Genetic analysis of sensor-based activity in Australian Merino sheep

The use of on-animalsensor data enables continuous monitoring, providing valuable behavioural observations with the potential to be used for assessing an individual’s health and productivity. This study aimed to estimate the genetic parameters of sheep general activity, measured as vector magnitude, which represents the overall magnitude of the movement to provide an objective activity level measurement. Activity records of 1,149 Merino ewes were collected with ActiGraph™ wGT3X-BT® devices over a period of 17 days (after editing 10 non-continuous days used in the analysis). Single trait estimates (STE) were initially considered for this analysis; however, random regression (RR) was also checked, given the patterns observed on raw records over time and days. The STE of heritability (h ² ) and repeatability (Rep) in the whole dataset were 0.19 ± 0.06 and 0.56 ± 0.08, respectively. Across 10 days, single estimates of h ² ranged from 0.15 ± 0.06 to 0.25 ± 0.08 and Rep from 0.45 ± 0.08 to 0.66 ± 0.08, while RR models resulted in lower h ² (0.11 to 0.23) and Rep (0.22 to 0.34) over these days. Genetic correlations between days (up to day 16) and times (2 hours) were high, which did not indicate the potential of random effects of time and day in genetic parameters for sheep activity. These estimated parameters can be considered for future welfarerelated breeding programs, and the amount of repeatability can guide the development of an effective measurement protocol.