Jackson, Kirby Shaw (2015-05). Associations of Feeding Behavior Patterns with Inter-Animal Variation in Feed Efficiency and Pre-Clinical Responses to Infectious Disease in Beef Cattle. Master's Thesis.
The objective of study 1 was to examine the associations between feeding behavior traits and performance and residual feed intake (RFI) in Brangus steers (N = 84). Steers with low-RFI phenotypes consumed 19% less (P < 0.01) DM intake while BW and ADG were similar compared to high-RFI steers. Steers with low RFI also spent 21% less time at the feed bunk, had 6% fewer (P < 0.05) bunk visit (BV) events, and tended (P = 0.08) to have 11% shorter meal durations per day than steers with high RFI. There were no differences in carcass quality or carcass income, therefore the reduction in feed cost of the low-RFI steers resulted in an increase (P < 0.05) in net revenue of $145 per animal compared to high-RFI steers. Time to bunk (TTB) was quantified on a daily basis as the interval length between feed truck delivery and the first BV event. Time to bunk was weakly correlated (P < 0.05) in a negative manner with ADG (-0.27) and positively correlated with exit velocity (0.25) and F:G ratio (0.25). Steers with low-TTB phenotypes gained 18% faster (P < 0.05), tended (P = 0.08) to have 11% more favorable F:G, and resulted in $88 more net revenue per animal (P < 0.05) than steers with high TTB. Results from this study demonstrated that animals with divergent phenotypes for RFI exhibited distinctive feeding behavior patterns, suggesting that feeding behavior traits could be useful as phenotypic biomarkers for RFI. The objective of study 2 was to characterize deviations in DM intake and feeding behaviors in bulls (N = 231) exhibiting clinical symptoms of bovine respiratory disease (BRD). The bulls were separated into 2 cohort groups based on observed clinical illness (N = 30) or those treated metaphylactically Draxxin (N = 201). A 2-slope broken-line regression model was applied separately on a population basis to the clinically-ill and metaphylaxis-treated cohorts to identify inflection points in DM intake and feeding behavior traits. The model detected inflection points for DM intake were 6.79 and 3.81 d prior to observed clinical illness or metaphylaxis treatment, respectively. Furthermore, the model detected inflection points for individual feeding behavior traits that (BV frequency and duration, Head down duration, maximum non-feeding interval, and non-feeding interval SD) ranged from 14.19 to 1.32 d prior to observed clinical illness, and from 12.59 to 3.79 d prior to metaphylaxis treatment. To further assess the value of monitoring deviations in feeding behavior traits as a method for pre-clinical detection of infectious disease individual CUSUM charts were constructed in a daily iterative manner to replicate real-time data analysis. The CUSUM model based on DM intake yielded a high proportion of true positives (87%; model predicted animal as ill) and high model test efficiency (89%) in the clinically-ill cohort, whereas, in the metaphylaxis-treated cohort the proportion of true positives detected (71%) and test efficiency (84%) were slightly lower. While time of model detection prior to observed clinical illness based on DM intake was not different (0.9 d; P > 0.10), time of model detection prior to metaphylaxis treatment was different (3.0 d; P < 0.05). Using BV duration, model times of detection were 2.7 (P < 0.05) and 7.9 d (P < 0.05) prior to clinical observation or metaphylaxis treatment, respectively. Results from study 2 demonstrated that use of statistical process control models to examine deviations in feeding behaviors were effective at predicting clinical symptoms of BRD, and that feeding behavior traits were more predictive than DM intake for pre-clinical detection of morbidity events in growing bulls.