Increased DNA typing success for feces and feathers of capercaillie (Tetrao urogallus) and black grouse (Tetrao tetrix)

Abstract Noninvasive sampling, for example, of droppings or feathers, is a promising approach for molecular genetic studies on endangered and elusive animal species. Yet, such specimens are known for containing only minute amounts of DNA, resulting in lower typing success rates relative to analyses on fresh tissues such as muscle or blood. Furthermore, artefactual signals as well as contamination are more likely to occur when DNA is limited. To increase the reliability of DNA typing from noninvasive samples, optimized DNA extraction and polymerase chain reaction protocols were developed, taking advantage of developments in the forensic field aiming at successful molecular genetic analysis of DNA templates being low in quality and quantity. In the framework of an extensive monitoring project on population dynamics of capercaillie and black grouse in the Tyrolean Alps, feces samples and molted feathers from both species were collected. On a subset comprising about 200 specimens of either species, eight polymorphic short tandem repeat (STR) markers were analyzed to test these improved protocols. Besides optimizing DNA yields, both lowered sample consumption and reduced hands‐on time were achieved, and the rates of informative profiles amounted to 90.7% for capercaillie and 92.4% for black grouse. Similarly, high success rates had not been achieved in earlier studies and demonstrate the benefit of the improved methodology, which should be easily adaptable for use on animal species other than those studied here. The STR genotypes were not only powerful enough to discriminate among unrelated birds but also appeared fit for telling apart closely related animals, as indicated by Pi and Pisib values. The software package allelematch aided analysis of genotypes featuring possible dropout and drop‐in effects. Finally, a comparison between molecular genetic and morphology‐based species‐of‐origin determination revealed a high degree of concordance.


| INTRODUC TI ON
Conservation and population genetic studies are indispensable sources of information in decision-making processes regarding environmental management and the protection of endangered species.
For suitable management and protection plans, reliable data on population sizes, trends, and conservation status are required.
A downside of routine genotyping approaches is that they typically require a direct sampling approach involving capture of the animals and invasive collection of DNA sources such as blood or muscle. This may cause unacceptably high levels of disturbance (Regnaut, 2004). On the contrary, collecting noninvasive sampling material (e.g., feces, feathers, or hairs) in an "indirect sampling approach" largely avoids unwanted interactions with the studied animals (Segelbacher, 2013;Taberlet, Waits, & Luikart, 1999). However, establishing reliable genotypes can be a nontrivial task, as DNA extracted from noninvasive samples may be (very) low both in quantity and in quality. For instance, fecal samples may not only contain just minute amounts of template DNA but also constitute a source of polymerase chain reaction (PCR) inhibitors.
Allelic dropout phenomena and/or false alleles are more likely to occur with such samples, and contamination is more of a concern with low-copy-number templates (Taberlet et al., 1999). Thus, compared to directly drawn samples (e.g., blood), noninvasively collected specimens can be considered a more difficult matrix for DNA extraction. As a consequence, genotyping success rates tend to be lowered and, to avoid generation of phantom individuals by declaring flawed profiles unique, error needs to be accounted for.
Extraction from challenging specimens should, therefore, yield the highest possible DNA quality and quantity to facilitate reliable genotyping in downstream analyses. By taking advantage of research in forensic genetics aiming at optimized DNA extraction and multiplexed DNA analysis techniques, extraction methods for feces and feathers were optimized and applied to a molecular genetic study on free-ranging capercaillie and black grouse in Tyrol.

| Study sites and sample collection
The study area was located in the Inner Alps of western Tyrol, Austria ( Figure 2). It was mainly situated in the subalpine and montane zones and ranged from 1,350 to 2,300 m above sea level. The arboreous area was heavily dominated by Picea abies, and the forest boundary consisted mainly of Larix decidua and Pinus cembra (Pitschmann, Reisigl, Schiechtl, & Stern, 1980). Species-specific habitat models based on Graf and Bollmann (2008) and Masoner (2012) were used to identify priority habitats for capercaillie and black grouse and to define areas for intense field collection. According to background information provided by local hunters, all such chosen areas-of-interest contained known lek sites. Systematic sampling of capercaillie and black grouse feces and feathers covered the entire area of interest, and lek site information was not forwarded to the field surveyors.
Fieldwork took place between March and May 2012, and study sites were visited twice within a period of 10-14 days. During this sampling period, both black grouse and capercaillie aggregated around lek sites for mating, which aided representative sampling of the local populations (Höglund & Alatalo, 1995). Per day and surveyor, an area of 56-139 ha was sampled along transects spanning the entire width of the study site and being 100 m apart from each other. However, in reaction to local conditions in topography, this uniform pattern was re-adjusted if necessary.
Furthermore, field surveyors were allowed to accommodate their tracks for covering suitable habitat structures. A short description of the recovery site along with its GPS coordinates was recorded for every specimen.
Based on our experience, the cooling effect of snow appears to decrease biodegradation and thus increases chances of successful DNA typing. Therefore, sample collection was restricted to feathers appearing intact and to feces on snow or displaying a visibly moist surface. All specimens were stored at −20°C until further use.
Initial morphological species assignments were performed by trained field surveyors on-site.

| Voucher specimens
Biological reference material for both bird species was provided by the Alpenzoo Innsbruck (Innsbruck, Austria). After collection from the wing vein by a staff veterinarian, blood of capercaillie (one individual of each sex) and black grouse (one male, two females) was airdried on Whatman FTA blood stain collection cards (GE Healthcare, Chalfont St Giles, UK) and stored at ambient temperature in the dark. Additionally, fresh feathers and feces were collected from all reference birds and stored at −20°C.

| Ethics statement
With prior notice, consultation, and agreement by the person entitled to the hunting right, the permission to enter the hunting grounds was granted for the purpose of conducting this study. As the aim of our study was not to find the birds themselves but to provide noninvasive sampling material, field surveyors only searched for droppings and feathers. Therefore, the disturbance of the investigated birds was kept to a minimum, and no specific permission to work with living animals was needed. Reference material of voucher animals was collected by a veterinarian in the "Alpenzoo" (Innsbruck, Austria) during a routine veterinary examination required by the law.

| DNA extraction
DNA extractions followed best practice lab routines and were performed in a dedicated pre-PCR area. Extraction blanks were run along the specimen of interest.
DNA from blood samples was extracted using the DNeasy Blood & Tissue kit (Qiagen, Hilden, Germany) following the manufacturer`s instructions. For feather samples, we used the EZ1 DNA Investigator kit (Qiagen). Therefore, 1 cm of the basal part of the feather was cut into small pieces as described by Segelbacher (2002). Lysis of large feathers (i.e., remiges and rectrices) was performed overnight at 56°C under constant shaking (550 rpm) in 1,000 μl TN ca buffer comprising 10 mmol/L Tris-HCl pH 8.0, 100 mmol/L NaCl, 1 mmol/L F I G U R E 2 Study sites (shaded) in the Inner Alps in the western part of Tyrol. Rivers are shown as dashed lines. 1: Inn Valley, 2: Kauner V., 3: Pitz V.; A: Austria, G: Germany, I: Italy CaCl 2 , 2% SDS (w/v), 40 μl proteinase K (>600 mAU/ml, Qiagen) and 40 μl dithiothreitol (1 mol/L, Qiagen) as described by Hellmann, Rohleder, Schmitter, and Wittig (2001). For smaller feathers, the lysis cocktail was halved in volume. Further steps were conducted according to manufacturer's instructions (EZ1 DNA Investigator kit large volume protocol).

DNA extraction from feces was performed with the DNA Stool
Mini kit (Qiagen) according to the vendor's protocol with modifications as specified below. Using sterile cotton sticks (Applimed, Châtel-Saint-Denis, CH), dry swabs were taken from the surface of the samples.
Regions with visible traces of uric acid were avoided (Segelbacher & Steinbrück, 2001). Cotton swabs were cut, and digestion was conducted overnight at 56°C in a shaker (550 rpm ;Segelbacher G., personal communication, February 12, 2013). Subsequently, 4 μl carrier RNA (1 μg/μl, Qiagen) was added, and the DNA was bound to the provided silica membrane (QIAamp spin column, Qiagen). After two washing steps with the provided buffer, the DNA was sequentially recovered in 2 × 75 μl buffer AE (Jacob et al., 2010) and stored at −20°C.

| PCR amplification and electrophoretic product sizing
Profiles featuring data for all eight loci are dubbed "full" hereafter, whereas partial profiles comprising information for 5-7 loci only were considered "informative."

| Molecular genetic species identification
Molecular genetic allocation to capercaillie or black grouse relied on species indicative allele size ranges observed for markers BG15, BG18, and sTuT2 (Jacob et al., 2010;G. Segelbacher, personal communication, February 12, 2013). Furthermore, using these loci, two other local grouse species, hazel grouse (Tetrastes bonasia L.) and rock ptarmigan (Lagopus muta Montin), can be identified, too (G. Segelbacher, personal communication, February 12, 2013). was used for identifying unique multilocus genotypes. This software enables the assignment of genotypes to individuals by also considering the possibility of missing or compromised data (Galpern et al., 2012). The R script utilized is to be found in the supporting information accompanying this study (Appendix S1).

| Statistical analyses
Allele frequencies, probability of identity (Pi), and probability of identity for siblings (Pi sib ) were estimated with GENECAP [version 1.4; (Wilberg & Dreher, 2004)]. Pi defines the likelihood of two randomly chosen, unrelated individuals sharing the same genotype, and Pi sib describes the probability of two identical genotypes in siblings (Waits, Luikart, & Taberlet, 2001 (Valière, 2002)]. Input files used in statistical analyses are to be found in the Data Files S1 and S2.

| STR genotyping of voucher animals
For capercaillie and black grouse, voucher specimens of both sexes were genotyped using three different types of biological reference material (feces, feathers, blood). As expected, these sample types yielded (1) full STR profiles (i.e., successful amplification of all eight targeted microsatellite loci), and (2) perfectly matching intraindividual genotyping results. Furthermore, our results for capercaillie and black grouse confirmed species indicative allele sizes at loci BG15, BG18, and sTuT2 (Tables 2 and S1).

| STR genotyping of field samples
Based on the initial species assignment relying on morphological fea-

| Comparison of genetic and morphological species identification
The rate of correct morphology-based species assignments was mainly influenced by the species of origin and sample type. Utilizing the molecular genetic species assignments as a reference, for caper-

| Reliability of genotyping
To test the reliability of analysis, PCR genotyping was repeated for twelve capercaillie and twelve black grouse samples six times each.
In the initial round of STR genotyping, per species six of these sam-

| Species assignments
When analyzing noninvasively collected feces and feather samples, reliable species-of-origin information is essential for data interpretation. For several grouse species, this information can be obtained by means of molecular genetic as well as morphologybased approaches. At the molecular genetic level, species indicative allele size ranges at the STR loci BG15, BG18, and sTuT2 allow for allocation of feathers and feces samples to capercaillie, black grouse, hazel grouse, and rock ptarmigan (Jacob et al., 2010;G. Segelbacher, personal communication, February 12, 2013). This observation was confirmed by our data (Table 2). Likewise, differences in size and appearance of feces and feathers can facilitate ascertainment of species of origin and sex (Klaus, 1986(Klaus, , 1990. Feces of male capercaillie are usually thicker (10-12 mm) than those of female individuals of the same species (8-9 mm, Figure 4; Klaus, 1986). The latter, however, may be confused with droppings from male black grouse, which were reported at an average thickness of 9.5 mm (Klaus, 1990). Finally, female source droppings of black grouse are usually smaller with an average diameter of 7.7 mm (Klaus, 1990).
A comparison between the genetic and morphological approach for species assignment of feces and feathers revealed a high degree of consistency. Field collection was performed by experienced field surveyors, and especially for feces, the rate of correct assignments was very high. A few difficulties associated with feathers were observed for several reasons. Feathers of female capercaillie and black grouse can be mixed up due to similar coloration and pattern. More critically, however, the plumage of both male black grouse and rock ptarmigan contains small white feathers that can be confused with each other. This may cause problems, as the habitats of both species overlap (Klaus, 1990;Storch, 2007).
TA B L E 3 Genetic parameters and rates of allelic dropout and false alleles for capercaillie and black grouse These aggravating factors have to be taken into consideration during the decision-making process. The success rates of the morphology-based approach, therefore, primarily depend on the experience level of field surveyors. However, if well-trained personnel is unavailable, sample examination and specimen-to-species assignments may also be realized, for example, by an expert in the laboratory. In summary, it can be said that the morphological approach not only acts as gatekeeper for molecular genetic analyses but also makes for an important stand-alone tool to establish the presence of particular species in a given habitat, particularly if financial resources and/or time for molecular genetic analyses are limited.

| DNA sampling and extraction
We set out to optimize and streamline DNA extraction from feces with particular regard to nucleic acid yield and quality as well as diminished consumption of biological material, time, and labor. The feces surface layer was assumed to contain the highest concentration of epidermal cells and less PCR inhibitors. Hence, a dry swab from the surface could provide sufficient amounts of valuable DNA and simultaneously offer two significant advantages over the original procedure involving lysis of the specimens. Firstly, swabbing takes considerably less hands-on time than processing entire feces samples or large portions thereof. This is of practical relevance when considering the high sample numbers usually required in ecological and conservation biological field studies. Secondly, only minute amounts of precious sample material are required by swabbing.
Specimens were not destroyed for DNA extraction, and the larger portion remained available for potential use in further analyses or studies.
For feathers, we employed a DNA extraction approach aiming at challenging forensic hair samples, which includes a dedicated extraction buffer containing calcium (Hellmann et al., 2001). Use of this particular buffer, which was specifically designed for DNA extractions of keratinized cells, resulted in complete lysis of the feather samples. DNA from feathers mainly derives from remaining keratinized pulp caps and from adherent cells (i.e., epidermal cells) from the surface of the calamus (Horváth, Martínez-Cruz, Negro, Kalmár, & Godoy, 2005). However, feathers, which have been exposed to adverse environmental conditions, may lack these adherent cells. Additionally, the DNA may already have been degraded, and DNA in keratin cells of the calamus may be the only source of DNA (Hellmann et al., 2001). Therefore, by digesting the whole feather tip, an additional source of DNA was made accessible, which may explain the observed increase in genotyping success.
Application of these improved protocols for DNA extraction is not restricted to the bird species analyzed here, but can be easily adapted for use on noninvasive samples from a broad range of animals.

| Genotyping and suitability of STR markers
It is generally known in forensic genetics that minute amounts of DNA in tissues exposed to environmental stress show degradation with fragment sizes above 250 bp being significantly less abundant in the extract [e.g., (Grubwieser et al., 2006)]. Therefore, shortening amplicon sizes by moving primers closer to the repeat region (Jacob et al., 2010) is expected to increase the amplification success rate (Grubwieser et al., 2006). Accordingly, in our study, a high rate of 8-locus profiles was achieved by testing polymorphic STR markers with multiplex PCR assays that yielded short amplicons (<220 bp).
Furthermore, the optimized feather and feces extraction methods bolstered genotyping success. For both capercaillie and black grouse, more than 90% of the samples were suitable for genetic individualization based on the criteria applying in this study (Table 4).
The success rate for capercaillie feathers was slightly reduced, maybe due to a stochastic effect arising from the low sample number. Eight  (Table 4).
The estimated Pi and Pi sib values (Table 3) were within the threshold recommended by Waits et al. (2001), who defined values between 1 × 10 −2 and 1 × 10 −4 as reasonably low for ecological studies. Therefore, the applied system of eight STR loci appeared powerful enough to discriminate among individuals, even in the case of close relatedness (e.g., parent-child or sibling relationship).
In capercaillie and black grouse, a statistically significant deviation from HWE was observed at locus mTuT1. Jacob et al. (2010) detected the same issue in similar studies on capercaillie. In black grouse, four additional loci (BG15, sTuD6, sTuD1, and sTuT4) deviated statistically significantly from HWE. Generally, such a deviation can indicate inbreeding or a distinct population structure. Other possible explanations are poor DNA quality resulting in allelic dropout or primer-binding site mutations causing reproducible null alleles. Yet, the short amplicon sizes generated with our approach and the observation that only particular loci were affected, rendered random dropout a rather unlikely explanation (Jacob et al., 2010). The software microchecker revealed the potential presence of null alleles for all loci deviating from HWE. In addition, the possibility of scoring errors caused by stutter bands of dinucleotide repeat markers cannot be completely excluded for sTuD6 and sTuD1 according to microchecker (data not shown). This class of microsatellites is known for its propensity to stutter peak formation, which can cause ambiguous allele calls (Butler, 2005;Taberlet et al., 1999). Nevertheless, due to their high capacity to discriminate among individuals (Table 2), all markers deviating from HWE were considered further on.
Based on GIMLET results, allelic dropout (ADO) and false alleles (FA) were the primarily observed error types (Table 3), as determined in the reliability testing experiment. In comparison with similar studies on capercaillie, the estimated error rates due to ADO (15.3%, Table 3), averaged over loci, were below the values reported by Regnaut, Lucas, and Fumagalli (2006) (Taberlet et al., 1996). In the reliability test, the percentage of FA was very low. Therefore, it can be assumed that both artefacts and contamination were rare in the main study as well. This again highlights the suitability of our DNA extraction and multiplex PCR amplification approaches for analysis of feces and feather samples from capercaillie and black grouse.

| Individualization of specimens
While being highly advantageous, noninvasive sampling of populations also entails analytical challenges. Due to the lack of a known one-to-one link between a specimen and an individual, multiple samples originating from a single animal may be analyzed. Thus, during data analysis, matching multilocus genotypes are pooled and collapsed into a unique consensus profile, which is subsequently attributed to a single, yet unknown, individual. In such an approach, genotyping error has to be accounted for to avoid creation of phantom individuals by declaring flawed STR profiles unique. To date, confirmation of genotyping results by a multiple tubes approach (Taberlet et al., 1996) represents the state-of-the-art attempt in controlling genotyping error and obtaining the accurate number of unique genotypes in a data set. In a multiple tubes approach, STR profiles are challenged in several independent rounds of genotyping. Errors are then revealed by building a consensus genotype for each sample. In some cases, however, missing information for some loci will persist, and due to limited financial resources, large-scale repeat PCR genotyping may not always be feasible. Moreover, global application of a multiple tubes approach appears not always to be absolutely necessary. On data sets containing low proportions of partial multilocus profiles and a considerable number of repeatedly sampled individuals, an alternative approach may be considered (Galpern et al., 2012). Under such circumstances, erroneous individualization due to ADO can be minimized by utilizing dedicated software, such as the allelematch package for R (Galpern et al., 2012).
Here, clusters of identical or similar genotypes potentially arising from repeatedly sampling single individuals are identified on basis of a pairwise dissimilarity matrix (Tables S6 and S7, Galpern et al., 2012). Such clusters help to control for genotyping error, as they may be considered a mimic of genotype alignments obtained by a multiple tubes approach. In our study, random allelic dropout has been identified as the major source of error. For different samples originating from the same individual, pairwise comparison of genotypes may yield mismatching allelic states at loci being actually heterozygous (e.g., pseudo-homozygous vs. actually heterozygous).
Allowing for such dropout-related pairwise mismatches, thus, will counteract this issue and help to reduce the number of phantom individuals. Nevertheless, to prevent assignment of a partial profile to multiple individuals, a maximum number of allowed pairwise mismatches have to be determined. This optimization process is aided by the allelematch software (Galpern et al., 2012). As the output from the allelematch software helps in pinpointing potentially erroneous STR profiles, it may be used in a "targeted multiple tubes" approach focusing on these questionable genotypes only.
False alleles were found to be much rarer than random allelic dropout events. Nevertheless, they constitute another form of genotyping error that may inflate the number of phantom individuals.
To counteract this, all multilocus genotypes pairwise differing by a single allele at a single locus only were challenged by re-amplification and re-analysis of the affected samples.
For data sets with a low expected proportion of multiple sampling and/or a high overall error rate, the "global" multiple tubes approach undoubtedly will be advantageous. Ultimately, depending on the design and aim of the study, it has to be evaluated, whether a global multiple tubes approach is required or if a software-based solution tied to target retyping of questionable samples appears sufficient.

| CON CLUS IONS
This study describes an optimized DNA extraction method for the molecular genetic analysis of noninvasive samples including feces and feathers. Importantly, the risk of sample cross-contamination and mix-up was minimized and the DNA extracts proved suitable for successful downstream genotyping applications. Multiplex PCR assays producing small amplicons from informative STR markers were developed. Additionally, advanced software solutions were used for analyzing genotyping data featuring missing information at some loci and a mild amount of false alleles. This approach fostered reliable genotyping of capercaillie and black grouse individuals. In addition to this, we are confident that our protocols will foster the use of noninvasively collected specimens in ecological and conservation studies, as they can be easily applied to a broad range of animal species.
We are also obliged to Michael Haupolter for modeling speciesspecific habitats. The authors thank the staff of the Alpenzoo Innsbruck (Austria), particularly Michael Martys, Matthias Seewald, and Dirk Ullrich for providing biological reference material of voucher animals. Anna König is greatly acknowledged for advice on DNA extraction from feather samples.
We are indebted to Werner Hecht, Peter de Knijff, and Gernot Segelbacher for valuable discussion.
This work was funded by the Provincial Government of Tyrol, department for hunting and fishing; therefore, the authors particularly wish to thank Franz Krösbacher. Additionally, we received financial support from Tyrolean Nature Protection Fund and Leopold-Franzens-University Innsbruck.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
SV, HN, BB, RL, and WP participated in study design. SV and RL collected the samples. SV and HN extracted DNA. SV, HN, BB, RL, and WP analyzed the data. SV, HN, and WP drafted the manuscript. All authors revised and added to the draft. All authors reviewed and approved the final manuscript.