A consensus map for Ug99 stem rust resistance loci in wheat

This consensus map of stem rust genes, QTLs, and molecular markers will facilitate the identification of new resistance genes and provide a resource of information for development of new markers for breeding wheat varieties resistant to Ug99. The global effort to identify new sources of resistance to wheat stem rust, caused by Pucciniagraminis f. sp. tritici race group Ug99 has resulted in numerous studies reporting both qualitative genes and quantitative trait loci. The purpose of our study was to assemble all available information on loci associated with stem rust resistance from 21 recent studies on Triticum aestivum L. (bread wheat) and Triticum turgidum subsp. durum desf. (durum wheat). The software LPmerge was used to construct a stem rust resistance loci consensus wheat map with 1,433 markers incorporating Single Nucleotide Polymorphism, Diversity Arrays Technology, Genotyping-by-Sequencing as well as Simple Sequence Repeat marker information. Most of the markers associated with stem rust resistance have been identified in more than one population. Several loci identified in these populations map to the same regions with known Sr genes including Sr2, SrND643, Sr25 and Sr57 (Lr34/Yr18/Pm38), while other significant markers were located in chromosome regions where no Sr genes have been previously reported. This consensus map provides a comprehensive source of information on 141 stem rust resistance loci conferring resistance to stem rust Ug99 as well as linked markers for use in marker-assisted selection.

1 3 destructive wheat diseases. It can cause up to 90 % yield loss in wheat production but has been effectively under control due to the successful deployment of resistance genes in wheat cultivars since the 1950s (McIntosh et al. 1995). However, the outbreak of a new stem rust race in Uganda named Ug99 (race TTKSK; Pretorius et al. 2000) spread throughout much of Africa, the Middle east and Iran and poses an imminent threat to wheat production worldwide Sharma et al. 2013).
To improve the efficiency of wheat breeding for durable resistance to stem rust, it is essential to understand the genetic basis. To-date, 58 stem rust resistance (Sr) genes have been numerically designated in wheat as part of the International Wheat Genetics Symposium Gene Catalog (McIntosh et al. 1995(McIntosh et al. , 2011. Several alleles conferring unique race specificities have been identified for many of these genes resulting in a total of 65 numerically designated resistance genes and alleles. Of these genes and alleles, phenotypic data have been published indicating that at least 27 are effective or partially effective to the Ug99 race group: Sr2 (Yr30), Sr13, Sr21,Sr22,Sr24,Sr25,Sr26,Sr27,Sr28,Sr32,Sr33,Sr35,Sr36,Sr37,Sr39,Sr40,Sr42,Sr44,Sr45,Sr46,Sr47,Sr51,Sr52,Sr53,Sr55 (Lr67/Yr46/ Pm46), Sr57 (Lr34/Yr18/Pm38), Sr58 (Lr46/Yr29/Pm39) (Faris et al. 2008;Ghazvini et al. 2012;Kolmer et al. 2011;Jin and Singh 2006;Jin et al. 2007;liu et al. 2011a, b;McIntosh et al. 2012;Singh et al. 2013b, c). Several additional resistance genes have been characterized as resistant to Ug99; however, their relationship to numerically designated genes has not been determined: SrAes7t, SrCad, SrND643, SrTA10171, SrTA10187, SrTA1662, SrTmp, SrWeb, Sr1RS Amigo (Hiebert et al. 2010(Hiebert et al. , 2011Jin and Singh 2006;Jin et al. 2007;Klindworth et al. 2012;Olson et al. 2013). It is possible that such genes with temporary designations are redundant with, or are alleles of, numerically designated genes. Several of the numerically designated Sr genes are qualitative and race specific (Jin et al. 2007;Singh et al. 2011). Qualitative genes are phenotyped as present or absent, often at the seedling stage by observing the characteristic low or high infection types displayed by them as described by Jin et al. (2007). One strategy for achieving durable resistance is to pyramid multiple qualitative resistance genes into each wheat variety. A major risk associated with the utilization of qualitative resistance genes is the ability of pathogens to defeat such genes when they are deployed alone in wheat cultivars as has been demonstrated by the original Ug99 defeating Sr24, Sr36, and resistance in cultivar 'Matlabas' resulting in "Boom and Bust" cycles (Jin et al. , 2009Pretorius et al. 2012).
Utilization of quantitative resistance, often based on multiple minor genes that slow down pathogen infection and colonization in adult plants, referred to as 'Adult Plant resistance' (APr) (Gustafson and Shaner 1982), is another approach for achieving durable resistance. Adult plant resistance to stem rust in wheat is a complex trait conferred by quantitative trait loci (QTl). This type of resistance can be more durable than single gene resistance due to race non-specificity of the resistance genes involved. A total of five numerically designated wheat stem rust resistance genes confer quantitative APr: Sr2, Sr55 (Lr67/Yr46/ Pm46), Sr56, Sr57 (Lr34/Yr18/Pm38), and Sr58 (Lr46/ Yr29/Pm39). In addition, several recent studies have identified numerous QTl associated with wheat stem rust resistance in diverse germplasm Bhavani et al. 2011;Crossa et al. 2007;Kaur et al. 2009;njau et al. 2013;rouse et al. 2014;Singh et al. 2013a, b, c;Yu et al. 2011Yu et al. , 2012.
The global effort to identify new sources of resistance to Ug99 has led to the identification of putative new qualitative and quantitative resistance loci reported in numerous studies in the past few years in different populations. As resistance loci are reported, it is important to determine their potential redundancy in order to prioritize those that can be deployed in a breeding program. The purpose of this study was to compile available information on Ug99 resistance loci and their map locations in a single consensus map to facilitate future mapping studies. The map locations from various sources were analyzed along with our recent association mapping projects involving 608 spring and winter wheat breeding lines from CIMMYT and the International Winter Wheat Improvement Program (Yu et al. 2011(Yu et al. , 2012. The consensus map contains Diversity Arrays Technology (DArT), Single nucleotide Polymorphism (SnP), Genotyping-by-Sequencing (GBS) and Simple Sequence repeat (SSr) markers to facilitate cross-referencing markers and Ug99 resistance loci with other maps.

Materials and methods
literature review and synthesis of stem rust resistance loci Stem rust resistance loci data were collected from 21 recent studies, for a total of 24 biparental mapping populations, 3 association panels, 2 BC 2 populations, and two from resistance gene cloning Bhavani et al. 2011;Crossa et al. 2007;Ghazvini et al. 2012;Haile et al. 2012;Hiebert et al. 2010Hiebert et al. , 2011Kaur et al. 2009;letta et al. 2013;Olson et al. 2013;njau et al. 2013;Periyannan et al. 2013;rouse et al. 2012, 2014Saintenac et al. 2013;Singh et al. 2013a, b, c;CIMMYT unpublished, Yu et al. 2011 (Table 1). The details of the reports, including the chromosome positions of stem rust resistance loci, their lOD (log (Odds)) scores and p values, parents used to develop the population as well as stem rust pathotypes for disease reaction are presented in Table 1. Three of the studies did not determine stem rust resistance loci by screening with Ug99 Crossa et al. 2007;Kaur et al. 2009). Though included these loci in our study, their effectiveness against Ug99 is unknown and are presented because of their relevance to Ug99 studies.

Construction of a consensus map
A consensus genetic map was constructed using the Wheat Interpolated Maps v4 (Diversity Arrays Technology Pty. ltd., personal communication) as a reference map.
(http://www.triticarte.com.au/). DArT, SnP, GBS and SSr markers from the Wheat KASPar SnP database (http://www.cerealsdb.uk.net/CerealsDB/SnPS/), the 2004 Wheat SSr Consensus Map (Somers et al. 2004), and the Thatcher/Mcneal map (Sherman et al. 2013) with concatenated DArT and SSr markers were integrated to the reference map using the ad hoc r package "lPmerge" (endelman and Plomion 2011) and BioMercator V3.0 (http://moulon.Inra.fr). lPmerge is an optimized "synthetic" (Wenzl et al. 2006) or "composite" (Hudson et al. 2012) approach to built a map across multiple populations. As opposed to minimizing an objective function based on the observed recombination frequencies between markers (JoinMap; Van Ooijen 2006) or (MultiPoint;ronin et al. 2012), this r package based its algorithm directly on the component linkage maps instead of the recombination frequencies. Other software uses the same approach, but lPmerge implements an additional algorithm for resolving ordinal conflicts found when the marker order was not consistent between the different linkage maps. resistance loci or QTl associated with stem rust resistance identified from 21 studies and one personal communication (Table 1) were projected onto the consensus map based on the position of the markers linked to the loci when the precision of the constructed map allowed it (Fig. 2). A star following the chromosome number on Table 1 tags the QTl not present on Fig. 2.

Overview of loci for stem rust resistance
The constructed consensus map contains a total of 1,433 markers distributed among the 21 chromosomes. Fourteen percent of these markers are SnP, 16.7 % are GBS, 39.4 % are DArT and 29.9 % are SSr. With a global average distance of 2.7 cM between markers, the maximum average distance between markers is observed on chromosome 5D (16 markers) with 7.6 cM and the minimum average distance between markers is observed on chromosome 3B (158 markers) with 1.9 cM. An expected negative correlation is observed between the marker number per chromosome and average distance in centimorgan between markers (r = 0.719). The average chromosome size was 150 cM with a maximum of 203.3 cM for chromosome 7A and a minimum of 83.4 cM. resistance gene and QTl maps and gene cloning information from 21 studies and one personal communication (Table 1) involving 24 mapping populations, three association panels, and two BC 2 populations were concatenated to construct a map for stem rust resistance loci (Fig. 1). The map consists of 141 stem rust resistance loci distributed across the genome, many of which were redundant loci detected in at least two studies. A total of 37, 86 and 18 resistance loci were located in the A, B, and D genomes, respectively (Fig. 1). Several hotspots of resistance loci were observed across the genome. nineteen were located on chromosome arm 3BS, while 6BS, 5Bl and 2Bl had nine, nine, and seven QTl, respectively. Among these hotspots, qualitative genes for stem rust resistance have been mapped on 3BS, 2Bl and 5Bl, while no qualitative genes have been characterized on 6BS. Clusters of QTls were located distally on 5BS, 6BS and 7AS in regions where no previously reported qualitative or quantitative genes are located.
Chromosome group 1 had 15 QTl and the cloned gene, Sr33, across the three homoeologous chromosomes. Six QTl were found on 1A (two on the short arm, three on the long arm and one is still in an undetermined arm location), five using bi-parental crosses ( (Schlegel and Kynast 1987;Jin and Singh 2006;McIntosh et al. 2012). Seven QTl were found on 1B (three on the short arm, three on the long arm and one is still in an undetermined arm location). Four were mapped using bi-parental crosses (PBW343/Crossbill, PBW343/Kenya nyangumi: CIMMYT unpublished; Avocet/ Pavon76: njau et al. 2013;Bhavani et al. 2011) and three using association mapping (eSWYT: Crossa et al. 2007;lD-SrrSn (spring): Yu et al. 2011 andthe AM durum panel: letta et al. 2013). Four of these QTl are represented in Fig. 2. Sr31 (marker XwPt-8949) is on a 1Bl.1rS rye translocation and is homologous with at least three of the QTl found in that region. Sr14, from tetraploid wheat origins (Heerman and Stoa 1956), was also reported on 1B, close to the centromere (McIntosh et al. 1995). The response of Sr14 to Ug99 has been reported as inconclusive by Jin et al. (2007) and is not displayed in Fig. 2. Two QTl were mapped on chromosome 1D (one on the short arm and one      Chromosome group 2 had 19 QTl across the three homoeologous chromosomes. The short arm of chromosome 2A has two QTl where the centromeric stem rust resistance gene Sr21 (The 1973) is located and the long arm had one QTl. One of these is represented in Fig. 2. Two Sr genes have been described on chromosome 2A, Sr21 (The 1973), derived from Triticum monococcum l., as well as Sr32 (McIntosh et al. 1995). It has to be noted that stem rust gene Sr32 has also been mapped on chromosome 2B ( On chromosome group 3, 26 QTl were reported as well as a cloned gene across the three homoeologous chromosomes. A single QTl was mapped on 3AS using association mapping (AM durum panel: letta et al. 2013) and two QTl were mapped on 3Al, one using a bi-parental cross (PBW343/Kenya Kudu: CIMMYT unpublished) and one by association mapping (AM durum panel: letta et al. 2013). Two of these QTl are shown in Fig. 2. Chromosome 3A carries two Sr genes, one cloned by Saintenac et al. (2013), Sr35 on the long arm and the other one, Sr27, is located on the short arm (Acosta 1963 Fig. 2. eleven of the QTl on chromosome 3BS are coincident with the slow rusting gene, Sr2, that contributes to APr . QTl near the centromere of chromosome 3B appear to be linked to the seedling resistance gene Sr12. Though Sr12 was not characterized as effective as Ug99 (Jin et al. 2007), recent data suggest that it interacts with other resistance loci to confer APr to Ug99 (rouse et al. 2014).
Ten QTl were located on group 4 chromosomes. One QTl was located on 4AS, three were on 4Al and one was at an unknown location. Among them, one was identified using a bi-parental cross (PBW343/Juchi: Bhavani et al. 2011) and four using association mapping (lD-SrrSn (winter) and lD-SrrSn (spring) Fig. 2 A consensus map of stem rust resistance loci in wheat. The map was constructed using the Wheat Interpolated Maps v4 (Diversity Arrays Technology Pty. ltd., personal communication) as a reference map. DArT (prefix "wPt"), SnP (prefix "BS" or "SnP") and SSr markers linked to stem rust resistance loci were integrated as described in the "Materials and methods". The bars with different colors or patterns on the left side of chromosome regions distinguish mapping populations used in each study as shown in Table 1 Fig. 2. Kaur et al. (2009) detected a distal 5BS QTl in three of the four environments tested using a bi-parental population. This finding was confirmed by Yu et al. (2011) in the detection of two QTl distally located on chromosome 5BS by association mapping in spring and winter CIMMYT germplasm between markers XwPt1149/XwPt 5346 and XwPt1302/XwPt3873, respectively. Both QTl were minor but overlapped with the QTl detected by Kaur et al. (2009) Fig. 2) and Ug99 effective gene Sr22 (The 1973) are located on 7Al. On 7B there were ten QTl, five of which were mapped using bi-parental crosses (PBW343/Huirivis#1: Bhavani et al. 2011

Discussion
Four maps including the Wheat Interpolated DArT Maps v4, the wheat consensus SSr map, the wheat KASPar

Fig. 2 continued
SnP map, and the Thatcher/Mcneal DArT/SSr markers map were used to build the consensus map for locating the stem rust resistance loci (Fig. 2). The consensus map consisted of 1,433 markers and an average marker interval of 11.5 cM. We focused on integrating markers in the regions spanning stem rust resistance QTl that can further facilitate fine mapping and cross referencing the locations with other maps. Therefore, the marker density in the QTl regions is higher than the average. For example, marker density is almost three times higher in the QTl region of rust resistance genes on 1B (Fig. 2, Chromosome 1B). The same is true for 2Bl, 3BS, 4Al, 6BS and 7Dl. However, the complexity and context dependency of QTl identified in different genetic backgrounds and environments can limit the accuracy of the locations. The accurate genome location of QTl and major genes across genetic backgrounds and environments is a prerequisite for the use of the QTl in MAS. Meta-analysis of QTl identified in different studies can locate QTl more precisely, thus facilitating the identification of closely linked markers for MAS. Because many of the populations from which APr was assessed lacked Ug99-effective qualitative resistance genes, the coincidence of APr with seedling resistance was not likely to be a result of the qualitative resistance genes conferring APr. For example, on 2BS, seven QTl were identified that overlapped with two Ug99-effective stem rust resistance genes, Sr39 and Sr36 (Fig. 2b). Since Sr39 was not present in the wheat cultivars used to map the QTl and Sr36 was only present in the germplasm analyzed by Yu et al. (2012) where no association was found between Sr36 and APr, the QTl on 2BS could be conferred by alleles of these qualitative genes, residual effects of other Sr genes on 2BS, or new genes. Similarly, the stem rust resistance gene Sr40 on 2BS is strongly associated with marker Xgwm388 and coincided with one QTl, however Sr40 was not present in the corresponding germplasm (Yu et al. 2011). Allelism testing utilizing both adult plant and seedling testing to identify both qualitative and quantitative resistance loci will be necessary to sort out the allelic relationships among many of the QTl and Sr genes reported on wheat chromosomes. In spite of the complexity of the meta-QTl analysis, using the available information on the Sr genes in the parents of biparental populations and accessions used for association mapping combined with the QTl location based on anchored markers, we were able to identify (a) QTl underlying some previously described Ug99 resistant Sr genes or residual effects of non Ug99 resistant genes and (b) putative locations of new Sr genes.
Colocation of Sr genes and QTl even though Sr31 carried by the rye introgression 1Bl.1rS is not effective to Ug99, the three QTl found on chromosome 1BS, homoeologous to 1rS, are possibly due to a residual effect of Sr31 or another gene on the rye translocation because wheat cultivar PBW343 possesses Sr31.  (Jin et al. 2007) was inconclusive, but because of its tetraploid origin and linkage to centromeric markers, the QTl on 1BS highlighted by letta et al. (2013) is likely conferred by Sr14. Chromosome arm 2B, with 15 QTl, has the second highest number of QTl per chromosome arm, but also the highest number of known qualitative Ug99 Sr genes (6 numerically designated: Sr39, Sr32, Sr36, Sr40, Sr47, Sr28 and SrWeb). Sorting out the allelic relationships among these QTl and at least 11 qualitative Sr genes (seven Ug99 resistant and four non Ug99 resistant Sr genes) on chromosome 2B will be necessary to determine whether any new Sr genes have been detected or if they are the results of residual effects. The three QTl found in the AM durum panel (letta et al. 2013) on chromosome 6Al overlap with Sr13, and even though Sr13 is not common in bread wheat, its presence in durum wheat suggests that the large region highlighted is most likely Sr13. Similarly, further studies are required to determine the allelic relationships among SrTA1662, SrCad, and Sr42 on 6DS (Hiebert et al. 2011;Ghazvini et al. 2012;lopez-Vera et al. 2014). Chromosome 7B has nine QTl, eight of these QTl mapped near Sr17 (position not represented on Fig. 2)  ). Though Sr17 is not effective to Ug99 in a qualitative manner (Jin et al. 2007) it is possible that Sr17 confers a residual APr effect. On chromosome 7D, early studies indicated that Sr57 (Lr34/Yr18/Pm38), Sr58 (Lr46/Yr29/ Pm39) enhanced stem rust resistance in cultivar 'Thatcher' (Dyck and Kerber 1977;Kerber and Aung 1999), and later reports suggested that Sr57 provided APr to stem rust in diverse backgrounds (McIntosh et al. 2012). Our previous studies consistently showed that the STS marker cslV34 was significantly associated with Ug99 resistance in winter and spring CIMMYT wheat panels with major effects or through gene-gene interactions (Yu et al. 2011(Yu et al. , 2012

Putative locations of new Sr genes
With six QTl found on 1A, that chromosome could be a potential new source of new APr. Because Sr1RS Amigo was introgressed from rye and confers qualitative resistance, it is most likely different from the six QTl on 1A. The same is true for the two QTl found on chromosome 1D. even though they overlap with four described Ug99 resistant Sr genes (SrTA1662, Sr33, Sr45, Sr50: rouse et al. 2011Sambasivam et al. 2008;Anugrahwati et al. 2008), the Aegilops tauschii or rye origin of these genes make it unlikely that these QTl are related. The two QTl found on chromosome 2A are likely to represent new Ug99 Sr resistance loci because the only known gene in the region, Sr21 (The 1973), is derived from T. monococcum, and because the parents of the population with the 2A QTl (PBW343/Pavon76: CIMMYT unpublished) are not known to possess T. monococcum in their pedigrees. Also, Sr21 was not in the pedigrees of the AM durum panel (letta et al. 2013). Chromosome group 3 has the second highest number of QTl, and chromosome 3B, with 23 QTl has the highest number of QTl by chromosome arm (Fig. 1). Chromosome arm 3BS has only one Ug99 Sr resistant gene, the slow rusting gene Sr2 and 19 of the QTl found on that arm are likely Sr2. In the Thatcher/Mcneal population, a QTl was coincident with Sr12 on 3Bl, a Ug99-ineffective resistance gene. It is possible that such defeated resistance genes could confer resistance when combined with genes such as Sr57 (lr34/Yr18/Pm38) that have been demonstrated to confer epistatic resistance to stem rust (Kolmer et al. 2011;rouse et al. 2014;Yu et al. 2012). Because Sr12 is present in the cultivar Thatcher and historically was used as an important source of resistance, Sr12 could be widespread in wheat germplasm. Since no resistance genes have been characterized on 3Bl, the QTl in this region are conferred by one or more new resistance genes. Ug99 resistance gene Sr51 has been introgressed into translocations on each of the group 3 homoeologous chromosomes (liu et al. 2011b). Since this gene is derived from Aegilops searsii and is not currently used in agriculture, none of the group 3 QTl are conferred by this gene. Chromosome 3A has three QTl, and the Sr35 cloned gene seems to overlap with the QTl in a CIMMYT population (CIMMYT unpublished) and in the durum panel (letta et al. 2013) on chromosome 3Al, these QTl are not Sr35 because seedling resistance to race TTKSK is absent in the mapping population used by CIMMYT. In addition, the T. monococcum origin of Sr35 is absent from the pedigrees of the panel used by letta et al. (2013). resistance gene Sr27 is also on 3A, but this gene is different from these QTl because it is located on a rye introgression not present in the population used to identify these QTl. On Chromosome 4A, the stem rust resistance gene designated SrND643 (CIMMYT unpublished) is a qualitative gene and the PBW343/Juchi population does not possess that resistance gene so the QTl from PBW343/Juchi is likely conferred by a new resistance gene or allele. The relationships among the QTl identified on 4Al, SrND643, and Sr7 Sears 1954) are not known. The distal end of chromosome 4B appears to be the source of a new Sr gene, although the variance explained for the QTl is low (9 %) and may represent a minor effect APr gene. Chromosome 5A does not carry any previously characterized Sr genes, so the three QTl identified on that chromosome arm likely represent new resistance loci. Although the phenotypic variation for the two QTl identified by letta et al. (2013) on durum wheat is small (R 2 = 4.1 and 4.4 %), further studies of that region could support the existence of new Ug99 resistant loci. The APr gene Sr56 (McIntosh et al. 2012) on chromosome 5Bl was previously designated QSr.Sun-5BL ) (not represented in Fig. 2). Among the 12 other QTl found on 5B, at least four are located on the short arm CIMMYT unpublished;Kaur et al. 2009;Yu et al. 2011) and likely represent new loci of importance. Although Kaur et al. (2009) did not test for APr to Ug99, Yu et al. (2011) also mapped distal 5BS QTl providing strong evidence for minor APr genes located in this region. resistance gene Sr26 on 6Al is present in conventional common wheat germplasm from Australia, but was not present in CIMMYT germplasm when these studies were conducted. Therefore Sr26 does not explain the QTl identified through association mapping (Crossa et al. 2007;Yu et al. 2011). A putative new resistance QTl on 6AS was mapped in the hexaploid biparental PBW343/ Kenya Swara population where no qualitative resistance genes are located (CIMMYT, unpublished). The QTl found on 6Al in the tetraploid population Kristal/Sebatel is likely a new source of resistance since Sr13-linked markers are in the distal region of 6Al. Chromosome 6B seems to be a rich source of new Sr genes with nine QTl detected. The relationship between the QTl on 6Bl and Sr11 is not known but Sr11 is ineffective against Ug99 and is not represented on Fig. 2. Chromosome arm 6BS appears to be a new source of Sr genes (Crossa et al. 2007;Yu et al. 2011;CIMMYT unpublished). The phenotypic variation explained by these QTl ranged from 5 to 14 % and most likely correspond to minor APr genes. Singh et al. (2011) detected a distal QTl that explained 56 % of the variance for stem rust on chromosome 6BS (XwPt4283/XwPt7207) in the bi-parental cross between PBW343/Kenya nyangumi. Validation of that QTl could be accomplished using the putative allele carrier, Kenya nyangumi, in another bi-parental cross. The QTl on 7AS, either from durum or bread wheat, once validated, would represent new stem rust resistance genes as Sr22 is located on 7Al. Since Sr22 was not present in the parents of the hexaploid mapping populations, these QTl are not conferred by Sr22. The more distal QTl found in the Kristal/Sebatel bi-parental tetraploid population (QSr.1PK-7A.2) is likely to be conferred by Sr22 (Haile et al. 2012). Since Sr genes have not been previously identified on 7BS, the QTl identified in this region are likely new.
Overall, the growing number of characterized Sr genes and QTl demonstrates that there is still potential for discovering new APr genes with varying levels of effect. This consensus map will facilitate the identification of new resistance genes and QTl and aid in the development of improved markers to increase breeding efficiency and the pool of alleles that are important for the control of Ug99.