Introduction

Japanese giant salamander (Andrias japonicus) is one of the largest amphibian species in the world, with a maximum body length of 1.5 m, and has been registered as a special Japanese natural treasure since 1952 (https://kunishitei.bunka.go.jp/heritage/detail/401/3129). This iconic salamander is distributed in streams with relatively moderate slopes in western Japan and reproduces from August to September (Okada et al. 2015). The giant salamander’s eggs hatch in winter and the larvae metamorphose after approximately 3 years. Sexual maturity is considered to take 8 years or more (Browne et al. 2014). Japanese giant salamander forages mainly on crustaceans and small fish, reaching at the top of the trophic chain in stream ecosystems in Japan (Browne et al. 2014; Matsui and Hayashi 1992; Matsui et al. 2008). However, recent climate and land use changes, as well as competition with the alien Chinese-originated Andrias spp. (e.g., Andrias davidianus and Andrias sligoi; Turvey et al. 2019), can result in habitat degradation and reduction of distribution for Japanese giant salamander (Browne et al. 2014; Fukumoto et al. 2015; Matsui et al. 2008). Moreover, introgression due to hybridization between the endemic and alien giant salamanders causes genetic contamination and can further decrease the population of the endemic giant salamander (Matsui et al. 2008; Yoshikawa et al. 2011).

Considering these crises, the Japanese giant salamander is now listed as vulnerable in both the International Union for Conservation of Nature and Natural Resources (IUCN) and the Japanese National Red List (Ministry of the Environment 2020). Japanese giant salamanders have no land phase, hide in riverbed gaps between rocks or in underwater halls in vegetated riverbanks during the day, and are active at night (Fukumoto et al. 2015). Owing to their aquatic and nocturnal features, conventional capture-based surveys are time-consuming, labor-intensive, and sometimes dangerous for researchers. In addition, capture-based surveys are potentially harmful to individuals and their habitats. Therefore, it is unlikely that these conventional methods can sufficiently assess the giant salamander’s current distribution and abundance, which limits the availability of information pertaining sites where conservation of the giant salamander should be prioritized.

Environmental DNA (eDNA; extra-organismal DNA fragments in aquatic and terrestrial environments) analysis has recently attracted considerable attention as a novel biomonitoring tool that can complement conventional methods (Clare et al. 2022; Deiner et al. 2017; Minamoto et al. 2012; Yao et al. 2022). In this analysis, target organisms’ presence and distribution can easily be estimated by detecting their eDNA fragments via PCR amplification. Thus, eDNA analysis allows for noninvasive, cost-effective, and sensitive biomonitoring (Czeglédi et al. 2021; Darling and Mahon 2011; Fediajevaite et al. 2021; Jo et al. 2020b) and several studies have demonstrated the usefulness of eDNA analysis for the surveillance of amphibian species (Bálint et al. 2018; Biggs et al. 2015; Ficetola et al. 2008; Iwai et al. 2019; Jo et al. 2020c; Li et al. 2021; Pilliod et al. 2013). Fukumoto et al. (2015) previously used eDNA analysis to survey the distribution of Japanese giant salamanders along the Katsura River basin, Japan, demonstrating the suitability of the analysis for rapid and extensive biomonitoring. Fukumoto et al. (2015) also stated that further application of eDNA analysis over a broader spatial scale would enhance its usefulness in the context of aquatic conservation.

Expanding the eDNA application in Fukumoto et al. (2015), the present study conducted an eDNA-based surveillance of Japanese giant salamander’s distribution in western Japan. We collected 410 water samples from 12 rivers and analyzed the presence or absence of target eDNA in the water samples. Although most eDNA studies have targeted mitochondrial DNA (mtDNA) as a genetic marker (Biggs et al. 2015; Jerde et al. 2011; Sakata et al. 2021), recent studies suggested the use of nuclear DNA (nuDNA), particularly multiple copies of ribosomal RNA (rRNA) genes, as a more sensitive genetic marker for eDNA analyses than mtDNA (Dysthe et al. 2018; Jo et al. 2020a; Minamoto et al. 2017). Despite multiple mitochondrial genomes per cell (tens to thousands of mtDNA copies), the multi-copy nuDNA marker has exhibited similar or higher eDNA detection sensitivity likely due to the tandem-repeated sequences of rRNA genes in eukaryotic cells (Bylemans et al. 2017; Jo et al. 2020a; Minamoto et al. 2017). We thus designed a novel eDNA assay for the multi-copy nuDNA of the Japanese giant salamander and compared the detection rate of target eDNA between mtDNA and nuDNA markers. We further examined the relationship between target eDNA detection rate and abiotic parameters to discuss the effects of environmental conditions on Japanese giant salamander distribution and their eDNA detection.

Methods

Field survey and water sampling

We collected water samples from 336 sites along 12 river systems in western Japan from June to December 2014 and June to November 2015 (Fig. 1). Water sampling was conducted three times in the Hatsuka River (Hyogo Pref.), twice in the Sayo River (Hyogo Pref.), and once in other rivers, resulting in a total of 410 water samples (Table S1). The rivers surveyed in our study were located in prefectures where the presence of the target species was confirmed by a previous national census and an individual study (Biodiversity Center of Japan 2001; Matsui et al. 2008). Considering the potential downstream transport of eDNA particles (Jo and Yamanaka 2022), most sampling sites were located 5–10 km apart along the mainstem, with some exceptions at the confluence of the tributaries. For each survey, we collected water samples using a 1-L plastic container from the river surface. The plastic containers were in advance carefully bleached with 0.1% sodium hypochlorite solution for at least 5 min (Jo et al. 2020b) and washed twice with the river water immediately before water sampling. Disposable gloves were put on during water sampling and were replaced at each sampling site. When collecting the water samples, environmental parameters (water temperature, pH, and electrical conductivity [EC]) were measured using Hanna Combo 2 (HI 98130; Hanna Instruments, Japan) at each site (Table S1). Water samples were transported to the laboratory (Kobe University, Hyogo, Japan) at low temperatures in a cooler box and filtered within a day following sampling (Jo et al. 2020c), except for water samples from Maruyama River, which were frozen and filtered a few days after sampling.

Fig. 1
figure 1

Map of all water sampling sites in this study with a photograph of Japanese giant salamander (at night survey in the Sayo River). Water samples were collected from 336 sites at 12 rivers in western Japan (shown as circles). Note that the detailed location with and without target eDNA detection is not disclosed because of the rarity and conservation importance of the target species

Full size image

We filtered water samples using a 47-mm diameter glass microfiber filter GF/F (nominal pore size 0.7 μm; GE Healthcare Life Science, Little Chalfont, UK). Each 1 L of water sample was filtered through one or two GF/F filters, depending on filter clogging. In each river system, one or two 1-L units of distilled water were filtered as a filtration negative control to assess the potential for contamination of target eDNA during and after water filtration. Before and after every use, all filtering equipment (i.e., filter funnels [Magnetic Filter Funnel, 500 mL capacity; Pall Corporation, Westborough, MA, U.S.], 1-L beakers, and tweezers) were bleached with 0.1% sodium hypochlorite solution for at least 5 min. All filtered samples were stored at − 20 °C until DNA extraction.

DNA extraction and real-time PCR

Total eDNA on the filter was extracted using the DNeasy Blood and Tissue Kit (Qiagen, Germany), following the method described by Minamoto et al. (2019). The presence of Japanese giant salamander eDNA in the water samples was evaluated using the StepOnePlus Real-Time PCR system (Applied Biosystems). We designed the primers and probe set for the nuclear DNA of the Japanese giant salamander, which amplifies 133 bp fragments of rRNA genes ranging from 18S to internal transcribed spacer-1 (ITS1) regions (Table 1; Table S2; Appendix S1). Each 20 μL TaqMan reaction contained 2 μL of template DNA, a final concentration of 900 nM of forward and reverse primers, and 125 nM of TaqMan probe in a 1 × TaqMan Gene Expression Master Mix (Thermo Fisher, U.S.). Thermal conditions of real-time PCR were as follows: 2 min at 50 °C, 10 min at 95 °C, and 55 cycles of 15 s at 95 °C and 60 s at 60 °C. Exuviae-derived DNA from target species (0.2 ng/template) and 2 μL of pure water were simultaneously analyzed as PCR-positive and -negative controls, respectively. All PCRs for the eDNA extracts, positive controls, and negative controls were performed in four replicates. Target eDNA was considered present in each water sample if any of the PCR replicates were clearly amplified. The primers and probe set developed in Fukumoto et al. (2015) was used as the assay for amplifying mtDNA of the target species (Table 1). The water samples collected in 2014 were analysed using both mtDNA and nuDNA markers, while those collected in 2015 were analyzed using only nuDNA markers because of the higher eDNA detection rate of the nuDNA marker (see below).

Table 1 The primers and probe set used in this study
Full size table

We estimated the limit of detection (LOD; defined as the lowest quantity of target DNA that could be detected in one of the PCR replicates) of each genetic marker. The LOD is based on detection/non-detection criteria and represents the ability of an assay to detect the low concentrations of target DNA fragments (Klymus et al. 2020). A dilution series of the A. japonicus tissue DNA (10–3–102 pg/2 µL PCR template) was prepared for estimating the LODs. PCRs for the dilution series were performed in six replicates and other PCR conditions were similar as described above.

Statistical analyses

All analyses were performed using R version 4.2.1 (R Core Team 2022) and the significance level (α) was set at 0.05. We compared the detection rate of target eDNA between mtDNA and nuDNA markers using water samples collected in 2014 (n = 246). We used a generalized linear mixed model (GLMM) with a binomial distribution to assess the effects of genetic markers and environmental parameters (EC and pH) on target eDNA detection using the lmerTest package (Kuznetsova et al. 2017). In the GLMM, the presence/absence of target eDNA in each water sample (1/0) was included as a dependent variable, and the type of genetic marker (mtDNA or nuDNA), pH, and EC were included as fixed effects. Given that water sampling was replicated in some rivers, each river was included as a random effect, and site IDs were nested within the river groups. Additionally, using the water samples collected in 2015 (n = 164), we also performed a binomial GLMM to assess the effects of environmental parameters (pH and EC) on target eDNA detection using the same formula as above (except for using only nuDNA markers). For both GLMMs, we confirmed that multicollinearity among the variables was negligible (variance inflation factor, VIF = 1.008–1.050). Water temperature was not used for the analyses because of its temporal variability within a day (Table S1).

Results

All PCR-positive controls were successfully amplified, and no filtration and PCR-negative controls were amplified throughout the study (Table S1). The LODs were estimated at 0.01 pg (per reaction) for both the genetic markers (Table S3). However, when 0.1 and 0.01 pg of tissue DNA were used, four and one replicates were amplified for the mtDNA marker and six and four replicates were amplified for the nuDNA marker.

In 2014 surveys, we collected 246 water samples in total from seven rivers and detected Japanese giant salamander eDNA in 74 samples using nuDNA marker (74/246 = 30.1%), 48 samples using mtDNA markers (48/246 = 19.5%), and 85 samples using either of the markers (85/246 = 34.6%; Table 2). Of the 246 samples, 37 showed target eDNA detections with both markers, 37 showed target eDNA detections with only the nuDNA marker, and 11 showed target eDNA detections with only the mtDNA marker. In contrast, the target eDNA was not detected in 161 samples using either of the markers (Table S1; results of target eDNA detections from each river are summarized in Fig. S1). The number of samples with target eDNA detection was higher for nuDNA than for mtDNA markers in all surveyed rivers. A binomial GLMM based on the 2014 dataset showed a higher eDNA detection rate of the nuDNA marker (P < 0.001) and significant negative effects of pH (P < 0.001) and EC (P = 0.024) on eDNA detection (Table 3). For the samples with eDNA detections, the pH and EC values were 7.08 ± 0.57 and 0.056 ± 0.031, respectively (Fig. 2). For the samples without eDNA detections, by contrast, the pH and EC values were 7.54 ± 0.74 and 0.076 ± 0.044, respectively.

Table 2 Summary of Japanese giant salamander eDNA detection from 12 river systems in western Japan and corresponding environmental parameters
Full size table
Table 3 Summary of binomial GLMMs
Full size table
Fig. 2
figure 2

Comparison of target eDNA detection between mitochondrial (mt) and nuclear (nu) markers (left) and EC (middle) and pH (right) values between detection/non-detection samples in 2014. Numbers in parentheses mean the sample size. Statistical differences are shown as asterisks (*P < 0.05; ***P < 0.001)

Full size image

In 2015 surveys, we collected 164 water samples in total from six rivers and detected Japanese giant salamander eDNA in 76 samples using nuDNA marker (76/164 = 46.3%; Table 2). A binomial GLMM based on the 2015 dataset showed a significantly negative effect of EC on eDNA detection (P < 0.001), whereas the pH effect was not significant (P = 0.673; Table 3; Fig. 3). For the samples with eDNA detections, the pH and EC values were 7.26 ± 0.73 and 0.051 ± 0.023, respectively. For the samples without eDNA detections, the pH and EC values were 7.31 ± 0.60 and 0.087 ± 0.037, respectively. In summary, during the 2-year survey from 2014 to 2015, target eDNA was detected in 150 of the 410 water samples (36.6%) using nuDNA markers and from 162 of the 410 water samples (39.5%) using either of the genetic markers (Table S1).

Fig. 3
figure 3

Comparison of EC (left) and pH (right) values between detection/non-detection samples in 2015. Statistical differences are shown as asterisks (***P < 0.001). n.s. non-significant

Full size image

Discussion

We succeeded in estimating the distributional information of Japanese giant salamanders in western Japan using eDNA analysis. The estimated distribution of the Japanese giant salamander generally corresponded to the known natural distribution of the species, as reported in previous studies (Biodiversity Center of Japan 2001; Matsui et al. 2008). Although Fukumoto et al. (2015) examined the distribution of Japanese giant salamander using eDNA analysis in a single river system over two days, our study was able to survey the distribution much more broadly over 19 days, further indicating the usefulness of eDNA analysis for rapid and extensive surveillance of aquatic rare species distribution. Moreover, we showed a higher sensitivity of the nuDNA marker for detecting giant salamander eDNA, which would allow for efficiently surveying the Japanese giant salamander’s entire distribution using eDNA analysis and facilitating their conservation.

The results of a binomial GLMM showed that the nuDNA marker developed in this study yielded higher detection of Japanese giant salamander eDNA than the mtDNA marker developed by Fukumoto et al. (2015). The result was also supported by its higher detection sensitivity in the LOD test, where the target DNA detection rate with low quantity (0.1 and 0.01 pg) was higher for the nuDNA marker than the mtDNA marker. Given that the previous reports have shown higher fish eDNA detection sensitivity through multi-copy nuDNA markers (Dysthe et al. 2018; Minamoto et al. 2017), our results supported the higher eDNA detection sensitivity of multi-copy nuDNA markers and indicated their usefulness for efficient eDNA-based biomonitoring for amphibian species.

Alternatively, the result may simply be accounted for by the difference in PCR amplicon lengths between the markers (116 bp for nuDNA and 133 bp for mtDNA) because longer DNA fragments are expected to degrade faster and persist shorter in the water (Hänfling et al. 2016; Jo et al. 2017). However, it is unlikely that such small differences in PCR amplification lengths between the markers would significantly affect the yields of target eDNA (Bylemans et al. 2018; Jo et al. 2020a). Altogether, the multi-copy nuDNA marker would allow for reducing the risk of false-negative eDNA detections and sensitively estimating the salamander’s distribution.

The binomial GLMM results also showed a significant negative effect of EC on the detection of the Japanese giant salamander eDNA in both years. High EC is an index of water quality decline, relating to the total dissolved solids and ion concentration in the water and negatively impacting the occurrence, abundance, and reproductive success of aquatic species including some salamanders and hellbenders (Alavi and Cosson 2006; Bodinof Jachowski et al. 2016; Bowles et al. 2006; Keitzer et al. 2013). Some studies have previously documented negative relationships between EC and detection rate of fish and amphibian eDNA (Jo et al. 2020b, c; Pitt et al. 2017), which supports our findings. Alternatively, high EC values can also relate to PCR inhibition (Schrader et al. 2012) and prevent the eDNA detection (Harper et al. 2019; Wineland et al. 2019), although Pitt et al. (2017) confirmed no evidence of PCR inhibition in their river water samples with higher EC values (> 0.5 mS/cm) than ours (< 0.25 mS/cm). We thus conclude that high EC is likely affecting the giant salamander’s occurrence at some sites, rather than detections of the target eDNA when the species are present there. This problem can be addressed to some extent by using a site occupancy model, which separates the probability of eDNA occurrence (not species occurrence) from that of eDNA detection (Jo et al. 2020c; Wineland et al. 2019), though the site occupancy modeling could not be applied to this study due to a lack of sampling replicates per survey site.

By contrast, the effect of pH on the eDNA detection was inconsistent between the survey years. In the 2014 survey, Japanese giant salamander eDNA tended to be detected less frequently in samples with higher pH. A higher pH (> 8.5) is also considered to reflect water quality (Boczkaj and Fernandes 2017) and can be observed in rivers with high algal productivity and limestone rocks (Bhateria and Jain 2016). However, it is less likely that eDNA persistence varies within the range of pH measured in this study (Jo et al. 2022; Strickler et al. 2015). Thus, the result may also indicate that alkaline conditions affect the giant salamander’s occurrence rather than target eDNA detection. For example, eDNA detection rate was low on average in the Sayo River, which had the highest pH and EC among the surveyed rivers. In contrast, the effect of pH on target eDNA detection was not significant in the samples collected in 2015. It is not clear what caused this difference, but this implies that, although EC was a strong variable determining the giant salamander’s distribution, a higher pH may not always be observed at sites that the salamanders do not prefer to inhabit. Their distribution can also be determined by other environmental factors (e.g., annual mean temperature, precipitation, elevation, land use, and geological features; Houlahan et al. 2000; Okada et al. 2008; Willson and Dorcas 2003). Precipitation can be related to the flow condition of the river, likely affecting the habitat suitability of river-dwelling species, including the target species, and land use and elevation can be related to anthropogenic impacts on amphibian habitats (Johnson et al. 2011). Other environmental parameters, such as biochemical oxygen demand (BOD) and dissolved organic carbon (DOC), may also associate with the giant salamander’s survival and distribution in the studied rivers. Future studies should measure relevant environmental variables simultaneously with eDNA sampling (Jo and Yamanaka 2022).

Although the presence/absence of target eDNA in a water sample was the focus of this study, we may have been able to infer a more detailed ecology of the giant salamanders in the studied rivers by quantifying target eDNA concentrations. Environmental DNA concentrations can represent their relative abundance and activity in the environment (Iwai et al. 2019; Jo et al. 2020a; Pilliod et al. 2013; Spear et al. 2015). Continuous eDNA-based quantitative monitoring will inform time-series changes not only regarding the giant salamander’s distribution but also their relative abundance in rivers. Moreover, as external fertilization (Kawamichi and Ueda 1998) can increase the relative concentration of nuDNA to mtDNA in the water, the ratio of nuclear to mitochondrial eDNA concentrations could be used to estimate the timing and location of their spawning (Bylemans et al. 2017; Wu et al. 2022). These efforts will advance our understanding of the life history and reproductive ecology of giant salamanders, providing useful information for conservation of endangered aquatic species with external fertilization.

For the practical use of nuDNA in eDNA-based biomonitoring, the limitation of nuclear gene sequences available in databases (e.g., GenBank) is a primary drawback, which is especially the case for non-model organisms (Jo et al. 2022; Minamoto et al. 2017; Sigsgaard et al. 2020). Phylogenetics studies targeting bacteria and fungi have mainly used nuclear ribosomal DNA (rDNA) sequences (Handelsman 2004; Toju et al. 2012) and accordingly a significant amount of them has been accumulated. In contrast, studies targeting macro-organism’s (e)DNA including fish and amphibians have mainly used mtDNA fragments and research interest in nuDNA is still limited. Some recent studies have challenged eDNA-based genetic diversity monitoring targeting aquatic vertebrates, but most of them were also based on mtDNA (Sigsgaard et al. 2016; Wakimura et al. 2023; but see Andres et al. 2021). While recent substantial advance of next-generation sequencing (NGS) technology has drastically reduced sequencing costs (Goodwin et al. 2016), achieving database enrichment and developing a framework for sharing data among the science communities will require further collaborative efforts of multiple researchers and laboratories in the future.

Including our study, eDNA analysis has enabled the collection of broad-scale species distribution data on a considerably shorter timescale compared to conventional methods (Biggs et al. 2015; Deiner et al. 2017; Yao et al. 2022). Such an advantage of eDNA analysis can be quite useful not only for a rapid understanding of the species’ distribution and abundance, but also for revealing the relationship between their suitable habitats and environmental conditions. A few studies recently applied eDNA analysis to species distribution models (SDMs; a.k.a. ecological niche models) to link species occurrence records with environmental conditions and then estimated habitat suitability, defined as the occurrence probability of a species at a site with a given environmental condition (Hashemzadeh Segherloo et al. 2022; Riaz et al. 2020; Wilcox et al. 2018). Predicting species distribution has become increasingly important for its conservation owing to the impacts of recent climate change, including global warming caused by anthropogenic CO2 emissions. These effects are considered to cause habitat shifts and disturbances for various animals and plants, which may further accelerate in the future (Butchart et al. 2010; Cardinale et al. 2012; Ceballos et al. 2015). Combined with statistical approaches such as SDMs, eDNA analysis can contribute to biodiversity conservation and ecosystem management. Such studies would help to preserve the habitats of various rare species, including the giant salamander, and save them from extinction.