Evaluation of sources of error in weight records of commercially raised growing pigs : a thesis presented in partial fulfillment of the requirements for the degree of Master of Veterinary Studies in Epidemiology at Massey University, Palmerston North, New Zealand

Abstract

The objective of this research was to investigate sources of errors in pig weight measurements. Three studies were conducted using data from one commercial New Zealand pig farm. In XChapter 4X, finisher pigs fed ad libitum or via a computerized liquid feeding system were weighed four times a day over a fourday period. Results showed that standardization of weighing time reduced diurnal fluctuations in pig weight. However, multivariate analysis showed that there was a significant interaction between day and time of day, which indicates that diurnal fluctuations in live weight are not consistent between days, particularly in ad libitum fed pigs. Hence, XChapter 5X investigated whether overnight feed withdrawal for 11 hours (weaners) or 17 hours (growers and finishers) is effective in reducing between-pig variation in live weight and growth rate. For grower and finisher pigs, feed withdrawal was associated with a reduction in variability in live weight and growth rate by up to 11.5%, whilst the effect was inconsistent in weaner pigs. It is recommended to repeat the investigation on other farms to assess long-term effects on pig performance before general recommendations can be made. XChapter 6X compared the magnitude of sampling error when sampling pens from batches of pigs, using different sample sizes and sampling methods. Increasing the portion of randomly selected pens reduced the sampling error, but in a diminishing manner. Purposive selection of two pens reduced sampling error by more than 64% compared with random sampling. However, purposive sampling introduces the risk of obtaining biased estimates. Thus, it is recommended to select pens from batches at random. These results may be used as an educational tool to demonstrate how to minimize errors in pig weights. Collecting more accurate weight records is likely to lead to improved interpretability of pig weights, and may promote better use of production data.