Magnaporthe oryzae populations in Sub-Saharan Africa are diverse and show signs of local adaptation

Rice blast caused by Magnaporthe oryzae is one of the most economically damaging diseases of rice worldwide. The disease originated in Asia but was detected for the first time in Sub-Saharan Africa (SSA) around 100 years ago. Despite its importance, the evolutionary processes involved in shaping the population structure of M. oryzae in SSA remain unclear. In this study, we investigate the population history of M. oryzae using a combined dataset of 180 genomes. Our results show that SSA populations are more diverse than earlier perceived, and harbor all genetic groups previously reported in Asia. While M. oryzae populations in SSA and Asia draw from the same genetic pools, both are experiencing different evolutionary trajectories resulting from unknown selection pressures or demographic processes. The distribution of rare alleles, measured as Tajima’s D values, show significant differences at the substructure level. Genome-wide analysis indicates potential events of population contraction strongly affecting M. oryzae in SSA. In addition, the distribution and haplotype diversity of effectors might suggest a process of local adaptation to SSA conditions. These findings provide additional clues about the evolutionary history of M. oryzae outside the center of origin and help to build customized disease management strategies.


138
downloaded from the Sequence Read Archive (SRA, http://www.ncbi.nlm.nih.gov/sra). A 139 summary of the dataset's sequencing yield and coverage can be found in Table S2.

141
Phylogenetic and population analysis 142 The phylogenetic tree was built with RAxML 8.2.9 (Stamatakis, 2014). Statistical confidence for 143 each node was set to 1000 bootstrap runs, utilizing the general time-reversible model of nucleotide 144 substitution with the Gamma model of rate heterogeneity. The phylogenetic tree was visualized 145 with the ggtree R package (Yu et al., 2017). We also performed a phylogenetic network analysis   Information Criterion were also performed in adegenet. We used another approach to infer the 153 optimum number of clusters by calculating the, Silhouette score in the R package factoextra 154 (Kassambara et al., 2017). To estimate genetic variation, the proportion of genetic variance due to 6 population differentiation and a significant departure from neutrality, we calculated the genome-156 wide nucleotide diversity (Pi), Wright's fixation index (Fst), and Tajima's D with the variants call 166 bwa-mem 0.7.17 (Li & Durbin, 2009). Samtools coverage from samtools 1.10 was used to 167 calculate the mean coverage of each gene in each of the 180 blast isolates, with the minimum read 168 depth set at 3x. The total number of mapped reads of each gene was divided by the length of that 169 gene in the reference (Li et al., 2009). The threshold set to determine the presence of an effector 170 gene was 80% coverage. A binary presence/absence matrix was created (Table S3)   manually curated before any test was performed. Effector diversity indices were obtained using 183 the R package pegas (Paradis, 2010

214
It is also suggested that this lineage emerged from a recombinant population with distinct genetic  Table   236 S4), Asia accumulate more diversity than SSA in every group ( Figure 2A; Table S5). The fixation    Table S5), which might point out to non-random removal of rare 256 alleles from the population. Since not all the SSA groups experience significant differences in 257 Tajima's D, one hypothesis is that a sudden population contraction occurred in the SSA region, 258 specifically affecting genetic groups 1 and 2. Interestingly, the differences in Tajima's D appears 259 to be scattered across the genome rather than concentrated in particular chromosomic regions 260 ( Figure 2D; Figure S5). For SSA genetic group 1, these differences appear to be more pronounced

291
We then used a subset of 75 effectors to assess presence/absence polymorphism in the M.

292
oryzae genomes (Table S3). We found that effector repertoires tend to have similar but not exact

300
To assess the diversity of M. oryzae effectors across regions, we extracted consensus sequences 301 from all genes and identified allelic variants. The number of haplotypes in a subset of 96 effectors 302 ranged from 1 to 16. The overall haplotype diversity (Hd), measured as the probability of finding 303 different alleles, was higher in Asia compare to SSA for all groups. The same was also true when 304 computing for nucleotide diversity ( Figure 4A-B). We then assessed the diversity of each effector 305 in each genetic group and found that genetic group 1 is statistically more diverse than any other 306 genetic groups ( Figure S7A-B). This observation aligned with the genome-wide diversity in Figure   307 S3A. While more effector haplotypes were observed in Asian populations, unique effector 308 haplotypes were present in SSA genomes ( Figure S8). The presence/absence distribution and 309 sequence differences in SSA genomes suggesting a certain level of adaptation.

336
The migration of plant pathogens to new agricultural ecosystems represent a major concern for 337 long-term strategies of food security. Understanding the events that shaped the pathogen 338 population structure in the new setup is likely to help to develop effective control measurements.

339
In this report, we reconstructed the evolutionary trajectory of the rice blast pathogen M.    as <80% of coverage. Gene names (rows) and isolate names (columns) are described in Table S1.

452
Low frequency (yellow) and high frequency (red) is based on the total number of isolates that have 453 that particular haplotype in each effector. The color in the text is based on the four genetic groups 454 inferred in Figure S2.

456
Additional Data S1. The complete effector sequences files for each of the180 strains in fasta file