Genetic diversity and population structure of Peronosclerospora sorghi isolates of Sorghum in Uganda

Sorghum is the third most important staple cereal crop in Uganda after maize and millet. Downy mildew disease is one of the most devastating fungal diseases which limits the production and productivity of the crop. The disease is caused by an obligate fungus, Peronosclerospora sorghi (Weston & Uppal) with varying symptoms. Information on the genetic diversity and population structure of P.sorghi in sorghum is imperative for the screening and selection for resistant genotypes and further monitoring possible mutant(s) of the pathogen. Isolates of P. sorghi infecting sorghum are difficult to discriminate when morphological descriptors are used. The use of molecular markers is efficient, and reliably precised for characterizing P. sorghi isolates. This study was undertaken to assess the level of genetic diversity and population structure that exist in P. sorghi isolates in Uganda. A total of 195 P. sorghi isolates, sampled from 13 different geographic populations from 10 different regions (agro-ecological zones) was used. Eleven (11) molecular markers, comprising of four Random amplified microsatellite (RAM) and seven (7) Inter-Simple Sequence Repeat (ISSR) markers were used in this study. The analysis of molecular variation (AMOVA) based on 11 microsatellite markers showed significant (P < 0.001) intra-population (88.9 %, PhiPT = 0.111) and inter-population (8.4 %, PhiPR = 0.083) genetic variation, while the genetic variation among regions (2.7 %, PhiRT = 0.022) was not significant. The highest genetic similarity value (0.987 = 98.7 %) was recorded between Pader and Lira populations and the lowest genetic similarity (0.913 = 91.3 %) was observed between Namutumba and Arua populations. The mean Nei's genetic diversity index (H) and Shannon Information Index (I) were 0.308 and 0.471 respectively. Seven distinct cluster groups were formed from the 195 P. sorghi isolates based on their genetic similarity. Mantel test revealed no association between genetic differentiation and geographical distance (R2 = 0.0026, p = 0.02) within the 13 geographic populations.

INTRODUCTION Downy mildew disease of sorghum, caused by an obligate soil-borne fungus Peronosclerospora sorghi (P. sorghi) [Weston and Uppal (Shaw)] (Frederiksen, 1980) poses a serious biotic stress to sorghum production and productivity worldwide (Williams, 1984). The prevalence and distribution of the disease is favoured by factors such as high relative humidity (Wang et al., 2000), low temperature  and rainfall which favours conidia production and subsequent development of the disease. When the dis ease is not managed at the early stages of infection, losses can reach 100 %. Infected plants show localized and systematic symptoms varying from lesions on leaf lamina to chlorosis . Previous reports suggest the prevalence and distribution of Sorghum downy mildew disease (SMD) in Uganda at varying incidence and severity levels (Bigirwa et al., 1998;Kumi et al., 2018), but no study has been done to unravel the state of genetic diversity of P. sorghi isolates in Uganda.
Morphological and molecular characterizations are two methods commonly used to assess genetic variability of P. sorghi (Bock, 1995). Unlike molecular characterization, morphological characterization is exclusively dependant on morphological traits of P. sorghi which come with several shortcomings ranging from; influence by environmental factors, time consuming and lack of efficient resolution power required to discriminate between genetically related isolates. More so, the use of morphological descriptors to assess genetic variation among isolates of P. sorghi is reported to be limited (Bock, 1995).
The advent of molecular markers has made it easier to study genetic diversity and population structure of P. sorghi with much precision and better accuracy (Perumal et al., 2006). Random Amplified Polymorphic DNA (RAPD) primers, Simple Sequence Repeats (SSRs) and Amplified Fragment Length Polymorphism (AFLP) have been extensively used to explore the variability among P. sorghi isolates (Bock et al., 2000) and different pathotypes of P. sorghi in sorghum (Perumal et al., 2008). Compared to other parts of the world, pathogenic and molecular variability among the P. sorghi isolates of sorghum have been well documented (Bock et al. 2000;Mathiyazhagan et al., 2008). Outcome of such research confirmed the existence of genetic diversity in P. sorghi and led to identification of new pathotypes (Perumal et al., 2006). Perumal et al. (2006) analyzed the genetic variability among the 14 isolates of P. sorghi including metaxyl resistant isolates and reported that approximately 25% of the bands were polymorphic across the isolates in the tested populations. Mathiyazhagan et al. (2006) also reported similar genetic variability between isolates from sorghum and corn in India, using restriction fragment length polymorphism analysis of the polymerase chain reaction (PCR). Sequence characterized amplified region (SCAR) marker have also been used to assess genetic variability of P. sorghi (Ladhalakshmi et al, 2009).
Knowledge regarding the extent of genetic diversity and genetic relationships of P. sorghi in Uganda will be valuable for designing a comprehensive breeding strategy for identifying resistant sorghum genotypes that will be resilient to SDM disease. Thus, studying and quantifying genetic diversity of P. sorghi from different agro-ecologies will offer a valuable marker assisted selection and breeding strategy. The objective of this study was to characterize the isolates of P. sorghi in Uganda by assessing the population structure, genetic diversity and/or relatedness using ISSR and RAMS markers.

II. METHODOLOGY Collection of samples
Hierarchical sampling technique was used to collect a total of 195 SDM samples from 13 districts (Table 1 and Fig 1) covering all the ten agro-ecologies in Uganda. These samples were taken from sorghum plants with leaves showing systemic and/or localized characteristic lesions of mildew with conspicuous conidia growth. Leaves samples were carefully covered with aluminum foil, labeled and stored in a cooler. DNA extraction and purification DNA extraction was done following the protocol described by McDermott et al. (1994). Conidia found on leaves were collected using a camel hair brush and transferred to a 1.5-ml microcentrifuge tube containing 500 μl of extraction buffer (50 mM Tris -HCl, pH 8.0; 0.7 M NaCl and 1% SDS) (1 × 108 spores/ml), vortexed for 30 s and incubated at 60 o C for 1 h. After incubation, the mixture was centrifuged at 12,000 × g for 10 min, and the aqueous phase was collected and extracted twice with an equal volume of phenol: chloroform: isoamylalcohol (25:24:1). The aqueous phase was transferred to a 1.5-ml microcentrifuge tube and the DNA was precipitated by addition of an equal volume of cold isopropanol and incubation at -20°C for 1 h.  The DNA was pelleted by centrifugation at 12,000 × g at 4°C for 10 min. The pellet was washed twice with cold 70% ethanol, air-dried and re-suspended in 50 μl of Tris -EDTA buffer (10 mM Tris-HCl and 1 mM EDTA, pH 8.0). The genomic DNA was checked by agarose gel electrophoresis and the concentrations of the DNA were determined using a NanoDrop spectrometer nm (Thermo Scientific, Waltham, MA) by measuring the absorbance at 260.
A total of twenty (20) primers, consisting of nine (9) Random Amplified Microsatellites (RAMS) and eleven (11) Inter Simple Single Repeats (ISSRs) ( Table 2) were used to screen the 195 P. sorghi isolates, but eleven (11) primers yielded clear visible PCR products and those were used in this study.

Data collection and analyses
Data collected from the amplified multiloci bands which were clearly visible were scored manually for each isolate using a binary format according to presence (1)   The binary data were subjected to analysis of molecular variance (AMOVA) ( Reyes-Valdés et al., Excoffier et al., 1992) with 999 permutations using GenAlEx6.5 (Peakall & Smouse, 2006;2012). The total genetic variation among 195 isolates was generated using phi-statistic through AMOVA. The genetic variations were partitioned into three; variation among regions (PhiRT), variation among population (PhiPR) and variation within population (PhiPT).
PhiPT coefficient values denote the proportion of estimate variance among population relative to the total variance and the pairwise between populations expressed as PhiPT = (VAP + VAR/ (VWP + VAP), where VAP is the estimate of variance among populations, VAR is the estimate of variance between the geographical regions, and VWP is the estimate of variance within the studied population. PhiPT was used to determine the genetic differentiation between the population, it is a measure which allows intra-individual variation suppression when comparing binary and codominant data (Teixeira et al., 2014).
The genetic variability of the populations (districts) was analyzed using the Hardy-Weinberg equilibrium (HWE) (Crow et al., 2008) assumption in GenAlEx 6.5 (Peakall & Smouse, 2006;2012). Pairwise Nei's genetic distances (Nei, 1972) and  4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 (Peakall & Smouse, 2006;2012) was performed to examine whether genetic isolation was associated with the geographic distance among P. sorghi populations. The pairwise Nei's population genetic distances were calculated based on gene frequency differences between populations, and these distances were then compared to geographic distances between populations and a correlation was run for these two parameters. The genetic distance matrix generated from 195 isolates in GenAlEx was used to perform a hierarchical cluster analysis based on Ward's hierarchical agglomerative clustering method (Ward, 1963) using R statistical package (R Core Team, 2013). A cluster dendogram was generated for the isolates (Maechler et al., 2016). In other to determine the optimal number of clusters for the dendogram, the Bayesian Inference Criterion (BIC) (Akogul, 2017) for k means method was used. This method deploys expectationmaximization, initialized by hierarchical clustering for parameterized Gaussian mixture models (Reynolds, 1992).
Following the cluster analysis, Discriminant Analysis of Principal Components (DAPC) (Jombart et al., 2010) was carried out in R statistical package to assess the relationships between the different clusters, using a method that focus on variability between-group, while neglecting variability within-group variation, which is precisely the rationale of Discriminant Analysis (DA) (Lachenbruch & Goldstein, 1979) to achieve the best discrimination of isolates into pre-defined groups. DAPC scatter-plot was generated which allows for a graphical assessment of the genetic structures between clusters.

III. RESULTS Genetic variation and relationship in P. sorghi at varied population levels
Statistical analyses which were performed on the total of 120 amplified polymorphic loci from PCR reaction for P. sorghi isolates using ISSR and RAMS revealed a wide genetic diversity and structure in the population. The overall genetic differentiation was examined for the 195 P. sorghi isolates from 13 geographic populations (Districts) covering 10 regions (AEZ) in Uganda ( Table 3). The AMOVA results are presented below. Df= degree of freedom, SS= sum of square, Phi statistic, P value is based 999 permutation. *** = Significant at P < 0.001 Partitioning of genetic differentiation at three levels (among population, within population and among regions) contributed varying degrees of genetic variation to the total variation observed. Genetic variation among and within population of the 195 isolates examined by AMOVA were significant (P <0.001), while genetic variation of isolates among regions where not significant. Variation within population accounted for 88.9 % of the total genetic variance observed in P. sorghi isolates, this means the highest genetic diversity of these isolates occurred within population level. Variation among population and regions contributed 8.4 % and 2.7 % genetic variation respectively to the total diversity of P. sorghi. Additionally, significant (P <0.001) Phi values for genetic diversity were recorded among population (PhiPR = 0.083) and within population (PhiPT = 0.111) while the Phi value recorded among regions (PhiRT = 0.022) was not significant.
The results from the analysis of genetic diversity among 195 isolates of P. sorghi from13 different populations are presented in Table 4.

Relationship for genetic distance and geographical distribution
The results of pairwise Nei's genetic distance between geographical population of P. sorghi (above diagonal) and pairwise Nei's genetic similarity between geographical populations of P. sorghi (below diagonal) are presented in Table 5. The general Nei's genetic distance (genetic difference) values recorded between the tested populations of P. sorghi were very low, ranging from a minimum of 0.014 to maximum of 0.091. The smallest genetic distance was observed between Lira and Pader population (0.14), both of which are in the same agro-ecological zone (Northern grassland), while the largest genetic distance was observed between Arua and Namutumba population in which fall within West Nile and Lake Victoria Crescent ecological zones respectively.  The results from Mantel test to examine whether genetic distance was associated with geographic distance among P. sorghi populations revealed no association (

Cluster analysis
The 195 P. sorghi isolates from 13 different geographic populations clustered into seven distinct cluster groups (Fig. 4) which indicates genetic diversity within the population. Each cluster was mutually exclusive of isolates identity but not in geographical y = 0.0085x + 1.4836 R² = 0.0026 Cluster 1 constituted isolates from Kabarole and Namutumba while cluster 4 constituted population of Arua, Iganga, Kumi, Masindi and Pader. Furthermore, clusters 5 and 6 shared nine (9) geographical populations namely; Arua, Pallisa, Kumi, Kabale, Hoima, Masindi, Lira Pader and Serere. But in addition to the aforementioned populations, cluster 6 also Iganga. Lastly, cluster 7 consisted of populations namely; Nebbi, Iganga and Pader. Following cluster analysis and mantel test, Discriminant analysis was carried using the detected number of clusters from the dendogram computed from the Bayesian Information Criterion (BIC, which employs k means) Model to determine whether or not population structure exist within the P. sorghi isolates. DAPC results (Fig. 5) showed genetic variability among the isolates but there was no clear separation pattern among the study population. The results from the DAPC analysis showed no clearly defined population structure and a total of seven discriminant eigenvalues for principal components (PCs) were retained. The proportion of variance conserved by the PCA principal components accounted for 36.9 % variance. Discriminant analysis eigenvalues for PC1, PC2 and PC3 were 40.436, 17.523 and 9.166 respectively. IV. DISCUSSION Genetic variability is significant to plant breeders for purposes of screening, selection and developing improved crop varieties that confers resistance to biotic stresses and are highly adaptable/ resilient to varied abiotic conditions while offering high quality yield and products for industrial purposes. In this present study, two different primer sets, Random Amplified Microsatellites (RAMs) and Inter-Simple Sequence Repeats (ISSRs) were used to assess genetic divers ity and population structure of 195 P. sorghi isolates from 13 different populations in Uganda. The combination of these two molecular markers was effective in evaluating the genetic diversity, population structure and estimating genetic variations.
The observed genetic variation in this study was partitioned within population rather than among population or among regions (Table 3). These results indicate that majority of the genetic differentiation (88.9 %) existed within the defined population of P. sorghi isolates and among populations (8.7 %). However, the insignificant genetic differentiation value of P. sorghi isolates among regions (2.7 %) in this study showed that genetic differentiation was independent of the geographic regions (agro-ecological zone) of the isolate. These result agreed with other findings on genetic diversity in P. Results from this study also showed high percentage of genetic polymorphism (94.04 %) (  (Mathiyazhagan et al., 2008) who also reported high percentage polymorphism (33 %-100%) in P. sorghi isolates.
Low genetic diversity index (0.304) ( Table 4) and high genetic similarity (98.7 %) ( Table 5) values recorded in this study was not surprising because P. sorghi is reported to exhibit high sexual recombination (Heffer-Link et al., 2002) and therefore explained the high level of inbreeding. Another contributing factor could be exchange of P. sorghi infected sorghum seeds among farmers for cultivation and thereby perpetuating the spread and development of the pathogen (seed-borne). These results were similar to findings of Sireesha & Velazhahan (2015) and Mathiyazhagan et al., (2008) who reported high genetic simmilarity values of 93 % and 90 % respectively within P. sorghi isolates. Their results further reported low genetic diversity.
Mantel test results between geographic distance and genetic differentiation among the 13 geographic populations of P. sorghum revealed no significant correlation (Fig. 3). This result showed that, the observed genetic variability of P. sorghi isolates (within population) from the 13 different populations was not structured in geographic space (there is no spatial structure). Cluster analysis result (Fig. 4) revealed that, P. sorghi isolates in Uganda could cluster into seven (7) genetically distinct groups according to genetic similarity/identity of the populations. These findings suggested that geographical origin of P. sorghi isolates has no influence on the clusters formation.
Discriminate analysis result (Fig. 5) showed no clear well-defined pattern of clusters of genetic structure among the study populations. DAPC analysis therefore confirmed the Mantel test results, that the observed genetic variability of P. sorghi isolates within the study populations was not spatially structured.

V.
CONCLUSION The study revealed a high genetic variation of P. sorghi within populations of Sorghi in Uganda. A weak association of genetic isolation by geographic distance was established for P. sorghi populations, which suggested that the observed genetic differences of P. sorghi was unaffected by the geographical populations. Seven genetically distinct clusters groups were formed from P. sorghi isolates according to the genetic similarities.