At-market sensor technologies to develop proxies for resilience and efficiency in dairy cows

Abstract

We hypothesise that at-market sensor technologies can be used to develop proxies for complex traits such as resilience and feed efficiency (FE). This was tested by comparing variables describing sensor data patterns (“curve-parameters”) from resilient or FE cows with non-resilient or non-FE cows. Sensor data included data from weighing scales, activity (steps) and rumination activity from neck collars, and milk production from the parlour or the milking robot. Curve-parameters were calculated for each sensor for each lactation for which data was available and included the mean, standard deviation (std), slope, skewness, and the autocorrelation. Data originated from a Wageningen Research farm, and included data from 1,800 cows with calvings between 1995-2016. During this time frame, there were 98 lactations with sufficient feed intake recordings to compute FE at lactation level (DMI (kg) / milk yield (kg)), and to rank them accordingly. The 1,800 cows that could be ranked according to their lifetime resilience (ability to re-calf in combination with the number of health and insemination events) based on scores for each of the, in total, 5,771 lactations. Subsequently, the 20% or 10% most and least FE or resilient lactations, respectively, were selected. Curve-parameters of these selected lactations were compared. Results imply that using a single sensor, or a single curve parameter, is likely to be insufficient as a proxy for resilience of efficiency. Future research should focus on studying which combination of curve parameters and sensors are most informative as proxy for these two complex traits.