GWAS Reveals the QTLome Complexity Governing Different Root Types in Adult Durum Wheat Plants

This study reports the high-throughput phenotyping and genetic dissection of 189 elite durum wheat for root system architecture (RSA) and shoot traits at an unprecedented level of details. Plants were grown up to the 7th leaf appearance (late tillering) using the GROWSCREEN-Rhizo platform (1-m deep rhizo-boxes) integrated with automatic root RGB imaging. The following primary RSA traits were measured: seminal, nodal and lateral total root length, root system convex hull, and root system width and depth distribution (twice/week). Measurements of leaf area, leaf number and tiller number were measured twice/week and SPAD was measured twice during the experiment. Root dry biomass and shoot fresh and dry biomass were collected at the end of the experiment. The time-course measurements of the above-mentioned traits allowed us to model root and shoot growth and thus to identify QTL describing the dynamic root growth, including the root/shoot. GWAS analysis was based upon the Illumina Infinium 90K SNP profiles and the Maccaferri et al. (2016) consensus map. The accessions are structured into five sub-groups representing the main breeding lineages worldwide, namely (i) ICARDA_dry, with accessions bred for dryland areas, (ii) ICARDA_temp, with accessions bred for temperate areas, (iii) ITALIAN, with accessions derived from crosses of Italian accessions with CIMMYT and Southwestern US accessions (Desert Durum®), (iv) CIMMYT_70, with accessions adapted from CIMMYT germplasm introduced to Mediterranean countries, and (v) CIMMYT_80, with accessions selected under irrigated conditions during the past three decades. In total, the 35 traits measured directly or computed as ratios, allowed for the identification of 75 significant QTL peaks, 13 of which affected both root and shoot traits, while 26 and 36 exclusively affected RSA or the shoot, respectively, hence indicating a limited functional overlap between the RSA and shoot QTLomes. Among the 39 QTL clusters shown to affect two or more traits, three were particularly noteworthy: (1) QCls3ubo on chr. 1B affected ten traits, with a marked effect on root growth rate and root length density at depth, (2) QCls25ubo on chr. 6A affected root system width and average leaf width, and (3) QCls32 on chr. 7A affected nine traits, in particular root depth. Notably, this QTL cluster revealed a strong, contrasting selection pattern between the accessions of the rainfed and the irrigated breeding programs conducted at ICARDA and CIMMYT. The accessions used in this experiment were previously tested for yield and its components in 15 environments across the Mediterranean Basin. The joint analysis of field and platform data provides valuable insights toward a better understanding and deployment of the RSA QTLome to enhance durum wheat yield in different environmental conditions. The selection signatures evidenced in contrasting environments for water regimes suggest the possible role of the RSA QTLome in wheat adaptation and breeding in such conditions. Fine mapping and candidate genes analysis are underway for the major root growth angle on chr. 6A QTL using in silico TILLING. Additionally, the QCls25ubo-6A haplotypes with contrasting effects on root growth angle are being introgressed in different genetic backgrounds