Summer temperature regimes in southcentral Alaska streams: watershed drivers of variation and potential implications for pacific salmon

Citation metadata

Publisher: NRC Research Press
Document Type: Report
Length: 10,802 words
Content Level: (Level 5)
Lexile Measure: 1540L

Document controls

Main content


Climate is changing fastest in high-latitude regions, focusing our research on understanding rates and drivers of changing temperature regimes in southcentral Alaska streams and implications for salmon populations. We collected continuous water and air temperature data during open-water periods from 2008 to 2012 in 48 nonglacial salmon streams across the Cook Inlet basin spanning a range of watershed characteristics. The most important predictors of maximum temperatures, expressed as mean July temperature, maximum weekly average temperature, and maximum weekly maximum temperature (MWMT), were mean elevation and wetland cover, while thermal sensitivity (slope of the stream-air temperature relationship) was best explained by mean elevation and area. Although maximum stream temperatures varied widely between years and across sites (8.4 to 23.7[degrees]C), MWMT at most sites exceeded established criterion for spawning and incubation (13[degrees]C), above which chronic and sublethal effects become likely, every year of the study, which suggests salmon are already experiencing thermal stress. Projections of MWMT over the next ~50 years suggest these criteria will be exceeded at more sites and by increasing margins.

Les changements climatiques sont les plus rapides dans les regions de haute latitude, et nos travaux tentent de comprendre les taux et facteurs associes a l'evolution des regimes thermiques dans les cours d'eau du centre-sud de l'Alaska et leurs repercussions sur les populations de saumons. Nous avons recueilli de maniere continue des donnees de temperature de l'eau et de l'air durant des periodes d'eau libre de 2008 a 2012 dans 48 cours d'eau non glaciaires a saumons a la grandeur du bassin versant du golfe de Cook representant un eventail de caracteristiques hydrographiques. Les plus importantes variables predictives des temperatures maximums, exprimees comme etant la temperature moyenne en juillet, la temperature moyenne hebdomadaire maximum et la temperature maximum hebdomadaire maximum (TMHM), etaient l'altitude moyenne et la couverture de milieux humides, alors que l'altitude moyenne et la superficie etaient les variables qui expliquaient le mieux la sensibilite thermique (la pente de la relation entre la temperature du cours d'eau et celle de l'air). Si les temperatures maximums du cours d'eau variaient beaucoup (de 8,4 a 23,7[degrees]C) d'une annee et d'un emplacement a l'autre, la TMHM dans la plupart des emplacements depassait le seuil etabli pour le frai et l'incubation (13[degrees]C) au-dela duquel des effets chroniques et subletaux deviennent probables, chaque annee de l'etude, donnant a penser que les saumons sont deja en situation de stress thermique. Les projections de la TMHM pour les ~50 prochaines annees portent a croire que ces seuils seront depasses dans un nombre croissant d'emplacements et par une plus grande marge. [Traduit par la Redaction]

Full Text: 


Climate is changing fastest in high-latitude regions like Alaska, where mean annual air temperature has increased nearly 2[degrees]C over the past 50 years and is projected to rise by an additional ~2-4[degrees]C in coming decades (Karl et al. 2009). This warming is associated with ongoing, interrelated changes in the hydrologic and temperature regimes of Alaska's fresh waters (Prowse et al. 2006), such as earlier ice breakup (Schindler et al. 2005; US EPA 2015), earlier depletion of snowmelt-derived streamflow (Stewart et al. 2005), and increasing water temperature (Taylor 2008).

The effects of water temperature on Pacific salmon (Oncorhynchus spp.) are pervasive. The abundance and taxonomic composition of prey and the seasonal timing of its availability are closely linked with temperature regimes (Caissie 2006; Burgmer et al. 2007). Physiological growth potential increases to an optimum temperature and declines as temperature increases (Brett et al. 1969). Timing of life history events, like spawning, emergence, the onset of exogenous feeding, and smolting, are adapted to prevailing environmental conditions and are largely driven by temperature (Brannon 1987; Quinn 2005). High temperatures can block migration corridors (Quinn et al. 1997; Salinger and Anderson 2006), increase disease virulence (Fryer and Pilcher 1974; Kocan et al. 2009), and cause stress or outright death (Richter and Kolmes 2005). As such, warming temperature has the potential to alter the suitability of water bodies for salmon populations.

Anticipating the magnitude and scope of ecological changes to freshwater habitats is essential for the management of Alaska's salmon populations, which contribute substantially to global wild salmon production (Ruggerone et al. 2010) and are exceedingly important to Alaska's ecology, economy, and societal well-being. Despite the rapidly warming climate and recent studies projecting decreases in suitable habitat for Pacific salmonids elsewhere (Morrison et al. 2002; Isaak et al. 2011; Ruesch et al. 2012; Jones et al. 2013), researchers are just beginning to understand rates and drivers of changing temperature regimes in Alaska streams (e.g., see Kyle and Brabets 2001; Davis and Davis 2008; Fellman et al. 2014) and the implications for salmon (e.g., see Lisi et al. 2013; Leppi et al. 2014; Wobus et al. 2015).

Regional analyses conducted outside of Alaska have relied on the compilation and synthesis of extensive data collected by multiple agencies and organizations over the last few decades (e.g., Isaak et al. 2010; Mantua et al. 2010). Long-term stream temperature data sets and regional monitoring networks in Alaska are limited, likely due to the minimal need for restoration planning in Alaska's relatively intact watersheds and relatively cool climate toward the northern end of Pacific salmon distribution. However, interest in developing stream and lake monitoring networks across Alaska is growing because a coordinated monitoring approach can generate regional-scale data useful for prioritizing sensitive areas for conservation and targeting other management strategies that can increase resiliency in aquatic ecosystems to a changing climate (Reynolds et al. 2013).

Summer stream temperature regimes and the sensitivity of streams to climatic warming are influenced by climate, landscape features, and characteristics of the stream and its channel (Mayer 2012). For the nonglacial salmon streams examined in this study, we expected the warmest and most thermally sensitive streams to drain low-elevation landscapes with features that enhance opportunities for absorption of solar radiation and equilibration to air temperature, such as low valley slope, large watershed area, and relatively high cover of wetlands, standing water, and nonforested areas (Caissie 2006; Lisi et al. 2013). We anticipated that lakes would have a warming, rather than moderating, effect on summer stream temperatures (Jones 2010; Lisi et al. 2013), particularly because lakes drained by our study streams tend to be shallow and dark-bottomed. However, we expected that higher streamflow, through the thermal mass it affords (Caissie 2006; Kelleher et al. 2012), and permeable landscapes conducive to the recharge and discharge of groundwater (O'Driscoll and DeWalle 2006; Chang and Psaris 2013; Johnson et al. 2013) would act to moderate water temperature and thermal sensitivity.

Our research focused on answering the following questions related to summer temperature regimes, specifically thermal maxima and thermal sensitivity, in salmon streams of the Cook Inlet region, southcentral Alaska:

(1) How do temperature regimes vary among streams and years?

(2) How are temperature regimes related to site and watershed characteristics and site-specific air temperature?

(3) What are the potential implications of these temperature regimes for different salmon species and life stages now and in the future?


Study area and temperature monitoring

This work focused on nonglacial salmon streams in southcentral Alaska's Cook Inlet basin (Fig. 1). Cook Inlet opens southward to the Gulf of Alaska, and its basin (121700 [km.sup.2]) consists of coastal and valley lowlands surrounded by rugged mountains, including some of North America's highest peaks. The climate ranges from continental to maritime, with mean annual temperatures between -6 and 6[degrees]C (Brabets 1999). Precipitation, much of which falls as snow during winter months, ranges annually from 50 cm across the continental zone to 180 cm across the maritime zone, with the greatest amounts falling in mountainous areas (Brabets 1999). The basin's major river systems drain alpine glaciers and therefore have high sediment loads and turbidity (Lloyd et al. 1987), although many of the tributary streams and smaller watersheds have little or no glacial influence and are clear. Cook Inlet basin has 12 000 km of documented salmon streams and supports substantial wild runs of all five North American Pacific salmon plus other anadromous and resident fishes (Johnson and Coleman 2014). The basin's salmon populations are harvested in commercial, personal use, subsistence, and sport fisheries. Most of the basin is free from anthropogenic watershed disturbances, although urban and suburban development occurs around Anchorage (Alaska's largest city) and several smaller communities, which, collectively, are home to more than two-thirds of Alaska's human population.

We monitored water temperature in 48 nonglacial salmon streams across the basin (Fig. 1) that spanned a range of watershed characteristics (Table 1) and, in most cases, could be accessed by the region's sparse road network. We focused on nonglacial streams because they tend to be warmer during summer than those that receive glacial meltwater (Kyle and Brabets 2001; Fellman et al. 2014) and thus likely more susceptible to thermal change over the next 50 years. Water temperature in the larger glacial systems of the region may be buffered in the short term, as monthly streamflows are projected to increase through the mid-21st century (Deb et al. 2015). We established one temperature monitoring site in the lowest accessible reach of each stream's watershed; 22 of the streams flow directly to the ocean (all sites on these streams are within 5 km of Cook Inlet), 23 streams flow into glacial river systems, and three streams flow into another nonglacial system. With only one site per watershed, we did not expect to capture the spatial heterogeneity of water temperature and cold-water refugia within each watershed; instead, our sampling approach allowed us to identify characteristics that influence temperature heterogeneity at a regional scale.

All 48 streams are listed in the Alaska Department of Fish and Game's Anadromous Waters Catalog (Johnson and Coleman 2014), with coho salmon (Oncorhynchus kisutch) being the most widespread (documented in 47 streams), followed by pink salmon (Oncorhynchus gorbuscha, 36 streams), Chinook salmon (Oncorhynchus tshawytscha, 34 streams), sockeye salmon (Oncorhynchus nerka, 31 streams), and chum salmon (Oncorhynchus keta, 23 streams). In 44 streams, the temperature monitoring site was located in documented spawning habitat for one or more salmon species, while all but one stream had documented spawning habitat upstream of the monitoring station (Johnson and Coleman 2014). Locations (latitude-longitude) and stream names for all sites are provided in online supplementary material, Table S1 (1).

From 2008 to 2012, we deployed paired water and air temperature loggers programmed to record at 15 or 30 min intervals in mid-May to mid-June, as conditions allowed, and collected them after 1 October (see Mauger 2008 for more details). For water temperature, we secured HOBO Pro v2 and StowAway TidbiT data loggers (Onset Computer Corporation, Bourne, Massachusetts) in well-mixed locations within the water column of the mainstem channel to minimize the influence of direct groundwater connection or other reach-scale dynamics. For air temperature, we placed a data logger within a solar shield and secured it 8-30 m away from the stream and at least 1.8 m off the ground. Our 5 years of open-water monitoring captured a period comparable to or slightly colder than the most recent 35 years in the Cook Inlet region (NOAA 2015). The regional summertime temperature anomaly (June-August mean temperature difference from 1980 to 2014 baseline) ranged from -1.1 to +0.3[degrees]C, with 2009 being the warmest year of the study period.

We checked data loggers against a NIST (National Institute of Standards and Technology) thermometer before and after each deployment in 0 and 20[degrees]C water baths to ensure we met our 0.25[degrees]C accuracy goal. We reviewed downloaded logger data and deleted erroneous measurements following Mauger et al. (2015). We attempted to collect 5 years of stream and air temperature data at 48 sites, thus potentially generating 240 "site-years" of data; however, data for 24 site-years were not collected due to loggers being lost during high flow events or other deployment failures. In addition, we removed 17 site-years due to missing or problematic local air temperature data and nine site-years due to deployment periods that likely missed the thermal maxima. The final data set included 48 sites with 1-5 years of data totaling 190 site-years of stream and air temperature data. Deployment dates spanned at least 16 June--31 August for 168 site-years, with 22 site-years having shorter deployment periods, which we evaluated to have captured the thermal maxima.

Stream temperature metrics

We aggregated the quality-controlled stream and air temperature time series for all days with at least 90% of daily measurements into daily means and maximums, which we then used to calculate temperature metrics. We selected three metrics to describe aspects of the summer temperature regimes in the 48 streams: mean July temperature, maximum weekly average temperature (MWAT), and maximum weekly maximum temperature (MWMT). We chose mean July stream temperature because this is typically the hottest month in southcentral Alaska. We calculated mean July temperature only for site-years when at least 90% of the days were captured (>27 days). We used MWAT and MWMT because they represent an intermediate time period over which transient high temperatures may affect fitness in aquatic organisms. We calculated MWAT by selecting the maximum from a weekly (7-day rolling) average of mean daily stream temperatures and calculated MWMT by selecting the maximum from a weekly (7-day rolling) average of maximum daily stream temperatures across the entire deployment period. All three of these temperature metrics have been useful for describing relative differences in thermal maxima between streams and potential impacts to salmonids (Wehrly et al. 2009; Isaak et al. 2010; Ruesch et al. 2012; Moore et al. 2013).

As a fourth response variable, we calculated thermal sensitivity for all sites with at least 3 years of stream and air temperature data (n = 45), which we expressed as the slope of the stream-air temperature relationship (Kelleher et al. 2012; Mayer 2012; Chang and Psaris 2013; Luce et al. 2014) based on weekly mean temperatures during July. Although others have shown the stream-air relationship to flatten out at high air temperatures (>20[degrees]C) due to the effect of evaporative cooling (Mohseni et al. 1998; Mayer 2012), southcentral Alaska's air temperatures in July rarely exceed 20[degrees]C. Because streams in colder climates do not show a substantial change of slope at these moderate air temperatures (Mohseni and Stefan 1999), we chose to use a linear relationship. We selected weekly measurements over daily because the relationship between stream and air temperatures becomes stronger as the time scale increases (Erickson and Stefan 2000). We used a linear mixed effects model with a random intercept for year to account for the temporal correlation between weekly values within a year. We used Akaike's information criterion (AIC) to compare results from a mixed effects model to one without a random intercept; the mixed effects model resulted in better model fit at 29 of the 48 sites. The slopes from the mixed effects model were generally lower than the slopes from the fixed effects model (37 of 48 sites), which we hypothesized was due to the variable influence of snowmelt on stream temperatures during July in some streams.

In addition to the temperature metrics described above, we calculated overall maximum temperature; weekly and monthly mean, maximum, and minimum temperature; daily minimum temperature; and maximum daily fluctuation for each site-year (Table S2 (1)).

Independent predictors of water temperature

We generated a suite of 12 site or watershed characteristics that we hypothesized directly or indirectly affect stream temperatures. Land-cover statistics (percent wetlands, forested, and open water) were derived for each watershed from 30 m resolution LANDSAT imagery (1999, 2003). Maximum watershed elevation, mean watershed elevation, logger site elevation, mean watershed slope, and watershed area were derived from the US Geological Survey (USGS) National Elevation Dataset 2-arc-second (about 60 m grid spacing) Digital Elevation Model (DEM). The 60 m DEM was chosen because it had seamless coverage of the project area. Stream gradient was derived from the USGS National Hydrography Dataset. Lake influence was calculated for 23 watersheds with lakes larger than 0.5 [km.sup.2] connected to the stream network. The area for each lake (n = 51) was divided by its distance to the monitoring site to generate an inverse-distance weighted lake effect and summed across each watershed, since some sites had more than one lake in their watershed. As a proxy for aquifer permeability and groundwater contribution to streams, we calculated the percent unconsolidated fluvial surficial geology (Wilson et al. 2012). Mean annual discharge was calculated using a regression equation developed for southcentral Alaska based on watershed area and mean annual precipitation (Verdin 2004); mean annual precipitation inputs to the equation were based on a mean for the years 1980-2009 (SNAP 2014). Watersheds for each site were delineated by manually editing the watershed boundary data set (sixth level Hydrologic Unit Codes), and all watershed attributes were developed using ArcGIS 10.1.

In addition to site and watershed predictors, we used sitespecific air temperatures to help explain variation in mean July water temperature, MWAT, and MWMT. For mean July water temperature, we used corresponding mean July air temperature; for MWAT and MWMT we used the concurrent 7-day moving average air temperature.

Data analysis

We used mixed effects models to approximate the relationships between temperature metrics (mean July temperature, MWAT, and MWMT) and the suite of predictors, with a random intercept for site because each site was sampled repeatedly for up to 5 years.

We selected our analysis methods so that we would be able to make site-specific predictions under climate change scenarios and also because they enabled us to understand the relative importance of different predictors and their effect sizes. We used multiple linear regressions to model thermal sensitivity because there was only one response for each stream. Given our relatively small sample size, we reduced the original list of 12 site and watershed characteristics by examining pairwise plots and variance inflation factors and sequentially removing highly correlated predictors (r > 0.7) while keeping predictors that we hypothesized would have the most direct effect on stream temperatures. The final set of predictors consisted of five watershed attributes: mean watershed elevation, wetland cover, watershed area, permeable geology cover, and lake influence, each with a variance inflation factor less than three.

We built a set of all possible models using the five watershed attributes, resulting in 32 models, and compared them using AIC. We used this approach for two reasons. First, we expected our five watershed attributes to have additive effects on stream temperatures, so constructing a smaller set of a priori models would not have been meaningful. Second, one of our primary goals was to identify the most important watershed attributes for our stream temperature responses, and a balanced model set allows for relative variable importance to be calculated and compared (Burnham and Anderson 2002). In addition to watershed attributes, the appropriate air temperature predictor was included in all models for mean July temperature, MWAT, and MWMT.

We defined our confidence set as all models with [DELTA]AIC < 4 while removing any models with uninformative parameters (adding an additional parameter to a better model; Arnold 2010). Parameter estimates were averaged over all models in the confidence set by substituting zero when a parameter was missing from a model (Anderson 2008; Grueber et al. 2011). Confidence intervals were estimated for each parameter using unconditional standard errors based on the final model set. Variable importance was calculated as the sum of the Akaike weights ([SIGMA][w.sub.i]) over all the models in which a parameter occurred. Model assumptions were evaluated using a global model with all five predictors and included normality of the residuals, normality of the random effects, homogeneity of variances (normalized residuals plotted against fitted values and each predictor), and a check for outliers using Cook's distance.

We used several measures to quantify model performance for each of the stream temperature responses. We report marginal and conditional [R.sup.2] for the responses modeled using mixed effect models, which are equivalent to the variance explained by the fixed effects and the variance explained by the fixed and random effects, respectively (Nakagawa and Schielzeth 2013). We used adjusted [R.sup.2] for each model in the confidence set for the thermal sensitivity response. Finally, we evaluated the accuracy of predictions by calculating the root-mean-squared error (RMSE) in addition to the observed versus predicted coefficient of determination ([r.sup.2]; Pineiro et al. 2008).

Stream temperature regimes under climate change scenarios

We predicted future MWMT by using air temperature projections for 2030-2039 and 2060-2069 from two emissions scenarios based on a mean of the five best-performing global climate models for Alaska (Walsh et al. 2008). The Scenarios Network for Alaska and Arctic Planning (SNAP) provided mean decadal July air temperature predictions for every decade in the 2000s for the A1B scenario (mid-range C[O.sub.2] emissions scenario) and A2 scenario (rapid increase in C[O.sub.2] emissions scenario) for each of our study sites. The 2000-2009 air temperatures were subtracted from the predictions for the two time periods and two emissions scenarios to calculate the predicted air temperature changes. The air temperature changes were then multiplied by each stream's July sensitivity to calculate the predicted stream temperature changes. The stream temperature changes were added to the maximum daily stream temperatures measured for each stream in 2009 because it was the year with the smallest deviation from the 35-year baseline. Future MWMT for each climate scenario were calculated from the adjusted maximum daily stream temperatures.


Mean July stream temperatures across all sites and years ranged from 6.5 to 18.9[degrees]C, with three sites having values above 18[degrees]C in 2009. Of the 15 warmest mean July temperatures, 11 occurred in 2009 and the remaining four occurred in 2011. Averaged across all years, the highest mean July temperature was 17.3[degrees]C at Jim Creek, which drains a low-elevation wetland and lake complex. Interannual variability in mean July temperatures ranged from 1.2 to 5.1[degrees]C for 44 sites with 3 or more years of data. The highest daily maximum stream temperatures across all years varied broadly among sites, ranging from 11.9 to 24.5[degrees]C and predominantly occurred in July 2009 (e.g., Fig. 2). Maximum daily ranges across sites varied from 3.5 to 11.6[degrees]C.

Maximum weekly average temperature (MWAT) ranged from 7.5 to 21.9[degrees]C across all sites and years. MWAT exceeded 18[degrees]C at 11 sites in 2009 and for two of these sites MWAT averaged over all years exceeded 18[degrees]C. There were 17 sites where MWAT averaged across all years was below 13[degrees]C. Interannual variability in MWAT across sites with 3 or more years of data ranged from 1.1 to 6.4[degrees]C.

Maximum weekly maximum temperature (MWMT) ranged from 8.4 to 23.7[degrees]C across all sites and years (Fig. 3). We recorded 18 MWMT measurements over 20[degrees]C at 13 different sites. When averaged across all years, 13 sites (27%) had mean MWMT above 18[degrees]C. On the cool end of the spectrum, 10 sites had mean MWMT below 13[degrees]C. Of the sites with 3 or more years of data, interannual variation in MWMT ranged from 2.1 to 7.3[degrees]C. MWMT typically occurred in late June and July and less frequently in August (Fig. 4). In 2010, two streams in southern Cook Inlet had MWMT measurements that occurred in September.

Thermal sensitivity ranged from 0.32 to 1.51. Fifteen sites had sensitivities greater than 1.0. (Fig. 5), demonstrating that July water temperature at these sites increased faster than air temperature and underscoring the fact that, although air temperature is highly correlated with water temperature, absorption of solar radiation is the primary driver (Johnson 2003).

For mean July temperature, MWAT, and MWMT, mean watershed elevation was the most important predictor ([SGIMA][w.sub.i] [greater than or equal to] 0.95) and negatively correlated (-); percent wetland cover (+) was of secondary importance ([SGIMA][w.sub.i], = 0.71-0.90); and geology, watershed area, and lakes were less important ([SGIMA][w.sub.i] [less than or equal to] 0.53; Table 2). Mean watershed elevation (-) was also the most important predictor for sensitivity ([SGIMA][w.sub.i] > 0.95), while watershed area (+) was secondary ([SGIMA][w.sub.i] > 0.74), and wetlands, geology, and lakes were of diminishing importance ([SGIMA][w.sub.i] [less than or equal to] 0.31).

The final confidence set for each response included two or three models (Table 3). The marginal [R.sup.2] was similar across the models in the confidence sets for mean July temperature and MWAT, ranging from 0.56 to 0.62. The marginal [R.sup.2] was slightly lower for MWMT and ranged from 0.48 to 0.53. The amount of variation explained for sensitivity was much lower than the other responses; the top model had an [R.sup.2] of 0.25. The [r.sup.2] for the observed versus predicted values closely matched the results for the marginal and adjusted [R.sup.2], with the best predictions for mean July temperature and MWAT, followed by MWMT, and relatively poor predictions for thermal sensitivity. The slopes of the observed-versus-predicted lines for mean July temperature, MWAT, and MWMT were all slightly less than 1 (0.89 to 0.97), indicating a tendency to underpredict these responses in the warmest streams.

All of the predictors except for lake influence were informative for one or more stream temperature responses (Table 4). To demonstrate effect sizes for each predictor, we calculated the change in each that would elicit a 0.1[degrees]C change in modeled mean July temperature, MWAT, or MWMT or a 0.1 change in thermal sensitivity. Mean watershed elevation was negatively correlated with all responses, and a change of roughly 20 m was associated with a 0.1[degrees]C change in mean July temperature, MWAT, and MWMT and a 0.1 change in thermal sensitivity. A small increase in wetland cover ([less than or equal to] 2%) was associated with a 0.1[degrees]C increase in mean July temperature, MWAT, and MWMT, while thermal sensitivity was essentially unresponsive to wetland cover. All temperature responses increased with watershed area, but increasing estimates of these responses by 0.1 units required a relatively large increase in area (i.e., 229-848 [km.sup.2]). Percent permeable geology was associated with decreasing mean July temperature and MWAT, although these effect sizes were also modest. The effect of air temperature predictors was approximately 1:1 on mean July temperature and MWAT, but MWMT was slightly less sensitive to changes in air temperature. It should be noted that the effects of some of the lesser important predictors are inconclusive, as their confidence intervals overlapped 0.

Predicted MWMT for 2030-2039 ranged from 13.1 to 23.7[degrees]C under the A1B emissions scenario and from 13.8 to 26.0[degrees]C under the A2 emissions scenario (Fig. 6). Twenty-nine sites (69%) had MWMT greater than 18[degrees]C under the A2 emissions scenario for 2030-2039. MWMT predictions for 2060-2069 ranged from 13.8 to 25.8[degrees]C under the A1B emissions scenario and were similar to the 2030-2039 predictions under the A2 emissions scenario. The predictions for 2060-2069 under the A2 emissions scenario ranged from 14.4 to 27.6[degrees]C, and 26 sites (62%) had MWMT greater than 20[degrees]C. It is important to note that, at the higher air temperatures predicted in future decades, stream temperature may not increase with air temperature at the same rate as in the past and a leveling off of stream temperatures (not seen during the study period) is likely (Mohseni and Stefan 1999). If this is the case, then we would expect our predicted MWMT to be biased high for the 13 sites with observed MWMT values already greater than 20[degrees]C.


Regional variation

The salmon streams we monitored varied widely in summertime thermal maxima, with a range of 10.1 to 21.0[degrees]C in MWMT across sites. Although the availability of temperature data for other streams in Alaska is limited, the temperatures we document here are higher than any presently reported data. A previous study of 32 glacial and nonglacial streams in Cook Inlet reported maximum weekly stream temperatures that ranged from 6.9 to 20.6[degrees]C (Kyle and Brabets 2001). MWAT for nine streams in southeast Alaska in 2011 ranged from 4.3 to 18.6[degrees]C, although only three of the sites were on nonglacial rivers (Fellman et al. 2014). A study of 33 streams in the Wood River watershed of western Alaska reported a maximum summertime mean temperature of 12.6[degrees]C (Lisi et al. 2013), but these systems were much smaller than those included in our study and did not include monthly or weekly values. Our maximum recorded temperatures, from the northern end of Pacific salmon distribution, are approximately 4[degrees]C less than thermal maxima observed in salmon streams across the Pacific Northwest (e.g., MWMT in Luce et al. 2014), >1000 km to the south.

The maximum thermal sensitivity we report for nonglacial streams in Cook Inlet (1.51) is greater than those reported in other studies. Thermal sensitivities calculated using year-round daily data for streams in the Columbia River basin ranged from 0.10 to 0.81 (Chang and Psaris 2013), sensitivities based on year-round daily data for streams in Pennsylvania ranged from 0.02 to 0.93 (Kelleher et al. 2012), and thermal sensitivities calculated from August weekly data ranged from 0.50 to 1.26 for streams in the Pacific Northwest (Mayer 2012). Our thermal sensitivities calculated from July weekly data may be higher because the slope of the stream-air temperature regression increases as the time step increases from daily to weekly (Erickson and Stefan 2000). Also, the change in stream temperatures decreases at higher air temperatures due to evaporative cooling (Mohseni et al. 1998), so slopes based on linear regressions from study areas with higher (>20[degrees]C) air temperatures may be biased low (Mohseni and Stefan 1999). Since southcentral Alaska's air temperatures in July are rarely above 20[degrees]C, this would not have affected our results.

Interannual variability in MWAT varied by more than 3[degrees]C during the study period for 33 sites (ranges in MWAT were 1.1 to 6.4 over all sites). Streams in British Columbia with 6 or more years of data had similar or higher variability in MWAT; two sites had standard deviations greater than 3[degrees]C (Moore et al. 2013), while the maximum standard deviation for MWAT in our data set was 2.8[degrees]C. We expect that interannual variability at our sites could be higher than our data set captured, given that the warmest year in our study period (2009) was only the 12th warmest summer over the 35-year baseline period (1980-2014).

Watershed drivers of variation in stream thermal regimes

Watershed elevation and wetland cover were the most important predictors of summer temperature regimes across our sites. We suspect that high-elevation watersheds tended to be colder because these sites had snowpacks that persisted into early June (, extending the inputs of cold meltwater (Caissie 2006; Luce et al. 2014; Lisi et al. 2015). Wetlands, by contrast, lead to warmer stream temperatures because of surface or near-surface flow paths and flat topographies that produce longer residence times and receive more direct radiation than groundwater flow paths (King et al. 2012; Callahan et al. 2014). Watershed elevation and, to a lesser extent, wetland cover were correlated with other predictors representing a gradient of characteristics associated with elevation. Mean watershed elevation was positively correlated with mean watershed slope and negatively correlated to percent forest cover, which decreases with elevation in this high-latitude region. Steeper watersheds receive less incoming solar radiation than horizontal surfaces (depending on aspect; Buffo et al. 1972) and also have shorter water residence times and increased water velocities, all of which lead to colder stream temperatures (Jones et al. 2013; Lisi et al. 2013).

The most important predictors of stream thermal sensitivity were mean watershed elevation and watershed area. The same mechanisms that lead to cold temperatures in high-elevation watersheds likely also moderate those temperatures to warming. Snowmelt contributions at higher elevations buffer stream temperatures to thermal loading during the early part of the summer (Lisi et al. 2015). High-elevation watersheds in our study area were also steeper, so they receive less direct solar radiation throughout the summer, and flowpaths have shorter water residence times during which they may equilibrate with air temperatures (Mayer 2012). We attribute higher thermal sensitivities in larger watersheds to decreased shading in larger and wider streams and also to longer water residence times allowing surface and near-surface flowpaths to equilibrate to surrounding air temperatures.

Our calculations of thermal sensitivity using local air temperatures at coastal sites may be biased high. Of the 15 sites with thermal sensitivity greater than 1.0, 11 were within 6 km of the coast and eight were within 2 km. Air temperatures recorded at these sites are moderated by Cook Inlet and are likely lower than the air temperatures experienced higher in the watershed where much of the warming occurs. Mayer (2012) also observed abnormally high summer sensitivities at coastal sites in Washington.

Other studies have found groundwater inputs to be the predominant driver of thermal sensitivity in streams (Kelleher et al. 2012; Mayer 2012). Groundwater contribution is commonly expressed as a baseflow index derived from continuous streamflow data. We calculated baseflow index for 16 of our streams for which historical or current streamflow data were available using the Webbased Hydrograph Analysis Tool (WHAT; Lim et al. 2005) and the daily baseflow index as calculated by Mayer (2012). Neither index correlated with the residuals of the sensitivity model, indicating that the baseflow indices could not explain variance in the thermal sensitivities of these 16 sites. The years of streamflow data for most sites did not overlap with each other or with our temperature monitoring efforts. However, both of the baseflow indices we calculated correlated with watershed coverage of permeable geology (r = 0.69 for WHAT and r =0.58 for daily baseflow index), which supports our original hypothesis that this watershed characteristic reflects groundwater inputs despite the fact that it was not an important predictor in the sensitivity model.

Snowmelt may be another important predictor controlling differences in stream sensitivity. Lisi et al. (2015) found that snowmelt-dominated streams in Bristol Bay were less sensitive than streams in rain-dominated low-elevation watersheds. They also observed statistically different sensitivities for streams before and after 15 July; high-elevation streams were less sensitive in the early part of the summer due to snowmelt, whereas low-elevation streams were less sensitive later in the summer due to shorter day length and, possibly, increasing rainfall. The magnitude of the previous winter's snowpack and timing of snowmelt likely affects stream sensitivities between years as well as over the summer season. Future investigations into controls on stream thermal sensitivity in Alaska may need to consider more complex models that account for differences in sensitivity as the summer progresses and between summers in addition to differences among sites. Our stream sensitivity response included temporal variability both within and between years that made it difficult to model relative differences between watersheds.

Our research does not explore the variability of stream temperatures likely present within some watersheds, but our results indicate that larger, low-elevation watersheds will be most impacted by increases in air temperature. These findings are further supported by research in other regions that identified the influence of elevation and slope angle on thermal properties of streams and indicate that steep, high-elevation tributaries are likely to provide future refugia for cold-water species (Isaak et al. 2010; Isaak and Rieman 2013; Isaak et al. 2015). We think that our monitoring locations integrated the general stream temperature regimes for each watershed, but a variety of factors (e.g., localized groundwater inputs, snowmelt-fed tributaries, and variable shading from riparian vegetation) undoubtedly create thermal mosaics at smaller spatial scales that were undetectable by our current study design. Future research intensively exploring temperature regimes and thermal sensitivity throughout watersheds will provide useful information for understanding patterns and drivers of variability.

Our summary of predictors controlling variation in stream temperature responses highlights the importance of developing better topographic and hydrologic data for Alaska. Improved DEMs and a networked hydrography data set are likely needed to develop stream temperature models for Alaska with similar performance as models developed in the Pacific Northwest. Including hydrologic predictors (e.g., discharge or baseflow index) in stream temperature models may be especially important in Alaska due to expected changes in the amount, timing, and form of precipitation (McAfee et al. 2014), in addition to interactions between stream hydrology and thawing permafrost (Walvoord and Striegl 2007; Jones and Rinehart 2010). Finally, as we continue to bring together existing temperature data and build new monitoring networks, stream network spatial statistical models are important tools that several studies have used to reduce spatial autocorrelation related to network topology, flow, and longitudinal connectivity (e.g., Isaak et al. 2010).

Potential implications of temperature regimes for salmon

We found that numerous nonglacial watersheds in the Cook Inlet region currently have stream temperatures that exceed threshold MWMT ranges identified by the US Environmental Protection Agency (US EPA) for the protection of salmon life stages. These criteria, above which chronic and sublethal effects become likely, are 13[degrees]C for spawning and egg incubation, 16-18[degrees]C for juvenile rearing, and 18-20[degrees]C for adult migration (US EPA 2003). Surprisingly, even in our relatively cool sampling period, MWMT at most sites exceeded the established criterion for spawning and incubation during every year of the study, which suggests salmon are already experiencing thermal stress in the Cook Inlet region.

While impacts can be expected for populations that are consistently exposed to elevated stream temperatures, a variety of complex physical factors could mediate the effects on salmon populations in the Cook Inlet region. Spawning sites are likely to be distributed across a diversity of stream reaches and habitat types that have different stream temperature regimes and, in some cases, may be driven by the presence of groundwater upwelling (Curry and Noakes 1995; Geist et al. 2002). Our temperature measurements from the lower section of each watershed will not be representative of every spawning location, especially in larger watersheds. In the absence of buffering effects (e.g., stream channel shade, groundwater inputs), stream temperature tends to increase in a downstream direction due to atmospheric warming that occurs along a stream's flow path (Sullivan et al. 1990). Further, a stream's physical structure exerts internal control over water temperature by influencing stream channel resistance to warming or cooling (Poole and Berman 2001) and creates a mosaic of stream temperature patterns within a watershed. While we would expect individuals that use high-elevation tributaries for spawning to be generally buffered from impacts, our research suggests that salmon that use main channel habitat to spawn are likely to be negatively impacted by elevated water temperatures.

Timing of spawning also varies widely among species, populations, and years (Quinn 2005), and each salmon species has a slightly different optimal temperature for growth and survival (Murray and McPhail 1988; Beacham and Murray 1990). Data that would allow us to relate the timing of MWMT to spawning in individual streams do not exist. However, available information from around Cook Inlet suggests that most pink, chum, and Chinook salmon populations and many sockeye salmon populations spawn during July and August (ERT 1984; Burger et al. 1985; Hollowell et al. 2015), during which time high temperatures may result in reduced gamete viability (US EPA 2003). It is also possible that accelerated embryonic development associated with increased incubation temperatures could lead to fry emergence prior to the seasonal onset of optimal rearing conditions (Taylor 2008) and that resulting selective pressure may lead to later spawning (Crozier et al. 2008). Previous research from the Cook Inlet region suggests that fry survival for coho salmon, which spawn during September and October, may be enhanced under warming scenarios, barring substantial modifications to the flow regime (Leppi et al. 2014).

Elevated summer stream temperature could have negative or positive impacts on the region's salmon populations, and the direction of the response will likely be determined by a suite of factors. Exceedances of thermal criteria for juvenile rearing (16-18[degrees]C) occurred in most streams during the warmest summer of the study (2009) and in several streams during cooler years. These exceedances would most directly impact juvenile Chinook and coho salmon, which rear in streams throughout the study area for 1 or 2 years, respectively, and may result in reduced growth and increased vulnerability to disease, predation, and competition (see US EPA 2003). For example, predation by invasive northern pike (Esox lucius), introduced to the Cook Inlet basin in the 1950s, appears to have contributed to a salmon population collapse in at least one low-gradient watershed, and salmonid susceptibility to pike predation is expected to increase as waters warm (Sepulveda et al. 2015). However, impacts on juvenile rearing could be mitigated to some extent by the availability of thermal refugia (Torgersen et al. 1999) or by shifting habitat use to higher elevations (Keleher and Rahel 1996; Isaak and Rieman 2013). Additionally, the impacts of warming waters may not be entirely negative. In colder streams, coho salmon have been shown to exploit spatial thermal heterogeneity by migrating to warmer areas after feeding, which increased their metabolic and growth rates (Armstrong and Schindler 2013; Armstrong et al. 2013). Further, temperatures in many of our study streams were continually below the optimum for salmon growth; warming in these streams may enhance growth in coming decades, given adequate food resources (Beer and Anderson 2011).

Projections for the next ~50 years suggest that if climate scenarios hold and our thermal sensitivity estimates are robust to extrapolation, MWMT in an increasing number of streams will exceed spawning, rearing, and migration temperature criteria by increasing margins. However, our projections also suggest that more than 30% of our study streams will be relatively insensitive to thermal change and continue to provide critical cold-water habitat for salmon populations into the future. We expect temperature-related impacts to be greatest at low-elevation sites because these had the warmest summer temperature regimes and are also warming the fastest (i.e., they showed the highest thermal sensitivity). MWMT in excess of 20[degrees]C during 1 or more years at 13 sites may have affected upstream migrations of adult salmon, five species of which move up the region's streams during summer. These impacts may include delayed migration (Quinn et al. 1997; Salinger and Anderson 2006), increased vulnerability to disease (Fryer and Pilcher 1974; Kocan et al. 2004, 2009), and reduced swimming performance (Brett 1995; Lee et al. 2003). In extreme cases, warming conditions coupled with low water may lead to mass salmon die-offs, as have been observed in Cook Inlet and elsewhere in Alaska (Murphy 1985; Woolsey 2013; Viechnicki 2013; Georgette 2014; Doogan 2015). Many of the region's largest salmon runs migrate up glacier-fed rivers to reach upstream spawning areas. Since glacial meltwater cools rivers considerably (Kyle and Brabets 2001; Fellman et al. 2014), these runs will avoid exposure to warm temperatures along some or all of their riverine migratory route, but the accelerating loss of glacial mass may reduce or eliminate this effect (Milner et al. 2009).

While we can reasonably suggest the above impacts associated with ongoing summer warming, a complex set of additional factors will help shape the responses of the region's salmon populations to increasing greenhouse gas emissions in coming decades. For example, increasing winter streamflow may lead to scouring of spawning redds and mortality of incubating embryos (Leppi et al. 2014) while at the same time increasing the availability of wintering habitat. Advancing spring freshets (Stewart et al. 2005) may lead to increased predation during low flows if smolts do not adjust migration timing or lead to mismatches with optimal ocean rearing conditions if they do. Ocean acidification may disrupt the food webs that support salmon at sea (Fabry et al. 2008). The overall effects of these phenomena, combined with increasing temperatures, are complex and uncertain. And although the Cook Inlet basin currently lacks the widespread human disturbance associated with salmon declines in the Pacific Northwest, high potential exists for future urban impacts especially in lowland, coastal areas where most Alaskans live and where streams have the highest summer temperatures and sensitivity to climate warming. However, targeted management strategies can increase resilience in aquatic ecosystems to a changing climate, such as improving riparian vegetation to shade streams, restoring fish passage to provide access to thermal refugia, and identifying sensitive areas for conservation (Rieman and Isaak 2010; Isaak et al. 2010). These strategies, in addition to maintaining habitat connectivity and complexity, along with salmon's inherent life history diversity and evolutionary potential, will help the long-term viability of the region's salmon populations (Hilborn et al. 2003; Crozier et al. 2008; Schindler et al. 2010; Reed et al. 2011).


This project was made possible in part by Alaska Clean Water Action grants from Alaska Department of Environmental Conservation; US Fish and Wildlife Service through the Alaska Coastal Program, Kenai Peninsula Fish Habitat Partnership, Mat-Su Basin Salmon Habitat Partnership; Alaska EPSCoR NSF award No. OIA-1208927 and the state of Alaska; Alaska Conservation Foundation, George H. & Jane A. Mifflin Memorial Fund, True North Foundation, and Patagonia. Special thanks go to Marcus Geist, Robert Ruffner, Branden Bornemann, Jeff and Gay Davis, and Laura Eldred for their assistance during this project, as well as the 63 individuals from federal and state agencies, nongovernmental organizations, Tribal entities, local communities, and businesses who helped collect temperature data. We also thank two anonymous reviewers for their valuable input to improve this manuscript.


Anderson, D.R. 2008. Model based inference in the life sciences: a primer on evidence. Springer, New York. doi:10.1007/978-0-387-74075-1.

Armstrong, J.B., and Schindler, D.E. 2013. Going with the flow: spatial distributions of juvenile coho salmon track an annually shifting mosaic of water temperature. Ecosystems, 16:1429-1441. doi:10.1007/s10021-013-9693-9.

Armstrong, J.B., Schindler, D.E., Ruff, C.P., Brooks, G.T., Bentley, K.E., and Torgersen, C.E. 2013. Diel horizontal migration in streams: juvenile fish exploit spatial heterogeneity in thermal and trophic resources. Ecology, 94: 2066-2075. doi:10.1890/12-1200.1. PMID:24279277.

Arnold, T.W. 2010. Uninformative parameters and model selection using Akaike's Information Criterion. J. Wildlife Manage. 74:1175-1178. doi:10.1111/ j.1937-2817.2010.tb01236.x.

Beacham, T.D., and Murray, C.B. 1990. Temperature, egg size, and development of embryos and alevins of five species of Pacific salmon: a comparative analysis. Trans. Am. Fish. Soc. 119: 927-945. doi:10.1577/1548-8659(1990)119<0927: TESADO>2.3.CO;2.

Beer, W., and Anderson, J. 2011. Sensitivity of juvenile salmonid growth to future climate trends. River Res. Appl. 27: 663-669. doi:10.1002/rra.1390.

Brabets, T.B. 1999. Water-quality assessment of the Cook Inlet Basin, Alaska--environmental setting. Water-Resources Investigations Report 99-4025. US Geological Survey, Anchorage, Alaska.

Brannon, E.L. 1987. Mechanisms stabilizing salmonid fry emergence timing. Can. Spec. Publ. Fish. Aquat. Sci. 96(1): 120-124.

Brett, J.R. 1995. Energetics. In Physiological ecology of Pacific Salmon. Edited by C. Groot, L. Margolis, and W.C. Clarke. University of British Columbia Press, Vancouver, B.C. pp. 1-68.

Brett, J.R., Shelbourn, J.E., and Shoop, C.T. 1969. Growth rate and body composition of fingerling sockeye salmon, Oncorhynchus nerka, in relation to temperature and ration size. J. Fish. Res. Board Can. 26(9): 2363-2394. doi:10.1139/ f69-230.

Buffo, J., Fritschen, L.J., and Murphy, J.L. 1972. Direct solar radiation on various slopes from 0 to 60 degrees north latitude. USDA Forest Service Research Paper PNW-142. Portland, Ore.

Burger, C.V., Wiimot, W.L., and Wangaard, D.B. 1985. Comparison of spawning areas and times for two runs of chinook salmon (Oncorhynchus tshawytscha) in the Kenai River, Alaska. Can. J. Fish. Aquat. Sci. 42(4): 693-700. doi:10.1139/ f85-089.

Burgmer, T., Hillebrand, H., and Pfenninger, M. 2007. Effects of climate-driven temperature changes on the diversity of freshwater macroinvertebrates. Oecologia, 151: 93-103. doi:10.1007/s00442-006-0542-9. PMID:16964502.

Burnham, K.P., and Anderson, D.R. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd ed. Springer, New York. doi:10.1007/b97636.

Caissie, D. 2006. The thermal regime of rivers: a review. Freshw. Biol. 51:1389-1406. doi:10.1111/j.1365-2427.2006.01597.x.

Callahan, M.K., Rains, M.C., Bellino, J.C., Walker, C.M., Baird, S.J., Whigham, D.F., and King, R.S. 2014. Controls on temperature in salmonidbearing headwater streams in two common hydrogeologic settings, Kenai Peninsula, Alaska. J. Am. Water Resour. Assoc. 51:84-98. doi:10.1111/jawr.12235.

Chang, H., and Psaris, M. 2013. Local landscape predictors of maximum stream temperature and thermal sensitivity in the Columbia River Basin, USA. Sci. Total Environ. 461-462: 587-600. doi:10.1016/j.scitotenv.2013.05.033. PMID: 23756218.

Crozier, L.G., Hendry, A.P., Lawson, P.W., Quinn, T.P., Mantua, N.J., Battin, J., Shaw, R.G., and Huey, R.B. 2008. Potential responses to climate change in organisms with complex life histories: evolution and plasticity in Pacific salmon. Evol. Appl. 1: 252-270. doi:10.1111/j.1752-4571.2008.00033.x. PMID: 25567630.

Curry, RA., and Noakes, D.L.G. 1995. Groundwater and the selection of spawning sites by brook trout (Salvelinus fontinalis). Can. J. Fish. Aquat. Sci. 52(8): 1733-1740. doi:10.1139/f95-765.

Davis, J.C., and Davis, G.A. 2008. Assessment and classification of Matanuska-Susitna fish habitat-stream water temperature. Final Report for the US Fish and Wildlife Service, Aquatic Restoration and Research Institute, Talkeetna, Alaska.

Deb, D., Butcher, J., and Srinivasan, R. 2015. Projected hydrologic changes under mid-21st century climatic conditions in a sub-arctic watershed. Water Resour. Manage. 29: 1467-1487. doi:10.1007/s11269-014-0887-5.

Doogan, S. 2015. Warm water killing fish in Mat-Su and Anchorage [online]. Available from [accessed 15 August 2015].

Erickson, T.R., and Stefan, H.G. 2000. Linear air/water temperature correlations for streams during open water periods. J. Hydrol. Eng. 5: 317-321. doi:10.1061/ (ASCE)1084-0699(2000)5:3(317).

ERT. 1984. Diamond Chuitna Project Aquatic Biology Baseline Studies Report. Vol. I. Text. Environmental Research and Technology, Inc., Fort Collins, Colo.

Fabry, V.J., Seibel, B.A., Feely, RA., and Orr, J.C. 2008. Impacts of ocean acidification on marine fauna and ecosystem processes. ICES J. Mar. Sci. 65: 414432. doi:10.1093/icesjms/fsn048.

Fellman, J.B., Nagorski, S., Pyare, S., Vermilyea, A.W., Scott, D., and Hood, E. 2014. Stream temperature response to variable glacier coverage in coastal watersheds of Southeast Alaska. Hydrol. Processes, 28: 2062-2073. doi:10. 1002/hyp.9742.

Fryer, J.L., and Pilcher, K.S. 1974. Effects of temperature on diseases of salmonid fishes. US Environmental Protection Agency, Office of Research and Development, Ecological Research Series. EPA-660/3-73-020.

Garrett, J.D. 2010. Pervasive thermal consequences of stream-lake interactions in small Rocky Mountain watersheds, USA. Utah State University.

Geist, D.R., Hanrahan, T.P., Arntzen, E.V., McMichael, G.A., Murray, C.J., and Chien, Y. 2002. Physicochemical characteristics of the hyporheic zone affect redd site selection by Chum salmon and fall Chinook salmon in the Columbia River. N. Am. J. Fish. Manage. 22: 1077-1085. doi:10.1577/1548-8675(2002) 022<1077:PCOTHZ>2.0.CO;2.

Georgette, S. 2014. Selawik: Refuge biologists investigate chum salmon die-off and discover a green culprit [online]. Available from fieldnotes/regmap.cfm?arskey=35220 [accessed 10 September 2014].

Grueber, C.E., Nakagawa, S., Laws, R.J., and Jamieson, I.G. 2011. Multimodel inference in ecology and evolution: challenges and solutions. J. Evol. Biol. 24: 699-711. doi:10.1111/j.1420-9101.2010.02210.x. PMID:21272107.

Hilborn, R., Quinn, T.P., Schindler, D.E., and Rogers, D.E. 2003. Biocomplexity and fisheries sustainability. Proc. Natl. Acad. Sci. U.S.A. 100: 6564-6568. doi: 10.1073/pnas.1037274100. PMID:12743372.

Hill, RA., Hawkins, C.P., and Carlisle, D.M. 2013. Predicting thermal reference conditions for U.S.A. streams and rivers. Freshw. Sci. 32: 39-55.

Hollowell, G., Otis, E.O., and Ford, E. 2015. 2014 Lower Cook Inlet area finfish management report. Alaska Department of Fish and Game, Fishery Management Report No. 15-32, Anchorage, Alaska.

Imholt, C., Soulsby, C., Malcolm, I.A., Hrachowitz, M., Gibbins, C.N., Langan, S., and Tetzlaff, D. 2013. Influence of scale on thermal characteristics in a large montane river basin. River Res. Appl. 29: 403-419. doi:10.1002/rra.1608.

Isaak, D.J., and Rieman, B.E. 2013. Stream isotherm shifts from climate change and implications for distributions of ectothermic organisms. Global Change Biol. 19: 742-751. doi:10.1111/gcb.12073. PMID:23504832.

Isaak, D.J., Luce, C.H., Rieman, B.E., Nagel, D.E., Peterson, E.E., Horan, D.L., Parkes, S., and Chandler, G.L. 2010. Effects of climate change and wildfire on stream temperatures and salmonid thermal habitat in a mountain river network. Ecol. Appl. 20(5): 1350-1371. doi:10.1890/09-0822.1. PMID:20666254.

Isaak, D.J., Wollrab, S., Horan, D., and Chandler, G.L. 2011. Climate change effects on stream and river temperatures across the northwest U.S. from 19802009 and implications for salmonid fishes. Clim. Change, 113: 499-524. doi: 10.1007/s10584-011-0326-z.

Isaak, D.J., Young, M.K., Nagel, D.E., Horan, D.L., and Groce, M.C. 2015. The cold-water climate shield: delineating refugia for preserving salmonid fishes through the 21st century. Global Change Biol. 21: 2540-2553. doi:10.1111/gcb. 12879.

Johnson, J., and Coleman, J. 2014. Catalog of waters important for spawning, rearing, or migration of anadromous fishes--Southcentral Region, Effective 1 July 2014. Anchorage, Alaska.

Johnson, M.F., Wilby, R.L., and Toone, JA. 2013. Inferring air-water temperature relationships from river and catchment properties. Hydrol. Processes, 28: 2912-2928. doi:10.1002/hyp.9842.

Johnson, S.L. 2003. Stream temperature: scaling of observations and issues for modelling. Hydrol. Processes, 17(2): 497-499. doi:10.1002/hyp.5091.

Jones, J.B., and Rinehart, A.J. 2010. The long-term response of stream flow to climatic warming in headwater streams of interior Alaska. Can. J. For. Res. 40(7): 1210-1218. doi:10.1139/X10-047.

Jones, L., Muhlfeld, C., Marshall, L., Mcglynn, B., and Kershner, J. 2014. Estimating thermal regimes of bull trout and assessing the potential effects of climate warming on critical habitats. River Res. Appl. 30: 204-216. doi:10.1002/ rra.2638.

Jones, N.E. 2010. Incorporating lakes within the river discontinuum: longitudinal changes in ecological characteristics in stream-lake networks. Can. J. Fish. Aquat. Sci. 67(8): 1350-1362. doi:10.1139/F10-069.

Karl, T.R., Melillo, J.M., and Peterson, T.C. 2009. Global climate change impacts in the United States. National Science and Technology Council, Cambridge University Press.

Keleher, C.J., and Rahel, F.J. 1996. Thermal limits to salmonid distributions in the Rocky Mountain region and potential habitat loss due to global warming: a geographic information system (GIS) approach. Trans. Am. Fish. Soc. 125: 1-13. doi:10.1577/1548-8659(1996)125<0001:TLTSDI>2.3.CO;2.

Kelleher, C., Wagener, T., Gooseff, M., McGlynn, B., McGuire, K., and Marshall, L. 2012. Investigating controls on the thermal sensitivity of Pennsylvania streams. Hydrol. Processes, 26(5): 771-785. doi:10.1002/hyp.8186.

King, R.S., Walker, C.M., Whigham, D.F., Baird, S.J., and Back, J.A. 2012. Catchment topography and wetland geomorphology drive macroinvertebrate community structure and juvenile salmonid distributions in south-central Alaska headwater streams. Freshw. Sci. 31: 341-364. doi:10.1899/11-109.1.

Kocan, R., Hershberger, P., and Winton, J. 2004. Ichthyophonus: an emerging disease of Chinook salmon in the Yukon River. J. Aquat. Anim. Health, 16: 58-72. doi:10.1577/H03-068.1.

Kocan, R., Hershberger, P., Sanders, G., and Winton, J. 2009. Effects of temperature on disease progression and swimming stamina in Ichthyophonusinfected rainbow trout, Oncorhynchus mykiss (Walbaum). J. Fish Dis. 32: 835-843. doi:10.1111/j.1365-2761.2009.01059.x. PMID:19570061.

Kyle, R.E., and Brabets, T.B. 2001. Water temperature of streams in the Cook Inlet basin, Alaska, and implications of climate change. US Geological Survey Water-Resources Investigation Report 01-4109.

Lee, C.G., Farrell, A.P., Lotto, A., MacNutt, M.J., Hinch, S.G., and Healey, M.C. 2003. The effect of temperature on swimming performance and oxygen consumption in adult sockeye (Oncorhynchus nerka) and coho (O. kisutch) salmon stocks. J. Exp. Biol. 206: 3239-3251. PMID:12909705.

Leppi, J.C., Rinella, D.J., Wilson, R.R., and Loya, W.M. 2014. Linking climate change projections for an Alaskan watershed to future coho salmon production. Global Change Biol. 20:1808-1820. doi:10.1111/gcb.12492. PMID:24323577.

Lim, K.J., Engel, B.A., Tang, Z., Choi, J., Kim, K., Muthukrishnan, S., and Tripathy, D. 2005. Automated web GIS based hydrograph analysis tool, WHAT. J. Am. Water Resour. Assoc. 41:1407-1416. doi:10.1111/j.1752-1688.2005. tb03808.x.

Lisi, P.J., Schindler, D.E., Bentley, K.T., and Pess, G.R. 2013. Association between geomorphic attributes of watersheds, water temperature, and salmon spawn timing in Alaskan streams. Geomorphology, 185:78-86. doi:10.1016/j.geomorph. 2012.12.013.

Lisi, P.J., Schindler, D.E., Cline, T.J., Scheuerell, M.D., and Walsh, P.B. 2015. Watershed geomorphology and snowmelt control stream thermal sensitivity to air temperature. Geophys. Res. Lett. 42: 3380-3388. doi:10.1002/2015GL064083.

Lloyd, D.S., Koenings, J.P., and LaPerriere, J.D. 1987. Effects of turbidity in fresh waters of Alaska. N. Am. J. Fish. Manage. 7: 18-33. doi:10.1577/1548-8659 (1987)7<18:EOTIFW>2.0.CO;2.

Luce, C., Staab, B., Kramer, M., Wenger, S., Isaak, D., and McConnell, C. 2014. Sensitivity of summer stream temperatures to climate variability in the Pacific Northwest. Water Resour. Res. 50: 3428-3443. doi:10.1002/2013WR014329.

Mantua, N., Tohver, I., and Hamlet, A. 2010. Climate change impacts on streamflow extremes and summertime stream temperature and their possible consequences for freshwater salmon habitat in Washington State. Clim. Change, 102: 187-223. doi:10.1007/s10584-010-9845-2.

Mauger, S. 2008. Water temperature data logger protocol for Cook Inlet salmon streams [online]. Homer, Alaska. Available from resources/contents/water-temperature-data-logger-protocol.

Mauger, S., Shaftel, R., Trammell, E.J., Geist, M., and Bogan, D. 2015. Stream temperature data collection standards for Alaska: minimum standards to generate data useful for regional-scale analyses. J. Hydrol. Reg. Stud. 4: 431-438. doi:10.1016/j.ejrh.2015.07.008.

Mayer, T.D. 2012. Controls of summer stream temperature in the Pacific Northwest. J. Hydrol. 475: 323-335. doi:10.1016/j.jhydrol.2012.10.012.

McAfee, S.A., Walsh, J., and Rupp, T.S. 2014. Statistically downscaled projections of snow/rain partitioning for Alaska. Hydrol. Processes, 28: 3930-3946. doi: 10.1002/hyp.9934.

Milner, A.M., Brown, L.E., and Hannah, D.M. 2009. Hydroecological response of river systems to shrinking glaciers. Hydrol. Processes, 23: 62-77. doi:10.1002/ hyp.7197.

Mohseni, O., and Stefan, H.G. 1999. Stream temperature/air temperature relationship: a physical interpretation. J. Hydrol. 218: 128-141. doi:10.1016/S00221694(99)00034-7.

Mohseni, O., Stefan, H.G., and Erickson, T.R. 1998. A nonlinear regression model for weekly stream temperatures. Water Resour. Res. 34: 2685-2692. doi:10. 1029/98WR01877.

Moore, R.D., Nelitz, M., and Parkinson, E. 2013. Empirical modelling of maximum weekly average stream temperature in British Columbia, Canada, to support assessment of fish habitat suitability. Can. Water Resour. J. 38:135147. doi:10.1080/07011784.2013.794992.

Morrison, J., Quick, M.C., and Foreman, M.G.G. 2002. Climate change in the Fraser River watershed: flow and temperature projections. J. Hydrol. 263: 230-244. doi:10.1016/S0022-1694(02)00065-3.

Murphy, M.L. 1985. Die-offs of pre-spawn adult pink salmon and chum salmon in southeastern Alaska. N. Am. J. Fish. Manage. 5: 302-308. doi:10.1577/15488659(1985)5<302:DOPAPS>2.0.CO;2.

Murray, C.B., and McPhail, J.D. 1988. Effect of incubation temperature on the development of five species of Pacific salmon (Oncorhynchus) embryos and alevins. Can. J. Zool. 66(1): 266-273. doi:10.1139/z88-038.

Nakagawa, S., and Schielzeth, H. 2013. A general and simple method for obtaining [R.sup.2] from generalized linear mixed-effects models. Methods Ecol. Evol. 4(2): 133-143. doi:10.1111/j.2041-210x.2012.00261.x.

NOAA. 2015. Climate at a glance [online]. National Oceanic and Atmospheric Administration. Available from [accessed 5 May 2015].

O'Driscoll, M.A., and DeWalle, D.R. 2006. Stream-air temperature relations to classify stream-ground water interactions in a karst setting, central Pennsylvania, USA. J. Hydrol. 329:140-153. doi:10.1016/j.jhydrol.2006.02.010.

Pineiro, G., Perelman, S., Guerschman, J.P., and Paruelo, J.M. 2008. How to evaluate models: observed vs. predicted or predicted vs. observed? Ecol. Model. 216(34): 316-322. doi:10.1016/j.ecolmodel.2008.05.006.

Poole, G.C., and Berman, C.H. 2001. An ecological perspective on in-stream temperature: natural heat dynamics and mechanisms of human-caused thermal degradation. Environ. Manage. 27(6): 787-802. doi:10.1007/s002670010188. PMID:11393314.

Prowse, T.D., Wrona, F.J., Reist, J.D., Gibson, J.J., Hobbie, J.E., Levesque, L.M.J., and Vincent, W.F. 2006. Climate change effects on hydroecology of arctic freshwater ecosystems. Ambio, 35: 347-358. doi:10.1579/0044-7447(2006)35[347: CCEOHO]2.0.CO;2. PMID:17256639.

Quinn, T.P. 2005. The behavior and ecology of Pacific salmon and trout. The American Fisheries Society and University of Washington Press.

Quinn, T.P., Hodgson, S., and Peven, C. 1997. Temperature, flow, and the migration of adult sockeye salmon (Oncorhynchus nerka) in the Columbia River. Can. J. Fish. Aquat. Sci. 54(6): 1349-1360. doi:10.1139/f97-038.

Reed, T.E., Schindler, D.E., Hague, M.J., Patterson, D.A., Meir, E., Waples, R.S., and Hinch, S.G. 2011. Time to evolve? Potential evolutionary responses of Fraser River sockeye salmon to climate change and effects on persistence. PLoS ONE, 6(6): e20380. doi:10.1371/journal.pone.0020380. PMID:21738573.

Reynolds, J.H., Murphy, K., and Smith, C. 2013. Alaska Stream and Lake Temperature Monitoring Workshop, Anchorage, Alaska, 5 and 6 November 2012 [online]. Western Alaska Landscape Conservation Cooperative. Available from [accessed 15 May 2013].

Richter, A., and Kolmes, S.A. 2005. Maximum temperature limits for Chinook, coho, and chum salmon, and steelhead trout in the Pacific Northwest. Rev. Fish. Sci. 13: 23-49. doi:10.1080/10641260590885861.

Rieman, B.E., and Isaak, D.J. 2010. Climate change, aquatic ecosystems, and fishes in the Rocky Mountain West: implications and alternatives for management. US Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, Colo.

Ruesch, A.S., Torgersen, C.E., Lawler, J.J., Olden, J.D., Peterson, E.E., Volk, C.J., and Lawrence, D.J. 2012. Projected climate-induced habitat loss for salmonids in the John Day River network, Oregon, U.S.A. Conserv. Biol. 26: 873-882. doi:10.1111/j.1523-1739.2012.01897.x. PMID:22827880.

Ruggerone, G.T., Peterman, R.M., Dorner, B., and Myers, K.W. 2010. Magnitude and trends in abundance of hatchery and wild pink salmon, chum salmon and sockeye salmon in the North Pacific Ocean. Mar. Coast. Fish. Dyn. Manage. Ecosyst. Sci. 2: 306-328. doi:10.1577/C09-054.1.

Salinger, D.H., and Anderson, J.J. 2006. Effects of water temperature and flow on adult salmon migration swim speed and delay. Trans. Am. Fish. Soc. 135: 188-199. doi:10.1577/T04-181.1.

Schindler, D.E., Rogers, D.E., Scheuerell, M.D., and Abrey, C.A. 2005. Effects of changing climate on zooplankton and juvenile sockeye salmon growth in Southwestern Alaska. Ecology, 86:198-209. doi:10.1890/03-0408.

Schindler, D.E., Hilborn, R., Chasco, B., Boatright, C.P., Quinn, T.P., Rogers, L.A., and Webster, M.S. 2010. Population diversity and the portfolio effect in an exploited species. Nature, 465: 609-612. doi:10.1038/nature09060. PMID: 20520713.

Sepulveda, A.J., Rutz, D.S., Dupuis, A.W., Shields, P.A., and Dunker, K.J. 2015. Introduced northern pike consumption of salmonids in Southcentral Alaska. Ecol. Freshw. Fish, 24: 519-531. doi:10.1111/eff.12164.

SNAP. 2014. SNAP data: historical monthly and derived precipitation products--771 m CRU TS [online]. Scenarios Network for Alaska and Arctic Planning, University of Alaska. Available from historical-monthly-and-derived-precipitation-products-771m-cru-ts [accessed 15 January 2014].

Stewart, I.T., Cayan, D.R., and Dettinger, M.D. 2005. Changes toward earlier streamflow timing across western North America. J. Climate, 18: 1136-1155. doi:10.1175/JCLI3321.1.

Sullivan, K., Tooley, J., Doughty, K., Caldwell, J.E., and Knudsen, P. 1990. Evaluation of prediction models and characterization of stream temperature regimes in Washington. Washington Department of Natural Resources Timber/Fish Wildlife Report TFW-WQ3-90-006.

Taylor, S.G. 2008. Climate warming causes phenological shift in Pink Salmon, Oncorhynchus gorbuscha, behavior at Auke Creek, Alaska. Global Change Biology, 14: 229-235. doi:10.1111/j.1365-2486.2007.01494.x.

Torgersen, C.E., Price, D.M., Li, H.W., and McIntosh, B.A. 1999. Multiscale thermal refugia and stream habitat associations of Chinook salmon in northeastern Oregon. Ecol. Appl. 9: 301-319. doi:10.1890/1051-0761(1999)009[0301:MTRASH]2.0.CO;2.

US EPA. 2003. EPA Region 10 Guidance for Pacific Northwest State and Tribal Temperature Water Quality Standards. EPA 910-B-03-002. US Environmental Protection Agency, Region 10 Office of Water, Seattle, Wash.

US EPA. 2015. Climate Change Indicators in the United States: Community Connection: Ice Breakup in Two Alaskan Rivers [online]. US Environmental Protection Agency. Available from science/indicators/snow-ice/river-ice.html [accessed 6 January 2016].

Verdin, K.L. 2004. Estimation of average annual streamflows and power potentials for Alaska and Hawaii. Idaho National Engineering and Environmental Laboratory, Idaho Falls, Idaho.

Viechnicki, J. 2013. Warm weather kills hatchery Chinook near Petersburg [online]. Available from [accessed 24 July 2013].

Walsh, J.E., Chapman, W.L., Romanovsky, V., Christensen, J.H., and Stendel, M. 2008. Global climate model performance over Alaska and Greenland. J. Climate, 21: 6156-6174. doi:10.1175/2008JCLI2163.1.

Walvoord, M.A., and Striegl, R.G. 2007. Increased groundwater to stream discharge from permafrost thawing in the Yukon River basin: Potential impacts on lateral export of carbon and nitrogen. Geophys. Res. Lett. 34: L12402. doi:10.1029/2007GL030216.

Wehrly, K.E., Brenden, T.O., and Wang, L. 2009. A comparison of statistical approaches for predicting stream temperatures across heterogeneous landscapes. J. Am. Water Resour. Assoc. 45: 986-997. doi:10.1111/j.1752-1688.2009. 00341.x.

Wilson, F.H., Hults, C.P., Schmoll, H.R., Haeussler, P.J., Schmidt, J.M., Yehle, L.S., and Labay, K.A. (Compilers). 2012. Geologic map of the Cook Inlet region, Alaska, including parts of the Talkeetna, Talkeetna Mountains, Tyonek, Anchorage, Lake Clark, Kenai, Seward, Iliamna, Seldovia, Mount Katmai, and Afognak 1:250,000-scale quadrangles: U.S. Geological Survey Scientific Inves tigations Map 3153, pamphlet [online]. Available from sim/3153/ [accessed 28 March 2014].

Wobus, C., Prucha, R., Albert, D., Woll, C., Loinaz, M., and Jones, R. 2015. Hydrologic alterations from climate change inform assessment of ecological risk to Pacific salmon in Bristol Bay, Alaska. PLoS ONE, 10(12): e0143905. doi:10.1371/ journal.pone.0143905. PMID:26645380.

Woolsey, R. 2013. As stream levels drop, salmon die before spawning [online]. Available from [accessed 5 September 2013].

Received 15 February 2016. Accepted 23 August 2016.

S. Mauger. Cook Inletkeeper, 3734 Ben Walters Lane, Homer, AK 99603, USA.

R. Shaftel. Alaska Center for Conservation Science, University of Alaska Anchorage, 3211 Providence Dr., Anchorage, AK 99508, USA.

J.C. Leppi. The Wilderness Society, 705 Christensen Dr., Anchorage, AK 99501, USA.

D.J. Rinella.* Alaska Center for Conservation Science, University of Alaska Anchorage, 3211 Providence Dr., Anchorage, AK 99508, USA; Department of Biological Sciences, University of Alaska Anchorage, 3211 Providence Dr., Anchorage, AK 99508, USA.

Corresponding author: Sue Mauger (email:

"Present address: US Fish and Wildlife Service, Anchorage Conservation Office, 4700 BLM Road, Anchorage, AK 99507, USA. 10.1139/cjfas-2016-0076

Published at on 27 September 2016.

(1) Supplementary data are available with the article through the journal Web site at

Caption: Fig. 1. Cook Inlet stream temperature monitoring network data logger locations.

Caption: Fig. 2. Daily mean stream temperatures from 2008 to 2012 at four sites during the open-water season. These sites represent much of the variation in thermal regimes and are examples of small streams with extensive permeable geology (Wasilla Creek); lowland coastal streams (Anchor River); high-elevation, lake systems (Byers Creek); and large, wetland-dominated streams (Deshka River).

Caption: Fig. 3. Maximum weekly maximum temperature (MWMT) for 48 sites over 5 years. The vertical dashed lines represent US EPA's (2003) recommended maximum temperatures for (i) salmon and trout spawning and egg incubation at 13[degrees]C, (ii) migration plus non-core juvenile rearing at 18[degrees]C, and (iii) adult migration at 20[degrees]C. Site locations (latitude-longitude) and complete stream names can be found in the online supplementary material, Table S1 (1).

Caption: Fig. 4. Daily occurrence of maximum weekly maximum temperature (MWMT) across all sites and years.

Caption: Fig. 5. Thermal sensitivity for sites with at least 3 years of stream and air temperature data (n = 45), calculated as the slope of the stream-air temperature relationship based on weekly mean July temperatures.

Caption: Fig. 6. Maximum weekly maximum temperature (MWMT) predictions for 42 streams in Cook Inlet. The range of predicted MWMT values for each time period is based on two climate change scenarios (A1B and A2), which represent possible outcomes depending on future emissions. Observed 2009 MWMT is also plotted for reference. The vertical dashed lines represent US EPA's (2003) recommended maximum temperatures for (i) salmon and trout spawning, egg incubation, and fry emergence at 13[degrees]C, (ii) migration plus non-core juvenile rearing at 18[degrees]C, and (iii) adult migration at 20[degrees]C.

Table 1. Watershed characteristics across the range
of 48 sites in the Cook Inlet basin.

Predictor           Mean   SD    Min.   Max.

Mean watershed      352    254   42     1158
elevation (m)

Watershed area      272    297   10     1623

Wetland cover (%)   12     11    0      40

Lake influence      0.2    0.4   0.0    2.7

Permeable geology   72     29    8      100
cover (%)

Predictor           Hypothesized effect         References

Mean watershed      Higher elevations           Isaak et al. 2010;
elevation (m)       contribute more snowmelt    Lisi et al. 2013;
                    and have colder air         Fellman et al. 2014
                    temperatures (-)

Watershed area      Larger watersheds have      Imholt et al. 2013;
([km.sup.2])        more discharge and higher   Hill et al. 2013;
                    thermal capacity (-), but   Lisi et al. 2013;
                    also have longer water      Chang and Psaris 2013
                    residence times (+)

Wetland cover (%)   Wetlands increase           Callahan et al. 2014
                    surface water
                    residence times (+)

Lake influence      Lakes increase water        Garrett 2010;
(km)                residence times (+)         Jones et al. 2014;
                                                Lisi et al. 2013

Permeable geology   Unconsolidated sediments    Kelleher et al. 2012;
cover (%)           retain groundwater (-)      Mayer 2012;
                                                Hill et al. 2013

Table 2. Relative variable importance ([SIGMA][w.sub.i]) by response
variable for each predictor.


                        Mean watershed   Wetland   Watershed
Temperature             elevation        cover     area
response variable

Mean July temperature   0.99             0.90      0.36
MWAT                    1.00             0.90      0.37
MWMT                    0.96             0.71      0.40
Sensitivity             0.95             0.31      0.74


                        Permeable   Lake
Temperature             geology     influence
response variable       cover

Mean July temperature   0.45        0.33
MWAT                    0.53        0.38
MWMT                    0.33        0.32
Sensitivity             0.28        0.28

Note: MWAT, maximum weekly average temperature; MWMT,
maximum weekly maximum temperature.

Table 3. Confidence set for each stream temperature response
along with model performance measures.

Response        Fixed effects                Model type      AAIC

Mean July       Air + geo + elev + wetland   Mixed effects   0
Mean July       Air + elev + wetland         Mixed effects   0.08
Mean July       Air + elev + area            Mixed effects   3.69
MWAT            Air + geo + elev + wetland   Mixed effects   0
MWAT            Air + elev + wetland         Mixed effects   1.01
MWAT            Air + elev + area            Mixed effects   3.95
MWMT            Air + elev + wetland         Mixed effects   0
MWMT            Air + elev + area            Mixed effects   1.63
MWMT            Air + elev                   Mixed effects   2.08
Sensitivity     Elev + area                  Multiple        0
Sensitivity     Elev                         Multiple        2.44

Response        Weight   [R.sup.2]   [R.sup.2]     [r.sup.2]    RMSE
                         marginal    conditional   observed
                                                   vs. fitted

Mean July       0.47     0.60        0.95          0.58         1.71
Mean July       0.45     0.59        0.95
Mean July       0.07     0.56        0.95
MWAT            0.58     0.62        0.93          0.61         1.83
MWAT            0.35     0.61        0.93
MWAT            0.08     0.58        0.93
MWMT            0.56     0.53        0.91          0.47         2.34
MWMT            0.25     0.51        0.91
MWMT            0.20     0.48        0.90
Sensitivity     0.77     0.25 *      NA            0.28         0.25
Sensitivity     0.23     0.19 *      NA

Note: Models are ordered for each response based on Akaike's
information criterion (AIC) values. The [r.sup.2] for observed versus
fitted values are based on model averaged parameter estimates. RMSE is
root-mean-square prediction error.

* Adjusted [R.sup.2].

Table 4. Parameter estimates (with 85% confidence intervals, CI)
and estimates of effect size.

Response      Predictor           Parameter   Lower
                                  estimate    85% CI

Mean July     Elevation (m)       -0.0044     -0.0062
temperature   Wetland (%)          0.0742      0.0324
              Area ([km.sup.2])    0.0001     -0.0011
              Geology (%)         -0.0070     -0.0218
              July mean air        1.0589      0.9915

MWAT          Elevation (m)       -0.0053     -0.0072
              Wetland (%)          0.0804      0.0347
              Area ([km.sup.2])    0.0001     -0.0012
              Geology (%)         -0.0110     -0.0269
              MWAT air             0.8938      0.8321

MWMT          Elevation (m)       -0.0052     -0.0076
              Wetland (%)          0.0401     -0.0107
              Area ([km.sup.2])    0.0004     -0.0012
              MWMT air             0.6098      0.5588

July          Elevation (m)       -0.0005     -0.0007
sensitivity   Wetland (%)          0.0004     -0.0061
              Area ([km.sup.2])    0.0002      0

Response      Upper     Estimate o
              85% CI    effect size

Mean July     -0.0026   -23
temperature    0.1159   1
               0.0014   848
               0.0078   -14
               1.1263   0.09

MWAT          -0.0033   -19
               0.126    1
               0.0015   727
               0.0048   -9
               0.9556   0.11

MWMT          -0.0028   -19
               0.091    2
               0.0021   229
               0.6609   0.16

July          -0.0003   -197
sensitivity    0.0069   239
               0.0004   525

* Change in predictor to elicit a 0.1[degrees]C increase in mean
July temperature, MWAT, or MWMT or a 0.1 increase in
thermal sensitivity.

Source Citation

Source Citation   

Gale Document Number: GALE|A490692927