Climate change impacts on native cutthroat trout habitat in Colorado streams

Headwater streams support vital aquatic habitat yet are vulnerable to changing climate due to their high elevation and small size. Coldwater fish are especially sensitive to the altered streamflow and water temperature regimes during summer low flow periods. Though previous studies have provided insights on how changes in climate and alterations in stream discharge may affect habitat availability for various native cutthroat trout species, suitable physical habitats have not been evaluated under future climate projections for the threatened Greenback Cutthroat Trout (GBCT) native to headwater regions of Colorado, USA. Thus, this study used field data collected from selected headwater streams across the current distribution of GBCT to construct one‐dimensional hydraulic models to evaluate streamflow and physical habitat under four future climate projections. Results illustrate reductions in both predicted streamflow and physical habitat for all future climate projections across study sites. The projected mean summer streamflow shows greater decline (−52% on average) compared to the projected decline in mean August flow (−21% on average). Moreover, sites located at a relative higher elevation with larger substrate and steeper slope were projected to experience more reductions in physical habitat due to streamflow reductions. Specifically, streams with step‐pool morphologies may experience grater changes in available habitat compared to pool‐riffle streams. Future climate change studies related to coldwater fish that examine spatial variation in flow alteration could provide novel data to complement the existing literature on the thermal characteristics. Tailoring reintroduction and management efforts for GBCT to the individual headwater stream with adequate on‐site monitoring could provide a more holistic conservation approach.

cutthroat trout species, suitable physical habitats have not been evaluated under future climate projections for the threatened Greenback Cutthroat Trout (GBCT) native to headwater regions of Colorado, USA. Thus, this study used field data collected from selected headwater streams across the current distribution of GBCT to construct one-dimensional hydraulic models to evaluate streamflow and physical habitat under four future climate projections. Results illustrate reductions in both predicted streamflow and physical habitat for all future climate projections across study sites. The projected mean summer streamflow shows greater decline (À52% on average) compared to the projected decline in mean August flow (À21% on average). Moreover, sites located at a relative higher elevation with larger substrate and steeper slope were projected to experience more reductions in physical habitat due to streamflow reductions. Specifically, streams with step-pool morphologies may experience grater changes in available habitat compared to pool-riffle streams. Future climate change studies related to coldwater fish that examine spatial variation in flow alteration could provide novel data to complement the existing literature on the thermal characteristics. Tailoring reintroduction and management efforts for GBCT to the individual headwater stream with adequate on-site monitoring could provide a more holistic conservation approach.

K E Y W O R D S
aquatic habitat, climate change, headwater streams, hydraulic modelling, Rocky Mountains, trout 1 | INTRODUCTION Headwater streams, defined as first-and second-order channels, account for nearly 80% of total river length in the United States (US) (E. Wohl, 2017). In mountainous regions, such as the Rocky Mountains, headwater streams are typically characterized by relatively high gradients with predominately gravel and cobble substrate (Jarrett, 1992;E. Wohl, 2010). They also serve as critical habitat for threatened endemic fish species, as well as food sources for fish and other aquatic and riparian organisms (Colvin et al., 2019;Meyer et al., 2007;Schlosser, 1995;Wipfli & Baxter, 2010).
The size and watershed position of headwater streams makes them especially sensitive to alterations in hydroclimatic conditions caused by changing climate (Beniston, 2003). Climate change in mountainous regions is of particular concern because seasonal snowpack, an important component of regional water supplies, is declining in the western United States (Pederson, Betancourt, & McCabe, 2013;Scalzitti, Strong, & Kochanski, 2016). Furthermore, warmer temperatures in winter and spring are shifting precipitation patterns from snow-dominant to rain-dominant hydrologic regimes in mountainous regions (Klos, Link, & Abatzoglou, 2014). As evapotranspiration increases and precipitation regimes shift because of warming climate, annual mean discharge will likely decrease (Berghuijs, Woods, & Hrachowitz, 2014;Furey, Kampf, Lanini, & Dozier, 2012;Hammond & Kampf, 2020;Jefferson, 2011;Milly & Dunne, 2020;White, Morrison, & Wohl, 2022), resulting in lower base flows in late summer months and a reduction of stream habitat to an extent that could significantly affect coldwater fish (Bradford & Heinonen, 2008;Watts, Grant, & Safeeq, 2016).
The native cutthroat trout species in the Southern Rocky Mountains have been declining with habitat loss from land-use changes, non-native trout species invasion, and water abstraction from human activities over the last 150 years (Roberts, Fausch, Hooten, & Peterson, 2017). Although increases in summer temperature might benefit age-0 cutthroat trout in the highest-elevation Colorado streams due to a longer growth period (Coleman & Fausch, 2007), diminishing streamflow in summer could amplify the already stressful environment for native cutthroat trout (Mantua, Tohver, & Hamlet, 2010;Roberts et al., 2017).
One of the native trout in the Southern Rocky Mountains is the Greenback cutthroat trout (Oncorhynchus clarkii stomias), which is the only salmonid native to the mountain and foothill waters of the South Platte River basin in Colorado (Metcalf et al., 2012). The greenback cutthroat trout (GBCT) were abundant in the late 19th century (Young, Harig, Rosenlund, & Kennedy, 2002), but their populations declined rapidly during the last century due to mining pollution, agriculture, harvesting for commercial sale, and non-native trout invasions (Young et al., 2002;Young & Harig, 2001). The GBCT are listed as threatened under the Endangered Species Act, and their present management focuses on establishing additional populations by propagating and reintroducing fish originating from the Bear Creek in the Arkansas River basin in central Colorado, which was the last remaining genetically pure, self-reproducing population of GBCT (Metcalf et al., 2012). Thus, identifying habitat characteristics suitable for reintroduction and long-term persistence of GBCT populations is paramount to the successful recovery of this threatened species.
However, uncertainties exist regarding how long-term habitat suitability is influenced in the changing climate and how it varies over space.
Previous studies show that populations of native cutthroat trout species in the western United States will likely be harmed due to hydrologic impacts of climate change (e.g., Kovach et al., 2016;Roberts, Fausch, Peterson, & Hooten, 2013;Williams, Haak, Neville, & Colyer, 2009). Existing native cutthroat trout populations in the western United States are already restricted to short headwater stream fragments due to habitat loss and non-native trout invasions (Harig & Fausch, 2002). The effects of climate change could further stress native cutthroat trout populations, as described by Roberts et al. (2017), who found that the combined outcome of climate change and non-native cutthroat trout invasion could extirpate 39% of the total Colorado River cutthroat trout populations and put another 37% of the populations at risk of extirpation in the Southern Rocky Mountains. Furthermore, a systematic review of 42 studies across nine countries that have quantified relationships between trout populations and temperature or streamflow suggested that climate-induced changes in hydrology are expected to have more influential consequences for trout than the summer and fall temperatures (Kovach et al., 2016), highlighting the importance of hydrologic changes likely to occur in headwater streams. The aforementioned studies demonstrate that native cutthroat trout populations are already reduced to small headwater streams, where climate-driven alterations in streamflow can affect the population persistence due to decreases in flow.
Moreover, they also emphasized the importance of taking multiple factors (e.g., stream-specific physical characteristics) into account when assessing climate impacts on coldwater trout habitat instead of only focusing on the thermal characteristics. Thus, evaluating the effects of climate change on native cutthroat trout populations could be improved by including possible changes in habitat driven by reduced streamflow.
Numerous studies have reported the negative impacts of reduced streamflow on fish species, including the loss of habitat (Bradford & Heinonen, 2008;Garbe, Beevers, & Pender, 2016;Hakala & Hartman, 2004;May & Lee, 2004), lowered water quality (Benejam, Angermeier, Munné, & García-Berthou, 2010;Guyette & Rabeni, 1995), and increased predation risk as a result of reduced water depth and velocity (Bradford & Heinonen, 2008;Harvey & Stewart, 1991;Heggenes & Borgstrom, 1988), which can lead to decreased population recruitment and survival (Jonsson & Jonsson, 2009). These studies highlight the importance of understanding the effects of climate-driven changes in streamflow on hydrological parameters that directly impact suitable habitat for trout. Because measuring all environmental conditions of interest over a timeframe that is adequate for robust statistical data analyses of the seasonal variations of hydrological parameters is particularly difficult (Meier & Reichert, 2005), utilizing hydraulic models to simulate various scenarios can be advantageous for quantifying climate change effects on trout habitat.

| Research objectives
Evaluating the effects of climate-induced streamflow changes on available habitat that support GBCT can be important for future management and preservation efforts. Previous work has included quantitative analysis to evaluate the minimum habitat requirements under climate change for native cutthroat trout species in the Rocky Mountain (Roberts et al., 2013(Roberts et al., , 2017Williams et al., 2009).
However, these studies quantified climate change effects at a broad scale relying mostly on GIS derived data, which might not be representative of individual sites that are candidates for reintroduction of GBCT. Hence, this study focuses on gathering more detailed stream morphologic and hydraulic measurements with the goal of evaluating the impact of future climate on instream habitat for threatened GBCT. To support this goal, we seek to answer the following research questions: (1) To what degree will GBCT habitat be reduced in headwater streams; and (2) Are there specific morphological or other characteristics that make headwater streams more susceptible to habitat loss? To answer these research questions, we used projections of future streamflow and one dimensional (1-D) hydraulic models to evaluate the current and future habitat metrics for GBCT (see Figure 1). Furthermore, we statistically related loss of habitat to relevant morphological stream attributes to determine site locations less suitable for GBCT persistence or future reintroduction.

| Study site characteristics
Twelve study sites located in the headwater regions along the Front Range in Colorado were selected with elevation ranging from 2156 to 3487 m ( Figure 2 and Table 1). These sites were selected as high priority GBCT conservation sites because they either support selfreproducing populations of this subspecies established via past reintroduction (Bear Creek, Herman Gulch, and Zimmerman Creek) or are candidates for future reintroduction (Harry Crockett, Colorado Parks and Wildlife, personal communication). Current efforts to restore GBCT populations focus on reintroducing them in headwater streams of their native South Platte basin, and reintroductions occur upstream of physical barriers (e.g., waterfalls and artificial barriers) to protect reintroductions populations from invasions by non-native trout species (Fausch, Rieman, Dunham, Young, & Peterson, 2009). Our 12 study sites represent habitats where most self-reproducing populations of GBCT occur in the foreseeable future.
The climate in the Front Range varies with altitude from an annual average of 100 cm of precipitation and 2 C at the highest elevations ($4000 m), to 40 cm and 10 C along the mountain front ($1500 m) (Wohl, 2001). High flows in the rivers throughout the  Step Note: These 12 sites were studied to determine the current and possible future hydraulic conditions of streams throughout the entire distribution of GBCT. morphology of the stream (Table 1). For instance, low gradient sites (e.g., Hauge Creek, Table 1) required longer survey lengths so that we could capture the variability in morphological features, such as pools and bars. A total of 10 cross sections for each site were surveyed, and the location of the thalweg, water surface elevation, and bank elevations were noted. The cross-section locations aligned with changes in channel morphology (e.g., cross-sections collected across riffles, pools, etc.). The total streamwise site length was the cumulative distance between surveyed cross sections. Channel width was calculated as the distance between right and left bank locations; average bankfull depth was obtained by averaging the weighted bankfull depth, which is calculated by subtracting elevation of each surveyed cross-section point from the averaged right and left bankfull elevations. The particle distribution for each site was measured within a riffle representative of general stream characteristics following the Wolman pebble count method (Wolman, 1954). Between 50 and 100 particles were sampled at roughly 0.5-m intervals across the channel. The elevation of each site was obtained using GPS data collected in the field using a Garmin handheld unit (model eTrex 22x), and the average slope for each site was calculated as the difference in upstream and downstream water surface elevations over the total surveyed distance. The aspect for each site was derived using a Geographic Information System (Google Earth). Finally, the morphological classification of each site was assigned using the Montgomery and Buffington channel classification system (Montgomery & Buffington, 1997). A photographic representation of the morphological classifications identified during field work are shown in Figure 3.

| Discharge time series
Time series data of water levels and atmospheric pressure were obtained using data loggers (Onset HOBO U20L) that were placed in stable stream location in each study site. Absolute pressure (P a ) was recorded in the channel at 30-min intervals from June 2019 to September 2020. A logger was also installed adjacent to the channel in open air to correct for atmospheric pressure (P atm ). A time series of flow depth, (h) was calculated using the gauge pressure (P abs ÀP atm ) as follows: Here, h is flow depth over the sensor (m), P abs is the hydrostatic pressure (P a ), P atm is the recorded atmospheric pressure (P a ), ρ w is water density (1000 kg/m 3 ), and g is acceleration due to gravity (9.81 m/s 2 ).
For each study site, a rating curve was developed using measured discharge data and the calculated depth at the time when the discharge was measured. Discharge was measured using an Acoustic  (Wenger et al., 2010).
Streamflow for historical conditions   Accuracy of Manning's roughness coefficient can significantly influence the calculated hydraulic characteristics in a natural channel (Ferguson, 2007

| Simulated flows
Discharges simulated in each HEC-RAS model include: (1) the calibrated discharge, which was measured during cross-sectional surveys; (2) M30MD, which is the measured mean 30-day minimum discharge from the rolling averages between June 14 and September 30; (3) 2040MS and 2040MAUG discharges; and (4) 2080MS and 2080MAUG discharges.

| Modelled habitat characteristics
HEC-RAS modelling values for water velocity (m/s), channel wetted perimeter (m), and maximum flow depth (m) were used to assess changes in habitat quality. The site-averaged and cross-sectional habitat values were examined for each site. Although specific habitat metrics necessary for GBCT population persistence have not been documented, the selected variables (i.e., velocity, wetted perimeter, and depth) are important for other trout species (Bjornn & Reiser, 1991). Specifically, trout compete for drifting food in summer (Nakano, Kitano, Nakai, & Fausch, 1998) and velocity determines drift food rate and capture efficiency by trout (Piccolo, Hughes, & Bryant, 2008). Water depth is critical as larger trout need deeper pools as cover (Heggenes, Northcote, & Peter, 1991). Plus, wetted perimeter is an index of total surface habitat area, and its reduction in  (Table 1). Latitude was initially included in the PCA but was removed because latitudinal gradient was too small relative to the overall range of GBCT. These variables were then reduced to two main principal components (PCs), which are linear functions of the original variables that are uncorrelated with each other and can maximize variance (Jolliffe & Cadima, 2016). Relationships between PCs and the percent change in simulated physical habitats under different future climate projections were then analysed with simple linear regression to quantify which sites would experience the greatest changes in habitat characteristics under future climate projections. The PCA was conducted using R statistical software (version 4.1; R Core Team, 2021), the "prcomp" function in the stats package (R Core Team, 2021) and the "ggbiplot" function in the ggplot2 package (Wickham, 2016) package. Statistical significance was set at α = 0.05.

| Streamflow under future climate projections
The percent reductions in MS flow obtained from the Western US Stream Flow Metrics database were À7% to À53% for 2040 and À18% to À80% for 2080. Similarly, the percent reduction in MAUG flow obtained from the database were À7% to À37% for 2040 and À3% to À46% for 2080.
Mean summer discharges were projected to decrease more than mean August discharges (

| Simulated habitat metrics
Mean summer projections experienced greater habitat reductions compared to mean August projections for both 2040 and 2080 future climate scenarios (Table 3 and Figure 4). Across all the sites, the habitat metrics decreased between À2% and À23% for 2040MS and À2% and À46% for 2080MS simulations. Single-site habitat metric reductions tended to be similar across all metrics such that velocity, wetted perimeter, and depth had similar percent reductions (e.g., reductions in velocity, wetted perimeter, and depth at Dry Creek were À19%, À20%, and À23%, respectively; Table 3 Table S1).

| Responses of physical habitat to projected flow reductions
Results from the PCA ( Figure 5) showed that approximately 68% of the variation among sites can be explained by the first two principal components (PC1 = 38.7%, PC2 = 29.2%). Contributing drainage area, average channel width, and average bankfull depth were positively correlated with PC1. Conversely, site elevation, substrate size (e.g., D50 and D84) and stream slope were negatively correlated with PC2.
There was a significant relationship between the PC2 and percent change in simulated physical habitats for both 2040 (R 2 = 0.64-0.79, p < 0.01) and 2080 (R 2 = 0.38-0.78, p < 0.01-0.058) mean summer climate projections (Figures 6 and 7). Because PC2 had high negative loadings of average slope, elevation, and substrate sizes, these results indicated that sites located at higher elevations with steeper slopes and larger substrates would experience greater changes in available habitat. Furthermore, step-pool morphology streams would experience greater changes and more variability in available habitats compared to streams with pool-riffle morphology (Figures 5 and 6). No significant relationships were found between PC1 and reductions in physical habitats for 2040 or 2080 climate projections (R 2 = 0-0.31; p = 0.31-0.96) (Figures 5 and 6). In contrast, the magnitude of habitat change was much smaller for mean August (results can be found in Figures S1 and S2).

| Implications for GBCT conservation
Conservations efforts for GBCT will be influenced by changes in streamflow and available habitat across a range of hydrologic time periods. The results from the projected streamflow reductions (Table 3 and Figure 4) demonstrate that substantial reductions occur in the mean summer projections but not in the mean August projections. This is likely because the mean summer flow projections capture large reductions in overall snowpack, snowmelt runoff, and earlier runoff timing that will occur in the southern Rocky Mountain (Clow, 2010;Harpold et al., 2012;Pederson et al., 2013) This interaction is likely to impair water quality (Ducharne, 2008) but could benefit recruitment of young-of-the-year fish in highelevation streams where cold temperature and short summer growing seasons currently limit their recruitment (Coleman & Fausch, 2007). Impacts of climate change varies among sites, but up to 20% reductions in velocity, wetted perimeter, and flow depths were projected by 2040 and 40% reductions in habitat were projected in 2080 based on changes in summer streamflow (Table 3).
These habitat reductions could have detrimental impacts on trout populations. For instance, stream salmonids show density-dependent body growth patterns in summer (Huntsman, Lynch, & Caldwell, 2021;Vøllestad & Olsen, 2008), and reduced wetted perimeter would result in a smaller carrying capacity of headwater habitats via summer flow reductions. In addition, larger trout require depth as cover in small streams (Penaluna, Dunham, & Andersen, 2021;Sotiropoulos, Nislow, & Ross, 2006). Heggenes et al. (1991) reported that cutthroat trout larger than 9 cm total length selected stream habitats with depths >25 cm, and maximum channel depths are already or projected to be below this threshold in many study streams. Finally, slower velocity would reduce drift food availability, which is the primary food source for stream salmonids in summer (Owens & Keeley, 2022;Uthe et al., 2019). Optimal foraging velocity of stream salmonids during base flow conditions range between 20 and 40 cm/s (Morita, Sahashi, & Tsuboi, 2016;Nislow, Folt, & Parrish, 1999), which again fall within the velocity range projected in the current and future climate scenarios for the study sites. Taken together, this analysis shows that the magnitude of available habitat alterations projected in this study could affect persistence of native cutthroat trout populations in highelevation Colorado streams.
It was found that climate change impacts on available habitat would vary among sites, with higher-elevation sites, characterized by steeper slopes and larger substrates, being more likely to experience greater degrees of available habitat changes owing to future flow reductions (Figures 6 and 7). This result aligns with elevationdependent climate impacts found in the Rocky Mountains and other mountain regions that are experiencing rapid climate-driven changes, such as flow reductions (e.g., Papadaki et al., 2016;Pederson et al., 2013;Tague & Dugger, 2010). Although the scope of this research was not to investigate the specific mechanisms of changes in available habitat across morphological stream types, steeper streams with larger substrate at higher elevations may be more sensitive to flow reductions because of the already shallow flows that tend to be present in high-gradient streams compared to low-gradient streams for a given discharge (Vezza, Parasiewicz, Spairani, & Comoglio, 2014).
Under large reductions of flows, this typically results in drying of step habitats and physical isolation of remaining pool habitats, where trout congregate and demographic rates such growth and survival are negatively impacted (Hakala & Hartman, 2004;Penaluna et al., 2021). Ultimately, flow reduction impacts on trout populations depends on availability of pools, which may differ by habitat morphology (cascade vs. step-pool vs. riffle-pool morphology) (Rosenfeld & Boss, 2001;Vezza et al., 2014). Since high-gradient streams with large substrate tend to be located at higher elevations (though not always), these sites, which also will have less contributing area to capture snowmelt runoff, may have habitat that is most impaired in future climates. In addition, since GBCT has already been confined to higher elevations (Cook, Rahel, & Hubert, 2010), the loss of habitat in steep, highelevation streams will exert more stress on the GBCT population that are often outcompeted by non-native trout.
In addition, conservation of native trout in the intermountain Western USA has frequently occurred in high-elevation, small headwaters because of the need for physically isolating their populations from invasion by non-native trout species (Fausch et al., 2009). Likewise, GBCT reintroductions in Colorado often occur at isolated higher elevations because they are protected from invasion by non-native species (Harig, Fausch, & Young, 2000;Young et al., 2002). These reintroduction sites coincide with streams expected to experience change in habitat characteristics due to flow reductions. Paradoxically, some of the highest-elevation study streams are limited by a cool and short summer growing season and warming climate could benefit these populations (Coleman & Fausch, 2007), especially if warming is accelerated by reduced summer flows. Conservation success of GBCT populations in Colorado in a changing climate could likely depend on the relative strengths of counteractive effects of warming temperature and reduced flows. Given the spatially heterogeneous effects of physical habitat change and uncertainties related to climate change projections, this study suggests that a GBCT conservation approach that establishes and protects populations along elevational and geomorphological gradients could build a portfolio that buffers against future uncertainties.

| Study limitations
The reliability of the results in this study are influenced by hydrologic and hydraulic uncertainties. The high elevation streamflow reductions that were used in this work are influenced by limitations in the VIC model's meteorological forcing data that are extrapolated from stations located at lower elevations. In addition, only 2 years of hydrologic field data were collected at each study site, constraining the conditions used for model calibrations and future flow projections. In addition, VIC model results fail to account for the heterogeneity in Sites with these characteristics are used for GBCT reintroduction efforts (Young et al., 2002), thus highlighting the importance of establishing conservation measures along elevational and geomorphological gradients to buffer against future uncertainties and create GBCT population resilience.

ACKNOWLEDGMENTS
We thank Marissa Karpack for her assistance in collecting field data.
This work was supported by funding from the US Geological Survey.
Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the US Government.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are availabile in a Sci-