net structiues had decreased flow velocities relative to control plots. Average flow velocities were highest in 

 between weir structures (132 mm s '), were lower in control plots (102 mm s '), and lowest in between net 

 structures (75 mm s '). There was no difference in flow treatments between the two replicate plots. 



Turbidity did not vary significandy among flow treatments, indicating flow manipulations did not alter the 

 concentration of suspended sediment. Turbidity differences were found among dates (p = 0.0001). No 

 interaction effects were detected. Turbidity samples for 23 April were not included in ANOVA because 

 sediment was resuspended during sample collection. 



Treatment Effects 



Macroinfauna community structure response over all sampling dates was clear, because there were no 

 significant treatment interactions for diversity (Nl), and evenness (El). There was no significant differences 

 among flow treatments for diversity, and evenness. There were significant differences for disturbance 

 frequency levels for diversity (p = 0.0001), and evenness (p = 0.005). Significant differences (0.0001) among 

 dates were detected for diversity and evenness. Diversity was highest for biweekly samples (1.82) and lowest 

 for undisturbed samples (1.31); both of which were significandy different from monthly (159) and bimonthly 

 (1.62) samples, which were the same fTable 1, Tukey test). Biweekly (0.59) samples had the highest average 

 evenness, which was different from monthly (0.44), bimonthly (0.37), and undisturbed levels (0.28) (Table 1, 

 Tukey test). Evenness in monthly and bimonthly samples were the same, and bimonthly and undisturbed 

 sample means were the same (Tukey test). 



Macroinfauna standing stock (i.e., abundance and biomass) response to the experimental treatments was 

 analyzed three ways: by all sample dates, by initiation dates, and by ending dates (Table 2). Regardless of the 

 design or analysis used, there were many significant interactions obscuring interpretation of main treatment 

 effects ^able 1). Total biomass and abundance did not have si milar results in regard to which interactions 

 existed. The two main experimental treatments (flow and disturbance frequency) contributed very litde 

 variance components regardless of the analysis technique, ranging from % to 12.7 %. 



In general, the full-rank analysis for two ending dates yielded similar results to the incomplete factorial design 

 for aU sampling dates for both abundance and biomass (Table 1). The difference between the contribution of 

 the date effect is particularly striking. Sampling date contributed the largest percentage of variance for 

 abundance and biomass for the incomplete factorial analysis and the fiill-rank analysis of ending dates, ranging 

 from 35.4 % - 86.9 %. In contrast, date contributed no variance to the full-rank analysis of initiating dates, and 

 nearly all the variance was contributed by the frequency*date interaction for abundance (91.9 %) and biomass 

 (57.7 %). The similarity between analyses of p-values and variance components for aU dates and ending dates 

 indicates macrobenthic response is similar among all treatments on given dates. S imil arly, sigmficant 

 frequency*date interactions found in the initial date analysis indicates that commumty change is different for 

 each experimental treatment due to samples being taken on different dates. 



Regardless of the analysis method, the flow*frequency interaction was always significant (Table 2). The nature 

 of the interaction is different responses to flow variation in frequent disturbances (at biweekly and monthly 

 scales) compared to less frequent disturbances at longer times scales ( bimonthly and undisturbed) (Fig. 6). 

 Abundance was very similar in flow treatments during short-term disturbance, but the increased flow treatment 

 had much higher abundances in the samples that had recovered the longest or were undisturbed (Fig. 6a). The 

 trend for biomass was similar, except that biomass was higher in the decreased flow treatment than in the 

 control or increased flow treatment over the short-term disturbances (Fig. 6b). 



Simple Main Effects 



Analysis of the full rank model yielded many significant interaction effects (Table 2), so simple main effects 

 models were run (Tables 3 and 4). Flow treatment effects were determined with separate analyses by each 

 disturbance frequency level (Table 3) and frequency treatment effects were determined with separate analyses 

 by each flow level (Table 4). 



Flow Effects 



Appendix F ♦ F-5 



