Development of mathematical models to predict calcium, magnesium and selenium excretion from lactating Holstein cows

Abstract

The aim of this study was to develop and evaluate mathematical models that predict mineral excretion, particularly calcium (Ca), magnesium (Mg) and selenium (Se), from lactating dairy cows. Mineral excretion can be affected by several dietary factors. A deficiency in Ca or Mg application to pasture, among other factors, can contribute to grass tetany or wheat pasture poisoning in cows, whereas an excess can cause runoff into water supplies. Manure application with high Se concentration can also result in runoff, causing the bioaccumulation of selenium in aquatic ecosystems, wetland habitats and estuaries, leading to toxic levels in fish. A database composed of studies relating to mineral utilisation in lactating dairy cows conducted after and including the year 2000 was compiled. A meta-analysis was conducted with the aim of creating multiple empirical equations to predict Ca, Mg and Se excretion from lactating dairy cows. Calcium intake, feed Ca content, milk yield, milk protein content and acid detergent fibre content in diet were positively and linearly related to Ca excretion. Dietary crude protein content and milk fat content were negatively related to Ca excretion. Magnesium intake, feed Mg content and milk yield were positively and linearly related to Mg excretion. Selenium content of diet and dry matter intake were linearly and positively related to Se excretion. Two sets of models were developed using or excluding the intake variable and both sets of models were evaluated with independent data originating from commercial herd or individual animals. In general, intake measurements improved prediction when evaluated with independent datasets (root mean square prediction error = 8% to 19% vs 14% to 26% of the average observed value). There were substantial mean biases, particularly those evaluated with data from a commercial farm, perhaps due to inaccurate feed intake measurements. Although there was generally good agreement between predicted and observed mineral excretion, model development and evaluation would benefit from an expanded database.