Disease & heat resistance

Metabolic signatures after genetic selection of litter size homogeneity and resilience in a maternal rabbit line

---

Abstract

Selection for livestock resilience is an attractive breeding goal to maintain productivity and welfare during environmental perturbations. However, direct individual measurement of resilience is challenging. In a divergent selection experiment in rabbits (University Miguel Hernández, Orihuela, Spain), resilience was measured indirectly as residual variance of litter size (Ve) after correcting for year-season and lactation status, establishing lines with homogeneous and heterogeneous litter sizes. Companion studies identified that the homogeneous line (high-resilient) showed improved resilience indicators under basal conditions, as well as in response to internal (post-partum stress) and external stimuli (vaccination and heat stress). In generation 8, the homogeneous (high-resilient) line exhibited superior on-farm performance, alongside reduced cortisol and C-reactive protein levels pre- and post-vaccination compared to the heterogeneous (low-resilient) line. Despite these findings, the biological mechanisms underlying these divergent physiological responses remain poorly understood. This study compared the comprehensive plasma metabolomic profiles of these two lines across two generations (13 and 17) to identify plasma metabolites responding to selection for resilience. Blood samples were collected 24 hours after first kindling from 80 randomly selected does balanced by generation and line (20 per line, 40 per generation). An untargeted plasma metabolomics study identified 1,397 metabolites in generation 13 and 1,170 in generation 17. Metabolic variables with consistent changes were identified in generation 17 and subsequently evaluated in generation 13. For that, metabolite relative abundances in generation 17 were subjected to a preliminary screening that retained 180 metabolites with relative abundance differential mean (n = 69) or variance (n = 131). Differential means were estimated with Bayesian linear models, while differential variance was assessed using a univariate Fisher's F-test. Lines were classified through logistic regression, partial least squares discriminant analysis (PLS-DA), and random forest following different data transformations: chi-squared normalization, pairwise log-ratio (LR), centered log-ratio (CLR), and additive log-ratio (ALR). After hyperparameter tuning, the model performance was evaluated across 100 iterations of repeated 70:30 train-test splits, incorporating algorithm-specific feature selection. The highest classification accuracy was achieved by PLS-DA across all log transformations. When an additional PLS regression analysis to predict Ve was applied, the same algorithm produced the highest correlation between observed and predicted values with a median value of 0.57 [interquartile range: 0.44-0.69] using 390 LRs. Among the overlapping metabolic variables between PLS-DA and PLS regression (167 LRs, 7 CLRs and 5 ALRs), only 17 LRs, 1 CLR, and 1 ALR were detected in generation 13. Two LRs, creatine/C10H8N8O2 and 3-methylglutarylcarnitine/N-(3-acetamidopropyl)pyrrolidin-2-one, showed consistent directional changes with increasing magnitude of differences between the lines across generations. Individually, the metabolites composing these variables participate in energy metabolism, and immune regulation, components enabling the animals to respond quickly to the environmental perturbations.

Keywords: 2026

How to Cite:

Chandrababu-Shailaja, C., Argente, M., García, M., Hernández, P. & Martínez-Álvaro, M., (2026) “Metabolic signatures after genetic selection of litter size homogeneity and resilience in a maternal rabbit line”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2285472. doi: https://doi.org/10.31274/wcgalp.23719

Rights: 1