A new fast real‐time PCR method for the identification of three sibling Apodemus species (A. sylvaticus, A. flavicollis, and A. alpicola) in Italy

Abstract The identification of field mice Apodemus flavicollis, Apodemus sylvaticus, and Apodemus alpicola represents a challenge for field scientists due to their highly overlapping morphological traits and habitats. Here, we propose a new fast real‐time PCR method to discriminate the three species by species‐specific TaqMan assays. Primers and probes were designed based on the alignment of 54 cyt‐b partial sequences from 25 different European countries retrieved from GenBank. TaqMan assays were then tested on 133 samples from three different areas of Italy. Real‐time PCR analysis showed 92 samples classified as A. flavicollis, 13 as A. sylvaticus, and 28 as A. alpicola. We did not observe any double amplification and DNA sequencing confirmed species assignment obtained by the TaqMan assays. The method is implementable on different matrices (ear tissues, tail, and blood). It can be used on dead specimens or on alive animals with minimally invasive sampling, and given the high sensitivity, the assay may be also suitable for degraded or low‐DNA samples. The method proved to work well to discriminate between the species analyzed. Furthermore, it gives clear results (amplified or not) and it does not require any postamplification handling of PCR product, reducing the time needed for the analyses and the risk of carryover contamination. It therefore represents a valuable tool for field ecologists, conservationists, and epidemiologists.

Apodemus sylvaticus is distributed throughout Europe (excluding Finland and northern parts of Scandinavia, the Baltics and Russia) and in some regions of North Africa (Schlitter et al., 2016). Apodemus flavicollis has a very similar distribution that extends northwards in southern Finland, in the Baltics, western Russia and some regions of Anatoly (whereas it is absent in Africa and Iceland) .
The physical resemblance of these two species is particularly marked in the southern part of their range, where discrimination based on the sole external characters such as body size and pelage color has proven to be unfeasible (Filippucci, Cristaldi, Tizi, & Contoli, 1984;Montgomery, 1980;Niethammer, 1978). Even though the two species have different ecologies, with A. flavicollis being more strictly associated to forested habitats and A. sylvaticus being also found in forest-edges, ecotones and in association with agricultural and anthropized environments, their ecological preferences partially overlap and the two species often live in syntopy (Marsh & Harris, 2000;Mitchell-Jones et al., 1999). The discrimination between these two siblings, albeit distinct species, is therefore very important for ecological, epidemiological, evolutionary, and B-chromosome studies. The situation is even more complex in the Alpine region, where a third very similar species, the Alpine field mouse (Apodemus alpicola) occurs in sympatry and often in syntopy with the other two.
The Alpine field mouse is distributed throughout the Alps (France, Switzerland, Germany, Italy, and Austria).
To date, different approaches have been developed and discussed to tackle the need of discriminating between Apodemus species, but the debate is still far from being solved. Cranial measurements have proven to be effective in many cases (e.g., Barčiová & Macholán, 2009;Debernardi et al. 2003;Jojić, Bugarski-Stanojević, Blagojević, & Vujošević, 2014;Reutter, Hausser, & Vogel, 1999), but their validity tends to be population-specific. Furthermore, they can only be used on dead animals or rests of them (e.g., on skulls found in owl pellets).
External morphological measures taken on dead or alive individuals, such as pelage color pattern, body size, length of tail, hear, hind foot, have also been used to identify species (e.g., Debernardi et al. 2003;Filippucci et al., 1984;Kuncová & Frynta, 2009;Niethammer, 1978), even though a high degree of uncertainty remains for individuals with overlapping characters (Bartolommei et al. 2016). Recently, an innovative bioacoustic approach, not needing animal sacrifice and carried out on awake animals, has been developed to help the classification of individuals based on distress calls emitted during handling (Ancillotto et al., 2017).
All the aforementioned methods, however, cannot provide a 100% classification success, as cranial, external, and bioacoustic traits present continuous gradients and animals with intermediate characters (including juveniles and sub-adults) cannot be classified correctly.
To date, the only classification techniques for Apodemus spp.
known to provide high rates of success rely on cytogenetic or molecular methods. Although the three species have similar karyotypes, Q/C banding has proved to be effective as a discrimination method (Engel et al., 1973;Hirning, Schulz, Just, Adolph, & Vogel, 1989), but it requires animal sacrifice. Molecular methods, including protein electrophoresis (Filippucci, 1992;Orlov, Bulatova, Nadjafova, & Kozlovsky, 1996;Vogel, Maddalena, Mabille, & Paquet, 1991) and DNA analysis (e.g., sequencing, PCR), can instead be applied on samples from alive animals (e.g., a small sample of ear or tail tissue). Michaux et al. (2001) described an assay based on a conventional PCR with species-specific primers targeting a fragment of cytochrome b mitochondrial gene (cyt-b). Given its simplicity and clarity of results (PCR product present or not), this method is one of the most commonly used in the literature, being preferred to other expensive methods such as DNA sequencing. However, in their recent work, Bugarski-Stanojević, Blagojević, Adnađević, Jovanović, and Vujošević (2013) showed that the method by Michaux et al. (2001) is subject to a certain degree of misidentification, probably due to the low specificity of the chosen primers or to the existence of nuclear copies of mitochondrial genes (pseudogenes) which determine false-positive results (Dubey, Michaux, Brunner, Hutterer, & Vogel, 2009). In their work, Bugarski-Stanojević et al.
(2013) compared the method with two alternative molecular assays with higher specificity: the arbitrarily primed-PCR (AP-PCR) and the intersimple sequence repeat-PCR (ISSR-PCR). Both methods result in species-specific DNA profiles that can be visualized through gel electrophoresis. Such methods, however, referring to arbitrary sequence PCR, generally exhibit poor interlaboratory reproducibility that hinders their widespread use. Moreover, postamplification handling of PCR products is required, increasing the time needed for the analyses and introducing the risk of carryover contamination.
Here, we propose a new fast real-time PCR method to distinguish between the three species. Real-time PCR using TaqMan probes has been reported by different authors as a fast and sensitive method for the identification of species (Cammà, Di Domenico, & Monaco, 2012;Di Domenico, Di Giuseppe, Wicochea Rodríguez, & Cammà, 2017;Overdyk, Braid, Naaum, Crawford, & Hanner, 2016). It does not require any postamplification step and can be easily automated allowing the analysis of large numbers of samples. Moreover, the application of specific primers in combination with fluorogenic probes considerably increases reaction specificity.

| Design of primers and probes
A total of 54 cyt-b partial sequences of A. flavicollis, A. sylvaticus, and A. alpicola from 25 different countries were retrieved from GenBank (Accession numbers in Table 1). Sequences were aligned with software MegAlign (DNASTAR Lasergene 10) and species-specific primers and TaqMan probes were designed based on differences between species (Table 2). Primer Express Software 3.0.1 (Applied Biosystems) was also used to exclude the presence of secondary structures between primers and probes that would reduce reaction efficiency.
TA B L E 1 Geographic origin, references and GenBank accession numbers of Apodemus cyt-b sequences used to design primers and probes for real-time PCR assays

| Samples and DNA extraction
We

| Sensitivity, specificity, and repeatability of the real-time PCR assays
The slope of the standard curve, the efficiency, and the correlation coefficient (R 2 ) of all three assays are reported in Table 3. The LOD calculated for the different species ranges between 10 and 100 pg as reported (Table 3). Moreover, the methods revealed a very high level of repeatability as assessed by the low values of the coefficient of variations as shown in the same Table 3.
For each assay, the other two species were tested as nontarget DNA, using the same amount, and no cross-amplifications were observed; human, M. musculus, Myodes glareolus, and Rattus norvegicus DNA did not produce fluorescent amplification signal in any of the three new developed assays.
In addition, in silico analyses confirmed the specificity also against other Apodemus species. In particular, primers and probe designed for the three different assays showed sequence identity against A. whiterbyi and A. uralensis lower than 92% even for the best records. An exception was A. sylvaticus probe against A. uralensis showing a 94% sequence identity; however, both forward and reverse primers mismatched in two nucleotides at the 3′ end, limiting the aspecific amplification of this nontarget species.

| Real-time PCR and DNA sequencing
Real-time PCR reactions obtained a 100% success of amplification.
Ninety-two samples were classified as A. flavicollis, 13 as A. sylvaticus and 28 as A. alpicola (Table 3). We did not obtain any double amplification (Table 3).
After sequencing, based on sequence quality and overlapping of the two strands, we selected a good central fragment of 677 bp (of 866 bp). For some samples, sequencing ended with poor-quality sequences that could not be used for the analysis or deposited. Other samples showed some clues of pseudogenes (i.e., double peaks) and also these sequences were excluded. We selected 33 good F I G U R E 1 Geographic origin of the 133 Italian Apodemus samples tested with real-time PCR for species discrimination. GP, Gran Paradiso National Park; VT, Viterbo area; TE, Teramo area sequences (about 25%) to be deposited in GenBank. All the 33 sequences were submitted on BLAST and confirmed the classifications obtained by TaqMan assays (Table 4). Accession numbers for deposited sequences are provided in the Data accessibility section.

| D ISCUSS I ON
In this work, we provided a new fast real-time PCR method for the discrimination of three Apodemus species. It can be used on dead specimens or on alive animals with minimally invasive sampling, a characteristic often required by ecology or conservation studies.
The three assays proved to be a useful tool on different matrices (tissues from ear, tail, and blood). Real-time PCR is characterized by a much higher sensitivity compared to conventional PCR, being able to detect even very low copies of DNA (Angelone-Alasaad et al., 2015). For this reason, this modern method is now commonly used to amplify small or degraded samples (Holt et al., 2016;Lee, McCord, & Buel, 2014) or for diagnostic purposes (Caraguel, Stryhn, Gagne′, Dohoo, & Hammell, 2011). Although we only tested our method on fresh or well-preserved samples, given its sensitivity, it is likely to work on degraded (e.g., museum specimen) or low-DNA samples obtained through noninvasive sampling, such as feces and hair (e.g., obtained through hair-tubes; Kanthaswamy, Premasuthan, Ng, Satkoski, & Goyal, 2012). Indeed, despite the DNA concentration used was 2 ng/μl, the assays show LODs three logs lower (2 pg/μl) then the input loaded in the test. The use of TaqMan probes considerably increases reaction specificity compared to other traditional methods (Kuboniwa et al., 2004). Accordingly, our assays proved to work well to discriminate between the species analyzed as we did not obtain any double amplification even in the presence of pseudogenes, a possible source of misclassification (Dubey et al., 2009). In silico analyses also suggest designed primers and probes to be species-specific also versus other nontarget Apodemus species (A. uralensis and A. witherbyi). Method efficacy is therefore comparable to or higher than other molecular assays such as those proposed We designed primers and probes based on sequences from sev- TA B L E 4 Results of real-time TaqMan assays for species discrimination and confirmation by DNA sequencing species, we did not directly test them and we cannot exclude that a certain level of cross-amplification could occur. In all these hypothetical cases, the comparison of C t values (not possible with traditional methods relying on visual interpretation of electrophoresis band patterns) may be a useful approach to identify correct species assignment even with doubtful amplifications.

ACK N OWLED G M ENTS
We are grateful to Sandro Bertolino (University of Turin), Emiliano Mori (University of Turin) and Leonardo Ancillotto (University of Naples Federico II) for providing samples from Gran Paradiso National Park. Samples from Viterbo area were collected during a previous project conducted by GS at Sapienza University of Rome.
Samples from Teramo were collected during project IZS AM 0412 RC founded by the Italian Ministry of Health.

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

DATA ACCE SS I B I LIT Y
DNA sequences: Genbank accession numbers KU975553-KU975564.

AUTH O R CO NTR I B UTI O N S
GS, CC, MDD, and IP conceived and designed the study; GS collected samples; GS, VC, and MDD performed analyses; GS and MDD wrote the manuscript; all the authors contributed to substantial manuscript improvement.