Feed represents the primary cost of broiler production, thus the formulation of cost-effective diets that meet broiler nutritional requirements is critical. To ensure this objective is met, nutrient specifications of feed ingredients must be accurately determined. However, Australian broiler nutritionists have expressed concern, as many nutrient specification databases contain dated information or lack Australian specific data. Therefore, the aim of this project was to compile current Australian and global data into a database of nutrient specifications for commonly used feed ingredients within the Australian chicken-meat industry. This database also evaluated the variation within feed ingredients for both Australian and global nutrient specifications, and identified areas which require further study.
Initially, integrated Australian broiler nutritionists were surveyed to identify the most common feed ingredients and key nutrients for consideration in the database. Within this survey, it was identified that information on digestible P, digestible Ca and fibre (all fractions) were rated as important by all nutritionists surveyed (9), as there is presently a lack of data for these specifications. Data were sourced and compiled from a total of 12 companies/databases, and, where data were lacking, references were sought from the literature. Data were collected for 42 ingredients with 102 nutrient specifications per ingredient, where all data were available. The mean value, sample number (n) and standard deviation (SD) were collected for each nutrient specification. From these data, the overall mean, total sample number and average standard deviation reported were calculated for both Australian data and global data. The sample size required to predict the mean value for each nutrient specification to 95% accuracy was also calculated.
Unsurprisingly, within the database there is substantially more global samples than Australian samples, and the number of samples for some Australian ingredients is quite low. Combined with the notable variability observed in the data, it is evident that published Australian sample numbers are not adequate to accurately predict mean values. For example, the mean protein value of wheat for Australian data is 111.6 g/kg (total n = 370; SD = 7.6) and global data is 119.2 g/kg (total n = 37,874; SD = 14.2), representing a 7% difference between the mean protein level. The SD for global data is larger than Australian data, which is expected as there is greater variation in agricultural practices, cultivars, environmental conditions etc., across multiple countries than within one country. However, upon calculation of the sample size required, it is evident that the number of Australian samples (n = 370) is inadequate, as 706 Australian samples are required to determine the mean protein content of wheat to 95% accuracy. When calculated for global data, it is determined that 2,177 samples are required, which is well below the actual sample number within the database (n = 37,874) and thus this figure is reliable. Overall, only 7% of the Australian data compiled meets the sample number required to accurately predict the mean value within 95% accuracy (13% within 90% accuracy), compared to 20% of global data (40% within 90% accuracy). Therefore, greater focus on determining the nutritive value of feed ingredients is required, particularly for Australian data.
ACKNOWLEDGEMENTS: The authors would like to sincerely thank AgriFutures Chicken Meat for funding this project. The authors are also extremely grateful for the support from the following companies to provide their data for inclusion and publication within the database; Adisseo, Ajinomoto, Cootamundra Oil Seeds, DuPont, Evonik, Poultry Hub, Novus, Premier Nutrition and RCI. The authors would also like to acknowledge the following open access sources which were also included; Feed grain Partnership, Feedipedia and INRA.
Abstract presented at the 30th Annual Australian Poultry Science Symposium 2020. For information on the next edition, click here.