Comparison of the data mining and machine learning algorithms for predicting the final body weight for Romane sheep breed

The current study aimed to predict final body weight (weight of fourth months of age to select the future reproducers) by using birth weight, birth type, sex, suckling weight, age at suckling weight, weaning weight, age at weaning weight, and age of final body weight for the Romane sheep breed. For this purpose, classification and regression tree (CART), multivariate adaptive regression splines (MARS), and support vector machine regression (SVR) algorithms were used for training (80%) and testing (20%) sets. Different data mining and machine learning algorithms were used to predict final body weight of 393 Romane sheep (238 female and 155 male animals) were used with different artificial intelligence algorithms. The best prediction model was obtained by CART model, both training and testing set. Constructed CART models indicated that sex, suckling weight, weaning weight, age of weaning weight, and age of final weight could be used as an indirect selection measure to get a superior sheep flock on the final body weight of Romane sheep. If genetically established, the Romane sheep whose sex is female, age of final weight is over 142 days, and weaning weight is over 28 kg could be chosen for affording genetic improvement in final body weight. In conclusion, the usage of CART procedure may be worthy of reflection for identifying breed standards and choosing superior sheep for meat yield in France.

Predicting feed efficiency traits in growing lambs from their ruminal microbiota

Selecting feed-efficient sheep could improve the sustainability of this livestock production. However, most sheep breeding companies cannot afford to record feed intake to select feed-efficient animals. Past studies underlined the potential of omics data, including microbiota metabarcoding data, as proxies for feed efficiency. The study involved 277 Romane lambs from two lines divergently selected for residual feed intake (RFI). There were two objectives: check the consequences of selecting for feed efficiency over the rumen microbiota, and assess the predictive ability of the rumen microbiota for host traits. The study assessed two contrasting diets (concentrate diet and mixed diet) and two microbial groups (prokaryotes and eukaryotes). Discriminant analyses did not highlight any significant effect of sheep selection for residual feed intake on the rumen microbiota composition. Indeed, prokaryotic and eukaryotic microbiota compositions poorly discriminated the RFI lines, with averaged balanced error rates ranging from 45% to 55%. Correlations between host traits (feed efficiency and production traits) and their predictions from microbiota data varied between -0.07 and 0.56, depending on the trait, diet and sequencing. Feed intake was the most accurately predicted trait. However, predictions from fixed effects and BW were more accurate than or as accurate as predictions from the microbiota. Environmental effects can greatly affect the variability of microbiota compositions. Considering batch and environmental effects should be paramount when the predictive ability of the microbiota is assessed. This study argues why metabarcoding the rumen microbiota is not the best way to predict meat sheep production traits: fixed effects and BW were more cost-effective proxies and they led to similar or better predictive accuracies than microbiota metabarcoding (16S and 18S sequencing).

The cost of host genetic resistance on body condition: evidence from divergently selected sheep

Trade‐offs between host resistance to parasites and host growth or reproduction can occur due to allocation of limited available resources between competing demands. To predict potential trade‐offs arising from genetic selection for host resistance, a better understanding of the associated nutritional costs is required. Here, we studied resistance costs by using sheep from lines divergently selected on their resistance to a common blood‐feeding gastro‐intestinal parasite (Haemonchus contortus). First, we assessed the effects of selection for high or low host resistance on condition traits (body weight, back fat, and muscle thickness) and infection traits (parasite fecal egg excretion and loss in blood haematocrit) at various life stages, in particular during the periparturient period when resource allocation to immunity may limit host resistance. Second, we analysed the condition–infection relationship to detect a possible trade‐off, in particular during the periparturient period. We experimentally infected young females in four stages over their first 2 years of life, including twice around parturition (at 1 year and at 2 years of age). Linear mixed‐model analyses revealed a large and consistent between‐line difference in infection traits during growth and outside of the periparturient period, whereas this difference was strongly attenuated during the periparturient period. Despite their different responses to infection, lines had similar body condition traits. Using covariance decomposition, we then found that the phenotypic relationship between infection and condition was dominated by direct infection costs arising from parasite development within the host. Accounting for these within‐individual effects, a cost of resistance on body weight was detected among ewes during their first reproduction. Although this cost and the reproductive constraint on resistance are unlikely to represent a major concern for animal breeding in nutrient‐rich environments, this study provides important new insights regarding the nutritional costs of parasite resistance at different lifestages and how these may affect response to selection.

Evaluation of the Links between Lamb Feed Efficiency and Rumen and Plasma Metabolomic Data

Feed efficiency is one of the keystones that could help make animal production less costly and more environmentally friendly. Residual feed intake (RFI) is a widely used criterion to measure feed efficiency by regressing intake on the main energy sinks. We investigated rumen and plasma metabolomic data on Romane male lambs that had been genetically selected for either feed efficiency (line rfi−) or inefficiency (line rfi+). These investigations were conducted both during the growth phase under a 100% concentrate diet and later on under a mixed diet to identify differential metabolite expression and to link it to biological phenomena that could explain differences in feed efficiency. Nuclear magnetic resonance (NMR) data were analyzed using partial least squares discriminant analysis (PLS-DA), and correlations between metabolites’ relative concentrations were estimated to identify relationships between them. High levels of plasma citrate and malate were associated with genetically efficient animals, while high levels of amino acids such as L-threonine, L-serine, and L-leucine as well as beta-hydroxyisovalerate were associated with genetically inefficient animals under both diets. The two divergent lines could not be discriminated using rumen metabolites. Based on phenotypic residual feed intake (RFI), efficient and inefficient animals were discriminated using plasma metabolites determined under a 100% concentrate diet, but no discrimination was observed with plasma metabolites under a mixed diet or with rumen metabolites regardless of diet. Plasma amino acids, citrate, and malate were the most discriminant metabolites, suggesting that protein turnover and the mitochondrial production of energy could be the main phenomena that differ between efficient and inefficient Romane lambs.

Variability in Global DNA Methylation Rate Across Tissues and Over Time in Sheep

Recent studies highlighted the influence of epigenetic marks in the variability of many complex traits, both in plants and animals. These studied focused only on specific sites of the genome having differentially methylated profiles among individuals and/or tissues. In contrast, we recently used the methylation rate of the entire genome as a unique measure considered as a novel quantitative phenotype in sheep. This phenotype named global DNA methylation rate (GDMR), measured by luminometric assay, integrates the methylation level of each CpG dinucleotide within the 6 million of CCGG sites along the ovine genome. GDMR measured in blood previously showed moderate heritability of 0.20 and provided evidence for a genetic determinism. The main objective of the present study was to better characterize the GDMR phenotype in various tissues and investigate its variability in several breeds of sheep reared in the same environment. GDMR was measured on blood samples collected monthly from 59 growing male and female lambs (24 Romane, 23 Blackbelly and 12 Charollais), between birth and 4 months of age. Blood GDMR was on average around 80% and was influenced by the sampling date (p < 0.001), the breed (p = 0.002) and the sex (p = 0.002). In addition, GDMR was determined in 12 somatic (frontal lobe, pituitary gland, heart, lung, sub cutaneous and perirenal adipose tissue, skeletal muscle, liver, spleen, adrenal gland, medulla and cortical kidney) and 6 reproductive tissues (ovary, oviduct, uterus, testis, epididymis and seminal vesicle). GDMR was on average 70% in somatic tissues but marked variation was observed depending on the tissue. The GDMR measured in blood was higher than that measured in other somatic tissues, and is not a good proxy of less accessible tissues. Female reproductive tissues had a 10% higher GDMR than male reproductive tissues. We demonstrated a significant influence of the breed on blood GDMR, certainly reflecting the influence of different genetic backgrounds. The effect of the breed on GDMR may be related to their specific abilities to adapt to and live in different conditions.

Apart From the Diet, the Ruminal Microbiota of Lambs Is Modified in Relation to Their Genetic Potential for Feed Efficiency or Feeding Behavior

Using two successive types of diets (100% concentrate and 67% forage), this study explores the relationship between the ruminal microbiota of 78 Romane lambs and their feed efficiency (residual feed intake trait) or feeding behavior (feeding rate trait). Analysis was carried out phenotypically by correlating feed efficiency or feeding behavior traits with the relative abundance of bacteria at the phylum, family, and genus levels, and then genetically by comparing the microbiota of lambs selected for extreme breeding values for residual feed intake or feeding rate. Our results confirmed the major effect of diet on the ruminal microbiota composition. The microbiota of lambs consuming a forage-based diet was distinguished by higher microbial diversity and also by higher relative abundance of Firmicutes, whereas Bacteriodetes and Actinobacteria were relatively more abundant in the microbiota of lambs consuming a concentrate-based diet. Moreover, the comparison of lambs divergent for residual feed intake breeding values revealed that regardless of diet, more efficient lambs possessed a ruminal microbiota enriched in Coprococcus, Moryella, [Eubacterium] Brachy group, and [Eubacterium] hallii group, but depleted in Lachnospiraceae FD2005 and Shuttleworthia. The connection between microbiota composition and feeding rate was more tenuous, with no link between the abundance of particular genera and lambs genetically divergent for feeding rate.

Genetic × environment variation in sheep lines bred for divergent resistance to strongyle infection

Drug-resistant parasites threaten livestock production. Breeding more resistant hosts could be a sustainable control strategy. Environmental variation linked to animal management practices or to parasite species turnover across farms may however alter the expression of genetic potential. We created sheep lines with high or low resistance to Haemonchus contortus and achieved significant divergence on both phenotypic and genetic scales. We exposed both lines to chronic stress or to the infection by another parasite Trichostrongylus colubriformis, to test for genotype-by-environment and genotype-by-parasite species interactions respectively. Between-line divergence remained significant following chronic stress exposure although between-family variation was found. Significant genotype-by-parasite interaction was found although H. contortus-resistant lambs remained more resistant against T. colubriformis. Growth curves were not altered by the selection process although resistant lambs were lighter after the second round of divergence, before any infection took place. Breeding for resistance is a sustainable strategy but allowance needs to be made for environmental perturbations and worm species.

Detection of unrecorded environmental challenges in high-frequency recorded traits, and genetic determinism of resilience to challenge, with an application on feed intake in lambs

BACKGROUND

Resilient animals can remain productive under different environmental conditions. Rearing in increasingly heterogeneous environmental conditions increases the need of selecting resilient animals. Detection of environmental challenges that affect an entire population can provide a unique opportunity to select animals that are more resilient to these events. The objective of this study was two-fold: (1) to present a simple and practical data-driven approach to estimate the probability that, at a given date, an unrecorded environmental challenge occurred; and (2) to evaluate the genetic determinism of resilience to such events.

METHODS

Our method consists of inferring the existence of highly variable days (indicator of environmental challenges) via mixture models applied to frequently recorded phenotypic measures and then using the inferred probabilities of the occurrence of an environmental challenge in a reaction norm model to evaluate the genetic determinism of resilience to these events. These probabilities are estimated for each day (or other time frame). We illustrate the method by using an ovine dataset with daily feed intake (DFI) records.

RESULTS

Using the proposed method, we estimated the probability of the occurrence of an unrecorded environmental challenge, which proved to be informative and useful for inclusion as a covariate in a reaction norm animal model. We estimated the breeding values for sensitivity of the genetic potential for DFI of animals to environmental challenges. The level and slope of the reaction norm were negatively correlated (- 0.46 ± 0.21).

CONCLUSIONS

Our method is promising and appears to be viable to identify unrecorded events of environmental challenges, which is useful when selecting resilient animals and only productive data are available. It can be generalized to a wide variety of phenotypic records from different species and used with large datasets. The negative correlation between level and slope indicates that a hypothetical selection for increased DFI may not be optimal depending on the presence or absence of stress. We observed a reranking of individuals along the environmental gradient and low genetic correlations between extreme environmental conditions. These results confirm the existence of a G [Formula: see text] E interaction and show that the best animals in one environmental condition are not the best in another one.

Lupin (Lupinus spp.) seeds exert anthelmintic activity associated with their alkaloid content

The growing range of drug resistant parasitic nematode populations threatens the sustainability of ruminant farming worldwide. In this context, nutraceuticals, animal feed that provides necessary dietary requirements while ensuring parasite control, could contribute to increase farming sustainability in developed and low resource settings. In this study, we evaluated the anthelmintic potential of lupin seed extracts against the major ruminant trichostrongylids, Haemonchus contortus and Teladorsagia circumcincta. In vitro observations showed that seed extracts from commercially available lupin varieties could significantly but moderately inhibit larval migration. This anthelmintic effect was mediated by the seed alkaloid content and was potent against both fully susceptible and multidrug resistant H. contortus isolates as well as a susceptible T. circumcincta isolate. Analytical chemistry revealed a set of four lupanine and sparteine-derivatives with anthelmintic activity, and electrophysiology assays on recombinant nematode acetylcholine receptors suggested an antagonistic mode of action for lupin alkaloids. An in vivo trial in H. contortus infected lupin-fed ewes and goats failed to demonstrate any direct anthelmintic effect of crude lupin seeds but infected lupin-fed goats suffered significantly less parasite-mediated blood losses. Altogether, our findings suggest that the anthelmintic potential of lupin remains limited. However, the potent alkaloids identified could lead to the development of novel drugs or may be used in combination with current anthelmintics to improve their efficacy.

Detailed genetic analysis of feeding behaviour in Romane lambs and links with residual feed intake

Breeding strategies based on feed efficiency are now implemented in most animal species using residual feed intake (RFI) criteria. Although relevant, the correlated responses of feeding behaviour traits resulting from such selection on RFI are poorly documented. We report the estimated feeding behaviour at three time levels (visit, meal and day) and genetic parameters between the feeding behaviour traits and their links with RFI and its components. Feed intake, feeding duration at three time levels (per visit, meal and day), feeding rate, number of visits and time-between-visits were estimated for 951 Romane lambs fed via automatic concentrate feeders. Heritability estimates of feeding behaviour traits ranged from 0.19 to 0.54 with higher estimates for the day level than the visit level. Daily feed intake was not genetically linked to feed intake at the visit level, whereas feeding duration between visit and day levels was moderately correlated (R_g  = +0.41 ± 0.12). RFI was not significantly correlated with feeding rate, but was positively linked to feed intake and feeding duration at the day level (+0.73 ± 0.09 and +0.41 ± 0.13, respectively) and negatively at the visit level (-0.33 ± 0.14 and -0.22 ± 0.17, respectively). Selecting animals with lower RFI values might modify their feeding behaviour by increasing feed intake and feeding duration at the visit level but decrease the number of visits per day (+0.51 ± 0.14).

Typing of strains from a single-source outbreak of Pseudomonas pickettii

Plasmid profiles, genome restriction fragment polymorphisms, carbohydrate oxidation-fermentation reactions, methylumbelliferyl substrate hydrolysis patterns, antimicrobial susceptibilities, and results obtained with the Biolog GN biochemical substrate kit were used to type 19 common-source, but mixed-biotype, outbreak strains and one epidemiologically distinct strain of Pseudomonas pickettii. Biotyping with conventional and methylumbelliferyl substrates failed to distinguish between strains. Plasmid profile testing was found to be inconsistent and not reproducible. The Biolog GN kit allowed greater strain differentiation than restriction fragment polymorphism did (12 biotypes versus 5 biotypes); antimicrobial susceptibility testing yielded 4 biotypes, and oxidation-fermentation tests gave 3 biotypes. Oxidation-fermentation results were consistent with restriction fragment polymorphs in all but 1 of the 20 strains tested. For ease of typing, comprehensive typeability, and reproducibility, oxidation-fermentation tests should be performed initially and followed if necessary by restriction fragment polymorph analysis for the elucidation of P. pickettii infection outbreaks.

Histocompatible miniature pig (d/d haplotype): generation of hybridomas secreting A or M monoclonal antibody

Two Hypoxanthine/Aminopterin/Thymidine-sensitive cell sublines (L142 and L231) have been derived from independent lymphoblastoid cell lines of B lineage. After propagation for more than 100 population doublings (1 year) in culture, these cells still retained a doubling time between 19 to 20 hours, near diploïdy and relatively low (L142) and high (L231) secretion rate of M immunoglobulins. Near diploid hybrid cells were easily generated with leukocytes from the spleen, the gut lamina propria or the mesenteric lymph nodes of pigs immunized against the transmissible gastroenteritis virus. Both the tumor sublines and the B cells were derived from histocompatible miniature pigs (d/d haplotype). Demonstration of fusion between the tumor sublines and B-cells was supported by the selection of hybridomas making the antigen-specific heavy (alpha isotype) and light chains from the B cell parent as well as the heavy and light chains of the lymphoblastoid parent. Moreover, some hybridomas were found to secrete only class A (dimeric) or class M immunoglobulins (0.2-10 micrograms/ml). Forty hybridomas secreted antibodies reactive in a virus-enzyme-linked cell immunoassay against cell-bound antigens and two were found to produce an antibody active only against the infected cell monolayer. Construction of intraspecies hybridoma can be used to perpetuate lymphocyte subsets useful for the study of the porcine immune system.