures have to be adapted to collect the data necessary to implement growth models in practical pig breeding applications.