Can citizen science fill knowledge gaps for the world’s most speciose and poorly-known insular amphibian fauna?

Introduction

Citizen science is a rapidly growing source of biodiversity information for scientific research and conservation planning (Amano et al. 2016; Roger et al. 2023). Easily accessible online platforms are increasingly underpinning regional biotic inventories in poorly-known areas (Suprayitno et al. 2022; Wangyal et al. 2022), detection of changes in the distribution of both invasive and non-invasive species (Pecl et al. 2019; Rowley et al. 2019), and even aspects of species discovery and description (Donnellan et al. 2023; Otani et al. 2024). Citizen science platforms may be particularly good for addressing knowledge gaps for rare or sporadically active species, or species that occur in areas that are populated but rarely visited by scientists (Roger et al. 2023). Carefully targeted and well-curated citizen science data are also informing inference of threat status for species of conservation concern (Farquhar et al. 2023).

Based on its successes, scientists and managers now view citizen science as one key strategy to help address some biodiversity knowledge gaps. However, this understanding is tempered by evidence that there are also many social, economic, and logistic constraints on the collection and use of data through citizen science (Collen et al. 2008; Amano and Sutherland 2013; Kass et al. 2022). The strengths and weakness of citizen science are particularly apparent in many tropical regions. These areas are home to the richest (Raz et al. 2024) yet most threatened and poorly-known concentrations of biodiversity on the planet (Amano and Sutherland 2013; Carlen et al. 2024) and as such, even casual observations have great potential to rapidly advance knowledge. However, in areas with low human development scores and relatively poor infrastructure, impediments to the effective use of citizen science platforms are also most pervasive (Daru and Rodriguez 2023; Carlen et al. 2024; López-Guillén et al. 2024). Identification of citizen science records for many taxa is further constrained by inadequate taxonomic frameworks and a shortage of expertise (the taxonomic impediment), issues that are again more pervasive in tropical biotas (Kass et al. 2022). In the context of these challenges, targeted reviews of how effectively citizen science is filling knowledge gaps in diverse tropical faunas may help to identify key strengths and weaknesses, and inform strategies to overcome or mitigate the latter.

Melanesia spans from eastern Indonesia around the islands of Maluku through to Fiji. The region is dominated by New Guinea, the world’s largest and highest tropical island that is home to the world’s most diverse insular frog, bird, and plant biotas (Camara-Leret et al. 2020; Oliver et al. 2022; Prasetya et al. 2023). The great biotic richness of Melanesia is juxtaposed against an incomplete biotic inventory with substantial gaps in taxonomic knowledge and distributional sampling for a range of different taxa (Camara-Leret et al. 2020; Kass et al. 2022; Oliver et al. 2022). These data gaps reflect long-term and ongoing logistical and social constraints on fieldwork in Melanesia (Camara-Leret et al. 2020). However, while most of Melanesia remains difficult to access for professional scientists, parts of the region are becoming increasingly developed and more accessible to tourists and other visitors (Tiku and Shimizu 2020). Access to mobile networks and smartphones across the region is also improving, although still patchy, providing opportunities for citizen science to fill biodiversity knowledge gaps in some parts of Melanesia (Maharani et al. 2022, 2023).

The frog fauna of Melanesia is exceptionally species-rich, with over 540 recognised species, and an estimated total exceeding 700 species (Oliver et al. 2022). Many of these appear to have localised distributions, making frogs arguably the best vertebrate group for understanding patterns of finer-scale endemism and turnover between regions of Melanesia (Kraus 2021; Dahl et al. 2023). However, the distributions of many species remain poorly understood, and a high proportion of the Melanesian frog fauna is assessed as Data Deficient on the IUCN Red List when compared with frog faunas of other tropical regions (Oliver et al. 2022). Many Melanesian frog species are also difficult to distinguish from each other without detailed comparisons of morphological, genetic, and/or acoustic features; and the number of taxonomic specialists with expertise in Melanesian frogs is very limited. However, in many ways, frogs are suited for targeting by citizen scientists because they are relatively small, approachable, easy to photograph, and can often be identified by calls that can be recorded on a smartphone (Rowley et al. 2019). Citizen science approaches have proven effective for compiling up-to-date and comprehensive distributional data for many species of frogs in nearby Australia (Rowley et al. 2019; Cutajar and Rowley 2022; Donnellan et al. 2023).

Here, we undertook a review and synthesis of records for Melanesian frogs that have been uploaded to the citizen science portal iNaturalist (www.iNaturalist.com) to better understand how effectively these publicly available data on citizen science platforms are filling knowledge gaps for this megadiverse but poorly-known fauna. We focus on the ability of these data to address the Wallacean shortfall (gaps in knowledge of distribution) and the degree to which gaps in taxonomic knowledge and capacity (the taxonomic impediment) impede identification of records. We also examine the capacity for iNaturalist records to improve knowledge of the conservation status of poorly-known frogs, especially Data Deficient species. Based on these syntheses, we highlight some strengths and limitations of citizen science platforms, and make a number of recommendations to strengthen the capacity of citizen scientists to fill knowledge gaps in poorly-known tropical regions such as Melanesia.

Materials and methods

We use the term Melanesia to refer to the region comprising New Guinea and islands to both the east (Solomon Islands) and west (Halmahera, Ceram, Buru, and surrounding islands) (Oliver et al. 2022). Although this definition differs from some other studies (especially on birds; e.g. Mayr and Diamond (2001)), it captures a biogeographically allied and relatively cohesive frog fauna that is dominated by two largely endemic radiations, the Asterophryinae and the Pelodryadidae (Oliver et al. 2022). For convenience this study excludes Fiji, which has a depauperate and relatively well-known frog biota.

We vetted all photographic records of Melanesian frogs uploaded to iNaturalist in the period up to and including 31 January 2024. Major contributors of Melanesian frog records to iNaturalist were invited to be co-authors. We follow the generic and species-level taxonomy used by AmphibiaWeb (https://amphibiaweb.org/, accessed 31 January 2024). This largely matches with the taxonomy of iNaturalist, with one difference being that we do not use the genus Ranoidea, and instead follow the traditional framework of recognising two genera of pelodryadid treefrogs in Melanesia, specifically Nyctimystes (sensu stricto as it applies to the radiation of torrent-breeding species with vertical pupils) and Litoria (all other species).

Where possible, we identified iNaturalist records to genus and species using external characters visible in photos. Identifications were based on our own experience with the Melanesian fauna supplemented by reference to relevant taxonomic literature where needed. Images of species that were difficult to identify were typically reviewed by two or more of the authors. To aid identifications, the people who uploaded images to iNaturalist were sometimes contacted to request additional information on ecology, calls, or localities. Images that were of insufficient quality or failed to illustrate sufficient diagnostic characters to permit confident assignment to a known (named or unnamed) species were typically identified to the level of genus (or more rarely just to family). Images of frogs determined to represent species that are new to science were also only identified to the level of genus.

As much as possible, we confirmed or suggested identifications on iNaturalist. However, because of the weighting system used by iNaturalist updated species identifications were sometimes downweighted and do not always show up as preferred names as of the time of writing.

We mapped the distributional data for all Melanesian frog species confirmed on iNaturalist as of 31 January 2024 using the online portal Simplemappr (www.simplrmappr.net), and distributions were then compared against data from the IUCN Red List assessments, original descriptions, field guides and other published resources to identify potential range extensions. Candidate range extensions were then reviewed in more detail, including further checks on species identifications and checks on the accuracy of distributional data on iNaturalist to confirm that iNaturalist records represented genuinely new distributional knowledge. We considered a single ‘site’ or ‘locality’ as having a maximum latitudinal and longitudinal range of <10 km or <0.1°, as outlined by Meiri et al. (2018). Accordingly in our final analysis, any records that we were confident were made >10 km from previously known localities were considered range extensions. In most cases, range extensions greatly exceeded spatial error associated with iNaturalist records, but where there was uncertainty, we checked with the uploader of the record.

Species with confirmed range extensions that are currently listed in one of the threatened categories (Critically Endangered, Endangered, and Vulnerable) or in the category Data Deficient were reassessed, while newly described species were assessed for the first time, using the IUCN Red List Categories and Criteria following protocols outlined in Oliver et al. (2022) and Luedtke et al. (2023). Data on population trends are not available for any of these species, so threat assessments were largely based around revised estimates for extent of occurrence coupled with the extent of habitat loss and degradation (Criteria B).

Results

Of the 1347 photographic frog records from the study region publicly available on iNaturalist as of 31 January 2024, 1341 were identified to family, 1340 were identified to a genus, and 1104 were provisionally identified to a species. Based on these records, images of at least 125 described species of frog found in the Melanesian region have been uploaded to iNaturalist, representing 23.6% of the 552 species recognised from the region as of 31 January 2024. Records from Indonesia (68.3%) far outnumbered records from Papua New Guinea (28.1%) and the Solomon Islands (3.6%) (Fig. 1a, b). Within Indonesia, records were concentrated into a small number of relatively accessible areas, including Halmahera and Obi, the Raja Ampat Islands (and especially Waigeo), parts of the Vogelkop Peninsula, and around the townships of Timika and Jayapura (Fig. 1b). Records from portions of the Central Cordillera in Indonesia were very sparse. Records from Papua New Guinea were more evenly spread across the country, although there was again a clear concentration of records around the national capital, Port Moresby (Fig. 1b).

Fig. 1.

Distribution of iNaturalist records for Melanesian frogs uploaded before 31 January 2024. (a) Overall distribution of records and (b) clustering of records for the four most diverse families of frogs in Melanesia. The dominance of records from the Vogelkop Peninsula and Waigeo Island is apparent in both maps. Notes: spatial accuracy of records on iNaturalist varies, so these figures should be seen as broad regional summaries.

Only four species were recorded more than 50 times, comprising two native species (Litoria infrafrenata (n = 163) and Cornufer papuensis (n = 83), and the introduced toads Duttaphrynus melanostictus (n = 86) and Rhinella marina (n = 86)). Together, these four species made up around 30% of records. All other species were represented by less than 50 records each, and usually less than ten records (range, 1–44; mean, 6; median, 3).

Among frog families native to Melanesia, Limnodynastidae had the highest percentage of records identifiable to species (100% of 21 records identified) followed by Pelodryadidae (96.4% of 490 records identified) (Table 1). The families Ranidae (83.0% of 188 records identified) and Ceratobatrachidae (81.5% of 216 records identified) were more challenging to identify, despite being only moderately species-rich (each comprising less than 60 recognised species in the study area). Identification of Microhylidae presented the greatest challenge (48.6% of 261 records identified).

In most cases where we could not identify a record to species, it was because we were simply unsure of whether the image was of a described or undescribed species. In contrast, we identified only a relatively small number of records that we are confident represent undescribed species; these were most obvious in the pelodryadids where records included at least six species from Indonesia that we consider as likely undescribed (see examples in Fig. 2a–d).

Fig. 2.

Examplar images of Melanesian frogs uploaded to iNaturalist that represent likely undescribed species. All records are from Indonesia. (a) Litoria sp., which is an undescribed species from Halmahera that was already known to us, (b–d) three different Litoria sp. from various sites in the Central Cordillera in Indonesia that we were unable to ascribe to any known described species or candidate species. Photo credit: (a–d), C. Lee.

Range extensions are inferred for around a third of the species recorded (42/125) (Table 2), including 12 Data Deficient species, one Vulnerable species (Litoria rueppelli) and five recently described species that have not previously been assessed (four of which were previously only known from a single site) (see Fig. 3a–f for examples). Based on new information on extent of occurrence (EOO) and habitat usage, we conclude that 15 of these previously Data Deficient, Vulnerable, and Not Evaluated species should now be considered Least Concern, one remains Data Deficient, and two should be considered Near Threatened based on their small known range and proximity to plausible threats.

Fig. 3.

Examplar images of Melanesian frogs uploaded to iNaturalist that represent range extensions. All records are from Indonesia. (a) Asterophrys pullifer from Waigeo Island (~450 km westwards range extension), (b) Austrochaperina gracilipes from Merauke area (first publicly available records from Indonesia, ~50 km westwards range extension), (c) Cophixalus salawatiensis (identified by image plus call) from east of Sorong (100 km range extension), (d) Litoria multicolor from northern edge of the Vogelkop Peninsula (~350 km westwards range extension), (e) Litoria pinocchio (identified by image plus call) from near Jayapura (~150 km eastwards range extension), and (f) Litoria verae from south of Nabire (~100 km eastwards range extension). Photo credit: b, e–f (C. Lee); a, c–d (J. Dashper).

iNaturalist records provide the first evidence of the widespread human-associated species Hylarana erythraea and Polypedates leucomystax from mainland New Guinea. Two toad species that are well established in New Guinea also continue to spread in the east (Rhinella marinus) and west (Duttophrynus melanostictus) of the island, respectively. The former species is now present in several areas in Western Province that are close to the border with Indonesia, suggesting it may cross this border soon (if it has not already). Fejervarya spp. also appear to be spreading into new areas of western New Guinea, such as the area around Timika. Two morphologically similar invasive species (Fejervarya cancrivora and Fejervarya limnocharis) are known in the genus and it is uncertain whether one or two species are present.

iNaturalist records also provide the first publicly available photos of several species that were until recently only known from specimens collected many decades ago. Examples include Callulops boettegeri from Halmahera, and possibly Nyctimystes montanus and Litoria pratti from the Arfak Mountains. Records on iNaturalist also demonstrate that Litoria rueppelli (currently listed as Vulnerable) from Halmahera and Obi is an explosive breeder in which breeding males exhibit temporary sexual dimorphism (Fig. 4). Males of this species are typically brown, but Images from Obi Island taken at 09:30 hours on 19 November 2022 show a dense aggregation of over 40 males showing bright yellow colouration clustered around a pond (Fig. 4). One image shows a bright-yellow male mating with what we presume is a dark-brown female (Fig. 4c). Based on observations by one of the authors (JCT) at the time, there were more than 140 males in this aggregation.

Fig. 4.

(a–c) Photographs from Obi Island, Indonesia, uploaded to iNaturalist that provide evidence that Litoria rueppelli is an explosive breeder that exhibits temporary male dimorphism. Males of this species are normally brown, but these images show a large aggregation of bright yellow males calling (a–b), and a bright yellow male in amplexus with a dark-brown animal that is presumably female (c). Photo credit: Joseph C. Thompson.

Discussion

Our synthesis of citizen science records for Melanesian frogs illustrates the potential for platforms such as iNaturalist to help address knowledge gaps and improve understanding of the conservation status of species from poorly-known biotas (Suprayitno et al. 2022; Wangyal et al. 2022). However, it also highlights how logistic and social constraints combine with knowledge gaps such as the taxonomic impediment to place major constraints on the collection and use of citizen science data.

Conclusion

Knowledge gaps for the Melanesian frog fauna (and indeed the Melanesia biota in general) are extreme, with large areas remaining little explored, many taxa known from one or few localities, and likely hundreds of additional species awaiting description (Oliver et al. 2022; Ferreira et al. 2024). Our synthesis of citizen science data for Melanesian frogs illustrates how this background creates potential for even a small number of observations to fill major gaps in our knowledge of distributions and conservation status. At the same time, incomplete taxonomic frameworks, sparse infrastructure, obvious geographic bias and social inequities constrain the utility of citizen science platforms to fill many knowledge gaps. Citizen science platforms do not remove the need for traditional survey and taxonomic work including collection of voucher specimens and tissues, but with ongoing engagement by professional scientists they do provide abundant scope to advance knowledge of diverse and poorly-known biotas.