Selecting for changes in average “parity curve” pattern of litter size in Large White pigs

Abstract

This study aimed to analyse genetic background of variation in reproductive performance between parities of a sow and to investigate selection strategies to change the “parity curve”. Total number born (TNB) recorded in Large White sows was provided by Topigs Norsvin. Analysis with basic (BM) and random regression (RRM) models was done in ASReml 4.1. The BM included only a fixed “parity curve”, while RRM included 3rd order polynomials for additive genetic and permanent sow effects. Parameters from RRM were used in simulations in SelAction 2.1. Based on Akaike information criterion, RRM was a better model for TNB data. Genetic variance and heritability estimates of TNB from BM and RRM were increasing with parity from parity 2. Genetically, parity 1 is the most different from parities 7 to 10, whereas most similar to parities 2 and 3. This indicates presence of genetic variation to change the “parity curve”. Based on simulations, the selection to increase litter size in parity 1 only increases TNB in all parities, but does not change the observed shape of “parity curve”, whereas selection for increased TNB in parity 1 and reduced TNB in parity 5 decreases differences between parities, but also reduces overall TNB in all parities. Changing the “parity curve” will be difficult as the genetic and phenotypic relationships between the parities are hard to overcome even when selecting for one parity.

Genetic parameter estimation for number born alive at different parities in Landrace and Large White pigs

“…The random regression model estimated them to be higher than 0.90, among which those between the first and second parities were the lowest in both breeds. A similar phenomenon has been observed in total number born (Sell‐Kubiak et al, ). Other studies have also reported that the genetic correlation of NBA between the first and second parities was slightly lower than those between other successive parities (e.g., David et al, ; Hanenberg et al, ; Lukovic et al, ).…”

“…Estimated heritability using the random regression model was lower than that using repeatability animal model at only the second parity. On the other hand, Sell‐Kubiak et al () analyzed the records of total number born from the first to the tenth parities in Large White sows provided by Topigs Norsvin, using a repeatability animal model, a multiple‐trait model regarding all parities as different traits each other, and a random‐regression model. These models assumed the heterogenous residual variances, and the magnitude relation of the heritability estimates among the models were different depending on parity.…”

“…Different models are available to estimate the genetic correlation between parities. Multiple‐trait model and random regression model approaches have been used in studies of the genetic correlations between different parities in litter size of pigs (e.g., Alfonso, Noguera, Babot, & Estany, ; David, Garreau, Balmisse, Bilon, & Canario, ; Hanenberg, Knol, & Merks, ; Lukovic, Uremovic, Konjacic, Uremovic, & Vincek, ; Sell‐Kubiak, Knol, & Mulder, ). Here, we estimated genetic parameters for NBA using three models: a single‐trait animal model, a two‐trait animal model, and a single‐trait random regression model.…”

“…correlations between different parities in litter size of pigs (e.g., Alfonso, Noguera, Babot, & Estany, 1997;David, Garreau, Balmisse, Bilon, & Canario, 2017;Hanenberg, Knol, & Merks, 2001;Lukovic, Uremovic, Konjacic, Uremovic, & Vincek, 2007;Sell-Kubiak, Knol, & Mulder, 2019). Here, we estimated genetic parameters for NBA using three models: a single-trait animal model, a two-trait animal model, and a single-trait random regression model.…”

Estimation of litter size variability phenotypes in Large White sows

“…Over the last years, the methodology introduced by Rönnegård et al (2010) using double hierarchical generalized linear model (DHGLM) and individual residual variances was applied to estimate phenotypes for variability (Felleki, Lee, Lee, Gilmour, & Rönnegård, 2012; Mulder, Rönnegård, et al, 2013). In this approach, it is crucial to obtain the most accurate estimates of residuals, which could differ significantly depending on the model used (Sell‐Kubiak, Knol, & Mulder, 2019). Nevertheless, methods based on DHGLM require high computing power and due to their complexity are in general difficult to implement in breeding programmes.…”