DEVELOPMENT OF A STANDARDIZED MONITORING PROTOCOL TO ASSESS THE EFFICACY OF NONNATIVE FISH SUPPRESSION IN THE LAMAR RIVER WATERSHED, YELLOWSTONE NATIONAL PARK by Keith David Wellstone A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Fish and Wildlife Management MONTANA STATE UNIVERSITY Bozeman, Montana July 2024 ©COPYRIGHT by Keith Wellstone 2024 All Rights Reserved ii ACKNOWLEDGMENTS I thank my advisor, Dr. Alexander Zale, for his unwavering support, patience, and guidance through this process. I also thank my committee members Todd Koel for giving me the tools and opportunities to advance my career, trusting me with this project, and working diligently to preserve native fishes in Yellowstone National Park, and Christopher Guy for encouraging me to think critically about fisheries science and always focus on the bigger picture. Jay Rotella helped me immensely by providing useful R code and explaining statistics in a way that even I could (somewhat) understand. Numerous biologists and technicians assisted with project planning and fieldwork. Without the help of Brian Ertel, Colleen Detjens, Andriana Puchany, and their crews, this research would have been impossible. My technicians Weston Neubauer and Morgan Krell were instrumental in the completion of this work; they will accomplish great things in their careers. The friends I have made in graduate school will be friends for life. Sam Fritz, Drew MacDonald, Mike Siemiantkowski, Jade Ortiz, Zach Maguire, Hayley Glassic, Ben Tumolo, Katie Furey, Robert Eckelbecker, Michelle Briggs, Colton Pipinich, Michael Throolin, Cody Vender, Kadie Heinle, Jose “Tosti” Sanchez, and many other friends made the past three years memorable and enjoyable. Last but certainly not least, I thank my family for their unwavering support, love, and encouragement throughout my life. Words cannot describe the gratitude I have towards all parties mentioned above. I use the term “we” in this thesis to acknowledge the collaborative effort required to carry out this work. iii TABLE OF CONTENTS 1. INTRODUCTION AND LITERATURE REVIEW ........................................................... 1 Freshwater Biological Invasions ......................................................................................... 1 Management of Nonnative Fish Populations ...................................................................... 2 Rainbow Trout .................................................................................................................... 3 Yellowstone Rocky Mountain Cutthroat Trout ................................................................... 4 Lamar River Watershed ...................................................................................................... 4 Project Goal ........................................................................................................................ 6 Literature Cited ................................................................................................................... 7 2. EVALUATING TROUT POPULATION MONITORING STRATEGIES IN A HIGH-ELEVATION, MEDIUM-SIZED RIVER SYSTEM: IMPLICATIONS FOR DETECTING TRENDS IN TROUT ABUNDANCE ................................................................................................................. 13 Introduction ....................................................................................................................... 13 Monitoring ............................................................................................................ 13 Components of Monitoring ................................................................................... 13 Abundance Estimation Methods ........................................................................... 15 Life-Stage Monitoring .......................................................................................... 16 Sampling Gear ...................................................................................................... 17 Objective ............................................................................................................... 18 Methods............................................................................................................................. 19 Study Area ............................................................................................................. 19 Pilot Snorkeling Study .......................................................................................... 21 Electrofishing CPUE ............................................................................................. 24 Abiotic Measurements .......................................................................................... 25 Mark Recapture ..................................................................................................... 25 Slough Creek ............................................................................................. 25 Lamar River. ............................................................................................. 26 Juvenile Sampling Pilot Study .............................................................................. 27 Slough Creek ............................................................................................. 27 Lamar River .............................................................................................. 27 Data Analysis ........................................................................................................ 28 Snorkel-Electrofishing Comparison Analysis ....................................................... 28 CPUE Analysis...................................................................................................... 29 Mark-Recapture Analysis...................................................................................... 29 Slough Creek. ............................................................................................ 29 Simulations ........................................................................................................... 30 CPUE. ....................................................................................................... 30 Mark Recapture. ........................................................................................ 31 Power Analyses ..................................................................................................... 33 iv TABLE OF CONTENTS CONTINUED CPUE. ....................................................................................................... 33 Mark Recapture. ........................................................................................ 34 Juvenile Trout Abundance Analysis ...................................................................... 35 Results ............................................................................................................................... 35 Snorkeling Pilot Study .......................................................................................... 35 Slough Creek ............................................................................................. 35 Lamar River .............................................................................................. 35 Snorkel-Electrofishing Comparison Analysis ....................................................... 36 Abiotic Measurements .......................................................................................... 36 CPUE Estimates .................................................................................................... 37 Slough Creek ............................................................................................. 37 Lamar River .............................................................................................. 37 Mark Recapture Estimates .................................................................................... 37 Slough Creek. ............................................................................................ 37 Lamar River. ............................................................................................. 38 Simulations ........................................................................................................... 38 CPUE ........................................................................................................ 38 Mark Recapture ......................................................................................... 38 Power Analyses ..................................................................................................... 38 Slough Creek ............................................................................................. 38 Lamar River .............................................................................................. 39 Juvenile Electrofishing Pilot Study ....................................................................... 39 Discussion ......................................................................................................................... 40 Snorkeling ............................................................................................................. 40 CPUE .................................................................................................................... 42 Slough Creek. ............................................................................................ 42 Lamar River. ............................................................................................. 43 Mark Recapture ..................................................................................................... 44 Power Analysis...................................................................................................... 44 Juvenile Sampling Feasibility ............................................................................... 45 Slough Creek. ............................................................................................ 45 Lamar River. ............................................................................................. 46 Management Recommendations ........................................................................... 46 Tables and Figures ............................................................................................................ 48 Literature Cited ................................................................................................................. 67 3. EVALUATION OF A FIELD-BASED HYBRID IDENTIFICATION KEY: IMPLICATIONS FOR NATIVE AND NONNATIVE TROUT MANAGEMENT IN YELLOWSTONE NATIONAL PARK .......................................... 76 Introduction ....................................................................................................................... 76 Questions............................................................................................................... 79 v TABLE OF CONTENTS CONTINUED Methods............................................................................................................................. 79 Study Area ............................................................................................................. 79 Sample Collection ................................................................................................. 80 Analysis................................................................................................................. 81 Discussion ......................................................................................................................... 82 Differentiating Between Individuals ≥ 100 mm with and without RBT Ancestry ........................................................................................................ 83 Differentiating Between Individuals < 100 mm with and without RBT Ancestry ........................................................................................................ 83 Differentiating Between RBT and CTX ............................................................... 84 Tables and Figures ............................................................................................................ 86 Literature Cited ................................................................................................................. 93 4. CONCLUSION: MONITORING AND MANAGEMENT RECOMMENDATIONS .................................................................................................. 96 Recommendations ............................................................................................................. 96 Juvenile Sampling ................................................................................................. 96 Adult Sampling ..................................................................................................... 97 Taxa Identification .................................................................................... 97 Snorkeling ................................................................................................. 97 CPUE ........................................................................................................ 97 Mark Recapture ......................................................................................... 98 Alternatives ....................................................................................................................... 98 Timing ................................................................................................................... 98 Electrofisher Configuration and Settings .............................................................. 99 Calculating Relative Taxa Proportion or the Ratio of Nonnative to Native Trout .......................................................................................................... 99 Literature Cited ............................................................................................................... 100 CUMULATIVE LITERATURE CITED ......................................................................... 101 vi LIST OF TABLES 1. Table 2.1 Sampling design and capture totals for mark-recapture surveys in Slough Creek, Yellowstone National Park, in 2022. Electrofishing events represent a single pass throughout the sampling area, and angling events represent a multi-day sampling effort throughout the sampling area. .............................. 48 2. Table 2.2. Catch-per-unit-effort (trout/100 m) summary statistics for Yellowstone Cutthroat Trout and nonnative trout in Slough Creek and the Lamar River from 2021 to 2022. Mean CPUE represents the mean across 18 sites in Slough Creek and 12 sites in the Lamar River. The SE represents the standard error of the CPUE estimates. ........................................................................ 48 3. Table 2.3. Coefficient of variation estimates for trout abundance estimates across simulated sampling designs. Sampling design 3A represents an all- angling approach, 2AE represents two angling events and one electrofishing event, 2EA represents two electrofishing events and one angling event, and EA represents a standard Lincoln-Peterson design with one electrofishing and one angling event. The CV denotes the coefficient of variation, and CI denotes 95% confidence intervals. .................................................................................... 49 4. Table 2.4. Estimated abundances, capture probabilities, and associated uncertainty (standard errors; SE) of juvenile trout in Slough Creek in 20-m long bank-habitat units extending 2 m perpendicular from the bank towards the main current. ............................................................................................................... 49 5. Table 3.1. Proportion of correct identifications of native and nonnative trout by length group and combined when differentiating among trout with and without Rainbow Trout ancestry in the Lamar River, Slough Creek, and Buffalo Creek, Yellowstone National Park in 2021. ......................................................... 86 6. Table 3.2. Sample sizes and proportions of correct and incorrect identifications among all trout phenotypes. Phenotypes identified represent YCT (Yellowstone Cutthroat Trout), RBT (Rainbow Trout), CTX-W (hybrids with white leading edges on the pelvic fins), CTX-H (hybrids with ≥ 6 head spots), CTX-WH (hybrids with both head spots and white fin tips), or CTX-S (hybrids exhibiting other traits suggesting introgression such as prominent pink coloration along the lateral line). Trout were collected in the Lamar River, Slough Creek, and Buffalo Creek in 2021. ................................................. 86 vii LIST OF TABLES CONTINUED 7. Table 3.3. Relative proportions (%) of field determinations by hybrid statuses (BC = backcrossed; F1 = first generation; F2 = second generation). Field identifications represent YCT (Yellowstone Cutthroat Trout), RBT (Rainbow Trout), CTX-W (hybrids with white leading edges on the pelvic fins), CTX-H (hybrids with ≥ 6 head spots), CTX-WH (hybrids with both head spots and white fin tips), or CTX-S (hybrids exhibiting other traits suggesting introgression such as prominent pink coloration along the lateral line). .................................................................................................................................. 87 viii LIST OF FIGURES 1. Figure 2.1. Map of the study area and section delineations of the Lamar River watershed, Yellowstone National Park. ................................................................... 50 2. Figure 2.2. Locations sampled by snorkeling in the Lamar River and Slough Creek, Yellowstone National Park. Sites in the Lamar River were sampled in 2021, and sites in Slough Creek were sampled in 2021 and 2022. We did not sample the Slough Creek site upstream of the Buffalo Creek confluence in 2022. ............................................................................................................................. 51 3. Figure 2.3. Snorkelers using a PVC pipe to space themselves equidistantly and maintain their snorkeling lanes during a survey in Slough Creek, Yellowstone National Park in 2021. ................................................................................. 52 4. Figure 2.4. Snorkelers drifting downstream to sample the Lamar River, Yellowstone National Park in 2021. ................................................................................. 53 5. Figure 2.5. Example of a juvenile sampling site in Slough Creek, Yellowstone National Park in 2021. ................................................................................. 54 6. Figure 2.6. Simulation procedures for determining the statistical power of sampling designs to detect declines in CPUE of Yellowstone Cutthroat and nonnative trout in the Lamar River and Slough Creek. In step one, a decline (e.g., 50% decline) was projected onto the bootstrapped, log-transformed CPUE estimates for each sampling design over a pre-specified time period (e.g., 10 years). The white circles represent the projected CPUE in year t (Nt). Step two projects stochasticity in the estimate based on the standard deviation of the initial CPUE and selects random points (black circles) from a normal distribution centered around the mean of the projected CPUE. Step three fits a linear regression line to the estimates. This process was repeated 10,000 times for each sampling design (number of sites) and combination of declines (e.g., 50% decline over 10 years as shown). The proportion of slopes that were negative and significant (P < 0.05) from the regressions represented our statistical power to detect a trend. This figure was adapted from Gibbs (1998) and Dauwalter (2009). ....................................................................... 55 7. Figure 2.7. Comparison of counts of trout taxa by electrofishing and snorkeling in Slough Creek in 2021 and 2022 (n=11; one site was omitted in 2022). Ctx (yellow) represents hybrid trout, yct (blue) represents Yellowstone Cutthroat Trout, and rbt (brown) represents Rainbow Trout........................ 56 ix LIST OF FIGURES CONTINUED 8. Figure 2.8. Predicted proportions of nonnative trout (Rainbow Trout and hybrids) sampled by electrofishing and angling in reference sites in Slough Creek, Yellowstone National Park in 2021 and 2022. Error bars represent 95% prediction intervals from the GLM. .......................................................................... 57 9. Figure 2.9. Changes in the CV of CPUE (trout/100 m) estimates for nonnative trout (circles) and Yellowstone Cutthroat Trout (triangles) in Slough Creek, Yellowstone National Park, when increasing numbers of sites are sampled. ...................................................................................................................... 58 10. Figure 2.10. Changes in the CV of CPUE (trout/100 m) estimates for nonnative trout (circles) and Yellowstone Cutthroat Trout (triangles) in the Lamar River, Yellowstone National Park, when increasing numbers of sites are sampled. ...................................................................................................................... 59 11. Figure 2.11. Monte Carlo simulation results for the probability of detecting (statistical power) 75% (A), 50% (B) and 25% (C) declines in Yellowstone Cutthroat Trout CPUE over 5- (squares), 10- (triangles), 15- (circles), and 25-year (asterisks) periods in Slough Creek. Results represent the proportion of simulated declines that were negative and significant (P ≤ 0.05). ................................................................................................................................. 60 12. Figure 2.12. Monte Carlo simulation results for the probability of detecting (statistical power) 75% (A), 50% (B) and 25% (C) declines in nonnative trout CPUE over 5- (squares), 10- (triangles), 15- (circles), and 25-year (asterisks) periods in Slough Creek. Results represent the proportion of simulated declines that were negative and significant (P ≤ 0.05). .................................... 61 13. Figure 2.13. Empirical (initial abundance) and simulated 50 and 75% declines in Yellowstone Cutthroat Trout abundance in Slough Creek. Fifty and 75% declines were simulated using the empirical estimates of abundance and capture probabilities from our electrofishing and angling mark-recapture study. Error bars represent 95% confidence intervals. ............................ 62 14. Figure 2.14. Monte Carlo simulation results for the probability of detecting (statistical power) 75% (A), 50% (B) and 25% (C) declines in Yellowstone Cutthroat Trout CPUE over 5- (gold squares), 10- (grey triangles), 15- (brown circles), and 25-year (blue asterisks) periods in the Lamar River. Results represent the proportion of simulated declines that were negative and significant (P ≤ 0.05). ................................................................................................. 63 x LIST OF FIGURES CONTINUED 15. Figure 2.15. Monte Carlo simulation results for the probability of detecting (statistical power) 75% (A), 50% (B) and 25% (C) declines in nonnative trout CPUE over 5- (gold squares), 10- (grey triangles), 15- (brown circles), and 25-year (blue asterisks) periods in the Lamar River. Results represent the proportion of simulated declines that were negative and significant (P ≤ 0.05). ................................................................................................................................. 64 16. Figure 2.16. Simulated 50% (triangles) and 75% (circles) declines in Yellowstone Cutthroat Trout abundance in the Lamar River across simulated sampling designs when compared to simulations informed by the empirical abundance estimate (squares). 3A represents an all-angling sampling design, 2AE represents two angling events and one electrofishing event, 2EA represents two electrofishing events and one angling event, and EA represents a standard Lincoln-Peterson design with one electrofishing and one angling event. Points (circles and triangles) represent the median abundance estimates from our simulations, and error bars represent the median upper and lower 95% confidence intervals. ......................................................... 65 17. Figure 2.17. Simulated 50% (triangles) and 75% (circles) declines in nonnative trout abundance in the Lamar River across simulated sampling designs when compared to simulations informed by the empirical abundance estimate (squares). 3A represents an all-angling sampling design, 2AE represents two angling events and one electrofishing event, 2EA represents two electrofishing events and one angling event, and EA represents a standard Lincoln-Peterson design with one electrofishing and one angling event. Points (circles and triangles) represent the median abundance estimates from our simulations, and error bars represent the median upper and lower 95% confidence intervals. ......................................................... 66 18. Figure 3.1. Example of the pictures taken of fish during sampling in Slough Creek, Yellowstone National Park in 2021. This trout was correctly identified as a “CTX-H”—a hybrid with 6 or more head spots but no white edges on the anal or pectoral fins. Genetic analysis confirmed this fish was a back-crossed Yellowstone Cutthroat Trout. ................................................................... 88 19. Figure 3.2. Length-frequency histogram of trout captured in lower Buffalo Creek by electrofishing. Age classes (ages 0, 1, and 2+) are defined by gaussian curves using a N-mixture model. Lengths are binned by 10-mm length intervals. Red lines indicate length cutoffs for age classes. Figure adapted from Puchany (2019). .......................................................................................... 89 xi LIST OF FIGURES CONTINUED 20. Figure 3.3. Hybrid status of individuals genotyped during our study. Rainbow Trout (RBT), BC (backcrossed) RBT, F1 (first generation hybrids), F2 (second generation hybrids), other (individuals with varying levels of introgression), BC Yellowstone Cutthroat Trout (YCT), and YCT are represented. ....................................................................................................................... 90 21. Figure 3.4. Predicted probabilities of correctly identifying trout > 100 mm of different genetic statuses. Genotypes represented include Rainbow Trout (RBT), BC (backcrossed) RBT, F1 (first generation hybrids), F2 (second generation hybrids), other (individuals with varying levels of introgression), BC Yellowstone Cutthroat Trout (YCT), and non-hybridized YCT. Error bars represent 95% prediction intervals. ........................................................................... 91 22. Figure 3.5. A first-generation hybrid that was incorrectly identified as a Yellowstone Cutthroat Trout because of the lack of both white leading edges on the anal and pelvic fins and head spots. ....................................................................... 92 23. Figure 3.6. A back-crossed Yellowstone Cutthroat Trout that was incorrectly identified as a Yellowstone Cutthroat Trout because of the lack of both white leading edges on the anal and pelvic fins and head spots. ................................................ 93 xii ABSTRACT Hybridization between native Cutthroat Trout and introduced Rainbow Trout is pervasive throughout western North America, and this hybridization has resulted in reduced abundances and range contractions of native Cutthroat Trout subspecies. In the Lamar River watershed in Yellowstone National Park, Yellowstone Cutthroat Trout × Rainbow Trout hybrids are abundant in the lower Lamar River watershed because of past stocking efforts. To mitigate the threat of hybridization in the Lamar River watershed, the National Park Service has acted to remove Rainbow Trout and hybrids and block the upstream movement of these nonnative taxa into the upper watershed. A standardized monitoring protocol is desired to assess the response of fish populations to these management actions and to monitor existing populations of Yellowstone Rocky Mountain Cutthroat Trout. We evaluated the efficacy of electrofishing, snorkeling, and angling to estimate the absolute abundances and catch-per-unit-effort of mixed-stock aggregations of trout in main-stem locations. We also evaluated the efficacy of targeted, juvenile sampling to estimate the abundances of juvenile trout of each taxon. We also assessed the accuracy and limitations of a field-identification key developed for differentiating among Yellowstone Rocky Mountain Cutthroat Trout, Rainbow Trout, and Yellowstone Rocky Mountain Cutthroat × Rainbow Trout hybrids in the Lamar River watershed. We observed agreement between the proportion of nonnative trout sampled by electrofishing and snorkeling in Slough Creek. Catch-per-unit effort estimates were highly variable for Yellowstone Rocky Mountain Cutthroat and nonnative trout, but error was reduced when minimum section lengths of 3,600 m in Slough Creek and 2,200 m in the Lamar River were sampled. In both streams, error in mark-recapture estimates was reduced when angling was incorporated. Statistical power to detect declines in Yellowstone Cutthroat and nonnative trout abundance was low for CPUE and mark- recapture surveys unless 50% declines occurred for both taxa. Electrofishing surveys are not feasible for estimating the abundance of juvenile trout because of high error in taxa identification (30% correct-identification rate). However, we differentiated among adult fish with high rates of accuracy (98%). These results will guide future long-term monitoring of trout populations in the Lamar River watershed. 1 CHAPTER ONE INTRODUCTION AND LITERATURE REVIEW Freshwater Biological Invasions Human-facilitated biological invasions are prevalent worldwide (Vitousek et al. 1997) and have resulted in novel species assemblages with reduced numbers of native species (Strayer 2010). Fresh water composes only 0.8% of the surface of the Earth, yet almost 6% of all described species are supported by freshwater ecosystems (Dudgeon et al. 2006). Anthropogenic activities have reduced freshwater biodiversity at a rapid rate (Albert et al. 2021), with the intentional and unintentional introduction of nonnative fishes as a significant contributing factor (Gozlan et al. 2010). Intentional introductions are often intended to enhance and diversify recreational angling opportunities (Pister 2001), create or improve aquaculture operations (Kerr et al. 2005), control undesired species by biological control (Kumar and Hwang 2006), or are the result of aquarium releases (Strecker et al. 2011). Unintentional introductions are often the result of aquaculture or baitfish escapement (Ludwig and Leitch 1996; Glover et al. 2017) and are facilitated by human transportation and the construction of commercial waterways (Jacobs and Keller 2017). Despite the motives (or lack thereof) for nonnative fish introductions, these actions can result in the establishment of self-sustaining, invasive populations. Established, invasive populations can drastically change aquatic biological communities by direct and indirect means, causing damage to ecosystems and economies (Koel et al. 2005; Pimentel et al. 2005). These introductions, along with other stressors including habitat degradation and fragmentation, climate change, and exploitation, have contributed to the decline of native aquatic species and 2 subsequent homogenization of aquatic species assemblages across global, aquatic landscapes (Rahel 2000; Dudgeon et al. 2006). Management of Nonnative Fish Populations Fisheries managers have progressively acknowledged the ecological value of preserving native fishes (Rahel 1997), and many state and federal fish and wildlife programs are tasked with preserving their long-term persistence. Common tools for managing nonnative fishes include isolation, containment, eradication, and suppression of established populations (Rahel and Smith 2018). Isolation is often achieved by constructing artificial barriers to prevent the movement of nonnative fish from nearby source populations (Thompson and Rahel 1998). Similarly, containment of nonnative populations often includes exclusion by barrier construction or other means of deterring movement (Clarkson 2004). Eradication is typically achieved using chemical piscicides (antimycin and rotenone) and is often used in conjunction with isolation strategies (Buktenika et al. 2013). Eradication of nonnative fish populations is also possible by mechanical means such as repeated electrofishing or gillnetting in small lakes or streams with limited habitat complexity (Knapp and Matthews 1998; Pacas and Taylor 2015); however, established populations are notoriously difficult to eradicate in large lakes and interconnected, fluvial systems (Simberloff 2014; Shepard et al. 2014). Mechanical reduction is an alternative strategy when eradication is too costly or improbable because of the complexity or size of the system or its proximity to other nonnative source populations (Koel et al. 2020). These reduction, eradication, and isolation strategies are commonly used in the western United States, where native Cutthroat Trout Oncorhynchus clarkii subspecies are threatened by predation, 3 displacement, competition, and hybridization with nonnative fishes (Al-Chokhachy et al. 2014; Meyer et al. 2017; Kovach et al. 2018). Rainbow Trout Rainbow Trout O. mykiss have been stocked globally for recreational purposes since the late 1800s and are one of the most widely introduced fishes in the world (Halverson 2010). Rainbow Trout are popular sport fish, but their introductions adversely affect native, freshwater species. For example, introduced Rainbow Trout displace and alter the distributions of amphibians in ponds and lakes (Hecnar and M’Closkey 1997; Knapp and Matthews 2000) and prey upon, compete with, and displace native fishes (McIntosh 2000; Seiler and Keeley 2009; Morita 2018). Rainbow Trout threaten the long-term persistence of native Westslope Cutthroat Trout O. lewisi and Yellowstone Rocky Mountain Cutthroat Trout O. virginalis by hybridization in the western United States (Kruse et al. 2000; Peacock and Kirchoff 2004; Gresswell 2011). Cutthroat and Rainbow Trout are closely related and often exhibit spatial and temporal reproductive overlap (Muhlfeld et al. 2009b) facilitating hybridization between the two taxa. This reproductive sympatry has resulted in the loss of locally adapted gene complexes and genetic diversity (Kovach et al. 2015), reduced fitness (Muhlfeld et al. 2009a), altered life-history expression and growth rates (Strait et al. 2020; Bourett et al. 2022), and in some cases, the genomic extinction of native Cutthroat Trout subspecies (Allendorf and Leary 1988). 4 Yellowstone Rocky Mountain Cutthroat Trout The Yellowstone Rocky Mountain Cutthroat Trout (hereafter Yellowstone Cutthroat Trout), a subspecies of Rocky Mountain Cutthroat Trout native to the Intermountain West, historically occupied the upper Snake and Yellowstone River watersheds (Behnke 2002). The Yellowstone Cutthroat Trout has been widely stocked outside of its native range for recreational purposes; however, habitat fragmentation and degradation, climate change, exploitation, and invasive species introductions (including hybridization; Gresswell 2011) have contributed to the reduced abundance and distribution of the subspecies. The subspecies currently occupies 43% of its native range (Endicott et al. 2016), with only 23% of the range occupied by non-hybridized populations (Endicott et al. 2016). Yellowstone National Park is at the center of the native range of the Yellowstone Cutthroat Trout and supports genetically unaltered and economically important populations of the subspecies (Al-Chokhachy et al. 2018). Despite the protected and relatively undisturbed status of aquatic habitats in this National Park, Yellowstone Cutthroat Trout populations are threatened by competition and displacement by nonnative Brook Trout Salvelinus fontinalis (Ertel et al. 2017), predation by nonnative Lake Trout Salvelinus namaycush (Ruzycki et al. 2003), and hybridization with nonnative Rainbow Trout (Heim et al. 2020b). Lamar River Watershed The Lamar River watershed was formerly a fluvial stronghold for genetically unaltered Yellowstone Cutthroat Trout. However, the National Park Service and other agencies intentionally stocked Rainbow Trout in the Lamar River watershed in the early 1900s to diversify sportfishing opportunities for visiting anglers (Varley 1981). Some of these Rainbow 5 Trout were stocked into the historically fishless headwaters of Buffalo Creek, a large tributary of Slough Creek in the Absaroka-Beartooth Wilderness of Montana. By the 1930s, the National Park Service shifted its focus from stocking nonnative fishes towards native fish conservation, adopting a policy “prohibiting the introduction of exotic species in national park or monument waters now containing only native species” (Madsen 1937). Although stocking of Rainbow Trout ceased nearly a century ago, legacy populations persist, especially in the headwater streams and lakes of the Buffalo Creek watershed. These fish continue to invade downstream reaches of Slough Creek and the Lamar River where they are hybridizing with native Yellowstone Cutthroat Trout. Rainbow Trout × Yellowstone Cutthroat Trout hybrids (hereafter “hybrids”) are now abundant in the lower Lamar River watershed, and because of the seasonal, fluvial connectivity of the system, appear to be invading the upper watershed where Yellowstone Cutthroat Trout populations of high conservation priority are present (Ertel 2017; Al-Chokhachy et al. 2018; Heim 2019). In addition to its conservation value, the watershed is one of the most popular angling destinations in the park, with over 10,000 anglers from around the world visiting the watershed annually to catch Yellowstone Cutthroat Trout (Heim et al. 2020a). The National Park Service is attempting to curtail the spread of invasive Rainbow Trout in the lower Lamar River watershed to slow the rate of hybridization in the upper watershed (Ertel et al. 2017). It has conducted single- and multiple-pass electrofishing to selectively remove Rainbow Trout and hybrids from the middle Lamar River and upper Slough Creek since 2013. Mandatory-kill angling regulations for Rainbow Trout and hybrids were instituted in 2014. A barrier was also constructed in lower Slough Creek in 2017 to slow upstream movement by Rainbow Trout and protect Yellowstone Cutthroat Trout populations in the three upper meadows 6 (Ertel et al. 2017). However, an overflow channel near the barrier reopened in a 2020 flood event; [K. Wellstone, personal observation]; the effects of this failure in facilitating further invasion by Rainbow Trout is unknown. Buffalo Creek was recently identified as the primary source of Rainbow Trout invading the Lamar River watershed (Heim et al. 2020b). The National Park Service and partner agencies plan to eradicate the Rainbow Trout population from Buffalo Creek in 2024 and 2025 using rotenone—a chemical piscicide. Following the rotenone treatment, genetically unaltered Yellowstone Cutthroat Trout from nearby fluvial populations are to be stocked in Buffalo Creek. Fish suppression and eradication efforts require quantitative measures of native and invasive fish population abundance over time to assess their efficacy (Syslo et al. 2016; Budy et al. 2020; Koel et al. 2020) but pilot studies may be required to assess the precision and bias of such measures in the absence of prior monitoring. Additionally, evaluating tradeoffs in resource expenditures versus precision and bias is necessary to inform feasible, long-term monitoring (Quist et al. 2006; Al-Chokhachy et al. 2009; Dauwalter et al. 2010). Project Goal The National Park Service desires a strategy for monitoring the success of nonnative Rainbow Trout and hybrid control actions in the Lamar River watershed. Specifically, the National Park Service desires a monitoring program that can detect whether their management actions will have reduced the abundance of Rainbow and hybrid trout, or increased the abundance of native Yellowstone Cutthroat Trout, in the lower Lamar River and Slough Creek. Heretofore, the study area had not been effectively sampled in a standardized way (NPS, 7 unpublished data); therefore, our goal was to determine the sampling gears, methods, and amount of effort needed to estimate the abundances of Yellowstone Cutthroat, Rainbow, and hybrid trout in the lower Lamar River watershed to be able to adequately detect changes in abundance resulting from the suppression actions. In chapter 2, we describe our evaluation of the efficacy of electrofishing, snorkeling, and angling to estimate the absolute abundances and catch-per-unit- effort of mixed-stock aggregations of trout in main-stem locations. We also evaluate the efficacy of targeted, juvenile sampling to estimate the abundances of juvenile trout of each taxon. In Chapter 3, we describe our evaluation of a field-identification key specifically designed for the Lamar River by determining the rate of correctly identifying trout taxa. In chapter 4, we make recommendations to aid National Park Service monitoring and management efforts in the watershed. Literature Cited Al-Chokhachy, R., C. C. Muhlfeld, M. C. Boyer, L. A. Jones, A. Steed, and J. L. Kershner. 2014. Quantifying the effectiveness of conservation measures to control the spread of anthropogenic hybridization in stream salmonids: a climate adaptation case study. North American Journal of Fisheries Management 34:642–652. Al-Chokhachy, R., B. B. Shepard, J. C. Burckhardt, D. Garren, S. Opitz, T. M. Koel, L. Nelson, and R. E. Gresswell. 2018. A portfolio framework for prioritizing conservation efforts for Yellowstone Cutthroat Trout populations. Fisheries 43:485–496. Albert, J. S., G. Destouni, S. M. Duke-Sylvester, A. E. Magurran, T. Oberdorff, R. E. Reis, K. O. Winemiller, and W. J. Ripple. 2021. Scientists’ warning to humanity on the freshwater biodiversity crisis. Ambio 50:85–94. Allendorf, F. W., and R. F. Leary. 1988. Conservation and distribution of genetic variation in a polytypic species, the Cutthroat Trout. Conservation Biology 2:170–184. Behnke, R. J. 2002. Trout and salmon of North America. Chanticleer Press, New York. 8 Bourret, S. L., R. P. Kovach., T. J. Cline., J. T. Strait, and C. C. Muhlfeld. 2022. High dispersal rates in hybrids drive expansion of maladaptive hybridization. Proceedings of the Royal Society B 289:1986 Buktenica, M. W., D. K. Hering, S. F. Girdner, B. D. Mahoney, and B. D. Rosenlund. 2013. Eradication of nonnative Brook Trout with electrofishing and antimycin-A and the response of a remnant Bull Trout population. North American Journal of Fisheries Management 33:117–129. Clarkson, R. W. 2004. Effectiveness of electrical fish barriers associated with the central Arizona project. North American Journal of Fisheries Management 24:94–105. Dudgeon, D., A. H. Arthington, M. O. Gessner, Z. Kawabata, D. J. Knowler, C. Lévêque, R. J. Naiman, A. Prieur-Richard, D. Soto, M. L. Stiassny, and C. A. Sullivan. 2006. Freshwater biodiversity: importance, threats, status, and conservation challenges. Biological Reviews 81:163–182. Endicott, C., L. Nelson, S. Opitz, A. Peterson, J. Burckhardt, S. Yekel, D. Garren, T. M. Koel, and B. B. Shepard. 2016. Rangewide status assessment for Yellowstone Cutthroat Trout (Oncorhynchus clarkii bouvieri): 2012. Yellowstone Cutthroat Trout Interagency Coordination Group, Helena, Montana. Ertel, B. D., K. C. Heim, J. L. Arnold, C. R. Detjens, and T. M. Koel. 2017. Preservation of native Cutthroat Trout in northern Yellowstone. Yellowstone Science 25:35–41. Glover, K. A., M. F. Solberg, P. McGinnity, K. Hindar, E. Verspoor, M. W. Coulson, M. M. Hansen, H. Araki, O. Skaala, and T. Svasand. 2017. Half a century of genetic interaction between farmed and wild Atlantic Salmon: status of knowledge and unanswered questions. Fish and Fisheries 18:890–927. Gozlan, R. E., R. Britton, I. G. Cowx, and G. H. Copp. 2010. Current knowledge on non-native freshwater introductions. Journal of Fish Biology 76:751–786. Gresswell, R. E. 2011. Biology, status, and management of the Yellowstone Cutthroat Trout. North American Journal of Fisheries Management 21:782–812. Halverson, M. A. 2010. An entirely synthetic fish: how Rainbow Trout beguiled America and overran the world. Yale University Press, New Haven, Connecticut. Hecnar, S. J., and R. T. M’Closkey. 1997. The effects of predatory fish on amphibian species richness and distribution. Biological Conservation 79:123–131. 9 Heim, K. C. 2019. Mechanisms of Rainbow Trout hybridization with native Cutthroat Trout in the Lamar River of Yellowstone National Park. Doctoral dissertation. Montana State University, Bozeman. Heim, K. C., T. E. McMahon, B. D. Ertel, and T. M. Koel. 2020a. Leveraging public harvest to reduce invasive hybridization in Yellowstone National Park: field identification and harvest of Cutthroat × Rainbow Trout hybrids. Biological Invasions 22:2685–2698. Heim, K. C., T. E. McMahon, S. T. Kalinowski, B. D. Ertel, and T. M. Koel. 2020b. Abiotic conditions are unlikely to mediate hybridization between invasive Rainbow Trout and native Yellowstone Cutthroat Trout in a high-elevation metapopulation. Canadian Journal of Fisheries and Aquatic Sciences 77:1433–1445. Jacobs, A. I., and R. P. Keller. 2017. Straddling the divide: invasive aquatic species in Illinois and movement between the Great Lakes and Mississippi basins. Biological Invasions 19:635–646. Kerr, S. J., C. S. Brousseau, and M. Muschett. 2005. Invasive aquatic species in Ontario: a review and analysis of potential pathways for introduction. Fisheries 30:21–30. Knapp, R. A., and K. Matthews. 2000. Non-native fish introduction and the decline of the mountain yellow-legged frog from within protected areas. Conservation Biology 14:428– 438. Knapp, R. A., and K. R. Matthews. 1998. Eradication of nonnative fish by gill netting from a small mountain lake in California. Restoration Ecology 6:207–213. Koel, T. M., J. L. Arnold, P. E. Bigelow, T. O. Brenden, J. D. Davis, C. R. Detjens, P. D. Doepke, B. D. Ertel, H. C. Glassic, R. E. Gresswell, C. S. Guy, D. J. MacDonald, M. E. Ruhl, T. J. Stuth, D. P. Sweet, J. M. Syslo, N. A. Thomas, L. M. Tronstad, P. J. White, and A. V. Zale. 2020. Yellowstone Lake ecosystem restoration: a case study for invasive fish management. Fishes 5:18. Koel, T. M., P. E. Bigelow, P. D. Doepke, B. D. Ertel, and D. L. Mahony. 2005. Nonnative Lake trout result in Yellowstone Cutthroat Trout decline and impacts to bears and anglers. Fisheries 30:10–19. Kovach, R. P., R. Al-Chokhachy, and T. Stephens. 2018. Proactive Rainbow Trout suppression reduces threat of hybridization in the upper Snake River basin. North American Journal of Fisheries Management 38:811–819. 10 Kovach, R. P., C. C. Muhlfeld, M. C. Boyer, W. H. Lowe, F. W. Allendorf, and G. Luikart. 2015. Dispersal and selection mediate hybridization between a native and invasive species. Proceedings of the Royal Society B 282:20142454. Kruse, C. G., W. A. Hubert, and F. J. Rahel. 2000. Status of Yellowstone Cutthroat Trout in Wyoming waters. North American Journal of Fisheries Management 20:693–705. Kumar, R., and J. S. Hwang. 2006. Larvicidal efficiency of aquatic predators: a perspective for mosquito biocontrol. Zoological Studies 45:447–466. Ludwig, H. R., Jr., and J. A. Leitch. 1996. Interbasin transfer of aquatic biota via anglers’ bait buckets. Fisheries 21:14–18. Madsen, D. H. 1937. Protection of native fishes in the national parks. Transactions of the American Fisheries Society 66:395–397. McIntosh, A. R. 2000. Habitat- and size-related variations in exotic trout impacts on native galaxiid fishes in New Zealand streams. Canadian Journal of Fisheries and Aquatic Sciences 57:2140–2151. Meyer, K. A., P. Kennedy, B. High, M. R. Campbell. 2017. Purifying a Yellowstone Cutthroat Trout stream by removing Rainbow Trout and hybrids via electrofishing. Transactions of the American Fisheries Society 146:1193–1203. Morita, K. 2018. Assessing the long-term causal effect of trout invasion on a native charr. Ecological Indicators 87:189–192. Muhlfeld, C. C., S. T. Kalinowski, T. E. McMahon, M. L. Taper, S. Painter, R. Leary, and F. W. Allendorf. 2009a. Hybridization rapidly reduces fitness of a native trout in the wild. Biology Letters 5:328–331. Muhlfeld, C. C., T. E. McMahon, D. Belcer, and J. L. Kershner. 2009b. Spatial and temporal spawning dynamics of native Westslope Cutthroat Trout, Oncorhynchus clarkii lewisi, introduced Rainbow Trout, O. mykiss, and their hybrids. Canadian Journal of Fisheries and Aquatic Sciences 66:1153–1168. Pacas, C., and M. K. Taylor. 2015. Nonchemical eradication of an introduced trout from a headwater complex in Banff National Park, Canada. North American Journal of Fisheries Management 35:748-754. Peacock, M., and V. Kirchoff. 2004. Assessing the conservation value of hybridized Cutthroat Trout populations in the Quinn River drainage, Nevada. Transactions of the American Fisheries Society 133:309–325. 11 Pimentel, D., R. Zuniga, and D. Morrison. 2005. Update on the environmental and economic costs associated with alien-invasive species in the United States. Ecological Economics 52:273–288. Pister, E. P. 2001. Wilderness fish stocking history and perspective. Ecosystems 4:279–286. Rahel, F. J. 1997. From Johnny Appleseed to Dr. Frankenstein: changing values and the legacy of fisheries management. Fisheries 22:8–9. Rahel, F. J. 2000. Homogenization of fish faunas across the United States. Science 288:854–856. Rahel, F. J., and M. A. Smith. 2018. Pathways of unauthorized fish introductions and types of management responses. Hydrobiologia 817:41–56. Roussel, J. M., A. Haro, and R. A. Cunjak. 2000. Field test of a new method for tracking small fishes in shallow rivers using passive integrated transponder (PIT) technology. Canadian Journal of Fisheries and Aquatic Sciences 57:1326–1329. Ruzycki, J. R., D. A. Beauchamp, and D. L. Yule. 2003. Effects of introduced Lake Trout on native Cutthroat Trout in Yellowstone Lake. Ecological Applications 13:23–37. Simberloff, D. 2014. Biological invasions: what’s worth fighting and what can be won? Ecological Engineering 65:112–121. Shepard, B. B., L. M. Nelson, M. L. Taper, and A. V. Zale. 2014. Factors influencing successful eradication of nonnative Brook Trout from four small Rocky Mountain streams using electrofishing. North American Journal of Fisheries Management 34:988–997. Strait, J. T., L. A. Eby, R. P. Kovach, C. C. Muhlfeld, and M. C. Boyer. 2020. Hybridization alters growth and migratory life history expression of native trout. Ecological Applications 14:821–833. Strayer, D. L. 2010. Alien species in fresh waters: ecological effects, interactions with other stressors, and prospects for the future. Freshwater Biology 55:152–174. Strecker, A. L., P. M. Campbell, and J. D. Olden. 2011. The aquarium trade as an invasion pathway in the Pacific Northwest. Fisheries 36:74–85. Thompson, P. D., and F. J. Rahel. 1998. Evaluation of artificial barriers in small Rocky Mountain streams for preventing the upstream movement of Brook Trout. North American Journal of Fisheries Management 18:206–210. 12 Varley, J. D. 1981. A history of fish stocking activities in Yellowstone National Park between 1881 and 1980. Information paper 35. Department of the Interior, National Park Service, Yellowstone National Park, Wyoming. Vitousek, P. M., C. M. D’Antonio, L. L. Loope, M. Rejmánek, and R. Westbrooks. 1997. Introduced species: A significant component of human-caused global change. New Zealand Journal of Ecology 21:1–16. 13 CHAPTER TWO EVALUATING TROUT POPULATION MONITORING STRATEGIES IN A HIGH- ELEVATION, MEDIUM-SIZED RIVER SYSTEM: IMPLICATIONS FOR DETECTING TRENDS IN TROUT ABUNDANCE Introduction Monitoring Monitoring is an essential component of fisheries management. Fisheries biologists often implement monitoring programs to estimate trends in the abundances and distributions of fish species of interest (Pope et al. 2010). Repeated measures of these estimates across time, space, or both can be compared to determine whether management actions such as habitat restoration, changes to harvest regulations, and invasive species suppression or eradication are having the desired effect on the population(s) of interest (Noble et al. 2007). Assessing the effectiveness of invasive species suppression and eradication efforts usually requires quantification of native and invasive fish abundances, or indices of abundances, over time (Syslo et al. 2016; Budy et al. 2020; Koel et al. 2020) and associated precision, bias, and statistical power to detect trends (Quist et al. 2006; Al-Chokhachy et al. 2009; Dauwalter et al. 2010; Dzul et al. 2013). Components of Monitoring Statistical power (1-β) is defined as the probability of correctly rejecting a null hypothesis; that is, the probability of rejecting the null hypothesis when it is truly false (probability of not committing a type-II error; β = type-II error). In the context of population 14 monitoring, statistical power determines the probability of the study design to detect changes in the population when they occur (Wagner et al. 2013). Components that affect statistical power include sampling design, effect size, sample size, and precision (Brown and Guy 2007). Sampling design is the structure of how the samples are collected, effect size is the desirable level of change between response or among response variables, sample size is the number of sampling units (e.g., sites or individuals) included in the study, and precision is the level of variability observed (i.e., the level of agreement among repeated samples) in the estimates (Gotelli and Ellison 2013). Natural demographic fluctuations of fish populations, sampling error, and sampling variation influence precision in a data set (Allen and Hightower 2010). Sampling error is often influenced by gear efficiency, crew experience, and the abundance estimator used in the study (Hangsleben et al. 2012; Gibson-Reinemer et al. 2016). Sampling variation is influenced by spatial and temporal heterogeneity in sampling units (multiple sites) or temporal heterogeneity in the sampling unit (single site). Among multiple sites, variation in parameter estimates (e.g., catch rates) and the abundance estimator used may lead to high variability across sites, leading to low precision (Hangsleben et al. 2012). At a single site, temporal heterogeneity in catch rates and the variance of the abundance estimator used may affect precision similarly (Dauwalter et al. 2009). Sampling error often masks fluctuations in fish populations, further confounding the difficulty and cost of detecting trends in population parameters (Ham and Pearsons 2000). 15 Abundance Estimation Methods Estimating abundance is often a primary objective for management agencies and researchers because abundance reveals the outcome of dynamic rate functions such as mortality and recruitment (Pope et al. 2010). Mark recapture is a commonly applied method for estimating the absolute abundances of fish populations (Otis et al. 1978). However, this estimator requires repeat sampling events to achieve precise estimates and is therefore often costly (Russell et al. 2012). Catch-per-unit-effort (CPUE) is an index of abundance commonly used to monitor fish populations and assemblages. Indices typically require less effort than absolute abundance estimates, allowing for increased replication and expansion of the geographic scope of the study. For example, reducing effort by using indices can be a cost-effective way of sampling more sites to reduce the variance of estimates for more effective trend detection (Mitro and Zale 2000; Hedger et al. 2013; Hanks et al. 2018). However, indices of abundance are inherently biased, because they do not allow for the estimation of capture probability. Consequently, the use of abundance indices assumes that changes in CPUE are proportional to changes in the absolute abundance of the population of interest (Hubert and Fabrizio 2007). Capture probabilities often change depending on the field conditions, sampling gears used, and personnel experience (Thompson and Rahel 1996; Rosenberger and Dunham 2005; Simonson et al. 2022); therefore, efforts to standardize sampling such as sampling during the same field conditions, using the same gears, and providing proper personnel training are essential to drawing the correct inference when using indices of abundance (Curry et al. 2009). Inland stream salmonid monitoring programs typically standardize sampling efforts by conducting surveys during base- discharge levels (spring, summer, and autumn, depending on the objective, focal species, or flow 16 regime) at the same time each year to avoid changes in environmental conditions and fish behavior. These fisheries are often sampled at either a single site or a network of sites to estimate the absolute abundance or CPUE of the salmonid populations present. Estimation at a single site may be used when the site is representative of the area of interest and is often favored when the location of the site is cost-effective and feasible given the access or conditions of the reach. Multiple sites are often sampled using a probabilistic approach such as random or systematic sampling when managers want to infer results to a large area of stream or watershed (Larson et al. 2001). Life-Stage Monitoring Monitoring programs often focus on adult fish abundance or CPUE in mixed-stock aggregations of main-stem rivers to make inferences about the status or trend of a fishery. However, assessing multiple life stages of fish populations can elucidate more detailed information on population dynamics. The abundance of age-0 (hereafter “juvenile”) salmonids is a measure of the number of individuals successfully hatched each year and can be used to determine spawning success, predict future recruitment to a population, and evaluate the population status of species of interest (Allen and Hightower 2010). This can be particularly important when applied concurrently with nonnative species suppression programs because it allows agencies and researchers to make early predictions about the success of their management efforts (Caudron and Champigneulle 2010; Kanno et al. 2016). Furthermore, in salmonid metapopulations, juveniles often represent distinct populations rather than mixed-stock aggregations, allowing for more targeted population assessments (Cegelski et al. 2006; Uthe et 17 al. 2016). However, estimating juvenile abundance is often difficult because of low capture probabilities and patchy distribution of juvenile fish in river systems (Korman et al. 2010). Estimating juvenile salmonid abundance in medium and large rivers (> 6 m wide) confounds these difficulties because of the ineffectiveness of traditional sampling methods for sampling juvenile fish (Hayes and Baird 1994; Hitt et al. 2020). Targeted juvenile sampling may be an alternative or supplemental option to monitoring adult abundance to expand the scope of inference of a monitoring program, if feasible. Sampling Gear Conducting a pilot study to identify locations where certain gears can be used for abundance and CPUE estimation is an essential step in the monitoring process (Noble et al. 2007). Stream-sampling gears are often limited by water depth, velocity, and accessibility (Bonar et al. 2009). In systems that are difficult or costly to access with certain sampling gears, the evaluation of methods for sampling fish is particularly important for developing effective monitoring programs (Korman et al. 2010; Neebling and Quist 2011; Pregler et al. 2015). Electrofishing, angling, and snorkeling are commonly used for describing the distribution and abundances of salmonids in clear, coldwater streams in the Intermountain West (Thurow et al. 2006; O’Neal 2007; Meyer et. Al. 2022); however, the applicability and effectiveness of these gears may only be feasible in certain locations. Multiple gears are often used concurrently to sample fish populations in large river systems where conditions are not suitable to a single gear (Zajicek and Wolter 2018; Dunn and Paukert 2020). Angling has been used by agencies to estimate the abundance of fishes, particularly in locations where traditional data collection (e.g., 18 electrofishing) is costly or not feasible (Zubik and Fraley 1988; Vidal et al. 2018), and citizen participation is often leveraged to reduce the cost of these efforts (Williams et al. 2015; Mycko et al. 2018; Oremland et al. 2022). A unique opportunity exists in Yellowstone National Park to leverage the support of volunteer anglers for structured data collection (Detjens et al. 2017). Therefore, incorporating the use of angling data for estimating abundance of Yellowstone Cutthroat, Rainbow, and hybrid trout in main-stem locations was an important aspect of our study. Objective The objective of our study was to determine a methodology and amount of effort needed to detect changes in the abundances of Yellowstone Rocky Mountain Cutthroat Trout O. virginalis bouvieri (hereafter Yellowstone Cutthroat Trout), Rainbow Trout O. mykiss, and Yellowstone Cutthroat × Rainbow Trout hybrids (hereby “hybrids”) in the lower Lamar River watershed. We started by evaluating the use of snorkeling and targeted juvenile electrofishing as potential monitoring methods. Next, we compared snorkeling and electrofishing estimates from paired sampling efforts in locations where snorkeling was feasible. Finally, we evaluated the precision of sampling designs implemented with electrofishing and angling gear for estimating the CPUE and absolute abundance of trout taxa in Slough Creek and the Lamar River and estimated the amount of effort needed to detect declines in the abundance of the taxa with sufficient statistical power. 19 Methods Study Area The Lamar River watershed in Yellowstone National Park originates in the Absaroka- Beartooth Mountains and has a drainage basin of 1,731 km2. The Lamar River flows for 78 km from its headwaters before reaching its confluence with the Yellowstone River. Most of the watershed is federally protected by the National Park Service and U.S. Forest Service. For this study, we refer to the sections of the Lamar River watershed (Figure 2.1) as (1) the upper watershed, encompassing all streams above the Lamar River canyon, (2) the lower Lamar River, encompassing the section of the Lamar River below the downstream end of the Lamar River Canyon to the confluence of the Yellowstone and Lamar rivers, (3) upper Slough Creek (a tributary to the Lamar River), encompassing everything above the Slough Creek barrier, (4) lower Slough Creek, encompassing everything below the Slough Creek barrier to the Lamar River and Slough Creek confluence, and (5) Buffalo Creek (a tributary to Slough Creek). The upper and lower Lamar River watersheds are separated by the Lamar River canyon, a hypothesized seasonal fish barrier (Figure 2.1). Yellowstone Cutthroat Trout are native to the Lamar River watershed and have persisted there for over 10,000 years (Benkhe 2002). The other native fishes present are Mottled Sculpin Cottus bairdii, Plains Sucker Pantosteus jordani, and Longnose Dace Rhinichthys cataractae. The National Park Service and other agencies intentionally stocked nonnative Rainbow Trout in the Lamar River watershed in the early 1900s to diversify sportfishing opportunities for visiting anglers (Varley 1981). However, by the 1930s the National Park Service shifted its focus from stocking nonnative fishes towards native fish conservation, adopting a policy “prohibiting the introduction of exotic species in national park or 20 monument waters now containing only native species” (Madsen 1937). Although stocking of Rainbow Trout ceased nearly a century ago, legacy populations still exist, and they continue to invade, hybridizing with native Yellowstone Cutthroat Trout. Hybrids are now abundant in the lower Lamar River watershed, and because of the seasonal, fluvial connectivity of the system, appear to be invading the upper watershed where Yellowstone Cutthroat Trout populations of high conservation priority are present (Ertel 2017; Al-Chokhachy et al. 2018; Heim 2019). Hybrid trout are primarily concentrated in the lower watershed, including Buffalo and Slough creeks and the lower Lamar River (Ertel et al. 2017). However, hybridization has been identified with increasing frequency in the upper watershed in recent years (Heim et al. 2020). This increased prevalence of hybridization is of concern to the National Park Service, which is tasked with preserving the persistence of genetically unaltered Yellowstone Cutthroat Trout populations in the upper Lamar River watershed. Hybrid trout spawn primarily in lower Buffalo and lower Slough creeks, with individuals moving more than 10 km from the upper watershed, through Lamar River canyon, to the lower watershed in the spring (Heim 2019). Buffalo Creek was identified as the primary source of Rainbow Trout in the watershed (Heim et al. 2020), and the National Park Service and agency partners plan to eradicate the Rainbow Trout population from Buffalo Creek in 2024 and 2025 using rotenone—a chemical piscicide. Following the rotenone treatment, genetically unaltered Yellowstone Cutthroat Trout from nearby fluvial populations will be stocked in Buffalo Creek. Selection of Sampling Locations. We scouted Slough Creek and the Lamar River prior to sampling, focusing our efforts on the lower Lamar River watershed downstream of Buffalo Creek according to National Park Service guidance. Our criteria for sampling included finding 21 locations that were suitable for multiple sampling methods and safely accessible. Our selected sampling area comprised a 4.3-km section of lower Slough Creek and a 2.8-km section of the lower Lamar River (Figure 2.1). Our Slough Creek section encompassed an unconstrained valley with a gradient of 0.004%. The vegetation in the low-gradient valley is primarily composed of Idaho fescue Festuca idahoensis and big sagebrush Artemesia tridentata. Riparian vegetation includes tall willow Salix gyeriana and sandbar willow S. exigua and S. melanopsis. Large herds of bison Bison bison graze the valley year-round, and browse pressure is evident along much of the stream banks. The streambed is primarily composed of large cobbles, gravel, and sand, with large meanders creating deep scours in the outside bends, point bars, and run-riffle-pool complexes. The average stream width of the valley part of the sampling section was 31.7 m. The lower Lamar River section had a gradient of 0.016% where large tertiary formations compose a canyon. The stream bed is composed of large boulders, cobbles, gravel, and sand. Dispersed large, woody debris and bedrock created deep plunge pools and long runs with scattered boulders. The average width of the canyon part of the sampling section was 31.2 m. Pilot Snorkeling Study We conducted a pilot snorkeling study in July 2021 to determine the efficacy of snorkeling as a sampling method for adult Yellowstone Cutthroat, Rainbow, and hybrid trout in the Lamar River and Slough Creek. We repeated the study in Slough Creek, but not the Lamar River, in July 2022 based on the results from 2021. We defined the minimum snorkeling thresholds to evaluate the utility of snorkeling as a potential sampling method in each stream. 22 These thresholds were defined as (1) underwater visibilities ≥ 3 m (Schill and Griffith 1984 [1.5 m is often used as the minimum threshold, but we determined this would be too low for snorkelers to identify key morphological characteristics such as white fin tips, coloration, and spotting patterns (Heim et al. 2020; Meyer et al. 2017)]); and (2) minimum water depths of 15 cm and the ability of snorkelers to see the bottom of the stream (Thurow 1994). We measured the visibility of each site by showing snorkelers the silhouette of a 25-cm, model trout marked with parr marks and spots underwater as they drifted downstream past it while holding a measuring tape. The minimum distance that parr marks and spots became visible to the snorkelers was recorded as the minimum visibility for each reach (Thurow 1994). Visibilities were measured to the nearest 10 cm, within fifteen minutes of the start or end of the survey, in depths that were representative of each site. We conducted snorkel surveys at six 400-m sites in 2021 in Slough Creek, but the site furthest upstream was omitted from the study in 2022 because of the difficulty of transporting gear to the location (Figure 2.2) and safety concerns during snorkel surveys; therefore, analyses included only five sites. We used these sites (hereafter “reference sites”) to compare the observed counts and proportions of trout taxa between snorkel and electrofishing surveys. We conducted snorkel surveys during the day (between 0900 and 1700 hours), 24 hours prior to electrofishing in Slough Creek. Snorkelers visually surveyed each reach by drift-diving downstream. Six snorkelers equipped with dry-suits, masks, and snorkels counted fish as they passed over them. Most sites were too swift to effectively sample in an upstream direction; water velocities made it difficult for snorkelers to maintain an upstream direction without being forced downstream by the current. Upstream sampling was only possible in riffle areas and backwaters, 23 but many riffles did not meet the minimum sampling depth. Snorkelers were randomly assigned to lanes to avoid bias in counting lanes. Snorkelers maintained their respective lanes by spacing themselves equidistantly along a 9-m long polyvinyl chloride (PVC) pipe (O’Neal 2007; Figure 2.3) in Slough Creek. Occasionally, snorkelers separated the pipe to split into groups in areas of different depths. In sections of stream too shallow to swim, snorkelers stood up, walked downstream, and waited until disturbed substrate cleared from the water before entering the next floatable section. Snorkelers noted whether they could see the bottom of the stream in each site. Trout were identified as Yellowstone Cutthroat, Rainbow, or hybrid trout. Lengths of trout were categorized as < 100 mm, 100–249 mm, 250–350 mm, and > 350 mm. All fish observations were recorded on dive slates attached to an arm of each snorkeler and transferred to a data sheet at the end of each reach. To avoid double counting, snorkelers used hand gestures to signal whether they had marked a fish that moved across multiple snorkeling lanes. We compared the number of trout identified to a taxonomic group (e.g., identified using morphological characteristics such as white fin tips, coloration, and spotting patterns [Heim et al. 2020; Meyers et al. 2017]) to the number of trout marked as unknown. These comparisons were used to assess how measured visibility, ability of snorkelers to see the bottom of the stream, and width of the stream affected the visual extent to which these characteristics were detectable. Snorkel surveys were conducted similarly in the Lamar River, but a PVC pipe was not used there (Figure 2.4) because of the presence of large boulders in the snorkeling lanes at some sites. We snorkeled seven 400-m sites and one 250-m site in late September 2021 in the Lamar River (Figure 2.2). Logistical constraints in 2021 precluded us from electrofishing after snorkeling in 24 the Lamar River; therefore, we did not conduct any comparisons between electrofishing and snorkeling surveys. Electrofishing CPUE We conducted electrofishing surveys in a 4.3-km section of Slough Creek, divided into 18 sites, in 2021 and 2022 to calculate CPUE (trout/100 m) of Yellowstone Cutthroat, Rainbow, and hybrid trout (Rainbow and hybrid trout hereby referred to collectively as “nonnative trout”); electrofishing surveys were conducted in late July in both years. Snorkel reference sites also served as starting and ending points during these surveys to compare electrofishing and snorkeling estimates; these sites encompassed multiple electrofishing sites, hence the discrepancy between the number of snorkel-reference and electrofishing sites. We conducted a single-pass electrofishing survey in a 2.6-km section of the Lamar River, divided into 12 sites, in late September 2022 to calculate CPUE of the trout taxa. We used VVP-15B electrofishers (Smith-Root, Vancouver, Washington) powered by 3,500- or 5,000-watt generators mounted to the frames of two 3-m, inflatable, self-bailing rafts in both streams. The electrofishers transferred power to the water through two electrode arrays attached to the bow of each raft (300 V, 30% duty cycle, and 30 Hz). The rafts floated downstream as one person in each operated the oars to steer and maintain a speed consistent with, or slightly faster than, the water velocity (Bonar et al. 2009). Two people netted fish from the bow of each raft using long-handled dip nets. Trout were processed at the downstream end of each site in a trailing raft. We identified each trout to taxon, recorded its total length (mm), and scanned it for an existing half-duplex passive integrated transponder (PIT) tag (Biomark FS-2001 ISO proximity reader; Biomark Inc., Boise, Idaho). A 25 syringe (Biomark MK7) was used to implant 23-mm tags into the dorsal sinuses of large, unmarked fish (> 120 mm). Small fish (70–120 mm) were implanted with 12-mm tags in their abdominal cavities using a scalpel. We used PIT-tag recoveries to evaluate the independence of sites (to assess if trout were moving among sites after being released, and therefore being included in calculations of the CPUE estimates of other sites). Trout were released at least 50 m upstream of the processing location to avoid recapture downstream. Abiotic Measurements We measured water temperature and conductivity was measured during electrofishing surveys. Water temperature was measured using Onset HOBOware data loggers (UA-001-08) placed at the upstream and downstream ends of the sections. Conductivity was measured using an Apera Instruments handheld conductivity meter (EC850). Mark Recapture Slough Creek. We conducted additional electrofishing and angling sampling in late July and early August 2022 to estimate the abundances of Yellowstone Cutthroat and nonnative trout. Four capture “events” (one event = a complete sampling of the 4.3-km section) were conducted over a 4-week period. We conducted two electrofishing events followed by two angling events (Table 2.1), with one week between each survey. Electrofishing methods were the same as described above for our CPUE analysis; the first electrofishing event was used for both the CPUE and mark-recapture analyses. Angling events involved 4–7 anglers divided into two groups. We randomly assigned each group to one of five stream segments within the 4.3-km section over a 4-day period. Each 4-day period of angling constituted a capture event, and 26 anglers sampled the segments in the same order during the second week to maintain consistency. Anglers proceeded upstream within their assigned segment, sampling strategically to ensure all available habitat was fished. They alternated among habitats and the use of fly and spinning gear to reduce bias in gear selectivity, lure presentation, and angler experience. Angled fish were released at their capture location. We identified each trout to taxon, recorded its total length (mm), and scanned it for an existing PIT tag. During each capture event, Yellowstone Cutthroat Trout without existing PIT tags were tagged using the methods described previously. We removed nonnative trout after the first capture event in each stream to assist with NPS nonnative trout suppression efforts. Yellowstone Cutthroat Trout captured during the electrofishing survey were released at the downstream end of each site. Yellowstone Cutthroat Trout captured during the angling surveys were released at the capture location. Lamar River. We conducted a 2-event, mark-recapture study over a 2-week period (Table 2.1) in a 2.6-km section of the Lamar River in 2022 (Figure 2.1) using techniques similar to those used in Slough Creek. Electrofishing surveys were conducted in late September. Angling was conducted by three anglers starting 10 days after electrofishing. Anglers started at the downstream end of the section and proceeded upstream, sampling the section over a 4-day period. We identified each trout to taxon, recorded its total length (mm), and scanned it for an existing PIT tag. We removed nonnative trout after the first capture event in each stream to assist with NPS nonnative trout suppression efforts. Yellowstone Cutthroat Trout captured during the angling surveys were released at the capture location. 27 Juvenile Sampling Pilot Study Slough Creek. We conducted three-pass-depletion, juvenile-trout electrofishing surveys at six sites of Slough Creek in October 2021 to determine the efficacy of estimating the absolute abundance of juvenile Yellowstone Cutthroat, Rainbow, and hybrid trout along channel margins. We sampled in October to allow individuals to reach their maximum growth potential for the year, ostensibly improving our ability to visually distinguish among trout taxa (Koenig 2006; Meyer et al. 2017). Sites were 20-m long bank-habitat units extending 2 m perpendicular from the bank towards the main current (Figure 2.5). Block nets were not used during these surveys; instead, we assumed sites were biologically closed because of the short sampling period (1 hour) and evidence that short-term juvenile emigration is limited (Mitro and Zale 2000; Korman et al. 2009). Two people operated backpack electrofishers while two netters captured trout with long-handled dip nets. Crews moved upstream, carefully sampling the substrate, and ensuring to shock interstitial spaces slowly among large substrates, when present. We conducted three electrofishing passes at each site unless we caught zero fish on the second pass (1 site). If more fish were caught on the third pass than on the second, we conducted a fourth pass (2 sites). During surveys, fish were placed in live boxes placed at least 5 m outside of the survey reach to avoid exposure to the electrical field during subsequent passes. At the end of each electrofishing pass, trout were identified to taxa, their total lengths were measured (mm), and small tissue samples were collected from their anal fins for genetic analyses (results of genetic analyses are detailed in Chapter 3). Lamar River. We conducted juvenile-trout electrofishing surveys at 16 sites in the Lamar River in October 2021 using the same methods as in Slough Creek. However, some sites were 28 extended to 50 m in length to increase the opportunity to catch fish. We also conducted opportunistic sampling in habitats deemed ideal for juvenile trout (boulder substrate with interstitial spaces along the bank). Data Analysis All data analyses were conducted using R statistical software (R Development Core Team, 2009) or Program MARK (White and Burnham 1999). Snorkel-Electrofishing Comparison Analysis We evaluated the accuracy of the taxonomic identifications assigned during snorkeling by comparing the raw counts and proportions of each taxon assigned by snorkeling versus electrofishing at sites where both techniques were used; the identifications of electrofished individuals were assumed to be correct because the fish were examined by hand. We first compared counts of each taxonomic group from each method at each site to a 1:1 line to determine if both methods recorded similar counts. We also compared the proportions of nonnative trout identified in reference reaches by snorkeling and electrofishing using a generalized linear mixed model. The model used a binomial random variable for the response, in which each trout observed was modeled as a Bernoulli trial with 1 = nonnative trout observed and 0 = Yellowstone Cutthroat Trout. We included method (snorkeling or electrofishing) as the explanatory variable, site as a fixed variable, and year as a random variable to account for non- independence of samples among years. Prediction intervals were calculated using the ggpredict function in the ggeffects package (Lüdecke et al. 2021). We assumed the observed proportions of 29 nonnative trout observed by each sampling methods to be statistically indistinguishable if the null hypothesis (H0) was accepted (P > 0.05) and 95% prediction intervals overlapped. CPUE Analysis Catch-per-unit-effort at each site was calculated as number of trout per 100 m in our simulated section (we multiplied the quotient by 100 to standardize our measurements to the same units) using the equation CPUE = 𝑐 𝑓𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 · 100 , where 𝑐 = the number of trout caught and 𝑓𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 = site length (m). Mark-Recapture Analysis Slough Creek. We fit closed mark-recapture models for Yellowstone Cutthroat Trout in Slough Creek with Program MARK using the RMark package (Laake 2013). For capture probability (p), three full-likelihood models were considered where p was constant (M0) and p varied by capture event (Mt; Otis et al. 1987). We used Akaike's Information Criterion adjusted for sample size (AICc) to evaluate model fit among the models (Burnham and Anderson 1998). The lowest AICc score was used to determine the best-ranked model compared to the lesser- ranked models. Goodness-of-fit testing was conducted using the CAPTURE interface in Program MARK. We calculated the weighted average for the derived estimates of �̂� and associated 95% confidence intervals because of the similar performances of the M0 and Mt models. Nonnative trout abundance estimates were also calculated using only the first two (electrofishing) events 30 and a Chapman-modified, Lincoln-Peterson (hereafter “LP”) estimator (Chapman 1952) using the mrClosed function in the FSA package (Ogle 2019). The LP estimator was calculated as �̂� = (𝑛1 + 1) · (𝑛2 + 1) (𝑚2 + 1) − 1, where �̂� = the estimated trout abundance, 𝑛1 = the number of fish caught during the first (marking) sampling event, 𝑛2 = the number of fish caught during the second (recapture) event, and 𝑚2= the number of marked fish recaptured during the second sampling event. We also estimated abundance of Yellowstone Cutthroat Trout from the first two sampling events using the LP estimator to compare to the models analyzed in Program MARK. All trout abundance estimates were standardized to density (trout/kilometer). Lamar River. We estimated Yellowstone Cutthroat Trout and nonnative trout abundances in the Lamar River using the LP estimator because we only had two sampling events. All trout abundance estimates were standardized to density (trout/kilometer). Simulations CPUE. We bootstrapped (with replacement) our CPUE estimates of each taxon from each stream to determine the sampling intensity (i.e., continuous site length) needed to precisely characterize trout CPUE in each stream. For our Slough Creek simulations, we averaged the CPUE estimates from each site across both years (2021 and 2022) to account for differing sampling conditions. For the Lamar River, we used our single year of data (2022). For each number of sites (e.g., 5, 6, 7, etc.), we conducted 10,000 simulations in which we (1) summed the 31 site lengths (e.g., 2 sites; 330 + 250 m = 580 m) and total number of individuals captured, (2) calculated CPUE as 𝐶𝑃𝑈𝐸𝑠𝑖𝑚 𝑖 = ∑𝑖𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙𝑠 ∑ 𝑠𝑖𝑡𝑒 𝑙𝑒𝑛𝑔𝑡ℎ𝑠 · 100, and (3) calculated the mean and standard deviation of all CPUE estimates for each sample size (number of sites) across the 10,000 simulations. The coefficient of variation (CV) was used to compare the effects of section length on the precision of CPUE. We calculated the CV of our CPUE simulations as 𝐶𝑉 = 𝑆𝐷 𝐶𝑃𝑈�̂� where SD = the population standard deviation of the simulations and 𝐶𝑃𝑈�̂� = mean CPUE across simulations for each sample size. Mark Recapture. We used abundance estimates of Yellowstone Cutthroat and nonnative trout (�̂�) and angling and electrofishing p to simulate alternative sampling designs and evaluate the precision of the resulting abundance estimates (Kéry and Royle 2016). Electrofishing p was estimated directly from the mark-recapture analysis ( 𝑚2 𝑛2 ) , and angling p was estimated by dividing the number of each taxon caught during angling surveys by the abundance estimate ( 𝑛2 �̂� ). For Yellowstone Cutthroat Trout, we constructed a hypothetical population using our abundance estimate (�̂�) of 399 individuals and our estimates of p (0.14 for electrofishing and 0.15 for angling) to inform our simulations. For nonnative trout, we used the �̂� of 224 individuals and the respective p for each sampling method (0.10 for electrofishing and 0.22 for angling). By creating these hypothetical populations, we had known abundances and were able to compare the point 32 estimates and confidence intervals generated by different sampling designs to the “true” population abundances. We generated p for each population across three simulated capture events (t = 3) assuming a study where captured trout receive a unique identifier (i.e., a PIT tag) during each event. For example, if a trout was simulated to be captured on the first occasion, not captured on the second occasion, and captured on the third occasion, it would have a capture history of 101. For each simulated capture event, we incorporated uncertainty into p by randomly selecting from a beta distribution centered around our specified p (SD = 0.02) for each method using the rbeta2 function in the IPMbook package (Schaub et al. 2023). We constructed this beta distribution based on a range of capture probabilities we determined to be realistic for each sampling gear. We used Program MARK with the RMark package to estimate abundance of the constructed populations using a full-likelihood Mt model (Otis et al 1978). We also estimated abundance using two events to evaluate agreement among our empirical estimates from the LP estimator and the simulations (EA) because maximum likelihood for the Mt model and LP estimator are equivalent when t = 2 (Otis et al. 1978). We evaluated sampling designs that incorporated combinations of angling and electrofishing including three angling events (3A), two angling events and one electrofishing event (2AE), and two electrofishing events and one angling event (2EA). Simulations were repeated 10,000 times for each sampling design. We used the simulation CVs and median 95% confidence interval lengths as measures of precision for comparing sampling designs. The CV of each sampling design was calculated as 𝐶𝑉 = 𝑆𝐷 �̂� , 33 where SD = the population standard deviation of the simulations and �̂� = mean abundance estimates across simulations for each sampling design. Median confidence interval lengths were calculated as the median upper – median lower 95% confidence intervals. Power Analyses We used different methodologies to determine the magnitude of change each sampling design could detect. We evaluated linear declines in CPUE by using time as a substitute for site replication because of the limited geographic scope of our study area. Therefore, we had no intra- annual variation in our estimates. For mark-recapture estimates, uncertainty is “built in” to the estimator, and therefore, we were able to use the confidence intervals to determine whether a change could be detected. CPUE. We used Monte Carlo simulations (recommended by Link and Hatfield 1990) to determine our statistical power to detect declining trends in the CPUE of Yellowstone Cutthroat and nonnative trout in Slough Creek and the Lamar River. We used our bootstrapped CPUE estimates to simulate declines in trout CPUE at a single site of varying lengths. We used an exponential decay function to simulate declines in CPUE using the equation Ni = N1 · (1 + 𝑟)t, where Ni = CPUE at time step i (e.g., year 5), N1 = the initial CPUE estimate (year 1), r = the rate of change for the prespecified declines, and t = the number of years. We simulated 25%, 50%, and 75% declines over 5-, 10-, 15-, and 20-year time periods. To account for demographic fluctuations in trout population abundances, we incorporated stochasticity into the estimate based 34 on the “proportional standard error” of CPUE at the initial time step calculated as ( 1 √𝐶𝑃𝑈𝐸 ; Gerrodette 1987) and selected random points from a normal distribution centered around the projected CPUE at time t+1 (Figure 2.6); this process was repeated 10,000 times for each scenario (site length, percentage decline, and number of years). The slope parameters of log- linear regression models were used to determine whether the decline was negative and detectible (H0: �̂� = 0; P < 0.05) for each simulation (Thompson et al. 1998; Dauwalter et al. 2009). Statistical power was calculated as the proportion of simulations that were negative and detectible out of our total simulations (Gibbs 1998). We used 0.8 (i.e., 80% of the simulations) as our threshold for acceptable power. Mark Recapture. We used our empirical and simulated abundance estimates and associated precision (95% confidence intervals) to determine whether we could detect 50 and 75% declines in Yellowstone Cutthroat and nonnative trout abundances across our simulated sampling designs. Rather than using a null hypothesis testing framework, we determined our ability to detect declines based on non-overlapping confidence intervals of our reduced and initial estimates. For the Lamar River, we repeated our simulations for each sampling design based on populations that were 50 and 25% (hereby “reduced estimates”) of the initial estimates. We then compared the median �̂�, lower 95%, and upper 95% confidence intervals of all 10,000 simulations of our reduced estimates to those of our initial estimates. We were unable to conduct the same population simulations for the LP estimates from Slough Creek because of low recapture rates (Table 2.1). We only conducted the simulations for Yellowstone Cutthroat Trout based on our estimated �̂� and p from RMark. 35 Juvenile Trout Abundance Analysis Juvenile abundance estimates were calculated with depletion models (Carle and Strub 1978) using the removal function in the FSA package (Ogle et al. 2019). Results Snorkeling Pilot Study Slough Creek. We determined that conditions in Slough Creek were conducive to snorkeling in 2021 but not in 2022. Mean snorkeler visibility was 3.3 m in 2021 and snorkelers observed 178 trout in 6 sites, 23 of which (12%) were identified as unknown (morphological characteristics could not be confirmed). Mean snorkeler visibility was 2.6 m in 2022 and snorkelers observed 135 trout in 5 sites, 55 of which (41%) were identified as unknown. A 0.7-m reduction in snorkeler visibility resulted in a nearly 4-fold increase in the proportion of trout that could not be identified to taxon. Snorkelers were unable to see the bottom of deep pools in three of five sites in both years. These results indicate that snorkeling may be useful for monitoring, but only when snorkeler visibility is ≥ 3 m. Lamar River. We determined that the lower Lamar River is not conducive to monitoring by snorkeling. The mean snorkeler visibility was 2.2 m and snorkelers observed 27 trout in seven sites, 22 of which (81%) were identified as unknown. Water velocities in this section made snorkeling difficult, and snorkelers were unable to slow themselves to observe trout long enough to identify morphological characteristics. Water visibility did not meet the minimum requirement of 3 m, snorkelers were unable to see the bottom of the stream in 3 sites, and most trout could not be identified to taxon. 36 Snorkel-Electrofishing Comparison Analysis We did not see agreement between the numbers of Yellowstone Cutthroat, Rainbow, and hybrid trout sampled (counts) by electrofishing and snorkeling in Slough Creek for 2021. Electrofishing and snorkeling counts in reference sites were not related when compared to a 1:1 line, and snorkeling counts were generally higher than electrofishing counts (Figure 2.7). Seventy-four trout were sampled by electrofishing, and 155 (excluding unknown) trout were observed during snorkel surveys in six reference sites in 2021. We sampled fewer trout during snorkeling and electrofishing surveys in 2022, with 63 trout sampled by electrofishing and 48 (excluding unknown) trout observed during snorkel surveys in five reference sites in 2022. We observed agreement between the observed proportions of trout taxa in the sample of each method, but the agreement was stronger in 2021. The total proportion of nonnative trout in reference sites was 0.65 (95% CI = 0.52 – 0.77) in 2021 for electrofishing surveys and 0.66 (95% CI = 0.58 – 0.74) for snorkeling. In 2022, we estimated the total proportion of nonnative trout as 0.79 (95% CI = 0.59 – 0.92) for electrofishing and 0.49 (95% CI = 0.37 – 0.60) for snorkeling. We have moderate evidence to suggest that proportions of nonnative trout did not differ between sampling methods based on results from the GLM (β = 0.77; P = 0.10; DF = 19; Figure 2.8). Abiotic Measurements Average water temperature was 17° C in 2021 and 16.8° C in 2022 and average conductivity was 162 µS in 2021 and 124 µS in 2022 during the electrofishing survey in the Slough Creek section. Average water temperature was 13.4° C and average conductivity was 217 µS during the 2022 electrofishing survey in the Lamar River section. 37 CPUE Estimates Slough Creek. Estimates of Yellowstone Cutthroat and nonnative trout CPUE were relatively low in Slough Creek. The frequency of zero catches of Yellowstone Cutthroat Trout increased from 4 to 13 from 2021 to 2022. Mean CPUE of Yellowstone Cutthroat Trout was 0.67 (SE = 0.24) in 2021 and 0.20 (SE = 0.09) in 2022. The frequency of zero catches of nonnative trout was 7 for both years. However, nonnative trout CPUE decreased from 1.14 (SE = 0.32) in 2021 to 0.86 (0.31) in 2022 (Table 2.2). Lamar River. Catch-per-unit-effort estimates were higher for both taxa in the Lamar River than in Slough Creek. Yellowstone Cutthroat Trout CPUE was 2.22 (SE = 0.42) with only one site having zero catches. Nonnative trout CPUE was 1.33 (0.42) and two sites had zero catches (Table 2.2). Mark Recapture Estimates Slough Creek. Yellowstone Cutthroat Trout abundance estimates were low (17 – 24% that of nonnative trout), depending on the model used. The weighted-average Yellowstone Cutthroat Trout abundance was 17 trout/km (95% CI of 12 – 22 trout/km). Estimates of abundance from the LP estimator were 24 trout/km (95% CI = 4 – 956 trout/km) for Yellowstone Cutthroat Trout and 100 trout/km (95% CI of 30 – 196 trout/km) for nonnative trout. Overall, recapture rates were very low for the electrofishing events (Table 2.1), resulting in wide confidence intervals for the LP estimator. However, Yellowstone Cutthroat Trout abundance estimates were precise when we conducted two extra capture events by angling, indicating that electrofishing alone may not be sufficient for estimating trout abundance in Slough Creek. 38 Lamar River. Yellowstone Cutthroat Trout abundance was 153 trout/km (95% CI = 93 – 230 trout/km) and nonnative trout abundance was 86 trout/km (95% CI = 43 – 185 trout/km). Angling captured more individuals of each taxon than electrofishing (Table 2.1). Simulations CPUE. The coefficient of variation of Yellowstone Cutthroat and nonnative trout CPUE estimates decreased with longer section lengths. In Slough Creek, the greatest reduction in error occurred when section length increased from 250 m to 500 m, and we observed a consistent decline in error as site length increased (Figure 2.9). In the Lamar River, the greatest reduction in error occurred when section length was increased from 200 to 400 m, and we observed a consistent decline in error as site length increased (Figure 2.10). Mark Recapture. For Lamar River estimates, an all-angling approach yielded the largest reduction in error, with a CV estimate of 0.18 and confidence interval width of 253 for Yellowstone Cutthroat Trout and a CV estimate of 0.14 and confidence interval width of 117 for nonnative trout (Table 2.3). Power Analyses