A multiple correspondence analysis on the feeding and management practices of Australian horse owners

This study aimed to investigate the current feeding and pasture management practices of Australian horse owners.
An online survey of 4512 horse owners was conducted from July to August in 2017, seeking information on participant demographics, horse demographics, current feed rations, management practices and pasture conditions. Participants were female (95%), and lived on the East coast of Australia (83.3%). Most horses (80%) grazed on pasture with ≥75% ground cover and 90% of horses were fed daily in addition to grazing access. The majority of horses were mature horses (2.5-20 years aged, n=2757). To better visualize, describe and cluster participant feeding and management practices a multiple correspondence analysis (MCA) was performed on the owners of mature horses. The MCA analysis assigns weights to variables creating a pointcloud of data; within the cloud, clusters represents a group of participants engaging in similar practices. The analysis revealed 4 clusters, using MCA active variables of body condition score, workload, stabling, pasture management and ration digestible energy (DE). Supplementary quantitative variables were participant location, years of experience, industry involvement, sources of information on feeding, how participants measured feed, and horse demographics. The supplementary qualitative variable was participant age. The MCA was conducted in R using the packages FactoMineR and factoextra. The first cluster included participants from Victoria who were non-competitive riders, whose horses were not in work, kept on ‘horse sick’ pasture with >75% ground cover, and receiving supplementary rations. Cluster two included competitive riders, whose horses were stabled for part of the day and given intermittent access to pasture when not stabled; pasture was short, with <25% ground cover. Horses were Warmblood and in moderate work (180-240 minutes work/week). Cluster three was composed of participants that did not respond to the all the survey questions resulting in a NA value for the MCA, highlighting the difficultly of conducting survey research. Cluster 4 was constructed of participants from South Australia, who kept their horses stabled with no pasture access. Horses were fed above the NRC (2007) DE recommendation and feed was measured with a container of a known quantity. The MCA was used to detect and represent underlying patterns in the data set that were not able to be represented through routine descriptive statistics. The results of the clustering reveal that within Australia horse management varies greatly depending on where horse owners live, the horse’s workload and how those horses are managed. The clusters identified from the MCA could be used for targeted education and resources for horse owners on equine nutrition and pasture management or as a guide for directing future research.