1 Environmental DNA in subterranean biology: range extension and taxonomic implications for Proteuspela ,*, David ,,,*,, Ale Snoj Kuntner, William R. Peter Trontelj, Milo Grizelj, Magdalena & Gregor Proteus anguinus Proteus Proteus Proteus Proteus e olm, Proteus anguinus Laurenti 1768, is a large amphibian endemic to subterranean waters of the Dinaric Karst, with a known range stretching between north-eastern Italy and southern Bosnia and Herzegovina. Despite the relatively broad expanse (60,000 km2) of karst topography in this region and decades of eld surveys, Proteus has only been documented at around 300 sites1. ese sites include caves where Proteus is recorded by visual observation or trapping, and springs where specimens may emerge during seasonal ooding5. While groundwater pollution and destruction of subterranean habitat are obvious threats to Proteus6 ,7, the negative anthropogenic impact cannot be determined without a reliable methodology to establish and monitor its presence. Even with advances in speleobiology, progress in dening the true geographic distribution and diversity of its populations has been slow8. For example, despite much speculation, no physical evidence of the presence of Proteus in the Dinaric Karst of Montenegro has been documented. Furthermore, as recently as in 1986, a unique, darkly pigmented non-troglomorphic population of Proteus was discovered in south-eastern Slovenia9 and described as the subspecies Proteus anguinus parkelj10. e results of subsequent morphometric analyses11 com) OPEN
2 supported its distinct taxonomic status, and phylogeographic analyses of mitochondrial DNA (mtDNA)14,15 con rmed the distinctiveness of its lineage, one, however, that is deeply nested in the Proteus phylogeny. e population of black Proteus has been documented at only four sites in an area of less than 2 km2 (refs 1, 7 and 16). In the same geographic region, but presumably in a dierent hydrogeological formation17,18, a closely related lineage14 of the troglomorphic, white Proteus subspecies (Proteus anguinus anguinus ) has also been recorded at nine sites1 ,9 (also A. Hudoklin, pers. comm. 20 July 2015). If two such morphs co-existed in the same local habitat without hybridizing, they would likely be reproductively isolated from each other by an intrinsic barrier. However, this simple and powerful test of species status is rarely available in obligate subterranean organisms, because their hab itat is patchy and their populations are usually fragmented and physically strongly isolated from each other19. To detect species like Proteus that are rare and dicult to observe with classical methods, detection of specic DNA released into the environment (environmental DNA or eDNA) is particularly useful22. e ubiquitous nature of DNA in aquatic environments and its rapid diusion from its source means that in theory the pres ence of a specic animal can be detected anywhere within the water body and not just at its point of origin28,29. Furthermore, as DNA released into most environments becomes quickly degraded, the eDNA approach detects the recent presence of target species30 without the need for direct observation or trapping. In this study we used Proteus individuals from the laboratory to develop a set of eDNA detection assays based on quantitative real-time polymerase chain reaction (qPCR) to examine the presence of Proteus in karst aquifers, where physical detection is dicult or impossible. First, we developed a SYBR green (Applied Biosystems) qPCR assay to search for Proteus eDNA in spring and cave water samples from the under-explored southernmost edge of the known range of Proteus in Herzegovina (southern Bosnia and Herzegovina) and outside of it in Montenegro. Second, we developed a TaqMan (Applied Biosystems) qPCR assay to discriminate the black Proteus eDNA from the white Proteus eDNA. Using this assay, we conducted a systematic inventory of Proteus in Bela Krajina (south-eastern Slovenia) to verify selected historic records of the white Proteus to determine the maximum range of the black Proteus and to test for possible co-occurrence of the two morphs.ResultsProteus eDNA detection by qPCR. e sample validation procedure for the observed outcomes of qPCR tests is illustrated in Fig.1. e Supplement lists the lower limit of detection and the conrmation of assay specicity for both SYBR and TaqMan qPCR assays. No false positives were observed. Out of 23 sites in Herzegovina examined for eDNA by the SYBR qPCR assay, four were veried to harbour Proteus (conrmed through visual encounter by a reliable informant). Out of these, two scored positive and one plausible for its eDNA while the fourth was negative. In Bela Krajina (Slovenia), only one veried site was included in the analyses by the TaqMan qPCR assay and it scored positive. Out of additional four likely sites (within the known range of the black Proteus ), three also were positive and one was negative. In our estimations during sampling, the ow of the springs where Proteus eDNA was detected varied from springs and wells with discharge rates as low as 0.1 L/s to as high as 2000 L/s (see SupplementaryTableS1). e maximum water temperature recorded during sampling was an exceptional 17 C, with the median at 11.7 C (see SupplementaryTableS1). Once sampled, DNA in water degraded within several days when stored at 4 C. By contrast, storage of dry lters at 20 C suciently preserved the integrity of the DNA for at least two months, and storage of isolated DNA at 20 C for at least four months.Detection of Proteus In Herzegovina (Fig.2), the springs Londa, Mua (no. 1 in SupplementaryTableS1), Nezdravica (no. 14) and Izvor Bregave (no. 20) showed weak signals (one positive signal in three replicates) representing only one of the genes and were negative in the re-run. Hence, these samples are categorised as uncertain for the presence of Proteus eDNA (possibly at the limit of detection, although we cannot completely exclude contamination despite the precaution mechanisms). A private well in Gornji Trebiat (no. 2) and Vrelo Vakuf (no. 4) showed a weak signal (one positive signal in three replicates) representing only the mitochondrial control region, and this signal (one positive signal in three replicates) was again observed in the re-run. erefore, these samples are categorised as plausible for containing Proteus eDNA. Here contamination is less likely as the signal was observed in independent runs. Peria Mlin (no. 3) and Kajtazovo Vrelo (no. 5) were positive for one gene, whereas the samples from Bunar kod Kue Mehe Dizdarevia (no. 10) and esma izpod Pogledovae (no. 11) were positive for both genes. us, these localities are interpreted as positive for the presence of Proteus eDNA. In Montenegro (Fig.3), the cave Sopot (no. 24 in SupplementaryTableS1) and the spring Izvor Grahovo 1 (no. 29) showed weak signals (one positive signal in three replicates) representing only one of the genes and were negative in the re-run. erefore, these signals are interpreted as uncertain for the presence of Proteus eDNA. e spring anik (no. 33) also showed a weak signal representing only the 16S rRNA gene (one positive out of three replicates), but in this case the same signal was observed again in three separate re-runs. erefore, this sample is interpreted as plausible to contain Proteus eDNA, and here contamination is less likely as the signal was observed in four independent runs. ProteusProteus eDNA was conrmed in six out of 19 samples analysed from Bela Krajina (Fig.4). Except for Otovski Breg (no. 45 in SupplementaryTableS1), a veried site with white Proteus all of the positive sites were springs along the Dobliica River. Following the direction of the Dobliica River ow, there was a gradient of relative concentration of Proteus eDNA in these springs, as deduced from cycle threshold (Ct) values in combination with the fraction of positive replicates in each sample. In the south-to-north and west-to-east directions, these fractions were as follows: Izvir ob Dobliici BK D3 (no. 51) 6/6, Izvir ob Dobliici BK D4 (no. 52) 4/6, Izvir ob Izlivu Jelevnice v Dobliico BK A2 (no. 47) and prajcarjev Zdenec (no. 41) 6/6, Izvir v Svibniku (no. 54) 2/6 2/3, Planinec (no.
3 55) 1/6 0/3. Outside the immediate area of the Dobliica River, the Otovski Breg (no. 45) sample was positive in 5/6 reactions, and Izvir Obrice (no. 40) in only 1/6 0/3 reactions. e latter and Planinec were not analysed further as the presence of Proteus eDNA in these springs was uncertain. All other analysed samples were negative for the presence of Proteus eDNA. Following our analyses, on the evening of November 1, 2016, a young black Proteus was observed by the last two co-authors in Planinec. Once Proteus eDNA was conrmed in a sample, it was tested further to determine if the eDNA belonged to the black or the white Proteus morph (Fig.5). We thus detected black Proteus eDNA in ve samples, all taken in springs along the Dobliica River, while the Otovski Breg (no. 45) sample was negative for black Proteus eDNA (0/6 reactions). e fraction of positive reactions again followed a northward and eastward gradient: Izvir ob Dobliici BK D3 (no. 51) 6/6, Izvir ob Dobliici BK D4 (no. 52) 3/6, Izvir ob Izlivu Jelevnice v Dobliico BK A2 (no. 47) 6/6, prajcarjev Zdenec (no. 41) 3/6, Izvir v Svibniku (no. 54) 2/6. We next analysed the same six samples for white Proteus eDNA. As expected, white Proteus eDNA was conrmed in the Otovski Breg sample (no. 45; 3/6), a known white Proteus site. Importantly, however, white Proteus eDNA was found in prajcarjev Zdenec (no. 41; 2/6 3/3), which was also positive for black Proteus eDNA. All other samples that were positive for black Proteus eDNA were negative for white Proteus eDNA. Sequencing of the PCR products amplied in the sample from prajcarjev Zdenec conrmed that both the black and the white haplotype were present in the sample. Figure 1. Evaluation of Proteus eDNA presence in a sample from the observed outcome of (a) SYBR qPCR assay and (b) TaqMan qPCR assay. Two mitochondrial DNA (mtDNA) regions (genes) were searched for in (a) and one mtDNA region was searched for in (b). e rst run in both (a) and (b) included two concentrations (dilutions) of the template (data were pooled in b). All assays were performed in three parallel reactions (wells). See also the Methods section.
4 Figure 2. Map of sampling sites and results of eDNA analyses in Bosnia and Herzegovina. Categorisation of sites is explained in Fig.1a and in the Methods section. e map was created using ArcGIS Desktop 10.3.1 (Esri 2015). Basemap used: World Street Map (Esri 2015). Source of hydrogeological data: http://diktas.iwlearn.org/ im/hydrogeological-map-of-the-dinaric-karst (last accessed 5 October 2016). Figure 3. Map of sampling sites and results of eDNA analyses in Montenegro. Categorisation of sites is explained in Fig.1a and in the Methods section. e map was created using ArcGIS Desktop 10.3.1 (Esri 2015). Basemap used: World Street Map (Esri 2015). Source of hydrogeological data: http://diktas.iwlearn.org/im/ hydrogeological-map-of-the-dinaric-karst (last accessed 5 October 2016).
5 DiscussionBelow we evaluate the basic parameters of our eDNA assay and its eectiveness in detecting Proteus in the laboratory as well as in dierent subterranean habitats. A suitable eDNA assay for Proteus must be able to detect trace amounts of highly diluted DNA released by potentially very small populations. e qPCR technique is commonly used in achieving this goal22,31. When com pared to classical PCR in combination with cloning and sequencing27, the real-time qPCR approach signicantly improves the eciency of eDNA detection and reduces the possibility of contamination as post-PCR analysis is omitted. Our tests showed that this method is suitable for detection of Proteus eDNA, including under the conditions encountered at karst springs. A compromise between at least three factors is necessary for an optimal use of the eDNA assay for Proteus : (1) the time of sampling, (2) the amount of water ltered and (3) the lower detection limit of the method. e discharge from karst springs typically varies signicantly through seasons in response to precipitation in the catchment area, with many springs inactive during the dry season32,33. As a consequence, sampling when water levels are optimal may be a challenge. When water levels are very high, eDNA may become too diluted or dispersed for detection. e latter may have been the case in a few springs in Herzegovina, which therefore required a repeated sampling to detect the presence of Proteus eDNA. Another concern when sampling for Proteus eDNA is collecting water samples without disturbing the sediment at very low water levels, while higher water levels may increase sediment transport through the karst aquifer. Even though sediment potentially holds more eDNA34,35, it also prevents ecient ltration. Furthermore, sediment can be a source of PCR inhibitors36,37. Monitoring PCR inhibition with the addition of synthetic DNA and corresponding primers and probes to each reaction, and diluting template DNA when inhibition is detected, is essential when analysing environmental samples for eDNA. Exposed DNA in water gradually decays predominantly due to the eects of UV-light, heat and decomposition by microorganisms31,38,39. Because of the absence of light and due to the relatively low and constant temperatures of groundwater inhabited by Proteus (SupplementaryTableS1; see also ref. 1), we expect eDNA to be Figure 4. Map of sampling sites and results of eDNA analyses for Proteus in Bela Krajina (Slovenia). Circles and squares depict the results of eDNA analyses; stars and triangles represent known localities of the black or the white Proteus respectively. Categorisation of sites is explained in Fig.1b and in the Methods section. Doblika Gora and Poljanska Gora are the south-eastern foothills of the Koevski Rog Plateau. e map was created using ArcGIS Desktop 10.3.1 (Esri 2015). Source of basemap: digital elevation model at 1:10,000 (http:// www.e-prostor.gov.si/si/zbirke_prostorskih_podatkov/topografski_in_kartografski_podatki/digitalni_model_ visin/digitalni_model_visin_5_x_5_m_dmv_5/, last accessed 5 October 2016).
6 relatively stable, which presumably facilitates detection. Furthermore, the eDNA fragments targeted in our assays are 100 bp long, i.e. short enough to persist in the environment40 despite degradation processes. As factors that aect DNA degradation may be less detrimental in subterranean streams, eDNA transport distance is likely increased in this habitat. Since eDNA transport in streams is greatly aected by discharge rates41,42, we timed the sampling, whenever and wherever it was possible, to the lowest water level and discharge rates that still allowed for ecient sampling. e mitochondrial control region and anking DNA and the 16S rRNA gene were chosen for our eDNA assay because a large set of sequences in Proteus is available for comparison and primer/probe design (see SupplementaryInformation). We believe this approach minimises the risk of not detecting a Proteus population due to an unknown variation in sequence. is especially applies to the primer set designed to bind to the conserved 16S rRNA gene, which appears to be general enough to be used in detection of any population within as well as outside the known range of Proteus e primer pair designed to amplify a fragment of the control region, on the other hand, targets a more variable region of the mitochondrial genome. erefore, the possibility that it could not bind eciently to the hypothetical Montenegro population, which may be genetically distinct from all other known populations, cannot be excluded. Compared to classical approaches of visual surveying or trapping, the eDNA analysis substantially improves our ability to detect Proteus in groundwater. is is most clearly observable in the results of the Bela Krajina sur vey, where the number of known sites with the black Proteus more than doubled aer a single sampling. Under optimal water-level conditions, the sampling protocol developed here for Proteus eDNA is expected to yield at least a 75% detection rate at densities of at least one animal per 256 m3 of water (see SupplementaryInformation). Although the number of sites used to validate the detection probability and accuracy of the method was low, Figure 5. (a) Distribution of eDNA specic for (b) black and (c) white Proteus in spring samples in Bela Krajina (Slovenia). Circles and squares depict the results of eDNA analyses; stars and triangles represent known localities of the black or the white Proteus respectively. e map was created using ArcGIS Desktop 10.3.1 (Esri 2015). Source of basemap: digital elevation model at 1:10,000 (http://www.e-prostor.gov.si/si/ zbirke_prostorskih_podatkov/topografski_in_kartografski_podatki/digitalni_model_visin/digitalni_model_ visin_5_x_5_m_dmv_5/, last accessed 5 October 2016). Sources of hydrogeological and geological layers: refs 17, 18 and 51.
7 laboratory tests (see SupplementaryInformation) and hydrogeological data strongly support the results obtained for springs with hitherto unknown status. Similar detection probabilities were reported for epigean aquatic ver tebrate species (e.g. refs 25, 42 and 43). Testing the usefulness of the eDNA assay in field research in Herzegovina resulted in the detection of Proteus at locations where it has not been previously recorded. In the most recent list of localities in Bosnia and Herzegovina3, only seven out of a total of 57 are known in the greater Trebiat River area. We discovered four new Proteus localities by the eDNA method and two very likely to harbour Proteus Following our analyses, the presence of Proteus was visually confirmed by cave diving at the spring Kajtazovo Vrelo (no. 5 in SupplementaryTableS1; Z. Vlaho, pers. comm. 29 June 2016). Here we also report the evidence for the existence of Proteus in Montenegro, based on a plausible sample of Proteus eDNA recorded at the spring anik (no. 33 in in SupplementaryTableS1). anik is located 1.7 km from the nearest Proteus locality in Herzegovina (B. Lewarne, pers. comm. 22 June 2014). e proximity of the two sites suggests that they may be hydrogeologically connected and therefore may share a common Proteus population. Alternatively, they may depend on the same catchment area, which could include an upstream Proteus locality potentially serving only as a donor site for eDNA inux into anik. Since no hydrogeological surveys have been conducted in the region, neither a connection of anik to the locality in Herzegovina nor the location of its catchment area has been determined. Nonetheless, preliminary results suggest that the two cave systems could be hydrogeologically isolated from each other (B. Lewarne, pers. comm. 2 October 2016). Our results therefore favour the possibility that Proteus individuals are actually present at the sampled location and that the southern range of Proteus extends into the Dinaric Karst of Montenegro. Next, using the eDNA assay, we report the discovery of new sites that may harbour the rare black Proteus while its presence was visually conrmed in yet another one, which showed a faint trace of Proteus eDNA. ree of these springs lie outside the limits of its known range and represent an extension of its presumed range north-eastward, along with the general direction of the ow of the Dobliica River. e distance of the new easternmost site, the spring Planinec (no. 55 in SupplementaryTableS1), from the nearest previously visually conrmed site, Kaniarica16, is 1.2 km (the actual extent of the cave system is unknown). Since we conrmed by visual observation that the occurrence of Proteus eDNA in a spring is indicative of its actual presence in close proximity, the present knowledge connes the black Proteus between the high plateau Koevski Rog which is lacking surface streams in the west, along the Dobliica River to the conuence with the stream Paki Potok in the east. As in most aquatic cave animals, ranges of Proteus populations are probably historically determined and highly restricted by the boundaries of present-day subterranean hydrogeological networks15. e white Proteus population appears to occupy a larger range, including sites in the Dolenjska region west of Koevski Rog14. It should be noted, however, that while the maximum span of the black Proteus range to the east, north and south was established in the present study, the extent of its distribution to the west remains unknown. e observed descending gradient in concentration of eDNA following the Dobliica main stream ow appears to reect a complex local network of underground connections that could receive inow through both southwest-to-northeast and northwest-to southeast oriented faults as proven also by the water tracings made in the region17,18 (see Fig.5). erefore, the core of the black Proteus distribution probably includes the contact zone between the Koevski Rog Plateau and the Bela Krajina Plain as well as the south-eastern parts of Koevski Rog. is conclusion is con sistent with earlier predictions, when only one9 or two sites10 were known. e discovery of the eDNA of both black and white Proteus syntopically in the spring prajcarjev Zdenec (no. 41 in SupplementaryTableS1) represents direct evidence suggesting that these two populations may be in contact with each other. e nding is strongly supported by an existing (intermittent) hydrogeological connection17 with Otovski Breg, a nearby site occupied by the white Proteus (no. 45 in SupplementaryTableS1). Because the relative concentration of eDNA of the white Proteus in prajcarjev Zdenec was similar to the relative concentration detected at Otovski Breg (sampled at around the same date), which cannot be explained by the dierence in discharge rates of the two springs alone, we believe that it reects the actual presence of the white Proteus in prajcarjev Zdenec. Passive eDNA dispersal would likely result in the presence of white Proteus eDNA in the spring closest to prajcarjev Zdenec (Izvir v Svibniku; no. 54 in SupplementaryTableS1), irrespective of a later sampling date and alongside black Proteus eDNA found there. As our analysis showed the relative concentration of white Proteus eDNA in prajcarjev Zdenec to be similar to the relative concentration of black Proteus eDNA in this same spring, but at the same time we failed to detect white Proteus eDNA in nearby Izvir v Svibniku, we expect repeated additional sampling would support this conclusion. Rare cases of subterranean syntopic occurrence of closely related lineages are valuable for the study of the poorly understood mechanisms of speciation and dierentiation within the subterranean realm (e.g. ref. 44). e distribution of the black and white Proteus eDNA in Bela Krajina is in agreement with the existence of a potential reproductive barrier between these two lineages, at least regarding female mating preferences. Assuming gene ow between the two lineages, substantial introgression of the white Proteus mtDNA would predict the presence of white Proteus eDNA in the springs to the west of prajcarjev Zdenec, which was not detected. Similarly, if inter-lineage mating regularly occurred in the other direction, we would expect black Proteus eDNA to appear in the sample at Otovski Breg, a site which was found to harbour only white Proteus eDNA. Signicantly, despite a low degree of sequence divergence between the two populations observed in the mitochondrial control region14,15,45, none of the comparative studies to date have detected any signs of their interbreeding, e.g. haplotype sharing14,15 or intermediate morphology10,12,13 ,4648. In combination with these observations, eDNA data suggest that the two populations may represent independent species, but additional analyses are needed to resolve the taxonomic status of the present as well as of other apparently monophyletic groups of Proteus In conclusion, we have demonstrated that the qPCR-based eDNA method can be utilised for a rapid detection of a rare subterranean species inhabiting karst groundwater. Due to its high sensitivity (see SupplementaryInformation) and general applicability, the SYBR qPCR-based eDNA assay is appropriate for
8 large-scale inventories of Proteus in groundwater throughout the Dinaric Karst, while the high specicity of the TaqMan qPCR-based eDNA assay makes this approach suitable for monitoring the distribution of closely related populations or taxa. Furthermore, as suggested by the results of the Bela Krajina survey in particular, the qPCR-based approach enables us to assess the relative abundance of Proteus eDNA in groundwater over a small spatial scale. Finally, we have shown that the eDNA approach can also be helpful in identifying potentially sympatric populations in the cryptic subterranean environment and therefore can be useful in the study of evolutionary history and taxonomy of subterranean taxa.Study Design. Development of our methodology to detect traces of Proteus eDNA in water involved the following steps (see SupplementaryInformation): (1) development of specic oligonucleotides for eDNA detection with qPCR, (2) testing the specicity of the oligonucleotides on tissue samples, (3) testing the lower detection limit of the method in laboratory conditions and (4) testing the performance of the method in nature, at three ver ied sites in Slovenia (SYBR qPCR only). Two broad geographic regions in the south-eastern part of the Dinaric Karst were then investigated for the presence and distribution of Proteus using the SYBR qPCR assay, while the distribution of two morphs of Proteus in south-eastern Slovenia was surveyed using the TaqMan qPCR assay. Our eDNA methodology is in line with general recommendations for eDNA sampling, analysis and reporting49.At eld sampling sites, 10 (exceptionally) to 20 L of water were collected taking care not to disturb the sediment during sampling. Samples were collected in brand-new 5or 10-L plastic canisters and stored in a dark cool room until ltration. Most samples were ltered within 24 hours aer collection. Water samples were ltered through sterile 0.45 m PES membrane lters (Sterlitech or Whatman) mounted on Nalgene polysulfone reusable bottle top lter holders (47 mm diameter), or through Nalgene MF75 series disposable bottle top lters with integral 0.45 m SFCA membrane (ermo Scientic) using a vacuum aspirator pump. Up to four lter membranes were used per sample, depending on the degree of clogging by sediment and other particles in the water. Aer ltration, the lters were rolled up using sterile disposable forceps, put into 5 ml tubes provided in the PowerWater DNA isolation kit (MoBio Laboratories/Qiagen) and stored at 20 C until DNA extraction. DNA was extracted following the kit manufacturers instructions, except for a minor adjustment to concentrate the eluted DNA: the nal elution volume was 50 l for 20-L samples and 30 l for 10-L samples.DNA amplification. SYBR chemistry eDNA assay. Two mtDNA regions (control region and 16S rRNA gene), were chosen to explore the presence of Proteus at the southernmost edge of its range (see SupplementaryInformation for details). A 106-bp fragment of the former and a 153-bp fragment of the latter were PCR-amplied using primer sets PangCRF (5 -GCGTTAATTACAAGGTGCACTTGG-3 ), PangCRR (5 -TGTACCAGGTATTACCTTTAATGTTGG-3 ), Pang16SF (5 -CTGCCTGCCCAGTGACAACA-3 ) and Pang16SR (5 -CACGAGGAGATCAATTTCGCAGA-3 ). Before PCR amplication, distilled water dilutions of each eDNA sample were prepared in ratios of 1/4 and 1/16. A reaction mix of 15 l total volume, which was applied to both target fragments, contained 7.5 l of 2X SYBR Green Real-Time PCR Master Mix (Applied Biosystems), 0.15 l of each of 100 M primers, 1.2 l of sterile PCR-grade water and 6 l of eDNA sample. All DNA amplications were performed on ViiA 7 Real-Time PCR System (Applied Biosystems) under the default thermo-cycling conditions for the Hold and Melt Curve Stages, while the PCR Stage involved 40 cycles with a 15-s denaturation step at 95 C and a 45-s annealing step at 62.5 C. For each template dilution, both target fragments were amplied in triplicate using separate plates for each primer pair. A single 384-well qPCR plate contained between four to 12 samples, so that individual samples were separated by at least one empty row and column. In addition, each plate included six negative controls (double distilled and tap water from outside of Proteus range) and two positive controls (tissue DNA and eDNA extracted from the laboratory water tanks), for a comparison of melting curves. Samples were scored positive for Proteus eDNA if at least two out of three replicate wells of at least one combination (dilution-primer pair) were positive by qPCR (see Fig.1A). If only one of the three wells was positive by qPCR, the sample was re-analysed using the same primer combination. Again, if at least two of the three replicate wells were positive in the re-run, the sample was scored positive for Proteus eDNA. If only one of the three wells was positive in the re-run, the sample was considered plausible to contain Proteus eDNA. If all wells were negative in the re-run, the presence of Proteus eDNA in the sample was considered uncertain. Finally, samples with all negative wells in the rst run were scored negative for Proteus eDNA and were not analysed further. TaqMan chemistry eDNA assay. ree primer-probe combinations were designed to rst test for the presence of Proteus eDNA in each sample and subsequently to determine whether the eDNA was characteristic for the black or the white Proteus (see SupplementaryInformation for details). To detect Proteus eDNA, we used the primer pair Pa16SF (5 -TACTGCCTGCCCAGTGACAA-3 ) and Pa16SR (5 -TGCACGAGGAGATCAATTTCG-3 ), which amplified 157 bp of the 16S rRNA gene, and a FAM-labelled TaqMan-MGB probe PROTEUS (5 -TTACGCTACCTTTGCACG-3 ) that binds to Proteus -specic complementary region within the amplicon. To recognise the black Proteus eDNA, we used the primer pair BPaCytbF (5 -CATCCTACTGACATGGATCGGA-3 ) and BPaCytbR (5 -GGCAGAGGTCTAGGAGTTTGTTTTC-3 ), which amplified 146 bp of the mitochondrial cytochrome b (cytb ) gene, and the TaqMan-MGB probe BLACK (5 -CATAATCCCATCAGCCGGA-3 ). The probe and forward primer contained two black Proteus -specific nucleotides at positions 2 & 7, and 5 & 13, respectively. To recognise the white Proteus eDNA, we used the primer pair WPaCytbF (5 -CAGATGCCATCGTACTGACCTG-3 ) and WPaCytbR (5 -TAGGAGTTTGTTTTCAGCCCATC-3 ), which
9 amplied 143 bp of the cytb sequence, and the TaqMan-MGB probe WHITE (5 -ATCGCCCTAATTCCATC-3 ). e probe and forward primer contained two white Proteus -specic nucleotides at positions 7 & 12, and 12 & 20, respectively. As all three probes were FAM-labelled, the three assays were performed in separate reactions. Each sample was tested undiluted and as a 1/4 dilution to minimise the eect of PCR inhibitors. Each probe-template com bination was run in a triplicate. All reactions included a synthetic control DNA, corresponding primers and a VIC-labelled probe (all part of the TaqMan Mutation Detection IPC Reagent Kit, Applied Biosysytems) to either detect the presence of PCR inhibitors or conrm that the assay was carried out. Each 384-well test plate included at least three wells of negative control (double distilled water, assay-non-specic Proteus tissue DNA in concentration of 10 pg or higher) and a positive control (assay-specic Proteus DNA in concentrations of 10 and/or 100 pg or higher). Individual eld samples were separated by at least two empty rows and columns. Samples that scored positive or plausible (see below) for Proteus eDNA were then tested for black and white Proteus eDNA on separate plates. For all samples and assays we used 10 l reaction mixtures containing 5 l of 2X TaqMan Environmental Master Mix 2.0 (Applied Biosystems), 0.5 l of each of 10 M primers, 0.5 l of 2.5 M FAM-labelled probe, 1 l of 10X TaqMan Mutation Detection IPC Reagent (Applied Biosystems), 0.2 l of 50X TaqMan Mutation Detection IPC Control DNA (Applied Biosystems) and 2.3 l of sample. qPCR reactions were performed on the ViiA 7 System (Applied Biosystems) under default conditions (1 min annealing at 60 C), except for an increase of the number of cycles to 60. If a specic product was observed in at least three out of six reactions containing either undiluted or diluted sample, the sample was scored positive for the presence of respective eDNA (see Fig.1B). On the other hand, samples with all negative wells were scored negative and were not analysed further. If, however, the specic product was observed in only one or two reactions with either undiluted or diluted sample, the assay was re-run in three replicates of 15 l reactions with the same proportions of reagents as above and 3.5 l of undiluted sample. If at least two out of three replicates were positive in the repeat, the presence of the respective eDNA marker in the sample was considered plausible. If none or one of the replicates were positive in the re-run, the presence of the respective eDNA marker in the sample was considered uncertain.Environmental DNA detectability assessment. e minimal density of Proteus in water at which its eDNA can still be detected (i.e. the lower detection limit of the SYBR and TaqMan qPCR assays) was determined from the animals hosted in controlled conditions as described in the Supplement. e approval for maintenance of live Proteus individuals was granted to the Laboratory by the Ministry of Environment and Spatial Planning of the Republic of Slovenia, Slovenian Environment Agency (Permit no. 35601-95/2009-4).Field survey. e authorities of all political entities visited were informed of the purpose of our eldwork and granted its approval. Trebiat River and Hutovo Blato areas in southern Bosnia and Herzegovina. Samples were collected in three rounds: March 31 April 8, 2014, April 27 May 10, 2014 and June 19, 2014. A total of 38 sites were visited (karst springs, caves and wells), of which 23 samples were analysed (see SupplementaryTableS1). Because of sample transportation delays during the rst eld trip, several sites were visited twice. Dinaric Karst in Montenegro. A wide area of the Dinaric Karst in Montenegro was sampled, including the Niki region, Skadar Lake region and the springs in Boka Kotorska Bay, Grahovo Polje and the territory of Banjan. Sampling in Montenegro was also organised in three rounds: October 3, 2013, November 16, 2013 and June 5, 2014. In total, 15 localities were visited, 11 of which were analysed (see SupplementaryTableS1). Bela Krajina, south-eastern Slovenia. Between July 20, 2015, 36 sites were visited, 13 of which were sampled (karst springs and a cave; see SupplementaryTableS1). We also sampled tap water in Dragatu, which is pumped from the groundwater source Dobliica, a known black Proteus site. Aer some precipitation, on November 2, 2015, ve additional springs were sampled. Sample collection. e sampling sites were selected on the basis of both published lists1 ,3 and unpublished sources of information on putative Proteus presence (reports in local media, interviews with local residents) as well as available information on hydrogeological connectivity to known localities17 ,18 ,50 (also http://diktas.iwlearn. org/im/hydrogeological-map-of-the-dinaric-karst) and, ultimately, hydrological conditions encountered during our visit. Samples were taken at low to medium water levels and during the lowest to average annual discharge rates of individual springs. Field samples were collected and ltered as randomly as possible, i.e. samples collected and ltered on the same day were never from adjacent sources. e investigators performing the qPCR assays were blinded with respect to detailed hydrogeological infor mation pertaining to individual samples. Rigorous controls for preventing and monitoring contamination were employed throughout the entire procedure (see SupplementaryInformation for details). e Supplement also provides the methods for GIS database construction and mapping of hydrogeological and geological data and information on Proteus sites.References1. Set, B. Distribution of Proteus (Amphibia: Urodela: Proteidae) and its possible explanation. J. Biogeogr. 24, 263 (1997). 2. letei, E., Jali B. & aa, T. Distribution of the Olm (Proteus anguinus Laur.) in Croatia. Mem. Biospeol. 23, 227 (1996). 3. otroan, D. asprostranjenje ovjeije ribice (Proteus anguinus Laurenti, 1768) na podruju Bosne i Hercegovine. [Distribution of the olm (Proteus anguinus Laurenti, 1768) in Bosnia and Herzegovina]. Na r 22, 57 (2002).
10 4. Bressi, N. Underground and unnown: updated distribution, ecological notes and conservation guidelines on the Olm Proteus anguinus anguinus in Italy (Amphibia, Proteidae). Ital J. Zool (Modena ) 71, 55 (2004). 5. Aljani, G., Aljani, M. & Golob, Z. Salvaging the washed-out Proteus Natura Sloveniae 18, 65 (2016). 6. Bulog, B., Mihajl, ., Jeran, Z. & Toman, M. Trace element concentrations in the tissues of Proteus anguinus (Amphibia, Caudata) and the surrounding environment. Water Air Soil Pollut. 136, 147 (2002). 7. Hudolin, A. Are we guaranteeing the favourable status of the Proteus anguinus in the Natura 2000 networ in Slovenia? In Pressures and Protection of the Underground arst Cases from Slovenia and Croatia (eds Prelove, A. & Zupan Hajna, N.) 169 (arst esearch institute ZC SAZU, Postojna, Slovenia, 2011). 8. Aljani, G., Gorii, ., Npru, M., Stanovi, D. & untner, M. Endangered Proteus: combining DNA and GIS analyses for its conservation In Dinaric arst Poljes Floods for Life (eds Sacl, P., Durst, ., otroan, D. & Stumberger, D.) 70 (EuroNatur, adolfzell, Germany, 2014). 9. Aljani, M., Habi, P. & Mihevc, A. rni moeril iz Bele rajine/e blac Olm from White Carniola. Nae jame 28, 39 (1986). 10. Set, B. & Arntzen, J. W. A blac, non-troglomorphic amphibian from the arst of Slovenia: Proteus anguinus parelj n. ssp. (Urodela: Proteidae). Bijdragen tot de Dierunde/Contrib Zool 64, 33 (1994). 11. Grillitsch, H. & Tiedemann, F. Die Grottenolm-Typen Leopold Fitzingers (Caudata: Proteidae: Proteus ). [e olm types of Leopold Fitzinger (Caudata: Proteidae: Proteus )]. Herpetozoa 7, 139 (1994). 12. Arntzen, J. W. & Set, B. Spea of the devil: the taxonomic status of Proteus anguinus parelj revisited (Caudata: Proteidae). Herpetozoa 8, 165 (1996). 13. Arntzen, J. W. & Set, B. Morphometric analysis of blac and white European cave salamanders, Proteus anguinus J Zool 241, 699 (1997). 14. Gorii, & Trontelj, P. Structure and evolution of the mitochondrial control region and aning sequences in the European cave salamander Proteus anguinus Gene 378, 31 (2006). 15. Trontelj, P. et al. A molecular test for cryptic diversity in ground water: how large are the ranges of macro-stygobionts? Freshw Biol 54, 727 (2009). 16. Ivanovi, M. Novo odritje tretje loacije habitata rnega moerila v Beli rajini. [A new discovery of the third location of the blac olm habitat in Bela rajina]. N-vestni 34, 1 (2012). 17. Habi, P., ogove, J., Bricelj, M. & Zupan, M. Izviri Dobliice in njihovo ire rao zaledje/Dobliica springs and their wider arst bacground. Acta Carsol 19, 5 (1990). 18. Nova, D. Padavinso zaledje izvira Jelevnice/echarge area of the resurgences of the Jelevnica Stream. Nae jame 38, 105 (1996). 19. Guzi, M. T. et al. Evidence for population fragmentation within a subterranean aquatic habitat in the Western Australian desert. Heredity 107, 215 (2011). 20. Niemiller, M. L. et al. Eects of climatic and geological processes during the pleistocene on the evolutionary history of the northern cavesh, Amblyopsis spelaea (Teleostei: Amblyopsidae). Evolution 67, 1011 (2013). 21. Asmyhr, M. G., Hose, G., Graham, P. & Stow, A. J. Fine-scale genetics of subterranean syncarids. Freshw Biol 59, 1 (2014). 22. ees, H. C., Maddison, B. C., Middleditch, D. J., Patmore, H. M. & Gough, C. e detection of aquatic animal species using environmental DNA a review of eDNA as a survey tool in ecology. J. Appl Ecol 51, 1450 (2014). 23. Ficetola, G. F., Miaud, C., Pompanon, F. & Taberlet, P. Species detection using environmental DNA from water samples. Biol Lett 4, 423 (2008). 24. Goldberg, C. S., Pilliod, D. S., Arle, S. & Waits, L. P. Molecular detection of vertebrates in stream water: a demonstration using ocy Mountain tailed frogs and Idaho giant salamanders. PLoS One 6, e22746, doi: 10.1371/journal.pone.0022746 (2011). 25. omsen, P. F. et al. Monitoring endangered freshwater biodiversity using environmental DNA. Mol Ecol 21, 2565 (2012). 26. Vences, M. et al. Freshwater vertebrate metabarcoding on illumina platforms using double-indexed primers of the mitochondrial 16S rNA gene, Conserv. Genet esour ., doi: 10.1007/s12686-016-0550-y (2016). 27. Vrs, J., Mrton, O., Schmidt, B. ., Gl, J. T. & Jeli, D. Surveying Europes only cave-dwelling chordate species (Proteus anguinus ) using environmental DNA. PLoS One 6, e0170945, doi: 10.1371/journal.pone.0170945 (2017). 28. Pilliod, D. S., Goldberg, C. S., Arle, S. & Waits, L. P. Factors inuencing detection of eDNA from a stream-dwelling amphibian. Mol Ecol esour 14, 109 (2014). 29. Deiner, ., Fronhofer, E. A., Mchler, E. & Altermatt, F. Environmental DNA reveals that rivers are conveyer belts of biodiversity information. Nat Commun 7, 12544, doi: 10.1038/ncomms12544 (2016). 30. Barnes, M. A. et al. Environmental conditions inuence eDNA persistence in aquatic systems. Environ Sci. Technol 48, 1819 (2014). 31. Bohmann, et al. Environmental DNA for wildlife biology and biodiversity monitoring. Trends Ecol. Evol 29, 358 (2014). 32. Bonacci, O. Analysis of the maximum discharge of arst springs. Hydrogeol. J. 9, 328 (2001). 33. Stevanovi, Z. Characterization of arst aquifer In arst Aquifers Characterization and Engineering (ed. Stevanovi, Z.) 47125 (Springer International, Switzerland, 2015). 34. Turner, C. ., Uya, L. & Everhart, C. Fish environmental DNA is more concentrated in aquatic sediments than surface water. Biol Conserv. 183, 93 (2015). 35. Jerde, L. J. et al. Inuence of stream bottom substrate on retention and transport of vertebrate environmental DNA. Environ. Sci. Technol. 50, 8770 (2016). 36. oc, C., Alum, A. & Abbaszadegan, M. PC inhibitor levels in concentrates of biosolid samples predicted by a new method based on excitation-emission matrix spectroscopy. Appl Environ Microbiol 76, 8102 (2010). 37. Schrader, C., Schiele, A., Ellerbroe, L. & Johne, PC inhibitors occurrence, properties and removal. J Appl Microbiol 113, 1014 (2012). 38. Dejean, T. et al. Persistence of environmental DNA in freshwater ecosystems. PLoS ONE 6, e23398, doi: 10.1371/journal. pone.0023398 (2011). 39. Stricler, M., Fremier, A. & Goldberg, C. S. Quantifying eects of UV-B, temperature, and pH on eDNA degradation in aquatic microcosms. Biol Conserv. 183, 85 (2015). 40. Deagle, B. E., Eveson, J. P. & Jarman, S. N. Quantication of damage in DNA recovered from highly degraded samples a case study on DNA in faeces. Front Zool 3, 11, doi: 10.1186/1742-9994-3-11 (2006). 41. Jane, S. F. et al. Distance, ow and PC inhibition: eDNA dynamics in two headwater streams. Mol Ecol esour 15, 216 (2015). 42. Wilcox, T. M. Understanding environmental DNA detection probabilities: a case study using a stream-dwelling char Salvelinus fontinalis Biol Conserv. 194, 209 (2016). 43. Piggott, M. P. Evaluating the eects of laboratory protocols on eDNA detection probability for an endangered freshwater sh. Ecol Evol. 6, 2739 (2016). 44. Deli, T., Trontelj, P., Zae, V. & Fier, C. Biotic and abiotic determinants of appendage length evolution in a cave amphipod. J Zool 299, 42 (2016). 45. Trontelj, P. et al. Age estimates for some subterranean taxa and lineages in the Dinaric arst. Acta Carsol 36, 183 (2007). 46. os, M., Bulog, B., Szl, A. & hlich, P. Immunocytochemical demonstration of visualpigments in the degenerate retinal and pineal photoreceptors of the blind cave salamander (Proteus anguinus ). Cell Tissue es 303, 15 (2001).
11 47. Schlegel, P. A., Steinfartz, S. & Bulog, B. Non-visual sensory physiology and magnetic orientation in the Blind Cave Salamander, Proteus anguinus (and some other cave-dwelling urodele species). eview and new results on light-sensitivity and non-visual orientation in subterranean urodeles (Amphibia). Anim Biol Leiden Neth. 59, 351 (2009). 48. Ivanovi, A., Aljani, G. & Artzen, J. W. Sull shape dierentiation of blac and white olms (Proteus anguinus anguinus and Proteus a. parelj ): an exploratory analysis with micro-CT scanning. Contrib Zool 82, 107 (2013). 49. Goldberg, C. S. et al. Critical considerations for the application of environmental DNA methods to detect aquatic species. Methods Ecol Evol ., doi: 10.1111/2041-210X.12595 (2016). 50. Sliovi, I. On the hydrogeological conditions of western Herzegovina (Bosnia and Herzegovina) and possibilities for new groundwater extractions. Geol Croat 47, 221 (1994). 51. inigoj, J., Lapanje, A. & Polja, M. Geologija obmoja zaledja Dobliice. [e geology of the Dobliica iver recharge area.] Dolenjsi ras 6, 46 (2012).eDNA analyses were carried out in the Molecular Genetics Laboratory at the Department of Animal Science, Biotechnical Faculty, University of Ljubljana and in the Evolutionary Zoology Laboratory at the Institute of Biology, Scientic Research Centre at the Slovenian Academy of Sciences and Arts. Additional support was provided by the Subterranean Biology Laboratory at the Department of Biology, Biotechnical Faculty, University of Ljubljana (Dr. Valerija Zakek), the Devon Karst Research Society (Brian Lewarne), the Institute of the Republic of Slovenia for Nature Conservation (Andrej Hudoklin), Omega d.o.o (Dr. Nataa Toplak), the Center for Karst and Speleology (Dr. Jasminko Mulaomerovi, Simone Milanolo and Dr. Ivo Lui), the Herzegovinian Mountain Rescue Service Mostar (Ivan Skaramuca, Zdenko Mari and ana Marijanovi), the Laboratory of Molecular Genetics, Institute of Forensic Medicine, Faculty of Medicine, University of Ljubljana (Dr. Irena Zupani Pajni), Ivan Bebek, Tajda Gredar, Luka Vodnik and Andrej Renelj. Marijan Govedi (Centre for Cartography of Fauna and Flora, Slovenia) and Dr. Lawrence B. Cohen (Yale School of Medicine and Korea Institute of Science and Technology) are thanked for reviewing the manuscript. e study was part of the project A survey of the distribution of Proteus anguinus by environmental DNA sampling, co-nanced by the Critical Ecosystem Partnership Fund, BirdLife International and DOPPS (2013), and the project With Proteus we share dependence on groundwater, co-nanced by the EEA Financial Mechanism and the Norwegian Financial Mechanism 2009 (SI03-EEA2013/MP-17).Idea of the study and management: G.A. Conceived and designed the study: .G., G.A., M.N.A., D.S. Conceived and designed eDNA experiments, analysed data: .G., D.S., G.A. Fieldwork preparation and sampling: G.A., M.N.A., .G., D.S., M.P., Z.G. Laboratory work: D.S., .G., G.A., Z.G., M.P., M.N.A. Contributing to laboratory materials & facilities: W.R.J., A.S., M.K., Z.G., G.A., M.P. GIS data processing and mapping: M.N.A. Wrote the paper: .G., D.S., W.R.J., G.A. Read and commented on the manuscript: P.T., A.S., M.N.A., M.K., Z.G., M.P.Accession codes: Proteus mitochondrial control region and rDNA sequences are deposited in GenBank, Accession Numbers KY523107KY523177. Supplementary information accompanies this paper at http://www.nature.com/srep Competing Interests: e authors declare no competing nancial interests. How to cite this article: Goriki, et al. Environmental DNA in subterranean biology: range extension and taxonomic implications for Proteus Sci. Rep. 7, 45054; doi: 10.1038/srep45054 (2017). Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional aliations. is work is licensed under a Creative Commons Attribution 4.0 International License. e images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ e Author(s) 2017