Crowdsourced Indicators of Flora and Fauna Species: Comparisons Between iNaturalist Records and Field Observations

Abstract

Cultural ecosystem services provide intangible benefits such as recreation and aesthetic enjoyment but are difficult to quantify compared to provisioning or regulating ecosystem services. Recent technologies offer alternative indicators, such as social media data, to identify popular locations and their features. This study demonstrates how large volumes of citizen science and social media data can be analyzed to reveal patterns of human interactions with nature through unconventional, scalable methods. By applying spatial statistical methods, data from the citizen science platform iNaturalist are analyzed and compared with ground-truth visitation data. To minimize data bias, records are grouped by taxonomic information and applied to the metropolitan area of Seoul, South Korea (2005–2022). The taxonomic information included in the iNaturalist data were investigated using a standard global biodiversity database. The results show citizen science data effectively quantify public preferences for scenic locations, offering a novel approach to mapping cultural ecosystem services when traditional data are unavailable. This method highlights the potential of large-scale citizen-generated data for conservation, urban planning, and policy development. However, challenges like bias in user-generated content, uneven ecosystem coverage, and the over- or under-representation of locations remain. Addressing these issues and integrating additional metadata—such as time of visit, demographics, and seasonal trends—could provide deeper insights into human–nature interactions. Overall, the proposed method opens up new possibilities for using non-traditional data sources to assess and map ecosystem services, providing valuable information for conservation efforts, urban planning, and environmental policy development.

1. Introduction

We obtain numerous benefits from nature through various forms of relationships with it [1,2]. These diverse values of nature include instrumental (e.g., health benefits of recreation), intrinsic (e.g., the non-use value of species), and relational values (e.g., sense of place, cultural meaning of activities) [2]. Collectively, these non-material contributions of nature to people are often referred to as cultural ecosystem services (CESs), encompassing physical and mental health benefits, landscape aesthetics, and social relations [3,4,5]. CESs play a significant role in enhancing human well-being, fostering enjoyment, and creating a sense of place imparted by nature [6].

Species, as integral components of ecosystems, are pivotal in sustaining the cultural benefits derived from nature [7]. The growing recognition of the role of species and biodiversity in human–nature relationships highlights their importance [8,9]. For instance, certain species enhance the aesthetic and recreational value of landscapes through activities such as wildlife watching [10,11], hiking [12,13], and photography [14,15]. In many regions, specific species hold cultural significance, forming the basis for traditional cultural practices, rituals, and festivals, thereby symbolizing cultural heritage and collective memory [16].

Despite these benefits, increasing urbanization has led to a growing disconnect between people and nature [17,18]. Urban areas, which host dense human populations, present unique challenges and opportunities for biodiversity conservation and the CESs derived from it [19,20]. Urban environments often support diverse species, making them critical for understanding human–nature interactions [21,22]. Activities such as species observation in urban areas not only raise public awareness but also provide valuable data on wildlife distribution and behavior in rapidly changing environments [23,24]. Additionally, the use of digital tools like citizen science platforms and mobile apps amplifies these efforts by enabling broader participation in biodiversity monitoring [25]. These tools have significant social impacts, fostering inclusivity and community engagement, bridging gaps between urban residents and nature, and promoting environmental stewardship among diverse populations [26]. As urbanization is predicted to expand to 68% of the global population by 2050 [27], understanding how urban residents engage with nature and species is essential for fostering a stronger human–nature connection. By enabling public participation in biodiversity monitoring, these tools can provide valuable data to inform conservation strategies, identify at-risk species, and track ecological changes in urban environments [28,29].

However, compared to other ecosystem services, CESs and biodiversity remain challenging to quantify and map, which hinders effective natural resource management [30,31]. To address this, alternative approaches, such as crowdsourced social media data, have been employed to explore human–species relationships and patterns in nature appreciation [32,33]. In recent years, crowdsourced data have proven invaluable for environmental monitoring and species conservation [34,35,36]. These data not only help analyze how people enjoy nature, e.g., [37], but also aid in data acquisition for biodiversity monitoring [38]. Engaging the public in data collection provides information on species distribution, behavior, and population trends that might otherwise be difficult to gather through traditional scientific methods [36,39]. For example, the use of crowdsourced data in species distribution models has more than doubled since the 2010s [36]. By engaging the public in data collection, this collaborative approach enhances our understanding of human–species connections while empowering communities to actively participate in the preservation of biodiversity and the health of our natural environments [23,40,41]. This is particularly significant in urban areas, where crowdsourced species data contribute to a deeper understanding of urban diversity [42].

South Korea is one of the countries experiencing strong industrialization and urbanization [43]. As a result, it has become one of the most urbanized countries globally, with 18.3% of its population residing in Seoul, the capital of the country. This figure rises to 45% when including the greater Seoul Capital Area [44]. Addressing the well-being of this rapidly growing urban population has driven efforts to better understand species dynamics and ecosystem services in the region [45]. For example, Park and Han [45] analyzed the ecological network functions in Seoul, and Jo et al. [46] highlighted the importance of biodiversity enhancement in urban forests for Seoul citizens. Additionally, Serret et al. [47] investigated pollinator families observed in Seoul from 2016 to 2018 using a citizen science program. Despite these efforts, comprehensive information about the distribution of various species and the specific locations people visit remains limited.

To address this gap, we investigate crowdsourced data to better understand the relationship between species and humans. The objectives of this study are to:

• Identify species distribution patterns in the Seoul Capital Area using crowdsourced data;
• Evaluate the consistency of these data with field observations;
• Determine visitors’ preferences for different types of species.

In addition, we explore methodological advancements to better analyze and interpret crowdsourced data.

2. Methods and Materials

2.1. Study Area

This study focuses on the Seoul Capital Area in South Korea (Figure 1). The Seoul Capital Area, located in northwest South Korea, includes the city of Seoul (the capital), the Incheon metropolis, and Gyeonggi Province (Figure 1). With an area of 12,685 km², the Seoul Capital Area is home to approximately 26 million people, of which nearly 10 million reside in Seoul. Tourism in this area is substantial as the area serves many domestic and international visitors [44].

2.2. Visitor Statistics

The Korean government publishes the visitor statistics regularly [48], which provide the number of visitors to major tourist destinations across the country. The dataset aims to generate national statistics to estimate tourist demand. Additionally, it evaluates the capacity of tourism facilities and supports planning for resource development. The raw data are collected quarterly by local governments and individual tourist sites, and subsequently reported to the Ministry of Culture, Sports, and Tourism (MCST). The dataset allows for comparisons across different regions and time periods [48].

From 2005 to 2022, the visitor statistics data in the study area covered the number of visitors and locations for the total 694 tourist sites, including natural and cultural attractions (Figure 1). We converted the tourism sites to a set of spatial points using the given coordinates and categorized sites by types of area, namely, ‘sightseeing’, ‘nature’, and ‘culture’, for further analysis following the definition provided by the Korea Culture and Tourism Institute [48]. Generally, indoor facilities such as museums are categorized into the ‘sightseeing’ category, outdoor sightseeing areas such as palaces and temples are categorized into the ‘culture’ category, and natural parks and mountains are categorized into the ‘nature’ category. The information of visitors was categorized into international visitors and domestic visitors depending on their home location. The proportion of domestic visitors was much larger than that of foreigners. Note that the distinction between domestic and foreign visitors is not utilized in this study.

2.3. iNaturalist Database

The iNaturalist platform was utilized to estimate the species distribution patterns by analyzing citizen-science records of biodiversity in the Seoul Capital Area. iNaturalist is a widely used cloud-based platform that enables users to upload photographs of flora, fauna, and insects, which can then be taxonomically identified within the platform. The data from the iNaturalist were acquired via the iNaturalist API and the ‘rinat’ package v0.1.9 [49] for the study period. In the Seoul Capital Area, the acquired data consist of a total of 75,373 records, with 4398 common names and 9049 scientific names. The dataset covers 13 taxonomic groups (kingdom/phylum/class). The breakdown by taxonomic level shows 6 kingdoms, 1 division, and 6 classes: Animalia, Plantae, Insecta, Fungi, Protozoa, and Chromista; Mollusca (phylum), Aves (class), Mammalia (class), Reptilia (class), Amphibia (class), Arachnida (class), and Actinopterygii (class). Most of the downloaded iNaturalist data (99.4%) include scientific names, while 71.6% include common names. We excluded records without scientific names and those labeled as ‘Homo Sapiens’ leaving us with 74,889 records.

2.4. GBIF Database

It is known that the iNaturalist database introduces some uncertainties in its taxonomic classification [28]. To address these uncertainties, we compared the iNaturalist data with the Global Biodiversity Information Facility (GBIF) [50], a global repository that aims to provide a more structured and standardized biodiversity database, incorporating data obscuration where necessary [28,51]. To retrieve standardized taxonomic information for iNaturalist species, we used the scientific names recorded in iNaturalist to query the GBIF database. The data retrieval process was carried out in GNU R, using the ‘rgbif’ package [52,53]. In addition to taxonomic data, we also retrieved the Red List categories from the International Union for Conservation of Nature (IUCN) for species found in iNaturalist. The IUCN Red List provides the conservation status of species, classifying them into nine categories: Extinct (EX), Extinct in the Wild (EW), Critically Endangered (CR), Endangered (EN), Vulnerable (VU), Near Threatened (NT), Least Concern (LC), Data Deficient (DD), and Not Evaluated (NE) [54].

2.5. Point Pattern Analysis

We performed a point pattern analysis using the Cross-K function to assess the spatial relationship between two distinct point patterns: the locations of MCST visitors and the locations of iNaturalist records. Cross-K, an extension of the K function, is utilized to investigate spatial clustering or dispersion between two datasets, primarily for understanding spatial dependencies in tourism and biodiversity data. The Cross-K function computes the expected number of points from one dataset within a specified distance of points from the other, allowing spatial relationships between the two point patterns to be evaluated based on the distribution of one set of points relative to the other [55,56].

Compared to traditional grid-based analyses (e.g., [31,37]), the Cross-K function directly accounts for the spatial configuration of individual points, rather than relying on aggregated gridded data. We adopted this approach to identify underlying spatial patterns that may not be captured by raster-based methods, thereby enabling a more nuanced examination of spatial dependencies.

In this study, the Cross-K function measures how the presence of one set of points (e.g., visitor statistics) is related to the presence of another set (e.g., iNaturalist records). The Cross-K statistic typically measures the difference in spatial patterns between two datasets relative to a Complete Spatial Random (CSR) process, which serves as a null model of spatial randomness [56,57]. Under CSR, points are distributed randomly and independently across the given area. In this study, the CSR model acts as a baseline against which the observed data are compared. The difference between the spatial distribution of the iNaturalist data and CSR represents the deviation of the observed patterns from those expected under CSR. This difference indicates whether the iNaturalist records are clustered, dispersed, or randomly distributed around the locations of visitors compared to a random pattern.

As our study is irregularly shaped, a translation correction was applied in the Cross-K analysis [56]. This correction addresses boundary effects in spatial point pattern analysis by adjusting or shifting the spatial data, thereby mitigating edge bias and ensuring a more accurate representation of spatial relationships [56].

The Cross-K function was evaluated for every 500 m from 0 to 10 km. The 500 m interval was chosen to capture fine-scale spatial relationships while encompassing a 10 km radius to account for broader spatial interactions. To summarize the metrics, we normalized the differences between the estimates of the Visitor–iNaturalist process and the CSR process. These differences were aggregated and normalized on a scale from −1 (representing the largest repulsion) to 1 (representing the largest attraction), across all taxonomic groups. The visitation counts from the MCST visitation statistics and the population size of the iNaturalist records were used as weights in the calculations. In addition to the Cross-K analysis, we visualized the patterns using kernel density estimation, employing a Gaussian kernel with a parameter of $\sigma =5$, which controls the width of the kernel [58]. All computations and visualizations were performed in GNU R, primarily using the ‘spatstat’ v3.3-0 [59] and ‘stats’ packages in GNU R v4.4.2. [60]

3. Results

3.1. Descriptive Statistics

The spatial distribution of the visitor statistics is shown in Figure 1. For the study period (2005–2022), the number of visitors in each category varies significantly. General sightseeing, such as museums and attractions, received the highest number of visitors (1175 million) compared to other types ‘culture’ (205 million) and ‘nature’(105 million). Visitors at these sightseeing spots are concentrated in central Seoul and the northern Gyeonggi-do area near the DMZ (Demilitarized Zone) (Figure 1). Areas classified as ‘nature’, which include mountains and parks, are distributed across the whole region, while ‘culture’ sites are concentrated in densely populated areas. Figure 2 shows the temporal variations in the number of visitors over the study period. The number of visitors significantly decreased starting in 2020, presumably due to the lockdown in South Korea, which lasted until 2022. The visitor statistics suggest that tourism activities were restricted due to the pandemic, leading to fewer movements.

Figure 3 shows the distribution of a total of 74,889 records in the study area. The majority of the iNaturalist data are categorized under the taxa Insecta (40.58%), Plantae (24.49%), and Aves (20.65%), followed by Arachnida (4.00%), Amphibia (3.15%), Fungi (2.35%), Animalia (1.4%), Mammalia (1.23%), Mollusca (0.81%), Reptilia (0.75%), Actinopterygii (0.60%), Protozoa (0.02%), and Chromista (0.01%) (Appendix A

Table A1. List of the top 30 species from iNaturalist data in the study area and their frequencies. Some taxonomic groups contain fewer than 30 species (e.g., Chromista). Note that missing common names in the original data are left intentionally to illustrate how the data are presented.

Taxa	Scientific Name	Common Name	Freq
Actinopterygii	Cyprinus rubrofuscus	Amur Carp	69
Actinopterygii	Cyprinus	Typical Carps	35
Actinopterygii	Actinopterygii	Ray-finned Fishes	20
Actinopterygii	Zacco platypus	Pale Chub	16
Actinopterygii	Oxudercinae	Mudskippers	13
Actinopterygii	Rhynchocypris oxycephalus	Chinese Minnow	13
Actinopterygii	Carassius auratus	Goldfish	12
Actinopterygii	Gobiidae	Gobies	10
Actinopterygii	Channa argus	Northern Snakehead	9
Actinopterygii	Micropterus salmoides	Largemouth Bass	9
Actinopterygii	Monopterus albus	Asian swamp eel	8
Actinopterygii	Misgurnus anguillicaudatus	Oriental Weatherfish	7
Actinopterygii	Misgurnus mizolepis	Chinese Weatherfish	7
Actinopterygii	Mugil cephalus	Sea Mullet	7
Actinopterygii	Periophthalmus modestus	Shuttles hoppfish	7
Actinopterygii	Silurus asotus	Amur Catfish	7
Actinopterygii	Odontobutis interruptus		6
Actinopterygii	Pseudogobio esocinus	Pike Gudgeon	6
Actinopterygii	Pungtungia herzi	Black Stripe Gudgeon	6
Actinopterygii	Acanthogobius hasta	Javelin Goby	5
Actinopterygii	Misgurnus	weatherfishes	5
Actinopterygii	Pseudorasbora parva	Topmouth Gudgeon	5
Actinopterygii	Carassius	Crucian Carps	4
Actinopterygii	Chanodichthys erythropterus	predatory carp	4
Actinopterygii	Coreoperca herzi		4
Actinopterygii	Cyprinidae	Cyprinids	4
Actinopterygii	Cyprinus carpio carpio	Mirror Carp	4
Actinopterygii	Koreocobitis rotundicaudata	White Nose Loach	4
Actinopterygii	Rhinogobius	Rhinogobies	4
Actinopterygii	Rhodeus suigensis		4
Amphibia	Hyla japonica	Japanese Tree Frog	615
Amphibia	Pelophylax nigromaculatus	Black-spotted Frog	289
Amphibia	Hyla suweonensis	Suweon Tree Frog	206
Amphibia	Pelophylax chosenicus	Golden Pond Frog	148
Amphibia	Bufo sachalinensis	Sakhalin toad	142
Amphibia	Kaloula borealis	Boreal Digging Frog	130
Amphibia	Hynobius leechii	Gensan Salamander	123
Amphibia	Rana uenoi	Ueno’s Brown Frog	123
Amphibia	Bombina orientalis	Oriental Fire-bellied Toad	98
Amphibia	Onychodactylus koreanus	Korean Clawed Salamander	79
Amphibia	Glandirana emeljanovi	Imienpo Station frog	77
Amphibia	Rana huanrensis	Huanren Frog	66
Amphibia	Rana	Pond Frogs	60
Amphibia	Rana coreana	Korean Brown Frog	52
Amphibia	Lithobates catesbeianus	American Bullfrog	50
Amphibia	Anura	Frogs and Toads	21
Amphibia	Ranidae	Typical Frogs	21
Amphibia	Bufo stejnegeri	Water Toad	19
Amphibia	Hyla	Holarctic Treefrogs	16
Amphibia	Amphibia	Amphibians	6
Amphibia	Pelophylax	Water Frogs	6
Amphibia	Ambystoma mexicanum	Axolotl	4
Amphibia	Ambystomatidae	Mole Salamanders	1
Amphibia	Caudata	Salamanders	1
Amphibia	Duttaphrynus melanostictus	Asian Common Toad	1
Amphibia	Leptodactylus fallax	Mountain Chicken	1
Amphibia	Rana latastei	Italian Agile Frog	1
Amphibia	Ranoidea caerulea	Australian Green Tree Frog	1
Animalia	Orthomorphella pekuensis		61
Animalia	Arthropoda	Arthropods	49
Animalia	Thereuonema tuberculata	Japanese House Centipede	40
Animalia	Chiromantes haematocheir	Red-clawed crab	33
Animalia	Armadillidium vulgare	Common Pill Woodlouse	31
Animalia	Scolopendra mutilans	Chinese Red-headed Centipede	31
Animalia	Oxidus gracilis	Greenhouse Millipede	28
Animalia	Brachyura	True Crabs	25
Animalia	Polydesmida	Flat-backed Millipedes	21
Animalia	Bipaliinae	Hammerhead Worms	19
Animalia	Orisarma dehaani		19
Animalia	Tubuca arcuata	Bowed Fiddler Crab	18
Animalia	Animalia	Animals	16
Animalia	Diversibipalium		15
Animalia	Eriocheir sinensis	Chinese Mitten Crab	15
Animalia	Oligochaeta	Earthworms and Allies	15
Animalia	Oxidus		15
Animalia	Paradoxosomatidae		15
Animalia	Cambaroides similis	Korean Crayfish	14
Animalia	Eriocheir		14
Animalia	Armadillidium	Pillbugs	13
Animalia	Scutigeridae	Typical House Centipedes	13
Animalia	Ligia	Sea Slaters	12
Animalia	Macrophthalmus japonicus		12
Animalia	Whitmania edentula		12
Animalia	Collembola	Springtails	11
Animalia	Megascolecidae	Giant Earthworms	11
Animalia	Pagurus minutus		11
Animalia	Armadillidium nasatum	Nosy Pill Woodlouse	10
Animalia	Ligia exotica	Wharf Louse	10
Arachnida	Trichonephila clavata	Joro Spider	421
Arachnida	Araneae	Spiders	100
Arachnida	Argiope bruennichi	Wasp Spider	93
Arachnida	Lycosidae	Wolf Spiders	84
Arachnida	Thomisidae	Crab Spiders	82
Arachnida	Araneidae	Orbweavers	79
Arachnida	Salticidae	Jumping Spiders	79
Arachnida	Orienticius vulpes		66
Arachnida	Araneus ventricosus		64
Arachnida	Araneomorphae	Typical Spiders	63
Arachnida	Carrhotus xanthogramma		61
Arachnida	Telamonia vlijmi		51
Arachnida	Philodromidae	Running Crab Spiders	47
Arachnida	Myrmarachne	Ant mimic Spiders	40
Arachnida	Agelenidae	Funnel Weavers	37
Arachnida	Lycosoidea	Wolf Spiders and Allies	37
Arachnida	Pisaura		34
Arachnida	Araneinae	Typical Orbweavers	33
Arachnida	Ebrechtella tricuspidata	Triangle Crab Spider	32
Arachnida	Parasteatoda tepidariorum	Common House Spider	32
Arachnida	Tetragnatha	Long-jawed Orbweavers	32
Arachnida	Pardosa	Thin-legged Wolf Spiders	31
Arachnida	Evarcha albaria	White-eyebrowed Jumping Spider	30
Arachnida	Neoscona scylloides		30
Arachnida	Araneoidea	Araneoid Spiders	27
Arachnida	Asianellus festivus		27
Arachnida	Linyphiidae	Sheetweb and Dwarf Weavers	27
Arachnida	Plexippoides regius		26
Arachnida	Agelena	Grass Funnel-web Spiders	25
Arachnida	Dolomedes sulfureus		25
Aves	Pica serica	Oriental Magpie	896
Aves	Ardea cinerea	Grey Heron	551
Aves	Anas zonorhyncha	Eastern Spot-billed Duck	525
Aves	Passer montanus	Eurasian Tree Sparrow	478
Aves	Streptopelia orientalis	Oriental Turtle dove	476
Aves	Columba livia domestica	Feral Pigeon	459
Aves	Hypsipetes amaurotis	Brown-eared Bulbul	452
Aves	Anas platyrhynchos	Mallard	388
Aves	Phalacrocorax carbo	Great Cormorant	329
Aves	Phoenicurus auroreus	Daurian Redstart	326
Aves	Larus crassirostris	Black-tailed Gull	307
Aves	Parus minor	Japanese Tit	301
Aves	Ardea alba	Great Egret	238
Aves	Egretta garzetta	Little Egret	216
Aves	Fulica atra	Eurasian Coot	188
Aves	Tachybaptus ruficollis	Little Grebe	186
Aves	Sinosuthora webbiana	Vinous-throated Parrotbill	182
Aves	Corvus macrorhynchos	Large-billed Crow	177
Aves	Sittiparus varius	Varied Tit	169
Aves	Poecile palustris	Marsh Tit	167
Aves	Cyanopica cyanus	Azure-winged Magpie	150
Aves	Emberiza elegans	Yellow-throated Bunting	150
Aves	Aix galericulata	Mandarin Duck	149
Aves	Mergus merganser	Common Merganser	149
Aves	Anas crecca crecca	Eurasian Green-winged Teal	140
Aves	Aythya ferina	Common Pochard	140
Aves	Anser albifrons	Greater White-fronted Goose	137
Aves	Falco tinnunculus	Eurasian Kestrel	137
Aves	Fringilla montifringilla	Brambling	135
Aves	Turdus naumanni	Naumann’s Thrush	135
Chromista	Botrydium		1
Chromista	Chromista	kelp, diatoms, and allies	1
Chromista	Noctiluca scintillans	sea sparkle	1
Chromista	Ochrophyta		1
Chromista	Phaeophyceae	brown algae	1
Chromista	Saprolegnia	Cotton Moulds	1
Fungi	Fungi	Fungi Including Lichens	298
Fungi	Polyporales	shelf fungi	87
Fungi	Agaricales	Common Gilled Mushrooms and Allies	73
Fungi	Trametes		55
Fungi	Trametes versicolor	turkey-tail	48
Fungi	Schizophyllum commune	splitgill mushroom	40
Fungi	Russula	brittlegills	38
Fungi	Amanita	amanita mushrooms	37
Fungi	Agaricomycetes		33
Fungi	Basidiomycota	Basidiomycete Fungi	27
Fungi	Coprinus comatus	Shaggy Mane	19
Fungi	Polyporaceae	bracket fungi	17
Fungi	Boletaceae	boletes	16
Fungi	Dacrymyces spathularia	Fan-shaped Jelly Fungus	16
Fungi	Ganoderma	Artist’s Brackets, Reishi, and Allies	16
Fungi	Lecanoromycetes	common lichens	15
Fungi	Trametes sanguinea	Cinnabar Bracket	15
Fungi	Boletales	boletes and allies	13
Fungi	Macrolepiota procera	Parasol	12
Fungi	Parmelioideae	typical shield lichens	12
Fungi	Pholiota	Scalycaps	12
Fungi	Russulaceae	Milkcaps, Brittlegills and Allies	12
Fungi	Amidella	Amanita Sect. Amidella	10
Fungi	Phallus luteus		10
Fungi	Cladonia	Pixie Cup Lichens	9
Fungi	Coprinellus micaceus	mica cap	9
Fungi	Ganoderma lucidum	lacquered bracket	9
Fungi	Lactarius	Common Milkcaps	9
Fungi	Lycoperdon		9
Fungi	Lycoperdon perlatum	common puffball	9
Insecta	Lepidoptera	Butterflies and Moths	678
Insecta	Harmonia axyridis	Asian Lady Beetle	246
Insecta	Polygonia c-aureum	Asian Comma	212
Insecta	Lycorma delicatula	Spotted Lanternfly	198
Insecta	Tenodera sinensis	Chinese Mantis	198
Insecta	Apis mellifera	Western Honey Bee	176
Insecta	Lymantria dispar	Spongy Moth	175
Insecta	Hyalessa maculaticollis	Robust Cicada	165
Insecta	Hierodula patellifera	Giant Asian Mantis	162
Insecta	Statilia maculata	Asian Jumping Mantis	150
Insecta	Halyomorpha halys	Brown Marmorated Stink Bug	138
Insecta	Geometridae	Geometer Moths	133
Insecta	Pseudozizeeria maha	Pale Grass Blue	132
Insecta	Oedaleus infernalis		131
Insecta	Exomala orientalis	Oriental Beetle	128
Insecta	Riptortus pedestris		128
Insecta	Ricania sublimata		109
Insecta	Tenodera angustipennis	Narrow-winged Mantis	108
Insecta	Camponotus japonicus	Japanese Carpenter Ant	106
Insecta	Timomenus komarowi		106
Insecta	Pieris rapae	Small White	103
Insecta	Acrida cinerea	Oriental Longheaded Locust	100
Insecta	Chironomidae	Non-biting Midges	100
Insecta	Coccinella septempunctata	Seven-spotted Lady Beetle	99
Insecta	Formica japonica	Japanese Wood Ant	99
Insecta	Insecta	Insects	98
Insecta	Atractomorpha lata		94
Insecta	Ichneumonidae	Ichneumonid Wasps	94
Insecta	Sinochlora longifissa		93
Insecta	Papilio xuthus	Chinese Yellow Swallowtail	92
Mammalia	Sciurus vulgaris	Eurasian Red Squirrel	134
Mammalia	Felis catus	Domestic Cat	133
Mammalia	Hydropotes inermis	Water Deer	93
Mammalia	Eutamias sibiricus	Siberian Chipmunk	72
Mammalia	Hydropotes inermis argyropus	Korean Water Deer	36
Mammalia	Nyctereutes procyonoides	Mainland Raccoon Dog	31
Mammalia	Canis familiaris	Domestic Dog	30
Mammalia	Sus scrofa	Wild Boar	19
Mammalia	Mammalia	Mammals	15
Mammalia	Mustela sibirica	Siberian Weasel	14
Mammalia	Lutra lutra	Eurasian Otter	12
Mammalia	Oryctolagus cuniculus	European Rabbit	11
Mammalia	Meles leucurus	Asian Badger	9
Mammalia	Mogera robusta	Large Mole	9
Mammalia	Rattus norvegicus	Brown Rat	9
Mammalia	Prionailurus bengalensis	Mainland Leopard Cat	8
Mammalia	Rattus	Old World Rats	7
Mammalia	Capreolus pygargus	Eastern Roe Deer	6
Mammalia	Cervus nippon	Sika Deer	6
Mammalia	Martes flavigula	Yellow-throated Marten	6
Mammalia	Muroidea	Muroids	6
Mammalia	Soricidae	Shrews	6
Mammalia	Ailuropoda melanoleuca	Giant Panda	5
Mammalia	Ailurus		5
Mammalia	Apodemus agrarius	Striped Field Mouse	5
Mammalia	Lepus coreanus	Korean Hare	5
Mammalia	Muridae	Old World Mice and Rats	5
Mammalia	Nyctereutes procyonoides koreensis	Korean Raccoon Dog	5
Mammalia	Panthera tigris	Tiger	5
Mammalia	Cervidae	Deer	4
Mollusca	Karaftohelix kurodana		67
Mollusca	Meghimatium		46
Mollusca	Gastropoda	Gastropods	35
Mollusca	Acusta redfieldi		31
Mollusca	Sinanodonta lauta		28
Mollusca	Acusta		20
Mollusca	Nesiohelix samarangae		18
Mollusca	Anemina arcaeformis		16
Mollusca	Pomacea canaliculata	Channeled Apple Snail	15
Mollusca	Ambigolimax	Threeband Slugs	14
Mollusca	Stylommatophora	Common Land Snails and Slugs	14
Mollusca	Nodularia douglasiae		13
Mollusca	Meghimatium fruhstorferi		12
Mollusca	Rapana venosa	Veined Rapa Whelk	12
Mollusca	Helicoidea		11
Mollusca	Littorina brevicula		11
Mollusca	Meghimatium bilineatum	Chinese Slug	8
Mollusca	Reishia clavigera		7
Mollusca	Aegista		6
Mollusca	Umbonium thomasi	Sand Snail	6
Mollusca	Bullacta caurina		5
Mollusca	Karaftohelix		5
Mollusca	Magallana gigas	Pacific Oyster	5
Mollusca	Pomacea	Common Apple Snails	5
Mollusca	Bekkochlamys subrejecta		4
Mollusca	Helicina		4
Mollusca	Ostreidae	True Oysters	4
Mollusca	Semisulcospira libertina		4
Mollusca	Acanthochitona achates		3
Mollusca	Batillariidae	Mud creeper Snails	3
Plantae	Plantae	plants	557
Plantae	Magnoliopsida	dicots	456
Plantae	Angiospermae	flowering plants	336
Plantae	Erigeron annuus	annual fleabane	202
Plantae	Tracheophyta	vascular plants	157
Plantae	Pinus	pines	138
Plantae	Bryophyta	mosses	134
Plantae	Setaria viridis	Green Bristle Grass	133
Plantae	Chelidonium asiaticum		124
Plantae	Asteraceae	sunflowers, daisies, asters, and allies	122
Plantae	Trifolium repens	white clover	112
Plantae	Commelina communis	Asiatic dayflower	110
Plantae	Pinus densiflora	Japanese red pine	110
Plantae	Taraxacum officinale	common dandelion	107
Plantae	Prunus	plums, cherries, and allies	103
Plantae	Rosa	roses	96
Plantae	Parthenocissus tricuspidata	Japanese creeper	91
Plantae	Rhododendron	rhododendrons and azaleas	89
Plantae	Acer palmatum	Japanese maple	86
Plantae	Rhododendron mucronulatum	Korean rhododendron	86
Plantae	Hibiscus syriacus	common hibiscus	80
Plantae	Prunus serrulata	Japanese Cherry	79
Plantae	Capsella bursa-pastoris	shepherd’s-purse	76
Plantae	Acer	maples	74
Plantae	Ginkgo biloba	ginkgo	72
Plantae	Cornus officinalis	Korean cornel dogwood	70
Plantae	Poaceae	grasses	69
Plantae	Coreopsis lanceolata	Lance-leaved Coreopsis	65
Plantae	Phytolacca americana	American pokeweed	65
Plantae	Quercus	oaks	62
Protozoa	Stemonitis	Chocolate Tube Slimes	3
Protozoa	Ceratiomyxa	Coral Slimes	2
Protozoa	Hemitrichia serpula	Pretzel slime mold	2
Protozoa	Mycetozoa	slime molds	2
Protozoa	Arcyria denudata	carnival candy slime mold	1
Protozoa	Lycogala		1
Protozoa	Myxomycetes	true slime molds	1
Protozoa	Physarum melleum		1
Reptilia	Elaphe dione	Steppe Ratsnake	65
Reptilia	Rhabdophis tigrinus	Tiger Keelback	52
Reptilia	Gloydius ussuriensis	Ussuri Mamushi	51
Reptilia	Oocatochus rufodorsatus	Frog-eating Rat Snake	37
Reptilia	Pelodiscus maackii	Amur Softshell Turtle	36
Reptilia	Hebius vibakari	Japanese Keelback	27
Reptilia	Takydromus wolteri	Mountain Grass Lizard	27
Reptilia	Gloydius brevicauda	Short-tailed Mamushi	25
Reptilia	Trachemys scripta	Pond Slider	23
Reptilia	Pseudemys concinna	River Cooter	21
Reptilia	Takydromus amurensis	Amur Grass Lizard	19
Reptilia	Elaphe schrenckii	Manchurian Black Ratsnake	18
Reptilia	Scincella vandenburghi	Tsushima Ground Skink	18
Reptilia	Lycodon rufozonatus	Red-banded Snake	16
Reptilia	Trachemys scripta elegans	Red-eared Slider	10
Reptilia	Eremias argus	Mongolia Racerunner	9
Reptilia	Takydromus	Grass Lizards	9
Reptilia	Mauremys reevesii	Chinese Pond Turtle	8
Reptilia	Gloydius intermedius	Central Asian Pitviper	5
Reptilia	Mauremys sinensis	Common thread turtle	5
Reptilia	Pseudemys nelsoni	Florida Redbelly Turtle	5
Reptilia	Pseudemys peninsularis	Peninsular Cooter	5
Reptilia	Trachemys scripta scripta	Yellow-bellied Slider	5
Reptilia	Colubridae	Colubrid Snakes	4
Reptilia	Deirochelyinae	Deirochelyine Turtles	4
Reptilia	Scincella	Ground Skinks	4
Reptilia	Pseudemys	Cooters	3
Reptilia	Rhabdophis tigrinus tigrinus		3
Reptilia	Aldabrachelys gigantea	Aldabra Giant Tortoise	2
Reptilia	Centrochelys sulcata	African Spurred Tortoise	2

). Given the high prevalence of taxonomic information in most records, iNaturalist data in the area primarily contain information on insects, plants, and birds (Figure 1). The iNaturalist taxonomic group information is not perfectly standardized as the taxonomic levels vary across groups. In particular, the Animalia category contains both freshwater fishes and unidentified Arthropods, leading to heterogeneity within the group. Note that we exclude Chromista in the following analysis due to the extremely low number of records (n = 6). Figure 4 shows the temporal variation in the number of iNaturalist records over the study period. In contrast to the visitor statistics, iNaturalist records actually increased over time, primarily driven by a higher number of bird-related data points in 2020. This suggests that more people went outdoors to photograph birds and uploaded the data to the platform.

3.2. Spatial Point Patterns

To visualize spatial point patterns, kernel density estimation was performed for the visitor statistics. The density estimates were visualized separately for four categories: (a) all visitors together, (b) visitors to cultural heritage sites, (c) visitors to natural sites, and (d) visitors to general tourist facilities (Figure 5). Kernel density estimation was also performed for iNaturalist records. Among all taxonomic groups, the densities of the groups clustered with natural sites are shown (Figure 6). A visual inspection indicates a potential correlation between the iNaturalist record density and the visitor statistics, which revealed iNaturalist records are somewhat concentrated around natural tourism sites.

3.3. Comparison of iNaturalist and Visitor Statistics

To quantify the relationships more solidly, a Cross-K analysis was conducted to provide a more rigorous statistical comparison between the two datasets. In Figure 7, Cross-K analysis is summarized to illustrate the relationship between the visitor statistics and the iNaturalist records, both in the form of spatial point patterns. When considering all iNaturalist records, no clear association between visitor statistics and the crowdsourced data was observed. However, when analyzed by type of tourism site, certain taxonomic groups showed meaningful associations with visitor statistics. As shown in Appendix A Figure A1, for the selected taxonomic groups, the observed point patterns (solid black line) show higher K values compared to the random process (dotted red line). At nature-based tourism sites, Animalia, Plantae, Amphibia, Mollusca, and Aves exhibited strong clustering. It should be noted that in the dataset, Animalia is somewhat inconsistently categorized, with species including worms and freshwater fishes as well as unidentified species (e.g., scientific name recorded simply ’Animalia’). For Mollusca, the observed associations suggest that snails and slugs are commonly found in nature protection areas that attract visitors, allowing for further interpretation (see species list in Appendix A Table A1).

In contrast, cultural and sightseeing sites exhibited a negative relationship with all taxonomic groups in relation to visitor statistics. The results of the Cross-K analysis showed that most taxonomic groups exhibited strong negative relationships with the number of visitors to cultural tourism sites. This was especially evident after the application of translation correction. Note that in the preliminary analysis, when the baseline Poisson model was applied, a positive relationship was observed across all site types.

3.4. Supplementing iNaturalist Data with Information from the GBIF Database

The comparisons between the taxonomic group information from iNaturalist and the GBIF database are presented in Figure 8 and Figure 9, illustrating how the iNaturalist taxonomic group data differs from the GBIF database at various taxonomic levels. Note that we queried GBIF data for each of the Top 30 species within each iNaturalist taxa group, leaving us with a total of 342 species. In Figure 8, the diagram on the left represents a kingdom-level comparison, showing that a large portion of iNaturalist records fall into the Animalia kingdom. Some records marked as Animalia and Mammalia in iNaturalist data belong to the Plantae kingdom, suggesting possible misclassification in the iNaturalist data. The Phylum-level comparison (Figure 8 on the right) breaks the iNaturalist taxonomic groups into more specific phyla, such as Arthropoda, Chordata, and Basidiomycota. This reveals that the Animalia group in iNaturalist data is substantially mixed up, for example, Arthropoda should not have been included. Except for species belonging to Arthropoda, most of the iNaturalist Animalia group species are worms and algae.

The class-level comparison (Figure 9 on the left) indicates that iNaturalist taxonomic groups are primarily classified at this level, with many groups closely matching their respective classes, except for Plantae and Chromista, aside from Animalia. The order-level comparison (Figure 9 on the right) provides evidence that iNaturalist groups are beyond this level.

We calculated the frequency and distribution of IUCN Red List categories for species found in the iNaturalist data (Figure 10 and Appendix A

Table A2. IUCN Red List categories for the top 30 species in each iNaturalist taxa group from the study area. Note that the IUCN code could not be retrieved from the GBIF database for 10 species.

IUCN Red List Category	Code	Number of Species	%
Not Evaluated	NE	216	65.06%
Data Deficient	DD	2	0.60%
Least Concern	LC	94	28.31%
Near Threatened	NT	4	1.20%
Vulnerable	VU	6	1.81%
Endangered	EN	7	2.11%
Critically Endangered	CR	3	0.90%
Extinct in the Wild	EW	0	0.00%
Extinct	EX	0	0.00%
(No Data)	-	10	0.03%
Total	-	342	100.00%

). Among the total 342 species from the top 30 species within each iNaturalist taxa group, we retrieved an IUCN class code from the GBIF database for 332 species. Of these, 216 species are categorized as Not Evaluated (NE), followed by Least Concern (LC) (n = 94). The Extinct (EX) and Extinct in the Wild (EW) categories were not identified (Figure 10). Smaller proportions are seen in the Near Threatened (NT) category (four species, 1.20%), Vulnerable (VU) category (six species, 1.81%), and Endangered (EN) category (seven species, 1.79%). The IUCN classification suggests that the study area may not host a large number of threatened species; however, this is insufficient evidence since 65.8% of the species remain unevaluated (NE), and their conservation status is therefore unknown.

4. Discussions

This study clustered visitor statistics in the Seoul Capital Area using data from a crowdsourced social media database, where users upload species observations. The iNaturalist data were found to be heavily concentrated in tourist hotspots within Seoul’s urban areas and northern Gyeonggi Province. However, categorizing the data by taxonomic groups revealed significantly different patterns. Notably, while the overall correlation between visitor statistics and iNaturalist data is weak, specific taxonomic groups showed strong clustering with ecotourism sites compared to cultural sites, whereas others exhibited negative patterns. These findings suggest that motivations for visiting cultural sites are less related to species observation, whereas natural sites show higher correlations with certain taxonomic groups. For instance, specific groups demonstrated strong positive cross-K relationships with natural sites, though preferences varied among species [61].

Animalia, Aves, Mollusca, and Plantae, in particular, demonstrated a positive cross-K relationship with visitor numbers in natural areas (Figure 7). Charismatic species appear to attract more visitors and foster pro-conservation behaviors [62]. Among the iNaturalist observations, Insecta, Plantae, and Aves were the most frequently recorded groups. Birdwatching, which accounted for the highest number of animal-related records, is especially popular in urban areas due to its psychological benefits and positive effects on human well-being [63]. However, iNaturalist user preferences often exhibit biases toward visually striking or culturally significant species. This leads to the over-representation of charismatic fauna and flora, such as large mammals and colorful birds, while less conspicuous organisms, like fungi and invertebrates, are under-represented [64,65]. Geographical factors, such as road density and accessibility, also influence observation patterns [66].

Given these biases, interpreting the causality of observed relationships requires caution. For example, while people may prefer freshwater environments where Mollusca and fish species are found [67], it remains unclear whether visitors specifically sought these areas for species observation. Over-reliance on simple correlations or iNaturalist data alone may lead to misleading conclusions, as demonstrated in Potsikas et al. [68], Cao and Hochmair [69]. The platform is more suited to exploring the discovery of plant and animal species in relation to the CES of ‘non-use values’ [61]. Complementing social media data with field observations and interviews could provide a deeper understanding of visitor motivations, particularly for cultural and natural sites. While crowdsourced data are rich, they may over- or under-represent specific activities or species depending on location [31,70]. Moreover, incorporating ground data at the national level [71] and enhancing global datasets could align local and global biodiversity records [72]. Such integrated approaches can prioritize conservation planning for endangered species and mitigate biodiversity loss.

Social media data have been widely used to analyze CES, often through aggregated methods like photo-user days (PUDs) using platforms such as Flickr [73]. While these methods offer insights into CES distribution, they risk obscuring finer-scale patterns. This study addresses these limitations by employing point pattern analysis, which preserves the granularity of data. This approach improves the accuracy of spatial analyses and helps identify localized patterns essential for biodiversity conservation and ecosystem management [74].

While the use of social media data for species distribution and CES analysis is cost-effective and captures spatiotemporal patterns of ecosystems [70,71,75], these findings must be contextualized within broader social–environmental frameworks. For instance, how can such patterns inform global discussions on sustainable tourism, biodiversity conservation, and equitable access to natural spaces? This study demonstrates the potential of social media platforms to engage broader audiences in biodiversity awareness but highlights the need for critical evaluations of biases and uneven ecosystem representations [71].

These comparisons reveal how social media-based observations may over-represent certain taxonomic groups, particularly those that are more visible or commonly observed, while GBIF’s more standardized records might show a more even distribution across different groups. The differences between these two datasets underscore the strengths and limitations of each, suggesting the complementary nature of combining both sources for a more holistic understanding of global biodiversity.

Despite its value, the iNaturalist database has limitations, including inconsistent taxonomic classifications and biases in species representation. Many species remain Not Evaluated (NE) under the IUCN Red List within the Seoul Capital Area, limiting biodiversity assessments. Moreover, geographical factors like road connectivity and accessibility may further influence data collection and observation patterns [61]. Future studies should prioritize integrating multiple databases to enhance reliability and reduce biases in conservation planning [31,76].

This research not only advances understanding within the Seoul Capital Area but also contributes to global knowledge on the intersections of visitor behavior, species observation, and ecosystem services. By integrating diverse data sources and considering local socio-ecological contexts, future studies can offer more comprehensive assessments. Ultimately, this study underscores the dual potential of social media data: as a research tool and as a means to inspire inclusive global dialogues on human–nature relationships.

5. Conclusions

This study highlights the potential of crowdsourced data, such as iNaturalist, to provide valuable insights into biodiversity and CESs while addressing its limitations. Patterns of species observations strongly align with ‘nature’ areas, reflecting ecotourism motivations, whereas cultural sites show weaker correlations. Charismatic species attract more visitors and promote conservation awareness, but biases in user preferences and geographic accessibility lead to under-representation of less conspicuous taxa. While social media data are cost-effective and capture spatial and temporal trends, their biases necessitate integration with field observations and alternative datasets for comprehensive assessments. Beyond the Seoul Capital Area, this research underscores the role of citizen science in advancing biodiversity conservation and sustainable tourism, advocating for inclusive approaches that bridge public engagement and equitable access to natural spaces. Moreover, the findings can inform governmental decision-making processes, supporting the development of evidence-based policies for biodiversity conservation and sustainable land use planning.