Biogeographic barriers can restrict gene flow, but variation in ecological drivers of dispersal influences the effectiveness of these barriers among different species. Detailed information about the genetic connectivity and movement of waterfowl across biogeographic barriers in northern Australia and Papua New Guinea is limited. We compared genetic connectivity for four species of Australasian waterfowl that vary in their capacity and predisposition for dispersal: Radjah Shelduck (Tadorna radjah), Wandering Whistling-Duck (Dendrocygna arcuata), Green Pygmy-Goose (Nettapus pulchellus), and Pacific Black Duck (Anas superciliosa). We obtained >2,700 loci from double-digest restriction-associated DNA sequencing for 15 to 40 individuals per species and found idiosyncratic patterns of population structure among the four species. The mostly sedentary Radjah Shelduck exhibited clear genetic differences between New Guinea and Australia as well as among locations within Australia. In contrast, the presumed sedentary Green Pygmy-Goose did not show obvious structure. Likewise, populations of the more dispersive Wandering Whistling Duck and Pacific Black Duck were unstructured and genetically indistinguishable between southern New Guinea and northern Australia. Our data suggest some Australo-Papuan biogeographical barriers are insufficient to impede gene flow in waterfowl species capable of dispersing great distances. In sedentary species like the Radjah Shelduck, these barriers, perhaps coupled with its ecology and natural history, restrict gene flow. Our findings bring new insight into the population ecology of Australo-Papuan waterfowl.

Double-digest restriction-site-associated DNA sequencing (ddRAD-seq) was used to sample loci from across the genome using the protocol of DaCosta and Sorenson (2014. PLoS ONE 9(9): e106713; https://doi.org/10.1371/journal.pone.0106713). Restriction enzymes SBfI and EcoRI were used to fragment the genome, and adapters containing sequences compatible with Illumina TruSeq reagents and barcodes were ligated to the sticky ends. Fragments were size-selected (300 – 450 bp) using gel electrophoresis and amplified using polymerase chain reaction (PCR). Purified PCR products were pooled in equimolar concentrations and sequenced on an Illumina HiSeq 2500.

Raw Illumina reads were demultiplexed and assembled using the pipeline of DaCosta and Sorenson (2014) [Scripts available at: http://github.com/BU-RAD-seq/ddRAD-seq-Pipeline]. Individual reads with an average Phred score of <20 were removed. Reads were clustered into putative loci using the UCLUST function in USEARCH v. 5 with an –id setting of 0.85, and aligned using MUSCLE V. 3. Homozygous genotypes were defined if greater than 93% of sequence reads were consistent with a single haplotype, whereas heterozygotes were defined if a second haplotype was represented by at least 29% of reads, or if a second haplotype was represented by as few as 10% of reads and the haplotype was present in other individuals. For individual genotypes that did not meet either criterion or contained more than two haplotypes, we retained the allele represented by the majority of reads and scored the second as missing data. The second allele was also scored as missing for genotypes based on fewer than 10 reads. We retained all loci with ≤10% missing genotypes across samples and ≤5% flagged genotypes. We categorized ddRAD-seq loci as either autosomal or sex-linked on the basis of alignments to the Mallard genome.

*.stru -- STRUCTURE v.2.3.4 or the R-script adegent

*spegedi.txt -- Spagedi v1.5