RIVER RESEARCH AND APPLICATIONS River Res. Applic. (2014) Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/rra.2843 RAPID RESPONSE OF A SAND-DOMINATED RIVER TO INSTALLATION AND REMOVAL OF A TEMPORARY RUN-OF-THE-RIVER DAM K. H. COSTIGANa,b*,†, C. M. RUFFINGa, J. S. PERKINc,d AND M. D. DANIELSa,e a Department of Geography, Kansas State University, Manhattan, Kansas, USA b School of Environment and Natural Resources, Ohio Agricultural Research and Development Center, The Ohio State University, Wooster, Ohio, USA c Division of Biology, Kansas State University, Manhattan, Kansas, USA d Department of Biology, Tennessee Technological University, Cookeville, Tennessee, USA e Stroud Water Research Center, Avondale, Pennsylvania, USA ABSTRACT Run-of-the-river dams (RORDs) comprise the vast majority of dams on river systems and are commonly removed as a part of stream resto- ration strategies. Although these dams are routinely removed, few studies have documented the geomorphological responses of sand-bed rivers to the removal of RORDs. We examined the response of a large sand-bed river located in South-Central Kansas, USA, to the installation and removal of a dam that is installed annually for seasonal recreational purposes. Channel adjustments were tracked using cross-sections sampled over the course of 7 months as the dam was installed and subsequently removed. Multivariate spatiotemporal analysis revealed emergence of channel stability when the dam was in place for most cross-sections, except for those immediately adjacent to or at great distances from the dam. Our results provide an approximation for how sand-bed rivers respond to RORD construction and removal and are useful for guiding management decisions involving preservation or restoration of connectivity. Results of this study suggest that sand-bed rivers are resilient and recover quickly when transient RORDs are removed. Copyright © 2014 John Wiley & Sons, Ltd. key words: dam removal; run-of-the-river dam; fluvial geomorphology; sand-bed river; Great Plains Received 15 May 2014; Revised 26 August 2014; Accepted 09 September 2014 INTRODUCTION (American Rivers et al., 1999; Gleick et al., 2009), 57% of which were in the USA (Pohl, 2002). Research re- Dams have long been recognized as essential contribu- garding hydrologic, hydraulic and geomorphic forms tions to societies that come at a pervasive and detrimen- and processes associated with dams are still burgeoning tal cost to riverscapes (Baxter, 1977). The mid-1900s fields and much remains to be learned about channel re- was a period of significant dam construction in the sponse to dam construction and removal. USA. Given that the average dam life span is 50 years Run-of-the-river dams extend the entire width of a (Doyle et al., 2003b), approximately 85% of dams are channel, have no structures that exceed the elevation of the nearing the end of their lifespan and are in need of main- bankfull channel and have no mechanism to regulate tenance or removal (Evans et al., 2000). Run-of-the-river discharge so that water flows freely over the crest of the dams (RORDs) represent 97% of dams within the USA structure (Csiki and Rhoads, 2010). The hydrologic effect (Federal Emergency Management Agency (FEMA) and of RORDs is negligible because there is no flow inhibiting USA Army Corps of Engineers (USACE), 1996) and structures; however, they can have dramatic hydraulic and pose regulatory challenges because there are few ecolog- geomorphic impacts. The main effects associated with ical or economic benefits to justify their continued RORDs are the development of low-velocity pools upstream operation beyond their useful lifespan (Vedachalam and of the dam (Stanley and Doyle, 2002), which can result in Riha, 2013). Dam removal is a common restoration strat- sediment deposition within the impounded area (Friedl and egy (e.g. Hart et al., 2002; Gleick et al., 2009), and over Wuest, 2002; Vanoni, 2006). Downstream of dams, intense 700 dams were removed worldwide in the last 100 years scour can occur at the base of structures as the stream reach degrades in order to re-establish sediment load through ero- *Correspondence to: K. H. Costigan, School of Environment and Natural sion of bed and bank materials (Bollaert and Schleiss, 2003; Resources, Ohio Agricultural Research and Development Center, The Ohio Doyle and Harbor, 2003). The river response to RORD State University, 1680 Madison Ave, Wooster, Ohio 44691, USA. removal depends upon the amount of sediment stored E-mail: [email protected]; [email protected] †Current address: School of Geoscience, University of Louisiana at Lafayette, behind the impoundment, which can range from negligible Lafayette, LA, 70504. to complete infilling (e.g. Wildman and MacBroom, 2005; Copyright © 2014 John Wiley & Sons, Ltd. K. H. COSTIGAN ET AL. Csiki and Rhoads, 2014). When low volumes of sediment MATERIALS AND METHODS are stored in the upstream impoundment, there is minimal Study site response to the removal of the dam (Skalak et al., 2009; Csiki and Rhoads, 2010, 2014). When large volumes of The Camp Mennoscah Dam (CMD) is located on the SFNR sediment are stored behind the impoundment, a large pulse of in South-Central Kansas, USA (Figure 1). At CMD, the sediment is introduced downstream, and the river responds as SFNR drains approximately 1680 km2 in a low-relief region. if a large dam was removed (Poff and Hart, 2002). To date, The SFNR flows east-southeast through tallgrass prairie studies have documented channel conditions prior to dam within the High Plains physiographic region, which is removal (e.g. Draut et al., 2011; Brenkman et al., 2012) and characterized by 3- to 5-m-thick loess deposits that overlie post-dam conditions (e.g. Doyle et al., 2003b; Wildman and thick deposits of Pleistocene and/or Pliocene alluvium MacBroom, 2005; Major et al., 2008; Pearson et al., 2011) (Mandel, 2008). The SFNR drains sandy areas, and as a of large dam removal projects. However, much less is known result, the river is wide, shallow and straight with a bed that about pre-dam conditions in general or what the effects of is continuously moving at most discharges and a bedload removing a RORD might have on a sand-bed river. that is sand (Costigan et al., 2014). Large rivers in the Great Plains are typically sand- Mean daily discharge is recorded at a US Geological dominated alluvial rivers without confining valley walls Survey gage (07145200) located approximately 18 river (Graf, 2005). Great Plains rivers are especially susceptible kilometers downstream of the CMD and has a mean annual to altered hydraulics and sediment regimes associated with discharge of 5.8 m3/s (minimum discharge = 2.8 m3/s; dams because of their low entrainment thresholds and maximum = 10.5 m3/s; 1951–2012) and mean annual flood weak bank materials (Montgomery and Buffington, discharge of 119 m3/s (minimum discharge = 7.2 m3/s; 1997). Great Plains rivers are highly fragmented by anthro- maximum = 510 m3/s; 1951–2012). During the surveying pogenic barriers and Plains states rank amongst those with the campaigns, the mean discharge was 2.4 m3/s (minimum = most in-stream barriers within the USA (Perkin et al., in press). 0.45 m3/s; maximum = 18 m3/s; Figure 2). The maximum Large rivers within the Great Plains have a high proportion observed discharge during the surveying campaign was be- of the largest reservoirs in the USA (Graf, 2005), and rivers tween the pre-dam survey and the first sample following with large dams (>10 m) have dramatically altered flow dam installation. During the surveying campaign, the Great regimes (Costigan and Daniels, 2012). Large dam struc- Plains was in the midst of a severe 2-year drought (Perkin tures on Great Plains rivers have resulted in rapid channel et al., in press), and flows in the SFNR were generally less narrowing as vegetation encroaches because of lower or than the 95% confidence intervals of flow for the 61 years absent flood events (Julian et al., 2011). Great Plains rivers of recorded discharge (Figure 2). No channel-forming flows with large, regulating dams on them may have up to 91% occured during the study period. The headwaters of the less standard active area than unregulated rivers in the SFNR are within the High Plains Aquifer, but this region region (Williams, 1978; Graf, 2006). Given that RORDs are of the aquifer has experienced no significant change in such common features on contemporary Great Plains groundwater levels (Sophocleous, 2000). riverscapes (Perkin et al., in press) and that RORD removal is The CMD is a 1-m-high dam that is installed annually for a growing restoration technique, understanding potential effects seasonal recreational purposes associated with Camp of dam removals on Great Plains rivers will help direct conser- Mennoscah (e.g. canoeing and swimming) and is in place vation action and inform management of sand-bed rivers. from April through October. There are permanent I-beams The goal of this work was to document how a sand- fixed to the bedrock approximately every meter across the dominated river responds to the construction and removal width of the river (Figure 3A and B). Wood planks are of a RORD. The specific objectives of this study were to placed horizontally between the I-beams and added in layers document the spatiotemporal extent of changes in channel until the dam is 1 m in height (Figure 3C and D). The dam morphological characteristics that occurred as the result of was installed on 4 April 2012 and was removed on 7 a RORD installation and removal. We quantify the geomor- October 2012. Because the dam has been added and phic conditions prior to, during and after installation of a removed from the river annually for at least fifty years, we temporary RORD on the South Fork Ninnescah River use ‘installation’ rather than ‘construction’ to capture its (SFNR) in South-Central Kansas, USA.