Laserfiche WebLink
<br /> <br /> <br /> <br />1 <br />1 <br /> <br />1 <br /> <br /> <br /> <br /> <br />11 <br /> <br />1 <br />1 <br /> <br />1 <br />the recovery of all endangered fishes. A major concern relative to continuation of the RIP <br />floodplain enhancement program was the outcome of hypothesis 114. If levee removal and <br />floodplain enhancement resulted in an increase in predators in the main channel that negatively <br />impacted native fishes, it would be detrimental to recovery of species such as Colorado <br />squawfish. Therefore, the flow chart in Figure 1.1 was generated to define possible outcomes <br />and serve as a basis for decisions on the benefits of floodplain enhancement. If no nonnative fish <br />response to floodplain inundation is detected then floodplain enhancement should proceed. In <br />addition, if nonnative fishes do increase, but native fish also increase or stay the same, then <br />floodplain enhancement can continue without harmful effects on endangered fishes. If nonnative <br />fish abundance increases and native fish decrease, the mechanisms affected need to be delineated <br />and engineering solutions or nonnative control recommendations need to be implemented to <br />compensate for potential nonnative fish impacts resulting from floodplain enhancement. <br />STATISTICAL DESIGN AND DATA ANALYSIS <br />Data will be dealt with in a series of a priori comparisons. First, linear combinations of <br />variables will be produced from those collected. These will ultimately become the response <br />variables using canonical correlation analyses or factor analyses, depending on the linearity of <br />the system. Water quality and macroinvertebrates will be put into 2-3 linear combinations to <br />reflect productivity. Nonnative fish responses will also be placed into 1-3 components <br />depending on the densities, the number of species collected and the range of age classes. Native <br />fishes will probably be left in their raw form since the numbers collected will be small (at least at <br />first). Once components have been generated, a series of binary comparisons will be made. <br />First, productivity changes observed in manipulated versus control areas will be compared using <br />repeated measures ANOVA's. Productivity will probably be represented by at least four axes, <br />including primary and secondary (invertebrates) productivity and native and nonnative fish <br />density or biomass estimates. Because samples will be collected weekly for 4-6 weeks, <br />short-term time series will be able to be generated. Because the sampling protocol calls for at <br />least three months of monthly sampling following the weekly sampling, we will have about five <br />months of 'monthly' data to put into a longer time-series (again, using repeated measures). <br />Because we are concerned with the effects of nonnative fishes on natives, <br />analysis-of-covariance will be used to test whether the repeated measures of native fish changes <br />were affected by sites with high versus low nonnative responses. This will be done either using <br />nonnatives as categories or as continuous covariates depending on the level and kind of response <br />we observe. Vegetation type will be used in a similar fashion since we are interested in whether <br />areas dominated by tamarisk result in different utilization by the native fish community as <br />compared to areas dominated by cottonwood and willow. Finally, to answer the overall question <br />of how floodplain enhancement affects riverine processes, both in terms of fish community <br />dynamics and productivity, we will use repeated measures ANOVA's (at both time scales) to <br />compare sampling sites with inundation to those without (controls for this year's sampling <br />effort).