Short‐term reservoir draining to streambed for juvenile salmon passage and non‐native fish removal

Abstract Fish passage out of reservoirs is a critical issue for downstream movement of juvenile salmonids and other migratory species. Reservoirs can delay downstream migrations by juvenile salmon for months or years. Here, we examine whether a novel management activity implementing annual short‐term draining of a reservoir to streambed improves timely downstream migration of juvenile salmonids. We analyse 12 years of fish capture data from a screw trap located downstream of Fall Creek Reservoir (Oregon, USA) to examine changes in timing of passage out of the reservoir and to compare fish species composition pre‐ and post‐draining. We observed a contraction in the timing of downstream migration for juvenile Chinook Salmon and reduction of yearlings in years following draining. We suggest that briefly draining the reservoir to streambed leads to reduced abundance of warm‐water invasive fishes in the reservoir after it refills. These changes could decrease predation and shift competition between invasive and resident riverine‐adapted native fishes in the reservoir. Collectively, our findings suggest that this low‐cost reservoir management option may improve passage and connectivity for juvenile Chinook Salmon while also decreasing the abundance of invasive fish species in the reservoir. This case study underscores the crucial need for further evaluations of reservoir draining in other systems and contexts.

because of difficulty locating downstream outlets and because outlets that attract fish may cause injury (Korn & Smith, 1971;Raymond, 1979;Schilt, 2007). Modification of dam outlet structures or construction of new structures to allow more effective downstream passage of fish are expensive (Williams, Armstrong, Katopodis, Larinier, & Travade, 2012) and can lead to additional problems with invasion of undesirable non-native species (Fausch, Rieman, Dunham, Young, & Peterson, 2009).
The establishment of invasive species in these areas is often facilitated by the creation of the artificial lentic, reservoir habitats associated with large dams (Havel, Lee, & Zanden, 2005;Vander Zanden, Lapointe, & Marchetti, 2015). In the PNW, invasive fish species in reservoirs include top piscivores such as Largemouth Bass (Micropterus salmoides) and crappie (Pomoxis sp.) that co-occur with native juvenile salmonids (Keefer et al., 2013). The abundance of invasive piscivores associated with large dams and reservoirs in this region also raises concerns about juvenile salmonid survival due to potential increases in predation (Monzyk, Romer, Emig, & Friesen, 2014). This predation risk likely acts in concert with passage problems negatively impacting juvenile salmon. Because of the low success and high cost of construction of downstream passage facilities, especially in highly fluctuating reservoirs, temporary draining of reservoirs to streambed during outmigration periods has been proposed. This is considered an operational alternative to improve downstream passage for juvenile Chinook Salmon (Oncorhynchus tshawytscha) at some PNW dams (Tiffan, Garland, & Rondorf, 2006; United States Army Corps of Engineers [USACE], 2015). By fully opening a dam outlet (e.g., gate valve), removing a physical barrier and temporarily converting the lentic reservoir into a lotic river, draining to streambed could improve the ability of juvenile salmon to locate the reservoir outlet and to pass downstream with minimal obstruction, high survival and without handling. This is effectively reservoir flushing, and may be accompanied by a washout of sediments (Schenk & Bragg, 2014) as well as other fish species.
Although there is some history of draining for passage and dam maintenance (Anderson, 2007), this strategy has not previously been evaluated for short-and longer-term impacts to the reservoir community.
Here, we quantify changes in timing of downstream passage of fish and in the fish community structure before and after brief fall annual draining of the reservoir to streambed using 12 years of data from fish collections downstream of a PNW reservoir. We also examine whether draining to streambed was followed by changes in the number of returning adult Chinook Salmon. We hypothesize that short-duration reservoir draining to streambed can be used to 1) increase the effectiveness of downstream passage of juvenile salmon, improving numbers of individuals available to recruit to the adult population. We predict that more effective passage will be evidenced by a reduction in passage in the months after reservoir draining to streambed (juvenile salmon will have already moved downstream) and that adult numbers will increase in subsequent years. 2) We also hypothesize that reservoir draining to streambed will ultimately reduce abundances of warm-water invasive species in the reservoir.
We expect warm-water species to be highly sensitive to the temporary lotic conditions experienced during draining, which might favour a transition to a community dominated by cool-water riverine-adapted native species in the reservoir over time. We expect a reduction in abundance of warm-water invasive species will result in shifting composition from invasive-dominated catches to nativedominated catches in the downstream screw trap. This work is an important first step in evaluating a novel management activity that could serve as an alternative to modification of dam outlet structures and may offer a cost-effective solution to downstream passage and invasive species management.

| Study site
This study was conducted in Fall Creek Reservoir, which was con- To facilitate downstream passage of salmonids, Fall Creek Dam was constructed with "fish horns." These are nine horn-shaped outlets arranged in groups of three at three elevations on the upstream face of the dam (Figure 1). To migrate downstream, fish must either locate or move through these fish horns or through the dam regulating outlet at the base of the dam; spillway structures are reserved only for emergencies at this location. Korn and Smith (1971) noted that the fish horns at Fall Creek Reservoir were not used by the majority of juvenile Chinook Salmon to migrate downstream. A recent study (Normandeau Associates, Inc., & Pierce, 2014) re-evaluated the performance of the fish horns and suggested they may have high immediate (56%) and delayed (89%) mortality rates for juvenile subyearling Chinook Salmon outmigrating in the fall. Juvenile Chinook Salmon grow larger in Fall Creek Reservoir than in nearby streams (Monzyk et al., 2014), and their large size may contribute to increased injuries and associated mortality.

| Reservoir draining to streambed
To remedy problems with limited passage through the fish horns (Korn & Smith, 1971), brief draining of Fall Creek Reservoir to streambed occurred annually from 1969 until 1979, and once more in 1989, after which reservoir operations changed to maintaining winter reservoir elevations at the "minimum conservation pool" (15-m deep; Figure 1). Contemporary reservoir draining to streambed was initiated in November 2011 and has occurred annually since. The duration of each annual contemporary draining to streambed event is approximately 1 week but varies depending on inflow conditions ( Figure   S1). Prior to the contemporary streambed draining, the composition of fishes in Fall Creek Reservoir roughly resembled those found in other nearby reservoirs (Table 1). June and August ( Figure 2). Trapping was also not possible for short periods during reservoir draining when the reservoir water level reached streambed, due to clogging from high sediment loads. Efficiency tests for trapping of fish noted up to 12.5% trap efficiency with higher efficiencies generally observed for live fish and during higher water discharge periods (G. Taylor, unpublished data). USACE personnel emptied the trap daily or every other day during operations in accordance with Oregon Department of Fisheries and Wildlife permits 15228, 16345, 17137, 17862, 18646, 19466, 20086, and 21274. Body sizes of juvenile Chinook Salmon collected were measured (fork

| Statistical methods
Data were cleaned to standardize coding across years and analysed using R Statistical Computing software (R. Core Team, 2014)  Salmon passage, we used data that included complete subyearling cohorts (i.e., April, when fry first start to enter the reservoir, to March, 2006March, -2017. Monthly and annual proportions and species-specific sizes were compared for years before and after reservoir draining to streambed using unpaired Wilcoxon tests implemented in R with the ggpubr package (Kassambara, 2018). We used the same methods to evaluate changes in the abundance and size structure of invasive species. As reservoir management outside of draining has not changed, we assume that changes in screw trap data reflect changes of the fish community inside of the reservoir. In addition to the proportion of invasive species with respect to native species, which we expect to be less affected, we compared before and after abundance and size within each species as described above. Numeric counts were highly variable across years but are presented in the case of species that were no longer detected in years following draining events. November 18, a significant contraction in the later end of the distribution (unpaired two-tail t test, p < .01; Figure 3, Table 2). The median passage date for juvenile Chinook Salmon before and after draining has not changed nor has the difference between the median and 75th percentile. However, the variability of the timing of the 75th percentile was reduced after reservoir draining to streambed. Though few to no fish passed after draining treatments in November, the timing of passage was variable before draining. We found no significant differences in the days elapsed between the 5th or 25th percentiles of the cumulative distribution compared to the median. Changes in timing were not accompanied by significant changes in the maximum Chinook Salmon size observed annually moving downstream nor in counts per year, though there was a trend towards smaller salmon, consistent with a reduction of reservoir-reared yearlings ( Figures S2 and S3).

| Invasive fish species downstream passage
The proportion of invasive species of the total number of fish exported daily across years decreased over time after 2011 following the initiation of reservoir draining to streambed (Figure 4). showed trends of larger maximum sizes over this same period ( Figure S3).

| Adult spring Chinook Salmon returns
To address whether the reservoir draining had longer term effects on

| DISCUSSION
This study shows that short-term draining of a reservoir during late fall increases downstream passage efficiency for juvenile Chinook Salmon.
Draining also passes non-native competitors and predatory fish out of the reservoir, which likely reduces their densities in the reservoir and their impact on incoming Chinook Salmon fry the following spring.
However, the increased outmigration has not yet resulted in increased returns of Chinook Salmon adults for spawning upstream of the dam and reservoir.
The absence of late migrants suggests that a high proportion of juvenile Chinook Salmon move through the dam during reservoir draining to streambed, leaving few to no juveniles remaining in the reservoir for outmigration later in winter or as yearlings. Because the juvenile Chinook Salmon grow quickly in reservoirs, and the fish horns in the dam result in injuries, outmigration as juveniles through the regulating outlet, rather than the horns, appears to lead to reduced mortality for young Chinook Salmon (Nesbit et al., 2014). These factors support draining to streambed as an "effective passage" option, defined by a high proportion of outmigrating juvenile Chinook Salmon passing through the dam with low passage-related mortality.
Draining to streambed may also help to reduce reservoir-related predation risk for juvenile Chinook Salmon through multiple mechanisms. In addition to changes in the abundance of invasive fishes being exported from the reservoir with draining, recent evidence indicates that both native and invasive reservoir resident piscivores switch diets to use lower trophic level sources, consistent with reduced prey fish densities and optimal foraging theory, following draining to streambed events . Thus, each spring following draining, juvenile Chinook Salmon may be entering a reservoir with a reduced predator density, and those remaining predators may be focused on capturing lower trophic level food items.
While the downstream export of large numbers of invasive fishes during draining is of concern for downstream river food webs, the same species have been passing through the dam throughout the year each year. The shift in timing of export of invasive species might lower their survival downstream, especially in river networks where conditions during the periods of export could be less favourable to their establishment (e.g., export of warm-water lentic fish downstream during the winter months with high discharge and cold conditions).
Although reservoir draining appears to reduce mortality for juvenile Chinook as they pass through the dam, as well as provide ecological benefits, such as a reduction in the size and number of invasive predatory fish, it is unclear what the long-term implications of this pulsed downstream migration will be for maintaining a diversity of FIGURE 4 Invasive species as a proportion of the total daily catch (open circles) before and after draining to streambed. The thicker line along the X-axis represents the period including reservoir draining to streambed. Water surface elevation in the reservoir is represented by the light grey line and is plotted against the secondary Y-axis FIGURE 5 Proportion of invasive fish species (number of invasives per total catch) by month in years before (light) and after (dark) draining to streambed. Boxes represent median and 25th to 75th percentiles, whiskers represent the largest and smallest values within 1.5 * the interquartile range (distance from the 25th to 75th percentile). Stars indicate significant differences before and after contemporary reservoir draining to streambed events (unpaired Wilcoxon Test, R, * < .01, ** < .001) Chinook Salmon life histories (see Schroeder, Whitman, Cannon, & Olmsted, 2015). After filling of Fall Creek Reservoir, as observed for other reservoirs (Romer, Monzyk, Emig, & Friesen, 2013), the majority of juvenile spring Chinook Salmon have exhibited downstream passage through the dam in the fall after spring and summer rearing; a smaller proportion of juvenile salmon outmigrated throughout the year as fry, subyearlings, or yearling fish. In years without draining, juveniles continued to exit the dam past the fall months, even when water levels increased above the typical winter minimum pool.
Draining results in the absence of these late migrants, including subyearlings passing in winter and yearling fish. Trade-offs will need to be weighed, given that Chinook Salmon life histories have likely already been modified by the dams and given potential benefits of the reservoir draining on reduced juvenile mortality.
We expected to see an increase in numbers of returning Chinook Creek Reservoir, an oligo-mesotrophic system, suggest that the downstream water quality impacts may be short-lived. Regardless, reservoir trophic status and amounts of sediment to be mobilized should be considered before draining. Timing of draining is also important to consider. In strongly seasonal environments, such as the PNW, a brief winter draining compared with a brief spring or summer draining would be expected to have very different consequences both within the reservoir and downstream. Concerns over the export of invasive species may be greater in some regions. In the PNW, invasive species are most associated with lentic habitats and are regularly exported from reservoirs regardless of draining (Keefer et al., 2013). Even so, they appear to be outcompeted in the upper Willamette River downstream of the dam described here (Williams, 2014) and the observed shift in export timing to unfavourable seasons may reduce risk. Finally, the large water level fluctuations associated with draining could be of concern for aquatic communities in reservoirs typically operated to have stable water levels, as even comparatively small water level fluctuations can have large consequences for lentic systems (Wantzen et al., 2008).
New options for reservoir management are increasingly under consideration as managers work to reduce ecological impacts of the dam and reservoirs on native species (Chen & Olden, 2017;Poff & Schmidt, 2016). Water levels in large reservoirs are often lowered for reasons other than downstream passage (e.g., for repairs or during drought years). These can provide unique opportunities that could be extended to additional studies of draining to streambed, especially to examine the impacts on reservoir conditions and resident fishes. Considering that draining may provide a low-cost option to improve passage and connectivity while also exporting invasive species, we highlight the critical need for further evaluations in other systems and contexts. Draining a reservoir to streambed is an interesting management alternative to aid downstream passage of juvenile salmon through large dams, and it needs to be evaluated on a case-by-case basis before broad implementation. For example, considering the variability in types of dam outlet structures and potential loss of hydropower generation during the draining and refill periods. Finding more ecologically sound reservoir management practices is a growing concern as we balance the societal benefits of large dams with their ecological costs and draining offers one option in a broad management portfolio.