A novel method to optimise the utility of underused moulted plumulaceous feather samples for genetic analysis in bird conservation

Non-invasive sampling methods are increasingly being used in conservation research as they reduce or eliminate the stress and disturbance resulting from invasive sampling of blood or tissue. Here we present a protocol optimised for obtaining usable genetic material from moulted plumulaceous feather samples. The combination of simple alterations to a ‘user-developed’ method, comprised of increased incubation time and modification of temperature and volume of DNA elution buffer, are outlined to increase DNA yield and significantly increase DNA concentration (W = 81, p < 0.01, Cohens’s d = 0.89). We also demonstrate that the use of a primerless polymerase chain reaction (PCR) technique increases DNA quality and amplification success when used prior to PCR reactions targeting avian mitochondrial DNA (mtDNA). A small amplicon strategy proved effective for mtDNA amplification using PCR, targeting three overlapping 314–359 bp regions of the cytochrome oxidase I barcoding region which, when combined, aligned with target-species reference sequences. We provide evidence that samples collected non-invasively in the field and kept in non-optimal conditions for DNA extraction can be used effectively to sequence a 650 bp region of mtDNA for genetic analysis.


Introduction
Working with cryptic, rare or elusive species can make obtaining invasive samples such as tissue or blood logistically difficult (Mills et al. 2000;Horváth et al. 2005). Moreover, for endangered and sensitive species, it can be difficult to obtain permits for more intrusive sampling methods, which in some cases are considered unethical (Segelbacher 2002). In these cases, biological samples such as feathers, hair, buccal cells, faecal matter and shed skin can be collected in the field with minimal disturbance to the study species (Mills et al. 2000;Bohmann et al. 2014). However, some types of non-invasive sample collection such as buccal swabbing and hair plucking still require trapping and handling of the animal (Broquet et al. 2006;Dai et al. 2015), and these methods have the potential to cause stress and affect the behaviour of an individual, even if such effects are short term (Broquet et al. 2006;Rudnick et al. 2009;Dai et al. 2015). Highly non-invasive sample collection of material such as moulted feathers, shed skin, faecal samples or environmental DNA (eDNA), which can be collected opportunistically in the field, can eliminate the need to interact with the study species (Bayard De Volo et al. 2008;Bohmann et al. 2014). This is advantageous, particularly for research on sensitive species and ecosystems, as it minimises the level of disturbance to the wildlife and prioritises the welfare of individuals being investigated (Dai et al. 2015).
Despite the advantages of using highly non-invasive sampling methods, it is often difficult to identify the biological material collected to species-level without genetic analysis Electronic supplementary material The online version of this article (https ://doi.org/10.1007/s1268 6-019-01117 -8) contains supplementary material, which is available to authorized users. (Mills et al. 2000;Rudnick et al. 2009). This is a particular problem for samples such as feathers, hair and faecal matter that can often come from a variety of species sharing the same habitat with similar somatic features (Waits and Paetkau 2005;Coghlan et al. 2012;Ahlers et al. 2017). Sample type also has an impact upon the likelihood of successful DNA extraction, for example, large primary, secondary and tail feathers are preferable for obtaining usable genetic material compared to smaller plumulaceous feathers (Dove 2000;Vili et al. 2013). Larger feathers are embedded deeper in the body of the bird and so are more likely to retain DNA containing biological material such as epithelial cells (Seki 2006;Gebhardt and Waits 2008). Primary feathers in particular can contain the umbilicus blood clot in the shaft of the quill which is a by-product of feather development and, if present, provides a plentiful source of DNA (Segelbacher 2002;Vili et al. 2013). However, opportunistic sample collection methods often remove the ability to select for sample type (Broquet et al. 2006). Furthermore, there is often no indication of how long the sample has been in the field and thus exposed to a variety of environmental conditions before collection (Hogan et al. 2008;Vili et al. 2013). Hot and humid environments provide non-optimal conditions for biological samples intended for genetic investigation (McNally et al. 1989;Hanson and Ballantyne 2005) as this can lead to a higher prevalence of decomposing microorganisms such as fungi, mould spores and keratin-degrading microorganisms, which can damage the DNA (Sangali and Brandelli 2000;Vili et al. 2013). The increased likelihood of degraded DNA in such samples reduces and often prevents the selection of non-invasively collected samples for use in genetic analyses (Vili et al. 2013). Therefore, an improved method is needed to increase the biological value of low quality samples which have been kept in non-optimal conditions; particularly for endangered or elusive species for which invasive sampling methods are not possible (Broquet et al. 2006;Hogan et al. 2008;Presti et al. 2013).
In this study we focused on the utility of non-invasively collected feather samples of the critically endangered Grenada dove (Leptotila wellsi). We provide a three-step process to allow successful extraction and amplification of mitochondrial DNA (mtDNA) from plumulaceous feather samples. Firstly, we describe improvements to a user-developed protocol for DNA extraction that increases DNA yield, followed by primerless PCR to improve quality, along with a small amplicon strategy that enabled effective mtDNA amplification using PCR, targeting three overlapping regions of the cytochrome oxidase I (COI) barcoding region.

Study species and sample collection
The Grenada dove is a critically endangered columbid, endemic to the island of Grenada (Rosenberg and Korsmo 2001; BirdLife International 2019) with a population size of 160 ± 30 individuals (Rivera-Milán et al. 2015). Habitat loss and degradation means that this species exists in small fragmented populations (Rosenberg and Korsmo 2001;Rusk 2008;Rivera-Milán et al. 2015). It is threatened by predation from a number of mammalian species as well as habitat destruction from natural damaging events such as hurricanes (Bolton et al. 2016). Low encounter rates and cryptic behaviour make surveying this species particularly difficult, requiring intensive monitoring that may cause disturbance (Rivera-Milán et al. 2015;Bolton et al. 2016;Rusk 2017). Therefore, non-invasive sampling methods are required to obtain samples for genetic analysis with minimal disruption to this endangered species.
Feather samples used for this study were obtained noninvasively, as moulted feathers, and collected opportunistically from known occupied dove territory: Mount Hartman estate and Perseverance, Grenada (Rusk 2008(Rusk , 2017. Habitat consists of both dry and mixed broadleaf evergreendeciduous second-growth forests (Rusk 2017). This tropical dry forest habitat has a minimum temperature of 22 °C and temperatures that can reach up to 32 °C, with a maximum rain fall of 259 mm in the rainy season and a minimum of 67 mm in the dry season (Meteostat 2018;Nelson et al. 2018). Due to the opportunistic nature of the sample collection, the feathers used in this study spent an unknown amount of time in the litter bed of this hot and humid environment before collection. Samples were stored in sample bags at 4 °C until transportation by airmail to the UK in June 2018. On arrival in the UK, the samples were cleaned with 70% ethanol and stored at − 20 °C. The sample set (n = 160) used in this study was comprised of 152 plumulaceous Fig. 1 Examples of the plumulaceous feathers used in this investigation ( Fig. 1), three secondary and three primary feathers, as well as two egg shells recovered from the forest floor.

DNA extraction
The commercially available QIAGEN DNeasy® Blood and Tissue kit was used to conduct DNA extraction. Extractions were carried out as per the 'User-Developed Protocol: Purification of total DNA from nails, hair, or feathers using the DNeasy® Blood & Tissue Kit' (QIAGEN Inc., Crawley) with the following alterations to incubation time, temperature and volume of DNA elution buffer. Feather samples were cut into 1 cm pieces directly into a sterile 1.5 ml microcentrifuge tube containing the lysis buffers using sterile scissors to increase surface area (Presti et al. 2013). The incubation step was increased to 48 hours in order to achieve complete sample lysis on samples that are particularly difficult to digest (Bush et al. 2005;Bayard De Volo et al. 2008). To increase DNA yield we heated buffer AE (elution buffer) at 70 °C for 10 min before use. Buffer AE contains the organic compound Tris [tris(hydroxymethyl) aminomethane, (HOCH 2 ) 3 CNH 2 )] and EDTA [Ethylenediaminetetraacetic acid (C 10 H 16 N 2 O 8 )] which functions to rehydrate the nucleic acids and release DNA from the silica membrane. This process is improved when the DNA and silica are exposed to higher temperatures (Bruns et al. 2007;Zhou and Ling 2011;Haddad et al. 2017). We used an elution volume of 100 µl in a two-step process, giving a final volume of 200 µl which was subsequently stored at − 20 °C. Although using half the recommended volume of elution buffer decreases the DNA yield, the aim was to increase the final concentration of DNA as it is well documented that a higher concentration of starting DNA in PCR reactions improves the likelihood of successful DNA amplification (Kishore et al. 2006;Rohland and Hofreiter 2007;Graziano et al. 2013).
In order to assess whether the alterations to the protocol had optimised DNA concentration following extraction, we compared the concentration of DNA from samples extracted using the standard manufacturer's protocol and the protocol we have outlined above. DNA concentrations for samples extracted using the standard protocol were obtained from a feather set that arrived in the laboratory in 2016 (n = 50). We used a number generator to randomly select 20 plumulaceous feather samples from the feather set obtained in 2016 and extracted using the standard protocol and from the feather set obtained in 2018 and extracted using the optimised extraction method. The DNA concentration of each sample was quantified using the Invitrogen™ Qubit™ 3.0 Fluorometer, which has a DNA detection range of 0.50-600 ng/ml, using dsDNA High Sensitivity settings following the manufacturer's protocol (Table 1). A Shapiro-Wilk test (Shapiro and Francia 1972) indicated that the data were not normally distributed (p < 0.05) therefore a Wilcox test (Wilcox 2008) to compare the difference in DNA extracted when using the standard protocol and the optimised protocol outlined in this study. All analyses were performed in R (R Core Team 2013).The results of the test showed that samples that underwent the optimised protocol had a significantly higher extracted DNA concentration than the standard protocol (W = 81, p value < 0.01, Cohens's d = 0.89). Samples that were lower than the detection range of the Invitrogen™ Qubit™ 3.0 Fluorometer were assigned the value 0.49 ng/ml. Table 1 shows that 30% of the samples extracted using the standard protocol were of too low concentration to evaluate. In comparison, only 5% of the samples that were extracted using the optimised protocol had a concentration below the range of the Invitrogen™ Qubit™ 3.0 Fluorometer. Therefore, both DNA extraction success rate and DNA concentration were increased by the optimised protocol. To test that target DNA, rather than that of subsidiary material, had been extracted and to ascertain the potential presence of PCR inhibitors, which are common when using non-invasively collected samples (Waits and Paetkau 2005;von Thaden et al. 2017), avian cytochrome b (cyt b), was amplified. We designed primers using Primer3 (Koressaar and Remm 2007;Untergasser et al. 2012) based on a cytochrome b sequence from the white-tipped dove (Leptotila verreauxi), whose whole mitochondrial genome is available on GenBank, accession number: NC_015190.1 (Pacheco et al. 2011). The Primer3 output for the designed primers is provided in Online Resource 1. This species was chosen as it has been suggested, based on the ecology of the species, that it is closely related to the Grenada dove (Blockstein and Hardy 1989), although molecular phylogenetic analysis has not yet been carried out for the Grenada dove. The primers were designed to include a 200-250 bp region of the cyt b gene, with a length of around 20 bases long, a GC content of around 50% and melting temperatures that are no more than 5 °C apart (Dieffenbach et al. 1993;Naqib et al. 2019). The chosen primers targeted a 203 bp region of cytochrome b gene: CYTB_2 Forward: 5′-CTG CCT ACT AAC CCA GAT CCT-3′ and CYTB_2 Reverse: 5′-AGG AGC CGT AGT AGA GTC CT-3′. To prevent contamination of samples, PCR was conducted in a PCR hood where tube racks, pipettes and tubes were exposed to UV light for 20 min prior to setting up the reaction (Bayard De Volo et al. 2008). PCR was conducted using illustra™ PuReTaq Ready-To-Go™ PCR Beads (GE HEALTHCARE; Chicargo) with a final volume of 25 µl containing ~ 2.5 units of recombinant PuReTaq DNA polymerase, 200 µM of dNTP's in 10 mM Tris-HCl, 50 mM KCl and 1.5 mM MgCl 2 stabilizers, BSA and reaction buffer, 5 µl of DNA template and a negative control using sterile H 2 O. PCR was conducted on a TECHNE TC-3000 thermocycler (Bibby Scientific Ltd; Stone) using the following conditions optimised for this primer set: initial denaturation at 95 °C for 5 min, 45 cycles of 95 °C for 30 s, 48 °C for 30 s, 72 °C for 60 s and a final extension at 72 °C for 5 min. PCR products were separated by electrophoresis (Westermeier 2005) on a 2% (Mardis and McCombie 2017) agarose gel (Thermo Fisher Scientific, Waltham) and visualised using a BioRad Gel Doc™ EZ Imager and Image lab 4.0 software (Bio-Rad Laboratories 2017). DNA extraction using the optimised method resulted in a visible band on the gel at the 203 bp target region of avian cyt b for 154 out of 158 (97.5%) of the feathers being used in this study (Table 2) thus confirming target DNA rather than subsidiary material had been amplified.

Primerless PCR
Primerless PCR, also known as 'DNA shuffling' or 'sexual PCR', exposes the DNA template to Taq DNA polymerase, dNTPs and a heating and cooling cycle which serves to denature the sample into smaller fragments which then anneal to each other (Stemmer 1994;Melnikov and Youngman 1999;Brakmann and Schwienhorst 2004;Suenaga et al. 2005;An et al. 2011). This is known as self-priming and functions to repair DNA damage such as nicks, fragmentation, abasic sites and blocked 3′-ends in degraded DNA samples that may inhibit amplification (Diegoli et al. 2012).
Primerless PCR reactions were conducted using illus-tra™ PuReTaq Ready-To-Go™ PCR Beads with a final volume of 25 µl including 5 µl of DNA template. Samples were subjected to a PCR cycle with the cycling parameters: initial denaturation at 95 °C for 5 min, 10 cycles of 95 °C for 30 s, 55 °C for 30 s, 72 °C for 60 s, 35 cycles of 95 °C for 30 s, 50 °C for 30 s, 72 °C for 60 s and a final extension at 72 °C for 5 min. A negative control was also generated at this stage by exposing 25 µl of PCR reaction mix and no DNA template to the same PCR cycle to ensure no contamination or false amplification occurred during the primerless PCR process.
In order to investigate the efficacy of the primerless PCR process we chose low quality and quantity samples for comparison. The samples chosen were identified by lack of gel band presence, brightness and/or clarity following optimised DNA extraction and amplification (Jacobs et al. 2013;Thiel et al. 2014) of the 203 bp region of cytochrome b gene outlined above. Samples consisted of: 10 plumulaceous feathers, one secondary feather, and one egg shell. Each sample was used as a substrate for the amplification of the barcoding region of the cytochrome oxidase I gene using primer set AWCF1 and AWCintR2 (C1; 328 bp) (Patel et al. 2010), as described in the small amplicon strategy section below, with and without a prior primerless PCR stage. A standardised dilution factor was used to ensure the same amount of DNA template was used in each primered PCR reaction. Following amplification, the samples were visualised on the UV transilluminator as a comparison for effective amplification with and without exposure to primerless PCR. This can be seen 1 3 for feather samples in Fig. 2 and the egg shell sample in Fig. 3, which show that more bands were present and were more defined after the addition of the primerless PCR step. Nine of the twelve samples failed to amplify without the addition of primerless PCR but successfully amplified and presented clear bands when exposed to primerless PCR. For example, sample 69 (Fig. 2) shows a brighter and more defined band is present after under-going the primerless PCR process. The increased prevalence of bands and the improved clarity, brightness and definition of the present bands after primerless PCR indicates the improved quality of DNA after exposure to this process.

Small amplicon strategy
The third step we adopted was a small amplicon strategy to successfully amplify and sequence three small overlapping amplicons, which were combined to construct a longer and more informative section of the gene. PCR products were separated and visualised using agarose-gel electrophoresis. All three primer sets successfully amplified DNA from our non-invasively collected samples (Fig. 4). Samples were purified using the QIAquick PCR Purification as per the manufacturer's protocol (QIAGEN Inc., Crawley). Samples were then prepared for sequencing using the Mix2Seq Kit (Eurofins Genomics, Luxembourg) according to the manufacturer's instructions and submitted to Eurofins Genomics, where they were sequenced using Sanger sequencing methods. Sequence data for the three overlapping regions were reconstructed to create an approximately 650 bp sequence (Fig. 5a). Firstly, consensus sequences were obtained for each amplicon by aligning the forward and reverse sequence data in NCBI Basic Local Alignment Search Tool (BLAST®) (Johnson et al. 2008) and were used in all further reconstructions and alignments. Obtaining consensus sequence is particularly important when using primerless PCR as random fragmentation and self-priming  (Patel et al. 2010) both with and without exposure to primerless PCR can introduce artificial recombinants or base errors but this can be counteracted by using consensus sequence (Weber et al. 2000). To assemble the longer combined sequence, each of the small amplicons were aligned using a global alignment algorithm with Emboss Explorer merger tool (Fig. 5b). A heuristic approach is used to select the base with the best local sequence quality score in the case of any mismatches (Bell and Kramvis 2013). Highly degraded DNA can produce poor read length or low quality base calls in the sequence data, particularly at the ends of the trace producing "messy" end sequences (Bell and Kramvis 2013). Therefore, most probably due to the highly degraded nature of the samples used in this study, some base inconsistencies were recognised. In these cases the Emboss Merger local quality score, along with the electropherogram obtained from Eurofins for each sequence (Fig. 5c), were assessed and the base with the highest quality score as per both the emboss and Eurofins sequence quality assessment was selected (Bell and Kramvis 2013). Due to the non-invasive sample collection method used for these samples, along with the inability to identify the species from the morphological features of the feather alone, each sequence was run in NCBI's Basic Local Alignment Search Tool (BLAST®) to predict species identification (Johnson et al. 2008). This search indicated that the sample, presented in Fig. 5, was from the Caribbean columbid, white-crowned pigeon (Patagioenas leucocephala), with a BLAST total score of 989 and a query coverage of 91% to Genbank sequence JJQ175689.1 (Schindel et al. 2011). Combined sequence data obtained from the three small amplicons were aligned with the corresponding COI barcoding region sequence data from a known L. wellsi specimen (unpublished data-Genbank reference sequence not yet available for this species) and P. leucocephala (accession number JQ175689.1) (Schindel et al. 2011) to confirm the expected nitrogenous base positions (Johnson et al. 2008), which is particularly important given the possibility of potential base errors introduced by primerless PCR (Weber et al. 2000). Target-species was confirmed for the 650 bp length of the COI barcoding region from non-invasively collected plumulaceous feather samples.

Discussion
The methods outlined in this study: an optimised userdeveloped DNA extraction protocol, use of a primerless PCR technique, and a small amplicon PCR strategy, facilitated the attainment of target-species mtDNA sequence data of 650 bp in length from non-invasively collected plumulaceous feather samples. Problems arising due to poor quality of DNA extracted from non-invasively collected samples can often dissuade researchers from proceeding with genetic analysis, causing the potential of such samples to go unrecognised (Horváth et al. 2005;Speller et al. 2011). This can also lead to a preference in invasive as opposed to non-invasive sample collection methods due to the higher confidence in invasive sample quality (Johnson and Clayton 2000a;Harvey et al. 2006). The methods we have outlined allow such samples, which previously may have been discounted for genetic analysis, to be successfully used to obtain informative sequence data. It is widely documented that feather type influences the success and yield of DNA extraction from feather samples (Bayard De Volo et al. 2008;Hogan et al. 2008). Primary feathers are widely used as a resourse for genetic monitoring of wild bird populations, as they often contain a blood clot located in the superior umbilicus of the feather shaft (Horváth et al. 2005). Along with assessing the genetic variation within populations (Nichols et al. 2001), feather samples have been used to investigate phylogeny and biogeography (Johnson and Clayton 2000b), and to understand the impact of anthropological processes on genetic structure and gene flow (Fourcade et al. 2016) of bird species. However, a study on molecular sexing and microsatellite genotyping of hyacinth macaw (Anodorhynchus hyacinthinus) found that despite using larger moulted feathers (with a size greater than 20 cm) than in this study, feathers that spent more than 7 days in the field and were of poor physical quality, had a low success rate in yielding sufficient DNA for genetic analysis (Presti et al. 2013). Nevertheless, we have demonstrated that plumulaceous feathers that are found in the field Sequence data from sample M79. a The 650 bp sequence obtained for sample M79 aligned with the corresponding COI barcoding region sequence data for Leptotila wellsi and Patagioenas leucocephala, labelled with its accession number JQ175689.1 (Schindel et al. 2011). b Emboss Explorer (Bell and Kramvis 2013) OUTPUT example for two overlapping sequences from sample M79 using forward and reverse consensus sequences. c Electropherogram example with base quality score as assigned by Eurofins for the forward sequence of COI_2 do not necessarily need to be excluded from sample sets as they can still be utilised to yield informative data.
The use of primerless PCR increased the amplification success in subsequent PCR reactions for samples that had proven difficult to amplify. Bands from samples exposed to primerless PCR were visually clearer and brighter, which suggests a higher quality of DNA (Hughes-Stamm et al. 2011;Jacobs et al. 2013;Lucena-Aguilar et al. 2016). Primerless PCR has been used in studies of ancient DNA, including that of Weber et al. (2000), who increased the successful amplification of ancient DNA from bone samples when investigating the population bottleneck of the northern elephant seal (Weber et al. 2000). The same technique was adopted using museum samples from the African horseshoe bat (Rhinolophus darlingi) to investigate the phenotypic convergence of its evolutionary history (Jacobs et al. 2013). Both studies reported that primerless PCR improved the recovery of DNA from ancient samples but did not comment on differences in DNA quality post primerless PCR (Weber 2004;Jacobs et al. 2013). To the best of our knowledge, our study is the first to demonstrate the utility of primerless PCR to increase DNA quality for degraded non-invasive feather samples and demonstrates that this technique can be applied to a variety of non-invasive samples collected in the field.
Although the primerless PCR technique recovers the quality of DNA obtained from the degraded samples this process does not completely repair damaged DNA, particularly when the DNA damage is highly fragmented, which is a common problem with samples kept in non-optimal conditions (Diegoli et al. 2012). The varying quality of sequence data at ends of the trace producing "messy" end sequences is a particular problem for overlapping regions resulting in base inconsistencies (Stiller et al. 2009). However, this is not specific to the techniques we describe here and is prevalent in many genetics datasets, though it is not widely documented (Sobel et al. 2002;Hackett and Broadfoot 2003;Bonin et al. 2004). Sequencing error can lead to misidentification of individuals, misinterpretations and erroneous conclusions to be drawn from genetic analysis of relatedness and population structure (Hogan et al. 2008). Sequencing error can occur at any stage when obtaining genetic information but there are a number of actions that can be taken to minimise these errors (Bonin et al. 2004). The following have been adopted in this study and are proposed for future studies to limit the risk of sequencing error following primerless PCR: (1) the amplification of small amplicons because genotyping error correlates with amplicon size (Vili et al. 2013); (2) the inclusion of negative controls (Waits and Paetkau 2005;Alda et al. 2007;Boonseub et al. 2012); (3) sequence quality assurance by using only forward and reverse consensus sequences with further analysis and inspection of electropherogram scoring levels; and (4) cross-reading and aligning sequence data with reference samples (Weber et al. 2000;Bonin et al. 2004). For downsteam analyses, it is also recommended to include a sequencing error rate when using genetic data for population and phylogenetic analyses (Bonin et al. 2004).
We have outlined methods that achieve successful extraction and amplification of an informative length of mtDNA from non-invasively collected plumulaceous feather samples. The analysis of mtDNA has many practical applications in conservation including species identification as well as presence/absence detection. The high mutation rate of mtDNA in comparison to nDNA allows the identification of Evolutionarily Significant Units (ESUs), giving an insight into the phylogenetics of a species (Cronin 1993;Wan et al. 2004;Gupta and Bhardwaj 2013). Data pertaining to phylogenetics have multiple conservation implications including evolutionary divergence and speciation, along with phylogeography and rates of change relating to morphology and behaviour of a species (Huang et al. 2009). Determining the point at which speciation occurs or defining a species as genetically distinct-known as the phylogenetic species concept-is particularly useful for resolving taxonomic uncertainties, outlining wildlife legislation, and in identifying conservation priorities (Hazevoet 1996;Wan et al. 2004;Pellens and Grandcolas 2016;Chen et al. 2018). Analysis of mtDNA can be used to make long-term species conservation action plans from an evolutionary perspective (Nielsen et al. 2017) and to inform short-term demographic management of populations through identification of population change and connectivity therefore providing information on the effects of habitat fragmentation (Cronin 1993;Moritz 1994;Nabholz et al. 2009). Therefore, our improved methods of extraction and amplification of mtDNA from non-invasively collected, low quality feather samples, have the potential to extend the applicability of molecular analyses in studies aimed at the conservation of endangered bird species, for which it is typically difficult to obtain high quality samples.

Conclusion
In conclusion, the optimised user-developed DNA extraction protocol, along with the use of the primerless PCR technique, and a small amplicon PCR strategy, are sufficient to enable DNA extraction and mtDNA amplification from moulted plumulaceous feathers collected non-invasively and opportunistically in the field. This not only provides evidence in support of using non-invasive sampling methods for genetic analyses, in particular when applied to endangered species, but also highlights the utility of biological material kept in non-optimal conditions, may previously have been discounted (Rawlence et al. 2009;Vili et al. 2013). Data collected in this manner is informative for species identification, presence/absence detection, population structure and phylogenetic analyses of rare and elusive species (Bonin et al. 2004;Marucco et al. 2011;Adam et al. 2014), all of which are key questions in conservation research.