The meaning of wild: Genetic and adaptive consequences from large-scale releases of domestic mallards.

The translocation of individuals around the world is leading to rising incidences of anthropogenic hybridization, particularly between domestic and wild congeners. We apply a landscape genomics approach for thousands of mallard (Anas platyrhynchos) samples across continental and island populations to determine the result of over a century of supplementation practices. We establish that a single domestic game-farm mallard breed is the source for contemporary release programs in Eurasia and North America, as well as for established feral populations in New Zealand and Hawaii. In particular, we identify central Europe and eastern North America as epicenters of ongoing anthropogenic hybridization, and conclude that the release of game-farm mallards continues to affect the genetic integrity of wild mallards. Conversely, self-sustaining feral populations in New Zealand and Hawaii not only show strong differentiation from their original stock, but also signatures of local adaptation occurring in less than a half-century since game-farm mallard releases have ceased. We conclude that 'wild' is not singular, and that even feral populations are capable of responding to natural processes. Although considered paradoxical to biological conservation, understanding the capacity for wildness among feral and feral admixed populations in human landscapes is critical as such interactions increase in the Anthropocene.

A Comparison of Host Responses to Infection with Wild-Type Avian Influenza Viruses in Chickens and Tufted Ducks.

Cross-species transmission of influenza A virus (IAV) from wild waterfowl to poultry is the first step in a chain of events that can ultimately lead to exposure and infection of humans. Herein, we study the outcome of infection with eight different mallard-origin IAV subtypes in two different avian hosts: tufted ducks and chickens. We found that infection and shedding patterns as well as innate immune responses were highly dependent on viral subtypes, host species, and inoculation routes. For example, intraoesophageal inoculation, commonly used in mallard infection experiments, resulted in no infections in contrast to oculonasal inoculation, suggesting a difference in transmission routes. Despite H9N2 being endemic in chickens, inoculation of mallard-origin H9N2 failed to cause viable infection beyond 1 day postinfection in our study design. The innate immune responses were markedly different in chickens and tufted ducks, and despite the presence of retinoic acid-inducible gene-I (RIG-I) in tufted duck transcriptomes, it was neither up nor downregulated in response to infection. Overall, we have revealed the heterogeneity of infection patterns and responses in two markedly different avian hosts following a challenge with mallard-origin IAV. These virus-host interactions provide new insights into important aspects of interspecies transmission of IAV. IMPORTANCE Our current findings highlight important aspects of IAV infection in birds that have implications for our understanding of its zoonotic ecology. In contrast to mallards where the intestinal tract is the main site of IAV replication, chickens and tufted ducks show limited or no signs of intestinal infection suggesting that the fecal-oral transmission route might not apply to all bird IAV host species. Our results indicate that mallard-origin IAVs undergo genetic changes upon introduction into new hosts, suggesting rapid adaptation to a new environment. However, similar to the mallard, chickens and tufted ducks show a limited immune response to infection with low pathogenic avian influenza viruses. These findings and future studies in different IAV hosts are important for our understanding of barriers to IAV transmission between species and ultimately from the wild reservoir to humans.

Evaluating Effects of AIV Infection Status on Ducks Using a Flow Cytometry-Based Differential Blood Count.

Ducks have recently received a lot of attention from the research community due to their importance as natural reservoirs of avian influenza virus (AIV). Still, there is a lack of tools to efficiently determine the immune status of ducks. The purpose of this work was to develop an automated differential blood count for the mallard duck (Anas platyrhynchos), to assess reference values of white blood cell (WBC) counts in this species, and to apply the protocol in an AIV field study. We established a flow cytometry-based duck WBC differential based on a no-lyse no-wash single-step one-tube technique, applying a combination of newly generated monoclonal antibodies with available duck-specific as well as cross-reacting chicken markers. The blood cell count enables quantification of mallard thrombocytes, granulocytes, monocytes, B cells, CD4+ T cells (T helper) and CD8+ cytotoxic T cells. The technique is reproducible, accurate, and much faster than traditional evaluations of blood smears. Stabilization of blood samples enables analysis up to 1 week after sampling, thus allowing for evaluation of blood samples collected in the field. We used the new technique to investigate a possible influence of sex, age, and AIV infection status on WBC counts in wild mallards. We show that age has an effect on the WBC counts in mallards, as does sex in juvenile mallards. Interestingly, males naturally infected with low pathogenic AIV showed a reduction of lymphocytes (lymphocytopenia) and thrombocytes (thrombocytopenia), which are both common in influenza A infection in humans. IMPORTANCE Outbreaks of avian influenza in poultry and humans are a global public health concern. Aquatic birds are the primary natural reservoir of avian influenza viruses (AIVs), and strikingly, AIVs mainly cause asymptomatic or mild infection in these species. Hence, immunological studies in aquatic birds are important for investigating variation in disease outcome of different hosts to AIV and may aid in early recognition and a better understanding of zoonotic events. Unfortunately, immunological studies in these species were so far hampered by the lack of diagnostic tools. Here, we present a technique that enables high-throughput white blood cell (WBC) analysis in the mallard and report changes in WBC counts in wild mallards naturally infected with AIV. Our protocol permits large-scale immune status monitoring in a widespread wild and domesticated duck species and provides a tool to further investigate the immune response in an important reservoir host of zoonotic viruses.

The swan genome and transcriptome, it is not all black and white.

BACKGROUND: The Australian black swan (Cygnus atratus) is an iconic species with contrasting plumage to that of the closely related northern hemisphere white swans. The relative geographic isolation of the black swan may have resulted in a limited immune repertoire and increased susceptibility to infectious diseases, notably infectious diseases from which Australia has been largely shielded. Unlike mallard ducks and the mute swan (Cygnus olor), the black swan is extremely sensitive to highly pathogenic avian influenza. Understanding this susceptibility has been impaired by the absence of any available swan genome and transcriptome information. RESULTS: Here, we generate the first chromosome-length black and mute swan genomes annotated with transcriptome data, all using long-read based pipelines generated for vertebrate species. We use these genomes and transcriptomes to show that unlike other wild waterfowl, black swans lack an expanded immune gene repertoire, lack a key viral pattern-recognition receptor in endothelial cells and mount a poorly controlled inflammatory response to highly pathogenic avian influenza. We also implicate genetic differences in SLC45A2 gene in the iconic plumage of the black swan. CONCLUSION: Together, these data suggest that the immune system of the black swan is such that should any avian viral infection become established in its native habitat, the black swan would be in a significant peril.

Comparative Genomics of the Waterfowl Innate Immune System.

Animal species differ considerably in their ability to fight off infections. Finding the genetic basis of these differences is not easy, as the immune response is comprised of a complex network of proteins that interact with one another to defend the body against infection. Here, we used population- and comparative genomics to study the evolutionary forces acting on the innate immune system in natural hosts of the avian influenza virus (AIV). For this purpose, we used a combination of hybrid capture, next- generation sequencing and published genomes to examine genetic diversity, divergence, and signatures of selection in 127 innate immune genes at a micro- and macroevolutionary time scale in 26 species of waterfowl. We show across multiple immune pathways (AIV-, toll-like-, and RIG-I -like receptors signalling pathways) that genes involved genes in pathogen detection (i.e., toll-like receptors) and direct pathogen inhibition (i.e., antimicrobial peptides and interferon-stimulated genes), as well as host proteins targeted by viral antagonist proteins (i.e., mitochondrial antiviral-signaling protein, [MAVS]) are more likely to be polymorphic, genetically divergent, and under positive selection than other innate immune genes. Our results demonstrate that selective forces vary across innate immune signaling signalling pathways in waterfowl, and we present candidate genes that may contribute to differences in susceptibility and resistance to infectious diseases in wild birds, and that may be manipulated by viruses. Our findings improve our understanding of the interplay between host genetics and pathogens, and offer the opportunity for new insights into pathogenesis and potential drug targets.

Reference genomes for conservation.

High-quality reference genomes for non-model species can benefit conservation.

A high-quality genome and comparison of short- versus long-read transcriptome of the palaearctic duck Aythya fuligula (tufted duck).

BACKGROUND: The tufted duck is a non-model organism that experiences high mortality in highly pathogenic avian influenza outbreaks. It belongs to the same bird family (Anatidae) as the mallard, one of the best-studied natural hosts of low-pathogenic avian influenza viruses. Studies in non-model bird species are crucial to disentangle the role of the host response in avian influenza virus infection in the natural reservoir. Such endeavour requires a high-quality genome assembly and transcriptome. FINDINGS: This study presents the first high-quality, chromosome-level reference genome assembly of the tufted duck using the Vertebrate Genomes Project pipeline. We sequenced RNA (complementary DNA) from brain, ileum, lung, ovary, spleen, and testis using Illumina short-read and Pacific Biosciences long-read sequencing platforms, which were used for annotation. We found 34 autosomes plus Z and W sex chromosomes in the curated genome assembly, with 99.6% of the sequence assigned to chromosomes. Functional annotation revealed 14,099 protein-coding genes that generate 111,934 transcripts, which implies a mean of 7.9 isoforms per gene. We also identified 246 small RNA families. CONCLUSIONS: This annotated genome contributes to continuing research into the host response in avian influenza virus infections in a natural reservoir. Our findings from a comparison between short-read and long-read reference transcriptomics contribute to a deeper understanding of these competing options. In this study, both technologies complemented each other. We expect this annotation to be a foundation for further comparative and evolutionary genomic studies, including many waterfowl relatives with differing susceptibilities to avian influenza viruses.

Health monitoring in birds using bio-loggers and whole blood transcriptomics.

Monitoring and early detection of emerging infectious diseases in wild animals is of crucial global importance, yet reliable ways to measure immune status and responses are lacking for animals in the wild. Here we assess the usefulness of bio-loggers for detecting disease outbreaks in free-living birds and confirm detailed responses using leukocyte composition and large-scale transcriptomics. We simulated natural infections by viral and bacterial pathogens in captive mallards (Anas platyrhynchos), an important natural vector for avian influenza virus. We show that body temperature, heart rate and leukocyte composition change reliably during an acute phase immune response. Using genome-wide gene expression profiling of whole blood across time points we confirm that immunostimulants activate pathogen-specific gene regulatory networks. By reporting immune response related changes in physiological and behavioural traits that can be studied in free-ranging populations, we provide baseline information with importance to the global monitoring of zoonotic diseases.

Avian Immunome DB: an example of a user-friendly interface for extracting genetic information.

BACKGROUND: Genomic and genetic studies often require a target list of genes before conducting any hypothesis testing or experimental verification. With the ever-growing number of sequenced genomes and a variety of different annotation strategies, comes the potential for ambiguous gene symbols, making it cumbersome to capture the "correct" set of genes. In this article, we present and describe the Avian Immunome DB (AVIMM) for easy gene property extraction as exemplified by avian immune genes. The avian immune system is characterised by a cascade of complex biological processes underlaid by more than 1000 different genes. It is a vital trait to study particularly in birds considering that they are a significant driver in spreading zoonotic diseases. With the completion of phase II of the B10K ("Bird 10,000 Genomes") consortium's whole-genome sequencing effort, we have included 363 annotated bird genomes in addition to other publicly available bird genome data which serve as a valuable foundation for AVIMM. CONSTRUCTION AND CONTENT: A relational database with avian immune gene evidence from Gene Ontology, Ensembl, UniProt and the B10K consortium has been designed and set up. The foundation stone or the "seed" for the initial set of avian immune genes is based on the well-studied model organism chicken (Gallus gallus). Gene annotations, different transcript isoforms, nucleotide sequences and protein information, including amino acid sequences, are included. Ambiguous gene names (symbols) are resolved within the database and linked to their canonical gene symbol. AVIMM is supplemented by a command-line interface and a web front-end to query the database. UTILITY AND DISCUSSION: The internal mapping of unique gene symbol identifiers to canonical gene symbols allows for an ambiguous gene property search. The database is organised within core and feature tables, which makes it straightforward to extend for future purposes. The database design is ready to be applied to other taxa or biological processes. Currently, the database contains 1170 distinct avian immune genes with canonical gene symbols and 612 synonyms across 363 bird species. While the command-line interface readily integrates into bioinformatics pipelines, the intuitive web front-end with download functionality offers sophisticated search functionalities and tracks the origin for each record. AVIMM is publicly accessible at https://avimm.ab.mpg.de .

Contrasting effects of host species and phylogenetic diversity on the occurrence of HPAI H5N1 in European wild birds.

Studies on the highly pathogenic avian influenza (HPAI) H5N1 suggest that wild bird migration may facilitate its long-distance spread, yet the role of wild bird community composition in its transmission risk remains poorly understood. Furthermore, most studies on the diversity-disease relationship focused on host species diversity without considering hosts' phylogenetic relationships, which may lead to rejecting a species diversity effect when the community has host species that are only distantly related. Here, we explored the influence of waterbird community composition for determining HPAI H5N1 occurrence in wild birds in a continental-scale study across Europe. In particular, we tested the diversity-disease relationship using both host species diversity and host phylogenetic diversity. Our results provide the first demonstration that host community composition-compared with previously identified environmental risk factors-can also effectively explain the spatial pattern of H5N1 occurrence in wild birds. We further show that communities with more higher risk host species and more closely related species have a higher risk of H5N1 outbreaks. Thus, both host species diversity and community phylogenetic structure, in addition to environmental factors, jointly influence H5N1 occurrence. Our work not only extends the current theory on the diversity-disease relationship, but also has important implications for future monitoring of H5N1 and other HPAI subtypes.

Targeted resequencing of coding DNA sequences for SNP discovery in nonmodel species.

Targeted capture coupled with high-throughput sequencing can be used to gain information about nuclear sequence variation at hundreds to thousands of loci. Divergent reference capture makes use of molecular data of one species to enrich target loci in other (related) species. This is particularly valuable for nonmodel organisms, for which often no a priori knowledge exists regarding these loci. Here, we have used targeted capture to obtain data for 809 nuclear coding DNA sequences (CDS) in a nonmodel organism, the Eurasian lynx Lynx lynx, using baits designed with the help of the published genome of a related model organism (the domestic cat Felis catus). Using this approach, we were able to survey intraspecific variation at hundreds of nuclear loci in L. lynx across the species' European range. A large set of biallelic candidate SNPs was then evaluated using a high-throughput SNP genotyping platform (Fluidigm), which we then reduced to a final 96 SNP-panel based on assay performance and reliability; validation was carried out with 100 additional Eurasian lynx samples not included in the SNP discovery phase. The 96 SNP-panel developed from CDS performed very successfully in the identification of individuals and in population genetic structure inference (including the assignment of individuals to their source population). In keeping with recent studies, our results show that genic SNPs can be valuable for genetic monitoring of wildlife species.

Fast and cost-effective single nucleotide polymorphism (SNP) detection in the absence of a reference genome using semideep next-generation Random Amplicon Sequencing (RAMseq).

Biodiversity has suffered a dramatic global decline during the past decades, and monitoring tools are urgently needed providing data for the development and evaluation of conservation efforts both on a species and on a genetic level. However, in wild species, the assessment of genetic diversity is often hampered by the lack of suitable genetic markers. In this article, we present Random Amplicon Sequencing (RAMseq), a novel approach for fast and cost-effective detection of single nucleotide polymorphisms (SNPs) in nonmodel species by semideep sequencing of random amplicons. By applying RAMseq to the Eurasian otter (Lutra lutra), we identified 238 putative SNPs after quality filtering of all candidate loci and were able to validate 32 of 77 loci tested. In a second step, we evaluated the genotyping performance of these SNP loci in noninvasive samples, one of the most challenging genotyping applications, by comparing it with genotyping results of the same faecal samples at microsatellite markers. We compared (i) polymerase chain reaction (PCR) success rate, (ii) genotyping errors and (iii) Mendelian inheritance (population parameters). SNPs produced a significantly higher PCR success rate (75.5% vs. 65.1%) and lower mean allelic error rate (8.8% vs. 13.3%) than microsatellites, but showed a higher allelic dropout rate (29.7% vs. 19.8%). Genotyping results showed no deviations from Mendelian inheritance in any of the SNP loci. Hence, RAMseq appears to be a valuable tool for the detection of genetic markers in nonmodel species, which is a common challenge in conservation genetic studies.

A history of hybrids? Genomic patterns of introgression in the True Geese.

BACKGROUND: The impacts of hybridization on the process of speciation are manifold, leading to distinct patterns across the genome. Genetic differentiation accumulates in certain genomic regions, while divergence is hampered in other regions by homogenizing gene flow, resulting in a heterogeneous genomic landscape. A consequence of this heterogeneity is that genomes are mosaics of different gene histories that can be compared to unravel complex speciation and hybridization events. However, incomplete lineage sorting (often the outcome of rapid speciation) can result in similar patterns. New statistical techniques, such as the D-statistic and hybridization networks, can be applied to disentangle the contributions of hybridization and incomplete lineage sorting. We unravel patterns of hybridization and incomplete lineage sorting during and after the diversification of the True Geese (family Anatidae, tribe Anserini, genera Anser and Branta) using an exon-based hybridization network approach and taking advantage of discordant gene tree histories by re-sequencing all taxa of this clade. In addition, we determine the timing of introgression and reconstruct historical effective population sizes for all goose species to infer which demographic or biogeographic factors might explain the observed patterns of introgression. RESULTS: We find indications for ancient interspecific gene flow during the diversification of the True Geese and were able to pinpoint several putative hybridization events. Specifically, in the genus Branta, both the ancestor of the White-cheeked Geese (Hawaiian Goose, Canada Goose, Cackling Goose and Barnacle Goose) and the ancestor of the Brent Goose hybridized with Red-breasted Goose. One hybridization network suggests a hybrid origin for the Red-breasted Goose, but this scenario seems unlikely and it not supported by the D-statistic analysis. The complex, highly reticulated evolutionary history of the genus Anser hampered the estimation of ancient hybridization events by means of hybridization networks. The reconstruction of historical effective population sizes shows that most species showed a steady increase during the Pliocene and Pleistocene. These large effective population sizes might have facilitated contact between diverging goose species, resulting in the establishment of hybrid zones and consequent gene flow. CONCLUSIONS: Our analyses suggest that the evolutionary history of the True Geese is influenced by introgressive hybridization. The approach that we have used, based on genome-wide phylogenetic incongruence and network analyses, will be a useful procedure to reconstruct the complex evolutionary histories of many naturally hybridizing species groups.

The Evolution of Innate Immune Genes: Purifying and Balancing Selection on ß-Defensins in Waterfowl.

In disease dynamics, high immune gene diversity can confer a selective advantage to hosts in the face of a rapidly evolving and diverse pathogen fauna. This is supported empirically for genes involved in pathogen recognition and signalling. In contrast, effector genes involved in pathogen clearance may be more constrained. ß-Defensins are innate immune effector genes; their main mode of action is via disruption of microbial membranes. Here, five ß-defensin genes were characterized in mallards (Anas platyrhynchos) and other waterfowl; key reservoir species for many zoonotic diseases. All five genes showed remarkably low diversity at the individual-, population-, and species-level. Furthermore, there was widespread sharing of identical alleles across species divides. Thus, specific ß-defensin alleles were maintained not only spatially but also over long temporal scales, with many amino acid residues being fixed across all species investigated. Purifying selection to maintain individual, highly efficacious alleles was the primary evolutionary driver of these genes in waterfowl. However, we also found evidence for balancing selection acting on the most recently duplicated ß-defensin gene (AvBD3b). For this gene, we found that amino acid replacements were more likely to be radical changes, suggesting that duplication of ß-defensin genes allows exploration of wider functional space. Structural conservation to maintain function appears to be crucial for avian ß-defensin effector molecules, resulting in low tolerance for new allelic variants. This contrasts with other types of innate immune genes, such as receptor and signalling molecules, where balancing selection to maintain allelic diversity has been shown to be a strong evolutionary force.

A tree of geese: A phylogenomic perspective on the evolutionary history of True Geese.

Phylogenetic incongruence can be caused by analytical shortcomings or can be the result of biological processes, such as hybridization, incomplete lineage sorting and gene duplication. Differentiation between these causes of incongruence is essential to unravel complex speciation and diversification events. The phylogeny of the True Geese (tribe Anserini, Anatidae, Anseriformes) was, until now, contentious, i.e., the phylogenetic relationships and the timing of divergence between the different goose species could not be fully resolved. We sequenced nineteen goose genomes (representing seventeen species of which three subspecies of the Brent Goose, Branta bernicla) and used an exon-based phylogenomic approach (41,736 exons, representing 5887 genes) to unravel the evolutionary history of this bird group. We thereby provide general guidance on the combination of whole genome evolutionary analyses and analytical tools for such cases where previous attempts to resolve the phylogenetic history of several taxa could not be unravelled. Identical topologies were obtained using either a concatenation (based upon an alignment of 6,630,626 base pairs) or a coalescent-based consensus method. Two major lineages, corresponding to the genera Anser and Branta, were strongly supported. Within the Branta lineage, the White-cheeked Geese form a well-supported sub-lineage that is sister to the Red-breasted Goose (Branta ruficollis). In addition, two main clades of Anser species could be identified, the White Geese and the Grey Geese. The results from the consensus method suggest that the diversification of the genus Anser is heavily influenced by rapid speciation and by hybridization, which may explain the failure of previous studies to resolve the phylogenetic relationships within this genus. The majority of speciation events took place in the late Pliocene and early Pleistocene (between 4 and 2millionyears ago), conceivably driven by a global cooling trend that led to the establishment of a circumpolar tundra belt and the emergence of temperate grasslands. Our approach will be a fruitful strategy for resolving many other complex evolutionary histories at the level of genera, species, and subspecies.

Genetic isolation between coastal and fishery-impacted, offshore bottlenose dolphin (Tursiops spp.) populations.

The identification of species and population boundaries is important in both evolutionary and conservation biology. In recent years, new population genetic and computational methods for estimating population parameters and testing hypotheses in a quantitative manner have emerged. Using a Bayesian framework and a quantitative model-testing approach, we evaluated the species status and genetic connectedness of bottlenose dolphin (Tursiops spp.) populations off remote northwestern Australia, with a focus on pelagic 'offshore' dolphins subject to incidental capture in a trawl fishery. We analysed 71 dolphin samples from three sites beyond the 50 m depth contour (the inshore boundary of the fishery) and up to 170 km offshore, including incidentally caught and free-ranging individuals associating with trawl vessels, and 273 dolphins sampled at 12 coastal sites inshore of the 50 m depth contour and within 10 km of the coast. Results from 19 nuclear microsatellite markers showed significant population structure between dolphins from within the fishery and coastal sites, but also among dolphins from coastal sites, identifying three coastal populations. Moreover, we found no current or historic gene flow into the offshore population in the region of the fishery, indicating a complete lack of recruitment from coastal sites. Mitochondrial DNA corroborated our findings of genetic isolation between dolphins from the offshore population and coastal sites. Most offshore individuals formed a monophyletic clade with common bottlenose dolphins (T. truncatus), while all 273 individuals sampled coastally formed a well-supported clade of Indo-Pacific bottlenose dolphins (T. aduncus). By including a quantitative modelling approach, our study explicitly took evolutionary processes into account for informing the conservation and management of protected species. As such, it may serve as a template for other, similarly inaccessible study populations.

The Value of Molecular vs. Morphometric and Acoustic Information for Species Identification Using Sympatric Molossid Bats.

A fundamental condition for any work with free-ranging animals is correct species identification. However, in case of bats, information on local species assemblies is frequently limited especially in regions with high biodiversity such as the Neotropics. The bat genus Molossus is a typical example of this, with morphologically similar species often occurring in sympatry. We used a multi-method approach based on molecular, morphometric and acoustic information collected from 962 individuals of Molossus bondae, M. coibensis, and M. molossus captured in Panama. We distinguished M. bondae based on size and pelage coloration. We identified two robust species clusters composed of M. molossus and M. coibensis based on 18 microsatellite markers but also on a more stringently determined set of four markers. Phylogenetic reconstructions using the mitochondrial gene co1 (DNA barcode) were used to diagnose these microsatellite clusters as M. molossus and M. coibensis. To differentiate species, morphological information was only reliable when forearm length and body mass were combined in a linear discriminant function (95.9% correctly identified individuals). When looking in more detail at M. molossus and M. coibensis, only four out of 13 wing parameters were informative for species differentiation, with M. coibensis showing lower values for hand wing area and hand wing length and higher values for wing loading. Acoustic recordings after release required categorization of calls into types, yielding only two informative subsets: approach calls and two-toned search calls. Our data emphasizes the importance of combining morphological traits and independent genetic data to inform the best choice and combination of discriminatory information used in the field. Because parameters can vary geographically, the multi-method approach may need to be adjusted to local species assemblies and populations to be entirely informative.

Population-level consequences of complementary sex determination in a solitary parasitoid.

BACKGROUND: Sex determination mechanisms are known to be evolutionarily labile but the factors driving transitions in sex determination mechanisms are poorly understood. All insects of the Hymenoptera are haplodiploid, with males normally developing from unfertilized haploid eggs. Under complementary sex determination (CSD), diploid males can be produced from fertilized eggs that are homozygous at the sex locus. Diploid males have near-zero fitness and thus represent a genetic load, which is especially severe under inbreeding. Here, we study mating structure and sex determination in the parasitoid Cotesia vestalis to investigate what may have driven the evolution of two complementary sex determination loci in this species. RESULTS: We genotyped Cotesia vestalis females collected from eight fields in four townships in Western Taiwan. 98 SNP markers were developed by aligning Illumina sequence reads of pooled DNA of eight different females against a de novo assembled genome of C. vestalis. This proved to be an efficient method for this non-model species and provides a resource for future use in related species. We found significant genetic differentiation within the sampled population but variation could not be attributed to sampling locations by AMOVA. Non-random mating was detected, with 8.1% of matings between siblings. Diploid males, detected by flow cytometry, were produced at a rate of 1.4% among diploids. CONCLUSIONS: We think that the low rate of diploid male production is best explained by a CSD system with two independent sex loci, supporting laboratory findings on the same species. Fitness costs of diploid males in C. vestalis are high because diploid males can mate with females and produce infertile triploid offspring. This severe fitness cost of diploid males combined with non-random mating may have resulted in evolution from single locus CSD to CSD with two independent loci.