4 | DISCUSSION
The rapid assessment procedure used in this study provided valuable information on structural integrity and function with relatively little effort in the field. All 137 structures within the 8-km study reach were surveyed within a two-day period each year, typically by a team of 3-4 individuals. Although the rankings were qualitative, the ordinal nature of data supported an analytical approach for evaluating structure performance, including investigating differences between years and structure types. Overall, instream structures met the performance objective of 90% stable and functional at the three year mark following completion of construction. However, structural integrity and function diminished over time and the likelihood of poorer rankings increasing following a 38-year flood that occurred in 2019. This suggests that the lifespan of some structures may be dependent on the magnitude and timing of flows that occur following completion of instream construction (Roper, Konnoff, Heller, & Wieman, 1998). Engineering design, including both the prescription and specifications for treatments, also played a role in structural performance, as some structure types performed better than others. Results indicate that some structure types should not be used in similar geomorphic settings and others would benefit from a more robust design approach.
The observed failure rate for instream structures (26%) was comparable or lower when compared to similar studies, which reported a mean failure rate of 40% across studies with a range of 15-75% (Babcock, 1982; Frissell & Nawa, 1992; Schmetterling & Pierce, 1999; Miller & Kochel, 2013). Boulder-toe structures were the most likely to exhibit issues with integrity and erosion. The study site is located in a wide alluvial valley where the Arkansas River has historically migrated across the floodplain. Prior to disturbance, the stream banks were dominated by woody riparian vegetation with willows stretching across much of the valley floor. Boulder toe was prescribed in some locations with the intention of preventing erosion into fluvial tailings deposits. However, this treatment did not perform well due to the dynamic nature of the system, and typically failed when erosion occurred behind the boulders that were intended to armor the stream bank. In similar systems with meandering riffle-pool morphology and stream banks dominated by woody riparian vegetation, we recommend that bioengineering techniques (Giordanengo et al., 2016) be utilized to stabilize stream banks rather than more hardened approaches, such as boulder toe. Log toe structures, considered a bioengineering approach, performed better than boulder toe with respect to integrity and erosion, but still exhibited failure rates over 20% during the study period. However, logs will degrade with time and, unlike boulders, will not be a permanent feature in the river as it evolves. Furthermore, the addition of large wood has been shown to improve stream habitat (White et al., 2011; Jones et al., 2014), while increasing hyporheic exchange and the retention of sediment, nutrients, and organic matter (Wohl et al., 2016).
High failure rates were also observed for log-vane structures. The failure of log vanes was likely associated with inadequate engineering design. The post hoc analysis of log-vane failures indicated that the likelihood of failure increased with flatter log-vane slopes. Flatter slopes were associated with a greater angle of departure from the stream bank, which may have exposed more surface area of the log to erosive forces in the channel. Boulders were used to anchor the end of the log vanes in the channel and typically protruded above the elevation of the streambed. These boulders were prone to movement and the number and size of anchor boulders was not clearly specified in the plans. Another common failure mechanism for log vanes was the undermining of footer logs, which may have been prevented by installing erosion control fabric on the upstream face of the header and footer logs. In addition, the length of logs utilized for the log vanes was relatively short, typically less than 5 m. Longer logs would have allowed the end of the log vanes to be buried into the stream bed and then anchored with appropriately sized boulders. The shorter logs entailed pinching the end of the vanes between two or more boulders, jeopardizing the integrity of the structures if boulders moved slightly. For future projects that utilize logs vanes, we recommend that specifications err on the side of longer logs and that the structures are engineered to withstand a specific design flow (Rafferty, 2013) that corresponds with project objectives and constraints.
Cobble toe, grade control in the main channel, and boulder clusters exhibited the best performance with respect to integrity and erosion. Cobble toe was typically utilized in locations where the river was relatively straight and erosional forces were lower, particularly when compared to the outside of meanders bends where scour and bank erosion may be more likely. Grade control structures in the main channel were typically placed in straight reaches as well, either to backwater an upstream pool or provide additional stability in locations where the channel had been realigned. Although some boulder clusters were placed in pool habitats, the majority (>80%) were placed in straight, riffle sections to increase channel complexity. The placement of these structure types in straight, less erosive, reaches may have contributed to the lower integrity and erosion rankings. At high flows, pools may scour with the coarser material mobilized from the pools being deposited over the riffles that form in the straight sections between meander bends (Lisle, 1979). The sediment transport processes in meandering rivers with riffle-pool morphology make pools more prone to erosion and riffles more prone to deposition, which may explain some of the observations from this study. For instance, log vanes that were placed in riffles were more likely to receive higher rankings for deposition than log vanes that were placed in pools. Deposition was also more likely in years with lower magnitude floods (i.e., 2017, 2018, and 2020), suggesting that depositional processes may have been more prevalent in these years, while scour and erosion would be more prevalent in years with larger magnitude floods (Backcock, 1982; Frissell & Nawa, 1992). Issues with erosion resulted in more problems with structural integrity, whereas deposition may have affected the function of structures but did not typically lead to failures. Sediment deposition may also have adverse impacts on fish and benthic macroinvertebrates (Henley, Patterson, Neves, & Lemly, 2000).
The change in RPD also varied between locations that were classified as riffle or pools prior to instream construction. Pools that were developed in riffle locations resulted in a greater initial change in RPD when compared to preexisting pools because the preexisting RPD was 0 m for all riffles. In addition, riffle locations are typically shallower than pools (Lisle, 1979), so it is not surprising that pool development in riffle locations resulted in larger RPD changes than the enhancement of existing pools. Furthermore, it is not surprising that the pools developed in riffle locations filled in to some degree following the first runoff cycle, whereas existing pools exhibiting less change in RPD between the as-built condition and first runoff cycle. The main effect of structure type did not have a significant effect on the change in RPD for either riffle or pools habitats, suggesting there were no differences among structure types regardless of the preexisting morphology. Years post construction was a significant effect for both the riffle and pool models, but this effect differed between morphologies. For riffles, the change in RPD associated with the as-built condition was greater than all other years, indicating that construction activities increased RPD and initially improved overwinter habitat for trout. However, pools that were developed in preexisting riffle locations exhibited a significant decline in RPD following the first runoff cycle. Despite this change after the first runoff cycle, the cumulative change in RPD within preexisting riffles was positive over the study period, increasing by 0.26 m. These results suggest that over excavating pools can still result in improved RPD even if those pools fill in to some degree.
The change in RPD at preexisting pools differed from the observations at riffle locations, with a smaller net increase of 0.13 m over the study period. Although years post construction was a significant main effect, the interpretation of that effect was confounded by the significant interaction between years post construction and structure type. Three structure types or combinations exhibited significant difference for the as-built condition at preexisting pool locations. The changes in RPD at log toe structures and log vane/log toe combination were greater than the changes at log vanes. Log vanes were either placed at the head of a pool, within riffle habitats to create pocket water and increase habitat complexity, or adjacent to fluvial tailings deposits to decrease near bank shear stress and reduce erosion risk. When log vanes were constructed in riffle locations, a relatively small pool was excavated downstream of the vane. These pools were typically shallower than the pools at meander bends where log toe and log toe/log vane combination treatments were often utilized, which may explain the difference in response between these structure types for the as-built condition. The boulder vane/boulder toe combination also differed from other structure types in year 1, exhibiting greater pool scour when compared six other structure types or combinations. The combination of a boulder vane and boulder toe treatment likely concentrated energy over the end of the vane while also armoring the stream bank, directing energy to the stream bed and resulting in greater scour. The hardened approach associated with the boulder vane/boulder toe combination resulted in a different RPD response, where scour appeared to be greater than other structure types in years 1 and 2, while deposition was greater in year 3.
Although the approach utilized in this study provided useful information regarding project objectives associated with the integrity and function of instream structures and their effect on pool development, the study was not without limitations. The rapid assessment procedure could be developed further by incorporated a qualitative habitat evaluation, which would provide valuable information for project managers when considering the need for maintenance at specific structures. As the project design was developed independently from the monitoring design, there was imbalance in sample sizes for some structure types, which may have affected the results from statistical analysis. The logistics of building such a large stream restoration project also entailed implementing the project over two construction seasons, which precluded the inclusion of discharge as a covariate and led us to utilize years post construction as a factor instead. The magnitude and duration of annual flood events surely affected the amount of sediment transport that occurred, which in turn affected the severity of erosion, deposition, and pool scour that occurred. Furthermore, the qualitative nature of the rapid assessment rankings introduced some subjectivity to the evaluation, but that was mitigated to some degree by having the same observer present for all surveys. Despite these limitations, the study indicated that structures met the performance objective for integrity and function and there was a positive increase in RPD and associated overwinter habitat. Our results suggest that the lifespan of some instream structure may be limited to a decade or less, as structural integrity and function declined over time. However, the longevity of structures may not be the most important consideration if they improved channel resilience by supporting the establishment and growth of woody riparian vegetation, which occurred within a few years of project completion (Cubley et al., 2022). Although the inference from this case study is somewhat limited, we hope that our results can inform any future maintenance needs for this project as well as structure selection and design for future stream restoration projects located in similar geomorphic settings.