How Big of an Effect Do Small Dams Have? Using Geomorphological Footprints to Quantify Spatial Impact of Low-Head Dams and Identify Patterns of Across-Dam Variation

Abstract

Longitudinal connectivity is a fundamental characteristic of rivers that can be disrupted by natural and anthropogenic processes. Dams are significant disruptions to streams. Over 2,000,000 low-head dams (<7.6 m high) fragment United States rivers. Despite potential adverse impacts of these ubiquitous disturbances, the spatial impacts of low-head dams on geomorphology and ecology are largely untested. Progress for research and conservation is impaired by not knowing the magnitude of low-head dam impacts. Based on the geomorphic literature, we refined a methodology that allowed us to quantify the spatial extent of low-head dam impacts (herein dam footprint), assessed variation in dam footprints across low-head dams within a river network, and identified select aspects of the context of this variation. Wetted width, depth, and substrate size distributions upstream and downstream of six low-head dams within the Upper Neosho River, Kansas, United States of America were measured. Total dam footprints averaged 7.9 km (3.0-15.3 km) or 287 wetted widths (136-437 wetted widths). Estimates included both upstream (mean: 6.7 km or 243 wetted widths) and downstream footprints (mean: 1.2 km or 44 wetted widths). Altogether the six low-head dams impacted 47.3 km (about 17%) of the mainstem in the river network. Despite differences in age, size, location, and primary function, the sizes of geomorphic footprints of individual low-head dams in the Upper Neosho river network were relatively similar. The number of upstream dams and distance to upstream dams, but not dam height, affected the spatial extent of dam footprints. In summary, ubiquitous low-head dams individually and cumulatively altered lotic ecosystems. Both characteristics of individual dams and the context of neighboring dams affected low-head dam impacts within the river network. For these reasons, low-head dams require a different, more integrative, approach for research and management than the individualistic approach that has been applied to larger dams.

Fig 1. Predictions about Dam Impact.

Predictions of geomorphic effects caused by low-head dams on (A) wetted width and depth, (B) channel widening, and (C) substrate size from the Web of Science literature on low-head dams. On all prediction plots, the X axis is the distance from the dam, the black trapezoid represents dam position, the area left of the dashed line represents habitat upstream of the dam and the area right of the dashed line represents habitat downstream of the dam. The impoundment is represented by grey shading.

Fig 2. Map of the Upper Neosho River network.

Map of our study area in the Upper Neosho river network (A) located in Kansas. Also shown are (B) six dam sites and two undammed reference sites along the Upper Neosho River and Lower Cottonwood Rivers. Major U.S. Army Corps of Engineers (USACE) reservoirs in the study river network are labeled for reference.

Fig 3. Evaluation of Substrate Selection among Individuals.

Comparison of (A) mean substrate size picked by four individuals at three randomly chosen points along a transect with ten replicates for each point and (B) average D₅₀ of three Wolman pebble counts of one riffle. NS indicates no significant difference between individuals. Statistics are the result of a Kruskal-Wallis test of sampler effect using (A) a Bonferroni corrected α = 0.016 (0.05 divided by 3 (for each location)) (B) α = 0.05.

Fig 4. Wetted Width.

Longitudinal profiles of wetted width for the six dams in the study reach showing upstream (left) and downstream (right) samples for our six study dams (black trapezoid); (A) Riverwalk, (B) Correll, (C) Ruggles, (D) Emporia, (E) Cottonwood Falls, and (F) Soden. Y axis arrows indicate mean wetted width. Small arrows in panel indicate footprint based on substrate size profiles see section on substrate size for details. P-values from Wilcoxon rank sum test for mean differences between upstream and downstream transects are shown. Asterisk indicates significance with Bonferroni corrected α = 0.01 (0.05 divided by five dams).

Fig 5. Depth.

Longitudinal profiles of depth for the six dams in the study reach showing upstream (left) and downstream (right) samples for our six study dams (black trapezoid). (A) Riverwalk, (B) Correll, (C) Ruggles, (D) Emporia, (E) Cottonwood Falls, and (F) Soden. Y axis arrows indicate mean depth. Small arrows in panel indicate footprint based on substrate size profiles (see section on substrate size for details). P-values from Wilcoxon rank sum test for mean differences between upstream and downstream transects are shown. Asterisk indicates significance with Bonferroni corrected α = 0.01 (0.05 divided by five dams).

Fig 6. Substrate size.

Longitudinal profiles of substrate size for the six dams in the study reach showing upstream undammed sites as available (left) and downstream (right) samples (black trapezoid). Each point represents a riffle’s location in relation to its distance from the dam. The end of the dam footprint is indicated by an arrow, where applicable, and was considered to be where riffles reached a baseline determined by the undammed sites, or the minimum value for (D₅₀) on its river (Upper Neosho or Lower Cottonwood). (A) Riverwalk, (B) Correll, (C) Ruggles, (D) Emporia, (E) Cottonwood Falls, and (F) Soden. The longitudinal profile for the two reference sites, Undammed-1 and Undammed-2 are plotted in the upstream panel for their corresponding dam sites, Riverwalk and Correll, respectively.

Fig 7. Cumulative Distribution Curves.

Upper panels display substrate size composition changes with increasing distance from dams for (A) Riverwalk, (B) Correll, (C) Ruggles, (D) Emporia, (E) Cottonwood Falls, (F) Soden, and undammed sites (G) Undammed-1 and (H) Undammed-2. Consecutive riffles below the dam are displayed until median substrate size (D₅₀) returned to 22.5 mm for Neosho River or 32 mm for Cottonwood River. For comparison, lower panels display substrate size compositions for riffles at reference sites located away from dams where distributions remained similar. Note: reference sites with riffles were not available for four of the six dams. P-values in the figures are based on Kolmogorov-Smirnov test of the distribution curves comparing the two riffles indicated in parentheses. See panel I for the legend—1° riffle indicates the first riffle downstream of a dam, 2° riffle indicates the second riffle downstream of a dam, 3° riffle indicates the third riffle downstream of a dam.

Fig 8. Low-head Dam Footprints.

Downstream (black bars) and upstream (gray bars) footprints of six dam sites and the average downstream and upstream footprints in the Neosho River network. The dashed lines are averages.

Fig 9. Regression for Environmental Correlates of Dam Footprints.

Univariate regressions for environmental correlates of (A) downstream, (B) upstream, and (C) total dam footprints expressed in terms of multiples of wetted widths. The relationship shown corresponds to the top model amongst competing univariate regressions testing dam height, distance to nearest upstream dam, number of upstream dams, and height of nearest upstream dam. The corresponding equation and correlation (adjusted R-sq) between the dam footprint and the corresponding explanatory variable are indicated in each panel.