Secondary contact zone and genetic introgression in closely related haplodiploid social spider mites

Abstract

How frequently hybridisation and gene flow occur in the contact zones of diverging taxa is important for understanding the speciation process. Stigmaeopsis sabelisi and Stigmaeopsis miscanthi high-aggression form (hereafter, S. miscanthi HG) are haplodiploid, social spider mites that infest the Chinese silver grass, Miscanthus sinensis. These two species are closely related and parapatrically distributed in Japan. In mountainous areas, S. sabelisi and S. miscanthi HG are often found in the highlands and lowlands, respectively, suggesting that they are in contact at intermediate altitudes. It is estimated that they diverged from their common ancestors distributed in subtropical regions (south of Japan) during the last glacial period, expanded their distribution into the Japanese Archipelago, and came to have such a parapatric distribution (secondary contact). As their reproductive isolation is strong but incomplete, hybridisation and genetic introgression are expected at their distributional boundaries. In this study, we investigated their spatial distribution patterns along the elevation on Mt. Amagi using male morphological differences, and investigated their hybridisation status using single-nucleotide polymorphisms by MIG-seq. We found their contact zone at altitudes of 150–430 m, suggesting that their contact zone is prevalent in the parapatric area, which is in line with a previous study. Interspecific mating was predicted based on the sex ratio in the contact zone. No obvious hybrids were found, but genetic introgression was detected although it was extremely low.

Introduction

Closely related species often overlap in their geographical distribution (Mayr 1963). Contact frequencies and reproductive, ecological, and genetic relationships at their distribution boundaries are important for their coexistence and evolutionary consequences (Coyne and Orr 2004; Johannesson et al. 2020). In particular, hybridisation is likely to occur if reproductive barriers are not completely established. In this case, they may fuse into a single species (Coyne and Orr 2004) or diversify further by character displacement and/or reinforcement of reproductive barriers (Dobzhansky 1959; Hoskin et al. 2005; Smadja and Butlin 2006; Pfennig and Rice 2014). If the fitness of the hybrids is high, a new species can be generated from hybridisation through developing reproductive isolation from its parental species (Rieseberg 1997; Seehausen 2004; Mallet 2008; Abbott et al. 2013). Even though this does not result in new species, genetic introgression by hybridisation may bring genetic diversity to the parent species and drive their subsequent evolution (Edelman et al. 2019). Therefore, to understand the evolutionary relationships and speciation of closely related species with overlapping distributions, it is important to determine the frequencies of contact, reproductive isolation, and genetic introgression at their distribution boundaries.

Speciation studies, including studies of hybrid zones, have mainly focused on diploid organisms, such as Drosophila flies (Coyne and Orr 2004). However, to confirm the generality of the findings in diploids and the relevance of interpretation to the findings, it would be necessary to target various taxa with different genetic systems and ecology. Haplodiploids, in which females develop from fertilised eggs and males develop from unfertilised eggs, account for about 15% of arthropods, the most species-rich taxon group on Earth (Blackmon et al. 2015). In recent years, haplodiploids have been expected to contribute to the advancement of speciation studies (Lohseand and Ross 2015; Nouhaud et al. 2020). This is because they are numerous and have a different genetic system from diploids and advantages in some analyses (Lohse and Ross 2015; Knegt et al. 2017; Nouhaud et al. 2020).

The Stigmaeopsis miscanthi species group (Acari: Tetranychidae) is a haplodiploid spider mite that infests Chinese silver grass, Miscanthus sinensis, in East Asia (Saito et al. 2018, 2019). The mites construct woven nests on the undersurface of the host plant leaves and live in groups within the nests. They are called social spider mites because there are two to three generations of overlap among nest members, and they show cooperative nest building, nest sanitation, and brood care (Saito 2010; Schausberger et al. 2021). Woven nests are protective against their natural enemies (predatory mites, predatory gall midges, etc.); however, some predators, such as the phytoseiid mite Typhlodromus bambusae, can intrude into the woven nests (Saito 2010). To protect nestmates and their offspring against predatory intruders, adult males and females counterattack the intruders cooperatively and sometimes kill the intruders if they are immature (Saitō 1986a, 1986b; Yano et al. 2011; Saito et al. 2011). Males are aggressive not only against predatory intruders but also against conspecific males. They kill each other to establish their own harem (Saitō 1990). The frequency of male-male killing (lethal male fight) varies among populations (Saito 1995; Saito and Sahara 1999; Sato et al. 2013a) and is associated with differences in male aggression and also with their reproductive, phylogenetic, and geographic relationships (Sato et al. 2000a, 2000b, 2015, 2015, 2018, 2019). Five species and two forms have been described in this species group so far (Saito et al. 2018, 2019).

In Japan, Stigmaeopsis sabelisi with lower male aggression, S. miscanthi high-aggression form (hereafter, S. miscanthi HG) with higher male aggression, and S. miscanthi mild-aggression form (hereafter, S. miscanthi ML) with intermediate male aggression are distributed (Saito 1995; Saito and Sahara 1999; Sato et al. 2013a; Sato et al. 2019). S. miscanthi ML is distributed in subtropical regions and is geographically isolated from the two other species (Sato et al. 2013a; Sato et al. 2019) (Fig. 1). On the other hand, S. sabelisi and S. miscanthi HG show overlap in their geographic distribution: S. sabelisi is distributed in colder regions (from Aomori Prefecture to Kyushu Islands), whereas S. miscanthi HG is distributed in warmer regions (from Shizuoka Prefecture to the main island of Okinawa) (Fig. 1). Japan is mountainous, and in areas where both species are distributed, S. sabelisi and S. miscanthi HG are found in the highlands and lowlands, respectively (parapatric distribution). A previous study inferred the population history of the species group using mtDNA (cytochrome c oxidase subunit I; COI) and estimated that S. sabelisi and S. miscanthi HG were derived from an ancestral group with mild male aggression in the subtropical region during the last glacial period (20,000–40,000 years BP for S. sabelisi and 5494–10,988 years BP for S. miscanthi HG) (Sato et al. 2019). Considering their inferred history together with their ecological and reproductive relationships and the migration history of their host plant (Clark et al. 2014), it is predicted that (1) S. sabelisi was derived south of Japan and migrated into the Japanese archipelago just after the host plant expanded its distribution into the Japanese archipelago; (2) as temperature increased more, the ancestral group expanded its distribution northward and migrated into the Ryukyus Islands; (3) S. miscanthi HG was derived from the ancestral group in and around the Japanese archipelago; and (4) S. miscanthi HG expanded its distribution in the Japanese archipelago and drove S. sabelisi to the colder region through competition and reproductive interference (Saito et al. 2013; Sato et al. 2013b; Sato et al. 2015), resulting in their present geographic distributions (Sato et al. 2019).

Fig. 1. The geographic distribution of the social spider mite Stigmaeopsis miscanthi species group in and around the Japanese Archipelago.

Three species are distributed in the Japanese Archipelago, and the geographic distributions of two species, S. sabelisi and S. miscanthi HG, overlap widely. In the overlapping area (parapatric region), S. sabelisi and S. miscanthi HG are distributed in the highlands and lowlands, respectively (Saito 1995, Saito and Sahara 1999, and Sato et al. 2013,b). The locations of Mt. Unzen and Mt. Amagi, where the distribution of the two species at different altitudes was studied in a previous study (Mt. Unzen; Sato et al. 2008) and in this study (Mt. Amagi), are also shown.

A previous field study at Mt. Unzen, one of the mountains on the Kyushu Islands at the southern end of the parapatric area (Nagasaki Prefecture; Fig. 1), found that the distributions of these two species broadly overlapped at intermediate altitudes (100–400 m), and both species were collected from the same host plant colonies in the contact zone (Sato et al. 2008). It is known that their reproductive isolation is strong but incomplete; there is a strong post-mating and pre-zygotic reproductive barrier, but a few hybrids are produced from interspecific crosses (proportion of hybrids: 0–30%) (Sato et al. 2000a, 2000b, 2015, 2018), and their hybrids are fertile (Sato 2004). Therefore, hybridisation and gene flow are likely to occur in the contact zones formed on each mountain in the parapatric area. In particular, males of S. miscanthi HG actively approach the females of S. sabelisi for mating, as they do for conspecific females (Sato et al. 2015). In addition, interspecific male fights occur readily between the two species, and males in the S. miscanthi HG tend to win interspecific male fights (Sato et al. 2013b). This suggests that S. sabelisi females have a higher chance of interspecific mating than S. miscanthi HG females, indicating that genetic introgression is likely asymmetric. However, it has not been confirmed whether hybridisation occurs in the contact zones. Furthermore, their contact zone has been reported only on Mt. Unzen, and it is unclear whether their contact zones are widespread in their parapatric areas.

In this study, to address whether the contact zone of S. sabelisi and S. miscanthi HG is widespread in their parapatric areas, we investigated their distribution along the elevation on and around Mt. Amagi in Shizuoka Prefecture, Japan. Mt. Amagi was selected as the study site because it is located at the northern end of the parapatric area (Fig. 1). The presence of contact zones at both the southern and northern ends of the parapatric areas (Mt. Unzen and Mt. Amagi) supports the hypothesis that contact zones are prevalent in parapatric areas. In addition, the geographic distribution of S. miscanthi HG has been moving further north due to global warming (Sato et al. unpublished data). Mt. Amagi was selected to see the impact of the northward shift in S. miscanthi HG distribution on S. sabelisi, especially through hybridisation.

To determine whether interspecific mating occurred in the contact zone, we analysed the sex ratio of mite colonies in the contact zone. The mites are haplodiploid, in which females develop from fertilised eggs and males develop from unfertilised eggs. Virgin females lay unfertilised eggs, and the number of eggs is significantly lower than that of mated females (Sato et al. 2000a, 2000b, 2018). Females mated with males of different species lay unfertilised eggs because of the strong post-mating pre-zygotic barrier (reproductive barrier in the egg fertilisation stage); however, the number of eggs was similar to that of females mated with conspecific males, possibly because females control the number of eggs by copulation stimuli (Sato et al. 2000a, 2000b, 2018). As females that mate with males of different species produce an overabundance of sons (Sato et al. 2000a, 2000b, 2018), the sex ratio would become relatively male-biased in mite colonies where interspecific mating occurs (Sato et al. 2008); although, the spider mite species originally showed an extremely female-biased sex ratio (proportion of males in offspring : 0.10–0.20) (Sato and Saito 2007; Saito et al. 2013). The increase in the male ratio may be temporary, as the higher male ratio may be controlled by male-male killing behaviour (lethal male fights). Related to this, the power of sex ratio to detect interspecific mating may depend on which species of female mated with heterospecific males. When females of S. miscanthi HG mated with heterospecific males, they produce more sons (S. miscanthi HG males), but the increased male ratio may decline rapidly due to the male-male killing behaviour. On the other hand, when S. sabelisi females mated with heterospecific males, they produce more sons (S. sabelisi males) and the increased male ratio probably remains high, as S. sabelisi males kill much less than S. miscanthi HG. However, the sex ratio still can be used for detecting the occurrence of interspecific mating, although it cannot be used for rejecting the possibility of interspecific mating. It is also known that male-male killing occurs between different species males, and the frequency of male-male killing in the interspecific male fight is intermediate between those in the intraspecific male fight of S. miscanthi HG (high frequency) and S. sabelisi (low frequency) (Sato et al. 2013b). It is therefore unlikely that interspecific male fight increases the male ratio in the contact zone.

To address whether hybridisation and genetic introgression occurred between the two mite species, we analysed the genetic population structure of the mite colonies in their contact zones. Spider mites are too small to obtain sufficient DNA from a single mite for molecular analysis (body length is less than 0.5 mm). Therefore, in the genetic analyses, we used multiplexed Inter-Simple Sequence Repeat (ISSR) genotyping by sequencing (MIG-seq), which is useful for small amounts of low-quality DNA (Suyama and Matsuki 2015). Single-nucleotide polymorphisms (SNPs) were detected using MIG-seq for genetic analysis. To confirm whether S. sabelisi and S. miscanthi HG collected from Mt. Amagi produced hybrids, we performed cross-experiments. Previous cross-experiments have found hybridisation; however, these studies used different populations, so it is necessary to confirm if the two species from Mt. Amagi produce hybrids or not. Finally, based on these findings, we discuss the prevalence of their contact zones, interspecific mating, gene flow, and the effect of the contact zone on their evolution.

Materials and methods

Study site and mite collection

We collected spider mites of S. miscanthi species group from the eastern area of Mt. Amagi (Shizuoka Prefecture, Japan) at altitudes of 20–680 m in July 2020 (Table 1, Fig. 1). This area is dominated by S. sabelisi; therefore, to ensure the collection of S. miscanthi HG, we collected mites from the area northeast of Mt. Amagi at an altitude of 20 m, additionally (Shiofuki in Table 1). We brought host plant leaves with mite nests to the laboratory and recorded the number of males, females, and immatures in each nest using stereoscopic microscopes. For species identification based on the male morphology, we prepared slide specimens of the collected males. For genetic analysis, we stored the collected females in microtubes containing 99% acetone. For species identification of females based on male morphology, we reared the collected immatures on detached M. sinensis leaves under controlled conditions (25 °C, 75–100% relative humidity, and 15:9 h light: dark). Since spider mites are haplodiploid, male genes are derived only from their mothers. Therefore, we prepared slide specimens of males developed from reared immatures within several days after mite collection and used them for species identification of females based on male morphology. We established laboratory mite cultures for each population using the remaining collected mites and used them for the cross-experiment.

Table 1.

Location where spider mites of Stigmaeopsis miscanthi species group were collected and the numbers of nests, females, males immatures, eggs and the male ratio collected from each location.

Location	Latitude	Longitude	Altitude (m)	No. of					Male ratio
				Nest	Female	Male	Immature	Egg
Shiofuki	34.96863	139.12685	20	8	48	4	23	133	0.08
Amagi 1	34.89088	139.13757	20	19	70	13	143	267	0.16
Amagi 2	34.84611	139.07361	50	18	74	17	157	331	0.19
Amagi 3	34.84833	139.06611	150	14	35	12	59	53	0.26
Amagi 4	34.88358	139.09616	160	10	13	6	26	79	0.32
Amagi 5	34.85333	139.06028	210	22	20	11	39	52	0.35
Amagi 6	34.88972	139.08833	250	7	15	3	25	61	0.17
Amagi 7	34.85750	139.05889	280	10	18	8	31	81	0.31
Amagi 8	34.88667	139.08917	290	4	6	2	22	20	0.25
Amagi 9	34.90222	139.10972	300	8	17	4	28	66	0.19
Amagi 10	34.88806	139.08306	350	6	7	2	10	25	0.22
Amagi 11	34.85582	139.06557	390	17	42	14	122	127	0.25
Amagi 12	34.90583	139.07472	400	28	80	33	139	293	0.29
Amagi 13	34.88917	139.07889	420	11	26	9	85	112	0.26
Amagi 14	34.85806	139.06444	430	21	40	4	50	172	0.09
Amagi 15	34.88583	139.07694	450	20	27	9	69	132	0.25
Amagi 16	34.90639	139.06500	460	35	103	29	164	228	0.22
Amagi 17	34.88361	139.07139	500	10	25	5	41	92	0.17
Amagi 18	34.86194	139.05944	530	11	20	6	47	45	0.23
Amagi 19	34.88139	139.06639	560	4	7	2	7	28	0.22
Amagi 20	34.87944	139.05972	680	10	19	5	30	94	0.21

Species identification by male morphology

As the relative length of leg I to leg III and body width were greater in males of the S. miscanthi HG than in those of S. sabelisi, we used morphological differences to identify the species of the collected mites (Saito 1995; Sato et al. 2000a,2019; Sato et al. 2013b). One to five males were placed on a drop of Hoyer’s solution on a slide glass and covered with cover glass. We placed a 10-g weight on the cover glass to flatten the mite body evenly for accurate measurement and dried it at 45 °C for more than 3 days. We took photos of the slide specimen with a microscope (Axioskop, Zeiss, Germany) and an eyepiece camera (Dino-Eye AM4023X, Anmo Electronics Corporation, Taiwan), after which we measured the lengths of the first and third legs (the four leg segments of the tarsus, tibia, genu, and femur) of the slide specimen using image-processing software (Image J ver. 1.53a; National Institute of Health, Bethesda, MD, USA) (Schneider et al. 2012). We calculated the lengths of the first and third legs (leg I and leg III) from the total of the four leg segment lengths and used the values in the analysis. The specimens used to construct the linear discriminant function were measured three times, and the average values were used. The specimens used for species identification were measured once, and the value was used for species identification.

In a previous study conducted on Mt. Unzen (Nagasaki Prefecture, Japan), a contact zone was found at altitudes of 100–400 m (Sato et al. 2008). Therefore, we assumed that the mite colonies at altitudes below 100 m and above 500 m were S. miscanthi HG and S. sabelisi, respectively. We constructed a linear discriminant function with the method of standard estimators of the mean and variance using lengths of the first and third legs of males from the colonies (S. miscanthi HG : 34 males and 41 males developed from immatures collected from Shiofuki and Amagi 1 and 2; S. sabelisi: 16 males and 41 males developed from immatures collected from Amagi 17–20; Table 1). We used the discriminant function for species identification of males and males developed from immatures collected from colonies at an altitude of 100–500 m (Amagi 3–16; Table 1). For discriminant analysis, we used statistic software R ver. 4.3.3 (R Core Team 2024) and the MASS package (Venables and Ripley 2002).

Interspecific mating inferred from the sex ratio in the field

The sex ratio is expected to be relatively male-biased in mite colonies where interspecific mating occurs (Sato et al. 2008), because females that mate with different species produce excessive numbers of sons (Sato et al. 2000a, 2000b, 2018). We estimated interspecific mating in the field by analysing the calculated sex ratio. Based on the results of species identification using male morphology, we categorised the mite colonies into three types: S. miscanthi HG, S. sabelisi alone, and the mixture. The male ratio (#male / (#male + #female)) was analysed with the colony type (S. miscanthi HG, S. sabelisi, mixture) using a generalised linear model (GLM). In the model, we used a binomial distribution as the error distribution and tested the explanatory variable (colony type) by using the likelihood ratio test (LRT). Multiple comparisons were performed using a post-hoc test (Tukey contrasts: all pairwise contrasts for the grouping variable). For the analysis, we used statistic software R ver. 4.3.3 (R Core Team 2024) and the multcomp package (Hothorn et al. 2008).

Genetic analysis

In the mites, females are much larger than males, and as a result, more DNA is obtained from females than males. In addition, as the mites are haplodiploids, hybrid males appear as the haploid offspring produced from hybrid females and therefore appear one generation later than hybrid females. To increase the detection power of hybrids, we performed genetic analysis using only female individuals. In the genetic analysis, we used a total of 237 females; 164 females from 11 colonies at altitudes of 100–500 m (contact zone), 38 females from two colonies below 100 m altitude (pure S. miscanthi HG), and 35 females from two colonies above 500 m altitude (pure S. sabelisi). DNA was extracted from a single female mite, which was stored in 99% acetone in a microtube, using PrepMan Ultra Sample Preparation Reagent (Applied Biosystems, Foster City, CA, USA).

To construct the MIG-seq library, we followed the protocol described by Suyama et al. (2022), which is modified from Suyama and Matsuki (2015). Briefly, a MIG-seq library was prepared using a two-step PCR method. In the first PCR, ISSR regions were amplified using MIG-seq primer set 1 (Suyama and Matsuki 2015). In the second PCR, indices and Illumina adapter sequences were added to the first PCR products. The Illumina MiSeq platform and MiSeq Reagent Kit v3 (150 cycles; Illumina) were used for sequencing. We skipped the sequencing of the first 17 bases of reads 1 and 2 (SSR primer region and anchors) using “DarkCycle”. We detected genome-wide SNPs from the sequenced raw reads and filtered SNPs and females, as described in Suyama et al. (2022). After the removal of extremely short reads and low-quality reads using trimmomatic 0.39 (Bolger et al. 2014), 42,183,313 reads (177,989 ± 4186 reads per sample) remained from 43,713,012 raw reads (184,443 ± 4,323 reads per sample). De novo SNP genotyping was performed using Stacks 2.55 (Rochette et al. 2019), utilizing the following parameters: minimum depth of coverage required to create a stack (m) of 3, maximum distance allowed between stacks (M) of 2, and number of mismatches allowed between sample loci while building the catalogue (n) of 2. Using the ‘populations’ program in Stacks, we removed SNP sites exhibiting high heterozygosity (Ho ≥ 0.6) and filtered out SNP sites with fewer than three minor alleles. SNPs that were retained by 50% or more samples were included in the SNP dataset. To avoid linked SNPs, only the first SNP from each locus was considered. Observed (H_O) and expected (H_E) heterozygosities as well as nucleotide diversities at SNP sites were calculated using the ‘populations’ program in Stacks. To determine the presence of hybrids and genetic introgression between the species, the selected SNPs were used to analyse the population structure using STRUCTURE Software ver. 2.3.4 (Pritchard et al. 2000) and Structure Harvester (Earl and von Holdt 2012). We conducted 30 independent runs with a burn-in of 100,000 steps, followed by an additional 100,000 steps using an admixture model. We estimated the log-likelihood of each cluster (K = 1–10). A Neighbour-Net network was constructed using SplitsTree4 4.14 (Huson and Bryant 2006), utilising the uncorrelated P-distance matrix ignoring ambiguous sites, which is the default setting and commonly used for MIG-seq data (e.g., Suetsugu et al. 2023).

Interspecific crossing experiments

We used laboratory cultures collected from Amagi 17 and Amagi 1 in the cross-experiment as S. sabelisi and S. miscanthi HG, respectively (Table 1). We placed a detached M. sinensis leaf (1.0 × 3.0 cm) on water-soaked cotton wool in a Petri dish (5.0 cm diameter, 1.5 cm high; SPL Life Sciences, Gyeonggi-do, Korea). We collected a teleiochrysalis female from the mite culture and placed it onto the prepared leaf arena. One day after mite introduction, we checked the emergence and construction of a woven nest and then introduced a male collected from the mite culture onto the leaf arena. We allowed them to mate and oviposit for 10 days, after which we removed the females and males from the leaf arena and recorded the number of eggs. We checked the development of their offspring daily until they reached the adult phase and recorded their survival and sex. We performed crossing experiments between the populations: Amagi 17 (female) × Amagi 1 (male) and Amagi 1 (female) × Amagi 17 (male), and within species as controls: Amagi 17 (female) × Amagi 17 (male) and Amagi 1 (female) × Amagi 1 (male). As another control, we observed virgin oviposition because spider mite females produce sons without mating, and the number of eggs is significantly fewer in the virgin oviposition than in the mated oviposition in the species group (Sato et al. 2000a, 2000b, 2018). We carried out the crossing experiments under controlled conditions (25 °C, 75–100% relative humidity, and 15:9 h light: dark).

To determine the pre-mating barrier, we compared the number of eggs among the cross combinations (interspecific crosses, intrapopulation crosses, and virgin oviposition) of each female species. For comparison, we used a GLM with a negative binomial error distribution because we detected overdispersion in the Poisson GLM. To determine post-mating and pre-zygotic barriers, we compared the offspring sex ratios between interspecific and intraspecific crosses using a GLM with a binomial error distribution. To evaluate whether there is a post-zygotic barrier to reproduction between these species, we compared the offspring survival ratios among the cross combinations (interspecific cross, intraspecific cross, and virgin oviposition) using a GLM with a quasibinomial error distribution, as we detected overdispersion in the binomial GLM. For the analysis, we used statistic software R ver. 4.3.3 (R Core Team 2024) and the package MASS for negative binomial GLM (Venables and Ripley 2002).

Results

Contact zone inferred by species identification by male morphology

The discriminant function was obtained from 33 S. miscanthi HG males and 105 S. sabelisi males as follows:

$$Discriminantscore=-5.730322-0.0916391legI+0.1076930legIII$$

where leg I and leg III are the lengths (μm) of the first and third legs, respectively. We checked whether the species predicted by the function matched the actual species by using the morphology of the males used to construct the function. As a result, the function provided 100% correct answers. Using the discriminant function, we identified 138 males for species identification in males and 175 males developed from the immatures for species identification in females, both of which were collected from colonies at altitudes of 100–500 m. For species identification in males, 33 and 105 individuals were identified as S. miscanthi HG and S. sabelisi, respectively. Both species were found together in eight colonies at altitudes of 160–420 m, although S. sabelisi dominated in seven colonies and S. miscanthi HG in one colony (Fig. 2). In the species identification of females, 34 and 141 individuals were identified as S. miscanthi HG and S. sabelisi, respectively. Both species were found together in six colonies at altitudes of 150–430 m, although S. sabelisi was dominant in nine colonies (Fig. 2). These results indicate that their contact zone exists at least in the area at altitudes of 150–430 m on Mt. Amagi.

Fig. 2. Distributional patterns of S. sabelisi and S. miscanthi HG along altitude in both sexes based on the discriminant scores in the species identification by male morphology.

Squares show the field points, and the numbers near the squares correspond to the location number in Table 1. The numbers with underlines are the colonies used to construct the discriminant function, and others are the colonies identified by the discriminant function. The colour of the square shows the species composition, and white, black, and grey indicate S. sabelisi only, S. miscanthi HG only, and the mixture, respectively. The right part of the square shows males and the left part shows females. Lines on the map are contour lines. Map source: GSI website (https://www.gsi.go.jp/).

We examined the morphology of 312 males in total and found a deformed male (Fig. 3) in the males developed from the immatures collected at an altitude of 350 m (Amagi 10; Table 1; Fig. 2). The legs usually consist of four segments (tarsus, tibia, genu, and femur); however, in the deformed male, only three segments were present on one of the third legs (Fig. 3).

Fig. 3. The third leg of a normal and abnormal males.

(a) A normal S. sabelisi male specimen, (b) a normal S. miscanthi HG male specimen, and (c) an abnormal S. miscanthi HG male specimen. The males of (a) and (b) were collected from 680 m altitude (Amagi 20) and 20 m altitude (Amagi 1), respectively, and the male of (c) was collected from 350 m altitude (Amagi 10) and identified as S. miscanthi HG by measuring the other leg. Legs usually consist of four segments (tarsus, tibia, genu, and femur), as shown in (a) and (b); however, in (c), only three segments were present on one of the third legs.

Interspecific mating inferred from the sex ratio in the field

The male ratio (#male / (#male + #female)) varied among the mite colonies (male ratio: 0.077–0.355; Fig. 4). The effect of colony type on the male ratio was significant (binomial GLM, df = 2, LRT = 13.095, P < 0.01). The male ratio in the colonies in which both species were found (mixture) was significantly higher than that in the colonies of pure S. miscanthi HG (Tukey contrasts, z = 3.243, P < 0.01). The male ratio of the mixture seemed to be higher than that in the colonies of pure S. sabelisi, but the difference was not statistically significant (Tukey contrasts, z = 2.100, P = 0.088). The male ratio was not different between the colonies of pure S. miscanthi HG and S. sabelisi (Tukey contrasts, z = 0.861, P = 0.661), indicating that differences in male aggression did not significantly affect the male ratio in the field. These results are consistent with the possibility that more sons are produced than expected by chance due to (at least) partial reproductive incompatibilities between species when they co-occur in the same colony.

Fig. 4. Sex ratio (proportion of males) in mite colonies along altitude.

White circles show the mite colonies of pure S. sabelisi, black circles show the mite colonies of pure S. miscanthi HG, and grey squares show the mite colonies of a mixture of S. sabelisi and S. miscanthi HG. Each bar shows the 95% confidence interval.

Hybridization and genetic introgression

A total of 111 SNPs from 189 loci in 237 females were used for the genetic analysis. All raw reads of MIG-seq data were submitted to the DDBJ Sequence Read Archive under the BioProject ID PRJDB17427. The S. miscanthi HG populations had lower heterozygosity and nucleotide diversity than the populations of S. sabelisi (Table S1). Neighbour-Net analysis demonstrated the distinction between S. sabelisi and S. miscanthi HG as two separate genetic clusters (Fig. 5). None of the samples occupied an intermediate position between S. sabelisi and S. miscanthi HG. Under the delta K method, the best STRUCTURE model assigned individuals to two genetic clusters (Fig. S1), corresponding to S. sabelisi and S. miscanthi HG, respectively (Table S2; Fig. 6). A total of 232 individuals were assigned as pure S. sabelisi or pure S. miscanthi HG, and five individuals collected from their contact zones were assigned as S. sabelisi with very small fragments of S. miscanthi HG. These results showed that there were no hybrids and that genetic introgression was very low.

Fig. 5. Neighbor-Net network of S. miscanthi species group collected from Mt. Amagi.

The network was reconstructed based on the uncorrected P-distance. The filled cycle represents the samples.

Fig. 6. Population genetic structure in mite colonies along altitude.

The proportion of the genome of every individual originating from each of the inferred clusters, K = 2: Orange is cluster 1 corresponding to S. miscanthi HG, and blue is cluster 2 corresponding to S. sabelisi. Individuals 1 to 60 were from the colonies where S. miscanthi HG was found, individuals 61 to 101 were from the colonies where both S. miscanthi HG and S. sabelis were found, and 102 to 237 were from the colonies where S. sabelis was found according to the results of identification by male morphology. Individuals 8, 12, 15, 24, and 54 have elements of S. miscanthi HG but with very few elements of S. sabelisi. Individuals 69, 75, 87, 117, 120 and 202 have elements of S. sabelisi but with very few elements of S. miscanthi HG. For details, see Table S2.

Reproductive isolation in laboratory interspecific crossing experiments

In the crossing experiments with S. sabelisi females, the number of eggs in interspecific crosses was significantly higher than that in virgin oviposition (negative-binomial GLM, z = 3.418, P < 0.001; Table 2) and similar to that in intrapopulation crosses (negative-binomial GLM, z = 0.793, P = 0.428; Table 2), strongly suggesting that mating occurred in the interspecific cross. A few hybrids were produced in two of the 24 pairs of interspecific crosses (Table 2). The two pairs produced one daughter each and the average female ratio in offspring (#daughter / (#daughter + #son)) was 0.155. Overall, however, the female ratio in the offspring of interspecific crosses was significantly lower than that in intrapopulation crosses (binomial GLM, z = 7.919, P < 0.001; Table 2). The survival rate of offspring from egg to adult in interspecific crosses was significantly lower than that in intrapopulation crosses binomial GLM, z = 8.832, P < 0.001; Table 2). These results indicate the presence of a strong pre-zygotic barrier.

Table 2.

The number of eggs laid in 10 days after female emergence, the survival rate from egg to adult, the female ratio, and the ratio of pair with daughters in interspecific crosses, intra-population crosses and virgin oviposition in the cross experiments using S. sabelisi females (a) and S. miscanthi HG females (b).

Cross type	Cross combination		No. of pairs	No. of eggs			Survival rate from egg to adult			Female ratio in offspring			Ratio of pairs with daughters
	Female	Male		(Mean	±	SE)	(Mean	±	SE)	(Mean	±	SE)
(a) Cross experiment of S. sabelisi female
Inter-specific	Amagi 17 (S. sabelisi)	Amagi 1 (S. miscanthi HG)	24	7.708	±	0.850	0.797	±	0.060	0.013	±	0.009	0.083
Intra-population	Amagi 17 (S. sabelisi)	Amagi 17 (S. sabelisi)	27	8.444	±	0.590	0.951	±	0.017	0.806	±	0.014	1.000
Virgin oviposition	Amagi 17 (S. sabelisi)	-	23	4.870	±	0.352	0.891	±	0.037	0.000	±	0.000	-
(b) Cross experiment of S. miscanthi HG female
Inter-species	Amagi 1 (S. miscanthi HG)	Amagi 17 (S. sabelisi)	25	8.320	±	0.725	0.773	±	0.038	0.030	±	0.030	0.040
Intra-population	Amagi 1 (S. miscanthi HG)	Amagi 1 (S. miscanthi HG)	22	7.818	±	0.737	0.932	±	0.032	0.820	±	0.023	1.000
Virgin oviposition	Amagi 1 (S. miscanthi HG)	-	21	4.143	±	0.416	0.900	±	0.044	0.000	±	0.000	-

The results of the crossing experiments with S. miscanthi HG females were similar to those of the crossing experiments with S. sabelisi females. The number of eggs in the interspecific cross was significantly higher than that in the virgin oviposition (negative-binomial GLM, z = 4.870, P < 0.001; Table 2) and was similar to that in the intrapopulation cross (negative-binomial GLM, z = 0.513, P = 0.608; Table 2), indicating that mating occurred in the interspecific cross. A few hybrids were produced in one of the 25 pairs of interspecific crosses (Table 2). The pair produced three daughters and the female ratio in offspring was 0.750. Overall, however, the female ratio in the offspring of interspecific crosses was significantly lower than that in intrapopulation crosses (quasibinomial GLM, t = 6.311, P < 0.001; Table 2). The survival rate of offspring from egg to adult in interspecific crosses was significantly lower than that in intrapopulation crosses (quasibinomial GLM, t = 2.692, P < 0.01; Table 2) but not significantly different from that in virgin oviposition (quasibinomial GLM, t = 1.565, P = 0.122; Table 2). These results indicate the presence of a strong pre-zygotic barrier.

Discussion

We found that the geographic distributions of S. sabelisi and S. miscanthi HG overlapped widely at intermediate altitudes (150–430 m) on Mt. Amagi. We conducted a field survey on Mt. Amagi at the northern end of their parapatric area because a previous study detected a contact zone on Mt. Unzen at the southern end of their parapatric area (Sato et al. 2008; Fig. 1). As their contact zone was confirmed at both the southern and northern ends of their parapatric area, we concluded that their contact zone would be prevalent in their parapatric area and that their contact zone may have a significant impact on their evolution and ecological relationships.

In several host-plant fields in their contact zones, both species were found together for both sexes. This opens the possibility that interspecific mating has occurred in these fields. Consistent with this possibility, the sex ratio of spider mites in the colonies of both species was male-biased compared to that of pure S. sabelisi or S. miscanthi HG colonies. As for their reproductive barrier, the post-mating pre-zygotic barrier contributes the most (Sato et al. 2000a, 2000b, 2015, 2018), which causes the overproduction of males owing to their haplodiploid genetic system and regulation of the number of eggs by copulation stimuli. The male-biased sex ratio might support interspecific mating in the contact zone. A previous study conducted on Mt. Unzen detected a male-biased sex ratio at intermediate altitudes (Sato et al. 2008). Therefore, we conclude that interspecific mating might be widespread in their parapatric area.

A previous study reported the presence of a pre-mating barrier, but it was very weak and asymmetric (Sato et al. 2015). The post-mating pre-zygotic barrier contributes most to the reproductive barrier; however, it is also incomplete, and a few hybrids were produced in their cross-experiments in the laboratory (proportion of hybrids: 0–30%) (Sato et al. 2000a, 2000b, 2015, 2018). In this study, we carried out crossing experiments using the populations collected from Mt. Amagi and confirmed that they produce hybrids, although the proportion of hybrids was slightly lower compared to those in previous studies (3–13%). Therefore, we expected that hybrids would exist in the contact zone and that gene flow would occur between them. However, our genetic analysis did not identify obvious hybrids, and it was estimated that genetic introgression was extremely low (Figs. 5 and 6). This suggests that gene flow is strongly restricted by mechanisms that act later than post-mating pre-zygotic isolation. Because their hybrids are fertile in the laboratory (Sato 2004), they may have low fitness in nature. For example, adult female spider mites enter diapause and overwinter. Diapause attributes differ between the two species: S. sabelisi takes a much longer time to emerge from diapause (high-intensity diapause), whereas the S. miscanthi HG emerges from diapause very quickly (low-intensity diapause) (Saito et al. 2002). The difference in diapause is considered the result of adaptation to colder and warmer regions in each species, and this difference contributes to maintaining their parapatric distribution. The diapause attributes of their hybrids have not yet been investigated; however, if they are not able to enter diapause or emerge from diapause at a suitable time in the local season, they will be selected against from the field, even though they are able to survive in the laboratory. It is also possible that hybrids have lower fertility in nature as compared to the parental species. As described before, males of this species kill each other to establish their own harems, and only the victorious male can reproduce (Saitō 1990). In this study, we identified a deformed male in the contact zone. We do not know if the deformity was caused by hybridisation, but it is worth investigating the influence of hybridisation on the morphology and fighting ability of males. Therefore, to understand why gene flow is extremely restricted in the contact zone, it is necessary to investigate hybrid traits, such as diapause attributes, morphology, and behaviour.

This field survey was conducted at the northern end of S. miscanthi HG distribution. It is common for populations at the edge of the distribution to have extremely low genetic diversity owing to bottlenecks and genetic drift. Because the genetic diversity of S. miscanthi HG populations has not been investigated in other regions, it is not possible to confirm whether the genetic diversity is lower in this area than in other areas. However, S. miscanthi HG populations showed lower heterozygosity and nucleotide diversity than S. sabelisi (Table S1). This may explain the low genetic diversity of S. miscanthi HG populations in this study. The low genetic diversity and harshness of the environment for S. miscanthi HG likely make it difficult to produce hybrids. Therefore, before concluding that gene flow is extremely restricted between them in their parapatric area, it is worth conducting this study on other mountains where both species are abundant.

Although not as much as expected, a small level of genetic introgression was detected between S. sabelisi and S. miscanthi HG (Table S2; Fig. 6). The question arises: Why are SNPs (and perhaps some genes), once incorporated into the genome by interspecific hybridisation, not lost in natural population and kept at the lowest frequency? Most SNPs have little effect on fitness as they are often in neutral regions of the genome or a silent mutation (mutations that occur at the genetic level but do not show up in the phenotype). In that case, they may persist in the population at low frequencies simply due to genetic drift etc. On the other hand, some SNPs can be coding regions and have an effect on fitness (e.g. Wood et al. 2021). As described previously, these two species have different distribution areas and male aggression levels: S. sabelisi is distributed in colder regions and shows lower male aggression, whereas S. miscanthi HG is distributed in warmer regions and shows higher male aggression (Saito 1995; Sato et al. 2013a). Male aggression varies among populations in each species, and the same relationship between winter coldness and male aggression was found in each species. S. miscanthi HG populations distributed in relatively colder regions showed lower male aggression compared to populations distributed in warmer regions, and the same clinal trend was found in S. sabelisi populations (Saito and Sahara 1999). Male aggression is genetically determined, as laboratory mite populations maintain similar aggression regardless of rearing temperature. If SNPs (and perhaps some genes) incorporated into the genome by interspecific hybridisation affect male aggression, then a clinical trend in each species could be generated by the genetic introgression. For example, S. miscanthi HG populations distributed in relatively colder regions are expected to encounter S. sabelisi much more frequently than those in warmer regions. Milder male aggression was likely caused by genetic introgression from S. sabelisi. To determine if SNPs (and perhaps some genes) incorporated into the genome by interspecific hybridisation affect male aggression and thus persist in the populations, it is necessary to investigate the inheritance of male aggression, genetic introgression status in other contact zones, the common SNPs (and perhaps some genes) among the contact zones and the fitness of males with genes from other species.

In this study, we found that S. sabelisi and S. miscanthi HG have broad contact zones throughout their parapatric area, and interspecific mating probably occurs in their contact zones. Gene flow is strongly restricted between them, however, a small level of genetic introgression was detected. This study is the first study revealing the hybridisation and genetic introgression in the secondary contact zone of closely related species in spider mites, although spider mites have been paid attention to as model organisms in evolutionary biology (Belliure et al. 2010; Saito 2011). Although a study reported the hybridisation among strains of the red spider mite, Tetranychus evansi (Boubou et al. 2012), this is an example of an agricultural pest whose distribution has been spread rapidly by artificial activities, therefore, differs from secondary contact between closely related species brought by a long history in the so-called natural environment. The contact zones of closely related species influence their evolution in various ways; they can cause species breakdown, promote diversification by character displacement and/or reinforcement of reproductive barriers, and generate new species by hybridisation (Coyne and Orr 2004; Johannesson et al. 2020). In the spider mites, contact zones are likely to contribute to diversification. In particular, reinforcement of reproductive isolation can occur when the fitness of a hybrid is low. The traits and fitness of their hybrids have not been well investigated; however, it is quite likely that hybrids have problems surviving and reproducing in nature, considering that we did not find obvious hybrids in their contact zones, despite incomplete reproductive isolation. Haplodiploids, including spider mites, are often able to adjust offspring sex ratio by controlling egg fertilisation, although the ability varies among species (Hardy 2002; West 2010; Macke et al. 2010). This suggests that even after mating, females may have a way to avoid producing hybrids by controlling egg fertilisation. Given this, if other conditions are met, reinforcement of reproductive isolation may easily evolve in haplodiploids. Reinforcement of reproductive isolation in the spider mites was suggested by a previous study, which found that post-mating, a pre-zygotic barrier evolved faster in S. sabelisi females collected from parapatric areas than in allopatric areas (Sato et al. 2018). This tendency was not found in S. miscanthi HG females; however, it fits their behavioural differences (Sato et al. 2018). Specifically, males readily fight with different species of males for nests, S. miscanthi HG males tend to win interspecific male fights (Sato et al. 2013b), and S. miscanthi HG males are active to mate regardless of female species much more than S. sabelisi males (Sato et al. 2015). These behavioural differences suggest that S. sabelisi females are likely exposed to the risk of reproductive interference by other male species, whereas S. miscanthi HG females are guarded by the same male species against such risk. We expect genetic analyses to provide insights into this hypothesis. However, in this study, we could not analyse the direction and details of genetic introgression because the gene flow was extremely low. As discussed above, their gene flow may be restricted by mechanisms acting later than post-mating pre-zygotic isolation. However, the amount of gene flow likely changes depending on the power relationships between S. sabelisi and S. miscanthi HG. In this study, genetic analyses were performed on Mt. Amagi, where S. sabelisi was found to be abundant. From this perspective, it would be worthwhile to conduct this study on other mountains.

Author contributions

Y. Sato conceived the study. Y. Sato, S. Konaka, Y. Tsumura, and Y. Suyama designed the study. S. Konaka, Y. Sato, and N. Matsumoto conducted the field surveys. S. Konaka performed the measurements, Y. Sato and S. Konaka analysed male morphology. S. Hirota, Y. Suyama, S. Konaka, Y. Sato, and Y. Tsumura performed molecular analyses. Y. Sato conducted the cross-experiment and analysed reproductive isolation. Y. Sato and S. Hirota wrote the first draft of the manuscript. All authors read and approved the final manuscript.

Data availability

All raw reads of the MIG-seq data are available from the DDBJ Sequence Read Archive under BioProject ID PRJDB17427. The morphology, sex ratio, and cross-experimental data are shown in the Supplementary Materials.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41437-024-00708-y.