Journal of Animal Science and Technology
Korean Society of Animal Sciences and Technology
RESEARCH ARTICLE

Connectedness rating among commercial pig breeding herds in Korea

Wonseok Lee1,#https://orcid.org/0000-0002-5103-3334, JongHyun Jung2,#https://orcid.org/0000-0003-3667-7710, Sang-Hyon Oh3,*https://orcid.org/0000-0002-9696-9638
1eGnome, Inc., Seoul 05836, Korea
2Jung P&C Institute, Yongin 16950, Korea
3Division of Animal Science, Gyeongsang National University, Jinju 52725, Korea
*Corresponding author: Sang-Hyon Oh, Division of Animal Science, Gyeongsang National University, Jinju 52725, Korea. Tel: +82-55-772-3285, E-mail: shoh@gnu.ac.kr

These authors contributed equally to this work.

© Copyright 2024 Korean Society of Animal Science and Technology. This is an Open-Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Received: Dec 08, 2023; Revised: Jan 11, 2024; Accepted: Jan 29, 2024

Published Online: Mar 31, 2024

Abstract

This study aims to estimate the connectedness rating (CR) of Korean swine breeding herds. Using 104,380 performance and 83,200 reproduction records from three swine breeds (Yorkshire, Landrace and Duroc), the CR was estimated for two traits: average daily gain (ADG) and number born alive (NBA) in eight breeding herds in the Republic of Korea (hereafter, Korea). The average CR for ADG in the Yorkshire breed ranges from 1.32% to 28.5% depending on the farm. The average CR for NBA in the Yorkshire herd ranges from 0% to 12.79%. A total of 60% of Yorkshire and Duroc herds satisfied the preconditions suggested for genetic evaluation among the herds. The precondition for the genetic evaluation of CR for ADG, as a productive trait, was higher than 3% and that of NBA, as a reproductive trait, was higher than 1.5%. The ADG in the Yorkshire herds showed the highest average CR. However, the average CR of ADG in the Landrace herds was lower than the criterion of the precondition. The prediction error variance of the difference (PEVD) was employed to assess the validation of the CR, as PEVDs exhibit fluctuations that are coupled with the CR across the herds. A certain degree of connectedness is essential to estimate breeding value comparisons between pig herds. This study suggests that it is possible to evaluate the genetic performance together for ADG and NBA in the Yorkshire herds since the preconditions were satisfied for these four herds. It is also possible to perform a joint genetic analysis of the ADG records of all Duroc herds since the preconditions were also satisfied. This study provides new insight into understanding the genetic connectedness of Korean pig breeding herds. CR could be utilized to accelerate the genetic progress of Korean pig breeding herds.

Keywords: Connectedness rating; Yorkshire; Landrace; Duroc; Average daily gain; Number born alive

INTRODUCTION

The accuracy of estimated breeding value (EBV) for economic traits is crucial in animal breeding programs. The accuracy of the estimation relies on connectedness to perform genetic analysis between pig breeding herds. Connectedness rating (CR) refers to the genetic similarity or genetic relationships among different herds or populations of animals [1]. In the context of animal breeding programs, it is crucial to assess the CR between herds or populations because it significantly impacts the accuracy of EBV for economic traits, which relies on having enough genetic information from the different herds or populations to perform a thorough genetic analysis. When herds are genetically connected, it means that there are shared ancestors or common genetic backgrounds between them. This genetic connectedness enables higher accuracy breeding value estimations because more genetic information is available to make predictions about an animal’s performance and breeding potential. On the other hand, if herds are genetically isolated or not well connected, the accuracy of EBV may be lower due to the limited genetic information available for analysis. In such cases, it becomes challenging to predict the performance of animals accurately, leading to less effective selection and breeding strategies. Therefore, understanding and evaluating these genetic links among herds or populations is essential in animal breeding programs to ensure accurate and successful genetic improvement of economically important traits in livestock affecting the prediction error variance of difference (PEVD) of EBV [2].

Many methods have been proposed to estimate CR [1,3-5]. Among them, Mathur et al. [6] suggested the CR, which has gained popularity as a reliable measure for evaluating connectedness. Subsequent studies have validated its consistency in producing accurate results in various connectedness analyses [2,7,8]. In Korea, genetic progress on swine has been achieved mainly by importing breeding pigs from other countries. However, as consolidation among pig breeding companies and farms has progressed, the number of pig genetic resources imported each year has decreased. In addition, multiple breeding farms plan to work together on pig genetic analysis to maximize genetic progress and mitigate the need to import breeding pigs into Korea. However, if there is no genetic link between herds, the analysis of EBV between different farms would not be reliable or accurate. It has been reported that the accuracy of the genetic evaluation increases when the CR between the herds is high [2,7-9].

In Korea, efforts are being made to utilize pig genetic resources at the national level. The Pig Improvement Network (PIN) is a project under the Ministry for Food, Agriculture, Forestry and Fisheries (MIFAFF) in Korea. It aims to improve and develop a unique Korean pig breed well-suited to the conditions in Korea by selecting, sharing, and evaluating excellent breeding stocks from various farms to secure genetic connections. The project is managed by the National Institute of Animal Science under the Rural Development Administration (RDA), which oversees project management and national-level genetic capacity evaluations. The Korean Animal Improvement Association (KAIA) is responsible for project implementation and performance testing. Currently, 19 pig farms are participating in this network, and three pig artificial insemination (AI) centers are in operation.

The PIN is focused on enhancing the competitiveness of the domestic pig industry and developing genetically superior pig breeds that are resilient and productive in the local environment. The collaboration between government agencies, research institutes, and pig farms allows for the comprehensive genetic evaluation and selective breeding required to achieve the project’s goals. The core of this project lies in how efficiently and effectively genetic resources are shared among different farms. Therefore, this study aims to estimate the connectedness among swine herds in Korea using three different breeds (Yorkshire, Landrace and Duroc) for average daily gain (ADG) and number born alive (NBA).

MATERIALS AND METHODS

Data preparation

Performance and reproduction data were collected from fifteen Korean pig breeding herds (8 Yorkshire herds, 5 Landrace herds and 4 Duroc herds) born between 1997 and 2016. Two traits were considered to calculate the connectedness between pairs of herds: ADG and NBA. The numbers of records per breed and farm are presented in Tables 1 and 2.

Table 1. Number of records for ADG
Farm Yorkshire Landrace Duroc
A 20,460 327 759
B 8,620 205 580
C 9,710 3,812 3,492
D 1,296 - -
E 17,888 357 -
F 2,971 1,094 -
G 5,476 - 2,261
H 14,138 10,484 -
TOTAL 80,559 16,279 7,092

ADG, average daily gain.

Download Excel Table
Table 2. Number of records for NBA
Farm Yorkshire Landrace Duroc
A 5,127 327 759
B 2,773 205 580
C 9,710 3,812 3,492
D 1,296 - -
E 17,888 357 -
F 2,971 1,094 -
G 5,476 - 2,261
H 14,138 10,484 -
TOTAL 59,379 16,279 7,092

NBA, number born alive.

Download Excel Table
Statistical model for breeding value

Both ADG and NBA datasets were analyzed for each breed to estimate the breeding value using the following statistical model (1).

y = Xb + Za + Hd + e
(1)

where y is the phenotype observation for ADG and NBA, b is a vector of fixed effects (herd effects), a is the vector of random effects (additive animal genetic effects), d is the vector of common litter effects, and e is a vector for environmental residuals (eN(0,Iσe2)). X, Z and H were used as incidence matrices corresponding to vectors b, a and d related to the random additive genetic effects (aN(0,Aσa2,dN(0,Iσd2)).σa2,σd2 and σe2 represent the additive genetic variance, litter variance, and environmental residual variance, respectively.

Mixed model equation construction

The above statistical model was used to construct the mixed model equation (MME) resulting in equation (2).

[ X X X Z X H Z X H H + I α 2 H Z H X Z H H H + A 1 α 1 ] [ h ^ d ^ a ] = [ X y Z y H y ]
(2)

where, A is the numerator genetic relationship matrix for animals, a1 refers to σe2/σa2 and a2 refers to σe2/σd2. The ASReml package [10] was used for solving equation (2).

Estimation of connectedness rating

CR was defined by the following equation (3):

C R i j = cov ( h ^ i , h j ) var ( h ^ i ) var ( h j )
(3)

The covariance for the herd and variance of estimation of each herd effect i and j were obtained by solving equation (2).

Evaluation of connectedness rating and its effect on comparison of estimated breeding value

According to Mathur et al. [11], the accuracy of an individual EBV is estimated using the prediction error variance corresponding to the animals. The PEVD can be used to validate the accuracy of EBVs of two individuals. PEVD is formularized as:

PEVD = Var ( a ^ i a ^ j ) = Var ( a ^ i ) + Var ( a ^ j ) 2 cov ( a ^ i j )
(4)

PEVD can then be substituted as the variance of estimates of difference between herd effects (VHD) [11]:

V H D i j = Average [ PEV ( a ^ i k a ^ j k ) ] Var ( h ^ i h ^ j ) Var ( h ^ i ) + Var ( h ^ j ) 2 cov ( h ^ i j )
(5)

RESULTS

Connectedness ratings for average daily gain

The average CRs for ADG in each breed are listed in Table 3. For the ADG in Yorkshire, a total of 8 herds were analyzed. The average CR between two herds ranges from 1.32 (B herd) to 28.05 (E herd). The maximum CR value was 93.44 between herds E and G, and the lowest CR value was 4.4 between herds B and G. For the ADG trait in Landrace, a total of 5 herds were used for analysis. The highest average CR was 2.50 between herds A and F, and the lowest average CR was 0.55 between herds C and H. All four herds A, B, C, and G were used in Duroc. The highest average CR was 16.14 between herds C and G, and the lowest average CR was 5.03 between herds A and C.

Table 3. Connectedness rating (CR) for ADG among herds
Herds Yorkshire connectedness rating (%) Landrace connectedness rating (%) Duroc connectedness rating (%)
Mean Max Most connected herd Mean Max Most connected herd Mean Max Most connected herd
A 2.18 9.56 F 2.08 12.49 F 5.03 10.77 C
B 1.32 4.41 G 0.55 2.44 H 11.21 27.81 G
C 1.68 7.42 H 0.88 0.32 H 13.8 30.92 G
D 18.29 86.81 F - - - - - -
E 28.05 93.44 G - - - - - -
F 21.18 86.81 E 2.50 12.49 A - - -
G 12.92 93.44 E - - - 16.14 30.92 C
H 2.01 7.42 C 1.31 4.1 C - - -

ADG, average daily gain.

Download Excel Table
Connectedness ratings for number born alive

The average CRs for NBA in each breed are presented in Table 4. The same numbers of herds that were used for ADG analysis were used to analyze NBA. The minimum average CR ranged from ~0 (herd A) to 12.79 (herd E) in Yorkshire. The maximum CR was 89.38 between herds E and G, and the lowest CR value was identified between herds A and F. The highest average CR was 0.09 (herd H), and the lowest average CR was ~0 (herds A and F) in Landrace. The highest average CR for NBA ranged from 1.17 to 4.70 in Duroc. According to Mathur et al. [6], the recommended minimum average CRs for ADG and NBA are 3% and 1.5%, respectively. When these criteria for both the performance and reproductive traits are met, the EBV comparison between herds can be performed accurately. The average CR values for the Landrace herd are below this criterion, so the values for the Landrace herd were excluded in the following evaluation step.

Table 4. Connectedness rating (CR) for NBA among herds
Herds Yorkshire connectedness rating (%) Landrace connectedness rating (%) Duroc connectedness rating (%)
Mean Max Most connected herd Mean Max Most connected herd Mean Max Most connected herd
A ~0 ~0 F ~0 ~0 F 1.17 3.40 G
B 0.82 3.65 G 0.02 0.1 H 1.27 4.01 G
C 0.17 0.86 H 0.08 0.37 H 3.23 11.4 G
D 8.40 59.60 F - - - - - -
E 12.79 89.38 G - - - - - -
F 8.71 59.55 E ~0 ~0 A - - -
G 11.7 89.38 E - - - 4.70 11.4 C
H 0.20 0.86 C 0.09 0.37 C - - -

NBA, number born alive.

Download Excel Table
Evaluation of connectedness ratings using prediction error variance of the difference

If two herds are highly connected, the PEVD decreases. The accuracy of the EBV is therefore greater when a pair of herds is evaluated jointly. According to Kennedy and Trus [4], the VHD is highly correlated with the average PEVD of pairwise comparisons of EBVs. Therefore, VHD can be used as an evaluation of CR. The VHDs for the Yorkshire and Duroc herds were calculated, but the VHD for the Landrace herd was not calculated due to its CR result. The PEVDs for the ADG and NBA traits in Yorkshire and Duroc breeds are shown in Tables 5, 6, 7, and 8. The tables show that the PEVD decreases as the CR increases, suggesting that the PEDV can be used as a validation indicator for the accuracy of the CR.

Table 5. Prediction error variance of difference (PEVD) for ADG among Yorkshire herds
Herds CR (%) > 3 PEVD
A,D 5.86 0.0024
A,F 9.56 0.0028
B,E 4.04 0.0022
B,G 4.41 0.0069
C,E 3.86 0.0017
C,H 7.42 0.0067
D,E 47.97 0.0010
D,F 86.81 0.0004
D,G 5.53 0.0059
E,F 65.86 0.0006
E,G 93.44 0.0001
E,H 7.22 0.0027

ADG, average daily gain.

Download Excel Table
Table 6. Prediction error variance of difference (PEVD) for ADG among Duroc herds
Herds CR (%) > 3 PEVD
A,B 3.53 0.00413
A,C 10.77 0.00400
A,G 5.82 0.00402
B,C 13.50 0.00462
B,G 27.81 0.00342

ADG, average daily gain.

Download Excel Table
Table 7. Prediction error variance of difference (PEVD) for NBA among Yorkshire herds
Herds CR (%) > 1.5 PEVD
B,E 2.40 0.0456
B,G 3.65 0.0762
D,E 7.39 0.0456
D,F 59.55 0.0203
E,F 9.86 0.0337
E,G 89.38 0.0048

NBA, number born alive.

Download Excel Table
Table 8. Prediction error variance of difference (PEVD) for NBA among Korean Duroc herds
Herds CR (%) > 1.5 PEVD
A,G 3.40 0.0809
B,G 4.01 0.0847
C,G 11.40 0.0469

NBA, number born alive.

Download Excel Table

Based on the results, it can be observed that there are significant differences in CR between different herds for each of the evaluated breeds. For instance, in the case of the Korean Yorkshire breed, the CR between herds E and G is very high at 93.44, whereas the CR between herds B and G is relatively low at 4.4. These findings indicate that the level of genetic exchange varies among different herds, which can have implications for the reliability of genetic evaluations and breeding programs. Another important observation is the relationship between the size of the contemporary group (CG) and the accuracy of the EBVs. When the size of the CG is less than ten animals, the accuracy of the EBVs significantly decreases. Therefore, it is crucial for each CG to consist of at least 10 or more pigs to achieve higher accuracy, which ensures a sufficient sample size for more reliable genetic evaluations and breeding decisions.

Increasing connectedness is crucial for breeding programs, which can be achieved by using common sires from multiple herds or sharing genetically superior AI boars. By doing so, it becomes possible to conduct extensive genetic comparisons across herds, explore the potential for large-scale selection, and achieve greater genetic progress. The continuous supply of genetically superior pigs is directly related to the active participation of swine producers in breeding programs, who can contribute by providing their own genetically superior animals or by participating in the formation of a pool of superior AI boars. Such active involvement allows for the improvement of the national breeding program’s structure, leading to increased genetic variability and connectedness. As a result, the participation of swine producers plays a crucial role in the field of animal breeding and helps foster greater genetic advancements and overall progress.

DISCUSSION

A certain level of connectedness is needed for accurate EBV comparisons between herds. In this study, 104,380 performance data items and 83,200 reproduction data items from three different swine breeds across a total of fifteen herds were used to analyze connectedness using the CR method. The range of the CR for ADG in Korean swine herds was between 0.55 and 28.05. The ranges in the Yorkshire and Duroc breeds were deemed satisfactory with an average CR greater than 3%, while those of the Landrace breed were lower than 3%. Therefore, it is possible to compare the genetic evaluation results of the Yorkshire and Duroc herds for the ADG trait. The efforts are significantly required to improve the CR between pig herds for a national swine genetic evaluation. Increasing CR between herds primarily involves enhancing the use of common boars. Leveraging AI technology in the swine industry, boar exchange among farms, establishing pools of superior AI boars, and using AI boars with high CR can be effective strategies. Ensuring an adequate proportion of offspring from common sires is crucial, as it enables accurate comparisons of animals across different herds. Increasing the proportion of offspring from common sires can enhance genetic connectedness and reduce bias in inter-herd comparisons.

Competing interests

No potential conflict of interest relevant to this article was reported.

Funding sources

This work was equally supported by the Korea Institute of Planning and Evaluation for Technology (IPET) in Food, Agriculture and Forestry through the Golden Seed Project, funded by the Ministry of Agriculture, Food and Rural Affairs (No. 213010051SB710) and the Cooperative Research Program for Agriculture Science & Technology Development (No. PJ0128172017) by the Rural Development Administration in the Korea.

Acknowledgements

Not applicable.

Availability of data and material

Upon reasonable request, the datasets of this study can be available from the corresponding author.

Authors’ contributions

Conceptualization: Jung JH, Oh SH.

Data curation: Lee W.

Formal analysis: Lee W.

Methodology: Lee W, Jung JH.

Software: Lee W, Jung JH.

Validation: Lee W, Jung JH.

Investigation: Lee W, Jung JH.

Writing - original draft: Lee W, Jung JH.

Writing - review & editing: Lee W, Jung JH, Oh SH.

Ethics approval and consent to participate

This article does not require IRB/IACUC approval because there are no human and animal participants.

REFERENCES

1.

Soga N. The effect of connectedness on the bias and accuracy for prediction of breeding value in swine herds [Master’s thesis]. Ames, IA: Iowa State University. 2009

2.

Škorput D, Gorjanc G, Luković Z. Evaluation of connectedness between the management units of landrace breed of pigs in croatia. Acta Agric Slov. 2012; 3:182

3.

Foulley JL, Hanocq E, Boichard D. A criterion for measuring the degree of connectedness in linear models of genetic evaluation. Genet Sel Evol. 1992; 24:315-30

4.

Kennedy BW, Trus D. Considerations on genetic connectedness between management units under an animal model. J Anim Sci. 1993; 71:2341-52

5.

Laloë D, Phocas F, Ménissier F. Considerations on measures of precision and connectedness in mixed linear models of genetic evaluation. Genet Sel Evol. 1996; 28:359-78

6.

Mathur PK, Sullivan B, Chesnais J. Estimation of the degree of connectedness between herds or management groups in the Canadian swine population. OttawaCanadian Centre for Swine Improvement 1998

7.

Soga N, Spangler ML, Schwab CR, Berger PJ, Baas TJ. Comparison of connectedness measures and changes in connectedness of the U.S. Duroc population. In: Proceedings of the 9th World Congress on Genetics Applied to Livestock Production. 2010 Leipzig, Germany.

8.

Sun CY, Wang CK, Wang YC, Zhang Y, Zhang Q. Evaluation of connectedness between herds for three pig breeds in China. Animal. 2009; 3:482-5

9.

Mathur PK, Sullivan BP, Chesnais JP. Measuring connectedness: concept and application to a large industry breeding program. In: Proceedings of the 7th World Congress on Genetics Applied to Livestock Production. 2002 Montpellier, France.

10.

Gilmour AR, Gogel BJ, Cullis BR, Welham SJ, Thompson R. ASReml user guide release 4.1 structural specification. Hemel hempstead: VSN International. 2015

11.

Mathur PK. Importance of ‘connectedness’ between herds for effective across herd genetic evaluation. J South China Agric Univ. 2005; 26:61-8