Laserfiche WebLink
<br />perimeter analyses. A one-way ANOVA randomized block design (blocking over <br /> <br /> <br />stage) was used to test for significant differences in WUA between flows. <br /> <br />Fisherls LSD, the MF was defined as the flow that resulted in a significane: <br /> <br />reduction in WUA (P(O.lO) from the preferred flow. <br /> <br />Comparison of Methods <br /> <br />No objective methods have been offered in the literature to distinguish bet- <br /> <br />ween results of methods that are compared in a study such as this. lnvestiga- <br /> <br />tors generally review the range of flow recommendations and select one for a, <br /> <br />usually subjec tive, reason. A "true valueH' instream flow recommendation mayor <br /> <br />may not exist for each stream or stream segment, however, we employed an objec- <br /> <br />tive approach for evaluating the relative bias of these 18 methods using a com- <br /> <br />monly accepted statistical procedure. <br /> <br />An unbiased MF range was defined for each stream as the mean and 95% con- <br /> <br />fidence interval of all recommendations for that stream. <br /> <br />The recommendation <br /> <br />from each method for each stream was then compared to this unbiased range to <br /> <br />de termine its predic tive tendency. <br /> <br />Chi square analyses were used to identify <br /> <br />r'-'~ <br /> <br />methods that were consistently within the unbiased range (P(0.05) as w~ll as to <br /> <br />evaluate the tendency of methods to overestimate or underestimate unbiased <br /> <br />ranges. <br /> <br />RESULTS AND DISCUSSION <br /> <br />Tennant <br /> <br /> <br />recommendation based on 30% AF proved to be one of the least biased <br /> <br />methods in this study. Recommendations were within the unbiased range for 10 of <br /> <br />, <br />,. <br /> <br />8 <br />