Edited by: Denis Milan, Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement (INRAE), France
Reviewed by: Mario Calus, Wageningen University and Research, Netherlands; Marie Lillehammer, Norwegian Institute of Food, Fisheries and Aquaculture Research (Nofima), Norway
This article was submitted to Livestock Genomics, a section of the journal Frontiers in Genetics
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Data for loin and backfat depth, as well as carcass growth of 126,051 three-way crossbred pigs raised between 2015 and 2019, were combined with climate records of air temperature, relative humidity, and temperature–humidity index. Environmental covariates with the largest impact on the studied traits were incorporated in a random regression model that also included genomic information. Genetic control of tolerance to heat stress and the presence of genotype by environment interaction were detected. Its magnitude was more substantial for loin depth and carcass growth, but all the traits studied showed a different impact of heat stress and different magnitude of genotype by environment interaction. For backfat depth, heritability was larger under comfortable conditions (no heat stress), as compared to heat stress conditions. Genetic correlations between extreme values of environmental conditions were lower (∼0.5 to negative) for growth and loin depth. Based on the solutions obtained from the model, sires were ranked on their breeding value for general performance and tolerance to heat stress. Antagonism between overall performance and tolerance to heat stress was moderate. Still, the models tested can provide valuable information to identify genetic material that is resilient and can perform equally when environmental conditions change. Overall, the results obtained from this study suggest the existence of genotype by environment interaction for carcass traits, as a possible genetic contributor to heat tolerance in swine.
The increased relevance of heat stress to livestock industries is due to concerns in animal welfare as well as its economic impact. Estimates of annual financial loss to the pork industry that are attributable to heat stress range between $299 and $316 million. Of these losses, those for growing–finishing pigs are estimated to be $202 million (
The biological mechanism by which heat stress impacts production and reproduction in pigs has been widely documented by different authors (
Carcass characteristics are crucial for the profitability of pork producers because the carcass price is often adjusted based on these components, particularly backfat (with negative emphasis) and loin depth (with positive emphasis). The knowledge of the extent of genotype by environment interactions for these traits and the ability to identify pigs that are less susceptible to heat stress would greatly increase the competitiveness and efficiency of the pork industry in the face of climatic changes.
Currently, selection tools for improving heat tolerance or adaptability are not implemented in swine genetic evaluations. Many studies explored the genetic variation in heat tolerance in several livestock species using reaction norm models (
Most of the studies regarding the G × E interaction performed in swine populations have been focused on live body weight and growth performance (
Studies on the carcass characteristics of the pig are not available, despite of their economic importance. Therefore, the objective of this study was to estimate genetic parameters for heat tolerance leveraging on a potential genotype × environment interaction for carcass quality traits of commercial crossbred pigs.
Animal use approval was not needed for this study because the data were from an existent database and were provided by The Maschhoffs, LLC (Carlyle, IL, United States), and Acuity Ag Solutions, LLC (Carlyle, IL, United States). Loin depth (cLD), backfat (cBF), and hot carcass weight were measured on terminal three-way crossbred pigs. Birth weight and date as well as harvest date were recorded, and harvest age was calculated. Carcass average daily gain (
All individuals were crossbred gilts and barrows, progeny of Duroc sires, and different purebred or crossbred dam lines. Animals were born in three sow farms and raised between September 2015 and November 2019 on two commercial grower–finisher flows (
Descriptive statistics for the traits used in the study (
cBF, mm | 18.7 | 4.11 |
cLD, mm | 67.3 | 7.03 |
cADG, kg/d | 0.54 | 0.12 |
Piglets were moved to different nursing/finishing facilities at weaning (18.7 ± 4.11 days). Individuals were considered ready for harvest at a target weight of approximately 136 kg. Harvest occurred on average at 178 days ± 10.6 days of age.
During the grow–finish period, a standard pelleted gender-specific dietary program was used. Individuals were monitored daily and received standard vaccination and emergency medication. For details on diet composition, vaccination, and medication during nursery, growth, and finish periods, see
For the data analyses, all animals were allocated into 57 contemporary groups (
Phenotyped individuals were progeny of 407 sires, and 279 of them were genotyped using the Illumina porcineSNP60 BeadChip (Illumina, Inc., San Diego, CA, United States). Crossbred individuals were born in a total of 20,525 litters; each sire was parent to 1–546 litters.
Meteorological data were extracted from the National Climatic Data Center Quality Controlled Local Climatological Data
The hourly temperature–humidity index was calculated using the formula proposed by
where T is the observed temperature in degree Celsius (°C) and RH is the observed relative humidity on a 0–100 scale. Average daily values for each climatic variable were then calculated.
In order to better investigate the patterns of thermal stress, three lifetime periods were defined. Using the birth date of the animal, thermal load was defined for three time intervals (60–92 days, 92–122 days, and 122–152 days) of age for each individual in the study. The average daily temperature, relative humidity, and temperature–humidity index for the daily values of each time interval were calculated. The nine resulting environmental covariates (three time periods by three climate variables) were then merged to each individual’s phenotypic record. For each of the 27 trait-by-environmental-covariate combination, linear models were used in order to evaluate the response of a considered trait to a specific environmental covariate. The first linear model was chosen, instead of Pearson correlation, for the possibility to adjust for other systematic effects. The model specifications for the first model were.
where y
where
Coefficients of determination and Bayesian information criterion values for the models assessing the impact of different heat load functions on the traits of interest.
THI | 60–92, d | 0.041 | −241192.87 | −241161.44 |
Temp | 60–92, d | 0.040 | −241186.62 | −241157.38 |
RH | 60–92, d | 0.041 | −241672.4 | −241649.27 |
THI | 92–122, d | 0.055 | −241775.51 | −241746 |
Temp | 92–122, d | 0.055 | −242126.2 | −242105.38 |
RH | 92–122, d | 0.048 | −241340.31 | −241352.62 |
THI | 122–152, d | 0.041 | −242626.65 | −242607.3 |
Temp | 122–152, d | −242805.43 | −242773.22 | |
RH | 122–152, d | 0.041 | − |
− |
THI | 60–92, d | 0.135 | 640065.622 | N.C. |
Temp | 60–92, d | 0.134 | 640037.831 | N.C. |
RH | 60–92, d | 0.139 | 639893.692 | N.C. |
THI | 92–122, d | 0.125 | 640068.172 | N.C. |
Temp | 92–122, d | 0.125 | 640140.87 | N.C. |
RH | 92–122, d | N.C. | ||
THI | 122–152, d | 0.135 | 639935.033 | N.C. |
Temp | 122–152, d | 0.136 | N.C. | N.C. |
RH | 122–152, d | 0.125 | N.C. | N.C. |
THI | 60–92, d | 0.014 | N.C. | N.C. |
Temp | 60–92, d | 0.015 | N.C. | N.C. |
RH | 60–92, d | 0.011 | N.C. | N.C. |
THI | 92–122, d | 0.022 | 814767.925 | N.C. |
Temp | 92–122, d | 0.022 | 814781.392 | N.C. |
RH | 92–122, d | 0.017 | 814757.367 | 814776.72 |
THI | 122–152, d | 0.014 | 814725.541 | N.C. |
Temp | 122–152, d | N.C. | ||
RH | 122–152, d | 0.011 | 814797.207 | N.C. |
The vectors for the sire effects were assumed as
where G is a 2 × 2 (co)variance matrix for the intercept and slope effects, respectively:
where
A total of 300,000 Gibbs samples were generated, while discarding the first 50,000 as burn-in and thinning every 50 samples. Posterior means and standard deviations of the remaining 5,000 samples were used as estimates and standard error for the (co)variance components. The goodness of fit was measured by the coefficient of determination (R2) for the model in 1 and by the Bayesian information criterion for the model in 2 (both for the first-order and second-order polynomials). Results are reported in
The additive genetic (co)variance structures of individual sire across the range of the environmental covariate (
where G is the estimated (co)variance matrix between the intercept and slope terms and Φ is a matrix containing a column of “1” (intercept) and the environmental covariate. Heritability at each single value
where Γmm is the
The summary of monthly mean temperatures and humidity is illustrated in
As reported in
For cBF, the best predictor (
Random regression models were used to evaluate the effect of heat stress and potential genetic control of heat tolerance for carcass quality traits of crossbred pigs. Heritability estimates from the current study indicates a possible genetic improvement for heat tolerance by selecting for the direct genetic component of carcass quality traits under heat-stressed conditions. Results illustrated in
Heritability estimates (95% empirical confidence intervals) for the three carcass quality traits of animals over the range of the respective climatic variable. The black dots report the estimates from the first-order Legendre polynomial random regression model. The blue dots report the estimates from the second-order Legendre polynomial random regression model (available for Carcass Average Daily Gain only).
Genetic correlations are summarized as a heat map for each analyzed trait in
While fit measures for cADG suggested the use of the second-order RRM, we noted poor convergence for the sire solutions for the quadratic term even if fixing the variance components. Therefore, we decide to use the first-order polynomial model for this trait.
Reaction norms for the twenty sires showing the higher and lower estimated breeding values (EBV) for the intercept term of the random regression model.
Reaction norms for the twenty sires showing the higher and lower estimated breeding values (EBV) for the slope term of the random regression model.
The phenotypic data used in this study came from a commercial system where three-way crossbred pigs are generated using single-sire semen, thus allowing the estimation of genetic parameters in a commercial environment. The growing–finishing units run on a fixed-weight system, where individuals are harvested when their body weight reaches the desirable value. However, harvest is not performed on an individual but on a batch-based basis, which allows some intra-batch variability in body weight and carcass composition. Descriptive statistics reported in
The heritability estimates from the current study indicate the potential to perform selection by selecting for the direct genetic component of carcass quality traits both under comfortable and heat-stressed conditions (
Although RRMs have been previously used to model animal weight (live weight) in dairy and beef cattle (
The impact of heat stress found in this study was considerable for cADG and cBF. Lower carcass fatness at slaughter connected to the decline in feed intake is generally reported in heat-stressed pigs (
Similar to feed intake, cADG shows a decreasing response during the thermal load and is affected by the animal’s body weight with heavier pigs more susceptible to heat stress than lighter ones (
The decrease in feed intake does not explain the weak loss in cLD when heat stress occurs. For this trait, the impact was approximately null and some families actually showed an increase in cLD when conditions involved heat stress. The increase in cLD was of 3 mm of cLD passing from −7 to 26°C of Temp, in contrast with the common observation on heat-stressed finishing pigs (
Differences in performance reported in this study between no thermal stress and maximum thermal stress scenario would have a large economic impact for producers in swine production, influencing carcass quality and consequently the profit function within the pork industry. The methodology proposed in this study using weather information to identify heat-tolerant animals could be a useful tool to improve the production system and implement the selection programs. Ideally, the use of this approach represents a breeding strategy to improve heat tolerance in relation to the farm resources.
The use of weather station climatic data was again proven valuable for a first estimation of reaction norm and heat tolerance of the different families. Outdoor records are a poor prediction of indoor condition, which does not allow the exact definition of comfortable and uncomfortable conditions. Further studies will need to consider indoor-recorded environmental data.
Farming systems could benefit more from including heat tolerance in the breeding programs of individuals that are resistant to extreme conditions. On the other hand, we found a partial antagonism between heat tolerance and productivity. Comparing the intHi and intLow sires for the three traits, we do not observe a strong difference in the slope of the reaction norms. Comparing the sloHi and sloLow sires, their performance under comfortable conditions appears to be different. For cBF, the most resilient individuals have a lower performance under comfortable conditions and approximately the same performance under heat stress. This could be related to their efficiency in converting feed into body weight instead of body heat. However, this hypothesis will need further studies to be proved.
If selection for increased resilience is performed, there will probably be a loss in performance under optimal conditions. This suggests that overall performance and tolerance to heat stress should be combined in an economic selection index, with different weights depending on the likelihood of certain conditions to occur in a particular system.
A random regression model including genomic information was used to evaluate the effect of heat stress on carcass quality traits of crossbred pigs. Data used for this study came from commercial operations, making the presented results representative of the swine industry.
Performance under heat stress seems to be less or equally heritable than under comfortable conditions, but genetic variation still exists even under heat stress, indicating that the identification and selection of the most resistant animals is possible in order to implement the selection programs. A graphical analysis of the reaction norms shows that genetic material with improved heat tolerance is easily identifiable. The use of outdoor-recorded environmental measures can be valuable for early studies on the subject, but indoor-recorded measures will be needed for further studies.
The three traits studied showed a different impact of heat stress and different magnitude of genotype by environment interaction. Because of this, it will be a task of the breeder to determine the stronger economic value of heat tolerance (vs overall performance) for each trait.
Further research is needed for the heat tolerance of swine to overcome the complexity of selection of heat-tolerant animals, due to a partial antagonism between heat tolerance and overall performance.
The datasets presented in this article are not readily available because the raw datasets are property of the swine breeding companies and this information is commercially sensitive. Requests to access the datasets should be directed to Clint Schwab.
Ethical review and approval was not required for the animal study because the data used in this study came from animal raised under commercial facilities for pork production. Written informed consent for participation was not obtained from the owners because the animals were property of some of the coauthors.
MU, NM, and FT conceived and designed this study. MU and MB carried out the analyses. MU, FT, NM, CM, and MB interpreted and discussed the results. MU wrote the first draft of the manuscript. ClS, JF, and CaS supervised the data collection and provided inputs for the analyses of the data. All the authors reviewed and approved the final manuscript.
The study used data that were provided as in kind by The Maschhoffs LLC. ClS, CaS, and JF were employed by The Maschhoffs LLC or Acuity Ag Solutions LLC at the time of submission. The results are commercially of interest to the above-mentioned companies but this interest did not influence the results presented in this manuscript in any matter. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We acknowledge the phenotypic, pedigree, and genomic datasets provided by the Maschhoffs LLC and Acuity Ag Solutions LLC companies.