CHANGING CLIMATES, PRODUCTION SYSTEMS AND TRAIT OBJECTIVES MAKE GENETIC RESOURCES VERY RELEVANT FOR PLANT BREEDING. HOWEVER, GERMPLASM COLLECTIONS ARE UNDERUTILIZED BECAUSE OF THE DIFFICULTY BREEDERS HAVE IN IDENTIFYING RELEVANT ACCESSIONS FOR SPECIFIC TRAITS OR ENVIRONMENTS. CARROT IS A GOOD MODEL CROP TO DEVELOP METHODOLOGY BECAUSE IT HAS AN UNUSUALLY HIGH AMOUNT OF GENOMIC INFORMATION AVAILABLE FOR A NON-COMMODITY CROP, INCLUDING A COMPLETE GENOME SEQUENCE. WE WILL DEVELOP AND TEST METHODS TO IMPROVE THE ACCURACY OF LIMITED PHENOTYPIC DATA TO CHARACTERIZE GERMPLASM COLLECTIONS. WE WILL ALSO DEVELOP SAMPLING STRATEGIES BASED ON HIGH-DENSITY MARKERS AND GENOMIC PREDICTIONS THAT WE HYPOTHESIZE WILL ALLOW BREEDERS TO MORE EFFECTIVELY SELECT ACCESSIONS TO INCLUDE IN BREEDING PROGRAMS THAN CURRENT COLLECTION SAMPLING METHODS, INCLUDING GEOGRAPHIC STRATIFICATION, MORPHOLOGICAL CLUSTERING OR NEUTRAL GENETIC DIVERSITY. WE WILL TEST THIS HYPOTHESIS BY COMPARING SAMPLING AND PREDICTIONS BASEDON GENOMIC MARKERS TO THESE THREE COMMON SAMPLING STRATEGIES FOR THEIR ABILITY TO IDENTIFY RELEVANT, HIGH VALUE ACCESSIONS ACROSS THE ENTIRE RANGE OF THE COLLECTION. WE WILL TEST THE EFFICIENCY OF USING GENOMIC-ASSISTED SELECTION COMPARED TO PHENOTYPIC SELECTION TO INCORPORATE USEFUL QUANTITATIVE TRAITS FROM ACCESSIONS TO ELITE BREEDING LINES. THIS WILL RESULT IN SOFTWARE TOOLS FOR PLANT BREEDERS AND ELITE CARROT BREEDING POOLS WITH TRAITS OF INTEREST TO BE RELEASED TO OTHER BREEDERS AND IN FINISHED CULTIVARS.