Get reliable rangeland science

Precision livestock technology (PLT) can improve sustainability of beef production on rangelands. Key to the advancement of PLT is the integration of technologies and data streams with animal nutrition models to better inform management decisions. Knowing dry matter intake (DMI) is essential for setting stocking rates and estimating forage removal by grazing beef cattle; however, estimating DMI for grazing cattle is difficult due to dynamic changes in forage quality and animal weight throughout a grazing season. A study was conducted from 2021-2023 at the South Dakota State University Cottonwood Field Station (Cottonwood, SD, USA) to estimate daily DMI for grazing steers. The objectives of our study were to 1) utilize machine learning (ML) to predict daily estimates of forage quality, 2) estimate average daily gain (ADG) using in pasture weighing systems, and 3) incorporate forage quality and ADG estimates into animal nutrition models to predict individual animal DMI. From 2021-2023, bi-weekly forage samples were collected and used to train a multivariate random forest model to predict daily acid detergent fiber (ADF) based on climate and imagery metrics derived from Google Earth Engine. Root mean square error of prediction was 2.6 with a 0. 81 correlation between predicted and observed values of ADF. SmartScalesTM (C-Lock Inc., Rapid City, SD, USA) were deployed in six pastures to estimate daily animal weights for grazing steers. Smoothing splines were used to estimate ADG allowing for non-linear changes in animal performance. Daily estimates of ADF and ADG were used to calculate daily DMI for individual animals using equations from the Nutrient Requirements of Beef Cattle. Overall, average DMI estimates for individuals ranged from 2-3% of body weight, which is within expectations for free ranging livestock. This paper address how big data, technology, and machine learning can be integrated to better aid grazing monitoring and forage demands for livestock.