Influence of Non-Native Trout on Native Non-Game Fish in Nebraska Headwater Streams

The direct predatory effects of introduced rainbow trout Oncorhynchus mykiss on native longnose dace Rhinichthys cataractae were examined using a series of in-stream enclosures to determine if presence, density, or acclimation period of rainbow trout influenced survival of longnose dace. The number of longnose dace remaining in enclosures over the first 72 hours after rainbow trout introduction differed in the presence and absence of rainbow trout, but did not differ between average and high densities of rainbow trout (F2,258.9 = 3.73, P = 0.03). Rainbow trout acclimated to the stream for longer periods had a greater initial influence on the number of longnose dace remaining in enclosures relative to those acclimated for shorter periods (F4,148.5 = 2.50, P = 0.04). Several factors likely influenced rainbow trout predation rates including predator experience, prey body length, and habitat availability. Future research should focus on both direct and indirect predatory interactions between rainbow trout and longnose dace in the context of whole assemblages. 34 INTRODUCTION Rainbow trout Oncorhynchus mykiss and longnose dace Rhinichthys cataractae are two species that often co-occur in small coolto cold-headwater streams. The native ranges of rainbow trout and longnose dace historically overlapped only in the Pacific Northwest. Today, rainbow trout have been introduced across nearly the entire native range of longnose dace. With these introductions, and with such common co-occurrence which could result in negative competitive or predatory interactions, it is surprising that no studies have directly examined interactions between rainbow trout and longnose dace. Understanding the underlying ecological interactions among these two species will provide the information necessary to ensure proper management and survival of both species in the future. Rainbow trout are opportunistic, generalist feeders, and consume a variety of aquatic and terrestrial invertebrates (Fenner et al. 2004, Metcalf et al. 1997), as well as some small fish (e.g., Blinn et al. 1993, Klammer 1984). In general, rainbow trout are found in coolto cold-water streams with distinct riffle-pool complexes and a variety of substrate. Microhabitat use within streams often shifts with both age and season (e.g., Baltz et al. 1991). Longnose dace are a small-bodied, riffle-dwelling cyprinid species. Longnose dace are abundant throughout their native range, which spans most of the North American continent (Scott and Crossman 1973a). Longnose dace prey on a variety of invertebrates including Diptera (e.g., Simuliidae, Chironomidae), Ephemeroptera (e.g., Baetidae, Siphlonuridae), and Tricopetera (e.g., Hydropsychidae; Reed 1959, Gee and Northcote 1963, Gerald 1966, Gibbons and Gee 1972, Pappantoniou and Dale 1982, and 35 Mullen 1991). Longnose dace shift microhabitat use with age, but are most often found in moderately fast to fast current and cobble or boulder substrate (Mullen and Burton 1995, Gee 1968, Gee and Northcote 1963, Gibbons and Gee 1972). Both rainbow trout and longnose dace are found in streams throughout Nebraska. These streams are somewhat unique compared to others in which these two species coexist. Nebraska streams are typically low gradient, consist of grassy riparian vegetation, and have little in-stream woody debris. Undercut banks and overhanging vegetation make up the majority of cover for fish in these streams. There is often little distinction in pool and riffle habitat, and little substrate complexity (largely dominated by sands). The lack of distinct pools and riffles may cause greater habitat overlap among rainbow trout and longnose dace compared to other systems, and the lack of cover for small-bodied species such as longnose dace suggests that there may be increased predatory risk from rainbow trout in Nebraska streams. Introductions of rainbow trout have negatively influenced several native species and have been implicated in the decline of many at-risk species in other systems (Turek et al. 2013). Competition and hybridization are the two most commonly cited mechanisms by which non-native rainbow trout influence native populations, yet rainbow trout are piscivorous and have been shown to feed on forage fish in Nebraska streams (Klammer 1984). Further, longnose dace may be at an increased risk of predation in headwater streams because large, native, piscivorous fish predators are not common in systems with longnose dace (Nebraska Game and Parks Commission unpublished data). Native predators of longnose dace in Nebraska streams consist almost entirely of birds (e.g., Belted Kingfishers and Great Blue Herons), mammals (e.g., Raccoons and Mink), 36 and macroinvertebrates (e.g., Odonates and Belostomids). Creek chub and grass pickerel are both native to these systems and likely prey on cyprinids to some extent (e.g., Schlosser 1988, Scott and Crossman 1973b). However, relative to other predators, they probably have little overall impact on native fish populations due the low abundance of large piscivorous individuals. Longnose dace populations are thought to be secure throughout their range, yet some populations have responded negatively to the introduction of non-native species. Introduction of non-native fishes was cited as a factor in the decline of the now extinct Banff longnose dace Rhinichthys cataractae smithi, once found only in a single marsh in Alberta, Canada (Miller et al. 1989). Similarly, rainbow trout are cited as a major factor in the extinction of Grass Valley Speckled dace Rhinichthys osculus reliquus, a closely related species to longnose dace, in Nevada (Miller et al. 1989). Concern over the potential negative influence of stocking non-native trout in Nebraska streams has led resource managers to prohibit stocking any trout species in streams containing species of concern, or at-risk species, until more is known about these interactions. Rainbow trout are still routinely stocked in streams with abundant native species. Examining interactions in these systems will lead to a better understanding of the ecological interactions between rainbow trout and longnose dace. Equally important, examining these interactions may also provide insight into the potential interactions among introduced trout and species of concern, and thus inform future management decisions involving stocking trout into streams with species of concern. The first step in understanding the complex interactions among these two species is to determine if there is a direct predatory threat of non-native rainbow trout to longnose 37 dace, and how that threat changes over time. Therefore, a series of in-stream enclosures were used to determine if the presence, density, or acclimation period of rainbow trout influenced the number of longnose dace remaining in enclosures. Prey size also likely affects escapement and predation rates. Therefore, the influence of longnose dace length on the probability that longnose dace escaped from control enclosures, and the probability that longnose dace survived in treatment enclosures was examined. We also examined the stomachs of rainbow trout three days after introduction into enclosures with longnose dace to determine if rainbow trout consumed longnose dace, and if rainbow trout density or length explained the presence of longnose dace in the stomachs of rainbow trout. METHODS Study Site Long Pine Creek is a second order, cold-water tributary to the Niobrara River, located on the edge of the Nebraska Sandhills and Northwestern Great Plains Ecoregions (Level III, US EPA). Long Pine Creek is Nebraska’s longest self-sustaining trout stream with approximately 30 km of trout-supporting water. It has historically been stocked with brook trout Salvelinus fontinalis, brown trout Salmo trutta, and rainbow trout. The study site was on private land approximately 5 km south of Long Pine, Nebraska. Brown trout and rainbow trout were prevalent during 2012 in the stream reach containing enclosures. Natural reproduction of both brown trout and rainbow trout was evident, as smaller than stocked size fish were collected. 38 Longnose dace are native to Long Pine Creek and were first recorded in the stream in 1939 (NGPC unpublished data). No longnose dace were collected from the study reach during 2012, although longnose dace were abundant just upstream of the study site during 2011 and records indicate they historically occupied the study site. Field Methods Twelve enclosures (1.5-m width x 3.0-m length x 0.9-m height) constructed of 2.54-cm PVC pipe and 0.6-cm hardware wire (Figure 3-1) were placed in Long Pine Creek during July and August 2013. Enclosures also included a fake undercut bank (0.6m width x 2.4-m length) made of landscaping fabric and PVC pipe that floated at the surface of the water and was anchored to one side of the enclosure. Longnose dace were collected from Plum Creek, Fairfield Creek, and Bone Creek, Brown County, Nebraska (Table 3-1) using a pulsed-DC backpack electrofisher and allowed to acclimate to Long Pine Creek in a temporary enclosure for 1 to 2 days. Five longnose dace (0.9 fish/m 2 ) were then weighed, measured, and randomly assigned to each enclosure. Longnose dace densities in enclosures were similar to previously reported natural densities of forage fish in Long Pine Creek (0.7 fish/m 2 , Klammer 1984). Abundance of longnose dace was monitored daily following introduction to ensure escapement was not possible. Additional longnose dace were added until all enclosures successfully held 5 longnose dace for 24 hours. All longnose dace in each round were from the same source stream. Following longnose dace introduction, a randomized complete block design (blocked by longitudinal position along the stream) was used to randomly assign rainbow trout density treatments to enclosures. Enclosures were blocked by longitudinal position 39 along the stream to account for any potential confounding effects based on spatial position along the stream. Rainbow trout density treatments were chosen to reflect a range of natural densities in Nebraska streams. Rainbow trout density treatments were 1) control (0 rainbow trout/enclosure), 2) average density (2 rainbow trout/enclosure), and 3) high density (4 rainbow trout/enclosure). The experiment was repeated four times (rounds) for a total of 16 replicates per treatment (4 within each round x 4 rounds; Figure 3-2). Rainbow trout were transported from Grove Trout Rearing Station, Antelope County, Nebraska and temporarily placed into two extra enclosures. Rainbow trout were allowed to acclimate for 3 to 22 days before being introduced into enclosures with longnose dace (Table 3-1). Following rainbow trout introduction, abundance of all fish was checked every 12 hours at approximately 0530 and 1730 (optimal light conditions) for 72 hours. Abundance was checked by quickly lifting enclosures from the stream and counting individuals. Dead, unconsumed longnose dace were immediately removed from enclosures and were not replaced during the experiment. All fish were removed from enclosures and euthanized after 72 hours. Rainbow trout stomachs were immediately checked for the presence of longnose dace and macroinvertebrates (with the exception that for round 1, only rainbow trout in enclosures with longnose dace missing were examined). Macroinvertebrates and other food items were recorded to determine if rainbow trout were feeding on alternative food sources. Water temperature, dissolved oxygen, and conductivity were measured in each enclosure once for each replicate. Depth and velocity was measured at three evenly spaced points along a transect positioned approximately 3.0 m upstream of enclosures, as 40 well as at 3 transects within the enclosures (front, middle, back) to determine if flows within enclosures were similar to natural conditions. Data Analysis Longnose Dace Mortality Generalized linear mixed models (PROC GLIMMIX, SAS v.9.2) were used to evaluate differences in the number of dace remaining in enclosures (Poisson distribution) between rainbow trout density treatments and number of days rainbow trout were allowed to acclimate to the stream prior to introduction into enclosures. Any change in the number of longnose dace remaining, relative to control enclosures, was assumed to be the direct result of rainbow trout treatments. Therefore, the number of longnose dace remaining in enclosures, relative to control enclosures, was assumed to be inversely correlated with mortality of longnose dace (i.e., a decrease in the number of dace remaining is representative of an increase in longnose dace mortality). Models included fixed effects of treatment (i.e., control, average, and high rainbow trout density), time (e.g., 12, 24, and 36 hours since stocking rainbow trout into enclosures), and rainbow trout acclimation period (i.e., 3, 4, 5, 11, and 22 days). Random effects included round, block, enclosure, enclosure (round*block), and time by enclosure (round*block). However, blocking by round, block, enclosure, and enclosure (round*block) accounted for very little variation and so were removed from the models. A first-order autoregressive covariance structure (AR1) and Kenward-Roger degrees of freedom correction were used to account for repeated measures in order to reduce the risk of a type I error. 41 Generalized linear models (PROC GLIMMIX, SAS v.9.2) were used to assess the influence of longnose dace length on the probability that longnose dace escaped (using only fish in control enclosures; binomial distribution), and the influence of longnose dace length on the probability that longnose dace survived in treatment enclosures (binomial distribution). The lengths of longnose dace present at the end of the experiment were measured and matched to pre-stocking lengths. Lengths of longnose dace that were not accounted for were determined via process of elimination. Generalized linear models were also used to evaluate the influence of rainbow trout density treatments on the number of unconsumed dead dace (Poisson distribution) removed from enclosures.