Research Note: Effects of high barn temperature on group-level dispersion and individual activity in broiler chickens.

Heat stress in broilers is a pressing issue in the changing climate. Data on broiler behavior might be useful for early detection of heat stress and subsequent intervention, and may provide potential indicators for heat tolerance that can be used in broiler breeding programs. Here, we used bird location data collected in a previous study during which broilers were inadvertently exposed to high ambient temperatures due to a local heat wave. We examined whether broiler behavior changed with increasing ambient temperatures, focusing on group-level dispersion behavior and individual-level locomotor activity. We observed that birds moved closer together with increasing temperatures up to 9 °C above the desired level, and remained in similar proximity or moved further apart at temperatures above that threshold. The activity level decreased or remained stable with increasing temperature during most parts of the day, but increased at the end of the day. Possibly, the birds exhibited compensatory behavior (such as drinking and eating) during the periods when the barn cooled down after a hot day, but that could not be confirmed as no behavioral observations were available. The difference in activity levels between individuals accounted for 8.4% of the total variation, suggesting that activity might be an interesting indicator trait for heat tolerance in broiler chickens. Overall, the results of this study can inform the development of behavior-based 1) early-warning systems for heat stress and 2) heat tolerance indicators, although data on behaviors that are more specific to heat stress are probably required.

Validation with single-step SNPBLUP shows that evaluations can continue using a single mean of genotyped individuals, even with multiple breeds.

BACKGROUND: In genomic prediction, it is common to centre the genotypes of single nucleotide polymorphisms based on the allele frequencies in the current population, rather than those in the base generation. The mean breeding value of non-genotyped animals is conditional on the mean performance of genotyped relatives, but can be corrected by fitting the mean performance of genotyped individuals as a fixed regression. The associated covariate vector has been referred to as a 'J-factor', which if fitted as a fixed effect can improve the accuracy and dispersion bias of sire genomic estimated breeding values (GEBV). To date, this has only been performed on populations with a single breed. Here, we investigated whether there was any benefit in fitting a separate J-factor for each breed in a three-way crossbred population, and in using pedigree-based expected or genome-based estimated breed fractions to define the J-factors. RESULTS: For body weight at 7 days, dispersion bias decreased when fitting multiple J-factors, but only with a low proportion of genotyped individuals with selective genotyping. On average, the mean regression coefficients of validation records on those of GEBV increased with one J-factor compared to none, and further increased with multiple J-factors. However, for body weight at 35 days this was not observed. The accuracy of GEBV remained unchanged regardless of the J-factor method used. Differences between the J-factor methods were limited with correlations approaching 1 for the estimated covariate vector, the estimated coefficients of the regression on the J-factors, and the GEBV. CONCLUSIONS: Based on our results and in the particular design analysed here, i.e. all the animals with phenotype are of the same type of crossbreds, fitting a single J-factor should be sufficient, to reduce dispersion bias. Fitting multiple J-factors may reduce dispersion bias further but this depends on the trait and genotyping rate. For the crossbred population analysed, fitting multiple J-factors has no adverse consequences and if this is done, it does not matter if the breed fractions used are based on the pedigree-expectation or the genomic estimates. Finally, when GEBV are estimated from crossbred data, any observed bias can potentially be reduced by including a straightforward regression on actual breed proportions.

Predicting the impact of genotype-by-genotype interaction on the purebred-crossbred genetic correlation from phenotype and genotype marker data of parental lines.

BACKGROUND: The genetic correlation between purebred (PB) and crossbred (CB) performances ([Formula: see text]) partially determines the response in CB when selection is on PB performance in the parental lines. An earlier study has derived expressions for an upper and lower bound of [Formula: see text], using the variance components of the parental purebred lines, including e.g. the additive genetic variance in the sire line for the trait expressed in one of the dam lines. How to estimate these variance components is not obvious, because animals from one parental line do not have phenotypes for the trait expressed in the other line. Thus, the aim of this study was to propose and compare three methods for approximating the required variance components. The first two methods are based on (co)variances of genomic estimated breeding values (GEBV) in the line of interest, either accounting for shrinkage (VCGEBV-S) or not (VCGEBV). The third method uses restricted maximum likelihood (REML) estimates directly from univariate and bivariate analyses (VCREML) by ignoring that the variance components should refer to the line of interest, rather than to the line in which the trait is expressed. We validated these methods by comparing the resulting predicted bounds of [Formula: see text] with the [Formula: see text] estimated from PB and CB data for five traits in a three-way cross in pigs. RESULTS: With both VCGEBV and VCREML, the estimated [Formula: see text] (plus or minus one standard error) was between the upper and lower bounds in 14 out of 15 cases. However, the range between the bounds was much smaller with VCREML (0.15-0.22) than with VCGEBV (0.44-0.57). With VCGEBV-S, the estimated [Formula: see text] was between the upper and lower bounds in only six out of 15 cases, with the bounds ranging from 0.21 to 0.44. CONCLUSIONS: We conclude that using REML estimates of variance components within and between parental lines to predict the bounds of [Formula: see text] resulted in better predictions than methods based on GEBV. Thus, we recommend that the studies that estimate [Formula: see text] with genotype data also report estimated genetic variance components within and between the parental lines.

Review: optimizing genomic selection for crossbred performance by model improvement and data collection.

Breeding programs aiming to improve the performance of crossbreds may benefit from genomic prediction of crossbred (CB) performance for purebred (PB) selection candidates. In this review, we compared genomic prediction strategies that differed in 1) the genomic prediction model used or 2) the data used in the reference population. We found 27 unique studies, two of which used deterministic simulation, 11 used stochastic simulation, and 14 real data. Differences in accuracy and response to selection between strategies depended on i) the value of the purebred crossbred genetic correlation (rpc), ii) the genetic distance between the parental lines, iii) the size of PB and CB reference populations, and iv) the relatedness of these reference populations to the selection candidates. In studies where a PB reference population was used, the use of a dominance model yielded accuracies that were equal to or higher than those of additive models. When rpc was lower than ~0.8, and was caused mainly by G × E, it was beneficial to create a reference population of PB animals that are tested in a CB environment. In general, the benefit of collecting CB information increased with decreasing rpc. For a given rpc, the benefit of collecting CB information increased with increasing size of the reference populations. Collecting CB information was not beneficial when rpc was higher than ~0.9, especially when the reference populations were small. Collecting only phenotypes of CB animals may slightly improve accuracy and response to selection, but requires that the pedigree is known. It is, therefore, advisable to genotype these CB animals as well. Finally, considering the breed-origin of alleles allows for modeling breed-specific effects in the CB, but this did not always lead to higher accuracies. Our review shows that the differences in accuracy and response to selection between strategies depend on several factors. One of the most important factors is rpc, and we, therefore, recommend to obtain accurate estimates of rpc of all breeding goal traits. Furthermore, knowledge about the importance of components of rpc (i.e., dominance, epistasis, and G × E) can help breeders to decide which model to use, and whether to collect data on animals in a CB environment. Future research should focus on the development of a tool that predicts accuracy and response to selection from scenario specific parameters.

Predicting the purebred-crossbred genetic correlation from the genetic variance components in the parental lines.

BACKGROUND: The genetic correlation between purebred and crossbred performance ([Formula: see text]) is an important parameter in pig and poultry breeding, because response to selection in crossbred performance depends on the value of [Formula: see text] when selection is based on purebred (PB) performance. The value of [Formula: see text] can be substantially lower than 1, which is partly due to differences in allele frequencies between parental lines when non-additive genetic effects are present. This relationship between [Formula: see text] and parental allele frequencies suggests that [Formula: see text] can be expressed as a function of genetic parameters for the trait in the parental lines. In this study, we derived expressions for [Formula: see text] based on genetic variances within, and the genetic covariance between parental lines. It is important to note that the variance components used in our expressions are not the components that are typically estimated in empirical data. The expressions were derived for a genetic model with additive and dominance effects (D), and additive and epistatic additive-by-additive effects (EAA). We validated our expressions using simulations of purebred parental lines and their crosses, where the parental lines were either selected or not. Finally, using these simulations, we investigated the value of [Formula: see text] for genetic models with both dominance and epistasis or with other types of epistasis, for which expressions could not be derived. RESULTS: Our simulations show that when non-additive effects are present, [Formula: see text] decreases with increasing differences in allele frequencies between the parental lines. Genetic models that involve dominance result in lower values of [Formula: see text] than genetic models that involve epistasis only. Using information of parental lines only, our expressions provide exact estimates of [Formula: see text] for models D and EAA, and accurate upper and lower bounds of [Formula: see text] for two other genetic models. CONCLUSION: This work lays the foundation to enable estimation of [Formula: see text] from information collected in PB parental lines only.

Bayesian neural networks with variable selection for prediction of genotypic values.

BACKGROUND: Estimating the genetic component of a complex phenotype is a complicated problem, mainly because there are many allele effects to estimate from a limited number of phenotypes. In spite of this difficulty, linear methods with variable selection have been able to give good predictions of additive effects of individuals. However, prediction of non-additive genetic effects is challenging with the usual prediction methods. In machine learning, non-additive relations between inputs can be modeled with neural networks. We developed a novel method (NetSparse) that uses Bayesian neural networks with variable selection for the prediction of genotypic values of individuals, including non-additive genetic effects. RESULTS: We simulated several populations with different phenotypic models and compared NetSparse to genomic best linear unbiased prediction (GBLUP), BayesB, their dominance variants, and an additive by additive method. We found that when the number of QTL was relatively small (10 or 100), NetSparse had 2 to 28 percentage points higher accuracy than the reference methods. For scenarios that included dominance or epistatic effects, NetSparse had 0.0 to 3.9 percentage points higher accuracy for predicting phenotypes than the reference methods, except in scenarios with extreme overdominance, for which reference methods that explicitly model dominance had 6 percentage points higher accuracy than NetSparse. CONCLUSIONS: Bayesian neural networks with variable selection are promising for prediction of the genetic component of complex traits in animal breeding, and their performance is robust across different genetic models. However, their large computational costs can hinder their use in practice.

Machine learning to further improve the decision which boar ejaculates to process into artificial insemination doses.

Current artificial insemination (AI) laboratory practices assess semen quality of each boar ejaculate to decide which ones to process into AI doses. This decision is aided with two, world-wide used, motility parameters that come available through computer assisted semen analysis (CASA). This decision process, however, still results in AI doses with variable and sometimes suboptimal fertility outcomes (e.g., small litter size). The hypothesis was that the decision which ejaculates to process into AI doses can be improved by adding more data from CASA systems, and data from other sources, in combination with a data-driven model. Available data consisted of ejaculates that passed the initial decision, and thus, were processed into AI doses and used to inseminate sows. Data were divided into a training set (6793 records) and a validation set (1191 records) for model development, and an independent test set (1434 records) for performance assessment. Gradient Boosting Machine (GBM) models were developed to predict four fertility phenotypes of interest (gestation length, total number born, number born alive, and number of stillborn piglets). Each fertility phenotype was considered as a numeric and as a binary outcome parameter, totaling to eight different fertility phenotypes. Data used to further improve the decision process originated from four sources: 1) CASA information, 2) boar ejaculate information, 3) breeding value estimations, and 4) weather information. These data were used to create seven prediction sets, where each new set added parameters to the ones included in the previous set. The GBM models predicted fertility phenotypes with low correlations (for numeric phenotypes) and area under the curve values (for binary phenotypes) on the test data. Hence, results demonstrated that a combination of more data and GBM did not enable further improvement of the AI dose quality checks, resulting in the rejection of our hypothesis. However, our study revealed parameters affecting boar ejaculate fertility which were not used in today's decision process. These parameters (listed in the top 10 in at least four GBM models) included one parameter associated with boar ejaculate information, two with breeding value estimations, five with CASA information, and one with weather information. These parameters, therefore, should be further investigated for their potential value when assessing the quality of boar ejaculates in daily routine AI doses processing.

The Impact of Non-additive Effects on the Genetic Correlation Between Populations.

Average effects of alleles can show considerable differences between populations. The magnitude of these differences can be measured by the additive genetic correlation between populations ([Formula: see text]). This [Formula: see text] can be lower than one due to the presence of non-additive genetic effects together with differences in allele frequencies between populations. However, the relationship between the nature of non-additive effects, differences in allele frequencies, and the value of [Formula: see text] remains unclear, and was therefore the focus of this study. We simulated genotype data of two populations that have diverged under drift only, or under drift and selection, and we simulated traits where the genetic model and magnitude of non-additive effects were varied. Results showed that larger differences in allele frequencies and larger non-additive effects resulted in lower values of [Formula: see text] In addition, we found that with epistasis, [Formula: see text] decreases with an increase of the number of interactions per locus. For both dominance and epistasis, we found that, when non-additive effects became extremely large, [Formula: see text] had a lower bound that was determined by the type of inter-allelic interaction, and the difference in allele frequencies between populations. Given that dominance variance is usually small, our results show that it is unlikely that true [Formula: see text] values lower than 0.80 are due to dominance effects alone. With realistic levels of epistasis, [Formula: see text] dropped as low as 0.45. These results may contribute to the understanding of differences in genetic expression of complex traits between populations, and may help in explaining the inefficiency of genomic trait prediction across populations.

Correction to: Validation of genomic predictions for body weight in broilers using crossbred information and considering breed-of-origin of alleles.

Following publication of original article [1], we noticed that there was an error: Eq. (3) on page 5 is the genomic relationship matrix that.

Validation of genomic predictions for body weight in broilers using crossbred information and considering breed-of-origin of alleles.

BACKGROUND: Pig and poultry breeding programs aim at improving crossbred (CB) performance. Selection response may be suboptimal if only purebred (PB) performance is used to compute genomic estimated breeding values (GEBV) because the genetic correlation between PB and CB performance ([Formula: see text]) is often lower than 1. Thus, it may be beneficial to use information on both PB and CB performance. In addition, the accuracy of GEBV of PB animals for CB performance may improve when the breed-of-origin of alleles (BOA) is considered in the genomic relationship matrix (GRM). Thus, our aim was to compare scenarios where GEBV are computed and validated by using (1) either CB offspring averages or individual CB records for validation, (2) either a PB or CB reference population, and (3) a GRM that either accounts for or ignores BOA in the CB individuals. For this purpose, we used data on body weight measured at around 7 (BW7) or 35 (BW35) days in PB and CB broiler chickens and evaluated the accuracy of GEBV based on the correlation GEBV with phenotypes in the validation population (validation correlation). RESULTS: With validation on CB offspring averages, the validation correlation of GEBV of PB animals for CB performance was lower with a CB reference population than with a PB reference population for BW35 ([Formula: see text] = 0.96), and about equal for BW7 ([Formula: see text] = 0.80) when BOA was ignored. However, with validation on individual CB records, the validation correlation was higher with a CB reference population for both traits. The use of a GRM that took BOA into account increased the validation correlation for BW7 but reduced it for BW35. CONCLUSIONS: We argue that the benefit of using a CB reference population for genomic prediction of PB animals for CB performance should be assessed either by validation on CB offspring averages, or by validation on individual CB records while using a GRM that accounts for BOA in the CB individuals. With this recommendation in mind, our results show that the accuracy of GEBV of PB animals for CB performance was equal to or higher with a CB reference population than with a PB reference population for a trait with an [Formula: see text] of 0.8, but lower for a trait with an [Formula: see text] of 0.96. In addition, taking BOA into account was beneficial for a trait with an [Formula: see text] of 0.8 but not for a trait with an [Formula: see text] of 0.96.

Estimating the purebred-crossbred genetic correlation of body weight in broiler chickens with pedigree or genomic relationships.

BACKGROUND: In pig and poultry breeding programs, the breeding goal is to improve crossbred (CB) performance, whereas selection in the purebred (PB) lines is often based on PB performance. Thus, response to selection may be suboptimal, because the genetic correlation between PB and CB performance ([Formula: see text]) is generally lower than 1. Accurate estimates of the [Formula: see text] are needed, so that breeders can decide if they should collect data from CB animals. [Formula: see text] can be estimated either from pedigree or genomic relationships, which may produce different results. With genomic relationships, the [Formula: see text] estimate could be improved when relationships between purebred and crossbred animals are based only on the alleles that originate from the PB line of interest. This work presents the first comparison of estimated [Formula: see text] and variance components of body weight in broilers, using pedigree-based or genotype-based models, where the breed-of-origin of alleles was either ignored or considered. We used genotypes and body weight measurements of PB and CB animals that have a common sire line. RESULTS: Our results showed that the [Formula: see text] estimates depended on the relationship matrix used. Estimates were 5 to 25% larger with genotype-based models than with pedigree-based models. Moreover, [Formula: see text] estimates were similar (max. 7% difference) regardless of whether the model considered breed-of-origin of alleles or not. Standard errors of [Formula: see text] estimates were smaller with genotype-based than with pedigree-based methods, and smaller with models that ignored breed-of-origin than with models that considered breed-of-origin. CONCLUSIONS: We conclude that genotype-based models can be useful for estimating [Formula: see text], even when the PB and CB animals that have phenotypes are closely related. Considering breed-of-origin of alleles did not yield different estimates of [Formula: see text], probably because the parental breeds of the CB animals were distantly related.

Required properties for markers used to calculate unbiased estimates of the genetic correlation between populations.

BACKGROUND: Generally, populations differ in terms of environmental and genetic factors, which can create differences in allele substitution effects between populations. Therefore, a single genotype may have different additive genetic values in different populations. The correlation between the two additive genetic values of a single genotype in two populations is known as the additive genetic correlation between populations and thus, can differ from 1. Our objective was to investigate whether differences in linkage disequilibrium (LD) and allele frequencies of markers and causal loci between populations affect the bias of the estimated genetic correlation. We simulated two populations that were separated by 50 generations and differed in LD pattern between markers and causal loci, as measured by the LD-statistic r. We used a high marker density to represent a high consistency of LD between populations, and lower marker densities to represent situations with a lower consistency of LD between populations. Markers and causal loci were selected to have either similar or different allele frequencies in the two populations. RESULTS: Our results show that genetic correlations were underestimated only slightly when the difference in allele frequencies between the two populations was similar for the markers and the causal loci. A lower marker density, representing a lower consistency of LD between populations, had only a minor effect on the underestimation of the genetic correlation. When the difference in allele frequencies between the two populations was not similar for markers and causal loci, genetic correlations were severely underestimated. This bias occurred because the markers did not predict accurately the relationships at causal loci. CONCLUSIONS: For an unbiased estimation of the genetic correlation between populations, the markers should accurately predict the relationships at the causal loci. To achieve this, it is essential that the difference in allele frequencies between populations is similar for markers and causal loci. Our results show that differences in LD phase between causal loci and markers across populations have little effect on the estimated genetic correlation.

Benefits of Dominance over Additive Models for the Estimation of Average Effects in the Presence of Dominance.

In quantitative genetics, the average effect at a single locus can be estimated by an additive (A) model, or an additive plus dominance (AD) model. In the presence of dominance, the AD-model is expected to be more accurate, because the A-model falsely assumes that residuals are independent and identically distributed. Our objective was to investigate the accuracy of an estimated average effect ([Formula: see text]) in the presence of dominance, using either a single locus A-model or AD-model. Estimation was based on a finite sample from a large population in Hardy-Weinberg equilibrium (HWE), and the root mean squared error of [Formula: see text] was calculated for several broad-sense heritabilities, sample sizes, and sizes of the dominance effect. Results show that with the A-model, both sampling deviations of genotype frequencies from HWE frequencies and sampling deviations of allele frequencies contributed to the error. With the AD-model, only sampling deviations of allele frequencies contributed to the error, provided that all three genotype classes were sampled. In the presence of dominance, the root mean squared error of [Formula: see text] with the AD-model was always smaller than with the A-model, even when the heritability was less than one. Remarkably, in the absence of dominance, there was no disadvantage of fitting dominance. In conclusion, the AD-model yields more accurate estimates of average effects from a finite sample, because it is more robust against sampling deviations from HWE frequencies than the A-model. Genetic models that include dominance, therefore, yield higher accuracies of estimated average effects than purely additive models when dominance is present.