Data management and analysis for two years of monitoring a deer slaughter premises
by
R. P. Littlejohn
AgResearch, Invermay
Coauthors: J. C. Pollard, J. M. Stevenson-Barry

Data was collected from an Otago deer slaughter premises on one day every week for two years, describing the whole experience of the deer from farm to carcass, including farm practices, transport, weather, pre-slaughter lairage, behaviour in race, animal details, bruising assessment and meat pH. The objective was to identify factors that could lead to improved meat quality and animal welfare.

The data set constituted 7953 animals in 275 loads from 157 farms on 81 days, using 43 carriers. Care was required to ensure alignment for each animal of data collected from different sources, to reconstruct partially reported data, and in many other data reporting issues. However, individual data verification was possible insofar as the data set was “large” but not “huge”.

About 60 response variables were reduced to 5-10 of the most important, which were analysed using mixed models with load as the random effect and 12-14 main effect explanatory factors. Load BLUPs were then regressed on variables describing farm history and weather (which were only partially reported over the first 5 months), and major terms were fed back into the full mixed model.

Some major differences were found with breed, gender, health status, isolation, time spent in yards, and some weather variables were significantly related to response variables, but no evidence was found relating on-farm factors to response variables.

Date received: April 23, 2002