Varroa mites escape the evolutionary trap of haplodiploidy

Abstract

Genetic diversity is essential for populations adapting to environmental-changes. Following genetic bottlenecks, invasive species often have reduced diversity, yet must rapidly adapt to novel environments. This paradox is especially pronounced in parasites, which face repeated transmission bottlenecks and anthropogenic countermeasures. The challenge of maintaining diversity in the short-term intensifies in haplodiploid species, where males inherit and transmit only half of their mother’s genome, limiting effective population size. To examine how such species may maintain adaptability, we study inheritance in Varroa mites (Varroa destructor), the globally invasive honey bee parasite. Using three-generation pedigrees, we find that Varroa is not haplodiploid, as previously thought. Rather, females produce diploid clone sons who transmit either copy of their mother’s genome to daughters. This system slows reduced heterozygosity loss under sib-mating, compared to ancestral haplodiploidy, increasing effective population size and adaptability. Our findings reveal a rare reversion from haplodiploidy, long considered an evolutionary end state, and suggest reproductive plasticity underlies the resilience of invasive parasites like Varroa.

Many parasites thrive despite severe inbreeding, challenging classic ideas about how they maintain genetic diversity. This study shows that Varroa mites, long assumed to be haplodiploid, instead use a previously unknown system where males are functionally diploid clonal sons that preserve maternal variation and boost evolutionary potential.

Introduction

Genetic diversity is the basis for biodiversity, being required for species to evolve in response to environmental changes¹. For a given selection coefficient, the effectiveness of natural selection vs. genetic drift is determined by the effective population size². While selection dominates at larger population sizes, beneficial alleles can be lost due to drift in small populations. Biological invasions are often accompanied by genetic bottlenecks that reduce effective population size. Parasites are particularly at risk, as they often experience bottlenecks as part of their life cycle: an introduction to a new host often involves one or a few individuals, and the host-parasite population dynamics also involve frequent and drastic changes in the population’s size³. In addition, because of difficulties involved in finding unrelated mates within a host, many parasites frequently inbreed, which further decreases genetic variability^4–6. Although all the above scenarios are expected to decrease the effective population size, by definition, invasive parasites are successful species that adapt to different environments, hosts, and, in many cases, the chemical pesticides used to control them^7,8. This ‘genetic paradox of invasion’^9,10 raises the question of how invasive species with expected low genetic diversity are so successful. A few solutions have been suggested for this paradox. These include physiological explanations^11–13 and multiple invasions that compensate for the initial low genetic diversity^13–15. Occasionally, organisms evolve unusual reproductive strategies that help them avoid inbreeding. For example, in some ants, males and females constitute genetically distinct clonal lineages that combine to produce sterile workers but do not intermix otherwise^16,17. Detecting unusual reproduction mechanisms typically requires multi-generation-controlled crosses, which are often difficult, if not impossible, to recreate (especially in non-domesticated organisms), suggesting that many unusual reproductive modes remain to be discovered.

Here we examined the reproductive biology of the Varroa mite (Varroa destructor), a global invasive parasite of honey bees. These mites have experienced a range of genetic bottlenecks. They went through a genetic bottleneck when switching from their original host, the Eastern honey bee (Apis cerana), to the Western honey bee (Apis mellifera)¹⁸. They also lose genetic diversity when colonizing new hives. Lastly, they routinely sib-mate in small families (typically one male mates with one to two female sisters)^19,20. This mating behavior further increases the strength of genetic drift, and makes natural selection less effective. Yet, Varroa shows high adaptability: it efficiently switches between host species^21–23 and can rapidly evolve resistance to pesticides via various mechanisms²⁴. While a high level of adaptability explains the mite’s successful global spread, the mechanisms underlying this adaptability remain puzzling in light of its regular genetic bottlenecks and routine inbreeding.

Varroa is believed to be haplodiploid, whereby fertilized eggs develop into diploid females, while unfertilized eggs develop into haploid males^19,25–27. Haplodiploid species have a smaller effective population size relative to diplodiploid systems^28,29, although haplodiploidy may confer long-term advantages in inbreeding systems, such as more effective purging of deleterious alleles³⁰. In Varroa, haplodiploidy was inferred from two cytogenetic studies that identified haploid (n = 7) and diploid (2n = 14) embryos in reproducing adult female mites^31,32. Although these studies referred to their material as V. jacobsoni, subsequent taxonomic work has shown that pre‑2000 reports of “V. jacobsoni” parasitizing A. mellifera actually correspond to V. destructor^33–35, so they are directly comparable to our data. Ruijter & Pappas (1983) further observed that early-laid eggs, corresponding to the male, which is typically the first offspring produced, yielded haploid embryos, while later-laid eggs were diploid. Together with reports that unmated females can produce only males¹⁹, these findings provided reasonable, if indirect, evidence for arrhenotokous haplodiploidy rather than paternal genome elimination (PGE) in Varroa.

Haplodiploidy is believed to be evolutionarily stable, sometimes being referred to as an “evolutionary trap”, suggesting limited potential for an evolutionary reversion to diploidy^36,37. Cruickshank and Thomas³⁸ used phylogenetic reconstruction to infer that Varroa evolved from diploid ancestors via intermediate pseudo-arrhenotokous species. While Varroa was included in that study, its haplodiploidy was assumed based on the fact that all members of its family were haplodiploid. However, the taxon sampling was sparse and, until recently, the phylogenetic placement of Varroa was uncertain, opening the study’s conclusions to criticism³⁹. Indeed, more complete investigations were hampered by the uncertain taxonomic placement of Varroa, until a recent study placed it as a subfamily within the family Laelapidae, allowing a more targeted comparison among species⁴⁰. Of the 19 sexually reproducing Laelapid mites in the Tree of Sex database⁴¹, only a single species is not haplodiploid: Dicrocheles phalaenodectes, a parasite inhabiting the ear canal of some Noctuid moths. However, the placement of Dicrocheles within Laelapidae is based solely on morphological data and has been debated⁴²; it is likely only distantly related to Varroa. Thus, based on the available evidence, haplodiploidy was ancestral to Varroa.

Theory suggests that reversal to diploidy from arrhenotoky is unlikely³⁷, and, indeed, no reversals of arrhenotokous sex determination are known to date⁴³. Yet, despite strong circumstantial evidence for arrhenotokous parthenogenesis in Varroa, its actual mode of genetic inheritance has never been directly tested, limiting our understanding of how this parasite maintains such high evolutionary potential under repeated genetic bottlenecks. Here, we explicitly examined this species’ mode of genetic inheritance using multi-generational pedigrees and revealed that Varroa is not haplodiploid, but instead females produce diploid clonal sons, a novel reproductive system that preserves maternal variation under sib-mating and enhances the parasite’s evolutionary potential.

Results

Two lines of evidence suggest that Varroa males are functionally diploid: [1] their somatic cells inherited and expressed both maternal alleles (Figs. 1 and 2), and [2] they randomly transmitted these alleles to their daughters (Fig. 1). Modeling shows that this increases the mite’s effective population size relative to ancestral haplodiploidy (Supplementary Fig. 1).

Fig. 1. Inheritance patterns in Varroa destructor mites show that males inherit two alleles from their mother and transmit both alleles to their daughters with equal probability. Females are produced sexually.

For each of the two cross types, we first filtered for informative genotypes in F0 and F1 and then observed the genotypes of those sites in F2 females and males, to infer the mode of genetic inheritance. a,c show the genotype flow for representative families (families 478 and 534, respectively) over three generations along the largest chromosome (NW_019211454) (“F0-F2 one family, one chromosome”). In cross 1 (panel A), the F0 female was homozygous (AA), and her F1 offspring were a heterozygous female (AB) and a homozygous male (AA). This filtering yielded 33 sites, whose genotypes were then observed in the F2 female and male offspring. In cross 2 (c), the F0 female was heterozygous (AB), and her F1 offspring were a homozygous female (AA) and a heterozygous male (AB). This filtering yielded 14 sites, whose genotypes were then observed in the F2 female and male offspring. b,d (“F2 pooled families, all sites”) show pooled genotype proportions in F2 females and males across all families and all seven chromosomes for cross 1 (AB × AA; panel B) and cross 2 (AA × AB; D). In both crosses, F2 females showed an approximately 1:1 ratio of AA:AB genotypes, consistent with sexual reproduction, whereas F2 males predominantly inherited their mother’s genotype, indicating obligatory maternal inheritance. As expected from AA × AB crosses, no BB genotypes were observed. Colors: yellow = AA, green = AB, blue = BB. F0 males were not collected; therefore, their genotypes are unknown and shown in gray.

Fig. 2. RNA-seq analysis of heterozygous site proportion in male and female Varroa mites.

The two sexes showed similar heterozygosity proportions in their RNA across five families (mean difference = 0.022 ± 0.10, two-sided paired t-test: t(4) = 0.46, p = 0.672, Cohen’s d = 0.20, 95% CI [−0.70, 1.08]). The y-axis shows the proportion of heterozygous sites of total sites analyzed. In each boxplot, the horizontal line indicates the median, the box boundaries represent the interquartile range (25–75th percentiles), and the whiskers extend to the most extreme values within 1.5 × the interquartile range; individual points show the heterozygosity values for each mite.

Males are produced by diploid arrhenotoky, preserving their mothers’ genotype

Adult males inherited both of their mother’s alleles, as evident from the three-generational pedigree study. Looking at the cross between a heterozygous F1 female (AB) and a homozygous F1 male (AA), we found that all variants of the F2 males maintained their mother’s heterozygosity (Fig. 1a, example of family 478). This was consistent across families, as pooled data showed that 97% of the genotypes in the F2 males maintained their mother’s heterozygosity (Fig. 1b). Similarly, when crossing a homozygous F1 female (AA) and a heterozygous F1 male (AB), variants of the F2 males were homozygous like the F1 mother (Fig. 1c), and the pooled-family’s data were consistent (Fig. 1d). Both copies of their mother’s genome were also expressed, as evidenced from heterozygosity in RNA-seq data of adult female mites and their sons (Fig. 2, Supplementary Table 1). Furthermore, the two sexes showed similar heterozygosity proportions in their RNA across five families (mean difference = 0.022 ± 0.10, paired t-test: t(4) = 0.46, p = 0.672, Cohen’s d = 0.20, 95% CI [−0.70, 1.08]).

Males transmit both copies of their mother’s genome to their daughters with equal probability

Males not only inherited both copies of their mother’s genome, but also transmitted each copy to their daughters with equal probability across all examined loci, in a typical Mendelian manner. Looking at the cross between a homozygous F1 female (AA) and a heterozygous F1 male (AB), we observed that about ~50% of the variants in the F2 females were homozygous (like their mother) and the rest were heterozygous (like their father) (Fig. 1c, d). F2 male offspring of the same cross were homozygous (AA), just like their mother. This was consistent across families, as shown by the pooled data (Fig. 1d).

All else being equal, Varroa diploid male arrhenotoky has a higher inbreeding effective population size relative to ancestral haplodiploidy

We conducted genetic transmission path modeling, which traced genetic transmission from parents to offspring and quantified genetic relatedness through path coefficients, following Wright⁴⁴. Conditional on the Varroa life cycle, fertility, and brood-cell occupancy rate, the model showed that Varroa has larger inbreeding effective population size than haplodiploids. However, the effective population size is directly linked to heterozygosity using this relationship:

$$N_e=-\frac{1}{2\left(\frac{p_t}{p_{t-1}}-1\right)}$$

where $p_t$ is heterozygosity at generation $t$ (see Mathematical Modeling in Methods), so the results are equivalent whether expressed as Ne or as heterozygosity decline. Therefore, Varroa loses heterozygosity more slowly than haplodiploids. This difference is pronounced during the early stages of colonizing a new host, when populations are small and bottlenecked. As populations approach the carrying capacity, the difference narrows, but Varroa still loses slightly, but consistently, less heterozygosity than haplodiploids (Supplementary Fig. 1). The difference arose because the two alleles in male genomes lowered the correlation between uniting gametes and attenuated inbreeding compared with haplodiploid mating, whereby the male inherited a single allele. Consequently, for the same census size, the effective population size was higher under diploid arrhenotoky.

Discussion

In this study, we found that V. destructor has evolved from ancestral haplodiploidy³⁸ to a novel reproductive strategy with diploid arrhenotokous males that are clones of their mothers (see Fig. 3 for a scheme of the inferred mode of genetic inheritance). This strategy potentially has short-, and long-term benefits for Varroa, each overcoming a different evolutionary “paradox”. In the short term, the production of clonal males increases the female’s fitness by (1) insulating her sons from the phenotypic expression of deleterious recessive alleles and de novo lethal mutations, and (2) acting as a demographic bet-hedge against genetic drift during the F1 reproductive bottleneck. By perfectly preserving both maternal alleles in the son’s sperm pool, the female minimizes the variance in her F2 genetic representation without suffering the demographic cost of producing a second son. This strategy can overcome the “paradox of sex” by blending the benefits of sexual and asexual reproduction. Namely, females produce diploid sons that inherit both copies of their genome, while daughters are produced sexually via meiosis. Consequently, introducing Mendelian segregation through the male lineage slows the loss of heterozygosity under sib-mating relative to ancestral haplodiploidy, thereby reducing the correlation between uniting gametes and slightly increasing the population size (Supplementary Fig. 1). In the long term, functional diplodiploidy can potentially increase the effective population size (vs. haplodiploidy) (Supplementary Fig. 1). A higher effective population size imparts greater evolutionary potential, and indeed Varroa has been remarkably adaptable despite repeated bottlenecks, exemplifying the “genetic paradox of invasion”.

Fig. 3. Hypothesized mode of genetic inheritance in Varroa male and female mites.

Females are produced via diploid sexual reproduction, while males are produced via diploid arrhenotoky. The left schematic shows diploid arrhenotoky, in which an unfertilized egg develops into a diploid son. The right schematic shows diploid sexual reproduction, in which fertilized eggs develop into diploid daughters. Orange chromosomes and segments denote maternal alleles, and blue chromosomes and segments denote paternal alleles. Female mites are indicated by the ♀ symbol and male mites by the ♂ symbol.

Our finding that adult Varroa males are functionally diploid creates an apparent tension with the karyotyping studies of Steiner et al. (1982) and Ruijter & Pappas (1983), which identified haploid embryos (n = 7) at developmental stages corresponding to the timing of male offspring. Notably, both studies are independently corroborated: they were conducted on different continents with different staining methods, and the association with male egg-laying order provides a meaningful indirect link to sex. It is therefore possible that a somatic ploidy reduction occurs at an early point during male development. This would have a partial precedent in phytoseiid mites, where paternal genome elimination (PGE) produces functionally haploid males from diploid embryos⁴⁵. However, the Varroa case would be mechanistically distinct from PGE: our pedigree data showed that males transmit both alleles in a true Mendelian manner, with no evidence of parent-of-origin-specific allele loss. Whether the embryonic n = 7 reflects a genuine ploidy transition during male development (i.e., through genome reduction, diploidization, or differential endoreduplication^46,47), cannot be resolved from the available data and warrants a separate dedicated investigation.

Another evolutionary surprise from this study is the effective loss of haploid arrhenotoky. Once evolved, asymmetric inheritance appears to reach an evolutionarily stable state, sometimes referred to as an “evolutionary trap”³⁷. Obstacles to re-evolving Mendelian inheritance are diverse. One obstacle is mechanistic, in that males would need to re-evolve meiosis, which has always been thought to be difficult⁴⁸. However, recent work from asexual stick insects indicates that male meiotic regulatory mechanisms are preserved in females⁴⁹, so re-evolving male meiosis may be easier than expected. The other obstacle is evolutionary, as examined in detail by Bull⁵⁰. In that scenario, females reach a fitness optimum in haplodiploid systems, being able to control male production and thus the sex ratio. Therefore, the re-emergence of diploid males would have to be a male trait, so biparental diploid males would have to carry mutations that would allow them to produce diploid sons. However, given the level of arrhenotokous female control over fertilization and sex ratios, those traits are theoretically unlikely to spread⁵⁰. Indeed, the only evolutionarily feasible transition from haplodiploidy appears to be towards parthenogenesis, though these are short lived⁵¹. Yet, Varroa overcame the constraint on the production of diploid sons by having females solely control both the sex ratio, and diploid male production. The production of meiotic sperm and the evolution of diploid arrhenotokous male production illustrate how the unforeseen complexity of biological mechanisms may challenge long-standing elegant mathematical models.

Although the mechanism of sex determination in Varroa remains unknown, other parthenogenetic systems suggest how diploid males might be produced from unfertilized eggs. Possible automatic mechanisms include central fusion of meiotic products, as in the Cape honey bee (A. m. capensis)⁵², and terminal fusion combined with inverted meiosis, as reported for some thelytokous mites⁵³. Diploid arrhenotoky also occurs in scale insects, but their diploidy is restored by gamete duplication, producing homozygous or functionally haploid males^54,55, contrasting with Varroa males, which retain maternal heterozygosity and transmit both copies of the maternal genome in a Mendelian manner (Fig. 1). Our results are not compatible with classical complementary sex determination: highly inbred Varroa populations would generate many diploid males, and our diploid heterozygous males would be predicted to develop as females under that system. A maternal factor is therefore more likely responsible, for example epigenetic modification, genomic imprinting, interactions with endosymbionts, or female control over fertilization and diploidy restoration^43,56,57.

Our work underscores the importance of comprehensive experimental inquiries into diverse reproductive mechanisms, even in organisms that appear to be well studied. Recent investigations have unveiled a broader spectrum of reproductive strategies than were previously recognized⁴⁸, exemplifying the dynamic nature of biological systems. Yet, the discovery of a novel reproductive system, such as that in Varroa, can have wide-ranging fundamental and applied consequences. For instance, Varroa’s haplodiploidy is assumed when considering how it evolves in response to countermeasures, such as using pesticides, breeding honey bees for parasite tolerance, or meiotic drive^58–60. Furthermore, Varroa has emerged as a compelling model for the study of the evolutionary dynamics in reproductive systems, offering insights into mechanisms for overcoming evolutionary constraints and the intricacies of sex determination. Our findings open promising avenues for future research, contributing to a deeper understanding of evolutionary innovation and reproductive plasticity in Varroa and beyond.

Methods

Honey bees and mites

For the pedigree study

Varroa mites and honey bees (Apis mellifera L.) were obtained from colonies at the Onna village apiary and the experimental apiary of the Okinawa Institute of Science and Technology (OIST) in Okinawa, Japan. All colonies used in the experiment were maintained without any treatment against Varroa and received a 60% w:v sugar solution and 70% pollen patties as supplemental feed when necessary. Experiments were conducted between July and November 2020.

For transcriptomic analysis of Varroa heterozygosity in mothers and sons

Varroa mites were collected from A. mellifera colonies at Zhejiang University in Hangzhou, China, in November 2024. A total of ten mites were collected from five families, each consisting of a mother and her son.

Artificial infestation for Varroa pedigree construction

To obtain a three-generational mite pedigree (F0, F1, and F2), we used a semi-natural infestation system. This started with the collection of a naturally infesting mated Varroa female (F0). Her adult daughter (F1) was then artificially introduced into a honey bee pre-pupal cell and allowed to commence the reproductive phase of its life cycle. The F1 mite and her offspring (F2) were then collected at the end of the reproductive cycle. For a detailed description of the experimental design, see Fig. 4 and the sections below. Using this approach, we could genotype all members of the pedigree, except for the male who mated with the F0 female.

Fig. 4. Pedigree experimental design.

A three-generation Varroa pedigree (F0-in red, F1-in orange, and F2-in yellow) of 30 families was constructed via the following steps: a Collecting a naturally infested family: foundress female mite (F0), and her two adult offspring: male and female F1 mites (F0 male’s genotype was not possible to study because the F0 female was previously mated, typically by a single male); b “Incubation phase”: Transferring the young F1 female onto a bee pupa for 1-4 days, till cuticle sclerotized and darkened; c “Dispersal phase”: Transferring F1 female onto a nurse adult bee for another 3 days; d Introducing the matured F1 female to a freshly capped 5th instar larva cell; e F1 female reproducing inside the bee cell; f 10 days post introduction, collecting the artificially infested family: F1 female and her F2 offspring. The F2 generation included a son and one or more daughter nymphs. This design provided known parents and genotypes for the F1 and F2 generations, allowing a detailed examination of inheritance.

Collection of naturally infesting families (F0 and F1 generations)

To obtain mites from the first two generations (F0 and F1), naturally infested families were collected from within the cells of 18- to 20-day-old worker bee pupae (Fig. 4a). At this developmental stage of the bee host, there is typically one dark brown female mite (the foundress, F0), one adult male (the foundress’ son, F1), at least one light-colored female (the first foundress’ daughter, F1), a few nymphs females, and an egg²⁵. We determined each mite’s sex and developmental stage according to external morphology and cuticle coloring based on⁶¹. For the color index, see Supplementary Fig. 2. To ensure that all offspring belonged to a single family, we thoroughly examined the cell content to confirm that it was infested by a single foundress mite and that no other adults were present. Except for the adult female daughters (F1), the rest of the family members were preserved in a separate microcentrifuge tube kept at −20 °C for later DNA extraction. The adult F1 daughters were easily distinguished from the F0 foundress mite by their lighter cuticle color (color index 0–2, see Supplementary Fig. 2), and were assumed to have already copulated with their brother, the F1 male, by the time of the collection¹⁹. Next, the F1 young adult females were prepared to be artificially introduced into a freshly capped 5^th instar larva cell to start their first reproductive cycle.

Incubation and dispersal phases of F1 female mites

Based on our and others’ observations⁶², the survival and fecundity of young adult mites increase if they stay for a few days on an adult nurse bee before starting their first reproductive cycle. This period is presumably imitating the “dispersal phase” (formerly called the “phoretic phase”), in which the adult female mite infests and feeds on an adult honey bee^62,63. Therefore, the collected young adult females (F1 generation) spent three days on a nurse bee before infestation. However, during the experiment, we noticed that the very young adult daughter’s cuticle was light and fragile, and most of the mites did not survive when they were transferred directly from the pupae onto a nurse bee. Therefore, we let the young mites first feed on a white-eyed pupa in a gelatin capsule for one to four days (“incubation phase”, Fig. 4b) until their cuticle was sclerotized and darker (color index 3–4, see Supplementary Fig. 2). After this intermediate period, the mites were marked on their dorsal idiosoma using a water-based paint marker pen (PC-1MR POSCA, Japan), and transferred onto an adult nurse bee in a cage for another three days, for the above-mentioned “dispersal phase” (Fig. 4c). In all incubation periods, the mites were kept in a controlled dark environment, imitating hive conditions (34–34.5 °C, 60–75% RH).

Artificial infestation of F1 female mites

After the artificial “dispersal phase”, the mites were brushed off the bees using a fine paintbrush following anesthesia via a 20-s exposure to CO₂, 0.2mPa at 10 L/hour, and placed on a moist filter paper to avoid desiccation until the infestation procedure. The mites were then introduced to a worker bee cell, one mite per cell, as described before⁶⁴ (Fig. 4d). In detail, six hours before infestation, the location of cells containing a 5th instar larva (about to be sealed) was marked. This was done by overlaying and pinning an acetate sheet on the wooden frame of a low‑mite hive. Each target cell was located visually and its position was traced on the acetate sheet by drawing a small circle at the corresponding cell location with a permanent marker. The frame was then placed back into the hive so that the bees could seal the cells. Five hours later, the frame was taken into a warm room (25 °C–28 °C) for the F1 females to be introduced into the freshly sealed cells. For the artificial infestation, a precise incision (3–5 mm) was made at the margin of the cell-capping using a surgical blade (N 11). After the incision was made, the cell capping was slightly lifted, just enough for the mite to crawl under the capping and into the cell. The mite was then placed at the entrance of the cell using a moist fine paintbrush, with its gnathosoma facing the cell opening. After the mite deliberately entered the cell, it moved under the larva to the bottom of the cell. The cap was then resealed by gently pushing it down using the back end of the paintbrush. The infested frame was then incubated in a dark controlled environment (34-34.5 °C, 60-75% RH) for ten days, during which the mites were reproducing (Fig. 4e). During this period, emerging bees were removed daily.

Collection of an artificially infested family (F2 generation)

Ten days after the F1 mites’ infestation, the cells were opened, and their contents were examined (Fig. 4f). The family members’ sex, developmental stage (adult, nymph, or egg), and viability were recorded. To ensure the parenthood of the offspring (F2), double-infested cells and unmarked mites were not included in the final analysis. In total, out of 78 successful infestations, 14 were double-infested by a non-marked mite. Each individual was snap-frozen in a separate microcentrifuge tube at -20 °C for later DNA extraction. In total, we collected 223 mites belonging to 30 families. Each family was composed of a foundress adult female (F0), her male and female adult offspring (F1), and their F2 offspring (at least one male/female mite in various developmental stages).

Genomic analysis of the Varroa pedigree

DNA extraction, library preparation, and sequencing

Genomic DNA was extracted from each mite following the protocol described previously²³. In brief, after surface sterilization using absolute ethanol, each mite was crushed in a 1.5 mL microcentrifuge tube dipped in liquid nitrogen. The crushed mite was then processed using a QIAamp DNA Micro Kit (© Qiagen) following the manufacturer’s instructions and eluted in 20 μL of water. Total dsDNA quantity was measured using a Qubit™ 4 Fluorometer with an Invitrogen dsDNA HS Assay Kit, and DNA quality was verified using a Bioanalyzer 2100 with high-sensitivity DNA kits (Agilent, Japan). Short inserts of 150-bp paired-end libraries were prepared for each individual using a Nextera XT DNA Library Preparation Kit (Illumina, Japan) and 16 PCR cycles. cDNA was then cleaned and size-selected using CA-magnetic beads (Dynabeads® MyOne Carboxylic Acid, Invitrogen; Thermo Fisher Scientific, USA), and 11–11.5% PEG 6000 (Sigma-Aldrich, Japan). The library’s quality was measured using a Qubit™ 4 Fluorometer with an Invitrogen dsDNA HS Assay Kit, and its size was assessed using a 4200 Tapestation with a high-sensitivity D5000 kit (Agilent, Japan). Libraries were run on a NovaSeq6000 in 150 bp x 2 paired-end mode (Illumina, Japan) at the OIST Sequencing Center. All biosamples are available in the Sequence Read Archive (SRA) under the accession number PRJNA794941. Biosample details including sex, developmental stage, pedigree information, and sequencing coverage, are provided in Supplementary Table 2.

Read mapping, variant calling, and filtering of genomic data

We mapped all reads to the V. destructor reference genome GCF_002443255.1¹⁸ using the very sensitive mode of NextGenMap v0.5.0⁶⁵. Reads were sorted, and duplicates were removed using SAMtools⁶⁶. We called variants against the reference genome using FreeBayes v1.1.0⁶⁷ with the following parameters: minimum alternative allele count = 5, minimum alternative allele fraction = 0.2, minimum mapping quality = 8, minimum base quality = 5, and using the four best SNP alleles. The variant calling resulted in a total of 73,903,468 variant sites. As our population consisted of closely related individuals (mite families from the same apiary, in a highly inbred system), we did not expect to detect many variants. Therefore, we set the minor allele frequency threshold to 0.2, which included sites with a minor allele frequency of 20% or more, to exclude any rare variants due to mutations or possible mapping errors. We included variants only for the seven chromosomes (excluding the mitochondrial DNA and unscaffolded regions). In addition, we filtered for the following site quality parameters: we included only bi-allelic sites, a depth range between 16 and 40, and a maximum missing data per site = 0.5. The variants’ quality distribution showed a clear bimodal distribution with a cutoff at ~10,000 (Supplementary Fig. 3). To ensure the quality of the variants for the genetic inheritance analysis, we chose a strict minimum quality filter of 15,000. Filters were applied using VCFtools v0.1.12b⁶⁸, resulting in 33,925 final variants from 223 individuals. The pipeline from FASTQ files down to the VCF file was constructed using Snakemake⁶⁹. For further security of site quality, we applied two additional site quality parameters to account for biases in alternate allele representation and to ensure high-confidence genotype calls based on allele frequency and quality. Variant allele frequency (VAF) was calculated as VAF = AO/DP, where AO (Alternate Observations) represented the number of reads supporting the alternate allele, and DP (Depth) was the total read depth at the site. Similarly, Quality Allele Frequency (QAF) was computed as QAF = QA/QA + QO, where QA (Quality of Alternate Allele) was the sum of Phred-scaled base quality scores for reads supporting the alternate allele, and QO (Quality of Reference Allele) was the sum of Phred-scaled base quality scores for reads supporting the reference allele. To minimize errors in genotyping due to sequencing artifacts or depth-related biases, we applied genotype-specific thresholds for VAF and QAF based on observation of values from F2 females of which parents and grandmother's genotypes were fixed, and so their own genotype was assumed with high confidence (Supplementary Figs. 4–6 for QAF and VAF distribution, in genotypes AA, BB, and AB, in accordance). The thresholds (Table 1), ensured that homozygous reference sites (AA) exhibited low alternate allele support (VAF and QAF < 0.01), heterozygous sites (AB) maintained a balanced allele distribution (0.2 < VAF and QAF < 0.8), and homozygous alternate sites (BB) demonstrated high alternate allele support (VAF and QAF > 0.9). This filtering approach enhanced the accuracy of genotype classification while reducing the inclusion of low-confidence variant calls.

Table 1.

Quality Allele Frequency (QAF) and Variant Allele Frequency (VAF) threshold and their ideal values are used for sites’ filtration in the genotype flow of mite pedigrees

Genotype	VAF and QAF thresholds	Ideal value, interpretation
AA (Homozygous Reference)	<0.01	~0, Strong reference support
AB (Heterozygous)	0.2–0.8	~0.5, Balanced allele representation
BB (Homozygous Alternate)	>0.9	~1, Strong alternate support

VAF and QAF are both defined as unitless frequencies ranging from 0 to 1.

Genotype validation using Sanger sequencing

Several genotype calls obtained from whole-genome sequencing were validated by Sanger sequencing using the same genomic DNA extracted. For the validation, we randomly selected three genes, located on different chromosomes, each containing at least one variant site. In total, we re-sequenced 14 individual mites from four families on six variant sites. The sequenced region was located on the exons of three genes: alpha-mannosidase 2x-like (LOC111243621 on chromosome NW_019211455.1), 5-oxoprolinase-like (LOC111252938, on chromosome NW_019211459.1), and probable E3 ubiquitin-protein ligase RNF144A-A (LOC111254119, on chromosome NW_019211460.1). Primers were designed with the NCBI primer design tool (utilizing Primer3 and BLAST), with default parameters and product size set to 300–1000 bp. Primer sequence, product length, and gene accession numbers are provided in Supplementary Table 3. The size of the PCR product was checked by running it in 1% agarose gel (135 V, 20 min), purified, and size-selected using the solid-phase reverse immobilization (SPRI) method. The purified PCR products were Sanger sequenced on a ThermoFisher SeqStudio Genetic Analyzer using the original reverse primer and BigDye® Direct Cycle Sequencing kit (Thermo Fisher Scientific, USA), following the manufacturer’s instructions. We then aligned the reverse complementary sequences to the Varroa reference genome (GCF_002443255.2_Vdes_3.0) to detect and validate the variant sites using Geneious Prime 2022.1.1. (https://www.geneious.com). For sequence alignments and the validated sites, see Supplementary Fig. 7).

Transcriptomic analysis of Varroa mother and son

RNA extraction, library preparation, and sequencing

Total RNA was extracted from every single mite and sequenced by Metware Biotechnology Co., Ltd. (Wuhan, China) following an in-house protocol. RNA integrity was assessed using a Qubit™ 4 Fluorometer (Invitrogen; Thermo Fisher Scientific, USA) and a Qsep400 Bioanalyzer. For mRNA enrichment, polyadenylated RNA was isolated using Oligo(dT) beads, followed by fragmentation with a fragmentation buffer. First-strand cDNA was synthesized using random hexamers and reverse transcriptase, while second-strand synthesis was carried out with dNTPs and DNA polymerase I. The resulting double-stranded cDNA was purified, subjected to end-repair, A-tailing, and adapter ligation, followed by PCR amplification. Library quality was assessed using a Qubit™ 4 Fluorometer with an Invitrogen dsDNA HS Assay Kit, and insert size distribution was confirmed using a Bioanalyzer (Agilent, Japan). The final libraries were pooled and sequenced on an Illumina platform in paired-end mode. All biosamples are available in the Sequence Read Archive (SRA) under the accession number PRJNA1216898. Biosample details, including sex, family, and mapped reads, are provided in Supplementary Table 1.

Read mapping, variant calling, and filtering of RNA-seq data

RNA-seq reads were mapped to the V. destructor reference genome (GCF_002443255.2_Vdes_3.0) using HISAT2 v2.2.1 in default settings⁷⁰. The mapping rate varied between samples, with uniquely mapped reads ranging from ~5.88% to 15.71% of the total reads. Clean reads were obtained after quality filtering using fastq v0.23.2⁷¹, with adapters removed, reads containing more than 10% N bases discarded, and bases with quality scores below 20 trimmed. Variant calling was performed using GATK⁷², and SNPs and InDels were annotated with ANNOVAR⁷³. To ensure high-confidence variant sites, we filtered out low-quality SNPs and retained only sites with adequate coverage and quality metrics, although the exact thresholds were not specified in the company report. We observed thousands of exonic variants per sample, including both synonymous and nonsynonymous SNVs, as well as stop-gain mutations. The resulting variant dataset included annotations for exonic, 5’ and 3’ UTR regions.

Mathematical modeling

We followed the classical approach used by Wright⁴⁴ to model declines in heterozygosity due to inbreeding within haplodiploid systems. This approach involves constructing path diagrams representing genetic transmission paths from parents to offspring and quantifying genetic relatedness through path coefficients. By enumerating paths and calculating their coefficients, we derived three mathematical relationships describing the expected declines in heterozygosity per generation under the reproductive life cycle of Varroa, one for each reproductive system (haplodiploidy, diploid arrhenotoky, and complete diploidy). The derivations yield three equations describing the per-generation decline in heterozygosity, translated into the inbreeding effective population size ($N_e=-\frac{1}{2\left(\frac{p_t}{p_{t-1}}-1\right)}$, where $p_t$ is heterozygosity at generation $t$). Supplementary Fig. 1 plots $N_e$ over generations for haplodiploidy, diploid arrhenotoky, and complete diploidy under two demographic scenarios that differ in absolute colony size while keeping the ratio of carrying capacity (C) to brood cells (K) constant (C/K = 2). $N_e$ values for all three systems at carrying capacity and at equilibrium are reported in the legend. Mathematical details are provided in Supplementary Information 1.

Statistical analysis

All analyses were carried out in the R statistical environment (version 4.2.2)⁷⁴. For the RNA‑seq heterozygosity comparison between mothers and sons, we calculated per‑individual heterozygosity and tested for differences using a paired t‑test with the family as the unit of replication. For the modeling results, we derived analytical expectations for heterozygosity decline and effective population size and evaluated them numerically across parameter ranges in R. All codes used to perform the analyses and generate figures are available in the accompanying R scripts and at the project GitHub repository, as described in the Code availability section⁷⁵.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

N.E., T.E., and M.A.T. performed the experimental work. N.E. and A.S.M. performed the data and statistical analyses. X.H. performed the modeling analyses. N.E., T.E., X.H., M.A.T., J.R., H.Z., E.P.E., and A.S.M. interpreted the results and contributed to writing the manuscript. All authors reviewed and approved the final version of the manuscript.

Peer review

Peer review information

Nature Communications thanks Bart Pannebakker and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

A.S.M. was supported by a Future Fellowship from the Australian Research Council (FT160100178). H.Z. was supported by the earmarked fund for CARS (CARS-44).

Data availability

The whole‑genome sequencing data generated in this study have been deposited in the Sequence Read Archive (SRA) under accession code PRJNA794941 (Varroa pedigree genomic data). The RNA‑seq data generated in this study have been deposited in the SRA under accession code PRJNA1216898 (Varroa mothers and sons). The raw sequencing data are publicly available without restrictions at these repositories. Processed variant calls, pedigree metadata, and modeling inputs are provided in the Supplementary Information (Supplementary Tables S1–S3 and Supplementary Text S1). Source data is provided as a Source Data file. Source data underlying all main text figures and Supplementary Figs. are provided in the accompanying Source Data file. Source data are provided with this paper.

Code availability

All analyses and figure generation are reproducible from the accompanying R scripts. The full analysis code is available at the GitHub repository “Varroa‑pedigree‑study” Zenodo, https://doi.org/10.5281/zenodo.21603686 and at the Rpubs companion document (https://rpubs.com/Nurit_Eliash/1432436).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77125-8.