Cereal diseases threaten food security . Increasing demand and changing climatic conditions cause extreme events that strongly affect yield stability [43, 44]. Zymoseptoria tritici is a major threat to European and Mediterranean bread and durum wheat production . Despite the increasing efforts to elucidate the genetic basis of tetraploid wheat resistance to STB [8, 40, 46], more studies are required for an effective breeding strategy in durum wheat for STB resistance. The complexity of mapping resistance genes in durum wheat is twofold; (i) the majority of durum wheat varieties is highly susceptible to Z. tritici [34, 46] and (ii) mapping resistance genes requires using specific isolates with pathogenicity to durum wheat as the majority of bread wheat derived Z. tritici isolates is non-pathogenic on durum wheat [36, 38, 47–49].
In our study, all data indicated and confirmed significant ‘isolate’ and ‘line x isolate’ interactions as determined in earlier studies [37, 38, 50, 51]; and recently proven in the bread wheat – Z. tritici pathosystem where both Stb6 and AvrStb6 genes were cloned [52–54]. Moreover, and comparably to other cereal diseases, namely to rust [55–58], we determined QTL that are detected for either seedling (Qstb1A and Qstb7A) or adult plant stage (Qstb2B_2), as well as a QTL that was detected at both stages (Qstb2B_1). Our findings confirm that specific plant physiological stage resistances are commonly observed in the Z. tritici – wheat pathosystem . Specific plant physiological stage resistances were also confirmed for other fungal diseases such as the powdery mildew and the leaf rust diseases [59, 60]. In fact, some Z. tritici resistance genes are uniquely effective at the seedling stage, such as the Stb7 gene mapped in the spring wheat cultivar ST6 , or at the adult plant stage, such as the Stb17 gene [62, 63]. In contrast, other resistance genes have proven to be effective at both seedling and adult plant stages alike the Stb4 and Stb5 qualitative genes [64, 65].
Subsequently, we compared the identified QTL to formerly identified Z. tritici genes using the reported literature. This comparison has revealed that a putative QTL for resistance to Z. tritici was mapped on chromosome 1A at 68 cM at the adult plant stage by Kidane et al.  through a genome-wide association study conducted on an Ethiopian durum wheat landrace population. Two other QTL mapped on chromosome 1A were also revealed by Goudemand et al.  in the bread wheat Apache/Balance population, and by Risser et al.  named as QStb.lsa_fb-1A in the bread wheat bi-parental mapping population Florett/Biscay. These QTL were mapped at the adult plant stage between 56 and 69 cM and could thus co-localize with the QTL mapped in the ‘Agili39’/Khiar population. However, and in contrast to the above-mentioned studies, the Qstb1A QTL mapped in the ‘Agili39’/Khiar population was solely detected at the seedling plant stage.
The Qstb7A QTL particularly conferred reduced necrosis values to Z. tritici isolate IIB123 and co-localizes with the major Stb3 gene that was mapped in the bread wheat cultivar Israel 493 [68, 69].
Two QTL for STB resistance were mapped on chromosome 2B in the ‘Agili39’/khiar population. Other studies have also revealed genomic regions on chromosome 2B associated with the STB resistance [8, 66, 70–73].
The Qstb2B_1 QTL identified in the ‘Agili39’/Khiar population likely co-localized with the known major gene Stb9 that was mapped in the French bread wheat cv. Courtot ; with the qSTB.2 QTL mapped in the Ethiopian durum wheat landrace population ; with the QStb.ihar-2B.2 QTL mapped in the Liwilla/Begra bread wheat doubled-haploid population ; and with the QStb.lfl-2B.1 mapped in the eight-founder MAGIC population of winter wheat . However, the Qstb2B_1 QTL is different from the QTL mapped at the long arm of chromosome 2B in the Nimbus/Stigg bread wheat mapping population .
The Qstb2B_2 QTL likely co-localized with the QTL identified on chromosome 2B in the mapping populations Apache/Balance and FD3/Robigus  associated with both necrosis and pycnidia resistance in the adult plant stage. However, due to the unavailability of marker sequences, we cannot conclude that Qstb2B_2 derived from ‘Agili39’ is the same locus that was mapped in the aforementioned bread wheat mapping populations.
Thus, the identified QTL in ‘Agili39’ co-localized with previously mapped QTL for STB resistance in bread and durum wheat populations, hence we cannot claim a new Stb gene in the ‘Agili39’ landrace accession. However, we clearly have identified QTL conferring resistance to a wide range of Z. tritici isolates under artificial inoculation conditions in seedlings and adult plants, known as field resistance .
Thus far, in durum wheat, only partial resistance to Z. tritici was reported [16, 75]. Here, we derived Qstb2B_1 from ‘Agili39’ that provides resistance to five Z. tritici isolates at the seedling stage, and to two isolates at the adult plant stage. Qstb2B_1 explains up to 61.6 % of the observed phenotypic variance and was characterized by a high heritability (0.98) with a dual action at the seedling and the adult plant stages. ‘Agili39’ is also the origin of Qstb2B_2, QTL providing a major adult resistance explaining up to 54.3 % of the observed phenotypic variance. Our findings confirm that Tunisian durum landraces harbor highly effective Z. tritici resistance QTL.
In fact, the initial screening of the Tunisian landrace accessions showed a remarkable genetic diversity for STB resistance, as claimed also by Ouaja et al.  proving that Tunisian durum wheat landraces encompass diverse and valuable sources of resistance to Z. tritici. Eight landrace accessions (Agili37; Agili38; Agili39, Sbei99; Derbessi 12, Mahmoudi101, JK85 and Azizi27) were highly resistant and one landrace showed an intermediate response (‘Agili41’). The different ‘Agili’ landrace accessions reacted differently to the deployed Z. tritici isolates, suggesting a different genetic background, which is in accord with Ferjaoui et al.  study hypothesizing that the tested ‘Agili’ accessions most likely carry different Stb genes combinations. Hence, and alike other durum wheat landrace populations , the Tunisian durum wheat landraces uncover untapped allelic diversity that is of a great value to support effective breeding strategies to enhance STB resistance in durum wheat.
Our data demonstrated that the broad efficacy of the observed STB resistance in ‘Agili39’ is due to several stacked QTL, both at seedling as well as adult plant stages, which was also commonly observed in inheritance studies in bread wheat [34, 62]. Pyramiding genes for disease resistance has been an effective strategy in preventing boom-and-bust cycles, and is now amenable through marker assisted breeding as a strategy to maintain disease resistance durability, such as for wheat stem rust where various resistance gene combinations have well controlled the disease since the mid-1950s and more recently to the devastating Ug99 race [60, 77–79]. A concrete illustration for Z. tritici is the effective resistance to a wide range of isolates in the bread wheat germplasm ‘KK4500’ and ‘TE11’ which is conferred by stacking several known Stb genes [80–82] and also in other germplasm several QTL have contributed to broad efficacy of resistance . Our data also confirm that stacking QTL in durum wheat results in broad efficacy of STB resistance. This study has identified genotypes harboring diverse resistance loci entailing dual actions at the different physiological stages constituting thus potential effective sources for Z. tritici resistance and will thus support sustainable breeding approach for Z. tritici resistance in durum wheat.
Finally, we explored QTL epistasis and identified four significant pairwise interactions of the identified QTL with an epistatic QTL mapped on chromosome 5B, designated as Qstb5B. Hence, the epistasis analysis has revealed other QTL that affects the expression of Z. tritici resistance in the ‘Agili39’/Khiar population. In fact, epistatic interactions between QTL are an important factor that affects the phenotypic expression of genes and genetic variation in populations [83–85]. Similarly, to many other studies [86–88], our data demonstrated interaction between QTL having main effect (Qstb2B_1 and Qstb2B_2) that are involved in epistasis with the Qstb5B QTL affecting the same trait. The epistasis analysis showed an additive-by-additive effect between the Qstb2B_1/ Qstb5B QTL and the Qstb2B_2/Qstb5B QTL, with a major effect of QTL mapped on chromosome 2B (Qstb2B_1 and Qstb2B_2) over the Qstb5B QTL. Interestingly, the epistasis analysis showed that the Qstb2B_2 QTL, proven to control pycnidia development at the adult plant stage by QTL analysis, has also an effect in controlling pycnidia development at the seedling stage when interacting with the epistatic Qstb5B QTL. Nonetheless, the epistasis analysis did not indicate an interaction between the two major QTL Qstb2B_1 and Qstb2B_2 mapped on chromosome 2B.