Sensitivity analyses were conducted using the regression models to test the sensitivity of the treatment method at 21 ± 2 °C. First, the inoculum was varied ±30% in the regression model (28.0-52.0 mg of P. ostreatus/g of stems) at constant moisture. Second, the moisture was separately varied ±30% (1.12-2.08 g of H2O/g of stems) at constant inoculum. The upper and lower bounds chosen represent very large variations in both inoculum and moisture. The predictions are plotted vs time in Fig. 6 for xylan deg­radation. The xylan degradation ranges at 12 wk for varied inoculum and moisture were 34.7-39.3 and 32.5-41.6% degraded, respectively. Similarly, the glucan degradation ranges at 12 wk for varied inoculum and moisture were 29.2-32.8 and 27.7-34.4% degraded, respectively (not shown). When varying only one parameter, the final compositions are predicted to be relatively insensitive to inoculum size, with the largest deviation of at most ±5% degradation at 12 wk. For moisture the system was predicted to be more sensitive, but it was less sensitive at shorter degradation times.

This indicates that initial moisture is the more critical parameter to control, and also that the system is less sensitive to initial moisture at shorter treat­ment times. Shorter treatment times could be used without compromising final compositions by increasing initial inoculum size, depending on final costs.

Additional sensitivity analyses were conducted by simultaneously varying both inoculum amount and moisture content (±30% for each). This analysis predicted maximum ranges at 6 and 12 wk of 24.5-31.9 and 30.1­43.8% xylan degraded, and 19.4-24.7 and 25.9-36.1% glucan degraded, respectively. This corresponds to ЛХ/AG ranges of 1.27-1.29 and 1.16-1.21 at 6 and 12 wk, respectively, and indicates that the expected AX/AG does not vary as widely as would be suggested by combining the individual sensitivity analyses. In addition, shorter treatment times were again fa­vored for minimum variation in the treated straw stem compositions.

In terms of selectivity and reduced sensitivity to initial moisture and inoculum, the results indicate that shorter treatment times are preferred, especially if moisture is either not controlled or poorly controlled. The regression models were next used to generate topographic plots of AX, AG, and AX/AG at the various combinations of inoculum and moisture.

The results are plotted in Figs. 7-9 for the regression model predictions after 6 wk of treatment. For locations of the parameter combinations used in this study on these plots, refer to Fig. 1 (which uses the same axes). Note that the parameter combinations (21.0, 0.77), (34.0, 0.90), and (41.0, 1.20) were not included in the statistical analyses because of poor distribution of the fungal inoculum onto the straw stems. The topographic plot of percentage xylan degraded after 6 wk of treatment is shown in Fig. 7. The diamond represents the conditions chosen for preparation of treated stems for the extrusion testing. The region of only 15-20% xylan degrada­tion roughly corresponds to the region in which the inoculated P. ostreatus was observed to be unable to outcompete the indigenous microbes, since about 15% xylan degradation was observed to occur without inoculum. Increased xylan removal is predicted as both moisture and inoculum increase. There are, however, wide ranges of parameter combinations that will give the same amount of xylan degradation, indicating a fairly insensitive system in terms of overall xylan degradation after 6 wk of treatment. The curvature of the dividing curve between 25-30 and 30­35% xylan degradation, and above 100 mg of P. ostreatus/g of stems, seems odd in that it curves back toward the inoculum axis. However, this was experimentally observed by comparing the results of the (149, 1.67) and (105, 2.24) parameter combinations (see Fig. 1). Since these experi­ments were independently replicated, the behavior appears to be real.

The topographic plot of percentage glucan degraded after 6 wk of treatment is shown in Fig. 8. Again, the diamond represents the condi­tions chosen for preparation of treated stems for the extrusion testing. The region of only 10-15% glucan degradation closely corresponds to the experimentally observed region in which the inoculated P. ostreatus was unable to outcompete the indigenous microbes. Increased glucan removal is predicted as both moisture and inoculum increase. As shown for xylan, there are again wide ranges of parameter combinations that give the same amount of glucan degradation, indicating that the system is also fairly insensitive in terms of overall glucan degradation after 6 wk of treatment.

Finally, the topographic plot of AX/AG after 6 wk of treatment is shown in Fig. 9. The diamond again represents the conditions chosen for preparation of treated stems for the extrusion testing. The region of AX/AG of 1.20-1.25 encompasses the region in which it was experimen­tally observed that AX/AG was about 1.0. A ratio of AX/AG of 1.0 indi­cates nonselective polysaccharide degradation and was taken as an indication of poor competition of the inoculated fungus with the indig­enous microbes. This reinforces the observations that at low moisture and inoculum, the regression model predictions are less accurate. Figure 9 also shows that AX/AG of 1.25-1.30 is predicted after 6 wk of treatment over a very large percentage of the possible moisture and inoculum com­binations. Thus, the system is very stable with respect to selectivity of polysaccharide degradation within the parameter ranges tested.