Feed efficiency is an economically important trait in the swine industry since feed accounts more than 50% of total production costs. A measure of feed efficiency, residual feed intake (RFI), is defined as the difference between observed and expected feed intake based on production and maintenance requirements. Since 2001 at Iowa State University (ISU), divergent selection for improved (Low RFI) and reduced feed efficiency (High RFI) has been conducted in Yorkshire pigs for ten generations. Using these selection lines, the over-arching objective of this dissertation was to further our knowledge of the biological and genetic basis of RFI in pigs. The main objectives were to investigate genotype-by diet interactions, identify genomic regions associated with RFI and component traits, validate insulin like growth factor I (IGF-I) as an early genetic indicator of grow-finish RFI, and to evaluate correlated responses to selection for grow-finish RFI on feed efficiency and performance of nursery pigs.

To quantify genotype by diet interactions for RFI, in generation (G) 8 through G10 of the high and low RFI lines, a lower-energy, higher-fiber (LEHF) diet was fed to a subset of pigs and compared to the performance of pigs fed a standard corn and soybean-meal based diet, similar to the diet fed during selection, which was higher in energy and lower in fiber (HELF). These diets differed in metabolizable energy (3.32 vs. 2.87 Mcal/kg for the HELF vs. LEHF diet) and neutral detergent fiber (9.4 vs. 25.9% NDF). When pigs were fed the HELF, the Low RFI pigs had lower average daily feed intake (-12%), energy intake (-12%), average daily gain (-6%), and backfat depth (-12%) than High RFI pigs (P < 0.05). Regardless of RFI line, performance was reduced when pigs were fed the LEHF compared to the HELF diet. For the LEHF diet compared to the HELF diet, differences between the RFI lines were smaller for average daily feed intake (-11%), energy intake (-10%), gain to feed ratio (+2%), and RFI (-6%) (P < 0.05). Feed digestibility was reduced when pigs were fed the LEHF diet, with the Low RFI pigs digesting significantly (P ≤ 0.04) more dry matter (+7%), gross energy (+7%), nitrogen (+10%) and NDF (+32%) than High RFI pigs when fed the LEHF diet. However, no line differences in digestibility were observed when the HELF diet was fed. Estimates of genetic correlations of performance traits across diets were high and positive for RFI and component traits, with a 0.87 à ± 0.28 genetic correlation for RFI across diets. The observed correlated response to selection in RFI when feeding the LEHF diet was 55% less than predicted based on the genetic correlation for RFI between diets. Genotype-by-diet interactions were further investigated by estimating single nucleotide polymorphism (SNP) by diet interactions, but these were found to account for less than 0.7% of the genetic variance in any given non-overlapping 1-Megabase window for RFI and component traits. By comparing the top genomic regions associated with RFI for the HELF and LEHF diets separately, we observed that the top associations were located on Sus scrofa chromosome (SSC) 2 near IGF2 (insulin like growth factor II) when pigs were fed the HELF but on SSC 6 for the LEHF diet, which demonstrates that at least some genomic regions associated with these traits were different between diets. This agrees with the estimated genetic correlation between diets for RFI (0.87 à ± 0.28) and provides more evidence of genotype-by-diet interactions for RFI.

Genomic regions associated with RFI and component traits given the HELF, LEHF and both diets combined were identified using a genome-wide association study (GWAS). Two genomic regions were associated with multiple traits, indicating pleiotropic effects, on SSC1 near MC4R (melanocortin-4 receptor) and on SSC2 near IGF2. Results showed that the genetic architecture of RFI was highly polygenic. Genomic regions were also identified by evaluating genomic regions under selection in the ISU and in an independent Large White population that was also divergently selected for RFI (INRA). Regions were identified on SSC 2 near CAST (calpastatin) and on SSC 13 near GAPBA (GA binding protein transcription factor alpha subunit), which were different than associations found using GWAS. However, findings also suggested that the differences in RFI between the ISU and INRA Low and High RFI lines may be the result of selection affecting different genes and biological pathways, as few common regions were identified to be under selective pressure in the two populations.

Using IGF-I data collected in G2 through G5, and in G10 and G11, IGF-I was found to have a positive genetic correlation with grow-finish RFI (0.54 à ± 0.19 for G2-G5 and 0.51 à ± 0.48 for G10-G11), indicating IGF-I is a good early biological marker for grow-finish RFI. In nursery-aged pigs in G10, the Low RFI line was found to eat less (-20%), grow slower (-10%) and have greater feed efficiency (+12% measured as gain to feed ratio) compared to the High RFI line, showing that selection for grow-finish RFI also improved nursery feed efficiency in the Low RFI line.

In conclusion, RFI is a biologically and genetically complex trait with no genes with major effects and many genes contributing small effects. Nutritional value of the diet fed during selection impacts feed efficiency and its genetic architecture. Therefore, genotype-by-diet interactions must be taken into consideration in selection programs, particularly those that desire to improve feed efficiency. Genomic selection would be a good strategy to improve feed efficiency because of the highly polygenic genetic architecture of RFI. In addition, juvenile IGF-I can be used as a genetic indicator for grow-finish RFI. Finally, correlated responses to selection for grow-finish RFI also led to improved feed efficiency of Low RFI pigs in the nursery.