Genomic patterns of strain-specific genetic structure, linkage, and selection across fall armyworm populations

Abstract

Background: Molecular genetic approaches have become vital to understanding the evolutionary processes that act on insect pest populations. From mapping the development of resistance to monitoring and predicting pest movement, genomic tools can inform and enhance pest management programs. Here, we used whole genome sequencing population genomics to unravel novel patterns of population structure, linkage, and selection across the genome of a notorious agricultural pest, the fall armyworm.

Results: Our data strongly support the existence of two genetically distinct strains of fall armyworm in North America, which have previously been referred to as the C-strain and the R-strain. Although these strains have diverged genetically, we find that differentiation is not uniform across the genome. The Z-chromosome appears to drive divergence between strains with high levels of linkage observed across this chromosome. We also show that a region of the Z-chromosome containing a circadian clock gene implicated in allochronic reproductive isolation is under strain-specific selection. Our data indicates that strains differ in their geographic distributions and exhibit distinct patterns of geographic sub-structuring indicative of unique dispersal patterns. We provide the first evidence for nuclear genomic differentiation between the two major overwintering populations of fall armyworm in the US. Finally, our data reveal population-specific patterns of selection on genomic regions containing putative insecticide resistance alleles, which could relate to their biogeography.

Conclusions: Our results support the existence of the fall armyworm as a pest dyad in the US, with genetically-distinct strains differing in their population structure, dispersal patterns, and genomic signatures of selection on regions likely involved reproductive isolation and insecticide resistance. These differences should be considered when devising and implementing management strategies.

Keywords: Spodoptera frugiperda; Population genomics; Strain divergence; Z-chromosome.

Fig. 1

The relative proportion of C-strain, R-strain, and putative hybrid individuals collected at each location. When locations were sampled at multiple timepoints, we did not observe any notable differences in strain composition across collections, so all individuals throughout the year are pooled within each location. Background shading designates western (clear), central (blue), and eastern (red) flyways. Hashed background designated area of intermixing between the central and eastern flyways

Fig. 2

PCA visualizing patterns of autosomal genetic structure within fall armyworm samples collected in the US. Circle markers represent the C-strain, square markers represent the R-strain, and triangle markers indicate putative hybrids. Colors indicate geographic regions where samples were collected. PC1 neatly divides the C-strain from the R-strain, while geographic regions within the C-strain cluster along PC2. No geographic clustering is seen in the R-strain. F_st values differentiating population clusters can be found in Table S1

Fig. 3

Comparison of F_st values between strains. (A) Comparison of F_st values calculated across the Z-chromosome and autosomes. The first comparison is within the C- and R-strains. The second comparison is between C- and R-strain individuals within a region (central or eastern flyway). The final comparison is between C-strain individuals and R-strain individuals collected in different flyways. (B) F_st values between C- and R-strain individuals calculated in 10 kb sliding windows across the genome. (C) Kernel smoothed absolute divergence (Dxy) between strains calculated in 10 kb windows across the Z-chromosome. Values above the dashed line exceed the maximum value of our permutation distribution and thus are significantly elevated

Fig. 4

Population statistics calculated across the autosomes (green) and the Z-chromosome (blue) for all three datasets (All individuals, R-strain, and C-strain). (A) Nucleotide diversity (π) and (B) Tajima’s D were calculated in 5 kb sliding windows. (C) Linkage Disequilibrium (LD) was calculated for all SNP pairs separated by 10,000 bp. This is past the LD decay asymptote, but LD remains high in the combined dataset

Fig. 5

Composite likelihood ratios (CLR) calculated in 1 kb intervals within a Z-chromosome region that showed evidence of a selective sweep for all three C-strain sub-populations; (A) Eastern US, (B) Central US and (C) Puerto Rico. Lines above each graph represent specific coding sequence for genes in the genomic region located on the forward (top) or reverse (bottom) DNA strand. Genes in black are coding regions with unknown function. Genes in red indicate Clk (forward strand) and Mdr49 (reverse strand)

Fig. 6

Genomic regions undergoing selective sweep in the C-strain eastern US sub-population. These regions were found on Chromosome 9 (A) Chromosome 22 (B), and Chromosome 27 (C). Lines above each graph represents annotated genes present within that genomic region that were located on the forward (top) or reverse (bottom) DNA strand. The genes colored in red indicate genes that could play a role in insecticide susceptibility and are indicated above the graph