Developing Regression Predictive Equations for Metabolizable Energy of High- Yielding Iranian Barley Varieties and Comparison with NIRS Method and the Values Indicated in the NRC Tables Based on the Performance of Male Broiler Chickens

Cereals are the main sources of calorie in poultry diets and corn is the most common cereal in poultry feed formulations; however, in some countries such as Iran, corn is mainly imported from other countries. In addition to import-associated problems, high volatility of corn price has recently resulted in a marked tendency between Iranian poultry producers to use other alternative grains in their formulations. Among the other cereals, wheat, rye, and barley are the most frequently used grains in poultry diets from which, barley is
believed to be a great alternative for corn due to its high productivity and good compatibility to the climatic conditions of the country. Barley is one of the most abundant grains raised in various areas of Iran and could be included in the formulations instead of corn. However, the extreme variability in nutrient contents observed within and between different barley varieties makes it difficult to achieve a good nutrient balance in barleycontaining diets. The energy content of feedstuffs is a topic of high importance for poultry nutritionists since birds regulate their feed intake based on dietary energy concentration. There are different methods to determine metabolizable energy (AME) content of feedstuffs including energy balance bioassay (excreta or ileal digestabased methods), referring to the standard tables describing feedstuff compositions (NRC and FEEDSTUFF tables), indirect AME determination using near-infrared spectroscopy (NIRS) technique and the use of multivariate prediction equations. Energy balance bioassay is the most reliable but time-consuming and expensive method while nutritionists need relatively simpler and faster methods for accurate feed AME estimation. On the other hand, contents of standard feed-describing tables are mean values obtained in a variety of previous studies performed under climatic conditions differing fairly from those of Iran. Most researchers agree that the values presented in the tables are not reliable and generalizable due to the extensive variability of feed types and varieties. During the last decades, various AME-predicting regression equations have been suggested for different feedstuffs but the data used for exploiting the equations have been obtained from animals and feeds genetically different from the modern commercial strains and varieties. Therefore, updating the equations using animals and feeds of today seems to be necessary. This study aimed at developing prediction equations for AME of the most producing Iranian barley varieties.

Three trials were conducted to develop regression predictive equations for apparent metabolizable energy (AME) of some of the most producing Iranian barley varieties in broiler chicken diets and to compare the outputs of the equations with the AMEn values estimated by infra-red spectrophotometry (NIRS) method as well as with the values published by the national research council (NRC, 1994). In the first experiment, 10 different barley varieties were analyzed for proximate composition. Then, in the second experiment, total tract AMEn values were determined for all of the barley varieties using 10 or 24-d-old broiler chickens and chromium oxide as an indigestible marker. Results of the two first trials were used to develop AMEn-predicting equations using SPSS software and "enter" procedure. To verify the accuracies of the predictive equations, the third trial was conducted using 400 broiler chicks in a completely randomized design consisting of five treatments with four replicates of 20 birds each. The AMEn content of the barley variety used in the third experiment was estimated according to the following five procedures: 1) The AMEn recommended by NRC (1994); 2) The AMEn predicted using the equation suggested by NRC (1994); 3) The AMEn values directly estimated in the balance trial (trial 2); 4) The AMEn values predicted by the equations developed in the 2nd trial; and 5) The AMEn estimated using NIRS method.

The equations obtained for 10 and 24-d-old broilers were: AMEn= 407.87*EE+27.27*NFE and AMEn= 271*EE + 33*NFE, respectively. The results showed that the AMEn values
exploited from the equations developed in the energy balance assay produced the closest performance to that of the AMEn values estimated directly during the same trial.

According to our findings, predictive equations can be used for accurate estimating of barley AMEn value for broiler diets formulation. In addition, our results showed that the old AMEn values and AMEnpredicting equations published by NRC (1994) and FEEDSTUFF (2014) are not accurate at least for Iranian barley varieties evaluated in the present study.