Inbreeding depression on growth and prolificacy traits in two lines of rabbit.

Inbreeding depression in closed populations impairs animal fitness, health, and productivity. However, not all inbreeding is expected to be equally damaging. Recent inbreeding is thought to be more harmful than ancient inbreeding because selection decreases the frequency of unfavourable alleles with time. Accordingly, selection efficiency is improved by inbreeding in a process called purging. This research aimed to quantify inbreeding depression on growth and prolificacy traits in two lines of rabbits selected for just one growth (Caldes line) or prolificacy (Prat line) trait, and also to find some evidence of purging of deleterious alleles by selection. Caldes line comprised 51 generations and 124,371 animals in the pedigree. Prat line comprised 34 generations and 161,039 animals in the pedigree. The effects of old, intermediate, and new inbreeding (Fold, Fint, and Fnew), as well as total cumulated classical inbreeding (F) and 3 measurements of ancestral inbreeding (AHC, Fa.K, and Fa.B) were estimated for average daily gain (ADG), slaughter weight (SW), weaning weight (WW), born alive (BA), the total number of kits (NT), and the number of weaned kits (NW). There was a clear inbreeding depression for all growth and prolificacy traits in the Caldes line (-7.19 g/d, -0.45 kg, -0.25 kg, -6 kits, -4 kits, and -4 kits per unit of increase in F for ADG, SW, WW, BA, NT, and NW, respectively) and also in Prat line (-7.48 g/d, -0.31 kg, -0.11 kg, -4 kits, -5 kits, and -4 kits per unit of increase in F for ADG, SW, WW, BA, NT, and NW, respectively). The inbreeding partition appears to be a reliable alternative for assessing inbreeding depression and purging. Thus, for example, in the Caldes line and for ADG the regression coefficients were -7.61, -5.41, and 7.76 g/d per unit of increase in Fnew, Fint, and Fold, respectively. In addition, AHC and Fa.B may provide more accurate evidence of purging than Fa.K. This study confirms the existence of inbreeding depression for growth and prolificacy traits in both lines of rabbits and shows evidence of purging of deleterious recessive alleles involved both in growth and prolificacy, independently of the selection criteria established in the line.

Impact of multi-output and stacking methods on feed efficiency prediction from genotype using machine learning algorithms.

Feeding represents the largest economic cost in meat production; therefore, selection to improve traits related to feed efficiency is a goal in most livestock breeding programs. Residual feed intake (RFI), that is, the difference between the actual and the expected feed intake based on animal's requirements, has been used as the selection criteria to improve feed efficiency since it was proposed by Kotch in 1963. In growing pigs, it is computed as the residual of the multiple regression model of daily feed intake (DFI), on average daily gain (ADG), backfat thickness (BFT), and metabolic body weight (MW). Recently, prediction using single-output machine learning algorithms and information from SNPs as predictor variables have been proposed for genomic selection in growing pigs, but like in other species, the prediction quality achieved for RFI has been generally poor. However, it has been suggested that it could be improved through multi-output or stacking methods. For this purpose, four strategies were implemented to predict RFI. Two of them correspond to the computation of RFI in an indirect way using the predicted values of its components obtained from (i) individual (multiple single-output strategy) or (ii) simultaneous predictions (multi-output strategy). The other two correspond to the direct prediction of RFI using (iii) the individual predictions of its components as predictor variables jointly with the genotype (stacking strategy), or (iv) using only the genotypes as predictors of RFI (single-output strategy). The single-output strategy was considered the benchmark. This research aimed to test the former three hypotheses using data recorded from 5828 growing pigs and 45,610 SNPs. For all the strategies two different learning methods were fitted: random forest (RF) and support vector regression (SVR). A nested cross-validation (CV) with an outer 10-folds CV and an inner threefold CV for hyperparameter tuning was implemented to test all strategies. This scheme was repeated using as predictor variables different subsets with an increasing number (from 200 to 3000) of the most informative SNPs identified with RF. Results showed that the highest prediction performance was achieved with 1000 SNPs, although the stability of feature selection was poor (0.13 points out of 1). For all SNP subsets, the benchmark showed the best prediction performance. Using the RF as a learner and the 1000 most informative SNPs as predictors, the mean (SD) of the 10 values obtained in the test sets were: 0.23 (0.04) for the Spearman correlation, 0.83 (0.04) for the zero-one loss, and 0.33 (0.03) for the rank distance loss. We conclude that the information on predicted components of RFI (DFI, ADG, MW, and BFT) does not contribute to improve the quality of the prediction of this trait in relation to the one obtained with the single-output strategy.

Disentangling the causal relationship between rabbit growth and cecal microbiota through structural equation models.

BACKGROUND: The effect of the cecal microbiome on growth of rabbits that were fed under different regimes has been studied previously. However, the term "effect" carries a causal meaning that can be confounded because of potential genetic associations between the microbiome and production traits. Structural equation models (SEM) can help disentangle such a complex interplay by decomposing the effect on a production trait into direct host genetics effects and indirect host genetic effects that are exerted through microbiota effects. These indirect effects can be estimated via structural coefficients that measure the effect of the microbiota on growth while the effects of the host genetics are kept constant. In this study, we applied the SEM approach to infer causal relationships between the cecal microbiota and growth of rabbits fed under ad libitum (ADGAL) or restricted feeding (ADGR). RESULTS: We identified structural coefficients that are statistically different from 0 for 138 of the 946 operational taxonomic units (OTU) analyzed. However, only 15 and 38 of these 138 OTU had an effect greater than 0.2 phenotypic standard deviations (SD) on ADGAL and ADGR, respectively. Many of these OTU had a negative effect on both traits. The largest effects on ADGR were exerted by an OTU that is taxonomically assigned to the Desulfovibrio genus (- 1.929 g/d, CSS-normalized OTU units) and by an OTU that belongs to the Ruminococcaceae family (1.859 g/d, CSS-normalized OTU units). For ADGAL, the largest effect was from OTU that belong to the S24-7 family (- 1.907 g/d, CSS-normalized OTU units). In general, OTU that had a substantial effect had low to moderate estimates of heritability. CONCLUSIONS: Disentangling how direct and indirect effects act on production traits is relevant to fully describe the processes of mediation but also to understand how these traits change before considering the application of an external intervention aimed at changing a given microbial composition by blocking/promoting the presence of a particular microorganism.

Use of Bayes factors to evaluate the effects of host genetics, litter and cage on the rabbit cecal microbiota.

BACKGROUND: The rabbit cecum hosts and interacts with a complex microbial ecosystem that contributes to the variation of traits of economic interest. Although the influence of host genetics on microbial diversity and specific microbial taxa has been studied in several species (e.g., humans, pigs, or cattle), it has not been investigated in rabbits. Using a Bayes factor approach, the aim of this study was to dissect the effects of host genetics, litter and cage on 984 microbial traits that are representative of the rabbit microbiota. RESULTS: Analysis of 16S rDNA sequences of cecal microbiota from 425 rabbits resulted in the relative abundances of 29 genera, 951 operational taxonomic units (OTU), and four microbial alpha-diversity indices. Each of these microbial traits was adjusted with mixed linear and zero-inflated Poisson (ZIP) models, which all included additive genetic, litter and cage effects, and body weight at weaning and batch as systematic factors. The marginal posterior distributions of the model parameters were estimated using MCMC Bayesian procedures. The deviance information criterion (DIC) was used for model comparison regarding the statistical distribution of the data (normal or ZIP), and the Bayes factor was computed as a measure of the strength of evidence in favor of the host genetics, litter, and cage effects on microbial traits. According to DIC, all microbial traits were better adjusted with the linear model except for the OTU present in less than 10% of the animals, and for 25 of the 43 OTU with a frequency between 10 and 25%. On a global scale, the Bayes factor revealed substantial evidence in favor of the genetic control of the number of observed OTU and Shannon indices. At the taxon-specific level, significant proportions of the OTU and relative abundances of genera were influenced by additive genetic, litter, and cage effects. Several members of the genera Bacteroides and Parabacteroides were strongly influenced by the host genetics and nursing environment, whereas the family S24-7 and the genus Ruminococcus were strongly influenced by cage effects. CONCLUSIONS: This study demonstrates that host genetics shapes the overall rabbit cecal microbial diversity and that a significant proportion of the taxa is influenced either by host genetics or environmental factors, such as litter and/or cage.

Analysis of the causal structure of traits involved in sow lactation feed efficiency.

BACKGROUND: Feed efficiency during lactation involves a set of phenotypic traits that form a complex system, with some traits exerting causal effects on the others. Information regarding such interrelationships can be used to predict the effect of external interventions on the system, and ultimately to optimize management practices and multi-trait selection strategies. Structural equation models can be used to infer the magnitude of the different causes of such interrelationships. The causal network necessary to fit structural equation models can be inferred using the inductive causation (IC) algorithm. By implementing these statistical tools, we inferred the causal association between the main energy sources and sinks involved in sow lactation feed efficiency for the first time, i.e., daily lactation feed intake (dLFI) in kg/day, daily sow weight balance (dSWB) in kg/day, daily litter weight gain (dLWG) in kg/day, daily back fat thickness balance (dBFTB) in mm/day, and sow metabolic body weight (SMBW) in kg0.75. Then, we tested several selection strategies based on selection indices, with or without dLFI records, to improve sow efficiency during lactation. RESULTS: The IC algorithm using 95% highest posterior density (HPD95%) intervals resulted in a fully directed acyclic graph, in which dLFI and dLWG affected dSWB, the posterior mean of the corresponding structural coefficients (PMλ) being 0.12 and - 0.03, respectively. In turn, dSWB influenced dBFTB and SMBW, with PMλ equal to 0.70 and - 1.22, respectively. Multiple indirect effects contributed to the variances and covariances among the analyzed traits, with the most relevant indirect effects being those involved in the association between dSWB and dBFTB and between dSWB and SMBW. Selection strategies with or without phenotypic information on dLFI, or that hold this trait constant, led to the same pattern and similar responses in dLFI, dSWB, and dLWG. CONCLUSIONS: Selection based on an index including only dBFTB and dLWG records can reduce dLFI, keep dSWB constant or increase it, and increase dLWG. However, a favorable response for all three traits is probably not achievable. Holding the amount of feed provided to the sows constant did not offer an advantage in terms of response over the other strategies.

Genotype by feeding regimen interactions for slaughter traits in rabbit and expected responses under restricted and full feeding.

The interaction between the genotype and feeding regimen (G×FR) for slaughter traits was estimated from data corresponding to 2557 animals under full (FF) and 2424 with restricted feeding (RF). Expected responses to selection under different scenario regarding feeding regimen were also calculated. Body weight at slaughter (SW), carcass weight (CW) and dressing out percentage (DoP) were analysed by using linear animal models in which records obtained under different feeding regimes were treated as different traits. Animals belonged to Caldes line, selected for average daily gain (G) under ad libitum feeding. The selection process information was included in the analyses. Marginal posterior mean of heritabilities were 0.102 for G, and 0.364, 0.257 and 0.167 for SW, CW and DoP under FF feeding. The corresponding values for animals fed on RF were 0.243, 0.203 and 0.379 for SW, CW and DoP, respectively. Genetic correlations between G and CW were positive and moderate, and those between G and DoP were low. The estimated genetic correlation between SW, CW and DoP under different feeding regimens were: 0.73, 0.69 and 0.87, respectively. These correlations cannot be said to be far enough from one to generate relevant G×FR interaction variance, which were estimated to be only 11.1%, 8.6% and 5.3% of the mean of the phenotypic variance for SW, CW and DoP, respectively. This lack of G×FR interaction variance, jointly with the higher heritability of DoP under RF, explains that the genetic improvement of DoP can be done more efficiently recording traits on animals under RF, even if the interest is on the performances under FF, i.e. by indirect selection.

Networks of inbreeding coefficients in a selected population of rabbits.

The correlation between pedigree and genomic-based inbreeding coefficients is usually discussed in the literature. However, some of these correlations could be spurious. Using partial correlations and information theory, it is possible to distinguish a significant association between two variables which is independent from associations with a third variable. The objective of this study is to implement partial correlations and information theory to assess the relationship between different inbreeding coefficients using a selected population of rabbits. Data from pedigree and genomic information from a 200K SNP chip were available. After applying filtering criteria, the data set comprised 437 animals genotyped for 114,604 autosomal SNP. Fifteen pedigree- and genome-based inbreeding coefficients were estimated and used to build a network. Recent inbreeding coefficient based on runs of homozygosity had 9 edges linking it with different inbreeding coefficients. Partial correlations and information theory approach allowed to infer meaningful associations between inbreeding coefficients and highlighted the importance of the recent inbreeding based on runs of homozygosity, but a good proxy of it could be those pedigree-based definitions reflecting recent inbreeding.

Machine learning applied to transcriptomic data to identify genes associated with feed efficiency in pigs.

BACKGROUND: To date, the molecular mechanisms that underlie residual feed intake (RFI) in pigs are unknown. Results from different genome-wide association studies and gene expression analyses are not always consistent. The aim of this research was to use machine learning to identify genes associated with feed efficiency (FE) using transcriptomic (RNA-Seq) data from pigs that are phenotypically extreme for RFI. METHODS: RFI was computed by considering within-sex regression on mean metabolic body weight, average daily gain, and average backfat gain. RNA-Seq analyses were performed on liver and duodenum tissue from 32 high and 33 low RFI pigs collected at 153 d of age. Machine-learning algorithms were used to predict RFI class based on gene expression levels in liver and duodenum after adjusting for batch effects. Genes were ranked according to their contribution to the classification using the permutation accuracy importance score in an unbiased random forest (RF) algorithm based on conditional inference. Support vector machine, RF, elastic net (ENET) and nearest shrunken centroid algorithms were tested using different subsets of the top rank genes. Nested resampling for hyperparameter tuning was implemented with tenfold cross-validation in the outer and inner loops. RESULTS: The best classification was obtained with ENET using the expression of 200 genes in liver [area under the receiver operating characteristic curve (AUROC): 0.85; accuracy: 0.78] and 100 genes in duodenum (AUROC: 0.76; accuracy: 0.69). Canonical pathways and candidate genes that were previously reported as associated with FE in several species were identified. The most remarkable pathways and genes identified were NRF2-mediated oxidative stress response and aldosterone signalling in epithelial cells, the DNAJC6, DNAJC1, MAPK8, PRKD3 genes in duodenum, and melatonin degradation II, PPARα/RXRα activation, and GPCR-mediated nutrient sensing in enteroendocrine cells and SMOX, IL4I1, PRKAR2B, CLOCK and CCK genes in liver. CONCLUSIONS: ML algorithms and RNA-Seq expression data were found to provide good performance for classifying pigs into high or low RFI groups. Classification was better with gene expression data from liver than from duodenum. Genes associated with FE in liver and duodenum tissue that can be used as predictive biomarkers for this trait were identified.

Use of group records of feed intake to select for feed efficiency in rabbit.

Models for genetic evaluation of feed efficiency (FE) for animals housed in groups when they are either fed ad libitum (F) or on restricted (R) feeding were implemented. Definitions of FE on F included group records of feed intake ( FI ¯ F ) and individual records of growth rate (GF ) and metabolic weight (MF ). Growth rate (GR ) as FE measurement on R was used. Data corresponded to 5,336 kits from a rabbit sire line, from 1,255 litters in 14 batches and 667 cages. A five-trait mixed model (also with metabolic weight on R, MR ) was implemented including, for each trait, the systematic effects of batch, body weight at weaning, parity order and litter size; and the random effects of litter, additive genetic and individual. A Bayesian analysis was performed. Conditional traits such as FI ¯ F | M F , G F and G F | M F , FI ¯ F were obtained from elements of additive genetics ( FI ¯ F | M F , G F g and G F | M F , FI ¯ F g ) or phenotypic ( FI ¯ F | M F , G F p and G F | M F , FI ¯ F p ) (co)variance matrices. In the first case, heritabilities were low (0.07 and 0.06 for FI ¯ F | M F , G F g and G F | M F , FI ¯ F g , respectively) but null genetic correlation between the conditional and conditioning traits is guaranteed. In the second case, heritabilities were higher (0.22 and 0.16 for FI ¯ F | M F , G F p and G F | M F , FI ¯ F p , respectively) but the genetic correlation between FI ¯ F | M F , G F p and G F was moderate (0.58). Heritability of GR was low (0.08). This trait was negatively correlated with G F | M F , FI ¯ F p and G F | M F , FI ¯ F g of animals on F, which indicate a different genetic background. The correlation between GR and GF was also low to moderate (0.48) and the additive variance of GF was almost four times that of GR , suggesting the presence of a substantial genotype by feeding regimen interaction.

Longitudinal analysis of direct and indirect effects on average daily gain in rabbits using a structured antedependence model.

BACKGROUND: Indirect genetic effects (IGE) are important components of various traits in several species. Although the intensity of social interactions between partners likely vary over time, very few genetic studies have investigated how IGE vary over time for traits under selection in livestock species. To overcome this issue, our aim was: (1) to analyze longitudinal records of average daily gain (ADG) in rabbits subjected to a 5-week period of feed restriction using a structured antedependence (SAD) model that includes IGE and (2) to evaluate, by simulation, the response to selection when IGE are present and genetic evaluation is based on a SAD model that includes IGE or not. RESULTS: The direct genetic variance for ADG (g/d) increased from week 1 to 3 [from 8.03 to 13.47 (g/d)2] and then decreased [6.20 (g/d)2 at week 5], while the indirect genetic variance decreased from week 1 to 4 [from 0.43 to 0.22 (g/d)2]. The correlation between the direct genetic effects of different weeks was moderate to high (ranging from 0.46 to 0.86) and tended to decrease with time interval between measurements. The same trend was observed for IGE for weeks 2 to 5 (correlations ranging from 0.62 to 0.91). Estimates of the correlation between IGE of week 1 and IGE of the other weeks did not follow the same pattern and correlations were lower. Estimates of correlations between direct and indirect effects were negative at all times. After seven generations of simulated selection, the increase in ADG from selection on EBV from a SAD model that included IGE was higher (~ 30%) than when those effects were omitted. CONCLUSIONS: Indirect genetic effects are larger just after mixing animals at weaning than later in the fattening period, probably because of the establishment of social hierarchy that is generally observed at that time. Accounting for IGE in the selection criterion maximizes genetic progress.

Genetic parameters and expected responses to selection for components of feed efficiency in a Duroc pig line.

BACKGROUND: Improving feed efficiency ([Formula: see text]) is a key factor for any pig breeding company. Although this can be achieved by selection on an index of multi-trait best linear unbiased prediction of breeding values with optimal economic weights, considering deviations of feed intake from actual needs ([Formula: see text]) should be of value for further research on biological aspects of [Formula: see text]. Here, we present a random regression model that extends the classical definition of [Formula: see text] by including animal-specific needs in the model. Using this model, we explore the genetic determinism of several [Formula: see text] components: use of feed for growth ([Formula: see text]), use of feed for backfat deposition ([Formula: see text]), use of feed for maintenance ([Formula: see text]), and unspecific efficiency in the use of feed ([Formula: see text]). Expected response to alternative selection indexes involving different components is also studied. RESULTS: Based on goodness-of-fit to the available feed intake ([Formula: see text]) data, the model that assumes individual (genetic and permanent) variation in the use of feed for maintenance, [Formula: see text] and [Formula: see text] showed the best performance. Joint individual variation in feed allocation to maintenance, growth and backfat deposition comprised 37% of the individual variation of [Formula: see text]. The estimated heritabilities of [Formula: see text] using the model that accounts for animal-specific needs and the traditional [Formula: see text] model were 0.12 and 0.18, respectively. The estimated heritabilities for the regression coefficients were 0.44, 0.39 and 0.55 for [Formula: see text], [Formula: see text] and [Formula: see text], respectively. Estimates of genetic correlations of [Formula: see text] were positive with amount of feed used for [Formula: see text] and [Formula: see text] but negative for [Formula: see text]. Expected response in overall efficiency, reducing [Formula: see text] without altering performance, was 2.5% higher when the model assumed animal-specific needs than when the traditional definition of [Formula: see text] was considered. CONCLUSIONS: Expected response in overall efficiency, by reducing [Formula: see text] without altering performance, is slightly better with a model that assumes animal-specific needs instead of batch-specific needs to correct [Formula: see text]. The relatively small difference between the traditional [Formula: see text] model and our model is due to random intercepts (unspecific use of feed) accounting for the majority of variability in [Formula: see text]. Overall, a model that accounts for animal-specific needs for [Formula: see text], [Formula: see text] and [Formula: see text] is statistically superior and allows for the possibility to act differentially on [Formula: see text] components.

Interaction of direct and social genetic effects with feeding regime in growing rabbits.

BACKGROUND: Most rabbit production farms apply feed restriction at fattening because of its protective effect against digestive diseases that affect growing rabbits. However, it leads to competitive behaviour between cage mates, which is not observed when animals are fed ad libitum. Our aim was to estimate the contribution of direct ([Formula: see text]) and social ([Formula: see text]) genetic effects (also known as indirect genetic effects) to total heritable variance of average daily gain ([Formula: see text]) in rabbits on different feeding regimens (FR), and the magnitude of the interaction between genotype and FR (G × FR). METHODS: A total of 6264 contemporary kits were housed in cages of eight individuals and raised on full ([Formula: see text]) or restricted ([Formula: see text]) feeding to 75% of the ad libitum intake. A Bayesian analysis of weekly records of [Formula: see text] (from 32 to 60 days of age) in rabbits on [Formula: see text] and [Formula: see text] was performed with a two-trait model including [Formula: see text] and [Formula: see text]. RESULTS: The ratio between total heritable variance and phenotypic variance ([Formula: see text]) was low (<0.10) and did not differ significantly between FR. However, the ratio between [Formula: see text] (i.e. variance of [Formula: see text] relative to phenotypic variance) and [Formula: see text] was ~0.52 and 0.86 for animals on [Formula: see text] and [Formula: see text], respectively, thus [Formula: see text] contributed more to the heritable variance of animals on [Formula: see text] than on [Formula: see text]. Feeding regimen also affected the sign and magnitude of the correlation between [Formula: see text] and [Formula: see text], i.e. -0.5 and ~0 for animals on [Formula: see text] and [Formula: see text], respectively. The posterior mean (posterior sd) of the correlation between estimated total breeding values (ETBV) of animals on [Formula: see text] and [Formula: see text] was 0.26 (0.20), indicating very strong G × FR interactions. The correlations between [Formula: see text] and [Formula: see text] in rabbits on [Formula: see text] and [Formula: see text] ranged from -0.47 ([Formula: see text] on [Formula: see text] and [Formula: see text] on [Formula: see text]) to 0.64. CONCLUSIONS: Our results suggest that selection of rabbits for [Formula: see text] under [Formula: see text] may completely fail to improve [Formula: see text] in rabbits on [Formula: see text]. Social genetic effects contribute substantially to ETBV of rabbits on [Formula: see text] but not on [Formula: see text]. Selection for [Formula: see text] should be performed under production conditions regarding the FR, by accounting for [Formula: see text] if the amount of food is limited.

Metabolic activity of sperm cells: correlation with sperm cell concentration, viability and motility in the rabbit.

The resazurin reduction test (RRT) is a useful technique to assess the metabolic rate of sperm cells. RRT depends on the ability of metabolically active cells to reduce the non-fluorescent dye resazurin to the fluorescent resorufin. The aim of this study was to develop a vital fluorometric method to evaluate metabolic activity of rabbit sperm cells. Twenty-five rabbit males were included in the study. Viability and morphology, motility and metabolic activity were evaluated using an eosin-nigrosin staining, a computer-assisted semen analysis (CASA) and the RRT, respectively. Spearman rank correlation analysis was used to determine the correlation between RRT and semen parameters. After evaluation, a concentration of 10 × 106 sperm cells/ml was selected for further experiments with RRT. No significant correlation was found between the RRT results and the motility parameters. However, after RRT a significant positive correlation between relative fluorescence units and the percentage of alive spermatozoa (r = 0.62; P = 0.001) and a negative one with the percentage of sperm cells with acrosomic abnormalities (r = -0.45; P < 0.05) were detected. The vital assessment of metabolic rate of sperm cells by RRT could provide more information about semen quality than other routine semen analysis, correlating with sperm viability and acrosome status information.

Integrating computer vision algorithms and RFID system for identification and tracking of group-housed animals: an example with pigs.

Precision livestock farming aims to individually and automatically monitor animal activity to ensure their health, well-being, and productivity. Computer vision has emerged as a promising tool for this purpose. However, accurately tracking individuals using imaging remains challenging, especially in group housing where animals may have similar appearances. Close interaction or crowding among animals can lead to the loss or swapping of animal IDs, compromising tracking accuracy. To address this challenge, we implemented a framework combining a tracking-by-detection method with a radio frequency identification (RFID) system. We tested this approach using twelve pigs in a single pen as an illustrative example. Three of the pigs had distinctive natural coat markings, enabling their visual identification within the group. The remaining pigs either shared similar coat color patterns or were entirely white, making them visually indistinguishable from each other. We employed the latest version of the You Only Look Once (YOLOv8) and BoT-SORT algorithms for detection and tracking, respectively. YOLOv8 was fine-tuned with a dataset of 3,600 images to detect and classify different pig classes, achieving a mean average precision of all the classes of 99%. The fine-tuned YOLOv8 model and the tracker BoT-SORT were then applied to a 166.7-min video comprising 100,018 frames. Results showed that pigs with distinguishable coat color markings could be tracked 91% of the time on average. For pigs with similar coat color, the RFID system was used to identify individual animals when they entered the feeding station, and this RFID identification was linked to the image trajectory of each pig, both backward and forward. The two pigs with similar markings could be tracked for an average of 48.6 min, while the seven white pigs could be tracked for an average of 59.1 min. In all cases, the tracking time assigned to each pig matched the ground truth 90% of the time or more. Thus, our proposed framework enabled reliable tracking of group-housed pigs for extended periods, offering a promising alternative to the independent use of image or RFID approaches alone. This approach represents a significant step forward in combining multiple devices for animal identification, tracking, and traceability, particularly when homogeneous animals are kept in groups.

In precision livestock farming, monitoring animal activity is crucial to ensure their health, well-being, and productivity. While digital cameras and computer vision algorithms offer a promising solution for this task, tracking individual animals of similar appearance when housed in groups can be challenging. Close interaction among animals can lead to a loss of individual identity, which affects tracking accuracy. To overcome this problem, we developed a framework that combines camera images with radio frequency identification (RFID) ear tags. This methodology was applied to a pen housing 12 pigs, with an RFID reader located inside the feeder. Among the pigs, three had unique coat markings, enabling them to be tracked most of the time without losing their identity (87% of the time). The remaining pigs could not be visually distinguished from each other, so information from the RFID system was used to recover lost IDs every time pigs entered the feeder. The framework achieves 97% accuracy in tracking, offering a reliable solution for monitoring group-housed pigs.

Feature Selection Stability and Accuracy of Prediction Models for Genomic Prediction of Residual Feed Intake in Pigs Using Machine Learning.

Feature selection (FS, i.e., selection of a subset of predictor variables) is essential in high-dimensional datasets to prevent overfitting of prediction/classification models and reduce computation time and resources. In genomics, FS allows identifying relevant markers and designing low-density SNP chips to evaluate selection candidates. In this research, several univariate and multivariate FS algorithms combined with various parametric and non-parametric learners were applied to the prediction of feed efficiency in growing pigs from high-dimensional genomic data. The objective was to find the best combination of feature selector, SNP subset size, and learner leading to accurate and stable (i.e., less sensitive to changes in the training data) prediction models. Genomic best linear unbiased prediction (GBLUP) without SNP pre-selection was the benchmark. Three types of FS methods were implemented: (i) filter methods: univariate (univ.dtree, spearcor) or multivariate (cforest, mrmr), with random selection as benchmark; (ii) embedded methods: elastic net and least absolute shrinkage and selection operator (LASSO) regression; (iii) combination of filter and embedded methods. Ridge regression, support vector machine (SVM), and gradient boosting (GB) were applied after pre-selection performed with the filter methods. Data represented 5,708 individual records of residual feed intake to be predicted from the animal's own genotype. Accuracy (stability of results) was measured as the median (interquartile range) of the Spearman correlation between observed and predicted data in a 10-fold cross-validation. The best prediction in terms of accuracy and stability was obtained with SVM and GB using 500 or more SNPs [0.28 (0.02) and 0.27 (0.04) for SVM and GB with 1,000 SNPs, respectively]. With larger subset sizes (1,000-1,500 SNPs), the filter method had no influence on prediction quality, which was similar to that attained with a random selection. With 50-250 SNPs, the FS method had a huge impact on prediction quality: it was very poor for tree-based methods combined with any learner, but good and similar to what was obtained with larger SNP subsets when spearcor or mrmr were implemented with or without embedded methods. Those filters also led to very stable results, suggesting their potential use for designing low-density SNP chips for genome-based evaluation of feed efficiency.

The value of gut microbiota to predict feed efficiency and growth of rabbits under different feeding regimes.

Gut microbiota plays an important role in nutrient absorption and could impact rabbit feed efficiency. This study aims at investigating such impact by evaluating the value added by microbial information for predicting individual growth and cage phenotypes related to feed efficiency. The dataset comprised individual average daily gain and cage-average daily feed intake from 425 meat rabbits, in which cecal microbiota was assessed, and their cage mates. Despite microbiota was not measured in all animals, consideration of pedigree relationships with mixed models allowed the study of cage-average traits. The inclusion of microbial information into certain mixed models increased their predictive ability up to 20% and 46% for cage-average feed efficiency and individual growth traits, respectively. These gains were associated with large microbiability estimates and with reductions in the heritability estimates. However, large microbiabililty estimates were also obtained with certain models but without any improvement in their predictive ability. A large proportion of OTUs seems to be responsible for the prediction improvement in growth and feed efficiency traits, although specific OTUs taxonomically assigned to 5 different phyla have a higher weight. Rabbit growth and feed efficiency are influenced by host cecal microbiota, thus considering microbial information in models improves the prediction of these complex phenotypes.

Machine Learning Prediction of Crossbred Pig Feed Efficiency and Growth Rate From Single Nucleotide Polymorphisms.

This research assessed the ability of a Support Vector Machine (SVM) regression model to predict pig crossbred (CB) performance from various sources of phenotypic and genotypic information for improving crossbreeding performance at reduced genotyping cost. Data consisted of average daily gain (ADG) and residual feed intake (RFI) records and genotypes of 5,708 purebred (PB) boars and 5,007 CB pigs. Prediction models were fitted using individual PB genotypes and phenotypes (trn.1); genotypes of PB sires and average of CB records per PB sire (trn.2); and individual CB genotypes and phenotypes (trn.3). The average of CB offspring records was the trait to be predicted from PB sire's genotype using cross-validation. Single nucleotide polymorphisms (SNPs) were ranked based on the Spearman Rank correlation with the trait. Subsets with an increasing number (from 50 to 2,000) of the most informative SNPs were used as predictor variables in SVM. Prediction performance was the median of the Spearman correlation (SC, interquartile range in brackets) between observed and predicted phenotypes in the testing set. The best predictive performances were obtained when sire phenotypic information was included in trn.1 (0.22 [0.03] for RFI with SVM and 250 SNPs, and 0.12 [0.05] for ADG with SVM and 500-1,000 SNPs) or when trn.3 was used (0.29 [0.16] with Genomic best linear unbiased prediction (GBLUP) for RFI, and 0.15 [0.09] for ADG with just 50 SNPs). Animals from the last two generations were assigned to the testing set and remaining animals to the training set. Individual's PB own phenotype and genotype improved the prediction ability of CB offspring of young animals for ADG but not for RFI. The highest SC was 0.34 [0.21] and 0.36 [0.22] for RFI and ADG, respectively, with SVM and 50 SNPs. Predictive performance using CB data for training leads to a SC of 0.34 [0.19] with GBLUP and 0.28 [0.18] with SVM and 250 SNPs for RFI and 0.34 [0.15] with SVM and 500 SNPs for ADG. Results suggest that PB candidates could be evaluated for CB performance with SVM and low-density SNP chip panels after collecting their own RFI or ADG performances or even earlier, after being genotyped using a reference population of CB animals.

Breeding farm, level of feeding and presence of antibiotics in the feed influence rabbit cecal microbiota.

BACKGROUND: The effect of the production environment and different management practices in rabbit cecal microbiota remains poorly understood. While previous studies have proved the impact of the age or the feed composition, research in the breeding farm and other animal management aspects, such as the presence of antibiotics in the feed or the level of feeding, is still needed. Characterization of microbial diversity and composition of growing rabbits raised under different conditions could help better understand the role these practices play in cecal microbial communities and how it may result in different animal performance. RESULTS: Four hundred twenty-five meat rabbits raised in two different facilities, fed under two feeding regimes (ad libitum or restricted) with feed supplemented or free of antibiotics, were selected for this study. A 16S rRNA gene-based assessment through the MiSeq Illumina sequencing platform was performed on cecal samples collected from these individuals at slaughter. Different univariate and multivariate approaches were conducted to unravel the influence of the different factors on microbial alpha diversity and composition at phylum, genus and OTU taxonomic levels. The animals raised in the facility harboring the most stable environmental conditions had greater, and less variable, microbial richness and diversity. Bootstrap univariate analyses of variance and sparse partial least squares-discriminant analyses endorsed that farm conditions exerted an important influence on rabbit microbiota since the relative abundances of many taxa were found differentially represented between both facilities at all taxonomic levels characterized. Furthermore, only five OTUs were needed to achieve a perfect classification of samples according to the facility where animals were raised. The level of feeding and the presence of antibiotics did not modify the global alpha diversity but had an impact on some bacteria relative abundances, albeit in a small number of taxa compared with farm, which is consistent with the lower sample classification power according to these factors achieved using microbial information. CONCLUSIONS: This study reveals that factors associated with the farm effect and other management factors, such as the presence of antibiotics in the diet or the feeding level, modify cecal microbial communities. It highlights the importance of offering a controlled breeding environment that reduces differences in microbial cecal composition that could be responsible for different animal performance.

Bayesian hierarchical model for comparison of different nonlinear function and genetic parameter estimates of meat quails.

This study aimed to compare different nonlinear functions to describe the growth curve of European quails and to estimate growth curve parameters, (co)variance components, and genetic and systematic effects that affected the curve using a hierarchical Bayesian model that allows joint estimation. Three different models were fitted in the first stage (Gompertz, Logístic, and von Bertalanffy). The analyzed data set had 45,965 records of 6,838 meat quails selected for higher body weight at 42 d of age for 15 successive generations, weighed at birth, 7, 14, 21, 28, 35, and 42 d of age. Comparisons of the overall goodness of fit were based on deviance information criterion (DIC) and mean square error. Gelfand's check function compared the models at different points of the growth curve. In the second stage, the systematic (sex and generation) and genetic effects were considered in an animal model. Random samples of the a posteriori distributions were obtained by Metropolis-Hastings and Gibbs sampling algorithms. The Gompertz function presented lower DIC and better adjustment at different ages and was defined as the best fit. The heritabilities of A, b, and k parameters were moderate (0.32, 0.29, and 0.18, respectively). The genetics correlations were A and b (0.25), A and k (-0.50), and b and k (0.03). The samples of the posterior marginal distributions for the differences between the estimates of the parameters of the Gompertz model, for generation, A, b, k, age at inflexion point (APOI), and weight at inflexion point (WPOI) showed differences in relation to sex, the females are heavier, A, WPOI, and APOI for females were also higher. In conclusion, 15 generations of selection and changes in the environmental conditions altered the growth curve, leaving the quails heavier and with greater WPOI and APOI, decreased growth rate, and increased the birth weight. The curve parameters could be used in a selection index, despite the difficulty in selecting quails with higher rate of growth and adult body weight.

Rabbit Microbiota Changes Throughout the Intestinal Tract.

To gain insight into the importance of carefully selecting the sampling area for intestinal microbiota studies, cecal and fecal microbial communities of Caldes meat rabbit were characterized. The animals involved in the study were divided in two groups according to the feed intake level they received during the fattening period; ad libitum (n = 10) or restricted to 75% of ad libitum intake (n = 11). Cecum and internal hard feces were sampled from sacrificed animals. Assessment of bacterial and archaeal populations was performed by means of Illumina sequencing of 16S rRNA gene amplicons in a MiSeq platform. A total of 596 operational taxonomic units (OTUs) were detected using QIIME software. Taxonomic assignment revealed that microbial diversity was dominated by phyla Firmicutes (76.42%), Tenericutes (7.83%), and Bacteroidetes (7.42%); kingdom Archaea was presented at low percentage (0.61%). No significant differences were detected between sampling origins in microbial diversity or richness assessed using two alpha-diversity indexes: Shannon and the observed number of OTUs. However, the analysis of variance at genus level revealed a higher presence of genera Clostridium, Anaerofustis, Blautia, Akkermansia, rc4-4, and Bacteroides in cecal samples. By contrast, genera Oscillospira and Coprococcus were found to be overrepresented in feces, suggesting that bacterial species of these genera would act as fermenters at the end of feed digestion process. At the lowest taxonomic level, 83 and 97 OTUs in feces and cecum, respectively, were differentially represented. Multivariate statistical assessment revealed that sparse partial least squares discriminant analysis (sPLS-DA) was the best approach for this purpose. Interestingly, the majority of the most discriminative OTUs selected by sPLS-DA were found to be differentially represented between sampling origins in univariate analysis. Our study provides evidence that the choice of intestinal sampling area is relevant due to important differences in some taxa's relative abundance that have been revealed between rabbits' cecal and fecal microbiota. An appropriate sampling intestinal area should be chosen in each microbiota assessment.