Complex multiple introductions drive fall armyworm invasions into Asia and Australia

Abstract

The fall armyworm (FAW) Spodoptera frugiperda is thought to have undergone a rapid 'west-to-east' spread since 2016 when it was first identified in western Africa. Between 2018 and 2020, it was recorded from South Asia (SA), Southeast Asia (SEA), East Asia (EA), and Pacific/Australia (PA). Population genomic analyses enabled the understanding of pathways, population sources, and gene flow in this notorious agricultural pest species. Using neutral single nucleotide polymorphic (SNP) DNA markers, we detected genome introgression that suggested most populations in this study were overwhelmingly C- and R-strain hybrids (n = 252/262). SNP and mitochondrial DNA markers identified multiple introductions that were most parsimoniously explained by anthropogenic-assisted spread, i.e., associated with international trade of live/fresh plants and plant products, and involved 'bridgehead populations' in countries to enable successful pest establishment in neighbouring countries. Distinct population genomic signatures between Myanmar and China do not support the 'African origin spread' nor the 'Myanmar source population to China' hypotheses. Significant genetic differentiation between populations from different Australian states supported multiple pathways involving distinct SEA populations. Our study identified Asia as a biosecurity hotspot and a FAW genetic melting pot, and demonstrated the use of genome analysis to disentangle preventable human-assisted pest introductions from unpreventable natural pest spread.

Figure 1

Maximum Likelihood cladograms of unique Spodoptera frugiperda C-strain and R-strain partial mitochondrial genomes based on concatenation of the 13 PCGs (11,393 bp) using IQ-Tree with 1000 UFBoot replications. Individuals in clades I, II, III, and IV (C-strain) and in Clades I, II, V (R-strain) that are in the same colour scheme (i.e., green, orange, blue, or pinks) shared 100% nucleotide identity. Mitogenome haplotypes from native individuals for both C- and R-strains are in khaki green colour. Red and dark grey dots at branch nodes represent bootstrap values of 87–100% and 74–86%, respectively. Bootstrap values < 74% are not shown. Red arrows indicate invasive individuals’ mitogenome haplotypes that are nested within native individuals. Country and sample codes are listed in Tables S1 and S2.

Figure 2

Maximum Likelihood (ML) phylogeny using 879 genome-wide non-coding SNPs with 1000 ultrafast (UF) bootstrap replications. Node support with < 74% are not shown, 74–86% are represented by grey circles, 87–100% are represented by red circles. Country of origin is colour coded.

Figure 3

Principal component analysis (PCA) showing variability between selected FAW populations from their invasive ranges. (a) China and Myanmar; (b) Kedah and Johor/Penang populations from Malaysia, (c) China and east African (Uganda/Malawi) populations, (d) Australia (Strathmore, Qld/Northern Territory + New South Wales), China, and Malaysia (Kedah) populations, (e) Australia (Strathmore, Qld) and PNG (Madang Province) populations, (f) Lao PDR/Vietnam and South Korea populations, (g) China and SE Asian (Lao PDR/Vietnam/Myanmar/Philippines/Malaysia) and Pacific/Australia (PNG) populations, and (h) Australia, China and Malaysia (Kedah) populations. Note the overall population genomic variability between countries (e.g., a, c–g) and within countries (e.g., Malaysia (b), Australia (d)). Populations with similar genomic variability are also evident, e.g., for Strathmore (e) and South Korea (f); and for Madang (e) and Lao PDR/Vietnam (f), further supporting potential different population origins of various FAW populations across the current invasive regions. The Southeast Asian and Chinese populations are overall different (g), Australia’s FAW populations showed similarity with both Southeast Asia and China (g, h).

Figure 4

(a, b) DivMigrate analyses with edge weight setting at 0.453 showing unidirectional (yellow arrow lines) and bidirectional (blue arrow lines) gene flow between countries in Africa and South Asia/East Asia/SE Asia. Significant migration rates (at alpha = 0.5) are in red and as provided in Table 3. Incidences of unidirectional migration were predominantly detected from China (CHN) Yunnan populations (CY, XP) to SE Asian populations (e.g., Myanmar (MMR), Laos PDR (Lao), Philippines (PHL)) and to east African populations (e.g., Uganda (UGA), Malawi (MWI)) (a, b). Source populations are CY (a) and XP (b). (c, d) DivMigrate analyses with edge weight setting at 0.453 showing unidirectional (yellow arrow lines) and bidirectional (blue arrow lines) migration between countries in Africa and South Asia/East Asia/SE Asia. Migration rates between populations are as provided in Table 3. (c) Vietnam (VNM) as the source population identified an incidence of unidirectional migration from Malaysia (MYS) Johor state (JB) to Vietnam, while bidirectional migration events were detected from Vietnam to other SE Asian (e.g., Philippines (PHL), Lao PDR (Lao), Myanmar (MMR)), to Pacific/Australia (i.e., Papua New Guinea (PNG)), as well as to east Africa (Uganda (UGA), Malawi (MWI)). (d) Lao PDR (LAO) as source population identified bidirectional migration events between various SEA populations and east African populations, while unidirectional migration events were identified from India (IND) and China (CHN) Yunnan populations (CY, YJ) to Laos PDR. No migration events were evident from SE Asian populations to China.

Figure 5

Admixture and corresponding CV plots for FAW populations from: (a) Australia, China, South Korea, Lao PDR, Myanmar, Malaysia, Philippines, PNG, and Vietnam, and (b) Benin, China, India, South Korea, Lao PDR, Myanmar, Malaysia, Philippines, PNG, Tanzania, and Vietnam. Optimal ancestral genetic clusters are K = 4 for both admixture plots. Boxed individuals have unique admixture patterns at K = 4 when compared with other populations. China FAW lacked Cluster 2 (navy blue colour; present in almost all SEA and Australian FAW), while in NSW and NT some individuals lacked cluster 3. South Korea ‘MF’ population generally lacked cluster 2, while Kedah (Malaysia) showed distinct (cluster 4) pattern for all individuals. The overall same observations are evident in the admixture plot in (b), with African FAW generally exhibiting admixture patterns similar to SEA populations than to Chinese FAW. With the exception of Kedah (Malaysia) and some Chinese FAW individuals, all FAW in the invasive range showed evidence of genomic admixture (i.e., hybrid signature). The figures were generated using the POPHELPER program < https://pophelpershiny.serve.scilifelab.se/ > and further manipulated in Microsoft PowerPoint for Mac v16.54.

Figure 6

Countries where invasive populations of FAW surveyed for genomic analyses in this study were collected from (see main text, Table S1). Native New World populations (not shown) and invasive populations from Benin, Uganda, Tanzania, Malawi, India, and China have been reported in Tay et al.. Geographic regions of the sampled countries were as defined by World Atlas and are provided in Table S1. The map was created by importing the colour-formatted map region from MapChart < https://www.mapchart.net > into Microsoft PowerPoint for Mac v16.54 where location details were added and further manipulated and formatted.