Supplementary Information This document contains supporting material for the manuscript: Quantifying Variety-specific Heat Resistance and the Potential for Adaptation under Warming Supplementary Tables S1-S8 Supplementary Figures S1-S25 Table S1. Characteristics of study sites in Kansas: location and observations included in sample Site 38.86°N/99.27°W Number of Years 25 Min/Max of Years 1985/2013 Number of Varieties 173 Number of Clusters 30 37.94°N/100.72°W 23 1985/2013 156 30 624 Abbreviation Latitude/longitude Hays, Ellis County ELD Garden City, Finney County FND Observations 717 Ottawa, Franklin County FRD 38.83°N/94.89°W 23 1985/2013 111 27 432 Tribune, Greeley County GRD 38.07°N/102.68°W 23 1986/2013 169 30 685 Hesston, Harvey County HVD 38.05°N/97.29°W 25 1985/2009 106 27 527 Parsons, Labette County LBD 37.30°N/95.5°W 27 1985/2013 114 27 498 Manhattan, Riley County RLD 39.16°N/96.67°W 17 1985/2011 102 27 359 Hutchinson, Reno County RND 38.08°N/97.87°W 20 1989/2012 108 27 433 Belleville, Republic County RPD 39.55°N/97.65°W 23 1985/2009 128 28 612 St. John, Stafford County STD 37.70°N/98.75°W 15 1985/2004 107 28 390 Colby, Thomas County THD 39.42°N/101.01°W 25 1985/2013 175 30 743 -- -- 29 1985/2013 197 30 6,020 Totals Table S2. Characteristics of study sites in Kansas: means of yield, temperature, and precipitation Tmin(°C) Site Tmax(°C) Cumulative Precip (in) Yield (bu/acre) Fall Winter Spring Fall Winter Spring Fall Winter Spring ELD 56.7 5.2 -7.1 4.6 20.2 7.2 19.9 4.1 1.4 5.9 FND 37.3 4.8 -7.1 3.9 20.4 8.1 19.7 3.3 1.3 5.1 FRD 50.5 7.0 -5.4 6.4 20.2 6.3 19.3 9.3 4.0 10.1 GRD 44.0 3.0 -8.1 2.2 19.4 7.2 18.9 2.7 1.0 3.6 HVD 40.8 7.3 -5.1 6.5 20.2 6.7 19.3 6.4 2.5 9.0 LBD 47.1 8.0 -3.7 7.8 20.7 7.8 20.0 8.1 4.4 10.6 RLD 51.3 6.8 -6.1 6.2 19.6 6.0 19.5 7.2 2.3 9.1 RND 44.5 6.7 -5.1 6.2 20.7 7.8 19.6 4.8 2.2 7.4 RPD 65.3 4.9 -7.7 4.5 19.1 5.3 18.4 4.9 1.9 8.3 STD 48.4 6.5 -5.4 6.2 19.9 7.5 19.6 4.9 2.4 7.8 THD 57.9 3.4 -8.1 2.6 19.2 6.7 18.5 3.5 0.9 5.4 Notes: Values were calculated across all years and seasons (Fall, Winter, and Spring) in the sample. Values for temperature and precipitation are shown by season. Seasons are September-November (Fall), December-February (Winter), and March-May (Spring). Table S3. Regression results: impacts of temperature on log wheat yield (bu/acre) Variables (1) (2) (3) (4) -0.0924*** [0.0221] -0.0324* [0.0173] -0.0358** [0.0154] -0.0109*** [0.00306] 0.00683** [0.00290] -0.00169* [0.000972] -0.0138** [0.00581] 0.0144** [0.00544] -0.00656** [0.00288] -0.00172** [0.000695] 0.00307*** [0.000966] -0.0773*** [0.0132] -0.0945*** [0.0226] -0.0389* [0.0192] -0.0355** [0.0159] -0.0112*** [0.00309] 0.00733** [0.00293] -0.00188* [0.00101] -0.0157** [0.00641] 0.0151** [0.00570] -0.00650** [0.00285] -0.00154** [0.000744] 0.00282** [0.00103] -0.0773*** [0.0147] -0.0652*** [0.0218] -0.0537*** [0.0172] -0.0357* [0.0208] -0.00970*** [0.00262] 0.00787** [0.00285] -0.00218* [0.00111] -0.0192*** [0.00563] 0.0146** [0.00581] -0.00454 [0.00312] -0.000886 [0.00100] 0.00150 [0.00101] -0.0730*** [0.0172] -0.0942*** [0.0226] -0.0390* [0.0191] -0.0357** [0.0160] -0.0111*** [0.00310] 0.00722** [0.00294] -0.00183* [0.00101] -0.0158** [0.00638] 0.0152** [0.00570] -0.00652** [0.00285] -0.00155** [0.000747] 0.00284** [0.00105] a -0.0866*** [0.0161] Field Trial Field Trial Farm Field Trial Trial Location Fixed Effects Y Y N Y Variety Fixed Effects Y N N N Cluster Fixed Effects N Y N Y County Fixed Effects N N Y N Heterogeneous High Spring DD effect N N N Y R-squared 0.470 0.431 0.429 0.433 Observations Freeze Days: Fall Freeze Days: Wint Freeze Days: Spring Degree Days Low: Fall Degree Days Med: Fall Degree Days High: Fall Degree Days Low: Winter Degree Days Med: Winter Degree Days High: Winter Degree Days Low: Spring Degree Days Med: Spring Degree Days High: Spring Source of Yield Measure 6020 6020 217 6020 Num Locations 11 11 11 11 Num Varieties 197 197 -- 197 Num Clusters -- 24 -- 24 Years 29 29 28 29 Notes: All models include a quadratic trend over time and Fall, Winter, and Spring precipitation and precipitation squared variables. Spatial and heteroskedasticity robust standard errors are reported in square brackets. Farm yields are countylevel yields from the National Agricultural Statistics Service. *, **, and *** denote statistical significance at the 10, 5, and 1 percent levels. a the estimate and standard error are reported for the average of the 24 cluster-specific parameter estimates for the effect of high degree days in the Spring. Table S4. Characteristics of the 24 genetic clusters included in the sample 1 Number of Locations 11 Number of Years 26 Min/Max of Years 1988/2013 Number of Varieties 5 2 11 27 1987/2013 17 511 3 11 12 1987/2013 5 88 4 11 24 1988/2013 8 228 5 11 18 1995/2012 5 192 6 11 26 1987/2013 11 219 7 11 26 1985/2010 19 519 8 11 28 1985/2013 21 557 Cluster Observations 323 9 11 26 1985/2013 11 266 10 11 29 1985/2013 8 348 11 11 29 1985/2013 8 281 12 11 24 1985/2012 8 346 13 11 24 1987/2013 8 187 14 11 11 2003/2013 8 229 15 11 15 1996/2013 6 107 16 4 5 2005/2009 1 14 17 11 24 1985/2008 8 293 18 5 3 2003/2005 1 12 19 11 9 1985/2004 9 99 20 11 18 1993/2013 3 171 21 11 16 1998/2013 6 173 22 11 28 1985/2012 11 503 23 11 20 1994/2013 5 195 24 11 19 1985/2010 5 159 Total 11 29 1985/2013 197 6020 Table S5. Clusters, release dates, traits, and pedigrees of included varieties Variety Release Source Year Special Pedigree Traits Cluster 1 2137 2154 2163 Everest TAM 302 1995 1986 1989 2009 1999 Kan. State U. Pioneer Pioneer Kan. State U. Texas A & M none none none none none Cluster 2 2145 Abilene Antero Avalanche Clara CL Danby Mesa NuGrain Nuplains Ogallala Rio Blanco Robidoux RonL Thunderbolt Trailblazer Trego Venango 2001 1988 2013 2001 2012 2005 1988 2006 1999 1992 1989 2010 2006 1999 1985 1999 2000 Kan. State U. AgriPro Col. State U. Colo. State U. Kan. State U. Kan. State U. AgriPro AgriPro USDA, U. Neb. AgriPro AgriPro U. Nebraka Kan. State U. AgriPro AgriPro Kan. State U. Goertzen none Abilene, Pioneer experimentals none Payne sib (Triumph type), Scout sibs white TAM 111, Trego/Betty sib white Rio Blanco, TAM 107 sib whi., CL RonL sib, Clearfield white Trego, Jagger sib none Payne sib (Triumph type), Scout sibs white AgriPro experimentals, Platte white Abilene, Kan. State U. experimentals none TAM 200 sib, Abilene white Sister to Mesa, Abilene none Wahoo, Cody, Scout66, same genetic family as Millineum white Trego, Colorado expt. none Abilene, modified TAM 107 none Sister to Mesa and Abilene white Rio Blanco, modified TAM 107 none Pioneer experimental Cluster 3 2157 2158 Billings Duster Iba 1983 1991 2009 2006 2012 Pioneer Okla. State U. Okla. State U. Okla. State U. Okla. State U. none none none none none Cluster 4 2172 2180 Endurance Gallagher NuFrontier Ok101 TAM 109 TAM 304 1985 1988 2004 2012 2000 2001 1991 2007 Pioneer Pioneer Okla. State U. Okla. State U. AgriPro Okla. State U. Texas A & M Texas A & M none TAM 101, Mexican Cimmyt experimentals none TAM 101, Mexican Cimmyt experimentals none 2180, Siouxland, TAM 105, experimentals none Duster, Oklahoma Expt. That includes 2180 white Improved Centurk type none 2180, Chisholm sib, Mesa no beard TAM 107, Texas experimentals none Pioneer germplasm, Arkan, experimentals Cluster 5 2174 Centerfield Custer Keota Ok102 1997 2006 1994 2005 2002 Okla. State U. Okla. State U. Okla. State U. Westbred Okla. State U. none CL none none none 2163, Experimental lines TAM 101, Mexican Cimmyt experimentals Mexican Cimmyt wheats. Related to 2180, without TAM 101 Pioneer experimentals, Betty ProBrand 812, Caldwell, TAM 106, experimentals Sturdy, Tascosa 2157 selection Pioneer experimentals, Ukranian line Pioneer experimentals, Texas expt., Nebraska expt. Duster, Oklahoma expt. That includes Karl 92, Tomahawk 2165, PL145, Illinois experimental 2174, Above sib Chisholm, TAM 105, Romanian variety Jagger, Custer 2174, Cimarron Table S5 continued. Variety Release Source Year Special Pedigree Traits Cluster 6 7805 Akron Ankor Brawl CL Plus Byrd Ruby Lee Stanton T153 T154 TAM 112 Thunder CL 1988 1994 2001 2011 2011 2011 2000 2011 2010 2005 2008 Agseco Col. State U. Col. State U. Col. State U. Col. State U. Okla. State U. Kan. State U. Trio Research Trio Research Texas A & M Col. State U none none none none none none none none none none white Probrand 817, TAM 105 TAM 107, Hail TAM 107, Hail, RWA-resistant germplasm Above, Canadian spring Clearfield wheat, experimentals TAM 112, Ike, Halt Endurance (Pioneer genetics), USDA experimental Modified TAM 107, TAM 200, experimentals, T136, T81, experimental line also used in T158 T811 (selection from T81), 2180, T88 TAM 110, TAM 200, WGRC experimental Kansas experiment line, Colorado experimental line, half Halt Cluster 7 7833 Alliance Arapahoe Brule Culver Harry HR64 Jules Millennium Mustang Niobrara Overland Rawhide Redland Siouxland Siouxland 89 Vista Wahoo Wrangler 1988 1994 1988 1982 1999 2002 1983 1994 2000 1983 1994 2007 1991 1988 1984 1989 1992 2001 1983 AGSECO Agseco U. Nebraska U. Nebraska U. Nebraska U. Nebraska Garst Col. State U. U. Nebraska AgriPro U. Nebraska U. Nebraska U. Nebraska U. Nebraska U. Nebraska U. Nebraska U. Nebraska U. Nebraska AgriPro none none none none none none none none none none none none none none none none none none none Olesen/Omaha/Homestead Arkan, Chisholm, Colt Brule, Agent, Parker 76, Lancer, Russian variety Gage, Nebraska experimentals Arapahoe, Scout, Buckskin, experimentals Newton, Brule, N/A Hawk, NE7666 Arapahoe, Abilene, Colt, Warrior, Agent, Kavkaz Colorado experimentals, Warrior, Sonora, Trapper Brule, TAM 107 sib Millineum, Seward, Archer, experimentals N/A Brule selection Warrior, Agent, Kavkaz Warrior, Agent, Kavkaz Centurk, Brule, Warrior, Ottawa, Atlas 66, Cheyenne, Arapahoe, Abilene Mustang sib Table S5 continued. Variety Release Source Year Special Pedigree Traits Cluster 8 7837 Denali Dumas Hondo Karl LCH08-80 Longhorn Neosho Norkan Oro Blanco Pony Rowdy Stallion T111 T83 TAM 108 TAM 111 Thunderbird Tonkawa Triumph 64 WB-Cedar 1988 2011 2000 1998 1988 2013 1990 2005 1986 1994 1985 1994 1985 2000 1995 1984 2003 1985 1994 1964 2011 AGSECO Col. State U. AgriPro AgriPro Kan. State U. Limagrain AgriPro AgriPro Kan. State U. AgriPro Rohm & Haas AgriPro AgriPro Trio Trio Texas A & M AgriPro, TA&M AgriPro Okla. State U. Danne Westbred none Trison/Prosper none TAM 111, experimental that is half Yumar none AgriPro experimentals none AgriPro experimentals none Plainsman V, Kaw, Atlas 50, Parker, Agent none Pioneer germplasm no beard Thunderbird none Victory, Thunderbird, Heyne sib none Scout, Plainsman V, Larned, Sage, Eagle white Rio Blanco, Colorado experimental related to Hawk none Scout, Sturdy none Mesa sib, AgriPro experimental none Bulk Selection none T67, Tecumseh, Plainsman V none TAM 107, TAM 108, Lancota none Sturdy sib, Triumph, Centurk none TAM 107, Centurk 78, Texas experimentals none Bulk selection none Chisholm, TAM 105, Romanian variety none Triumph, Blackhull, Kanred none TAM 302, experimentals with 2180 in pedigree Cluster 9 7846 Bill Brown Colby 94 Hatcher Lamar Sierra Vona 1988 2007 1994 2005 1988 1990 1976 Agseco Col. State U. Agseco Col. State U. Col. State U. AgriPro Col. State U. none none none none none none none Windstar Wings Yuma Yumar 1997 1977 1991 1997 U. Nebraksa AgriPro Col. State U. Col. State U. none none none none Vona, Centurk, Sturdy, Agent Yumar, Arlin N/A Yumar, Vista, Kansas experimental Vona, Wings, Colorado experimental Century, Scout, Payne, Mexican spring wheat, English wheat Kaw, Sonora 64, Tacuari, Warrior, Kenya 58, Newthatch, Cheyenne, Tenmarq, Mediterranean, Hope, Parker, Lancer, Norin Siouxland, Brule, Caldwell, Vona sib Vona, Yugoslavian wheats Yuma, RWA-resistant germplasm Cluster 10 Above Halt Prairie Red Ripper T158 TAM 105 TAM 107 TAM 110 2001 1994 1999 2007 2009 1979 1984 1996 Col. State U. Col. State U. Col. State U. Col. State U. Trio Research Texas A & M Texas A & M Agseco, T A&M CL none none none none none none none TAM 110, French variety TAM 107, Sumner, RWA-resistant genetics TAM 107, RWA-resistant germplasm Prairie Red, experimentals T81, KS93U206 exptl. Scout, short wheats TAM 105, Amigo TAM 107, Largo Table S5 continued. Variety Release Source Year Special Pedigree Traits Cluster 11 Arkan Coronado Garrison Pecos Settler CL Voyager Wesley Winterhawk 1982 1994 2011 1991 2008 1994 1999 2007 Kan. State U. AgriPro Okla. State U. AgriPro U. Nebraska Goertzen USDA, U. Neb. WestBred none none none none none none none none Sage, Arthur Mustang, experimentals, some spring wheat Betty, Hickok, Oklahoma experimental that is half 2180 Arkan, AgriPro experimental Millennium, Wesley, Above sibs N/A Plainsman V, Colt, Cody, Russian experimental WestBred experimentals Cluster 12 Arlin Camelot Dodge Lakin Newton 1992 2008 1986 2000 1978 Kan. State U. U. Nebraska Kan. State U. Kan. State U. Kan. State U. white none none white none NuHorizon Snowmass Tiger 2000 2009 2010 General Mills Col. State U. Kan. State U. white white white Selection from hard red winter and hard red spring wheat Arlin sib, Redland, Experimetnals Arkan sib, Newton sib Arlin, experimentals Scout, Sonora 64, Pitic, Kenya 58, Newthatch, Thatcher, Frontana, Klein Rendidor Arlin, AgriPro experimentals Trego and experimentals, with some Arlin Experimentals, including some Ike Cluster 13 Armour Guymon Hickok Intrada Spartan TAM 113 TAM 200 TAM 202 2008 2005 1994 2000 2008 2010 1986 1991 Westbred Okla. State U. AgriPro Okla. State U. Westbred Texas A & M Texas A & M Texas A & M none white none white none none none none Westbred experimentals Intrada, Platte, AgriPro experimental TAM 200 sib, AgriPro experimental Rio Blanco, TAM 200 2180, Experimentals TAM 200 sib, TAM 105 sib, TAM 202, TAM 200 Amigo, Scout types Amigo gene, Siouxland outcross Cluster 14 Art Fuller Jackpot OK Bullet Overley Postrock Tarkio WB-Stout 2007 2006 2008 2005 2003 2006 2006 2010 AgriPro Kan. State U. Agripro Okla. State U. Kan. State U. Agripro WestBred WestBred none none none none none none none none Jagger, AgriPro experimentals Jagger, experimentals W98-232 and WGRC experimetnal Jagger, KS96WGRC39 experimental Jagger, Heyne, modified TAM 107 Jagger, Ogallala, Kansas experimental Jagger, Snow White, Oklahoma expt. Jagger, Santa Fe sib, experimentals Cluster 15 Big Dawg Hitch Santa Fe SY Wolf TAM 203 W99-194 1996 2008 2003 2011 2007 2004 AgriPro WestBred WestBred Agripro Texas A & M U. Nebraska none none none none none none Bulk selection Abilene, karl 92, Jagger, Kansas experimentals Jagger, G1878 Agripro experimentals Romanian variety, experimentals N/A Cluster 16 Bond CL 2005 Col. State U. CL Above sib, Yumar Table S5 continued. Variety Release Source Year Special Pedigree Traits Cluster 17 Bronco Century Chisholm Cimarron Discovery Enhancer Onaga Payne 1989 1996 1983 1990 1995 1998 1998 1977 AgriPro Okla. State U Okla. State U. Okla. State U. Goertzn Goertzen Agseco Okla. State U. none none none none none none none none Payne, half-parent of Hawk Payne, TAM 101, Amigo Sturdy sib, Triumph, Oro, Tenmarq Payne, Wings (Vona sib) N/A Includes Pioneer genetics Pioneer genetics Sturdy, Triumph 64 Cluster 18 Burchett 2003 Farmer Direct white AgriPro experimentals Cluster 19 Carson Centura Centurk 78 1986 1983 1978 Col. State U. U. Nebraska U. Nebraska none none none Cody Goodstreak 1986 2002 U. Nebraska U. Nebraska none none Hawk Parker 76 Ram Sandy 1981 1977 1983 1981 AgriPro Kan. State U. AgriPro Col. State U. none none none none Centurk, Scout, Anza Centurk 78, Warrior, Agent Turkey, Cheyenne, Parker, Newthatch (HRS), Kenya 58 (HRS), Hope (HRS) Centurk 78, Agent Len, Butte, Agent, Waldron, Bluebird, Dular, Eagle, Cheney, Larned, TAM107, Colt, Patrizanka Baca, Colorado experimentals Parker*5/Agent Hawk sib Centurk, Mexican varieties Cluster 20 Cisco Deliver Karl 92 2003 2004 1992 Goertzen Okla. State U. Kan. State U. none Karl 92, Bounty experimentals no beard Karl, Chisholm, experimentals none Plainsman V, Kaw, Atlas 50, Parker, Agent Cluster 21 CJ Cutter Heyne Jagalene SY Southwind 2010 2001 1998 2001 2012 Agripro AgriPro Kan. State U. AgriPro Agripro none none white none none WB-Grainfield 2012 Westbred none Cluster 22 Colt Eagle Hallam Ike Infinity CL Larned McGill Nekota Scout 66 1984 1971 2005 1993 2005 1976 2010 1994 1967 U. Nebraska Kan. State U. U. Nebraska Kan. State U. U. Nebraska Kan. State U. U. Nebraska U. Nebraska U. Nebraska none none none none CL none none none none T163 Tut 2010 1989 Triio Research Pharaoh none none Agripro experimentals Jagger, TAM 200, W81-296 Experimentals, including some Plainsman V Jagger, Abilene Agripro experimentals experimentals Westbred and Kan. State U. experimentals Related to Vona. Nebred, Hope, Turkey, Cheyenne, Ponca Niobrara, Brule, Bennett Dular, Eagle, Cheney, Larned, Colt Windstar, Millennium sib, Above sib Scout, Ottawa Ike, Redland, Chisholm, Vona, and Plainsman TAM 107, Bennett Scout, Nebred, Hope (HRS), Pawnee, Marquillo (HRS), Kawvale T811 (selection from T81), 93WGRC27 experimental Field selection of soil-borne resistant Scout type Table S5 continued. Variety Release Source Year Special Pedigree Traits Cluster 23 Greer Jagger Shocker Smoky Hill TAM 401 2010 1994 2006 2006 2008 Agripro Kan. State U. WestBred WestBred Texas A&M none Experimentals, including Pioneer genetics none Karl sib, Stephens (SWW) none Jagger, Tomahawk, Freedom (soft) none Bulk selection no beard Jagger, Mason Cluster 24 Hawken NuDakota Ponderosa Tomahawk Victory 2007 2006 1992 1990 1985 AgriPro AgriPro AgriPro AgriPro AgriPro none white none none none AgriPro experimentals Jagger, Romanian variety Thunderbird, W81-133 Related to Victory Bulk selection Table S6. Estimated heat effects and average yields for the 24 genetic clusters included in the sample Cluster Stand Err Avg Yield 1 -0.082*** Estimate 0.0182 52.9 Varieties 2 -0.071*** 0.0202 51.7 3 -0.099*** 0.0166 49.9 2145,ABILENE,ANTERO,AVALANCHE,CLARA CL,DANBY,MESA,NUGRAIN,NUPLAINS,OGALLALA,RIO BLANCO,ROBIDOUX,RONL,THUNDERBOLT,TRAILBLAZER,TREGO,VENANGO 2157,2158,BILLINGS,DUSTER,IBA 4 -0.095*** 0.0189 50.4 2172,2180,ENDURANCE,GALLAGHER,NUFRONTIER,OK101,TAM 109,TAM 304 5 -0.067*** 0.0203 52.0 2174,CENTERFIELD,CUSTER,KEOTA,OK102 6 -0.087*** 0.0177 51.9 7805,AKRON,ANKOR,BRAWL CL PLUS,BYRD,RUBY LEE,STANTON,T153,T154,TAM 112,THUNDER CL 7 -0.075*** 0.0143 52.3 8 -0.089*** 0.0128 50.4 9 -0.086*** 0.0150 53.5 7846,BILL BROWN,COLBY 94,HATCHER,LAMAR,SIERRA,VONA,WINDSTAR,WINGS,YUMA,YUMAR 10 -0.065*** 0.0165 51.6 ABOVE,HALT,PRAIRIE RED,RIPPER,T158,TAM 105,TAM 107,TAM 110 11 -0.087*** 0.0276 48.9 ARKAN,CORONADO,GARRISON,PECOS,SETTLER CL,VOYAGER,WESLEY,WINTERHAWK 12 -0.063*** 0.0171 46.6 ARLIN,CAMELOT,DODGE,LAKIN,NEWTON,NUHORIZON,SNOWMASS,TIGER 13 -0.091*** 0.0255 49.3 ARMOUR,GUYMON,HICKOK,INTRADA,SPARTAN,TAM 113,TAM 200,TAM 202 14 -0.086** 0.0343 53.0 ART,FULLER,JACKPOT,OK BULLET,OVERLEY,POSTROCK,TARKIO,WB-STOUT 15 -0.113** 0.0421 53.5 BIG DAWG,HITCH,SANTA FE,SY WOLF,TAM 203,W99-194 16 -0.154*** 0.0374 58.8 BOND CL 17 -0.080*** 0.0177 50.6 BRONCO,CENTURY,CHISHOLM,CIMARRON,DISCOVERY,ENHANCER,ONAGA,PAYNE 18 -0.075** 0.0296 54.8 BURCHETT 19 -0.133*** 0.0230 50.0 CARSON,CENTURA,CENTURK 78,CODY,GOODSTREAK,HAWK,PARKER 76,RAM,SANDY 20 -0.109*** 0.0307 50.4 CISCO,DELIVER,KARL 92 21 -0.077*** 0.0247 53.0 CJ,CUTTER,HEYNE,JAGALENE,SY SOUTHWIND,WB-GRAINFIELD 22 -0.043** 0.0182 48.8 COLT,EAGLE,HALLAM,IKE,INFINITY CL,LARNED,MCGILL,NEKOTA,SCOUT 66,T163,TUT 23 -0.089*** 0.0296 52.5 GREER,JAGGER,SHOCKER,SMOKY HILL,TAM 401 24 -0.064*** 0.0160 51.8 HAWKEN,NUDAKOTA,PONDEROSA,TOMAHAWK,VICTORY 2137,2154,2163,EVEREST,TAM 302 7833,ALLIANCE,ARAPAHOE,BRULE,CULVER,HARRY,HR 64,JULES,MILLENNIUM,MUSTANG,NIOBRARA, OVERLAND,RAWHIDE,REDLAND,SIOUXLAND,SIOUXLAND 84,VISTA,WAHOO,WRANGLER 7837,DENALI,DUMAS,HONDO,KARL,LCH08-80,LONGHORN,NEOSHO,NORKAN,ORO BLANCO,PONY,ROWDY,STALLION,T111,T83,TAM 108,TAM 111,THUNDERBIRD,TONKAWA,TRIUMPH 64,WB-CEDAR Notes: Highest and lowest heat effects are in italics. Estimates are from the preferred model (Table S3, Col 4). Average yields (bu/acre) are predicted using this specification, with yield replacing log yield as the dependent variable. The model is re-estimated and predictions are made at the sample average of covariates, with the exception that the trend variable is held at its 2013 value to reflect modern practices. Table S7. Joint hypothesis tests for regression parameters Null Hypothesis P-value Equality of Spring DD high effect across genetic clusters 0.000 Precipitation Effects Jointly Zero 0.055 Trend Parameters Jointly Zero 0.006 Equality of Fixed Effects All location effects equal 0.000 All cluster effects equal 0.000 Notes: Hypotheses were tested using F-tests with spatial and heteroskedasticity robust standard errors. *, **, and *** denote statistical significance at the 10, 5, and 1 percent levels. Table S8. Prediction accuracy for alternative models Model rmse Preferred 0.3274 All spring temperature effects heterogeneous 0.3277 All temperature effects heterogeneous 0.3702 Notes: We consider models with varying degrees of heterogeneity in the temperature coefficients. The preferred model allows the effect of extreme heat in the spring to vary across genetic clusters. The second model allows the effect of all spring temperatures to vary, and the third allows the effect of all temperature variables to vary. Each model is estimated 1000 times, randomly choosing 80 percent of the 6,020 observations. Parameter estimates are then used to predict yields for the omitted 20 percent in each subsample. Root-mean-squared-errors (rmse) are reported for these predictions. The smaller the rmse, the better the performance of the model. 150 100 50 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 0 Figure S1 Spatial and temporal variation of yields. We observe yields across varieties at the location-year level, and construct boxplots for each year. Each box is defined by the upper and lower quartile, with the median depicted as a horizontal line within the box. The endpoints for the whiskers are the upper and lower adjacent values, which are defined as the relevant quartile +/- three-halves of the interquartile range, and circles represent data points outside of the adjacent values. Fall Wint Avg Min Temp (°C) -2 10 8 6 4 -6 -8 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 -10 2 15 -4 Spri 10 5 0 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Avg Min Temp (°C) Avg Min Temp (°C) 12 Figure S2 Seasonal average daily minimum temperature by year. We average observed daily minumimum temperaure within each season for each location-year. Seasons are SeptemberNovember (Fall), December-February (Winter), and March-May (Spring). The location specific measures are then used to construct boxplots for each year. Each box is defined by the upper and lower quartile, with the median depicted as a horizontal line within the box. The endpoints for the whiskers are the upper and lower adjacent values, which are defined as the relevant quartile +/- three-halves of the interquartile range, and circles represent data points outside of the adjacent values. Fall Avg Max Temp (°C) 12 22 20 18 16 Wint 10 8 6 4 2 24 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 14 Spri 22 20 18 16 14 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Avg Max Temp (°C) Avg Max Temp (°C) 24 Figure S3. Seasonal average daily maximum temperature by year. We average observed daily maximum temperaure within each season for each location-year. Seasons are SeptemberNovember (Fall), December-February (Winter), and March-May (Spring). The location specific measures are then used to construct boxplots for each year. Each box is defined by the upper and lower quartile, with the median depicted as a horizontal line within the box. The endpoints for the whiskers are the upper and lower adjacent values, which are defined as the relevant quartile +/- three-halves of the interquartile range, and circles represent data points outside of the adjacent values. Fall 10 Cumulative Precip (in) 20 15 10 5 25 8 6 4 2 0 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 0 Wint 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 25 Spri 20 15 10 5 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 0 Figure S4. Seasonal cumulative precipitation by year. We sum observed daily precipitation within each season for each location-year. Seasons are September-November (Fall), DecemberFebruary (Winter), and March-May (Spring). The location specific measures are then used to construct boxplots for each year. Each box is defined by the upper and lower quartile, with the median depicted as a horizontal line within the box. The endpoints for the whiskers are the upper and lower adjacent values, which are defined as the relevant quartile +/- three-halves of the interquartile range, and circles represent data points outside of the adjacent values. 60 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 10 20 30 40 50 Wint Spri 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 0 5 10 15 20 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Freeze Exposure (24 hours) 20 0 5 10 15 Fall Figure S5. Seasonal freeze days. Cumulative exposure to below 0°C temperatures within each day are constructed using a sinusoidal interpolation of temperature exposure. Daily values are summed within each season for each location-year. Seasons are September-November (Fall), December-February (Winter), and March-May (Spring). The location specific measures are then used to construct boxplots for each year. Each box is defined by the upper and lower quartile, with the median depicted as a horizontal line within the box. The endpoints for the whiskers are the upper and lower adjacent values, which are defined as the relevant quartile +/- three-halves of the interquartile range, and circles represent data points outside of the adjacent values. 1,800 200 300 400 500 Wint 2010 2005 2010 2010 2000 2005 2005 1995 2000 2000 1990 1995 1995 1985 1990 1990 1,000 1,200 1,400 1,600 1,800 1985 100 1,000 1,200 1,400 1,600 DD above 0 (24 hours) 600 Fall 1985 800 Spri Figure S6. Seasonal degree days above 0°C. Degree days for each day are constructed using a sinusoidal interpolation of temperature exposure. Daily values are summed within each season for each location-year. Seasons are September-November (Fall), December-February (Winter), and March-May (Spring). The location specific measures are then used to construct boxplots for each year. Each box is defined by the upper and lower quartile, with the median depicted as a horizontal line within the box. The endpoints for the whiskers are the upper and lower adjacent values, which are defined as the relevant quartile +/- three-halves of the interquartile range, and circles represent data points outside of the adjacent values. -5 5 0 -5 Wint_DD0 -10 5 Spri_DD0 0 -5 -10 ALL ELD FND FRD GRD HVD LBD RLD RND RPD STD THD -10 0 -10 Trend (% of mean) -5 Trend (% of mean) 0 Fall_DD0 Spri_FR ALL ELD FND FRD GRD HVD LBD RLD RND RPD STD THD -10 ALL ELD FND FRD GRD HVD LBD RLD RND RPD STD THD Trend (% of mean) 5 5 Trend (% of mean) -5 ALL ELD FND FRD GRD HVD LBD RLD RND RPD STD THD -10 0 Wint_FR ALL ELD FND FRD GRD HVD LBD RLD RND RPD STD THD -5 5 ALL ELD FND FRD GRD HVD LBD RLD RND RPD STD THD 0 Fall_FR Trend (% of mean) Trend (% of mean) 5 Figure S7. Trends in freeze days (FR) and degree days above 0°C (DD0) over time. Each bar corresponds to the estimated trend coefficient in a model that regresses the weather variable against time. The estimates are normalized to a percentage of the sample mean for that variable. The term, “ALL” refers to a pooled model across all locations, the others denote a regression using only that locations data. Bars show 95% confidence intervals. 0 A 15 B -.05 Number of varieties 10 -.1 5 -.15 1960 1970 1980 1990 2000 Release year 2010 0 1960 1970 1980 1990 2000 Release year 2010 Figure S8. Effect of extreme heat across seed variety release years. (A) Degree days above 34°C are interacted with polynomials of release year in the regression model. Lines measure the estimated marginal effect of an additional degree day above 34°C. The horizontal line is the constant effect in absence of the interactions. (B) The number of varieties by release year A. Varieties Ranked by Observed Trial Years in Data Number of Years 30 20 10 0 1 Varieties 197 B. Clusters Ranked by Observed Trial Years in Data Number of Years 30 20 10 0 1 Clusters 24 Figure S9. Number of observed years for seed varieties and clusters of varieties. Bars report the total number of trial years that each variety or cluster appeared in. 2 1 0 -1 0 50 100 150 Varieties, ranked by effect size 200 Figure S10. Effect of extreme heat across seed varieties. Degree days above 34°C are interacted with a dummy (indicator) variable for each seed variety. Droplines report estimated marginal effects of an additional degree day above 34°C for each variety. -.15 -.2 -.25 -.3 -.35 15 16 17 18 19 20 21 22 23 24 25 Number of clusters 26 27 28 29 30 Figure S11. Optimal number of genetic clusters to include in the regression model. We consider models with 15-30 genetic clusters, and interact dummy (indicator) variables for these clusters with degree days above 34°C. Root mean squared errors (Rmse) are reported as a percentage reduction relative to the baseline model that does not interact degree days above 34°C with genetic clusters. Each of the 30 models is estimated 1000 times, randomly choosing 80 percent of the 6,020 observations. Parameter estimates are then used to predict yields for the omitted 20 percent in each subsample. The larger the reduction the better the performance of the model. N=15 N=16 N=17 N=18 N=19 N=20 N=21 N=22 N=23 N=24 N=25 N=26 N=27 N=28 N=29 N=30 0 -.05 -.1 -.15 0 -.05 -.1 -.15 0 -.05 -.1 -.15 0 -.05 -.1 -.15 0 10 20 30 0 10 20 30 0 10 20 30 0 10 20 30 Clusters, ranked by effect size Graphs by Num Figure S12. Heterogeneous effect of extreme heat across various groupings of genetic similarity. We consider models with 15-30 categories of genetic clusters. For example, when N=15 each variety is classified into one of fifteen groups, when N=30 each variety is classified into one of thirty groups. A separate heat effect is estimated for each cluster by interacting the categorical variable with degree days above 34°C. The Droplines report estimated marginal effect of an additional degree day above 34°C for each cluster. HR 64 MUSTANG NIOBRARA WAHOO OE ARAPAH M NNIU MILLE CISCO ER DELIVL 92 KAR LIN AR GE DOD T ELO CAMRIZON R HO TIGE S NU S MA IN OW AK L SN ON WT TT NE CHE AN K S R BU AR CO L PE R C EY LE L K TT ES AW N SE W H ISO O R D TE R A IN GAR ON R O C W TE R FI EL O 217 D K1 4 21 02 BI L 21 57 DU LIN 58 ST GS ER OR H O B ON IBA LA DO RO NCO TA WDY M THU STAL 108 NDE LIO N LON RBIR GHO D NEO RN SHO DEN A TAM LI 1 LCH08 11 WB-CED -80 AR TRIUMPH 64 PONY TONKAWA KARL EN C 76 78 K R W KE RK HA AR TU RA P EN TU C EN Y C OD Y K C ND ON REA SA RS ST CA OD A GOERR SI MAR 94 LA LBY CO 6 R 784 DSTA WIN CHER HAT A M YU AR YUM S WING VONA OWN R BILL B L BOND C OSA PONDER HAWKEN ORY VICT TOMAHAWK NUDAKOTA SY SOUTHWIND CJ HEYNE WB-GRAI NFIELD CUTTER JAGALE OVER NE L E Y FULL STO ER TAR UT OK KIO JAC BULLE PO KPOT T AR STRO CK SA T HI NTA TA TCH FE W M2 S 99- 03 B Y W 194 S IG OL SH MO DAW F G O KY G TA RE CK HI M ER ER LL 40 1 ER N US G O PL G NT JA TA 5 CL L S 0 78 RAWOR B NK N A KRO A 54 T1 53 LEE T1 BY 2 RU M 11 L TA RD ER C BY UND TH 8 T15 PER RIP T HAL 107 TAM 105 TAM IE RED PRAIR 10 TAM 1 ABOVE DUMAS 7837 T83 NORKAN T111 VO EV YA ER GE E R 21 ST 21 37 TA 21 63 M 54 3 TA 21 02 M 72 10 9 2 EN OK 180 D GA URA 101 NU LLAG NCE FRO HE R N TAM TIER ARM 304 GUY OUR M INTR ON AD TAM 1 A 1 TAM 20 3 0 TAM 20 2 HICKOK SPARTAN AVALANCHE CLARA CL TREGO ANTERO DANBY RONL RBOLT DE THUN NCO RIO BLA ZER BLA TRAIL LLALA A G O X IDOU ROB NGO A VEN 2145 AIN GR NU LAINS E P NU ILEN A AB MES CO ON LM R B HO RY IS TU N H C EN RO E C AR YN M PA ER CI NC RY HA VE GA EN CO NA TER A S O S T I D O CU KE WRANGLER OVERLAND 7833 REDLAND BRULE ALLIAN VISTA CE HAR R CUL Y JUL VER SIO ES U SIO XLAN RA UXLA D T1 WHID ND 89 E INF 63 NE INIT HA KOT Y CL A L LA LA EA RN M S GL ED TU COU E IK T T 66 C E M OL R CG T AM IL L Dendrogram of 197 wheat lines Figure S13. Dendrogram for the clustered wheat varieties. 25 frequency 20 15 10 5 0 -.15 -.1 -.05 Effect of extreme heat Figure S14. Heterogeneity of heat sensitivity across genetic clusters. Figure provides a kernel density plot for the 24 heat effects estimated under our preferred model. The underlying effects are provided in Table S6. .1 0 -.1 -.2 -.3 -.4 -.5 -.6 -.7 0 10 20 Spring temperature (°C) 30 40 Figure S15. Heterogeneous effect of temperatures above 34°C across genetic clusters. Graph displays changes in log yield if the crop is exposed for one day to a particular 1°C temperature interval. Solid black line reports the average effect, other colors correspond to the 24 genetic clusters under our preferred model. 0 -20 -40 -60 -80 Most resilient cluster Least resilient cluster Average across clusters +1°C +2°C +3°C Warming scenario +4°C +5°C Figure S16. Replication of Figure 3 for an alternative model using 22 genetic clusters. The alternative model is same as the preferred model except it uses the grouping scheme with 22 clusters instead of 24. Bars show 95% confidence intervals. 0 -20 -40 -60 -80 Most resilient cluster Least resilient cluster Average across clusters +1°C +2°C +3°C Warming scenario +4°C +5°C Figure S17. Replication of Figure 3 for an alternative model using 30 genetic clusters. The alternative model is same as the preferred model except it uses the grouping scheme with 30 clusters instead of 24. Bars show 95% confidence intervals. 0 -20 -40 -60 -80 -100 Most resilient cluster Least resilient cluster Average across clusters +1°C +2°C +3°C Warming scenario +4°C +5°C Figure S18. Replication of Figure 3 for an alternative model using location specific trend variables. The alternative model is same as the preferred model except it includes separate quadratic trends for each location. Bars show 95% confidence intervals. -28 -27 -26 -25 1995 2005 2015 Harvest Year +3°C Warming 1985 1995 2005 2015 Harvest Year 1985 1995 2005 2015 Harvest Year -56 -57 -58 +4°C Warming -55 -54 -53 1985 -29 Yield impact (%) -14.5 -15 -15.5 Yield impact (%) Yield impact (%) -40 -41 -42 -43 -44 Yield impact (%) +2°C Warming 1995 2005 2015 Harvest Year -39 1985 -14 -13.5 -4.4 -4.6 -4.8 -5 Yield impact (%) -5.2 +1°C Warming +5°C Warming 1985 1995 2005 2015 Harvest Year Figure S19. Replication of Figure 4 for an alternative model using 22 genetic clusters. The alternative model is same as the preferred model except it uses the grouping scheme with 22 clusters instead of 24. -28 -27 -26 -25 1995 2005 2015 Harvest Year +3°C Warming 1985 1995 2005 2015 Harvest Year -54 -56 -42 -41 Yield impact (%) -40 -39 -52 1985 -29 Yield impact (%) -14 -14.5 -15 +2°C Warming 1995 2005 2015 Harvest Year +4°C Warming 1985 1995 2005 2015 Harvest Year -58 -43 -44 Yield impact (%) -15.5 Yield impact (%) 1985 -13.5 -13 -4.4 -4.6 -4.8 -5 -5.2 Yield impact (%) -5.4 +1°C Warming +5°C Warming 1985 1995 2005 2015 Harvest Year Figure S20. Replication of Figure 4 for an alternative model using 30 genetic clusters. The alternative model is same as the preferred model except it uses the grouping scheme with 30 clusters instead of 24. -25 -26 -27 -28 +3°C Warming 1985 1995 2005 2015 Harvest Year +4°C Warming 1985 1995 2005 2015 Harvest Year -58 -56 -54 -52 -50 -29 Yield impact (%) +2°C Warming 1985 1995 2005 2015 Harvest Year -60 Yield impact (%) -38 -40 -42 -44 Yield impact (%) -36 1985 1995 2005 2015 Harvest Year -24 -12 -13 -14 -15 Yield impact (%) -4 -4.5 -5 Yield impact (%) -5.5 +1°C Warming +5°C Warming 1985 1995 2005 2015 Harvest Year Figure S21. Replication of Figure 4 for an alternative model using location specific trend variables. The alternative model is same as the preferred model except it includes separate quadratic trends for each location. 0 -20 -40 -60 -80 2014 planted varieties Recently released varieties Most resilient varieties +1°C +2°C +3°C Warming scenario +4°C +5°C Figure S22. Replication of Figure 5 for an alternative model using 22 genetic clusters. The alternative model is same as the preferred model except it uses the grouping scheme with 22 clusters instead of 24. Bars show 95% confidence intervals. 0 -20 -40 -60 -80 2014 planted varieties Recently released varieties Most resilient varieties +1°C +2°C +3°C Warming scenario +4°C +5°C Figure S23. Replication of Figure 5 for an alternative model using 30 genetic clusters. The alternative model is same as the preferred model except it uses the grouping scheme with 30 clusters instead of 24. Bars show 95% confidence intervals. 0 -20 -40 -60 -80 2014 planted varieties Recently released varieties Most resilient varieties +1°C +2°C +3°C Warming scenario +4°C +5°C Figure S24. Replication of Figure 5 for an alternative model using location specific trend variables. The alternative model is same as the preferred model except it includes separate quadratic trends for each location. Bars show 95% confidence intervals. B. 1°C warming A. Baseline C. 2°C warming Most Most Most Avg Avg Avg Least Least Least D. 3°C warming E. 4°C warming F. 5°C warming Most Most Most Avg Avg Avg Least Least Least 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 Average yield (bu/acre) Figure S25. Yield predictions for least, average, and most heat resistant varieties across warming scenarios. Yields are predicted under the preferred model at the cluster level. Trend variables are set to their highest observed value in the data to reflect current production practices. “Most” combines the three most heat resistant clusters, “Least” combines the three least resistant, and all others are combined in “Average”.