The Common or European Starling, Sturnus vulgaris (hereafter starling), is considered one of the most successful invasive avian species globally, with invasive populations on all human-populated continents [1]. The establishment of invasive starling populations is the result of intentional human-mediated introductions that occurred in the mid to late 19th century [2]. Acclimatisation societies were responsible for the majority of these introductions. These societies were groups specifically formed to introduce plant and animal species across the colonies, aiming to transform these ‘new’ environments into something more reminiscent of ‘home’, or to control agricultural or horticultural pests [3]. Unfortunately, in many cases, they were ignorant of the ecological consequences of their actions. Repeated introductions of starlings to multiple locations around Aotearoa occurred between 1862 and 1883 to control insect grain pests, contributing to the present-day invasive population [4].

Successful invasions like those of the starling may appear paradoxical. Despite the often-small size of the introduced population and hence limited genetic diversity to facilitate adaptation, invasive species do not only become established but can flourish under novel environmental conditions. What enables certain invasive species to adapt and thrive in novel environments remains an unresolved question within the context of invasion biology [5].

Many different competing ecological hypotheses exist concerning successful invasives. Examples include the exploitation of vacant niches associated with human activity or the competitive exclusion of native species [5]. An important component of addressing these questions is to understand which historical introduction sites were the most successful and why, and which of these sites are facilitating ongoing invasive range expansion. Genetic data can help to inform an understanding of this invasion history, but despite biological invasions continuing to escalate at an alarming rate, population genomic data are notably lacking for many of the most pervasive and invasive species around the world [6]. Such studies will be crucial to enable evidence-based decision-making if we are to predict, prevent, or manage biological invasions successfully.

We used genome-wide DNA sequencing data to produce the first high-resolution genomic dataset of starlings in Aotearoa to characterise and examine their genetic diversity and population structure. We found a relatively high level of genetic differentiation and population substructuring for samples taken from across the Auckland region, which is congruent with documented introduction histories [7] while also implying restricted gene flow with other regions. Other locations presented with more variable genetic population structure, suggesting potential connectivity between regions south of Auckland. This first insight into the interplay between the biotic and abiotic factors that are shaping invasive starling populations in Aotearoa could potentially inform proactive control strategies.

Methods

Sample Collection, DNA Extraction and Sequencing

106 starling specimen samples were obtained from various sources within Aotearoa from five geographically distinct regions (Figure 1). These consisted of three locations in the North Island: Auckland (AUK: n=18), Palmerston North (PLM: n=12), Upper Hutt (UHT: n=40), and two in the South Island: Blenheim (BLN: n=15) and Canterbury (CAN: n=21). Muscle tissue was subsampled from each individual, and DNA was extracted from these tissue samples and sent to Diversity Arrays Technology Pty Ltd (DArT P/L) for processing and genome-wide sequencing [8]. The DArT method sequences DNA from a random sample of the genome, a method termed ‘reduced representation sequencing’, to capture a representative snapshot of genome-wide diversity.

Data Processing, Mapping, Variant Calling and Filtering

Once we received the raw DNA sequence outputs from DArT, a series of software programs and pipelines were used to process these raw outputs. We identified a list of variable sites in the genome where there was a single DNA base change within a sequence (single nucleotide polymorphisms, or ‘SNPs’). SNPs capture diversity between individuals and between sampling locations, and SNP variants can be used to cluster individuals by their similarity and determine which sampling sites are genetically dissimilar to each other. In brief, we used Fastp v0.23.2 [9] to trim DNA sequence lengths to 70 bases and discard all sequences shorter than 40 bases. We used the Stacks v2.2 [10] pipeline to discard low-quality sequences, remove sequences with uncalled bases, and also remove the restriction enzyme recognition sites (Pstl & Sphl). Next, we used BWA v0.7.17 [11] to index our reference genome S. vulgaris vAU1.0 [12], align the sequences, and produce output files containing the alignments and their respective base qualities. Alignments were sorted and indexed using SAMtools v1.16.1 [13], and SNPs were called and annotated using BCFtools v1.16 [14]. We then systematically removed replicates and siblings from the dataset using DartR v2.9.7 [15], resulting in a final individual count of 76 (N=76). Finally, we used VCFtools v0.1.15 [16] to filter out poor-quality SNPs and remove SNPs present in less than 50% of the individuals in each sampling location. We ran one final filtering step to ensure low levels of missing data (due to technical errors during sequencing or issues sequencing particular segments of DNA), resulting in a final dataset containing 29,117 unique SNP sites.