Introduction

Wildlife diseases are a significant contributor to population declines and can lead to local and global extinctions1,2,3,4. Wildlife diseases can drive biodiversity loss, adversely impact economies and pose a threat to human and livestock health5,6,7,8. Effective disease surveillance, management and conservation strategies require accurate estimates of pathogen prevalence in wild populations9. Yet pathogen detection in wild animals faces many challenges, such as those associated with the logistics and costs of capture-based field sampling, and the increasing need for non-invasive methods, as animal welfare concerns increase10,11,12.

Scat sampling has been proposed as a health-monitoring methodology that can be (1) deployed at a large-scale, (2) cost effective, and (3) less invasive compared to direct sampling of wildlife13,14,15,16. Scat surveys may also reduce biases associated with capture and sighting probability, and those associated with “convenience sampling”, which relies on samples from wildlife hospitals or carcasses17. The efficiency of scat surveys can further be increased when coupled with the use of detection dogs, where trained sniffer dogs are used to detect the presence of scats18. In some species, such as the koala, this method can be 150% more accurate and 20 times quicker than human searches18.

The koala (Phascolarctos cinereus) is an iconic Australian marsupial and a key piece of the Australian tourism industry19. Koalas also serve as a flagship species, aiding in the protection of habitat for other species20. Recently, koala populations in Queensland, New South Wales and the Australian Capital Territory were listed as ‘Endangered’ under the Environment Protection and Biodiversity Conservation Act (EPBC Act 1999). Disease, especially Chlamydia infection, is an important contributor to the decline of koala populations. Chlamydia infection is currently recognised as a top cause for koala admission to veterinary hospitals, and as the illness with the highest morbidity and mortality21,22,23. Most prevalent is Chlamydia pecorum, which can cause blindness, urinary tract disease, and infertility in both male and female koalas, driving population declines24,25. Visual signs of C. pecorum infections range from incontinence and associated staining of the fur around the cloaca, to erythematous and proliferative conjunctivitis and keratitis25. However, observing these signs in an arboreal mammal in the wild is difficult and many infected animals do not present clinically observable signs26; for example, disease of the reproductive tract is not visible externally. Consequently, measures obtained from visual assessments may not be accurate. To date, detection of Chlamydia by species-specific quantitative PCR (qPCR) represents the gold standard laboratory diagnosis, relying on the use of ocular and urogenital swab samples27. However, collection of such samples requires physical handling of the animals, which is not practical for large-scale monitoring of wild koalas, as well as being time consuming and requiring specialised equipment and staff. Recent methods using DNA isolated from non-invasive scat collections may offer an alternative that addresses these constraints28,29. Scat samples have been used to screen for the presence of viruses30,31 and pathogenic bacteria29,32,33,34, and are increasingly used to monitor health and disease in wildlife12,29.

With the advancement of genomic sequencing technologies, the development of molecular markers, such as single nucleotide polymorphisms (SNPs), has offered a cost-effective approach for genetic screening. SNP genotyping is increasingly being used to monitor wild populations using non-invasive samples such as scats35,36,37,38,39,40. The use of SNPs may be preferable to the use of microsatellites when working with lower-quality non-invasive samples, as SNPs are more abundant and dispersed throughout the genome - they produce fewer genotyping errors, and their small size may allow for a greater amplification success when samples are somewhat degraded41,42.

Technologies such as reduced-representation whole-genome genotyping using DArTseq (Sequencing-based Diversity Array Technology) enable large-scale discovery of sequence variation without the need for previously established genomics resources43. DArTseq uses restriction enzymes to fragment DNA samples, producing highly reduced portions of the genome, and combines these with next-generation sequencing platforms44,45,46,47. This technology has been successfully applied to DNA in koala scat samples to detect the presence of C. pecorum29. However, while the specificity (i.e., the pathogen is not detected when absent) of C. pecorum detection from scat using DArTseq was 100%, the sensitivity (i.e., the pathogen is detected when present) was low, ranging between 38 and 50% for urogenital infections29.

Recently, restriction-enzyme associated methods have been paired with targeted sequence-capture protocols, combining relatively inexpensive and fast library preparation with sequence enrichment to detect loci of particular interest48,49. DArTcap combines elements of the DArTseq protocol with sequence capture probes to target loci of interest from an existing DArTseq library. This approach was designed to increase sensitivity and specificity with higher on-target sequencing coverage levels50. To our knowledge,few studies have attempted targeted-sequence-capture methods to identify the presence of pathogens using non-invasive samples from wild animals51,52. Implementing markers of infection alongside host genetic markers and other specific markers of interest (e.g. diet) into one panel, thus creating a one-stop-shop for scat analyses, rather than multiple detached processes, would considerably increase efficiency and decrease processing times and costs.

Here, we adapted and employed DArTcap to detect C. pecorum infection using DNA extracted from koala scats. We assessed its performance, in terms of sensitivity and specificity, by comparing DArTcap outcomes to those obtained by the gold standard method, qPCR assay on high quality samples (i.e., urogenital swabs), and to qPCR assay directly from scat samples.

Methods

Sample collection

This research was conducted under a current animal ethics permit, approved by the Animal Ethics Committee at the University of the Sunshine Coast, QLD, Australia. All activities were performed in accordance with the permit guidelines and regulations. Permit number: ANA20162. The authors also complied with the ARRIVE guidelines.

Sample collection was conducted following the same procedure as described in Cristescu, et al.29 by trained veterinary personnel. Sampled individuals were wild koalas from two monitored populations. The first population was monitored to mitigate the impacts of the construction of the Section D of the Bruce Highway, Southeast Queensland. The second population is located in Redland City Council, and was monitored as part of the local koala management plan. Wild koalas were caught and transported to the Endeavour Veterinary Ecology clinic, where they underwent a veterinary examination, and samples were taken. A total of 28 sample sets were collected from 23 koalas, each set consisting of urogenital swabs and scats collected within the same day for each koala. Some koalas (n = 4) were sampled more than once, in which case sample sets for the same koala were at least 13 days apart (min = 13, max = 82, Table S1). Veterinary personnel collected 13 sample sets from sick koalas (i.e. those showing obvious signs of C. pecorum infection) and 15 sample sets from healthy koalas (i.e. those not showing any signs of C. pecorum infection). While veterinary examinations were used for the preliminary assessment of koala disease status, qPCR results from urogenital swab samples constituted the reference for the Chlamydia status of samples (i.e., the investigated C. pecorum detection methods were compared to results from qPCR of swab samples). Urogenital swab samples were collected during veterinary examinations under general anaesthesia. Koalas were anaesthetised with alfaxalone 10 mg/ml (Alfaxan CD-RTUR, Jurox Pty Ltd), injected intramuscularly into the quadriceps muscle at a dose rate of 3-5 mg/kg. Anaesthesia was maintained as required with additional doses of Alfaxan injected either intramuscularly or intravenously, or inhalation of a combination of isoflurane (Isoflo™, AbboR) and medical oxygen. Fresh scats were also collected either during veterinary examination or from the transport cage and stored in individual 50 mL tubes. Swabs and scats were frozen at -18 °C until DNA was extracted.

DNA isolation

Genomic DNA extractions from swab samples were conducted at the Koala Health Hub (KHH, University of Sydney). Briefly, DNA from swabs was extracted using a MagMAX CORE Nucleic Acid Purification Kit (Thermo Fisher cat# A32702; Thermo Fisher Scientific, Waltham, MA, USA) by shaving them into a 1.5 mL tube containing 350 µL of MagMAX CORE Lysis Solution and 10 µL of Proteinase K, and incubating the tubes at 56 °C for 1 h. The lysate was then added to a 96DW-plate containing 350 µL of MagMAX CORE Binding Solution and 20 µL of MagMAX CORE Magnetic Beads, then immediately processed on a KingFisher™ Flex automated extraction instrument, using the MagMax_Core_Flex protocol. DNA was eluted to a final volume of 100 µL.

DNA from scat samples was extracted following two methods at two different laboratories. First, at KHH, DNA extraction was carried out using the ISOLATE Fecal DNA Kit (Meridian Bioscience®, Australia cat # BIO-52082) according to the manufacturer’s instructions except for the following modifications as described in Wright, et al.53: Lysis buffer (700 µL) and approximately 200 mg of scat from the collection tube were added to a Bashing Bead Lysis Tube and homogenized at 6 m/s for 30 s using a FastPrep®-24 instrument. Prior to samples being loaded onto a Spin II A Filter, an added centrifugation step (10 000 x g, 3 min) was carried out to pellet colloidal material using a 1.5 mL micro-centrifuge tube. DNA was eluted to a final volume of 100 µL and stored at -80 °C.

Second, at the University of the Sunshine Coast (UniSC), DNA extraction from koala scat was conducted using the QIAamp PowerFecal Pro DNA Kit (Qiagen), following the manufacturer’s protocol, and the method described by Schultz, et al.54 with the following variations. The outermost layer of the scats was sliced off to collect the epithelial cells from the scat surface. After adding CD1 buffer, samples were incubated at 65 °C for 10 min, and then vortexed for seven minutes at maximum speed using Genie 2 Vortex Mixer (Scientific Industries). Final DNA isolates were eluted in 100 µl of C6 elution buffer. Each isolate was tested for DNA isolation quality on a 1.5% agarose gel. DNA isolates that passed this quality control were then stored at -80 °C until sent for genotyping.

Diagnostic qPCR assay for C. pecorum

Diagnostic qPCR assays were conducted for all swab samples and both DNA extractions from scat samples (i.e. KHH and UniSC extractions, explained above) at KHH. Presence of Chlamydia in swabs and scat DNA was determined by C. pecorum species-specific qPCR assay, a protocol practiced routinely at the KHH laboratories. Briefly, real-time PCR was conducted using SensiFAST™ Probe No-ROX Kit (Meridian Bioscience®, Australia) on a CFX96 Touch™ Real-Time PCR Detection System with the corresponding CFX Maestro software (BioRad, Australia) and 2 µL of template DNA. The multiplex qPCR reaction at a final volume of 20 µL included primers and probes targeting C. pecorum (ompB gene), Chlamydia genus (23 S rRNA) and koala β-actin reference gene. All primers and probes described in Hulse et al.55 were used at a final concentration of 400 nM and 200 nM respectively. PCR conditions were 3 min at 95 °C, followed by 40 cycles of 10 s at 95 °C and 40 s at 58 °C. The limit of detection (LOD) of this multiplex assay is 86 copies of C. pecorum target per reaction (2 µL of template). The LOD of this assay was determined using Probit analysis (95%) as described by Pum61 and known concentrations of a pUCIDT-AMP vector (Integrated DNA Technologies, USA) containing an insert of the target sequence and flanking regions. All DNA samples were assayed neat and in a 10-fold dilution to check for changes in qPCR efficiency due to the presence of inhibitors in the sample.

Targeted genotyping using DArTcap

DNA isolated from scat samples following the UniSC extraction method (described above) were submitted to Diversity Arrays Technology Pty Ltd (Canberra, Australia) to detect C. pecorum presence using DArTcap genotyping technology. Initial C. pecorum markers were identified using the DArTseq approach29, and a total of 55 specific ‘capture probes’ were designed to bind to sequences expected in the DArTseq representations when C. pecorum is present.

Data analyses

Results from qPCR were classified as ‘Negative’ or ‘Positive’ for C. pecorum based on the presence or absence of qPCR amplification of the C. pecorum target gene, within the assay’s LOD. Each sample was considered adequate in host DNA quantities and absent of significant inhibition if the sample was positive (Ct ≤ 32) for koala β-actin and its corresponding 10-fold dilution was 3.3 qPCR cycles apart; otherwise, it was marked as failed for qPCR quality control (QC-failed). For DArTcap, the presence of the C. pecorum pathogen was assessed based on the detection of the expected C. pecorum-specific sequences. The total read depth of the expected sequences was calculated to make the call for each sample. The sample was classified as “potentially positive” when any signal for the sequences were detected and confirmed as “C. pecorum positive” when the sum of the read depth of the C. pecorum specific sequences passed a threshold of higher than 1 in 10,000 reads of the total reads of the sample.

Results of scat qPCR assay and DArTcap were compared to swab qPCR assay (i.e. gold standard). For each method, we calculated sensitivity as the ability of the different methods to detect the pathogen when present:

$$\:Sensitivity=\frac{True\:positives}{True\:positives+False\:negatives}\times\:100$$

And specificity as the ability of a test to remain negative when the pathogen is absent:

$$\:Specificity=\frac{True\:negatives}{True\:negatives+False\:positives}\times\:100$$

Results

Descriptive data on DArTcap markers

DArT provides data on two main types of variation. The SNP-based data sheet was mainly used for genetic analysis of the samples, while the presence absence variation (PAV) output was used for microbial DNA detection. A total of 38 PAV markers that were consistently only present in koala with known Chlamydia infection were selected for DArTcap-based Chlamydia detection. Due to the variable quantity and sheared nature of DNA extracted from scats, a large number of loci were tried for Chlamydia detection. The initial group of markers were later confirmed as C. pecorum sequences through BLAST alignment to C. pecorum genomes available in National Centre for Biotechnology Information (NCBI: https://www.ncbi.nlm.nih.gov/). From these, 26 markers detected C. pecorum in at least 45% of extracted koala DNA samples from scat. On average, positive samples had 22 markers with a read depth greater than one, and the median total read depth was 845. However, for scats, this number is likely dependent on infection levels, environmental circumstances, and the quality of DNA extracted from scats.

Comparison of chlamydial detection

Results from the qPCR assay of urogenital swabs showed a total of n = 13 C. pecorum positive samples and n = 15 C. pecorum negative samples (Tables S1 and S2).

Classification of chlamydial status from veterinary examinations resulted in three false negatives and three false positives, resulting in a sensitivity of 76.9% and a specificity of 80.0% (Table 1 and S1).

DArTcap results from scat DNA were obtained for 27 samples. These revealed a sensitivity of 91.7% and a specificity of 100% in comparison to the gold standard (Table 1, S1 and S2), which includes detection of one false negative sample.

Two scat DNA samples extracted at KHH failed sample quality control for qPCR assays (Tables S1 and S3; QC-fail), hence these two samples could not be used. Excluding the aforementioned two samples, qPCR results for the scat DNA extracted at KHH (n = 26; i.e. 13 positive samples and 13 negative samples for the gold standard) indicated a sensitivity of 84.6%, and a specificity of 100% (Table 1). Here, two samples were false negatives (Table 1 and Table S1). All scat samples extracted at UniSC were successfully amplified for qPCR assays (n = 28). Here, qPCR results indicated a specificity of 100%, but relatively low level of sensitivity (69.2%), as four false negatives were detected (Table 1, S1 and S4).

Table 1 Comparison of methods for Chlamydia (C. pecorum) detection to the current gold standard (i.e., qPCR of DNA extracted from urogenital swabs). N is the total number of samples used for specificity and sensitivity calculations. Veterinary examinations were performed under general anaesthesia, when swabs were also collected. Scat DNA extractions were performed in two different laboratories, using different scats from the same individuals, and using different extraction protocols: KHH is the Koala Health Hub (University of Sydney), and UniSC is the University of the Sunshine Coast. Scat DNA for DArTcap analysis was extracted at the University of the Sunshine Coast.
Full size table

Discussion

Chlamydial disease is among the most significant drivers of koala infertility and mortality21. To effectively manage wild koala populations, the prevalence of the pathogen Chlamydia must be appropriately and accurately monitored. However, detecting Chlamydia in wild koalas remains difficult, with current approaches relying on the capture of koalas and a comprehensive veterinary examination under anaesthesia. In this study, we propose a non-invasive, cost-effective and large-scale approach for detection of the most prevalent of the Chlamydia species, C. pecorum, using DNA isolated from koala scats and the DArTcap genotyping. We found that this approach had similar accuracy to the current gold standard method, namely qPCR of urogenital swabs.

Genetic screening is increasingly used as a monitoring tool to inform the conservation of wild populations56. Genetic screening using technologies such as restriction enzyme-associated sequencing and next generation sequencing allows for estimation of fundamental attributes of wild populations, such as population size, inbreeding levels, sex ratios and population structure, which are critical to effective management56. Detection of C. pecorum using a combination of restriction-enzyme-associated and targeted-sequence capture methods, such as DArTcap, is convenient and cost-effective in that it can be incorporated into existing koala monitoring programs that use genetic markers. In other words, presence of C. pecorum can be obtained together with genetic marker data of individuals, using a single DNA isolation from the same sample. Once a complexity reduced library is developed using technologies similar to DArTseq, relevant probes can be identified to target specific markers for C. pecorum, which can then be captured, for instance, using DArTcap. This allows researchers to cost-effectively obtain genomic data to estimate relevant population attributes (using developed host SNP libraries) and simultaneously estimate prevalence of C. pecorum (using C. pecorum-targeted probes) in a single genotyping run without added cost, using DNA obtained non-invasively from scats (Fig. 1). Genetic markers also allow to for the identification and removal of samples originating from the same individual, avoiding the inclusion of duplicated samples and therefore increasing the accuracy of pathogen prevalence estimates when using samples collected in the wild. Moreover, this method can be implemented in large-scale surveys using detection dogs18 or citizen science57 for scat collection.

Opportunities open up to include genetic markers for additional pathogens or other elements of interest, further increasing the scope of the SNP panel and further advancing the vision of a one-stop-shop scat panel (Fig. 1). Suitability for inclusion would depend on the maximum multiplexing capability of the genotyping technology and availability in the chosen complexity-reduction method used to make the library. In this case, the DArTseq library enzymes PstI / SphI were optimised for koala genotyping; however, DNA extractions from koala scats still include a large number of sequences which can be potential microbial markers. Furthermore, while the hereby presented tool proved to be an effective and accurate diagnostic tool for detecting C. pecorum presence, the current approach does not produce information on the amount of C. pecorum present in scats, which could indicate infection load. Use of improved techniques to measure the level of infection based on the known amount of koala DNA, or other consistent bacterial DNA, could aid in the quantification of the pathogen.

Fig. 1
figure 1

Applicability of the proposed method for C. pecorum detection using fresh scats collected from the wild and SNP genotyping. Koala DNA can be analysed using the same samples, to obtain population and individual metrics, which can be used to correct disease prevalence estimates. Additional markers of interest can be added to further broaden the scope of information obtained.

Full size image

Based on the diagnostic qPCR results of urogenital swabs, we confirmed that veterinary clinical observations were not a perfect diagnostic tool for C. pecorum infection58. Discrepancies exist between veterinary clinical observations and diagnostic qPCR results due to the complexity of chlamydial disease. Some animals do not progress to disease59, others develop chronic structural disease but cease to shed bacteria, while others continue to shed after clinical disease has subsided62.

Results from qPCR of scats did not perform better than those obtained by veterinary clinical observations in terms of sensitivity, although their specificity was greater (100%), producing false negatives but no false positives. Importantly, scat DNA extractions conducted at two different laboratories achieved different results for qPCR sensitivity. The UniSC laboratory protocol was optimised for koala DNA while the KHH laboratory protocol was optimised for chlamydial DNA – and the latter extraction provided higher sensitivity (84.6% versus 69.2% sensitivity, respectively). This highlights the importance of maximising optimisation of extraction protocols and testing different protocols – just as it is important to compare the results of different sampling and storage procedures – to assess true method performance35,36,60. Given the intrinsic lower quality of non-invasive genetic samples, care should be taken every step of the way when working with these; from sample detection to collection, transport, storage, extraction and genotyping. This is especially important because non-invasive samples are particularly susceptible to degradation, with a variety of factors driving variation in their quality, including scat age and environmental degradation factors35,54. Here, we expect scat freshness to be relevant for obtaining the highest genetic quality54. The use of detection dogs trained on fresh scat odour can aid in both increasing sampling efficiency and the quality of samples18 (Fig. 1). Future research avenues include investigating the impact of environmental exposure on C. pecorum detectability (especially, sensitivity), to determine how fresh samples need to be for successful testing, as has been done for koala genetics54. Moreover, while the extractions used in this study resulted in lower sensitivity than desired, further optimisation may increase qPCR sensitivity for C. pecorum detection from scats, which could make this method suitable as a diagnostic tool. This would be especially useful when genetic information is not required, for instance, when used as a non-invasive alternative to swabs in veterinary care settings. Quantitative PCR may also inform the load of the infection, which can be useful.

The complexities of chlamydial shedding require that tests be interpreted according to the sample collected. All tests used in this study reflect chlamydial shedding at the site sampled, not infection of the animal as a whole. The dynamics of chlamydial infection, its movement between the urogenital and digestive tract, and drivers of intestinal shedding, are poorly understood59,62. Some factors that could potentially influence these include animal anatomy, age, sex, and the stage of infection and disease. Yet, while these variables and their effect on test results remain to be investigated, the sensitivity and specificity of assays presented in this study are appropriate for monitoring of populations.

Conclusion

Here, we present a non-invasive and cost-effective approach to detect the presence of C. pecorum in koala scat using a combination of restriction-enzyme associated and targeted-sequence-capture methods, namely DArTcap. This panel has the potential to be multiplexed across many pathogens and incorporated to existing population genetic studies of koala monitoring programs to conveniently and cost-effectively estimate the prevalence of infections in wild populations. Our findings have important implications for koala monitoring and conservation and open opportunities for new research linking host genetic diversity to pathogen infection, and disease vulnerability, both at the individual and population level. The proposed method of integrating host and pathogen marker panels in a one-stop shop style can be extended to other wild species where disease is a major threat.