Global FAW population genomic signature supports complex introduction events across the 1 Old World 2 3

Accurate genomic knowledge can elucidate the global spread patterns of invasive pests. The high- 17 profile invasive agricultural pest Spodoptera frugiperda (fall armyworm; FAW) is a case in point . Native to the 18 Americas, the FAW was first reported in West Africa in 2016 and has rapidly spread to over 64 countries across 19 the Old World, resulting in significant economic losses. The chronological order of reported detections has led 20 to the hypothesis that the FAW moved eastwards across Africa and then Asia, however genomic evidence 21 remains lacking to test this hypothesis and to identify the potential origin of invasive populations. Using a whole 22 genome sequencing approach, we explored the population genomic signatures of FAW populations from the 23 Americas and the Old World. Analyses of complete mitochondrial DNA genomes identified 12 maternal lineages 24 across the invasive range, while genomic signatures from 870 high-quality nuclear genome-wide single 25 nucleotide polymorphic (SNP) DNA markers identified five distinct New World populations that broadly reflected 26 their native geographical ranges and absence of host-plant preference status. Unique FAW populations in the 27 Old World were also identified that suggested multiple introductions underpinned their rapid global spread. We 28 identified in Asian FAW individuals, genomes lacking evidence of admixture; while analysis of identified complex 29 substructure revealed significant directional geneflow from Asia into East Africa, in contrast to a simple east-to- 30 west spread. Our study highlights the need for population genomics approaches in analysing complex pest 31 invasions, and the importance of international partnership to address global biosecurity challenges presented 32 by emerging high priority insect pests.

3 What is missing from current research into the spread of FAW is analysis of broader genomic evidence. 4 Genome-wide single nucleotide polymorphic (SNP) markers aligned to well-annotated genomes can provide 5 powerful genomic evidence for understanding introduction pathways (e.g., 1 ) and eliminate candidate 6 populations 14 as well as elucidate hybrid signatures 13 . The genomes of both Sfr and Sfc have been sequenced 7 and annotated 44 , allowing higher resolution analysis of genetic structure, migration patterns and sub-species 8 status based on a high number of genome-wide SNPs to enable identification of the potential New World origins, 9 and the species and admixture status of the invasive Sfc and Sfr populations.

10
In this study, we provide an assessment of global FAW movement history based on genomic data that 11 incorporates populations from both Northern, Central, and Southern Americas, and the Caribbean (i.e.,  Table 1). The 27 initial differentiation of these individuals as 'corn-preferred' or 'rice-preferred' was based on the partial mtDNA 28 COI gene region (see 55 (corresponding to BC55 nt6065-6092; nt9544-9580;   2   nt12807-12838; nt15047-15276) were trimmed due to alignment difficulties and low genome assembly   3   confidence associated with simple repeat units, resulting in all samples having final mitochondrial DNA genome   4 length of 15,059 bp. We identified unique mitogenome haplotypes using the DNAcollapser in FaBox (1.5) 5 <https://users-birc.au.dk/~palle/php/fabox/dnacollapser.php> 73

15
In this study, we used the original assembled genome of Gouin et al. 44 for our raw data processing. Genomic 16 raw data was cleaned and trimmed using Trimmomatic and aligned to the S. frugiperda (rice v1) genome 44

46
To estimate directional gene flow between the populations, as well as the relative magnitudes of these flows, the function divMigrate in the R package diveRsity 100 online version was used 48 <https://popgen.shinyapps.io/divMigrate-online/> 101 . Gene flows between all sites were calculated and then 49 normalized to obtain relative migration rates (between 0 and 1). The program divMigrate searches for gene flow 50 directionality between each pair of populations by identifying significant asymmetry using a hypothetically 51 . CC-BY-NC-ND 4.0 International license (which was not certified by peer review) is the author/funder. It is made available under a The copyright holder for this preprint this version posted August 25, 2020. . https://doi.org/10.1101/2020.06.12.147660 doi: bioRxiv preprint 5 defined pool of migrants for each pair. This pool is then compared to pairs in the overall population to calculate 1 directional genetic differentiation and relative migration. To evaluate the significance of asymmetric migration, 2 1,000 bootstraps were performed. Resulting migration matrices were then plotted using Gephi 3 <https://gephi.org/> to generate network graphs. These show directional gene flows between populations 4 (located at the nodes), with the thickness of the lines showing relative strength of gene flow.

7
A total of 197 FAW individuals were sequenced, 102 from the native New World range and 95 from the 8 invasive Old World range (Fig. 1). From the pest's native range, we detected 25 'rice' mitochondrial genome (i.e., 9 mitogenome) haplotypes, and 51 'corn' mitogenome haplotypes. All FAW from Mexico and Peru had the 'corn' 10 mitogenome while FAW from Guadeloupe and French Guiana were all 'rice' mitogenomes. Of the FAW from the 11 invasive range nine 'corn' and 'rice' mitogenome haplotypes were identified; one of the 'corn' mitogenome  The trimmed (15,059bp) mitochondrial DNA genome phylogeny of all individuals in our study identified 2 two distinct clades that corresponded to the 'rice-preferred' and 'corn-preferred' clusters ( Fig. 2). Based on the 3 near complete mitogenome phylogeny, a minimum of four and five introduction events were likely associated 4 with the 'rice' and 'corn' maternal lineages, respectively (Fig. 2). Except for the 'corn' specimen (CH06) from 5 Yunnan that clustered strongly with an individual from Mississippi (UM04), all 'corn' individuals from the invasive         (Figs. 5, 6B), one of which originated from Asia and involved genetic 7 contribution from the Yunnan Cangyuan (CC) population (Fig. 7), as well as gene flow from Malawi (Fig. 5). While 8 the Malawi population overall showed admixture patterns similar to Peru (Fig. 5)  The genomic analysis of FAW from native and invasive ranges in this work contradicts recent published 2 theories on the pathway, origin, and direction of spread of this pest across the Old World. Using neutral and 3 unlinked genome-wide SNPs obtained from material available at early stages of the invasion, we showed, 4 through population admixture analysis, ML distance network, and gene flow directionality analyses, that there 5 were likely multiple introductions to both Africa and Asia. Studies to date have relied on analyses of limited 6 partial mitochondrial DNA (e.g., partial COI and CYTB; 21,60 ) and the nuclear Tpi partial gene (e.g., Nagoshi  Myanmar, Thailand, Vietnam, Malaysia, Indonesia, etc.) meant that other candidate FAW invasive populations 5 could have been the 'source invasive populations' for the Asian/Old World invasions. One example is the 6 modelling of FAW spread via monsoon wind patterns from Myanmar into southern China 119 , a hypothesis that 7 could be tested using genomic evidence. International trade pathways are increasingly being identified as 8 responsible for accidental introductions of invasive plant pests and pathogens (e.g., Lopes-da-Silva 120 ).

9
Evaluation of introduction pathways will therefore need to also include trade data, as has been undertaken for 10 the invasion by H. armigera from the Old World into the New World 3,15 (reviewed also in Jones et al. 9

55
. CC-BY-NC-ND 4.0 International license (which was not certified by peer review) is the author/funder. It is made available under a The copyright holder for this preprint this version posted August 25, 2020.

51
. CC-BY-NC-ND 4.0 International license (which was not certified by peer review) is the author/funder. It is made available under a The copyright holder for this preprint this version posted August 25, 2020. . https://doi.org/10.1101/2020.06.12.147660 doi: bioRxiv preprint