Production responses of dairy cows to precision feeding based on historical performance during short-term changes in supplementation

The increasing automation of dairy systems enables precision feeding, where dietary supplements can be tailored to individual animals. However, data describing short-term production responses to such supplementation are lacking, limiting the development of individualized feeding algorithms. This study aimed to 1) examine dairy cow responses to short-term changes in top dress supplementation and 2) evaluate two simple precision feeding strategies compared to a conventional total mixed ration (TMR). Twenty-four lactating Holstein cows were assigned to one of four top-dress treatments: soybean meal (SBM), corn grain (CG), corn gluten feed (GLT), or no supplement in a replicated 4 × 4 Latin square during a 36-d training phase. Subsequently, cows were reassigned to three feeding groups: Control (conventional TMR), algorithm 1 (based on mean past performance), or algorithm 2 (based on the slope of past responses), and evaluated over four 7-d testing periods. During training, dry matter intake (DMI) and milk yield (MY) increased on all top dress treatments relative to the Control treatment (P < 0.001), but feed efficiency (FE; MY/DMI) declined (P < 0.001). SBM reduced milk fat percentage and yield (P < 0.05), while milk protein, body weight (BW), and energy-corrected milk (ECM) were unaffected (P > 0.27). During testing, DMI tended to increase in Algorithm 1 vs. Control (25.2 vs. 22.7 kg/d; P = 0.08), with a corresponding tendency for reduced FE (1.45 vs. 1.61; P = 0.07). No performance differences were detected between algorithm 2 and Control (P > 0.50). Milk composition, ECM, and BW were similar across feeding groups (P > 0.33). Feed cost, milk revenue, and income over feed cost (IOFC) did not differ between precision-fed and control groups (P > 0.33). These findings suggest that short-term historical production responses may be insufficient to inform future supplementation decisions and that more advanced approaches, including new data streams, may be necessary to improve individual cow performance using precision feeding strategies.