River connectivity reestablished: Effects and implications of six weir removals on brown trout smolt migration

Today's river systems have been extensively modified, requiring us to rethink how we approach the management of these important ecosystems. We evaluated the effects of removing 6 weirs in River Villestrup (Jutland, Denmark) on the smolt run of brown trout (Salmo trutta) over the course of 12 years. During 5 of these years, we evaluated the number, size, and timing of smolts during their downstream migration. We found an increase in smolt output following the weir removals, along with a decrease in average length and indications of an earlier peak migration. Our results suggest that barrier removal has led to an increase in spawning success by adults, fry survival, recruitment, and smolt migration success. Weir removal is therefore a viable management approach to restore connectivity in freshwater streams and rivers, which promotes the passage of smolts as they migrate to marine environments.

The diversity, abundance, and sustainability of aquatic species have long been threatened by the human-induced fragmentation of rivers (Khan & Colbo, 2008;Saunders, Hobbs, & Margules, 1991). Barriers in the form of dams, weirs, and culverts have become so prominent in today's river systems that the majority of them have lost their original connectivity and natural characteristics (Jager, Chandler, Lepla, & Van Winkle, 2001;Jungwirth, Schmutz, & Weiss, 1998). These barriers exacerbate the current poor state of many freshwater ecosystems.
Efforts to mitigate the impacts of barriers, such as fishpasses, have seen limited success (Bunt, Castro-Santos, & Haro, 2012) and are usually costly (Gibson, Haedrich, & Wernerheim, 2005). Furthermore, such approaches do not repair the damage done to the ecosystems as a whole (Birnie-Gauvin, Aarestrup, Riis, Jepsen, & Koed, 2017); rather, they provide an opportunity for some fish to move upstream or downstream past the barrier. This is particularly relevant for migratory fish species such as salmonids, which depend on freshwater Kennedy, Ladle, & Milner, 2003;Shirvell & Dungey, 1983). Juvenile trout generally spend between 1 and 5 years in freshwater, after which individuals differentiate phenotypically (Nielsen, Aarestrup, Nørum, & Madsen, 2003). Some individuals will assume a resident phenotype and remain in freshwater their entire life, whereas others will assume the migratory phenotype and migrate to marine environments (Jonsson & Jonsson, 1993;Nielsen, Aarestrup, & Madsen, 2006). This phenomenon is known as partial migration (Chapman, Brönmark, Nilsson, & Hansson, 2011).
Although the drivers for partial migration remain poorly understood (though many hypotheses exist, Chapman et al., 2011), the benefits of migrating to sea appear to be linked to a larger availability of food items in marine environments, thus allowing migratory individuals to attain larger sizes and a greater reproductive potential (Chapman et al., 2011;Northcote, 1984;Shrimpton, 2013). Juveniles that become migratory individuals are known as smolts and differ from their resident counterparts both behaviorally and physiologically. For example, smolts appear to be less aggressive (Jonsson & Jonsson, 2011;Thorstad et al., 2012), have greater sodium-potassium ATPase activity in their gills (Aarestrup, Nielsen, & Madsen, 2000), and appear to have greater levels of blood-circulating antioxidants .
Smolts typically migrate during the months of March to May depending on latitude (peak smolt migration period, e.g., Bohlin, Dellefors, & Faremo, 1993), though some migrate during the autumn (Winter et al., 2016. The downstream smolt migration is thought to be triggered by a range of environmental factors, such as photoperiod, temperature, and discharge (Hoar, 1988). Furthermore, smolts are thought to migrate downstream during the "smolt window." This window is thought to be affected by factors such as physiological and ecological readiness to enter marine environments, risk of predation, and growth potential (McCormick, Hansen, Quinn, & Saunders, 1998). It is thus essential that smolts be able to reach marine waters as quickly and easily as possible, with their passage unhindered.

| The restoration project
Barriers cause the upstream portion of the river to become inundated and thus can hinder the passage of smolts heading downstream due to the slowing of water (e.g., Schwinn, Aarestrup, Baktoft, & Koed, 2017) and difficulties associated with finding a safe passage route past the structure itself (e.g., Thorstad, Økland, Kroglund, & Jepsen, 2003). Furthermore, barriers hinder the upstream passage of adult trout during their spawning migration. In Denmark, such barriers often occur in the form of weirs in conjunction with fish farms. River Villestrup (northeast Jutland, Denmark) historically had 17 fish farms. In an attempt to restore the river to its original state and reinstate connectivity on the lower two thirds of the river, six weirs (five in the mainstem and one in a tributary) were removed. All associated fish farms were simultaneously closed. The weirs were likely to have been several hundred years old, though precise years of origin are not available. Each weir was originally made of concrete or wood and removed  Table 1 for specific details on each weir). Today, only one weir remains in the upstream portion of the river (Figure 1c, no. 6). This study investigated the effectiveness of this restoration approach with regard to the smolt run over the course of 12 years (five study years).

| Study site and trap set up
River Villestrup is located in northeast Jutland (Denmark), where it runs for 20 km before entering the Mariager Fjord ( Figure 1). The river is fed by groundwater and rainfall and has a mean annual discharge of 1.1 m 3 s −1 . It is home to a wild population of partially anadromous brown trout, with both resident and migratory phenotypes. Before the weir removals, river Villestrup was characterized mostly by sandy and muddy substrates in the close vicinity of the weirs, with little pool/riffle habitat. As in several Danish rivers, river Villestrup had and still has a relatively low gradient (approximately 1.0%) and meandering form. However, following the removals, the river bed is Unfortunately, given the expenses and time required to maintain a trap for 2 months, we could not perform the study continuously between 2004 and 2016. Thus, specific study years were selected to provide the most representative data to evaluate the effects of weir removal through a before-after approach.

| Fish processing
Every day during the study period, the trap was emptied to count and measure (±0.1 cm) all smolts. Fish were anaesthetised with benzocaine (0.03 g l −1 ) for measurements and fin clipped (adipose fin). Fish were then released just downstream of the trap. Although it was unlikely, fish could return upstream after having been measured. In that case, finclipping allowed us to detect if a fish had already been measured and counted, and that individual was then removed from the day's count.

| Environmental variables
Water discharge data were obtained from a monitoring station located 750 m upstream of the trap. Temperature data were obtained using an underwater temperature data logger (Onset HOBO Tidbit v2 UTBI-001, range: −20°C to 70°C, Massachusetts, USA).

| Data analysis
All trout between 10.0 and 21.0 cm caught in the trap were considered to be smolts (despite coloration) for the purpose of the analysis. This is a fair assumption given the close distance between the trap and the fjord. Furthermore, a follow-up electrofishing pass downstream of the trap after the end of the smolt season showed very few trout.
Mean length between years was compared using a simple linear regression model: Lengths were log-transformed to meet assumptions of normality and homoscedacity.
All statistical analyses were performed using R version 3.4.1.

| RESULTS
The size of the smolt run increased following the removal of weirs, with the largest class in 2015, followed by 2016 (Table 2; Figure 2).
Average length of downstream migrating trout was different across study years, decreasing significantly every year (p < .05; Figure 3).
We note an indication of earlier peak migration following weir removal (Figures 2 and 4). where ponded zones previously were. These environmental changes presumably restored or even created new grounds ideal for spawning and early development which adults and fry did not have access to for centuries, when fish farms and mills were first established in the river system. Adult sea trout are also able to spawn farther upstream than Note. Height (m), width (m), length of ponded zones (m), presence or absence of fishway, and date of removal for the weirs found in River Villestrup.  class of brown trout can affect one or more subsequent age classes through intraspecific competition between cohorts (Elliott, 1994;Nordwall, Näslund, & Degerman, 2001). In this case, the 1+ age class which migrated in 2015 may have significantly reduced the abundance of the 0+ age class which would have migrated in 2016, either through predation, density-dependent mortality, or intraspecific competition (Elliott, 1994). (2) It is possible that the decrease was due to variation in the annual smolt production, which may vary from year to year due to variation in biotic and abiotic factors (Chadwick, 1982;Warren, Dunbar, & Smith, 2015). In this case, we would expect the number of smolts to increase again in the upcoming years. (3) It is possible that the population suffered high overwinter mortality due to harsh environmental conditions (Elliott, 1993). At least one other Danish stream Birnie-Gauvin, Gudsø stream).

| Smolt size and peak migration
We observed a decrease in the average smolt size through the years.
It is possible that following weir removal, smaller fish were also successful in migrating downstream, rather than larger fish only, which are presumably more apt at escaping predators in ponded zones or overcoming weirs (Winstone, Gee, & Varallo, 1985). In other words, smaller fish no longer get stuck at weirs and/or penetrate the grid used to prevent fish from entering the water intake channel at fish farms and are capable of descending downstream. Another possibility for progressively smaller fish following weir removal is that a greater number of fish caused higher intraspecific competition for food and may have resulted in smaller fish (Holm, Refstie, & Bø, 1990). Additionally, it is likely that spawning success and recruitment increased, which simply increased the number of migrating fish, with a wide range of sizes. Our findings likely reflect a combination of all three possible explanations. Alternatively, it is possible that the removals impacted the invertebrate community, and thus, may have reduced food availability. Although we cannot rule out this explanation, it is rather unlikely that the post-removal invertebrate community had diminished so much that fish were smaller. Because fastflowing water is typically inhabited by different invertebrate types than slower moving water (Doisy & Rabeni, 2001), we argue that We expected the peak migration to occur earlier following the removal of the weirs through a reduction in delays at ponded zones but cannot make that conclusion for certain. Although our results indicate a trend for an earlier peak migration, flood events during the study years make it impossible to make a meaningful analysis. Evidence suggests that dams delay the passage of migrating fish greatly (Aarestrup & Koed, 2003;Gauld, Campbell, & Lucas, 2013), and that these effects are worse when multiple dams must be overcome (Caudill et al., 2007). Ponded zones can cause smolts to lose their orientation due to diminished flow, thus delaying them (Schilt, 2007). The removal of five of the six weirs in the main stem of river Villestrup likely prevented such delays in downstream migration, thus enabling fish to reach marine environments faster.

| Implications
Our results suggest that complete barrier removal has several important implications for freshwater fisheries and river management. Weir removal presumably increases the number of adult fish able to successfully migrate upstream and spawn, perhaps due to a reduced incidence of injuries at obstacles, diminished energy expenditure to attain spawning grounds (i.e., adults no longer have to invest energy to surpass barriers), and by making impassable stretches into passable ones (Castro-Santos & Letcher, 2010). Furthermore, weir removal may increase reproductive output through successful egg emergence (i.e., unhindered by sedimentation), which would then lead to an increased recruitment rate and an increased smolt output in the following 2+ years. Weir removal also makes smolts more successful in their downstream migration via reduced mortality at fish farm intake grids (Aarestrup & Koed, 2003), reduced predation at ponded zones (Jepsen, Aarestrup, Økland, & Rasmussen, 1998), decreased delays (Aarestrup & Koed, 2003;Schilt, 2007), and presumably decreased energy expenditure. In addition, barriers may induce an artificial population structure by favoring larger individuals; removal can reinstate a more natural population structure, with a wider size range.
Many of the fish species that migrate between freshwater and marine waters, including trout, are used as indicator species for good environmental and ecological status, as they experience many habitats during their movement from upland streams to lowland rivers and then to the sea (Gough, Philipsen, Schollema, & Wanningen, 2012;Lasne, Bergerot, Lek, & Laffaille, 2007). Their importance in the context of management cannot be understated. Fish usually migrate for one of three reasons; migrations are either for spawning, feeding, or refuge seeking (Northcote, 1984). Regardless of the causes for migration, barriers diminish the ease of access to spawning and feeding grounds and hinder passage to refuge areas. These effects are likely exacerbated in rivers with numerous barriers (Lucas & Batley, 1996).
Extensive fragmentation of river connectivity limits dispersal of many fish species (McLaughlin et al., 2006). Furthermore, dams impact the hydrogeomorphology of streams in some places. For example, barriers cause a decrease in water velocity, an increase in water temperature, a decrease in oxygen availability, and sedimentation (Baxter, 1977;Petts, 1984). Because most diadromous species exhibit homing behaviour, and because the latter is directly related to predictable environmental conditions such as temperature, water chemistry, and rhythmic patterns of environmental changes, their homing behavior is likely to be greatly impacted by the presence of obstacles (Lucas & Baras, 2008).
In the present study, we demonstrate that weir removal is an appropriate approach to reinstate river connectivity and to increase long-term population sustainability of fish species. We provide some of the first data evaluating the full river system effects of barrier removal and further emphasize the need to implement this approach in management schemes whenever possible.