The genetic consequences of the subdivision of populations are regarded as significant to long-term evolution, and research has shown that the scale and speed at which this is now occurring is critically reducing the adaptive potential of most species which inhabit human-impacted landscapes. Here, we provide a rare, and to our knowledge, the first analysis of this process while it is happening and demonstrate a method of evaluating the effect of mitigation measures such as fauna crossings. We did this by using an extensive genetic dataset collected from a koala population which was intensely monitored during the construction of linear transport infrastructure which resulted in the subdivision of their population. First, we found that both allelic richness and effective population size decreased through the process of population subdivision. Second, we predicted the extent to which genetic drift could impact genetic diversity over time and showed that after only 10 generations the resulting two subdivided populations could experience between 12–69% loss in genetic diversity. Lastly, using forward simulations we estimated that a minimum of 8 koalas would need to disperse from each side of the subdivision per generation to maintain genetic connectivity close to zero but that 16 koalas would ensure that both genetic connectivity and diversity remained unchanged. These results have important consequences for the genetic management of species in human-impacted landscapes by showing which genetic metrics are best to identify immediate loss in genetic diversity and how to evaluate the effectiveness of any mitigation measures.

The study koala population.

We were able to ask these questions because of the rareness of the dataset which was obtained during and after construction of a linear transport infrastructure (e.i. rail line) which resulted in the subdivision of a large population of koala. The dataset was collected as part of an extensive Koala Management Program (Beyer et al. 2018), in which all free-ranging koalas (Phascolarctos cinereus; n = 503 koalas) were captured and healthy animals released and monitored between 2013 and 2017 for the purpose of a rail infrastructure project located in the eastern Moreton Bay Region (-27.234°; 153.036°, Queensland, Australia). These koalas formed part of a single breeding population (Schultz et al. 2022) inhabiting a mixture of urban and peri-urban koala habitat remnants along 12.6 kms of the linear transport infrastructure project footprint (Hanger et al. 2017). Koala-proof fencing was installed alongside the rail line corridor to prevent koala death from crossing attempts, and underpasses were built at strategic locations along the rail line (Hanger et al. 2017). Because all koalas were VHF and/or GPS tracked during all phases of construction (pre, during and post), it afforded us detailed knowledge of how and why the impacted koala population’s size varied throughout the rail line construction phases: (1) all deaths and births and causes of death (Hanger et al. 2017; Beyer et al. 2018), (2) which koalas were translocated as a result of the rail line infrastructure project, (3) their locations pre and post-construction (whether koalas occupied habitat on one side or the other of the rail line), and (4) rail crossing events by koalas after establishment of the rail barrier using both dedicated fauna crossings and hydrology culverts (Dexter et al. 2017). In total, 291 koalas out of the 461 processed (57.8% population size decline) died or were euthanised during the four-year monitoring program: 182 from predation, 84 from disease, 14 from trauma, and 11 unknown causes of death (Beyer et al. 2018). Twenty-eight were translocated as a result of their core home range directly overlapping with the rail line and/or because current land-use and imminent future land-use precluded their long-term viability if left in situ.

Full protocols are available in the technical report by Hanger et al. (2017), and scientific permits and ethics approvals for catching, handling, veterinary examination and treatment, and monitoring of koalas as follows: scientific research permits issued by Queensland Department of Environment and Heritage Protection WISP-11525212, WISP-16125415, WISP-13661313, WITK-14173714, WISP-17273716; animal ethics approvals from Queensland Department of Agriculture and Fisheries CA-2012/03/597, CA-2013/09/719, CA-2014/06/777, CA-2015/03/852 and CA-2016/03/950.

Koala subsampling, genotyping and final genetic dataset.

Single nucleotide polymorphism (SNP) genotyping data were generated from blood or tissue samples, collected during routine veterinary examinations from 452 of the 503 monitored individual koalas for which blood or tissue samples remained. Blood samples were stored at -20 °C, and tissue samples were stored in 70% ethanol. DNA was extracted using the DNeasy Blood and Tissue Kit (QIAGEN), following the manufacturer’s protocol, and DNA extracts were stored at -80° C. SNP genotyping was conducted as in Kjeldsen et al. (Kjeldsen et al. 2019) by Diversity Arrays Technology, Canberra, using their proprietary DArTseq^TM technology. DArTseq^TM utilises a combination of next-generation sequencing platforms and DArT complexity-reduction methods (Kilian et al. 2012; Courtois et al. 2013; Cruz et al. 2013; Raman et al. 2014). The protocol is also optimised for organism and application by selecting the most appropriate complexity reduction method. This is assessed based on minimising skewed size ranges, non-ideal numbers of fragments, and percentages of repetitive elements. Samples were then processed as per Kilian et al. (2012). SNP genotyping produced a total of 8649 SNPs. We then filtered those SNPs for minor allele frequency ≥ 0.05, 90% individual call-rate, ≥ 99% technical replication average which resulted in 3655 SNPs from 367 koalas in total. We further removed 42 koalas which, while included in the Koala Management Programme, where found in areas not contiguous with the main population, and subsequently used as translocation sites for koalas whose home ranges overlapped the rail infrastructure footprint. Finally, to increase the accuracy of our simulations (see below), we removed individual koalas with any missing data in their genotypes, resulting in 270 individuals.

To isolate the effect of population subdivision caused by the construction of the linear infrastructure from mortality events caused by predation, disease, trauma from roads and fighting and unknown causes, we organised the koala genotypes into three genetic datasets. As our point of reference, we created Dataset 1 (n = 270) which contains all the genotypes from the monitored koalas collected during the 4 years. To account for death by predation, disease, trauma and unknown events, we created Dataset 2 (n = 114) which contains all genotyped koalas from dataset 1 minus koalas which died of from predation, disease, trauma and unknown causes. Comparing changes in the genetic composition of dataset 1 and 2 allows us to understand the impact of population decline due to mortality events caused by predation, disease, trauma and unknown causes. To then estimate the immediate and longer-term effect of the linear infrastructure on our studied koala population, we created Dataset 3 which contains all genotyped koalas from dataset 2 minus those that were translocated because their core home range sat in the linear transport infrastructure footprint and any new koalas born after the linear transport infrastructure was built. Koalas in dataset 3 were then divided into two to represent their locations above (n = 27) and below (n =75) the linear transport infrastructure. Comparing changes in the genetic composition of dataset 2 and 3 (above and below) allowed us to isolate the immediate and predict the longer-term genetic consequences of the linear transport infrastructure project (i.e. population size reduction by translocation and population subdivision). This left 102 koalas that were previously genetically connected, then became subdivided on either side of the rail-line.