Factors affecting pre-weaning mortality of piglets

Authors: RAHMAN, TOWFIQUR - University Of Nebraska; BROWN-BRANDL, TAMI - University Of Nebraska; Rohrer, Gary; SHI, YEYIN - University Of Nebraska; SHARMA, S. RAJ - University Of Nebraska; MANTHENA, VAMSI - University Of Nebraska

Submitted to: American Society of Agricultural and Biological Engineers
Publication Type: Abstract Only
Publication Acceptance Date: 1/11/2022
Publication Date: 7/20/2022
Citation: Rahman, M.T., Brown-Brandl, T.M., Rohrer, G.A., Shi, Y., Sharma, S.R., Manthena, V. 2022. Factors affecting pre-weaning mortality of piglets [abstract]. In: Proceedings of the American Society of Agricultural and Biological Engineers. July 17-20, 2022, Houston, Texas. Paper 220581. pg. 23.

Interpretive Summary:

Technical Abstract: High pre-weaning mortality rates for piglets are a significant concern for the worldwide pork industries. This issue is not only an economic loss for the producers but also associated with well-being concerns for the industry. There are multidimensional causes for pre-weaning mortality (PWM) like mummies (fetal death) and stillborn (dead at birth), health condition of sow and piglets, environmental factors (season, controlled environment), congenital abnormalities, disease, and crushing by sow. The present study focuses on determining the impact of sow, piglet, and environmental factors on the occurrence of PWM and specifically overlays. Data were from 1,982 litters located at the U.S. Meat Animal Research Center over the years 2016-2021. The sows were housed in a new farrowing building with 3 separate rooms, each with 20 – 1.8 by 2.7m farrowing crates. PWM was calculated as PWM = (number of piglets born - number of piglets at weaning)/number of piglets born. The number of piglets born included stillborn, but not mummy piglets. Sow factors that were considered included parity (1-4), gestation length (109-119 days), litter line (Yorkshire, Landrace sire lines), and health diagnosis (healthy and unhealthy). Piglet factors that were considered included the number of stillborn, born alive, and mean birth weight. Environmental factors that were considered included farrowing crate placement within the room (10 different locations), and season (4 seasons). Litters were classified according to the sow parity and mean birth weight. ANOVA was performed to determine impacts of season, parity, and the interaction of these two factors. In addition, three different models (Beta regression, Binomial, Random Forest algorithm) were used to determine the association with PWM with all listed contributing factors. The ANOVA analysis found PWM had significant effects of season and parity (P<0.01) but not the interaction (P>0.3). Overlays were significantly affected by season (P<0.01) and the interaction of parity and season (P<0.1). On average, PWM was found to be 19.39% and 7.5% was caused by overlays from the dataset. ANOVA model (Tukey’s HSD) shows season impacted both the PWM, and overlays in the fall season having significantly higher mortality rates compared with the other seasons (P<0.1). For parity, PWM was higher in the fourth parity compared to parties 1 and 2 (P<0.01). Among the three machine learning models, Random Forest (RF) performed the best and showed the highest linearity and correlation between PWM and the predicted PWM. The mean squared error (MSE) for RF was 12.92% which is lower than the other two models (Binomial: 4569.41 & Beta regression: 41.67). The coefficient of determination was 0.92 for RF. A quantitative index of feature importance from RF shows that the mean birth weight of piglets is the most significant factor for PWM (Mean Decrease Accuracy (MDA) = 12.14). The PWM increases as the mean birth weight decreases. Then, sow health diagnosis (MDA= 5.89), gestation length (MDA= 5.12), season (MDA= 4.67), and parity (MDA= 4.09) also have significant effects on mortality. Locations within the room and litter lines have no significant effects on mortality. Considering the challenges to reduce pre-weaning mortality and limited studies with the multiple major contributing factors, this study is valuable for further investigations on large production datasets. USDA is an equal opportunity employer.