Productive differentiation of dairy farms according to feeding strategies in the Central Santa Fe Basin, Argentina

Abstract

The intensification of dairy farming in the Pampas region has led to increasing diversification of feeding strategies. In this context, this study aimed to evaluate whether these strategies allow consistent discrimination of dairy farms in the Central Santa Fe Basin according to their daily milk production. Linear Discriminant Analysis was applied using data from 71 farms surveyed in the INTA Dairy Sector Survey. Farms were grouped into three production strata defined a priori: small (≤2,500 L day^-1), medium (2,501–4,000 L day^-1) and large (>4,000 L day^-1). Two discriminant functions were obtained, of which the first explained 87% of the variance and was mainly associated with the intake of concentrate feed, maize grain, silage, and hay, as well as the contribution of grazing. Overall classification accuracy was 67.6%, decreasing to 54.9% under cross-validation. Performance by stratum, assessed using the F1-score, was higher for small and large farms (0.81 and 0.60, respectively) and lower for medium farms (0.48). Based on the discriminant structure, one representative farm per stratum was identified using Mahalanobis distance. It is concluded that feeding strategies allow moderate discrimination among dairy farms and enable the identification of feeding profiles that provide useful empirical references for subsequent applications, particularly in productive, economic, and environmental modelling.