Growth modeling of dairy heifers in Québec based on random regression

Abstract

Cue, R. I., Pietersma, D., Lefebvre, D., Lacroix, R., Wade, K., Pellerin, D., de Passillé, A-M. and Rushen, J. 2012. Growth modeling of dairy heifers in Québec based on random regression. Can. J. Anim. Sci. 92: 33–47. A total of 144 006 weight (calculated from tape girth measurements) and height data records from Québec dairy heifers were analyzed using random regression to estimate growth curve parameters of Ayrshires, Brown Swiss and Holstein animals to permit prediction of individual heifer growth from 0 to 32 mo. There were, on average, 5.15 records per heifer (minimum 3 records, maximum 25 records). The body weight data were analyzed using linear and quadratic fixed and random regressions, with a power-of-the-mean (POM) function to model the residual variance. The POM was 1.2 for Holstein and Ayrshire and slightly less than 1 for Brown Swiss. Estimated body weight at 24 mo was 507, 564, 624 kg, for Ayrshires, Brown Swiss and Holstein, respectively. The height data were analyzed with a Brody, monomolecular non-linear growth curve model. Mature height was estimated to be 148 cm in both Holstein and Ayrshires, and 150 cm in Brown Swiss. Random regression models were shown to be able to predict individual growth, and can be incorporated in decision-support tools to help producers reducing the average age at first calving.

Linear body measurements and productivity of subtropicalHolstein–Friesiandairy calves

“…The combination of all body measurements increased marginal and conditional R 2 by 29% and 31%, respectively. The results of this study are in agreement with the results of those earlier studies that reported higher R 2 values for models that contained a single measurement, and combinations of body measurement using different types of cattle [38][39][40] or calves. 41 It was concluded that HG was more highly correlated with LW than other body measurements.…”

“…An observation such as this allows farm management to recognise and address problems promptly, thus improve individual calf performance. LW has also been found to be a reliable indicator of calf progress and, by proxy, allows producers to adjust management to improve productivity, 38 and future lactation performances. 43,44 Disease events cause losses in LW as it promotes energy diversion away from growth to immune defences, and diarrhoea causes nutrient loss, dehydration, and reduction in feed intake by suppressing appetite.…”

Predictive equations for early-life indicators of future body weight in Holstein dairy heifers

“…Even though feeder, season the calf was born in, and the year the calf was born in significantly affected the BW of heifers up to 400 d of age, they were excluded from the final model because they were random factors that cannot be replicated; therefore, they would not be able to be determined for a predictive equation, which was similar to reports from (Cue et al, 2012). Our objective was to use the significant variables from early life to generate a predictive model that farms could use to select dairy replacement heifers at an earlier age.…”

“…Year effect was highly significant in the model up to 400 d with a decrease of 40.5, 8.6, 2.6, and 11.1 kg from 2015 to 2019. Other research has found year to also be significant when measuring BW over time (Donovan et al, 1998;Cue et al, 2012). Based on evidence seen by Dietrich (2015) looking at standard plate and coliform count in autofeeders pre-and postcleaning, bacteria buildup in the autofeeders as they age could explain some of the differences seen among years.…”

“…To generate an equation to better monitor measurable variables that affect heifer performance, all 3 variables were removed from the model. Cue et al (2012) removed herd, year, and month because they wanted to generate a predictive model. Excluding significant variables introduces error, but including them will diminish the usefulness of the model when implementing it on-farm.…”