Genomic signatures of native (Japanese) and introduced (Oceanian) populations of the Asian longhorned tick Haemaphysalis longicornis

Abstract

The Asian longhorned tick (Haemaphysalis longicornis) impacts the health of humans, livestock, and wildlife, both as a parasite and a vector. Over more than a century, H. longicornis has become a widespread invasive species across much of the Asia-Pacific region, with a recent emergence in North America. This dispersal success could be associated with the ability of certain lineages of H. longicornis to reproduce via parthenogenesis. We analyzed complete mitochondrial genomes and genome-wide single-nucleotide polymorphisms (SNPs) from ticks collected within the native range of H. longicornis (Japan), invasive populations in the Asia-Pacific (Australia, Fiji, and New Caledonia), and laboratory-maintained parthenogenetic and bisexual strains. We also compared our data with DNA sequences in GenBank from invasive populations in other countries (e.g., United States). Two distinct clades (Mt-Clade 1 and Mt-Clade 2) were observed in the phylogenetic tree based on mitochondrial genomes. All specimens from the bisexual laboratory strains were assigned to Mt-Clade 1 and those from the parthenogenetic laboratory strains were assigned to Mt-Clade 2, whereas the field-collected male specimens were included in both clades. No geographical cline associated with the phylogenetic characteristics was observed. In contrast, analyses based on genome-wide SNPs demonstrated a cluster composed of Oceanian samples apart from Japanese samples and detected gene flow between the two mitogenomic clades. Our results suggest that parthenogenesis may have evolved multiple times within H. longicornis rather than originating from a single evolutionary event.

Significance statement.

The global and rapid expansion of Haemaphysalis longicornis is not only an issue of invasive species dispersal but a critical One Health concern. Our study demonstrates that parthenogenetic strains have likely emerged multiple times in H. longicornis, contradicting the assumption of a single origin. This fact may represent an evolutionary mechanism that promotes invasion success and increases public health risks. Furthermore, a genomic difference between tick populations in native and introduced regions was revealed. By revealing the evolutionary dynamics of a globally important vector, our work advances understanding of invasion biology and provides critical insights for predicting and managing the spread of H. longicornis and other invasive species of public health significance.

Introduction

The Asian longhorned tick (Haemaphysalis longicornis) is recognized as one of the most significant vectors of tick-borne diseases in Asia. It transmits a diverse range of pathogens, including severe fever with thrombocytopenia syndrome virus, Rickettsia, and Theileria (1–3). Furthermore, it has been implicated in the emergence of tick-bite-induced mammalian meat allergy (4).

The natural range of H. longicornis is within temperate East Asian (Japan, South Korea, North Korea, China, and Russia). However, it has become established in Oceania (Australia, New Zealand, New Caledonia, and Fiji) due to accidental human introductions, and in 2017, it was detected in the United States of America (5). Haemaphysalis longicornis has since spread to 18 of the 50 states in the United States (6).

Two distinct reproductive phenotypes are known from H. longicornis, namely bisexual and parthenogenetic (7). Outside of its native range, only the parthenogenetic phenotype has been detected (8–10). This reproductive phenotype may have facilitated the introduction and establishment of H. longicornis in Oceania and North America, as a single female can rapidly generate a large asexual population.

Typical bisexual populations of H. longicornis are diploid with 20 autosomes along with two sex chromosomes in females and one in males (7). While the majority of parthenogenetic H. longicornis populations are partially or wholly triploid with 30–35 chromosomes, though typically 32–33 in both sexes (7). Finally, some H. longicornis female populations that can reproduce both sexually and asexually exhibit aneuploidy, in which some of the chromosomes are triploid while others are diploid with the karyotype ranging from 23 to 28 chromosomes (7). Breeding trials coupled with karyotyping in H. longicornis by Oliver et al. (7) found that triploid females could not produce viable offspring when they mate with diploid males. In addition to both the ploidies, aneuploid females mated with diploid males could successfully produce viable offspring.

Population genetics is a useful tool for investigating the geographical origins of species by comparing the genetic features of the populations from both native and non-native localities (11). In previous studies, phylogenetic analyses of H. longicornis based on a partial mitochondrial cytochrome c oxidase subunit 1 (cox1) locus distinguished bisexual and parthenogenetic clades into distinct haplogroups (12, 13). Moreover, the invasive populations in the United States were inferred to be parthenogenetic based on their phylogenetic position and the apparent absence of male ticks (10). However, these conclusions were made solely on a maternally inherited mitochondrial gene, warranting further verification of the parthenogenetic population hypothesis using more robust genetic markers, such as those encoded in the nuclear genome.

High-throughput DNA-sequencing technologies have enabled the effective discovery and mapping of large sets of genome-wide genetic markers (14). Several methods have been employed to sample single-nucleotide polymorphisms (SNPs) from genome-wide regions, which are inherited from both parents. In particular, multiplexed inter-simple sequence repeats genotyping by sequencing (MIG-seq) has been developed as a cost-effective method to extract genome-wide SNPs (15, 16). The genome-wide SNPs obtained by MIG-seq have been successfully applied to population genetics and phylogeographic analyses in nonmodel species of various taxonomic groups (17, 18). Moreover, its high discriminatory power has proven to be useful in understanding the population genetics of highly asexual populations (15, 19).

This study was designed to understand the genetic structure of native and introduced H. longicornis populations based on genome-wide SNPs and complete mitochondrial genomes (mitogenomes), as well as their relationships with reproductive phenotypes. Through this, we sought to understand the origin of parthenogenesis in H. longicornis and elucidate the likely origin of the exotic populations of this tick.

Results

Tick sampling

Haemaphysalis longicornis were collected at 24 sampling localities in 13 different prefectures in Japan, and six sites in the Oceanian region (Fiji, New Caledonia, and Australia). From one and four individuals were sampled for each locality, with 38 individuals from Japan and 7 from Oceania (n = 45; Fig. 1, Table 1). In addition to the field-collected specimens, we included 11 samples from five laboratory strains (two bisexual and three parthenogenetic). These strains were established and maintained at Hokkaido University, the National Research Center for Protozoan Diseases (NRCPD) at the Obihiro University of Agriculture and Veterinary Medicine, Japan, and the USA Centers for Disease Control and Prevention (CDC), and were used as reference specimens for reproductive phenotypes (Table 1).

Figure 1.

Maps showing the sampling localities of H. longicornis in this study. The left map shows sampling countries and the number of samples in parentheses. The right map shows the sampling localities in Japan and the color dots indicate the phylogenetic clades based on the mitogenome to which the specimens were assigned: the green is Mt-Clade 1, the purple is Mt-Clade 2, and the black is both the clades (Fig. 3).

Table 1.

Sampling localities and genetic information of H. longicornis.

No.	Sample ID	Sex	Origin	Locality	Latitude	Longitude	CDS haplotype	cox1 haplotype
01	Hl511	F	Vegetation	Japan: Hokkaido	42.8719	142.1231	MitC2_1	H3
02	Hl681	F	Vegetation	Japan: Shizuoka	34.8926	138.2698	MitC1_1	H43
03	Hl683	F	Vegetation	Japan: Shizuoka	34.8926	138.2698	MitC1_2	H44
04	Hl903	N	Vegetation	Japan: Fukushima	37.7999	140.4025	MitC2_2	H3
05	Hl1693	M	Vegetation	Japan: Kagoshima	31.9355	130.7960	MitC2_3	H45
06	Hl1729	F	Vegetation	Japan: Kagoshima	32.1209	130.5235	MitC2_4	H2
07	Hl1974	F	Vegetation	Japan: Wakayama	33.6598	135.6638	MitC2_5	H46
08	Hl1976	F	Vegetation	Japan: Wakayama	33.6598	135.6638	MitC1_3	H18
09	Hl2011	F	Vegetation	Japan: Wakayama	33.8488	135.4862	MitC1_4	H18
10	Hl2014	M	Vegetation	Japan: Wakayama	33.8488	135.4862	MitC1_5	H47
11	Hl2242	M	Vegetation	Japan: Mie	34.8485	136.2947	MitC1_6	H23
12	Hl2269	M	Vegetation	Japan: Mie	34.3555	136.7038	MitC1_7	H18
13	Hl2271	F	Vegetation	Japan: Mie	34.3555	136.7038	MitC1_8	H18
14	Hl2272	F	Vegetation	Japan: Mie	34.3555	136.7038	MitC2_6	H48
15	Hl2273	F	Vegetation	Japan: Mie	34.3555	136.7038	MitC1_9	H49
16	Hl2311	M	Vegetation	Japan: Mie	34.4558	136.7880	MitC1_10	H18
17	Hl2589	M	Vegetation	Japan: Wakayama	34.3719	135.6438	MitC1_11	H50
18	Hl2618	F	Vegetation	Japan: Shimane	35.2235	132.5823	MitC1_12	H18
19	Hl2619	F	Vegetation	Japan: Shimane	35.2235	132.5823	MitC2_7	H51
20	Hl2782	F	Vegetation	Japan: Kochi	33.2658	132.9895	MitC1_13	H23
21	Hl2994	M	Vegetation	Japan: Ehime	33.3670	132.4143	MitC1_14	H52
22	Hl3018	F	Vegetation	Japan: Ehime	33.3670	132.4143	MitC1_15	H18
23	Hl3020	F	Vegetation	Japan: Ehime	33.3670	132.4143	MitC1_16	H18
24	Hl3329	F	Vegetation	Japan: Hiroshima	34.4886	132.4531	MitC2_8	H2
25	Hl3339	M	Vegetation	Japan: Hiroshima	34.4886	132.4531	MitC2_9	H45
26	Hl3512	M	Vegetation	Japan: Nagasaki	32.9854	129.1659	MitC1_17	H18
27	Hl3525	F	Vegetation	Japan: Kochi	33.6542	134.0828	MitC2_10	H53
28	Hl3600	M	Vegetation	Japan: Nagasaki	32.9405	129.1395	MitC1_18	H18
29	Hl3601	M	Vegetation	Japan: Nagasaki	32.9405	129.1395	MitC1_1	H43
30	Hl3616	F	Vegetation	Japan: Chiba	35.1894	140.2534	MitC1_19	H18
31	Hl3619	M	Vegetation	Japan: Chiba	35.1894	140.2534	MitC1_20	H54
32	Hl3626	F	Vegetation	Japan: Wakayama	33.8324	135.3036	MitC1_21	H54
33	Hl3627	F	Vegetation	Japan: Wakayama	33.8324	135.3036	MitC1_22	H54
34	Hl3628	M	Vegetation	Japan: Wakayama	33.8324	135.3036	MitC1_22	H54
35	Hl3640	M	Vegetation	Japan: Wakayama	34.3141	135.3980	MitC1_23	H54
36	Hl3646	F	Vegetation	Japan: Kagawa	34.1534	133.6500	MitC1_24	H55
37	Hl3647	M	Vegetation	Japan: Goto Islands	32.8677	129.0808	MitC1_25	H18
38	Hl3648	M	Vegetation	Japan: Kochi	33.2875	134.1725	MitC2_11	H4
39	Hl_FJ1	F	Cattle	Fiji: Farm A	−17.7900	178.4700	MitC2_12	H4
40	Hl_FJ14	F	Cattle	Fiji: Farm B	−17.9800	178.4500	MitC2_12	H4
41	Hl_FJ28	F	Cattle	Fiji: Farm C	−17.8000	178.5600	MitC2_11	H4
42	Hl_FJ7	F	Cattle	Fiji: Farm D	−17.8000	178.4900	MitC2_13	H4
43	Hl_NC01	F	Sheep	New Caledonia	−22.1100	166.3100	MitC2_14	H4
44	Hl_NC02	F	Sheep	New Caledonia	−22.1100	166.3100	MitC2_11	H4
45	Hl_AU_B4503	F	Dog	Australia: New South Wales	—	—	MitC2_15	H4
46	Hl_HU_LabP01	PF	Kochi parthenogenetic strain	Hokkaido University	—	—	MitC2_16	H45
47	Hl_HU_LabB01	BM	Kochi bisexual strain	Hokkaido University	—	—	MitC1_22	H54
48	Hl_HU_LabB02	BF	Kochi bisexual strain	Hokkaido University	—	—	MitC1_22	H54
49	Hl_CDC_LabP01	PF	CDC parthenogenetic strain	Centers for Disease Control and Prevention	—	—	MitC2_17	H3
50	Hl_CDC_LabP02	PF	CDC parthenogenetic strain	Centers for Disease Control and Prevention	—	—	MitC2_17	H3
51	Hl_OU_LabB01	BM	Oita bisexual strain	National Research Center for Protozoan Diseases, Obihiro University of Agriculture and Veterinary Medicine	—	—	MitC1_26	H50
52	Hl_OU_LabB02	BM	Oita bisexual strain	National Research Center for Protozoan Diseases, Obihiro University of Agriculture and Veterinary Medicine	—	—	MitC1_26	H50
53	Hl_OU_LabB03	BF	Oita bisexual strain	National Research Center for Protozoan Diseases, Obihiro University of Agriculture and Veterinary Medicine	—	—	MitC1_26	H50
54	Hl_OU_LabB04	BF	Oita bisexual strain	National Research Center for Protozoan Diseases, Obihiro University of Agriculture and Veterinary Medicine	—	—	MitC1_26	H50
55	Hl_OU_LabP01	PF	Okayama parthenogenetic strain	National Research Center for Protozoan Diseases, Obihiro University of Agriculture and Veterinary Medicine	—	—	MitC2_18	H4
56	Hl_OU_LabP02	PF	Okayama parthenogenetic strain	National Research Center for Protozoan Diseases, Obihiro University of Agriculture and Veterinary Medicine	—	—	MitC2_18	H4

Superscripts indicate sex, developmental stage, or reproductive phenotype: B and P denote specimens from bisexual parthenogenetic laboratory strains, respectively, while M, F, and N represent male, female, and nymph, respectively.

Mitogenome characteristics

The complete mitogenome sequences of 56 H. longicornis specimens were amplified, yielding lengths that varied between 14,692 and 14,699 bp. The average nucleotide composition was 38.2% (A), 39.0% (T), 13.0% (C), and 9.7% (G). Each mitogenome encoded 13 protein-coding sequences (CDSs), two ribosomal RNAs (12S and 16S), and 22 transfer RNA genes, as well as one noncoding control region, arranged in the same order and strand orientation as the H. longicornis reference mitogenome (NC_037493). The concatenated sequences of the 13 CDSs, excluding start and stop codons (10,740 bp), were collapsed into 45 haplotypes. The Oceanian samples from Fiji and New Caledonia were resolved into four haplotypes, one of which was identical to that of a single male specimen from Kochi (no. 38 in Table 1), located in Shikoku Island, southwestern Japan. The haplotype reported in the United States (MW602986) was identical to that of a single nymph collected in Fukushima (no. 04 in Table 1), northeastern Japan. The haplotypes generated from laboratory strains were identical within each strain and distinct from one another. The genetic variation based on p-distances varied between 0.0006 and 0.02553 in all CDSs and the two ribosomal regions except for nad3 (Fig. 2). The average p-distance in the cox1 region (as mentioned below) was 0.009, similar to that of the total CDS regions (0.0098).

Figure 2.

Genetic variation based on p-distances across all CDSs and two ribosomal regions. The black lines and red points represent the median and mean values, respectively.

The Bayesian phylogenetic tree based on the CDSs was reconstructed with three H. longicornis mitogenome sequences from China (MK450606 and MK439888 (20)), the United States (MW602986 (21)), and one Haemaphysalis hystricis mitogenome sequences (NC_039765) and showed two distinct clades (Mt-Clades 1 and 2) with a high posterior probability of 1.00 (Fig. 3A). All specimens from the bisexual laboratory strains were assigned to Mt-Clade 1, and those from the parthenogenetic laboratory strains and Oceania were assigned to Mt-Clade 2, whereas the field-collected male specimens were assorted in both the clades (for instance, nos. 12, 17, and 28 in Mt-Clade 1 and nos. 05, 25, and 38 in Mt-Clade 2 in Fig. 3A). The average p-distance between the clades was 0.0154, whereas the average p-distances within Mt-Clade 1 and Mt-Clade 2 were 0.0042 and 0.0015, respectively. The p-distances among all haplotypes are summarized in Table S1. The nucleotide diversity of the mitogenome sequences was higher in Mt-Clade 2 than in Mt-Clade 1 (Table 2). Specimens included in Mt-Clade 1 were obtained from the central to southwestern regions of Japan, while those in Mt-Clade 2 were distributed throughout Japan. The sympatric presence of ticks from both clades was observed at three localities (Fig. 1).

Figure 3.

Phylogenetic trees by Bayesian analysis based on (A) the concatenated sequences of 13 mitochondrial CDSs (10,743 bp) and (B) partial cox1 sequences (621 bp). The numbers shown in tree (A) correspond to the tick specimens listed in Table 1, while the haplotypes shown in tree (B) were designated according to the previous description by Egizi et al. (10), with those marked by an asterisk newly identified in the present study. Superscripts indicate sex, developmental stage, or reproductive phenotype: B and P denote specimens from bisexual and parthenogenetic laboratory strains, respectively, while M, F, N, and L indicate male, female, nymph, and larva, respectively. Oceanian samples are highlighted in orange.

Table 2.

Genetic diversity indices of H. longicornis mitogenome CDSs.

	Clade	Number of ticks	Number of haplotypes	Number of polymorphic sites	Haplotype diversity (SD)	Average number of nucleotide differences	Nucleotide diversity (SD)
All specimens	Mt-Clade 1	33	26	122	0.975 (0.016)	16.08	0.0015 (0.0001)
	Mt-Clade 2	23	18	206	0.976 (0.020)	30.72	0.0028 (0.0010)
Japanese specimens only	Mt-Clade 1	27	25	120	0.994 (0.012)	15.89	0.0015 (0.0001)
	Mt-Clade 2	11	11	200	1.000 (0.039)	49.69	0.0046 (0.0016)

The phylogenetic tree based on the partial cox1 sequences (621 bp) was constructed using sequences generated in this study along with those available in GenBank (n = 562). The partial sequences of cox1 obtained in this study were collapsed into 18 haplotypes. Five haplotypes (H2, H3, H4, H18, and H23) were identical to those reported in a previous study (10), while the remaining 13 newly identified in the current study were assigned as H43–H55. The cox1 sequences of all Oceanian specimens, the parthenogenetic strain in the NRCPD and the field-collected male specimen from Kochi prefecture were identical to haplotype H4. This haplotype was previously reported in South Korea, Australia, New Zealand, and the Kingdom of Tonga (10). Moreover, the specimen from the parthenogenetic laboratory strain (Kochi strain), and field-collected males from Kagoshima and Hiroshima prefectures grouped into a single haplotype H45. The phylogenetic tree based on these cox1 haplotypes showed two distinct clades (designated as Mt-Clade 1′ and Mt-Clade 2′), consistent with the mitogenome-based tree, supported by a high posterior probability of 1.00 (Fig. 3B). The average p-distance between these two clades was 0.0246, while the within-clade distances were 0.0059 within Mt-Clade 1′ and 0.0062 within Mt-Clade 2′. All haplotypes of the bisexual laboratory strains were assigned to Mt-Clade 1′ and those from parthenogenetic laboratory strains belonged to Mt-Clade 2′ (Fig. 3B). However, both clades contained both female and male specimens collected from the field. The cox1 haplotypes obtained from the specimens in this study are summarized in Table 1.

Genome-wide SNP variations

The genome-wide SNPs were also identified using MIG-seq for 56 tick samples (Table 1). Out of the 19,898,972 raw Illumina reads obtained, a total of 149,199–344,923 reads/specimen (with an average of 280,998) were obtained after quality and length filtering. After SNP calling, 1,146 SNP loci which were shared among >80% of the specimens, were identified. Notably, each specimen exhibited a unique genotype. The mean heterozygosity based on these SNPs was 0.233 in Mt-Clade 1 and 0.241 in Mt-Clade 2. Maximum likelihood (ML) phylogenetic analysis based on these genome-wide SNPs was performed under the TIM3e + ASC + R2 substitution model, with 10,000 generations for ultrafast bootstrap replications (UFBoot) and SH-like approximate likelihood ratio test (SH-aLRT). The specimens assigned to Mt-Clades 1 and 2 in the mitogenomic tree were not completely resolved in the SNP-based tree (Fig. 4A). Most specimens from Mt-Clade 2 and two specimens from Mt-Clade 1 (nos. 18 and 36) clustered together without strong support (UFBoot <95% and SH-aLRT <80%). In contrast, the Oceanian specimens except the one from Australia constituted a clade strongly supported by both criteria. The specimens from each laboratory strain also assembled into highly supported independent clades, despite harboring different genotypes within each strain (Fig. 4A). Interestingly, the Australian specimen clustered with the parthenogenetic laboratory strains from the NRCPD (nos. 55 and 56). No clear separation between the mitogenome clades was resolved in either the ML phylogenetic tree or the network (Fig. 4B).

Figure 4.

A) Maximum likelihood (ML) phylogenetic tree and B) phylogenetic network constructed using the NeighborNet algorithm. The numbers in OTUs and superscripts correspond to those in Fig. 3. Oceanian samples are highlighted in orange. The color of the boxes indicates the phylogenetic clades based on the mitogenome to which the specimens were assigned: the green and purple boxes represent Mt-Clade 1 and Mt-Clade 2, respectively.

For the genetic clustering, the principal component analysis (PCA) showed clear clusters corresponding to the localities where the specimens were collected (Fig. 5). Some field-collected specimens from Japan and Australia were roughly aggregated in the same cluster regardless of their mitogenome-based clade assignments. In contrast, specimens from the Southwest Pacific islands and the CDC laboratory strains were independently clustered. Discriminant analysis of principal components (DAPC) yielded results consistent with the PCA, dividing all specimens into three posterior groups (PG1, PG2, and PG3) based on two discriminant functions. For this analysis, the first nine PCs were retained with 38.9% of the total cumulative genetic variance (Figs. 5 and S1). Japanese specimens were separated into two posterior groups (PG1 and PG2); however, both groups included admixed specimens belonging to the two clades identified in the mitogenome-based tree. PG1 consisted of some Japanese specimens and a bisexual laboratory strain from the NRCPD, while PG2 consisted of some Japanese specimens, the Australian and parthenogenetic laboratory strains from the NRCPD and CDC. Specimens from Fiji and New Caledonia comprised a single posterior group (PG3). Individual-based genetic structure analysis by ADMIXTURE identified K = 3 as the lowest mean cross-validation (CV) error value aside from K = 1 (Fig. 6A). Japanese field specimens and the two bisexual laboratory strains clustered into a single group, whereas the specimens from Fiji and New Caledonia belonged to the other group (Fig. 6B). The Australian specimen and the parthenogenetic laboratory strains indicated a substantial proportion of admixture (Fig. 6B). However, the Australian individual and one specific specimen from the parthenogenetic laboratory strain exhibited only minor variability across replicates (Fig. S2). Gene flow analysis within and between the mitogenome clades indicated higher migration rates from Mt-Clade 1 to Mt-Clade 2 (0.0534) than in the opposite direction (0.0286; Table 3), indicating that the gene flow from Mt-Clade 1 to Mt-Clade 2 occurred approximately twice as frequently as the reverse.

Figure 5.

The results of PCA of individual genotypes. Plot colors correspond to those in Fig. 4. Symbols show the sampling localities: circle from Japan, square from Oceanian countries, and triangle from laboratory strain. The proportion of variance explained by PC1 and PC2 is shown in parentheses on the respective axes and those of the remaining PCs are shown in the barplot. The colors of the numbers demonstrate the posterior groups defined by DAPC.

Figure 6.

A) Boxplots of CV errors with SD. The red line and points show the mean CV error values across 10 replicates. B) Population genetic structure of the 56 specimens using ADMIXTURE based on genome-wide SNPs. The numbers on the x-axis correspond to the tick specimen numbers provided in Table 1. Mt-Clade 1 and Mt-Clade 2 represent the phylogenetic clades based on mitogenome sequences (Fig. 2).

Table 3.

Estimates of migration rates between mitogenome clades using BA3-SNPs.

		From
		Mt-Clade 1	Mt-Clade 2
To	Mt-Clade 1	0.9714 (0.0155)	0.0286 (0.0155)
	Mt-Clade 2	0.0534 (0.0240)	0.9466 (0.0240)

SDs of the values are given in parentheses.

Discussion

Classification and origin of reproductive phenotypes based on genetic structure

The two distinct clades observed in the mitogenome-based phylogenetic tree were consistent with the analyses relying on partial cox1 sequences in previous studies (12, 13, 20). Superficially, the two separated clades seem to align well with reproductive phenotypes. However, three of the adults included in Mt-Clade 2 were field-collected males in our study (Fig. 3). The occurrence of males in parthenogenetic populations is extremely rare, according to Oliver et al. (7) who reported that the ratio of males to females within parthenogenetic Japanese H. longicornis populations is 1:1,500, while in Australian populations it is 1:400. Under the simplistic assumption that H. longicornis Mt-Clade 1 is bisexual while Mt-Clade 2 is parthenogenetic, the number of males in Mt-Clade 2 (three individuals) was unexpectedly high. This result suggests that the relationship between reproductive phenotypes and phylogenetic characteristics may not be so strict and that a more nuanced explanation is required. The present phylogenetic analysis and PCA using genome-wide SNPs showed no clear clusters corresponding to the mitogenome clades (Figs. 4 and 5), and the admixture and gene flow analyses detected gene flow from Mt-Clade 1 to Mt-Clade 2 (Fig. 6, Table 3). The presence of field-collected males in a supposedly parthenogenetic clade coupled with apparent gene flow between clades which should be reproductively isolated is unusual but may be caused by one or more of the following phenomena: (i) the irregular emergence of males from parthenogenetic population, (ii) parthenogenetic females have irregularly emerged multiple times from bisexual population in Mt-Clade 2, and (iii) supposedly parthenogenetic female in Mt-Clade 2 can mate with male and reproduce both male and female offspring (Fig. 7).

Figure 7.

Three hypotheses regarding the presence of male individuals in Mt-Clade 2. (i) The males can emerge from parthenogenic females irregularly in Mt-Clade 2. (ii) A cryptic bisexual population can reproduce parthenogenetic females multiple times in Mt-Clade 2. (iii) The females in Mt-Clade 2 can mate with bisexual male in Mt-Clade 1, resulting in the male offspring having Mt-Clade 2 mitogenome.

The first hypothesis is the simplest and has already been reported in previous studies, as mentioned above. Under this hypothesis, no mating occurs between the two tick populations with distinct reproductive phenotypes. Consequently, the genetic characteristics of males would be similar to those of females in Mt-Clade 2. However, genome-wide SNP composition of one male specimen (no. 38) was more closely related to those of Mt-Clade 1 individuals (Fig. 6). Although only a limited number of male specimens from Mt-Clade 2 were available in this study, this result suggests that the first hypothesis may not be supported.

The second hypothesis suggests the multiple emergences of triploidy within Mt-Clade 2. Diploidy is considered the ancestral state in H. longicornis (as it is in the entire genus Haemaphysalis), whereas triploidy is likely a derived state (7, 22). Triploidy in H. longicornis may have arisen multiple times due to unreduced (diploid) gametes from one sex combining with haploid gametes from the other. Mason and Pires (23) highlighted that genetic regulation sometimes governs unreduced gamete production, and as such Mt-Clade 2 may have a higher frequency of this phenomenon. Based on mitogenome phylogenetic analysis, the male specimens showed distinct haplotypes that clustered into different subclades. Furthermore, the genome-wide SNP composition of one female specimen (no. 14) was more similar to that of Mt-Clade 1 than Mt-Clade 2 individuals (Fig. 6). These facts may indicate the signals of gene flow between two lineages. Therefore, one explanation for multiple males within Mt-Clade 2 is a small number of ancestral diploid and/or bisexual lineages scattered within multiple derived triploid lineages which now dominate the majority of Mt-Clade 2.

Perhaps the third hypothesis may provide the most plausible explanation that Mt-Clade 2 is not entirely triploid but instead comprises aneuploid populations which engage in facultative parthenogenesis and sexual reproduction. Although Oliver et al. (7) collected 15 separate H. longicornis populations from Japan, Russia, and Korea, only one of these was aneuploid. Additionally, aneuploidy in Japanese H. longicornis has never been documented, and of the nine Japanese populations capable of parthenogenesis which were studied by Oliver et al. (7), all were triploid. Interestingly, Kitaoka (24) reported successful F1 progeny from hybridization between parthenogenetic and bisexual H. longicornis. However, as noted by Oliver et al. (7), Kitaoka (24) did not check the karyotypes of these ticks to assess whether his parthenogenetic population was triploid or aneuploid. Assuming the presence of aneuploidy within Mt-Clade 2, the occurrence of individuals possessing an Mt-Clade 2 mitogenome alongside a nuclear SNP profile more similar to Mt-Clade 1 than to other Mt-Clade 2 individuals (e.g. no. 38) could be explained by gene flow from Mt-Clade 1 to Mt-Clade 2. Therefore, although the only known population of aneuploid H. longicornis occurs in Korea (Jeju Island), it is possible that other aneuploid populations also occur in Japan. Further investigations should be performed to clarify the underlying mechanisms of the emergence of parthenogenesis in H. longicornis by analyzing karyotypes of field-collected ticks and mating experiments using karyotyped laboratory strains of each mitochondrial clade.

Global population structure of H. longicornis

Since the number of specimens from each locality is limited (one to four specimens), analyses based on genetic differentiation among sampling localities (Φ_ST and F_ST) and isolation by distance were not conducted, which represents the primary limitation of this study. Another constraint is that the field samples were collected between 2013 and 2018; this time span could potentially involve genetic shifts or new introductions over the years. Despite these limitations, our phylogenetic analyses based on the mitogenome CDS and cox1 sequences detected no clear geographical cline in genetic characteristics across Japan (Fig. 3). In fact, specimens from Mt-Clades 1 and 2 were collected from the same sites at three localities (e.g. nos. 07 and 08). Even within the clades, the CDS haplotype MitC1_1 was detected in two distinct locations, Shizuoka prefecture, Honshu Island (central Japan), and Nagasaki prefecture, Goto Islands (southwestern Japan). Consistent with this, phylogenetic and network analyses showed no significant geographical clustering in Japan (Fig. 4). These results might suggest that there is no genetic structure within Japanese H. longicornis populations. Such lack of population structure has also been reported in H. longicornis populations in China and Japan based on mitochondrial DNA (13, 25). This lack of population structure might be driven by the broad host range of H. longicornis which has been recorded from a wide range of mammals and birds (26, 27). In particular, the dispersal of immature stages of H. longicornis via migratory birds across Japan may facilitate widespread admixture. Another factor contributing to this genetic homogeneity may be anthropogenic transportation through the livestock industry, as infested livestock have been moved and traded across Japan for centuries.

Based on mitogenome analysis, all Oceanian and the US H. longicornis belonged to a single subclade within Mt-Clade 2 (Fig. 3). The Fijian and New Caledonian specimens possessed haplotypes identical to a single Japanese specimen (no. 38) from Kochi prefecture, Japan. The Oceanian population was previously suggested to have been introduced to Australia from a northern Japanese population in the 19th century (8). However, as mentioned above, the Japanese H. longicornis population is highly admixed, suggesting that the geographical origin of the introduced tick population into Oceanian countries in the 19th century may be untraceable even when using complete mitogenomes and genome-wide SNPs. Oceanian specimens except those from Australia constitute a single cluster that was differentiated from Japanese clusters based on genome-wide SNP analyses. The individual-based genetic structure analysis by admixture also showed the biased ancestry proportion for Oceanian specimens (Fig. S3). These results suggested that Fijian and New Caledonian populations have been transported from native regions or introduced population in other localities, and after colonization there, they have accumulated unique genetic variations, diverging from the source population. This intercontinental-level genetic differentiation among tick populations based on genome-wide SNPs, which could not be resolved using mitogenomes alone, may provide an important reference for future population genetic analyses. For instance, Australian specimen was more closely related to the parthenogenetic laboratory strain in the NRCPD than the other Oceanian samples. Interestingly, the considerable genetic difference observed between the Australian sample and the other Oceanian specimens (Fiji and New Caledonia) suggests that multiple parthenogenetic lineages may be present in Oceania, potentially resulting from multiple introduction events, although the populations in the Southwest Pacific Islands likely share a common parthenogenetic lineage. Therefore, future population genetic studies based on genome-wide SNPs using a large number of tick specimens that represent the genetic diversity within the targeted populations in Australia and the Southwest Pacific Islands will help to evaluate the intercontinental-level origin of newly introduced populations.

The CDS haplotype (MitC2_2) of the tick deposited from the United States (MW602986) was shared with a specimen (no. 04) from Fukushima prefecture, Japan. The cox1 haplotype (H3) of MitC2_2 has also been reported from other native range, including China and South Korea (10) but not from Oceania. Moreover, the analysis based on genome-wide SNPs separated the CDC laboratory strain from the specimens from Oceania (Fig. 5). These facts might indicate that H. longicornis population in the United States is likely to be introduced directly from its native range. Further analyses using specimens collected throughout tick distribution will be needed to investigate the details of intercontinental-level origin of ticks that have invaded the United States.

Conclusion

The reproductive phenotypes of H. longicornis have previously been considered to align strictly with the two phylogenetic cox1 clades (i.e. a bisexual clade and a parthenogenetic clade), which based on a limited analysis of a single genetic locus. However, our analyses of complete mitogenomes and genome-wide SNPs challenge this long-held assumption, suggesting that the relationship between mitochondrial phylogeny and reproductive phenotypes is far more complex than previously recognized. Specifically, Mt-Clade 2 appears to show patterns of widespread bisexual reproduction as well as parthenogenesis as evidenced by the presence of field-collected males. This supports the hypothesis that aneuploidy may be common within this clade. Furthermore, the lack of significant genetic differentiation using genome-wide SNPs among sampling localities in Japan suggests extensive historical admixture within native populations. This therefore implies that the geographical origin of the tick population introduced into Oceania in the 19th century is likely untraceable by genomics. Nevertheless, the spread of ticks outside of Northeast Asia may be informed using genomic analysis as the population in Oceania constituted a single unique cluster separated from those in Japan. The genomic information presented in this study will serve as a critical reference and will assist in mapping the patterns of spread of H. longicornis outside its native range.

Materials and methods

Tick samples and DNA extraction

Haemaphysalis longicornis were collected by flagging a flannel cloth over the vegetation during periods of high tick activity between 2013 and 2018 at 24 sampling localities in 13 different prefectures in Japan (n = 38; Fig. 1, Table 1). We also included seven samples from Oceanian origin; four adult female ticks were collected from cattle on four farms in Fiji in 2016; and two adult females were from a single farm in New Caledonia in 2019. Additionally, one adult female tick was also obtained from a pet dog in New South Wales, Australia, in 2017. Tick specimens were identified morphologically to species and sex under an M205C Leica stereo microscope (Leica Microsystems, Wetzlar, Germany) using keys and descriptions by Yamaguti et al. (26). In addition, to the field-collected specimens, we established two laboratory strains, one bisexual and one parthenogenetic derived from field-collected ticks in Kochi prefecture, Japan. The field-collected ticks were raised to adults by blood feeding on Japanese white rabbits (female, specific pathogen free; Japan SLC, Shizuoka, Japan) at Hokkaido University, and the sex ratio in their progeny was confirmed after they matured into adults using (26). The animal experiments were approved by the Animal Experiment Committee of the Graduate School of Veterinary Medicine, Hokkaido University (approval no. 22-0030, approved on 2022 March 27). The bisexual strain produced both male and female offspring, while the parthenogenetic produced only female offspring. Both laboratory strains have been maintained for several generations without the introduction of any external individuals. From the two established laboratory strains, one male and female each from the bisexual strain and one female from the parthenogenetic strain were utilized for subsequent analyses. Moreover, we also utilized eight samples of one bisexual and two parthenogenetic laboratory-maintained strains as the reference specimens of reproductive phenotypes; two males and two females of the Oita bisexual strain maintained at the NRCPD, Obihiro University of Agriculture and Veterinary Medicine, Japan, two females of the Okayama parthenogenetic strain maintained at the NRCPD, and two females of the CDC parthenogenetic strain (USA).

Ticks were individually washed with 70% ethanol followed by washing with sterile PBS twice, then homogenized in 100 μL of high-glucose Dulbecco's Modified Eagle Medium (Gibco, Life Technologies) by using Micro Smash MS-100R (TOMY, Tokyo, Japan) for 30 s at 2,500 rpm, as described previously (28). Total DNA was extracted from 50 μL of the tick homogenate using a blackPREP Tick DNA/RNA Kit (Analytikjena, Germany), according to the manufacturer's instructions and kept at −30 °C until used.

Mitogenome analyses

The complete mitogenome sequences of H. longicornis were amplified in two overlapping PCRs (long and short-range PCRs). Long-range PCR primers: mtG_K23: 5′-TCCTACATGATCTGAGTTYAGACCG-3′ and mtG_K26: (5′-ACGGGCGATATGTRCATATTTTAGAGC-3′) (29) and short-range PCR primers (H_gap_F1: 5′-YAAYTCCAAAAATTGATGCAAA-3′ and H_gap_R1: 5′-AAGTCAAGRTGCARCAWAAR-3′) designed by aligning complete mitogenomes of genus Haemaphysalis deposited in the database were employed. The long-range PCRs were performed in a 50-μL reaction mixture containing 10 μL of 5× PrimeSTAR GXL Buffer (Mg²⁺ Plus; TaKaRa Bio Inc., Shiga, Japan), 4.0 μL of dNTP Mixture (2.5 mM each), 200 nM of each primer, 1.0 μL of PrimeSTAR GXL DNA Polymerase (TaKaRa Bio Inc.), and 2.0 μL of template DNA. Thermal cycling was performed by an initial denaturation step at 94 °C for 60 s, followed by 45 cycles of 98 °C for 10 s, 55 °C for 15 s, 68 °C for 60 s, and a final extension of 68 °C for 5 min. The short PCRs were performed in a 25-μL reaction mixture containing 12.5 μL of 2× Gflex PCR Buffer (Mg²⁺, dNTP plus; TaKaRa Bio Inc.), 0.5 μL of Tks Gflex DNA Polymerase (1.25 units/μL; TaKaRa Bio Inc.), 200 nM of each primer, and 1.0 μL of template DNA. The reaction conditions were 94 °C for 60 s, 45 cycles of 98 °C for 10 s, 55 °C for 15 s, 68 °C for 60 s, and a final extension of 68 °C for 5 min. PCR amplicons were confirmed by electrophoresis on a 1.5% agarose gel stained with Gel-Red (Biotium, Hayward, CA, United States) and purified with a NucleoSpin Gel and PCR Clean-Up Kit (TaKaRa Bio Inc.).

Illumina sequencing libraries were constructed from the purified amplicons of long and short-range PCRs using the Nextera DNA Library Prep Kit (Illumina, Hayward, CA, United States) and were sequenced with the MiSeq reagent kit v3 for 600 cycles, 2 × 301 bp paired-end, on an Illumina MiSeq platform. The raw reads were mapped against a reference mitogenome (NC_037493) by Geneious v10.2.6 (Biomatters Ltd, Auckland, New Zealand). Consensus sequences were then generated by mitigating the potential sequencing errors derived from PCR amplification. The resulting sequences were annotated using MITOS webserver (30) followed by manual editing. The complete mitogenome sequences obtained were submitted to the DNA Data Bank of Japan (http://www.ddbj.nig.ac.jp) with the accession numbers LC667696–LC667751.

A phylogenetic analysis was performed based on the 13 mitochondrial protein-coding genes (cox1, cox2, cox3, nad1, nad2, nad3, nad4, nad4L, nad5, nad6, cytb, atp6, and atp8) of the mitogenome CDSs, excluding the start and stop codons of each gene. The analysis included three H. longicornis mitogenome sequences from China (MK450606, MK439888 (20)), the United States (MW602986 (21)), and one H. hystricis mitogenome sequence (NC_039765) as an outgroup. The sequence alignment was conducted with the MAFFT (31), and the best nucleotide substitution models were estimated based on the Bayesian information criterion by Kakusan4 (32). Gene- and codon-partitioned models were employed as the optimal nucleotide substitution models for the Bayesian phylogenetic analysis. All selected models for each partition are summarized in Table S2. The Bayesian tree was estimated over 5,000,000 generations via the simultaneous Metropolis-coupled Markov Chain Monte Carlo algorithm (n chains = 4) with a sampling frequency of 100 generations using MrBayes5D (33). Assessments of stationarity and the sample size adequacy for each parameter (effective sample size 200) were performed using Tracer v.1.7 (34). The first 25% of trees were discarded as burn-in. The remaining trees were used to reconstruct the 50% majority-rule consensus tree and to estimate nodal support values as posterior probabilities. The p-distances between haplotypes were calculated using MEGA v. 7 (35) and genetic diversity indices were calculated using DnaSP v. 6.12.03 (36). An additional tree was constructed based on the partial cox1 sequences (621 bp) obtained in this study along with those available in GenBank (n = 562), originating from a total of eight countries (Australia, China [PRC], Japan, the Kingdom of Tonga, New Zealand, South Korea [ROK], and the United States). The selected nucleotide substitution model was a codon-proportioned model; F81, HYK85, and K80 models were for the first, second, and third codon positions, respectively.

Genome-wide SNP analysis

Preparation of the MIG-seq libraries from a total of 56 ticks (Table 1) was performed using the method reported by Suyama et al. (16) with some modifications. Briefly, a 20.0-μL reaction mixture of the first PCR included 1.0 μL of sample DNA, 0.2 μM of each MIG-seq set-1 primer, 0.1 μL of Multiplex PCR Enzyme, and 10.0 μL of 2× Multiplex PCR Buffer from the Multiplex PCR Assay Kit Ver. 2 (TaKaRa Bio Inc.). Thermal cycling conditions of the first PCR were as follows: 94 °C for 60 s followed by 25 cycles at 94 °C for 30 s, 38 °C for 60 s, and 72 °C for 60 s, with a final extension at 72 °C for 10 min. The first PCR products were visualized by electrophoresis in a 2.0% agarose gel stained with Gel-Red and purified using a 1.8-fold volume of AMPure XP (Beckman Coulter, Brea, CA, United States). The second PCR was conducted in a PCR mixture containing 2.5 μL of purified first PCR product, 0.2 μM each of forward and reverse indexing primers, 0.24 μL of PrimeSTAR GXL DNA Polymerase, 0.96 μL of dNTP mixture, and 2.4 μL of 5× PrimeSTAR GXL Buffer. Thermal cycling conditions for the second PCR were as follows: 94 °C for 60 s followed by 12 cycles at 98 °C for 10 s, 54 °C for 15 s, and 68 °C for 60 s. The second PCR products were visualized by electrophoresis on a 2.0% agarose gel stained with Gel-Red and equivalently pooled among samples as a MIG-seq DNA library. The DNA library was size-selected at ∼400–800 bp by NucleoMag NGS Clean-up and Size Select (TaKaRa Bio Inc.). Agilent 2200 TapeStation system with High Sensitivity D5000 ScreenTape (Agilent Technologies, Santa Clara, CA, United States) and Qubit fluorometer (Life Technology, Carlsbad, CA, United States) were used to measure the fragment size and quantity of the libraries, respectively. An Illumina MiSeq platform and a MiSeq Reagent Kit v3 were used to generate 2 × 301-bp paired-end sequences.

The 17-base sequences (comprising SSR and anchor motifs) corresponding to the first PCR primers and low-quality reads were removed using Trimmomatic v. 0.39 (37). Quality-based trimming was conducted with a 4-bp sliding window with a minimum average quality score of 15, and all remaining reads were trimmed to a uniform length of 100 bp from the 5′ end. Subsequently, Stacks v. 2.55 (38, 39) was utilized for de novo SNP calling. Reads from each sample were bundled together using the ustacks program with the minimum depth of coverage required to create a stack (m) set to 3, the maximum distance in nucleotides allowed between stacks (M) set to 2. A consensus loci catalog across individuals was constructed using the cstacks program with the number of mismatches allowed between loci (n) set to 2, and the SNPs from each sample that matched against the catalog were called using the sstacks program. Finally, SNPs were filtered and exported using the populations program with a minimum minor allele frequency (min_maf) threshold of 0.05 and a minimum percentage of individuals required to process a locus (R) set to 0.8. The “write-random-snp” option was enabled to randomly select one SNP per locus to represent the genotype. All other parameters used in the current study were set to default values according to Suyama et al. (16).

ML phylogenetic analysis based on the selected SNPs was carried out using IQ-TREE v. 2.1.3 (40). Automatic identification of the most appropriate substitution models for the SNPs with an ascertainment bias correction model was done by -m MFP + ASC option according to Bayesian Information Criterion model selection (41, 42). Ultrafast bootstrap (UFBoot) percentages (43) and the SH-like approximate likelihood ratio test (SH-aLRT) (44) were performed with 10,000 replicates to calculate node support values. A phylogenetic network based on the genome-wide SNPs was constructed using the NeighborNet algorithm (45) with Kimura's two-parameter model of nucleotide substitution (46) by the SplitsTree v. 5.1.7 (47). Expected heterozygosity, as an index of within-population genetic diversity, was calculated using GenAlEx v. 6.5 (48). PCA was also performed using plink v.1.90 (49) and visualized in R (50). Additionally, DAPC was performed using the xvalDapc function in the R package “adegenet” (51, 52). Prior clusters were defined using find.clusters with the percVar in pca.select and perc.pca = 80 options. To infer individual-based genetic structure, the population structure was analyzed by ADMIXTURE v. 1.3.0 (53). The number of ancestral populations (K) ranged from one to 20. The termination criterion for each run was defined by a log-likelihood increase (C) of less than 0.0001 between iterations. ADMIXTURE runs were repeated 10 times for each K and the optimal K was determined based on the lowest mean CV error value across replicates. The results of the admixture analysis were visualized in R. To infer gene flow patterns, BA3-SNPs (54, 55) was used to estimate recent migration rates between the two mitogenome clades. The analysis was performed with 50,000,000 MCMC iterations, sampling every 1,000 steps after a burn-in period of 5,000,000 iterations. The mixing parameters were adjusted as follows: migration rates between populations (m) to 0.3, allele frequencies (a) to 0.6, and inbreeding coefficients (f) to 0.03. Assessments of stationarity and the sample size adequacy for each parameter (effective sample size 200) were performed using Tracer v.1.7 (34).

Supplementary material

Supplementary material is available at PNAS Nexus online.

Competing interest

The authors declare no competing interests.

Funding

This study was funded by the Japan Society for the Promotion of Science (JSPS) KAKENHI (JP19K23703, JP20KK0151, JP20K15674, JP23K23770, JP23K14090, JP23K20041, JP24KK0133, JP25K02166, JP26K09121, and JP26H02351), the Japan Agency for Medical Research and Development (AMED), Japan (JP20wm0125008, JP23wm0225034, JP23jf0126002, and JP25fk0108717), and AMED Strategic Center of Biomedical Advanced Research and Development for Preparedness and Response (SCARDA), Japan (JP223fa627005). This work was also supported by the World-leading Innovative and Smart Education (WISE) Program (1801) from the Ministry of Education, Culture, Sports, Science, and Technology, Japan.

Author contributions

Yuma Ohari (Conceptualization, Investigation, Methodology, Visualization, Writing—original draft), Kodai Kusakisako (Methodology), Mackenzie L. Kwak (Writing—review & editing), Rika Umemiya-Shirafuji (Resources, Writing—review & editing), Wessam Mohamed Ahmed Mohamed (Methodology, Writing—review & editing), Mohamed Abdallah Mohamed Moustafa (Methodology, Writing—review & editing), Elisha Chatanga (Investigation, Writing—review & editing), Samuel Kelava (Investigation, Writing—review & editing), Ketan Christi (Resources, Writing—review & editing), Dinesh Kumar (Resources, Writing—review & editing), Nariaki Nonaka (Resources, Writing—review & editing), Yasuko Orba (Resources, Writing—review & editing), Keita Matsuno (Funding acquisition, Resources, Writing—review & editing), and Ryo Nakao (Conceptualization, Data curation, Funding acquisition, Supervision, Writing—review & editing)

Data availability

The data that support the findings of this study are openly available in the DNA Data Bank of Japan at http://www.ddbj.nig.ac.jp.