Laserfiche WebLink
I/ <br />Research Design <br />Sam lin <br />The upper Colorado system is made up of a wide 'variety of habitats from <br />sluggish backwaters-and shallow sandy runs to series of waterfalls and <br />rapids, with extensive reaches inaccessible except by air or by floating <br />down the river considerable distances. <br />Since this study area encompasses some 570 miles of the Colorado, Gunnison <br />and Green Rivers, it was impossible to study the entire length with the <br />allocated manpower; therefore, it was necessary to rely on a statistically <br />sound unbiased sampling scheme from which to infer habitat/fishery life <br />history relationships. <br />The river system consists of several 30 to 70 mile sections of relatively <br />homogeneous fishery habitats, e.g., high-gradient canyons,-low-gradient <br />braided winding `sections and transition zones between these two main <br />types. Our sampling scheme identified these homogeneous sections and <br />handled them as discrete strata of a random sampling scheme. This <br />system provides a good distribution of samples and tends to reduce <br />within-strata sampling variability (Table 3). Criteria used to delineate <br />strata included both geophysical (river channel configuration,.substrate <br />types, gradient and depth) and biological parameters (productive bottom <br />lands, sterile sand runs, productive backwaters, swift canyons and point <br />sources of allochthonous materials). <br />Because we could work sequentially downstream with less investment of <br />time, we did not use a completely random design Ito select the sequence