Machine learning models reveal Saccharomyces yeasts are associated with poor piglet growth in early development

Authors: CHHETRI, NISAN - North Carolina State University; Summers, Katie; Campos, Philip; Postnikova, Olga; Rivera-Colon, Israel; HARLOW, KALYNN - Oak Ridge Institute For Science And Education (ORISE); Oliver, William; Wells, James; Davies, Cary

Submitted to: Journal of Animal Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/21/2025
Publication Date: 4/29/2025
Citation: Chhetri, N., Summers, K.L., Campos, P.M., Postnikova, O.A., Rivera-Colon, I., Harlow, K., Oliver, W.T., Wells, J., Davies, C.L. 2025. Machine learning models reveal Saccharomyces yeasts are associated with poor piglet growth in early development. Journal of Animal Science. 103. Article eskaf133. https://doi.org/10.1093/jas/skaf133.
DOI: https://doi.org/10.1093/jas/skaf133

Interpretive Summary: Piglets which are born small often grow slowly, reach slaughter at a late date, and are more susceptible to diseases, resulting in financial losses for industry. Under well designed nutrition and management practices, a subset of slow-growers may exhibit compensatory growth and reach weights comparable to those of faster-growing piglets. Despite a broad understanding of the factors which impact growth, knowledge regarding the mechanisms which underlie compensatory growth is incomplete. The microbiome is a collection of bacteria, fungi, and other microbes which reside in the gut and interact with the host to shape health. Emerging evidence suggests a role for the microbiome in growth, as gut bacteria are associated with multiple performance metrics. However, few studies have explored the impact of fungal components of the microbiome. Yeasts, one fungal group, have been used in the swine industry to improve health and growth, but only a handful of species have been considered. Here, we profile all fungi present in the gut of pre-wean piglets and determine associations between these organisms and growth rate. We apply machine learning (ML), an information-rich method, to model the abundances of fungi. Using this approach, we identify fungi previously used as probiotics such as Pichia, Rhodotorula, and Aspergillus, as well as taxa previously not associated with piglet growth, such as Wallemia which may serve as probiotic candidates. We also identify species which may have a negative impact on growth, including Lodderomyces, Clavispora, and Ustilago. Together, these findings provide a basis for the design of tailored therapeutics which target the microbiome of the weanling pig and advance our understanding of fungi and growth.

Technical Abstract: Modern swine production relies on consistent growth rates across individuals to maximize efficiency and earnings, but a subset of piglets are born small and grow slowly. Nutrition and management practices can augment growth of slow-growers but there remains a substantial portion of piglets which never reach their full growth potential. Traditionally, in-feed antibiotics were administered to enhance growth but with limitations on use, alternatives are needed. Emerging evidence suggests a role for bacterial members of the gut microbiome in growth, but fungal members have been largely overlooked in microbiome studies. Yeasts have been used in the swine industry to improve health and growth, but a limited number of species have been utilized, and study results are mixed. Here, we use ITS2 sequencing to profile the mycobiome of piglets at two timepoints in early development, postnatal days 14 (D14) and 21 (D21), just before weaning. We apply machine learning algorithms and differential abundance analysis to identify fungi which are associated with average daily gain (ADG). We identified several genera already in use as probiotics including Pichia, Rhodotorula, and Aspergillus, as well as taxa previously not associated with piglet health, such as Wallemia, which was significantly more abundant in good-growers at D14, and Lodderomyces, Clavispora, and Ustilago which was associated with poor-growers. These results contribute to our understanding of the mycobiome and growth and will be useful in the design of tailored therapeutics which target the weanling pig to improve lifetime performance.