An experimental study: effects of boulder placement on hydraulic metrics of instream habitat complexity

Effects of grid spacing on habitat hydraulic complexity metrics

The sensitivity of the habitat hydraulic complexity metrics to Δs was examined by calculating the metrics for Δs = 0.06, 0.12, 0.18, and 0.24 m (for M₄, Δs = Δx = Δy). Figure 3 shows the variation of the metrics with grid spacing for scenarios with boulders. A preliminary assessment of no-boulder scenarios (S1-L and S1-H) showed that all the metrics decreased by increasing the grid spacing. However, because the metrics are mostly used in complex rather than non-obstructed and 1-D flows, the plots only include scenarios with boulder placement to highlight the effects of grid spacing on the metrics in complex flows. All the metrics generally decreased as Δs increased. At the low flow rate, by changing the Δs from the smallest to largest, i.e., 0.06 m to 0.024, the mean decreases in the M₁, M₂, and M₄ metrics (averaged over all the scenarios with boulders) were 45.1, 9.9, and 74.7%, respectively. At the high flow rate, these reductions were 34.8, 14.7, and 82.5% for M₁, M₂, and M₄, respectively. Table 2 shows the p-values associated with the changes in the metrics due to increasing Δs from 0.06 to 0.24 m for all scenarios. The table indicates that changes in M₁ and M₄ were statistically significant while for M₂ they were not (p-values > 0.05 for all scenarios except for S2-H). This result indicated the considerable influence of grid spacing on M₁ and M₄ metrics in the reaches with boulder placement. Additionally, the differences in the reported average reductions due to changing the flow rate were less than 10%, indicating an insubstantial effect of flow rate on the habitat hydraulic complexity metrics’ sensitivity to the grid spacing. The significant sensitivity of the metrics M₁ and M₄ to the grid spacing in this study is contrary to the findings of a previous study in which an insignificant correlation was found between the habitat hydraulic complexity metrics and Δs²⁹. This difference can be attributed to different topographic features in the studied reaches. In the previous findings, measurements were mainly taken around the bends and reaches with no significant obstruction²⁹, in which a more uniform flow with smaller velocity gradients is expected. However, in this study, the systematic boulder placement generated more complex flow patterns with noticeable velocity gradients. Therefore, due to the variations of flow velocities in the zone studied, substantially different values for the metrics are anticipated by computing the metrics over different spatial scales.

Statistical distribution of habitat hydraulic complexity metrics

Table 3 lists the mean, minimum, maximum, and standard deviations of the habitat hydraulic complexity metrics (Δs = 0.06 m) for all the scenarios. To complement the results from Table 3 and assess whether the influences of solely changing the boulder concentration or flow rate on the metrics were statistically significant, Table 4 shows p-values associated with changing flow rate from low to high for a given boulder concentration, and Table 5 shows p-values associated with gradually increasing the boulder concentration for a given flow rate.

For metric M₁, the mean M₁ values for scenarios incorporating boulders showed the same order of magnitude as values from previous studies for reaches with single and multiple boulders²⁷ but they were about one order of magnitude larger than calculated values in the confluence of two rivers¹¹. Using a larger grid spacing in the study in the confluence of two rivers¹¹ can be the reason for this difference. For a scenario at the higher flow rate, the mean M₁ was on average (averaged for all the scenarios) 36% greater than its counterpart at the lower flow rate and this change in M₁ values was statistically significant with p < 0.05 (Table 4). Adding boulders and subsequently increasing the boulder concentration resulted in a slight increase in the mean M₁ values at both flow rates. This increase was statistically significant (p < 0.05) except for changes from S2-L to S3-L, S1-H to S2-H, and S3-H to S4-H with p = 0.105, 0.135, and 0.065, respectively (Table 5).

The minimum values of M₁ did not show a specific trend with boulder concentration and flow rate. For a given boulder concentration, the maximum values of M₁ were generally larger at the higher flow rate except for the lowest boulder concentration. Comparing the standard deviations revealed that for a given boulder concentration increasing the flow rate increased the variability of M₁ as well. Furthermore, the standard deviations showed that for a given flow rate, increasing boulder concentration generally increased the variability of M₁ values. The only exception was the change from the medium to the highest boulder concentration at the higher flow rate (S3-H to S4-H) where the standard deviation slightly decreased. For the lower and higher flow rates, the largest variability occurred at the highest boulder concentration (S4-L) and the medium boulder concentration (S3-H), respectively.

For metric M₂, the mean M₂ values for scenarios with boulders were of the same order of magnitude as those observed for a series of complex habitats, some including boulders^9,29. According to another study, after the construction of several reefs in the river, the mean M₂ increased due to variation in the local flow²⁵. However, even after the reef construction, the calculated mean M₂ values in their study²⁵ were about two orders of magnitude smaller than the values estimated in this study. For a given boulder concentration, increasing the flow rate resulted in an average drop of 21% in the mean M₂ values. This decrease was only significant (Table 4) for the lowest boulder concentration (S2-L to S2-H). Similar to M₁, for a given flow rate, the mean M₂ values gradually increased by increasing the boulder concentration. This increase was statistically significant for all scenarios except for the change from S2-L to S3-L (p = 0.265; Table 5).

Adding boulders (i.e., change from S1-L to S2-L as well as S1-H to S2-H) generally increased the minimum observed M₂ values but similar to M₁, minimum M₂ values did not show a specific trend with boulder concentration and flow rate. The maximum M₂ values were one to two orders of magnitude higher than the mean M₂ values, indicating the presence of isolated large M₂ values in the studied zone. They showed a similar order of magnitude with the calculated maximum M₂ values in reaches with boulders in the previous findings²⁷. The extremely large M₂ values can be attributed to small values in the denominator of the M₂ (Eq. 2) due to the significantly lower velocities in the wake of boulders²⁹. This also can be associated with the drop in the mean M₂ values after increasing the flow rate for a given boulder concentration. By increasing the flow rate, the low velocities in the wake of the boulders increased resulting in larger values in the denominator of the M₂ equation and subsequently smaller M₂ values. As expected, due to varying local flows, the variabilities of M₂ values for scenarios with boulders were significantly higher. The largest standard deviations were for scenarios S2-L and S4-L, which were the same scenarios with the highest observed maximum M₂ values.

For scenarios with boulders, the mean M₄ values were one order of magnitude higher than the scenarios without boulders. The mean M₄ values for scenarios with boulders showed the same order of magnitude as reported values in areas with vortices¹⁴, around boulders¹⁵, and in complex habitats⁹. However, the obtained values within the bends²⁹ were about an order of magnitude smaller than the mean M₄ values in this study. After adding boulders to the flume, a significant elevation in the mean M₄ can be observed, specifically at the lower flow rate. After adding boulders, the mean M₄ increased 438% and 320% for the lower and higher flow rates, respectively.

The changes in M₄ values due to increasing the flow rate for a given boulder concentration did not follow a specific trend and were statistically insignificant (Table 4). Regardless of the flow rate, by gradually increasing the boulder concentration, the mean M₄ also increased. These increases were statistically significant (Table 5) except for the change from S3-L to S4-L (p = 0.174; Table 5). The increase due to increasing boulder concentration was more noticeable at the higher flow rate compared with the lower flow rate. For instance, at the higher flow rate, by changing the boulder concentration from the lowest (S2-H) to the highest (S4-H), the increase in mean M₄ was 95% while at the lower flow rate the same change in the boulder concentration (S2-L to S4-L) resulted in only 31% increase in mean M₄.

Spatial variations of habitat hydraulic complexity metrics

To examine the spatial distribution of the habitat hydraulic complexity metrics in a reach with boulders, contour maps of the metrics M₁ and M₂ in the detailed measurement zone were generated. The metrics shown by contour maps were computed by Δs = 0.06 m because the preliminary evaluation of the results showed that changing the grid spacing did not noticeably change the spatial distribution of the metrics as also reported by another study¹¹. Figure 4 shows the contour maps of M₁ for scenarios with boulders. High-M₁ regions generally appeared downstream and at the sides of the boulders. At the higher flow rate, the high-M₁ regions were more extended in comparison with the lower flow rate. For instance, while the high-M₁ regions at the lower flow rate for scenarios S2-L, S3-L, and S4-L covered about 1D, 1D, and 2D downstream of the boulders, respectively, at the higher flow rate these regions extended to approximately 1.5D, 2.5D, and 3D downstream of the boulders. This can be attributed to the generally higher average velocities between two points (U_avg) at the higher flow rate. At the lower flow rate, by increasing the boulder concentration from 3.4 to 5.4%, no substantial change in the spatial pattern of M₁ values occurred; however, by further increasing the boulder concentration to 8.3%, high-M₁ regions extended more noticeably. At the higher flow rate, the spatial patterns of M₁ for the lowest boulder concentration (S2-H) differed from those patterns at the higher boulder concentrations (S3-H and S4-H) where high-M₁ regions appeared more frequently over a larger area. These are compatible with the results from Table 3, in which the scenarios with the higher boulder concentration generally resulted in the larger variability (standard deviation) of M₁ values in the detailed measurement zone.

As shown in Fig. 5, similarly, high-M₂ regions can be seen in the vicinity of the boulders, specifically in the wake of the boulders. For the smallest boulder concentration, λ = 3.4%, increasing the flow rate (i.e., change from S2-L to S2-H) reduced the extent of the high-M₂ region in the boulder’s wake from 2.5D to 1.5D. For the medium boulder concentration, λ = 5.4%, by increasing the flow rate a high-M₂ region appeared downstream of the leftmost boulder in the detailed measurement zone, and for the highest boulder concentration, λ = 8.3%, no specific change in the spatial patterns of M₄ was observed as a result of increasing the flow rate. At both flow rates, increasing the boulder concentration generally resulted in the appearance of more high-M₂ regions. The extremely large values of M₂ occurred immediately downstream of the boulders, where extremely low (near-zero) velocities were observed in the recirculation zone of the boulders resulting in very small values in the M₂ equation, as discussed in the previous section.

Figure 6 shows the contour map of ({{M}}_{{4, com}}) in detailed measurement zone computed for cells with Δx = Δy = 0.06 m. It provides information about the local modified recirculation and the regions that highly contribute to the computed M₄ for the entire reach. Generally, zones with higher ({{M}}_{{4, com}}) were in the wake regions and at the sides of the boulders. For the scenario with λ = 3.4%, by increasing the flow rate, the extent of cells with higher ({{M}}_{{4, com}}) in the wake of the boulders slightly decreased from about 2D to 1.5D. For λ = 5.4%, by increasing the flow rate, the extent of the region with higher local modified recirculation downstream of the leftmost boulder was reduced from 2.5D to 1.5D; however, a region with high ({{M}}_{{4, com}}) appeared at the right side of that boulder. For λ = 8.3%, the number of cells with higher modified recirculation generally increased in-between boulders at both flow rates. Generally, by increasing the boulder concentration at both flow rates, the number of cells with higher ({{M}}_{{4, com}}) increased. These findings corroborate that increasing the boulder concentration results in higher M₄ values in the detailed measurement zone as seen in Table 3.

Previous findings also indicated increased metrics downstream, at sides, and in-between randomly placed boulders with different sizes and submergence ratios^14,15,27. However, to the best of the authors’ knowledge, the spatial variation in habitat hydraulic complexity metrics due to systematic boulder placement has not been assessed in previous studies.

Although flows in two experimental scenarios (S3-H and S4-H) were classified in different flow regimes (Regime 2) than the other scenarios (Regime 3), no significant influence of submergence ratio on the spatial distribution of metrics was identified. This might be due to the fact that the submergence ratios of the experimental scenarios were relatively close and the influence of this factor on the flow field remained insignificant even though the flows in S3-H and S4-H were technically classified as a different flow regime.

Estimation of habitat hydraulic complexity metrics

To further investigate the dependence between the derived dimensionless terms, the relationships between the dimensionless metrics and other dimensionless parameters were examined. Figure 7 shows the variation of the dimensionless habitat hydraulic complexity metrics M₁H/({{U}}_{{reach}})², M₂H, and M₄H/({{U}}_{{reach}}) with the other dimensionless terms, Q/({{U}}_{{reach}}) H², H/Δs, λ, and Fr. The coefficient of determination, R², was found based on the best available fit between the parameters. It should be mentioned that the dimensionless metrics were computed for Δs = 0.06, 0.12, 0.18, and 0.24 m (for M₄, Δs = Δx = Δy). A negative correlation between the dimensionless metrics and Q/({{U}}_{{reach}}) H² can be seen; however, R² values for M₂H, and M₄H/({{U}}_{{reach}}) were 0.55 and 0.28, respectively, indicating a weak to moderate correlation⁵⁸. By increasing H/Δs, the dimensionless metrics also increased. The correlation coefficient for M₁H/({{U}}_{{reach}})², and M₂H were 0.35 and 0.25, respectively, which indicated weaker correlations with H/Δs. As λ increased, the dimensionless metrics also increased, and the correlation coefficients for M₁H/({{U}}_{{reach}})², M₂H, and M₄H/({{U}}_{{reach}}) were 0.79, 0.92, and 0.52, which indicated moderate to strong correlations between the dimensionless metrics and the boulder concentration. A negative correlation between the dimensionless metrics and Fr can be seen. The R² values were 0.61, 0.80, and 0.48 for M₁H/({{U}}_{{reach}})², M₂H, and M₄H/({{U}}_{{reach}}), respectively, showing a moderate to a strong relationship between the parameters.

Although in some cases, especially for M₂H, and M₄H/({{U}}_{{reach}}), weak to moderate correlations were observed; the derived dimensionless groups from the dimensional analysis can be used for estimating the habitat hydraulic complexity metrics. A multiple regression analysis was performed and the following expressions were found:

By comparing the predicted values from Eqs. (9) to (11) and the measured values from the experiments (Fig. 8), the R² values of M₁H/({{U}}_{{reach}})², M₂H, and M₄H/({{U}}_{{reach}}) for a 95% confidence level were 0.97, 0.91, and 0.88, respectively. These strong correlations showed acceptable performance of the proposed equations to predict the habitat hydraulic complexity metrics. Using Eqs. (9) to (11), the average habitat hydraulic complexity metrics for a reach with boulders can be estimated by only obtaining information about the reach-averaged depth and velocity, flow rate, boulder concentration (in rock-ramp arrangement), and the desired grid spacing.

However, the proposed equations were only based on the limited tested dataset and parameters and are only applicable to the investigated range of parameters in this study. It should be emphasized that in this study boulder concentration was calculated for boulders placed in a rock-ramp arrangement. One may obtain similar boulder concentrations for completely different arrangements such as boulder clusters, random (non-uniform) boulder placement, or even for an isolated boulder that covers a large area; however, the applicability of these relationships should be limited to rock-ramp arrangements due to significant changes in the local flow fields for the other configurations⁵⁹. There are multiple other potentially influential factors, including but not limited to embedded depth, bed slope, and substrate composition that should be taken into account for designing a boulder placement scenario. Further numerical and experimental studies that incorporate these factors as well as assess the parameters investigated (e.g., boulder concentration, flow rate, etc.) over a broader range are needed to improve the accuracy and applicability of these equations in the field and under more complex conditions.

Potential implications of habitat hydraulic complexity metrics for instream species

The habitat hydraulic complexity metrics may provide important information about the instream habitat quality and availability. Possible influences of habitat hydraulic complexity metrics, calculated over the default grid size (0.06 m) for the flume model (1:1 scale), on the instream habitat, as well as potential effects of increasing the grid size to the largest in this study (0.24 m) need further elaboration. As pointed out earlier, the spatial scale on which the metrics are calculated can noticeably influence the metric values. This makes comparing the metrics across the studies challenging as a variety of spatial scales have been used in different studies and sometimes scales are not incorporated. In the following comparisons, the spatial scales over which the metrics were computed were mostly larger or unreported and also were not necessarily based on an ecologically relevant scale. Even with these limitations, interpreting the metrics in this work in relation to habitat quality and availability for instream species may be helpful and provide insightful information for future work and projects.

Regions with noticeable longitudinal velocity and energy gradients as well as recirculation can be used to delineate suitable habitats such as spawning grounds in a reach^11,15,27. Previous findings indicated a partial relationship between M₁ values and spawn density per unit area of Chinese Sturgeons (Acipenser sinensis)³⁰. It was found that the spawns were mainly observed in areas where M₁ > 0.029 J kg⁻¹ m⁻¹. The M₁ values in the investigated scenarios of this study exceeded 0.029 J kg⁻¹ m⁻¹ except for a small number of points between the boulders; however, as the boulder concentration and flow rate increased the frequency of regions with low M₁ values reduced. Earlier, it was mentioned that increasing the grid size from 0.06 to 0.24 m resulted in an average reduction of M₁ values up to 45.1%. Even by applying this reduction, most M₁ values remained above 0.029 J kg⁻¹.

Higher M₂ values can be used to identify locations with noticeable biological richness and ideal feeding habitats^9,28. It was predicted that M₂ values in the range of 4–14 m⁻¹ can be distinguished as a suitable location for brown trout feeding¹⁴. In this study, before placing boulders, the observed M₂ values were mainly below this range but after adding boulders the regions with M₂ values in the range of 4–14 m⁻¹ appeared. These regions were located mainly in the middle of wake regions, i.e., about 1.5D downstream of the boulders. Increasing the boulder concentration, generally expanded regions with M₂ values in this specific range. The extremely large M₂ values in the near-wake region did not fall in this range; however, they should not be necessarily assumed as unsuitable regions because the significantly lower velocities in these regions can provide resting zones and refuge for many species. By increasing the grid size to 0.24 m and applying a 14.7% drop in M₂ values, which was mentioned earlier as the average reduction due to increasing the grid size to the largest, no significant change in the spatial pattern of M₂ values was observed and they mostly remained in 4–14 m⁻¹ range as before. Small-scale regions with high M₁ and M₂ values which are adjacent to higher velocity zones may provide a suitable zone for place-specific activities, in which fish such as juvenile salmon and steelhead can forage while minimizing the bioenergetics cost of swimming^19,26. Boulder placement, especially with higher densities, resulted in the more frequent appearance of such areas in this study.

For a complex region between boulders with several brown trout redds, it was reported that M₄ values were greater by two orders of magnitude from a nearby region with a homogenous flow and without any redds¹⁵. In this study, after boulder placement, the M₄ values increased by only one order of magnitude, and scenarios with the highest boulder concentration, S4-L, and S4-H, resulted in the largest M₄ values, which indicated more flow complexity. It was indicated that M₄ = 0.5 s⁻¹ might be an upper threshold for brook trout (Salvelinus fontinalis) habitat⁹. In this study, after boulder placement, M₄ values for all the arrangements exceeded 0.5 s⁻¹. However, unlike the M₄ values for the default grid size (0.06 m), for a larger grid spacing of 0.24 m, the M₄ values mainly remained under 0.5 s⁻¹.

It should be noted that habitat availability and selection may not be attributed to only the spatial flow patterns and subsequently the habitat hydraulic complexity metrics. There are several other factors such as cover, temperature, substrate composition, food availability, etc., that may substantially affect the instream habitat selection; however, considering the effects of all parameters together is difficult as they vary significantly in different sites⁷. For instance, substrate composition has a substantial influence on instream preferred habitat^35,60,61. In this work, the bed was not movable which resulted in simplifying the in-situ conditions as different local hydraulics and substrate compositions are expected in a mobile bed around large roughness elements. In addition, other works have evaluated the habitat hydraulic complexity metrics over a variety of bed material sizes. These dissimilarities reduce the reliability of comparisons across studies and, again, highlight the need to consider a wider range of parameters with both hydraulic and structural complexities in relation to instream habitat for future work. More field data or experimental data with presence of live fish are needed to establish strong correlations between habitat hydraulic complexity metrics and influential factors for instream habitat assessment such as fish density, and availability of spawning grounds or feeding zones. Additionally, for a more accurate and useful comparison, based on the target species and their life stage as well as studied reach features, the effects of grid spacing should be considered and reported, if feasible.

Source: Ecology - nature.com