Breeding Strategy for Resistance to Striga Asiatica Based on Genetic Diversity and Population Structure of Tropical Maize Lines


 Maize (Zea mays L.) is a major staple crop in southern Africa and is produced on millions of hectares. However, its yield is greatly reduced by Striga spp, a parasitic weed which is causing US$ 7 billion losses annually. Use of host resistance could be an effective way of controlling Striga and resistance to Striga is quantitative, mainly controlled by additive gene action. Understanding the population structure and genetic diversity is therefore key in designing an effective breeding program targeting grain yield heterosis and resistance to Striga. The aim of this study was to determine the genetic diversity and population structure of the key germplasm from tropical Africa. This information could guide in the identification of heterotic groups and potential testers required to kick start a maize breeding program for Striga asiatica in southern Africa. A total of 222 maize inbred lines from IITA and CIMMYT were used in this study. The materials were genotyped using the genotyping-by-sequencing method. A total of 45 000 SNP markers were revealed, and these were subjected to analysis of molecular variance, structure analysis and clustering using the Gower’s distance and neighbor joining algorithm. Molecular variance was lager within individuals (91%) than among populations (9%). The inbred lines clustered into three major groups, with the IITA germplasm clustering separately from CIMMYT germplasm. A breeding strategy for Striga asiatica resistance was proposed with the aim of increasing genetic gains in both the resistance and grain yield.


Introduction
Maize is the main preferred staple in southern Africa with consumption rate averaging about 100kg per capita per annum (Epka, 2019). However, Striga spp is one of the major biotic factors affecting maize production in Africa (Ejeta and Gressel, 2007) and is considered among the world's worst weeds (Shayanowako et al., 2018) causing up to US$ 7 billion loss annually in Africa (Rubiales et al., 2009). Striga spp is an obligate parasite that draws nutrients and water from its host (Ejeta and Gressel, 2007). There are two types of Striga spp which are Striga asiatica and Striga hermonthica that are prevalent in southern Africa and the rest of Africa, respectively. The widely reported cultural, biological and chemical control options for Striga spp are not feasible for the resource limited farmers in sub-Saharan Africa (Joel et al., 2007). Use of host resistance has been effective in controlling pests, diseases and weeds in various crops.
The International Institute of Tropical Agriculture (IITA) managed to incorporate resistance from wild relatives of maize, Zea diploperennis and Tripsacum dactyloides (Rispail et al., 2007). Resistance was also sourced from maize populations in east Africa where the cereals have co-existed with this parasite for long. Subsequently, a number of inbred lines and hybrids with resistance to Striga hermonthica were developed and were shared across various maize breeding programs in the rest of Africa. Resistant genotypes usually show few Striga root attachments and little Striga germination stimulant (strigolactones) production (Rank et al., 2004). Other resistance mechanisms include reduced owering and reduced seed set of the Striga species (Awad et al., 2006). These mechanisms of resistance were found to be effective in controlling Striga asiatica in southern Africa (Gasura et al. In Press). Resistance to Striga spp was found to be controlled mainly by additive gene action (Gasura et al. In Press).
Germplasm from IITA has novel sources of resistance to Striga spp while germplasm from the International Maize and Wheat Improvement Centre (CIMMYT) is widely adapted to many regions including southern Africa. In order to maximize heterosis using germplasm from IITA and CIMMYT, there is need to understand the population structure and genetic diversity of these materials (Mengesha et al., 2017). This information is crucial in the identi cation of potential testers and prediction of potential heterotic groups, which are some key determinants of an effective breeding program in maize (Laborda et al., 2005;Nyombayire, 2016).
Plant breeders need to cross inbred lines that are from different heterotic groups to maximize heterosis.
The longest and expensive period during a hybrid production is when selecting parents that when crossed To begin a maize hybrid program one has to use a well-documented germplasm with well-known heterotic groups and patterns (Moose and Mumm, 2008). Methods such as geographical information, phenotypic traits, pedigree information, combining ability and the use of molecular markers (Wende et al., 2013) have been widely used to classify genotypes. In the temperate regions, the Reid * Lancaster heterotic pattern has been developed using pedigree analysis and also in Europe, the European int * Maize Belt dent have been developed based on phenotypic markers (endosperm types) ( . The aim of this study was to determine the genetic diversity and population structure of the key germplasm from CIMMYT and IITA using SNP markers. This information could guide in the identi cation of heterotic groups and potential testers required to kick start a maize breeding program for Striga asiatica in southern Africa.

Plant materials
A total of 222 maize inbred lines comprising of 192 inbred lines from CIMMYT, Harare, Zimbabwe and 30 inbred lines from IITA, Ibadan, Nigeria were used in the study. The names and pedigrees of this germplasm is provided in Table 1.

Genomic DNA isolation
Seeds were shipped to the Biosciences for East and Central Africa (BeCA) at the International Livestock Research Institute (ILRI) (BeCA-Hub-ILRI) Kenya. The seeds of the 222 inbred lines were germinated and the DNA was extracted from fresh tissues that were one week old using the modi ed CTAB method (Saghai-Maroof et al., 1986). The DNA was checked for quality using the agarose gel and quantity using a spectrophotometer. Genotyping was done using Genotyping-By-Sequencing ( (Earl, 2012) visualized the structure analysis results following the Evanno approach. RStudio software was then used for cluster analysis to depict the inferred groups using the Gower's distance (Gower, 1971) and neighbor joining algorithm. The silhouette plots using RStudio like structure results also suggested three groups and the dendrogram was then sub-divided into the 3 groups using the cutree option in RStudio (Team R, 2015).

Genetic diversity among genotypes and populations
The total molecular variation was partitioned into among population and within populations. Larger genetic variability (91%) was attributed to variation within populations and the remaining 9% variation was explained by variation among populations (Table 2). Population structure, cluster analysis and genetic distances The population structure of the germplasm was suggested following the admixture model. From the prosed K = 10 groups, the Evanno criterion (Earl, 2012) suggested three distinct groups (Fig. 1).
The dendrogram generated using the Neighbor joining algorithm based on the 45 000 SNP markers grouped the 222 inbred lines into three major clusters (Fig. 2) as suggested by the silhouette plots and the Evanno criterion. The rst cluster had two inbred lines, one from CIMMYT (T396-326) and another one from IITA (TZEI4). The remaining 220 inbred lines belonged to the second and the third clusters that are also partitioned clearly into IITA and CIMMYT maize lines. These two groups had some several subgroups within them. Inbred lines within some sub-groups were also clustered based on their heterotic groups for example CML 395 in group A, clustered separately from CML 444 in group B. Furthermore, in most cases inbred lines in CIMMYT heterotic group A grouped together while inbred lines that are in CIMMYT heterotic group B also grouped separately within their sub-cluster.

Discussion
In a panmictic population, the among population variance is expected to be minimal or absent (Meirmans, 2006). The small among population variance observed could be due to the different selection systems being conducted by IITA and CIMMYT since they target different breeding agro-ecological regions. This could explain why the IITA materials grouped separately from the CIMMYT materials, thus the groups are in total agreement with the sources of germplasm. The occurrence of T396-326 and TZEI4 in the same group yet they come from different sources is an example of some common inconsistencies in clustering. Although mutation, selection and genetic drift can lead to the alignment of inbred lines from The high within population genetic variation observed is mainly due to the enrichments efforts to widen the genetic base of the breeding materials of these organizations. Indeed, within each group there were many sub-clusters that shows the existence of huge variation within the group. Inbred line TZEI4 that clustered with only one CIMMYT line T396-326 shows that it has unique properties as compared to all other inbred lines from IITA. This shows that TZEI4 can be crossed to the rest of CIMMYT lines in this collection and still expressing high heterosis. However, it is unclear to predict the heterotic groups of this line together with the rest of the CIMMYT lines because most of the IITA lines and CIMMYT lines clustered separately. We expected lines of the same heterotic groups to cluster together thus showing the relationship between IITA and CIMMYT germplasm. One possibility could be that all the IITA materials belong to a heterotic group completely different to that of CIMMYT, hence the existence of these two major groups.
Genetic diversity information enables breeders to take stock of available genetic variation, conserve and e ciently utilize their materials in various breeding programs (Bidhendi et al., 2012). Resistance to Striga was shown to be controlled by mainly additive gene action, suggesting that resistant lines must be crossed when formulating hybrids. However, grain yield, the nal target trait, is mainly controlled by nonadditive gene action. To guide the future breeding programs that aim at the simultaneous improvement of Striga resistance and grain yield in maize, the following program is proposed for tester identi cation, as well as line and hybrid development; 1. Cross eight (8) IITA materials from different sub-cluster (including TZEI4 that is highly distinct) to 20 CIMMYT lines in different sub-clusters using a line x tester scheme 2. Evaluate the 160 testcross materials under optimum condition to get the speci c combini ability (SCA ) effects for grain yield. Also evaluate the 160 materials in the laboratory and glasshouse to get the Striga asiatica resistance parameters.
3. Select two testers one for group A lines and two testers for group B lines. Lines with negative SCA effects are in the same group, while lines with positive SCA effects are in different groups. The testers must have positive general combining ability (GCA) effects for Striga asiatica resistance. 4. Line improvement will be done within heterotic group using Striga asiatica resistance materials from IITA that are within that group.
5. Line selection will be done by testcross performance evaluation using a tester from a different heterotic group. A desirable line must have both positive GCA effects for grain yield and negative GCA effects for Striga resistance parameters. 6. Desirable hybrids will be made by crossing lines from different heterotic groups but having positive GCA effects for grain yield and negative GCA effects for Striga resistance parameters.
7. Hybrids will be evaluated for Striga asiatica resistance in the laboratory and greenhouse. Only hybrids with minimum thresholds in terms of resistance to Striga will be taken to the eld for preliminary yield trials thus reducing the number of hybrids to be taken to the eld. These will be followed by multi-locational testing on at least ve locations for two seasons as required by the variety release committee.
The ultimate goal of every breeding program is to improve its e ciency as determined by the number of hybrids produced per unit time. In this regard, to improve the selection e ciency and genetic gains, we propose the use of molecular markers for screening of Striga resistance, major diseases, and then complement that information from the eld evaluation of grain yield performance based on the combining ability of the testcross hybrids.

Conclusions
Molecular genetic diversity has been clearly seen in this study and is largely located within individuals (91%) with 9% among populations. The inbred lines were clustered into three major groups, with the IITA germplasm clustering separately from CIMMYT germplasm. A breeding strategy for Striga asiatica resistance was proposed with the aim of increasing e ciency in genetic gains for both the resistance and maize grain yield.  Table   Table 1 not available with this version. Figure 1 Page 11/11

Figures
Number of groups inferred from structure software Inferred groups and clustering of the 222 IITA and CIMMYT maize inbred lines based on 45 000 SNP markers using the Gower's distance and the neighbor joining algorithm.