Introduction

Monitoring species distribution and abundance accurately and efficiently is vital for conservation efforts. Traditional survey methodologies examine morphological characteristics to distinguish between species, requiring proficient personnel with specialized taxonomic knowledge (Thomsen and Willerslev 2015; Mena et al. 2021). However, traditional survey methods may disrupt the environment during the survey process, and scaling up investigations and broadening the range of surveyed species necessitates significant resources, time, and effort (Zimmermann et al. 2015; Port et al. 2016).

Using environmental samples in a noninvasive technique known as environmental DNA (eDNA) metabarcoding is an emerging method of species monitoring. One of the key benefits of this technology lies in its capacity to identify the presence of living organisms, encompassing both fauna and flora, utilizing minute quantities of environmental substances, such as water, soil, and air (Buxton et al. 2017; Deiner et al. 2017; Jarman et al. 2018). The term eDNA encompasses genetic material, including DNA from the skin, mucus, saliva, secretions, urine, pollen, microorganisms, and corpses, released into the environment by organisms (Ruppert et al. 2019). Additionally, eDNA analysis has been effectively demonstrated in identifying rare species that are challenging to detect through traditional survey methods because it uses unique genetic material for each species with high accuracy and sensitivity (Jerde et al. 2011; Hänfling et al. 2016; Duarte et al. 2023). In its early days, the primary application of eDNA was focused on threatened species in protected areas and the surveillance of fish species within river ecosystems (Valentini et al. 2016; Miya et al. 2015). Since then, its application has broadened to include more taxonomic categories, such as avian and mammalian species (Ushio et al. 2017, 2018; Franklin et al. 2019; Sales et al. 2020; Broadhurst et al. 2021). Nevertheless, study areas have usually focused on river systems.

The urban ecosystem exhibits high fragmentation due to anthropogenic influences, resulting in heterogeneous ecological characteristics. The presence of invasive species and the occurrence of key native species vary in each specific area, highlighting the necessity of monitoring microhabitats within the urban ecosystem. Commonly used biological indicator species for assessing the ecological integrity of urban ecosystems include birds and mammals (Hausner et al. 2003; Tulloch et al.2011; Sattler et al. 2014). Notwithstanding the advancements made using eDNA, the identification of terrestrial animals presents distinct challenges owing to their life histories (Ruppert et al. 2019; Sales et al. 2020; Gregory et al. 2021) In contrast to aquatic organisms, which inhabit and regularly interact with water, terrestrial organisms typically have limited availability of water and sporadic encounters with it, primarily while drinking, traversing aquatic environments, or eliminating waste (Seeber et al. 2019; Lynggaard et al. 2023). Bird species are especially challenging, but according to Ushio et al. (2018), recent research has indicated the possibility of utilizing eDNA for identifying avian species. Indeed, several investigations have subsequently employed metabarcoding as a methodology to examine avian species (Schütz et al. 2020; Alfaro et al., 2022; Mariani et al. 2021; Johnson et al. 2023; Jønsson et al. 2023). However, research on avian species using eDNA remains limited, requiring additional investigations. Therefore, in this study, we attempted to use eDNA metabarcoding to monitor birds and mammals. Extending the role of eDNA, this study specifically focused on urban biological environments.

Ponds located within urban areas offer various ecological benefits to semi-aquatic and terrestrial mammals, including opportunities for drinking, feeding, dispersal, and breeding (Klymus et al. 2017). Urban ecological parks and their associated ponds are crucial in providing habitats for various mammalian and avian species within the urban ecosystem. Consequently, the analysis of eDNA in water samples collected from these ponds can be utilized to detect diverse organisms and assess the level of biodiversity in urban areas (Deiner et al. 2017; Harper et al. 2019). With over 2800 species, Gildong Ecological Park is renowned for its great ecological significance. This park, surrounded by an urban area, is the first ecological park designated in Seoul, and it was established in 1999. Every year, volunteers and specialists monitor the park to promote species diversity and provide a habitat for urban wildlife, resulting in an abundance of monitoring data for existing species.

Research using eDNA metabarcoding in urban areas has been conducted in various settings. For instance, Charvoz et al. (2021) conducted a study in which they sampled 45 marsh locations within the Geneva canton, Switzerland, focusing on 5 newt’s community. In another study by Kim et al. (2022), sampling was conducted in 19 urban park water tanks and four natural marshes to compare with camera trapping data in Gwanggyo, Korea. Moreover, in Beijing, Zhang et al. (2023) collected 109 samples across the entire city, detecting 39 bird and 11 mammalian species.

This study’s main objective was to use eDNA metabarcoding to monitor mammal and bird species in an urban ecological park. Our study referenced the sampling frequencies and sizes from previous studies, particularly those conducted on marshes in other urban areas (Charvoz et al. 2021; Kim et al. 2022; Zhang et al. 2023). We collected 10 to 20 samplings at once and repeated this process five times over one year. We had three specific research goals:

  1. 1.

    To compare the species list obtained from eDNA sampling with the results of existing studies and assess the usability of eDNA for species detection.

  2. 2.

    To examine whether the species composition differs depending on the types of specific zones within the park.

  3. 3.

    To perform time-series monitoring using eDNA examining mammal and bird species changes over time.

Materials and methods

Data collection

Seoul is a metropolis with a population of over 10 million people, and our study site is an ecological park located within this urban ecosystem. Gildong Ecological Park, situated in Gangdong-gu, Seoul, Korea, is the first ecological park designated in Seoul, and it was established in 1999. With over 2800 species present, this park is renowned for its great ecological significance. It covers an area of 80,683 m2 and is on a low marsh. The park is segregated into three sections: a marsh area (the marsh zone), an ecological forest area (the mountain zone), and a rural experience learning center (the rural experience zone). The park encompasses nine distinct marshes that span a total of 26,127 m2 (Fig. 1). They receive groundwater from three wells flowing from the ecological forest area to the marshes. A permanent water system is maintained throughout the year due to the topography and groundwater characteristics. At Gildong Ecological Park, monitoring is performed through self-assessment by experts and volunteers covering the entire area. Bird species are monitored three times a year, while continuous observation is conducted for mammals. Each year, a species list is published in the monitoring report. Since its opening in 1997, constant monitoring surveys have accumulated rich ecological information. Utilizing this monitoring data, we aimed to compare and analyze the occurrence species information obtained through eDNA.

Fig. 1
figure 1

Environmental DNA sampling sites in Gildong Ecological Park, Seoul. The white number corresponds to the number of species detected

Full size image

The Gildong Ecological Park marsh zone serves as a habitat for several species of water birds, including herons (Ardea cinerea) and mandarin ducks (Aix galericulata). Conversely, the mountain zone is primarily inhabited by mammals, such as raccoon dogs (Nycetereuts procyonoides koreansis), water deer (Hydropotes inermis), and squirrels (Sciurus vulgaris coreae), as well as forest birds, such as the great spotted woodpecker (Dendrocopos major). Moreover, amphibians and resident birds, such as finches (Fringilla montifringilla), are present in the rural experience zone.

From August 2020 to September 2021, eDNA sampling was conducted five times, with a total of 65 samples (including five negative control groups; purified water) taken from the nine marshes. All samplings were conducted using Strevix 0.45 μm filters (Millipore). To minimize contamination, sampling was carried out using a separate 30 mL disposable syringe and disposable gloves each time, and 500–1000 mL of water was collected per sample. Once collected, samples were immediately stored in a portable ice box and transported to the laboratory, where they were stored at − 18 °C until DNA extraction.

DNA metabarcoding

The primers targeting mitochondrial 12 S rDNA were used for the DNA metabarcoding. Of the 65 samples, 63 were analyzed using the MiMammal primer set (MiMammal-U-F, 5′-GGGTTGGTAAATTTCGTGCCAGC-3′; MiMammal-U-R, 5′-CATAGTGGGGTATCTAATCCCAGTTTG-3′, Ushio et al. 2017), while 52 were analyzed using the MiBird primer set (MiBird-U-F, 5′-GGGTTGGTAAATCTTGTGCCAG C-3′, MiBird-U-R, 5′-CATAGTGGGGTATCTAATCCCA GTTTG-3′, Ushio et al. 2018). Sequencing was performed using the MiSeq platform (Illumina, San Diego, CA, USA) in collaboration with the National Instrumentation Center for Environmental Management (NICEM). The detailed metabarcoding experimental procedures are in the Appendix (Method S1).

Data analysis

Species identification was carried out using established pipeline database (Sato et al. 2018), retaining only database matches with 97% or greater similarities. For both mammalian and avian species, proportional read counts were calculated for each sample to reduce false positives due to inter-sample contamination sample (Harper et al. 2019; Muri et al. 2020; Lin et al. 2020), and species represented by less than 0.01% of the total read counts were removed (Port et al. 2016; Hänfling et al. 2016; Harper et al. 2023). Species identities were cross-checked against Korea’s National Biological Species List (https://species.nibr.go.kr/, National Biological Resource Center) to reassign them to domestic subspecies and correct any incorrect nomenclature.

We calculated the proportional read count of each detected species within each eDNA sample from total raw read counts to assess the eDNA detection in each. As an added measure, the total read counts were logarithmically transformed to facilitate the comparison of occurrence levels among different zones. The total count of eDNA reads indirectly indicates the total detection amounts within the target area (Buxton et al. 2017; Polanco et al. 2021; Kang et al. 2024). Since neither the log-transformed read count data (Shapiro–Wilk normality test W = 0.2, P < 0.001) nor the proportional read counts data (Shapiro–Wilk normality test W = 0.7, P < 0.001) conformed to a normal distribution, we employed the Mann–Whitney U test and the Kruskal–Wallis test for statistical comparisons. When comparing species detection levels using eDNA metabarcoding across park zones, samples from the rural experience zone (neither mountain nor marsh), which had only a small number of samples, were eliminated to compare the eDNA detection levels better. It was postulated that the detection rate of the three common mammal species (Water deer, Racoon dog and Wild boar) would be higher in the mountain zone, whereas the two species of ducks (mandarin ducks and herons) would exhibit greater detection rates in the marsh zone. We conducted all statistical analyses using R v.4.1.3 (R Core Team, 2017).

Results

eDNA metabarcoding

The MiMammal primer yielded 7,071,446 sequence reads, of which 4,513,973 remained after trimming, merging, and length filtering, while the MiBird primer generated 5,781,721 sequence reads, with 3,843,725 reads retained after the same processing steps. After clustering to remove chimeras and duplicates, the library contained 1,109,672 sequences for MiMammal, with an average of 18,496 reads per sample (excluding negative controls), and 2,640,723 sequences for MiBird, with an average of 52,814 reads per sample (excluding negative controls)

Since fish were not the target species of this study, they were subsequently removed from the PCR-negative control group (Appendix Table S2). The remaining contaminants included mice (Mus musculus) and humans (Homo sapiens). It was determined that human DNA contamination occurred during the sampling process, and mouse DNA was considered cross-contaminated during the PCR amplification and metabarcoding processes. Therefore, they were excluded from all analyses. To reduce false positives, species-specific read counts with relative abundance values of less than 0.01% were removed (Hänfling et al. 2016; Muri et al. 2020). Through this process, we aimed to remove results with a small amount of amplification due to inter-sampling contamination.

Each species was reassigned to a domestic subspecies based on the National List of Korean Species (https://species.nibr.go.kr/, National Institute of Biological Resources). A total of four species were changed: Among birds, the genera Tadorna were identified as the mandarin duck, based on pre-survey results at the Seoul Wildlife Center using the MiBird primer (Appendix Table S2). Wild cat (Felis silvestris) and rock pigeons (Columba livia) were detected, they were reclassified under the taxonomies for domestic cats (Falis catus) and domestic pigeons (Columba liva domestica), respectively. The Eurasian tit was designated in the database as Poecile palutris. Since this database is from Japan, the we modified the scientific name Parus palustris for the Korean subspecies. The eDNA metabarcoding analysis detected 31 terrestrial animal species, including 20 bird species and 11 mammal species (Table 1; Fig. 2).

Table 1 Comparing environmental DNA (eDNA) and monitoring surveys: Mammals
Full size table

Comparing the eDNA metabarcoding results to traditional survey results

The eDNA metabarcoding results for mammalian species detected 11 species, accounting for 73% of the 15 species recorded in the Gildong Ecological Park Annual Report (East Seoul Park Management Office 2020). We found several mammal species considered to be important in urban habitats, including water deer (Hydropotes inermis), wild boar (Sus scrofa), Siberian weasels (Mustela sibirica), and raccoon dogs (Nyctereutes procyonoides). For example, the city of Seoul lists the Siberian weasel as a protected species, showing that it is possible to monitor protected species in urban areas using eDNA. It also appears that it is possible to monitor wild boars, which have recently become a problem due to African swine fever (ASF) and are considered a risk factor in the urban ecosystem (Ko et al. 2021; Baek et al. 2023).

Fig. 2
figure 2

Top 20 dominant bird species detected via eDNA at Gildong Ecological Park based on proportional reads in comparison with annual monitoring reports

Full size image

For birds, 20 species were identified. The monitoring report states that during the course of a year, observations of 77 different bird species were made in the target region. It was considered, however, that it would be more appropriate to compare occurrence rates across various species due to the seasonal limits on bird surveys. Major bird species, including dominant species, represented 67% of the total 6268 individuals observed in annual bird surveys, such as vinous-throated parrotbill (Paradoxornis webbianus), Japanese tit (Parus major), marsh tit (Parus palustirs), common magpie (Pica Pica), were detected using eDNA (Fig. 2). Additionally, seasonal birds, such as the mandarin duck (Aix galericulata), common kestrel (Falco tinnunculus), brambling (Fringilla montifringilla), and Eurasian teal (Anas crecca), were also detected, demonstrating the potential for eDNA to detect dominant bird species at specific times.

Comparison of species detection levels using eDNA metabarcoding across park zones

When we examined the presence or absence of the 26 species (all species with proportional read counts higher than 0.01) across all samples, water deer (15), mandarin duck (12), brown rat (Rattus norvegicus) (10), and Eurasian teal (9) were most frequently detected across sampling sites (Fig. 3). Regarding the specific sampling locations, 13 species were detected in marsh areas F and D and forested area B. For bird species, area B had the highest levels of eDNA detection, followed by marsh areas D and C. The marsh area G had the highest level of eDNA detection for mammals, followed by the forested area B.

The mountain samples (0.117) and marsh samples (0.108) shared the same median proportional read count across all species, and there was no statistically significant difference between them (Mann–Whitney U test: U = 1360, P = 0.9) (Fig. 4(a)). When analyzing the detection data for each taxonomic group, birds exhibited no statistically significant difference between the marsh (0.123) and mountain (0.061) zones (Mann–Whitney U test: U = 265, P = 0.2) (Fig. 4(b)). Similarly, there was no statistically significant difference between the marsh (0.082) and mountain (0.151) zones for mammals (Mann–Whitney U test: U = 388, P = 0.2) (Fig. 4(c)). This indicates little difference in species detections between marsh and mountain zones, suggesting that mammals and birds make roughly equal use of every area within Gildong Ecological Park.

Fig. 3
figure 3

Heatmap showing proportional read counts for eDNA samples. The heatmap is faceted by sampling site (marsh, mountain, or rural zone) and the universal primer used (Bi: MiBird primer; M: MiMammal primer) Each sub-location represents a marsh (A–I). For those with the same number, they indicate the left and right sides of the marsh, while in the mountain zone, which is divided into five terraced marshes, they are assigned from B1 to B4

Full size image
Fig. 4
figure 4

Differences in proportional eDNA read counts in all samples and when using (a) both, (b) Mibird, and (c) Mimammal primers

Full size image
Fig. 5
figure 5

Differences in log-transformed eDNA read counts in all samples and when using (a) both, (b) MiBird, and (c) MiMammal primers

Full size image

We aimed to discern differences in read count magnitude. Consequently, a logarithmic transformation was applied to the read count, followed by a comparative analysis (Evans et al. 2016; Polanco et al. 2021; Kang et al. 2024). Samples with log-transformed read counts below 0.01 were excluded to account for potential contamination. The mountain zone samples (1.59) and marsh zone samples (2.36) had significantly different median read counts when compared across all species (Mann–Whitney U test: U = 1677, P = 0.03) (Fig. 5(a). Birds exhibited a sizeable difference between the mountain zone (2.41) and marsh zone (3.41), although the difference was not statistically significant (Mann–Whitney U test: U = 314, P = 0.1) (Fig. 5(b). Similarly, there was no statistically significant difference between the mountain (1.29) and marsh (1.95) zones for mammals (Mann–Whitney U test: U = 554, P = 0.2) (Fig. 5(c). This result means that more read counts were detected in the marshes, although there was no difference in the specific taxonomic groups.

The Kruskal–Wallis test was employed to examine the variation in occurrence rates among two park zones for the five species with the highest detection frequency. Nonetheless, the proportional read counts for the mountainous and marsh samples did not exhibit a statistically significant difference across the five species (P = 0.8) (Fig. 6). Hence, it was determined that the sampling area did not significantly influence the level of eDNA detection, even for the most prominent species.

Fig. 6
figure 6

Relationships between proportional read counts of the five most abundant species and zone nested within species. A boxplot illustrating the distribution of proportional read counts for the five species

Full size image

Furthermore, a comparative analysis assessed the average number of detections for each of the five major species. The results revealed that the mandarin duck exhibited a higher detection frequency in the marsh zone habitats, whereas the forested mountain zone areas yielded fewer detections. Conversely, the water deer displayed greater detection rates in the forested environments, with relatively fewer detections observed in the marsh areas (Fig. 6).

Seasonal variation in the eDNA results

Detecting mammalian species utilizing eDNA exhibited seasonal variation, with two species recorded during summer, 9 during autumn, and 11 during winter. The greatest median proportional read counts were found during the summer (0.109), followed by the fall (0.108), and then the winter (0.061) (Fig. 7(b)). However, the Kruskal–Wallis test revealed that the observed differences were not statistically significant (P = 0.8) (Fig. 7(b)). This implies a lack of statistically significant variation in the overall detection rate of mammals across different seasons.

Considering a total of 20 avian species that were detected, the median proportional read count values exhibited a higher magnitude during the summer (0.703), followed by winter (0.112) and then autumn (0.108). However, the Kruskal–Wallis test indicated no statistical significance (P = 0.1) (Fig. 7(a)). These findings suggest that there was no statistically significant variation in the overall detection rate of avian DNA across different seasons. Seasonal variations in avian populations have been observed due to the presence of migratory bird species. During the summer, the migratory grey heron (Ardea cinerea) and the non-migratory resident, the vinous-throated parrotbill (Paradoxomis webbianus) were observed. Notably, resident species, such as the domestic pigeon (Columba livia domestica) and the blue magpie (Cyanopica cyanus), were present during autumn. Furthermore, the white-cheeked starling (Sturnus cineraceus), a migratory species generally observed during the summer, was also recorded in the autumn, specifically in the September sampling. During the winter, there was a notable increase in detections of the mandarin duck (Aix galericulata), a migratory bird known to visit during this season. Additionally, the Eurasian teal (Anas crecca), another winter migratory bird, was observed more frequently (Fig. 7).

Fig. 7
figure 7

Relationships between proportional read counts and sampling season. The observed data (colored points) are displayed for each bird (a) or mammal (b) species. Migration type—including residents (Res; circles), summer visitors (SV; triangles), and winter visitors (WV; squares)—are displayed for the bird species

Full size image

Discussion

Differences in eDNA detection potential based on life histories and characteristics of each species

The undetected species known to be present at Gildong Ecological Park primarily encompass ground-dwelling organisms, such as hedgehogs and moles, as well as arboreal creatures, such as squirrels, all of which exhibit limited interactions with aquatic environments (Tables 1 and 2). In two recent studies, Roger et al. (2022) and Clare et al. (2022) used airborne eDNA to facilitate the detection of ground-dwelling organisms. Rather than solely relying on water sampling, these findings suggest that alternative sampling methods, such as soil or air analysis, should be considered when detecting ground-dwelling organisms since these species are important in urban ecology (Lynggaard et al. 2022; Allen et al., 2023; Johnson et al. 2023). Mammalian eDNA detection results revealed no new, previously unknown species as compared to traditional survey results. This is thought to be owing to the properties of the target site. Although the target location is a high-quality natural park, it is located in a city, and the number of mammal species that can appear is limited. However, by employing eDNA, we were able to determine the frequency of occurrence in each specific target area (Fig. 3).

In the case of avian species, species compositions varied between the outcomes of eDNA metabarcoding and conventional bird surveys (Fig. 2), with several factors hypothesized as the potential reason. Firstly, the life history traits of certain animals may result in higher contact frequencies with water and higher amounts of eDNA in the environment. Thus, the eDNA detection levels of waterfowl, especially Anatidae, such as mandarin duck, mallard, and Eurasian teal, were high (Leempoel et al. 2020). Secondly, due to the propensity of eDNA detection levels to increase with biomass (Jo et al. 2019), large species that had relatively low dominance in traditional bird surveys may have high detection levels in eDNA, including pheasant, brown-eared bulbul, and common kestrel. Additionally, the Vinous-throated parrotbill was detected prominently in both eDNA metabarcoding and conventional bird surveys. Thus, eDNA detection levels tend to rise as a species appears in its habitat more frequently, regardless of its size or level of contact with water (Williams et al. 2018).

When analyzing eDNA, no waterbirds of passage were observed. However, compared with traditional survey results, out of the total of 1 to 16 summer-visitor bird species, 3 to 14 winter-visitor bird species were detected using eDNA analysis. In contrast, 16 of 35 resident avian species were observed, representing 46% of the passerine population. Amplification of eDNA is more effective when higher quantities of DNA remain in the environment. Thus, the DNA from accumulated detritus of resident birds within the study area is clearly proportionally higher. Furthermore, it was noted that five resident bird species belonging to the woodpecker family were not detected among the 20 species surveyed. Employing eDNA in water poses challenges for monitoring avian species inhabiting mountainous regions, where birds, including woodpeckers, exhibit a comparatively lower dependence on water resources and primarily occupy forested ecosystems. Although no eDNA studies have reported the observation of woodpecker species utilizing water sources, previous research has successfully detected the presence of the order of Piciformes, which encompasses woodpeckers, through the eDNA analysis of air samples (Lynggaard et al. 2022).

Consideration of differences in the detection rate of each species in a small sub-region

Based on the results of existing ecological surveys, we hypothesized that the detection rate of bird species using eDNA would be higher in marsh ecosystems than in forest areas and that mammals would show the opposite trend. This expectation is rooted in the understanding that marshes typically serve as the primary habitats for waterbirds. Regarding mammals, it was anticipated that their prevalence would be greater in the densely forested regions within the park, given that numerous species inhabit Seungsan Mountain, which is contiguous with the Gildong Ecological Park. Mammals such as wild boars and water deer do not live within the park but are introduced from the mountain. Upon examination of the read count data, it became evident that avian species exhibited a higher prevalence in marsh ecosystems, whereas mammalian species demonstrated a greater prevalence in mountainous regions (Fig. s4(b, c) and 5(b, c). Nevertheless, the results were not considered statistically significant. This variation in eDNA detection levels can be attributed to a range of environmental factors—including the specific characteristics of a species’ habitat and water pH, temperature, and flow—rather than solely the broad type of ecosystem (Jo et al. 2019; Hauger et al. 2020; Troth et al. 2021).

No apparent differences in detection were observed between regions for each species (Fig. 6). The study site encompassed a park covering roughly 80,000m2, where water flows from mountainous regions to marsh areas (East Seoul Green Park Management Office, 2020). This observation implies that genetic material deposited in the forested regions can potentially migrate into and persist within the marsh areas. Deiner and Altermatt (2014) and Thomsen and Willerslev (2015) provided estimates for the potential range of eDNA migration in water, which spans from hundreds of meters to several kilometers. Zhang et al. (2019) used eDNA to study fish spatial patterns in lake ecosystems and discovered that while spatial autocorrelation happens in lakes bigger than 100 ha, fish community distributions are similar in relatively small lakes of about 3 ha. This is because, in small water spaces, water properties exhibit greater homogeneity. In accordance with this, the small size of water bodies within our study area probably facilitated a relatively effortless transfer of genetic material, resulting in less notable variations in detection levels across the specific regions. Therefore, if a research site size is roughly 80,000m2, it can be viewed as one ecosystem unit, and sampling can be conducted without consideration of microhabitat influence. It is believed that quantification on this scale can increase the applicability of eDNA for wide use in urban biodiversity monitoring.

Table 2 Sum of proportional read counts and the number of detections of each species by sampling season
Full size table

Seasonal change effects in the detection rates of species

The findings of this study indicate that the detection rate of water deer (Hydropotes inermis) peaked during summer, followed by a gradual decline (Fig. 6(b)). Conversely, the detection rate of the wild boar species (Sus scrofa) was lowest during the summer months, with a subsequent gradual increase into winter. If there are mammalian species that hibernate, it is expected that their activity will decrease significantly during winter. However, it is believed that no species are engaging in hibernation within the study area of the park, resulting in an absence of significant variation in the presence of mammals throughout the year.

For birds, the lack of variability in the overall read count data suggests that avian DNA quantities in the surrounding ecosystem remain consistent irrespective of the season. Nevertheless, the observed species counts exhibited seasonal fluctuations, with the detection of 2 species during summer, 9 during autumn, and 11 during winter. This observation implies that there is a low abundance of bird species interacting with water in Gildong Ecological Park during the summer season, whereas a greater number did during the autumn and winter seasons. This aligns with the pattern observed in animals that exhibit increased water utilization in response to low temperatures. Additionally, in winter, the lack of non-frozen water sources makes the availability of drinking water concentrated on ponds that consistently remain unfrozen. Moreover, an additional consideration is that in Korea, water volumes increase enormously during the summer due to the rainy season. As a result, eDNA may be more diluted in summer. Indeed, when this research was conducted in the summer of 2021, marshes were swollen due to heavy rain.

In this study, we analyzed detection rates of mammal and bird species within the urban ecosystem. Previous studies by Harper et al. (2019) and Mena et al. (2021) analyzed the likelihood of detection variation across species and seasons. Interestingly, their results corroborate our findings, with no significant differences in detection rates. However, it is worth considering the reasons for the absence of differences in detection rates among species in the urban area of our study.

As mentioned earlier, we believe that even small-scale areas within the study site may contribute to these findings. An 80,000 m2 area is relatively small for detecting eDNA variation. This might reflect abundant animal activity throughout the study site and/or a limited impact on microhabitats. These results could provide valuable insights into using eDNA to study urban wildlife species. For example, when conducting basic monitoring surveys of urban ecosystems using eDNA, this information could help determine the number and spatial distribution of sampling sites based on the size of park watersheds. Therefore, further research is crucial for understanding the subtle differences in detection rates in and between specific areas, considering how the life histories of animals affect their movement and the changing seasons, especially in urban areas. Overall, our study demonstrates the potential of eDNA metabarcoding as a tool for the efficient and versatile monitoring of species in urban ecological spaces.

Conclusion

Environmental DNA metabarcoding offers an efficient means to comprehensively assess species presence and frequency in a given space with relatively minimal effort. In this study, utilizing eDNA metabarcoding centered around the ecologically valuable Gildong Ecological Park, which allowed the demonstration of the potential of eDNA metabarcoding in monitoring urban animal populations. Through five rounds of repeated sampling and the analysis of 65 samples, we detected 73% (11/15) of mammalian species and 67% (represented 67% of the total 6268 individual) of the dominant bird species. Notable urban mammals, such as raccoons and squirrels, as well as species of concern, such as wild boars, and natural monuments designated that culturally significant species such as the mandarin duck and yellow bittern, were detected. Additionally, we observed changes in species composition over time by conducting the monitoring during the summer, autumn, and winter seasons.

Although we aimed to analyze differences in species detection across specific zones within the park, there were no significant differences in detection levels between marsh and forested areas. This suggests that microhabitat influence is minimal in relatively small parks, such as Gildong Ecological Park, which animals heavily use, and the space functions as a single ecosystem. Therefore, sampling for species monitoring is versatile, without considering the specific location within the park. However, more frequent observations were beneficial in some areas, especially for birds, where seasonal variations in species composition were detected. Furthermore, we found variations in detection levels for different species across seasons, indicating a potential for using eDNA in temporal species monitoring. While this study focused on mammals and birds, future research should be expanded to include other indicator species of urban ecosystems, such as amphibians, reptiles, and insects. This study can be a foundational exploration into using eDNA for future urban ecosystem research.