MICHIGAN STREAM SALMONID GROWTH AND SURVIVA L IN A CHANGING CLIMATE: PREDIC TIONS AND IMPLICATIONS FOR RESILIENCE - BASED MANAGEMENT By Andrew Kenneth Carlson A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Fisheries and Wildlife Doctor of Philosophy Ecology, Evolutionary Biology, and Behavio r Dual Major 2019 ABSTRACT MICHIGAN STREAM SALMONID GRO WTH AND SURVIVA L IN A CHANGING CLIMATE: PREDIC TIONS AND IMPLICATIONS FOR RESILIENCE - BASED MANAGEMENT By Andrew Kenneth Carlson From biodiversity and aesthetic beauty to recreation and water for human use (e.g., municipal, industr ial, agricultur al ), streams and rivers are socioeconomically and ecologically vital ecosystems. Coldwater streams and their biota are particularly unique, but they are increasingly threatened by climate change and associated temperature warming, changing hydr ology (e.g., grou ndwater input, temperature; precipitation magnitude, intensity, frequency), and modifications to thermal and physical habitats that support aquatic organism growth, reproduction, and survival. As such, monitoring and modeling of stream thermal - hydrological regimes are important for sustainable management of coldwater fishes including Brook Trout ( Salvelinus fontinalis ), Brown Trout ( Salmo trutta ), and Rainbow Trout ( Oncorhynchus mykiss ) in a changing climate. Ultimately, stream salmonid management decisions should foster social - ecological resilience the ability to retain robust stream ecosystems and human systems amid stressors such as climate change and promote resilience - based management. Using Michigan trout streams as a case stud y, the objectives of this dissertation were to: 1) Develop stream - specific temperature models to forecast stream thermal regimes and project thermal habitat suitability for Brook Trout, Brown Trout, and Rainbow Trout growth and survival throughout Michigan a mid climate change; 2) Compare stream - specific and generalized (i.e., region - specific) temperature models relative to their accuracy (i.e., exactness of temperature prediction) and efficiency (i.e., applicability at management - relevant spatial extents) to develop a model implementation and evaluation approach that can be used for salmonid management programs in Michigan and beyond; 3) Integrate stream temperature modeling results with other thermal habitat information (e.g., groundwater input, watershed and ri parian land use/land cover) and trout relative abundance to create a decision - support tool to assist fisheries professionals in operationalizing resilience - based salmonid management within and beyond Michigan in a changing climate; 4) Develop an approach for incorporating precipitation and groundwater into stream temperature modeling and thermal habitat management amid climate change. Climate change will affect Michigan stream trout in ways that vary among streams and populations. In most groundwater - dominat ed streams, thermal habitats can be accurately modeled and effectively managed using a generalized (i.e., region - specific) approach. However, stream - specific temperature modeling is considerably more accurate than a generalized approach in surface runoff - d ominated systems, where the increased resource expenditure (e.g., money, time, personnel) associated with stream - specific modeling may be justified in systems containing high - priority fisheries resources (e.g. trophy individuals, endangered species) . Decis ion - support tool s are valuable for synthesizing biological, hydrological, and thermal data in ways that foster informed management decision - making on local and regional scales. Similarly, developing precipitation - and groundwater - corrected stream temperatu re models is important for accurate, efficient thermal habitat projections that promote resilience - based salmonid management in a changing climate. iv To Mom and Dad, for guidance, patience, and unwavering love v ACKNOWLEDGEMENTS This dissertation is dedicated to Laura and Dan Carlson. Mom and Dad have always been been blessed with. My success in college and graduate school and now, a bu rgeoning career of my dreams reflect their success as parents. Thank you so much, Mom and Dad. Thinking back to my earliest aspirations for a scientific career, I am indebted to many, many people. Thank you especially to Jim Christensen (Mr. C), who saw my scientific potential journey to a Ph.D. Thank you, Mr. C , for l ighti ng a spark and fueling my passion for science. Like it was yesterday, I remember the first time I read a paper written by Dr. Bill Taylor. passion for fisheries. As I searched for Ph.D. advisors, Bill was on the top of my list. What a life - changing experience it was to first meet Bill in Quebec City at the 2014 AFS meeting. Much like the moment I learned I had received the University Distinguished Fellowship to work with Bill. And the moment he and I set foot in Barcelona to present research before an international audience. Bill has made possible the very finest fisheries doctoral program I can imagine. He has broadened and deepened my knowledge of fisherie s in so ma ny ways. Words can scarcely describe how thankful I am for his mentorship and friendship these last four years. I would not write these words without the steadfast support and guidance of those who invested in me early in my life. Thank you to Dan Kovacich, Jacob Montgomery, and Robert Rantanen (White Bear Lake Area High School); Drs. Len Ferrington, Rebecca Montgomery, Jim vi Perry, Andrew Simons, Bruce Vondracek, Peter Sorensen, and Peter Jordan (University of Minnesota); Julie Grecian, Ron Lawre nz, and Geoff Urban (Lee and Rose Warner Nature Center); and Kacie Carlson, Dave Crawford, Guy Schmidt, Will French, Jerry Johnson, Tom Landwehr, Don Pereira, Brad Parsons, Scott Mackenthun, Brian Nerbonne, Jim Levitt, Joel Stiras, and Heidi Wolf (Minnesot a Department of Natural Resources). My Ph.D. would not have been possible without Dr. Jianguo (Jack) Liu, whose brilliance I was fortunate to experience in many ways first through his scientific manuscripts, then through his guidance and mentorship, and now through papers co - authored together. I am also indebted to Dr. Liu for helping me secure a postdoc at Princeton to apply the metacoupling framework (Liu 2017) to build resilient fisheries and a sustainable agricultural system across the world . Thank y ou so much, Dr. Liu! I thank Dana Infante, Kelley Smith, Jess Mistak, Gary Whelan, Jim Dexter, and others at the Michigan Department of Natural Resources for helping me succeed as a recipient of the 2015 - 2016 Janice Lee Fenske Excellence i n Fisheries Fello wship . I also wholeheartedly thank Doug Beard, Gary Roloff, Troy Zorn, Dan Hayes, Bill Porter, David Williams, Mark Rey, Francesca Rey, Abby Bennett, Chuck Krueger, John Hesse, Abby Lynch, Zeenatul Basher, Scott Winterstein, Shawn Carter, Craig Paukert, Ph ilip Strong, Chris Wood, Steve Moyer, and Brian Zupancic for mentoring me in various but invariably important ways throughout my Ph.D. I thank the many devoted fisheries scientists involved with the InFish and NC1189 multistate networks for their camar aderie and collaboration these past four years. In addition, I thank my mentors and colleagues in the MSU Biological Sciences Program particularly Sonya Lawrence, Chuck Elzinga, Jon Stoltzfus, Sibyl Lee, Sachi Gorzenski, S helly Smith, and Tom Harpstead as well as the many students and mentees I was fortunat e to interact with. You added an invaluable service element to my Ph.D. experience one rooted in teaching, vii learning, and mentorship and for that, I am forever grateful. I thank Ben Hollingsworth, Tim Brady, and many other mentors in the Boone and Crock ett Club for providing me an outstanding opportunity to write and publish popular articles with Presidents of this legendary organization. And last, but certainly not least, I thank my friends and col leagues in the MSU Department of Fisheries and Wildlife and EEBB progra m especially teammates on THEE Graduate Assistants championship - winning softball team and those involved in Sunday Football and Wednesday Basketball for providing happiness, joy, an d a sense of balance to a rewarding Ph.D. experience. My doctoral program would not have been possible without generous fieldwork assistance from Betsy Riley, Esox (Essie), and Sam Betances. I am thankful for data acquisition and computer support provided by James Vatter, Kyle Herreman, and Wesley Daniel (Michigan State University); Jennifer Moore Myers (United States Forest Service) ; Tracy Kolb and Todd Wills (Michigan Department of Natural Resources); Henry Quinlan (United States Fish and Wildlife Servic e) ; and Stacy Nelson and Ernie Hain (Nort h Carolina State University). I also thank MSU FW and CSIS staff (especially Sue Nichols, Jennifer Carducci, and Jill Cruth) and the many organizations that provided funding for my research: MSU Graduate School, Dep artment of Fisheries and Wildlife, FW Graduate Student Organization, Michigan DNR, United States Geological Survey, Trout Unlimited, Red Cedar Fly Fishers, Fly Fishers International, and the NASA - MSU program. Finally, I am grateful for the wisdom and guida nce of the late Drs. David W. Willis and Robert A. Klumb, who provided invaluable insight for conducting fisheries research and navigating graduate school. Through students like myself they inspired, their legacies will live on forever. viii TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ......................... xi LIST OF FIGURES ................................ ................................ ................................ ...................... xv INTRODUCTION ................................ ................................ ................................ .......................... 1 LITERATURE CITED ................................ ................................ ................................ ................... 8 CHAPTER 1: PROJECTED I MPACTS OF CLIMATE CHANGE ON STREAM SALMONIDS WITH IMPLICATIONS FOR RESILIENCE - BASED MANAGEMENT ................................ .. 12 Abstract ................................ ................................ ................................ ................................ ..... 13 Introduction ................................ ................................ ................................ ............................... 14 Methods ................................ ................................ ................................ ................................ ..... 16 Study area ................................ ................................ ................................ .............................. 16 Stream - specific regression models ................................ ................................ ........................ 20 Air temperature projections ................................ ................................ ................................ ... 21 Stream temperature and thermal habitat suitability projections ................................ ............ 23 Analyses ................................ ................................ ................................ ................................ . 24 Results ................................ ................................ ................................ ................................ ....... 26 Model accuracy ................................ ................................ ................................ ...................... 26 Thermal habitat suitability: Brook charr ................................ ................................ ............... 26 Thermal habitat suitability: Brown trout ................................ ................................ ............... 33 Thermal habitat suitability: Rainbow trout ................................ ................................ ............ 37 Discussion ................................ ................................ ................................ ................................ . 39 Management implications ................................ ................................ ................................ ...... 42 Acknowledgments ................................ ................................ ................................ ..................... 44 LITERATURE CITED ................................ ................................ ................................ ................. 45 CHAPTER 2: COMPARING STREAM - SPECIFIC TO GENERALIZED TEMPERATURE MODELS TO GUIDE COLDWATER SALMONID MANAGEMENT IN A CHANGING CLIMATE ................................ ................................ ................................ ................................ ..... 52 Abstract ................................ ................................ ................................ ................................ ..... 53 Introduction ................................ ................................ ................................ ............................... 55 Methods ................................ ................................ ................................ ................................ ..... 60 Study area ................................ ................................ ................................ .............................. 6 0 Baseflow measurements ................................ ................................ ................................ ........ 60 Stream - specific regression models ................................ ................................ ........................ 64 Generalized regression models ................................ ................................ .............................. 66 Air temperature projections ................................ ................................ ................................ ... 67 Stream temperature and thermal habitat suitability projections ................................ ............ 68 Analyses ................................ ................................ ................................ ................................ . 69 Results ................................ ................................ ................................ ................................ ....... 70 Temperature projections ................................ ................................ ................................ ........ 70 ix Thermal habitat suitability projections ................................ ................................ .................. 71 Generalized K versus SP models ................................ ................................ ........................... 75 Data - limited streams ................................ ................................ ................................ .............. 78 Model applicability to streams outside Michigan ................................ ................................ .. 78 Discussion ................................ ................................ ................................ ................................ . 81 Acknowledgements ................................ ................................ ................................ ................... 88 LITERATURE CITED ................................ ................................ ................................ ................. 89 CHAPTER 3: SCIENCE TO ACTION: DECISION - SUPPORT TO ADVANCE STREAM TROUT MANAGEMENT IN A CHANGING CLIMATE ................................ ......................... 96 Abstract ................................ ................................ ................................ ................................ ..... 97 Introduction ................................ ................................ ................................ ............................... 98 Methods ................................ ................................ ................................ ................................ ... 102 Study area ................................ ................................ ................................ ............................ 102 Fisheries professional survey ................................ ................................ ............................... 104 Stream temperature ................................ ................................ ................................ .............. 105 Trout population characteristics ................................ ................................ .......................... 111 Groundwater contribution to streamflow ................................ ................................ ............ 111 Watershed and riparian land cover ................................ ................................ ...................... 112 Stream prioritization tool development ................................ ................................ ............... 112 Results ................................ ................................ ................................ ................................ ..... 116 Stream temperature ................................ ................................ ................................ .............. 116 Trout population characteristics ................................ ................................ .......................... 116 Groundwater contribution ................................ ................................ ................................ .... 120 Watershed and riparian land cover ................................ ................................ ...................... 120 Discussion ................................ ................................ ................................ ............................... 121 Acknowledgements ................................ ................................ ................................ ................. 128 APPENDIX ................................ ................................ ................................ ................................ . 130 LITERATURE CITED ................................ ................................ ................................ ............... 140 CHAPTER 4: DEVELOPING PRECIPITATION - AND GROUNDWATER - CORRECTED STREAM TEMPERATURE MODELS TO IMPROVE TROUT MANAGEMENT IN A CHANGING CLIMATE ................................ ................................ ................................ ............ 146 Abstract ................................ ................................ ................................ ................................ ... 147 Introduction ................................ ................................ ................................ ............................. 148 Methods ................................ ................................ ................................ ................................ ... 152 Study area ................................ ................................ ................................ ............................ 152 Temperature and precipitat ion measurements ................................ ................................ ..... 154 Temperature projections ................................ ................................ ................................ ...... 155 Stream temperature models and projections ................................ ................................ ........ 156 Thermal habitat suitability predictions ................................ ................................ ................ 160 Analyses ................................ ................................ ................................ ............................... 161 Results ................................ ................................ ................................ ................................ ..... 162 Stream temperature models and thermal sensitivity ................................ ............................ 162 Stream temperature projections ................................ ................................ ........................... 168 Thermal habitat suitability predictions ................................ ................................ ................ 176 x Discussion ................................ ................................ ................................ ............................... 181 Acknowledgments ................................ ................................ ................................ ................... 190 APPENDIX ................................ ................................ ................................ ................................ . 191 LITERATURE CITED ................................ ................................ ................................ ............... 197 SYNTHESIS ................................ ................................ ................................ ............................... 203 Climate change will affect Michigan stream trout ................................ ................................ .. 205 Stream hydrology influences the effectiveness of thermal habitat modeling approaches ...... 206 Decision - support tools help integrate climate change predictions into trout management ..... 207 Monitoring and modeling precipitation and groundwater are critical for resilience - based salmonid management ................................ ................................ ................................ ............. 208 LITERATURE CITED ................................ ................................ ................................ ............... 210 xi LIST OF TABLES TABLE 1.1. Descriptive information about 30 salmonid streams used for temperature modeling in Michigan, USA. Region refers to Michigan location (i.e., Upper Peninsula [UP], northern Lower Peninsula [NLP], southern Lower Peninsula [SLP]). Subbasin denotes the National Hydrography Dataset subbasin (i.e., 8 - digit Hydrologic Unit Code). Map number refers to stream identifiers in Figure 1.1. Species refers to salmonids present in each stream (i.e., brook charr [BKC], brown trout [BNT], rainbow trout [RBT]). BFI denotes baseflow index, the mean rate of base flow divided by the corresponding mean rate of total streamflow (Neff et al. 2005). ......... 1 8 TABLE 1.2. Stream - specific temperature regression models with standard error (SE) and coefficient of determination ( R 2 ). F (i.e., F 1,7 ) and P values refer to the air temperature (A) parameter used to predict stream temperature (S). R. denotes River and Crk. denotes Creek. .... 2 2 TABLE 1.3. Thermal habitat suitability status (Habitat status) designations and corresponding temperature ranges (Temperature) and growth conditions (Growth) for juvenile and adult brook charr (BKC; Fry et al. 1946; Baldwin 1957; Raleigh 1982a ), brown trout (BNT; Raleigh et al. 1986; Elliott and Hurley 2000; Hay and Young 2006 ), and rainbow trout (RBT; Wurtsbaugh and Davis 1977; Raleigh 1982b ). ................................ ................................ ................................ ........ 2 5 TABLE 1.4. Actual versus projected stream temperatures and thermal habitat suitability (THS) status for br ook charr (BKC), brown trout (BNT), and rainbow trout (RBT) as predicted by stream - specific air - water temperature regression models. Temperatures are predicted in 2006 under the A2 (820 ppm atmospheric CO 2 by 2100) and B1 (550 ppm atmospheric CO 2 by 2100) obtained from the Northeast Climate Science Center. The first and second THS numbers represent statuses associated with actual and projected temperatures, respectively . .................... 2 7 TABLE 1.5. Historical warming (1976 2006) and backcasted stream temperature change (2006 - over corresponding time periods. 2006 and 2012 stream temperatures were backcasted and compared to known temperatures to measure the accuracy of stream - specific temperature models. For 2006 2012, predicted changes in A and S are provided for the A2 and B1 climate forcing scenarios (B1 in parentheses). R. denotes River and Crk . denotes Creek. ....................... 3 0 TABLE 1.6. Projected future warming (2012 2036, 2036 Predicted changes in A and S are provided for the A2 and B1 climate forcing scenarios (B1 in parentheses). R. denotes River and Crk. denotes Creek. ................................ .............................. 3 4 xii TABLE 2.1. Descriptive information about 52 streams used for temperature modeling in Michigan, USA. Sub - basin denotes the National Hydrography Dataset sub - basin (i.e. 8 - digit Hydrologic Unit Code). Map number refers to stream identifiers in Figure 2. 1. Type denotes hydrological influence: groundwater - dominated (GD) streams were distinguished from surface - runoff dominated (RD) systems using a threshold base - flow index (i.e. > 0.60; Neff et al. 2005) . Species: brook charr [BKC], brown trout [BNT], rainbow trout [RBT]). Area: drainage area, km 2 . Elevation: mean catchment elevation, m. Asterisks indicate that streams had historical field - collected air and stream temperatures for development of stream - specific regression models. R. Peninsula, Southern Lower Peninsula, and Upper Peninsula, respectively. ................................ . 6 2 TABLE 2.2. Stream - specific temperature regressions with standard errors (SE), F values (i.e. F 1,7 ), P values, and R 2 values. S and A denote stream temperature and air temperature, Lower Peninsula and Northern Lower Peninsula, respectively. ................................ ................... 6 5 TABLE 2.3. Projected temperature s (° C) and associated thermal habitat suitability (THS) statuses for growth and survival of brook charr (BKC), brown trout (BNT), rainbow trout (RBT), and all three species (All) in groundwater - dominated streams (1 = optimal, 2 = suboptimal, 3 = no growth , 4 = high mortality risk). Actual (i.e., field - measured) temperatures and THS statuses in 2006 are also included. Predictions are derived from stream - specific and generalized models (separated by commas) under the A2 carbon dioxide emissions scenario and are similar to those from the B1 scenario (not included here). R. denotes river, Crk. denotes creek, and NLP denotes the Northern Lower Peninsula of Michigan. ................................ ................................ ................. 7 3 TABLE 2.4. Projected temperatures (° C) and associated thermal habitat suitability (THS) stat uses for growth and survival of brook charr (BKC), brown trout (BNT), rainbow trout (RBT), and all three species (All) in surface runoff - dominated streams (1 = optimal, 2 = suboptimal, 3 = no growth, 4 = high mortality risk). Actual (i.e. field - measured) t emperatures and THS statuses in 2006 are also included. Predictions are derived from stream - specific and generalized models (separated by commas) under the A2 carbon dioxide emissions scenario and are similar to those from the B1 scenario (not included her e). R. denotes river, Crk. denotes creek, and SLP denotes the Southern Lower Peninsula of Michigan. ................................ ................................ ................. 7 6 TABLE 2.5. List of groundwater - dominated streams without actual (i.e. field - measured) 2006 stream temperatures for which generalized models were used to predict temperatures and associated associated thermal habitat suitability (THS) statuses in 2006, 2012, 2036, and 2056 under A2 and B1 emissions (separated by commas). 1 = optimal, 2 = suboptimal, 3 = no growth, 4 = high mortality risk. R. denotes river and Crk. denotes creek. Species abbreviations are as follows: ALL = brook charr, brown trout, and rainbow trout; BKC = brook charr; BNT = brown trout; RBT = rainbow trout. ................................ ................................ ................................ .......... 79 xiii TABLE 2.6. List of brook charr (BKC) and brown trout (BNT) streams located outsi de Michigan with actual temperatures (i.e., field - measured in 2006) and associated thermal habitat suitability (THS) statuses. 2 = suboptimal, 3 = no growth, 4 = high mortality risk. The table also includes projected temperatures (stream - specific [SS] mode l, generalized [Gen] model) and associated THS statuses in 2006. State abbreviations are as follows: Connecticut (CT), Maine (ME), North Carolina (NC), Wisconsin (WI). ................................ ................................ ................................ ... 80 TABLE 3.1. Stream prioritization tool criteria and associated variables, data source s, and weights (i.e., relative importance values) for evaluating and prioritizing streams for fisheries management according to trout population and thermal habitat quality. Abbreviations are as follows: MDNR (Michigan Department of Natural Resources), USG S (United States Geological Survey), and NLCD (National Land Cover Database). ................................ .............................. 10 6 TABLE 3.2. Overall stream rankings based on six stream prioritization tool (SPT) criteria. Columns correspond to six SPT criteria: water temperature (Temp); trout population characteristics (Trout); groundwater contribution to stream flow (GW) expressed as baseflow index, t he mean baseflow divided by total streamflow ; percent watershed (WS) and riparian (Rip) land cover composed of deciduous forest, evergreen forest, mixed fo rest, and grassland; and projected 2056 water temperature (FutTemp). The Trout column was calculated by adding two scores across species (i.e., Brook Trout, Brown Trout, Rainbow Trout): presence/absence (score = 1/0) and relative abundance (score = 1 [low] , 2 [medium], 3 [high]) as defined by Wills Northern Lower Peninsula, Southern Lower Peninsula, and Upper Peninsula. ......................... 117 TABLE 4.1. Michigan stream information and model parameters. Map number refers to stream identifiers in Figure 4. 1. BFI represents base flow index, the mean rate of base flow divided by the corresponding mean rate of total streamflow (Neff et al. 2005). Species denotes the trout species present in each stream (BKT [brook trout], BNT [brown trout], RBT [rainbow trout] , All [all three species]). Year denotes the baseline year and corresponding weather conditions (2016: warm, dry; 2017: cool, wet) from which the model was developed. Other abbreviations denote model intercepts (Int); coefficients for mean daily air temper ature (MDAT), accumulated degree - days above mean summer air temperature (ADD, a measure of groundwater input), and cumulative daily precipitation since July 1 (PR); P values; bias - criterion scores (AICc); adjusted R 2 values ( Adj R 2 ) for groundwater - and precipitation - corrected models; and adjusted R 2 values for MDAT - only models (MDAT Adj R 2 ). ............................... 163 TABLE 4.2. Michigan stream temperatures and projected future temperatures in three conditions of modeled air temperature warming (+1.7°C, +3.4°C, +5.1°C) and thermal sensitivity of groundwater/precipitation (0.0, 0.66, 0.1; in parentheses) based on 2016 weather conditions (i.e., warm, dry). ................................ ................................ ................................ ................................ .. 1 70 T ABLE 4.3. Michigan stream temperatures and projected future temperatures in three conditions of modeled air temperature warming (+1.7°C, +3.4°C, +5.1°C) and thermal sensitivity of xiv groundwater/precipitation (0.0, 0.66, 0 .1; in parentheses) based on 2017 weather conditions (i.e., cool, wet). ................................ ................................ ................................ ................................ .... 174 TABLE 4.4. Thermal habitat suitability for brook trout (BKT), brown trout (BNT), and rainbow trout (RBT) survival and growth in Michigan streams based 2016 weather conditions (i.e ., warm, dry). Table entries report the proportion of streams that are suitable for survival (numerator) and growth (denominator) in three conditions of modeled air temperature warming (+1.7°C, +3.4°C, +5.1°C) and thermal sensitivity of groundwater/precipi tation (0.0, 0.66, 0.1; in parentheses). Streams are classified as groundwater - dominated (GD; base flow index > 0.60), runoff - dominated (RD; base flow index < 0.60), or systems with intermediate groundwater input (Int; base flow index = 0.60; Neff et al., 2005). ................................ ................................ ................. 177 TABLE 4.5. Thermal habitat suitability for brook trout (BKT), brown trout (BNT), and rainbow trout (RBT) survival and growth in Michigan streams based 2017 weather conditions (i.e., cool, wet). Table entries report the proportion of streams that are suitable fo r survival (numerator) and growth (denominator) in three conditions of modeled air temperature warming (+1.7°C, +3.4°C, +5.1°C) and thermal sensitivity of groundwater/precipitation (0.0, 0.66, 0.1; in parentheses). Streams are classified as groundwater - dom inated (GD; base flow index > 0.60), runoff - dominated (RD; base flow index < 0.60), or systems with intermediate groundwater input (Int; base flow index = 0.60; Neff et al., 2005). ................................ ................................ ................. 179 TABLE 4.6 ( Appendix 4.1 ) . Daily water temperatures in Michigan trout streams from late - May to late - October 2018. Abbreviations are as follows: AUS = Au Sable River, BLK = Black River, CED = Cedar River (Upper Peninsula), EBF = East Branch Fox River, MAN = Manistee River, PAN = Paint River, PIG = Pigeon Riv er, PIN = Pine River (Southern Lower Peninsula), STJ = St. Joseph River, STU = Sturgeon River (Upper Peninsula), TAM = Tamarack River. ............. 192 xv LIST OF FIGURE S FIGURE 1. 1. Map of brook charr, brown trout, and rainbow trout streams used for air - stream temperature modeling in Michigan, USA. Streams and corresponding identification numbers are listed in Table 1.1. Triangles denote locations of MDNR stream temperature gauges. ............... 1 7 FIGURE 1.2. Relative influence of baseflow index on thermal sensitivity (i.e., air temperature regression coefficients) of streams in the State of Michigan, USA. The regression equation is: Thermal sensitivity = - 1.78*Baseflow index + 1.38 ( P < 0.01, R 2 = 0.44). ................................ . 29 FIGURE 1.3. Projected temperatures in individual Michigan brook charr streams in 2006, 2012, 2036, and 2056 under A2 (820 ppm atmospheric CO 2 by 2100) and B1 (550 ppm by 2100) scenarios. Plots are organized by region and emissions scenario: (a) Upper Peninsula, A2 scenario; (b) Upper Peninsula, B1 scenario; (c) northern Lower Peninsula, A2 scenario; (d) northern Lower Peninsula, B1 scenario; (e) southern Lower Peninsula, A2 scenario; (f) southern Lower Peninsula, B1 scenario. Dotted lines represent transitions between thermal habitat s uitability statuses. Stream abbreviations are as follows: BRC = Bryan Creek, BLR = Black River, BOR = Boardman River, CAC = Canada Creek, CAR = Carp River, CEC = Cedar Creek, DUC = Duke Creek, EFR = East Branch Fox River, ESR = Escanaba River, IRR = Iron River, LIR = Little Indian River, MAR = Manistee River, MAC = Martin Creek, PIR = Pine River (northern Lower Peninsula), STR = Salmon Trout River, WSR = West Branch Sturgeon River. 3 2 FIGURE 1.4. Projected temperatures in individual Michigan brown trout s treams in 2006, 2012, 2036, and 2056 under A2 (820 ppm atmospheric CO 2 by 2100) and B1 (550 ppm by 2100) scenarios. Plots are organized by region and emissions scenario: (a) Upper Peninsula, A2 scenario; (b) Upper Peninsula, B1 scenario; (c) northern Lowe r Peninsula, A2 scenario; (d) northern Lower Peninsula, B1 scenario; (e) southern Lower Peninsula, A2 scenario; (f) southern Lower Peninsula, B1 scenario. Dotted lines represent transitions between thermal habitat suitability statuses. Stream abbreviations are as follows: BEC = Bear Creek, BOR = Boardman River, BRC = Bryan Creek, CAR = Carp River, CEC = Cedar Creek, CER = Cedar River, DUC = Duke Creek, EFR = East Branch Fox River, ESR = Escanaba River, MAR = Manistee River, PMR = Pere Marquette River, PGR = Pigeon River, PNR = Pine River (southern Lower Peninsula), PIR = Pine River (northern Lower Peninsula), PRC = Prairie Creek, ROR = Rogue River, TAR = Tahquamenon River, WSR = West Branch Sturgeon River. ................................ . 3 6 FIGURE 1.5. Projected temperatures in individua l Michigan rainbow trout streams in 2006, 2012, 2036, and 2056 under A2 (820 ppm atmospheric CO 2 by 2100) and B1 (550 ppm by 2100) scenarios. Plots are organized by region and emissions scenario: (a) Upper Peninsula, A2 scenario; (b) Upper Peninsula, B1 s cenario; (c) northern Lower Peninsula, A2 scenario; (d) northern Lower Peninsula, B1 scenario; (e) southern Lower Peninsula, A2 scenario; (f) southern Lower Peninsula, B1 scenario. Dotted lines represent transitions between thermal habitat xvi suitability stat uses. Stream abbreviations are as follows: CAR = Carp River, CHR = Chocolay River, DAC = Davenport Creek, ELR = Elm River, LMR = Little Muskegon River, MAR = Manistee River, PGR = Pigeon River, PIR = Pine River, ROR = Rogue River, WSR = West Branch Sturge on River, YDR = Yellow Dog River. ................................ ................................ .... 3 8 FIGURE 2.1. Map of 52 brook charr, brown trout, and rainbow trout streams used for air stream temperature modeling in Michigan. Streams and corresponding identification numbers are listed in Table 1. Modified from Carl son et al. (2017). ................................ ................................ .......... 61 FIGURE 2.2. Temperature predictions in representative Michigan streams in 2006, 2012, 2036, and 2056 using stream - specific (dotted lines) and generalized (solid lines) temperature models. Plots are organized by species and emissions scenario: (a) brook charr, A2 scenario; (b) brook charr, B1 scenario; (c) brown trout, A2 scenario; (d) brown trout, B1 scenario; (e) rainbow trout, A2 scenario; (f) rainbow trout, B1 scenario. In each plot, the upper stream is surface runoff - dominated, an d the lower stream is groundwater - dominated. Dotted lines represent transitions between thermal habitat suitability statuses. Stream abbreviations are as follows: BAR = Bark River, BOR = Boardman River, CAR = Carp River, DAV = Davenport Creek, ESR = Escana ba River, MAN = Manistee River. ................................ ................................ ................................ .... 7 2 FIGURE 3.1. Map of 52 Brook Trout, Brown Trout, and Rainbow Trout streams in Michigan, USA. The map was produced using the stream prioritization tool described herein. Map numbers refer to individual streams: Au Sable R. (1), Bark R. (2*), Bear Creek (3*), Black R. (4*), Boardman R. (5*), Brule R. (6), Bryan Creek (7*), Canada Creek (8*), Carlton Creek (9), Carp R. (10*), Cedar Creek (11*), Cedar R. (12*), Cedar R. (13), Chocolay R. (14*), Coldwater R. (15), Cooks Run (16 ), Davenport Creek (17*), Duke Creek (18*), East Branch Fox R. (19*), Elm R. (20*), Escanaba R. (21*), Falls R. (22), Hersey R. (23), Iron R. (24*), Little Indian R. (25*), Little Muskegon R. (26), Manistee R. (27*), Mann Creek (28), Martin Creek (29), Men ominee R. (30), Miller Creek (31), Mosquito Creek (32), Muskegon R. (33), Ogontz R. (34), Paint R. (35), Pere Marquette R. (36*), Pigeon R. (37*), Pine R. (38*), Pine R. (39*), Prairie Creek (40*), Rapid R. (41), Rogue R. (42*), Salmon Trout R. (43*), Silv er Creek (44), South Branch Pine R. (45), St. Joseph R. (46), Sturgeon R. (47), Tahquamenon R. (48*), Tamarack Creek (49), West Branch Maple R. (50), W. Branch Sturgeon R. (51*), Yellow Dog R. (52*). Asterisks indicate that streams had historical field - col lected air and stream temperatures for development of stream - specific regression models. ................................ ................................ ................................ ......... 10 3 FIGURE 3.2. Conceptual map depicting connections between important thermal drivers in Michigan trout streams (i.e., stream prioritization tool criteria [boxes]) a nd resilience - based management that promotes thermally resilient streams. ................................ ............................. 10 7 FIGURE 3.3. Example of how normalized output scores (N_score) for each input criterion (Table 1) were multiplied by manager - defined weights to produce weighted scores (W_score). These scores were then added for all criteria to produce a final output score for each stream reflecting its relative importance for trout management in a changing climate. ........................ 115 xvii FIGURE 3.4. Michigan stream rankings for a stream prioritization tool criteri on about high - quality riparian land cover conditions. Streams were ranked according to the proportion of their riparian zones containing habitats that promote cool, thermally favorable conditions for trout (i.e., deciduous/evergreen/mixed forest, grassland ), with higher rankings indicating greater coverage by thermally favorable habitats. The ten highest - ranked streams by the SPT with respect to riparian land cover conditions are included above. ................................ .................... 122 F IGURE 3.5. Overall Michigan stream importance rankings c onsidering all six stream prioritization tool (SPT) criteria (i.e., water temperature, trout relative abundance, groundwater contribution to stream flow, high - quality watershed land cover, high - quality riparian land cover, projected 2056 water temperature) . Streams with greater scores for individual criteria received higher importance rankings. The ten highest - ranked streams with respect to all six SPT criteria are included above. ................................ ................................ ................................ ..................... 123 FIGURE 4.1. Map of 15 trout streams used for water temperature modeling in Michigan. Streams and corresponding identification numbers are listed in Table 4.1. ............................... 153 FIGURE 4.2. Relationships between predicted mean daily water temperature and observed mean daily air temperature for Michigan trout streams. Graphs (a), (b), and (c ) display examples that span the range of air - stream temperature relationships and corresponding regression statistics (including slope, y - intercept, adjusted R 2 ) observed in this study. ................................ ............. 165 FIGURE 4.3. Comparison of predictions of mean daily water temperature in Michigan streams between linear regressions that used only mean daily air temperature (MDAT) as an independent variable, and models that used both MDAT and an additional predictor (i.e., acc umulated degree - days above mean summer air temperature [ADD], cumulative daily precipitation since July 1 [PR]). Graphs (a), (b), and (c) show streams spanning a gradient of baseflow from runoff - dominated to groundwater - dominated and encompassing the range of air - stream temperature relationships observed in this study. R 2 values are adjusted R 2 . ................................ ................. 166 FIGURE 4.4. Comparison of the distribution of adjusted R 2 values for Michigan stream temperature linear regressions that used only mean daily air temperature ( MDAT only) as an independent variable, and models that used both MDAT and an additional predictor (i.e., accumulated degree - days above mean summer air temperature [ADD], cumulative daily precipitation since July 1 [PR]). ................................ ................................ ................................ .. 167 FIGURE 4.5. Relationships betw een modeled stream thermal sensitivity (°C water/°C air) and groundwater input for the 13 Michigan streams best modeled with MDAT + PR (surface runoff - dominated [RD] streams) and MDAT + ADD (groundwater - dominated [GD] streams). Graphs (a), (b), and (c) dis play stream thermal sensitivities for three conditions of increasing groundwater thermal sensitivity (TS gw = 0.0, 0.66. 1.0). Dotted lines denote transitions between RD and GD streams. ................................ ................................ ................................ ................... 169 xviii FIGURE 4.6. Distribution of projected mean daily water temperature s based on 2016 weather conditions (warm, dry) in three conditions of modeled air temperature warming (+1.7°C, +3.4°C, +5.1°C) and thermal sensitivity of groundwater/precipitation (0.0, 0.66, 0.1) for the 13 Michigan streams best modeled with MDAT + PR and MDAT + ADD. Within each box plot, small dashed lines are means and solid lines are medians. Long dashed lines spanning the entire panel represent upper thermal thresholds for brook trout (BKT), brown trout (BNT), and rainbow trout (RBT), whereas long dott ed lines denote upper thermal thresholds for growth. ........................ 1 71 FIGURE 4.7. Distribution of projected mean daily water temperatures based on 2017 weather conditions (cool, wet) in three conditions of modeled air temperature warming (+1.7°C, +3.4°C, +5.1°C) and thermal sensitivity of groundwater/precipitation (0.0, 0.66, 0.1) for the 13 Michigan streams best modeled with MDAT + PR and MDAT + ADD. Within each box plot, small dashed lines are means and solid lines are medians. Long dashed lines spanning the entire panel represent upper thermal thresholds for brook trout (BKT), brown trout (BNT), and rainbow trout (RBT), whereas long dotted lines denote upper thermal thresholds for growth. ........................ 173 1 INTRODUCT ION 2 and oceans and in the process provide important ecological goods and services to humanity (e.g., recreation; water for municipal, industrial, and agricultural use; L oomis et al. 2000). However, stressors such as climate change, invasive species introductions, land - use alterations, and habitat fragmentation (e.g., dam installation) are threatening riverine ecosystems across the globe (Dudgeon et al. 2006, Reid et al. 2 018). As these stressors continue to alter water availability, water quality, and related ecosystem structure and function in streams and rivers, aquatic and fisheries managers have started to embrace resilience as a guiding principle for maintaining and e nhancing aquatic ecosystem conservation (Hansen et al. 2015). Resilience is the capacity of a system to withstand disturbance and thereby retain its structure and function (Holling 1973) and continue to provide benefits desired by humans. Currently, climat e change is particularly threatening to the resilience of streams and rivers worldwide as it increases air and water temperature and alters local and regional hydrology (e.g., increased groundwater temperature, more or less frequent precipitation), often r educing the availability and quality of physical, chemical, and thermal habitats for aquatic organisms (Woodward et al. 2010; Snyder et al. 2015). Brook Trout ( Salvelinus fontinalis ), Brown Trout ( Salmo trutta ), and Rainbow Trout ( Oncorhynchus mykiss ) are icons of coldwater streams and excellent indicator species for evaluating the effects of climate change on coldwater stream fisheries. Short - and long - term changes in climate that increase stream temperatures can negatively affect the growth, reproduction , survival, and population resilience of these salmonids and other coldwater fishes 2009; Dunham et al. 2009). Hence, it is important to predict the impacts of climate change on coldwater streams and their fisheries to maintain healthy coldwater ecosystems and associated 3 ecological services (e.g., recreational fishing, flood control) locally, regionally, and globally. An effective way to project climate change ef fects on fish growth, reproduction, survival, and population resilience is to develop models that predict stream temperatures as a function of changing climatic, hydrological , and landscape variables (e.g., air temperature, precipitation, groundwater , riparian/watershed land cover ). Stream temperature predictions can then be integrated with other local and regional dat a (e.g., fish abundance, growth ) to develop informed strategies for groundwater conservation, riparian habitat protection/rehabilitation, and other trout management activities to increase the resilience of coldwater streams and trout fisheries that would otherwise be vulnerable to thermal warming. Sustaining coldwater streams and salmonid fisheries in a changing climate requires management and governance programs that promote thermal resilience (i.e., ability of coldwater streams to absorb thermal changes and retain their original thermal regimes). In turn, thermal resilience promotes social - ecological resilience (i.e., ability of salmonid management programs to retain the structure and function of coldwater ecosystems and human systems amid change). By focusing on thermal resilience and social - ecological resilience, fisheries and aquatic resource professionals can implement resilience - based salmonid management: maintaining and enhancing the structure and function of coldwater stream ecosystems and allied human systems via collaborations among aquatic and terrestrial scientists, managers, policymakers, and public and private stakeholders. For example, fisheries professionals can promote resilience - based salmonid management by protecting and rehabilitating riparian zones to ensure that they have abundant trees, shrubs, grasses, and wildflowers in streams and rivers containing coldwater species projected to be negatively affected by climate change. Such riparian habitat management can greatly increase the thermal resilience of coldwater streams, thereby maintaining and enhancing 4 the growth, reproduction, and survival of coldwater fishes ( Waco & T aylor 2010) . In addition, fisheries professionals can promote resilience - based salmonid management by developing other forms of freshwater recreation in ways th at extend beyond single species or groups (e.g., coldwater fishes) to encompass entire fish communities and ecosystem services (Paukert et al. 2016). Such programs can promote resilience - based salmonid management by increasing stakeholder adaptiveness in a changing climate and thereby enhancing social - ecological resilience, allowing fisheries and aquatic resource managers to focus their limited resources (e.g., time, money) on conserving the highest - priority coldwater streams and trout populations. The goa l of this dissertation is to forecast the effects of climate change on coldwater streams and trout populations in Michigan to provide fisheries professionals with a knowledge base for implementing resilience - based salmonid management programs. In particula r, there is a need to develop stream temperature monitoring and modeling approaches that enable fisheries professionals to understand and predict thermal habitat conditions for trout growth and survival now and in the future . In turn, fisheries professiona ls need to be able to use this information in a decision - support framework, laying a foundation for resilience - based salmonid management in Michigan and throughout the United States and the world . Brook Trout , Brown Trout, and Rainbow Trout are distributed throughout more than 12 ,000 miles of designated coldwater trout streams in Michigan, supporting socioeconomically valuable recreational fisheries in which more than 585,000 anglers spent 8.2 million angling days in 2011 ( Godby et al. 2007; USFWS 2011; MLS A 2018). In addition to the numerical abundance, wide geographic distribution, and socioeconomic importance of Michigan stream trout fisheries, coldwater stream temperatures have been monitored for long time periods (i.e., 10 20 years), making Michigan an ideal study 5 area for investigating how thermal warming will impact coldwater streams and trout populations. However, climate change is projected to increase stream temperatures in Michigan and other Midwestern United States by 0.8 4.0 °C (Pilgrim et al. 19 98; Lyons et al. 2010). This will likely cause some streams to exceed thermal optima for Brook Trout, Brown Trout, or Rainbow Trout, causing declines in growth, reproduction, and survival, particularly during the warmest period of the year (i.e., July; Zor n et al. 2011). Hence, the need for science - driven, resilience - based salmonid management has never been greater. Such a management program requires the following (i.e., objectives of this dissertation): 1) predicting future stream temperatures and trout th ermal habitat conditions; 2) comparing the advantages and disadvantages of different stream temperature projection models; and , 3) using stream temperature predictions in combination with hydrological, biological (i.e., trout population), and human dimensions data to make informed, resilience - based salmonid management decisions . The four chapters of this dissertation address the objectives described above. In collaboration with the Michigan Department of Natural Resources (MDNR), I selected Michigan coldwater streams that spanned a thermal gradient from north to south and a hydrological gradient from surface runoff to groundwater dominance so that my study systems encompassed the diversity of thermal and hydrological conditions experienced by Michigan stream trout. In addition, all streams were important from a fisheries management standpoint as they supported the wild production of, and recreational fisheries for, Brook Trout , Brown Trout, Rainbow Trout, or a combination of these species. In these streams, I measured hourly temperatures from late May through mid - October for three years (2016 - 2018) and developed stream - specific temperature models to forecast trout thermal habitat suitability until 2056 in a changing climate 6 (Ch apter 1 ), with the goal of enabling fisheries professionals to anticipate future thermal habitat conditions and thereby create resilienc e - based salmonid management programs. Although stream - thermal regime (e.g., solar radiation, groundwater input), they are not always possible to use amid limitations in management resources (e.g., time, money, personnel). Hence, I developed generalized (i.e., region - specific) temperature models, compared them to stream - specific models with respect to accuracy (i.e., exactness of temperature prediction) and efficiency (i.e ., applicability at management - relevant spatial extents), and made recommendations for trout thermal habitat monitoring and management (Chapter 2 ). In conducting research for Chapters 1 and 2 and interacting with the MDNR and the United States Geological S urvey ( USGS), I came to understand that when managing stream trout populations, fisheries professionals must integrate diverse sources of information, including stream temperatures, the factors that influence those temperatures (e.g., stream hydrology, rip arian and watershed land cover), and trout population parameters (e.g., fish abundance, growth) while considering the perspectives and needs of fisheries stakeholders (e.g., policy makers, anglers, general public). To help facilitate this process, I design ed and implemented a decision - support tool based on a formal survey of MDNR fisheries professionals regarding their opinions on and attitudes toward trout management in Michigan given a changing climate ( Chapter 3 ) . The decision - support tool synthesized st ream temperature projections with other trout management decision - making criteria (e.g., groundwater input, riparian and watershed land cover, trout relative abundance) to prioritize streams for resilience - based salmonid management in a changing climate (Rohweder et al. 2015a,b). Finally, recognizing that projected changes in climate could significantly impact trout thermal habitat quality and quantity via changes to precipitation and groundwater regimes 7 ( generally not i ncorporated into stream temperature models ) , I developed an approach for predicting the effects of changing precipitation and groundwater on growth and survival of Brook Trout , Brown Trout, and Rainbow Trout (Chapter 4 ). In particular, I evaluated the accuracy and eff iciency of precipitation - and groundwater - corrected stream temperature models relative to conventional air - stream temperature models to determine if fisheries professionals would benefit from monitoring and modeling precipitation and groundwater in their t rout management programs. This research suggests that climate change will affect Michigan stream trout populations in ways that are expected (e.g., direct stream temperature warming) and ways that are less anticipated (e.g., changes in precipitatio n and groundwater dynamics ). Whereas the thermal habitat suitability of some systems for trout growth and survival will decline from present and historical levels, thermal habitats will likely become more suitable in other streams. Overall, Michigan stream s and trout populations are changing and will continue to change (Carlson et al. 2017a, b; Carlson et al. 2018; Zorn et al. 2018). This dissertation provides fisheries professionals and allied stakeholders throughout Michigan, the United States, and the wo rld with a knowledge base to address these changes via stream temperature monitoring and modeling approaches and decision - support tools. In turn, these approaches and tools are the raw material for resilience - based salmonid management that maintains and en hances the structure and function of coldwater stream ecosystems and allied human systems via collaborations among aquatic and terrestrial scientists, managers, policymakers, and public and private stakeholders . 8 LITERATURE CITED 9 LITERATURE CITED Carlson, A. K., W. W. Taylor, K. M. Schlee, T. G. Zorn, and D. M. Infante. 2017a (Chapter 1) . Projected impacts of climate change on stream salmonids with implications for resilience - based management. Ecology of Freshwater Fish 26: 190 20 4. Carlson, A. K., W. W. Taylor, K. M. Hartikainen, D. M. Infante, T. Do uglas Beard, Jr., and A. J. Lynch. 2017 b (Chapter 2) . Comparing stream - specific to generalized temperature models to guide salmonid management in a changing climate. Reviews in Fish Biology and Fisheries 27: 443 462. Carlson, A. K., W. W. Taylor, Z. Basher, T. Do uglas Beard, Jr., and D. M. Infante. 2018 (Chapter 3). Science to a ction: d ecision - support to advance stream t rout m anagement in a c hanging c limate . Proceedings of the Wild Trout XII Symposium 12:85 92. Carlson, A. K., W. W. Taylor, D. M. Infante. In review (Chapter 4). Developing precipitation - and groundwater - corrected stream temperature models to improve management of brook charr amid climate change. Dudgeon, D., A. H. Arthington , M. O. Gessner , Z. I. Kawabata , D. J. Knowler, C. Lévêque, R. J. Naiman, A. H. Prieur - Richard , D. Soto, M. L. J. Stiassny , and C. A. Sullivan. 2006. Freshwater bio diversity: importance, threats, status and conservation cha llenges. Biological Rev iews 81: 163 182. Dunham, J. B., A. E. Rosenberger, R. F. Thurow , C. A. Dolloff , and P. J. Philip . 2009. Coldwater fish in wadeable streams [Chapter 8]. In: S. A. Bonar, W. A. Hubert, D. W. Willis, eds. Standard methods for sampling North American freshwater fishes. Bethesda, MD: American Fisheries Society, pp. 119 138. Godby, N. A., Jr, E. S. Rutherford, and D. M. Mason. 2007. Diet, feeding rate, growth, mortal ity, and production of juvenile steelhead in a Lake Mic higan tributary. North American Jour nal of Fisheries Management 27: 578 592. Hansen, G. J. A., J. W. Gaeta, J. F. Hansen, and S. R. Carpenter. 2015. Learning to manage and managing to learn: sustaining freshwater recreational fisheries in a changing environment. Fisheries 40:56 64. Holl ing, C. S. 1973. Resilience and stability of ecological systems. Annual Review of Ecology and Systematics 4:1 23. Loomis, J., P. Kent, L. Strange, K. Fausch, and A. Covich. 2000. Measuring the total economic value of restoring ecosystem services in an imp aired river basin: results from a contingent valuation survey. Ecol ogical Economics 33:103 - 117. 10 Lyons, J . , T. Zorn , J. Stewart, P. Seelbach, K. Wehrly, and L. Wang. 2009. Defining and chara cterizing coolwater streams and their fish assemblages in Michigan and Wisconsin, USA. North American Journal of Fisheries Management 29 :1130 1151. Michigan Lake Stewardship Associations (MLSA) . 2018. Managing our streams and rivers. Accessed 11 December 2018 < https://www.mymlsa.org/managing - michigan - streams/ >. Paukert, C. P., B. A. Glazer, G. J. A. Hansen, B. J. Irwin, P. C. Jacobson, J. L. Kershner, B. J. Shuter, J. E. Whitney, and A. J. Lynch. 2016. Adapting inland fisheries management to a Pilgrim, J. M., X. Fang, and H. G. Stefan. 1998. Stream temperature correlations with air temperatures in Minnesota: implications for climate warming. Journal of the American Water Resources Association 34:1109 1121. Raleigh, R. F. 1982a. Habitat suitability index models: brook trout: U.S. Fish and Wildli fe Service Biological Report 82 (10.24). 42 pp. Raleigh, R. F. 1982b. Habitat su itability index models: rainbow trout: U.S. Fish and Wildli fe Service Biological Report 82 (10.60). 76 pp. Raleigh, R. F., L. D. Zuckerman, and P. C. Nelson. 1986. Habitat suitability index models and instream flow suitability curves: brown trout: U.S. Fish and Wildlife Service Biological Report 82(10.124). 65 pp. Reid, A. J., A. K. Carlson, I. F. Creed, E. J. Eliason, P. A. Gell, P. T. J. Johnson, K. A. Kidd, T. J. MacCormack, J. D. Olden, S. J. Ormerod, J. P. Smol, W. W. Taylor, K. Tockner, J. C. Vermaire, D. Dudgeon, and S. J. Cooke. 2018. Emerging threats and persistent conservation challenges for freshwater biodiversity. Biological Reviews: XX:X XX XXX. Rohweder, J., S. Vacek, S. M. Crimmins, and W. E. Thogmartin. 2015a. A case study of assigning conservation value to dispersed habitat units for conservation planning. Journal of Conservation Planning 11:13 27. Rohweder, J., S. Vacek, W. E. Thog martin, and S. M. Crimmins. 2015b. Management unit prioritization tools. U.S. Fish and Wildlife Service, La Crosse, Wisconsin, USA. Snyder, C. D., N. P. Hitt, and J. A. Y oung. 2015. Accounting for groundwater in stream fish therm al habitat responses to climate cha nge. Ecological Applications 5: 1397 1419. T homas, S. M., S. W. Griffiths, and S. J. Ormerod, 2015. Beyond cool: adapting upland streams for climate change using riparian woodlands. Global Change Biology 22: 310 3 4. USFWS (United States Fish and Wildlife Service). 2011. 2011 National survey of fishing, hunting, and wildlife - associated recreation. Washington, DC: U.S. Department of the 11 Interior, U.S. Fish and Wildlife Service, and U.S. Department of Commerce, U.S. Cens us Bureau. Waco, K. E., and W. W. Taylor. 2010. The influence of groundwater withdrawal and land use changes on brook charr ( Salvelinus fontinalis ) thermal habitat in two coldwater tributaries in Michigan, USA. Hydrobiologia 650:101 116. Woodward, G., D. M. Perkins, and L. E. Brown, 2010. Climate change and freshwater ecosystems: impacts across multiple levels of organization. Philosophical Transactions of the Royal Soci ety B: Biological Sciences 365: 2093 2106. Zorn, T. G., P. W. Seelbach, and M. J. W iley. 2011. Developing user - friendly habitat suitability tools from regional stream fish survey data. North American Journal of Fisheries Management 31:41 55. Zorn, T. G., T. A. Cwalinski, N. A. Godby, Jr., B. J. Gunderman, and M. A. Tonello. 2018. Manage ment plan for inland trout in Michigan. Michigan DNR Fisheries Division, Fisheries Report No. 30, Lansing. 12 CHAPTER 1: PROJECTED IMPACTS OF CLIMATE CHANGE ON STREAM SALMONIDS WITH IMPLICATIONS FOR RESILIENCE - BASED MANAGEMENT Carlson, A. K., W. W. Taylor, K. M. Schlee, T. G. Zorn, and D. M. Infante. 2017. Projected impacts of climate change on stream salmonids with implications for resilience based management The content of this chapter is intended to be identical to the published manuscript cited above and reflects specifications (e.g. , formatting ) for Ecology of Freshwater Fish . Any differences should be minor and are unintended. 13 Abstract The sustainability of freshwater fisheries is increasingly affected by climate warming, habitat alteration, invasive species, and other drivers of global change. The State of Michigan, USA, contains ecologically, socioeconomically valuable coldwater stream salm onid fisheries that are highly susceptible to these ecological alterations. Thus, there is a need for future management approaches that promote resilient stream ecosystems that absorb change amidst disturbances. Fisheries professionals in Michigan are responding to this need by designing a comprehensive management plan for stream brook charr ( Salvelinus fontinalis ), brown trout ( Salmo trutta ), and rainbow trout ( Oncorhynchus mykiss ) populations. To assist in developing such a plan, we used stream - speci fic regression models to forecast thermal habitat suitability in streams throughout Michigan from 2006 2056 under different predicted climate change scenarios. As baseflow index (i.e., relative groundwater input) increased, stream thermal sensitivity (i.e., relative susceptibility to temperature change) decreased. Thus, the magnitude of temperature warming and frequency of thermal habitat degradation were lowest in streams with the highest basefl ow indices. Thermal habitats were most suitable in rainbow trout streams as this species has a wider temperature range for growth ( 12.0 22.5°C ) compared to brook charr ( 11.0 20.5°C ) and brown trout ( 12.0 20.0°C ). Our study promotes resilience - based salmoni d management by providing a methodology for stream temperature and thermal habitat suitability prediction. Fisheries professionals can use this approach to protect coldwater habitats and drivers of stream cooling and ultimately conserve resilient salmonid populations amidst global change. KEYWORDS: brook charr, brown trout, rainbow trout, Michigan, temperature, resilience 14 Introduction As climate warming , habitat fragmentation, invasive species, and other drivers of global change alter aquatic ecosystem s throughout the world, managing fisheries for resilience has become an important conservation framework (Hansen et al. 2015). Resilience is the capacity of a syst em to absorb disturbances and retain its structure and function ( Holling 1973 ) . Managing aquatic ecosystems for resilience is particularly important when they contain species that are sensitive to ecological stressors. For instance, with relatively low thermal tolerance thresholds ( Raleigh 1982a,b; Raleigh et al. 1986), salmonid fishes may exp erience reductions in growth and survival due to temperature elevation caused by climate change. Thus, managing these species and their ecosystems for thermal resilience is an important task. Collaboration among scientists , managers, policy makers, and pub lic stakeholders will be important for developing management approaches commensurate with the wide geographic distribution and high socioeconomic value of salmonid populations ( Isaak et al. 2015; Snyder et al. 2015) . Tem perature exerts a fundamental ph ysi ological influence on fish metabolism, which regulates growth, survival, and reproduction of individuals ( Dodds and Whiles 2010) and ultimately the dynamics of populations and communities. For example , water temperature is an important factor determining fish distribution and assembla ge composition (Magnuson et al. 1979). Temperatures above species - specific thermal maxima cause mortality; temperatures below maxima alter fish growth, reproduction, abundance, and population size structure (Magnuson et al. 19 97). In addition, t hermal warming can indirectly decrease fish growth and survival by degrading water quality (e.g., reduced d issolved oxygen; Ficklin et al. 2013). Climate change is predicted to increase stream temperatures and thereby alter thermal habi tat suitability (Lyons et al. 2010) and fish comm unity composition (Isaak et al. 2012). 15 Moreover, dams, culverts, and other anthropogenic barriers increase water temperatures and decrease population and habitat connectivity ( Lessard 2000; Hayes et al. 2006 ). Regardless of mechanism, stream salmonids are sensitive to thermal warming because they are adapted to cold and coolwater environments ( Raleigh 1982a,b; Raleigh et al. 1986 ; Wehrly et al. 2007) . Climate - driven increases in water temperature may reduce salmonid growth, reproduction, and survival in streams currently near thermal optima . Thus, it is imperative that fisheries professionals develop management strategies to promote thermal resilience and thereby conserve salmonid populations in a warming climate. The S tate of Michigan, USA, has ecologically and socio economically valuable p opulations of brook charr ( Salvelinus fontinalis ), brown trout ( Salmo trutta ), and rainbow trout ( Oncorhy nchus mykiss ) distributed throughout 31,000 km of streams (Godby et al. 2007; Tyler and Rutherford 2007). Projected air temperature warming resulting from climate change is predicted to increase stream temperatures in the Midwestern United States by 0.8 4. 0°C ( Pilgrim et al. 1998 ; Lyons et al. 2010). If temperatures exceed thermal optima for these species, growth, reproduction, and/or survival will decline, particularly during the warmest month of the year (i.e., July ; Zorn et al. 2011) . Thus, Michigan is an ideal study area for investigating how thermal warming will impact coldwater streams and salmonid population dynamics. This information will enable fisheries professionals to develop resilience - based management programs: collaborative efforts among sci entists, managers, policy makers, and public stakeholders to maintain stream ecosystem structure and function amidst global change. The goal of this study was to evaluate effects of projected air temperature warming on salmonid thermal habit at suitabili ty in coldwater streams in Michigan to facilitate development of a resilience - based salmonid management program. Our first objective was to measure the 16 accuracy of stream - specific air - water temperature regression models by backcasting stream temperatures in 2006 and 2012, years with pre - existing air and stream temperature metrics. Our second objective was to forecast stream temperatures in 2036 and 2056 a nd project thermal habitat suitability for brook charr, brown trout, and rainbow trout growth and survival . We predicted that water temperatures would increase overall from 2006 2056 but unevenly among streams and time periods (e.g., 2012 2036, 2036 2056) due to system - specific patterns of thermal warming and temporal variability in projected carbon dioxide (CO 2 ) emissions (Arblaster et al. 2014 ) . We expected that thermal habitat impairment would occur more frequently in streams with extensive surface runoff than in groundwater - dominated, thermally buffered systems (Sear et al. 1999; Krider et al. 2013). Methods Study area Our study encompassed 30 coldwater streams throughout the S tate of Michigan, USA (F igure 1. 1) . Streams were chosen in three Michigan Department of Natural Resources (MDNR ) management regions (i.e., Upper P eninsula [UP] , northern Lower Peninsula [ N LP] , and southern Lower Peninsu la [ SLP ] ; Table 1 .1 ) to conduct the study at a statewide scale relevant for salmonid management that also spanned a latitudinal thermal gradient in which temperatures increased from north (47.03°N) to south (42.64°N ). In addition, streams were selected to encompass a hydrological gradient from surface - runoff to groundwat er dominance by evaluating their relative base flow, the component of streamflow attributable to groundwater. We obtain ed base flow index (BFI), the mean rate of base flow divided by the corresponding mean rate of total streamflow, using a United States Geolog ical Survey report (Neff et al. 2005) . 17 FIGURE 1. 1. Map of brook charr, brown trout, and rainbow trout streams used for air - stream temperature modeling in Michigan, USA. Streams and corresponding identification numbers are listed in Table 1 .1 . Triangles denote locations of MDNR stream temperature gauges. 18 TABLE 1 .1 . Descriptive information about 30 salmonid streams used for temperature modeling in Michigan, USA. Region refers to Michigan location (i.e., Upper Peninsula [UP], northern Lower Peninsula [NLP], southern Lower Peninsula [S LP]). Subbasin denotes the National Hydrography Dataset subbasin (i.e., 8 - digit Hydrologic Unit Code). Map number refers to stream identifiers in Figure 1. 1. Species refers to salmonids present in each stream (i.e., brook charr [BKC], brown trout [BNT], rainbow trout [RBT]). BFI denotes baseflow index, the mean rate of base flow divided by the corresponding mean rate of total streamflow (Neff et al. 2005). Stream Region Sub - basin Map number Species BFI Bark River UP Cedar - Ford 18 BNT 0.60 Bear Creek SLP Kalamazoo 14 BNT 0.58 Black River NLP Black 1 BKT 0.63 Boardman River NLP Boardman - Charlevoix 2 BKT, BNT 0.58 Bryan Creek UP Escanaba 19 BKT 0.62 Canada Creek NLP Black 3 BKT 0.63 Carp River UP Carp - Pine 20 BKT, BNT, RBT 0.64 Cedar Creek SLP Lower Grand 10 BKT, BNT 0.50 Cedar River SLP Cedar - Ford 11 BNT 0.60 Chocolay River UP Betsy - Chocolay 25 RBT 0.62 Davenport Creek UP Brevoort - Millecoquins 26 RBT 0.65 Duke Creek SLP Lower Grand 15 BKT, BNT 0.50 East Branch Fox River UP Manistique 27 BKT, BNT 0.73 Elm River UP Keweenaw Peninsula 21 RBT 0.45 Escanaba River UP Escanaba 22 BKT, BNT 0.44 Iron River UP Brule 28 BKT 0.59 Little Indian River UP Manistique 29 BKT 0.73 Little Muskegon River NLP Muskegon 4 RBT 0.62 Manistee River NLP Manistee 5 BKT, BNT, RBT 0.65 19 Stream Region Sub - basin Map number Species BFI Martin Creek SLP Pere Marquette - White 16 BKT 0.61 Pere Marquette River NLP Pere Marquette - White 6 BNT 0.61 Pigeon River NLP Cheboygan 7 BNT, RBT 0.65 Pine River SLP Pine 12 BNT 0.65 Pine River NLP Manistee 8 BKT, BNT, RBT 0.49 Prairie Creek SLP Lower Grand 13 BNT 0.50 Rogue River SLP Lower Grand 17 BNT, RBT 0.50 Salmon Trout River UP Keweenaw Peninsula 23 BKT 0.45 Tahquamenon River UP Tahquamenon 24 BNT 0.55 West Branch Sturgeon River NLP Cheboygan 9 BKT, BNT, RBT 0.65 Yellow Dog River UP Dead - Kelsey 30 RBT 0.52 20 Baseflow index range s from zero to one with increasing groundwater input (Wahl and Wahl 1988); streams in our study ranged from 0.44 0.73 (Table 1 .1 ). All BFI calculations were made using a digital filter hydrograph separation tec hnique (Arnold and Allen 1999, Kelleher et al. 2012) whereby daily streamflow records were partitioned into groundwater and surface - runoff components . Moreover, all streams were important from a management standpoint as they supported recreational fisherie s for brook charr, brown trout, rainbow trout, or a combination of these coldwater species . Only streams that had necessary historical data for development of stream - specific temperature regression models (i.e., field - measured air and water temperatures) w ere selected. We developed a list of streams meeting these criteria (i.e., latitudinal gradient, hydrological gradient, recreational importance, historical data) using information from the MDNR 2015) and through person al communication with MDNR Fisheries Division personnel (Tracy Kolb, Todd Wills) and employees of the Michigan Council of Trout Unlimited (Trout Unlimited 2015). A minimum of five streams per region were selected for each species to ensure regional replication . One exception was the SLP, where brook charr and rainbow trout are not widely distributed and only four streams contained one or both of these species. In total, 16, 18, and 11 streams contained brook charr , brown trout, and rainbow trout, res pectively. Eleven streams supported more than one species, and four supported all species (Table 1 .1 ) . Stream - specific regression models Historical air and water temperatures for each stream were used to create stream - specific regression models. Daily air temperatures collected in summer months (i.e., June, July, August) from 1990 2010 were compiled using the United States Historical Climate Network online 21 interface ( http://cdiac.ornl.gov/ ) . Air temperatures were measured at the gauging station closest MDNR water temperature gauges recorded daily temperatures in summer months from 1990 2010. T he most upstream ga u ge on each stream was selected because temper atures are typical ly coolest and most optimal for salmonids in headwater reaches. As such, we focused on these areas because if their thermal habitat is degraded, downstream habitat and salmonid populations will also be impaired. Hydrologic Unit Codes (HUC s) subbasin (HUC8) and subwatershed (HUC 12) were identified using the National Hydr ography Dataset Plus Version 1 ( NHDPlusV1 ) and the Wa tershed Boundary Dataset (USEPA 2005). The North American Anthropogenic Barrier Dataset was used to l ocate and omit ga u ges directly below dams , which elevate temperatures compared to upst ream reaches (Lessard 2000). For streams without MDNR gauges, a substitute ga u ge on the nearest stream within the same subwatershed was used. Stream - specific regression models were developed by pairing mean summer air and water temperatures from recent years (i.e., 2002 2010) in Microsoft Excel (Table 1. 2). Air temperature coefficients represented indices of stream thermal sensitivity (i.e., relative susceptibility to temperature change; Kelleher et al. 2012) and were coupled with air temperature projections (see below) to predict future water temperatures. In addition, historical warming in each stream was evaluated by com paring air temperatures in 1976 det ermine the magnitude of stream warming. Air temperature projections Three coupled climate models (CCMs) were used to backcast mean June, July, and August air temperatures in 2006 and 2012 and forecast mean temperatures in the same months in 22 TABLE 1. 2 . Stream - specific temperature regression models with standard error (SE) and coefficient of determination ( R 2 ). F (i.e., F 1,7 ) and P values refer to the air temperature (A) parameter used to predict stream temperature (S). R. denotes River and Crk. denotes Creek. Stream name Regression SE F P R 2 Bark R. S = 0.32A + 12.98 0.05 49.41 < 0.01 0.86 Bear Crk. S = 0.23A + 11.61 0.03 46.74 < 0.01 0.85 Black R. S = 0.29A + 8.74 0.04 57.04 < 0.01 0.88 Boardman R. S = 0.14A + 11.99 0.02 72.85 < 0.01 0.90 Bryan Crk. S = 0.42A+7.69 0.08 28.86 < 0.01 0.78 Canada Crk. S = 0.60A + 6.91 0.08 61.72 < 0.01 0.88 Carp R. S = 0.28A + 12.06 0.04 41.63 < 0.01 0.84 Cedar Crk. S = 0.56A + 6.32 0.09 39.10 < 0.01 0.83 Cedar R. (SLP) S = 0.25A + 11.42 0.03 97.82 < 0.01 0.92 Chocolay R. S = 0.23A + 10.29 0.03 77.99 < 0.01 0.91 Davenport Crk. S = 0.13A + 8.97 0.01 99.55 < 0.01 0.92 Duke Crk. S = 0.48A + 4.45 0.08 37.27 < 0.01 0.82 E. Branch Fox R. S = 0.33A + 7.73 0.04 86.45 < 0.01 0.91 Elm R. S = 0.82A + 2.11 0.06 193.20 < 0.01 0.96 Escanaba R. S = 0.88A + 3.03 0.14 39.26 < 0.01 0.83 Iron R. S = 0.30A + 12.76 0.04 52.58 < 0.01 0.87 Little Indian R. S = 0.06A + 14.86 0.01 83.65 < 0.01 0.91 Little Muskegon R. S = 0.34A + 12.09 0.04 69.13 < 0.01 0.89 Manistee R. S = 0.13A + 10.67 0.02 57.70 < 0.01 0.88 Martin Crk. S = 0.38A + 9.61 0.05 62.93 < 0.01 0.89 Pere Marquette R. S = 0.18A + 12.50 0.02 51.89 < 0.01 0.86 Pigeon R. S = 0.11A + 11.93 0.01 60.49 < 0.01 0.88 Pine R. (NLP) S = 0.22A + 10.89 0.03 73.41 < 0.01 0.90 Pine R. (SLP) S = 0.40A + 9.23 0.04 119.15 < 0.01 0.94 Prairie Crk. S = 0.74A + 0.95 0.07 106.03 < 0.01 0.93 Rogue R. S = 0.23A + 13.17 0.04 27.46 < 0.01 0.77 Salmon Trout R. S = 0.29A + 8.23 0.05 33.60 < 0.01 0.80 Tahquamenon R. S = 0.50A + 12.29 0.05 115.88 < 0.01 0.93 W. Branch Sturgeon R. S = 0.06A + 11.93 0.01 67.49 < 0.01 0.89 Yellow Dog R. S = 0.38A + 9.90 0.06 45.58 < 0.01 0.85 23 2036 and 2056 for each subbasin: the Third Generation Coupled Global Climate Model (CGCM3, Canadian Centre for Climate Modelling and Analysis), the CM2 Global Coupled Climate Model (CM2, Geophysical Fluid Dynamics Laboratory at the National Oceanic and Atmospheric Administration), and the Hadley Centre Coupled Model version 3 (HadCM3, Met . All CCMs were based on the Wo rld Climate Research Programme (WCRP) Coupled Model Intercomparison Project phase 3 (CMIP3) multi - model dataset . Spatial downscaling was performed using the Bias - Correction Spatial Disaggregation (BCSD) approach to adjust the resolution of the climate model (~200km x 200km) to a scale germane for Michigan streams (12km x 12km; Maurer et al. 2007). T he Eastern Forest Environmental Threat Assessment Cente r (EFETAC) in North Carolina supplied mean June, July , and August air temperatures for Michigan subbasins containing coldwater salmonid streams. Projections were made based on the CGCM3, CM2, and HadCM3 models using area - weighted means for all years assuming the Special Report of Emission Scenarios (SRES) A2 and B1 climate forcing scenarios. The A2 scenario describes a future world with rapid economic growth and efficient energy technologies and predicts atmospheric CO 2 concentrations to be 820 ppm in 2100. In contrast, the B1 scenario projects a convergent world with a servic e and informatio n economy and reduced material consumption and predicts atmospheric CO 2 concentrations 550 ppm in 2100. Combining scenarios was in formative as they represent upper and lower emissions thresholds for stream tempe rature prediction. Stream temperature and thermal habita t suitability projections Stream temperatures were backcasted in 2006 and 2012 and forecasted i n 2036 and 2056 by inputting CCM air temperature pr edictions i nto stream - specific regression models. Mean July 24 stream temperatures were projected b ecause this month is typically the warmest and most thermally stressful for salmonids in Michigan (Zorn et al. 2011). To incorporate the range of air temperatures projected by each CCM and the intrinsic uncertainty and unique characteristics (i.e., atmosph eric pressure, sea ice rheology, forest canopy density, soil layering) of each model, predictions were averaged across the three S pecies - specific the rmal habitat suitability status was assigned for each stream based on conditions for growt h and surv ival associated with projected July temperatures . Status 1, 2, 3, and 4 corresponded with optimal growth, suboptimal (i.e., reduced) growth, no growth, and extirpation , respectively (Table 1. 3). United States Fish and Wildlife Service ( USFWS) Biological Report s (e.g., Raleigh 1982a,b; Raleigh et al. 1986 ) contained thermal habitat status temperature ranges for juveniles and adults of each species . Other sources (i.e., Fry et al. 1946; Baldwin 1957 ; Wurtsbaugh and Davis 1977 ; Elliott and Hurley 2000 ; Hay and Young 2006 ) contained temperature ranges for juveniles and/or adults and were used to confirm temperatures reported in the USFWS reports. When t hreshold tempera tures (e.g., thermal minima, maxima ) differed between juveniles and adults, we reporte d juvenile temperatures under the premise that resilient salmonid fisheries can only be conserved if young fish survive to adulthood. Analyses The accuracy of stream - specific regression models was evaluated by comparing each re and thermal habitat suitability stat us in 2006 to its actual (i.e., field - measured) temperature and associated habitat status obtained from the MDNR database. The association between BFI and air temperature regression coefficients, which are indices of stream thermal sensitivity , was assessed using simple linear regression ( Kelleher et al. 2012). 25 TABLE 1. 3 . Thermal habitat suitability status (Habitat status) designations and corresponding temperature ranges (Temperature) and grow th conditions (Growth) for juvenile and adult brook charr (BKC; Fry et al. 1946; Baldwin 1957; Raleigh 1982a ), brown trout (BNT; Raleigh et al. 1986; Elliott and Hurley 2000; Hay and Young 2006 ), and rainbow trout (RBT; Wurtsbaugh and Davis 1977; Raleigh 1 982b ). Species Habitat status Temperature Growth BKC 1 Optimal 2 Suboptimal 3 None 4 Extirpation BNT 1 Optimal 2 Suboptimal 3 26.2 None 4 Extirpation RBT 1 Optimal 2 Suboptimal 3 None 4 Extirpation 26 Results Model accuracy Stream - specific models accurately projected temperature and thermal habitat suitability status in brook charr, brown trout, and rainbow trout streams. Under the A2 scenario, the mean deviation between predicted and actual temperatures was - 0.46°C (SD = 0.5 6; Table 1.4). Under the B1 scenario, the mean deviation between predicted and actual temperatures was - 0.58°C (SD = 0.59; Table 1.4). Under the A2 and B1 scenarios, stream - specific models predicted thermal habitat status with 93.0% percent overall accurac y in streams with brook charr (94.0% accuracy, n = 15), brown trout (89.0% accuracy, n = 16), and rainbow trout (100.0% accuracy, n = 11; Table 1.4). Thermal habitat suitability: Brook charr Stream - specific regression models projected that climate - induc ed air temperature elevation will have substantial effects on stream temperature and thermal habitat suitability, with impacts varying by BFI (i.e., streams with lower BFI were more thermally sensitive; Figure 1.2), species, region, time period, and climat e forcing scenario. From 1976 to 2006, the mean temperature of brook charr streams increased by 0.74°C (Table 1.5). In the UP under the A2 and B1 scenarios, thermal habitat suitability was predicted to be optimal in the East Branch Fox and Little Indian ri vers and suboptimal in the Iron River from 2006 to 2056 (Figure 1.3a,b). From 2006 to 2012, thermal habitat was projected to be optimal in Bryan Creek and the Salmon Trout River and suboptimal in the Carp and Escanaba rivers under both scenarios. From 2012 to 2036, thermal habitat was projected to become suboptimal in Bryan Creek with predicted warming by 1.19°C (A2 scenario ) and 1.31°C ( B1 scenario ). Similarly, thermal habitat was forecasted to 27 TABLE 1. 4 . Actual versus projected stream temperatures and thermal habitat suitability (THS) status for brook charr (BKC), brown trout (BNT), and rainbow trout (RBT) as predicted by stream - specific air - water temperature regression models. Temperatures are predicted in 2006 under the A2 (820 ppm atmospheric CO 2 by 2100) and B1 (550 ppm atmospheric CO 2 by 2100) obtained from the Northeast Climate Science Center. The first and second THS numbers represent s tatuses associated with actual and projected temperatures, respectively. Stream name Actual A2 THS B1 THS Bark River 19.91 18.95 - 0.96 BNT: 2,2 18.88 - 1.03 BNT: 2,2 Bear Creek 15.75 16.54 0.79 BNT: 1,1 16.54 0.79 BNT: 1,1 Black River 15.05 14.46 - 0.59 BKC: 1,1 14.36 - 0.69 BKC: 1,1 Boardman River 15.02 14.88 - 0.14 BKC: 1,1 14.85 - 0.17 BKC: 1,1 BNT: 1,1 BNT: 1,1 Bryan Creek 16.66 15.27 - 1.39 BKC: 2,1 15.06 - 1.60 BKC: 2,1 Canada Creek 19.95 19.41 - 0.54 BKC: 2,2 19.09 - 0.86 BKC: 2,2 Carp River 17.61 17.05 - 0.56 BKC: 2,2 17.00 - 0.72 BKC: 2,2 BNT: 2,2 BNT: 2,2 RNT: 2,2 RNT: 2,2 Cedar Creek 18.30 18.27 - 0.04 BKC: 2,2 18.32 0.02 BKC: 2,2 BNT: 2,2 BNT: 2,2 Cedar River (SLP) 17.15 16.59 - 0.56 BNT: 2,1 16.46 - 0.69 BNT: 2,1 Chocolay River 15.23 14.77 - 0.46 RNT: 1,1 14.52 - 0.71 RNT: 1,1 Davenport Creek 11.50 11.39 - 0.11 RNT: 1,1 11.34 - 0.16 RNT: 1,1 Duke Creek 15.32 14.69 - 0.63 BKC: 1,1 14.73 - 0.59 BKC: 1,1 BNT: 1,1 BNT: 1,1 E. Branch Fox River 14.23 13.76 - 0.47 BKC: 1,1 13.62 - 0.61 BKC: 1,1 BNT: 1,1 BNT: 1,1 Elm River 17.48 16.96 - 0.52 RNT: 2,2 16.40 - 1.10 RNT: 2,2 Escanaba River 19.95 18.88 - 1.07 BKC: 2,2 18.45 - 1.50 BKC: 2,2 BNT: 2,2 BNT: 2,2 Iron River 18.61 18.53 - 0.08 BKC: 2,2 18.34 - 0.27 BKC: 2,2 Little Indian River 16.16 16.09 - 0.07 BKC: 1,1 16.03 - 0.13 BKC: 1,1 Little Muskegon River 19.89 19.24 - 0.65 RNT: 2,2 19.22 - 0.67 RNT: 2,2 Manistee River 13.55 13.29 - 0.26 BKC: 1,1 13.29 - 0.26 BKC: 1,1 BNT: 1,1 BNT: 1,1 RNT: 1,1 RNT: 1,1 Martin Creek 18.33 17.86 - 0.47 BKC: 2,2 17.87 - 0.46 BKC: 2,2 Pere Marquette River 16.62 16.41 - 0.21 BNT: 2,2 16.41 - 0.21 BNT: 2,2 Pigeon River 14.45 14.12 - 0.33 BNT: 1,1 14.10 - 0.35 BNT: 1,1 RNT: 1,1 RNT: 1,1 28 TABLE 1. 4 Stream name Actual A2 THS B1 THS Pine River (NLP) 15.97 15.26 - 0.71 BKC: 1,1 15.27 - 0.70 BKC: 1,1 BNT: 1,1 BNT: 1,1 RNT: 1,1 RNT: 1,1 Pine River (SLP) 18.42 17.55 - 0.87 BNT: 2,2 17.56 - 0.86 BNT: 2,2 Prairie Creek 18.35 16.57 - 1.78 BNT: 2,1 16.64 - 1.71 BNT: 2,1 Rogue River 18.36 17.95 - 0.41 BNT: 2,2 17.97 - 0.39 BNT: 2,2 RNT: 2,2 RNT: 2,2 Salmon Trout River 14.62 13.51 - 1.11 BKC: 1,1 13.31 - 1.31 BKC: 1,1 Tahquamenon River 20.02 21.16 1.14 BNT: 3,3 20.86 0.84 BNT: 3,3 W. Branch Sturgeon River 13.30 13.12 - 0.18 BKC: 1,1 13.11 - 0.19 BKC: 1,1 BNT: 1,1 BNT: 1,1 RNT: 1,1 RNT: 1,1 Yellow Dog River 18.00 17.33 - 0.67 RNT: 2,2 16.96 - 1.04 RNT: 2,2 29 FIGURE 1.2. Relative influence of baseflow index on thermal sensitivity (i.e., air temperature regression coefficients) of streams in the State of Michigan, USA. The regression equation is: Thermal sensitivity = - 1.78*Baseflow index + 1.38 ( P < 0.01, R 2 = 0.44). 30 TABLE 1. 5 . Historical warming (1976 2006) and backcasted stream temperature change (2006 - temperatures were backcasted and compared to known temperatures to measure the accuracy of stream - specific temperature models. For 2006 2012, predicted changes in A and S are provided for the A2 and B1 climate forcing scenarios (B1 in parentheses). R. deno tes River and Crk. denotes Creek. Stream name - 2006 - 2006 - 2012 - 2012 Bark R. + 1.72 + 0.55 + 1.09 (+ 1.23) + 0.35 (+ 0.39) Bear Crk. + 0.94 + 0.22 + 0.58 (+ 0.82) + 0.13 (+ 0.19) Black R. + 3.17 + 0.92 + 0.60 (+ 0.59) + 0.17 (+ 0.17) Boardman R. + 3.17 + 0.44 - 0.56 ( - 0.77) - 0.08 ( - 0.11) Bryan Crk. + 2.33 + 0.98 + 1.01 (+ 1.48) + 0.43 (+ 0.62) Canada Crk. + 3.17 + 1.90 - 0.79 ( - 0.59) - 0.47 ( - 0.36) Carp R. + 2.00 + 0.56 + 0.65 (+ 1.14) + 0.18 (+ 0.32) Cedar Crk. + 0.72 + 0.40 + 0.58 (+ 0.65) + 0.33 (+ 0.37) Cedar R. (SLP) + 2.61 + 0.65 - 0.68 ( - 0.23) - 0.17 ( - 0.06) Chocolay R. + 2.72 + 0.63 - 0.64 (+ 0.43) - 0.15 (+ 0.10) Davenport Crk. + 2.00 + 0.26 + 0.55 (+ 0.79) + 0.07 (+ 0.10) Duke Crk. + 0.72 + 0.35 + 0.65 (+ 0.74) + 0.31 (+ 0.36) E. Branch Fox R. + 2.00 + 0.66 + 0.56 (+ 0.87) + 0.19 (+ 0.29) Elm R. + 2.72 + 2.23 + 0.99 (+ 1.79) + 0.82 (+ 1.47) Escanaba R. + 2.72 + 2.40 + 1.01 (+ 1.48) + 0.89 (+ 1.30) Iron R. + 1.89 + 0.57 - 0.77 ( - 0.03) - 0.23 ( - 0.01) Little Indian R. + 2.72 + 0.16 - 0.61 (+ 0.21) - 0.04 (+ 0.01) Little Muskegon R. + 2.56 + 0.87 - 0.79 ( - 0.80) - 0.27 ( - 0.27) Manistee R. + 3.17 + 0.41 + 0.76 (+ 0.49) + 0.10 (+ 0.06) Martin Crk. + 2.11 + 0.80 - 0.82 ( - 0.83) - 0.31 ( - 0.31) Pere Marquette R. + 2.11 + 0.38 - 0.82 ( - 0.83) - 0.15 ( - 0.15) Pigeon R. + 3.17 + 0.35 + 0.67 (+ 0.50) + 0.07 (+ 0.06) Pine R. (NLP) + 2.11 + 0.46 + 0.76 (+ 0.49) + 0.17 (+ 0.11) 31 TABLE 1.5 Stream name - 2006 - 2006 - 2012 - 2012 Pine R. (SLP) + 2.56 + 1.02 + 0.68 (+ 0.72) + 0.27 (+ 0.29) Prairie Crk. + 1.94 + 1.44 + 0.58 (+ 0.65) + 0.43 (+ 0.48) Rogue R. + 0.72 + 0.17 + 0.65 (+ 0.74) + 0.15 (+ 0.17) Salmon Trout R. + 2.33 + 0.68 + 0.99 (+ 1.79) + 0.29 (+ 0.52) Tahquamenon R. + 2.00 + 1.00 + 0.75 (+ 1.32) + 0.38 (+ 0.66) W. Branch Sturgeon R. + 3.17 + 0.19 + 0.67 (+ 0.50) + 0.04 (+ 0.03) Yellow Dog R. + 2.33 + 0.89 - 0.63 (+ 0.39) - 0.24 (+ 0.15) 32 FIGURE 1.3. Projected temperatures in individual Michigan brook charr streams in 2006, 2012, 2036, and 2056 under A2 (820 ppm atmospheric CO 2 by 2100) and B1 (550 ppm by 2100) scenarios. Plots are organized by region and emissions scenario: (a) Upper Peninsula, A2 sc enario; (b) Upper Peninsula, B1 scenario; (c) northern Lower Peninsula, A2 scenario; (d) northern Lower Peninsula, B1 scenario; (e) southern Lower Peninsula, A2 scenario; (f) southern Lower Peninsula, B1 scenario. Dotted lines represent transitions between thermal habitat suitability statuses. Stream abbreviations are as follows: BRC = Bryan Creek, BLR = Black River, BOR = Boardman River, CAC = Canada Creek, CAR = Carp River, CEC = Cedar Creek, DUC = Duke Creek, EFR = East Branch Fox River, ESR = Escanaba River, IRR = Iron River, LIR = Little Indian River, MAR = Manistee River, MAC = Martin Creek, PIR = Pine River (northern Lower Peninsula), STR = Salmon Trout River, WSR = West Branch Sturgeon River. 33 become unsuitable for growth in the Escanaba River with p rojected warming by 2.49°C (A2 scenario) and 2.73°C (B1 scenario; Table 1. 6 ). Habitat suitability was predict ed to remain the same from 2036 to 2056 under both scenarios in all brook charr streams evaluated. In the NLP from 2006 to 2 056, thermal habitat was projected to be optimal in most streams (Black, Boardman, Manistee, Pine, West Branch Sturgeon rivers) under the A2 and B1 scenarios ( Figure 1. 3c,d). From 2012 to 2036, thermal habitat was forecasted to become unsuitable for growth in Canada Creek under the B1 scenario with predicted warming by 2.27 °C (Figure 1. 3d; Table 1. 6 ). In the SLP under the A2 and B1 scenarios , thermal habitat was predicted to be suboptimal from 2006 to 2056 in Martin Creek and Cedar Creek (Fig ure 1. 3 e,f ). From 2012 to 2036 u nder the B1 scenario , thermal habitat was forecasted to beco me suboptimal in Duke Creek with predicted warming by 1.51 °C ( Fig ure 1. 3f; Table 1. 6 ). Thermal habitat suitability: Brown trout From 1976 to 2006, the mean temperature of brown trout streams increased by 0.65°C as mean air temperature increased by 2.09°C ( Table 1. 5 ). In the UP under the A2 and B1 scenarios, thermal habitat suitability was predicted to be optimal in the East Branch Fox River from 2006 to 2056 (Figure 1. 4a,b ). From 2006 to 2012, thermal habitat suitability was projected to be optimal in the Bark River, suboptimal in the Escanaba River, and unsuitable in the Tahquamenon River under both scenarios (Figure 1. 4a,b). During the same time period, thermal habitat wa s forecasted to be suboptimal in the Carp River under the B1 scenario with predicted warming by 0.32°C (Table 1.5). From 2012 to 2036, thermal habitat was projected to be unsuitable for growth in the Escanaba River with predicted warming by 2.49 °C ( A2 scen ario) and 2.73 °C ( B1 scenario; Table 1.6). From 2036 to 2056, habitat suitability was predicted to remain the same 34 T ABLE 1. 6 . Projected future warming (2012 2036, 2036 air and stream temperatures over corresponding time periods. Predicted changes in A and S are provided for the A2 and B1 climate forc ing scenarios (B1 in parentheses). R. denotes River and Crk. denotes Creek. Stream name - 2036 - 2036 036 - 2056 - 2056 Bark R. + 2.74 (+ 3.16) + 0.88 (+ 1.01) - 0.69 ( - 0.65) - 0.22 ( - 0.21) Bear Crk. + 2.40 (+ 3.10) + 0.55 (+ 0.71) + 0.72 ( - 0.29) + 0.17 ( - 0.07) Black R. + 2.58 (+ 3.79) + 0.75 (+ 1.10) - 0.77 ( - 1.15) - 0.22 ( - 0.33) Boardman R. + 2.49 (+ 3.77) + 0.35 (+ 0.53) - 0.87 ( - 1.04) - 0.12 ( - 0.15) Bryan Crk. + 2.83 (+ 3.11) + 1.19 (+ 1.31) - 0.62 ( - 0.83) - 0.26 ( - 0.35) Canada Crk. + 2.58 (+ 3.79) + 1.55 (+ 2.27) - 0.77 ( - 1.15) - 0.46 ( - 0.69) Carp R. + 2.78 (+ 3.62) + 0.78 (+ 1.01) - 0.49 ( - 1.29) - 0.14 ( - 0.36) Cedar Crk. + 2.48 (+ 3.14) + 1.39 (+ 1.76) + 0.48 ( - 0.49) + 0.27 ( - 0.28) Cedar R. (SLP) + 2.74 (+ 3.16) + 0.69 (+ 0.79) - 0.69 ( - 0.65) - 0.17 ( - 0.16) Chocolay R. + 2.71 (+ 3.36) + 0.62 (+ 0.77) - 0.42 ( - 1.31) - 0.10 ( - 0.30) Davenport Crk. + 2.63 (+ 3.53) + 0.34 (+ 0.46) - 0.54 ( - 1.26) - 0.07 ( - 0.16) Duke Crk. + 2.48 (+ 3.14) + 1.19 (+ 1.51) + 0.48 ( - 0.49) + 0.23 ( - 0.24) E. Branch Fox R. + 2.61 (+ 3.44) + 0.86 (+ 1.14) - 0.50 ( - 1.27) - 0.17 ( - 0.42) Elm R. + 3.19 (+ 2.65) + 2.62 (+ 2.17) - 0.75 ( - 0.39) - 0.62 ( - 0.32) Escanaba R. + 2.83 (+ 3.11) + 2.49 (+ 2.73) - 0.62 ( - 0.83) - 0.54 ( - 0.73) Iron R. + 3.13 (+ 2.82) + 0.94 (+ 0.85) - 0.74 ( - 0.35) - 0.22 ( - 0.11) Little Indian R. + 2.61 (+ 3.44) + 0.16 (+ 0.21) - 0.50 ( - 1.27) - 0.03 ( - 0.08) Little Muskegon R. + 2.48 (+ 3.41) + 0.84 (+ 1.16) - 0.12 ( - 0.72) - 0.04 ( - 0.24) Manistee R. + 2.47 (+ 3.56) + 0.32 (+ 0.46) - 0.54 ( - 0.81) - 0.07 ( - 0.11) Martin Crk. + 2.48 (+ 3.32) + 0.94 (+ 1.26) + 0.05 ( - 0.55) + 0.02 ( - 0.21) Pere Marquette R. + 2.48 (+ 3.32) + 0.45 (+ 0.60) + 0.05 ( - 0.55) + 0.01 ( - 0.10) Pigeon R. + 2.52 (+ 3.78) + 0.28 (+ 0.42) - 0.83 ( - 1.11) - 0.09 ( - 0.12) Pine R. (NLP) + 2.47 (+ 3.56) + 0.54 (+ 0.78) - 0.54 ( - 0.81) - 0.12 ( - 0.18) 35 TABLE 1.6 Stream name - 2036 - 2036 - 2056 - 2056 Pine R. (SLP) + 2.49 (+ 3.25) + 1.00 (+ 1.30) + 0.23 ( - 0.69) + 0.09 ( - 0.27) Prairie Crk. + 2.48 (+ 3.14) + 1.83 (+ 2.32) + 0.48 ( - 0.49) + 0.35 ( - 0.37) Rogue R. + 2.48 (+ 3.14) + 0.57 (+ 0.72) + 0.48 ( - 0.49) + 0.11 ( - 0.11) Salmon Trout R. + 3.19 (+ 2.65) + 0.92 (+ 0.77) - 0.75 ( - 0.39) - 0.22 ( - 0.11) Tahquamenon R. + 2.72 (+ 3.51) + 1.36 (+ 1.75) - 0.41 ( - 1.36) - 0.21 ( - 0.68) W. Branch Sturgeon R. + 2.52 (+ 3.78) + 0.15 (+ 0.23) - 0.83 ( - 1.11) - 0.05 ( - 0.07) Yellow Dog R. + 2.98 (+ 2.96) + 1.13 (+ 1.13) - 0.63 ( - 0.81) - 0.24 ( - 0.31) 36 FIGURE 1.4. Projected temperatures in individual Michigan brown trout streams in 2006, 2012, 2036, and 2056 under A2 (820 ppm atmospheric CO 2 by 2100) and B1 (550 ppm by 2100) scenarios. Plots are organized by region and emissions scenario: (a) Upper Peninsula, A2 scenario; (b) Upper Peninsula, B1 scenario; (c) northern Lower Peninsula, A2 scenario; (d) northern Lower Peninsula, B1 scenario; (e) southern Lower Peninsula, A2 scenario; (f) southern Lower Penin sula, B1 scenario. Dotted lines represent transitions between thermal habitat suitability statuses. Stream abbreviations are as follows: BEC = Bear Creek, BOR = Boardman River, BRC = Bryan Creek, CAR = Carp River, CEC = Cedar Creek, CER = Cedar River, DUC = Duke Creek, EFR = East Branch Fox River, ESR = Escanaba River, MAR = Manistee River, PMR = Pere Marquette River, PGR = Pigeon River, PNR = Pine River (southern Lower Peninsula), PIR = Pine River (northern Lower Peninsula), PRC = Prairie Creek, ROR = Rog ue River, TAR = Tahquamenon River, WSR = West Branch Sturgeon River. 37 under both scenarios in all brown trout stream s evaluated . In the NLP from 2006 to 2056, thermal habitat was forecasted to be optimal under the A2 and B1 scenarios in all brown trout streams evaluated , including the Boardman, Manistee, Pere Marquette, Pigeon, Pine, and West Branch Sturgeon rivers (Figure 1. 4 c,d). In the SLP under the A2 and B1 scenarios , thermal habitat suitability was predicted to be optimal in Duke Creek and suboptimal in the Rogue River from 2006 to 2056 (Figure 1. 4e,f). From 2006 to 2012, thermal habitat was projected to be optimal in the Cedar River and suboptimal in Cedar Creek and the Pine Rive r under both scenarios. During the same time period , ther mal habitat was predicted to be suboptimal in Prairie Creek under the B1 scenario with projected warming by 0 .48°C ( Table 1. 5 ). From 2012 to 2036, thermal habitat was forecasted to be suboptimal in the Cedar River and Bear Creek and unsuitable for growth in Cedar Creek under both scenarios. From 2036 to 2056 under the A2 scenario, thermal habitat was proj ected to be optimal in the Cedar River and unsuitable for growth in Cedar Creek wi th predicted cooling and warming by 0.17°C and 0.27°C, respectively (Figure 1. 4e; Table 1. 6 ). Thermal habitat suitability was forecast ed to remain the same from 2036 to 2056 under the B1 scenario in all brown trout streams evaluated (Figure 1. 4f). Therma l habitat suitability: Rainbow trout From 1976 to 2006, the mean water temperature of brown trout streams increased by 0.64°C (air temperature +2.42°C; Table 1.5). In the UP under the A2 and B1 scenarios, thermal habitat suitability was predicted to be optimal in the Chocolay River and Davenport Creek and suboptimal in the Carp, Elm, and Yellow Dog Rivers from 2006 to 2056 (Figure 1.5a,b). Similarly, thermal habitat was projected to be optimal (Manistee, Pigeon, Pine, and West Branch Sturgeon rivers) and suboptimal (Little Muskegon River) in the NLP from 2006 to 2056 under 38 FIGURE 1.5. Projected temperatures in individual Michigan rainbow trout streams in 2006, 2012, 2036, and 2056 under A2 (820 ppm atmospheric CO 2 by 2100) and B1 (550 ppm by 2100) scenarios. Plots are organized by region and emissions scenario: (a) Upper Peninsula, A2 scenario; (b) Upper Peninsula, B1 scenario; (c) northern Lower Peninsula, A2 scenario; (d) northern Lower Peninsula, B1 scenario; (e) southern Lower Peninsula, A2 sce nario; (f) southern Lower Peninsula, B1 scenario. Dotted lines represent transitions between thermal habitat suitability statuses. Stream abbreviations are as follows: CAR = Carp River, CHR = Chocolay River, DAC = Davenport Creek, ELR = Elm River, LMR = L ittle Muskegon River, MAR = Manistee River, PGR = Pigeon River, PIR = Pine River, ROR = Rogue River, WSR = West Branch Sturgeon River, YDR = Yellow Dog River. 39 the A2 and B1 scenarios (Figure 1. 5c,d). T hermal habitat in the Rogue River, the only rainbow trout stream in the SLP , was predicted to be suboptimal under both emissions scenarios f rom 2006 to 2056 (Figure 1. 5 e,f). Discussion Our study is the first broad - scale investigation of climate - driven stream temperature warming in Michigan with implications for salmonid management. Water t emperatures were projected to increase by 0.19 5.94°C in 30 coldwater salmonid streams over the next 40 years due to predicted air temperature warming. This finding supports a previous study in the State Mi nnesota, USA ( Pilgrim et al. 1998 ), in which temperatures in 39 streams increased by 0.30 6.90°C under a doubling of atmospheric CO 2 concentrations comparable to the A2 climate forcing scenario. In addition, our study supports a previous investigation in t he State of Wisconsin, USA (Lyons et al. 2010), in which thermal habitat degradation was projected for brook charr and brown trout in 282 streams forecasted to warm by 0.80 4.00°C over the next 45 years. In our research , the magnitude of warming and therma l habitat impairment varied among streams and time periods, with the greatest warming and habitat degradation occurring from 1976 200 6 (2.72°C) and projected from 2012 2036 (2.49 2.62°C) in the Elm and Escanaba rivers, systems with the lowest BFI (< 0.45). In contrast, the magnitude of warming from 1976 200 6 (0.16 0.66 °C) and projected warming from 2012 2036 (0.16 0.86°C) was considerably lower the East Branch Fox and Little Indian rivers, systems with the highest BFI (0.73). This finding supported our hyp othesis that temperature and thermal habitat alterations would vary spatially and temporally due to system - specific patterns of warming. It also reinforces previous research indicating fish assemblages in Michigan streams with stable discharge from 40 groundwater inputs are more resilient to thermal warming than those in streams with less stable discharge (Zorn et al. 2012 ). Thermal habitat degradation was projected to occur least frequently in rainbow trout streams as this species has a wider temperatu re range ( 12.0 22.5°C ) for growth and survival compared to brook charr ( 11.0 20.5°C ) and brown trout ( 12.0 20.0°C ) (Wurtsbaugh and Davis 1977) . Warmer air temperatures are projected to increase stream temperatures both directly (i.e., advection) and indirectly through effects on precipitation, evaporation, and transpiration. Predicted effects of climate change include more variable (and overall lo wer) precipitation (Parry et al. 2007; Stoner et al. 2012), increased evaporation (Compagnucci et al. 2001), and decreased discharge. These impacts would increase stream temperature by reducing the volume of water exposed to solar radiation . However, in Mi chigan streams near the Great Lakes, precipitation will likely intensify in fall and winter due to predicted increases in snowfall (Norton and Bolsenga 1993; Primack 2000), increasing surface runoff and water volume and decreasing temperature compared to s ystems without this source of precipitation ( Kurylyk et al. 2013) . I ncreased canopy and soil surface evaporation is projected to decrease groundwater recharge and increase wate r temperature (Ge et al. 2013), although streams high BFI are expected to warm a t a slower rate than those with low BFI due to groundwater - driven thermal buffering ( Menberg et al. 2014). Transpiration of water from plants has been reported to contribute to stream warming by reducing streamflow and groundwater recharge ( Federer and Las h 1978; Bond et al. 2002 ). Effects of climate warming on transpiration and stream temperature will depend largely on changes in the species composition and leaf characteristics (e.g., stomatal conductance, water use - efficiency) of riparian plants and trees ( Compagnucci et al. 2001 ; Kirschbaum 2004). We encourage researchers to incorporate the thermal effects of transpiration, riparian shading, and 41 watershed land cover ( Wehrly et al. 1997; Wiley et al. 2010 ) into future stream temperature modeling to improve predictive accuracy . Projected increases in Michigan stream temperatures will result in thermal habitat conditions less conducive for salmonid growth and survival during the warmest period of the year (Raleigh 1982a,b; Raleigh et al. 1986 ), particularly in thermally sensitive streams with low BFI values. This will likely be manifested by declines in availability and changes in spatial arrangement of optimal thermal habitat, with potential effects on salmonid distribution and survival . For instance, if str eam warming restricts coldwater habitats to specific areas (e.g., headwater reaches), salmonid populations will become isolated, which could increase interspecific competition and decrease survival ( Tsuboi et al. 2013; Dugdale et al. 2015 ). P revious resear ch i suggested warmer water temperatures resulting from climate change will restrict the distribution and decrease the survival of brook charr , brown trout, and rainbow trout (Steen et al. 2010). Researchers in the State of Wisco nsin, USA, predicted that by 2060 the total length of streams suitable for brown trout would decline by 8 , 33, and 88 percent under limited (summer air temperatures increase 1.0°C and water 0.8°C) , moderate (air 3.0°C and water 2.4°C) , and major (air 5.0°C and water 4.0°C) climatic warming (Lyons et al. 2010). Results were more extreme for brook charr , with distributional declines of 4 4 and 94 percent under limited and moderate warming, and complete extirpation under major warming. Although our m odels did not project extirpation of any species in the Michigan streams evaluated, summer growth limitation may be coupled with decreased reproduction as lower oxygen levels resulting from stream temperature warming and/or groundwater withdrawal may reduc e egg survival or prevent spawning (Raleig h 1982a,b; Raleigh et al, 1986). Although annual salmonid growth may remain stable or increase due to longer growing season length and 42 increased prey availability, projected reductions in growth and survival during the warmest period of the year have important management implications. Management implications We encourage scientists, biologists, policy makers , and public stakeholders to collaboratively develop resilience - based management programs to conserve salmon id populations amidst global change. Our research represents a step in this direction as it projects consequences of stream thermal warming for salmonid growth and survival in Michigan. Our results indicate it is important for fisheries professionals to manage streams for thermal resilience by forming public - private partnerships to protect watershed land cover types that facilitate high ground water recharge (e.g., grasslands; Waco and Ta ylor 2010 ; Siitari et al. 2011), preserve riparian vegetation and associated shading (Blann et al . 2002), and maintain longitudinal connectivity to promote salmonid movement to cold headwater reaches (Drake and Taylor 1996; Hayes et al. 1998). We recommend these strategies play prominent roles in resilience - based management programs for coldwater streams and their important salmonid fisheries. As documented in our study, streams with low thermal sensitivity (i.e., high BFI) will likely maintain thermal cond itions (e.g., cool summer temperatures, seasonal flow stability ; Wiley et al. 1997; Baker et al. 2003) that are more conducive for salmonid growth and survival than systems with high thermal sensitivity in the next 40 years. Thus, we encourage managers to allocate resources (i.e., time, money, personnel) to prioritize protection of streams with low thermal sensitivity and inform public stakeholders about realistic expectations for stream fish communities amidst global change (e.g., salmonid decline, centrar chid expansion; Pease and Paukert 2014). Moreover, because streams with high BFI were less susceptible to temperature 43 change, we suggest managers use spatial BFI maps as tools for understanding stream thermal sensitivity across large geographic areas. More over, we encourage managers to increase the spatial and temporal coverage of air and stream temperature monitoring networks and thereby expand their utility for salmonid management. Fisheries professionals and public stakeholders can also collaboratively implement additional strategies for resilience - based salmonid management. They can promote thermally resilient salmonid populations by removing dams and installing fish ladders at roadside crossings and culverts to restore stream habitat connectivity, part icularly in cold headwater reaches, which function as thermal refugia during warm summer months. Fisheries professionals can also foster salmonid population resilience by implementing regulations (e.g., protected slot limits, reduced creel limits) that red uce harvest and increase survival during thermally stressfu l periods. In addition, we recommend fisheries professionals protect a diversity of salmonid size classes , genetic stocks, and prey species that tolerate a wide range of temperatures predicted from climate change models (Hansen et al. 2015). We encourage fisheries managers and policy makers to provide incentives (e.g., financial assistance, open space tax deduction, fast - track permitting; Knight 2009) for land developers and property owners to prote ct coldwater habitat and thermal buffering mechanisms on their lands. Moreover, because it is infeasible to protect salmonid populations and thermal habitat in all streams, it will be necessary for fisheries professionals to implement a triage approach wit h specific criteria for stream protection (e.g., species composition, habitat qua lity, recreational importance). In summary, resilience - based salmonid management programs will require effective collaboration among scientists, biologists, policy makers , and public stakeholders . Our research promotes resilience - based salmonid management by providing a methodology to project stream temperature and thermal habitat suitability. 44 Fisheries professionals can use this approach to protect coldwater habitats and drive rs of stream cooling and ultimately conserve resilient salmonid populations amidst global change. Acknowledgments We thank Jennifer Moore Myers ( USFS EFETAC ) for assistance with air temperature projection models and data acquisition. We thank Kyle Herreman and Wesley Daniel ( Michigan State University ), Tracy Kolb and Todd Wills ( Michigan Department of Natural Resources ), and Henry Quinlan ( United S tates Fish and Wildlife Service) for providing initial environmental and trout population data critical for development of this study. We acknowledge the Program for Climate Model Diagnosis and Intercomparison (PCMDI) and the WCRP' s Working Group on Coupled Modeling for assisting in procurement of the WCRP CMIP3 multi - model dataset. 45 LITERATURE CITED 46 LITERATURE CITED Arblaster, J., Dufresne, J.L., Fichefet, T., Fiedlingstein, P., Gao, X., Gutowski, Jr., W.J., Johns, T., Krinner, G., Shongwe, M., Tebaldi, C., Weaver. A.J. & Wehner, M. 2014. Long - term climate change: projections, commitments and irreversibility. In: Stocker, T.F., Qin, D., Plattner, G.K., Tignor, M., Allen, S.K., Boschung, J., Nauels, A., Xia, Y., Bex, V. & Midgley, P.M., eds. Climate change 2013: the physica l science basis. Contribution of working group I to the fifth assessment report of the Intergovernmental Panel on Climate Change . pp. 1029 1136. Arnold, J.G. & Allen, P.M., 1999. Validation of automated methods for estimating base flow and ground water r echarge from streamflow records. Journal of the American Water Resources Association 35: 411 424. Baker, M.E., Wiley, M.J., Carlson, M.L. & Seelbach, P.W. 2003. A GIS model of subsurface water potential for aquatic resource inventory, assessment, and envi ronmental management. Environmental Management 32: 706 719. Baldwin, N.S. 1957. Food consumption and growth of brook trout at different temperatures. Transactions of the American Fisheries Society 86: 323 328. Blann, K., Nerbonne, J.F. & Vondracek, B. 2002. Relationship of riparian buffer type to water temperature in the driftless area ecoregion of Minnesota. North American Journal of Fisheries Management 22: 441 451. Bond, B.J., Jones, J.A., Moore, G., Phillips, N., Post, D. & McDonnell, J.J. 2002. The zone of vegetation influence on baseflow revealed by diet patterns of streamflow and vegetation water use in a headwater basin. Hydrological Processes 16 : 1671 1677 . Compagnucci, R., da Cunha, L., Hanaki, K., Howe, C., Mailu, G., Shiklomanov, I. & Stakhiv, E. 2001. Hydrology and water resources. In: McCarthy, J.J., Canziani O.F., Leary N.A., Dokken D.J. & White K.S., eds. Climate change 2001: impacts, adaptation, and v ulnerability. International Panel on Climate Change Third Assessment Report. pp. 191 233. Dodds, W.K. & Whiles, M.R. 2010. Freshwater ecology, second edition: concepts and environmental applications of limnology. Burlington: Academic Press. xv + 829 pp. Drake, M.T. & Taylor, W.W. 1996. Influence of spring and summer water temperature on brook charr, Salvelinus fontinalis , growth and age structure in the Ford River, Michigan . Environmental Biology of Fishes 45 : 41 51. 47 Dugdale, S.J., Bergeron, N.E., & St - Hilaire, A. 2015. Spatial distribution of thermal refuges analysed in relation to riverscape hydromorphology using airborne thermal infrared imagery. Remote Sensing of Environment 160 : 43 55. Elliott, J.M. & Hurley, M.A. 2000. Daily energy intake and growth of piscivorous brown trout, Salmo trutta . Freshwater Biology 44: 237 245. Federer, C.A. & Lash, D . 1978. Simulated streamflow response to possible differences in transpiration among species of hardwood tr ees. Water Resources Research 14: 1089 1097 . Ficklin D . L ., Stewart, I . T . & Maurer, E . P. 2013. Effects of climate change on stream temperature, dissolved oxygen, and sediment concentration in the Sierra Nevada in California. Water Resources Research 49: 276 5 2782. Fry, F.E.J., Hart, J.S. & Walker, K.F. 1946. Lethal temperature relations for a sample of young speckled trout, Salvelinus fontinalis Vol. 54 The University of Toronto Press, Toronto. Ge, Z.M., Kellomäki, S., Zhou, X., Wang, K.Y., Peltola, H., Väisänen, H. & Strandman, H. 2013. Effects of climate change on evapotranspiration and soil water availability in Norway spruce forests in southern Finland: an ecosystem model based approach. Ecoh ydrology 6: 51 63. Godby, N.A., Jr., Rutherford, E.S. & Mason, D.M. 2007. Diet, feeding rate, growth, mortality, and production of juvenile steelhead in a Lake Michigan tributary. North American Journal of Fisheries Management 27: 578 592. Hansen, G.J.A., Gaeta, J.W., Hansen, J.F. & Carpenter, S.R. 2015. Learning to manage and managing to learn: Sustaining freshwater recreational fisheries in a changing environment. Fisheries 40: 56 64. Hay, J., Hayes, J. & Young, R. 2006. Water quality gui delines to protect trout fishery values. Prepared for Horizons Regional Council. Cawthorn Report No. 1025. 17 pp. Hayes, D.B., Dodd, H. & Lessard, J. 2006. Effects of small dams on cold water stream fish communities. In: Nelson, J., Dodson, J.J., Friedl and, K., Hamon, T.R., Musick, J. & Verspoor, E., eds. Reconciling f isheries with c onservation: p roceedings of the f ourth w orld f isheries c ongress . Bethesda: American Fisheries Society Symposium 43, pp. 587 602. Hayes, D.B., Taylor, W.W., Drake, M., Marod, S. & Whelan, G. 1998. The value of headwaters to brook trout ( Salvelinus fontinalis ) in the Ford River, Michigan, USA. In: Haigh, M. J., Krecek, J., Rajwar , G.S & Kilmartin , M.P. (eds). Headwaters: w ater resources and soil conservation. New Delhi : Oxford an d IBH Publishing Co., 495 pp. 48 Holling, C. S. 1973. Resilience and stability of ecological systems. Annual Review of Ecology and Systematics 4: 1 23. Isaak, D.J., Muhlfeld, C.C., Todd, A.S., Al - Chokhachy, R., Roberts, J., Kershner, J.L., Fausch , K.D. & Hostetler, S.W. 2012. The past as prelude to the future for understanding 21st - century climate effects on Rocky Mountain trout. Fisheries 37: 542 556. Isaak, D.J., Young, M.K., Nagel, D.E., Horan, D.L., Groce, M.C. 2015. The cold - water climate shield: delineating refugia for preserving salmonid fishes through the 21st century. Global Change Biology 21 : 2540 2553. Kelleher, C ., Wagener, T., Gooseff, M., McGlynn, B., McGuire, K. & Marshall, L. Investigating controls on the the rmal sensitivity of Pennsylvania streams. Hydrological Processes 26: 771 785. Kirschbaum, M.U.F. 2004. Direct and indirect climate change effects on photosynthesis and transpiration. Plant Biology 6 : 242 253. Knight, K. 2009. Land use planning for salmon, steelhead and trout. Washington Department of Fish and Wildlife. Olympia, Washington. Available at: http://wdfw.wa.gov/publications/ 00033/wdfw00033.pdf (accessed 1 October 2015). Krider, L.A., Magner, J.A., Perry, J., Vondracek, B. & Ferrington, L.C., Jr. 2013. Air - water - sandstone landscape. Journal of the American Water Resources Association 49: 896 907. Kurylyk, B.L., Bourque, C.P.A. & MacQuarrie, K.T.B. 2013. Potential surface temperature and shallow groundwater temperature response to climate change: an example from a small forested catchment in east - central New Brunswick (Canada). Hydrology and Earth System Sciences 17: 2701 2716. L essard, J.L. 2000. Temperature effect of dams on coldwater fish and macroinvertebrate communities in Michigan. Master of Science, Michigan State University. Lyons, J., Stewart, J.S. & Mitro , M. 2010. Predicted effects of climate warming on the distribution of 50 stream fishes in Wisconsin, U.S.A. Journal of Fish Biology 77: 1867 1898. Magnuson, J.J., Crowder, L.B. & Medvick, P.A. 1979. Temperature as an ecological resource. American Zoologi st 19: 331 343 . Magnuson, J.J., Webster, K.E., Assel, R.A., Bowser, C.J., Dillon, P.J., Eaton, J.G., Evans, H.E., Fee, E.J., Hall, R.I., Mortsch, L.R., Schindler, D.W. & Quinn, F.H. 1997. Potential effects of climate change on aquatic systems: Laurentian Great Lakes and Precambrian Shield Region. Hydrological Processes 11: 825 871 . 49 Maurer, E.P., Brekke, L., Pruitt, T. & Duffy, P.B. 2007. Fine - resolution climate projections enhance regional climate change impact studies. Eos Transactions, American Geophysi cal Union 88 : 504. MDNR (Michigan Department of Natural Resources). 2015. Fishing Michigan.gov/fishing. Michigan Department of Natural Resources. Available at: http://www.michigan.gov/ dnr/0,4570,7 - 153 - 10364 --- ,00.html (accessed 19 June 2015). Menberg, K., Blum, P., Kurylyk, B.L. & Bayer, P. 2014. Observed groundwater temperature response to recent climate change. Hydrology and Earth System Sciences 18: 4453 4466. Neff, B.D., Day, S.M., Piggott, A.R. & Fuller, L.M. 2005. Base flow in the Great Lakes basin: U.S. Geological Survey Scientific Investigations Report 2005 5217, 23 p. Norton, D.C. & Bolsenga, S.J. 1993. Spatiotemporal trends in lake effect and continental snowfall in the Laurentian Great Lakes, 1951 1980. Journal of Climate 6 : 1943 1956. Ozaki, N., Fukushima, T., Harasawa, H., Kojiri, T., Kawashima, K. & Ono, M. 2003. Statistical analyses on the effects of air temperature fluctuations on river water qualities. Hydrological Processes 17: 2837 2853. Parry, M., Canziani, O., Palutikof , J., van der Linden, P. & Hanson, C. 2007. Climate change 2007: impacts, adaptation and vulnerability. International Panel on Climate Change Fourth Assessment Report. Pease, A.A. & Paukert, C.P. 2014. Potential impacts of climate change on growth and prey consumption of stream - dwelling smallmouth bass in the central United States. Ecology of Freshwater Fish 23 : 336 346. Pilgrim, J.M., Fang, X. & Stefan, H.G. 1998. Stream temperature correlations with air temperatures in Minnesota: i mplications for climate warming. Journal of the American Water Resources Association 34: 1109 1121. Primack, A.G.B. 2000. Simulation of climate - change effects on riparian vegetation in the Pere Marquette River, Michigan. Wetlands 20 : 538 547. Raleigh, R. F. 1982a. Habitat suitability index models: brook trout: U.S. Fish and Wildlife Service Biological Report 82 (10.24). 42 pp. Raleigh, R. F. 1982b. Habitat suitability index models: rainbow trout: U.S. Fish and Wildlife Service Biological Report 82 ( 10.60). 76 pp. Raleigh, R.F., Zuckerman, L.D. & Nelson, P.C. 1986. Habi tat suitability index models and instream flow suitability curves: brown trout : U.S. Fish and Wildlife Service Biological Report 82(10.124). 65 pp. Rutledge, A.T. 1998. Computer programs for describing the recession of ground - water discharge and for estimating mean ground - water recharge and discharge form streamfl ow data update: U.S. Geological Survey Water - Resources Investigations Report 98 4148, 43 p. 50 Sear, D.A., Armitage, P.D. & Dawson, F.H. 1999. Groundwater dominated rivers. Hydrological Processes 13: 255 276. Siitari, K.J., Taylor, W.W., Nelson, S.A.C. & Weaver, K.E. 2011. The influence of land cover composition and groundwater on thermal habitat availability for brook charr ( Salvelinus fontinalis ) populations in the United States of America. Ecology of Freshwater Fish 20: 431 437. Snyder, C.D., Hitt., N. P. & Young J.A. 2015. Accounting for groundwater in stream fish thermal habitat responses to climate change . Ecological Applications 5:1397 1419. Steen, P.J., Wiley, M.J. & Schaeffer, J.S. 2010. Predicting future changes in Muskegon River watershed game fish distributions under future land cover alteration and climate change scenarios. Transactions of the American Fisheries Society 139: 396 412. Stoner, A.M.K., Hayhoe, K., Yang, X.H . & Wuebbles, D.J. 2013. An asynchronous regional regression model for statistical downscaling of daily climate variables. International Journal of Climatology 33 : 2473 2494. Trout Unlimited. 2015. Michigan TU roster. Available at: http://www.michigantu.org/images /pdffiles/documents/MITU%20Council%20Roster092315pfm.pdf ( accessed 19 June 2015). Tsuboi, J., Iwata, T., Morita, K., Endou, S., Oohama, H. & Kouichi, K. 2013. Strategies for the conservation and management of isolated salmonid populations: lessons from Japanese streams. Freshwater Biology 58 : 908 917. Tyler, J.A. & Rutherford, E.S. 2007. River restoration effects on steelhead populations in the Manistee River, Michigan: Analysis using an individual - based model. Transactions of the American Fisheries Society 136: 1654 1673. USEPA (United States Env ironm ental Protection Agency). 2005. National Hydrography Dataset Plus - NHDPlus. Waco, K.E. & Taylor, W.W. 2010. The influence of groundwater withdrawal and land use changes on brook charr ( Salvelinus fontinalis ) thermal habitat in two coldwater tributaries in Michigan, USA. Hydrobiologia 650: 101 116. Wahl, K.L. & Wahl, T.L. 1988. E ffects of regional ground water declines on streamflows in the Oklahoma Panhandle. In: Waterstone, M. & Burt R.J., eds. Proceedings of the s ympos ium on water - use data for water r esources management. Tucson: American Water Resources Association, pp. 239 - 249. Wehrly, K.E., Wiley, M.J. & Seelbach, P.W. 1997. Landscape - based models that predict July thermal characteristics of lower Michigan rivers. Michigan Department of Natural Reso urces, Fisheries Research Report 2037, Ann Arbor. 51 Wehrly, K.E., L. Wang & M. Mitro. 2007. Field - based estimates of thermal tolerance limits for trout: Incorporating exposure time and temperature fluctuation. Transactions of the American Fisheries Society 136 : 365 374 . Wiley, M.J., Kohler, S.L. & Seelbach, P.W. 1997. Reconciling landscape and local views of aquatic communities: Lessons from Michigan trout streams. Freshwater Biology 37: 133 148. Wiley, M.J., Hyndman, D.W., Pijanowski, B.C., Kendall, A.D., Riseng, C., Rutherford, E.S., Cheng, S.T., Carlson, M.L., Tyler, J.A., Stevenson, R.J., Steen, P.J., Richards, P.L., Seelbach, P.W., Koches, J.M. & Rediske, R.R. 2010. A multi - modeling approach t o evaluating climate and land use change impacts in a Great Lakes River Basin. Hydrobiologia 657 : 243 262. Wurtsbaugh, W.A. & Davis , G.E . 1977. Effects of fish size and ration level on the growth and food conversion efficiency of rainbow trout, Salmo gairdneri Richardson. Journal of Fish Biology 11: 99 104. Zorn, T.G., Seelbach, P.W. & Rutherford, E.S. 2012. A regional - scale habitat suitability model to assess the effects of flow reduction on fish assemblages in Michigan streams. Journal of the America n Water Resources Association 48: 871 895. Zorn, T. G., Seelbach, P.W. & Wiley , M.J . 2011. Developing u ser - f riendly h abitat s uitability t ools from r egional s tream f ish s urvey d ata. North American Jour nal of Fisheries Management 31: 41 55. 52 CHAPTER 2: COMPARING STREAM - SPECIFIC TO GENERALIZED TEMPERATURE MODELS TO GUIDE COLDWATER SALMONID MANAGEMENT IN A CHANGING CLIMATE Carlson, A. K., W . W. Taylor, K. M. Hartikainen , D. M. Infante , T. D. Beard, Jr., A. J. Lynch . 2017. Comparing stream - specific to generalized temperature models to inform salmonid management in a changing climate . Reviews in Fish Biology and Fisheries 27:443 462 . The content of this chapter is intended to be identical to the published manuscript cited above and reflects specifica tions (e.g., formatting ) for Reviews in Fish Biology and Fisheries . Any differences should be minor and are unintended. 53 Abstract Global climate change is predicted to increase air and stream temperatures and alter thermal habitat suitability for growth and survival of coldwater fishes, including brook charr ( Salvelinus fontinalis ) , brown trout ( Salmo trutta ) , and rainbow trout ( Oncorhynchus mykiss ). In a changing climate, a ccurate stream temperature modeling is increasingl y important for sustainable salmonid management throughout the world. However, finite resource availability (e.g. funding , personnel) drives a tradeoff between thermal model accuracy and efficiency (i.e. cost - effective applicability at management - relevant spatial extents). Using different projected climate change scenarios, we compared the accuracy and efficiency of s tream - specific and generalized (i.e. region - specific ) temperature models for coldwater salmonids within and outside the State of Michigan , USA, a region with long - term stream temperature data and productive coldwater fisheries. Projected s tream temperature warming between 2016 and 2056 ranged from 0.1 to 3.8 ° C in groundwater - dominated streams and 0.2 to 6.8 ° C in surface - runoff dominated systems in the State of Michigan. Despite their generally lower accuracy in predicting exact stream temperatures, generalized models accurately projected salmonid thermal habitat suitability in 82 % of groundwater - dominated streams, includ ing those with brook charr ( 80 % accuracy), brown trout (89 % accuracy ), and rainbow trout ( 75 % accuracy). In contrast, generalized models predicted thermal habitat suitability in runoff - dominated streams with much lower accuracy (54 %). These results suggest that, a midst climate change and constraints in resource availabili ty, generalized models are appropriate to forecast thermal conditions in groundwater - dominated streams within and outside Mich igan and inform regional - level salmonid management strategies t hat are practical for coldwater fisheries managers, policy makers, and the public. We recommend fisheries professionals reserve resource - intensive 54 stream - specific models for runoff - dominated sy stems containing high - pri ority fisheries resources (e.g. trophy individuals, endangered species) that will be directly impacted by projected stream warming. KEYWORDS: brook charr, brown trout, climate change, coldwater fisheries, rainbow trout, temperatu re 55 Introduction S treams and rivers cover 0.30 0.56% of globe across a fluvial area of 485,000 682,000 km 2 ( Downing et al. 2012). Climate change is projected to increase stream and river temperatures and alter their thermal habitat suitability for growth, reproduction, and survival of resident fishes (Pilgrim et al. 1998; Lyons et al. 2010; Carlson et al. 2017 ) . In addition, warmer air temperatures are predicted to increase stream and river temperatures indirectly through more variable (and overall lower) precipitation (Stoner et al. 2013), increased evaporation (Compagnucci et al. 2001), and thus decreased groundwater discharge into these river systems. Salmonids such as brook char r ( Salvelinus fontinalis ), brown trout ( Salmo trutta ), and rainbow trout ( Oncorhynchus mykiss ) are coldwater fishes with laboratory temperature p referenda less than or equal to 20 ° C and critical therm al maxima less than or equal to 31 ° C (Raleigh 1982a, b ; Raleigh et al. 1986; Lyons et al. 2009). These species support ecologically, socioeconomically, and culturally important fisheries in North America, South America, Europe, and parts of Asia and Australia (MacCrimmon 1972; Budy et al. 2013 ; Weber et al. 2015 ). As such, understanding the effects of climate change on these valuable species will have economic and social ramifications for the management of coldwater stream fisheries throughout the world. Temperature regulates fish metabolism, growth, reproduc tion, and survival ( Dodds and Whiles 2010 ; I saak et al. 2012), all of which affect fish recruitment and fisheries productivity. Temperatures above species - specific thermal maxima cause mortality, whereas temperatures at or below maxima alter individual growth and reproduction (Magnuson et al. 1997 ) . Predicted climate - driven i ncrease s in stream temperature s are likely to decrease the thermal habitat suitability of coldwater streams for growth and survival of brook charr , brown trout, and rainbow trout because these species have relatively low thermal tolerances to warm temperatures ( Fry et 56 al. 1946 ; Raleigh 1982a, b; Raleigh et al. 1986). Despite the importance of climate - change - induced temperature increases of riverine ecosystems through out the world (Kaushal et al. 2010; van Vliet et al. 2013), there is widespread scarcity in the availability of long - term stream temperature datasets throughout the world. This data deficiency impedes or prevents development of region - and stream - specific temperature models and inhibits the prediction and mitigation of the ecological and social effects of stream warming on salmonid fisheries (e.g. fragmented species distribution, reduced socioeconomic output). Collectively, the global distribution and impor tance and stream salmonids, the global scope of climate change, and the general dearth of long - term stream temperature data indicate the importance of developing accurate , efficient approaches for projecting effects of climate change on coldwater streams and designing data - limited strategies for sustainable salmonid management in a warming world. Fisheries management is broadly defined as the pro cess of using information (e.g. ecolog ical, economic, social, political ) to develop strategies for achieving goals established for fisheries resources (Kruger and Decker 1999). Decision - making is fundamental to fisheries management and nec essitates informed choices regarding allocation of fini te resources (e. g. funding, time, personnel) to achieve the goals set for specific fisheries systems . For example, to mitigate the effects of climate change on coldwater stream ecosystems, fisheries professionals must understand how stream temperature is r egulated by ambient atmospheric conditions (e.g. air temperature) and also influenced by meteorological (e.g. solar radiation, wind, h umidity) and hydrological (e.g. discharge, depth, groundwater input) conditions in stream watershed s (Gu et al. 1998; Pilg rim et al. 1998). With this knowledge, fisheries professionals can make informed temperature modeling decisions, including which models and variables to use, amidst constraints 57 in resource availability and, ultimately, which thermal habitat management stra tegies to implement for mitigation purposes. Heat budget models generally predict water temperatures with high accuracy (i.e. exactness of temperature projection) because they prioritize the relative influence of the various atmospheric, meteorological, a DeWalle 2006; Wehrly et al. 2009 ). However, these models generally require extensive and expensive data collection protocols for small spatial extents (e.g. stream reaches), limiting their appli cability at the regional scales (e.g. watersheds) where fisheries management agencies typically operate ( Anonymous 2000; WDNR 2002; MNDNR 2011) . As a result, alternative approaches such as air - stream temperature regression models are often less complex, mo re cost - effective (i.e. same accuracy at lower cost), and more spatially appropriate than heat budget models (Mohseni et al. 1998; Benyahya et al. 2007), making them more useful for fisheries management agencies when managing for most fish species of inter est. In lieu of heat budget models, fisheries professionals currently use two broad approaches to describe the relationship between air tem perature and stream temperature : stream - specific models and generalized models. Stream - specific temperature models account for the unique discharge, groundwater input), whereas generalized models are region - specific in representing the thermal regimes of all streams in a p articula r area. S tream - specific models treat each stream as a distinct system with a thermal regime influenced by discrete factors, whereas generalized models assume that regio nal patterns in air temperature, as opposed to system - specific characteristi cs, are the primary drivers of water temperature ( 1993 ; Krider et al. 2013) . Each modeling approach has advantages and disadvantages . G eneralized 58 models may be less accurate than stream - specific models in predicting exact water temperatu res, but their development requires significantly lower investments of funding, time, and personnel as streams do not have to be monitored individually. In contrast, stream - specific models are generally more accurate than generalized models but require sig nificantly more resources to develop. In addition, fisheries management strategies (e.g. harvest regulations) implemented based on stream - specific models may be cumbersome to implement because they will likely be unique to and thus variable among indiv idual streams reaches. Many U.S. states (e.g. Michigan, Wisconsin, Minnesota; Anonymous 2000; WDNR 2002; MNDNR 2011) already manage streams on a regional basis, aligning more closely with the generalized , rather than stream - specific, temperature modeling framework. However, the tradeoffs associated with st ream - specific and generalized models are impor tant to consider in evaluating the accuracy and efficiency (i.e. cost - effective applicability at management - relevant spatial extents) of each modeling approac h to achieve specific management objectives for sustaining stream salmonid fisheries in a changing climate. Comparing stream - specific and generalized temperature models necessitates the availability of long - term stream temperature data. Although long time - series stream temperature data sets are relatively uncommon, fisheries professionals in the State of Michigan, USA, have monitored temperatures in many trout streams for 10 20 years , making it an ideal study area in which to compare s tream - specific and generalized models. Brook charr , brown trout, and rainbow trout are ecologically and socio - economically important fishes in Michigan ( Godby et al. 2007 ), serving as keystone predators in coldwater streams and suppor ting valuable recreational fisheries in which more than 585,000 anglers spent 8.2 million angling days in 2011 ( USFWS 201 1 ). In Michigan, projected air temperature warming is predicted to inc rease 59 summer temperatures in coldwater streams by 0.19 5.49° C from 2016 to 2056 ( Carlson et al. 2017 ) and decrease thermal habitat availability and salmonid growth in systems that exceed thermal optima during the warmest period of the year (i.e. July; Zorn et al. 2011). Although Carlson et al. (2017 ) used stream - spec ific models to project future water temperatures in Michigan streams, the present study expands upon previous work by explicitly comparing stream - specific and generalized models to inform stream salmonid management amidst climate change and limitations in the availability of funding, time, and personnel. The goal of this study was to evaluate the impact of warming water temperatures on salmonid growth and survival in coldwater streams in select areas of North America. Two modeling approaches were used to p redict thermal habitat changes: stream - specific and generalized models . These models were chosen as they are spatially and temporally robust and thus can inform salmonid management approaches amidst a changing climate and resource limitations . As such, thi s study was intended to lay a conceptual foundation for future studies in other areas of the world with coldwater streams and fish populations susceptible to climate change. The specific objectives of this study were to: (1) measure the accuracy of stream - specific and generalized models in predicting water temperature and thermal habitat suitability for salmonid growth and survival in streams in the State of Michigan and the eastern USA that span latitudinal and hydrological gradients and support socioecono mically valuable salmonid populations ; (2) forecast future water temperatures and thermal habitat suitability in these streams a in select future years (i.e. 2036, 2056); and, (3) evaluate the accuracy and efficiency of stream - specific and generalized models to develop a model comparison approach that can be used for salmonid management programs within and outside Michigan. 60 Methods Study area Fifty - two coldwater salmonid streams were selected throughout Michigan based on latitudinal, hydrological, and recreational criteria (Figure 2. 1, Table 2. 1). Streams spanned a latitudinal thermal gradient from north to south. Base flow, the component of streamflow attributable to groundwater, varied among streams such that they covered a gradient from surface - runoff to groundwat brook charr , brown trout, and rainbow trout occur in Michigan. In addition, all streams were important from a fisheries management perspective as they support productive recreational fisheries for brook charr , brown trout, or rainbow trout. These species are widely distributed throughout Michigan (Zorn et al. 2011, 2012) , making them effective indicator species for evaluating the impact of climate change on coldwater stream fishes adapted to groundwater - dominated and surface runoff - dominated streams. Overall, brook charr were found in 28 of the streams evaluated, brown trout in 26 streams, and rainbow trout in 21 streams. Seventeen of the streams studied s upported more than one salmonid species, and six streams supported all three species (Table 2. 1). Baseflow measurements A United States Geolog ical Survey report of base flow (Neff et al. 2005 ) was used to obtain base flow index (BFI), a value that represents the mean rate of base flow (mm*yr - 1 ) divided by the corresponding mean rate of total streamflow (mm*yr - 1 ) and ranges from zero (i.e. no groundwater) to one (i.e. all groundwater; Wahl and Wahl 1988). All BFI c alculations were made using a digital filter hydrograph separation tec hnique (Arnold and 61 FIGURE 2 .1. Map of 52 brook charr, brown trout, and rainbow trout streams used for air stream temperature modeling in Michigan. Streams and corresponding identifica tion numbers are listed in Table 1. Mo dified from Carlson et al. (2017 ). 62 TABLE 2.1 . Descriptive information about 52 streams used for temperature modeling in Michigan, USA. Sub - basin denotes the National Hydrography Dataset sub - basin (i.e. 8 - digit Hydrologic Unit Code). Map number refers to stream identifiers in Figure 2. 1. Type d enotes hydrological influence: groundwater - dominated (GD) streams were distinguished from surface - runoff dominated (RD) systems using a threshold base - flow index (i.e. > 0.60; Neff et al. 2005) . Species: brook charr [BKC], brown trout [BNT], rainbow trout [RBT]). Area: drainage area, km 2 . Elevation: mean catchment elevation, m . Asterisks indicate that streams had historical field - collected air and stream temperatures for development of stream - specific regression models. R. denotes river and Crk. denotes creek. NLP, S LP , and UP refer to M Peninsula, Sou thern Lower Peninsula, and Upper Peninsula, respectively. Stream Map Type Species BFI Area Elevation Au Sable R. 1 GD BKC, BNT, RBT 0.72 5003.86 401.68 Bark R.* 2 RD BNT 0.6 0 114.74 238.14 Bear Crk.* 3 GD BNT 0.65 25.23 256.07 Black R.* 4 RD BKC 0.55 1180.62 368.47 Boardman R.* 5 GD BKC, BNT 0.63 626.78 319.63 Brule R. 6 RD BNT 0.52 2719.49 515.46 Bryan Crk.* 7 RD BKC 0.6 0 175.6 432.32 Canada Crk.* 8 RD BKC 0.55 472.24 284.2 0 Carlton Crk. 9 GD BKC 0.73 20.1 235.04 Carp R.* 10 RD BKC, BNT, RBT 0.6 0 675.99 260.19 Cedar Crk.* 11 RD BKC, BNT 0.5 0 48.17 267.04 Cedar R. (SLP)* 12 RD RBT 0.6 0 114.74 336.84 Cedar R. (UP) 13 RD BNT 0.6 0 735.56 306.49 Chocolay R.* 14 RD RBT 0.56 404.04 379.63 Coldwater R. 15 RD BNT 0.45 432.53 271.94 Cooks Run 16 RD BKC, BNT 0.52 183.63 529.42 Davenport Crk.* 17 RD RBT 0.57 20.12 250.71 Duke Crk.* 18 RD BKC, BNT 0.5 0 85.73 276.93 E. Branch Fox R.* 19 GD BKC, BNT 0.73 310.8 274.93 Elm R.* 20 RD RBT 0.45 31.34 424.23 Escanaba R.* 21 RD BKC, BNT 0.44 1012.69 424.08 Falls R. 22 RD BKC 0.52 118.62 377.38 Hersey R. 23 GD BNC 0.62 297.85 368.4 0 Iron R.* 24 RD BKC 0.52 181.04 488.86 Little Indian R.* 25 GD BKC 0.73 220.93 277.8 0 Little Muskegon R. 26 GD RBT 0.62 966.07 313.04 Manistee R.* 27 GD BKC, BNT, RBT 0.65 1916.59 406.05 Mann Crk. 28 GD BKC 0.65 44.29 202.39 Martin Crk. 29 GD BKC 0.61 52.84 259.35 Menominee R. 30 RD RBT 0.57 10308.2 477.73 63 Stream Map Type Species BFI Area Elevation Miller Crk. 31 GD BKC 0.65 2.8 241.26 Mosquito Crk . 32 GD BKC 0.62 29.01 193.09 Muskegon R. 33 GD RBT 0.62 6604.47 359.73 Ogontz R. 34 GD RBT 0.74 53.16 214.53 Paint R. 35 RD BNT 0.52 1613.56 496.64 Pere Marquette R.* 36 GD BNT 0.61 1901.05 310.87 Pigeon R.* 37 GD BNT, RBT 0.65 944.5 380.18 Pine R. (SLP)* 38 RD BNT 0.49 1622.37 318.84 Pine R. (NLP)* 39 GD BKC, BNT, RBT 0.65 688.94 395.08 Prairie Crk.* 40 RD BNT 0.5 270.83 246.3 Rapid R. 41 GD BKC, BNT, RBT 0.63 212.64 355.73 Rogue R.* 42 RD BNT, RBT 0.5 678.58 278.19 Salmon Trout R.* 43 RD BKC 0.45 104.12 442.89 Silver Crk. 44 GD BKC 0.65 8 224.22 S. Branch Pine R. 45 GD RBT 0.72 114.34 273.14 St. Joseph R. 46 GD BKC, RBT 0.63 10722.6 346.56 Sturgeon R. 47 GD BKC, RBT 0.74 505.05 276.66 Tahquamenon R.* 48 RD BNT 0.55 2027.96 285.95 Tamarack Crk. 49 GD BNT, RBT 0.62 375.55 302.21 W. Branch Maple R. 50 GD BKC 0.65 598.29 282.57 W. Branch Sturgeon R.* 51 GD BKC, BNT, RBT 0.65 473.97 394.9 Yellow Dog R.* 52 RD RBT 0.52 178.71 537.07 64 Allen 1999; Kelleher et al. 2012) whereby daily streamflow records were partitioned into groundwater and surface - runoff components to determine the relative contribution of each. A BFI of 0.60 w as treated as a threshold for stream s to be categorized as groundwater - dominated (GD ; BFI > 0.60 ) or surface runoff - . Stream - specific regression models Historical a ir and water temperatures were used to develop stream - specific temperature regression models ( Table 2. 2). D aily air temperatures measured in July from 1990 to 2010 were obtained using the United States Department of Energy Historical Climate Network ( CDIAC 2016). Air temperature m easurements were reported from the gauging station closest to each , where MDNR gauges recorded daily stream temperatures i n July from 1990 to 2010. July temperatures were used because this month is typically the warmest and most thermally stressful for salmonids in Michigan (Zorn et al. 2011) and likely to be the time period that will first impact salmonid thermal habitat quality and quantity in a changing climate. In July, te mp era tures are typically more suitable (i.e. cool er ) for salmonids that live in GD, heavily forested headwater reaches of Michigan streams ( Drake and Taylor 1996; Hayes et al. 1998), so emphasis was placed on these reaches with the understanding that if they become warmer, temperatures in downstream reaches will also generally increase. The National Hydrography Dataset Plus Version 1 (NHDPlusV1) and the Watershed Boundary Dataset (USEPA 2005) were used to identify each stre sub - basin ( 8 - digit Hydrologic Unit Code [ HUC8 ] ) and sub - watershed (HUC 12) . In addition, NHDPlusV1 and the United States Geological Survey StreamStats interactive map application (USGS 2015) were used to measure drainage area ( km 2 ) 65 TABLE 2.2. Stream - specific temperature regressions with standard errors (SE), F values (i.e. F 1,7 ), P values, and R 2 values. S and A denote stream temperature and air temperature, respectively. R. denotes river and Crk. denotes creek. Southern Lower Peninsula and Northern Lower Peninsula, respectively. Stream Model SE F P R 2 Runoff - dominated Bark R. Water = 12.98 + 0.32*air 0.05 49.41 < 0.01 0.86 Bryan Cr k . Water = 7.69 + 0.42*air 0.08 28.86 < 0.01 0.78 Carp R. Water = 12.06 + 0.28*air 0.04 41.63 < 0.01 0.84 Cedar Cr k . Water = 6.32 + 0.56*air 0.09 39.10 < 0.01 0.83 Cedar R. (SLP) Water = 11.42 + 0.25*air 0.03 97.82 < 0.01 0.92 Elm R. Water = 2.11 + 0.82*air 0.06 193.20 < 0.01 0.96 Escanaba R. Water = 3.03 + 0.88*air 0.14 39.26 < 0.01 0.83 Pine R. (SLP) Water = 9.23 + 0.40*air 0.04 119.15 < 0.01 0.94 Prairie Cr k . Water = 0.95 + 0.74*air 0.07 106.03 < 0.01 0.93 Salmon Trout R. Water = 8.23 + 0.29*air 0.05 33.60 < 0.01 0.80 Tahquamenon R. Water = 12.29 + 0.50*air 0.05 115.88 < 0.01 0.93 Groundwater - dominated Bear Cr k . Water = 11.61 + 0.23*air 0.03 46.74 < 0.01 0.85 Black R. Water = 8.74 + 0.29*air 0.04 57.04 < 0.01 0.88 Boardman R. Water = 11.99 + 0.14*air 0.02 72.85 < 0.01 0.9 0 Canada Cr k . Water = 6.91 + 0.60*air 0.08 61.72 < 0.01 0.88 Chocolay R. Water = 10.29 + 0.23*air 0.03 77.99 < 0.01 0.91 Davenport Cr k . Water = 8.97 + 0.13*air 0.01 99.55 < 0.01 0.92 Duke Cr k . Water = 4.45 + 0.48*air 0.08 37.27 < 0.01 0.82 East Branch Fox R. Water = 7.73 + 0.33*air 0.04 86.45 < 0.01 0.91 Iron R. Water = 12.76 + 0.30*air 0.04 52.58 < 0.01 0.87 Little Indian R. Water = 14.86 + 0.06*air 0.01 83.65 < 0.01 0.91 Manistee R. Water = 10.67 + 0.13*air 0.02 57.70 < 0.01 0.88 Pere Marqu ette R. Water = 12.50 + 0.18*air 0.02 51.89 < 0.01 0.86 Pigeon R. Water = 11.93 + 0.11*air 0.01 60.49 < 0.01 0.88 Pine R. (NLP) Water = 10.89 + 0.22*air 0.03 73.41 < 0.01 0.90 Rogue R. Water = 13.17 + 0.23*air 0.04 27.46 < 0.01 0.77 West Branch Sturgeon R. Water = 11.93 + 0.06*air 0.01 67.49 < 0.01 0.89 Yellow Dog R. Water = 9.90 + 0.38*air 0.06 45.58 < 0.01 0.85 66 and mean elevation (m) for each stream. Stream - specific regression models were developed by pairing mean July air and water temperatures from recent years (i.e. 2002 2010) for the 28 streams for which historical stream temperatures were available (Table 2.1, 2.2). To predict future stream temperatures, the product of air temperature regression coefficients and a ir temperature projections (descri bed below) was calculated and added to model intercepts (i.e. stream temperature = air temperature coefficient*projected future air temperature + intercept). Air temperature coefficients represented indices of stream thermal sensitivity (i.e. relative susceptibility to temperature change ) because larger positive coefficients produced warmer stream temperatures ( Kelleher et al. 2012 ). Generalized regression models S tream temperatures were also predicted by converting sub - basin air temperature p r ojection s to water temperatures using two generalized regression equations. The Stefan and referred to as SP) model, developed specifically for RD streams , estimates weekly stream temperature by: T w = 2.9 + 0.86 T a , where T w is water temperature (° C) and T a is air temperature (° C). The standard deviation (SD) of the model is 2.16 ° C. The Krider et al. (2013 , hereafter referred to as K) model, developed specifically for G D streams , estimates weekly stream temperature by : T w = 6.63 + 0.38 T a , where T w is water temperature (° C) and T a is air temperature (° C). The model SD is 4.80° C. 67 T he K model was used to project GD stream temperatures, whereas the SP model was used to predict RD stream tempe ratures, as these models were developed specifically for those corresponding streams types. Although both models were developed for weekly temperatures, daily and monthly air and water temperatures have been reported to be associated through a nearly 1:1 relationship ( Ozaki et al . 2003 ), suggesting a strong positive weekly - to - monthly relationship and validating use of monthly average air temperatures to project stream temperatures. Air temperature projections M ean July air temperatures were forecast ed in future years ( 2036 and 2056 ) for each sub - basin evaluated. Air temperatures were projected using three coupled climate models : the Third Generation Coupled Global Climate Model (CGCM3, Canadian Centre for Climate Modelling and Analysis), the CM2 Global Coupled Climate Model (CM2 , Geophysical Fluid Dynamics Laboratory at the National Oceanic and Atmospheric Administration), and the Hadley Centre All models were based on the World Climate Resea rch Programme (WCRP) Coupled Model Intercomparison Project phase 3 (CMIP3) multi - model dataset and had variable thermal input parameters (e.g. solar radiation, volcanic activity, trace gases, sulfate aerosols ). S patial downscaling was performed using the B ias - Correction Spatial Disaggregation (BCSD) approach to adjust the resolution of the climate model (~200 km x 200 km) to a scale appropriate for Michigan streams (12 km x 12 km; Maurer et al. 2007 ). Mean July air temperatures were obtained for Michigan sub - basins containing salmonid streams from the United States Forest Eastern Forest Environmental Threat Assessment Center (EFETAC) in North 68 Carolina. T emperatures were projected using the Special Report of Emission Scenarios A2 and B1 cli mate forcing scenarios and the CGCM3, CM2, and HadCM3 models using area - weighted means for all years . The A2 scenario predicts atmospheric CO 2 concentrations to be 820 ppm in 2100 in a world characterized by rapid economic grow th and efficient energy technologies (IPCC 2007). In contrast, the B1 scenario projects atmospheric CO 2 concentrations to be 550 ppm in 2100 in a convergent world with a service and information economy and reduced material consu mpti on (IPCC 2007). Stream temperature and thermal habitat suitability projections S tream - specific and generalized models were used to backcast July stream temperatures in 2006 and 2012 and forecast July temperatures in 2036 and 2056 based on air temperature predictions in these years. Backcasting was performed to evaluate the accuracy of stream - specific and generalized models by comparing predicted and actual temperatures in years with pre - existing stream temperature metrics. Mean air temperatures were calculated from the three forest canopy density, atmospheric pressure, s oil layering ), and range of predicted air temperatures. S pecies - specific thermal habitat suitabili ty statuses were assigned for each stream based on temperature thresholds for growth and survival associated wi th projected July temperatures. Status 1 streams had maximum July average temperatures that were optimal for growth of brook charr (11.0 16.5 ° C; Raleigh 1982a), brown trout (12 .0 17.0° C; Raleigh et al. 1986 ; Hay and Young 2006), and rainbow trout (12.0 16.4° C; Wurtsbaugh and Davis 1977; Raleigh 1982b). Temperatures of status 2 streams resulted in reduced growth of brook charr (16.5 20.5° C; Ra leigh 1982a), brown trout (17.0 20.0° C; Elliott and Hurley 2000), and rainbow trout (16.4 22.5° C; Wurtsbaugh and 69 Davis 1977) . S tatus 3 streams had temperatures that were too warm for growth to occur : 20.5 25.3° C for brook charr (Baldwin 1957; Raleigh 1 982a), 20.0 26.2° C for brown trout (Hay and Young 2006), and 22.5 25.0° C for rainbow trout (Wurtsbaugh and Davis 1977; Raleigh 1982b). Finally, status 4 streams had thermal conditions that create high mortality risk for brook charr (>25.3° C; Fry et al. 1946 ; Raleigh 1982a), brown trout (>26.2° C; Hay and Young 2006), and rainbow trout (>25. 0° C; Wurtsbaugh and Davis 1977 ; Ra leigh 1982b). Analyses T he accuracy of stream - specific regression models was evaluated by calculating the deviation between projected and actual stream temperatures and salmonid thermal habitat suitability statuses in 2006 . Similarly, t he accuracy of generalized regression models was assessed by calculating differences between projected and actual stream temperatures and salmo nid thermal habitat suitability statuses . In comparing generalized K and SP models, temperature projections were considered inaccurate if they were lower (i.e. under - prediction) or higher ( i.e. over - prediction) than actual temperatures by an amount greater than the standard deviation of the model with the lowest standard deviation (i.e. SP, SD = 2.16 ° C ). T he relative accuracy of stream - specific and generalized models was evaluated by deviation between projected and actual temperatures and salmonid thermal habitat suitability statuses with those of the corresponding model (i.e. generalized or stream - specific) in each of the 28 streams with historical water temperatures (Table 2. 1). Stream - specific models could not be develop ed for the 24 streams (17 GD, 7 RD) for which historical water temperatures were not available. However, it was deemed appropriate to project stream temperatures and habitat suitability statuses using either the generalized K model (for the additional 17 G D streams) or the 70 SP model (for the additional 7 RD streams) if the accuracy of these models in predicting habitat suitability in their respective streams was greater than 80%. Temperature deviations (i.e. between actual temperatures and model projections and between stream - specific and generalized model predictions) were considered biologically significant if they produced changes in thermal habitat suitability statuses. Such changes were necessary for biological significance as defined in this manuscript, so temperature deviations rather than absolute temperatures were the appropriate measurement and were required for defining management - relevant categories for growth and survival . To investigate the applicability of stream - specific and generalized models in regions outside Michigan, temperatures and thermal habitat suitability projections from the two model types were compared with historical (i.e. 2006) temperatures and habitat sui tability statuses in five reported high - quality brook charr and brown trout streams located in the States of Connecticut, Maine, North Carolina, and Wisconsin , USA (Schlee 2014 ). Historical data for these streams were obtained from the United States Geolog ical Survey Na tional Water Information System (USGS 2016). Results Temperature projections Stream - specific models projected historical water temperatures more accurately than generalized models in all 28 streams that had historical water temperatures available. In GD streams, t he mean deviation of stream - specific model projections from actual temperatures in 2006 was - 0.30 ° C (SD = 0.35, range = - 0.71 0.79) under the high CO 2 emission scenario and - 0.38 ° C (SD = 0.41, range = - 1.04 0.79) under the low emission scenario . Compared to stream - 71 specific model predictions, the mean deviation of generalized K model projections in GD streams was larger under high emissions ( - 1.21° C, S D = 2.10, range = - 5.40 2.60) and low emissions ( - 1.36° C, S D = 2.13, range = - 5.60 2.30). Similarly, the mean deviation of stream - specific model projections from actual temp eratures in RD streams was - 0.70 ° C ( high emissions, S D = 0.77,range = - 1.78 1.14) ) and - 0. 88 ° C ( low emissions; S D = 0.75,range = - 1.70 0.84) , compared to 2.57 ° C ( high emissions , S D = 1.92, range = - 0.70 5.30 ) and 2.10 ° C ( low emissions, S D = 2.11, range = - 1.50 4.60 ) for the generalized SP model. For streams in the State of Michigan, s tream - specific models predicted that stream temperatures will increase by an average of 1.5 ° C ( GD streams) and 3.1 ° C ( RD streams ) under high emissions and 1.2 ° C (G D streams ) and 3.1 ° C ( RD streams ) under low emissi ons from 2016 to 2056. Throughout Michigan, g eneralized models projected stream temperatures will increase by an av erage of 0.6° C (GD streams) and 1.5° C (RD streams) under high emissions and 0.8° C (GD streams) and 1.9° C (RD str eams) under low emissions from 2016 to 2056. Thermal habitat suitability projections Although stream - specific models predicted temperatures of GD and RD streams more accurately than generalized models, both model types projected accurate salmonid therm al habitat suitability statuses in GD streams (Fi gure 2. 2). Under high and low CO 2 emissions, stream - specific models projected thermal habitat suitability with 100 % accuracy in GD streams with brook charr , brown trout, and rainbow trout (Table 2. 3). Deviations between actual temperatures and stream - specific model projections were not b iologically significant as they produced the same thermal habitat suitability statuses. In comparison, the overall accuracy of the generalized K model in projecting thermal habitat suitability was 82 % in GD streams for brook 72 FIGURE 2.2. Temperature predictions in representative Michigan streams in 2006, 2012, 2036, and 2056 using stream - specific (dotted lines) and generalized (solid lines) temperature models. Plots are organized by species and emissions scenario: (a) brook charr, A2 scenario; (b) bro ok charr, B1 scenario; (c) brown trout, A2 scenario; (d) brown trout, B1 scenario; (e) rainbow trout, A2 scenario; (f) rainbow trout, B1 scenario. In each plot, the upper stream is surface runoff - dominated, and the lower stream is groundwater - dominated. Do tted lines represent transitions between thermal habitat suitability statuses. Stream abbreviations are as follows: BAR = Bark River, BOR = Boardman River, CAR = Carp River, DAV = Davenport Creek, ESR = Escanaba River, MAN = Manistee River. 73 TABLE 2.3. Projected temperatures (° C) and associated thermal habitat suitability (THS) statuses for growth and survival of brook charr (BKC), brown trout (BNT), rainbow trout (RBT), and all three species (All) in groundwater - dominated streams (1 = optimal, 2 = subo ptimal, 3 = no growth, 4 = high mortality risk). Actual (i.e., field - measured) temperatures and THS statuses in 2006 are also included. Predictions are derived from stream - specific and generalized models (separated by commas) under the A2 carbon dioxide em issions scenario and are similar to those from the B1 scenario (not included here). R. denotes river, Crk. denotes creek, and NLP denotes the Northern Lower Peninsula of Michigan. Stream Species Year Temperature THS Actual Actual THS Bear Crk. BNT 2006 16.5,15.4 1,1 15.8 1 2012 16.7,15.1 1,1 2036 17.8,16.0 2,1 2056 18.2,16.2 2,1 Black R. BKC 2006 14.5,14.6 1,1 15.1 1 2012 15.1,14.3 1,1 2036 16.3,15.3 1,1 2056 16.0,15.0 1,1 Boardman R. BKC, BNT 2006 14.9,14.5 1,1 15 1 2012 14.9,14.3 1,1 2036 15.8,15.3 1,1 2056 15.5,14.9 1,1 Canada Crk. BKC 2006 19.4,14.6 2,1 20 2 2012 19.5,14.3 2,1 2036 22.9,15.3 3,1 2056 21.9,15.0 3,1 Duke Crk. BKC, BNT 2006 14.7,15.3 1,1 15.3 1 2012 15.8,14.9 1,1 2036 18.5,15.9 2,1 2056 19.1,16.0 2,1 E. Branch Fox R. BKC, BNT 2006 13.8,14.0 1,1 14.2 1 2012 14.1,13.8 1,1 2036 15.6,14.8 1,1 2056 15.3,14.6 1,1 Iron R. BKC 2006 18.5,14.1 2,1 18.6 2 2012 17.8,13.8 2,1 2036 19.7,15.0 2,1 2056 19.2,14.7 2,1 74 Stream Species Year Temperature THS Actual Actual THS Little Indian R. BKC 2006 16.1,14.0 1,1 16.2 1 2012 16.1,13.8 1,1 2036 16.5,14.8 2,1 2056 16.4,14.6 1,1 Manistee R. All 2006 13.3,14.6 1,1 13.6 1 2012 13.7,14.3 1,1 2036 14.3,15.3 1,1 2056 14.2,15.1 1,1 Pigeon R. BNT, RBT 2006 14.1,14.4 1,1 14.5 1 2012 14.2,14.2 1,1 2036 14.8,15.1 1,1 2056 14.6,14.8 1,1 Pine R. (NLP) All 2006 15.3,14.6 1,1 16 1 2012 15.9,14.3 1,1 2036 17.1,15.3 2,1 2056 16.9,15.1 2,1 Rogue R. BNT, RBT 2006 18.0,15.3 2,1 18.4 2 2012 17.9,14.9 2,1 2036 19.0,15.9 2,1 2056 19.2,16.0 2,1 W. Branch Sturgeon R. All 2006 13.1,14.4 1,1 13.3 1 2012 13.2,14.2 1,1 2036 13.5,15.1 1,1 2056 13.4,14.8 1,1 Yellow Dog R. RBT 2006 17.3,14.1 2,1 18 2 2012 19.7,13.9 2,1 2036 21.8,15.0 2,1 2056 21.3,14.8 2,1 75 charr ( 80 % accuracy), brown trout (89 % accuracy ), and rainbow trout ( 75 % accuracy; Table 2.3 ) under both emissions scenarios . In contrast to GD streams, differences in temperature predictions between stream - specific and generalized models in RD systems were more often biologically significant (Figure 2.2). S tream - specific models were more accurate than generalized models in pred icting salmonid thermal habitat suitability statuses in RD streams. Compared to 100 % accuracy in GD systems, stream - specific models predicted thermal habitat suitability with 8 0 % accuracy in RD streams for brook charr (80 % accuracy), brown trout (75 % accuracy), and rainbow trout (100 % accuracy; Table 2. 4) under high CO 2 emissions. Under low emissions, stream - specific models predicted thermal habitat suitability with 93% accuracy in GD streams for brook charr ( 80 % accuracy), brown trout (100 % accuracy), and rainbow trout (100 % accuracy; Table 2. 4). In comparison, under high emissions, the overall accuracy of the generalized SP model was 47 % in RD streams for brook charr (60 % acc uracy), brown trout ( 25 % acc uracy ), and rainbow trout (100 % accuracy; Table 2. 4 ) . U nder low emissions, the overall accuracy of the generalized SP model was 60 % in RD streams for brook charr (60 % accuracy), brown trout ( 50 % accuracy ), and rainbow trout (100 % accuracy; Table 2. 4). Generalized K versus SP models The generalized K model predicted stream temperatures more accurately than the generalized SP model. The generalized K model predicted historical GD stream temperatures with 75 % accuracy under high and low CO 2 emissio ns (Table 2. 3 ), whereas the generalized SP model backcasted temperatures with only 36 % accuracy under high and low CO 2 emissions (Table 2. 4). The generalized SP model over - predicted temperatures by 2.4 5.3 ° C (average 76 TABLE 2.4. Projected temperatures (° C) and associated thermal habitat suitability (THS) statuses for growth and survival of brook charr (BKC), brown trout (BNT), rainbow trout (RBT), and all three species (All) in surface runoff - dominated streams (1 = optimal, 2 = suboptimal, 3 = no growt h, 4 = high mortality risk). Actual (i.e. field - measured) temperatures and THS statuses in 2006 are also included. Predictions are derived from stream - specific and generalized models (separated by commas) under the A2 carbon dioxide emissions scenario and are similar to those from the B1 scenario (not included here). R. denotes river, Crk. denotes creek, and SLP denotes the Southe rn Lower Peninsula of Michigan. Stream Species Year Temperature THS Actual Actual THS Bark R. BNT 2006 18.9,20.4 2,3 19.9 2 2012 19.7,19.8 2,2 2036 21.8,22.2 3,3 2056 21.2,21.6 3,3 Bryan Crk. BKC 2006 15.3,19.9 1,2 16.7 2 2012 17.2,19.3 2,2 2036 19.9,21.8 2,3 2056 19.4,21.2 2,3 Carp R. All 2006 17.1,19.7 2,2 17.6 2 2012 17.9,19.0 2,2 2036 19.3,21.4 2,3 2056 19.0,21.0 2,3 Cedar Crk. BKC, BNT 2006 18.3,22.5 2,3 18.3 2 2012 18.7,21.7 2,3 2036 19.9,23.8 2,3 2056 20.1,24.2 2,3 Cedar R. (SLP) BNT 2006 16.6,20.4 1,3 17.1 2 2012 16.0,19.8 1,2 2036 17.3,22.2 2,3 2056 16.9,21.6 1,3 Escanaba R. BKC, BNT 2006 18.9,19.9 2,2 19.9 2 2012 18.6,19.3 2,2 2036 21.1,21.8 3,3 2056 20.5,21.2 3,3 Elm R. RBT 2006 17.0,19.9 2,2 17.5 2 2012 20.7,19.4 2,2 2036 25.4,22.2 4,2 2056 24.3,21.5 3,2 77 Stream Species Year Temperature THS Actual Actual THS Pine R. (SLP) BNT 2006 17.5,22.2 2,3 18.4 2 2012 18.5,21.4 2,3 2036 20.5,23.6 3,3 2056 20.7,23.8 3,3 Prairie Crk. BNT 2006 16.6,22.5 1,3 18.3 2 2012 17.9,21.7 2,3 2036 21.4,23.8 3,3 2056 22.0,24.2 3,3 Salmon Trout R. BKC 2006 13.5,19.9 1,2 14.6 1 2012 15.9,19.4 1,2 2036 17.8,22.2 2,3 2056 17.3,21.5 2,3 Tahquamenon R. BNT 2006 21.2,19.3 3,2 20 3 2012 21.6,18.7 3,2 2036 24.0,21.1 3,3 2056 23.6,20.7 3,3 78 3. 8 °C) in seven RD streams (i.e. Bryan Creek, Cedar Creek, Cedar River, Elm River, Pine River, Prairie Creek, Salmon Trout River) under high emissions and by 2.5 4.6 ° C (average 3. 7 ° C ) in six RD streams (i.e. Bryan Creek, Cedar Creek, Cedar River, Pine River, Prairie Creek, Salmon Trout River ) under low emissions ( Table 2.4 ). The generalized SP model predicted thermal habitat suitability in RD streams with 35 % lower accuracy (high emissions) and 22 % lower accuracy (low emissions) than the generalized K model in GD systems. Therefore, generalized models are best suited for use in GD streams. Data - limited streams Because the generalized K model projected historical thermal habitat suitability statuses in GD streams with relatively high accuracy (> 80 %), the K model was used to predict temperature and habitat suitability in the 17 GD systems for which historical water temperatures were unavailable. Under both high and low CO 2 emissions, all of these streams were projected to maintain optimal growing conditions in July for brook charr , brown trout, and rainbow trout until the year 2056 ( Table 2. 5). Model applicability to streams outside Michigan Thermal habitat suitability projections from stream - specific and generalized models were compared with historical habitat suitability statuses in five streams outside the State of Michigan to gauge the applicability of the two model types in different regi ons of the USA. For these streams, generalized models predicted salmonid thermal habitat suitability with comparable accuracy to stream - specific models. In five brook charr streams, 67 % of generalized model predictions were accurate, compared to 0 % of st ream - specific model predictions (Table 2. 6). 79 TABLE 2.5. List of groundwater - dominated streams without actual (i.e. field - measured) 2006 stream temperatures for which generalized models were used to predict temperatures and associated associated thermal ha bitat suitability (THS) statuses in 2006, 2012, 2036, and 2056 under A2 and B1 emissions (separated by commas). 1 = optimal, 2 = suboptimal, 3 = no growth, 4 = high mortality risk. R. denotes river and Crk. denotes creek. Species abbreviations are as follo ws: ALL = brook charr, brown trout, and rainbow trout; BKC = brook charr; BNT = br own trout; RBT = rainbow trout. Stream 2006 Pr THS 2012 Pr THS 2036 Pr THS 2056 Pr THS Au Sable R. 14.6,14.4 ALL: 1,1 14.2,14.1 ALL: 1,1 15.2,15.5 ALL: 1,1 14.9,15.1 ALL: 1,1 Carlton Crk. 14.0,13.7 BKC: 1,1 13.8,13.8 BKC: 1,1 14.8,15.1 BKC: 1,1 14.6,14.6 BKC: 1,1 Hersey R. 14.7,14.7 BNT: 1,1 14.4,14.4 BNT: 1,1 15.3,15.7 BNT: 1,1 15.3,15.3 BNT: 1,1 Little Muskegon R. 14.7,14.7 RBT: 1,1 14.4,14.4 RBT: 1,1 15.3,15.7 RBT: 1,1 15.3,15.4 RBT: 1,1 Mann Crk. 15.4,15.4 BKC: 1,1 15.1,15.1 BKC: 1,1 16.0,16.3 BKC: 1,1 16.2,16.2 BKC: 1,1 Martin Crk. 14.9,14.9 BKC: 1,1 14.6,14.6 BKC: 1,1 15.5,15.8 BKC: 1,1 15.5,15.6 BKC: 1,1 Miller Crk. 15.4,15.4 BKC: 1,1 15.1,15.1 BKC: 1,1 16.0,16.3 BKC: 1,1 16.2,16.2 BKC: 1,1 Mosquito Crk. 14.7,14.7 BKC: 1,1 14.4,14.4 BKC: 1,1 15.3,15.7 BKC: 1,1 15.3,15.4 BKC: 1,1 Muskegon R. 14.7,14.7 RBT: 1,1 14.4,14.4 RBT: 1,1 15.3,15.7 RBT: 1,1 15.3,15.4 RBT: 1,1 Ogontz R. 14.1,13.9 RBT: 1,1 13.9,13.9 RBT: 1,1 14.9,15.2 RBT: 1,1 14.7,14.7 RBT: 1,1 Rapid R. 14.5,14.5 ALL: 1,1 14.3,14.2 ALL: 1,1 15.3,15.6 ALL: 1,1 14.9,15.2 ALL: 1,1 S. Branch Pine R. 14.6,14.4 RBT: 1,1 14.2,14.1 RBT: 1,1 15.2,15.5 RBT: 1,1 14.9,15.1 RBT: 1,1 Silver Crk. 15.4,15.4 BKC: 1,1 15.1,15.1 BKC: 1,1 16.0,16.3 BKC: 1,1 16.2,16.2 BKC: 1,1 St. Joseph R. 15.6,15.5 BKC: 1,1 15.2,15.2 BKC: 1,1 16.0,16.4 BKC: 1,1 16.4,16.5 BKC: 1,2 RBT: 1,1 RBT: 1,1 RBT: 1,1 RBT: 2,2 Sturgeon R. 14.1,13.9 BKC: 1,1 13.9,13.9 BKC: 1,1 14.9,15.2 BKC: 1,1 14.7,14.7 BKC: 1,1 RBT: 1,1 RBT: 1,1 RBT: 1,1 RBT: 1,1 Tamarack Crk. 14.7,14.7 BNT: 1,1 14.4,14.4 BNT: 1,1 15.3,15.7 BNT: 1,1 15.3,15.4 BNT: 1,1 RBT: 1,1 RBT: 1,1 RBT: 1,1 RBT: 1,1 W. Branch Maple R. 14.4,14.3 BKC: 1,1 14.2,14.0 BKC: 1,1 15.1,15.5 BKC: 1,1 14.8,15.1 BKC: 1,1 80 TABLE 2.6. List of brook charr (BKC) and brown trout (BNT) streams located outside Michigan with actual temperatures (i.e., field - measured in 2006) and associated thermal habitat suitability (THS) statuses. 2 = suboptimal, 3 = no growth, 4 = high mortality risk. The table also includes projected temperatures (stream - specific [SS] model, generalized [Gen] model) and associated THS statuses in 2006. State abbreviations are as follows: Connecticu t (CT), Maine (ME), North Carolina (NC), Wisconsin (WI). Stream Actual BKT THS BNT THS SS Gen BKT THS BNT THS Broad Swamp Brook, CT -- -- 24.9 21.8 3,3 3,3 Meduxnekeag River, ME 19.6 2 2 19.8 18.4 2,2 2,2 Deep Creek, NC 26.1 4 3 24.1 23.6 3,3 3,3 Embarrass River, WI 19.7 2 2 21.3 20.2 3,2 3,3 Red River, WI -- -- 21.1 20.2 3,2 3,3 81 Thermal habitat suitability projections were consistent between generalized and stream - specific models in three of five brook charr streams. In five brown trout streams outside Michigan, 67 % of generalized and stream - specific model predictions were accurate , and t hermal habitat suitability projections were consistent between model types in all five streams ( Table 2.6 ). The accuracy of generalized mode l thermal habitat suitability predictions was similar in streams outside and inside Michigan. At least two thirds of predictions were accurate in brook charr streams (outside Michigan: 67 % accuracy ; inside Michigan: 73 % accuracy) and brown trout streams (outside Michigan: 67 % accuracy ; inside Michigan: 71 % accuracy). The relatively high accuracy of generalized models in predicting thermal habitat suitability in streams within and outside Michigan indicates their applicability in coldwater fisheries management programs in the eastern and north central United States of America. Discussion Salmonids are ecologically, socioeconomically, and culturally important throughout the world. However, their comparatively low thermal optima ( Raleigh 1982a, b; Raleigh et al. 1986) combined with global climate warming and widespread scarcity in long - term stream temperature data indicate a pressing need for reliable, cost - effective methods for projecting stream t emperatures with limited data to manage thermal habitats. Although heat budget models represent a highly accurate method for predicting stream temperatures, they are expensive, data - intensive, and require physical, hydrological, and meteorological measurem ents at small spatial extents (e.g. stream reaches; Mohseni et al. 1998; Benyahya et al. 2007), which most fisheries management agencies do not have the resources to collect or analyze. As a result, fisheries 82 management agencies need efficient, cost - effect ive alternatives to accurately project future thermal habitat conditions. This study demonstrates the utility of stream - specific and generalized temperature models for salmonid management amidst global climate change and resource limitations . This study shows that s tream - specific models, reflecting the unique influence of thermal drivers in each system (e.g. air te mperature, solar radiation, riparian shading, groundwater input, discharge, precipitation, evaporation; Bartholow 1989 ; Gu et al. 1 998) , predict temperatures more accurately than generalized, region - specific models. However, developing temperature profiles for stream - specific models requires considerable investments of funding , time, and personnel (e.g. purchase, installation, and monitoring of temperature gauges for each stream) . Despite the higher accuracy of stream - specific models for prediction of stream temperatures in this study , generalized models were comparably accurate in projecting therm al habitat suitability for salmonid growth and survival in GD streams. Consequently, generalized models can be useful f or fisheries professionals seeking to optimize the expenditure of finite resources on research and management efforts necessary to conser ve coldwater stream fisheries with climate change. The magnitude of stream warming projected in this study is similar to previous investigations conducted in and near the study area in Michigan. Water temperatures were predicted to increase by 0.1 3.8 ° C i n GD streams and 0.2 6.8 ° C in RD streams in Michigan due to projected climate change from 2016 to 2056 . These results corroborate projected stream temperature warming in other Upper Midwestern states (0.3 6.9 ° C in the State of Minnesota, USA: Pilgrim et al. 1998; 0.8 4.0 ° C in the State of Wisconsin, USA: Lyons et al. 2010). Previous research in Michigan indicates that GD streams are more thermally resilient than RD streams due to groundwater - driven thermal buffering and flow stability, which causes coldw ater 83 fishes to be more susceptible to summer growth reductions in RD systems ( Zorn et al. 2012). The present study supports this finding as the magnitude of temperature warming was projected to be greater in RD streams than GD systems with thermal buffering. Generalized temperature models accurately projected stream thermal dynamics in GD streams wi thin and outside Michigan, indicating their broad utility for salmonid management in other regions of the world with coldwater streams vulnerable to climate - induced warming. G eneralized models projected thermal habitat suitability with 82 % accuracy in GD Michigan streams and were as accurate as stream - specific models in predicting thermal habitat suitability in systems outside Michigan . This indicates that g eneralized models are useful for forecasting temperature ranges for salmonid growth and survival in broad regions of the eastern and north central USA and suggests that g eneralized models are widely applicable in other regions that contain stream salmonid populations. T hese findings are important for stream salmonid management because by using generaliz ed models, fisheries professionals can invest fewer resources (e.g. funding, time, personnel) to achieve a comparable level of accuracy in projecting thermal conditions for salmonid growth and survival and may then invest more in implementing programs that enhance the thermal resilience of streams that are likely to be impacted by a warming environment. In addition, generalized models applied on regional scales would promote salmonid management strategies (e.g. harvest regulations) that, by virtue of their regional scale, would be would be more practical for fisheries stakeholders compared to the complex assortment of site - specific regulations that would result from using stream - specific models . Moreover, fisheries professionals can use generalized models as simple tools to inform anglers and other stakeholders about probable effects of climate change on coldwater stream fish communities, 84 including reduced salmonid abundance and increase d abundance of warmwater fishes (e.g. centrarchids; Pease and Paukert 201 4). Compared to stream - specific models, generalized models could be efficiently incorporated into existing regional - level stream salmonid management programs in the States of Mich igan ( Anonymous 2000 ), Wisconsin ( WDNR 2002 ), Minnesota ( MNDNR 2011 ), and elsewhere throughout the world (e.g. Spain; Antunes et al. 1999). Facing cost - benefit tradeoffs, fisheries professionals may willingly exchange the lower accuracy of generalized models in predicting temperature for their cost - effectiveness and effic iency, particularly their ability to project thermal habitat suitability with comparable accuracy and lower resource expenditure relative to stream - specific models. Results from this study have other important implications for stream salmonid management throughout the world . As climate change increases air temperatures and decreases stream thermal habitat suitability for salmonid growth, reproduction, a nd survival (Lyons et al. 2010; Isaak et al. 2012), managing streams and their salmonid populations for thermal resilience will become an ever important task for fisheries professionals. This study indicates that by using generalized models, fisheries professionals can reduce costs associated with temperature modeling in GD streams and may then be able to al locate more resources toward thermal habitat management. Potential management actions include protect ing riparian vegetation that provides shading (Blann et al. 2002), preserving grasslands and other land cover types that promote groundwater recharge (Siitari et al. 2011 ), and maintain ing longitudinal stream connectivity that allows salmonid s to move to cooler habitats (e.g. headwater reaches; Waco and Taylor 2010) during the warm summer months . Generalized mo dels are advantageous for watershed - level salmonid management as they enable relatively accurate, cost - effective stream temperature forecasting at regional scales. 85 Fi sheries professionals can use results from generalized models to develop climate vulnerabi lity maps that facilitate prioritization of streams for thermal habitat management in a changing climate. In addition, fisheries professionals can integrate generalized model temperature projections with socioeconomic information (e.g. angler values and be haviors, stream - specific resource availability) to design decision - support tools (DSTs) that streamline decision - making for stream salmonid management. As the applications of DSTs continue to be discovered (Azadivar et al. 2009, Bitunjac et al. 2016), they have proven to be important for ecologically, sociologically informed fisheries management in a chan ging climate (Lynch et al. 2015; Lynch et al. 2016). Accurate, readily interpretable generalized models can enable fisheries professionals to collaborate w ith policy makers to ensure that stream temperature modeling is used to inform the development of policies that sustain stream salmonid populations amidst predicted changes in the This study provides a foundation for future coldwater fish eries research throughout the world regarding the effects of increased air and water temperatures on stream salmonid populations and approaches t o mitigate and adapt to t he impacts of global climate change . Although this study focused on relatively dat a - rich streams in Michigan, Wisconsin , and the eastern United States, it employed methods for developing and comparing stream - specific and generalized stream temperature models that are widely applicable in other regions of the world. Stream temperature da ta spanning multiple years are necessary for using the methods described herein, thus we encourage fisheries professionals to expand the spatial and temporal coverage of air and stream temperature monitor ing networks and thereby enhance their utility for f isheries management . Increased temperature data collection and installation of air and stream temperature gauging stations will enable comparisons between the present study and research conducted in 86 the western United States, Canada, Europe, and other regi ons of the world where stream salmonids are abundant yet stream - specific and generalized temperature models have not been juxtaposed. In addition to stream temperature modeling and thermal habitat management, further r esearch is needed to assess other pote ntial tools for sustaining stream salmonid populations and other species amidst climate change. For instance, the effects of management actions to restore stream habitat connectivity (e.g. dam removal, fish ladder installation at roadside crossings and cul verts ) on salmonid populations need to be more comprehensively assessed. Further research is needed to determine whether maintaining a diversity of salmonid genetic stocks, size classes, and prey species that to lerate temperatures projected under climate c hange (Hansen et al. 2015) is a useful strategy to increase the resilience of stream salmonid populations . The goal of this study was to construct and compare stream - specific and generalized temperature models for Michigan streams to project future thermal habitat conditions (i.e. suitability statuses) for salmonid growth and survival, develop a model comparison approach that is spatially and temporally robust and broadly applicable within and beyond Michigan , and thereby inform salmonid management t hroughout the world amidst climate change and resource limitations. Our purpose was not to assess model performance in terms of absolute temperatures. Fisheries managers in Michigan and other regions with coldwater salmonid streams are primarily concern ed with maintaining and/or improving thermal habitat conditions for salmonid growth and survival rather than evaluating model performance in terms of absolute temperatures. In these management systems, models assess ing and compar ing thermal habitat suitabilit y statuses for growth and survival are more useful than those looking at absolute temperature. We structured the present manuscript correspondingly and believe a thermal habitat suitability approach is most appropriate for communicating with fisheries mana gers and other decision - 87 makers who are principally concerned with thermal habitat conditions. These individuals are not temperature modelers per se but rather decision - makers who require information on current and future thermal habitat conditions. However , we acknowledge that other fisheries management systems throughout the world operate differently and may require distinct thermal modeling approaches, including absolute temperature assessment, that reflect unique research goal s and objectives. In summary , this study demonstrates the efficacy of generalized temperature models for stream salmonid management in a changing climate, particularly in GD streams. In RD systems, t he lower accuracy of generalized models than stream - specific models in predicting wat er temperature was often biologically significant, leading to differences in projected thermal habitat suitability statuses between model types. This suggests that fisheries professionals can reserve stream - spec ific models for systems in which the accuracy of temperature prediction is especially important, including streams that support trophy fishing opportunities, threatened/endangered species, or other ecologically or socioeconomically valuable resources. In these streams, the added costs of using stream - specific models and installing additional air and stream temperature monitoring stations may be justifiable for fisheries management agencies, given that these systems are especially valuable. I n regions that contain both GD and RD st reams , integrating generalized and stream - specific models will be an effective strategy for balancing model accuracy and efficiency . Overall, this study illustrates how fisheries professionals throughout the world can use different temperature modeling fra meworks to optimize the expenditure of finite resources as they project future stream thermal conditions and develop more effective strategies for coldwater fisheries management amidst global climate change. 88 Acknowledgements We thank T. Kwak (North Ca rolina State University) for comments that improved the quality of this manuscript. We thank K. Herreman and W. Daniel (Michigan State University); T. Zorn, T. Kolb, and T. Wills (Michigan Department of Natural Resources); and H. Quinlan (United States Fis h and Wildlife Service) for providing trout population and environmental data critical for development of this study. We thank J. Myers (USFS EFETAC) and S. Nelson and E. Hain (North Carolina State University) for assistance with data acquisition and air t emperature projections. We acknowledge the Programme for Climate Model Diagnosis and Intercomparison of the WCRP CMIP3 multimodel data set. Any use of trade, firm, or pr oduct names is for descriptive purposes only and does not imply endorsement by the U.S. Government. 89 LITERATURE CITED 90 LITERATURE CITED Anonymous (2000) Michigan stream classification: 1967 system. In: Schneider JC (ed.) 2000. Manual of fisheries survey methods II: with periodic updates. Michigan Department of Natural Resources, Fisheri es Special Report 25, Ann Arbor Antunes A, Apostolidis A, Berrebi P, Duguid A, Ferguson A, García - Marín JL, Guyomard R, Hansen MM, Hindar K, Koljonen ML, Laikre L, Largiader C, Martínez P, Nielsen EE, Palm S, Ruzzante D, Ryman N, Triantaphyllidis C (1999). Conservation genetic management of brown trout ( Salmo trutta ) in Europe. Report by the Concerted Action on Identification, Management and Exploitation of Genetic Resources in the Brown trout ( Salmo trutta ). 91 pp Arnold JG, Allen PM (1999) Validation of automated methods for estimating base flow and ground water recharge from streamflow records. J Am Water Resour Assoc 35(2): 4 11 424. doi:10.1111/j.1752 - 1688.1999.tb03599.x Azadivar F, Truong T, Jiao Y operations research and systems science approach. Expert Syst Appl 36 (2): 2971 2978. doi:10.1016/j.eswa.2008.01.080 B aldwin NS (1957) Food consumption and growth of brook trout at different temperatures. T Am Fish Soc 86 (1):323 328. doi: 10.1577/1548 - 8659(1956)86[323:FCAGOB]2.0.CO;2 Bartholow JM (1989) Stream t emperature i nvestigations: f ield and a nalytic m ethods. Instream flow information paper number 13. U.S. Fish and Wildlife Service , Biological Report Number 89, Washington , D.C., 139 pp Benyahya L, Caissie D, St - Hilaire A, Ouarda TBMJ, Bobée B (2007) A r eview of statistical w ater t emperature m odels . Can Water Resour J 32(3):179 192. doi: 10.4296/cwrj3203179 Bitunjac I, Jajac N, Katavic I (2016) Decision s upport to s ustainable m anagement of b ottom t rawl f leet. Sustainability 8(3) :1 - 23 . doi:10.3390/su8030204 Blann K, Nerbonne JF, Vondracek B (2002) Relationship of riparian buffer type to water temperature in the driftless area ecoregion of Minnesota. N Am J Fish Manage 22(2): 441 451. d oi: 10.1577/1548 - 8675(2002)022<0441:RORBTT>2.0.CO;2 Budy P, Thiede GP, Lobón - Cerviá J, Fernandez GG, McHugh P, McIntosh A, Vøllestad LA, Becares E, Jellyman P (2013) Limitation and facilitation of one of the world's most invasive fish: an int ercontinental comparison. Ecology 94:356 367. doi:10.1890/12 - 0628.1 Carlson AK , Taylor WW, Schlee KM, Zorn TG, Infante DM (2017 ) Projected impacts of climate change on stream salmonids with implications for resilience - based management. Ecol Freshw Fish 25. doi: 10.1111/eff.12267 91 CDIAC (Carbon Dioxide Information Analysis Center) ( 2016 ) Carbon dioxide information analysis center. United States Department of Energy, Oak Ridge National Laboratory. http://cdiac.ornl.gov/ . A ccessed 6 June 2016 Compagnucci R, da Cunha L, Hanaki K, Howe C, Mailu G, Shiklomanov I, Stakhiv E (2001) Hydrology and water resources. In: McCarthy JJ, Canziani OF, Leary NA, Dokken DJ, White KS (eds) Climate Change 2001: i mpacts, a daptation, and v ulnerability. International Panel on Cli mate Change Third Assessment Report. pp. 191 233 Downing JA, Cole JJ, Duarte CM, Middelburg JJ, Melack JM, Prairie YT, Kortelainen P, Striegl RG, McDowell WH, Tranvik LJ (2012) Global abundance and size distribution of streams and rivers. Inland Waters 2 :229 236. doi:10.5268/IW - 2.4.502 Drake MT, Taylor WW (1996) Influence of spring and summer water temperature on brook charr, Salvelinus fontinalis , growth and age structure in the Ford River, Michigan . Environ Biol Fish 45(1) :41 51. doi: 10.1007/BF00000626 Duki V, V (2012) Analysis of groundwater regime on the basis of streamflow hydrograph . Facta Univ 10(3):301 314. d oi: 10.2298/FUACE1203301D Elliott JM, Hurley MA (2000) Daily energy intake and growth of piscivorous brown trout, Salmo trutta . Freshwater Biol 44(2): 237 245. doi:10.1046/j.1365 - 2427.2000.00560.x Evans DH, Claiborne JB (2005) The physiology of fishes, third edition. Taylor & Francis, Boca Raton Fry FEJ, Hart JS, Walker KF (1946) Lethal temperature relations for a sample of young speckled trout, Salvelinus fontinalis Vol. 54 The University of Toronto Press, Toronto Godby NA, Jr, Rutherford ES, Mason DM (2007) Diet, feeding rate, growth, mortality, and production of juvenile steelhead in a Lake Michigan tributary. N Am J Fish Manage 27(2): 578 592. doi: 10.1577/M06 - 077.1 Gu R, Montgomery S, Austin TA (1998) Quantifying the effects of stream discharge on summer river temperature. Hydrol Sci J 43(6):885 904. doi:10.1080/02626669809492185 Hansen GJA, Gaeta JW, Hansen JW, Carpenter SR (2015) Learning to manage and managing to learn: sustaining freshwater recreational fisheries in a changing environment. Fisheries 40(2):56 - 64. doi: 10.1080/03632415.2014.996804 Hay J, Hayes J, Young (2006) Water quality guidelines to protect trout fishery values. Prepared for Horizons Regional Council. Caw thorn Report Number 1025, 17 pp Hayes DB, Taylor WW, Drake M, Marod S, Whelan G (1998) The value of headwaters to brook trout ( Salvelinus fontinalis ) in the Ford River, Michigan, USA. In: Haigh MJ, Krecek J , 92 Rajwar GS, Kilmartin MP (eds). Headwaters: water resources and soil conservation. , Oxford an d IBH Publishing Co., New Delhi, pp. 75 185 IPCC (Intergovernmental Panel on Climate Change) ( 2007 ) Climate C hange 2007: Synthesis Report. Contribution of Working Groups I, II, and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. IPCC, Geneva, Switzerland. 104 pp Isaak DJ, Muhlfeld CC, Todd AS, Al - Chokhachy R, Roberts J, Kershner JL, Fausch KD, Hostetler SW (2012) The past as prelude to the future for understanding 21st - century climate effects on Rocky Mountain trout. Fisheries 37(12):542 556 . doi: 10.1080/03632415 .2012.742808 Kaushal SS, Likens GE, Jaworski NA, Pace ML, Sides AM, Seekell D, Belt KT, Secor DH, Wingate RL (2010) Rising stream and river temperatures in the United States. Front Ecol Environ 8(9):461 466. doi: 10.1890/090037 Kelleher C , Wagener T, Goo seff M, McGlynn B, McGuire K, Marshall L (2012) Investigating controls on the thermal sensitivity of Pennsylvania streams. Hydrol Process 26(5):771 785. doi: 10.1002/hyp.8186 Knight, K ( 2009 ) Land use pla nning for salmon, steelhead and trout. Washington Department of Fish and Wildlife. Olympia, Washington. http://wdfw.wa.gov/publications/00033/ wdfw00033 . pdf . A ccessed 14 August 2015 Krider LA, Magner JA, Perry J, Vondracek B, Ferrington LC, Jr (2013) Air - water temperature relationships in the trout strea - sandstone landscape. J Am Water Resour Assoc 49(4):896 907. doi : 10.1111/jawr.12046 Kruger CC, Decker DJ (1999) The process of fisheries management. In: Kohler CC, Hubert WA (eds) Inland Fisheries Management in Nort h America, 2nd edn. American Fisheries Society, Bethesda, pp. 31 59 Lynch AJ, Varela - Acevedo E, Taylor WW (2015) The need for decision - support tools for a changing climate: application to inland fisheries management. Fisheries Management and Ecology 22 (1)14 - 24. doi : 10.1111/fme.12013 Lynch AJ, Myers BJE, C hu C, Eby LA, Falke JA, Kovach RP, Krabbenhoft TJ, Kwak TJ, Lyons J, Paukert CP , Whitney JE ( 2016 ) Climate change effects on North American inland fish populations and assemblages. Fisheries 41 (7) :346 - 3 61. doi : 10.1080/03632415. 2016.1186016 Lyons J, Zorn T, Stewart J, Seelbach P, Wehrly K, Wang L (2009) Defining and characterizing coolwater streams and their fish assemblages in Michigan and Wisconsin, USA. N Am J Fish Manage 29(4):1130 1 1 51 . doi: 10.1577/M08 - 118.1 93 Lyons J, Stewart JS, M itro M (2010) Predicted effects of climate warming on the distribution of 50 stream fishes in Wisconsin, U.S.A. J Fish Biol 77(8):1867 1898 . doi: 10.1111/j.1095 - 8649.2010.02763.x MacCrimmon, HR (1971) World distribution of rainbow trout ( Salmo gairdneri ): further observations. J Fish Res Board Can 29(12):1778 1791. doi:10.1139/f72 - 287 Magnuson JJ, Webster KE, Assel RA, Bowser CJ, Dillon PJ, Eaton JG, Evans HE, Fee EJ, Hall RI, Mortsch LR, Schindler DW, Quin n FH (1997) Potential effects of climate change on aquatic systems: Laurentian Great Lakes and Precambrian Shield Region. Hydrol Process 11, 825 871 . doi: 10.1002/(SICI)1099 - 1085(19970630)11:8<825::AID - HYP509>3.0.CO;2 - G Maurer EP, Brekke L, Pruitt T,Duffy PB (2007) Fine - resolution climate projections enhance regional climate change impact studies . Eos Trans Am Geophys Uni on 88(47): 504 . doi: 10.1029/2007EO470006 McKergow L, Parkyn S, Collins R, Pattinson, P (2005) Small h eadwater s treams of the Auckland Region. Volume 2: Hydrology and water quality. Auckland Regional Council Technical Publication No. 312. pp. 1 67 MNDNR (Minnesota Department of Natural Resources) (2015) Fisheries long - range plan for trout stream resource management in southeast Minnesota 2010 2015 and progress report. St . Paul, Minnesota: Minnesota Departm ent of Natural Resources, 50 pp Mohseni O, Stefan HG, Erickson TR. ( 1998 ) A n onlinear r egression m odel for w eekly s tream temperatures. Water Resour Res 34(10): 2685 2692. doi: 10.1029/98WR01877 Neff BD, Day SM, Piggott AR, Fuller L M ( 2005) Base flow in the Great Lakes basin. U.S. Geological Survey Scientific Investigations Report 2005 5217, 23 pp , DeWalle DR (2006) Stream - air temperature re lations to classify stream - ground water interactions in a karst setting, central P ennsylvania, USA. J Hydrol 329: 140 153 . doi: 10.1016/j.jhydrol.2006.02.010 Ozaki N, Fukushima T, Harasawa H, Kojiri T, Kawashima K, Ono M (2003) Statistical analyses on the effects of air temperature fluctuations on river water qualities. Hydrol Process 17:2837 2853 . doi: 10.1002/hyp.1437 Pease AA, Paukert CP (2014) Potential impacts of climate change on growth and prey consumption of stream - dwelling smallmouth bass in the central United Stat es. Ecol Freshw Fish 23:336 346 . doi: 10.1111/eff.12084 Pilgrim JM, Fang X, Stefan HG (1998) Stream temperature correlations with air temperatures in Minnesota: implications for climate warming. J Am Water Resour Assoc 34:1109 1121 . doi: 10.1111/j.1752 - 1688.1998.tb04158.x 94 Raleigh RF (1982a) Habitat suitability index models: brook trout. Washington, D.C.: U.S. Fish and Wildlife Service, Biol ogical Report Number 82, 42 pp Raleigh RF (1982b) Habitat suitability index models: rainbow trout. Washington, D.C.: U.S. Fish and Wildlife Service, Biol ogical Report Number 82, 76 pp Raleigh RF, Zuckerman LD , Nelson PC (1986) Habitat suitability index models and instream flow suitability curves: brown trout . Washington, D.C.: U.S. Fish and Wildlife Service, Biological Report Number 82, 65 pp Schlee KM (2014) The impact of climate change on brook trout ( Salvelinus fontinalis ) thermal habitat in their native range in the United States. MSc Thesis, Michigan State University, East Lansing, 98 pp Siitari KJ, Taylor WW, Nelson SAC. Weaver KE (2011) The influence of land cover composition and groundwater on thermal habitat availability for brook charr ( Salvelinus fontinalis ) populations in the United States of America. Ecol Freshw Fish 20:431 437 . doi: 10.1111/j.1600 - 0633.2011.00487.x Stacey, NE (1984) Control of the timing of ovulation by exogenous and endogenous factors. In: Potts GW, Wootton RJ (eds) Fish reproduction: strategies and tactics. Academic Press, London, pp. 207 222 Stefan HG , Preud'homme EB (1993) Stream temperature estimation from air temperature. Water Res Bul Am Water Res Assoc 29:27 45 . doi: 10.1111/j.1752 - 1688.1993.t b01502.x Stoner AMK, Hayhoe K, Yang XH , Wuebbles DJ (2013) An asynchronous regional regression model for statistical downscaling of daily climate variables. Internat J Climat 33 :2473 2494 . doi: 10.1002/joc.3603 USEPA (United States Environmental Protection Agency) (2005) National Hydrography Dataset Plus - NHDPlus. http://ww w.horizon - systems.com/nhdplus/ A ccess ed 6 June 2016 USFWS ( United States Fish and Wildlife Service) ( 2011) 2011 National survey of f ishing, h unting, and w ildlife - a ssociated r ecreation. Washington , D.C.: U.S. Department of the Interior, U.S. Fish and Wildlife Service, and U.S. Department of Comme rce, U.S. Census Bureau. 172 pp USGS ( United States Geological Survey) ( 2015) StreamStats. http://water.usgs.gov/osw/ streamstats/ Accessed 19 December 2016 USGS ( 2016) National Water Information System. http://waterdata.usgs.gov/nwis Accessed 6 June 2016 95 van Vliet MTH, Franssen WHP, Yearsley JR, Ludwig F, Haddeland I, Lettenmaier DP, Kabat P (2013) Global river discharge and water temperature under climate change. Glob Environ Chang 23(2):450 1 464 . doi: 10.1016/j.gloenvcha.2012.11.002 Waco KE, Taylor WW (2010) The influence of groundwater withdrawal and land use changes on brook charr ( Salvel inus fontinalis ) thermal habitat in two coldwater tributaries in Michigan, USA. Hydrobiologia 650:101 116 . doi: 10.1007/s10750 - 010 - 0204 - 0 Wahl KL, Wahl TL (1988) E ffects of regional ground water declines on streamflows in the Oklahoma Panhandle. In: Waterstone M, Burt RJ (eds). Proceedings of the s ympos ium on water - use data for water r esources management. Tucson: American Water Resources Association, pp. 239 - 249 Weber GM, Hostuttler MA, Semmens KJ, Beers BA (2015) Indu ction and viability of tetraploids in brook trout ( Salvelinus fontinalis ). Can J Fish Aquat Sci 72(10):1443 1449. doi:10.1139/cjfas - 2014 - 0536 Wehr ly KE, Brenden TO, Wang L (2009) A comparison of statistical approaches for predicting stream temperatures across heterogeneous landscapes. J Am Water Resour Assoc 45(4):986 997 . doi: 10.1111/j.1752 - 1688.2009.00341.x WDNR (Wisconsin Department of Natural Resources) (2002) Wisconsi n trout streams . Madison, Wisconsin: Wisconsin Department of Natural Resources, Report PUB - FH - 806. 103 pp Wurtsbaugh WA, Davis GE (1977) Effects of fish size and ration level on the growth and food conversion efficiency of rainbow trout, Salmo gairdneri Richardson. J Fish Biol 11: 99 104 . doi: 10.1111/j.1095 - 8649.1977.tb04102.x Zorn TJ, Seelbach PW, Rutherford ES (2012) A regional - scale habitat suitability model to assess the effects of flow reduction on fish assemblages in Michigan streams. J Am Water Resour Assoc 48:871 895 . doi: 10.1111/j.1752 - 1688.2012.00656.x Zorn TJ, Seelbach PW, Wiley MJ (2011) Developing u ser - f rie ndly h abitat s uitability t ools from r egional s tream f ish s urvey d ata. North Am J Fish Manage 31:41 5 . doi: 10.1080/02755947.2011.557965 96 CHAPTER 3: SCIENCE TO ACTION: DECISION - SUPPORT TO ADVANCE STREAM TROUT MANAGEMENT IN A CHANGING CLIMATE Carlson, A. K., W . W . Taylor, Z. Basher, T. D. Beard, Jr., D. M. Infante. 2017. Science to action: decision - support to advance stream trout management in a changing climate. Proceedings of the Wild Trout XII Symposium 12:85 - 92. The content of this chapter is base d on the published manuscript cited above and reflects specifications (e.g., formatting ) for Proceedings of the Wild Trout XII Symposium . 97 Abstract Decis ion - making with limited information is commonplace in fisheries management, stemming from the need to sustain fisheries ecosystems in the face of changing environmental and human conditions. Decision support tools (DSTs) facilitate decision - making by systematically in tegrating environmental an d socioeconomic information and accounting for variability in human and natural systems, yet they have not been widely applied in freshwater recreational fisheries management . As such, w e collaborated with fisheries research and management professionals to develop a DST specifically, a stream prioritization tool (SPT) to inform fisheries management am id climate change in Michigan coldwater streams inhabited Brook Trout Salvelinus fonti nalis , Brown Trout Salmo trutta , and Rainbow Trout Oncorhynchus mykis s . The SPT ranked streams by synthesizing management decision - making c riteria that affect trout thermal habitat quality (e.g., current and future stream temperature, trout abundance, groundwater input ). Productive, socioeconomically important trout streams with high thermal habitat quality such as the Au Sable and Manistee rivers were predictably the highest - ranked streams by the SPT and thus warrant continued trout population and thermal habitat management (e.g., groundwater conservation). However, certain streams currently important for recreational fishing (e.g., Muskegon River, Pere Marquette River) were projected to have relatively low thermal habitat quality by 2056, whereas other streams without top - tier fisheries (e.g., Rapid River, Davenport Creek) were predicted to have high - quality thermal habitats, suggesting they merit increased management efforts. Revealing unexpected yet management - relevant findings under different scenarios of climate change, the SPT is a flexible instrument to help sustain th ermally resilient trout populations and streamline fisheries management decision - making amid climate change. 98 Introduction The complexity and variability of coupled human and natural systems (CHANS) such as fisheries complicates their investigation and management by generating uncertainty about system dynamics (Liu et al. 2007). For example, the intricate interactions among fishes, fish habitats, and fisheries stakeholders often compel managers to make decision s with incomplete current and future kn owledge of fisheries ecology (e.g., population size, habitat use) or human dimensions (e.g., stakeholder values, behavior, demographics). As climate change, invasive species, habitat alteration, and other stressors continue to increase the complexity and u ncertainty of fisheries management (Hansen et al. 2015), it is important to develop approaches and tools that enable managers and allied stakeholders to make efficient, effective decisions to enhance fisheries sustainability . Decision support tools (DSTs) broadly defined as information systems that organizations and individuals use to make informed decisions can facilitate fisheries decision - making amid uncertainty by allowing information synthesis (e.g., biological, social, economic, political) and eva luation of trade - offs among different choices that fisheries professionals face (NRC 2010), thereby streamlining fisheries management. The complexity of fisheries as CHANS demands that fisheries professionals and stakeholders make decisions by integrating b iological, social, economic, and political information to manage and conserve fisheries. Informed decision - making using biological information requires that fisheries professionals consider the abundance, growth, and diversity of fish populations and the ir habitats and threats they face to make policy and management choices that enhance fisheries productivity and sustainability. Socially, evaluation of networks among professionals and stakeholders (e.g., policy makers, fishers, general public) and the ethics, values, and norms th at influence these networks is necessary for socially informed 99 management of fisheries as CHANS (Ward et al. 2016; Arlinghaus et al. 2017) . Commercial, recreational, and artisanal f isheries also support markets and economies sp anning local to international scales , thus decision - quality of life and the well - being of entire communities and nations ( Fréon et al. 2014 ; Carlson et al. 2017a, 2017b ). Finally , fisheries are generally managed by government agencies (e.g., state, federal) whose decisions are inf luenced to varying degrees by government structure, the timescale of political appointments, and associated political conditions ( Lynch et al. 2015 ). With such complex, multidimensional factors to consider, fisheries professionals would benefit from using DSTs that systematically and efficiently integrate b iological, social, economic, and political information to facilitate management decision - maki ng. Fisheries decision support is a nascent field ( Lynch et al. 2015 ) due to a historical research emphasis on the biology and ecology of fishes and their habitats as opposed to their integration with human dimensions that impact the societal value of fisheries. However, extant fisheries DSTs demonstrate the value of these tools for systematically informing integrated decision - making processes. For instance, Dichmont et al. (2013) develop ed a DST to aid Northern Prawn Pandalus borealis managers in selecting spatial closure strate gies for Marine Protected Areas (MPAs) that optimized the attainment of socio economic and ecological objectives (e.g., consistent revenue for fishers, productive fish stocks). Likewise , Stortini et al. (2015) designed a DST to allow fisheries managers to v isualize options for setting no - take boundaries to maximize fishery sustainability while minimizing socioeconomic costs to fishers in the St. Anns Bank MPA, Scot ian Shelf, Canada. Although they have not been broadly applied in fisheries management, particu larly in freshwater systems, DSTs have numerous applications with the potential to inform and advance inland fisheries management programs. Not only do 100 they synthesize ecological, economic, social, and political information to promote robust decisions, DST s are compatible with structured decision - making, adaptive management, and related frameworks being used or proposed for use in fisheries management systems (Lynch et al. 2015). As a result , fisheries professionals can use DSTs to model diverse environmental and socioeconomic conditions and complement extant decision - making approaches to make better informed policy and management choices. Fisheries decision support is particularly important amid climate change, a complex issue with diver se (and often uncertain) ecological, social, and political consequences that are affecting fisheries productivity at multiple scales. On an individual level, climate change can decrease fish growth and survival by increasing cortisol levels and reduce aero bic scope by depleting dissolved oxygen levels ( Whitney et al. 2016 ). Warmwater fishes, particularly those in families Cyprini dae, Catostomidae, Ictaluridae, Centrarchidae, and Percidae that require warm temperatures for individual and population persisten ce (Lyons 1996; Lyons et al. 2009), may experience physiological benefits (e.g., higher reproductive success) resulting from warmer water temperatures . However , climate change is projected to decrease the recruitment, growth, abundance, and distribution of coldwater fishes those in families Salmonidae and Cottidae with temperature p C and critical therm 31 °C (Raleigh 1982a, 1982 b ; Raleigh et al. 1986; Lyons et al. 2009; Carlson et al. 201 7c ). As such, c limate change will also impact fisheries stakeholders (e.g., fisheries professionals , commercial and recreational fishers ) by altering fish abundance , growth, and distribution , which may interfere with ability to catch, eat, research, and manage fish ( Hunt et al. 20 16) . Given these multi - scale effects of climate change on fisheries , adaptation strategies should focus on enhancing the social - ecological resi lience of fisheries ecosystems and human 101 systems (i.e., their ability to retain structure and function amid climate change; Holling 1973; Paukert et al. 2016 ). Such resilience - based management will require collaboration among scientist s, managers, biologists, policy makers, and public stakeholders within and outside the fisheries discipline whose decisions individual or organizational affect the viability and productivity of fish populations and their habitats ( Carlson et al. 2017c ). By integrating ecologi cal, social, economic, and political information and providing a platform for multidisciplinary tea mwork, DSTs facilitate fisheries decision - making that builds social - ecological resilience . The goal of this study was to develop a DST specifically, a stream prioritization tool (SPT) to rank coldwater streams based on to management decision - making cr iteria that affect trout thermal habitat quality (e.g., current and future stream temperature, relative abundance of trout, groundwater input). In turn, fisheries professionals can use the SPT to plan management programs that promote thermally resilient streams and sustain socio - ecologically valuable Michigan stream trout populations (i.e., Brook Trout Salvelinus fontinalis , Brown Trout Salmo trutta , Rainbow Trout Oncorhynchus mykiss ; USFWS 2011 ) amid climate change. These species were selected because t hey are ecologically and recreationally important and widely distributed throughout Michigan ( Zorn et al. 2011, 2012 ), making them effective indicator species for assessing the impacts of climatic warming on coldwater stream organisms (Carlson et al. 2017 b ). Our objectives were to: 1) forecast effects of air temperature warming due to climate change on future stream temperatures and thermal habitat suitability for trout growth and survival ; 2) survey Michigan fisheries professionals regarding their perspect ives about climate change effects on stream trout populations, coldwater habitats, fisheries management, and associated decision - making; and, 3) evaluate the perceptions of fisheries professionals regarding SPT development 102 (e.g., important management issue s and decisions) and design (e.g., effective ways to create and deliver an SPT). We then used these results to develop an SPT to support Michigan fisheries professionals as they make decisions for resilience - based salmonid management amid climate change . Methods Study area Michigan was an ideal study area in which to develop a n SPT for coldwater fisheries management, as Brook Trout, Brown Trout, and Rainbow Trout are distributed throughout 31,000 km of streams (Godby et al. 2007; Tyler and Rutherford 2007) and support productive recreational fisheries in which more than 585,000 anglers spent 8.2 million angling days in 2 011 (USFWS 2011). In addition, Michigan trout streams were important to study as coldwater stream temperatures throughout the Midwestern United States are projected to increase by 0.8 4.0 °C amid air temperature warming, with potentially effects on trout g rowth, reproduction, and survival (Pilgrim et al. 1998; Lyons et al. 2010). Latitud inal location , hydrology (i.e., groundwater/surface - runoff dominance) , and recreational importance were used as criteria for selecting 52 streams throughout the State of Mic higan for this study (Figure 3.1). Streams were selected across a latitudinal thermal gradient spanning the State of Michigan from north to south to reflect the existing and projected variation in average air temperature amid climate change. Streams were a lso selected to encompass a hydrological gradient (i.e., groundwater dominance to surface - runoff dominance) over which Brook T rout, Brown Trout , and Rainbow Trout occur in Michigan. Moreover, all streams had 103 FIGURE 3.1. Map of 52 Brook Trout, Brown Trout, and Rainbow Trout streams in Michigan, USA. The map was produced using the stream prioritization tool described herein. Map numbers refer to individual streams: Au Sable R. (1), Bark R. (2*), Bear Creek (3*), Black R. (4*), Boardman R. (5*), Brule R . (6), Bryan Creek (7*), Canada Creek (8*), Carlton Creek (9), Carp R. (10*), Cedar Creek (11*), Cedar R. (12*), Cedar R. (13), Chocolay R. (14*), Coldwater R. (15), Cooks Run (16), Davenport Creek (17*), Duke Creek (18*), East Branch Fox R. (19*), Elm R. (20*), Escanaba R. (21*), Falls R. (22), Hersey R. (23), Iron R. (24*), Little Indian R. (25*), Little Muskegon R. (26), Manistee R. (27*), Mann Creek (28), Martin Creek (29), Menominee R. (30), Miller Creek (31), Mosquito Creek (32), Muskegon R. (33), Ogo ntz R. (34), Paint R. (35), Pere Marquette R. (36*), Pigeon R. (37*), Pine R. (38*), Pine R. (39*), Prairie Creek (40*), Rapid R. (41), Rogue R. (42*), Salmon Trout R. (43*), Silver Creek (44), South Branch Pine R. (45), St. Joseph R. (46), Sturgeon R. (47 ), Tahquamenon R. (48*), Tamarack Creek (49), West Branch Maple R. (50), W. Branch Sturgeon R. (51*), Yellow Dog R. (52*). Asterisks indicate that streams had historical field - collected air and stream temperatures for development of stream - specific regress ion models. 104 recreational fisheries for one or more of these species, making them important for fisheries professionals with the Michigan Department of Natural Resources (MDNR). Fisheries professional survey In its 2013 2017 Fisheries Division Strategic Plan (MDNR 2013 ) , the MDNR explicitly identified the need for DSTs and described their value for optimizing fisheries management programs, which provided justification and enthusiasm for the fisheries professional survey and SPT developed herein. Fisheries managers and biologists employed by the MDNR were surveyed to evaluate their opinions and perspectives regarding management of Brook Trout , Brown Trout , and Rainbow Trout in Michigan given the threat of a warming climate. A 30 - question survey instrument approved by the Michigan State University Institutional Review Board (IRB # x16 - 1438e ; i052807) was designed using SurveyMonkey® and emailed to 40 fisheries professional s (23% of MDNR fisheries workforce) encompassing all fisheries management offices throu ghout the state of Michigan . Reminder emails were sent every two to three weeks during a 2.5 - month time span from November 2016 to February 2017 in which the survey was open. In total, 31 fisheries professionals responded to the survey, for a response rate of 78%. The survey encompassed a range of questions designed to assess the perspectives of fisheries professionals regarding current stream trout management strategies in Michigan; how resource availability (e.g., money, time, personnel) and thermal, hydr ological, and biological conditions influence the management strategies they select (e.g., stocking, habitat protection/rehabilitation) ; and their perspectives regarding the essential components of a n SPT to inform resilience - based salmonid management . Fis heries professionals were also asked to rank the relative importance of SPT criteria for evaluating and ultimately prioritizing streams for 105 current and future resil ience - based management actions . These criteria included six factors that regulate thermal ha bitat quality and resilience amid climate change (Siitari et al. 2011, Carlson et al. 2017b ,c ): stream temperature, trout population characteristics (i.e., presence/absence, relative abundance), groundwater contribution, high - quality watershed land cover, high - quality riparian land cover, and projected changes in stream temperature resulting from climate change (Table 3.1, Figure 3.2). Four land cover types typically associated with optimal or near - optimal trout habitats were used to define the - qualit watershed and riparian land cover SPT criteria: deciduous forest, evergreen forest, mixed forest, and grassland ( Siitari et al. 2011 ; Carlson et al. 2016 ). Criteria rankings were then used to calculate criterion - specific weights (i.e., relative importan ce values) for incorporation into the SPT (Table 3.1). Survey questions were designed in consultation with communications and survey specialists from the MDNR and Michigan State University to ensure that they were succinct, yet detailed enough to provide n ecessary information f or stream trout management and SPT development. Stream temperature Daily water temperatures for all 52 streams studied herein were obtained from an MDNR database that contained temperature records from multiple reaches within each stream from 1990 to 2010. The temperature selected because these reaches are generally coolest and most optimal for Michigan trout in summer ( Drake an d Taylor 1996; Hayes et al. 1998). As such, headwaters we re the focus of this study because if their temperatures increase, temperatures in downstream temperatures in downstream reaches will also generally rise. The National Anthropogenic Barrier Dataset ( Ostroff et al. 2013) was used to locate and omit stream temperature gauges directly below dams, 106 TABLE 3.1. Stream prioritization tool criteria and associated variables, data sources, and weights (i.e., relative importance values) for evaluating and priori tizing streams for fisheries management according to trout population and thermal habitat quality. Abbreviations are as follows: MDNR (Michigan Department of Natural Resources), USGS (United States Geological Survey), and NLCD (National Land Cover Database ). Criterion Variable(s) Source Weight Water temperature Temperature (°C) MDNR 0.23 Trout population characteristics Presence/absence, catch - per - unit - effort (# fish/mile) MDNR 0.20 Groundwater contribution Base flow index USGS, Neff et al. (2005) 0.17 High - quality watershed land cover % deciduous/evergreen/mixed forest and grassland in watershed NLCD 2011 (Homer et al. 2015) 0.14 High - quality riparian land cover % deciduous/evergreen/mixed forest and grassland in riparian zone NLCD 2011 (Homer et al. 2015) 0.14 Future/projected water temperature Temperature (°C) Carlson et al. 2017 b,c 0.11 107 FIGURE 3.2. Conceptual map depicting connections between important thermal drivers in Michigan trout streams (i.e., stream prioritization tool criteria [boxes]) and resilience - based management that promotes thermally resilient streams. 108 which generally increase temperatures relative to upstream reaches in Michigan streams (Lessard and Hayes 200 3 ) and thus bias water temperature models and associated predictions. Mean July stream temperatures were calculated for each stream because this month, generally the warmest and most thermally stressful for trout in Michigan (Zorn et al . 2011), is likely to be the time period during which trout thermal habitat quality and quantity is first affected by climate change. Mean daily water temperatures in July from the year 2006 were used to define the water temperature criterion for the SPT because records for this year were most spa tially extensive and complete. The pr edicted change in w ater temperature resulting from climate change projections for each strea m was also used as a n SPT criterion (Figure 3.2) . S tream temperatures in the year 2056 were projected by integrating future air temperature pr edictions into air - water temperature regression models. Stream - specific regression models were developed by pairing historical air and water temperatures for each stream. The United States Department of Energy Historical Climate Network (CDIAC 2016) was used to obtain daily July air temperatures measured from 1990 to 2010 for each stream . Air temperatures we re collected from the gauging station closest - specific regression models were generated by pairing mean July air and water temperatures from recent years (i.e., 2002 2010) f or the 28 streams for which historical stream temperatures were available ( Figure 3. 1 ). If streams did not have historical water temperatures, generalized (i.e., region - specific) air - water temperature models were used (Carlson et al. 2017b). In particular, t he stream temperature model was developed specifically for surface runoff - dominated streams and estimates weekly stream temperature by: 109 T w = 0.86T a + 2.9, where T w is water temperature (°C) and T a is air temperature (°C). To predict water temperatures in groundwater - dominated streams, the Kride r et al. (2013) model estimates weekly stream temperature is estimated by: T w = 0.38T a + 6.63, where T w is water temperature (°C) and T a is air temperatur e (°C). After developing air - water temperature models for all 52 streams, future stream temperatures were predicted and incorporated into the SPT as indices of projected changes in water temperature resulting from changes in air temperature. Mean July air temperatures were forecasted in 2056 using three coupled climate models (CCMs) based on the World Climate Research Programme (WCRP) Coupled Model Intercomparison Project phase 3 (CMIP3): the Third Generation Coupled Global Climate Model (CGCM3, Canadian C entre for Climate Modelling and Analysis), the CM2 Global Coupled Climate Model (CM2, Geophysical Fluid Dynamics Laboratory at the National Oceanic and Atmospheric Administration) and the Hadley Centre Coupled Model version 3 (HadCM3, Met Office, United Ki Service). The bias - correction spatial disaggregation (BCSD) approach was used to spatially suitable for Michigan streams (Maurer et al. 2007). Air temp eratures were projected for each stream s sub - basin (HUC8) as defined in the National Hydrography Dataset Plus Version 1 (NHDPlusV1). Mean projected air temperatures from the three CCMs were calculated to account for the 110 uncertainty, temperature drivers (e .g. , forest canopy density, atmospheric pressure, soil layering), and range of predicted air temperatures unique to each model. Finally, future stream temperatures wer e predicted by calculating the products of air temperature regression coefficients and CC M air temperature means and adding them to model intercepts (i.e., stream temperature = air temperature coefficient*projected future air temperature + intercept). Future stream temperature projections were then compared with s pecies - specific temperature t hresholds for growth and survival to assign thermal habitat suitability statuses for each stream. We obtained threshold temperatures from United States Fish and Wildlife Service (USFWS) Biological Reports (e.g., Raleigh 1982a, 1982 b ; Raleigh et al. 1986) containing thermal habitat status temperature ranges for juveniles and adults of each species. Other sources (i.e., Fry et al. 1946; Baldwin 1957; Elliott and Hurley 2000; Hay et al. 2006) contained temperature ranges for juveniles or adults and were used to confirm temperatures reported in the USFWS reports. When threshold temperatures (e.g., thermal minima, maxima) differed between juveniles and adults, we reported juvenile temperatures under the premise that resilient salmonid fisheries can only be conse rved if young fish survive to adulthood. We designated status 1 streams as those that had m ean July temperatures that were optimal for growth of Brook Trout (11.0 16.5 °C; Raleigh 1982a), Brown Trout (12.0 17.0 °C; Raleigh et al. 1986 ; Hay et al. 2006), an d Rainbow Trout (12.0 16.4 °C; Raleigh 1982b). Temperatures of status 2 streams were suitable, but not optimal, for growth of Brook Trout (16.5 20.5 °C; Raleigh 1982a), Brown Trout (17.0 20.0 °C; Elliott and Hurley 2000), and Rainbow Trout (16.4 22.5 °C; R aleigh 1982b). Status 3 streams were too warm for growth (in July) of Brook Trout ( 20.5 25.3 °C ; Baldwin 1957 ; Raleigh 198 2a ), Brown Trout ( 20.0 26.2 °C ; Hay et al. 2006 ), and Rainbow Trout ( 22.5 25.0 °C ; Raleigh 1982b). Finally, temperatures of status 4 streams resulted in high 111 mortality risk for Brook Trout (> 25.3 °C; Fry et al. 1946 ; Raleigh 1982 a), Brown Trout ( > 26.2 °C; Hay et al. 2006), and Rainbow Trout (> 25.0 °C; Raleigh 1982b). Trout population characteris tics The presence/absence and relative abundanc e of Brook Trout , Brown Trout , and Rainbow Trout (i.e., catch per unit effort [number of fish/ mile ] from standardized electrofishing surveys ) were also used as an input criterion for the SPT (Figure 3.2). These trout population characteristics were obtained for all 52 streams from a MDNR report documenting the status and trends of Michigan stream resources ( Wills et al. 2015). Relative abundance was converted to a categorical measurement a s defined by the MDNR (Wills et al. 2015 ) : high abundance (>75 th percentile ; Brook Trout: > 560 fish/mile; Brown Trout: > 548 fish/mile; Rainbow Trout: > 128 fish/mile ), medium abundance (25 th - 75 th percentile ; Brook Trout: 29 - 560 fish/mile; Brown Trout: 39 - 548 fish/mile; Rainbow Trout: 13 - 128 fish/mile ), and low abundance (< 25 th percentile ; Brook Trout: < 29 fish/mile; Brown Trout: < 39 fish/mile; Rainbow Trout: < 13 fish/mile). Groundwater contribution to streamflo w Groundwater was us ed as an input crit erion for the SPT (Figure 3.2) because it has been reported to provide temperature buffering and flow stability that often make streams thermally suitable for trout throughout the year (Menberg et al. 2014; Carlson et al. 2017c ). Stream - specific groundwater inputs were assessed using baseflow, the proportion of streamflow attributable to groundwa ter. Baseflow was expressed as baseflow index (BFI), the mean baseflow (mm*year - 1 ) divided by total streamflow (mm*year - 1 ), values of which were obtained 112 from a United States Geological Survey report (Neff et al. 2005). BFI values range from zero (i.e., no groundwater) to one (i.e., all groundwater; Wahl and Wahl 1988). A digital filter hydrograph separation technique was used to calculate BFI by partitioni ng daily streamflow records into their groundwater and surface - runoff components. Groundwater - dominated streams were those with BFI > 0.60, whereas surface runoff - 0.60 (McKergow et al. 2005 ; ). Watershed and riparian land cover deciduous forest, evergreen forest, mixed forest, or grassland w ere used as SPT input criteri a (Figure 3.2) because these land cover types are typica lly associated with optimal or near - optimal trout thermal habitats (Siitari et al. 2011; Carlson et al . 2016 ). W atershed and riparian land cover were evaluated using the National Land Cover Database (NLCD) 2011 (Homer et al. 2015). Riparian zone width was defined as 100 m to be consistent with previous trout stream research (Vondracek et al. 2005 ; Carlson et al. 2016 ). Stream prioritization tool development The SPT was dev eloped in collaboration with fisheries professionals from MDNR and the United States Geological Survey ( USGS ) using the Management Unit Prioritization Tool framework ( Rohweder et al. 2015a, 2015b). R esults from the fisheries professional survey were used to define six SPT criteria important for resilience - based s almonid management (Figure 3.2), along with their weights (Table 3.1). Higher weights corresponded to higher priority, with the sum of weights for all criteria equaling one. 113 Output scores for each criterion were calculated such that higher scores correspo nded with higher - quality thermal habitat conditions for trout as defined by temperature, groundwater input, and watershed/riparian land cover or, for the trout population characteristics criterion, higher - quality trout populations (as defined by species presence and relative abundance). Because higher BFI values cor respond ed with higher - quality trout thermal habitat conditions, BFI values were not adjusted to calculate scores for the groundwater contribution criterion. However, water temperatures in 2006 and 2056 had to be adjusted so that lower (i.e., more optimal) temperatures produced greater SPT scores, and vice versa. This was achieved by subtractin g each temperature from a reference value (i.e., 25 °C) greater than the maximum temperature so that cooler temperatures received higher scores and warmer temperatures received lower scores. Scores for the trout population characteristics criterion were assigned in binary format for presence (score = 1) and absence (score = 0) and increasing numerical format for low relative abundance (score = 1), medium relative abunda nce (score = 2), and high relative abundance (score = 3 ; Wills et al. 2015). Scores were summed across species such that streams with Brook Trout , Brown Trout , and Rainbow Trout in high abundance received higher scores than streams with low abundance or ab sent trout sp ecies. C umulative percentage s o f each comprised of deciduous forest, evergreen forest, mixed forest, and grassland represented score s for the watershed and riparian land cover criteri a, respectively . For e ach stream, output scores for each input criterion were normalized to a consistent scale of 0 to 100 using the following formula (Rohweder et al. 2015a, 2015b): 114 s individual stream score performed indi vidually for each input criterion . Normalized scores for each criterion were multi plied by manager - defined weights to produce weighted scores. Then, weighted scores were added to produce a final output score for each stream representing its importance for current and Figure 3.3). Lastl y , streams were ranked by final output scores The SPT was designed and delivered to Michigan fisheries professionals (e.g., managers, biologists, policy makers ) for use in stre am trout management via Data Basin ( https://databasin.org ) , a n open - access mapping and analysis platform that allows users to specify input criteria, set constraints, define management area boundaries, and visualize ou tcome s of potential decisions . Spatial data representing each of the six input criteria were assembled in ArcMap 10.4 (Environmental Systems Research Institute, Redlands, California) an d exported to Data Basin for display and dissemination to fisheries professionals. Choosing an open - access platform such as Data Basin, as opposed to a proprietary geographic information systems platform with limited accessibility, was an important prerequisite for delivering the SPT to Michigan fisheries professionals in an accessible, user - friendly manner. Data Basin was configured to display six map layers corresponding to each of the SPT criteria. Each criterion layer was e quipped with clearly defined symbols and color codes to depict important difference among streams (e.g., water temperatures, trout population characteristics, watershed land cover) . A final map layer was created to display stream importance scores and allow comparisons among streams. SPT users have the ability to enable/disable these seven map layers a t their discretion to create map combinations in accordance with their decision - making needs (e.g., evaluating stream 1 15 FIGURE 3.3. Example of how normalized output scores (N_score) for each input criterion (Table 1) were multiplied by manager - defined weights to produce weighted scores (W_score). These scores were then added for all criteria to produce a final output score for each stream reflecting its relative importance for trout management in a changing climate. 116 thermal regimes, assessing st ream riparian habitat conditions, deciding which streams warrant habitat rehabilitation). The SPT is available for fisheries professionals and other interested parties at http://bit.ly/2KPPojq . Results Stream t emperature Streams which currently have optimal temperatures for trout growth and survival were projected to sustain their high - quality thermal conditions through at least 2056. Davenport Creek (eastern Upper Peninsula; Figure 3.1) had the coolest current July water temperature (11.5°C) and therefore received the highest ranking for the water temperature criterion , making it the most thermally optimal trout stream in the present (Table 3.2). Other highly - ranked streams with respect to water temperat ure included the West Branch Sturgeon River (13.3 °C , northern Lower Peninsula) , Manistee River (13.6 °C , northern Lower Peninsula) , and Carlton Creek (14.0 °C , western Lower Peninsula; Figure 3.1). Fo r the projected future water temperature criterion , Davenp ort Creek and the West Branch Sturgeon and Manistee rivers were again the highest - ranked streams by the SPT , with pr edicted mean July water temperatures in the year 2056 that ( Table 3.2). Trout population c haracteristics Th e highest - ranked systems for the trout population characteristics criterion were three of Michigan ( Table 3.2 ). High rankings for the Au Sable, Manistee, and West Branch Sturgeon rivers ower Peninsula (Figure 3.1 ) reflected the fact that all three trout species (i.e., Brook Trout , Brown Trout , 117 TABLE 3.2. Overall stream rankings based on six stream prioritization tool (SPT) c riteria. Columns correspond to six SPT criteria: water temperature (Temp); trout population characteristics (Trout); groundwater contribution to stream flow (GW) expressed as baseflow index, t he mean baseflow divided by total streamflow ; percent watershed (WS) and riparian (Rip) land co ver composed of deciduous forest, evergreen forest, mixed forest, and grassland; and projected 2056 water temperature (FutTemp). The Trout column was calculated by adding two scores across species (i.e., Brook Trout, Brown Trout, Rainbow Trout): presence/a bsence (score = 1/0) and relative abundance (score = 1 [low], 2 [medium], 3 [high]) as defined by Wills et al. (2015). R. denotes river and Crk. denotes Peninsula, and Upp er Peninsula . Stream Overall Temp Trout GW WS Rip FutTemp Au Sable R. 1 14.5 (9) 12 (1) 0.72 (6) 35.8 (25) 76.4 (10) 14.9 (9) Bark R. 47 19.9 (45) 2 (41) 0.60 (27) 32.0 (30) 31.3 (39) 21.2 (43) Bear Crk. 32 15.8 (28) 4 (17) 0.65 (8) 23.3 (36) 9.4 (49) 18.2 (31) Black R. 27 15.1 (21) 4 (17) 0.55 (35) 7.0 (48) 68.3 (17) 16 (20) Boardman R. 10 15.0 (20) 8 (6) 0.63 (17) 39.2 (20) 60.1 (20) 15.5 (18) Brule R. 45 19.7 (42) 3 (34) 0.52 (38) 32.1 (29) 58.5 (21) 21.1 (40) Bryan Crk. 25 16.7 (33) 2 (41) 0.60 (27) 58.4 (10) 89.2 (3) 19.4 (36) Canada Crk. 49 20.0 (47) 2 (41) 0.55 (35) 16.5 (40) 55.8 (23) 21.9 (47) Carlton Crk. 16 14.0 (4) 3 (34) 0.73 (3) 17.8 (39) 35.1 (37) 14.6 (4) Carp R. 30 17.6 (35) 8 (6) 0.60 (27) 20.5 (37) 23.8 (44) 19.0 (32) Cedar Crk. 44 18.3 (37) 7 (8) 0.50 (44) 8.4 (47) 21.0 (45) 20.1 (37) Cedar R. (SLP) 48 16.3 (31) 3 (34) 0.60 (27) 39.8 (19) 46.3 (30) 23.2 (49) Cedar R. (UP) 37 21.8 (51) 2 (41) 0.60 (27) 12.5 (43) 42.9 (31) 16.9 (27) Chocolay R. 14 15.2 (22) 4 (17) 0.56 (34) 54.0 (12) 80.1 (8) 16.3 (24) Coldwater R. 52 22.6 (52) 4 (17) 0.45 (49) 18.2 (38) 5.8 (51) 24.4 (52) Davenport Crk. 8 11.5 (1) 3 (34) 0.57 (32) 96.3 (1) 41.6 (34) 12.4 (1) Duke Crk. 43 15.3 (23) 4 (17) 0.50 (44) 11.0 (45) 11.64 (48) 19.1 (33) E. Branch Fox R. 3 14.2 (7) 6 (11) 0.73 (3) 79.6 (5) 93.9 (1) 15.3 (12) Elm R. 35 17.5 (34) 4 (17) 0.45 (49) 66.8 (7) 89.89 (2) 24.3 (51) Escanaba R. 46 20.0 (47) 5 (14) 0.44 (52) 49.0 (13) 36.4 (35) 20.5 (38) Falls R. 33 19.9 (45) 4 (17) 0.52 (38) 83.4 (4) 67.7 (18) 21.4 (45) 118 Stream Overall Temp Trout GW WS Rip FutTemp Hersey R. 18 14.7 (14) 4 (17) 0.62 (20) 38.3 (22) 46.9 (28) 15.3 (12) Iron R. 42 18.6 (41) 2 (41) 0.52 (38) 30.6 (32) 57.2 (22) 19.2 (34) Little Indian R. 11 16.2 (30) 3 (34) 0.73 (3) 62.7 (9) 82.2 (7) 16.4 (25) Little Muskegon R. 24 14.7 (14) 4 (17) 0.62 (20) 28.3 (33) 33.4 (38) 15.3 (12) Manistee R. 5 13.6 (3) 12 (1) 0.65 (8) 2.9 (50) 68.5 (16) 14.2 (3) Mann Crk. 7 15.4 (24) 4 (17) 0.65 (8) 84.0 (3) 88.6 (4) 16.2 (21) Martin Crk. 20 14.9 (19) 4 (17) 0.61 (25) 24.7 (35) 54.8 (24) 15.5 (18) Menominee R. 50 20.5 (50) 2 (41) 0.57 (32) 37.2 (23) 25.4 (43) 21.7 (46) Miller Crk. 19 15.4 (24) 2 (41) 0.65 (8) 56.2 (11) 50.3 (27) 16.2 (21) Mosquito Crk. 29 14.7 (14) 2 (41) 0.62 (20) 32.2 (28) 30.4 (40) 15.3 (12) Muskegon R. 31 14.7 (14) 4 (17) 0.62 (20) 14.9 (42) 9.2 (50) 15.3 (12) Ogontz R. 15 14.1 (5) 2 (41) 0.74 (1) 38.8 (21) 28.0 (41) 14.7 (6) Paint R. 41 19.7 (42) 3 (34) 0.52 (38) 32.7 (26) 70.2 (14) 21.1 (40) Pere Marquette R. 22 16.6 (32) 4 (17) 0.61 (25) 40.2 (18) 69.3 (15) 17.4 (30) Pigeon R. 12 14.5 (9) 7 (8) 0.65 (8) 6.5 (49) 53.0 (26) 14.6 (4) Pine R. (NLP) 17 16.0 (29) 11 (4) 0.65 (8) 32.3 (27) 27.8 (42) 16.9 (39) Pine R. (SLP) 38 18.4 (38) 3 (34) 0.49 (48) 30.8 (31) 41.7 (33) 20.7 (27) Prairie Crk. 51 18.4 (38) 2 (41) 0.50 (44) 0.0 (51) 2.5 (52) 22 (48) Rapid R. 4 14.5 (9) 11 (4) 0.63 (17) 26.5 (34) 86.3 (5) 14.9 (9) Rogue R. 36 18.4 (38) 7 (8) 0.50 (44) 43.0 (16) 35.5 (36) 19.2 (34) S. Branch Pine R. 21 14.6 (12) 4 (17) 0.45 (49) 95.1 (2) 53.2 (12) 17.3 (29) Salmon Trout R. 23 14.6 (12) 2 (41) 0.65 (8) 48.5 (14) 46.6 (25) 16.2 (21) Silver Crk. 9 15.4 (24) 4 (17) 0.72 (6) 36.9 (24) 75.2 (29) 14.9 (9) St. Joseph R. 28 15.6 (27) 6 (11) 0.63 (17) 11.6 (44) 13.1 (47) 16.4 (25) Sturgeon R. 6 14.1 (5) 6 (11) 0.74 (1) 42.9 (17) 74.0 (13) 14.7 (6) Tahquamenon R. 39 20.0 (47) 2 (41) 0.55 (35) 65.1 (8) 78.9 (9) 23.6 (50) Tamarack Crk. 26 14.7 (14) 5 (14) 0.62 (20) 16.1 (41) 19.0 (46) 15.3 (12) 119 Stream Overall Temp Trout GW WS Rip FutTemp W. Branch Maple R. 13 14.4 (8) 4 (17) 0.65 (8) 0.0 (51) 83.3 (6) 14.8 (8) W. Branch Sturgeon R. 2 13.3 (2) 12 (1) 0.65 (8) 69.3 (6) 42.6 (32) 13.4 (2) Yellow Dog R. 34 18.0 (36) 4 (17) 0.52 (38) 47.1 (15) 75.4 (11) 21.3 (44) 120 Rainbow Trout) were abundant in these systems. The Pine River (southern Lower Peninsula) and Rapid River (northern Lower Peninsula) were both ranked fourth, and the Boardman River (northern Lower Peninsula) and Carp River (eastern Upper Peninsula) were ran ked sixth with respect to trout population characteristics (Table 3.2, Figure 3.1). As such, the three trout species were relatively abundant in these streams compared to the 45 lower - ranked systems, likely due to the presence of thermally favorable condit ions (e.g., groundwater, watershed/riparian land cover). Groundwater c ontribution Groundwater inputs are important for trout thermal habitat quality because they buffer against temperature extremes in the summer and winter. With a comparatively high BFI of 0.74 , the Ogontz and Sturgeon rivers (central Upper Peninsula) were the highest - rank ed streams by the SPT in terms of groundwater contribution to streamflow ( Table 3.2, Figure 3.1) . Five other trout streams (i.e., East Branch Fox and Little Indian rivers, central Upper Peninsula; Au Sable and South Branch Pine rivers, northern Lower Peninsula; Carlton Creek; Figure 3. 1) ranked among the top seven systems had BFI values > 0.70, meaning t hey are thermally buffered by groundwater inp ut and thus likely to maintain optimal or suitable thermal habitat conditions for Brook Trout, Brown Trout, and Rainbow Trout in a warming climate. Watershed and riparian land c over Streams with the greatest proportion of watershed land cover types tha t promote high - quality trout thermal habitats (i.e., deciduous forest, evergreen forest, mixed forest, grassland ) included Davenport Creek (96.3%) and the South Branch Pine River (95.1%), both of which are 121 located in relatively rural, undeveloped parts of Michigan (i.e., eastern Upper Peninsula, northeastern Lower Peninsula; Table 3.2, Figure 3.1) . Other highly - ranked streams for the watershed land cover criterion included Mann Creek (southern Lower Peninsula), the Falls River (western Upper Peninsula, Figure 3.1), and the East Branch Fox River, each with 79.6 84.0% of their watersheds composed of land cover types associated with high - quality stream thermal ha bitats. Similar to watershed land cover, the highest - rank ed stream for the riparian land cover c riterion (i.e., East Branch Fox River; Table 3.2, Figure 3.4) was a rural , undeveloped central Upper Peninsula system with 93.9% of its riparian zone containing thermally favorable trout habitat types. The Elm River (western Upper Peninsula), Bryan Creek ( central Upper Peninsula) , and Mann Creek were also highly - ranked streams with 88.6 89.9% of their riparian zones comprised of thermally favorable habitat types . Considering all six criteria simultaneously, the highest - rank ed streams by the SPT (i.e., grea test importance for current and future trout management ) were the Au Sable, West Branch Sturgeon, East Branch Fox, Rapid, and Manistee rivers ( Table 3.2 , Figure 3.5). Discussion Uncertainty is ubiquitous in the decision - making processes of fisheries management as fisheries are ecologically and socioeconomically complex systems, yet decisions must be made continually to sustain fish communities and the human systems they support. P revious studies in the fledgling field of fisheries decision support ha ve focused primarily on marine fisheries, particularly how DSTs can help fisheries professionals balance economic and ecological objectives in fisheries management decision - making. For instance, Dichmont et al. (2013) used a DST to develop MPA closure strategies that balanced economic and ecological objectives for 122 FIGURE 3.4. Michigan stream rankings for a stream prioritization tool criterion about high - quality riparian land cover conditions. Streams were ranked according to the proportion of their riparian zones containing habitats that promote cool, thermally favorable conditions for trout (i.e., deciduous/evergreen/mixed forest, grassland), with higher rankings indicating greater coverage by thermally favorable habitats. The ten highest - ranked streams by the SPT with respect to riparian land cover conditions are included above. 123 FIGURE 3.5. Overall Michigan stream importance rankings considering all six stream prioritization tool (SPT) criteria (i.e., water temperature, trout relative abundance, groundwater contribution to stream flow, high - quality watershed land cover, high - quality riparian land cover, projected 2056 water temperature). Streams with greater scores for individ ual criteria received higher importance rankings. The ten highest - ranked streams with respect to all six SPT criteria are included above. 124 marine fisheries, similar to how Azadivar et al. (2009) and Stortini et al. (2015) used DSTs to identify management approaches that minimized mortality of marine fishes while maximizing socioeconomic benefits (e.g., food, revenue, employment). Despite the scarcity of DSTs in freshwater fisheries , the present study helps fill this knowledge gap by illustrating how colla boration among freshwater fisheries scientist s, managers, biologists, and policy makers generated an SPT that integrates diverse information for robust decision - making regarding stream trout thermal habitat management. By providing information on stream - specific thermal habitat conditions (e.g., current/future temperature, groundwater input, riparian habitat), the SPT enable s fisheries professionals to select streams and time periods for which resource allocation (e.g., money, personnel, equipment) will most effic iently and effectively promote resilience - based management objectives. In the present study, some of the SPT rankings supported extant knowledge of Michigan trout streams , whereas other s were unexpected and enhanced understanding of these s ystems. For instance, the Au Sable River was the highest - ranked stream with respect to all SPT criteria collectively (Figure 3.2) , which confirms the renowned highly regarded trout fishery, a National Wild and Sce nic River, and a Michig an Blue Ribbon Trout Stream (Zorn and Sendek 2001 ; Canale and Chapra 2016). Similarly, the Manistee fifth - highest ranking substantiated the fact that this stream currently supports a productive, highly valued trout fishery (Tyler and Rutherford 200 7 ; Danhoff et al. 2017) . Sixty percent of the ten highest - rank ed trout streams were located in the Northern Lower Peninsula (NLP) of Michiga n, where geological, climatic, hydrological, and thermal conditions are known to provide cold, well - oxygenated strea ms for productive trout fisheries (T. Zorn, MDNR, personal communication). However, certain socioeconomically important systems currently 125 valuable for angling (e.g., Muskegon River, Pere Marquette River) were projected to have relatively poor thermal habitat quality in a changing climate. Moreover, streams without top - tier fisheries (e.g., Rapid River, Davenport Creek) were predicted to have excellent habitat quality for the trout populations studied. These results indicate that the Rapid River, Davenp ort Creek, and other highly - ranked, lesser - known systems merit management efforts to maintain and enhance trout populations and their thermal habitats. In addition, these findings illustrate how the SPT revealed surprising, management - relevant findings and represents a flexible instrument for fisheries professionals as they make decisions for resilience - based salmonid management in a changing climate. An important consideration in designing the SPT was the primacy of providing fisheries professionals with integrative information (e.g., stream temperatures, groundwater, land cover, trout population characteristics), rather than prescribing decisions for them. After all, s tream - specific ecological and socioeconomic circumstances will ultimately dictate how fi sheries profes sionals use the information the SPT supplies , in accordance with the eight steps of the fisheries management process (Taylor et al. 1995). For instance, high - importance streams (as ranked herein) may be high - priority when socio - political circumstances favor trophy fisheries management, threatened/ endangered species conservation, or other outcomes that require active, on - the - ground management approaches. However, high - importance streams according to our rankings may be low - priority when managers deem it most important to rehabilitate lower - ranked systems that are unlikely to sustain themselves without management intervention. The intention of co - producing a n SPT with fisheries professionals was to ensure that they receiv ed the information they deemed most important via an integrative tool to inform not execute decision - making. By supporting, rather than making, stream fisheries man agement decisions, the 126 SPT is an adaptable instrument for balancing ecological and socio economic objectives given changes in the environment and human systems over time. The SPT was also timely as it was concurrent with development of statewide inland trout management plan by the MDNR ( Carlson and Zorn 2018 ; Zorn et al. 2018 ). At a time when ecological and social data on trout fisheries are being assimilated and important policy and management decisions are imminent , it is advantageous for Michigan fisheries professionals to have a n SPT to assist them in synthesizing multiple management priori ties for efficient, effective decision - making. As a mechanism for enhancing trout management in a changing climate, the SPT described in this study enables fisheries professionals to foreca st future thermal habitat conditions and proactively manage trout populations, thereby promoting a resilience approach to stream trout management. Resilience - based fisheries management recognizes the importance of maintaining or restoring the capacity of habitats, populations, communities, and ecosystems to resist and re cover from environmental disturbances, including those associated with a changing climate ( Waldman et al. 2016 ; Carlson et al. 2017c ). Today, warming air and water temperatures are impacting fisheries from the individual to ecosystem levels (Paukert et al. 2016) and opportunities (e.g., commercial, recreational, subsistence) and livelihoods via changes in fish abundance and distribution (Hunt et al. 2016). Given that the natural and human components of fisheries systems are changing as the climate warms, it is important to manage fisheries for resilience . Because fisheries affect and are affected by extant b iological, social, economic, and political conditions, r esilience - based management requires collaboration among div erse fisheries and land management professional s and stakeholders and consideration of ecological and socioeconomic information to facilitate robust , integrated management strategies (Carlson et al. 2017c ). As illustrated herein , the SPT is an 127 instrument f or the development and assessment of resilience - based salmonid management programs because it arose from science - management collaboration and integrate s information about the structure and function of fisheries ecosystems and their interactions with human systems, thereby facilitating social - ecological resilience. Through a case study involving coldwater stream trout management , this investigation advances previous re search on fisheries decision support by demonstrating the utility of a n SPT for Michigan trout management in a changing climate. The SPT developed herein enables fisheries professionals to predict the effects of climate change i n Michigan trout streams (e. g., magnitude and spatiotemporal distribution of warming), evaluat e t he need for and appropriateness of fisheries ma nagement actions (e.g., thermal habitat management, f ish stocking), and prioritize future habitat manage ment and rehabilitation activities i n multiple watersheds. Despite the value of the SPT (and, more broadly, DSTs) for informing fisheries management , decision support is an emerging subdiscipline in the fisheries profession and thus requires further research to broaden and deepen its contrib ution to fisheries resilience and sustainability. For instance, most fisheries DSTs (Azadivar et al. 2009; Dichmon t et al. 2013 ; Stortini et al. 2015 ; Bitunjac et al. 2016 ) have been applied to inform decision - making in marine c ommercial systems rather tha n freshwater or recreational fisheries (but see Canale and Chapra 2016). Hence, there is a pressing need to develop DSTs applicable to freshwater fisheries in both commercial and recreational contexts and usable by diverse stakeholder groups (e.g., state a nd federal fisheries management agencies, non - governmental organizations, anglers, general public) to enhance fisheries sustainability. R esearch at broader spatial and longer temporal scales is necessary to advance the science and practice of freshwater fisheries decision support . Although we focused our research 128 socioeconomic value, we conducted this study with the understanding that our approach for SPT develo pment and implementation could be readily applied to other fisheries outside Michigan. Indeed, there are numerous opportunities for fisheries professionals to apply our methods to enhance the efficiency and effectiveness of fisheries management for differe nt species and ecosystems (e.g., lakes, rivers, wetlands). Climate change adaptation formed the conceptual underpinning for the present study, but climate change is certainly not the only stressor affecting fisheries worldwide. There is a need for DSTs tha t inform fisheries decision - making amid additional ecosystems threats (e.g., invasive species, hydropower expansion), the development of which will require co - production among scientist s, managers, biologists, policy makers, and other fisheries stakeholder s. Finally, continuing to engage the public in the process of DST development will make fisheries conservation a more collaborative, socially acceptable endeavor (Irv in and Stansbury 2004; Azadivar et al. 2009 ; NRC 2010; Carlson et al. 2 017c ) that promotes resilience - based fisheries fisheries resources. Acknowledgements We thank committee members of the Science to Action Fellowship (S. Carter, D. Infante, A. Lynch, C. Paukert, S. Winterstein), funded by Michigan State University and the USGS National Climate Adaptation Science Center, for their guidance and support as the first author completed his fellowship experience. We thank members of the National Climate Adaptation Science Center (E. Argo, J. Cushing, K. Donahue, E. Fort, N. Hartke - K. Malpeli, M. Matty, K. Maxwell, E. Minder, B. Myers, H. Padgett, C. Romulo, M. 129 Rubenstein, L. Thompson, J. Thornley, S. Weiskopf) for their warm collegiality as the first author spent time in residence at headquarters in Reston, Virginia. We acknowledge other USGS professionals (K. Jenni, C. Thatcher) for t heir helpful insights regarding development of the stream prioritization tool (SPT) described herein. 130 APPENDIX 131 Appendix 3 .1 . Survey instrument sent to Michigan fisheries professionals. MICHIGAN TROUT STREAM MANAGEMENT IN A CHANGING CLIMATE A Survey of Michigan Fisheries Professionals You are being asked to take part in a research study on stream trout management in Michigan given the potential for a changing climate regime. The survey is designed to evaluate the opinions an d perspectives of Michigan Department of Natural Resources (MDNR) fisheries professionals regarding management of inland stream trout (i.e., brook trout, brown trout, rainbow trout) in Michigan given the threat of a warming climate. Results from this sur vey will be combined with results from ongoing stream temperature modeling performed by the survey authors to project the effects of a changing climate regime on growth and survival of brook trout, brown trout, and rainbow trout in Michigan. Survey results will also be integrated with those from the 2015 MDNR Michigan Inland Trout Angler Survey. The ultimate goal of this project is to design a user - friendly, map - based decision - support tool in collaboration with the United States Geological Survey that assist s Michigan fisheries professionals in planning trout stream management programs that promote thermally resilient streams and productive stream trout populatio ns. Decision - support tools aid decision making by systematically incorporating information, accounting for uncertainties, and facilitating evaluation of tradeoffs between alternatives. In the 2013 - 2017 MDNR Fish eries Division Strategic Plan, Goal Four (Obj ective Two) expresses the need for decision - support tools and . By completing this survey, you are contributing to an important task of enhancing coldwater trout stream management amidst a changing climate regime . If you choose to participate, you will fill out a voluntary, anonymous survey on your perceptions of current and future trout stream management, including the potential use of decision - support tools for management decision - making. Your answers will be confidential. Your participation in the project is completely anonymous, voluntary, and uncompensated. Your participation will assist with the development of a useful decision - support tool to info rm stream trout management in a changing climate . There is no penalty or loss of benefits if you chose not to participate. You can skip any questions or withdraw at any time. If you have any questions about the research study, please contact Andrew Carlson (carls422@msu.edu) or Bill Taylor ( taylorw@msu.edu ). If you have any questions or concerns about your role and rights as a research participant, would like to obtain information or offer input, or would like to register a complaint about this resea rch study, you may contact, anonymously if you wish, the Michigan State University Human Research Protection Program at PHONE: 517 - 355 - 2180, FAX: 517 - 432 - 4503, EMAIL: irb@msu .edu, or MAIL: 207 Olds Hall, MSU, East Lansing, MI 48824. Please ask any question s you may have before agreeing to participate in this study. Thank you for your contribution to this important study. STATEMENT OF CONSENT : I have read the above information and have received answers to any questions I asked. I consent to take part in this study. Your Signature ___________________________________ Date ________________________ Your Name (printed) ____________________________________________________________ 132 Appendix 3.1 MICHIGAN TROUT STREAM MANAGEMENT IN A CHANGING CLIMATE A Survey of Michigan Fisheries Professionals Please complete the survey by checking the appropria te box or filling in the blank. Thank you for your time and effort in completing th e survey. Sincerely, Andrew Carlson Ph.D. Student Michigan State University Please list your agency name. ___________________________________________________ Please check the Michigan peninsula and region you work in (Upper, Northern Lower, Southern Lower ) . ____ Upper ____ Northern Lower ____ Sout hern Lower Please list the Michigan fisheries management unit you work in (Western Lake Superior, Eastern Lake Superior, Northern Lake Michigan, Northern Lake Huron, Central Lake Michigan, Southern Lake Huron, Southern Lake Michigan, Lake Erie). _____________________________ _____________________________________________________________________________ Please list the M DNR office/station/service center you work in (e.g., Escanaba , Cadillac , Traverse City, Harrietta ) _____________________________________________________________________________ 133 PART A. Stream trout management A1. Does your office manage stream trout fisheries (i.e., brook trout, brown trout, or rainbow trout)? ____Yes (go A2) ____No (skip to Part B ) A2. How important do you think the following criteria for prioritizing streams for trout management in your region? Please specify other criteria if they are not listed here. Criteria Very important Important So mewhat Important Unimportant Not sure Abundance of brook trout Abundance of brown trout Abundance of rainbow trout Angler use Proximity to office In - stream habitat quality Riparian habitat quality Groundwater input Stream is currently thermally optimal for trout Other: 134 A3. Based on the criteria you specified in question A2, p lease list the five trout streams counties that you think are most important to manage in your district? Please include the name of the county in which each stream is located. 1)_______________________________________________________ 2 )_ ______________________________________________________ 3 )_______________________________________________________ 4 )_______________________________________________________ 5 )_______________________________________________________ Personnel resources A4. Ho w many employees in your office are responsible for managing stream fisheries resources (all fish species included , not just trout )? ____________ A5. Of these fisheries employees, how many manage stream trout fisheries as a full - time or part - time job responsibility? _______Full - time _______Part - time activities? ____Yes ____No please briefly explain or list how personnel limitations affect stream trout management. Financial resources A8. Approximately what percent management? ___________% 135 A9. Of this budget (stream trout budget), estimate the percent that your office expends for the following activities. 0 - 5% 6 - 10% 11 - 25% 26 - 50% >50 % know a) Management (survey stream fish populations, implement management plans, maintain property, etc. ) b) Research (agency stream studies, university - funded studies , NGO - funded studies, etc. ) c) Program services (financial management, personnel management, data processing , etc. ) d) Education (publications, exhibits, sport shows, talks, media programs, etc. ) e) Hatcheries (fish production and stocking, etc. ) f) Land acquisition (stream access areas for the public, including parking and ramps, etc. ) g) Physical habitat improvem ent (in - stream structures, riparian planting , etc. ) h) Consultative services (cooperative efforts with other agencies, NGOs, public, etc.) activities? ____Yes ____No briefly explain or list how financial limitations affect stream trout management. Time stream trout fisheries for: Sum to 100% a) Management (surveys, plans, maintenance, etc.) __________% b) Research (with agency, universities, NGOs, etc.) __________% i) Program services (financial management, personnel management, data processing , etc. ) __________% c) Education (publications, exhibits, talks, etc.) __________% d) Hatcheries (fish p roduction and stocking, etc.) __________% e) Land acquisition (stream access areas for the public) __________% f) Habitat improvement (in - stream structures, riparian planting , etc.) __________% Total 100% f s a d f x x x x x x g g x z x 136 A13. Do ____Yes ____No briefly explain or list how time limitations affect stream trout management. A15. Do limitations in e q uipment /f acilities ( e.g., vehicles, stream sampling gear, hatcheries ) ____Yes ____No briefly explain or list how limitations in e qu ipment /f acilities affect stream trout management. PART B. Stream temperature monitoring B1. Does your office monitor trout stream tem peratures? ____ Yes ____No to question B1, how often does office monitor trout stream tem peratures? ____Seasonally ____Monthly ____Weekly ____Daily ____Hourly There are two broad approaches that fisheries professionals generally use to describe the relationship between a ir tem perature and stream temperature : stream - specific models and generalized models. Stream - specific temperature models account for the unique factors that whereas generalized models are region - specific in representing the thermal regimes of all streams in a particula r region. Stream - specific models are generally more accurate in predicting exact temperatures for individual streams as compared to generalized models, but generalized models are more efficient as they require fewer resources (e.g., money, time, personnel ) to develop. B3. Does your office use stream - specific models, generalized models, both , or neither ? ____ Stream - specific models ____ Generalized models ____ Both ____ Neither B4. Given your office trout management and stream temperature monitoring/modeling, which model type s do you think would be most useful for your office (even ? ____ Stream - specific models (always) ____ Generalized models (always) ____ Stream - specific models (in certain situations) 137 ____ Generalized models (in certain situations) ____ Models are equally useful ____ Models are not useful - briefly describe those situations. PART C. Stream thermal habitat management C1. Does your office manage stream thermal habitats (via groundwater conservation, riparian or watershed protection/rehabilitation, etc.), including collabo rative management with other organizations? ____Yes ( go to C 2) ____No (skip to Part D ) C2. W hat thermal habitat management strategies for stream trout does your office currently use or plan to use in the future? Strategy Current Future Groundwater conservation Riparian habitat protection Riparian habitat rehabilitation Watershed habitat protection Watershed habitat rehabilitation Other: 138 C3. Imagine that your office is developing a stream thermal habitat management program. How important do you believe each of the following factors would be? Include a number for each factor indicating its RELATIVE PRIORITY, where larger numbers indicate h igher priority. The SUM of numbers for the eight factors should be 100 . Sum to 100 a) Current water temperature ____ b) Future/projected water temperature ____ c) Current groundwater contribution ____ d) Future/projected groundwater contribution ____ e) Current s urface water contribution ____ f) Future/projected s urface water contribution ____ g) Trout population characteristics ( presence/absence, relativ e abundance, size structure, recreational importance , etc.) ____ h) Riparian attributes (e.g., species composition, shading) ____ i) Watershed land cover (e.g., grassland, forest, agricultural) ____ j) Presence/absence of temperature gauges ____ k) Resource availability (e.g., personnel, finances, time, equipment/facilities) ____ Total = 100 PART D. Decision - support tools Results from this survey will be used to develop a decision - support tool to assist Michigan fisheries professionals in planning management programs that promote thermally resilient streams and trout populations . Decision - support tools (DSTs) enable people and organizations to identify policy and management options for achieving their objectives a midst uncertainty. For example, fisheries managers could use a DST to prioritize streams for trout population and thermal habitat management in a changing climate. Imagine a DST for stream trout management in a changing climate . This DST combines data on stream temperature and trout populations with information on agency resource availability and trout anglers (e.g., values, opinions, demographics). D1. How useful do you believe the DST described above would be for stream trout management? Not useful Not very useful Moderately useful Useful Very useful know/no opinion 139 D2. How can this DST be made as user - friendly as possible? Please specify other features if _____Color coding _____Symbol coding _____Maps included _____Delivered via the Internet _____Delivered via a GIS program _____ Written case studies for streams _____Background information on DSTs _____Other: Please share additional comments you have about this survey related to trout stream management or decision - support tools in Michigan. Thank you for participating! Please return this survey by email to: Andrew Carlson Michigan State University Email: carls422@msu.edu Phone: 651 - 280 - 7013 140 LITERATURE CITED 141 LITERATURE CITED Arlinghaus, R., J. Al ós, B. Beardmore, K. Daedlow, M. Dorow, M. Fujitani, D. H ü hn, W. Haider , L. M. Hunt, B. M. Johnson, F. Johnston, T. Klefoth, S. Matsumura, C. Monk, T. Pagel, J. R. Post, T. Rapp , C. Riepe, H. Ward , and C. Wolter. 2017. Understanding and m anaging f reshwater r ecreational f isheries as complex a daptive s ocial - e cological s ystems. Reviews in Fisheries Science & Aquaculture 25:1 41. Azadivar F., T. Truong, and Y. Jiao. 2009. using operations research and systems science approach. Expert System s with Applications 36: 2971 2978. Baldwin, N. S. 1957. Food consumption and growth of br ook trout at different temperature s. Transactions of the American Fisheries Society 86: 323 328. Bitunjac, I., N. Jajac, and I. Katavic. 2016. Decision s upport to s ustainable Management of Bottom Trawl Fleet. Sustainability 8 :1 2 3. Canale, R. P., and S. C. Chapra. Decision support models for assessing the impact of aquaculture on river water quality. Journal of Environmental Engineering 142(10):1 1. Carbon Dioxide Information Analysis Center ( CDIAC). 2016. Unit ed States Departmen t of Energy Historical Climate Network . United States Department of Energy, Oak Ridge National Laboratory. http://cdiac.ornl.gov/ . Accessed 26 June 2018. Carlson, A. K., and T. G. Zorn. 2018. Values, opinions, and behavior of inland trout anglers in Michigan . Fisheries Report 29. Michigan Department of Natural Resources , Lansing, Michigan, USA. Carlson A. K., W. E. French, B. Vondracek, L. C. Ferrington, Jr., J. E. Mazack, and J. L. Cochran - Biederman. 2016 . Brown trout growth in Minnesota streams as related to landscape and local factors. Jo urnal of Freshwater Ecology 31 :421 429. Carlson, A. K., W. W. Taylor, J. Liu, and I. Orlic. 2017a. The telecoupling framework: an integrative tool for enhancing fisheries management. Fisheries 42:395 397. Carlson, A. K., W. W. Taylor, K. M. Hartikainen, D. M. Infante, T. D. Beard Jr., and A. J. Lynch. 2017b. Comparing stream - specific to generalized temperature models to guide salmonid management in a changing climate. Reviews in Fish Biology and Fisheries 2 7:443 462. Carlson, A. K., W. W. Taylor, K. M. Schlee, T. G. Zorn, and D. M. Infante. 2017c . Projected impacts of climate change on stream salmonids with implications for resilience - based management. Ecology of Freshwater Fish 26:190 204. Danho ff, B. M., C. J. Huckins, N. A. Auer, C. W. Goble, S. A. Ogren, and J. M. Holtgren. 2017. Abiotic habitat assessment for Arctic Grayling in a portion of the Big Manistee River, Michigan. Transactions of the American Fisheries Society 146:645 662. 142 Dichmont, C. M., N. Ellis, R. H. Bustamante, R. Deng, S. Tickell, R. Pascual, H. Lozano - Montes, and S. Griffiths. 2013. Evaluating marine spatial closures with conflicting fisheries and conservation objectives. Journal of Applied Ecology 50:1060 1070 . Drake, M. T., and W. W. Taylor. 1996. Influence of spring and summer water temperature on brook charr, Salvelinus fontinalis , growth and age structu re in the Ford River, Michigan. Environ mental Biol ogy of Fish es 45 :41 51. . Analysis of groundwater regime on the basis of streamflow hydrograph. Facta U niversitatis , Series: Architecture and Civil Engineering 10 :301 314 . Elliott, J. M., and M. A. Hurley. 2000. Daily energy intake and growth of piscivorous brown trout, Salmo tru tta . Freshwater Biology 44: 237 245. Fréon, P., J. C. Sueiro, F. Iriarte, O. F. Miro Evar, Y. Landa, J. Mittaine, and M. Bouchon. 2014. Harvesting for food versus feed: a review of Peruvian fisheries in a global context. Reviews in Fish Biology and Fisheries 24:381 398. Fry, F. E . J., J. S. Hart, and K. F. Walker . 1946. Lethal temperature relations for a sample of y oung speckled trout, Salvelinus fontinalis , Vol. 54. The University of Toronto Press, Toronto, Ontario, Canada. Godby, N. A., Jr., E. S . Rutherford, and D. M. Mason. 2007. Diet, feeding rate, growth, mortality, and p roduction of juvenile steelhead in a Lake Michiga n tributary. North American Journal of Fisheries Management 27 :578 592. H ansen, G. J. A., J. W. Gaeta , J. F. Hansen, S. R. Car penter . 2015. Learning to manage an d managing to learn: sustaining freshwater recreational fishe ries in a changing environment. Fisheries 40:5 6 64. Hay, J ., J. Hayes, and R. Young. 2006. Water quality guidelines to protect trout fishery values. Prepared for Horizons Regional Council. Cawthorn Report No. 1025. Cawthorn Institute, Nelson, New Zealand. Hayes, D. B . , W. W. Taylor , M. Dr ake, S. Marod, and G. Whelan. 1998. The value of headwaters to brook trout ( Salvelinus fontinalis ) in the Ford Riv e r, Michigan, USA. Pages 75 185 in M. J. Haigh, J. Krecek, G. S. Rajwar, and M. P. Kilmartin, editor. He adwaters: water resources and so il conservation. Oxford and IBH Publishing Co., New Delhi , India. Holling, C. S. 1973. Resilience and stability of ecological systems. Annual Review of E cology and Systematics 4: 1 23. Homer, C. G., J. A. Dewitz , L. Yang , S. Jin , P. Danielson , G. Xian , J. Coulston , N. D. Herold , J. D. Wickham , and K. Megown. 2015. Completion of the 2011 National Land Cover Database for the conterminous United States - Representing a decade of land cover change information. Photogrammetric Eng ineering and Remote Sensing 81 :345 354 . 143 Hunt, L. M., E. P. Fenichel, D. C. Fulton, R. Mendelsohn, J. W. Smith, T. D. Tunney, A. J. Lynch, C. P. Pauke rt, and J. E. Whitney. 2016. Identifying alternate pathways for climate change to impact inland recreational fishers. Fisheries 41:362 372. Irvin, R. A., and J. Stansbury. 2004. Citizen participation in decision making: is it worth the effort? Public Administration Review 64:55 65. Krider L . A . , J. A. Magner , J. Perry , B. Vondracek , and L. C. Ferrington, Jr . 2013. Air - water tempera ture relationships in the trout - sandstone landscape. Journal of the American Water Resources Association 49 :896 907. Lessard, J. L., and D. B. 2003. Effects of elevated water temperatures on fish and macroinvertebrate communities below small dams. River Research and Applications 19:721 732. Liu, J. G., T. Dietz, S. R. Carpenter, C. Folke, M. Alberti, C. L. Redman, S. H. Schneider, E. Ostrom, A. N. Pell, J. Lubchenco, W. W. Taylor, Z. Ouyang, P. Deadman, T. Kratz, and W. Provencher. 2007 . Coupled human and natural systems. Ambio 36:639 649. Lynch, A. J., E. Varela - Acevedo, and W. W. Taylor. 2015. The need for decision - support tools for a changing climate: application to inland fisheries management. Fisheries Management and Ecology 22 : 14 24. Lyons, J. 1996. Patterns in the species composition of fish assemblages in Wisconsin streams. Environmental Biology of Fishes 45:329 - 341. Lyons, J. , T. Zorn , J. Stewart, P. Seelbach, K. Wehrly, and L. Wang. 2009. Defining and chara cterizing coolwater stre ams and their fish assemblages in Michigan and Wis consin, USA. North American Journal of Fisheries Management 2 9:1130 1151. Lyons, J. , J. S. Stewart, and M. Mitro. 201 0. Predicted effects of climate warming on the distribution of 50 stream fishes in Wisco nsin, USA. Journal of Fish Biology 77 :1867 1898. Maurer, E. P., L. Brekke , T. Pruitt, and P. B. Duffy . 2007. Fine resolution climate proje ctions enhance regional climate change impact studies. Eos 88: 504. McKer gow, L . , S. Parkyn , R. Collins , and P. Pattinson. 2005. Small headwater streams of the Auckland Region. Volume 2: Hydrology and water quality. Auckland Regional Council Technical Publication 312: 1 67 . Menberg, K . , P. Blum , B. L. Kuryly k, and P. Bayer. 2014. Observed groundwater temperat ure response to recent climate change. Hydrology and Earth System Sciences 18:4453 4466. Michigan Department of Natural Resources [MDNR]. 2013. Charting the course: Fisheries al Resources, Ann Arbor, Michigan, USA. National Research Council [NRC]. 2010. Informing an effective response to climate change. The National Academies Press, Washington, D.C., USA. 144 Neff B. D . , S. M. Day , A. R. Piggott , and L. M. Fuller. 2005. Base flow in the Great Lakes basin. U. S. Geological Survey Scientific Investigations Report 2005 5217 . U. S. Geological Survey, Reston, Virginia, USA. Ostroff, A., D. Wieferich, A. Cooper, D. Infante, and USGS Aquatic GAP Program. 2 013. 2012 National Anthropogenic Barrier Dataset (NABD). U.S. Geological Survey Aquatic GAP Program: Denver, Colorado, USA. Paukert, C. P., B. A. Glazer, G. J. A. Hansen, B. J. Irwin, P. C. Jacobson, J. L. Kershner, B. J. Shuter, J. E. Whitney, and A. J. L ynch. 2016. Adapting inland fisheries management to a changing climate. Fisheries 41:374 384. Pilgrim, J . M . , X. Fang , and H. G. Stefan. 1998. Stream temperature correlations with air temperat ures in Minnesota: implications for climate w arming. Journal of the American Water Resources Association 34:1109 1121. Raleigh, R. F. 1982 a . Habitat suitability index models: brook trout. U.S. Fish and Wildlife Service Biological Report 82 (10.24). U.S. Fish and Wildlife Service , Fort Collins, Colorado, USA. Raleigh, R. F. 1982 b . Habitat suitability index models: rainbow trout. U.S. Fish and Wildlife Service Biological Report 82 (10.60). U.S. Fish and Wildlife Service , Fort Collins, Colorado, USA. Raleigh, R. F., L. D. Zuckerman, and P. C. Nelson. 1986. Habitat suitabi lity index models and instream flow suitability curves: brown trout. U.S. Fish and Wildlife Service Biological Report 82(10.124). U.S. Fish and Wildlife Service , Fort Collins, Colorado, USA. Rohweder, J., S. Vacek, S. M. Crimmins, and W. E. Thogmartin. 201 5 a . A case study of assigning conservation value to dispersed habitat units for conservation planning. Journal of Conservation Planning 11:13 27. Rohweder, J., S. Vacek, W. E. Thogmartin, and S. M. Crimmins. 2015 b . Management unit prioritization tools. U.S . Fish and Wildlife Service, La Crosse, Wisconsin, USA. Siitari , K . J . , W. W. Taylor , S. A. C. N elson, and K. E. Weaver. 2011. The influence of land cover composition and groundwater on thermal habitat availabil ity for brook charr ( Salvelinus fontinalis) populations i n the United States of America. Ecol ogy of Freshwater Fish 20:431 437 . Stefan H . G . , from air temperature. Water Resources Bulletin of the American Water Resources Association 29:27 4 5 . - support tool to facilitate discussion of no - take boundaries for Marine Protected Areas during stakeholder consultation processes. Journal for Nature Conservation 23:45 52. Taylor, W. W., C. P. Ferreri , F. L. Poston, and J. M. Robertson. 1995. Educating fisheries professionals using a watershed approach to emphasize the ecosystem paradigm. Fisheries 20:6 8. 145 Tyler, J. A., and E. S. Rutherford. 2007. River restoration effects on steelhead populations in the Manistee River, Michigan: analysis using an individual - based model. Transactions of the American Fisheries Society 136 :1654 1673. USFWS (United States Fish an d Wildlife Service). 2011. 2011 National survey of fishing, h unting, and wild life - associated recreation. U.S. Department of the Interior, U.S. Fish and Wildli fe Service, and U.S. Department of Commerce, U.S. Census Bureau , Washing ton, D.C., USA. Vondracek B . , K. L. Blann , C. B. Cox , J. F. Nerbonne , K. G. Mumford , B. A. Nerbonne , L. A. Sovel l, and J. K. H. Zimmerman. 2005. Land use, spatial scale, and stream systems: lessons from an agricultura l region. Environmental Management 36:775 791. Wahl, K. L., and T. L. Wahl. 1988. Effects of re gional ground water declines on streamf lows i n the Oklahoma Panhandle. Pages 239 249 i n M. Waterstone, and R. J. Burt , editor. Proceedings of the symposium on water - use data for water resources management. American W ater Resources Association, Tucson, Arizona, USA. Waldman J., K. A. Wilson, M. Mathe r, and N. P. Snyder. 2016. A resilience approach can improve anadromous fish restoration. Fisheries 41:116 126. Ward, H. G. M, M. S. Allen, E. V. Camp, N. Cole, L. M. Hunt, B. Matthia s, J. R. Post, K. Wilson, R. Arlinghaus. 2016. Understanding and m anaging s ocial e cological f eedbacks in s patially s tructured r ecreational f isheries: the overlooked behavioral dimension. Fisheries 41:524 535. Whitney, J. E., R. Al - Chokhachy, D. B. Bunnell, C. A. Caldwell, S. J. Cooke, E. J. Eliason, M. Rogers, A. J. Lynch, and C. P. Paukert. 2016. Physiological basis of climate change impacts on North American inland fishes. Fisheries 41:332 345. Wills, T. C., T. G. Zorn, D. B. Hayes, and K. E. Wehrly. 2015. Status and trends of Michigan stream resources, 2002 2007. Fisheries Re port 09. Michigan Department of Natural Resources, Ann Arbor, Michigan, USA. Zorn, T. G., and A. J. Nuhfer. 2007. Influences on brown trout and brook trout population dynamics in a Michigan river. Transactions of the Ame rican Fisheries Society 136:691 705 . Zorn, T. G., and Sendek, S. P. 2001. Au Sable River assessment. Fisheries Special Report 26. Michigan Department of Natural Resources, Ann Arbor, Michigan, USA. Zorn, T. G., P. W. Seelbach , and M. J. Wiley . 2011. Developing user - friendly habitat suitab ility tools from regional stream fish survey data. North American Journal of Fisheries Management 31:41 45. Zorn, T. G., P. W. Seelbach , and E. S. Ruther ford . 2012. A regional - scale habitat suitability model to assess the effects of flow reduction on fish assemblages in Michigan streams. Journal of the American Water Resources Association 48:871 895. Zorn T. G., T. A. Cwalinski, N. A. Godby, Jr., B. J. Gunderman, and M. A. Tonello. Management plan for inland trout in Michigan. Fisheries Report 30. Michigan Department of Natural Resources , Lansing, Michigan, USA. 146 CHAPTER 4: DEVELOPING PRECIPITATION - AND GROUNDWATER - CORRECTED STREAM TEMPERATURE MODELS TO IMPROVE TROUT MANAGEMENT IN A CHANGING CLIMATE Carlson, A. K., W . W . Taylor, D. M. Infante. In review . Developing precipitation - and groundwater - corrected stream temperature models to improve management of brook charr amid climate change . The content of this chapter is based on the manuscript cited above , which is currently in review in Hydrobiologia . The chapter reflects specifications (e.g., formatting ) of this journal and includes additional results and conclusion related to brown trout and rainbow trout. 147 Abstract Ensuring the ecological integrity of coldwater streams in a warming world require s understanding how water temperatures changes will affect the sustainability of coldwater fish populations such as brook trout ( Salvelinus fontinalis ), brown trout ( Salmo trutta ), and rainbow trout ( Oncorhynchus mykiss ). However, current models for predic ting stream temperature have common flaws, such as assuming spatially uniform (inaccurate) air - stream temperature relationships or requiring measurement of expensive hydrometeorological drivers (e.g., solar radiation, convection) in a manner impractical fo r fisheries management. Hence, we developed an accurate, cost - effective, management - relevant approach for modeling effects of changes in air temperature, precipitation, and groundwater inputs on coldwater stream temperatures and trout survival and growth i n Michigan, USA. Precipitation - and groundwater - corrected regressions were more accurate than air - stream temperature models for predicting stream temperatures. Projected stream warming intensified in proportion to simulated air temperature warming and was most extreme in surface runoff - dominated streams, given their limited groun dwater - driven thermal buffering. However, groundwater - dominated streams will not invariably provide coldwater habitats needed by trout if groundwater temperatures increase or groundwater inputs decline due to reduced precipitation. Amid resource limitation s, fisheries managers can use our stream temperature modeling approach to predict effects of climate change on trout survival and growth and take actions to facilitate their sustainability in riverine systems. KEYWORDS: brook trout; brown trout; climate change; groundwater; growth; precipitation; survival; rainbow trout 148 Introduction Streams provide important ecological goods and services to humanity (e.g., recreation; water for municipal, industrial, and agricultural use; Loomis et al., 2000), but they are highly vulnerable to certain stressors, including climate change (Woodward et al., 2010) and associated threats to fishes and their habitats (Hershkovitz et al., 2015; Kanno et al., 2015). Climate change has been projected to impact streams through nu merous mechanisms, including increased water temperatures (including groundwater) and alterations to hydrological regimes (e.g., more frequent heavy precipitation, reduced snowpack), resulting in changes in thermal and physical habitat for aquatic organism s (Woodward et al., 2010; Snyder et al., 2015). Stream temperature is a fundamental factor influencing the suitability and productivity of stream habitats for aquatic biota. Hence, projected increases in stream temperatures resulting from short - and long - t erm changes in climate (Thomas et al., 2015; Carlson et al., 2016 ) are cause for concern among fisheries professionals, policy makers, and allied stakeholders (e.g., non - governmental organizations, general public), particularly those charged with conservin g coldwater fishes. Stream fishes in the family Salmonidae (e.g., brook trout Salvelinus fontinalis , brown trout Salmo trutta , rainbow trout Oncorhynchus mykiss ) are adapted to coldwater and coolwater environments and have relatively low thermal tolerance thresholds (Raleigh 1982a,b; Raleigh et al. 1986), making them sensitive to induced stream temperature warming. In addition, stream trout are valuable from ecological, economic, recreational, cultural perspectives (Godby et al. 2007; Tyler & Rutherford 20 07). Hence, projecting the effects of climate change on stream trout population viability and productivity is necessary for developing management strategies that will conserve these species in a warming world. 149 Historically, regression models for predict ing stream temperature have included air temperature as a primary driver of water temperature because it is surrogate for solar radiation, the factor that most strongly influences stream temperature (Webb et al., 2008). Despite the importance of understand ing how projected changes in air temperature affect stream temperature (i.e., thermal sensitivity), these simple air - stream temperature models ignore other drivers including groundwater input, precipitation dynamics (e.g., magnitude, intensity), watershe d land cover, and riparian shading (Constantz, 1998; Ebersole et al. 2003) that have the potential to significantly affect stream thermal regimes. Until recently, stream temperature models have largely ignored variability in groundwater dynamics (e.g., m agnitude, temperature) among streams and stream reaches, thereby decreasing accuracy of thermal forecasting and efficacy of thermal habitat management in a changing climate, particularly in headwater areas where groundwater inputs tend to be relatively lar ge (Snyder et al., 2015). By accounting for stream - and reach - level heterogeneity in groundwater dynamics, groundwater - corrected stream temperature models should provide a more realistic, reliable method for evaluating stream temperature warming than simpl e air - stream temperature models. Ultimately, this should allow groundwater - corrected stream temperature models to inform thermal habitat management actions needed to facilitate trout population sustainability (e.g., forest canopy rehabilitation, riparian p rotection). As a buffer to daily and seasonal temperature alterations, groundwater generally causes stream temperature to be cooler in summer and warmer in winter than streams dominated by surface runoff, particularly in headwater reaches (Webb et al., 20 08). Thermal buffering is ecologically important because it may mitigate effects of climate change on coldwater fishes and their habitats. Despite the ecological significance of groundwater, its incorporation into stream 150 temperature models can be confounde d by the complexity of groundwater dynamics, especially heterogeneity in groundwater temperatures and input magnitudes among stream reaches (Snyder et al., 2015). Although stream heat budget models incorporate groundwater and other atmospheric, meteorologi cal, and hydrological variables to predict water temperature (Leach & Moore, 2011; Westhoff et al., 2011), they are expensive, data - intensive, and generally impractical for use in fisheries management (Dunham et al., 2005; Snyder et al., 2015). Groundwater - corrected stream temperature regressions were recently developed to inform brook trout management in the Eastern United States of America (USA; Snyder et al., 2015), but these models did not include other important thermal drivers (e.g., precipitation), n or was their applicability evaluated in other areas that have socio - ecologically valuable populations of brook trout, brown trout, rainbow trout, and other coldwater fishes (e.g., Midwestern USA). Developing a methodology to integrate groundwater and preci pitation dynamics into stream temperature modeling is important because it can promote stream management for thermal resilience using readily measureable temperature drivers. More broadly, this endeavor would support resilience - based management programs fo r coldwater streams that enhance the ability of these ecosystems and associated human systems to absorb disturbances (e.g., temperature changes) and yet retain their overall structure and function (Carlson et al., 2016; Paukert et al., 2016). Compared to groundwater, effects of precipitation on stream temperature are infrequently studied. Precipitation is rarely included as an explanatory variable in stream temperature models (Snyder et al., 2015), perhaps because potential processes through wh ich precipitation affects water temperature (e.g., changes in timing and magnitude of surface runoff delivered to channels, reduced relative influence of groundwater inputs on temperature, changes in turbidity) 151 are indirect and difficult to measure. As suc h, the effects of climate change on precipitation, and resultant effects on stream temperature, have not been widely studied. In the Great Lakes region, climate change is expected to increase the frequency and intensity of precipitation events, particularl y during winter and spring ( Cherkauer & Sinha, 2010; Hayhoe et al., 2014), with potential effects on stream temperature (e.g., precipitation increases the discharge and volume of water exposed to solar radiation, causing temperatures to decrease or rise at a slower rate; precipitation increases sediment erosion, water turbidity, and absorption of solar radiation, causing temperatures to rise; Merriam et al., 2017). However, the extent to which precipitation regimes in a changing climate will affect groundwa ter recharge and associated thermal buffering in coldwater streams, and the degree to which managers can influence these relationships via water and land use management practices to sustain coldwater fisheries, have not been thoroughly investigated in the Great Lakes region. The State of Michigan has a diversity of coldwater stream ecosystems that experience different air temperature patterns and hydrological regimes (i.e., groundwater/surface - runoff dominance) and currently support productive trout fishe ries that are recreationally and culturally renowned. Hence, Michigan was an ideal study area for addressing our goal: to develop an accurate, cost - effective, management - relevant approach for modeling effects of changes in air temperature, precipitation, a nd groundwater on coldwater stream temperatures and trout survival and growth to help fisheries professionals sustainably manage trout populations amid climate change. Our objectives were to: 1) create stream - specific regression models that account for the influence of air temperature, precipitation patterns, and groundwater input on coldwater stream temperatures in Michigan ; 2) compare precipitation - and groundwater - corrected models to simple air - stream temperature models in terms of accuracy (i.e., exactn ess of temperature 152 projection); and, 3) use precipitation - and groundwater - corrected models to predict effects of climate change on stream temperature and thermal habitat suitability for trout survival and growth out to the year 2056. Methods Study area T his study included coldwater streams (n = 15) containing trout populations located throughout the State of Michigan (Figure 4. 1). These streams were distributed across most of Michigan from north to south (46.41°N to 42.64°N) to encompass latitudinal variation in air temperatures and thus stream thermal regimes. In addition, these streams spanned a hydrological gradient from groundw ater to surface runoff dominance, which was evaluated according to base flow index, the proportion of streamflow represented by groundwater. Base flow index was calculated using a digital filter hydrograph separation method described by Neff et al. (2005). Streams were partitioned according to base flow index as: groundwater - dominated (base flow index > 0.60); surface - - and intermediate groundwater input (base flow index = 0.60; McKergow because they contained viable, productive populations of brook trout (n = 9 streams), brown trout (n = 11), and rainbow trout (n = 9). These coldwater species are threatened by climate change (Carlson et al., 2016; Merriam et al., 2017) and distributed throughout 31,000 km of streams in Michigan, making them suitable indicator species for predicting how warmer water temperatures will affect str eam fishes in groundwater - dominated and runoff - dominated systems. 153 FIGURE 4.1. Map of 15 trout streams used for water temperature modeling in Michigan. Streams and corresponding identification numbers are listed in Table 4.1. 154 Temperature and precipitat ion measurements Water temperature was measured hourly throughout July and August 2016 and 2017 in headwater portions of all 15 streams. These months were selected because they are generally the warmest and most thermally stressful for Michigan stream trout (Zorn et al., 2011) and would likely encompass the period during which predicted climatic changes would most strongly affect thermal habitat quality and quantity in this region. Moreover, headwater reaches were selected because they are typically the coolest, most thermally optimal habitats for trout during warm summer months in Michigan ( Hayes et al., 1998); if they become warmer, temperatures in downstream reaches will also generally increase. Water temperature was measured using HOBO Pro v2 data lo ggers that are accurate within 0.2°C and have a drift of <0.1°C every year (Onset Computer Corporation, 2009). Data loggers were installed in habitats of intermediate velocity and depth, and were shielded from debris and direct sunlight using white polyvin yl chloride (PVC) pipes. Water was allowed to flow into the pipes through a series of holes. Hourly water temperatures were used to calculate the mean daily stream temperature (MDST) as a 24 - h average. Hourly water temperatures were also measured in 2018 a nd are expressed as MDST values in Appendix 4.1. Hourly air temperatures and daily precipitation measurements were collected throughout the study period using the Michigan State University Enviro - weather Automated Weather Station Network (EAWSN, 2018) at temperature data were summarized as mean daily air temperatures (MDAT) for each stream reach. Likewise, precipitation measurements were summarized as cumulative daily precipitation for each reach. 155 Temper ature projections Three coupled climate models were used to project future (i.e., 2036, 2056) July and August air temperatures in each stream reach studied. These models included the Third Generation Coupled Global Climate Model (CGCM3, Canadian Centre for Climate Modelling and Analysis), the CM2 Global Coupled Climate Model (CM2, Geophysical Fluid Dynamics Laboratory at the National Oceanic and Atmospheric Administration), and the Hadley Centre National Weather Service). These models were selected because they differ in their thermal input parameters (e.g. solar radiation, trace gases, sulfate aerosols), thereby encompassing a range of climatic conditions that Michigan streams could experience in the future. However, all coupled climate models were based on the World Climate Research Programme (WCRP) Coupled Model Intercomparison Project phase 3 (CMIP3) multi - model dataset. The spatial resolution of climate models (~200 x 200 km) was downscaled to a level relevant for Michigan streams (12 x 12 km; Maurer et al., 2007) using the Bias - Correction Spatial Disaggregation approach. Projected air temperatures were supplied by the United States Forest Service Eastern Forest Environmental Threat Assessment Center in North Carolina, USA, and calculated using the A2 (820 ppm atmospheric CO2 by 2100) and B1 (550 ppm atmospheric CO2 by 2100) climate forcing scenarios from the Special Report of Emission Scenarios (IPCC, 2007). Modeling air temperatures under A2 a nd B1 conditions represented upper and lower CO2 emission thresholds for stream temperature prediction. Air temperature predictions from coupled climate models were used to define three modeled air temperature warming (MATW) increments (+1.7°C, +3.4°C, +5. 1°C) for projecting future MDSTs. These increments covered the range of predicted air temperature increases for 156 Michigan over the next 40 years (Carlson et al., 2016). Stream temperatures were projected under different prevailing weather conditions by appl ying MATW increments to both 2016 (relatively warm, dry) and 2017 (relatively cool, wet) air temperatures (EAWSN, 2018). These conditions were useful for forecasting future stream thermal regimes because they represented weather extremes for Michigan strea ms under predicted changes in climate (i.e., temperature, precipitation; Primack, 2000; Parry et al., 2007; Stoner et al., 2013). Stream temperature models and projections Least - squares linear regression was used to model MDST as a function of MDAT, groundwater, and precipitation. Initial modeling focused on stream - specific relationships between MDST and MDAT. Streams generally warmed throughout summer, but varied in the deg ree to which MDST was correlated with MDAT, indicating stream - specific variability in how groundwater and precipitation affect stream temperature. Hence, in addition to MDAT, thermal effects of groundwater input and precipitation were modeled. In groundwa ter - dominated streams, the thermal influence of groundwater was calculated as accumulated degree - days above mean summer air temperature (ADD) because it is directly related to summer ground surface temperature, the driver of groundwater temperature during this time year (Kurylyk et al., 2013). Previous researchers have successfully used ADD to incorporate groundwater dynamics into stream temperature modeling in the eastern United States (Snyder et al., 2015). The ADD approach is readily applicable in ground water - dominated Michigan streams and offers benefits for resource managers (e.g., accuracy along with practicality and inexpensive data collection) compared to complex heat budget models that require detailed atmospheric, meteorological, and hydrological d ata for a limited number of sites 157 (Webb et al., 2008). Stream temperature models for groundwater - dominated streams took the following form: (1) where MDST i is projected MDST (°C) on day i , MDAT i is projected MDAT (°C) on day i , ADD i is the ADD (degree - days) on day i , m 1 and m 2 are regression coefficients, and b 0 is the model intercept. In runoff - dominated streams, MDST was modeled as a function of MDAT and precipitation according to the following formula: (2) where PR i is the cumulative precipitation (since July 1) on day i and other model components are the same as above. Models including ADD were applied to runoff - dominated streams and those including precipitation were applied to groundwater - dominated streams, but these models were not sufficiently parsimonious (according to bias - see Analyses section below) to warrant further consideration. In addition, both ADD and precipitation models were applied to streams with intermediate groundwater input, but their relatively high AIC c scores (i.e., low parsimony) compared to MDAT - only models rendered it statistically appropriate to model water temperatures in these streams with MDAT alone. Effects of groundwater inputs and precipitation on stream temperatures under projected climate change scenarios may not be the same as those that currently occur (Kurylyk et al., 2013; Menberg et al., 2014). For example, although the amount of precipitation may remain relatively stable in summer, as predicted for Michigan under both high and low CO 2 emissions scenarios 158 (Hayhoe et al., 2014), the temperature of precipitation (or groundwater) may change in a warming climate. Hence, we modeled effects of changes in thermal sensitivity of groundwater (TS gw ; change in groundwater temperature per 1°C air temperature increase; Snyder et al., 2015) and thermal sensitivity precipitation (TS pr; change in precipitation temperature per 1°C air temperature increase) on future stream temperatures. In practice, this involves increasing model y - intercepts by the pro duct of MATW and TS gw (in groundwater - dominated streams; Snyder et al., 2015) or TS pr (in runoff - dominated streams). The stream - specific y - intercept increase is a function of TS gw or TS pr and the proportion of streamflow comprised of groundwater ( R 2 ADD ) or precipitation ( R 2 PR ), calculated as: ( 3) where m 2 is regression coefficient for the ADD i term in (1), S ADD is the standard deviation of ADD, S MDST is the standard deviation of MDST, n is the number of days, ADD i is the ADD at day i , is the mean ADD, MDST i is the MDST at day i , and is the mean MDST (Snyder et al., 2015). Similarly, R 2 PR was calculated as: (4) where m 2 is the regression coefficient for PR i (cumulative precipitation since July 1) in (2), S PR is the standard deviation of PR, S MDST is the standard deviation of MDST, n is the number of days, PR i is the PR at day i , is the mean PR, MDST i is the M DST at day i , and is the mean 159 MDST. To incorporate R 2 ADD and R 2 PR into stream temperature models, linear regressions were developed between model y - intercepts and R 2 ADD (for groundwater - dominated streams) and y - intercepts and R 2 PR values (for runoff - dominated streams): (5) (6) These models explained 68% and 72% of the variation in model y - intercepts, respectively, and residuals were uncorrelated and randomly distributed around ze ro. Hence, these models were considered suitable for use in stream temperature projection (Snyder et al., 2015). To model how changes in air temperature, TS gw , and TS pr would affect model y - intercepts, the following equations were used for groundwater - dominated and runoff - dominated streams: (7) (8) In turn, these adjusted model y - intercepts were used in place of those in equations (1) and (2) to predict stream - specific MDST under different climate change scenarios: (9) (10) 160 Stream temperatures were modeled under three TS gw and TS pr conditions (0.0, 0.66, 1.0) that span a range from insensitive streams (i.e., groundwater and precipitation temperature do not change substantially with air temperature) to highly sensitive streams (i.e., groundwater and precipita tion temperature change in proportion to air temperature). The 0.0 condition was included as a reference point and is likely less realistic than 0.66 and 1.0, which encompass values reported or used in recent studies (Kurylyk et al., 2013; Snyder et al., 2 015). Adjusted multiple regression models (Eq. 7, 8) were used to project MDST in each stream reach based on all combinations of MATW (+1.7°C, +3.4°C, +5.1°C) and TS gw/ TS pr (0.0, 0.66, 1.0). Stream thermal sensitivity the increase in stream temperature resulting from a 1.0°C air temperature increase (Snyder et al., 2015) was also evaluated for each stream under three TS gw conditions. Stream thermal sensitivity measurements were derived by first calculating stream - specific mean summer water temperatures from baseline (i.e., 2016, 2017) data. These values were subtracted from MDSTs calculated using equations (9) and (10) under 1.0°C air temperature warming compared to baseline temperatures. The difference between these two values was treated as an empiric al measurement of stream thermal sensitivity (Snyder et al., 2015). Thermal habitat suitability predictions Stream - specific temperature projections were compared with temperature ranges for trout survival and growth to assess future thermal habitat suita bility under different combinations of MATW and TS gw /TS pr . Temperature thresholds (i.e., thermal minima, maxima ) were defined in reference to juvenile trout (if they differed from those of adults) under the premise that resilient trout fisheries can only b e conserved if young fish survive to adulthood. Streams with optimal 161 trout growing conditions were those that had mean July - August temperatures ranging from 11.0 °C to 16.4 °C (brook trout; Baldwin, 1957; Raleigh, 1982a), 12.0 °C to 16.9 °C (brown trout; H ay et al. 2006 ), and 12.0 °C to 16.3 °C (rainbow trout; Wurtsbaugh and Davis 1977; Raleigh 1982b). Streams with suitable (but not optimal) temperatures for trout growth ranged from 16.5 °C to 20.4 °C (brook trout; Baldwin, 1957; Raleigh, 1982a), 17.0 °C to 19.9 °C (brown trout; Hay et al. 2006 ), and 16.4 °C to 22.4 °C (rainbow trout; Wurtsbaugh and Davis 1977; Raleigh 1982b). Streams that were too warm for trout growth in July - August had temperatures ranging from 20.5 °C to 25.2 °C (brook trout; Baldwin, 19 57 ; Raleigh, 1982a), 20.0 °C to 26.1 °C (brown trout; Hay et al. 2006 ), and 22.5 °C to 24.9 °C (rainbow trout; Wurtsbaugh and Davis 1977; Raleigh 1982b). Finally, streams that were too warm for trout survival in July - August had k trout; Fry et al. 1946 ; Raleigh 1982a), 26.2 °C (brown trout; Hay et al. 2006 ), and 25.0 °C (rainbow trout; Wurtsbaugh and Davis 1977; Raleigh 1982b ). These temperature ranges were used to calculate the proportion of streams studied that would remain sui table for trout survival and growth under alternative climate change scenarios. Analyses Four stream temperature models (i.e., MDAT, MDAT + ADD, MDAT + precipitation, MDAT + ADD + precipitation) were developed and compared for each stream using informati on - theoretic model selection and bias - corrected AICc (Burnham & Anderson, 2002). Invariably, groundwater - dominated streams were modeled most accurately (i.e., lowest AICc - domi nated streams with the precipitation model (MDAT + precipitation), thus equations (9) and (10) were applied to these stream types. Models including both ADD and precipitation were 162 occasionally within two AICc units of top - supported models, wherein the addi tional parameter (relative to the top - performing model) was uninformative (i.e., did not reduce model deviance) such that interpreting the extra parameter would have caused modeling bias (Burnham & Anderson, 2001; Arnold, 2010). Hence, by only including th e top model for each stream, all models received substantial AICc support and only contained informative parameters (i.e., those that reduced model deviance). All analyses were performed in RStudio Desktop version 1.1.423 (RStudio, 2015). Results Stream temperature models and thermal sensitivity Relationships between MDAT and MDST were highly variable among the Michigan coldwater streams evaluated (Table 4. 1). Air and stream temperatures were positively correlated in runoff - dominated systems (e.g., Paint River; R 2 = 0.86; Figure 4. 2a) and in streams with intermediate groundwater input (e.g., Tamarack Creek; R 2 = 0.55; Figure 4. 2b) but not significantly correlat ed in groundwater - dominated streams (e.g., East Branch Fox River; R 2 = 0.09; Figure 4. 2c). Temperatures in most streams (87%) were modeled most accurately (i.e., lowest AICc values) by combining ADD and MDAT in groundwater - dominated streams and precipitati on and MDAT in runoff - dominated streams. Including ADD and precipitation improved model accuracy, with adjusted R 2 values increasing by 0.06 0.75 (Table 4. 1, Figure 4. 3a, b, c). Models including ADD and precipitation had an average adjusted R 2 of 0.83 (ran ge 0.75 0.95), compared to 0.58 (range 0.06 0.80) for unadjusted models with only MDAT (Table 4. 1, Figure 4. 4). Water temperatures in streams with intermediate groundwater input (i.e., Pigeon 163 TABLE 4.1. Michigan stream information and model parameters. Map number refers to stream identifiers in Figure 4. 1. BFI represents base flow index, the mean rate of base flow divided by the corresponding mean rate of total streamflow (Neff et al . 2005). Species denotes the trout species present in each stream (BKT [brook trout], BNT [brown trout], RBT [rainbow trout], All [all three species]). Year denotes the baseline year and corresponding weather conditions (2016: warm, dry; 2017: cool, wet) from which the mo del was developed. Other abbreviations denote model intercepts (Int); coefficients for mean daily air temperature (MDAT), accumulated degree - days above mean summer air temperature (ADD, a measure of groundwater input), and cumulative daily precipitation si nce July 1 (PR); P values; bias - R 2 values (Adj R 2 ) for groundwater - and precipitation - corrected models; and adjusted R 2 values for MDAT - only models (MDAT Adj R 2 ). Stream Map BFI Species Y ear Int MDAT ADD PR P AICc Adj R 2 MDAT Adj R 2 Au Sable R. 1 0.67 All 2016 15.91 0.14 - 0.06 -- <0.01 43.80 0.79 0.60 2017 14.45 0.18 - 0.05 -- <0.01 38.67 0.84 0.70 E. Branch Fox R. 2 0.61 BKT, BNT 2016 11.13 0.12 - 0.04 -- <0.01 19.29 0.75 0.46 2017 10.64 0.02 - 0.03 -- <0.01 36.87 0.75 0.10 Manistee R. 3 0.61 All 2016 14.38 0.20 - 0.08 -- <0.01 51.55 0.80 0.62 2017 15.53 0.09 - 0.06 -- <0.01 41.25 0.77 0.71 Pigeon R. 4 0.60 BNT, RBT 2016 10.52 0.14 -- -- <0.01 47.04 0.56 -- 2017 9.08 0.21 -- -- <0.01 77.94 0.58 -- W Br Sturgeon R. 5 0.60 All 2016 9.86 0.18 -- -- <0.01 57.76 0.45 -- 2017 9.45 0.19 -- -- <0.01 87.34 0.52 -- Sturgeon R. 6 0.59 BKT, RBT 2016 14.04 0.40 -- - 0.64 <0.01 75.86 0.89 0.64 2017 13.35 0.35 -- - 0.55 <0.01 100.93 0.86 0.65 Tamarack Creek 7 0.55 BNT, RBT 2016 15.23 0.47 -- - 0.52 <0.01 40.23 0.90 0.55 2017 11.37 0.37 -- - 0.31 <0.01 54.86 0.88 0.80 Black R. 8 0.51 BKT 2016 11.98 0.22 -- - 0.40 <0.01 33.69 0.83 0.77 2017 8.42 0.22 -- - 0.60 <0.01 51.52 0.85 0.72 Canada Creek 9 0.51 BKT 2016 15.72 0.11 -- - 0.71 <0.01 49.40 0.78 0.55 2017 14.22 0.21 -- - 0.60 <0.01 59.42 0.75 0.62 164 Stream Map BFI Species Year Int MDAT ADD PR P AICc Adj R 2 MDAT Adj R 2 Rapid R. 10 0.50 All 2016 11.55 0.08 -- - 1.44 <0.01 25.72 0.85 0.47 2017 11.76 0.04 -- - 2.51 <0.01 69.46 0.84 0.49 Paint R. 11 0.49 BNT 2016 18.50 0.32 -- - 0.76 <0.01 46.37 0.95 0.72 2017 21.00 0.18 -- - 1.09 <0.01 67.65 0.81 0.06 Rogue R. 12 0.47 BNT, RBT 2016 20.28 0.14 -- - 0.46 <0.01 77.50 0.83 0.57 2017 16.83 0.22 -- - 1.57 <0.01 69.38 0.84 0.66 Pine R. 13 0.44 BNT 2016 15.58 0.08 -- - 0.20 <0.01 37.06 0.75 0.50 2017 13.18 0.06 -- - 0.93 <0.01 31.03 0.79 0.43 Cedar R. 14 0.38 BNT 2016 14.60 0.27 -- - 0.68 <0.01 51.24 0.77 0.57 2017 12.76 0.35 -- - 0.63 <0.01 74.87 0.87 0.74 St Joe R. 15 0.35 BKT, RBT 2016 17.56 0.23 -- - 0.62 <0.01 19.65 0.86 0.71 2017 15.96 0.28 -- - 1.20 <0.01 62.32 0.89 0.76 165 FIGURE 4.2. Relationships between predicted mean daily water temperature and observed mean daily air temperature for Michigan trout streams. Graphs (a), (b), and (c) display examples that span the range of air - stream temperature relationships and corresponding re gression statistics (including slope, y - intercept, adjusted R 2 ) observed in this study. 166 FIGURE 4.3. Comparison of predictions of mean daily water temperature in Michigan streams between linear regressions that used only mean daily air temperature (MDAT) as an independent variable, and models that used both MDAT and an additional predictor (i.e., accumulated degree - days above mean summer air temperature [ADD], cumulative daily precipitation since July 1 [PR]). Graphs (a), (b), and (c) show streams spannin g a gradient of baseflow from runoff - dominated to groundwater - dominated and encompassing the range of air - stream temperature relationships observed in this study. R 2 values are adjusted R 2 . 167 FIGURE 4.4. Comparison of the distribution of adjusted R 2 values for Michigan stream temperature linear regressions that used only mean daily air temperature (MDAT only) as an independent variable, and models that used both MDAT and an additional predictor (i.e., accumulated degree - days above mean summer air tem perature [ADD], cumulative daily precipitation since July 1 [PR]). 168 River, West Branch Sturgeon River) were modeled most accurately with MDAT alone (Table 4. 1). These findings suggest that fisheries managers should model groundwater - dominated stream temper atures with ADD and MDAT, runoff - dominated stream temperatures with precipitation and MDAT, and temperatures of streams with intermediate groundwater input with MDAT alone. Stream thermal sensitivity tended to decline with increasing groundwater input but was highly influenced by TSgw conditions. The decrease in stream thermal sensitivity with increasing groundwater input was most pronounced for TSgw = 0 (Figure 4. 5a) and considerably weaker for TSgw = 0.66 (Figure 4. 5b). For TSgw = 1.00, thermal sensitivity remained stable regardless of groundwater input (Figure 4. 5c), indicating removal of groundwater - driven temperature buffering and thus less favorable thermal conditions for trout survival and growth. Stream temp erature projections Projected future water temperatures in Michigan coldwater streams varied among climate change scenarios (i.e ., combinations of MATW and TS gw /TS pr ) and between baseline weather conditions (i.e., warm/dry, cool/wet) used to construct stre am temperature models. In warm, dry conditions, projected stream temperat ures warmed as both MATW and TS gw /TS pr increased (Table 4. 2). In groundwater - dominated streams, mean projected stream temperature across TS gw categories was 16. 58°C for MATW = 1.7°C, 17.36°C for MATW = 3.4°C, and 18.15°C for MATW = 5.1°C (Figure 4. 6). Within all MATW categories, predicted groundwater - dominated stream temperatures warmed as TS gw increased. The magnitude of this warming increased as MATW intensifie d from +1.7°C (0.94°C) to +3.4°C (1.88°C) and +5.1°C (2.82°C; Figure 4. 6). These results indicate that in warm, dry future weather conditions, water temperatures in 169 FIGURE 4.5. Relationships between modeled stream thermal sensitivity (°C water/°C air) a nd groundwater input for the 13 Michigan streams best modeled with MDAT + PR (surface runoff - dominated [RD] streams) and MDAT + ADD (groundwater - dominated [GD] streams). Graphs (a), (b), and (c) display stream thermal sensitivities for three conditions of increasing groundwater thermal sensitivity (TS gw = 0.0, 0.66. 1.0). Dotted lines denote transitions between RD and GD streams. 170 TABLE 4.2. Michigan stream temperatures and projected future temperatures in three conditions of modeled air temperature warming (+1.7°C, +3.4°C, +5.1°C) and thermal sensitivity of groundwater/precipitation (0.0, 0.66, 0.1; in parentheses) based on 2016 weathe r conditions (i.e., warm, dry). Stream 2016 +1.7(0) +1.7(0.66) +1.7(1) +3.4(0) +3.4(0.66) +3.4(1) +5.1(0) +5.1(0.66) +5.1(1) Au Sable R. 17.25 17.50 18.33 18.75 17.73 19.39 20.24 17.97 20.45 21.73 E. Branch Fox R. 12.79 13.14 13.48 13.66 13.35 14.04 14.40 13.56 14.60 15.14 Manistee R. 17.22 17.55 18.23 18.58 17.89 19.26 19.96 18.23 20.29 21.34 Pigeon R. 13.54 13.78 13.78 13.78 14.02 14.02 14.02 14.26 14.26 14.26 W Br Sturgeon R. 13.65 13.95 13.95 13.95 14.25 14.25 14.25 14.55 14.55 14.55 Sturgeon R. 23.81 24.30 25.04 25.42 24.97 26.45 27.21 25.64 27.87 29.01 Tamarack Creek 19.80 20.87 21.82 22.31 21.67 23.57 24.55 22.48 25.32 26.79 Black R. 15.91 16.35 17.16 17.59 16.71 18.35 19.19 17.08 19.53 20.80 Canada Creek 16.99 17.26 18.10 18.53 17.44 19.12 19.99 17.62 20.15 21.45 Rapid R. 12.87 13.30 13.87 14.17 13.43 14.58 15.17 13.56 15.28 16.17 Paint R. 22.94 23.38 24.37 24.87 23.92 25.90 26.91 24.46 27.43 28.95 Rogue R. 22.08 22.38 23.26 23.72 22.61 24.39 25.30 22.85 25.51 26.88 Pine R. 16.92 17.24 18.03 18.43 17.38 18.95 19.76 17.51 19.87 21.08 Cedar R. 19.25 19.68 20.48 20.89 20.14 21.73 22.55 20.59 22.98 24.22 St Joe R. 22.28 22.61 23.51 23.97 23.00 24.80 25.72 23.40 26.09 27.48 171 FIGURE 4.6. Distribution of projected mean daily water temperatures based on 2016 weather conditions (warm, dry) in three conditions of modeled air temperature warming (+1.7°C, +3.4°C, +5.1°C) and thermal sensitivity of groundwater/precipitation (0.0, 0.66, 0.1) for the 13 Michigan streams best modeled with MDAT + PR and MDAT + ADD. Within each box plot, small dashed lines are means and solid lines ar e medians. Long dashed lines spanning the entire panel represent upper thermal thresholds for brook trout (BKT), brown trout (BNT), and rainbow trout (RBT), whereas long dotted lines denote upper thermal thresholds for growth. 172 groundwater - dominated Michi gan streams will rise in proportion to the magnitude of air temperature warming and groundwater thermal sensitivity amid climate change. However, predicted increases in groundwater - dominated stream temperatures will be small relative to those projected for runoff - dominated streams studied herein. In runoff - dominated streams under warm, dry weather conditions, projected water temperatures were appreciably higher than those in groundwater - dominated systems (Figure 4. 6) despite similar trajectories of thermal change (i.e., stream temperatures increase in proportion to MATW and TS pr ; Table 4. 2). In runoff - dominated streams, mean projected stream temperature across TS pr categories was 20.43°C for MATW = 1.7°C, 21.52°C fo r MATW = 3.4°C, and 22.60°C for MATW = 5.1°C (Figure 4. 6). Predicted runoff - dominated stream temperatures warmed in proportion to TS pr within MATW categories. As in groundwater - dominated streams, the magnitude of this increase became larger in runoff - domin ated systems as MATW increased from +1.7°C (1.25°C) to +3.4°C (2.51°C) and +5.1°C (3.76°C; Figure 4. 6). These findings suggest that in warm, dry future weather conditions, water temperatures in runoff - dominated Michigan streams will increase more than those in groundwater - dominated streams while following a similar proportional relationship with the magn itude of air temperature warming and groundwater thermal sensitivity in a changing climate. In cool, wet weather conditions, projected stream temperatures were lower than those predicted in warm, dry conditions (Figure 4. 6, 4. 7). However , temperatures wa rmed with increasing MATW and TS gw / TS pr (Table 4. 3). In groundwater - dominated streams, mean projected stream temperature across TS gw categories was 15.15°C for MATW = 1.7°C, 15.88°C for MATW = 3.4°C, and 16.61°C for MATW = 5.1°C (Figure 4. 7). Predicted gro undwater - dominated stream temperatures warmed as TS gw increased within all MATW categories. The 173 FIGURE 4.7. Distribution of projected mean daily water temperatures based on 2017 weather conditions (cool, wet) in three conditions of modeled air temperatu re warming (+1.7°C, +3.4°C, +5.1°C) and thermal sensitivity of groundwater/precipitation (0.0, 0.66, 0.1) for the 13 Michigan streams best modeled with MDAT + PR and MDAT + ADD. Within each box plot, small dashed lines are means and solid lines are medians . Long dashed lines spanning the entire panel represent upper thermal thresholds for brook trout (BKT), brown trout (BNT), and rainbow trout (RBT), whereas long dotted lines denote upper thermal thresholds for growth. 174 T ABLE 4.3. Michigan stream temperatures and projected future temperatures in three conditions of modeled air temperature warming (+1.7°C, +3.4°C, +5.1°C) and thermal sensitivity of groundwater/precipitation (0.0, 0.66, 0.1; in parentheses) based on 2017 weathe r conditions (i.e., coo l, wet). Stream 2017 +1.7(0) +1.7(0.66) +1.7(1) +3.4(0) +3.4(0.66) +3.4(1) +5.1(0) +5.1(0.66) +5.1(1) Au Sable R. 16.52 16.72 17.55 17.97 17.03 18.68 19.53 17.34 19.82 21.09 E. Branch Fox R. 10.86 11.51 11.98 12.22 11.55 12.49 12.98 11.59 13.00 13.73 Manistee R. 15.37 15.52 16.24 16.61 15.68 17.12 17.86 15.84 18.00 19.11 Pigeon R. 12.90 13.18 13.18 13.18 13.53 13.53 13.53 13.88 13.88 13.88 W Br Sturgeon R. 12.81 13.10 13.10 13.10 13.43 13.43 13.43 13.75 13.75 13.75 Sturgeon R. 21.76 22.17 22.98 23.40 22.76 24.38 25.21 23.35 25.77 27.02 Tamarack Creek 17.84 18.59 19.09 19.35 19.23 20.23 20.75 19.86 21.36 22.14 Black R. 14.33 14.79 15.12 15.29 15.15 15.82 16.16 15.52 16.52 17.03 Canada Creek 16.91 17.17 17.97 18.39 17.52 19.13 19.96 17.86 20.29 21.54 Rapid R. 11.85 12.23 12.80 13.10 12.30 13.44 14.03 12.37 14.09 14.97 Paint R. 21.21 21.65 22.53 22.99 21.96 23.73 24.64 22.27 24.93 26.30 Rogue R. 19.51 20.32 21.24 21.71 20.69 22.53 23.48 21.06 23.82 25.24 Pine R. 13.04 13.40 14.22 14.64 13.50 15.13 15.98 13.59 16.05 17.32 Cedar R. 18.11 18.30 19.09 19.50 18.89 20.49 21.31 19.49 21.88 23.11 St Joe R. 19.80 20.20 21.15 21.63 20.67 22.56 23.53 21.15 23.97 25.43 175 magnitude of this increase became larger as MATW increased from +1.7°C (1.02°C) to +3.4°C (2.03°C) and +5.1°C (3.05°C; Figure 4. 7). These results indicate that in cool, wet future weather conditions, water temperatures of groundwater - dominated Michigan str eams will rise in proportion to the magnitude of air temperature warming and groundwater thermal sensitivity amid climate change. Nevertheless, predicted increases in groundwater - dominated stream temperatures will be small relative to: 1) those predicted f or groundwater - dominated streams in warm, dry weather conditions; and 2) those projected for runoff - dominated streams in cool, wet weather conditions. In runoff - dominated streams under cool, wet weather conditions, projected water temperatures were higher than those in groundwater - dominated systems (Figure 4. 7) notwithstanding similar trajectories of thermal change (i.e., stream temperatures increase in proportion to MATW and TS pr ; Table 4. 3). Mean projected stream temperature across TS pr categories in runoff - dominated streams was 18.50°C for MATW = 1.7°C, 19.50°C for MATW = 3.4°C, and 20.51°C for MATW = 5.1°C (Figure 4. 7). Predicted runoff - dominate d stream temperatures warmed as TS pr increased within MATW categories. As in groundwater - domi nated streams, the magnitude of this increase became larger in runoff - dominated systems as MATW increased from +1.7°C (1.12°C) to +3.4°C (2.24°C) and +5.1°C (3.36°C; Figure 4. 7). These findings suggest that in cool, wet future weather conditions, water tem peratures in runoff - dominated Michigan streams will increase in proportion to air temperature warming and groundwater thermal sensitivity in a changing climate. Such temperature changes will likely be large relative to those in groundwater - dominated stream s in similar weather conditions but small relative to those in runoff - dominated streams in warm, dry conditions. 176 Thermal habitat suitability predictions In 2016, all streams evaluated in this study had MDST that were suitable for survival of brook trout, brown trout, and rainbow trout. Most streams (73%, n = 11) had temperatures that were optimal or suitable for summer trout growth. Four runoff - dominated s treams (Paint, Rogue, St. Joseph, and Sturgeon rivers) had temperatures that were unsuitable for summer growth of brook trout and brown trout, including two streams (Paint and Sturgeon rivers) that were unsuitable for rainbow trout growth (Table 4. 2). Base d on 2016 weather conditions (i.e., relatively warm, dry), thermal habitats in groundwater - dominated streams were predicted to be suitable for survival of all trout species regardless of MATW/ TS gw conditions (Table 4. 4, Figure 4. 6). Projected groundwater - dominated stream temperatures were suitable for summer growth of rainbow trout in all MATW/ TS gw conditions, brook trout in all but one condition (MATW = 5.1°C, TS gw = 1), and brown trout in all but three conditions (MATW = 3.4°C, TS gw = 1; MATW = 5.1°C, TS gw = 0.66; MATW = 5.1°C, TS gw = 1; Table 4. 2, 4. 4, Figure 4. 6). Hence, even if climate change results in warm, dry weather conditions, Michigan stream trout will continue to survive and generally maintain summer growth in the groundwater - domina ted streams studied herein. In warm, dry weather conditions, runoff - dominated streams became progressively less suitable for survival and growth of all trout species as MATW and TS pr increased (Table 4. 4, Figure 4. 6). As MATW increased from +1.7°C t o +5.1°C, the mean percentage of runoff - dominated streams suitable for brook trout survival and growth declined from 93% (survival) and 60% (growth) to 67% (survival) and 47% (growth). Likewise, the mean percentage of runoff - dominated streams suitable for brown trout survival and growth decreased from 100% (survival) and 39% (growth) to 77% (survival) and 28% (growth; Table 4. 4, Figure 4. 6). 177 TABLE 4.4. Thermal habitat suitability for brook trout (BKT), brown trout (BNT), and rainbow trout (RBT) survival and growth in Michigan streams based 2016 weather conditions (i.e., warm, dry). Table entries report the proportion of streams that are sui table fo r survival (numerator) and growth (denominator) in three conditions of modeled air temperature warming (+1.7°C, +3.4°C, +5.1°C) and thermal sensitivity of groundwater/precipitation (0.0, 0.66, 0.1; in parentheses). Streams are classified as groundwater - dom inated (GD; base flow index > 0.60), runoff - dominated (RD; base flow index < 0.60), or systems with intermediate groundwater input (Int; base flow in dex = 0.60; Neff et al., 2005). Species Stream +1.7(0) +1.7(0.66) +1.7(1) +3.4(0) +3.4(0.66) +3.4(1) +5.1(0 ) +5.1(0.66) +5.1(1) BKT GD (3) 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/0.33 RD (5) 1/0.6 1/0.6 0.8/0.6 1/0.6 0.8/0.6 0.6/0.6 0.8/0.6 0.6/0.6 0.6/0.2 Int (1) 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 BNT GD (3) 1/1 1/1 1/1 1/1 1/1 1/0.67 1/1 1/0.33 1/0.33 RD (6) 1/0.5 1/0.33 1/0.33 1/0.33 1/0.33 0.83/0.33 1/0.33 0.83/0.33 0.5/0.17 Int (2) 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 RBT GD (2) 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 RD (5) 1/0.6 0.8/0.4 0.8/0.4 1/0.4 0.8/0.2 0.4/0.2 0.8/0.4 0.2/0.2 0.2/0.2 Int (2) 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 178 Rainbow trout exhibited a similar trend, as 87% and 47% of runoff - dominated streams were suitable for survival and growth when MATW = +1.7°C, compared to 40% (survival) and 27% (growth) when MATW = +5.1°C. Streams with intermediate groundwater input (i.e., Pigeon and West Branch Sturgeon rivers) were projected to be suitable for survival and growth of all trout species in all MATW/ TS pr conditions (Table 4. 4, Figure 4. 6). Thus, in warm, dry future weather conditions, trout survival and growth will decrease in runoff - dominated Michigan streams to a greater degree than groundwater - dominated streams in similar weather conditions. In relatively cool, wet weather conditions of 2017, streams were generally more suitable for trout survival and growth than in 2016 ( i.e., warm, dry). All streams were suitable for trout survival in 2017, and most streams (87%, n = 13) had temperatures that were optimal or suitable for summer trout growth (Table 4. 3). Only two streams (Paint River, Sturgeon River) were unsuitably warm f or summer growth. Based on 2017 conditions, thermal habitats in groundwater - dominated streams were predicted to be suitable for survival of all trout species regardless of MATW/ TS gw conditions (Table 4. 5, Figure 4. 7). Projected groundwater - dominated stream temperatures were suitable for summer growth of rainbow trout in all MATW/ TS gw conditions and brook trout and brown trout in all but one condition (MATW = 5.1°C, TS gw = 1; Table 4. 3, 4. 5, Figure 4. 7). Hence, trout will continue to survive and generally ma intain summer growth in groundwater - dominated Michigan streams if climate change generates relatively cool, wet future weather conditions. In cool, wet weather conditions, runoff - dominated streams became less suitable for trout survival and growth a s MATW and TS pr increased, but to a lesser extent than in warm, dry conditions (Table 4. 4, 4. 5, Figure 4. 6, 4. 7). As MATW increased from +1.7°C to +5.1°C in cool, wet conditions, the mean percentage of runoff - dominated streams suitable for brook trout 179 TABLE 4.5. Thermal habitat suitability for brook trout (BKT), brown trout (BNT), and rainbow trout (RBT) survival and growth in Michigan streams based 2017 weather conditions (i.e., cool, wet). Table entries report the proportion of streams that are sui tab le for survival (numerator) and growth (denominator) in three conditions of modeled air temperature warming (+1.7°C, +3.4°C, +5.1°C) and thermal sensitivity of groundwater/precipitation (0.0, 0.66, 0.1; in parentheses). Streams are classified as groundwate r - dominated (GD; base flow index > 0.60), runoff - dominated (RD; base flow index < 0.60), or systems with intermediate groundwater input (Int; base flow in dex = 0.60; Neff et al., 2005). Species Stream +1.7(0) +1.7(0.66) +1.7(1) +3.4(0) +3.4(0.66) +3.4(1) + 5.1(0) +5.1(0.66) +5.1(1) BKT GD (3) 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/0.67 RD (5) 1/0.8 1/0.6 1/0.6 1/0.6 1/0.6 1/0.6 1/0.6 0.8/0.6 0.6/0.4 Int (1) 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 BNT GD (3) 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/0.67 RD (6) 1/0.67 1/0.67 1/0.67 1/0.67 1/0.33 1/0.33 1/0.67 1/0.33 0.83/0.33 Int (2) 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 RBT GD (2) 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 RD (5) 1/1 1/0.8 1/0.8 1/0.8 1/0.4 0.8/0.4 1/0.8 0.8/0.4 0.4/0.4 Int (2) 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 180 survival and growth declined from 100% (survival) and 67% (growth) to 80% (survival) and 53% (growth). Brown trout exhibited a similar trend, as 100% and 67% of runoff - dominated streams were suitable for survival and growth when MATW = +1.7°C, compared to 94% (sur vival) and 44% (growth) when MATW = +5.1°C (Table 4. 5, Figure 4. 7). Likewise, the mean percentage of runoff - dominated streams suitable for rainbow trout survival and growth decreased from 100% (survival) and 87% (growth) to 73% (survival) and 53% (growth) as MATW increased from +1.7°C to +5.1°C. Streams with intermediate groundwater input were projected to be suitable for survival and growth of all trout species regardless of MATW/ TS pr conditions (Table 4. 5, Figure 4. 7). Thus, in cool, wet future weather co nditions, runoff - dominated streams will exhibit reductions in trout survival and especially growth that are large relative to those in groundwater - dominated streams in similar weather conditions but small relative to those in runoff - dominated streams in wa rm, dry conditions. Overall, Michigan coldwater stream temperatures were projected to increase in proportion to the magnitude of air temperature warming and groundwater thermal sensitivity under various climate change scenarios. Stream temperature warming was predicted to be more severe under warm, dry future weather conditions than cool, wet conditions. However, regardless of weather conditions, groundwater - dominated stream temperatures were projected to increase by an appreciably smaller magnitude than r unoff - dominated stream temperatures such that brook trout, brown trout, and rainbow trout will continue to survive and generally maintain summer growth in groundwater - dominated systems studied herein. In contrast, trout survival and growth were projected t o decrease in runoff - dominated streams with continued air temperature warming and increasing groundwater thermal sensitivity, particularly in warm, dry future weather conditions. 181 Discussion Michigan coldwater streams vary widely in temperature and thermal habitat suitability for summer trout survival and growth. Whereas runoff - dominated streams are highly sensitive to changes in air temperature, groundwater - dominated streams have thermal regi mes that are less affected by air temperature alterations due to groundwater - driven thermal buffering. This indicates a need for a stream temperature modeling approach that accommodates among - stream heterogeneity in thermal drivers (e.g., air temperature, precipitation, groundwater) in an accurate, accessible, management - relevant way. Although heat budget models incorporate a wide array of stream temperature drivers (Leach & Moore, 2011; Westhoff et al., 2011) including groundwater and precipitation the y are extremely data - intensive, site - restrictive (i.e., only applied in a limited number of streams), expensive to develop, and thus often impractical for fisheries and aquatic resource professionals who seek to efficiently allocate limited fisheries manag ement resources across relatively large areas containing diverse streams and watersheds. Likewise, although air - stream temperature models are inexpensive and readily applicable for climate change forecasting, they overlook hydrological influences on stream temperature including climate - driven changes in precipitation and groundwater (Menberg et al., 2014) and ignore reach - level variation in stream thermal sensitivity (Tague et al., 2007; Kelleher et al., 2012). Hence, in this study we developed an accur ate, cost - effective, management - relevant methodology for predicting effects of changes in air temperature, precipitation, and groundwater on water temperatures in brook trout, brown trout, and rainbow trout streams amid climate change. Our results indica te that basic modifications to air - stream temperature models greatly improve model fit while incorporating stream - specific effects of groundwater and precipitation 182 on water temperature. Moreover, models adjusted for ADD and precipitation can be used to pro ject future stream temperatures and trout thermal habitat suitability under varying climate change scenarios (i.e., combinations of MATW and TS gw /TS pr ). Importantly, these adjusted models can be applied to hydrologically diverse streams (i.e., both groundw ater - dominated and runoff - dominated), thereby expanding the scope of previous research wherein adjusted models were only applied to groundwater - dominated systems (Snyder et al., 2015). As observed herein and in previous research (Westhoff & Paukert, 2014), status (groundwater - dominated or runoff - dominated) and its most accurate temperature model (MDAT + ADD or MDAT + precipitation) allows fisheries and aquatic resource professionals and policy makers to use stream hydr ology as a criterion to select variables to include in stream temperature models (i.e., MDAT + ADD in groundwater - dominated streams, MDAT + precipitation in runoff - dominated streams). Other researchers have employed network models to accurately predict str eam temperatures in the western USA based on thermal spatial autocorrelation driven by elevation gradients and tributary effects (Peterson & Ver Hoef, 2010). However, stream thermal dynamics in Michigan are different from those in western USA streams, as M ichigan has relatively little elevation change, larger groundwater influence, more precipitation as rain, and less snowmelt over a shorter annual period. Therefore, stream temperature modeling in the present study required an approach based on groundwater and summer precipitation (i.e., rain). Previous authors have acknowledged the importance of accounting for changes in precipitation patterns in stream climate change modeling (Snyder et al., 2015), but few accurate, cost - effective, management - relevant mode ls that include precipitation have been generated prior to this research. The stream temperature modeling approach developed herein for Michigan streams is readily transferable to other areas where the 183 primary drivers of stream temperature are air temperat ure, groundwater, and precipitation (primarily as rain) rather than elevation change or snowmelt. were consistently negative in 2016 and 2017 in all streams (Table 4. 1), indicating thermal buffering effects of both groundwater and precipitation in summer. On average, precipitation weather in the latter year (EAWSN, 2018) and demonstrating an overall cooling effect of summer precipitation on stream temperature in runoff - dominated streams. This suggests precipitation, by increasing stream discharge and water volume, can offset effects of increased air temperature on stream temp eratures (Merriam et al., 2017). In addition, modeling stream temperatures under divergent weather conditions (i.e., warm/dry, cool/wet) enabled us to generate a diversity of models that encompassed the range of temperature and precipitation conditions tha t Michigan streams currently experience and are projected to experience in the future. This, in turn, provides fisheries and aquatic resource professionals with a flexible thermal modeling approach for forecasting stream temperatures along a gradient of fu ture temperature and precipitation regimes (Primack, 2000; Parry et al., 2007; Stoner et al., 2013). Effects of groundwater and precipitation on stream temperatures may be different in the future compared to the present (Kurylyk et al., 2013; Menberg et a l., 2014), indicating a need to account for these changes by incorporating stream thermal sensitivity and associated variables (e.g., TS gw , TS pr ) into stream temperature modeling. Although stream thermal sensitivity declined with increasing groundwater inp ut, as expected, this relationship was dependent on TS gw conditions. For instance, if groundwater temperature is not affected by warming air temperatures in a changing climate, increased groundwater input will continue to have a 184 buffering effect on stream temperature (Figure 4. 5a). However, more realistically, groundwater - driven thermal buffering will become weaker or non - existent (Figure 4. 5b,c) if TS gw falls between 0.66 and 1.0 as predicted by recent research regarding effects of climate change on groundwater dynamics in coldwater streams (Kurylyk et al., 2013; Snyder et al., 2015). Hence, although groundwater is an important source of cool, oxygen - rich water that trout managers must consider in stream thermal habitat management , groundwater - dominated streams are not static systems that will invariably remain cold in a changing climate. In addition to groundwater, precipitation affects stream thermal regimes, yet it is rarely accounted for in stream temperature modeling studies (Snyder et al., 2015). This likely reflects the indirect nature and logistical difficulty of measuring mechanisms whereby precipitation influences stream temperature (e.g., changes in timing and magnitude of discharge, turbidity). In the present study, pre cipitation was projected to have a cooling effect on water temperatures in runoff - dominated Michigan streams via increased stream discharge, as also documented in a West Virginia trout stream (Merriam et al., 2017). Such an important finding would have bee n masked in the absence of a modeling approach that explicitly accounted for effects of climate - induced changes in precipitation on stream temperature. Hence, precipitation - corrected stream temperature models represent tools that fisheries and aquatic reso urce professionals can use to reliably forecast effects of changing precipitation regimes on stream trout survival and growth. In turn, they can make corresponding improvements to fisheries management programs and policies and ultimately enhance fisheries sustainability amid climate change. For instance, fisheries managers could use models of current or future precipitation to identify trout streams with precipitation - driven cooling in summer and prioritize them for management activities (e.g., riparian zon e protection/rehabilitation to maintain/increase stream shading). 185 The magnitude of stream warming predicted herein was comparable to previous research in Michigan and surrounding Great Lakes states. Projected stream temperature warming based on the most extreme predicted climatic changes (MATW = 5.1°C, TS gw /TS pr = 1.0) was 0.7 7.0°C for warm, dry weather conditions and 0.9 5.7°C for cool, wet conditions. These ranges are similar to those documented in studies in Michigan (0.1 6.8°C; Carlson et al., 2017), Wisconsin (0.8 4.0°C; Lyons et al., 2010), and Minnesota (0.3 6.9°C; Pilgrim et al., 1998). However, in the present study, explicit consideration of hydrological alterations driven by climate change drivers increased the reliability of stream temperature projections relative to previous research. For example, stream temperatures were predicted to increase by 0.1 3.8°C in groundwater - dominated Michigan streams and 0.2 6.8°C in runoff - dominated streams by Carlson et al. (2017). However, probable changes in t hermal sensitivity of groundwater and precipitation were not accounted for in prior Michigan studies, so projected warming reported herein for groundwater - dominated streams (2.4 4.5°C [warm, dry]; 2.9 4.6°C [cool, wet]) and runoff - dominated systems (3.3 7. 0°C [warm, dry]; 2.7 5.7°C [cool, wet]) is likely more reliable and thus of greater assistance to fisheries managers for trout thermal habitat management. Overall, this study corroborates the observation that groundwater - dominated streams are more thermall y resilient than runoff - dominated streams and thus offer better growing conditions for trout during warm summer months (Zorn et al., 2012) and other thermally stressful times of year. These findings suggest that groundwater conservation efforts (e.g., prot ecting groundwater recharge by conserving grasslands and restricting groundwater withdrawal; Waco & Taylor, 2010) should be a key component of coldwater stream fisheries management programs in Michigan and elsewhere. 186 Results of this study indicate that c urrent climate change drivers will likely increase Michigan stream temperatures and degrade thermal habitats for summer trout survival and growth, with the extent of degradation dependent on the magnitude of air temperature warming and changes in groundwat er input and precipitation patterns. Overall, groundwater - dominated streams were predicted to be more thermally resilient than runoff - dominated streams, yet the thermal buffering capacity of groundwater was appreciably lower when TS gw = 0.66 or 1.0, both o f which are more realistic than TS gw = 0.0 (Kurylyk et al., 2013; Snyder et al., 2015). Nevertheless, all groundwater - dominated streams studied herein were projected to remain suitable for survival of brook trout, brown trout, and rainbow trout regardless of climate change conditions (i.e., MATW, TS gw /TS pr ) and baseline weather (i.e., warm/dry, cool/wet). Projected summer trout survival and growth were lower in runoff - dominated streams than groundwater - dominated streams and declined progressively as climati c changes intensified (i.e., MATW = +3.4°C or 5.1°C, TS gw /TS pr = 0.66 or 1.0). Rainbow trout were relatively more resistant to reductions in summer growth compared to brook trout and especially brown trout because they can grow in a moderately wider temper ature range than the latter species (Wurtsbaugh and Davis 1977). Although the greatest declines in trout survival and growth were predicted in runoff - dominated streams in warm, dry weather conditions, survival and growth were still affected in cool, wet co nditions. Therefore, resource managers should prioritize trout management in runoff - dominated streams capable of thermal habitat rehabilitation particularly those predicted to experience increased precipitation and an associated cooling effect and in g roundwater - dominated streams where management activities produce tangible returns on investment relative to trout survival and growth. For example, it would be inefficient to expend resources (e.g., money, time, personnel) in runoff - dominated streams that exceed trout temperature tolerances or, 187 alternatively, groundwater - dominated streams that can sustain their thermal habitat conditions independent of human intervention (e.g., cold streams such as the East Branch Fox River). Projected declines in thermal habitat suitability for summer trout growth and, to a lesser extent, survival will likely be manifested by changes in thermal habitat availability and connectivity. For instance, reduced overall availability of thermal habitats suitable for trout grow th will likely make it more difficult for fish to move to cold habitats, with possible effects on trout population distribution in Michigan (Carlson et al., 2016), where dam - induced habitat fragmentation is already extensive (Cooper et al. 2016). By isolat ing coldwater habitats in particular locations (e.g., headwaters), climate change could subdivide stream trout populations and further reduce survival (Steen et al., 2010). For instance, in Wisconsin, USA, the length of streams suitable for brook trout was projected to decline by 44%, 94%, and 100% under climate change classified as limited (summer air temperatures increase 1.0°C and water 0.8°C), moderate (air +3.0°C, water +2.4°C), and major (air +5.0 °C, water +4.0 °C; Lyons et al., 2010). Although annua l (as opposed to summer) trout growth could stabilize or increase in a changing climate due to a lengthened growing season (and perhaps greater availability of warmwater prey), fisheries and aquatic resource professionals should account for predicted reduc tions in trout survival and growth during the warmest period of the year as they develop management programs to increase trout sustainability. Results of this study have important implications and applications for trout management within and beyond Michi gan. Managing coldwater streams and their thermally sensitive fish populations for thermal resilience (i.e., ability to absorb temperature changes and retain ecosystem structure and function; Holling, 1973) will be increasingly important as climate change continues to degrade coldwater habitats throughout the world (Almodóvar et al., 2012; 188 Isaak et al., 2012; Santiago et al., 2015). Amid limitations in money, time, personnel, and other resources for fisheries management, fisheries and aquatic resource profe ssionals will benefit from accurate, cost - effective, management - relevant stream temperature models such as those described herein. In turn, lower resource expenditure on temperature data collection and model development will allow greater resource allocati on toward thermal habitat management activities. For instance, resource managers can form public private partnerships to protect key groundwater resources (e.g., springs, seeps); conserve grasslands and other watershed - level vegetation and soil types that promote groundwater recharge; preserve and rehabilitate riparian trees and plants that promote water infiltration, percolation, and stream shading; and sustain longitudinal stream connectivity to allow trout movement to coldwater habitats during summer (Wa co & Taylor, 2010; Siitari et al., 2011). Collectively, these and related strategies can be used to develop resilience - based management programs for coldwater streams and their important trout fisheries (Carlson et al., 2016). Resilience - based management involves collaboration among scientists, biologists, policy makers, and public stakeholders to cultivate fisheries ecosystems that are robust to local and global change (e.g., climate change, land use alteration) and, likewise, management systems that can withstand environmental and socioeconomic stressors (Paukert et al., 2016). For example, resource managers can increase the resilience of fisheries ecosystems by protecting a diversity of trout age and size classes and prey resources that tolerate a wide r ange of temperatures expected in a changing climate (Hansen et al., 2015), instituting angling regulations (e.g., reduced creel limits, protected slot limits) that preserve trout populations during thermally stressful times, and monitoring precipitation le vels and associated thermal effects in runoff - dominated streams. Similarly, resource managers can enhance management system resilience by developing public 189 outreach programs that inform stakeholders about thermal habitat management, prepare them for realis tic stream fish community outcomes amid climate change (e.g., salmonid decline, centrarchid expansion; Pease & Paukert, 2014), and garner their support for resilience - based salmonid management. In conclusion, we developed a methodology for, and demonstra ted the advantages of, modeling effects of air temperature, groundwater input, and precipitation patterns on stream thermal regimes in an accurate, cost - effective, management - relevant manner. The modeling approach described herein allows fisheries and aqua tic resource professionals to forecast and respond to effects of climate change on populations of trout and other fish species in management - relevant ways. Methods used in this study are widely applicable throughout the range of brook trout, brown trout, a nd rainbow trout (and other species) because they only require readily available (or easily collectable) air/stream temperature and precipitation data and readily accommodate other thermal drivers (e.g., riparian/watershed land cover) . This stream temperat ure modeling approach is also flexible because it can be applied in streams with diverse air temperature and hydrological regimes (i.e., groundwater - dominated and runoff - dominated systems), thereby expanding the scope of previously developed groundwater mo dels (Snyder et al., 2015) and helping fill knowledge gaps regarding the effects of precipitation changes (e.g., thermal sensitivity, magnitude, timing) on stream temperature. Overall, precipitation - and groundwater - corrected stream temperature models are useful for reliable thermal forecasting and associated management efforts ranging from groundwater conservation to riparian habitat rehabilitation and public engagement. In turn, these activities will promote sustainable, resilience - based management of col dwater streams and their socio - ecologically important trout fisheries within and beyond Michigan in a changing climate. 190 Acknowledgments The lead author thanks Bruce Vondracek (emeritus USGS Minnesota Cooperative Fish and Wildlife Research Unit, University of Minnesota) for inspiring him to pursue a fisheries research career. We thank Jennifer Moore Myers (United States Forest Service Eastern Forest Environmental Threat Assessment Center) and Stacy Nelson and Ernie Hain (North Carolina State Unive rsity) for assisting with air temperature data acquisition and projection models. We thank Kyle Herreman and Wesley Daniel (Michigan State University [MSU]); Troy Zorn, Tracy Kolb, and Todd Wills (Michigan Department of Natural Resources); and Henry Quinla n (United States Fish and Wildlife Service) for assisting in procurement of environmental and trout population data for this study. Further, we acknowledge the Programme for Climate Model n Coupled Modelling for their helpful guidance regarding use of the WCRP CMIP3 multimodel data set. We especially wish to thank T. Hitt (United States Geological Survey) for thought - provoking ology of Stream Salmonids funding sources that made it possible to conduct the research leading to this paper, including the University Distinguished Fellowship (MSU), the MSU Graduate School, the MSU Department of Fisheries and Wildlife, the Robert C. Ball and Betty A. Ball Fisheries and Wildlife Fellowship (MSU), the Schrems West Michigan Chapter of Trout Unlimited Fellowship, the Red Cedar Fly Fishers Graduate Fellow ship, and the Fly Fishers International Conservation Scholarship. 191 APPENDI X 192 T ABLE 4.6 ( Appendix 4.1 ) . Mean d aily water temperatures in Michigan trout streams from late - May to late - October 2018. Abbreviations are as follows: AUS = Au Sable River, BLK = Black River, CED = Cedar River (Upper Peninsula), EBF = East Branch Fox River, MAN = Manistee River, PAN = Paint River, PIG = Pigeon River, PIN = Pine River (Southern Lower Peninsula), STJ = St. Joseph River, STU = Sturgeon River (Upper Penins ula), TAM = Tamarack River. Date AUS BLK CED EBF MAN PAN PIG PIN STJ STU TAM 5/26/2018 N/A N/A N/A N/A N/A N/A N/A 17.93 20.59 N/A 18.47 5/27/2018 N/A N/A N/A N/A N/A N/A N/A 18.10 21.36 N/A 18.71 5/28/2018 N/A N/A N/A N/A N/A N/A N/A 18.28 22.55 N/A 18.53 5/29/2018 N/A N/A N/A N/A N/A N/A N/A 18.75 23.11 N/A 19.29 5/30/2018 N/A N/A N/A N/A N/A N/A N/A 18.26 22.66 N/A 18.48 5/31/2018 N/A N/A N/A N/A N/A N/A N/A 19.03 22.45 N/A 20.10 6/1/2018 N/A N/A N/A N/A N/A N/A N/A 18.25 22.30 N/A 19.00 6/2/2018 15.87 13.06 N/A N/A 15.03 N/A 11.57 16.21 20.87 N/A 16.35 6/3/2018 14.48 11.95 N/A N/A 13.27 N/A 10.66 14.89 19.49 N/A 14.90 6/4/2018 14.76 12.49 N/A N/A 14.39 N/A 11.22 14.92 19.34 N/A 15.63 6/5/2018 13.22 11.14 N/A N/A 12.88 N/A 10.02 14.74 18.93 N/A 15.48 6/6/2018 13.24 11.67 N/A N/A 13.26 N/A 10.42 13.67 18.19 N/A 14.80 6/7/2018 14.88 12.97 N/A N/A 14.95 N/A 11.32 15.07 19.26 N/A 16.53 6/8/2018 15.37 13.21 N/A N/A 15.12 N/A 11.49 16.10 19.90 N/A 17.35 6/9/2018 15.36 13.53 N/A N/A 15.24 N/A 11.71 14.98 19.25 N/A 15.09 6/10/2018 15.44 13.62 N/A N/A 14.74 N/A 11.68 14.35 18.20 N/A 14.36 6/11/2018 15.54 13.36 N/A N/A 15.19 N/A 11.59 14.89 17.85 N/A 16.11 6/12/2018 16.26 14.16 N/A N/A 16.21 N/A 12.12 15.68 18.17 N/A 17.50 6/13/2018 16.94 14.77 N/A N/A 16.63 N/A 12.47 16.98 19.19 N/A 18.58 6/14/2018 16.23 14.24 N/A N/A 15.78 N/A 12.26 16.01 19.48 N/A 17.40 6/15/2018 14.64 13.20 N/A N/A 14.22 N/A 11.35 15.12 19.97 N/A 16.29 6/16/2018 15.27 13.69 N/A N/A 14.43 N/A 11.67 15.56 20.19 N/A 16.22 6/17/2018 17.34 15.35 N/A N/A 16.67 23.07 13.11 17.21 21.42 N/A 18.83 6/18/2018 17.65 15.54 17.38 13.88 16.92 22.77 13.09 18.56 23.25 22.18 19.98 6/19/2018 16.25 13.86 16.51 13.24 15.12 22.64 12.22 16.85 22.19 22.40 17.08 6/20/2018 16.96 14.28 16.27 13.16 16.27 22.72 12.54 15.73 19.40 22.90 16.25 6/21/2018 16.61 14.37 15.92 13.18 16.06 22.49 12.45 15.74 18.65 23.10 17.10 6/22/2018 14.99 12.61 14.70 12.54 14.68 22.15 11.27 15.01 18.28 23.21 15.87 6/23/2018 14.02 12.01 14.50 12.51 13.54 22.64 10.95 14.85 18.27 23.68 15.38 6/24/2018 15.50 12.58 14.35 12.35 15.37 22.74 11.73 15.25 18.97 21.82 15.70 6/25/2018 15.88 13.06 14.32 11.69 15.72 21.84 11.84 15.68 19.66 21.40 17.35 6/26/2018 15.23 13.00 14.72 12.08 15.21 21.57 11.63 14.69 19.77 22.02 16.57 6/27/2018 15.60 14.03 17.41 12.79 15.72 22.50 12.21 15.69 19.52 23.56 16.92 6/28/2018 17.26 15.15 18.98 13.65 17.06 24.10 13.04 16.65 20.28 25.17 18.07 193 TABLE 4.6 (Appendix 4.1) Date AUS BLK CED EBF MAN PAN PIG PIN STJ STU TAM 6/29/2018 18.15 15.97 20.90 14.03 17.94 24.59 13.56 17.94 22.02 25.81 19.92 6/30/2018 20.57 18.53 22.94 14.43 20.51 26.61 15.23 19.70 21.71 26.73 21.94 7/1/2018 20.86 19.01 20.48 14.37 20.18 25.77 15.46 20.27 21.58 26.19 22.27 7/2/2018 19.89 18.16 19.17 14.30 19.33 24.68 14.87 19.52 21.19 25.66 21.44 7/3/2018 18.98 17.20 20.10 14.23 18.57 25.31 14.31 18.35 20.99 26.44 20.67 7/4/2018 19.19 17.70 19.33 14.33 18.61 25.26 14.65 19.04 21.55 26.89 22.13 7/5/2018 19.76 17.65 19.80 14.78 18.86 25.70 15.12 19.79 21.47 27.21 22.32 7/6/2018 17.75 15.69 16.45 13.21 16.98 24.98 13.75 18.05 20.35 25.34 20.63 7/7/2018 16.80 15.00 16.93 12.89 16.50 25.16 13.23 16.19 19.52 25.22 19.37 7/8/2018 17.18 15.58 18.53 13.54 16.93 25.45 13.48 16.25 19.91 24.94 19.53 7/9/2018 17.82 16.19 20.56 13.80 17.47 26.34 13.87 17.26 20.32 25.92 19.93 7/10/2018 17.98 15.87 17.72 13.02 17.36 26.20 13.68 18.05 21.01 25.43 20.74 7/11/2018 17.02 15.25 16.78 12.69 16.70 25.08 13.25 16.91 20.42 24.97 20.31 7/12/2018 15.38 14.14 16.99 12.35 15.02 23.72 12.46 16.15 20.22 24.19 19.37 7/13/2018 17.36 15.60 18.03 13.01 17.26 24.09 13.73 17.68 20.92 24.58 20.67 7/14/2018 17.98 15.80 19.38 13.44 17.76 24.95 13.86 17.43 20.75 26.15 20.27 7/15/2018 18.36 16.35 19.88 14.04 18.35 25.82 14.10 17.59 21.21 26.71 19.79 7/16/2018 18.51 16.29 21.05 14.29 17.49 26.10 14.09 18.26 20.87 26.68 21.20 7/17/2018 16.64 14.74 15.82 12.98 15.97 24.54 13.33 16.78 20.41 23.74 19.37 7/18/2018 15.81 14.20 14.52 11.77 15.38 24.15 12.79 15.77 19.78 23.32 18.51 7/19/2018 16.11 14.56 16.75 12.72 15.87 23.81 12.82 15.60 19.95 24.10 18.54 7/20/2018 16.21 14.95 18.28 12.96 16.33 22.61 13.12 16.29 19.82 24.27 18.72 7/21/2018 15.96 14.41 17.02 13.10 15.79 21.27 13.39 16.26 19.86 22.96 18.15 7/22/2018 16.69 15.29 17.87 13.71 16.82 21.48 14.27 16.35 19.89 22.16 17.45 7/23/2018 16.56 14.41 18.43 12.97 16.29 22.88 13.28 15.72 19.75 22.49 17.00 7/24/2018 16.80 14.72 18.60 13.36 17.02 23.92 13.54 16.10 20.29 23.70 17.15 7/25/2018 17.15 14.81 17.48 13.27 17.12 23.41 13.52 16.64 20.30 23.59 18.46 7/26/2018 16.95 14.92 15.97 13.22 16.96 21.96 13.51 16.77 19.96 22.55 18.91 7/27/2018 15.91 13.61 15.00 12.12 15.41 20.56 12.75 15.57 19.57 20.88 17.62 7/28/2018 16.03 13.85 15.08 11.97 15.70 20.71 12.94 15.30 19.31 21.18 17.13 7/29/2018 15.33 13.64 15.68 12.25 15.52 22.26 12.82 15.42 19.52 22.14 17.91 7/30/2018 15.38 14.00 16.25 12.26 15.68 23.19 12.90 15.51 19.57 23.28 18.51 7/31/2018 16.10 14.19 17.31 12.75 15.61 23.64 12.97 15.54 19.48 23.76 18.64 8/1/2018 16.40 14.48 15.94 13.13 15.62 23.32 13.08 16.20 19.80 23.61 18.77 8/2/2018 16.06 14.09 13.57 12.38 15.39 20.91 13.03 16.34 20.09 21.44 18.78 8/3/2018 17.09 14.23 16.19 11.95 16.57 20.64 13.19 16.96 20.19 21.51 19.54 8/4/2018 17.35 14.42 16.21 12.34 16.56 20.89 13.38 16.71 20.52 22.44 19.96 8/5/2018 18.99 16.74 17.47 12.79 18.68 22.06 14.50 18.14 21.21 22.80 21.15 194 TABLE 4.6 (Appendix 4.1) Date AUS BLK CED EBF MAN PAN PIG PIN STJ STU TAM 8/6/2018 17.43 15.68 17.96 12.74 16.69 23.72 13.97 17.63 20.66 23.47 20.23 8/7/2018 17.26 14.94 17.19 12.79 16.56 23.57 13.56 17.04 20.52 23.04 18.85 8/8/2018 17.33 14.75 17.51 12.19 16.46 23.99 13.32 17.45 20.37 23.34 19.64 8/9/2018 16.90 15.26 18.32 13.28 16.88 24.86 13.63 17.31 19.96 24.44 20.11 8/10/2018 16.55 14.73 17.65 12.61 16.35 24.99 13.36 17.13 20.49 24.43 20.64 8/11/2018 16.96 14.91 18.57 12.94 17.01 25.38 13.54 16.57 20.40 25.01 20.15 8/12/2018 16.64 14.78 17.97 12.64 16.63 25.48 13.42 16.76 20.47 25.16 20.46 8/13/2018 16.81 14.63 18.82 12.76 16.66 25.77 13.26 17.07 20.39 25.41 20.47 8/14/2018 16.80 15.11 19.21 12.97 16.95 26.08 13.70 17.03 20.32 25.63 20.20 8/15/2018 16.95 15.39 18.57 13.00 16.70 25.09 13.68 17.23 20.47 24.44 20.47 8/16/2018 17.00 15.06 18.48 12.83 16.76 24.53 13.63 16.87 20.22 24.71 20.42 8/17/2018 17.06 15.08 19.21 12.93 16.50 24.29 13.56 16.14 20.09 24.54 18.98 8/18/2018 16.37 14.36 17.69 12.68 15.98 24.28 13.34 16.15 20.54 23.45 17.88 8/19/2018 16.82 14.23 16.33 11.71 16.19 23.57 13.20 16.73 20.15 23.01 19.22 8/20/2018 16.12 14.44 17.37 11.87 15.83 22.89 13.22 16.67 19.86 23.31 19.30 8/21/2018 14.98 13.33 16.47 11.82 14.46 21.70 12.72 17.27 20.17 21.31 18.73 8/22/2018 15.34 13.65 15.02 11.42 15.22 20.42 13.14 16.58 19.39 20.45 17.90 8/23/2018 14.80 13.57 16.01 11.35 14.81 20.82 12.87 15.37 19.19 21.05 17.53 8/24/2018 14.43 13.86 14.90 11.78 14.51 20.51 12.54 15.06 19.20 20.58 17.33 8/25/2018 14.29 13.89 16.10 11.64 14.21 20.63 12.80 15.83 20.08 20.69 17.32 8/26/2018 16.18 14.44 16.53 11.76 15.79 21.02 13.35 16.90 20.86 21.19 18.69 8/27/2018 16.44 14.62 18.47 12.45 15.98 21.61 13.44 19.00 21.42 22.21 20.08 8/28/2018 17.50 15.61 17.81 12.93 16.76 21.35 14.45 19.90 21.13 21.67 20.92 8/29/2018 17.14 15.27 15.68 11.95 16.14 19.69 14.32 19.44 20.24 19.53 19.78 8/30/2018 14.99 13.41 13.49 10.59 14.24 18.58 12.52 16.50 19.05 18.68 17.02 8/31/2018 14.56 13.16 14.64 10.86 14.42 18.68 12.49 15.88 19.67 18.67 17.17 9/1/2018 15.35 14.27 17.81 11.93 15.11 20.27 13.03 16.60 20.70 19.93 17.24 9/2/2018 16.62 14.69 17.97 12.40 16.05 21.49 13.58 17.73 21.11 20.84 19.05 9/3/2018 17.41 15.18 17.75 12.18 16.64 21.60 13.90 19.66 20.69 21.62 20.11 9/4/2018 17.80 15.58 17.18 12.07 16.92 21.66 14.76 20.08 21.03 21.46 20.57 9/5/2018 18.28 16.62 17.81 15.18 17.00 21.27 14.76 20.47 21.25 20.97 20.43 9/6/2018 16.71 15.54 15.62 13.56 15.74 20.06 13.46 18.89 19.81 20.08 18.77 9/7/2018 14.21 12.41 13.92 11.53 13.40 19.64 11.74 16.00 19.22 19.84 15.85 9/8/2018 13.41 11.77 13.22 11.33 13.09 18.82 11.51 14.94 18.69 19.08 15.06 9/9/2018 12.58 10.90 12.28 10.31 12.53 18.06 10.85 13.92 18.20 18.76 14.46 9/10/2018 12.70 10.97 11.82 10.13 12.76 18.10 10.89 13.99 18.42 18.69 14.99 9/11/2018 13.69 12.02 13.23 10.82 13.70 19.15 11.62 14.34 18.77 19.23 15.50 9/12/2018 14.31 13.10 15.32 11.55 14.15 19.77 12.09 14.35 18.78 20.27 16.13 195 TABLE 4.6 (Appendix 4.1) Date AUS BLK CED EBF MAN PAN PIG PIN STJ STU TAM 9/13/2018 14.54 13.01 15.75 11.66 14.25 20.19 12.16 14.38 19.38 20.79 16.18 9/14/2018 15.70 13.93 15.85 11.97 15.53 20.99 12.84 14.74 20.00 21.47 16.83 9/15/2018 15.22 13.44 18.36 12.31 14.84 22.35 12.59 15.03 19.94 22.28 16.58 9/16/2018 15.19 13.75 18.30 12.44 15.10 22.16 12.78 15.02 20.34 22.90 17.55 9/17/2018 15.63 14.82 18.06 12.50 15.49 22.25 13.27 15.19 20.68 22.79 18.26 9/18/2018 15.31 13.92 12.55 11.25 14.92 20.59 12.80 15.41 20.38 19.64 17.68 9/19/2018 13.65 12.69 11.49 10.49 13.19 18.82 11.92 15.21 19.90 18.54 16.21 9/20/2018 13.23 12.34 14.13 10.73 12.78 17.67 11.92 14.91 20.33 18.28 16.43 9/21/2018 14.59 13.88 15.10 11.68 14.39 16.75 12.97 15.15 20.34 17.98 17.86 9/22/2018 12.16 11.10 11.59 9.43 11.71 15.04 10.82 14.75 17.76 15.85 14.39 9/23/2018 11.49 10.57 11.47 9.97 11.34 14.84 10.62 13.62 17.98 16.21 13.70 9/24/2018 12.62 11.78 13.21 10.06 12.75 14.97 11.46 13.39 18.42 16.30 14.52 9/25/2018 14.39 13.38 13.99 10.40 14.54 15.36 12.49 13.84 20.04 16.16 16.84 9/26/2018 12.99 12.01 11.98 9.61 12.68 14.12 11.67 14.34 18.54 14.91 15.93 9/27/2018 11.19 10.51 10.23 8.40 10.99 13.17 10.54 13.59 17.31 13.78 13.07 9/28/2018 11.10 10.82 10.42 8.96 11.18 12.21 10.69 13.24 17.85 13.47 13.23 9/29/2018 9.76 9.54 8.44 7.98 9.60 10.64 9.67 12.67 17.33 12.03 11.31 9/30/2018 9.13 8.91 7.87 7.51 9.06 10.28 9.19 11.96 18.17 11.50 10.65 10/1/2018 9.18 9.14 8.23 7.82 9.49 10.21 9.42 10.84 18.70 11.25 10.53 10/2/2018 9.81 9.81 9.38 8.25 9.93 10.14 9.84 11.48 19.12 11.69 11.85 10/3/2018 10.59 10.55 11.03 9.03 10.62 10.67 10.38 13.03 19.80 12.34 14.16 10/4/2018 11.93 12.22 11.50 10.08 12.00 10.83 11.43 14.61 19.72 11.90 15.65 10/5/2018 8.70 8.73 8.34 7.73 8.45 9.60 8.86 10.67 17.34 10.46 10.92 10/6/2018 9.90 9.79 8.92 8.46 9.74 9.33 9.78 12.03 19.33 10.84 12.86 10/7/2018 9.58 9.45 8.61 8.19 9.18 8.91 9.50 11.97 18.57 10.05 12.44 10/8/2018 10.38 10.32 9.03 8.15 10.36 8.83 10.15 12.47 19.03 10.03 13.52 10/9/2018 12.96 12.92 9.82 9.09 13.00 9.01 11.73 14.48 N/A 10.16 15.81 10/10/2018 13.73 13.35 9.16 9.16 13.31 8.50 11.99 15.39 N/A 9.83 15.97 10/11/2018 12.35 12.03 8.97 9.57 11.47 8.00 10.99 14.12 N/A 10.03 13.99 10/12/2018 8.40 8.60 6.47 7.60 8.02 6.03 8.62 10.12 N/A 8.56 10.40 10/13/2018 8.01 7.88 5.10 6.95 7.86 4.92 8.17 8.70 N/A 7.97 9.15 10/14/2018 8.35 8.36 5.95 7.43 8.27 5.09 8.66 N/A N/A 7.83 N/A 10/15/2018 8.18 8.31 4.72 6.36 7.91 4.33 8.36 N/A N/A 6.96 N/A 10/16/2018 7.26 7.31 4.59 6.16 7.38 3.87 7.78 N/A N/A 6.57 N/A 10/17/2018 6.95 6.96 4.68 5.91 7.07 3.67 7.38 N/A N/A 5.88 N/A 10/18/2018 6.16 6.15 3.40 5.32 6.43 3.60 6.90 N/A N/A 5.41 N/A 10/19/2018 7.07 7.41 5.17 6.26 7.26 5.00 7.98 N/A N/A 5.97 N/A 10/20/2018 6.89 7.21 5.02 6.23 6.90 4.61 7.60 N/A N/A 5.45 N/A 196 TABLE 4.6 (Appendix 4.1) Date AUS BLK CED EBF MAN PAN PIG PIN STJ STU TAM 10/21/2018 5.74 5.96 3.99 5.47 5.61 3.77 6.78 N/A N/A 4.64 N/A 10/22/2018 5.83 6.12 3.43 5.14 6.15 3.91 6.72 N/A N/A 4.76 N/A 10/23/2018 6.56 6.64 4.52 6.15 6.75 4.66 7.26 N/A N/A 5.34 N/A 10/24/2018 6.54 6.31 4.20 5.82 6.63 4.74 7.10 N/A N/A 5.12 N/A 10/25/2018 5.11 5.07 4.03 5.09 4.91 5.28 6.01 N/A N/A 4.51 N/A 10/26/2018 N/A N/A 5.05 N/A N/A N/A N/A N/A N/A N/A N/A 197 LITERATURE CITED 198 LITERATURE CITED Almodóvar, A., G. G. Nicola, D. Ayllón & B. Elvira, 2012. Global warming threatens the persistence of Mediterranean brown trout. Global Change Biology 18: 1549 1560. Criterion. Journal of Wildlife Management 74: 1175 1178. Baldwin, N. S., 1957. Food consumption and growth of brook trout at different temperatures. Transactions of the American Fisheries Society 86: 323 328. Burnham, K. P. & D. R. Anderson , 2002. Model selection and multimodel inference: a practical information - theoretic approach. Springer - Verlag, New York, New York. Carlson, A. K., W. W. Taylor, K. M. Schlee, T. G. Zorn & D. M. Infante, 2016. Projected impacts of climate change on stream salmonids with implications for resilience - based management. Ecology of Freshwater Fish 26: 190 204. Carlson, A. K., W. W. Taylor, K. M. Hartikainen, D. M. Infante, T. Douglas Beard, Jr. & A. J. Lynch, 2017. Comparing stream - specific to generalized temperature models to guide salmonid management in a changing climate. Reviews in Fish Biology and Fisheries 27: 443 462. Cherkauer, K. A. & T. Sinha , 2010. Hydrologic impacts of pr ojected future climate change in the Lake Michigan region. Journal of Great Lakes Research 36: 33 50. Constantz, J., 1998. Interaction between stream temperature, streamflow, and groundwater exchanges in Alpine streams. Water Resources Research 34 : 1609 1615. Cooper, A. R., D. M. Infante, K. E. Wehrly, L. Wang & T. O. Brenden, 2016. Identifying indicators and quantifying large - scale effects of dams on fishes. Ecological Indicators 61: 646 657. e on the basis of streamflow hydrograph. Facta Universitatis 10: 301 314. Dunham, J., G. Chandler, B. Rieman, and D. Martin, 2005. Measuring stream temperature with - GTR - 150WWW. USDA Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. Enviro - weather Automated Weather Station Network (EAWSN), 2018. Michigan State University. Accessed 13 June 2018, https://mawn.geo.msu.edu/. Ebersole, J. L., W. J. Liss & C. A. Frisse ll, 2003. Cold water patches in warm streams: physicochemical characteristics and the influence of shading. Journal of the American Water Resources Association 39 : 355 368. Fry, F. E. J., J. S. Hart & K. F. Walker, 1946. Lethal temperature relations for a sample of young speckled trout, Salvelinus fontinalis Vol. 54. The University of Toronto Press, Toronto, Ontario. 199 Godby, N. A., Jr., E. S. Rutherford & D. M. Mason, 2007. Diet, feeding rate, growth, mortality, and production of juvenile steelhead in a Lak e Michigan tributary. North American Journal of Fisheries Management 27: 578 592. Hansen, G. J. A., J. W. Gaeta, J. W. Hansen & S. R. Carpenter, 2015. Learning to manage and managing to learn: sustaining freshwater recreational fisheries in a changing env ironment. Fisheries 40: 56 64. Hay J., J. Hayes, and R. Young. 2006. Water quality guidelines to protect trout fishery values. Prepared for Horizons Regional Council. Cawthorn Report Number 1025, 17 pp. Hayes, D. B., W. W. Taylor, M. Drake, S. Marod & G . Whelan, 1998. The value of headwaters to brook trout ( Salvelinus fontinalis ) in the Ford River, Michigan, USA. In Haigh, M. J., J. Krecek, G. S. Rajwar & M. P. Kilmartin (eds), Headwaters: Water Resources and Soil Conservation. Oxford and IBH Publishing Co., New Delhi, India: 75 185. Hershkovitz, Y., V. Dahm, A. W. Lorenz & D. Hering, 2015. A multi - trait approach for the identification and protection of European freshwater species that are potentially vulnerable to the impacts of climate change. Ecologic al Indicators 50: 150 160. Holling, C.S., 1973. Resilience and stability of ecological systems. Annual Review of Ecology and Systematics 4: 1 23. IPCC (Intergovernmental Panel on Climate Change) , 2007. Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II, and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Intergovernmental Panel on Climate Change, Geneva, Switzerland, 104 pp. I saak, D. J., S. Wollrab, D. Horan & G. Chandler, 2012. Climate change effects on streamand river temperatures across the northwestern US from 1980 2009 and implications for salmonid fishes. Climatic Change 113: 499 524. Kanno, Y., B. H. Letcher, A. L. Ro factors affecting brook trout occurrence in headwater stream segments. Transactions of the American Fisheries Society 144: 373 382. at native trout its history, biology, and angling possibilities (revised edition). Skyhorse Publishing, New York, New York. Kelleher, C., T. Wagener, M. Gooseff, B. McGlynn, K. McGuire & L. Marshall, 2012. Investigating controls on the thermal sensitivi ty of Pennsylvania streams. Hydrological Processes 26: 771 785. Kurylyk, B. L., S. P. A. Bourque, and K. T. B. MacQuarrie, 2013. Potential surface temperature and shallow groundwater temperature responses to climate change: an example from a 20 0 small foreste d catchment in east - central New Brunswick (Canada). Hydrology and Earth Systems Sciences 17: 2701 2716. Leach, J. A. & R. D. Moore, 2011. Stream temperature dynamics in two hydrogeomorphically distinct reaches. Hydrological Processes 25: 679 690. Loomis, J., P. Kent, L. Strange, K. Fausch & A. Covich, 2000. Measuring the total economic value of restoring ecosystem services in an impaired river basin: results from a contingent valuation survey. Ecological Economics 33: 103 117. Lyons, J., J. S. Stewart & M. Mitro, M. 2010. Predicted effects of climate warming on the distribution of 50 stream fishes in Wisconsin, U.S.A. Journal of Fish Biology 77: 1867 1898. Maurer, E. P., L. Brekke, T. Pruitt & P. B. Duffy , 2007. Fine - resolution climate projections enhanc e regional climate change impact studies. Eos Transactions, American Geophysical Union 88: 504 504. McKergow, L., S. Parkyn, R. Collins & P. Pattinson, 2005. Small headwater streams of the Auckland Region. Volume 2: Hydrology and water quality. Auckland Regional Council Technical Publication No. 312: 1 67. Menberg, K., Blum, P., Kurylyk, B.L. & Bayer, P. 2014. Observed groundwater tempera ture response to recent climate change. Hydrology and Earth System Sciences 18:4453 4466. Merriam, E. R., R. Fernandez, J. T. Petty & N. Zegre, 2017. Can brook trout survive climate change in large rivers? If it rains. Science of the Total Environment 607 608: 1225 1236. Neff, B. D., S. M. Day, A. R. Piggott & L. M. Fuller, 2005. Base flow in the Great Lakes basin. U.S. Geological Survey Scientific Investigations Report 2005 5217, 23 pp. Onset Computer Corporation. 2009. HOBO U22 water temp pro v2: 10366 - C. Onset Computer Corporation, Bourne, Massachusetts, USA. Parry, M., O. Canziani, J. Palutikof, P. van der Linden & C. Hanson, 2007. Climate change 2007: impacts, adaptation and vulnerability. International Panel on Climate C hange Fourth Assessment Report. Paukert, C. P., B. A. Glazer, G. J. A. Hansen, B. J. Irwin, P. C. Jacobsen, J. L. Kershner, B. J. Shuter, J. E. Whitney & A. J. Lynch, 2016. Adapting inland fisheries management to a changing climate. Fisheries 41: 374 384. Pease, A.A. & C. P. Paukert, 2014. Potential impacts of climate change on growth and prey consumption of stream - dwelling smallmouth bass in the central United States. Ecology of Freshwater Fish 23: 336 346. Peterson, E. E., and J. M. Ver Hoef. 2010. A mixe d - model moving - average approach to geostatistical modeling in stream networks. Ecology 91: 644 651. 201 Pilgrim, J. M., X. Fang & H. G. Stefan, 1998. Stream temperature correlations with air temperature in Minnesota: implications for climate warming. Journal o f the American Water Resources Association 34: 1109 1121. Primack, A. G. B., 2000. Simulation of climate - change effects on riparian vegetation in the Pere Marquette River, Michigan. Wetlands 20: 538 547. Raleigh R.F. 1982a. Habitat Suitability Index Models : Brook Trout. Washington, D.C.: U.S. Fish and Wildlife Service, Biological Report Number 82, 42 pp. Raleigh R.F. 1982b. Habitat Suitability Index Models: Rainbow Trout. Washington, D.C.: U.S. Fish and Wildlife Service, Biological Report Number 82, 76 pp . Raleigh R.F., Zuckerman L.D. & Nelson P.C. 1986. Habitat Suitability Index Models and Instream Flow Suitability Curves: Brown Trout. Washington, D.C.: U.S. Fish and Wildlife Service, Biological Report Number 82, 65 pp . RStudio. 2015. Boston (MA): RStudio, Inc; [accessed 13 April 2018]. http://www.rstudio.com/. Santiago, J. M., D. G. de Jalón, C. Alonso, J. Solana, J. Ribalaygua, J. Pórtoles & R. Monjo, 2015. Brown trout thermal niche and climate change: expected changes in the distribution of cold water fish in central Spain. Ecohydrology 9: 514 528. Siitari, K. J., W. W. Taylor, S. A. C. Nelson & K. E. Weaver, 2 011. The influence of land cover composition and groundwater on thermal habitat availability for brook charr ( Salvelinus fontinalis ) populations in the United States of America. Ecology of Freshwater Fish 20: 431 437. Snyder, C. D., N. P. Hitt & J. A. Young, 2015. Accounting for groundwater in stream fish thermal habitat responses to climate change. Ecological Applications 25 : 1397 1419 Steen, P. J., M. J. Wiley & J. S. Schaeffer, 2006. Predicting future changes in Muskegon River watershed game fish distributions under future land cover alteration and climate change scenarios. Transactions of the American Fisheries Society 139: 396 412. Stoner, A. M. K., K. Hayhoe, X. H. Yang & D. J. Wuebbles, 2013. An asynchronous regional regression model for statistical downscaling of daily climate variables. International Journal of Climatology 33: 2473 2494. Tague, C., M. Farrell, G. Grant, S. Lewis & S. Rey, 2007. Hydrolgeologic controls on summer stream temperatures in the McKenzie River basin, Oregon. Hydrological Processes 21: 3288 3300. Thomas, S. M., S. W. Griffiths & S. J. Ormerod, 2015. Beyond cool: adapting upland streams for climate change u sing riparian woodlands. Global Change Biology 22: 310 324. Tyler, J. A. & E. S. Rutherford, 2007. River restoration effects on steelhead populations in the Manistee River, Michigan: Analysis using an individual - based model. Transactions of the American F isheries Society 136: 1654 1673. 202 Waco, K. E. & W. W. Taylor, 2010. The influence of groundwater withdrawal and land use changes on brook charr ( Salvelinus fontinalis ) thermal habitat in two coldwater tributaries in Michigan, USA. Hydrobiologia 650: 101 116 . Webb, B. W., D. M. Hannah, R. D. Moore, L. E. Brown & F. Nobilis, 2008. Recent advances in stream and river temperature research. Hydrological Processes 22 , 902 918. Westhoff, J. T. & C. P. Paukert, 2014. Climate change simulations predict altered bioti c response in a thermally heterogeneous stream system. PLoS One 9: e111438. Westhoff, M. C., M. N. Gooseff, T. A. Bogaard, and H. H. G. Savenije. 2011. Quantifying hyporheic exchange at high spatial resolution using natural temperature variations along a first - order stream. Water Resources Research 47: W10508. Woodward, G., D. M. Perkins & L. E. Brown, 2010. Climate change and freshwater ecosystems: impacts across multiple levels of organization. Philosophical Transactions of the Royal Society B: Biologica l Sciences 365: 2093 2106. Wurtsbaugh, W. A., and G. E. Davis. 1977. Effects of fish size and ration level on the growth and food conversion efficiency of rainbow trout, Salmo gairdneri Richardson. Journal of Fish Biology 11:99 104. Zorn, T. G., P. W. Seelbach & M. J. Wiley, 2011. Developing user - friendly habitat suitability tools from regional stream fish survey data. North American Journal of Fisheries Management 31: 41 55. Zorn, T. G., P. W. Seelbach & E. S. Rutherford, 2012. A regional - scale h abitat suitability model to assess the effects of flow reduction on fish assemblages in Michigan streams. Journal of the American Water Re sources Association 48: 871 89 203 SYNTHESIS 204 streams and their trout populations (Chapter 1). However, these effects will likely vary according to stream - specific groundwater regimes, precipitation patterns, and riparian and watershed land cover. I found that thermal habitats in groundwater - dominate d streams (baseflow index > 0.60; managed using a generalized (i.e., region - specific) approach that encompasses regional groundwater dynamics. However, in surface ru noff - groundwater is not a primary hydrological driver, a more resource - intensive stream - specific approach is needed , particularly in streams containing high - priority fisheries resources (e.g., trophy individuals, endang ered species; Chapter 2). Regardless of stream hydrology, decision - support tools information systems that organizations and individuals use to make science - based fisheries management decisions facilitate trout management decision - making by systematical ly integrating thermal, hydrological, and biological (i.e., fish abundance and species composition) information about trout streams across local and regional scales (Carlson et al. 2018). For instance, I developed a decision - support tool that synthesizes f isheries manager - defined decision - making criteria (e.g., current and future stream temperature, trout abundance, groundwater input) and thereby enables fisheries professionals to prioritize streams according to their importance for trout management (Chapte r 3). Ultimately, decision - support tools promote resilience - based salmonid management by integrating thermal, hydrological, and biological factors in streams and forecasting future biotic and abiotic conditions, allowing fisheries professionals to make informed decisions to increase stream thermal resilience and social - ecologica l resilience (e.g., groundwater conservation, riparian habitat protection/rehabilitation, public outreach efforts). Likewise, precipitation - and groundwater - corrected stream temperature 205 models provide an effective method to accurately forecast future therm al habitat conditions for trout growth and survival, enabling fisheries professionals to enhance the thermal resilience of trout streams and the social - ecological resilience of trout management programs. For instance, fisheries professionals can promote re silience - based salmonid management by forming public private partnerships to rehabilitate riparian vegetation and associated shading (Blann et al. 2002), protect watershed land cover types with high groundwater recharge (e.g., grasslands; Waco & Taylor 201 0; Siitari et al. 2011), and inform fisheries stakeholders about realistic expectations for stream fisheries amid climate change (e.g., salmonid decline, centrarchid expansion; Pease & Paukert 2014). Climate change will affect Michigan stream trout As c oldwater fishes with limited tolerance of warm water conditions, stream salmonids are particularly susceptible to short - and long - term changes in climate that increase stream temperatures (Thomas et al. 2015). In particular, trout growth, reproduction, and survival are regulated by temperature and thus are likely to be affected by warming stream thermal regimes in a changing climate. Hence, it is important for fisheries professionals to predict how trout will be affected by warming air and water temperature s and other climatic changes (e.g., more variable and intense precipitation, increased groundwater temperature and reduced inflow; Parry et al. 2007; Stoner et al. 2013; Menberg et al. 2014) and use this information to establish resilience - based salmonid m anagement programs. As such, I measured water temperatures in a representative sample of coldwater streams throughout Michigan and developed stream - specific temperature models to forecast thermal habitat suitability in a changing climate from 2006 to 2056 (Chapter 1). For this study, I selected 30 trout streams throughout Michigan to represent the 206 variability in air and water temperatures, hydrological conditions (i.e., groundwater to surface runoff dominance), and trout species composition (i.e., combinati ons of Brook Trout, Brown Trout, and Rainbow Trout) experienced by Michigan stream trout. Stream temperatures were projected to increase between 0.19 and 5.94 °C by 2056 due to predicted air temperature warming, with reduction in growth and survival of Bro ok Trout, Brown Trout, and Rainbow Trout (Chapter 1). Thermal habitat impairment for growth was predicted to occur least frequently in Rainbow Trout streams as this species has a wider temperature range for growth (12.0 22.5 °C) compared to Brook Trout (11 .0 20.5 °C) and Brown Trout (12.0 20.0 °C; Wurtsbaugh & Davis 1977). Overall, projected increases in Michigan stream temperatures will result in thermal habitat conditions less conducive for trout growth and survival during the warmest period of the year, particularly in surface runoff - dominated streams with minimal groundwater - driven thermal buffering capacity, suggesting a need for resilience - based salmonid management (explored further below). Stream hydrology influences the effectiveness of thermal habitat modeling approaches Despite the utility of stream - specific regression models for projecting future stream temperatures with high accuracy (i.e., exactness of temperature prediction; Chapter 1), stream - specific models are resource - intensive in term s of the time, money, and personnel required for data collection, management, and modeling. As such, I compared stream - specific and generalized (i.e., region - specific) temperature models in Michigan trout streams to evaluate the potential tradeoff between model accuracy and efficiency (i.e. applicability at management - relevant spatial extents; Chapter 2). For this study, I selected 52 streams throughout Michigan that represented the thermal, hydrological, and biological (i.e., fish abundance and species 207 com position) conditions experienced by Michigan stream trout. Generalized models were less accurate than stream - specific models in projecting stream temperatures, particularly in runoff - dominated streams, where they had 54% overall accuracy. However, generali zed models accurately predicted thermal habitat suitability (82% overall accuracy) in groundwater - dominated streams containing Brook Trout (80% accuracy), Brown Trout (89% accuracy), and Rainbow Trout (75% accuracy; Chapter 2). Hence, amid climate change a nd constraints in resource availability, fisheries professionals can use generalized models to forecast thermal conditions in groundwater - dominated streams, reserving resource - intensive stream - specific models for surface runoff - dominated systems containing high - priority fisheries resources (e.g. trophy individuals, endangered species). Decision - support tools help integrate climate change pr ediction s into trout management Fisheries professionals face the difficult task of synthesizing diverse types of information (e.g., thermal, hydrological, biological) to make effective trout management decisions. To help facilitate this data integration process, I collaborated with fishe ries professionals from the MDNR and the USGS to produce a decision - support tool for stream trout management within and beyond Michigan in a changing climate (Chapter 3). The tool ranked trout streams by their management priority based on manager - defined d ecision - making criteria, including current and projected 2056 stream temperature, groundwater input, trout relative abundance, and watershed and riparian land cover. The decision - support tool highlighted a spectrum of stream trout fisheries, ranging from r ecreationally significant systems whose future warming is likely to reduce the importance of trout fisheries (e.g., Muskegon River) to lesser - known streams projected to become more thermally suitable for trout and thus significant for trout management 208 (e.g ., Davenport Creek; Chapter 3). Overall, the decision - support tool promotes resilience - based salmonid management by providing fisheries professionals with a knowledge base to integrate diverse data (e.g., biological, hydrological, thermal) and forecast fut ure stream conditions within and beyond Michigan, ultimately ranking streams according to their priority for management interventions to increase thermal and social - ecological resilience (e.g., groundwater conservation, riparian habitat protection/rehabili tation, public outreach efforts). Monitoring and modeling precipitation and groundwater are crit ical for resilience - based salmonid management Incorporating precipitation and groundwater dynamics (e.g., amount, temperature) into stream temperature monitor ing and modeling programs is important for resilience - based salmonid management in a changing clima te because these factors have significant effect s on stream thermal regimes. However, precipitation and groundwater are rarely included together in stream te mperature models. Until recently, stream temperature models have largely ignored variability in precipitation and groundwater dynamics (e.g., magnitude, timing, temperature) among streams and stream reaches (Snyder et al. 2015). This inhibits the accuracy of stream temperature forecasting and thus the efficacy of trout management in a changing climate. To help fill this knowledge gap, I developed an approach for predicting the effects of changes in air temperature, precipitation, and groundwater on coldwater stream temperatures and Brook Trout, Brown Trout, and Rainbow Trout growth and survival in Michigan ( Chapter 4) . Precipitation - and groundwater - corrected stream temperature models (mean R 2 = 0.83 ) were more accurate than conventional air - stream t emperature models (mean R 2 = 0.58) in predicting stream temperatures to the year 2056 . Projected stream warming intensified in proportion to simulated 209 air temperature warming and was most extreme in surface runoff - dominated streams, given their limited gro undwater - driven thermal buffering ( Carlson et al. in review , Chapter 4). However, groundwater - dominated streams will not invariably provide coldwater habitats needed by Brook Trout, Brown Trout, and Rainbow Trout if groundwater temperatures increase or gro undwater inputs decline due to reduced precipitation or groundwater withdrawal by humans for other activities (e.g., irrigation, municipal use). Fisheries professionals throughout Michigan, the United States, and the world can use precipitation - and ground water - corrected stream temperature models to better forecast effects of climate change on trout growth and survival and ultimately develop resilience - based salmonid management programs, thereby enhancing the sustainability of trout fisheries now and in the years to come. 210 LITERATURE CITED 211 LITERATURE CITED Blann, K., J. F. Nerbonne, and B. Vondracek. 2002. Relationship of riparian buffer type to water temperature in the driftless area ecoregion of Minnesota. North American Journal of Fisheries Management 22:441 451. Carlson, A. K., W. W. Taylor, K. M. Schlee, T. G. Zorn, and D. M. Infante. 2017a (Chapter 1) . Projected impacts of climate change on stream salmonids with implications for resilience - based management. Ecology of Freshw ater Fish 26: 190 204. Carlson, A. K., W. W. Taylor, K. M. Hartikainen, D. M. Infante, T. Do uglas Beard, Jr., and A. J. Lynch. 2017 b (Chapter 2) . Comparing stream - specific to generalized temperature models to guide salmonid management in a changing climate. Reviews in Fish Biology and Fisheries 27: 443 462. Carlson, A. K., W. W. Taylor, Z. Basher, T. Do uglas Beard, Jr., and D. M. Infante. 2018 (Chapter 3). Science to a ction: d ecision - support to advance stream t rout m anagement in a c hanging c limate . P roceedings of the Wild Trout XII Symposium 12:85 92. Carlson, A. K., W. W. Taylor, and D. M. Infante. In rev iew (Chapter 4). Developing precipitation - and groundwater - corrected stream temperature models to improve management of brook charr amid climate cha nge . hydrograph. Facta Univ 10:301 314. McKergow, L., S. Parkyn, R. Collins, and P. Pattinson 2005. Small headwater streams of the Auckland Region. Volume 2: Hydrology and water quality. Auckland Regional Council Technical Publication No. 312. pp. 1 67. Menberg, K., P. Blum , B. L. Kurylyk, and P. Bayer. 2014. Observed groundwater tempera ture response to recent climate change. Hydrolog y and Earth System Sciences 18: 4453 4466. Parry, M., O. Canziani , J. Pal utikof, P. dervan Linden, and C. Hanson. 2007. Climate change 2007: impacts, adaptation and vulnerability. Intern ational Panel on Climate Change Fourth Assessment Report. Pease, A. A., and C. P. Paukert. 2014. P otential impacts of climate change on growth and prey consumption of stream - dwelling smallmouth bass in the central United States. Ecology of Freshwater Fish 23:336 346. Siitari, K. J., W. W. Taylor, S. A. C. Nelson, and K. E. Weaver. 2011. The influence of land cover composition and groundwater on thermal habitat availability for brook charr ( Salvelinus fontinalis ) populations in the United States of America. Ecology of Freshwater Fish 20:431 437. Snyder, C. D., N. P. Hitt, and J. A. Y oung. 2015. Accounti ng for groundwater in stream fish therm al habitat responses to climate cha nge. Ecological Applications 5: 1397 1419. 212 Stoner, A. M. K., K. Hayhoe , X. H. Yang , and D. J. Wuebbles. 2013. An asynchronous regional r egression model for statistical downscaling of daily c limate variables. International Journal of Climatology 33: 2473 2494. T homas, S. M., S. W. Griffiths, and S. J. Ormerod, 2015. Beyond cool: adapting upland streams for climate change using riparian woodlands. Global Change Biology 22: 310 3 4. Waco, K. E., and W. W. Taylor. 2010. The influence of groundwater withdrawal and land use changes on brook charr ( Salvelinus fontinalis ) thermal habitat in two coldwater tributaries in Michigan, USA. Hydrobiologia 650:101 116. Wurtsbaugh, W. A., and G. E. Davis. 1977. Effects of fish size and ration level on the growth and food conversion efficiency of rainbow trout, Salmo gairdneri Richardson. Journal of Fish Biology 11: 99 104.