Mosquito iDNA reveals landscape patterns of birds and mammals

Changes in fire regimes and the introduction of invasive species are negatively impacting Northern Australia's biota, highlighting the need for reliable biodiversity monitoring data. Here we used mosquito‐ingested DNA (iDNA) to detect birds and mammals within Kakadu National Park. Six sites were sampled on two occasions (wet and dry season), with three traps deployed for two consecutive nights of sampling at each site. Mosquitoes were processed in bulk samples per trap ranging from 31 to 1987 mosquitoes. Extracted DNA was amplified using primers targeting vertebrates (12SV5), mammals (16SMam1/2) and birds (Aves01) with iDNA‐derived richness and composition measured. A total of 50 avian and 19 mammalian taxa were detected, including threatened species such as the white‐throated grasswren (Amytornis woodwardi) and the ghost bat (Macroderma gigas), along with difficult‐to‐monitor groups such as small‐bodied and arboreal mammals. Community composition was shown to be distinct between sampling events, and avian richness was higher during the dry season. Importantly for the management of the park, our findings suggest that fire regimes modified both bird and mammal communities during the dry season, with vegetation structure being correlated with avian and mammalian community composition during the wet season. This study highlights the capacity for iDNA metabarcoding to provide fine‐scale ecologically relevant information within a large, dynamic and difficult‐to‐access landscape

Methods
Bulk mosquito samples were collected from six 1ha sites across two field campaigns undertaken during a wet and dry season. At each site three EVS traps were set up and samples were collected for two consecutive nights. A total of 71 bulk samples were collected and processed using a salting out procedure that could be adapted for the variability in the size of the bulk samples. Samples were PCR'd using three primers (12SV5, 16SMam1/2 and Aves01), with all PCRs performed in triplicates. PCR product was cleaned using a Minielute kit and the three primers were pooled equamolar. Before being sent to the Ramaciotti centre for genomics where they were sequenced on a Novaseq 6000 sequencing run. Raw sequencing data was firstly processed using OBITools, with the steps briefly including reassembling paired ended reads, removal of poor quality reads, assignment of reads to their relvant primer and sample, dereplication of reads and initial taxonomic assignments. Following processing through OBITools samples were clustered into MOTUs using SUMACLUST applying a 97% similarity threshold. Data was then further cleaned using MetabaR. The three primers were then merged to form an avian and mammalian dataset.