Balancing selection could be supported by the TSP phenomenon

The sex locus should be under balancing selection which predicts that alleles could persist as trans-species polymorphisms (TSP). At the honeybee sex locus, TSP is also present in some related species (Cho et al., 2006). Cho et al showed that the type 2 csd alleles of A. mellifera are more similar to A. dorsata, suggesting balancing selection works at the honeybee sex determination locus. In addition, the major histocompatibility complex of vertebrates (Takahata and Nei, 1990) and the self-incompatibility locus of plants (Vekemans and Slatkin, 1994) also show the TSP phenomenon under balancing selection.

I tested whether the TFA alleles might exhibit TSP by sequencing three different sub-fragments of the hypervariable region, the gene EGF-like, and neutral regions LG1N, LG3N, LG5N1, LG5N4 and LG10N. Indeed, this analysis revealed that the alleles for hypervariable region and the gene EGF-like for TFA are dispersed in the phylogenetic tree with RIFA except the sub-fragment, UFO. Yet, the homologous sequence of the out-group could not be found, and consequently the trees of sub-fragments were unroot trees.

In contrast, none of the neutral regions exhibited the TSP phenomenon; all were clustered in one group which could be split from RIFA sequences.

Although consistent with balancing selection, TSP may also arise by recent speciation, introgression, and convergence. Recent speciation can be excluded because TFA is a basal lineage in the fire ants and split from the lineage leading to RIFA a long time ago (Gotzek et al., 2010; Pitts and McHugh Ross, 2005). Recent introgression is also unlikely because the two species are not known to hybridize where they overlap.

Convergence means the independent evolutional lineages have with similar features.

An argument against convergence is that evidence from another ant suggests that the sex locus may be ancient. In the ant V. emeryi, which uses 2 locus CSD, the location of one of the sex determination genes is close to the homologous hypervariable region and may suggest SDL is conserved from ancestral species. Thus, balancing selection is the most likely explanation.

The fact that V. emeryi uses 2-loci CSD raises an interesting possibility regarding sex determination evolution in ants. The second QTL locus maps to the homologous tra gene, which is homologous to the honeybee csd gene. One intriguing possibility is that fire ants in the native range actually use 2-locus sex determination, like V. emeryi, but because of the bottlenecks associated with fire ant invasions into the USA and subsequently elsewhere, there was loss of alleles at the tra locus. Consequently, the invasive range now only has single-locus CSD. This scenario, termed sex determination meltdown event, has been demonstrated for a parasitoid wasp, Cotesia rubecula (de Boer et al., 2012). If this is true, then the evolution of 2-locus sex determination followed by meltdown of one allele may suggest a mechanism of how CSD loci appear to "move"

during evolution.

Figure 1. Comparison of the candidate sex locus within the SDL region of TFA and RIFA.

The TFA draft genome contains 3 overlapping fragments that align to the RIFA candidate sex locus. Using microsatellite loci which are linked to the RIFA SDL region as markers labels the homologous TFA SDL region. In addition to the hypervariable region (HVR) and the genes THUMP and EGF-like, the relative positions of all markers are similar between these two species. Only the SDL11 marker could not be confirmed because it is not assembled onto the same TFA contig as the other markers. Units are in kilobasepairs (kb).

A.

B.

Figure 2. Hardy-Weinberg simulation analysis

(A)The simulation examined three parameters: allele number, sample size and the exact p-value. (B) When the type I error is fixed at 0.05, the largest exact p-value is similar to a line described by Y = 1.28 * X - 0.58.

suppose all allele frequencies are one divided by all allele number.

one colony extract one worker(diploid individual) as the sample.

0 20 40 60 80 100

020406080100120

For exact p-value less than 0.05,

how many heterozygosities should be got while no homozygosity

detected allele number

suppose all allele frequencies are one divided by all allele number.

one colony extract one worker(diploid individual) as the sample.

0 20 40 60 80 100

020406080100120

For exact p-value less than 0.05,

how many heterozygosities should be got while no homozygosity

detected allele number

minimum needed sample size(colonies)

y=1.28301830183018*x-0.582424242424304

Figure 3. The hypervariable and surrounding region.

(A) Schematic of genomic region where all TFA females are never homozygous. This region encompasses 5 microsatellite markers, THUMP, SDL6-1, SDL6-7, SDL6-3 and SDL6-4. (B) Zoom of the region between the two genes, THUMP and EGF-like. (C) Detailed view of the region including CoP, HVR and UFO. Red, orange, and yellow are a rough indication of the diversity within each sub-region.

Figure 4. The shared “insertion” between TFA and RIFA.

A shared “insertion” between a TFA (W1F3HVR ) and a RIFA (SinvTWH1) sample (red box). No other samples share this insertion. The length of this insertion is 167 bp.

Figure 5. Phylogenic trees of the hypervariable and surrounding regions.

TFA samples, red; RIFA samples, black; out-group, orange. Purple dots indicate node support >70% by bootstrap replication. (A) The gene THUMP, (B) The intervening region, (C) CoP1, (D) CoP2, (E) HVR1, (F) HVR2, (G) UFO, (H) The gene EGF-like.

Figure 6. Phylogenic trees of the neutral regions.

TFA samples, red; RIFA samples, black; out-group, orange. Purple dots indicate node support >70% by bootstrap replication. (A) LG1N, (B) LG3N, (C) LG5N1, (D) LG5N4, (E) LG10N.

A.

B.

C.

D.

E.

Table 1. Descriptive information for the microsatellites used in this study.

Location Locus Allele number Size Core unit

LG3 SDL1 5 183~271 (CT)n

LG3 Soli125 2 196~202 (CGA)n

LG3 SDL5 3 247~253 (AGG)n

LG3 THUMP 16 310~347 (AT)n

LG3 SDL6-1 7 117~145 (GT)n

LG3 SDL6-7 2 359~361 (AT)n

LG3 SDL6-3 3 109~133 (GCG)n

LG3 SDL6-4 4 248~273 (GAG)n

LG3 SDL7 3 232~236 (CA)n

LG3 SDL11 10 376~402 (AG)n

LG10 Sol20 6 117~135 (TC)n

LG10 SdagC487 3 304~315 (CT)n

LG5 SiMS2A-65 5 137~153 (AG)n

LG1 Tramsa2 3 201~211 (AG)n

This study examined 256 diploid females and 60 haploid males (total 316 samples) from 39 colonies. The linkage group (LG) refers to the RIFA linkage map (Huang et al., unpublished).

Table 2. Summary of the Hardy-Weinberg analysis of diploid female microsatellite data.

Allele number Replicates Proportion reject H₀, α = 0.05

P-value

SDL1 5 1000 0 1

Soli125 2 1000 0 1

SDL5 3 1000 0 1

THUMP 16 1000 0.326 0.674

SDL6-1 7 1000 0.011 0.989

SDL6-7 2 1000 0.201 0.799

SDL6-3 3 1000 0.478 0.522

SDL6-4 4 1000 0 1

SDL7 3 1000 0 1

SDL11 10 1000 0.646 0.354

Sol20 6 1000 0.002 0.998

SdagC487 3 1000 0 1

SiMS2A-65 5 1000 0.243 0.757

Tramsa2 3 1000 0.427 0.573

256 diploid female data from 35 families were used. The analysis was tested with 1000 adjusted-bootstrap replicates, with the final P-value defined as 1 – [ (number of tests rejecting H0)/replicate number]. SDL1 to SDL11 are loci near the hypervariable region, and the four others are independent loci on different linkage groups.

Table 3. Haplotype patterns for all 3, 4, or 5 adjacent loci combinations in the focal sex locus region

Haplotypes consisting of 3, 4, or 5 adjacent loci were constructed and revealed only one region, THUMP to SDL6-4, with no homozygous female individuals. NA, excluded data (due to the unphased data); allele number means total haplotype number; homo. number means total homozygote count; hetero. number means total heterozygote count. Bold boxed regions include the hypervariable region. (A) 3-loci haplotypes. (B) 4-loci haplotypes. (C) 5-loci haplotypes. The microsatellite marker order is: SDL1, Soli125, SDL5, THUMP, SDL6-1, SDL6-7, SDL6-4, SDL6-3, SDL7, and SDL11. The yellow block is the only no homozygote case (p-value ≅ 0.12).

Table 4. The average nucleotide divergence (per site) within and between RIFA and TFA

The within/between-species nucleotide divergence (π and Dxy) for each locus examined are indicated. Within species divergence (π), is the average divergence of two randomly picked sequences from one species. Between species nucleotide diversity (Dxy) is the average divergence of two randomly picked sequences selected respectively from these two species. Tajima’s D is a summary statistic for the state of the evolution force. For example, positive value could indicate balancing selection or population contraction and negative value could indicate purify selection, balancing selection, bottleneck or others.

Values in red are significant (p < 0.05). Fragment, the locus name; N, the number of sequences; site (bp), the length of locus after alignment.