Joint analysis of longitudinal feed intake and single recorded production traits in pigs using a novel Horizontal model

A Random Regression Model Based on a Single-Step Method for Improving the Genomic Prediction Accuracy of Residual Feed Intake in Pigs

Residual feed intake (RFI) is considered as a measurement of feed efficiency, which is greatly related to the growth performance in pigs. Daily feeding records can be obtained from automatic feeders. In general, RFI is usually calculated from the total measurement records during the whole test period. This measurement cannot reflect genetic changes in different growth periods during the test. A random regression model (RRM) provides a method to model such type of longitudinal data. To improve the accuracy of genetic prediction for RFI, the RRM and regular animal models were applied in this study, and their prediction performances were compared. Both traditional pedigree-based relationship matrix (A matrix) and pedigree and genomic information-based relationship matrix (H matrix) were applied for these two models. The results showed that, the prediction accuracy of the RRM was higher than that of the animal model, increasing 24.2% with A matrix and 40.9% with H matrix. Furthermore, genomic information constantly improved the accuracy of evaluation under each evaluation model. In conclusion, longitudinal traits such as RFI can describe feed efficiency better, and the RRM with both pedigree and genetic information was superior to the animal model. These results provide a feasible method of genomic prediction using longitudinal data in animal breeding.

Genomic and Phenotypic Agreement Defines the Use of Microwave Dielectric Spectroscopy for Recording Muscle Lipid Content in European Seabass (Dicentrarchus labrax)

Recording the fillet lipid percentage in European seabass is crucial to control lipid deposition as a means toward improving production efficiency and product quality. The reference method for recording lipid content is solvent lipid extraction and is the most accurate and precise method available. However, it is costly, requires sacrificing the fish and grinding the fillet sample which limits the scope of applications, for example grading of fillets, recording live fish or selective breeding of fish with own phenotypes are all limited. We tested a rapid, cost effective and non-destructive handheld microwave dielectric spectrometer (namely the Distell fat meter) against the reference method by recording both methods on 313 European seabass (Dicentrarchus labrax). The total method agreement between the dielectric spectrometer and the reference method was assessed by Lin’s concordance correlation coefficient (CCC), which was low to moderate CCC = 0.36–0.63. We detected a significant underestimation in accuracy of lipid percentage 22–26% by the dielectric spectrometer and increased imprecision resulting in the coefficient of variation (CV) doubling for dielectric spectrometer CV = 40.7–46% as compared to the reference method 27–31%. Substantial genetic variation for fillet lipid percentage was found for both the reference method (h² = 0.59) and dielectric spectroscopy (h² = 0.38–0.58), demonstrating that selective breeding is a promising method for controlling fillet lipid content. Importantly, the genetic correlation (r_g) between the dielectric spectrometer and the reference method was positive and close to unity (r_g = 0.96), demonstrating the dielectric spectrometer captures practically all the genetic variation in the reference method. These findings form the basis of defining the scope of applications and experimental design for using dielectric spectroscopy for recording fillet lipid content in European seabass and validate its use for selective breeding.

Simultaneous Bayesian estimation of genetic parameters for curves of weight, feed intake, and residual feed intake in beef cattle

Rates of gain and feed efficiency are important traits in most breeding programs for growing farm animals. The rate of gain (GAIN) is usually expressed over a certain age period and feed efficiency is often expressed as residual feed intake (RFI), defined as observed feed intake (FI) minus expected feed intake based on live weight (WGT) and GAIN. However, the basic traits recorded are always WGT and FI and other traits are derived from these basic records. The aim of this study was to develop a procedure for simultaneous analysis of the basic records and then derive linear traits related to feed efficiency without retorting to any approximation. A bivariate longitudinal random regression model was employed on 13,791 individual longitudinal records of WGT and FI from 2,827 bulls of six different beef breeds tested for their own performance in the period from 7 to 13 mo of age. Genetic and permanent environmental covariance functions for curves of WGT and FI were estimated using Gibbs sampling. Genetic and permanent covariance functions for curves of GAIN were estimated from the first derivative of the function for WGT and finally the covariance functions were extended to curves for RFI, based on the conditional distribution of FI given WGT and GAIN. Furthermore, the covariance functions were extended to include GAIN and RFI defined over different periods of the performance test. These periods included the whole test period as normally used when predicting breeding values for GAIN and RFI for beef bulls. Based on the presented method, breeding values and genetic parameters for derived traits such as GAIN and RFI defined longitudinally or integrated over (parts of) of the test period can be obtained from a joint analysis of the basic records. The resulting covariance functions for WGT, FI, GAIN, and RFI are usually singular but the method presented here does not suffer from the estimation problems associated with defining these traits individually before the genetic analysis. All the results are thus estimated simultaneously, and the set of parameters is consistent.

New residual feed intake criterion for longitudinal data

Background

Residual feed intake (RFI) is one measure of feed efficiency, which is usually obtained by multiple regression of feed intake (FI) on measures of production, body weight gain and tissue composition. If phenotypic regression is used, the resulting RFI is generally not genetically independent of production traits, whereas if RFI is computed using genetic regression coefficients, RFI and production traits are independent at the genetic level. The corresponding regression coefficients can be easily derived from the result of a multiple trait model that includes FI and production traits. However, this approach is difficult to apply in the case of multiple repeated measurements of FI and production traits. To overcome this difficulty, we used a structured antedependence approach to account for the longitudinality of the data with a phenotypic regression model or with different genetic and environmental regression coefficients [multi- structured antedependence model (SAD) regression model].

Results

After demonstrating the properties of RFI obtained by the multi-SAD regression model, we applied the two models to FI and production traits that were recorded for 2435 French Large White pigs over a 10-week period. Heritability estimates were moderate with both models. With the multi-SAD regression model, heritability estimates were quite stable over time, ranging from 0.14 ± 0.04 to 0.16 ± 0.05, while heritability estimates showed a U-shaped profile with the phenotypic regression model (ranging from 0.19 ± 0.06 to 0.28 ± 0.06). Estimates of genetic correlations between RFI at different time points followed the same pattern for the two models but higher estimates were obtained with the phenotypic regression model. Estimates of breeding values that can be used for selection were obtained by eigen-decomposition of the genetic covariance matrix. Correlations between these estimated breeding values obtained with the two models ranged from 0.66 to 0.83.

Conclusions

The multi-SAD model is preferred for the genetic analysis of longitudinal RFI because, compared to the phenotypic regression model, it provides RFI that are genetically independent of production traits at all time points. Furthermore, it can be applied even when production records are missing at certain time points.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12711-021-00641-2.

Genetic parameters and genomic prediction for feed intake recorded at the group and individual level in different production systems for growing pigs

Background

In breeding programs, recording large-scale feed intake (FI) data routinely at the individual level is costly and difficult compared with other production traits. An alternative approach could be to record FI at the group level since animals such as pigs are normally housed in groups and fed by a shared feeder. However, to date there have been few investigations about the difference between group- and individual-level FI recorded in different environments. We hypothesized that group- and individual-level FI are genetically correlated but different traits. This study, based on the experiment undertaken in purebred DanBred Landrace (L) boars, was set out to estimate the genetic variances and correlations between group- and individual-level FI using a bivariate random regression model, and to examine to what extent prediction accuracy can be improved by adding information of individual-level FI to group-level FI for animals recorded in groups. For both bivariate and univariate models, single-step genomic best linear unbiased prediction (ssGBLUP) and pedigree-based BLUP (PBLUP) were implemented and compared.

Results

The variance components from group-level records and from individual-level records were similar. Heritabilities estimated from group-level FI were lower than those from individual-level FI over the test period. The estimated genetic correlations between group- and individual-level FI based on each test day were on average equal to 0.32 (SD = 0.07), and the estimated genetic correlation for the whole test period was equal to 0.23. Our results demonstrate that by adding information from individual-level FI records to group-level FI records, prediction accuracy increased by 0.018 and 0.032 compared with using group-level FI records only (bivariate vs. univariate model) for PBLUP and ssGBLUP, respectively.

Conclusions

Based on the current dataset, our findings support the hypothesis that group- and individual-level FI are different traits. Thus, the differences in FI traits under these two feeding systems need to be taken into consideration in pig breeding programs. Overall, adding information from individual records can improve prediction accuracy for animals with group records.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12711-021-00624-3.

Why breed disease-resilient livestock, and how?

BACKGROUND

Fighting and controlling epidemic and endemic diseases represents a considerable cost to livestock production. Much research is dedicated to breeding disease resilient livestock, but this is not yet a common objective in practical breeding programs. In this paper, we investigate how future breeding programs may benefit from recent research on disease resilience.

MAIN BODY

We define disease resilience in terms of its component traits resistance (R: the ability of a host animal to limit within-host pathogen load (PL)) and tolerance (T: the ability of an infected host to limit the damage caused by a given PL), and model the host's production performance as a reaction norm on PL, depending on R and T. Based on this, we derive equations for the economic values of resilience and its component traits. A case study on porcine respiratory and reproductive syndrome (PRRS) in pigs illustrates that the economic value of increasing production in infectious conditions through selection for R and T can be more than three times higher than by selection for production in disease-free conditions. Although this reaction norm model of resilience is helpful for quantifying its relationship to its component traits, its parameters are difficult and expensive to quantify. We consider the consequences of ignoring R and T in breeding programs that measure resilience as production in infectious conditions with unknown PL-particularly, the risk that the genetic correlation between R and T is unfavourable (antagonistic) and that a trade-off between them neutralizes the resilience improvement. We describe four approaches to avoid such antagonisms: (1) by producing sufficient PL records to estimate this correlation and check for antagonisms-if found, continue routine PL recording, and if not found, shift to cheaper proxies for PL; (2) by selection on quantitative trait loci (QTL) known to influence both R and T in favourable ways; (3) by rapidly modifying towards near-complete resistance or tolerance, (4) by re-defining resilience as the animal's capacity to resist (or recover from) the perturbation caused by an infection, measured as temporal deviations of production traits in within-host longitudinal data series.

CONCLUSIONS

All four alternatives offer promising options for genetic improvement of disease resilience, and most rely on technological and methodological developments and innovation in automated data generation.

Boosted trees to predict pneumonia, growth, and meat percentage of growing-finishing pigs1

In pig production, efficiency is benefiting from uniform growth in pens resulting in single deliveries from a pen of possibly all animals in the targeted weight range. Abnormalities, like pneumonia or aberrant growth, reduce production efficiency as it reduces the uniformity and might cause multiple deliveries per batch and pigs delivered with a low meat yield or outside the targeted weight range. Early identification of pigs prone to develop these abnormalities, for example, at the onset of the growing-finishing phase, would help to prevent heterogeneous pens through management interventions. Data about previous production cycles at the farm combined with data from the piglet’s own history may help in identifying these abnormalities. The aim of this study, therefore, was to predict at the onset of the growing-finishing phase, that is, at 3 mo in advance, deviant pigs at slaughter with a machine-learning technique called boosted trees. The dataset used was extracted from the farm management system of a research center. It contained over 70,000 records of individual pigs born between 2004 and 2016, including information on, for example, offspring, litter size, transfer dates between production stages, their respective locations within the barns, and individual live-weights at several production stages. Results obtained on an independent test set showed that at a 90% specificity rate, the sensitivity was 16% for low meat percentage, 20% for pneumonia and 36% for low lifetime growth rate. For low lifetime growth rate, this meant an almost three times increase in positive predictive value compared to the current situation. From these results, it was concluded that routine performance information available at the onset of the growing-finishing phase combined with data about previous production cycles formed a moderate base to identify pigs prone to develop pneumonia (AUC > 0.60) and a good base to identify pigs prone to develop growth aberrations (AUC > 0.70) during the growing-finishing phase. The mentioned information, however, was not a sufficient base to identify pigs prone to develop low meat percentage (AUC < 0.60). The shown ability to identify growth aberrations and pneumonia can be considered a good first step towards the development of an early warning system for pigs in the growing-finishing phase.

Macro-environmental sensitivity for growth rate in Danish Duroc pigs is under genetic control

The aim of this study was to examine (i) the genetic variation in macro-environmental sensitivity (macro-ES) for ADG in Danish Duroc pigs, (ii) the genetic heterogeneity among sexes, and (iii) residual variance heterogeneity among herds. Record of ADG for 32,297 boars (19 herds) and 42,724 gilts (16 herds) was used for analysis. The data were provided by the National Danish Pig Research Centre. The analysis was performed by fitting univariate reaction norm models with the herd-year-month on test (HYM) effect as environmental covariates and herd-specific residual variance for boars and gilts separately under a Bayesian setting. The environmental covariate was inferred simultaneously with other parameters of the model. Gibbs sampling was used to sample model dispersion and location parameters. The posterior means and highest posterior density intervals of the additive genetic variance, genetic correlations for ADG, and heritability were calculated over the continuous environmental range of -3σh to +3σh (SD of the HYM effect). The coheritability of ADG at the average environmental level and ADG in the environments along the -3σh to +3σh environmental gradient were also calculated. The analysis showed significant variation in macro-ES, revealing genotype by environment interactions (G × E) for ADG. The presence of G × E resulted in changes in additive genetic variance and heritability across the -3σh to +3σh range. The genetic correlations were high and positive between ADG in environments differing by 1σh units or less and decreased to moderately positive between ADG in the extreme environments in both sexes. The coheritability of ADG in the environment at the average level and the -3σh environment for boars were greater than the heritability in the environment at the average level, while it was less for gilts. The coheritability of ADG in the environment at the average level and the +3σh environment for boars was less than heritability in the environment at the average level, while it was either the same or greater for gilts, depending on the residual variance. Boars had larger additive genetic and residual variances than gilts. Heterogeneity of residual variances across herds was shown for both sexes. In conclusion, this study shows the presence of macro-ES, genetic variance heterogeneity among sexes for ADG in pigs, and residual variance heterogeneity across herds.

Estimation of variance components and prediction of breeding values based on group records from varying group sizes

Background

Records on groups of individuals rather than on single individuals could be valuable for predicting breeding values (BV) of the traits that are difficult or costly to measure individually, such as feed intake in pigs or beef cattle. Here, we present a model, which handles group records from varying group sizes and involves multiple fixed and random effects, for estimating variance components and predicting BV. Moreover, using simulation, we investigated the efficiency of group records for predicting BV in situations with various group sizes and structures, and factors that affect the trait.

Results

The results show that the presented model for group records worked well and that variances estimated from group records with varying group sizes were consistent with those estimated from individual records, but with larger standard errors. Ignoring litter and pen effects had very little or no influence on the accuracy of estimated BV (EBV) obtained from group records. However, ignoring litter effects resulted in biased estimates of additive genetic variance and EBV. The presence of litter and pen effects on phenotypes decreased the accuracy of EBV although the prediction model fitted both effects. Having more littermates in the same pen led to a higher accuracy of EBV. The decay of EBV accuracy with increasing group size was more marked for scenarios with litter and pen effects than without. When litters of six individuals were divided into two pens, accuracies of EBV obtained from group records with a size up to 12 (average 9.6) and up to 24 (average 19.2) were 66.6 and 57.6% of those estimated from individual records in the scenario with litter and pen effects on phenotypes. These percentages reached 77.0 and 68.4% in the scenario without litter and pen effects on phenotypes.

Conclusions

Our results indicate that the model works appropriately for the analysis of group records from varying group sizes. Using group records for genetic evaluation of traits such as feed intake in pig is feasible and the efficiency of the resulting estimates depends on the size and structure of the groups and on the magnitude of the variances for litter and pen effects.

Bayesian estimation of direct and correlated responses to selection on linear or ratio expressions of feed efficiency in pigs

Background

This study aimed at (1) deriving Bayesian methods to predict breeding values for ratio (i.e. feed conversion ratio; FCR) or linear (i.e. residual feed intake; RFI) traits; (2) estimating genetic parameters for average daily feed consumption (ADFI), average daily weight gain (ADG), lean meat percentage (LMP) along with the derived traits of RFI and FCR; and (3) deriving Bayesian estimates of direct and correlated responses to selection on RFI, FCR, ADG, ADFI, and LMP. Response to selection was defined as the difference in additive genetic mean of the selected top individuals, expected to be parents of the next generation, and the total population after integrating genetic trends out of the posterior distribution of selection responses. Inferences were based on marginal posterior distributions obtained from the Bayesian method for integration over unknown population parameters and “fixed” environmental effects and for appropriate handling of ratio traits. Terminal line pigs (n = 3724) were used for a multi-variate model for ADFI, ADG, and LMP. RFI was estimated from the conditional distribution of ADFI given ADG and LMP, using either genetic (RFI_G) or phenotypic (RFI_P) partial regression coefficients. The posterior distribution of the FCR’s breeding values was derived from the posterior distribution of “fixed” environmental effects and additive genetic effects on ADFI and ADG.

Results

Posterior means of heritability were 0.32, 0.26, 0.56, 0.20, and 0.15 for ADFI, ADG, LMP, RFI_P, and RFI_G, respectively. Selection against RFI_G showed a direct response of − 0.16 kg/d and correlated responses of − 0.16 kg/kg for FCR and − 0.15 kg/d for ADFI, with no effect on other production traits. Selection against FCR resulted in a direct response of − 0.17 kg/kg and correlated responses of − 0.14 kg/d for RFI_G, − 0.18 kg/d for ADFI, and 0.98% for LMP.

Conclusions

The Bayesian methodology developed here enables prediction of breeding values for FCR and RFI from a single multi-variate model. In addition, we derived posterior distributions of direct and correlated responses to selection. Genetic parameter estimates indicated a genetic basis for the studied traits and that genetic improvement through selection was possible. Direct selection against FCR or RFI_P resulted in unexpected responses in production traits.