Table 3.
NC + phytase level, FTU/kg | Probability of contrast | Dose response model1 | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Item | PC | NC | 250 | 500 | 1,000 | 2,000 | SEM | PC vs. NC | NC vs. phytase | Linear | Exponential |
AID P | 48.5 | 50.6 | 56.4 | 63.6 | 68.3 | 69.6 | 5.47 | 0.583 | 0.0002 | <0.001 | 0.016# |
AID Ca | 52.3 | 66.3 | 66.4 | 63.2 | 69.8 | 66.2 | 4.55 | 0.0013 | 0.7515 | 0.799 | 0.939 |
AID N | 74.2 | 71.5 | 75.4 | 78.7 | 78.9 | 80.4 | 1.39 | 0.142 | <.0001 | <0.001 | 0.012# |
AID IP6 | 35.7 | 38.8 | 53.8 | 67.2 | 75.9 | 82.8 | 4.8 | 0.3352 | <.0001 | <0.001 | <0.001# |
ATTD P | 38.3 | 38.1 | 54.3 | 61.2 | 68.1 | 71.1 | 1.75 | 0.9313 | <.0001 | <0.001 | <0.001# |
ATTD Ca | 49.7 | 61.9 | 69.9 | 73.5 | 74.1 | 69.9 | 3.74 | 0.0009 | 0.002 | 0.198 | 0.294 |
ATTD DM | 93.6 | 94.3 | 94.8 | 94.6 | 95 | 95.1 | 0.67 | 0.1485 | 0.1079 | 0.115 | 0.501 |
ATTD N | 81.6 | 83.2 | 82.1 | 83.8 | 84.6 | 85.1 | 1.11 | 0.0624 | 0.494 | 0.031# | 0.74 |
ATTD Na | 82.5 | 85.3 | 85.6 | 88.7 | 88.5 | 91.9 | 1.45 | 0.2258 | 0.2465 | <0.001# | 0.997 |
ATTD energy | 82.7 | 84.6 | 84.1 | 84.3 | 84.7 | 85.9 | 0.58 | 0.0207 | 0.8414 | 0.034# | 0.62 |
ME, kcal/kg DM | 3516 | 3506 | 3528 | 3530 | 3544 | 3611 | 35.9 | 0.7533 | 0.265 | 0.031# | 0.744 |
ret P, g/day | 1.31 | 1.05 | 1.42 | 1.55 | 1.65 | 1.63 | 0.23 | 0.1453 | 0.0027 | 0.022 | 0.165# |
Total P excretion, g/day | 1.89 | 1.61 | 1.16 | 0.93 | 0.88 | 0.82 | 0.01 | 0.123 | <.0001 | <0.001 | 0.004# |
NC, negative control; PC, positive control.
1Linear and exponential regression analysis were performed with increasing phytase dose from 0 (NC) to 2,000 FTU/kg, excluding PC, using JMP fit Y by X for linear response (where P-value is for phytase dose slope) and modeling-nonlinear - exponential growth and decay: fit exponential 3P for exponential response (= a+b * EXP (c * phytase dose)). The P-value in the table is for growth rate. When P-value is below <0.1 for one of the models, a goodness of fit test was done, and the optimal model is the prediction equation with the lowest AIC (measure of fit) and root mean square error (RMSE, measure of precision) and marked with #. An exponential model was used instead of a quadratic response to estimate the asymptote level. The asymptote (P < 0.0001) was reached at 70.3% for AID P, 80% for AID N, 84% for AID IP6, 71% for ATTD P and 0.83 g/d for total P excretion.