Prediction of Eggshell Ultrastructure via Some Non-destructive and Destructive Measurements in Fayoumi Breed

Possibilities of predicting eggshell ultrastructure from direct non-destructive and destructive measurements were examined using 120 Fayoumi eggs collected from the flock at 45 weeks of age. The non-destructive measurements included weight, length and width of the egg. The destructive measurements were breaking strength and shell thickness. The eggshell ultrastructure traits involved the total thickness of eggshell layer, thickness of palisade layer, cone layer and total score. Prediction of total thickness of eggshell layer based on non-destructive measurements individually or simultaneously was not possible (R2 = 0.01 to 0.16). The destructive measurements were far more accurate than the non-destructive in predicting total thickness of eggshell layer. Prediction based on breaking strength alone was more accurate (R2 = 0.85) than that based on shell thickness alone (R2 = 0.72). Adding shell thickness to breaking strength (the best predictor) increased the accuracy of prediction by 5%. The results obtained indicated that both non-destructive and destructive measurements were not useful in predicting the cone layer (R2 not exceeded 18%). The maximum accuracy of prediction of total score (R2 = 0.48) was obtained from prediction based on breaking strength alone. Combining shell thicknesses and breaking strength into one equation was no help in improving the accuracy of prediction.

Eggshell quality was very important in field of poultry industry at either table eggs or hatching domain. Because, the poor eggshell quality causes economic losses due to the distance between the farms and marketing places. On the other hand, the integrity of eggshell and avoiding any cracks are very important for incubation eggs. Few studies have recorded the relationship among some non-destructive and destructive measurements of the eggs of chickens (Abanikannda et al., 2007;Aygun and Yetisir, 2010), Guinea fowl (Obike and Azu, 2012;Alkan et al., 2013) and quails (Kul and Seker, 2004;Rathert et al., 2011;Ojedapo, 2013). However, no published reports on the use of nondestructive and destructive measurements in constructing prediction equations aiming predicting of eggshell ultrastructure properties of the laying hens.
The aim of this study was to develop prediction equations to predict the total, palisade and cone thickness of eggshell layers from destructive and nondestructive properties instead of the high costs needed to inspect the ultrastructure of eggshell using scanning electron microscope (SEM) technique.

MATERIAL AND METHODS
This experiment was conducted at Poultry Production Department, Faculty of Agriculture, Ain Shams University. The Fayoumi hens were housed in individual cages from 16 weeks of age up to the end of the experiment at 50 weeks of age. All birds were reared under similar environmental, managerial and hygienic conditions. To assess eggshell parameters, a total of 120 Fayoumi eggs at the 45 weeks, were taken. Each egg was weighed and its dimensions (width and length) were measured using a digital caliper to calculate shape index. The thickness (mm) of the shell with intact membranes was measured at three different points in the middle part of the egg using a dial gauge micrometer. Shell breaking strength (N) were measured by quasistatic compression using an Instron (UK527, High Wycombe, Buckinghamshire, UK) fitted with a 50 N load capture at compression speed of 5mm/min. Breaking strength was measured as the maximum force (N) required fracturing each egg.
Samples of eggshell were chosen to investigate thickness layer eggshell (measure effective thickness). The specimens were prepared by cutting a piece (1 cm 2 ) of shell from the equatorial region of each egg. The shell membranes were removed by chemical solution (Radwan, 2007). Following these preparative treatments, two samples from each egg were mounted in inner side uppermost and in vertically manner on aluminum stubs, coated with gold for 3 min in an Emscope Sputter Coater. These samples were examined using JEOL JSM-T330A scanning electron microscopy at 15 Kv. The cross-sectional thickness of palisade and cone layers were directly measured in μm using scaling software provided with the SEM at a magnification of ×200. The total thickness of each specimen was measured as the distance from its' outermost surface to the point where the basal caps inserted into the shell membranes. The thickness of the cone layer was also assessed, this being the distance from the basal caps to the point at which the palisade columns first fused. Subtraction of these two measures provided a thickness of the palisade thickness or effective thickness (Bain, 1990;Solomon, 1991). Triplicate measures were performed in each case and the mean values were used in the statistical analysis. Ultrastructure of eggshell was assessed according to the procedures outlined by Robert and Brackpool (1994).

Statistical analysis
Data on non-destructive (weight, length and width of egg) and destructive (breaking strength and shell thickness) measurements, as predictors, and the ultrastructure eggshell properties (total, thickness of palisade layer, cone layer and total score), as response variables, were analyzed according to the following regression model of SAS (2005): Y i = a+b 1 X 1i +b 2 X 2i + ... +b p X pi +e i Where: Y i = the dependent variable (ultrastructure eggshell traits) of the i th egg; a = intercept; X pi = the p th independent variable (non-destructive and destructive traits) of the i th egg; b 1 , b 2 , ..., b P = partial regression coefficients of Y on X's; and e i = error assumed to be normally independent distributed with mean = 0 and variance = 2 e  .
The regression analysis was performed using the REG procedure of SAS (2005).

Detecting multicollinearity
To indicate multicollinearity, a high degree of correlation among the independent variables, as among the considered predictors in the present study, tolerance value and variance inflation factor value (VIF) were calculated according to Montgomery (2001).

RESULTS AND DISCUSSION
Means, coefficients of variation and the range for nondestructive, destructive and eggshell ultrastructure are given in Table 1. It appeared that the coefficients of variability for the traits describing the destructive and ultrastructure of eggshell were comparable (9.92% to 19.61%) and much higher than those for non-destructive properties (3.79% to 6.75%).

Correlations
Correlation coefficients between the traits describing the non-destructive, destructive and ultrastructure eggshell properties are given in Table 2.
The reason that may be advanced for this negative relationship is the fact that egg length is the denominating factor in estimating shape index according to Panda (1996) and Gunlu et al. (2003). This observation agrees with reports of Choprakarn et al. (1998).
Egg width shows positive correlation with shape index (0.59), this is because shape index in directly related to egg width, and this result is similarly observed by Obike and Azu (2012); Kul and Seker (2004); Rathert at al. (2011) and Aygun and Yetisir (2010). The reason for this could be as a result of the denser part of the egg (yolk) occupying the width area, which translates to heavier weight for the egg.
It seems that the non-destructive traits of the egg (weight, length, width, and shape index) were independent (p>0.05) of breaking strength (r = -0.28 to +0.09) and eggshell thickness (r = -0.24 to +0.24).
The eggshell thickness and breaking strength in Fayoumi breed were highly correlated positively (0.77, Table 2). This value was much higher than the value of 0.47 obtained on Bandara, Mandarah and Norfa native Egyptian strain as package by Fathi et al. (2010).
The traits describing eggshell ultrastructure were weakly correlated (p>0.05) with non-destructive traits of the egg (r = -0.35 to +0.24). These results indicated that the non-destructive traits were not useful in predicting the eggshell ultrastructure properties. The strong correlation coefficients obtained between thickness of total and

Multicollinearity
Values of tolerance and VIF of the predictors are given in Table 3. Tolerance value represents the amount of variability in independent variable that is not explained by other independent variables. The tolerance values indicated that 28% of the variability in egg weight is not explained by the other predictors. The corresponding figures were 37% for egg length, 31% for egg width, 16% for breaking strength and 15% for shell thickness. The values of VIF illustrated that 96.43% of the variance in egg weight could be explained by the other predictors. The corresponding figures were 97.27% for egg length, 96.80% for egg width, 93.67% for breaking strength and 93.34% for shell thickness. These results indicate that the degree of multicollinearity among the four predictors could be negligible. So, these findings can be trusted and applied to other samples.

Prediction equations
Regression equations of total thickness of eggshell layer on non-destructive (weight, length, and width of the egg) and destructive (breaking strength and eggshell thickness) measurements together with their accuracy of prediction (R 2 ) values are given Table 4.  Prediction of total layers: Prediction of total thickness of eggshell layer based on non-destructive measurements (weight, length, and width of the egg) individually (E 1 , E 2 , and E 3 ) or simultaneously (E 4 ) was not possible (R 2 = 0.01 to 0.16). This is due to the low correlations between the dependent variable and the three predictors (r = -0.29, -0.26, and -0.09, respectively). It appeared that the destructive measurements (breaking strength and shell thickness) were far more accurate than the non-destructive in predicting total thickness of eggshell layer. Prediction based on breaking strength alone (E 5 ) was more accurate (R 2 = 0.85) than that based on shell thickness alone (E 6 ) (R 2 = 0.72). Adding shell thickness to breaking strength (the best predictor) to formulate E 7 was useful in increasing the accuracy of prediction (R 2 = 0.90). The high accuracy of prediction obtained from these equations were due to the strong relationship between total thickness of eggshell layer and both of breaking strength (0.92) and shell thickness (0.85). The limited improvement in accuracy of prediction obtained from combining shell thickness to breaking strength into one equation was due to the strong correlation (0.77) between the two predictors.
Prediction of palisade layer: It appeared that weight, length and width of the egg together (E 11 ) or individually (E 8 , E 9 , and E 10 , respectively) were not efficient in predicting the thickness of palisade layer (R 2 = 0.04 to 0.12). The maximum accuracy of prediction (R 2 = 0.79) was obtained with the inclusion of length of egg with the two direct destructive measurements into one equation (E 15 ). Dropping length of egg from this equation to form E 14 was associated with slight reduction in accuracy of prediction (R 2 = 0.76). Prediction based on breaking strength alone (E 12 ) would yield higher accuracy (R 2 = 0.69) than that based on shell thickness alone (R 2 = 0.64).
Prediction of cone layer: The results presented in Table  4 indicated that both non-destructive and the destructive measurements were not useful in predicting the cone layer (R 2 not exceeded 18%). This is due to the weak correlation between cone layer and these predictors ( Table 2).
Prediction of total score: The maximum accuracy of prediction (R 2 = 0.48) was obtained from prediction based on breaking strength alone (E 27 ). Adding shell thicknesses to E 27 to form E 29 not help in moving the accuracy of prediction.

CONCLUSION
It is not possible to predict ultrastructure properties of the Fayoumi egg from non-destructive measurements. The destructive measurements (breaking strength and shell thickness) individually or together are accurate predictors for total and Palisade layers.